当前位置:首页 >> 研究生入学考试 >>

2011 Recruitment of TIF1r to Chromatin


Molecular Cell

Article
Recruitment of TIF1g to Chromatin via Its PHD Finger-Bromodomain Activates Its Ubiquitin Ligase and Transcriptional Repressor Activities
Eleonora Agricola,1 Rebecca A. Randall,1 Tessa Gaarenstroom,1 Sirio Dupont,2 and Caroline S. Hill1,*
1Laboratory 2Department

of Developmental Signalling, Cancer Research UK London Research Institute, 44 Lincoln’s Inn Fields, London WC2A 3LY, UK of Histology, Microbiology and Medical Biotechnologies, University of Padua School of Medicine, viale Colombo 3, 35131

Padua, Italy *Correspondence: caroline.hill@cancer.org.uk DOI 10.1016/j.molcel.2011.05.020

SUMMARY

加固

停留 时间

The interplay between sequence-speci?c DNAbinding transcription factors, histone-modifying enzymes, and chromatin-remodeling enzymes underpins transcriptional regulation. Although it is known how single domains of chromatin ‘‘readers’’ bind speci?c histone modi?cations, how combinations of histone marks are recognized and decoded is poorly understood. Moreover, the role of histone binding in regulating the enzymatic activity of chromatin readers is not known. Here we focus on the TGF-b superfamily transcriptional repressor TIF1g/ TRIM33/Ectodermin and demonstrate that its PHD ?nger-bromodomain constitutes a multivalent histone-binding module that speci?cally binds histone H3 tails unmethylated at K4 and R2 and acetylated at two key lysines. TIF1g’s ability to ubiquitinate its substrate Smad4 requires its PHD ?nger-bromodomain, as does its transcriptional repressor activity. Most importantly, TIF1g’s E3 ubiquitin ligase activity is induced by histone binding. We propose a model of TIF1g activity in which it dictates the residence time of activated Smad complexes at promoters of TGF-b superfamily target genes.
INTRODUCTION In eukaryotic cells the genome is organized into chromatin, which both solves the problem of packaging DNA and plays a central role in regulating all DNA-associated processes. Epigenetic modi?cations of the histone tails and recruitment of chromatin-remodeling complexes are key factors in this regulation (Taverna et al., 2007). Posttranslational modi?cations (PTMs), which occur mainly on the N-terminal tails of the histones, constitute a ‘‘histone code’’ that characterizes the transcriptional state of each gene (Jenuwein and Allis, 2001). The enzymes that add and remove these PTMs are referred to as ‘‘writers’’ and ‘‘erasers,’’ respectively, while proteins such as chromatin-remodeling enzymes and transcriptional regulators

recruited by the PTMs and bound via speci?c domains are known as ‘‘readers’’ (Jenuwein and Allis, 2001). The combination of the activities of writers, erasers, and readers with sequencespeci?c DNA-binding transcription factors and the basal transcription machinery governs transcriptional regulation. A number of domains have been identi?ed that recognize individual PTMs on histones, for example, bromodomains and the plant homeodomain (PHD) zinc ?nger (Ruthenburg et al., 2007; Taverna et al., 2007). Bromodomains recognize acetylated lysines, and proteins may have single or double bromodomains that recognize and directly bind different patterns of acetylated lysines (Taverna et al., 2007). PHD ?ngers have been recently divided into three groups: those that recognize trimethylated K4 of H3, those that speci?cally bind unmodi?ed K4 of H3, and those that do not recognize histones (Chakravarty et al., 2009; Taverna et al., 2006). However, many of the chromatin readers contain combinations of domains that in principle could decode speci?c patterns of PTMs, but how this works is poorly understood. The transcriptional intermediary factor 1 (TIF1) family is a group of proteins with multiple histone-binding domains. In humans, this family comprises four proteins, TIF1a/TRIM24, TIF1b/ TRIM28/KAP1, TIF1g/TRIM33/Ectodermin, and TIF1d/TRIM66, which are characterized by an N-terminal tripartite motif (TRIM) domain consisting of a RING domain, two B boxes and a coiled coil, and a C-terminal PHD ?nger and bromodomain (Meroni and Diez-Roux, 2005). Despite their similar overall structure, these proteins have diverse functional roles in transcriptional regulation. TIF1a is a ligand-dependent nuclear receptor coregulator and more recently has been implicated in regulating p53 stability (Jain and Barton, 2009). TIF1b is a component of the corepressor complex N-CoR1 and the NuRD nucleosome-remodeling complex and promotes heterochromatin formation by interacting with the heterochromatin protein HP1 (Groner et al., 2010). Its PHD ?nger is of the non-histone-binding class (Chakravarty et al., 2009) and has been shown to act as a SUMO ligase, sumoylating the adjacent bromodomain (Ivanov et al., 2007). TIF1d is testis speci?c and interacts with HP1g (Khetchoumian et al., 2004). In contrast, TIF1g does not interact with any HP1 family members or chromatin-remodeling/modifying complexes. Instead it has been shown to function in the TGF-b superfamily pathways (Dupont et al., 2005, 2009; He et al., 2006; Levy
Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc. 85

Molecular Cell
Mechanism of Action of TIF1g

et al., 2007; Morsut et al., 2010), and recently, the zebra?sh ortholog moonshine was shown to regulate transcriptional elongation of erythroid genes (Bai et al., 2010). The TGF-b superfamily ligands are a group of growth and differentiation factors including the TGF-bs, Activin, Nodals, BMPs, and GDFs that signal to the nucleus via complexes of activated Smads comprising C-terminally phosphorylated receptor-regulated Smads (R-Smads) and Smad4 (Schmierer and Hill, 2007). The pathway was traditionally split into two branches in which TGF-b/Activin/Nodal ligands signal via Smad2 and 3, whereas BMP/GDF ligands signal via Smad1, 5, and 8, although recent work has indicated that TGF-b additionally activates Smad1/5 (Wu and Hill, 2009). One study suggested that TIF1g competes with Smad4 to form complexes with activated Smad2/3 and was thus proposed to function as an alternative Smad4 in controlling hematopoietic cell fate (He et al., 2006). Other studies, however, have concluded that TIF1g is a repressor of TGF-b superfamily responses (Dupont et al., 2005, 2009; Morsut et al., 2010; Levy et al., 2007). Knockdown of TIF1g in Xenopus embryos enhances Nodal signaling and hence promotes mesendoderm induction (Dupont et al., 2005). Knockout of TIF1g in mouse results in embryonic lethality resulting from excessive Nodal signaling (Morsut et al., 2010). TIF1g was also revealed as a potent repressor of TGF-b-induced transcription in a highthroughput siRNA screen (Levy et al., 2007). It is thought to inhibit TGF-b/Nodal signaling by monoubiquitinating nuclear Smad4 on K519 and, to a lesser extent, K507, via its N-terminal RING domain, which disrupts active Smad complexes in the nucleus (Dupont et al., 2009). Here we dissect the mechanism of action of TIF1g and reveal the fundamental importance of the tandem PHD ?nger and bromodomain in TIF1g’s transcriptional repressor activity in TGF-b superfamily responses. We demonstrate that TIF1g is a multivalent chromatin reader, which binds a speci?c epigenetic code on histone H3. Moreover, we show that the E3 ubiquitin ligase activity of TIF1g toward Smad4 is regulated by histone binding and propose that it acts as a repressor of TGF-b superfamilyinduced transcription by restricting the residence time of activated Smad complexes on the promoters of target genes. RESULTS TIF1g Is a Repressor of TGF-b Superfamily Transcriptional Responses, and This Requires Its PHD Finger-Bromodomain TIF1g is a potent repressor of transcription downstream of TGF-b superfamily ligands (Dupont et al., 2005, 2009; Levy et al., 2007; Morsut et al., 2010), but it is unclear how speci?c it is for these responses. We therefore tested the effect of TIF1g knockdown on TGF-b- and BMP-induced transcription using stable cell lines containing either an integrated Smad3dependent reporter, CAGA12-Luc (Dennler et al., 1998), or an integrated Smad1-dependent reporter, BRE-Luc (Korchynskyi and ten Dijke, 2002), and compared this with the effect on an integrated reporter driven by serum response elements (SRELuc), which bind serum response factor (SRF) and the cofactors MRTF-A and B in response to serum or cytochalasin D (Posern and Treisman, 2006). Knockdown of TIF1g increased both
86 Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc.

