Article Text

Download PDFPDF

The hedgehog pathway regulates remodelling responses to biliary obstruction in rats
  1. A Omenetti1,
  2. Y Popov2,
  3. Y Jung1,
  4. S S Choi1,
  5. R P Witek1,
  6. L Yang1,
  7. K D Brown1,
  8. D Schuppan2,
  9. A M Diehl1
  1. 1
    Division of Gastroenterology, Department of Medicine, Duke University Medical Center, Durham, USA
  2. 2
    Division of Gastroenterology, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, USA
  1. Dr A M Diehl, Duke University Medical Center, Division of Gastroenterology, GSRB 1, 595 LaSalle Street, Suite 1073, Box 3256, Durham, NC 27710, USA; annamae.diehl{at}duke.edu

Abstract

Background: Chronic biliary obstruction provokes fibrosis and accumulation of immature ductular cells. This fibroductular reaction resolves following biliary decompression, suggesting that it may also be involved in the repair of biliary damage. The hedgehog (Hh) pathway becomes activated in liver after bile duct ligation (BDL), and might modulate hepatic remodelling because Hh ligands are potent morphogens.

Objective: To study the induction of the Hh pathway during progression and resolution of biliary fibrosis, and to clarify whether Hh signalling regulates accumulation of bile duct progenitor cells.

Design and main outcome measures: Livers from rats with BDL were examined by quantitative real-time polymerase chain reaction analysis and immunohistochemistry to identify factors that might stimulate Hh signalling. BDL rats were subjected to Roux-en-Y hepaticojejunostomy (R-Y) to relieve biliary obstruction in order to determine whether these factors and Hh signalling declined as ductular populations and concomitant fibrosis regressed. Cultures of immature ductular cells were treated with putative Hh inducers and Hh ligands to confirm their functional relevance.

Results: BDL increased expression of platelet-derived growth factor-BB (PDGF-BB) and sonic hedgehog (Shh), downregulated hedgehog-interacting protein (Hip), activated Hh signalling, and expanded populations of Hh-responsive ductular cells that expressed pancyotkeratin, a liver progenitor cell marker. After R-Y, Hip remained suppressed, expression of PDGF-BB and Shh gradually declined, and populations of hedgehog-responsive ductular cells regressed. In cultured ductular cells, PDGF-BB treatment induced Shh expression, and incubation with Shh inhibited apoptotic activity.

Conclusions: These results identify a mechanism for activation of the Hh pathway during cholestasis and suggest that Hh signalling regulates ductular cell accumulation after biliary injury.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Biliary injury initiates extensive cross-talk amongst neighbouring cells within portal tracts, presumably initiating a programme that is aimed at reconstructing the damaged bile ducts.1 During this process, the injured biliary epithelium releases soluble factors that promote accumulation and activation of matrix-producing myofibroblastic cells in peri-portal mesenchyme.28 Activated myofibroblastic cells, in turn, release soluble factors that promote the growth of bile ductular cells.9 During chronic biliary injury, there is protracted activation of these tissue repair mechanisms. Consequently, the usually well-organised portal tracts expand into the liver lobule, and the hepatic architecture is gradually remodelled by a fibroductular reaction that consists of proliferating cholangiocytes, fibroblastic cells and associated stroma.

Complete reconstitution of normal liver structure and function is uncommon as long as the factors inciting biliary injury persist.1 7 Hence, by examining liver samples obtained during ongoing biliary injury, it has been difficult to differentiate mechanisms that perpetuate liver injury from those that predominantly mediate repair responses. Reparative mechanisms would likely remain prominent for some time after the injury-provoking insult has been removed, and gradually dissipate as normal liver architecture is restored. The rat model of reversible biliary obstruction is useful for elucidating reparative responses to biliary injury. In this system, the common bile duct is ligated for 4 weeks to produce chronic biliary obstruction; then a Roux-en-Y anastomosis is constructed to decompress the biliary system and animals are monitored for weeks to months until restitution of liver architecture is complete.10 11

In the present study, we evaluated rats with reversible biliary obstruction to determine if hedgehog (Hh) signalling might regulate repair of biliary injury. We focused on the Hh pathway because it mediates mesenchymal–epithelial interactions that modulate tissue development,12 13 as well as remodelling of various adult organs,1416 and we recently reported that hepatic Hh activity increases after BDL in rodents9 and in patients with primary biliary cirrhosis.17 The present study had three goals. The first was to identify potential mechanisms for Hh pathway activation following biliary obstruction. The second was to assess whether or not Hh signalling abated when biliary obstruction was relieved. The third was to directly evaluate the effects of putative Hh pathway inducers and Hh ligands on cholangiocytes in culture.

