Disclosure of potential conflicts of interest is found at the end of this article.
Author contributions: Y.S.C. and W.H.H.: experimental design, collection and/or assembly data, data analysis and interpretation, manuscript writing; S.W.P.: experimental design, collection and/or assembly data, data analysis and interpretation; S.D.P.: collection and/or assembly data; C.H.H.: collection and/or assembly data; P.C.H.: collection and/or assembly data; L.N.W.: conceptual development, experimental design, data analysis and interpretation, manuscript writing. Y.-S.C. and W.-H.H. contributed equally to this article.
First published online in STEM CELLSEXPRESS February 25, 2011.
Promyelocytic leukemia (Pml) protein is required for Oct4 gene expression and the maintenance of its open chromatin conformation in stem cells. In proliferating stem cells, Pml-nuclear body, along with transcription factors TR2, steroidogenic factor 1 (SF1) and Sp1, and Brg1-dependent chromatin remodeling complex (BRGC), associates with conserved region 1 (CR1) of this promoter to maintain a nucleosome-free region for gene activity. Retinoic acid (RA) rapidly downregulates Pml, resulting in the replacement of BRGC with Brm-containing remodeling complex, disassociation of SF1 and Sp1, retaining of TR2, recruitment of receptor-interaction protein 140, G9a and HP1γ, and sequential insertion of two nucleosomes on CR1 that progressively displays repressive heterochromatin marks. This study demonstrates a functional role for Pml in maintaining a specific open chromatin conformation of the Oct4 promoter region for its constant expression in stem cells; and illustrates the mechanism underlying RA-induced chromatin remodeling of Oct4 gene in differentiating cells, in which Pml plays a critical role. The study also demonstrates a novel mode of chromatin remodeling, which occurs by repositioning and sequentially inserting nucleosomes into a specific region of the gene promoter to compact the chromatin in differentiating cells. STEM Cells 2011;29:660–669
Oct4, or POU-domain class 5 transcription factor 1 (Pou5f1), is critical for the maintenance of ESC and embryonal carcinoma (EC) cells that share similar stem cell properties, as well as induced pluripotent stem cell reprogramming [1–5]. In ESC or EC cell cultures, it is critical not only to have this gene constantly activated but also to tightly regulate its expression level. The gene, Pou5f1 or Oct4, is regulated by multiple regulatory regions including the distal enhancer (DE), proximal enhancer (PE), and proximal promoter (PP) . In stem cells, PP regulates this gene through the activities of multiple activating transcription factors (TFs) including specificity protein 1 (Sp1) , steroidogenic factor 1 (SF1) , and testis receptor 2 (TR2) , of which the binding sites are clustered in the most proximal conserved region 1 (CR1). In retinoic acid (RA)-exposed cells, TR2 is rapidly stimulated for interaction with promyelocytic leukemia (Pml) protein and is subsequently sumoylated, then functions as a repressor for this gene . RA at a physiological concentration (0.1–1 μM) suppresses this gene and triggers cell differentiation in ESC and EC [10, 11]. Changes in histone methylation and acetylation on PP are found in differentiating EC cells . However, neither the chromatin conformation of this gene promoter in stem cells nor its chromatin remodeling process for gene silencing during stem cell differentiation was clear. Further, as multiple TFs can activate this gene, one crucial question concerns how they are coordinated.
Pml proteins, together with other nuclear proteins including Fas death domain-associated protein (Daxx) and nuclear dot-associated Sp100 protein (Sp100), CBP/p300, etc. , form Pml-nuclear bodies (NBs) [13, 14]. Pml-NBs appear near highly acetylated chromatin and can associate with certain nascent RNAs, suggesting a relationship of Pml-NBs with gene activation . Our earlier studies showed RA-stimulated recruitment of TR2 to Pml for its sumoylation and repressing the Oct4 gene in RA-exposed cells ; however, it was unclear if Pml played additional roles in the maintenance or regulation of this gene activity, and if it had any relationship with the chromatin conformation of gene or coordination of its multiple TFs and potential chromatin remodelers. A study of the major histocompatibility complex (MHC) locus indicated that Pml-NBs organize genes located within this locus into a high-order chromatin-loop structure . Interestingly, the Oct4 gene is found within the MHC gene cluster . Other studies suggested that Pml-NBs regulate transcription by recruiting TFs  and participating in chromatin remodeling , but there has been no direct evidence for either functional role of Pml or its mechanism of action in remodeling chromatin.
