Poised Lineage Specification in Multipotential Hematopoietic Stem

Cell Stem Cell Short Article

Poised Lineage Speciﬁcation in Multipotential Hematopoietic Stem and Progenitor Cells by the Polycomb Protein Bmi1

Hideyuki Oguro,1,2 Jin Yuan,1,2 Hitoshi Ichikawa,4 Tomokatsu Ikawa,5 Satoshi Yamazaki,3,6 Hiroshi Kawamoto,5 Hiromitsu Nakauchi,3,6 and Atsushi Iwama1,2,* 1Department of Cellular and Molecular Medicine, Graduate School of Medicine, Chiba University, Chiba 260-8670, Japan 2JST, CREST 3JST, ERATO Sanbancho, Chiyoda-ku, Tokyo 102-0075, Japan 4Genetics Division, National Cancer Center Research Institute, Tokyo, 104-0045, Japan 5Laboratory for Lymphocyte Development, RIKEN Research Center for Allergy and Immunology, Yokohama 230-0045, Japan 6Laboratory of Stem Cell Therapy, Center for Experimental Medicine, Institute of Medical Sciences, University of Tokyo, Tokyo 108-8679, Japan *Correspondence: [email protected] DOI 10.1016/j.stem.2010.01.005

SUMMARY Pietersen and van Lohuizen, 2008). They reside in two main complexes, termed Polycomb repressive complex 1 and 2 Polycomb group (PcG) proteins are essential regula- (PRC1 and PRC2). PRC2 and trithorax group (trxG) proteins tors of stem cells. PcG and trithorax group proteins mark developmental regulator gene promoters with bivalent mark developmental regulator gene promoters with domains consisting of overlapping repressive and activating bivalent domains consisting of overlapping repres- histone modifications to keep developmental regulators sive and activating histone modifications to keep ‘‘poised’’ for activation in embryonic stem cells (ESCs) (Bernstein them poised for activation in embryonic stem cells. et al., 2006; Spivakov and Fisher 2007; Mendenhall and Bern- stein, 2008). Likewise, in adult stem cells, developmental regula- Bmi1, a component of PcG complexes, maintains tors that govern lineage specification are supposedly repressed the self-renewal capacity of adult stem cells, but its epigenetically to maintain their multipotency (Pietersen and van role in multipotency remains obscure. Here we Lohuizen, 2008). Among PcG proteins, Bmi1, a component of show that Bmi1 is critical for multipotency of hemato- PRC1, and its role in the inheritance of the stemness of adult poietic stem cells (HSCs) and multipotent progeni- stem cells have been well characterized (Molofsky et al., 2003; tors (MPPs). B cell lineage developmental regulator Park et al., 2003; Iwama et al., 2004). Bmi1 maintains the self- genes, Ebf1 and Pax5, appeared to be transcription- renewal capacity of adult stem cells, at least partially, by repres- ally repressed by bivalent domains before lineage sing the Ink4a/Arf locus, which encodes a cyclin-dependent Ink4a Arf commitment. Loss of Bmi1 resulted in a resolution kinase inhibitor, p16 , and a tumor suppressor, p19 of bivalent domains at the Ebf1 and Pax5 loci, leading (Jacobs et al., 1999; Oguro et al., 2006). Deletion of both Ink4a to their premature expression in HSC/MPPs accom- and Arf in Bmi1-deficient mice substantially restored the defective self-renewal capacity of HSCs (Oguro et al., 2006) and neural panied by accelerated lymphoid specification and stem cells. On the other hand, comprehensive genome-wide a marked reduction in HSC/MPPs. Thus, Bmi1 is analyses demonstrated that PcG proteins directly repress a large required to reinforce bivalent domains at key devel- cohort of developmental regulators in ESCs and, by doing so, opmental regulator gene loci to maintain lineage maintain ESCs in an undifferentiated pluripotent state (Boyer specification poised for activation in adult stem cells. et al., 2006; Lee et al., 2006). In adult stem cells, however, target genes for Bmi1 involved in stem cell functions other than Ink4a/ Arf have not been well documented, and the contribution of Bmi1 INTRODUCTION to the multipotency and lineage specification of stem cells remains unclear. Here we show that Bmi1 plays a crucial role All hematopoietic cells are generated from hematopoietic stem in preventing premature lineage specification of HSCs and multi- cells (HSCs) via a hierarchy of progenitor populations with potent progenitors (MPPs) through reinforcement of bivalent restricted differentiation potential. Although many transcription domains at key developmental gene loci. factors that play key roles in the lineage commitment of HSCs have been identified, the precise regulation of their expression RESULTS in HSCs remains to be elucidated (Dias et al., 2008; Laslo et al., 2008; Orkin and Zon, 2008). Polycomb group (PcG) Premature Derepression of B Cell Lineage-Regulator proteins are involved in the maintenance of gene silencing Genes in Bmi1-Deficient Multipotent Progenitors through chromatin modifications and have been implicated in Target genes for Bmi1 involved in stem cell functions other than stem cell self-renewal (Sparmann and van Lohuizen, 2006; Ink4a/Arf have not been well documented. To address this issue,

Cell Stem Cell 6, 279–286, March 5, 2010 ª2010 Elsevier Inc. 279 Cell Stem Cell Bmi1 Inhibits Premature Differentiation of HSCs

-/- Figure 1. Premature Derepression of B -/- À/À AB Lineage Regulator Genes in Bmi1 Ink4a- -/- À/À -/- Arf HSCs/MPPs Ink4a-Arf -/- -/- Ink4a-Arf -/- -/- (A) Heat map of the expression of B lineage-

