Fibrogenesis in pediatric cholestatic liver disease: Role of taurocholate and hepatocyte-derived monocyte chemotaxis protein-1 in hepatic stellate cell recruitment


  • Potential conflict of interest: Nothing to report.


Cholestatic liver diseases, such as cystic fibrosis (CF) liver disease and biliary atresia, predominate as causes of childhood cirrhosis. Despite diverse etiologies, the stereotypic final pathway involves fibrogenesis where hepatic stellate cells (HSCs) are recruited, producing excess collagen which initiates biliary fibrosis. A possible molecular determinant of this recruitment, monocyte chemotaxis protein-1 (MCP-1), an HSC-responsive chemokine, was investigated in CF liver disease and biliary atresia. The bile-duct-ligated rat and in vitro coculture models of cholestatic liver injury were used to further explore the role of MCP-1 in HSC recruitment and proposed mechanism of induction via bile acids. In both CF liver disease and biliary atresia, elevated hepatic MCP-1 expression predominated in scar margin hepatocytes, closely associated with activated HSCs, and was also expressed in cholangiocytes. Serum MCP-1 was elevated during early fibrogenesis. Similar observations were made in bile-duct-ligated rat liver and serum. Hepatocytes isolated from cholestatic rats secreted increased MCP-1 which avidly recruited HSCs in coculture. This HSC chemotaxis was markedly inhibited in interventional studies using anti-MCP-1 neutralizing antibody. In CF liver disease, biliary MCP-1 was increased, positively correlating with levels of the hydrophobic bile acid, taurocholate. In cholestatic rats, increased MCP-1 positively correlated with taurocholate in serum and liver, and negatively correlated in bile. In normal human and rat hepatocytes, taurocholate induced MCP-1 expression. Conclusion: These observations support the hypothesis that up-regulation of hepatocyte-derived MCP-1, induced by bile acids, results in HSC recruitment in diverse causes of cholestatic liver injury, and is a key early event in liver fibrogenesis in these conditions. Therapies aimed at neutralizing MCP-1 or bile acids may help reduce fibro-obliterative liver injury in childhood cholestatic diseases. (HEPATOLOGY 2008.)

Cholestatic liver diseases such as cystic fibrosis liver disease (CFLD), a cause of intrahepatic cholestasis, and biliary atresia (BA), a cause of extrahepatic cholestasis, are common and important causes of cirrhosis in children.1–4 Disrupted bile flow via intrahepatic or extrahepatic biliary obstruction results in variably progressive liver fibrosis. In CFLD, this process is slow, causing significant morbidity from focal biliary cirrhosis in about 20% of patients by the second decade of life.3, 4 The apical localization of the cystic fibrosis transmembrance conductance regulator (CFTR) in cholangiocytes5 suggests that abnormal ion transport leads to cholestasis, with focal fibrosing destruction of intrahepatic bile ducts.6 The resulting pathognomonic pattern of injury is focal biliary cirrhosis, for which there is no known treatment and which may evolve into multilobular biliary cirrhosis and liver failure. In BA, a fibro-obliterative extrahepatic cholangiopathy causing cholestasis in newborns, there is a rapid progression to biliary cirrhosis over the first months of life.1–3 Unless a timely hepatoportoenterostomy succeeds in achieving bile drainage, liver failure ensues. BA is the most common cause of cirrhosis and the single most frequent indication for liver transplantation in children.

Despite the diverse origins and disease progression in CFLD, in BA, and indeed in other cholestatic diseases in children, progression to cirrhosis is the major determinant of outcome, and the process by which this occurs appears stereotypic. We previously demonstrated that activated hepatic stellate cells (HSCs) and periportal myofibroblasts are the principal collagen-producing cells causing cirrhosis in CFLD7 and BA.8 It is generally recognized that in normal liver, HSCs exist in a quiescent phenotype in a perisinusoidal location, closely associated with hepatocytes.9 In liver injury, HSCs are recruited and activated into “myofibroblast-like” cells which have a central role in the deposition of fibrillar collagens and fibrotic matrix and expression of proinflammatory and profibrogenic cytokines and associated receptors.9, 10 The mechanisms involved in HSC recruitment in vivo are unclear. Both cholangiocytes and hepatocytes may produce soluble factors capable of both activating HSCs and causing their recruitment,11 and we have demonstrated the expression of transforming growth factor-β1 in peri-scar hepatocytes in association with disease progression in CFLD7 and in BA.8 However, the search for earlier events in cholestasis which might stimulate recruitment of portal fibroblasts and perisinusoidal HSCs, thus initiating fibrogenesis, might reveal potential targets for therapy.11, 12 In particular, HSCs are known to respond to a variety of chemokines in vitro, with monocyte chemotaxis protein-1 (MCP-1) being one of the most potent.13, 14

In this study, we examined the role of MCP-1 in the early recruitment of HSCs and the genesis of hepatic fibrosis in clinical studies in children with CFLD and BA, and investigated the potential mechanisms associated with MCP-1 induction, using both an animal model of cholestatic liver injury, the bile-duct-ligated (BDL) rat, as well as in vitro cell culture models.


α-SMA, alpha-smooth muscle actin; AUC, area under curve; BA, biliary atresia; BDL, bile duct ligated; CFLD, cystic fibrosis liver disease; ELISA, enzyme-linked immunosorbent assay; ESI-MS/MS, electrospray ionization-tandem mass spectrometry; HSC, hepatic stellate cell; MCP-1, monocyte chemotaxis protein-1; ROC, receiver operating characteristic; TCA, taurocholic acid.

Patients and Methods

Human Studies

Patient Details.

