Erythropoietin exposure disrupts B cell development and drives the emergence of myeloid-biased biphenotypic progenitors
Andrada Chiron-Margerie, Stéphanie Bessoles, Guillaume Sarrabayrouse, Kutaiba Alhaj-Hussen, Corentin Joulain, Thomas Darde, Pierre de la Grange, Amine M. Abina, Bruno Canque, Roman Krzysiek, Salima Hacein-Bey-Abina
TL;DR
Erythropoietin treatment disrupts B cell development in mice, leading to the emergence of cells with mixed myeloid and B cell traits.
Contribution
The study reveals a novel effect of EPO on early B cell development, promoting myeloid-like B cell progenitors during stress hematopoiesis.
Findings
EPO promotes a transition from common lymphoid progenitors to pre-pro-B cells but impairs the next developmental stage.
EPO exposure leads to the emergence of atypical B cell precursors expressing myeloid markers.
EPO reprograms B cell precursors by upregulating myeloid genes and downregulating B cell identity genes.
Abstract
Recombinant human erythropoietin (EPO) is widely used to treat anemia. EPO has immunomodulatory effects extending beyond erythropoiesis, but its impact on lympho-hematopoiesis remains insufficiently explored. The objective of this study was to investigate B lymphopoiesis in the context of hyper-EPOemia-induced stress hematopoiesis. Using an EPO supplementation model in C57BL/6 mice, we found that EPO exerts contrasting effects on early B cell development. EPO supplementation promotes the transition from common lymphoid progenitors (CLPs) to pre-pro-B cells but impairs the pre-pro-B to pro-B transition, in part by downregulating interleukin-7 (IL-7) receptor expression. Remarkably, EPO promotes the emergence of atypical B cell precursors, including M-CSFR/CD115-expressing CLPs and CD11b and CD16/32-expressing pre-pro-B cells. Gene expression profiling revealed that EPO reprograms early B…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsErythropoietin and Anemia Treatment · Hematopoietic Stem Cell Transplantation · Platelet Disorders and Treatments
Introduction
Homeostatic hematopoiesis is tightly regulated, yet stress conditions induce cytokine- and infection-driven reprogramming that activates accelerated hematopoietic trajectories, transiently biasing output toward erythroid or myeloid lineages (Swann et al., 2024). Erythropoietin (EPO), like other key growth factors, not only drives erythroid differentiation of committed progenitors but also influences hematopoietic stem and progenitor cell (HSPC) fate (Grover et al., 2014; Mossadegh-Keller et al., 2013; Singh et al., 2018), biasing them toward the erythroid lineage, reshaping clonal output (Eisele et al., 2022; Grover et al., 2014; Singh et al., 2018), and impairing (B) lymphopoiesis (Singbrant et al., 2011). Beyond its role in erythroid development (Grover et al., 2014), EPO also exerts immune-modulatory/immunosuppressive effects, including promotion of regulatory T cells (Cravedi et al., 2014), inhibition of T lymphocyte proliferation and cytokine production (Donadei et al., 2019), inhibition of macrophage activation (Nairz et al., 2011), and reduction of B cell counts in the peripheral blood (Nagashima et al., 2018).
B lymphopoiesis takes place in specialized bone marrow (BM) niches through stepwise differentiation of HSCs into common lymphoid progenitors (CLPs) and then B cell precursors (Hoffmann et al., 2007). Alternative progenitors like common myeloid progenitors (CMPs), c-Kit^−^ CLPs, or Lin^−^Sca1^+^c-Kit^+^Flt3^−^ progenitors (Fossati et al., 2010; Yang et al., 2007) can also generate functionally distinct B cells. Although erythroid/megakaryocyte (Ery/Mk) potential is lost at the lymphoid-biased multipotent progenitor (LMPP) stage, lymphoid and myeloid potentials remain coupled (Adolfsson et al., 2005), with myeloid potential persisting until the pre-pro-B stage (Guo et al., 2018; Rumfelt et al., 2006). Consistent with this view, bipotent B/macrophage progenitors link B lymphoid and macrophage lineages in fetal and adult mouse BM (Audzevich et al., 2017; Zriwil et al., 2016). Additionally, unique CD11b^+^ “myeloid-like” B cells (M-B cells) were recently reported during emergency myelopoiesis (Kanayama et al., 2023). Furthermore, committed B precursors can reprogram to monocytes/macrophages (Chen et al., 2022; Rolink et al., 2000; Xie et al., 2004) or osteoclasts, with EPO enhancing osteoclast reprogramming (Deshet-Unger et al., 2020). Collectively, this highlights a regulatory network connecting erythropoiesis, B lymphopoiesis, and skeletal homeostasis (Deshet-Unger et al., 2020).
However, EPO’s impact on gene regulation during B cell differentiation beyond LMPPs remains largely unexplored (Grover et al., 2014; Singh et al., 2018). Given the complex developmental relationship between B lymphoid and myeloid lineages and the clinical relevance of hyper-EPOemia, this study examines how elevated EPO levels modulate the balance between these lineages. The impact of sustained EPO exposure was analyzed by combining immunophenotypic and gene expression profiling of HSPCs and B cell precursors, tracking differentiation across multiple compartments, including BM, spleen, blood, and peritoneal cavity (PC). Our findings reveal the emergence of B lymphoid progenitors/precursors that co-express myeloid markers, including CD115/macrophage colony-stimulating factor (M-CSF) receptor (M-CSFR), and exhibit a mixed lympho-myeloid gene signature. Furthermore, EPO exposure disrupted lymphopoiesis and myelopoiesis in the BM while promoting extramedullary hematopoiesis (EMH) in the spleen.
Results
Sustained EPO exposure promotes the emergence of myeloid-imprinted CD115+ CLP populations
To investigate EPO’s non-erythropoietic effects, C57BL/6 mice were treated for 2 weeks with recombinant human EPO (rhEPO) or PBS (control). By day 14, EPO-treated mice showed increased hematocrit and slightly reduced BM cellularity (Figures S1A and S1B). Immunophenotypic characterization of BM HSPCs from EPO-treated mice showed expansion of Lin^−^Sca1^+^c-Kit^+^ (LSK, 1.2-fold increase) and Lin^−^Sca1^−^c-Kit^+^ (LK-containing erythro-myeloid progenitors) (Grover et al., 2014) compartments (Figures 1A–1C). The LK compartment increased only in relative proportion not in absolute number. Within the LSK pool, EPO mainly increased the percentages and absolute values of Flt3^−^CD127/IL7Rα^−^ LSK (primarily composed of HSCs), without affecting downstream CD127^+^ or CD127^−^ LMPPs. Classically, LMPPs are defined as Lin^−^Sca-1^+^c-Kit^+^Flt3^hi^ cells that lack surface CD127/IL-7Rα (Adolfsson et al., 2005; Mansson et al., 2008). The earliest CD127^+^ stage with full lymphoid potential corresponds to Ly6D^−^ CLPs, all-lymphoid progenitors (ALP) (Inlay et al., 2009). A minor CD127^+^ LMPP fraction has been described, particularly in neonates, with enhanced capacity for T cell and ILC generation via CLP-independent pathways (Ghaedi et al., 2016). To account for this heterogeneity, LMPPs were stratified into CD127^−^ and CD127^+^ subsets, with “LMPPs” in the main text referring to the CD127^−^ population.Figure 1. Impact of EPO supplementation on hematopoietic stem cells/progenitors in the bone marrow(A) Schematic illustration of B lymphoid ontogeny in the bone marrow based on the literature. LSK: lineage (Lin)^−^Sca1^+^c-Kit^+^ progenitors; HSC, hematopoietic stem cell; MPP, multipotent progenitor; LMPP, lymphoid-primed multipotent progenitor; CLP, common lymphoid progenitor; ALP, all-lymphoid progenitor; BLP, B-cell-biased lymphoid progenitor (BLP1–BLP3). Created with BioRender.com.(B) Representative zebra plots showing various bone marrow stem cell/progenitor populations: LSK (Lin^−^Sca1^+^c-Kit^+^), Flt3^−^CD127^−^ LSK fraction, LK (Lin^−^Sca1^−^c-Kit^+^), Lin^−^Sca1^+^c-Kit^−^, LMPP (CD127^+^ and CD127^−^ fraction), CLP, ALP (Ly6D^−^ CLP fraction), and BLP (Ly6D^+^ CLP fraction) in EPO-treated (red) vs. PBS-treated mice (black). ALP and BLP fractions were defined based on Ly6D expression, according to Inlay et al. (2009). Ly6D fractions of BLP were subdivided to identify Ly6D^hi^ fractions. Zebra plots also depict “unconventional” c-Kit^low^Sca1^low^Flt3^+^CD127^−^ “CLP-like” progenitors. Lineage negative cells were delineated using FITC-coupled antibodies against CD3, CD11c, CD19, Ly6G, NK1.1, and TER119 and BV650-coupled anti-CD11b.(C–E) Scatterplots showing the proportions and absolute cell numbers of bone marrow stem cell/progenitor populations of LSK, LK, Lin^−^Sca1^+^c-Kit^−^, and c-Kit^low^Sca1^low^ cells (C); Flt3^−^CD127^−^ LSK fraction, LMPP CD127^−^ and CD127^+^ fractions, CLP, and c-Kit^low^Sca1^low^Flt3^+^CD127^− “^CLP-like” progenitors (D); and ALP (Ly6D^−^ CLP fraction) and BLP (total Ly6D^+^ CLP fraction, including both Ly6D^low^ and Ly6D^hi^ fractions) (E) in EPO vs. PBS (control)-treated mice. Graphs depict cell percentages among lineage-negative (Lin^−^) bone marrow cells.(F) Scatterplot showing the ratio of BLP to ALP progenitors in EPO-treated (red) vs. PBS-treated (gray) mice.(G) Scatterplots showing the proportions and absolute cell numbers of bone marrow Ly6D^hi^ and Ly6D^low^ BLPs in EPO-treated (red) vs. PBS-treated (gray) mice. Graphs depict cell percentages among lineage-negative (Lin^−^) bone marrow cells.(H) Representative zebra plots showing CD115 expression among CLP, ALP, BLP, and c-Kit^low^Sca1^low^ Flt3^+^ CD127^−^ “CLP-like” cells in the bone marrow of EPO-treated (red) compared to PBS-treated (black) mice.(I) Scatterplots illustrating the proportions and absolute cell numbers of CD115-expressing bone marrow lymphoid progenitors, as shown in (H). (n = 25 mice per group, pooled from five independent experiments).Gating strategy is depicted in supplemental methods. For all graphs, horizontal bars represent the mean and standard deviation (SD); ^∗^p < 0.05; ^∗∗^p < 0.01; ^∗∗∗^p < 0.001.
