Transforming Growth Factor-β-mediated Signaling via the p38 MAP Kinase Pathway Activates Smad-dependent Transcription through SUMO-1 Modification of Smad4*

Post-translational modifications such as ubiquitination, phosphorylation, and acetylation play important roles in the regulation of Smad-mediated functions. Here, we demonstrate that Smad4 is covalently modified by SUMO-1, which was characterized recently as a key modulator of many transcription factors. Sumoylation of Smad4 mainly occurs at lysine 159, located in the linker region, and facilitates Smad-dependent transcriptional activation. Furthermore, we show that the PIAS family proteins, PIAS1 and PIASxβ, function as E3 ligase factors for Smad4. Intriguingly, sumoylation of Smad4 was strongly enhanced by TGF-β-induced activation of the p38 MAP kinase pathway but not the Smad pathway. Activation of p38 not only stabilized PIASxβ protein but also enhanced PIASxβ gene expression, suggesting that PIAS-mediated sumoylation of Smad4 is regulated by the p38 MAP kinase pathway. These findings illustrate a novel regulatory mechanism by which Smad-dependent transcriptional activation cooperatively modulates Smad proteins through receptor-mediated phosphorylation and sumoylation.


Post-translational modifications such as ubiquitination, phosphorylation, and acetylation play important roles in the regulation of Smad-mediated functions.
Here, we demonstrate that Smad4 is covalently modified by SUMO-1, which was characterized recently as a key modulator of many transcription factors. Sumoylation of Smad4 mainly occurs at lysine 159, located in the linker region, and facilitates Smad-dependent transcriptional activation. Furthermore, we show that the PIAS family proteins, PIAS1 and PIASx␤, function as E3 ligase factors for Smad4. Intriguingly, sumoylation of Smad4 was strongly enhanced by TGF-␤-induced activation of the p38 MAP kinase pathway but not the Smad pathway. Activation of p38 not only stabilized PIASx␤ protein but also enhanced PIASx␤ gene expression, suggesting that PIAS-mediated sumoylation of Smad4 is regulated by the p38 MAP kinase pathway. These findings illustrate a novel regulatory mechanism by which Smad-dependent transcriptional activation cooperatively modulates Smad proteins through receptor-mediated phosphorylation and sumoylation.
Members of the transforming growth factor-␤ (TGF-␤) 1 superfamily regulate diverse biological functions including cell differentiation, growth inhibition, migration, survival, and apoptosis (1)(2)(3). The cellular effects of TGF-␤ are mediated by binding type I and type II serine/threonine kinase receptors. Upon ligand binding and activation, the type II receptor kinase phosphorylates the type I receptor kinase. The activated type I receptor then phosphorylates receptor-regulated Smads (R-Smad), Smad2 and Smad3, in the TGF-␤/activin pathways. Smads, the central molecules in TGF-␤ signaling, act as tran-scription factors or coactivators for regulating target gene expression (4). Receptor-mediated phosphorylation of R-Smad additionally creates a complex with the common-mediator Smad (Co-Smad; Smad4). These complexes translocate and accumulate in the nucleus where they are directly involved in the transcriptional regulation of various target genes (5,6).
TGF-␤ also activates members of the mitogen-activated protein (MAP) kinase family, including TGF-␤-activated kinase, c-Jun N-terminal kinase, and extracellular signal-regulated kinase (7)(8)(9)(10). TGF-␤-activated kinase then activates the stress-activated kinase p38, which plays an important role in regulating cellular processes such as inflammation, cell differentiation, and apoptosis (11)(12)(13)(14). A recent study (15) using a mutant type I receptor, which was incapable of activating the Smad pathway but still retained signaling via the MAP kinase pathway, showed that TGF-␤ receptor-activated p38 is involved in TGF-␤-induced apoptosis but not growth arrest in mouse mammary gland epithelial (NmuMG) cells (15). Thus, the p38 pathway seems to be sufficient to induce apoptosis in these cells, whereas a fully epithelial-to-mesenchymal transition response requires additional TGF-␤ signaling and possibly requires Smad-mediated transcriptional activation (16). These observations suggest that diverse biological responses regulated by TGF-␤ are mediated by different downstream signaling pathways, dependent on either Smad or MAP kinase or both.
Post-translational modifications regulate the function of many proteins. In the case of Smads, hetero-oligomeric formation by Smads is dependent on phosphorylation of R-Smads. Furthermore, recent findings have underscored critical functions of ubiquitination and acetylation in the control of Smaddependent gene regulation. Ubiquitination plays a role in regulating Smad function as Smads are rapidly targeted for degradation by the proteasome (17,18). The ultimate degradation of Smads after ligand stimulation has been firmly established as a mechanism to terminate Smad signaling. However, acetylation of Smad7 by coactivator p300, one of the inhibitory Smads (I-Smad), occurs on two lysine residues that are also targeted by ubiquitination and thereby prevents ubiquitin-mediated degradation (19).
