BMP signaling balances murine myeloid potential through SMAD-independent p38MAPK and NOTCH pathways

Abstract

Bone morphogenetic protein (BMP) signaling regulates early hematopoietic development, proceeding from mesoderm patterning through the progressive commitment and differentiation of progenitor cells. The BMP pathway signals largely through receptor-mediated activation of Mothers Against Decapentaplegic homolog (SMAD) proteins, although alternate pathways are modulated through various components of mitogen-activated protein kinase (MAPK) signaling. Using a conditional, short hairpin RNA (shRNA)-based knockdown system in the context of differentiating embryonic stem cells (ESCs), we demonstrated previously that Smad1 promotes hemangioblast specification, but then subsequently restricts primitive progenitor potential. Here we show that co-knockdown of Smad5 restores normal progenitor potential of Smad1-depleted cells, suggesting opposing functions for Smad1 and Smad5. This balance was confirmed by cotargeting Smad1/5 with a specific chemical antagonist, LDN193189 (LDN). However, we discovered that LDN treatment after hemangioblast commitment enhanced primitive myeloid potential. Moreover, inhibition with LDN (but not SMAD depletion) increased expression of Delta-like ligands Dll1 and Dll3 and NOTCH activity; abrogation of NOTCH activity restored LDN-enhanced myeloid potential back to normal, corresponding with expression levels of the myeloid master regulator, C/EBPα. LDN but not SMAD activity was also associated with activation of the p38MAPK pathway, and blocking this pathway was sufficient to enhance myelopoiesis. Therefore, NOTCH and p38MAPK pathways balance primitive myeloid progenitor output downstream of the BMP pathway.

Figure 1

Conditional depletion of Smad1 and Smad5 reveals differing contributions of each to primitive hematopoiesis. Cre/loxP-mediated homologous recombination was used to target insertion of constructs containing shRNA hairpins corresponding to sequences within the MH1 domain of Smad1 alone (miSMAD1), Smad5 alone (miSMAD5), or both Smad5 and Smad1 (miSMAD51). Each hairpin construct is targeted to the identical site downstream of an operator/promoter, which allows tet-on activation via dox treatment (miSMAD1 was previously characterized). (A) Representative western blotting analysis of SMAD5 and SMAD1 protein expression in lysates derived from EBs after 24 hours of dox treatment (DOX, compared with uninduced control). (B) Quantitative densitometric analysis of samples induced with 2 μg/mL dox normalized to control β-actin levels from at least 3 separate experiments to derive mean ± SEM. (C-E) Embryoid bodies derived from parental AinV18 cells, miSMAD5, miSMAD51, or miSMAD1 ESCs were induced with dox on day 2 (dox d2) or day 4 (dox d4) of EB differentiation and evaluated by hematopoietic colony assays to measure primitive erythroid (EryP [C]) and myeloid (macrophages, MacP [D]; megakaryocytes, MegaP [E]) potential. Colony assays were each performed at least 3 separate times, with samples counted in triplicate, and values are reported as normalized to untreated controls. Statistical significance compared with no dox controls, with *P < .05 and **P < .01, respectively. CTL, control; dox, doxycycline.

Figure 2

Small-molecule–based inhibition of BMP signaling results in increased primitive myeloid potential. (A) Murine EBs were treated for 24 hours with 0.25 μM LDN, harvested on the indicated days of EB differentiation, and subjected to western blotting analysis of pathway-restricted R-SMAD phosphorylation. Phospho-SMAD levels were compared with total expression of SMAD1, and β-actin was used as a loading control. Embryoid bodies treated with LDN on day 2 or 4 of EB differentiation were cultured to day 6, harvested, and seeded on cytokine-supplemented methylcellulose medium to gauge primitive hematopoietic potential, including (B) erythroid progenitors and early myeloid lineages, (C) macrophage progenitors, and (D) megakaryocyte progenitors. Results of hematopoietic progenitor assays are reported as mean ± SEM from at least 3 experimental repeats, with values normalized to relevant controls. **P < .01. EryP, erythroid progenitor; MacP, macrophage progenitor; MegaP, megakaryocyte progenitor.

Figure 3

Control of myeloid potential by BMP signaling is SMAD-independent. (A) Titration of LDN compound and western blotting analysis of SMAD1/5/9 phosphorylation compared with total SMAD1 expression shows dose dependence of BMP signaling in day 4/5 EBs, with ∼50% inhibition achieved under conditions of 0.01 μM LDN (B, henceforth noted as LDN50). Analysis of myeloid potential by methylcellulose colony assays to examine (C) macrophage, (D) megakaryocyte, and (E) mixed lineage progenitors shows persistence of myeloid phenotype after LDN treatment regardless of whether baseline is untreated control (black bar compared with thin-striped bar) or dox-induced Smad5/Smad1 knockdown (gray bar compared with wide-striped bar). (F) Effects of BMP signaling inhibition on myeloid potential were examined on progenitors by sorting Flk1⁺ cells from day 3 EBs, and reculturing them 24 hours later in the presence of LDN50. Shown is a representative sort. (G) Macrophage potential was measured; gray bars correspond to FLK1⁻ cells, striped bars to a 50/50 mix of FLK1⁻ and FLK1⁺, and black bars to cultures containing FLK1⁺ only. Colony assays were each performed at least 3 separate times, with samples counted in triplicate. *P < .05; **P < .01.

