THE BILIARY EPITHELIUM PRESENTS ANTIGENS TO AND ACTIVATES NATURAL KILLER T CELLS

Abstract

Cholangiocytes express antigen-presenting molecules but it has been unclear whether they can present antigens. Natural killer T (NKT) cells respond to lipid antigens presented by the major histocompatibility complex (MHC) class I-like molecule CD1d and are abundant in the liver. We investigated whether cholangiocytes express CD1d and present lipid antigens to NKT cells and how CD1d expression varies in healthy and diseased bile ducts. Murine and human cholangiocyte cell lines as well as human primary cholangiocytes expressed CD1d as determined by flow cytometry and western blotting. Murine cholangiocyte cell lines were able to present both exogenous and endogenous lipid antigens to invariant (i) and non-invariant (ni) NKT cell hybridomas and primary NKT cells in a CD1d-dependent manner. A human cholangiocyte cell line, cholangiocarcinoma cell lines and human primary cholangiocytes also presented exogenous CD1d-restricted antigens to iNKT cell clones. CD1d expression was down-regulated in the biliary epithelium of patients with late primary sclerosing cholangitis (PSC), primary biliary cirrhosis (PBC) and alcoholic cirrhosis compared to healthy controls.

Conclusions

Cholangiocytes express CD1d, present antigens to NKT cells and CD1d expression is down-regulated in diseased biliary epithelium. These findings show that the biliary epithelium can activate an important lymphocyte subset of the liver. Our study describes the presence of a potentially important immune pathway in the biliary system, which may be capable of regulating inflammation in the context of biliary disease.

Introduction

Clinically important biliary diseases such as primary sclerosing cholangitis (PSC) and primary biliary cirrhosis (PBC) are characterised by inflammation and destruction of bile ducts (1,2). It is evident that cholangiocytes not only represent passive targets of the immune system, but play an active role in immune responses (3). The biliary epithelium constitutively express MHC class I molecules while MHC class II molecules are up-regulated in the biliary epithelium in several cholestatic diseases and during liver allograft rejection (4–7). In spite of this clear MHC class II expression it has proven difficult to show that the cholangiocytes can actually present peptide antigens, while their ability to present lipid antigens has not yet been investigated (8,9).

Natural killer T (NKT) cells are a subset of lymphocytes that respond to lipid antigens presented by the MHC class I-like molecule CD1d (10). These lymphocytes have functional qualities of both adaptive and innate immune cells with innate-like features that include rapid cytokine production such as interleukin-4 (IL-4) and interferon-γ (IFN-γ) following activation (11). NKT cells are divided into two main subsets, type I and type II. Type I or invariant (i) NKT (iNKT) cells are defined by their ability to recognize the glycosphingolipid α-galactosylceramide (α-GalCer) and their expression of a canonical (invariant) T cell receptor (TCR)-α chain, while type II NKT cells can not recognize this glycosphingolipid and are non-invariant (ni) in that they express a diverse array of TCR-α chains (11). NKT cells are enriched in human and murine liver (12), making them interesting in the context of biliary disease. NKT cells can play either a protective or detrimental role in different autoimmune diseases such as inflammatory bowel diseases (IBD) and diabetes (12–14), and these diseases in particular are associated with the biliary disease primary sclerosing cholangitis (PSC) (15). Expression of CD1d has been described at the hepatocyte-biliary border of hepatitis C patients, as well as by the biliary cells of primary biliary cirrhosis (PBC) patients (16,17) raising the possibility that they can activate NKT cells. Indeed, further data culminated in the Koch’s Postulate finding that a PBC-like disease can be caused by iNKT cells in model Novosphingobium infection, bacteria which produce α-GalCer-like ligands, and this was not observed in mice lacking iNKT cells (18). CD1d expression in the biliary epithelium has not been reported in other liver diseases or cholangiopathies such as PSC.

In the present study we sought to evaluate the expression of CD1d in murine and human biliary epithelium and to determine whether cholangiocytes are able to present lipid antigens to and activate NKT cells. Our data demonstrate that cholangiocytes can indeed function as antigen presenting cells, and that CD1d is down-regulated in the biliary epithelium of diseased livers. This suggests that cholangiocytes exhibit a potential regulatory function through the activation of NKT cells.

Materials and methods

Cells and cell lines

Two murine cholangiocyte cell lines were originally isolated from the small and large intrahepatic bile ducts of BALB/c mice, respectively small and large cholangiocytes. These murine cell lines and a human cholangiocyte cell line, H69, had previously been immortalized by introduction of the SV40 large T antigen (19,20). Cholangiocarcinoma cell lines EGI-1, TFK-1 (21) (DSMZ, Braunschweig, Germany) and HuCCT1 (JCRB Cell Bank, Osaka, Japan) were acquired commercially. Cholangiocarcinoma cell lines KMBC and KMCH-1 (22,23) were kind gifts from Prof Gregory Gores (Mayo Clinic Medical Center, Rochester, MN), while cholangiocarcinoma and gall bladder carcinoma cell lines Sk-ChA-1, Mz-ChA-1 and Mz-ChA-2 (24) were kindly given to us by Prof Alexander Knuth (University Hospital Zürich, Zürich, Switzerland).

