Search

G-CSF mobilizes hematopoietic stem cells (HSCs) from the bone marrow (BM) into the blood by suppressing a subset of HSC niche supportive macrophages. As macrophages are the central component of erythropoietic islands in BM, spleen and liver, we examined the effect of G-CSF on erythropoiesis in C57BL/6 mice. Mobilizing doses of G-CSF caused a marked whitening of the BM, a 15-fold decrease in the number of phenotypic erythroblasts, a 1.5-fold decrease in polychromatic and orthochromatic erythroblasts, and a 4.5-fold reduction in reticulocytes in the BM. Conversely, more immature pro-erythroblasts increased 4.4-fold. As the cell surface antigen ER-HR3 identifies erythroid island macrophages in mouse liver and spleen, while VCAM-1 and CD169 on macrophages have been independently reported to be critical for erythropoiesis, we followed the expression on these antigens on BM macrophages during mobilization. G-CSF treatment caused a 35-fold reduction in the number of CD11b+ F4/80+ VCAM1+ ER-HR3+ CD169+ Ly6G+ macrophages that paralleled the loss of erythroblasts. As a result, splenic erythropoiesis was up-regulated to compensate for the loss of medullary erythropoiesis with a 4-5 fold increase in pro-erythroblasts, all erythroblast subsets and reticulocytes. In another set of experiments, we quantified medullar erythropoiesis and macrophages during recovery after a 4 day G-CSF treatment. Erythroblasts and supportive macrophages significantly recovered as early as 24 hours after cessation of G-CSF but it took 7 days to normalize to pre-mobilization values. This suggests that mobilizing doses of G-CSF transiently block erythroblast differentiation specifically in the BM (but not the spleen) by affecting central macrophages in erythroid islands. To confirm that CD11b+ F4/80+ VCAM1+ ER-HR3+ CD169+ Ly6G+ macrophages are critical to the maturation of pro-erythroblasts into erythroblasts, we next performed a broad macrophage depletion by injecting clodronate-loaded liposomes, or a selective depletion of CD169+ macrophages in mice knocked-in with diphtheria toxin receptor into the Siglec1 (CD169) gene. Both clodronate liposome treatment in wild-type mice, and diphtheria toxin treatment in Siglec1DTR/+ mice caused a concomitant depletion of CD11b+ F4/80+ VCAM1+ ER-HR3+ CD169+ macrophages, loss of erythroblasts and accumulation of pro-erythroblasts. Unlike G-CSF, these two treatments also blocked splenic erythropoiesis. In conclusion, we propose that 1) CD11b+ F4/80+ VCAM1+ ER-HR3+ CD169+ Ly6G+ macrophages include nursing macrophages at the centre of erythroid islands and are essential for the maturation of pro-erythroblasts to erythroblasts and 2) mobilizing doses of G-CSF transiently stop medullary erythropoiesis by depleting CD11b+ F4/80+ VCAM1+ ER-HR3+ CD169+ Ly6G+ macrophages in erythropoietic islands in the BM, but not in the spleen. Disclosures: No relevant conflicts of interest to declare.

Abstract 2402 The MYB oncogene is widely expressed in acute leukaemias and is important for the continued proliferation of leukemia cells, raising the possibility that MYB may be a therapeutic target. However realization of this potential requires (i) a significant therapeutic window for MYB inhibition, given its essential role in normal haemopoiesis; and (ii) an approach for developing an effective therapeutic. We previously showed that the interaction of Myb with the coactivator CBP/p300 is essential for Myb's intrinsic transforming activity (1). Here we use haemopoietic cells from the Booreana mouse strain, which carries a mutation in Myb that prevents interaction with CBP/p300 (see Fig. 1 and ref. 2), to examine the requirement for this interaction in myeloid transformation and leukaemogenesis. Using this strain and a strain (plt6) carrying a “complementary” mutation in p300 (3), we show that the Myb-p300 interaction is essential for in vitro transformation by the myeloid leukaemia oncogenes AML1-ETO, AML1-ETO9a, MLL-ENL, and MLL-AF9. We further show that unlike cells from wild-type (WT) mice, Booreana cells fail to induce leukaemia upon transplantation into irradiated recipients following transduction with an AML1-ETO9a retrovirus (Fig. 2). These data, as well as reinforcing the notion that MYB is an essential cofactor for myeloid leukaemogenesis, highlight disruption of the Myb-p300 interaction as a potential therapeutic strategy for AML. Furthermore, our results suggest that such a strategy would have a useable therapeutic index since Booreana mice, unlike Myb null mice, are viable. We have started to explore the use of small peptides to provide proof-of-principle for this approach. We have also begun to explore the molecular basis of the requirement for MYB, and the Myb-p300 interaction, in AML using gene expression profiling. This has highlighted several Myb target genes – identified by our previous ChIP-Seq analysis (4) - that have been previously implicated in myeloid leukaemogenesis, as being differentially expressed between WT and Booreana cells transduced with AML1-ETO9a. One striking example is the apparent Myb-p300 dependent activation of Ets2 and Gfi1, accompanied by repression of Gata2 and Gfi1b; this pattern has previously been reported in MLL-AF9 and MOZ-TIF-induced murine AML (5). Disclosures: No relevant conflicts of interest to declare.

