Differential usage of transcriptional repressor Zeb2 enhancers, distinguishes adult and embryonic hematopoiesis.

SUMMARY

The transcriptional repressor ZEB2 regulates development of many cell fates among somatic, neural and hematopoietic lineages, but the basis for its requirement in these diverse lineages is unclear. Here, we identified a 400 basepair (bp) region located 165 kilobases (kb) upstream of the Zeb2 transcriptional start site (TSS) that binds the E proteins at several E-box motifs and was active in hematopoietic lineages. Germline deletion of this 400bp region (Zeb2^Δ−165 mice) specifically prevented Zeb2 expression in hematopoietic stem cell (HSC) derived lineages. Zeb2^Δ−165 mice lacked development of plasmacytoid dendritic cells (pDCs), monocytes, and B cells. All macrophages in Zeb2^Δ−165 mice were exclusively of embryonic origin. Using single-cell chromatin profiling, we identified a second Zeb2 enhancer located at +164-kb that was selectively active in embryonically derived lineages, but not hematopoietic stem cells (HSCs) derived ones. Thus, Zeb2 expression in adult, but not embryonic, hematopoiesis is selectively controlled by the −165kb Zeb2 enhancer.

eTOC blurb

ZEB2 is required for the development and functions of multiple hematopoietic lineages. Huang et al. demonstrate that the distinct usage of several enhancers identified in this study is the molecular basis for the differential regulation of ZEB2 expression between adult and embryonic hematopoiesis.

Graphical Abstract

INTRODUCTION

Mammalian hematopoiesis occurs in three phases. Primitive hematopoiesis begins in the yolk sac (YS) around embryonic age 7.0 (E7.0) producing nucleated erythrocytes and macrophages (MFs) (Moore and Metcalf, 1970). Transient definitive hematopoiesis begins in the YS around E8.5, producing erythro-myeloid progenitors (EMPs) and lympho-myeloid progenitors (Palis et al., 1999; Frame et al., 2013; Lin et al., 2014). Finally, definitive hematopoiesis begins when hematopoietic stem cells (HSCs) colonize the fetal liver (FL) around E10.5 and persists in the bone marrow (BM) throughout adult life.

Hematopoiesis requires precise control by lineage-determining transcription factors (TFs) whose expression is regulated mainly by cis-acting enhancer elements (Graf and Enver, 2009; Orkin and Zon, 2008; Hoppe et al., 2016; Long et al., 2016). Multiple enhancers may act within a single locus to control gene expression in different cell types (Frankel et al., 2010; Osterwalder et al., 2018). Enhancer usage can change dynamically during the differentiation of progenitor cells, reflecting distinct gene expression profiles enforced by transcriptional circuitries at different stages of lineage development (Lara-Astiaso et al., 2014; Luyten et al., 2014).

Dendritic cells (DCs) and tissue-resident macrophages (RTMs) are two closely-related mononuclear phagocytic lineages that are important in antigen presentation and induction of adaptive immunity (Steinman and Cohn, 1973; Ginhoux and Jung, 2014; Lavin and Merad, 2013). DCs are comprised of classical dendritic cells (cDCs) and plasmacytoid DCs (pDCs) (Steinman and Cohn, 1973; Cella et al., 1999), and are derived from HSCs during definitive hematopoiesis (Fogg et al., 2006; Naik et al., 2007). By contrast, RTMs can develop both from transient definitive hematopoiesis during embryogenesis and from definitive hematopoiesis in the adult (Ginhoux and Guilliams, 2016). During transient definitive hematopoiesis, EMPs can differentiate locally into YS MFs or migrate to the FL to produce monocytes (Ginhoux and Guilliams, 2016). YS MFs and FL monocytes are the precursors that give rise to the majority of RTMs in the adult (Hoeffel et al., 2015; Hoeffel and Ginhoux, 2015). During adult definitive hematopoiesis, BM-derived circulating monocytes can also contribute to the RTM population, but it is not known whether embryonic RTM development relies on the same transcriptional mechanisms as in the adult (Liu et al., 2019; Ginhoux and Guilliams, 2016).

Zinc finger E-box binding homeobox 2 (ZEB2) is a transcriptional repressor required for both DC and MF lineages (Wu et al., 2016; Scott et al., 2016; Scott et al., 2018). ZEB2 was identified as a SMAD-interacting protein regulating epithelial-mesenchymal transition (Verschueren et al., 1999; Remacle et al., 1999; Comijn et al., 2001; van Grunsven et al., 2003). Germline deletion of Zeb2 in mice causes embryonic lethality (Higashi et al., 2002; Van de Putte et al., 2003), and ZEB2 acts in development of the nervous system and melanocytes (Hegarty et al., 2015; Denecker et al., 2014). Conditional deletion of Zeb2 in hematopoietic lineages causes a defect in HSC mobilization and results in neonatal lethality (Goossens et al., 2011). ZEB2 functions in NK cell maturation (van Helden et al., 2015), in terminally differentiated T cells (Dominguez et al., 2015; Guan et al., 2018; Omilusik et al., 2015; Omilusik et al., 2018), and inducible deletion of Zeb2 in adult mice with an Mx1-cre system results in a 80% reduction in B cell numbers in peripheral blood (Li et al., 2016). Zeb2 is also required for development of dendritic cells(Wu et al., 2016; Scott et al., 2016; Bagadia et al., 2019a), monocytes (Wu et al., 2016), and for the maintenance of RTM identity (Scott et al., 2018), suggesting that ZEB2 is active in cells derived both from embryonic and adult hematopoiesis.

Despite its essential role in the development of multiple hematopoietic lineages (Bagadia et al., 2019b; Scott and Omilusik, 2019), the transcriptional regulation of Zeb2 is incompletely characterized. Some studies have suggested that the TF T-bet might regulate Zeb2 expression in T cells and NK cells (van Helden et al., 2015; Omilusik et al., 2015; Dominguez et al., 2015), while another study implicates c-Myb as important for Zeb2 expression in CD8⁺ T cells (Gautam et al., 2019). We recently described a transcriptional circuit of mutual repression between Id2 and Zeb2 operating in the common dendritic cell progenitor (CDP) to resolve the fate choice between pDCs and type 1 cDCs (cDC1)(Bagadia et al., 2019a; Durai et al., 2019). Since ID2 can act by inhibiting the activity of E-proteins, we proposed that E proteins may positively regulate Zeb2 expression in the CDP, but did not identify the regulatory elements that might mediate this control.

Here, we used chromatin immunoprecipitation and sequencing (ChIP-seq) and assay for transposase-accessible chromatin using sequencing (ATAC-seq) to identify potential regulatory regions in the Zeb2 locus. We found an enhancer located 165kb upstream of the Zeb2 transcriptional start site (TSS) that was bound by E proteins and contains a cluster of 4 E-box consensus motifs (CANNTG). Deletion of the 400 bp region containing these 4 sites using CRISPR/Cas9 genome editing showed that it was required for all expression of Zeb2 in the bone marrow (BM) and essential for the development of pDC, monocytes, and B cells, suggesting that these three lineages share similar transcriptional circuitry for their control of Zeb2 expression. Notably, this enhancer was not required for development of embryonically derived RTMs or for ZEB2 expression in RTMs, but was required for monocyte-derived tissue MF development. Single cell ATAC-seq identified a second enhancer located 164-kb downstream of the Zeb2 TSS that is selectively active in embryonically derived MF progenitors and RTM arising from transient definitive hematopoiesis, but not adult definitive hematopoiesis. Thus, the use of the - 165-kb Zeb2 enhancers by cell lineages distinguishes adult and embryonic hematopoiesis.

RESULTS

E proteins family transcription factors bind a −165-kb Zeb2 enhancer in both mouse and human.

We previously found a mutually repressive circuit for Zeb2 and Id2 operating during DC development(Bagadia et al., 2019a). Since ID2 acts by forming inactive heterodimers with E-proteins, we wondered whether E-proteins could positively regulate Zeb2 expression. To test this, we examined ChIP-seq data for E2-A (encoded by Tcf3) performed in mouse HPC-7 cells (Calero-Nieto et al., 2014). This analysis identified two prominent E2-A-binding peaks in the Zeb2 locus, a peak located 59-kb upstream of Zeb2 TSS (−59-kb Zeb2) and another peak located 165-kb upstream of Zeb2 TSS (−165-kb Zeb2) (Figure 1A). ATAC-seq of hematopoietic lineages revealed that the −59-kb Zeb2 enhancer was active only in HSCs but not in any terminally differentiated lineages tested, suggesting an early function of this enhancer (Figure 1B). The - 165-kb Zeb2 enhancer was active in multiple lineages, including pDCs and monocytes, and its accessibility largely correlated with the amount of Zeb2 mRNA in these lineages (Wu et al., 2016) (Figure 1B).

Figure 1. E proteins engage the −165–kb Zeb2 enhancer to regulate Zeb2 expression.

(A) Normalized sequencing tracks of ChIP–seq with anti–CTCF or anti–TCF3 in mouse HPC–7 cell line (mm9) (Calero-Nieto et al., 2014) or of placental mammal basewise conservation from PhyloP(Siepel et al., 2005). The –59–kb and −165–kb Zeb2 enhancers are shown as boxed (n = 1 biological replicate per population). (B) Normalized sequencing tracks of ATAC–seq in the indicated populations. The –59–kb and −165–kb Zeb2 enhancers are shown as boxed, the +164-kb Zeb2 enhancer is indicated with a red triangle at the bottom of the track (adapted from Immunological Genome Project Open Chromatin Regions(Yoshida et al., 2019), data shown are merged from two replicates). (C and D) BM progenitor cells (Lin⁻ CD117^hi) isolated from Cas–9 transgenic mouse expressing a pair of sgRNA targeting the –59–kb or −165–kb Zeb2 enhancers or scramble controls were cultured with Flt3L for 7 days and analyzed by flow cytometry to identify DC subsets. (C) Numbers indicate the percentage of cells in the indicated gates (representing four independent experiments, n=4 for all sgRNA pairs). (D) Statistical analysis of the frequency of pDCs from the cultures in C, small horizontal lines indicate the mean (representing four independent experiments, n=4 for all sgRNA pairs). (E) Electrophoretic mobility shift assay (EMSA) of E2–A DNA binding activity with indicated probes and nuclear extracts from WEHI cells or HEK293–T cells overexpressing GFP or Tcf3. (F) Flow cytometry analysis showing GFP-reporter activities in BM-derived cDC1s, cDC2s, and pDCs infected with empty retrovirus (black) or retroviruses expressing WT or mutant −165–kb Zeb2 enhancer as indicated. 4 mut, −165–kb Zeb2 enhancer with all 4 E boxes deleted. E1–E4 mut, 165–kb Zeb2 enhancer with E box 1-4 deleted respectively. (G) Statistical analysis of GFP MFI in indicated cell types in F. Small horizontal lines indicate the mean (representing three independent experiments, n=3 for each population) *P < 0.05; **P < 0.01, unpaired, two–tailed, Student’s t–test. Please also see Figure S1.

