Ezh1 Is Required for Hematopoietic Stem Cell Maintenance and Prevents Senescence-Like Cell Cycle Arrest

Cell Stem Cell Article

Ezh1 Is Required for Hematopoietic Stem Cell Maintenance and Prevents Senescence-like Cell Cycle Arrest

Isabel Hidalgo,1,6 Antonio Herrera-Merchan,1,6 Jose Manuel Ligos,2 Laura Carramolino,3 Javier Nun˜ ez,4 Fernando Martinez,4 Orlando Dominguez,5 Miguel Torres,3 and Susana Gonzalez1,* 1Stem Cell Aging Group 2Cellomics Unit 3Genetic Control of Organ Development and Regeneration 4Bioinformatics Unit Centro Nacional de Investigaciones Cardiovasculares (CNIC), E-28029 Madrid, Spain 5Genomics Unit, Centro Nacional de Investigaciones Oncologicas (CNIO), E-28029 Madrid, Spain 6These authors contributed equally to this work *Correspondence: [email protected] http://dx.doi.org/10.1016/j.stem.2012.08.001

SUMMARY known about how epigenetic mechanisms dictate the activities of Polycomb factors to ensure blood homeostasis. Polycomb group (PcG) proteins are key epigenetic PcG proteins assemble in multimeric complexes and induce regulators of hematopietic stem cell (HSC) fate. The transcriptional repression of target genes through chromatin PcG members Ezh2 and Ezh1 are important determi- changes such as methylation of lysine 27 on histone 3 (H3K27). nants of embryonic stem cell identity, and the tran- Biochemical analysis revealed that PcG proteins assemble in at script levels of these histone methyltransferases least two complexes, Polycomb repressive complexes 1 and 2 are inversely correlated during development. How- (PRC1 and PRC2). Studies of mouse mutants showed that these factors are essential for normal development, with deletion of ever, the role of Ezh1 in somatic stem cells is largely Polycomb genes resulting in prenatal or early postnatal death, unknown. Here we show that Ezh1 maintains repopu- depending on inactivated gene and how the inactivation is lating HSCs in a slow-cycling, undifferentiated state, achieved (Sauvageau and Sauvageau, 2010). The PRC2 complex protecting them from senescence. Ezh1 ablation contains one of two homologous enhancer of zeste (Ez) pro- induces signiﬁcant loss of adult HSCs, with concom- teins (Ezh1 or Ezh2), which contain near identical catalytic SET itant impairment of their self-renewal capacity due domains (Laible et al., 1997). Ezh1 and Ezh2 are chromatin-modi- to a potent senescence response. Epigenomic and fying histone lysine methyltransferases (Margueron and Rein- gene expression changes induced by Ezh1 deletion berg, 2011). Most research on H3K27 has focused on the in senesced HSCs demonstrated that Ezh1-medi- trimethylated form (H3K27me3), while the monomethylated and ated PRC2 activity catalyzes monomethylation and dimethylated forms (H3K27me1 and H3K27me2, respectively) dimethylation of H3K27. Deletion of Cdkn2a on the remain poorly characterized (Margueron and Reinberg, 2011). Although Ezh2 knockout in embryonic stem cells (ESCs) reduces Ezh1 null background rescued HSC proliferation H3K27 dimethylation and trimethylation, H3K27 methylation at and survival. Our results suggest that Ezh1 is an PcG target promoters is not abolished. This has been attributed important histone methyltransferase for HSC mainte- to the activity of Ezh1, which forms an alternative PRC2 com- nance. plex (Shen et al., 2008) and has distinct chromatin-binding prop- erties (Margueron et al., 2008). The importance of the H3K27me3 mark has been characterized through conditional targeting of INTRODUCTION Ezh2 in somatic stem cells in the brain, pancreas, epidermis, and hematopoietic system (Chen et al., 2009; Ezhkova et al., Epigenetic regulation of chromatin structure is thought to con- 2009; Mochizuki-Kashio et al., 2011; Pereira et al., 2010; Su trol cell memory, and Polycomb group (PcG) proteins are impor- et al., 2003). Double knockout of Ezh1 and Ezh2 in mouse skin tant components of diverse epigenetic phenomena, including abolishes the H3K27me3 modiﬁcation and severely compro- genomic imprinting, cell fate determination, stem-cell plasticity mises formation and maintenance of hair follicles (Ezhkova and renewal, and cancer development (Goldberg et al., 2007). et al., 2011). However, the role of PRC2-Ezh1 in the maintenance PcG-mediated epigenetic alteration of hematopoietic stem cells of HSCs is unknown. (HSCs) is considered a driving force behind many age-related Cell senescence is a permanent cell-cycle arrest that is re- HSC changes and is often dysregulated in human malignancies fractory to growth factors and other proliferation signals. Un- (Sauvageau and Sauvageau, 2010). Protection of the transcrip- like apoptotic cells, which rapidly disintegrate, senescent cells tional ‘‘stemness’’ network is thus essential for the maintenance remain viable in long-term culture and are found with increasing of a healthy HSC compartment throughout life. However, little is frequency in aged tissues and at sites of age-related pathology,

Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. 649 Cell Stem Cell HSC Senescence Protection by Ezh1

A 4 young HSCs B C Ezh1 wt locus aged HSCs 14 15 16 17 18 3 19 20 21

Ezh1 Additional Additional 2 HindIII ScaI ScaI * HindIII * Ezh2 1 14 15 16 17 18 Neo 19 20 21 Relative Ezh1 levels mRNA WB Suz12 0 Ezh1 targeting vector 2 Eed 1 β-actin 0 13 kDa

Relative Ezh2 levels mRNA wt HSCs (LSK) Lin- LSK LT-HSC MP CLP 350 pb 8.5 kDa 300 pb 7.6 kDa D HSCs E 5.7 kDa + - (LSKCD150 CD48 ) Ezh1 Ezh2 WB H3K27me3 H3K27me2

H3K27me1

H3 HSCs - ΔΔ Ct (Ezh1-KO vs. Control) INK4a INK4b (LSKCD150+CD48-) p15 Ezh2 Ezh1 p16 ARF

F H&E Ki67 Ezh1 Bone marrow Ezh1-KO Control

G Control Ezh1 MP LSK MP LSK MPP 46.7 % MPP 32.2 %

9.2 % CLP 3.7 % LT LT ST ST 38.8 % 43.1 % 23.4 % 26.2 % cKit cKit CD150 CD150 CD135 CD135 Sca1 CD34 Sca1 CD34 GMP CD48 CD48 CLP GMP 15.3 % 18.8% CLP MEP 32.9 % MEP 23.6 % CMP CMP 41.6 % 47.3 %

CD16/32 68.4%

CD16/32 63.7 %

CD34 CD127 CD34 CD127

H Control I 15 Control Ezh1-KO Ezh1-KO

cells) 10 6

(x10 5 Total in BM Total Total in BM cells Total 0 LSK LT ST MPP CD150+ CLP CMP MEP GMP CD48-

Figure 1. Characterization of Primitive Hematopoietic Populations in Ezh1-KO Mice (A) Expression of total Ezh2 and Ezh1 mRNA was measured by qRT-PCR in sorted bone marrow (BM) subpopulations from young (8-week-old) and aged (50-week-old) WT mice: lineage negative (Lin); Lin, Sca1+, c-Kit+ (LSK); long-term repopulating HSCs (LSKCD34lowFlk-2low; LT-HSCs), myeloid progenitors (LinKit+Sca1; MPs), and common lymphoid progenitors (LinKitlowSca1lowCD127+; CLPs). Results are standardized to b-actin and are expressed as the fold change of the indicated subpopulations from aged mice compared with the same gated subpopulations from young mice (means ± SD; n = 12, *p < 0.05).

