ALBERT

Description: Abstract 2476 The inappropriate expression of EVI1 is associated with aggressive myelodysplastic syndromes (MDS) and acute myeloid leukemia. We previously showed that in vivo EVI1 causes an MDS-like disease in mice and that in vitro it alters cell cycling, self-renewal, and differentiation pathways of primary bone marrow (BM) cells. It is likely that these alterations contribute to the oncogenic potential of EVI1. MicroRNAs (miRNA) are small noncoding RNAs that are 21 to 25 nt in length that up- and down-regulate gene expression during cell development, proliferation, differentiation, and apoptosis. MiRNAs play a critical role in solid and hematopoietic cancers. We found that in hematopoietic cells the expression of several miRNAs is significantly altered by EVI1. Among them, miRNA-124 is strongly downregulated in EVI1-expressing primary BM and 32Dcl3 cells. To evaluate the role of miRNA-124 in primary murine BM cells, we isolated Lin- cells and infected the cells with the empty vector, EVI1, or both EVI1 and miRNA-124. We confirmed that EVI1 alters cellular pathways regulating self-renewal, S1 phase entry, and differentiation. More importantly, however, we found that when miRNA-124 is co-expressed with EVI1, the expression of proteins such as Cyclin D3 and BMI-1, that control some of these pathways, is normalized. These results suggest that the effects of EVI1 are mediated by miRNA-124 repression. We hypothesized that the down-regulation of miRNA-124 could be associated with inappropriate DNA methylation of the miRNA. Both in man and mouse, there are three known alleles of miRNA-124. In the mouse, the alleles are located on chromosome bands 2H4, 3A1, and 14D1. Analysis of the three alleles indicates that miRNA-124-3 harbors the larger number of CpG dinucleotides, including 24 CpG within a stretch of 190 bp upstream of and overlapping the stem-loop start-site. By using the bisulfite DNA sequencing method, we found that in BM cells the number of methylated CpG dinucleotides in this region is significantly increased after EVI1 expression. The DNA methylation is mostly centered in two clusters, which we named cluster I and cluster II. These data clearly show that miRNA-124 is a critical target of EVI1 in human diseases. Disclosures: No relevant conflicts of interest to declare.

Description: Target antigen density has emerged as a major factor influencing the potency of CAR T cells. Our laboratory has demonstrated that the activity of numerous CARs is highly dependent on target antigen density (Walker et al., Mol Ther, 2017), and high complete response rates in a recent trial of CD22 CAR T cells for B-ALL were tempered by frequent relapses due to decreased CD22 antigen density on lymphoblasts (Fry et al., Nat Med, 2018). To assess if antigen density is also a determinant of CD19 CAR T cell therapeutic success, we analyzed CD19 antigen density from fifty pediatric B-ALL patients treated on a clinical trial of CD19-CD28ζ CAR T cells. We found that patients whose CD19 expression was below a threshold density (2000 molecules/lymphoblast) were significantly less likely to achieve a clinical response than those whose leukemia expressed higher levels of CD19. In order to further understand this limitation and how it may be overcome, we developed a model of variable CD19 antigen density B-ALL. After establishing a CD19 knockout of the B-ALL cell line NALM6, we used a lentivirus to reintroduce CD19 and then FACS sorted and single cell cloned to achieve a library of NALM6 clones with varying CD19 surface densities. CD19-CD28ζ CAR T cell activity was highly dependent on CD19 antigen density. We observed decreases in cytotoxicity, proliferation, and cytokine production by CD19 CAR T cells when encountering CD19-low cells, with an approximate threshold of 2,000 molecules of CD19 per lymphoblast, below which, cytokine production in response to tumor cells was nearly ablated. Given that a CD19-4-1BBζ CAR is FDA approved for children with B-ALL and adults with DLBCL, we wondered whether CARs incorporating this alternative costimulatory domain would have similar antigen density thresholds for