Suppression of Fas-FasL coexpression by erythropoietin mediates erythroblast expansion during the erythropoietic stress response in vivo

Abstract

Erythropoietin (Epo) is the principal regulator of the erythropoietic response to hypoxic stress, through its receptor, EpoR. The EpoR signals mediating the stress response are largely unknown, and the spectrum of progenitors that are stress responsive is not fully defined. Here, we used flow cytometry to identify stress-responsive Ter119+CD71highFSChigh early erythroblast subsets in vivo. In the mouse spleen, an erythropoietic reserve organ, early erythroblasts were present at lower frequencies and were undergoing higher rates of apoptosis than equivalent cells in bone marrow. A high proportion of splenic early erythroblasts coexpressed the death receptor Fas, and its ligand, FasL. Fas-positive early erythroblasts were significantly more likely to coexpress annexin V than equivalent, Fas-negative cells, suggesting that Fas mediates early erythroblast apoptosis in vivo. We examined several mouse models of erythropoietic stress, including erythrocytosis and beta-thalassemia. We found a dramatic increase in the frequency of splenic early erythroblasts that correlated with down-regulation of Fas and FasL from their cell surface. Further, a single injection of Epo specifically suppressed early erythroblast Fas and FasL mRNA and cell-surface expression. Therefore, Fas and FasL are negative regulators of erythropoiesis. Epo-mediated suppression of erythroblast Fas and FasL is a novel stress response pathway that facilitates erythroblast expansion in vivo.

Figure 1.

Definition of flow cytometric erythroblast subsets. (A) Flow cytometric analysis of freshly isolated bone marrow cells labeled with antibodies against Ter119 and CD71. Dead cells were excluded with 7-AAD. The left panel shows all viable bone marrow cells. Ter119^high cells are further analyzed with respect to their forward scatter (FSC) in the right panel. (B) Cytospin preparations of cells sorted from the ProEs and Ery.A, B, and C subsets shown in panel A are stained for hemoglobin with diaminobenzidine and counterstained with Giemsa. Scale bar represents 5 μm.

Figure 2.

Epo-responsive erythroblast subsets in bone marrow and spleen. Wild-type mice were injected with a single dose of Epo (50 mg/kg) or an equal volume of saline, and followed for up to 96 hours. Bone marrow and spleen cells were freshly isolated and analyzed. The results of 7 independent experiments were pooled. Each data point is the mean ± SEM of 2 to 9 mice. Each experiment contained mice of a single sex or an equal number of females and males. There was no significant difference between the Epo response of male and female mice. (A-C) Epo caused a transient increase in spleen weight (A), hematocrit (B), and in the proportion of Ter119⁺ cells in both bone marrow (Ci) and spleen (Cii). (D-E) Examples of flow cytometric histograms of bone marrow (D) and spleen (E) cells labeled with Ter119 and CD71 antibodies and analyzed as illustrated in Figure 1. There is a clear increase in the proportions of the ProE and Ery.A subsets in both tissues, with a moderate increase in Ery.B. (F) Changes in the relative frequencies of erythroblast subsets within the Ter119⁺ compartment in bone marrow or spleen, following Epo injection or a control saline injection. Data were analyzed as illustrated in panels D-E, and mean values for data pooled from 7 experiments are presented.

Figure 3.

Lower frequency and higher apoptosis of splenic early erythroblast subsets in the basal state. (A) The ProE and Ery.A subsets are a lower proportion of Ter119⁺ cells in spleen than in bone marrow. Data are the mean ± SEM from Balb/C mice (n = 7). ^*Student t test (2-tailed, unequal variance), P < .01. (B) Higher annexin V binding of splenic Ery.A compared with equivalent bone marrow cells. Viable cells (impermeable to 7-AAD) from either bone marrow (top panels) or spleen (bottom panels) were analyzed for Ter119/CD71 expression. Ter119^high cells were gated and further analyzed with respect to FSC. FSC^highCD71^high cells (Ery.A) were examined for annexin V binding (right panel). Background fluorescence in the annexin V channel was determined by comparison with Ery.A cells stained for all colors except for annexin V (“fluorescence minus one” or FMO-annexin V, gray curve). The following fluorescent conjugates were used: CD71-FITC, Ter119-APC, 7-AAD, annexin V-Alexa Fluor 350. Controls included unstained cells, single-color controls, and FMO controls. (C) Higher annexin V binding in splenic erythroblasts compared with equivalent bone marrow erythroblasts. Staining strategy and data analysis as in panel B, for each of the indicated erythroblast subsets. Mean ± SEM, Balb/C mice (n = 4); ^* P < .003; ^** P < .001. (D) Higher proportion of cells positive for activated caspase 3 in splenic than in bone marrow Ery.A. The proportion (%) of cells staining positive for caspase 3 in each tissue is indicated. Representative experiment; tissue from 2 Balb/C mice.

Figure 4.

