ALBERT

Description: Mammalian erythropoiesis has long been established to occur within erythroblastic islands (EBIs), niches where erythroblasts differentiate in close contact with a central macrophage. While it is generally accepted that EBI macrophages play an important role in regulation of erythropoiesis, very little is known about the specific macrophage populations involved in EBI formation, the regulation that occurs within EBIs, or how this niche fits into the broader context of hematopoiesis. We analyzed native EBIs isolated from mouse bone marrow using multispectral imaging flow cytometry (Seu et. al. Front Immunol 2017). Consistent with historical observations, the EBIs were heterogeneous and many contained a number of closely CD11b+ cells in addition to erythroblasts and a central F4/80+ macrophage. Flow cytometry analysis of cells dissociated from native bone marrow EBIs indicated these niches are also enriched 2-3 fold in myeloblasts and granulocytic precursors up to metamyelocytes relative to the total bone marrow while they are depleted of mature granulocytes (bands and segmented cells). Bulk RNAseq of the CD11b+ population isolated from EBIs showed high expression of genes characteristic of the granulocytic lineage (e.g. Elane, Mpo, Gfi1, Cebpe, Camp, and Mmp9), indicating the EBI macrophages may regulate myelopoiesis along with erythropoiesis and that EBIs should really be considered as erythro-myeloblastic islands (EMBIs). To critically document the various hematopoietic cell populations that constitute EMBIs, we used the 10x Genomics Chromium system to obtain single cell gene expression data on ~3,500 total cells from isolated EMBIs along with at least 1,000 sorted cells from each of the 3 major EMBI-associated populations (F4/80+, CD71+, and CD11b+) (Fig 1a, b). The data were analyzed using 10x Genomics' Loupe cell Browser and Iterative Clustering and Guide-gene Selection (ICGS, http://www.AltAnalyze.org, Olsson et. al. Nature 2016). From the ICGS analysis, ~30% of the total EMBI-associated cells were myeloid cells that segregated into at least 3 transcriptionally distinct clusters representing granulocytic progenitors and precursors. As expected, erythroblasts with a progressive maturation pattern made up the bulk (60%) of the EMBI-associated cells, while up to 10% were a heterogeneous population of cells that exhibited expression of macrophage markers such as Csf1R and Irf8, along with genes previously described to characterize resident macrophages, such as Fn1and Fsp1/S100A4 (Fig 1c). In order to investigate the balance of myeloid cells with erythroid cells within the EMBIs, we examined the ratio of CD71+ cells to CD11b+ and how this ratio changes in models of altered granulopoiesis. While the number of myeloid cells at any island varied, the overall ratio of CD11b+ area to CD71+ within the EMBIs was relatively constant at steady state. In three different murine models of anemia of inflammation (AoI), we found that this ratio of CD11b+ to CD71+ cells within the EMBI increases dramatically indicating that the increased granulopoiesis and suppression of erythropoiesis noted in AoI is a result of altered balance of the hematopoiesis within the EMBI unit. Similarly, stimulation of granulopoiesis with GCSF also results in a shift within the EMBIs to CD11b+ myeloid cells and suppression of erythroid cells. Alternatively, in gfi1 KO mice, a model of congenital neutropenia in which granulopoiesis fails at an early stage, the ratio shifts toward CD71+ erythroid cells with paucity of the granulocytic precursors that are typically found at the EMBIs. Taken together, these data indicate that granulocyte progenitors and precursors are specifically associated with EMBI macrophages in the mouse bone marrow. The preferential localization of myeloid precursors within EMBIs suggests this niche is a site for granulopoiesis as well as erythropoiesis and production of these lineages is dynamically regulated within this niche. Our work with multiple murine models of altered granulopoiesis demonstrates that pathological expansion of one of the lineages within this niche may suppress the other and that the interactions within the EMBI could be a useful therapeutic target for AoI. These novel findings significantly broaden our understanding of the role of this hematopoietic niche in the regulated development of lineage committed erythroid and myeloid cells. Disclosures No relevant conflicts of interest to declare.