activation. Surprisingly, CD19-4-1BBζ CAR T cells made even less cytokine, proliferated less, and had further diminished cytolytic capacity against CD19-low cells compared to CD19-CD28ζ CAR T cells. Analysis by western blot of protein lysates from CAR T cells stimulated with varying amounts of antigen demonstrated that CD19-CD28ζ CAR T cells had higher levels of downstream signals such as pERK than CD19-4-1BBζ CAR T cells at lower antigen densities. Accordingly, calcium flux after stimulation was also significantly higher in CD19-CD28ζ than CD19-4-1BBζ CAR T cells. In a xenograft model of CD19-low B-ALL, CD19-4-1BBζ CAR T cells demonstrated no anti-tumor activity, while CD19-CD28ζ CAR T cells eradicated CD19-low leukemia cells. Therefore, the choice of costimulatory domain in CAR T cells plays a major role in modulating activity against low antigen density tumors. CD28 costimulation endows high reactivity towards low antigen density tumors. We confirmed the generalizability of this finding using Her2 CAR T cells; Her2-CD28ζ CAR T cells cleared tumors in an orthotopic xenograft model of Her2-low osteosarcoma, while Her2-4-1BBζ CAR T cells had no effect. This finding has implications for CAR design for lymphoma and solid tumors, where antigen expression is more heterogeneous than B-ALL. To enhance the activity of CD19-4-1BBζ CAR T cells against CD19-low leukemia, we designed a CAR with two copies of intracellular zeta in the signaling domain (CD19-4-1BBζζ). T cells expressing this double-zeta CAR demonstrated enhanced cytotoxicity, proliferation, cytokine production, and pERK signaling in response to CD19-low cells compared to single-zeta CARs. Additionally, in a xenograft model, CD19-4-1BBζζ CAR T cells demonstrated enhanced activity against CD19-low leukemia compared to CD19-4-1BBζ CAR T cells, significantly extending survival. The addition of a third zeta domain (CD19-4-1BBζζζ) further enhanced the activity of CAR T cells. However, inclusion of multiple copies of the costimulatory domains did not improve function. In conclusion, CD19 antigen density is an important determinant of CAR T cell function and therapeutic response. CD19-CD28ζ CARs are more efficient at targeting CD19-low tumor cells than CD19-4-1BBζ CARs. The addition of multiple zeta domains to the CAR enhances its ability to target low antigen density tumors. This serves as proof of concept that rational redesign of CAR signaling endodomains can result in enhanced function against low antigen density tumors, an important step for extending the reach of these powerful therapeutics and overcoming a significant mechanism of tumor escape. Disclosures Lee: Juno: Consultancy.

Description: Mature red blood cells (RBC) contain approximately 95% cytosolic hemoglobin for the purpose of blood oxygen transport. This specialized state is achieved during erythropoiesis by regulated gene expression and protein degradation. During late-stage erythropoiesis, ubiquitin ligases eliminate unnecessary proteins and maintain quality control by degrading unstable proteins, including unpaired hemoglobin subunits. However, ubiquitin ligases are expressed at all stages of erythropoiesis and the functions of most are unknown. To study ubiquitin ligases involved in RBC formation, we performed a Cas9/single guide (sg) RNA screen for functional ubiquitin-proteasome components in HUDEP-2 cells, an immortalized human cell line that proliferates as immature erythroblasts and can be induced to undergo terminal maturation. We identified the E3 ubiquitin ligase FBXO11 as a top-ranked candidate. FBXO11 is a member of the F-box protein family that assembles into a SKP1-CUL1-F-box (SCF) E3 ubiquitin ligase complex. Depletion of FBXO11 by 2 different sgRNAs in HUDEP-2 cells inhibited erythroid maturation, as evidenced by reduced hemoglobinization, failure to induce the maturation marker Band3 and persistence of immature cell morphology. In primary human CD34+ cells, suppression of FBXO11 expression by Cas9 + two independent sgRNAs inhibited erythroid maturation, as evidenced by reduced Band3 expression (5.3% vs. 15.6% for non-targeting sgRNA, P 〈 0.001; n=4). RNA-seq analysis of FBXO11-depleted HUDEP-2 cells