With Institutional Review Board approval and informed consent, 30 children with CFLD (age, 10.7 ± 0.8) and 29 with BA (age, 2.3 ± 0.6) undergoing clinically indicated liver biopsy and endoscopy, had liver, serum, and bile (CFLD n = 19) stored for research. Age-matched and gender-matched pediatric controls for serum analysis comprised subjects attending for minor plastic procedures (n = 27; age, 8.9 ± 0.6); those for liver analysis comprised three subjects, aged 3.5–5.0 years with histologically normal liver, biopsied to exclude liver disease for extrahepatic portal hypertension; and those for bile analysis comprised subjects without liver disease endoscoped for gastrointestinal symptoms. CFLD was defined if two of three criteria were met in children with CF: (i) hepatomegaly with or without splenomegaly; (ii) persistent elevation of serum alanine aminotransferase (>1.5× upper limit of normal) >6 months; and (iii) ultrasound scan with abnormal echogenicity or nodular edge suggestive of cirrhosis. BA was confirmed by operative cholangiogram. Liver sections were frozen and/or formalin-fixed and paraffin-embedded.

Ribonuclease Protection Assay.

Frozen liver tissue was available from 8 patients with BA, 14 with CFLD, and 3 controls. Ribonuclease protection assay for MCP-1 messenger RNA (mRNA) expression was standardized to, and expressed as a percentage of the housekeeping gene GAPDH (glyceraldehyde 3-phosphate dehydrogenase). Liver histology was stratified according to the Scheuer stage of fibrosis.15 In CFLD, these were graded as fibrosis stage 0, n = 4, stage 1–2, n = 5, and stage 3–4, n = 5. All infants with BA had fibrosis stages 3–4.


Liver sections (n = 30 CFLD, n = 29 BA) were subjected to heat retrieval at 95°C for 20 minutes in citrate buffer, then incubated with polyclonal goat anti-MCP-1 antibody (1:8, AF-279-NA; R&D Systems, Minneapolis, MN) overnight at 4°C.16 Bound antibody was detected using a biotinylated rabbit anti-goat secondary antibody (Dako, Denmark), followed by streptavidin-peroxidase (StrepABC; Dako) as previously described.7, 8 Negative control used nonimmune anti-human antisera (Dako) in place of primary antibody. Immunohistochemistry was also performed for α-smooth muscle actin (α-SMA) and cytokeratin-7 (CK-7) on human liver sections to assess the role of HSCs and liver progenitor cells,17 respectively, in MCP-1 expression. The Vector VIP peroxidase substrate (SK4600, Vector Laboratories, Burlingame, CA) and Envision horseradish peroxidase–labeled anti-mouse polymer (K4001, Dako) kits were used with either a mouse monoclonal anti–α-SMA (1:400, clone 1A4; Sigma, St. Louis, MO) or an anti-CK-7 (1:100, OV-TL12/30; Dako) primary antibody.

In Situ Hybridization for MCP-1 mRNA.

For detection of MCP-1 mRNA, a 200 base-pair fragment of human MCP-1 complementary DNA was subcloned into pBluescript KSII vector. Digoxigenin-labeled riboprobes for sense (control) and antisense for MCP-1 mRNA were produced by in vitro transcription with SP6 and T7 polymerases (Boehringer-Mannheim, Germany), and in situ hybridization was performed as previously described.7, 8

Serum and Biliary MCP-1 and Taurocholic Acid Measurement.

Serum from 30 children with CFLD, 20 with BA, 27 controls, and bile from 19 patients with CFLD and 13 controls, were assayed for MCP-1 using a human MCP-1 enzyme-linked immunosorbent assay (ELISA) (Pierce Biotechnology, Rockford, IL). The hydrophobic bile acid Taurocholic acid (TCA) has been shown to induce MCP-1 protein release in vitro in an immortalized murine bile duct epithelial cell line,18 and is a bile acid of interest in cholestatic injury. TCA concentration was measured in bile and matching serum samples, following extraction, using electrospray ionization–tandem mass spectrometry (ESI-MS/MS), as described.19, 20 Serum and biliary MCP-1 levels were correlated with the Scheuer stage of hepatic fibrosis15 and TCA levels.

Animal Studies

Bile Duct Ligation in Rats.

The BDL rat model of cholestatic liver injury was utilized to further characterize the role of MCP-1 as a determinant of cholestasis-induced hepatic fibrogenesis. With institutional Animal Ethics Committee approval and compliance with Australian regulatory guidelines, male Sprague-Dawley rats aged 6–8 weeks, weighing 200–300 g, underwent bile duct ligation. Sham-operated rats served as controls. Rats were euthanized at 3–14 days and 6–18 weeks, with blood, bile, and liver stored at −70°C. Bile was aspirated from the “bile cyst” formed above the ligature in BDL rats. In sham controls, the common bile duct was cannulated and bile collected for 30 minutes.

Immunohistochemistry for MCP-1.

Immunohistochemistry for MCP-1 expression was performed as described earlier for human liver sections7, 8 using a rabbit polyclonal anti-MCP-1 primary antibody (1:100, AB1834P; Millipore, Billerica, MA) and Envision horseradish peroxidase–labeled anti-rabbit polymer kit (K4003, Dako).

Real-Time RT-PCR for MCP-1 mRNA Expression in BDL Rats.

RNA was extracted from BDL and sham livers and real-time reverse transcription polymerase chain reaction (RT-PCR) was performed as previously described7 with the following modifications. Primers for rat MCP-1 and the housekeeping gene β-actin (Table 1) were designed using Primer Express software (PerkinElmer, Foster City, CA) and were BLAST searched to avoid complementarity to other genes. Complementary DNA was diluted five-fold prior to amplification. MCP-1 mRNA expression was standardized to β-actin using the two standard curve method.

