Distinct Roles of Smad2-, Smad3-, and ERK-dependent Pathways in Transforming Growth Factor-β1 Regulation of Pancreatic Stellate Cellular Functions*

Pancreatic stellate cells (PSCs) play a major role in promoting pancreatic fibrosis. Transforming growth factor-β1 (TGF-β1) regulates PSC activation and proliferation in an autocrine manner. The intracellular signaling pathways of the regulation were examined in this study. Immunoprecipitation and immunocytochemistry revealed that Smad2, Smad3, and Smad4 were functionally expressed in PSCs. Adenovirus-mediated expression of Smad2, Smad3, or dominant-negative Smad2/3 did not alter TGF-β1 mRNA expression level or the amount of autocrine TGF-β1 peptide. However, expression of dominant-negative Smad2/3 inhibited PSC activation and enhanced their proliferation. Co-expression of Smad2 with dominant-negative Smad2/3 restored PSC activation inhibited by dominant-negative Smad2/3 expression without changing their proliferation. By contrast, co-expression of Smad3 with dominant-negative Smad2/3 attenuated PSC proliferation enhanced by dominant-negative Smad2/3 expression without altering their activation. Exogenous TGF-β1 increased TGFβ1 mRNA expression in PSCs. However, PD98059, a specific inhibitor of mitogen-activated protein kinase kinase (MEK1), inhibited ERK activation by TGF-β1, and consequently attenuated TGF-β1 enhancement of its own mRNA expression in PSCs. We propose that TGF-β1 differentially regulates PSC activation, proliferation, and TGF-β1 mRNA expression through Smad2-, Smad3-, and ERK-dependent pathways, respectively.

possess fat droplets containing vitamin A and are quiescent. In the quiescent state they are characterized by desmin-positive but ␣-smooth muscle actin (␣-SMA)-negative staining (1). When cultured in vitro, PSCs are autoactivated (autotransformed), changing their morphological and functional features (2). PSCs commence losing vitamin A containing lipid droplets, highly proliferating, increasing expression of ␣-SMA, and producing and secreting extracellular matrix components such as collagen and fibronectin. Namely, PSCs are autotransformed to myofibroblast-like cells. In vivo, PSCs are also activated during both human and experimental pancreatic fibrosis (3). Therefore, PSCs are believed to play an important role in pancreatic fibrogenesis.
TGF-␤ 1 is one of major profibrogenic cytokines in various tissues. Recently, evidence for TGF-␤ participation in pancreatic fibrogenesis has been mounting. In this regard, it has been noted that transgenic mice overexpressing TGF-␤ 1 in islet cells develop fibrosis of exocrine pancreas (4). Moreover, inhibition of TGF-␤ 1 by anti-TGF-␤ 1 antibody reduced extracellular matrix production in rat cerulein pancreatitis (5). TGF-␤ 1 has also been shown to promote PSC activation and collagen production and to inhibit proliferation of PSCs in an autocrine manner (6,7). In human chronic pancreatitis tissue, TGF-␤ 1 expression was observed in acinar cells adjacent to areas of fibrosis and in spindle cells in fibrotic bands (3). Thus, TGF-␤ 1 is thought to promote pancreatic fibrosis, in part by modulating PSC functions. However, the intracellular signaling pathway(s) through which TGF-␤ 1 regulates PSC functions is still uncertain.
Sma-and Mad-related proteins (Smads) are a group of recently identified molecules that function as intracellular signaling mediators and modulators of TGF-␤ family members (8,9). Smads can be classified into three groups: receptor-regulated Smads (R-Smads), common mediator Smad (Co-Smads), and inhibitory Smads (I-Smad). In TGF-␤ signaling pathway, Smad2 and Smad3 function as R-Smads, Smad4 functions as a Co-Smad, and Smad7 functions as an I-Smad. Upon TGF-␤ binding to TGF-␤ type II receptor, the type II receptor kinase phosphorylates the GS domain of TGF-␤ type I receptor, leading to activation of the type I receptor. The activated type I receptor kinase phosphorylates Smad2 and Smad3 (R-Smads) at two serine residues in the SSXS motif at their extreme C termini (10,11). Phosphorylated Smad2 and Smad3 form oligomeric complexes with Smad4 (Co-Smad); the complexes then translocate into the nucleus. These complexes then activate the transcription of target genes. Thus, TGF-␤ intracellular signaling involves dual Smad-dependent pathways, namely, Smad2-and Smad3-dependent pathways. In addition to Smad-mediated signaling pathways, other signaling pathways have also been shown to mediate TGF-␤ signaling. For example, TGF-␤ activates Rho-GTPase, mitogen-activated protein kinases, and protein kinase B (12). However, a direct link between these mediators and TGF-␤ receptors has not been demonstrated unequivocally (12).
