|Home | About | Journals | Submit | Contact Us | Français|
Smad2 and Smad3 (Smad2/3) are key intracellular signal transducers for TGF-β signaling and their transcriptional activities are controlled through reversible phosphorylation and nucleocytoplasmic shuttling. However, the precise mechanism underlying nuclear export of Smad2/3 remains elusive. Here we report the essential function of RanBP3 in selective nuclear export of Smad2/3 in the TGF-β pathway. RanBP3 directly recognizes dephosphorylated Smad2/3, which results from the activity of nuclear Smad phosphatases, and mediates nuclear export of Smad2/3 in a Ran-dependent manner. As a result, increased expression of RanBP3 inhibits TGF-β signaling in mammalian cells and Xenopus embryos. Conversely, depletion of RanBP3 expression or dominant negative inhibition of RanBP3 enhances TGFβ-induced anti-proliferative and transcriptional responses. In conclusion, our study supports a definitive role of RanBP3 in mediating Smad2/3 nuclear export and terminating TGF-β signaling.
The transforming growth factor-beta (TGF-β) superfamily signaling controls a wide variety of biological processes from development to pathogenesis primarily through intracellular Smad proteins (Derynck and Miyazono, 2007). Smads are structurally and functionally grouped into three subfamilies: five receptor-activated Smads (R-Smads), one common Smad and two inhibitory Smads. Of R-Smads, Smad2 and Smad3 (Smad2/3 thereafter for simplicity) transduce activin and TGF-β signals, whereas Smad1, Smad5 and Smad8 transduce signals from the BMP/GDF (Bone Morphogenetic Proteins/Growth Differentiation Factor) family. Upon binding to their receptor complex at the cell surface, TGF-β and related growth factors activate Smad proteins and promote accumulation of Smads in the nucleus, where they mediate transcriptional responses (ten Dijke and Hill, 2004; Massague et al., 2005; Feng and Derynck, 2005).
Smads possess intrinsic nucleocytoplasmic shuttling capacity and their subcellular distribution is determined through association with nuclear transport factors and retention proteins (Watanabe et al., 2000; Pierreux et al., 2000; Inman et al., 2002b; Nicolas et al., 2004; Schmierer and Hill, 2005). Although it has been reported that TGF-β-induced phosphorylation favors the nuclear import of phosphorylated R-Smads (P-Smads) by enhancing their association with importin-β (Xiao et al., 2000; Kurisaki et al., 2001) and/or disassociation with cytoplasmic retention factors such as SARA (Xu et al., 2000), live cell microscopy analysis suggest a pivotal role of nuclear export in regulating TGF-β signaling. TGF-β appears not to affect the nuclear import rate of Smad2 (Schmierer and Hill, 2005). Instead, it decreases Smad2 nuclear export, thereby retaining more the phosphorylated Smad2 in the nucleus (Xu et al., 2002; Schmierer and Hill, 2005). Conversely, dephosphorylation of P-Smad2/3 by phosphatase PPM1A promotes nuclear export of Smad2/3 (Lin et al., 2006). Therefore, the phosphorylation status of R-Smads is coupled with its nuclear accumulation (Xu et al., 2002; Inman et al., 2002b; Lin et al., 2006). It is generally thought that the level of nuclear R-Smads determines the duration and strength of signaling.
Unlike common Smad4 and BMP-specific Smad1, whose nuclear export depends on the function of CRM1, a Leptomycin B (LMB)-sensitive export receptor (Watanabe et al., 2000; Pierreux et al., 2000; Xiao et al., 2001; Xiao et al., 2003), TGF-β-specific Smad2/3 undergo nuclear export in a LMB-insensitive manner and appear to be independent of CRM1 function or Smad4 export (Pierreux et al., 2000; Inman et al., 2002b; Nicolas et al., 2004; Schmierer and Hill, 2005). It has been reported that TGF-β-specific Smad3 nuclear export involves TGF-β-dependent interaction with Exportin 4 (in the case of Smad3) (Kurisaki et al., 2006) and TGF-β independent interaction with nucleoporins (Xu et al., 2002). Given the highly conserved putative NESs and Exportin4-binding motifs between BMP- and TGF-β-specific R-Smads (Xiao et al., 2001; Kurisaki et al., 2006), it remains unresolved how export mediators discriminate these R-Smads.
In this study, we provide compelling evidence that Ran-binding protein 3 (RanBP3) serves as a direct mediator for nuclear export of Smad2/3. Originally identified as an interaction partner with the guanine nucleotide exchange factor (RCC1) (Mueller et al., 1998), RanBP3 has been implicated in the process of nucleocytoplasmic trafficking by its ability to bind a few proteins including Ran, nuclear pore components and CRM1 (Mueller et al., 1998; Englmeier et al., 2001; Lindsay et al., 2001). We found RanBP3 directly interacts with Smad2/3 in their dephosphorylated form, but not with Smad1. Consistent with its role in exporting Smad2/3, RanBP3 negatively regulates TGF-β signaling. Knockdown of RanBP3 or blockade of the RanBP3-Smad2/3 interaction inhibits Smad2 nuclear export and enhances TGF-β-induced responses. In accordance, RanBP3 represses activin/Smad2, but not BMP-Smad1 signaling in Xenopus embryos. These findings support the notion that nuclear export of R-Smads is selectively controlled by specific export mediators in the process of TGF-β signal termination.
