Ezh2 spares KitL from the cutter

In this issue of Blood, Neo et al identify a cell-extrinsic role for the epigenetic regulator enhancer of zeste 2 (Ezh2) in maintaining fetal liver erythropoiesis.¹  Using conditional Ezh2 deletion approaches, these investigators demonstrate that Ezh2 maintains appropriate levels of membrane-bound Kit ligand (mKitL) by repressing endothelial expression of matrix metallopeptidase 9 (Mmp9). These data provide a striking demonstration of how epigenetic factors not only act as cell-intrinsic regulators of hematopoietic stem cell (HSC) fate but can also impact cell-extrinsic signals provided by the hematopoietic niche.

Ezh2 is a key enzymatic component of the multimeric polycomb repressive complex 2 (PRC2), which also includes the subunits Eed, Suz12, and RbAp48.²  Ezh2 (and its paralog, Ezh1) is a histone methyltransferase, an epigenetic “writer” enzyme that methylates histone 3 at lysine 27 (H3K27) and leads to chromatin silencing.²  Hence, regulation of this histone mark is crucial for governing the balance between self-renewal and differentiation in a number of stem cell types.³  Given its critical role as a transcriptional repressor, it comes as little surprise that EZH2 is mutated or overexpressed in a wide array of cancers, including hematological malignancies. Loss-of-function mutations in EZH2 have been identified in myelodysplastic syndrome, myeloproliferative neoplasms (including juvenile myelomonocytic leukemia), and some T-cell acute lymphoblastic leukemias.^4,5  On the other hand, an array of solid tissue malignancies and some B-cell lymphomas feature gain-of-function mutations or EZH2 overexpression, which often correlates with poor disease prognosis.⁴  This suggests EZH2 plays a critical role in regulating hematopoiesis in normal and disease contexts.

In their study, Neo et al use Vav-Cre hematopoietic-specific and Tie2-Cre hematopoietic/endothelial–specific Cre recombinases to delete Ezh2 in mouse embryos. Although germ line Ezh2 deletion leads to embryonic lethality by embryonic day 7.5, prior studies have shown that hematopoietic Ezh2 deletion using the Vav-Cre system, or deletion in hematopoietic adult mice using an interferon-inducible Mx-Cre allele, leads to relatively minor hematopoietic effects, particularly impaired lymphoid differentiation.⁶  Such a mild phenotype has been largely attributed to functional redundancy between Ezh1 and Ezh2,⁷  as deletion of the central PRC2 component Eed using Vav-Cre leads to profound hematopoietic defects due to a failure of HSC to properly differentiate.⁶  Hence, while cell-intrinsic PRC2 is absolutely required for proper hematopoiesis, Ezh2 is largely dispensable. In contrast, studies using the Tie2-Cre and tamoxifen-inducible full-body Rosa26-Cre-ERT allele to delete Ezh2 yield an embryonic lethal phenotype due to severe anemia and reduction in Lin⁻c-Kit⁺Sca-1⁺ (LSK) populations, which are enriched for multipotent progenitors and HSCs.⁸  However, HSC isolated from the fetal livers of these mice were capable of blood reconstitution following transplantation, leading to the conclusion that Ezh2 may be functioning non–cell autonomously. Against this background, Neo et al systematically reproduce these findings in their Vav-Cre and Tie2-Cre systems, confirming functional hematopoiesis in Vav-Cre::Ezh2^Δ/Δ embryos, while Tie2-Cre::Ezh2^Δ/Δ embryos develop anemia leading to lethality at embryonic day 13.5, despite relatively normal stromal cell numbers and elevated levels of erythropoietin (Epo) and KitL mRNA in the fetal liver.

A particularly striking observation is the functional distinction between soluble and membrane-bound KitL in fetal hematopoiesis. Strikingly, mKitL levels (but not total KitL levels) are significantly decreased in Tie2-Cre::Ezh2^Δ/Δ fetal livers. Neo et al also show using an in vitro coculture system of wild-type LSK cells with endothelial cells from Tie2-Cre::Ezh2^Δ/Δ and mKitL-deficient mice that mKitL is obligatory for erythroid expansion in the fetal liver. Indeed, this finding reiterates a sometimes-overlooked point that cytokines often exist in a membrane-bound form that may have distinct biological effects and/or target cells relative to the soluble form.⁹  Such membrane-bound cytokines are not always detectable and/or distinguishable by traditional enzyme-linked immunosorbent assays. mKitL, much like other cytokines with membrane-bound forms such as tumor necrosis factor or FasL, induces a stronger and more sustained level of signaling than the soluble form. Taken together, these results highlight the importance of mKitL in fetal liver erythropoiesis; notably, as fetal HSC numbers are not depleted in Tie2-Cre::Ezh2^Δ/Δ mice, they also imply that different hematopoietic stem and progenitor populations have distinct requirements for Kit signaling. As deletion of PRC2 components can also result in loss of c-Kit expression and failure to correctly generate certain lineages,⁶  future studies may focus on the relative importance of c-Kit signaling in lineage specification and how PRC2 regulates KitL-c-Kit pathway activity in development and disease.

A key highlight of this study is the link between Ezh2, regulation of Mmp9 expression and mKitL proteolytic cleavage into the soluble form by Mmp9. Prior analysis of Tie2-Cre::Ezh2^Δ/Δ embryos identified defects in vascular integrity related to elevated Mmp9 expression and activity; this defect could be corrected by genetic inactivation of Mmp9 in Tie2-Cre::Ezh2^Δ/Δ embryos.¹⁰  Here, Neo et al demonstrate that pharmacological Mmp9 blockade rescues mKitL surface expression and can reverse the erythropoiesis defect in their coculture system. Moreover, chromatin immunoprecipitation (ChIP) analysis of a fetal endothelial cell line demonstrated that Ezh2 binds to the Mmp9 promoter. Importantly, the authors show that treatment of such cells with a pharmacological Ezh2 inhibitor reduces Ezh2 binding and H3K27 trimethylation at the Mmp9 promoter. Similarly, histological analysis Tie2-Cre::Ezh2^Δ/Δ::Mmp9^−/− embryos show rescue of erythropoiesis. Altogether, these findings identify a critical regulatory circuit by which Ezh2 suppresses Mmp9 expression, thereby limiting cleavage of mKitL. As demonstrated above, proper regulation of this circuit is crucial for maintaining erythroid development in embryonic life.

Because of its oncogenic role, EZH2 inhibitors are currently in clinical trials.⁴  However, these inhibitors carry some problematic hematopoietic side effects, including anemia. Along these lines, Neo et al show that EZH2 inhibition in an adult endothelial cell line leads to Mmp9 overexpression and reduced erythropoiesis. Hence, more detailed preclinical investigations using in vivo systems may be desirable to better understand how off-target misregulation of Mmp9 and/or other Ezh2 targets may contribute to bone marrow suppression and whether modulation of MMP9 activity or careful titration of EZH2 inhibitors can minimize treatment-related comorbidity. Altogether, this study provides a striking window into how Ezh2 functions as a cell-extrinsic hematopoietic regulator and prompts further examination of how epigenetic modifiers mediate the interplay between HSCs and their niche.