Polycomb repressive complex 2 component Suz12 is required for hematopoietic stem cell function and lymphopoiesis

Hematopoiesis is the process of blood cell formation that generates approximately 1 trillion blood cells per day in humans¹  that is sustained by rare HSCs resident in the bone marrow (BM). The balance between HSC self-renewal and differentiation is highly dynamic, allowing for the production of phenotypically distinct blood cells at steady state, as well as rapid expansion of specific cell types during stress. This balance is a tightly regulated process governed by both cell-extrinsic factors (eg, cytokine signaling and the BM microenvironment)^2,3  and cell-intrinsic cues (eg, transcription factors and chromatin modifiers).^4-6  Hence, it is not surprising that disruption to this delicate balance can result in life-threatening hematologic disorders, such as leukemia.

Polycomb group proteins are a major class of epigenetic regulators that repress target genes by coordinating changes in the chromatin structure via posttranslational modifications of histone tails. They exists in several multiprotein complexes and are important in fine-tuning gene expression.⁷  PRC2 consists of 3 core components: embryonic ectoderm development (Eed), suppressor of zeste 12 protein homolog (Suz12), and either enhancer of zeste homolog 2 (Ezh2) or its close homolog enhancer of zeste homolog 1 (Ezh1). Ezh1 and Ezh2 both possess histone methyltransferase activity for di- and tri-methylation of histone H3 at position lysine 27 (H3K27me2/3). Although these core factors are essential for PRC2 complex stability and function, the presence of several accessory factors including Rbbp4/7, Jarid2, Mtf2, Phf1, and Phf19 have been implicated in the recruitment of PRC2 to its target genes and modulation of its activity.⁸  PRC2 core components are required for embryonic development: Eed, Suz12, and Ezh2 knockout mice develop abnormally and die prior to gastrulation.^9-11  Moreover, PRC2 is essential for proper differentiation of stem cells from diverse lineages including embryonic, skin, liver, and neuronal stem cells.^12-17 

Conditional knockout studies in mice have enabled dissection of the roles of PRC2 in the hematopoietic system. In the lymphoid lineages, Ezh2 deletion resulted in impairment of B-lymphoid development,^18-20  germinal center B-cell formation,²¹  and terminal differentiation of naïve CD4⁺ T-cells,²²  but it was largely dispensable for T-lymphoid development in the thymus.²³  Although several studies have shown that conditional inactivation of Ezh2 appears to have minimal impact on HSC function in adult mice, the role of Ezh2 in fetal hematopoiesis is less clear. Mochizuki-Kashio et al¹⁸  argued that fetal hematopoiesis relies exclusively on Ezh2-PRC2, and this dependency shifts toward Ezh1-PRC2 during the transition from neonates to adults, evidenced by increased Ezh1 expression and a concomitant decline in Ezh2 expression in aging HSCs. Although Ezh1 is not required for fetal hematopoiesis, Hidalgo et al²⁴  further demonstrated that Ezh1-deficient adult mice progressively developed BM failure as a result of abnormal activation of senescence programs in HSCs. Xie et al²⁰  recently generated an Eed conditional knockout mouse and showed that Eed is essential for adult but not fetal HSC function. HSCs overexpressing Ezh2 also possess enhanced capacity for serial transplantation in vivo, and recipient mice develop myeloproliferative-like disorders.^25,26  Somewhat paradoxically, a number of studies have demonstrated that partial inhibition of PRC2 also enhances HSC function^2,27  and can predispose to leukemia.^28,29  Collectively, these observations illustrate that hematopoietic cells respond sensitively to altered PRC2 dosage, and any slight deviation can dramatically alter the hematopoietic system. Indeed, a spectrum of mutations in EED, EZH2, and SUZ12 has been described in many human cancers including lymphoid and myeloid malignancies, highlighting the association of altered PRC2 activity with oncogenesis.^22,30-35  Thus, although PRC2 has become a compelling therapeutic target in malignancy, a more detailed understanding of the complex functional consequences of altered PRC2 activity in the hematopoietic system will be a prerequisite for effective manipulation of PRC2 activity in disease.