650 Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. Cell Stem Cell HSC Senescence Protection by Ezh1

including preneoplastic lesions. Cells senesce in response to the regulation of histone H3K27 methylation. Our results indicate many potentially oncogenic stressors, including severe DNA that Ezh1 promotes a slow cycling state in primitive HSCs and damage, dysfunctional telomeres, chromatin alterations, oxida- prevents differentiation and senescence of adult HSCs. tive stress, and strong mitogenic signals such as those delivered by some oncogenes (Collado et al., 2007). Induction of the RESULTS senescence program normally requires activation of tumor suppressor pathways, including the p19ARF (p14ARF in humans), p53, Expression of the Histone Methyltransferase Ezh1 and RB-p16INK4a signaling cascades, and the senescence pro- Increases during HSC Aging gram is now regarded as a potent cell-autonomous barrier to Since expression of several PcG genes involved in chromatin tumorigenesis (Prieur and Peeper, 2008). Accordingly, loss of organization becomes dysregulated during the aging of HSCs the senescence response increases the incidence of cancer in (Chambers et al., 2007b), we hypothesized that histone methyl- humans and mice (Nardella et al., 2011). Moreover, cell se- transferase activity might be coordinately distributed between nescence also entails increased secretion of proinflammatory Ezh1 and Ezh2 in young and old HSC populations. Ezh1 and proteins, inducing the senescence-associated (SA) secretory Ezh2 mRNAs were detected in primitive long-term HSCs (LT- phenotype (SASP), and might contribute significantly to chronic HSCs) isolated from the bone marrow (BM) of young (8-week- inflammation (Freund et al., 2010). old) wild-type C57BL/6 mice; but while levels of Ezh1 mRNA Epigenetic mechanisms play an important role in senescence. were 3-fold higher in LT-HSCs isolated from BM of 50-week- In senescing human fibroblasts, heterochromatic regions con- old mice, Ezh2 transcript levels did not change with age (p < dense to form senescence-associated heterochromatic foci 0.05; Figure 1A). A similar divergence was seen with protein that limit the DNA damage response (Di Micco et al., 2011; Narita expression (Figure 1B), as noted in other studies on human et al., 2003). Expression of Jhdm1b, a demethylase specific for and mouse HSCs (Mochizuki-Kashio et al., 2011; Wagner the H3K36me2 modification, causes cell immortalization or et al., 2009) and mouse kidney (Margueron et al., 2008). Protein leukemic transformation depending on its demethylase activity levels of the other PRC2 components, Suz12 and Eed, did not on p15INK4b, and its knockdown results in cell senescence (He change with age (Figure 1B). Given this pattern of Ezh1 expres- et al., 2008). In human and mouse fibroblasts, the Cdkn2a region sion in repopulating HSCs, we hypothesized that Ezh1 might act (encoding p16INK4a and ARF) is repressed by PcG-mediated primarily as a homeostatic regulator of HSCs. H3K27me3, and the repressive mark is lost during oncogene- To test this, we generated a conditional knockout Ezh1 allele in induced senescence, resulting in Cdkn2a expression (Bracken which exons encoding the SET catalytic domain were flanked et al., 2007). with loxP sequences (ckoEzh1fl/fl)(Figure 1C). To induce spe- The incomplete loss of H3K27me3 in Ezh2/ ESCs, the rela- cific Ezh1 deletion in the hematopoietic system, we crossed tively mild phenotypes of tissues lacking Ezh2, and the inverse ckoEzh1fl/fl mice with vavCre transgenic mice. From here on, expression patterns of Ezh1 and Ezh2 during development Ezh1fl/fl-vavCretg/+ knockout mice and their Ezh1fl/fl-vavCre+/+ together suggest that Ezh1- and Ezh2-mediated H3K27 methyl- wild-type (WT) littermates are referred to as Ezh1-KO and ation are differentially coordinated during tissue development Control. Analysis of primitive HSCs (LSKCD150+CD48; 20- and homeostasis. It is therefore important to determine the role week-old) showed that complete Ezh1 deletion correlated with of Ezh1-containing PRC2 complexes in HSC maintenance. We increased expression of p16INK4a, ARF, and p15INK4b (p < report here that Ezh1 is essential for adult hematopoiesis through 0.001), whereas expression of Ezh2 was unaffected (Figure 1D).

(B) Representative western blots (WBs) showing the expression of Ezh1, Ezh2, Suz12, Eed, and b-actin in HSCs (LSK cells) from young (8-week-old) and aged (50-week-old) WT mice. (C) Top: Ezh1 genomic locus and the Ezh1-targeted locus before and after Cre-mediated recombination. Red arrows mark the loxP sequences. The restriction enzyme-digested DNA fragments recognized by the 30and 50 probes are indicated. Bottom: Southern blots show HindIII-digested genomic DNA isolated from both ckoEzh1+/+ and ckoEzh1fl/+ ESCs hybridized with the 30and 50 probes (left panels). Genomic PCR analysis of ckoEzh1+/+, ckoEzh1fl/fl, and ckoEzh1fl/+ mice are shown in the right panels. See also Table S1. (D) qRT-PCR analysis of the mRNA expression of Ezh1, p16INK4a, ARF, and Ezh2 in HSCs (LSKCD150+CD48) from Ezh1-KO and Control mice. Data are standardized to b-actin levels, and are expressed as the fold change of the indicated subpopulation from Ezh1-KO mice relative to Control mice (means ± SD; n = 12, **p < 0.001, *p < 0.05). See also Table S2. (E) Representative WBs showing the protein levels of Ezh1, Ezh2, H3K27me3, H3K27me2, H3K27me1, and histone3 (H3) in HSCs (LSKCD150+CD48) from Ezh1-KO and Control mice. (F) Histological and immunostaining characterization of paraffin-embedded BM sections from Ezh1-KO mice and Control littermates, using hematoxylin and eosin staining (H&E), and anti-Ki67 and anti-Ezh1 antibodies. Scale bars represent 500 and 50 mm for H&E and 100 mm for immunohistochemistry. (G) Representative FACS profiles of BM cells from Ezh1-KO mice and Control littermates. BM cells were immunostained with antibodies to hematopoietic lineage markers, c-Kit and Sca1. FACS plots were gated on the LSK supopulation (orange gate), which was subclassified for expression of CD34 and CD135 to give the percentages of MPPs, LT-HSCs, and ST-HSCs, and for expression of CD150+ and CD48 to give the percentage of the LSK-SLAM population. The CLP fraction (green gate) was gated on LKlowSlowCD127+. Myeloid progenitors, defined as the LK+S (LKS) population (MP, blue gate), were subclassified into common myeloid progenitors (CMPs), granulocytic macrophage progenitors (GMPs), and megakaryocytic-erythroid progenitors (MEPs) based on CD16/32 and CD34 expression. At least five mice per genotype were tested, over at least two separate experiments. Numbers indicate percentages of the total nucleated BM cell population, except for the subclassification of LSK cells, where numbers indicate the percentage of the total LSK subpopulation. (H and I) Absolute numbers per femur of LSK, LT, ST, MPP, and LSK-SLAM primitive populations (H), and CLP, CMP, MEP, and GMP subpopulations (I) (as described in Figure 1G) in BM from Control and Ezh1-KO mice. Data are means ± SD (n = 10, *p < 0.05, **p < 0.001). See also Figure S1.

Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. 651 Cell Stem Cell HSC Senescence Protection by Ezh1

A F Control Ezh1-KO 100 CD45.1

CD45.2 SA 80 㻙 β Gal 60

40 78% 8.1% 20 * p16 % Engraftment in PB cells 0 CD45.245.2 CD45.25.2

Control Ezh1 INK4a B Control

MPP 71% MPP 100 Control

80 Ezh2

CD135 LT ST 60 CD34 CD45.2 40

20 % Engraftment in BM cells 79% 81% 0 LT ST LSK LT ST MPP

CD45.2 Ezh1-KO Bmi1

MPP 9.1% MPP 100

80 Ezh1-KO CD135 LT ST CD45.2 60 CD34 40

0.0% 2.1% % Engraftment in BM cells 20 0 LT ST LSK LT ST MPP Dcr2

CD45.2

C D 25 50 p15 20 40 INK4b

15 30 * 10 20

0 Proliferative rate (%) 0 LT ST MPP LSK CLP MP AnnexinV+ /Hoechst- cells (%) γ 㻙

Control H2AX E LT 12.5 5.6 Control: Controloll Ezh1-KO Ezh1-KOKKOO

ST 17.9 Control 4.7 Ezh1-KO 45 LSK Ezh1-KO Control Ezh1-KO 30 cells (%)

MPP + 9.6 8.4 Control Ezh1-KOControlContro l 15

Ezh1-KOO FDG CD135 12

C 0 CD34 C FDG 12 LSK LT ST MPP

Figure 2. Ezh1 Deletion Induces Cellular Senescence (A and B) Average chimerism in posttransplant competitive recipients transplanted with 1:1 ratios of donor and competitor cells, analyzed by the contribution of donor-derived (CD45.2) cells to peripheral blood (PB) cells (A), and the indicated subclassiﬁed LSK populations (B). FACS plots show a representative experiment. The bar graphs show the percentages of the indicated populations 12 weeks posttransplant. Data are means ± SD (n = 8, *p < 0.05, **p < 0.001). See also Figure S2. (C) Percentage of early apoptotic cells (annexinV-Hoechst staining) in the indicated populations. Data are means ± SD (n = 6, *p < 0.05).