To test the function of these regions for DC development, we used an in vitro Fms-related tyrosine kinase ligand (Flt3L) culture system and CRISPR/Cas9 genome editing (Figures 1C and 1D). A pair of single guide RNA (sgRNA) flanking each region was introduced by retroviral vector into CD117^hi BM progenitor cells isolated from Cas9 transgenic mice (Platt et al., 2014), cultured with Flt3L for seven days and pDC output was analyzed. Targeting the −59-kb Zeb2 enhancer caused no significant changes in pDC output compared to samples with scramble guide controls, suggesting this enhancer was not required for pDC differentiation from CD117^hi progenitors (Figures 1C and 1D). However, targeting the −165-kb Zeb2 enhancer caused a significant reduction in the output of pDCs (Figures 1C and 1D), suggesting that this region may be required for Zeb2 expression in CD117^hi BM progenitor cells.

To test whether the −165-kb Zeb2 enhancer binds E-proteins, we performed electrophoretic mobility shift assay (EMSA) using oligonucleotide probes representing the sequences flanking each of the four E-box motifs (Figure 1E). We used nuclear extracts from HEK293T cells overexpressing either Tcf3 or gfp and from the WEHI murine B cell line, which expresses high amount of endogenous Tcf3. As a positive control, we used the IgH μE5 enhancer element known to bind E proteins as an electrophoretic mobility shift assay (EMSA )probe (Steininger et al., 2011). E2-A formed strong complexes with the IgH μE5 probe and with three of the four E-box probes in HEK293T nuclear extracts expressing Tcf3, but not in extracts expressing only gfp (Figure 1E). E2-A also formed complexes with three of the four E-box probes in nuclear extracts from untransformed WEHI cells, although weaker than that observed using E2-A overexpression (Figure 1E). These results suggest that the −165-kb Zeb2 enhancer can bind both endogenous and overexpressed E-proteins.

We also tested the dependency of −165-kb Zeb2 enhancer activity on E-box motifs. The native enhancer was weakly active in cDC1s, intermediately active in cDC2s, and highly active in pDCs, a pattern similar to the amount of Zeb2 mRNA in these cell lineages (Figure 1F and 1G). We next examined the effect on enhancer activity of mutations within the different E-box motifs. We found that individual deletions of E box 2, 3 and 4, reduced enhancer activity, but that deletion of E box 1 did not reduce activity (Figure 1F and 1G). This pattern was similar to the pattern of E protein binding determined by EMSA for these E-box motifs, where E box1 failed to form a complex with E2A, but E boxes 2, 3 and 4 formed strong complexes (Figure 1E). Finally, deletion of all E boxes within the enhancer regions caused a reduction in activity to the background amount of that seen in cDC1 cell, which did not express Zeb2. In summary, these results indicate that a set of E box motif contribute to the activity of the −165-kb Zeb2 enhancer.

A ChIP-seq for E2–2 performed in human CAL-1 pDC cell line showed strong E2–2 binding at a site located 183-kb upstream of human ZEB2 TSS, in a 500bp region that is highly homologues to the mouse −165-kb Zeb2 enhancer region (Figures S1A and S1D). The four E box motifs were conserved between human and mouse and present in similar locations within each enhancer (Figures S1B-D).

−165-kb Zeb2 enhancer is required for Zeb2 expression in vivo in BM.

To analyze the function of −165-kb Zeb2 enhancer in vivo, we generated mice with a germline deletion of this region by electroporating zygotes with Cas9-sgRNA complex (RNP complex) using the same sgRNA as above (Figure 1C). We generated the enhancer deletion in two genetic backgrounds (Figures 2A, S2A and S2B). First, we used C57BL6/J mice to produce the Zeb2^Δ−165 line, which was then crossed with a reporter mouse line expressing an Id2-IRES-GFP cassette (Id2^gfp) to generate the Zeb2^Δ−165Id2^gfp mouse line. We also directly targeted Zeb2^egfp mice (Nishizaki et al., 2014), harboring a ZEB2-EGFP fusion protein reporter, to produce the Zeb2 ^{Δ−165 egfp} line.

Figure 2. Deletion of the −165kb Zeb2 enhancer eliminates Zeb2 expression and increase Id2 expression in BM progenitor cells.

(A) Schematics showing the −165kb Zeb2 enhancer with four E–box motifs and sgRNA target sequences depicted (left) and strategies of generating the Zeb2 ^Δ−165, Zeb2 ^{Δ−165 egfp}, and Zeb2 ^Δ−165Id2^gfp mice lines (right). (B) Flow cytometry analyzing BM progenitor cells from mice of the indicated genotype to identify lineage negative progenitors and dendritic cell progenitors (DCP). Lineage markers include: CD3, CD105, and Ter119. Numbers indicate the percentage of cells in the indicated gates (representing four independent experiments, n=4 for WT and Zeb2 ^Δ−165 mice). (C and D) Flow cytometry analyzing BM cells from Zeb2 ^Δ−165egfp and Zeb2 ^Δ−165Id2^gfpmice for GFP expression in populations as gated in (B). Numbers indicate the geometric mean fluorescence intensity (MFI) of GFP in the indicated populations (representing four independent experiments n=4 for Zeb2^egfp and Zeb2 ^Δ−165egfp mice and n=1 for Zeb2 ^Δ−165Id2^gfpmice). (C) Statistical analysis of percentage of GFP⁺ cells in the indicated populations (middle panel), and of MFI of the indicated populations (right panel). Small horizontal lines indicate the mean (representing four independent experiments n=4 for Zeb2^egfp and Zeb2 ^{Δ−165 egfp} mice). ***P < 0.001; ****P < 0.0001, unpaired, two–tailed, Student’s t–test. Please also see Figure S2

The BM progenitor cells from these mice were analyzed to test the importance of the - 165-kb Zeb2 enhancer on expression of Zeb2 and Id2. Dendritic cell progenitors (DCPs) that have not specified to the cDC1 lineage are identified as Lin^- CD117^hi CD135⁺ CD226^- bone marrow (BM) cells (Durai et al., 2019). In Zeb2^egfp mice, 70% of DCPs expressed high amount of ZEB2-EGFP (Figures 2B and 2C), as expected (Bagadia et al., 2019a). By contrast, Zeb2 ^{Δ−16 5egfp} mice had equivalent numbers of DCPs, but failed to express ZEB2-EGFP in either DCPs or in any BM cells. Previously, conditional deletion of Zeb2 in a 1-cre system resulted in neonatal lethality related to impaired HSC mobility during embryonic development (Goossens et al., 2011). In contrast, we found that Zeb2 ^Δ−165 and Zeb2 ^{Δ−165 egfp} mice were born at expected Mendelian ratios, excluding embryonic lethality, and showed no signs of morbidity. The number of Lin^-SCA1⁺c-KIT⁺ (LSK) HSC/progenitor cells was similar between Zeb2 ^Δ−165 mice and WT mice, but LSK cells from Zeb2 ^{Δ−165 egfp} mice showed no detectable ZEB2-EGFP expression (Figures S2C-E). These results suggest that the −165-kb Zeb2 enhancer is required for Zeb2 expression in all HSC-derived lineages, but is not required for Zeb2 expression in the progenitors of HSCs during embryonic development.

−165-kb Zeb2 enhancer deletion increases ID2 expression in BM progenitor cells.

Our genetic analysis indicates that Zeb2 acts downstream of Id2 with respect to cDC1 specification, but acts upstream of Id2 with respect to Id2 expression (Bagadia et al., 2019a). These results implied that a mutually repressive circuit between Zeb2 and Id2 is in operation in the CDP whose outcome determines commitment to either the pDC or cDC1 fate. This model predicts that loss of Zeb2 expression in CDPs should increase Id2 expression.

To test this, we measured EGFP as a surrogate for ID2 expression in BM progenitors from Zeb2 ^Δ−165Id2^gfp mice (Figure 2D). In total Lin^- BM progenitors, there was a increase in EGFP expression in BM progenitors from Zeb2 ^Δ−165Id2^gfp mice compared to Id2^gfp control mice, raising the EGFP mean fluorescence intensity (MFI) from 79 to 159. Also, in the non-specified DCPs fraction, EGFP was increased from 93 MFI in Id2^gfp control mice to 309 in Zeb2 ^Δ−165Id2^gfp mice (Figure 2D), consistent with the predicted repression of Id2 by ZEB2.

−165-kb Zeb2 enhancer is required for development of pDCs, monocytes and B cells.

We next examined the impact of the −165-kb Zeb2 enhancer deletion on the development of mature hematopoietic lineages (Figure 3). Evidence suggests that Zeb2 is required for the development of pDCs (Wu et al., 2016; Scott et al., 2016) and monocytes (Wu et al., 2016), and inducible deletion of Zeb2 in adult mice with the Mx1-cre system also results in a 80% reduction in B cell number in peripheral blood (Li et al., 2016). First, we found a 67% reduction in total splenic cellularity in Zeb2 ^Δ−165 mice compared to controls (Figures S3C and S4I). pDCs and monocytes were absent, and B cells were reduced by more than 97% in Zeb2 ^Δ−165 mice compared to controls (Figures 3A-C and S3C). By contrast, cDC1s and neutrophils were increased as a percentage of splenocytes in Zeb2 ^Δ−165 mice, but their absolute numbers were relatively unchanged (Figures 3A, 3C and S3C). Red pulp macrophages (RPMs) were present in normal numbers in Zeb2 ^Δ−165 mice (Figures 3C and S3C). T cells and NK cells were increased in Zeb2^Δ−165 mice as a percentage of splenocytes (Figure 3B), but not in absolute numbers (Figure S3C).

