Biliary epithelial cells and primary biliary cirrhosis: The role of liver-infiltrating mononuclear cells


  • Potential conflict of interest: Nothing to report.


Primary biliary cirrhosis (PBC) is characterized by the highly selective autoimmune injury of small intrahepatic bile ducts, despite widespread distribution of mitochondrial autoantigens. On this basis, it has been suggested that the targeted biliary epithelial cells (BECs) play an active role in the perpetuation of autoimmunity by attracting immune cells via chemokine secretion. To address this issue, we challenged BECs from patients with PBC and controls using multiple Toll-like receptor (TLR) ligands as well as autologous liver-infiltrating mononuclear cells (LMNCs) with subsequent measurement of BEC phenotype and chemokine production and LMNC chemotaxis by quantifying specific chemokines. Our data reflect that BECs from PBC patients and controls express similar levels of TLR subtypes, CD40, and human leukocyte antigen DRα (HLA-DRα) and produce equivalent amounts of chemokines in our experimental conditions. Interestingly, however, BEC-expressed chemokines elicit enhanced transmigration of PBC LMNCs compared with controls. Furthermore, the addition of autologous LMNCs to PBC BECs led to the production of higher levels of chemokines and enhanced the expression of CD40 and HLA-DRα. Conclusion: We submit that the proinflammatory activity of BECs in PBC is secondary to the intervention of LMNCs and is not determined per se. These data support the hypothesis that BECs are in fact “innocent victims” of autoimmune injury and that the adaptive immune response is critical in PBC. (HEPATOLOGY 2008.)

Primary biliary cirrhosis (PBC) is characterized by a highly selective autoimmune injury of intrahepatic bile ducts.1 Evidence suggests that there is a direct role of biliary epithelial cells (BECs) in the T cell–mediated insult through features of BECs that render such cells unique targets.2 However, the recurrence of PBC following liver transplantation3 militates against this thesis, and study of autoreactive T cells suggests that BECs are “innocent victims”.4, 5 BECs are known to contribute to the defense mechanisms against infectious agents recognizing pathogen-associated molecular patterns via specific membrane Toll-like receptors (TLRs)6 and produce chemokines that activate both the innate and adaptive immunity compartments.7, 8 BECs also regulate the migration of immunocompetent cells to the liver9 and thus contribute to the perpetuation of autoimmunity.10

An interaction between autoreactive T cells and BECs via CD40 or human leukocyte antigen DRα (HLA-DRα) and facilitated by chemokines has been postulated,11 but data to prove this view in PBC are limited. We thus undertook an extensive study to determine if the recognition of specific pathogen-associated molecular patterns by BECs elicited a different chemokine production and membrane molecule expression in patients with PBC. We report herein that liver-infiltrating mononuclear cells (LMNCs) from patients with PBC have unique and significant differences in responses to chemoattractants and that they enhance the BEC response to TLR stimulation compared with controls. These results suggest that the role of BECs in PBC are secondary to autoreactivity within the liver and are not due to a unique PBC BEC phenotype.


BEC, biliary epithelial cell; HBV, hepatitis B virus; HCV, hepatitis C virus; HLA-DRα, human leukocyte antigen DRα; LMNC, liver-infiltrating mononuclear cell; PBC, primary biliary cirrhosis; TLR, Toll-like receptors.

Materials and Methods

The explanted livers were all derived from patients with cirrhosis. There were a total of 7 explanted livers from patients with PBC, 3 explanted livers from patients with hepatitis B virus (HBV) and 5 explanted livers from patients with hepatitis C virus (HCV). All patients had end-stage liver cirrhosis without signs of unrelated acute liver injury. The diagnosis of PBC was based on established criteria1 and all patients with PBC had positive serum antimitochondrial antibodies. All patients provided written informed consent using a study approved by the ethics committee of Kyushu University Hospital.

Isolation of Intrahepatic BECs and LMNCs.