EPO exposure increased the c-Kit^low^Sca1^low^ fraction, including CLPs (defined as c-Kit^low^Sca1^low^Flt3^+^CD127^+^ cells, 1.6-fold increase) as well as an unconventional c-Kit^low^Sca1^low^ Flt3^+^CD127^−^ “CLP-like” population (Figures 1A–1D). The CLP stage phenotypically resembles LMPPs but exhibits reduced Sca-1 and c-Kit and robust interleukin-7Rα (IL-7Rα) expression (Mansson et al., 2008). The “CLP-like” population shares key phenotypic features with classical CLPs but lacks IL-7Rα/CD127 expression. Both Ly6D^−^ ALPs and Ly6D^+^ B-cell-biased lymphoid progenitors (BLPs) (Inlay et al., 2009) increased, whereas the BLP to ALP ratio remained unchanged (Figures 1E and 1F), suggesting that EPO exposure does not alter the balance between B and T/NK/ILC progenitors. Notably, the Ly6D^hi^ BLP fraction showed a significant increase in EPO-treated mice (2-fold increase; Figure 1G), a previously unreported finding.
EPO exposure also decreased c-Kit and Flt3 receptor expression (geometric mean fluorescence intensity, gMFI) in LSK, LMPPs, and CLPs (ALPs/BLPs) (Figures S1C–S1F), without affecting Sca1 and CD127 expression. Expression of Ly6D, an early B cell specification marker (Inlay et al., 2009), significantly increased in EPO-treated mice, distinguishing Ly6D^low^ and Ly6D^hi^ BLP fractions (Figures S1G and S1H). These findings suggest that hyper-EPOemia might facilitate the transition from LMPPs to BLPs.
Subsequent analyses revealed that EPO exposure induced atypical expression of the monocyte/macrophage differentiation receptor CD115/M-CSFR/CSF1R on CLPs, ALPs, BLPs, and c-Kit^low^Sca1^low^Flt3^+^CD127^−^ cells (Figure 1H). Furthermore, increased CD115 expression and CD115-to-Flt3 and CD115-to-CD127 gMFI ratios in CLPs suggest myeloid imprinting (Figures S1I–S1L). Increased lymphoid-to-myeloid progenitor ratio indicates in addition that EPO favors CLP generation (Figure S1M). The minimal CD115 expression observed in B cell progenitors and LMPPs across both EPO-treated and control groups (data not shown) suggests that aberrant EPO-induced CD115 acquisition is specific to CLPs. Overall, our findings suggest that hyper-EPOemia reshapes early B lymphopoiesis, generating a distinct population of myeloid-imprinted CD115^+^ CLPs.
EPO impairs B cell development during pre-pro-B to pro-B transition
To investigate EPO’s impact on early B lymphopoiesis, we next analyzed progression from pre-pro-B to mature B cells. EPO-treated mice showed significant increase of the B220^+^CD19^−^ pre-pro-B fraction, with a “pre-pro-B-like” B220^+/low^CD19^−^c-Kit^+^CD43^+/hi^ (CD11c^−^DX5^−^) population displaying aberrant CD127^low/−^ phenotype (Figures 2A–2C and 2F). The term “pre-pro-B-like” denotes a progenitor population phenotypically related to classical pre-pro-B cells but lacking normal IL-7Rα/CD127 expression. Because IL-7R is a key marker and driver of early B-lineage commitment, its absence is notable, as IL-7- or IL-7R-deficient mice fail to progress beyond the pre-pro-B stage (Kikuchi et al., 2008; Morgan and Tergaonkar, 2022). Nonetheless, substantial heterogeneity exists within the CLP/Pre-Pro-B compartment (Chiron-Margerie et al., 2025), and CD127 is not universally used to define these cells (Hardy et al., 1991; Rolink et al., 1994; Singbrant et al., 2011).Figure 2. Impact of EPO supplementation on B lymphoid and biphenotypic B lymphoid/myeloid precursors in the bone marrow(A) Representative zebra plots displaying the identification of total B lymphoid cells and progenitors (B220^+^CD19^−^ and B220^+^CD19^+^) in the bone marrow of PBS (black) and EPO (red)-exposed mice.(B and C) Scatterplots illustrating the frequencies and absolute cell numbers of B220^+^CD19^−^ cells (including pre-pro-B precursors) (B) and of B220^+^CD19^+^ cells (C) in the bone marrow of PBS- (black) and EPO (red)-exposed mice.(D) Representative zebra plots displaying the identification of pre-pro-B, pro-B, pre-BI, pre-BIIL, and pre-BIIS lymphoid precursors, along with immature and mature B cells in the bone marrow of PBS- (black) and EPO (red)-exposed mice.(E and F) Scatterplots illustrating the frequencies and absolute cell numbers of pre-pro-B, pro-B, pre-B precursors, immature, and mature B cells (E) and B220^+^CD43^+^CD127^−^ “pre-pro-B-like” cells (F) in EPO-treated (red) vs. PBS (gray)-treated mice. “Pre-pro-B-like” precursors exhibited a phenotype similar to classical pre-pro-B cells (CD11c^−^DX5^−^B220^+^CD19^−^CD43^+^c-Kit^+^), but with low or absent CD127 expression (CD127^low/–^). Graphs show cell percentages among total live bone marrow cells (n = 25 mice/group, obtained from five independent experiments).(G) Representative zebra plots showing the frequencies of pre-pro-B and pro-B+pre-BI cells co-expressing CD11b and CD16/32 (biphenotypic B/myeloid progenitors). Lineage markers used included FITC-coupled antibodies against CD3, IgD, IgM, Ly6G, TER119, and NK1.1, as well as BV785-coupled anti-CD11c.(H) Mean frequencies (within Lin⁻CD93⁺ cells) and absolute cell counts of CD11b⁺CD16/32⁺ pre–pro-B cells (biphenotypic B/myeloid pre–pro-B cells) in EPO (red) compared with PBS (gray) mice; (n = 20 mice per group, pooled from five independent experiments).(I) Scatterplot showing gMFI of CD16/32 in CD16/32^+^CD11b^+^ pre-pro-B precursors from EPO- (red) and PBS (gray)-exposed mice (n = 8 mice per group, obtained from three independent experiments).CD16/32 = FcγR receptor.Cells were identified using gating strategies shown in supplemental methods.For all graphs, horizontal bars represent the mean and standard deviation (SD); ^∗^p < 0.05, ^∗∗^p < 0.01, ^∗∗∗^p < 0.001, ^∗∗∗∗^p < 0.0001.
