Requirement for the Dynein Light Chain km23-1 in a Smad2-dependent Transforming Growth Factor-β Signaling Pathway*

We have identified km23-1 as a novel transforming growth factor-β (TGFβ) receptor (TβR)-interacting protein that is also a light chain of the motor protein dynein (dynein light chain). Herein, we demonstrate by sucrose gradient analyses that, in the presence of TGFβ but not in the absence, km23-1 was present in early endosomes with the TβRs. Further, confocal microscopy studies indicate that endogenous km23-1 was co-localized with endogenous Smad2 at early times after TGFβ treatment, prior to Smad2 translocation to the nucleus. In addition, immunoprecipitation/blot analyses showed that TGFβ regulated the interaction between endogenous km23-1 and endogenous Smad2 in vivo. Blockade of km23-1 using a small interfering RNA approach resulted in a reduction in both total intracellular Smad2 levels and in nuclear levels of phosphorylated Smad2 after TGFβ treatment. This decrease was reversed by lactacystin, a specific inhibitor of the 26 S proteasome, suggesting that knockdown of km23-1 causes proteasomal degradation of phosphorylated (i.e. activated) Smad2. Blockade of km23-1 also resulted in a reduction in TGFβ/Smad2-dependent ARE-Lux transcriptional activity, which was rescued by a km23-1 small interfering RNA-resistant construct. In contrast, a reduction in TGFβ/Smad3-dependent SBE2-Luc transcriptional activity did not occur under similar conditions. Furthermore, overexpression of the dynactin subunit dynamitin, which is known to disrupt dynein-mediated intracellular transport, blocked TGFβ-stimulated nuclear translocation of Smad2. Collectively, our findings indicate for the first time that a dynein light chain is required for a Smad2-dependent TGFβ signaling pathway.

TGF␤ is the prototype of a large family of structurally related growth and differentiation factors that initiates its signals from a receptor complex consisting of TGF␤ RI (T␤RI) and TGF␤ RII (T␤RII) serine/threonine kinase receptors (20 -24). Activated T␤RII recruits, phosphorylates, and activates T␤RI. Then, the activated receptor complex can phosphorylate Smads 2 and 3, and these receptor-activated Smads (RSmads) then form a complex with Smad4. The TGF␤-activated, heteromeric Smad complexes are translocated to the nucleus, where they induce or repress transcription of defined genes (20, 24 -27). Additional data indicate that the interactions among T␤Rs, Smads, adaptor/scaffolding proteins, and cytoskeletal elements represent important regulatory mechanisms in TGF␤ signaling (26,28).
We have shown that T␤RII is absolutely required for phosphorylation of the DLC km23-1, as well as for the recruitment of km23-1 to the rest of the dynein motor through the DIC (13). Further, km23-1 undergoes rapid phosphorylation on serine residues after T␤R activation, in keeping with the kinase specificity of the T␤Rs (13). Moreover, specific mutants of km23-1 block km23-1 binding to the DIC and disrupt TGF␤-mediated transcriptional events (29,30). In addition, consistent with a role for km23-1 in TGF␤ signaling, small interfering RNA (siRNA) blockade of km23-1 expression resulted in a decrease in specific TGF␤-mediated cellular responses, including an induction of fibronectin expression and an inhibition of cell cycle progression (31).
Because we have shown that km23-1 is required for mediating specific TGF␤ responses, and because it is well established that the Smads are key TGF␤ signaling components, here we investigated the role of km23-1 in controlling Smad compartmentalization and transcriptional activation. We provide the first evidence that TGF␤ can induce the interaction of endogenous km23-1 with endogenous Smad2. Further, we show that endogenous km23-1 and endogenous Smad2 are co-localized in a TGF␤and time-dependent manner, prior to Smad2 translocation to the nucleus. Endogenous km23-1 was also localized in early endosomal compartments with the T␤Rs after TGF␤ treatment. siRNA-specific blockade of km23-1 resulted in a depletion of intracellular Smad2, which was partially blocked by the proteasomal inhibitor lactacystin, suggesting that 26 S proteasomal degradation of Smad2 can occur in the absence of km23-1. In keeping with these results, blockade of km23-1 also reduced TGF␤/Smad2-dependent transcriptional regulation. Finally, we demonstrate for the first time that dynein-dependent intracellular events are required for Smad2 nuclear translocation after TGF␤ treatment, because overexpression of dynamitin inhibited these effects. Thus, our results demonstrate for the first time that the DLC km23-1, as well as dynein motor activity, are required for a Smad2-dependent TGF␤ signaling pathway.
Cell Culture-Mv1Lu (CCL-64) cells were purchased from the American Type Culture Collection (Manassas, VA) and were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. Madin-Darby canine kid-ney cells (CCL-34) were also obtained from ATCC and were grown in Minimum Essential Medium-␣ supplemented with 10% fetal bovine serum. Cultures were routinely screened for mycoplasma using Hoechst 33258 staining (13).
