Comparison with Ezh2 reveals the PRC2-dependent functions of Jarid2 in hematopoietic stem Cell lineage commitment

Summary

Previous studies showed the polycomb repressive complex 2 (PRC2) co-factor Jarid2 represses self-renewal transcriptional networks in mouse multipotent progenitor cells (MPPs). But only a fraction of de-repressed HSC-specific genes were associated with loss of H3K27me3, implying Jarid2 may have non-canonical (PRC2-independent) functions in hematopoiesis. We sought to delineate these potential PRC2-independent functions by comparing stem and progenitor cells genetically deficient for either Jarid2 or Ezh2 (enzymatic component of PRC2). Loss of Ezh2 increased myeloid differentiation but with a defect in lymphopoiesis. In contrast, loss of Jarid2 enhanced multi-lineage differentiation proportionally. Single-cell transcriptomics showed loss of Jarid2 had minimal impact across progenitor populations, but loss of Ezh2 led to accumulation of lymphoid-biased MPP4 cells and B cell progenitors in the bone marrow. Functional assays confirmed a differentiation block at the pre-pro-B cell stage. These data suggest the major PRC2-dependent function of Jarid2 in hematopoietic progenitors is restriction of myeloid differentiation potential.

Highlights

• •Loss of Ezh2 enhances myeloid differentiation with a defect in lymphopoiesis
• •Loss of Jarid2 enhances multi-lineage hematopoietic differentiation
• •Restriction of myeloid lineage commitment by Jarid2 is PRC2-dependent
• •Loss of Ezh2 leads to accumulation of lymphoid-biased MPPs and developmental arrest of B cell progenitors

Bjeije and colleagues compare the molecular and functional properties of hematopoietic stem and progenitor cells genetically deficient for Ezh2 (PRC2 enzymatic component) or Jarid2 (PRC2.2 co-factor) to define the PRC2-dependent functions of Jarid2 in hematopoietic lineage commitment.

Introduction

Hematopoietic stem cells (HSCs) maintain blood homeostasis through the balance of self-renewal to sustain the HSC pool and differentiation that gives rise to all mature blood cell lineages (Orkin and Zon, 2008; Wilson and Trumpp, 2006). This delicate balance is governed by networks of transcriptional and epigenetic regulators (Buenrostro et al., 2018; Lara-Astiaso et al., 2014). Among the most critical epigenetic regulators is the polycomb repressive complex 2 (PRC2), a histone methyltransferase complex that catalyzes trimethylation of histone H3 on lysine 27 (H3K27me3), a modification associated with transcriptional repression. The enzymatic subunit of PRC2 is enhancer of zeste homolog 2 (Ezh2), which catalyzes the methylation reaction (Margueron and Reinberg, 2011; Simon and Kingston, 2009). Previous studies have demonstrated that Ezh2 is essential for maintenance of HSC identity and function. Conditional deletion of Ezh2 in the hematopoietic system results in loss of HSC quiescence, premature differentiation, and stem cell exhaustion (Mochizuki-Kashio et al., 2011; Xie et al., 2014). The importance of PRC2 in regulating hematopoiesis is underscored by the prevalence of EZH2 mutations in a range of blood cancers (Morin et al., 2010; Nikoloski et al., 2010; Score et al., 2012).

PRC2 does not intrinsically possess genome binding selectivity. Locus specificity is provided by the binding of different accessory co-factors that recruit this complex to distinct genomic sites in different cell types (Kasinath et al., 2021; Pasini et al., 2010; Sarma et al., 2008). PRC2 is divided into two subcomplexes defined by the co-factor binding: PRC2.1 (PHF1, MTF2, and PHF19) and PRC2.2 (JARID2 and AEBP2) (Glancy et al., 2023; Healy et al., 2019; Loh et al., 2021). These variants modulate PRC2 activity, chromatin targeting, and context-specific gene regulation during development and stem cell fate decisions (Kloet et al., 2016; Peng et al., 2009). Our previous studies highlighted a critical role for the PRC2.2 accessory protein Jarid2 in regulation of hematopoiesis. Jarid2 is essential for repression of self-renewal transcriptional networks in mouse multipotent progenitor (MPP) cells. Conditional deletion of Jarid2 results in de-repression of HSC-specific genes in MPPs, which conveys ectopic self-renewal potential (Celik et al., 2018). But molecular analysis revealed that only about half of the de-repressed HSC-specific genes were associated with loss of H3K27me3 in Jarid2-null MPPs (Celik et al., 2018). This implies Jarid2 may potentially have non-canonical (PRC2-independent) functions in hematopoiesis.

Here, we sought to define specific PRC2-dependent functions of Jarid2 in blood regeneration by comparing the functional and molecular properties of stem and progenitor cells genetically deficient for either Jarid2 or Ezh2. Loss of Ezh2 leads to myeloid skewing and peripheral lymphoid deficiency in competitive transplantation assays. In contrast, loss of Jarid2 enhanced engraftment in all peripheral blood lineages. Single-cell transcriptomics showed that while loss of Jarid2 had minimal impact across early progenitor populations, loss of Ezh2 led to accumulation of lymphoid-biased MPP4 cells in the bone marrow (BM). CITE-seq analysis further revealed an increase in early B cell progenitors in Ezh2-deficient BM, with functional assays showing a differentiation block at the pre-pro-B cell stage. Cumulatively, our data show that the main molecular function of Jarid2 in relation to PRC2 co-factor activity is restricting myeloid differentiation in hematopoietic progenitor cells. Moreover, the contrasting differences between Ezh2 and Jarid2 loss-of-function may imply Jarid2 has non-canonical activities that regulate HSC fate and lineage specification.

Results

Genetic deletion of Jarid2 and Ezh2 leads to reduction of phenotypically defined HSCs

To compare the consequences of Jarid2 and Ezh2 loss of function for hematopoiesis, we crossed Jarid2^fl/fl and Ezh2^fl/fl mice to the Vav-Cre driver to delete these alleles in all developing blood cells. However, we were unable to obtain any viable Vav-Cre:Ezh2^fl/fl pups, suggesting Ezh2 is critical for establishing hematopoiesis. To circumvent this, we generated inducible conditional knockout mouse models by crossing Jarid2^fl/fl and Ezh2^fl/fl mice to the Mx1-Cre strain. Recombination of floxed alleles (=“Δ”) was induced in eight-week-old mice by injection of polyinosinic:polycytidylic acid (pIpC). Successful gene knockout was confirmed by PCR (Figure 1A) and was highly efficient in hematopoietic stem cells (HSCs; Figure 1B). Immunophenotypic analysis (Figure 1C) revealed no major differences in the abundance of most hematopoietic stem and progenitor cell (HSPC) populations, but both Jarid2^Δ/Δ and Ezh2^Δ/Δ mice showed reduced numbers of phenotypically defined (Lineage− Sca-1+ c-Kit+ Flk2− CD48− CD150+) HSCs (Figure 1D), in line with previous studies (Celik et al., 2018; Yang et al., 2016). These genetic mouse models provided powerful tools to compare the functional roles of Ezh2 and Jarid2 in hematopoiesis.

