Abstract
The erythroid terminal differentiation program couples sequential cell divisions with progressive reductions in cell size. The erythropoietin receptor (EpoR) is essential for erythroblast survival, but its other functions are not well characterized. Here we use Epor−/− mouse erythroblasts endowed with survival signaling to identify novel non-redundant EpoR functions. We find that, paradoxically, EpoR signaling increases red cell size while also increasing the number and speed of erythroblast cell cycles. EpoR-regulation of cell size is independent of established red cell size regulation by iron. High erythropoietin (Epo) increases red cell size in wild-type mice and in human volunteers. The increase in mean corpuscular volume (MCV) outlasts the duration of Epo treatment and is not the result of increased reticulocyte number. Our work shows that EpoR signaling alters the relationship between cycling and cell size. Further, diagnostic interpretations of increased MCV should now include high Epo levels and hypoxic stress.
Similar content being viewed by others
Introduction
Red-cell formation (erythropoiesis) is continuous throughout life, replenishing senescent red cells and responding to increased demand during anemia, bleeding, or hypoxic stress. Anemia resulting from nutritional deficiencies, malaria, chronic disease, cancer, or hereditary hemoglobinopathies, accounts for 8.8% of all disabilities globally1. Erythropoietin (Epo) is the principal and essential regulator of definitive (adult-type) erythropoiesis, regulating erythropoietic rate in the basal state and during the stress response. Epo acts through its receptor, EpoR, a transmembrane type I cytokine receptor2, first expressed in the earliest erythroid-committed progenitors. EpoR expression peaks in colony-forming-unit-erythroid (CFU-e) progenitors3,4 (Supplementary Fig. 1) with the onset of erythroid terminal differentiation (ETD)5, a process that starts with the induction of erythroid gene transcription4. During ETD, erythroblasts undergo 3–5 maturational cell divisions in which they become smaller, express hemoglobin, and enucleate to form reticulocytes. EpoR rescues proerythroblasts and basophilic erythroblasts (here collectively termed ‘early erythroblasts’) and CFU-e from apoptosis6,7, a principal mechanism of erythropoietic rate regulation8,9. EpoR is downregulated in late erythroblasts, which no longer depend on its signaling for survival10,11,12 (Supplementary Fig. 1). Epo−/− or Epor−/− mice die on embryonic day 13 (E13) as a result of severe anemia5,13,14. Their fetal liver, the site of hematopoiesis at mid-gestation, contains CFU-e progenitors but is entirely devoid of cells undergoing ETD5,13,14.
The absolute dependence of definitive early erythroblasts on EpoR signaling for survival makes it challenging to identify other essential functions of EpoR in these cells. Key open questions include a role for EpoR in cell-cycle regulation. Although early reports suggested that Epo does not alter the erythroblast cell cycle15, EpoR signaling induces cell-cycle genes in these cells16, and is essential for the cycling of Epo-dependent cell lines17,18 and cultured CFU-e19. EpoR also promotes cycling in yolk-sac-derived primitive erythroblasts during early embryonic development20. Therefore, EpoR may also be required for the cycling of adult-type erythroblasts, a function that may contribute to the erythropoietic stress response.
A second open question is whether EpoR is required for induction of erythroid genes. EpoR and similar cytokine receptors do not instruct lineage choice and are instead required for essential permissive functions21,22,23,24. It is not clear, however, whether these include signals that facilitate erythroid gene transcription. EpoR signaling was shown to phosphorylate GATA1, a key erythroid transcriptional regulator, but the broad impact of this on GATA1 function is not clear25.
To address these gaps, we developed a genetic system that identifies essential non-survival functions of EpoR signaling. We rescued mouse Epor−/− fetal liver progenitors from apoptosis by transduction with the anti-apoptotic protein Bcl-xL, and compared their ensuing differentiation with that of Epor−/− progenitors that were rescued by re-introduction of the EpoR. We found that the Bcl-xL survival signal, in the absence of any EpoR signaling, supported expression of the erythroid transcriptional program and formation of enucleated red cells. However, key ETD features were abnormal. First, erythroblasts underwent slower and fewer cell cycles, suggesting a cell-cycle role for EpoR. We confirmed this role in adult mice in vivo, finding that Epo administration shortened the cycle of early erythroblasts, cells that are already amongst the fastest cycling cells in the bone marrow26,27,28. Second, we found that, unexpectedly, despite stimulating rapid cycling, EpoR signaling increases cell size in both erythroblasts and red cells. This contrasts with the well-established inverse relationship between the number of erythroblast cell divisions and red-cell size29,30,31,32. Using mice doubly deleted for both EpoR and HRI, we found that EpoR regulation of red-cell size is also independent of the well-described iron and heme-regulated pathway33,34,35. We confirmed these findings in healthy human volunteers that were administered Epo, finding an increased MCV that persisted long after Epo and reticulocyte levels returned to baseline. Our work reveals novel EpoR functions, and suggests hypoxia, anemia, and other high-Epo syndromes as new diagnostic interpretations of increased red-cell size in the clinic.
Results
Non-survival EpoR signals are essential for normal erythroid differentiation
Erythroid differentiation in Epor−/− fetal liver is arrested at the CFU-e stage5,13,14,20. Epor−/− CFU-e can be rescued in vitro by transduction with EpoR or a similar cytokine receptor5,22. Here we asked whether transducing Epor−/− CFU-e with Bcl-xL, an anti-apoptotic transcriptional target of EpoR signaling36,37,38,39, would be sufficient to support erythroid differentiation. As control, we transduced Epor−/− cells from the same fetal livers with the EpoR. The use of bicistronic retroviral expression vectors allowed us to track transduced cells (Fig. 1a).
As expected, Epor−/− cells transduced with ‘empty’ vector failed to give rise to CFU-e-derived colonies in semi-solid medium, whereas EpoR-transduced Epor−/− cells (EpoR-Epor−/−) generated CFU-e colonies in an Epo-dependent manner. Bcl-xL-transduced Epor−/− cells (Bcl-xL-Epor−/−) failed to give rise to CFU-e colonies of the usual size and appearance (Fig. 1b). Instead, they generated a similar number of much smaller colonies with fewer cells (colony areas were 439 ± 208 μm2 versus 217 ± 106 μm2, mean ± SD, for EpoR-Epor−/− v. Bcl-xL-Epor−/−, p = 3.6 × 10−13; Fig. 1c, d). Co-transduction of Epor−/− cells with both Bcl-xL and a constitutively active form of Stat5, an EpoR-activated transcription factor, was also not sufficient to support the formation of normally-sized Epor−/− CFU-e colonies (Fig. 1b).
Liquid cultures of Bcl-xL-Epor−/− in the presence or absence of Epo, and of EpoR-Epor−/− erythroblasts with Epo, contained hemoglobinized cells by 36 h post transduction, while EpoR-Epor−/− erythroblasts without Epo did not (Fig. 1e). However, differentiation of Bcl-xL-Epor−/− erythroblasts appeared to be accelerated, with cultures containing smaller and morphologically more mature erythroblasts, including many enucleated cells; there were few if any enucleated cells in cultures of EpoR-Epor−/− erythroblasts at this time (Fig. 1e).
Differentiation abnormalities of Bcl-xL-Epor−/− erythroblasts were also evident from flow cytometric analysis. In wild-type progenitors, the transition from the CFU-e stage to ETD is marked by sharp upregulation of CD71 (encoded by the transferrin receptor, Tfrc), followed by upregulation of Ter1194,26,40. Epor−/− progenitors arrest in development prior to CD71 upregulation (the small number of Ter119+ cells in Epor−/− fetal liver are yolk-sac-derived erythroblasts40, Fig. 1f). Transduction of Epor−/− fetal liver cells with EpoR allowed them to resume the expected sequence of cell surface marker expression, upregulating CD71 by 18 h and Ter119 by 36 h (Fig. 1g). By contrast, Bcl-xL-Epor−/− cells failed to upregulate CD71 at any point of the culture although they did upregulate Ter119 (Fig. 1g).
Thus, our initial analysis showed that, when rescued from apoptosis by Bcl-xL, Epor−/− progenitors can differentiate into hemoglobinized, enucleated red cells in the absence of additional EpoR signals. However, their ETD is abnormal, failing to upregulate CD71, and differentiating prematurely into fewer and smaller red cells.
Erythroblasts undergo fewer and slower cell cycles in the absence of EpoR signaling
CFU-e express the receptor tyrosine kinase Kit and the Interleukin-3 (IL3) receptor22,41,42. Addition of stem cell factor (SCF, the Kit ligand) and IL3 to the media increased the overall yield of transduced Epor−/− fetal liver cells, but the difference in cell number between Bcl-xL-Epor−/− and EpoR-Epor−/− erythroblasts remained (Supplementary Fig. 2a). We modified our transduction protocol to make use of this improvement in yield, culturing freshly transduced Epor−/− progenitors for 15 h in SCF and IL3 before transitioning the cells to an Epo-containing medium for the remainder of differentiation. Since SCF and IL3 also promote the growth of myeloid cells, all analysis was performed on cells that were both negative for non-erythroid lineage markers and positive for reporters of transduction (hCD4 and/or GFP, Supplementary Fig. 2b, Fig. 2a).
Pre-incubation with SCF and IL3 did not ameliorate the abnormalities of Bcl-xL-Epor−/− erythroblast differentiation. In particular, these cells failed to upregulate CD71 (Fig. 2a, b). We examined the possibility that these abnormalities were the result of overexpression of Bcl-xL, rather than the absence of EpoR signaling, by co-transducing Epor−/− progenitors with both EpoR and Bcl-xL, each linked to a distinct reporter (Supplementary Fig. 2c). The doubly transduced progenitors were indistinguishable from cells transduced with only the EpoR, indicating that the lower cell number and failure to upregulate CD71 were not the result of Bcl-xL overexpression, but rather, of absent EpoR signaling (Supplementary Fig. 2d-g).
