Coculture of human liver macrophages and cholangiocytes leads to CD40-dependent apoptosis and cytokine secretion


  • Potential conflict of interest: Nothing to report.


In the vanishing bile duct syndromes (VBDS), primary biliary cirrhosis and chronic allograft rejection, cholangiocyte apoptosis is associated with sustained macrophage infiltration of the liver, suggesting that these cells may mediate tissue damage and contribute to bile duct destruction. We have previously reported that activation of CD40 on cholangiocytes with either soluble CD154 or cross-linking monoclonal antibody to CD40 induces apoptosis in vitro. We have now developed a novel primary human cell coculture model and used it to investigate (1) how macrophages kill cholangiocytes; (2) how paracrine cell interactions can shape the local cytokine milieu within the liver. We report that lipopolysaccharide (LPS) and interferon (IFN) induce sustained expression of CD154 on liver-derived macrophages (LDM) in vitro. Coculture of activated LDM expressing high levels of CD154 (CD40 ligand) with human cholangiocytes resulted in (1) CD40-dependent secretion of proinflammatory cytokines; (2) apoptosis of cholangiocytes that was abolished by antagonistic antibodies directed against human CD40 or human CD154. Conclusion: Macrophages are important effector cells in bile duct destruction in VBDS, and this role is dependent on CD40-mediated mechanisms. Thus activation of CD40 on cholangiocytes by activated macrophages provides a molecular mechanism to amplify chronic inflammation and bile duct destruction in liver disease. These data suggest that effective targeting strategies to antagonize CD40/CD154 may have beneficial effects in patients suffering from the VBDS. (HEPATOLOGY 2008.)

The vanishing bile duct syndromes (VBDS) are a group of diseases characterized by bile duct destruction and fibrosis. They include chronic allograft rejection (CR), graft-versus-host disease, and the chronic inflammatory biliary diseases primary sclerosing cholangitis and primary biliary cirrhosis (PBC). These diseases are associated with an inflammatory infiltrate that is believed to promote local inflammation and tissue destruction by secreting cytokines and chemokines and mediating effector mechanism that destroy bile ducts.1 However, the mechanisms of these effector responses are poorly understood. We have demonstrated that CD40, a member of the tumor necrosis factor receptor superfamily, is a critical regulator of cholangiocyte apoptosis in vitro when engaged by its soluble ligand CD154 or activating anti-CD40 antibodies. Ligation of cholangiocyte CD40 induces apoptosis via nuclear factor kappa B (NF-κB) and AP-1 dependent activation of Fas.2 The physiological importance of the CD154–CD40 pathway in cholangiocytes is emphasized by the phenotype of patients with hyperimmunoglobulin M syndrome that lack functional CD154. Such patients suffer recurrent bacterial sepsis and show increased susceptibility to hepatobiliary cancer,3, 4 and the few who survive into the 3rd decade of life have a high prevalence of liver complications.5

In PBC and CR,2, 6 CD40 is strongly expressed on inflamed bile ducts, and its ligand, CD154, is detected on CD68-positive macrophages and a subpopulation of T cells. Most models of bile duct injury in VBDS propose T cells as the main effector cells, and the role of activated macrophages is unclear. The fact that macrophages are the predominant CD154-expressing cells in inflammatory liver disease made us reevaluate this model to test the hypothesis that intrahepatic macrophages may be the dominant effector cells in mediating bile duct destruction in VBDS through CD40-dependent mechanisms.2, 6, 7

In the current study, we used a novel coculture assay using primary human cholangiocytes and macrophages to determine whether the coculture of activated liver-derived macrophages (LDM) expressing high levels of CD154 can directly induce apoptosis of primary cholangiocytes and modulate proinflammatory cytokine release, via CD40/CD154-dependent interactions.


AP-1, activator protein 1; CCL, chemokine (C-C motif) ligand; CR, chronic allograft rejection; CXCL, chemokine (C-X-C) ligand; GM-CSF, granulocyte–macrophage colony-stimulating factor; IFN, interferon; IL, interleukin; LDM, liver-derived macrophages; LPS, lipopolysaccharide; mAb, monoclonal antibody; mRNA, messenger RNA; NF-κB, nuclear factor kappa B; PBC, primary biliary cirrhosis; RPMI, Roswell Park Memorial Institute; siRNA, short interfering RNA; TNF-α, tumor necrosis factor alpha; VBDS, vanishing bile duct syndromes.

