Transcriptional Reprogramming of CD11b+Esamhi Dendritic Cell Identity and Function by Loss of Runx3

Classical dendritic cells (cDC) are specialized antigen-presenting cells mediating immunity and tolerance. cDC cell-lineage decisions are largely controlled by transcriptional factor regulatory cascades. Using an in vivo cell-specific targeting of Runx3 at various stages of DC lineage development we show that Runx3 is required for cell-identity, homeostasis and function of splenic Esamhi DC. Ablation of Runx3 in DC progenitors led to a substantial decrease in splenic CD4+/CD11b+ DC. Combined chromatin immunoprecipitation sequencing and gene expression analysis of purified DC-subsets revealed that Runx3 is a key gene expression regulator that facilitates specification and homeostasis of CD11b+Esamhi DC. Mechanistically, loss of Runx3 alters Esamhi DC gene expression to a signature characteristic of WT Esamlow DC. This transcriptional reprogramming caused a cellular change that diminished phagocytosis and hampered Runx3-/- Esamhi DC capacity to prime CD4+ T cells, attesting to the significant role of Runx3 in specifying Esamhi DC identity and function.


Introduction
Classical dendritic cells (cDC) are a sparsely distributed, migratory group of bone-marrow-(BM) derived immune cells specialized in antigen uptake and presentation [1,2] and as such important mediators of immunity and tolerance [3,4]. cDC are divided into two main groups: migratory DC (mDC) and stationary DC (sDC). mDC are scattered in most body organs including lung, gut and skin and act as sentinels by sampling environmental pathogens. Upon activation, mDC migrate to secondary lymphoid organs and present processed antigens to T cells. sDC, on the other hand, sample and process antigens directly in the lymphoid organs. These sDC are heterogeneous with respect to their cell-surface markers and function and have based on surface molecules been divided into three major types: XCR1 + CD8α + /CD4 -/CD11b -, CD4 + /CD11b + /CD8α -and CD4 − /CD8α − CD11b + [5]. In addition, splenic CD8α + and CD4 + subsets differ in MHCI and MHCII processing pathways, reflected in their ability to preferentially induce CD4 or CD8 T cell responses [6].
Despite their phenotypic and functional heterogeneity all splenic cDC are derived from a common DC precursor (CDP) [7]. BM CDP give rise to plasmacytoid DC (pDC) and circulating pre-DC that seed the peripheral tissues [7]. The current classification defines two distinct subclasses of splenic CD11b + cDC, as CD11b + Esam hi and CD11b + Esam low cells [8]. Gene expression profiling and functional analysis of these two CD11b + cDC subclasses have suggested that CD11b + Esam low DC are related to monocytes rather than to cDC and therefore could have derived from the monocyte/dendritic cell precursor (MDP) without involving a CDP intermediate [8]. Generation of CD11b + Esam hi /CD11b + Esam low DC is controlled by the Notch2 pathway, so that loss of Notch2 signaling selectively affects the development of CD11b + Esam hi DC [8].
Several transcription factors (TF) affect DC lineage development and function. Mice lacking the TF Id2, Irf8 or Batf3 are deficient in CD8 + DC [9][10][11] and loss of Irf4 TF affects CD4 + DC development [12]. On the other hand, loss of the cDC-specific TF zDC (Zbtb46) [13,14] did not impair DC development, but was associated with increased activation of naive cDC [15].
Runx3 TF is highly expressed in mature BM derived cultured DC and mediates their response to TGF-ß [16]. Here we explored Runx3 function throughout DC lineage development using mice that lack Runx3 specifically in DC and their progenitors. Runx3 ablation at defined developmental DC stages led to largely impaired splenic CD4 + /CD11b + DC compartment. Combined chromatin immunoprecipitation sequencing (ChIP-seq) and gene expression analysis revealed that Runx3 acts as a key gene expression regulator of CD11b + Esam hi DC homeostasis. Accordingly, loss of Runx3 altered gene expression profile of the residual Runx3 deleted (Runx3 Δ ) Esam hi DC to a profile characteristic of WT Esam low DC subtype. Moreover, this transcriptional reprogramming yielded functionally impaired Runx3 Δ Esam hi DC compromising their ability to prime CD4 + T cells. The results defined Runx3regulated target genes that participate in Runx3-mediated DC lineage development and function.