Figure 1.

Genetic deletion of Jarid2 and Ezh2 leads to reduction of phenotypically defined HSCs

(A) Genotyping PCR confirming successful excision of Jarid2 (exon 4) and Ezh2 (exons 16 and 17) in single-HSC-derived colonies.

(B) Percentage of recombined single-HSC-derived colonies per 96-well plate. Two biological replicates per three independent experiments.

(C) Representative flow cytometry gating strategy to identify indicated hematopoietic stem and progenitor cell populations.

(D) Absolute numbers of indicated HSPC populations from control (n = 18), Jarid2^Δ/Δ (n = 17), and Ezh2^Δ/Δ (n = 16) mice 8 weeks post-pIpC. Data are mean ± SEM. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. One-way ANOVA with Tukey correction for multiple comparisons.

Loss of Ezh2 does not convey long-term reconstitution to MPPs unlike loss of Jarid2

A defining functional property resulting from conditional deletion of Jarid2 in hematopoietic cells is the ability of MPPs to engraft long-term in serial repopulation assays. To determine if this phenotype was related to PRC2, 100 MPPs (Lineage− Sca-1+ c-Kit+ Flk2− CD48− CD150−) were isolated from Mx1-Cre control, Jarid2^Δ/Δ, and Ezh2^Δ/Δ mice (CD45.2) and competitively transplanted with 2.5 × 10⁵ congenic (CD45.1) BM cells into lethally irradiated mice. Peripheral blood analysis revealed reduced overall chimerism from Ezh2^Δ/Δ MPPs (Figures 2A and S1A). However, multi-lineage analysis identified this was due to a specific deficit in peripheral lymphoid output of Ezh2^Δ/Δ MPPs as myeloid differentiation was enhanced (Figure 2B). Transplantation of Jarid2^Δ/Δ MPPs produced the highest proportion of recipient mice with long-term multi-lineage reconstitution (LTMR), defined as >1% donor-derived engraftment in myeloid, B cell, and T cell blood populations (Figure 2C). Despite overall lower peripheral blood engraftment, Ezh2^Δ/Δ MPPs showed increased BM engraftment (Figure 2D), which is composed mostly of mature myeloid cells (Figures S1B–S1D). Analysis 18 weeks post-transplant demonstrated increased numbers of donor-derived Jarid2^Δ/Δ and Ezh2^Δ/Δ MPPs in the BM of recipient mice (Figure 2E). To determine if these Ezh2^Δ/Δ MPPs possessed self-renewal ability, 3.0 × 10⁶ BM cells were transferred from primary recipients to secondary irradiated mice. Jarid2^Δ/Δ MPPs were the only genotype to show substantial peripheral blood engraftment in secondary recipients (Figure 2F), with chimerism in all major blood lineages (Figures 2G and S1E) and LTMR (Figure 2H). While Ezh2^Δ/Δ MPPs showed some BM engraftment (Figures 2I and S1F–S1H) and regeneration of donor-derived MPPs (Figure 2J), the engraftment was variable and myeloid-biased. Thus, the majority of Ezh2^Δ/Δ recipients could not be classified as LTMR (Figure 2H). The contrasting outputs of Jarid2^Δ/Δ and Ezh2^Δ/Δ MPPs suggest Jarid2 may restrain self-renewal and lymphoid output of MPPs in a PRC2-independent fashion.

Figure 2.

Loss of Ezh2 does not convey long-term reconstitution to MPPs unlike loss of Jarid2

(A) Percentage of donor-derived (CD45.2) peripheral blood (PB) engraftment from transplantation of 100 control (n = 8), Jarid2^Δ/Δ (n = 7), or Ezh2^Δ/Δ (n = 8) MPPs in primary recipients.

(B) Donor-derived chimerism in myeloid, B cell, and T cell PB lineages 16 weeks post-primary transplantation.

(C) Percentage of primary recipient mice with long-term multi-lineage reconstitution (LTMR).

(D) Donor-derived bone marrow (BM) engraftment 18 weeks post-primary transplantation.

(E) Donor-derived MPP cell count from BM of primary recipient mice 18 weeks post-transplantation.

(F) Percentage of donor-derived PB engraftment from secondary transplantation of primary BM from control (n = 8), Jarid2^Δ/Δ (n = 8), or Ezh2^Δ/Δ (n = 8) MPP-transplanted primary recipients.

(G) Donor-derived chimerism in myeloid, B cell, and T cell PB lineages 16-week post-secondary transplantation.

(H) Percentage of secondary recipient mice with LTMR.

(I) Donor-derived BM engraftment 18 weeks post-secondary transplantation.

(J) Donor-derived MPP cell count from BM of secondary recipient mice 18 weeks post-transplantation. Data are mean ± SEM. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. One-way ANOVA with Tukey correction for multiple comparisons. See also Figure S1.

Jarid2 regulates PRC2-dependent restriction of myeloid potential in hematopoietic progenitors

To compare the functional effects of loss of Jarid2 versus loss of Ezh2 in an unbiased fashion, competitive transplantation of unfractionated whole BM was performed by transplanting 5.0 × 10⁵ BM cells from pIpC-treated Mx1-Cre control, Jarid2^Δ/Δ, and Ezh2^Δ/Δ mice with 5.0 × 10⁵ BM cells from congenic mice. Loss of Jarid2 enhanced peripheral blood engraftment (Figure 3A) and chimerism in all major blood lineages (Figure 3B). In contrast, Ezh2 deficiency increased myeloid differentiation without significantly altering lymphoid output compared to control BM (Figure 3B). Both Jarid2^Δ/Δ and Ezh2^Δ/Δ genotypes showed significantly enhanced BM chimerism relative to control cells at 18 weeks post-transplant (Figures 3C and S2A–S2C). While loss of Ezh2 promoted expansion of the collective HSPCs (Lineage− Sca-1+ c-Kit+; Figure 3D), Ezh2^Δ/Δ cells showed reduced donor-derived HSC (Lineage− Sca-1+ c-Kit+ CD48− CD150+) regeneration in the BM of recipient mice (Figure 3E), similar to Jarid2^Δ/Δ cells.

Figure 3.

Jarid2 regulates PRC2-dependent restriction of myeloid potential in hematopoietic psrogenitors

(A) Percentage of donor-derived (CD45.2) peripheral blood (PB) engraftment from competitive transplantation of WBM from control (n = 14), Jarid2^Δ/Δ (n = 16), or Ezh2^Δ/Δ (n = 19) mice in primary recipients.

(B) Donor-derived chimerism in myeloid, B cell, and T cell PB lineages 16-weeks post-primary transplantation.

(C) Donor-derived bone marrow (BM) engraftment 18 weeks post-primary transplantation.

(D) Donor-derived HSPC (CD45.2+ Lineage− Sca-1+ c-Kit+) cell count from BM of primary recipient mice 18 weeks post-transplantation.

(E) Donor-derived HSC (CD45.2+ Lineage− Sca-1+ c-Kit+ CD48− CD150+) cell count from BM of primary recipient mice 18 weeks post-transplantation.