However, EPO significantly decreased the total B220^+^CD19^+^ fraction, selectively impairing all downstream B cell differentiation stages (Figures 2A–2E). Consistent with this, pre-pro-B, pro-B, and pre-B cells from EPO-treated mice downregulated surface CD127/IL7Rα and CD25/IL2Rα, while upregulating c-Kit and/or B220 markers (Figures S1N–S1P). These findings suggest EPO exposure might inhibit B lymphopoiesis by deregulating key maturation receptors.
To investigate EPO-induced myeloid imprinting in B precursors, we analyzed biphenotypic B/macrophage and B/osteoclast Pro-B precursors as well as B1-specified (B1P) progenitors that generate B1 cells (Audzevich et al., 2017; Blin-Wakkach et al., 2004; Deshet-Unger et al., 2020). EPO exposure did not expand pro-B, pre-BI, or pre-BII cells expressing myeloid markers (CD11b, CD16/32, F4/80, and CD115) or B1P progenitors (Figures S2A and S2B). Interestingly, EPO exposure led to a 2-fold increase in the proportion of CD11b^+^CD16/32^+^ pre-pro-B cells (Figures 2G and 2H), displaying elevated CD16/32 expression (Figure 2I), suggesting potential myeloid rewiring of early B cell precursors (Audzevich et al., 2017).
EPO impairs monocyte lineage differentiation
Observed expansion of biphenotypic CLP and pre-pro-B precursors prompted us to examine lymphoid-to-myeloid output balance. Consistent with previous findings (Grover et al., 2014; Singh et al., 2018), EPO-treated mice exhibited reduced myeloid progenitor fractions (Flt3^+^CMP, GMP, MDP, GP, and MP/cMoP) and CD115^+^/M-CSFR monocyte lineage populations, especially TrpMono (precursors of Ly6C^+^/Ly6C^−^ monocytes), with significant decreases in absolute cell numbers for Flt3^+^CMP, MDP, MP/cMoP, TrpMono, and Ly6C^−^ monocytes (Figure 3). Additionally, c-Kit and CD115 expression decreased at all myeloid/monocytic stages, whereas CD11b selectively decreased in TrpMono and mature monocytes (Figures S2C–S2I), highlighting EPO’s negative effect on monocyte differentiation in the BM.Figure 3. Impact of EPO supplementation on myeloid progenitors in the bone marrow(A) Schematic illustration of myeloid ontogeny in the bone marrow, based on Yáñez et al. (2017) and Chong et al. (2016). The illustration shows hematopoietic stem cells (HSCs), multipotent progenitors (MPPs), common myeloid progenitors (CMPs), granulocyte-monocyte progenitors (GMPs), monocyte-dendritic progenitors (MDPs), granulocyte progenitors (GPs), monocyte progenitors (MPs), common monocyte progenitors (cMoPs), common dendritic progenitors (CDPs), transitional premonocytes (TrpMono), Ly6C^hi^ and Ly6C^−^ monocytes (Mono), neutrophils (Neutro), and dendritic cells (DCs). Analyzed populations are highlighted in red. Created with BioRender.com.(B) Representative zebra plots showing bone marrow myeloid progenitor populations: CMP (Flt3^+^ and Flt3^−^), GMP, MDP, GP, and MP + cMoP in EPO-treated (red) vs. PBS-treated (black) mice. Lineage markers included FITC-coupled antibodies against CD3, CD11c, CD19, Ly6G, NK1.1, and TER119, as well as BV650-coupled anti-CD11b.(C) The scatterplots on the right show the mean frequencies and the absolute cell numbers of myeloid progenitors among lineage-negative (Lin−) bone marrow cells.(D) Representative zebra plots showing frequencies of Lin^−^CD115^+^ cells, transitional premonocytes (TrpMono, immediate precursors of Ly6C^+^ monocytes), and Ly6C^+^ and Ly6C^−^ mature monocytes (Mono) in the bone marrow of PBS-treated (black) and EPO-treated (red) mice. Lineage markers included FITC-coupled antibodies against CD3, CD11c, CD19, Ly6G, NK1.1, and TER119.(E and F) Frequencies (among total live bone marrow cells) and absolute cell numbers of CD115^+^(Lin^−^) cells (E) and of TrpMono and mature Ly6C^+^ and Ly6C^−^ Mono (F), in PBS-treated vs. EPO-treated mice. n = 8 mice per group, obtained from three independent experiments.CD16/32 = FcγR receptor.Cells were identified using gating strategy from supplemental methods.For all graphs, horizontal bars represent the mean and standard deviation (SD); ^∗^p < 0.05; ^∗∗^p < 0.01.
EPO exposure induces extensive extramedullary B lymphopoiesis and monopoiesis
Given similarities between our hyper-EPOemia model and stress-erythropoiesis models featuring EMH (Perry et al., 2009), we next examined the effect of sustained EPO exposure on spleen progenitor compartments. Unlike the BM, spleen from EPO-treated mice showed increased cellularity (Figures S3A and S3B). Phenotypic assessment revealed significant increases in LSK, Flt3^−^CD127^−^ LSK, LK, Lin^−^Sca1^+^c-Kit^−^, and c-Kit^low^Sca1^low^ fractions, in the c-Kit^low^Sca1^low^ Flt3^+^CD127^−^ “CLP-like” population as well as enriched LMPP and CLP progenitors (2- to 2.6-fold increase in EPO vs. control), without an imbalance between BLP and ALP (Figures 4A–4E).Figure 4. Impact of EPO supplementation on hematopoietic stem cells/progenitors (HSPCs) and lymphoid progenitors in the spleen(A) Representative zebra plots showing spleen stem cell/progenitor populations of LSK (lineage (Lin)^−^Sca1^+^c-Kit^+^), Flt3^−^CD127^−^ LSK fraction, LK (Lin^−^Sca1^−^c-Kit^+^), Lin^−^Sca1^+^c-Kit^−^, LMPP (and CD127^+^ LMPP fraction), CLP, and the respective ALP (Ly6D^−^ CLP fraction) and BLP (Ly6D^+^ CLP fraction) progenitors in EPO-treated (red) vs. PBS-treated mice (black). ALP and BLP fractions were defined based on Ly6D expression, according to Inlay et al. (2009). Lineage markers included FITC-coupled antibodies against CD3, CD11c, CD19, Ly6G, NK1.1, and TER119, as well as BV650-coupled anti-CD11b.(B–D) Scatterplots illustrating the proportions and absolute cell numbers of spleen stem cell/progenitor populations of LSK, LK, Lin-Sca1+c-Kit−, and c-Kit^low^Sca1^low^cells (B); Flt3^−^CD127^−^ LSK fraction, CD127^−^ and CD127^+^ LMPP fractions, CLP and c-kit^low^Flt3^+^CD127^- "^CLP-like cells" (C); and of ALP and BLP (D) in PBS- (gray) and EPO (red)-treated mice; n = 20 mice in each group, obtained from five independent experiments.(E) Scatterplot showing the ratio of BLP to ALP progenitors in EPO- (red) vs. PBS (gray)-treated mice.(F) Representative zebra plots showing the identification of pre-pro-B, pro-B + pre-BI, pre-BIIL, and pre-BIIS precursor cells in the spleen of PBS- (black) and EPO (red)-treated mice. Lineage markers included FITC-coupled antibodies against CD3, CD5, IgKappa, IgLambda, and Ly6G, as well as BV650-coupled anti-CD11b.(G and H) Scatterplots showing frequencies and absolute cell numbers of CD93^+^Lin^−^ progenitors (containing B1 and B2 progenitors) (G) and of various stages of B precursors (H) in the spleen of PBS-treated (gray), compared to EPO-treated (red) mice; n = 25 mice in each group, obtained from five independent experiments. Plots show frequency among total live cells of the spleen.Gating strategies are depicted in supplemental methods.For all graphs, horizontal bars indicate the mean and standard deviation (SD). ^∗^p < 0.05; ^∗∗^p < 0.01; ^∗∗∗^p < 0.001; ^∗∗∗∗^p < 0.0001.