Materials and Methods

Source of Liver Tissue

Hepatectomy specimens were obtained from patients undergoing liver transplantation for end-stage liver diseases including PBC, primary sclerosing cholangitis, and CR. Cell isolation was done within 24 hours of tissue collection. Tissue procurement was done with approval (06/Q702/61) by the Local Ethics committee.

Isolation and Culture of Liver-Derived Macrophages and Cholangiocytes

For LDM isolation, finely diced liver (100 g) was incubated at 37°C in Gey's balanced salt solution containing 0.2% Pronase and 0.8 μg/mL deoxyribonuclease. The digest was filtered, washed, resuspended in Gey's balanced salt solution, layered onto a 16% isotonic Nycodenz (Axis-Shield) gradient, and centrifuged at 600g for 20 minutes. Viability was greater than 98%, using trypan blue dye exclusion. LDMs were grown on 13-mm-diameter plastic coverslips (Nunc, Denmark) in Roswell Park Memorial Institute (RPMI) 1640 medium (containing 10% v/v heat-inactivated human serum, 2 mM L-glutamine, and 100 U/mL penicillin and 100 μg/mL streptomycin). After 2 hours' incubation, nonadherent cells were washed off with sterile phosphate-buffered saline. LDMs were used after 10 days in culture.

Cholangiocytes were isolated as described2, 8 and used between passages 2 and 6. Before coculture, monocultures of LDM were assessed for CD154 expression after stimulation for 24 hours with 10 ng/mL lipopolysaccharide (LPS) and 10 ng/mL Interferon-gamma (IFN-γ). LDM seeded onto coverslips were carefully “settled out” over confluent monolayers of cholangiocytes seeded in 24-well plates. Control monoculture wells were set up where “unseeded” coverslips were placed onto the monolayer of cholangiocytes in the 24-well plate or seeded coverslips of LDMs were placed into wells without cholangiocytes. The LDM and cholangiocytes were washed in fresh medium before coculture and subsequently incubated for up to 72 hours in the presence or absence of LPS and IFN. LDM were subsequently carefully removed from the coculture wells for separation and analysis of either cell type.

Flow Cytometric Analysis of Liver-Derived Macrophages

LDMs were resuspended in fluorescence-activated cell sorting media containing 0.5 mM MgCl2, 1.0 mM CaCl2, phosphate-buffered saline, 10% fetal bovine serum, and 1 mg/mL sodium azide. Surface marker staining was first done before fixation, permeabilization, and intracellular antibody staining. Surface markers included monoclonal CD14-PE (Dako, UK), CD56-PE (Dako, UK), and CD154-PE (Santa Cruz, NM). Intracellular staining was done using monoclonal CD68-fluorescein isothiocyanate (Dako, UK). Positive staining was defined against the background obtained with isotype-matched irrelevant antibodies. Cells were analyzed on a Coulter Epics XL flow cytometer (Coulter, UK). Results were analyzed on Summit software (Dako, UK).

Isolation of RNA

Total RNA was purified from LDM using RNeasy Mini Kit (Qiagen, UK) according to the manufacturer's protocol. RNA was immediately reverse transcribed using Superscript II RNase H reverse transcriptase and random hexamers, and then stored at −20°C.

Conventional Polymerase Chain Reaction

Conventional polymerase chain reaction (PCR) was performed using a commercially available primer for CD154 (Stratagene, UK) and β-actin housekeeping gene (Alta Biosciences, UK). Samples were held at 4°C before PCR amplification according to the following cycling program: 5 minutes' denaturation (95°C), 5 minutes' annealing (60°C), 35 cycles of 1 minute 55 seconds (72°C), 1 minute 10 seconds (94°C), and 1 minute 10 seconds (60°C). Final extension was 10 minutes at 72°C, and samples were held at 4°C. After amplification, samples were electrophoresed in 2% agarose gel and stained with ethidium bromide, and photographic negatives were assessed. Negative PCR control was provided by endothelial cells. Positive controls were provided by stimulated lymphocytes (overnight with 1.5 μM ionomycin and 10 ng/mL phorbol 12-myristate 13-acetate [PMA]) and stimulated peripheral blood mononuclear cells [10 μg/mL phytohemagglutinin (PHA) (4 days), 50 μg/mL interleukin (IL)-2 (3 days), 1 ng/mL PMA, and 500 ng/mL concanavalin A (1 day)].