Fas and FasL are coexpressed by early erythroblasts and mediate early erythroblast apoptosis in vivo. (A-B) Expression of Fas (A) or FasL (B) in spleen and bone marrow cells. Both Fas and FasL are expressed at higher frequencies in ProE and Ery.A subsets. Both are also expressed at higher frequencies in spleen than in bone marrow. Freshly isolated spleen and bone marrow were labeled with antibodies against Ter119, CD71, and either Fas (Jo2) or FasL (MFL3). Staining strategy and analysis were similar to those described in Figure 3. Data from C57BL6/129 mice, n = 5, mean ± SEM. ^* P < .004; ^** P < .001. (C) Fas and FasL are coexpressed by Ery.A in spleen (top left panel) and bone marrow (top right panel). Ter119/CD71/FSC analysis was used to gate Ery.A, as illustrated in Figure 1. Bottom panels show FMO controls (Figure 3B), which were used to set background fluorescence in Ery.A. These lack either Fas or FasL primary antibodies, as indicated, but contain secondary and all other primary antibodies. Representative of 3 experiments. Two different combinations of fluorescent conjugates gave similar results. (D-E) Single-cell analysis of Fas expression and annexin V binding in spleen and bone marrow Ery.A. Freshly isolated bone marrow and spleen cells were simultaneously labeled with annexin V and with antibodies to Ter119, CD71, and Fas. Fas-positive Ery.A are several-fold more likely to bind annexin V than Fas-negative cells (D). In panel E, Ery.A were subdivided into 8 subsets of increasing Fas expression, measured as geometric mean fluorescence per cells. Fas expression is correlated with annexin V binding, in both spleen and bone marrow. Data are from Balb/C mice (n = 2).

Figure 5.

Down-regulation of cell-surface Fas and FasL in early erythroblasts from β-thal mice. (A) Representative flow cytometric histograms showing an example of the changes found in erythroblast frequencies within Ter119⁺ cells in spleen in a β-thal mouse compared with a littermate wild-type mouse. (B) Representative flow cytometric histograms of Fas and FasL expression in spleen Ery.A, B, and C, in β-thal erythroblasts, and in erythroblasts from a wild-type littermate. The percentage of Ery.A positive for Fas relative to FMO control (not shown) is shown in red. The percentage of Ery.B positive for Fas is shown in blue. (C-D) Changes in erythroblast subset frequencies, and in Fas and FasL expression, in spleen (C) and bone marrow (D), in β-thal mice (n = 8) compared with littermate wild-type mice (n = 8). Data are mean ± SEM. Spleen and bone marrow cells were analyzed as illustrated in panels A-B. ^*P < .05, ^**P < .01, ^***P < .001, ^****P < .001.

Figure 6.

Suppression of erythroblast Fas and FasL and decreased apoptosis following Epo injection. Wild-type Balb/C mice were injected with a single dose of Epo (50 mg/kg) or with an equal volume of saline, and followed as described in Figure 2. Freshly isolated spleen cells were analyzed for Fas and FasL cell-surface expression and mRNAs or for apoptotic markers. (A) Representative histograms of cell-surface Fas (left panels) and FasL (middle panels) expression or annexin V binding (right panels) in Ery.A, B, and C, 48 hours following injection of either saline (top panels) or Epo (bottom panels). The percentage of Ery.A positive for Fas, FasL or annexin V is indicated in red, and that of Ery.B in blue. (B) Time course of cell-surface Fas and FasL following Epo or saline injection. Results pooled from 5 experiments, normalized to initial Fas or FasL levels; mean ± SEM, n = 2 to 7 mice per time point. (C) Time course of annexin V binding (geometric mean fluorescence) in ProEs (top panel) and in Ery.A (bottom panel) following Epo administration. Results from duplicate mice are shown for each time point. (D) Changes in activated caspase 3 in splenic erythroblasts following Epo injection. Splenic cells were labeled for Ter119 and CD71, were fixed and permeabilized, and then labeled with an antibody specific for activated caspase 3. Results from duplicate mice are shown for each time point. (E) Time course of Fas, FasB, and FasL mRNA expression in splenic Ery.A following injection with Epo, measured using quantitative real-time PCR, using RNA from freshly sorted splenic Ery.A. Bottom panels show the ΔΔCT for the indicated mRNA between mice injected with Epo and those given saline injections, normalized to actin mRNA. Each data point is the mean ± SEM of triplicate measurements in a single experiment using 1 to 2 Epo-injected mice and 1 to 2 saline-injected mice. Two to 4 independent experiments were conducted per time point. Top panels show the mean fold change in mRNA at each time point, calculated as 2^ΔΔCT, using the ΔΔCT values shown in the bottom panels. Each data point is the mean ± SEM for all the mice examined at each time point.

Figure 7.

Frequency of splenic Ery.A is a function of their Fas expression. (A) The frequency of splenic Ery.A is a function of their Fas expression. Each data point is derived from the mean values of Ery.A frequency and Fas expression presented for each erythropoietic stress model in Figures 5, 6 and Tables 1, 2. The same data are presented with either linear axes (left panel) or logarithmic axes (right panel). Ery.A frequency is expressed as a ratio of mean Ery.A frequency in the stressed mice to mean Ery.A frequency in matched controls. The fraction of Fas-positive Ery.A is calculated as the ratio of mean Fas-positive Ery.A in the stressed mice to the mean Fas-positive cells in matched controls. The slope of the line in the left panel is (-1.9), and the data fit the equation y = 0.82x^-1.9 (R = 0.912), where y is Ery.A frequency and x is Ery.A Fas expression. If only chronic erythropoietic stress models are considered (ie, pregnancy and Epo injection are omitted), the equation describing the data becomes y = 1.00x^-1.9 (R = 0.9999). (B) The homeostatic role of Fas in splenic early erythroblasts and in T cells. Splenic early erythroblasts coexpress Fas and FasL in the basal state, resulting in continuous apoptosis. Erythropoietic stress leads to Epo-mediated down-regulation of Fas and FasL expression and enhanced erythroblast survival, increasing erythropoietic rate. In contrast, naive T cells do not express Fas and FasL in the basal state. Fas/FasL expression is induced following T-cell activation and clonal expansion, when Fas-mediated apoptosis is key in activation-induced cell death.