Liver specimens were first digested with 1 mg/mL of collagenase type I (Sigma-Aldrich, Tokyo, Japan) and digested cells were then gradient-separated to obtain LMNCs.12 BECs were separated from LMNCs using magnetic bead-conjugated BerEp4 antibodies (Invitrogen, Carlsbad, CA) specific for epithelial cells. Cells were then resuspended in a medium obtained from a 1:1 mixture of Ham's F12 and Dulbecco's modified Eagle's medium, supplemented with 5% fetal bovine serum, epithelial growth factor (10 ng/mL), cholera toxin (10 ng/mL), hydrocortisone (0.4 μg/mL), tri-iodothyronine (1.3 ng/mL), transferrin (5 μg/mL), insulin (5 μg/mL), adenine (24.3 μg/mL) (all from Sigma), and 10 ng/mL hepatocyte growth factor (R&D Systems, Minneapolis, MN) and cultured as described.13 Cell phenotype was verified via immunohistochemistry with antibodies against cytokeratins 7 and 19 (Dako, Glostrup, Denmark), and cell purity exceeded 99%. The viability of all cells was greater than 95% as verified by Trypan blue exclusion observation.

Cells (104/mL in 48-well plates) were first cultured for 48 hours to study molecule expression and chemokine production in the presence of multiple TLR stimuli, including 10 μg/mL lipoteichoic acid, 10 μg/mL polyinosinic:polycytidylic acid [poly(I:C)], 10 μg/mL purified lipopolysaccharide, or 10 μg/mL flagellin. All TLR ligands were purchased from Invivogen (San Diego, CA). LMNC studies were performed as described below.

Isolation of Messenger RNA and Expression Analysis.

Total RNA was isolated from cultured BECs using RNAeasy columns (Qiagen, Valencia, CA) and quantitative analyses performed with real-time polymerase chain reaction assay using SYBR Green PCR Master Mix (Invitrogen) and an ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Tokyo, Japan). Expression data were obtained for TLRs, CD40, HLA-DRα, CD80, CD86, and glyceraldehyde 3-phosphate dehydrogenase using primers illustrated in Table 1. Data are expressed as the fold change versus glyceraldehyde 3-phosphate dehydrogenase.

Table 1. Primer Sequences Used for Real-Time Polymerase Chain Reaction Analyses
GeneForward PrimerReverse Primer
  1. Abbreviation: GAPDH, glyceraldehyde 3-phosphate dehydrogenase.


Cytokine and Chemokine Production Following TLR Stimulation.

Supernatants from cultured BECs with different TLR stimuli were analyzed for chemokine production using sandwich enzyme-linked immunosorbent assay kits (R&D Systems) using a combination of unlabeled and biotin- or enzyme-coupled monoclonal antibody to GRO-α/CXCL1, ENA-78/CXCL5, GCP-2/CXCL6, IL-8/CXCL8, MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5, and IP-10/CXCL10. The Raybio Human Cytokine Antibody Array G Series 2000 (Norcross, GA) was used to analyze cytokine levels in cell supernatants.

LMNC Chemotaxis.

The chemoattractant activity of CXCL1, CXCL5, CXCL6, CXCL8, CCL3, CCL4, CCL5, and CXCL10 (R&D Systems) on LMNCs from PBC cases and controls was assessed in 24-well plates assembled with Chemotaxicell (8-μm pore size; Kurabou, Osaka, Japan). LMNC suspensions (1 × 106 cells/1 mL) were seeded and incubated for 2 hours with 1 μg/mL of each chemokine. Transmigrated LMNCs were counted in 10 randomly chosen high-power microscopy fields (×400), as described previously.14

Effect of LMNCs on BEC Surface Marker Expression and Chemokine Production.

PBC and non-PBC LMNCs were cultured (2 × 106/well) in 24-well plates precoated with 5 μg/mL of human anti-CD3 antibody (R&D Systems) with 1 mL of complete RPMI 1640 with 10% fetal bovine serum for 72 hours. Supernatants and LMNCs were harvested individually, and LMNCs were irradiated (3000 rad) before being added to cultured BECs. BECs added with anti-CD3–stimulated LMNC supernatant for 48 hours were used for membrane marker expression, and irradiated LMNCs were used for the study of BEC chemokine production with poly(I:C) as described.

