The chromatin remodeling subunit Baf200 promotes normal hematopoiesis and inhibits leukemogenesis

Background Adenosine triphosphate (ATP)-dependent chromatin remodeling SWI/SNF-like BAF and PBAF complexes have been implicated in the regulation of stem cell function and cancers. Several subunits of BAF or PBAF, including BRG1, BAF53a, BAF45a, BAF180, and BAF250a, are known to be involved in hematopoiesis. Baf200, a subunit of PBAF complex, plays a pivotal role in heart morphogenesis and coronary artery angiogenesis. However, little is known on the importance of Baf200 in normal and malignant hematopoiesis. Methods Utilizing Tie2-Cre-, Vav-iCre-, and Mx1-Cre-mediated Baf200 gene deletion combined with fetal liver/bone marrow transplantation, we investigated the function of Baf200 in fetal and adult hematopoiesis. In addition, a mouse model of MLL-AF9-driven leukemogenesis was used to study the role of Baf200 in malignant hematopoiesis. We also explored the potential mechanism by using RNA-seq, RT-qPCR, cell cycle, and apoptosis assays. Results Tie2-Cre-mediated loss of Baf200 causes perinatal death due to defective erythropoiesis and impaired hematopoietic stem cell expansion in the fetal liver. Vav-iCre-mediated loss of Baf200 causes only mild anemia and enhanced extramedullary hematopoiesis. Fetal liver hematopoietic stem cells from Tie2-Cre+, Baf200f/f or Vav-iCre+, Baf200f/f embryos and bone marrow hematopoietic stem cells from Vav-iCre+, Baf200f/f mice exhibited impaired long-term reconstitution potential in vivo. A cell-autonomous requirement of Baf200 for hematopoietic stem cell function was confirmed utilizing the interferon-inducible Mx1-Cre mouse strain. Transcriptomes analysis revealed that expression of several erythropoiesis- and hematopoiesis-associated genes were regulated by Baf200. In addition, loss of Baf200 in a mouse model of MLL-AF9-driven leukemogenesis accelerates the tumor burden and shortens the host survival. Conclusion Our current studies uncover critical roles of Baf200 in both normal and malignant hematopoiesis and provide a potential therapeutic target for suppressing the progression of leukemia without interfering with normal hematopoiesis. Electronic supplementary material The online version of this article (10.1186/s13045-018-0567-7) contains supplementary material, which is available to authorized users.


Background
Hematopoiesis is a continuous process with a rare population of hematopoietic stem cells (HSCs) giving rise to all blood cell types. It is finely orchestrated by both cellintrinsic factors and microenvironmental clues [1]. In mammals, hematopoiesis occurs sequentially during the development in the yolk sac (YS), aorta-gonadmesonephros (AGM) region, placenta, fetal liver (FL), and bone marrow (BM). After birth, hematopoiesis continues with balanced proliferation, differentiation, quiescence, and apoptosis of HSCs. Various transcription factors, signaling pathways, and epigenetic regulators are involved in the regulation of these processes, and abnormalities occurring in these processes cause hematopoietic disorders including anemia or malignant transformation [2][3][4][5][6][7][8][9][10][11]. Therefore, defining factors involved in the regulation of hematopoiesis is an important issue in HSC biology.
Adenosine triphosphate (ATP)-dependent chromatin remodeling complexes include five different classes, SWI/ SNF, ISWI, NuRD/Mi2/CHD, INO80, and SWR1, and are evolutionarily conserved. The SWI/SNF complexes of mammals include the BRG1-associated factor (BAF) complex and polybromo BRG1-associated factor (PBAF) complex. These two complexes consist of several common subunits and specific subunits. The common subunits include BRG1, BAF170, BAF155, BAF60a/b/c, BAF57, BAF53a/b, BAF47, BAF45a/b/c/d, and actin. BRM and BAF250a/b subunits are found only in the BAF complex; BAF200, BAF180, and BRD7 subunits are limited to the PBAF complex [12]. BRG1, BAF53a, BAF45a, BAF180, and BAF250a are known to be involved in hematopoiesis [13][14][15][16][17][18][19][20][21][22], but the hematopoietic activity of other BAF or PBAF subunits has not been determined. BAF200, also known as ARID2, encoded by the BAF200 gene, is a unique subunit of the PBAF chromatin remodeling complex, and inactivating mutations have been reported in a variety of human cancers [23][24][25]. Baf200-null mutant mice displayed embryonic lethality with multiple cardiac defects [26]. A recent study revealed a correlation between BAF200 defect and myelodysplasia [27]. However, the precise role of Baf200 in hematopoiesis has not been examined so far. Here, we investigated the function of Baf200 in hematopoiesis through conditional deletion approach using the Tie2-Cre, Vav-iCre, and Mx1-Cre mice. Tie2-Cre-mediated loss of Baf200 causes embryonic lethality with defective fetal erythropoiesis, a reduced HSC pool in the FL, and impaired long-term reconstitution capacity of FL HSCs. Vav-iCre-mediated loss of Baf200 causes mild anemia and increased extramedullary hematopoiesis in adult mice. The long-term reconstitution potentials of the FL or BM HSCs from Vav-iCre + , Baf200 f/f mice are severely impaired. The loss of Baf200 alters the transcription of a cohort of genes involved in the maintenance of HSC homeostasis. In addition, Baf200 deficiency accelerates the progression of MLL-AF9-induced leukemia. Taken together, the results demonstrate the involvement of Baf200 in both normal and malignant hematopoiesis and provide additional knowledge of the cellular and genetic activity of the chromatin remodeling complex in HSC function.