The transferrin receptor is critical for iron import into erythroid cells. Iron deficiency leads to microcytic anemia. We therefore examined whether iron deficiency might account for the abnormal differentiation of Bcl-xL-Epor−/− erythroblasts, by co-transducing Epor−/− progenitors with Tfrc, in addition to either Bcl-xL or EpoR (Fig. 2c–e). In an alternative approach, we added iron-loaded ferric-salicylaldehyde isonicotinoyl hydrazone (Fe-SIH) to the culture medium of both Bcl-xL-Epor−/− and EpoR-Epor−/− erythroblasts. SIH is a cell-membrane-permeable synthetic iron chelate, which, when pre-loaded with iron, will deliver iron intracellularly for heme synthesis, bypassing defects in Tfrc iron transport in erythroid cells43. Neither of these approaches altered the proliferative defect of Bcl-xL-Epor−/− erythroblasts (Fig. 2c, e). The viability of all erythroblasts was high with no significant difference between Bcl-xL-Epor−/− and EpoR-Epor−/− erythroblasts (Fig. 2d), suggesting that the proliferative deficit of Bcl-xL-Epor−/− erythroblasts is the result of fewer cell divisions. In the first 26 h of culture, there was a substantial difference in doubling time (6.1 h v. 8.6 h for EpoR-Epor−/− v. Bcl-xL-Epor−/− erythroblasts, Fig. 2c, e). The doubling time of 6 h for EpoR-Epor−/− is in good agreement with our recent finding of a 6 h cell cycle in wild-type early erythroblasts in vivo26, and with the finding that early erythroblasts have the shortest cell cycle amongst bone-marrow hematopoietic progenitors27.
Iron may affect cell growth by acting as a cofactor in ribonucleotide reductase (RNR) catalysis of deoxyribonucleotide synthesis44. However, supplementation of the culture medium with deoxyribonucleosides (dN), which bypass RNR via the deoxyribonucleoside kinase salvage pathway45, had little effect on the proliferative defect (Fig. 2c, e). Taken together, in the absence of EpoR signaling, erythroblasts fail to upregulate CD71 and also undergo fewer and longer cell divisions. Supplementation with iron or deoxyribonucleosides does not rescue these deficits.
Epo administration shortens cell-cycle duration in early erythroblasts in vivo
To test whether EpoR stimulation alters cell-cycle length in vivo, we used a mouse transgenic for histone H2B fused to a fluorescent timer protein (H2B-FT, Fig. 2f), which fluoresces blue when first synthesized but matures over 1–2 h into a red fluorescent protein27. The ratio of blue fluorescence to total fluorescence (red + blue) is an indicator of cell-cycle length27. Administration of Epo (100 U) once daily resulted in a clear shift in the ratio of blue to total fluorescence at 36 h, in all bone-marrow early erythroblast subsets (Fig. 2f, g and Supplementary Fig. 3). These data confirm that Epo/EpoR signaling increases cell-cycle speed in wild-type erythroblasts in vivo.
EpoR shortens both G1 and S phase through an iron-independent mechanism
The onset of ETD is associated with cell-cycle shortening, from ~15 h in CFU-e, to 6 h in early erythroblasts4,26,40, including a shortened, 4-h-long S phase26. We asked whether the cell-cycle shortening effect of EpoR (Fig. 2e, g) is exerted in G1 or in S phase. The shortening of G1 by cytokine receptor signaling is well documented46,47,48,49,50. However, to our knowledge, there are no reports of cytokine signaling altering S phase speed.
To examine this, we pulsed cultures of EpoR or Bcl-xL-transduced Epor−/− erythroblasts with bromodeoxyuridine (BrdU), a nucleoside analog that is incorporated into DNA during S phase, and analyzed the cells 30 min following the pulse. The fraction of cells that are labeled with an anti-BrdU antibody indicates the proportion of cells in S phase at the time of the pulse. Further, the amount of BrdU incorporated into S phase cells during the 30 min pulse, as measured by the BrdU mean fluorescence intensity (MFI) of S phase cells, indicates the intra-S phase rate of DNA synthesis, which is inversely related to S phase duration26. We found that, in the first 10 h of ETD, BrdU MFI in S phase cells was 50% higher in EpoR-Epor−/−, compared with Bcl-xL-Epor−/− erythroblasts, suggesting that EpoR signaling increases S phase speed (Fig. 3a, b).
If the slowing of S phase alone could account for the increased cell-cycle length of Bcl-xL-Epor−/− erythroblasts, S phase would constitute a larger fraction of total cell-cycle duration. However, the fraction of Bcl-xL-Epor−/− erythroblasts in S phase was actually somewhat lower, with a corresponding increase in the fraction of cells in G1 (Fig. 3b), and little change in the fraction of cells in G2 or M (not shown). These observations suggest that, in the absence of EpoR signaling, both S and G1 phases lengthen.
Supplementing the culture medium with Fe-SIH increased S phase speed modestly in all Epor−/− erythroblasts (Fig. 3c). There was no rescue of S phase speed in Bcl-xL-Epor−/− erythroblasts by either the addition of deoxyribonucleosides or double transduction with both Bcl-xL and Tfrc (Fig. 3d), although there was a small increase in the number of cells in S phase in the latter (Fig. 3e). Taken together, these results indicate that EpoR is essential for accelerating both G1 and S phases of the cycle in early ETD, via mechanisms that are largely independent of iron and the nucleotide pool.
Imaging flow cytometry shows Epor −/− erythroblasts and reticulocytes are smaller
Nutritional deficiencies, drugs, or genetic perturbations that reduce the number of cell divisions lead to the formation of larger red cells (macrocytosis29,30,31,32,51). Therefore, we expected that the fewer cell divisions of Bcl-xL-Epor−/− erythroblasts would result in larger size for these cells. Instead, they appeared to be smaller (Fig. 1e). To address this question quantitatively, we measured cell and nuclear size in EpoR-Epor−/− and Bcl-xL-Epor−/− erythroblasts by imaging flow cytometry (Fig. 4a, b). We calibrated the measured cell areas by comparing them with beads of known diameter (Supplementary Fig. 4a). We found that both cell and nuclear size were significantly smaller in the absence of EpoR (7.5 ± 0.6 μm and 6.7 ± 0.7 μm for EpoR-Epor−/− and Bcl-xL-Epor−/− erythroblasts, respectively, mean ± sem, p = 0.001, t = 46 h). Although Bcl-xL-Epor−/− erythroblasts express significantly lower CD71 (Fig. 2a, b), the addition of Fe-SIH to the culture did not alter their smaller cell or nuclear size (Fig. 4b).
We asked whether the smaller size of Bcl-xL-Epor−/− erythroblasts could reflect an accelerated process of differentiation. If at any given time of the culture Bcl-xL-Epor−/− erythroblasts were smaller only as a result of being at a more advanced maturation stage, they should give rise to normally-sized enucleated reticulocytes, albeit at an earlier time. However, imaging flow cytometry showed that Bcl-xL-Epor−/− reticulocytes were significantly smaller (5.6 ± 0.5 μm vs. 4.5 ± 0.15 μm for of EpoR-Epor−/− vs. Bcl-xL-Epor−/−, mean ± sem, p = 0.002, Fig. 4c, d).
To assess whether the smaller size of Bcl-xL-Epor−/− erythroblasts is the result of overexpression of Bcl-xL, rather than absent EpoR signaling, we doubly transduced Epor−/− fetal liver cells with both EpoR and Bcl-xL. We used the Bcl-xL-linked GFP and the EpoR-linked hCD4 fluorescence reporters to quantify expression and ensured that all comparisons were made between cells expressing similar levels of each retroviral vector (Supplementary Fig. 4b–d). We found that erythroblasts and reticulocytes transduced with both EpoR and Bcl-xL were similar in size to those transduced with only the EpoR, and significantly larger than those transduced with only Bcl-xL (Fig. 4e; Supplementary Fig. 4c, d). Therefore, Bcl-xL overexpression is not the cause of the smaller size of Bcl-xL-Epor−/− erythroblasts and reticulocytes.
The level of EpoR expression in transduced Epor−/− cells positively correlated with erythroblast cell diameter (Supplementary Fig. 5). The relationship follows classical dose/response kinetics (Spearman correlation = 0.97, p-value = 0.004). By contrast, there was no correlation between Bcl-xL expression and cell diameter.
EpoR regulation of red-cell size is independent of HRI
HRI is activated by iron and heme deficiency and mediates the formation of smaller, hypochromic red cells, by inhibiting translation52. Bcl-xL-Epor−/− erythroblasts failed to upregulate CD71 (Tfrc), the principal iron transporter. Although iron supplementation did not rescue the smaller size of Bcl-xL-Epor−/− erythroblasts (Fig. 4b), it remained possible that intracellular iron delivery was somehow incomplete.
To determine definitively the relevance of the iron/heme/HRI pathway to cell size regulation by EpoR, we generated Epor−/−Hri−/− mice (Fig. 5a). Similar to Epor−/− mice, Epor−/−Hri−/− mice died at mid-gestation with severe anemia. We rescued both Epor−/− and Epor−/−Hri−/−-fetal liver cells in parallel, by transduction with either Bcl-xL or EpoR (Fig. 5b–e). In agreement with the known role of HRI as a negative regulator of erythroblast size, both Bcl-xL-transduced and EpoR-transduced erythroblasts were larger on the Epor−/−Hri−/− genetic background than on the Epor−/− background. Importantly, for a given genetic background, either Epor−/−Hri−/− or Epor−/−, the difference in size between Bcl-xL and EpoR-rescued cells remained (Fig. 5b–e). These results clearly show that EpoR signaling regulates cell size independently of the HRI pathway, since, even in the absence of HRI, EpoR signaling promotes the formation of larger erythroblasts (Fig. 5b–d) and reticulocytes (Fig. 5e).
Accelerated maturation in the absence of EpoR, assessed independently of cell size
Cell size is frequently used as indicator of erythroid maturational stage53,54,55,56. Our initial impression was that Bcl-xL-Epor−/− erythroblasts completed their maturation sooner than EpoR-Epor−/− erythroblasts (Fig. 1e, g). However, the finding that Bcl-xL-Epor−/− erythroblasts are smaller throughout maturation makes cell size an unreliable indicator of maturational stage in these cells. We therefore assessed maturation using two alternative measures.
First, the cell surface marker Ter119, whose expression increases with maturation53,54,55,56, reached significantly higher levels in Bcl-xL-Epor−/− erythroblasts than in EpoR-Epor−/− erythroblasts at 48 h (Supplementary Fig. 6a). Second, we looked at nuclear offset, a quantitative measure of nuclear eccentricity that is independent of cell size56 (Supplementary Fig. 6b–d). The nuclear offset is the ratio of the delta centroid (the distance between the geometrical centers of the cell and the nucleus) to the cell diameter (Supplementary Fig. 6b). Nuclear offset increased continuously throughout ETD, but did so earlier in Bcl-xL-Epor−/− erythroblasts, with the difference between Bcl-xL-Epor−/− and EpoR-Epor−/− erythroblasts peaking at 48 h (p = 0.02) (Supplementary Fig. 6c, d). Taken together, both Ter119 expression and nuclear offset suggest that EpoR signaling prolongs erythroblast maturation.