O Wild-type Ink4a-Arf Bmi1 Bmi1 2 associated genes in a microarray analysis of H Wild-typeInk4a-ArfBmi1 Bmi1 À Pou2af1 (OBF1) IL-7Ra KSL HSCs/MPPs. Red, higher expression; Sfpi1 Igll1 (l5) green, lower expression than wild-type average. Vpreb1 Ebf1 Tcfe2a (B) Semiquantitative RT-PCR analysis of B lineage Rag1 St6gal1 (CD75) Il7r developmental genes, Bmi1, Ink4a, and Arf in wild- Ms4a1 (CD20) À/À À/À À/À Ebf1 type, Ink4a-Arf , Bmi1 , and Bmi1 Ink4a- Cd19 À/À À Vpreb3 Arf IL-7Ra KSL cells. Pax5 Blnk (C and D) Quantitative RT-PCR analysis of Ebf1 (C) Irf4 Pax5 Cd79a and Pax5 (D). mRNA levels in indicated progenitor Lef1 Blk Igll1 fractions puriﬁed from recipient mouse BM repo- Zfpn1a3 (Aiolos) À/À À/À Vpreb1 pulated by wild-type, Ink4a-Arf , and Bmi1 Ighg À/À Vpreb1/Vpreb2 Ink4a-Arf BM cells were normalized to Hprt1 Bmi1 Prdm1 (Blimp1) expression. Expression levels relative to those Cd79a (mb-1) Il7r Ink4a in the wild-type pro-B cells are shown as the Cd79b (B29) mean ± standard deviation (SD) for triplicate Tcfe2a (E2A) Arf Zfpn1a1 (Ikaros) analyses. **p < 0.01. See also Figure S1. Gapdh Sfpi1 (PU.1)

<1/8 1/4 1/2 1 2 4 8< C D to those in wild-type pre-B cells (Figures 1 Wild-type 1 Wild-type 1C and 1D). To confirm these findings at -/- -/- ** Ink4a-Arf Ink4a-Arf a single-cell level, we compared the dere- -/- -/- -/- -/- Hprt1 ) Hprt1 ) Bmi1 Ink4a-Arf ** Bmi1 Ink4a-Arf 0.1 0.1 pressed Pax5 protein levels by immunos- taning with an ArrayScan, a cell image analyzer designed for high-capacity 0.01 0.01 (normalized to (normalized to automated fluorescence imaging and Relative Ebf1 expression Relative Pax5 expression quantitative analysis. Two hundred cells were scanned and the fluorescence - KSL KSL CLP - KSL KSL CLP + pre-B CMP GMP MEP + pre-B CMP GMP MEP

Flt3 Flt3 intensity was plotted in Figure S1C. Dere- Flt3 ETP/DN2 Flt3 ETP/DN2 pression of Pax5 protein was evident in HSC/MPPs and LMPPs compared to we compared gene expression profiles of HSC/MPP cells in IL- levels of Pax5 in wild-type cells. Common lymphoid progenitors 7RaÀc-Kit+Sca-1+LineageÀ (IL-7RaÀKSL) fractions from bone (CLPs) that possess both B and T lymphoid potential also ex- marrow (BM) of wild-type, Ink4a-ArfÀ/À, Bmi1À/À, and Bmi1À/À pressed significantly higher levels of Ebf1 and Pax5 in the Ink4a-ArfÀ/À mice. Unexpectedly, a set of genes specific to the absence of Bmi1, whereas Ebf1 and Pax5 were strictly silenced B lymphoid lineage, but not T lymphoid or myeloid lineage, in T cell lineage progenitors. These results suggest that Bmi1 were prematurely derepressed in both Bmi1À/À and Bmi1À/À functions in the silencing of Ebf1 and Pax5 in HSC/MPP/LMPPs Ink4a-ArfÀ/À HSC/MPPs (Figure 1A; Figures S1A and S1B avail- and myeloid progenitors and that Ink4a and Arf are not involved able online). Among these genes, Ebf1 and Pax5 are B cell- in the transcriptional regulation of these genes. specific master transcription factor genes that function in B lineage specification at the earliest stage of B cell differentiation, Direct Repression of Ebf1 and Pax5 Expression and other genes are their downstream targets (Dias et al., 2008; by Bmi1 through Repressive Histone Modifications Laslo et al., 2008). Semiquantitative RT-PCR confirmed that at Their Promoters Ebf1, Pax5, and downstream targets were derepressed in To confirm this, we characterized the Ebf1 and Pax5 promoters Bmi1À/À and Bmi1À/ÀInk4a-ArfÀ/À IL-7RaÀKSL cells, while the by conducting chromatin immunoprecipitation (ChIP) assays. To expression of upstream genes, Sfpi1 (PU.1), Tcfe2a (E2a), and obtain enough cells, we used BM lineage marker-negative (Lin- Il7r, was not altered (Figure 1B). We next quantified the Ebf1 eageÀ) progenitors depleted of cells committed to the lymphoid, and Pax5 expression in each progenitor fraction by quantitative myeloid, and erythroid lineages. Binding of Bmi1 to both of the RT-PCR. Because the BM environment of Bmi1À/À and Bmi1À/À Ebf1 promoters (Roessler et al., 2007) and to the Pax5 promoter Ink4a-ArfÀ/À mice is unable to support HSCs (Oguro et al., 2006), was detected, but not to the promoter of Il7r, a gene upstream of we purified progenitors from wild-type recipient BM reconsti- Ebf1 (Figures 2A and 2B). Bivalent domains in ESCs consist of tuted with Bmi1À/ÀInk4a-ArfÀ/À BM cells to exclude the environ- a repressive mark, trimethylation of histone H3 at lysine 27 mental effect. Ebf1 and Pax5 were derepressed in Flt3ÀIL- (H3K27me3) catalyzed by PRC2, and an active mark, trimethyla- 7RaÀKSL HSC/MPPs and Flt3+IL-7RaÀKSL lymphoid-primed tion of histone H3 at lysine 4 (H3K4me3) catalyzed by trxG multipotent progenitors (LMPPs), and even in myeloid proteins. PRC1 catalyzes the monoubiquitination of histone committed progenitors including common myeloid progenitors H2A (H2Aub1) at lysine 119 and thereby impedes RNA poly- (CMPs) and megakaryocyte/erythroid progenitors (MEPs), merase II elongation. Bivalent domains that carry both PRC2 although the expression levels were considerably low compared and PRC1 are efficiently retained upon the differentiation of

280 Cell Stem Cell 6, 279–286, March 5, 2010 ª2010 Elsevier Inc. Cell Stem Cell Bmi1 Inhibits Premature Differentiation of HSCs

Figure 2. Recruitment of Bmi1 and Histone A Ebf1 exon1a Ebf1 exon1b Pax5 exon1 +1 +1 +1 Modiﬁcation Status at the Ebf1 and Pax5 -1000 -2000 -1000 -4000 -3000 -2000 -1000 Promoter Loci (A) Diagram illustrating the genomic context of