To understand the molecular mechanisms of Smad transcriptional function, we explored the possible modification of Smads by SUMO-1 through post-translational modifications and attempted to define the biological significance of cross-talk between Smad activation and TGF-␤ signaling. Here, we report that Smad4 is a target for SUMO-1 modification both in vivo and in vitro. Additionally, we identified PIAS proteins as E3 ligases for sumoylation, which activate Smad4-dependent transcriptional function. Furthermore, the TGF-␤-mediated p38 MAP kinase pathway regulates PIAS gene expression and protein stabilization, which thereby enhances Smad4 sumoylation.

MATERIALS AND METHODS
Cell Culture, Transfections, and Luciferase Reporter Assay-Cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal calf serum, 100 units/ml of penicillin, and 100 g/ml of streptomycin. Transfections were performed using Fu-GENE 6 (Roche Diagnostics Corp.), according to the manufacturer's instructions. Luciferase activities were normalized to Renilla luciferase activities derived from cotransfected pRL-CMV-Luc (Promega). All reporter assays were performed in triplicate, and standard errors (S.E.) are denoted by the bars in figures.
Protein Purification and in Vitro Sumoylation Analysis-GST and GST fusion proteins were expressed in the Escherichia coli strain BL21 (DE3) and affinity purified with glutathione S-Sepharose beads according to the manufacturer's instructions (Amersham Biosciences). Expression and purification of GST-Sua1 and His-Uba2 in the baculovirus system were performed as described previously (44). For in vitro sumoylation assays, GST-fused Smad protein substrates were incubated with E1 (GST-Sua1/His-Uba2) and E2 (GST-Ubc9) enzymes and recombinant SUMO(GG) in reaction buffer consisting of 50 mM Tris-HCl (pH 8.0), 1 mM dithiothreitol, 3 mM MgCl 2 , and 5 mM ATP at 30°C for 30 min. For Western blot analysis, proteins were fractionated by SDS-PAGE and electroblotted onto an Immobilon-P membrane (Millipore). The blot was incubated sequentially with mouse anti-Myc antibody and horseradish peroxidase-conjugated anti-mouse IgG and detected using enhanced chemiluminescence (ECL; Amersham Biosciences).
Immunoprecipitations-COS7 cells (1 ϫ 10 5 per 6-cm-diameter dish) were transfected using FuGENE 6 according to the manufacturer's instructions. After incubation, cells were lysed in 1 ml of RIPA buffer (25 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% Nonidet P-40, 0.1% SDS, 0.5% sodium deoxycholate, 1 mM dithiothreitol, 5 mM EDTA, 10 mM N-ethylmaleimide, 200 M indole-3-acetic acid, and a complete protease inhibitor mixture tablet (Roche Applied Science)) for 30 min on ice. Cell debris was removed by centrifugation for 15 min. Lysates were first cleared with protein G beads for 30 min, followed by incubation with antibodies for 1 h at 4°C. Finally, the antibody complexes were captured with protein G beads for 1 h. Beads were washed four times with the same buffer, and immunoprecipitates were eluted and analyzed by Western blot.
Semiquantitative RT-PCR Analysis-Total RNA was extracted using ISOGEN (Nippon Gene) according to the manufacturer's instructions. First-strand cDNA was synthesized with Superscript II (Invitrogen) using 2.5 g of total RNA and an oligo(dT) primer in a 20-l reaction volume. PCR was performed using 1 l of the reaction mixture for cDNA synthesis in a 10-l reaction volume of PCR buffer containing 200 M of each dNTP, 2 M of each set of primer, and 0.2 units of AmpliTaq at 94°C for 30 s, 55°C for 30 s, and 72°C for 1 min. The sequences of the primers used are mouse PIASx␤, 5Ј-AGAGAGCTCT-TCTGATGAAGAGG-3Ј (forward) and 5Ј-TCTCACTCCTGCTGCTG-GATGAAC-3Ј (reverse); SP6, 5Ј-ATTTAGGTGACACTATAGAATAG-3Ј (reverse); human GAPDH, 5Ј-ATG GGGAAGGTGAAGGTCGG-3Ј (forward) and 5Ј-TGGAGGGATCTCGCTCCTGG-3Ј (reverse). An aliquot of PCR products was electrophoresed on 2% agarose gels containing ethidium bromide and quantitated by UV-illumination.
Pulse-Chase Analysis-COS7 cells were plated to 50% confluency in 6-well dishes, 1 day before transfection. Approximately 24 -30 h posttransfection, cells were cultured with methionine/cysteine-deficient Dulbecco's modified Eagle's medium (ICN) for 30 min and then metabolically labeled with 0.1 mCi of [ 35 S]methionine/cysteine (ICN) in Dulbecco's modified Eagle's medium, supplemented with 10% fetal calf serum for 1 h. After washing with phosphate-buffered saline, cells were chased for the indicated time intervals in complete medium. Cells were then lysed in RIPA buffer as described above, and soluble extracts were subjected to immunoprecipitation with anti-FLAG antibody. Immunoprecipitates were analyzed by SDS-PAGE followed by autoradiography, and the relative intensity of each band was quantitated by scanning densitometry.

