A requirement for Lim domain binding protein 1 in erythropoiesis

To examine Ldb1 expression in the hematopoietic lineage, adult bone marrow and fetal liver cells were stained intracellularly with affinity-purified polyclonal Ldb1 antisera. FACS analysis revealed that Ldb1 levels are highest in lineage^lo/−Sca-1⁺c-kit⁺ (LSK) cells (Fig. S1 A), which are known to contain hematopoietic progenitors including HSCs (Ikuta and Weissman, 1992; Morrison et al., 1995). Indeed, in a separate study we found that Ldb1 is critical and continuously required for HSC maintenance (unpublished data). Slightly lower levels of Ldb1 were detected in lineage-committed progenitor populations (common myeloid progenitors [CMPs], megakaryocyte-erythroid progenitors [MEPs], and granulocyte-macrophage progenitors [GMPs]; Fig. S1 A). Interestingly, Ldb1 was also expressed at all stages of erythroid development but was strongly down-regulated in mature myeloid (Gr1⁺) and lymphoid (B and T) cells (Fig. S1, B and C), suggesting that Ldb1 may also have an important and specific function in erythropoiesis (Fig. S1 C).

It was reported previously that benzidine-positive cells are undetectable in ldb1^−/− yolk sacs, suggesting absent or defective primitive erythropoiesis (Mukhopadhyay et al., 2003). In agreement with this observation, in vitro–cultured embryoid bodies derived from ldb1^−/− ES cells did not generate erythroid lineage cells, although myeloid lineage cells could be generated in the absence of Ldb1 (Fig. S2 A). Mesoderm/hemangioblast marker genes and genes involved in Ldb1 complex formation (lmo2, scl, gata2, and gata1) were expressed in day 5 ldb1^−/− embryoid bodies, but no expression of erythroid-specific genes was detectable at any time point analyzed (Fig. S2, B and C; and not depicted). Consistent with these data, there was no evidence of blood formation in E9 ldb1^−/− embryos (Fig. 1 A, left). E9 ldb1^−/− yolk sac blood islands contained extremely low numbers of cells (Fig. 1 A, middle) that included rare myeloid blasts but no erythroblasts (Fig. 1 A, right). To evaluate the differentiation potential of progenitor cells in ldb1^−/− yolk sacs, single cell suspensions from E9 yolk sacs were cultured in vitro in hematopoietic differentiation medium. Cells from E9 ldb1^−/− yolk sacs generated ∼10-fold fewer total colonies than cells from control yolk sacs (Fig. 1 B and not depicted). In addition, although control yolk sac cells gave rise to both erythroid and nonerythroid colonies that were apparent by day 6 of culture, ldb1^−/− yolk sac cells generated only nonerythroid colonies consisting of myeloid blasts or mature macrophages that were apparent only after 10–14 d of culture (Fig. 1, B and C and not depicted). Together, these results demonstrate that hematopoietic specification is impaired and primitive erythropoiesis is abrogated in the absence of ldb1.

To examine the role of Ldb1 in definitive erythropoiesis, we used a conditional deletion allele of ldb1 (ldb1^fl) generated in ES cells by homologous recombination (Suleiman et al., 2007). Although ldb1^−/− mice die between E9 and E10, Ldb1^fl/fl mice were viable and fertile and exhibited no hematopoietic defects (unpublished data). Cre-recombinase–mediated deletion of the ldb1^fl alleles in oocytes recapitulated the early embryonic lethality observed in ldb1^−/− mice (unpublished data). Because our results indicated that Ldb1 was required for normal hematopoietic specification and primitive erythropoiesis, we generated mice in which conditional deletion of ldb1 was mediated by the tyrosine kinase with immunoglobulin and epidermal growth factor homology domains 2 (Tie2) Cre transgene (Kisanuki et al., 2001). In a previous study, it was shown that HSCs are generated in normal numbers and primitive erythropoiesis is partially rescued in Tie2Cre scl^fl/fl embryos but that Tie2Cre-mediated gene deletion occurs in fetal liver hematopoietic progenitor cells that give rise to definitive erythrocytes (Schlaeger et al., 2005).