This study determines the specific chromatin conformation of Oct4 PP in stem cells, and establishes a functional role for Pml in maintaining this active chromatin conformation of gene by recruiting specific TFs and chromatin remodelers. The study further delineates the chromatin remodeling process of this gene in RA-induced cell differentiation process where cells are rapidly depleted of Pml during early RA-induction, and uncovers a specific mode of chromatin remodeling by ordered nucleosome insertion/reposition on Oct4 PP in differentiating cells.
MATERIALS AND METHODS
Cell Culture and Treatment
The P19 embryonal carcinoma (EC) cell line was purchased from ATCC. P19 cells were maintained in α-MEM containing 7.5% calf serum, 2.5% fetal bovine serum, and 1% penicillin streptomycin at 37°C in 5% CO2. CJ7 ESCs were maintained in ES medium (Dulbecco's modified Eagle's medium), supplemented with 17% ESC-qualified fetal bovine serum, 2 mM glutamine, 0.1 mM nonessential amino acids, 6 μM β-mercaptoethanol, 2 mM HEPES, and 1,000 U/ml recombinant leukemia inhibitory factor. Cells were grown on irradiated mouse embryonic feeder cells (MEFs) in 0.1% gelatin-coated plates. For RA treatment, experiments were conducted with 1 μM RA for 72 hours.
Luciferase Reporter Assay
Luciferase reporter assays were performed as previously described . Briefly, P19 cells were cultured in 24-well plates and transfected with 0.3 μg Oct4-reporter plasmid, proper amount of expression plasmids (such as CMV-Pml, CMV-Sp1, CMV-SF1, and CMV-TR2) and 0.1 μg of SV40-LacZ by using Lipofectamine2000 (Invitrogen, Carlsbad, CA, www.invitrogen.com). Twenty-four hours after transfection, cells were lysed, and luciferase and lacZ activities were determined.
Pml-siRNA, TR2-siRNA, Sp1-siRNA, and SF1-siRNA were purchased from Qiagen. siRNAs were introduced into cells by HiPerfect (301704, Qiagen, Hilden, Germany, www1.qiagen.com) for 72 hours and then mRNA or protein was collected for RT-qPCR or Western blot.
Reverse Transcription and Real-Time Polymerase Chain Reaction
Total RNA was extracted from P19 EC cells using Trizol reagent according to the manufacturer's instructions (Invitrogen). RNA was reverse transcribed by using the Omniscript RT kit (205113, Qiagen). For real-time polymerase chain reaction (PCR) analysis, 2X Brilliant II Master Mix (600804, Agilent Technologies, Santa Clara, CA, www.home.agilent.com) was used, and PCR was performed on an MX3000P Stratagene thermocycler. The relative values were normalized to β-actin and presented as ΔΔCt methods . Primer sequences are listed below.
Chromatin Immunoprecipitation Assay
Chromatin immunoprecipitation (ChIP) assays were performed as described , using the following antibodies: Pml (sc-18423, Santa Cruz Biotechnology, Santa Cruz, CA, www.scbt.com), Daxx (07-471, Millipore, Billerica, MA, www.millipore.com), TR2 (sc-9087, Santa Cruz Biotechnology), Sp1 (07-645, Millipore), SF1 (sc-10976X, Santa Cruz Biotechnology), BAF155 (sc-9746X, Santa Cruz Biotechnology), Brg1 (sc-10768X, Santa Cruz Biotechnology), Brm (sc-28710X, Santa Cruz Biotechnology), G9a (3306, Cell Signaling Technology, Beverly, MA, www.cellsignal.com), receptor-interaction protein 140 (RIP140; ab42126, Abcam, Cambridge, U.K., www.abcam.com), H3 (ab1791, Abcam), H3.3 (ab62642, Abcam), H3K9me3 (CS200604, Millipore), H3K27me3 (ab6002-25, Abcam), AcH3 (06-599, Millipore), HP1γ (sc-10213, Santa Cruz Biotechnology), and RNAPII (05-623, Millipore). Primer set 8, used to monitor CR1 in most ChIP assays, was 5′-CCCACAGCTC TGCTCCTCC-3′ (forward) and 5′-AGCCCTGCTGACCCAT CACC-3′ (reverse). All primers were listed in Supporting Information.