SUMO-1 Modification of Smad4 at Lys-159 in Vivo and in
Vitro-SUMO-1 modification of certain transcriptional regulators is stimulated by extracellular environmental insults in-cluding cytokines, hormones, heat shock, and irradiation (31,(45)(46)(47), and this modification regulates transcriptional function. We investigated whether the SUMO-1 modification system acts on the TGF-␤ signaling pathway. We first examined whether the transcription factors Smad2, Smad3, and Smad4, which are elements of the TGF-␤-dependent pathway, are modified by SUMO-1 in TGF-␤-treated NIH3T3 cells transiently expressing FLAG-Smads and HA-SUMO-1. Western blot analysis using anti-FLAG antibody revealed production of every Smad form in the cell lysates. However, in cells producing ectopic Smad4, an additional more slowly migrating band was observed when cells coexpressed HA-SUMO-1 (Fig. 1A, lane 6). This more slowly migrating band was not detected in cells producing ectopic Smad2 and Smad3 (lanes 1-4), suggesting that Smad4 may be specifically targeted for SUMO-1 modification.
To further examine whether Smad4 is indeed a substrate for SUMO-1 modification, we performed an in vitro sumoylation assay using recombinant Smad4 as a substrate. Smad4 has a consensus sumoylation sequence, VKDE, which is conserved among other species (Fig. 1B). We speculated that the Lys-159 in human Smad4 is a likely target for sumoylation. To assess this hypothesis, mutant Smad4, in which lysine 159 was converted to Arg, and wild-type Smad4 were prepared as GST-Myc-Smad4(K/R) and GST-Myc-Smad4 fusion proteins, respectively. These Smad4 proteins were incubated with various combinations of recombinant E1 and E2, together with GST-SUMO(GG) or GFP-SUMO(GG), as shown in Fig. 1B. Analysis of the reaction products by Western blot with anti-Myc antibody revealed a more slowly migrating Smad4 band only when E1 and E2, as well as GST-SUMO(GG), were present. However, the reaction containing Smad4(K/R) did not give rise to the slower migrating form (lane 5), indicating that Lys-159 is a target for modification. To confirm that this slower migrating band was indeed sumoylated Smad4, GFP-SUMO(GG) was substituted for GST-SUMO(GG) in the reaction. As expected, a slightly higher molecular weight form of Smad4 was detected, reflecting the difference in molecular weight of GFP-SUMO(GG) and GST-SUMO(GG) (lanes 6 and 7). These data strongly suggest that Smad4 is a substrate for SUMO-1 and that Lys-159 is a major SUMO-1 conjugation site.
To evaluate the state of Smad4 sumoylation in response to TGF-␤ signaling, expression plasmids containing FLAG-Smad4, FLAG-Smad4(K/R), Myc-Smad2, HA-SUMO-1, and the constitutively active form of the TGF-␤ type-I receptor, T␤R-I(T/D), were cotransfected in various combinations into COS7 cells, as shown in Fig. 1C. Western blotting with anti-FLAG antibody revealed the presence of Smad4 in all cells transfected with plasmid expressing FLAG-Smad4 (Fig. 1C, top panel). When SUMO-1 was also expressed, a higher molecular weight form of Smad4 was detected, along with a substantial amount of unmodified Smad4 (Fig. 1C, top panel, lanes 4 -7, white arrowhead). This form was not present in cells expressing Smad4(K/R) (lanes 8 and 9). Moreover, to determine whether this higher molecular weight form representing Smad4 was conjugated to SUMO-1, cell extracts were subjected to immunoprecipitation with anti-FLAG antibody and analyzed by Western blot using either anti-FLAG or anti-HA antibodies. The slower migrating form of Smad4 (about 90 kDa) was detected when SUMO-1 was coproduced (Fig. 1C, second panel from top, lanes 4 -7, white arrowhead). The filter was then reprobed with anti-HA antibody, demonstrating that the slower migrating form of Smad4 (about 90 kDa) was indeed the sumoylated form (Fig. 1C, third panel from top, white arrowhead). Several SUMO-1-reactive bands migrating slower than SUMO-1-conjugated Smad4 were observed in cells producing Smad4 but not in cells producing Smad4(K/R) (Fig. 1C, third  panel from top, lanes 4 -7). These bands were not likely to be sumoylated forms of Smad4,because the anti-FLAG antibody did not recognize them (Fig. 1C, second panel from top). Intriguingly, the amount of sumoylated Smad4 was increased when a T␤R-I(T/D) expression plasmid was cotransfected, and this was not changed by expressing ectopic Smad2 (Fig. 1C, top three panels, lanes 5 and 7), suggesting that TGF-␤ signaling plays a role in controlling Smad4 sumoylation, possibly through a pathway independent of TGF-␤-mediated Smad activation.
Smad4 is a common mediator for members of the TGF-␤ superfamily, including TGF-␤, activins, and BMPs (3). To investigate whether SUMO-1 modification of Smad4 has any effect on the TGF-␤ superfamily-mediated Smad pathway, we examined the effects of wild-type and mutant Smad4(K/R) on expression of a p3TP-Luc reporter gene, which contains a Smad-responsive element in its promoter. HepG2 cells were cotransfected with the reporter plasmid, together with wildtype or Smad4(K/R) expression plasmids, and then treated with TGF-␤ or BMP-4. Reporter activity was enhanced by treatment with TGF-␤ or BMP-4, even in cells not expressing ectopic Smad4 (Fig. 1D). Cells ectopically expressing wild-type Smad4 showed higher reporter gene activity than those expressing Smad4(K/R). This suggests that SUMO-1 modification of Smad4 plays a role in activation of the transcriptional response by the TGF-␤ superfamily.