The differences between EpoR-transduced and Bcl-x-transduced Epor −/− erythroblasts are maintained across a wide range of EpoR expression levels
The Epor−/− transduction model allows confident identification of essential non-survival functions of the EpoR. This model suffers, however, from two drawbacks. First, it is in vitro; we therefore tested whether our conclusions hold in vivo (Fig. 2f, Supplementary Fig. 3, and below). Second, retroviral expression of EpoR and Bcl-xL may differ in timing or expression level from their physiological profiles. We have shown above that the slower cycles, lower CD71 expression and smaller cell size of Bcl-xL-Epor−/− erythroblasts result from absence of EpoR signaling rather than Bcl-xL overexpression (Supplementary Figs. 2, 4). To investigate the potential effect of EpoR overexpression, we transduced Epor−/− erythroblasts with high-titer, undiluted retroviral supernatant, resulting in a ~3.5-fold higher expression of EpoR, compared with physiological expression in fresh or cultured erythroblasts (Supplementary Fig. 7a). We then transduced Epor−/− fetal liver cells with either undiluted, high-titer retroviral supernatant or with five-fold and ten-fold dilutions of the same supernatant. EpoR expression decreased 8-fold by qRT-PCR in cells transduced with a 10-fold diluted supernatant. We found that the differences between EpoR-transduced and Bcl-xL-transduced Epor−/− erythroblasts, in CD71 expression, cell number, maturation rate, and cell size, were all maintained regardless of retroviral titer (Supplementary Fig. 7b–f). Therefore, EpoR functions identified with the Epor−/− transduction model are not narrowly dependent on EpoR or Bcl-xL expression levels. Further, the viability of EpoR or Bcl-xL-transduced Epor−/− erythroblasts is comparable to that of wild-type erythroblasts (Supplementary Fig. 8).
Epo increases cell size and prolongs maturation of wild-type erythroblasts in vitro and in vivo
To test our conclusions outside the Epor−/− transduction model, we asked whether Epo concentration affects cell size and maturation rate in wild-type erythroblasts in culture, and during Epo administration to mice in vivo. We differentiated wild-type fetal liver CFU-e (‘S0’ in Fig. 6a40) in vitro in the presence of a range of Epo concentrations. Cell size increased in an Epo-concentration-dependent manner, at every stage of differentiation, including reticulocytes (Fig. 6b, Supplementary Fig. 9a). The Epo-concentration range affecting cell size, from 0.01 to 10 Units/ml, corresponds to the entirety of the physiological and stress range in vivo57,58. Higher Epo also increased reticulocyte size heterogeneity (Supplementary Fig. 9b).
Epo also caused a dose-dependent delay in maturation. As expected, higher Epo resulted in higher cell number at all stages of differentiation (Supplementary Fig. 9c). However, the distribution of erythroblasts at higher Epo concentrations was increasingly skewed in favor of earlier differentiation subsets (Supplementary Fig. 9d). Similarly, the intensity of Ter119 expression decreased at higher Epo concentrations (Supplementary Fig. 9e). We further assessed cell maturation by measuring the nuclear offset, which decreased with increasing Epo concentration at all flow-cytometric stages (Supplementary Fig. 9f–g). Together these findings show that Epo prolongs ETD in a dose-dependent manner.
We also assessed the effect of Epo on erythroblast cell size in vivo. We injected mice with a range of Epo doses and used nuclear offset as a size-independent measure of maturational stage (Fig. 6c). We divided the nuclear offset distribution of all Ter119+ bone-marrow erythroblasts from saline-injected mice into quintiles (Fig. 6d). Increasing nuclear offset quintiles corresponded to increasingly mature erythroblast subsets as judged by the established criteria of decreasing CD71 and cell area, confirming the utility of this approach (Fig. 6e). We then used the nuclear offset quintiles values from these control mice to classify Ter119+ erythroblasts from Epo-injected mice into five maturational stages. We found that for a given nuclear offset-defined maturational stage, there was an Epo dose-dependent increase in cell diameter. This effect was particularly striking in erythroblasts that corresponded to the two most mature quintiles (Fig. 6f, g), confirming that Epo dose regulates erythroblast cell size.
Taken together, graded increases in Epo/EpoR signaling result in graded increases in cell size, shown by varying either the ligand concentration in wild-type erythroblasts (Fig. 6b, Supplementary Fig. 9a) or receptor expression in Epor−/− erythroblasts (Supplementary Fig. 5a).
EpoR signaling delays induction of p27KIP1, leading to increased number of cell cycles
To investigate the molecular mechanisms underlying EpoR-regulated functions, we compared gene expression in differentiating EpoR-Epor−/− and Bcl-xL-Epor−/− erythroblasts, using RT-qPCR (Supplementary Fig. 10). ETD markers Slc4a1 (Band3) and Hbb1 were induced similarly in both cell types. There were no significant differences in transcription factor expression, with the exception of Tal1, whose levels were 30% lower in Bcl-xL-Epor−/− (p < 0.005). Tal1 was previously linked to cell-cycle regulation in hematopoietic cells59,60.
Among cell-cycle regulators, the CDK inhibitor p27KIP1 (Cdkn1b) was induced prematurely in Bcl-xL-Epor−/−, reminiscent of its premature expression in Epor−/− primitive erythroblasts20. A second member of the CIP/KIP family, p57KIP2 (Cdkn1c), was also expressed at somewhat higher levels. The induction of p27KIP1 toward the end of ETD in wild-type erythroblasts contributes to mitotic exit61,62,63,64. To determine the effect of its premature induction, we transduced wild-type fetal liver S1 cells (CD71highTer119neg) with either p27KIP1 or ‘empty vector’ (Supplementary Fig. 11a). p27KIP1-transduced cells showed reduced proliferation, without affecting cell viability, suggesting they underwent fewer cell cycles (Supplementary Fig. 11b, c). Unlike Bcl-xL-Epor−/− erythroblasts, however, p27KIP1-transduced cells were larger, and slower to undergo maturation, as judged by lower nuclear offset (Supplementary Fig. 11d–f). Therefore, while the EpoR-mediated negative regulation of p27KIP1 increases cell-cycle number, its regulation of cell size and maturation rate are mediated by other pathways. Addition of the phosphatidylinositol 3-kinase (PI3K) inhibitor, LY29400265, to wild-type erythroblasts resulted in premature induction of p27KIP1 mRNA (Supplementary Fig. 12), suggesting that negative regulation of p27KIP1 by EpoR is mediated via PI3K.
Several EpoR signaling pathways are implicated in the regulation of cell size
EpoR activates three principal signaling pathways: ras/MAP kinase, Stat539,66, and PI3K67,68. Neonatal mice hypomorphic for Stat5 have microcytic anemia69. Here we found that, similarly, circulating red cells from E13.5 Stat5-deficient embryos are smaller than those of wild-type littermates (Supplementary Fig. 13a, b). Using U0126, a MEK1- and MEK2-specific inhibitor70, and the PI3K inhibitor LY294002, we examined the role of these pathways in the regulation of cell size. PI3K inhibition significantly decreased the size of early (‘S2’) and late (‘S3') erythroblasts and reticulocytes, but MEK1/2 inhibition had no consistently significant effect (Supplementary Fig. 13c, d). Therefore, it is likely that cell size regulation by EpoR is the integrated result of multiple signaling pathways.
Epo administration increases red-cell size (MCV) and size variation (RDW) in human volunteers
We examined the effect of Epo on red-cell size in healthy volunteers in three intervention studies. Participants were either given Epo (studies #1 and #2, Fig. 6h, Supplementary Figs. 14 and 15), or subjected to phlebotomy (study #3, Supplementary Fig. 16). In studies #1 and #2, the effect of Epo on athletic performance was examined, and will be either reported elsewhere (study #1) or was previously reported (study #271). Here we present the detailed blood parameters associated with these studies.
In study #1 (Fig. 6h, Supplementary Fig. 14), baseline parameters were established during four weekly blood samplings, followed by injection with Epo (20 IU/kg every other day, 25 subjects) or placebo (9 subjects) for 3 weeks. On average, hemoglobin increased by 5% over baseline values in the Epo group during the treatment period. Blood sampling continued for an additional 5 weeks following cessation of treatment. In study #2 (Fig. 6h, Supplementary Fig. 15), baseline measurements were followed by weekly dosing with Epo (24 subjects) or placebo (24 subjects) for 7 weeks, with Epo dosing adjusted to achieve an increase of 10–15% in hemoglobin. Follow-up continued for a month after cessation of treatment. In study #3 (Supplementary Fig. 16), 21 subjects participated in a randomized double-blind placebo-controlled crossover study in which 900 ml of whole blood was withdrawn from the treatment group by venipuncture. Subjects were then followed for 25 days.
In all three studies, there was a significant increase in MCV in the treatment groups compared with baseline values and with the placebo group, which persisted well beyond the treatment period (Supplementary Figs. 14–16, Supplementary statistical analysis). There was no correlation between MCV and the reticulocyte count, whose time courses were clearly divergent (r < 0.1 between MCV and reticulocyte count in all three studies, Pearson’s product-moment correlation, Supplementary statistical analysis). In studies #1 and #2, the reticulocyte count increased during Epo treatment, but declined sharply below baseline values as soon as Epo treatment ceased. By contrast, MCV values remained high (Fig. 6h). Similarly, in study #3, MCV values continued to climb at a time when the reticulocyte count was declining (Supplementary Fig. 16). Thus, the increase in MCV is not the result of an increase in the number of reticulocytes. Together with the increase in MCV, there was an increase in red-cell distribution width (RDW-SD, Fig. 6i, Supplementary Figs. 14 and 15; no RDW is available for study #3). There was a significant, positive correlation between MCV and RDW-SD (r = 0.51, p = 2 × 10−28 for study #1; r = 0.52, 2 × 10−24 for study #2).
Red-cell volume declines continuously as red cells age72,73,74,75,76. Therefore, we considered the possibility that the persistently elevated MCV following Epo administration might be the result of the expected increase in the relative number of younger red cells, rather than an increase in their size. To address this, we simulated the expected increase in MCV that would arise only from an increase in the proportion of younger red cells, assuming no effect of EpoR signaling on red-cell size (Supplementary analysis: ‘Simulation of MCV’). This simulation indicated that an increased proportion of younger red cells cannot fully account for the extent or duration of the observed increase in MCV following Epo administration, consistent with a direct role for EpoR signaling in the regulation of cell size.
Discussion
Using a genetic model in which we provide Epor−/− erythroblasts with exogenous survival signaling, we identified novel non-redundant functions for EpoR during ETD. EpoR signaling determines the number and speed of cell divisions and duration of terminal differentiation. While it has little effect on the broad ETD transcriptional program, it drives the formation of qualitatively different, larger red cells. In wild-type erythroblasts, EpoR signaling increases cell size in an Epo dose-dependent manner at every stage of erythroid terminal differentiation (ETD), leading to the production of larger reticulocytes and RCs. Human intervention studies are consistent with a similar effect of EpoR signaling on red-cell size in human erythropoiesis. In the discussion below we integrate the apparently disparate EpoR functions into a coherent model (Fig. 7). We also discuss previously unexplained instances of macrocytic and heterogeneously-sized red cells, now interpretable as the result of increased EpoR signaling during hypoxic stress.