TGF-b- and BMP-induced transcription, having a stronger effect on TGF-b-dependent responses (Figures 1A and 1B). In contrast, knockdown of TIF1g had no signi?cant effect on cytochalasin D-induced transcription via the SRE (Figure 1C). This suggests that TIF1g is not a general transcriptional repressor, but has speci?city for TGF-b superfamily responses. Previous work suggested that the repressive effect of TIF1g on TGF-b-dependent transcriptional responses is primarily mediated through monoubiquitination of Smad4 at K519, which destabilizes activated R-Smad-Smad4 complexes in the nucleus (Dupont et al., 2009). Smads constantly shuttle between the cytoplasm and nucleus in the absence and presence of signaling (Schmierer and Hill, 2005), with dissociation of activated nuclear Smad complexes preceding export of monomeric Smads to the cytoplasm, where they will accumulate if signaling is terminated. If TIF1g promoted disruption of bulk activated nuclear Smad complexes, the lifetime of these complexes after termination of signal would be expected to be prolonged when TIF1g was knocked down. We investigated this in a HaCaT cell line stably expressing a TIF1g shRNA (Figure S1A) and compared the behavior of the Smads in this cell line with that in wild-type HaCaTs. We measured the kinetics of Smad3 phosphorylation and Smad2/3 nuclear accumulation after TGF-b induction and Smad3 dephosphorylation and reaccumulation of Smads 2 and 3 in the cytoplasm after signal termination by addition of the ALK5 inhibitor SB-431542 (Inman et al., 2002). Surprisingly, no differences were seen between the two cell types (Figure S1B). We also investigated the TGF-b induction of PAI-1 (Levy and Hill, 2005) in these two cell lines and measured the decay of PAI-1 mRNA levels after signal termination. Although PAI-1 was induced more strongly in the cell line depleted of TIF1g, there was no difference in the mRNA decay kinetics after signal termination between the two cell lines (Figures S1C and S1D), again arguing against gross stabilization of activated nuclear Smad complexes when TIF1g is depleted. From these results it appeared that TIF1g was unlikely to mediate its repressive effects on TGF-b superfamily-dependent transcription by simply controlling levels of activated nuclear Smad complexes. We hypothesized instead that it might function at the level of chromatin, as it contains a tandem PHD ?nger and bromodomain, both of which are known in other proteins to recognize PTMs on histones (Ruthenburg et al., 2007; Taverna et al., 2007). Indeed, mutation of key amino acids in either the PHD ?nger or the bromodomain abolished the repressive activity of TIF1g on Activin-induced transcription of the CAGA12-Luc reporter in 293T cells (Figure 1D), indicating that both of these domains are essential for repression of TGF-b superfamilyinduced transcription. TIF1g Is Recruited to Chromatin upon TGF-b Stimulation, and This Depends on Smad4 Given the importance of the PHD ?nger and bromodomain in TIF1g’s transcriptional repressive activity, we investigated whether TIF1g was recruited to chromatin at TGF-b-dependent promoters using the PAI-1 gene as a model. Chromatin immunoprecipitations (ChIPs) were performed in MDA-MB-231 cells, which express PAI-1 in response to TGF-b (Figure S2) in a Smad3-dependent manner (Dennler et al., 1998). The gene

Molecular Cell
Mechanism of Action of TIF1g

Figure 1. The Repressive Activity of TIF1g on TGF-b Superfamily Transcriptional Responses Requires Its PHD Finger-Bromodomain
(A and B) MDA-MB-231 cells containing integrated CAGA12-Luc (A) or BRE-Luc (B) and TK-Renilla reporters were transfected with individual siRNA oligos or pools and treated with the indicated ligands for 8 hr. (C) HaCaT-SRE-Luc cells were transfected with siRNAs and induced ± cytochalasin D for 7 hr. NT, nontargeting. (D) 293Ts were transfected with CAGA12-Luc and TK-Renilla along with the indicated pCS2-TIF1g constructs. In all cases, luciferase levels were normalized to Renilla levels. In all cases the data are means and standard deviations for a representative experiment performed in triplicate. Protein overexpression and knockdown were veri?ed by western blot. In (D), wild-type TIF1g protein has a slower mobility due to its FLAG tag.

was analyzed for RNA polymerase II (Pol II) and TIF1g enrichment at the transcription start site (TSS); at the Smad binding region (SBR) of the promoter, which is about 725 bp upstream of the TSS; and in the coding region (CR), approximately 8800 bp downstream of the TSS. In uninduced cells, Pol II density was highest at the TSS compared with the SBR and CR. TGF-b further increased Pol II density as expected at all three sites (Figure 2A). TIF1g was detected at all three sites in the absence of TGF-b stimulation, and its density increased after TGF-b induction, most strongly at the SBR (Figure 2A). The TGF-b-induced enrichment of TIF1g at the SBR and TSS was inhibited when the signal was terminated by incubation with SB431542 for 2 hr, suggesting that it correlated with Smad activation (Figure 2B). TIF1g interacts with Smad4 (Dupont et al., 2005), so we investigated whether Smad4 was required for the TGF-b-dependent