EPO exposure also increased the CD93/AA4.1^+^ Lin^−^ fraction containing B1/B2 progenitors, enhanced B cell development from pre-pro-B to pre-BIIS stages (1.2- to 1.5-fold increase) (Figures 4F–4H) and increased CD127^+^ B1Ps and CD19^+^B220^−/low^CD43^+^c-Kit^+^CD127^+^ B1Ps (Figures S2J and S2K) (Ghosn et al., 2011). In contrast to the BM, splenic B-lymphoid progenitors did not upregulate myeloid markers. Interestingly, exposure to EPO increased the numbers of total B cells (B220^+^CD19^+^), type II follicular (Fol II) B2 cells, marginal zone (MZ) B cells, MZ progenitor (MZP) cells, M-B-like cells, as well as atypical IgM^+^IgD^−^ and IgM^−^IgD^−^ CD21^−^CD23^low/−^CD11b^+^ B cells (Figures 5A, 5B, and S3C–S3J and data not shown). Notably, spleen CD19^+^B220^+^CD23^low/−^CD21^−^ B cells are phenotypically similar to age-associated B cells (ABCs), which are expanded in autoimmune diseases and viral infections (Mouat et al., 2022; Woodruff et al., 2020).Figure 5. Impact of EPO supplementation on B lymphoid cells, myeloid progenitors, and monocytes in the spleen(A) Representative zebra plots showing spleen M-B-like cells (defined as CD19^+^B220^+^CD11b^+^IgD^+^IgM^^low/-^^CD5^−^ cells) and B1 cells (defined as CD19^+^B220^+^CD11b^+^IgM^+^CD5^+/−^) in PBS- (black) and EPO (red)-exposed mice. Right side scatterplots depict frequencies and cell numbers of spleen M-B like cells in EPO (red) compared to PBS (gray) mice (n = 12 mice/group, obtained from three independent experiments).(B) Histograms showing geometric mean intensity (gMFI) of CD11b, CD21, CD23, IgM, and IgD in M-B-like cells from PBS- (gray) and EPO (red)-treated mice, compared to gMFI in B1 cells.(C) Representative zebra plots showing spleen myeloid progenitors: common myeloid progenitor (CMP, Flt3^+^, and Flt3^−^), granulocyte-monocyte progenitor (GMP), monocyte-dendritic progenitor (MDP), granulocyte progenitor (GP), and monocyte progenitor + common monocyte progenitor (MP + cMoP) in EPO-treated (red) vs. PBS-treated (black) mice. The lineage markers included FITC-coupled antibodies against CD3, CD11c, CD19, Ly6G, NK1.1, and TER119, as well as BV650-coupled anti-CD11b.(D) Scatterplot depicting the mean frequencies of lineage-negative (Lin^−^) c-Kit^+^CD16/32^−^ cells (including erythroid progenitors) among total live cells in EPO-treated (red) versus PBS-treated (gray) mice.(E) Scatterplot showing the mean frequencies of myeloid progenitors in the spleen, among total live cells in EPO-treated (red) versus PBS-treated (gray) mice; n = 22 mice per group, obtained from five independent experiments.(F and G) Representative zebra plots showing the frequencies of Lin^−^CD115^+^ cells, transitional premonocytes (TrpMono), and Ly6C^+^ and Ly6C^−^ mature monocytes (Mono) in the spleen of PBS-treated (gray) versus EPO-treated (red) mice. The lineage markers included FITC-coupled antibodies against CD3, CD11c, CD19, Ly6G, NK1.1, and TER119. The right side scatterplots (G) show the frequencies of TrpMono and mature Ly6C^+^ and Ly6C^−^ monocytes among total live cells; n = 22 mice per group, obtained from five independent experiments.Gating strategies are depicted in supplemental methods.For all graphs, horizontal bars indicate the mean and standard deviation (SD). ^∗^p < 0.05; ^∗∗^p < 0.01; ^∗∗∗^p < 0.001.
Regarding the myeloid lineage, we observed increases in GMP, GP, MP/cMoP, and TrpMono fractions (1.8-, 2.5-, 4-, and 2.3-fold increase, respectively) and c-Kit^+^CD16/32^−^ progenitors (including erythroid progenitors) (Grover et al., 2014), with no effect on MDP or mature monocytes (Figures 5C–5G). The CLP-to-GMP ratio was unchanged (data not shown).
Similar to the BM, EPO also decreased the expression of c-Kit in Flt3^+^CMPs and GMPs and of CD11b in TrpMono and monocytes (Figures S3K and S3L). Collectively, these findings suggest that EPO supports stress erythropoiesis while also promoting spleen extramedullary B lymphopoiesis and monopoiesis.
EPO induces blood Ly6C+ monocytosis and B lymphocytosis, without altering peritoneal cavity composition
Furthermore, EPO exposure induced moderate Ly6C^+^ monocytosis and B lymphocytosis, while white blood cell and neutrophil counts remained unchanged (Figures S3M–S3P). EPO-exposed mice showed increased counts of circulating Fol II B cells and CD21^−^CD23^−^CD11b^+^IgM^+^ atypical B cells, similar to the spleen, whereas circulating M-B-like cells were slightly increased (Figures S3Q–S3S).
Since innate B1 cells express the myeloid marker CD11b (Ghosn et al., 2011; Kanayama et al., 2023), we assessed EPO’s impact on B1 and myeloid populations in the PC. No significant changes were observed in innate B1 (B1a and B1b), follicular B2, or myeloid populations, including monocytes and macrophages (large and small peritoneal macrophages), indicating hyper-EPOemia does not affect PC-associated B and myeloid cells in our model (data not shown).
Exposure to EPO induces myeloid lineage imprinting at the molecular level in B-cell-committed precursors
To explore underlying mechanisms of impaired B lymphopoiesis in EPO-exposed mice, we performed expression profiling of FACS-sorted B lymphoid precursors (pro-B to pre-BIIS) using microarray. Gene set variation analysis (GSVA) confirmed stage-specific signatures (Figures S4A and S4B). Gene set enrichment analysis (GSEA) based on log_2_ fold-change gene rank revealed that EPO exposure downregulated stage-specific signatures in pro-B and pre-BI precursors (Figures 6A–6C), affecting expression of genes regulated by key transcription factors (TFs) IRF4, FOXO3, FOXO1, ERG, and IRF8 (Figure S4C). EPO-exposed Pre-BIIL precursors displayed downregulation of PAX5 targets and overexpression of PAX5-repressed genes (Figure 6D) (Delogu et al., 2006; Pridans et al., 2008; Schebesta et al., 2007). TF network analysis of EPO-exposed B cell precursors showed that downregulation of B cell signatures coincided with overexpression of myeloid-related TFs Gfi1, Spi1/PU.1, *Cebpα/*C/EBPα, Cebpβ/C/EBPβ, and *Cebpε/*C/EBPε target genes, suggesting a potential module driving biphenotypic B/myeloid stages (Figure S4C). Low PAX5 expression promotes biphenotypic differentiation with selective myeloid gene expression, including Cebpα, Pu.1, and Csf1r in pro-/pre-B cells (Simmons et al., 2012). Additionally, erythroid-development-related genes targeted by Gata1 and Tal1 were also overexpressed (data not shown).Figure 6. Microarray analysis of FACS-sorted B-lymphoid-committed precursors of the bone marrow(A–C) Gene set enrichment analysis (GSEA) from log2fold-change gene rank in EPO-treated versus PBS-treated (control) mice using publicly available MSigDB collections. These collections represent stage-specific gene signatures of human and mouse B cell precursors. The title of each GSEA figure corresponds to the B precursor population of interest. NES, normalized enrichment score. Legends indicate the MSigDB signature used (depicted in Figure S4B).(D) GSEA analysis of PAX5 target genes, categorizing genes as PAX5-activated (target genes) or PAX5-repressed genes. This is based on studies by Delogu et al. (2006), Pridans et al. (2008), and Schebesta et al. (2007).(E) GSEA in EPO-treated versus control mice using pre-granulocyte-macrophage (pre-GM) progenitor and pre-megakaryocyte-erythrocyte (pre-MegE) progenitor cell signatures from Pronk et al. (2007) (GEO: [GSE8407](GSE8407)). The title of each GSEA figure corresponds to the B precursor population of interest. Signature names are shown in green. NES, normalized enrichment score.(F) Heatmap displaying Z scores from log2 normalized expression of selected myeloid genes representing monocyte, neutrophil, eosinophil, and mast cell lineages in FACS-sorted B precursors (pro-B to pre-BIIS). These genes are characteristic of atypical monocytes (segregated-nucleus-containing atypical monocytes or SatM) from Satoh et al. and Nature, 2017.(G) Heatmap showing Z scores from log2 normalized expression of selected genes representing myeloid-like B cells (M-B cells) from Kanayama et al. (2023) in FACS-sorted B precursors (pro-B to pre-BIIS).All GSEA/enrichment results shown have a false discovery rate (FDR)-adjusted p value (q value) < 0.05. Each FACS-sorted progenitor population has n = 2 replicates. Cells were sorted using the gating strategy depicted in supplemental methods.
