Casein Kinase I (cid:1) Plays a Functional Role in the Transforming Growth Factor- (cid:2) Signaling Pathway*

The transforming growth factor- (cid:2) (TGF- (cid:2) ) signaling pathway is known to be involved in a wide range of biological events, including development, cellular differentiation, apoptosis, and oncogenesis. The TGF- (cid:2) signal is mediated by ligand binding to the type II receptor, leading to the recruitment and activation of the type I receptor, and subsequent activation of a family of intracellular signal transducing proteins called Smads. Here we report a regulatory role for casein kinase I (cid:1) (CKI (cid:1) ) in the TGF- (cid:2) signaling cascade. We find that CKI (cid:1) binds to all Smads and the cytoplasmic domains of the type I and type II receptors both in vitro and in vivo . The interaction of CKI (cid:1) with the type I and type II receptors is independent of TGF- (cid:2) stimulation, whereas the CKI (cid:1) / Smad interaction is transiently disrupted by ligand treatment. Additionally, CKI (cid:1) is able to phosphorylate the receptor-activated Smads (Smads 1–3 and 5) and the type II receptor in vitro . Transcriptional reporter assays reveal that transient overexpression of wild type CKI (cid:1) dramatically reduces basal reporter activity but enhances TGF- (cid:2) -stimulated transcription. Furthermore, overexpression of a kinase-dead mutant of CKI (cid:1) inhibits both basal and ligand-induced

The transforming growth factor-␤ (TGF-␤) signaling pathway is known to be involved in a wide range of biological events, including development, cellular differentiation, apoptosis, and oncogenesis. The TGF-␤ signal is mediated by ligand binding to the type II receptor, leading to the recruitment and activation of the type I receptor, and subsequent activation of a family of intracellular signal transducing proteins called Smads. Here we report a regulatory role for casein kinase I⑀ (CKI⑀) in the TGF-␤ signaling cascade. We find that CKI⑀ binds to all Smads and the cytoplasmic domains of the type I and type II receptors both in vitro and in vivo. The interaction of CKI⑀ with the type I and type II receptors is independent of TGF-␤ stimulation, whereas the CKI⑀/ Smad interaction is transiently disrupted by ligand treatment. Additionally, CKI⑀ is able to phosphorylate the receptor-activated Smads (Smads 1-3 and 5) and the type II receptor in vitro. Transcriptional reporter assays reveal that transient overexpression of wild type CKI⑀ dramatically reduces basal reporter activity but enhances TGF-␤-stimulated transcription. Furthermore, overexpression of a kinase-dead mutant of CKI⑀ inhibits both basal and ligand-induced transcription, whereas inhibition of endogenous CKI catalytic activity with IC261 blocks only TGF-␤-stimulated reporter activity. Finally, knocking down CKI⑀ protein levels results in a significant increase in basal and TGF-␤-induced transcription. These results suggest that CKI⑀ plays a liganddependent, differential, and dual regulatory role within the TGF-␤ signaling pathway.
The transforming growth factor-␤ (TGF-␤) 1 signaling cascade is critical for the processes of embryologic development and homeostasis of organisms as diverse as fruit flies and humans. This pathway begins its regulatory functions at the earliest stages of development and acts to coordinate the complex mechanisms of cellular differentiation that will ultimately result in a mature organism. The power that this signal transduction pathway wields over cellular fate is necessary for its ability to regulate the critical events of development and differentiation. However, when regulatory controls are lost, the result is usually uncontrolled growth and proliferation. Therefore, it is not surprising that mutations within the TGF-␤ pathway have been implicated in a wide range of clinically observed pathological abnormalities (1)(2)(3)(4).
The TGF-␤ superfamily of ligands includes bone morphogenetic proteins (BMPs), activins, and TGF-␤s. This signaling pathway is a relatively simple cascade that consists of the ligand, the type I and type II receptors, and the cytoplasmic signal transducers called Smads (5)(6)(7). The type I and type II receptors are serine/threonine kinases that, upon ligand binding, form a heterotetrameric complex in which the constitutively active type II receptor phosphorylates the GS domain of the type I receptor resulting in catalytic activation. Once activated, the type I receptor then transiently associates with and phosphorylates the receptor-activated Smads (R-Smads) at their two most carboxyl-terminal serine residues. Smads consist of two highly conserved Mad homology domains, termed MH1 and MH2, connected by a relatively divergent linker region. The MH1 domain is involved in DNA binding, whereas the MH2 domain is important for protein/protein interactions. Furthermore, in the absence of ligand the MH1 and MH2 domains can interact to form an inhibitory conformation that is relieved by type I receptor phosphorylation. The phosphorylation of Smads 2 and 3 by the activated TGF-␤ and activin type I receptors or Smads 1, 5, and 8 by the activated BMP type I receptors leads to R-Smad association with the co-Smad, Smad 4. This ligand-induced R-Smad-co-Smad heteromeric complex can then translocate to the nucleus and regulate gene transcription usually through additional associations with coactivators such as p300/CBP, corepressors such as c-Ski, or other transcription factors such as AP-1 (7).
The casein kinase I (CKI) family consists of seven known isoforms (␣, ␤, ␥1, ␥2, ␥3, ␦, and ⑀) that possess a highly homologous serine/threonine kinase domain and a divergent and variable length carboxyl-terminal tail (8,9). These kinases have been implicated in a wide range of cellular functions, including vesicular trafficking, DNA damage repair, cell cycle progression, and cytokinesis (8). Additionally, the enormous amount of research conducted on the various CKI isoforms has yielded invaluable information regarding kinase activity, substrate specificity, tissue distribution, subcellular localization, and catalytic regulation (8). Furthermore, the identification and characterization of a CKI consensus recognition sequence has been an area of particular interest (8, 10 -12). However, it is becoming increasingly apparent that one universal canonical recognition sequence probably does not exist, and thus identifying potential CKI phosphorylation sites within a given substrate remains a laborious endeavor.
The majority of the research conducted to date on the CKI family has centered primarily on characterizing its biochemical properties and cellular functions. However, recently there have been significant developments in the CKI field, and these once obscure kinases have become important players in the area of signal transduction. This idea is clearly supported by the numerous papers that have been published in the last several years that show a significant role for CKI⑀ and CKI␦ in the circadian rhythms of mammals (13)(14)(15)(16)(17)(18)(19), the regulation of a G q/11 -coupled receptor, the inhibition of the nuclear translocation of NF-AT4, the negative regulation of the Wnt and Hedgehog signaling pathways by CKI␣ (20 -28), the regulation of the ␤-platelet-derived growth factor receptor by CKI␥2 (29), and the regulation of the Wnt and JNK signaling pathways by CKI⑀ (30 -41).