(F) Percentage of donor-derived PB engraftment from secondary transplantation of primary BM from control (n = 7), Jarid2^Δ/Δ (n = 8), or Ezh2^Δ/Δ (n = 10) WBM-transplanted primary recipients.

(G) Donor-derived chimerism in myeloid, B cell, and T cell PB lineages 16-weeks post-secondary transplantation.

(H) Donor-derived BM engraftment 18 weeks post-secondary transplantation.

(I) Donor-derived HSPC count from BM of secondary recipient mice 18 weeks post-transplantation.

(J) Donor-derived HSC cell count from BM of secondary recipient mice 18 weeks post-transplantation. Data are mean ± SEM. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. One-way ANOVA with Tukey correction for multiple comparisons. See also Figure S2.

To verify the observed phenotypes were heritable, secondary transplantation was performed by transferring 3.0 × 10⁶ unfractionated BM cells from individual primary recipients into unique secondary recipient mice. While overall peripheral blood engraftment was not significantly different (Figure 3F), lineage bias was dramatically affected by loss of Ezh2. Differentiation of Ezh2^Δ/Δ cells was strongly biased toward myelopoiesis with a severe deficiency in peripheral lymphoid output (Figure 3G). In contrast, loss of Jarid2 significantly enhanced T cell potential (Figure 3G). Due to the myeloid bias, Ezh2^Δ/Δ cells were dominant in the overall BM (Figures 3H and S2D–S2F). HSPC expansion (Figure 3I) and HSC depletion (Figure 3J) of Ezh2^Δ/Δ cells was observed analogous to primary transplantation. These results confirm the observed phenotypes are programmed at the level of long-term HSCs. Collectively, these data imply the major PRC2-related function of Jarid2 in hematopoiesis is restriction of myeloid differentiation potential in HSPCs, as both Jarid2^Δ/Δ and Ezh2^Δ/Δ genotypes show increased myeloid output relative to control cells. Conversely, lymphoid priming in HSPCs may be licensed by other PRC2 co-factors or regulated by potential non-canonical functions of Jarid2.

Loss of Ezh2 leads to accumulation of lymphoid-biased MPP4 cells

While loss of Jarid2 increased overall differentiation output into all blood lineages, loss of Ezh2 manifested an expansion of total HSPCs, which was associated with lymphoid deficiency. To understand the mechanisms underlying this, single-cell transcriptomic analysis was performed on donor-derived HSPCs (CD45.2+ Lineage− Sca-1+ c-Kit+) from primary recipients of competitive whole BM (WBM) transplantation (Figure 3). Cell clustering and annotation were performed based on transcriptomic profiles (Figure S3A), with dimensionality reduction visualized using uniform manifold approximation and projection (UMAP; Figure 4A). Differential gene expression analysis between clusters was conducted using the Wilcoxon rank-sum test, applying thresholds of adjusted p < 0.01 and absolute log2 fold change >0.25. Marker genes for each cluster were identified using the same criteria (Figure S3B). Clusters were annotated by comparing enriched marker genes with established signatures of HSPC populations (Table S1) (Collins et al., 2024; Giladi et al., 2018).

Figure 4.

Loss of Ezh2 leads to accumulation of lymphoid-biased MPP4 cells

(A) UMAP clustering and cell annotation of scRNA-seq data generated from donor-derived HSPCs (CD45.2+ Lineage− Sca-1+ c-Kit+) post-primary transplant.

(B) Dotplot showing expression of Mycn and Runx1t1 in MPP1 cells of indicated genotypes.

(C) HSC module score showing average expression of HSC-specific genes in cell populations of indicated genotypes.

(D) Number of consensus H3K27me3 peaks from two biological replicates of CUT&TAG in HSPCs of indicated genotypes.

(E) UMAP split by genotype showing percentage of cells for each genotype within each cluster. Dashed red outline indicates MPP4 population.

(F) Percentage of cells from indicated genotypes that fall within MPP4 cell cluster. Data are mean ± SEM. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. One-way ANOVA with Tukey correction for multiple comparisons. See also Figures S3–S5 and Tables S1, S2, S3 and S4.

The ectopic self-renewal observed in Jarid2^Δ/Δ MPPs was previously shown to be driven by de-repression of HSC-specific genes such as Mycn and Runx1t1 (Celik et al., 2018). We confirmed upregulation of Mycn and Runx1t1 in Jarid2^Δ/Δ MPP1 in this single-cell RNA sequencing (scRNA-seq) dataset (Figure 4B). These genes were also upregulated in Ezh2^Δ/Δ MPP1 (Figure 4B), suggesting PRC2 might be broadly responsible for suppressing HSC self-renewal networks during hematopoietic differentiation. We collated a list of HSC-defining genes to compute an HSC module score (Collins et al., 2024; Herault et al., 2021; Nestorowa et al., 2016), which was significantly increased in Ezh2^Δ/Δ MPPs across multiple cell clusters (Figure 4C). This residual expression of HSC-specific genes in downstream progenitor cells suggests PRC2-mediated silencing is a major mechanism of suppressing HSC-defining transcriptional networks during hematopoietic commitment.

In general, PRC2 core components were consistently expressed across most clusters (Figure S4A). We hypothesized that the different phenotypes between Jarid2 and Ezh2 loss-of-function HSPCs may be partially due to compensation by other PRC2 co-factors in the absence of Jarid2. The only PRC2.1 co-factor that was significantly expressed in HSPCs was Mtf2 (Figure S4A), although it was not significantly different between the genotypes in most clusters. However, the PRC2.2 co-factor Aebp2 was modestly upregulated in HSPCs lacking Jarid2 (Figures S4B and S4C), presenting a possible compensatory mechanism. Similarly, Ezh1 was modestly upregulated in HSCs in the absence of Ezh2 (Figure S5B), suggesting not all PRC2 activity was lost after inactivation of Ezh2.

To determine if differential gene expression was associated with changes in repressive chromatin, CUT&TAG was performed for H3K27me3 in HSPCs. The consensus peaks from two biological replicates for each genotype identified the expected pattern with 68,268; 58,206; and 27,014 H2K27me3 peaks identified for control, Jarid2^Δ/Δ, and Ezh2^Δ/Δ HSPCs, respectively (Figures 4D and S5A; Table S2). To integrate chromatin and transcriptional data, we generated a PRC2 module score for known PRC2 target genes in HSCs (Majewski et al., 2008; Xie et al., 2014). The average expression of this gene module was consistently increased in Ezh2^Δ/Δ cells from all HSPC clusters compared to control and Jarid2^Δ/Δ cells (Figure S5B). Curiously, PRC2 target genes were decreased in some Jarid2^Δ/Δ HSPC clusters compared to control cells (Figure S5B), providing further support for potential non-canonical regulatory functions of Jarid2.