Our results demonstrated that Hh signalling followed the expected kinetics of regenerative mechanisms. The pathway was activated by biliary obstruction, remained robust for weeks after obstruction was relieved, but gradually subsided as normal hepatic architecture was restored. Activation of Hh signalling followed repression of the Hh inhibitor, Hh-interacting protein (Hip), and accompanied induction of the sonic hedgehog (Shh) ligand. Expression of Shh, in turn, paralleled that of platelet-derived growth factor-BB (PDGF-BB), a cytokine that upregulates expression of Shh in hepatic stellate cells.18 Treatment of cultured cholangiocytes with PDGF-BB also directly increased their expression of Shh. In the rat model of reversible cholestasis, populations of Hh-responsive ductular cells expanded when Shh expression was high and regressed when Shh mRNA levels declined. Treatment of cultured cholangiocytes with recombinant Shh inhibited their apoptotic activity. These findings complement and extend the literature results demonstrating that Hh ligands stimulate growth and repair,1316 and support the hypothesis that the Hh pathway regulates liver remodelling following biliary injury.

MATERIALS AND METHODS

Animals and experimental design

Male Sprague–Dawley rats (Charles River Laboratories, Wilmington, Massachusetts, USA) underwent BDL and scission (n = 32)10 or sham surgery (n = 4). Four weeks later, a Roux-en-Y biliary–enteric anastomosis (R-Y) was constructed10 11 and recovery was monitored over the ensuing 12 weeks. Liver samples were collected at 4 weeks post-BDL (n = 8), and at 3 days (n = 4), 1, 2, 4 and 12 weeks (n = 5 per group) after R-Y. Tissues were processed for RNA analysis and immunohistochemistry. Results were compared to those from sham-operated rats.

Cell culture

The murine cholangiocyte 603B line19 (from GJ Gores, Mayo Clinic, Rochester, Minnesota, USA), was maintained as described,9 20 and then cultured in serum-free media with and without Shh (0.01–1 nmol/l) or Ihh (1–100 nmol/l) (StemCell Technologies, Vancouver, Canada) or PDGF-BB (20 ng/ml, R&D Systems, Minneapolis, Minnesota, USA) for 24 h (Shh, Ihh) or 48 h (PDGF-BB) and harvested for isolation of mRNA or assessment of apoptotic activity.

Caspases 3/7 activity

Apoptotic activity was assayed using the Apo-ONE Homogeneous Caspase 3/7 Apoptosis Assay (Promega, Madison, Wisconsin, USA), according to the manufacturer’s instructions.9 A FLUOstar OPTIMA microplate reader (BMG Labtech, Durham, North Carolina, USA) was used for luminescence and fluorescence measurements.

Two-step real-time reverse transcription-polymerase chain reaction

Total RNA was extracted using TRIzol (Invitrogen, Carlsbad, California, USA). After RNase-free DNase I treatment (Qiagen, Valencia, California, USA), RNA was reverse transcribed to cDNA templates using random primers and Superscript RNase H-reverse transcriptase (Invitrogen) and amplified using a SYBR Green PCR Master Mix (Applied Biosystems, Foster City, California USA). Quantitative real-time polymerase chain reaction (QRT-PCR) was performed.9 18 For all primers pairs (table 1), specificity was confirmed by cloning and sequencing PCR products. Threshold cycles (Ct) were automatically calculated by the iCycler iQ Real-Time Detection System (iCycler, Hercules, California, USA). Each sample was analysed in triplicate. Target gene levels in treated cells or tissues are presented as a ratio to levels detected in corresponding control cells or tissues, according to the ΔΔCt method.