RNA isolation and microarray gene expression analysis
Total RNA from sorted DC was extracted using the miRNeasy Mini Kit (QIAGEN) including DNase digest (QIAGEN). RNA purity was assessed with a BioAnalyzer 2100 (Agilent Technologies). Samples of 250ng RNA from each isolated DC population was labeled and hybridized to Affymetrix mouse exon ST 1.0 microarrays according to manufacturer instructions. Microarrays were scanned using GeneChip scanner 3000 7G. Each experiment was conducted in duplicates or triplicates using RNA from independent FACS isolated cells. Statistical analysis was performed using the Partek® Genomics Suite (Partek Inc., St. Louis, Missouri 63141) software. CEL files (containing raw expression measurements) were imported to Partek GS. The data was preprocessed and normalized using the RMA (Robust Multichip Average) algorithm [21] with GC correction. To identify differentially expressed genes, One-Way ANOVA was applied. Differential gene lists were created by filtering the genes based on: absolute fold change >= 1.5, p <=0.05 and signal above background in at least one microarray (log 2 intensity >=6.5). Log 2 gene intensities were used for scatter plots (Partek) and matrix plots containing: correlation, histogram and scatter plots (using R language). Gene expression clusters for 224 transcripts were generated using log 2 ratios with the Expander SOM tool [22]. These transcripts were differentially expressed in our and in [8] experiments and were above background in both datasets. Microarray data is available at Gene Expression Omnibus (GEO) under accession numbers GSE48589 (Runx3 function in CD4 + splenic DC) and GSE48590 (The affect of specific ablation of Runx3 from Esam splenic DC).

ChIP-seq data acquisition and analysis
Two biological replicate ChIP-Seq experiments were conducted for detection of Runx3-bound genomic regions using in-house anti Runx3 Ab and 30x10 6 positive CD11c MACS isolated (Miltenyi Biotec) DC. These cells were further purified by FACS to CD11c hi MHCII + CD4 + cDC subset. Isolated cells were fixed in 1% formaldehyde and sonicated to yield DNA fragments of ~300bp according to standard procedures previously summarized in [23]. For immunoprecipitation, 40ul of anti-Runx3 Ab were added to 15 mL of diluted, fragmented chromatin; whole cell extract fragmented chromatin served as control. DNA was purified using QIAquick spin columns (QIAGEN). For ChIP-seq analysis, Illumina sequencing of short reads (50 bp) was performed in one lane of Hiseq 2000 using v2 clustering and sequencing reagents. For data analysis 70 million Runx3 IP and 74 million whole genome extract sequences were aligned uniquely to the mouse genome (mm9) using bowtie [24]. Bound regions were detected using MACS2 [25]. Runx3 bound peaks and coverage data (bigWig files) were uploaded to the UCSC genome browser [26]. GREAT algorithm was applied to determine genes corresponding to Runx3-bound peaks and ontology enrichment analysis [27]. Cistrome CEAS [28] was used to calculate average profile read near TSS and to generate venn diagrams. De-novo discovery of TF binding sites was conducted using DREME [29]. Analysis of co-bound zDC -Runx3 peaks was done using bedtools [30]. Statistical evaluation of fold enrichment distributions was done using the R package. ChIP-seq data is available at Gene Expression Omnibus (GEO) under accession number GSE48588 (Genome-wide maps of Runx3 bound regions in CD4 + splenic DC).

RT-qPCR and Western blotting
Total RNA was reverse-transcribed using miScript reverse transcription kit (QIAGEN) according to manufacturer's instructions. Quantitation of cDNAs was performed by qPCR using Roche LC480 LightCycler with sequence-specific primers and miScript SYBR Green PCR kit (QIAGEN). Target transcript quantification was calculated relative to ACTB mRNA, which served as an internal control. Standard errors were calculated using REST [31]. For Western blot analysis FACS isolated DC sub-populations were collected, washed once in PBS, proteins extracted with RIPA buffer and analyzed by Western blotting with anti Runx3 Ab [17]. GAPDH was used as an internal loading control.