Given the fact that many CKI family members have been found to play important regulatory functions in diverse signaling pathways, it is not surprising that we have identified CKI⑀ as a Smad3-interacting protein. Our functional characterization of this interaction has revealed that CKI⑀ is able to bind to Smads and TGF-␤ receptors in vitro and in vivo. We have also found that CKI⑀ transient overexpression represses basal activity but enhances ligand-stimulated activity of a TGF-␤-regulated transcriptional reporter. Furthermore, we have observed that ectopic expression of a kinase-deficient mutant of CKI⑀ causes a dominant negative phenotype that is illustrated by its inhibition of both basal and ligand-induced reporter activity. Additionally, inhibition of endogenous CKI kinase activity using a highly specific inhibitor causes no change in basal reporter activity but dramatically blocks TGF-␤-mediated transcription. Finally, our findings show that depletion of CKI⑀ by siRNA leads to increased Smad3 protein levels and consequently an increase in both basal and TGF-␤-stimulated transcriptional reporter activity. These data provide the first evidence demonstrating that CKI⑀ plays a critical regulatory role within the TGF-␤ signaling cascade.

MATERIALS AND METHODS
Plasmids-The GST fusion constructs of Smads 1-5 and Smad3N, Smad3NL, Smad3C, and Smad3CL have been described previously (42). The GST fusion construct of the cytoplasmic domain of the TGF-␤ type I has been described previously (43). The GST fusion construct of the TGF-␤ type II receptor was generated by digesting the full-length cDNA with HpaI and EcoRI and then subcloning the fragment containing the cytoplasmic domain into the SmaI/EcoRI sites of pGEX-3X. The mammalian expression vectors pRK5-Smad2-FLAG, pRK5-Smad-3-FLAG, and pRK5-Smad4-FLAG were provided by Dr. Rik Derynck. The pRc-CKI⑀-HA construct was a gift from Dr. David Virshup. pcDNA3-CKI⑀ was generated by PCR using pRc-CKI⑀-HA as template and the following primers: CKI⑀-F 5Ј-GGG AAG CTT AGC GGC CGC ATG GAG CTA CGT GTG GGG AAC AAG-3Ј, CKI⑀-R 5Ј-GCT CTA GAC TCA CTT CCC GAG ATG GTC AAA TGG C-3Ј. The CKI⑀ PCR product was digested with HindIII and XbaI, subcloned into pBKS, and sequenced to confirm the absence of mutations. CKI⑀ was then subcloned into the NotI and XbaI sites of pcDNA3 (Invitrogen). CKI⑀-GST was constructed by subcloning the CKI⑀ PCR product from pBKS into the NotI site of pGEX-4T-3 (Amersham Biosciences), oriented, and sequenced to confirm in-frame fusion with GST. The pGEX-4T-3-CKI⑀ construct was used to make carboxyl-terminal truncation mutants by introducing stop codons at various points within the tail region. The stop codons were introduced using the QuikChange TM site-directed muta-genesis protocol from Stratagene. The primer sequences used to generate these mutations are identical to those described previously (44). pcDNA3-CKI⑀ kinase-dead (KD) was generated by mutating lysine 38 to arginine using the QuikChange TM site-directed mutagenesis protocol from Stratagene. The following primer sequences were used for mutagenesis: CKI⑀K38R-F 5Ј-GGA AGT CGC CAT CAG GCT GGA GTG TGT G-3Ј and CKI⑀K38R-R 5Ј-CAC ACA CTC CAG CCT GAT GGC GAC TTC C-3Ј. Isolated clones were sequenced to confirm the correct mutation.
Cell Culture and Generation of Stable CKI⑀ Overexpressing Cell Lines-HepG2 cells were cultured in DMEM (Invitrogen) supplemented with 10% FBS, nonessential amino acids, and antibiotics and grown at 37°C in 5% CO 2 . HaCaT cells were cultured in ␣-minimum Eagle's medium, supplemented with 10% FBS, nonessential amino acids, and antibiotics and grown at 37°C in 5% CO 2 . HaCaT cells stably overexpressing HA-tagged CKI⑀ were generated using LipofectAMINE Reagent (Invitrogen) to transfect pRc-CKI⑀-HA according to the manufacturer's protocol. Following transfection, selection media containing 400 g/ml G418 was added, and individual clones were then isolated and analyzed by Western blot using anti-CKI⑀ (Transduction Laboratories). Positive clones were maintained in ␣-minimum Eagle's medium supplemented with 10% FBS, nonessential amino acids, antibiotics, and 200 g/ml G418 and grown at 37°C in 5% CO 2 .
GST Pulldown Assays-Smad-GST, receptor-GST, or CKI⑀-GST fusion proteins were bacterially expressed, and the bacteria were pelleted, resuspended in MTPBS (4 mM Na 2 HPO 4 , 16 mM NaH 2 PO 4 ⅐H 2 O, 150 mM NaCl, 1% Triton X-100, 1 mM PMSF, 1 mM DTT, 0.5 mM Na 3 VO 4 , protease inhibitors, pH 7.3), and lysed by sonication. The bacterial lysates were cleared by centrifugation and incubated for 2-4 h at 4°C with glutathione-Sepharose 4B (Amersham Biosciences). The Sepharose beads containing bound GST fusion proteins were washed 4ϫ with lysis buffer and analyzed by SDS-PAGE and Coomassie Brilliant Blue staining in order to determine relative yields. pcDNA3-CKI⑀, pRK5-Smad2, pRK5-Smad3, and pRK5-Smad4 were radiolabeled with [ 35 S]methionine using the Promega TNT® Coupled Reticulocyte Lysate System according to the manufacturer's protocol. An equal amount of purified GST fusion protein and a constant volume of reticulocyte lysate containing 35 S-labeled protein were diluted together in 400 l of ULB ϩ (50 mM Tris, 150 mM NaCl, 50 mM NaF, 0.5% Nonidet P-40, 1 mM PMSF, 1 mM DTT, 0.5 mM Na 3 VO 4 , protease inhibitors, pH 7.5) and incubated for 2-4 h at 4°C. The glutathione-Sepharose beads with bound GST fusions proteins were then washed 5ϫ with binding buffer, separated on SDS-PAGE gels, dried, and analyzed for the presence of 35 S-labeled proteins by autoradiography. Co-immunoprecipitation Assays-HaCaT cells or HaCaT cells stably overexpressing HA-CKI⑀ were grown to 50 -60% confluence on 10-cm plates under the culturing conditions described above and then treated with 100 pM TGF-␤ for variable lengths of time. TGF-␤-treated and -untreated cells were washed 2ϫ with phosphate-buffered saline, scraped, pelleted, and lysed with ULB ϩ (50 mM Tris, 150 mM NaCl, 50 mM NaF, 0.5% Nonidet P-40, 1 mM PMSF, 1 mM DTT, 0.5 mM Na 3 VO 4 , 10 mM ␤-glycerophosphate, 2 mM sodium molybdate, protease inhibitors, pH 7.5). Lysates were cleared by centrifugation, and protein concentrations were quantified using the modified Bradford protein assay kit from Bio-Rad. 400 -600 g of protein was incubated at 4°C for 2 h with either anti-TGF-␤ type I receptor (R-20, Santa Cruz Biotechnology), anti-TGF-␤ type II receptor (L-21, Santa Cruz Biotechnology), or anti-Smad2/3 (N-19, Santa Cruz Biotechnology), followed by incubation with a 50:50 mixture of protein A-Sepharose/protein G-Sepharose beads (Amersham Biosciences) for an additional 1 h at 4°C. The beads were pelleted and washed 4ϫ with ULB ϩ , and the precipitated proteins were eluted by boiling for 5 min in sample loading buffer, separated by SDS-PAGE, transferred to Immobilon-P polyvinylidene difluoride membrane (Millipore), and analyzed for the presence of CKI⑀ by immunoblotting with anti-CKI⑀ (Transduction Laboratories). Blots were then stripped and reprobed using anti-Smad1/2/3 (H-2 Santa Cruz Biotechnology) to analyze the efficiency of the Smad2/3 immunoprecipitation.