Mice
The Baf200 LacZ/+ mice line was described previously [26]. Baf200 LacZ/+ mice were crossed with Actin-Flpe transgenic mice to generate Baf200 f/+ mice. Then, Baf200 f/+ mice were further crossed with heterozygous Tie2-Cre, Vav-iCre, or Mx1-Cre transgenic mice to generate Tie2-Cre + , Baf200 f/f , Vav-iCre + , Baf200 f/f , or Mx1-Cre + , Baf200 f/f mice. All mice were bred under specific pathogen-free conditions. The protocols were approved by the Institutional Animal Care and Use Committee (IACUC) in Institut Pasteur of Shanghai. Genotyping and gene deletion efficiency were performed by polymerase chain reaction (PCR) using primers specific for wild-type (WT) Baf200 alleles, floxed exon4 or deleted exon4. Gene deletion efficiency was also determined by reverse transcription-quantitative PCR (RT-qPCR) using primers in exon3 and exon4 (see Additional file 1: Table S2 and  Table S3 for the primers).

Flow cytometry
FL, BM, spleen, and thymus cells were isolated and passed through a 40-μm nylon cell strainer (BD Biosciences) and stained for 20 min on ice in PBS supplemented with 2% FBS. Dead cells were discarded from analysis by 4,6-diamino-2-phenylindole (DAPI) (Molecular Probes). All the antibodies used in the experiments are summarized in Additional file 1: Table S4. Flow cytometric analysis was performed on LSRII or Fortessa (BD Biosciences), and flow sorting was performed on FACSAriaII (BD Biosciences). Data were analyzed by FlowJo software (Tree Star, Ashland, OR).

FL cell counting
Embryos were collected from female mice at days 12.5 to 17.5 of pregnancy, and the FLs dissected from each embryo were removed into 1 mL PBS supplemented with 2% FBS. To obtain single cells, the FLs were pipetted by 1 mL pipette gently and passed through a 40-μm nylon cell strainer (BD Biosciences). Then, the cell number was counted by hemocytometer.

Cell cycle and apoptosis analysis
For FL cell cycle analysis, pregnant female mice were injected intraperitoneally with BrdU for 1 h, and for BM LSK cell cycle analysis, WT or Vav-iCre + , Baf200 f/f mice were injected intraperitoneally with BrdU for 2 h. Then, FL or BM cells were isolated and stained with antibodies, followed by fixation, permeabilization, and stained with anti-BrdU antibody and Hoechst 33342 (Molecular Probes) according to the manufacturer's instruction (BD Biosciences). For Ki67 staining, FL or BM cells were stained with surface markers to define subsets, followed by fixation, permeabilization, and stained with anti-Ki67 antibody and Hoechst 33342.
To detect apoptotic cells, FL or BM cells were stained with surface markers to define subsets, followed by staining with DAPI and anti-annexin V antibody in accordance with the manufacturer's instruction (BD Biosciences).

Histology
The FL, spleen, liver, and bone (femur and tibia) tissues were fixed in 4% paraformaldehyde overnight at room temperature. To decalcify, the bone tissues were placed into 10% EDTA solution and the solution was changed every other day for about 2 weeks. Then, the tissues were embedded in paraffin, sectioned at 6 μm, and stained with hematoxylin-eosin.
Retroviral production, transformation, and leukemia mouse model Retroviral supernatants were produced using HEK293T cells as described previously [28]. For leukemia cell production, Lin − BM cells were isolated, pre-stimulated overnight, and infected with retrovirus. Infected cells were injected into lethally (9.5 Gy, X-ray) irradiated recipients. Leukemia cells were harvested from the sick mice and transplanted into sub-lethally (5 Gy, X-ray) irradiated recipients in secondary transplantation.