Figure 3. Zeb2 −165kb enhancer is required for pDC, B cell and monocyte development.

(A–C), Flow cytometry analysis of (A, left panel) pDCs and cDC1s, (B, left panel) B cells, T cells, and NK cells, and (C, left panel) neutrophils (neutro), monocytes (MO) and red pulp macrophages (RPM) in spleen of mice of indicated genotypes. Statistical analysis of these lineages as percentage of total splenocytes (A-C, right panels). Small horizontal lines indicate the mean (representing three independent experiments, n=4–8 for each lineage analyzed). (D) Normalized sequencing tracks of ATAC–seq in the WT and Zeb2 ^Δ−165 splenic B cells and normalized sequencing tracks of ChIP–seq with anti–TCF3 in mouse HPC–7 cell line (mm10) (Calero-Nieto et al., 2014). The +161–kb, +164–kb, and −165–kb Zeb2 enhancers are shown as boxed. (n = 2–3 biological replicates per population). N.S. not significant; *P < 0.05; ****P < 0.0001, unpaired, two–tailed, Student’s t–test. Please also see Figure S3 and S4.

Next, we examined development of B cells. We found a 97% reduction in splenic B cells as a percentage of spenocytes and a reduction of more than 99% in absolute numbers compared to WT mice (Figures 3B and S3C). This severe reduction contrasts with a report of only a 80% reduction described for Mx1-cre⁺ Zeb2^fl/fl mice (Li et al., 2016). This more severe reduction could be a results of the earlier action of −165-kb Zeb2 enhancer compared with induced deletion caused by polyI:C treatment in Mx1-cre mice, in which B cell progenitors may reach later stages of development. We also used ATAC-seq to determine the global accessibility of enhancers in both WT B cells and in the small number of B cells that remained in Zeb2 ^Δ−165 mice (Figure 3D). Deletion of the Zeb2 ^Δ−165 enhancer did not significantly alter the overall accessibility of enhancers in the Zeb2 locus in splenic B cells (Figure 3D). However, one enhancer located 164-kb downstream of the Zeb2 TSS was accessible in Zeb2 ^Δ−165 but not in WT B cells (Figure 3D, bottom left panel). This result suggests that the +164-kb may represent a site for an enhancer that may weakly compensate for loss of the −165-kb enhancer in some settings. The few B cells developed in Zeb2 ^Δ−165 mice are similar to B cells from WT mice by transcriptional profiling (Figure S3D), with less than 100 genes differing in expression more than 4-fold between Zeb2 ^Δ−165 and WT mice. Zeb2^Δ−165 B cells express higher amount of Id2 mRNA (Figure S3E), in agreement with results from Id2^gfp reporter mice (Figure 2D).

Since Id2 is required for the development of innate lymphoid cells (ILCs) (Verykokakis et al., 2014; Yokota et al., 1999), we quantified NK cells and all three ILC subsets in the small intestine (SI) lamina propria in Zeb2 ^Δ−165 mice (Figures S3A and S3B). There were no significant changes in NK cells or ILC3 cells and less than 2-fold increases in ILC1 and ILC2 cells. In summary, the −165-kb Zeb2 enhancer is strictly required for the development of pDCs, monocytes and B cells, but not for several other hematopoietic lineages.

−165-kb Zeb2 enhancer is required for early lineage specification.

To determine the stage of pDC development impacted by Zeb2^Δ−165 enhancer deletion, we analyzed pDC progenitors in BM (Figures 4A and 4E). A lymphoid pDC progenitor was recently identified as a Ly6D⁺ SiglecH⁺ double positive (DP) fraction of Lin^-CD127⁺ CD135⁺ lymphoid progenitors (LPs) (Rodrigues et al., 2018) (Figure 4A). This pDC progenitor comprised 10% of LPs in WT mice, but was completely absent in Zeb2^Δ−165 mice (Figures 4A and 4E). pDCs can also develop from common dendritic cell progenitor (CDP) but no clonogenic pDC progenitor has been defined within the CDP (Bagadia et al., 2019b). However, cDC1 specification in the CDP (identified as Lin^- CD117^int CD117⁺ CD135⁺ CD226⁺ cells (Durai et al., 2019)) was increased by 3-fold in Zeb2 ^Δ−165 mice compared with WT control mice (Figure 4B), suggesting an increase in alternative fate specification caused by the loss of Zeb2 expression. These results suggest that deletion of the Zeb2−165kb enhancer blocks pDC development at its earliest known stage.

Figure 4. Zeb2 −165kb enhancer is required for BM pDC and B cell progenitors.

(A-D), Flow cytometry of BM cells from WT and Zeb2 ^Δ−165 mice to identify (A) double positive lymphoid progenitors (DP LPs), (B) pre–cDC1, (C) Hardy fraction A–C, or (D) ILC–2 progenitor (ILC2P). Lineage markers include: CD3, CD105, and Ter119. Numbers indicate the percentage of cells in the indicated gates (representing three independent experiments, n=3 for WT mice, n=6 for Zeb2 ^Δ−165 mice). (E and F) Statistical analysis (E) of populations analyzed in (A) and (B), or (F) of population analyzed in (C) and (D) as percentage of Lin⁻ BM cells (representing three independent experiments, n=3 for WT mice, n=6 for Zeb2 ^Δ−165 mice). **P < 0.01; ***P < 0.001; ****P < 0.0001, unpaired, two–tailed, Student’s t–test. Please also see Figure S3 and S4.

Mx1-cre induced deletion of Zeb2 was reported to arrest B cell development at Hardy fraction A (Fr. A) (Li et al., 2016). In Zeb2 ^Δ−165 mice, Zeb2 expression was lost beginning in HSCs (Figures S2C-E), and we found a significant reduction in all Hardy fractions (Figures 4C and 4F). This result suggests that Zeb2 ^Δ−165 mice have a defect in B cell specification from the common lymphoid progenitor (CLP). Notably, the ILC2 progenitors (ILC2P) that arise from the CLP (Hoyler et al., 2012; Stier et al., 2018) were also increased in Zeb2 ^Δ−165 mice (Figure 4D), suggesting that the CLP is present but lacks the capacity for B cell specification. Since Id2 is required for development of NK cells (Yokota et al., 1999; Delconte et al., 2016) and ILCs (Klose et al., 2014; Xu et al., 2019), the elevated Id2 expression in Zeb2 ^Δ−165 mice may deviate progenitors away from B cell specification in favor of ILC development at the CLP stage.

Id2 reportedly inhibits expression of Csf1r (Iavarone et al., 2004), which is required for monocyte development (Dai et al., 2002). In agreement, Zeb2 ^Δ−165 mice lacked expression of Csf1r on Lin^- CD135⁺ BM progenitors (Figure 4A), consistent with an effect of increased ID2 expression (Figure 2C). Since Csf1r expression is required to identify the monocyte-dendritic cell progenitor (MDP) (Fogg et al., 2006) and the common monocyte progenitor (cMoP) (Hettinger et al., 2013), we were unable to directly evaluate these two progenitor stages in Zeb2^Δ−165 mice (Figure 4A). However, the development of cDC1 in Zeb2^Δ−165 mice (Figures 3A and S3C) suggests that the MDP does develop, since the MDP gives rise to the CDP in which the cDC1 lineage is specified.

−165-kb Zeb2 enhancer is required for normal lymphocyte function.

To test whether the few B cells developing in Zeb2 ^Δ−165 mice are functionally normal, we first tested their ability to undergo antibody class switch recombination (CSR). We found that Zeb2 ^Δ−165 B cells had severely impaired switching to IgG1compared to WT B cells induced by LPS and IL-4 treatment (Figure S4A and S4B). As a functional test, we examined the of Zeb2^Δ−165 mice to survive infection by West Nile Virus, which relies on B cells for protective antibodies (Diamond et al., 2003). Consistent with defective B cell responses, Zeb2 ^Δ−165 mice uniformly succumbed by 10 days after West Nile infection, while 80% of WT mice survived beyond 22 days (Figure S4C). Zeb2 ^Δ−165 B cells also formed fewer and smaller follicles in the spleen than WT ones (Figure S4I).

Zeb2 has also been reported to be required for T cell and NK cell function (Omilusik et al., 2015; Dominguez et al., 2015; van Helden et al., 2015). At steady state, we found that NK cell maturation was reduced by 84% in Zeb2 ^Δ−165 spleen (Figure S4D and S4E), similar to results of conditional Zeb2 deletion in NK cells using Zeb2^fl/fl Ncr1^icre mice (van Helden et al., 2015). We also examined T cell responses in the context of infection by the lymphocytic choriomeningitis virus Armstrong strain (LCMV-Arm), which was previously examined in the context of T-cell specific Zeb2 deletion using Gzmb-cre and Cd4-cre (Omilusik et al., 2015; Dominguez et al., 2015). In agreement with these reports, we found significantly reduced numbers of LCMV-specific KLRG1^hi CD127^lo terminal effector CD8⁺ T cells in Zeb2 ^Δ−165 mice 8 days after LCMV-Arm infection compared to WT mice (Figure S4F-S4H). Together, these results indicate that the −165-kb Zeb2 enhancer is required for Zeb2 expression in a range of immune lineages, and that loss of this enhancer recapitulates the previously reported functional defect generated by lineage-specific Cre-mediated Zeb2 inactivation.

RTM populations are present in Zeb2 ^Δ−165 mice.

Since splenic RTMs were present in Zeb2 ^Δ−165 mice, we asked whether RTMs were present in other tissues (Figure 5). We found that Zeb2 ^Δ−165 mice had normal populations of lung alveolar and interstitial MFs, liver Kupffer cells (Figures 5A and 5B), and peritoneal cavity macrophages (PCMs) (Figures 5C and 5D). Consistently, Zeb2 mRNA was expressed normally in RPM of Zeb2 ^Δ−165 mice compared with control WT mice (Figure S4J). In contrast, Zeb2 mRNA was undetectable in Zeb2 ^Δ−165 BM DCPs (Figure S4J), consistent with the lack of EGFP expression by DCPs in Zeb2 ^{Δ−165 egfp} reporter mice (Figure 2A). In splenic neutrophils and BM pre-DC1 population, where Zeb2 is expressed at a low amount in WT mice, there was more than 80% reduction in Zeb2 mRNA concentration in Zeb2 ^Δ−165 mice compared with WT mice (Figure S4J). In summary, Zeb2 mRNA expression is maintained in RTMs, but not in HSC-derived lineages, in Zeb2 ^Δ−165 mice.