Abstract 216 Up to 5% allogeneic healthy donors and up to 40–60% of chemotherapy-treated patients in autologous setting, fail to reach minimal threshold of 2×106 blood CD34+cells/kg in response to G-CSF, precluding transplantation. Plerixafor, a small inhibitor of the chemokine receptor CXCR4, used for 4 days in combination with G-CSF enables this minimal threshold to be reached in up to 60% patients who previously failed to mobilise in response to G-CSF alone. However, the remaining 40% of patients who failed to mobilise in response to G-CSF alone, still fail to mobilize adequately with G-CSF + Plerixafor. In an attempt to further boost HSC mobilization in response to combinations of G-CSF and Plerixafor, we have investigated the role of the hypoxia-sensing pathway in HSC mobilization. HIF-1α (Hypoxia-inducible factor-1α) controls HSC proliferation and self-renewal in poorly perfused hypoxic bone marrow (BM) niches where very quiescent HSC with highest self-renewal potential reside. When O2 concentration is above 2% in the cell microenvironment, HIF-1α protein is rapidly hydroxylated on Pro residues by prolyl hydroxylases PHD1-3. This recruits the E3 ubiquitin ligase VHL, which targets HIF-1α to rapid proteasomal degradation. When O2concentration is below 2% (hypoxia), PHDs are inactive; HIF-1α protein is stabilized, associates with its β subunit ARNT, translocates to the nucleus and activates of transcription and hypoxia-responsive genes. In this study, we have investigated the effect of pharmacological stabilization of HIF-1α protein on HSC mobilization in mice using the HIF-PHD inhibitor FG-4497. We report that FG-4497 treatment stabilizes HIF-1α protein in mouse BM. We find that FG-4497 synergizes with G-CSF and Plerixafor to enhance HSC mobilization. C57/Bl6 mice were in 4 treatment groups: (G) 250μg/kg/day G-CSF alone for 2 days; (GF) G-CSF for 2 days + 20mg/kg/day FG-4497 for 3 days; (GP) G-CSF for 2 days together with16mg/kg Plerixafor 1 hour prior harvest; (GPF) G-CSF together with Plerixafor and FG-4497 with same dosing as above. Mobilization of colony-forming cells (CFC), phenotypic Lin-CD41-Sca1+Kit+CD48-CD150+ HSC, and functional HSC in long-term competitive transplantation assays were measured. Mice in the GF group (G-CSF + FG-4497) mobilized CFC to the blood 4-fold and phenotypic HSC 3-fold more than mice mobilized with G-CSF alone (p To determine which cell types drive HSC mobilization in a HIF-1α-dependant manner, we crossed HIF1αflox/floxmice with osxCre (HIF-1α gene deletion in osteoprogenitors), LysMCre (deletion in myeloid cells), or with SclCreER mice (tamoxifen-induced deletion in HSC). While studies in LysMCre and SclCreER mice are ongoing, we find that deletion of HIF-1α gene in osteoprogenitors (osxCre mice) decreased 2.5-fold the number of CFU/mL blood following 2 and 3 days treatment with G-CSF. This suggests that HIF-1α in osteoprogenitors and their osteoblastic progenies is necessary for optimal mobilization in response to G-CSF. In conclusion, our data highlight the importance of HIF-1α in HSC mobilization and provide a novel therapeutic strategy for increasing HSC mobilization above levels obtained with combinations of G-CSF and Plerixafor. Thus PHD inhibitors could be useful agents in patients who still fail to mobilize in response to G-CSF and plerixafor. Disclosures: Walkinshaw: Fibrogen Inc.: Employment, Equity Ownership.