T-cell proliferation and Phagocytosis assays
CD4 + and CD8 + OVA-specific T cells were isolated from spleens and lymph nodes of OT-II and OT-I TCR-transgenic mice and CD4 or CD8 cells isolated by MACS (Miltenyi Biotec). Cells were labeled with CFSE (Invitrogen) and coinjected into the tail veins of recipient mice (2 X 10 6 cells/mouse). Twentyfour hours later, 20ug of soluble OVA (Sigma-Aldrich) per mouse was injected. Analysis of T-cell proliferation in spleens of recipient mice was performed 96 hours following T-cell transfer. For phagocytosis assay 200µl of 2.66% solid Fluoresbrite Carboxylate were injected into the tail veins with FITC labeled 0.5µm latex microspheres (Polysciences, Warrington, PA). Analysis of splenic DC microspheres uptake was performed 5 and 18 hours following injection.

Statistical analysis
Statistical significance was determined with unpaired, two tailed Student's t test. To identify differentially expressed genes, One-Way ANOVA was applied. Spearman correlation between microarray intensities was calculated using the R package. zDC peaks were found to be overlapping Runx3 peaks more than expected by chance using the hyperbrowser [32]. Specifically the test applied preserved the zDC regions and the number of Runx3 peaks and randomized (Monto Carlo test) the positions of Runx3 peaks with a constrain to the same chromosome and to a similar distance from the TSS. Same p value resulted in the opposite test (i.e. randomizing the zDC peaks). Kolmogorov-Smirnov test was applied on the Runx3 peaks fold enrichment distributions using the R package (ks.test) (http://www.R-project .org/).
Reporter gene expression was absent from blood monocytes and plasmacytoid DC (pDC) ( Figure S1B). Among splenic CD11c hi MHCII + cDC Runx3/GFP expression was readily detected in CD4 + CD8 -DC and DN populations, but absent from CD8 + CD4 -DC ( Figure 1C). Further analysis indicated that Runx3 expression was confined to CD11c + CD11b + cDC ( Figure  S1C) with Runx3/GFP highly expressed in both Esam hi and Esam low sub-populations ( Figure S1D). Lewis et al. recently proposed that Esam hi and Esam low DC subsets might originate along different pathways with only Esam hi CD11c hi CD11b + DC, but not Esam low CD11c hi CD11b + DC involving a CDP intermediate [8].
Interestingly, a similar reporter gene expression pattern was detected in cells derived from either Runx3 P1-AFP/+ or Runx3 P2-EGFP/+ mice ( Figure S1A and E), indicating that Runx3 expression in BM DC-progenitor subsets and mature CD11c + CD11b + DC was driven by both P1 and P2 promoters. In summary, Runx3 expression was readily detected in subpopulations of BM CMP, MDP and pre-DC as well as in splenic CD11b + DC, whereas Runx3 expression was absent from CD8 + DC, monocytes and pDC populations.
FACS analysis of Cebpα-DC-Runx3 Δ or CD11c-DC-Runx3 Δ splenocytes revealed a pronounced reduction in CD11c hi MHCII + cells (Figure 2A). Detailed analysis of the splenic DC compartment indicated that loss of Runx3 specifically affected the CD11b + subclass in both DC-Runx3 Δ mouse strains ( Figure  2B-D). No other hematopoietic cell populations besides CD11c hi MHCII + CD11b + DC seemed to be compromised, as evidenced by unaffected frequencies of B cells, T cells and pDC in both DC-Runx3 Δ mice compared to littermate controls ( Figure 2C and D). Of note, the reduction in splenic CD4 + and DN DC affected both Esam hi and Esam low DC in the Cebpα-DC-Runx3 Δ and CD11c-DC-Runx3 Δ mouse strains ( Figure 2E). The reduction of Esam hi and Esam low DC in these mice was however not associated with increased apoptosis ( Figure 2F). FACS analysis of Runx3 -/-mice [18] that lacked Runx3 throughout myelopoiesis recapitulated the DC phenotype observed in the DC-Runx3 Δ mouse strains ( Figure S2B-D).