Sucrose Gradient Assays-Madin-Darby canine kidney cells were plated at 1.5 ϫ 10 4 cells/cm 2 in 10-cm plates. Twenty-four hours after plating, the medium was replaced with serum-free minimum essential-␣ medium. Thirty minutes after incubation, Madin-Darby canine kidney cells were cultured in the absence or presence of TGF␤ (5 ng/ml) for 5 min (ten 10-cm plates each). Early-endosome-containing fractions were then prepared as described previously (36).
siRNAs-km23-1 siRNA and the negative control (NC siRNA) were prepared as described previously (31). The siRNA was designed in a region of km23-1 where the mink and human forms do not differ.
Immunofluorescence Microscopy Analyses-For km23-1 and Smad2 co-localization experiments, Mv1Lu cells were fixed with 4% paraformaldehyde in phosphate-buffered saline for 20 min at room temperature, and permeabilized with 0.5% Triton X-100 in phosphate-buffered saline for 5 min. Subsequently, these cells were incubated with km23-1 rabbit anti-serum (1:200) and 5 g/ml anti-Smad2 monoclonal Ab for 1 h, respectively. The bound primary antibodies were visualized with 2 g/ml Alexa Fluor 488-conjugated goat anti-rabbit IgG (green) and cy3-conjugated goat anti-mouse IgG (red). Co-localization of km23-1 and Smad2 is indicated by a yellow color (merge). Images were collected with a Leica TCS SP2 AOBS confocal microscope. The images in supplemental Fig. S1 were de-convoluted using Huygens Essential software from Scientific Volume Imaging (Exton, PA). Co-localization of km23-1 and Smad2 puncta was quantified using the co-localization function in Image Pro Plus 4.1 software (Media Cybernetics, Inc., Silver Spring, MD). A similar approach has been used to quantify the co-localization of other proteins as described previously (37). At least five cells in each group from each double-labeled experiment were analyzed for co-localization of km23-1 and Smad2. For the immunofluorescence analyses to study the effects of siRNAs on nuclear expression and translocation of Smad2 by TGF␤, the cells were fixed and permeabilized as for the co-localization studies. Subsequently, these cells were incubated with 5 g/ml anti-Smad2 monoclonal Ab for 2 h, and then the bound Ab was visualized with 2 g/ml Alexa 594 goat anti-mouse IgG. Immunofluorescence images were captured using a Nikon Diaphot microscope with a Retiga 1300 chargecoupled device camera (BioVision Technologies, Inc., Exton, PA) running IPLab v3.6.3 software (Scanalytics, Inc., Fairfax, VA). One hundred green fluorescence protein (GFP)-positive cells were counted for cultures of both km23-1 siRNA-transfected and NC siRNA-transfected cells. 4Ј,6-Diamidino-2-phenylindole staining designates individual cells. Triplicate fields are shown for each condition. For the immunofluorescence analyses to study the effects of overexpression of dynamitin on nuclear translocation of Smad2 by TGF␤, the cells were analyzed as for the studies of the effects of siRNAs on nuclear expression and translocation of Smad2, except that the cells were co-transfected with GFP and empty vector (EV) or dynamitin.
Cellular Fractionation-The NE-PER Nuclear and Cytoplasmic Extraction Reagent kit (78833, Pierce) was used to fractionate Mv1Lu cells according to the manufacturer's protocol.
Luciferase Reporter Assays-Mv1Lu cells were plated at 1 ϫ 10 4 cells/cm 2 in 12-well plates. Twenty-four hours after plating, the cells were transfected with the indicated amounts of either km23-1 siRNA or NC siRNA, together with the activin-responsive element (ARE)-Lux and FAST-1 (38), or the Smad-binding element (SBE) 2 -Luc (39). Renilla was used to normalize transfection efficiencies, and pcDNA3.1 was used to normalize the amount of total DNA transfected as described previously (40). Twenty-four hours after transfection, the medium was replaced with serum-free Dulbecco's modified Eagle's medium. 1 h after incubation, Mv1Lu cells were cultured in the absence or presence of TGF␤ (5 ng/ml) for an additional 18 h. Luciferase activity was measured using Promega's Dual-luciferase Reporter Assay System following the manufacturer's instructions. All assays were performed in triplicate. Data are expressed as mean Ϯ S.E.

