Bcl11a Controls Flt3 Expression in Early Hematopoietic Progenitors and Is Required for pDC Development In Vivo

Bcl11a is a transcription factor known to regulate lymphoid and erythroid development. Recent bioinformatic analysis of global gene expression patterns has suggested a role for Bcl11a in the development of dendritic cell (DC) lineages. We tested this hypothesis by analyzing the development of DC and other lineages in Bcl11a −/− mice. We found that Bcl11a was required for expression of IL-7 receptor (IL-7R) and Flt3 in early hematopoietic progenitor cells. In addition, we found severely decreased numbers of plasmacytoid dendritic cells (pDCs) in Bcl11a −/− fetal livers and in the bone marrow of Bcl11a −/− fetal liver chimeras. Moreover, Bcl11a −/− cells showed severely impaired in vitro development of Flt3L-derived pDCs and classical DCs (cDCs). In contrast, we found normal in vitro development of DCs from Bcl11a −/− fetal liver cells treated with GM-CSF. These results suggest that the persistent cDC development observed in Bcl11a −/− fetal liver chimeras reflects derivation from a Bcl11a- and Flt3-independent pathway in vivo.

A bioinformatic analysis of global gene expression patterns has identified groups of transcription factors that may be involved in fate decisions along the DC lineage [16]. Among genes that increase in expression from the macrophage-DC progenitor (MDP) to the CDP, those that do not increase in expression from the MDP to the monocyte were labeled in that analysis as possible promoters of DC commitment. Transcription factors identified by these criteria include some previously associated only with pDC development, including E2-2 and Spi-B [17,18], and some previously associated only with cDCs, including Zbtb46 [19,20]. Other factors identified in this analysis include Irf8, Bcl11a, and Runx2. Recently, it has been demonstrated in the setting of competitive BM reconstitution that Irf8 promotes the development of all DC subsets [21], even though Irf8 2/2 mice in other settings do not show defects in CD4 + cDC development [12,13]. We wondered, therefore, whether a similar early role in DC development could be identified for another factor such as Bcl11a.
Bcl11a was first described as a gene located at a common proviral integration site in BXH2 myeloid leukemias, and its human ortholog was found to be a recurrent target of translocations in B cell malignancies [22,23]. This gene encodes a Krüppellike zinc finger transcription factor expressed in neural and lymphoid tissues that is essential for the development of B cells and for thymocyte maturation [24]. In the erythroid lineage, BCL11A acts in trans to silence the fetal hemoglobin locus in cooperation with the transcription factor SOX6 [25,26]. Indeed, differences in stage-specific expression between human BCL11A and mouse Bcl11a account at least in part for interspecies differences in fetal hemoglobin expression patterns [25].
Although Bcl11a has been recognized as a useful marker of pDCs [27,28], its actual role in DC development remains unreported. Thus, we sought to examine DC development in the setting of Bcl11a deficiency in vivo and in vitro. We found that Bcl11a was required for normal expression of IL-7 receptor (IL-7R) as well as Flt3 in early hematopoietic progenitors. In addition, we observed a strict requirement for Bcl11a in pDC development and found evidence for a Bcl11a-independent pathway of cDC development in vivo.