Here, we have generated a novel conditional knockout allele of Suz12 in mice. We show that unlike partial loss of Suz12 function, deletion of Suz12 in hematopoiesis results in dysfunction of both fetal and adult HSCs, which become unable to contribute effectively to blood cell production. Moreover, via restricted deletion in specific blood cell lineages, we establish that Suz12 is dispensable for production of myeloid cells, megakaryocytes, and platelets, while it is essential for B- and T-lymphoid development.

Experiments were approved by the Walter and Eliza Hall Institute of Medical Research Animal Ethics Committee. A targeting vector was generated carrying loxP sites flanking exon 5 of the murine Suz12 gene that included a neomycin-selection cassette flanked by 2 Flip recombinase target sites (Figure 1A). To verify correct targeting, genomic DNA from transfected C57BL/6 ES cells were digested with HindIII and the fragments were separated on 0.8% agarose gel to detect targeted clones, which was confirmed by Southern blotting (8.4 kb). A neomycin-resistant embryonic stem cell clone carrying the targeted locus was then used to generate chimeric mice. The neomycin cassette was removed by crossing to FlpE transgenic mice³⁶  to generate the Suz12 floxed (fl) allele, herein referred to as Suz12^fl. A polymerase chain reaction (PCR)-based strategy was developed for further genotyping to distinguish the wild-type (WT), FL, and recombined (Δ) alleles (PCR primers: forward: CAAAAACCTTGTCTCGGAAA, reverse: TCATCATCGGGATATGGGTA). The DelCre,³⁷ LysMCre,³⁸ Pf4Cre,³⁹ Rag1Cre,⁴⁰ Rosa26CreERT2⁴¹  (herein referred to as CreER^T2), and VavCre⁴²  mice have been previously described. All studies were performed with mice on a C57BL/6 genetic background.

Single-cell suspensions from BM, spleen, thymus, mesenteric lymph nodes, and E13.5 fetal livers were prepared in balanced salt solution (150 mM NaCl, 3.7 mM KCl, 2.5 mM CaCl₂, 1.2 mM MgSO₄, 7.4 mM HEPES-NaOH, 1.2 mM KH₂PO₄, 0.8 mM K₂HPO₄) supplemented with 5% (vol/vol) fetal calf serum. For analysis of blood, red cells were lysed by treatment with lysis buffer (150 mM NH₄Cl, 0.1 mM EDTA, and 12 mM NaHCO₃) and the remaining cells were washed with balanced salt solution/5% fetal calf serum. Monoclonal antibodies used were: B220 [clone RA3-6B2], CD4 [GK1.5], CD8a [53-6.7], CD16/32 [24G.2], CD19 [1D3], CD25 [PC61.5], CD34 [RAM34], CD43 [S7], CD44 [IM7], CD45.1 [A20], CD45.2 [104], CD48 [HM48-1], CD127/IL7Rα [A7R34], CD135/Flt3 [A2F10], CD150 [9D1], c-Kit [2B8], Gr-1 [RB6-8C5], IgM [II/41], Mac-1 [M1-70], Ly6D [49-H4], Ly6G [Ly6G], Sca-1 [D7], and Ter119 [Ly76]. Antibodies were conjugated to Ag-presenting cell, AlexaFluor-590, AlexaFluor-700, brilliant violet 421, FITC, PE, PECy7, PerCPCy5.5, or biotin, followed by streptavidin-conjugated phycoerythrin-TexasRed (BD Biosciences, eBioscience, or Biolegend). Propidium iodide or fluorogold was used for dead cell exclusion. Samples were processed using LSR II and LSR Fortessa (BD Biosciences) instruments with data analysis performed using FlowJo software (Treestar Inc., Ashland).

Peripheral blood was collected from the retro-orbital plexus for automated cell count (Advia3120, Bayer). Megakaryocyte and macrophage cultures are described in detail in the supplemental Methods, available on the Blood Web site.