Micrococcal Nuclease Nucleosome Mapping and Restriction Enzyme Accessibility Assays
Micrococcal nuclease (MNase) digestion was performed as described . Nuclei isolated from P19 cells were digested with 20 and 80 U of MNase (Worthington, Lakewood, NJ, www.worthington-biochem.com) at 37°C for 5 minutes, followed by proteinase K treatment at 37°C overnight. The purified DNA was subjected to Southern blot analysis. Restriction enzyme accessibility assays were carried out as described . Isolated nuclei from P19 cells were digested with 100 U of XbaI, StuI, HhaI, NcoI, and AvrII (New England Biolabs, Ipswich, MA, www.neb.com) for 30 minutes. The purified genomic DNA was redigested with 100 U of HindIII. The digested fragments were analyzed by Southern blot using 32P-labeled probe 2 (Fig. 5B).
Ligation-mediated PCR (LM-PCR) was performed as described . Nuclei were digested with 20 U of MNase (Worthington) for 15 minutes at 37°C, followed by proteinase K treatment overnight at 37°C. Mononucleosomal DNA (∼150 bp) was then purified from 1.5% agarose gels and 1 μg of the purified DNA was phosphorylated at 5′-termini with T4 polynucleotide kinase and then ligated to the universal linker using T4 DNA ligase. Purified DNA fragments were amplified by PCR with the universal linker oligonucleotide (25 bp) and 32P-labeled Oct4-specific primers.
RA Reduces the Expression of Pml, a Positive Regulator of Oct4 Gene
In RA-induced P19 EC cell differentiation process, Oct4 protein was reduced beginning at 24 hours of RA treatment, whereas Pml protein level was downregulated even before 24 hours (Fig. 1A), revealing that RA downregulated Pml before it suppressed Oct4 expression. We then examined the potential functional role for Pml in the regulation of Oct4 gene expression using siRNA-mediated knockdown approaches. As shown in Figure 1B and 1C (left panel), proliferating P19 stem cells treated with siRNA of Pml (siPml), or RA, expressed significantly lower levels of Oct4 mRNA (∼70% reduction) and its protein (∼90% reduction). To decipher whether this phenomenon holds true for ESCs, we used the same strategy to differentiate and manipulate a mouse ESC line CJ7. As shown in Figure 1C (right panel), Pml was dramatically reduced by siRNA knockdown of Pml and by RA treatment (∼95% and ∼80%, respectively). Similarly, Oct4 level was proportionally reduced by knockdown of Pml and by RA treatment (∼85% and ∼95%, respectively; Fig. 1C). To obtain further functional evidence for Pml to support Oct4 expression, we conducted gain-of-function studies by ectopic expression. As shown in Figure 1D, overexpression of Pml in control cells had no significant effects on Oct4 expression, but was able to rescue, at least partially, Oct4 expression in siPml- (left) and RA-treated (right) cells. It is clear that silencing Pml or RA treatment reduces both Pml and Oct4 levels. But expressing an siRNA-resistant Pml vector effectively rescues Oct4 expression (left). However, expressing the Pml expression vector could only partially rescue downregulation of Oct4 by RA (right). This demonstrates the specific role of Pml in maintaining Oct4 level. The lack of significant effect of overexpressing Pml to further activate Oct4 in the control (scrambled siRNA transfected cells, in Scr panels) cells might be due to saturation of the system with other endogenous components that limit overactivation of Oct4. The merely partial rescue of Oct4 in RA-treated cells by overexpressing Pml could be due to the suppression of other endogenous components affected by RA. Overall, these data show a potential functional role for Pml in the maintenance of Oct4 gene activity in stem cells including embryonic stem cells, and RA downregulates Pml, thereby rapidly, and at least partially, contributing to early suppression of Oct4 gene in differentiating cells. In addition, a similar pattern of downregulation was also observed for other pluripotency-related genes such as Nanog and Sox2 (Supporting Information Fig. S1), implying that Pml may play a general role in regulating the genes maintaining stem cell property. For a technical consideration, we set up mechanistic experiments to start the investigation of this phenomenon using P19 cell culture, in particular, for studying the remodeling process of Oct4 gene chromatin as described in the following.