Ebf1-1 Ebf1-2 Ebf1-3Ebf1-4 Pax5-1 Pax5-2 Pax5-3 Ebf1 and Pax5 promoter loci. Regions ampliﬁed - by site-speciﬁc quantitative PCR are indicated by B D Wild-type (Lin ) Wild-type -/- -/- - a-Bmi1 a-H3K4me3 Bmi1 Ink4a-Arf (Lin ) bars. -/- -/- 1.0 Bmi1 Ink4a-Arf 12 Wild-type (B) (B and C) ChIP analyses of wild-type (gray bar) or * Bmi1-/-Ink4a-Arf-/- (B) Bmi1À/ÀInk4a-ArfÀ/À (green bar) LineageÀ cells in 0.8 10 Ebf1 and Pax5 promoter regions and Il7r and 8 * 0.6 Actb control promoter regions via anti-Bmi1 (B) 6 * or anti-H2Aub1 antibody (C). Percentages of input 0.4 % Input % Input * 4 DNA are shown as the mean ± SD for triplicate ** ** 0.2 2 analyses. *p < 0.05, **p < 0.01. (D and E) ChIP analyses of wild-type (gray bar) or 0 0 Bmi1À/ÀInk4a-ArfÀ/À (green bar) LineageÀ cells

Il7r Il7r À/À Actb Actb and wild-type (faint yellow bar) or Bmi1 Ink4a- Ebf1-1 Ebf1-2 Ebf1-3 Ebf1-4 Pax5-1 Pax5-2 Pax5-3 Ebf1-1 Ebf1-2 Ebf1-3 Ebf1-4 Pax5-1 Pax5-2 Pax5-3 ArfÀ/À (yellow bar) IgMÀB220+ B cells via anti- - H3K4me3 (D) or anti-H3K27me3 (E) antibody. C E Wild-type (Lin ) Wild-type -/- -/- - a-H2Aub1 a-H3K27me3 Bmi1 Ink4a-Arf (Lin ) Percentages of input DNA are shown as the -/- -/- Bmi1 Ink4a-Arf 16 Wild-type (B) mean ± SD for triplicate analyses. *p < 0.05; 1.2 -/- -/- 14 Bmi1 Ink4a-Arf (B) **p < 0.01. See also Figure S2. 1.0 12 ** * 0.8 10 ** ** ** ** 0.6 8 ** ** % Input % Input 6 0.4 ** 4 * differentiation block at the pre-B stage in * ** 0.2 ** 2 * Bmi1À/À BM, and restored their numbers 0 0 only partially (Figure 3A). As described

Il7r Il7r À/À Actb Actb above, the Bmi1 BM environment is Ebf1-1 Ebf1-2 Ebf1-3 Ebf1-4 Pax5-1 Pax5-2 Pax5-3 Ebf1-1 Ebf1-2 Ebf1-3 Ebf1-4 Pax5-1 Pax5-2 Pax5-3 unable to support HSC self-renewal and this appeared to be true for B cell devel- ESCs into somatic stem/progenitor cells and account for the vast opment in Bmi1À/À mice. Wild-type BM cells were transplanted majority of developmental regulator gene promoters (Ku et al., into sublethally irradiated Bmi1À/À recipient mice, and as ex- 2008). The Ebf1 and Pax5 promoters carried both H3K27me3 pected, B cell differentiation was significantly compromised at and H3K4me3 along with H2Aub1 (bivalent state) in wild-type the pre-B cell stage in the Bmi1À/À BM environment BM LineageÀ progenitors, but lost H3K27me3 in wild-type B (Figure S3A). A similar differentiation block was also reported cells (active monovalent state) (Figures 2C–2E). Of note, they in B cells deficient in PRC2 genes, Ezh2 and Yy1, and has lost, at least partially but to a significant extent, not only been attributed to defective somatic rearrangements of the H2Aub1 but also H3K27me3 in Bmi1À/ÀInk4a-ArfÀ/À LineageÀ immunoglobulin heavy chain gene (Igh)(Su et al., 2003; Liu progenitors compared with levels in wild-type cells, whereas et al., 2007). However, this was not the case in Bmi1À/À B cells H3K4me3 levels were left unchanged. These results indicate (Figure S3B). that Bmi1 directly represses the expression of both Ebf1 and Pax5 by keeping their promoters in a bivalent state and suggest Enhanced B Lymphopoiesis at the Expense of that PRC1 affects PRC2 function or stabilizes the H3K27me3 T Lymphopoiesis in Bmi1À/ÀInk4a-ArfÀ/À BM Cells modification (Figure S2). To evaluate the real impact of loss of Bmi1 on lymphocyte differentiation, we transplanted Bmi1À/ÀInk4a-ArfÀ/À BM cells into BM Environment Unable to Support B Lymphopoiesis wild-type recipient mice and analyzed their differentiation in in Bmi1-Deficient Mice a normal BM environment. In wild-type recipients, the develop- Based on these findings, we next investigated the impact of mental arrest of Bmi1À/ÀInk4a-ArfÀ/À cells at the pre-B stage derepression of Ebf1 and Pax5 in HSC/MPP/LMPPs. Overex- was completely released, and, as expected from the derepression of Ebf1 and Pax5 in hematopoietic progenitors is pressed expression of Ebf1 and Pax5, Bmi1À/ÀInk4a-ArfÀ/À reportedly sufficient to skew the developmental potential of hematopoietic cells generated 4-fold more CLPs and increased lymphoid progenitors toward the B cell lineage at the expense numbers of B lineage progenitors (Figure 3B). These results of T lymphopoiesis (Souabni et al., 2002; Zhang et al., 2003; Pon- confirmed that an environment unable to support B lymphopoi- gubala et al., 2008). In contrast, loss of each PRC1 component in esis largely accounts for the impaired lymphopoiesis in Bmi1À/À mice commonly caused arrested lymphoid development at pre- Ink4a-ArfÀ/À mice and demonstrated that lymphoid specification B and CD4ÀCD8À double-negative (DN) stages (van der Lugt, and subsequent B lymphopoiesis is unexpectedly enhanced by et al., 1994; Raaphorst et al., 2001). Deletion of both Ink4a and loss of Bmi1. In contrast, thymic reconstitution by Bmi1À/ÀInk4a- Arf or Arf only in Bmi1À/À mice partially restored splenocyte ArfÀ/À BM cells was severely compromised to one-tenth and thymocyte numbers (Jacobs et al., 1999) and largely (Figure 3C). Numbers of thymic progenitors were significantly released a differentiation block in thymic T cells (Miyazaki decreased in the earliest progenitor fraction, early thymic et al., 2008). However, deletion of Ink4a/Arf did not release the progenitors (ETPs), but the reduction in T cells at the later stages