In an attempt to identify novel intracellular factors controlling TGF-β signaling, we discovered that RanBP3 was a strong candidate for mediating nuclear export of Smad2/3. As a first step in evaluating a possible role of RanBP3 in TGF-β signaling, we examined the effects of RanBP3 on Smad2/3-mediated transcriptional activation using various Smad-dependent gene reporters. Increased expression of human or mouse RanBP3 caused a decrease in TGF-β-dependent transcription from the Smad-binding element (SBE)-Luc reporter (Figure 1A; data not shown). RanBP3 also inhibited TGF-β-induced activation of the natural p21 promoter in HaCaT cells (Figure 1B), as well as the p15 promoter and the plasminogen activator inhibitor type 1 (PAI-1) promoter in HepG2 cells (data not shown). However, RanBP3-wv, a mutant that loses its Ran-binding activity (Englmeier et al., 2001), failed to block TGF-β-induced promoter activation, although the mutant was expressed at a similar level to that of RanBP3 (Figure 1C). In sharp contrast to its ability to regulate TGF-β signaling, RanBP3 did not interfere with BMP-induced activation of the Id1 promoter in C2C12 cells (Figure 1D) or HepG2 cells (Figure S3), suggesting that RanBP3 specifically acts on TGF-β, but not BMP, signaling.
Consistent with the ability of RanBP3 to inhibit activation of TGF-β-responsive promoters, HaCaT cells stably expressing RanBP3 (Figure 1H, blot a) diminished TGF-β-mediated induction of endogenous p15, p21 and PAI-1 mRNAs (Figure 1E-G). In addition, the protein level of p21 was also reduced (Figure 1H, blot b). Overexpression of RanBP3, however, did not disrupt the TGF-β-induced C-terminal phosphorylation of Smad2/3 (Figure 1H, blot c and d), nor did it affect the total level of Smad2/3 in the same stable cells (Figure 1H, blot e). These data suggest that RanBP3 acts downstream of Smad2/3 activation.
During early vertebrate development, activin/nodal and BMP signals are known to control embryonic patterning and cell fate determination (Chang et al., 2002). Activation of TGF-β signaling induces mesoderm- and endoderm-specific gene expression in a dose-dependent fashion in early Xenopus embryos. To examine whether RanBP3 can regulate TGF-β signaling during Xenopus embryogenesis, we examined the effects of RanBP3 on activin- or BMP-induced endogenous gene expression in Xenopus ectodermal explants (animal caps). As shown in Figure 1I, activin induced a whole range of mesendodermal markers, including the endodermal marker Sox17α, the dorsal mesodermal marker chordin, the ventrolateral mesodermal marker XWnt8, and the pan-mesodermal marker XBra at gastrula stages (lane 2), whereas BMP4 induced the expression of Sox17α, XWnt8 and Xbra (lane 5). Co-expression of wildtype RanBP3, but not RanBP3-wv, completely suppressed activin-induced expression of Sox17α and chordin, and moderately reduced the expression of Xbra and XWnt8 (lanes 3 and 4). In contrast, RanBP3 or RanBP3-wv failed to block the expression of the genes activated by BMP4, including Sox17α, Xwnt8 and XBra (lane 6). These results further support the notion that RanBP3 preferentially blocks activin but not BMP signaling in a Ran-binding dependent manner.
We next investigated whether RanBP3 depletion enhances TGF-β responses. We first established HaCaT cell lines that stably express distinct shRNAs against separate target sequences of RanBP3, i.e. shRNA494 (designated RanBP3-KD1) or shRNA-676 (designated RanBP3-KD2 and KD3)(Figure 2A and 2E). It is clear that depletion of RanBP3 enhanced TGF-β-induced expression of both the SBE-Luc reporter (Figure 2B) and the natural genes including those encoding PAI-1, and CDK inhibitors p15 and p21 (Figure 2C). Consistent with the increased mRNA levels of p15 and p21, depletion of RanBP3 rendered cells more sensitive to TGF-β-induced growth inhibition (Figure 2D and 2E). Collectively, these data established a novel role of RanBP3 as negative modulator in TGF-β-mediated transcriptional responses.
Because RanBP3 has been implicated in nuclear export by its ability to bind Ran, nuclear pore components and CRM1 (Mueller et al., 1998; Englmeier et al., 2001; Lindsay et al., 2001), we sought to evaluate if RanBP3 has a direct role in regulating Smad2/3 nuclear export. The ability of RanBP3 to affect TGF-β-induced nuclear accumulation of endogenous Smad3 was examined in HaCaT cells. As seen in Figure 3A, indirect immunostaining of Smad3 revealed that overexpression of RanBP3 decreased the intensity of nuclear Smad3 by 55.3%(+/-4.2%). Similar results were observed with mouse RanBP3 (data not shown). To further confirm this observation, TGF-β-induced nuclear accumulation of endogenous Smad2 was examined in parental HaCaT cells, RanBP3-expressing stable cells (RanBP3-OE), and RanBP3 knockdown cells (RanBP3-KD1). The results demonstrate that the level of nuclear Smad2/3 is inversely proportional to that of RanBP3 in the nucleus. As shown in Figure 3B, TGF-β stimulation rapidly induced Smad2 accumulation in the nucleus in parent HaCaT cells. Increased expression of RanBP3 reduced the level of nuclear Smad2 in RanBP3-OE cells, whilst knockdown of endogenous RanBP3 expression rendered more endogenous Smad2 accumulated in the nucleus of RanBP3-KD1 cells.