Table 1 Primers for quantitative real-time polymerase chain reaction analysis

Immunohistochemistry

Formalin-fixed, paraffin-embedded liver sections were incubated with primary antibodies to glioblastoma 2 (Gli2, 1:1000; Abcam, Cambridge, Massachusetts, USA) and pancytokeratin (1:500; Dako, Carpinteria, California, USA) at 4°C overnight. Other sections were similarly exposed to non-immune sera. Polymer–horseradish peroxidase (HRP) anti-rabbit (Dako) was used as secondary antibody, and antigens were detected by diaminobenzidine (DAB; Dako). The tissue was counterstained with haematoxylin (Sigma, St Louis, Missouri, USA). Omitting primary antibodies from reactions eliminated staining, demonstrating staining specificity.

Statistical analysis

Results are expressed as means with the standard error of the mean (SEM), unless indicated otherwise. Comparisons between groups were performed using the Student t test. Significance was accepted at the 5% level.

RESULTS

Induction and reversal of the fibroductular response in rats

Expression of procollagen α1(I) mRNA was induced by BDL, and gradually declined as normal hepatic architecture was reconstituted after R-Y biliary decompression (fig 1A), confirming that liver fibrogenesis resolves once biliary injury is alleviated. In addition, cholestatic liver injury triggered peri-portal accumulation of ductular cells that expressed pancytokeratin, a marker of activated cholangiocytes and liver progenitors.21 22 This ductular response also regressed as normal liver architecture was restored (fig 1B–G). These findings verify that the rat BDL/R-Y model is useful for examining mechanisms that regulate the fibroductular response to chronic biliary injury.

Figure 1 Reversible biliary fibrosis model. Adult rats underwent bile duct ligation (BDL) (n = 32) or sham surgery (n = 4). Four weeks later, a Roux-en-Y biliary–enteric anastomosis (R-Y) was constructed. (A) Quantitative real-time polymerase chain reaction analysis of procollagen α1(I) mRNA expression. **p<0.005 vs sham control. (B–G) Immunohistocheminstry for pancytokeratin in sham-operated control (B,E), after 4 weeks of BDL (C,F), and 12 weeks after R-Y (D,G) at ×20 (B–D) and ×40 magnification (E–G).

Differential expression of hedgehog ligands and Hip during induction and reversal of the fibroductular response

Because the growth of immature bile ductular cells in culture is regulated by Hh ligands9 and Hh pathway activity increases after BDL,9 we evaluated our model of reversible cholestatic liver damage to identify factors that might regulate hepatic Hh pathway activity. Expression of Shh, Indian hedgehog (Ihh) and Hip, a factor that inhibits Hh signalling by preventing Shh and Ihh from interacting with their receptors, were assessed by QRT-PCR analysis. BDL upregulated Shh expression. Shh mRNA levels were more than 20-fold higher in BDL rats than in sham-operated controls (fig 2A). After R-Y, Shh expression gradually fell, returning to baseline levels by 12 weeks. Ihh mRNA levels also increased after BDL, although induction of Ihh was much less dramatic than that of Shh (fig 2B). Ihh expression declined to baseline within the first week after R-Y, but steadily rose from 2 weeks onward after biliary decompression, and by 12 weeks after R-Y expression of Ihh was as high as it had been at 4 weeks of BDL, well above the sham-operated controls. Interestingly, while BDL increased levels of Hh ligands, it caused a significant reduction in hepatic expression of Hip (fig 2C), which remained below control values even at 12 weeks post-biliary decompression. These findings identify two plausible mechanisms for Hh pathway activation during cholestasis: repression of Hip (a Hh signalling inhibitor) and induction of Hh ligands.

Figure 2 Hedgehog expression during induction and reversal of biliary fibrosis (A–C). Quantitative real-time polymerase chain reaction analysis of liver RNA: (A) sonic hedgehog (Shh); (B) Indian hedgehog (Ihh); (C) hedgehog-interacting protein (Hip). Data are normalised to sham control (n = 8, BDL; n = 4, 1 week after R-Y; n = 5, per other groups). *p<0.05 vs sham control, **p<0.005 vs sham control, †p<0.05 vs BDL. R-Y, Roux-en-Y.