Cell Stem Cell 6, 279–286, March 5, 2010 ª2010 Elsevier Inc. 281 Cell Stem Cell Bmi1 Inhibits Premature Differentiation of HSCs

Figure 3. Bmi1À/ÀInk4a-ArfÀ/À BM Cells ABWild-type Ink4a-Arf-/- Show Enhanced B Lymphopoiesis at the -/- Bmi1 Wild-type Expense of T Lymphopoiesis and Reduced -/- -/- Bmi1 Ink4a-Arf 5 Ink4a-Arf-/- 2.0 HSC/MPP Numbers in a Normal Environ- Bmi1-/-Ink4a-Arf-/- ment 4 1.5 (A) Numbers of BM B lineage progenitors and 3 splenic B cells per body weight of 4-week-old 1.0 2 mice. Data were normalized relative to the wild-

Cell number type and are shown as the mean ± SD (5 % n % 7). 0.5 1 (relative to wild-type) (relative to wild-type) (B) Numbers of donor-derived BM B cell lineage Cell number / body weight 0 0 progenitors and splenic B cells recovered from

.D) the wild-type recipient mice repopulated by BM KSL CLP HSC MPP CLP pro-B LMPP pro-B cells of indicated mutant mice. Data were normal- pre-pro-B mature B splenic B pre-pro-B mature Bsplenic B immature B immature B pre-B (Fr.C')pre-B (Fr.D) pre-B (Fr.C')pre-B (Fr ized relative to the wild-type and are shown as the CD mean ± SD (5 % n % 9). (C) Cell counts of BM, spleen, thymus, and periph- Wild-type Wild-type Ink4a-Arf-/- Ink4a-Arf-/- eral blood (PB) of wild-type recipient mice repopu- 1.5 Bmi1-/-Ink4a-Arf-/- 2.0 Bmi1-/-Ink4a-Arf-/- lated by BM cells of indicated mutant mice. Data were normalized relative to the wild-type and are 1.5 1.0 shown as the mean ± SD (8 % n % 20). 1.0 (D) Numbers of donor-derived thymic T cell lineage +

Cell number progenitor cells and B220 B cells recovered from 0.5 Cell number 0.5 the wild-type recipient mice repopulated by BM (relative to wild-type) (relative to wild-type) 0 0 cells of indicated mutant mice. Data were normal-

BM Spleen Thymus PB + ized relative to the wild-type and are shown as the DP ETP DN3 DN4 DN2a DN2b % % B220 mean ± SD (4 n 7). CD4SP CD8SP E (E) Thymic reconstitution potential of LMPPs on Intrathymic Intravenous intrathymic and intravenous injection. Thymic reconstitution potential was evaluated by intrathy- 100 100 mic and intraveneous injections of LMPPs recov- 75 75 ered from the wild-type recipient mice repopulated by BM cells of indicated mutant mice. 250 50 50 freshly isolated donor LMPPs (B6-Ly5.2) were mixed with the same number of competitor wild- 25 25 % of donor cells % of donor cells type LMPPs (B6-Ly5.1/Ly5.2 F1) and injected

0 0 into each thymic lobe of recipient B6-Ly5.1 mice irradiated with a dose of 6.0Gy. 1000 freshly iso- -/- -/- -/- -/- -/- -/- lated donor LMPPs (B6-Ly5.2) were mixed with Bmi1 Bmi1 Wild-type Wild-type the same number of competitor wild-type LMPPs Ink4a-Arf Ink4a-Arf Ink4a-Arf Ink4a-Arf (B6-Ly5.1/Ly5.2 F1) and intravenously injected into recipient B6-Ly5.1 mice irradiated with a dose of 6.0Gy. Each dot shows the percent chimerism of donor cells in the recipient thymic DP fraction analyzed at 20 days after transplantation. Bars indicate the mean (intrathymic injection, WT; n = 6, others; n = 10, intravenous injection, n = 10). See also Figure S3. was relatively moderate (Figure 3D). Of importance, Bmi1À/À ient mice. As expected, Ebf1 knockdown efficiently suppressed Ink4a-ArfÀ/À LMPPs did not show any defects in migration/ B cell repopulation by both Ink4a-ArfÀ/À and Bmi1À/ÀInk4a- homing to and expansion in the thymus in experiments with intra- ArfÀ/À HSCs and significantly restored a drop in the KSL cell thymic and intravenous injections (Figure 3E). Furthermore, the fraction (HSCs/MPPs/LMPPs) repopulated by Bmi1À/ÀInk4a- growth and differentiation of most immature thymic progenitors, ArfÀ/À HSCs (Figure S3D). Unexpectedly, however, Ebf1 knock- ETPs and DN2 cells, from mice reconstituted with Bmi1À/ÀInk4a- down profoundly affected the thymic T cell repopulation of both ArfÀ/À BM cells, were not compromised in culture (Figure S3C). Ink4a-ArfÀ/À and Bmi1À/ÀInk4a-ArfÀ/À HSCs, probably because Thus, the impaired thymic reconstitution was assumed to be of some adverse effects of the lentiviral knockdown system on due to the reduced production of T cell progenitors in BM that T cell repopulation (data not shown). These findings suggest subsequently migrate and home to the thymus. Of interest, that premature expression of Ebf1 is one of the direct mecha- HSCs, MPPs, and LMPPs were considerably decreased in abso- nisms responsible for premature differentiation of Bmi1À/À lute numbers (Figure 3B), further supporting enhanced prema- Ink4a-ArfÀ/À HSCs/MPPs and their enhanced commitment to ture differentiation of Bmi1À/ÀInk4a-ArfÀ/À multipotent stem/ B lineage lymphocytes. progenitor cells induced by derepressed expression of Ebf1 and Pax5. Loss of Bmi1 Enhances Commitment of LMPPs To confirm the direct effect of derepressed Ebf1 on deregu- and CLPs and trans-Differentiation of CMPs lated differentiation of Bmi1À/ÀInk4a-ArfÀ/À HSCs, we infected into the B Cell Lineage Ink4a-ArfÀ/À and Bmi1À/ÀInk4a-ArfÀ/À HSCs purified from To further evaluate the capacity for lineage commitment and primary recipient mice with a lentivirus expressing shRNA differentiation of Bmi1À/ÀInk4a-ArfÀ/À multipotent stem/progen- against Ebf1 and transplanted them into lethally irradiated recip- itor cells, we conducted lymphoid differentiation assays in vitro.

