Search

Somatic mutations in splicing factor genes are drivers of myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML). The splicing factors U2AF1 and U2AF2 form the U2AF heterodimer that is critical in the 3' splice site (3'SS) recognition and in the recruitment of U2 small nuclear ribonucleoproteins for the activation of the spliceosome complex. U2AF1 carries hotspot mutations in its two RNA binding motifs; yet the molecular mechanisms affecting the splicing process and promoting clonal advantage remain unclear, albeit necessary to develop effective targeted therapies. We applied a multi-omics approach comparing the activities of two U2AF1 mutants (S34F and Q157R) in MDS/AML cell lines and primary samples. Using a novel approach of fractionated enhanced crosslinking immunoprecipitation coupled with deep RNA-sequencing (freCLIP-seq), we mapped transcriptome-wide binding at nucleotide resolution and we identified conformational changes in mutant vs wild-type U2AF1 binding. Specifically, we observed an emergent peak in position -3 of the 3'SS for the S34F mutant and in position +1 for the Q157R mutant, matching the critical positions observed by differential splicing analysis on RNA-seq data. Altered U2AF1-RNA binding compromised U2AF2-RNA interactions, resulting predominantly in exon exclusion and intron retention. Combined binding-splicing analysis showed that while the Q157R mutant mainly exhibits loss of binding, the S34F mutant follows a gain-of-binding pattern, where splicing progression appears impaired by increased mutant binding. Functional analysis of genes affected by both binding and splicing alterations revealed that U2AF1 mutants alter RNA granule biology, affecting in particular stress granule-enriched transcripts and proteins. Stress granules are membrane-less cytoplasmic assemblies of RNAs and RNA binding proteins that improve cellular adaptation in response to stress conditions. Increased stress granule formation has been linked to tumorigenesis as a strategy exploited by cancer cells to regulate gene expression and signal transduction, enhancing their fitness under stress. To probe how aberrant binding and splicing of stress granule components affected stress granule biology, we assessed stress granule formation in U2AF1 mutant vs wild-type cells at steady state and after stress induction with sodium arsenite treatment. Immunofluorescent staining followed by confocal imaging demonstrated that U2AF1 mutations enhance stress granule formation upon arsenite stress in both cell lines and primary samples. RNA turnover analysis by TimeLapse-seq confirmed that U2AF1 S34F and Q157R mutations promote stability/synthesis of transcripts that are enriched in stress granules and determine degradation/shutdown of transcripts that are depleted in stress granules, providing a molecular explanation for the increase in stress granules observed by imaging. Finally, we were able to corroborate our observations by single-cell RNA-seq in patient-derived U2AF1-mutant MDS blasts, establishing the causal link between U2AF1 mutations and upregulation of stress granule components. Collectively, this multi-omics analysis identified biological processes directly influenced by mutant U2AF1 binding and splicing, laying the foundation for a new paradigm where splicing factor mutations enhance stress granule formation by acting on the availability of their RNA and protein components. The enhanced formation of stress granules potentially fosters the stress adaptation of U2AF1-mutant cells, contributing to their clonal advantage in MDS/AML. Stress granule perturbations may therefore represent a novel therapeutic vulnerability in U2AF1-mutant MDS/AML patients and possibly in patients carrying other splicing factor mutations. Disclosures Hunck: Boehringer Ingelheim: Other: Fellowship.