Competitive BM transplantation studies were performed using 1 × 10⁶ test cells (CD45.2) and 1 × 10⁶ WT competitor cells (CD45.1) from 8- to 10-week-old mice. BM cells were injected intravenously into 8- to 10-week-old irradiated CD45.1/CD45.2 or CD45.1 recipients (2 × 550 Gy). Three recipient mice were used per test donor (n = 3-4 donors). For secondary transplants, 2 × 10⁶ cells from individual primary transplant recipients were injected into 3 CD45.1 recipients and analyzed 12 weeks later. For Suz12/CreER^T2 studies, recipient mice were bled 5-weeks posttransplant to assess baseline chimerism, and 1-week later, tamoxifen (Sigma) or vehicle (10% ethanol in peanut oil; Sigma) was administered via oral gavage for 3 consecutive days (4.2 mg/day). For gene expression analysis, a single dose of tamoxifen (3 mg) was administered by gavage.

Cells were homogenized in radio immunoprecipitation assay buffer (1% NP-40, 0.1% sodium dodecyl sulfate, 0.5% sodium deoxycholate, 150 mM NaCl, 50 mM Tris-HCl) and supplemented with Complete Protease Inhibitors (Roche). Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to polyvinylidene fluoride membranes, and blotted with antibodies against Suz12 (P-15, Santa Cruz), Ezh2 (AC22, Cell Signaling Technology), H3K27me3 (07-449, Millipore), histone H3 (AS3, Millipore), and actin (I-19, Santa Cruz).

Data were analyzed using GraphPad Prism 5.0. When multiple statistical tests were performed, the new threshold for statistical significance was established by dividing the standard significance threshold by the number of tests (P value < .05). A two-tailed Student t test was performed in 2 group comparisons. When comparing multiple groups, one-way analysis of variance was performed, followed by Tukey’s post-hoc test.

Suz12^Δ/+ mice were generated by crossing animals bearing a Suz12^fl allele with DelCre transgenic mice. The DelCre transgene was bred out and intercrosses established between Suz12^Δ/+ mice. No Suz12^Δ/Δ offspring were weaned from these matings (supplemental Table 1), consistent with the embryonic lethality previously reported in homozygous Suz12 knockout⁴³  and gene-trap¹¹  mice. To confirm conditional Suz12 deletion, we generated Suz12^fl/fl/CreER^T2 mice, where deletion of Suz12 exon 5 can be induced in the presence of tamoxifen. For 3 consecutive days, tamoxifen (4.2 mg/day) was administered, and organs were harvested 8 days after the last injection. Genomic PCR analysis showed efficient excision of the loxP-flanked exon in both the BM and spleen (Figure 1B). Immunoblot analysis confirmed the absence of Suz12 protein and a concomitant reduction in Ezh2 protein in the BM, spleen, and thymus (Figure 1C). Similarly, when freshly isolated Suz12^fl/fl/CreER^T2 E13.5 lineage-depleted fetal liver cells were cultured in the presence of 4-hydroxytamoxifen (1 × 10⁻⁷ M) or vehicle (100% ethanol) for 24 hours, immunoblot analysis confirmed the absence of Suz12 protein, as well as a significant reduction in Ezh2 and global H3K27me3 levels in 4-hydroxytamoxifen–treated cells (Figure 1D).

Previous reports have demonstrated that the PRC2 catalytic subunit Ezh2 is essential for fetal hematopoiesis, but is dispensable for adult BM HSC function,¹⁸  due to compensation by Ezh1.²⁴  In contrast, a recent study suggests that Eed is not required for fetal hematopoiesis.²⁰  To assess the specific role of Suz12 in fetal hematopoiesis, Suz12^Δ/+/VavCre^T mice were bred with Suz12^fl/fl mice to restrict Suz12 deletion to hematopoietic cells. Genotyping revealed no Suz12^fl/Δ//VavCre^T mice were present at weaning (supplemental Table 2A), but these animals were present and viable at embryonic day E13.5 (supplemental Table 2B). Suz12^fl/Δ//VavCre^T embryos displayed significantly reduced fetal liver cellularity, and a concomitant decrease in the numbers of hematopoietic stem cells (lineage⁻ Sca-1⁺ c-Kit⁺ [LSK] CD48⁻ CD150⁺) and myeloid progenitor (common myeloid progenitor [CMP], granulocyte macrophage progenitor [GMP], and megakaryocyte erythroid progenitor [MEP]) populations (Figure 2A), as well as compromised erythroid development (Figure 2B; supplemental Figure 1).