UMAP clustering (Figure 4E) identified a striking enrichment of Ezh2^Δ/Δ cells in the cluster annotated as MPP4 (Figure 4F). This cell population, defined by expression of marker genes such as Dntt, Flt3, and Notch1 (Figure S5C), are known to be lymphoid-biased MPPs (also called MPP^Ly; Figure 1D) (Challen et al., 2021; Pietras et al., 2015). The accumulation of Ezh2^Δ/Δ MPP4 was associated with increased expression of genes typically associated with HSC identity, such as Mecom, Hlf, and Meis1 (Figure S5D; Table S3), possibly implying that in the absence of Ezh2, these MPP4 acquire transcriptional features of HSCs that may convey some measure of self-renewal. The most striking of these was Pbx3, upregulation of which is associated with abnormal self-renewal of leukemic stem cells (Guo et al., 2017; Li et al., 2016), which was upregulated in most Ezh2^Δ/Δ HSPC clusters (Figure S5E), including MPP4 and lymphoid-primed progenitors (Figure S5F). The upregulation of these genes was associated with loss of H3K27me3 in Ezh2^Δ/Δ HSPCs (Figure S5G; Table S4). Taken together, these data indicate loss of Ezh2 leads to differentiation block and accumulation of lymphoid-biased MPP4 cells due to de-repression of HSC-specific gene signature genes. This may underlie the lymphoid-deficient output of these cells in serial BM transplantation (Figure 3).

Loss of Ezh2 leads to accumulation of B cell progenitors in the bone marrow

To investigate the potential impacts of Jarid2 and Ezh2 loss of function on HSPC lineage priming in a native setting (non-transplanted mice), Cellular Indexing of Transcriptomes and Epitopes by Sequencing (CITE-seq) was performed on HSPCs (Lineage− Sca-1+ c-Kit+) from sex-matched Mx1-Cre control, Jarid2^Δ/Δ, and Ezh2^Δ/Δ mice to gain simultaneous information on gene expression and surface proteins at the single-cell level. Antibody-derived tags (ADTs) were included for well-established cell surface markers of HSPC populations (c-Kit, Sca-1, CD48, CD150, CD201, and CD135; Figure S6A). UMAP clustering and cell-type annotation were performed using both transcriptomic and ADT expression profiles (Table S5) (Collins et al., 2024; Giladi et al., 2018). Dataset cross-validation showed the increased expression HSC-specific genes previously observed in Ezh2^Δ/Δ MPP4 cells was conserved in Ezh2^Δ/Δ MPPs in CITE-seq data (Figure S6B).

As an initial quality control, potential sex-specific effects were examined. No significant differences were observed between males and females within each genotype (Figure 5A). Given the absence of sex biases, data from male and female mice were combined for each genotype in downstream analyses. The most striking result was enrichment of two distinct B cell progenitor cell clusters in Ezh2^Δ/Δ HSPCs (consistent across both sexes; Figures 5B and S6C), defined by expression of B lineage markers such as Ebf1, Pax5, and CD19 (Figure 5C). Gene ontology enrichment analysis of differentially expressed genes in Ezh2^Δ/Δ B cell progenitors (Table S6) suggested defects in lymphocyte proliferation and B cell activation in the absence of Ezh2 (Figure 5D). The genes underlying these differences in Ezh2^Δ/Δ B cell progenitors were conserved in lymphoid-primed progenitors identified in the scRNA-seq analysis (Figure S6D). Gene set enrichment analysis (GSEA) demonstrated that Ezh2 loss of function led to activation of B cell survival pathways (Figure 5D), driven by upregulation of critical survival genes such as Mcl1 and IL7r (Figure 5E; Table S6). These data suggest loss of Ezh2 promotes expansion of early B cell progenitors in the BM through upregulation of factors that alter B cell survival and proliferation.

Figure 5.

Loss of Ezh2 leads to accumulation of B cell progenitors in the bone marrow

(A) UMAP clustering and cell annotation of CITE-seq data generated from HSPCs (Lineage− Sca-1+ c-Kit+) of indicated genotypes isolated from mice 8 weeks post-pIpC.

(B) Frequency of cell of indicated genotypes within B cell progenitor 1 and B cell progenitor 2 cell clusters.

(C) Expression of B cell markers CD19, Ebf1, and Pax5 in cells across the UMAP.

(D) Gene ontology analysis of differentially expressed genes in Ezh2^Δ/Δ B cell progenitors showing defects in lymphocyte proliferation and activation gene expression signatures.

(E) Gene set enrichment analysis (GSEA) of B cell progenitor 1 and B cell progenitor 2 clusters showing enrichment of B cell survival pathway gene expression in Ezh2^Δ/Δ cells.

(F) Pseudobulk differential gene expression analysis in B cell progenitor 1 and B cell progenitor 2 clusters showing overexpression of early B cell survival genes in Ezh2^Δ/Δ cells. Data are mean ± SEM. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. One-way ANOVA with Tukey correction for multiple comparisons. See also Figure S6 and Tables S5 and S6.

Loss of Jarid2 increases B cell output, whereas loss of Ezh2 leads to developmental arrest of pre-pro B cells

To determine if the transcriptional alterations in Ezh2^Δ/Δ HSPCs contributed to the functional deficiencies in lymphoid output observed from Ezh2^Δ/Δ cells in transplantation assays, B cell developmental stages were examined in primary recipient mice by flow cytometry (Figures 6A and S7). Ezh2^Δ/Δ recipient mice exhibited a significant accumulation of pre-pro B cells (B220+ IgM− CD43+ CD19−) compared to both control and Jarid2^Δ/Δ recipients (Figure 6B). Further dissection of the pre-pro-B cell compartment by CD24 expression (Ayre et al., 2015) demonstrated Ezh2 loss led to a marked increase in both early (CD24⁻) and late (CD24⁺) pre-pro-B cell populations (Figure 6B). However, at later stages of B cell development, there was a significant reduction of Ezh2^Δ/Δ cells, in stark contrast to Jarid2^Δ/Δ cells that were significantly increased (Figure 6B). To directly assess if loss of Ezh2 induces maturational arrest of B cell progenitors, in vitro B cell differentiation assays were performed. Purified control, Jarid2^Δ/Δ, and Ezh2^Δ/Δ HSCs were cultured on OP9 stromal cells with lymphoid cytokines and assessed for differentiation status by flow cytometry (Figure 6C). After 10 days in vitro, there was a significant increase in early pre-pro-B cells in Ezh2^Δ/Δ co-cultures with a concomitant decrease in late pre-pro-B cells compared to control and Jarid2^Δ/Δ cells (Figure 6D), directly confirming that Ezh2 is required for appropriate B cell maturation. In contrast, Jarid2 appears to restrict B cell potential in early HSPCs, as loss of Jarid2 consistently leads to increased B cell differentiation output.

Figure 6.

Loss of Jarid2 increases B cell output, whereas loss of Ezh2 leads to developmental arrest of pre-pro-B cells

(A) Representative flow cytometry gating to identify B cell developmental stages in the bone marrow (BM) of recipient mice.

(B) Frequency of different donor-derived B cell progenitor populations in the BM of primary transplant recipient mice.

(C) Representative flow cytometry analysis of day 10 cells from B cell differentiation assays.

(D) Frequency of early and late pre-pro-B cells derived from HSCs of indicated genotypes at day 10 of B cell differentiation assay. Data are mean ± SEM. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. One-way ANOVA with Tukey correction for multiple comparisons. See also Figure S7.