Table 1. Oligonucleotide Sequences Used in Real-Time RT-PCR Amplification for Gene Expression of Rat and Human MCP-1 and β-Actin
GeneForward PrimerReverse Primer
Rat MCP-1 (Amplicon length = 70 bp)cagatgcagttaatgccccacagccgactcattgggatcat
Human MCP-1 (Amplicon length = 89 bp)ccaagcagaagtgggttcagcttgggttgtggagtgagtg
Rat and human β-actin (Amplicon length = 77 bp)actatcggcaatgagcggttcatgccacaggattccataccc

Serum and Biliary MCP-1 and TCA Measurement.

MCP-1 and TCA levels were determined as described earlier using Rat MCP-1 ELISA (Pierce Biotechnology) and ESI-MS/MS,19, 20 respectively. Biliary MCP-1 was standardized to bile protein correcting for bile flow in BDL versus sham rats. MCP-1 levels were correlated with duration of bile duct ligation, METAVIR stage of fibrosis,21 and TCA levels.

Isolation of HSCs and Hepatocytes.

Rat HSCs were isolated from normal rat liver by sequential pronase (Roche Diagnostics, Mannheim, Germany) and collagenase B (Boehringer Mannheim) digestion and separation on an arabinogalactan (Larcoll-Sigma, St Louis, MO) gradient with >95% purity, as described.22, 23 HSCs were cultured on 8.0 μM pore BD Falcon HTS FluoroBlock 24-multiwell insert membranes (Becton Dickinson, Mansfield, MA) coated with rat tail collagen type I (Sigma, St. Louis, MO) at 0.5 × 106 cells/insert in M199 media with supplements,23 including 10% horse and 10% calf serum (CSL, Sydney, Australia), for 5 days which resulted in their culture-activation.10 HSCs were washed and placed in serum-free medium prior to coculture with hepatocytes in migration assays.

Hepatocytes were isolated from sham rats using 0.025% Collagenase B (Boehringer Mannheim) perfusion as described,24 followed by centrifugation on an arabinogalactan gradient, showing >98% purity. Hepatocytes were also isolated from BDL rat livers (12 weeks post-BDL) with Collagenase B increased to 0.05%. Sham or BDL hepatocytes were cultured in collagen-coated Falcon 24-well plates (Becton Dickinson) at 0.5 × 106 cells/well. Hepatocytes were allowed to adhere for 24 hours, then placed in serum-free medium for a further 24 hours prior to coculture migration assay.

In Vitro Coculture Model for HSC Migration Assay.

HSCs were labeled with 10 μM 5-(and 6)-carboxyfluorescein diacetate, succinimidyl ester (CFDA-SE) (C1157; Molecular Probes, Eugene, OR) for 45 minutes at 37°C, washed to remove unincorporated dye, and incubated in serum-free media at 37°C for 30minutes to allow acetate groups of CFDA-SE to be cleaved by intracellular esterases, yielding highly fluorescent amine-reactive carboxyfluorescein succinimidyl ester. These succinimidyl ester groups react with intracellular amines to form fluorescent conjugates.25 Culture-activated HSCs on FluoroBlock cell culture inserts were placed in the 24-well culture plate containing hepatocytes from either sham-ligated or BDL rats, forming the upper and lower chambers of the migration assay insert system, respectively. Cells were cocultured for 0–4 hours during which cell migration of CFDA-SE–labeled HSCs through the FluoroBlock membrane toward hepatocytes in the lower chamber was measured spectrofluorimetrically at regular intervals using a TECAN SpectraFluor Plus plate-reader, with excitation and emission wavelengths of 485 nm and 535 nm, respectively.

Interventional studies were designed to assess the contribution of MCP-1, derived from hepatocytes isolated from BDL rats, to HSC recruitment. Anti-MCP-1 neutralizing immunoglobulin G (2 and 8μg/mL; AB1834P; Millipore, Billerica, MA) or control nonimmune immunoglobulin G was added to the lower hepatocyte chamber for 2 hours prior to coculture. All migration assay results were standardized by subtracting baseline CFDA-SE labeled HSC migration in the absence of hepatocyte coculture. HSC migration was standardized to hepatocyte total cellular protein (assessed by bicinchoninic protein assay) to allow for variation between hepatocyte isolation and culture confluency.

Treatment of Hepatocytes and HepG2 Cells with Taurocholic Acid.

Hepatocytes isolated from control rats were cultured on collagen-coated or Matrigel-coated six-well plates at 0.5 × 106 cells/well for 48 hours. HepG2 cells were plated at 0.25 × 106 cells/well until 70% confluent. Cells were then treated with 15–300 μM TCA (Sigma) in serum-free media for 0–24 hours. Conditioned media was collected, cellular RNA extracted, and MCP-1 protein and mRNA quantitated by ELISA and RT-PCR, respectively.

Statistical Analyses.

Statistical analysis was performed using SPSS version 15.0 (SPSS Inc., Chicago, IL) and GraphPad Prism 4.03 (GraphPad Software Inc., San Diego, CA). Quantitative data was analyzed using the One Sample t-test (for comparison to sham rats), the Independent Samples t-test, the one-way analysis of variance (ANOVA), and the Between Subjects Effects Repeated Measures ANOVA where appropriate. Post hoc comparisons were performed using the least significant difference method. Associations between MCP-1 measures, histological fibrosis stage, and grade of ductular proliferation were conducted using either Spearman's rho or Pearson's R correlation. Receiver operating characteristic (ROC) analysis assessed the diagnostic utility of elevated serum MCP-1 levels in identifying the early stages of hepatic fibrosis in BDL rats and in children with CFLD. For all statistical analyses, a two-tailed P < 0.05 was considered statistically significant.