RNA-seq and RT-qPCR
Total RNA was isolated from flow-sorted LSK cells or S3 cells using the RNeasy Mini kit (QIAGEN, Valencia, CA), and the RNA-seq analysis was performed at BGI (The Beijing Genomics Institute) via Illumina HiSeq™ 2000. Briefly, the mRNA was enriched by using the oligo (dT) magnetic beads, then fragmented into fragments. Then, the double-stranded complementary DNA (cDNA) was synthesized and purified by magnetic beads. End reparation and 3′-end single nucleotide A (adenine) addition was then preformed. Finally, sequencing adaptors were ligated to the fragments. The fragments were enriched by PCR amplification, and the quality of the library was assessed by using Agilent 2100 Bioanalyzer and ABI StepOnePlus Real-Time PCR system. Then, the library was ready for sequencing. Raw data were deposited in the Sequence Read Archive database (accession number SRP117796). Functional Gene Ontology analysis of differential genes was determined by using DAVID Bioinformatics Resources 6.8 [29,30].
For RT-qPCR analysis of gene expression in LSK cells, cDNA was produced by EZ-press Cell to cDNA Kit (EZbioscience). For RT-qPCR analysis of gene expression in other cells, total RNA was isolated by TRIzol (invitrogen) and converted into cDNA by using FastKing RT Kit (Tiangen). Primers for PCR were designed using Primer3 software or selected from PrimerBank database (Additional file 1: Table S3). All the reactions were performed at ABI Prism 7900HT (Applied Biosystems) using SYBR® Premix Ex Taq ™ II (Takara).

Statistical analysis
The significance of differences was determined with twotailed paired Student's t tests by using Prism software (GraphPad Software). Survival data from leukemia mouse models were analyzed by using a log-rank nonparametric test, and the data were expressed as Kaplan-Meier survival curves. Statistical significance is indicated by *p < 0.05, **p < 0.01, or ***p < 0.001.

