The bi-directional regulation of TGF-β1 (transforming growth factor-β1) on fibroblast proliferation with stimulation at low concentration, but inhibition at high concentration, has important significance during tissue repair. The mechanism has not been defined. c-Ski is a major co-repressor of TGF-β1/Smad3 signalling; however, the exact role of c-Ski in the bi-directional regulation of fibroblast proliferation remains to be determined. In the present study, we established a dose–effect relationship of bi-directional regulation of TGF-β1-mediated proliferation in rat skin fibroblasts, and found that c-Ski overexpression promoted fibroblast proliferation by inhibiting Smad3 activity. Importantly, c-Ski expression was decreased at the high concentration of TGF-β1, but increased at the low concentration of TGF-β1. This dose-dependent change in TGF-β1 action did not affect Smad3 phosphorylation or nuclear translocation, but altered Smad3 DNA-binding activity, transcriptional activity and expression of the downstream gene p21 that both increased at the high concentration and decreased at the low concentration. Furthermore, c-Ski overexpression exerted synergistic stimulation with TGF-β1 at the low concentration, but reversed the inhibitory effect of TGF-β1 at high concentrations, while knockdown of c-Ski by RNA interference abrogated bi-directional role of TGF-β1 on fibroblast proliferation. Thus our data reveal a new mechanism for this bi-directional regulation, i.e. c-Ski expression change induced by low or high TGF-β1 concentration in turn determines the promoting or inhibiting effects of TGF-β1 on fibroblast proliferation, and suggests an important role of c-Ski that modulates the local availability of TGF-β1 within the wound repair microenvironment.
Fibroblasts are the most common cells in connective tissue and are derived from mesenchymal cells during embryonic development. The important physiological functions of fibroblasts include the deposition of ECM (extracellular matrix) and regulation of epithelial differentiation . In addition, fibroblasts have a prominent role during wound repair. They proliferate and invade lesions, modulate local inflammation, generate ECM to serve as a scaffold for other cells and possess cytoskeletal elements that facilitate contractions of healing wounds . Disorder of fibroblast proliferation would result in delayed wound healing or tissue fibrosis, even fibroma [3,4].
As a growth factor, TGF-β1 (transforming growth factor β1) is released by platelets, macrophages and fibroblasts within wound tissues, and exerts many important functions, including modulation of fibroblast proliferation, enhancement of fibroblast migration, maturation and ECM synthesis . However, some of these TGF-β1 functions are apparently paradoxical. For example, in mesenchymal cells (such as smooth muscle cells, osteoblasts, chondrocytes and, especially, fibroblasts), TGF-β1 exhibits a bi-directional effect, e.g. TGF-β1 can stimulate or inhibit cell proliferation by low and high concentrations of TGF-β1 respectively [6–9]. In fact, this bi-directional regulation of fibroblast proliferation by TGF-β1 is critical to tissue damage and remodelling processes. During wound healing, TGF-β1 showed none or low expression at an early stage, and kept increasing until 100–1000-fold elevation at later stages [10–12]. Thus there was increased cell proliferation and chemotactic effects at the periphery of the lesioned tissues where the TGF-β1 concentration was low or non-existent. As the cells reach the core of the lesioned tissues while higher TGF-β1 concentrations were detected, cell proliferation and migration become inhibited. Consequently, at the core of lesioned tissues, fibroblasts devote their energy to perform differentiated functions of ECM deposition [13,14].
However, why a growth factor such as TGF-β1 produces an apparent opposing effect in the same cells is not clear. The proliferative response induced by TGF-β1 in mesenchymal cells has been reported to be mediated indirectly through autocrine secretion of PDGF (platelet-derived growth factor). Specifically, at low concentrations, TGF-β1 induces cell proliferation by stimulating autocrine PDGF-AA secretion or expression of PDGF receptor α subunit [15–17]. At high concentrations, TGF-β1 limits the extent of this proliferative response by reducing the expression PDGF receptor α subunit [7,18]. However, in these studies, when neutralizing antibodies raised against PDGF were tested, they produced only modest effects on the bi-directional action of TGF-β1 on the autocrine proliferative effect. Moreover, in human adult skin fibroblasts, anti-PDGF neutralizing antibody was unable to block the mitogenic effect of TGF-β1 . Finally, this is unlikely to be a general mechanism, since many cell types, such as epithelial cells, do not normally express PDGF receptors and are unresponsive to PDGF . The exact mechanism underlying the bi-directional modulation of TGF-β1 in fibroblast proliferation remains to be defined.
c-Ski, a homologue of v-Ski in cells , is reported to be a complex regulator of fibroblast proliferation: c-Ski can promote proliferation of chicken embryo fibroblasts, but inhibit that of mouse embryo fibroblasts [21,22]. Our previous studies found that c-Ski was a new tissue-repair-related gene, beginning its expression in fibroblasts after wounding, and showed a similar expression tendency to fibroblast proliferation . c-Ski has been reported to be a major co-repressor of TGF-β1/Smad3 signalling (Smad3 is a direct downstream transcriptional factor of TGF-β1, and regulates gene transcription through forming a complex with co-activator or co-repressor) [24,25]. It is thus suggested that c-Ski may play an important role in the bi-directional regulation of fibroblast proliferation by TGF-β1.
