Splitting of circulating red blood cells as an in vivo mechanism of erythrocyte maturation in developing zebrafish, chick and mouse embryos

From an evolutionary perspective, mammalian red blood cells (RBCs) are enucleated (Orkin and Zon, 1997) to increase hemoglobin levels (Ji et al., 2011). Because the committed production of the erythroid lineage via definitive erythropoiesis uniquely includes nuclear extrusion in mammals, the red cell population in the peripheral circulation of, say, adult rodents or humans consists entirely of nucleus-free erythrocytes. Nucleated and larger-sized RBCs, in contrast, are found in the peripheral blood of non-mammalian vertebrates such as birds (Schwartz and Stansbury, 1954) and teleost fish (Randall et al., 2014), as well as in prenatal vertebrate stages, including early mammalian embryos, as a result of primitive erythropoiesis.

Aside from murine models, the zebrafish is another lab species that is increasingly utilized in hematopoietic research because of its amenability to genetic manipulations and regenerative capacity (Hlushchuk et al., 2016; Poss et al., 2003). Primitive erythropoiesis in the zebrafish takes place in various regions, most prominently in the inner cell mass (Orkin and Zon, 2008). Once blood circulation starts at 24 h post-fertilization (hpf), primitive RBCs enter the bloodstream and mature (Kulkeaw and Sugiyama, 2012). It is thought that primitive erythropoiesis accounts for all circulating RBCs until 4 days post-fertilization (dpf) (Kulkeaw and Sugiyama, 2012). From 26 to 48 hpf, definitive hematopoietic precursors emerge from endothelial cells lining the ventral wall of the dorsal aorta. The space between the dorsal aorta and axial vein is equivalent to the aorta–gonad–mesonephros region of mouse ontogeny at embryonic days E10.5–E12.5. These definitive precursors progressively appear between 2 and 6 dpf in the circulation and subsequently expand within the caudal hematopoietic tissue of larval fish. Hematopoiesis of the adult fish shifts to the kidney marrow and thymus (Kulkeaw and Sugiyama, 2012).

In chick development, primitive hematopoiesis occurs in the blood islands of the yolk sac (Lassila et al., 1982). As a result, immature RBCs (erythroblasts) enter the circulation and undergo several mitoses along with their terminal differentiation (Baumann and Meuer, 1992). These yolk sac-derived primitive RBCs remain the principal red cell population until day 6 of embryonic development. By that time, definitive RBCs originate from stem cells located in the wall of the aorta and enter the vascular bed, thus gradually replacing primitive lineage constituents. Definitive RBCs are unable to produce embryonic hemoglobin; instead they mainly synthesize the two adult hemoglobins HbA and HbD (Baumann and Meuer, 1992).

In the mouse embryo, primitive erythroid cells, with their nucleated morphology and expression of embryonic plus adult hemoglobins (Kingsley et al., 2004), start developing in yolk sac blood islands between days 7 and 8 of embryonic development and, shortly thereafter, are seen circulating in the bloodstream until E16.5 (Baron et al., 2013; Fraser et al., 2007; Kingsley et al., 2004). From E9.5 to E12.5, the embryonic blood is dominated by large basophilic proerythroblasts, while during the 24 h period from E12.5 to E13.5 a striking change in Giemsa reactivity, illustrated by a switch to orthochromatophilic erythroblasts as the predominant erythroid entity, takes place (Fraser et al., 2007). Eventually, primitive mouse erythroblasts extrude their nuclei between E12.5 and E16.5 (Kingsley et al., 2004; McGrath et al., 2008). Definitive erythrocytes, with their smaller size, enucleated appearance and accumulation of adult, but not embryonic, hemoglobin originate in the fetal liver at E12.5 and rapidly predominate as circulating RBCs.

This is the first study documenting the tangling and splitting of circulating primitive RBCs as an in vivo mechanism for RBC division in zebrafish or mouse embryos and RBC maturation in chicken.

Zebrafish embryos, Danio rerio (F. Hamilton 1822), were dechorionated around 28 hpf and mounted in E3 medium (5 mmol l⁻¹ NaCl, 0.17 mmol l⁻¹ KCl, 0.33 mmol l⁻¹ CaCl₂, 0.33 mmol l⁻¹ MgSO₄) containing 0.4% low-melting point agarose, 0.003% phenylthiourea and 0.02% MS-222 (all reagents from Sigma-Aldrich, Buchs, Switzerland). Overnight imaging of several caudal vein plexus (CVP) regions was performed at 28.5°C using a Zeiss LSM880 confocal microscope with an LD C-apochromat 40×/1.1 W objective (Zeiss, Oberkochen, Germany). All images and videos were processed using Fiji v1.50i (Schindelin et al., 2012). To assess RBC mitosis, whole-mount immunohistochemistry using phospho-histone H3 (pH3) antibodies (Merck Millipore, Schaffhausen, Switzerland; Ref: 06-570) was performed with double-transgenic embryos Tg(fli1a:GFP;gata1:DsRed) at 48 hpf. Because of the presence of a fli1a promoter, these transgenics express a GFP reporter selectively in their vascular endothelial cells (green fluorescing vascular network), while erythroid precursors and RBCs are highlighted in vivo through expression of a DsRed fluorophore under the control of the gata1 promoter (see Fig. 1). Our zebrafish experiments were ethically approved by the Cantonal Veterinary Office (license number: BE59/15; validity: 31 October 2018).