Humanized mouse model is a powerful tool for the study of human hematopoiesis and hematologic diseases in vivo. However, existing models continue to be limited by absent engraftment of stem cells with inherent growth defects, especially in bone marrow failure (BMF) disorders such as low-grade myelodysplastic syndromes (MDS) and Fanconi Anemia (FA). Previously, we presented a highly efficient MDS/AML PDX model in cytokine-humanized MISTRG mice, that express human M-CSF, IL-3, GM-CSF and THPO in the background of the SIRPα - humanized rag -/- IL2Rg -/- mice. MISTRG mice support efficient, faithful and long-term MDS hematopoietic stem and progenitor cell (HSPC) engraftment with multi-lineage representation. Similar to MDS, FA stem cells are defective and sensitive to toxic stress conferred by inflammation or irradiation. Irradiation, required in conventional mice as pre-conditioning, increases the risk of secondary malignancies, induces activation of inflammatory pathways and may damage the bone marrow microenvironment. In this study, we present MISTRG6kit W41 mice that in addition to the above cytokines express human IL-6, which improves lymphoid development, and that were CRISPR modified to introduce the white spotting 41 mutation in c-Kit (c-kit W41), which results in partial loss of c-kit function in the murine host stem and progenitor cells. To characterize this mouse model, we compare the engraftment capacity between MISTRG6 and MISTRG6kit W41 mice. We find that the HSPCs from both healthy donor and low grade MDS patient can be more efficiently propagated in MISTRG6Kit W41 mice and that they can be serially transplanted. Human cells are detectable at higher levels in PB of MISTRG6Kit W41 mice as early as 8 weeks post transplantation. As previously described, c-kit mutation enhances human erythroid engraftment while multi-lineage engraftment is maintained in engrafted MISTRG6Kit W41 mice. Therefore, MISTRG6Kit W41 mice offer a better host environment that supports human hematopoietic expansion without irradiation. FA is an inherited genetic disease carrying FA gene mutations associated with DNA damage and cancer predisposition and marked by BMF, MDS and AML. FA HSPCs cannot be expanded or grown for sufficient periods in vitro, and existing mouse models have failed to demonstrate engraftment of primary cells from FA patient. We transplanted a MDS sample from a FANCC mutant patient into MISTRG6Kit W41 mice giving rise to a myeloid predominant graft that could be successfully serially transplanted. Moreover, non-transformed FA HSPCs from 2 FANCA and 2 FANCG mutant patients were successfully engrafted in MISTRG6Kit W41 mice. In summary, MISTRG mice with human IL-6 expression and c-Kit W41 mutation provide a more efficient and convenient way for pre-clinical studies of BMFs such as FA and low-grade MDS with maintenance of an intact humanized stem cell niche, and this is the first time we saw primary cells from FA patient propagate in mouse model. We believe that our model would be highly beneficial to the study of additional diseases associated with DNA damage or cancer predisposition, such as dyskeratosis congenita or Schwachman-Diamond syndrome. Disclosures No relevant conflicts of interest to declare.

During hematopoietic differentiation from hematopoietic stem cells (HSCs) to mature blood cells, cells undergo a metabolic shift from glycolysis to mitochondrial respiration. The mechanisms by which hematopoietic cells adjust their energy metabolism are still under investigation. N6-mehyladenosine (m 6A) mRNA modification has been reported to regulate numerous fundamental cellular processes through control of RNA stability or translational efficiency. The fat mass and obesity-associated protein (FTO), an m 6A m and m 6A mRNA demethylase, has been reported to affect cellular metabolism in acute myeloid leukemia (AML). ALKBH5, the specific RNA m 6A demethylase, controls oncogene expression in