GSEA showed positive enrichment for pre-GM, pre-MegE, and “myeloid-like” B (M-B) cell signatures in EPO-exposed pre-BII precursors (Figures 6E and S4D). EPO increased the expression of myeloid differentiation genes (*Csf1r/*M-CSFR, Lyz1/2, Mpo, Elane, etc.) (Satoh et al., 2017) (Figure 6F) and of M-B cell signature genes (Ly6c2, Ccr2, Ngp, *Fcgr3/*CD16, Ly6g, Itgam/CD11b, etc.) (Kanayama et al., 2023) (Figure 6G), suggesting reprogramming of early B-precursors toward a myeloid lineage-imprinted state, resembling M-B cells (Kanayama et al., 2023).
Gene ontology (GO) and pathway enrichment analysis confirmed myeloid differentiation, upregulation of cell migration, adhesion, ROS metabolism, angiogenesis, heme scavenging, inflammation (tumor necrosis factor [TNF] signaling), hypoxia response (HIF1A targets), clonal expansion (MYC pathway), and metabolic reprogramming (mTORC1 signaling, oxidative phosphorylation). Pro-B and pre-BIIS stages showed decreased proliferation, while pre-BI cells upregulated proliferation-associated genes (Figures S5A–S5C). Transcriptional priming, driven by key TFs, shapes lineage potential while bifurcation is guided by antagonistic TF interactions (Huang et al., 2007). To provide a comprehensive picture of the effect of EPO exposure on HSPC priming and B-lineage commitment, we extended our analysis on FACS-sorted BM HSPCs/precursors, including Flt3^−^CD127^−^ LSKs and B-lineage stages from LMPP to immature B cells. We used a recently developed mini-RNA-seq approach (Alhaj Hussen et al., 2023) and improved statistical robustness using five replicates per condition. GMPs were analyzed to assess potential EPO-induced effects and served as a control for the myeloid signature (Figure S6A).
Uniform manifold approximation and projection (UMAP) analysis and lineage-specific signatures validated transcriptional profiles (Table S1; Figures S6B and S6C). To assess EPO-induced changes, we evaluated enrichment of DEGs using published profile data (Pronk et al., 2007) and MsigDB gene sets.
EPO exposure was associated with upregulated myeloid signatures and downregulated lymphoid signatures in lymphoid precursors (Figure S7; Table S2). Corroborating GeneChip data, EPO exposure upregulated myeloid markers at all B lymphoid developmental stages: (1) pre-pro-B cells upregulated markers like Anxa1, Anxa2, S100a6, Ms4a6b, and Fcgr3/CD16, consistent with higher proportions of CD11b^+^CD16/32^+^ cells; (2) pro-B cells upregulated markers like Mybl2, Sod2, Siglech, Junb, and Gfi1; (3) pre-BI cells upregulated Ms4a3, Hdc, Ctsc, Steap4, Vamp1, and Ncf4; (4) pre-BII cells upregulated genes Epha2, Myadm, Maf, Cd209b, Adam19, and Vsir; and (5) immature B cells Trim33, Ncf1, Creg1, Ctsh, Pou2f2, and Ms4a6b. Pre-BII cells also upregulated genes related to Ery/Mk differentiation and stress hematopoiesis (Trib2, Blvrb, and Pdgfrb), and Pre-BIIL cells overexpressed the B cell inhibitor Id3 (Jaleco et al., 1999). Trib2 notably regulates myeloid commitment factors C/EBPα and C/EBPβ (Fang et al., 2021) (Figures 7A and 7E–7J; Table S3).Figure 7. Mini-RNA-seq expression profiling of FACS-sorted progenitors/precursors in the bone marrow(A) Heatmap illustrating Log2 fold changes (FCs) of 1,842 differentially expressed genes (DEGs) between EPO- and control (PBS-treated) groups across progenitor/precursor populations. Each population includes three to five replicates. The heatmap was generated using the k-means clustering method. The left-side table displays representative genes from all clusters (C1–C20).(B–J) Heatmaps showing log2 normalized expression of selected myeloid and lymphoid genes that are differentially regulated between EPO- and PBS-exposed mice. Each heatmap corresponds to a specific FACS-sorted hematopoietic progenitor/precursor population: (B) LSK; (C) LMPP; (D) CLP; (E) pre-pro-B; (F) pro-B; (G) pre-BI, (H) pre-BIIL; (I) pre-BIIS; and (J) immature B cells. Gating strategy for progenitor/precursor sorting is depicted in supplemental methods.
EPO-induced transcriptional rewiring in B cell precursors was marked by deregulated expression of key genes throughout the B-lineage trajectory: (1) pre-pro-B cells showed altered lymphoid gene expression with overexpression of Dok3, Sdc4, Prkd2, and Nedd9 and downregulation of crucial factors like Fos (AP-1 family TF), Fcrl1 (BCR co-receptor), Bach1 (transcription repressor inhibiting the myeloid program to support B lymphopoiesis), and Gfra2 (BLP marker) (Itoh-Nakadai et al., 2014); (2) pro-B cells upregulated B-lymphoid genes such as Bach1, Bst1 (B cell commitment), Btg1, Btg2, Malt1 (BCR signaling), and the pre-B marker *Il2ra/*CD25, while downregulating TFs associated with lymphoid fate Fos, Runx3, Foxo3, Bcl11a, and BCR-signaling component Fyn; (3) pre-BI cells downregulated lymphoid genes like Erg, Irf8, JunD, and Lck; (4) pre-BIIL cells upregulated key B-lymphoid genes (Cd19, *Ms4a1/*CD20, Cd79a, Pax5, Rag1, Cd74, and Cd2) and S1pr1, which supports BM egress; and (5) pre-BIIS showed increased expression of the B-lymphoid TF Spib and lineage-associated genes Dok3, Cd2, and Bach1. The pre-BII stage upregulated early lymphoid-stage-associated genes (Cnn3, Bst1, Sesn1, Slamf6, Ephb2, St6gal1, Gpat3) but downregulated critical B cell markers/TFs like Foxm1, Irf4, Malt1, Cd44, and Ctcf (in pre-BIIL) and *Il2ra/*CD25 *(*confirming decreased expression in phenotypic study), Vpreb1 pre-BCR component, Tnfrsf13c (BAFFR), Sdc4, and Prkd2 (in pre-BIIS). Immature B cells exhibited upregulation of B-lymphoid genes (Btg2, Cr2/Cd21, Cxcr5, Fcer2a/Cd23, and Cd40) but downregulated various B-lymphoid genes, including *Ikzf3/*Aiolos, Fos, Marks, Ciita, Birc3, Hes1, Bach2, Fyn, and Junb (BCR signaling), receptors involved in B-lymphoid differentiation (Igfr1, Il5ra, and Sirpa) and BM egress (S1pr1) (Figures 7A and 7E–7J; Table S3).
Overall, these results indicate that EPO exposure imprints a myeloid-like signature in B cell precursors, without fully transitioning them to the myeloid lineage.