PIAS Family Proteins Act as E3 Ligases for Smad4 Sumoylation-Recently, two distinct SUMO-1 E3 ligases, PIAS family proteins and RanBP2, have been identified (35)(36)(37). PIAS family proteins act as either positive or negative regulators for many transcriptional factors or cofactors (48). Therefore, we investigated whether PIAS family proteins function as E3 ligases for Smad4. We first examined the effect of PIASx␤ and PIAS1 expression on SUMO-1 modification of Smad4 in COS7 cells. Sumoylation of Smad4 was enhanced by both PIASx␤ and PIAS1 ( Fig. 2A, lanes 3 and 4), indicating that these PIAS proteins targeted Smad4 for SUMO-1 modification. PIAS family proteins also sumoylate p53 (49), androgen receptor (50), and cytomegalovirus IE2 protein (51). Based on the data here, Smad4 is also likely to be a target for PIAS family proteins.
To further clarify the importance of PIAS protein enzymatic activity on Smad4 sumoylation, we generated mutant PIASx␤, PIASx␤(C/S), in which the conserved cysteine residue at position 353 within the RING finger domain has been changed to serine. This mutant completely lacks E3 ligase activity. Either mutant or wild-type PIASx␤ were produced in cells, and Smad4 sumoylation was analyzed. Small amounts of Smad4 were covalently conjugated upon expression of SUMO-1, even in the absence of ectopic PIASx␤ expression (Fig. 2B, lane 5, top panel), but Smad4 sumoylation was enhanced by ectopic expression of PIASx␤ (lane 6). Sumoylated Smad4 was scarcely detected when SUMO-1 was not ectopically produced, suggesting a limiting amount of free endogenous SUMO-1 in COS7 cells. Immunoprecipitation with anti-FLAG antibody followed by Western blotting with the same antibody revealed the slower migrating form of Smad4 (Fig. 2B, lane 6, second panel from top). Reprobing this filter with anti-HA antibody, after stripping the anti-FLAG antibody, showed slower migrating bands, one of which coincided with the band observed in the second panel and thus was likely to be SUMO-1-conjugated Smad4. We also observed a band migrating even more slowly than sumoylated Smad4 (Fig. 2B, white arrowhead, third panel from top). Because this band was not detected with anti-FLAG antibody, we believe this represents sumoylated cellular protein(s) that coimmunoprecipitated with wild-type and/or sumoylated Smad4 as described above. In contrast, Smad4 sumoylation was not observed in cells expressing PIASx␤(C/S) (Fig. 2B, lane 7).
To confirm that PIASx␤ functions as an E3 ligase factor for Smad4, we performed in vitro sumoylation assays. Recombinant GST-Myc-Smad4 was incubated with limited amounts of E1 and E2 (10 ng of E1 and 5 ng of E2 per 20-l reaction volume) in the presence or absence of bacterially expressed GST-PIASx␤ or GST-PIASx␤(C/S) and analyzed by Western blotting using anti-Myc antibody. Small amounts of sumoylated Smad4 were detected in the reaction without PIASx␤ addition (Fig. 2C, lane 1). As expect, addition of GST-PIASx␤ (lane 2), but not GST-PIASx␤(C/S) (lane 3), enhanced Smad4 sumoylation. Taken together, these results clearly indicate that PIAS family proteins, specifically via their RING finger domains, function as E3 ligases for Smad4, both in vivo and in vitro.
SUMO-1 Conjugation to Smad4 Positively Regulates the Expression of Genes Downstream of TGF-␤-We showed previously that wild-type Smad4 has higher transcriptional activity than Smad4(K/R) on a reporter gene that responds to TGF-␤ and BMP-4 (Fig. 1). Thus, we speculated that SUMO-1 modification of Smad4 enhances Smad-mediated transcription. To investigate the physiological significance of SUMO-1 conjugation of Smad4 on TGF-␤ signaling, we analyzed the effects of Smad4 sumoylation on TGF-␤-responsive transcription. For this purpose, we used cells expressing T␤R-I(T/D), a constitutively active form of the TGF-␤ type I receptor, which mimics TGF-␤ signaling. COS7 cells were transfected with various combinations of plasmids expressing Smad2, Smad4, Smad4 (K/R), SUMO-1, PIASx␤, and PIASx␤(C/S), together with p3TP-Luc and a plasmid expressing T␤R-I(T/D) (Fig. 3A). Coproduction of SUMO-1 and Smad2/4 increased the activity by about 1.5-fold relative to cells producing Smad2/4. Additional production of PIASx␤ increased the activity about 3-fold relative to cells producing Smad2/4 alone. However, expression of PIASx␤(C/S) did not confer any further activation. In contrast, cells producing Smad2/Smad4(K/R) did not show significant differences in luciferase activity, even when they coproduced SUMO-1 and PIASx␤ or PIASx␤(C/S). These results suggest that Smad4 sumoylation by PIASx␤, in a RING finger domaindependent manner, enhances Smad-dependent signaling. Similar data were observed using pSBE-Luc as a reporter plasmid, which contains four Smad binding elements in its promoter (data not shown).