The finding that the two DC-lineage stage-specific Runx3 Δ mouse strains, as well as germ-line Runx3-mutants displayed a similar impairment of their DC compartment suggested that Runx3 may not be required at a specific lineage decision stage, but rather be involved in DC homeostasis. To address this issue we examined the ability of CD11c-DC-Runx3 Δ BM cells to replenish the DC compartment by conducting a competitive BM-repopulation assay. A 1:1 mixture of CD11c-DC-Runx3 Δ BM cells (CD45.2) and WT BM cells (CD45.1) was transferred into lethally irradiated recipient mice (CD45.1). Six to eight weeks following the transfer spleens from experimental (CD11c-DC-Runx3 Δ /wt > wt) and control (Runx3 +/+ /wt > wt) chimeras were analyzed for the distribution of CD4 + /CD8 + and DN DC. CD11c-DC-Runx3 +/+ BM cells of the control group efficiently repopulated the three DC subsets successfully competing with the WT BM ( Figure 2G left). In striking contrast, CD11c-DC-Runx3 Δ BM cells were severely impaired in generating CD11c hi MHCII + cells ( Figure 2G right). Specifically, CD4 + Runx3 Δ DC and DN Runx3 Δ DC were found outcompeted by their WT counterparts, whereas Runx3 Δ CD8 + were overrepresented relative to WT CD8 + DC ( Figure 2G right). Collectively, these findings demonstrate a cell-autonomous Runx3 function in CD4 + DC homeostasis with Runx3 loss significantly reducing splenic CD4 + DC numbers (Figure 2).

Runx3 is required for lineage identity and homeostasis of Esam hi DC
To elucidate the cell-intrinsic regulatory role of Runx3 during splenic CD4 + DC homeostasis we sought to identify Runx3responsive genes driving this process. Comparison of Runx3 -/and WT CD4 + DC expression profiles revealed 412 differentially expressed genes ( Figure 3A and Table S1). Significantly, several of these Runx3-responsive genes had previously been implicated in DC lineage development and homeostasis. For example, Irf8 [10] and Spic/PU.1 [33] were upregulated while Irf4 [12] was downregulated in Runx3deficient CD4 + DC ( Figure 3A and Figure S3A). Moreover, comparative analysis of WT and Runx3 Δ CD4 + DC revealed elevated levels of Esam and CD11b and reduced expression of CD11c in CD11c + -Runx3 Δ DC ( Figure 3B and Figure S3B), suggesting that loss of Runx3 particularly affected the Esam hi DC population. This conclusion is supported by the earlier findings that Runx3 is highly expressed in both CD11b + Esam hi and CD11b + Esam low DC ( Figure S1C and D), but its loss preferentially affected the CD11b + /Esam hi population ( Figure 2).
Subsequently, we analyzed differential gene expression in Runx3 Δ and WT CD11b + Esam hi DC ( Figure 3C). Interestingly, 66 (30%) of the 202 Runx3-responsive genes upregulated in Runx3 Δ CD11b + Esam hi DC were previously identified as Esam low DC specific [8] (Table S2). Consequently, the gene expression signature of Runx3 Δ CD11b + Esam hi DC corresponded to that of WT Esam low DC (Spearman correlation values of 0.56 and 0.57 for Runx3 Δ Esam hi /WTEsam low ) ( Figure  3D), while its correlation value with WT CD11b + Esam hi (Runx3 Δ Esam hi /WT-Esam hi ) was markedly diminished to 0.34 and 0.37 ( Figure 3D). These findings indicated that loss of Runx3 caused a shift of the CD11b + Esam hi population towards a CD11b + Esam low cell-identity manifested in altered gene expression. This conclusion was supported by the observation that the Runx3 -/-CD4 + DC gene expression profile highly correlated with CD11b + Esam low DC ( Figure S3C), but also with that of Rbpj Δ CD11b + Esam low DC [8]. These latter Esam low DC were derived from mice lacking the Notch2-pathway TF Rbpj that regulates Esam expression [8]. Specifically, the correlation values of Runx3 -/-Esam hi to WT Esam low DC and that of Runx3 -/-Esam hi to Rbpj Δ Esam low DC were 0.72 and 0.70, respectively, compared to a correlation value of 0.52 for Runx3 -/-Esam hi to WT Esam hi DC ( Figure S3C).