To determine the role of Suz12 in adult hematopoiesis, we transplanted BM cells from Suz12^fl/fl/CreER^T2 or Suz12^+/+/CreER^T2 (test, CD45.2) mice, mixed with an equal number of cells from wt (competitor, CD45.1) marrow, into irradiated CD45.1/CD45.2 recipients (Figure 3A). Peripheral blood was analyzed 5 weeks after transplantation to establish the baseline chimerism, which was equivalent in both groups. A week later, tamoxifen or vehicle was administered via oral gavage and the percentage of test donor-derived cells in the peripheral blood was monitored at 6, 10, and 16 weeks after tamoxifen treatment. Recipients of the mixture of Suz12^+/+/CreER^T2 and WT BM receiving either tamoxifen or vehicle showed no significant changes in the level of chimerism at all 3 time points (Figure 3B, top panel). In contrast, recipients of the Suz12^fl/fl/CreER^T2 and wt BM mixture that were treated with tamoxifen showed progressive loss of test donor-derived CD45.2 cells over the 16-week period, an effect that was not evident in vehicle-treated mice (Figure 3B, bottom panel). Similarly, analysis of major hematopoietic organs 16 weeks after tamoxifen treatment revealed almost no contribution from test donor-derived CD45.2⁺ cells in tamoxifen-treated recipients of the Suz12^fl/fl/CreER^T2 and wt BM mixture (Figure 3D), including the BM LSK population (Figure 3C), suggesting that Suz12 is required for steady-state maintenance and function of adult HSCs. Analysis of gene expression in LSK cells and lineage⁻c-Kit⁺ cells sorted 7 days after tamoxifen treatment confirmed marked induction of Cdkn2a (p16 and p19), Cdkn1a, and other PRC2 target genes (supplemental Figure 2), which was consistent with the changes in gene expression observed in Eed knockout HSCs.²⁰  The strong induction of Cdkn2a and Cdkn1a likely contribute to the significant functional defects in Suz12-deficient HSCs.

Previous findings have shown that hematopoietic stem and progenitor cells are very sensitive to PRC2 dosage, with partial reduction^2,20,28  or elevation^25,26  allowing significantly increased HSC function. Moreover, mice carrying a heterozygous loss-of-function allele of Suz12 have elevated blood lymphocytes attributed to increased lymphoid progenitor activity.^2,44  Consistent with these findings, HSCs from tamoxifen-treated Suz12^fl/+/CreER^T2 mice showed enhanced self-renewal capacity, outcompeting HSCs from vehicle-treated Suz12^fl/+/CreER^T2 mice in serial competitive BM transplantation assays (supplemental Figure 3A). Suz12^Δ/+ mice had higher blood leukocyte counts (supplemental Figure 3B) and this phenotype was observed in various combinations of either germline or hematopoietic-specific heterozygous deletion of Suz12 (supplemental Table 3).