The ETD is a time of rapid change in many aspects of the cell. Our results support a model in which ETD has two-phases: an early, Epo-dependent phase, and an Epo-independent late phase39 (Fig. 7). EpoR expression peaks in early erythroblasts, which are highly dependent on EpoR signaling for survival6,53,77, and exquisitely sensitive to Epo, as judged by Stat5 phosphorylation66. By contrast, late erythroblasts downregulate EpoR12 and are relatively resistant to apoptosis53,77. The functions we identified here for EpoR signaling in ETD are similarly focused on early erythroblasts. In addition to EpoR signaling, ETD is also supported by the erythroblastic island niche, an area that was not addressed in our model.
We identified five key non-survival functions of EpoR signaling, in Epor−/− and in wild-type erythroblasts, in vitro and in vivo: (1) EpoR prolongs ETD, as determined by delayed expression of Ter119 and delayed increase in nuclear offset; (2) it increases the number of cell cycles; (3) it skews the distribution of developing erythroblasts in favor of earlier erythroblasts; (4) it increases cell-cycle speed; and (5) it increases cell size throughout ETD, generating larger and more heterogeneous red cells. The prolongation of ETD is consistent with the increase in the number of cycles. Neither informs us directly regarding the stage(s) of ETD that are being prolonged. However, the skewed distribution in favor of early erythroblasts indicates, based on the ergodic principle78 (see the “Methods” section), that EpoR signaling prolongs early ETD relative to the late ETD phase. Together, these observations suggest that EpoR prolongs the early phase of ETD by increasing the number of early ETD cell cycles. This conclusion is consistent with our data, showing the largest differences in cell-cycle number in response to EpoR occur in the first 24 h of ETD; and with the known responsiveness of early ETD to EpoR signaling. In addition, it explains the observation that EpoR increases cell-cycle speed, since early ETD cell cycles are unusually fast26,27, and much faster than cycles in late ETD26,27,79; our observations show that EpoR signaling regulates the speed of these unique cycles.
Therefore, of the five EpoR functions, the first four are outcomes of an EpoR-driven increase in the number and speed of early ETD cell cycles (Fig. 7). One of the factors known to regulate the onset of late ETD is p27KIP1, whose induction promotes slower cycling and cell-cycle exit61,80,81. Here we found that EpoR signaling increases cell-cycle number by inhibiting p27KIP1 mRNA induction through the PI3K pathway, which was previously reported to also lead to p27KIP1 proteosomal degradation63. A similar role for EpoR, delaying p27KIP1 induction and morphological maturation, was noted in primitive yolk-sac erythroblasts20. The converse was found in Klf1−/− erythroblasts, which fail to induce p27KIP1 and fail to undergo cell-cycle exit64. Here we found that exogenous premature expression of p27KIP1 in wild-type erythroblasts reduced their cycling, but did not accelerate maturation, and like other factors that reduce cycling, resulted in larger erythroblasts. Therefore, the effects of EpoR signaling on erythroblast maturation rate and cell size are unrelated to its suppression of p27KIP1.
The most surprising of our findings was the effect of EpoR signaling on cell size. We found that erythroblasts differentiating in the absence of EpoR gave rise to smaller red cells, in spite of undergoing fewer cell cycles. Further, in wild-type fetal liver erythroblasts, cell size was sensitive to Epo concentration within the physiological and stress range. These findings appear contrary to the well-established link between the loss in cell size and the number of cell divisions during ETD. Thus, deletions of E2F429, cyclin D330, CDK2, or CDK431 each reduce the number of cell divisions during ETD and result in macrocytic red cells. Similarly, macrocytic red cells are seen when nucleotide pools limit DNA synthesis rate, as in patients treated with hydroxyurea32, or in B12 or folate deficiencies. The EpoR effect on red-cell size was also independent of a second established pathway, in which red-cell size is regulated by iron status via HRI34,52. Neither iron supplementation nor deletion of HRI corrected the cell size deficit of Epor−/− erythroblasts. While these experiments do not exclude an interaction between HRI and EpoR signaling35, they show conclusively that EpoR stimulation of larger red-cell size is independent of HRI.
Our data therefore suggest that EpoR regulates red-cell size through a novel mechanism. The finding that the EpoR-driven increase in cell size begins in early erythroblasts suggests that it takes place in the very same cells in which EpoR signaling also induces additional rapid cycles. We propose that the well-established coupling of cell size loss with cell divisions is a default state, seen in cells where EpoR signaling is weak or absent. We further suggest that strong EpoR signaling, as may occur in early erythroblasts66, can override this default state and maintain cell size in spite of rapid cycling (Fig. 7). The maintenance of cell size in dividing cells is the norm in most tissues82,83 and so it is possible that EpoR signaling permits early erythroblasts to employ similar pathways of size control as those found outside ETD. The mechanisms that determine the characteristic size of a cell and that maintain it through cell divisions are not fully understood, but are thought to depend on strong growth factor signaling to promote the metabolic pathways required for building biomass83. To maintain their size, cells must attain a size threshold before committing to cell division; in avian erythroblasts and other cell types, a larger size correlates with a longer G1 phase82,84. The ability of EpoR signaling to increase cell size in early erythroblasts, which are some of the most rapidly dividing cells in vivo26,27, predicts that these cells have exceptionally efficient mechanisms for growth. Conversely, this also implies that impairments in growth pathways would have a specifically deleterious effect, potentially contributing to the selective damage of ribosomopathies in the erythroid lineage85.
Together with an increase in cell size, high Epo also increased cell size heterogeneity, in mouse and human. Unlike low Epo levels, which generate only weak signaling and therefore relatively uniform small cells, high-Epo levels might be expected to support the survival of erythroblasts with varying Epo sensitivities53,86, in which the strength of EpoR signaling may vary, giving rise to a range of red-cell sizes (Fig. 7).
The relationship between high MCV, high RDW, and high levels of Epo may have been overlooked previously by being attributed to an increase in reticulocytes. We have excluded this possibility, finding no correlation between reticulocyte numbers and MCV. We also found that the extent and duration of increase in MCV following Epo administration cannot be accounted for solely by the skewing in the age distribution of circulating red cells in favor of younger cells (see Supplementary Analysis: Simulation of MCV). Indeed, our mouse data show increased cell size throughout terminal differentiation, including larger than normal reticulocytes, and not simply more numerous reticulocytes.
Recent GWAS and other studies have linked multiple genomic loci to the regulation of MCV87,88,89,90. These include Epo, Epor, and Lnk, all expected to alter EpoR signaling strength91. An Epo-mediated increase in MCV in clinical settings might be tempered by iron status or by pathology affecting terminal differentiation. Nevertheless, our work predicts that in the absence of erythroid pathology or nutritional deficiencies, Epo levels might be a key determinant of MCV. Indeed, an increase in Epo might account for the unexplained macrocytosis in hypoxemic patients with chronic obstructive pulmonary disease92,93 and in iron-replete pregnancy94,95. An increase in RDW was recently proposed as a potential longer-term biomarker for brief hypoxemic episodes in conditions such as acute respiratory distress, sepsis, or congestive heart failure96,97. Indeed, clinically, the RDW may prove to be a more sensitive marker of EpoR signaling than the MCV. The regulation of MCV by Epo also clarifies unexplained changes in red-cell volume associated with Kit function. Kit regulates the proliferation of early erythroid progenitors but is downregulated with entry into ETD. Gain of function Kit mutations in mice lead to erythrocytosis as a result of excess progenitors entering ETD; the red cells are microcytic98, presumably in response to a compensatory decrease in Epo. Conversely, loss of function Kit mutations are associated with an increased MCV, which is in proportion to the severity of anemia98,99, and can be now be explained by a paucity of progenitors entering ETD and the expected compensatory increase in Epo100. Transgenic expression of Epo rescues the lethal c-KitW/W mutation, also resulting in macrocytic red cells99. Given the persistence of higher MCV and RDW beyond the period in which Epo is elevated, these markers may be useful additions to a panel of diagnostic markers for detecting hypoxic stress in the clinic as well as Epo doping by athletes.
The adaptive value, if any, of a higher MCV in erythropoietic stress is not yet clear. Surprisingly, the increase in MCV in our human intervention studies was not associated with increased corpuscular hemoglobin (MCH). On the contrary, we found a statistically significant decrease in mean corpuscular hemoglobin concentration (MCHC) in both Epo intervention studies, though not in the phlebotomy intervention. Interestingly, a lower MCHC may enhance the action of 2, 3, diphosphoglycerate (2,3-DPG), an allosteric regulator that binds hemoglobin and lowers its affinity for oxygen. Red-cell 2,3-DPG increases in response to anemia or hypoxia, improving oxygen unloading in tissues101,102. The affinity of 2,3-DPG to hemoglobin increases significantly at lower MCHC103. A lower MCHC may therefore improve the 2,3-DPG-dependent unloading of oxygen. Indeed, a lower MCHC is also an HRI-regulated outcome characteristic of microcytic iron-deficiency anemia, possibly for similar reasons. The EpoR-regulated increase in MCV might therefore provide a mechanism for lowering MCHC and improving oxygen unloading in tissues during hypoxic stress.
Methods
Explanation of the ergodic principle
The ergodic principle can be applied in biology to a multi-stage process in the steady state (e.g., steady-state differentiation in tissue, or the cell cycle104,82). It suggests that in a snapshot in time of cells undergoing the process, the number of cells at each stage is inversely proportional to the length of time that cells spend at that stage. Hence, finding that a differentiation stage contains many cells suggests that cells spend a longer period of time in that stage; conversely, if a differentiation stage is sparsely populated, this would suggest that transit through that stage is fast. Therefore, as applied here, finding that EpoR signaling skews the erythroblast population in favor of early erythroblasts suggests that cells are spending proportionally more time in the early erythroblast stage.
Mice
Stat5−/− mice were obtained from Dr. Lothar Hennighausen (National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD). Epor+/− mice were obtained from the Lodish laboratory, Whitehead Institute for Biomedical Research, Cambridge, MA. Balb/C mice were obtained from the Charles River Laboratories, Wilmington, MA. The Epo/Saline injection experiment on adult mice was conducted on male C57BL6 fluorescence timer (FT) transgenic mice. Mice were housed at a dedicated facility, with regulated temperature (range 20–26 °C), a 12 h/12 h dark/light cycle, and 30–70% humidity. Mice were fed on Iso Pro 3000 irradiated rodent diet #5P76. All experiments were conducted in accordance with animal protocol A-1586 approved by the University of Massachusetts Chan Medical School Institutional Animal Care and Use Committee.