To specifically evaluate Smad2/3 export without interference of continuous import, we next examined the TGF-β-induced nuclear accumulation of Smad2/3 in the presence of TβRI inhibitor SB431542, which blocks Smad2/3 phosphorylation in the cytoplasm and nuclear accumulation thereafter (Laping et al., 2002; Inman et al., 2002a). In parental HaCaT cells, Smad2 and Smad4 were accumulated in the nucleus after 2 h of TGF-β stimulation, but redistributed throughout the whole cells after removal of TGF-β and simultaneous addition of SB431542 for 2 h. Treatment with CRM1 inhibitor LMB blocked Smad4 but not Smad2 redistribution (Figure 3C, top panels). In RanBP3-KD1 cells, sustained Smad2 nuclear localization was observed in the absence or presence of LMB (Figure 3C, second row panels from the bottom). On the contrary, knockdown of RanBP3 did not affect the Smad4 cellular redistribution (Figure 3C, bottommost panels).
We next sought to specifically examine how much Smad2/3 are exported into the cytoplasm. To this end, cells were similarly treated as in Figure 3C and harvested at different time point during 4 h export process. Smad2/3 levels in the cytoplasmic vs. nuclear fractions at each time point were analyzed by Western blotting. The quality of cytoplasmic/nuclear fractionation was determined by the presence of cytoplasmic GADPH and nuclear Lamin A/C in the fractions. At time point 0 (immediately after 60 min TGF-β treatment), a very low level of Smad2/3 in the cytoplasmic fractions of both wildtype and knockdown cell lines was detected, indicating most of the Smad2/3 resided in the nucleus. SB431542 treatment promoted the cycling back of Smad2/3 in the cytoplasm in parental HaCaT cells, as indicated by the gradual increase in the level of cytoplasmic Smad2/3 and the corresponding decrease in the nuclear Smad2/3 (Figure 3D). In contrast, knockdown of RanBP3 slowed down the cytoplasmic-to-nuclear shift of Smad2/3 in RanBP3-KD1 cells (Figure 3D). Taken together, the data in Figure 3 support the notion that RanBP3 positively regulates the nuclear export of Smad2/3.
Since RanBP3 is a Ran-binding protein, we sought to examine whether RanBP3 requires Ran to export Smad2/3 by using in vitro export assay. In these assays, HaCaT cells stably expressing GFP-Smad2 were first treated with TGF-β for 2 h to induce Smad2 nuclear accumulation, and then pulsed with digitonin for 5 min to permeabilize cell membrane. After washing, the remaining level of GFP-Smad2, which represented nuclear Smad2 that was not exported, was determined by Western blotting analysis. Several dosages of digitonin were tested to produce substantial loss of cytoplasmic proteins yet retain the intact nuclear envelope (Figure S1). To avoid over-permeabilization, we chose 30 ng/ml of digitonin in the subsequent assays. Proper cell membrane permeabilization was confirmed by lack of GAPDH in the total lysates of permeabilized cells (Figure 4A). In addition, functional integrity of the nuclear envelope was assessed by similar levels of GFP-Smad2 in digitonin-permeabilized and non-permeabilized cells at time point 0 (normalized against that of Lamin A/C) as well as the disappearance of cytoplasmic GADPH after 5 min of digitonin treatment (Figure 4A, lane 4). In vitro export of GFP-Smad2 was progressive over time, as indicated by a reduced level of GFP-Smad2 protein (Figure 4A, lanes 4-6) or diminished GFP fluorescence (data not shown) in the permeabilized cells. Retention of GFP-Smad2 in the nucleus appeared to partially, if not completely, depend on Ran activity and energy because it decreased after addition of recombinant Ran proteins to the cells (lanes 9 and 10) and increased when incubated on ice (lanes 7 and 8).
We then analyzed the effect of the Ran-binding deficiency of RanBP3-wv on Smad2 export. Albeit at low efficiency, recombinant RanBP3 and RanBP3-wv proteins could similarly enter the nucleus of permeabilized cells (Figure S2). It is apparent that RanBP3 was capable of exporting Smad2 (Figure 4B, lanes 4 and 5). In sharp contrast, RanBP3-wv had no effect on Smad2 export in vitro (lanes 6 and 7). To further compare the effect of RanBP3 or RanBP3-wv mutant, we also used a highly sensitive quantitative export assay (Cullen, 2004), which has been previously used to analyze Smad2 nuclear export (Xu et al., 2002; Lin et al., 2006). Consistent with our data in vitro, we found that RanBP3 increased Smad2 export by 5.2-folds, whereas RanBP3-wv failed to produce any effects (Figure 4C). These results strongly suggest that RanBP3 promotes Smad2 nuclear export in a Ran-dependent manner.