Because TGF-␤ 1 stimulus is transduced through the multiple intracellular pathways described above, the elucidation of the signaling pathways through which TGF-␤ 1 regulates PSC functions is likely to provide new insights related to the molecular pathogenesis of pancreatic fibrosis. We therefore conducted the present study to examine these pathways by applying adenovirus-mediated overexpression of Smad2 and Smad3. However, because Smad2 and Smad3 compete with each other at the receptor and for Smad4-binding steps for their activation, overexpression of Smad2 and Smad3 blocks endogenous Smad3 and Smad2 functions by competing at the binding steps to TGF-␤ receptor and Smad4. Thus, the possibility remains that the observed effects of Smad2 and Smad3 overexpression on PSC functions may result from the blockade of endogenous Smad3 and Smad2 function but may not result from the enhancement of Smad2 and Smad3 activity by their overexpression. To exclude this possibility, we investigated Smad2-and Smad3-specific roles in TGF-␤ 1 regulation of PSC functions by co-expression of dominant-negative Smad2/3 with Smad2 or Smad3. Although the dominant-negative Smad2/3 mutant was generated by substituting Glu for Asp-407 of Smad3, which is defective in TGF-␤ receptor-dependent phosphorylation, this mutant possesses a dominant-negative effect on both Smad2 and Smad3 (13). Thus, we designated the mutant as dominantnegative Smad2/3. In this way, expression of dominant-negative Smad2/3 blocks both endogenous Smad2 and Smad3 functions at TGF-␤ receptor-dependent phosphorylation step. Therefore, the co-expression of Smad2 or Smad3 with dominant-negative Smad2/3 rescues only Smad2-or Smad3-dependent pathways, respectively. Thus, we can examine Smad2-and Smad3-specific signaling pathways. Using this method, we demonstrated that TGF-␤ 1 activates PSCs through a Smad2dependent pathway and inhibits their proliferation through a Smad3-dependent pathway. Moreover, TGF-␤ 1 enhanced its own mRNA expression and peptide secretion of PSCs through an ERK-dependent pathway.
Isolation and Culture of Rat Pancreatic Stellate Cells-Rat pancreatic stellate cells were prepared as described (1). Briefly, rat pancreas was digested in Gey's balanced salt solution supplemented with 0.05% collagenase P, 0.02% Pronase, and 0.1% DNase. After filtration through nylon mesh, the cells were centrifuged on a 13.2% Nycodenz gradient at 1400 ϫ g for 20 min. PSCs were collected from the band just above the interface of the Nycodenz solution and the aqueous layer, washed, and resuspended in Iscove's modified Dulbecco's medium containing 10% fetal calf serum, 100 units/ml penicillin, and 100 mg/ml streptomycin. PSCs were cultured in a 5% CO 2 atmosphere at 37°C. All of the experiments were carried out using PSCs from passages 2 and 3.
Immunoprecipitation and Western Blotting-Immunoprecipitation was performed as described previously (14). Western blotting was carried out as described before (15), using enhanced chemiluminescence reagent to visualize the secondary antibody.
Adenovirus Infection-Recombinant adenoviruses containing recom-binant Smad DNAs were kindly provided by Dr. Miyazono (University of Tokyo, Tokyo, Japan). For a single adenovirus infection, the cells were infected with a recombinant adenovirus at a dose of 10 plaqueforming units (pfu)/cell in the culture media described above. In the experiments using double adenovirus infection, the cells were infected with dominant-negative Smad2/3 adenovirus (AdDNSmad2/3) at a dose of 10 pfu/cell, simultaneously with Smad2 (AdSmad2) or Smad3 (AdSmad3) adenovirus at doses of 1, 5, or 10 pfu/cell. Subsequent experiments were performed 48 h after infection. An adenovirus expressing ␤-galactosidase (AdLacZ) was used as an infection control. Immunofluorescence Microscopy-Immunofluorescence microscopy was performed as described previously (15,16). The samples were examined by epifluorescence microscopy (see Fig. 2) and confocal fluorescence microscopy (see Fig. 3) (Fluoview FV300; Olympus, Tokyo, Japan) using an Olympus BX51 microscope. The images were digitized and then processed using Photoshop 5.0 software (Adobe Systems Inc., Mountain View, CA).