Mini-RNA-seq analysis reveals significant transcriptional deregulation in EPO-exposed HSC and early lymphoid progenitors
EPO-exposed mice showed downregulation of lymphoid (CLP and B progenitors) and myeloid (PreGM) signatures in Flt3^−^CD127^−^LSK, with shifts toward Ery/Mk lineages (Grover et al., 2014) in LSK and LMPPs (Figure S7; Table S2). This shift is supported by the upregulation of Ery/Mk differentiation-associated genes (Hoxa10, Tfr2, Dapk1, Pf4, F13a1, Ptgs1, Emilin1, and Ccl5). Despite upregulation of lymphoid priming genes (Itgb7) and T/NK/ILC differentiation genes (Ikzf4, Irf2, and Notch2), EPO-exposed LSK downregulated key genes related to B lymphoid (Lyl1, Cnn3, St6gal1, Gpat3, Cd22, Ikzf3/Aiolos, and CD74), T-lymphoid (Jam2 and Il21r), and myeloid differentiation (Ms4a3, Ctsh, Alas1). Normal LMPP differentiation requires loss of Ery/Mk potential and gain of lymphoid potential (Adolfsson et al., 2005). EPO-exposed LMPPs upregulated myeloid genes (Ctsd, S100A6, and Ltbr), lymphoid (Ikzf3 and Cd2) genes, and the T/NK lineage specification marker Notch1. Notably, LMPP also upregulated the erythroid differentiation factor Tal1, essential for balancing myeloid and T lymphocyte differentiation (De Pooter et al., 2019), while downregulating markers associated with dendritic cell (DC) maturation (Cd83 and Epas1), possibly indicating altered differentiation potential (Figures 7A–7C; Table S3).
In line with impaired lymphopoiesis in vivo, EPO-exposed CLP downregulated lymphoid genes Erg, Fos, Bfsp2, and Hells, as well as pre-BCR complex components Igll1 (λ5) and Vpreb2. Conversely, EPO-exposed CLP overexpressed myeloid/macrophage differentiation markers (*Csf1r/*M-CSFR and Csf2rb2/βCR) and ILC/T lymphocyte precursor genes (Notch2, Il18rap, Batf, Il21r, and Ccr7), reflecting EPO’s impact on lymphoid differentiation (Figures 7A and 7D; Table S3).
Overall, these findings indicate that EPO exposure significantly reprograms the hematopoietic landscape, affecting both lymphoid and myeloid identities at the transcriptomic level in response to stress signals.
EPO exposure disrupts gene networks controlling progenitor proliferation and iron homeostasis and induces markers of oxidative stress
GO and GSEA analyses revealed stage-specific effects of EPO on BM progenitors, primarily downregulating pathways related to metabolism, development, proliferation, and stimulus response across most stages. However, CLPs and pre-BII cells (particularly pre-BIIL) upregulated genes tied to development, activation, and adhesion. LSK and CLPs showed reduced MYCN and mTOR-C1 targets, while Pre-BI upregulated MYC targets (Table S2; Figure S8). EPO did not affect mitosis-associated genes in LSK, suggesting transcriptomic reprogramming without cell-cycle changes (Grover et al., 2014). Confirming microarray results, pre-BI upregulated cell-cycle genes, while other stages downregulated cell-cycle/cell division and E2F target genes. Pre-pro-B, pre-BIIS, and immature B cells also downregulated cell-death-related genes, which might reduce negative selection of B cell precursors (Figure S8). EPO also induced markers of oxidative stress and altered iron metabolism genes. Progenitors showed altered expression of oxidative stress and redox homeostasis genes (like for instance Mapk14, Meis1, Sod2, Txndc11, etc.) and iron metabolism genes (Tfr2, Acp5, Glrx3, Trf, Steap3, Lamp2, Ncoa4, and Tfrc), indicating a complex response to EPO exposure (Table S3).
Discussion
Hematopoiesis is a complex process that maintains homeostasis by adapting HSPC proliferation and differentiation to environmental cues including hormones, cytokines, and metabolites.
Although EPO has been shown to impair B cell lymphopoiesis (Grover et al., 2014; Singbrant et al., 2011; Singh et al., 2018), its effects on B-cell-lineage-committed and upstream uncommitted/lymphoid-biased progenitors remain incompletely understood. Here, we investigated the impact of sustained EPO exposure on B lymphopoiesis in mice, focusing on upstream progenitor populations. Given the plasticity of early progenitors, we also assessed the balance between B lymphoid and myeloid lineage output through phenotypic and gene profiling analysis. The connection between these two lineages is highlighted by (1) the existence of biphenotypic B/myeloid progenitors (Audzevich et al., 2017; Zriwil et al., 2016) and of “myeloid-like” B cells (M-B cells) (Kanayama et al., 2023), (2) the persistence of myeloid potential up to the pre-pro-B stage (Guo et al., 2018; Rumfelt et al., 2006), and (3) the reprogramming ability of B progenitors into myeloid cells (Deshet-Unger et al., 2020; Rolink et al., 2000; Xie et al., 2004).
Our quantitative analysis shows that sustained hyper-EPOemia profoundly skews hematopoiesis toward the erythroid/megakaryocytic lineages at the expense of B-lymphoid and monocytic progenitors. Absolute counts demonstrated a robust expansion of the LSK population and a moderate increase in the LK compartment (which harbors the earliest erythroid and megakaryocytic progenitors, including pre-MegE, MEP, and pre-CFU-E) (Grover et al., 2014; Pronk et al., 2007), together with a reduction in B and monocytic outputs. At the lymphoid level, B cell development is specifically impaired at the pre-pro-B to pro-B transition, whereas Flt3^−^HSCs/MPPs and early CLP/Pre-Pro-B compartments are expanded. This pattern is consistent with the differential effects of acute versus chronic EPO exposure: while acute stimulation mainly boosts erythroid progenitors without mobilizing HSCs (Grover et al., 2014; Singh et al., 2018), chronic hyper-EPOemia drives erythroid skewing through profound transcriptional reprogramming of HSCs and MPPs, including repression of lymphoid and myeloid differentiation programs, as demonstrated in Tg6 mice (Singh et al., 2018).
Notably, EPO expanded previously unreported “biphenotypic” lymphoid/myeloid populations of CD115+/M-CSFR CLPs and CD11b+CD16/32+ pre-pro-B cells characterized by myeloid imprinting at the transcriptional level. Therefore, our data indicate that biphenotypic B/myeloid-like precursors in the BM under hyper-EPOemia arise from early progenitors between the CLP and pre-pro-B stages, where EPO induces CD115 expression and a partial myeloid program. We also observed an expansion of “atypical” lymphoid-biased progenitors lacking CD127 expression, including CD115^+^c-Kit^low^Sca1^+^Flt3^+^CD127^−^ “CLP-like” and CD11c^−^DX5^−^B220^+/low^CD19^−^CD43^hi^c-Kit^+^CD127^−^ “Pre-Pro-B-like” cells, without a shift toward myeloid over lymphoid fate. These CLP-like or pre-pro-B-like cells may represent transitional intermediates between LMPPs/CLPs and pre-pro-B cells or atypical B-lymphoid progenitors with downregulated CD127.
Gene profiling also revealed that EPO induced myeloid gene upregulation in lymphoid progenitors, starting in CLPs and persisting throughout B lymphopoiesis. EPO-exposed B precursors, particularly at the pre-B stage, exhibited a mixed lymphoid-myeloid transcriptional profile, shared with M-B cells involved in emergency myelopoiesis (Kanayama et al., 2023). This suggests that EPO may promote M-B cell differentiation.
The generation of CD115/M-CSF1R-expressing CLPs may be a hallmark of hyper-EPOemia-induced stress hematopoiesis, influencing cell fate decisions toward an M-B type population. Indeed, M-CSF promotes the growth and differentiation of mononuclear phagocytes, macrophages, and DCs, while also acting as a myeloid fate-instructive cytokine by driving the expression of the myeloid master regulator PU.1 (Mossadegh-Keller et al., 2013). Whereas CD115 is expressed at low levels in normal hematopoietic progenitors, it undergoes epigenetic silencing during B lymphopoiesis (Tagoh et al., 2004). Additionally, active repression mechanisms prevent CD115 expression in B-lymphoid progenitors, as PAX5 represses myeloid-specific genes by directly interfering with transcription factor binding and the basal transcription machinery (Ingram et al., 2011; Tagoh et al., 2006). CD115-expressing pro-B cells with residual myeloid potential and lympho-myeloid progenitors co-expressing CD127 and CD115 are found in the developing fetal liver but do not persist into adulthood (Kajikhina et al., 2015; Zriwil et al., 2016). Therefore, in adult mice, CD115 expression in lymphoid progenitors suggests a potential connection between EPO treatment and stress hematopoiesis, possibly by ineffective epigenetic silencing of CD115 expression (Tagoh et al., 2004). We speculate that the EPO-induced expansion of CD115^+^ CLPs and CD16/32^+^CD11b^+^ pre-pro-B cells may be driven by M-CSF, as EPO has also been shown to increase M-CSF production within the BM microenvironment (Gorodov et al., 2024).