Having established the role of RanBP3 in mediating nuclear export of Smad2/3, we investigated if RanBP3 could physically associate with Smads. Co-immunoprecipitation (co-IP) assay was used to examine the ability of RanBP3 to interact with Smads in transfected HEK293 cells. We detected the presence of RanBP3 in the immuno-complex of Smad2 (Figure 5A, lanes 2 and 3), but not in that of Smad1 (Figure 5A, lane 4-5). RanBP3 could also bind to Smad3 (Figure 5B, lanes 1 and 2), which is the closest homolog of Smad2 and also functions as TGF-β signal transducer. The interaction of RanBP3 with Smad4, a common Smad for both TGF-β and BMP signaling, was comparatively weaker (Figure 5B, lanes 3 and 4). Furthermore, immunoprecipitation of endogenous RanBP3 retrieved endogenous Smad2 and Smad3 proteins in human keratinocytes HaCaT (Figure 5C, lane 4). Taken together, these data reveal an interaction between RanBP3 and TGF-β-specific Smad2/3 under physiological conditions. Notably, activation of the TGF-β receptor appeared to decrease the Smad2/3-RanBP3 interaction (Figure 5C, lane 5, to be discussed later).
We further examined the RanBP3-Smad3 interaction by mapping their respective interacting domains. As shown in Figure 5D, both the N-terminal MH1 domain and C-terminal MH2 domain could bind to RanBP3 in vitro. The domains in RanBP3, namely the N-terminal domain (N), intermediate domain domain (F) and C-terminal Ran:GTP-binding domain (R) (Mueller et al., 1998; Lindsay et al., 2001), all associated with Smad3 yet weakly than the full-length RanBP3 (Figure 5E). Among all the domains, R apparently exhibited stronger affinity to Smad3 than N or F (Figure 5E, lane 4) and directly interacted with Smad3 in vitro (Figure 5F, lane 3).
To further identify the structural determinants that define the RanBP3-interacting specificity of Smad3, we generated a series of chimeras between Smad1 (unable to bind to RanBP3) and Smad3 (able to bind to RanBP3). In these chimeras, one or two domains of Smad3 were replaced with the counterpart(s) of Smad1 (Figure 5G, right panel). The ability of these chimeras to bind to RanBP3 was assessed and compared with wildtype Smad 3 in co-IP experiments. As observed earlier, RanBP3 interacted with wildtype Smad3, but not Smad1 (Figure 5G, lanes 2 and 3). RanBP3 also interacted with Smad1-Smad3 chimeras containing the MH1 and/or MH2 domain of Smad3 (Figure 5G, lanes 4, 5 and 7-9). However, RanBP3 did not interact with Smad131 with the linker region only of Smad3 (Figure 5G, lane 6), suggesting the MH1 and MH2 domain mediate the RanBP3-Smad3 interaction. Interestingly, among all the chimeras, Smad313, which contains both the MH1 and MH2 domains of Smad3, exhibited the strongest interaction with RanBP3 that is comparable to that of wildtype Smad3 (Figure 5G, lanes 3 and 8). Taken together, data in Figure 5 suggest that at least two different epitopes in Smad3 are recognized by RanBP3 and cooperatively mediate the Smad3-RanBP3 interaction.
As shown in Figure 5, the interaction between RanBP3 and Smad2/3 was apparently weaker in the presence of active TGF-β signaling, either by expression of a constitutively active mutant form of TβRI, i.e. ALK5(T202D) (Figure 5A, lane 3) or TGF-β ligand stimulation (Figure 5C, lane 5). Since RanBP3 resides in the nucleus (Mueller et al., 1998) (Figures 3A and 3B), where Smad2/3 dephosphorylation by nuclear phosphatase PPM1A also occurs (Lin et al., 2006), we examined the effect of PPM1A on the RanBP3-Smad3 interaction. The level of TGF-β-induced Smad3 phosphorylation was reduced by PPM1A, but was not affected by overexpressed RanBP3 (Figure S4). Notably, increased expression of PPM1A clearly enhanced the association between RanBP3 and Smad3 (Figure 6A, lanes 3 and 4). These data support the notion that RanBP3 preferentially binds to dephosphorylated Smad2/3.
We then carried co-IP experiments to examine the Smad2 SXS motif mutant that either lacks the C-terminal serine phosphorylation (inactive, 2SA) or harbors phosphorylation-mimetic residues (active, 2SD). Analysis of the results revealed different binding affinity of these mutants toward RanBP3. As shown in Figure 6B, RanBP3 pulled down more Smad2(2SA) than Smad2(2SD) mutant (lanes 3 and 4). Since the SXS motif phosphorylation induces Smad2/3-Smad4 interaction, we sought to determine whether TGF-β-induced Smad2-Smad4 interaction contributed to the decreased association between Smad2 and RanBP3. As shown in Figure 6C, overexpression of Smad4 did not affect the interaction between RanBP3 and the Smad2(2SD) mutant (lanes 2 and 3), which readily interact with Smad4 in the absence of TGF-β stimulation (data not shown). Overexpression of RanBP3 also had no affect on the interaction between Smad2(2SD) mutant and Smad4 (Figure 6D). These data suggest that Smad4 and RanBP3 do not interfere with each other for binding to Smad2/3.
To further explore how the phosphorylation affects the RanBP3-Smad2/3 interaction, we carried out in vitro binding assays to examine the RanBP3 interaction with the recombinant Smad2 MH2 (aa 241–467) protein either in its phosphorylated or unphosphorylated form. The results revealed that GST-RanBP3 could bound to unphosphorylated Smad2 MH2 domain (Figure 6E, lane 1), but had no binding affinity to the phosphorylated Smad2 MH2 domain (Figure 6E, lane 2). Collectively, these data establish that RanBP3 preferentially binds to the unphosphorylated and recently dephosphorylated Smad2/3.
