SUMO-1/Ubc9 promotes nuclear accumulation and metabolic stability of tumor suppressor Smad4.

Tumor suppressor Smad4/DPC4 is a central intracellular signal transducer for transforming growth factor-beta (TGF-beta) signaling. We recently reported that transcriptional potential of Smad4 was regulated by SUMOylation in transfected HeLa cells (1), but the precise mechanism and function of Smad4 SUMOylation in TGF-beta signaling remain to be elucidated. Here, we describe the regulation of TGF-beta signaling by SUMOylation through the control of Smad4 metabolic stability and subcellular localization. We found that SUMO-1 overexpression strongly increases Smad4 levels, while inhibition of SUMOylation by small interfering RNA (siRNA)-mediated knockdown of the E2 enzyme Ubc9 reduces endogenous Smad4 levels. Concomitantly, SUMO-1 overexpression enhances and Ubc9 knockdown reduces levels of intranuclear Smad4, growth inhibitory response, as well as transcriptional responses to TGF-beta. Comparison of wild type and mutant forms of Smad4 for SUMOylation, ubiquitination, and half-life allows the conclusion that SUMO-1 modification serves to protect Smad4 from ubiquitin-dependent degradation and consequently enhances the growth inhibitory and transcriptional responses of Smad4.

The strength and intensity of TGF-␤ 1 signaling require a tight control of the activity of each signaling component, including the central signal transducing Smad proteins. Tumor suppressor Smad4/DPC4 is the common mediator for TGF-␤ signaling by forming a complex with R-Smads in response to ligand stimulation (2)(3)(4)(5)(6). The heteromeric complexes of R-Smads and Smad4 are then translocated into the nucleus where they exert ligand-induced changes in transcription of a variety of genes involved in cell responses, including cell proliferation, differentiation, and extracellular matrix remodeling (for reviews, see Refs. [2][3][4][5][6][7]. The heteromeric Smad complex activates transcription through its ability to functionally cooperate with several promoter-specific transcription factors and/or to bind specific DNA sequences (7,8).
Recent studies have shown R-Smads are regulated by the proteasome-mediated degradation system (9 -13). Ubiquitination, the covalent attachment of ubiquitin to proteins, predominantly serves to target proteins for their degradation by proteasomes (14). Interestingly, a number of ubiquitin-related proteins are also present in eukaryotic cells (15). These proteins, including the small ubiquitin-like modifier-1 (SUMO-1), utilize a conjugation system that is similar to ubiquitination (16 -18). In contrast to ubiquitination, SUMO-1 modifications of target proteins do not promote their degradation, but modulate the subcellular localization or biological activities of targets (19 -25).
Smad4 is the central mediator for signaling of TGF-␤ superfamily, and thus it is important to study the regulation of Smad4. Recently, we and another group identified Smad4 as a substrate of SUMOylation pathway (1,33). Here we further elucidate the mechanism of how the SUMOylation regulates Smad4 activity under physiological conditions. We have found that RNA interference (RNAi)-mediated silencing of the human Ubc9 gene disrupts Smad4 SUMOylation, decreases Smad4 stability, reduces Smad4 accumulation in the nucleus, and consequently blocks TGF-␤ signaling. Thus, SUMOylation of tumor suppressor Smad4 provides a novel mechanism to control TGF-␤ antiproliferative signaling.
Ni-NTA Precipitation and Western Blot-Ni-NTA precipitation and Western blot analysis were essentially as described previously (1). Briefly, HeLa cells were transfected with expression plasmids for Histagged Smad4 and FLAG-tagged SUMO1 and harvested in guanidinium lysis buffer (6 M guanidinium HCl, 0.1 M NaPO 4 , 0.01 M Tris⅐Cl, pH 8). His-tagged Smad4 was retrieved from the lysates by using Ni-NTA beads (Qiagen), followed by standard gel electrophoresis and Western blotting analysis using anti-Smad4 (Santa Cruz Biotechnology) and anti-FLAG (Sigma) antibody.
Transcription Reporter Assays-Transfections, TGF-␤ treatment, and reporter assays were carried out as described previously (27). Generally, HeLa cells at 25-30% confluence were transfected with expression plasmids for Smads and/or reporter plasmids. Reporter plasmid SBE-Luc contains the luciferase gene under control of the Smadbinding elements (SBE) (28). The amount of transfected DNA was always the same by adding vector DNA, whenever needed. 40 -45 h after transfection, cells were treated with 200 pM TGF-␤ for 12 h. Cells were then harvested for measurement of luciferase and ␤-galactosidase activities. All assays were done in triplicate and all values were normalized for transfection efficiency against ␤-galactosidase activity.