Figure 5. RTM development is intact in Zeb2 −165kb enhancer deletion mice.

(A) Flow cytometry of the indicated tissues from WT and Zeb2 ^Δ−165 mice to identify MF populations (representing four independent experiments, cells on the lower panel were pre-gated on CD64 and F4/80 double positive, n=4 for WT and Zeb2 ^Δ−165 mice). (B) Statistical analysis of populations in (A) as percentage of CD45⁺ cells. (C) Flow cytometry analyzing peritoneal cavity macrophage (PCM) populations as a percentage of CD45⁺ cells from WT and Zeb2 ^Δ−165 mice with indicated treatments (representing three independent experiments, n=3–7 for WT and Zeb2 ^Δ−165 mice). (D) Statistical analysis of populations in (C) as percentage of CD45⁺ cells. (E) Flow cytometry of the indicated tissues at the indicated time from WT and Zeb2 ^Δ−165 embryos to identify MF and monocyte (MO) populations (representing three independent experiments, n=3–5 for WT and Zeb2 ^Δ−165 mice). (F) Statistical analysis of populations in (E) as percentage of CD45⁺ cells. (G) Flow cytometry analyzing ZEB2 amount by EGFP expression in fetal liver monocytes (MO) from Zeb2^egfp and Zeb2 ^{Δ−165 egfp} mice. Numbers indicate the geometric mean fluorescence intensity (MFI) of GFP in the indicated populations (representing three independent experiments n=3 for Zeb2^egfp and Zeb2 ^{Δ−165 egfp} mice). (H) Statistical analysis of MFI of fetal liver monocytes in G. Small horizontal lines indicate the mean (representing three independent experiments n=3 for Zeb2^egfp and Zeb2 ^{Δ−165 egfp} fetuses). N.S. not significant; **P < 0.01, unpaired, two–tailed, Student’s t–test. Please also see Figure S6.

Recent work has shown that RTM populations in adult mice are largely of embryonic origin (Liu et al., 2019; Sawai et al., 2016; Ginhoux and Guilliams, 2016; Hoeffel et al., 2015). For example, lineage tracing using Ms4a3-cre suggests that less than 40% of RPMs are derived from monocytes from the adult HSC (Liu et al., 2019). To determine the origin of RTMs in Zeb2 ^Δ−165 mice, we used a model of sterile peritonitis in which intraperitoneal (i.p) administration of thioglycolate induces the acute loss of peritoneal cavity macrophages (PCM) (Zhang et al., 2019; Gautier et al., 2013; Liu et al., 2019). After 72 hours, recently formed PCM are generated from HSC-derived monocytes that migrate into the peritoneal cavity and differentiate into MFs (Liu et al., 2019). We found that at steady state, untreated Zeb2 ^Δ−165 mice and WT mice had equivalent numbers of PCM. Further, thioglycolate treatment i.p. caused complete depletion of PCM in 12 hours in both WT and Zeb2 ^Δ−165 mice. However, after 72 hours, PCM were regenerated in WT mice, but failed to regenerate in Zeb2 ^Δ−165 mice (Figures 5C and 5D). These results indicated that steady state RTMs in Zeb2 ^Δ−165 mice are embryonically derived, and that Zeb2 expression in embryonically derived RTM is not dependent on the −165-kb Zeb2 enhancer.

Embryonic MF development is intact in Zeb2 ^Δ−165 mice.

During transient definitive hematopoiesis, EMPs generate both YS MFs and FL monocytes in two phases (Ginhoux and Guilliams, 2016). EMPs in the first phase, labeled by Runx1^cre/EYFP lineage tracing at E7.5, differentiate into YS MFs. A late phase of EMPs, labeled by Runx1^cre/EYFP at E8.5, can differentiate locally, but can also enter blood circulation and migrate to the FL to differentiate into monocytes (Hoeffel et al., 2015).

To determine if embryonic development of MFs is affected in Zeb2 ^Δ−165 mice, we examined Zeb2 ^Δ−165 and WT embryos at E10.5 for YS MFs, and at E15.5 for the presence and numbers of FL monocytes and MFs. We found no significant change in the development of YS MFs at E10.5 or in development of FL MFs at E15.5 between Zeb2 ^Δ−165 and WT embryos (Figure 5E and 5F). FL monocytes were also present at E15.5 in Zeb2 ^Δ−165 embryos, although slightly reduced in numbers compared to WT embryos (Figure 5E and 5F). The FL monocytes from Zeb2^Δ−165 embryos also expressed Zeb2 at amount similar to WT embryos (Figure 5G and 5H). Together, these results indicate that Zeb2 expression supporting embryonic macrophage and monocyte development during transient definitive hematopoiesis is independent of the −165-kb Zeb2 enhancer for Zeb2 expression.

Embryonic MF precursors may activate an alternative Zeb2 enhancer.

Since Zeb2 expression during embryonic macrophage development does not require the - 165-kb Zeb2 enhancer, we asked if alternative Zeb2 enhancers could function during embryonic hematopoiesis. One potential candidate is locate at +164-kb in the Zeb2 locus that showed increased ATAC-seq accessibility in the B cells from Zeb2 ^Δ−165 mice (Figure 3D). To test this, we carried out single-cell ATAC-seq (Satpathy et al., 2019) on yolk sac macrophages (YS MFs) isolated from E10.5 WT embryos, on fetal liver (FL) monocytes isolated from E15.5 WT embryos, and adult splenic macrophages and monocytes isolated from WT mice (Figures 6 and S5). High quality scATAC-seq profiles were recovered from 1,484 YS MFs, 6,179 FL monocytes, 8,751 splenic monocytes, and 6,562 splenic macrophages, with a median of 10,554 fragments per cell and a median of 14.3 for enrichment of Tn5 transposon insertions at transcription start sites (Satpathy et al., 2019; Granja et al., 2020). Clustering identified four major cell clusters representing each cell type, and two minor clusters comprised of mixed cell types (Table S1 and Figure S5). The clusters were defined by 37,311 marker peaks and 9,756 marker gene scores that were consistent with their cellular identities (Figure S5). Cluster 1 represented adult red pulp macrophages (RPMs), cluster 2 represented YS MFs, cluster 3 represented FL monocytes, and cluster 4 represented splenic monocytes.

Figure 6. Embryonic MF precursors engage a +164kb Zeb2 enhancer to regulate Zeb2 expression.

Normalized pseudo-bulk ATAC-seq tracks in the indicated lineages from the indicated clusters, bulk ATAC-seq of EMP(Wu et al., 2020) and ChIP-seq tracks of indicated histone marks or transcription factor in indicated lineages. The +161–kb and +164–kb (dashed line) and −165–kb (solid red line) Zeb2 enhancers are shown as boxed (representing one biological replicate for ATAC-seq and 2–3 replicates for ChIP-seq). Please also see Figure S5 and S6

We next compared the ATAC-seq profile of the four major clusters, focusing on peaks in the Zeb2 locus (Figure 6). An ATAC-seq peak was present in the −165-kb region in splenic monocytes, as expected, which was also present in each of the other cell types. All other ATAC-seq peaks present in these other three cell types were also present in adult monocytes, with one exception. As expected, the +164-kb Zeb2 peaks and a nearby peak located at +161-kb were present in RPMs, YS MFs, and FL monocytes, but were barely detectable in splenic monocytes. RPM in un-manipulated adult mice are predominantly of fetal origin (Hoeffel et al., 2015; Schulz et al., 2012), suggesting that the +164-kb and +161-kb peaks could function as enhancers sufficient to activate Zeb2 expression in macrophage progenitors during embryonic hematopoiesis, independently of the −165-kb enhancer. In agreement, ATAC-seq analysis of erythro-myeloid progenitors (EMPs) (Wu et al., 2020), which generate YS MFs and FL monocytes, showed significant accessibility at both the +164-kb and +161-kb regions (Figure 6). In addition, ChIP-seq analysis showed the +164-kb region to have heavy H3K27 histone acetylation in embryonically derived Kupffer cells, but not in adult BM-derived adult monocytes, (Lavin et al., 2014) (Figure 6). Together, these results indicate that +164kb region may function as an embryonic-specific enhancer for Zeb2.

To test the functional activity of the +164-kb Zeb2 region, we examined its reporter activity directly in yolk sac macrophages and adult BM-derived monocytes (Figure S6A). First, when examined in YS macrophages, the +164-kb Zeb2 region had higher reporter compared to the −165-kb enhancer region. Second, the +164-kb Zeb2 region was more active in YS macrophages compared with its activity in adult BM-derived monocytes (Figure S6A). Finally, the +161-kb Zeb2 region showed little activity in either population. Since there are distinct TF motifs used by the different cell types (Figure S5C), the +164-kb region may conceivably bind TFs that are active during embryonic hematopoiesis, but are not active in adult HSCs.

In particular two HOMER predicted NUR77 motifs are present in the +164-kb enhancer but not the +161kb or the −165kb enhancers (Figure S3F), and notably, the transcription factor NUR77 (encoded by Nr4a1) is highly expressed in YS MFs and FL monocytes but not adult BM-derived monocytes(van de et al., 2016) (Figure S6B), and ChIP-seq analysis of NUR77 in RAW macrophage cell line showed strong binding at the +164-kb Zeb2 enhancer (Koenis et al., 2018) (Figure 6). Consistently, NUR77 deficient (Nr4a1^–/–) embryos showed significantly reduced FL monocytes and macrophage production (Figure S6C). Taken together these results suggest a potential role of NUR77 in supporting Zeb2 expression at the +164-kb Zeb2 enhancers in fetal macrophage progenitors.

Another TF of the same family that is preferentially expressed in fetal macrophage progenitors is LXRα (encoded by Nr1h3) (Mass et al., 2016). ChIP-seq analysis of LXRα in the macrophage-like RAW cell line indicated strong binding at the +164-kb Zeb2 enhancer(Sakai et al., 2019) (Figure 6). The transcriptional profile of Nr1h3^–/– Kupffer cells resembles that of Zeb2 ^–/–Kupffer cells, suggesting that LXRα might also regulate Zeb2 expression at the +164-kb enhancer(Scott et al., 2018). In this way, adult hematopoiesis would be uniquely dependent the - 165-kb enhancer, while embryonic expression of Zeb2 is driven by embryonic specific factors acting at the +164-kb enhancer. Testing this concept will require the specific mutations of the +164-kb region and assaying its role in embryonic and adult MF development.

