CHIP controls the sensitivity of transforming growth factor-beta signaling by modulating the basal level of Smad3 through ubiquitin-mediated degradation.

Transforming growth factor-beta (TGF-beta) signaling is critical in a variety of biological processes such as cell proliferation, differentiation, and apoptosis. TGF-beta signaling is mediated by a group of proteins including TGF-beta receptors and Smads. It is known that different cells can exhibit different sensitivities to TGF-beta. Several molecular mechanisms, such as the differential expression of the receptor levels, have been suggested as contributing to these differences. Here, we report evidence for a novel mechanism of regulating TGF-beta sensitivity that depends on the role of CHIP (carboxyl terminus of Hsc70-interacting protein) in regulating the basal level of Smad3 via the ubiquitin-dependent degradation pathway. First, using a luciferase assay we found that overexpression of CHIP inhibited TGF-beta signaling, whereas silencing CHIP expression by small interfering RNAs led to increased TGF-beta signaling sensitivity. Second, based on the results of cell proliferation assays and JunB expression, we found that TGF-beta signaling could be abolished by stably overexpressing CHIP. Third, in those cell lines with stably expressed CHIP, we observed that the Smad3 protein level was dramatically decreased. Finally, we demonstrated that CHIP served as a U-box dependent E3 ligase that can directly mediate ubiquitination and degradation of Smad3 and that this action of CHIP was independent of TGF-beta signaling. Taken together, these findings suggest that CHIP can modulate the sensitivity of the TGF-beta signaling by controlling the basal level of Smad3 through ubiquitin-mediated degradation.

It has been reported that degradation of Smad proteins through ubiquitination is an important mechanism for regulating cellular responsiveness to TGF-␤ family ligands (11,12,15). The effects of most of the known ubiquitination ligases, however, are mainly for terminating the TGF-␤ signaling. For example, several E3 ligases, such as SCF/Roc1 (28) and Smurf1/2 (24, 25, 29 -32), are thought to mediate ubiquitination and degradation of the activated form of Smad proteins. Roc1 is a Cullin-binding protein with RING finger domain and has been reported to associate with Smad3 in a Smad3-SCF/ Roc1 complex in the nucleus, which is then transported into the cytoplasm for degradation (28). Smurf1 and Smurf2 are HECT (homologous to E6-AP C terminus) domain-containing E3 ligases (24, 25, 29 -32). Smurf1 selectively interacts with receptor-regulated Smads specifically in the BMP (bone morphogenitic protein) pathway to trigger Smad1 ubiquitination and degradation (32). Smurf1 has also been reported to interact with TGF-␤ type I receptor through Smad7 to regulate receptor degradation (30). It has also been shown that TGF-␤-induced Smad2-Smurf2 association with the transcriptional co-repressor SnoN leads to Smurf2 targeting SnoN for ubiquitin-mediated degradation by proteasome (29). These findings provided a mechanism for the down-regulation of TGF-␤ signaling via degradation of activated R-Smads (15).
It will be interesting to investigate whether some of the E3 ligases may also mediate the degradation of Smads in the absence of TGF-␤ signaling and, thus, could regulate the basal state of Smads. At present, whether Smurf2 could target unstimulated Smads is still unclear (25). Recently, based on a yeast two-hybrid study, we found that CHIP interacted with Smad1/4 and regulated the bone morphogenetic protein signal pathway (23). CHIP was originally identified as a novel tetratricopeptide repeat-containing protein and was confirmed to interact with heat shock proteins (33). This protein has been reported to inhibit Hsp40-stimulated ATPase activity of Hsc70. It may also block the Hsc70-substrate-binding cycle to reduce chaperon efficiency (33) or enhance Hsp70-dependent folding activity (34). It has been demonstrated that CHIP interacts directly with a tetratricopeptide repeat domain in Hsp90 to induce ubiquitination and degradation of the glucocorticoid receptor through the proteasome (35). Also, CHIP was identified as mediating the degradation of the cystic fibrosis transmembrane conductance regulator (36), Parkin-associated endothelin-receptorlike receptor (Pael-R) (37), the androgen receptor (38), and ErbB2 (39). In these previous studies, CHIP was considered as a novel E3 ligase to change the function of Hsp70/Hsp90 chaperons from a protein-folding machine into a degradation factor (35, 36, 39 -42). We would like to explore whether CHIP may have a functional role in the TGF-␤ signaling mechanism. Here, we present evidence that CHIP can be an important regulator in controlling the sensitivity of the TGF-␤ signaling by reducing the basal level of Smad3 through ubiquitination.