TIF1g chromatin association using a HaCaT cell line expressing a tetracycline-inducible shRNA against Smad4 (HaCaT-TRS4) and the corresponding parental cell line (HaCaT-TR) (Levy and Hill, 2005) (Figure 2C). Both cell lines were treated with tetracycline, then induced or not with TGF-b for 1 hr. The results in the parental cell line mirrored what we observed in MDA-MB-231 cells. However, when Smad4 was knocked down, TGF-b-induced enrichment of Pol II and TIF1g was reduced, suggesting that Smad4 is required for their recruitment upon TGF-b stimulation. TIF1g Does Not Form a Stable Complex on Naked DNA with Activated Smads Although the binding of TIF1g to the PAI-1 promoter upon TGF-b stimulation requires Smad4, this does not necessarily mean that it is directly recruited by Smad4. To investigate this further, we
Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc. 87

Molecular Cell
Mechanism of Action of TIF1g

Figure 2. TIF1g Binding to the TGF-b-Induced PAI-1 Gene Depends on Both Ligand Activation and Smad4
(A) This ?gure shows qPCR of the PAI-1 Smad binding region (SBR), transcription start site (TSS), and coding region (CR) from ChIP assays using RNA Pol II, TIF1g, and Smad4 antibodies. ChIPs were performed on extracts from MDA-MB-231 cells treated ± TGF-b for 1 hr. (B) As in (A) except extracts were additionally obtained from cells treated for 2 hr with SB-431542 after the TGF-b induction. (C) Extracts were prepared from HaCaT-TR and HaCaT-TRS4 cells treated with tetracycline (Tet) overnight and treated ± TGF-b for 1 hr. Smad4 knockdown was veri?ed by western blot. In all cases data correspond to the average of duplicate qPCRs from a representative experiment normalized to input.

examined the ability of TIF1g to interact with the Smads. Endogenous TIF1g interacted with Smad4 in both the presence and absence of TGF-b, but only interacted with Smad2 upon
88 Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc.

TGF-b stimulation, and this was independent of Smad4 (Figure 3A) (He et al., 2006). TIF1g also interacted with phosphorylated Smad3 (PSmad3) in a ligand-inducible manner (Figure 3B).

Molecular Cell
Mechanism of Action of TIF1g

Figure 3. TIF1g-Smad Complex Formation Does Not Occur on Naked DNA
(A and B) Extracts were prepared from treated HaCaT-TR and HaCaT-TRS4 cells (A) or 293Ts (B). Extracts were analyzed by western blot using the indicated antibodies, either directly (Input) or after IP with protein A beads (Beads) or anti-TIF1g antibody-coupled beads (a-TIF1g). Note that in (B), the antibody against phosphorylated Smad3 (PSmad3) also recognizes phosphorylated Smad1 (PSmad1). Only PSmad3 interacts with TIF1g. (C) Nuclear extracts were prepared from HaCaT cells treated ± TGF-b for 1 hr and either analyzed by western blot directly (Input) or after DNA pull-down assay using wild-type (WT) c-Jun SBR oligos or those mutated in the Smad3/Smad4 binding site (Mut).

Although these immunoprecipitations (IPs) con?rm that Smads 2, 3, and 4 interact with TIF1g, they do not reveal whether the activated Smads simultaneously form a complex on DNA and bind TIF1g. We investigated this in a DNA pull-down assay using Smad-binding elements, which bind activated Smad3-Smad4 complexes upon TGF-b stimulation (Inman and Hill, 2002; Levy et al., 2007). As previously reported, Smad4 bound the wildtype Smad binding elements in the absence of signal, and a 1 hr TGF-b stimulation induced an increase in Smad4 binding, along with PSmad3 binding (Levy et al., 2007) (Figure 3C). However, we did not detect any binding of TIF1g in either the absence or presence of TGF-b, suggesting that TIF1g does not form a stable complex with activated Smads on DNA (Figure 3C). Thus, although Smad4 is required for the enrichment of TIF1g at the PAI-1 promoter upon ligand stimulation, its stable recruitment to chromatin must require additional interactions. The PHD Finger-Bromodomain of TIF1g Constitutes a Histone-Binding Reader Module with Speci?city for H3 The most likely mechanism underlying TIF1g’s recruitment to chromatin would involve direct interaction between its tandem PHD ?nger-bromodomain and PTMs on histones. In other proteins, PHD ?ngers bind either unmodi?ed or trimethylated K4 of H3, whereas bromodomains bind acetylated lysines on

histone tails (Chakravarty et al., 2009; Taverna et al., 2007). We used peptide pull-down assays to examine the interaction of TIF1g with different H3 and H4 peptides (Figure 4A). 293T extract was incubated with the immobilized peptides, and the binding of endogenous TIF1g was detected by western blot. TIF1g bound weakly to unmodi?ed H3 peptide and strongly to an H3 peptide acetylated at ?ve lysines (Figure 4B). It failed to bind the H3 peptide containing trimethylated K4 (H3 1–38 K4me3) and did not bind either the unmodi?ed or acetylated H4 peptides (Figure 4B). As a control for H3K4me3 binding, we used the C-terminal PHD ?nger of BPTF, which is a subunit of NURF (Li et al., 2006). This PHD ?nger speci?cally bound to H3 1–38 K4me3 (Figure 4B). To control for binding to the acetylated H4 peptide, we used the bromodomain BD1 of BRDT, which selectively recognizes the H4 tail ` bearing two or more acetylated lysines (Moriniere et al., 2009). This bromodomain speci?cally bound H4 1–34 (4ac) (Figure 4B). Binding of TIF1g to the H3 peptides was entirely dependent on the presence of the PHD ?nger-bromodomain (Figure 4C). The ability of TIF1g to bind an H3 tail containing unmethylated K4 and its failure to bind when K4 is trimethylated suggested that its PHD ?nger is another member of the group I PHD ?ngers that recognize H3K4me0 (Chakravarty et al., 2009). Alignment of the sequence of the TIF1g PHD ?nger with those of other group I PHD ?ngers (AIRE and BHC80) reveals a high degree of similarity. Importantly, residues in AIRE and BHC80 crucial for H3 binding are conserved in the TIF1g PHD ?nger, but absent in the PHD ?nger of the related protein, TIF1b, that does not bind histones (Chakravarty et al., 2009) (Figure S3A). We next investigated the relative importance of the TIF1g PHD ?nger and bromodomain for binding different regions of H3 and to the acetylated lysines. Peptide pull-downs were performed with extract made from 293Ts expressing FLAG-tagged PHD ?nger-bromodomain (PHD/Bromo) or derivatives mutated in either the PHD ?nger or bromodomain (Figure S3B). The wildtype PHD/Bromo bound to H3 1–38 and H3 1–38 5ac, and also to H3 1–21, but not to H3 22–38 (Figure S3B), consistent with the observation that the AIRE PHD ?nger contacts the N-terminal
Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc. 89