Hedgehog target gene expression changes during induction and reversal of biliary fibrosis

Hh ligands regulate Hh-responsive cells by interacting with the cell surface receptor Patched (Ptc), to de-repress a co-receptor, Smoothened (Smo), activating an intracellular signalling cascade that culminates in the nuclear translocation of Gli family transcription factors (Gli1, Gli2, Gli3). This, in turn, increases the expression of Hh target genes, including certain Hh pathway components, such as Gli2.12 13 2325 To determine if changes in the expression of the Hh ligands and Hip influenced Hh signalling, we evaluated the effects of BDL and biliary decompression on levels of Gli2 mRNA (fig 3A). BDL induced Gli2 expression significantly. After R-Y, Gli2 mRNA levels gradually fell and eventually normalised by 12 weeks. To verify that these mRNA changes were accompanied by changes in Gli2 protein expression, and to localise the cell types that were Hh responsive (ie, Gli2 positive), immunohistochemistry was performed (fig 3B–E). The livers of sham-operated rats, exhibited a few Gli2(+) cells that localised in portal tracts (PTs). In contrast, livers of BDL rats contained many peri-portal Gli2(+) ductular cells and stromal cells. Patterns of Gli2 expression changed after relief of biliary obstruction. By 1 week after R-Y, for example, many periportal hepatocytes had become Gli2(+). By 12 weeks after R-Y, when normal liver architecture was nearly restored, the pattern of Gli2 expression was similar to sham-operated controls, with only rare PT cells expressing this protein. Interestingly, at this late time point, levels of Gli1 mRNA were still somewhat increased, but expression of Gli3 had also reverted to baseline (fig 3F). Unfortunately, lack of reliable Gli1 or Gli3 antibodies for immunohistochemistry precluded cellular localisation of these factors. Nevertheless, the differential effects of BDL and biliary decompression on the size of Gli2(+) ductular cell populations support the concept that coordinated changes in the expression of the Hh ligands and Hip influenced the ductular response to biliary injury by modulating Hh signalling in Hh-responsive ductular cells.

Figure 3 Glioblastoma 2 (Gli2) gene and protein expression during induction and reversal of biliary fibrosis. (A) Quantitative real-time polymerase chain reaction analysis of Gli2 after bile duct ligation (BDL) and BDL + R-Y vs sham-operated rats (n = 8 BDL, n = 4 1 week after R-Y, n = 5 in each other group), **p<0.005 vs sham control. (B–E) Immunohistochemistry for Gli2 (brown nuclear staining) in representative sham control, BDL rat, BDL rat 1 week after R-Y, and BDL rat 12 weeks after R-Y. Arrows demonstrate Gli2(+) cells. (F) QRT-PCR analysis of Gli1 (white bars) and Gli3 (grey bars) after BDL and BDL + R-Y vs sham-operated rats (n = 8 BDL, n = 4, 1 week after R-Y, n = 5 in each other group), **p<0.005 vs sham control, †p<0.05 vs BDL.

The ductular cell population that expands after BDL is enriched with pancytokeratin(+) immature cells that can generate either hepatocytes or cholangiocytes.22 26 Because Gli2 expression paralleled the expansion and contraction of the ductular cell compartment after BDL, we examined serial liver sections from rats after BDL, 1 week and 12 weeks after R-Y to determine whether or not Gli2 was expressed by the pancytokeratin(+) population. At 4 weeks of BDL, pancytokeratin staining was localised predominantly in peri-portal ductular cells, and most of these pancytokeratin(+) ductular cells expressed Gli2 (fig 4A,B). Gli2 was also detected in pancytokeratin(−) stromal cells and some endothelial cells in PT of BDL rats. One week after R-Y, when many hepatocytic cells near PT strongly expressed Gli2 (figs 3D and 4C), most peri-portal hepatocytic cells also stained positively for pancytokeratin (fig 4D). Thus, the pattern of pancytokeratin and Gli2 expression shifted during the early recovery period from cholestatic liver injury, with hepatocytic expression of these markers becoming prominent once biliary obstruction was relieved. Results obtained in immunostained sections from rats 12 weeks after R-Y biliary decompression demonstrated further changes in the pancytokeratin/Gli2(+) population. By this time, liver architecture had been largely restored to normal, hepatocytic expression of Gli2/pancytokeratin was negligible, and only occasional small cells in PT stained positive for Gli2 (fig 4E) or pancytokeratin (fig 4F). Thus, the phenotype and localisation of Gli2/pancytokeratin(+) cells changed with the onset and relief of biliary obstruction.