Mammalian Cell Lines and Transfections-Mv1Lu (mink lung epithelial cells), HEK293T ,COS-7, HepG2, and SkOV3 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum at 37°C in 5% CO 2 -containing atmosphere. BaF3 cells from a bone marrow-derived murine interleukin-3 dependent pro-B cell line were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 50 M ␤-mercaptoethanol, and 5% conditioned medium from the WEHI3 cells as a source of inteleukin-3. The U937 cell line was maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum and 50 M ␤-mercaptoethanol. Lipofectamine 2000 (Invitrogen) and the CalPhos mammalian transfection system (Clontech) were used for transient transfection for Mv1Lu, COS-7, and 293T cells. For establishment of the HA-CHIP-expressing stable Mv1Lu cells, pEFneo/HA-CHIP was transfected into Mv1Lu cells using Lipofectamine 2000 according to the manufacturer's instructions. As for BaF3/HA-CHIP cells, the pEFneo/HA-CHIP plasmid was transfected into the BaF3 cells using an electroporation procedure. Briefly, 1 ϫ 10 6 cells were mixed with 15 g of plasmid in ice-cold PBS, placed in a chamber, and subjected to 240 V (t Ϸ 35-40 ms) with a BTX ECM 399 electroporation system (BTX division of Genetronics Inc.). Selection was initiated 24 h after transfection using 2 mg/ml G418 (Invitrogen), and the resultant clones were characterized by Western blot and maintained in medium containing 0.5 mg/ml G418.
RNA Isolation and RT-PCR Analysis-Total RNAs from 5 ϫ 10 6 cells were extracted with 1 ml of Trizol (Invitrogen) plus 200 l of chloroform. The supernatant was precipitated with an equal volume of isopropanol. The RNA pellet was re-suspended in sterile water. 1 g of total RNA was used as template for one-step RT-PCR analysis (Takara Biotechnology Co., Ltd) according to the protocol. The primers designed to amplify the fragment in the C-terminal (U-box domain) of human and mouse CHIP were 5Ј-tataggatccaagaagaagcgctggaacagc-3Ј (forward) and 5Ј-tatactcgagtcagtagtcctccacccagccatt-3Ј (reverse). The reaction mixture was first incubated at 50°C for 30 min for reverse transcription and then denatured at 94°C for 2 min before PCR cycles. 26 cycles of PCR were conducted under the conditions of 94°C for 50 s, 55°C for 50 s, and 72°C for 1 min. The reaction products were extended at 72°C for 10 min before storing at 4°C. The RT-PCR for ␤-actin was performed with the same aliquoted samples as an internal control.
Pulse-Chase Assay-Transiently transfected COS-7 cells growing in the 12-well plate were pre-incubated in methionine/cysteine-deficient medium for 1 h and then pulsed for metabolic labeling for 10 min with 200 Ci/ml of [ 35 S]methionine/[ 35 S]cysteine (PerkinElmer Life Sciences) and chased at different time points. The cells were lysed and subjected to anti-FLAG (M2) immunoprecipitation and separated by SDS-PAGE. The gels were dried and 35 S-labeled F-Smad3 was visualized with a PhosphorImager.
Ubiquitination Assay in Vivo-293T cells were transfected with the related constructs to express HA-ubiquitin, His-tagged Smad3, and Myc-tagged CHIP proteins as indicated in the presence of 50 M MG132 for 5 h. 7.5 ng/ml TGF-␤1 was used as a final concentration for treatment of the cells. The cells were lysed, precipitated, and immunoblotted as described previously (23).