Inhibition of the CD154/CD40 Pathway Using Gene Silencing With Short Interfering RNA to CD154

Transfection of CD154-Specific Short Interfering RNA Into Liver-Derived Macrophages.

The following polyacrylamide gel electrophoresis purified and desalted CD154-specific short interfering RNA (siRNA) sequences were transfected at 200 nM: 5′GGAGAAAGAUCCUUAUCCUdTdT3′; 5′CAAGGACUCUAUUAUAUCUdTdT3′; 5′GGAUCCUCCUUAUGGAGAAdTdT3′ using JetsiEndo (Eurogentec, Belgium) according to the manufacturer's instruction. Cells were used in coculture experiments 72 hours after transfection. The knockdown of CD154 achieved was quantified by real-time PCR as described below. Cell viability was assessed using trypan blue staining.

Real-time PCR.

Seventy-two hours after transfection, total RNA from LDM was extracted with RNeasy Mini Kit and reverse transcribed as described. CD154 messenger RNA (mRNA) levels were assayed by real-time PCR using CD154 TaqMan Gene Expression Assays Hs00163934_m1 (Applied Biosystems) and 18S rRNA control. The reactions were carried out in MicroAmp Optical 96-well reaction plates using TaqMan Universal PCR Master Mix, and the plates were run in a 7500 Real-time PCR sequence detector systems (Applied Biosystems). The reaction mixtures were subjected to the following amplification: 50°C for 2 minutes, 95°C for 10 minutes, followed by 40 cycles at 95°C for 15 seconds and 60°C for 1 minute. Data were analyzed using the delta-delta-threshold cycle method.

Inhibition of CD154/ CD40 Interaction with Antagonistic Antibodies

The CD154/CD40 pathway blockade was also carried out using antagonistic monoclonal antibodies (mAb) to human CD40 (clone ch5D12 isotype IgG4) and antagonistic mAb to human CD154 (clone 5c8, isotype IgG2a). Both mAbs were provided by PanGenetics (Utrecht, The Netherlands) and used over a range of 1 to 40 μg/mL final concentration.

Prior to coculture, all cells were washed in RPMI 1640 media; some LDM were preincubated with anti-CD154 mAb for 1 hour, and some cholangiocytes were preincubated with anti-CD40 mAb for 1 hour. All cocultures were performed for 72 hours in RPMI 1640. In selected experiments, LPS and IFN were included in the culture medium at a final concentration of 10 ng/mL each. All experiments using the antagonistic antibodies were matched by control experiments employing the appropriate isotype-matched control immunoglobulin.

Assessment of Apoptosis

Apoptosis of cholangiocytes was quantified using morphological criteria (cell shrinkage, chromatin condensation, Councilman bodies), and in situ DNA end-labeling as described.9


Coverslips of LDM were stained by standard alkaline phosphatase anti-alkaline phosphatase technique10 using CD68 mAb (Dako, UK).

Paraffin-embedded liver tissue sections were processed by antigen retrieval as described11 and incubated with anti-CD68 mAb. Antibody was detected using the substrate Diaminobenzidine (Sigma) according to manufacturer's protocol.

Assessment of Cytokine and Chemokine Secretion

Cell culture supernatants were centrifuged to remove cell debris and stored at −80°C until analysis. Twenty-two cytokines and chemokines [tumor necrosis factor alpha (TNF-α), IL-1β, IL-5, Eotaxin, RANTES, IL-13, IL-2, CCL2, CXCL28, IFN, IL-10, IL-1α, IL-6, IL-7, IL-12p70, CCL3, IL-4, CXCL10, granulocyte–macrophage colony-stimulating factor (GM-CSF), IL-3, IL-15, and IL-12p40] were measured simultaneously in supernatant samples using Beadlyte cytokine suspension arrays multiplex enzyme-linked immunosorbent assay detection kit (Upstate, NY) according to the manufacturer's instructions and read using the Luminex100 LabMAP system (Luminex Corporation). Cytokine and chemokine concentrations were calculated by reference to the standard curve. Values were reported as median fluorescence intensity.