Molecular Cell
Mechanism of Action of TIF1g

Figure 4. TIF1g Speci?cally Binds an Acetylated H3 Tail in which R2 and K4 Are Unmethylated
(A) Sequences of histone H3 and H4 tails. Colored residues indicate modi?cations analyzed. (B) Peptide pull-down using differently modi?ed H3 and H4 peptides and extracts made from untransfected 293Ts (upper panel) or 293Ts transfected with FLAGtagged BPTF PHD ?nger (middle panel) or with FLAG-tagged BRDT bromodomain (BD1; lower panel). FLAG-tagged proteins were detected by western blot and binding of endogenous TIF1g using a TIF1g antibody. (C) As in (B) except that extract was prepared from 293Ts stably expressing FLAG-tagged TIF1g DPHD/Bromo. Endogenous TIF1g was detected with the antiTIF1g antibody. (D) Peptide pull-down using extract prepared from untransfected 293Ts and H3 peptides with the indicated mutations. (E) As in (D) except that H3 peptides with single acetylated lysines or a combination of two acetylated lysines were used. (F) As in (D) except that single H3 peptides or combinations of H3 peptides were used. In all cases the lanes marked Input are extract prior to pull-down, and those marked Beads were pulled down with beads alone. Beads conjugated with peptides were analyzed by SDS-PAGE, and peptides were detected by Coomassie (peptide loading). (G) A model indicating how the TIF1g PHD ?nger-bromodomain binds to the H3 tail.

eight amino acids of H3 (Chakravarty et al., 2009). Mutation of the PHD ?nger reduced binding to H3 1–38, H3 1–38 5ac, and H3 1–21 and promoted nonspeci?c binding to H3 22–38, while mutation of the bromodomain abolished binding to the H3 1– 38 5ac, but had little effect on binding to the unmodi?ed peptides (Figure S3B). We conclude that the bromodomain recognizes the acetylated lysines, while the PHD ?nger is required for binding to the N-terminal region of the H3 peptide, as well as for optimal binding by the bromodomain. To de?ne how the TIF1g PHD ?nger binds to H3, we made mutations or modi?cations of key residues in H3 shown to be important for the interaction of the AIRE PHD ?nger with the H3 tail (Chakravarty et al., 2009; Chignola et al., 2009). In the
90 Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc.

context of H3 1–21, mutation of R2, K4, R8, and K9 all inhibited TIF1g binding, as did asymmetric methylation of R2 (Figure 4D), substantiating our hypothesis that TIF1g’s PHD ?nger shares binding speci?city with AIRE. To identify which acetylated lysines are required for binding of the bromodomain, we analyzed endogenous TIF1g binding to a panel of peptides each acetylated at single lysines. TIF1g failed to bind any of the monoacetylated peptides as strongly as it did the 5ac peptide (Figure 4E, left panels), although it did bind a peptide acetylated at K23 and, to a lesser extent, one acetylated at K18 (Figure 4E, left panels). In the protein BRDT, a single bromodomain cooperatively ` binds two sequential acetylated lysines in the H4 tail (Moriniere

Molecular Cell
Mechanism of Action of TIF1g

et al., 2009). To investigate whether this binding mechanism is shared by TIF1g, we performed pull-downs using peptides with two sequential acetylated lysines. TIF1g had a signi?cantly higher af?nity for these diacetylated peptides compared to the monoacetylated ones, and combinations that included acetylation of K18 and K23 bound at least as strongly as the 5ac peptide (Figure 4E; for quantitations, see Figure S3C). TIF1g binds the N-terminal region of the H3 tail via its PHD ?nger, and two sequential acetylated lysines via its bromodomain. This raised the question of whether binding was in cis, i.e., both domains binding the same H3 tail, or in trans, with the PHD ?nger binding one H3 tail and the bromodomain another. Binding of the bromodomain requires binding of the PHD ?nger, as demonstrated by the fact that an H3 peptide containing the bromodomain-interacting region (K18 and K23 acetylated), but truncated in its N terminus (H3 8–38 K18/K23ac) and thus missing the PHD ?nger binding region, failed to bind TIF1g (Figure 4F). We reasoned that if the PHD ?nger-bromodomain bound to H3 tails in trans, we should be able to rescue the binding of the H3 8–38 K18/K23ac peptide by mixing it with the H3 1–38 or H3 1–21 peptides that bind the PHD ?nger. However, no rescue was evident (Figure 4F), leading us to conclude that binding is in cis. These data establish the TIF1g PHD ?nger-bromodomain as a chromatin-binding reader module with speci?city for the H3 tail, whose epigenetic code is de?ned as unmodi?ed K4 and R2 and acetylation of at least two sequential lysines, for example, K18 and K23 (Figure 4G). Binding of TIF1g to Histones via Its PHD FingerBromodomain Regulates Its E3 Ubiquitin Ligase Activity TIF1g was previously shown to monoubiquitinate Smad4 (Dupont et al., 2005, 2009). However, its ability to bind the H3 tail and the fact that it is recruited to chromatin in vivo prompted us to investigate whether it might also ubiquitinate histones. For these experiments we used full-length FLAG-tagged TIF1g puri?ed from 293Ts. Our preliminary in vitro ubiquitination experiments using puri?ed histones did not reveal any histone ubiquitination (data not shown). However, we observed ubiquitination of a protein migrating above the 116 kDa marker, speci?cally in the presence of TIF1g and histones. This ubiquitination was dependent on the dose of TIF1g and on the histones (Figure 5A), and importantly, the ubiquitinated protein had identical migration on SDS-PAGE to TIF1g, suggesting that it was autoubiquitinated TIF1g. We proved this by immunoprecipitating TIF1g after the ubiquitination reaction (Figure 5B). These results strongly suggested that the binding of TIF1g to histones activates its E3 ligase activity. The histones used were puri?ed from chicken erythrocyte nuclei, and identical results were obtained with core histones puri?ed from HeLa cells (data not shown). The histones are a heterogeneous mixture of histones posttranslationally modi?ed at different residues. We show, for example, that they contain H3 acetylated at K14, K18, and K23 (Figure 5C). We reasoned that if TIF1g binding to the modi?ed histones resulted in its activation, we should observe a much weaker activation, or none, if completely unmodi?ed histones were used instead. We thus generated recombinant octamers, puri?ed from E. coli (Luger et al., 1997),

