Expression of Interleukin-10 by in Vitro and in Vivo Activated Hepatic Stellate Cells*

Activated hepatic stellate cells (HSC) participate in matrix remodeling and deposition in liver fibrosis. The present study demonstrates that interleukin (IL)-10 is expressed by HSC upon activation in vitro or in vivo and that autocrine effects of this cytokine include inhibition of collagen production. Culture activation of HSC caused a distinct increase in IL-10 mRNA level compared with freshly isolated quiescent HSC. Treatment of cultured HSC with tumor necrosis factor-α, transforming growth factor-β, or lipopolysaccharide further increased IL-10 mRNA by 2-fold and resulted in the release of IL-10 protein into the medium. HSC isolated from rats after bile duct ligation (BDL) showed prominent increases in IL-10 mRNA (× 100) and protein (× 30) levels at 7 days after BDL, but such induction disappeared in advanced liver fibrosis (19 days after BDL). IL-10 expression correlated positively with mRNA expression of interstitial collagenase and inversely with that of α1(I) collagen. Addition of anti-IL-10 IgG to cultured HSC caused enhanced collagen production under a basal or stimulated condition with TGF-β, tumor necrosis factor-α, or lipopolysaccharide. These effects were associated with increased α1(I) collagen mRNA and reciprocally reduced collagenase mRNA levels. Co-transfection of HSC with an IL-10 expression vector and collagen reporter genes showed a 40% inhibition of α1(I) collagen promoter activity. These results demonstrate that activation of HSC causes enhanced autocrine expression of IL-10 which possesses a negative autoregulatory effect on HSC collagen production mediated at least in part by α1(I) collagen transcriptional inhibition and stimulation of collagenase expression. These findings, along with the demonstrated early induction of HSC IL-10 expression and its late disappearance during biliary liver fibrosis, suggest itsin vivo role in matrix remodeling and a possibility that failure for HSC to sustain IL-10 expression underlies pathologic progression to liver cirrhosis.

Hepatic stellate cells (HSC) 1 are vitamin A-storing perisinu-soidal cells in the liver. These cells participate in matrix remodeling and wound healing of the liver via their myofibroblastic activation (see Ref. 1 for review). Several plausible mechanisms have been proposed which underlie HSC activation (1). One such mechanism involves soluble factors such as cytokines and inflammatory mediators, which seem to induce different aspects of the cellular activation. Fox example, platelet-derived growth factor, IL-1, TNF-␣, TGF-␣ are all mitogenic to HSC (2,3), while TGF-␤ is a potent fibrogenic cytokine that not only induces expression of matrix genes (3)(4)(5) and tissue inhibitors of metalloprotease (TIMP) (6), but may also confer HSC a myofibroblastic phenotype by up-regulating ␣-smooth muscle actin expression (7,8). These soluble factors are released by effector cells such as hepatic macrophages (9 -11), endothelial cells (12), hepatocytes (13), or platelets (14) to establish a paracrine mode of action or produced by HSC to achieve autocrine effects (15). It also seems important to recognize that one of the primary activities of many of these cytokines resides in their modulation of inflammation and immune responses. As integral part of wound repair processes, monocytes, fibroblasts, and myofibroblasts are recruited to the injury site by platelet-derived growth factor (16) and TGF-␤ (17). HSC are shown to express MCP-1 (18), which chemoattracts monocytes; M-CSF (19), which induces proliferation and differentiation of monocytes; PAF (20) and CINC (21), which recruit neutrophils; and ICAM (22), which causes adhesion and transmigration of neutrophils. Thus, HSC may also actively participate in regulation of inflammation in the liver.