282 Cell Stem Cell 6, 279–286, March 5, 2010 ª2010 Elsevier Inc. Cell Stem Cell Bmi1 Inhibits Premature Differentiation of HSCs

A B C Figure 4. Enhanced Commitment of LMPP MPP LMPP À/À À/À 80 7 7 Bmi1 Ink4a-Arf LMPPs and CLPs to Wild-type -/- pro-B pro-B the B Cell Lineage In Vitro ) 6 ) 6 ) Ink4a-Arf 5 5 5 myeloid myeloid 60 Bmi1-/- 5 5 (A) Growth of LMPPs from indicated 4-week-old -/- -/- mutant mice. Freshly isolated cells were cultured Bmi1 Ink4a-Arf 4 4 40 in the presence of SCF, Flt3L, and IL-7, and the 3 3 results are shown as the mean ± SD for triplicate 2 2 20 cultures. Cell number (x10 Cell number (x10 Cell number (x10 1 1 (B and C) Numbers of CD43+B220+ pro-B cells and 0 0 0 Gr1+Mac1+ myeloid cells in culture. Freshly iso- 0 2 4 6 8 101214 -/- -/- Wild-type Ink4a-Arf-/- Bmi1 Wild-type Ink4a-Arf-/- Bmi1 Days Ink4a-Arf-/- Ink4a-Arf-/- lated MPPs (B) and LMPPs (C) from wild-type recipient BM repopulated by BM cells of indicated D LMPP on TSt-4 E CLP on TSt-4 F CMP on TSt-4 mutant mice were cultured in the presence of SCF, 100 100 100 N.D. Flt3L, IL-7, and IL-3, and were counted and N.D. analyzed by ﬂow cytometry after 14 days of

37 37 37 culture. Data are shown as the mean ± SD for triplicate cultures. (D–F) Limiting dilution analyses of B cell development from donor-derived LMPPs (D), CLPs (E), 10 10 10 and CMPs (F). Freshly isolated cells were cocul- % negative wells (B cell) % negative wells (B cell) 1/6 1/89 1/117 1/1.4 1/18 1/29 % negative wells (B cell) 1/112 tured with TSt-4 stromal cells, and the generation + + 0 25 50 75 100 125 150 0 5 10 15 20 25 30 0 100 200 300 400 of B220 CD43 pro-B cells in each well was de- Input cells per well Input cells per well Input cells per well tected by flow cytometry after 14 days of cocul- G LMPP on TSt-4/DLL1 H CLP on TSt-4/DLL1 K ture. The calculated frequencies are indicated 100 100 inside of the plot. Gray, wild-type; blue, Ink4a- ArfÀ/À; green, Bmi1À/ÀInk4a-ArfÀ/À. N.D., not PU.1 detected. 37 37 (G–J) Limiting dilution analyses of T cell development from donor-derived LMPPs (G) and CLPs IL-7Ra ? (H). Freshly isolated cells were cocultured with 10 10 1/17 TSt-4/DLL1 stromal cells, and the generation of Bmi1 + + % negative wells (T cell) % negative wells (T % negative wells (T cell) % negative wells (T 1/13 1/16 1/20 1/29 1/76 E2A CD4 CD8 DP cells in each well was detected by flow cytometry after 21 days of coculture. 0 102030405060 0 25 50 75 100 125 150 + + Input cells per well Input cells per well Ebf1 B220 CD43 pro-B cells were detected in the same culture of LMPPs (I) and CLPs (J). The calcu- I LMPP on TSt-4/DLL1 J CLP on TSt-4/DLL1 lated frequencies are indicated inside of the 100 100 À/À N.D. Pax5 plot. Gray, wild-type; blue, Ink4a-Arf ; green, N.D. Bmi1À/ÀInk4a-ArfÀ/À. 37 37 (K) Transcriptional networks involved in B cell specification and commitment. This study Rag1 l5 mb-1 unveiled that Bmi1 is a crucial negative regulator B29 of Ebf1 and Pax5 transcription. See also Figure S4. 10 10 VpreB CD19 1/240 Blk % negative wells (B cell) % negative wells (B cell) 1/37 1/11 1/240

0 20 40 60 80 100 120 0 50 100 150 200 250 Input cells per well Input cells per well B cell lineage differentiation Wild-type Ink4a-Arf-/- Bmi1-/-Ink4a-Arf-/-

We first cultured LMPPs and CLPs under B cell-inducing condi- elevated, with Bmi1À/ÀInk4a-ArfÀ/À LMPPs and CLPs being tions. In contrast to the LMPPs from Bmi1À/À mice, which re- more than 10-fold higher than with wild-type and Ink4a-ArfÀ/À sponded little to IL-7, the LMPPs from Bmi1À/ÀInk4a-ArfÀ/À cells (Figures 4D and 4E). This finding was also confirmed by mice were hypersensitive to IL-7 and showed enhanced prolifer- a single-cell analysis (Figure S4C). To our surprise, Bmi1À/À ation compared to those from wild-type and Ink4a-ArfÀ/À mice Ink4a-ArfÀ/À CMPs, in which Ebf1 and Pax5 expression was (Figure 4A). Comparable growth responses were observed with ectopically derepressed, gave rise to B cells, although at a low LMPPs and CLPs purified from wild-type recipients repopulated frequency (Figure 4F), suggesting that B cell potential is not by Bmi1À/ÀInk4a-ArfÀ/À BM cells (Figures S4A and S4B). Under completely suppressed in CMPs. In contrast, Bmi1À/ÀInk4a- conditions that allowed both myeloid and B cell differentiation, ArfÀ/À LMPPs and CLPs showed T cell production at levels Bmi1À/ÀInk4a-ArfÀ/À MPPs and LMPPs exhibited almost normal comparable to those of wild-type and Ink4a-ArfÀ/À mice under myeloid differentiation, but again generated B cells more inten- T cell-inducing conditions that involved TSt-4/DLL1 stromal cells sively (Figures 4B and 4C). To determine the probability of expressing a Notch ligand, Delta-like 1 (DLL1) (Figures 4G and Bmi1À/ÀInk4a-ArfÀ/À progenitors committing to the B cell 4H; Masuda et al., 2005). Of interest, however, intensive B cell lineage, we next performed limiting dilution assays via a cocul- production was observed with Bmi1À/ÀInk4a-ArfÀ/À LMPPs ture system with TSt-4 stromal cells (Masuda et al., 2005). The and CLPs but not with wild-type or Ink4a-ArfÀ/À cells under the frequency of progenitors producing B cells was markedly T cell conditions (Figures 4I and 4J). These results suggest that