and con?rmed that these histones were indeed unmodi?ed (Figure 5C). These histone octamers did not induce autoubiquitination of TIF1g (Figure 5C), indicating that binding of TIF1g to modi?ed histones induces its activation. To strengthen this conclusion, we puri?ed TIF1g deleted in its PHD ?nger-bromodomain (TIF1g DPHD/Bromo) from the stable 293T cell line and used it in the autoubiquitination reaction. TIF1g DPHD/Bromo has no transcriptional repressive activity (data not shown) and does not interact with the H3 tail (Figure 4C). It was not autoubiquitinated in the presence of puri?ed histones (Figure 5D), con?rming our hypothesis that binding of TIF1g to histones via its PHD ?nger-bromodomain induces its activation. Autoubiquitination of TIF1g was also readily detected in vivo, both in the absence and presence of TGF-b signaling (Figure 5E). We mapped the region of TIF1g that is ubiquitinated using a series of deletion mutants and showed that the sites of ubiquitination lie outside the RING domain and PHD ?nger-bromodomain (Figure S4A). In fact, they reside in the TRIM domain, as a construct comprising just the TRIM domain was readily autoubiquitinated (Figure S4B). In view of our in vitro ubiquitination results and the requirement of the RING domain for ubiquitination activity, we were surprised that TIF1g DRING and TIF1g DPHD/Bromo were ubiquitinated. The most likely explanation would be homo-oligomerization of the TIF1g deletion mutants with endogenous TIF1g via the coiled-coil region of the TRIM domain, as has been observed for the related protein, TIF1b (Peng et al., 2000). Indeed, all of the TIF1g derivatives, with the exception of that deleted in the TRIM domain, interacted with wild-type TIF1g (Figure S4C). Ubiquitination of Smad4 by TIF1g Requires the RING Domain and Interaction of the TIF1g PHD FingerBromodomain with Histones We have demonstrated by autoubiquitination that binding of the TIF1g PHD ?nger-bromodomain activates TIF1g. We next sought evidence that histone binding also activated TIF1g’s ability to bind its primary substrate, Smad4 (Dupont et al., 2009). We could readily demonstrate that puri?ed Gal4-Smad4 was ubiquitinated in vitro by TIF1g only in the presence of puri?ed histones (Figure 6A). If histone engagement activates TIF1g, then in vivo Smad4 ubiquitination by TIF1g must occur on chromatin. If this is the case, we would expect TIF1g deleted in its PHD ?nger-bromodomain to be de?cient in its ability to ubiquitinate Smad4 in vivo. In 293Ts, monoubiquitinated Smad4 is readily detected, which is dependent on TIF1g (Dupont et al., 2009). Full-length TIF1g promoted monoubiquitination of Smad4, and weak diubiquitination of Smad4 was also observed, indicating that TIF1g actively ubiquitinates Smad4 in this assay (Figure 6B). In contrast, neither TIF1g DPHD/Bromo nor TIF1g DRING promoted Smad4 ubiquitination, indicating that they were inactive (Figure 6B) and highlighting the importance of these domains for TIF1g activity. Finally, we addressed how chromatin interaction might activate TIF1g. We noticed that TIF1g DPHD/Bromo interacted much more strongly with Smad4 than did wild-type TIF1g (Figure 6B). This suggested that in the context of full-length TIF1g, the Smad4 interaction region, which has been mapped to the linker between the TRIM domains and PHD ?nger-bromodomain
Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc. 91

Molecular Cell
Mechanism of Action of TIF1g

Figure 5. Histone Binding Induces TIF1g Autoubiquitination Activity
(A) In vitro ubiquitination assay. Recombinant E1, E2, and HA-Ubiquitin (HA-Ub) were incubated with puri?ed core histones and TIF1g puri?ed from FLAG-TIF1g 293Ts. Autoubiquitinated TIF1g was detected with the HA antibody and unmodi?ed TIF1g with TIF1g antibody. A Coomassie-stained gel shows puri?ed TIF1g (left panel). Molecular weight (MW) markers are shown in kDa. (B) Assays were performed as in (A) using 5 ml puri?ed TIF1g, but additionally the samples were IPed using FLAG and TIF1g antibodies prior to western blot. (C) As in (B) but also including incubation with recombinant octamers. The histones used in the assay were western blotted using antibodies recognizing histone H3 acetylated at different lysines as indicated and are also shown stained with Coomassie. Note that the migration of H2B in the puri?ed chicken erythrocyte histones differs slightly from that of the recombinant Xenopus histones, as previously observed (Luger et al., 1997). (D) As in (B) but including incubation with TIF1g DPHD/Bromo puri?ed from FLAG-TIF1g DPHD/Bromo 293Ts. (E) 293Ts were transfected with HA-Ubiquitin and FLAG-TIF1g and incubated overnight ± SB-431542. Extracts were IPed (IP) with FLAG-antibody and blotted as indicated. In (A) and (C), the migration of a 116 kDa marker is shown.

(Dupont et al., 2005), is masked in the inactivated state. Indeed, we could demonstrate that an isolated PHD ?nger-bromodomain interacts with both TIF1g DPHD/Bromo and an isolated TIF1g
92 Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc.

TRIM domain (Figure 6C). Thus, in the inactive state, the TIF1g PHD ?nger-bromodomain participates in an intramolecular interaction with the TRIM domain. Upon recruitment of TIF1g to

Molecular Cell
Mechanism of Action of TIF1g

Figure 6. Smad4 Ubiquitination by TIF1g Depends on Activation of TIF1g by Binding to Modi?ed Histones and on the TIF1g PHD Finger-Bromodomain and the RING Finger
(A) In vitro ubiquitination assay performed as in Figure 5A, but with the inclusion of Gal4-Smad4 as the TIF1g substrate. Autoubiquitinated TIF1g and ubiquitinated Gal4-Smad4 are detected in the HA blot. (B) In vivo ubiquitination assay. Extracts were prepared from 293Ts transfected with Myc-Smad4, FLAG-TIF1g, or mutants thereof ± a plasmid expressing HA-Ubiquitin (HA-Ub). Extracts were IPed (IP) using anti-Myc antibody, and the IPs were western blotted with the indicated antibodies. Inputs and positions of molecular weight markers are shown. A lighter exposure of the Smad4 blots indicated equal loading. (C) The N-terminal TRIM domain and the C-terminal PHDbromodomain interact in solution. 293T cells were transfected with the constructs shown. Extracts were IPed using anti-Myc antibody, and the IPs were western blotted with FLAG antibody. Inputs are shown.

DISCUSSION TIF1g is a potent repressor of TGF-b superfamily signaling pathways. Our demonstration that the tandem PHD ?nger and bromodomain are critical for its repressive activity prompted us to investigate whether TIF1g exerts its effects on chromatin. TIF1g is enriched at the promoter of the PAI-1 gene upon TGF-b stimulation. Although this correlates with activated Smad binding and requires Smad4, we show that the Smads are not suf?cient to recruit TIF1g to chromatin. Instead we have shown that TIF1g is recruited to chromatin through direct interactions with modi?ed histones. Our results reveal that the PHD ?nger-bromodomain of TIF1g speci?cally recognizes an H3 tail that is unmodi?ed at K4 and R2 and acetylated at two or more lysines, with combinations including K18 and K23 being the most favorable. Importantly, in vitro ubiquitination assays demonstrate that binding of TIF1g to histones containing a heterogeneous mixture of PTMs activates its E3 ubiquitin ligase activity, as revealed by autoubiquitination and by its ability to ubiquitinate its substrate Smad4. We further demonstrate that ubiquitination of Smad4 by TIF1g in vivo requires the PHD ?ngerbromodomain. A Model of TIF1g’s Activity Considering all our results, we propose the following model for TIF1g’s activity (Figure 7). In the absence of a TGF-b signal, the PHD ?nger-bromodomain of TIF1g interacts with the TRIM domain, masking the Smad4 interaction region in the linker and keeping TIF1g in an inactive state (Figure 7A). TGF-b superfamily signals lead to nuclear accumulation of activated Smad complexes (Schmierer and Hill, 2007), which bind the SBR of the PAI-1 gene (Dennler et al., 1998)
Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc. 93

chromatin through engagement of the PHD-bromodomain with H3 tails, the TRIM domain would be released, leading to TIF1g activation and Smad4 ubiquitination.