RESULTS
Previous reports have shown that the early endosome pathway plays a critical role in T␤R endocytosis and subsequent TGF␤ signal transduction (42)(43)(44)(45)(46). Specifically, the T␤Rs are known to be internalized with Smad2 into early endosomes within minutes of TGF␤ addition to cells (42,45). Further, our previous data suggested that T␤RII kinase activity was required for the ability of the DLC km23-1 to bind the dynein motor through the DIC, as well as for TGF␤ responses downstream (13). Thus, the DLC km23-1 may recruit early endosomal TGF␤ signaling complexes during intracellular transport and downstream effects, following T␤R endocytosis. Accordingly, it was of interest to examine whether endogenous km23-1 might be co-localized with the T␤Rs in early endosomes after TGF␤ treatment. To assess this, we performed sucrose flotation gradients to isolate endosomal compartments enriched for early endosome antigen-1 (EEA1) (36), followed by Western blotting with km23-1-specific rabbit anti-serum or T␤RI/RII Abs as described under "Materials and Methods." As expected, in the absence of TGF␤ (left panel), the majority of T␤RII (top panel) and T␤RI (second panel) were present in fractions 6 -8. However, upon TGF␤ activation (right panel), the amount of T␤RII and T␤RI present in EEA1-enriched fractions was increased (fractions 4 and 5), consistent with a previous report (45). The bottom panel indicates the localization of EEA1, designating the fractions containing early endosomes. In terms of km23-1 localization (third panel), in the absence of TGF␤ (left panel), km23-1 was not present in the early endosomal fractions (fractions 4 and 5). Instead, the majority of km23-1 accumulated in fractions 6 -8. However, as early as 5 min after TGF␤ addition (right panel), km23-1 was present in the EEA1-enriched early endosomal fractions (fractions 4 and 5). Moreover, even upon a much longer exposure (not shown) of the km23-1 expression data in the absence of TGF␤ (Fig. 1, left), no km23-1 was detectable in the EEA1-enriched early endosomal fractions. We also noticed that total km23-1 levels were higher in the presence of TGF␤. However, using EEA1 levels as an expression control to compare two different gradient runs (42), it was clear that the EEA1 levels were also higher in the presence of TGF␤, suggesting that the increase in km23-1 was due to differential loading. Thus, the results in Fig. 1 demonstrate that endogenous km23-1 is present in early endosomes with endogenous T␤Rs in the presence of TGF␤.
Because Smad2 is a critical intracellular mediator of TGF␤ responses (47,48), co-localized with T␤RII in EEA1-positive early endosomes after T␤Rs endocytosis (42,45), it is conceivable that km23-1 might be co-localized with Smad2 in a punctate staining pattern, indicative of the co-existence of km23-1 and Smad2 in endosomal compartments. Thus, we performed immunofluorescence studies using confocal microscopy after TGF␤ treatment of TGF␤-responsive Mv1Lu cells, using km23-1-specific rabbit anti-serum (left panels) or a Smad2 Ab (middle panels), respectively. As indicated in Fig. 2, in the absence of TGF␤ (top panel), Smad2 (middle panel) was concentrated in cytoplasmic punctate vesicular structures, consistent with a previous report (49). Similarly, endogenous km23-1 displayed a punctate staining pattern that was present throughout the cytoplasm in the absence of TGF␤ (left, top panel). Colocalization is shown as a yellow color (right, top panel), and was quantified using Image Pro Plus 4.1 software from Media Cybernetics, Inc. The percentages of co-localization were obtained from multiple images as described under "Materials and Methods." In the absence of TGF␤, co-localization of km23-1 with Smad2 was ϳ13%. There was a slight increase in co-localization of km23-1 with Smad2 at 2 min (ϳ18%) after TGF␤ treatment. However, TGF␤ treatment resulted in greater co-localization of km23-1 and Smad2 at 5 min (ϳ28%) after TGF␤ addition to Mv1Lu cells (Fig. 2, third panels). This level of co-localization was similar to that previously reported for quantitation of co-localization of other early endosome proteins (50). In addition, a partial redistribution of km23-1 and Smad2 toward the perinuclear region was observed at 5 min after TGF␤ stimulation. However, once Smad2 had translocated to the nucleus by 15 min after TGF␤ treatment (bottom panel), km23-1 was still localized in the cytoplasm and was no longer co-localized with Smad2. For all studies, the pre-immune serum and relevant IgG controls were negative (data not shown), confirming the specificity of the km23-1 and Smad2 Abs. High quality TIFF files for the data in Fig. 2 can be found in supplemental Fig. S1. Our immunofluorescence results obtained using confocal microscopy clearly indicate that km23-1 and Smad2 are co-localized intracellularly at early times after TGF␤ treatment, prior to entry of Smad2 into the nucleus.
