ALBERT

Description: Diamond Blackfan anemia (DBA) is a rare inherited bone marrow failure syndrome characterized by red blood cell hypoplasia, congenital anomalies and cancer predisposition. In addition, short stature and poor skeletal growth are found in a subset of DBA patients, suggesting similar developmental abnormalities in erythropoiesis and osteogenesis in that subset. Furthermore it has been shown recently that osteoblasts secrete erythropoietin, linking the marrow niche to the modulation of erythropoiesis. DBA has been shown in the majority of cases to result from haploinsufficiency of large or small ribosomal subunit proteins. The p53 pathway, known to be activated by abortive ribosome assembly, contributes to the erythroid failure of DBA. We studied two DBA genotypes in vitro using murine embryonic stem (ES) cell lines harboring gene trap mutations in ribosomal proteins RPS19 and RPL5, respectively. Both mutants had decreased embryoid body (EB) formation, decreased definitive erythroid colony formation and similar p53-dependent primitive erythroid differentiation defects (see Figure A). Cell cycle analyses were normal in the Rps19 mutant ES cells, but there was a significant G2/M arrest in the Rpl5 mutant ES cells, which was unaffected by p53 knockdown. In addition, the Rpl5 mutant cells had a more pronounced growth defect in culture compared to the Rps19 mutant cells (Figure B). ES cells were differentiated, in vitro, to osteoblasts using established culture conditions, and confirmed both by morphology and molecular characterization (e.g. RUNX2 and Osteopontin). Following 14 days of osteogenic differentiation, bone mineralization was confirmed via Alizarin Red staining. A marked reduction in Alizarin Red staining was seen in the Rpl5 mutant cells while there was only a slight diminution of staining in the Rps19 mutant ES cultures (see Figure C). Therefore the erythroid differentiation defect appears similar in both the Rps19 and Rpl5 mutant ES cells. However the Rpl5 mutant appears to have a more severe phenotype at the ES stage, as evidenced by a pronounced p53-independent G2/M arrest and slower growth rate and subsequently during osteogenic differentiation. These data suggest an explanation for the more severe non-erythroid phenotype seen in a subset of DBA patients. Disclosures: No relevant conflicts of interest to declare.

Description: Pomalidomide, a second-generation immunomodulatory drug, is a fetal hemoglobin (HbF) inducing agent with potential implications for the treatment of β-hemoglobinopathies such as sickle cell disease (SCD). However, its mechanism of action remains unknown. Through an in-depth characterization of human erythropoiesis and globin gene regulatory networks, we now provide evidence that pomalidomide alters transcription networks involved in erythropoiesis and globin switching, thereby leading to a partial reprogramming of adult hematopoietic progenitors toward fetal-like erythropoiesis. Adult peripheral blood CD34+ cells from normal individuals were differentiated toward the red cell lineage using an adapted 3-phase culture system. At day 14 of culture, we observed a reciprocal globin gene switch at the mRNA and protein levels. These results were confirmed by high performance liquid chromatography of hemolysates (HbF/(HbF+HbA): 31.7 ± 1.4% vs. 6.5 ± 0.7% pomalidomide and vehicle, respectively). Next, we studied erythroid differentiation using flow cytometric analyses of the cell surface markers interleukin-3R (IL-3R), glycophorin A (GPA), CD34 and CD36 for early erythroid precursors (BFU-E and CFU-E) as well as GPA, α4-integrin and band3 for terminal erythroid differentiation. While there were no changes in terminal erythroblast maturation, an accumulation of BFU-E in pomalidomide-treated cultures at days 2 and 4 of differentiation was seen, indicating a delay at the BFU-E to CFU-E transition, and also, that pomalidomide exerts its effect in the early-stages of erythropoiesis. Indeed, treatment with pomalidomide during the phase of the culture system that generates erythroid progenitors led to significantly more γ-globin expression than treatment during the phase which proerythroblasts undergo terminal erythroid differentiation. At the molecular level, pomalidomide was found to rapidly and robustly decrease Ikaros (IKZF1) expression exclusively by post-translational targeting to the proteasome. Moreover, pomalidomide selectively reduced the expression of components of key globin regulatory pathways including BCL11A, SOX6, KLF1, GATA1 and LSD1 while not affecting others (e.g. CoREST, GATA2, GFI1B, and HDAC1). Pomalidomide had a transient effect on GATA1 and KLF1 expression. While shRNA knockdown of Ikaros using two different lentiviral constructs delayed erythroid differentiation, it failed to appreciably stimulate HbF production or alter BCL11A expression. These results suggest that the loss of Ikaros alone is insufficient to recapitulate the phenotype observed in