IL-10 is a cross-regulatory cytokine produced by Th2 cells, macrophages, mast cells, and B cells. It mediates several key functions of multiple cell types. IL-10 inhibits functions of Th1 cells and their expression of IL-2 and ␥-interferon (23), suppresses macrophages, including antigen presentation to Th1 cells, cytokine production, and cytotoxic activities (24,25). In contrast, IL-10 stimulates mast cells (26) and B cells (27). In addition, IL-10 has been shown recently to down-regulate type I collagen gene expression and to increase matrix metalloprotease-1 (interstitial collagenase) and matrix metalloprotease-3 (stromelysin-1) (MMP-1 and -3) expression in cultured skin fibroblasts, suggesting a role of IL-10 in the breakdown and remodeling of the extracellular matrix (28). In contrast, exogenous IL-10 inhibits synthesis of MMP-9 (92-kDa gelatinase) and blocks LPS-stimulated MMP-1 expression by human macrophages while it stimulates their TIMP-1 production (29). Thus, these findings suggest fibrogenic effects of IL-10 on macrophages, which seem to oppose the aforementioned effects on fibroblasts.
In this report, we demonstrate for the first time, induced expression of IL-10 by rat HSC upon activation in vitro by culturing on plastic dish and in vivo by cholestatic liver injury. IL-10 expression by HSC is up-regulated by TNF-␣ and TGF-␤1 in vitro and induced conspicuously during the early phase of cholestatic injury followed by a disappearance of the induction at the late fibrogenic phase. In vitro neutralization experiments demonstrate autocrine stimulation of interstitial collagenase expression and inhibition of ␣1(I) collagen expression by IL-10 in HSC, suggesting its role in initiation of matrix remodeling.

MATERIALS AND METHODS
Cholestatic Liver Injury-Cholestatic liver injury was induced in male Wistar rats weighing 500 -600 g by aseptic ligation and transection of the common bile duct (BDL) as described previously (30). Another group of the rats was sham-operated to serve as controls (Sham). The animal protocol described in this study was approved by the Institutional Care and Use Committee of the University of Southern California.
HSC Isolation and Culture-HSC were isolated from normal male Wistar rats, BDL, and Sham animals by in situ digestion of the liver and arabinogalactan gradient ultracentrifugation as reported previously (31,32). The purity and the viability of the cells from all animals exceeded 98 and 97%, respectively. The cells from normal rats were cultured in RPMI with 10% fetal calf serum in 24-well plates for 5-6 days after isolation. For experiments testing effects of TNF-␣ (0.1-10 ng/ml), TGF-␤ (0.1-10 ng/ml), LPS (1-100 g/ml), and anti-IL-10 IgG (20 g/ml), the cells were washed with phosphate-buffered saline twice and incubated with serum-free RPMI and test substances for 42 h. TNF-␣, TGF-␤, and goat anti-mouse IL-10 IgG were purchased from R & D System (Minneapolis, MN), and LPS and non-immune goat IgG were from Sigma.
RNA Extraction, RT-PCR-Total RNA was extracted from freshly isolated and cultured cells by a method of Chomczynski and Sacchi (33). For RT-PCR, total RNA was reverse-transcribed using 600 units of Moloney murine leukemia virus reverse transcriptase and oligo(dT) at 37°C for 60 min. The synthesized cDNA for IL-10, interstitial collagenase, ␣1(I) collagen, and ␤-actin were amplified using specific sets of primers designed from published cDNA sequences (34 -36) as follow: IL-10, CTGGCTCAGCACTGCTAT and ATTCATGGCCTTGTAGACAC; rat interstitial collagenase, CGAACACTCAAATGGTCCCA and TCCA-CATGGTTGGGAAGTTC; collagen ␣1(I), ACAGCACGCTTGTGGAT and GTCTTCAAGCAAGAGGACCA; ␤-actin, GAGCTATGAGCTGCCT-GACG and AGCACTTGCGGTCCACGATG. A competitive template containing the specific primer sequences for IL-10 was constructed using a PCR MIMIC™ (CLONTECH). For competitive PCR, sample cDNA was amplified in the presence of the increasing amount of the competitor (5 ϫ 10 Ϫ21 to 1.6 ϫ 10 Ϫ19 mol). The mRNA quantity was assessed by calculating the amount of the competitor added to achieve an equimolar amount of the PCR products of the competitor and IL-10.
