Identification and characterization of human erythroblastic islands (EBIs) at single cell resolution Free

Nearly all red cell production takes place within erythroblastic islands (EBIs), a structure comprised of a central macrophage (CM) and ~10-50 adherent erythroid cells. Although essential for erythropoiesis, its function is unclear and characterization of the CM is lacking. A normal adult makes 2.4x10⁶ red cells per second; each cell contains 270x10⁶ hemoglobin (Hgb) molecules; each Hgb molecule contains 4 heme moieties. Thus, the erythron requires 2.6 quadrillion heme moieties per second and needs a containment strategy to avoid ROS and other heme toxicities. This realization led us to hypothesize that CM functions as a safe and efficient heme-iron recycling center, accepting heme from early erythroid precursors while providing the iron to co-adherent iron-avid late stage precursors, enhancing erythropoiesis.

As an initial test of our hypothesis, we modeled EBIs in vitro with CD34-derived erythroid precursors and CD14-derived EBI macrophage using published methods (Haematologica 103:395, 2018). CD34-derived erythroid precursors expanded 8-fold more (p<0.01) in the EBI culture than if cultured without EBI macrophages and EBI macrophages had 12-fold more heme (p<0.01) than macrophages cultured without erythroid precursors. The EBI macrophages upregulated HMOX1 (p<0.001) and ferroportin (SLC40A1, p<0.05), indicating that they were degrading heme and exporting the iron.

To better test our hypothesis, we developed methods to collect marrow from normal volunteers and enrich intact EBIs. We characterized the enriched EBIs using scRNAseq and identified three clusters of macrophages, abundant erythroid cells at all stages of maturation, and smaller numbers of cells in other hematopoietic lineages. Unlike mouse EBIs, human EBIs contain relatively few neutrophils. We noted that one macrophage cluster expressed the published markers VCAM1, CD163, and CD206 and considered this the likely CM cluster. Importantly, this cluster uniquely and highly expresses CD91 (LRP1; imports heme), HRG1 (SLC48A1; imports heme to cytoplasm), SPIC (heme-induced transcription factor), HMOX1 (metabolizes heme to iron), and ferroportin (SLC40A1; exports iron). Despite exposure to, and uptake of, large amounts of heme, the EBI CM does not express inflammatory cytokines (TNF, IL6, IL1β, & TGFβ). However, these genes are highly expressed in other marrow macrophages that are geographically distinct per UMAP. We next used logistic regression classifier algorithms to develop an unbiased 17-gene signature panel that specifically identifies the EBI CM. As a validation study, we screened surface marker expression and identified 5 surface markers that uniquely labeled this cluster. The five-marker flow panel labeled 0.1-1% of normal human marrow cells, consistent with the expected frequency of EBI macrophages in marrow. We sorted macrophages from marrow aspirates using the surface marker panel and showed that they highly express the 17-gene EBI CM signature, thus confirming the gene panel and flow panel identified the same macrophage. We then demonstrated that these macrophages expressed CD91, HRG1, SPIC, HMOX1, and ferroportin and enhanced erythroid expansion 10-fold in EBI cultures, similar to that observed with CD14-derived EBI macrophages, confirming that the EBI central macrophage that we flow-sorted functions to degrade excess heme and enhance erythropoiesis as hypothesized.

Our prior work demonstrated that in MDS-5q patient marrow both the mutant and normal erythroid precursors are abnormal and have alterations consistent with excess heme (Blood Adv 7:4848, 2023). We modeled ribosomal haploinsufficiency of MDS-5q with RPS14 shRNA expressed in the CD34-derived erythroid precursors during EBI cultures. The mutant precursors expanded poorly (only 2X vs 9X in control cultures, p<0.001) and the EBI macrophages contained 3-fold more heme (p<0.05) than those in control EBI cultures, suggesting the macrophages could not rapidly process excessive erythroblast-derived heme. If the CM dysfunctions and loses its nurse cell capacity, co-adherent MDS and residual normal erythroid cells would be damaged, perhaps explaining why ineffective erythropoiesis and anemia occurs early in MDS when many normal cells persist. We are using the 17-gene signature to reliably identify and study CM in cryopreserved clinically-obtained marrow samples from MDS patients and directly test this. The surface marker panel will let us isolate EBI macrophages for confirmatory functional studies.