pomalidomide-treated conditions. We next compared the expression levels of proteins involved in globin gene regulation among untreated peripheral blood, pomalidomide-treated peripheral blood and untreated cord blood-derived erythroid cells. We found striking similarities between cord blood and pomalidomide-treated adult cells at day 4 of differentiation. Indeed, BCL11A, KLF1, SOX6, LSD1 and GATA1 showed decreased expression levels both in cord blood and pomalidomide-treated adult peripheral blood, while the levels of CoREST, HDAC1 and GATA2 remained unchanged indicating that pomalidomide partially reprograms adult erythroid cells to a fetal-like state. Taken together, our results show that the mechanism underlying reactivation of HbF by pomalidomide involves Ikaros-independent reprogramming of adult erythroid progenitors. Finally, we found that this mechanism is conserved in SCD-derived CD34+ cells. Our work has broad implications for globin switching, as we provide direct evidence that Ikaros does not play a major role in the repression of γ-globin during adult erythropoiesis, and further supports the previously held notion that globin chain production is determined prior to or at the level of CFU-E. Disclosures Allen: Celgene: Research Funding; Bristol Myers Squibb: Equity Ownership; Onconova: Membership on an entity's Board of Directors or advisory committees; Alexion: Membership on an entity's Board of Directors or advisory committees; Takeda: Membership on an entity's Board of Directors or advisory committees.

Description: The mechanisms underlying the development of erythropoietin (EPO)-refractory anemia in the setting of chronic inflammatory states are largely unknown. Elevated levels of the classical inflammatory mediators decrease red cell output. However, pathologic concentrations of many of these molecules do not persist beyond the acute phase, indicating that specific mediators are likely to play a role in the anemia associated with chronic inflammation. High mobility group box protein 1 (HMGB1) is a potent alarmin able to induce tissue injury during the acute and chronic phases of inflammation, and recently, shown to contribute to anemia in a murine model of sepsis. Here, we show that HMGB1 directly inhibits erythropoiesis by modulating EPO signal transduction in human erythroid cells through a newly identified HMGB1 receptor, which is surprisingly the erythropoietin receptor (EPOR). Surface plasmon resonance (SPR) reveals that HMGB1 binds the extracellular domain of EPOR (Kd = 130nM) with an affinity comparable to that of EPO. Cysteine residues contained within the A- and B-box domains of HMGB1 that have previously been shown to mediate HMGB1-receptor interactions are also responsible for the EPOR-HMGB1 interaction since a mutant form of HMGB1 lacking these cysteine residues (i.e. 3S HMGB1) fails to bind the EPOR. Cell-based assays suggest that the direct binding of HMGB1 to the EPOR and the subsequent degradation of EPOR accounts for altered EPO signaling by HMGB1. Biologically, HMGB1 reduces the phosphorylation of intracellular EPO effectors including JAK2 (2-fold reduction), STAT5 (4-fold), and ERK1/2 (4-fold). Decreased effector phosphorylation is not due to the increased activity of SHP1/2 phosphatases further implicating inhibition at the receptor level. Loss of EPO signaling due to HMGB1 binding results in decreased erythroid proliferation of differentiated CD34+ cells at the EPO-dependent stages of erythropoiesis: Day 14: 1.03x108 ± 4.67x107 cells/mL vs 1.87x106 ± 9.70x105 cells/mL, vehicle vs HMGB1, respectively. In addition, HMGB1 decreases the numbers of colony forming unit-erythroid (CFU-E) progenitors by 60%, and these progenitors fail to undergo terminal erythroid differentiation with a block at the basophilic erythroblast stage and apoptosis of late-stage erythroblasts as determined by flow cytometric analysis of annexin V staining. To understand the consequences of HMGB1-EPOR interactions on the EPO-induced transcriptome, RNA-sequencing was performed on purified human CFU-E dosed with HMGB1 and EPO. HMGB1 reduces the expression of known EPO target genes (ERFE, CISH, EGR1), and concomitantly, upregulates a number of unique transcripts (ETS2, VMP1, NFKBIZ) suggesting that HMGB1-EPOR interactions may alter receptor conformation in manner that differentially activates the EPOR and consequently, gene expression. Finally, in a mouse model of sepsis survival, bone marrow-derived erythroid precursor cells contain diminished phosphorylated STAT5 levels at a time when elevated HMGB1 plasma concentrations are observed, thereby demonstrating that the loss of EPO signal transduction also occurs in vivo. Taken together, our work identifies HMGB1 as a novel inhibitor of EPO signaling through its interaction with the EPOR, and strongly implicates HMGB1 as a previously undiscovered effector of EPO-refractory anemia associated with chronic inflammation. Disclosures No relevant conflicts of interest to declare.