IL-10 cDNA Fragment Subcloning and Sequencing-The 475-bp IL-10 cDNA fragment generated by PCR was isolated from 1.5% agarose gel with an elution buffer (0.5 M ammonium acetate, 10 mM magnesium acetate, 1 mM EDTA, 0.1% SDS, pH 8.0). The eluted cDNA was ligated into a PCR™ vector from a TA cloning kit (Invitrogen Corp., San Diego, CA). After a large scale plasmid preparation, the sequence of the cDNA was determined by the chain termination method (U. S. Biochemical Corp.). A partial EcoRI fragment (326 bp) of the cDNA insert was purified from agarose gel as described above and used as a probe for Northern blot hybridization.
Northern Blot Analysis-For Northern blot analysis for IL-10, 10 g of total RNA was electrophoresed in 1% agarose gel containing formaldehyde and transferred to nylon filter (Micron Separations, Westboro, MA) as described (31). The 326-bp IL-10 cDNA fragment was labeled with [␣-32 P]dCTP using a random primer labeling kit (Life Technologies, Inc.). Prehybridization, hybridization, washing, and autoradiography were performed as described previously (31). Equivalent RNA loading was confirmed by rehybridization of the filter for 18 S rRNA.
Enzyme-linked Immunosorbent Assay and Western Blot Analysis-For determination of the IL-10 concentration in the HSC culture medium, the media from three wells were combined, dialyzed, lyophilized, and analyzed for IL-10 using a mouse IL-10 enzyme-linked immunosorbent assay kit (Quantikine™, R & D Systems). To detect IL-10 in HSC, Western blot analysis was performed. Freshly isolated HSC were sub-jected to protein extraction using a 2 ϫ lysis buffer (100 mM Tris-HCl, pH 6.8, 4% SDS, 20% glycerol, and 2% ␤-mercaptoethanol). Samples (100 g of protein per each sample) were separated by 8% polyacrylamide gel electrophoresis using reducing conditions and transferred to nitrocellulose filters (Bio-Rad). The filters were first treated with 10% non-fat milk in 20 mM Tris base, pH 7.6, 137 mM NaCl, and 0.1% Tween 20 (TBST) and incubated with goat polyclonal anti-mouse IL-10 antibodies (R & D Systems) at 1:300 dilution in TBST with 1% bovine serum albumin, followed by incubation with rabbit anti-goat IgG antibodies (1:2000) conjugated with horseradish peroxidase. The immobilized IL-10-antibody complex was detected by chemiluminescence using ECL kit (Amersham Corp.).
Collagen Synthesis Assay-To examine whether neutralization of IL-10 produced by HSC affects collagen production, the cultured cells were incubated for 48 h with serum-free RPMI with ascorbic acid (50 g/ml) and ␤-aminopropionitrile fumarate (50 g/ml) in the presence of anti-IL-10 IgG or nonimmune IgG (20 g/ml) and incubated with [2,3,4,5-3 H]proline (10 Ci/ml) for the last 18 h to radiolabel newly synthesized collagen (3). The cell and media proteins were precipitated with 10% trichloroacetic acid. Collagen production was determined by a collagenase digestion method (37).
Transfection and Reporter Gene Assays-To examine whether IL-10 has any effects on collagen gene promoter activity, cultured HSC after the first passage were transiently co-transfected with a collagen reporter gene plasmid (pGLCO 2 or pGLCO3) (38) and a murine sense or antisense IL-10 expression vector (39), which was kindly provided by Dr. Lili Feng at the Scripps Research Institute. The collagen reporter gene, pGLCO 2 , contains 2200 bp of the murine ␣1(I) collagen gene 5Ј-flanking region linked to the luciferase reporter gene, while pGL-COL3 contains 220 bp of ␣1(I) collagen gene 5Ј-flanking region. Liposomes were prepared using 3 g of a reporter gene plasmid and 5 g of the expression vector along with 32 l of LipofectAMINE reagent (Life Technologies, Inc.). The cells were incubated with the liposomes for 8 h, the liposome mixture removed, and fresh medium containing 10% fetal calf serum added. The cells were incubated for additional 36 h, washed with phosphate-buffered saline, and lysed for luciferase assay as described previously (38).
