RNA Helicase DDX5 Inhibits Reprogramming to Pluripotency by Mirna-Based Repression of RYBP and Its PRC1-Dependent and -Independent Functions

Article

RNA Helicase DDX5 Inhibits Reprogramming to Pluripotency by miRNA-Based Repression of RYBP and its PRC1-Dependent and -Independent Functions

Graphical Abstract Authors Huanhuan Li, Ping Lai, Jinping Jia, ..., Duanqing Pei, Andrew P. Hutchins, Hongjie Yao

Correspondence [email protected]

In Brief RNA-binding proteins have poorly deﬁned roles in somatic cell reprogramming. Li et al. show that the RNA-binding protein DDX5 erects an epigenetic barrier to reprogramming. DDX5 controls RYBP through microRNA- 125b to suppress speciﬁc somatic genes through deposition of H2AK119ub1 while activating an OCT4-KDM2B pluripotent gene program.

Highlights d DDX5 acts as a barrier to somatic cell reprogramming d DDX5 loss of function upregulates RYBP through microRNA-125b d Upregulated RYBP enhances H2AK119ub1 deposition at lineage-speciﬁc genes via PRC1 d DDX5 silencing activates the OCT4-KDM2B network through RYBP independently of PRC1

Li et al., 2017, Cell Stem Cell 20, 462–477 April 6, 2017 ª 2016 Elsevier Inc. http://dx.doi.org/10.1016/j.stem.2016.12.002 Cell Stem Cell Article

RNA Helicase DDX5 Inhibits Reprogramming to Pluripotency by miRNA-Based Repression of RYBP and its PRC1-Dependent and -Independent Functions

Huanhuan Li,1,2,7 Ping Lai,1,7 Jinping Jia,3,7 Yawei Song,1 Qing Xia,1 Kaimeng Huang,1 Na He,4 Wangfang Ping,1 Jiayu Chen,5 Zhongzhou Yang,1 Jiao Li,1 Mingze Yao,1 Xiaotao Dong,1 Jicheng Zhao,6 Chunhui Hou,4 Miguel A. Esteban,1 Shaorong Gao,5 Duanqing Pei,1 Andrew P. Hutchins,4 and Hongjie Yao1,8,* 1CAS Key Laboratory of Regenerative Biology, Guangdong Provincial Key Laboratory of Stem Cell and Regenerative Medicine, CAS Center for Excellence in Molecular Cell Science, Guangzhou Institutes of Biomedicine and Health, Chinese Academy of Sciences, Guangzhou 510530, China 2GZMU-GIBH Joint School of Life Sciences, Guangzhou Medical University, Guangzhou 511436, China 3Laboratory of Translational Genomics, National Cancer Institute, NIH, Bethesda, MD 20892, USA 4Department of Biology, Southern University of Science and Technology of China, Shenzhen 518055, China 5School of Life Sciences and Technology, Tongji University, Shanghai 200092, China 6National Laboratory of Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, China 7Co-ﬁrst author 8Lead Contact *Correspondence: [email protected] http://dx.doi.org/10.1016/j.stem.2016.12.002

SUMMARY followed by the upregulation of pluripotency genes (Brambrink et al., 2008; Li et al., 2010; Samavarchi-Tehrani et al., 2010; RNA-binding proteins (RBPs), in addition to their Stadtfeld et al., 2008). The phased and timed nature of these functions in cellular homeostasis, play important transitions suggests tight regulation of the reprogramming pro- roles in lineage specification and maintaining cellular cess (Golipour et al., 2012; Polo et al., 2012). It is now clear identity. Despite their diverse and essential functions, that the OSKM factors need to overcome a series of epigenetic which touch on nearly all aspects of RNA metabolism, barriers and pass through a sequence of distinct molecular the roles of RBPs in somatic cell reprogramming are and cellular events. Tremendous effort has been focused on studying the roles of chromatin-binding proteins and modifying poorly understood. Here we show that the DEAD- enzymes in reprogramming and pluripotency (Apostolou and box RBP DDX5 inhibits reprogramming by repressing Hochedlinger, 2013; Rais et al., 2013). However, much less has the expression and function of the non-canonical pol- been reported regarding the roles of RNA-binding proteins ycomb complex 1 (PRC1) subunit RYBP. Disrupting (RBPs) in reprogramming, pluripotency, and differentiation Ddx5 expression improves the efficiency of iPSC gen- despite the ability of the RBP LIN28A, to replace KLF4 in human eration and impedes processing of miR-125b, leading reprogramming (Yu et al., 2007), suggesting that other RBPs to Rybp upregulation and suppression of lineage- may also have critical roles in regulating the reprogramming specific genes via RYBP-dependent ubiquitination process. of H2AK119. Furthermore, RYBP is required for RBPs are involved in a wide range of regulatory pathways, PRC1-independent recruitment of OCT4 to the pro- from RNA metabolism to epigenetic regulation (Guallar and moter of Kdm2b, a histone demethylase gene that Wang, 2014). Importantly, an increasing number of studies have demonstrated that RBPs not only play constitutive roles promotes reprogramming by reactivating endoge- but also have important functions in the maintenance of cell iden- nous pluripotency genes. Together, these results tity (Zhang et al., 2016). DEAD-box RBPs have essential roles in reveal important functions of DDX5 in regulating re- cellular RNA metabolism, including transcription, pre-mRNA programming and highlight the importance of a splicing, ribosome biogenesis, transport, translation, and RNA Ddx5-miR125b-Rybp axis in controlling cell fate. decay (Linder and Jankowsky, 2011). DEAD-box RBP DDX5 is a component of the Drosha complex and plays important roles in regulating microRNA processing INTRODUCTION (Dardenne et al., 2014; Gregory et al., 2004). DDX5 associates with the long non-coding RNA (lncRNA) RNA component of Induced pluripotent stem cells (iPSCs) provide a powerful in vitro mitochondrial RNAase P (RMRP) and modulates Th17 cell experimental model to investigate the mechanisms controlling effector functions (Huang et al., 2015) and regulates CCCTC- cell fate conversion. The transduction of the ‘‘Yamanaka factors’’ binding factor (CTCF) transcriptional insulation (Yao et al., OCT4, SOX2, KLF4, and c-MYC (OSKM) into somatic cells 2010). In addition, p53 and SMAD proteins can both modulate sets off a series of phased transitions, initiated by the reduction microRNA metabolism through their association with DDX5 of somatic genes, a mesenchymal-epithelial transition (MET), (Davis et al., 2008; Suzuki et al., 2009).

462 Cell Stem Cell 20, 462–477, April 6, 2017 ª 2016 Elsevier Inc. AB C OSKM Ddx5 mRNA Ddx5 mRNA *** *** 1.2 shDdx5 #1 shDdx5 #2 10 *** shCtrl *** 1 level 8 0.8 AP staining 6 0.6

4 0.4

2 0.2 Oct4-GFP Oct4-GFP 12.4% Oct4-GFP Relative expression level 4.10% 13.7%

Relative expression 0 0 FACS MEFs iPSCs shCtrl mESCs shDdx5 shDdx5 #1 #2 α-DDX5 α-DDX5 FSC GFP α-β-Actin α-β-Actin

D E F OSKM OSKM Secondary MEFs +Dox +Vc Ctrl DDX5 *** 350 *** *** 600 *** 300 AP staining 250 400 200

150 5.45% 1.02% Oct4-GFP Oct4-GFP 100 200 FACS 50 Oct4-GFP colony number Oct4-GFP colony number

0 0 FSC

#1 #2 #1 #2 shCtrl shCtrl GFP Ddx5 Ddx5 Ddx5 Ddx5 h sh sh sh s

G OSKM HISecondary MEFs 100 OSKM +Dox +Vc ** OSKM+DDX5 ** 100 *** 80 180 80 ** 60 135 60 40 90 40 20

Oct4-GFP colony number *** 45 20 0 Oct4-GFP colony number Oct4-GFP colony number D0 D3 D6 D9 D12 0 0 Ctrl DDX5 rl 5 t X C D D J K MEFs MEFs OSKM+shCtrl 1000 OSKM 10000 OSKM+DDX5 OSKM+shDdx5 #1 *** *** *** 1000 OSKM+shDdx5 #2 ***

level 100 100 *** *** ression level 10

xp Snai1 Zeb2 10 1

** *** ee v Epcam Krt8 Snai1 Zeb2 ti 0.1 a Epithelial genes 1 Rel 0.01

Relative expression Epcam Cldn3 0.001

0.1 Mesenchymal genes Epithelial genes 0.0001 Mesenchymal genes

Figure 1. DDX5 Loss of Function Promotes Somatic Cell Reprogramming (A) qRT-PCR and western blot analysis for endogenous Ddx5 mRNA and protein levels in MEFs, mESCs, and iPSCs. (B) qRT-PCR and western blot to test DDX5 knockdown efficiency in MEFs transduced with OSKM together with either control shRNA or two Ddx5 shRNAs. (C) Top: AP-stained wells of a representative reprogramming experiment transfected with OSKM and either a control or Ddx5-targeting shRNA on day 9. Bottom: GFP+ cells analyzed by fluorescence-activated cell sorting (FACS) in reprogramming cells transfected with OSKM and either a control or Ddx5-targeting shRNA on day 12. (legend continued on next page) Cell Stem Cell 20, 462–477, April 6, 2017 463 In this study, we showed that DDX5 is a barrier for somatic cell Next, we transfected DDX5 along with OSKM into MEFs and reprogramming. DDX5 loss of function upregulates the non-ca- assessed the effects of DDX5 overexpression in MEFs (Fig- nonical PRC1 component RING1 and YY1 binding protein ure S1B). DDX5 significantly inhibited the formation of both (RYBP) through a mechanism involving microRNA (miRNA) AP+ and GFP+ colonies compared with the controls (Figures 125b, which leads to increased RYBP-mediated ubiquitination 1F and 1G). Furthermore, we observed that DDX5 overexpres- of histone H2A at K119 (H2AK119ub1) at lineage-specific loci sion significantly inhibited the timing of the appearance of iPSCs and repression of these genes. DDX5 loss of function enhances (Figure 1H). In addition, we observed a similar result with DDX5 reprogramming efficiency and facilitates RYBP-mediated OCT4 overexpression in the secondary MEF system (Figure 1I). The recruitment at the Kdm2b locus, which promotes reprogram- blocking effect of DDX5 during reprogramming was verified ming. This effect is independent of the RYBP-PRC1 complex with another overexpression system (pW-TRE tet-on inducible and, instead, an OCT4-RYBP complex forms, illustrating system) (Figures S1C–S1E). Consistent with the knockdown ex- the context-specific functions of RYBP. DDX5 loss of func- periments, exogenous DDX5 did not affect cell proliferation in the tion not only results in the enrichment of RYBP-stimulated presence of OSKM induction, indicating a growth-independent H2AK119ub1 at lineage-specific genes, thus suppressing effect of DDX5 on reprogramming efficiency (Figures S1F and the expression of these genes, but also activates the OCT4- S1G). In the early stage of reprogramming, the cells pass KDM2B pluripotent network and so assists reprogramming through a MET (Li et al., 2010; Samavarchi-Tehrani et al., cells to overcome an epigenetic barrier in the late phase of 2010). Overexpression of DDX5 perturbed the downregulation reprogramming. of Snai1 and Zeb2 (mesenchymal genes) and the upregulation of Epcam and Cldn3 (epithelial genes) on day 5 of reprogram- RESULTS ming (Figure 1J), whereas the pattern was reversed in the Ddx5 knockdown (Figure 1K). These results suggest that DDX5 Is A Negative Regulator of Somatic Cell DDX5 impairs the MET, whereas its loss promotes the MET. Reprogramming Overall, these data indicate that DDX5 is a negative regulator To investigate the roles of DDX5 in regulating reprogramming, of somatic cell reprogramming at both the early and late phases we first examined DDX5 expression levels in mouse embryonic of reprogramming. fibroblasts (MEFs), mouse embryonic stem cells (mESCs), and iPSCs by western blot and qRT-PCR. Ddx5 was significantly DDX5 Interacts with RYBP and Negatively Regulates the higher expressed in mESCs and iPSCs compared with MEFs Expression of RYBP (Figure 1A). We also noticed a gradual increase in Ddx5 expres- To investigate the mechanisms by which DDX5 inhibits reprog- sion during the reprogramming of MEFs with OSKM factors ramming, we searched for proteins that could interact with (Figure S1A). To investigate DDX5 function in reprogramming, DDX5. We generated DDX5-FLAG-tagged stable cell lines and we constructed two independent retroviral short hairpin performed protein purification and mass spectrometry (MS). RNA (shRNA) vectors that showed robust downregulation of Interestingly, MS analysis identified subunits of PRC1, including both the protein and mRNA of DDX5 (Figure 1B). Then, we RYBP, CBX7, and the catalytic subunit RING1B (Figure 2A). transduced these shRNA vectors together with OSKM into RYBP and CBX7 belong to mutually exclusive PRC1 com- OG2 MEFs bearing an Oct4-GFP transgenic reporter. Notably, plexes; RYBP has been identified as a noncanonical PRC1 sub- Ddx5 shRNAs, along with OSKM factors, significantly increased unit, whereas CBX7 is a canonical PRC1 subunit (Gao et al., the number of alkaline phosphatase (AP)-positive colonies 2012; Tavares et al., 2012). To confirm the MS results, we per- (an early marker of reprogramming) and the number of GFP+ formed FLAG co-immunoprecipitation (co-IP), which indicated colonies (a late marker of reprogramming) (Figures 1C and 1D), that RING1B, CBX7, and RYBP were retained in FLAG-DDX5- indicating that DDX5 negatively regulates reprogramming. expressing cells following IP with an anti-FLAG antibody but Furthermore, by using a secondary MEF reprogramming system not from cells expressing FLAG alone (Figure 2B). To verify an (Gao et al., 2013), we found that DDX5 loss of function signifi- endogenous interaction between DDX5 with CBX7, RYBP, and cantly promoted iPSCs generation in contrast to the control RING1B, co-IP experiments were performed using nuclear ex- (Figure 1E). tracts (NEs) from mESCs, which indicated that DDX5 interacts