VBDS Are Associated with Activated Macrophages in Portal Tracts

Immunohistochemical staining of tissue sections for CD68 show periductal macrophage infiltration around inflamed bile ducts (Fig. 1A) and periductal granuloma formation (Fig. 1B) around diseased ducts in PBC. In CR, macrophages were detectable in large numbers around dysplastic bile ducts (Fig. 1C) and in perivenular areas of hepatocyte apoptosis (Fig. 1D).

Figure 1.

Identification of CD68 positive macrophages in human liver tissue using immunohistochemistry (×20 magnification). The panels in this figure demonstrate macrophage involvement in ductopenic liver disease. In all the panels, macrophages are identified with white arrows and the relevant area(s) of bile duct or hepatocyte destruction are identified with black arrows. (A) shows CD68+ve periductal macrophage infiltration around an inflamed bile duct in PBC; (B) shows the characteristic bile duct destruction with periductal granuloma formation seen in the periportal and septal stages of PBC. (C, D) show CD68+ve macrophages surrounding dysplastic bile ducts and in centrilobular areas of hepatocyte apoptosis respectively in end-stage chronic allograft rejection.

Characterization of Liver-Derived Macrophages

LDM in culture typically showed epithelioid morphology with a characteristic indistinct cell membrane (Fig. 2A), and all cells stained strongly for CD68 (Fig. 2A). Cells were gated on size and the dual expression of CD68 and CD14 (Fig. 2B), or CD68 and CD56 (Fig. 2B). Using this gating strategy, a significant proportion (22%-47%) of LDMs isolated from a variety of livers (n = 3 per “type”) were CD68+/CD14+ but all were CD68+/CD56 confirming they were macrophages and not natural killer cells (Fig. 2C). The apparent differences in percentage of CD68+/CD14+ cells isolated from different liver disease groups are not statistically significant and probably reflect varying degrees of inflammation or sampling error (due to small group size) rather than a disease-specific phenomenon.

Figure 2.

Phenotypic characterization of liver derived macrophages. (A) panel (a) shows a phase-contrast image of LDM after 10 days in culture on the plastic coverslips. They appear as population of large morphologically diverse cells of flattened epithelioid appearance. Panel (b) shows that all LDM stained positively for CD68 in situ (×40 mag.). LDM phenotype was also confirmed using flow cytometry with directly conjugated primary monoclonal antibodies in relevant combinations (CD68-FITC, CD14-PE, and CD56-PE). (B) shows the dual expression parameters on which the cells were gated. A significant portion of the CD68+ cells were CD14+ and all were CD56-ve. (C) summarizes the co-expression of CD68/CD14 and CD68/CD56 by isolates from a range of resected livers.

Expression of CD154 on Liver-Derived Macrophages Is Induced by LPS and IFN In Vitro

CD154 expression by the LDM derived from PBC and CR livers was confirmed using PCR to detect CD154 mRNA (Fig. 3). We have consistently found that there is no CD154 expression in normal liver. Isolated LDM generally expressed low but detectable levels of CD154 protein (Fig. 4A), and expression was reduced to basal levels after 5 to 7 days of in vitro culture. We therefore tested the ability of various combinations of proinflammatory cytokines to increase CD154 expression on LDM in vitro. Dose-dependent titration (data not shown) showed that 24 hours' stimulation with LPS and IFN at minimum concentrations of 10 ng/mL increased CD154 protein expression, which was detectable by immunocytochemistry (Fig. 4B). In contrast, stimulation of LDM for 24 hours using IL-4 or the mitogens PMA, PHA, and concanavalin A did not induce expression of CD154 (Fig. 4B). We confirmed the up-regulation of CD154 mRNA expression in LDM stimulated by LPS + IFN via real-time PCR (Fig. 4C). Sustained CD154 expression was dependent on continuing cytokine stimulation and was rapidly lost on removing the LPS or IFN as demonstrated by the time course studies showing that, after overnight stimulation with LPS and IFN, surface expression falls by 50% within 1.5 hours of removing the stimulating factors and to baseline levels by 12 to 24 hours (Fig. 4D).

Figure 3.

Activated liver derived macrophages express CD154 mRNA. CD154 mRNA was detected in variable levels in LDM isolated from liver tissue. This figure shows a representative PCR gel on cells isolated from end-stage CR and PBC liver tissue. Lane 1) CR liver; Lane 2) PBC liver; Lane 3) Liver endothelial cell control (routinely negative for CD154 mRNA); Lane 4) stimulated PBMNC positive control; Lane 5) Stimulated lymphocyte positive control. The corresponding lanes below show β-actin mRNA control to correct for mRNA loading. The molecular weight markers are shown on the left-hand side of the gels.