Next, we wished to determine whether endogenous km23-1 and endogenous Smad2 were present in the same complex after TGF␤ treatment. To assess this, we performed immunoprecipitation/blot analyses in the absence or presence of TGF␤. As shown in Fig. 3, TGF␤ induced a rapid interaction of endogenous km23-1 with endogenous Smad2 (lane 3, top panel). The kinetics were similar to those for km23-1 binding to the DIC (13), with some basal interaction, but with increased association at 5 min after TGF␤ addition. In contrast, the association between km23-1 and Smad2 was significantly decreased at 15 min after TGF␤ addition to Mv1Lu cells, at a time when Smad2 is translocated to the nucleus in these cells (lane 4, top panel). The results in this figure depict Smad2-km23-1 interactions in total cell lysates, which would include data from cells that have not yet translocated to the nucleus. Therefore, the higher interaction levels at 15 min in this figure, compared with Fig. 2, might be the result of co-localized km23-1 and Smad2 in the cytoplasm. Overall then, our results are consistent with those shown in Fig. 2 regarding the kinetics for co-localization of km23-1 and Smad2. As expected, there were no bands in the IgG-negative control (lane 1, top panel). Equal loading and FIGURE 2. TGF␤ induces the co-localization of km23-1 with Smad2 prior to Smad2 nuclear translocation. Mv1Lu cells were cultured in the absence or presence of TGF␤ for the indicated times, and then were fixed and permeabilized as described under "Materials and Methods." Endogenous km23-1 was detected using rabbit km23-1 anti-serum, followed by Alexa Fluor 488conjugated goat anti-rabbit IgG (left panel). Smad2 was detected using a mouse monoclonal Smad2 Ab and cy3-conjugated goat anti-mouse IgG (middle panel). The cells were analyzed using a Leica TCS SP2 AOBS confocal microscope at a magnification of 630 with appropriated filter sets. The merge photos show potential co-localization of endogenous km23-1 and endogenous Smad2  expression of endogenous Smad2 was confirmed by re-probing with an anti-Smad2 Ab, as shown in the middle panel. Equal expression of endogenous km23-1 was confirmed by Western blot analysis, as shown in the third panel. These results were scanned by densitometry and are shown in the bottom panel.
Our results indicate for the first time that TGF␤ regulates the interaction between endogenous km23-1 and endogenous Smad2 in vivo in a time-dependent manner, suggesting that km23-1 may provide a novel link between a motor light chain and Smad-dependent TGF␤ signaling.
DLCs have been shown to function in the recruitment of the cargo (i.e. signaling complexes) to the dynein motor for intracellular transport prior to downstream effects (3,51,52). Further, our results in Figs. 1-3 support an association of the km23-1 DLC with Smad2-containing early endosomal signaling complexes. Thus, it was of interest to determine whether disruption of dynein motor activity by dynamitin overexpression would block Smad2 nuclear translocation after TGF␤ treatment. Thus, we performed immunofluorescence studies after transiently transfecting Mv1lu cells with dynamitin or EV in the presence of GFP. The presence of the GFP signal designates the cells that were transfected with dynamitin-myc or EV (left panels, Fig. 4). In the absence of TGF␤, both the EV-transfected, GFP-positive cells and the dynamitin-transfected, GFPpositive cells displayed the same pattern of diffuse punctate staining (data not shown). In contrast, as expected, the EVtransfected, GFP-positive cells displayed largely nuclear expression of Smad2 in response to TGF␤ (top panel, Fig. 4), indicative of ligand-induced translocation of Smad2 to the nucleus. However, in the dynamitin-transfected, GFP-positive cells, Smad2 displayed a diffuse punctate staining pattern, with reduced nuclear expression of Smad2 (bottom panel, Fig. 4). As expected, untransfected (GFP-negative) cells also responded to TGF␤ with significant Smad2 translocation. Our results demonstrate that disruption of the dynein motor complex blocks the ability of Smad2 to reach the nucleus after TGF␤ treatment, suggesting that Smad2 intracellular transport requires dynein.
To quantify the effects of dynamitin overexpression on TGF␤-mediated nuclear translocation of Smad2 from Fig. 4, we counted 100 GFP-positive cells in cultures of either EV-transfected or dynamitin-transfected cells treated with TGF␤ (5 ng/ml). Of these 100 GFP-positive cells, the cells showing nuclear translocation of Smad2 were counted as described under "Materials and Methods." In EV-transfected cells, 94% of the GFP-positive cells displayed Smad2 nuclear translocation in response to TGF␤. As expected, 100% of the GFP-negative cells also responded to TGF␤ with Smad2 nuclear translocation, whether EV or dynamitin had been co-transfected with GFP. However, only 20% of the dynamitin-transfected, GFPpositive cells still displayed Smad2 nuclear translocation after TGF␤ treatment. That is, in the dynamitin-transfected cells, Smad2 displayed a diffuse punctate staining pattern in 80% of the GFP-positive cells. The percentage of disruption of intracellular transport of Smad2 caused by dynamitin overexpression was similar to that observed previously for other dyneindependent events (53)(54)(55). Thus, quantification of our results confirmed what was observed in the immunofluorescence pho-tographs, and suggested that disruption of dynein-dependent functions reduced intracellular transport of Smad2, thereby preventing Smad2 accumulation in the nucleus after TGF␤ treatment.
The data in Fig. 4 indicate that disruption of the dynein motor complex blocked the ability of Smad2 to reach the nucleus after TGF␤ treatment. If the DLC km23-1 is needed to recruit the TGF␤ signaling complexes to the rest of the dynein motor, eventually leading to downstream nuclear events, it might be expected that blockade of endogenous km23-1 would block the transcriptional activation of TGF␤/Smad-dependent target genes in the nucleus. To establish that the km23-1 siRNA could block endogenous km23-1 expression, we transiently transfected Mv1Lu cells with km23-1 siRNA or NC siRNA as described previously. Western blot analysis was then performed as shown in Fig. 5A. Transfection with km23-1 siRNA (lanes 4 -6) resulted in a marked decrease in endogenous km23-1 levels compared with controls (lanes 1-3). In addition, we have previously shown that km23-1 siRNA could specifically knock down km23-1 expression in two different epithelial cell lines (31).