Discussion

Our prior work revealed the critical importance of the PRC2 co-factor Jarid2 in normal and neoplastic hematopoiesis. While Jarid2 was demonstrated to be a bona fide tumor suppressor in chronic myeloid neoplasms, loss of Jarid2 in normal MPPs leads to de-repression of HSC-specific transcriptional programs, conveying ectopic long-term self-renewal potential to this progenitor cell population (Celik et al., 2018). Molecular analyses suggested only a fraction of the transcriptional changes arising from loss of Jarid2 in MPPs were associated with changes in distribution of the repressive chromatin mark H3K27me3 catalyzed by PRC2. The goal of this study was to determine if Jarid2 may have PRC2-independent functions in hematopoiesis by comparing the functional and molecular properties of HSPCs lacking either Jarid2 or Ezh2, the enzymatic component of PRC2. Our findings define the major PRC2-dependent function of Jarid2 in hematopoiesis to be silencing of pro-myeloid differentiation pathways during lineage commitment. In the absence of both Jarid2 and Ezh2, HSPCs show enhanced myeloid output in functional assays. The developmental arrest of B cell maturation in Ezh2-deficient BM and subsequent resulting peripheral lymphoid deficiency was not observed in Jarid2-deficient BM, suggesting that other PRC2 co-factors mediate PRC2-dependent lymphoid priming in HSPCs. While the possibility remains for non-canonical functions of Jarid2 in licensing lymphoid commitment in HSPCs, this needs to be confirmed with combinatorial Jarid2-/Ezh2-deficient mutant mouse crosses.

The critical importance of PRC2-mediated gene regulation in hematopoiesis is underscored by the recurrence of EZH2 somatic mutations in blood cancers. EZH2 gain-of-function mutations are recurrent drivers of lymphoma (Beguelin et al., 2013; Morin et al., 2010), whereas EZH2 loss-of-function mechanisms (either point mutations or deletions of chromosome 7q) are common in myeloid neoplasms such as myelodysplastic syndromes (MDSs) and myelofibrosis (MF) (Ernst et al., 2010; Nikoloski et al., 2010; Score et al., 2012). While prior studies have indicated a role for Ezh2 in B cell development (Su et al., 2003) and germinal center formation (Beguelin et al., 2013), there has been surprisingly little functional characterization of how Ezh2 loss-of-function influences normal HSC fate decisions. A previous study suggested Ezh2 loss of function did not dramatically alter function of HSCs in transplant assays (Xie et al., 2014). But these experiments were done with fetal-liver- or neonatal-derived HSCs. This study also used Ezh2^fl/fl mice crossed to the Vav-Cre driver, which we were unable to generate in this study, perhaps due to strain or animal house differences. Another study reported that overexpression of Ezh2 by retroviral transduction increased transplantability of HSCs (Kamminga et al., 2006). We attempted to reconcile these differences and provide a definitive functional assessment of Ezh2-deficient HSPCs by using conditional knockout adult Mx1-Cre:Ezh2^fl/fl mice. Our data clearly show that loss of Ezh2 leads to a drastic influence on lineage output by driving myeloid differentiation at the expense of lymphopoiesis. In contrast, transplantation of Jarid2-deficient BM enhanced engraftment in all major blood lineages. These studies using the Mx1-Cre strain corroborated our prior data using Vav-Cre to conditionally inactivate Jarid2 showing a reduction in HSC regeneration after transplant and endowment of MPPs with enhanced repopulating activity (Celik et al., 2018). However, the lineage differentiation of Jarid2-deficient HSPCs was enhanced in the Mx1-Cre model, which may reflect an influence of the interferon response induced by injection of pIpC. Future studies will determine if there is a specific function of Jarid2 in HSPCs in response to inflammatory stress.

This work provides clear resolution into the function of Ezh2 in adult HSCs during normal hematopoiesis and provides evidence supporting the concept of non-canonical functions of Jarid2 regulating HSC fate decisions. This adds to the growing body of literature, suggesting that many epigenetic regulators that are molecularly defined in biochemical or isolated cell culture assays may have functions outside these described roles that regulate stem cell populations in vivo. Together, this work positions Jarid2 and Ezh2 as critical, mechanistically distinct regulators of HSC fate, with broad relevance to the regulation of normal and neoplastic hematopoiesis.

Methods

Animals

All animal procedures were approved by the Institutional Animal Care and Use Committee of Washington University. All mice used in this study were on C57Bl/6 background. Jarid2^fl/fl mice were previously described (Mysliwiec et al., 2006). Ezh2^fl/fl mice (Neff et al., 2012) were obtained from Dr. Jeffrey Magee (Washington University). Jarid2 and Ezh2 carrying homozygous floxed alleles were crossed to the Mx1-Cre driver. For strains carrying Mx1-Cre, recombination of floxed alleles (=“Δ”) was induced by intraperitoneal injection of six doses (300 mg/mouse) of polyinosinic-polycytidylic acid (pIpC; Sigma #P1350) given every other day. Ten- to twelve-week-old mice were typically used for experimentation. Equal numbers of male and female mice were used, and no gender biases were noted. Eight-week-old male and female C57Bl/6 CD45.1 mice (The Jackson Laboratory # 002014) were used as recipients for BM transplantation assays.

Methocult-based HSC genotyping

Mx1-Cre-induced recombination efficiency was assessed by sorting single HSCs from either Jarid2^Δ/Δ or Ezh2^Δ/Δ mice into 96-well plates containing Methocult M3434 medium (Stem Cell Technologies #03434). After 2-weeks culture, individual colonies were collected and washed with Dulbecco’s phosphate buffer saline (Sigma #D8537), and genomic DNA was isolated using the KAPA Express Extract Kit (Sigma # KK7103). To confirm the deletion of Jarid2 (exon 3) and Ezh2 (exons 16 and 17), PCR was performed as previously described using primer pairs: Jarid2-FloxOut-F//Jarid2-FloxOut-R (Mysliwiec et al., 2006) and Ezh2 Excision F1//Ezh2 Excision R1//Ezh2 Excision R2 (Neff et al., PNAS, 2012), respectively. The recombination efficiency was checked using the following PCR conditions: 95°C for 3 min, 95°C for 15 s, 60°C for 15 s, 72°C for 20 s, and 72°C for 5 min.

Bone marrow transplantation

CD45.1 congenic C57Bl/6 mice were used as recipients for all transplantation experiments. Mice received a lethal dose of total body irradiation administered in two split doses of 5.25 Gy (10.5 Gy total) delivered ∼4 h apart using a gamma irradiator. For primary competitive transplantation experiments, 100 donor-derived MPP cells (CD45.2⁺ Lin^– Sca-1⁺ c-Kit⁺ CD48^– CD150^–) were isolated via fluorescence-activated cell sorting (FACS) on a MoFlo cell sorter (Beckman Coulter) and transplanted via retro-orbital injection along with 2.5 × 10⁵ unfractionated wild-type WBM competitor cells. For primary whole bone marrow transplants, 5 × 10⁵ total CD45.2 WBM cells were co-transplanted with 5 × 10⁵ wild-type CD45.1 WBM competitor cells into lethally irradiated recipients. Donor and competitor cells were distinguished by differential expression of CD45 allelic isoforms: CD45.2 (donor) and CD45.1 (competitor and recipient). At 18 weeks post-transplant, 3 × 10⁶ WBM cells were harvested from individual primary recipients transferred into uniquely identified lethally irradiated secondary recipients to assess long-term repopulating capacity.

