Erythrocytes in the trunk of zebrafish embryos

In our previous study (Jin et al., 2009), we presented evidence indicating that, in zebrafish embryos, hematopoietic stem cells (HSCs) in the ventral wall of the dorsal aorta (VDA), where HSCs originate, and in the caudal hematopoietic tissue (CHT)/posterior blood island (PBI), where HSCs later home, possess distinct differentiation outputs. Using expression analysis and fate mapping, we showed that HSCs only give rise to myeloid cells and not to erythroid cells in the VDA, whereas HSCs differentiate into erythroid cells upon migrating to the CHT/PBI (Jin et al., 2009). By analyzing the differentiation of HSCs in embryos in which HSCs were trapped in the VDA, we further showed that the erythropoietic ability of developing HSCs is dependent on the CHT/PBI (Jin et al., 2009).

In their Correspondence (Falenta and Rodaway, 2011), Falenta and Rodaway have provided additional evidence to support their earlier finding (Zhang and Rodaway, 2007) and questioned the non-erythropoietic nature of the VDA because they have detected αe1- and βe1-globin-expressing cells (using αe1- and βe1-globin probes) and cells positive for hemoglobin peroxidase activity [using diaminofluorene (DAF) staining] in the trunk of zebrafish embryos. As suggested by Falenta and Rodaway, we stained 4 days post-fertilization (dpf) embryos for a longer time period of 24 hours (see Appendix S1 in the supplementary material), rather than 2 hours as in the original paper, by whole-mount in situ hybridization (WISH) to detect αe1- and βe1-globin. Whereas staining for 2 hours in our original study failed to detect αe1- and βe1-globin expression in the trunk, staining for 24 hours allowed the detection of αe1- and βe1-globin expression in the trunk of some 4 dpf embryos. However, these αe1- or βe1-globin-positive cells in the trunk were rare in comparison with the numbers of αe1- and βe1-globin-positive cells that were present in the CHT/PBI and a significant portion of 4 dpf embryos that were stained for 24 hours with αe1- or βe1-globin probes lacked staining in the trunk (see Table S1 in the supplementary material). Furthermore, `per cell' staining intensity for αe1- and βe1-globin was much weaker in the trunk compared with that in CHT/PBI (see Fig. S1A,B in the supplementary material). By contrast, all the embryos stained with lyz or mpx, two myeloid markers, for a short period of 2 hours had detectable signals in the trunk (see Appendix S1 and Table S1 in the supplementary material) and `per cell' staining intensity was comparable in the trunk and CHT/PBI (see Fig. S1C,D in the supplementary material). Because of the finding that staining intensity for αe1- and βe1-globin in the trunk was weaker than that observed in the CHT/PBI, whereas staining intensity for lyz and mpx was comparable in these two sites, we think that cells in the trunk might express genuinely low levels of αe1- and βe1-globin. Additionally, in these embryos stained for 24 hours, primitive erythrocytes were also weakly stained by αe1-globin and βe1-globin probes (see Fig. S1A,B in the supplementary material), which raises the possibility that the erythroid cells reported by Falenta and Rodaway to be present in the trunk are primitive erythrocytes. It is also possible that these rare erythrocytes come from the CHT/PBI and stop randomly in the blood vessel during the fixation step of WISH. Therefore, the nature and origin of these cells weakly positive for αe1- and βe1-globin in the trunk are not clearly addressed. Specifically, it is unknown whether these cells are derived from the differentiation of HSCs in the VDA, from primitive erythrocytes that carry residual globin transcripts, or from other sources. A lineage-tracing study would help to resolve the origin of these trunk erythrocytes in the future. Irrespective of the origin of trunk erythrocytes, embryos at 4 dpf, on average, contained many fewer αe1- and βe1-globin⁺ clusters in the trunk than lyz⁺ or mpx⁺ clusters (see Table S1 in the supplementary material). Moreover, Falenta and Rodaway mentioned that there was no detectable hemoglobin expression between the DA and the posterior cardinal vein (PCV) at 2 dpf, which is consistent with our previous finding that myeloid cells but not erythroid cells are generated in situ at this time point in the VDA in our previous report (Jin et al., 2009). Altogether, the contention that HSCs in the VDA produce erythroid cells in situ is not well supported.

Falenta and Rodaway also raised concerns over the requirement of CHT/PBI for definitive erythropoiesis. They suggested that definitive erythrocytes could arise in the absence of CHT/PBI because the proportion of rugby ball-shaped `definitive' erythrocytes was normal in the circulation of 7 dpf tail-transected embryos. Although the tail transection experiment is elegant, it suffers from two potential drawbacks. First, in embryos in which the CHT/PBI has been removed, other alternative definitive hematopoietic compartments might still function, such as the pronephros, which could compensate for the loss of the CHT/PBI. Second, the identification of rugby ball-shaped circulating cells as definitive erythrocytes is still questionable and is not generally accepted within the field, as these cells have not been shown to be lost from embryos that lack definitive hematopoiesis. Indeed, the high percentage (∼94%) of definitive-like cells among the circulating erythrocytes of 7 dpf wild-type embryos reported by Falenta and Rodaway conflicts with findings reported by Weinstein et al. that 75% of circulating cells are primitive erythrocytes at 7 dpf, and that at 10 dpf, 50% of circulating cells remain as primitive cells (Weinstein et al., 1996). Given such ambiguity, we revisited the identity of these rugby ball-shaped `definitive' cells in this study. If these cells are of definitive origin, we anticipate that they would be absent or greatly reduced in runx1 mutant embryos (runx1^w84x), in which definitive hematopoiesis is reported to be blocked at an early stage (Jin et al., 2009; Sood et al., 2010). All runx1^w84x mutants lack HSC- and lineage-specific markers in the AGM, CHT and thymus during early definitive hematopoiesis and, consequently, all runx1^w84x mutants become bloodless after 11-12 dpf, although 20% of runx1^w84x mutants regain circulating blood cells after 15 dpf (Jin et al., 2009; Sood et al., 2010). Thus, the runx1^w84x mutant is an appropriate system to evaluate whether rugby ball-shaped cells at 7 dpf are definitive hematopoietic cells. Largely consistent with the finding reported by Falenta and Rodaway, we found that 82.9% of circulating blood cells isolated from 7 dpf wild-type embryos were of rugby ball shape (see Appendix S1 and Fig. S1E,F in the supplementary material). However, the percentage of rugby ball-shaped cells did not decrease, but was instead maintained, in 7 dpf runx1^w84x mutants (in which 95.5% of circulating blood cells were of rugby ball shape), which would not be predicted if they are definitive in origin (see Appendix S1 and Fig. S1E,F in the supplementary material). Thus, it appears more likely that the rugby ball-shaped cells observed at 7 dpf reflect a more matured state of primitive erythrocytes rather than a definitive erythroid population. Our data also suggest that rugby ball shape is not a reliable criterion for scoring definitive erythrocytes in related experiments.

As a result of these findings, we stand by our original conclusions.