652 Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. Cell Stem Cell HSC Senescence Protection by Ezh1

The overall level of H3K27me3 was not altered in Ezh1-KO tent with a requirement for Ezh1 in the self-renewal and mainte- mice; however, the levels of H3K27me2 and H3K27me1 were nance of adult HSCs. markedly lower (Figure 1E). HSC Senescence Is Increased in Ezh1-KO BM Ezh1 Deletion Causes BM Failure Analysis of freshly isolated primitive stem- and progenitor-cell We next examined the effect of Ezh1 deficiency on primitive BM populations revealed no difference between genotypes in HSC subpopulations in BM. Immunostaining for Ezh1 in BM the frequency of annexinV+/Hoechst cells, indicating that pro- sections revealed specific vavCre-mediated Ezh1 deletion in grammed cell death does not underlie the disappearance of hematopoietic cells, while fibrovascular stroma (sinusoid endo- Ezh1-KO LSK HSCs from the BM or their inability to compete thelial cells and fibroblasts) remained positive, a finding also with WT cells in transplantation experiments (Figure 2C). We reflected in weak Ezh1 protein expression in whole Ezh1-KO therefore analyzed whether Ezh1 deletion affects the prolifera- BM (Figure 1F; data not shown). Histological analysis detected tion of primitive hematopoietic cells in vivo. Examination of severe femoral hypoplasia in BM of adult Ezh1-KO mice (Fig- cell-cycle status by BrdU incorporation, normalized for total ure 1F), reflecting a 2-fold lower BM cellularity than in Controls numbers, showed that the LT- and ST-HSC/MPP Ezh1-KO pop- (p < 0.001; Figure S1A available online). Nonlymphoid tissues ulations cycled less than corresponding Controls (p < 0.05; Fig- showed no significant differences in Ezh1 staining between ure 2D and data not shown). genotypes (Figure S1B). Furthermore, adult Ezh1-KO animals Given that hematopoietic-specific Ezh1 inactivation trig- had a smaller HSC-enriched LSK pool than Controls (p < gered accumulation of transcripts for p16INK4a and ARF (Fig- 0.05; Figures 1G and 1H). Analysis of LSK subfractions in ure 1D), which are critical activators of the senescence pro- Ezh1-KO mice showed significantly lower frequencies (p < gram (Lowe et al., 2004), we tested whether Ezh1 deletion 0.001) and absolute numbers (p < 0.001) of all three primi- triggered cellular senescence in HSCs. Consistent with the anti- tive repopulating LSK populations: LT-HSCs, short-term HSCs senescence role of other Polycomb proteins such as Bmi1 (ST-HSCs), and multipotent progenitors (MPPs) (Figures 1G (Jacobs et al., 1999), depletion of endogenous Ezh1 in isolated and 1H). Further enrichment of the LSK population with SLAM mouse embryonic fibroblasts (MEFs) resulted in increased tran- markers (CD48 and CD150+) confirmed reduction of the script and protein expression of p16INK4a and ARF (p < 0.001; LSK-SLAM population in Ezh1-KO BM (p < 0.001; Figures 1G Figure S2B), as well as increased activity of senescence- and 1H). associated SA-b-galactosidase (SA-b-gal) (Figure S2C). Con- The more mature lymphoid progenitor pool was also signifi- versely, reexpression of Ezh1 should recapitulate the immortal- cantly smaller in KO mice (p < 0.05; Figures 1G and 1I), while ization phenotype of p16INK4a/ARF null MEFs (Gonzalez et al., frequencies and numbers of myeloid progenitors and colony- 2006). Evaluation of this in colony-formation assays in Ezh1- forming cells (CFCs) were unaffected (Figure 1I and Figure S1C). deleted MEFs confirmed that Ezh1 reexpression immortalizes Moreover, pyrorin Y staining of LSK CD34low cells showed a fibroblasts by downregulating expression of p16INK4a and ARF smaller pool of quiescent (G0) HSCs in Ezh1-KO mice (p < (Figure S2C). 0.05; Figure S1D). Analysis of BM cells for the side popula- These in vitro results suggested that adult Ezh1-KO BM might tion (SP) phenotype, a marker of quiescent HSCs in adult BM, have a senescent phenotype. To test this in vivo, we first revealed 2-fold fewer SP cells in Ezh1-KO mice (p < 0.05; measured SA-b-gal activity in primitive HSC subpopulations

Figure S1E). by FACS-based assay using the fluorogenic substrate C12FDG To determine the long-term reconstitution capacity of Ezh1- (Debacq-Chainiaux et al., 2009). Approximately 34% of Ezh1- deficient BM cells, we crossed the ckoEzh1fl/fl line with KO LT-HSCs were positive for SA-b-gal, comparable to Ras- Rosa26Cre-ERT2 mice to generate Rosa26Cre-ERtg/tg;Ezh1fl/fl induced MEF senescence (Kuilman et al., 2010), while fewer (Ezh1-KO-ER) and Rosa26Cre-ERtg/tg;Ezh1+/+ (Control-ER) than 7% of HSCs from Control littermates stained positive (p < mice. After tamoxifen induction of Cre-mediated Ezh1 deletion, 0.001) (Figure 2E and Figure S2D). In contrast, the frequency BM cells from these mice were used in competitive transplanta- of MPP senescence was similar in WT and Ezh1-KO BM (Fig- tion experiments, thus excluding influence from the Ezh1-defi- ure 2E). We also examined SA-b-gal activity and DNA damage cient BM microenvironment on hematopoiesis. Donor-derived by immunohistochemistry in BM sections. Immunostaining for (CD45.2) Ezh1-KO-ER BM cells did not contribute significantly gH2AX DNA repair foci revealed activation of the DNA-damage to peripheral blood (PB) at 12 weeks posttransplant (wpt; p < response in Ezh1-BM mice (Figure 2F), consistent with recent 0.05; Figure 2A) or LSKs in BM at 16 wpt (p < 0.001; Figure 2B). evidence linking DNA damage to cellular senescence (Di Micco Moreover, KO donor cells contributed weakly to lymphoid pro- et al., 2011). Immunostaining also revealed greater expression genitors in BM (p < 0.05; Figure S2A). These results are consis- of Dcr2, p15INK4b, and p16INK4a in Ezh1-KO BM, whereas

(D) Proliferation rate of LSK, MP, and CLP subpopulations measured by in vivo BrdU incorporation over 24 hr. Data are means ± SD (n = 5, *p < 0.05). (E) Left: Percentage of senescence-associated b-gal (SA-b-gal) positive cells determined by ﬂow cytometry in LSK, LT, ST, and MPP primitive populations in

Control and Ezh1-KO mice. Fluorescence histograms for C12-fluorescein show the relative levels of b-gal in LSK subpopulations; the values above the peaks are the median fluorescence intensities of the respective populations. Right: Percentage of SA-b-gal-positive cells in LSK, LT, ST, and MPP populations. Data are means ± SD (n = 6, *p < 0.05, **p < 0.001). See also Figure S2. (F) Cytochemical staining of SA-b-gal activity in paraffin sections of BM from Control and Ezh1-KO mice, and accompanying panels showing representative paraffin-embedded BM sections from Control and Ezh1-KO mice immunostained with the indicated antibodies. Scale bars, 50 mm. See also Figure S2.

Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. 653 Cell Stem Cell HSC Senescence Protection by Ezh1

5 Figure 3. Characterization of Primitive A B Control Hematopoietic Populations in Ezh1-KO 4 Ezh1-KO ) Mice 6 (A) vavCre expression in a sagittal section of 3 an E11.5 Rosa26-eYFPtg/+/vavCretg/+ embryo. Expression is detected in most hematopoietic 2 cells and organs. YFP-positive cells are brown and sections were hematoxylin counterstained. 1 The right panel shows an enlarged view of the Total FL cells (x10 FL Total fetal liver (FL) area. 0 (B) Absolute cell numbers in whole FLs from Control and Ezh1-KO mice at E11.5 (n = 12). (C) Left: Representative FACS proﬁles of FL cells from Ezh1-KO mice and Control littermates were C D Control Control 18.1% gated on the LSK subpopulation; the CLP fraction MP LSK low low + 0.47 % 0.18% Control was gated on LK S CD127 , and myeloid )

CLP 6 0.23 % Ezh1-KO 1.6% CD45.1 progenitors were deﬁned as the LKS population 40 (MP). Numbers indicate percentages of the 80.8% 30 * total nucleated BM cell subpopulations. At least eight mice per genotype were tested, over at Ezh1-KO 20 * CD45.2 MP LSK least two separate experiments. Right: Absolute 0.31 % 0.11% Ezh1-KO CLP 10 13.9% numbers per FL of LSK, MP, and CLP primitive 0.2 %

Total FL cells (x10 FL Total 0 populations are indicated. Data are means ± SD CD45.1 cKIT cKIT LSK MP CLP 1.3% (n = 16, *p < 0.05).