(D) Number of GFP+ colonies of MEFs reprogrammed with OSKM and two independent shRNAs for Ddx5 (shDdx5) on day 12. (E) Number of GFP+ colonies in doxycycline-inducible secondary MEFs transduced with control shRNA or two shRNAs targeting Ddx5 on day 12. (F) Top: AP-stained wells of a representative reprogramming experiment transfected with OSKM and either empty plasmid (Ctrl) or pMXs-DDX5 on day 12. Bottom: GFP+ cells analyzed by FACS in reprogramming cells transfected with OSKM and either empty plasmid (Ctrl) or pMXs-DDX5 on day 12. (G) Number of GFP+ colonies of MEFs reprogrammed with OSKM and either empty plasmid (Ctrl) or pMXs-DDX5 on day 12. (H) Number of GFP+ colonies of MEFs reprogrammed with OSKM and either empty plasmid (Ctrl) or pMXs-DDX5 at different time points. (I) Number of GFP+ colonies in doxycycline-inducible secondary MEFs transduced with empty vector (Ctrl) or pMXs-DDX5 on day 12. (J) qRT-PCR analysis for the MET genes using RNA lysates from MEFs reprogrammed with OSKM and empty vector (Ctrl) or pMXs-DDX5 on day 5. Untransduced MEFs were used for normalization. (K) qRT-PCR analysis for the MET genes using RNA lysates from MEFs reprogrammed with OSKM and control shRNA or two shRNAs for Ddx5 on day 5. Untransduced MEFs were used for normalization. The data in (A), (B), (D), (E), and (G)–(K) are reported as mean values ± SD with the indicated signiﬁcance by using Student’s t test (*p < 0.05, **p < 0.01, ***p < 0.001). The data in (A), (B), (D), (G), (H), (J), and (K) are plotted from three biological replicates experiments, each with three technical repeats. The data in (E) and (I) are plotted from four biological replicates, each with three technical repeats. See also Figure S1.

464 Cell Stem Cell 20, 462–477, April 6, 2017 A B C

kDa Marker Ctrl-Flag DDX5-Flag 170 Input Flag-IP Co-IP 130 Ctrl-Flag + - + - 100 DDX5-Flag - + - + 70 DDX5 α-Flag (DDX5) Input IgG CBX7 RING1BRYBP 55 α-DDX5 α-RING1B 40 RING1B α-RYBP mESCs 35 RYBP α-CBX7 CBX7 25

E mESCs D Input Flag-IP #1 #2 #3 +/+ -/- -/- -/- Ctrl-Flag - + - - + Ddx5 Ddx5 Ddx5 Ddx5 DDX5-Flag - + - + + α-DDX5 RNase A - - - - + α-RYBP α-Flag (DDX5) 11.702.10 1.72

α-CBX7 α-CBX7 1 1.02 1.20 1.04 α-RING1B α-RING1B α-RYBP 11.041.06 1.07 α-β-Actin

F Rybp mRNA G OSKM OSKM+shCtrl * 10 OSKM+shDdx5#1 D3 D6 D9 D12 OSKM+shDdx5#2 8 * Ddx5 Ddx5 Ddx5 shCtrl sh shCtrl sh shCtrl sh shCtrl 6 α-DDX5 4 ** ** α-RYBP 2

Relatvie expression level expression Relatvie 1 1.15 1.01 1.58 1.161.56 1.91 2.18 0 α-β-Actin D3 D6 D9 D12

Figure 2. DDX5 Interacts with PRC1 Complexes, and Loss of DDX5 Leads to Upregulation of RYBP (A) FLAG-tagged DDX5 was puriﬁed from a stably transduced cell line and visualized by Coomassie blue staining, and interacting proteins were identiﬁed by mass spectrometry. (B) Detection of the interaction between DDX5 and PRC1 subunits by FLAG-IP with DDX5 using nuclear extracts. FLAG-tagged empty vector was used as a control. Membranes were immunoblotted with anti-FLAG, RING1B, RYBP, and CBX7 antibodies. (C) Detection of interaction between endogenous DDX5 and RING1B, RYBP, and CBX7 in mESCs by co-IP with anti-RING1B, RYBP, and CBX7 antibody, respectively, followed by immunoblotting for DDX5. Immunoglobulin G (IgG) was used as a negative control. (D) Detection of the interaction between DDX5 and PRC1 subunits treatment with/without RNase A. FLAG-tagged empty vector was used as a control. Mem- branes were immunoblotted with anti-FLAG, RING1B, RYBP, and CBX7 antibodies. (E) Western blot for PRC1 subunits using cell lysates from wild-type and three different Ddx5/ mESC lines. b-Actin was used as the loading control. (F) qRT-PCR analysis for Rybp mRNA using RNA lysates from MEFs reprogrammed with OSKM and control shRNA or two shRNAs targeting Ddx5 at different time points. MEFs transduced with OSKM and control shRNA were used for normalization. The data are reported as mean values ± SD with the indicated

(legend continued on next page)

Cell Stem Cell 20, 462–477, April 6, 2017 465 with PRC1 complex (Figure 2C). Furthermore, RNase A treat- To investigate which time point is essential for RYBP to pro- ment of lysates disrupted the DDX5/RING1B association but mote reprogramming, we constructed a doxycycline-inducible failed to disrupt the interaction of DDX5 with RYBP and CBX7, tet-on RYBP plasmid and examined the effects of RYBP on re- indicating that RNA is required for the DDX5/RING1B interac- programming by adding or withdrawing doxycycline at different tion but not for the DDX5/RYBP and DDX5/CBX7 interactions time points (Figures 3F and 3G). These time course treatments (Figure 2D). indicated that RYBP was beneficial for iPSC generation in To study the biological functions of DDX5, we generated both the early stage and the later stage of reprogramming (Fig- DDX5 knockout mESC clones using clustered regularly inter- ures 3F and 3G). We further found that RYBP overexpression spaced short palindromic repeats (CRISPR)/Cas9 technology decreased the expression of Snai1 and Zeb2 (mesenchymal (Figure S2A). DDX5 knockout was confirmed by DNA genes) but promoted the expression of Cdh1 and Krt8 (epithelial sequencing and western blot (Figures S2B and S2C). Loss of genes) on day 5 (Figure S3H), whereas RYBP knockdown DDX5 had no effect on the expression of pluripotent genes achieved the opposite result (Figure S3I), suggesting that (Oct4, Nanog, Sox2, and Utf1), indicating that DDX5 loss of RYBP facilitates the MET transition at the early stage of reprog- function has no effect on maintaining pluripotency (Figure S2D). ramming. Additionally, in agreement with the enhanced reprog- Moreover, we could not detect any obvious change in cell ramming kinetics, exogenous RYBP induced a substantial proliferation or cell cycle after DDX5 knockout (Figures S2E increase in the expression of pluripotent genes on day 11, sug- and S2F). gesting that RYBP promotes the activation of the pluripotency To investigate the genes regulated by loss of DDX5, we per- network at the late stage of reprogramming (Figure 3H). Neither formed RNA sequencing (RNA-seq) experiments in both wild- gain of function nor loss of function of RYBP caused any obvious type and Ddx5/ mESCs. The RNA-seq data indicated that effect on cell proliferation (Figures S3J and S3K). The resulting loss of DDX5 slightly upregulated RYBP expression (Figure S2G; OSKM with RYBP iPSCs were karyotypically normal (Figure S3L), Table S1). Loss of DDX5 had no effect on the protein levels of fully pluripotent, and capable of generating chimeras (Fig- other members of the PRC1 complex, whereas RYBP was upre- ure 3I). These data indicate that RYBP is a positive regulator of gulated in three separate mESC clones (Figure 2E). Importantly, reprogramming. both the mRNA and protein levels of RYBP were upregulated Our data indicate that DDX5 and RYBP play opposite roles in by DDX5 knockdown during reprogramming (Figures 2F reprogramming (Figures 1 and 3). RYBP is negatively regulated and 2G; Figure S2H), indicating enhanced RYBP expression by DDX5 in both mESCs and reprogramming cells (Figure 2). in both Ddx5/ mESCs and Ddx5 knockdown cells during Therefore, we speculated that RYBP might be downstream reprogramming. Therefore, these experiments suggest that of DDX5 in regulating reprogramming. We transfected DDX5 RYBP might be a downstream target of DDX5 in the process of and RYBP alone or together to establish clearer relationships reprogramming. between DDX5 and RYBP with respect to their importance during reprogramming. Our data indicated that overexpression of RYBP Enhances Reprogramming and Overwrites the RYBP could overcome the inhibitory effects of DDX5 overex- Negative Effect of DDX5 pression on reprogramming (Figure 3J). Conversely, loss of To assess the roles of RYBP in reprogramming, we knocked DDX5 had no influence on reprogramming in the absence of down the Rybp expression level in OSKM-mediated reprogram- RYBP, suggesting that RYBP loss of function abolishes the ming using two shRNA vectors that displayed approximately enhancing effect of DDX5 knockdown on reprogramming (Fig- 80%–90% knockdown efficiency at both the mRNA and protein ure 3K). These results indicate that RYBP is a downstream target levels (Figures S3A and S3B). RYBP knockdown greatly dimin- of DDX5 during the reprogramming process. ished the formation of AP+ and GFP+ colonies (Figures 3A and 3B). Moreover, the negative effect on reprogramming efficiency miRNA-125b Is Regulated by DDX5 and Targets Rybp at of RYBP knockdown was confirmed by using secondary MEFs the 30 UTR to Inhibit Reprogramming (Figure S3C). In addition, when we reprogrammed MEFs with DDX5 plays a vital role in cellular processes by modulating OSKM plus RYBP (Figure S3D), it resulted in a higher efficiency miRNA processing and gene transcription (Dardenne et al., of reprogramming compared with OSKM without RYBP, and 2014). Because the RYBP reprogramming phenotype so closely we observed an increased number of AP+ and GFP+ colonies matched the DDX5 loss-of-function phenotype, and because (Figures 3C and 3D). Reprogramming with RYBP also acceler- DDX5 and RYBP seemed mutually antagonistic to reprogram- ated the formation of GFP+ colonies compared with control ming (Figures 3J and 3K), we speculated that RYBP could cells (Figure 3E). A similar enhancing effect was confirmed be regulated by miRNA suppression and so we could iden- with RYBP overexpression in secondary MEFs (Figure S3E). tify DDX5 as an upstream regulator of RYBP. To search for In addition, the positive effect of RYBP on reprogramming miRNAs regulated by DDX5, we performed a comparative anal- was verified by the pW-TRE overexpression system (Figures ysis of miRNA-seq between wild-type and Ddx5/ mESCs and S3F and S3G). identified 73 miRNAs with significantly differential expression significance by using Student’s t test (*p < 0.05, **p < 0.01, ***p < 0.001). The data are plotted from three biological replicates experiments, each with three technical replicates. (G) Western blot for RYBP protein level using cell lysates from MEFs reprogrammed with OSKM and control shRNA or shRNAs for Ddx5 at different time points. b-Actin was used as the loading control. See also Figure S2 and Table S1.