Figure 4.

Expression of CD154 on liver derived macrophages is induced by LPS and IFNγ. The data shown in (A) obtained by flow cytometry indicates low variable expression of CD154 in unstimulated LDM isolated from a variety of different livers. (B) shows representative cultures of LDM stained in situ using immunocytochemistry to detect CD154 after 24 hours without stimulation (a). (b) 24 hours with LPS and IFNγ; (c) 24 hours with IL4; (d) after 24 hours with PMA; (e) 24 hours with Con A; (f) 24 hours with PHA. These data show that only a combination of LPS and IFNγ was able to enhance LDM CD154 expression. (C) shows the confirmed alterations of CD154 expression in LDM stimulated by LPS + IFNγ by real time PCR. There was a 16-fold increase in CD154 mRNA expression following treatment of LDM with LPS + IFNγ. The results represent the mean value +/− SEM (n = 3). (D) shows flow cytometry plots quantifying CD154 expression in a time-course experiment after the withdrawal of stimulation with LPS and IFNγ. The point of withdrawal of stimulation is represented by the first flow cytometry plot and this is designated as “time zero”. We subsequently analysed CD154 expression at five additional time points. These data clearly demonstrate that withdrawal of LPS and IFNγ from LDM culture medium resulted in a rapid initial reduction in expression of CD154 with an eventual return to baseline levels within 24 hours.

Liver-Derived Macrophage-Induced CD40-Dependent Apoptosis of Cholangiocytes

Coculture of primary human cholangiocytes with LDM resulted in cholangiocyte apoptosis. Unstimulated monoculture of cholangiocytes in RPMI 1640 medium showed a baseline level of 7.5% (±0.5%) apoptosis after 72 hours in culture. In the presence of LPS and IFN alone, this increased to 20% (±0.4%). Unstimulated cocultures showed 25.5% (±1.2%) cholangiocyte apoptosis at 72 hours, whereas cocultures in the presence of LPS and IFN induced an almost 2-fold increase in cholangiocyte apoptosis to 50.5% (±1.1%) (Fig. 5A).

Figure 5.

Liver derived macrophages in coculture induce CD154/CD40 dependent cholangiocyte apoptosis. Treatment of stimulated LDM with CD154 siRNA resulted in a relatively small reduction in mRNA expression (approx. 20%). However the reduction was statistically significant and commensurate with the observed reduction in cholangiocyte apoptosis in subsequent coculture. No reduction in cholangiocyte apoptosis was observed when CD154 siRNA was substituted for the scrambled sequence control. These data are summarized in (A). Incubation with antagonistic antibodies directed against CD40 and CD154 proved to be highly successful, and in combination, completely abrogated the cholangiocyte apoptosis observed when stimulated LDM were incubated with cholangiocytes. These data are summarized in (B). (C) shows representative ISELS of cholangiocyte monoculture (a); cholangiocyte monoculture with added LPS and IFNγ (b); LDM-cholangiocyte coculture (c); LDM-cholangiocyte coculture with added LPS and IFNγ (d); LDM-cholangiocyte coculture in the presence of CD154 and CD40 antagonistic antibodies (e); LDM-cholangiocyte coculture with added LPS and IFNγ and in the presence of CD154 and CD40 antagonistic antibodies (f). The proportion of apoptotic cells were 7%, 22%, 26%, 34%, 24%, and 4%, respectively, in the experiment illustrated.