Pml-NBs Regulate Oct4 Gene Expression by Associating with Its PP
To substantiate the functional role for Pml in activating Oct4 gene, and to determine the gene's regulatory region targeted by Pml, we generated an Oct4 transcriptional reporter containing its promoter and enhancers, and monitored the reporter activity in gain-of-function transfection studies. As shown in Figure 2A, we detected a Pml dose-dependent activation of this reporter. Four conserved regions (named CRs) are present in the Oct4 upstream regulatory region (Fig. 2B, left). We generated a series of Oct4 reporters deleted in various CRs and determined their activities in the presence of Pml. As shown in Figure 2B (right), the most proximal CR, which is the only one located in PP, pCR1, readily and fully activated this reporter in the presence of Pml, indicating that CR1 of PP is sufficient to mediate Pml's effect and probably is the physical target of Pml.
To examine Pml's binding sites on any specific Oct4 regulatory sequences, we conducted ChIP assays to scan the entire regulation region for binding by Pml, as well as key proteins associated with the Pml-NB structures, such as Daxx and CBP/P300. Figure 2C (left) shows that Pml itself most strongly associated with CR1 and CR4 in stem cells (Ctrl), which was apparently reduced in either siPml- or RA-treated cells. Figure 2C (middle) shows the ChIP-PCR result of one Pml-NB marker (Daxx), and the right panel shows the ChIP-PCR result of another Pml-NB associated protein (CBP/P300). Interestingly, in stem cell cultures, both the two Pml-NB markers associated only with the CR1 region, which was drastically reduced in either siPml- or RA-treated cells. These results suggest that for Oct4, on which multiple regulatory sequences can be associated with Pml protein itself, only CR1 (PP) is a physical target of Pml-NBs. Importantly, RA, or siPml, treatment effectively reduces the formation of Pml-NBs on this region, likely due to its rapid downregulation of Pml.
Pml Activates Oct4 Gene Along with TFs TR2, SF1, and Sp1
Studies in the past have shown Oct4 gene expression activated, principally, by TR2, SF1, and Sp1 binding to their cognate binding sites, the hormone response elements and Sp1 site, within CR1. To determine the contribution of each one of these three major TFs, as well as Pml, to the Oct4 gene activity, we conducted both gain-of-function reporter assays and siRNA-mediated knockdown of these three TFs, in combination with Pml. Figure 3A shows that all the three TF and Pml could by itself each activate the Oct4 reporter, and the combination of each individual TF with Pml could more effectively activate this reporter than by each individual TF or Pml alone, suggesting a slightly enhancing role for these TFs in the action of Pml, and probably vice versa. Consistently, as shown in Figure 3B, silencing Pml, or each individual TF, reduced the endogenous Oct4 level, and silencing Pml together with each TF even more effectively suppressed Oct4 expression. Interestingly, Pml silencing alone reproducibly decreased SF1 and TR2 protein levels, suggesting that Pml can also affect SF1 and TR2 protein expression (see “Discussion” section). All together, these gain-of-function and loss-of-function data validate the combinatorial effects of Pml with each individual TF in activating Oct4.