Culture medium and growth factors
Fetal liver cells were cultured in IMDM with added L-glutamine and 25 mM HEPES (Gibco), 20% fetal calf serum (Hyclone), 1% penicillin/streptomycin (ThermoFisher Scientific), 2 × 10−4 M β-Mercaptoethanol (Sigma), supplemented when indicated with 0.5 IU/ml Epo (Procrit, Amgen; 1 IU/ml = 1.2 ng/ml) and 100 ng/ml SCF (Peprotech), and 10 ng/ml IL3 (Peprotech).
Isolation of mouse erythroid progenitors
To isolate wild-type S0 cells, fetal liver cells were depleted of lineage-positive cells by labeling with biotin-conjugated CD71, Ter119, Gr1, Mac1, and CD41 antibodies followed by magnetic separation using either EasySep beads a (StemCell Technologies) or MojoSortTM Streptavidin Nanobeads (BioLegend) according to the manufacturers’ instructions.
Flow cytometry
Fetal liver cells were analyzed on LSRII (BD Biosciences) cytometers using DIVA software (BD Biosciences). Dead cells were excluded using DAPI (Roche). FACS data were analyzed using FlowJo software (Tree Star Inc., CA).
Antibodies used:
PE Mouse Anti-Human CD4 (RPA-T4) (BD Biosciences) dilution 1:50
PE/Cy7 Rat Anti-Mouse CD71 (RI7217) (BioLegend) dilution 1:100
APC/Cyanine7 Rat Anti-Mouse Ter119 (Ter119) (BioLegend) dilution 1:100
PE Rat Anti-Mouse Ter119 (Ter119) (BD Biosciences) dilution 1:100
APC Rat Anti-Mouse Ter119 (Ter119) (BD Biosciences) dilution 1:100
biotin Rat Anti-Mouse CD71 (C2) (BD Biosciences) dilution 1:100
biotin Rat Anti-Mouse Ter119 (BD Biosciences) dilution 1:100
biotin Rat Anti-Mouse Ly-6G and Ly-6C/Gr1 (RB6-8C5) (BD Biosciences) dilution 1:100
biotin Rat Anti-Mouse CD11b/Mac1 (M1/70) (BD Biosciences) dilution 1:100
biotin Rat Anti-Mouse CD41 (MWReg30) (Thermo Scientific) dilution 1:100
FITC Rat Anti-Mouse Ly-6G and Ly-6C/Gr1 (RB6-8C5) (BD Biosciences) dilution 1:100
FITC Rat Anti-Mouse CD11b/Mac1 (M1/70) (BD Biosciences) dilution 1:100
FITC Rat Anti-Mouse CD41 (MWReg30) (BD Biosciences) dilution 1:100
FITC Rat Anti-Mouse CD45R/B220 (RA3-6B2) (BD Biosciences) dilution 1:100
FITC Hamster Anti-Mouse CD3e (145-2C11) (BD Biosciences) dilution 1:100
PE Rat Anti-Mouse Ly-6G and Ly-6C/Gr1 (RB6-8C5) (BioLegend) dilution 1:100
PE Rat Anti-Mouse CD11b/Mac1 (M1/70) (BioLegend) dilution 1:100
PE Rat Anti-Mouse CD41 (MWReg30) (BD Biosciences) dilution 1:100
PE Rat Anti-Mouse CD45R/B220 (RA3-6B2) (BD Biosciences) dilution 1:100
PE Hamster Anti-Mouse CD3e (500A2) (BioLegend) dilution 1:100
Imaging flow cytometry
Imaging flow cytometry was used to analyze cell fluorescence in conjunction with morphological parameters. It was performed on an Amnis Flowsight cytometer (Luminex Corporation, TX) using INSPIRE software v6.5 (Luminex Corporation, TX). Live nuclear diameter was measured using the cell-permeable far-red fluorescent DNA dye, DRAQ5® (Cell Signaling). Amnis data was analyzed using IDEAS software v6.0 (Luminex Corporation, TX). New mask functions were generated to analyze bright-field cell area (Definition: Object (M01, Ch01, Tight)) as well as Draq5 fluorescence nuclear area (Definition: Morphology (M11, Ch11)). Raw mean Draq5 pixel intensity feature was generated using Draq5 Morphology mask for nuclear area. Raw flow cytometric feature data were exported and analyzed in R programming language.
Calibration of nuclear and cell diameters measured by imaging flow cytometry
Imaging flow cytometry was performed on standardized bead sizes, 2.0μ, 3.4μ, 5.1μ, 7.4μ, 9.96μ, and 14.3μ (Spherotech Inc.). IDEAS bright-field cell area mask (Definition: Object(M01, Ch01, Tight)) was fitted to the bead image acquisition. The data were analyzed using R, and within each bead group, values that lie greater or less than 3 standard deviations from the mean were removed (0.9% of events were removed with this threshold). To correct biases in the cell area values calculated by the Amnis software, we fit a linear model (polynomial curve) using the manufacturer bead sizes as a predictor for the Amnis calculated cell area (Stats, base R, degree = 2), with an R2 value of 0.97. This model was then used to predict cell diameters from experimental cell areas.
Analysis of imaging flow-cytometry data
Further analysis of exported imaging flow-cytometry data was done using RStudio Version 1.2.1335, RStudio, Inc. Population distributions were log normalized. Population dataset were filtered by removing outliers that are 3 or more standard deviations from the mean.
Fluorescence quantile analysis (Supplementary Fig. 4)
Events whose cell areas were 3 standard deviations from the mean were removed. For CD4 and GFP intensities, quantiles were calculated across all samples using the quantile function (Stats, base R). To visualize the data, a density plot was drawn using ggplot2 (geom_density)105 and colors chosen from viridis106. Each event was then categorized by which bin it fell into (for CD4 and GFP respectively). After this, each event had 2 associated values, which quantile of GFP and which quantile of CD4 that it belonged to. For each of the 3 samples (Bcl-xLGFP + VhCD4, VGFP + EpoRhCD4, and Bcl-xLGFP + EpoRhCD4), the mean cell diameter was then calculated within each of these composite bins (i.e., the mean cell diameter in Sample X, for GFP quantile Y and CD4 quantile Z). Next, the ratio of the mean cell diameters within each CD4/GFP bin were calculated between Bcl-xLGFP + VhCD4 and Bcl-xLGFP + EpoRhCD4 using VGFP + EpoRhCD4 as a reference. These data were plotted as a heatmap using ggplot2 and the geom_tile function. To show the distributions of cell diameters for all samples within each composite bin, example density plots were drawn using ggplot2 (geom_density). Example insets were colored by sample, and separate panels were drawn for each composite quantile bin.
Nuclear offset
The intensity weighted delta centroid XY feature was used to measure the distance between the centroid features of two images: CD71 fluorescence for the cell image and DRAQ5 fluorescence for the nucleus. To calculate the cell diameter, first the correlation between CD71 area feature and the bright-field-based area feature was obtained by plotting both values for each event. This allowed us to assign a bright-field area value to each event based on the CD71 area, and then use this value, in combination with the bead calibration curve (see “Calibration of nuclear and cell diameters” above), to calculate cell diameter. Nuclear offset was then calculated by dividing the delta centroid by cell diameter.
Identification of enucleated reticulocytes
Cells were selected by gating on focused, single cell, live, lineage (Gr1, Mac1, CD41, B220, CD3e) negative, hCD4 and GFP positive, and Ter119 positive events. The raw mean pixel intensity of Draq5 (nuclei) was plotted against the total Draq5 intensity (nuclei), giving two clearly distinct populations. We visually confirmed lack of Draq5 signal in the enucleated reticulocyte population.
Cytospins
Cells were spun onto coated ShandonTM Cytoslides (Thermo Scientific) using a ShandonTM Cytospin3 at 800 rpm for 5 minutes. The slides were dried, fixed and stained40. Cytospin preparations were examined using a Zeiss Axioskop 40 microscope using a SPOT Flex Camera (Diagnostic Instruments, Inc.) and imaged using SPOT v.5.6 software (SPOT Imaging).
Cell-cycle analysis
Cell-cycle status and S phase speed were analyzed using BrdU incorporation26. Briefly, cells were pulsed at a final concentration of 33 μM BrdU for 30 min. Cells were immediately labeled with the LIVE/DEAD Kit (Invitrogen L23105), fixed, and permeabilized. Erythroid subsets were identified using anti-CD71 (BD Biosciences 113812) and anti-Ter119 (BD Biosciences 553673). BrdU incorporation was measured by biotin-conjugated anti-BrdU (MOBU-1, BioLegend) followed by a secondary stain with Brilliant Violet 421™ Streptavidin (BioLegend). DNA content was measured by 7AAD (BD Biosciences).
Retroviral Transduction and in vitro differentiation of fetal liver cells
Epor, Bcl-xL, and Tfrc were subcloned into MSCV-IRES-hCD4 retroviral vector. Bcl-xL and Tfrc were also subcloned into MSIG-IRES-GFP retroviral vector (MSIG 1.1 SK). High-titer viral supernatants were prepared by co-transfecting the pCL-Eco packaging vector and desired plasmid into Phoenix cells using Lipofectamine 2000 transfection reagent (ThermoFisher Scientific). High-titer virus was collected in ‘erythroid medium’: IMDM (L-glutamine, 25 mM HEPES) (Gibco), 20% fetal calf serum, 1% penicillin/streptomycin, 10−4 M β-Mercaptoethanol.
Retroviral transduction was done by spin infection of Epor−/− or Epor−/− Hri−/− fetal liver cells at 2000 rpm, 30 °C for 1 h on 50 µg/ml fibronectin (GIBCO) coated dishes in 4 µg/ml polybrene (Sigma), supplemented in some experiments with 0.5 U/ml Epo (Amgen). Transduced cells were incubated for 15 h with 100 ng/ml SCF and 10 ng/ml IL3 (Peprotech). Cells were then transferred to differentiation medium: IMDM (L-glutamine, 25 mM HEPES) (Gibco), 20% fetal calf serum, 1% penicillin/streptomycin, 10−4 M β-Mercaptoethanol, and 0.5 U/ml Epo (Amgen) for the indicated times. In the case of experiments that include Epor−/− Hri−/−, the media was also supplemented with 1 mg/ml iron-saturated human transferrin (Sigma). Where indicated, liquid cultures were also supplemented with Fe-loaded salicylaldehyde isonicotinoyl hydrazone (Fe-SIH, 10 µM, a lipophilic iron chelator, a gift from the late Dr. Prem Ponka (McGill University, Montréal, Québec, Canada), with 0.7 µM deoxyribonucleosides (3′-Deoxythymidine, 2′-Deoxyguanosine monohydrate, 2′-Deoxyadenosine monohydrate, 2′-Deoxycytidine, Sigma),
In vitro differentiation of fetal liver cells with PI3K and MEK1/MEK2 inhibitors
Isolated wild-type S0 cells were cultured in differentiation media (Epo 0.5 U/ml) with 1 µM or 10 µM PI3K inhibitor, LY294002 (EMD Millipore) or MEK1/MEK2 inhibitor, U0126 (EMD Millipore). Inhibitors were replenished every 24 h.
