E2F-2 Promotes Nuclear Condensation and Enucleation of Terminally Differentiated Erythroblasts

Abstract

E2F-2 is a retinoblastoma (Rb)-regulated transcription factor induced during terminal erythroid maturation. Cyclin E-mediated Rb hyperphosphorylation induces E2F transcriptional activator functions. We previously reported that deregulated cyclin E activity causes defective terminal maturation of nucleated erythroblasts in vivo Here, we found that these defects are normalized by E2F-2 deletion; however, anemia in mice with deregulated cyclin E is not improved by E2F-2-loss, which itself causes reduced peripheral red blood cell (RBC) counts without altering relative abundances of erythroblast subpopulations. To determine how E2F-2 regulates RBC production, we comprehensively studied erythropoiesis using knockout mice and hematopoietic progenitors. We found that efficient stress erythropoiesis in vivo requires E2F-2, and we also identified an unappreciated role for E2F-2 in erythroblast enucleation. In particular, E2F-2 deletion impairs nuclear condensation, a morphological feature of maturing erythroblasts. Transcriptome profiling of E2F-2-null, mature erythroblasts demonstrated widespread changes in gene expression. Notably, we identified citron Rho-interacting kinase (CRIK), which has known functions in mitosis and cytokinesis, as induced in erythroblasts in an E2F-2-dependent manner, and we found that CRIK activity promotes efficient erythroblast enucleation and nuclear condensation. Together, our data reveal novel, lineage-specific functions for E2F-2 and suggest that some mitotic kinases have specialized roles supporting enucleation of maturing erythroblasts.

Keywords: E2F; cell cycle; cell differentiation; enucleation; erythropoiesis; gene regulation; mouse models.

FIG 1

Bone marrow erythroid cell maturation defects associated with deregulated cyclin E-Cdk2 activity are E2F-2 dependent. (A) Ter119-positive bone marrow cells were sorted by expression of CD44 versus FSC (gating as shown in panel C) and collected for immunoblot analysis; relative abundance of E2F-2 compared to loading control is indicated. (B) Hematopoietic progenitors obtained from wild-type fetal livers were differentiated to the erythroid lineage in culture as shown (right panels). Left, cells were harvested at the indicated time points and immunoblotted for E2F transcription factors, with β-actin shown as a loading control. Right, representative micrographs (magnification, ×100) of fetal liver progenitors (top) and cells obtained after 2 days in erythroid differentiation culture (bottom), in which the dominant morphologies are mature, nucleated erythroid cells (E, orthochromatic erythroblast form) and enucleated erythroid cells. p, pyrenocyte; M, mature myeloid cell (neutrophil). (C) Erythroid maturation was studied using Ter119⁺ bone marrow cells collected from age- and sex-matched mice of the indicated genotypes, and representative CD44/FSC plots are shown with the percentage of cells in each subpopulation labeled. The frequency of proerythroblasts, gated based on CD44^high Ter119^dim, is shown as mean values with corresponding standard deviations derived from 4 to 8 mice of each genotype. (D) Percentages of cells in each erythroid subpopulation region (RII to RV) shown in panel C, representing averages from 4 to 8 mice per genotype. P values for cyclin E^{T74A T393A} versus E2F-2^−/−; cyclin E^{T74A T393A} comparison: RII, 0.26; RIII, 0.001; RIV, ≤0.0001. Comparisons between relative abundances of erythroid subpopulations revealed no statistically significant differences between wild-type and E2F-2 knockout mice or between wild-type and E2F-2^−/−; cyclin E^{T74A T393A} mice (P values for the latter comparison: RII, 0.56; RIII, 0.95; RIV, 0.69). (E) The indicated bone marrow erythroid subpopulations from the specified genotypes were sorted as shown in panel C, and lysates were prepared and immunoblotted. Samples were electrophoresed and blotted together using two gels and membranes imaged simultaneously. A representative immunoblot demonstrating cyclin E protein expression and HDAC2 as a nucleoprotein used for normalization is shown from three separate experiments, with relative abundance measurements indicated.

