Enucleation is the hallmark of erythropoiesis in mammals. Previously, we determined that yolk sac-derived primitive erythroblasts mature in the bloodstream and enucleate between E14.5–16.5 of mouse gestation (

Kingsley, et al.
), however, it is unclear by what mechanism or even where primitive erythroblasts enucleate. Definitive erythroblasts in the fetal liver and adult marrow enucleate by nuclear extrusion, generating reticulocytes and small, nucleated cells with a thin rim of cytoplasm referred to as “extruded nuclei”, that rapidly lose cell membrane phosphatidylserine asymmetry and are engulfed by macrophages. Since the cell membrane plays an important role in the biology of these cells, “extruded nucleus” is a misleading term. We propose that these cells be termed “pyrenocytes”, derived from the Greek “pyren” (the pit of a stone fruit) and “cyte” (cell), reflecting their highly condensed nucleus and high nuclear to cytoplasmic ratio. Careful examination of murine fetal blood by immunohistochemistry and multispectral imaging flow cytometry (Amnis ImageStream) revealed a transient population of εy-globin-positive pyrenocytes temporally coincident with the enucleation of primitive erythroblasts (E14.5–15.5). A high percentage (40%) of these circulating primitive pyrenocytes were annexin V-positive, suggesting that they, like their definitive counterparts, generate a signal that can facilitate engulfment by macrophage cells. At E14.5–15.5, the highest frequency of macrophage cells within the conceptus is found in the liver. Immunohistochemical studies with anti-εy-globin and F4/80 antibodies revealed that many primitive erythroblasts in the liver, but not the spleen, are in close proximity to macrophages. Furthermore, the frequency of nucleated primitive erythroblasts was higher in the liver than in the bloodstream at E15.5. These results, taken together, suggest that late-stage primitive erythroblasts do not passively flow through the liver but may, rather, interact with macrophage cells there. Surprisingly, primitive erythroblasts, but not co-circulating fetal definitive erythrocytes, can reconstitute erythroblast islands by attaching to fetal liver-derived macrophages in vitro. α4 integrin blocking antibodies, but not isotype control antibodies, reduce erythroblast island reconstitution, indicating that these primitive erythroid-macrophage interactions are mediated in part by α4 integrin. Finally, we found that unlike definitive erythroblasts, late-stage primitive erythroblasts fail to autonomously enucleate in vitro unless co-cultured with macrophage cells. We conclude that primitive erythroblasts in the mammalian fetus enucleate by nuclear extrusion as a semi-synchronous cohort and generate a transient population of pyrenocytes. Furthermore, our studies suggest that this process occurs in the fetal liver in association with macrophage cells. Continued investigation of the differences and similarities between primitive and definitive erythropoiesis will lead to an improved understanding of the terminal steps of erythroid maturation.

Author notes

Disclosure: No relevant conflicts of interest to declare.