Human Studies

Cellular Localization of Elevated MCP-1 Expression.

In CFLD and BA liver sections, MCP-1 protein was principally demonstrated in hepatocytes at the growing margin of scar tissue as well as in cholangiocytes (Fig. 1A-H). In addition, HSCs and an occasional single CK-7 stained cell, most likely liver progenitor cells,17 demonstrated positive MCP-1 staining (results not shown). In pediatric control liver, MCP-1 was minimally detected in bile duct epithelial cells, but not in hepatocytes (Fig. 1I).

Figure 1.

Cellular localization of elevated hepatic MCP-1 expression in BA. (A,B) Immunohistochemistry for MCP-1 protein localized to hepatocytes at the scar interface and throughout the regenerative nodule. Immunohistochemistry for MCP-1 protein showing localization to the apical membrane (black arrows) of bile duct epithelial cells (C) and pericanalicular membranes (black arrowheads) of hepatocytes (D). (E) In situ hybridization for MCP-1 mRNA in bile duct epithelial cells in expanded bile ducts. (F) In situ hybridization using the sense (control) probe for MCP-1 mRNA confirming specificity of antisense probe in (E). (G) In situ hybridization for MCP-1 mRNA localized to cholangiocytes within proliferating bile ductules and peri-scar hepatocytes. (H) Dual immunohistochemistry for MCP-1 (brown) in hepatocytes and α-SMA (purple, identifying activated HSCs), showing close spatial association between MCP-1–expressing hepatocytes (black arrows) and activated HSCs in an area of active fibrogenesis along the scar margin. In addition, numerous perisinusoidal activated HSCs (black arrowheads), closely associated with fibrotic bands are surrounded by hepatocytes expressing MCP-1. (I) Immunohistochemistry for MCP-1 protein was localized to bile duct epithelial cells within bile ducts in the portal tract in pediatric control liver. Original magnifications: ×100 (A); ×200 (B, G, H, I); ×400 (C, E, F); ×1000 (D).

In CFLD, hepatocytes were the predominant cell type expressing MCP-1 (18 of 30 biopsies, 60%). In 11 (37%), MCP-1 staining predominated in cholangiocytes within the portal region, although in most of these sections, both hepatocyte and bile duct cells stained positive. In a few cases, MCP-1 expression was not evident at all. In BA, MCP-1 expression was considerably greater, where 25 of 29 (86%) biopsies showed abundant hepatocyte MCP-1 at the interface of scar tissue and throughout the regenerative nodule (Fig. 1A,B) and in cholangiocytes within the portal region (17 of 29 sections, 59%; Fig. 1C).

In situ hybridization confirmed the cellular source of MCP-1 expression. MCP-1 mRNA was principally expressed in hepatocytes along the growing margin of scar tissue and within the regenerating nodule, as well as cholangiocytes within proliferating bile ductules and occasional perisinusoidal cells in expanding regions of the biliary fibrotic scar (Fig. 1G). MCP-1 mRNA was expressed in bile duct epithelial cells in expanded bile ducts within the periportal regions of the liver (Fig. 1E). The specificity of the antisense MCP-1 mRNA probe is shown using the sense control (Fig. 1F). MCP-1 protein was predominantly localized to the apical (Fig. 1C) and pericanalicular membranes (Fig. 1D) of bile duct epithelial cells and hepatocytes, respectively, suggesting for the first time that MCP-1 protein may be actively excreted into bile. Of interest, there appeared to be a close spatial association between α-SMA–positive HSCs and MCP-1–expressing hepatocytes in areas of active fibrogenesis along the growing margin of the scar (Fig. 1H).

Quantitation of Hepatic MCP-1 mRNA.

MCP-1 mRNA in BA livers was increased by 5.2-fold (Fig. 2A) and 2.5-fold in CFLD compared to controls. When expressed relative to the stage of hepatic fibrosis in CFLD, MCP-1 levels were 2.7-fold elevated in fibrosis stage 0, i.e., before any histological evidence of hepatic fibrosis (Fig. 2A).

Figure 2.

MCP-1 and TCA levels in BA and CFLD compared to pediatric controls. (A) Hepatic MCP-1 mRNA expression levels in infants with BA (n = 8) or children with CFLD (n = 14) were compared to pediatric controls (n = 3). Results are expressed standardized to the housekeeping gene GAPDH and are represented as mean ± s.e.m. Children with CFLD were stratified into fibrosis stages15 0 (no fibrosis, n = 4), stages 1-2 (mild-moderate fibrosis, n = 5) and stages 3-4 (severe fibrosis, n = 5). (*P = 0.003, BA versus Controls; # P = 0.007, all CFLD subjects versus controls). (B) Serum MCP-1 concentration was increased in BA (P < 0.0001, n = 20) and CFLD (P < 0.0001, n = 30) compared to Controls (n = 27). (C) Correlation analysis showing a weak negative relationship between serum MCP-1 and the stage of fibrosis (r = −0.39, P = 0.048) in CFLD. Mean ± 2SD for controls is shown. (D) ROC curve for serum MCP-1 in CFLD distinguishing stage 0 fibrosis versus stages 1-4 fibrosis. The ROC AUC is 0.73 (P = 0.08). (E) Biliary MCP-1 levels were increased in CFLD (n = 19) compared to controls (n = 13) (P = 0.003). TCA levels in serum (F) and bile (G), were increased in CFLD compared to controls (P = 0.002 and P = 0.02, respectively). (H) There was a weak positive correlation between MCP-1 and TCA levels in bile in CFLD (n = 19) (r = 0.46, P = 0.046).

Serum and Biliary MCP-1 in Children with BA and CFLD, versus Controls.