Statistical Analysis.

All experiments were performed in triplicate and all data are expressed as the mean ± standard error of the mean. Comparisons between parameters were performed using Student t tests. All analyses were 2-tailed, and P values < 0.05 were considered as statistically significant. Statistical analyses were performed using Intercooled Stata version 8.0 (Stata Corp, College Station, TX).


TLR-Induced BEC Expression.

Figure 1 reflects data on expression of CD40 and HLA-DRα molecules on BECs from patients with PBC and controls under basal and TLR stimulation conditions. No significant differences were observed in the expression of all studied TLRs. CD80 and CD86 expression was not detected in either of the BEC populations (data not shown). Our preliminary study of chemokine production demonstrated that BECs [either unstimulated or challenged with poly(I:C)] produce significant amounts of GRO-α/CXCL1, ENA-78/CXCL5, GCP-2/CXCL6, IL-8/CXCL8, MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5, and IP-10/CXCL10; a representative protein assay is depicted in Fig. 2. No differences were observed in BEC chemokine production in any of the stimulation conditions between PBC and controls (Fig. 3). The other chemokines studied herein had negligible production levels by BECs and were not studied further.

Figure 1.

CD40 and HLA-DRα expression in BECs in 7 patients with PBC and 8 viral disease controls (3 with HBV, 5 with HCV). No significant differences were observed between patients and controls in any of the examined conditions.

Figure 2.

Preliminary study of cytokines and chemokines produced by cultured BECs. BECs were studied either under basal conditions or after stimulation with poly(I:C) for 48 hours. Cell-free culture supernatants were then analyzed using a protein array kit to evaluate 174 different proteins simultaneously. Unstimulated cells produced detectable amounts of GRO-α/CXCL1, ENA-78/CXCL5, GCP-2/CXCL6, and IL-8/CXCL8, whereas poly(I:C) stimulation led to enhanced MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5, and IP-10/CXCL10. These molecules were thus chosen for further experiments.

Figure 3.

BEC chemokine production under different TLR stimulation conditions using BECs from 7 patients with PBC and 8 viral disease controls (3 with HBV, 5 with HCV). None of the TLR ligands elicited a different chemokine response in the 2 populations.

LMNC Chemotaxis.

The chemokines identified in our previous experiments on BECs were used for the study of LMNC chemotaxis (Fig. 4). PBC was similar to controls with respect to GRO-α/CXCL1, ENA-78/CXCL5, GCP-2/CXCL6, and IL-8/CXCL8 as attractants. Conversely, PBC LMNCs manifested a significantly higher transmigration rate (expressed as the number of transmigrated cells per high-power microscopy field) compared with control cells when challenged with MIP-1α/CCL3 (19.2 ± 6.1 in PBC versus 11.2 ± 2.9 in controls; P = 0.01), MIP-1β/CCL4 (17.6 ± 5.4 versus 10.6 ± 2.7; P = 0.01), RANTES/CCL5 (18.4 ± 6.9 versus 11.1 ± 5.3; P = 0.04), and IP-10/CXCL10 (23.8 ± 5.9 versus 11.2 ± 3.2; P < 0.01).

Figure 4.

Chemotactic effects of specific chemokines on LMNCs from 7 patients with PBC and 8 viral disease controls (3 with HBV, 5 with HCV). PBC cells challenged with MIP-1α/CCL3, MIP-1B/CCL4, RANTES/CCL5, and IP-10/CXCL10 manifest a significantly enhanced transmigration when compared with control LMNCs. *P < 0.05.

LMNC-Induced BEC Molecule Expression.