Furthermore, our results showed that EPO-exposed CLPs upregulated Csf2rb2, encoding the βCR/CD131 (GM-CSF receptor β and tissue protective receptor [TPR] subunit), which is crucial for hematopoietic progenitor survival and myeloid differentiation, especially after BM damage (Chen et al., 2016). Also, ectopic GM-CSFR expression on CLPs promotes myelomonocytic differentiation, while GM-CSFR downregulation is essential for lymphoid lineage potential (Iwasaki-Arai et al., 2003). Overall, these findings suggest that EPO induces broader stress hematopoiesis beyond fast-track erythropoiesis, promoting generation of a unique population of B cell progenitors bearing myeloid imprinting at transcriptional and phenotypic levels. While the effects of EPO on BM microenvironment are recognized, impaired lymphopoiesis and myelopoiesis may also involve intrinsic mechanisms, as evidenced from altered expression of key receptors including c-Kit, Flt3, CD127/IL-7Rα, and CD115/M-CSFR. As IL-7 signaling plays a critical role in B lymphopoiesis (Kaiser et al., 2023), decreased CD127/IL-7Rα expression and increased CD115-to-CD127 ratio in CLPs may hinder IL-7R-STAT5 signaling, leading to reduced pro-B cell survival and early B cell block.
Ultimately, the myeloid-like B cell precursors observed under EPO reflect the intrinsic plasticity of early lymphoid progenitors, consistent with prior evidence that B-lymphoid and myeloid lineages share developmental potential (Kanayama et al., 2023; Rolink et al., 2000; Rumfelt et al., 2006). This plasticity is regulated by a balance between B-lineage TFs (PAX5 and EBF1) and myeloid-inducing factors (C/EBPs and PU.1) and can be modulated by cytokines such as CSF1, GM-CSF, or IL-2, which override intrinsic programs (Chen et al., 2022; Nguyen et al., 2024; Stanley and Chitu, 2014; Xie et al., 2004). EPO-exposed progenitors upregulate CSF1R/CD115 and C/EBP TFs while downregulating IL-7R and PAX5 targets, suggesting a shift toward myeloid potential.
Additionally, EPO exposure altered B-cell-identity-associated transcriptional signatures and reduced B progenitors’ proliferation, further explaining impaired B lymphopoiesis. Downregulation of pre-BCR component genes (Igll1, Vpreb2, and Vpreb1) and pre-BCR-induced TFs (IRF4, IRF8, and Aiolos) in EPO-exposed precursors suggests impaired pre-BCR signaling and/or pre-BCR-induced proliferation, potentially reducing immunoglobulin lambda (Igλ) transcription and BCR diversity. Interestingly, lymphoid precursors from patients with IL-7Rα deficiency overexpress myeloid genes, whereas mature B cells present decreased BCR diversity (Kaiser et al., 2023).
Overall, our myeloid-like B-cell precursors closely resemble the M-B cells previously described by Kanayama et al., 2023, during infection-induced emergency myelopoiesis, which also display a mixed B/myeloid transcriptional profile (Kanayama et al., 2023). To our knowledge, the interplay between erythropoietic and inflammatory stress has not yet been comprehensively investigated at both central (BM) and peripheral levels. The seminal study by Mirchandani et al., 2022, primarily focused on peripheral compartments, demonstrated in a mouse model of hypoxic acute lung injury and in acute respiratory distress syndrome (ARDS) patients that hypoxemia promotes erythroid bias while impairing monopoiesis, paralleling our observations under hyper-EPOemia (Mirchandani et al., 2022). Importantly, this study also showed that CSF1-Fc treatment restores monopoiesis and accelerates inflammation resolution, underscoring the pivotal role of the M-CSF axis in stress-induced myelopoiesis (Mirchandani et al., 2022).
When viewed alongside (1) the findings of Mirchandani et al., 2022, (2) earlier work on B-lineage plasticity showing that early pro-B cells can transdifferentiate into macrophages (Audzevich et al., 2017), (3) evidence that CSF1R^+^PAX5^low^ pre-B/immature B cells can convert into macrophage-like cells in response to tumor-derived M-CSF (Chen et al., 2022), and (4) the well-established pro-resolving actions of EPO mediated through its TPR (EpoR/CD131) complex on macrophages (Peng et al., 2020), our results support a unified model. We propose a model in which combined hypoxemic and inflammatory stress positions EPO as a central hematopoietic biasing signal, predisposing early lymphoid progenitors toward myeloid and regenerative fates while simultaneously coordinating stress-adaptive erythropoiesis. This coordinated adaptation would ensure both efficient oxygen transport and inflammation resolution through the remarkably pleiotropic action of EPO. Ongoing in vivo fate mapping and transplantation studies should determine whether these EPO-induced progenitors can generate mature B or myeloid cells and whether they functionally resemble populations generated during infection- or stress-induced myelopoiesis.
In contrast to the BM, EPO enhanced splenic lymphopoiesis and myelopoiesis, likely due to HSC/MPP mobilization during stress-induced erythropoiesis. This may explain the blood monocytosis and B lymphopoiesis observed in EPO-treated mice. An alternative extramedullary erythropoiesis pathway, involving BMP4 and Hedgehog signaling, supported by HSC mobilization from the BM and driven by stress erythroid progenitors, occurs during recovery from acute anemia or inflammation (Perry et al., 2009). Whether supraphysiological EPO engages the BMP4-dependent splenic stress erythropoiesis pathway (Paulson et al., 2020) remains unknown, and its effects on splenic progenitors warrant further investigation. Overall, these findings suggest that EPO reshapes early B cell trajectories centrally while the spleen compensates to maintain immune output.
Furthermore, EPO exposure increased total, M-B like and CD21^−^CD23^−^ ABC-like B cells in the spleen and blood, suggesting potential disruption in B cell distribution and/or extrafollicular B cell development. Unique CD21^−/low^ CD11c^+^ and/or CD11b^+^ B cells called ABCs have been associated with autoimmune diseases and severe COVID-19 (Mouat et al., 2022; Woodruff et al., 2020).
This study suggests that high EPO levels may induce myeloid reprogramming of lymphoid progenitors, but some limitations preclude a clear distinction between the direct and indirect effects of EPO. However, while our transcriptomic data do not allow us to discriminate between intrinsic and extrinsic mechanisms, existing literature provides a strong framework to interpret the lympho-myeloid shift observed under sustained EPO exposure. First, EPO can act directly on lymphoid progenitors, as CD19^+^ B-lineage cells and pro-B cells express EpoR (Deshet-Unger et al., 2020; Zhang et al., 2021) and in vitro EPO biases MPPs toward erythro-myeloid fates independently of niche signals (Eisele et al., 2022). Concurrently, EPO alters the BM microenvironment by downregulating CXCL12, IL-7, and VCAM1 in stromal and endothelial cells, reducing support for early B cell progenitors (Ito et al., 2017; Singbrant et al., 2011). Similar effects occur with G-CSF, which reprograms niches, decreases trophic factors, and induces B cell precursor apoptosis (Day et al., 2015). EPO may also indirectly impair B cell maturation via hypoxia and HIF-dependent pathways, as HIF stabilization or constitutive HIF-1α activation blocks immature B cell development (Burrows et al., 2020). Together, these findings indicate sustained EPO signaling suppresses B lymphopoiesis through direct, stromal, and hypoxia-driven mechanisms, which likely also contribute to myeloid reprogramming in lymphoid progenitors. The preliminary phenotypic and gene expression results require validation through in vitro assays and lineage tracing of EpoR-expressing progenitors (Zhang et al., 2021).
Furthermore, the myeloid imprinting of lymphoid progenitors under EPO may reflect not only transcriptional skewing but also deeper epigenetic and chromatin remodeling within early progenitors. Hematopoietic stress can reshape enhancer accessibility, DNA methylation, and 3D chromatin architecture, biasing lineage potential (Meng and Nerlov, 2024). During hyper-EPOemia, direct and indirect signals may increase epigenomic plasticity in early lymphoid progenitors. B lymphopoiesis is driven by a hierarchical network of TFs—including Ikaros, PU.1, E2A, FoxO1, EBF1, and Pax5—that cooperatively shape the epigenetic landscape and direct lineage commitment (Boller and Grosschedl, 2014; Bossen et al., 2015). EBF1 and Pax5, in particular, function as epigenetic regulators to activate B cell programs and repress alternative fates (Busslinger and Tarakhovsky, 2014; Hagman et al., 2011; Li et al., 2018; Lin et al., 2010). Because Pax5-deficient Pro-B cells readily adopt myeloid identities (Cobaleda et al., 2007; Delogu et al., 2006; Heavey, 2003), EPO-induced reprogramming at the CLP-Pre-Pro-B stage may involve transient epigenetic modulation of essential B-lineage regulators. Given that epigenetic state often predicts fate more accurately than transcription, the EPO-driven shift toward myeloid trajectories may stem from altered chromatin priming and suppression of lymphoid regulatory landscapes. Dedicated epigenomic studies will be required to test this hypothesis.