Description: Autophagy maintains hematopoietic stem cell integrity and prevents malignant transformation. In addition to bulk degradation, selective autophagy serves as an intracellular quality control mechanism and requires autophagy receptors, such as p62 (SQSTM1), to specifically bridge the ubiquitinated cargos into autophagosomes. Here, we investigated the function of p62 in acute myeloid leukemia (AML) in vitro and in murine in vivo models of AML. Loss of p62 impaired expansion and colony-forming ability of leukemia cells and prolonged latency of leukemia development in mice. High p62 expression was associated with poor prognosis in human AML. Using quantitative mass spectrometry, we identified enrichment of mitochondrial proteins upon immunoprecipitation of p62. Loss of p62 significantly delayed removal of dysfunctional mitochondria, increased mitochondrial superoxide levels, and impaired mitochondrial respiration. Moreover, we demonstrated that the autophagy-dependent function of p62 is essential for cell growth and effective mitochondrial degradation by mitophagy. Our results highlight the prominent role of selective autophagy in leukemia progression, and specifically, the importance of mitophagy to maintain mitochondrial integrity.

Description: Acute myeloid leukemia (AML) is a genetically heterogeneous disease where multiple mutations coincide in hematopoietic stem and progenitor cells leading to malignant transformation. One important class of mutations alters the function of signaling intermediates such as Fms-like tyrosine kinase 3 (FLT3), thereby helping AML cells to overcome the physiological communication with their microenvironment. Activating mutations in FLT3 are found in approximately 30% of adult AML cases. Particularly common are internal tandem duplications (ITD) in the juxtamembrane domain of FLT3, which are associated with poor clinical outcome. Recently, a phase II study of the second-generation FLT3 inhibitor AC220 (quizartinib) showed a complete remission rate of 44% to 54% in relapsed/chemotherapy refractory AML. However, secondary point mutations in the FLT3 tyrosine kinase domain have been reported as common causes of acquired clinical resistance to the FLT3 inhibitor AC220. We used quantitative mass-spectrometry-based phosphoproteomics to elucidate and compare the signaling out-put of FLT3-ITD and its AC220-resistant mutants harbouring either the F691L 'gatekeeper' substitution or the D835V activation loop mutant in AML cell models. Our comprehensive signaling analyses profiled thousands of phosphorylation events in a site-specific manner and revealed marked differences in the signaling profiles of the FLT3 mutant variants. In general, we found differential activation of signal transducer and activator of transcription 5 (STAT5) and mitogen-activated protein (MAP) kinase signaling when comparing FLT3-ITD and the AC220-resistant mutants. Interestingly, some cytosolic tyrosine kinases showed differential activation patterns. For instance, spleen tyrosine kinase (SYK) signaling was significantly enhanced downstream of FLT3-ITD-F691L and cells harbouring this mutant showed increased responsiveness to compounds targeting SYK. Our resource study shows, how point mutations conferring resistance to AC220 impact the signaling output of FLT3-ITD and uncovers pathways whose inhibition might be useful to disrupt oncogenic signaling elicited by FLT3-ITD-F691L and -D835V mutants. Disclosures No relevant conflicts of interest to declare.