−165-kb Zeb2 enhancer is required for structure integrity of the Zeb2 locus

To evaluate the impact of −165-kb Zeb2 enhancer deletion on 3D chromatin structure, we performed two sets of Hi-C analysis. The in situ Hi-C analysis of WT and Zeb2 ^Δ−165 splenic CD8⁺ T cells showed no significant global changes on the structure of chromosome 2 (Figure S7B). We performed HYbrid Capture Hi-C (Hi-C²) on WT and Zeb2 ^Δ−165 RPMs as well as WT splenic pDCs to examine the Zeb2 locus specifically by probing an 810-kb region around Zeb2. In both WT pDCs and WT RPMs, we observed similar interaction profiles around Zeb2 gene, including a strong interaction between the Zeb2 promoter region and the −165-kb enhancer region (Figure 7). We observed that the −165-kb Zeb2 enhancer was located at the distal end of an extended region of heavy interactions with the Zeb2 promoter (Figure 7 and S7A). Notably, deletion of 400 bp within the −165-kb enhancer reduced interactions with the promoter across this entire region (Figures 7 and S7A). The Zeb2 promoter also showed an interaction with the +164-kb Zeb2 enhancer region, both in WT RPMs and in Zeb2 ^Δ−165 RPMs, which was somewhat weaker in WT pDCs. Of note, this interaction between the +164-kb Zeb2 and the promoterwas not affected by the deletion of the −165-kb Zeb2 enhancer (Figures 7 and S7A). This result might suggest a compensatory activity of the +164-kb Zeb2 enhancer in place of the −165-kb enhancer. Together, these data indicate that the −165-kb Zeb2 enhancer can form a loop-like structure with the Zeb2 promoter that participates in the structural integrity of a functional Zeb2 locus in adult hematopoietic lineages.

Figure 7. 165-kb Zeb2 enhancer is required for the structural integrity of Zeb2 locus.

− Normalized Hi-C² interaction maps, virtual 4C tracks from the viewpoint of Zeb2 promoter in the indicated lineages, and RPM ATAC-seq track from Figure 6(representing one biological replicate). The +161–kb, +164–kb and −165–kb Zeb2 enhancers are shown as boxed in the tracks. The regions of interactions with Zeb2 promoter are shown as boxed in the interaction map (dashed line). Track heights for virtual 4C tracks were normalized based on the signal intensity of Zeb2 promoter (representing three biological replicates for WT pDC and two biological replicates for WT and Zeb2 ^Δ−165 RPMs). Please also see Figure S7

Discussion

This study was prompted by our attempt to explain how ID2 suppresses Zeb2 expression during cDC1 development(Bagadia et al., 2019a; Bagadia et al., 2019b). cDC1 specification in the CDP is initiated by a transient pulse of Nfil3 expression that first suppresses Zeb2, which de-represses Id2 and leads to stable loss of Zeb2 even in the absence of Nfil3(Bagadia et al., 2019a). Since ID2 does not bind DNA, we wondered whether the stable loss of ZEB2 was due to the squelching of positive regulation at the Zeb2 locus mediated by E-proteins. E2A and E2–2 are both expressed in the CDP and are inhibited by ID2. Therefore we carried out a functional in vitro test for several candidate E-protein binding sites in the Zeb2 locus (Calero-Nieto et al., 2014) for their role in DC development using in vitro CRISPR/Cas9 targeting of BM progenitors. This approach identified the −165-kb region as possibly supporting pDC development, which we confirmed in vivo by generation and analysis of mice harboring a germline deletion in this region. This region contains a cluster of 4 E-box motifs, which we confirmed as binding to E2A in vitro using EMSA. This Zeb2 enhancer is necessary for development of pDCs, B cells, and monocytes, and monocyte-derived macrophages, but not for embryonically derived macrophages or other hematopoietic lineages.

A role for ZEB2 in pDC, B cell and monocyte development has been previously shown using deletion of its coding regions, but previous studies did not address the mechanism controlling Zeb2 expression in these three distinct lineages. E2A is the predominant E protein in B cells, while E2–2 is predominant in pDCs, suggesting that both of these lineages could maintain Zeb2 expression via the −165-kb enhancer using these factors. However, neither of these E proteins is highly expressed in monocytes, which instead express PU.1. Conceivably, monocyte progenitors such as the MDP use an E protein to activate Zeb2 expression via the - 165-kb enhancer, which may be maintained subsequently by another factor such as PU.1. In agreement, E2A is highly and uniformly expressed in the MDP (Durai et al., 2019). Alternately, PU.1 expressed in the MDP could initiate the activity of the −165-kb enhancer, since there is a consensus motif (GAGGGA) sequence capable of binding PU.1 that partially overlaps the first E-box motif. Subsequent Zeb2 expression could then be maintained by either E2A or E2–2 after pDC and B cell specification. Testing these possibilities will require the generation of additional germline mutations within the −165-kb enhancer.

The present study also investigated the molecular mechanisms governing the expression of Zeb2 throughout hematopoiesis and highlights the differential usage of enhancers of the same gene by lineages arising from different stages of hematopoiesis. A recent study, assaying Irf8 enhancers during cDC1 development, has discovered how different enhancers of the same gene can be required at different stages during the development of a single lineage (Durai et al., 2019). Although the development of cDC1 requires both the +41-kb and +32-kb Irf8 enhancer, it has been demonstrated that the +41-kb Irf8 enhancer is required for the specification of CDPs to pre-cDC1s, while the +32-kb Irf8 enhancer is essential for commitment of pre-cDC1s to mature cDC1s. Our study expands that discovery by examining the usage of Zeb2 enhancers throughout hematopoiesis. While all HSC-derived lineages that emerged during adult hematopoiesis are dependent on −165-kb Zeb2 enhancer for Zeb2 expression, RTMs and their embryonic precursors express normal amount of Zeb2 despite the absence of this enhancer. Instead, these cells, arising from embryonic hematopoiesis might rely on a +164-kb Zeb2 enhancer to maintain their ZEB2 amount. These stage-specific activities of enhancers to differentially regulate the expression of a single gene could also exist for other genes important during hematopoiesis. The analysis of such activities could reveal molecular mechanisms that regulate early hematopoiesis that are not yet fully understood.

RTMs from embryonic and BM origin have been shown to have different gene expression profiles (Lavin et al., 2014; Bruttger et al., 2015), reflecting their differences in functions and self-maintenance. Other reports have demonstrated that monocyte-derived MFs can acquire almost identical transcriptional signatures in tissues as embryonically derived MFs, arguing that tissue-specific cues rather than ontogenies are more dominant in imprinting MF transcriptional programs (Gibbings et al., 2015; Scott et al., 2018). The Zeb2 ^Δ−165 mouse provides a unique model to interrogate these two models since only embryonically derived MFs are present. These mice allow for the examination of functions of MFs derived from different sources by comparing the responses of these mice with control mice containing normal BM-derived MFs to various infection models where MFs play a key role.

Finally, the results of the present study also identify +164-kb Zeb2 enhancer as a cis-element potentially important for early embryonic hematopoiesis. Vav1-cre induced deletion of Zeb2 in mice results in neonatal lethality that does not occur in Zeb2 ^Δ−165 mice. This indicates that there are alternative enhancer elements that likely control the expression of Zeb2 in other tissues such as neurons and melanocytes, and during early hematopoiesis.

Limitations of study

This study identifies an enhancer required for Zeb2 expression in adult hematopoietic lineages, but does not exclude the possibility that other enhancers may also be required. Our screen for enhancer candidates was based on ChIP-seq for E2A followed by a CRISPR/Cas9 targeting of in vitro pDC development from early BM progenitors. Conceivably, the ChIP-seq may have missed some enhancer candidates, and the sgRNA guides used to test the candidates that were identified could have produced false negative results in vitro. For example, the −59-kb enhancer bound E2A in HPC7 cells, but in vitro targeting had no impact on pDC development, unlike the in vitro targeting of the −165-kb enhancer. This may have been a result of poor targeting based on sgRNA guide failure. Nonetheless, it may be of value to carry out a future in vivo deletion of the −59-kb to evaluate this possibility. We also identified enhancers that appear to be active only during embryonic hematopoiesis, based on scATAC-seq. Testing whether these enhancers are required for embryonic macrophage development will require not only that they be targeted in vivo, but may also require that they be targeted in the setting of an inactivated −165-kb enhancer, which is also active in the embryonic setting, so potentially providing redundant support in the embryo. Finally, we have shown that E proteins, such as E2A and E2–2, are likely candidates for factors supporting the −165-kb enhancer in B cells and pDCs. However, we still need to test which factors support this enhancer in the myeloid lineages, such as monocytes, that also require Zeb2 expression.

STAR METHODS

LEAD CONTACT

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Kenneth Murphy (kmurphy@wustl.edu).

MATERIALS AVAILABILITY

All unique/stable reagents generated in this study are available from the Lead Contact without restriction.

DATA AND CODE AVAILABILITY

B cells microarrays are available on the GEO database with the following accession number: GSE167458.This data was utilized in Figure 3. The sc-ATAC-seq and bulk ATAC-seq data is available on the GEO database with the following accession number: GSE167330. This data is utilized in Figure 3, Figure 6, and Figure S5. The Hi-C and Hi-C² data is available on the GEO database with the following accession number: GSE171103 and GSE171564. This data is utilized in Figure 7 and Figure S7. Following data sets were downloaded and reanalyzed: Microarray data sets for YS macrophages, FL monocytes and BM monocytes (GSE76999), ChIP-seq data sets for mouse E2-A (GSE48086), H3K27Ac (GSE63339), NUR77 (GSE102394), LXRα (GSE128662), for human E2–2 (GSE43876), ATAC-seq data set for EMP (GSE144243) and Immunological Genome Project (ImmGen) data sets for ATAC-seq (GSE100738).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice

WT C57BL6/J mice and R26^Cas9/+ mice (B6N.129(Cg)-Gt(ROSA)26Sor^{tm1.1(CAG-cas9*,-EGFP)Fezh}/J) were obtained from the Jackson Laboratory. ZEB2-EGFP fusion protein reporter (STOCK Zfhxlb^tm2.1Yhi) mice were derived from biological material provided by the RIKEN BioResource Center through the National BioResource Project of the Ministry of Education, Culture, Sports, Science and Technology, Japan. Id2–IRES–GFP mice(Jackson et al., 2011) were generously donated by G. Belz.