Transfection of CD154-specific siRNA into LDM was relatively inefficient, resulting in at best 20% knockdown of CD154 mRNA expression. However, when the transfected LDM were used in the coculture experiments with cholangiocytes, there was a significant reduction in apoptosis from 50.5% ±1.5% to 40.0% ±3.9% (P < 0.001) when compared with cocultures treated with scrambled siRNA (Fig. 5B). This reduction in apoptosis correlated well with the level of CD154 knockdown. However, because the overall effects were small, we proceeded to experiments employing antagonistic antibodies to CD40 and or CD154. When CD154 or CD40 antagonistic mAbs were used alone or in combination, there was almost complete inhibition of cholangiocyte apoptosis in the cocultures. In the absence of the antagonistic mAbs, a level of apoptosis of 36.2% (±2.9%) was observed. In the presence of 20 μg/mL anti-CD154 mAb, the observed apoptosis fell to 10.8% (±1.0%) and in the presence of 40 μg/mL anti-CD154 mAb to 5.7% (±1.1%). Apoptosis of 6.3% (±0.5%) was observed with the addition of 20 μg/mL anti-CD40 mAb, and the addition of 40 μg/mL anti-CD40 mAb led to a further reduction of observed apoptosis to 4.3% (±0.5%). When both mAbs were present simultaneously at a concentration of 40 μg/mL, the observed apoptosis was minimal at 3% (±0.6%) and no different from baseline levels in untreated cells, demonstrating that the LDM-induced cholangiocyte apoptosis was CD40-dependent.

Coculture of Liver-Derived Macrophages and Cholangiocyte-Induced CD154-CD40–Dependent Cytokine Secretion

We used a microbead enzyme-linked immunosorbent assay to measure secretion of multiple cytokines in the cocultures. Five cytokines were significantly (P < 0.001) reduced in coculture (IL-3; IL-12p70; IL-10; chemokine (C-C motif) ligand 3 [CCL3]; GM-CSF) (Fig. 6A); and 3 were significantly (P < 0.001) increased (chemokine (C-X-C motif) ligand 10 [CXCL10]; IL-6; and CCL2) (Fig. 6B). We then used antagonistic antibodies to determine to what extent the changes in cytokines in coculture were dependent on CD40/CD154 activation (Fig. 6C). Of the cytokines that were reduced in coculture, the levels of IL-3, IL-12p70, IL-10, and GM-CSF were further reduced (P < 0.001) by CD40/CD154 antagonism, suggesting that CD40-mediated signals were not responsible for repression of these cytokines in coculture and indeed that CD40 is involved in their induction. However, levels of CCL3 were significantly increased (P < 0.001) by CD40/CD154 antagonism, suggesting that CD40 was actively suppressing the secretion of this cytokine in coculture. The levels of IL-6 and CCL2 secreted in coculture were all reduced (P < 0.05) by CD40/CD154 antagonism, whereas CXCL10 levels were unaffected. The levels of several cytokines that did not change in coculture were nevertheless markedly reduced (P < 0.01) in coculture by antagonizing CD40/CD154, namely IL-1β; IL-2; IL-4; IL-5; IL-13; CXCL8; Eotaxin; and TNF-α, and to a lesser extent (P > 0.01) IL-1α; IL-12p40; and RANTES. Thus, CD40-dependent pathways have complex effects on cytokine secretion in cocultures.

Figure 6.

Coculture of Liver Derived Macrophages and cholangiocytes modulated cytokine secretion via CD154/CD40 interaction. LDM and cholangiocyte monoculture or coculture supernatants were collected and assayed for 23 cytokines using Multiplex Cytokine Array ELISA. To simplify the data in the Panels, we have designated the sum total of the cytokine secretion by LDMs and by cholangiocytes in monoculture in specific conditions (stimulation with LPS and IFNγ as described in methods) as the “expected level” in coculture of the two cell types under the same conditions. This would be representative of the coculture cytokine concentration assuming that it was entirely due to the additive effect of the cytokines generated by the composite cells in the cocultures under the same specific culture conditions. The measured concentration in coculture is designated as the ‘actual concentration’. The result in (A) and (B) show the effect of coculture (without CD154-CD40 antagonism) on cytokine levels, and, (C) shows the effect of CD154-CD40 antagonism on cytokine levels in coculture. The 3 panels show: cytokines whose concentrations decreased in coculture ie cytokines whose actual concentrations in coculture were less than the expected concentrations (PA): cytokines whose concentrations increased in coculture ie cytokines whose actual concentrations in coculture were greater than the expected concentrations (PB): and finally we show the effect of CD154l-CD40 antibody antagonism on cytokines levels in coculture (PC). The results represent the mean value ± SEM (n = 3).


This study provides the first direct evidence that macrophages can kill cholangiocytes directly via CD154/CD40. Our coculture model permits paracrine cell interaction, allowing CD154/CD40 interaction to take place. In chronic liver diseases, LDM are activated by proinflammatory cytokines such as IFN and bacterial products such as LPS. This is likely to modulate their phenotype and thus affect cytokine secretion, resulting in redistribution of activated macrophages to sites of inflammatory damage. We have used LDM as a broad term to define the population of cells isolated in this study because it is unclear whether these tissue-derived cells arise solely from a resident population of macrophages or also include cells derived from recruited circulating monocytes.

Several properties have been ascribed to macrophages found in inflamed tissues, including local immune regulation and the resolution of liver injury.12, 13 We have concentrated on a third function: their role as effector cells mediating liver injury directly and maintaining chronic inflammation.

The macrophage isolation technique we used was based on the method described by Heuff et al14 and gave us a high yield of viable and pure cells with typical phenotypic and morphological characteristics of macrophages in culture.15, 16 The difficulty in detaching the LDM from plastic led us to grow them on tissue culture–grade plastic coverslips.17 The coverslips proved to be easy to handle and provide a means of growing and transferring these cells into coculture systems without the need for detachment. Confluent coverslips of LDM were allowed to settle over monolayers of cholangiocytes, allowing paracrine interaction. Unseeded coverslips did not affect cholangiocyte viability or influence chemokine secretion (data not shown).

Unstimulated LDM expressed low levels of CD154, which was readily inducible in response to LPS and IFN. Consistent with the findings in T cells, where sustained expression of CD154 requires specific enhancer/maintenance signals,18 we found that CD154 expression decreases rapidly after withdrawal of LPS and IFN from the coculture medium. Thus, the CD154/CD40 system is exquisitely able to be regulated, and loss of local proinflammatory signals results in the rapid switch off of the pathway. LPS and IFN play an important role in promoting CD40-mediated interactions between LDM and cholangiocytes in the cocultures because TNF-α generated by LPS activation of LDM in the cocultures is capable of specifically increasing CD40 expression on cholangiocytes,2 and IFN itself is known to increase cholangiocyte CD40 expression.19

The primary human cell coculture system we developed allowed us to show that LDM presenting CD154 can directly cause death and inflammatory responses in CD40-bearing cholangiocytes. This suggests that liver macrophages have an important effector function in VBDS. To demonstrate the involvement of CD154-CD40 in this process, we used CD154-specific siRNA and antagonistic antibodies targeting CD154 and CD40. Transfection, though inefficient, was sufficient to suppress gene expression by 10% to 20%, and this correlated well with reduction in apoptosis observed in the siRNA-treated cells. However, the definitive confirmation of the involvement of CD40 came from antibody blocking studies in which we were able to inhibit apoptosis to below 5%, a level seen in untreated cells. Thus, CD40/CD154 is responsible for virtually all cholangiocyte apoptosis in this setting.

In support of our findings that CD40 can amplify Fas-mediated apoptosis of liver-derived epithelial cells,2, 6, 7 it has been demonstrated that CD40 ligation, in the presence of IFN, promotes Fas-dependent apoptosis of human salivary epithelial cells,20 suggesting that CD40-mediated Fas-dependent apoptosis is a common mechanism in the control of epithelial cell survival. If our model is correct, the major role for macrophages in perpetuating cholangiocyte injury and loss might explain why the VBDS respond poorly to current immunosuppressive drugs that predominantly target lymphocytes.21 We have now provided direct evidence that macrophages are not only associated with portal tracts of end-stage PBC and CR2, 6, 7 but they are also able to directly kill cholangiocytes and promote cytokine secretion in a CD154–CD40-dependent manner.

Treatment with IFN and LPS doubled the baseline level of cholangiocyte apoptosis in monoculture. One possible explanation is that IFN increases the reactivity of cholangiocytes to LPS22 via toll-like receptor 4 (TLR4), TRAF6 recruitment, and activation of downstream signalling pathways including activator protein-1 (AP-1) and NF-κB activation, resulting in increased susceptibility to apoptosis as has been described in other systems.23 Few studies have investigated the regulation of growth of human cholangiocytes, and most of these have employed immortalized cells. One study showed that the immortalized human cholangiocyte cell line H69 responded to LPS by secreting IL-6 and undergoing proliferation.24 A separate study showed that exposure of the same cell line to C. parvum infection induces downstream release of IkappaB but not activation of c-Jun N-terminal kinase (one of the kinases involved in AP-1 activation). This led to NF-κB dominating AP-1 signalling.25 A direct comparison of responses to IL-6 between the human cholangiocarcinoma cell line (SG231) and primary cholangiocytes showed the cell line to have a more sustained response to IL-6 stimulation.26 Thus, immortalized cell lines may lose their sensitivity to apoptosis and show survival responses after TLR4 ligation, whereas primary cells respond by activating downstream signaling via AP-1 instead of NF-κB, favoring apoptosis over survival. This sensitivity to apoptosis is more likely to reflect the response of non-malignant cells in vivo.

