▪Severe sepsis is a leading cause of death and disability. Anemia in sepsis survivors affects close to 100% of patients after the third day of in-hospital stay, regardless of blood levels on admission. Circulating levels of Erythropoietin (Epo) are low; paradoxically, administration of recombinant Epo is ineffective, and related to increased morbidity. During sepsis, bone Marrow is hypoproliferative. While transfusions can improve outcome in the short term, its use increases the risk of infection and mortality without any sustained beneficial effect. The pathogenesis of anemia during sepsis is unclear. High mobility group box 1 (HMGB1), a cytokine that is a critical mediator of sepsis, is released into circulation a few days after sepsis onset, remaining increased for 8 weeks after severe sepsis. HMGB1 levels are increased for at least 8 weeks in murine models of sepsis survival.To induce severe sepsis, cecal ligation and puncture (CLP) was performed in BALB/c mice. Three days after CLP, mice developed persistent anemia, represented by a significant reduction in hematocrit (Sham=49.8±3.2 vs. CLP=29.7±6.7%; p≤0.001), hemoglobin (16.7±1.2 vs. 9.9±2.4mg/dL; p≤0.001), and red blood cells mass (10.2±0.7 vs. 5.4±1.7 x106/µL; p≤0.001). Anemia persisted for at least 25 days after CLP. In CLP survivors, reticulocyte counts were erratic, and insufficient to the degree and duration of anemia (8.2±0.8 vs. 6.6±2.1%; p=ns). Analysis of terminal erythroid differentiation using CD44 and Ter119 or CD44 and FSC as markers demonstrated a significant decrease in all erythroid progenitors, from proerythroblast to orthochromatic erythroblast. Concomitantly, mice surviving CLP developed splenomegaly. Splenic architecture was disrupted after CLP, with expansion of the red pulp, characteristic of stress erythropoiesis. Analysis of terminal erythroid differentiation demonstrated an increase in the quantity of erythroid progenitors.An anti-HMGB1 mAb (2G7) was administered after CLP. Strikingly, 2G7-treated septic mice were significantly protected from developing anemia, and had levels of hemoglobin and hematocrit similar to sham-operated mice. These results highlight a critical role for HMGB1 as key modulator of stress erythropoiesis in a murine model of sepsis survivors.To get further insight into the function of HMGB1 and translate our findings to the pathophysiology of human erythropoiesis, we used CD34+ cells derived from cord blood.Cord blood-derived CD34+ cells were incubated in MethoCult in the presence or not of HMGB1. HMGB1 induced a dose dependent decrease in CFU-E.In murine sepsis, there is a stepwise elevation of different redox forms of HMGB1, with an early increase in all-thiol (inflammatory), followed by a partially oxidized before a fully oxidized (with no known inflammatory activity) appears. At day 7, all-thiol HMGB1 reduced significantly the number of CFU-E, while the fully oxidized had no significant effect. At day 14, the number of BFU-E was reduced in the presence of HMGB1, and further decreased with all-thiol HMGB1.In conclusion, our findings suggest that CLP is a reproducible model to study anemia of sepsis. In mice surviving sepsis, stress erythropoiesis is consistently found. Administration of anti-HMGB1 monoclonal antibody reverses anemia of murine sepsis, demonstrating that HMGB1 can be a potential target in the anemia of sepsis survivors. Translating the findings to the human system, we found that HMGB1 impairs differentiation of CD34+ cells towards the BFU-E and CFU-E stages in colony formation assays, implying that HMGB1 might play a role early during differentiation. The redox status of HMGB1 is critical for its biological function, since its effects are not retrieved when HMGB1 is fully oxidized. Disclosures:No relevant conflicts of interest to declare.