All mice were generated, bred, and maintained in the Washington University in St. Louis School of Medicine specific pathogen-free animal facility. Animals were housed in individually ventilated cages covered with autoclaved bedding and provided with nesting material for environmental enrichment. Up to five mice were housed per cage. Cages were changed once a week, and irradiated food and water in autoclaved bottles were provided ad libitum. Animal manipulation was performed using standard protective procedures, including filtered air exchange systems, chlorine-based disinfection, and personnel protective equipment including gloves, gowns, shoe covers, face masks, and head caps. All animal studies followed institutional guidelines with protocols approved by the Animal Studies Committee at Washington University in St. Louis.

Unless otherwise specified, experiments were performed with mice between 6 and 10 weeks of age. No differences were observed between male and female mice in any assays performed and so mice of both genders were used interchangeably throughout this study. Within individual experiments, mice used were age- and sex-matched whenever possible.

Generation of Zeb2 enhancer deletion mice

Enhancer deletion mice were generated as described (Durai et al., 2019). sgRNAs flanking the - 165-kb Zeb2 enhancer were identified using CHOPCHOP (http://chopchop.cbu.uib.no/), and the following sgRNA sequences were used: Zeb2 −165 5’: GAGTGAGAGATCATCAAATG and Zeb2 +165 3’: GATAACGTTCTTGAAGCATA. sgRNA with the desired sequences were synthesized and conjugated with purified Cas9 protein to form the RNP complex by the Department of Pathology Micro-Injection Core at Washington University in St Louis. Day 0.5 single cell zygotes from C57Bl/6 mice were isolated and underwent electroporation at the Department of Pathology Micro-Injection Core at Washington University in St Louis. Around 60 single cell zygotes were electroporated with 8μM of RNP complex using 1mm gap cuvette (BioRad). Electroporated zygotes were then transferred into the oviducts of pseudopregnant recipient mice.

The resulting pups were screened by PCR with the following primers to identify those that had successful deletion of the enhancers of interest: 165mutant-Screening-5’: CTGCAGCAGGTTGACAAAGA; 165mutant-Screening-3’: CCTGAAGTGTACGCTCACCA. Mice with the desired deletion were then outcrossed to wildtype (WT) C57BL6/J mice, and the resulting heterozygous mice were intercrossed to generate homozygous enhancer deletion mice.

Antibodies and flow cytometry.

Cells were kept at 4ºC while being stained in PBS supplemented with 0.5% BSA and 2mM EDTA in the presence of antibody blocking CD16/32 (2.4G2; BD).

The following antibodies were from BD: Brilliant Ultraviolet 395–anti-CD117 (2B8), PE-CF594–anti-CD135 (A2F10.1), V500–anti-MHC-II (M5/114.15.2), Alexa Fluor 700 or allophycocyanin–anti-Ly6C (AL-21), Brilliant Violet 421–anti-CD127 (SB/199), biotin or FITC–anti-CD19 (1D3), Brilliant Violet 421–anti-CCR9 (CW-1.2), allophycocyanin –anti-CD43 (S7), PE mouse anti-mouse CD249 (Lγ−51) (BP-1), PE–anti-NK1.1 (PK136), PE–anti-Ly6G (1A8). The following antibodies were from eBioscience: PE-Cy7–anti-TCRβ (H57–597), allophycocyanin–eFluor 780–anti-CD11c (N418), violetFluor 450–anti-MHC Class II (I-A/I-E) (M5/114.15.2), eFluor 450–anti-CD11b (M1/70), allophycocyanin–anti-CD317 (eBio927), peridinin chlorophyll protein (PerCP)–eFluor 710–anti-CD172a (P84), PerCP-Cy5.5–anti-SiglecH (eBio-440c), PE–anti-IL-25R (MUNC33), APC-eFluor 780-anti-F4/80 (BM8), biotin–anti-CD105 (MJ/718), PE-Cy7–anti-CD25 (PC61.5).The following antibodies were from BioLegend: PE–anti-Ly6D (49-H4), PE-Cy7–anti-CD24 (M1/69), Brilliant Violet 711–anti-CD115 (AFS98), PE or Brilliant Violet 421–anti-XCR1 (ZET), PE–anti-CD45.2 (104), PE/Dazzle 594–anti-CD45R (RA3–6B2), biotin–anti-Ly6G (1A8), biotin–anti-Ter119 (TER-119), PE–anti-CD226 (10E5), Brilliant Violet 605–anti-CD64 (X54–5/7.1), biotin or APC–anti-CD3e (145–2c11). Cells were analyzed on a FACSCanto II or FACSAria Fusion flow cytometer (BD), and data were analyzed with FlowJo v10 software (TreeStar).

BM progenitor analysis

Bone marrow (BM) was harvested as described (Grajales-Reyes et al., 2015). For FACS analysis, BM cells was isolated and depleted of CD3-, CD105-, Ter119- and in the instance of HSC analysis CD19- and Ly6G-expressing cells by staining with corresponding biotinylated antibodies followed by depletion with MagniSort Streptavidin Negative Selection Beads (Thermo Fisher). All remaining BM cells were then stained with fluorescent antibodies prior to sorting.

Tissue harvest and population analysis

Tissues were harvested as described (Durai et al., 2019). Briefly, spleens were minced and digested in 5 ml of Iscove’s modified Dulbecco’s media (IMDM) + 10% FCS (cIMDM) with 250 μg/ml of collagenase B (Roche) and 30 U/ml of DNaseI (Sigma-Aldrich) for 45 min at 37 °C with stirring. Lungs were minced and digested in 5 ml of cIMDM with 4 mg/ml of collagenase D (Roche) and 30 U/ml of DNaseI (Sigma-Aldrich) for 1.5 h at 37 °C with stirring. Livers were minced and digested in 5 ml of cIMDM with 125μg/ml of Liberase TL (Roche) and 30 U/ml of DNaseI (Sigma-Aldrich) for 1h at 37 °C with stirring. After digestion was complete, single-cell suspensions from all organs were passed through 70-μm strainers and red blood cells were lysed with ammonium chloride–potassium bicarbonate (ACK) lysis buffer. Cells were subsequently counted with a Vi-CELL analyzer (Beckman Coulter) and 3–5 × 10⁶ cells were used per antibody staining reaction.

For peritoneal cell analysis, 5 ml of MACS buffer (Dulbecco’s phosphatebuffered saline + 0.5% BSA +2 mM ethylenediaminetetraacetic acid (EDTA)) was injected into the peritoneum of mice using a 27-g needle. After injection, the mice were shaken gently to dislodge peritoneal cells. A 25-g needle was then used to collect the peritoneal fluid. Cells were lysed with ACK buffer and counted as described above.

For small intestine analysis, small intestines were collected and flushed with HBSS to remove the fecal contents. Peyer’s patches were removed, intestines were openned lengthwise, washed and cut into small pieces. Cut pieces were subjected to gentle agitation for 20 mins in a solution containing HBSS, HEPES, BCS and EDTA and then vortexed. This agitation step was repeated once more. Tissues pieces were then rinsed in HBSS before being digested with collagenase type IV in complete RPMI medium for 40–45 minutes at 37°C with gentle shaking. Digests were filtered, centrifuged, and subjected to density gradient centrifugation using 40% and 70% percoll solutions. Cells from the interface were recovered, washed and used for analysis.

Yolk sac and fetal liver harvest and population analysis

Yolk sacs were harvested from embryos at E10.5 and digested in 1ml of MACS buffer with 250 μg/ml of collagenase B (Roche) and 30 U/ml of DNaseI (Sigma-Aldrich) for 30 min at 37 °C with stirring. Fetal liver was harvested from embryos at E15.5 and mechanically dissociated by passing through a 20-g needle three times. The single cell suspensions were then filtered and cells were lysed with ACK buffer as described above. For YS macrophage culture, the YS macrophages were sorted and then cultured with 25ng/ml MCSF and the culture was analyzed after four days.

METHOD DETAILS

Thioglycollate induced sterile peritonitis.

The sterile peritonitis was induced as described (Liu et al., 2019). Briefly, 1 mL of 4% sterile thioglycollate broth (BD Biosciences) was injected i.p. into 8-week-old mice of the indicated genotypes and then analyzed at the indicated time points.

Expression microarray analysis.

RNA from WT and Zeb2 ^Δ−165 B cells was extracted with a NucleoSpin RNA XS Kit (Machery-Nagel), amplified with WT Pico System (Affymetrix) and hybridized to GeneChip Mouse Gene 1.0 ST microarrays (Affymetrix) for 18 h at 45 °C in a GeneChip Hybridization Oven 640. The data was analyzed with the Affymetrix GeneChip Command Console. Microarray expression data was processed using Command Console (Affymetrix, Inc) and the raw (.CEL) files generated were analyzed using Expression Console software with Affymetrix default RMA Gene analysis settings (Affymetrix, Inc). Probe summarization (Robust Multichip Analysis, RMA), quality control analysis, and probe annotation were performed according to recommended guidelines (Expression Console Software, Affymetrix, Inc.). Data were normalized by robust multiarray average summarization and underwent quartile normalization with ArrayStar software (DNASTAR).

Quantitative RT-PCR (qPCR)

RNA from indicated populations from WT and Zeb2 ^Δ−165was extracted with a NucleoSpin RNA XS Kit (Machery-Nagel). cDNA was generated using SuperScript IV One-Step RT-PCR system (Invitrogen) and Oligo(dT)20 Primer (Invitrogen). qPCR was performed using SYBR Green-based detection and the following previously published primers(Scott et al., 2016): Zeb2-qPCR-F: GGCAAGGCCTTCAAGTACAA; Zeb2-qPCR-R: AAGCGTTTCTTGCAGTTTGG; GAPDH-qPCR-F: TGCCCCCATGTTTGTGATG; GAPDH-qPCR-R: TGTGGTCATGAGCCCTTCC. The amount of Zeb2 mRNA detected from each sample was normalized to that of GAPDH mRNA.

