Gdf15 regulates murine stress erythroid progenitor proliferation and the development of the stress erythropoiesis niche

Abstract

Anemic stress induces the proliferation of stress erythroid progenitors in the murine spleen that subsequently differentiate to generate erythrocytes to maintain homeostasis. This process relies on the interaction between stress erythroid progenitors and the signals generated in the splenic erythroid niche. In this study, we demonstrate that although growth-differentiation factor 15 (Gdf15) is not required for steady-state erythropoiesis, it plays an essential role in stress erythropoiesis. Gdf15 acts at 2 levels. In the splenic niche, Gdf15^-/- mice exhibit defects in the monocyte-derived expansion of the splenic niche, resulting in impaired proliferation of stress erythroid progenitors and production of stress burst forming unit-erythroid cells. Furthermore, Gdf15 signaling maintains the hypoxia-dependent expression of the niche signal, Bmp4, whereas in stress erythroid progenitors, Gdf15 signaling regulates the expression of metabolic enzymes, which contribute to the rapid proliferation of stress erythroid progenitors. Thus, Gdf15 functions as a comprehensive regulator that coordinates the stress erythroid microenvironment with the metabolic status of progenitors to promote stress erythropoiesis.

Graphical abstract

Figure 1.

Gdf15 signaling is required for recovery from PHZ induced acute anemia. (A) Steady-state erythropoiesis is represented by red blood cell (RBC) counts, hemoglobin (Hb) concentrations, and hematocrit levels. n = 9 for WT and n = 11 for Gdf15^−/−. Data are shown as individual subject and the mean ± standard error of the mean (SEM). (B-F) Analysis of recovery from PHZ-induced acute anemia. (B) Survival analysis of WT and Gdf15^−/− after receiving different doses of PHZ with P = .0019 when Gdf15^−/− is compared with control (Gehan-Breslow-Wilcoxon test). (C) Hematocrit levels during the entire recovery period. Each time point reflects >5 mice analyzed. Data are shown as the mean ± SEM. (D) Spleen weights at key time points during recovery. (E-F) Spleen stress BFU-E assay was conducted at key time points following PHZ injection. (E) Quantification of stress BFU-E colonies. (F) Representative images of stress BFU-E colonies of WT and Gdf15^−/− mice. Pictures were taken with 40× magnification using an Olympus CK40 microscope. Each time point reflects >3 mice analyzed. Data are shown as individual subject and the mean ± SEM. *P < .05, **P < .01, and ***P < .001.

Figure 2.

Gdf15 signaling is required for erythroid short-term radioprotection following bone marrow transplant. Analysis of short-term radioprotection by WT or Gdf15^−/− bone marrow cells following transplant into lethally irradiated recipients. (A) Survival curve with P < .001 (Gehan-Breslow-Wilcoxon test). Data are representative of 4 individual transplants with total n = 9 for WT and n = 16 for Gdf15^−/−. (B) Hematocrit levels during the recovery period. Each time point reflects >5 mice analyzed. Data are shown as the mean ± SEM. (C-E) Donor mice were CD45.2 and recipient mice were CD45.1. Spleens were isolated at day 8 and day 10 post-BMT for analysis. (C) Analysis of the proliferation of cell in the spleen following transplant, spleen weight (left), and total cell number (right) on the indicated days after transplant are shown. (D) Analysis of donor cell proliferation, total CD45.2⁺ donor cells (left), Kit⁺Sca1⁺ progenitors (middle), and CD34⁺CD133⁺Kit⁺Sca1⁺ progenitors (right) in each spleen. (E) Number of stress BFU-E in the spleen on the indicated days after transplant. Each time point reflects >3 mice analyzed. Data are shown as individual subject and the mean ± SEM. *P < .05, **P < .01, and *** P < .001.

Figure 3.