CKI⑀ in Vitro Kinase Assays-Smad-GST, receptor-GST, or CKI⑀-GST fusion proteins were bacterially expressed, and the bacteria were pelleted, resuspended in MTPBS (4 mM Na 2 HPO 4 , 16 mM NaH 2 PO 4 ⅐H 2 O, 150 mM NaCl, 1% Triton X-100, 1 mM PMSF, 1 mM DTT, 0.5 mM Na 3 VO 4 , protease inhibitors, pH 7.3), and lysed by sonication. The bacterial lysates were cleared by centrifugation and incubated for 2-4 h at 4°C with glutathione-Sepharose 4B (Amersham Biosciences). The Sepharose beads containing bound GST fusion proteins were washed four times with lysis buffer and then incubated for 6 -8 h in elution buffer (50 mM Tris, 25 mM glutathione, pH 8.0) at 4°C. The eluted proteins were quantified using a modified Bradford protein assay from Bio-Rad and analyzed by SDS-PAGE and Coomassie Brilliant Blue staining to visualize the quality of the purified protein. 2 g of purified and eluted substrate protein or 1 g of ␣-casein (Sigma) was combined with 250 ng of purified and eluted GST-CKI⑀-319 in 20 l of 2ϫ kinase buffer (60 mM HEPES, 14 mM MgCl 2 , 200 g/ml bovine serum albumin, 1 mM DTT, pH 7.5). The volume was adjusted to 38.5 l using sterile H 2 O, and the reaction was initiated by the addition of 1.5 l of ATP (1 l of 1 mM cold ATP and 0.5 l of [␥-32 P]ATP). The reactions were incubated at 37°C for 30 min and terminated by the addition of 4ϫ SDS-PAGE loading buffer and incubation at 95°C for 5 min. 50% of each kinase reaction was separated on an SDS-PAGE gel, dried, and visualized by autoradiography.
Transcriptional Reporter Assays-The 3TP-Lux reporter construct was provided by Dr. Joan Massagué, and the SBE-Lux reporter was constructed in our laboratory and has been described previously (45). 250 -300 ϫ 10 3 HepG2 cells/well were plated into 6-well plates and cultured as described above for 24 h or until an approximate confluence of 30 -40% was reached. By using the calcium phosphate co-precipitation method, 2 g of total DNA per well was transiently transfected (this included 0.250 g/well of the indicated reporter construct and 0.125 g/well of a CMV-␤-galactosidase construct) for 12-16 h. Cells were then washed two times with phosphate-buffered saline, allowed to recover for 6 -12 h in fresh media, and then incubated for an additional 18 -24 h with or without 100 pM TGF-␤. Following TGF-␤ treatment, cells were lysed, and supernatants were analyzed for luciferase activity using a luminometer. Luciferase activities were normalized to ␤-galactosidase activity to correct for variations in transfection efficiency. The luciferase assays in which IC261 (Santa Cruz Biotechnology) was used to inhibit the kinase activity of CKI⑀ were performed in HepG2 cells that had been transiently transfected with either SBE-Lux or 3TP-Lux reporters exactly as described above. IC261 was dissolved in Me 2 SO to create a stock concentration of 50 mM, and following transfection the cells were pretreated for at least 2 h with a range of IC261 concentrations (10, 20, and 40 M) prior to the addition of 100 pM TGF-␤.
CKI⑀ Transient Knockdown Experiments-HepG2 cells were plated into 10-cm plates, cultured as described above, and grown to 50 -60% confluence. Prior to transfection, cells were washed one time with serum-free DMEM, and then 3.6 ml of serum-free media was added to each plate. The sequence of the siRNA oligo used in this study was designed from the human CKI⑀ coding region and has been described previously (26). The siRNA oligo was purchased in the duplexed, desalted, 2Ј-deprotected, and purified form from Dharmacon. 1.44 nmol of the siRNA duplex was diluted in 1200 l of serum-free DMEM, whereas 72 l of OligofectAMINE (Invitrogen) was diluted separately in 288 l of serum-free DMEM and incubated at room temperature for 10 min. The diluted OligofectAMINE was added to the diluted siRNA and incubated for an additional 20 min at room temperature. The siRNA/ OligofectAMINE complex mixture was further diluted to 2.4 ml after 20 min using serum-free DMEM, and the entire volume was then added to the 10-cm plate. Transfection was terminated after 4 h by the addition of 6 ml of DMEM containing 20% serum and cultured overnight. The transfected HepG2 cells were then split, and the entire procedure was repeated twice. Following the third transfection, the HepG2 cells were seeded into 6-well plates to test for changes in TGF-␤-mediated transcriptional activity using the SBE-Lux and 3TP-Lux transcriptional reporter constructs exactly as described above. Following these luciferase assays, the remaining supernatants from each related condition were pooled, and protein concentrations were determined by using the modified Bradford protein assay kit from Bio-Rad. An equal amount of protein (40 g for CKI⑀, Smad3, and ␤-catenin) from each siRNA condition was separated by SDS-PAGE, transferred to Immobilon-P polyvinylidene difluoride membrane, and blotted with either anti-CKI⑀ (Transduction Laboratories), anti-Smad1/2/3 (Santa Cruz Biotechnology, Inc.), or anti-␤-catenin (Zymed Laboratories Inc.).