Immunofluorescence-HeLa and RI-14 cells, untransfected or trans-fected as specified in the text and figure legends (Figs. 1, C and E, and 3C), were grown on cover slips, fixed with cold methanol, and blocked with 2% bovine serum albumin in phosphate-buffered saline, pH 7. Cells were then stained with anti-Smad4 or anti-His monoclonal antibodies, followed with FITC-conjugated anti-mouse antibody (Sigma), and visualized under a Zeiss Axioplan II microscope.
Pulse-Chase and Ubiquitination Assay-HeLa cells were transfected with His-tagged wild type, K113R/K159R or R100T mutant. 48 h later, cells were pulsed for 30 min with 400 Ci ml Ϫ1 [ 35 S]methionine/cysteine, and then chased in regular medium supplemented with nonradioactive methionine/cysteine for varying time periods, as indicated in Fig. 2A. Cell lysates were harvested and subjected to Ni-NTA precipitation. The precipitated Smad4 proteins were analyzed on SDS-PAGE and visualized by autoradiography.
For ubiquitination assay, we also performed precipitation of Smad4 from lysates of the HeLa cells, which was co-transfected with HAtagged ubiquitin. Precipitated Smad4 were subjected to SDS-PAGE, and its ubiquitination was recognized by immunoblotting with anti-HA (ubiquitin) antibody.
RNA Interference-The target sequence of the Ubc9 siRNA was CAAAAAATCCCGATGGCAC (Dharmacon Research), which corresponds with nucleotides 86 -104 downstream from initiation codon of the human Ubc9 coding region. siRNA was transfected into HeLa cells using LipofectAMINE 2000 (Invitrogen), as described previously (29). For immunofluorescence, 48 h upon siRNA transfection, cells were treated with TGF-␤ for 1 h, fixed with ice-cold methanol, and immunostained with FITC-conjugated anti-Smad4 antibody (B8, Santa Cruz) to visualize the endogenous Smad4. For reporter assays, SBE-luc reporter plasmid was cotransfected with siRNA. 24 h later, cells were treated with TGF-␤ for 12 h and harvested for luciferase analysis.
For precipitation-Western analysis, HeLa cells were first transfected with plasmid DNA, as specified in the text. Twenty-four hours upon DNA transfection, the same cells were transfected with Ubc9 siRNA. Forty-eight h later, cells were harvested, and cell lysates were used to detect Ubc9 protein in anti-Ubc9 Western blot or further processed to analyze Smad4 SUMOylation.
Cell Proliferation-HeLa cells were first transfected with siRNA as described above. After 24 h, 1 ϫ 10 3 cells were then replated into each well of a 96-well flat-bottom plate. Upon attachment to the well (3 h), cells were treated with or without TGF-␤ for 2 or 3 days. Measurement of cell proliferation was then carried out using MTS (3-(4,5dimethyl-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2Htetrazolium)-based CellTiter96® Aqueous assay (Promega). Tetrazolium compound MTS was added to the cells, which were then returned to the cell culture incubator (37°C, 5%CO 2 ) for 1.5 h. The absorbance of cells was read at 590 nM in a microplate reader. The TGF-␤-induced inhibition of cell proliferation was calculated by the following formula:  Fig. 1, B and C, respectively. Smad4 NES mutant was exclusively localized in the nucleus (Fig. 1C), and its SUMOylation was enhanced (Fig. 1B, lane 6). In sharp contrast, conjugation of SUMO1 to the NLS mutant, which was primarily localized in the cytoplasm (Fig. 1C), could not be detected (Fig. 1B, lane 7). In addition, we compared the SUMOylation of wild type Smad4 in the presence or absence of nuclear export inhibitor leptomycin B. Leptomycin B treatment increased the level of SUMOylation of Smad4 (Fig. 1B, lane 4). Thus, our results suggest that Smad4 SUMOylation occurs in the nucleus. Previous studies also point to the nucleus as SUMOylation site (31), although SUMOylation may take place in the cytoplasm. Alternatively, SUMOylated Smad4 is more stable in the nucleus than in the cytoplasm.