Abstract 1889 We have previously shown that G-CSF administration impairs HSC niches in the mobilized bone marrow (BM). G-CSF causes rapid suppression (within 2 days) of endosteal osteoblasts and bone formation with concomitant down-regulation of Kit ligand, CXCL12 and angiopoietin-1. This effect is mediated by the depletion of specific populations of BM macrophages1. Considering the very rapid impairment of HSC niches in response to G-CSF, we hypothesized that 1) the most primitive HSC should also mobilize at this very early stage within the first 48 hours of G-CSF treatment, and 2) that down-regulation of HSC niche function should also alter the number or function of HSC remaining in the mobilized BM. To test this, 125μg/kg rhuG-CSF was injected twice daily to C57BL/6 mice; blood and BM harvested at days 2 and 5 of G-CSF treatment to be transplanted into congenic recipients in a long-term competitive repopulation assay (LT-CR). Transplantation of 25μL blood showed a gradual increase in the number of LT-CR cells mobilized in response to G-CSF as measured by donor chimerism in myeloid and B lineages at 16 weeks post-transplant. Expectedly repopulating units (RU) per mL blood progressively increased from 0.2 ± 0.0 (n=6) in steady-state to 2.9 ± 1.6 (n=9) and 82.6 ± 40.4 (n=9) at days 2 and 5 of G-CSF treatment respectively. At 16 weeks post-transplant, BM from primary recipients were transplanted into secondary recipients. Surprisingly, secondary recipients of blood samples collected after 2 and 5 days of G-CSF treatment had equivalent levels of donor chimerism (37.2% ± 6.6% for 2 days G-CSF and 47.1% ± 7.8% for 5 days G-CSF, n = 8 per group). Therefore, although the absolute number of RU mobilized at day 2 of G-CSF was 28-fold lower than at day 5 of G-CSF administration, more primitive serially reconstituting HSC were mobilized at equivalent levels at days 2 and 5 of G-CSF treatment. This supports our hypothesis that most potent serially reconstituting HSC are mobilized as early as day 2 of G-CSF treatment consistent with the disappearance of osteoblasts1. To test the potential of HSC remaining in the BM, BM cells from G-CSF mobilized mice were transplanted in competition with BM cells from congenic mice in steady-state. Donor chimerism at 16 weeks post-transplant showed that competitive repopulation of BM cells was severely impaired at day 5 of G-CSF treatment with the number of RU per 200,000 BM cells decreasing from 4.1 ± 1.4 in steady-state and 5.2 ± 1.6 at day 2 of G-CSF treatment, to only 0.14 ± 0.05 at day 5 of G-CSF treatment. To test whether this 29-fold decrease in competitive repopulation was due to increased HSC proliferation, we measured BrdU incorporation for the last 2.5 days prior to BM harvest as well as cell cycle analysis with Ki67 and Hoechst33342. The proportion of quiescent Lin- Sca1+ Kit+ CD48- phenotypic HSC in G0 phase decreased from 62.8 ±4.0% in steady-state to 43.5±8.2% at day 2 of G-CSF, but surged back to 80.5±1.9% and 75.1±3.5% at days 3.5 and 5 of G-CSF treatment. The proportion of HSC in G1 and S/G2/M phases followed the opposite pattern, up at day 2, down at days 3.5 and 5. This was confirmed by BrdU incorporation for 2.5 days with the number of BrdU+ cells among Lin- Sca1+ KIT+ CD48- cells rising from 35.1±4.0% in steady-state, to 51.2±4.5% at day 2 of G-CSF and going down to 18.1±1.9% at day 3.5 and 23.3±5.5% at day 5 of G-CSF. Therefore, G-CSF recruits phenotypic HSC into cell cycle within the first 2 days of administration, but HSC return to quiescence despite continuing G-CSF. Therefore decreased repopulation potential at day5 of G-CSF is not due to increased cycling. Finally, we noted that the number of Lin-Sca1+KIT+CD48-CD150+ HSC and Lin-Sca1+KIT+CD48-CD150- multipotent progenitors were reduced 2.4- and 2.8-fold respectively (p Disclosures: No relevant conflicts of interest to declare.