Overall, our results reinforce the concept of a shared developmental pathway between B lymphoid and myeloid lineages, highlight the versatility of progenitor branching points and the remodeling role of stress-related cytokines and contribute to understanding hematopoietic emergencies in response to various stimuli. Additionally, better understanding of the hematopoietic impact of rhEPO/ESA use will help tailor these therapies to individual patient needs.
Methods
Mice
C57BL/6 mice (Envigo, Harlan Laboratories) were housed at the Faculté de Pharmacie de Paris. Groups of 3–5 male mice (4–6 weeks old) received subcutaneous injections every 2 days for 2 weeks with either 40 IU/mouse (2,000 IU/kg) of rhEPO (Eprex, 10,000 IU/mL, Janssen) or PBS (control). This EPO regimen represents an intermediate dosing strategy within the highly variable mouse protocols reported in the literature, which span 300–10,000 IU/kg and 3 days to 4 months (Deshet-Unger et al., 2020; Singbrant et al., 2011; Wang et al., 2013; Zhang et al., 2018). This previously validated protocol reliably raises hematocrit and induces splenomegaly, confirming strong systemic and stress-erythropoietic activity (Bessoles et al., 2024; Sarrabayrouse et al., 2025). The selected dose is sufficient to activate both the high-affinity EPOR and potentially the lower-affinity EPOR-CD131 complex, which requires higher EPO levels (Broxmeyer, 2013). Additional details are in supplemental methods.
Flow cytometry and fluorescent-activated cell sorting
Immunophenotypic characterization and isolation of hematopoietic populations were performed as described (Bessoles et al., 2024). Additional details are in supplemental methods.
Gene expression profiling by GeneChip microarray
RNA was extracted (Qiagen RNAeasy Micro Kit) and cDNA hybridized to the Affymetrix Clariom S Mouse microarray (Plateforme Génomique, Centre de Recherche de l’Institut Curie, Paris). Bioinformatics analysis was conducted by GenoSplice technology (Bessoles et al., 2024), as detailed in supplemental methods.
Gene expression profiling by mini-RNA-seq
Mini-RNA sequencing (RNA-seq; DGE-seq) was performed as adapted from Cacchiarelli et al. (2015) and Alhaj Hussen et al. (2023). Libraries were sequenced on a NextSeq 2000 (Illumina) at the Genome East Platform (Strasbourg). Technical details and bioinformatic analysis are in supplemental methods.
Statistical analysis
Data were analyzed using GraphPad Prism (v.9.5.1, GraphPad Software Inc., San Diego, CA). Statistical significance was assessed with Mann-Whitney tests (^∗^p < 0.05; ^∗∗^p < 0.01; ^∗∗∗^p < 0.001; ^∗∗∗∗^p < 0.0001). Results are presented as mean ± standard deviation, with significance indicated in figures and legends.
Resource availability
Lead contact
Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Prof. Salima Hacein-Bey-Abina ([email protected]).
Materials availability
This study did not generate new unique reagents.
Data and code availability
Microarray data and Mini-RNA-Seq data have been deposited at GEO, under accession numbers GEO: [GSE288756](GSE288756) and GEO: [GSE289311](GSE289311), respectively, and will be publicly available as of the date of publication.
All other data reported in this paper and any additional information required to reanalyze the data will be shared by the lead contact upon request. This paper does not report original code.
Acknowledgments
This study was funded by 10.13039/501100004099Ligue Contre le Cancer (RS18/75-109) and by the nonprofit organizations ADRPP and Adebiopharm. We express our sincere gratitude to Professors Michel Plotkine and Dominique Bellet for their support. We are grateful to Dr. Stéphane Mancini for his invaluable advices and insightful discussion on our study. We would also like to thank the staff at the animal facility of the UFR de Pharmacie, Faculté de Santé, Université Paris Cité (Paris, France), for their expert animal care, as well as the team at the Centre d'Histologie, d'Imagerie et de Cytométrie (CHIC), Centre de Recherche des Cordeliers (CRC), with special thanks to Dr. Helene Fohrer-Ting. Additionally, we are grateful to the staff of the Plateforme Génomique, Centre de Recherche de l’Institut Curie and the Paris Genome East Platform at the IGBMC (Strasbourg, France).
Author contributions
A.C.-M. designed and performed the experiments, analyzed the data, and wrote the manuscript; S.B. helped designing and performing the experiments, participated in transcriptional analyses, and critically revised the manuscript; G.S. and C.J. helped performing experiments and critically revised the manuscript; K.A.-H. helped performing mini-RNA-seq experiments and critically revised the manuscript; P.G. and T.D. performed the transcriptional analyses; A.M.A. conceived the project and critically revised the manuscript; B.C. contributed to data analysis, helped interpreting the results, and critically revised the manuscript; R.K. contributed to data analysis, helped interpreting the results, and wrote the manuscript; S.H.-B.-A. conceived the project, ensured the scientific supervision of the project, designed the experiments, interpreted the results, and wrote the manuscript. All authors read and approved the manuscript.
Declaration of interests
All authors declare that they have no conflicts of interest to disclose.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Adolfsson J.Månsson R.Buza-Vidas N.Hultquist A.Liuba K.Jensen C.T.Bryder D.Yang L.Borge O.-J.Thoren L.A.M.Identification of Flt 3+ Lympho-Myeloid Stem Cells Lacking Erythro-Megakaryocytic Potential Cell 121200529530610.1016/j.cell.2005.02.01315851035 · doi ↗ · pubmed ↗
- 2Alhaj Hussen K.Chabaane E.Nelson E.Lekiashvili S.Diop S.Keita S.Evrard B.Lardenois A.Delord M.Verhoeyen E.Multimodal cartography of human lymphopoiesis reveals B and T/NK/ILC lineages are subjected to differential regulationi Science 26202310789010.1016/j.isci.2023.107890 PMC 1052054037766969 · doi ↗ · pubmed ↗
- 3Audzevich T.Bashford-Rogers R.Mabbott N.A.Frampton D.Freeman T.C.Potocnik A.Kellam P.Gilroy D.W.Pre/pro-B cells generate macrophage populations during homeostasis and inflammation Proc. Natl. Acad. Sci. USA 1142017 E 3954 E 396310.1073/pnas.161641711428461481 PMC 5441795 · doi ↗ · pubmed ↗
- 4Bessoles S.Chiron A.Sarrabayrouse G.De La Grange P.Abina A.M.Hacein-Bey-Abina S.Erythropoietin induces tumour progression and CD 39 expression on immune cells in a preclinical model of triple-negative breast cancer Immunology 173202436038010.1111/imm.1383238953295 · doi ↗ · pubmed ↗
- 5Blin-Wakkach C.Wakkach A.Rochet N.Carle G.F.Characterization of a Novel Bipotent Hematopoietic Progenitor Population in Normal and Osteopetrotic Mice J. Bone Miner. Res.1920041137114310.1359/JBMR.04031815176997 · doi ↗ · pubmed ↗
- 6Boller S.Grosschedl R.The regulatory network of B-cell differentiation: a focused view of early B-cell factor 1 function Immunol. Rev.261201410211510.1111/imr.1220625123279 PMC 4312928 · doi ↗ · pubmed ↗
- 7Bossen C.Murre C.S.Chang A.N.Mansson R.Rodewald H.-R.Murre C.The chromatin remodeler Brg 1 activates enhancer repertoires to establish B cell identity and modulate cell growth Nat. Immunol.16201577578410.1038/ni.317025985234 PMC 4474778 · doi ↗ · pubmed ↗
- 8Broxmeyer H.E.Erythropoietin: multiple targets, actions, and modifying influences for biological and clinical consideration J. Exp. Med.210201320520810.1084/jem.2012276023401569 PMC 3570099 · doi ↗ · pubmed ↗