To increase the probability that ldb1 would be completely deleted after Cre expression, one Ldb1 allele was deleted in the germline to generate ldb1^+/Δ mice. Tie2Cre ldb1^+/Δ mice were then crossed with ldb1^fl/fl mice to generate Tie2Cre ldb1^fl/Δ embryos. Tie2Cre-mediated deletion of the ldb^fl allele in fetal liver cells was confirmed by PCR of genomic DNA (Fig. S3 C). Tie2Cre ldb1^fl/Δ embryos developed normally through E12. However, by E12.5, all Tie2Cre ldb1^fl/Δ embryos exhibited widespread hemorrhage and edema (Fig. S3 A). Fetal liver size and cellularity were markedly decreased in E11.5–14.5 Tie2Cre ldb1^fl/Δ mice (Fig. S3, A and B) and no embryos survived beyond E14.5.

EryPs were detected in the blood of all E12.5 Tie2Cre ldb1^fl/Δ embryos; however, a maturation defect was apparent from the high percentage of mitotic and basophilic cells and the high frequency of immature erythroid blasts (Fig. 2 A). This phenotype was strikingly similar to that observed in E12.5 Tie2Cre scl^fl/fl mice (Schlaeger et al., 2005). The reduced cellularity in Tie2Cre ldb1^fl/Δ fetal livers was due primarily to low numbers of erythroid (Ter119⁺) cells (Fig. 2 B) and was most obvious on E13.0 and thereafter, when the definitive erythroid pool was rapidly expanding in control embryos but not in Tie2Cre Ldb1^fl/Δ embryos (Fig. 2 C and Fig. S3 B). Importantly, although the total number of fetal liver cells was decreased in E12.5 Tie2Cre ldb1^fl/Δ mice, the number of LSKs was normal at this time point indicating that the erythropoietic defects were not secondary to HSC depletion (Fig. 2 B). Analysis of downstream populations revealed a reduction in Lin⁻Sca1⁻c-kit⁺ (LK) fetal liver cells in Tie2Cre ldb1^fl/Δ mice, reflecting a marked decrease in the number of MEP (Fig. 2, B and C). In contrast, the number of CMP and GMP fetal liver LK cells was only slightly reduced in E12.5 Tie2Cre ldb1^fl/Δ mice, pointing to an early erythroid-specific defect (Fig. 2, B and C). Immature CD71^hiTer119^low–med proerythroblasts and total Ter119⁺ cells were also strongly reduced in Tie2Cre ldb1^fl/Δ embryos (Fig. 2, B and C).

To evaluate the developmental potential of E12.5 Tie2Cre ldb1^fl/Δ fetal liver cells, we next performed in vitro methylcellulose progenitor assays. After 8 d of culture, colonies were identified by visual inspection and enumerated. The number of erythroid colonies originating from Tie2Cre Ldb1^fl/Δ fetal liver cells was strongly reduced relative to controls (Fig. 2, D and E). In addition, erythroid colonies from Tie2Cre ldb1^fl/Δ mice were derived almost exclusively from EryP–colony forming cells (CFCs; Fig. 2 E). In contrast, most erythroid colonies from control fetal livers were derived from definitive burst forming unit–erythroid (BFU-E) progenitors (Fig. 2, D and E; and not depicted). The near absence of definitive BFU-E colonies in Tie2Cre Ldb1^fl/Δ fetal liver cultures is consistent with the reduced number of fetal liver erythroid progenitors (MEPs; Fig. 2, B and C). In contrast, the number of nonerythroid CFU-macrophage (M) and CFU-GM progenitors were not significantly reduced in Tie2Cre ldb1^fl/Δ fetal livers, identifying a selective requirement for ldb1 in erythroid development (Fig. 2, D and E).

To determine if Ldb1 is also required for adult erythropoiesis, we generated Mx1Cre ldb1^fl/fl mice so that ldb1 deletion could be induced by injection of polyinosinic-polycytidylic acid (pI:pC; Kühn et al., 1995). As shown in Fig. 3 A, most Mx1Cre ldb1^fl/fl mice, but not control ldb1^fl/fl mice, died within 20 d of pI:pC injection. As reported previously, injection of pI:pC causes a transient anemia (Hall et al., 2003), and this was observed in both control and Mx1Cre Ldb1^fl/fl mice (Fig. 3 B). However, although the hematocrit normalized 14 d after pI:pC injection in control mice, the hematocrit continued to decline in Mx1Cre ldb1^fl/fl mice (Fig. 3 B). The total number of bone marrow cells was also dramatically decreased in Mx1Cre ldb1^fl/fl mice after pI:pC-mediated Ldb1 deletion (unpublished data). Similar to Tie2Cre ldb1^fl/Δ mice, Lin^lo/−Sca1⁻c-kit^hiCD34^loFcγR^lo MEP populations, as well as CD71⁺Ter119⁺erythroblasts, were significantly reduced in pI:pC-treated Mx1Cre ldb1^fl/fl mice (Fig. 3 C and Fig. S4 A). BFU-E and mixed (erythroid/myeloid) colonies were virtually absent in day 8 methylcellulose cultures of bone marrow cells from pI:pC-treated Mx1Cre Ldb1^fl/fl mice (Fig. S4, B and C). Similar to the results obtained with Tie2Cre ldb1^fl/Δ fetal liver cells, granulocyte/macrophage colony formation was much less affected by inactivation of ldb1 in adult bone marrow cells (Fig. S4 C).