CKI⑀ Physically Interacts with Smads and TGF-␤ Receptors
in Vitro-The relatively simplistic mechanism by which TGF-␤ regulates a large and diverse set of cellular events raises the possibility that there may be additional proteins that play an important role in modulating the activity of this pathway. Therefore, we undertook the task of identifying pathway interactors and effectors by carrying out a yeast two-hybrid screen using Smad3 as bait as described previously (45,46). The results of this screen have led to the identification of many potential Smad3-interacting proteins, including the CKI␥2 iso-form. To confirm the yeast two-hybrid results, we first analyzed several CKI isoforms for their ability to interact directly with Smad3. In vitro binding assays were performed in which [ 35 S]methionine-labeled CKI␣, CKI␦, CKI⑀, and CKI␥2 were each independently incubated with purified GST-fused full-length Smad proteins. As shown in Fig. 1B, CKI⑀ is able to interact strongly not only with Smad3 but also with Smads1, -2, and -4, whereas interaction with Smad5 is extremely weak. The in vitro interactions of CKI␣ and CKI␦ with purified Smad proteins are relatively weak, whereas binding of full-length CKI␥2 to Smads is virtually undetectable (data not shown). Furthermore, by using Smad3 deletion mutants fused to GST and depicted in Fig. 1A, we found that CKI⑀ interacts most strongly with the MH2 domain. This is consistent with previous findings that have shown the MH2 domain of the R-Smads to be important in protein/protein interactions (7,47,48). The findings from these initial binding studies, combined with previous reports that CKI⑀ plays a functional role in other signaling pathways, led us to focus our investigation on this particular CKI isoform.
It has been reported previously (49) that immunoprecipitated T␤RII possesses an intrinsic casein kinase I activity. Close examination of the data from this study shows that when the T␤RII was immunoprecipitated from cells labeled with [ 35 S]methionine, a protein of approximately the same size as CKI⑀ was also precipitated. This observation led us to investigate the possibility that CKI⑀ might interact with T␤RI and/or the T␤RII. CKI⑀ was again radiolabeled with [ 35 S]methionine, and in vitro binding assays were performed with purified GST fusions of the cytoplasmic domains of the T␤RI and the T␤RII as well as GST fusions of deletion mutants of the T␤RII cytoplasmic domain. A schematic of the full-length receptors is depicted in Fig. 1C with the cytoplasmic domains bracketed. As shown in Fig. 1D, CKI⑀ is able to bind to the T␤RI and the T␤RII as well as the T␤RII deletion mutants with relatively equal affinities.
The tail of CKI⑀ constitutes approximately a third of the length of the protein and is relatively long compared with other CKI isoforms. Previous reports (44,50,51) have shown that this region is capable of undergoing intracellular autophosphorylation that results in autoinhibition of the kinase activity. Furthermore, it has been shown that removal of a significant portion of the tail region, mutation of potential autophosphorylation sites, or treatment with protein phosphatases can relieve this autoinhibitory effect (44,50,51). Additionally, there have been suggestions that the tail region of the CKI isoforms may help to direct substrate specificity, as well as participate in substrate binding (8). Therefore, to determine whether the carboxyl terminus of CKI⑀ is important for its interaction with Smads, we performed in vitro binding assays in which [ 35 S]methionine-labeled Smads 2-4 were independently incubated with purified GST fusions of CKI⑀ mutants that are missing portions of the tail region. Fig. 1E depicts the relative size and location of the domains of the CKI⑀ isoform and illustrates where the premature stop codons were introduced in order to generate the truncation mutants. As shown in Fig.  1F, even the loss of ϳ70% of the tail region of CKI⑀ (i.e. CKI⑀-319) does not significantly alter its ability to bind to Smads. The data provide additional evidence of in vitro binding between CKI⑀ and the Smad proteins, while also demonstrating that the tail region of CKI⑀ is not required for these interactions.
TGF-␤ Transiently Disrupts CKI⑀-Smad Binding but Does Not Alter CKI⑀-receptor Binding in Vivo-Because CKI⑀ can bind to Smads and the TGF-␤ receptors in vitro, we next tested whether this interaction also occurred in mammalian cells.
HaCaT cells, a spontaneously immortalized human keratinocyte cell line that is responsive to TGF-␤, were used to create a cell line that stably overexpresses CKI⑀ (data not shown). By using cell lysates from parental HaCaT cells and HaCaT cells overexpressing CKI⑀ (HaCaT-CKI⑀3) for co-immunoprecipitation (co-IP) assays, we found that CKI⑀ can interact with the T␤RI and the T␤RII, as well as Smad2 and Smad3 in vivo ( Fig.  2A). Most interesting, the strongest interaction occurred between CKI⑀ and the receptors, whereas a weaker interaction was observed with the R-Smads. In addition, co-IP assays were also performed using antibodies against Smad4 and Smad1/5, and consistent with the in vitro binding studies, CKI⑀ was also found to interact with these Smad proteins in vivo (data not shown). It should be noted that the parental HaCaT cells express CKI⑀ endogenously, and the interaction with Smads and the TGF-␤ receptors can be observed when co-IP experiments are performed using lysates from these cells (lanes labeled WT in Fig. 2A). However, the observed weak signal of in vivo interaction between endogenous Smads and endogenous CKI⑀ prompted us to use the HaCaT-CKI⑀3 cell line for future Smad2/3-CKI⑀ binding studies. In contrast, the strong in vivo binding of endogenous T␤RI and T␤RII with endogenous CKI⑀ allowed us to utilize the parental HaCaT cells for future receptor-CKI⑀ co-IP experiments.