Luciferase Assays-Luciferase assays were carried out in Mv1Lu cells by co-transfecting (CAGA) 12 -MLP-Lux and different constructs as indicated. The vectors were used to balance the transfection. 6 h after transfection, the cells were treated with TGF-␤1 (7.5 ng/ml) for 24 h. Luciferase activities were measured by the Dual luciferase assay system (Promega) in Top Count (Packard). The Renilla luciferase internal control vector (pRL-TK) (Promega) was always used for calibration of the even transfection efficiency. 20 ng/ml leukemia inhibitor factor (LIF) (R&D Systems, Inc.) was used in the experiment for the STAT3 luciferase activity assay.

CHIP Can Significantly Reduce the Sensitivity of TGF-␤ Signaling As Demonstrated by Smad3-dependent Transcrip-
tional Activities-To test the role of CHIP on TGF-␤ signal transduction, we determined whether CHIP had any effect on Smad3-dependent transcriptional activities that could be measured by the luciferase reporter (CAGA) 12 -MLP-Lux. We performed the experiments in TGF-␤ response cells (Mv1Lu) and found that 7.5 ng/ml TGF-␤ stimulated the luciferase activity 5.6-fold over control (Fig. 1A, left columns). The same dose of TGF-␤, on the other hand, did not stimulate the reporter activity in cells overexpressing CHIP (Fig. 1A, right  columns). This result suggests that overexpression of CHIP can desensitize cells in response to TGF-␤ signaling.
To determine whether such an effect of overexpressed CHIP is specific, we conducted a similar experiment using the STATresponsive reporter pGL2-M67/sie, which responses to several cytokines (e.g. LIF) in the presence of STAT3 (43). Cells were co-transfected with the STAT3 reporter and STAT3 in the presence or absence of the overexpressed CHIP protein. When these cells were treated with LIF, we observed that the overexpressed CHIP failed to inhibit the LIF-induced STAT3 reporter (Fig. 1B). These results, together with what we observed as described in the above paragraph, indicate that CHIP functions specifically in the Smad-regulated pathway.
To further confirm the ability of overexpressed CHIP to reduce TGF-␤ signaling, we tested CHIP effects in another overexpression system that involves the constitutively active TGF-␤ type I receptor (TGF-␤RI(T204D)). In this system, the overexpressed TGF-␤RI(T204D) stimulated the (CAGA) 12 -MLP-Lux reporter luciferase activities to 82-fold over the control (Fig. 1C, open columns). Upon co-transfection of CHIP with TGF-␤RI(T204D), the luciferase activity was decreased to the background level (Fig. 1C, filled columns). In a third overexpression system, Smad3 and Smad4 were co-expressed with the luciferase reporter in the presence or absence of CHIP. Whereas Smad3 and Smad4 overexpression in Mv1Lu cells led to 153-fold increase in the luciferase activities (Fig. 1D, open columns), similar to results reported by others (44 -46), CHIP co-expression dramatically decreased the Smad3 and Smad4 responses to the background level (Fig. 1D, filled columns). These results suggest that overexpressing CHIP could effectively block the transcriptional modulator activities of Smad3 and Smad4. These experiments were repeated four times in triplicate, and the results were consistent. Similar data were also obtained in COS-7 and 293T cells (data not shown). In yet another experiment, we found that overexpression of CHIP also blocked the activity of Smad2 (Fig. 1E). These data together suggest that overexpressed CHIP can negatively regulate TGF-␤ signaling.