Pml Recruits TFs and Chromatin-Remodeling Complex and Is Required for Maintaining the Open Chromatin Conformation of Oct4 Promoter
The stronger effect of Pml, when combined with each TF, on activating Oct4 gene activity suggested a potential function for Pml in recruiting these TFs, and possibly, chromatin remodelers to the Oct4 gene. In ChIP assays, we indeed detected association of the three TFs and Pml on the CR1 region (Fig. 4A, left six panels and Supporting Information Fig. S2A). Importantly, in siPml-treated cells, none of these TFs were effectively recruited, suggesting Pml's functional role in facilitating the recruitment of these TFs to this critical region of Oct4 promoter. Consistently, Pml, SF1, and Sp1 (except TR2, see following) were also poorly recruited to this gene in RA-treated cells, supporting that these activating TFs could no longer be recruited to Oct4 promoter because of RA-triggered downregulation of Pml. Of notice is the association of TR2 with this promoter even in RA-treated cells but not in siPml-treated cells. This was due to RA-triggered conversion of activating TR2 into a sumoylated TR2 that became a repressor and could still associate with this chromatin . Failure of TR2 association with this promoter in siPml cells was due to the requirement for Pml in mediating TR2 sumoylation . As predicted, the recruitment of Pml complexes with TR2, SF1, and Sp1 on the CR1 region was detected in repeated ChIP (ReChIP) as shown in Figure 4A, middle four panels and Supporting Information Fig. S2B. But the formation of these Pml-TF complexes was dramatically reduced under Pml silencing or RA treatment. To examine whether the recruitment of these factors to the Oct4 promoter held true in ESCs, we also carried out these ChIP experiments in ESCs. We first ruled out the possibility that the MEF feeders might complicate the ChIP experiments by conducting these experiments using the MEF cells alone, which ruled out the concern (Supporting Information Fig. S2A, bottom). Importantly, recruitment of Pml-NB components and these TFs to the Oct4 regulatory region also happens in ESCs (Fig. 4A, right), suggesting that the dynamics of these transcription regulatory molecules to regulate the Oct4 gene are similar between P19 and ESCs.
We then determined the possible chromatin remodeling complexes on the Oct4 gene. Mammalian Brg1/Brm-associated factor (BAF) chromatin remodeling complexes are essential for self-renewal and pluripotency of ESCs . In proliferating ESCs, the BAF complex is mainly composed of Brahma-related gene one (Brg1) and BAF155, and is named Brg1-dependent chromatin remodeling complex (BRGC). During differentiation, Brm is recruited to replace Brg1 in the BAF complex that is named Brm-containing remodeling complex (BRMC) [25, 26]. BRGC is indispensable for Oct4 gene expression and Brg1 deficiency in mice leads to early embryonic lethality [27, 28]. As shown in ChIP assays (Fig. 4B, upper five panels and Supporting Information Fig. S2C), BAF155 was effectively recruited on this region in stem cells, which was only slightly reduced in siPml- and RA-treated cells. Interestingly, Brg1 associated with this region only in stem cells, but Brm associated with this region only in cells treated with siPml or RA. While this may merely be a correlative phenomenon, the mutually exclusive recruitment of Brg1 and Brm would suggest different roles for these two proteins, that is, Brg1 functions in Oct4 gene activation while Brm acts in its repression. The corecruitment of both Brg1 and BAF155 to CR1 in stem cells confirms BRGC's role in Oct4 expression for stem cell proliferation, whereas displacement of Brg1 by Brm in the BAF complex in siPml- or RA-treated cells would imply that BRMC replaces BRGC to repress Oct4 gene when cells lose Pml and undergo differentiation. Recruiting Brg1 only in the presence of Pml is supported by ReChIP as shown in the bottom two panels of Figure 4B and Supporting Information Fig. S2D. As RA rapidly reduced Pml protein level, it should also reduce Pml/Brg1 complex formation. This is confirmed in reciprocal co-immunoprecipitation (Fig. 4C). Further, it is known that Brg1 complex usually contains actin; this has been confirmed here because actin is detected in the Brg1 complex (Fig. 4C).
All together, the results show that, in stem cells, Pml recruits TFs such as TR2, Sp1, SF1, and the activating chromatin remodeler BRGC to Oct4 PP. In differentiating cells (RA-induction), Pml level decreases, the activating TFs and Brg1 then dissociate from this promoter, but repressive TR2 and BAF155 stays on, and Brm is recruited to form BRMC complex to repress this gene. This provides the molecular explanation for Pml's action in maintaining Oct4 gene activity in stem cells.
Loss of Pml Contributes to Chromatin Remodeling and Specific Nucleosome Rearrangement on Oct4 PP
To determine the chromatin conformation of the regulatory region of Oct4 gene in stem cells when this gene is active, we conducted MNase digestion assays using probes detecting DE, PE, and PP. As shown in Figure 5A, clear regular nucleosomal arrays were detected with all three probes in stem cells (Ctrl), siPml-treated (siPml), and RA-treated cells. This demonstrates that the Oct4 regulatory region forms regular nucleosome arrays in both stem and differentiating cells. However, the PI probe detected apparent changes (increased intensity) of mono- and di-nucleosomes in siPml- or RA-treated cells (lower right panel), suggesting that RA treatment, or Pml silencing, has triggered some form of chromatin remodeling without drastically disturbing the nucleosomes on Oct4 PP.