Colony-formation assays in methylcellulose
Retroviral transduction was done by spin infection of Epor−/− fetal liver cells as described above. From each transduced sample (4 h post infection), 200,000 cells were mixed with 1 ml MethoCult (M3234, STEMCELL Technologies) supplemented with 2 U/ml Epo (Amgen). Erythroid (CFU-e) was scored from duplicate plates on day 3 of culture. Expression of hemoglobin in erythroid colonies was confirmed by staining with diaminobenzidine (Sigma) in situ before scoring. Colony area was measured using ImageJ version: 2.0.0-r-54/1.51 h.
Quantitative RT-PCR assay
Total RNA was isolated from in vitro cultured fetal liver cells using the AllPrep DNA/RNA Micro Kit (Qiagen) and quantified by Quant-iT RiboGreen RNA reagent kit (Thermo Scientific) on the 3300 NanoDrop Fluorospectrometer. Reverse transcription was done using the SuperScript III first-strand synthesis system (Invitrogen) with random hexamer primers. Quantitative PCR was performed using the ABI 7300 sequence detection system with TaqMan reagents and TaqMan MGB probes (Applied Biosystems). Each reaction was carried out on a dilution series of the template cDNA to ensure linearity of signal.
TaqMan MGB probes used: β-actin (Mm02619580_g1), PU.1 (Mm00488140_m1), GATA1 (Mm01352636_m1), GATA-2 (Mm00492300_m1), Alas2 (Mm01260713_m1), Band3 (Mm01245920_g1), β-globin (Mm01611268_g1), p21 (Mm00432448_m1), p27 (Mm00438168_m1), p57 (Mm01272135_g1), p16ink4a (Mm01257348_m1), p15ink4b (Mm00483241_m1), p18ink4c (Mm00483243_m1), p19ink4d (Mm00486943_m1), CCND1 (Mm00432359_m1), CCND2 (Mm00438071_m1), CCND3 (Mm01612362_m1), CCNE1 (Mm00432367_m1), CCNE2 (Mm00438077_m1), CCNA2 (Mm00438063_m1), CCNA1 (Mm00432337_m1), E2F2 (Mm00624964_m1), E2F4 (Mm00514160_m1), Tfrc (Mm00441950_m1), Bcl-xL (Mm00437783_m1), DNMT1 (Dnmt100599784), Ifitm1 (Mm01279023_m1), Ifitm3 (Mm00847057_s1), Tal1 (Mm00441665_m1), NFE2 (Mm00801891_m1), LMO2 (Mm00493153_m1), cdk6 (Mm00438163_m1), cdk6 (Mm01311342_m1), cdk4 (Mm00726334_s1), cdk2 (Mm00443947_m1), cdc25a (Mm00483162_m1), cdc25b (Mm00499136_m1), cdc25c (Mm00486880_m1), Klf1 (Mm00516096_m1).
Epo stimulation in vivo
Epo (Epoetin alfa; Amgen) was injected subcutaneously in a total volume of 300 μL in sterile isotonic saline, at the indicated doses and frequencies.
Human intervention studies
Human intervention studies 1 and 3 were performed at the University of Copenhagen. In intervention study 1, subjects received recombinant human erythropoietin (rhEPO). In intervention study 3, participants were subjected to phlebotomy. Thirty-four healthy non-smoking males (n = 19) and females (n = 15) of European descent (age 25 ± 3 years, height 179 ± 10 cm and weight 70 ± 10 kg) participated in the erythropoietin treatment intervention: n = 25 received Epo, n = 9 received Placebo. Another 21 healthy non-smoking male subjects of European descent (age 29 ± 6 years, 184 ± 7 cm, and 77 ± 8 kg) participated in the phlebotomy intervention. No participant had donated blood for at least three months prior to the start of the study or been exposed to high altitude (>1000 m) for at least two months.
The human studies were conducted in Copenhagen, Denmark according to all applicable national and international rules and regulations including the Helsinki II declaration. Ethics approval letters for the studies (protocol numbers H-2-2014-109 & H-17036662, enclosed with the Supplementary Information files; registration number, registration number NCT04227665 for Study #1) were granted by the Regional Branch (Copenhagen Region) of the Danish National Committee on Health Research Ethics (https://en.nvk.dk/). Both studies aim to identify novel biomarkers following either phlebotomy (Study #3) or Epo administration (Study #1). All participants were informed both orally and in writing of potential risks and discomforts associated with participation before written consent was obtained. Participants were compensated for their participation (Study #1: sports equipment equivalent to ~5500 Danish kroner; Study #2, 5000 Danish kroner). Participants were recruited via advertising on social media, dedicated web-pages, and flyers. There is a potential selection bias toward healthier than average participants since the studies examined the effect of Epo on athletic performance. This appears unlikely to influence the results.
Experimental design
rhEPO treatment: The study used a randomized single-blinded placebo-controlled design. After weekly baseline collection of venous blood for 4 weeks, the participants received eleven intravenous injections of 20 IU·kg bw−1 epoetin alpha (Eprex, Janssen, Birkerød, Denmark) (rhEPO group, 25 participants; 13 male and 12 female) or saline (placebo group, 9 participants; 6 male and 3 female) every second day. Venous blood samples were collected weekly during the treatment and for 5 weeks following treatment.
Phlebotomy: The intervention applied a randomized single-blinded placebo-controlled crossover design. The week before phlebotomy, two baseline venous blood samples were collected with 4 days apart. Next, the participants were phlebotomized of two whole-blood units, corresponding to 900 mL or sham-phlebotomized followed by venous blood collection 3, 14, and 25 days later. A recovery period of >4 months was applied before the participants crossed over and repeated the experiment.
Blood sample analysis
All venous blood samples were collected in 2 mL EDTA-anticoagulated vacutainers (Becton Dickinson, New Jersey, USA) after at least 10 min of rest in a seated position and with <30 s use of tourniquet. In the rhEPO trial, samples were immediately analyzed for a complete blood count using a Sysmex XN-450 (Sysmex, Kobe, Japan) including mean cell volume, hemoglobin concentration, reticulocyte count, reticulocyte percentage, and red-cell distribution width. In the phlebotomy trial, samples were stored at 4 °C and analyzed within 2 h of collection for mean cell volume, hemoglobin concentration, reticulocyte count, and reticulocyte percentage using a Sysmex XE-2100 (Sysmex, Kobe, Japan).
Human intervention study 2 was performed at the Centre for Human Drug Research, Leiden, Netherlands. This study was reported elsewhere71, but reporting did not include MCV and RDW information. Briefly, non-professional well trained male cyclists ages 28–50 were randomly assigned to placebo or recombinant human Epo (epoetin β) groups. Baseline measurements were followed by weekly dosing with Epo (24 subjects) or placebo (24 subjects) for 7 weeks. Epo dosing (5000–10,000 IU) was adjusted for each subject, to achieve an increase of 10–15% in hemoglobin over baseline. Follow-up continued for a month after cessation of treatment.
Statistics
For the human studies, we computed baseline-corrected values at each post-baseline time point for each subject by subtracting the corresponding subject-level mean baseline measurement, which was used to fit linear mixed-effect models using the nlme package107. For intervention studies 1 and 2, the model includes subject as random effect, treatment, time, and the interaction of treatment by time as fixed effects. To test whether Epo treatment and placebo differ significantly at each post-baseline time point, a set of pre-defined contrasts were performed using the multcomp package108 followed by multiplicity adjustment using Benjamini–Hochberg procedure109. For intervention study 3, each post-baseline time point was analyzed separately with the model that includes subject as random effect, treatment, period, and sequence of treatments as fixed effects. See Supplementary Information.
For mouse and in vitro experiments, we used both parametric and non-parametric statistical significance tests for sample comparisons as indicated in each figure legend.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
Data availability
Complete blood-count source data for the human studies are provided in the ‘Supplementary statistical analysis of human intervention studies’ in the Supplementary Information file. Additional flow-cytometry data is available upon request. Source data are provided with this paper.
Code availability
The Supplementary MCV simulation python script is deposited in https://github.com/socolovm/Simulation-of-MCV.
References
Kassebaum, N. J. et al. A systematic analysis of global anemia burden from 1990 to 2010. Blood 123, 615–624 (2014).
D’Andrea, A. D., Fasman, G. D. & Lodish, H. F. Erythropoietin receptor and interleukin-2 receptor b chain: a new receptor family. Cell 58, 1023–1024 (1989).
Stephenson, J. R., Axelrad, A. A., McLeod, D. L. & Shreeve, M. M. Induction of colonies of hemoglobin-synthesizing cells by erythropoietin in vitro. Proc. Natl. Acad. Sci. USA 68, 1542–1546 (1971).
Tusi, B. K. et al. Population snapshots predict early haematopoietic and erythroid hierarchies. Nature 555, 54–60 (2018).
Wu, H., Liu, X., Jaenisch, R. & Lodish, H. F. Generation of committed erythroid BFU-E and CFU-E progenitors does not require erythropoietin or the erythropoietin receptor. Cell 83, 59–67 (1995).
Koury, M. J., Bondurant, M. C., Graber, S. E. & Sawyer, S. T. Erythropoietin messenger RNA levels in developing mice and transfer of 125I-erythropoietin by the placenta. J. Clin. Invest. 82, 154–159 (1988).
Koury, M. J. & Bondurant, M. C. Erythropoietin retards DNA breakdown and prevents programmed death in erythroid progenitor cells. Science 248, 378–381 (1990).
Koury, M. J. & Bondurant, M. C. The molecular mechanism of erythropoietin action. Eur. J. Biochem. 210, 649–663 (1992).
Koulnis, M., Porpiglia, E., Hidalgo, D. & Socolovsky, M. In A Systems Biology Approach to Blood Vol. 844 (eds. Corey, S. J., Kimmel, M. & Leonard, J. N.) 37–58 (Springer New York, 2014).