Sca-1 SSC 82.2% (D) Recipient mice were injected with pooled FL

B220 cells from Ezh1-KO mice or Control littermates CD45.2 (CD45.2). Average chimerism in posttransplant recipients: values on the right panels are the SA-β-gal mean percentages of donor-derived (CD45.2) E F lymphoid (B220) cells 8 weeks after secondary LSK MP CLP Control transplant. FACS plots show a representative Ezh1-KO 9.4 8.1 9.6 9.1 experiment (n = 6). (E) SA-b-gal-positive cell content in transplanted 7.9 11.6 chimeras harboring FL cells from Control or Ezh1- Control KO mice, determined by ﬂow cytometry of LSK, CLP, and MP primitive populations. Top: C -

Counts 12 ﬂuorescein ﬂuorescence histograms show the

C12FDG relative levels of b-gal in LSK subpopulations; values above the peaks are the median ﬂuores- 15 cence intensities of the respective populations (n = 6). Bottom: Percentages of SA-b-gal-positive 10 cells in LSK, MP, and CLP populations. Data are cells (%) Ezh1-KO + means ± SD (n = 6; bottom panel). 5

FDG (F) Representative cytochemical staining of SA-

12 b-gal-positive cells on BM paraffin sections. C 0 LSK MP CLP Scale bars, 50 mm. expression of Bmi1 and Ezh2 was the same in Control and Ezh1 Is Not Required for Fetal Hematopoiesis Ezh1-KO BM (Figure 2F). RT-PCR analysis confirmed greater Ezh2 appears to be required only in fetal HSCs, since its deletion expression of the senescence markers Dcr2, p15INK4b, and does not compromise adult hematopoiesis except for lympho- Dec1 (Collado and Serrano, 2006) in Ezh1-KO LSKCD150+ poiesis (Mochizuki-Kashio et al., 2011; Su et al., 2003). To inves- cells (Figure S2E). To confirm the role of Ezh1 in hematopoie- tigate the requirement of Ezh1 during fetal hematopoiesis, we sis, we retrovirally reexpressed Ezh1 in Ezh1-KO-ER HSCs first tested the specificity of vavCre activity in fetal liver (FL) by (LSKCD150+) and examined the ability of these cells to res- crossing vavCretg/+ mice with Rosa26-eYFPtg/ mice. Consistent cue hematopoiesis in competitive transplantation experiments. with previous studies (Stadtfeld and Graf, 2005), immunohisto- Ezh1 overexpression substantially decreased p16INK4a expres- chemistry revealed constitutive vav-promoter-driven Cre ex- sion in LSKCD150+ cells but had no effect on Ezh2 levels (Fig- pression in hematopietic cells from E9.5, mostly restricted to ure S2F). In the transplantation experiments, restored Ezh1 FL (Figure 3A). Total FL cell numbers in mutant mice were indis- expression rescued the capacity of donor Ezh1-KO cells to tinguishable from those of Control littermates (Figure 3B). How- engraft (Figure S2G). Moreover, mice engrafted with these cells ever, FL of mutant mice contained lower proportions of all were protected from the accumulation of senescent LSK primitive FL-LSK cells (p < 0.05; Figure 3C). In repopulation cells observed in recipients of vector-overexpressing Ezh1-KO experiments, Ezh1-deficient and Control FL cells showed only cells (p < 0.05; Figure S2H). These results suggest a primary func- a slight difference in the ability to reconstitute the lymphoid-con- tion of Ezh1 in the protection of adult HSCs from senescence. taining (B220) population in PB (Figure 3D). Analysis of SA-b-gal

654 Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. Cell Stem Cell HSC Senescence Protection by Ezh1

expression in E12.5 FL by FACS and cytochemical staining did Comparable results were obtained by injecting tamoxifen- not show accumulation of senescent FL cells of either genotype treated Ezh1-KO-ER HSCs (CD45.2) into CD45.1 hosts (p < (Figures 3E and 3F). These data thus suggest that while Ezh1 0.05; Figure S3C). The retention of Ezh1-deficient HSCs in deletion in FL influences HSC number, Ezh1 is not required for the BM niche was examined by reciprocal transplant experi- maintenance of fetal hematopoiesis. ments in which WT donor BM cells (CD45.1) were transplanted into tamoxifen-treated Control-ER or Ezh1-KO-ER recipients Ezh1 Deficiency Impairs B Cell Development (CD45.2). Analysis of recipient PB and immunohistochemical Given the differences in primitive subpopulations in Ezh1-KO staining of paraffin BM sections revealed no differences in mice, we analyzed whether specific Ezh1 deletion affects adult engraftment between the genotypes (Figures 5E and 5F), indi- hematopoiesis in vivo. PB cell counts and flow cytometry of cating that an Ezh1-deficient microenvironment does not affect spleen and thymus in Ezh1-KO animals revealed marked and HSC engraftment. Together, our transplantation experiments significant loss of B lymphocytes (p < 0.05) and a slight drop demonstrate that the development of Ezh1/-induced senes- in myeloid cells (Figures 4A–4C). Ezh1-KO spleens were about cence is hematopoietic cell autonomous, with defective migra- half the size of spleens from Control littermates (p < 0.05; Fig- tion of HSCs to the BM that likely compromises their self- ure 4B), consistent with the reduced proportion of mature renewal. lymphoid cells (p < 0.05; Figure 4B). Histopathological analysis of Ezh1-KO spleen mice was consistent with this phenotype Regulation of Multipotency Genes by Ezh1 (Figure 4D). Hematoxylin and eosin staining showed manifest To investigate the mechanism by which Ezh1 deletion impairs white and red pulp hypoplasia and an absence of germinal HSC activity, we profiled gene expression in highly enriched re- centers in Ezh1-KO spleen, accompanied by lymphoid cell populating HSCs (LSKCD150+CD48) isolated from Ezh1-KO depletion. Analysis of Ezh1-deficient BM cells showed altered and Control mice. Stringent gene array analysis identified 788 development of pro-B cells (IgMB220+CD43+), pre-B cells differentially expressed genes in Ezh1-KO HSCs (297 downregu- (IgMB220+CD43), and immature B cells (IgM+B220+CD43), lated and 491 upregulated; Table S4). A complete list of these with a concomitantly reduced proportion of recirculating lym- genes has been deposited in the Gene Expression Omnibus, phoid cells (IgM+B220highCD43) (p < 0.05; Figure 4E). The accession number GSE36288. The downregulated genes in- alterations in B cell development indicate that Ezh1 is required cluded many crucial for HSC function, such as Gata3, Runx1, for proper differentiation of HSCs into lymphoid progeny, and Meis1, Myb, Pten, Foxo3a, Thy1.1, Abcg2, Srgap2, Pbx1, and raise the possibility that PRC2-Ezh1 restricts certain HSC/pro- Cdkn1a (Chambers et al., 2007a; Forsberg et al., 2010). Also genitor activities, similar to the impaired B cell maturation from downregulated were genes for essential lymphoid differentiation pro-B to pre-B cells observed in mice lacking Ezh2 (Su et al., factors (Ng et al., 2009; Oguro et al., 2010), such as Dntt, Flk2, 2003). Igh6, Ikaros, Sfpi1, and Mef2c, and cell-cycle related genes such as Cdc6, Cdk1, and Mcm5, suggesting growth arrest (Kuil- BM Failure upon Loss of Ezh1 Is Hematopoietic Cell man et al., 2010)(Figures S4A and S4B). The set of upregulated Autonomous genes included genes related to secreted proteins and differen- To define the impact of Ezh1 on the maintenance of the stem cell tiation/development (e.g., Bmp2 and Igfbp3), consistent with the pool, we performed sequential BM transplantations. BM from central role of secreted factors in senescence (Kang et al., 2011). Ezh1-KO or WT littermate mice (CD45.2) was injected into irradi- Moreover, there was robust upregulation of genes encoding ated CD45.1 hosts, which were monitored for reconstitution of proinflammatory factors associated with SASP, including IL-6, PB, spleen, and BM. Ezh1-KO cells showed significantly weaker IL-7, granulocyte macrophage colony-stimulating factor (GM- PB chimerism at 4 wpt (p < 0.05; Figure S3A), as well as less CSF), and certain insulin-like growth factor (IGF)-binding pro- engraftment of primitive HSCs at 16 wpt (p < 0.05; Figure S3B). teins (Coppe´ et al., 2010)(Figures 6B and 6C). These findings PB lymphoid cell counts were significantly lower in Ezh1-KO-in- suggest that Ezh1 controls the secretome by repressing secre- jected hosts than in control-injected counterparts (p < 0.05; Fig- tion of key senescence factors. Furthermore, we detected ure 5A). Donor-derived CD45.2+ BM cells were then purified and altered expression of genes regulated by other Polycomb injected into secondary recipients. Spleens of secondary recipi- members, including upregulated expression of Dkk2, Cyp1b1, ents showed a consistent decrease in the lymphoid-containing Vsnl1, and Il6R, and downregulated expression of H2Afx, (B220) population (p < 0.05; Figure 5B). Moreover, the senes- Gmnn, and Foxc1 (Bracken et al., 2006; Majewski et al., 2008) cence phenotype was far more marked in secondary recipients (Figures 6B and 6C). The genes differentially regulated in Ezh1- of Ezh1-KO cells than in secondary recipients of Control BM cells depleted cells were objectively classified by Gene Ontology (Figure 5C). Strikingly, Ezh1/ HSCs were unable to engraft ter- analysis (Figure 6A). Changes in expression for several genes tiary recipients, while recipients of Control cells survived (data from each category were validated by real-time PCR using not shown). sorted HSCs (LSKCD150+CD48 cells) from Ezh1-KO and Control mice (Figure 6B), confirming the array analysis. Ezh1 Deletion Results in Homing Defects in BM Cells We next conducted chromatin immunoprecipitation (ChIP) To evaluate the mechanism of action of Ezh1 loss on HSC experiments on sorted Ezh1-KO and WT HSCs (LSKCD150+ engraftment, we assessed the homing capacity of CFSE-labeled cells) to detect the effect of Ezh1 deletion on H3K27 methyla- Ezh1-KO and Control HSCs in transplantation assays. CFSE+ tion and the specific binding of Ezh1, Ezh2, and Bmi1 to the Ezh1-KO LSK cells homed to the BM of recipient mice at a lower promoter regions of Ezh1 target genes. Lack of Ezh1 increased frequency than cells from Control mice (p < 0.05; Figure 5D). the amount of bound H3K27me2 in the promoter regions of the

Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. 655 Cell Stem Cell HSC Senescence Protection by Ezh1

Control A B

2% 58.8% 58.8%58.58588.8%8%

/µL) 0.6 %

3 Control Ezh1-KO * * Ezh1-KO 1.3 % Cell counts (x10

SSCA SSCA 1.4 % 34.4 % WBC LYM MID GRA B220 Ly6G F4/80 /µL) /µL) 6 3

40 8 Control ) ) 5 6 30 6 Ezh1-KO 20 * 4 * 10 2 cells (x 10 cells (x 10 Cell counts (x10 Cell counts (x10 0 0 spleen B220 Ly6G F4/80 RBC PLT cells

Control Ezh1-KO C Control Control Ezh1-KO D 20.5% 39.5% 20.5200050.50..55 27.527.27275.55

15.4% H&E

15.4 Ezh1-KO 80 27.7% 43.8% Control ) 6 60 Ezh1-KO 20.5200050.50..5 20.52020.20 5 40 cells (x 10

CD4 12..2% 20 CD8 0 Thymus CD4 CD4CD8 CD8 Control Ezh1-KO cells E

Control Ezh1-KO 21.2% 14.5% 40 Control

2.1% 12.1% 18.1% 9.5% Ezh1-KO Ki67 30 ) 5

20 * B220 IgM *

cells (x 10 * 10

38.2% 48.2% 28.2% 31.2% 0 proB preB immature recirculating CD43

Figure 4. Lymphoid Differentiation Defects in Ezh1-KO Hematopoiesis (A) Peripheral blood counts in 20- to 22-week-old Ezh1-KO and Control mice. WBC, white blood cells; LYM, lymphocytes; MID, monocytes; GRA, granulocytes; RBC, red blood cells; PLT, platelets. Data are means ± SD (n = 8, *p < 0.05). (B) Top: Representative FACS plots showing the percentages of lymphoid (B220) and myeloid (Ly6G and F4/80) positive cells in the Ezh1-KO and Control spleens. Bottom: Bar charts show absolute numbers of total spleen, lymphoid, and myeloid cell populations in the spleens of Ezh1-KO and Control mice; data are means ± SD (n = 11, *p < 0.05). The photograph shows representative spleens from Ezh1-KO and Control mice. (C) Left: Representative FACS plots showing the percentages of CD4-positive, CD4+CD8-positive, and CD8-positive cells in the thymus of Ezh1-KO and Control animals. Right: The photograph shows thymus from Ezh1-KO and Control mice, and the bar charts show thymus cell populations and absolute numbers of mature lymphoid (CD4, CD4+CD8, and CD8) cells. Data are means ± SD (n = 10). (D) Histological (H&E) and immunostaining (anti-Ki67 antibody) characterization of representative spleen sections from Ezh1-KO mice and Control littermates. Scale bars, 1,000 mm (upper panels) and 200 mm (magniﬁed views in lower panels). (E) Left: Representative FACS plots showing the percentages of pro-B cells (IgMB220+CD43+), pre-B cells (IgMB220+CD43), immature B cells (IgM+ B220+CD43), and recirculating cells (IgM+B220highCD43) in Ezh1-KO and Control BM. Right: Absolute numbers per femur of mature lymphoid cell populations in Ezh1-KO and Control mice. Data are means ± SD (n = 12, *p < 0.05).

656 Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. Cell Stem Cell HSC Senescence Protection by Ezh1

Control

A B 0.3

60 62 % CD45.1 10 Control

Control ) Ezh1-KO 6 8 Ezh1-KO 98.8 45 8 6 CD45.2 4 30 6 Ezh1-KO 2 B220 cells (x10 4 0 15 Control Ezh1-KO 0.5 * 2 * 37% CD45.1

Donor-derived cells (%) ( secondary recipients) 0 0 CD3 CD4 CD8 B220 Mac-1+/Gr-1 LY6G F4-80

SSC 99

B220 CD45.2

Control Ezh1-KO C D Control

0.35%

0.5 SA- Control 0.4 Ezh1-KO cells + -gal 0.3 Ezh1-KO 0.2 0.17% 0.1 % of CFSE

0 Count

CFSE

Control-ER Ezh1-KO-ER E F Control-ER 60 Ezh1-KO-ER

H&E 30

15 Donor-derived cells (%)

0 WBC Lymphocytes Neutrophils Monocytes Ki67

Figure 5. Ezh1 Function in Hematopoiesis Is Cell Autonomous (A) Average chimerism in posttransplant secondary recipients (separated by a 10-week reconstitution period) analyzed for the contribution of donor-derived (CD45.2) cells to circulating mature myeloid cells (Mac-1/Gr-1), macrophages (F4-80), granulocytes (LY6G), and lymphoid subfractions (CD3, CD4, CD8, and B220). Data are means ± SD (n = 8; *p < 0.05). See also Figure S3. (B) Subpopulation profiles of B cells (B220; CD45.1/CD45.2) in spleens of secondary recipient mice transplanted with BM cells from Ezh1-KO mice or Control littermates. The bar chart shows the percentages (mean ± SD) of the indicated populations (n = 8; *p < 0.05). (C) Cytochemical staining of SA-b-gal-positive cells in paraffin-fixed BM sections of recipient mice transplanted with BM cells from Ezh1-KO mice or Control littermates. Scale bars, 50 mm. (D) Representative FACS plots of BM cells in recipient mice 24 hr after injection with CFSE-labeled Control or Ezh1-KO LSK cells. The bar chart shows the mean ± SD of two independent experiments with four mice per experiment (n = 8; *p < 0.05). See also Figure S3. (E) H&E and anti-Ki67 staining of BM sections of Control-ER and Ezh1-KO-ER recipient mice injected with WT BM cells. Scale bars, 500 mm and 50 mm. (F) Average chimerism in posttransplant Control-ER and Ezh1-KO-ER recipients analyzed by the contribution of donor-derived (CD45.1) cells to PB populations (WBC, white blood cells). Data are means ± SD (n = 9). downregulated genes Cdc6, Runx1, and Mef2c, but not in the were enriched for the H3K27me1 signal in Ezh1-KO cells, promoters of p16INK4a, Bmp2, and Igfbp3 (Figure S4A). Interest- whereas this signal was absent from the p16INK4a, Bmp2, and ingly, the entire promoter regions of Cdc6, Runx, and Mef2c Igfbp3 promoters. In contrast, the levels of H3K27me3 and

Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. 657 Cell Stem Cell HSC Senescence Protection by Ezh1

Percentage HSCs (LSKCD150+CD48-) A changed B Cell-To-Cell Signaling and Interaction 55.2 MT3 MAP3K6 CALCB LY6a CD19 CXCL12 CAMP CFH FASLG CYBR KDR IL6 Cdc2 Cdkn1a Cdkn2a ITGA7 JAM2 PTPN11 GM-CSF STAT5 SNRPN GLDC EWSR1 Hematological System Development and Function 47.4 RAP1A Immune Cell Trafficking 38.2 Inflammatory Response 58.2 Cell Cycle, and Cellular Organization 42.4 Cellular Movement 48.3 Hematological Disease 42.2 Cellular Function and Maintenance 32.4

0 2.0 4.0 6.0 8.0 10.0 Relative expression level Cellular System Cellular Disease Signaling Response Trafficking Trafficking Cell Cycle Movement Cell-to-Cell Maintenance Immune Cell Inflammatory Hematological Hematological Cellular Organiz.