Flow cytometry

Peripheral blood (PB) was collected from recipient mice, and red blood cells were lysed using ACK lysis buffer (Gibco). Leukocytes were stained for donor chimerism and lineage analysis using flow cytometry (Cytek Northern Lights). Surface markers used to assess lineage contribution included myeloid lineage markers Gr-1 (Ly6G/Ly6C) and Mac-1 (CD11b), B cell lineage marker B220 (CD45R), and T cell lineage marker CD3ε. CD45.1 and CD45.2 were used to discriminate recipient/competitor and donor-derived populations, respectively. At 18 weeks post-transplantation, recipient mice were sacrificed, and WBM was harvested from femurs, tibias, and iliac crests. BM cells were stained for HSPCs and mature lineage markers. Lineage markers included Gr-1, Mac-1, Ter119, CD3ε, and B220. HSPC panel included c-Kit (CD117), Sca-1 (Ly6A/E), Flk2 (Flt3), CD150 (Slamf1), and CD48 (SLAMF2). LSK (Lin^– Sca-1⁺ c-Kit⁺) populations and HSPC subfractions including MPP^Ly (Lin^– Sca-1⁺ c-Kit⁺ Flk2⁺ CD48⁺ CD150^–), MPP^G/M (Lin^– Sca-1⁺ c-Kit⁺ Flk2^– CD48⁺ CD150^–), MPP^Mk/E (Lin^– Sca-1⁺ c-Kit⁺ Flk2^– CD48⁺ CD150⁺), MPP (Lin^– Sca-1⁺ c-Kit+ Flk2− CD48− CD150−), and HSC (Lin− Sca-1+ c-Kit+ Flk2− CD48− CD150+) were identified and quantified. BM B cell development was analyzed using B220, CD19, CD43, CD24, immunoglobulin G (IgM), and IgD to identify pre-pro-B, early pre-pro-B, late pre-pro-B, pro-B, pre-B, immature, and mature B cell subsets. The standard Hardy fraction flow cytometry gating strategy was used to classify B cell developmental stages in the BM.

All antibody staining was performed in Hank’s balanced salt solution (HBSS) buffer (Corning #21021CV) containing Pen/Strep (100 Units/mL; Thermo Fisher Scientific #MT30002CI), HEPES (10 μM; Life Technologies # 15630080), and fetal bovine serum (FBS) (2%; Sigma #14009C). Briefly, cells were suspended in complete HBSS at a concentration of 5 × 10⁸ cells/mL and incubated on ice for 20 min with the desired antibodies listed in the table below. For cell sorting, magnetic enrichment was carried out using the AutoMACS Pro Seperator (Miltenyi Biotec) with mouse CD117-conjugated microbeads (Miltenyi Biotec #130-091-224). Post-enrichment, the positive cell fraction was stained with appropriate antibodies and sorted by FACS (Beckman Coulter MoFlo). All antibodies utilized in this study were used at 1:100 dilutions and were obtained from BioLegend, eBioscience, or BD Biosciences unless otherwise stated. Samples were analyzed using a Cytek Northern Lights flow cytometer. Data were analyzed using FlowJo software (BD Biosciences). Fluorescence minus one (FMO) controls and single-color controls were included for compensation and gating.

Flow Cytometry Antibody	Source	Identifier
BV605 anti-mouse CD45.2	BioLegend	Cat# 109841
FITC anti-mouse CD45.1	BioLegend	Cat# 110706
APCy7 anti-mouse Gr-1	BioLegend	Cat# 108424
APCy7 anti-mouse Mac-1	BioLegend	Cat# 101226
APCy7 anti-mouse B220	BioLegend	Cat# 103224
APCy7 anti-mouse CD3e	BioLegend	Cat# 100330
APCy7 anti-mouse Ter119	BioLegend	Cat# 116223
BV421 anti-mouse c-Kit	BioLegend	Cat# 105827
APC anti-mouse Sca-1	BioLegend	Cat# 122512
PECy7 anti-mouse CD48	BioLegend	Cat# 103424
PE anti-mouse CD150	BioLegend	Cat# 115904
BV421 anti-mouse CD45.2	BioLegend	Cat# 84208
APC anti-mouse CD3e	BioLegend	Cat# 100312
APC anti-mouse B220	BioLegend	Cat# 103212
PECy7 anti-mouse Gr-1	BioLegend	Cat# 108416
PECy7 anti-mouse Mac-1	BioLegend	Cat# 101216
PECy7 anti-mouse B220	BioLegend	Cat# 103222
PE anti-mouse Flk2	BD Biosciences	Cat# 553842
FITC anti-mouse c-Kit	BD Biosciences	Cat# 561680
BV421 anti-mouse CD150	BioLegend	Cat# 115925
PECy7 anti-mouse IgM	eBioscience	Cat# 25-5790-82
BV421 anti-mouse CD19	BioLegend	Cat# 115538
PE anti-mouse CD43	eBioscience	Cat# 12-0431-83
APCcy7 anti-mouse CD24	BioLegend	Cat# 101850
FITC anti-mouse CD45.2	BioLegend	Cat# 109806
PE anti-mouse IgD	eBioscience	Cat# 12-5993-82
Biotin anti-mouse Sca-1	BioLegend	Cat# 108103

Single-cell RNA sequencing

c-kit+ cells were enriched from the BM of 18-week post-primary transplant recipient mice. Donor-derived (CD45.2+) KSL (c-Kit+ Sca1+ Lineage−) cells were subsequently sorted from the enriched population and resuspended at a concentration of 1,200 cells/mL in PBS supplemented with 0.04% BSA. Single-cell libraries were prepared using the Chromium Next GEM Single Cell 3′ Reagent Kits v3.1 (Dual Index) (10× Genomics, PN-1000268), following the manufacturer’s protocol (User Guide CG000317 Rev D) with minor modifications. Cell partitioning and barcoding were performed using the Chromium Next GEM Chip G Single Cell Kit (PN-1000127) and Chromium Dual Index Kit TT Set A (PN-1000215). GEM generation and reverse transcription (GEM-RT), followed by post-GEM cleanup, were conducted according to standard procedures. Purified cDNA was amplified for 11–13 cycles based on input cell number and quality and then cleaned using SPRIselect beads (Beckman Coulter, Cat. No. B23318). Amplified cDNA was assessed using the Agilent Bioanalyzer High Sensitivity DNA Kit (Agilent, Cat. No. 5067-4626) to determine concentration and size distribution. Final gene expression (GEX) libraries were constructed as per the 10× Genomics protocol, adjusting the number of amplification cycles according to cDNA input. The libraries were quantified using the KAPA Library Quantification Kit for Illumina platforms (Roche/KAPA Biosystems, Cat. No. KK4824) and sequenced on the Illumina NovaSeq X Plus system using the XP workflow on an S4 Flow Cell with a paired-end 28 × 10 × 10 × 150 bp sequencing strategy. A target median depth of 50,000 reads per cell was achieved.