Bcl11a is Required for Development of CLPs and CDPs
During hematopoiesis, Bcl11a is expressed at similar levels in the hematopoietic stem cell (HSC), multipotent progenitor (MPP), common lymphoid progenitor (CLP), common myeloid progenitor (CMP), and megakaryocyte-erythroid progenitor (MEP) [16]. To study the function of Bcl11a in hematopoietic progenitors, we used mice targeted for deletion of the first exon of Bcl11a [24]. Since Bcl11a 2/2 mice die in utero or perinatally, we compared hematopoietic progenitor populations present in wild type (WT) and Bcl11a 2/2 fetal livers at embryonic day 14.5. First, we analyzed development of Lin 2 Sca-1 + c-Kit + (LSK), CLP, granulocyte-macrophage progenitor (GMP), MEP, and CDP populations ( Fig. 1). WT and Bcl11a 2/2 fetal livers showed comparable frequencies of GMPs and MEPs. However, Bcl11a 2/2 fetal livers showed a greater than twofold decrease in the frequency of LSK cells and more marked decreases in frequencies of IL-7R + CLPs and Flt3 + CDPs relative to WT fetal livers (Fig. 1A, B); within the LSK fraction, Bcl11a 2/2 fetal livers showed defects in both CD150 (Slamf1) + and CD150populations (Fig. 1C). One study has demonstrated that a Sca-1 lo c-Kit + Flt3 + CD150population with granulocyte and macrophage potential (SL-GMP) can be identified which excludes mast cell potential [29]; GMPs in the Bcl11a 2/2 fetal liver, however, lacked Flt3 expression (data not shown) and no SL-GMP population could be identified (Fig. 1D). Next, we analyzed hematopoietic development in chimeras produced by transferring WT or Bcl11a 2/2 fetal liver cells into lethally irradiated congenic recipient mice (Fig. 2). Four to six weeks after transfer, donor-derived Bcl11a 2/2 BM showed decreased frequencies of LSK cells, CLPs, and CDPs but comparable frequencies of GMPs and MEPs relative to donorderived WT BM ( Fig. 2A, B); within the LSK fraction, donorderived Bcl11a 2/2 BM showed a greater proportion of CD150 + cells than did donor-derived WT BM, corresponding to a decrease in the overall frequency of the more differentiated CD150population (Fig. 2C). As in Bcl11a 2/2 fetal livers, no SL-GMP population could be identified in donor-derived Bcl11a 2/2 BM (Fig. 2D). In summary, the loss of Bcl11a in hematopoietic progenitors resulted in impaired development of LSK cells as well as a selective loss of CLPs and CDPs; these effects were observed both in the fetal stage and in the adult chimera, demonstrating that this factor is required in fetal and adult hematopoiesis.
Conceivably, the absence of IL-7R + CLPs and Flt3 + CDPs in Bcl11a 2/2 fetal livers and BM could result from a requirement for Bcl11a in the development of the CLP and CDP or from a more restricted requirement for Bcl11a in the expression of IL-7R and Flt3, the surface markers that identify these populations. In either case, however, the loss of Bcl11a should result in DC defects because Flt3 ligand (Flt3L) signaling is essential for DC development in the steady state [30][31][32].

Bcl11a Regulates Expression of Il7r and Flt3
To identify Bcl11a target genes that explain its role in hematopoietic progenitors, we compared global gene expression by microarray for donor-derived WT and Bcl11a 2/2 populations isolated from chimeric BM (Fig. 3). Since we observed that IL-7R-and Flt3-expressing populations were affected by the loss of Bcl11a, we avoided the use of these surface markers in order to allow for comparison of equivalent populations across genotypes. Thus, we isolated multipotent progenitors (MPPs) as identified by the lack of CD150 expression within the LSK fraction [33][34][35]. We also isolated GMPs from the same BM, since the size of this population was unaffected by loss of Bcl11a.
We found that WT and Bcl11a 2/2 GMPs were more similar to each other in gene expression than WT and Bcl11a 2/2 MPPs were to each other (Fig. 3A). One hundred and thirty-four probe sets showed a greater than twofold change in expression between WT and Bcl11a 2/2 GMPs. In contrast, 1020 probe sets showed a greater than twofold change in expression between WT and Bcl11a 2/2 MPPs; of these, only 38 also show a greater than twofold change between WT and Bcl11a 2/2 GMPs (Fig. 3B). These data suggest that GMP population size is unaffected by loss of Bcl11a because this transcription factor regulates relatively few genes in GMPs.
Since the loss of Bcl11a impaired development of CDPs but not GMPs, we examined Bcl11a target genes which showed expression patterns that distinguish DCs from monocytes and macrophages. Thus, we compared the ratio of gene expression in CDPs relative to monocytes against the ratio of gene expression in Bcl11a 2/2 MPPs relative to WT MPPs (Fig. 3C). Of genes most highly expressed in CDPs relative to monocytes, those most affected by loss of Bcl11a included Flt3, Cnn3 (encoding calponin 3), Dntt (encoding the template-independent DNA polymerase TdT), Il7r, and Blnk (encoding B-cell linker protein, which links components of B-cell receptor signaling). We also compared changes in gene expression between WT and Bcl11a 2/2 MPPs for members of the core cDC transcriptional signature identified in a published bioinformatic analysis [16] (Fig. 3D). Within this core signature, we found only three genes-Ass1, Amica1, and Flt3-that showed a greater than twofold decrease in expression in Bcl11a 2/2 MPPs relative to WT MPPs. Taken together, the decreased expression of Flt3 and Il7r in Bcl11a 2/2 MPPs suggests that Bcl11a may be specifically required for the expression of these genes.