revealed 951 decreased transcripts (enriched for erythroid genes) and 339 increased transcripts (enriched for genes expressed in activated T-cells) compared to control cells expressing non-targeting gRNA (P 〈 0.05). Thus, FBXO11 is required for erythroid maturation and facilitates erythroid gene expression. We sought to establish how FBXO11 modulates erythropoiesis and erythroid gene expression by identifying the relevant ubiquitination substrate(s). Combined quantitative proteome analysis with RNA-seq of FBXO11-depleted HUDEP-2 cells identified several proteins that are upregulated with no change in their corresponding mRNA. We tested whether reduction of these candidate substrates could alleviate the erythroid maturation block conferred by FBXO11 depletion. In FBXO11 gene-disrupted HUDEP-2 cells, suppression of the heterochromatin-associated protein BAHD1 partially rescued hemoglobinization and Band3 expression (4.2% for Cas9 + non-targeting sgRNA vs. 21.7% for Cas9 + BAHD1 sgRNAs, P 〈 0.01; n=3) . Conversely, stable overexpression of V5-epitope-tagged BAHD1 in WT HUDEP-2 cells reduced Band3 expression from 25.0% to 11.4% (P 〈 0.001; n=3) and inhibited hemoglobinization. Transcriptome analysis demonstrated a significant inverse correlation between genes deregulated by BAHD1-V5 overexpression and FBXO11 deficiency in HUDEP-2 cells, particularly for erythroid genes that are downregulated (P 〈 0.0001). BAHD1, named after its bromo-adjacent homology domain that interacts with H3K27me3, is part of a transcriptional repressor complex. We showed that BAHD1 and FBXO11 co-immunoprecipitated in cells and that BAHD1 amino (N)-terminal segments of 188 or 240 amino acids were robustly modified with ubiquitin by SCFFBXO11 complex. Chromatin immunoprecipitation-sequencing analysis of BAHD1-V5-expressing WT HUDEP-2 cells showed strong enrichment for BAHD1 occupancy on erythroid gene promoters that were downregulated by FBXO11-deficiency (P 〈 0.0001). We next investigated whether a specific set of histone marks distinguish FBXO11-regulated genes in normal erythroblasts. We found that most FBXO11-regulated genes in both HUDEP-2 and primary CD34+ derived erythroblasts harbor histone marks H3K4me3 and H3K27me3, indicating a "bivalent" epigenetic state that supports low level transcription in stem or progenitor cells. Together, these data indicate that FBXO11 activates expression of erythroid genes by ubiquitinating and degrading bivalent promoter-bound BAHD1 repressor complexes with likely resolution to a monovalent transcriptionally active state. Overall, our findings identify FBXO11 as a ubiquitin ligase that utilizes a novel mechanism to activate erythroid genes during RBC formation. This newly identified pathway may contribute to known activities of FBXO11 as a tumor suppressor and developmental regulator in non-erythroid tissues. Disclosures No relevant conflicts of interest to declare.

Description: The primitive hematopoietic stem/progenitor cells (HSPCs) during embryonic hematopoiesis are thought to be short-lived (SL) with limited self-renewal potential. The fate and consequence of these short-lived HSPCs, once reprogrammed into “long-lived” in a living animal body, remain unknown. Here we show that targeted expression of a dominant-negative C/ebpα (C/ebpαDN) in the primitive SL-HSPCs during zebrafish embryogenesis extends their life span, allowing them to survive to later developmental stage to colonize the definitive hematopoietic sites, where they undergo a proliferative expansion followed by erythropoietic dysplasia and embryonic lethality because of circulation congestion. Mechanistically, C/ebpαDN binds to a conserved C/EBP-binding motif in the promoter region of bmi1 gene, associated with a specific induction of bmi1 transcription in the transgenic embryos expressing C/ebpαDN. Targeted expression of Bmi1 in the SL-HSPCs recapitulates nearly all aberrant phenotypes induced by C/ebpαDN, whereas knockdown of bmi1 largely rescues these abnormalities. The results indicate that Bmi1 acts immediately downstream of C/ebpαDN to regulate the survival and self-renewal of HSPCs and contribute to the erythropoietic dysplasia.