Retroviral infection and culture

Retroviruses were produced by transfecting retroviral vectors into Plat-E cells essentially as described (Bagadia et al., 2019a) and collecting viral supernatants 2 days later. For in vitro CRISPR/Cas9 deletion, a Thy1.1-hU6-gRNA-BbsI stuffer RV vector and a GFP-hU6-gRNA-BbsI stuffer RV vector were used (Theisen et al., 2018). The following primers containing the sgRNA sequence and BbsI overhangs were annealed and cloned into the BbsI digested vector, all 5’ oligonucleotides were cloned into the Thy1.1 vector and all 3’ oligonucleotides were cloned into the GFP vector. For the Zeb2 −165-kb 5’ sgRNA: CACCGgagtgagagatcatcaaatg and AAACcatttgatgatctctcactcC. For the Zeb2 −165-kb 3’ sgRNA: CACCGgataacgttcttgaagcata and AAACtatgcttcaagaacgttatcC. For the Zeb2 −59-kb 5’ sgRNA: CACCGtcaatgtgcaaataccacga and AAACtcgtggtatttgcacattgaC. For Zeb2 −59-kb 3’ sgRNA: CACCGgccatttcctgcactcagga and AAACtcctgagtgcaggaaatggcC. Lin^— CD117^hi BM cells from R26^Cas9/+ mice were then sorted and transduced with viral supernatants by spin infection at 1800 RPM for 1 hour in the presence of 2 ug/mL polybrene. Infected cells were then cultured in Flt3L for 8 days before DCs were analyzed by flow cytometry. For in vitro reporter assay, a Thy1.1 pA GFP CMVp_min PmeI MSCV vector was used as control, enhancer elements were cloned into this vector to test their activity in vitro. Lin^— CD117^hi BM cells and YS macrophages from WT mice were sort purified and and transduced with viral supernatants by spin infection at 1800 RPM for 1 hour in the presence of 2 ug/mL polybrene. Infected cells were then cultured in Flt3L for 8 days or MCSF for 4 days before GFP activities were analyzed by flow cytometry.

Electrophoretic mobility shift assay (EMSA).

EMSAs were performed as described (Iwata et al., 2017). Oligonucleotide pairs were annealed to generate probes that were labeled with ³²P-dCTP using Klenow polymerase. HEK293FT cells were transiently transfected with retroviral vectors for Tcf4 or GFP using TransIT-LT1. After 48 h, cells were lysed with buffer A (10 mM HEPES-KOH, pH 7.9, 1.5 mM MgCl₂ and 10 mM KCl) containing 0.2% NP40 and protease inhibitors. Nuclei were pelleted, resuspended in buffer C (20 mM HEPES-KOH, pH 7.9, 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA and 25% glycerol), and centrifuged to obtain nuclear extracts (Dignam et al., 1983). EMSA was performed using nuclear extracts from HEK293T cells transfected with Tcf4 or GFP (up to 1.5 μg total) or Wehi cells, 0.25 μg poly dI-dC (Sigma) and ³²P-labeled probes in 10 μl binding reactions for 20 min at 4 °C. Reactions were separated on 4–7%T 3.3%C polyacrylamide mini-gels in 0.4× TBE for 50 min at 250 V and 4 °C and were analyzed by autoradiography.

B cell class switching recombination assay.

Splenic B cells were sort purified from and cultured at the density of 500,000 cells/ml in 96 well plate with either LPS (10μg/ml, Sigma-Aldrich) alone, or LPS (10μg/ml) and IL-4 (25ng/ml, Peprotech). Cells were collected on day 4, and surface IgG1 expression was analyzed by FACS.

Analysis of E-box motifs in human −165-kb Zeb2 enhancer.

The occurrence of E-box motifs in the element −165-kb relative to the Zeb2 transcription start site was found with the FIMO motif-identification program (Bailey et al., 2009) at a P-value threshold of 1 × 10⁻² with the E-box position weight matrix obtained from Homer software packge(Heinz et al., 2010). Human and mouse elements were aligned via Clustal Omega W.

Bulk ATAC-seq

Bulk atac-seq samples were processed according to the previously described Omni-ATAC protocol(Corces et al., 2017). Briefly, cells were thawed, live/dead stained, and sorted for viability. Sorted cells were lysed and then nuclei were transposed for 30 minutes. For samples with less than 50,000 live cells recovered, the amount of Tn5 enzyme was scaled down according to the cell number, but the transposition reaction volume was maintained at 50 uL. Transposed fragments were then amplified and prepared for sequencing. Samples were quantified by bioanalyzer and 2×75 paired end sequencing was performed using an Illumina Nextseq.

Viral infections

For LCMV infection, mice were infected with 2 × 10⁵ plaque-forming units (PFU) of LCMVArm strain via the intraperitoneal route. Splenic T cells were harvested on day 8 after infection, and analyzed by FACS. For WNV infection, were inoculated subcutaneously with 100 PFU of WNV by footpad injection after anesthetization with xylazine and ketamine. The survival of these animals was monitored over a 22 day time period.

Computational analysis for ChIP-seq and ATAC-Seq

ChIP-seq and ATAC-seq datasets were aligned and mapped on the mouse reference genome (GRCm38/mm10) using Bowtie software (version 1.1.1) with the following command:–sam–best -p 4 -m 1 mm10–chunkmbs 1000. Making tag directories was performed by HOMER (version 4.9) software. Tag directories for ChIP-seq and ATAC-seq were generated by ‘makeTagDirectory’ and duplicated reads in each dataset were discarded using the argument -tbp 1. Browser Extensible Data (BED) files with genomic locations for each specific gene set were generated using ‘table browser’ tool in the UCSC genome browser web site. The transcriptional regions (gene body) were extended by 50 kb upstream and downstream using ‘slopBed’ in BedTools2 software. ChIP-seq or ATAC-seq peaks merged with genomic regions for specific gene sets were extracted by ‘intersectBed’. Bedgraph files for visualizing as UCSC genome browser tracks were created by ‘makeUCSCfile’ with following arguments: -fsize 5e7 -res 1

Single cell ATAC-seq.

For library preparation, cells were thawed and then transposed and processed following the 10X Genomics single cell ATAC kit and protocol as previously described (Satpathy et al., 2019). Libraries were sequenced on an Illumina HiSeq 4000 at the Stanford Functional Genomics Facility. The generated reads were aligned to the mm10 reference genome assembly using the 10X cellranger-atac count pipeline, yielding aligned, de-duplicated ATAC-seq fragments for each sample and emulsion barcode. R package ÀrchR` was used for all downstream analysis (Granja et al., 2020). Fragments files were loaded into ArrowFiles and barcodes containing at least 1000 fragments and having an enrichment of Tn5 insertions in transcription start sites of at least 4 (relative to +/− 2000 bp from each TSS) were called as cells.

Data processing steps were performed using the default ArchR workflow (Granja et al., 2020). Briefly, dimensionality reduction and clustering were performed on the tile matrix. To capture cell type specific peaks, peaks were called on each cluster individually and merged into a union peak set. After determining the union peak set, peaks were annotated by transcription factor motifs.

For track visualization and heat maps, group coverages (pseudobulk) were created for each sample, each cluster, and each sample within each cluster. These coverages were normalized such that each coverage had equal fragments in TSS and visualized using the ArchR Browser(Granja et al., 2020). Marker peaks and marker GeneScores for each cluster were computed using `getMarkerFeatures` with cutoffs FDR <= 0.01 and Log2FC >= 0.5. All features passing these cutoffs are shown in the respective heatmaps. Motifs enriched within the marker peaks for each cluster were computed using `peakAnnoEnrichment` with cutoffs FDR <= 0.1 and Log2FC >= 0.1. The top 8 motifs passing these thresholds for each cluster were visualized in a heatmap.

Hi-C² probe design

To design probes targeting the Zeb2 region for HYbrid Capture Hi-C (Hi-C²), we followed a similar approach as published(Sanborn et al., 2015). In short, we (i) identified all MboI restriction sites within the target region, (ii) designed our bait probe sequences to target sequences within a certain distance of the MboI restriction sites as Hi-C ligation junctions occur between them, and (iii) followed a similar three-pass probe design strategy sequentially relaxing various probe-quality filtering parameters including the distance of the probe from the MboI restriction site, the number of repetitive bases, the GC content, and increasing probe density in gaps of probe coverage. We then removed overlapping probes or probes with identical sequences. After all three passes, we identified 1328 unique probes covering the Zeb2 region (chr2:44.71–45.52 Mb; 1.64 probes/kb). A fifteen-base pair primer sequence (details in Table S2) was then appended to both ends of the 120-bp probe sequence to facilitate single oligo pool synthesis and subsequent amplification of this region-specific sub-pool. Probe construction and hybrid selection was then followed with sequences specific to this study using the published strategy(Sanborn et al., 2015).

Hi-C and Hi-C² experiments

The Hi-C datasets were generated using a protocol derived from the in situ Hi-C protocol standardized by the 4DN consortia. In brief, we developed and optimized a low-input HiC protocol to process low cell numbers sorted freshly from mouse spleen. Freshly sorted cells were kept on ice and typically diluted to 100K-1M cells/ml in media (min volume of 1ml). The cells were immediately crosslinked with 2% formaldehyde for 10 min, quenching with 0.2M glycine for 5 min at room temperature and 5 min on ice, followed by permeabilizing them with nuclei intact, digesting the DNA with MboI (4-cutter restriction enzyme), filling the 5′-overhangs while incorporating biotin-14-dATP (a biotinylated nucleotide), followed by ligating the resulting blunt-end fragments, shearing the DNA to a 400–700-bp fragment size using a BioRuptor Pico in a “30s ON 30s OFF” setting for 8 cycles, capturing the biotinylated ligation junctions with streptavidin beads, building an Illumina library followed by 10–12 rounds of PCR amplification, and finally analyzing the resulting fragments with paired-end sequencing. The resulting library was always shallow sequenced to 300K–1.5M reads to check for library build quality looking at key statistics such as complexity, number of Hi-C contacts, inter vs. intrachromosomal interactions, and long-range v/s short-range intrachromosomal interactions. Libraries that passed the quality check can be either sequenced deeper and/or used as pools for subsequent Hi-C² experiments.