To identify the subcellular compartment where RanBP3 interacts with Smad2/3, we used a YFP-based fluorescence complementation system that allows easy visualization of protein interactions with microscopy (Hu et al., 2002). In this system, YFP fragment YN (aa 1-154) and YC (aa 155-238) were separately fused with each of the interacting protein partners. Protein interaction brings together the two YFP halves to reconstitute a functional YFP that produces a fluorescent signal. In HaCaT cells co-expressing YC-RanBP3 and YN-Smad3, YFP fluorescence was detected in the nucleus in the absence of TGF-β (Figure 6F, set a); similar results were observed when the reverse pair, i.e. YN-RanBP3 and YC-Smad2, were used (Figure 6F, set b). As negative controls, the YC-RanBP3/YN-vector pair and YN-Smad2/YC-vector pair did not yield fluorescence (Figure 6F, set c and d).
We next examined how the RanBP3-Smad2 interaction responds to TGF-β. As shown in Figure 6G, endogenous Smad2 resided in both the cytoplasm and the nucleus in HaCaT cell at the resting state (0.2% FBS treatment overnight). Whilst TGF-β stimulation for 2 h induced nuclear accumulation of Smad2, TβRI inhibitor SB431542 rendered Smad2 mostly accumulated in the cytoplasm. The distribution of transiently expressed YC-Smad2 mimics endogenous Smad2. When YC-Smad2 and YN-RanBP3 were co-expressed, YFP fluorescence resided exclusively in the nucleus under all the conditions examined, strongly suggesting that RanBP3 interacts with Smad2/3 in the nucleus regardless TGF-β signaling.
As RanBP3 also associates with the MH1 domain of Smad3 (Figure 5D, lane 2), we sought to examine whether RanBP3 would affect the Smad3's DNA-binding ability using DNA pulldown assay. The biotin-labeled SBE oligo comprises of four copies of the SBE sequences and was used to retrieve SBE-containing complex from the lysates of HaCaT control cells and RanBP3-KD1 stable cells. As shown in Figure 6H, TGF-β induced binding of Smad3 and Smad4 to biotinylated SBE (lane 3 and 5). RanBP3 was undetectable in the SBE-containing complex (lane 2 and 3), which is consistent with its position downstream of the Smad complex dissociation and dephosphorylation of Smad3. Moreover, RanBP3 depletion did not affect the level of SBE-bound Smad3 (lane 5), suggesting that RanBP3 would not affect the Smad3's DNA-binding activity.
To evaluate how the RanBP3-Smad interaction modulates Smad2/3 export as well as TGF-β signaling, we explored the use of the Smad-interacting region in RanBP3 as a competitive inhibitor for this interaction. The RanBP3-R resides in the cytoplasm when exogenously expressed (Welch et al., 1999). Since RanBP3 interacts with Smad2/3 in the nucleus (Figure 6F and 6G) and partly through the R domain of RanBP3 (Figure 5E and 5F), we introduced the nuclear localization signal (NLS) in the N terminus of RanBP3-R (generating NLS-R) in order to force RanBP3-R to be localized in the nucleus. When NLS-R was co-transfected with RanBP3, the association of RanBP3 with Smad2 (Figure 7A) and Smad3 (Figure 7B) was disrupted in HEK293T cells, clearly demonstrating that RanBP3-R competes with RanBP3 for Smad2/3 binding. We then examined the effect of NLS-R in Smad2 export and TGF-β-induced transcriptional responses. Quantitative Smad2 export assay revealed NLS-R attenuated RanBP3-mediated Smad2 export in a dose-dependent manner (Figure 7C). Consistently, NLS-R enhanced the TGF-β-induced SBE-luc response in HaCaT cells (Figure 7D). Collectively, these data suggest that through its direct interaction with Smad2/3, RanBP3 mediates the nuclear export of Smad2/3 and hence function as a negative modulator in TGF-β signaling.
Stringent control of activation and termination of TGF-β signaling is critical for normal cell physiology. Deregulation of the TGF-β signaling pathway has been associated with pathogenesis of cancer and other diseases (Massague et al., 2000; Bierie and Moses, 2006; ten Dijke and Arthur, 2007). Though much effort during the past decade has focused on activation of TGF-β signaling, increasing evidence suggests that modulation of TGF-β signaling decay also plays important roles in maintenance of a balanced signaling level. TGF-β signaling can be terminated by different ways, such as dephosphorylation of R-Smads, degradation of Smad proteins, or nuclear exclusion of Smad signal transducers. Recent work demonstrates that a phosphatase-dependent mechanism operates in the export of dephosphorylated Smad2 out of the nucleus during termination of TGF-β signaling (Inman et al., 2002b; Nicolas et al., 2004; Schmierer and Hill, 2005). Although PPM1A has been identified as a Smad2 phosphatase and can promote nuclear export of Smad2 (Lin et al., 2006), the exact export machinery that transports dephosphorylated Smad2 out of the nucleus has not been elucidated. In this study, we identify RanBP3 as the key component that mediates Smad2/3 nuclear export.
