Transcriptional pause release is a rate-limiting step for somatic cell reprogramming.

Article

Transcriptional Pause Release Is a Rate-Limiting Step for Somatic Cell Reprogramming Graphical Abstract

Authors Longqi Liu, Yan Xu, ..., Jun Wang, Miguel A. Esteban

Correspondence [email protected] (Y.X.), [email protected] (M.A.E.)

In Brief Liu et al. demonstrate that the transition of Pol II from a paused to a productive elongation stage is a rate-limiting step for inducing pluripotency genes in reprogramming. This transition is promoted by the recruitment of active P-TEFb to pluripotency gene promoters, which is simultaneously coordinated by BRD4 and KLF4.

Highlights

Accession Numbers

Pol II is paused at pluripotency genes during reprogramming

GSE59833

P-TEFb induces transcriptional elongation at pluripotency genes in reprogramming BRD4 and HEXIM1 have opposite roles in reprogramming KLF4 helps recruit P-TEFb to pluripotency genes in reprogramming and ESCs

Liu et al., 2014, Cell Stem Cell 15, 1–15 November 6, 2014 ª2014 Elsevier Inc. http://dx.doi.org/10.1016/j.stem.2014.09.018

Please cite this article in press as: Liu et al., Transcriptional Pause Release Is a Rate-Limiting Step for Somatic Cell Reprogramming, Cell Stem Cell (2014), http://dx.doi.org/10.1016/j.stem.2014.09.018

Cell Stem Cell

Article Transcriptional Pause Release Is a Rate-Limiting Step for Somatic Cell Reprogramming Longqi Liu,1,2,5,14 Yan Xu,1,2,5,14,* Minghui He,3,14 Meng Zhang,1,2,4 Fenggong Cui,1,2 Leina Lu,6 Mingze Yao,2 Weihua Tian,1,2 Christina Benda,1,2,5 Qiang Zhuang,1,2,5 Zhijian Huang,1,2 Wenjuan Li,1,2,5 Xiangchun Li,3 Ping Zhao,1,2,5 Wenxia Fan,1,2 Zhiwei Luo,1,2,5 Yuan Li,2 Yasong Wu,2,5 Andrew P. Hutchins,2 Dongye Wang,1,2,7 Hung-Fat Tse,8,9,10 Axel Schambach,11 Jon Frampton,12 Baoming Qin,2,9 Xichen Bao,1,2 Hongjie Yao,2 Biliang Zhang,13 Hao Sun,6 Duanqing Pei,2,9 Huating Wang,6 Jun Wang,3 and Miguel A. Esteban1,2,9,* 1Laboratory of Chromatin and Human Disease, Chinese Academy of Sciences, South China Institute for Stem Cell Biology and Regenerative Medicine, Guangzhou Institutes of Biomedicine and Health, Guangzhou 510530, China 2Key Laboratory of Regenerative Biology, Chinese Academy of Sciences, South China Institute for Stem Cell Biology and Regenerative Medicine, Guangzhou Institutes of Biomedicine and Health, Guangzhou 510530, China 3BGI-Shenzhen, Shenzhen 518083, China 4School of Life Sciences, Anhui University, Hefei 230601, China 5University of Chinese Academy of Sciences, Beijing 100049, China 6Li Ka Shing Institute of Health Sciences, Department of Orthopaedics and Traumatology, The Chinese University of Hong Kong, Hong Kong SAR, China 7Drug Discovery Pipeline Group, Guangzhou Institutes of Biomedicine and Health, Guangzhou 510530, China 8Cardiology Division, Department of Medicine, Queen Mary Hospital, the University of Hong Kong, Hong Kong SAR, China 9Hong Kong-Guangdong Joint Laboratory on Stem Cell and Regenerative Medicine, the University of Hong Kong and Guangzhou Institutes of Biomedicine and Health, China 10Shenzhen Institutes of Research and Innovation, the University of Hong Kong, Hong Kong SAR, China 11Department of Experimental Hematology, Hannover Medical School, Hannover D-30625, Germany 12Institute for Biomedical Research, College of Medical and Dental Sciences, University of Birmingham, Birmingham B15 2TT, UK 13Laboratory of RNA Chemical Biology, State Key Laboratory of Respiratory Diseases of the Chinese Academy of Sciences, Guangzhou Institutes of Biomedicine and Health, Guangzhou 510530, China 14Co-first author *Correspondence: [email protected] (Y.X.), [email protected] (M.A.E.) http://dx.doi.org/10.1016/j.stem.2014.09.018

SUMMARY

Reactivation of the pluripotency network during somatic cell reprogramming by exogenous transcription factors involves chromatin remodeling and the recruitment of RNA polymerase II (Pol II) to target loci. Here, we report that Pol II is engaged at pluripotency promoters in reprogramming but remains paused and inefficiently released. We also show that bromodomain-containing protein 4 (BRD4) stimulates productive transcriptional elongation of pluripotency genes by dissociating the pause release factor P-TEFb from an inactive complex containing HEXIM1. Consequently, BRD4 overexpression enhances reprogramming efficiency and HEXIM1 suppresses it, whereas Brd4 and Hexim1 knockdown do the opposite. We further demonstrate that the reprogramming factor KLF4 helps recruit P-TEFb to pluripotency promoters. Our work thus provides a mechanism for explaining the reactivation of pluripotency genes in reprogramming and unveils an unanticipated role for KLF4 in transcriptional pause release. INTRODUCTION The enforced expression of defined transcription factors can change cell fate (Sindhu et al., 2012). A striking example of this

phenomenon is the reprogramming of induced pluripotent stem cells (iPSCs) from somatic cells by the Yamanaka factors (OCT4, SOX2, KLF4, and c-MYC; OSKM) (Takahashi and Yamanaka, 2006). Because of their ability to differentiate into diverse cell lineages, iPSCs provide a potential supply of cells for regenerative medicine and also an excellent platform for in vitro disease modeling and toxicology screening (Saha and Jaenisch, 2009). Reprogramming is characterized by the existence of roadblocks that tend to derail the process (Apostolou and Hochedlinger, 2013). Gene expression analyses of bulk populations have helped clarify these roadblocks and contributed to dividing reprogramming into distinct phases, e.g., initiation, maturation, and stabilization (Samavarchi-Tehrani et al., 2010). The notion that gene transcription in reprogramming is phased has also been validated using single-cell profiling at different time points (Buganim et al., 2012) and with expression arrays of defined intermediate cell populations (Polo et al., 2012). Elucidating the rate-limiting steps regulating the different phases of gene transcription in reprogramming is important because this may help improve the methodology. Gene transcription in metazoans is highly regulated and comprises multiple steps (Min et al., 2011). Formation of the preinitiation complex involves the recruitment of several general transcription factors in addition to Pol II and has traditionally been considered the major rate-limiting step of gene transcription. Consistent with this idea, overexpressing specific subunits of the general transcription factor TFIID potentiates reprogramming by facilitating pluripotency gene transcription (Pijnappel Cell Stem Cell 15, 1–15, November 6, 2014 ª2014 Elsevier Inc. 1


Cell Stem Cell Pol II Pause Release Is Limiting for Reprogramming

A

B

C

D

E

F

G

H

I

J

K

Figure 1. Pol II Is Paused at Pluripotency Genes in Reprogramming (A) GRO-sequencing maps of Rcor1 and Pttg1 in ESCs from the reanalysis of a previous data set (GSE27037). (B) Venn diagrams show the number of genes upregulated in the first and second wave of reprogramming that belong to class II in MEFs, ESCs, or both. (C) Schematic representation describing how to calculate the TR. (D) Percentage of second wave genes that experience Pol II pausing at day 8 of reprogramming. D indicates day. (E) Pol II TR distribution of second wave genes paused at day 8 of reprogramming compared to day 5 and untransduced MEFs. (F) Box plots showing the average TR of second wave genes paused at day 8 of reprogramming compared to day 5 and untransduced MEFs. (G) Percentage of second wave genes paused at day 8 of reprogramming that experience pause release in ESCs. (legend continued on next page)