SUMOylation of Smad4 Occurs in the Nucleus and Is Regulated by TGF-␤-Our
We noticed that TGF-␤ had stimulatory effects on SUMOylation of Smad4 (Fig. 1B, lane 2). We then determined how TGF-␤ treatment regulated SUMOylation of Smad4. Addition of 200 pM TGF-␤ to the culture medium increased the level of SUMO1 modification of Smad4 (Fig. 1D). After 4 -8-h treatment, the Smad4-SUMO conjugate reached the maximum level; at 12 h, the level dropped to basal or below basal level (Fig.  1D). Thus, the results support the notion that the increase in Smad4 SUMOylation could be attributed to the TGF-␤-induced translocation of Smad4 into the nucleus, where SUMOylation occurs.
SUMOylation Enhances TGF-␤-induced Nuclear Accumulation of Smad4 -One key regulation of Smad4 activity is its nuclear translocation in response to TGF-␤. However, Smad4 also has NES (see Fig. 1A), and thus, is frequently shuttled between the nucleus and cytoplasm (30). In our work, we examined whether SUMOylation influenced Smad4 subcellular accumulation in RI-14 cells. Results from immunofluorescence microscopy are shown in Fig. 1E. Since Smad4 is a shuttling protein, we divided the subcellular localization of Smad4 into two categories: exclusive nuclear (N) or non-exclusive nuclear staining (N/C ϩ C); the latter included cytoplasmic staining (C) or a mixed distribution between the nucleus and cytoplasm (N/C). In the absence of TGF-␤ stimulation, there were about 30% (63/213) of cells that expressed wild type Smad4 exclusively in the nucleus and 70% of cells in the cytoplasm or in both compartments. With TGF-␤ stimulation, 60% (106/177) of cells expressed Smad4 exclusively in the nucleus, while 40% were in the cytoplasm or both, supporting the notion that TGF-␤ causes a nuclear shift of Smad4 protein. We also examined the effect of SUMO1 expression on subcellular distribution of Smad4. In the absence of TGF-␤ stimulation, SUMO1 ex- erted a moderately stimulatory effects on Smad4 nuclear localization. In comparison with cells without exogenous SUMO1, 42% (45/107) of SUMO1-transfected cells contained Smad4 in the nucleus. More significantly, SUMO1 caused an increase of TGF-␤-induced nuclear accumulation of Smad4, with 80% (90/ 113) of cells exhibiting exclusively nuclear localization of Smad4. Analysis of cytoplasmic and nuclear fractions of Smad4 also gave similar results (data not shown). Since SUMO expression had no effect on Smad4 nuclear staining in the absence of TGF-␤, our results suggest that SUMO1 appear to increase Smad4 retention and/or stability in the nucleus.
Correlation between Loss of SUMOylation and Rapid Degradation of Cancer-derived Smad4 Mutant-Unlike ubiquitin, which promotes proteasome-dependent degradation of substrates, SUMO1 has not been linked to protein degradation. It has been suggested that SUMOylation may stabilize conjugated proteins by competing for the same lysine residues for ubiquitination (19). During the course of our SUMOylation experiments, we observed that the steady state level of Smad4 is low in the absence of SUMO1, but overexpression of SUMO1 can increase the steady state level of Smad4 (1).
To examine the metabolic stability of Smad4, we performed pulse-chase experiment. We compared the stability of wild type, K113R/K159R mutant, and a cancer-derived Smad4 mutant with Arg 3 Thr mutation (R100T) that has previously been shown to undergo faster degradation (11). The half-life of wild type Smad4 was found to be around 9 h, whereas the level of the K113/159R mutant remained 50% upon 24 h of chase ( Fig. 2A). The R100T mutant had only a half-life of 3 h. Similar results were obtained in cycloheximide chase experiments (data not shown). These results suggest that Lys 113 , Lys 159 , or both, could also serve as a negative element for regulating Smad4 degradation. Thus, the SUMOylation on Lys 113/159 residues or mutation of the same sites simply prevent Smad4 from degradation. This conclusion is consistent with the reduced ubiquitination on the K113R/K159R mutant ( Fig. 2B; Ref. 1) and increased ubiquitination on the R100T mutant ( Fig. 2B;  Ref. 11). This observation is reminiscent of IB, of which SUMOylation antagonizes ubiquitination through the lysine residue Lys 21 (19). Jab1, a component of COP9 signalosome, was found to promote Smad4 degradation (32). However, over- HeLa cells were similarly transfected as described in the legend to A, with the exception of the addition of expression plasmids for Smad4 (wild type (WT) or K113R/K159R (KR)). C, a working model for the function of Smad4 SUMOylation. In response to TGF-␤, Smad4 forms a complex with phosphorylated R-Smads (not shown) and is transported into the nucleus. Nuclear Smad4 undergoes SUMO-1 modification, which can stabilize the protein and perhaps inhibits its nuclear export. Although SUMOylation and de-SUMOylation are dynamic processes, constant SUMOylation of Smad4 increases its total level in the nucleus and consequently sustaining the activated Smad signaling. expression of Jab1 still induced the degradation of K113R/ K159R mutant (data not shown), suggesting that additional ubiquitin E3 ligases or factors (other than Jab1 pathway) are responsible for Smad4 degradation through the Lys 113 and/or Lys 159 residues.