FIG 2

RBC abnormalities associated with E2F-2 deletion in vivo. (A) Red cell parameters from peripheral blood of age- and sex-matched mice in the indicated genotypes. Each symbol represents an individual mouse. ****, P ≤ 0.0001; ***, P ≤ 0.001; **, P ≤ 0.01. (B) Peripheral blood from wild-type and E2F-2-null age- and sex-matched mice was collected, stained with thiazole orange, and analyzed by flow cytometry. Data are shown as the percentage of reticulocytes and are derived from 5 mice per genotype. Error bars represent standard deviations; P = 0.85. (C) RBCs were isolated from the peripheral blood of wild-type and E2F-2-null mice, stained with CFSE, and injected into wild-type recipient animals. Peripheral blood was collected from recipient animals at the indicated time points, and the frequency of labeled cells remaining was determined by flow cytometry. Half-lives were calculated using the slope of the trend line. The graph shows results of a representative experiment using four animals of each genotype; the composite P value was calculated from 3 separate experiments.

FIG 3

E2F-2 regulates cell cycle progression during stress erythropoiesis. (A) Bone marrow cells from wild-type and E2F-2-null mice were harvested, manually counted, stained for Ter119⁺ expression, and enumerated by flow cytometry. **, P ≤ 0.01. Inset, Ter119⁺ bone marrow cells were stained with anti-annexin V and 7-AAD and analyzed by flow cytometry, and the average percentages of annexin V-positive, 7-AAD-negative cells are displayed with corresponding standard deviations (n = 3 per genotype). (B) Age- and sex-matched wild-type and E2F-2 knockout animals were administered PHZ and their spleens harvested at 7 or 11 days postinjection. Representative splenic erythroid cell maturation profiles are shown, with the relative contributions of each cell subpopulation indicated. Total Ter119⁺ splenocyte representations are listed with corresponding standard deviations (n = 3 per genotype and time point). (C) Splenocytes from mice of the indicated genotypes and treatment time points were manually counted. **, P ≤ 0.01; *, P ≤ 0.05; not significant (ns), P > 0.05. (D) Wild-type and E2F-2-null spleens were harvested as for panel B and stained for Ter119 expression, and positive cells were enumerated by flow cytometry. **, P ≤ 0.01; *, P ≤ 0.05; ns, P > 0.05. (E) Peripheral blood from animals treated as for panel A was collected and complete blood counts performed; RBC values are shown. ***, P ≤ 0.001; **, P ≤ 0.01. (F) Reticulocytes were enumerated from peripheral blood collected as for panel D using thiazole orange staining. **, P ≤ 0.01; *, P ≤ 0.05 (using one-tailed t test). (G) Spleens from animals treated as for panel B were harvested, and cells were stained for Ter119 expression then labeled with propidium iodide for DNA content measurement to model cell cycle profiles. Shown are frequencies of cells in each cell cycle phase as an average from 3 animals. (H) Untreated animals or those treated with PHZ as for panel B were administered BrdU and spleens harvested 2 h later. Cells were stained for Ter119 expression and 7-AAD and BrdU incorporation and analyzed by flow cytometry. S-phase transit time was calculated as described previously (15, 29). Frequencies of BrdU⁺ cells and S-phase durations are indicated as an average from 3 mice per genotype, with corresponding P values indicated.

FIG 4

E2F-2 regulates erythroblast enucleation. (A) Hematopoietic progenitors were harvested from wild-type and E2F-2 knockout bone marrow, differentiated to the erythroid lineage in culture for 2 days, and stained with Hoechst 33342 and anti-Ter119 antibody. The plots shown are representative from an age- and sex-matched pair, with gates outlining the enucleated cell population (Ter119⁺ and Hoechst negative). (B) Hematopoietic progenitors were obtained from E14.5 wild-type and E2F-2 knockout fetal livers and differentiated in culture from 1 to 3 days, and enucleation was analyzed as for panel A. (C) Hematopoietic progenitors (day 0) were directly stained for Ter119 expression following lineage depletion or following differentiation over the indicated time points. Shown is the average frequency of Ter119⁺ cells at each time point with corresponding standard deviations (n = 4 mice). (D) Percentage of enucleated cells in fetal liver-derived erythroid cultures differentiated for 2 days, as described for panel B. Each symbol represents an individual experiment. ****, P ≤ 0.0001. (E) Micrographs (magnification, ×40) of wild-type and E2F-2-null fetal liver erythroid cells, sorted based on CD44 versus FSC within the Ter119⁺ population to obtain RIV cells.