Knocking Down CHIP by siRNA Can Sensitize TGF-␤ Signaling-The above experiments showed that overexpression of the CHIP protein could inhibit (or attenuate) TGF-␤ signaling. Conversely, we wanted to examine whether knocking down the endogenous CHIP could promote or facilitate the TGF-␤ signaling. To address this question, we created two siRNA constructs with specific sequences of CHIP driven by the U6 promoter (designated pBS/U6/CHIPi1 (23) and pBS/U6/CHIPi2). Transfection of these two constructs into 293T cells significantly blocked the expressed CHIP protein, whereas the vector (pBS/U6) and the construct for EGFP siRNA (pBS/U6/EGFPi) did not ( Fig. 2A).
Knowing that 293T cells had a moderate level of endogenous CHIP expression (Fig. 2B), we transfected the CHIP siRNA constructs with the (CAGA) 12 -MLP-Lux reporter to this cell line. We observed that transfection of CHIP siRNA reduced the endogenous CHIP protein levels (Fig. 2, C and D, bottom sections). Interestingly, in the paired luciferase assays we found that the presence of CHIP siRNAs (1 and 2) enhanced TGF-␤RI(T204D)-stimulated luciferase activities by 2.5-fold (Fig.  2C, right two columns), whereas the vector (pBS/U6) and EGFP siRNA (pBS/U6/EGFPi) had no effect (Fig. 2B, middle two  columns). Furthermore, when we treated the Mv1Lu cells with TGF-␤ we also found that the CHIP siRNAs (1 and 2) could increase the luciferase activities (Fig. 2D). These experiments were repeated three times in triplicate. Our results indicate that CHIP siRNAs can facilitate TGF-␤ signaling through blocking CHIP expression. This finding provides the independent evidence that CHIP is a negative regulator in TGF-␤ signaling.
CHIP Can Desensitize TGF-␤-induced Cell Growth Arrest-If CHIP attenuates or inhibits TGF-␤ signaling, then overexpression of CHIP would block the response of the cells to TGF-␤. We examined this possibility by a cell proliferation assay using two cell lines, Mv1Lu and BaF3, which were reported to respond to TGF-␤ induced growth arrest (47). When increasing amounts of TGF-␤ were added to the medium, the BaF3 mock cells exhibited growth arrest (Fig. 3B, line with squares). Cells stably expressing CHIP (Fig. 3A, second lane from left), however, remained resistant to the same doses of TGF-␤ (Fig. 3B, line with diamonds). In Mv1Lu cells, we used three independent clones stably expressing the CHIP protein (Fig. 3C, first three lanes from the left) to repeat the proliferation experiment. Similar to what was observed in BaF3 cells, these CHIP-expressed cells lost the ability to respond to TGF-␤-induced growth arrest (Fig. 3D). These results suggest that cells with overexpressed CHIP are less sensitive to TGF-␤ mediated cell cycle arrest, implying that CHIP can attenuate TGF-␤ signaling.
CHIP Can Attenuate Native Gene Expression Induced by TGF-␤-The above data demonstrated that overexpressed CHIP could inhibit both the gene responses and the growth arrest mediated by TGF-␤. We further investigated whether CHIP had any effect on the expression of the native TGF-␤ response genes. We examined JunB, an early TGF-␤ response gene (4,48,49), in both BaF3 and Mv1Lu wild type cells (mock cells) with or without overexpressing CHIP. Whereas TGF-␤ could stimulate junB gene expression in the mock cells (Fig. 4,  A and B, second and sixth lane from the left), in cell lines stably expressing CHIP the JunB protein level was significantly decreased. Such effects were detectable at either 4 or 20 h after TGF-␤ exposure (Fig. 4, A and B). These data clearly suggest that the increased expression of CHIP could inhibit the native gene expression induced by TGF-␤.
CHIP-mediated Degradation of Smad3 Can Occur Independent of TGF-␤ Signaling-To understand the mechanisms underlying the effect of CHIP-mediated inhibition of TGF-␤ signaling, we speculated that CHIP might mediate the degradation of Smad3 through ubiquitination based upon our previous studies (23). To test this hypothesis, we first examined the degradation of Smads induced by CHIP. As expected, overexpression of CHIP indeed decreased Smad3 protein levels in a manner sensitive to the proteasome inhibitor MG132, implying that CHIP may directly or indirectly induce the proteasomal degradation of Smad3 (Fig. 5A). This degradation was further confirmed by pulse-chase assays as shown in Fig. 5, B and C.