RanBP3 is a nuclear Ran-binding protein. In addition to its C-terminal domain that has the Ran-binding capacity, the N-terminal sequences of RanBP3 contain several FXFG motifs that are characteristic of a subgroup of nucleoporins (Mueller et al., 1998). The ability of RanBP3 to mediate nuclear export of Smad2/3 depends on its direct association with Smad2 and Smad3, as demonstrated by a series of biochemical and immunofluorescence experiments. RanBP3 interacts with Smad2/3 primarily with its C-terminal R domain, and it shows strong preference to unphosphorylated Smad2/3 (Figures 5 and and6).6). The binding occurs in the nucleus (Figures 6F and 6G), suggesting that dephosphorylation of activated Smad2/3 by phosphatases may be a prerequisite for RanBP3 binding and nuclear export. Indeed, the nuclear Smad phosphatase PPM1A facilitates, while receptor activation decreases, the RanBP3-Smad2/3 association (Figure 6A). As un- or de-phosphorylated Smad2 is most likely not in complex with Smad4, these findings support the notion that RanBP3-mediated Smad2/3 nuclear export is downstream of Smad dephosphorylation and/or the disassociation of Smad2/3 from Smad4. Thus, either unphosphorylated or recently dephosphorylated Smad2/3 serve as cargos for RanBP3-mediated nuclear export.
The function of RanBP3 in regulating Smad nuclear export is restricted to the TGF-β pathway. RanBP3 selectively inhibits activin/TGF-β, but not BMP, transcriptional responses in mammalian cells (Figure 1H) and Xenopus embryos (Figure 1I), in agreement with the cargo selection of RanBP3 (Figures 5A, 5B and 5G). This supports the notion that distinct mechanisms underlying R-Smad nuclear export operate in TGF-β and BMP pathways. It has been shown that both Smad1 and Smad4 utilize CRM1 to associate with NES in these Smads for their nuclear export (Xiao et al., 2001; Xiao et al., 2003). This process is thus susceptible to treatment with LMB, which impairs the function of CRM1. In contrast, nuclear export of Smad2/3 is resistant to LMB (Inman et al., 2002b; Nicolas et al., 2004; Schmierer and Hill, 2005). It is reported that Exportin 4, which is LMB-insensitive, interacts with Smad3 and promotes its nuclear export. It is of importance to note that the Exportin 4-binding regions locates in the Smad3 MH2 domain and is well conserved among all R-Smads (Kurisaki et al., 2006). It is thus possible that Exportin 4 may represent a general exporter for all Smads. In contrast, our data presented herein clearly suggest that RanBP3 selectively interacts with Smad2/3, through the cooperative binding to both the MH1 and MH2 domains, and specifically mediates the export of Smad2/3 out of the nucleus.
The functional consequence of RanBP3-dependent nuclear export of Smad2/3 is also assessed in our studies. The activity of RanBP3 controls the nuclear Smad2/3 level, which correlates well with the strength of the TGF-β signaling. Despite a few studies on Smad nuclear export, the physiological consequences of Smad nuclear export in TGF-β signaling have rarely been examined. It is reasonable to speculate that TGF-β signaling should be short-lived or impaired when Smad2/3 proteins are forced out of the nucleus. Indeed, overexpression of RanBP3 dampens Smad2/3-mediated transcriptional responses both in mammalian cells and in Xenopus embryos (Figures 1 and and2).2). Conversely, knockdown of RanBP3 expression causes retention of Smad2/3 in the nucleus and enhances TGF-β responses. Similar outcome can be achieved by disrupting the endogenous RanBP3-Smad2/3 interaction (Figure 7). Consistent with its inability to bind to Smad1 (and presumably Smad5 and Smad8), RanBP3 has no effects on the BMP-induced transcriptional responses in mammalian cells and in Xenopus embryos (Figures 1D and 1I). This further supports the notion that distinct mechanisms underlying R-Smad nuclear export operate in TGF-β and BMP pathways.
The observation that inhibition of RanBP3 enhances TGF-β signaling seems to contradict with a previous model, which implicates that continuous nucleocytoplasmic shuttling is required for continuous TGF-β signaling (Schmierer and Hill, 2005). The simplified model has not taken into account of the newly synthesized Smads (that can transmit signals into the nucleus) and the turnover of existing Smads. In the RanBP3 knockdown cells, there is significant accumulation of dephosphorylated Smad2/3. Could the dephosphorylated Smad2/3 in the nuclear be partially active? Indeed, unphosphorylated Smad2 exhibits some transcriptional activity on in vitro chromatin templates (Ross et al., 2006). Alternatively, dephosphorylated Smad2/3 could be re-phosphorylated by unknown kinase(s) in the nucleus; however, this seems unlikely based on our observation that knockdown of RanBP3 did not dramatically increase the level of P-Smad2/3 in the nucleus (Figure 3D). Thus, the exact underlying mechanism for how accumulation dephosphorylated of Smad2/3 in the nucleus ultimately leads to increased TGF-β signaling remains to be further elucidated.