Intensification of mono- and di-nucleosomes signals suggested nucleosome addition/insertion to this region in siPml- or RA-treated cells. We then conducted restriction accessibility (Fig. 5B) assay. It appeared that Pml silencing (lower left) clearly altered restriction sensitivity on StuI, HhaI, and NcoI sites (marked with * signs on the right of the diagnostic fragments) but not on XbaI or AvrII sites, suggesting nucleosome reposition in the siPml-treated cells. RA treatment resulted in a very similar restriction accessibility pattern (lower right), suggesting that these two treatments have elicited very similar chromatin remodeling processes. Therefore, in stem cells where Oct4 gene is active, nucleosomes do assemble on its promoter and regulatory regions; in differentiating cells (such as by RA exposure or silencing Pml), PP maintains the nucleosome array but undergoes chromatin remodeling (such as nucleosome reposition or sliding) so that restriction sensitivity changes on certain regions.
We then conducted nucleosome scanning  to roughly map the positions of nucleosomes on Oct4 PP before and after differentiation. The preliminary scanning results detected two nucleosomes at the two termini of PP (approximately 600 base pairs), three nucleosomes in siPml-treated cells, and four nucleosomes in RA-treated cells. To determine the fine map of nucleosome positions, we carried out LM-PCR as shown in Figure 6. Actual sizes of diagnostic fragments (depicted on the right of each gel) were calculated by subtracting 25 bp (the length of the universal linker ligated to the purified mononucleosomal DNA and used as a reverse primer in PCR) from the lengths shown on these gels. LM-PCR using F1 amplified 75-, 53-, and 45-bp fragments from stem, siPml-treated, and RA-treated cells, respectively. Accordingly, the 3′-border of the 5′ terminal nucleosome in the three types of cells is assigned to −366 (stem cells), −388 (siPml treated), and −396 (RA treated) positions relative to transcription initiation site (TIS). This is confirmed by LM-PCR using F2 primer, which generated 42-bp (stem cells), 20-bp (siPml-treated), and undetectable (RA-treated, due to the small size, 12 bp, of this fragment, which migrated out of the gel) fragments. Therefore, the 3′ border of this 5′-terminal nucleosome has shifted upward (toward the 5′ direction) for 22 nucleotides in Pml-silenced cells, and 30 nucleotides in RA-treated cells. This is consistent with the inaccessibility of XbaI site in all three groups (Fig. 5B) because this site remains covered by the sliding nucleosome even when cells are differentiating.
LM-PCR using R1 primer mapped the 3′-terminal nucleosome covering TIS. Its 5′-border moved from −80 to −32 and −22 in cells treated with siPml and RA, respectively. The NcoI site initially was near the border of this nucleosome, but was then entirely covered by the downward (toward 3′ direction) sliding nucleosome in cells treated with siPml or RA. This is consistent with the result of Figure 5B showing much reduced accessibility of NcoI site in siPml- or RA-treated cells. LM-PCR using F3 or F4 primers generated no fragments in stem cells, supporting the absence of nucleosome between two terminal nucleosomes of PP in stem cells. Very interestingly, in RA-treated cells, F3 and F4 both generated specific fragments; but in siPml-treated cells, only F4 generated a specific fragment, confirming addition of one nucleosome (3′ border at -82) in siPml-treated cells and two nucleosomes (3′ border at -247 and -55, respectively) in RA-treated cells. This is consistent with the results of Figure 5B showing accessibility of StuI and HhaI sites only in stem cells. Therefore, the initially nucleosome-free middle region (−366 to −80) of PP assembles one new nucleosome in cells depleted of Pml and adds two nucleosomes in cells treated with RA for differentiation. In both situations, the original two terminal nucleosomes are pushed further apart. In differentiating cells, the denser nucleosome array presumably will facilitate chromatin compaction.