Wickrema, A., Bondurant, M. C. & Krantz, S. B. Abundance and stability of erythropoietin receptor mRNA in mouse erythroid progenitor cells. Blood 78, 2269–2275 (1991).
Broudy, V. C., Lin, N., Brice, M., Nakamoto, B. & Papayannopoulou, T. Erythropoietin receptor characteristics on primary human erythroid cells. Blood 77, 2583–2590 (1991).
Zhang, J., Socolovsky, M., Gross, A. W. & Lodish, H. F. Role of Ras signaling in erythroid differentiation of mouse fetal liver cells: functional analysis by a flow cytometry-based novel culture system. Blood 102, 3938–3946 (2003).
Kieran, M. W., Perkins, A., Orkin, S. & Zon, L. Thrombopoietin rescues in vitro erythroid colony formation from mouse embryos lacking the erythropoietin receptor. Proc. Natl. Acad. Sci. USA 93, 9126–9131 (1996).
Lin, C. S., Lim, S. K., D’Agati, V. & Costantini, F. Differential effects of an erythropoietin receptor gene disruption on primitive and definitive erythropoiesis. Genes Dev. 10, 154–164 (1996).
Iscove, N. N. The role of erythropoietin in regulation of population size and cell cycling of early and late erythroid precursors in mouse bone marrow. Cell Tissue Kinet. 10, 323–334 (1977).
Fang, J. et al. EPO modulation of cell-cycle regulatory genes, and cell division, in primary bone marrow erythroblasts. Blood 110, 2361–2370 (2007).
Ferro, F. Jr., Kozak, S. L., Hoatlin, M. E. & Kabat, D. Cell surface site for mitogenic interaction of erythropoietin receptors with the membrane glycoprotein encoded by Friend erythroleukemia virus. J. Biol. Chem. 268, 5741–5747 (1993).
Spivak, J. L. et al. Cell cycle-specific behavior of erythropoietin. Exp. Hematol. 24, 141–150 (1996).
von Lindern, M. et al. The glucocorticoid receptor cooperates with the erythropoietin receptor and c-Kit to enhance and sustain proliferation of erythroid progenitors in vitro. Blood 94, 550–559 (1999).
Malik, J., Kim, A. R., Tyre, K. A., Cherukuri, A. R. & Palis, J. Erythropoietin critically regulates the terminal maturation of murine and human primitive erythroblasts. Haematologica 98, 1778–1787 (2013).
Socolovsky, M., Dusanter-Fourt, I. & Lodish, H. F. The Prolactin receptor, as well as severly truncated erythropoietin receptors support differentiation of erythroid progenitors. J. Biol. Chem. 272, 14009–14013 (1997).
Socolovsky, M., Fallon, A. E. J. & Lodish, H. F. The prolactin receptor rescues EpoR−/− erythroid progenitors and replaces EpoR in a synergistic interaction with c-kit. Blood 92, 1491–1496 (1998).
Socolovsky, M., Lodish, H. F. & Daley, G. Q. Control of hematopoietic differentiation: lack of specificity in signaling by cytokine receptors. Proc. Natl. Acad. Sci. USA 95, 6573–6575 (1998).
Brisken, C., Socolovsky, M., Lodish, H. F. & Weinberg, R. The signaling domain of the erythropoietin receptor rescues prolactin receptor-mutant mammary epithelium. PNAS 99, 14241–14245 (2002).
Kadri, Z. et al. Phosphatidylinositol 3-kinase/Akt induced by erythropoietin renders the erythroid differentiation factor GATA-1 competent for TIMP-1 gene transactivation. Mol. Cell Biol. 25, 7412–7422 (2005).
Hwang, Y. et al. Global increase in replication fork speed during a p57KIP2-regulated erythroid cell fate switch. Sci. Adv. 3, e1700298 (2017).
Eastman, A. E. et al. Resolving cell cycle speed in one snapshot with a live-cell fluorescent reporter. Cell Rep. 31, 107804 (2020).
Hwang, Y., Hidalgo, D. & Socolovsky, M. The shifting shape and functional specializations of the cell cycle during lineage development. Wiley Interdiscip Rev. Syst. Biol. Med. 13, e1504 (2020).
Humbert, P. O. et al. E2F4 is essential for normal erythrocyte maturation and neonatal viability. Mol. Cell 6, 281–291 (2000).
Sankaran, V. G. et al. Cyclin D3 coordinates the cell cycle during differentiation to regulate erythrocyte size and number. Genes Dev. 26, 2075–2087 (2012).
Jayapal, S. R. et al. Hematopoiesis specific loss of Cdk2 and Cdk4 results in increased erythrocyte size and delayed platelet recovery following stress. Haematologica 100, 431–438 (2015).
Burns, E. R., Reed, L. J. & Wenz, B. Volumetric erythrocyte macrocytosis induced by hydroxyurea. Am. J. Clin. Pathol. 85, 337–341 (1986).
Suragani, R. N. et al. Heme-regulated eIF2alpha kinase activated Atf4 signaling pathway in oxidative stress and erythropoiesis. Blood 119, 5276–5284 (2012).
Chen, J. J. & Zhang, S. Heme-regulated eIF2alpha kinase in erythropoiesis and hemoglobinopathies. Blood 134, 1697–1707 (2019).
Zhang, S. et al. HRI coordinates translation by eIF2alphaP and mTORC1 to mitigate ineffective erythropoiesis in mice during iron deficiency. Blood 131, 450–461 (2018).
Silva, M. et al. Erythropoietin can promote erythroid progenitor survival by repressing apoptosis through Bcl-XL and Bcl-2. Blood 88, 1576–1582 (1996).
Motoyama, N., Kimura, T., Takahashi, T., Watanabe, T. & Nakano, T. bcl-x prevents apoptotic cell death of both primitive and definitive erythrocytes at the end of maturation. J. Exp. Med. 189, 1691–1698 (1999).
Socolovsky, M., Fallon, A. E. J., Wang, S., Brugnara, C. & Lodish, H. F. Fetal anemia and apoptosis of red cell progenitors in Stat5a−/−5b−/− mice: a direct role for Stat5 in bcl-XL induction. Cell 98, 181–191 (1999).
Koulnis, M. et al. Contrasting dynamic responses in vivo of the Bcl-xL and Bim erythropoietic survival pathways. Blood 119, 1228–1239 (2012).
Pop, R. et al. A key commitment step in erythropoiesis is synchronized with the cell cycle clock through mutual inhibition between PU.1 and S-phase progression. PLoS Biol. 8, e1000484 (2010).
von Lindern, M., Schmidt, U. & Beug, H. Control of erythropoiesis by erythropoietin and stem cell factor: a novel role for Bruton’s tyrosine kinase. Cell Cycle 3, 876–879 (2004).
Umemura, T., al-Khatti, A., Donahue, R. E., Papayannopoulou, T. & Stamatoyannopoulos, G. Effects of interleukin-3 and erythropoietin on in vivo erythropoiesis and F-cell formation in primates. Blood 74, 1571–1576 (1989).
Garrick, L. M. et al. Ferric-salicylaldehyde isonicotinoyl hydrazone, a synthetic iron chelate, alleviates defective iron utilization by reticulocytes of the belgrade rat. J. Cell. Physiol. 146, 460–465 (1991).
Nyholm, S. et al. Role of ribonucleotide reductase in inhibition of mammalian cell growth by potent iron chelators. J. Biol. Chem. 268, 26200–26205 (1993).
Eriksson, S., Munch-Petersen, B., Johansson, K. & Ecklund, H. Structure and function of cellular deoxyribonucleoside kinases. Cell. Mol. Life Sci. CMLS 59, 1327–1346 (2002).
Zhu, L. & Skoultchi, A. I. Coordinating cell proliferation and differentiation. Curr. Opin. Genet Dev. 11, 91–97 (2001).
Dalton, S. Linking the cell cycle to cell fate decisions. Trends Cell Biol. 25, 592–600 (2015).
Quelle, F. W. Cytokine signaling to the cell cycle. Immunologic Res. 39, 173–184 (2007).
Khaled, A. R. et al. Cytokine-driven cell cycling is mediated through Cdc25A. J. Cell Biol. 169, 755–763 (2005).
Matsumura, I. et al. Transcriptional regulation of the cyclin D1 promoter by STAT5: its involvement in cytokine-dependent growth of hematopoietic cells. EMBO J. 18, 1367–1377 (1999).
Nagao, T. & Hirokawa, M. Diagnosis and treatment of macrocytic anemias in adults. J. Gen. Fam. Med. 18, 200–204 (2017).
Han, A. P. et al. Heme-regulated eIF2alpha kinase (HRI) is required for translational regulation and survival of erythroid precursors in iron deficiency. EMBO J. 20, 6909–6918 (2001).
Liu, Y. et al. Suppression of Fas-FasL coexpression by erythropoietin mediates erythroblast expansion during the erythropoietic stress response in vivo. Blood 108, 123–133 (2006).
Chen, K. et al. Resolving the distinct stages in erythroid differentiation based on dynamic changes in membrane protein expression during erythropoiesis. Proc. Natl. Acad. Sci. USA 106, 17413–17418 (2009).
Kalfa, T. & McGrath, K. E. Analysis of erythropoiesis using imaging flow cytometry. Methods Mol. Biol. 1698, 175–192 (2018).
McGrath, K. E., Bushnell, T. P. & Palis, J. Multispectral imaging of hematopoietic cells: where flow meets morphology. J. Immunol. Methods 336, 91–97 (2008).
Erslev, A. J., Wilson, J. & Caro, J. Erythropoietin titers in anemic, nonuremic patients. J. Lab Clin. Med. 109, 429–433 (1987).
Kojima, S., Matsuyama, T. & Kodera, Y. Circulating erythropoietin in patients with acquired aplastic anaemia. Acta Haematol. 94, 117–122 (1995).
Dey, S., Curtis, D. J., Jane, S. M. & Brandt, S. J. The TAL1/SCL transcription factor regulates cell cycle progression and proliferation in differentiating murine bone marrow monocyte precursors. Mol. Cell Biol. 30, 2181–2192 (2010).
Chagraoui, H. et al. SCL-mediated regulation of the cell-cycle regulator p21 is critical for murine megakaryopoiesis. Blood 118, 723–735 (2011).
Hsieh, F. F. et al. Cell cycle exit during terminal erythroid differentiation is associated with accumulation of p27(Kip1) and inactivation of cdk2 kinase. Blood 96, 2746–2754 (2000).
Rylski, M. et al. GATA-1-mediated proliferation arrest during erythroid maturation. Mol. Cell Biol. 23, 5031–5042 (2003).
Bouscary, D. et al. Critical role for PI 3-kinase in the control of erythropoietin-induced erythroid progenitor proliferation. Blood 101, 3436–3443 (2003).