AML. ALKBH5 becomes highly expressed in hematopoietic progenitors during hematopoietic development but the physiological role of RNA m 6A demethylase during hematopoiesis remains unknown. To investigate the function of the RNA m 6A demethylase ALKBH5 in hematopoiesis, we generated Vav-iCre +; Alkbh5fl/fl (vcAlkbh5-/-) mice, resulting in deletion of Alkbh5 specifically in the hematopoietic system. vcAlkbh5-/-mice showed no hematopoietic defects at steady states up to 12 months of age. We applied TimeLapse-seq on lineage-depleted bone marrow cells of WT and vcAlkbh5-/- mice to determine whether loss of ALKBH5 perturbed mRNA stability and/or RNA turnover. Ogdh mRNA was the most destabilized transcript resulting in significantly reduced OGDH protein levels. OGDH is the rate-limiting enzyme in the tricarboxylic acid (TCA) cycle. Inhibition of OGDH subsequently induces production of L-2-hydroxyglutarate (L-2-HG), whose metabolism is closely coupled to energy metabolism through inhibition of oxygen consumption. L-2-HG, the enantiomer of D-2-HG, inhibits the function of a-ketoglutarate (a-KG)-dependent enzymes, including TET and KDM enzymes. We measured L- and D-2-HG in the plasma of WT and vcAlkbh5-/- mice by chiral derivatization to distinguish the two enantiomers. Although D-2-HG levels were similar in the plasma of WT and vcAlkbh5-/- mice, L-2-HG levels were significantly increased in the plasma of vcAlkbh5-/- mice. We therefore determined the function of Jumonji C-domain lysine demethylases (JmjC-KDMs) by measuring histone methylation: H3K9me3, H3K27me3 and H3K36me3 modifications were all significantly increased in Alkbh5-deficient hematopoietic cells. We next sought to understand whether reduction of OGDH expression and resulting increased L-2-HG levels production could impair energy metabolism via perturbation of the TCA cycle and oxidative phosphorylation (OXPHOS) in the mitochondria. We isolated lineage negative hematopoietic stem and progenitor cells (HSPCs) from WT and vcAlkbh5-/- mice and subjected these to the Seahorse ATP Rate Assay. Comparing oxygen consumption rate (OCR) data and the kinetics of the Extracellular Acidification Rate (ECAR) of both groups, we found that less ATP was produced by mitochondria of the vcAlkbh5-/- cells, while ATP produced by glycolysis showed no difference between the two groups. In the meantime, the ultrastructure of mitochondria in the Alkbh5-deficient cells remains normal. We next determined whether the attenuated energy metabolism of Alkbh5-deficient HSPCs was functionally relevant by testing HSPC function in competitive transplantation assays. Interestingly, vcAlkbh5-/- cells showed a significant competitive defect at all differentiation stages except in phenotypic long-term HSCs (LT-HSCs). This suggests that LT-HSCs, thought to preferentially rely on glycolysis as opposed to OXPHOS for their energy source, are protected from loss of ALKBH5 and OGDH. In conclusion, our study demonstrates that ALKBH5 modulates energy metabolism by regulating mRNA stability of metabolic enzymes through its m 6A demethylation activity during hematopoiesis. This finding links Alkbh5 expression kinetics to the metabolic shift from glycolysis to mitochondrial OXPHOS during hematopoietic development. Disclosures No relevant conflicts of interest to declare.

In vivo models of human erythropoiesis with generation of circulating mature human red blood cells (huRBC) have remained elusive, limiting studies of primary human red cell disorders. In our prior study, we have generated the first combined cytokine-liver humanized immunodeficient mouse model (huHepMISTRG-Fah) with fully mature, circulating huRBC when engrafted with human CD34+ hematopoietic stem and progenitor cells (HSPCs)1. Here we present for the first time a humanized mouse model of human sickle cell disease (SCD) which replicates the hallmark pathophysiologic finding of vaso-occlusion in mice engrafted with primary