FACS analysis of the few CD71⁺Ter119⁺erythroblasts present in Mx1Cre Ldb1^fl/fl mice after pI:pC-induced deletion of ldb1 revealed an increase in the percentage of apoptotic (annexin V⁺) cells as well as a reduced percentage of cycling cells, indicating a requirement for Ldb1 in cell survival (Fig. 3 C and Fig. S4 A). To determine if Ldb1 is also important for megakaryocyte survival, sternal sections were obtained from pI:pC-treated Mx1Cre ldb1^fl/fl and control ldb1^fl/fl mice and stained with hematoxylin and eosin (H&E). Strikingly, as early as 8 d after pI:pC injection, no megakaryocytes were detectable in sternum sections from pI:pC-treated Mx1Cre ldb1^fl/fl mice, pointing to a critical role for Ldb1 in megakaryocyte survival (Fig. 3 D). In agreement with this finding, platelet counts in the peripheral blood of Mx1Cre ldb1^fl/fl mice were markedly reduced relative to control ldb1^fl/fl mice 8 d after pI:pC injection (Fig. 4 B).

To evaluate the impact of Ldb1 down-regulation on the expression of erythroid- and megakaryocytic-specific genes, we analyzed gene expression in a Friend mouse erythroleukemia (MEL) cell line expressing a stable construct encoding ldb1 shRNA (Song et al., 2007). Differentiation of the MEL cell line was stimulated by DMSO, which has been shown to induce the expression of erythroid- and megakaryocyte-specific transcripts (Friend et al., 1973). As shown in Fig. 4 A, ldb1 transcripts were reduced by 20–50% in both uninduced and DMSO-induced Ldb1 shRNA MEL cells. Interestingly, knockdown of ldb1 strongly inhibited the induced expression of several erythroid genes including α- and β-globin, epb4.2, slc4α1, and alas2 (Fig. 4 A). In addition, expression of the megakaryocyte-specific transcript itga2b was also markedly decreased (Fig. 4 A). These results confirm and extend previous studies supporting a critical requirement for Ldb1 in erythroid gene expression (Xu et al., 2003; Anguita et al., 2004; Lahlil et al., 2004; Song et al., 2007). The dramatic increase in apoptotic erythroblasts after in vivo deletion of ldb1 (Fig. 3 C) also prompted us to examine the effect of ldb1 knockdown on the expression of erythroid prosurvival factors. As shown in Fig. 4 B, expression of three prosurvival genes, epor, bcl-xl, and sox6, was also significantly decreased in ldb1 shRNA MEL cells, especially at later time points in cell culture when these genes were highly induced in control cells (Fig. 4 B). These results indicate that Ldb1 protein complexes regulate the expression of prosurvival as well as erythroid- and megakaryocyte-specific genes.

In the present study, we relate data demonstrating a critical, continuous, and specific role for Ldb1 in both erythropoiesis and megakaryopoiesis. These results are consistent with the high expression of Ldb1 in erythroid progenitors (Fig. S1) and with the observation that deletion of ldb1 leads to a severe developmental block at the common megakaryocyte/erythroid progenitor stage (Fig. 2 C and Fig. 3 C). Deletion of ldb1 severely impacted the erythroid/megakaryocyte lineages but had little effect on the development of myeloid cells. Using B cell– or T cell–specific Cre transgenes, we also found no evidence of a requirement for ldb1 in lymphocyte development (unpublished data). A similar nonmandatory role for lmo2 in T cell development was also demonstrated (McCormack et al., 2003).