Gnanapragasam, M. N. et al. EKLF/KLF1-regulated cell cycle exit is essential for erythroblast enucleation. Blood 128, 1631–1641 (2016).
Vlahos, C. J., Matter, W. F., Hui, K. Y. & Brown, R. F. A specific inhibitor of phosphatidylinositol 3-kinase, 2-(4-morpholinyl)-8-phenyl-4H-1-benzopyran-4-one (LY294002). J. Biol. Chem. 269, 5241–5248 (1994).
Porpiglia, E., Hidalgo, D., Koulnis, M., Tzafriri, A. R. & Socolovsky, M. Stat5 signaling specifies basal versus stress erythropoietic responses through distinct binary and graded dynamic modalities. PLoS Biol. 10, e1001383 (2012).
Kuhrt, D. & Wojchowski, D. M. Emerging EPO and EPO receptor regulators and signal transducers. Blood 125, 3536–3541 (2015).
Lodish, H. F., Ghaffari, S., Socolovsky, M., Tong, W. & Zhang, J. In Erythropoietins, Erythropoietic Factors, and Erythropoiesis: Molecular, Cellular, Preclinical, and Clinical Biology (eds. Elliott, S. G., Foote, M. & Molineux, G.) 155–174 (Birkhäuser, Basel, 2009).
Socolovsky, M. et al. Ineffective erythropoiesis in Stat5a(−/−)5b(−/−) mice due to decreased survival of early erythroblasts. Blood 98, 3261–3273 (2001).
Favata, M. F. et al. Identification of a novel inhibitor of mitogen-activated protein kinase kinase. J. Biol. Chem. 273, 18623–18632 (1998).
Heuberger, J. et al. Effects of erythropoietin on cycling performance of well trained cyclists: a double-blind, randomised, placebo-controlled trial. Lancet Haematol. 4, e374–e386 (2017).
Bosch, F. H. et al. Characteristics of red blood cell populations fractionated with a combination of counterflow centrifugation and Percoll separation. Blood 79, 254–260 (1992).
Willekens, F. L. et al. Hemoglobin loss from erythrocytes in vivo results from spleen-facilitated vesiculation. Blood 101, 747–751 (2003).
Gifford, S. C., Derganc, J., Shevkoplyas, S. S., Yoshida, T. & Bitensky, M. W. A detailed study of time-dependent changes in human red blood cells: from reticulocyte maturation to erythrocyte senescence. Br. J. Haematol. 135, 395–404 (2006).
Franco, R. S. et al. Changes in the properties of normal human red blood cells during in vivo aging. Am. J. Hematol. 88, 44–51 (2013).
d’Onofrio, G. et al. Simultaneous measurement of reticulocyte and red blood cell indices in healthy subjects and patients with microcytic and macrocytic anemia. Blood 85, 818–823 (1995).
Socolovsky, M. et al. Negative autoregulation by FAS mediates robust fetal erythropoiesis. PLoS Biol. 5, e252 (2007).
Thomas, P. Making sense of snapshot data: ergodic principle for clonal cell populations. J. R. Soc. Interface 14, 20170467 (2017).
Shearstone, J. R. et al. Global DNA demethylation during mouse erythropoiesis in vivo. Science 334, 799–802 (2011).
Panzenböck, B., Bartunek, P., Mapara, M. Y. & Zenke, M. Growth and differentiation of human stem cell factor/erythropoietin-dependent erythroid progenitor cells in vitro. Blood 92, 3658–3668 (1998).
Gnanapragasam, M. N. & Bieker, J. J. Orchestration of late events in erythropoiesis by KLF1/EKLF. Curr. Opin. Hematol. 24, 183–190 (2017).
Ginzberg, M. B., Kafri, R. & Kirschner, M. On being the right (cell) size. Science 348, 1245075 (2015).
Björklund, M. Cell size homeostasis: Metabolic control of growth and cell division. Biochimica et. Biophysica Acta (BBA) - Mol. Cell Res. 1866, 409–417 (2019).
Dolznig, H., Grebien, F., Sauer, T., Beug, H. & Müllner, E. W. Evidence for a size-sensing mechanism in animal cells. Nat. Cell Biol. 6, 899–905 (2004).
Narla, A. & Ebert, B. L. Ribosomopathies: human disorders of ribosome dysfunction. Blood 115, 3196–3205 (2010).
Kelley, L. L. et al. Survival or death of individual proerythroblasts results from differing erythropoietin sensitivities: a mechanism for controlled rates of erythrocyte production. Blood 82, 2340–2352 (1993).
Ludwig, L. S. et al. Transcriptional states and chromatin accessibility underlying human erythropoiesis. Cell Rep. 27, 3228–3240.e3227 (2019).
Timmer, T. et al. Associations between single nucleotide polymorphisms and erythrocyte parameters in humans: a systematic literature review. Mutat. Res. 779, 58–67 (2019).
Read, R. W. et al. GWAS and PheWAS of red blood cell components in a Northern Nevadan cohort. PLoS ONE 14, e0218078 (2019).
Seiki, T. et al. Association of genetic polymorphisms with erythrocyte traits: verification of SNPs reported in a previous GWAS in a Japanese population. Gene 642, 172–177 (2018).
Tumburu, L. & Thein, S. L. Genetic control of erythropoiesis. Curr. Opin. Hematol. 24, 173–182 (2017).
Pavlović-Kentera, V., Bogdanović, M., Miladinović, D. & Slavković, V. Erythropoietin level and macrocytosis in patients with chronic pulmonary insufficiency. Respiration 34, 213–219 (1977).
Tsantes, A. E. et al. Red cell macrocytosis in hypoxemic patients with chronic obstructive pulmonary disease. Respir. Med. 98, 1117–1123 (2004).
Chanarin, I., McFadyen, I. R. & Kyle, R. The physiological macrocytosis of pregnancy. Br. J. Obstet. Gynaecol. 84, 504–508 (1977).
Hoffbrand, V. & Provan, D. ABC of clinical haematology. Macrocytic Anaemias Bmj 314, 430–433 (1997).
Yčas, J. W., Horrow, J. C. & Horne, B. D. Persistent increase in red cell size distribution width after acute diseases: a biomarker of hypoxemia? Clin. Chim. Acta 448, 107–117 (2015).
Schepens, T., De Dooy, J. J., Verbrugghe, W. & Jorens, P. G. Red cell distribution width (RDW) as a biomarker for respiratory failure in a pediatric ICU. J. Inflamm. 14, 12 (2017).
Geissler, E. N., McFarland, E. C. & Russell, E. S. Analysis of pleiotropism at the dominant white-spotting (W) locus of the house mouse: a description of ten new W alleles. Genetics 97, 337–361 (1981).
Waskow, C., Terszowski, G., Costa, C., Gassmann, M. & Rodewald, H. R. Rescue of lethal c-KitW/W mice by erythropoietin. Blood 104, 1688–1695 (2004).
Kabaya, K. et al. Improvement of anemia in W/WV mice by recombinant human erythropoietin (rHuEPO) mediated through EPO receptors with lowered affinity. Life Sci. 57, 1067–1076 (1995).
Benesch, R. & Benesch, R. E. The effect of organic phosphates from the human erythrocyte on the allosteric properties of hemoglobin. Biochem Biophys. Res. Commun. 26, 162–167 (1967).
Bunn, H. F. Evolution of mammalian hemoglobin function. Blood 58, 189–197 (1981).
Garby, L. & De Verdier, C. H. Affinity of human hemoglobin A to 2,3-diphosphoglycerate. Effect of hemoglobin concentration and of pH. Scand. J. Clin. Lab Invest. 27, 345–350 (1971).
Hu, X., Eastman, A. E. & Guo, S. Cell cycle dynamics in the reprogramming of cellular identity. FEBS Lett. 593, 2840–2852 (2019).
Wickham, H. ggplot2 Elegant Graphics for Data Analysis (Springer-Verlag, 2016).
Garnier, S. viridis: Default Color Maps from ‘matplotlib’ (2018).
Lindstrom, M. J. & Bates, D. M. Newton-Raphson and EM algorithms for linear mixed-effects models for repeated-measures data. J. Am. Stat. Assoc. 83, 1014–1022 (1988).
Hothorn, T., Bretz, F. & Westfall, P. Simultaneous inference in general parametric models. Biometrical J. 50, 346–363 (2008).
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B 57, 289–300 (1995).
Acknowledgements
The authors would like to thank the UMASS Chan Medical School Flow-cytometry core and Susanne Pechhold for her help with Imaging flow cytometry. Authors of the human intervention studies in Copenhagen (J.B. and N.B.N.) wish to thank all participants of the studies as well as Thomas Christian Bonne and Andreas Breenfeldt Andersen (Department of Nutrition, Exercise and Sports, University of Copenhagen, Denmark), Mikkel Gybel-Brask (Section for Transfusion Medicine, Capital Region Blood Bank, Copenhagen University Hospital, Denmark) and Carl-Christian Howard Kitchen (Department of Anesthesiology, Copenhagen University Hospital, Denmark) for excellent assistance throughout the studies. This work was funded by NIH R01DK100915, R01DK120639 and R01HL141402 (M.S.), R25GM113686 (D.H.), and by R01DK087984 (J.J.C.). J.B. was funded in part by Partnership for Clean Competition and Anti-Doping Denmark. Partnership for Clean Competition and Anti-Doping Denmark funded the phlebotomy trial and World Anti-Doping Agency funded the Copenhagen erythropoietin treatment trial.
Author information
Authors and Affiliations
Contributions
M.S. conceived, designed, and supervised all of the mouse-based experiments and wrote the Supplementary MCV simulation script. N.B.N. conceived, designed, and supervised human studies #1 and #3. D.H., R.P., Y.H., S.M.S., and A.E.E. designed and performed the mouse experiments. J.B. designed and performed human studies #1 . J.A.A.C.H. contributed data from human study #2. K.G. contributed analysis and code for the mouse experiments. S.G. and J.J.C. contributed mouse models and experimental design. L.J.Z. performed statistical analysis of the human studies. M.J.K. conceived mouse experiments. All of the authors contributed to manuscript preparation.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review information
Nature Communications thanks Constance Noguchi and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Hidalgo, D., Bejder, J., Pop, R. et al. EpoR stimulates rapid cycling and larger red cells during mouse and human erythropoiesis. Nat Commun 12, 7334 (2021). https://doi.org/10.1038/s41467-021-27562-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-021-27562-4
- Springer Nature Limited
This article is cited by
-
The role of specialized cell cycles during erythroid lineage development: insights from single-cell RNA sequencing
International Journal of Hematology (2022)
-
The path from stem cells to red blood cells
International Journal of Hematology (2022)