Results
Tie2-Cre + , Baf200 f/f mice display embryonic lethality with defective fetal erythropoiesis The expression of Baf200 was examined in representative hematopoietic lineages. Baf200 is widely expressed in all hematopoietic lineages and particularly in FL LSK (Lin − Sca1 + c-Kit + ), FL hematopoietic progenitor cell (HPC) and granulocyte-macrophage progenitor (GMP) (Lin − Sca1 − c-Kit + CD34 + CD16/32 high ) lineages (Additional file 1: Figure S1). To investigate its role in hematopoiesis in vivo, Baf200 was conditionally deleted by crossing Baf200 f/f mice with Tie2-Cre transgenic mice which express Cre-recombinase in endothelial cells and hematopoietic lineages [31]. Only four Tie2-Cre + , Baf200 f/f mice were obtained among 239 offspring from intercross between male Tie2-Cre + , Baf200 f/+ mice and female Baf200 f/f mice at weaning age, which was considerably lower than the expected Mendelian ratio (Additional file 1: Figure S2a). The four Tie2-Cre + , Baf200 f/f mice exhibited low recombination efficiency of the LoxP sites in hematopoietic cells (data not shown). Examination of the genotypes in the litters from timed mating intercross found that Tie2-Cre + , Baf200 f/f embryos were alive and were recovered at the expected Mendelian ratio at embryonic day (E) 17.5, but that the majority died by E18.5 ( Fig. 1a and Additional file 1: Figure S2a).
Genomic PCR or RT-qPCR confirmed the deletion of Baf200 in the FL cells of E14.5 Tie2-Cre + , Baf200 f/f embryos (Additional file 1: Figure S2c, d), suggesting that Baf200 was efficiently deleted in the hematopoietic lineages. Tie2-Cre + , Baf200 f/f embryos showed severe anemia and profound growth retardation as early as E15.5 compared with their WT Tie2-Cre − , Baf200 f/f or Tie2-Cre − , Baf200 f/+ littermates ( Fig. 1a and Additional file 1: Figure S2b). The absolute cell number of FL from Tie2-Cre + , Baf200 f/f embryos was also reduced since E13.5 (Fig. 1b). The apoptosis of FL cells in Tie2-Cre + , Baf200 f/f and WT embryos at E14.5 and E16.5 was measured using annexin-V and DAPI staining followed by flow cytometric analysis. Although the apoptosis of FL cells were comparable between Tie2-Cre + , Baf200 f/f and WT embryos at E14.5, the percentage of annexin-V + , DAPI − and annexin-V + , DAPI + FL cells in Tie2-Cre + , Baf200 f/f embryos at E16.5 is significantly higher than that in WT embryos (Additional file 1: Figure S2e). These results suggested that increased apoptosis might account, at least in part, for the reduced FL cell number in Tie2-Cre + , Baf200 f/f embryos.
Evaluation of erythropoiesis in the FL from WT and Tie2-Cre + , Baf200 f/f embryos by staining with Ter119 and CD71 surface markers revealed that the percentages of S0, S1, and S2 cells were increased and the percentages of S4 and S5 cells were decreased in E15.5 Tie2-Cre + , Baf200 f/f FL (Fig. 1c, d). Although the total number of FL cells in the Tie2-Cre + , Baf200 f/f embryos was reduced, the absolute numbers of S0, S1, and S2 cells remained unchanged, whereas the absolute numbers of S3, S4, and S5 cells were reduced (Fig. 1e). Notably, similar phenotype was also observed at E13.5 and E14.5 (Additional file 1: Figure S2f-g) and became much more apparent at E16.5 ( Fig. 1f-h). Though the percentage of FL MEP (Lin − Sca-1 − c-Kit + CD16/32 − CD71 − ) was increased, the absolute cell number remained unchanged in E14.5 Tie2-Cre + , Baf200 f/f FL (Additional file 1: Figure S2 h). We found that the cell cycle status and apoptosis of FL erythrocytes from Tie2-Cre + , Baf200 f/f embryos were comparable to those from WT embryos (Additional file 1: Figure S2i-k), suggesting that the defective erythropoiesis in Tie2-Cre + , Baf200 f/f FL was not caused by reduced cell proliferation or increased apoptosis. These results indicate that Baf200 plays a pivotal role in the erythroid terminal differentiation in the FL.
To determine the potential mechanism by which Baf200 regulates fetal erythropoiesis, we performed RNA-seq analysis of the transcriptomes of S3 cells from E14.5 WT or Tie2-Cre + , Baf200 f/f embryos. A total 161 upregulated genes and 60 downregulated genes were observed in Tie2-Cre + , Baf200 f/f S3 cells (Fig. 1i). Functional Gene Ontology (GO) analysis of differentially expressed genes indicated significant enrichment in translation, proteolysis involved in cellular protein catabolic process, glycosphingolipid metabolic process, erythrocyte homeostasis, intracellular signal transduction, response to oxidative stress, and oxidation-reduction process terms (Additional file 1: Figure S3). During the transition from S2 to S3, the expression level of Myb and Kit is downregulated [32], and the downregulation of Myb is important for erythroid terminal differentiation [33,34]. In addition, Hmox1 overexpression in erythroid cells decreases hemoglobin synthesis [35]. Moreover, the expression of Pla2g16 is induced during erythroid maturation, and Pla2g16 deficiency leads to impaired proerythroblasts (ProE) terminal differentiation [36]. We found that the expression of Myb, Hmox1, and Kit was increased, while the expression Pla2g16 was decreased in Tie2-Cre + , Baf200 f/f S3 cells (Fig. 1i-k).These may partially explain the impaired FL erythropoiesis in Tie2-Cre + , Baf200 f/f embryos. Baf200 f/f S3 cells. j, k Upregulation genes (j) and downregulation genes (k) in Tie2-Cre + , Baf200 f/f S3 cells were confirmed by RT-qPCR (n = 3 per genotype). Data are shown as means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001 The expansion of FL HSCs in Tie2-Cre + , Baf200 f/f embryos is severely impaired We then examined the hematopoietic stem and progenitor cells in WT or Tie2-Cre + , Baf200 f/f FL (Fig. 2a). The absolute number of LSK (Lin − Sca1 + c-Kit + ) cells and longterm (LT)-HSCs (CD150 + CD48 − LSK) from Tie2-Cre + , Baf200 f/f FL was similar to those from WT FL at E12.5 (Fig. 2b, c). However, the absolute number of LSK cells and LT-HSC was significantly decreased in E13.5 and E15.5 Tie2-Cre + , Baf200 f/f FL (Fig. 2b, c), indicating that the expansion, but not the initial seeding, of HSCs in the FL was impaired in Tie2-Cre + , Baf200 f/f embryos. We next detected the cell proliferation and apoptosis of LSK and CD150 + LSK cells from E14.5 Tie2-Cre + , Baf200 f/f embryos.
There was a significant increase in apoptosis of the LSK and CD150 + LSK cells of Tie2-Cre + , Baf200 f/f embryos, while the cell proliferation remains unchanged (Fig. 2d, e). Hence, Tie2-Cre-mediated loss of Baf200 causes aberrant apoptosis of FL LSK and LT-HSC.