Cell Stem Cell 6, 279–286, March 5, 2010 ª2010 Elsevier Inc. 283 Cell Stem Cell Bmi1 Inhibits Premature Differentiation of HSCs

although Bmi1À/ÀInk4a-ArfÀ/À progenitors retain the potential to observation that forced expression of Bmi1 enhances the self- generate T cells in vitro, their commitment status is highly renewal and multipotency of HSCs (Iwama et al., 2004), rein- skewed to the B cell lineage in BM, where the Notch signal is forcement of bivalent domains by Bmi1 at promoters of develop- relatively weak, because of derepressed expression of Ebf1 mental regulator genes and tumor suppressor genes might and Pax5. enhance the stemness of HSCs. This might be true for cancer We next analyzed the capacity of Bmi1À/ÀInk4a-ArfÀ/À MPPs stem cells. BMI1 expression is often upregulated in human and LMPPs to differentiate into the myeloid lineage. Colony cancers, particularly in cancer stem cells (Sparmann and van assays did not show any differences in the potential to differen- Lohuizen, 2006; Chiba et al., 2008). Higher levels of BMI1 could tiate into neutrophils, macrophages, erythroblasts, and mega- enhance self-renewal capacity and keep developmental regu- karyocytes but showed a slight reduction in numbers and colony lator genes poised for activation, leading to an augmentation of size (Figures S4D and S4E). A reduced proliferative capacity of the tumorigenicity of cancer stem cells. Bmi1À/ÀInk4a-ArfÀ/À cells was apparent in liquid cultures under In this study, we also demonstrated that Bmi1 balances myeloid culture conditions (Figures S4F and S4G). Ectopic B versus T cell lineage commitment of uncommitted lymphoid expression of Ebf1 and Pax5 reportedly blocks myeloid differen- progenitors by tuning the expression of Ebf1 and Pax5. Bmi1À/À tiation (Souabni et al., 2002; Pongubala et al., 2008). Thus, it is Ink4a-ArfÀ/À lymphoid progenitors preferentially committed to possible that even under myeloid culture conditions, commit- the B lymphoid lineage, but their T lineage commitment was ment into B cells is enhanced by the ectopic expression of severely inhibited in BM because of derepressed Ebf1 and Ebf1 and Pax5, resulting in reduced proliferation of myeloid Pax5 expression. Unexpectedly, Bmi1À/ÀInk4a-ArfÀ/À LMPP/ lineage cells. CLPs could produce T cells at a comparable level to controls when cocultured with TSt-4/DLL1 stromal cells. However, DISCUSSION Bmi1À/ÀInk4a-ArfÀ/À LMPPs did not show any defects in migration/homing to and expansion in the thymus, suggesting that the In this study, we unexpectedly identified that loss of Bmi1 leads reduced production of T cell progenitors in BM accounts for the to premature expression of B cell lineage developmental regu- impaired thymic reconstitution. It is possible that a strong Notch lator genes, Ebf1 and Pax5, and their downstream target genes signal from TSt-4/DLL1 cells overrode moderate levels of Ebf1 in HSCs/MPPs. In an Ink4a/Arf null genetic background (Bmi1À/À and Pax5 derepressed in Bmi1À/ÀInk4a-ArfÀ/À LMPPs. By Ink4a-ArfÀ/À), Bmi1-deficient HSCs that reacquired self-renewal contrast, moderate levels of Ebf1 and Pax5 were enough to capacity showed accelerated lymphoid specification, resulting overcome a weak Notch signal in BM and restrict commitment in a drastic increase in CLPs at the expense of the HSC/MPP of LMPPs to the B cell lineage. We further observed lineage pool size. In fact, HSCs, MPPs, and LMPPs were considerably conversion of CMPs to the B lineage. All these findings suggest decreased in absolute numbers. These findings clearly indicate that the PRC1 mark, H2Aub1, is required to maintain gene that Bmi1 functions as a negative regulator of lymphoid specifi- silencing of Ebf1 and Pax5 not only in HSCs/MPPs but also in cation in HSCs/MPPs and, by postponing lineage specification, myeloid progenitors (Figure S4H). contributes to the maintenance of their multipotent status Our findings also provide important insight into how PcG- (Figure 4K; Figure S4H). mediated repression of developmental regulator genes is Ebf1 and Pax5 appeared to be transcriptionally repressed by released during normal differentiation. The IL-7 signal activates bivalent domains coregulated by both PRC1 and PRC2 before Ebf1 transcription (Dias et al., 2008; Laslo et al., 2008) and lineage commitment. Bivalent domains in ESCs consist of Bmi1 function is supposedly regulated by its phosphorylation a repressive PRC2 mark, H3K27me3, and an active trxG mark, status (Voncken et al., 1999). Based on our findings that an H3K4me3. Recently, PRC1 was revealed to coregulate the vast absence of Bmi1 triggers a resolution of bivalency at the Ebf1 majority of developmental regulator gene promoters marked and Pax5 promoters, it would be of importance to uncover the with bivalent domains (Ku et al., 2008). Of note, the Ebf1 and crosstalk between the external signals, including IL-7 signal, Pax5 promoters lost, to a significant extent, not only the PRC1 and Bmi1 function in terms of the epigenetic regulation of devel- mark, H2Aub1, but also H3K27me3 in Bmi1À/ÀInk4a-ArfÀ/À Lin- opmental regulator gene transcription (Figure 4K). eageÀ progenitors, while H3K4me3 levels were left unchanged (Figure S2). Although the PRC1 mark, H2Aub1, reportedly EXPERIMENTAL PROCEDURES contributes to gene silencing by impeding RNA polymerase II elongation, our findings indicate that Bmi1 is required to maintain Mice Bmi1+/À mice and Ink4a-ArfÀ/À mice in a C57BL/6 (B6-Ly5.2) background were the Ebf1 and Pax5 promoters in a bivalent state, thus suggesting used. C57BL/6 (B6-Ly5.2) mice were purchased from Japan SLC (Shizuoka, that PRC1 or its epigenetic mark, H2Aub1, collaborates with Japan). C57BL/6 mice congenic for the Ly5 locus (B6-Ly5.1) were purchased PRC2 in the maintenance/stabilization of the H3K27me3 modifi- from Sankyo-Lab Service (Tsukuba, Japan). Mice were bred and maintained in cation at key developmental regulator gene loci. the Animal Research Facility of the Graduate School of Medicine, Chiba Stem cells are defined as cells with self-renewal and multiline- University in accordance with our institutional guidelines. age differentiation capacity. Bmi1 maintains the self-renewal capacity of HSCs by repressing the tumor suppressor genes Flow Cytometric Analysis and Sorting Antibodies used in flow cytometric experiments and cell surface protein represented by Ink4a and Arf (Sparmann and van Lohuizen, expression used to define hematopoietic cell types are listed in Supplemental 2006). The findings of this study demonstrate that Bmi1 keeps Experimental Procedures. Lineage marker antibodies for BM cells were Gr-1, differentiation programs poised for activation in HSCs/MPPs to Mac-1, Ter119, B220, CD4, and CD8a. Those for thymocytes were Gr-1, Mac-1, maintain their multipotency. Taken together with our previous Ter119, B220, CD4, CD8a, CD3, TCRgd, CD11c, and NK1.1. Flow cytometric