Figure 4 Shift in pancytokeratin/Gli2 positive populations from ductular to hepatocytic cells during liver remodelling. Immunohistochemistry in serial sections immunostained for Gli2 (A,C,E) and pancytokeratin (B,D,F) in 4 week BDL-operated rats (A,B) and after 1 week (C,D) or 12 weeks (E,F) of R-Y (×40). Pancytokeratin(+) ductular cells co-expressing Gli2 (arrows). Gli, glioblastoma.

PDGF-BB induces ductular cell expression of Shh, and Shh inhibits apoptosis of ductular cells

The temporal correlations that were noted between variations in Shh expression and the size of Hh-responsive, immature ductular cell populations suggested that Shh might be driving ductular cell accumulation in cholestatic livers. This stimulated efforts to identify an injury-related factor that might induce Shh expression. Injured biliary epithelia produce PDGF-BB,3 a growth factor for myofibroblastic stellate cells,3 and PDGF-BB plays a major role in myofibroblast accumulation and liver fibrosis after BDL.3 4 27 28 Recently, it was shown that treating primary rat stellate cell cultures with PDGF-BB upregulates their expression of Shh mRNA via PI3K/AKT-dependent mechanisms.18 Although Shh mRNA levels dramatically increased during BDL as compared to after R-Y, it is not known if PDGF-BB regulates Shh expression during cholangiocyte/progenitor cell activation in biliary fibrogenesis. Therefore, we used QRT-PCR analysis to determine how changes in PDGF-BB expression related to variations in Shh expression in the rat BDL/R-Y model. Notably, expression of PDGF-BB and Shh mRNA varied in parallel with each other throughout the injury and repair process (fig 5A).

Figure 5 Platelet-derived growth factor-BB (PDGF-BB) induces expression of sonic hedgehog (Shh) and Shh inhibits apoptosis of ductular cells. Quantitative real-time polymerase chain reaction (QRT-PCR) analysis of PDGF-BB and Shh expression in sham controls, bile duct ligated (BDL) rats, and BDL rats 1 and 12 weeks after Roux-en-Y (R-Y) anastomosis (A). QRT-PCR analysis of Shh and Indian hedgehog (Ihh) mRNA in cholangiocytes treated with PDGF-BB (20 ng/ml) (B). Cholangiocyte caspase 3/7 activity after 24 h treatment with vehicle, Shh (1 nmol/l) or Ihh (1 nmol/l) (C). *p<0.05 vs control, **p<0.005 vs control, †p<0.05 vs BDL.

Because assessment of Hh ligand expression is not possible on formalin-fixed rat tissues, we could not localise production of Shh by immunohistochemistry. Therefore, to determine if, like stellate cells, cholangiocytes also respond to PDGF-BB exposure by inducing Shh expression, we treated a well-characterised, immature cholangiocyte cell line (603B)9 19 with PDGF-BB. To determine the specificity of PDGF-BB effects on Hh ligand production, we also monitored cholangiocyte expression of Ihh, because Ihh mRNA levels did not consistently change in conjunction with PDGF-BB expression after BDL/R-Y (figs 2 and 5). PDGF-BB treatment doubled levels of Shh mRNA in cultured cholangiocytes, but had little effect on Ihh expression (fig 5B).

Shh functions as an autocrine viability factor for myofibroblastic stellate cells, and is necessary for PDGF-BB to induce stellate cell proliferation.18 In our rat model of reversible biliary fibrogenesis, mRNA levels of PDGF-BB, Shh and Gli2 were greatest when populations of pancytokeratin/Gli2(+) ductular cells were largest and declined as these cell populations regressed. Because those findings suggest that Shh may exert direct, growth regulatory effects on immature ductular cells, we treated the immature cholangiocyte line with Shh or Ihh and monitored proliferation (evidenced by BrdU incorporation) and apoptosis (evidenced by caspase 3/7 activation). Neither recombinant Hh protein affected proliferative activity under these culture conditions (data not shown). However, Shh, but not Ihh, significantly inhibited caspase 3/7 activity (fig 5C). Indeed, cholangiocytes were exquisitely sensitive to the anti-apoptotic actions of Shh, exhibiting dose-related decreases in caspase 3/7 activity when exposed to picomolar concentrations of Shh. In contrast, 100-fold higher doses of Ihh did not reduce caspase 3/7 activity in these cells.