Baf200 is required for the FL HSC maintenance
A competitive repopulation assay was performed by cotransplanting FL cells from CD45.2 + Tie2-Cre + , Baf200 f/f embryos or CD45.2 + WT controls with CD45.1 + BM cells at a ratio of 2:1 into lethally irradiated CD45.1 + CD45.2 + recipient mice (Fig. 3a). Reconstitution by donor cells was evaluated in the peripheral blood, spleen, thymus, and BM of the recipient mice 16 weeks post-transplantation. As expected, WT donor cells competed effectively with the competitor cells. In contrast, Tie2-Cre + , Baf200 f/f FL cells failed to contribute the long-term reconstitution of the myeloid, T, and B cell lineages or the LSK cells in the chimeras ( Fig. 3b and Additional file 1: Figure S5a, b). The homing capacity of Tie2-Cre + , Baf200 f/f FL cells remains unchanged (Additional file 1: Figure S5c). We next generated an inducible Baf200-deletion mouse model by crossing Mx1-Cre with Baf200-flox mice. FL cells from Mx1-Cre + , Baf200 f/f (before deletion) or control embryos were transplanted together with competitor BM cells into lethally irradiated recipients. Six to 8 weeks later, the recipient mice received poly(I:C), and the peripheral blood was analyzed for 3 months (Fig. 3c) Baf200 led to gradual but significant decrease of lymphoid and myeloid lineages in the recipients (Fig. 3d). These results suggest that a cell-intrinsic role of Baf200 is required for FL HSC function.

Baf200 contributes to adult erythropoiesis
To further explore the in vivo function of Baf200 in hematopoiesis, we generated Vav-iCre + , Baf200 f/f mice by utilizing Vav-iCre mice strain which express Cre-recombinase in hematopoietic lineages [42]. The Vav-iCre + , Baf200 f/f mice were recovered at the expected Mendelian ratio and displayed an appearance indistinguishable from the Vav-iCre − , Baf200 f/f or Vav-iCre − , Baf200 f/+ WT controls. Efficient deletion of Baf200 in the BM, spleen, and thymus cells from Vav-iCre + , Baf200 f/f mice was confirmed by genomic PCR (Additional file 1: Figure S6a). Peripheral blood counts revealed that Vav-iCre + , Baf200 f/f mice exhibited decreased RBC number, hemoglobin concentration, hematocrit, and mean corpuscular hemoglobin along with increased RDW, suggesting a mild anemia may occur in Vav-iCre + , Baf200 f/f mice ( Fig. 4a and Additional file 1: Figure S6b). Consistently, Vav-iCre + , Baf200 f/f mice exhibited slightly paler bone and splenomegaly (Fig. 4b). However, the total number of nucleated cells in the BM and spleen from Vav-iCre + , Baf200 f/f mice were comparable with those from WT mice (Fig. 4c). Vav-iCre + , Baf200 f/f mice (Fig. 4d, e). The activation of erythropoiesis in the spleen might have been an attempt to compensate for insufficient erythropoiesis in the BM. Surprisingly, an increase of immature erythrocytes defined as Ter119 + CD71 + and a decrease of mature erythrocytes defined as Ter119 + CD71 − were detected in the peripheral blood of Vav-iCre + , Baf200 f/f mice (Fig. 4f). Taken together, the results indicate that Baf200 contributes to adult erythropoiesis.

Baf200 is involved in maintaining the adult hematopoiesis in steady state
Flow cytometric analysis of the HSC compartments demonstrated decreases in the percentages and absolute numbers of LT-HSCs, ST-HSCs, MPP cells, common myeloid progenitor (CMP) cells, and MEP cells in the BM of Vav-iCre + , Baf200 f/f mice (Fig. 5a, b). Enhanced extramedullary hematopoiesis in the spleen usually accompanies a decrease in BM hematopoiesis. Indeed, the percentages and numbers of CMP, GMP, and MEP cells in the spleen of Vav-iCre + , Baf200 f/f mice were increased (Fig. 5c, d), indicating that Baf200 deficiency promotes extramedullary hematopoiesis. Flow cytometric analysis found no significant differences in the cell cycle status or apoptosis of LSK cells in Vav-iCre + , Baf200 f/f mice and WT controls (Additional file 1: Figure S7a-g). The data reveal that Baf200 is not required for cell cycle process or apoptosis of the adult LSK compartment in steady-state hematopoiesis.
To determine the potential molecular pathways by which Baf200 regulates BM HSCs function, we conducted RNA-seq assay on purified BM LSK cells from Vav-iCre + , Baf200 f/f and WT mice. A total of 264 genes were upregulated and 198 genes were downregulated in Baf200-deficient LSK cells (Fig. 5e). Although the top GO terms are related to immune response, cell adhesion, and metabolic process, the erythrocyte development and hematopoiesis terms were also observed (Additional file 1: Figure S8). RT-qPCR experiment confirmed the altered expression of several hematopoiesis-associated genes (Fig. 5f ), suggesting that Baf200 has an impact on adult HSC function, at least in part through regulation of the hematopoiesis-related genes.