284 Cell Stem Cell 6, 279–286, March 5, 2010 ª2010 Elsevier Inc. Cell Stem Cell Bmi1 Inhibits Premature Differentiation of HSCs

experiments were performed with FACSVantage SE, FACSAria II (BD Biosci- Dias, S., Xu, W., McGregor, S., and Kee, B. (2008). Transcriptional regulation of ences), MoFlo (Beckman Coulter), and JSAN (Bay Bioscience). lymphocyte development. Curr. Opin. Genet. Dev. 18, 441–448. Iwama, A., Oguro, H., Negishi, M., Kato, Y., Morita, Y., Tsukui, H., Ema, H., Bone Marrow Transplantation Kamijo, T., Katoh-Fukui, Y., Koseki, H., et al. (2004). Enhanced self-renewal For analyses of the ability of mutant hematopoietic cells to differentiate in a of hematopoietic stem cells mediated by the polycomb gene product Bmi-1. normal environment, 2 3 106 BM mononuclear cells from 4-week-old B6- Immunity 21, 843–851. Ly5.2 mutant mice were transplanted into wild-type B6-Ly5.1 mice irradiated Jacobs, J.J., Kieboom, K., Marino, S., DePinho, R.A., and van Lohuizen, M. À/À with a dose of 9.5 Gy. The ability of the Bmi1 microenvironment to support (1999). The oncogene and Polycomb-group gene bmi-1 regulates cell prolifer- 6 B lymphopoiesis was evaluated by transplanting 2 3 10 wild-type BM cells ation and senescence through the ink4a locus. Nature 397, 164–168. (B6-Ly5.1) into 4-week-old wild-type and Bmi1À/À mice (B6-ly5.2) irradiated Ku, M., Koche, R.P., Rheinbay, E., Mendenhall, E.M., Endoh, M., Mikkelsen, with a dose of 6.0 Gy. T.S., Presser, A., Nusbaum, C., Xie, X., Chi, A.S., et al. (2008). Genomewide analysis of PRC1 and PRC2 occupancy identifies two classes of bivalent Cell Culture domains. PLoS Genet. 4, e1000242. For growth assays, sorted cells were cultured for 14 days in the presence of Laslo, P., Pongubala, J.M., Lancki, D.W., and Singh, H. (2008). Gene regula- 20 ng/ml stem cell factor (SCF), 20 ng/ml Flt3 ligand (Flt3L), and 20 ng/ml tory networks directing myeloid and lymphoid cell fates within the immune IL-7, with or without 20 ng/ml IL-3 (Peprotech). For the detection of B cells, system. Semin. Immunol. 20, 228–235. sorted cells were cultured for 14 days with TSt-4 stromal cells (Masuda et al., 2005) in 48-well plates. For the detection of T cells, sorted cells were Lee, T.I., Jenner, R.G., Boyer, L.A., Guenther, M.G., Levine, S.S., Kumar, R.M., cultured for 21 days with TSt-4/DLL stromal cells (Masuda et al., 2005). The Chevalier, B., Johnstone, S.E., Cole, M.F., Isono, K., et al. (2006). Control of frequencies of B and T cell commitment were determined by limiting dilution developmental regulators by Polycomb in human embryonic stem cells. Cell assays and calculated with L-Calc software (StemCell Technologies). All 125, 301–313. culture experiments were performed with RPMI 1640 medium supplemented Liu, H., Schmidt-Supprian, M., Shi, Y., Hobeika, E., Barteneva, N., Jumaa, H., with 10% fetal calf serum, 2 mM L-glutamine, 100 U/ml penicillin, 100 mg/ml Pelanda, R., Reth, M., Skok, J., Rajewsky, K., and Shi, Y. (2007). Yin Yang 1 is streptomycin, 50 mM2-b-mercaptoethanol (Sigma), 1 mM sodium pyruvate, a critical regulator of B-cell development. Genes Dev. 21, 1179–1189. and 0.1 mM nonessential amino acid solution (GIBCO BRL). Masuda, K., Kubagawa, H., Ikawa, T., Chen, C.C., Kakugawa, K., Hattori, M., Kageyama, R., Cooper, M.D., Minato, N., Katsura, Y., and Kawamoto, H. Statistical Analysis (2005). Prethymic T-cell development defined by the expression of paired Statistical significance was analyzed with an unpaired t test. immunoglobulin-like receptors. EMBO J. 24, 4052–4060. Mendenhall, E.M., and Bernstein, B.E. (2008). Chromatin state maps: New SUPPLEMENTAL INFORMATION technologies, new insights. Curr. Opin. Genet. Dev. 18, 109–115. Miyazaki, M., Miyazaki, K., Itoi, M., Katoh, Y., Guo, Y., Kanno, R., Katoh-Fukui, Supplemental Information includes Supplemental Experimental Procedures Y., Honda, H., Amagai, T., van Lohuizen, M., et al. (2008). Thymocyte prolifer- and four figures and can be found with this article online at doi:10.1016/ ation induced by pre-T cell receptor signaling is maintained through polycomb j.stem.2010.01.005. gene product Bmi-1-mediated Cdkn2a repression. Immunity 28, 231–245. Molofsky, A.V., Pardal, R., Iwashita, T., Park, I.K., Clarke, M.F., and Morrison, ACKNOWLEDGMENTS S.J. (2003). Bmi-1 dependence distinguishes neural stem cell self-renewal from progenitor proliferation. Nature 425, 962–967. We thank Maarten van Lohuizen and Ronald A. DePinho for providing Bmi1+/À Oguro, H., Iwama, A., Morita, Y., Kamijo, T., van Lohuizen, M., and