Because km23-1 siRNA could specifically knock down endogenous km23-1 expression, we transiently transfected Mv1Lu cells with either km23-1 siRNA or NC siRNA, and then performed ARE-Lux luciferase reporter assays in the absence FIGURE 4. The dynein motor is required for TGF␤-mediated Smad2 nuclear translocation. Mv1Lu cells were transiently co-transfected with either GFP and EV, or GFP and dynamitin-myc. Twenty-four hours after transfection, cells were incubated in serum-free medium before addition of TGF␤ (5 ng/ml) for 15 min. Cells were fixed, and endogenous Smad2 was detected using a mouse monoclonal Smad2 Ab and Alexa 594 goat anti-mouse IgG (red). 4Ј,6-Diamidino-2-phenylindole (DAPI) staining permitted visualization of nuclei of individual cells (blue). GFP was used as a marker to designate cells transfected with siRNA (green). The cells were analyzed by a Nikon Diaphot microscope at a magnification of 400 with appropriated filter sets. Duplicate fields are shown for each condition. and presence of TGF␤. The ARE-lux reporter was previously shown to be activated by TGF␤ or activin in a Smad2-dependent manner (38). As shown in Fig. 5B, TGF␤ induced ARE-lux activity in the EV and NC siRNA cells. In contrast, the cells transfected with km23-1 siRNA displayed a dose-dependent decrease in the fold induction of ARE-lux activity by TGF␤ with increasing doses of km23-1 siRNA, relative to NC siRNA. Although 0.1 g/well of km23-1 siRNA reduced TGF␤-inducible ARE-lux activity to levels that were 67% of control values, higher concentrations of km23-1 siRNA resulted in greater reductions in the -fold induction by TGF␤ (to levels that were 44 and 23% of NC siRNA values, respectively). Thus, km23-1 is required for TGF␤ induction of Smad2-dependent transcriptional activity.
Thus far, our data have focused on the role of km23-1 in mediating Smad2-specific events. It was also of interest to determine whether blockade of km23-1 would affect Smad3-dependent TGF␤ transcriptional responses. Thus, we examined the effects of km23-1 siRNA on the Smad3-specifc SBE2-Luc luciferase reporter (39) in the absence and presence of TGF␤, after transiently transfecting Mv1Lu cells with either km23-1 siRNA or NC siRNAs. The results in Fig. 5D demonstrate that blockade of km23-1 had no effect on Smad3-dependent transcriptional activation, indicating that km23-1 is relatively specific for TGF␤/Smad2-dependent transcriptional activation. These findings further support a specific role for km23-1 in mediating TGF␤and Smad2-dependent TGF␤ signaling events.
Because our results have shown that intracellular transport of Smad2 is dynein-dependent, and that blockade of km23-1 specifically inhibited TGF␤/Smad2dependent transcriptional activity, it was of interest to determine whether blockade of endogenous km23-1 would block Smad2 nuclear translocation after TGF␤ treatment. Accordingly, we performed immunofluorescence studies to examine TGF␤-dependent Smad2 translocation to the nucleus in individual cells after siRNA knockdown of km23-1 (Fig. 6). Mv1Lu cells were transiently transfected with either NC siRNA or km23-1 siRNA in the presence of GFP. The presence of the GFP signal designates the cells that were transfected with the relevant siRNAs (left panels, Fig. 6). In the absence of TGF␤, the NC siRNA-transfected cells and the km23-1 siRNA-transfected cells displayed the same pattern of diffuse punctate staining (Fig. 6A). In contrast, as expected, the NC siRNA-transfected, GFP-positive cells displayed largely nuclear expression of Smad2 in response to TGF␤ (Fig. 6B), indicative of ligand-induced translocation of Smad2 to the nucleus. However, in the km23-1 siRNA-transfected, GFP-positive cells, Smad2 expression was barely detectable in the nucleus (Fig. 6C). As expected, untransfected (GFP-negative) cells also responded to TGF␤ with significant Smad2 translocation. Our results demonstrate that knockdown of km23-1 results in a decrease in nuclear expression of Smad2 after TGF␤ treatment.
To quantify the siRNA effects on TGF␤-mediated nuclear  Table 1, in the NC siRNA-transfected cells, 93% of the GFP-positive cells displayed Smad2 nuclear expression in response to TGF␤. In contrast, in the km23-1 siRNA-transfected cells, Smad2 expression was barely detecta-ble in the nucleus of GFP-positive cells, with only 11% of the km23-1 siRNA, GFP-positive cells still displaying Smad2 nuclear expression. As expected, 100% of the GFPnegative cells responded to TGF␤ with increased Smad2 expression, whether the NC or the km23-1 siRNA had been co-transfected with GFP. Thus, the quantitation of our immunofluorescence results confirmed that nuclear Smad2 levels were specifically reduced by km23-1 siRNA in the presence of TGF␤.