In the present study, we demonstrated that c-Ski not only promoted skin fibroblast proliferation, but also interacted with the Smad3-mediated transcriptional mechanism to produce the bi-directional effect of TGF-β1 on fibroblast proliferation. Moreover, the expression of c-Ski was under regulation by the low or high concentration of TGF-β1, in turn, determining the promoting or inhibiting effects of TGF-β1 in rat fibroblasts. Our data thus revealed an essential role for c-Ski in mediating directional regulation of fibroblasts by low and high doses of TGF-β1 through a feedback regulation of expression of c-Ski. This provides mechanistic insight into the apparently paradoxical bi-directional effect of TGF-β1 on fibroblast proliferation. Thus the control of c-Ski expression in physiological and pathological conditions may represent an effective therapeutic strategy for wound healing and treatment of tumour development.
TGF-β1 was purchased from PeproTech of America. BrdU (bromodeoxyuridine) ELISA kit was from Roch Company. Antibodies specific to Smad3 (sc-6032), p21 (sc-397), c-Ski (sc-33693), β-actin (sc-1616) and their respective horseradish-peroxidase-coupled secondary antibodies were purchased from Santa Cruz Biotechnology. Rabbit polyclonal anti-phospho-Smad3 (Ser433/Ser435) was purchased from Cell Signaling Technology. pEGFP-N1 fluorescence plasmid was obtained from Clontech. FLAG-tagged Smad3 and c-Ski expression plasmid was provided by Dr Kohei Miyazono (Department of Biochemistry, Cancer Institute of the Japanese Foundation for Cancer, Tokyo, Japan). The SBE (Smad-binding element)–Luc (luciferase) reporter plasmid was provided by Dr Carl-Henrik Heldin (Ludwig Institute for Cancer Research, Uppsala, Sweden).
Primary fibroblasts culture
Primary fibroblasts were obtained from back skin of 4–6-week-old Wistar rats. Fibroblast culture was established as described previously . Fibroblasts were incubated at 5% CO2 in a humidified 37 °C incubator, cultured with DMEM (Dulbecco's minimum essential medium) (Gibco) supplemented with 10% heat-inactivated FBS (fetal bovine serum) (Gibco) three times a week. When cells reached confluence, they were passaged by trypsinization (trypsin was obtained from Gibco), resuspended and re-plated. Fibroblasts at passage 3–8 were used.
Plasmid construction for c-Ski interference
To design target-specific siRNAs (small interfering RNAs) against c-Ski, sequences from the coding region of rat (which was cloned by our laboratory and deposited in GenBank® under accession number DQ409171) were selected using the siRNA Target Finder tool from Ambion (http://www.ambion.com/techlib/misc/siRNA_finder.html). The specificity of the siRNAs was verified by a search of the sequences deposited at the GenBank® with the basic local alignment search tool (BLAST). The selected oligonucleotides were cloned into a BglII/HindIII-cut pSuper vector (Oligoengine, Seattle, WA, U.S.A.). Sequences of oligonucleotides for c-Ski were sense, 5′-GATCCCCGAAGGAGTTGGCGGCCAGCTTCAAGAGAGCTGGCCGCCAACTCCTTCTTTTTA-3′, and antisense, 5′-AGCTTAAAAAGAAGGAGTTGGCGGCCAGCTCTCTTGAAGCTGGCCGCCAACTCCTTCGGG-3′; and for control RNAi (RNA interference) were sense, 5′-GATCCCCGACTTCATAAGGCGCATGCTTCAAGAGAGCATGCGCCTTATGAAGTCTTTTTA-3′, and antisense, 5′-AGCTTAAAAAGACTTCATAAGGCGCATGCTCTCTTGAAGCATGCGCCTTATGAAGTCGGG-3′.