Cell-intrinsic Baf200 is required for HSC maintenance
Competitive hematopoietic reconstitution assays were performed to further characterize the requirement of Baf200 for adult HSC maintenance by transplanting BM cells from Vav-iCre + , Baf200 f/f or WT controls together with competitor BM cells into lethally irradiated recipients (Fig. 6a). Sixteen weeks post-transplantation, Baf200deficient HSCs failed to repopulate lymphoid and myeloid lineages in the peripheral blood, BM, spleen, and thymus of the recipients (Fig. 6b and Additional file 1: Figure S9a). Moreover, the mutant donor-derived LSK cells were barely detectable in chimeras despite the comparable homing capacity of WT and Baf200-deficient BM cells (Fig. 6b and Additional file 1: Figure S9b). The results indicate that Baf200 is required for the maintenance of adult BM HSCs.
To further elucidate whether Baf200 regulates hematopoiesis in a cell-intrinsic manner, we firstly utilized inducible Baf200-deletion mice (Mx1-Cre + , Baf200 f/f ) and performed competitive transplantation experiments as above (Fig. 6c). Before poly(I:C) injection, Mx1-Cre + , Baf200 f/f and WT BM cells had similar reconstitution capacity, while loss of Baf200 also led to gradual but significant decrease of lymphoid and myeloid lineages in the recipients (Fig. 6d). In addition, we transplanted CD45.1 + BM cells into lethally irradiated Vav-iCre + , Baf200 f/f or WT mice. Two months later, BM cells from the recipients were harvested and cotransplanted into lethally irradiated recipients together with competitor cells. Sixteen weeks later, flow cytometry assays were performed to detect the lineage contribution of transplanted cells (Additional file 1: Figure S9c). The results revealed no difference in contribution ratios of any lineages (Additional file 1: Figure S9d). Therefore, these data demonstrate that Baf200 is a cell autonomously required for the adult HSC maintenance.
Considering that Tie2-Cre-mediated deletion of Baf200 led to embryonic lethality, we then detected the effects of Baf200 deletion on fetal hematopoiesis from Vav-iCre + , Baf200 f/f mice. Although there was no difference in the absolute number of FL cells in E14.5 Vav-iCre + , Baf200 f/f and WT embryos (Additional file 1: Figure S10a), detailed flow cytomeric analysis showed that there was a blockade of S3 to S4 transition during erythroid differentiation in the Vav-iCre + , Baf200 f/f embryos (Additional file 1: Figure S10b). The frequency and total number of LSK cells and LT-HSCs in the FL from Vav-iCre + , Baf200 f/f embryos remained unchanged compared with those in the littermate controls (Additional file 1: Figure S10c). However, competitive transplantation assay demonstrated that FL cells from Vav-iCre + , Baf200 f/f embryos also showed impaired long-term reconstitution capacity (Additional file 1: Figure S10d-f).