Our results indicate that an important function of Ldb1 during erythropoiesis is to induce the expression of lineage-specific and prosurvival genes, presumably via its documented role as a key subunit of Ldb1/Lmo2/Scl/E2A/Gata-1 multimeric complexes (Fig. 4). Ldb1 nucleated complexes have been shown to regulate the expression of several erythroid genes including α- and β- globin, epb4.2, and gypA (Xu et al., 2003; Anguita et al., 2004; Lahlil et al., 2004; Song et al., 2007). In addition, genome-wide profiling studies have detected Ldb1 at most Gata-1–bound DNA elements in erythroid cell lines, suggesting that Ldb1 complexes represent a major instrument for Gata-1–mediated gene activation (Tripic et al., 2009; Soler et al., 2010). In this paper, we provide evidence that down-regulation of Ldb1 severely impairs the induced expression of multiple erythroid- and/or megakaryocyte-specific genes in erythroleukemia cells. Interestingly, the expression of genes encoding other Ldb1 complex subunits, including lmo2, gata-1, and scl, was not as severely affected by down-regulation of ldb1, indicating that failure to assemble functional Ldb1 complexes, rather than reduced expression of Ldb1 complex subunits, is responsible for the reduction in erythroid gene expression (Fig. 4). In summary, these results establish an essential function for Ldb1 during erythrocyte and megakaryocyte development and survival and suggest that Ldb1-nucleated complexes represent an important mechanism by which Gata-1 activates gene transcription in these lineages.

Tie2Cre and Mx1Cre transgenic mice were purchased from The Jackson Laboratory. All mice were bred and maintained in a National Institutes of Health (NIH) Research Animal Facility in accordance with the specifications of the Association for Assessment and Accreditation of Laboratory Animal Care. Mouse protocols were approved by the NIH Animal Care and Use Committee.

Polyclonal rabbit anti-sera were raised against Ldb1 peptide MLDRDVGPTPMYPPTYLEPC-amide, affinity purified, and conjugated with Alexa Fluor 647 (A20186; Invitrogen). Purified rabbit IgG (AB-105-C; R&D Systems) conjugated in the same way served as background control. Freshly isolated single cell suspensions were surface stained, fixed with 2% formaldehyde, permeabilized with 0.5% Triton X-100 (Bio-Rad Laboratories) in PBS, stained with Alexa Fluor 647–conjugated rabbit anti-Ldb1 or rabbit IgG, and analyzed by flow cytometry.

The generation of ldb1^+/− ESCs (Mukhopadhyay et al., 2003), ldb1^−/− ESCs (Hwang et al., 2008), and ldb1^fl ESCs (Suleiman et al., 2007) has been described previously.

Undifferentiated ESCs were cultured in modified DME (Invitrogen) containing 15% FCS and LIF on mitomycin-treated embryonic fibroblast feeders. To generate embryoid bodies, ESC cultures were trypsinized until 5–10 cell clusters were formed. Trypsinization was stopped by addition of 10 ml IMDM medium containing 10% fetal calf serum and cells were incubated for 45 min at 37°C to allow reattachment of embryonic fibroblasts. ESCs were transferred to 10-cm bacterial Petri dishes and incubated in IMDM plus 15% FCS for 3–6 d in the presence of 40 ng/ml SCF and observed for the formation of embryoid bodies. Embryoid bodies generated in the primary culture were harvested by gravity sedimentation and plated in secondary methylcellulose cultures containing 50 ng/ml of recombinant mouse (rm) SCF, 10 ng/ml rm IL-3, and 10 ng/ml rh IL-6 or 50 ng/ml rm SCF and 3 U/ml rh Epo (STEMCELL Technologies Inc.).

Assays for hematopoietic progenitors in fetal liver or adult bone marrow were performed by culturing enumerated cells in methylcellulose-based IMDM medium containing 3 U/ml of recombinant human (rh) Epo, 50 ng/ml rm SCF, 10 ng/ml rm IL-3, and 10 ng/ml rh IL-6 (STEMCELL Technologies Inc.). For cultures of fetal liver cells and bone marrow cells, 2.5 × 10⁴–10⁵ cells were plated in duplicate and scored for colony formation at day 8. For yolk sac cultures, single cell suspensions of E9 yolk sacs were made, divided in half, and plated in duplicate.