We next integrated the Runx3 ChIP-seq data (see below) with the differential gene expression results of Runx3 Δ CD11b + Esam hi DC and Runx3 -/-CD4 + DC. Using this dataset we performed clustering analysis using 224 Runx3-responsive gene transcripts compared to WT CD11b + Esam low and Rbpj Δ CD11b + Esam low DC datasets [8]. Six clusters were identified ( Figure 3E). Cluster 1-5 contained 186 Runx3responsive genes (either down or up -regulated), of which 127 (~70%) were bound by Runx3. Importantly, clusters 1-4 included 156 genes that were upregulated in both Runx3deficient CD11b + Esam hi DC and CD4 + DC experiments and displayed a similar gene expression profile to that of WT CD11b + Esam low DC previously reported [8]. This observation indicated that in WT CD11b + Esam hi DC Runx3 negatively regulates these genes. We next used RT-qPCR analysis of RNA from Runx3 Δ CD11b + Esam hi , WT CD11b + Esam hi or WT CD11b + Esam low DC or FACS analysis to validate the clustering microarray gene expression data. This analysis confirmed that key WT CD11b + Esam low DC-specific genes, including Clec12a, Cx 3 cr1, Irf8 and Flt3 were markedly upregulated in Runx3 Δ CD11b + Esam hi DC ( Figure 3F and G) along with genes such as Kit and Itgax that were downregulated upon loss of Runx3 ( Figure 3B and G). Taken together the data support the conclusion that Runx3 normally represses a number of critical genes to facilitate the maintenance of Esam hi cell identity. Upon loss of Runx3, Runx3 Δ Esam hi DC acquired a gene expression profile characteristic to Esam low DC. Interestingly, this cell identity shift of Esam hi to Esam low DC did not affect their Impaired Homeostasis of Runx3-/-CD11b+Esamhi DC PLOS ONE | www.plosone.org  Splenic CD4 + DC from three Runx3 Δ and WT mice were isolated by FACS, RNA was isolated and subjected to microarray analysis. Shown are the mean intensities (log 2 ) of WT and Runx3 Δ CD4 + DC. Genes that were up-or down-regulated due to Runx3 Δ are marked by red or blue, respectively. Differential expression cut-off was set to minimal absolute fold-change of 1.5, and p-value<=0.05. Selected relevant genes are indicated. (B) Expression of Esam (upper panel) CD11b (middle panel) and CD11c (lower panel) in splenic DC subsets from Runx3 +/+ (Runx3 fl/fl ) mice (red line) and CD11c-DC-Runx3 Δ mice (blue line). (C) Scatter plot of differentially expressed genes in CD4 + Esam hi DC from WT and CD11c-DC-Runx3 Δ mice. Splenic Esam hi DC were obtained from two individual mice. Genes that were up-or down-regulated due to Runx3 Δ are marked by red or blue, respectively. Differential expression cut-off was set to minimal absolute fold-change of 1.5, and p-value<=0.05. Selected relevant genes are indicated. (D) Gene expression log 2 pairwise comparisons between WT-CD4 + Esam hi , CD11c-Runx3 Δ CD4 + Esam hi subsets and the published Esam experiment [8]. maturation state, as evidenced by the level of activation markers ( Figure S3D).

Identification of Runx3-regulated genes in splenic CD4 + DC
We next sought to identify among Runx3-responsive genes those that were directly regulated by Runx3 and could therefore contribute to the cell-phenotypic shift of Esam hi DC to Esam low DC. ChIP-seq was conducted using FACS-isolated CD11c hi MHCII + CD4 + splenic WT DC and anti-Runx3 antibodies (Ab). Data analysis revealed that of the 15121 Runx3-bound regions 9938 were associated with annotated genes ( Figure  4A). Location analysis of these Runx3-occupied regions relative to the nearest transcription start sites (TSS) of the annotated genes, revealed that ~50% were placed ±10kb from a gene TSS ( Figure S4A and B). Runx3 ChIP-seq data analysis using GREAT [27] indicated that Runx3-bound regions were highly enriched for genes that belong to specific functional categories, including antigen processing and presentation ( Figure S4C). To identify among the CD4 + DC Runx3-responsive genes those that were Runx3-regulated we cross-analyzed Runx3 ChIP-Seq and differential gene expression datasets. Of the 412 Runx3-responsive genes 316 were bound by Runx3 and defined as Runx3-regulated ( Figure  4A). Forty percent (127) of these Runx3-regulated genes belonged to the subset characteristic of WT CD11b + Esam low DC in Runx3-deficient CD4 + DC ( Figure 3E), underscoring the crucial role of Runx3 in specifying the Esam hi DC gene expression signature. Among these Runx3-responsive genes, Runx3 was found to occupy the genomic loci of Clec12a, Cx 3 cr1, Flt3, Rbpj, Itgam and Itgax ( Figure 4B).