Chicken embryos, Gallus gallus domesticus (Linnaeus 1758), were pre-incubated at 37°C and placed in 100×20 mm Petri dishes (Corning, NY, USA) at post-incubation day 3. At day 7 (n=3), day 10 (n=5) and day 13 (n=3), 100 µl of 10% FITC-dextran (2 MDa; Sigma-Aldrich) was injected below the CAM for contrast enhancement. Images and videos were acquired using a Leica DFC365X camera mounted on a Leica stereomicroscope M205FA (Leica Microsystems, Heerbrugg, Switzerland). Blood smears were stained with Giemsa May–Grünwald (Sigma-Aldrich) on day 5 (n=4), day 7 (n=6), day 10 (n=7) and day 13 (n=2). For transmission electron microscopy, blood was collected on day 10 (n=4), centrifuged (1000 rpm, 4 min) in 2.5% glutaraldehyde (540 mOsm, pH 7.4) and post-fixed in osmium tetroxide (340 mOsm, pH 7.4). Next, samples were embedded in epon812 and stained using uranyl acetate and lead citrate. For the time period used herein (day 7–day 13), CAM assays are not considered animal tests by Swiss legislature.

Six to 12 mouse embryos, Mus musculus Linnaeus 1758, were harvested from gravid C57BL/6JRj females killed by cervical dislocation at developmental days E9.5, E11, E12.5 and E14. Smears of embryonic blood from any of these stages were prepared, mitotic figures of cells were stained with Giemsa May–Grünwald and images were documented using a Zeiss Axio Imager M2 light microscope (Zeiss) with an Olympus UC50 color-CCD camera (Olympus Schweiz AG, Volketswil, Switzerland). The mitotic frequency of primitive RBCs as proportion of 1000 counted erythroid cells was determined per developmental stage in comparison with maternal blood by five different individuals through manual counting. Evidence for in vivo mitosis and dumbbell morphology of nucleated primitive RBCs was obtained using whole-mount preparations of embryonic yolk sacs and umbilical cord tissue. Following fixation of the tissue in Karnovsky buffer (2.5% glutaraldehyde + 2% paraformaldehyde in 0.1 mol l⁻¹ cacodylate buffer, pH 7.4), cells were DAPI stained and visualized in a Zeiss LSM880 confocal microscope with a Plan Apochromat 63×/1.4 NA oil objective (Zeiss). The mouse experiments were ethically approved by the Cantonal Veterinary Office (license number: BE43/18; validity: 01 June 2020).

All statistical analyses were performed with GraphPad Prism v5.04. P-values of two-tailed paired t-tests were considered statistically significant if P<0.05 (*P<0.05, **P<0.01, ***P<0.001). Values are given as means±95% confidence interval.

Blood is a major site of erythrocyte proliferation and maturation in lower vertebrates and early ontogenic stages. Yamamoto and Iuchi (1975) were, to the best of our knowledge, the first to demonstrate by electron microscopy the maturation of primitive erythroid cells in the circulation of rainbow trout embryos. Although later studies pointed to the importance of mitoses to promote the maturation of circulating RBCs in fish (Glomski et al., 1992), further mechanistic insights into molecular or biomechanical cues to initiate and govern both proliferation and maturation of red cells in the bloodstream were slow to emerge. Here, we demonstrate for the first time that splitting of circulating RBCs marks a quantitatively meaningful in vivo mechanism of RBC division in zebrafish embryos and that vascular bifurcations and transluminal pillars are key biomechanical components in triggering the process.

Between 28 and 72 hpf, the CVP (Isogai et al., 2001) of zebrafish embryos consists of a highly interconnected vascular network (Fig. 1A,B; Movie 1). RBC splitting in the CVP proceeds as a four-step process (Movie 2). (i) Division: mitotic activity of circulating RBCs was determined using whole-mount pH3 staining of 48 hpf Tg(fli1a:GFP;gata1:DsRed) zebrafish (Fig. 1B,C). Gata1 is a marker for red blood cells and erythroid precursors (Traver et al., 2003). On average, 1.49±0.28% (n=9) of the intravascular gata1⁺ cells in the CVP were pH3⁺ (Fig. 1D). (ii) Tangling: long-term confocal imaging of the CVP was performed between 36 and 72 hpf. On average, 11.24±0.93 (n=11) dumbbell-shaped RBCs tangled up per hour at vascular bifurcations and transluminal pillars, per region (Fig. 1F). We identified around 3–4 tangling regions per embryo, suggesting ∼1000 splitting events occur per day between 32 and 72 hpf. This high frequency would render RBC splitting a biologically meaningful mechanism for RBC division. (iii) Elongation: once tangled, RBCs remained at the vascular structure where the soma-to-soma connection subsequently elongated (Fig. 1E). Afterwards, elongation proceeded by exponential growth, indicating the process occurs faster with longer-lasting tangles (Fig. 2E,F). Tangled RBCs receive constant blood flow with strong shear forces as potent morphogenic cues, which are known to regulate a variety of processes (Brönnimann et al., 2016; Hoefer et al., 2013; North et al., 2009). However, elevated blood flow velocity, a key determinant of shear forces, altered neither splitting time nor elongation speed, suggesting that RBC splitting itself is unaffected by incident shear stress (Fig. 2A–D). (iv) Splitting: after 129.56±1.29 s (n=11), the connection between the somata eventually broke, thus releasing two separate cell bodies into circulation (Fig. 1G).