Since our data suggest a role of SUMOylation in Smad4 stability, we set out to determine whether the R100T mutant might have a loss or reduction in SUMOylation. As shown in Fig. 2C, the R100T mutant failed to be SUMOylated, while the wild type was efficiently SUMOylated in HeLa cells (Fig. 2C), suggesting that the rapid proteolysis of this mutant could be responsible for its inability to be SUMOylated. In light of the observation that the R100T mutant has normal nuclear localization with Smad2 in response to TGF-␤ (11), its loss of SUMOylation may be attributed to a conformation change so that efficient SUMO modification cannot occur on the Lys 113/159 residues. As SUMOylation competes or blocks ubiquitination on Smad4, the lack of SUMOylation thus contributes to the enhanced ubiquitination and faster degradation of the R100T mutant (Fig. 2, A and B). More importantly, mutations of ubiquitination sites at Lys 113 and Lys 159 greatly enhanced the metabolic stability of R100T; the cycloheximide experiment clearly showed that the triple mutant (R100T/K113R/K159R) had a much longer half-life than the unstable R100T mutant (Fig. 2D). A similar observation was reported recently (33). Therefore, insufficient SUMOylation may serve as one mechanism for cells to escape TGF-␤ growth inhibitory control during cancer development.
Reduction in Smad4 SUMOylation Decreases the Steady State Level of Endogenous Smad4 -Taken together, data described above would suggest that SUMO modification plays a positive role in regulating Smad4 activity. Indeed, reporter assays demonstrated that overexpression of SUMO1 and Ubc9 stimulated TGF-␤-induced transcriptional responses (1). To test this more directly in a physiologically relevant context, we sought to assess the impacts of disrupting endogenous Ubc9 expression using RNAi. Since Ubc9 is the only E2 enzyme for modification by members of SUMO family, inhibition of its expression should disrupt SUMOylation of its substrates.
We used a 21-nucleotide siRNA that specifically targets to the human Ubc9 coding region and examined its effectiveness in silencing Ubc9 expression in HeLa cells. 48 h after transfection with the siRNA, HeLa cells expressed a significantly low level of endogenous Ubc9 protein, in comparison with the mocktransfection control (Figs. 3, A and B, and 4B). In contrast, expression of ␤-actin remained unchanged (Figs. 3, A and B,  and 4B). Treatment of HeLa cells with unrelated RNAi oligonucleotides did not reduce the level of Ubc9 protein (data not shown).
We then determined whether diminished Ubc9 expression could affect the SUMOylation of Smad4 and reverse the effects of SUMOylation on Smad4 stability. Our results showed that increasing doses of Ubc9 siRNA caused gradual reduction in the steady state level of SUMO-conjugated Smad4 in transfected HeLa cells (Fig. 3A). Accordingly, the steady state level of endogenous Smad4 decreased upon siRNA transfection (Fig.  3, B and C). Smad4 was present in both the nucleus and cytoplasm and displayed a diffused and weak staining (Fig.  3C). TGF-␤ treatment induced a concentration of Smad4 in the nucleus, causing a brighter image under microscope. Inhibition of Ubc9 expression with siRNA reversed the effect of TGF-␤ on Smad4 accumulation. Levels of Smad2 and Smad3 (controls) were the same with or without the siRNA (Fig. 3B).
Ubc9 Is Required for TGF-␤-induced Growth Inhibition and Endogenous Plasminogen Activator Inhibitor-1 (PAI-1) Expression-TGF-␤ exerts its responses by regulating gene transcrip-tion involved in growth regulation and extracellular matrix remodeling. We then investigated the consequence of Ubc9 gene silencing on TGF-␤ induced gene responses and cell proliferation. We first utilized the synthetic SBE-luc reporter (28). Treatment of 100 nM siRNA exhibited 70% reduction of TGF-␤-induced SBE-luc reporter expression in HeLa cells (Fig. 4A) as well as HepG2 cells (data not shown). As controls, the sense oligonucleotide of Ubc9 or unrelated GFP siRNA had no effects on SBE-luc activity (Fig. 4A).