Figure 5 illustrates the relative fold changes in CD40 and HLA-DRα expression in PBC and non-PBC BECs when the supernatant from autologous CD3-stimulated LMNC and single TLR stimuli were used. In the absence of TLR stimulation, PBC BECs demonstrate a significantly higher expression of both CD40 (1.25 ± 0.37 for PBC versus 0.80 ± 0.20 for controls; P = 0.02) and HLA-DRα (0.23 ± 0.07 for PBC versus 0.16 ± 0.04 for controls; P = 0.02) when autologous LMNC supernatant was added to the culture. Furthermore, this difference was confirmed for both molecules in all experimental conditions [that is, when lipopolysaccharide, lipoteichoic acid, poly(I:C), or flagellin were used as TLR ligands]. BECs did not express CD80 or CD86 in both populations (data not shown).

Figure 5.

CD40 and HLA-DRα expression in BECs in the presence of TLR stimulation and autologous LMNC supernatant in 7 patients with PBC and 8 viral disease controls (3 with HBV, 5 with HCV). The addition of autologous LMNC supernatant elicited a significantly higher expression of CD40 and HLA-DRα in PBC BECs compared with controls. No change was observed in the expression of CD80 and CD86 (data not shown). *P < 0.05.

LMNC-Induced BECs Chemokine Production.

The production of MIP-1α/CCL3, MIP-1β/CCL4, RANTES/CCL5, and IP-10/CXCL10 was investigated in BECs following interaction with autologous irradiated LMNCs. These chemokines were chosen based on recent data11 as well as data from the present study (Fig. 2). We note that irradiated LMNCs did not secrete any of the mentioned chemokines (data not shown) and were added to BECs. Chemokine production from poly(I:C)-challenged BECs following the addition of irradiated autologous LMNCs are illustrated in Fig. 6. PBC-cultured BECs produced significantly higher levels of chemokines compared with control BECs when autologous LMNCs were used. In fact, the production was higher in PBC cells compared with controls for MIP-1α/CCL3 (5.1 ± 1.5 ng/mL in PBC versus 3.1 ± 0.5 ng/mL in controls; P = 0.01), MIP-1β/CCL4 (5.4 ± 1.4 ng/mL versus 3.1 ± 0.9 ng/mL; P = 0.01), RANTES/CCL5 (5.1 ± 1.5 ng/mL versus 3.2 ± 0.8 ng/mL; P = 0.01), and IP-10/CXCL10 (5.7 ± 1.2 ng/mL versus 3.2 ± 0.5 ng/mL; P < 0.01).

Figure 6.

BEC chemokine production in the presence of TLR stimulation and cell contact of autologous LMNCs in 7 patients with PBC and 8 viral disease controls (3 with HBV, 5 with HCV). BECs from PBC cocultured with irradiated autologous CD3-stimulated LMNCs produced significantly higher levels of MIP-1α/CCL3, MIP-1B/CCL4, RANTES/CCL5, and IP-10/CXCL10 compared with control BECs. *P < 0.05.


We herein report the unique observation that LMNCs from patients with PBC manifest significant differences in their response to chemoattractants proven to be produced by BECs and, more importantly, enhance the BEC response to TLR stimulation when compared with controls. Conversely, no differences in such responses were observed between PBC and non-PBC cultured BECs in the absence of autologous LMNCs. Our data strongly support the hypothesis that BECs exert a role in the disease pathogenesis that is secondary to cellular autoreactivity within the liver.

Available evidence suggests that multiple mechanisms lead to tolerance breakdown (disease initiation) and autoimmunity perpetuation15–18 in PBC. It is therefore important to determine whether BECs play a direct role in PBC or are mere targets of the autoimmune reaction mediated by T cells and autoantibodies. Knowledge of the pathophysiology of the biliary epithelia has increased considerably in recent years and clearly suggests an active role in liver immunopathology.19 Indeed, it is now well established that BECs express class II HLA molecules, secrete proinflammatory chemokines, and express multiple TLR subtypes on their membranes, possibly connecting the innate and acquired arms of the immune response.20, 21 These functions have obvious implications in the defense against infectious agents and prevention of bacterial cholangitis, but they may also contribute to the loss of tolerance, along with other cellular compartments.22, 23 Indeed, although there are no significant data on BEC expression of major histocompatibility complex molecules, the interactions between BECs and autoreactive T cells, and their peculiar apoptotic properties, such data still have failed to provide a clear answer on the role of BECs in either the primary or secondary events of PBC.13, 24–26 We do note, however, the exciting pilot data on the possible remodeling of injured epithelium through the hedgehog signaling pathway.27