466 Cell Stem Cell 20, 462–477, April 6, 2017 A OSKM B OSKM C OSKM

shCtrl shRybp #1 shRybp #2 *** Ctrl RYBP 150 **

120

60 6.45% 1.44% 1.87% 4.06% 14.0% Oct4-GFP Oct4-GFP Oct4-GFP Oct4-GFP Oct4-GFP 30 Oct4-GFP colony number 0 FSC FSC #1 #2 shCtrl GFP Rybp Rybp GFP sh sh

D E F OSKM+ tet-on RYBP *** OSKM *** OSKM 400 ** *** 300 300 OSKM+RYBP 350 *** 250 300 250 200 250 200 150 *** 200 150 150 100 100 100 50 ** 50 Oct4-GFP colony number

50 Oct4-GFP colony number

Oct4-GFP colony number 0 0 0 D0 D3 D6 D9 D12 Ctrl RYBP - Dox D0-D13 D1-D13 D3-D13 D5-D13 D7-D13 D9-D13 D11-D13 OSKM+ tet-on RYBP G 400 H +DOX 350 *** 300 *** *** l 90 MEFs ** ** e 250 v e OSK M *** l 80 n

200 o 70 OSKM+RYBP i *** 150 60 ess r

p 50

100 x 40 *** *** ve e Oct4-GFP colony number 50 i 30 0 20

Relat 10 - Dox 0 D0-D1 D0-D3 D0-D5 D0-D7 D0-D9 D0-D11 D0-D13 Nanog endo-Sox2 endo-Oct4 Esrrb Utf1 Lin28a +DOX

I J K *** 450 *** 800 RYBP-OSKM-iPSCs 360 600 Bright field GFP Germline 270 400 180 200 Nanog Merge DAPI 90 Oct4-GFP colony number Oct4-GFP colony number

0 0 Ctrl + - - - shCtrl + - - - DDX5 - + - + shDdx5 - + - + RYBP - - + + shRybp - - + +

Figure 3. RYBP Promotes Reprogramming and Overcomes the Negative Effect of DDX5 (A) Top: AP-stained wells of a representative reprogramming experiment transfected with OSKM and either a control or Rybp-targeting shRNA on day 10. Bottom: GFP+ cells analyzed by FACS in reprogramming cells transfected with OSKM and either a control or Rybp-targeting shRNA on day 12. (B) Number of GFP+ colonies of MEFs reprogrammed with OSKM and two independent shRNAs for Rybp (shRybp) on day 12. (C) Top: AP-stained wells of a representative reprogramming experiment transfected with OSKM and either empty plasmid or pMXs-RYBP on day 10. Bottom: GFP+ cells analyzed by FACS in reprogramming cells transfected with OSKM and either empty plasmid or pMXs-RYBP on day 12. (D) Number of GFP+ colonies on day 10 in MEFs reprogrammed with OSKM and empty vector or the indicated RYBP vector. (E) Number of GFP+ colonies of MEFs reprogrammed with OSKM and empty vector or the indicated RYBP vector at different time points. (legend continued on next page)

Cell Stem Cell 20, 462–477, April 6, 2017 467 (p < 0.01) (Figure S4A; Table S2). 23 miRNAs were selected we conclude that loss of DDX5 activates RYBP by inhibiting based on a fold change greater than 4 (Figure 4A). Our miRNA- miR-125b. seq data, in combination with miRNA target analysis (http:// To gain further insight into the role of miR-125b in reprogram- www.targetscan.org), identified two miRNAs: miR-125b-5p ming, we transduced mimics of either negative control or miR- (shortened as miR-125b) and miR-29a-3p (shortened as 125b together with OSKM and observed that overexpression miR-29a) as top candidates (Figure 4B) with the following prop- of miR-125b impaired iPSCs generation, as measured by a erties: significantly differentially expressed in Ddx5/ mESCs reduction in AP+ and GFP+ colonies (Figures 4H and 4I). Further, compared with wild-type cells (false discovery rate [FDR] < miR-125b gain of function had little effect on cell proliferation 0.01), downregulated at least 4-fold in Ddx5/ mESCs (log2 (Figure S4I). In addition, we observed that miR-125b overexpres- [fold change] % 2.0), and predicted to bind to the 30 UTR of sion significantly inhibited the endogenous expression of plurip- Rybp based on Target Scan (Figure 4B; Figure S4A). Interest- otent genes (Oct4, Sox2, Utf1, and Nanog) induced by OSKM ingly, among these 28 predicted miRNAs targeting Rybp,we (Figure 4J). found that miR-125b has the longest seed sequence (13 nucleo- To study miR-125b loss of function on the regulation of RYBP tides [nt]) complementary to the 30 UTR of Rybp, whereas and reprogramming, we synthesized and transfected miR-125b the typical seed sequence includes 7–8 nt, indicating that inhibitors (anti-miR-125b) together with OSKM into MEFs. Rybp miR-125b might target Rybp more stably (Figure 4C; Table S3). was upregulated by miR-125b loss of function in MEFs (Figures Further, we tested whether these two miRNAs function in S4J and S4K). In accordance with the RYBP gain of function, reprogramming by performing ectopic expression of their pri- miR-125b inhibition not only accelerated iPSC generation miRNAs. miR-125b significantly reduced the appearance of but also activated the expression of pluripotent genes (Fig- GFP+ colonies (Figures S4B and S4C), whereas miR-29a caused ures 4K and 4L; Figure S4L). miR-125b knockdown caused no significant change in the number of GFP+ colonies (Figures little difference in cell proliferation (Figure S4M). Together, we S4D and S4E). Therefore, we only focused on miR-125b for conclude that miR-125b targets Rybp at the 30 UTR and inhibits further study. reprogramming. Because DDX5 is known to modulate transcription as well as To establish the relationships of Ddx5/miR-125b and miR- miRNA processing, we measured both the levels of pri-miRNA 125b/Rybp during reprogramming, we performed several rescue and mature miRNA and found that DDX5 knockout resulted in experiments. We found that ectopic expression of miR-125b a significant reduction in mature miRNA but had no effect on abolished the reprogramming-enhancing effect by Ddx5 knock- the expression of pri-miR-125b-1 or pri-miR-125b-2, indicating down, indicating that miR-125b is a downstream target of Ddx5 that DDX5 regulates the processing but not transcription of (Figure 4M). Consistently, miR-125b loss of function overwrote miR-125b (Figure 4D; Figure S4F). Further, the positive regula- the reprogramming-blocking effects of Ddx5 overexpression tion of DDX5 on miR-125b was verified with two shRNAs target- (Figure S4N). These data suggest that miR-125b is positively ing Ddx5 both in MEFs and reprogramming cells (Figures S4G regulated by DDX5 and that it is a downstream target of DDX5 and S4H). Together, the above results indicate that miR-125b in reprogramming. Further, our data indicated that overexpres- is regulated by DDX5. sion of the coding sequence of Rybp (which lacks the 30 UTR To investigate whether miR-125b directly targets Rybp at the and, hence, is no longer a target of miR-125b and no longer 30 UTR, a pGL3-Rybp 30 UTR firefly luciferase reporter vector degraded) could mostly overcome the inhibitory effects of miR- containing the miR-125b binding site was constructed. Co- 125b overexpression on reprogramming (Figure 4N). In addition, transfection of pGL3-Rybp 30 UTR (wild-type [WT]) together we found that, in the absence of Rybp, miR-125b loss of function with miR-125b mimics and a Renilla luciferase plasmid into caused little difference in colony number (Figure S4O), indicating NIH 3T3 cells resulted in inhibition of luciferase activity (Figures that Rybp is a downstream target of miR-125b in reprogram- 4C and 4E), and this inhibition could be abolished when the ming. Taken together, we conclude that DDX5 loss of function miR-125b binding region at the 30 UTR of Rybp was mutated promotes reprogramming by upregulating Rybp through sup- (Figure 4E). To determine whether miR-125b negatively regu- pressing miR-125b processing. lates Rybp during reprogramming, we transfected a pMXs- miR-125b plasmid along with OSKM into MEFs and found DDX5 Loss of Function Stimulates the Recruitment of that miR-125b significantly downregulated RYBP expression Histone H2AK119ub1 at Lineage-Specific Genes at both the protein and mRNA levels (Figures 4F and 4G). through RYBP These data demonstrate that miR-125b directly targets the 30 RYBP-containing PRC1 complexes are responsible for depos- UTR of Rybp and that it regulates RYBP expression. Thus, iting the majority of the H2AK119ub1-repressive chromatin

(F and G) Number of GFP+ colonies on day 13 in MEFs reprogrammed with OSKM and a tet-on inducible RYBP. Doxycycline (dox, 2 mg/ml) was added (F) or withdrawn (G) daily for the indicated time periods. (H) qRT-PCR analysis for the indicated genes using RNA from the reprogrammed cells. Endo, endogenous expression levels. (I) Immunoﬂuorescence, phase contrast micrographs, and chimeric mice with germline transmission obtained with a representative iPSC colony produced with OSKM plus RYBP. Scale bar, 50 mm. (J) Number of GFP+ colonies on day 12 in MEFs reprogrammed with OSKM and DDX5, RYBP, or empty vector. (K) Number of GFP+ colonies on day 12 in MEFs reprogrammed with OSKM and control shRNA, or shRNA targeting either DDX5 or RYBP. The data in (B), (D)–(H), (J), and (K) are reported as mean values ± SD with the indicated signiﬁcance levels (*p < 0.05, **p < 0.01, ***p < 0.001) by using Student’s t test. The data in (B), (D), (E), and (H) are plotted from at least three independent experiments, each with three technical replicates. The data in (F), (G), (J), and (K) are plotted from four independent experiments, each with three technical replicates. See also Figure S3.

468 Cell Stem Cell 20, 462–477, April 6, 2017 AB D -4.0 0.0 4.0 log2(fold-change) miR-125b-5p 52 miR-125b miR-29a-3p miRNAs significantly *** differentially expressed 1E-07 *** in Ddx5-/- mESCs *** 1E-07 compared to WT (p<0.01) 2 1E-07 0 17 2 24 8E-08 miRNAs miRNAs predicted downregulated 0 6E-08 -/- to target RYBP in Ddx5 mESCs by TargetScan 4E-08 (log2<=-2.0) 2E-08 0E+00 Relative expression level to U6 to level expression Relative +/+ #2 C -/- #1 -/- -/- #3 Ddx5 3`AGUGUUCAAUCCCAGAGUCCCU 5` miR-125b Ddx5 Ddx5 Ddx5 ||||||||||||| mESCs 2034 : 5` CUUUUAUUAAGGGUCUCAGGGA 3` Rybp #3 Rybp -/- #1-/- #2-/- -/- #4 2034nt 3` UTR WT-1WT-2 CUUUUAUUAAGGGUUAUCAAGA Mutant-Rybp 3’UTR Ddx5Ddx5Ddx5Ddx5 Rybp mRNA E WT Mutant F G OSKM ** OSKM 1.2 *** 1 Ctrl miR-125b 0.8 α-RYBP RLU 0.6 1 0.68 0.4 α-β-Actin 0.2 Relative expression level 0 NC miR-125b Ctrl miR-125b H I KL OSKM OSKM OSKM OSKM *** anti-NC anti-miR-125b 300 135 NC miR-125b *** 250

200 90 150

100 3.60% 16.0% 45 Oct4-GFP Oct4-GFP 6.46% 2.13% 50

Oct4-GFP Oct4-GFP Oct4-GFP colonynumber

Oct4 -GFP number colony 0

FSC b C 5 0 2 1 FSC NC miR-125b GFP anti-N

GFP anti-miR- J OSKM MN800 *** 400 *** Ctrl miR-125b 1.2 ** ** * 600 ** 300

evel 1 l n o i 0.8 400 200 s s

pre 0.6 x 200 100 ve e

0.4 colonynumberOct4-GFP colonynumberOct4-GFP i at

el 0.2 R 0 0 0 shCtrl + - + - NC + - + - endo-Oct4 endo-Sox2 Utf1 Nanog shDdx5 - + - + miR-125b - + - + NC + + - - Ctrl + + - - miR-125b - - + + RYBP - - + +

(legend on next page) Cell Stem Cell 20, 462–477, April 6, 2017 469 mark (Gao et al., 2012), and loss of DDX5 results in the upregu- by H2AK119ub1 have a strong association with cerebellar cortex lation of RYBP through downregulating miR-125b. Therefore, development (Figure S5E). Together, DDX5 loss of function we investigated whether DDX5 regulates H2AK119ub1 through dynamically changed the distribution of H2AK119ub1. RYBP. Western blot indeed showed that loss of DDX5 led to Because DDX5 loss of function promotes reprogramming and an increase in histone H2AK119ub1 levels compared with con- increases H2AK119ub1 levels through RYBP, we hypothesized trol cells both in mESCs (Figure 5A) and reprogramming cells that loss of DDX5 may silence lineage-specific genes by recruit- (Figure S5A). This increase in H2AK119ub1 levels could be ing H2AK119ub1 to their transcription start sites (TSSs). Consis- rescued by the co-transfection of an siRNA targeting Rybp (Fig- tent with this idea, examination of individual gene tracks and ure 5A), suggesting that DDX5 loss of function increases the ChIP-qPCR confirmed that DDX5 knockout in mESCs caused H2AK119ub1 level through RYBP. significant increases in H2AK119ub1 near the TSSs of lineage- To examine the effects of DDX5 loss of function, miR-125b in- specific genes, including Egr4 and Jun (ectoderm markers), hibition, as well as RYBP overexpression on H2AK119ub1 in Bmp4 and Tril (mesoderm markers), and Klf15 and Rasd1 (endo- cells, we performed a western blot (Figure 5B) and found that derm markers) (Figures 5C and 5D; Figure S6A). In accordance both RYBP overexpression and miR-125b loss of function with H2AK119ub1, ChIP experiments indicated that the recruit- increased H2AK119ub1 levels in contrast with the control and ment of RYBP to the TSSs of lineage-specific genes (Egr4, showed comparable levels of H2AK119ub1 in three Ddx5/ Jun, Bmp4, Tril, Klf15, and Rasd1) were also markedly increased mESC lines (lanes 2–4, Figure 5B). in Ddx5/ mESCs in contrast to Ddx5+/+ mESCs (Figure 5E; To investigate whether DDX5 alters the genome-wide distribu- Figure S6B). tion of H2AK119ub1, we performed chromatin immunoprecipita- Importantly, we found that, during the reprogramming pro- tion followed by deep sequencing (ChIP-seq) for H2AK119ub1 in cess, knockdown of DDX5 increased the level of H2AK119ub1 both Ddx5+/+ and Ddx5/ mESCs. Overall, 5,908 H2AK119ub1 compared with control cells (Figure S5A). Interestingly, co-trans- regions were found in both Ddx5+/+ and Ddx5/ mESCs, 1,591 fection of OSKM with shDdx5 led to increased levels of binding sites were specific to Ddx5+/+ mESCs, and 1,335 bind- H2AK119ub1 at the TSSs of Egr4, Bmp4, and Klf15 and conse- ing sites were only found in Ddx5/ mESCs, suggesting a spe- quent suppression of their expression (Figure 5F; Figure S6C). cific process rather than a global alteration of H2AK119ub1 levels Our results further indicated that the recruitment of RYBP on (Figure S5B). We next asked whether these cohorts of genes TSS regions of the selected genes increased (Figure 5G; Fig- represent different biological functions. Gene ontology (GO) ure S6D). In addition, to confirm the RYBP function as the analysis of the specific categories revealed that genes with same route as DDX5 knockdown during reprogramming, we H2AK119ub1 both in Ddx5+/+ and Ddx5/ mESCs were performed a corresponding ChIP-qPCR by using an anti- involved in pattern specification processes, regionalization, and H2AK119ub1 antibody and found that RYBP gain of function morphogenesis (Figure S5C). Genes specifically marked by increased recruitment of H2AK119ub1 on TSS regions of H2AK119ub1 in Ddx5+/+ mESCs were related to a mesoderm selected developmental genes, which is consistent with DDX5 cell fate (Figure S5D), whereas, in Ddx5/ mESCs, genes marked deficiency (Figure S6E). This suggests that at least part of the