In aggregate, these results suggest that increased cholangiocyte production of PDGF-BB during biliary obstruction, apart from driving myofibroblast and stellate cell proliferation,29 acts in an auto- and paracrine fashion to stimulate cholangiocyte production of Shh that, in turn, promotes expansion of ductular cell populations by inhibiting their apoptosis.

DISCUSSION

Hh ligands are potent morphogens that orchestrate tissue construction.12 13 15 Hh pathway activation generally promotes accumulation of progenitor cells.30 31 When appropriately regulated, this may accelerate wound healing. For example, injection of Hh ligands promotes healing of myocardial infarcts and other types of ischaemic tissue injury.32 However, excessive or insufficient Hh activity is detrimental. Hh pathway inhibition in early embryogenesis causes cyclopia, for example, while excessive developmental Hh signalling results in extra digits.33 In adults, inhibition of the Hh pathway contributes to gastric atrophy,34 but many adult cancers exhibit excessive Hh activity.3537 These data argue that appropriate modulation of the Hh pathway is required to shrink and expand populations of Hh-responsive progenitors in accordance with changing demands for generation/replacement of mature cells during tissue development and regeneration.

Consistent with this concept, small numbers of Hh-responsive cells have been demonstrated within progenitor niches of many healthy adult tissues. These Hh-responsive populations expand and differentiate when growth is required.13 Until very recently, the possibility that Hh-responsive progenitors might play a role in regenerating injured adult livers had not been considered. However, emerging evidence that myofibroblastic stellate cells18 38 and ductular cells9 are capable of producing and responding to Hh ligands9 18 38 raises the intriguing possibility that the Hh pathway might regulate repair responses in adult livers.

Populations of myofibroblasts and ductular cells expand dramatically during many types of chronic liver injury in adults, and gradually dissipate as normal hepatic architecture is restored.1 27 39 A role for Hh signalling in regulating this process is supported by a recent study that demonstrated an increased fibroductular response to BDL in mice with genetic impairments that impede downregulation of Hh pathway activity.9 However, viable adult mice with targeted disruption of Hh signalling in liver myofibroblasts or ductular cells have not been generated. Coupled with the protracted time course of recovery from chronic biliary injury, and the expense of Hh neutralising antisera and pharmacological Hh pathway inhibitors, this has made it impractical to evaluate the converse (ie, that inhibiting Hh signalling blocks the fibroductular reaction and delays repair of cholestatic liver damage).

Therefore, the present study took an alternative approach to assess the relevance of Hh pathway activation during biliary injury. The strategy was based on the concept that there must be injury-related mechanisms in adult livers that control the expansion and contraction of Hh-reactive liver cell populations if such cells are truly involved in liver repair after BDL. To identify these mechanisms, a rat model of reversible cholestatic liver damage was examined at multiple time points during the onset and regression of the fibroductular response. Because bile duct progenitors are known to be involved in biliary repair,1 40 induction and regression of putative Hh pathway regulators were correlated with changes in the size of Hh-responsive populations of immature ductular cells in the rat model. Cultures of immature cholangiocytes were then treated with these factors to determine the direct effects of the putative Hh pathway regulators.

BDL perturbed several factors that are known to regulate Hh signalling. It significantly induced hepatic expression of two Hh ligands, Shh and Ihh, and also repressed expression of Hip, a Hh ligand antagonist. These findings suggest that biliary injury leads to events that increase ligand-initiated Hh signalling. Our results identify PDGF-BB as one of the key injury-related factors that promote Hh pathway activation after biliary obstruction. Variations in Shh mRNA levels tightly paralleled fluctuations in PDGF-BB expression in rats with reversible cholestatic liver damage, and PDGF-BB treatment increased Shh expression in cultured cholangiocytes. In contrast, levels of Ihh mRNA correlated poorly with changes in PDGF-BB expression in the rat BDL/R-Y model, and PDGF-BB treatment failed to induce Ihh expression in cultured cholangiocytes. Hence, the effects of PDGF-BB on Hh ligand production after BDL were specific for Shh. Other, yet to be identified, factors regulate Ihh expression during cholestatic liver damage. In mouse uterine tissues, for example, Ihh expression is inhibited by oestrogens and stimulated by progesterone.41 42 Because we demonstrated that Hip expression was consistently repressed after BDL, regardless of changes in PDGF-BB expression, factors other than PDGF-BB must also have influenced Hip mRNA levels during the onset and regression of biliary injury.