The in vivo interaction we observed between components of the TGF-␤ pathway and CKI⑀ suggested that CKI⑀ may play a role within this signaling cascade. To investigate this possibility further, we next tested whether TGF-␤ treatment of Ha-CaT-CKI⑀3 cells or parental HaCaT cells could alter Smad-CKI⑀ or receptor/CKI⑀ interactions. As shown in Fig. 2B, the [ 35 S]Methionine-labeled CKI⑀ was incubated with purified, GST-fused receptor cytoplasmic domains, and binding was determined by autoradiography. E, schematic depiction of the structure of CKI⑀. The diagram illustrates the short amino-terminal region, the highly conserved kinase domain, and the more divergent carboxyl-terminal tail region. The arrows indicate the approximate location of the premature stop codons that were introduced to create the carboxyl-terminal truncations, and the number represents the codon that was mutated. F, the tail region of CKI⑀ is not required for binding to Smads in vitro. CKI⑀ full-length and CKI⑀ truncation mutants fused to GST were purified and incubated with [ 35 S]methionine-labeled Smad2, Smad3, or Smad4, and binding was determined by autoradiography.
interaction between CKI⑀ and Smad2/3 is transiently disrupted by TGF-␤ treatment, with a nearly complete disassociation by 2 h and re-association by 4 h. However, when a similar co-IP experiment was performed using an anti-T␤RII antibody, there was no detectable change in CKI⑀ binding within the same time frame (data not shown). To determine whether a ligand-induced change in CKI⑀-receptor binding might be a slower or delayed phenomenon, we again performed co-IP experiments using lysates from parental HaCaT cells that had been treated over a 12-h period with TGF-␤. As shown in Fig. 2C, there is no discernible change in T␤RI-CKI⑀ or T␤RII-CKI⑀ binding resulting from TGF-␤ treatment.
CKI⑀ Phosphorylates R-Smads and the T␤RII in Vitro-Because CKI⑀ is a serine/threonine kinase (the same as the type I and type II receptors), we tested if CKI⑀ could phosphorylate purified Smads and/or receptors in vitro. As discussed earlier, previous reports have shown that CKI⑀ is capable of autophosphorylation resulting in autoinhibition. We also observed a significant amount of autophosphorylation by fulllength CKI⑀ in our kinase assays; however, we did not have any difficulties with autoinhibition. The autophosphorylated band of full-length CKI⑀ runs at approximately the same molecular weight as the purified Smad and receptor proteins making it difficult to clearly observe the level of phosphorylation of each of these substrates. Therefore, to eliminate the CKI⑀ autophosphorylation band, we used the CKI⑀-319 truncation mutant that shows significant reduction in autophosphorylation (44) but retains the ability to bind Smad proteins in vitro (Fig. 1F). We found that CKI⑀ can phosphorylate the TGF-␤activated Smads (Smads 2 and 3) and the BMP-activated Smads (Smads 1 and 5) to a lesser extent but not the co-Smad (Smad4) (Fig. 3). In addition, by using the purified Smad3 deletion mutants depicted in Fig. 1A, we were able to map the CKI⑀ phosphorylation sites to the linker region and the MH1 domain (Fig. 3). Finally, we observed that CKI⑀ can also phosphorylate the cytoplasmic domain of the T␤RII but not that of the T␤RI (Fig. 3). These in vitro kinase assay results further suggest a role for CKI⑀ within the TGF-␤ signaling cascade.
CKI⑀ Plays a Ligand-dependent, Differential, and Dual Regulatory Role in TGF-␤ Signaling-The interaction studies described above prompted us to determine whether CKI⑀ is capable of playing a functional role in the TGF-␤ pathway. To address this question, we used the classic transcriptional reporter assay as a means of measuring changes in TGF-␤ signaling when CKI⑀ was transiently overexpressed. By using the

FIG. 2. TGF-␤ transiently disrupts CKI⑀-Smad binding but does not alter CKI⑀-receptor binding in vivo.
A, CKI⑀ physically interacts with the TGF-␤ type I and the TGF-␤ type II receptors as well as Smad2 and/or Smad3 in vivo. Wild type (WT) HaCaT cells and HA-CKI⑀ overexpressing (OE) HaCaT cells were lysed, and T␤RI, T␤RII, and Smad2/3 were immunoprecipitated (IP) followed by immunoblotting (Blot) using anti-CKI⑀. Ab, antibody. B, TGF-␤ treatment transiently disrupts the CKI⑀/Smad interaction. Ha-CaT cells stably overexpressing HAtagged CKI⑀ were treated with 100 pM TGF-␤ for 0, 0.5, 1, 2, and 4 h. The cells were then harvested and lysed, and Smad2/3 was immunoprecipitated followed by immunoblotting using anti-CKI⑀ (Lower Blot). The same membrane was stripped and reblotted using anti-Smad1/ 2/3 (Upper Blot). C, TGF-␤ treatment does not alter the binding of CKI⑀ with the type I or the type II receptor. Wild type HaCaT cells were treated with TGF-␤ for 0, 2, 4, 6, 8, and 12 h. The cells were then harvested and lysed, and the type I and the type II receptors were immunoprecipitated followed by immunoblotting for CKI⑀. SBE-Lux transcriptional reporter construct, a plasmid composed of four consecutive repeats of SBE fused to a luciferase gene, we found that transiently overexpressing CKI⑀ in HepG2 cells, a TGF-␤-responsive human hepatocellular carcinoma cell line, resulted in a significant reduction in basal transcriptional activity (Fig. 4A). In contrast, when TGF-␤ was added to HepG2 cells transiently overexpressing CKI⑀, the transcriptional reporter activity was actually enhanced (Fig. 4A). This ligand-dependent, differential, and dual response to CKI⑀ overexpression produced a dramatic increase in fold induction of TGF-␤-induced transcriptional reporter activity compared with the control cells lacking ectopic CKI⑀ expression (Fig. 4B). Furthermore, the transient overexpression of a CKI⑀ kinasedead construct (CKI⑀-K38R) produced a dominant negative effect leading to a reduction in both basal and TGF-␤-stimulated transcriptional activity of the SBE-Lux reporter (Fig. 4A). These results strongly suggest that CKI⑀ plays a differential and dual function that appears to not only depend on the presence or absence of ligand but also on the catalytic activity and the expression level of CKI⑀ itself. This critical point is conclusively supported by the data showing that the reduction in basal activity of the SBE-Lux construct depends only on increased expression of CKI⑀ regardless of the catalytic state, whereas the ligand induced activity of this reporter is enhanced only when catalytically active CKI⑀ is ectopically expressed (Fig. 4A).