FIG 5

E2F-2 regulates cell cycle- and chromosome organization-related genes but does not alter cell cycle kinetics in terminally maturing erythroid subpopulations. (A) Hematopoietic progenitors were isolated from wild-type and E2F-2 knockout fetal livers (E14.5) and either harvested for RNA or differentiated in culture. Orthochromatic erythroblasts were collected and RNA harvested for RNA sequencing. Differentially expressed genes between E2F-2 knockout and wild-type samples were defined as those with a log₂ fold change of ≥1. (B) Functional annotation analysis of differentially expressed genes in E2F-2 knockout erythroblasts was performed (NIH DAVID tool [30]), and shown are nonsynonymous, Gene Ontology (GO) biological process terms with a Benjamini-Hochberg (FDR) score of <1.0E−6. (C) Wild-type and E2F-2 knockout bone marrow cells were harvested, stained for Ter119 and CD44 expression, fixed, and labeled with propidium iodide for DNA content measurement for cell cycle profiles. Representative cell cycle profiles are shown for RII and RIII subpopulations. Cell cycle distributions are listed with standard deviations from three age- and sex-matched mice per genotype. (D) Left, wild-type fetal liver-derived progenitors were differentiated in culture to erythroid cells and treated with Cdk inhibitors (purvalanol A or roscovitine) during the final 14 h of the differentiation time course. Cells were collected and assayed for enucleation as for Fig. 4. Representative plots are shown. Right, fetal liver-derived progenitors were differentiated in culture for 1 day, and Cdk inhibitors were added and left for 14 h overnight. Cells were lysed, and histone kinase assays were performed.

FIG 6

E2F-2 regulates nuclear condensation during erythroid enucleation. (A) Fetal liver-derived progenitor cells from wild-type and E2F-2-null mice were differentiated for 2 days in culture. Cells were placed on polylysine-coated coverslips by cytospins and stained with DAPI. Confocal z-stack images were acquired, and the nuclear area was calculated from maximally projected images; each symbol represents an individual nucleus. Approximately 400 nuclei per genotype were analyzed from at least two biological replicates (P < 0.0001 using the Mann-Whitney test). (B) Wild-type and E2F-2-null fetal liver-derived erythroid cells were fixed, stained with anti-Ter119 antibody and DRAQ5, and analyzed using imaging flow cytometry. Left, representative histograms of DRAQ5 area measurements for each genotype; at least 10,000 events were collected per experimental condition. Right, representative bright-field and DRAQ5 (red)/Ter119 (yellow) images of wild-type and E2F-2-null erythroid cells within indicated erythroid subpopulations, gated using Ter119 and cell size as described previously (37). Mean nuclear area measurements for each genotype are displayed on the right with corresponding P values. (C) Wild-type and E2F-2 knockout fetal liver-derived erythroid cells were cultured, differentiated, fixed, and stained with Alexa Fluor-488-conjugated phalloidin and DAPI. Cells were imaged, and actin ring formation was counted. Left, frequencies of cells with an actin ring are shown for both genotypes; over 500 cells were counted per genotype from 4 separate experiments (P = 0.142). Right, representative images of actin rings (arrowheads) in wild-type and E2F-2 knockout cells. (D) Wild-type and E2F-2-null fetal liver-derived erythroid cells were differentiated for 1 or 2 days, fixed, and stained as for panel B. A minimum of 10,000 events per experimental condition were collected using imaging flow cytometry. Delta centroid values were calculated and corrected for variations in nuclear size by dividing by the DRAQ5 radius. ****, P ≤ 0.0001.