Serum MCP-1 levels were significantly increased by 2.1-fold in BA and 1.5-fold in CFLD compared to age-matched controls (Fig. 2B). When serum MCP-1 levels in CFLD were plotted against the stage of hepatic fibrosis, there was a weak negative correlation (Fig. 2C). ROC curve analysis showed that a serum MCP-1 of 1060 pg/mL was 77% sensitive and 67% specific for distinguishing between children with no histological evidence of fibrosis (stage 0) and those with any fibrosis (stages 1-4) (Fig. 2D). Thus, MCP-1 appears to be elevated early in cholestasis in CFLD. However, further studies on a larger cohort of patients is required to assess the potential clinical utility of serum MCP-1 in predicting early fibrogenesis. There was no statistically significant correlation between serum MCP-1 levels and the inflammation grade in patients with CFLD (r = 0.02, P = 0.9; data not shown).

Bile was only available from children with CFLD and controls. Biliary MCP-1 levels were increased 1.6-fold in CFLD versus controls (Fig. 2E). There was no statistically significant correlation between biliary MCP-1 levels and the stage of hepatic fibrosis (results not shown).

In Vivo Association Between MCP-1 and Taurocholate in CFLD.

In CFLD, serum TCA levels were elevated by 11-fold (Fig. 2F) and biliary TCA levels were increased by 2.6-fold (Fig. 2G) compared to controls. There was a weak, but statistically significant correlation between MCP-1 and TCA levels in the bile (Fig. 2H); however, there was no association evident in the serum (results not shown). In pediatric controls, there was no association between MCP-1 and TCA levels in serum or bile (results not shown).

Animal Studies

Immunohistochemistry for MCP-1 in BDL Versus Sham Rats.

Immunohistochemistry for MCP-1 expression in the BDL rat liver revealed a similar distribution to that seen in BA and CFLD. MCP-1 was predominantly expressed in hepatocytes and cholangiocytes, with expression also evident in occasional HSCs (Fig. 3A). MCP-1 was not detected in sham control rat livers (Fig. 3B).

Figure 3.

Cellular localization of elevated hepatic MCP-1 expression in BDL rat liver. (A) Immunohistochemistry for MCP-1 protein localized to hepatocytes throughout the regenerative nodule and cholangiocytes within proliferating bile ductules. (B) No MCP-1 expression was observed in sham control rat liver. Original magnifications, ×200.

Quantitation of MCP-1 mRNA in BDL Rats.

Real-time RT-PCR revealed a significant relationship between hepatic MCP-1 mRNA expression and the duration of bile duct ligation, with a six-fold increase demonstrated as early as 3 days after ligation, indicating MCP-1 is induced very early in cholestasis (Fig. 4A). MCP-1 mRNA preceded other indicators of HSC activation and hepatic fibrosis, with α-SMA and procollagen α1(I) mRNA not significantly elevated until 2 weeks after ligation. Peak MCP-1 mRNA expression of 15.8-fold versus sham controls occurred up to 2 weeks after ligation, returning to control levels by 18 weeks, whereas α-SMA and procollagen α1(I) mRNA peaked at 2 weeks and 6 weeks after ligation, respectively (Fig. 4A). In addition, there was a weak negative correlation between hepatic MCP-1 mRNA expression and the stage of hepatic fibrosis (r = −0.34; P = 0.06), providing further support for a role for MCP-1 in the early events associated with the genesis of fibrosis (results not shown).

Figure 4.

Hepatic, serum, and biliary MCP-1 levels in BDL rats. (A) There was a significant relationship between the hepatic MCP-1 (P = 0.001), α-SMA (P = 0.001) and procollagen α1 (I) (P = 0.0001) mRNA levels and the duration of bile duct ligation. Results expressed as fold-difference relative to sham-controls. MCP-1 mRNA was increased at day 3 (*, P = 0.005), with peak expression at 2 weeks (**, P = 0.0001). α-SMA and procollagen α1 (I) mRNA expression were not increased until 2 weeks (#, P = 0.00002 and ##, P = 0.0007, respectively), with peak expression of procollagen α1 (I) at 6 weeks (***, P = 0.0009). Data are mean ± standard error of the mean of >5 rats/group/time point. (B) Serum MCP-1 levels were increased in BDL versus sham rats (P = 0.002) and (C) there was a significant negative correlation with the duration of bile duct ligation (r = −0.72, P = 0.0002). Mean ± two standard deviations of normal range is shaded. (D) There was a negative correlation between serum MCP-1 levels and the stage of fibrosis in BDL rats (r = −0.54, P = 0.01). Mean + two standard deviations for sham controls is shown. (E) ROC curve for serum MCP-1 distinguished mild-moderate fibrosis (stages 0–2) from severe fibrosis (stages 3–4) in BDL rats (AUC = 0.87, P = 0.01). (F) Biliary MCP-1 levels were increased in BDL versus sham rats (P = 0.002). (B-F; BDL rats, n = 30 and sham controls, n = 12).

Serum and Biliary MCP-1 in BDL Rats.

Serum MCP-1 was elevated by 1.6-fold in BDL rats versus controls (Fig. 4B) and there was a significant negative correlation between serum MCP-1 levels and the duration of ligation (Fig. 4C). A negative correlation was also demonstrated between serum MCP-1 and the stage of hepatic fibrosis in BDL rats (Fig. 4D), similar to that observed in CFLD (Fig. 2C). MCP-1 levels were markedly elevated by 3 days after ligation and by ∼50 days had returned to the normal range (Fig. 4C). All BDL rats with fibrosis stages 0–1 had elevated serum MCP-1 levels outside 2 standard deviations of the mean of sham controls, with 93% for fibrosis stages 0–2 (Fig. 4D). ROC curve analysis for early fibrogenesis, fibrosis stages 0–1 versus established fibrosis, stages 2–4, revealed an area under the curve (AUC) of 0.76 (P = 0.04) (results not shown). The diagnostic accuracy was improved if assessing early fibrosis stages 0–2 versus severe fibrosis stages 3–4, where ROC analysis revealed a serum MCP-1 of 16,207 pg/mL was 83% sensitive and 93% specific for distinguishing BDL rats with early fibrosis, with an AUC of 0.87 (P = 0.01) (Fig. 4E).