Statistical Analysis-All data are expressed as means Ϯ S.E. The significance for the difference between the groups was assessed using standard t test. For quantification of competitive RT-PCR data, a linear regression analysis was performed between the concentration of the competitive template used in the assay and the ratio of densitometric results for the competitor product over than of the target gene product.

RESULTS
IL-10 RT-PCR for Cultured HSC-RT-PCR analysis of RNA from freshly isolated HSC from normal rats showed no detectable product for IL-10 using 35 cycles of amplification (first lane, Fig. 1 (upper panel)). However, HSC cultured on plastic wells for 7 days showed an increase in IL-10 mRNA as indicated by a detectable PCR product (second lane, Fig. 1 (upper  panel)). Furthermore, incubation of HSC with TNF-␣ or TGF-␤, cytokines known to stimulate HSC (3), caused further increases in IL-10 mRNA (lane 3-8, Fig. 1 (upper panel)). Semiquantitative analyses of the RT-PCR results were performed by scanning densitometry of the IL-10 PCR product and standardization with ␤-actin results (Fig. 1, lower panel). These analyses show 2-fold increases in IL-10 mRNA expression by TGF-␤ (10 ng/ml) or TNF-␣ (10 ng/ml) and a 50% increase by LPS (10 g/ml).
IL-10 Release by Cultured HSC-To assess the levels of IL-10 released by cultured HSC following treatment with several agonists, and enzyme-linked immunosorbent assay was performed on the medium samples (Table I). No detectable IL-10 was measured in the medium from untreated HSC. However, TNF-␣, TGF-␤, and LPS all stimulated IL-10 release by HSC.
HSC IL-10 mRNA Expression in Cholestatic Liver Injury-To determine whether IL-10 is expressed by HSC during the course of cholestatic liver injury, HSC were isolated from rats at 2, 7, and 19 days after bile duct ligation or sham operation, and RT-PCR was performed on HSC RNA samples for detection of IL-10 mRNA (Fig. 2). Sham HSC show undetectable or minimal IL-10 mRNA levels at each time point. In contrast, HSC from BDL show a distinct increase in IL-10 mRNA at 2 days, which was further accentuated at 7 days. Interestingly, this induction of IL-10 mRNA expression was completely abrogated at 19 days. To quantitatively assess the induction at 7 days, competitive PCR was performed using a specifically constructed competitive template. As shown in the upper panel of Fig. 3A, addition of an increasing amount of the competitor (from right to left) resulted in a progressive reduction in the level of IL-10 PCR product from the Sham sample while reciprocally raising the level of the competitor product. For the BDL samples, which are expected to have much lower levels, 40 cycles of amplification was used. Even though the level of IL-10 product was still lower, the similarly effective competition by the competitor was shown but with the amounts of the competitor that were approximately 2 orders of magnitude lower than those used for Sham. Linear regression analysis was performed for three pairs of competitive PCR data (Fig. 3B). As predicted, the level of IL-10 mRNA in 7-day BDL HSC was 100-fold higher than that in Sham HSC.
Subcloning and Sequencing of IL-10 PCR Product-To verify that the PCR product detected was truly a IL-10 cDNA fragment, we subcloned the product into a TA vector and sequenced a partial EcoRI fragment (326 bp) following a large scale plasmid preparation and cDNA purification. The sequence of the fragment showed a perfect match with the published nucleotide sequence of rat IL-10 cDNA (34), demonstrating that our PCR indeed detected IL-10 mRNA in HSC. Furthermore, we have utilized the purified 326-bp IL-10 cDNA fragment as a probe to perform Northern blot analysis on HSC RNA samples. Northern blot analysis clearly confirmed prominent induction of IL-10 mRNA expression in HSC from 7-day BDL as compared with corresponding Sham (Fig. 4).
Detection of IL-10 Protein in Activated HSC-Western blot analysis was performed to examine whether IL-10 protein level is coordinately increased in HSC from 7-day BDL (Fig. 5). The analysis detected an immunoreactive band with a distinct increased intensity in BDL (last two lanes), which corresponded to the molecular size (17 kDa) of authentic recombinant mouse IL-10 standard (first lane). As an internal control, desmin was immunoblotted using the same samples, which showed the relatively similar immunoreactivity between the two groups of the samples (bottom panel).
Relationship of IL-10 Expression to Collagen or Collagenase Expression in BDL-We were very intrigued by the time-dependent induction of IL-10 in HSC at 7 days in BDL animals. Since recent studies suggested regulation of collagen and collagenase genes by IL-10 in other cell types (28,29), we examined ␣1(I) collagen and interstitial collagenase mRNA expression in the same HSC RNA samples used for IL-10 RT-PCR analysis (Fig. 2, lower two panels). Induction of interstitial collagenase mRNA expression was shown to coincide with that of IL-10 at 7 days as was the disappearance of induction of both genes at 19 days. On the contrary, marked induction of ␣1(I) collagen expression occurred when IL-10 expression ceased at 19 days. Freshly isolated HSC show very low levels of IL-10 (first lane), whereas culture-activated HSC grown in plastic dish for 6 days expressed appreciably more IL-10 mRNA (second lane). Treatment of cultured HSC with TNF-␣ and TGF-␤ further increased IL-10 mRNA levels (third to eighth lanes). Lower panel, densitometric analysis of IL-10 mRNA expression by cultured HSC. Densitometric RT-PCR data for IL-10 mRNA were standardized with ␤-actin signals and statistically compared between different treatment groups. Treatment of cultured HSC with TNF-␣ (10 ng/ml) and TGF-␤ (10 ng/ml) resulted in significant 2-fold increases in IL-10 mRNA expression, while LPS (10 g/ml) caused a 40% increase. *, p Ͻ 0.05 as compared with control.  2. IL-10 mRNA expression in HSC isolated from rats with cholestatic liver injury. IL-10 RT-PCR analysis was performed with HSC isolated from rats at 2 days, 7 days, and 19 days after ligation of bile duct (BDL) (B) or sham operation (S). IL-10 mRNA expression in HSC is slightly and prominently enhanced in BDL animals at 2 and 7 days, respectively, while such induction disappeared at 19 days. RT-PCR also demonstrates time-dependent induction of interstitial collagenase mRNA expression by HSC at 7 days after BDL, which coincides with IL-10 mRNA expression. ␣1(I) collagen mRNA levels were mildly increased in HSC from BDL animals at 7 days, followed by its more accentuated increase at 19 days when IL-10 expression ceased.

IL-10 Neutralization Enhances HSC Collagen Production-
The above results suggested a possible link between IL-10 expression by activated HSC and matrix homeostasis. To examine this possibility, collagen production was assessed by incorporation of [ 3 H]proline with cultured HSC exposed to FIG. 6. IL-10 neutralization stimulates HSC collagen production. Cultured HSC were incubated with anti-IL-10 IgG (20 g/ml) or nonimmune IgG (20 g/ml) in the absence or presence of TGF-␤ (10 ng/ml), TNF-␣ (10 ng/ml), or LPS (10 g/ml). Collagen production by these cells was determined by incorporation of [ 3 H]proline into collagenase sensitive peptides as described under "Materials and Methods." Addition of IL-10 antibodies alone resulted in a 50% increase in basal collagen production (Control) by HSC. TGF-␤-mediated up-regulation of collagen production was enhanced by 2-fold by IL-10 neutralization. TNF-␣ or LPS, which did not stimulate HSC collagen production by itself, caused significant enhancement of collagen production if added together with anti-IL-10 IgG. *, p Ͻ 0.05.

FIG. 3.
A, representative data for IL-10 competitive PCR for HSC from rats at 7 days after bile duct ligation (BDL) or sham operation (Sham). As shown in this example, addition of an IL-10 competitor concentration-dependently suppressed the level of IL-10 PCR product while it increased the level of the competitor product in both samples from Sham and BDL animals. However, the BDL sample required higher concentrations of the competitor to suppress generation of IL-10 product than the Sham sample, indicating the higher IL-10 mRNA level in the BDL sample. B, data from three sets of competitive PCR were plotted and a linear regression analysis was performed. Arrows indicate the estimated levels of IL-10 mRNA as determined by the concentration of the competitor, which achieved generation of the equal amount of IL-10 and competitor products (the ratio of IL-10 over competitor product ϭ 1). Note the estimated increase in IL-10 mRNA in BDL HSC is 100-fold. TGF-␤ (10 ng/ml), TNF-␣ (10 ng/ml), and LPS (10 g/ml) in the presence of anti-IL-10 IgG or nonimmune IgG (Fig. 6). The addition of anti-IL-10 IgG alone caused a 50% increase in basal collagen production. TGF-␤-mediated stimulation of collagen production was doubled by the addition of the antibodies. Even though TNF-␣ or LPS alone did not increase collagen production, concomitant IL-10 neutralization resulted in significant enhancements in collagen synthesis. These results clearly demonstrate an inhibitory autocrine effect of IL-10 on collagen synthesis by culture-activated HSC.
IL-10 Neutralization Affects ␣1(I) Collagen and Collagenase mRNA Expression-To investigate mechanisms underlying the observed inhibitory role of IL-10 in HSC collagen production, we have examined effects of IL-10 neutralization on mRNA expression of ␣1(I) collagen and interstitial collagenase by cultured HSC. Exposure of HSC to TNF-␣ (10 ng/ml) or LPS (10 g/ml) stimulated mRNA expression of collagenase (Fig. 7,  upper left panel). However, addition of anti-IL-10 IgG clearly suppressed these stimulatory effects (Fig. 7, upper right panel). TGF-␤ (10 ng/ml) stimulated ␣1(I) collagen mRNA expression in cultured HSC and LPS marginally showed the effect (Fig. 7,  upper panel). Addition of anti-IL-10 antibodies further promoted the increases in ␣1(I) collagen mRNA levels in TGF-␤-or LPS-stimulated HSC (Fig. 7, upper panel). Densitometric data from at least three sets of experiments were standardized with ␤-actin results and statistically compared between the differ-ent treatments (Fig. 7, lower panel). Addition of anti-IL-10 antibodies slightly, but significantly, reduced basal interstitial collagenase mRNA expression. Furthermore, it suppressed an increase in interstitial collagenase mRNA levels induced by TNF-␣ and LPS (Fig. 7, lower left panel). On the contrary, IL-10 neutralization significantly enhanced by 2-fold the increase in ␣1(I) collagen mRNA expression by HSC exposed to TGF-␤ and LPS (Fig. 7, lower right panel). These results suggested that IL-10 secreted by culture-activated HSC not only induces basal expression of collagenase but also mediates up-regulation of collagenase expression caused by TNF-␣ and LPS. At the same time, IL-10 expression induced by TGF-␤ and LPS appears to exert an inhibitory effect on ␣1(I) collagen expression.
IL-10 Suppresses COLL Promoter Activity-To examine whether the observed inhibitory effect of IL-10 on collagen expression is mediated via its effect at transcriptional level, transient co-transfections were performed using an IL-10 expression vector (sense or antisense) with an ␣1(I) collagen promoter-luciferase construct (pGLCO2 or pGLCO3). As compared with transfection with the antisense IL-10 expression vector, collagen promoter activity was inhibited 40% with the sense IL-10 expression vector (Fig. 8). This suggests that IL-10's negative autoregulatory effects on HSC collagen expression is mediated at least in part via its transcriptonal inhibition of collagen gene.

DISCUSSION
The present study is the first to demonstrate that HSC express IL-10 upon their activation in vivo and in vitro. HSC are considered as pericytes in the liver (40), which are also known to serve as principal cells to participate in liver fibrogenesis via their myofibroblastic activation (1). A recent report demonstrated expression of IL-10 by mesangial cells, the pericytes in the glomerulus, which are incriminated as the major source of fibroproliferative responses in glomerulonephritis (41). Since HSC and mesangial cells are considered analogous due to their similar functionality and pathophysiologic roles, we hypothesized that HSC may express IL-10. Our results demonstrate IL-10 is expressed by HSC upon activation in culture and during the early stage of biliary liver injury. Our RT-PCR specificity was verified by sequencing of the IL-10 PCR product, with which we further confirmed induced mRNA expression via Northern blot analysis. Western blot analysis of HSC protein extracts revealed a prominently expressed 17-kDa IL-10 protein at 7 days after BDL, and the cultured HSC were shown to express and release IL-10 in response to TNF-␣, TGF-␤, and LPS.
Our previous work showed enhanced expression of TNF-␣ by hepatic macrophages at 1 and 2 weeks after BDL but an almost complete disappearance of such induction at 3 weeks (42). Since our in vitro experiment demonstrates induction of IL-10 in HSC by TNF-␣, it may be assumed that macrophage-derived TNF-␣ in the liver might have induced IL-10 expression by HSC in the time-dependent manner in the BDL model. However, the concomitant induction (7 days) and repression (19 days) of macrophage TNF-␣ and HSC IL-10 expression in this in vivo model also suggest IL-10 derived from HSC may not function as an anti-inflammatory cytokine toward hepatic macrophages. This assumption led us to think of other biological significance that HSC-derived IL-10 may possesses in the liver. To this end, we were intrigued by recent studies that showed IL-10-mediated regulation of the genes involved in matrix remodeling and homeostasis such as MMP-1, MMP-3, MMP-9, TIMP-1, and ␣1(I) collagen in skin fibroblasts (28) and macrophages (29). Indeed, our culture study clearly demonstrates IL-10 released by HSC suppresses their collagen production and this effect is mediated at least in part by transcriptional inhibition of collagen gene and enhanced expression of interstitial collagenase. These findings suggest a negative autoregulatory role of IL-10 in HSC collagen production in matrix remodeling. In support of this view, our in vivo data reveals concomitant induction of IL-10 and interstitial collagenase in HSC during the early stage of cholestatic liver fibrosis (7 days after BDL) and the lack of HSC IL-10 expression in association with marked ␣1(I) collagen induction in advanced liver fibrosis at 19 days. This raises an intriguing secondary hypothesis that the failure of HSC to continue their expression of IL-10 may underlie progressive fibrogenesis leading to liver cirrhosis.
Interplay between soluble factors of paracrine and autocrine sources is complex in regulation of HSC biology. Among several cytokines implicated in activation of HSC, TGF-␤ is considered a potent fibrogenic cytokine that seems capable of conferring HSC most aspects of cellular activation (3)(4)(5)(6)(7)(8). This cytokine can be released by hepatic macrophages (9) or HSC by themselves (15), and the paracrine and autocrine interaction can be established via its ability to autoinduce its expression (15). In our in vitro study, TGF-␤ induced IL-10 in cultured HSC. However, IL-10 induction was abolished in HSC at 19 days after BDL despite HSC (43) and hepatic macrophages 2 continue to express TGF-␤1 at this time point in this model. Thus, unlike the close in vivo association between IL-10 and TNF-␣ expression in the BDL model, this dissociation of IL-10 and TGF-␤ expression suggests complex regulation of IL-10 expression by TGF-␤. Another discrepancy noted in the present study was the continued induction of IL-10 expression by cultureactivated HSC as compared with the time-dependent induction seen in vivo. This obviously suggests cellular or molecular differences in in vitro and in vivo activated HSC or reflects the absence of other in vivo factors that may cause the time-dependent expression in the culture system. It is attractive to speculate that these factors may be derived from other cell types in the liver including hepatic macrophages. Additional studies are obviously needed to test this hypothesis. FIG. 8. Effects of IL-10 expression on ␣1(I) collagen promoter activity. Cultured HSC were co-transfected with a IL-10 expression vector (sense or antisense) and an ␣1(I) collagen promoter-luciferase construct (pGLCOL2 or pGLCOL3) to examine effects of IL-10 expression of ␣1(I) collagen promoter activity. The transfection with the antisense vector served as a control. Note that the promoter activity of both pGLCOL2 and pGLCO3 was suppressed by 40% by the IL-10 sense transfection, indicating expression of IL-10 suppresses ␣1(I) collagen promoter activity.