Figure 4

SMAD-independent BMP control of myeloid potential is dependent on NOTCH, associated with increased expression of Δ-like ligands. (A) Quantitative RT-PCR analysis of Δ-like genes Dll1 and Dll3 in day 5 EBs after 24-hour treatment with dox to deplete Smad5 and Smad1 (black bar); LDN only to inhibit BMP signaling (gray bar); or dox + LDN demonstrating that enhanced expression is SMAD-independent (striped bar). (B) qPCR analysis of NOTCH effector genes 24 hours after BMP inhibition with LDN. (C) Western blotting analysis of changes in NOTCH activation via cleaved receptor intracellular domain levels (NICD), 24 hours after BMP inhibition by LDN50 in day 4 EBs. (D) Western blotting analysis of NICD 24 hours after 5 μM DAPT treatment of day 4 EBs. Representative blots are shown with densitometric analysis of relative levels compared with untreated controls. Methylcellulose colony assay analyses of (E) macrophage and (F) megakaryocyte potential under conditions of LDN50 treatment alone (compare small striped gray bars to black control bars) or in combination with 5 μM DAPT (compare gray bars to large striped bars), with each experiment performed at least 3 separate times. (G) qPCR analysis of transcript levels for myeloid regulatory transcription factor C/EBPα on day 8 of EB differentiation after day 4 treatment with LDN alone (compare black control bar to gray bar), or cotreatment with LDN + 5 μM DAPT (compare gray bar to striped bar). Values in qPCR assays are reported as mean ± SEM from at least 3 experimental repeats, normalized to untreated controls, and compared with Gapdh as a reference gene. In all experiments, *P < .05 and **P < .01.

Figure 5

BMP inhibition by LDN is associated with deactivation of p38MAPK. Phosphorylation states of reported downstream kinases responsive to SMAD-independent BMP signaling were probed in western blotting experiments and compared with total protein levels with β-actin as a loading control. (A) p38MAPK Thr/Tyr phosphorylation was markedly downregulated after overnight treatment with either full-dose LDN or LDN50 to achieve ∼100% or 50% BMP signaling inhibition, respectively, with co-depletion of SMAD1/5 by dox induction resulting in no significant changes in p38 phosphorylation. In contrast, (B) Thr/Tyr phosphorylation of ERK1/2, (C) Thr/Tyr phosphorylation of JNK, and (D) Ser phosphorylation of AKT displayed no significant changes compared with controls for any treatment. (E) These relationships are presented as quantified densitometry data, where **P < .01.

Figure 6

Direct inhibition of p38MAPK signaling is sufficient to control primitive myeloid potential. (A) Western blotting analysis of total and phospho-p38 levels after treatment of day 5 EBs with DMSO (vehicle control) or SCIO469, a small-molecule inhibitor specific to the p38-α isoform. (B) Densitometric analysis from 3 separate experiments, with phospho-p38 normalized to total p38 protein expression levels. Primitive myeloid potential was analyzed using methylcellulose-based progenitor colony assays, which showed that (C) MacP and (D) MegaP potential after direct p38MAPK inhibition is enhanced to levels similar to those observed after BMP inhibition. Macrophage (E) and megakaryocyte (F) potential was analyzed in serum-free EB cultures after addition at day 4 of recombinant BMP4, with or without the p38-α inhibitor SCIO469. BMP4 treatment directly attenuated myeloid potential and this was rescued by simultaneous p38 inhibition. Furthermore, BMP4 increased p38 activation and decreased NOTCH pathway signaling, measured by western blotting analysis of phospho-p38 and cleaved NOTCH intracellular domain, respectively (G). Western blotting results are quantified from 3 separate experiments in panel H. Values are normalized to DMSO-treated (A-F) or untreated (G-H) controls; asterisks indicate significance, with *P < .05 and **P < .01, respectively. DMSO, dimethylsulfoxide.

Figure 7

NOTCH and p38 function in parallel pathways to balance primitive myelopoiesis. (A) p38MAPK activation was examined in day 5 EBs by western blotting experiments under conditions of LDN50 treatment alone or in combination with 5 μM DAPT treatment, with β-actin as loading control. (B) Phospho-p38 levels were analyzed in comparison with total p38, and reported as mean ± SEM from 3 separate experiments. *P < .05 and **P < .01, respectively. Although DAPT alone decreases pp38 levels, it is not as significant as LDN, and does not impact myeloid output in the absence of LDN. (C) qPCR analysis of Dll1 and Dll3 Notch ligand genes after p38 inhibition with SCIO469. (D) Model diagram of BMP signaling control of myeloid potential proposing parallel pathways that function downstream through NOTCH and p38MAPK and converge to balance C/EBP levels. (E) The schematic summarizes ways that BMP signaling modulates primitive hematopoiesis through both SMAD-dependent and SMAD-independent mechanisms. Based on ESC differentiation studies, SMAD1 normally functions in hematovascular precursors to generally restrict myeloid and erythroid (panhematopoietic) potential. SMAD5 normally balances this effect while at the same time uniquely promoting primitive erythroid potential. BMP receptor activation separately functions to restrict C/EBPα-regulated myeloid capacity, independent of SMAD activity, through p38MAPK activation and Notch pathway inhibition.