To specifically explore the role of Suz12 in lymphopoiesis, we conditionally inactivated Suz12 in the early stages of lymphoid development using Rag1Cre mice.⁴⁰  Adult Suz12^fl/fl/Rag1Cre^KI mice appeared normal and healthy, but displayed lymphopenia (supplemental Table 4) and had dramatically reduced spleen and thymus cellularity (supplemental Figure 4A). Genotyping of T-lymphocytes in the blood of Suz12^fl/fl/Rag1Cre^KI mice suggested that the residual lymphocyte production was attributable to cells that had escaped recombination of the Suz12 fl allele (supplemental Figure 4B). Analysis of BM showed that the numbers of HSC-enriched LSK cells, lymphoid-primed multipotent progenitors, common lymphoid progenitors (CLPs), and myeloid progenitors were normal in Suz12^fl/fl/Rag1Cre^KI mice (Figure 4A). However, B-lymphopoiesis was arrested at the pre-pro–B to pro-B transition in Suz12^fl/fl/Rag1Cre^KI mice as shown by the absence of CD19-expressing B-lymphocytes (Figure 4B). Similarly, Suz12^fl/fl/Rag1Cre^KI mice had defective T-lymphopoiesis (Figure 4C). The lymphoid defects were cell-intrinsic because in competitive transplantation assays production of lymphoid lineages from Suz12^fl/fl/Rag1Cre^KI, BM was outcompeted by WT marrow with effects on both B- and T-lymphopoiesis during the transition from the CLPs to pre-pro–B cells in BM, and the DN1/ETP cells in the thymus, respectively (Figure 4D). As expected, Suz12^fl/fl/Rag1Cre^KI BM contributed robustly to the myeloid lineage (Figure 4D).

We have previously shown that partial loss of Suz12 was able to ameliorate thrombocytopenia in c-Mpl^−/− mice.²  Although it was not possible in these previous studies to dissect contributions from stem cell and platelet-specific effects, here we used Pf4Cre^T transgenic mice³⁹  to specifically delete Suz12 in megakaryocytes and platelets. Both Suz12^fl/+/Pf4Cre^T and Suz12^fl/fl/Pf4Cre^T mice displayed normal numbers of platelets (supplemental Table 5). The number of BM megakaryocytes was also unchanged in Suz12^fl/fl/Pf4Cre^T mice, and platelet turnover, as assessed by thiazole orange staining, was normal (supplemental Figure 5). In addition, Lin⁻ progenitors isolated from the BM of Suz12^fl/fl/Pf4Cre^T mice were able to efficiently differentiate into mature megakaryocytes, in which near absolute deletion of Suz12 was evident, using a short-term in vitro culture system (supplemental Figure 6A). Suz12 deficient megakaryocytes had significantly reduced levels of H3K27me3 and displayed modest activation of PRC2 target genes (supplemental Figure 6B-D). Similarly, in Suz12^fl/fl/LysMCre^T mice, in which Suz12 was deleted specifically in granulocytes and macrophages,³⁸  no differences in the numbers of mature neutrophils, monocytes, or eosinophils were evident relative to control littermates (supplementary Table 6). Suz12^fl/fl/LysMCre^T BM cells were also able to differentiate into mature macrophages in vitro in the absence of Suz12. Genomic PCR analysis from BM-derived macrophages showed efficient recombination of the fl allele (supplemental Figure 7A), which was further supported by immunoblot analysis showing decreased levels of Suz12, Ezh2, and global H3K27me3 (supplemental Figure 7B).

The recent discoveries of recurring mutations in PRC2 core components EZH2, EED, and SUZ12 in various hematologic malignancies^30-35  have underscored the need for more precise understanding of PRC2 function in normal and malignant hematopoiesis. Although a number of studies have documented roles for PRC2 in murine hematopoiesis, these reports paint a complex picture with seemingly diverse effects of manipulating expression of different PRC2 components, cell type-specific effects and profound biological consequences of apparently subtle alterations in PRC2 complex activity. Contributing to this complexity is the functional redundancy between Ezh2-PRC2 and Ezh1-PRC2. Loss of Ezh2-PRC2 can be compensated by Ezh1-PRC2, and vice versa, in HSCs and in other stem cells.^{14,17,18,24,45} Ezh2 expression in HSCs declines with age, whereas Ezh1 expression increases, and homozygous deletion of Ezh2 in adult BM appears to have minimal impact on HSC function mainly due to compensation by Ezh1-PRC2,^18,20  which appears to be the dominant enzymatic component for PRC2 in adult HSCs. Therefore, it is not surprising that Ezh1-deficient HSCs are unable to engraft in lethally irradiated mice, and mice with hematopoietic-specific deletion in Ezh1 eventually develop progressive BM failure.²⁴  To circumvent the complexities of Ezh1/Ezh2 redundancy, we generated Suz12 conditional knockout mice for the inactivation of PRC2. In adult chimeric mice with a mixture of Suz12-conditional and WT hematopoiesis, induced deletion of Suz12 resulted in a near complete absence of Suz12-null hematopoietic cells in the BM and peripheral organs accompanied by a precipitous of loss of Suz12-deleted HSCs, establishing a critical requirement for Suz12 in the maintenance of adult HSCs and their capacity to generate blood. Suz12-deficient HSCs activate expression of classic PRC2 target genes, including Cdkn2a (p16 and p19) and Cdkn1a, which likely explains their marked functional impairment. Our data are consistent with that of a recent study using Eed conditional knockout mice,²⁰  emphasizing a critical role for PRC2 in adult HSCs.

