Corepressor Rcor1 is essential for murine erythropoiesis

Abstract

The corepressor Rcor1 has been linked biochemically to hematopoiesis, but its function in vivo remains unknown. We show that mice deleted for Rcor1 are profoundly anemic and die in late gestation. Definitive erythroid cells from mutant mice arrest at the transition from proerythroblast to basophilic erythroblast. Remarkably, Rcor1 null erythroid progenitors cultured in vitro form myeloid colonies instead of erythroid colonies. The mutant proerythroblasts also aberrantly express genes of the myeloid lineage as well as genes typical of hematopoietic stem cells (HSCs) and/or progenitor cells. The colony-stimulating factor 2 receptor β subunit (Csf2rb), which codes for a receptor implicated in myeloid cytokine signaling, is a direct target for both Rcor1 and the transcription repressor Gfi1b in erythroid cells. In the absence of Rcor1, the Csf2rb gene is highly induced, and Rcor1(-/-) progenitors exhibit CSF2-dependent phospho-Stat5 hypersensitivity. Blocking this pathway can partially reduce myeloid colony formation by Rcor1-deficient erythroid progenitors. Thus, Rcor1 promotes erythropoiesis by repressing HSC and/or progenitor genes, as well as the genes and signaling pathways that lead to myeloid cell fate.

Figure 1

Targeted disruption of murine Rcor1 results in embryonic lethality. (A) Gene targeting strategy for the Rcor1 locus: schematic representation of the wild-type (WT) Rcor1⁺ allele, targeting vector, Rcor1^flox allele (Floxed), and mutant Rcor1^– allele (KO). Rcor1^flox/+ mice were crossed with Meox2-Cre and subsequently outcrossed to obtain Rcor1^+/− mice. Arrows indicate positions of the A3, A2, and B2 genotyping primers. (B) Top panel: PCR analysis with primers A3 and B2 to resolve WT and mutant Rcor1 alleles (KO) in E13.5 embryos. Bottom panel: PCR analysis with primers A2 and B2 to resolve WT and floxed (F) Rcor1 allele in E13.5 embryos. (C) Genotypes resulting from Rcor1 heterozygote matings. (D) Western blot analysis showing the expression levels of Rcor1, -2, and -3 proteins in different tissues of E13.5 WT and KO embryos. FRT, flippase recognition target sites; GAPDH, loading control; pA, polyadenylation site; PGK-Neo, neomycin resistance cassette; SD, splice donor.

Figure 2

Rcor1 null embryos exhibit defective embryonic erythropoiesis. (A) E13.5 control (Ctrl) and mutant (Mut) littermates. Note pale liver in mutant (arrow). Scale bar: 1 mm. (B) Comparison of E13.5 control and mutant fetal livers. The mutant fetal liver is smaller and paler than the control fetal liver. Scale bar: 1 mm. (C) May-Grunwald Giemsa staining of E13.5 fetal liver cytospin preparations. The control liver contains early erythroid progenitors (i: BFU-E or CFU-E like cells; ii: proerythroblast) and late erythroid precursors (iii: early basophilic erythroblast; iv: late basophilic erythroblast; v: orthochromatophilic erythroblast). The mutant fetal liver contains primarily early-stage erythroid progenitors. Scale bar: 10 μm. (D) May-Grunwald Giemsa staining of E15.5 peripheral blood smears showing enucleated definitive erythrocytes (arrow) in the control that are lacking in the mutant. In the mutant, nearly all the circulating blood cells are primitive wave erythroid cells of normal appearance. Scale bar: 20 μm. (E) Mean of enucleated RBC frequency in control (n = 5) and mutant (n = 3) E15.5 peripheral blood. Error bars show standard deviation (SD). ***P < .001. Images for (A-B) were taken with a Zeiss SteREO Lumar.V12 microscope, a Neo-Lumar S ×0.8 FWD 80-mm objective and an AxioCam HRc camera. Images for (C-D) were acquired with a Zeiss Axiovert S-100 (Carl Zeiss), an AxioCam HRc camera, and either (C) a Zeiss plan-neofluar ×100/1.30 oil lens or (D) a Zeiss plan-neofluar ×40/1.30 oil lens.

Figure 3

Erythropoietic differentiation in Rcor1 knockout mice is blocked at the transition from proerythroblast to basophilic erythroblast. (A) Flow cytometry profiles of fetal liver cells stained with CD71 and TER119. R0-R5: Gates of different erythroblast populations according to their expression levels of CD71 and TER119. R0 contains mixed populations of hematopoietic stem cell (HSC) and early progenitors, such as CMP, GMP, and MEP; R1 consists of mostly immature RBC progenitors, including BFU-E and CFU-E; R2 comprises mainly proerythroblasts and early basophilic erythroblasts; R3 contains early and late basophilic erythroblasts; R4 is composed of chromatophilic and orthochromatophilic erythroblasts; and R5 contains late orthochromatophilic erythroblasts and reticulocytes. Note that the transition from R2 to R3 is arrested in the mutant fetal liver. (B) May-Grunwald Giemsa staining showing similar morphology of FACS-sorted mutant and control E14.5 fetal liver cells. Scale bar: 20 μm. (C) Schematic diagram showing transplant of fetal liver cells that express CD45.2 and β-globin haplotype Hbbs (donor) into irradiated mice double congenic for CD45.1 and β-globin haplotype Hbbd (host). (D) Donor cell contribution to circulating leukocytes of adult WT mice transplanted with 2 million E13.5 mutant or control fetal liver cells. (E) Hemoglobin electrophoresis analysis indicates that mutant fetal liver cells cannot generate RBCs after transplantation into WT adult mice. (F) Immunostaining for Rcor1 protein in FACS-sorted E14.5 fetal liver cells. Mutant R0 cells serve as a negative control (NC). Nuclei labeled with 4′,6 diamidino-2-phenylindole (DAPI). Scale bar: 20 μm. Images in this figure were acquired with a Zeiss Axiovert S-100, a Zeiss plan-neofluar ×63/1.25 oil lens, and an AxioCam HRc camera. D, donor; H, host.