Proliferation assay and transient transfection
Cell proliferation was determined by the measurement of BrdU incorporation into newly synthesized cellular DNA using a cell proliferation ELISA kit (Roche). The assay was performed as described in the manufacturer's instructions. Briefly, for TGF-β1 dose–effect experiments, fibroblasts were seeded in DMEM/0.4% FBS. After 24 h of pre-incubation, the medium from each well was replaced with fresh DMEM/0.4% FBS and TGF-β1 at six different concentrations (2.5 pg/ml, 25 pg/ml, 250 pg/ml, 2.5 ng/ml, 25 ng/ml or 250 ng/ml). Fibroblasts exposed to DMEM/0.4% FBS were used as controls. Cells were incubated for 48 h. At 4 h before analysis, 10 μl of BrdU labelling solution per well was added to cells. Cells were then fixed with 100 μl of fix solution per well for 30 min at room temperature (20–25 °C) and incubated with anti-BrdU antibody conjugated with horseradish peroxidase for 90 min. A substrate solution was then added into each well, and absorbance of samples was measured at 470 nm wavelength using an ELISA plate reader.
For transfection experiments, cells were first seeded into 96-well plates at a density of 104 cells/100 μl per well and incubated for 24 h before transfection. Cells were then transfected with 0–200 ng of c-Ski plasmid, Smad3 plasmid, or a combination of c-Ski and Smad3 plasmids with Lipofectamine™ transfection reagent (Invitrogen), and pcDNA3.0 transfection was used as control. Co-transfection of pEGFP-N1 was used to control transfection efficiency. At 48 h after transfection, BrdU incorporation was assayed as mentioned above.
To observe the role of c-Ski overexpression or knockdown in TGF-β1 bi-directional regulation of fibroblast proliferation, fibroblasts were transfected with c-Ski plasmids or pSuper-c-Ski-RNAi. At 6 h after transfection, cells were synchronized with DMEM/0.4% FBS for 24 h, then DMEM/0.4% FBS with TGF-β1 at 25 pg/ml or 25 ng/ml was added, then cell proliferation was assayed after 24 h. pcDNA3.0 empty vector or pSuper-control-RNAi transfection was used as control.
At 24 h after synchronization with DMEM/0.4% FBS, fibroblast were left untreated or treated with 25 pg/ml or 25 ng/ml TGF-β1 for the times indicated. Whole-cell lysates were collected with M-PER™ protein extraction reagent (Pierce) containing protease inhibitor mixture (Calbiochem) to detect Smad3 phosphorylation, c-Ski and p21 expression. Equal amounts of protein were separated by SDS/PAGE (10% gels) and transferred on to an Immobilon-P PVDF membrane (Millipore). The expression of actin or non-phosphorylated Smad3 in each sample was also detected to show equal protein loading.
Real-time PCR assays
Fibroblasts were left untreated or treated with 25 pg/ml or 25 ng/ml TGF-β1 for 1 h. Total RNA was isolated from fibroblasts using TRIzol® (Invitrogen) according to the manufacturer's instructions. Reverse transcription was performed with AMV (avian myeloblastosis virus) reverse transcriptase (TaKaRa) and oligodT18 primers (TaKaRa). Real-time PCR was performed using SYBR Green I (Molecular Probes) and the iQCycler thermocycler (Bio-Rad). The primers for Smad3 were sense, 5′-TCCTGGCTACCTGAGTGAAGA-3′, and antisense 5′-GTTGGGAGACTGGACGAAAA-3′; for c-Ski were sense, 5′-TCAACTCGGTGTGCGATG-3′, and antisense, 5′-CGTCCGTCTTGGTGATGAG -3′; and for GAPDH (glyceraldehyde-3-phosphate dehydrogenase) were sense, 5′-AGGTTGTCTCCTGCGACTTCA-3′, and antisense, 5′-TGGTCCAGGGTTTCTTACTCC-3′. Real-time PCR was carried out as follows: initial denaturation for 5 min at 94 °C and 40 cycles of PCR consisting of 15 s at 94 °C, 15 s at 59 °C and 20 s at 72 °C. Quantification was always normalized using endogenous control GAPDH .
EMSAs (electrophoretic mobility shift assays)
Protein binding activity to the Smad3 DNA element was measured by EMSAs as described previously . Briefly, the cells were transfected with control RNAi or c-Ski RNAi. After 6 h, cells were synchronized with DMEM/0.4% FBS for 24 h, then treated with control medium (DMEM/0.4% FBS) or DMEM/0.4% FBS containing TGF-β1 at 25 pg/ml or 25 ng/ml for 1 h. Nuclear protein was extracted with NE-PER™ nuclear protein extraction kit (Pierce) and stored at −70 °C until analysis. The sequences of the SBE were sense, 5′-CTAGGATAGCGTCTAGACATAGTCTAGACTGAGT-3′, and antisense, 5′-CTAGGACTCAGTCTAGACTATGTCTAGACGCTAT-3′. Synthetic complementary oligonucleotides were 3′-biotinylated using the biotin 3′-end DNA labelling kit (Pierce). Binding reactions were carried out for 20 min at room temperature in the presence of 50 ng/μl poly(dI-dC)·(dI-dC), 0.05% Nonidet P40, 5 mM MgCl2, 10 mM EDTA and 2.5% glycerol in 1× binding buffer (LightShift™ chemiluminescence EMSA kit; Pierce) using 20 fmol of biotin-end-labelled target DNA and 4 μg of nuclear extract. The DNA-binding reaction without nuclear protein was used as protein control. For competition EMSA, a 100-fold molar excess of non-labelled oligonucleotide probe was added to the binding reaction. For supershift experiments, nuclear extracts were incubated overnight with anti-Smad3 antibody before the binding reaction. The samples were fractionated through a 4% polyacrylamide gel and transferred on to a nylon membrane (Hybond™-N+). Transferred DNAs were cross-linked to the membrane and detected using horseradish-peroxidase-conjugated streptavidin (LightShift™ chemiluminescence EMSA kit) according to the manufacturer's instructions.