Abstract 2334 The endosteal region of the bone marrow (BM) is hypoxic in steady-state and most quiescent hematopoietic stem cells (HSC) reside in hypoxic, poorly perfused niches. Mobilizing doses of G-CSF renders most of the BM space hypoxic. Most cellular effects of hypoxia are mediated by O2-labile hypoxia-inducible transcription factors (HIF). At O2 concentration above 2%, HIF-α is rapidly hydroxylated on Pro residues by the prolyl hydroxylases PHD1, PHD2, and PHD3. HIF-α prolyl hydroxylation recruits the E3 ubiquitin ligase VHL, which targets HIF-α to the proteasome. In hypoxia (O2 < 2%), HIF-α proteins are stable, associate with the β subunit ARNT, and translocate to the nucleus to activate transcription. It has emerged that HIF-1α regulates HSC proliferation and is critical to maintain long-term HSC self-renewal in vivo. In this study, we investigated the effect of pharmacological stabilization of HIF-1α protein on HSC cycling and mobilization in mice using two different HIF prolyl hydroxylase (PHD) inhibitors, dimethyloxalyl glycine (DMOG) and FG-4497. We first assessed whether DMOG and FG-4497 stabilized HIF-1α protein in BM leukocytes in vivo by western-blot. Following a single injection of 400mg/kg DMOG, HIF-1α protein was stabilized for up to 6 hrs in BM leukocytes. With 20mg/kg FG-4497, HIF-1α protein persisted over 12 hours. HIF-1α protein was below detection in the BM from saline injected animals. C57BL/6 mice were injected daily with 400mg/kg DMOG to measure effect on HSC cycling and BrdU incorporation by flow cytometry. Using Hoecht33342 and FITC-conjugated anti-Ki67 antibody, a 18 day DMOG treatment increased the proportion of Lin-negative Kit+ Sca1+ CD48- (L-K+S+48-) HSC in phase G0 from 61±11% to 84±6% (p Since HIF-PHD inhibitors slow HSC cycling in vivo, we tested whether they could protect HSC from sublethal irradiation. Mice were treated with 400mg/kg DMOG or saline for 22 days and then irradiated with 9.0Gy. Both cohorts were leukopenic betweens days 7 and 14 post-irradiation but DMOG treated mice had significantly higher blood leukocytes at days 22 and 30, and higher platelet numbers day 22 and all subsequent time-points suggesting that DMOG enhances HSC survival with higher blood recovery. Finally mice were injected with DMOG for 4 days and pegylated rhuG-CSF 3 and 1.5 days before harvest. DMOG doubled mobilization in response to G-CSF with 1,620±530 CFC / mL blood in response G-CSF alone, and 3,250±830 CFC / mL in the G-CSF + DMOG group (p In conclusion, these data highlight the importance of the hypoxia pathway and HIF in the regulation of HPSC cycling and trafficking in vivo. Furthermore HIF-PHD inhibitors may provide therapeutic opportunities to protect HSC from deleterious effects of irradiation as well as increasing mobilization efficiency for transplantation. Disclosures: Walkinshaw: Fibrogen Inc.: Employment, Equity Ownership.

Hematopoietic stem cells (HSC) reside in specialized niches in the bone marrow (BM), that regulate their survival, proliferation and differentiation. Two types of HSC niches have been reported: endosteal niches in close contact with osteoblasts, and endothelial niches near vascular sinuses. Whether these niches have distinct functions in controlling HSC fate remains unknown. One difference between these two niches is the constitutive expression of E-selectin and P-selectin by BM endothelial cells. E- and P-selectin are two cell adhesion molecules that modulate hematopoietic progenitor cell (HPC) survival, proliferation and differentiation in vitro. We now show that deletion of E-selectin, but not P-selectin, delays HSC turn-over in the BM in vivo. Mice lacking either E-selectin (E−/ −), P-selectin (P−/ −) or both (PE−/ −) were given bromo-deoxyuridine (BrdU) in their drinking water for up to 14 days. Lineage-negative c-KIT+ Sca-1+ CD34− (LKS34) cells were sorted from the BM and stained for BrdU incorporation into genomic DNA. Although it took only 3.6 days for 50% of LKS34 cells from wild-type (WT) and P−/ − mice to incorporate BrdU, 9 days were required for 50% BrdU incorporation in LKS34 cells from E−/ − and PE−/ − double KO mice. Thus, HSC cycling time is 2.5 times slower in the absence of E-selectin. To confirm these findings, LKS cells were stained with rhodamine123, a vital dye that is retained by metabolically active cells but effluxed from quiescent HSC. A higher proportion of LKS cells from E−/ − mice were rhodamine dull (34±2%) than WT LKS (23±1%; p=0.037) confirming that a greater proportion of HSC from E−/ − mice are quiescent. To further support these findings, we determined the effect of E-selectin deletion on HSC recovery following cytotoxic stress with a single dose of 5-fluorouracil (5FU 150mg/kg). As KIT is strongly down-regulated in the BM of 5FU-treated mice, we examined frequency and BrdU incorporation in Lin− Sca1+ CD41− CD48− CD150+ long-term reconstituting HSC. We found HSC recovery to be enhanced in E−/ − mice with a 5-fold increase in HSC numbers per femur compared to WT mice at day 7 post-5FU. Despite the more rapid recovery of E−/ − HSC, BrdU incorporation remained