Sequence analysis revealed that 86% of Runx3-occupied genomic regions contained the canonic RUNX motif TGTGGT ( Figure 4C). Interestingly, Runx3-bound regions were also enriched for the motif TGACGT comprising the binding site of DC-specific TF zDC/zbtb46 [15] ( Figure 4C). Moreover, comparison of the published zDC ChIP-seq dataset [15] to Runx3-occupied regions in splenic CD4 + DC revealed that a high proportion (66%) of zDC bound genes was co-bound by Runx3 and that the zDC/Runx3 co-occupied regions were significantly enriched in Runx3 binding ( Figure S4D). This observation is consistent with the possibility that Runx3 and zDC cooperate in regulating DC specific gene expression particularly as similar to Runx3, zDC participates in homeostasis of splenic CD4 + DC [13].

Runx3 is required for CD4 + DC-mediated T-cell priming
T cell priming and activation by DC link innate and adaptive immunity and modulate tolerance vs. immune response. CD4 + T cells are preferentially primed by CD4 + DC in a MHC-II dependent manner [6,34]. CD4 + DC-dependent T cell priming was reported to be impaired in DC-Rbpj Δ mice that lack CD11b + Esam hi DC [8], raising the possibility that also the Runx3 Δdependent phenotypic shift of CD4 + Esam hi to CD4 + Esam low DC would be associated with impaired T cell immunity. To address this scenario, we isolated Ovalbumin (OVA)-specific CD8 + and CD4 + T cells from TCR transgenic OT-I and OT-II mice, labeled them with CFSE, to allow monitoring of in vivo proliferation and transferred the cells into WT control and CD11c-DC-Runx3 Δ mice. Donor mice bore the congenic marker CD45.1, allowing detection of grafted cells in the CD45.2 + recipient mice. Four days following immunization with OVA the OT-I CD8 + T cell graft underwent pronounced proliferation in both WT and CD11c-DC-Runx3 Δ recipients ( Figure 5A upper panel). Thus, the lack of Runx3 did not influence the cross-priming efficiency of CD8 + DC. In contrast, the response of grafted OT-II CD4 + T cells to the OVA challenge was impaired in CD11c-DC-Runx3 Δ mice as compared to control recipients ( Figure 5A lower panel, B and Figure S5A), attesting to the important role of Runx3 in CD4 + DC ability to prime T cells. As Esam low CD11b + DC lack MHC-II priming capacity [8] the diminished CD4 + T cell priming in DC-Runx3 Δ mice is most probably due to an intrinsic cell defect in the Esam hi CD4 + DC. Consistent with this conclusion, Runx3 bound ( Figure 5C) and regulated expression ( Figure  S4C) of classical MHC-II genes and its loss could therefore affect MHC-II presentation. Together, the data indicate that Runx3 not only participates in promotion and maintenance of Esam hi DC identity, but also contributes to their CD4 + T cell priming activity.