Loss of Baf200 accelerates leukemogenesis in MLL-AF9-induced AML
The aforementioned results showed that Baf200 plays important roles in normal hematopoiesis; we then asked whether Baf200 also has a role in leukemogenesis. To test this, we utilized a previously described acute myeloid leukemia (AML) mouse model induced by MLL-AF9 fusion protein (Fig. 7a) [43]. Leukemia cells were harvested from two cohorts of primary recipients, and 50,000, 200,000, or 500,000 cells were transplanted into sublethally irradiated recipients in the secondary transplantation. Notably, Baf200 deficiency significantly accelerated the leukemia progression in the secondary transplantation. All mice in Vav-iCre + , Baf200 f/f group died at around 25 days post-transplantation, whereas the average survival lifespan of recipients in WT group behaved in a dose-dependent manner (Fig. 7b). We then adopted 200,000 donor cells as a standard in the followup experiments. Vav-iCre + , Baf200 f/f group mice showed a b c d e f much paler bones and splenomegaly at 20 days posttransplantation (Fig. 7c). Histology evaluation revealed more leukemia blasts in the peripheral blood, BM, spleen, and liver in Vav-iCre + , Baf200 f/f group recipients than WT group (Fig. 7d), and flow cytometry confirmed much more GFP + leukemia cells in the peripheral blood, BM, and spleen from Vav-iCre + , Baf200 f/f group recipients (Fig. 7e). The Vav-iCre + , Baf200 f/f group mice consistently exhibited increased myeloid cells and decreased lymphoid cells in the peripheral blood compared with WT mice (Fig. 7f,  g). In addition, Vav-iCre + , Baf200 f/f group mice showed significantly accelerated accumulation of leukemia cells in the peripheral blood (Fig. 7h). We then detected the leukemia stem cell (LSC) frequency in WT and Vav-iCre + , Baf200 f/f recipients and found that the percentage of LSC-enriched population (GFP + Lin − c-Kit + ) were much higher in the BM and spleen of Vav-iCre + , Baf200 f/f group (Fig. 8a). Another set of markers for enriched LSCs confirmed an increase in the percentage of GFP + Gr1 − c-Kit + cells in Vav-iCre + , Baf200 f/f group (Fig. 8b). To explore the underlying mechanisms by which Baf200 deficiency accelerates the leukemia progression, we sorted GFP + BM leukemia cells from Vav-iCre + , Baf200 f/f and WT group and detected the expression of hematopoiesis-related genes by RT-qPCR assay. Interestingly, several genes associated with hematopoietic malignancies, including Gata2, Cebpa, Pdk1, Myb, Myc, and Meis1 [44][45][46][47][48], were upregulated in Vav-iCre + , Baf200 f/f AML cells (Fig. 8c). The expression of Cebpe, a key transcription factor in granulopoiesis [49], was decreased in Vav-iCre + , Baf200 f/f AML cells (Fig. 8d). Furthermore, loss of Baf200 promotes the progression of leukemogenesis partially by inhibiting the expression of p57 and p16 (Fig. 8d). Taken together, the results indicate that Baf200 acts as a tumor suppressor in MLL-AF9-induced AML.