The second major player in the PBC scenario is the immune cell population that infiltrates the affected liver. Recent data have demonstrated a qualitative defect of the T regulatory compartment in patients with PBC,28 while memory B cell defects lead to the PBC-associated hyper–immunoglobulin M,29 possibly through changes in potassium channels.30 PBC monocytes are more reactive to TLR stimuli than controls.31 Similarly, the phenotype of autoreactive CD8+ and natural killer T cells has been widely studied.32, 33 It is noteworthy, however, that most available data are obtained from peripheral blood samples and do not necessarily reflect the scenario of liver-infiltrating cells. Interestingly, autoreactive clone frequency is significantly higher in PBC livers compared with patient peripheral blood.33 For these reasons, we submit that the isolation and ex vivo use of LMNCs is a critical step in understanding the cellular interactions that contribute to the pathology of PBC and is a major strength of the present study. One limitation of our data is the use of patients with HBV and HCV as controls. A second limitation is that all patients had end-stage cirrhosis. The use of such patient material was based primarily on the availability of explanted livers, because such studies could not be performed on liver biopsies. Future studies should include healthy livers as controls, perhaps through the use of regional lymph nodes removed during transplantation.

Several additional lines of evidence obtained in the present study warrant further discussion. First, it is noteworthy that CD40 expression on BECs was significantly enhanced by autologous LMNC supernatants while not differing between PBC and control cells in the presence of TLR stimuli. This is particularly interesting since CD40 has been demonstrated as critical to induce apoptosis in BECs,25 whereas CD40 ligand (CD40L) levels were not different in PBC and control sera34 despite being recently suggested to play a pivotal role in B cell autoimmunity.35 Second, the chemokine pattern produced by PBC BECs is also similar to data in other autoimmune diseases in which the chemokine production by targeted cells contributes to the development of pathology.36, 37 Finally, our data suggest an important role of the TLR3 pathway challenge in PBC immunopathology. To this regard, it is intriguing that data from other autoimmune diseases support an anti-inflammatory effect of TLR3 stimulation.38

Our results have 2 major implications in the understanding of PBC pathogenesis. First, we demonstrated that PBC cholangiocytes are not functionally different from control cholangiocytes when stimulated with TLR ligands in terms of membrane marker expression and chemokine production, despite the reported differences in specific chemokine receptors.39 Second, PBC BECs manifest a significantly different response to TLR stimulations when cocultured with autologous LMNCs. Taken together, these data clearly suggest that BECs and LMNCs remain, respectively, the “innocent victims” and “usual suspects” in the natural history of PBC. These data therefore suggest that the etiology of PBC depends on an understanding of the relationship between autoaggressive and protective T cell responses, sex as a risk factor, and the selective genetic basis of autoimmunity.40–47

It is difficult to speculate on the primum movens of T cell tolerance breakdown and their chemotaxis to PBC liver. Data on the use of xenobiotics,16 epidemiological48 and familial49 studies, as well as new animal models50 appear to support one unifying theory. This states that tolerance to mitochondrial antigens is broken in the presence of a permissive genetic background and the right environmental mimicry on the lipoic acid residue of pyruvate dehydrogenase.51 Tolerance may be broken in the liver during the catabolism of molecules or, alternatively, in other organs or tissues and then spread to the bile duct cells, possibly based on their unique apoptotic features and their relationship with Tregs.17, 40, 52 At this point, chemokines released by injured BEC would cause the accumulation and expansion of autoreactive T clones in the portal tract and ultimately cause PBC pathology, favored by the promiscuity defects in T regulatory cells.