Molecular Cell
Mechanism of Action of TIF1g

Figure 7. Model of TIF1g Action
(A–C) Model for chromatin-dependent TIF1g ubiquitination of Smad4. The TRIM domain is indicated in yellow when TIF1g is inactive; in active TIF1g, it is orange. The key histone H3 modi?cations are labeled on both the blue box and the pink circle. The arrow indicates H3 acetylation by a HAT. SBE, Smad binding element; Ub, ubiquitin. For discussion, see text.

and promoters/enhancers of other target genes (Ross et al., 2006). Activated Smads recruit, among other factors, histone acetyltransferases (HATs) such as p300/CBP, which acetylate the H3 and H4 tails (Ross et al., 2006). This would promote binding of TIF1g to the chromatin at, for example the SBR of the PAI-1 promoter, since the PHD ?nger interacts sequence speci?cally with the N-terminal region of H3, and the bromodomain would engage with multiple acetylated lysines on the H3 tail. We propose that this binding activates TIF1g’s E3 ubiquitin ligase activity, as it could release the TRIM domain from the intramolecular interaction with the PHD ?nger-bromodomain. Activated TIF1g would then ubiquitinate chromatin-bound Smad4 at lysines 519 and 507 (Figure 7B). Ubiquitination at these sites is known to disrupt Smad complexes (Dupont et al., 2009) and hence would promote their release from the promoter (Figure 7C). Consistent with this model, the HAT p300 has been shown to promote monoubiquitination of Smad4 (Wang et al., 2008). We therefore propose that TIF1g represses TGF-b superfamily transcription by restricting the residence time of individual activated Smad complexes on the promoters/enhancers of target genes. It is now accepted that the recruitment and assembly of the transcription machinery at promoters/enhancers are highly dynamic events and that modulation of dynamic interactions of chromatin-bound transcription factors is an important regulatory mechanism for gene expression (Hager et al., 2009). Although the residence time of Smad binding to chromatin has not been measured, it is estimated that transcription factor residence times on chromatin are in the range of 30–120 s (Hager et al., 2009). We propose that modulation of the Smad residence time on the promoter by TIF1g will dictate the ef?ciency of transcriptional initiation, which will determine levels of mRNA synthesis. Consistent with this idea, slower exchange dynamics of the glucocorticoid receptor at a promoter correlates with more mRNA synthesis from that promoter (Stavreva et al., 2004). The
94 Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc.

short timescale of seconds for residence times of individual Smad complexes on the PAI-1 promoter ?ts well with the fact that we are unable to detect a stabilizing effect of TIF1g knockdown when we assayed the kinetics of PAI-1 mRNA decay after signal termination on a timescale of minutes/hours. Our observation that TIF1g was enriched on chromatin upon TGF-b stimulation was counterintuitive, as we anticipated that a repressor would be bound when the gene was not transcribed. In the TGF-b signaling pathway, the repressors Ski and SnoN act in this way, being bound to promoters/enhancers of target genes in the absence of signal and recruiting transcriptional corepressors such as N-CoR or mSin3A to keep the genes silent. Upon TGF-b stimulation, Ski and SnoN are degraded to allow the activated Smad complexes access to promoters/enhancers (Le Scolan et al., 2008; Levy et al., 2007). Thus, this mode of repression can only be reversed by repressor degradation. However, another type of transcriptional repressor can be envisaged. For biological processes to be continuously ?nely tuned, activatory processes have to be counteracted by continuous deactivation, the balance of activation/deactivation determining output. We propose that TIF1g continuously opposes Smadinduced transcriptional activation using just such a mechanism. By triggering histone modi?cations, activated Smad complexes indirectly promote recruitment of the repressor TIF1g that functions to directly control the time that the Smad complexes remain bound and active at a promoter/enhancer. By bringing HATs like p300/CBP to TGF-b target gene enhancers/promoters (Ross et al., 2006), the Smads indirectly act as the ‘‘writers’’ of the histone code that is read by TIF1g. Recognition of the H3 Tail by TIF1g’s PHD Finger-Bromodomain Our data identify TIF1g’s tandem PHD ?nger-bromodomain as a histone-binding module, which recognizes the H3 tail and ‘‘reads’’ a speci?c histone code consisting of unmodi?ed K4

Molecular Cell
Mechanism of Action of TIF1g

and R2 and acetylation of two or more of K9, 14, 18, 23, and 27. This speci?city highlights the equal importance of unmodi?ed and modi?ed amino acids in the histone code. Peptide pulldowns demonstrated that the TIF1g PHD ?nger recognizes the N-terminal region of the H3 tail and requires K4 and R2 to be unmethylated. This speci?c H3 sequence recognition may provide the observed speci?city in TIF1g binding for H3 over H4. The bromodomain binds C-terminally on the same H3 tail. It displays some speci?city for acetylated K23 alone, but binds optimally to two consecutive acetylated lysines. The recent structure of the BRDT bromodomain revealed that two acetylated lysine residues spaced three residues apart cooperate to interact with one ` binding pocket (Moriniere et al., 2009). Our observations suggest that the TIF1g bromodomain similarly recognizes two acetylated lysines, albeit with a spacing of four or ?ve residues. Our experiments indicate that the TIF1g PHD ?nger-bromodomain binds to a single H3 tail and binding is thus intranucleosomal. This is in contrast to the BPTF tandem PHD ?nger-bromodomain, which is proposed to bind two different histone tails, with the PHD ?nger binding an H3 tail methylated at K4 and the bromodomain binding an acetylated H4 tail (Ruthenburg et al., 2007). Binding to Modi?ed Histones Activates TIF1g’s E3 Ubiquitin Ligase Activity For activated Smad complexes to accumulate effectively in the nucleus after ligand stimulation, factors that promote dissociation of Smad complexes must be rigorously controlled. We previously proposed that a Smad disruption step was required prior to Smad dephosphorylation in the nucleus, as the phosphates on the R-Smads are buried in the Smad complex (Schmierer et al., 2008). By ubiquitinating Smad4 at K519 and K507, TIF1g promotes disruption of Smad complexes. However, our demonstration that the E3 ubiquitin ligase activity of TIF1g is induced upon histone binding allows us to conclude that TIF1g only induces dissociation of chromatin-associated Smad complexes. TIF1g not associated with chromatin would be in the inactive masked conformation, and thus Smad complexes in the nucleoplasm are protected from dissociation and hence dephosphorylation. In summary, our work provides insights into the mechanism of transcriptional repression. We show that histone binding and ‘‘reading’’ of a speci?c histone code has a direct effect on the enzymatic activity of the reader, which mediates its repressive activity by inactivating a transcriptional activator.
EXPERIMENTAL PROCEDURES Cells, Plasmids, siRNAs, and Antibodies HaCaT, HaCaT-TR, HaCaT-TRS4, 293T, and MDA-MB-231 cells were cultured in Dulbecco’s modi?ed Eagle’s medium containing 10% fetal calf serum. The HaCaT shTIF1g cell line was generated by transfecting the shTIF1g/Ectodermin plasmid (Dupont et al., 2009) using FuGENE HD (Roche) according to the manufacturer’s instructions, followed by selection with 3 mg/ml puromycin (Invivogen). To generate the FLAG-TIF1g and FLAGTIF1g DPHD/Bromo 293T stable cell lines, 293Ts were transfected using FuGENE HD with the appropriate plasmids together with pSUPER-retropuro (OligoEngine) for puromycin resistance. Cells were selected with 0.8 mg/ml puromycin. The MDA-MB-231 CAGA12-Luc/TK-Renilla and the MDA-MB-231 BRE-Luc/TK Renilla cell lines have been described (Wu et al.,