TGF-␤ signaling has been shown to enhance Smad3 turnover (15). We tested whether the CHIP-mediated degradation of Smad3 is dependent on TGF-␤ signaling. We assayed the level of exogenously expressed FLAG-tagged Smads and treated the cells with or without TGF-␤ in the presence or absence of overexpressed CHIP. Western blot analysis showed that TGF-␤ indeed initiated the degradation of the exogenous Smad3 (Fig.  5D, upper section, compare lanes 1 and 3). However, overexpression of CHIP did not show significant alteration of the degradation of Smad3 (Fig. 5D, upper section, lane 4). Actually, Smad3 degradation triggered by TGF-␤ remained at the same rate in the presence or absence of CHIP overexpression (Fig.  5D, upper section, compare lanes 1 and 2 without overexpression of CHIP and lanes 2 and 4 with overexpression of CHIP). The quantified densities of Smad3 bands (Fig. 5D, lower section) further showed that the difference between lanes 1 and 2 (without TGF-␤) and that between lanes 2 and 4 (with TGF-␤) were almost the same. These data suggest that CHIP-mediated Smad3 degradation might be independent of TGF-␤ signaling.
CHIP Can Mediate Smad3 Ubiquitination Independent of TGF-␤ Signaling-Previously, we had observed that CHIP can mediate Smad1 ubiquitination (23). By optimizing our experimental procedures, we have now observed that CHIP can also induce ubiquitination of Smad3 (Fig. 6A, smear bands) in the presence of 50 M MG132. To further test whether the CHIPmediated Smad3 ubiquitination is subjected to the regulation of TGF-␤ signaling, we performed an in vivo ubiquitination experiment by overexpressing Smad3 and CHIP in 293T cells treated with or without TGF-␤. We found that CHIP induced ubiquitination of Smad3 with or without TGF-␤ treatment (Fig.  6B). The Smad3(3S-A) mutant, which could not be phosphorylated by TGF-␤ (50), was also tested for its ubiquitination by CHIP. Results from our in vivo ubiquitination experiment indicated that the mutant Smad3(3S-A) was also ubiquitinated by CHIP (Fig. 6C). These data implied that the CHIP-induced Smad3 ubiquitination occurred independently of TGF-␤ signal activation.
Because CHIP was reported as an E3 ligase containing a U-box domain, we tested the role of the U-box in CHIP-mediated Smad3 ubiquitination. We used a mutant of CHIP (H260Q), which contained a point mutation within the U-box and lacked its ubiquitin E3 ligase, to perform the ubiquitination experiment. As shown in Fig. 6B, the CHIP (H260Q) mutant exhibited significantly reduced ability to mediate Smad3 ubiquitination. In luciferase assays, the mutant CHIP (H260Q) also failed to inhibit Smad3-mediated transcriptional activation (Fig. 6D). These data indicate that CHIP functions as an E3 ligase for Smad3 and that such activity is dependent on its U-box.
CHIP Maintains Basal Level of Smad3 but Does Not Affect the Kinetics-The above experiments (Fig. 5) demonstrated that CHIP mediates degradation of the exogenously overexpressed Smad3. To investigate whether CHIP can change the basal level of the endogenous Smad proteins, we stably expressed CHIP in both Mv1Lu and BaF3 cells and tested the levels of the endogenous Smad3 in the presence or absence of TGF-␤. Results of our Western blot analysis showed that the endogenous Smad3 protein level in cells with stably overexpressed CHIP was greatly decreased compared with that in wild type cells (Fig. 7A, compare lanes 1 with 5, upper two  sections). The presence of CHIP, however, did not prevent TGF-␤ to enhance the accumulation of Smad3 (Fig. 7A, comparing lanes 1 to 4 with lanes 5 to 8, upper two sections), even though the basal levels under CHIP were dramatically lower.