Assay of Luc reporter gene expression
Fibroblasts were transfected with pcDNA3.0, Smad3 or a combination of Smad3 and c-Ski expression plasmids, together with SBE–Luc reporter and pRL-TK (Promega) plasmids by the Lipofectamine™ method. At 24 h after transfection, cells were starved in serum-free medium for 8 h and stimulated with control medium, 25 pg/ml or 25 ng/ml TGF-β1 for 24 h . Luciferase activity was detected using a dual-luciferase reporter assay system (Promega). Relative luciferase activity was normalized with Renilla luciferase activity.
Bi-directional effects of TGF-β1 on rat skin fibroblast proliferation
Watelet et al.  reported that TGF-β1 was undetectable or showed low expression (approx. 55 pg/ml) at the beginning of nasal surgery, and reached approx. 15 ng/ml 1 week after surgery . However, others have reported that TGF-β1 concentration could reach dozens or hundreds of ng/ml at a later stage of wound healing using a different model of wound healing [10,12]. To determine the effect of TGF-β1 on skin fibroblasts, we isolated primary rat skin fibroblasts and determined the effects of increasing concentrations of TGF-β1 from 2.5 pg/ml to 250 ng/ml on its proliferation using a BrdU ELISA method. Consistent with previous reports in human fetal lung and Tenon's fibroblasts [9,30], TGF-β1 stimulated rat skin fibroblast proliferation at low concentrations (from 2.5 pg/ml to 2.5 ng/ml), with a maximum at 25 pg/ml. However, at the high concentrations of 25 ng/ml and 250 ng/ml, TGF-β1 inhibited rat skin fibroblast proliferation (Figure 1).
Effects of TGF-β1 on the proliferation of rat skin-derived fibroblasts
c-Ski promotes and Smad3 suppresses rat skin fibroblast proliferation
Smad3 can mediate the inhibitory effects of TGF-β1 on various cell types . c-Ski by itself is active, but acts as a co-repressor of Smad3 to exert a complex regulation of fibroblast proliferation [21,22]. To investigate the effects of c-Ski and Smad3 on skin fibroblast proliferation, we transfected skin fibroblasts with increasing amounts of expression plasmids containing Smad3 or c-Ski or the combined Smad3 and c-Ski. The transfection efficiency was approx. 55.9% (see Supplement 1 at http://www.BiochemJ.org/bj/409/bj4090289add.htm). The effects of Smad3 and c-Ski on fibroblast proliferation were examined at 48 h after transfection. The results showed that c-Ski significantly increased fibroblast proliferation in a dose-dependent manner (Figure 2A). At the highest concentration tested (200 ng/ml), c-Ski increased fibroblast proliferation more than 3-fold. In contrast, Smad3 inhibited dose-dependently the fibroblast proliferation (Figure 2B).
Effect of c-Ski and Smad3 on fibroblast proliferation
Because c-Ski acts as a co-repressor of Smad3 transcription in epithelial cells, we also investigated the effect of co-transfection of Smad3 and c-Ski on skin fibroblast proliferation. c-Ski functions in a similar manner. As shown in Figure 2(C), we co-transfected fibroblasts with a fixed amount (50 ng) of Smad3 expression plasmid with different amounts (0, 20 and 50 ng) of c-Ski expression plasmid. We found that Smad3 alone decreased cell proliferation compared with the cells of the mock transfection. Importantly, co-transfection of c-Ski with Smad3 plasmids reversed dose-dependently the inhibitory effect of Smad3, i.e. the stimulatory effects of c-Ski predominate the inhibitory effects of Smad3 in their regulation of skin fibroblast proliferation. These data suggest that, in skin fibroblasts, c-Ski acts as a co-repressor of Smad3 to promote fibroblast proliferation.