Description: MicroRNAs (miRNAs) are small RNAs that bind Argonaute (Ago) family proteins and recruit them to target mRNAs based on seed sequence complementarity, thereby causing mRNA degradation and/or translational repression. Multiple miRNAs are dynamically expressed during erythropoiesis. Deletion of miR-144/451, the most abundantly expressed erythroid miRNA gene, causes anemia and increased red cell sensitivity to oxidant stress, in part by de-repressing the target mRNA Ywhaz. However, the total number of mRNAs targeted by miR-144/451 is unknown. To identify erythroblast miR-144/451 target genes comprehensively, we used a technique named HITS-CLIP to sequence mRNA fragments bound to Ago proteins in erythroblasts from miR-144/451 wild-type (WT) and knockout (KO) mouse fetal livers. Using a novel peak-calling algorithm (YODEL), we determined that Ago binds to 6,651 peaks on 3,533 mRNAs (Fig 1A). Of these, 1/3rd (2,212 peaks on 1,414 mRNAs) were depleted in KO, indicating that Ago binding to these sites depended on the presence of miR-144/451. Seed sequences for deleted miRNAs (451, 144-5p, 144-3p) were enriched in the depleted peak set (Fig 1B), while seeds for non-deleted miRNAs (486a, 16-5p, 122-5p) were concentrated in the non-depleted set (Fig 1C), validating the specificity of our analysis. We then performed RNA-Seq and quantitative proteomics by TMT-mass-spectrometry on WT and KO erythroblasts. mRNAs directly targeted by miRNAs are expected to be stabilized in KO, compared to mRNAs altered in abundance through indirect transcriptional effects. We inferred mRNA stability from RNA-Seq data by calculating the ratio of exonic and intronic signals in KO and WT cells (Fig 2). Using these criteria, 131 mRNAs showed increased stability (genes within red dotted oval) in KO erythroblasts compared to WT, 100 of which also showed increased protein levels by mass spectrometry. In contrast, only 12 genes showed reduced RNA stability and protein in KO. Notably, very few genes were altered solely at the protein level, indicating that repression by miR-144/451 occurs largely through mRNA degradation, not translation inhibition. Surprisingly, most mRNAs bound by miR-144/451-Ago complexes did not show altered stability or translation (1,414 bound vs. 131 altered mRNAs). To investigate this mismatch, we examined HITS-CLIP peaks for additional features predictive of target regulation. Peaks in 3prime UTRs were more likely to regulate mRNA levels than peaks in coding exons (p=10-7), and peaks containing canonical seed sequences matching miR-144/451 were more likely to regulate mRNA levels than those lacking them (p=10-5). This indicates that while miRNAs recruit Ago to a large number of mRNA sites, the location of the binding site within the mRNA, and the degree of seed match, are important determinants of target regulation. Our combined studies identified numerous mRNAs that are targeted directly by miR-144/451. Ndufb5, Cox10 and Hccs showed increased mRNA stability and increased protein in KO, and showed miR-144/451 dependent HITS-CLIP peaks with canonical seed sequences. All three mRNAs encode components of the mitochondrial electron transfer chain (ETC), or are required for normal ETC function. Preliminary studies show that KO erythroblasts exhibit increased ETC activity consistent with de-repression of its component genes. Overall, our results demonstrate that miRNA-guided binding of Ago proteins to mRNAs is insufficient to produce mRNA repression, and that additional modifying variables determine whether physical interaction leads to repression. This finding is of general relevance to miRNA biology. Moreover, our studies provide a more comprehensive set of erythroblast miR-144/451 mRNA targets for further study, including components of the mitochondrial electron transfer chain. Figure 1 (A) Volcano plot showing HITS-CLIP peaks in KO vs. WT erythroblasts. Known targets of miR-144/451 (Ywhaz, Cab39 and Vapa) are indicated. (B) Distribution of canonical seed sequences of miRNAs deleted in KO mice. (C) Distribution of canonical seed sequences of miRNAs not deleted in KO mice. Figure 1. (A) Volcano plot showing HITS-CLIP peaks in KO vs. WT erythroblasts. Known targets of miR-144/451 (Ywhaz, Cab39 and Vapa) are indicated. (B) Distribution of canonical seed sequences of miRNAs deleted in KO mice. (C) Distribution of canonical seed sequences of miRNAs not deleted in KO mice. Figure 2 Alteration in gene expression between KO vs WT erythroblasts at the mature mRNA level (X-axis) and the RNA stability level (Y-axis). Known targets of miR-144/451 (Ywhaz, Cab39 and Vapa) are indicated. mRNAs with increased stability and abundance are indicated as black dots within a red dotted oval. Figure 2. Alteration in gene expression between KO vs WT erythroblasts at the mature mRNA level (X-axis) and the RNA stability level (Y-axis). Known targets of miR-144/451 (Ywhaz, Cab39 and Vapa) are indicated. mRNAs with increased stability and abundance are indicated as black dots within a red dotted oval. Disclosures No relevant conflicts of interest to declare.