There was a 26-fold increase in biliary MCP-1 levels in BDL versus sham rats (Fig. 4F), which was negatively correlated with the stage of fibrosis (r = −0.54, P = 0.002) and positively correlated with the grade of bile ductular proliferation (r = 0.47, P = 0.009) (results not shown).

There was no significant association between inflammation grade and either serum MCP-1 levels (r = 0.23, P = 0.33) or biliary MCP-1 levels (r= −0.07, P = 0.71) in BDL rats (data not shown).

In Vitro Coculture Chemotaxis Assays Using Hepatocytes and HSCs

Because hepatocytes, particularly those at the growing margin of scar tissue, appeared to be a major source of MCP-1 in both CFLD and BA, we investigated whether HSC chemotaxis could be induced by MCP-1 produced by hepatocytes in a cholestatic event. After 4 hours, HSC chemotaxis was increased by 5.8-fold ± 1.5-fold when cocultured with BDL-derived hepatocytes versus sham-derived hepatocytes (Fig. 5).

Figure 5.

Role of MCP-1, derived from hepatocytes isolated from either BDL or sham rats, in HSC chemotaxis. There was a significant increase in HSC chemotaxis toward hepatocytes from BDL rats (–●– ● –) versus hepatocytes from sham controls (–×–×–) over the duration of the experiment (ANOVA, P = 0.009) with a 5.8-fold increase at 4 hours (*, P = 0.005). In interventional studies, we assessed the role of hepatocyte-derived MCP-1 to this enhanced HSC chemotaxis through the use of neutralizing antibody to MCP-1. A total of 2 μg/mL (–▴–▴–) or 8 μg/mL (–▪–▪–) neutralizing antibody to MCP-1 was added to the bottom coculture chamber containing hepatocytes from BDL or sham rats to inhibit HSC chemotaxis. When BDL hepatocytes were pretreated with 8 μg/mL anti-MCP-1 neutralizing antibody (–▪–▪–), there was a significant decrease in HSC migration versus untreated BDL hepatocytes (– ●–●–) after 4 hours (#, P = 0.03). There was no significant difference between BDL-hepatocytes treated with either 2 μg/mL (–▴ –▴ –) or 8 μg/mL (–▪ –▪–) neutralizing MCP-1 antibody versus sham hepatocytes (–×–×–) after 4 hours (P = 0.3 and P = 0.4, respectively), indicating that HSC chemotaxis was returned to baseline (sham) levels by inhibiting the action of BDL hepatocyte-derived MCP-1. Data shown are mean ± standard error of the mean, n = >3/treatment/time point.

In interventional studies, incubation of BDL hepatocytes with anti-MCP-1 neutralizing antibody (2 μg/mL or 8 μg/mL) resulted in markedly decreased HSC migration. Four hours after treatment with 2 μg/mL anti-MCP-1 antibody, HSC migration was decreased by 50.4% ± 18.3%. However, using 8 μg/mL anti-MCP-1 antibody, HSC migration was inhibited by 52.6% ± 19.4% after only 2 hours; after 4 hours, this inhibition had increased to 79.4% ± 7.3% (Fig. 5). Importantly, there was no significant difference in HSC migration, when HSCs were cocultured with BDL hepatocytes treated with 8 μg/mL anti-MCP-1 antibody versus sham-ligated hepatocytes over the 4-hour duration.

Role of TCA in Hepatocyte MCP-1 Induction

TCA induced a dose-dependent increase in MCP-1 mRNA expression in human HepG2 cells and primary rat hepatocytes isolated from sham controls (Fig. 6A,B). From these studies, 150 μM TCA produced optimal results. TCA induced a 3.9-fold ± 0.7-fold increase in MCP-1 mRNA expression in rat hepatocytes after 30 minutes (Fig. 6C). This translated into a 3.4-fold ± 0.3-fold increase in MCP-1 protein after 6 hours in hepatocytes cultured on rat tail collagen (Fig. 6D) and a 2.4-fold ± 0.4-fold increase in cells cultured on Matrigel (Fig. 6E), returning to control levels after 24 hours.

Figure 6.

Role of TCA in MCP-1 induction. (A) Human HepG2 cells and (B) primary rat hepatocytes were treated with 15 μM, 150 μM, or 300 μM TCA for 0-6 hours. (A) TCA induced a dose-dependent increase in MCP-1 mRNA expression in HepG2 cells (P = 0.003), with peak expression at 30 minutes using 150 μM TCA (P = 0.03) and 300 μM TCA (P = 0.058), versus untreated controls. (B) TCA induced MCP-1 mRNA expression in rat hepatocytes with an optimal dose of 150 μM TCA for 30 minutes (P = 0.03). When rat hepatocytes were treated with 150 μM TCA for 10 minutes to 24 hours, there was time-dependent increase in (C) MCP-1 mRNA expression at 30 minutes (P = 0.0002), as well as MCP-1 protein secreted into conditioned media from (D) hepatocytes cultured on rat tail collagen (P = 0.0009), or (E) hepatocytes cultured on Matrigel (P = 0.01) at 6 hours.