Bcl11a is Required for pDC but not cDC Development in vivo
Next, we examined the development of mature hematopoietic subsets in WT and Bcl11a 2/2 fetal liver chimeras (Fig. 4). In accordance with previous reports [24], we observed atrophic thymi in Bcl11a 2/2 chimeras (data not shown). In the BM, the size of the donor-derived compartment was comparable in WT and Bcl11a 2/2 chimeras; in the spleen and skin-draining lymph nodes, Bcl11a 2/2 cells were somewhat impaired in their competition against the residual host population (Fig. 4A). Within the donorderived compartment of the spleen, a profound defect in pDC development was apparent in Bcl11a 2/2 chimeras relative to WT chimeras (Fig. 4B). In contrast, donor-derived cDCs were present in Bcl11a 2/2 chimeras with no significant decrease relative to WT chimeras (Fig. 4C).
Among lymphoid subsets, donor-derived B cells, CD4 T cells, CD8 T cells, and cd T cells were decreased in frequency by at least tenfold in the spleens of Bcl11a 2/2 chimeras as compared to WT chimeras, consistent with previous reports [24], while NK cells were decreased by slightly more than threefold (Fig. 5A). Among myeloid subsets other than pDCs, donor-derived CD8 2 cDCs showed a modest threefold decrease in the spleens of Bcl11a 2/2 chimeras as compared to WT chimeras; other myeloid populations examined, including CD8 + cDCs, were not decreased in frequency (Fig. 5B). Thus, Bcl11a was strictly required for the development of pDCs but not cDCs in vivo.

Flt3-dependent, but not GM-CSF-dependent, DC Development Requires Bcl11a in vitro
We compared the development of WT and Bcl11a 2/2 cells in vitro in response to treatment with Flt3L or granulocyte macrophage colony-stimulating factor (GM-CSF) (Fig. 6). The observation that Flt3 2/2 mice retain DC development [36] suggests an alternative receptor for Flt3L or a Flt3L-independent pathway for DC development. Thus, we supplied excess Flt3L or GM-CSF to distinguish between these possibilities in the context of Bcl11a deficiency. As expected, pDCs developed from WT fetal liver cells (Fig. 6A) and from the donor-derived BM cells of WT chimeras (Fig. 6B) in response to Flt3L treatment. In contrast, pDCs developed in markedly decreased numbers from Bcl11a 2/2 fetal liver cells and completely failed to develop from the donorderived BM cells of Bcl11a 2/2 chimeras under these conditions (Fig. 6A, B), demonstrating that Bcl11a is required for pDC development in response to Flt3L both in vivo and in vitro. We also examined cDC development from WT and Bcl11a 2/2 fetal liver cells in vitro in response to treatment with Flt3L or GM-CSF. Flt3L-derived cDCs were markedly reduced in cultures of Bcl11a 2/2 fetal liver cells relative to cultures of WT fetal liver cells (Fig. 6C, E). However, GM-CSF-derived DCs developed in normal numbers from cultures of Bcl11a 2/2 fetal liver cells relative to cultures of WT fetal liver cells (Fig. 6D, F). These results suggest that Flt3L cannot signal through an alternative receptor to rescue cDC development in Bcl11a 2/2 cells, but that an alternative pathway of DC development may be mediated by GM-CSF.

Loss of Flt3L Results in Lineage-specific Defects in pDC Development
Next, we examined the development of splenic pDCs in the context of IL-7R or Flt3L deficiency. A previous study has demonstrated that splenic pDCs in Il7r 2/2 mice or Il7 2/2 mice are decreased in absolute number when compared to WT controls [37]. We found that splenic pDCs in Il7r 2/2 mice were not decreased in proportion to total splenocytes when compared to WT controls matched for strain, age, and sex (Fig. 7A). This result suggests that the hematopoietic defects in these mice may have relatively few lineage-specific consequences for pDC development.
Accordingly, and in contrast to our observations in Il7r 2/2 mice, we found that Flt3l 2/2 mice showed a greater than fourfold reduction in splenic pDC frequency as compared to WT littermate controls (Fig. 7B), in addition to reductions in absolute spleen size (data not shown).