We generated 14 in situ Hi-C libraries (Table S3): 6 from pDC-WT, 4 from WT-Mac and 4 from KO-Mac cells, shallow sequencing on average to 1M reads per Hi-C library. Subsequently, we also generated 14 in situ Hi-C² libraries from each of the Hi-C libraries on which hybrid selection was performed. All in situ Hi-C libraries generated as part of this study are detailed in Table S4. All the Hi-C data was processed using the computational pipeline described before(Rao et al., 2014). Hi-C libraries were sequenced to a depth of between 376K and 1.46M reads (on average, 971K reads). Individual Hi-C² libraries were sequenced to a depth of between 5M and 12.2M reads (on average, 7.2M reads). All data was initially processed using the pipeline published before(Rao et al., 2014), and visualized on the desktop and web version of Juicebox. All Hi-C contact maps from WT-pDC, WT macrophage (WT RPM) and Zeb2 ^Δ−165 macrophage (KO RPM) were combined respectively, using juicer’s mega.sh, as these are in essence “biological” replicates, to generate higher resolution megamaps. All Hi-C² contact matrices were also combined this way and normalized using VC_SQRT method. These sample-type specific megamaps consisted of contact maps between 30.7M and 37.4M reads deep. All scripts and virtual 4C analyses were performed similar to previously published work(Choudhary et al., 2020).

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis for single cell ATAC-sequencing data is described above. Horizontal lines in figures indicate the mean. Results from independent experiments were pooled as indicated in figure legends. Data were analyzed using Prism (GraphPad), using unpaired two-tailed Student’s t-tests when comparing two groups. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
BUV395 rat anti-mouse CD117 (clone: 2B8)	BD Biosciences	Cat#: 564011
PE-CF594 rat anti mouse-Flt3 (clone: A2F10.1)	BD Biosciences	Cat#: 562537
PE mouse anti-rat CD90/mouse CD90.1 (clone: OX-7)	BD Biosciences	Cat#: 554898
V500 rat anti-mouse I-A/I-E (clone: M5/114.15.2)	BD Biosciences	Cat#: 562366
Alexa Fluor® 700 rat anti-mouse Ly6C (clone: AL-21)	BD Biosciences	Cat#: 561237
Brilliant Violet 421 rat anti-mouse CD127 (clone:A7R34)	BD Biosciences	Cat#: 135027
Biotin or FITC rat anti-mouse CD19 (clone: 1D3)	BD Biosciences	Cat#: 553784 Cat#: 553785
BV421 mouse anti-mouse CCR9 (clone: CW-1.2)	BD Biosciences	Cat#: 565412
APC rat anti-mouse CD43 (clone: S7)	BD Biosciences	Cat#: 560663
PE mouse anti-mouse CD249 (Lγ−51) (clone: BP-1)	BD Biosciences	Cat#: 553735
PE mouse anti-mouse NK-1.1 (clone: PK136)	BD Biosciences	Cat#: 557391
PE rat anti-mouse Ly-6G (clone: 1A8)	BD Biosciences	Cat#: 551461
PE rat anti-mouse Ly-6A/E (clone: D7)	BD Biosciences	Cat#: 553108
PE rat anti-mouse IgG1 (A85-1)	BD Biosciences	Cat#: 550083
eFluor 450 rat anti-mouse CD11b (M1/70),	eBioscience	Cat#: 48-0112-82
APC anti-mouse CD317 (clone: eBio927)	eBioscience	Cat#: 17-3172-82
PerCP-eFluor 710 anti-mouse CD172a (clone: P84)	eBioscience	Cat#: 46-1721-82
PerCP-eFluor 710 anti-mouse SiglecH (clone: eBio440c)	eBioscience	Cat#: 46-0333-82
PE anti-mouse IL-25R (clone: MUNC33)	eBioscience	Cat#: 50-112-2389
APC-eFluor 780 anti-mouse F4/80 (clone: BM8)	eBioscience	Cat#: 47-4801-82
Biotin-anti-mouse CD105 (clone: MJ7/18)	eBioscience	Cat#: 13-1051-82
V450 anti-mouse I-A/I-E (clone: M5/114.15.2)	eBioscience	Cat#: 48-5321-82
APC-eFluor 780 anti-mouse CD11c (clone: N418)	eBioscience	Cat#: 47-0114-80
PE-Cy7 anti-mouse TCRβ (clone: H57-597),	eBioscience	Cat#: 25-5961-82
PE-Cy7 anti-mouse CD25 (clone: PC61.5)	eBioscience	Cat# 25-0251-82
PerCP-eFluor 710 anti-mouse Klrg1 (clone: 2F1)	eBioscience	Cat# 46-5893-82
PE/Cyanine7 anti-mouse CD24 (clone: M1/69)	BioLegend	Cat#: 101822
PE anti-mouse Ly-6D (clone: 49-H4)	BioLegend	Cat#: 138604
Brilliant Violet 711 anti-mouse CD115 (clone: AFS98)	BioLegend	Cat#: 135515
PE or Brilliant Violet 421 anti-mouse XCR1 (clone: ZET)	BioLegend	Cat#: 148204 Cat#: 148216
PE anti-mouse CD45.2 (clone: 104)	BioLegend	Cat#: 109807
PE/Dazzle 594 anti-mouse/human CD45R/B220 (clone: RA3-6B2)	BioLegend	Cat#: 103258
Biotin anti-mouse TER-119 (clone: TER-119)	BioLegend	Cat#: 116204
Biotin anti-mouse Ly6G (clone: 1A8)	BioLegend	Cat#: 127604
PE anti-mouse CD226 (clone: 10E5)	BioLegend	Cat#: 128805
Brilliant Violet 605 anti-mouse CD64 (clone: X54-5/7.1)	BioLegend	Cat#: 139323
Biotin or APC anti-mouse CD3ε (clone: 145-2C11)	BioLegend	Cat#: 100311
Chemicals, Peptides, and Recombinant Proteins
TransIT-LTI	MIRUS Bio	Cat#: MIR 2300
Iscove’s Modified Dulbecco’s Medium (IMDM)	GIBCO	Cat#: 12440–046
Opti-MEM Reduced Serum Medium	GIBCO	Cat#: 31985–070
Hexadimethrine bromide (Polybrene)	Sigma-Aldrich	Cat#: H9268
Fetal Bovine Serum (Characterized)	HyClone	Cat#: SH30071.03
Sodium pyruvate	Corning	Cat#: 25-000-CI
L-Glutamine	Gibco	Cat#: 25030–164
Pen Strep (Penicillin Streptomycin)	Gibco	Cat#: 15140–122
2-Mercaptoethanol	Sigma-Aldrich	Cat#: M3148
MEM Non-essential Amino Acid Solution (100X)	Sigma-Aldrich	Cat#: M7145
Dnase I	Sigma-Aldrich	Cat#: D4527
Collagenase B	Sigma-Aldrich	Cat#: COLLB-RO
Thioglycollate Medium	Sigma-Aldrich	Cat#: T-9032
Aprotinin bovine recombinant	Sigma-Aldrich	Cat#: A6103
Leupeptin hydrochloride microbial, R 90% (HPLC)	Sigma-Aldrich	Cat#: L9783
HEPES	Sigma-Aldrich	Cat#: H3375
Lipopolysaccharide from Escherichia coli (055:B5)	Sigma-Aldrich	Cat#: L2880
Recombinant murine IL-4	PeproTech	Cat#: 214–14
Recombinant murine M-CSF	PeproTech	Cat#: 315–02
PE conjugated LCMV GP33-44 tetramer	MBL international
Critical Commercial Assays
NucleoSpin RNA XS Kit	Machery-Nagel	Cat#: 740902.50
WT-Pico Kit	Affymetrix
Deposited Data
Microarray of WT and Zeb2 Δ−165 B cells	This study	GSE167458
Microarray for YS macrophages, FL monocytes and BM monocytes	Van de Laar et al., 2016	GSE76999
Sc-ATAC-seq and ATAC-seq of WT and Zeb2 Δ−165 B cells	This study	GSE167330
ATAC-seq	ImmGen	GSE100738
ATAC-seq (mouse EMP)	Wu et al., 2020	GSE144243
ChIP-seq (mouse E2-A)	Calero-Nieto et al., 2014	GSE48086
ChIP-seq (human E2-2)	Ghosh et al., 2014	GSE43876
ChIP-seq (mouse H3K27Ac)	Lavin et al., 2014	GSE63339
ChIP-seq (mouse NUR77)	Koenis et al., 2018	GSE102394
ChIP-seq (mouse LXRα)	Sakai et al., 2019	GSE128662
Capture Hi-C and in situ Hi-C	This study	GSE171564 and GSE171103
Experimental Models: Organisms/Strains
Mouse: B6.129S4-Gt(ROSA)26Sortm1(FLP1)Dym/JRainJ	Jackson Laboratory	JAX: 009086
Mouse: Zfhxlbtm2.1Yhi	Riken
Mouse: C57BL/6	Jackson Laboratory	JAX: 000664
Mouse: Zeb2 Δ−165	This study
Mouse: Zeb2 Δ−165 egfp	This study
Mouse: Id2-IRES-GFP	G. Belz
Oligonucleotides
Quantitative RT-PCR primers for Zeb2 and GAPDH	This study	Method Details
sgRNA targeting enhancers for Zeb2	This study	Method Details
Probes for EMSA	This study	Method Details
Genotyping primers for Zeb2 Δ−165 and Zeb2 Δ−165 egfp mice	This study	Method Details
Software and Algorithms
GraphPad Prism 7.0	GraphPad	https://www.graphpad.com/scientific-software/prism/
FlowJo v10	Tree Star	https://www.flowjo.com/solutions/flowjo
FACSDiva	BD Biosciences	https://www.bdbiosciences.com/en-us/instruments/research-instruments/research-software/flowcytometryacquisition/facsdiva-software
ArrayStar 15	DNASTAR	https://www.dnastar.com/
ArchR	Granja et al., 2020	www.ArchRProject.com

Highlights:

• Zeb2 −165-kb enhancer is required for the expression of ZEB2 in HSC derived lineages.

• pDC, monocyte, and B cell development requires Zeb2 −165-kb enhancer.

• Tissue resident macrophages (RTMs) in Zeb2 ^Δ−165 mice are entirely of embryonic origin.

• Embryonic expression of ZEB2 may rely on alternative +164-kb Zeb2 enhancer.