Description: While hematopoietic stem cells (HSCs) can sustain the production of all types of mature blood cells throughout the life, there also exists HSC-independent hematopoiesis, which partially supports embryonic hematopoiesis and generation of specific types of adult hematopoietic cells (e.g., macrophages). Examples of the HSC-independent hematopoiesis include (i) the primitive wave of hematopoiesis that produces unipotent progenitors for erythrocytes, megakaryocytes or macrophages, and (ii) the "pro-definitive" hematopoiesis that produces multipotent erythro-myeloid progenitors (EMPs). Given that HSCs and HSC-independent progenitors are both derived from endothelial cells in distinct or overlapping hematopoietic sites, tracing their developmental origins and clarifying the regulatory mechanism will enhance our understanding of the profound difference between them and may improve in vitro generation of HSCs. Human HSCs have been refined based on the expression of CD49f (ITGA6). In combination with other HSC markers (CD34+CD38-CD45RA-CD43+CD90+), high expression of CD49f identifies long-term multilineage engrafting HSCs, whereas the cells with low CD49f represent a subtype of hematopoietic progenitor cells (HPCs) that possess transient engrafting activity. Meanwhile, CD49f has also been shown to be heterogeneously expressed in hemogenic endothelial cells (HECs), which give rise to both HSCs and EMPs via endothelial-to-hematopoietic transition (EHT). Thus, determining the changes (i.e., persistence, gain or loss) of CD49f expression during EHT is a key step in tracing the origins of HSCs and HSC-independent HPCs. In this study, using an in vitro system of HSC differentiation from human embryonic stem cells (hESCs), we observed that, while CD49f is highly expressed in all hESCs, only a portion of HECs express CD49f. Importantly, live cell imaging analysis revealed that CD49f expression persists during EHT, which is accompanied by initiating CD43 expression. To test whether the differential CD49f expression is associated with HSC versus HPC functions, we sorted the CD49fhigh and CD49flow cells and performed colony forming assay and gene expression profiling. The results showed that the CD49fhigh cells have multilineage potential, whereas the CD49flow cells lack lymphoid potential but show a strong erythroid preference. Gene expression analysis confirmed that the CD49fhigh and CD49flow cells represent HSCs and erythroid-biased HPCs, respectively, and that the Wnt and Notch signaling pathways may play a role in their functions. Collectively, these observations suggest that the CD49fhigh and the CD49flow cells are concurrently derived from the CD49fhigh and CD49flow HECs, thus modeling the in vivo generation of HSCs and HSC-independent HPCs. Based on the in vitro observations, we proposed that CD49f in vivo may also specify the distinct HSPCs emerged at different developmental stages/sites. To test this hypothesis, we isolated mouse primitive HPCs, EMPs and definitive HSCs, as well as their parental HECs, from yolk sac, embryo, and aorta-gonad-mesonephros (AGM) of different embryonic stages and determined their CD49f expression. The results showed that the primitive erythroid progenitors have lowest, whereas the definitive AGM HSCs have highest, CD49f levels; this trend was also observed in the related HECs isolated from various stages/sites. Thus, it is likely that the embryonic hematopoiesis is recapitulated, at least partially, by the in vitro system in terms of the sequential emergence of HSPCs ranging from unipotent erythroid progenitors to multipotent definitive HSCs, and this may also underlie the situation that EMPs and HSCs can be produced at the same stage/site but independently from different HECs. In summary, using the in vitro HSC differentiation system, we found that the differential expression of CD49f discriminates HSCs and HSC-independent progenitors, which are concurrently emerged from HECs. The persistent CD49f expression during EHT suggests that the fates of HSCs and HSC-independent HPCs are pre-defined in their parental HECs. Combining our in vivo data, the differential expression of CD49f also provide a possible regulatory mechanism for the multi-wave hematopoiesis. Further exploring the function and mechanism of CD49f in these regulations should be important for fully understanding the precisely regulated HSC generation and activities. Disclosures No relevant conflicts of interest to declare.

