Stimulation of Proliferation of Rat Hepatic Stellate Cells by Galectin-1 and Galectin-3 through Different Intracellular Signaling Pathways

We found that the expression of galectin-1 and galectin-3 was significantly up-regulated in hepatic stellate cells (HSCs) both in the course of their transdifferentiation into myofibroblasts, a process of “self-activation,” and in the fibrosis of liver tissues. Recombinant galectin-1 and galectin-3 stimulated the proliferation of cultured HSCs via the MEK1/2-ERK1/2 signaling pathway. However, galectin-3 utilized protein kinases C and A to induce this process, whereas galectin-1 did not. We also found that thiodigalactoside, a potent inhibitor of β-galactoside binding, attenuated the effects of both galectins. In addition, galectin-1, but not galectin-3, promoted the migration of HSCs. Thus, it appears that galectin-1 and galectin-3, generated by activated HSCs, could participate in β-galactoside binding and induce different intracellular signaling pathways leading to the proliferation of HSCs.

These growth factors trigger the proliferation and the secretion of ECMs by HSCs. Recent studies have elucidated that MAPK and phosphatidylinositol 3-kinase (PI3K) are key signaling pathways involved in the growth factor-induced stimulation of HSCs (4,5).
There have been a few reports on the function of galectins in the physiologic and pathologic processes of the liver (23,(32)(33)(34). Galectin-3 expression was induced in regenerative nodules of cirrhotic liver tissues and in hepatocellular carcinomas (33). The expression of galectin-1 and galectin-3 was examined in relation to the progression and infiltration of intrahepatic cholangiocarcinomas and to preneoplastic and early neoplastic changes of biliary epithelial cells (23). A study on concanavalin A-induced acute liver injury showed that galectin-1 administration protects mice from liver injury by selectively eliminating activated T cells and preventing the synthesis of proinflammatory Th1-derived cytokines (34).
Our previous proteome analysis of rat HSC proteins revealed that the amount of galectin-1 is greatly increased in activated HSCs compared with quiescent HSCs; the secretion of galectin-1 is also much enhanced by activated HSCs (32). Although a detailed analysis has not yet been performed, these results suggest that an enhanced level of galectin-1 modifies the function of HSCs themselves as well as other hepatic constituent cells in chronically injured liver.
In this study, we investigated in detail the expression pattern of galectin-1 and galectin-3 in HSCs and in fibrotic liver tissues. We further demonstrated that both types of galectins activate MAPK pathways presumably by cross-linking with target molecules through their ␤-galactoside-containing glycoconjugates, leading to the proliferation of HSCs.
Animals-Pathogen-free male Wistar rats were obtained from Japan SLC, Inc. (Shizuoka, Japan). Animals were housed at a constant temperature and supplied with laboratory chow and water ad libitum.
Production of Recombinant Galectin-1 and Galectin-3-Recombinant human galectin-1 (r-galectin-1) and r-galectin-3 were generated in our laboratory as described previously (35). Briefly, DNA fragments encoding either human galectin-1 or galectin-3 were amplified by PCR using cloned cDNA as a template. The amplified fragments were ligated to pET21a. Generated prokaryotic expression vectors were used to transform Escherichia coli BL21(DE3) cells. Recombinant proteins were induced in the cells by 1 mM isopropyl-␤-D-thiogalactopyranoside. They were purified by affinity chromatography on asialofetuin-Sepharose 4B, which was prepared according to Oda et al. (36) and Hirabayashi and Kasai (37).
Liver Fibrosis Models-Liver fibrosis was induced in rats either by intraperitoneal injection of TAA (50 mg/body) twice a week for 8 weeks (38) or by ligation of the common bile duct for 2 weeks. After fibrosis developed, the peritoneal cavities of the rats were opened under ether anesthesia. The liver was perfused with phosphate-buffered saline via the portal vein to remove the blood completely and subsequently removed. Part of the liver was fixed in 4% paraformaldehyde and used for histological evaluation. The remaining parts were quickly frozen in liquid nitrogen and stored at Ϫ80°C until used.
Preparation of Hepatic Constituent Cells-Hepatic constituent cells were isolated from rat livers as previously described in detail (39). Kupffer cells and sinusoidal endothelial cells were used immediately after isolation. HSCs were plated on plastic dishes in DMEM supplemented with 10% FBS (FBS/DMEM). Hepatocytes were cultured on type I collagen-coated dishes in Williams' Medium E supplemented with 10% FBS (FBS/Williams' Medium E). The culture media were changed every day. HSCs isolated from normal and fibrotic livers are referred to as quiescent HSCs and in vivo activated HSCs, respectively, in this study (40). Quiescent HSCs cultured for 7 days are referred to as in vitro activated HSCs (41).
Northern Blotting-Total RNA was extracted from cultured HSCs and liver tissues using Isogen. Total RNA (10 g) was separated on a 1% agarose gel and transferred onto a nylon membrane. After prehybridization, the membrane was incubated in buffer supplemented with PCR-amplified double-stranded cDNAs for galectin-1, galectin-3, and glyceraldehyde-3-phosphate dehydrogenase, which were labeled with [␣-32 P]dCTP using the Rediprime DNA labeling system; this was followed by autoradiography on Kodak XAR-5 x-ray film.
Immunoblotting-HSCs were cultured in the presence or absence of test agents and then homogenized in buffer consisting of 62.5 mM Tris, 0.1% glycerol, 2% SDS, and 5% 2-mercaptoethanol (pH 6.8). After the samples (10 g of protein) were heat-denatured, they were analyzed by 7.5-15% SDS-PAGE and then transferred onto Immobilon P membranes. After blocking, the membranes were treated for 2 h at room temperature with individual antibodies. After washing, they were incubated with horseradish peroxidase-conjugated secondary antibodies. Immunoreactive bands were visualized on Kodak XAR-5 film using ECL detection reagent.
Immunohistochemistry-Immunohistochemical detection of galectin-1 and galectin-3 in rat liver tissues was performed according to the method described in detail by Nakatani et al. (42). Immunoprecipitates were visualized using 0.025% 3,3Ј-diaminobenzidine tetrahydrochloride and 0.003% H 2 O 2 . Specimens were observed under an Olympus IX70 microscope.
Cell Growth Assays-Isolated HSCs were cultured on plastic dishes for 3 days in FBS/DMEM and then maintained for 24 h in serum-free DMEM. Isolated hepatocytes were cultured on type I collagen-coated dishes for 24 h in FBS/Williams' Medium E and then maintained for 24 h in serum-free Williams' Medium E. The cells were successively stimulated with test agents for 24 h and pulse-labeled with 1.0 Ci/ml [ 3 H]thymidine during the last 6 h. The incorporated radioactivity was subjected by liquid scintillation counting as previously described (38). In another experiment, isolated HSCs (1 ϫ 10 5 cells/well) were plated in six-well culture plates, incubated for 2 days in FBS/DMEM, and maintained for 24 h in serum-free DMEM. The cells were successively stimulated with either r-galectin-1 or r-galectin-3 for 24 or 48 h. They were successively fixed in 100% methanol and stained with a Giemsa solution. The number of HSCs was counted in a microscopic field (0.84 mm 2 ) at a magnification of ϫ100. Five microscopic fields were randomly chosen for each specimen.
Migration Assay-The migration activity of HSCs was assayed using cell culture insert as previously described (43). HSCs (1 ϫ 10 5 cells) cultured for 2 days were detached from the plates using trypsin, suspended in 400 l of FBS/DMEM, introduced into the insert, and allowed to adhere to the upper surface of the membrane. HSCs were then maintained in serum-free DMEM for 24 h. Recombinant rat PDGF-BB (r-PDGF-BB; 20 ng/ml) or r-galectin-1 or r-galectin-3 (10 g/ml) was successively introduced into the upper chamber. After 48 h of incubation, the culture medium was removed. Cells adhering to the membrane were fixed in 100% methanol and stained with a Giemsa solution. The number of HSCs on the upper surface of the membrane was counted at a magnification of ϫ400, and that on the lower surface was counted similarly by changing the focus. Five microscopic fields were randomly chosen for each specimen. The proportion of migrated cells (termed migration index) was calculated as follows: migration index (%) ϭ (number of cells on lower surface of membrane)/(number on upper and lower surfaces of membrane) ϫ 100.
Statistical Analysis-Data presented as bar graphs are means Ϯ S.D. of three independent experimental series. Statistical analysis was performed by Student's t test at a significance level of p Ͻ 0.05.

Expression of Galectin-1 and Galectin-3 in Rat Hepatic Constituent Cells and Liver
Tissues-HSCs, hepatocytes, Kupffer cells, and sinusoidal endothelial cells were isolated from normal rat livers. The expression of galectin-1 and galectin-3 mRNAs was determined by RT-PCR (Fig. 1A). Galectin-1 mRNA was not detectable in any of the cells, whereas galectin-3 mRNA was detectable in Kupffer cells. Hepatocytes, Kupffer cells, and sinusoidal endothelial cells underwent apoptosis after a few days of culture as reported previously (44,45). In contrast, HSCs increased in number and morphologically transdifferentiated into myofibroblasts after 4 days of culture. The expression of both galectin-1 and galectin-3 mRNAs was increased in HSCs in a time-dependent manner as revealed by RT-PCR and Northern blot analysis (Fig. 1B). Western blotting revealed that the protein levels of galectin-1 and galectin-3 increased in a time course similar to that for the expression of smooth muscle ␣-actin and PDGF receptor-␤, well characterized markers for activated HSCs. In accordance with these observations on in vitro activated HSCs, in vivo activated HSCs isolated from TAA-induced fibrotic livers expressed mRNAs and proteins for galectin-1 and galectin-3, both of which were undetectable in quiescent HSCs isolated from normal livers (Fig. 1C). Immunohistochemistry showed that galectin-1 was negligible in intact liver tissues (Fig. 1D, panel a), whereas in the septa of TAA-induced fibrotic livers, where activated HSCs were localized, it was prevalent (panel b). Galectin-3 immunoreactivity, which was sporadically positive in Kupffer cells in normal livers (panel c), was augmented around periportal areas and septa in fibrotic livers (panel d). These observations support the results obtained with HSCs and Kupffer cells isolated from normal and fibrotic livers described above. RT-PCR showed that the mRNA levels of both galectin-1 and galectin-3 were up-regulated in two models of liver fibrosis, the model obtained by TAA administration and the model obtained by common bile duct ligation (Fig. 1E). These results conclusively demonstrate that the expression of galectin-1 and galectin-3 is augmented in activated HSCs.
Effects of Galectin-1 and Galectin-3 on HSC Proliferation-To clarify the role of galectin-1 and galectin-3 in the process of liver fibrosis, we examined their effects on the proliferation of HSCs and hepatocytes. As depicted in Fig. 2A, both galectin-1 and galectin-3 were found to stimulate DNA synthesis in HSCs in a dose-dependent manner, with the stimulation by the former higher than that by the latter. Their mitogenic activity at 10 g/ml was less than that at 20 ng/ml PDGF-BB, the most potent mitogen for HSCs (4,5). Neither galectin-1 nor galectin-3 affected DNA synthesis in hepatocytes (Fig. 2B). Additionally, the cell number of HSCs treated with either galectin-1 or galectin-3 (10 g/ml) significantly increased at 2 or 4 days after stimulation compared with that of unstimulated control cells (Fig. 2C, panel a). In fact, microscopic observation showed that the cell density of HSCs increased at 4 days after stimulation with either galectin-1 or galectin-3 (10 g/ml) compared with the unstimulated control culture of HSCs (panels b-e). The mitogenic activity of both galectins was again less than that of PDGF-BB (20 ng/ml).
Intracellular Signaling Pathways in Galectin-stimulated HSCs-To reveal the molecular mechanism underlying the mitogenic activity of galectin-1 and galectin-3 for HSCs, we investigated their effects on the activity of MAPK cascades in HSCs. Both galectin-1 and galectin-3 induced the phosphorylation of ERK1/2 in a time-and dose-dependent manner (Fig.  3A, panel a). No such activation was observed with p38 MAPK, SAPK/JNK, and Akt (Fig. 3B). ERK1/2 was phosphorylated rather gradually, with a maximum intensity at 120 min of galectin (10 g/ml) treatment. It was interesting that this time course of ERK1/2 phosphorylation was sharply different from that of PDGF-BB-induced phosphorylation, where the initial and maximum level of ERK1/2 phosphorylation was observed as early as 6 min after treatment. Phosphorylation of MEK1/2, an upstream signal of ERK1/2, also occurred later in galectinstimulated HSCs than in PDGF-BB-stimulated HSCs (Fig. 3A,  panel b).
Thus, it appears that galectin-1 and galectin-3 stimulate HSCs through the phosphorylation of MEK1/2 and ERK1/2. In fact, galectin-induced DNA synthesis in HSCs was abolished by PD98059 or U0126 in a dose-dependent manner (Fig. 4C).
Effects of Inhibitors of PKC and PKA on ERK1/2 Phosphorylation in Galectin-stimulated HSCs-To obtain further in- Closed and open arrowheads indicate positive staining for galectin-1 and galectin-3, respectively. PV, portal vein. Magnification is ϫ200. E, expression of galectin-1 and galectin-3 mRNAs in normal and fibrotic liver tissues. Liver fibrosis was induced either by intraperitoneal TAA administration for 8 weeks (wk) or by ligation of common bile duct for 2 weeks. RT-PCR was performed on total RNA isolated from these tissues.
sight into the mechanism of the galectin-induced activation of MEK1/2 and ERK1/2 in HSCs, the phosphorylation of ERK1/2 in galectin-stimulated HSCs was investigated in the presence of GF109203X, a specific PKC inhibitor, and H-89, a specific PKA inhibitor. The PDGF-BB-induced phosphorylation of ERK1/2 was suppressed by GF109203X, but not by H-89 (Fig.  5A), implying that this activation was PKC (but not PKA)-dependent, as previously reported (4). Interestingly, the phosphorylation of ERK1/2 induced by galectin-1 was not affected by GF109203X or by H-89, whereas that induced by galectin-3 was. These results indicate that galectin-1 and galectin-3 activate different signaling pathways leading to the phosphorylation of ERK1/2.
Inhibition of ERK1/2 Phosphorylation and DNA Synthesis by Thiodigalactoside-Thiodigalactoside (TDG) is one of the potent hapten inhibitors of ␤-galactoside binding (46,47). Preliminary studies were performed to ensure that TDG has no stimulatory effect on both ERK1/2 activation and DNA synthesis in HSCs. Fig. 6A shows the results of such experiments. In fact, TDG at 10 and 20 mM did not affect either process at all, whereas PDGF-BB at 20 ng/ml stimulated them as expected. Next, galectin-1 and galectin-3 (10 g/ml each) were pretreated with TDG at 20 mM, and HSCs were successively challenged with the pretreated galectin-1 or galectin-3 in the presence of excess TDG and examined for the phosphorylation of ERK1/2 (Fig. 6B). TDG significantly decreased the potential of galectin-1 or galectin-3 to activate the phosphorylation of ERK1/2 in HSCs. TDG (20 mM) also significantly decreased the potential of both galectins to stimulate DNA synthesis in HSCs (Fig. 6C). These results indicate that both galectin-1 and galectin-3 HSCs were then treated with 10 g/ml r-galectin-1 or r-galectin-3 for the indicated time periods or stimulated with either stimulus at the indicated doses for 120 min. Stimulation with r-PDGF-BB (20 ng/ml) was used as a positive control. Panel b, detection of total and phospho-MEK1/2. HSCs were cultured for the indicated time periods in the presence of 20 ng/ml r-PDGF-BB or 10 g/ml r-galectin-1 or r-galectin-3. B: detection of total and phospho-p38 MAPK, total and phospho-SAPK/JNK, and total and phospho-Akt in HSCs. HSCs were cultured for 72 h in FBS/DMEM and maintained for 24 h in serum-free DMEM. The cells were then treated with 10 g/ml r-galectin-1 or r-galectin-3 for the indicated time periods. Note that p38 MAPK, SAPK/JNK, and Akt were not activated by these galectins. might modulate ERK1/2 activation and DNA synthesis in HSCs through their ␤-galactoside-binding potential.
Effect of Galectins on HSC Migration-HSCs are mobilized when the liver is injured, migrate out of their residence to the site of necrosis, and accumulate there (48,49). Here, we further tested the effects of galectins on the migratory activity of HSCs using cell culture insert (43). PDGF-BB at 20 ng/ml significantly augmented the migration of HSCs (Fig. 7). Galectin-1 (but not galectin-3) at 10 g/ml also enhanced their migratory activity.

DISCUSSION
Galectin-1 has been reported to promote the proliferation of rat pulmonary arterial endothelial cells and vascular smooth muscle cells via binding between the ␤-galactoside glycoconjugates of the lectin and the ␤-galactoside-containing moiety of their cell-surface receptors (15,50). There have been reports that galectin-3 can also modulate the growth of fibroblasts by interacting with their surface glycoconjugates (17). Our present study revealed that both galectin-1 and galectin-3 promote the proliferation of HSCs. Galectin-1 was found to be more potent as a mitogen for HSCs than galectin-3. Galectin-1 forms a homodimer of 14-kDa subunits, and each subunit has a single carbohydrate-binding site (51). On the other hand, galectin-3 is a monomer of 32-kDa subunits with a single carbohydratebinding site (52). Galectin-1 has an extensive affinity for complex-type N-glycans. In contrast, galectin-3 has a high affinity for repeating lactosamine units (53). Herein, we speculated that the counter-receptor of galectins on the HSC surface might derive from glycoproteins such as N-glycans. Although the precise molecular mechanism is not known, the greater sensitivity of HSCs to stimulation by galectin-1 compared with galectin-3 may be due to the differences in structure, ligands, and means of clustering between the two types of galectins.
In general, cells regulate their susceptibility to galectins by regulating the expression of glycoprotein counter-receptors that bear the oligosaccharide ligands for galectins (54). Previous reports revealed that counter-receptors for galectin-1 could be the T cell glycoproteins CD43 and CD45 (24), lysosomeassociated membrane glycoprotein-1 and -2 (55,56), fibronectin (57), laminin (58), and ␤-lactosamine-containing glycolipid (59). Previous studies listed the following as counter-receptors for galectin-3: lysosome-associated membrane glycoprotein-1 FIG. 4. Effects of MEK1/2 inhibitors on ERK1/2 phosphorylation and DNA synthesis in galectin-stimulated HSCs. A, effects of PD98059 and U0126 on the appearance of phospho-ERK1/2. HSCs were incubated with r-galectin-1 and r-galectin-3 (10 g/ml each) and with either of the inhibitors at the indicated doses for 2 h. The cells were subjected to Western blotting for detection of total phospho-ERK1/2. Note that both PD98059 and U0126 dose-dependently hampered the galectin-induced appearance of phospho-ERK1/2. p44 and p42 indicate molecular masses. B, effects of SB203580 and LY294002 on the appearance of phospho-ERK1/2 in galectin-stimulated HSCs. Experiments similar to those described for A were performed using SB203580 and LY294002, both at 10 M. Note that neither of these agents affected the phosphorylation of ERK1/2. C, effects of PD98059 and U0126 on galectin-induced DNA synthesis in HSCs. HSCs were treated with either r-galectin-1 or r-galectin-3 (10 g/ml) and with either PD98059 or U0126 at the indicated doses for 24 h. The cells were pulse-labeled with HSCs were preincubated for 30 min with or without these inhibitors at the indicated doses and stimulated with r-PDGF-BB (20 ng/ml) for 6 min. Note that the PDGF-BB-dependent ERK1/2 activation was PKC (but not PKA)-dependent in HSCs, as reported previously (4). B, effects of GF109203X and H-89 on the activation of ERK1/2 in galectin-1stimulated HSCs. Experiments similar to those described for A were performed for r-galectin-1-stimulated HSCs. HSCs were incubated for 2 h with 10 g/ml r-galectin-1 in the presence or absence of either GF109203X or H-89 at the indicated doses. Note that galectin-1-dependent ERK1/2 activation was unaffected by these inhibitors. C, effects of GF109203X and H-89 on the activation of ERK1/2 in galectin-3-stimulated HSCs. Experiments similar to those described for B were performed for r-galectin-3-stimulated HSCs. Note that galectin-3-dependent ERK1/2 activation was both PKC-and PKA-dependent. and -2 (60, 61), immunoglobulin ⑀ constant region (62), Fc receptors (62), laminin (61), carcinoembryonic antigen (61), Mac-2-binding protein (60,61), and lipopolysaccharide (63). Highly specific receptors for galectin-1 or galectin-3 have not been discovered to date. Our Western blot analysis revealed that neither CD43 nor CD45 was expressed in quiescent and activated HSCs (data not shown). Although we have not yet identified the counter-receptors for galectin-1 and galectin-3 in cultured HSCs, we speculate that the ␤-galactoside-containing molecules on the HSC surface might play a role in regulating the initiation of signal transduction when galectin-1 and galectin-3 specifically encounter oligosaccharide ligands.
Generally, ERK1/2 phosphorylation initiated by PDGF or epidermal growth factor takes place within minutes (4,67). In contrast, our present study showed that both galectin-1 and galectin-3 induced ERK1/2 phosphorylation with a slight increase at 30 min and a maximum response at 120 min. Such a rather slow occurrence of ERK1/2 activation has been similarly observed in integrin-dependent signaling pathways triggered by ECMs (68). Activation of Eph family receptors by their cell-surface ligands was also reported to require at least 1 h for maximum receptor phosphorylation (69). In addition, tyrosine phosphorylation of discoidin domain receptor-2 caused by type I collagen was reported to take place at 30 min after stimulation and to last for 120 min (70,71), showing a time course very similar to that for our present observations. Thus, the time course of signal activation seems to be varied among the combination of ligands and their receptors, although the precise molecular mechanism for this variation is not known.
In addition to the ERK1/2 pathway, the PI3K-Akt pathway is a key signaling pathway involved in the growth factor-induced proliferation of HSCs (4,5). In this study, however, we failed to detect the phosphorylation of Akt in HSCs stimulated with galectins. Furthermore, tyrosine phosphorylation of PI3K was hardly detected by immunoprecipitation analysis (data not shown). These facts indicate that galectin-triggered HSC proliferation is predominantly through a MEK1/2-ERK1/2 pathway, although the contribution of the PI3K-Akt pathway can never be disregarded.
Matricellular proteins are involved in regulating the migration and proliferation of cells (72). Galectin-1 was previously reported to be involved in the migration of myoblasts and astrocytes (28,73). In the present in vitro study, we found that galectin-1, but not galectin-3, significantly promoted the migration of HSCs. This activity of galectin-1 might be mediated by the interaction of galectin-1 with ␣ 1 ␤ 1 integrin and ECM proteins such as laminin (74), resulting in the disruption of the function of other adhesion molecules (30) and thus the prevention of the cell adhesion.
In summary, our study has shown that the expression of galectin-1 and galectin-3 is induced in activated HSCs and FIG. 6. Attenuation of galectin-1-and galectin-3-dependent ERK1/2 activation and DNA synthesis by TDG. A, effects of TDG on ERK1/2 phosphorylation and DNA synthesis in unstimulated HSCs. HSCs were cultured in the presence of TDG at the indicated concentrations for 2 h or at 20 mM for the indicated time periods, and their ERK1/2 phosphorylation (panel a) and DNA synthesis (panel b) were investigated. Stimulation with r-PDGF-BB at 20 ng/ml was used as a positive control. The results clearly show that TDG alone had no effects on ERK1/2 phosphorylation and DNA synthesis in HSCs. B, effects of TDG on ERK1/2 phosphorylation in galectin-stimulated HSCs. HSCs were cultured for 72 h in FBS/DMEM and maintained for 24 h in serum-free DMEM. A mixture containing 10 g/ml r-galectin-1 or r-galectin-3 and 20 mM TDG was preincubated for a few minutes and introduced into the above culture of HSCs. The cultures were maintained for an additional 2 h. Note that preincubation with TDG remarkably attenuated the appearance of phospho-ERK1/2 in galectin-stimulated HSCs. C, effects of TDG on DNA synthesis in galectin-stimulated HSCs. Experiments on DNA synthesis were performed as described for fibrotic livers. These galectins can cross-link with ␤-galactoside-containing glycoconjugates on the cell surface, mainly initiating the MEK1/2-ERK1/2 signaling pathway. This leads to the induction of the proliferation of HSCs. Thus, the induction of galectin-1 and galectin-3 in activated HSCs seems to play an important role in the development of liver fibrosis via this autocrine loop.