Although our results establish an important consensus with previous studies on the role of PRC2 in adult HSCs, differences remain regarding the role of PRC2 in fetal hematopoiesis. Although Ezh1 appears to be dispensable for fetal hematopoiesis,²⁴  Mochizuki-Kashio et al¹⁸  found that deletion of Ezh2 (driven by Tie2Cre) led to compromised HSC function and defective erythroid differentiation at E12.5. However, a later study by Xie et al²⁰  suggested that deletion of Ezh2 via VavCre conditional targeting had no obvious fetal hematopoietic defect, and the authors concluded that Ezh2 is dispensable for both adult and fetal hematopoiesis. The use of different cre-recombinases and Ezh2 conditional alleles could be the simplest explanation for the contradictory findings. Xie et al²⁰  also found that fetal hematopoiesis in embryos with VavCre-mediated loss of Eed was undisturbed, but that Eed was critical after birth, with neonatal Eed^−/− mice displaying pancytopenia, anemia, and BM failure. In contrast, we show here that VavCre-mediated deletion of Suz12 during fetal hematopoiesis significantly reduced the number of long-term HSCs and myeloid progenitors, as well as severely compromised erythroid development. These data prompt the surprising conclusion that fetal HSC are highly dependent on Suz12 without the requirement for Eed. New evidence has recently emerged that supports the existence of a noncanonical complex that contains Ezh1 and Suz12 but lacks Eed.⁴⁶  A second, and more intriguing hypothesis is that Suz12 may regulate fetal hematopoiesis independent of Ezh1/Ezh2. Support for PRC2-independent actions of core PRC2 components has emerged with a recent report that Ezh2 modulates transcription in a non–PRC2-dependent fashion in prostate cancer depending on its phosphorylation state.⁴⁷ 

Increased blood lymphocyte numbers were observed in Suz12 heterozygous mice generated by ubiquitous (DelCre) or hematopoietic-specific (VavCre) cre-recombinase–mediated gene deletion, consistent with our previous study in mice carrying the hypomorphic Suz12^Plt8 allele.⁴⁴  Although previous studies have clearly demonstrated impaired lymphopoiesis as a result of complete loss of PRC2 core components, we have extended this finding by showing that this effect is due to PRC2 loss in specific progenitor cells, independent of actions in HSCs. Although PRC2 inactivation in maturing lymphocytes had little impact on B- or T-cell numbers,^19,23  we demonstrate here that Rag1Cre-mediated inactivation of Suz12 in CLPs results in severe defects in both B- and T-lymphoid development. The contrasting increase in self-renewal of B-lymphoid precursors in mice with partial loss of PRC2 function⁴⁴  is consistent with a specific role for PRC2 in lymphoid progenitor cells and further suggests that the influence of PRC2 on lymphoid development is highly sensitive to gene dosage, as it is in HSCs.