In the rat BDL/R-Y model, accumulation and regression of the fibroductular response to biliary injury correlated best with changes in Shh expression: populations of Gli2(+) cells were largest when Shh expression was highest and smallest when Shh expression was lowest. Apart from Shh activation of Gli2 via PDGF-BB, PDGF-BB may also have enhanced Gli2 levels by activating AKT,18 since AKT stabilises cellular levels of Gli, thereby enhancing Gli transcriptional activity in Hh-responsive cells.43 Hip repression likely further promoted Shh function, contributing to the upregulation of Gli2 expression and accumulation of Hh-responsive, Gli2(+) cells when Shh expression increased after BDL. However, downregulation of Hip was insufficient to maintain Gli2 expression when Shh levels declined. Our data also demonstrate that Ihh was incapable of substituting for Shh as a Gli2-inducing factor. Indeed, when livers neared recovery from cholestatic damage, numbers of Gli2(+) cells declined as expression of Ihh rose, suggesting that ligand switching (ie, swapping Shh for Ihh) actually helped to downregulate Hh signalling in cholangiocytes at the end of the repair response. Evidence that Shh, but not Ihh, inhibited apoptotic caspases in cultured cholangiocytes supports the concept that these two Hh ligands exerted divergent effects on the recovery process from cholestatic damage. At present, the mechanisms underlying such differences remain unexplained. However, they were not likely explained by opposing effects on Gli3 expression. This Hh-regulated factor sometimes functions as a transcriptional repressor.44 However, Gli3 mRNA levels increased only modestly after BDL and returned to basal levels despite increased expression of Ihh late in the recovery period.

In summary, the present study is the first to demonstrate that changes in Hh pathway activity parallel the induction and regression of the fibroductular response to chronic biliary injury. The results show that in damaged livers, as in culture, immature ductular cells are Hh-responsive, upregulating expression of Hh target genes and accumulating when Shh increases. Several complementary mechanisms that likely collaborate to activate Hh signalling in these cells were also identified. These include PDGF-mediated induction of Shh gene expression and stabilisation of Gli2, a Shh-induced transcriptional activator, as well as repression of the Shh receptor antagonist, Hip. The present evidence supports the concept that PDGF-BB collaborates with Shh to expand ductular cell populations after biliary injury. This is particularly noteworthy because cholangiocyte-derived PDGF-BB is already known to play a major role in the expansion of myofibroblast populations after BDL,3 28 and it was recently proven that Hh signalling is necessary for PDGF-BB to induce proliferation of liver myofibroblasts in culture.18 By demonstrating that Hh ligands control the growth of two of the key cell populations involved in biliary repair, these findings refine current understanding of autocrine–paracrine signals that regulate liver remodelling after biliary injury and suggest novel diagnostic and therapeutic targets in patients with cholestatic liver disease.

Acknowledgments

We thank Dr GL Gores (Mayo Clinic, Rochester, USA) and Dr Y Ueno (Tohoku University School of Medicine, Sendai, Japan) for providing the murine cholangiocyte cell line.

REFERENCES

Footnotes

  • See Commentary, p 1198

  • Funding: This work was supported by National Institutes of Health Grants 5RO1-AA010154-11 and 5RO1-DK053792-07 to AMD. AO is a recipient fellow of Dottorato di Ricerca in and Alimenti e salute (VI ciclo), Universita’ Politecnica delle Marche, Ancona, Italy.

  • Competing interests: None.

  • Ethics approval: Experiments using rats were approved by the Government of Lower Franconia, Germany, and the Institutional Review Board of Beth Israel Deaconess Medical Center on the 8 February 2007.

Linked Articles