Figure 4

In vitro colony-forming assays reveal a cell-autonomous defect in erythropoiesis and enhanced myeloid potential in Rcor1-deficient erythropoietic progenitors. (A-B) Numbers of colonies generated from (A) FACS-sorted R2 fetal liver cells or (B) R1 fetal liver cells in methylcellulose culture. Results from 4 experiments for R2 and 5 experiments for R1 are shown (mean ± SD). Equal numbers of control and mutant cells were plated in each experiment. (C) Representative myeloid colonies generated from methylcellulose cultures of mutant R2 cells. Scale bar: 200 μm. (D) Representative cells from cytospin preparations of mutant myeloid colonies stained with May-Grunwald Giemsa. (i) Macrophage, (ii-iii) mast cells, and (iv-vii) granulocytes. Scale bar: 10 μm. (E) Schematic diagram for generating inducible Rcor1 deletions in R1 fetal liver (FL) cells. (F) Numbers of colonies generated from FACS-sorted R1 Mx1-Cre; Rcor1^flox/– fetal liver cells cultured in methylcellulose with or without IFN-α. Results from 6 experiments are shown (mean ± SD). All images were acquired by using a Zeiss Axiovert S-100 and AxioCam HRc cameras. A Zeiss Fluar ×10/0.5 objective was used for the images in (C) and a Zeiss plan-neofluar ×100/1.30 oil objective was used for the images in (D). *P < .05; **P < .01; ***P < .001. Myeloid colonies, colonies containing mast cells, granulocytes, and/or macrophages; N.S., nonsignificant.

Figure 5

Knockout of Rcor1 results in the upregulation of myeloid genes. (A) Volcano plot from RNA-Seq showing all expressed genes. The y-axis shows statistical significance. A false discovery rate (FDR) –adjusted P value of .05 is 1.3 on this scale. The x-axis shows the magnitude of change (mutant/control). Red dots, all upregulated genes with fold change ≥2 (P < .05) and an adjusted tag number of >50 (average mutant reads minus average control reads; for details, see “Materials and methods”). Green dots, all downregulated genes with fold change ≤2 (P < .05) and an adjusted tag number of >50 (average control reads minus average mutant reads). Blue dots, all other genes. (B) Confirmation of representative RNA-Seq results by qPCR. (C) Correlation between qPCR and RNA-Seq data based on data points from (B). (D) GSEA showing that genes that are upregulated in mutant Rcor1^−/− cells correlate positively with genes that are enriched in monocytes. Heat map: genes are ranked according to their relative expression levels in monocytes to “other” hematopoietic cell types (LT-HSC, NK cells, monocytes, granulocytes, erythrocytes, naive CD4 cells, naive CD8 cells, active CD4 cells, active CD8 cells, B cells). Red, highly enriched in monocytes. Black vertical lines represent single genes from the upregulated Rcor1^−/− gene set, which are positioned with respect to the ranked list of the reference cell type expression data set. The green line is the running sum for calculating the enrichment score (ES). Positive and negative ES indicate enrichment at the top and bottom of the ranked reference list, respectively. ES is normalized (NES) on the basis of differences in gene set size and correlations between gene sets and is used to compare GSEA results across experiments. (E) GSEA showing that genes upregulated in Rcor1^−/− cells are correlated positively with genes that are highly expressed in HSCs. Heat map: genes are ranked according to their relative expression levels in HSCs relative to CMPs, GMPs, and MEPs. Colors and NES as described in (D). (F) Examples of de-repressed genes from Rcor1^−/− cells. Red colored genes have been reported to block erythropoiesis when de-repressed in RBC progenitors.

Figure 6

Rcor1 mutants exhibit a hypersensitive CSF2 signaling pathway. (A-B) Rcor1 and Gfi1b occupy the promoters of indicated target genes measured by (A) Rcor1 ChIP and (B) Gfi1b ChIP analysis in the MEL cell line. Binding is represented as fold enrichment relative to a negative region from the β-actin gene intron. For each gene, a site 2 to 8 kb away from the positive binding site served as the internal NC. Results from 3 experiments are shown (mean ± SD). ‡ indicates comparisons to β -actin: ‡P < .05; ‡‡P < .01; ‡‡‡P < .001. *Indicates comparison with internal NC: *P < .05; **P < .01; ***P < .001. (C) Rcor1 ChIP and (D) Gfi1b ChIP analysis in primary control R2 fetal liver cells (FLCs) and mutant total FLCs. Binding is represented as fold enrichment relative to a negative region from the β-actin gene intron. The mean of 2 independent experiments is shown. (E) Representative flow cytometry analysis of Csf2rb expression level in E13.5 control fetal liver (n = 10) and mutant fetal liver (n = 5). PE-conjugated anti-Csf2rb or PE-conjugated immunoglobulin G1 (isotype control) were used. (F) Western blot analysis showing that purified TER119^– mutant cells have higher levels of p-Stat5, but similar levels of total Stat5 and Jak2 protein, following treatment with CSF2. This experiment was repeated once more with the same result. (G) Colony-forming assay results showing that the Jak2 inhibitor TG101348 reduces CFU-GM colonies generated from mutant fetal liver R2 cells. Results from 4 independent control or mutant fetal livers treated with TG101348 or dimethylsulfoxide (DMSO) are shown (mean ± SD). ***P < .001.