Our studies suggest that dephosphorylation of Smad2 and Smad3 by PPM1A (and/or other Smad phosphatases) precedes their nuclear export by RanBP3. However, a question arises on how RanBP3 facilitates nuclear export of Smad2/3. RanBP3 can partner with CRM1 in some cases, and it also binds weakly to nuclear pore components in vitro (Englmeier et al., 2001) as well as the small GTPase Ran (Mueller et al., 1998; Englmeier et al., 2001). Since inhibition of CRM1 by LMB does not promote nuclear accumulation of Smad2/3 (Xu et al., 2002; Kurisaki et al., 2006), the chance that CRM1 is involved in RanBP3-mediated export of Smad2/3 is rather slim. Ranbp3-R can disrupt the Ranbp3-Smad2/3 interaction, dominantly inhibit Smad2 export and enhance TGF-β transcriptional responses (Figures 7C and 7D), suggesting that the N-terminal region of RanBP3 is apparently needed for the export process. As the N-terminal sequences contain FXFG motifs that are homologous to those in certain nucleoporins, we favor the idea that RanBP3 may act directly as a nuclear exporter for Smad2/3 (Figure 7E). This notion is supported by recent discoveries that RanBP3 enhances nuclear export of β-catenin to terminate Wnt-signaling (Hendriksen et al., 2005) and is a negative regulator for STAT92E nuclear accumulation in Drosophila (Baeg et al., 2005). The similar dependence of Smad2/3 and β-catenin on RanBP3 for nuclear export suggests that RanBP3 may represent a new class of nuclear exporter, which regulates the localization of key signaling components to determine the duration and the strength of the signals in diverse signaling pathways.
Details on plasmids and antibodies used/made in the study are provided in the online supplemental materials.
Human HaCaT, HEK293T and HepG2 cells were grown as described previously (Feng et al., 1998; Feng et al., 2000; Lin et al., 2006). HaCaT cells stably expressing EGFP-Smad2 were kindly provided by Caroline Hill (Nicolas et al., 2004). C2C12 cells were maintained as undifferentiated myoblasts in Dulbecco's modified Eagle's medium with high glucose supplemented with 10% fetal bovine serum. HaCaT cells were transfected with SuperFect (Qiagen), HEK293T cells and HepG2 cells with Lipofectin (Invitrogen) and C2C12 cells with LipofectAmine (Invitrogen).
Reporter plasmids SBE-Luc, p21-Luc (gift of Bert Vogelstein), p15-Luc (gift of Xiao-Fan Wang) and p800-Luc (gift of David Luskutoff) were used to measure TGF-β-induced transcription. Plasmid Id1-Luc (gift of Peter ten Dijke) was used to measure BMP-induced transcription. pSVβgal (Promega), which expresses β-galactosidase under the control of the SV40 early promoter, was co-transfected to allow normalization of transfection efficiency. The total amount of transfected DNA was adjusted to the same amount by addition of vector DNA when necessary. Reporter assays were carried out as described (Feng et al., 2000). Briefly, 20 to 24 h after transfection, cells were treated with TGF-β1 (2 ng/ul) or BMP2 (50 ng/ul) for 24 h. Cells were then harvested for measurement of luciferase and β-galactosidase activities. All assays were done in duplicates and all values were normalized for transfection efficiency against β-galactosidase activities.
The target sequences of shRNA-494 (nt 494-512 of coding region, AGAGCCCCAGAAAAATGAG) and shRNA-676 (nt 676-694, CCGCAACGAACTATTTCCT) (targeting both human RanBP3a and RanBP3b) were previously described (Hendriksen et al., 2005). Annealed sense and antisense DNA oligonucleotides were cloned into pSRG vector (Lin et al., 2006). Stable HaCaT cell lines expressing shRNA-494/676 or pSRG vector were generated and selected with 2 ug/ul of puromycin.
Glutathione S-transferase fusion proteins were prepared using a commercial kit (Amersham Pharmacia Biotech). In vitro translated (TNT kit, Promega) [35S]-labeled protein was pre-cleared with 5 ug of GST protein for 1 h and then incubated with 5 ug of various GST-fusion proteins for 2 h in the binding buffer (50 mM Tris•HCl [pH7.5], 120 mM NaCl, 2 mM EDTA, 0.1% NP40). [35S]-labeled proteins bound to GST-fusion proteins were retrieved on glutathione-Sepharose beads, separated by SDS-PAGE, and visualized by autoradiography. For comparing the binding of phosphorylated or unphosphorylated Smad2 to RanBP3, 500 ng of recombinant Smad2MH2 or phosphorylated Smad2MH2 (aa 241–467) (Lin et al., 2006) were incubated with GST-RanBP3. RanBP3-bound Smad2MH2 was examined via Western blot analysis.
HaCaT parental or its stable cells were treated with SB431542 with indicated time periods in the text and harvested with fractionation buffer (10 mM HEPES [pH7.9], 1.5 mM MgCl2, 10 mM KCl, 0.5 mM DTT, 0.5% NP-40 and protease inhibitors) for 20 min on ice, and then fractionated by centrifugation. The fractions were analyzed on SDS-PAGE and subsequent Western blot analysis (Figure 3D).
HaCaT cells stably expressing GFP-Smad2 cells were first treated with TGF-β to induce nuclear accumulation of Smad2 (Figure 4A & 4B). Cells were permeabilized with digitonin on ice for 5 min, and then thoroughly washed 3 times with a cold transport buffer (20 mM HEPES [pH7.3], 110 mM potassium acetate, 5 mM sodium acetate, 2 mM DTT, 1 mM GTP, 1.0 mM EGTA and proteinase inhibitor). Cells were incubated for 60 min at 30°C in the transport buffer with addition of ATP regeneration system (Sigma; 1 mM ATP, 5 mM creatine phosphate, and 20 U/ml creatine phosphokinase) and RanBP3 or Ran. RanBP3 proteins were prepared from GST-RanBP3 by removing GST with Precission (Amersham), and Ran proteins were enriched by anti-Flag immunoprecipitation of cell lysates from HEK293T cells expressing Flag-Ran and eluted with Flag-peptide (Sigma). Cells were then rinsed 3 times with the cold transport buffer. Excess buffer was removed and permeabilized cells were resolved in SDS-loading buffer. The level of GFP-Smad2 was then analyzed by Western blot analysis.