To further support our hypothesis that CKI⑀ plays a regulatory role in the TGF-␤ signaling pathway, we employed a second TGF-␤-responsive transcriptional reporter construct. The 3TP-Lux reporter utilizes a portion of the plasminogen activator inhibitor-1 promoter that is known to contain both Smad-binding elements and AP-1-binding elements (52,53). In an attempt to replicate the CKI⑀ effect on the SBE-Lux reporter described above, we conducted the same experiment shown in Fig. 4A using the 3TP-Lux construct. However, in contrast to the SBE-Lux results, overexpression of CKI⑀ alone produced a slight reduction in both basal and TGF-␤stimulated 3TP-Lux activity (data not shown). Although these results do not support the data obtained with the SBE-Lux reporter, they are consistent with previous reports (38,40,41) demonstrating that CKI⑀ is a potent antagonist of JNK signaling and thus might act as an inhibitor of AP-1 activation. Therefore, any Smad-dependent enhancement of TGF-␤-stimulated 3TP-Lux transcriptional reporter activity that may result from CKI⑀ overexpression is most likely negated by its simultaneous inhibition of the JNK pathway. In light of these findings, the 3TP-Lux reporter proved useful for monitoring changes in CKI⑀ expression and/or catalytic activity.
Because the results of the in vitro kinase assays indicate that CKI⑀ can phosphorylate Smad3 (Fig. 3), we also tested whether concomitantly overexpressing CKI⑀ and Smad3 could effect TGF-␤-induced and Smad3-mediated transcription. HepG2 cells were transiently transfected with the plasmids indicated in Fig. 4C as described under "Materials and Methods." In this assay, the SBE-Lux reporter shows the usual stimulated response to TGF-␤ treatment, and the addition of Smad3 results in a significantly enhanced activity of the reporter. However, the simultaneous transient overexpression of Smad3 and CKI⑀ results in a near doubling of the reporter activity compared with that produced by Smad3 ectopic expression alone (Fig.  4C). Furthermore, overexpressing Smad3 in combination with CKI⑀-KD resulted in a complete inhibition of the ability of Smad3 to stimulate SBE-Lux reporter activity (Fig. 4C). These results from the CKI⑀ wild type and kinase-null overexpression experiments suggest that CKI⑀ enhances TGF-␤ signaling by phosphorylating Smad3 and improving its ability to drive transcriptional reporter activity.
Inhibition of CKI⑀ Kinase Activity Blocks TGF-␤-induced Transcription-To demonstrate further the importance of CKI⑀ catalytic activity in regulating the TGF-␤ signaling pathway, we used a specific inhibitor of casein kinase I, 3-[(2,4,6-trimethoxyphenyl)methylidenyl]-indolin-2-one (IC261), to block the kinase activity of endogenous CKI⑀ in HepG2 cells. This inhibitor has been shown previously to be ϳ10 times more specific for CKI⑀, and the highly homologous CKI␦ isoform, than for the other CKI isoforms (54). Transcriptional reporter assays were again used to monitor any changes in TGF-␤ signaling in the presence of this specific CKI⑀ inhibitor. As shown in Fig. 5A, the effect of IC261 treatment on the basal transcriptional activity of the SBE-Lux reporter was negligible. However, when HepG2 cells were transfected with the SBE-Lux construct and treated with the CKI⑀ inhibitor prior to the addition of TGF-␤, the transcriptional reporter activity was significantly inhibited (Fig. 5A). These results are consistent with the overexpression studies described in Fig. 4 showing that a block in catalytic activity (i.e. CKI⑀-KD) leads to an inhibition in TGF-␤-induced transcriptional reporter activity again suggesting that a CKI⑀mediated phosphorylation event is extremely important. Furthermore, a lack of change in basal reporter activity in the presence of IC261 is also consistent with our previous findings since a reduction in basal reporter activity appears to require only an increase in CKI⑀ protein expression levels regardless of its catalytic state. Therefore, as expected, in the assays where the inhibitor was used there was no change in SBE-Lux basal activity because there was no change in CKI⑀ expression (Fig.  5A). Finally, to provide additional evidence that CKI⑀ kinase activity was effectively inhibited by IC261, we repeated the transcriptional reporter assays using the 3TP-Lux construct. Given the detailed explanation above regarding the structure of the 3TP-Lux construct and the ability of CKI⑀ to inhibit JNK signaling, it is not surprising that HepG2 cells transfected with this reporter and then treated with IC261 show a dramatic increase in both basal and ligand-stimulated transcriptional activity (Fig. 5B).
Transient Knockdown of CKI⑀ Results in Deregulation of the TGF-␤ Signaling Pathway-To demonstrate conclusively that CKI⑀ plays an important role in the TGF-␤ pathway, we used a

FIG. 3. CKI⑀ phosphorylates R-Smads and the T␤RII in vitro.
Purified and eluted GST-fused Smads and receptors were used as substrates (2 g) for purified and eluted GST-fused CKI⑀-319 (0.250 g) as described under ''Materials and Methods.'' Kinase assays were performed, and the level of substrate phosphorylation was determined by separating the kinase reactions on an SDS-PAGE gel followed by autoradiography.
small interfering RNA oligo (siRNA) to transiently knockdown CKI⑀ expression in HepG2 cells. The sequence and effectiveness of this siRNA has been reported previously (26), and as shown in Fig. 6A, transfection of this duplexed siRNA oligo significantly reduces CKI⑀ protein levels in HepG2 cells. The effects of the reduction in CKI⑀ expression on TGF-␤ signaling were again monitored using the SBE-Lux and 3TP-Lux transcriptional reporters. In support of the luciferase assays showing that ectopic expression of CKI⑀ reduces the basal activity of the SBE-Lux reporter (Fig. 4A), the transient knockdown of CKI⑀ resulted in an increase in basal transcriptional activity of both SBE-Lux and 3TP-Lux (Fig. 6, A and B). Furthermore, when CKI⑀ protein levels were reduced, the TGF-␤ stimulated activity of the SBE-Lux and the 3TP-Lux reporters increased significantly over the control (Fig. 6, A and B). This last finding was surprising because CKI⑀ overexpression also produced an increase in SBE-Lux reporter activity in HepG2 cells that were treated with TGF-␤ (Fig. 4A). However, as shown in Fig. 6D, CKI⑀ depletion leads to an increase in the basal expression level of Smad3, which may explain why an increase in both basal and ligand-stimulated transcriptional reporter activity is observed under this condition. ␤-Catenin was used as a loading control because HepG2 cells have been shown to express a mutant form ␤-catenin (55) that is resistant to degradation and thus is not affected by changes in CKI⑀ expression (Fig. 6D). It has been well established that ectopically expressed Smad3 is capable of nuclear translocation and DNA binding in the absence of TGF-␤ stimulation; therefore, it is not surprising that an increase in Smad3 protein levels would lead to elevated basal reporter activity (42, 53, 56 -62). Furthermore, this observation can be extended to explain how TGF-␤-stimulated reporter activity increases when CKI⑀ is depleted and Smad3   FIG. 4. CKI⑀ plays a ligand-dependent, differential, and dual regulatory role in TGF-␤ signaling. A, CKI⑀ transient overexpression reduces basal reporter activity and enhances TGF-␤-stimulated reporter activity. HepG2 cells were transiently transfected with the TGF-␤ transcriptional reporter SBE-Lux (0.250 g), ␤-galactosidase as an internal control (0.125 g), and either wild type CKI⑀ (0.125 and 0.250 g) or KD CKI⑀ (0.125 and 0.250 g), treated with or without 100 pM TGF-␤ overnight, and then analyzed for luciferase activity. B, CKI⑀ significantly enhances the fold increase in transcriptional reporter activity in HepG2 cells. The fold increase in luciferase activity in response to TGF-␤ treatment was determined by dividing the average TGF-␤-stimulated reporter activity (black bars from A) by the average basal reporter activity (gray bars from A). C, CKI⑀ enhances TGF-␤-induced and Smad3-mediated transcription. HepG2 cells were transiently transfected with SBE-Lux (0.250 g), ␤-galactosidase (0.125 g), Smad3 (0.250 g), and either CKI⑀ (0.500 g) or CKI⑀-KD (0.500 g), treated with or without 100 pM TGF-␤ overnight, and then analyzed for luciferase activity. levels increase, because it can be safely assumed that an increase in basal Smad3 protein would also lead to an increase in ligand-activated Smad3. Additionally, as shown in Fig. 2B, CKI⑀ is able to re-associate with Smad3 after ϳ4 h of TGF-␤ treatment; however, if CKI⑀ protein levels are reduced this re-association would be impaired, thus allowing receptor-activated Smad3 to continue to enter the nucleus unimpeded and sustain transcription for a longer period. Finally, Fig. 6E illustrates how the changes in CKI⑀ protein levels and catalytic state affect TGF-␤-mediated transcriptional reporter activity, while also providing a convenient summary of the results of the related transcriptional reporter experiments described in Figs. 4 -6.