significantly lower in E−/ − HSC on days 3 and 7 post-5FU suggesting the decreased HSC turn-over in the absence of E-selectin protects them from the cytotoxic effect of 5FU. To determine whether this effect was mediated by the two described E-selectin receptors PSGL-1 and/or CD44, BrdU incorporation experiments were repeated with mice lacking both the PSGL-1 and CD44 genes. LKS cell turnover in these mice was identical to that of WT suggesting that the effect is mediated by a distinct unknown receptor(s) on HSC. The fact that a novel E-selectin receptor on HSC/HPC is involved was confirmed both using flow cytometry with selectin-IgM chimeras as well as cell adhesion assays using plastic-adsorbed selectin-IgG chimeras. In both assays, 90-95% of LKS cells from CD44−/ − PSGL-1−/ − double KO mice bound E-selectin whereas adhesion to P-selectin was completely lost. Taken together our findings suggest that E-selectin, whose constitutive expression is restricted to BM endothelial cells, plays an important role in the regulation HSC turnover in vivo, endothelial niches, where E-selectin is expressed, support more rapid HSC turn-over within the BM, and this effect is mediated by unknown E-selectin receptors distinct from PSGL-1 or CD44.

Since the publication of our paper “Regulation of p21WAF1/Cip1 Stability by Wisp39, an Hsp90 Binding Protein” by Jascur et al., (2005, Molecular Cell 17, 237–249), it has come to our attention that the WISp39 gene reported in our paper is identical to the DIR1 gene (Robson et al., 1999xA novel human stress response-related gene with a potential role in induced radioresistance. Robson, T., Joiner, M.C., Wilson, G.D., McCullough, W., Price, M.E., Logan, I., Jones, H., McKeown, S.R., and Hirst, D.G. Radiat. Res. 1999; 152: 451–461Crossref | PubMed | Scopus (36)See all ReferencesRobson et al., 1999).

Abstract 564 The behaviour of a hematopoietic stem cell (HSC) is regulated by its immediate micro-environment or niche. We have identified a novel function for the adhesion molecule E-selectin which is constitutively expressed on bone marrow (BM) vasculature. Using mice knocked-out for E- (E-/-) or P-selectin (P-/-) genes, we investigated whether selectin absence alters HSC behaviour in vivo. We found HSC cycling in the absence of E-selectin to be significantly delayed 2.5-fold in BrdU incorporation assays compared to either P-/- or WT (mice were administered BrdU for 3d then BrdU incorporation in BM Lineage-KIT+Sca1+(LKS+)CD34- or LKS+CD48-CD150+cells measured). To confirm these findings, LKS+ cells were stained with rhodamine123, a vital dye retained by metabolically active cells but not quiescent HSC. More LKS+ cells from E-/- mice were rhodamine dull (34±2%) than WT (23±1%; p=0.037) confirming that a greater proportion of HSC from E-/- mice are quiescent. We then determined whether administration of E-selectin antagonists alone could similarly delay HSC turnover. Mice were administered the glycomimetic E-selectin antagonist GMI-1070, for set periods of time before harvest. We found HSC turnover to be significantly delayed following GMI-1070 administration (1.4 fold less BrdU incorporation, p=0.011) with a concomitant 1.4-fold increase in the number of Rho123 dull LSK+ quiescent HSC per femur (p=0.020). Non-cycling, quiescent HSC are known to be more resistant to chemotherapy and irradiation. Indeed 7 days following 5-FU administration, we found that E-/- mice had faster BM HSC recovery / less HSC damage compared to WT mice, both by phenotype analysis and in a competitive long-term reconstituting assay. Following 5-FU administration the number of reconstituting units/femur in WT mice decreased 5.1-fold but only decreased 2.3-fold in similarly treated E-/- mice. Interestingly, when mice were pre-treated with GMI-1070 before 5-FU, there was significantly enhanced blood neutrophil recovery compared to mice administered 5-FU alone (blood neutrophils were 710±205 ×103/mL with GMI-1070, compared to 234±141 ×103/mL without, at day 9 post-5-FU, p=0.0001). Similarly when mice were severely irradiated and test bleeds performed weekly, a more rapid haematopoietic recovery was observed in E-/- compared to WT mice. In summary, we have identified a novel function for the adhesion molecule E-selectin. HSC turnover is dramatically reduced in E-/- mice an effect that can be replicated by transient administration of E-selectin antagonist mimetics. Furthermore blood leukocyte and HSC numbers recover faster following cytotoxic or irradiation injury in the absence or blockage of E-selectin-mediated cell adhesion. Thus E-selectin may well be a crucial component of the proliferative HSC niche regulating HSC turnover. Blockage of E-selectin adhesive interaction by GMI-1070, a novel E-selectin antagonist that has completed phase I clinical trails, may represent a promising treatment for the protection of HSC during chemotherapy. Disclosures: Winkler: Glycomimetics Inc: Research Funding. Smith:GlycoMimetics, Inc: Employment. Patton:GlycoMimetics, Inc: Employment. Magnani:GlycoMimetics, Inc.: Employment. Levesque:Glycomimetics Inc.: Research Funding.