As an independent method of verifying whether km23-1 knockdown reduced TGF␤-mediated nuclear expression of Smad2, we performed Western blot analyses after subcellular fractionation of the cells as described under "Materials and Methods." As shown in Fig. 6D 4, top panel). However, in the km23-1 siRNA-transfected cells, phospho-Smad2 levels in the nuclear fraction were significantly decreased after TGF␤ treatment (lane 6, top panel). Expression of lamins A/C demonstrate equal loading of nuclear extracts (bottom panel) (56). Thus, the results in Fig. 6D further demonstrate that blockade of km23-1 results in a decrease of phosphorylated Smad2 in the nuclear fraction in the presence of TGF␤, consistent with the results obtained in the immunofluorescence analyses.
It is noteworthy when comparing Fig. 6 (B and C) that a corresponding increase in cytoplasmic Smad2 was not observed, when nuclear Smad2 expression was blocked by the km23-1 siRNA in the presence of TGF␤. This finding would suggest that degradation of Smad2 might occur when km23-1 functions are blocked. This effect was not observed upon overexpression of dynamitin (Fig. 4). Because a previous report has shown that TGF␤-activated Smad2 can be degraded through the ubiquitin-proteasomal-degradation pathway (57), it was of interest to determine whether blockade of this degradation pathway would reverse the km23-1 siRNA-mediated blockade of the TGF␤-dependent nuclear accumulation of Smad2. In addition, previous reports have shown that cells treated with the ubiquitin proteasomal degradation inhibitor lactacystin Twenty-four hours after transfection, cells were fixed and endogenous Smad2 was detected using a mouse monoclonal Smad2 Ab and Alexa 594 goat anti-mouse IgG (red). 4Ј,6-Diamidino-2-phenylindole (DAPI) staining permitted visualization of individual cells (blue). GFP was used as a marker to designate cells transfected with siRNA (green). The cells were analyzed by a Nikon Diaphot microscope at a magnification of 1000 with appropriated filter sets. Triplicate fields are shown for each condition. B, Mv1Lu cells were transiently cotransfected with GFP and NC siRNA. Twenty-four hours after transfection, cells were incubated in serum-free medium before addition of TGF␤ (5 ng/ml)) for 15 min, and then analyzed as in A. C, Mv1Lu cells were transiently co-transfected with GFP and km23-1 siRNA. Twenty-four hours after transfection, cells were incubated in serum-free medium before addition of TGF␤ (5 ng/ml)) for 15 min, and then analyzed as in A. The cells marked by an arrowhead in Fig. 3C display barely detectable nuclear Smad2 expression. D, Mv1Lu cells were either mock transfected or transiently transfected with either NC siRNA or km23-1 siRNA. Twenty-four hours after transfection, cells were incubated in serum-free medium before addition of TGF␤ (5 ng/ml)) for the indicated times, followed by cell fractionation as described under "Materials and Methods." Top panel, nuclear fractions were subjected to SDS-PAGE (15%), transferred to a polyvinylidene difluoride membrane, and blotted with a rabbit phospho-Smad2 Ab. Bottom panel, membrane was then re-probed with an anti-Lamin A/C Ab as a nuclear marker. The results shown are representative of two similar experiments. remained fully viable for 8 h after treatment and that, generally, after a 2-h exposure, a 50% inhibition was obtained at 1-10 M (58). Moreover, pretreatment of Mv1Lu cells with lactacystin (10 M) for 8 h before cell lysis was shown to cause a significant inhibition of 26 S proteasome degradation (59). Thus, we chose lactacystin (10 M) pretreatment for 8 h for these experiments. Mv1Lu cells were transiently transfected either NC siRNA or km23-1 siRNA in the absence or presence of TGF␤ (5 ng/ml) and/or lactacystin (10 M). As expected for the NC siRNAtransfected cells, TGF␤ induced a rapid increase in phosphorylated Smad2 in the nuclear fraction (Fig. 7A, lanes 2-4, left  panel). Phosphorylated Smad2 was detectable within 5 min of TGF␤ treatment and continued increasing for at least 15 min. After addition of lactacystin to the NC siRNA-transfected cells, the levels at 5-15 min after TGF␤ treatment were slightly higher than for TGF␤ treatment alone (lanes 6 -8, left panel). Consistent with the results in Fig. 6 (C and D), in the km23-1 siRNA-transfected cells, phospho-Smad2 levels in the nuclear fraction were significantly decreased at all time points after TGF␤ treatment (lanes 2-4, right panel), with respect to those for the NC siRNA. This decrease was reversed by the proteasomal inhibitor lactacystin (lanes 7 and 8, right panel). To quantify these results, Western blots from two independent experiments were scanned by densitometry, and the results were expressed graphically (Fig. 7B). As shown in this figure, lactacystin treatment resulted in higher levels of phospho-Smad2, especially for the cells receiving km23-1 siRNA. Thus, blockade of ubiquitin proteasomal degradation prevented the loss of phospho-Smad2 that occurs when km23-1 is blocked. Taken together, our results indicate that a proteasomal degradation mechanism is responsible, at least in part, for the reduced levels of TGF␤-activated (i.e. phosphorylated) Smad2 that are observed when km23-1 expression is knocked down.