Figure 4. DDX5 Downregulates RYBP by Modulating miR-125b (A) Heatmap showing miRNAs significantly regulated (FDR < 0.01), filtered with 4-fold changes in wild-type and four different Ddx5/ mESC lines. Log2 relative gene expression is visualized as shades of red (higher than Ddx5+/+ mESCs) and shades of blue (lower than Ddx5+/+ mESCs) in Ddx5/ mESCs. (B) A Venn diagram depicting two miRNAs that were significantly downregulated by loss of DDX5 and predicted to target RYBP. (C) A schematic showing the sequence of miRNA-125b and the target site in the 30 UTR mRNA of Rybp. Also shown is the mutated Rybp 30 UTR that ablates the miRNA-125b binding site. (D) qRT-PCR analysis for miR-125b using RNA from wild-type and three Ddx5/ mESC lines. U6 was used for normalization. (E) A wild-type (WT) or mutant Rybp 30 UTR luciferase reporter construct was transfected into NIH 3T3 cells along with the indicated miRNA mimics. Luciferase activities were determined 48 hr post-transfection and normalized to Renilla luciferase activity. NC, negative control of miRNA mimics. (F) Western blot for RYBP on day 10 in MEFs reprogrammed with OSKM and empty plasmid or the indicated miR-125b plasmid. b-Actin was used as the loading control. (G) qRT-PCR analysis for Rybp mRNA expression in MEFs reprogrammed with OSKM and empty plasmid or the indicated miR-125b plasmid on day 10. (H) Top: AP-stained wells of a representative reprogramming experiment transfected with OSKM and either NC mimics (NC) or miR-125b mimics (miR-125b)on day 10. Bottom: GFP+ cells analyzed by FACS in reprogramming cells transfected with OSKM and NC mimics or miR-125b mimics on day 10. (I) Number of GFP+ colonies in MEFs reprogrammed with OSKM and NC mimics (NC) or miR-125b mimics (miR-125b) on day 10. (J) qRT-PCR analysis for the indicated genes using RNA from MEFs reprogrammed with OSKM and the empty plasmid or miR-125b plasmid on day 10. Values are presented in reference to MEFs transduced with OSKM plus empty plasmid. (K) Top: AP-stained wells of a representative reprogramming experiment transfected with OSKM and either NC inhibitors or miR-125b inhibitors on day 10. Bottom: GFP+ cells analyzed by FACS in reprogramming cells transfected with OSKM and NC inhibitors (anti-NC) or miR-125b inhibitors (anti-miR-125b)on day 10. (L) Number of GFP+ colonies in MEFs reprogrammed with OSKM and NC inhibitors (anti-NC) or miR-125b inhibitors (anti-miR-125b) on day 10. (M) Number of GFP+ iPSC colonies counted on day 10 of OSKM-mediated reprogramming of MEFs transduced with miR-125b mimics (miR-125b), control mimics (NC), Ddx5 shRNA, or control shRNA. (N) Number of GFP+ iPSC colonies counted on day 10 of OSKM-mediated reprogramming of MEFs transduced with miR-125b mimics (miR-125b), control mimics (NC), RYBP, or control vector. The data in (D), (E), (G), (I), (J), and (L)–(N) are reported as mean values ± SD with the indicated significance by using Student’s t test (*p < 0.05, **p < 0.01, ***p < 0.001). The data in (D) and (L)–(N) are plotted from four independent experiments, each with three technical replicates. The data in (E), (G), (I), and (J) are plotted from three independent experiments, each with three technical replicates. See also Figure S4 and Tables S2 and S3.

470 Cell Stem Cell 20, 462–477, April 6, 2017 A siRybp B

#1 #2 #3 #1 #2 #3 +/+ -/- #1 -/- #2 -/- #3 mESCs +/+ -/- -/- -/- -/- -/- -/- mESCs Ddx5 Ddx5 Ddx5 Ddx5 Ddx5 Ddx5 Ddx5 Ddx5 Ddx5 Ddx5 Ddx5 RYBP-Flag-HAanti-miR-125b α-DDX5 α-DDX5 α-RYBP α-RYBP 1 1.74 1.27 1.47 0.570.42 0.43 11.341.28 1.30 1.37 1.49 α-β-Actin α-β-Actin α-H2AK119ub1 α-H2AK119ub1 1 1.77 1.63 1.88 1.12 1.15 1.06 11.431.69 1.56 1.60 1.57 α-H2B α-H2B lane 1 2 3 4 5 6 7 lane 1 2 3 4 5 6

C D EF Fragment near Egr4 TSS Fragment near Egr4 TSS Fragment near Egr4 TSS Ectoderm Egr4 IgG H2AK119ub1 IgG RYBP IgG H2AK119ub1 Chr6: 85508574-85514994 1 kbp 18 * 0.15 * ** +/+ 43 30 Ddx5 H2AK119ub1 15 0.12 25

12 ut 0.09 p 20 1 9 % Input 15

-/- %In Ddx5 H2AK119ub1 43 %Input 0.06 6 10 3 0.03 5 1 0 0 +/+ -/- 0 Egr4 Ddx5 Ddx5 Ddx5+/+ Ddx5-/- shCtrl shDdx5 mESCs mESCs OSKM

Fragment near Bmp4 TSS Fragment near Bmp4 TSS Fragment near Bmp4 TSS Bmp4 Mesoderm IgG RYBP IgG H2AK119ub1 Chr14: 46363520-46410669 1 kbp IgG H2AK119ub1 0.1 +/+ Ddx5 H2AK119ub1 28 7 * 16 ** * 0.08 6 12 5 0.06 put 1 4 In Ddx5-/- H2AK119ub1 28 % Input 8 0.04 % 3 %Input 2 0.02 4 1 1 0 0 Gm15217 0 Bmp4 Ddx5+/+ Ddx5-/- Ddx5+/+ Ddx5-/- shCtrl shDdx5 mESCs mESCs OSKM Fragment near Klf15 TSS Fragment near Klf15 TSS Fragment near Klf15 TSS Endoderm Klf15 IgG RYBP IgG H2AK119ub1 10 kbp IgG H2AK119ub1 0.04 Chr6: 90442576-90495238 14 ** Ddx5+/+ H2AK119ub1 30 6 * * 12 5 0.03 10 4 8 1

0.02 %Input

-/- 3 % Input 6 Ddx5 H2AK119ub1 30 % Input 2 4 0.01 2 1 0 1 0 0 shCtrl shDdx5 Klf15 Ddx5+/+ Ddx5-/- +/+ -/- Ddx5 Ddx5 OSKM mESCs mESCs G Egr4 Bmp4 Klf15 IgG shCtrl IgG shDdx5 IgG shCtrl IgG shDdx5 IgG shCtrl IgG shDdx5 RYBP shCtrl RYBP shDdx5 RYBP shCtrl RYBP shDdx5 RYBP shCtrl RYBP shDdx5 0.12 * 0.3 * 0.16 * ** 0.08 0.12 0.2 * 0.08 % Input % Input 0.04 % Input 0.1 0.04

0 0 0 D5 D11 D5 D11 D5 D11 OSKM OSKM OSKM

(legend on next page)

Cell Stem Cell 20, 462–477, April 6, 2017 471 A C mESCs Sucrose gradient: 10%-30% Co-IP 2224 6 8 10 12 14 16 18 20 24

Input IgG RYBP α-RYBP α-RING1B α-RING1B α-OCT4 α-OCT4

B D E Co-IP Co-IP Co-IP

Input IgG RYBP IgG Input IgG OCT4 Input RYBP α-DDX5 α-RING1B α-RING1B

α-RYBP α-OCT4 α-OCT4 sucrose gradient sucrose gradient

Figure 6. RYBP Forms Protein Complexes with PRC1 and OCT4 Separately (A) Detection of interaction between endogenous RING1B, OCT4, and RYBP in mESCs by co-IP with anti-RYBP antibody, followed by immunoblotting for RING1B and OCT4. (B) Detection of interaction between endogenous DDX5, RYBP, and OCT4 in mESCs by co-IP with anti-OCT4 antibody, followed by immunoblotting for DDX5 and RYBP. (C) Whole-cell lysates from mESCs were sedimented on a 10%–30% sucrose gradient by ultracentrifugation and divided into 24 fractions. The gradients with even numbers were fractioned and analyzed by western blot. The indicated antibodies were used. (D and E) Detection of interaction between endogenous RING1B, OCT4, and RYBP in fractions 6–10 (D) and fractions 14–18 (E) as indicated in (C) by co-IP with anti-RYBP antibody. reason why DDX5 impairs reprogramming is inappropriate main- ures 2 and 6A; Gao et al., 2012; Tavares et al., 2012; van den tenance of somatic developmental genes and that, with the loss Berg et al., 2010). We further showed that DDX5 interacts of DDX5, these genes can then be suppressed in an RYBP- with OCT4 as well (Figure 6B). Because RYBP can form a dependent manner to possibly facilitate activation of the plurip- protein complex with both PRC1 and OCT4 in mESCs, we otency network. wondered whether RYBP, PRC1, and OCT4 form a single protein complex or whether RYBP forms two different protein RYBP Forms Separate Protein Complexes with PRC1 complexes containing either PRC1 or OCT4 in mESCs. To and OCT4, which Have Context-Speciﬁc Functions better decipher the relationship between RYBP-PRC1 and RYBP is not only a noncanonical subunit of PRC1 complexes RYBP-OCT4, we fractionated proteins by differential centrifu- and interacts with DDX5 but also interacts with OCT4 (Fig- gation on a 10%–30% sucrose gradient in mESCs (Figure 6C).

Figure 5. DDX5 Loss of Function Stimulates the Recruitment of H2AK119ub1 at Lineage-Speciﬁc Genes through RYBP (A) Western blot analysis of H2AK119ub1 from histones and RYBP from whole-cell lysates of Ddx5+/+ mESCs, Ddx5/ mESCs, and Ddx5/ mESCs transduced with Rybp siRNA. b-Actin and histone H2B were used as the loading controls. (B) Western blot analysis of H2AK119ub1 from histones and RYBP from whole-cell lysates of Ddx5+/+ mESCs, Ddx5/ mESCs, and RYBP-FLAG-hemagglutinin (HA) overexpression mESCs and mESCs transduced with miR-125b inhibitors (anti-miR-125b). b-Actin and histone H2B were used as the loading controls. (C) ChIP-seq data showing H2AK119ub1 occupancy at TSS regions of three representative genes (ectoderm, Egr4; mesoderm, Bmp4; endoderm, Klf15)in Ddx5+/+ and Ddx5/ mESCs. (D) The relative enrichment of H2AK119ub1 levels at TSS regions of three representative genes (Egr4, Bmp4, and Klf15) in Ddx5+/+ and Ddx5/ mESCs. The enrichment was measured by ChIP-qPCR. IgG was used as a negative control. (E) The relative enrichment of RYBP levels at TSS regions of three representative genes (Egr4, Bmp4, and Klf15) in Ddx5+/+ and Ddx5/ mESCs. IgG was used as a negative control. (F) The relative enrichment of H2AK119ub1 levels at TSS regions of three representative genes (Egr4, Bmp4, and Klf15) in reprogrammed MEFs with OSKM with either control shRNA or Ddx5 shRNA on day 10. IgG was used as a negative control. (G) The relative enrichment of RYBP levels at TSS regions of three representative genes (Egr4, Bmp4, and Klf15) in MEFs reprogrammed with OSKM with either control shRNA or Ddx5 shRNA on days 5 and 11. IgG was used as a negative control. The data in (D)–(G) are reported as mean values ± SD with the indicated signiﬁcance by using Student’s t test (*p < 0.05, **p < 0.01, ***p < 0.001). The data in (D)–(G) are plotted from at least three independent experiments, each with three technical replicates. See also Figures S5 and S6.