Suppression of the Ubc9 expression also affected the induction of endogenous target gene products of TGF-␤ such as PAI-1. In HeLa cells, TGF-␤ induced the production of PAI-1 protein (Fig. 4B, compare lanes 1 and 4). Simultaneous treatment with 20 nM Ubc9 siRNA sufficed to abolish the PAI-1 induction (Fig. 4B, compare lanes 4 and 5). As control, the level of ␤-actin remained unchanged regardless of TGF-␤ and/or siRNA treatment.
TGF-␤ also inhibits proliferation of epithelial cells. Growth of HeLa cells can be inhibited by TGF-␤, achieving up to 30% growth inhibition (data not shown). Since HeLa cells can be readily transfected using siRNA, we assessed the growth inhibitory response in these cells in the presence or absence of Ubc9 siRNA. Addition of TGF-␤ (for 2 or 3 days) caused a 25% inhibition on the proliferation of HeLa cells treated with transfection reagent (Vehicle) (Fig. 4C). In sharp contrast, the growth inhibitory response of TGF-␤ was nearly abolished in Ubc9 siRNA-transfected cells, which exhibited only 4 -6% growth inhibition by TGF-␤. During the siUbc9 treatment period, HeLa cells displayed a normal proliferation status in comparison to non-siUbc9-treated cells in the absence of TGF-␤ stimulation (data not shown). Our results suggest that disruption of Ubc9 expression using RNAi can effectively disrupt cellular SUMOylation machinery and consequently inhibit the SUMO-mediated stimulatory effect on Smad4 activity in growth inhibition.
Increased Smad4 Expression Restores TGF-␤ Signaling in Ubc9-depleted Cells-To investigate further the critical role of Ubc9 in Smad4 functions, we determined whether overexpression of Ubc9 could rescue TGF-␤ signaling in Ubc9-depleted cells. Since mouse and human Ubc9 share identical amino acid sequences, but the selected siUbc9 target nucleotide sequence differs in the human and mouse Ubc9 genes, we reasoned that mouse Ubc9 could substitute for human counterpart without being silenced by siUbc9. As shown in Fig. 5A, overexpression of the mouse Ubc9 gene could rescue the loss of TGF-␤-induced SBE-luc reporter expression. Furthermore, if depletion of Ubc9 caused Smad4 instability and subsequent loss of TGF-␤ signaling, one might be able to rescue the loss of TGF-␤ response by overexpressing Smad4. To this end, we transfected HeLa cells with Smad4 or the Smad4(K113R/K159R) mutant. It is observed that expression of Smad4 or KR mutant can reverse the silencing effect of Ubc9 siRNA (Fig. 5B), strongly suggesting that the defective phenotype in TGF-␤ signaling caused by Ubc9 siRNA is not due to the indirect pleiotrophic effects of other SUMO targets. In summary, our current study elucidated the function of SUMOylation in TGF-␤ signaling. SUMOylation has positive effects on the growth inhibitory and transcriptional functions of Smad4, at least largely due to an increase in Smad4 stability. SUMOylation appears to compete with ubiquitination at Lys 113 and Lys 159 . Reduced SUMOylation by knocking down Ubc9 expression causes instability of the endogenous Smad4 and abolishes the growth inhibitory and transcriptional functions of Smad4. In consistent with this is the biochemical property of a cancer-derived mutant R100T, which fails to be modified by SUMO and undergoes faster degradation. Therefore, this supports the notion that SUMO-ylation may be an important gatekeeper that prevents Smad4 from degradation in normal cells. SUMOylation of Smad4 occurs in the nucleus. The stimulatory effects of Smad4 SUMOylation on TGF-␤ signaling may also be connected to an inhibition of its nuclear export by SUMOylation. The SUMOylation on Lys 113 and Lys 159 , which spans the NES (amino acids 143-148) of Smad4, may block the interaction of the NES with the export machinery. Fig. 5C illustrates our working model on the relationship of SUMOylation, stability, and nuclear retention of Smad4. Further delineation on the mechanism of Smad4 nuclear import and export will provide more insights into the functions of SUMOylation in Smad4-dependent signaling events.