Finally, we examined the kinetics of degradation of phosphorylated Smad3 under the treatment of TGF-␤ with or without CHIP. We found that TGF-␤ could mediate phosphorylation of Smad3 in the Mv1Lu cell lines regardless of whether or not CHIP was stably expressed (Fig. 7C). Apparently, the degradation rate of the phosphorylated Smad3 was not affected by the overexpressed CHIP protein (Fig. 7, D and E). These results thus suggest that CHIP mainly regulates the basal Smad3 protein level and that its effect does not seem to affect the kinetics of the TGF-␤ signaling. DISCUSSION TGF-␤ signaling has been intensively studied during the past two decades (1)(2)(3). It is now known that TGF-␤ signaling is mediated by Smad proteins in the cytoplasm where Smads are phosphorylated and translocated into the nucleus after activation of TGF-␤ receptors (2,9). It has been documented that several proteins at different levels could control the activities of Smads (R-Smads and common Smads) to regulate TGF-␤ signaling (2, 3). One of the most advanced signs of progress in understanding the regulation of TGF-␤ signal transduction is the study of the termination mechanism of the signaling regulated by ubiquitin-proteasome-dependent degradation (11,51). TGF-␤ signaling could be tightly controlled by positive and negative regulators (10) via the accumulation and subsequent degradation of the phosphorylated Smads (R-Smads) (15). The down-regulation of activated Smad proteins could be a critical mechanism for effectively turning off TGF-␤ signaling to avoid excess stimulation. To date, several E3 ligases have been identified as participating in terminating the TGF-␤ signaling. Among them, SCF/Roc1 and Smurf1/2 were reported as mediating the degradation of the phosphorylated Smads proteins (24,28,29). In this report, we proposed that CHIP, a newly recognized E3 ubiquitin ligase, could regulate the basal level of Smad3 through a ubiquitin-dependent degradation pathway and, thereby, desensitized the cells to responding to TGF-␤ signaling. We observed that the overexpression of CHIP could significantly attenuate TGF-␤ signaling based on luciferase reporter assays, cell growth arrest experi-ments, and measurements of TGF-␤ induced gene expression. Conversely, knocking down the endogenous CHIP protein was found to significantly enhance the sensitivity of TGF-␤ signaling. These data thus indicate that CHIP is a negative regulator for TGF-␤ signal transduction.
Interestingly, unlike Roc1 and Smurf1/2, in which the degradation of Smads was dependent on TGF-␤ stimulation (24,25,29,30,32), CHIP could decrease the total Smad3 level independently of the TGF-␤ activation. Our data demonstrated that TGF-␤ signaling was unnecessary for CHIP to mediate Smad3 ubiquitination and degradation. It appears that the CHIP-mediated degradation of Smad3 was independent of the phosphorylation of Smad3. For example, we observed that the Smad3(3S-A) mutant, which lost the feature of phosphorylation, could be equally ubiquitinated (Fig. 6C) and degraded by CHIP (data not shown). Taken together, our results suggest that CHIP may desensitize the cell's response to TGF-␤ by decreasing the basal Smad3 level.
Our findings have added to the general understanding that TGF-␤ signaling could be controlled by a broad spectrum of E3 ligases. Differing from SCF/Roc1, a typical E3 ligase containing a RING finger domain, and Smurf1/2, a newly identified E3 ligase containing a HECT domain, CHIP contains a U-box domain, which was classified as a new member of the E3 ligase family. Originally, CHIP was reported as mediating the degradation of misfolded or unfolded proteins (36, 37, 40 -42). This work proposes a new function for CHIP. Earlier, we had observed that CHIP could interact with Smad1, 2, and 4 and mediate their ubiquitination (23). We now have evidence that CHIP could also degrade Smad3 in a manner independent of TGF-␤ signaling. Our findings suggest that CHIP may regulate the basal levels of multiple Smads through a proteasome degradation mechanism and, in turn, can modulate the sensitivity of TGF-␤ signaling.