Both siPml and RA Each Can Induce Repressive Chromatin Formation on the Oct4 Gene Promoter
Addition of nucleosomes into PP might trigger this region to adopt a more repressive conformation, such as heterochromatin. We then conducted ChIP to monitor repressive chromatin markers including corepressor RIP140 , HP1γ , H3K9me3 and H3K27me3 , and, for a comparison, active chromatin marker AcH3, and RNA PolII  (Fig. 7A and Supporting Information Fig. S3A). In stem cells, AcH3 and RNA PolII were both clearly detected on this region, consistent with the expression of this gene. Pml silencing and RA treatment drastically reduced AcH3 mark and RNA PolII recruitment on this region. In both siPml- and RA-treated cells, all the repressive markers monitored, including HP1γ, H3K9me3, H3K27me3, and RIP140 were enhanced. As euchromatic histone lysine N-methyltransferase 2 (Ehmt2 or G9a)  can mediate H3K9 methylation, and RIP140 can be involved in heterochromatin formation , we speculated G9a corecruitment with RIP140 to this promoter. This is confirmed in ReChIP (Fig. 7A bottom two panels and Supporting Information Fig. S3B).
This study demonstrates a functional role for Pml-NBs in supporting the expression of Oct4 gene in stem cells including EC and ESCs. The results also provide mechanistic details in maintaining a specific open chromatin conformation of Oct4 PP in P19 EC cells, which is to provide a platform to recruit Sp1, SF1, TR2 and activating chromatin remodeler BRGC to maintain Oct4 gene activity. In proliferating cells, this chromatin segment exists in a partially open and nucleosome-scarce conformation, and, upon RA treatment (which first lowers the Pml level), remodeling occurs through inserting and repositioning nucleosomes and forming heterochromatin. The process involves dissociating Sp1 and SF1, retaining TR2, exchanging Brg1 with Brm, and recruiting other repressive factors such as RIP140 and G9a. This would explain the requirement for Pml in the maintenance of Oct4 gene activity and its open chromatin conformation in stem cells. However, it remains to be determined whether this interesting mechanism can be generalized for all the ESCs.
RA can rapidly lower Pml thereby depleting Pml-NB structures in differentiating cells, which contributes to repressive chromatin remodeling in the early cell differentiation process. A model (Fig. 7B) is proposed that in stem cells, Pml-NBs form on CR1 to provide platforms for recruiting activating TFs and remodeler BRGC to maintain the nucleosome-free region of PP. Upon RA-treatment, Pml is rapidly lost, which destroys Pml-NBs; therefore, SF1, Sp1, and Brg1 cannot be maintained. TR2, initially associated with Pml-NBs, rapidly turns into a sumoylated repressor by RA's effect and can still bind this chromatin to recruit corepressors such as RIP140 and histone deacetylases. Discrepancy on the expression level of BAF155 in RA-treated cells has been reported [26, 34], which may be due to difference in the duration of RA treatment. However, within the time window, we examined here (day 3), BAF155 level is relatively constant and stays on the chromatin. Brm is recruited, so that BRGC is replaced by BRMC. Together with RIP140, G9a and HP1γ are also recruited to occupy CR1. Ultimately, this facilitates the formation of a more compacted chromatin with a denser nucleosome array on CR1.
It is also of interest to explore the specific action of Pml on the CR1 region for Pml-NB formation, as Pml associates strongly with both CR1 and CR4 regions. It is possible that Pml associates with other proteins to facilitate the formation of NB on CR1. For example, SATB1 is a protein that can rearrange chromatin into loops, and Pml can interact with this protein to reorganize the MHC class I gene locus, resulting in high-order chromatin loops . Further, we do not exclude the possibility that Pml forms a complex with other component on CR4 region to exert a different activity. It has been shown that Pml can interact with numerous proteins including Sp1 . Our ChIP-ReChIP and co-immunoprecipitation data (Fig. 4) show that Pml physically associates with TR2 and SF1, as well as Brg1. It would be interesting to investigate whether these proteins all directly interact with Pml for their recruitment.