Measurement of DNA Synthesis-DNA synthesis was determined by measuring [ 3 H]thymidine incorporation into cells. [ 3 H]
Thymidine was added to the culture medium and incubated for 2 h, and the incorporation of radioactivity was measured as described previously (17).
Measurement of TGF-␤ 1 Peptide Secretion-TGF-␤ 1 peptide secretion was examined by determining the concentration of TGF-␤ 1 peptide in a culture medium of PSCs using a commercial kit from DRG International (Mountainside, NJ), according to the manufacturer's instructions.
Competitive Reverse Transcription-PCR of TGF-␤ 1 mRNA-Total RNA was obtained from PCS cells by using ISOGEN (Wako, Tokyo, Japan), followed with synthesis of double-stranded DNA as described FIG. 1. Immunoprecipitation of Smad2, Smad3, and Smad4. Smad proteins were immunoprecipitated (IP) from PSC lysate using polyclonal antibodies against each Smad protein. The immunoprecipitates were fractioned on a 10% sodium dodecyl sulfate-polyacrylamide gel and transferred to a nitrocellulose membrane. Western blotting of the immunoprecipitates was carried out with the same antibodies used for each immunoprecipitation.

FIG. 2. TGF-␤ 1 induces nuclear translocation of both Smad2
and Smad3 in PSCs. Immunocytochemical analysis of nuclear accumulation of Smad2 and Smad3 was performed using PSCs that were transfected with Smad2 or Smad3 adenoviral vectors and overexpressed the proteins. A and B, infection efficiency of adenovirus was determined by using AdLacZ infection and in situ staining with X-gal. C-F, PSCs were infected with AdSmad2 (C and D) or AdSmad3 (E and F). The cells were stained with anti-Smad2 (C and D) or anti-Smad3 (E and F) antibodies before (C and E) and after (D and F) 2 h of stimulation with 10 pM TGF-␤ 1 . Bars, 40 m.
previously (18). Competitive PCR of TGF-␤ 1 mRNA was performed using the competitive PCR kit for rat TGF-␤ 1 (Maxim Biotech Inc., San Francisco, CA) according to the manufacturer's instructions. In this method, 189-and 250-bp PCR fragments are generated by amplifying a DNA competitor and rat TGF-␤ 1 cDNA, respectively.

Expression of Smad2, Smad3, and Smad4 Proteins in PSCs-
We first examined the expression of R-Smads (Smad2 and Smad3) and Co-Smad (Smad4) in rat PSCs. To this end, we performed immunoprecipitation of Smad2, Smad3, and Smad4 from crude extract of rat PSCs, using antibodies specific to each Smad. As shown in Fig. 1, all of the three Smads were immunoprecipitated from rat PSCs, suggesting that essential components of both Smad2-and Smad3-dependent TGF-␤ 1 signaling pathways are present in PSCs.
TGF-␤ Induced the Nuclear Accumulation of both Smad2 and Smad3 in PSCs-We next examined whether Smad2-and Smad3-dependent TGF-␤ 1 signaling pathways are functioning in PSCs. Because the levels of endogenously expressed Smad2 and Smad3 are not sufficient to detect by immunochemistry using specific antibodies, we used adenovirus-mediated overexpression of Smad2 and Smad3 in PSCs. We first determined the infection efficiency by using AdLacZ infection and in situ staining with X-gal. As shown in Fig. 2, more than 98% of   AdLacZ-infected PSCs expressed ␤-galactosidase (Fig. 2, A and  B). In PSCs infected with AdSmad2 and AdSmad3, sufficient expression of Smad2 and Smad3 was observed by immunocytochemistry (Fig. 2, C and E). When treated with 10 pM TGF-␤ 1 , both Smad2 and Smad3 accumulated in the nucleus (Fig. 2, D  and F). These data indicate that functional Smad2-and Smad3dependent signaling pathways are present in PSCs.
Infection Efficacy and Characterization of AdDNSmad2/3 in PSCs-We subsequently attempted to elucidate the intracellular signaling pathways through which autocrine TGF-␤ 1 modulates PSC functions using AdSmad2, AdSmad3, and AdDNS-mad2/3. As described above, infection of AdSmad2 and AdSmad3 into cells resulted in expression of sufficient quantities of functional Smad2 and Smad3 proteins in PSCs, respectively. Thus, we next characterized AdDNSmad2/3 infection in PSCs with confocal immunofluorescence microscopy. In this experiment, we utilized anti-Smad3 antibody to identify dominant-negative Smad2/3 protein expression because AdDNS-mad2/3 was generated by substituting Glu for Asp-407 of smad3. As shown in Fig. 3, AdDNSmad2/3 infection induced expression of sufficient quantities of dominant-negative Smad2/3 protein throughout the cytoplasm in PSCs (Fig. 3A). When stimulated with 10 pM TGF-␤ 1 , dominant-negative Smad2/3 did not accumulate in the nucleus but remained in the cytoplasm (Fig. 3B). These data demonstrate that the expressed dominant-negative Smad2/3 protein is functionally defective in TGF-␤ signaling. In COS7 cells, the expression of dominant-negative Smad2/3 has been shown to inhibit both Smad2-and Smad3-dependent TGF-␤ signaling pathways by blocking Smad2 and Smad3 phosphorylation by TGF-␤ type 1 receptor (13). We thus performed the next experiments to confirm the dominant-negative nature of AdDNSmad2/3 infection in TGF-␤ signaling in PSCs. To this end, we investigated the effect of AdDNSamd2/3 and AdSmad2 double infection on Smad2 function in PSCs. When AdLacZ (10 pfu/cell) and AdS-mad2 (1 pfu/cell) were co-infected, overexpressed Smad2 accumulated in the nucleus after stimulation with 10 pM TGF-␤ 1 (Fig. 3, C and D). In contrast, when AdDNSmad2/3 (10 pfu/cell) was co-infected with AdSmad2 (1 pfu/cell) into PSCs, overexpressed Smad2 did not accumulate in the nucleus but remained in the cytoplasm (Fig. 3, E and F). These data imply that dominant-negative Smad2/3 blocks Smad2 nuclear accumulation after TGF-␤ stimulation, indicating the dominant-negative effect of AdSmad2/3 infection on Smad-dependent TGF-␤ signaling pathway in PSCs. We further examined the effect of higher dose infection of AdSmad2 on the dominant-negative effect of AdSmad2/3 infection. When AdSmad2 was infected at the dose of 10 pfu/cell with AdDNSmad2/3 (10 pfu/cell), overexpressed Smad2 accumulated in the nucleus after stimulation with 10 pM TGF-␤ 1 , although it remained in the cytoplasm to some extent (Fig. 3, G and H). These data indicate that the Smad2-and Smad3-dependent signaling pathways attenuated by AdDNSmad2/3 infection can be rescued by higher doses of co-infected AdSmad2 or AdSmad3.
Infection of AdSmad2, AdSmad3, or AdDNSmad2/3 Did Not Alter TGF-␤ 1 mRNA Expression or TGF-␤ 1 Peptide Secretion of PSCs-After the evaluation of the effects of infection of these adenoviruses on TGF-␤ signaling pathways in PSCs as described above, we examined whether their infection into PSCs alters TGF-␤ 1 mRNA expression level and the amount of autocrine TGF-␤ 1 peptide of PSCs. As shown in Fig. 4, TGF-␤ 1 mRNA expression level of PSCs was not altered by the infection of AdSmad2, AdSmad3, or AdDNSmad2/3. In addition, TGF-␤ 1 peptide secretion from PSCs was not affected by their infection (Table I). These data indicate that we can observe the effect of these adenoviruses infections on diverse PSC functions modulated by autocrine TGF-␤ 1 , regardless of the effect of the infection on the amount of autocrine TGF-␤ 1 .