Description: Background: Diamond Blackfan anemia (DBA) is a congenital anemia characterized by failure of adequate erythrocyte expansion from hematopoietic precursors. The genetic basis of DBA is largely established, with mutation or deletion of at least 19 structural ribosomal protein (RP) genes, a RP chaperone (TSR2), or a pivotal erythroid transcription factor (GATA1) identifiable in most DBA cases. However, the marked clinical variability in DBA-including varying ages of presentation, severity of anemia, responsiveness to corticosteroids, and sporadic hematologic remissions-remains unexplained by genotype and may be modulated by epigenetic factors. Further understanding of this variability is of potential therapeutic relevance for biomarkers of steroid response and remission as well as in application to novel treatment approaches. Aim: We characterized genome-wide methylation and chromatin accessibility of erythroid progenitors from normal controls and DBA patients during erythroid commitment in order to identify the epigenetic features associated with erythroid failure in DBA, steroid response, and remission. Methods: We expanded and sorted CD71+/CD235- (transferrin receptor/glycophorin A) and CD71+/CD235+ erythroid cell fractions from DBA patients and controls after isolation of primary circulating CD34+ cells from peripheral blood (O'Brien et al, Blood 129(23):3111, 2017). We performed DNA methylation analysis using Illumina Epic arrays in 9 control and 22 DBA subjects (11 transfusion-dependent, 6 steroid responsive, 5 remission), characterizing differentially methylated probes and regions among groups. To define broad chromatin domains, we identified A/B chromatin compartments (representing open/closed chromatin) using long-range correlations in methylation data as previously described (Fortin et al. Genome Biol 16:180, 2015). To identify discrete local changes in chromatin accessibility, we performed ATAC-sequencing in 9 controls and 17 DBA patients (10 transfusion, 6 steroid, 1 remission). Results: Global DNA methylation falls during erythroid commitment, with 258,618 differentially hypermethylated CpG sites in normal control GlyA- cells compared to their more differentiated GlyA+ counterparts. This pattern is exaggerated in DBA samples, with 297,926 sites hypermethylated in GlyA- cells. We identified 17,392 CpC sites that distinguish GlyA- DBA progenitors from normal progenitors (16,953 hyper- and 439 hypomethylated). We identified 1,749 differentially methylated sites in comparison of transfusion dependent and remission DBA, as well as 79 sites different between transfusion dependent and steroid responsive DBA. Using genome-wide methylation data, we evaluated A/B compartment organization among these groups to identify large regions of open and closed chromatin during normal and DBA early erythroid differentiation. We observe significant shifts in A/B compartments in normal cells concurrent with the acquisition of GlyA surface expression. At genome scale, transfusion dependent and steroid resistant DBA samples are generally similar to each other, with thousands of regions where A/B identity are closely matched in DBA, but diametrically opposed to the configuration in stage-matched normal controls. Intriguingly, remission samples generally matched A/B compartments of other DBA samples in GlyA- fractions but GlyA+ compartments more closely resemble those of the controls, indicating that normalization of chromatin structural maturation accompanies hematologic remission. We generated a uniform set of 8,877 enriched ATAC-seq peaks on autosomes for differential chromatin accessibility analysis. As with methylation data, a large proportion (25%; 1085 up and 1114 down, B-H adj. P 〈 0.1) showed differential accessibility in normal control GlyA- vs GlyA+ cells. Steroid-responsive cases showed additional regions of differential accessibility during early maturation, with 31% of regions (1515 up and 1248 down) differentially accessible. Among transfusion dependent DBA patients, this count was much higher, with over half of peak regions (52%, 2400 up and 2216 down, B-H adj. P 〈 0.1) showing differential accessibility. Conclusion: Epigenetic maturation is broadly altered in DBA erythroid progenitors compared to stage matched normal controls, with specific changes identifiable in patients responding to steroids and in remission. Disclosures No relevant conflicts of interest to declare.