To assess whether RBC splitting adds to the renewal of nucleated RBCs in other models, we utilized the chick CAM with its dense capillary plexus forming around day 10 (Djonov et al., 2000; Ribatti et al., 2001). Contrary to our findings with zebrafish embryos, we observed thousands of simultaneous tangling events in the CAM (Fig. 3A,B; Movie 3). In blood smears (Fig. 3C,D), the fraction of immature RBCs with mitotic figures decreased dramatically between day 7 (8.48±1.39%) and day 10 (0.19±0.1%; P=0.0013) (Fig. 3E). Pre- and post-tangling RBCs were present only after day 7. After day 7, immature RBCs polarize and become dumbbell-shaped structures (Fig. 3D), with the nucleated soma and the non-nucleated appendix remaining connected by a weakly labeled cytoplasmic bridge. The frequent occurrence of post-tangle figures in the blood at day 10 and day 13 strongly suggests RBC splitting (Fig. 3D), although we were unable to document an entire event in vivo. Possibly, chick elongation phases markedly exceed those in zebrafish.

Recruitment of erythroblasts from extravascular yolk sac sites along with mitotic maturation plus terminal differentiation of primitive RBCs inside the circulation of the chicken embryo have already been reported (Baumann and Meuer, 1992). Immature avian and zebrafish RBCs contain functional mitochondria and numerous free ribosomes for protein synthesis (Moritz et al., 1997). From the differentiation of basophilic erythroblasts to polychromaphilic erythroblasts and mature RBCs, the amount of free ribosomes progressively decreases (Brasch et al., 1974; Glomski and Pica, 2011). Consequently, protein synthesis and mitochondrial oxygen consumption slow down, which effectively decreases reactive oxygen species production and increases oxygen availability (Stier et al., 2013; Zhang et al., 2011). In electron micrographs, immature (pre-tangle) RBCs displayed higher ribosomal and mitochondrial content than mature (post-tangle) RBCs (Fig. S1a–d), suggesting that RBC splitting underlies the reduction of ribosome numbers, mitochondria and possibly other organelles in chick RBCs undergoing maturation.

Abundant evidence for the division of circulating nucleated RBCs, including dumbbell-shaped daughter cells that retained a cytoplasmic bridge between two nucleated somata, was also obtained in Giemsa-stained blood smears from E9.5-, E11- (Fig. 3F), E12.5- (Fig. 3G) and E14-staged embryos of C57BL/6JRj mice. The frequency of mitotic events ranged from 2.7% (E.9.5) to 3.5% (E14), without varying significantly between developmental stages (data not shown). As E12.5 embryos contain a total of 4.8 million Coulter-counted circulating RBCs (Kingsley et al., 2004), the mean 2.9% proportion of dividing cells obtained here amounts to a total of ∼139,000 mitotic cells within the peripheral blood of these embryos. Assuming a doubling time of approximately 8 h for primitive E12.5 RBCs (Isern et al., 2011), the 2.9% mitotic frequency yields an estimated daily renewal of a considerable proportion (1.1 million or ∼23%) of the total population of circulating RBCs in E12.5 mouse embryos. DAPI staining of nucleated RBCs in conjunction with confocal microscopy allowed us to detect dividing RBCs and post-tangled dumbbell figures of RBCs within the umbilical vascular network of the embryonic yolk sac (Fig. 3H).

Collectively, RBC splitting was shown to underlie and promote the division of circulating RBCs in zebrafish and mouse embryos and the maturation of immature RBCs in the developing chick embryo.

Microscopy was performed on equipment supported by the Microscopy Imaging Center (MIC), University of Bern, Switzerland. The work was carried out during study or qualification programs of the (i) University of Bari: Transplantation of organs and tissues and cell therapies; and (ii) University of Bern: (a) Graduate School for Cellular and Biomedical Sciences; (b) BNF National qualification programme. The authors thank Prof. Nadia Mercader Huber for antibodies and fruitful discussions, Dr Roman Schönauer for help with confocal microscopy, Jeannine Wagner and Werner Graber for support in electron microscopy, and Regula Bürgy, Eveline Yao and Séverin Yao for their cell staining and cell counting assistance.