Effects of Smad2, Smad3, and Dominant-negative Smad2/3 Overexpression on PSC Activation-We examined involvement of Smad2 and Smad3 in PSC activation using adenovirusmediated overexpression of the proteins. PSC activation was examined by determining the amount of ␣-SMA protein in PSCs with Western blotting. Overexpression of dominant-negative Smad2/3 inhibited PSC activation (Fig. 5), indicating that TGF-␤ 1 activates PSCs through a Smad-dependent pathway. Moreover, overexpression of Smad2 but not Smad3 enhanced PSC activation, suggesting that TGF-␤ 1 activates PSCs through a Smad2-dependent but not a Smad3-dependent pathway. However, because both Smad2 and Smad3 compete with each other for receptor and for Smad4 binding steps for the activation of their pathways as described above, the possibility remained that the positive effect of Smad2 overexpression on PSC activation may have resulted from inhibition of endogenous Smad3 function and not the involvement of Smad2-dependent pathway in TGF-␤ 1 induced PSC activation. To exclude this possibility, we investigated Smad2-and Smad3specific roles in TGF-␤ 1 -induced PSC activation by co-infection of AdDNSmad2/3 with AdSmad2 or AdSmad3. As shown in Fig.  5, AdSmad2 co-infection with AdDNSmad2/3 rescued PSC activation inhibited by AdDNSmad2/3. On the other hand, AdS-mad3 co-infection with AdDNSmad2/3 did not alter PSC activation inhibited by AdDNSmad2/3. These data suggest that TGF-␤ 1 activates PSCs through a Smad2-dependent pathway .
Effects of Smad2, Smad3, and Dominant-negative Smad2/3 Overexpression on PSC Proliferation-We next examined the pathway through which TGF-␤ 1 inhibits PSC proliferation. PSC proliferation was examined by determining DNA synthesis by means of [ 3 H]thymidine incorporation. AdDNSmad2/3 infection enhanced PSC proliferation, whereas AdSmad3 infection inhibited it (Fig. 6), suggesting that TGF-␤ 1 inhibits PSC proliferation through a Smad3-dependent pathway. Moreover, results of the co-infection method showed that AdSmad3 coinfection with AdDNSmad2/3 inhibited PSC proliferation enhanced by AdDNSmad2/3. On the other hand, AdSmad2 coinfection with AdDNSmad2/3 did not alter PSC proliferation augmented by AdDNSmad2/3 infection. These data suggest that TGF-␤ 1 inhibits PSC proliferation through a Smad3-dependent pathway.
TGF-␤ 1 Enhanced TGF-␤ 1 mRNA Expression of PSCs through an ERK-dependent Pathway-As described above, TGF-␤ 1 mRNA expression of PSCs was not affected by the infection of AdSmad2, AdSmad3, or AdDNSmad2/3, suggesting that the regulation of TGF-␤ 1 mRNA expression in PSCs is independent of Smad-dependent signaling pathways. Thus, we next attempted to elucidate the regulatory mechanism of TGF-␤ 1 mRNA expression in PSCs. To this end, we first examined whether TGF-␤ 1 modulates its own mRNA expression in PSCs. As shown in Fig. 7A, the addition of exogenous TGF-␤ 1 into the culture medium of PSCs enhanced TGF-␤ 1 mRNA expression in a dose-dependent manner, indicating TGF-␤ 1 autoinduction independent of Smad-mediated signaling. Because mitogen-activated protein kinases including ERKs are also TGF-␤ signaling mediators (12), we examined the participation of ERK-dependent pathway in the autoinduction of TGF-␤ 1 mRNA in PSCs. For this purpose, we blocked ERK activation by using the MEK1 inhibitor PD98059. PD98059 pretreatment decreased TGF-␤ 1 mRNA expression in PSCs (Fig. 7B, first and second lanes). Moreover, the addition of exogenous TGF-␤ 1 into the culture medium could not enhance TGF-␤ 1 mRNA expression in PSCs pretreated with PD98059 ( Fig. 7B, first and second lanes). Consistent with these data, PD98059 pretreatment decreased TGF-␤ 1 peptide secretion from PSCs (Fig. 7C). Finally, we confirmed that TGF-␤ 1 activates ERK in PSCs (Fig. 7D, first and second lanes), and PD98059 pretreatment blocked ERK activation (Fig. 7D, third  and fourth lanes) . These data indicate that TGF-␤ 1 autoinduction in PSCs is regulated through an ERK-dependent pathway.
Although TGF-␤ signaling is mediated by both Smad2-and Smad3-dependent pathways, their functional difference has been uncertain. However, studies using targeted homozygous deletion of Smad2 and Smad3 genes in mice revealed their distinct functions in embryo development. Smad2 knockout mice are embryonic lethal because of the defects of left-right patterning and mesoderm induction (19,20). In contrast, Smad3 knockout mice are viable but are smaller than wild-type littermates and show forelimb malformation and die because of immune function defects (21,22). In addition, using hepatic stellate cells (HSCs) derived from the Smad3 knockout mice, Schnabl et al. (23) recently elucidated the Smad3 specific role in cellular function. They reported that Smad3 is necessary for TGF-␤-mediated inhibition of HSC proliferation but not for HSC activation, which is consistent with our data on Smad3 function in PSC proliferation. However, the specific role of Smad2 in HSC function has not yet been demonstrated. Recently, using fibroblasts derived from both embryos of Smad2 and Smad3 knockout mice, Piek et al. (24) reported that Smad2 and Smad3 mediated the transcription of distinct genes in fibroblasts stimulated by TGF-␤. Indeed, these cell systems are useful for comparing the functions of Smad2 and Smad3 in cells derived from embryos of knockout mice. However, because Smad2 knockout is lethal to embryonic mice because of embryo development defects, it had been difficult to examine specific roles of Smad2 and Smad3 concurrently in fully differentiated cells derived from matured organs. In the present study, however, we have demonstrated specific roles of Smad2 and Smad3 in TGF-␤ 1 regulation of PSC functions isolated from the mature pancreas by employing adenovirus-mediated co-expression of Smad2 or Smad3 with dominant-negative Smad2/3. It is noteworthy that this method can be widely applied for the study on TGF-␤ intracellular signaling pathway in a variety of mature organs because adenovirus-mediated gene transfer is highly effective in various cell types.
Our present observations that TGF-␤ 1 activates PSCs through a Smad2-dependent pathway and inhibits PSCs proliferation through a Smad3-dependent pathway provide a novel therapeutic strategy for pancreatic fibrosis. Because TGF-␤ 1 is a key activator of PSCs (7) and a main inducer of pancreatic fibrosis (4), the therapeutic effect of TGF-␤ 1 stimulus inhibition on pancreatic fibrosis has been extensively studied. For example, Menke et al. (5) reported that inhibition of TGF-␤ 1 by injection of neutralizing TGF-␤ 1 antibody reduced extracellular matrix formation in pancreatitis in vivo. However, TGF-␤ 1 is also an autocrine inhibitor of PSC proliferation (6). Thus, blockade of TGF-␤ 1 activity promotes PSC proliferation. If TGF-␤ 1 stimuli on PSC activation could be selectively blocked without diminishing the TGF-␤ 1 inhibitory effect on PSC proliferation, it could be a more potent therapeutic method for pancreatic fibrosis. In this respect, our present data indicate that selective blockade of the Smad2-dependent pathway without affecting the Smad3-dependent pathway can be a novel strategy for the treatment of pancreatic fibrosis .
Our data on TGF-␤ 1 mRNA expression and peptide secretion of PSCs are important. Because TGF-␤ 1 mRNA expression and TGF-␤ 1 peptide secretion of PSCs were not affected by overexpression of Smad2, Smad3, or dominant-negative Smad2/3, we could apply their overexpression and co-expression to observe Smad2-and Smad3-specific roles in the regulation of PSC function by autocrine TGF-␤ 1 . We also demonstrated that TGF-␤ 1 enhanced its own mRNA expression through a Smad-independent but ERK-dependent pathway. To date, TGF-␤ 1 has been shown to augment the expression of its own mRNA (25) in both normal and transformed cells, and the promoter sequences of TGF-␤ 1 gene responsive to the autoinduction have been identified (26). In addition, Yue and Mulder (27) reported that MEK-ERK pathway activation is required for TGF-␤ 1 expression induced by TGF-␤ 3 , a TGF-␤ isoform derived from a gene distinct from that of TGF-␤ 1 . However, the intracellular signaling pathway of TGF-␤ 1 autoinduction has never been demonstrated. Thus, to our knowledge, this is the first report that has elucidated the intracellular signaling pathway of TGF-␤ 1 autoinduction.
In conclusion, we showed that TGF-␤ 1 regulates PSC activation, proliferation, and TGF-␤ 1 mRNA expression through Smad2-, Smad3-, and ERK-dependent pathways, respectively. These observations provide new insights for understanding the mechanism of pancreatic fibrosis and developing a novel therapeutic strategy for its treatment.