Not children from a lesser god

During mammalian ontogenesis, erythroid cells are sequentially produced in distinct anatomic sites. This temporal site succession is paralleled by profound changes in morphologic and functional properties of the cells being generated.¹  In fact, embryonic red cells are large (approximately 200 μm) megaloblastic cells that mature in the yolk sac synchronously in the absence of erythropoietin (EPO)²  and retain, as in lower vertebrates, their nuclei after being released in the circulation. On the other hand, fetal macrocytic and adult normocytic (produced in liver and bone marrow, respectively) red cells are smaller (approximately 125 μm and approximately 80 μm, respectively) enucleated cells whose production is strictly EPO dependent.³  To emphasize such differences, mammalian embryonic and fetal/adult red cells have been designated “primitive” and “definitive” red cells, respectively, and it was even questioned whether the 2 cell types derive from the same stem cell pool. The erythroid system was, in fact, considered one of the best examples of “embryonic development” recapitulating “phylogenetic evolution,” an erudite expression that means: we do not have the faintest idea of what is going on.

The molecular mechanism controlling the formation of primitive and definitive red cells has been extensively investigated by Stuart Orkin's and Masi Yamamoto's labs. Both investigators have reported that loss of GATA-1 impairs formation of both primitive and definitive cells, while loss of GATA-2 affects primarily that of the definitive ones. Definitive red cell formation is, in fact, controlled in sequence (although partial overlap is not excluded) by GATA-2, which determines the amplification of the progenitor cell pool, and by GATA-1, which controls the apoptotic rate of the precursor cells.

The concept that primitive and definitive erythroid cells are completely different has been recently challenged by the observation that (1) mouse embryonic stem cells give rise in vitro to a common precursor capable of generating both cell types⁴ ; (2) megaloblasts that have undergone enucleation are present in the blood at a later stage of mouse development⁵ ; and (3) the formation of primitive red cells also requires both GATA-2 and GATA-1. In fact, Fujiwara and colleagues (page 583) report in this issue of Blood that GATA-1 and GATA-2 exert overlapping functions in primitive erythroid development. Therefore, the difference between the control of primitive and definitive erythropoiesis is that in the fetal/adult stage the functions of the 2 genes are spread over the numerous divisions necessary to go from progenitor to precursor cells, while, in the embryos, their functions are shared during the few divisions necessary to make the megaloblast. The main consequence of this difference is the extent of amplification allowed for primitive and definitive erythropoiesis. Fetal/adult erythropoiesis, because of the separation in a GATA-2– and GATA-1–dependent phase, has the high expansion potential required for an efficient response to stress, an expansion potential not needed by the embryos in which it could even be deleterious. It would be interesting to know whether, during the phylogenesis, the separation into 2 phases of the control of adult erythropoiesis has represented a selective advantage by providing to the species a better response to stress.