Mutation of Gdf15 impairs expansion of stress erythroid progenitors in vitro. (A) Analysis of relative Gdf15 expression levels in unfractionated bone marrow (control), isolated erythroid progenitors (progenitors), and stromal cells (stroma) after 7 days of SEEM culture. One-way analysis of variance followed by Tukey pairwise comparison was performed. (B-E) Unfractionated WT or Gdf15^−/− bone marrow cells were cultured for 7 days in SEEM. Nonadherent SEPs were collected and stained with fluorescent anti-Kit, anti-Sca1, and anti-CD133 antibodies. (B) Flow gating strategy schematic. (C) Representative flow cytometry analysis. (D) Percentage (left) and total number (right) of Kit⁺Sca1⁺ cells. (E) Percentage (left) and total number (right) of CD34⁺CD133⁺Kit⁺Sca1⁺ cells, Data are shown as individual subject and the mean ± SEM. (F-H) Unfractionated WT or Gdf15^−/− bone marrow cells were labeled with PKH26 dye at the beginning of a 5-day culture in SEEM. (F) Schematic of the gating strategy. (G) Representative flow cytometry analysis. (H) Percentage of PKH26^loCD133^negKit⁺Sca1⁺ cells (top) and PKH26^hiCD133⁺Kit⁺Sca1⁺ cells (bottom). Data are shown as individual subject and the mean ± SEM. *P < .05, **P < .01, and ***P < .001. ns, not significant.

Figure 4.

Gdf15^−/−defects in steady state and in splenic niche after 75% dosage of PHZ challenge. (A) Analysis of the concentration and percentage of monocytes in the peripheral blood at steady state. n = 9 for WT and n = 11 for Gdf15^−/−. (B) Analysis of F4/80⁻CD11b⁺Ly6C⁺ monocytes in the homeostatic bone marrow. (C) Flow cytometry analysis of steady-state spleen resident red pulp macrophages, F4/80⁺Vcam1⁺ (left) and percentage of F4/80⁺Vcam1⁺ cells (right). (D-N) WT and Gdf15^−/− mice were challenged with 75 mg/kg body weight PHZ through intraperitoneal injection. (D-F) Spleen cells were collected 24 hours after PHZ injection and stained for intracellular Ccl2 and cell surface CD11b and Ly6C. (D) Percentage of Ccl2⁺ splenocytes. (E) Flow cytometry analysis of CD11b and Ly6C expression on Ccl2⁺ cells (left) and the percentage of CCL2⁺CD11b⁺Ly6C⁺ monocytes in the spleen (right). (F) MFI of F4/80 expression on CCL2⁺CD11b⁺Ly6C⁺ monocyte population. (G-N) Blood and spleens were collected 48 hours after PHZ challenge. (G) Hematocrit. (H) Monocyte frequency (left) and percentage (right) in peripheral blood. (I) Flow cytometry analysis of spleen cells stained with anti-F4/80, anti-CD11b and anti-Ly6C antibodies (left) and the percentage of CD11b⁺Ly6C⁺ monocytes in the spleen (right). (J) The percentage of F4/80⁺CD11b⁻Ly6C⁻ macrophages in the spleen. (K) Flow cytometry analysis of macrophage surface markers, F4/80 and Vcam1 (left), and the percentage of F4/80⁺Vcam1⁺ red pulp macrophages. (L) Flow cytometry analysis CD169 expression in splenocytes (left) and the percentage of F4/80⁺CD169⁺ macrophages in the spleen. (M) Changes in macrophage and monocyte populations between 24-hour and 48-hour time points after PHZ treatment. Percentage of CD11b⁺Ly6C⁺ monocytes (far left), percentage of F4/80⁺CD11b⁻Ly6C⁻ macrophages (mid-left), percentage of F4/80⁺Vcam1⁺ RPMs (mid-right) and percentage of F4/80⁺CD169⁺ macrophages (far right) are shown as fold changes. (N) Change in relative expression levels of CD163, Hmox1, and Spic in the spleen between 24-hour and 48-hour time points are shown as fold changes. Data are shown as individual subject and the mean ± SEM. *P < .05, **P < .01, and ***P < .001. (O) Schematic of differences of monocyte mobilization in the peripheral blood, homing to the spleen, and monocytes differentiation into RPMs in WT and Gdf15^−/− mice. MFI, mean fluorescent intensity.

Figure 5.