Description: Background: RPL35A, a gene encoding a large ribosomal subunit protein located at the telomeric end of chromosome 3q (3q29-qter) is essential for rRNA processing, ribosomal biogenesis, cell proliferation, and apoptosis, and accounts for a subset of patients with Diamond Blackfan anemia (DBA). Reported pathogenic RPL35A mutations include single-nucleotide variants (SNVs), small insertion/deletions (indels), and large contiguous gene deletions associated with 3q29 microdeletion syndrome. 3q29 deletion syndrome is an overlapping syndrome that consists of developmental and intellectual disability with or without dysmorphic features, and other congenital anomalies but no anemia or cytopenia. The clinical phenotype and disease severity of patients with RPL35A-related DBA may be influenced by other genes deleted within 3q29 and be different in patients with large deletions compared to those with SNVs or small indels. Objectives: To determine whether DBA patients with large deletions of the 3q29 region have a more severe disease phenotype than those with SNVs or small indels in RPL35A, and whether other genes deleted within the 3q29 region might contribute to some of the features. Methods and Results: We identified 40 patients in a multi-institutional, international collaborative study of patients with DBA with RPL35A haploinsufficiency: 21 had deletion of RPL35A as part of 3q29 contiguous gene deletion, ranging in size from 0.012 Mb to 11 Mb in 16 patients; the extent of the deletion beyond RPL35A in either direction was unknown in 5 patients. Nineteen patients had SNVs or small indels (7 missense, 1 nonsense, 3 splice site, 6 indels and 2 unclear pathogenicity). Thirty-nine of 40 patients had severe anemia, 32 had neutropenia at some time and 3 had thrombocytopenia. Compared to the patients with SNVs or small indels, a significantly higher proportion of patients with RPL35A haploinsufficiency due to 3q29 deletion had steroid-resistant anemia (17 vs 7; p=0.009), severe chronic or intermittent neutropenia requiring treatment with G-CSF (7 vs 0; p=0.009), and humoral and/or cellular immunodeficiency (7 vs 1; p=0.046) diagnosed in some patients due to recurrent infections requiring hospitalizations (10 vs 2; p=0.03). Learning difficulties (12 vs 2; p=0.003), craniofacial abnormalities (11 vs 3; p=0.02), skeletal and limb defects (9 vs 2; p=0.03) or multiple physical anomalies (≥3) were also more frequent in patients with large deletions than in those with SNVs or indels (11 vs 3; p=0.02). Microcephaly (28%), short stature (33%), cardiac defects (28%) and/or urogenital abnormalities (23%) were equally distributed. The potential genes of interest near RPL35A that may be associated with immune dysregulation and/or neutropenia are RNF168, TFRC, PAK2, PIGZ, DLG1 and LMLN. One or more of these genes were deleted in at least 6 of 7 patients with neutropenia or immunodeficiency. Eight of 9 patients with malformations involving extremities, skeleton and ribs had deletions of TCTEX1D2 which is associated with rib/thoracic dysplasia and polydactyly. The genes of interest deleted in patients with developmental delay and learning disabilities included PAK2 and DLG1 in 9 patients, as well as RNF168, PPP1R2, TNK2 and q29 KIAA0226 in 8 of 12 patients in whom the extent of the deletion was known. Summary and Conclusion: Patients with DBA due to RPL35A contiguous gene deletions are clearly different from those with SNVs or small indels and have increased frequency of steroid-resistant transfusion-dependent anemia, severe neutropenia, immunodeficiency, learning/developmental delay, and craniofacial/skeletal/limb anomalies. Distinction of this subtype of DBA with RPL35A haploinsufficiency due to 3q29 contiguous gene deletion is relevant to their management, and evaluations should include a work-up for immunodeficiency. Further studies are