FIG 7

Cit expression is directly regulated by E2F-2 during erythroblast differentiation and controls nuclear condensation. (A) Calculated reads per kilobase per transcript per million mapped reads (RPKM) from RNA sequencing (described for Fig. 5A) for Cit are shown; values represent averages from two biological replicates, and error bars represent standard deviations. (B) Fetal liver-derived hematopoietic progenitors and differentiated, purified Ter119⁺ cells of the indicated genotypes were harvested and immunoblotted for CRIK. A representative Western blot is shown; HSC70 was used as a loading control. (C) E2F-2 chromatin immunoprecipitation was performed on wild-type fetal liver-derived erythroid cells. Real-time quantitative PCR was performed with primers 10 kb upstream of Cit (negative control) and at Cit and Ccne1 promoters. Ccne1 is shown as a canonical E2F target for comparison (26). Values shown are averages from 4 experiments with corresponding standard deviations. **, P ≤ 0.01; *, P ≤ 0.05. (D) β-Estradiol was added to G1-ER cells to induce differentiation over 24 h. Cells were removed during the course of differentiation, lysed, and immunoblotted for CRIK; HSC70 is shown as a loading control. Undifferentiated G1-ER cells were retrovirally transduced with MigR1-Cit and immunoblotted in parallel for expression comparison. The splice mark indicates irrelevant samples from the same gel. (E) G1-ER cells were retrovirally transduced with a control shRNA (shCtrl) or one of two hairpins designed to target E2f2. Cells were immunoblotted for E2F-2 and CRIK expression, with β-actin shown as a loading control. Samples were run on a single gel. (F) G1-ER cells were transduced as described for panel E, fixed, stained with anti-Ter119 antibody and DRAQ5, and analyzed using imaging flow cytometry (≥10,000 events per sample). Cells were gated based on GFP and DRAQ5 staining and spherical aspect ratio. Left, nuclear area was measured as the area of the Draq5 image as for Fig. 6 (****, P ≤ 0.0001). Middle, nuclear polarity was measured by imaging flow cytometry as for Fig. 6. One hairpin, shE2f2-B, showed a significant reduction in polarity (P < 0.0001). Right, transduced G1-ER cells were imaged for actin ring formation as for Fig. 6. Actin rings were enumerated in at least 150 cells per sample; no significant differences were found between shCtrl and shE2f2 samples. (G) G1-ER cells were retrovirally transduced with shCtrl or one of two hairpins targeting Cit. Cells were harvested for RNA, and quantitative reverse transcription-PCR was performed. Left, inset, Cit mRNA expression values relative to Rn18s. Left, nuclear area (P < 0.0001). Middle, nuclear polarity; no-shCit samples showed a significant reduction in polarization. Right, actin rings were enumerated; no significant differences were observed.

FIG 8

CRIK activity regulates erythroblast condensation and enucleation. (A) Wild-type and E2F-2^−/− fetal liver-derived Ter119⁺ erythroid cells were immunoblotted for total myosin regulatory light chain 2 (MLC2) and phospho-T18/S19 MLC2 (pMLC2). Shown is a representative blot; pMLC2 abundance was quantified in ImageJ relative to total MLC2 levels, with normalized values indicated. (B) Wild-type fetal liver-derived cells were differentiated for 2 days in culture. Kinase inhibitors (A-674563 and RAF265; concentrations are listed in Materials and Methods) were added during the final 14 h of culture. Enucleation was assayed as for Fig. 4. (C) Wild-type fetal liver-derived erythroid cells were treated with kinase inhibitors and an AKT-specific inhibitor (MK-2206) for panel B. Cells were immunoblotted for total and phospho-MLC2 expression. Representative blots are shown, with the splice mark indicating removal of lanes containing cell lysates treated with lower concentrations of MK-2206. Bottom, immunoblots were quantified as for panel A; error bars represent standard deviations. ***, P ≤ 0.001; *, P ≤ 0.05; there was a nonsignificant difference in pMLC2 abundance with MK-2206. (D) Wild-type fetal liver-derived erythroid cells were treated as described for panel B. Cells were fixed, stained with anti-Ter119 antibody and DRAQ5, and analyzed using imaging flow cytometry. Nuclear area was measured by the area of the DRAQ5 stain. ****, P ≤ 0.0001; ns, P > 0.05. Nuclear polarity was measured as for Fig. 6. ****, P ≤ 0.0001; ns, P > 0.05. (E) Wild-type and E2F-2^−/− fetal liver-derived hematopoietic progenitors were transduced with empty vector (MigR1) or MigR1-Cit. Cells expressing MigR1-Cit exhibited approximately 2-fold overexpression of Cit, measured by quantitative reverse transcription-PCR. Transduced cells were differentiated for 2 days, and enucleation was measured within the GFP⁺ population. E2F-2^−/− cells showed a significant increase in enucleation compared to MigR1 controls (*, P = 0.03 [three separate experiments]); a significant difference was not observed for Cit-expressing wild-type cells compared to MigR1.