Bcl11a is Required for Flt3 Expression in cDCs and Binds the Flt3 Genomic Locus
Because we observed cDC development in Bcl11a 2/2 chimeras, we assessed whether these cells might express Flt3 in a Bcl11aindependent manner; however, unlike WT cDCs in the BM, Bcl11a 2/2 cDCs in the same compartment showed no discernible Flt3 expression by flow cytometry (Fig. 7C), again suggesting that Shown is a Venn diagram of probe sets (excluding normalization controls) with a greater than twofold change in expression between WT and Bcl11a 2/2 MPPs. (C) Shown are log 2 -transformed ratios of gene expression in Bcl11a 2/2 MPPs relative to WT MPPs (x-axis) plotted against log 2 -transformed ratios of gene expression in WT CDPs relative to WT monocytes (ImmGen; y-axis). For clarity, probe sets with less than twofold changes in expression (log 2 -transformed ratios between 21 and 1) along either dimension are omitted (gray). (D) Shown is a heat map of log 2 -transformed gene expression in WT and Bcl11a 2/2 GMPs and MPPs for probe sets that constitute an ImmGen core cDC signature. Highlighted are genes that show a greater than twofold change in expression between WT and Bcl11a 2/2 GMPs (red) or between WT and Bcl11a 2/2 MPPs (green). doi:10.1371/journal.pone.0064800.g003 a Flt3-independent pathway is instead responsible for their development. Finally, to assay Bcl11a binding at the Flt3 locus, we performed chromatin immunoprecipitation (ChIP) using mouse pro-B cells. By quantitative real-time polymerase chain reaction (qPCR), we detected an approximately threefold enrichment at a region in the first intron of the Flt3 locus in DNA precipitated using anti-Bcl11a antibody as compared to isotype control (Fig. 7D). In summary, our results document a strict requirement for Bcl11a in pDC development both in vivo and in vitro; further, the requirement for Bcl11a in cDC development may differ based on the cytokine stimulus to which progenitors are exposed. The actions of Bcl11a include regulation of Flt3 expression by direct binding to the Flt3 locus, and Bcl11a is required for Flt3 expression in DCs and their progenitors.