A quantitative analysis of nuclear transport of Smad2 was previously described (Xu et al., 2002; Lin et al., 2006), using MS2 coat protein-fused Smad2 (gift of Joan Massagué) and pDM128/8xMS2 export reporter system (gift of Bryan Cullen) (Cullen, 2004). Briefly, HEK293T cells were transfected with plasmids for MS2-Smad2, CAT reporter pDM128/8xMS2 and β-galactosidase plasmid (for normalization) and additional factors under examination. 45 h after transfection, cell lysates were analyzed for CAT activity using ELISA-based assay (Roche). Relative export activity is calculated by normalized the CAT activity with β-galactosidase activity in each samples.
DNA pulldown assay was essentially carried out as previously described (Feng et al., 2000; Feng et al., 2002). Briefly, HaCaT parental or its stable cells were treated with or without TGF-β for 2 h and collected in lysis buffer (10 mM HEPES pH 7.5, 150 mM NaCl, 1 mM MgCl2, 0.5 mM EDTA, 0.5 mM DTT, and 0.1% NP-40, 10% glycerol). The cell lysates were incubated with 30 pmol biotinylated SBE oligonucleotides and 5 ug of poly(dI-dC) at 4°C for 16 hours. DNA-bound proteins were collected with streptavidin-beads (Dynabeads M-280 Invitrogen) for 2 h, washed extensively with the lysis buffer, and examined by Western blotting.
Endogenous or epitope-tagged proteins were immunoprecipitated from cell lysates by appropriate antibody affinity gel as indicated in the text and figure legends. After extensive washes, immunoprecipitated proteins were eluted in SDS sample loading buffer (Bio-Rad), separated by SDS-PAGE, transferred to nitrocellulose (Pierce), and detected in Western blots with appropriate primary antibodies coupled with horseradish peroxidase-conjugated secondary antibody by chemiluminescence (Pierce).
Cells were grown in 96-well plates (3,000 cells per well) in cell culture medium supplemented with 0.2% fetal bovine serum (FBS) for overnight before TGF-β treatment. Cell proliferation was measured in duplicates for 3 days. Each harvest was incubated with 20 ul of CellTiter 96 AQueous One Solution Reagent (Promega) for 3 h at 37°C in the 5% CO2 incubator. The absorbance at 490nm was obtained using the VersaMax microplate reader (Molecular Devices).
RNAs used for microinjection into Xenopus embryos were synthesized with SP6 RNA polymerase using mMessage mMachine kit (Ambion). DNA templates were: AscI-linearized pCS105-RanBP3, pCS105-RanBP3-wv, EcoRI-linearized pSP64T-activin, and AscI-linearized pCS105-BMP4. RNAs were injected alone or in combination into the animal poles of two-cell stage embryos. The doses of RNAs used were RanBP3 or RanBP3-wv (1 ng), activin (1 pg), and BMP4 (20 pg). Ectodermal explants (animal caps) from injected embryos were obtained at blastula stages (stage 9) and incubated to gastrula stages (stage 11), at which time total RNA was extracted from these caps, and the gene expression pattern was analyzed by reverse transcription (RT)-PCR. The primers used in RT-PCR were as described (Chang et al., 1997).
qRT-PCR was carried out using Assay-on-Demand kits (Applied Biosystem) and previously described (Lin et al., 2006). Briefly, total RNAs were prepared using TriZol reagent (Invitrogen) from HaCaT cells. mRNA levels of target genes p15, p21 and PAI-1 were normalized against 18S RNA. Each target was measured in triplicates.
To set up a BiFC system (Hu et al., 2002), we created two expression plasmids pXF1YN and pXF1YC that contain aa 1-154 of enhanced YFP (designated YN) or 155-238 (designated YC), respectively. YN or YC-tagged Smads or RanBP3b were constructed by inserting the corresponding coding sequences into EcoRI and SalI sites of pXF1YN or pXF1YC. These expression plasmids were transfected into HaCaT cells, and 48 h after transfection, the fluorescence production was detected with a Zeiss Axiovert 200M microscope.
Cells grown on coverslips were fixed with 4% formaldehyde for 30 min at 4°C, followed by 0.5% Triton X-100 treatment for 30 min and blocked with 5% milk in PEM buffer (400 mM Potassium PIPES [pH 6.8], 0.8 mM EGTA, 5 mM MgCl2). Cells were then probed with primary antibodies, followed by Alexa Fluor 546 or Alexa Fluor 594 secondary antibody (Molecular Probes). Fluorescence images were acquired with a Zeiss Axiovert 200M microscope.
We thank Bryan Cullen, Maarten Fornerod, Caroline S. Hill, Mien-Chie Hung, Tom Kerppola, Ed Leof, Joan Massagué, Yigong Shi, Peter ten Dijke, and Bert Vogelstein for numerous reagents that make this study possible. Thanks also go to Hongyan Xu for helpful discussion with statistics analysis, Feng Liu for technical assistance, and members of Feng and Lin labs for continuous stimulating discussion. This research was supported by NIH grants (R01AR053591 and R01CA108454 to X.-H.F., R01DK073932 to X.L., R01HD43345 to C.C.). X.-H.F is a Leukemia and Lymphoma Society Scholar.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.