472 Cell Stem Cell 20, 462–477, April 6, 2017 A B C D 10 kbp Kdm2b promoter Kdm2b promoter chr5:122871321-122907116 IgG RYBP IgG OCT4 OCT4 131 0.015 0.06

7709 4999 25465 4 0.010 0.04 % Input

RYBP 34 % Input 0.005 0.02

RYBP OCT4 4 0.000 0.00 Primers 1 2 Primers 1 2 Kdm2b

EGHKdm2b promoter F OSKM IgG Rybp+/+ IgG Rybp-/- mESCs 300 +/+ -/- Rybp+/+ mESCs OCT4 Rybp OCT4 Rybp 250 0.30 Rybp-/- mESCs +/+ -/- *** 1.2 ** ** Rybp Rybp 200 1 α-KDM2B 150 0.20 ** 0.8 1 0.81 ** 100 0.6 α-RYBP % Input 50 0.10 0.4 α-β-Actin Oct4-GFP colony number

Relative expression level 0.2 0 Flag + + - - 0 shLuc + - + - 0.00 Rybp Kdm2b shRybp - ++- Primers 1 2 KDM2B - - + +

I J K OSKM ** Kdm2b promoter Kdm2b promoter 250 IgG Ddx5+/+ IgG Ddx5-/- IgG Ddx5+/+ IgG Ddx5-/- OCT4 Ddx5+/+ OCT4 Ddx5-/- RYBP Ddx5+/+ RYBP Ddx5-/- 200 0.15 0.04 * 150 * * 0.12 ** * 0.03 100 0.09 0.02 % Input 50 % Input Oct4-GFP colony number 0.06

0 0.01 0.03 Flag + + - - shLuc - - + + 0.00 0.00 shKdm2b - ++- Primers 1 2 Primers 1 2 RYBP - - + + L DDX5 Loss of Function K119 Ubiquitylation K4 Methylation K27 Methylation miR-125b

RYBP RYBP RYBP RYBP Ring1B Nspc1 RYBP OCT4 KDM2B

Development-specific genes Pluripotency-associated genes

(legend on next page) Cell Stem Cell 20, 462–477, April 6, 2017 473 Interestingly, RYBP was present in two major peaks in fractions mESCs significantly reduced Kdm2b expression at both the 6-–8 and 14–18 (Figure 6C). The RYBP in fractions 6–8 overlaps mRNA and protein levels (Figures 7F and 7G). with RING1B, which appeared in fractions 6–10, and is pre- Because both RYBP and KDM2B overexpression promote sumably the PRC1 complex. Intriguingly, we also detected an reprogramming, and RYBP regulates Kdm2b expression RYBP complex in fractions 14–18 that overlaps not with (Figures 7F and 7G), we investigated whether this phenome- RING1B but with OCT4 (Figure 6C). Co-IP experiments indicate non also occurred in reprogramming. OG2 MEFs were co- that, in fractions 6–10, RYBP interacts with RING1B but transfected with OSKM and shRNA retroviruses targeting not OCT4 (Figure 6D). However, in fractions 14–18, RYBP either Rybp or Kdm2b, along with a KDM2B or RYBP overex- associates with OCT4 but not RING1B (Figure 6E). These pression plasmid, in combination (Figures 7H and 7I). The data indicate that RYBP forms different protein complexes knockdown and overexpression efficiencies of RYBP and with either PRC1 or OCT4 in mESCs, suggesting that RYBP KDM2B in MEFs were determined by qRT-PCR (Figures S7B may acquire specific functions by interacting with different and S7C). Our data indicated that RYBP knockdown had protein co-factors. little effect on reprogramming when KDM2B was overexpressed in conjunction with OSKM (Figure 7H). However, RYBP Functions Upstream of KDM2B and Loss of DDX5 KDM2B knockdown significantly inhibited reprogramming Facilitates the Binding of RYBP and OCT4 at the when RYBP was overexpressed together with OSKM (Fig- Kdm2b Locus ure 7I), suggesting that RYBP functions upstream of KDM2B To investigate whether RYBP co-binds to specific genes with in reprogramming. OCT4, we performed ChIP-seq with an anti-RYBP antibody in The qRT-PCR and western blot results indicated that both mESCs and compared the results with the published results of mRNA and protein levels of Kdm2b were upregulated by genome-wide localizations of OCT4 (Whyte et al., 2013). Inter- DDX5 loss of function (Figures S7D and S7E). To understand estingly, we found that 39.3% of the RYBP-binding sites overlap- the mechanism involved in the activation of Kdm2b expression ped with 16% of the OCT4-binding sites (Figure 7A). From the by DDX5 loss of function, ChIP-qPCR experiments were per- co-binding gene list, we found that Kdm2b is a target gene for formed using anti-RYBP and anti-OCT4 antibodies in both both RYBP and OCT4. KDM2B is another non-canonical subunit Ddx5+/+ and Ddx5/ cells. Our results indicated that OCT4 of the PRC1 complex and has been reported to promote reprog- bound more strongly to the Kdm2b promoter in Ddx5/ cells ramming (Liang et al., 2012; Wang et al., 2011; Wu et al., 2013). than in Ddx5+/+ cells (Figure 7J). Similarly, the recruitment of Therefore, we were curious to know how RYBP and KDM2B RYBP to the Kdm2b promoter was also markedly increased in function together in regulating reprogramming. Consistently, Ddx5/ cells in contrast to Ddx5+/+ cells (Figure 7K). Taken analysis of the ChIP-seq data demonstrated that the occu- together, these results suggest that DDX5 is a negative regu- pancies of both OCT4 and RYBP were enriched at the promoter lator for the recruitment of RYBP and OCT4 to the Kdm2b pro- of the Kdm2b gene (Figure 7B), a result supported by ChIP- moter and then regulates the reactivation of endogenous plurip- qPCR (Figures 7C and 7D). otent gene expression during reprogramming. To investigate the effect of RYBP on the binding of OCT4 at the Kdm2b promoter, we took advantage of RYBP knockout DISCUSSION mESCs, which had no effect on the expression of pluripotent genes (Figure S7A). ChIP-qPCR was performed using an anti- In this study, we report that the RBP DDX5 acts as a barrier to OCT4 antibody followed by qPCR in both Rybp+/+ and Rybp/ somatic cell reprogramming. DDX5 inhibits RYBP function by mESCs. Interestingly, we found that the loss of RYBP signifi- modulating miR-125b and, thereby, affects the deposition of cantly reduced the recruitment of OCT4 to the Kdm2b promoter inhibitory ubiquitin on histones through a non-canonical RYBP- (Figure 7E). Furthermore, we found that the loss of RYBP in PRC1 complex. It is counterintuitive that DDX5 is upregulated

Figure 7. RYBP Regulates KDM2B Expression, and DDX5 Loss of Function Facilitates the Recruitment of RYBP and OCT4 at the Kdm2b Locus (A) Venn diagram showing the overlap of genes co-bound by RYBP (yellow) and OCT4 (blue) with the published datasets (NCBI GEO: GSE44288). (B) An analysis of ChIP-seq data showing that OCT4 and RYBP occupancy is enriched at the promoter of the mouse Kdm2b gene. (C and D) The relative enrichment of RYBP (C) and OCT4 (D) levels at the Kdm2b promoter in mESCs. The enrichment was measured by ChIP-qPCR. IgG was used as a negative control. (E) The relative enrichment of OCT4 levels at the Kdm2b promoter in Rybp+/+ and Rybp/ mESCs. The enrichment was measured by ChIP-qPCR. IgG was used as a negative control. (F) qRT-PCR analysis of Rybp and Kdm2b expression in both Rybp+/+ and Rybp/ mESCs. (G) Western blot for RYBP and KDM2B using cell lysates from both Rybp+/+ and Rybp/ mESCs. b-Actin was used as the loading control. (H) Number of GFP+ colonies on day 10 in MEFs reprogrammed with OSKM plus either/both Rybp shRNA and/or KDM2B vector. (I) Number of GFP+ colonies on day 10 in MEFs reprogrammed with OSKM plus either/both Kdm2b shRNA and/or RYBP vector. (J and K) The relative enrichment of OCT4 (J) and RYBP (K) at the Kdm2b promoter in Ddx5+/+ and Ddx5/ mESCs. The enrichment was measured by ChIP- qPCR. IgG was used as a negative control. (L) A proposed working model schematically representing the main message of our work. The data in (C)–(F) and (H)–(K) are reported as mean values ± SD with the indicated signiﬁcance by using Student’s t test analysis (*p < 0.05, **p < 0.01, ***p < 0.001). The data in (C)–(F), (J), and (K) are plotted from three independent experiments, each with three technical replicates. The data in (H) and (I) are plotted from four independent experiments, each with three technical replicates. See also Figure S7.

474 Cell Stem Cell 20, 462–477, April 6, 2017 during reprogramming and acts as a barrier. However, there is no ming. This study highlights how DDX5 acts to regulate the direct correlation between iPSC generation and gene expression epigenetic environment and so regulates cell fate transitions, level; for example, p53 and Setdb1 are both highly expressed in a function mainly attributed to transcription factors and other embryonic stem cells (ESCs) compared with somatic cells, yet proteins that interact directly with chromatin. they have both been reported as reprogramming roadblocks (Chen et al., 2013; Li et al., 2009). STAR+METHODS In mESCs, two major variations of the PRC1 complex are present, defined by the mutually exclusive binding of CBX7 or RYBP Detailed methods are provided in the online version of this paper (Gao et al., 2012; Tavares et al., 2012). miR-125b has been re- and include the following: ported to target CBX7 and regulate ESC differentiation (O’Logh- len et al., 2012), and, in this study, miR-125b binds RYBP at the 30 d KEY RESOURCES TABLE UTR, inhibits RYBP expression, and further blocks reprogram- d CONTACT FOR REAGENT AND RESOURCE SHARING ming (Figure 4). It is intriguing that the two PRC1 complexes d EXPERIMENTAL MODEL AND SUBJECT DETAILS have a common upstream regulator in miR-125b. In addition to B Mice miR-125b, miRNA let-7 was significantly downregulated after B Cell Lines and Cell Culture Condition DDX5 loss of function in mESCs (Figure 4) and has been reported d METHOD DETAILS to act as a barrier to reprogramming (Worringer et al., 2014). B iPSCs Generation However, our analysis indicated that let-7 does not target B Immunofluorescence Rybp. Therefore, we propose that inhibition of let-7 miRNAs by B Karyotyping and Generation of Chimeric Mice loss of DDX5 promotes reprogramming in an Rybp-independent B Generation of DDX5 Knockout mESCs manner. B shRNAs, siRNAs, miRNA mimics and inhibitors It has been shown that genes occupied by RYBP are primarily B Quantitative RT-PCR Analysis involved in the regulation of metabolism and cell cycle progres- B Luciferase Assay sion (Morey et al., 2013). Here we found that, in reprogramming B Western Blot and Endogenous Co-IP Assay cells, the loss of DDX5 and consequent enhanced RYBP lead B Histone Extraction to enhanced suppression of lineage-specific genes through B Sucrose Gradient Ultracentrifugation H2AK119ub1, facilitating reprogramming from MEFs to iPSCs, B ChIP-Seq and Bioinformatic Analysis potentially mediated by the formation of context-specific protein B RNA-Seq, miRNA-Seq and Bioinformatic Analysis complexes comprising either RYBP-PRC1 or RYBP-OCT4. We d QUANTIFICATION AND STATISTICAL ANALYSIS suggest that DDX5 controls reprogramming at both early and d DATA AND SOFTWARE AVAILABILITY late phases of reprogramming through two different RYBP- dependent mechanisms (Figure 7L). SUPPLEMENTAL INFORMATION Interaction of RYBP with OCT4 suggests that RYBP might be involved in the regulation of the OCT4 pluripotency Supplemental Information includes seven figures and six tables and can be found with this article online at http://dx.doi.org/10.1016/j.stem.2016.12.002. network. Indeed, we show that the loss of RYBP has no effect on Oct4 expression but results in the downregulation of AUTHOR CONTRIBUTIONS Kdm2b expression, which is consistent with upregulation of Kdm2b expression by loss of DDX5, and also reduces the H.Y., H.L., P.L., and J.J. initiated the study and designed the experiments. H.L. binding of OCT4 to the Kdm2b promoter. Unexpectedly, these and P.L. conducted most of the experiments. J.J. analyzed part of the data. effects suggest that RYBP may also act as an activator (at A.P.H., N.H., and Z.Y. performed the bioinformatics analysis. Y.S., Q.X., least indirectly by promoting the binding of OCT4) and not K.H., W.P., J.C., J.L., M.Y., X.D., J.Z., C.H., M.A.E., S.G., and D.P. contributed to the work. H.Y., H.L., P.L., and J.J. wrote the manuscript. H.Y. conceived and just a repressor in reprogramming. Interestingly, KDM2B, a supervised the entire study. histone lysine 4/lysine 36 on histone H3 protein (H3K4/K36) demethylase (He et al., 2008), is linked to PRC1 because it ACKNOWLEDGMENTS can also form a non-canonical KDM2B-PRC1 complex that is involved in the deposition of histone H2AK119ub1 (Wu We thank Profs. Xiangdong Fu and Guohong Li for their critical readings et al., 2013). Moreover, KDM2B is a downstream target of and helpful discussions and Dr. Jiekai Chen for providing the anti-KDM2B OCT4 and has been shown to promote reprogramming (Liang antibody. This work was supported in part by the Ministry of Science et al., 2012; Wang et al., 2011). Our experiments indicate that and Technology of the People’s Republic of China (2016YFA0100400, 2015CB964800, and 2016YFA0100302), the National Natural Science RYBP is an upstream regulator of both OCT4 and KDM2B in Foundation of China (31471210, 31271391, 31601050, 31550110206, and regulating reprogramming and the enrichments of OCT4 at 31471242), a Guangdong Frontier and Key Technology Innovation the Kdm2b promoter. Special Grant (2016B030229006), Guangdong Natural Science Funds Overall, we identified a cascade of regulation, beginning with (2015A030308003 and 2015A030310041), the Guangzhou Municipal Science the RBP DDX5, which, through miR125b, is an upstream regu- and Technology Bureau (201510010061), and the One Hundred Talents Proj- lator of RYBP, that, in context-specific complexes with PRC1 ect of the Chinese Academy of Sciences (to H.Y.). or OCT4, can act to pose a barrier to reprogramming. We pro- Received: January 27, 2016 pose that DDX5-mediated regulation of RYBP acts to suppress Revised: August 8, 2016 development-specific genes and activate pluripotency-promot- Accepted: December 2, 2016 ing genes (such as Kdm2b) in successive steps of reprogram- Published: January 19, 2017; corrected online: April 6, 2017

Cell Stem Cell 20, 462–477, April 6, 2017 475 REFERENCES Li, H., Collado, M., Villasante, A., Strati, K., Ortega, S., Can˜ amero, M., Blasco, M.A., and Serrano, M. (2009). The Ink4/Arf locus is a barrier for iPS cell reprog- Apostolou, E., and Hochedlinger, K. (2013). Chromatin dynamics during ramming. Nature 460, 1136–1139. cellular reprogramming. Nature 502, 462–471. Li, R., Liang, J., Ni, S., Zhou, T., Qing, X., Li, H., He, W., Chen, J., Li, F., Zhuang, Brambrink, T., Foreman, R., Welstead, G.G., Lengner, C.J., Wernig, M., Suh, Q., et al. (2010). A mesenchymal-to-epithelial transition initiates and is required H., and Jaenisch, R. (2008). Sequential expression of pluripotency markers for the nuclear reprogramming of mouse fibroblasts. Cell Stem Cell 7, 51–63. during direct reprogramming of mouse somatic cells. Cell Stem Cell 2, Liang, G., He, J., and Zhang, Y. (2012). Kdm2b promotes induced pluripotent 151–159. stem cell generation by facilitating gene activation early in reprogramming. Chen, J., Liu, H., Liu, J., Qi, J., Wei, B., Yang, J., Liang, H., Chen, Y., Chen, J., Nat. Cell Biol. 14, 457–466. Wu, Y., et al. (2013). H3K9 methylation is a barrier during somatic cell reprog- Linder, P., and Jankowsky, E. (2011). From unwinding to clamping - the DEAD ramming into iPSCs. Nat. Genet. 45, 34–42. box RNA helicase family. Nat. Rev. Mol. Cell Biol. 12, 505–516. Dardenne, E., Polay Espinoza, M., Fattet, L., Germann, S., Lambert, M.P., Neil, McLean, C.Y., Bristor, D., Hiller, M., Clarke, S.L., Schaar, B.T., Lowe, C.B., H., Zonta, E., Mortada, H., Gratadou, L., Deygas, M., et al. (2014). RNA heli- Wenger, A.M., and Bejerano, G. (2010). GREAT improves functional interpre- cases DDX5 and DDX17 dynamically orchestrate transcription, miRNA, and tation of cis-regulatory regions. Nat. Biotechnol. 28, 495–501. splicing programs in cell differentiation. Cell Rep. 7, 1900–1913. Morey, L., Aloia, L., Cozzuto, L., Benitah, S.A., and Di Croce, L. (2013). RYBP Davis, B.N., Hilyard, A.C., Lagna, G., and Hata, A. (2008). SMAD proteins con- and Cbx7 define specific biological functions of polycomb complexes in trol DROSHA-mediated microRNA maturation. Nature 454, 56–61. mouse embryonic stem cells. Cell Rep. 3, 60–69. Esteban, M.A., Wang, T., Qin, B., Yang, J., Qin, D., Cai, J., Li, W., Weng, Z., O’Loghlen, A., Mun˜ oz-Cabello, A.M., Gaspar-Maia, A., Wu, H.A., Banito, A., Chen, J., Ni, S., et al. (2010). Vitamin C enhances the generation of mouse Kunowska, N., Racek, T., Pemberton, H.N., Beolchi, P., Lavial, F., et al. and human induced pluripotent stem cells. Cell Stem Cell 6, 71–79. (2012). MicroRNA regulation of Cbx7 mediates a switch of Polycomb orthologs € Friedlander, M.R., Mackowiak, S.D., Li, N., Chen, W., and Rajewsky, N. (2012). during ESC differentiation. Cell Stem Cell 10, 33–46. miRDeep2 accurately identifies known and hundreds of novel microRNA Polo, J.M., Anderssen, E., Walsh, R.M., Schwarz, B.A., Nefzger, C.M., Lim, genes in seven animal clades. Nucleic Acids Res. 40, 37–52. S.M., Borkent, M., Apostolou, E., Alaei, S., Cloutier, J., et al. (2012). A mo- Gao, Z., Zhang, J., Bonasio, R., Strino, F., Sawai, A., Parisi, F., Kluger, Y., and lecular roadmap of reprogramming somatic cells into iPS cells. Cell 151, Reinberg, D. (2012). PCGF homologs, CBX proteins, and RYBP define func- 1617–1632. tionally distinct PRC1 family complexes. Mol. Cell 45, 344–356. Rais, Y., Zviran, A., Geula, S., Gafni, O., Chomsky, E., Viukov, S., Mansour, Gao, Y., Chen, J., Li, K., Wu, T., Huang, B., Liu, W., Kou, X., Zhang, Y., Huang, A.A., Caspi, I., Krupalnik, V., Zerbib, M., et al. (2013). Deterministic direct H., Jiang, Y., et al. (2013). Replacement of Oct4 by Tet1 during iPSC induction reprogramming of somatic cells to pluripotency. Nature 502, 65–70. reveals an important role of DNA methylation and hydroxymethylation in reprogramming. Cell Stem Cell 12, 453–469. Risso, D., Schwartz, K., Sherlock, G., and Dudoit, S. (2011). GC-content normalization for RNA-Seq data. BMC Bioinformatics 12, 480. Golipour, A., David, L., Liu, Y., Jayakumaran, G., Hirsch, C.L., Trcka, D., and Wrana, J.L. (2012). A late transition in somatic cell reprogramming requires Robinson, M.D., McCarthy, D.J., and Smyth, G.K. (2010). edgeR: a regulators distinct from the pluripotency network. Cell Stem Cell 11, 769–782. Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140. Gregory, R.I., Yan, K.P., Amuthan, G., Chendrimada, T., Doratotaj, B., Cooch, N., and Shiekhattar, R. (2004). The Microprocessor complex mediates the gen- Samavarchi-Tehrani, P., Golipour, A., David, L., Sung, H.K., Beyer, T.A., Datti, esis of microRNAs. Nature 432, 235–240. A., Woltjen, K., Nagy, A., and Wrana, J.L. (2010). Functional genomics reveals a BMP-driven mesenchymal-to-epithelial transition in the initiation of somatic Guallar, D., and Wang, J. (2014). RNA-binding proteins in pluripotency, differ- cell reprogramming. Cell Stem Cell 7, 64–77. entiation, and reprogramming. Front. Biol. (Beijing) 9, 389–409. He, J., Kallin, E.M., Tsukada, Y., and Zhang, Y. (2008). The H3K36 demethy- Shao, Z., Zhang, Y., Yuan, G.C., Orkin, S.H., and Waxman, D.J. (2012). lase Jhdm1b/Kdm2b regulates cell proliferation and senescence through MAnorm: a robust model for quantitative comparison of ChIP-Seq data sets. Genome Biol. 13, R16. p15(Ink4b). Nat. Struct. Mol. Biol. 15, 1169–1175. Huang, K., Jia, J., Wu, C., Yao, M., Li, M., Jin, J., Jiang, C., Cai, Y., Pei, D., Pan, Stadtfeld, M., Maherali, N., Breault, D.T., and Hochedlinger, K. (2008). Defining G., and Yao, H. (2013). Ribosomal RNA gene transcription mediated by the molecular cornerstones during fibroblast to iPS cell reprogramming in mouse. master genome regulator protein CCCTC-binding factor (CTCF) is negatively Cell Stem Cell 2, 230–240. regulated by the condensin complex. J. Biol. Chem. 288, 26067–26077. Sun, Z., Evans, J., Bhagwate, A., Middha, S., Bockol, M., Yan, H., and Kocher, Huang, W., Thomas, B., Flynn, R.A., Gavzy, S.J., Wu, L., Kim, S.V., Hall, J.A., J.P. (2014). CAP-miRSeq: a comprehensive analysis pipeline for microRNA Miraldi, E.R., Ng, C.P., Rigo, F., et al. (2015). DDX5 and its associated lncRNA sequencing data. BMC Genomics 15, 423. Rmrp modulate TH17 cell effector functions. Nature 528, 517–522. Suzuki, H.I., Yamagata, K., Sugimoto, K., Iwamoto, T., Kato, S., and Miyazono, Hutchins, A.P., Jauch, R., Dyla, M., and Miranda-Saavedra, D. (2014). glbase: K. (2009). Modulation of microRNA processing by p53. Nature 460, 529–533. a framework for combining, analyzing and displaying heterogeneous genomic Tavares, L., Dimitrova, E., Oxley, D., Webster, J., Poot, R., Demmers, J., and high-throughput sequencing data. Cell Regen. (Lond.) 3,1. Bezstarosti, K., Taylor, S., Ura, H., Koide, H., et al. (2012). RYBP-PRC1 com- Hutchins, A.P., Takahashi, Y., and Miranda-Saavedra, D. (2015). Genomic plexes mediate H2A ubiquitylation at polycomb target sites independently of analysis of LPS-stimulated myeloid cells identifies a common pro-inflamma- PRC2 and H3K27me3. Cell 148, 664–678. tory response but divergent IL-10 anti-inflammatory responses. Sci. Rep. van den Berg, D.L., Snoek, T., Mullin, N.P., Yates, A., Bezstarosti, K., 5, 9100. Demmers, J., Chambers, I., and Poot, R.A. (2010). An Oct4-centered protein Kumar, V., Muratani, M., Rayan, N.A., Kraus, P., Lufkin, T., Ng, H.H., and interaction network in embryonic stem cells. Cell Stem Cell 6, 369–381. Prabhakar, S. (2013). Uniform, optimal signal processing of mapped deep- Wang, T., Chen, K., Zeng, X., Yang, J., Wu, Y., Shi, X., Qin, B., Zeng, L., sequencing data. Nat. Biotechnol. 31, 615–622. Esteban, M.A., Pan, G., and Pei, D. (2011). The histone demethylases Langmead, B., Trapnell, C., Pop, M., and Salzberg, S.L. (2009). Ultrafast and Jhdm1a/1b enhance somatic cell reprogramming in a vitamin-C-dependent memory-efficient alignment of short DNA sequences to the human genome. manner. Cell Stem Cell 9, 575–587. Genome Biol. 10, R25. Whyte, W.A., Orlando, D.A., Hnisz, D., Abraham, B.J., Lin, C.Y., Kagey, M.H., Li, B., and Dewey, C.N. (2011). RSEM: accurate transcript quantification from Rahl, P.B., Lee, T.I., and Young, R.A. (2013). Master transcription factors and RNA-Seq data with or without a reference genome. BMC Bioinformatics mediator establish super-enhancers at key cell identity genes. Cell 153, 12, 323. 307–319.

476 Cell Stem Cell 20, 462–477, April 6, 2017 Worringer, K.A., Rand, T.A., Hayashi, Y., Sami, S., Takahashi, K., Tanabe, K., Yu, J., Vodyanik, M.A., Smuga-Otto, K., Antosiewicz-Bourget, J., Frane, J.L., Narita, M., Srivastava, D., and Yamanaka, S. (2014). The let-7/LIN-41 pathway Tian, S., Nie, J., Jonsdottir, G.A., Ruotti, V., Stewart, R., et al. (2007). Induced regulates reprogramming to human induced pluripotent stem cells by control- pluripotent stem cell lines derived from human somatic cells. Scioence 318, ling expression of prodifferentiation genes. Cell Stem Cell 14, 40–52. 1917–1920. Wu, X., Johansen, J.V., and Helin, K. (2013). Fbxl10/Kdm2b recruits polycomb Zhang, Y., Liu, T., Meyer, C.A., Eeckhoute, J., Johnson, D.S., Bernstein, B.E., repressive complex 1 to CpG islands and regulates H2A ubiquitylation. Mol. Nusbaum, C., Myers, R.M., Brown, M., Li, W., and Liu, X.S. (2008). Model- Cell 49, 1134–1146. based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137. Yao, H., Brick, K., Evrard, Y., Xiao, T., Camerini-Otero, R.D., and Felsenfeld, G. Zhang, J., Ratanasirintrawoot, S., Chandrasekaran, S., Wu, Z., Ficarro, S.B., (2010). Mediation of CTCF transcriptional insulation by DEAD-box RNA-bind- Yu, C., Ross, C.A., Cacchiarelli, D., Xia, Q., Seligson, M., et al. (2016). LIN28 ing protein p68 and steroid receptor RNA activator SRA. Genes Dev. 24, Regulates Stem Cell Metabolism and Conversion to Primed Pluripotency. 2543–2555. Cell Stem Cell 19, 66–80.

Cell Stem Cell 20, 462–477, April 6, 2017 477 STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies anti-DDX5 antibody Millipore Cat#05-850; RRID: AB_327052 Monoclonal ANTI-FLAG M2 antibody Sigma-Aldrich Cat#F1804; RRID: AB_262044 produced in mouse anti-Histone H2B antibody Cell Signaling Technology Cat#12364; RRID: N/A anti-rabbit monoclonal Ubiquityl-Histone Cell Signaling Technology Cat#8240S; RRID: AB_10891618 H2A (Lys119) (D27C4) antibody anti-Histone H3 antibody Abcam Cat#ab1791; RRID: AB_302613 anti-NANOG antibody Abcam Cat#ab106465; RRID: AB_10858563 anti-OCT4 antibody Abcam Cat#ab19857; RRID: AB_445175 anti-OCT3/4 antibody Santa Cruz Biotechnology Cat#sc-5279; RRID: AB_628051 Normal Rabbit IgG Santa Cruz Biotechnology Cat#sc-2027; RRID: AB_737197 anti-RING1B antibody Cell Signaling Technology Cat# 5694S; RRID: AB_10705604 anti-RYBP antibody Millipore Cat#ab3637; RRID: AB_2285466 anti-RYBP antibody Abcam Cat#ab5976; RRID: AB_305212 anti-CBX7 antibody Abcam Cat#ab21873; RRID: AB_726005 anti-b-ACTIN antibody Sigma-Aldrich Cat#A5441; RRID: AB_476744 anti-KDM2B antibody Laboratory of Jiekai Chen N/A Chemicals, Peptides, and Recombinant Proteins Anti-FLAG M2 Magnetic Beads afﬁnity Sigma-Aldrich Cat#M8823 isolated antibody RNase A QIAGEN Cat#19101 mLIF Medium Supplement Millipore Cat# ESG1107 CHIR99021 HCl Selleck Cat# S2924; CAS: 252917-06-9 PD0325901 Selleck Cat# S1036; CAS: 391210-10-9 Vitamin C Sigma-Aldrich Cat# 49752; CAS: 66170-10-3 NBT/BCIP Stock Solution Roche Cat# 11681451001 EDTA-free Protease Inhibitor Cocktail Roche Cat# 04693132001 DAPI Sigma-Aldrich Cat# D9542; CAS: 28718-90-3 G418 Sigma-Aldrich Cat# A1720; CAS: 108321-42-2 Doxycycline hyclate Sigma-Aldrich Cat# D9891; CAS: 24390-14-5 Demecolcine Aladdin Cat# C131506; CAS: 477-30-5 Polyethylenimine ‘‘Max,’’ (Mw 40,000) - High Polysciences Cat# 24765-2; CAS#:26338-45-4, Potency Linear PEI 9002-98-6, 26913-06-4 Goat Serum Sigma-Aldrich Cat# G6767 Dynabeads Protein G for Immunoprecipitation Life Technologies Cat# 10004D Dynabeads Protein A for Immunoprecipitation Life Technologies Cat# 10001D Critical Commercial Assays Cell Cycle Assay Kit Vazyme Biotech Cat# A411-01 GoScript Reverse Transcription System Promega Cat# A5000 Dual-Luciferase Reporter Assay Systems Promega Cat# E1910 Eastep Super Total RNA Extraction Kit Promega Cat# LS1040 Deposited Data RNA-seq data This paper GEO: GSE76825 ChIP-seq data This paper GEO: GSE76825 (Continued on next page)

e1 Cell Stem Cell 20, 462–477.e1–e6, April 6, 2017 Continued REAGENT or RESOURCE SOURCE IDENTIFIER microRNA-seq data This paper GEO: GSE76825 OCT4 ChIP-Seq (Whyte et al., 2013) GEO: GSE44288 Experimental Models: Cell Lines HEK293T N/A RRID: CVCL_0063 NIH 3T3 N/A RRID: CVCL_0594 Plat-E N/A RRID: CVCL_B488 OG2 MEFs This paper (Esteban et al., 2010) OG2 secondary MEFs Laboratory of Shaorong Gao (Gao et al., 2013) R1 mES cells ATCC RRID: CVCL_2167 RYBP/ R1 mES cells Laboratory of Guohong Li N/A MPI-II mES cells Laboratory of Duanqing Pei RRID: CVCL_2H58 DDX5/ MPI-II mES cells This paper N/A Experimental Models: Organisms/Strains CBA/CaJ mice The Jackson Laboratory RRID: IMSR_JAX:000654 C57BL/6J mice The Jackson Laboratory RRID: IMSR_JAX:000664 129S4/SvJaeJ mice The Jackson Laboratory RRID: IMSR_JAX:009104 ICR/HaJ mice The Jackson Laboratory RRID: IMSR_JAX:009122 Recombinant DNA pMXs-SOX2/KLF4/OCT4/c-MYC-3Flag Laboratory of Miguel A. Esteban (Esteban et al., 2010) pMXs-DDX5-3Flag This paper N/A pMXs-RYBP-3Flag This paper N/A pMXs-KDM2B-3Flag Laboratory of Duanqing Pei (Wang et al., 2011) pMXs-miR125b-3Flag This paper N/A pMXs-miR29a-3Flag This paper N/A pSUPER-shDdx5#1/#2 This paper N/A pSUPER-shRybp#1/#2 This paper N/A pSUPER-shKdm2b Laboratory of Duanqing Pei (Wang et al., 2011) pW-TRE-RYBP-Flag (Tet-on) This paper N/A pW-TRE-DDX5-Flag (Tet-on) This paper N/A pcDNA3.3-hCas9 Addgene Addgene: 41815 U6-sgRNA-DDX5#1/#2 This paper N/A pGL3-Control vector Promega CAT#E1741 pGL3-RYBP-30UTR This paper N/A pGL3-RYBP-30UTR-MUTANT This paper N/A pRL-null Vector Promega CAT#E2271 Sequence-Based Reagents siRNA targeting sequence: RYBP UACAGU This paper N/A CUGCUAACGCUACdTdT miRNA mimics for miR-125b ucccugaga This paper N/A cccuaacuuguga miRNA inhibitors for miR-125b CAGUAC This paper N/A UUUUGUGUAGUACAA Oligo Set 1 for targeting genomic Ddx5 Forward This paper N/A 50-CACCGAGACCGCGGCCGGGATCGA-30 Oligo Set 1 for targeting genomic Ddx5 Reverse This paper N/A 50-AAACTCGATCCCGGCCGCGGTCTC-30 Oligo Set 2 for targeting genomic Ddx5 Forward This paper N/A 50-CACCGGAAAGAAGTTTGGAAATCC-30 Oligo Set 2 for targeting genomic Ddx5 Reverse This paper N/A 50-AAACGGATTTCCAAACTTCTTTCC-30 (Continued on next page)

Cell Stem Cell 20, 462–477.e1–e6, April 6, 2017 e2 Continued REAGENT or RESOURCE SOURCE IDENTIFIER Primer for Rybp 3‘UTR fragment amplification This paper N/A Forward primer, 50-AGTTCTAGAGGAT GCAAGCTGTACTTACC-30 Primer for Rybp 3‘UTR fragment amplification This paper N/A Reverse primer, 50-AGTTCTAGACTAA TCTAGGGGGTCCAATC-30 Sequences of shRNA oligos, see Table S4 This paper N/A Primers used for qRT-PCR analysis, see Table S5 This paper N/A Primers used for ChIP-qPCR analysis, see Table S6 This paper N/A Software and Algorithms ImageJ NIH, USA https://imagej.nih.gov/ij FlowJo 7.6 Single Cell Analysis Software http://www.flowjo.com/ Bowtie2 (v2.2.8) (Langmead et al., 2009) http://bowtie-bio.sourceforge.net/ bowtie2/index.shtml MACS (v1.4.2) (Zhang et al., 2008) http://liulab.dfci.harvard.edu/MACS/ DFilter (v1.0) (Kumar et al., 2013) https://omictools.com/dfilter-tool MAnorm (Shao et al., 2012) http://bcb.dfci.harvard.edu/gcyuan/ MAnorm/MAnorm.htm GREAT (McLean et al., 2010) http://bejerano.stanford.edu/great/ public/html/ RSEM (v1.2.19) (Li and Dewey, 2011) http://deweylab.github.io/RSEM/ CAP-miRSeq pipeline (Sun et al., 2014) http://bioinformaticstools.mayo.edu/ research/cap-mirseq/ miRDeep2 mapper (Friedlander€ et al., 2012) https://www.mdc-berlin.de/8551903/en/ glbase (Hutchins et al., 2014) https://bitbucket.org/oaxiom/glbase/ wiki/Home EDASeq (v2.4.1) (Risso et al., 2011) http://www.bioconductor.org/packages/ release/bioc/html/edgeR.html

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for reagents may be directed to, and will be fulﬁlled by, the Lead Contact, Dr. Hongjie Yao in Guangzhou Institutes of Biomedicine and Health, Chinese Academy of Sciences ([email protected]).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mice C57BL/6J, CBA/CaJ, 129S4/SvJaeJ and ICR/HaJ mice were purchased from the Jackson Laboratory. C57BL/6J, CBA/CaJ, 129S4/ SvJaeJ were used for generating OG2-MEFs and ICR/HaJ mice were used for chimera generation. Animals were individually housed under a 12 hr light/dark cycle and provided with food and water ad libitum. Our studies followed the guidelines for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocols were approved by the Committee on the Ethics of Animal Experiments of Guangzhou Institutes of Biomedicine and Health. All efforts were made to minimize animal discomfort.

Cell Lines and Cell Culture Condition HEK293T, NIH 3T3 and Plat-E cells were maintained in DMEM high-glucose media (Hyclone) supplemented with 10% FBS (Excell). OG2 MEFs (Esteban et al., 2010) were obtained from E13.5 mouse embryos from crossing male Oct4-GFP transgenic allele-carrying mice (CBA/CaJ 3 C57BL/6J) to 129S4/SvJaeJ female mice. Embryos were sexed by inspecting gonads for the pattern of Oct4-GFP expression. Gonads and internal organs were removed before processing the embryos for MEF isolation. To generate iPS cells that can be identiﬁed in mouse chimeras after blastocyst injection, we derived MEFs from E13.5 embryos that are hemizygous for Oct4-GFP and heterozygous for the Rosa26-lacZ reporter allele. OG2-MEFs and secondary MEFs were maintained in DMEM high-glucose media containing 10% FBS (GIBCO), 1 mM sodium pyruvate (GIBCO), 1 mM non-essential amino acids (GIBCO), 1 3 GlutaMAX (GIBCO). mESCs and iPSCs were cultured on feeders-coated plates in DMEM high-glucose media containing 15% FBS (GIBCO), 1mM sodium pyruvate (GIBCO), 1mM non-essential amino acids (GIBCO), 1 3 GlutaMAX (GIBCO), 0.1 mM 2-mercaptoethanol e3 Cell Stem Cell 20, 462–477.e1–e6, April 6, 2017 (GIBCO), 1000 U/ml leukemia inhibitory factor (LIF) (Millipore), and the 2i inhibitors, 3 mM CHIR99021 (Selleck), and 1 mM PD0325901 (Selleck).

METHOD DETAILS iPSCs Generation For OG2 MEFs reprogramming, 15,000 OG2 MEFs at passage 2 were plated in a 12-well plate and then infected twice with retroviral supernatants generated with Plat-E cells. Plat-E cells were plated at 70%–80% confluency and transfected using modified polyethylenimine (PEI) transfection method as follows. The medium of a 10-cm dish was replaced with 9 mL fresh Plat-E medium. A mixture of 10 mg plasmid and 960 mL Opti-MEM (GIBCO) was mixed vigorously after adding 40 mL PEI (Polysciences, 1 mg/ml) and was incubated at room temperature for 10 min, and then was added into the medium in the 10-cm dish. The supernatant of the transfected Plat-E cells were harvested using a syringe and filtered through a 0.45 mm filter. Equal volumes of the 4 supernatants of each OSKM transcription factor were mixed with 1 volume of fresh Plat-E medium containing polybrene at a final concentration of 8 mg/ml. The multiplicity of infection of virus was controlled between 106-107. The infection efficiency was tested by MEFs transduced with virus of pMXs-dsRed. For infecting OG2 MEFs, the medium of OG2 MEFs was replaced with 2 mL of infection mixture per well in a 12-well plate. Infected cells were cultured with mESC medium containing Vitamin C (Sigma; 50 mg/ml) post infection and renewed daily. iPSCs colonies appeared about 6 days post infection. GFP positive colonies were counted at day 10 to day 12 after infection. Fluo- rescence-activated cell sorting (FACS) was performed to test the GFP+ efficiency. For secondary MEF reprogramming (Gao et al., 2013), 12,500 MEFs were seeded and transduced twice using viral supernatants in 12-well plates. The medium was changed to mESC medium containing Vitamin C and doxycycline (Sigma, 1 mg/ml) at 24 hr after infection. GFP+ colonies were counted at 12 days after infection. NBT/BCIP (Roche) was used for AP staining according the instructions of the manufacturer. GFP+ colonies were picked and maintained with mESC medium. AP staining and Immunofluorescence were used for iPSCs characterization.

Immunofluorescence Cells were washed twice with PBS and were fixed in 4% paraformaldehyde at room temperature for 20 min. The samples were then permeabilized with 0.2% Triton X-100 containing 10% goat serum (Sigma) in PBS at room temperature for 15 min. Samples were then washed three times with PBS and incubated with primary antibodies at 4C overnight. The antibody used for cell immunofluorescence was against NANOG (Abcam). The cells were then washed for four times and were incubated with secondary antibody in a cassette at room temperature for 1.5 hr. Then the cells were washed for four times with PBS and a final concentration of 0.1 mg/ml DAPI (Sigma) was included in the final wash to stain nuclei. Images were captured with an inverted microscope (DMI4000, Leica Microsystems).

Karyotyping and Generation of Chimeric Mice For karyotype analysis, cells were first plated onto a 6-well plate at 70%–80% confluence and demecolcine (Aladdin) was added to a final concentration of 20 mg/ml for 1 hr. Cells were then trypsinized, pelleted by centrifugation at 2,000 3 g for 5 min, resuspended in 8 mL of 0.075 M KCl, and incubated at 37C for 20 min. Fixative solution composed of 1 part of acetic acid and 3 parts methanol was added to a final volume of 10 ml, mixed gently, and incubated at 37C for 10 min. After further centrifugation, the supernatant was removed, and ice-cold fixative solution composed of 1 part acetic acid and 3 parts methanol were added to a final volume of 10 ml. Cells were dropped on a cold slide and incubated at 75C for 3 hr. Belts were treated with trypsin and colorant, and metaphases were analyzed on a Olympus BX51 microscope. For chimeric mice generation, blastocysts were obtained through mating of hormone primed female ICR/HaJ and male ICR/HaJ mice. Chimeras were produced by injecting 10-15 iPSCs into the cavity of the blastocysts, followed by implantation into pseudo pregnant ICR/HaJ mice. After implantation, chimeric mice were obtained 4 weeks later.

Generation of DDX5 Knockout mESCs Two sets of oligonucleotides were separately cloned into an U6-sgRNA vector to create paired sgRNA target sites for SpCas9-nick- ase in the first or second exon of Ddx5, encoding the DDX5 protein. CRISPR/Cas9 target sites in Ddx5 (NM_007840.3) were designed by http://zifit.partners.org/ZiFiT/CSquare9Nuclease.aspx. Then these two sets of oligonucleotides were synthesized and inserted into the Bbs I site of the U6-gRNA vector. The sequences of two sets of oligonucleotides are listed in the Key Resources Table. To generate stable Ddx5/ mESC lines, 2 3 106 mESCs were transfected with 20 mg U6-sgRNA vector and 50 mg pcDNA3.3- hCas9 plasmids. Cells were selected with 500 mg/ml of G418 (Sigma) after 24 hr of infection and were maintained in selection for 72 hr. Single clones were separated and were seeded in each well of a 48-well plate. Individual well was inspected by microscopy to exclude polyclonal cell lines. Monoclonal cell lines were expanded. Then, DNA of monoclonal cell lines was extracted to detect the genotype through DNA sequencing and protein was analyzed by western blot. shRNAs, siRNAs, miRNA mimics and inhibitors shRNA oligos, siRNAs, miRNA mimics and miRNA inhibitors were designed and synthesized by Shanghai GenePharma (GenePharma, China). Two pairs of shRNA oligos targeting on Ddx5 or Rybp gene were designed and constructed into pSUPER plasmid. Three siRNAs targeting on Rybp gene were designed and synthesized, the most effective siRNA (siRybp) identified by

Cell Stem Cell 20, 462–477.e1–e6, April 6, 2017 e4 qPCR was applied for the further experiments. 24 hr prior to transfection, cells were plated onto a 6-well plate at 40%–60% conﬂu- ence. Transfection was performed with Lipofectamine 2000 (Invitrogen, USA) according to the manufacturer’s protocol. 7.5 mL Lipofectamine 2000 reagent and siRNAs, miRNA mimics or inhibitors were diluted in Opti-MEM (GIBCO) and incubated at room temperature for 10 min. Then the mixtures were added to cells, and the ﬁnal concentration of siRNAs, miRNA mimics or inhibitors was 50 nM. The medium was replaced 4-6 hr after transfection with fresh culture medium. The sequences of shRNA oligos, siRNAs, miRNA mimics and miRNA inhibitors are listed in Table S4 and the Key Resources Table.

Quantitative RT-PCR Analysis RNA extraction was performed with Eastep Super Total RNA Extraction Kit (Promega). 1 mg of total RNA was then reverse tran- scribed with Superscript First-Strand Synthesis system (Promega). cDNAs of interest were then quantiﬁed with real-time qPCR ampliﬁcation. The primers used in the qRT-PCR assays are listed in Table S5. All the experiments were repeated for three times.

Luciferase Assay To construct a Rybp-30 UTR reporter plasmid, a 401-bp fragment encompassing the miR-125b binding site was amplified by PCR and cloned into pGL3 vector (Promega) at the Xba I site. To construct a mutant reporter plasmid, the seed region was mutated from CTCAGGGA to TATCAAGA. Renilla luciferase reporter was obtained from Promega and luciferase assay was conducted according to the manufacturer’s instructions. For luciferase assay, NIH 3T3 cells were plated into 24-well plate culture dishes. 2 mL of lipofectamine 2000 reagent, 0.8 mg of pGL3 vector and 80 ng of Renilla luciferase plasmid were diluted in Opti-MEM (GIBCO) and incubated at room temperature for 10 min. The mixtures were added to NIH 3T3 cells and the final concentration of miRNA mimics or inhibitors was 20 mM. Luciferase assays were repeated for three times. The amplification primers are listed in the Key Resources Table.

Western Blot and Endogenous Co-IP Assay Whole cell extracts were obtained with cell lysis buffer (50 mM Tris-HCl (pH 7.6), 1% Triton X-100, 1 mM EDTA, 10% glycerol, 1 mM DTT, 1 mM PMSF and protease inhibitor cocktail). Nuclear protein extracts were prepared as described (Yao et al., 2010). After centrifugation at 13,000 rpm for 10 min, soluble proteins were quantiﬁed by BCA, and 1 mg of proteins was used for each co-IP experiment. Samples were incubated overnight with 3 mg of antibodies and 20 mL of protein A- or G- dynabeads (Life technologies). Immnuno- precipitated materials were washed 3 times with high-salt buffer, loaded onto SDS-PAGE and transferred to PVDF membrane (In- vitrogen). Then the membrane was washed with TBS-T buffer and immunoblotting was performed with the indicated antibodies (listed in the Key Resources Table).

Histone Extraction Approximately 5 3 106 cells were collected and used to extract histones. Brieﬂy, cells were pelleted by slow spinning, washed once with 5 mL ice-cold PBS and re-suspended with 2 mL of hypotonic lysis buffer (10 mM Tris-HCl (pH 7.4), 1 mM KCl, 1.5 mM MgCl2, 1 mM DTT, 1 mM PMSF and protease inhibitor cocktail), incubated on ice for 30 min. Then, the nucleus was pelleted by spinning at 5000 rpm at 4 C for 10 min, resuspended in 0.4 mL H2SO4 (0.4 N) and rotated at 4 C for 30 min to overnight. After centrifugation at full speed for 10 min, the supernatant was precipitated with 20% TCA and incubated on ice for 30 min. Pellets were collected and washed twice with cold acetone. Pellets were resuspended in 30-50 mL of TE buffer and analyzed by western blot.

Sucrose Gradient Ultracentrifugation Whole cell extracts from mESCs were sedimented on a 10%–30% sucrose gradient by centrifugation using an OPTIMA L-100XP rotor (Beckmann) at 38,000 rpm at 4C for 16 hr. The gradient was fractioned and analyzed by SDS-PAGE. Indicated antibodies were used for western blot.

ChIP-Seq and Bioinformatic Analysis ChIP experiments were performed as previously described (Huang et al., 2013). Briefly, 1 3 107 cells were crosslinked with 1% form- aldehyde at room temperature for 10 min. Then the reaction was stopped by adding glycine (final concentration, 0.125 M). The cells were sonicated in SDS lysis buffer containing 1 3 protease inhibitor cocktails and 1 mM PMSF to achieve a chromatin sized of 100- 300 bp. The sonicated chromatin was incubated with indicated antibodies coupled with dynabeads protein A and G (1:1 mixed) at 4C overnight with rotation. Immune complexes were washed with the following buffers: low salt wash buffer (0.1% SDS, 1%Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.0), 150 mM NaCl), high salt wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl (pH 8.0), 500 mM NaCl), LiCl wash buffer (0.25 M LiCl, 1% IGEPAL-CA630, 1% deoxycholic acid (sodium salt), 1 mM EDTA, 10 mM Tris-HCl (pH 8.0)) and TE buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA). Antibody-bound chromatin was reverse-crosslinked, and the ChIPed DNA samples were purified for sequencing. The ChIP DNA libraries were constructed following the Illumina ChIP-Seq library generation protocol. ChIP-seq data was aligned to the mm10 mouse genome assembly using bowtie2 (Langmead et al., 2009) and peaks were discovered using either MACS (v1.4.2) (Zhang et al., 2008) for RYBP or DFilter (v1.0) (Kumar et al., 2013) for H2AK119ub1 data. Overlaps between different ChIP-seq libraries were performed using MAnorm (Shao et al., 2012). Gene ontology analysis was performed using GREAT (McLean et al., 2010). ChIP-seq data was verified with ChIP-qPCR. The primers used in the ChIP-qPCR assays are listed in Table S6. e5 Cell Stem Cell 20, 462–477.e1–e6, April 6, 2017 RNA-Seq, miRNA-Seq and Bioinformatic Analysis RNA-Seq data was processed as described in (Hutchins et al., 2015). Briefly, reads were aligned to a transcriptome index built from the Ensembl transcriptome version 76 (mm10), using RSEM (v1.2.19) (Li and Dewey, 2011) and bowtie2 (v2.2.5) (Langmead et al., 2009) and normalized with EDASeq (Risso et al., 2011). For a gene to be considered expressed the gene must have at least 15 normalized tags in any 2 samples. Other analysis was performed using glbase (Hutchins et al., 2014). miRNA-seq data was processed using the CAP-miRSeq pipeline (Sun et al., 2014), using the miRDeep2 mapper (Friedlander€ et al., 2012). Differential expression of miRNAs was called using edgeR (Robinson et al., 2010).

QUANTIFICATION AND STATISTICAL ANALYSIS

Data are presented as mean values ± SD unless otherwise indicated in figure legends. Sample number (n) indicates the number of independent biological samples in each experiment. Sample numbers and experimental repeats are indicated in figures and figure legends or methods section above. Statistical tests were selected based on appropriate assumption (Levene’s Test for Equality of Variances) with respect to data distribution and variance characteristics. Statistical significance was determined by Student’s t test analysis (two-tailed) for two groups. Differences in means were considered statistically significant at p < 0.05. Significance levels are: * p < 0.05; ** p < 0.01; *** p < 0.001. For RNA-seq analysis, differential gene expression calling was done using a q value < 0.05 according to the statistical framework implemented by DESeq2. For microRNA-seq, differential microRNAs were called using edgeR and a FDR < 0.01.

DATA AND SOFTWARE AVAILABILITY

The accession number for the RNA-sequencing, microRNA-sequencing, and ChIP sequencing data reported in this paper is NCBI GEO: GSE76825.

Cell Stem Cell 20, 462–477.e1–e6, April 6, 2017 e6