Low and high concentrations of TGF-β1 increase and decrease respectively the protein and mRNA levels of c-Ski, but not Smad3, in skin fibroblasts
To investigate further a possible modulation of c-Ski and Smad3 expression by TGF-β1 at low and high concentrations, we also determined the protein and mRNA levels for c-Ski and Smad3 by Western blotting and real-time PCR respectively. On the basis of the results shown in Figure 1, we selected TGF-β1 at the concentrations of 25 pg/ml and 25 ng/ml for their stimulatory and inhibitory effects respectively. The results showed that TGF-β1 increased c-Ski protein levels at the low concentration (25 pg/ml), but decreased c-Ski levels at the high concentration (25 ng/ml), at 0.5 and 1 h after TGF-β1 treatment (Figure 3A). In parallel with the changes in c-Ski protein levels, c-Ski mRNA expression displayed a similar expression pattern 1 h after TGF-β1 treatment, i.e. the low-concentration of TGF-β1 increased c-Ski mRNA, while the high-concentration of TGF-β1 decreased c-Ski mRNA in fibroblasts (Figure 3C).
Protein and mRNA expression of c-Ski and Smad3 during bi-directional regulation of fibroblast proliferation by TGF-β1
Following TGF-β1 treatment, phosphorylated Smad3 proteins increased at the low as well the high concentration of TGF-β1, with the peak activation at 1 h post-treatment (Figure 3B). On the other hand, non-phosphorylated Smad3 remained unchanged after TGF-β1 treatment. Similarly, expression of Smad3 mRNA remained unchanged 1 h after TGF-β1 treatment (Figure 3C).
Regulation of translocation into the nucleus of Smad3 by TGF-β1 treatment in a dose-dependent manner
Furthermore, we also examined the subcellular location of Smad3 after TGF-β1 treatment by immunohistochemistry. Smad3 immunoreactivity was mainly detected in the cytoplasm when there was no TGF-β1 stimulation (see Supplement 2A at http://www.BiochemJ.org/bj/409/bj4090289add.htm). At 1 h after TGF-β1 treatment, Smad3 immunoreactivity was mainly detected in the nucleus, with diminishing staining in the cytoplasm. This subcellular translocation of Smad3 immunoreactivity from cytoplasm to nucleus was more evident after treatment with the high concentration of TGF-β1 (see Supplement 2C) rather than that of the low concentration of TGF-β1 (see Supplement 2B). These results suggest a nuclear translocation of Smad3 after TGF-β1 treatment. On the other hand, the change in c-Ski expression did not affect the nuclear translocation of Smad3.
A low concentration of TGF-β1 decreases, but a high concentration of TGF-β1 increases, Smad3 DNA-binding ability
c-Ski forms a complex with Smad3 to regulate Smad3 DNA binding and transcription. To investigate further the effect of c-Ski change in Smad3 DNA-binding activity after TGF-β1 treatment, we performed EMSAs using an SBE probe, a putative sequence to which Smad3 can bind . Fibroblasts were transfected with pSuper-c-Ski-RNAi or pSuper-control-RNAi, and Smad3 DNA-binding activity was measured 1 h after the addition of 25 pg/ml or 25 ng/ml TGF-β1. Figure 4 shows that, in the control RNAi group, the Smad3–DNA complex expression increased by 55% at 25 ng/ml (in parallel with the decreased level of c-Ski, Figure 3), but decreased by 44% at 25 pg/ml TGF-β1 (accompanied by increased c-Ski expression, Figure 3) compared with the control. In contrast, efficient c-Ski knockdown (see Figures 6B and 6C) significantly enhanced Smad3–DNA complex expression as compared with the control RNAi group, but the variation range of Smad3–DNA complex expression induced by 25 ng/ml or 25 pg/ml TGF-β1 was obviously attenuated, only increased 26% or decreased 17% respectively.
Change in Smad3 DNA-binding activity during bi-directional regulation of fibroblast proliferation by TGF-β1
Smad3 transcriptional activity decreased at low concentrations of TGF-β1, but increased at high concentrations of TGF-β1
The SBE–Luc reporter gene expression system has been used widely to evaluate the transcription activity of Smad3. Overexpression of Smad3 can increase SBE-mediated transcription . To assess whether the alteration in c-Ski expression results in changes in Smad3-mediated transcription after involving treatment with TGF-β1, we examined the effect of the low and high concentrations of TGF-β1 on Smad3-mediated transcription using a SBE reporter gene assay. We transfected fibroblasts with pcDNA3.0, Smad3 or Smad3 together with c-Ski, together with the SBE reporter. The transfected fibroblasts were then treated with TGF-β1 at the low (25 pg/ml) or high concentration (25 ng/ml). As shown in Figure 5(A), 25 pg/ml TGF-β1 decreased, whereas 25 ng/ml TGF-β1 increased basal SBE reporter activity. Smad3 transfection alone significantly activated the SBE reporter. TGF-β1 at 25 pg/ml significantly inhibited the Smad3-activated SBE reporter activity. In contrast, TGF-β1 at 25 ng/ml significantly increased Smad3-activated SBE reporter activity. Interestingly, co-transfection of c-Ski with Smad3 reversed/prevented the TGF-β1-induced SBE reporter activity. Specifically, overexpression of c-Ski reversed the inhibition and stimulation of Smad3-activated SBE reporter activity after treatment with TGF-β1 at the low and high concentrations respectively.
Effects of low and high concentrations of TGF-β1 on Smad3 transcription activity
In several cell types of human origin, such as normal and transformed epithelial or endothelial cells, TGF-β1 inhibits their proliferation via the induction of cyclin-dependent kinase inhibitors, such as p21CIP1/WAF1 or p15INK4B . It has been reported that p21 and p15 expression depends on Smad3 activity . To further examine the functional consequence of altered SBE-binding activity and Smad3-mediated transcription activity, we selected p21 as a Smad3-dependent target gene, and examined the effects of TGF-β1 at low and high concentrations on p21 expression in fibroblasts. We found that TGF-β1 at 25 pg/ml concentration inhibited, whereas TGF-β1 at 25 ng/ml concentration promoted p21 expression in skin fibroblasts by 60% and 115% respectively (Figure 5B).
Expression of c-Ski modulates the bi-directional actions of TGF-β1 on skin fibroblast proliferation
c-Ski variation induced by low or high concentrations of TGF-β1 is closely relevant to stimulation or repression of fibroblast proliferation. Next, we investigated the effect of c-Ski overexpression or knockdown on the bi-directional actions of TGF-β1 on fibroblast proliferation. Figure 6(A) showed that c-Ski overexpression increased fibroblast proliferation from 66 to 95% at 25 pg/ml, but decreased fibroblast proliferation from 51 to 26% at 25 ng/ml of TGF-β1 compared with that transfected with the pcDNA3.0 mock control. On the other hand, efficient c-Ski knockdown (Figures 6B and 6C) abolished the stimulatory effect of the low concentration of TGF-β1 (25 pg/ml), while attenuating the inhibitory effect of the high concentration of TGF-β1 (25 ng/ml) (Figure 6D).
Effect of c-Ski overexpression or knockdown on the bi-directional proliferative response of TGF-β1 on rat skin fibroblasts
Molecular dissection of the mechanism underlying the bi-directional modulation of TGF-β1 on rat skin fibroblast proliferation reveals a novel feedback loop by which TGF-β1 regulates c-Ski expression (and its interaction with Smad3-mediated transcription) to exert the bi-directional actions. This proposed/working model is supported by the following findings. (i) In parallel with the bi-directional actions of TGF-β1 on cell proliferation, TGF-β1 induced and reduced the expression of c-Ski mRNA and protein in fibroblasts at low and high concentrations. In particular, we provided the first experimental evidence that, at the low concentration, TGF-β1-stimulated cell proliferation is associated with the induction of c-Ski expression. (ii) The cell proliferative effect by low doses of TGF-β1 and c-Ski expression is accomplished by suppressing the Smad3-mediated signalling cascades, including Smad3 binding to SBE sites, SBE-mediated transcription and expression of p21 in fibroblasts. (iii) Overexpression of c-Ski augments the stimulatory effect of TGF-β1 at low concentrations while attenuating the inhibitory effect of TGF-β1 at high concentrations, and knockdown of c-Ski abrogated the bi-directional effect of TGF-β1, indicating that c-Ski is a key determinant of the bi-directional actions of TGF-β1 on fibroblast proliferation. Together, these results delineate a novel regulator model by which TGF-β1 regulates c-Ski expression to exert bi-directional actions on fibroblast proliferation via Smad3–SBE–p21 transcriptional regulation. The identification of the feedback loop via c-Ski provides a plausible explanation for the bi-directional actions of TGF-β1 at low and high concentrations respectively, and thus control of c-Ski expression in physiological and pathological conditions may be an effective therapeutic strategy for wound healing and treatment of tumour development.
TGF-β1 has been shown to produce a bi-directional effect in many cell types, including human fetal lung fibroblasts (HFL-1) , human Tenon's fibroblasts , human newborn thoracic aorta smooth muscle cells and human chondrocytes . Similarly, we have demonstrated a bi-directional effect of TGF-β1 on the proliferation of normal rat skin fibroblasts in vitro, with TGF-β1 stimulating fibroblast proliferation at low concentrations (2.5 pg/ml–2.5 ng/ml) and inhibiting proliferation at high concentrations (25 ng/ml and above) (Figure 1). However, the molecular mechanisms underlying this bi-directional action of TGF-β1 on cell proliferation are not known. The stimulatory action of TGF-β1 was ascribed to a down-regulation of p21CIP1/WAF1 [34,35]. The inhibitory effect of TGF-β1 has been attributed to the up-regulation of p21CIP1/WAF1 and p15INK4B [32,33]. However, these studies did not observe p21 expression after low or high concentrations of TGF-β1 were administered at the same time and in the same cell. We unambiguously demonstrated that the stimulatory and inhibitory effects of TGF-β1 are associated with Smad3-mediated down- and up-regulation of p21 in fibroblasts. This bi-directional regulation of TGF-β1 also involves Smad3 nuclear translocation (see Supplement 2), SBE binding (Figure 4) and transcriptional activities (Figure 5).
What controls bi-directional regulation of p21 expression after treatment with TGF-β1? Previous studies suggesting that an autocrine PDGF loop  and Rb (retinoblastoma protein)  may play an important role in the bi-directional action of TGF-β1 on cell proliferation of connective tissues (see the Introduction), a lack of effect of neutralizing antibody against PDGF and limited expression of PDGF receptors in specified cell types indicate that additional mechanisms are needed to account for the bi-directional regulation of TGF-β1 on cell proliferation. Our studies focused on two key regulators of the TGF-β1 signalling cascade, Smad3 and c-Ski. Smad3 was identified as a specific downstream molecule of the TGF-β1 signalling cascade  and can mediate the inhibitory effect of TGF-β1 on proliferation of epithelial cells  and fibroblasts (Figure 2) by inducing the expression of cyclin-dependent kinase inhibitors, such as p21CIP1/WAF1 and p15INK4B . Following phosphorylation of the TGF-β1 receptor, Smad3 dissociates from the receptor, forms a complex with Smad4 and translocates to nucleus, where they further form complexes with co-activator and co-repressor to regulate the transcription of target genes . Indeed, we demonstrated that this inhibition on fibroblasts by Smad3 is associated with the increased Smad3-mediated inhibition on transcriptional activity and expression of Smad3-mediated downstream genes, such as p21 (Figure 5), which in turn inhibits cell proliferation. However, total endogenous Smad3 mRNA and unphosphorylated protein levels did not change after treatment with low and high doses of TGF-β1. Furthermore, activated Smad3, i.e. phosphorylated Smad3 was increased in a dose-dependent manner 0.5 and 1 h after TGF-β1 treatment at both low and high concentrations, and TGF-β1 at low and high doses both increased nuclear translocation of Smad3. These results indicate changes in the mRNA and total protein levels, and phosphorylation of Smad3 could not account for the bi-directional regulation of TGF-β1 on p21 expression and cell proliferation. Thus Smad3 is not the direct cause of the bi-directional effect of TGF-β1 on cell proliferation. The lack of bi-directional changes in Smad3 protein, phosphorylation and nuclear translocation after the low and high doses of TGF-β1, together with the bi-directional changes in Smad3 DNA-binding activity, Smad3 transcriptional activity strongly suggest that additional regulators come into play to affect Smad3-mediated transcription after TGF-β1 treatment.
We proposed and provide experimental evidence that c-Ski plays an essential role in mediating bi-directional regulation of fibroblast proliferation by low and high doses of TGF-β1. c-Ski is widely distributed in various species and tissues, especially in nervous and muscular tissues. It is involved in various physiological and pathological processes, such as early development of the central nervous system, maturation of macrophages and tissue regeneration [37,38]. c-Ski has attracted much more attention since it was identified as a transcriptional co-repressor of Smad3 in TGF-β1 signalling. c-Ski can form a complex with Smad3 and inhibits Smad3 transcriptional activity. This is achieved by interacting directly with Smad3 to competitively bind SBE sites, recruit NCoR (nuclear receptor co-repressor) and HDAC1 (histone deacetylase 1) and stabilize inactive Smad complexes on SBE sites [24,25,39]. Our data provide the first experimental evidence for the essential role of c-Ski in mediating the bi-directional regulation of fibroblast proliferation through a feedback regulation of c-Ski expression. This novel working model is supported by the following experimental observations. (i) In skin fibroblasts, c-Ski markedly promotes fibroblast proliferation (Figure 2A), and reverses the inhibitory effect of Smad3 on fibroblast proliferation (Figure 2C). The opposing effects of c-Ski and Smad3 on fibroblasts is consistent with previous finding that c-Ski in fact acts a co-repressor of Smad3 in the regulation of epithelial cell proliferation . Moreover, the strong stimulatory effect by c-Ski compared with relatively weak inhibition by Smad3 and the overriding stimulatory effect of c-Ski (the reversal of inhibitory effect of Smad3 by co-transfection of c-Ski) after the co-transfection of c-Ski and Smad3 indicate that the c-Ski effect may predominate over the inhibitory effect by Smad3 in cell proliferation. (ii) c-Ski overexpression enhanced the stimulatory effect of the low concentration of TGF-β1 and abolished the inhibitory effect of the high concentration of TGF-β1, whereas knockdown of c-Ski abrogated the bi-directional effect of TGF-β1. (iii) The critical regulatory role of c-Ski and Smad3 give rise to the question of whether TGF-β1 may produce bi-directional actions on fibroblasts by modulating the expression of c-Ski and Smad3. Indeed, we demonstrated that, in parallel to their stimulatory and inhibitory effects on cell proliferation, the low and high concentrations of TGF-β1 increase and decrease c-Ski mRNA and protein levels. Thus TGF-β1 produces bi-directional actions on cell proliferation by modulating the expression of c-Ski.