2011). The HaCaT SRE-Luc/TK Renilla cell line was a gift from Richard Treisman. Plasmids, antibodies, and siRNAs are described in the Supplemental Experimental Procedures. Cell Treatments Cells were induced as indicated with 2 ng/ml TGF-b1 (PeproTech), 20 ng/ml Activin (R&D Systems), 100 ng/ml BMP7 (R&D Systems), or 2 mg/ml tetracycline. 293Ts were treated overnight with 10 mM SB-431542 (Tocris) to inhibit autocrine signaling. Where cells were stimulated with ligand, the SB-431542 was washed out with PBS, and then fresh media was added before ligand induction. The HaCaT SRE cells were starved overnight in media containing 0.3% FCS and stimulated with 2 mM Cytochalasin D (Calbiochem). Luciferase Assays, Reverse Transcription, and Quantitative PCR Dual luciferase/Renilla assays were performed as described (Levy et al., 2007). Reverse transcription and qPCR were essentially as described (Ross et al., 2006), with the following modi?cations. Total RNA (2 mg) was reverse transcribed using Af?nity Script QPCR cDNA Synthesis Kit (Stratagene), and qPCR was performed with Fast SYBR Green Master Mix (Applied Biosystems) on an ABI 7500 FAST real-time PCR system. The sequences of the primers are given in the Supplemental Experimental Procedures. Assays were performed in duplicate. For each time point, the average PAI-1 mRNA values were normalized to average GAPDH values, and the relative values were quantitated with the formula 2(average PAI-1) ? (average GAPDH). SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, Supplemental References, and four ?gures and can be found with this article online at doi:10.1016/j.molcel.2011.05.020. ACKNOWLEDGMENTS ¨ We thank Eva Gronroos for the MDA-MB-231 reporter cell lines, Tamara Gruner and Richard Treisman for the MRTF antibody and the SRE reporter cell line, Nicola O’Reilly and the peptide synthesis facility for peptide synthesis, and Ozan Aygun and Jesper Svejstrup for advice on protein puri?cation and the HA-ubiquitin plasmid. We acknowledge the London Research Institute Equipment Park and Cell Services for their assistance. We thank Stefano Piccolo, Erik Sahai, Jesper Svejstrup, Helle Ulrich, and members of the Hill lab for discussions and/or very useful comments on the manuscript. The work was funded by Cancer Research UK, the EU (MRTN-CT-2004-005428), and an AIRC grant (S.D.). Received: July 7, 2010 Revised: March 5, 2011 Accepted: May 2, 2011 Published: July 7, 2011 REFERENCES Bai, X., Kim, J., Yang, Z., Jurynec, M.J., Akie, T.E., Lee, J., LeBlanc, J., Sessa, A., Jiang, H., DiBiase, A., et al. (2010). TIF1g controls erythroid cell fate by regulating transcription elongation. Cell 142, 133–143. Chakravarty, S., Zeng, L., and Zhou, M.M. (2009). Structure and site-speci?c recognition of histone H3 by the PHD ?nger of human autoimmune regulator. Structure 17, 670–679. Chignola, F., Gaetani, M., Rebane, A., Org, T., Mollica, L., Zucchelli, C., Spitaleri, A., Mannella, V., Peterson, P., and Musco, G. (2009). The solution structure of the ?rst PHD ?nger of autoimmune regulator in complex with non-modi?ed histone H3 tail reveals the antagonistic role of H3R2 methylation. Nucleic Acids Res. 37, 2951–2961. Dennler, S., Itoh, S., Vivien, D., ten Dijke, P., Huet, S., and Gauthier, J.M. (1998). Direct binding of Smad3 and Smad4 to critical TGF b-inducible

Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc. 95

Molecular Cell
Mechanism of Action of TIF1g

elements in the promoter of human plasminogen activator inhibitor-type 1 gene. EMBO J. 17, 3091–3100. Dupont, S., Zacchigna, L., Cordenonsi, M., Soligo, S., Adorno, M., Rugge, M., and Piccolo, S. (2005). Germ-layer speci?cation and control of cell growth by Ectodermin, a Smad4 ubiquitin ligase. Cell 121, 87–99. Dupont, S., Mamidi, A., Cordenonsi, M., Montagner, M., Zacchigna, L., Adorno, M., Martello, G., Stinch?eld, M.J., Soligo, S., Morsut, L., et al. (2009). FAM/USP9x, a deubiquitinating enzyme essential for TGFbeta signaling, controls Smad4 monoubiquitination. Cell 136, 123–135. ? Groner, A.C., Meylan, S., Ciuf?, A., Zangger, N., Ambrosini, G., Denervaud, N., Bucher, P., and Trono, D. (2010). KRAB-zinc ?nger proteins and KAP1 can mediate long-range transcriptional repression through heterochromatin spreading. PLoS Genet. 6, e1000869. Hager, G.L., McNally, J.G., and Misteli, T. (2009). Transcription dynamics. Mol. Cell 35, 741–753. He, W., Dorn, D.C., Erdjument-Bromage, H., Tempst, P., Moore, M.A., and ? Massague, J. (2006). Hematopoiesis controlled by distinct TIF1g and Smad4 branches of the TGFbeta pathway. Cell 125, 929–941. Inman, G.J., and Hill, C.S. (2002). Stoichiometry of active smad-transcription factor complexes on DNA. J. Biol. Chem. 277, 51008–51016. ? Inman, G.J., Nicolas, F.J., Callahan, J.F., Harling, J.D., Gaster, L.M., Reith, A.D., Laping, N.J., and Hill, C.S. (2002). SB-431542 is a potent and speci?c inhibitor of transforming growth factor-b superfamily type I activin receptorlike kinase (ALK) receptors ALK4, ALK5, and ALK7. Mol. Pharmacol. 62, 65–74. Ivanov, A.V., Peng, H., Yurchenko, V., Yap, K.L., Negorev, D.G., Schultz, D.C., Psulkowski, E., Fredericks, W.J., White, D.E., Maul, G.G., et al. (2007). PHD domain-mediated E3 ligase activity directs intramolecular sumoylation of an adjacent bromodomain required for gene silencing. Mol. Cell 28, 823–837. Jain, A.K., and Barton, M.C. (2009). Regulation of p53: TRIM24 enters the RING. Cell Cycle 8, 3668–3674. Jenuwein, T., and Allis, C.D. (2001). Translating the histone code. Science 293, 1074–1080. ? Khetchoumian, K., Teletin, M., Mark, M., Lerouge, T., Cervino, M., OuladAbdelghani, M., Chambon, P., and Losson, R. (2004). TIF1d, a novel HP1-interacting member of the transcriptional intermediary factor 1 (TIF1) family expressed by elongating spermatids. J. Biol. Chem. 279, 48329–48341. Korchynskyi, O., and ten Dijke, P. (2002). Identi?cation and functional characterization of distinct critically important bone morphogenetic protein-speci?c response elements in the Id1 promoter. J. Biol. Chem. 277, 4883–4891. Le Scolan, E., Zhu, Q., Wang, L., Bandyopadhyay, A., Javelaud, D., Mauviel, A., Sun, L., and Luo, K. (2008). Transforming growth factor-b suppresses the ability of Ski to inhibit tumor metastasis by inducing its degradation. Cancer Res. 68, 3277–3285. Levy, L., and Hill, C.S. (2005). Smad4 dependency de?nes two classes of transforming growth factor b (TGF-b) target genes and distinguishes TGF-b-induced epithelial-mesenchymal transition from its antiproliferative and migratory responses. Mol. Cell. Biol. 25, 8108–8125. Levy, L., Howell, M., Das, D., Harkin, S., Episkopou, V., and Hill, C.S. (2007). Arkadia activates Smad3/Smad4-dependent transcription by triggering signal-induced SnoN degradation. Mol. Cell. Biol. 27, 6068–6083. Li, H., Ilin, S., Wang, W., Duncan, E.M., Wysocka, J., Allis, C.D., and Patel, D.J. (2006). Molecular basis for site-speci?c read-out of histone H3K4me3 by the BPTF PHD ?nger of NURF. Nature 442, 91–95.

Luger, K., Rechsteiner, T.J., Flaus, A.J., Waye, M.M., and Richmond, T.J. (1997). Characterization of nucleosome core particles containing histone proteins made in bacteria. J. Mol. Biol. 272, 301–311. Meroni, G., and Diez-Roux, G. (2005). TRIM/RBCC, a novel class of ‘single protein RING ?nger’ E3 ubiquitin ligases. Bioessays 27, 1147–1157. ` ? Moriniere, J., Rousseaux, S., Steuerwald, U., Soler-Lopez, M., Curtet, S., Vitte, A.L., Govin, J., Gaucher, J., Sadoul, K., Hart, D.J., et al. (2009). Cooperative binding of two acetylation marks on a histone tail by a single bromodomain. Nature 461, 664–668. Morsut, L., Yan, K.P., Enzo, E., Aragona, M., Soligo, S.M., Wendling, O., Mark, M., Khetchoumian, K., Bressan, G., Chambon, P., et al. (2010). Negative control of Smad activity by ectodermin/Tif1g patterns the mammalian embryo. Development 137, 2571–2578. Peng, H., Begg, G.E., Schultz, D.C., Friedman, J.R., Jensen, D.E., Speicher, D.W., and Rauscher, F.J., 3rd. (2000). Reconstitution of the KRAB-KAP-1 repressor complex: a model system for de?ning the molecular anatomy of RING-B box-coiled-coil domain-mediated protein-protein interactions. J. Mol. Biol. 295, 1139–1162. Posern, G., and Treisman, R. (2006). Actin’ together: serum response factor, its cofactors and the link to signal transduction. Trends Cell Biol. 16, 588–596. Ross, S., Cheung, E., Petrakis, T.G., Howell, M., Kraus, W.L., and Hill, C.S. (2006). Smads orchestrate speci?c histone modi?cations and chromatin remodeling to activate transcription. EMBO J. 25, 4490–4502. Ruthenburg, A.J., Li, H., Patel, D.J., and Allis, C.D. (2007). Multivalent engagement of chromatin modi?cations by linked binding modules. Nat. Rev. Mol. Cell Biol. 8, 983–994. Schmierer, B., and Hill, C.S. (2005). Kinetic analysis of Smad nucleocytoplasmic shuttling reveals a mechanism for transforming growth factor b-dependent nuclear accumulation of Smads. Mol. Cell. Biol. 25, 9845–9858. Schmierer, B., and Hill, C.S. (2007). TGFbeta-SMAD signal transduction: molecular speci?city and functional ?exibility. Nat. Rev. Mol. Cell Biol. 8, 970–982. Schmierer, B., Tournier, A.L., Bates, P.A., and Hill, C.S. (2008). Mathematical modeling identi?es Smad nucleocytoplasmic shuttling as a dynamic signal-interpreting system. Proc. Natl. Acad. Sci. USA 105, 6608–6613. ¨ Stavreva, D.A., Muller, W.G., Hager, G.L., Smith, C.L., and McNally, J.G. (2004). Rapid glucocorticoid receptor exchange at a promoter is coupled to transcription and regulated by chaperones and proteasomes. Mol. Cell. Biol. 24, 2682–2697. Taverna, S.D., Ilin, S., Rogers, R.S., Tanny, J.C., Lavender, H., Li, H., Baker, L., Boyle, J., Blair, L.P., Chait, B.T., et al. (2006). Yng1 PHD ?nger binding to H3 trimethylated at K4 promotes NuA3 HAT activity at K14 of H3 and transcription at a subset of targeted ORFs. Mol. Cell 24, 785–796. Taverna, S.D., Li, H., Ruthenburg, A.J., Allis, C.D., and Patel, D.J. (2007). How chromatin-binding modules interpret histone modi?cations: lessons from professional pocket pickers. Nat. Struct. Mol. Biol. 14, 1025–1040. Wang, B., Suzuki, H., and Kato, M. (2008). Roles of mono-ubiquitinated Smad4 in the formation of Smad transcriptional complexes. Biochem. Biophys. Res. Commun. 376, 288–292. Wu, M.Y., and Hill, C.S. (2009). Tgf-b superfamily signaling in embryonic development and homeostasis. Dev. Cell 16, 329–343. Wu, M.Y., Ramel, M.-C., Howell, M., and Hill, C.S. (2011). SNW1 is a critical regulator of spatial BMP activity, neural plate border formation, and neural crest speci?cation in vertebrate embryos. PLoS Biol. 9, e1000593.

96 Molecular Cell 43, 85–96, July 8, 2011 ?2011 Elsevier Inc.


相关文章:
更多相关标签: