Acetylation of Smad2 by the Co-activator p300 Regulates Activin and Transforming Growth Factor β Response*

Transforming growth factor β (TGFβ) signals primarily through the Smad proteins to regulate cell growth, differentiation, and extracellular matrix production. Post-translational modifications, such as phosphorylation, play an important role in the regulation of the Smad proteins. TGFβ signaling results in the phosphorylation of Smad2 and Smad3 that then oligomerize with Smad4 and translocate into the nucleus to initiate transcription of TGFβ target genes. The initiation of transcription is significantly enhanced by the direct interaction of the Smad complex with p300/CBP (CREB-binding protein), a co-activator with intrinsic acetyltransferase activity. However, how p300/CBP enhances transcription through this interaction is not entirely understood. In this report, we show that Smad2, but not the highly homologous Smad3, can be acetylated by p300/CBP in a ligand-dependent manner. At least three lysine residues, Lys19, Lys20, and Lys39, are required for efficient acetylation of Smad2, as mutations altering these lysines abolished Smad2 acetylation in vivo. This acetylation event is required for the ability of Smad2 to mediate activin and TGFβ signaling. Mutation of the three key lysine residues did not alter the stability of Smad2 or the ability of Smad2 to form a complex with Smad4 on promoter DNA, but it prevented nuclear accumulation of Smad2 and subsequent TGFβ and activin responses. Thus, our studies reveal a novel mechanism of modulating Smad2 activity and localization through protein acetylation.

The transforming growth factor ␤ (TGF␤) 2 signaling pathway plays complex roles in the regulation of many diverse biological processes including cell cycle arrest, differentiation, apoptosis, and epithelial to mesenchymal transition. TGF␤ signaling is initiated when ligands of the TGF␤ signaling superfamily such as activin and TGF␤ bind to and activate the TGF␤ serine/threonine kinase receptor complex (1,2). The activated receptor complex then phosphorylates receptor-activated Smads, such as Smad2 and Smad3, on the SSXS motif in the C termini, enabling nuclear translocation and oligomerization with the common mediator Smad4 (3,4). In the nucleus, this Smad heteromeric complex binds to target promoter sequences and recruits transcriptional coactivators to regulate transcription of TGF␤ target genes (1,5).
Smad2 and Smad3 mediate both activin and TGF␤ signaling and share 92% sequence identity. However, there are several salient differences between the two. First of all, Smad2 contains two extra peptide inserts, named the GAG and TID regions, respectively, that are not present in Smad3 (6). The TID domain (also known as exon3) in Smad2 prevents the direct binding of Smad2 to DNA, whereas Smad3 can bind DNA directly (7). Secondly, mice lacking Smad2 or Smad3 display very different phenotypes. Smad2 knock-out mice are embryonic lethal at embryonic day 10.5 (E10.5) with vascular and cranial abnormalities and impaired left-right patterning (8 -10). Smad3-null mice, however, are viable but suffer from impaired immune function and chronic inflammation (11,12). Finally, the stability and intracellular localization of Smad3 and Smad2 are also regulated by different mechanisms (13). Thus, many functional and regulatory differences exist between these two proteins.
Both Smad2 and Smad3 bind to and recruit the coactivator p300/CBP to enhance transcriptional activity of the activated Smad complex. p300 was first identified as an E1A-associated protein and displays high levels of sequence and functional homology with the CREB-binding protein, CBP (14). Initially identified as a scaffolding protein to bridge two proteins together, p300 also functions to transfer the acetyl group from acetyl coenzyme A to the lysine residues in histones allowing remodeling of chromatin to a more open relaxed conformation for transcription (15)(16)(17). Acetylation of histones can be reversed by the activity of histone deacetylases (HDAC), which returns the active chromatin back to its closed, inactive form by removal of acetyl groups (18).
In the past decade, many non-histone proteins, such as p53, ␤-catenin, importin-␣, E1A, and Smad7 (19 -21), have been shown to be acetylated by p300 and other acetyltransferase proteins such as P/CAF and GCN. Acetylation of non-histone proteins can result in alterations of biochemical and functional activities of the substrate proteins. For example, acetylation appears to inhibit the interaction of E1A with importin-␣, decreasing its nuclear import (22). Acetylation can alter the intracellular localization of proteins such as c-Abl and can also compete with other modifications such as ubiquitination and sumoylation in proteins like p53 and Smad7 (21,23,24). Acetylation of Smad7, an inhibitory Smad, increases its stability by modifying the lysine residues required for ubiquitin attachment, preventing ubiquitination and proteasomal degradation (25). Finally, acetylation may alter the intracellular localization, DNA binding, and other post-translational modifications of proteins (26).
The MH2 domains of Smad1, Smad2, Smad3, and Smad4 interact directly with the C-terminal domain of p300/CBP, resulting in increases in the transcriptional activity of the Smad complex (27)(28)(29)(30)(31). Disruption of this interaction through overexpression of adenovirus E1A significantly decreases transcription by the Smad complex (32). Smad3 has also been shown to interact with HDAC1 directly through its MH1 domain and recruit it to the promoter DNA to inhibit transcription (33). Based on these observations, it has been proposed that the Smad proteins activate transcription by recruiting the p300/ CBP histone acetyltransferase to the chromatin to remodel it into the active, open confirmation (34). However, whether this is the only mechanism by which p300/CBP enhances Smad transcription is not clear. Indeed, Inoue et al. (35) recently reported acetylation of Smad3 on lysine 378 by p300/CBP in cells treated with trichostatin A (TSA). While this manuscripts was being prepared, a new study by Simonsson et al. (36) also reported identification of an acetylation site in Smad2 that appears to be important for the DNA binding activity of the alternatively spliced form of Smad2 but has no effect on the signaling activity of WT Smad2. In this study, we carried out a detailed analysis of acetylation of Smad proteins by p300. We show that Smad2, but not Smad3, can be acetylated in a p300dependent manner and that this modification requires lysines 19, 20, and 39 in Smad2. We further show that this acetylation event plays a role in promoting the nuclear accumulation of Smad2 upon TGF␤ stimulation, leading to the increase in downstream TGF␤ responses. Our work thus has uncovered a novel mechanism by which p300/CBP enhances TGF␤ signaling through direct modification of Smad2.

EXPERIMENTAL PROCEDURES
Cell Culture and Reagents-293T and NIH3T3 cells were maintained in Dulbecco's modified Eagle's medium containing 10% FBS. Hep3B human heptoma cells (American Type Culture Collection) were maintained in minimum Eagle's medium supplemented with 10% FBS. Smad2-deficient mouse embryonic fibroblasts obtained from Dr. Anita Roberts were cultured in Dulbecco's modified Eagle's medium containing 10% FBS, sodium pyruvate (1 mM), and glutamate (2 mM).
Monoclonal antibodies specific for acetylated lysine were purchased from Cell Signaling Technologies. Smad2 antibodies were from BD Transduction Laboratories and Smad3 (FL-425) and TGF␤ receptor (TGF␤RI; V22) antibodies from Santa Cruz Biotechnology. Anti-FLAG, anti-hemagglutinin (HA) antibodies and TSA were purchased from Sigma. Anti-phospho-Smad2 antisera were a generous gift from Aristidis Moustakas (Ludwig Institute for Cancer Research, Uppsala, Sweden). SB-431542 was purchased from Tocris.
Transfection and Stable Cell Lines-Transient transfections were performed using Lipofectamine Plus reagents (Invitro-gen) according to manufacturer's protocol. To generate stable cell lines, Smad2-null mouse embryo fibroblast (MEF) cells were co-transfected with wild type or mutant FLAG-Smad2 and pBABEpuro, which contains a puromycin resistance gene. After 48 h, transfected cells were selected by growing in medium containing 2 g/ml puromycin.
In Vitro Acetylation Assay-Purified GST-Smad fusion proteins were incubated at 30°C for 30 min in acetylation assay buffer (50 mM Tris, pH 7.8, 1 mM DTT, 10% glycerol, 1 mM PMSF, 10 mM sodium butyrate, 1 mM EDTA, and 0.05 Ci of 14 C-acetyl coenzyme A (PerkinElmer Life Sciences) in the presence of GST-tagged histone acetylation domain of p300 (GST-p300-HAT). Samples were resolved on a SDS-polyacrylamide gel and visualized by autoradiography using a phosphorimaging system.
Mass Spectrometry-FLAG-tagged Smad2, isolated from 293T cells co-transfected with full-length p300 by immunoprecipitation with anti-FLAG, were eluted with FLAG peptide and resolved on a 10% SDS-polyacrylamide gel. The Smad2 band was then excised from the gel, digested with chymotrypsin, and run through a Thermo Finnigan LCQ DECA XP Plus ion trap mass spectrometer interfaced with a Shimadzu Binary HPLC to carry out liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis.
Luciferase Assay-MEF cells were transiently transfected with the promoter reporter constructs and FLAG-tagged wild type or mutant Smad2. At 24 h after transfection, cells were serum-starved and treated with 50 pM TGF␤ for 16 h. Luciferin levels were measured as described previously (38).
Growth Inhibition Assay-1 ϫ 10 4 MEF cells were seeded in 6-well plates and stimulated with different concentrations of TGF␤1 for 4 days. Cells were counted and compared with the number of unstimulated cells to determine relative cell growth.
Pulse-Chase Assay-293T cells were co-transfected with fulllength p300 and FLAG-tagged wild type or mutant Smad2. At 24 h after transfection, cells were pulsed with 0.5 Ci/ml [ 35 S]methionine (PerkinElmer Life Sciences) in pulse media (cysteine/methionine-free Dulbecco's modified Eagle's medium plus 10% dialyzed FBS) for 30 min. Cells were then washed with regular media and chased for different periods of time. 35 S-Labeled Smad2 was then isolated by immunoprecipitation, resolved on a 10% SDS-polyacrylamide gel, and detected by autoradiography.
Electrophoretic Mobility Shift Assay (EMSA)-DNA fragments containing the Smad-binding element as described previously (40) were end-labeled with 32 P, gel-purified, and incubated with affinity-purified Smad2-Smad4 complex to test for DNA binding ability. The protein-DNA complexes were resolved on a 5% nondenaturing gel. For antibody supershift assays, the Smad2-Smad4 complexes were preincubated with 4 g of the specified antibody for 1 h at 4°C before EMSA was performed.
Immunofluorescence-Cells were seeded on sterile glass slides and treated with 100 ng/ml TSA for 8 h before staining. Cells were fixed with 4% paraformaldehyde, permeablized with 0.1% Triton X-100, blocked with blocking buffer (10% newborn calf serum, 1% bovine serum albumin, and 0.02% Triton X-100 in phosphate-buffered saline), and stained with anti-FLAG as described previously (41). Proteins were visualized on an Axiophot epifluorescence microscope or a confocal LSM 510 microscope (Zeiss).
To study protein shuttling, cells were treated for 30 min with 100 g/ml cycloheximide before stimulation with 100 pM TGF␤. Cells were then treated with 10 M SB-431542 prior to immunostaining.

RESULTS
Smad2 Is Acetylated in the Presence of p300-To determine whether Smad proteins could be acetylated, a panel of FLAGtagged Smad proteins were transiently transfected into 293T cells together with full-length p300 (Fig. 1A). Smad7 had been shown previously to be acetylated in the presence of p300 (21) and therefore was included as a positive control. Smad proteins were isolated by immunoprecipitation with anti-FLAG beads, and their acetylation status was evaluated by Western blotting with anti-acetyl lysine-specific antiserum. Under this condition, in addition to Smad7, only Smad2 was acetylated when coexpressed with p300. Interestingly, the highly homologous Smad3 was not significantly acetylated. The observed difference in acetylation between Smad2 and Smad3 is particularly interesting because of the high sequence identity between the two.
To determine whether p300 could directly acetylate Smad2, an in vitro acetylation assay was performed using bacterially expressed Smad2, Smad3, and Smad4 and the purified acetyltransferase domain of p300 ( Fig. 1B) or CBP (data not shown). Again, strong acetylation of Smad2 was observed in the presence of p300 or CBP, suggesting that p300/CBP can directly acetylate Smad2. In the same reaction, Smad3 and Smad4 also appeared to be labeled, but these signals were much weaker if significant at all when compared with that of Smad2. Thus, we decided to focus on the acetylation of Smad2 for the rest of the study.
Endogenous Smad2 Is Also Acetylated-To demonstrate that endogenous Smad2 is also acetylated and that this acetylation event is subjected to regulation by TGF␤, endogenous Smad2 FIGURE 1. Smad2 is acetylated by p300. A, FLAG-tagged Smads (F-Smad) were transiently transfected into 293T cells in the absence and presence of full-length p300 and isolated by immunoprecipitation as described under "Experimental Procedures." Acetylation of these Smad proteins was detected by Western blotting (WB) with an acetyl lysine-specific antibody. The membrane was stripped and reprobed using an anti-FLAG antibody. B, in vitro acetylation assay. Bacterially expressed GST-p53 and GST-Smad proteins were incubated with the acetyltransferase domain of p300 and 14 C-labeled acetyl-CoA. The 14 C-labeled proteins were resolved on an SDS-polyacrylamide gel, dried, and detected by phosphorimaging. C, Hep3B cells were serum-starved and treated with 2.5 M TSA for 8 h. Cells were then stimulated with 100 pM TGF␤ and harvested. Endogenous Smad2 was isolated by immunoprecipitation with anti-Smad2 followed by Western blotting with the indicated antibodies. D, Hep3B cells were transiently transfected with FLAG-Smad2 (F-Smad2) in the absence and presence of full-length p300. 24 h after transfection, cells were treated with 1 M TSA for 12 h. Smad2 was isolated by immunoprecipitation with anti-FLAG followed by Western blotting with the indicated antibodies. E, Hep3B cells were transiently transfected with FLAG-Smad2 and treated with 1 M TSA for 12 h. Cells were stimulated with 100 pM TGF␤, cytoplasmic and nuclear fractions were isolated, and lysates were probed for tubulin and HDAC1 as a control. Smad2 was isolated by immunoprecipitation with anti-FLAG followed by Western blotting with an anti-acetyl lysine antibody. JULY 20, 2007 • VOLUME 282 • NUMBER 29 was isolated from Hep3B cells treated with or without TGF␤ by immunoprecipitation with anti-Smad2 and subjected to Western blotting with anti-acetyl lysine-specific antibody. In the absence of TGF␤, Smad2 is acetylated but at a very low level. Treatment with TGF␤ induced increased acetylation of Smad2 (Fig. 1C). This was expected because p300 is localized only in the nucleus, and its interaction with Smad2 increases upon nuclear accumulation of Smad2 as a result of TGF␤ stimulation. This acetylation of Smad2 was further enhanced by treatment with TSA, an inhibitor of HDACs, suggesting that Smad2 acetylation is reversible and can be regulated by HDACs (Fig. 1D).

Smad2 Acetylation
To further confirm that acetylation of Smad2 indeed occurs in the nucleus, cytoplasmic and nuclear fractions were prepared from Hep3B cells, and acetylation of Smad2 was determined as described above. A low level of acetylated Smad2 could be detected in the nuclear fraction in the absence of TGF␤ stimulation, but this acetylation was significantly enhanced in the presence of TGF␤ (Fig. 1E). In contrast, no acetylated Smad2 was detected in the cytoplasmic fractions either in the presence of absence of TGF␤. Thus, acetylation of Smad2 occurs primarily in the nucleus.
The GAG Region of Smad2 Is Required for Acetylation-To identify the sites of acetylation in Smad2, we co-transfected Smad2/Smad3 chimeras swapping the MH1 or MH2 domains of Smad2 together with p300 into 293T cells and evaluated the acetylation of these mutants by Western blotting (Fig. 2, A and B). The N-terminal MH1 domain appeared to be necessary and sufficient for Smad2 acetylation (Fig.  2B). Within the MH1 domain, the major difference between Smad2 and Smad3 is the presence of the GAG and TID domains in Smad2 but not in Smad3. Deletion of the GAG domain from Smad2 significantly impaired Smad2 acetylation, whereas removal of the TID domain had no effect (Fig.  2C), suggesting that acetylation of Smad2 requires the GAG region. Consistent with this, when the GAG region was inserted into Smad3, acetylation of Smad3 was observed at levels comparable with that of wild type Smad2 (Fig. 2D). Thus, the GAG domain of Smad2 is both necessary and sufficient for Smad2 acetylation.
Identifying Smad2 Acetylation Sites-There are many lysine residues in the MH1 domain of Smad2, most of which are conserved in Smad3 with the exception of Lys 144 , which is uniquely present in Smad2. However, mutation of this lysine did not affect the acetylation of Smad2 (data not shown), indicating that Lys 144 is not the site of acetylation. It is likely, therefore, that the GAG region of Smad2 confers a difference in conformation between Smad2 and Smad3 that causes the surrounding lysines in Smad2 to be more accessible to acetylation by p300.
Two complementary approaches were taken to identify the sites of acetylation in Smad2. In the first approach, we employed mass spectrometry to detect lysine residues with N-linked acetylation modifications in FLAG-Smad2 isolated from 293T cells co-overexpressing p300. Mass spectrometry identified three lysine residues, lysines 19, 20, and 39, as possible acetylated residues (Fig. 3B). In the second approach, point mutations changing combinations of lysines to arginines (Fig.  3A) in the area proximal to the GAG region were generated and examined for acetylation. Mutation of lysines 19 and 20, either by themselves or in combination with other lysine residues (for example 3K13R; see Fig. 3A for diagram of Smad2 Lys-Arg point mutants), resulted in a significant loss of Smad2 acetylation, whereas mutations of all other lysine residues in this area did not affect acetylation (Fig. 3C). This suggests that Lys 19 and Lys 20 are required for Smad2 acetylation. However, they are not the only sites of acetylation, because the 2K19R mutant was still acetylated effectively in an in vitro acetylation assay (Fig. 3D). This is consistent with the result obtained from the mass spectrometry analysis showing that Lys 39 is also an acetylation site. Although mutation of Lys 39 alone did not significantly affect Smad2 acetylation and only partially inhibited its transcriptional activity (see Fig. 5), mutation of Lys 19 , Lys 20 , and Lys 39 abolished acetylation and Smad2 signaling (Figs. 3C and 5). Taken together, our study suggests that Smad2 can be acetylated on three lysine residues, Lys 19 , Lys 20 , and Lys 39 .
Smad2 Phosphorylation and Acetylation Occur Independently of Each Other-Stimulation of TGF␤ results in the phosphorylation of Smad2 in the C-terminal serine residues, leading to its oligomerization with Smad4 and nuclear translocation. To elucidate whether the C-terminal phosphorylation of Smad2 is necessary for acetylation, a Smad2 3S-3A mutant lacking the C-terminal phosphorylation sites was transfected into 293T cells along with full-length p300 (Fig.  4A). Although endogenous Smad2 needs to be phosphorylated to accumulate in the nucleus, immunofluorescence staining confirmed that overexpressed Smad2 3S-3A can bypass this requirement and accumulate in the nucleus even without TGF␤ stimulation (Fig. 4B). When thus overexpressed, the Smad2 3S-3A mutant was acetylated to a similar level as the wild type protein, indicating that Smad2 in the nucleus does not need to be phosphorylated for acetylation to occur.

Smad2 Acetylation
Similarly, acetylation of Smad2 did not alter the phosphorylation state of the protein. Wild type and 3K19R mutant Smad2 were phosphorylated to similar levels when co-transfected with p300 into 293T cells (Fig. 4C). Thus, although phosphorylation of Smad2 is necessary to for its nuclear translocation, once in the nucleus acetylation of Smad2 can occur independently of its phosphorylation status.
Acetylation of Smad2 Is Required for Activin and TGF␤ Signaling-To determine the effects of Smad2 acetylation on TGF␤ signaling, we first measured TGF␤-and activin-dependent transcription using luciferase reporter constructs specific for activin (ARE-lux) or TGF␤ (p3TP-lux) signaling. Western blotting was performed to ensure equal levels of Smad2 and 3K19R expression in the transfected cells (data not shown). MEFs isolated from the Smad2-null mice (43) were used to avoid any potential interference by endogenous Smad2.
Unlike WT MEF, which displayed strong activation of ARElux upon activation of activin signaling, Smad2-null MEF no longer supported transcription from ARE-lux (Fig. 5A). Transfection of wild type Smad2 restored the transcription activation to the Smad2-deficient MEF. In contrast, introduction of the 3K19R mutant did not rescue activin-induced transcriptional activation. Similarly, wild type but not the 3K19R mutant conferred TGF␤-induced transactivation to Smad2-null MEF (Fig.  5B). Thus, acetylation of Smad2 is required for activin-and TGF␤-induced transcriptional responses. Consistent with this, a mutant that mimics the acetylated state of Smad2 (3K19Q) exhibited a moderate but reproducible increase in transcription (Fig. 5, A and B) when expressed at a similar level as WT Smad2 (data not shown).
To discern the contribution of each of the three acetylated lysine residues to activin-and TGF␤-induced transcription responses, mutations altering each of the individual lysine residues (K19R, K20R, K39R) or two in combination (K19R/K20R, K19R/K39R, K20R/K39R) (Fig. 3A) were generated and tested in the luciferase reporter assay. All of these mutants partially inhibited the transcriptional activity of Smad2, suggesting that acetylation of all three lysine residues contribute to the activity  (WB). B, FLAGtagged Smad constructs were transiently transfected into Hep3B cells. Wild type F-Smad2 (F-Smad2) and F-Smad2 3S-3A were localized through immunostaining with anti-FLAG and visualized using confocal microscopy. C, wild type and 2K19R mutant Smad2 were transfected into 292T cell in the presence and absence of full-length p300 and constitutively active TGF␤ type I receptor (Alk5*). Phosphorylation of Smad2 was assessed by Western blotting with anti-phospho-Smad2 antibodies. FIGURE 5. Acetylation of Smad2 is required for activin and TGF␤ signaling. WT or mutant Smad2 was co-transfected into Smad2-null MEF together with either ARE-lux (A) or p3TP-lux (B). Cells were treated with 50 pM TGF␤ for16 h, and a luciferase assay was performed as described under "Experimental Procedures." C, WT or 3K19R Smad2 was stably introduced into Smad2-null MEF. Cells were treated with increasing concentration of TGF␤ for 4 days, and cell growth was evaluated by cell counting.

Smad2 Acetylation
of Smad2. The Lys 19 mutants demonstrated the most significant decrease in transcriptional activation, most likely because this residue is necessary for the cooperative acetylation of other lysines. Thus, mutating this lysine also affects the acetylation potential of the other identified acetylated lysines (Fig. 5, A and B).
We next examined how Smad2 acetylation affects the growth inhibitory response of TGF␤. Wild type or the 3K19R mutant Smad2 was transfected stably into the Smad2-null MEF. Cells were treated with different concentrations of TGF␤, and the growth of cells was assessed after 4 days (Fig. 5C). WT MEFs expressing endogenous Smad2 were responsive to TGF␤, exhibiting ϳ60% growth inhibition, whereas Smad2-null MEF showed very little growth inhibition. Stable expression of wild type Smad2 suppressed cell growth ϳ50% whereas cells expressing the 3K19R mutant did not exhibit any cell cycle arrest. Taken together, these results indicate that acetylation of Smad2 is required for activin and TGF␤ signaling.
Acetylation Does Not Affect the Stability, Smad4 Oligomerization, and DNA Binding Ability of Smad2-We next turned to the molecular mechanism by which acetylation of Smad2 affects its signaling activity. In particular, we examined the effects of Smad2 acetylation on its stability, the ability to hetero-oligomerize with Smad4, its presence in the DNA-binding complex, and intracellular localization. In pulse-chase experiments using transfected 293T cells, both the wild type Smad2 and the 3K19R mutant have similar half-lives (Fig. 6A), suggesting that Smad2 stability is not affected by acetylation.
Because Smad2 must oligomerize with Smad4 to carry out its transcription activity, we next examined whether acetylation affected its interaction with Smad4 by a co-immunoprecipitation assay (Fig. 6B). In 293T cells transfected with FLAG-tagged wild type or mutant Smad2 together with Myc-Smad4 and the constitutively active TGF␤ receptor (Alk5*), wild type and 3K19R mutant associated with similar amounts of Smad4, indicating that acetylation does not alter the oligomerization of Smad2 with Smad4.
Similarly, in an EMSA using the Smad2-Smad4 complex purified from transiently transfected 293T cells, no difference in DNA binding ability was detected between the wild type Smad2 and the 2K19R mutant, suggesting that acetylation of Smad2 does not affect DNA binding (Fig. 6C).
Acetylation of Smad2 Affects Intracellular Localization after TGF␤ Treatment-Acetylation has been shown to play a role in the intracellular localization of many proteins including p53, HNF-␣, and importin-␣. To elucidate whether Smad2 acetylation also influences its intracellular localization, we performed indirect immunofluorescence staining in Hep3B cells transfected with wild type or mutant Smad2 (Fig. 7). In the absence of TGF␤, both wild type and mutant Smad2 were distributed throughout the cell in both cytoplasm and nucleus, consistent with reports that under nonstimulated conditions Smad2 quickly shuttles in and out of the nucleus (44). Upon TGF␤ treatment, wild type Smad2 rapidly accumulated in the nucleus as expected. In contrast, the 3K19R mutant remained in the cytoplasm and did not accumulate in the nucleus. Similar results were also observed in NIH3T3 cells and Smad2-null MEF cells (data not shown). Single lysine to arginine and double lysine to arginine mutants also exhibited a failure to fully accumulate in the nucleus, whereas the 3K19Q mutant, which mimics the acetylation state of Smad2, localized readily in the nucleus upon TGF␤ stimulation (Fig. 7). Thus, acetylation of Smad2 appears to be required for its nuclear translocation in response to TGF␤. This decreased nuclear accumulation in response to TGF␤ most likely accounts for the inability of the mutant Smad2 to mediate TGF␤ and activin signaling.
Acetylation May Affect Nuclear Export-If acetylation of Smad2 increases its nuclear accumulation, one would predict that the acetylation-mimicking 3K19Q mutant will linger in the nucleus longer following TGF␤ stimulation. To test this, we performed a time course experiment examining the localization of WT Smad2 and 3K19Q at various times after TGF␤ stimulation. Hep3B cells transiently expressing WT or mutant Smad2 were pretreated with cycloheximide prior to TGF␤ treatment to prevent new protein synthesis. After 1 h of TGF␤ treatment, the ALK5 receptor inhibitor SB-431542 was added to the cells to block further activation of the Smad proteins. Localization of Smad2 proteins at various time points after these treatments was determined by immunofluorescence staining. By this approach we were able to follow the fate of activated Smad2. FIGURE 6. Smad2 acetylation does not affect Smad2 stability, oligomerization with Smad4, or its presence on promoter DNA. A, pulse-chase assay. 293T cells were transfected with wild type or 3K19R Smad2 in the presence and absence of full-length p300, pulse-labeled, and chased. 35 S-Labeled Smad2 were isolated and resolved by SDS-PAGE as described under "Experimental Procedures." B, wild type or 3K19R Smad2 was co-transfected into 293T cells together with Myc-tagged Smad4 in the absence and presence of full-length p300 and the constitutively active TGF␤ type I receptor (RI), Alk5*. Smad2 was isolated by immunoprecipitation with anti-FLAG, and the associated Smad4 was detected by Western blotting (WB) with anti-Myc antibody. C, EMSA. The Smad2-Smad4 complex was isolated from 293T cells transfected with FLAG-Smad4, a constitutively active TGF␤ type I receptor (Alk5*), and HA-tagged wild type or 2K19R Smad2. An EMSA and a supershift assay were performed as described under "Experimental Procedures." Consistent with previously published results (45), WT Smad2 accumulated in the nucleus at 1 h after TGF␤ stimulation but relocated back to the cytoplasm at 2 h post-stimulation. In contrast, the 3K19Q mutant readily accumulated in the nucleus in response to TGF␤ but failed to be redistributed back to the cytoplasm even after 3 h of TGF␤ treatment (Fig. 8). This observation confirms that acetylation promotes nuclear localization of Smad2 and further suggests that it may do so by decreasing nuclear export of Smad2.

DISCUSSION
Reversible protein acetylation has been shown to affect a diverse array of biochemical properties including protein-protein interactions, DNA binding, protein stability, and intracellular localization. In this study we have demonstrated that Smad2, but not Smad3, can be robustly acetylated in the presence of p300/CBP both in vitro and in vivo to enhance TGF␤ and activin signaling. This acetylation requires at least three key lysine residues, lysines 19, 20, and 39, and appears to promote Smad2 nuclear accumulation, leading to enhanced transcription activity. We have shown that acetylation of Smad2 appears to affect the localization of Smad2 upon TGF␤ signaling possibly by decreasing the rate of Smad2 nuclear export following TGF␤ stimulation. The 3K19R mutant that could no longer be acetylated failed to accumulate in the nucleus, whereas the 3K19Q mutant that mimicked a constitutively acetylated state remained in the nucleus much longer than WT Smad2 after TGF␤ signaling ends.
Smad2 has been shown to continuously shuttle in and out of the nucleus (45), and TGF␤-induced nuclear accumulation occurs primarily as a result of a decrease in the rate of nuclear export (46). Our result that acetylation affects nuclear export is entirely consistent with these earlier observations and also to be expected because import of Smad2 precedes acetylation by nuclear p300. It is not clear how acetylation directly influences nuclear export. There are at least two possibilities. First, acetylation may affect the dephosphorylation of Smad2, which is necessary for relocation of Smad2 to the cytoplasm (47). Alternatively, acetylation may directly impact the export machinery. Xu et al. (48) have shown that Smad2 directly binds to the nucleoporins CAN/Nup214 and Nup153 and that Nup153 is necessary for Smad2 export in an importin/exportin-independent manner. Whether this mechanism is the only one responsible for Smad2 export in vivo is not clear. Much more work is clearly needed in the future to determine the exact mechanism by which acetylation prevents nuclear export. Nevertheless, our data suggest that in conjunction with Smad phosphorylation, Smad2 acetylation contributes to the slowing of nuclear export following TGF␤ stimulation and therefore constitutes an important step in the regulation of TGF␤ and activin signaling.
Smad2 is not the only protein in the TGF␤ signaling pathway that can be acetylated. In addition to Smad7, Inoue et al. (35) recently reported acetylation of Smad2 and Smad3 by p300/CBP in cells treated with TSA. They found that acetylation of Smad3 occurred in the MH2 domain on Lys 378 , a lysine conserved among all R-Smads. Mutation of this lysine to arginine (K378R) appeared to result in a decrease in receptor-mediated phosphorylation of Smad3 and subsequently a reduction in transcriptional activity. However, we did not observe very robust acetylation of Smad3, especially compared with that of Smad2, nor did mass spectrometry experiments identify Lys 378 as an acetylated residue under our FIGURE 7. Smad2 acetylation is required for nuclear accumulation. Hep3B cells were transfected with 0.1 g of WT or 3K19R-Smad2 in the absence and presence of full-length p300. Cells were treated with TSA for 10 h and with 100 pM TGF␤ for 1 h prior to staining. Smad2 localization was determined by immunostaining with anti-FLAG and visualized using confocal microscopy. FIGURE 8. Smad2 acetylation may decrease the rate of nuclear export. Hep3B cells transiently expressing WT or mutant Smad2 in the presence of full-length p300 were pretreated with 100 g/ml cycloheximide for 30 min before treatment with 100 pM TGF␤. Cells were treated either with or without SB-451542 for the indicated times prior to immunostaining with anti-FLAG and visualization using confocal microscopy. experimental conditions. While this manuscripts was being prepared, a new report by Simonsson et al. (36) also found acetylation of Smad3 to occur at a significantly lower level than that of Smad2. In contrast to our observation that mutant Smad2 defective in acetylation exhibited impaired signaling activity, they detected no reduction in transcriptional activity of the K19R mutant Smad2 in HepG2 cells using the ARE-lux reporter assay. We speculate that the reason that they failed to observe a difference between wild type and mutant Smad2 in transcription may be because of the interference of endogenous Smad2 in HepG2 cells, as we did not observe any difference in transcription activity between wild type and mutant Smad2 in Hep3B cells either (Data not shown). The endogenous Smad2 in these cells could be functioning at saturation levels, masking the effect of the mutant Smad2. When we used Smad2-null MEF, a clear difference could be readily observed.
Smad2 exists in two isoforms: the full-length protein form containing both the GAG and TID (exon3) domains and an alternatively spliced short form lacking exon 3 (Smad2⌬E3) (49). Unlike full-length Smad2, this shorter isoform is able to bind DNA directly. Simonsson et al. (36) showed that acetylation on Lys 19 in the short isoform Smad2(⌬E3) is required for its DNA binding activity. However, because the fulllength Smad2 does not bind to DNA directly, acetylation is unlikely to have the same effect on full-length Smad2. Indeed, we showed that the 3K19R mutant Smad2 can still be recruited to DNA through oligomerization with Smad4. Instead, acetylation of full-length Smad2 enhances Smad2 activity by promoting its nuclear accumulation through decreasing the rate of nuclear export. Because the full-length Smad2 is found in all adult and embryonic tissues but Smad2(⌬E3) is present primarily in mouse cells during development, acetylation may regulate the activity of these Smad2 forms through different mechanisms in embryos and adult tissue.
Like phosphorylation, acetylation can occur on multiple lysine residues within the same protein and often can occur in different combinations depending on the specific conditions of the cells and the given acetyltransferase enzymes, leading to various downstream consequences. Although lysines 19, 20, and 39 were found to be required for Smad2 acetylation, they may not be the only residues that are acetylated. Additional Smad2 acetylation sites may exist that may conceivably affect Smad2 activity. Identification of specific acetylation sites may be difficult, as the specificity of acetyltransferase for substrate lysine residues is not stringent so that, in the absence of the targeted lysine, other lysine residues in the proximity may serve as substrates. Despite this difficulty, a continuing effort to decipher Smad2 acetylation sites under different stimuli and conditions will surely increase our understanding of how the TGF␤ signaling pathway is regulated by acetylation. Our findings also provide a possible new direction for pharmaceutical intervention of TGF␤ signaling in cancer and inflammatory diseases by targeting Smad2 acetylation. Treatment of cells with HDAC inhibitors may increase the nuclear accumulation of Smad2, leading to improved growth-inhibitory responses of cells response to TGF␤.