2 Cell Stem Cell 15, 1–15, November 6, 2014 ª2014 Elsevier Inc.



et al., 2013). Pausing of activated Pol II 30–50 nucleotides downstream of +1 and its release by the P-TEFb complex constitutes another rate-limiting step of gene transcription that has received growing attention in recent years (Levine, 2011). Although Pol II pause release was originally thought to represent a rather specialized mechanism of transcriptional regulation, analysis of nascent RNAs using global run-on sequencing (GRO-sequencing) (Min et al., 2011) has shown that around one-third of genes in the metazoan genome experience this mode of regulation at some point during the organism’s lifetime (Levine, 2011). In this regard, genes regulated through pause release are particularly prevalent in stress responses and during embryonic development (Smith and Shilatifard, 2013). Potentially, this is because Pol II pausing and its subsequent release enables a method of rapid and synchronous activation of gene expression programs. Pol II pausing might also be a mechanism to hold the activation of specialized gene programs during cell fate conversions if not stimulated properly through pause release. Here, we demonstrate that the transition of Pol II from a paused to a productive elongation stage is a rate-limiting step for inducing pluripotency genes in the late phase of reprogramming. This transition is promoted by the recruitment of active P-TEFb to pluripotency gene promoters, which is simultaneously coordinated by BRD4 and KLF4. Our study thus proposes a revised model for the reactivation of the pluripotency network in reprogramming. RESULTS A Major Change in Transcriptional Regulation Characterizes the Late Phase of Reprogramming It has been reported that reprogramming comprises two major waves of gene transcription (Polo et al., 2012). The first wave involves the upregulation of genes related to proliferation, metabolism, and cytoskeletal organization, while the second wave is mostly genes related to the pluripotency network. Because mouse embryonic fibroblasts (MEFs) are very different from pluripotent cells, we envisaged that variations in the mode of transcriptional regulation occur during reprogramming. To test this, we initially contrasted the genes upregulated in the two waves of reprogramming with a previous GRO-sequencing data set of MEFs and embryonic stem cells (ESCs) (Min et al., 2011). This study divided genes into four classes: (1) not paused and transcribed, which show concentration of GRO-sequencing reads along the gene body (class I, Figure 1A); (2) paused and transcribed, which show concentration of GRO-sequencing reads in the proximal promoter and less pronouncedly along the gene body (class II, Figure 1A); (3) paused and not transcribed; and (4) not paused and not transcribed. Roughly 35%–40% of all RefSeq genes were shown to belong to class II in both MEFs and ESCs, but the kind of genes varied in each cell type. Our comparative analysis showed that genes upregulated in the first wave of reprogramming mostly belong to class II in both ESCs and MEFs (606 genes out of 816) (Figure 1B and Fig-

ure S1A available online). Conversely, the second wave is substantially depleted of class II genes in both MEFs and ESCs and is enriched in class II genes in ESCs only (Figure 1B and Figure S1A). Average density of nascent RNA along the promoter and gene body regions of all class II genes induced in the second wave of reprogramming confirmed higher levels of Pol II pausing and pause release in ESCs than in MEFs (Figure S1B). We conclude that many genes activated in the second wave of reprogramming must experience a change in their mode of transcriptional regulation compared to MEFs. This change consists in the pausing of Pol II at their proximal promoters and the subsequent pause release to enter productive elongation. Pol II Is Paused at Pluripotency Promoters during Reprogramming We then hypothesized that productive transcriptional elongation of pluripotency genes could represent a rate-limiting step of reprogramming. To explore this, we performed ChIP-sequencing for Pol II in MEFs reprogrammed with OSKM at days 5 and 8; ESCs and untransduced MEFs were used as controls. The Pol II traveling ratio (TR), which compares the average density of Pol II in the proximal promoter and gene body (Figure 1C), was employed to calculate the degree of pausing among different genes. Consistent with previous reports, we used a TR > 4 as indicative of Pol II pausing (Zeitlinger et al., 2007). Notably, approximately half of the genes upregulated in the second wave of reprogramming displayed a TR > 4 in OSKM reprogramming at day 8 (Figure 1D), but the same genes showed a significantly lower value at day 5 and in untransduced MEFs (Figures 1E and 1F). There was good correlation between second wave genes paused at day 8 of reprogramming and those classified as class II in ESCs in the GRO-sequencing data analysis (Figure S1C). In addition, comparison with our Pol II ChIP-sequencing data in ESCs showed that a large fraction (55%) of second wave genes has lower TR in ESCs than at day 8 of reprogramming, indicating that they experience productive elongation in ESCs (Figures 1G–1I). Among those genes there are key pluripotency regulators including Oct4 and Nanog (Figure 1G). Interestingly, Oct4 showed substantial pausing at day 5 as well, indicating that for some pluripotency regulators Pol II is recruited early to their promoters but remains paused until the late phase (Figure 1J). Notably, the noncoding pluripotency regulator Mir290 also exhibited Pol II pausing at day 8 of reprogramming (Figure S1D). The Pol II binding pattern of a class II gene in MEFs, Thy1, which is downregulated early in reprogramming, is shown as a control (Figure 1K). Therefore, Pol II pausing at pluripotency promoters characterizes reprogramming, strongly suggesting that productive transcriptional elongation is limiting for this process. Pause Release Factor P-TEFb Controls the Reactivation of Pluripotency Genes in Reprogramming To demonstrate that transcriptional pause release of pluripotency genes is limiting for reprogramming, we focused on

(H) Pol II TR distribution of second wave genes paused at day 8 of reprogramming that are released in ESCs. (I) Box plots showing the average TR of second wave genes paused at day 8 of reprogramming that are released in ESCs. (J and K) Pol II occupancy on Oct4 and Thy1 in MEFs, in ESCs, and at two different time points during reprogramming. See also Figure S1.

Cell Stem Cell 15, 1–15, November 6, 2014 ª2014 Elsevier Inc. 3



B Brightfield

ME Fs ES Cs

D0 3

H3

H3

6

9

AP

12 CDK9-DN

***

80 60 40 20 0

Brightfield

C

E

OSKM

OSKM AP

GFP

Number of GFP+ colonies

D

D

C t K 9 rl -D N

CDK9

GFP

Ctrl

OSKM

CDK9

OSKM

C

OSKM


A

shCtrl

shCdk9 #1

F

Early phase D3 CDK9-DN

-

D4

+

-

+

Late phase D5

-

D10

+

-

+

G

20

-D

−2

1st wave

D

2nd wave

D

9

+Dox

+Dox

K

En

rrb Es

8 n2

No change

0

Li

Down

*

15

N

Up

**

2

59.9%

79.1%

**

30

ox

49 2483

33.5%

**

-S

64 2355

10.7%

*

45

do

Early phase Late phase

6.6%

OSKM + CDK9-DN

En

10.2%

Down

2nd wave

OSKM

do -O ct 4

RefSeq genes

og

Late Phase RefSeq genes

Ctrl

an

J

Relative expression level

I

Up

2

0

11

D

20

D

0-

40

D

D

60

0-

5

7 D

D

0-

3

D

D

0-

1

D 0-

D

D

-D

D

0-

ox

0

>6

Expression (log2)

40

**

80

9D 1 7- 1 D 11


*

60

OSKM + teton CDK9-DN

** *

80

ox

OSKM + teton CDK9-DN Number of GFP+ colonies

OSKM

H

s sh hC C trl dk sh 9 C #1 dk 9 #2

shCdk9 #2

*** ***

70 60 50 40 30 20 10 0

Figure 2. P-TEFb Regulates the Activation of Pluripotency Genes in the Late Phase of Reprogramming (A) Western blot for endogenous CDK9 in MEFs and ESCs (left) or during reprogramming (right). Histone H3 (H3) was used as loading control. (B) Phase contrast (left), fluorescence photographs (middle), and AP-stained wells (right) of MEFs reprogrammed with OSKM and CDK9-DN at day 10. An empty vector was used as a control (Ctrl; also hereafter in similar experiments). Scale bar represents 50 mm. (C) Number of GFP+ colonies at day 12 in MEFs reprogrammed as in (B). Average colony numbers of three independent experiments are shown. (D) Phase contrast (left), fluorescence photographs (middle), and AP-stained wells (right) of MEFs reprogrammed with OSKM and two independent shRNAs for Cdk9 (shCdk9) at day 10. shRNA against firefly luciferase was used as a control (shCtrl; also hereafter in similar experiments). Scale bar represents 50 mm. (E) Number of GFP+ colonies at day 12 in MEFs reprogrammed as in (D). Average colony numbers of three independent experiments are shown. (F and G) Number of GFP+ colonies at day 11 in MEFs reprogrammed with OSKM and an inducible CDK9-DN (teton CDK9-DN). Doxycycline (dox; 2 mg/ml) was administered daily for the indicated time periods. The dotted lines are used to highlight deviations from the mean number of colonies in untreated control wells (also hereafter in selected experiments). Average colony numbers of two independent experiments are shown. (H) Heatmap of RNA-sequencing data showing the expression level of genes corresponding to the first (for days 3, 4, and 5) and second (for day 10) wave of reprogramming in MEFs reprogrammed with OSKM and empty vector or CDK9-DN. (I) Number of RefSeq genes significantly differentially regulated (fold change >1.25, false discovery rate or FDR < 0.01) by CDK9-DN in the early phase (in all three time points) and late phase of OSKM reprogramming. (legend continued on next page)




CDK9, the catalytic subunit of the P-TEFb complex, because it stimulates elongation by phosphorylating Pol II and the negative elongation factors DSIF and NELF (Zhou and Yik, 2006). We observed increased CDK9 expression (assessed by western blotting and quantitative PCR [qPCR]) in ESCs compared to that in MEFs, and we also observed such during the reprogramming of MEFs with OSKM factors delivered as retroviruses (Figure 2A and Figures S2A and S2B). To reduce CDK9 activity, we prepared a retrovirus producing a dominantnegative form (a kinase inactive mutant) of CDK9 (CDK9-DN) and two independent retroviral shRNA vectors (Figure S2C), which were transduced along with OSKM into MEFs bearing an Oct4-GFP transgenic reporter. Reducing CDK9 activity (tested using CDK9-DN) did not alter exogenous OSKM expression level and only impaired cell proliferation moderately (Figures S2D and S2E). Notably, OSKM reprogramming with CDK9-DN or the two shRNAs allowed the formation of AP+ colonies (an early marker of reprogramming) (Buganim et al., 2013), but these were mostly GFP compared to the controls (Figures 2B–2E), indicating incomplete reprogramming. We achieved the same result using a different reprogramming system, Oct4-GFP transgenic MEFs with an integrated OSKM cassette in the Col1a1 locus, hereafter termed secondary MEFs (Figures S2F and S2G). We also overexpressed CDK9 during OSKM reprogramming but observed no change in the number of GFP+ colonies (Figure S2H). To confirm the selective role of CDK9 in the late phase of reprogramming, we employed a doxycycline-inducible lentivirus producing CDK9-DN (Figure S2I) and a known CDK9 inhibitor (flavopiridol) (Rahl et al., 2010). Both approaches impaired the appearance of GFP+ colonies if induced/ administered in the late phase (day 7 to day 11) of reprogramming, but not in the early phase (day 0 to day 5) (Figures 2F and 2G and Figures S2J and S2K). We also performed RNA-sequencing of MEFs reprogrammed with OSKM in the presence or absence of CDK9-DN. Interestingly, CDK9-DN induced rather modest changes in gene transcription in the early phase of reprogramming compared to the control (Figures 2H and 2I). Of note, the well-known CDK9 target Hexim1 was among the downregulated genes (He et al., 2006) (Table S1). Conversely, CDK9-DN led to the downregulation of a large number of genes at day 10, many of which belong to the second transcriptional wave of reprogramming (Figures 2H–2J and Table S1). The selective effect of ablating CDK9 function on the late phase of reprogramming was validated by qPCR of mesenchymal and epithelial genes in MEFs reprogrammed with OSKM and CDK9-DN at day 5 (Figure S2L) and by qPCR of pluripotency genes at day 9 (Figure 2K). Overall our data support the idea that high levels of P-TEFb activity are dispensable in the early phase of reprogramming but instrumental for reactivating the pluripotency network in the late phase.

BRD4 Is a Positive Regulator of Reprogramming The experiments above showing that CDK9 overexpression does not enhance the formation of GFP+ colonies led us to speculate that CDK9 enzymatic activity, but not its overall expression level, is a limiting factor for reprogramming. Supporting this possibility, a big proportion of CDK9 is held inactive in a complex containing HEXIM1 in many cell types (Zhou and Yik, 2006). It is also well known that the bromodomain and extraterminal (BET) family member BRD4 activates CDK9 by displacing HEXIM1. BRD4 thus appeared as a candidate for further exploration. There are two different splice isoforms of BRD4 (short and long). They share the amino terminal domain (two tandem bromodomains [BDs]), responsible for interacting with acetylated histones, and the extraterminal (ET) domain (Belkina and Denis, 2012). However, only the carboxy terminal domain (CTD) of the long isoform allows interaction with CDK9 and its activation. The short isoform of BRD4 displayed low expression level by qPCR (based on the qPCR cycle threshold [Ct] value) in MEFs and ESCs (Figure S3A), so we did not study it further. Notably, qPCR and western blot of the long isoform of BRD4 (hereafter named BRD4) showed higher expression in ESCs compared to MEFs and an increase of such during OSKM reprogramming (Figure 3A and Figures S3A and S3B). We then reduced BRD4 levels in OSKM reprogramming with two independent retroviral shRNA vectors (Figure S3C). The two shRNAs enabled the formation of AP+ colonies but greatly diminished the appearance of GFP+ colonies (Figures 3B and 3C), matching the pattern previously observed for CDK9 inhibition. Moreover, we could not detect any obvious change in cell proliferation (Figure S3D). The negative effect on reprogramming efficiency of Brd4 knockdown was confirmed using secondary MEFs (Figure S3E). In addition, we observed that the BRD4 inhibitor JQ1 (Belkina and Denis, 2012) did not affect reprogramming efficiency when administered in the early phase, but significantly reduced reprogramming efficiency in the late phase (Figures 3D and 3E). Next, we reprogrammed MEFs with OSKM and a retroviral vector producing BRD4. This increased the number of GFP+ colonies (3-fold) without noticeably changing the appearance of AP+ colonies (Figures 3F and 3G). Exogenous BRD4 also accelerated the formation of GFP+ colonies compared to the control (Figure 3H) but did not affect cell proliferation (Figure S3F). The enhancing effect on reprogramming of BRD4 overexpression was confirmed using secondary MEFs (Figure S3G). Moreover, iPSC colonies generated with BRD4 and OSKM contained the Brd4 retrovirus in their genome and were fully pluripotent (Figure 3I and Figures S3H–S3K). These data indicate that BRD4 is a positive regulator of the late phase of reprogramming. BRD4 Potentiates Reprogramming through P-TEFb Because BRD4 has CDK9-independent functions (Belkina and Denis, 2012), we tested whether it regulates reprogramming

(J) Percentage of RefSeq and second wave genes upregulated and downregulated (fold change >1.25, FDR < 0.01) or unchanged by CDK9-DN at day 10 of reprogramming. (K) qPCR for the indicated genes using RNA lysates from MEFs reprogrammed with OSKM and empty vector or CDK9-DN at day 9. Endo- indicates endogenous. Values are referred to untransduced MEFs. The data in (C), (E), (F), (G), and (K) are reported as mean values ± standard deviation (SD) with the indicated significance (*p < 0.05, **p < 0.01, ***p < 0.005). See also Figure S2 and Table S1.




30

F

G

**

60 40 20 0


*

Ctrl

*

100

0- 7 D 9 0D D 11 0D 13

5

D

D

0D

D

C

D

BRD4 inhibitor

I BRD4-C11

BRD4

AP

NANOG

DAPI

Merge

Brightfield

GFP

Germline

80

H

60 40

OSKM Ctrl BRD4

20 0

D0 3 6 9 12 15 GFP+ colony appearance

D

C tr 9- l D D 11 7D 11


80

OSKM

OSKM

OSKM

C tr BR l D 4

E

3

0

s sh hC B r trl d sh 4 # Br 1 d4 #2

H3

20 10 0

D

shBrd4 #2

0-

12

D

9

1

6

BRD4

*

90 60

D

OSKM

*

0-

shBrd4 #1

120

trl

H3

70 60 50 40 30

D

shCtrl

OSKM

*** ***

D

AP

GFP

BRD4

D0 3

D OSKM


Brightfield

M EF ES s C s

C

OSKM

0-

B


A

BRD4 inhibitor

Figure 3. BRD4 Enhances Reprogramming (A) Western blot for endogenous BRD4 in MEFs and ESCs (top) or during reprogramming (bottom). Histone H3 (H3) was used as the loading control. (B) Phase contrast (left), fluorescence photographs (middle), and AP-stained wells (right) of MEFs reprogrammed with OSKM and shRNA control or two shRNAs for Brd4 (shBrd4) at day 10. Scale bar represents 50 mm. (C) Number of GFP+ colonies at day 12 in MEFs reprogrammed as in (B). Average colony numbers of three independent experiments are shown. (D and E) Number of GFP+ colonies in MEFs reprogrammed with OSKM and the BRD4 inhibitor JQ1 (100 nM). DMSO was used as a control. Colonies were counted at day 13 (D) or day 11 (E). Average colony numbers of two independent experiments are shown. (F) Number of GFP+ colonies at day 11 in MEFs reprogrammed with OSKM and empty vector or BRD4. Average colony numbers of three independent experiments are shown. (G) AP-stained wells of a representative experiment as in (F) at day 10. (H) Schematic representation of the time of GFP+ colony appearance during OSKM reprogramming with or without exogenous BRD4. (I) Immunofluorescence, phase contrast photographs, and chimeric mice with germline transmission (red arrow) obtained with a representative iPSC colony produced with OSKM and BRD4. Scale bar represents 50 mm. The data in (C)–(F) are reported as mean values ± SD with the indicated significance (*p < 0.05, **p < 0.01, ***p < 0.005). See also Figure S3.

specifically through the activation of CDK9. We prepared retroviral vectors producing a series of truncations that correspond to the major functional domains of BRD4 (Figure 4A). Only those BRD4 constructs retaining the CTD, including BRD4-CTD, were able to pull down endogenous CDK9 when overexpressed in HEK293T cells (Figure 4B). We overexpressed these different truncations together with OSKM in MEFs. Interestingly, BRD4BD acted as a dominant-negative for endogenous BRD4, reducing GFP+ colonies (Figure 4C) but without obviously changing the number of AP+ colonies (Figure S4A). This effect is likely related to the ability of this construct to diminish the interaction of endogenous BRD4 with chromatin. As predicted, BRD4-D-CTD also failed to enhance, and even reduced, the number of GFP+ colonies compared to the control (Figure 4C). In addition, we observed that BRD4-CTD reproducibly enhanced the formation of GFP+ colonies 3-fold over intact BRD4 (Figure 4C). This suggested that the ET domain of BRD4 might contain a module that impairs reprogramming, perhaps through 6 Cell Stem Cell 15, 1–15, November 6, 2014 ª2014 Elsevier Inc.

interaction with negative regulators. Supporting this idea, a truncated form of BRD4 (BRD4-D-ET) that only lacks the ET domain (Figure 4A) increased the number of GFP+ colonies to a level closer to that of BRD4-CTD (Figure 4C). One possible candidate mediating this negative effect is CHD4, which belongs to the MBD3/NuRD complex with repressive function in reprogramming (Rais et al., 2013), because it is known to interact with BRD4 through the ET domain (Rahman et al., 2011). We also detected that overexpressing BRD4-ET inhibited rather than enhanced the number of GFP+ colonies (Figure 4C). This implies that overexpressed BRD4-ET might interact as well with positive regulators of reprogramming, thus titrating down their interaction with endogenous BRD4. To further show that the effect of BRD4-CTD is mediated through CDK9, we prepared two mutant forms (CTD-FEE/AAA and CTD-R/P) that cannot interact with CDK9 (Figure 4D). Consistent with previous studies (Bisgrove et al., 2007), these mutants could not pull down endogenous CDK9 in



47

62

IP: FLAG IB: CDK9

13

10

0 60

CTD 8

ET 67

1 58 16 9 34 9 46 1

BD I BD II

BRD4 BRD4-BD BRD4-ET BRD4-CTD BRD4-Δ-CTD BRD4-Δ-ET

C

Ct rl BR D4 BR D4 BR -BD D4 BR -ET D4 BR -CT D D BR 4-ΔD 4 CT -Δ D -E T

B

OSKM Number of GFP+ colonies

A

CDK9

FLAG

**

** ***

200 150 *

100

*** ***

50 0

C t B rl BR R D D 4 4 BR -B D BR D4BR D4 ET D -C 4- TD BR Δ-C D TD 4ΔET

Input

250

CDK9

WT 1335 ERRRREAMAATIDMNFQSDLLSIFEENLF 1362 FEE/AAA ERRRREAMAATIDMNFQSDLLSIAAANLF ERPRREAMAATIDMNFQSDLLSIFEENLF R/P

Input

200

G

OSKM

160 120 80 40 0

80 60 40 20 0

15

30

20

10

15

10

5

0

0

0

BR

D

C

tr 4- l C TD

30

C tr 4- l C TD

45

BR

C t BR rl D E 4 BR /A D AA 4R /P FE

4-

D

CTD-UC8 NANOG

DAPI

Merge

BRD4-CTD SSEA-4

DAPI

Merge

Ctrl

20

BR

C **

40

I

**

C tr 4- l C TD

**

hUCs #3 (♂ ♂) + OSKM

D

60

hUCs #2 (♀ ♀) + OSKM

BR

hUCs #1 (♀ ♀) + OSKM

D

Number of ESC-like colonies

H

*

100

BR C TD D4 trl -F -CT EE D /A C AA TD -R /P

FLAG

OSKM **


IP: FLAG IB: CDK9

trl BR D C 4-C TD T C -FE D TD E -R /A /P AA

CDK9 binding site

C

CTD BRD4


F

E

D

Figure 4. BRD4 Enhances Reprogramming through P-TEFb (A) Schematic depiction of wild-type BRD4 and the different truncated BRD4 mutants used in this study. (B) Immunoprecipitation (IP) of the different BRD4 variants (FLAG-tagged) shown in (A) in HEK293T cells. FLAG-tagged GFP was used as control. Membranes were immunoblotted (IB) for CDK9 and FLAG. Black arrows indicate specific bands. (C) Number of GFP+ colonies at day 11 in MEFs reprogrammed with OSKM and empty vector or the indicated BRD4 variants. Average colony numbers of three independent experiments are shown. (D) Schematic depiction of the location of the CDK9 binding site on BRD4 and the corresponding modifications performed to abolish their interaction. (E) Immunoprecipitation of the indicated BRD4 variants (FLAG-tagged) in HEK293T cells. FLAG-tagged GFP was used as a control. Membranes were immunoblotted for CDK9 and FLAG. (F and G) Number of GFP+ colonies at day 11 in MEFs reprogrammed with OSKM and empty vector or the indicated BRD4 variants. Average colony numbers of three (F) or two (G) independent experiments are shown. (H) Number of human ESC-like colonies at day 28 in human urinary cells (hUCs) from three different donors reprogrammed with OSKM and BRD4-CTD. (I) Phase contrast and immunofluorescence photographs of human iPSCs produced from urinary cells transduced with OSKM and BRD4-CTD. Scale bar represents 50 mm. The data in (C) and (F)–(H) are reported as mean values ± SD with the indicated significance (*p < 0.05, **p < 0.01, ***p < 0.005). See also Figure S4.

immunoprecipitation assays (Figure 4E) and failed to increase the formation of GFP+ colonies over the baseline in MEFs reprogrammed with OSKM (Figure 4F). Introducing the same mutations into full-length BRD4 (BRD4-FEE/AAA and BRD4-R/P)

had the same outcome (Figure 4G). We also validated the effect of all BRD4 constructs in secondary MEFs (Figures S4B and S4C). Likewise, we employed a polycistronic OSKM lentiviral vector and BRD4-CTD to reprogram a noninvasive Cell Stem Cell 15, 1–15, November 6, 2014 ª2014 Elsevier Inc. 7



B

Li

-S

1

2

0.30

0.25

0.25

0.20

0.20

*

0.15

*

0.01

0

0

1 **

**

**

0.18

D

2

3

1 ai

2

Sn

4

2

0.02 0

3

2 **

1

3

2 **

0.08

**

0.06 0.04

0.15

0.06

0.10

0.03

0.05

0.02 0

0

1

3

2

1

F

2

2

1

3

Nanog

1 kb

shBrd4 #1

0.10

0.04

0.20

ESCs shCtrl

*

0.06

0.25

0

0

3

0.30

0.09

0.05

2

1

0.08

0.35

0.12

0.10

*

1

3

2

0.15 0.15

1

on

0.12

0.03 0.02

Ex

3

0.04

0.05

0.20

3

0.05

0.05

3

Ex

on

0.06

*

0.10

2

2 **

0.07

0.15

0.10

% input

1

1

3

0.30

Sox2

1 kb

on

Ex

3

Sall4

5 kb 2 3 4 5 on on on on Ex Ex Ex Ex

1

on

Ex 2 o Ex n 3 on 4

on

1 on Ex

2

Ex

on

on

Oct4

1 kb

Ex

Nanog

Ex

Ser2P BRD4-CTD

1

Ze b

En

Ser2P Ctrl

Pol II BRD4-CTD

1

3

rrb

do

do En

Pol II Ctrl

**

Primers

0.01

IgG BRD4-CTD

1

Ser2P

1 0.1

IgG Ctrl

0.35

Primers

10

Es

n2

2 ox

4 -O ct

an N

8

0

1 kb

% input

**

100

ld

***

100

C

Pol II

**

OSKM + BRD4-CTD

1

**

200

OSKM

1000

C

OSKM + BRD4-CTD

dh

***

Ex

300

Ctrl


OSKM

og


Ctrl

C

A

1 kb

3

Oct4

shBrd4 #2 100

150

BRD4

AP

1

1

0.6 **

0.4 0.2

** ** *** ***

**

**

og O ct 4 So x2 Li n2 8 Es rrb

0 N

15 4

*** **

3 *

2

**

** **

**

shCtrl *

shBrd4 #1 shBrd4 #2

1 0 Fg f5 Pa x6 So x1 T/ Br a C dx 2 Eo m es G at a4 G at a5 Fo xa 2

0.8


1.0

an


E

(legend on next page)




reprogrammable human cell source, urinary cells (Zhou et al., 2012). BRD4-CTD significantly increased the number of human ESC-like colonies using cells from three independent donors (Figure 4H). We picked eight independent colonies from one donor; they could be expanded retaining ESC-like morphology (Figure 4I), had integrated the BRD4-CTD transgene, and were fully pluripotent (Figure 4I and Figures S4D–S4F). Collectively, these experiments demonstrate that BRD4 enhances reprogramming (mouse and human) through CDK9. BRD4 Stimulates Productive Elongation at Pluripotency Genes in Reprogramming and ESCs To demonstrate that BRD4 facilitates the activation of the pluripotency network in the late phase of reprogramming, we performed qPCR of MEFs reprogrammed with OSKM and BRD4-CTD at day 9. As expected, BRD4-CTD induced a substantial increase in the expression of pluripotency genes belonging to the second transcriptional wave of reprogramming (Figure 5A), but there was no significant change in epithelial or mesenchymal genes at day 5 (Figure 5B). Next, we studied whether the effect of BRD4 on pluripotency genes is due to stimulation of transcriptional elongation. For this purpose, we selected eight candidates induced in the second wave of reprogramming: Nanog, Oct4, Sall4, Sox2, Dppa2, Rex1, Esrrb, and Utf1. ChIP-qPCR along the promoter and coding region of these genes showed that BRD4-CTD increases Pol II binding to the gene body compared to the control, indicating more productive elongation (Figure 5C and Figure S5A). Similar results were observed performing ChIP for Pol II phosphorylated on serine 2 (Ser2P), a posttranslational modification triggered by CDK9 that specifically identifies elongating Pol II (Zhou and Yik, 2006) (Figure 5C and Figure S5A). If our hypothesis that BRD4 is a major regulator of pluripotency genes during reprogramming were true, one might also expect that reducing BRD4 expression should blunt pluripotency genes in ESCs. However, a previous genome-wide siRNA screening in ESCs showed cell death but not differentiation upon Brd4 knockdown (Fazzio et al., 2008). To clarify this issue, we reduced Brd4 expression in ESCs with two independent lentiviral shRNA vectors (Figure S5B). This caused quick (within 4 days) acquisition of a differentiated morphology and concomitant loss of AP staining and OCT4-GFP signal, which was associated with reduced expression of pluripotency genes and enhanced expression of lineage-specific genes (Figures 5D and 5E and Figure S5C). In addition, we overexpressed BRD4-CTD in ESCs before letting them differentiate by withdrawing LIF. BRD4-CTD had a moderate effect in both sustaining pluripotency genes and preventing the increase of lineage-specific genes, but it could not prevent

the quick (within 4 days) loss of ESC morphology and AP staining (Figures S5D and S5E). Finally, to further demonstrate that BRD4 regulates pluripotency genes, we analyzed a BRD4 ChIPsequencing data set from ESCs. This showed strong binding of BRD4 at the promoter regions of multiple pluripotency genes including Nanog and Oct4 (Figure 5F). Hence, BRD4 regulates pluripotency gene expression in reprogramming and in ESCs. Differences between our work and the study by Fazzio et al. (2008) may rely on the extent of Brd4 knockdown, which is likely stronger with the siRNA, and may point to an additional function of BRD4 in regulating the ESC cell cycle. Inhibition of P-TEFb Activity by HEXIM1 Impairs the Reactivation of Pluripotency Genes in Reprogramming Because the reprogramming booster BRD4 enhances P-TEFb activity by displacing HEXIM1 bound to CDK9 (Zhou and Yik, 2006), we postulated that HEXIM1 is a roadblock to reprogramming. To evaluate this, we reprogrammed MEFs with OSKM and a retroviral vector producing HEXIM1. As expected, this reduced the interaction between endogenous BRD4 and CDK9, as assessed by immunoprecipitation of CDK9 at day 9 of reprogramming (Figure 6A). HEXIM1 overexpression also reduced cell proliferation (Figure S6A), in agreement with previous work in other contexts (Hong et al., 2012). Nevertheless, a significant number of AP+ colonies still formed, though these were mostly GFP (Figures 6B and 6C). A similar effect was observed with HEXIM1 overexpression in secondary MEFs (Figure S6B). We then tested the effect on OSKM reprogramming of two independent retroviral shRNA vectors for Hexim1. These shRNAs knocked down Hexim1 efficiently and substantially enhanced the interaction between endogenous BRD4 and CDK9 (Figure 6D). They also increased the formation of GFP+ colony formation by 3- to 4-fold, which was not associated with an obvious change in the number of AP+ colonies (Figures 6E and 6F). Similar results were obtained knocking down Hexim1 in secondary MEFs (Figure S6C). Moreover, we reprogrammed MEFs with OSKM and Hexim1 shRNA in optimized culture conditions (Chen et al., 2010). This achieved over 71% GFP+ cells assessed by FACS (Figures 6G and 6H). We confirmed that the iPSCs produced by knocking down Hexim1 had integrated the shRNA vectors into their genome and were fully pluripotent (Figures S6D–S6H). We also employed qPCR of MEFs reprogrammed with OSKM in standard medium at day 9 to show that HEXIM1 overexpression prevents the activation of pluripotency genes while its knockdown enhances it (Figures 6I and 6J). In addition, we studied whether Hexim1 knocking down with a lentiviral shRNA vector maintains pluripotency gene expression in ESCs deprived of LIF by sustaining CDK9 activation through BRD4.

Figure 5. BRD4 Regulates Transcriptional Pause Release at Pluripotency Genes (A and B) qPCR for the indicated genes using RNA lysates from MEFs reprogrammed with OSKM and empty vector or BRD4-CTD at day 9 (A) and day 5 (B). Values are referred to untransduced MEFs. (C) Upper: schematic representation of the Nanog, Oct4, Sall4, and Sox2 loci. The amplified regions are shown below. Lower: ChIP-qPCR of Pol II (total and Ser2P) binding at these loci in MEFs reprogrammed with OSKM and empty vector or BRD4-CTD. (D) AP staining of ESCs transduced with shRNA control or two shRNAs for Brd4. Scale bar represents 50 mm. (E) qPCR for the indicated pluripotency genes using lysates from ESCs transduced with shRNA control or two shRNAs for Brd4. (F) BRD4 occupancy on Nanog and Oct4 from the reanalysis of a previous data set (GSE36561). The data in (A)–(C) and (E) are reported as mean values ± SD with the indicated significance (*p < 0.05, **p < 0.01, ***p < 0.005). See also Figure S5.




B

OSKM 1

OSKM AP

GFP

EX IM H

1 trl C

EX IM

trl

H

C

Brightfield Ctrl

HEXIM1 IB

C

OSKM


A

BRD4 HEXIM1

CDK9 IP: CDK9

60 40 20 0

H

C EX trl IM 1

Input

***

80

D

F

E

OSKM

AP OSKM

HEXIM1

shHex #1 OSKM

BRD4 CDK9

shHex #2 IP: CDK9

30 shHex 0

J

Ctrl OSKM

OSKM

*

**

50

rrb Es

8

0

n2

rrb

8 n2 Li

og an

do

-O ct En 4 do -S ox 2

0

**

**

100

Li

10

150

og

20

200

-O ct En 4 do -S ox 2

**

**

250

an

**

30

***

300

N

40


OSKM + shHex

***

***

En

71.8%

50

N

shHex


OSKM + HEXIM1

GFP

GFP

shCtrl

60

Ctrl

OSKM, iSF1

FSC

Brightfield

do

I

22.6%

**

En

H

shCtrl

OSKM, iSF1

s sh hC H trl ex sh # H 1 ex #2

Input

**

90

Es

IB


sh C tr sh l H ex #1 sh H ex #2 sh C trl sh H ex sh #1 H ex #2

shCtrl

G

Figure 6. HEXIM1 Reduces P-TEFb Activity and Blocks Reprogramming (A) Immunoprecipitation of CDK9 using lysates from MEFs reprogrammed with OSKM and empty vector or HEXIM1 at day 9. Membranes were immunoblotted for HEXIM1, BRD4, and CDK9. (B) Phase contrast (left), fluorescence photographs (middle), and AP-stained wells (right) of MEFs reprogrammed with OSKM and HEXIM1 at day 10. Scale bar represents 50 mm. (C) Number of GFP+ colonies at day 13 in MEFs reprogrammed as in (B). Average colony numbers of three independent experiments are shown. (D) Immunoprecipitation of CDK9 using lysates from MEFs reprogrammed with OSKM and shRNA control or two shRNAs for Hexim1 (shHex) at day 9. Membranes were immunoblotted for HEXIM1, BRD4, and CDK9. (E) AP-stained wells of MEFs reprogrammed with OSKM and shRNA control or two shRNAs for Hexim1 at day 10. (F) Number of GFP+ colonies at day 11 in MEFs reprogrammed as in (E). Average colony numbers of two independent experiments are shown. (G) Phase contrast (left) and fluorescence photographs (right) of MEFs reprogrammed with OSKM and shRNA control or shRNA for Hexim1 in iSF1 medium at day 14. Scale bar represents 50 mm. (H) OCT4-GFP+ cells analyzed by FACS in MEFs reprogrammed as in (G) at day 14. (I and J) qPCR for the indicated genes using RNA lysates from MEFs reprogrammed with OSKM and empty vector or HEXIM1 (I) and shRNA control or shRNA for Hexim1 (J) at day 9. Values are referred to untransduced MEFs. The data in (C), (F), (I), and (J) are reported as mean values ± SD with the indicated significance (*p < 0.05, **p < 0.01, ***p < 0.005). See also Figure S6.

There was only a mild change in pluripotency or lineage-specific genes in ESCs transduced with Hexim1 shRNA (Figure S6I). Therefore, HEXIM1 constitutes a barrier for the reactivation of pluripotency genes in the late phase of reprogramming but does not seem to control the dismantling of the pluripotency network in the early stages of ESC differentiation. 10 Cell Stem Cell 15, 1–15, November 6, 2014 ª2014 Elsevier Inc.

KLF4 Recruits CDK9 to Pluripotency Gene Promoters in Reprogramming and ESCs Transcription factors help recruit CDK9 to specific loci (Romano and Giordano, 2008). We thus predicted that the reprogramming factors support CDK9 binding at pluripotency genes in the late phase of reprogramming and in ESCs. In this regard, c-MYC is



known to recruit CDK9 to a subset of gene promoters (mostly related to cell cycle and metabolism) in ESCs (Rahl et al., 2010). Yet, exogenous c-MYC did not seem a good candidate because it is necessary at the beginning of reprogramming but dispensable afterward (Polo et al., 2012; Sridharan et al., 2009). To confirm this, we reprogrammed MEFs with retroviruses producing the three other Yamanaka factors (OSK) together with BRD4, BRD4-CTD, or Hexim1 shRNA. All three approaches significantly enhanced the number of GFP+ colonies (Figures 7A and 7B). Then, we analyzed the role of endogenous c-MYC in OSK reprogramming using a c-MYC inhibitor (10058-F4) (Rahl et al., 2010), which blunted GFP+ colony formation if added in the early or late phase of reprogramming (Figure S7A). This effect was associated with reduced proliferation and increased apoptosis (Figures S7B and S7C), differing significantly from the mode of action shown above for CDK9 and BRD4 inhibition. We also analyzed the binding of c-MYC on genes belonging to the second wave of reprogramming using a ChIP-sequencing data set from ESCs (Chen et al., 2008). This showed that fewer than 20% of the genes in the second wave are bound by c-MYC in ESCs, among which core pluripotency factors such as Nanog and Oct4 are not included (Figure S7D). Although these results don’t formally exclude the possibility that endogenous c-MYC participates in CDK9 recruitment to pluripotency loci, they strongly suggest that other mechanisms coexist. To study whether CDK9 interacts with any of the other three reprogramming factors, we transfected OCT4, SOX2, KLF4, and c-MYC (as a positive control) individually in HEK293T cells. Exogenous c-MYC and KLF4, but not OCT4 and SOX2, could pull down endogenous CDK9 in immunoprecipitation assays (Figure 7C and Figure S7E). We verified the interaction between KLF4 and endogenous CDK9 in MEFs reprogrammed with OSKM at day 9 (Figure 7D). In addition, we compared previous ChIP-sequencing data sets for CDK9 and 13 ESC transcription factors (among them OCT4, SOX2, KLF4, and c-MYC) or chromatin regulators (Chen et al., 2008; Whyte et al., 2013). CDK9 binding genome-wide showed prominent correlation with KLF4 binding (and also ESRRB and TFCP2L1) but weaker correlation with the core pluripotency module containing OCT4 and SOX2 (in green) or the MYC module containing c-MYC and n-MYC (in blue) (Figure 7E). Accordingly, the top 1,000 KLF4-bound genes in ESCs had significantly higher CDK9 ChIP-sequencing signals than the bottom 1,000 (Figure S7F), and the same was observed when the reverse comparison was performed (Figure S7G). Likewise, we analyzed the average enrichment of CDK9, KLF4, and c-MYC around promoter regions genome-wide in ESCs, which showed higher levels of co-occupancy by CDK9/KLF4 than CDK9/c-MYC (Figure 7F). We also observed that 70% of the genes upregulated in the second wave of reprogramming are KLF4 targets in ESCs (Figure S7H) and CDK9 also binds to many of them (Figure S7I). Consequently, differences in the promoter enrichment between CDK9/KLF4 and CDK9/c-MYC became more obvious when only second wave genes were used for the analysis (Figure 7G), which is illustrated by CDK9 and KLF4 binding (but not c-MYC) at Nanog and Oct4 promoters (Figure 7H and Figure S7J). We further confirmed that KLF4 and CDK9 bind together at pluripotency genes using sequential ChIP (first for KLF4 and then for CDK9) in ESCs, MEFs, and two time points during reprogramming (Figures 7I and 7J and Figure S7K).

To demonstrate that CDK9 binding and pause release at pluripotency genes is KLF4 dependent, we also transduced ESCs with an effective lentiviral shRNA vector for Klf4 (Figure S7L). In agreement with a previous report (Zhang et al., 2010), Klf4 knockdown caused quick reduction of pluripotency gene expression (within 4 days) (Figure S7M). This was associated with reduced loading of CDK9 onto the corresponding promoters (Figure 7K) and of Pol II (total and Ser2P) on the gene body as early as 48 hr after transduction (Figure 7L). Therefore, KLF4 helps stabilize CDK9 binding at pluripotency gene promoters, contributing to promote Pol II pause release at pluripotency loci. DISCUSSION We have demonstrated that Pol II pause release mediated by P-TEFb is a rate-limiting step for the reactivation of pluripotency genes in reprogramming. This process is mediated by the coordinated effects of KLF4 and BRD4 (Figure 7M). We employed bulk populations rather than specific reprogramming intermediates. Yet, the rather high levels of Pol II enrichment and pausing at many pluripotency genes suggest that this may be a general phenomenon occurring as well in cell populations refractory to reprogramming. In this regard, it is worth noting that induction of pause release in standard medium only enhances reprogramming in a proportion of cells, indicating the need to simultaneously remove other roadblocks to achieve high reprogramming efficiency. Our model showing that KLF4 has a more dominant effect than c-MYC in recruiting P-TEFb to pluripotency promoters in the late phase of reprogramming and in ESCs is compatible with the proposed activating role of KLF4 in the second transcriptional wave of reprogramming (Polo et al., 2012). It also correlates with the observation that treating ESCs with flavopiridol reduces the expression of many genes controlled by pause release that are not bound by c-MYC (Rahl et al., 2010). Moreover, a recent report proposed that in oncogenic transformation c-MYC mostly controls Pol II recruitment but not pause release (Sabo` et al., 2014), suggesting that a similar phenomenon participates in reprogramming. From a pluripotency perspective, the new role of KLF4 in controlling pause release may help explain why ESC function is so well synchronized (Min et al., 2011). From a somatic cell perspective, it may provide clues into how KLF4 orchestrates gene activation in normal epithelial homeostasis and disease (e.g., in cancer) (Rowland and Peeper, 2006). Notably, reprogramming can be achieved with cocktails of exogenous factors that don’t include KLF4 (Buganim et al., 2012). In these cases, it is possible that one or more of these alternative transcription factors participate in pluripotency gene pause release. Likewise, it should be considered that induction of endogenous KLF4 in the late phase of reprogramming could substitute for exogenous KLF4. Understanding further how Pol II pausing and pause release at pluripotency promoters are regulated, and specifically how they relate to newly acquired chromatin modifications (Apostolou and Hochedlinger, 2013; Buganim et al., 2013; Koche et al., 2011), may provide new ways to improve reprogramming. In this regard, it would be interesting to study whether Pol II pausing at pluripotency promoters is caused by the coexistence of bivalent Cell Stem Cell 15, 1–15, November 6, 2014 ª2014 Elsevier Inc. 11



s sh hCt H rl e sh x # H 1 ex #2

OCT4 NANOG SOX2 c−MYC n−MYC ZFX E2F1 STAT3 SMAD1 CTCF TFCP2L1 ESRRB KLF4 CDK9

RefSeq genes

F

CDK9 KLF4 ESRRB TFCP2L1 CTCF SMAD1 STAT3 E2F1 ZFX n−MYC c−MYC SOX2 NANOG OCT4

2.5

CDK9 KLF4 c-MYC

2.0 1.5 1.0 0.5 -2000

-1000 0 1000 Distance from TSS bp

0.2

0.6

1

OSKM D5

4 3

*

*

OSKM D12

*

K

2

CDK9 ChIP

0.10

g1 cl

a2 D

Su

pp

og

1 rn

an N

Lr

Ex on Ex 2 on Ex 3 on 4

1 on Ex

3 4

on

N

2 3 4 5 on on on on Ex Ex Ex Ex

1

1

2

3

M

4

Inactive P-TEFb complex

0.4

0.2

1

3

2

TSS

0.20

1

2

*

*

3

4

Pluripotency genes

*

0.15 0.10 0.05 0

×

4

0.25

*

Chromatin remodeling and Pol II recruitment

KLF 4

0

4

*

*

0.1 3

Active P-TEFb complex

*

0.3 *

**

O ct 4

0

Ex

CDK9 shKlf4

0.15

Oct4

1 kb

CDK9 shCtrl

*

0.20

Ser2P shKlf4

*

IgG shKlf4

0.25 0

Pol II shKlf4

**

IgG shCtrl

0.30

1

Nanog

2

K9

c-MYC

MEFs

IgG shKlf4

% input

0.14 0.12 0.10 0.08 Ser2P 0.06 0.04 0.02 0 Primers

1 5

0.05

1

D

KLF4

Ser2P shCtrl

% input

Pol II

1000 bp

-1000 0 Distance from TSS

CDK9

Pol II shCtrl

12 0.30 0.25 0.20 0.15 0.10 0.05 0 Primers

1.0 -2000

Nanog

IgG shCtrl

1 kb

2.0

1 kb

Sequential ChIP (KLF4>CDK9)

N

rn Lr

L

3.0

1 8

an og O ct D 4 pp a Su 2 cl g1

0

CDK9 KLF4 c-MYC

4.0

1

O ct 4

4

2nd wave 5.0

H

Correlation

Fold enrichment over a negative control (Lrrn1)

**

1

Fold enrichment over a negative control (Lrrn1)

*

G

30

J

**

8

C

IB

0

**

12

t

KLF4

IB FLAG

Sequential ChIP (KLF4>CDK9) in ESCs 16

pu In

CDK9

E

I

IP

C trl

0

60 50 40 30 20 10

og

20

IP: FLAG

an

40

IP: FLAG Input

ChIP-sequencing density

60

IP: FLAG Input

% input

80

*

70

Input

trl O C T C 4 trl O C T4 C trl SO X C 2 trl SO X2 C trl KL F4 C trl KL F4

*

*

C

100


120

D

C

OSK

C trl BR B R D D4 4C TD


*

Ig G

B

OSK

ChIP-sequencing density

A

TSS

Paused

Late phase of reprogramming 1

2

Released

**

3

Pluripotent state

4

(legend on next page)




marks (Apostolou and Hochedlinger, 2013), i.e., acquisition of active H3K4me3 without complete removal of inhibitory H3K27me3 during the course of reprogramming. In addition, the role of BRD4, an acetylated histone reader and component of super-enhancers (Loveń et al., 2013; Whyte et al., 2013), in promoting reprogramming is also particularly attractive. Among other possibilities, it is possible that while exogenous transcription factors form super-enhancers at distal regions of pluripotency genes, they also induce the acetylation of nearby histones in order to first recruit BRD4 and then promote pause release through P-TEFb. This process may also be modulated by the interaction of BRD4 and the reprogramming factors with other regulators such as the MBD3/NuRD complex (Rais et al., 2013). In summary, our work provides a mechanism that explains how the pluripotency network is activated in reprogramming and how it is sustained in ESCs. It is tempting to speculate that stimulating pause release through BRD4 also improves other types of transcription-factor-based cell fate conversions (Sindhu et al., 2012). EXPERIMENTAL PROCEDURES Cell Culture and iPSC Generation OG2 MEFs (Esteban et al., 2010) and OG2 secondary MEFs (Carey et al., 2010) were obtained from E13.5 embryos carrying the Oct4-GFP transgene by following standard procedure. Secondary MEFs were derived from mice carrying an inducible OSKM cassette inserted into the Col1a1 locus that had been crossed with OG2 mice. MEFs were maintained in DMEM/high glucose (Hyclone) containing 10% fetal bovine serum (FBS, PAA), GlutaMax (Invitrogen), nonessential amino acids (Invitrogen), and penicillin/streptomycin (Hyclone). HEK293T cells and Plat-E cells were maintained in DMEM/high glucose (Hyclone) containing 10% FBS (PAA). Mouse ESCs and iPSCs were cultured in mouse ESC medium (DMEM with 15% FBS [GIBCO], GlutaMax, nonessential amino acids, sodium pyruvate, b-mercaptoethanol, and 1,000 U/ml LIF [Millipore]) on mitomycin-C-treated MEFs or 0.1% gelatin. Human urinary cells were isolated and cultured as previously described (Zhou et al., 2012). Mouse and human iPSCs were generated as previously described (Zhou et al., 2012; Zhuang et al., 2013). Further details are provided in Supplemental Experimental Procedures.

gies). Full-length Brd4 was amplified by PCR from Flag-HA-Brd4 cDNA purchased from Addgene (#14441). The 4.3 kb PCR product was cloned into the pMXs vector. Truncated deletion mutants were generated by performing PCR with different primers and were cloned into pMXs. Brd4-FEE/AAA, Brd4R/P, Brd4-CTD-FEE/AAA, and Brd4-CTD-R/P mutants were generated using QuikChange site-directed mutagenesis kit. Flag-HA-tagged Brd4 mutants for coimmunoprecipitation studies were also purchased from Addgene: Brd4 (#22304), Brd4-D-ET (#21938), Brd4-D-CTD (#31352), Brd4-ET (#31353), Brd4-CTD (#31354), and Brd4-BD (#32886). Full-length Hexim1 was amplified by PCR from cDNA obtained from MEFs and cloned into pMXs. The lentiviral Brd4-CTD was constructed by subcloning pMXs-Brd4-CTD into pRlenti-based vector. Cdk9 shRNA inserts were cloned into pRetroSuper. Brd4 and Hexim1 shRNA inserts were cloned into both pRetroSuper vector and pLKO.1 vector. Klf4 shRNA inserts were cloned into pLKO.1 vector. shRNA target sequences are listed in Table S2. All plasmids were validated by sequencing. qPCR, Immunofluorescence, Western Blotting, Coimmunoprecipitation, ChIP, and Apoptosis Detection qPCR was performed using SYBR Green (Takara) on an ABI 7500 machine. Assays were run in triplicate and values were normalized on the basis of Actb values. Immunofluorescence microscopy was performed using a Leica TCS SP2 Spectral confocal microscope. Western blotting was performed using ECL or ECL plus (Amersham). Coimmunoprecipitation and ChIP were done using protein A/G Dynabeads (Life Technologies). Apoptosis was detected using Annexin V: PE Apoptosis Detection Kit (BD Biosciences). Primary antibodies and primers are listed in Tables S3 and S4. Further details are provided in Supplemental Experimental Procedures. RNA-Sequencing, ChIP-Sequencing, and Bioinformatics Analysis RNA-sequencing was performed by RiboBio Co., Ltd. ChIP-sequencing was performed as described with small variations. Further experimental details and explanations about the bioinformatics analysis are provided in Table S5 and Supplemental Experimental Procedures. Statistical Analysis The Student’s t test was used throughout the manuscript. ACCESSION NUMBERS RNA-sequencing and ChIP-sequencing data are available in the Gene Expression Omnibus (GEO) database under the accession number GSE59833. SUPPLEMENTAL INFORMATION

Plasmids Full-length Cdk9 was amplified by PCR from cDNAs obtained from MEFs and was cloned into pMXs vector. Cdk9-DN mutant (Fujinaga et al., 1998) was generated using QuikChange site-directed mutagenesis kit (Agilent Technolo-

Supplemental Information for this article includes seven figures, five tables, and Supplemental Experimental Procedures and can be found with this article online at http://dx.doi.org/10.1016/j.stem.2014.09.018.

Figure 7. KLF4 Interacts with CDK9 and Shares Overlapping Targets throughout the Genome (A and B) Number of GFP+ colonies at day 12 in MEFs reprogrammed with OSK and the indicated vectors. Average colony numbers of three (A) or two (B) independent experiments are shown. (C) Immunoprecipitation of the indicated proteins (FLAG-tagged) in HEK293T cells. FLAG-tagged GFP was used as control. Membranes were immunoblotted for CDK9 and FLAG. (D) Coimmunoprecipitation of endogenous CDK9 with KLF4 in MEFs reprogrammed with OSKM at day 10. Antibodies against IgG were used as a control. (E) Hierarchical clustering of the indicated transcription factors and chromatin regulators from the reanalysis of previous data sets (GSE11431 and GSE44286). Colors in the heatmap indicate the correlation between the ChIP-sequencing data sets. (F and G) Distribution of CDK9, KLF4, and c-MYC ChIP-sequencing tags relative to the transcription start site (TSS) of RefSeq genes (F) or second wave genes (G) obtained from the reanalysis of previous data sets (GSE11431 and GSE44286). (H) CDK9, KLF4, and c-MYC occupancy on Nanog obtained from the reanalysis of previous data sets (GSE11431 and GSE44286). (I and J) Sequential ChIP of KLF4 and then CDK9 at the indicated gene promoters in ESCs (I) and MEFs reprogrammed with OSKM at two different time points (J). Untransduced MEFs were used as a control in (J). (K) ChIP-qPCR for CDK9 on the Nanog and Oct4 promoters 48 hr after ESCs were transduced with shRNA control or shRNA for Klf4 (shKlf4). (L) Upper: schematic representation of the Nanog and Oct4 loci and the regions amplified in the ChIP-qPCR. Lower: ChIP-qPCR of Pol II (total and Ser2P) binding to these loci 48 hr after transducing ESCs with shRNA control or shRNA for Klf4. (M) Schematic representation of the main message of our work. The data in (A), (B), and (I)–(L) are reported as mean values ± SD with the indicated significance (*p < 0.05, **p < 0.01, ***p < 0.005). See also Figure S7.




AUTHOR CONTRIBUTIONS L.L., Y.X., and M.A.E. conceived the idea. L.L. and Y.X. conducted most of the experiments. M.H. performed all the bioinformatics analysis. L.L., Y.X., M.H., and M.A.E. analyzed the data. All other authors contributed to the work. M.A.E. supervised the entire study, provided financial support, and wrote the manuscript. ACKNOWLEDGMENTS We thank all members of the Esteban lab for their support and Jiekai Chen, Hong Song, Yongqiang Chen, Tianran Peng, Jing Liu, Keyu Lai, Yingying Li, Xiujuan Cai, Rongping Luo, and Shijuan Huang for technical assistance. This work was supported by grants to M.A.E. from the Ministry of Science and Technology of China 973 program (2011CB965201), the Strategic Priority Research Program of the Chinese Academy of Sciences (XDA01020106), the National Science Foundation of China (31371513), the Bureau of Science, Technology, and Information of Guanghzou Municipality (2012J5100040), and the International Science and Technology Cooperation Program of China (2013DFE33080). H.W. is funded by Research Grants Council (RGC) of the Hong Kong Special Administrative Region, China (476310). H.Y. is funded by the National Science Foundation of China (31271391). H.T. is funded by Theme Based Research Scheme (T12-705/11) and Innovation and Technology Support Programme (Tier 3) (ITS/303/12). Received: February 18, 2014 Revised: August 7, 2014 Accepted: September 26, 2014 Published: October 9, 2014 REFERENCES Apostolou, E., and Hochedlinger, K. (2013). Chromatin dynamics during cellular reprogramming. Nature 502, 462–471. Belkina, A.C., and Denis, G.V. (2012). BET domain co-regulators in obesity, inflammation and cancer. Nat. Rev. Cancer 12, 465–477. Bisgrove, D.A., Mahmoudi, T., Henklein, P., and Verdin, E. (2007). Conserved P-TEFb-interacting domain of BRD4 inhibits HIV transcription. Proc. Natl. Acad. Sci. USA 104, 13690–13695. Buganim, Y., Faddah, D.A., Cheng, A.W., Itskovich, E., Markoulaki, S., Ganz, K., Klemm, S.L., van Oudenaarden, A., and Jaenisch, R. (2012). Single-cell expression analyses during cellular reprogramming reveal an early stochastic and a late hierarchic phase. Cell 150, 1209–1222. Buganim, Y., Faddah, D.A., and Jaenisch, R. (2013). Mechanisms and models of somatic cell reprogramming. Nat. Rev. Genet. 14, 427–439. Carey, B.W., Markoulaki, S., Beard, C., Hanna, J., and Jaenisch, R. (2010). Single-gene transgenic mouse strains for reprogramming adult somatic cells. Nat. Methods 7, 56–59.

He, N., Pezda, A.C., and Zhou, Q. (2006). Modulation of a P-TEFb functional equilibrium for the global control of cell growth and differentiation. Mol. Cell. Biol. 26, 7068–7076. Hong, P., Chen, K., Huang, B., Liu, M., Cui, M., Rozenberg, I., Chaqour, B., Pan, X., Barton, E.R., Jiang, X.C., and Siddiqui, M.A. (2012). HEXIM1 controls satellite cell expansion after injury to regulate skeletal muscle regeneration. J. Clin. Invest. 122, 3873–3887. Koche, R.P., Smith, Z.D., Adli, M., Gu, H., Ku, M., Gnirke, A., Bernstein, B.E., and Meissner, A. (2011). Reprogramming factor expression initiates widespread targeted chromatin remodeling. Cell Stem Cell 8, 96–105. Levine, M. (2011). Paused RNA polymerase II as a developmental checkpoint. Cell 145, 502–511. Loveń, J., Hoke, H.A., Lin, C.Y., Lau, A., Orlando, D.A., Vakoc, C.R., Bradner, J.E., Lee, T.I., and Young, R.A. (2013). Selective inhibition of tumor oncogenes by disruption of super-enhancers. Cell 153, 320–334. Min, I.M., Waterfall, J.J., Core, L.J., Munroe, R.J., Schimenti, J., and Lis, J.T. (2011). Regulating RNA polymerase pausing and transcription elongation in embryonic stem cells. Genes Dev. 25, 742–754. Pijnappel, W.W., Esch, D., Baltissen, M.P., Wu, G., Mischerikow, N., Bergsma, A.J., van der Wal, E., Han, D.W., Bruch, Hv., Moritz, S., et al. (2013). A central role for TFIID in the pluripotent transcription circuitry. Nature 495, 516–519. Polo, J.M., Anderssen, E., Walsh, R.M., Schwarz, B.A., Nefzger, C.M., Lim, S.M., Borkent, M., Apostolou, E., Alaei, S., Cloutier, J., et al. (2012). A molecular roadmap of reprogramming somatic cells into iPS cells. Cell 151, 1617– 1632. Rahl, P.B., Lin, C.Y., Seila, A.C., Flynn, R.A., McCuine, S., Burge, C.B., Sharp, P.A., and Young, R.A. (2010). c-Myc regulates transcriptional pause release. Cell 141, 432–445. Rahman, S., Sowa, M.E., Ottinger, M., Smith, J.A., Shi, Y., Harper, J.W., and Howley, P.M. (2011). The Brd4 extraterminal domain confers transcription activation independent of pTEFb by recruiting multiple proteins, including NSD3. Mol. Cell. Biol. 31, 2641–2652. Rais, Y., Zviran, A., Geula, S., Gafni, O., Chomsky, E., Viukov, S., Mansour, A.A., Caspi, I., Krupalnik, V., Zerbib, M., et al. (2013). Deterministic direct reprogramming of somatic cells to pluripotency. Nature 502, 65–70. Romano, G., and Giordano, A. (2008). Role of the cyclin-dependent kinase 9-related pathway in mammalian gene expression and human diseases. Cell Cycle 7, 3664–3668. Rowland, B.D., and Peeper, D.S. (2006). KLF4, p21 and context-dependent opposing forces in cancer. Nat. Rev. Cancer 6, 11–23. Sabo`, A., Kress, T.R., Pelizzola, M., de Pretis, S., Gorski, M.M., Tesi, A., Morelli, M.J., Bora, P., Doni, M., Verrecchia, A., et al. (2014). Selective transcriptional regulation by Myc in cellular growth control and lymphomagenesis. Nature 511, 488–492. Saha, K., and Jaenisch, R. (2009). Technical challenges in using human induced pluripotent stem cells to model disease. Cell Stem Cell 5, 584–595.

Chen, X., Xu, H., Yuan, P., Fang, F., Huss, M., Vega, V.B., Wong, E., Orlov, Y.L., Zhang, W., Jiang, J., et al. (2008). Integration of external signaling pathways with the core transcriptional network in embryonic stem cells. Cell 133, 1106–1117.

Samavarchi-Tehrani, P., Golipour, A., David, L., Sung, H.K., Beyer, T.A., Datti, 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 cell reprogramming. Cell Stem Cell 7, 64–77.

Chen, J., Liu, J., Han, Q., Qin, D., Xu, J., Chen, Y., Yang, J., Song, H., Yang, D., Peng, M., et al. (2010). Towards an optimized culture medium for the generation of mouse induced pluripotent stem cells. J. Biol. Chem. 285, 31066– 31072.

Sindhu, C., Samavarchi-Tehrani, P., and Meissner, A. (2012). Transcription factor-mediated epigenetic reprogramming. J. Biol. Chem. 287, 30922–30931.

Esteban, M.A., Wang, T., Qin, B., Yang, J., Qin, D., Cai, J., Li, W., Weng, Z., Chen, J., Ni, S., et al. (2010). Vitamin C enhances the generation of mouse and human induced pluripotent stem cells. Cell Stem Cell 6, 71–79.

Sridharan, R., Tchieu, J., Mason, M.J., Yachechko, R., Kuoy, E., Horvath, S., Zhou, Q., and Plath, K. (2009). Role of the murine reprogramming factors in the induction of pluripotency. Cell 136, 364–377.

Fazzio, T.G., Huff, J.T., and Panning, B. (2008). An RNAi screen of chromatin proteins identifies Tip60-p400 as a regulator of embryonic stem cell identity. Cell 134, 162–174.

Takahashi, K., and Yamanaka, S. (2006). Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell 126, 663–676.

Fujinaga, K., Cujec, T.P., Peng, J., Garriga, J., Price, D.H., Granã, X., and Peterlin, B.M. (1998). The ability of positive transcription elongation factor B to transactivate human immunodeficiency virus transcription depends on a functional kinase domain, cyclin T1, and Tat. J. Virol. 72, 7154–7159.

Whyte, W.A., Orlando, D.A., Hnisz, D., Abraham, B.J., Lin, C.Y., Kagey, M.H., Rahl, P.B., Lee, T.I., and Young, R.A. (2013). Master transcription factors and mediator establish super-enhancers at key cell identity genes. Cell 153, 307–319.


Smith, E., and Shilatifard, A. (2013). Transcriptional elongation checkpoint control in development and disease. Genes Dev. 27, 1079–1088.



Zeitlinger, J., Stark, A., Kellis, M., Hong, J.-W., Nechaev, S., Adelman, K., Levine, M., and Young, R.A. (2007). RNA polymerase stalling at developmental control genes in the Drosophila melanogaster embryo. Nat. Genet. 39, 1512– 1516. Zhang, P., Andrianakos, R., Yang, Y., Liu, C., and Lu, W. (2010). Kruppel-like factor 4 (Klf4) prevents embryonic stem (ES) cell differentiation by regulating Nanog gene expression. J. Biol. Chem. 285, 9180–9189. Zhou, Q., and Yik, J.H. (2006). The Yin and Yang of P-TEFb regulation: implications for human immunodeficiency virus gene expression and global

control of cell growth and differentiation. Microbiol. Mol. Biol. Rev. 70, 646–659. Zhou, T., Benda, C., Dunzinger, S., Huang, Y., Ho, J.C., Yang, J., Wang, Y., Zhang, Y., Zhuang, Q., Li, Y., et al. (2012). Generation of human induced pluripotent stem cells from urine samples. Nat. Protoc. 7, 2080–2089. Zhuang, Q., Qing, X., Ying, Y., Wu, H., Benda, C., Lin, J., Huang, Z., Liu, L., Xu, Y., Bao, X., et al. (2013). Class IIa histone deacetylases and myocyte enhancer factor 2 proteins regulate the mesenchymal-to-epithelial transition of somatic cell reprogramming. J. Biol. Chem. 288, 12022–12031.


Brd4 and JMJD6-associated anti-pause enhancers in regulation of transcriptional pause release.

reprogramming for regenerative medicine.

Delivering factors for reprogramming a somatic cell to pluripotency.

YAP controls transcriptional elongation through CKD9 recruitment for proximal pause release: "Hippo-thetical", new therapeutic targets?

Somatic Cell Reprogramming into Cardiovascular Lineages.

p73 is required for appropriate BMP-induced mesenchymal-to-epithelial transition during somatic cell reprogramming.

Reprogramming to pluripotency through a somatic stem cell intermediate.

miR-6539 is a novel mediator of somatic cell reprogramming that represses the translation of Dnmt3b.

Promyelocytic Leukemia Protein Is an Essential Regulator of Stem Cell Pluripotency and Somatic Cell Reprogramming.

Differential effects of Akt isoforms on somatic cell reprogramming.

The use of small molecules in somatic-cell reprogramming.

Vitamin C modulates TET1 function during somatic cell reprogramming.

Reprogramming somatic cells to a kidney fate.

Human somatic cell reprogramming: does the egg know best?

Somatic cell reprogramming-free generation of genetically modified pigs.

The piggyBac Transposon as a Platform Technology for Somatic Cell Reprogramming Studies in Mouse.

ERRs Mediate a Metabolic Switch Required for Somatic Cell Reprogramming to Pluripotency.

A pause for reflection.

Transcriptional reprogramming in cellular quiescence.

The reprogramming of transcriptional competence.

Tet and TDG mediate DNA demethylation essential for mesenchymal-to-epithelial transition in somatic cell reprogramming.

To pause for a moment.

Reprogramming of human somatic cells by bacteria.

Cell-free extract from porcine induced pluripotent stem cells can affect porcine somatic cell nuclear reprogramming.