FastQ files were demultiplexed, aligned to the mouse reference genome (mm10), and processed using Cell Ranger v.3.1.0. After generation of the gene-barcode matrix, low-quality cells were filtered out based on quality control metrics: cells with >10% mitochondrial gene expression (indicative of stress or apoptosis) and cells in the top 5% of total UMI or feature counts (to exclude potential multiplets or doublets) were excluded from downstream analysis. Further analysis was performed using Seurat v.3.0 in R and Partek Flow software. Dimensionality reduction was conducted via principal-component analysis (PCA), and clusters were visualized using both t-distributed stochastic neighbor embedding (t-SNE) and uniform manifold approximation and projection (UMAP). Marker genes for each cluster were identified using a Wilcoxon rank-sum test, with significance set at adjusted p < 0.01 and absolute log2 fold change >0.25. Clusters were annotated based on known HSPC gene markers. Differential gene expression analysis within clusters was performed to compare cells from control, Jarid2^Δ/Δ, and Ezh2^Δ/Δ mice. Enriched pathways were identified using the GSEA software package. All raw read data (FASTQ files) are publicly available at the Gene Expression Omnibus database under accession number GSE302605.

Single-cell CITE-seq

c-Kit+ Sca1+ Lineage− (KSL) cells were sorted by FACS. Cells were first stained with biotin-conjugated anti-Sca1 (Ly-6A/E) antibody for 20 min at 4°C, then washed and incubated with a panel of six TotalSeq-B barcoded antibodies targeting surface markers: streptavidin-PE (Sca-1; BioLegend #405287), CD117 (BioLegend #105849), CD48 (BioLegend #103457), CD150 (BioLegend #115951), CD135 (BioLegend #135319), and CD201 (BioLegend #141511). Antibody staining was performed using 1 μL of each antibody in 50 μL of PBS containing 0.04% BSA, for 30 min on ice, protected from light. Following staining, cells were washed with cold PBS containing 0.04% BSA, filtered through a 40 μm cell strainer, and assessed for viability (>90%) using Trypan Blue exclusion. After that, cells were resuspended at 1,200 cells/μL in PBS + 0.04% BSA. Single-cell transcriptome and protein profiling were performed using the Chromium Next GEM Single Cell 3′ Reagent Kits v.3.1 (Dual Index) (PN-1000268) in conjunction with the Feature Barcode Kit for Cell Surface Protein (PN-1000262), following the manufacturer’s protocol (User Guide CG000317 Rev D) with slight optimizations to the staining and preparation steps. Cells were loaded onto the Chromium Controller using the Chromium Next GEM Chip G (PN-1000127), targeting recovery of approximately 10,000 cells per channel. Gel bead-in-emulsion (GEM) generation, barcoding, and reverse transcription were carried out using the recommended thermal cycling conditions: lid temperature 53°C, 53°C for 45 min, and 85°C for 5 min; hold at 4°C. Post-RT GEMs were broken, and barcoded cDNA was purified using Dynabeads MyOne Silane (PN-2000048). Amplified cDNA containing both mRNA- and antibody-derived tags (ADTs) was generated via PCR using cycling conditions recommended by 10× Genomics and cleaned up using SPRIselect reagent (Beckman Coulter, Cat. No. B23318).

Separate libraries were prepared for gene expression and Feature Barcode (protein) profiling: (1) Gene Expression Library: a portion of the cDNA was enzymatically fragmented, end-repaired, A-tailed, and ligated to adapters, followed by sample index PCR amplification using the Dual index Kit TT Set A (PN-1000215). (2) Feature Barcode Library: a separate cDNA aliquot was amplified using primers targeting the Feature Barcode constructs using Dual index Kit NT Set A (PN-1000242). Final libraries were quantified using a Qubit 4 Fluorometer and assessed for size distribution using the Agilent Bioanalyzer High Sensitivity DNA Kit (Cat. No. 5067-4626). Libraries were pooled in appropriate ratios and sequenced on the Illumina NovaSeq X Plus system using an S4 flow cell with the following read configuration: 28 bp (Read 1), 10 bp (i7 index), 10 bp (i5 index), and 90 bp (Read 2). A target median sequencing depth of ∼50,000 reads per cell was used for gene expression and 5,000–10,000 reads per cell for antibody-derived tags.

FastQ files were demultiplexed and processed using Cell Ranger v.3.1.0. Gene expression and antibody capture reads were analyzed simultaneously using the cellranger count pipeline with a custom reference genome built from the mm10 genome assembly and the appropriate Feature Barcode reference file. Cells with >10% mitochondrial transcript content and those within the top 5% of UMI counts (potential doublets) were excluded from downstream analysis.

Further analysis was performed using Seurat v.3.0 (R package) and Partek Flow software. Dimensionality reduction was conducted using PCA on the gene expression matrix, and antibody-derived tag (ADT) data were analyzed in parallel following centered log-ratio (CLR) normalization. Clustering was performed using graph-based algorithms and visualized via t-SNE and UMAP. Cluster-specific gene and surface protein markers were identified using the Wilcoxon rank-sum test, with significance thresholds of adjusted p < 0.01 and absolute log2 fold change >0.25. Cluster annotation was guided by known HSPC transcriptomic and surface marker profiles. Differential gene expression analysis within clusters was performed to compare cells from control, Jarid2^Δ/Δ, and Ezh2^Δ/Δ mice. Enriched pathways were identified using the GSEA software package. All raw read data (FASTQ files) are publicly available at the Gene Expression Omnibus database under accession number GSE302605.

Cut and tag

Approximately 20,000 HSPCs (Lineage− Sca-1+ c-Kit+) were purified by flow cytometry. Immediately following isolation, cells were fixed on ice for 5 min using fixation buffer containing 0.5% formaldehyde (final concentration) and 10% BSA in PBS. Nuclei were then permeabilized on ice for 5 min using an isotonic permeabilization buffer. Fixed and permeabilized nuclei were incubated overnight at 4°C with a primary antibody targeting H3K27me3 (Rabbit mAb, Cat# 9733S, Cell Signaling Technology) prepared in low-salt buffer. Normal Rabbit IgG (Cat# 2729S, Cell Signaling Technology) was used as a negative control. The following day, nuclei were incubated for 1 h at room temperature with a guinea pig anti-rabbit secondary antibody (Cat# ABIN101961) diluted in low-salt buffer. After washing once with low-salt buffer, nuclei were incubated with Protein A-Tn5 Transposase (pA-Tn5, Transposase-Loaded, Cat# C01070001-T30, Diagenode) for 1 h at room temperature. Nuclei were subsequently washed once with high-salt buffer, and tagmentation was initiated by adding 2 μL of 250 mM MgCl₂ to nuclei resuspended in 50 μL of high-salt buffer. Samples were incubated in a thermomixer at 37°C and 550 rpm for 1 h. The reaction was stopped by adding 4 μL of 0.5 M EDTA, followed by a single wash with cold PBS. Genomic DNA was then extracted using the PureLink Genomic DNA Mini Kit (Cat# 3059073, Thermo Fisher Scientific). DNA concentration was quantified using the Qubit dsDNA High Sensitivity Assay Kit (Cat# Q32851, Thermo Fisher Scientific). Indexed libraries were generated via PCR amplification using specific i5 and i7 index primers and the NEBNext High-Fidelity 2× PCR Master Mix (Cat# M0541S, New England BioLabs). Library size selection was performed using SPRIselect beads (Cat# B23318, Beckman Coulter).