In Vivo Association Between MCP-1 and TCA in BDL Rats

TCA levels were elevated by 41-fold in serum (Fig. 7A) and 2.8-fold in bile (Fig. 7B) in BDL versus sham rats. There was a highly significant negative correlation between serum TCA and the duration of bile duct ligation (Fig. 7C), similar to the correlation seen earlier between serum MCP-1 and ligation (Fig. 4C). Thus, both TCA and MCP-1 are elevated early in cholestatic injury and with the demonstrated positive correlation between serum TCA and MCP-1 in BDL rats (Fig. 7D), suggests a potential causal association. A significant negative correlation was observed between MCP-1 and TCA in bile (Fig. 7E). Hepatic MCP-1 mRNA expression was also negatively correlated with biliary TCA (Fig. 7F), but there was no association evident in serum (results not shown). In sham rats, there was no association between TCA levels in serum or bile and either hepatic MCP-1 mRNA, or serum or biliary MCP-1 levels (results not shown).

Figure 7.

In vivo association between MCP-1 and TCA in BDL rats. (A) Serum and (B) biliary TCA levels were increased in BDL rats (P < 0.0001 and P = 0.003, respectively) versus sham controls. (C) There was significant negative correlation between serum TCA and duration of bile duct ligation (r = −0.75, P < 0.0001). (D) Serum TCA levels were significantly correlated with serum MCP-1 levels (r = 0.81, P < 0.0001) in BDL rats, whereas (E) in bile, there was a significant negative correlation between TCA and MCP-1 (r = −0.63, P = 0.0002). (F) Hepatic MCP-1 mRNA was also negatively correlated (r = −0.47, P = 0.01) with biliary TCA levels. (n = 21 for A,C,D; n = 29 for B,E; n = 17 for F).


This study has identified that a key therapeutic target chemokine in cholestatic liver injury, MCP-1, is highly expressed in the liver, and is elevated in the serum in two diverse causes of cholestatic liver injury in children, CFLD and BA. In CFLD, its excretion into both serum and bile appears to be an early event in cholestatic liver injury, prior to or in conjunction with early histological evidence of fibrosis. These findings were similar to and consistent with those observed in the liver, serum, and bile in an animal model of cholestatic liver injury. Moreover, hepatocytes isolated from BDL rats displayed up-regulated MCP-1, which caused markedly increased HSC chemotaxis that was inhibited by up to 80% in interventional studies using a neutralizing antibody to MCP-1. In addition, a potential stimulus for the induction of MCP-1 in hepatocytes is suggested by the in vivo association with the hydrophobic bile acid TCA which is elevated during cholestasis, and the in vitro observation that MCP-1 is induced by TCA in isolated normal human and rat hepatocytes.

Elevated expression of MCP-1 has not previously been demonstrated in causes of childhood cirrhosis, although studies have shown increased expression in adult liver diseases.26, 27 The finding of MCP-1 expression in hepatocytes, particularly those at the edge of active fibrosis, is novel and may be unique to cholestatic liver diseases such as BA and CFLD, because previous studies in adult chronic viral hepatitis have reported expression of MCP-1 limited to inflammatory cells, cholangiocytes, and HSCs. Chemokines generally regulate leukocyte recruitment to various tissues of the body, including the liver. Specifically, MCP-1 is thought to be responsible for recruitment of monocytes and activated T lymphocytes secreted by diverse tissues in response to injury,28 and is one of the principal chemokines associated with HSC recruitment in vitro.13, 14 Although chemokine (C-C) receptor 2 (CCR2) is the major known receptor required for effecting MCP-1-elicited cell signaling events, MCP-1 appears to exert its effects in the absence of CCR2 via unknown receptors in HSCs,13 portal fibroblasts,29 and in aortic smooth muscle cells.30 In this study, we did not find an association between serum MCP-1 and inflammation, but did demonstrate a spatial relationship with activated HSCs. It is possible that excreted MCP-1 may spill into the sinusoidal blood supply having a paracrine effect on HSC recruitment to the scar interface and peribiliary regions, leading to active fibrogenesis in these areas. Alternatively, as suggested by the association with taurocholate levels, MCP-1 expression may occur locally in hepatocytes, particularly those at the edge of the area of active fibrosis as a result of cholestasis, perhaps in the absence of inflammation, which is not a prominent feature in these conditions.

Similarly, serum and biliary MCP-1 have not been studied in causes of childhood cirrhosis, with the exception of a study in BA,31 where patients with moderate liver dysfunction versus normal or severe liver dysfunction had higher serum levels. In our study, the ROC for serum MCP-1 in CFLD suggests elevation of serum MCP-1 levels early in the disease process, verified by the weak but statistically significant negative correlation with hepatic fibrosis stage, and a similar sequence in the BDL rat. Early up-regulation of CC-chemokines such as MCP-1, macrophage inflammatory protein 1α (MIP-1α) and MIP-1β, and RANTES, have been demonstrated in mouse models of fulminant hepatic failure including concanavalin A and galactosamine/lipopolysaccharide, with expression occurring prior to hepatic inflammatory cell infiltration and liver damage.32

The observations in CFLD and BA liver, serum, and bile are supported by experiments using the BDL rat model. Hsieh and colleagues have previously shown increased MCP-1 expression in BDL rats treated with Roux-en-Y choledochojejunostomy and E. coli-induced cholangitis.33 In our experiments, as well as MCP-1 protein expression, elevated MCP-1 mRNA expression was seen prior to histological hepatic fibrosis (six-fold increase on day 3, peaking at 15.8-fold at 2 weeks, and returning to control levels after 18 weeks), biliary MCP-1 levels were increased versus controls, and serum and biliary MCP-1 levels and biliary TCA concentration all negatively correlated with fibrosis. Of note, MCP-1 mRNA expression preceded the expression of both α-SMA (7 days) and procollagen α1(I) (2 weeks) mRNA, which are established markers of HSC activation. In addition, although some MCP-1 was detected in bile from sham rats, substantially elevated levels of biliary MCP-1 (26-fold) were found in BDL rats, supporting our hypothesis that excess MCP-1 protein present along the hepatocyte pericanalicular membrane and apical membrane of cholangiocytes appears to be actively excreted into bile.