Nakauchi, mice and Ink4a+/À mice, respectively, Naohito Nozaki for the Bmi1 antibody, H. (2006). Differential impact of Ink4a and Arf on hematopoietic stem cells and and Yuji Yamazaki for flow cytometric sorting. H.O. is supported by a postdoc- their bone marrow microenvironment in Bmi1-deficient mice. J. Exp. Med. toral fellowship from the Japanese Society for the Promotion of Science. This 203, 2247–2253. work was supported in part by Grants-in-Aid for Scientific Research (#20052009) and for the Global COE Program (Global Center for Education Orkin, S.H., and Zon, L.I. (2008). Hematopoiesis: An evolving paradigm for and Research in Immune System Regulation and Treatment), MEXT, Japan, stem cell biology. Cell 132, 631–644. a Grant-in-Aid for the Core Research for Evolutional Science and Technology Park, I.K., Qian, D., Kiel, M., Becker, M.W., Pihalja, M., Weissman, I.L., Morri- (CREST) from the Japan Science and Technology Corporation (JST), and son, S.J., and Clarke, M.F. (2003). Bmi-1 is required for maintenance of adult a grant from the Takeda Science Foundation. self-renewing haematopoietic stem cells. Nature 423, 302–305. Pietersen, A.M., and van Lohuizen, M. (2008). Stem cell regulation by poly- Received: May 20, 2009 comb repressors: Postponing commitment. Curr. Opin. Cell Biol. 20, 201–207. Revised: November 25, 2009 Pongubala, J.M., Northrup, D.L., Lancki, D.W., Medina, K.L., Treiber, T., Ber- Accepted: January 8, 2010 tolino, E., Thomas, M., Grosschedl, R., Allman, D., and Singh, H. (2008). Tran- Published: March 4, 2010 scription factor EBF restricts alternative lineage options and promotes B cell fate commitment independently of Pax5. Nat. Immunol. 9, 203–215. REFERENCES Raaphorst, F.M., Otte, A.P., and Meijer, C.J. (2001). Polycomb-group genes as regulators of mammalian lymphopoiesis. Trends Immunol. 22, 682–690. Bernstein, B.E., Mikkelsen, T.S., Xie, X., Kamal, M., Huebert, D.J., Cuff, J., Fry, B., Meissner, A., Wernig, M., Plath, K., et al. (2006). A bivalent chromatin Roessler, S., Gyo¨ ry, I., Imhof, S., Spivakov, M., Williams, R.R., Busslinger, M., structure marks key developmental genes in embryonic stem cells. Cell 125, Fisher, A.G., and Grosschedl, R. (2007). Distinct promoters mediate the regu- 315–326. lation of Ebf1 gene expression by interleukin-7 and Pax5. Mol. Cell. Biol. 27, 579–594. Boyer, L.A., Plath, K., Zeitlinger, J., Brambrink, T., Medeiros, L.A., Lee, T.I., Levine, S.S., Wernig, M., Tajonar, A., Ray, M.K., et al. (2006). Polycomb Souabni, A., Cobaleda, C., Schebesta, M., and Busslinger, M. (2002). Pax5 complexes repress developmental regulators in murine embryonic stem cells. promotes B lymphopoiesis and blocks T cell development by repressing Nature 441, 349–353. Notch1. Immunity 17, 781–793. Chiba, T., Miyagi, S., Saraya, A., Aoki, R., Seki, A., Morita, Y., Yonemitsu, Y., Sparmann, A., and van Lohuizen, M. (2006). Polycomb silencers control cell Yokosuka, O., Taniguchi, H., Nakauchi, H., and Iwama, A. (2008). The poly- fate, development and cancer. Nat. Rev. Cancer 6, 846–856. comb gene product BMI1 contributes to the maintenance of tumor-initiating Spivakov, M., and Fisher, A.G. (2007). Epigenetic signatures of stem-cell iden- side population cells in hepatocellular carcinoma. Cancer Res. 68, 7742–7749. tity. Nat. Rev. Genet. 8, 263–271.

Cell Stem Cell 6, 279–286, March 5, 2010 ª2010 Elsevier Inc. 285 Cell Stem Cell Bmi1 Inhibits Premature Differentiation of HSCs

Su, I.H., Basavaraj, A., Krutchinsky, A.N., Hobert, O., Ullrich, A., Chait, B.T., Voncken, J.W., Schweizer, D., Aagaard, L., Sattler, L., Jantsch, M.F., and van and Tarakhovsky, A. (2003). Ezh2 controls B cell development through histone Lohuizen, M. (1999). Chromatin-association of the Polycomb group protein H3 methylation and Igh rearrangement. Nat. Immunol. 4, 124–131. BMI1 is cell cycle-regulated and correlates with its phosphorylation status. van der Lugt, N.M., Domen, J., Linders, K., van Roon, M., Robanus-Maandag, J. Cell Sci. 112, 4627–4639. E., te Riele, H., van der Valk, M., Deschamps, J., Sofroniew, M., van Lohuizen, M., et al. (1994). Posterior transformation, neurological abnormalities, and Zhang, Z., Cotta, C.V., Stephan, R.P., deGuzman, C.G., and Klug, C.A. (2003). severe hematopoietic defects in mice with a targeted deletion of the bmi-1 Enforced expression of EBF in hematopoietic stem cells restricts lymphopoi- proto-oncogene. Genes Dev. 8, 757–769. esis to the B cell lineage. EMBO J. 22, 4759–4769.

286 Cell Stem Cell 6, 279–286, March 5, 2010 ª2010 Elsevier Inc.