needed to determine whether the complex phenotypes and severe disease manifestations in these patients are solely due to RPL35A haploinsufficiency or to the potential effect of other genes deleted in the 3q29 region. Genotype-phenotype characterization and comparison of DBA patients with 3q29 deletion with those due to deletions in other ribosomal protein genes may determine similarities or differences in disease phenotypes related to large gene deletions versus the influence of multigenic contiguous deletions. Supported by: AZV 16-32105A Disclosures Kattamis: Novartis: Consultancy, Honoraria; CELGENE: Consultancy, Honoraria; ApoPharma: Honoraria; Vifor Pharma: Consultancy. Niemeyer:Celgene: Consultancy, Membership on an entity's Board of Directors or advisory committees.

Description: Diamond Blackfan anemia (DBA) is a rare, inherited bone marrow failure syndrome characterized by anemia, congenital anomalies and a predisposition to cancer. The patients usually present during infancy or early childhood, but can also present in adulthood. In the majority of cases DBA is due to a mutation in a small or large ribosomal protein (RP) subunit leading to RP haploinsufficiency. The only treatments for the anemia of DBA are red cell transfusions (accompanied by iron chelation), oral corticosteroid therapy or stem cell transplantation. Pospisilova et al. (Haematologica 2007; 92(5):e66-67) reported one complete and two partial erythroid responses after the use of the branched chain amino acid L-leucine in 6 select patients. In skeletal muscle, leucine supplementation can upregulate components of the protein synthetic machinery, including ribosomal proteins, promoting protein translation. Mouse and fish models of DBA respond with amelioration of anemia to L-leucine. We therefore proposed to study the effect of L-leucine on transfusion dependence and growth in subjects with DBA. Methods: The primary objectives were to determine the feasibility of administering L-leucine in subjects with DBA who are red cell transfusion-dependent and to determine the efficacy of L-leucine to produce a hematologic and growth response. The secondary objective was to determine the safety profile of L-leucine. Twelve study sites were involved in this multi-center, Phase I/II study with an anticipated accrual of 50 subjects. A dose of 700 mg/M2 orally three times per day for 9 months was used. Inclusion criteria included age 〉 2 years, the diagnosis of DBA and transfusion dependence with adequate kidney and liver function. Response was evaluated at 9 months with Complete Response (CR) defined as no further transfusions required and Hb 〉9; Partial Response (PR): Hb 〈 9 gm/dL with an increase in reticulocyte count and transfusion interval; and No Response (NR): no change in transfusion needs, Hb or reticulocyte count . Growth percentiles were evaluated at baseline and at completion of 9 months of treatment and the growth velocity change was calculated. Results: The study opened July 2014 and closed February 2017; 55 subjects were consented; 12 were screen failures; 43 subjects were evaluable. There were 21 males; the median age was 9 years 1 month (2 years 5 months - 46 years 1 month). There were no untoward side effects experienced by any subject that were attributable to the L-leucine. Two subjects had an erythroid CR and 1 subject had a PR. The CRs occurred at 1 month and 3 months after start of L-leucine. The subject with PR had an elevated reticulocyte count but was not able to maintain a Hb 〉9 gm/dL without a transfusion and thus was not transfusion independent. Of the 30 subjects with growth potential who received L-leucine 10 experienced a positive growth velocity change at 9 months of therapy compared to baseline. At a median age of 7.5 years, the mean pre-leucine height percentile was 27 +/- 17.9 and the post-leucine height percentile was 35 +/- 19.9 (p