Genes downregulated C Control Ezh1-KO D in Ezh1-KO HSCs Sfpi1 297 297 Lymphoid Flk2 development Mef2c 42 29 Foxc1 71 161 Polycomb- Gmnn related Il6R Polycomb-related Lymphoidmphohoid devde developmentelopmp genes regulatorregulatort r genesgenesne Dkk2 Genes upregulated Bmp2 Senescence- in Ezh1-KO HSCs associated IGF-BP1 GM-CSF 491 491

310-1-3 Relative expression (log2 scale) 18 9 75 51

Polycomb-related Senescence-associated genes genes

Figure 6. Ezh1 Maintains BM Homeostasis (A) Enrichment scores for GO categories in the set of genes differentially expressed in Ezh1-deficient HSCs identified high percentage changes in genes regulating HSC maintenance. (B) Relative expression levels of selected transcripts (horizontal axis) from each GO category (A) measured by qRT-PCR in sorted HSCs (LSKCD150+CD48) from Ezh1-KO and Control mice. Data are means ± SD from three replicates in Ezh1 null HSCs normalized to expression in Control HSCs (dashed line). (C) Representative heat maps show the expression of genes with lymphoid-development, senescence-associated, and Polycomb-related functional annotations that are significantly downregulated and upregulated in Ezh1 mutant LSKCD150+CD48 cells. The rows correspond to genes and the columns to samples. Gene expression values (relative to the mean expression in Control cells) are indicated on a log2 scale according to the color scheme shown. (D) Venn diagram illustrating the overlap between genes downregulated and upregulated in these cells and genes with senescence-, lymphoid- and Polycomb- related functional annotations according to the studies cited in the main manuscript. See also Figure S4 and Table S2. bound Ezh2 and Bmi1 in the Cdc6, Runx, and Mef2c pro- data support the view that the actions of Ezh2 and Bmi1 likely moters were not affected by the absence of Ezh1 (Figure S4A). require the presence of Ezh1, and that Ezh1-dependent repres- Furthermore, whereas Control HSCs showed site-specific sion induces monomethylation of H3K27 as the primary product binding of Ezh1, Ezh2, and Bmi1 to the promoter regions of of the PRC2-Ezh1 reaction, with subsequent dimethylation. The p16INK4a (Bracken et al., 2007), Bmp2, and Igfbp3, binding gene expression data thus correlate strongly with the hemato- was absent from these upregulated genes in Ezh1-KO cells. poietic phenotype of Ezh1-KO mice and suggest a mechanism Comparable results were observed with the upregulated Ezh1- by which the histone methytansferase Ezh1 protects against target genes IL-6, Cdkn1a, and Zmat3 (data not shown). These senescence.

658 Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. Cell Stem Cell HSC Senescence Protection by Ezh1

C D

Figure 7. Recovery of Defective Hematopoiesis in Ezh1/Cdkn2a-TKO Mice (A) qRT-PCR analysis of the expression of Ezh1, p16INK4a, ARF, Runx1, and Cdc6 mRNA in total HSCs (LSKCD150+CD48) from Ezh1/Cdkn2a-TKO mice (Ezh1fl/fl;p16INK4a-ARFfl/fl;vavCretg/+). Results are standardized to b-actin and expressed relative to littermate Control-TKO mice (Ezh1fl/fl;p16INK4a-ARFfl/fl; vavCre+/+). Data are means ± SD; n = 10, **p < 0.001, *p < 0.05). (B) Absolute numbers of mature lymphoid (B220 and CD4/CD8) cells and total cellularity in spleen, thymus, and BM of Ezh1/Cdkn2a-TKO and Control-TKO mice. Data are means ± SD (n = 5). (C) Left: Representative FACS plots of BM cells, indicating mean percentages of the LSK populations, in Ezh1/Cdkn2a-TKO mice and Control-TKO littermates. Right: Absolute numbers per femur of the LSK (LS+K+), CLP (LKlowSlowCD127+), and MP (LK+S) subpopulations. Data are means ± SD (n = 8).

(D) C12-fluorescein fluorescence histograms showing the relative levels of b-gal in LSK subpopulations from Control-TKO and Ezh1/Cdkn2a-TKO mice. The values above the peaks are the median fluorescence intensities of the respective populations (n = 6). (E) Contribution of donor-derived (CD45.2) cells from Ezh1/Cdkn2a-TKO and Control-TKO mice to PB populations in recipients at 4 weeks posttransplant (WBC, white blood cells). Data are means ± SD (n = 7).

Ezh1 Limits the HSC Senescence Response by tory upregulation of other Polycomb members (p < 0.05; Fig- Repressing Cdkn2a ure 7A; data not shown). We also observed altered HSC expres- To test whether repression of Cdkn2a by Ezh1 is essential for its sion of the Ezh1 targets Runx1 and Cdc6 (Figure 7A). Total cell antisenescence function, we generated Ezh1/Cdkn2a triple- numbers in BM, thymus, and spleen of Ezh1/Cdkn2a-TKO knockout (TKO) mice (Ezh1fl/fl;p16INK4a-ARFfl/fl;vavCretg/+) and mice were similar to numbers in Control-TKO littermates (Fig- Control-TKO mice (Ezh1fl/fl;p16INK4a-ARFfl/fl;vavCre+/+) to condi- ure 7B), and mutant and Control mice also had similar propor- tionally eliminate hematopoietic Cdkn2a expression in the Ezh1- tions of all major BM populations, including LSK, common KO background (Figure 7A). Complete deletion of Ezh1 and lymphoid progenitor (CLP), and MP stem/progenitor populations p16INK4a/ARF was not accompanied by significant compensa- (Figure 7C). The deletion of p16INK4a and ARF in Ezh1-KO HSCs

Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. 659 Cell Stem Cell HSC Senescence Protection by Ezh1

led to a near complete bypass of the senescence pathway repressive H3K27 mark, which activates the master senescence as determined by the decrease in the proportion of LSK cells ex- regulator Bmp2 (Kaneda et al., 2011) and selectively represses hibiting SA-b-gal activity (4%–7% SA-b-gal+ cells, compared expression of Runx1 and Cdc6, which regulate critical hemato- with 5% in Control-TKO mice) (Figure 7D). To test the function- poietic processes (Swiers et al., 2010). Our results reveal dy- ality of Ezh1/Cdkn2a-TKO HSCs, we generated competitive namic and harmonized H3K27me1 and H3K27me2 alterations, chimeras. Remarkably, flow cytometry of PB in stably engrafted repression of Runx1 without gain of H3K27me3, and selective chimeric mice showed that donor contribution to lymphocytes activation of downstream target genes, such as Smad1, that declined slightly after 8 weeks without affecting PB myeloid cells may contribute to growth arrest (Kaneda et al., 2011). Despite (Figure 7E). In summary, these data suggest that removal of this broad association of H3K27me1 H3K27me2 with silenced p16INK4a and ARF from the Ezh1 null background rescues hema- genes in different tissue lineages, loss of Ezh1-mediated repres- topoiesis by increasing proliferation and alleviating senescence sion reduces the proliferation of these lineages by upregulating in the Ezh1-KO BM. the Cdkn2a locus, and tempers acquisition of the senescence barrier. Indeed, deletion of the p16INK4a/p19ARF locus rescues DISCUSSION the HSC defects in these mice. Our findings underline the importance of epigenetic changes in the regulation of the criti- Chromatin modifiers such as the PcG proteins determine the cal physiological barrier to inducing the senescence that blocks transmission of genetic information and regulate cell differentia- oncogenic transformation. Further understanding of how PRC2- tion. Disruption of PcG function prevents differentiation of ESCs Ezh1 maintains specific gene-expression patterns may come and correct embryonic development. The PRC2 methyltrans- from consideration of overall chromatin structure, an essential ferase Ezh2 was thought to be unique and essential for all issue that has garnered attention in the context of ESC differen- H3K27me3 marks in ESCs (Bernstein et al., 2006; Boyer et al., tiation (Margueron and Reinberg, 2011). 2006). However, its homolog Ezh1 is also indispensable for the Ezh1 and Ezh2 proteins are implicated in H3K27 monomethy- establishment of pluripotent cells and for the maintenance of lation, dimethylation, and trimethylation, with each methylated ESC self-renewal (Margueron et al., 2008; Shen et al., 2008). form likely to be functionally distinct. The regulation of the level Here we show that this dual requirement also holds true for of H3K27 methylation is, however, unknown. The answer may at least one population of somatic stem cells, the HSCs. The lie in the particular protein composition of PRC2 complexes. extreme sensitivity of hematopoietic cells to Ezh1 inactivation For example, Suz12 has been found to be essential for is remarkable when compared with the weaker impact of inacti- H3K27me3, but dispensable for H3K27me1 (Pasini et al., 2004). vating its homolog Ezh2 in a similar system (Mochizuki-Kashio Similarly, the mammalian protein PHF1, a Drosophila PCL et al., 2011; Su et al., 2003). Ezh1 deficiency severely reduces homolog, associates with PRC2 components and stimulates the size of the adult HSC fraction and impairs the capacity of formation of H3K27me3 in vitro and in vivo (Cao et al., 2008). HSCs for self-renewal and quiescence, effects not observed in Our work suggests that H3K27me1 plays a critical role in main- mice lacking Ezh2 (Mochizuki-Kashio et al., 2011). We thus taining heterochromatin condensation and gene silencing, a conclude that Ezh1 plays a dominant role among histone meth- mechanism reminiscent of the silencing imposed by two Arabi- yltransferases in the maintenance of adult HSCs. dopsis histone methyltransferase genes, ATXR5 and ATXR6 We show that in the hematopoietic system, where stem cell (Jacob et al., 2010). It remains unknown, however, how this activation is a transitory state that requires self-renewal coupled histone modification signals such events. One possibility is with the prevention of differentiation and senescence, Ezh1 that H3K27me1 simply recruits an H3K27me1-binding protein. plays an unanticipated and critical role. Ezh1 protects primitive Alternatively, monomethylation of H3K27 (and possibly further hematopoietic cells from senescence by keeping them in a H3K27me2 and H3K27me3 modifications) may act in a negative slow-cycling state through its PRC2-associated Polycomb func- manner, either by preventing unmethylated H3K27-interacting tion. Deletion of Ezh1 profoundly compromises BM hematopoi- proteins from binding chromatin or by blocking alternative his- esis, but not the self-renewal capacity of HSCs in FL. This tone modifications to H3K27. We thus suggest that H3K27me1 reflects the specialized function of Ezh1, which is mediated is an important intermediary PRC2-Ezh1 product, since it not through monomethylation and dimethylation of H3K27, thereby only constitutes the substrate for subsequent H3K27me2, but preventing proliferation and repressing pathways required for also prevents H3K27 acetylation. In fact, our results, while vali- terminal differentiation and senescence of adult HSCs. Our dating only a small proportion of Ezh1-regulated transcripts, data suggest that Ezh1-deficient cells exhibit delayed lympho- shed light on this matter, and demonstrate the binding of Ezh1, cyte differentiation at specific stages of hematopoiesis, and it Ezh2, and Bmi1 to upregulated Igfbp3, IL-6, Cdkn1a, Zmat3, may be that this effect arises from the reduced size of the prim- Bmp2, and p16INK4a transcripts in WT HSCs (Bracken et al., itive HSC population. 2007). The action of Ezh1 might thus involve other chromatin Epigenetic mechanisms, including DNA methylation and his- targets, such as Bmi1 or Ezh2, enabling fine regulation of cell tone modification, are important regulators of gene expression specification processes. These data provide a starting point for in senescence. The initial challenge in revealing Ezh1-mediated further studies to uncover the underlying mechanisms of Ezh1- regulation of gene expression was to determine what Ezh1 target mediated hematopoiesis. genes do. Through genome-wide array analyses combined with Our results imply that Ezh1-regulated senescence ability is a quantitative ChIP assays, we have examined epigenomic and limiting factor in the maintenance of functional HSC reserves. gene expression changes induced by Ezh1 deletion in senesced It thus seems plausible that some of the decline in HSC re- HSCs. Our results identify a coordinated methylation of the constituting potential associated with defective self-renewal in