Although there are several steps in the TGF-β1 signalling cascade by which c-Ski can influence Smad3-mediated transcription, the finding of the lack of bi-directional change in Smad3 protein phosphorylation and nucleus translocation, and demonstration of bi-directional changes in Smad3 DNA-binding activity and transcription activity, indicate that c-Ski competitively binds the SBE sequence and represses Smad3 transcription. This is consistent with the proposed model of c-Ski that acts as co-repressor of Smad3 in regulating DNA transcription. Thus we propose the following working model: at low concentrations, TGF-β1 treatment increases c-Ski expression, leading to the inhibition of c-Ski on Smad3 DNA binding and transcriptional activity and in turn down-regulation of p21 and increased cell proliferation. After treatment with high concentrations of TGF-β1, c-Ski expression decreases, leading to the dis-inhibition of c-Ski on Smad3 DNA binding and transcriptional activity and in turn up-regulation of p21 and suppression of cell proliferation. That is, c-Ski expression regulated by low or high concentrations of TGF-β1 results in a change of p21 expression, leading to alteration in the composition of p21WAF and p27KIP in complexes with cyclins D1 and E, and therefore Rb phosphorylation switching, resulting in cell proliferation or inhibition. This provides an explanation for early studies that suggest that the different Rb phosphorylation state is the switch for stimulatory and inhibitory effects of TGF-β1 [41,42].
How does the high concentration of TGF-β1 reduce c-Ski expression? In mink lung epithelial cells, Ski and Ski-related protein, Sno, exhibited a rapid disappearance in a manner dependent on the concentration of TGF-β1 . Proteasomes are involved in the rapid turnover of Ski protein [43,44]. In some cells, ligand activation of the TGF-β1 receptor may cause the rapid degradation of pre-existing Ski–Sno, resulting in the removal of an important obstacle to TGF-β1-mediated transcriptional activation . However, in other cells, the Ski–SnoN protein complex may persist after TGF-β1 receptor activation, leading to continued blocking of the TGF-β1 signalling cascade involving the Smad3/Smad4 transcription factors. Edmiston et al.  found an inability of TGF-β1 to cause SnoN degradation, leading to resistance to TGF-β1-induced growth arrest in oesophageal cancer cells. Accordingly, the well-documented highly variable effects of TGF-β1 on various cell types may be attributable to the background of ancillary proteins, such as Ski and SnoN, which are capable of blunting and redirecting specific downstream effectors pathways emanating from the TGF-β1 pathway . In our bi-directional regulation of TGF-β1 on skin fibroblast proliferation, we supposed a pattern that a low concentration of TGF-β1 promotes c-Ski expression and a high concentration of TGF-β1 decreases c-Ski expression, resulting in the decrease and increase in Smad3 transcriptional activity and down- and up-regulation of p21. Thus our experiments suggest an important role for c-Ski in modulating the local availability of TGF-β1 within the tissue microenvironment of cells and exemplify the delicate homoeostasis maintained for cell proliferation. Given the critical involvement of the TGF-β1-mediated signalling cascade in tumour development and wound healing [23,46], the identification of the essential role of c-Ski in mediating the bi-directional regulation of TGF-β1 on fibroblast proliferation may lead to an effective therapeutic strategy for wound healing and treatment of tumour development.
We thank Ping Zhou and Hua Wang for their technical assistance. We thank Dr Jiangfan Chen for his reading and suggested revisions of the manuscript and Dr Miyazono and Dr Heldin for kindly providing us with the plasmids. This work was supported by National Key Basic Research Program of China (numbers G1999054205, 2005cb522602) and the National Natural Science Foundation of China (number 30400468).
Dulbecco's modified essential medium
electrophoretic mobility-shift assay
fetal bovine serum
platelet-derived growth factor
small interfering RNA
transforming growth factor β1
These authors contributed equally to this work.
The nucleotide sequence data reported for rat c-Ski mRNA will appear in GenBank®, EMBL, DDBJ and GSDB Nucleotide Sequence Databases under the accession number DQ409171.