Discussion
This study extends the known actions of Bcl1la in immune lineage development and provides a mechanism for its effects. Although Bcl11a has been recognized as a factor required for normal lymphoid development [24], the basis for this requirement has been unclear. It has been shown that Bcl11a acts upstream of the B cell factors Ebf1 and Pax5 and that Il7r mRNA is not expressed in Bcl11a 2/2 fetal livers [24]. Lack of IL-7Ra or the cytokine receptor common c chain (c c , encoded by Il2rg) severely impairs T and B cell development [39][40][41][42][43]. In T cell development, IL-7R signaling is thought to promote thymocyte survival, since Bcl-2 rescues impaired T cell development in Il7r 2/2 or Il2rg 2/2 mice [44][45][46][47][48]. In B cell development, Bcl-2 does not rescue development in the absence of IL-7R or c c [44][45][46][47]49], and IL-7R signaling is thought to induce expression of the transcription factor Ebf in the CLP [50][51][52].
Here, we demonstrate that Bcl11a is required for normal expression of IL-7R as early as the CLP and we add the novel observation that Bcl11a promotes the development of Flt3dependent lineages. Together, these actions provide a more complete account for previously observed defects in lymphocyte development in Bcl11a 2/2 mice, since T cell potential is preserved in IL-7-deficient CLPs in a Flt3L-dependent manner [53,54]. The mechanisms by which Bcl11a deficiency impairs T and B cell development, however, still remain incompletely explored. Consistent with a previous report [24], we confirmed the presence of residual T and B cells in the spleen of chimeras reconstituted with Bcl11a 2/2 fetal liver cells. By contrast, tamoxifen-induced deletion of Bcl11a in chimeras that have been reconstituted with Rosa26-CreERT2;Bcl11a flox/flox BM cells results in a more profound loss of T and B cells [55]. Thus, synchronous deletion of Bcl11a within a previously intact hematopoietic compartment produces a different outcome than does sustained deficiency throughout hematopoiesis. These results may point to a crucial lymphopoietic role for cells in which Bcl11a is dispensable for survival but necessary for development or maturation, or vice versa; these cells could include HSCs, mature T and B cells, or even residual CLPs undetectable due to a lack of IL-7R and Flt3 expression.
In line with a previous finding that E2-2 regulates Bcl11a expression [18], we also document a strict requirement for Bcl11a in the development of pDCs. The development of pDCs in vivo was lost in Bcl11a 2/2 fetal liver chimeras. In agreement, Flt3L cultures of BM derived from these Bcl11a 2/2 chimeras showed a complete loss of pDC development in vitro. Because mature pDCs are shortlived, non-proliferative, and continuously replenished from progenitor populations [56,57], the nearly complete loss of this population is most attributable to a developmental defect and not merely to cell survival defects in mature pDCs. This interpretation would be consistent with a finding that rescue of Bcl11a-deficient progenitors from increased apoptosis by p53 deficiency is unable to restore lymphoid potential [55].
Notably, however, in vitro development of cDCs was eliminated in Flt3L cultures of Bcl11a 2/2 fetal liver cells but was maintained in GM-CSF cultures of Bcl11a 2/2 fetal liver cells. Flt3L and GM-CSF have distinct, non-redundant actions in supporting cDC development [58,59]. The combined loss of Flt3L and GM-CSF causes a more severe cDC deficiency than loss of Flt3L alone; indeed, Flt3L-deficient mice retain an appreciable population of Flt3-expressing progenitors [38]. The maintenance of cDCs in Bcl11a 2/2 fetal liver chimeras suggests that these cells may rely on a Bcl11a-and Flt3-independent pathway for their development, survival, or expansion. Conditional knockout models would clarify which of these alternatives underlie the observed phenotype in Bcl11a 2/2 mice. Since DCs developed normally in vitro from Bcl11a 2/2 progenitors treated with GM-CSF, it is possible that Bcl11a 2/2 cDCs in vivo indeed represent development from GM-CSF-dependent progenitors, related perhaps to monocyte-derived dendritic cell lineages [60].

Ethics Statement
This study was carried out in strict accordance with recommendations in the U.S. National Research Council Guide for the Care and Use of Laboratory Animals. The protocol was approved by the Washington University Animal Studies Committee (#20090320).

Single-cell Suspensions of Fetal Liver
At embryonic day 14.5, fetal livers were mechanically dissociated with a syringe plunger and sterile 70-mm cell strainer (Fisher) into IMDM +10% (v/v) FCS (I10F). For subsequent cell culture or flow cytometry, red blood cells were lysed in ACK lysing buffer before counting by Vi-CELL (Beckman Coulter).

Flow Cytometry and Sorting
Staining was performed at 4uC in the presence of Fc block (clone 2.4G2, BD Biosciences or BioXCell) in FACS buffer (DPBS +0.5% BSA +2 mM EDTA). Cells were analyzed using a FACSCanto II (BD Biosciences) or sorted using a FACSAria (BD Biosciences); data were visualized using FlowJo software (TreeStar).

Chimeras
B6.SJL mice were lethally irradiated (1200 rad) and injected intraorbitally with 4610 6 fetal liver cells isolated from WT or Bcl11a 2/2 fetuses. After 4 or 6 weeks, BM was isolated by grinding and Histopaque-1119 (Sigma-Aldrich) centrifugation and either sorted by flow cytometry or cultured. From these mice, thymi were minced and digested in 250 mg/ml collagenase B (Roche) and 30 U/ml DNase I (Sigma-Aldrich) and analyzed by flow cytometry.

Microarray Analysis
MPP and GMP populations were sorted from fetal liver chimeras and pooled by donor genotype. RNA was isolated using an RNAqueous-Micro Kit (Ambion) and submitted for amplification, labeling and hybridization. Expression values were analyzed after RMA quantile normalization using ArrayStar software (DNASTAR). Data were deposited in the Gene Expression Omnibus (GEO) repository under accession no. GSE46270.

Statistics
Statistical differences were identified using Prism 6 (GraphPad) by multiple unpaired Student's t tests, controlling the false discovery rate (Q) by the method of Benjamini and Hochberg. *, Q = 0.05; **, Q = 0.01.