660 Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. Cell Stem Cell HSC Senescence Protection by Ezh1

Ezh1-KO mice could be due to senescence, rather than genomic ACKNOWLEDGMENTS damage or telomere erosion. Loss of HSC potential might be INK4a fl/fl mediated by activation of senescence or cytostatic tumor We thank Dr. Berns for the ckop16 /ARF mice; A. Hidalgo, A.R. Ramiro, and J.A. Enriquez for helpful discussions; P. Arribas, M. Sonseca, R. Diges, suppressor pathways at cell-cycle checkpoints, leading to dele- and G. Merfu for excellent technical assistance; G. Giovinazzo, L.-M. Criado, tion or growth arrest of defective progenitor clones. M. Montoya, M. Can˜ amero, and their teams for technical support; and S. Bar- On the basis of our results, we propose that Ezh1 regulates tlett for English editing. S.G. is funded by the Human Frontiers Science the balance between HSC maintenance and a protective senes- Program Organization and the Spanish Ministries of Science and Innovation cence response to prevent uncontrolled proliferation. Abundant (SAF2010-15386) and Health (FIS PI06/0627). The CNIC is supported by the evidence indicates that cell senescence is a natural physio- Ministerio de Economia y Competitividad and the Pro-CNIC Foundation. logical protection against tumor development, and therapeutic Received: February 22, 2012 engagement of senescence may present a valuable treatment Revised: June 8, 2012 strategy for cancer. Targeting the PRC2-Ezh1 complex is an Accepted: August 4, 2012 attractive approach to the manipulation of these intrinsic senes- Published: November 1, 2012 cence pathways. The ability to promote senescence induction upstream of Cdkn2a is likely to have a substantial therapeutic REFERENCES effect, especially given the frequent mutations that inactivate these pathways. Understanding molecular mechanisms by Bernstein, B.E., Mikkelsen, T.S., Xie, X., Kamal, M., Huebert, D.J., Cuff, J., Fry, which Ezh1 affects stem cell fate will provide new insights into B., Meissner, A., Wernig, M., Plath, K., et al. (2006). A bivalent chromatin structure marks key developmental genes in embryonic stem cells. Cell 125, HSC biology and will also increase understanding of leukemic 315–326. transformation, with the potential to facilitate eradication of Boyer, L.A., Plath, K., Zeitlinger, J., Brambrink, T., Medeiros, L.A., Lee, T.I., leukemic stem cells. Levine, S.S., Wernig, M., Tajonar, A., Ray, M.K., et al. (2006). Polycomb complexes repress developmental regulators in murine embryonic stem cells. EXPERIMENTAL PROCEDURES Nature 441, 349–353. Bracken, A.P., Dietrich, N., Pasini, D., Hansen, K.H., and Helin, K. (2006). Mice Genome-wide mapping of Polycomb target genes unravels their roles in cell Generation and analysis of ckoEzh1fl/+ mice is detailed in the Supplemental fate transitions. Genes Dev. 20, 1123–1136. Information. The vavCre and Rosa26-Cre-ERT2 mouse lines were maintained after backcrossing to C57BL/6 over two generations, and were crossed with Bracken, A.P., Kleine-Kohlbrecher, D., Dietrich, N., Pasini, D., Gargiulo, G., ckoEzh1fl/fl mice to generate Ezh1-KO (Ezh1fl/fl/vavCretg/+), Control (Ezh1fl/fl/ Beekman, C., Theilgaard-Mo¨ nch, K., Minucci, S., Porse, B.T., Marine, J.C., vavCre+/+), Ezh1-KO-ER (Ezh1fl/fl-CreERtg/tg) and Control-ER (Ezh1+/+- et al. (2007). The Polycomb group proteins bind throughout the INK4A-ARF CreERtg/tg) mice. All genotyping primers are listed in Table S1. ckop16INK4a/ locus and are disassociated in senescent cells. Genes Dev. 21, 525–530. ARFfl/fl mice, kindly provided by Anton Berns, were crossed with Ezh1-KO Cao, R., Wang, H., He, J., Erdjument-Bromage, H., Tempst, P., and Zhang, Y. mice to generate Ezh1/Cdkn2a-TKO and Control-TKO mice. Additional details (2008). Role of hPHF1 in H3K27 methylation and Hox gene silencing. Mol. Cell. are provided in the Supplemental Experimental Procedures. Biol. 28, 1862–1872. Chambers, S.M., Boles, N.C., Lin, K.Y., Tierney, M.P., Bowman, T.V., Transplantation Assays Bradfute, S.B., Chen, A.J., Merchant, A.A., Sirin, O., Weksberg, D.C., et al. + Cells were transplanted (via retro-orbital injection) into the congenic CD45.1 (2007a). Hematopoietic fingerprints: an expression database of stem cells + B6.SJL mouse strain, and hosts were monitored for the presence of CD45.2 and their progeny. Cell Stem Cell 1, 578–591. donor-derived cells. Engraftment efficiency in recipients was monitored by Chambers, S.M., Shaw, C.A., Gatza, C., Fisk, C.J., Donehower, L.A., and FACS analysis of the contribution of donor CD45.1-positive cells to PB popu- Goodell, M.A. (2007b). Aging hematopoietic stem cells decline in function lations at 8 weeks posttransplant. Additional details are provided in the and exhibit epigenetic dysregulation. PLoS Biol. 5, e201. Supplemental Experimental Procedures. Chen, H., Gu, X., Su, I.H., Bottino, R., Contreras, J.L., Tarakhovsky, A., and Flow Cytometry, Cell Sorting, ChIP Assays, Protein Analysis, and Kim, S.K. (2009). Polycomb protein Ezh2 regulates pancreatic beta-cell Immunohistochemistry Ink4a/Arf expression and regeneration in diabetes mellitus. Genes Dev. 23, Details are described in the Supplemental Experimental Procedures. 975–985. Collado, M., and Serrano, M. (2006). The power and the promise of oncogene- Statistical Analysis induced senescence markers. Nat. Rev. Cancer 6, 472–476. Computational processing of experimental data is described in the Supple- Collado, M., Blasco, M.A., and Serrano, M. (2007). Cellular senescence in mental Information. All data are expressed as means ± SD. Statistical signifi- cancer and aging. Cell 130, 223–233. cance of differences was measured by unpaired two-tailed Student’s t test; Coppe´ , J.P., Desprez, P.Y., Krtolica, A., and Campisi, J. (2010). The senes- *p < 0.05, or **p < 0.001. cence-associated secretory phenotype: the dark side of tumor suppression. Annu. Rev. Pathol. 5, 99–118. ACCESSION NUMBERS Debacq-Chainiaux, F., Erusalimsky, J.D., Campisi, J., and Toussaint, O. A complete list of the differentially expressed genes in Ezh1-KO HSCs has (2009). Protocols to detect senescence-associated beta-galactosidase (SA- been deposited in the Gene Expression Omnibus under accession number betagal) activity, a biomarker of senescent cells in culture and in vivo. Nat. GSE36288. Protoc. 4, 1798–1806. Di Micco, R., Sulli, G., Dobreva, M., Liontos, M., Botrugno, O.A., Gargiulo, G., SUPPLEMENTAL INFORMATION dal Zuffo, R., Matti, V., d’Ario, G., Montani, E., et al. (2011). Interplay between oncogene-induced DNA damage response and heterochromatin in senes- Supplemental Information for this article includes four figures, four tables, and cence and cancer. Nat. Cell Biol. 13, 292–302. Supplemental Experimental Procedures and can be found with this article Ezhkova, E., Pasolli, H.A., Parker, J.S., Stokes, N., Su, I.H., Hannon, G., online at http://dx.doi.org/10.1016/j.stem.2012.08.001. Tarakhovsky, A., and Fuchs, E. (2009). Ezh2 orchestrates gene expression

Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc. 661 Cell Stem Cell HSC Senescence Protection by Ezh1

for the stepwise differentiation of tissue-specific stem cells. Cell 136, 1122– Margueron, R., Li, G., Sarma, K., Blais, A., Zavadil, J., Woodcock, C.L., 1135. Dynlacht, B.D., and Reinberg, D. (2008). Ezh1 and Ezh2 maintain repressive Ezhkova, E., Lien, W.H., Stokes, N., Pasolli, H.A., Silva, J.M., and Fuchs, E. chromatin through different mechanisms. Mol. Cell 32, 503–518. (2011). EZH1 and EZH2 cogovern histone H3K27 trimethylation and are essen- Mochizuki-Kashio, M., Mishima, Y., Miyagi, S., Negishi, M., Saraya, A., tial for hair follicle homeostasis and wound repair. Genes Dev. 25, 485–498. Konuma, T., Shinga, J., Koseki, H., and Iwama, A. (2011). Dependency on Forsberg, E.C., Passegue´ , E., Prohaska, S.S., Wagers, A.J., Koeva, M., Stuart, the polycomb gene Ezh2 distinguishes fetal from adult hematopoietic stem J.M., and Weissman, I.L. (2010). Molecular signatures of quiescent, mobilized cells. Blood 118, 6553–6561. and leukemia-initiating hematopoietic stem cells. PLoS ONE 5, e8785. Nardella, C., Clohessy, J.G., Alimonti, A., and Pandolfi, P.P. (2011). Pro-senes- Freund, A., Orjalo, A.V., Desprez, P.Y., and Campisi, J. (2010). Inflammatory cence therapy for cancer treatment. Nat. Rev. Cancer 11, 503–511. networks during cellular senescence: causes and consequences. Trends Narita, M., Nunez,~ S., Heard, E., Narita, M., Lin, A.W., Hearn, S.A., Spector, Mol. Med. 16, 238–246. D.L., Hannon, G.J., and Lowe, S.W. (2003). Rb-mediated heterochromatin Goldberg, A.D., Allis, C.D., and Bernstein, E. (2007). Epigenetics: a landscape formation and silencing of E2F target genes during cellular senescence. Cell takes shape. Cell 128, 635–638. 113, 703–716. Gonzalez, S., Klatt, P., Delgado, S., Conde, E., Lopez-Rios, F., Sanchez- Ng, S.Y., Yoshida, T., Zhang, J., and Georgopoulos, K. (2009). Genome-wide Cespedes, M., Mendez, J., Antequera, F., and Serrano, M. (2006). Oncogenic lineage-specific transcriptional networks underscore Ikaros-dependent lym- activity of Cdc6 through repression of the INK4/ARF locus. Nature 440, phoid priming in hematopoietic stem cells. Immunity 30, 493–507. 702–706. Oguro, H., Yuan, J., Ichikawa, H., Ikawa, T., Yamazaki, S., Kawamoto, H., He, J., Kallin, E.M., Tsukada, Y., and Zhang, Y. (2008). The H3K36 demethy- Nakauchi, H., and Iwama, A. (2010). Poised lineage specification in multipoten- lase Jhdm1b/Kdm2b regulates cell proliferation and senescence through tial hematopoietic stem and progenitor cells by the polycomb protein Bmi1. p15(Ink4b). Nat. Struct. Mol. Biol. 15, 1169–1175. Cell Stem Cell 6, 279–286. Jacob, Y., Stroud, H., Leblanc, C., Feng, S., Zhuo, L., Caro, E., Hassel, C., Pasini, D., Bracken, A.P., Jensen, M.R., Lazzerini Denchi, E., and Helin, K. Gutierrez, C., Michaels, S.D., and Jacobsen, S.E. (2010). Regulation of hetero- (2004). Suz12 is essential for mouse development and for EZH2 histone meth- chromatic DNA replication by histone H3 lysine 27 methyltransferases. Nature yltransferase activity. EMBO J. 23, 4061–4071. 466, 987–991. Pereira, J.D., Sansom, S.N., Smith, J., Dobenecker, M.W., Tarakhovsky, A., Jacobs, J.J., Kieboom, K., Marino, S., DePinho, R.A., and van Lohuizen, M. and Livesey, F.J. (2010). Ezh2, the histone methyltransferase of PRC2, regu- (1999). The oncogene and Polycomb-group gene bmi-1 regulates cell prolifer- lates the balance between self-renewal and differentiation in the cerebral ation and senescence through the ink4a locus. Nature 397, 164–168. cortex. Proc. Natl. Acad. Sci. USA 107, 15957–15962. Kaneda, A., Fujita, T., Anai, M., Yamamoto, S., Nagae, G., Morikawa, M., Tsuji, Prieur, A., and Peeper, D.S. (2008). Cellular senescence in vivo: a barrier to S., Oshima, M., Miyazono, K., and Aburatani, H. (2011). Activation of Bmp2- tumorigenesis. Curr. Opin. Cell Biol. 20, 150–155. Smad1 signal and its regulation by coordinated alteration of H3K27 trimethy- Sauvageau, M., and Sauvageau, G. (2010). Polycomb group proteins: multi- lation in Ras-induced senescence. PLoS Genet. 7, e1002359. faceted regulators of somatic stem cells and cancer. Cell Stem Cell 7, Kang, T.W., Yevsa, T., Woller, N., Hoenicke, L., Wuestefeld, T., Dauch, D., 299–313. Hohmeyer, A., Gereke, M., Rudalska, R., Potapova, A., et al. (2011). Shen, X., Liu, Y., Hsu, Y.J., Fujiwara, Y., Kim, J., Mao, X., Yuan, G.C., and Senescence surveillance of pre-malignant hepatocytes limits liver cancer Orkin, S.H. (2008). EZH1 mediates methylation on histone H3 lysine 27 and development. Nature 479, 547–551. complements EZH2 in maintaining stem cell identity and executing pluripo- Kuilman, T., Michaloglou, C., Mooi, W.J., and Peeper, D.S. (2010). The tency. Mol. Cell 32, 491–502. essence of senescence. Genes Dev. 24, 2463–2479. Stadtfeld, M., and Graf, T. (2005). Assessing the role of hematopoietic plas- Laible, G., Wolf, A., Dorn, R., Reuter, G., Nislow, C., Lebersorger, A., Popkin, D., ticity for endothelial and hepatocyte development by non-invasive lineage Pillus, L., and Jenuwein, T. (1997). Mammalian homologues of the Polycomb- tracing. Development 132, 203–213. group gene Enhancer of zeste mediate gene silencing in Drosophila hetero- Su, I.H., Basavaraj, A., Krutchinsky, A.N., Hobert, O., Ullrich, A., Chait, B.T., chromatin and at S. cerevisiae telomeres. EMBO J. 16, 3219–3232. and Tarakhovsky, A. (2003). Ezh2 controls B cell development through histone Lowe, S.W., Cepero, E., and Evan, G. (2004). Intrinsic tumour suppression. H3 methylation and Igh rearrangement. Nat. Immunol. 4, 124–131. Nature 432, 307–315. Swiers, G., de Bruijn, M., and Speck, N.A. (2010). Hematopoietic stem cell Majewski, I.J., Blewitt, M.E., de Graaf, C.A., McManus, E.J., Bahlo, M., Hilton, emergence in the conceptus and the role of Runx1. Int. J. Dev. Biol. 54, A.A., Hyland, C.D., Smyth, G.K., Corbin, J.E., Metcalf, D., et al. (2008). 1151–1163. Polycomb repressive complex 2 (PRC2) restricts hematopoietic stem cell Wagner, W., Bork, S., Horn, P., Krunic, D., Walenda, T., Diehlmann, A., Benes, activity. PLoS Biol. 6, e93. V., Blake, J., Huber, F.X., Eckstein, V., et al. (2009). Aging and replicative Margueron, R., and Reinberg, D. (2011). The Polycomb complex PRC2 and its senescence have related effects on human stem and progenitor cells. PLoS mark in life. Nature 469, 343–349. ONE 4, e5846.

662 Cell Stem Cell 11, 649–662, November 2, 2012 ª2012 Elsevier Inc.