Activation of p38 MAP Kinase by TGF-␤-Signaling Enhances SUMO-1 Conjugation to Smad4 -We observed that Smad4 sumoylation was enhanced by ectopic expression of T␤R-I(T/D).
To verify the involvement of TGF-␤ signaling, we carried out experiments to dissect the role of downstream signaling through TGF-␤. As in a previous report (15), we generated a combined mutant type I receptor, T␤R-ImL45(T/D), which has a constitutively active kinase domain but lacks the ability to phosphorylate Smads. This mutant molecule provides a useful tool for dissecting the molecular mechanisms underlying different TGF-␤ intracellular signaling pathways. In fact, in contrast to T␤R-I(T/D), T␤R-ImL45(T/D) was unable to reconstitute Smad-mediated transcription because of defective Smad2 phosphorylation but was able to activate p38 MAP kinase in COS7 cells (Fig. 4A, lanes 2 and 3), in agreement with a previous report (15). We examined the requirement of downstream elements of the TGF-␤ signaling pathway in Smad4 sumoylation by using these mutant receptors. Plasmids expressing FLAG-Smad4, HA-SUMO-1, T␤R-I(T/D), and T␤R-ImL45(T/D) were transfected in various combinations into COS7 cells, and Smad4 sumoylation was examined. Sumoylation was not detectable in cells lacking ectopic SUMO-1 (Fig.  4B, lanes 1, 4, and 5) but became detectable in cells coexpressing SUMO-1 (lane 6). Smad4 sumoylation was enhanced by coexpression of T␤R-I(T/D) or T␤R-ImL45(T/D) (Fig. 4B, lanes  7 and 8). These results indicate that TGF-␤-dependent Smad activation is not required for Smad4 sumoylation. Interestingly, the amount of sumoylated Smad4 was higher in cells producing T␤R-ImL45(T/D) than in those producing T␤R-I(T/ D), which is likely related to higher p38 phosphorylation activity in T␤R-ImL45(T/D) (Fig. 4A). This observation led us to speculate that p38 may play a role in activation of SUMO-1 conjugation to Smad4.
We examined the effect of p38 on Smad4 sumoylation by using an inhibitor of p38. COS7 cells plated in duplicate were transfected with plasmids expressing FLAG-Smad4, HA-SUMO-1, GFP-PIASx␤, and T␤R-ImL45(T/D) in various combinations as shown in Fig. 4C. One set was then treated with the p38 inhibitor, SB203580, for 24 h before cell harvest. In the other set of cells, not treated with inhibitor, phosphorylated p38 was detected at varying levels, with highest amounts in lysates producing T␤R-ImL45(T/D) (Fig. 4C, lanes 4 -6, fourth  panel from the top). SUMO-1-conjugated Smad4 was detected in cells producing SUMO-1, SUMO-1 plus PIASx␤, SUMO-1 plus T␤R-I(T/D), SUMO-1 plus T␤R-ImL45(T/D), or SUMO-1, PIASx␤, and T␤R-ImL45(T/D), although the level of sumoylated Smad4 varied. Levels were highest in the lysate coproducing SUMO-1, PIASx␤, and T␤R-ImL45(T/D) (Fig. 4C, lane  6). In addition to sumoylated Smad4, several bands reactive to the anti-HA antibody were detected, which may correspond to cellular protein(s) interacting with wild-type and/or sumoylated Smad4 (Fig. 4C, white arrowhead, third panel from the  top). In lysates prepared from cells treated with SB203580, the amount of SUMO-1-conjugated Smad4 was lower than in cells not treated with SB203580. In particular, levels of SUMO-1conjugated Smad4 were dramatically enhanced when both T␤R-ImL45(T/D) and GFP-PIASx␤ were expressed together (lane 6), whereas SB203580 treatment of cells coproducing T␤R-ImL45(T/D) and PIASx␤ were reduced to nearly one-quar- ter that in untreated lysates (lane 6 versus 12). Taken together, these findings suggest that activation of p38 MAP kinase, but not Smads, by TGF-␤ signaling enhances PIAS-mediated Smad4 sumoylation.
We further examined the effect of SB203580 on Smad-mediated transcription in TGF-␤ signaling. NIH3T3 cells were transfected with a p3TP-Luc reporter plasmid and were treated with either TGF-␤ or SB203580, or both (Fig. 4D, left). Treatment with SB203580 significantly repressed Smad-mediated transcription. Furthermore, we performed the reporter assay with various combinations of T␤R-I(T/D), Smad2 and Smad4, SUMO-1, and PI-ASx␤ (Fig. 4D, right). COS7 cells were then treated with or without 30 M SB203580 for 12 h, 24 h after transfection, and luciferase activities were measured. As expected, luciferase activity was inhibited by treatment with SB203580 (black bars). Collectively, these findings suggest that enhanced Smad4 sumoylation by TGF-␤ signaling is regulated by modulating PIAS functions through p38 MAP kinase activation, which thereby promotes Smad-mediated transactivation.