patient-derived SCD HSPCs. SCD is an inherited blood disorder caused by a single point mutation in the beta-globin gene. Murine models of SCD exclusively express human globins in mouse red blood cells in the background of murine globin knockouts2 which exclusively contain murine erythropoiesis and red cells and thus fail to capture the heterogeneity encountered in patients. To determine whether enhanced erythropoiesis and most importantly circulating huRBC in engrafted huHepMISTRG-Fah mice would be sufficient to replicate the pathophysiology of SCD, we engrafted it with adult SCD BM CD34+ cells as well as age-matched control BM CD34+ cells. Overall huCD45+ and erythroid engraftment in BM (Fig. a, b) and PB (Fig. c, d) were similar between control or SCD. Using multispectral imaging flow cytometry, we observed sickling huRBCs (7-11 sickling huRBCs/ 100 huRBCs) in the PB of SCD (Fig. e) but not in control CD34+ (Fig. f) engrafted mice. To determine whether circulating huRBC would result in vaso-occlusion and associated findings in SCD engrafted huHepMISTRG-Fah mice, we evaluated histological sections of lung, liver, spleen, and kidney from control and SCD CD34+ engrafted mice. SCD CD34+ engrafted mice lungs showed an increase in alveolar macrophages (arrowheads) associated with alveolar hemorrhage and thrombosis (arrows) but not observed control engrafted mice (Fig. g). Spleens of SCD engrafted mice showed erythroid precursor expansion, sickled erythrocytes in the sinusoids (arrowheads), and vascular occlusion and thrombosis (arrows) (Fig. h). Liver architecture was disrupted in SCD engrafted mice with RBCs in sinusoids and microvascular thromboses (Fig. i). Congestion of capillary loops and peritubular capillaries and glomeruli engorged with sickled RBCs was evident in kidneys (Fig. j) of SCD but not control CD34+ engrafted mice. SCD is characterized by ineffective erythropoiesis due to structural abnormalities in erythroid precursors3. As a functional structural unit, erythroblastic islands (EBIs) represent a specialized niche for erythropoiesis, where a central macrophage is surrounded by developing erythroblasts of varying differentiation states4. In our study, both SCD (Fig. k) and control (Fig. l) CD34+ engrafted mice exhibited EBIs with huCD169+ huCD14+ central macrophages surrounded by varying stages of huCD235a+ erythroid progenitors, including enucleated huRBCs (arrows). This implies that huHepMISTRG-Fah mice have the capability to generate human EBIs in vivo and thus represent a valuable tool to not only study the effects of mature RBC but also to elucidate mechanisms of ineffective erythropoiesis in SCD and other red cell disorders. In conclusion, we successfully engrafted adult SCD patient BM derived CD34+ cells in huHepMISTRG-Fah mice and detected circulating, sickling huRBCs in the mouse PB. We observed pathological changes in the lung, spleen, liver and kidney, which are comparable to what is seen in the established SCD mouse models and in patients. In addition, huHepMISTRG-Fah mice offer the opportunity to study the role of the central macrophage in human erythropoiesis in health and disease in an immunologically advantageous context. This novel mouse model could therefore serve to open novel avenues for therapeutic advances in SCD. Reference 1. Song Y, Shan L, Gybli R, et. al. In Vivo reconstruction of Human Erythropoiesis with Circulating Mature Human RBCs in Humanized Liver Mistrg Mice. Blood. 2019;134:338. 2. Ryan TM, Ciavatta DJ, Townes TM. Knockout-transgenic mouse model of sickle cell disease. Science. 1997;278(5339):873-876. 3. Blouin MJ, De Paepe ME, Trudel M. Altered hematopoiesis in murine sickle cell disease. Blood. 1999;94(4):1451-1459. 4. Manwani D, Bieker JJ. The erythroblastic island. Curr Top Dev Biol. 2008;82:23-53. Disclosures Xu: Seattle Genetics: Membership on an entity's Board of Directors or advisory committees. Flavell:Zai labs: Consultancy; GSK: Consultancy.

Spliceosomal gene mutations function as drivers of hematologic malignancies and other cancers with an occurrence of more than 50% in myelodysplastic syndromes and secondary acute myeloid leukemia. Hotspot mutations S34F and Q157R in the two zinc finger domains of the splicing factor U2AF1, forming with U2AF2 the U2AF complex that recognizes 3' splice site (3'SS) of U2 introns, alter exon usage in a sequence-specific manner. However, how pathological U2AF1 mutations disrupt ordered splicing, from binding to recruitment of cooperating RNA binding proteins and ultimately splicing kinetics, is still not known at the molecular level. To obtain unique insights into in vivo RNA binding mechanisms, we performed fractionated enhanced crosslinking immunoprecipitation coupled with deep RNA sequencing (freCLIP-seq) on human erythroleukemia (HEL) cells expressing wild-type (WT) or mutant (S34F, Q157R) U2AF1. Transcriptome-wide analysis of binding at single nucleotide resolution in light and heavy fractions, corresponding respectively to U2AF1 only and U2AF complex, allowed to: i) deconvolute U2AF1 signal peaking over the AG dinucleotide at the intronic end of the 3'SS region, and U2AF2 signal sitting on the adjacent polypyrimidine tract (PPT); ii) identify conformational changes in mutant U2AF1 binding with a novel peak in position -3 of the 3'SS region for S34F and in position +1 for Q157R. Alternative splicing analysis on newly collected RNA-seq data showed that less included exons present higher probability of U in position -3 for S34F and A in position +1 for Q157R, pinpointing a match with nucleotide positions affected by aberrant binding in freCLIP-seq. In both U2AF1 mutants, aberrant binding and splicing mechanisms affected genes involved in mRNA processing and transport (P-value We then performed a combined analysis of differential binding and aberrant splicing in U2AF1 mutants vs WT considering 4 categories: ">inclusion/>binding", "binding", ">inclusion/binding" with 123 events out of 309 (Figure 1A). Moreover, differential binding was not dependent on specific nucleotides in position -3: events characterized by increased S34F binding (Figure 1B, top), as well as events characterized by decreased S34F binding (Figure 1B, bottom), showed -3U in less included exons or -3C in more included exons. The binding analysis across the 4 categories showed that increased S34F binding was associated with reduced U2AF2 binding (Figure 1C, top) particularly in less included exons, while decreased S34F binding was associated with increased U2AF2 binding (Figure 1C, bottom) especially in more included exons. Finally, analysis of branch point and splice junction features revealed that PPT strength influences the splicing outcome with "binding" category characterized by a weak PPT that impairs U2AF2 binding in the presence of skewed U2AF1 S34F binding (Figure 1D). Additionally, transcriptome-wide RNA kinetics analysis by TimeLapse-seq demonstrated that U2AF1 S34F and Q157R, compared to WT, globally decrease synthesis of aberrantly spliced and bound 3'SS regions. Of note, this shutdown effect was particularly evident in the downstream exons pointing towards a role of U2AF1 mutations in a widespread alteration of RNA synthesis and splicing dynamics. Collectively, these results disclose novel molecular mechanisms of pathogenic U2AF1 mutations in the context of myeloid malignancies and provide the basis for the development of effective U2AF1 directed therapeutic strategies. Disclosures Hunck: Boehringer Ingelheim Fonds: Other: MD Fellowship.

Myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML) are heterogeneous clonal disorders. Isocitrate dehydrogenase-1 and -2 (IDH1/2) mutations are detected in ~20% of AML and ~5% of MDS, in which they confer gain of a neomorphic function that leads to the production of (R)-2-hydroxyglutarate (2HG). Targeted inhibition of mutant IDH1/2 has resulted in significant responses in IDH1/2 mutant MDS and AML but is not curative and patients relapse (Stein et al. Blood 2016, DiNardo et al. N Engl J Med 2018). 2HG accumulation inhibits the function of histone demethylases (KDM4A and KDM4B) that are critical for the homologous recombination (HR) DNA repair pathway and consequently for the repair of DNA double strand breaks (DSBs) (Mallette et al. EMBO J 2012, Sulkowski et al. Sci Transl Med 2017). In HR deficient tumors, Poly-ADP-Ribose Polymerase (PARP) is essential for DNA single strand break (SSB) repair. In IDH mutant tumors PARP inhibitors induce synthetic lethality by suppressing the repair of SSBs, which eventually get converted into DSBs (Javle and Curtin Br J Cancer 2011). We previously demonstrated that in AML, IDH1/2 mutations impair DNA damage response by inducing a defect in HR, and that this renders leukemia cells susceptible to PARP inhibitors in vitro. We hypothesized that this vulnerability would also exist in IDH mutant MDS and more importantly, that this vulnerability would persist in MDS/AML resistant to IDH1/2 inhibitors. To determine whether PARP inhibition targets IDH mutant MDS/AML in vivo, we took advantage of 2 syngeneic mouse models of MDS and AML relying on co-mutation of SRSF2/IDH2 and FLT3/IDH2, respectively. Olaparib (PARP inhibitor) effectively targeted IDH2 mutant but not IDH2 wild type MDS/AML (Fig. 1A). We next sought to determine whether PARP inhibition mediated synthetic lethality persists in MDS/AML resistant to targeted IDH inhibition. We transduced IDH2 mutant murine cells with IDH2 WT or IDH2 MUT lentiviral vectors carrying one of two published resistance mutations. While these resistance mutations conferred resistance to the targeted IDH2 inhibitor Enasidenib, cells remained sensitive to Olaparib (Fig. 1B). Patient MDS/AML is heterogeneous and in general carries additional genetic mutations and epigenetic alterations. We therefore engrafted IDH1/2 WT and MUT MDS/AML patient samples in cytokine humanized immunodeficient mice and treated with vehicle of Olaparib. Engrafted mice were assigned to vehicle or Olaparib 8 weeks after transplantation based on equal engraftment levels determined by BM aspiration. Mice were treated with vehicle or Olaparib via IP injection for 21 days. Human engraftment levels and plasma 2-HG levels were significantly reduced in Olaparib treated animals when compared to pre-treatment and vehicle-treated mice (Fig. 1C). Of note, when equal numbers of huCD34+ cells from vehicle or Olaparib treated mice were transplanted into next generation mice, engraftment was significantly higher for recipients of human cells from vehicle than Olaparib treated mice, suggesting that Olaparib is toxic to leukemia initiating cells. In contrast, IDH WT MDS/AML was insensitive to Olaparib treatment (Fig. 1C). In conclusion, PARP inhibition is effective in vivo against IDH mutant MDS/AML and can overcome targeted IDH inhibitor resistance. Disclosures Flavell: Zai labs: Consultancy; SMOC: Equity Ownership; Troy: Equity Ownership; Artizan Biosciences: Equity Ownership; GSK: Consultancy; Rheos Biomedicines: Equity Ownership. Prebet:Boehringer Ingelheim: Research Funding; Boehringer Ingelheim: Research Funding; novartis: Honoraria; pfizer: Honoraria; Boehringer Ingelheim: Research Funding; Agios: Consultancy, Research Funding; novartis: Honoraria; Jazz Pharmaceuticals: Consultancy, Honoraria, Research Funding; pfizer: Honoraria; Genentech: Consultancy; pfizer: Honoraria; novartis: Honoraria; Tetraphase: Consultancy; novartis: Honoraria; pfizer: Honoraria; Bristol-Myers Squibb: Honoraria, Research Funding; novartis: Honoraria; pfizer: Honoraria. Bindra:Cybrexa: Consultancy, Equity Ownership.

The murine host has remained a readily available and ethically acceptable model for the study of human diseases and therapeutic testing. Immunodeficient mouse models support engraftment of human hematopoietic stem cells (HSC) but with limitation in efficiency and mature lineage representation. Combined knock-in of several non-crossreactive human cytokines (M-CSF, IL3/GM-CSF, and Thrombopoietin) into the corresponding murine loci in the SRG strain (in short termed "MISTRG") has enhanced engraftment and maintenance of human HSCs with multi-lineage differentiation (Rongvaux et al. Annu Rev Immunol 2013, Deng et al. Nature 2015, Saito et al. Blood 2016, Theocharides et al. Haematologica 2016). Despite robust HSC engraftment and myelo- and erythropoiesis in bone marrow (BM), all humanized immunodeficient mouse models lack of mature human red blood cells (RBC), platelets, and myeloid cells in peripheral blood (PB) (Rahmig et al. Stem Cell Reports 2016, Yurino et al. Stem Cell Reports 2016, Song et al. Nat Commun 2019). Yet, full maturation and representation of all myeloid lineages in PB is essential to study diseases of the HSC, such as MDS, and of the RBC, such as sickle cell anemia or malaria. With universal absence of a murine adaptive immune system the culprit is likely the murine host's innate immune system. Previous studies have shown that treatment of engrafted mice with liposomal clodronate that abrogates murine (and human) macrophages, with or without cobra venom factor that eliminates complement, can increase mature human circulating RBC, but only transiently and with significant toxicity. We first sought to determine the site of huRBC sequestration and destruction. Intravital imaging after injection of CFSE labelled huRBC identified the murine liver as the major site of RBC destruction. While muRBC rapidly circulate through the liver circulation, huRBC have greatly increased transit times and are sequestered in liver vessels. We hypothesized that humanization of the murine host's liver could potentially alleviate huRBC sequestration and significantly increase circulating huRBC. In previous studies deletion of fumarylacetoacetate hydrolase (Fah) in the Rag-/-Il2rg-/- background has allowed humanization of the liver and served to study diseases such as malaria (Vaughan et al. J Clin Invest 2012). The liver is the site of synthesis of numerous proteins, some of which directly impact hematopoiesis and blood cells, such as complement. We deleted the Fah gene via CRISPR/Cas9 in MISTRG mice and crossed MISTRG-Fah-/- mice to homozygosity (MISTRGFah). MISTRGFah are viable, fertile, and healthy when maintained on drinking water supplemented with 2-(2-nitro-4-trifluoromethylbenzoyl)-1, 3-cyclohexanedione (NTBC), that blocks tyrosine metabolism upstream of Fah and prevents buildup of hepatotoxic metabolites. At 8 weeks of age we implanted MISTRGFah mice with commercially available, adult human hepatocytes (HuHep) via direct injection into the splenic vein, followed by gradual withdrawal of NTBC water. Regulated buildup of intracellular fumarylacetoacetate results in death of murine Fah-/- hepatocytes and regeneration with HuHep with up to 90% repopulation by HuHep (Azuma et al. Nature biotechnology 2007). When plasma human albumin levels reached 2mg/dL, indicative of significant (~80%) HuHep repopulation, we sublethally (80cGy) irradiated HuHepMISTRGFah mice and engrafted each mouse with 105 fetal liver (FL) derived CD34+ cells. 10 weeks post transplantation, mice were analyzed for engraftment and specifically erythroid maturation in PB. HuHepMISTRGFah mice had significantly higher levels of BM and interestingly spleen erythropoiesis and circulating huRBC in PB (Fig.1 a). CD235a+ huRBC in HuHepMISTRGFah mice are enucleated (Hoechst neg) and mature as evident by loss of CD49d (ITGA4) and gain of Band3 staining (Hu et al. Blood 2013) (Fig.1 b). Interestingly, human erythroid cells in MISTRG but not HuHepMISTRGFah mice are coated with murine complement C3 (muC3) (Fig.1 c) suggesting that liver humanization results in loss of muC3 expression. In conclusion, we have generated the first humanized mouse model with fully mature, circulating huRBC when engrafted with human CD34+ stem and progenitor cells. Ongoing studies are testing the applicability of this model to MDS and sickle cell disease. Disclosures Flavell: SMOC: Equity Ownership; Zai labs: Consultancy; GSK: Consultancy; Artizan Biosciences: Equity Ownership; Troy: Equity Ownership; Rheos Biomedicines: Equity Ownership.