DISCUSSION
km23-1 was previously identified to be both a T␤R-interacting protein and a light chain of the motor protein dynein (13). Further, kinase-active T␤Rs were shown to be required for km23-1 phosphorylation and for recruitment of km23-1 to the dynein motor complex through the DIC (13). In addition, blockade of km23-1 is known to reduce specific TGF␤ responses downstream (31). These previous results suggested that, subsequent to T␤R activation and endocytosis, TGF␤ signaling components such as Smads might represent one type of cargo that could be transported intracellularly by dynein-dependent mechanisms, involving the DLC km23-1. In the current report, we demonstrate for the first time that dynein motor activity is required for TGF␤-dependent Smad2 accumulation in the nucleus. In this regard, we show that overexpression of dynamitin, which is known to disrupt dynein-dynactin functions and to block dynein-dependent intracellular transport, blocked Smad2 nuclear accumulation after TGF␤ treatment. We also describe herein the novel findings that after TGF␤ treatment, km23-1 is present in early endosomes with T␤Rs, and it is co-localized with Smad2 prior to Smad2 translocation to the nucleus. Further, we report that km23-1 and Smad2 are present in the same complex after TGF␤ treatment and that knockdown of km23-1 reduces both nuclear levels of phospho-Smad2, as well as Smad2-dependent ARE-Lux transcriptional activity. Collectively, our results provide the first evidence that dynein motor activity and the DLC km23-1 are required for Smad2-dependent TGF␤ signaling events.
Although this is the first report of a positive role for a DLC in a TGF␤-and Smad-dependent signaling pathway, several examples of motor protein light chain regulation of other signaling pathways have been reported previously (9,10,60,61). For example, it has been reported that kinesin light chain 1 is the link between kinesin motor proteins and the c-Jun NH 2terminal kinase (JNK)-interacting proteins, motor receptors known to be important for JNK and p38 mitogen-activated protein kinase signaling (60 -62). In addition, JNK-associated leucine zipper protein was shown to serve as a link between the kinesin motor proteins and their cargo, namely JNK signaling components (63). Similarly, light chains for the motor protein dynein have been shown to regulate the movement of signaling complexes along MTs (3,51). For example, the DLC Tctex-1 (DYNLT1) has been shown to associate with the Trk neurotrophin receptors for the transport of neurotrophins during vesicular trafficking, an effect that is thought to result from the direct interaction between the Trk receptors and the dynein motor machinery (64). Further, DLC1 (LC8, DYNLL1) has been shown to have a facilitation role in the nuclear translocation of the estrogen receptor in breast cancer cells (65). In addition, recent evidence suggests that the interaction of DLC1 (DYNLL1) with the RasGRP3 exchange factor for Ras-like small GTPases could play an important role in controlling downstream signaling from diacylglycerol (66). Finally, LC8 (DYNLL1) has been shown to function as a versatile acceptor (i.e. motor receptor) to facilitate dynein-mediated nuclear accumulation of p53 after DNA damage (51).
In addition to the motor light chains themselves, other components of the motor machinery are important for cargo recognition. For example, dynactin has been shown to play a critical role in both cargo binding and regulation of dynein-mediated transport (67,68). Overexpression of one of the dynactin subunits, termed dynamitin, is known to disrupt dynein-dynactin functions, thereby diminishing dynein motor activities required for the intracellular transport of cargoes. Along these lines, overexpression of dynamitin has been used as an effective tool for examining the requirements of dynein-dependent cargo transport for intracellular signaling events (6,55,69,70). For example, disruption of dynein motor activity by overexpressing dynamitin impaired the accumulation of p53 in the nucleus following DNA damage (71). In addition, overexpression of p50/dynamitin impaired the nuclear accumulation of STAT5B after growth hormone induction (72), as well as the nuclear translocation of the glucocorticoid receptors after ligand stimulation (73). Because our results have shown that overexpression of dynamitin blocked Smad2 nuclear translocation after TGF␤ treatment, dynein-dependent intracellular events also appear to be required for TGF␤/Smad2 downstream effects.