Description: MicroRNAs (miRs), a class of 19- to 23-nucleotide non-coding RNA molecules, are involved in tumorigenesis by regulating tumor cells and microenvironment. Our study revealed serum miR21 expression in a large cohort of B-cell lymphoma patients and the biological function of miR21 both in vitro and in vivo. Comparing with healthy volunteers, serum miR21 was significantly increased in patients with B-cell lymphoma (Figure 1A). As revealed by immunohistochemistry in 50 tumor samples of DLBCL, CD31-positive microvessels were more frequently observed in high miR21 group than in low miR21 group (Figure 1B). High miR21 expression patients displayed more peripheral blood Treg cells than low miR21 expression patients, instead of natural killer (NK) cells (Figure 1C). We futher study the biological function of miR21, B-lymphoma cell SU-DHL-4 were transfected with miR21 mimics and treated with chemotherapeutic agents. Under the monoculture condition, as compared to the control mimics, ectopic expression of miR21 significantly diminished the cytotoxic effect of doxorubicin and cisplatin, but sensitized lymphoma cells to ABT-199. Under the direct co-culture system, mimicking lymphoma microenvironment, miR21 overexpression resulted in lymphoma cell resistance to chemotherapeutic agents, but sensitivity to ABT-199 in the direct co-culture system. ABT-199 remarkably downregulated miR21 expression in both the monoculture system and the direct co-culture system, irrespective to Bcl-2 expression. To clarify the underlying mechanism of miR21-mediated sensitization of ABT-199 on B-cell lymphoma, we studied the effect of miR21 on HUVEC sorted from the direct co-culture system. Co-culturing with miR21-overexpressing lymphoma cells significantly stimulated HUVEC growth, which was retarded by ABT-199 (Figure 2A). As detected by ELISA, VEGFA was increased by miR21 transfection and reduced by ABT-199 in both control siRNA-transfected HUVEC and Bcl-2 siRNA-transfected HUVEC (Figure 2B). Accordingly, similar changes of tube formation and endothelial cell migration towards lymphoma cells were present (Figure 2C). We hereafter studied the effect of miR21 on Treg cells sorted from the direct co-culture system. Consistent with change of VEGFA, co-culturing with miR21-overexpressing lymphoma cells significantly increased VEGFR2 expression on Treg cells, which were decreased by ABT-199 (Figure 2D). ABT-199-induced downregulation of VEGFA/VEGFR2 signaling was associated with Treg cell growth inhibition, resulting in reduction of immunosuppressive cytokine TGF-¦Â and molecule IL-2 (Figure 2E and 2F). In the direct co-culture system, miR21 overexpression induced ICOS expression on Treg cells and ICOSL expression on HUVEC and, both of which were inhibited by ABT-199 (Figure 3A). To confirm the role of ICOS/ICOSL axis, antibody against ICOS was added to the direct co-culture system. Pharmacological inhibition of ICOS/ICOSL interaction significantly abrogated the sensitivity of miR21-overexpressing cells to ABT-199, as well as HUVEC and Treg cell growth (Figure 3B). Blockade of ICOS/ICOSL also interfered the action of ABT-199 on VEGFA/VEGFR2 signaling between Treg cells and endothelial cells (Figure 3C). Murine xenograft model was established with subcutaneous injection of B-lymphoma cells, ABT-199 particularly retarded the growth of miR21-overexpressing tumors, consistent with the inhibition of ICOS/ICOSL axis, VEGFA/VEGFR2 signaling, tumor angiogenesis and Treg cell growth. Collectively, these data demonstrated that miR21 plays an oncogenic role in B-cell lymphoma by modulating tumor microenvironment and supported clinical rationale for using miR21 as a biomarker to select chemoresistant B-lymphoma patients who may benefit from treatments containing ABT-199. Disclosures No relevant conflicts of interest to declare.

Description: The microRNA (miRNA) locus miR-144/451 is abundantly expressed in erythrocyte precursors, facilitating their terminal maturation and protecting against oxidant stress. However, the full repertoire of erythroid miR-144/451 target messenger RNAs (mRNAs) and associated cellular pathways is unknown. In general, the numbers of mRNAs predicted to be targeted by an miRNA vary greatly from hundreds to thousands, and are dependent on experimental approaches. To comprehensively and accurately identify erythroid miR-144/451 target mRNAs, we compared gene knockout and wild-type fetal liver erythroblasts by RNA sequencing, quantitative proteomics, and RNA immunoprecipitation of Argonaute (Ago), a component of the RNA-induced silencing complex that binds miRNAs complexed to their target mRNAs. Argonaute bound ∼1400 erythroblast mRNAs in a miR-144/451–dependent manner, accounting for one-third of all Ago-bound mRNAs. However, only ∼100 mRNAs were stabilized after miR-144/451 loss. Thus, miR-144 and miR-451 deregulate