PRC2 appears to be dispensable in maturing granulocytes and macrophages, as well as in megakaryocytes and platelets. Although we have not identified the precise stage in myelopoiesis at which dependency on Suz12/PRC2 is lost, this insensitivity to PRC2 loss in the myeloid lineages is intriguing, and is in stark contrast to the absolute requirement for PRC2 during lymphopoiesis. The inability of lymphoid cells to tolerate deletion of Suz12 is likely due to the involvement of PRC2 at a number of key points during differentiation. For example, Ezh2 is required for effective TCR-signaling in developing thymocytes²³  and for rearrangement of the immunoglobulin heavy chain (IgH) gene in B-lymphoid progenitors.¹⁹  Indeed, committed B-lymphocytes that have completed immunoglobulin rearrangement tolerate loss of Ezh2.¹⁹  It may be that repressive marks established at key target genes by PRC2 in early hematopoietic stem and progenitor cells are sufficient to enable myeloid differentiation, but not lymphoid differentiation, or it may be that there is compensation by other epigenetic factors that are specific to the myeloid lineage. Thus, in addition to a critical requirement for PRC2 in HSCs, our data establish additional lineage-specific roles in committed hematopoietic cells.

Although a profound loss of PRC2 activity results in defective adult HSC function, we and others have previously shown that a partial reduction in the activity of PRC2 via genetic manipulation of Eed, Ezh2, or Suz12 results in enhanced HSC self-renewal in serial transplantation assays,^2,20,27,28  which are findings we have confirmed here in mice with induced heterozygous deletion of Suz12 in adult mice. Interestingly, Ezh2-overexpressing murine HSCs display increased proliferation and self-renewal, and eventually caused a myeloproliferative disorder similar to that found in humans.^25,26  Collectively, these studies establish that PRC2 dosage needs to be tightly regulated for hematopoietic homeostasis, with slight perturbations resulting in pathological consequences. Indeed, in myeloid malignancies, the observation that heterozygous loss-of-function mutations in EZH2, EED, and SUZ12 are more frequent than homozygous deletion^30,32,34  implies that PRC2 depletion favors the initial clonal expansion of leukemic stem cells, whereas complete PRC2 loss abrogates self-renewal and proliferation, as was observed in MLL-AF9–driven leukemic stem cells.⁴⁸  With PRC2 increasingly being considered as a potential target for cancer therapy, and EZH2 inhibitors currently under evaluation,^49,50  the biological consequences of altered PRC2 activity, as well as the cell and developmental stage-specific effects evident from mouse genetic studies, such as those presented here, will need to be carefully considered in any therapeutic context.

The authors thank Stefan Glaser for Rosa26CreER^T2 mice and Stephen Nutt for Rag1Cre mice; Takashi Ushiki, Ashley Ng, and Rhys Allan for helpful discussions; Jason Corbin, Jasmine McManus, Janelle Lochland, Lauren Wilkins, Keti Stoev, Elizabeth Kyran, Hayley Backhaus, Carolina Alvarado, and Liana Mackiewicz for skilled technical assistance and animal husbandry; and Dr Lachlan Whitehead for assistance with immunofluorescence image analysis.

This work was supported by program (1016647 and 575500) and project (1011663) grants, fellowships (W.S.A., I.J.M., M.E.B.), and Independent Research Institutes Infrastructure Support Scheme grant (361646) from the Australian National Health and Medical Research Council, fellowships from the Australia Research Council (M.E.B.), an Australian Postgraduate Award (S.C.W.L.), the Carden Fellowship (D.M.) of the Cancer Council, Victoria, the Australian Cancer Research Fund, and Victorian State Government Operational Infrastructure Support.

Contribution: S.C.W.L., I.J.M., and W.S.A. designed experiments and wrote the manuscript; S.C.W.L., S.M., C.H., M.K., M.L., E.C.J., L.D.R., D.M., and I.J.M. performed experiments and analyzed data; M.E.B. and S.A.K. provided critical conceptual input and feedback on data interpretation and the manuscript; and I.J.M. and W.S.A. supervised research.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Ian J. Majewski, Cancer and Haematology Division, The Walter and Eliza Hall Institute of Medical Research, 1G Royal Parade, Parkville, VIC 3052, Australia; e-mail: majewski@wehi.edu.au.