As mentioned above, our results are consistent with km23-1 being present in early endosomes after T␤R-mediated endocytosis. However, our data suggest, further, that km23-1 may be one of the factors required for the intracellular movement of the endosomal T␤R/Smad2 signaling complexes toward the nucleus, based upon the known direction of movement of dynein motors (13). Along these lines, it is well established that T␤Rs are endocytosed through the clathrin-mediated pathway, which is important for promoting signaling (42)(43)(44)(45)(46). During clathrin-mediated endocytosis, the T␤R complex is targeted to clathrin-coated pits, where it binds to the ␤2-adaptin subunit of AP2 (45,74,75). Dynamin 2ab functions downstream of T␤RI activation, where it excises the budded vesicle from the plasma membrane (75). After clathrin-mediated endocytosis, T␤Rs are found for extended periods of time in EEA1-enriched early endosomes (45). The clathrin-mediated endocytic pathway is thought to promote the co-localization of T␤Rs with downstream signaling components (i.e. Smad2) in early endosomes. In addition, although T␤R phosphorylation and association with Smad2 can occur at the plasma membrane, RSmad phosphorylation and downstream signaling only appear to occur after clathrin-dependent endocytosis, requiring an unknown activity or activities downstream of dynamin 2ab function (45,76). Based upon our current results, km23-1 may participate in the recruitment of Smad2-containing TGF␤ signaling endosomes to the rest of the dynein motor for intracellular transport, prior to both nuclear translocation and downstream nuclear events. In this way, km23-1 may represent one of the additional steps, downstream from dynamin 2ab function, that is required for Smad signaling after T␤R activation (13).
Based upon our results and those of others, we propose a model for km23-1 action in the recruitment of TGF␤ signaling endosomes for intracellular transport along MTs. According to this model, within minutes of ligand binding, activated T␤Rs are internalized into EEA1/SARA-enriched endosomes, where Smad2 is recruited by SARA (42,45). Once km23-1 is phosphorylated by T␤RII (13) and Smad2 is phosphorylated by T␤RI (21,24,25), km23-1 selectively interacts with the T␤R/Smad2 complex, and recruits the TGF␤ signaling endosome to the dynein motor through the DIC-km23-1 interaction. In this regard, our previous results have shown that kinase-active T␤RIIs are absolutely required for the interaction of km23-1 with DIC (13). In addition, dynein is known to mediate the association of endosomal membranes with MTs (77,78). After attachment of the TGF␤ signaling endosomes to the rest of the motor, km23-1/dynein transports the TGF␤ signaling endosomes along MTs to the next endosomal compartment. The requirement of dynein motor function for intracellular movement of Smad2 was established by our results in Fig. 4 involving dynamitin overexpression. Upon reaching subsequent compartments, the signaling complex may active downstream components or be translocated to the nucleus for transcriptional regulation of target genes.
Although Smads 2 and 3 are highly homologous and share some overlapping activities, they have distinct functions and are regulated differentially (79,80). For example, previous work has indicated that Smad2 activates ARE-Lux (38), whereas Smad3 activates SBE2-Luc (39). In addition, Smad2 and Smad3 may be phosphorylated in different endocytic locales (81), and this distinct compartmentalization is in keeping with their divergent mechanisms of oligomerization (82), intracellular degradation (83), and regulation of TGF␤ cellular effects (84,85). Our studies have shown that blockade of km23-1 reduced TGF␤-and Smad2-dependent ARE-Lux transcriptional activity, but not TGF␤and Smad3-dependent SBE2-Luc activity, suggesting that km23-1's role in mediating Smad transcriptional activation is somewhat specific for this RSmad. Similarly, we have previously shown that km23-1 is regulated by TGF␤, but not by EGF (31). Further, others have found that receptors for another TGF␤ superfamily member (BMPRII) interact with another DLC (Tctex-1, DYNLT1), but not with km23-1. 3 Thus, the DLCs also show specificity with regard to the growth factors and receptors that activate them.
Ubiquitin proteasomal-mediated degradation is known to control the levels of Smads transcriptionally and post-translationally (26,83). Here we have shown that the proteasomal inhibitor lactacystin partially restored TGF␤-stimulated nuclear Smad2 expression, to levels more similar to those observed without km23-1 blockade. Therefore, blockade of km23-1 appears to stimulate a Smad2 ubiquitin proteasomalmediated degradation pathway, possibly due to the inability of km23-1 to recruit Smad2 to the rest of dynein motor. Smad ubiquitin regulatory factor (Smurf)-mediated ubiquitination pathways have been shown to play critical roles in the degradation of Smads. For example, Smurf1 has been shown to selectively interact with Smads1 and 5, targeting them for degradation (86). In addition, Smurf2 induces ubiquitin-mediated degradation of Smads 1 and 2 (87). Further, a recent report has shown that neural precursor cell-expressed, developmentally down-regulated 4-2 (NEDD4-2), a new member of the Smurflike E3 ligases, also induces Smad2 degradation via a ubiquitindependent degradation pathway. More importantly, it has been suggested previously that some Smad degradation can occur in the cytoplasm through Smurf-mediated ubiquitination pathways (86,88). Because our data demonstrate that blockade of km23-1 results in a depletion of Smad2 expression in the presence of TGF␤, through a ubiquitin proteasomal degradation pathway, it will be of interest to determine in future studies whether Smurfs are involved in this pathway.