Conjugated antibodies, including mouse and hamster IgG isotype controls, were purchased from BD or eBioscience. The lineage marker (Lin) mixture for fetal liver cells included the following biotinylated antibodies: CD3-ε (145-2C11), CD4 (GK1.5), CD8-α (53–6), CD8-β (53–5.8), TCR-β (H57-597), TCR-γδ (GL3), CD19 (1D3), B220 (RA3-6B2), Gr1 (RB6-8C5), Ter119, CD49b (Dx5), and NK1.1 (PK136). Lin mixture for adult bone marrow cells included all of these plus Mac-1 (M1/70). Other conjugated antibodies used for surface staining included: CD34 (RAM34), CD43 (S7), CD71 (C2), c-kit (2B8), Sca1 (D7), and FcγIII/IIR (2.4G2). Biotinylated primary antibodies were detected by incubation of antibody-coated cells with streptavidin-PerCP-Cy5.5 or APC-Cy7 in a two step staining procedure. Percentage of apoptotic cells was determined by annexin V (BD) staining according to the manufacturer’s instructions. For cell cycle analysis, cells were surface stained, fixed in 2% formaldehyde, permeabilized in 0.1% NP-40, and then stained with DAPI (Invitrogen). Data were acquired with a FACSCalibur or LSR II flow cytometer and analyzed with FlowJo software (Tree Star, Inc.).

The MEL cell line expressing ldb1 shRNA has been described previously (Song et al., 2007). For gene expression studies, total cell RNA was isolated using PicoPure RNA isolation kit (Arcturus), and 100 ng of each RNA sample was reverse transcribed using SuperScript first-strand synthesis system (Invitrogen) and assayed by RT-PCR. Transcript quantification was performed with a LightCycler 480 (Roche). Duplicates were run for each sample in a 96-well plate. β-Actin was used as the endogenous reference gene. All quantitative RT-PCR reactions were run in three independent experiments. The relative quantification method was used, with the ratio of the mRNA level for the gene of interest normalized to the level of β-actin and the mean of control bone marrow samples as the calibrator value. The specificity of the products was confirmed based on melting curves and electrophoresis.

PCR genotyping of Tie2Cre Ldb^fl/Δ mice and Mx1Cre Ldb^fl/fl mice was performed on tail DNA. Primers used for detection of the Cre transgene were 5′-CGATGCAACGAGTGATGAGG-3′ and 5′-GACTTGCTGTCACTTGGTCGT-3′. For Tie2Cre Ldb^fl/Δ mice, a combination of the following three oligonucleotide primers were used to detect the wild-type Ldb1, the Ldb1^fl, and the Cre-mediated deleted (Ldb^Δ) alleles: Ldb1S 5′-CAGCAAACGGAGGAAACGGAAGATGTCAG-3′, Ldb1AS 5′-CTTATGTGAC­CACAGCCATGC­ATGCATGTG-3′, and Ldb1A 5′-TCAGGCTGGCCTTTAAACCTAA-3′. Ldb1S-Ldb1AS amplified a 320-bps wild-type allele fragment and 445-bps floxed allele fragment. Ldb1A-LdbAS amplified a 390-bps Ldb1 Cre-mediated recombination allele fragment. For Mx1Cre ldb1^fl/fl mice, a combination of the following three oligonucleotide primers were used to detect the wild-type Ldb1, the Ldb1^fl, and the Cre-mediated deleted (ldb^Δ) alleles: Ldb1A 5′-TCAGGCTGGCCTTTAAACCTAA-3′, Ldb1B 5′-TGGGACTACAAGGCTGAGAACA-3′, and Ldb1C 5′-TGGCTGAGCTTATGTGACCA-3′. Ldb1B-Ldb1C amplified a 458-bp wild-type allele fragment and a 534-bp floxed allele fragment. Ldb1A-LdbC amplified a 342-bp Ldb1 deleted allele fragment.

To inactivate ldb1 in adult hematopoietic cells, 250 µl pI:pC (1 mg/ml; GE Healthcare) was injected intraperitoneally into each mouse on days 1, 3, and 5 (the day of the first injection was designated day 1). Mice were sacrificed on day 8 unless otherwise noted. Cells obtained from bone marrow or spleen were either analyzed directly or cultured in methylcellulose to analyze their colony forming potential.

Fig. S1 shows Ldb1 expression in hematopoietic cells. Fig. S2 shows impaired erythroid developmental potential of ldb1^−/− ESCs. Fig. S3 shows defective erythropoiesis in Tie2Cre ldb^fl/Δ embryos. Fig. S4 shows that Ldb1 is continuously required for survival and expansion of erythroblasts in adults.

We thank Heiner Westphal for providing mouse and cell lines and Dalal El-Khoury and Tsadok Cohen for technical support.

This research was supported by the Intramural Research Program of the National Institutes of Health, Eunice Kennedy Shriver National Institute of Child Health and Human Development, and National Institutes of Diabetes and Digestive and Kidney Diseases.

The authors have no conflicting financial interests.