Discussion
Although ATP-dependent SWI/SNF complex have been demonstrated to play pivotal roles in normal or malignant hematopoiesis [13][14][15][16][17][18][19][20][21], little is known about the function of PBAF-specific subunit Baf200 in these processes. Here, we utilized several hematopoietic lineagespecific Cre mouse lines to investigate the role of Baf200 in fetal and adult hematopoiesis, as well as in leukemogenesis. Our current study clearly showed the cell-intrinsic requirement of Baf200 in hematopoietic cells, as FL HSCs from Tie2-Cre + , Baf200 f/f or Vav-iCre + , Baf200 f/f mice exhibited impaired capacity of long-term hematopoietic reconstitution. Moreover, we utilized adoptive transfer strategy with inducible Mx1-Cre system and verified the cell-intrinsic requirement for Baf200 in the FL or adult BM HSC maintenance.
We noticed that Tie2-Cre + , Baf200 f/f mice displayed much more severe phenotypes than Vav-iCre + , Baf200 f/f mice. Tie2-Cre + , Baf200 f/f embryos died at embryonic stage with defective fetal erythropoiesis, whereas Vav-iCre + , Baf200 f/f mice were born at expected Mendelian ratio with an appearance indistinguishable from their littermate controls. Tie2-Cre is originally expressed in hemogenic endothelium as early as E6.5, which leads to efficient deletion in endothelial cells and their progeny (HSCs and all definitive blood cells) [31,50]. In contrast, Vav-iCre begins to be expressed at E9.5 and the expression was greatly enhanced at E12.5; thus, it causes high deletion efficiency in hematopoietic lineages but low deletion efficiency in endothelial cells [42,51]. So we speculated one possibility is that Baf200 in endothelial cells may also play a role in fetal hematopoiesis. Endothelial cell-specific Cre strain, such as VE-Cadherin-CreERT2, would be useful for dissecting the precise role of Baf200 in endothelial cells in the future. In mouse, there are two waves of fetal hematopoiesis, primitive, and definitive. Primitive hematopoiesis produces erythroid and myeloid cells in the YS blood island at around E7.5. Definitive hematopoietic cells (transient erythroid and myeloid progenitors) arise from hemogenic endothelium in the YS at around E8.25, and these cells can migrate into the FL to give rise to erythroid and myeloid progeny. Generation of de novo HSCs takes place in the AGM region at around E10.5, soon the HSCs migrate into the FL, expand, and differentiate [1,52]. Previous study revealed that Tie2-Cre mediates recombination in almost all primitive and definitive hematopoietic cells [50], while only definitive hematopoietic cells that have already homed into FL can express recombinase in Vav-iCre strain [42]. Therefore, another possibility of the distinct phenotypes observed in Tie2-Cre and Vav-iCre strain is the different timing of Cre activity. Several SWI/SNF complex subunits have been reported to play important roles in hematopoiesis. Brg1, a catalytic subunit of the BAF and PBAF complexes, is involved in both primitive and definitive erythropoiesis [13,14]. Baf250a, a subunit of the BAF complex, controls the pool size of FL HSCs [20]. Baf53a, a subunit of both BAF and the PBAF complex, is required for HSC a b c d Data are shown as means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001 maintenance [18]. Baf180, another unique subunit of PBAF complex, is required for long-term reconstitution potential of HSCs [22]. Our current study provided additional evidence that Baf200, a PBAF specific subunit, plays a key role in hematopoiesis. Although Baf200 and Baf180 are both the PBAF-specific subunits, Baf200 seems to have a more important role than Baf180 in hematopoiesis. Baf200-deficient HSCs were poorly competitive in long-term hematopoietic reconstitution following primary transplantation, whereas Baf180-deficient HSCs had a compromised long-term reconstitution potential only in secondary transplantation. BAF200 interacts with BAF180 and some other identical subunits within PBAF complex to exert their function [12]. However, a previous study showed that BAF200, but not BAF180, is essential for the stability of PBAF complex [53]. This may be an explanation for the appeared more vital role of Baf200 in hematopoiesis. However, whether Baf200 regulates hematopoiesis through a PBAF-independent mechanism needs to be determined in future studies. Epigenetic modifiers may have adverse effects on normal and malignant hematopoiesis. For example, Asxl1, a polycomb group protein, is essential for preserving hematopoietic reconstitution capacity, whereas haploinsufficiency of Asxl1 can cause MDS-like disease [54,55]. Utx, a di-and trimethyl H3K27 demethylase, is required for normal hematopoiesis [56]; meanwhile, Utx serves as a tumor suppressor in NOTCH1-induced T cell acute lymphoblastic leukemia (T-ALL) [57]. Moreover, differentially expressed genes regulated by Utx were found in normal Utx-deficient BM cells or in Utx-deficient leukemia cells [56,57], indicating Utx may regulate the target genes in a dual context-dependent manner. Our current study found that Baf200 deficiency impairs normal hematopoiesis but accelerates MLL-AF9-driven leukemogenesis, suggesting that Baf200 may play contrast roles in normal and malignant hematopoiesis. We also revealed diverse genes regulated by Baf200 in normal LSK cells and AML cells. However, whether Baf200 regulate its target genes or pathways in a dual context-dependent manner in normal or leukemia cells still need further investigation.
Several potential Baf200 targets were identified by RNA-seq analysis and RT-qPCR in our study. RNA-seq results revealed diverse patterns of gene differentially regulated by Baf200 in different cell lineages, including FL LSK, FL erythrocytes, adult BM LSK, and leukemia cells. These results indicate that Baf200 may regulate different subsets of target genes in a cell contextdependent manner. It may recruit different transcription factors or histone modifiers or be recruited by them to the specific target genes in different cell types. Whether Baf200 controls these genes through a direct or indirect manner still needs further investigations. Future studies will aim to explore the underlying mechanisms of Baf200 in normal and malignant hematopoiesis by using ChIP-based analysis.

Conclusions
In summary, this study reveals that loss of Baf200 impairs the normal fetal and adult hematopoiesis while accelerates the progression of MLL-AF9-induced leukemia, which provide a potential therapeutic target for leukemia treatment.

Additional file
Additional file 1: Table S1. The immunophenotypes of the tested subsets in Figure S1. Table S2. List of primers used in mice genotyping. Table S3. List of primers used in RT-qPCR experiments. Table S4. List of commercial available antibodies used in these studies. Figure S1. Expression pattern of the Baf200 gene in FACS-purified populations from mouse FL, BM, and spleen (n = 3). Figure S2. Baf200 is dispensable for the proliferation and apoptosis of FL erythrocytes. Figure S3. Gene Ontology analysis of Baf200-regulated genes in FL S3 cells. Figure S4. RNA-seq analysis of FL LSK cells from WT and Tie2-Cre + , Baf200 f/f embryos. Figure S5. FL HSCs from Tie2-Cre + , Baf200 f/f embryos show impaired long-term reconstitution potential, related to Fig. 3. Figure S6. Baf200 contributes to adult erythropoiesis and hematopoiesis, related to Figs. 4 and 5. Figure S7. Baf200 is dispensable for the cell cycle status or apoptosis of BM LSK compartment in steady state. Figure S8. Gene Ontology analysis of Baf200-regulated genes in BM LSK cells. Figure S9. Cell-intrinsic role of Baf200 in HSCs function, related to Fig. 6. Figure