Cytotoxic chemotherapies used to treat acute leukemia and high-grade lymphoma target rapidly-dividing cells including the malignant cells but also normal HSC needed to replenish the blood and immune system, as well as mucosal cells lining the respiratory and gastrointestinal tracts. Consequences include therapy-induced mucositis (providing a portal of entry for bacteria) and neutropenia leading to increased morbidity and potentially life-threatening infections. In acute myeloid leukemia and high-grade lymphoma patients, treatment-related mortality can reach 20% (Elmaagacli, BMT, 2007). We now report a novel therapy (therapeutic blockade of E-selectin) that may enhance survival after repeated rounds of high-dose chemotherapy. In mice we find the mechanism contributing to this enhanced survival to be a combination of accelerated neutrophil recovery and a dramatic reduction in therapy-related weight loss and intestinal mucositis when E-selectin is therapeutically blocked or genetically-deleted. Our studies were performed using two treatment regimens (chemotherapy or radiation). For chemotherapy, cohorts of mice received repeated rounds of the anti-metabolite, 5-fluorouracil (5-FU) at a dose of 150mg/kg every 10 days without prophylactic antibiotic support. Using this regime, 80% of E-selectin gene-deleted mice (E-/-) survived >6 weeks. In sharp contrast all wild-type mice died (p Deletion of the E-selectin gene also prevented therapy-induced weight loss (from 16% average weight loss in wild-type mice, to only 2% in E-/- mice after 2 rounds 5-FU, p Therapy-induced mucositis is greatly exacerbated by a wave of secondary inflammation following initial direct damage. We found administration of the selective E-selectin antagonist GMI-1271 (20 mg/kg bi-daily) for 5 days following each round of chemotherapy enhanced neutrophil recovery and similarly protected mice from weight loss and mucositis following chemotherapy. Importantly the administration of GMI-1271 effectively blocked secondary migration of inflammatory F4/80+ Ly-6C+ macrophages to intestines of mice following chemotherapy or irradiation. In conclusion, our results suggest that inflammatory macrophages play a major role in chemotherapy-induced mucositis and are recruited to damaged intestine following E-selectin up-regulation. Importantly there is no current highly-effective treatment to prevent mucositis. Therapeutic blockade of E-selectin by administration of GMI-1271 at the same time as chemotherapy represents a potential effective and novel therapy to block recruitment of inflammatory macrophages to damaged mucosa whilst also enhancing neutrophil recovery, thus protecting patients from two chemotherapy-induced side-effects. Therapeutic blockade of E-selectin should provide an opportunity for improved completion rates of toxic chemotherapy or radiotherapy regimes and reduce therapy-related morbidity and mortality. Disclosures: Magnani: GlycoMimetics, Inc.: Employment, Equity Ownership.