DISCUSSION
The complex and diverse signals conveyed by the relatively simple TGF-␤ pathway prompted us to investigate the possibility that other proteins may participate in regulating, maintaining, or propagating the basal and ligand-stimulated activity of this signaling cascade. The data presented here provide the first evidence that casein kinase I⑀ plays an important ligand-dependent, differential, and dual regulatory role in the TGF-␤ signaling pathway.
We have demonstrated that CKI⑀ can physically interact with Smads in vitro and in vivo and that the interaction between CKI⑀ and Smad3 is transiently disrupted by TGF-␤ treatment. The observed association between Smad3 and CKI⑀ is yet another example of Smad3 physically interacting with a known component of the canonical Wnt signaling pathway. Furuhashi et al. (63) have shown that Smad3 and axin, a scaffolding protein critical for ␤-catenin degradation, can interact in vivo and that this association is disrupted by TGF-␤ stimulation. Furthermore, we have found that Smad3 and GSK-3␤ also interact in vivo and that this association is disrupted by TGF-␤ treatment. 2 In addition, a recent report (64) identified Dishevelled (Dvl), a positive regulator of Wnt signaling, in a yeast two-hybrid screen as a Smad3-interacting protein. It has also been reported that receptor-activated Smad3 is ubiquitinated and degraded by the proteasome machinery fol- FIG. 5. Inhibition of CKI⑀ kinase activity blocks TGF-␤-induced transcription. A, IC261 inhibition of CKI⑀ has no significant effect on SBE-Lux basal activity but does dramatically inhibit TGF-␤-stimulated transcriptional activity. HepG2 cells were transfected with the SBE-Lux reporter (0.250 g), ␤-galactosidase as an internal control (0.125 g), and pGCN (1.625 g) as filler DNA. Following transfection, the cells were pretreated with IC261 (10, 20, and 40 M) for at least 2 h, followed by an overnight treatment with or without 100 pM TGF-␤ and then analyzed for luciferase activity. B, IC261 inhibition of CKI⑀ results in a significant increase in both basal and TGF-␤-stimulated 3TP-Lux transcriptional reporter activity. HepG2 cells were transfected with the 3TP-Lux reporter (0.250 g), ␤-galactosidase as an internal control (0.125 g), and pGCN (1.625 g) as filler DNA. Following transfection, the cells were pretreated with IC261 (10, 20, and 40 M) for at least 2 h, followed by an overnight treatment with or without 100 pM TGF-␤ and then analyzed for luciferase activity.
FIG. 6. Transient knockdown of CKI⑀ results in deregulation of the TGF-␤ signaling pathway. A, HepG2 cells transfected with a CKI⑀-specific siRNA oligo leads to reduced CKI⑀ protein expression. Western blot analysis using anti-CKI⑀ was performed on lysates from HepG2 cells that had been transfected with an siRNA oligo specific for human CKI⑀ as described under ''Materials and Methods.'' B, reduced CKI⑀ expression results in increased basal and ligand-stimulated SBE-Lux transcriptional activity. HepG2 cells were serially transfected for three consecutive passages with an siRNA oligo directed against human CKI⑀. The cells were then transfected with the SBE-Lux reporter (0.250 g), ␤-galactosidase as an internal control (0.125 g), and pGCN (1.625 g) as filler DNA, treated with or without 100 pM TGF-␤, and then analyzed for luciferase activity. C, reduced CKI⑀ expression results in increased basal and ligand-stimulated 3TP-Lux transcriptional activity. HepG2 cells were treated and analyzed exactly as described in B, except that 0.250 g of 3TP-Lux was used as the transcriptional reporter. D, depletion of CKI⑀ leads to increased Smad3 expression in HepG2 cells. Western blot analysis using anti-Smad1/2/3 was performed on lysates from HepG2 cells transfected with an siRNA oligo specific for human CKI⑀ as described under ''Materials and Methods.'' The membrane was then stripped and re-probed using anti-␤-catenin in order to confirm equal protein loading. E, summary of the effects on TGF-␤ signaling resulting from changes in CKI⑀ protein expression and catalytic activity. lowing association with ROC1-SCF Fwb1a , a ubiquitin-protein isopeptide ligase complex, the same complex that promotes the ubiquitination and degradation of ␤-catenin (65). Finally, while this manuscript was in preparation, Inoue et al. (66) published a report confirming our unpublished findings that steady state Smad3 protein levels are tightly regulated and that in addition to Smad3 ubiquitination and degradation following TGF-␤ treatment, Smad3 also undergoes ubiquitination and degradation in the absence of ligand stimulation. These previously published findings, in combination with the data presented here, strongly support a model in which Smad3 associates with the axin complex and may explain how Smad3 is regulated in the absence of ligand stimulation. Although it remains to be determined whether Smad3 and ␤-catenin bind to axin at the same time or whether their interactions with axin are mutually exclusive, mounting evidence supports the notion that Smad3 is intimately associated with the axin degradation complex. Furthermore, this body of data demonstrates that TGF-␤ treatment destabilizes this multiprotein complex, suggesting that the stability and activity of Smad3 is probably regulated in a similar fashion as that observed for ␤-catenin. Most interesting, this hypothesis also explains why we consistently observe a dramatic increase in Smad3 protein levels in TGF-␤-responsive cells following ligand treatment (Fig. 2B, upper panel). It is clear that further investigation into the exact role of CKI⑀ in the regulation of Smad3 steady state protein levels will be necessary, and our laboratory is currently engaged in research to elucidate how CKI⑀, and possibly other CKI isoforms, may participate in Smad3 turnover.