Defects in Gdf15^−/−stroma contribute to impaired expansion of SEPs in vitro. (A-E) Gdf15^−/− BM cells were cultured in SEEM in vitro with or without the addition of WT CD11b⁺Ly6C⁺ monocyte (WT MO). After coculturing for 7 days, total cell numbers are presented in panel A. (B) Representative flow cytometry analysis of erythroid progenitor surface markers. (C) The percentages of Kit⁺Sca1⁺ SEPs (left) and absolute numbers of Kit⁺Sca1⁺ SEPs (right). (D) The percentages (left) and absolute numbers (right) of CD34⁺CD133⁺Kit⁺Sca1⁺ SEPs. Addition of WT monocytes promoted the expansion of Gdf15^−/− SEPs. (E) Stress BFU-E colony production in cultures ± WT monocytes. (F-I) WT and Gdf15^−/− BM cells were cultured in SEEM for 3 days. Nonadherent progenitor cells were collected with culture media and plated on the stromal layer of the indicated genotype. Progenitors were cultured for another 4 days before analysis. (F) Flow cytometry analysis of living vs dead progenitor cells. (G) Flow cytometry analysis of Kit⁺Sca1⁺ stress erythroid progenitors of the indicated genotype grown on the indicated stromal genotype. (H) Statistical analyses on percentages (left) and absolute numbers (right) of Kit⁺Sca1⁺ SEPs. (I) Statistical analyses on percentages (left) and absolute numbers (right) of CD34⁺CD133⁺Kit⁺Sca1⁺ SEPs. Data are shown as individual subject and the mean ± SEM. *P < .05, **P < .01, and ***P < .001.

Figure 6.

Gdf15 signaling activates Hif2α-dependent Bmp4 expression via inhibition of Vhl during the recovery from anemia. (A-B) Analysis of relative mRNA expression levels of Bmp4 in the spleen at key time points during the recovery from BMT (A) and PHZ challenge (B), respectively. Each time point reflects >3 mice analyzed. Data are shown as individual subject and the mean ± SEM. (C) Primary spleen cells isolated from Gdf15^−/− mice were treated ±10 ng/mL GDF15 for the indicated time. Relative Bmp4 and Vhl mRNA expression levels were determined by qRT-PCR. Each time point reflects >3 mice analyzed. Data are shown as individual subject and the mean ± SEM. (D) Primary spleen cells derived from Gdf15^−/− mice were transduced with short hairpin RNAs targeting Vhl gene for 48 hours and then cultured in stress erythropoiesis media for 72 hours. Relative mRNA expression levels of Bmp4 and Vhl were analyzed by qRT-PCR. Each treatment was done in triplicate and the histogram is representative of 2 independent experiments. Data are shown as the mean ± SEM. (E-F) Analysis of relative mRNA expression levels of Vhl in the spleen at key time points during the recovery from (E) BMT and (F) PHZ challenge, respectively. Each time point reflects >3 mice analyzed. Data are shown as individual subject and the mean ± SEM. (G-H) Chromatin immunoprecipitation analysis of Hif2α binding to HRE4 of Bmp4 gene in the spleen during the recovery from BMT (G) and PHZ (H) treatment. Binding affinity is expressed as fraction of input DNA. Each time point reflects >3 mice analyzed. Data are shown as the mean ± SEM. (I-J) Analysis of Vhl and Hif2α protein expression in the spleen during the recovery from BMT (I) and PHZ (J) challenge, respectively. Western blots were probed with anti-VHL, anti-HIF2α, and anti-β-actin (control) antibodies. Three independent experiments were performed. In quantification graphs, data from 1 independent experiment are shown as individual subject and the mean ± SEM. *P < .05, **P < .01, and ***P < .001.

Figure 7.

Mutation of Gdf15 alters the expression of metabolic enzymes during stress erythropoiesis. (A) Five × 10⁵ bone marrow cells of WT or Gdf15^−/− were transplanted to lethally irradiated WT recipients. Relative expression levels of Hif1α, Pdk1, Pdk3, Glut1, and Gls1 at indicated time points are plotted in a base-10 logarithmic scale. (B) WT and Gdf15^−/− bone marrow cells were cultured in expansion media for 7 days. Stress erythroid progenitors were collected and analyzed with qRT-PCR. Relative expression levels of Hif1α, Pdk1, Pdk3, Glut1, and Gls1 are shown. (C) qRT-PCR analysis of Hif1α, Pdk1, Pdk3, Glut1, and Gls1 relative expression levels in control and mutant progenitors cultured on control or mutant stroma as indicated (see Figure 5F-G for description). Paired Student t test was performed to evaluate differences in gene expression levels. (D-E) Gdf15^−/− bone marrow cells were cultured in vitro with or without 2 mM glutamate supplementation. (D) Flow cytometry analysis of (left) Kit⁺Sca1⁺ progenitor cells, (middle) percentage, and (right) absolute number of CD133^negKit⁺Sca1⁺ SEPs are shown. (E) Analysis of stress BFU-E colony formation assay in hypoxia (2% O₂). Data are shown as individual subject and the mean ± SEM. *P < .05, **P < .01, and ***P < .001.