Cut and tag sequencing and analysis

Final libraries were quantified using a Qubit 4 Fluorometer and assessed for size distribution using Agilent TapeStation 4150 (Agilent Technologies, Cat# G2992A) and Agilent 631 Bioanalyzer High Sensitivity DNA Kit (Cat# 5067–4626). Libraries were pooled in appropriate ratios and sequenced on the Illumina NovaSeq X Plus system using an S4 633 flow cell with the following read configuration: 28 bp (Read 1), 10 bp (i7 index), 10 bp (i5 index), and 90 bp (Read 2). A target sequencing depth of 150 × 10⁶ reads was used per sample. Raw sequencing reads (FASTQ files) were first evaluated for quality using FastQC (v.0.12.1) and summarized with MultiQC (v.1.16) to assess per-base quality, adapter contamination, and GC content. Adapter trimming and low-quality read removal were performed using Trim Galore (v.0.6.10). Trimmed reads were aligned to the mouse reference genome (mm39) using Bowtie2 (v.2.5.2) with parameters optimized for CUT&Tag’s short and low-background fragments. SAM files were converted to BAM, sorted, and indexed using SAMtools (v.1.20). PCR duplicates were removed using Picard (v.3.1.1). Only properly paired, uniquely mapped reads were retained for downstream analysis. Filtered BAM files were used to generate fragment files (BEDPEformat) using bedtools bamtobed. To visualize the CUT&Tag signal, RPGC normalized genome coverage tracks were generated using deepTools (v.3.5.5). These BigWig files and the bam files were used for global signal visualization and representative genome browser tracks. Peak calling was performed using MACS2 (v.2.2.9.1) with parameters optimized for CUT&Tag and for the broad distribution characteristic of H3K27me3. Peaks (broad domains) were annotated to the nearest genes and genomic features using ChIPseeker (v.1.38.0). Shared and condition-specific peak sets across experimental groups were identified using bedtools intersect and ChIPpeakAnno and visualized using VennDiagram or Intervene (v.0.6.5). Global enrichment or depletion of H3K27me3 signal was quantified using deepTools. Signal matrices, heatmaps, and average profiles (“tornado” or aggregate plots) were generated to visualize mean CUT&Tag signal intensity across peak regions or gene bodies, revealing global shifts between conditions. Representative loci were visualized using Integrative Genomics Viewer (IGV). BigWig tracks from each condition were loaded, and normalized signal intensities were displayed across selected gene regions of interest, highlighting the representative depletion of H3K27me3 peaks. Differential peak intensity analysis was performed using Limma (v.3.66.0), normalizing for effective genome size, library size, and sequencing depth. Statistical significance of global H3K27me3 depletion between genotypes was determined using Wald tests with an false discovery rate (FDR) threshold of 0.05. Plots were generated using R (v.4.3) and ggplot2 for consistency and reproducibility. Motif enrichment analysis was performed with findMotifsGenome.pl function from Homer (v.5.0).

OP9-based B cell differentiation assay

To assess B cell differentiation potential, an OP9 stromal cell co-culture system was employed. On day 1, OP9 cells were seeded at a density of 5 × 10⁴ cells per well in 24-well plates using OP9 culture medium composed of α-MEM (Gibco), supplemented with 20% FBS (Gibco) and 1% penicillin/streptomycin (Gibco). On day 2, the medium was replaced with B cell differentiation medium consisting of OP9 culture medium supplemented with 5 ng/mL Flt3 ligand (Flt3L; PeproTech) and 5 ng/mL interleukin-7 (IL-7; PeproTech). Immediately following the medium change, 250 HSCs were plated per well onto the OP9 monolayer. Cultures were maintained for 10 days, with B cell differentiation medium replenished every 3 days. On day 10, non-adherent and loosely adherent cells were harvested and filtered through a 40 μm cell strainer (Corning) to remove residual OP9 stromal cells. Single-cell suspensions were stained with fluorophore-conjugated antibodies against CD45.2 and B cell progenitor markers (B220, CD43, and CD24) for flow cytometric analysis.

Quantification, UMAP cell clustering, and statistical analysis

Statistical comparisons were performed using one-way ANOVA followed by Tukey’s post hoc test for multiple comparisons. Cell clustering and annotation from scRNA-seq and CITE-seq data were performed based on integrated gene expression and antibody-derived tag (ADT) profiles. Cluster-specific marker identification used the Wilcoxon rank-sum test with thresholds of adjusted p < 0.01 and absolute log2 fold change >0.25. Pseudobulk differential expression analysis and GSEA were conducted using thresholds of adjusted p < 0.05 and absolute log2 fold change >0.25. Statistical significance is represented as follows: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Error bars on graphs represent the standard error of the mean (SEM).

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact Grant A Challen (grantchallen@wustl.edu).

Materials availability

Mice and other materials used in this study are available from the lead contact with a completed materials transfer agreement.

Data and code availability

• •All raw read data (FASTQ files) for single cell sequencing and CUT&TAG data are publicly available at the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE302605.
• •No original code was generated in the study.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We thank all members of the Challen laboratory, particularly Samantha Burkart for laboratory management. We thank the Alvin J. Siteman Cancer Center, supported in part by an NCI Cancer Center Support Grant P30CA091842. This publication is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. G.A.C. was supported by the National Institutes of Health (NIH; HL147978, CA236819, and DK124883), the Edward P. Evans Foundation, the Leukemia and Lymphoma Society (6667-23), and the American Cancer Society (CSCC-RSG-23-991417-01-CSCC). H.B. was supported by the Edward P. Evans Center for MDS at Washington University in St. Louis. I.X.R. was supported by NIH P30 CA091842. A.K. was supported by the American Society of Hematology (Graduate Hematology Award). J.A.M. was supported by the NIH (HL152180) and the Children’s Discovery Institute of Washington University and St. Louis Children’s Hospital. J.A.M. is a Scholar of the Leukemia and Lymphoma Society.

Author contributions

Conceptualization and study design, G.A.C. Experimentation and data acquisition, H.B., W.H., N.I., A.K., I.X.R., J.A., and Y.L. Data analysis, H.B., W.H., S.L., W.Y., J.A.M., and G.A.C. Funding acquisition, G.A.C. Project administration and supervision, G.A.C. Manuscript preparation, H.B. and G.A.C.

Declaration of interests

The authors declare the following competing interests (unrelated to this work): G.A.C. has performed consulting and received research funding from Incyte, Ajax Therapeutics, Atavistik Bio, and ReNAgade Therapeutics Management and is a co-founder, member of the scientific advisory board and shareholder of Pairidex, Inc.

Published: February 5, 2026