Distinct changes were induced in cytokine responses in the cocultures, and some were regulated by CD40/CD154. The mechanism responsible for modulation of cytokine secretion in the coculture might involve NF-κB activation in response to CD40 activation and also phagocytosis of apoptotic cholangiocytes by LDM. The normal response of macrophages to apoptotic cells is to down-regulate inflammatory responses mainly by secretion of transforming growth factor beta. However, the presence of “danger signals” in the form of endogenous IFN allows macrophages to override the system and mount an inflammatory response on internalizing apoptotic cells.27 Our data suggest that enhanced CD40-mediated signaling may be permitted by this response, providing further evidence that local factors determine whether the outcome of macrophage activation is resolution or perpetuation of inflammation.28, 29

Perhaps surprisingly, most cytokines were unaffected in coculture including IL-12 and TNF-α. However, the levels of 5 cytokines were reduced: IL-3, IL-4, IL-10, CCL3, and GM-CSF, and 3 were significantly increased: CXCL10, IL-6, and CCL2. The cytokines that were increased would all favor persistent inflammation.

The chemokines CXCL10 and CCL2 are associated with Th1-type inflammatory responses and recruit effector lymphocytes and monocytes respectively, thereby promoting inflammation. Cholangiocytes secrete these chemokines in response to proinflammatory cytokines.30, 31 CXCL10 can be posted on sinusoidal endothelium, where it recruits effector T cells to the parenchyma in chronic hepatitis.31 CCL2 is secreted by cytokine-stimulated cholangiocytes in vitro, and increased expression is detected on bile ducts in portal inflammation.32 CCL2 plays a major role in liver fibrogenesis by recruiting leukocytes, activating and transdifferentiating fibroblasts.33 Human cholangiocytes express and secrete functional CCL2 in response to proinflammatory signals,30 and its ability to recruit effector leukocytes supports a role in inflammatory bile duct damage.34 Cholangiocytes secrete and respond to IL-6 in an autocrine/paracrine manner.35 IL-6 is strongly implicated in chronic inflammation, and IL-6–deficient mice are resistant to several models of chronic liver damage.36 In other circumstances, IL-6 has been implicated in the limitation of hepatic injury, mediation of hepatic regeneration,37 and the regulation of cholangiocyte growth.24, 26, 38 The consequences of increased IL-6 production are thus likely to be complex and dependent on other local microenvironmental signals.

The use of antagonistic antibodies to CD40/CD154 did not affect the secretion of CXCL10 in coculture. CXCL10 transcription is dependent on IFN, and increased secretion in cocultures probably reflects IFN secretion by activated LDM, in addition to contact-dependent mechanisms mediated by CD54 and CD44, all of which can trigger IFN secretion by LDM.39 The increased secretion of CCL2 was reduced to baseline by CD40 antagonism, suggesting that CD40 activation triggers the secretion of CCL2. The transcription of CCL2 is regulated by NF-κB, whose nuclear translocation is induced by CD40 activation.7 The increased secretion of IL-6 was also prevented by CD40 antagonism, which again may be a consequence of NF-κB activation on which IL-6 secretion is dependent.40, 41 The ability of CD40/CD154 antagonism to effectively inhibit some cytokines that were not increased by coculture is likely to reflect complex networks of autocrine and paracrine regulation in the cocultures involving exogenous IFN and endogenous cytokines such as IL-10 that may themselves further regulate cytokine transcription.

Our data demonstrate that CD154-bearing LDM interact with CD40 on cholangiocytes, resulting in apoptosis of the cholangiocytes and increased secretion of cytokines involved in the development and persistence of chronic inflammation. The study provides further evidence to support the role of CD40 in liver inflammation by amplification of cholangiocyte apoptosis and the promotion of a microenvironment that supports chronic inflammation.


We thank Professor Stefan Hubscher for kindly assessing and photographing the human liver tissue immunohistochemical images used in the manuscript.