Suppression of Medullar Erythropoiesis in Response to Bacterial Lipopolysaccharides (LPS) Involves Two Distinct TLR4-Dependent Mechanisms with Contrasted Requirements for G-CSF Receptors

Erythropoiesis is a highly controlled process partly regulated in erythroblastic islands by the central erythroblastic island macrophage (EI MΦ) which provides iron, growth factors and mediates enucleation of the maturing erythroblasts. As macrophages are key effectors of inflammation, we investigated the effect of bacterial LPS in vivo on erythropoiesis and EI MΦ defined as CD11b⁺ F4/80⁺ VCAM1⁺ CD169⁺Ly6G⁺ in mice. C57BL/6 mice were injected i.p. with 2.5 mg/kg/day LPS from E. coli for 2 days and the effect on medullar erythropoiesis examined at various time-points after this. LPS administration caused a marked whitening of the bone marrow (BM) with decreased numbers of basophilic (9-fold), polychromatic (3.7-fold), orthochromatic erythroblasts (2.2-fold) and reticulocytes (2.5-fold) 48hrs after LPS challenge. Those remained significantly reduced up to 6 days post-LPS. Likewise, EI MΦ were suppressed in the BM 24-48hrs after LPS challenge and remained significantly reduced 6 days post-LPS. This loss of medullar erythropoiesis was compensated by increased number of EI MΦ (13.6-fold), pro-erythroblasts (1.5-fold), polychromatic (1.9-fold), orthochromatic (3.2-fold) and reticulocytes (2.3-fold) in the spleen. As this phenotype resembled suppression of medullar erythropoiesis following G-CSF treatment, we examined whether the mechanism could be indirect via endogenous G-CSF release. LPS induced a transient 80-fold increase in G-CSF concentration in the blood from 123pg/mL (n=6) to 10ng/ml (n=6) 2 days post-LPS. LPS was administered to TLR4 KO and G-CSF receptor (GCSFR) KO mice. These LPS-mediated responses were abrogated in TLR4 KO mice demonstrating that erythropoiesis suppression in response to LPS is fully TLR4-dependant. However responses in GCSFR KO mice were more contrasted. EI MΦ numbers did not change in GCSFR KO mice in response to LPS demonstrating that suppression of EI MΦ in response to LPS is an indirect effect of endogenous G-CSF release. In contrast, medullar erythropoiesis was still suppressed in GCSFR KO mice with significantly reduced numbers of basophilic, polychromatic and orthochromatic erythroblasts demonstrating that medullar erythroblast suppression 1) persists despite the presence of EI MΦ, and 2) is not G-CSF-dependent. Unexpectedly, the BM from GCSFR KO mice treated with LPS was not whitened with high numbers of reticulocytes/erythrocytes. To further understand how BM erythrocytes could be increased whilst erythropoiesis is suppressed in LPS-treated GCSFR KO mice, we measured vascular leakage by injecting Evans Blue i.v. Blood plasma volume in the BM of LPS-treated GCSFR KO mice was 2.9-fold higher compared to LPS-treated wild-type mice and untreated WT and KO mice (6.0±2.0 µL vs 2.1±0.9 µL blood plasma/femur, p=0.005) suggesting that GCSFR-mediated signaling is necessary to maintain the integrity of the BM vasculature in response to LPS. In conclusion LPS-mediated medullar erythropoiesis suppression involves at least two different TLR4-dependent mechanisms in regards to their requirement for GCSFR: 1) GCSFR-dependent suppression of EI MΦ, 2) GCSFR-independent and EI MΦ-independent suppression of maturing erythroblasts. We are currently investigating whether the latter mechanism involves hepcidin. Finally we also discovered that GCSFR-mediated signaling is necessary to maintain the BM vasculature integrity following LPS challenge.