PIASx␤ Protein Is Stabilized, and Its mRNA Is Up-regulated by p38 MAP Kinase Activation-The preceding data suggested that activation of p38 MAP kinase by TGF-␤ signaling might promote the function of PIAS family proteins and subsequently enhance SUMO-1 conjugation to Smad4. To investigate the molecular link between p38 MAP kinase and PIAS function, we first examined the effect of TGF-␤-mediated MAP kinase acti-  Fig. 1. Phospho-p38 (P-p38) and phospho-Smad2 (P-Smad2) were analyzed by using specific antibodies. The levels of proteins expressed in whole cell lysates were analyzed and shown as indicated. B, Smad activation pathway in TGF-␤ signaling is not required for sumoylation of Smad4. COS7 cells were cotransfected with (ϩ) or without (Ϫ) plasmids expressing FLAG-Smad4 (3 g), HA-SUMO-1 (1 g), or T␤R-I(T/D) or T␤R-ImL45(T/D) (3 g), respectively. A plasmid expressing Myc-Smad2 was also transfected into cells in all dishes. Lysates prepared from cells 36 h after transfection had their total protein concentration adjusted to the same level. Then, lysates were immunoprecipitated (IP) with anti-FLAG antibody followed by Western blotting with anti-FLAG antibody. The relative sumoylated ratio of Smad4 was indicated as a percent of total Smad4, based on densitometric quantitation. (Ϫ) shows below the detection limit. The protein levels of Smad2 and ␣-tubulin in each lysate were shown as indicated. The asterisk indicates the immunoglobulin heavy chain. C, treatment with a p38 MAP kinase inhibitor reduces Smad4 sumoylation. COS7 cells were cotransfected with (ϩ) or without (Ϫ) plasmids expressing FLAG-Smad4 (3 g), HA-Smad4 (1 g), GFP-PIASx␤ (2 g), and T␤R-ImL45(T/D) (3 g), respectively. Twelve h after transfection, cells were treated with or without 30 M SB203580 for 24 h, and cell lysates were prepared. Immunoprecipitation and Western blot analysis were performed as shown in Fig. 1C. The levels of protein in whole cell lysates are shown as indicated. Sumoylated cellular protein(s) coimmunoprecipitated with Smad4 are indicated as a white arrowhead. D, inhibition of p38 MAP kinase activity suppresses Smad-mediated transactivation. Left, NIH3T3 cells were transfected with 50 ng of p3TP-Luc. Twenty-four h after transfection, cells were treated with solubilized solution without TGF-␤ (control; cont.), 2 ng/ml TGF-␤, 30 M SB203580 (SB), and TGF-␤ plus SB203580 (TGF-␤ ϩ SB). Luciferase activities in cell lysates prepared after 12 h of treatment were measured. The activity of the treatment with control medium was arbitrarily given a value of 1, and the activities of the other transfections were adjusted relative to this assay. Right, COS7 cells were cotransfected with 25 ng of p3TP-Luc and 50 ng of plasmid expressing T␤R-I(T/D) in the presence (ϩ) or absence (Ϫ) of 50 ng of plasmids expressing FLAG-Smad2, together with FLAG-Smad4, and HA-SUMO-1, and 100 ng of FLAG-PIASx␤. Twenty-four h after transfection, cells were treated with (black bars) or without (white bars) 30 M SB203580 for 12 h, and luciferase assays were performed. The activity of the reporter plasmid alone was arbitrarily given a value of 1, and relative activities were calculated. vation on PIASx␤ production. COS7 cells were cotransfected with plasmids expressing FLAG-PIASx␤ in combination with HA-SUMO-1 and T␤R-ImL45(T/D). Lysates prepared from these cells were then subjected to Western blotting using anti-FLAG antibody (Fig. 5A). Multiple bands reactive to the anti-FLAG antibody, which migrate slower than PIASx␤ and seem to be sumoylated forms of PIASx␤, were noted, in agreement with previous reports (49,52). The amount of sumoylated, as well as unmodified, PIASx␤ was significantly increased when cells were transfected with T␤R-ImL45(T/D) (lane 2, 4), whereas a weak induction in response to SUMO-1 expression was observed (lane 3), because the amount of the transcript was the same in each case from the plasmid expressing FLAG-PIASx␤ (Fig. 5A, lower panel). This suggested that activation of the MAP kinase pathway by TGF-␤-signaling contributes to PIASx␤ protein stability.