It is interesting that the effects of RA treatment (rapidly downregulating Pml) and Pml silencing on this gene chromatin remodeling are similar, but not identical (Figs. 5 and 6). This indicates that Pml depletion recapitulates a fraction of RA's overall effects, which include nongenomic and extensive genomic activities. Presumably, early reduction in Pml initiates the first phase of repressive remodeling by inserting a nucleosome and sliding two terminal nucleosomes apart to gradually cover TIS on CR1. Prolonged RA effects would induce other genes to elicit a more complete and permanent repressive remodeling by inserting the second additional nucleosome, which pushes two terminal nucleosomes further apart to completely cover TIS. This region is ultimately compacted and heterochromatinized. With regards to the detailed mechanisms, several possibilities can be tested, such as nucleosome sliding and nucleosome exchange. For instance, H2A/H2B dimer removal and H2A/H2B dimer exchange may be examined. In the future, it would be useful to use tagged histone proteins to investigate these several possibilities.
For Oct4 gene, Pml acts as a positive regulator, which is supported by Pml-NBs association with nascent Oct4 mRNA (Supporting Information Fig. S4). But Pml-NBs can also be involved in transcriptional repression and can colocalize with HP1 . Therefore, the function of Pml may be gene-specific and related to its subnuclear localization and associated proteins. We have shown that TR2 protein is recruited to Pml-NBs initially as an activator, but, through RA's nongenomic activity, TR2 can be sumoylated by associating with Pml and becomes a repressor that remains binded to this promoter in differentiating cells . Our current data reveal another activity of Pml in downregulating SF1 and TR2 (Fig. 3B, middle). Therefore, Pml has multiple functional roles, presumably depending upon its associated proteins.
Brg1 or Brm could be involved in transcriptional activation or repression, depending upon the target gene [25, 28]. For the Oct4 gene, Brg1 is involved in activation in stem cells, whereas Brm is probably involved in chromatin compaction and heterochromatin formation after RA treatment [24, 28]. A recent study has shown that the BAF complex associates with the Oct4 promoter and that BAF155 is required for heterochromatin formation and chromatin compaction on this promoter . As both Brg1 and Brm, along with BAF155 to form BAF complexes, it is predicted that BAF155 would be constantly detected on Oct4 PP. This is consistent with our data (Fig. 4). Therefore, in stem cells, Brg1 associates with Pml, contributing to the recruitment of BRGC by Pml-NBs to Oct4 promoter to maintain a nucleosome-scarce, active chromatin conformation. In differentiating cells, Brm is recruited to replace Brg1 when Pml is reduced, probably contributing to repressive chromatin remodeling. This may also be enhanced by the continuing binding of TR2 repressor, which recruits corepressor RIP140, histone deacetylases, and chromatin modifying enzyme G9a, all involved in heterochromatin formation . The mechanism underlying the recruitment of Brm on Oct4 gene promoter and the insertion of the second nucleosome in RA-induced differentiating cells remains to be determined.
Maintaining a constant level of Oct4 is crucial to healthy proliferation of ESCs or EC cells. In RA-induced cultures, Oct4 gene is rapidly suppressed. This study demonstrates the functional role for Pml protein and Pml NBs in recruiting specific activating transcription factors TR2, SF1, and Sp1, as well as activating chromatin remodeler BRGC to CR1 region of the Oct4 promoter, thereby maintaining a specific open chromatin conformation of this promoter for its constant expression in stem cells. The study also illustrates that rapid downregulation of Pml by RA treatment initiates the repressive chromatin remodeling process, which includes sequentially inserting two nucleosomes into the initially nucleosome-scarce CR1 region, pushing one 3′ nucleosome to completely cover the transcription initiation site, and ultimately rendering heterochromatin formation on this promoter in more differentiated cells. Thus, in RA-induced cell differentiation, chromatin remodeling of the Oct4 gene occurs sequentially, and is initiated by downregulating Pml and condensing its proximal promoter through nucleosome insertion and sliding to cover its transcription initiation site.
We thank Y.-C. Tsui for technical help. This work was supported by National Institutes of Health Grants DK54733, DK054733-09S1, DK060521-07S1, DK60521, DA11190, DA11806 and K02-DA13926; the Philip Morris USA Inc. and Philip Morris International; and the Distinguished McKnight University Professorship to L.-N. W.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflict of interest.