To further explore proof of concept, we designed chemotaxis experiments to determine whether hepatocytes have a role in MCP-1-induced HSC recruitment. We were particularly interested in the role of scar interface hepatocytes in HSC recruitment, because this expanding region of periportal fibrosis has the most abundant, active fibrogenesis and HSC activation in CFLD and BA,7, 8 and as Fig. 1A,B,D demonstrates, hepatocytes appear to be one of the major sources of MCP-1 expression in this study. Indeed, hepatocytes isolated from BDL rats caused HSC recruitment at a rate 5.8-fold greater than hepatocytes from sham controls. In interventional studies designed to assess the contribution of MCP-1 to this HSC chemotaxis, we demonstrated that a neutralizing antibody to MCP-1 caused a dose-dependent inhibition in HSC chemotaxis of up to 80% in the presence of BDL rat-derived hepatocytes, providing further support to the hypothesis that MCP-1 plays a key role in HSC chemotaxis in cholestatic liver injury. These data may also explain, in part, the recruitment of portal (myo)fibroblasts to the periductular region in cholestasis. Previous studies using whole bile duct segments isolated from BDL rats also demonstrated induction in HSC chemotaxis,34 and neutralizing antibody to platelet-derived growth factor caused a 60% inhibition of chemotaxis, while anti-endothelin-1 had no effect. Kruglov and colleagues showed that MCP-1 derived from BDL rat cholangiocytes induced the transformation of portal fibroblasts to myofibroblasts, although they did not specifically evaluate the role of cholangiocyte-derived MCP-1 in HSC or portal fibroblast chemotaxis.29

A potential mechanism for MCP-1 induction relates to its association with TCA. A recent study using an immortalized murine cholangiocyte cell line investigated the effects of taurine-conjugated hydrophobic bile acids on cell viability, proliferation, apoptosis, and cytokine release.18 TCA was the only bile acid to have any effect on chemokine release with a four-fold and three-fold induction in MCP-1 and interleukin-6 release, respectively, while having no effect on cell viability, proliferation, or apoptosis.18In vitro, we found that TCA induced the expression of MCP-1 at the message level in both human hepatocytes (HepG2 cells) and in primary cultures of normal rat hepatocytes. This increased mRNA expression translated into a three-fold increase in MCP-1 protein secreted by primary hepatocytes after 6 hours. The in vivo associations between MCP-1 and TCA in both serum and bile in CFLD and BDL rats also suggests a causal link. In this context, the observed difference in slope of biliary MCP-1 and TCA correlations between CFLD and the BDL rat deserves comment. Bile flow in CFLD, where there is no jaundice, is less affected than in the complete biliary obstruction in BDL rats, the former progressing from a focal to more generalized intrahepatic cholestasis. We found a positive biliary correlation in CFLD, where intrahepatic cholestasis results in retention of bile acids in the liver, stimulation of MCP-1 production with release into bile, with reabsorption of both TCA and MCP-1 back into serum via the hepatobiliary plexus. In the BDL rat, we found positive serum and hepatic correlations but a negative biliary correlation, as would be expected. There was an initial increase in MCP-1 expression and production, (Fig. 4), then a reduction, with increasing fibrosis with release into bile and serum, associated with an increase of TCA in serum and bile (Fig. 7). It is, of course, possible that other toxic bile acids may have a role on hepatocyte function and on the expression of a number of different cytokines, chemokines, and growth factors. We have previously demonstrated a potential role for cholic acid in liver disease with serum cholic acid levels reflecting disease progression in CFLD,35 an observation also demonstrated in BDL rats.36 Further studies will be of interest in assessing the contribution of distinct bile acids and their conjugates, both in isolation and in a variety of different combinations.

In summary, this study has suggested a key role for hepatocyte-derived MCP-1, potentially induced by elevated biliary and serum TCA in cholestasis, in the recruitment of HSCs in several diverse causes of cholestatic liver injury. We propose that due to the bile duct obstruction in BA and altered bile flow in CF, MCP-1 excreted into bile by both hepatocytes and bile duct epithelial cells may then “spill” into the sinusoidal blood supply, thus having a paracrine effect on HSC chemotaxis. This may be a potential mechanism for the recruitment of inflammatory cells and HSCs to the peribiliary regions of the acinus, which could contribute to the focal biliary cirrhosis in CFLD, the fibrotic biliary obstruction seen in BA, and the active fibrogenesis along the expanding scar interface in both conditions. The data presented rather necessarily represents correlative analyses of novel observations in human disease, supported by the similar, but consistent observations in an animal model of cholestatic liver disease and in vitro cell culture systems. The findings raise an important and strong hypothesis which will require mechanistic proof beyond the scope of these initial observations. It would be of great interest to further assess the role of MCP-1 in hepatocyte-specific knockout models, in parallel with genetic and chemical interventional studies. Intervention aimed at detoxifying TCA, as highlighted by the therapeutic use of ursodeoxycholic acid to displace toxic bile acids from the bile acid pool37 or neutralizing MCP-1, as suggested by our in vitro experiment, may help prevent progression of the fibrotic liver injury which accompanies these cholestatic diseases in children. Continued investigation of disease mechanisms may not only lead to further development of traditional management approaches, but importantly provide new directions in the prevention and treatment of these diseases.