To further examine whether p38 MAP kinase is involved in increasing PIASx␤ levels, COS7 cells were transfected with FIG. 5. Activation of p38 enhances PIASx␤ mRNA expression and protein stability. A, accumulation of PIASx␤ by activation of the MAP kinase pathway in TGF-␤ signaling. COS7 cells were transfected with 3 g of plasmid expressing FLAG-PIASx␤, together with (ϩ) or without (Ϫ) 1 and 3 g of plasmids expressing HA-SUMO-1 and T␤R-ImL45(T/D), respectively. Cell lysate was prepared 36 h after transfection, and equivalent amounts of each lysate were analyzed by Western blotting with anti-FLAG antibody. The levels of mRNA expressed from the FLAG-PIASx␤ expression plasmid were analyzed by semi-quantitative RT-PCR using a forward primer in the PIASx␤ coding region and a reverse primer in the region downstream of the multiple cloning site in the pcDNA3 vector (SP6). The signals for FLAG-PIASx␤ at 12 cycles of PCR and GAPDH at 15 cycles of PCR are shown. A control lacking RT is shown (Ϫ). The levels of mRNA for GAPDH are shown as an internal control. B, the activation of p38 MAP kinase increased PIASx␤ protein levels. COS7 cells were cotransfected with (ϩ) or without (Ϫ) 2 g of plasmids expressing FLAG-PIASx␤, MKK6(DE), a constitutively active form of MKK6, and p38-HA, respectively. Cell lysates were prepared 36 h after transfection, and equivalent amounts of each lysate were immunoprecipitated with anti-FLAG antibody, followed by Western blotting using anti-FLAG antibody. Phospho-p38 (P-p38) and phospho-Elk-1 (P-Elk-1) were analyzed using specific antibodies. The asterisk indicates the immunoglobulin heavy chain. C, effect of p38 inhibitor on degradation of PIASx␤. COS7 cells were cotransfected with 2 g of plasmid expressing FLAG-PIASx␤, together with or without 3 g of plasmid expressing T␤R-ImL45(T/D). Twenty-four h after transfection, cells were treated with (ϩ) or without (Ϫ) 30 M SB203580. After an additional 12 h of incubation, cells were metabolically labeled with [ 35 S]methionine/cysteine for 1 h. Cells were then chased for 0, 1, 3, and 5 h with unlabeled methionine/cysteine. Cells were lysed, immunoprecipitated with anti-FLAG antibody, and analyzed by autoradiography. The amount of PIASx␤ was quantified by densitometry and plotted as a percent of the zero time values under each condition. D, p38 MAP kinase up-regulates the mRNA level of endogenous PIASx␤. NIH3T3 cells were transfected with (ϩ) or without (Ϫ) 5 g of plasmid expressing T␤R-ImL45(T/D). Twenty-four h after transfection, cells were treated with (ϩ) or without (Ϫ) 30 M SB203580 for 12 h. Total RNA was prepared, and mRNA levels of the endogenous PIASx␤ were determined by semi-quantitative RT-PCR using the set of primers described under "Materials and Methods." The signals for PIASx␤ at 27 cycles of PCR and GAPDH at 15 cycles of PCR are shown. The levels of mRNA for GAPDH are shown as an internal control. plasmids expressing MKK6(DE), a constitutively active form of MKK6 that has the ability to phosphorylate p38, p38-HA, and FLAG-PIASx␤. Cell lysates obtained from these cells were subjected to Western blotting using anti-FLAG antibody (Fig. 5B). Coproduction of MKK6(DE) and p38-HA dramatically enhanced the level of PIASx␤ protein, indicating that p38 MAP kinase activation is likely to be a critical determinant of PI-ASx␤ levels in cells. Cell lysates were subjected to Western blotting using anti-phospho-p38 and anti-phospho-Elk-1 antibodies to verify the function of MKK6(DE) and p38-HA (Fig.  5B). Based on these observations, we hypothesized that activation of p38 MAP kinase by TGF-␤ signaling is important for PIASx␤ stability. To address this possibility, we performed pulse-chase analysis to determine the half-life of PIASx␤, together with or without mL45(T/D), in the presence or absence of SB203580. PIASx␤ was relatively stable upon coexpression of mL45(T/D) in the absence of SB203580 but was rapidly degraded upon treatment with SB203580 (Fig. 5C). These results suggest that PIASx␤ is stabilized by TGF-␤-mediated activation of p38 MAP kinase. Next, to determine whether p38 MAP kinase is involved in regulating the expression of endogenous PIASx␤, T␤R-ImL45(T/D) was produced in NIH3T3 cells, and cells were analyzed with or without SB203580. mRNA levels of PIASx␤ were then analyzed by semi-quantitative RT-PCR (Fig. 5D). Interestingly, expression of PIASx␤ was up-regulated by expression of T␤R-ImL45(T/D) in the absence of SB203580 (compare lane 1 versus lane 2 in Fig. 5D) and repressed by treatment with SB203580 (lanes 3 and 4). These results indicate that activation of p38 MAP kinase not only enhances PIASx␤ protein stability but also up-regulates the expression of endogenous PIASx␤ mRNA. DISCUSSION In this study, we found that SUMO-1 modification of Smad4 is a novel mechanism of gene regulation in TGF-␤ signaling. Smad4 is predominantly modified by SUMO-1 at Lys-159, within the linker region between MH1 and MH2, in vivo and in vitro. We identified PIAS1 and PIASx␤ as E3 ligase factors for SUMO-1 conjugation to Smad4. Our observations in luciferase reporter assays indicate that PIASx␤ activated SUMO-1 conjugation to Smad4 in a RING finger domain-dependent manner and that this modification is necessary for full activation in response to PIAS-mediated transactivation. On the other hand, the transcriptional activity of the sumoylation-deficient mutant Smad4(K/R) was weakly potentiated by expressing wildtype PIASx␤ but not the RING finger mutant (Fig. 3), suggesting that PIASx␤ functions to not only sumoylate Smad4 but also to conjugate SUMO-1 to other cellular factor(s) involved in Smad-mediated transactivation. Thus, SUMO-1 conjugation by PIASx␤ at positions other than Lys-159 on Smad4 may also play a role in transcriptional activation. Further studies to clarify the molecular basis of transcriptional activation of Smad4dependent transcription by PIASx␤ are warranted. During review of this manuscript, two reports showed that Smad4 was sumoylated at multiple sites and that Smad-mediated transactivation was up-regulated by the modification (53,54). However, we could not clearly detect an additional sumoylated form of Smad4(K/R) in COS7 cells and could not to determine whether sumoylation site(s) of Smad4 is only Lys-159 or not. Although we do not know the reasons for differences in the results at present, they may be because of cell line variation and/or expression levels of SUMO-1.