Description: To sustain lifetime hematopoiesis, adult hematopoietic stem cells (HSCs) are required to remain quiescent. Proper control over cell cycle status underlies HSC fitness. Although the identity of cell cycle mediators which control HSC cycling are basically understood, the gene regulatory networks controlling them are largely unknown. Gfi1 encodes a transcriptional repressor protein that is required to maintain HSC quiescence. Over-cycling of stem-progenitors causes Gfi1-/- HSC exhaustion and prevents long-term engraftment in lethally-ablated recipients. Gfi1 controls myeloid progenitor maintenance through the direct regulation of HoxA9, but the transcriptional programming necessary to control HSC function is not known. Gfi1-mediated repression is critically dependent upon the recruitment of known corepressors; including the Lsd1/CoREST complex, and Eto proteins. Moreover, Gfi1 directly represses a suite of microRNA genes, including miR-21, miR-196b and miR-302b. We were intrigued by a series of papers in which miR-302b was shown to target the Lsd1 corepressor, which was subsequently shown to be required for HSC fitness. Lsd1 serves as a transcriptional co-repressor for Gfi1, Gfi1b, Scl1/Tal1, Bcl11a, Sall4, and Runx1. Knockdown and genetic ablation studies demonstrated that loss of Lsd1 results in a derepression of stem/progenitor genes, and leads to defects in HSC self-renewal and marrow failure. Given this published example, we next sought to determine whether miR-196b and miR-21 might also affect Gfi1 cofactors. First, we find that miR-196b targets CBFA2T3/Mtg16, encoding Eto2 (in human and mouse, respectively). Eto2 is the most abundant Eto family protein in hematopoietic stem and progenitors, and functions as a transcriptional co-repressor with Gfi1, Gfi1b, Scl1/Tal1, Gata-1, Lmo2, and Ldb1. In luciferase-sensor assays for the CBFA2T3-3'UTR, cotransfection with a miR-196b mimic reduces luciferase activity, dependent upon a canonical miR-196b binding site. In agreement with published data, we find that Mtg16-/- HSC are reduced in number and are impaired in competitive transplant assays, but this defect can be partially alleviated by increasing Mtg16-/- HSC numbers. We also find that the initial serial transplant of Mtg16-/- HSC leads to immediate exhaustion. Second, we find that miR-21 targets Ski which functions as a transcriptional corepressor for Smad, Gli, MAD, and thyroid hormone receptor transcription factors. In luciferase-sensor assays for the Ski 3'UTR, cotransfection with a miR-21 mimic reduces luciferase activity, dependent upon a canonical miR-21 binding site. We find that both synthetic and endogenous Ski and Gfi1 proteins directly and specifically interact with one another, implicating Ski as a new Gfi1-corepressor. Recently, overexpression of Ski was shown to enhance HSC repopulation in transplant recipients and caused myeloproliferative disease by promoting HSC gene signatures independent of Tgf-β signaling. Ski-/- is perinatal lethal due to cleft palate, so we analyzed Ski-/- hematopoiesis by transplanting E14.5 fetal liver cells. Similar to Mtg16-/- HSC, Ski-/- HSCs display impaired competitive engraftment that can be rescued by increasing Ski-/- HSC numbers. We also find that the initial serial transplant of Ski-/- HSC leads to immediate exhaustion. Finally, in Gfi1-/- lineage negative bone marrow (which expresses elevated levels of miR-21 and miR-196b) the level of Ski and Eto2 protein are significantly decreased. Our findings suggest that Gfi1 repression of individual Gfi1-target miR genes is required for the proper control of Gfi1 co-factors and subsequent HSC biology. To test this concept we generated mice with tetracycline-inducible miR-196b expression (in mice with intact Gfi1 expression). After only one month of transgene induction we find significant HSC exhaustion. Our data suggests that repression of Gfi1 target genes miR-21, miR-196b and miR-302b is critical for normal HSC function, because these miR target essential transcriptional cofactors. Disclosures No relevant conflicts of interest to declare.

Description: Activating mutations of Rho-family of small GTPases have been linked to lymphoproliferative disorders, although the pathogenesis mechanism involved is unknown. BCR-ABL (p190) B-cell acute lymphoblastic leukemia (B-ALL) arises from the expression of the oncofusion protein BCR-ABL in a B-cell progenitor. The transforming effect of BCR-ABL is dependent on the tyrosine kinase (TK) activity of the fusion protein that leads to autophosphorylation, recruitment of adaptor proteins, and subsequent activation of downstream signaling. TK inhibitors (TKIs) have been used as frontline treatment for Ph+ B-ALL patients. However, relapse is common in Ph+ B-ALL despite high rates of complete response with initial therapy, probably because of survival of leukemic progenitors. These BCR-ABL+ progenitors appear to develop additional epigenetic and genetic alterations that result in proliferative advantage frequently associated with silencing of the cyclin dependent kinase inhibitor Cdkn2a, even before mutant Cdkn2a gene deleted cells are selected during clonal evolution. Recent work by our group (Chang KH et al., Blood 2012) identified the Rho GTPase guanine nucleotide exchange factor Vav3 in BCR-ABL mediated lymphoid leukemogenesis. We showed that the deficiency of the guanosine nucleotide exchange factor Vav3 delays leukemogenesis and phenocopies the effect of Rac2 (and combined Rac2/Rac1) deficiency (Thomas EK et al., Cancer Cell 2007; Sengupta A et al., Blood 2010), a downstream effector of Vav3. Upregulated Vav3 expression and activation only partly depend on ABL TK activity, and Vav3 deficiency collaborates with tyrosine kinase inhibitors to impair leukemogenesis in vitro and in vivo through impaired proliferation and survival. On the other hand, our group has demonstrated that Bmi1 overexpression frequently found in BCR-ABL+ B-ALL results in B-cell progenitor reprogramming through acquisition of a stem cell-like phenotype (Sengupta A et al., Blood 2012). Bmi1 forms part of the classical polycomb repression complex 1 (PRC1) where its component Ring1A/B catalyzes histone H2A mono-ubiquitination at lysine 119, which in conjunction with the PRC2 complex activity leads to chromatin compaction and repression of target genes. Through epistasis experiments, we found that Vav3 or Rac2 deficiency abrogates the oncogenic effect of Bmi1 overexpression. Co-immunoprecipitation experiments in nuclear and cytoplasmic cell extracts demonstrated that Vav3 and Rac1/Rac2 co-immunoprecipitate with Bmi1 in the nucleus but not in the cytosol of BCR-ABL+ leukemic cells. Interestingly, control non-BCR-ABL expressing nuclear extracts show minimal, if any, level of co-immunoprecipitation. This co-immunoprecipitation is not directly induced by BCR-ABL since BCR-ABL does not co-immunoprecipitate with Vav3/Rac1/Rac2 but does with Bmi1, suggesting that nuclear Vav3 activity may be dissected from the TK activity of BCR-ABL. Biochemically, the overexpression of Bmi1 results in increased activation of nuclear Rac which is practically abrogated by the deficiency of Vav3 as assessed in cellular pulldown assays of primary leukemic B-cell progenitors. As expected, downstream expression of Cdkn2a is repressed by overexpression by Bmi1. Deficiency of Vav3 restores the expression of Cdkn2a to control levels. This data suggests a transcriptional regulatory role of the signaling proteins Vav3/Rac2 in the nucleus. Chromatin immunoprecipitation (ChIP)-qPCR for Bmi1, Ring1B and polycomb repressive histone marks (H2AK119 and H3K27me3) and the assay for Tn5-transposase accessible chromatin (ATAQ)-qPCR for the Cdkn2a locus in Vav3- or Rac2-deficient, BCR-ABL+ primary B-cell progenitors were compared with their BCR-ABL, Vav3/Rac2 expressing counterparts. These assays confirmed that Vav3 and Rac2 are essential for PRC dependent transcriptional repression of Cdkn2a through occupancy of the Cdkn2a promoter and decreased accessibility to Cdkn2a chromatin. In conclusion, our studies establish for the first time an association between nuclear Vav3/Rac and polycomb repressive activities in p190-BCR-ABL+ leukemogenesis through their activity on the Cdkn2a locus. Vav3 may represent a novel target for adjuvant therapy with TKI in BCR-ABL+ lymphoblastic leukemia. Disclosures No relevant conflicts of interest to declare.

Description: The transforming growth factor beta (TGF-β) signaling pathway controls hematopoietic stem cell (HSC) behavior in the marrow niche; however, TGF-β signaling becomes chronic in early-stage myelodysplastic syndrome (MDS). Although TGF-β signaling normally induces negative feedback, in early-stage MDS, high levels of microRNA-21 (miR-21) contribute to chronic TGF-β signaling. We found that a TGF-β signal–correlated gene signature is sufficient to identify an MDS patient population with abnormal RNA splicing (eg, CSF3R) independent of splicing factor mutations and coincident with low HNRNPK activity. Levels of SKI messenger RNA (mRNA) encoding a TGF-β antagonist are sufficient to identify these patients. However, MDS patients with high SKI mRNA and chronic TGF-β signaling lack SKI protein because of miR-21 activity. To determine the impact of SKI loss, we examined murine Ski−/− HSC function. First, competitive HSC transplants revealed a profound defect in stem cell fitness (competitive disadvantage) but not specification, homing, or multilineage production. Aged recipients of Ski−/− HSCs exhibited mild phenotypes similar to phenotypes in those with macrocytic anemia. Second, blastocyst complementation revealed a dramatic block in Ski−/− hematopoiesis in the absence of transplantation. Similar to SKI-high MDS patient samples, Ski−/− HSCs strikingly upregulated TGF-β signaling and deregulated expression of spliceosome genes (including Hnrnpk). Moreover, novel single-cell splicing analyses demonstrated that Ski−/− HSCs and high levels of SKI expression in MDS patient samples share abnormal alternative splicing of common genes (including those that encode splicing factors). We conclude that miR-21–mediated loss of SKI activates TGF-β signaling and alternative splicing to impair the competitive advantage of normal HSCs (fitness), which could contribute to selection of early-stage MDS-genic clones.

Description: Efforts to understand the genomic impact of human-disease-relevant genetic lesions and how they disrupt the normal sequence of cell-state transitions is hampered by a lack of defined hierarchical cellular states and corresponding networks of regulatory genes (transcription factors). Severe congenital neutropenia (SCN) patients display inherited and de novo mutations in Growth factor independent-1 (GFI1), which encodes a zinc-finger transcription factor. We identified known (e.g. N382S in zinc finger 5) and novel GFI1 sequence changes in SCN patients, then used lentiviral mediated expression to functionally evaluate them. GFI1-N382S, GFI1-K403R and GFI1-R412X mutations (in zinc finger 6) significantly elevated the expression of the Gfi1 target gene, Irf8. We generated mice with these patient-derived SCN-associated mutations in the murine Gfi1 locus. Neonatal and adult Gfi1N382S/- and Gfi1R412X/-mice are neutropenic, but Gfi1K403R/- mice have normal steady-state neutrophil levels. The resulting steady-state dysgranulopoiesis in adult mice was further pronounced in neonates. We noted that Gfi1R412X/-mice accumulate less Gfi1 protein than Gfi1+/+, while Gfi1R412X/R412Xhomozygous alleles genetically rescued both the hypomorphic protein defect and substantially restored neutrophil numbers (though not to normal). In contrast, functional challenge with neutrophil-dependent pathogens in vivo revealed a broad susceptibility for all Gfi1-mutant mice. To determine the underlying mechanism of neutropenia and immune defects, we first used novel flow cytometry analyses and Fluidigm C1 single cell RNA-Seq to establish the successive genomic states encompassing normal granulocyte specification and commitment. Independent CITE-Seq/10x sequencing analysis provided direct correlation between flow cytometry populations and genomic information, while also establishing the trajectory through genomic states traversed during terminal granulopoiesis. Next, using a novel bioinformatics algorithm (cellHarmony) we assigned Gfi1-mutant cells to their respective wild-type cell states and then determined differential gene expression. We find few genes deregulated across granulopoiesis, and that the bulk of transcriptional impact on Gfi1 target genes is specific to successive granulopoietic cell states. These insights facilitated Gfi1R412X/- Irf8+/-genetic rescue of granulocytic specification, but not post-commitment defects. We noted that a portion of Gfi1R412X/-gene deregulation unrepaired by genetic rescue was enriched for chromosome organization, proteolysis, and innate immune effectors. Electron microscopy revealed uncondensed chromatin in mature Gfi1R412X/-neutrophils while SWATH proteomics identified a loss of neutrophil granule proteins and members of the NADPH oxidase complex (potentially linking SCN with chronic granulomatous disease genes). To this end, we functionally validated impaired NADPH oxidase complex function in neutrophils from Gfi1-mutant mice. We noted Gfi1 mutant mice have consistently elevated levels of granulocyte colony stimulating factor (but not other cytokines), and so we extended our analysis of oxidative burst to GCSF-rescued human SCN patients to find profound defects; underscoring the inability of genetic or cytokine rescued specification to resolve post-commitment defects. We illustrate a work flow that can be broadly applied to molecularly dissect translationally relevant mouse models of disease, and underscores the necessity of evaluating mutations within the context of relevant cell states. Disclosures Dwivedi: Abbvie: Employment. Myers:Bellicum Pharmaceuticals: Membership on an entity's Board of Directors or advisory committees; Novartis: Membership on an entity's Board of Directors or advisory committees. Nazor:BioLegend: Employment.

Description: Key Points Xenografted ALL cells faithfully recapitulate CNS leukemia and are characterized by high expression of VEGF, mediating CNS entry of ALL cells. VEGF captured by bevacizumab in vivo specifically reduces CNS leukemia, providing a novel strategy to target CNS involvement in ALL.

Description: Differentiation from hematopoietic stem and progenitor cells (HSPCs) to committed blood lineages is dependent on lineage specific transcription factors that simultaneously promote gene expression that commits progenitors to specific lineages while repressing genes associated with alternative lineages. In addition to transcription factors, small non-coding microRNAs (miRNAs) also have the potential to influence cell fate decisions through negative regulation of lineage specific genes. We previously observed that germline knockout of the mirn23a miRNA cluster (which codes for mature miRNAs miR-23a, miR-27a, and miR-24-2) resulted in increased common lymphoid progenitors (CLPs) and B cells with a concomitant decrease in granulocyte/monocyte progenitors (GMPs) and their progeny. This was the first evidence of a miRNA being able to influence a lymphoid/myeloid cell fate decision using a genetic knockout model. To follow up these results, we sought to identify a detailed molecular mechanism of the mirn23a-/- mouse phenotype. Evaluation of HSPC populations by flow cytometry revealed that while mirn23a-/- mice have no difference s in their LT-HSC populations, they show imbalanced levels of MPP3 and MPP4 populations, suggesting that bifurcation from the MPP2 to the MPP3/MPP4 is the earliest cell type regulated by mirn23a to influence hematopoietic cell fate decisions. RNA and protein analysis of multipotent EML cell lines generated from wildtype and mirn23a-/- mice revealed that mirn23a negatively regulated critical HSPC genes Runx1, Satb1, Ikzf1, Mef2c, Bach1, and Bach2 that are involved in committing MPPs to CLPs. Additionally, genes associated with the commitment of CLPs to B cells, EBF1 and Pax5, were also increased. We observed that miR-24-2 target, Trib3, antagonizes PI3K/AKT signaling to promote EBF1 and Pax5 expression through nuclear accumulation of FoxO1. Trib3 also agonizes the BMP/Smad pathway through negative regulation of E3-ubiquitinase Smurf1. Ex vivo OP9 cultures with primary mirn23a-/- cells cultured with FoxO1 and BMP inhibitors revealed that both the PI3K/Akt and BMP/Smad pathway are critical for mirn23a-/- phenotypes. Consistent with mirn23a being a critical gene for myeloid commitment of hematopoietic progenitors, we observe that B Cell factor EBF1 represses transcription of mirn23a, creating a regulatory feedback loop between mirn23a and EBF1. In addition to mirn23a, there is a homologous mirn23b miRNA cluster that is expressed at lower levels in hematopoietic cells. To investigate compound loss of mirn23a and mirn23b in adult hematopoiesis, we generated mirn23a-/-mirn23bf/f Mx-1 cre mice to circumvent variable embryonic/ neonatal lethality. These mice showed a further increase in B lymphopoiesis and decrease in myelopoiesis compared to mirn23a-/- mice. Interestingly, these mice also exhibited decreased bone marrow cellularity at 3 weeks post mirn23b excision. As judged by overall numbers and percent of bone marrow, LT-HSC, MPP, and LSK populations were decreased. We are currently investigating the underlying mechanism for the decreased stem cells and overall cellularity. Overall, these results show that mirn23a/b miRNAs bias cell fate decisions at the MPP through negative regulation of critical lymphoid transcription factors. Sustained commitment to the B cell lineage is dependent on both the PI3K/Akt and BMP/Smad signaling pathways, both of which are regulated by mirn23a target Trib3. In turn, EBF1 negatively regulates mirn23a, creating a regulatory feedback loop between EBF1 and mirn23a. Compound loss of mirn23a/mirn23b results in decreased bone marrow cellularity and stem cell loss. Disclosures No relevant conflicts of interest to declare.

Description: In acute lymphoblastic leukemia (ALL), central nervous system (CNS) directed therapy is required to achieve long-term remission and survival even in patients without detectable CNS disease, indicating subclinical CNS manifestation in many patients. Hence, prophylactic CNS therapy is indispensable but bears the risk of secondary neoplasms and neurocognitive deficits in ALL survivors. Therefore, a better understanding of mechanisms mediating CNS involvement in ALL are needed in order to identify potential targets for prophylactic and therapeutic intervention. In this study, we transplanted primary patient B cell precursor (BCP) ALL cells onto NOD/SCID mice and investigated engraftment of human leukemia cells in the recipient's peripheral blood (PB), bone marrow (BM), spleen (S), and meninges by flowcytometry staining for huCD19. Upon disease onset, we identified meningeal infiltration of human ALL cells together with leukemia engraftment in BM, S and PB in a subset of samples (CNSpos) in contrast to absent CNS involvement despite high leukemia cell infiltration of BM, S and PB in other recipients (CNSneg). CNSpos and CNSneg phenotypes were consistently observed in subsequent xenograft passages. In CNSpos animals meningeal leukemia infiltration was also detected by immunohistochemistry on brain sections and meningeal enhancement was detected by small animal magnetic resonance imaging in CNSpos recipients. We further characterized ALL cells isolated from meningeal and BM infiltrates by gene expression profiling and identified the gene coding for vascular endothelial growth factor A (VEGF) to be highly expressed in ALL cells isolated from the CNS as compared to BM derived cells. Differential expression of VEGF was validated by qPCR and confirmed in independent sample cohorts. Most interestingly, reported functions of VEGF include regulation of cellular growth, vascular permeability, and trans-endothelial cell migration, and elevated VEGF protein levels have previously been described in cerebrospinal fluid specimens collected from ALL and AML patients with CNS leukemia. VEGF signals through its receptors 1 or 2 (VEGFR1/2). On all primary ALL samples, only expression of VEGFR1 but not VEGFR2 was detected. We used the BCP-ALL cell line Nalm-6, which also expresses VEGF and VEGFR1 and mediates a clear CNSpos phenotype upon engraftment. However, cellular survival, proliferation, apoptosis, and metabolic activity were not affected, neither upon exposure to VEGF or the antagonizing VEGF antibody bavacizumab, nor by stable VEGF overexpression or knock-down, thus indicating absence of autocrine VEGF/VEGFR1 signaling. We further analyzed whether paracrine VEGF signaling through VEGFR2, which upon activation mediates endothelial cell permeability, is involved in trans-endothelial migration of ALL cells leading to leukemia infiltration of the CNS. Brain endothelial cells (bEND.3) incubated with VEGF showed increased phosphorylation of VEGFR2 downstream signaling molecules (Src, AKT), which indicates activated signaling mediating cellular permeability. Moreover, we modeled migration of ALL cells through brain endothelial cells in a transwell assay and observed significantly increased migration of Nalm-6 ALL cells through bEND.3 monolayers upon exposure to VEGF or upon overexpression of VEGF as compared to controls. Vice versa, significantly lower numbers of migrated leukemia cells were detected after incubation with bevacizumab or upon VEGF knock down indicating VEGF dependent trans-endothelial migration. Thus, we identified high expression of VEGF in CNS derived leukemia cells, absent autocrine signaling of VEGF on ALL cells and, most importantly, VEGF dependent trans-endothelial migration of ALL cells indicating VEGF as a possible mediator of CNS leukemia. Finally, we investigated the impact of VEGF on CNS leukemia manifestation in vivo. Recipient animals transplanted with 3 CNSpos primograft samples (4 experiments, one in repetition) were treated with bevacizumab or control. In all experiments, anti-VEGF treatment significantly reduced the leukemia burden exclusively in the CNS but not in BM, S, and PB compartments, indicating that transmigration of leukemia cells and CNS manifestation in ALL is mediated by VEGF. Thus, targeting of VEGF signaling may serve as a novel strategy to control CNS leukemia in patients. Disclosures No relevant conflicts of interest to declare.