Reduced phagocytic capacity of Runx3-deficient splenic CD11b + DC
DC-mediated adaptive immune responses involve antigen phagocytosis followed by MHC-dependent T cell priming. The crucial contribution of Runx3 to the ability of Esam hi CD4 + DC to prime CD4 + T cells posed the question as to whether it was also required for phagocytic DC activity. To directly evaluate this possibility we injected 1x10 10 FITC-labeled latex beads to WT and DC-Runx3 Δ mice, and monitored uptake of fluorescent beads by splenic DC, 5 and 18 hours later ( Figure 5A and Figure S5B). Of note, Runx3 Δ CD4 + /CD11b + DC displayed a reduced capacity to phagocytose the beads compared to WT CD4 + /CD11b + DC, while Runx3 Δ CD8 + DC exhibited phagocytic capacity similar to WT DC ( Figure 5D). Moreover, when we divided the CD11b + DC population and analyzed the phagocytic ability of WT Esam hi and Esam low DC independently, we found that Esam hi DC exhibited a higher phagocytosis potential than their Esam low DC counterparts ( Figure 5E and Figure 5B). However, this higher phagocytic capacity of WT Esam hi DC diminished upon Runx3 ablation ( Figure 5E and Figure S5C). Even more convincingly, injection of 1x10 10 FITClabeled latex beads into chimeras of a 1:1 mixture of CD11c-DC-Runx3 Δ BM cells (CD45.2) and WT BM cells (CD45.1) recapitulated the aberrant phagocytic phenotype of CD4 + / CD11b + DC ( Figure 5F). Collectively these results indicated that in naive CD11b + /CD4 + /Esam hi DC, Runx3 regulates basic cDC functions including phagocytosis and antigen-processing.

Discussion
Extensive efforts have been made over the past decade to elucidate the mechanisms underlying the production of DCprogenitors and to delineate the heterogeneity and functions of their downstream lineage cell subsets [7,35,36]. However, information about the role of specific TF in DC development, subset specification and homeostasis is still incomplete. Using Impaired Homeostasis of Runx3-/-CD11b+Esamhi DC PLOS ONE | www.plosone.org   In good correlation with the expression pattern, inactivation of Runx3 at distinct stages of DC lineage development led to profound reduction of the CD4 + Esam hi DC subset. Interestingly, this Runx3-ablation-dependent reduction in CD4 + Esam hi DC was associated with extensive changes in CD4 + Esam hi DC gene expression, causing a shift in cell identity. Specifically, a considerable number of Runx3-regulated genes that are normally repressed in WT CD4 + Esam hi DC became activated in Runx3 Δ CD4 + Esam hi DC. Consequently, Runx3 Δ CD4 + Esam hi DC acquired a gene expression signature resembling that of WT Esam low DC. Importantly, this gene expression shift of Runx3 Δ CD4 + Esam hi led to acquisition of phenotypic features characterizing the WT Esam low DC subset [8], including a compromised CD4 + T cell priming capacity, as well as decreased phagocytosis.
The finding that in CD4 + DC the zDC/Runx3 co-bound regions comprised more than 50% of the genomic regions occupied by this recently identified DC-specific TF [15] suggest that the two TF cooperate in driving the CD4 + Esam hi transcriptional program and may collaborate in repressing Runx3-regulated Esam low DC genes. Of particular relevance to these recent findings is the observation that inactivation of the E2-2 TF during pDC maturation led to a major change in pDC gene expression [37,38] so that E2-2-deficient pDC displayed induced or repressed expression of genes characteristic to CD8 + DC and pDC respectively [37]. Both TGF-β and Notch signaling pathways are important for DC development and function [20,39]. These two pathways crosstalk through protein-protein interaction of Notch intracellular domain with Smad3 [40]. Fainaru et al. have shown involvement of Runx3 in TGF-β dependent DC function [16] and Runx3 participates in Notchmediated endothelial/mesenchymal transition [41]. Expression of the Esam receptor by CD11b + DC is regulated by the Notch2-Rbpj signaling [8]. In this context, it is of note that Runx3 Δ CD4 + Esam hi mice display elevated expression of Esam indicating Runx3 involvement in DC Notch2 signaling pathway. The idea that Runx3 participates in Notch2/TGF-β crosstalk during DC development is further supported by data documenting significant increased of Smad3 expression in Runx3 Δ CD4 + DC. In summary, Runx3 play a crucial role in specifying Esam hi DC identity and function. Accordingly, loss of Runx3 induces transcriptional reprogramming manifested in phenotypic shift of Esam hi to Esam low DC that compromises CD4 + T cell priming activity.