The cytokines granulocyte colony-stimulating factor (G-CSF) and Flt3 ligand (Flt3-L) mobilize hematopoietic stem and progenitor cells into the peripheral blood of primates, humans, and mice. We recently reported that G-CSF administration causes a transient blockade of medullar erythropoiesis by suppressing erythroblastic island (EI) macrophages in the bone marrow. In the study described here, we investigated the effect of mobilizing doses of Flt3-L on erythropoiesis in mice in vivo. Similar to G-CSF, Flt3-L caused whitening of the bone marrow with significant reduction in the numbers of EI macrophages and erythroblasts. This was compensated by an increase in the numbers of EI macrophages and erythroblasts in the spleen. However, unlike G-CSF, Flt3-L had an indirect effect on EI macrophages, as it was not detected at the surface of EI macrophages or erythroid progenitors.

The vascular adhesion molecule E-selectin is a key component of the bone marrow (BM) vascular niche, awakening otherwise dormant hematopoietic stem cells (HSC) (Winkler et al., Nat Med 2012). Vascular niches also mediate malignant cell survival. We find E-selectin becomes upregulated on BM vasculature during leukemia and that adhesion of BM AML blasts to E-selectin promotes survival signaling. This is unique to E-selectin - not observed with adhesion to P-selectin, PECAM1, or integrin ligands ICAM1, VCAM1. In vivo we find absence (in SELE-/- mice) or therapeutic blockade of E-selectin by administration of E-selectin antagonist GMI-1271 to mice, induces 9-fold greater chemosensitivity in 11q23-rearranged AML Initiating Cells (LIC) to cytarabine therapy in vivo and doubles survival over that achieved by chemotherapy alone. Together these data support the Phase I/II clinical trial of GMI-1271 in combination with intensive chemotherapy to improve efficacy of therapy for AML (NCT02306291), and raise the further questions: 1) what are the AML LIC ligands interacting with E-selectin at the vascular niche and 2) what are the pathways initiated by E-selectin adhesion that mediate chemoresistance. Using murine models of AML generated by retroviral transduction of the 11q23 fusion oncogene MLL-AF9 into HSC, we find leukemic blasts rapidly upregulate E-selectin binding potential upon oncogenic transformation. To investigate if this is due to aberrant glycosylation facilitating the generation of de novo AML cell surface receptors, we generated AML from HSC knocked-out for the two canonical E-selectin receptors PSGL1 and CD44. Surprisingly we found although HSC from CD44-/-PSGL1-/- mice still bound to E-selectin (indicating HSC express additional E-selectin ligands), AML cells generated from these same HSCs no longer did. Thus the functional E-selectin ligands utilized by murine HSC and AML blasts differ. When transplanted into recipient mice CD44-/-PSGL1-/- AML engrafted with similar disease progression to matching wildtype AML but demonstrated ~100-fold greater sensitivity to high-dose cytarabine chemotherapy. When GMI-1271 was co-administered together with chemotherapy no further boost in chemosensitivity was observed in CD44-/-PSGL1-/- AMLs. This complements in vitro data showing E-selectin adhesion-mediated chemoresistance is lost when CD44 and/orPSGL1are absent in AML cells. Together suggesting that CD44 and/or PSGL1 are the key receptors involved in E-selectin-mediated pro-survival signaling. To understand the pro-survival signaling pathways involved, E-selectin antagonist GMI1271 or saline control were administered 5 days to mice with AML and BM KIT+ leukemic blasts sorted. RNA sequencing revealed only a small number (~200) of mRNA transcripts differentially regulated following therapeutic E-selectin blockade. Of note these included down-regulation in several components of the PI3K/AKT/NF-kB signaling pathway frequently involved in cell survival. Indeed AKT Ser473 and NF-kB p65 Ser536 were found to be rapidly phosphorylated within 15 minutes of E-selectin adhesion in AML blasts in vitro. This phosphorylation was specific to E-selectin adhesion and not observed with other vascular adhesion molecules tested suggesting that direct E-selectin-mediated AKT activation in AML cells at the vascular niche may be the potential mechanism driving vascular mediated chemoresistance. To confirm whether therapeutic blockade of E-selectin alone dampens intracellular AKT signaling in AML blasts, cohorts of mice with wildtype or matching CD44-/-PSGL1-/- AML were administered GMI-1271 for 5 days then BM AML blasts collected for signaling studies. Both the administration of E-selectin antagonist or absence of CD44 and/or PSGL-1 dampened AKTSer473 phosphorylation in BM AML blasts in vivo, correlating with their heightened sensitivity to chemotherapy . In summary, these findings suggest E-selectin ligands on specific CD44 and/or PSGL1 glycoforms are the predominant determinants through which the induction/amplification of chemoresistant pathways in AML blasts occur following vascular adhesion and further that these pathways are coupled to an early phosphorylation event of AKT - and finally explain potential mechanism how GMI-1271 administration is able to attenuate vascular-mediated signaling and restore susceptibility of AML to chemotherapy. Disclosures Magnani: GlycoMimetics, Inc.: Employment, Equity Ownership, Membership on an entity's Board of Directors or advisory committees.

Matrix metalloproteinase (MMP) activity is regulated by tissue inhibitor of metalloproteinases (TIMPs). We found that while TIMP-1 and -2 expressions were unaffected, and TIMP-4 was not expressed, TIMP-3 mRNA expression decreased 10-fold within the bone marrow (BM) during G-CSF induced mobilization. In addition, through reverse zymography, the level of biologically active TIMP-3 protein was also shown to decrease during mobilization. Down-regulation of TIMP-3 may contribute to the accumulation of active MMPs within the BM, allowing for the release of hematopoieticstem/progenitor cells (HSPC) from the BM matrix. By qRT-PCR we have shown 10-fold greater TIMP-3 expression in endosteal mRNA compared to central BM mRNA in mouse femur (p=0.008). To assess which bone associated cell populations expressed the majority of TIMP-3, pooled bones were crushed, collagenase treated and FACS sorted. Mesenchymal progenitors (CD45-Lin-Sca1+) expressed the highest level of TIMP-3 followed by endothelial cells (CD45-Lin-CD31+) and mature osteoblasts (CD45-Lin-Sca1-CD51+). Erythroid progenitors (CD45+Ter119+Kit+), megakaryocyte progenitors (CD45+CD41+Kit+) and megakaryocytes (CD45+CD41+Kit−) from BM were also found to express TIMP-3, but at a level at least 10-fold lower than those of non-hematopoietic stromal cells. All other BM hematopoietic cell types tested were negative for TIMP-3 expression. Immunohistofluorescence on bone sections validated TIMP-3 expression in megakaryocytes, endothelial cells and osteoblasts. Expression of TIMP-3 in mouse platelets was confirmed by reverse zymography. To investigate TIMP-3 function we over-expressed huTIMP-3 in mice via retroviral transduction with MND-X-IRES-eGFP (MXIE) retroviral vector. BM cells retrovirally transduced with MXIE-huTIMP-3 or empty MXIE control was transplanted into lethally irradiated congenic mice. Engraftment and transduction levels were determined by GFP expression. At 3-months post-transplant there were no significant differences in body weight, total blood, spleen or BM cell counts between the two groups. qRT-PCR data showed that over-expressing huTIMP-3 did not alter the expression level of endogenous mTIMP-3. Flow cytometry analysis showed that in mice transduced with MXIE-huTIMP-3, the frequency of GFP+ B cells (CD11b-B220+) was reduced by 50% in the blood from 23.88±12.00% to 11.94±7.85% (p=0.0315) and by 64% in the BM from 25.06±13.78% to 9.02±7.67% (p=0.0188) when compared to MXIE controls. Conversely, the frequency of GFP+ huTIMP-3 expressing myeloid cells (CD11b+) was significantly increased in the blood from 55.69±17.13% to77.91±6.31% (p=0.0005), BM from 58.67±16.32% to 77.32±12.02% (p=0.0244) and spleen from 14.07±3.75% to 28.82±6.85% (p=0.0002). Unexpectedly, the frequency of untransduced GFP- myeloid and B cells were similar between the two groups. Although huTIMP-3 over-expression did not significantly alter the number of GFP+ HSPC (Linage-Sca1+Kit+, LSK) per femur (MXIE 0.03±0.03%, MXIE-huTIMP-3 0.01±0.01%, p=0.1139), LSK turnover in huTIMP-3 over-expressing cells was increased in vivo from 4.36±2.83% to 13.31±5.61% (p=0.0159) as determined by BrdU incorporation following 3 days of BrdU administration. Similarly, a trend was also observed in vitro after 12days of culture, LSK sorted from MXIE-huTIMP-3 mice proliferate faster than MXIE controls from 2.55^6cells/ml±1.05 to 9.6^6cells/ml±0.54 (p=0.1). In summary, huTIMP-3 over-expression in mice increased HSPC proliferation in vivo and in vitro. And whilst the huTIMP-3 over-expression in mice was not at a sufficient level to observe a global effect on total BM haematopoiesis, our data suggests that forced huTIMP-3 over-expression in vivo skews differentiation towards myelopoiesis at the detriment of lymphopoiesis.