Our data clearly show that sumoylation of Smad4, despite affecting a relatively small proportion of Smad4, had a significant function in TGF-␤-mediated gene expression. SUMO-1 modification is reversed by specific proteases that were still enzymatically active in cell lysates prepared for analysis of SUMO-1-conjugated Smad4. Supplementation of lysates with the SUMO-1 protease inhibitor, N-ethylmaleimide, often restored the level of SUMO-1 conjugation, suggesting that the level of sumoylated Smad4 in these studies is underrepresented and that the dynamics of equilibrium between sumoylation and de-sumoylation of Smad4 may regulate Smad-dependent transcription in vivo. Alternatively, SUMO-1-conjugated Smad4 may recruit the transcriptional coactivator(s) complex by providing novel interaction sites. Recent studies showed that the linker region adjacent to the SUMO-1 conjugation site of Smad4, known as the Smad4 activation domain (SAD), is required for activation of Smad4-dependent signaling responses and plays a role in recruiting the transcriptional coactivator CBP/p300 (55,56). Our findings have revealed that sumoylated cellular protein(s) were coimmunoprecipitated with Smad4, but not Smad4(K/R), from cells expressing ectopic SUMO-1 (Fig. 1C). SUMO-1-conjugated Smad4 may recruit additional cellular factors, in addition to CBP/p300, which enhance Smad4-dependent transcription. Further identification and characterization of these protein(s) should provide significant insight into the transactivation of Smad4 by SUMO-1 modification.
Certain types of cytokine/receptor signaling can activate downstream elements using plural pathways (57). For example, type-I IFNs, such as IFN-␣, IFN-␤, and IFN-␥, that activate the Jak/Stat pathway, also referred to as the canonical IFNs pathway, also activate the non-canonical IFNs pathway including p38 MAP kinase. The TGF-␤ superfamily also activates two signaling pathways, the Smad family and MAP kinase cascade, that are referred to as canonical and non-canonical, respectively, to regulate its downstream genes. In this study, we demonstrated that the non-canonical pathway, p38 MAP kinase activation, activated by TGF-␤ signaling, played an important role in Smad4 sumoylation. We found enhanced PIASx␤ gene expression and protein accumulation following activation of p38 MAP kinase. An inhibitor of p38, SB203580, inhibited Smad4 sumoylation and, simultaneously, suppressed transcriptional enhancement by sumoylated Smad4. Accumulation of PIASx␤ by activation of p38 MAK kinase seemed to be the result of prolonged half-life of this molecule and of increased expression of this gene. p38 has been shown to activate transcription factors by phosphorylation (58). These factor(s) are likely to mediate increased PIASx␤ expression at the transcriptional level. Because stabilization of PIASx␤ was not correlated with sumoylation (Fig. 5C), it is likely that phosphorylation of PIASx␤ by p38 itself or a kinase activated by p38 may contribute to its stability. The molecular mechanisms of the prolonged half-life of PIASx␤ by p38 activation need to be clarified.
Association of Smad4 sumoylation with canonical and noncanonical pathways of TGF-␤ signaling may be an efficient way to activate downstream Smad signaling. Previous studies have shown that transcriptional activation of target genes by TGF-␤, including the collagenase-3 and biglycan genes, requires the activation of both the MAP kinase and Smad pathways (59,60). One explanation for this cross-talk between Smad and MAP kinase pathways is that activated p38 is directly involved in enhancing transcription via Smad4 sumoylation by PIAS family proteins. It has also been shown that Smad4 functions as a transcriptional cofactor for transcription factors such as AP-1 (61) and nuclear hormone receptors (1,62,63), in addition to functioning in a heterodimeric form with the R-Smad-dependent transcription factor, a well characterized signaling pathway of the TGF-␤ superfamily. In this case, Smad4 may also contribute to its own transcriptional regulation by sumoylation through a similar mechanism to that described in this paper. In this regard, cross-talk between two independent signaling pathways, one of which includes p38 MAP kinase activation, may play an important role in regulating gene expression through sumoylation of transcriptional or related factors.