In addition to the binding studies discussed above, we found that depletion of CKI⑀ protein using small inhibitory RNA oligos resulted in deregulation of the TGF-␤ pathway, which was demonstrated by a dramatic increase in both basal and ligand-stimulated transcriptional reporter activity (Fig. 6, B and C). This increase in reporter activity is most likely the result of an increase in the basal expression level of Smad3, which further supports the conclusion that CKI⑀ is critical for properly regulating Smad3 protein levels in the absence of ligand (Fig. 6D). The importance of properly maintaining Smad3 expression levels within a cell becomes clear when it is considered that Smad3 possesses a functional nuclear localization sequence and is capable of nuclear translocation even in the absence of ligand stimulation (58 -62). Additionally, unlike Smad2 which requires Smad4 for DNA binding, Smad3 is able to bind to DNA independently and drive transcription (53,56). Furthermore, we have found that transfecting as little as 10 ng of a Smad3 expression vector in HepG2 cells is sufficient to dramatically increase transcriptional reporter activity even in the absence of TGF-␤. 3 These findings and observations demonstrate the necessity of regulating Smad3 protein levels within the cell in the absence of ligand stimulation, but it remains to be determined how CKI⑀ might regulate Smad3 expression levels, localization, and function. Through careful analysis of the data presented in this study, we conclude that the negative basal control imposed on TGF-␤ signaling by CKI⑀ is probably achieved passively by helping to recruit and sequester Smad3 in the axin complex ultimately resulting in Smad3 turnover. Our model of passive negative regulation of basal TGF-␤ signaling by CKI⑀ is supported by the transcriptional reporter data that demonstrate that ectopic expression of either wild type CKI⑀ or kinase-null CKI⑀ is able to reduce basal reporter activity (Fig. 4A). Furthermore, when HepG2 cells were treated with IC261, an inhibitor of CKI kinase activity, there was no effect on basal signaling, again suggesting that the reduction in basal reporter activity simply requires the presence of CKI⑀ protein regardless of its catalytic state (Fig. 5A).
CKI⑀ clearly plays a role as a negative regulator of basal activity; however, our findings also suggest that CKI⑀ is a critical modulator of TGF-␤-induced, Smad3-mediated transcription. Although the mechanism by which a dual regulatory function might be achieved by the same protein within the same signaling pathway remains to be explained, previous research conducted on other signaling pathways provides a point of reference for the formulation of a working model. There are extensive data to suggest that CKI⑀ is a positive regulator of the Wnt signaling pathway; however, the means by which this positive effect is achieved requires further clarification. Several studies have independently shown that CKI⑀ exerts its positive influence on Wnt signaling by binding to and phosphorylating Dvl, leading to its ability to antagonize GSK-3␤ activity and thus stabilizing ␤-catenin (30,31,33,35,36,39). In addition to the evidence showing CKI⑀ binds and phosphorylates Dvl, there is also a report that CKI⑀ phosphorylates components of the axin degradation complex, leading to its destabilization and ultimately an increase in ␤-catenin protein levels (35). Furthermore, this same group has shown recently (67) that Wnt stimulation alleviates CKI⑀ autoinhibition resulting in increased catalytic activity. Considering these reports, it is conceivable that TGF-␤ stimulation might also lead to CKI⑀ activation, resulting in destabilization of the axin complex and ultimately the release of receptor activated Smad3. This hypothesis is supported by our transcriptional reporter data showing that ectopic expression of CKI⑀ enhances reporter activity only in the presence of ligand, whereas overexpression of the kinase-null CKI⑀ blocks TGF-␤-induced reporter activity (Fig. 4A). Moreover, the treatment of HepG2 cells with IC261 potently inhibits TGF-␤-mediated transcription suggesting that CKI⑀ catalytic activity is critical for the propagation of a ligand-induced signal from the membrane to the nucleus (Fig. 5A).
Finally, in addition to the reports demonstrating a role for CKI⑀ in the Wnt pathway, there are also several reports showing that CKI⑀ plays an important role in modulating the circadian rhythms of mammals by phosphorylating mPer1 and mPer3 leading to their degradation by the ubiquitin-proteasome machinery (14,17). Most interesting, it has also been reported (17) that CKI⑀ phosphorylation of mPer3 promotes its nuclear translocation and thus enhances the ability of mPer3 to antagonize the transcriptional activity of BMAL1-CLOCK. Therefore, considering our observation that CKI⑀ can phosphorylate Smad3 in vitro (Fig. 3), it is possible that CKI⑀ may also phosphorylate Smad3 in vivo in a ligand-dependent manner and thus enhance the ability of Smad3 to translocate to the nucleus and initiate transcription of TGF-␤ target genes. This hypothesis is further supported by the transcriptional reporter data showing that simultaneous ectopic expression of CKI⑀ and Smad3 leads to significant enhancement of Smad3-mediated reporter activity compared with overexpression of Smad3 alone (Fig. 4C). Although it appears that CKI⑀ phosphorylation of Smad3 improves the ability of Smad3 to drive ligand-induced transcription, a more detailed investigation will be needed in order to determine the significance and necessity of this phosphorylation event in TGF-␤ signaling. Consequently, our laboratory is currently involved in research aimed at identifying and characterizing the CKI⑀ phosphorylation sites on Smad3.
In conclusion, the results presented in this study, when considered in combination with previous reports illustrating the role of CKI⑀ in other signaling pathways, have begun to reveal a complex scenario in which the CKI⑀ isoform negatively reg-ulates the basal state of several major signal transduction pathways, while also acting to enhance and fine-tune these same pathways when they become ligand-activated. Therefore, we propose a model in which CKI⑀ plays the dual role of an inhibitor, in the absence of ligand, and an enhancer, in the presence of ligand, within the TGF-␤ signaling pathway.