Murine iNKT cell (DN32.D3, 24.7, 24.8) and niNKT cell hybridomas (14S.6, 14S.7, 14S.10 and 14S.15) and two human iNKT cell clones (JC2.7 and J3N.5) have been previously described (25–27). CD1d transfected murine RMA-S cells and an Epstein-Barr virus (EBV) cell line ectopically expressing human CD1d were used as positive controls (25). Primary murine NKT cells were extracted from the livers of C57BL/6 mice. Primary dendritic cells were isolated from the spleens of C57BL/6 mice using CD11c MicroBeads, (Miltenyi Biotec, Bergisch Gladbach, Germany). Primary cholangiocytes were extracted from explanted livers from liver recipients. Cells were cultured in conditions summarised in the Supplementary material and in Table S1.

Mice

C57BL/6 mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and Taconic Farms (Germantown, NY). The mice were housed in a Minimal Disease Unit at the animal facility at Oslo University Hospital, Rikshospitalet, Oslo. All animal experiments were approved by the Norwegian Animal Research Committee and all animals received human care in line with "Guide for the Care and Use of Laboratory Animals" (National Institutes of Health Publication, 8th Edition, 2011).

Flow cytometry

All cells were incubated with an antibody against Fc-receptors to avoid non-specific binding. Cholangiocytes were stained with anti-CD1d antibody or isotype control for 30 minutes. To detect iNKT cells, lymphocytes were stained with anti-TCR β and loaded or unloaded PBS-57 CD1d tetramers (kindly provided by the NIH Tetramer Core, Emory, GA) for an hour. Flow cytometric analysis was performed using BD FACS Calibur or a BD FACS Verse flow cytometer. The results were analyzed in FlowJo version 9.5.3 (TreeStar, Ashland, OR). Antibodies and NKT cell reagents are described in more detail in the Supplementary material.

Western blotting

Murine and human cholangiocyte cell lines and primary human cholangiocytes were analyzed for CD1d and CK19 expression by Western blotting. Detailed methods are provided in the Supplementary material.

Antigen presentation assays

To assess antigen presentation, cholangiocytes were seeded into 96 well plates, with each condition in triplicate. Exogenous antigen presentation was evaluated by loading cholangiocytes with the model antigen α-GalCer (KRN 7000) (Avanti Polar Lipids, Alabaster, AL) for 4 hours, while for endogenous antigen presentation no pre-incubation was performed. After incubation the plates were centrifuged and washed three times in medium and NKT cells/hybridomas were added. In some experiments a monoclonal anti-CD1d antibody or an isotype control were added to the wells to confirm CD1d restriction. Supernatants were collected and cytokines were measured with ELISA. All experiments were performed a minimum of three times. Detailed methods are described in the Supplementary material.

Extraction of primary lymphocytes from mice

The detailed methods have been described previously (28) and are provided in the Supplementary material.

Patient Specimens

Surgical specimens of explanted livers from 39 patients with either PSC (n=12), PBC (n=12), alcoholic cirrhosis (n=10), familial amyloidotic polyneuropathy (FAP) (n=2), colorectal cancer metastasis (n=2, non-tumorous areas, obtained after liver resection surgery), and from one healthy donor liver were used for immunohistological examinations. Patients with FAP and colorectal cancer metastasis and the healthy donor were considered to have healthy biliary epithelium, and their liver tissue was used as control material. Primary cholangiocytes were extracted from two explanted livers with PSC and non-alcoholic steato-hepatitis (NASH). Written informed consent was obtained from all study participants. Ethical approval was obtained from the research ethics committees at each participating medical centre in accordance with the declaration of Helsinki (S-0 88726, 2013/188-31/1).

Re-stimulation of human iNKT cell clones

iNKT cell clones were stimulated with irradiated peripheral blood mononuclear cells from healthy donors as described in the Supplementary material.

Extraction of primary cholangiocytes from human tissue

Primary cholangiocytes were isolated from tissue of the portal area of explanted livers of liver transplant recipients (29,30) as described in the Supplementary material.

Immunofluorescence

Sections of formalin-fixed paraffin-embedded liver tissue were stained with anti-CD1d or an isotype control and co-stained with an anti-CK19 antibody. Detailed methods are included in the Supplementary material.

Quantification of CD1d expression and image processing

For each immunofluorescently stained patient sample images were captured from two areas with large bile ducts and two areas with small bile ducts, with the exception of some samples where no large bile ducts were found. All images were acquired with identical exposure times and camera settings and the raw images were analyzed with Fiji image processing program (31). The CK19 positive cells were used to define the area for quantification so that only the CD1d expression of the CK19 positive cells was measured. The mean pixel intensity (on a scale from 0 to 255) of CD1d staining in CK19 positive cells in each slide was determined and the same analyses were performed on neighbouring sections stained with an isotype control antibody. The mean pixel intensity of the isotype control staining was subtracted from the CD1d mean pixel intensity. The mean value of the two measurements of CD1d expression in the large and small bile ducts respectively were used in further analysis. In images used for illustration the exposure time for CK19 was adjusted to get similar signal between groups, while all CD1d and isotype control images were captured with the same exposure settings. The same contrast stretching was performed on all the images shown using Adobe Photoshop CS5.

Statistical analysis

All values are presented as mean ± SEM except biochemistry values that are presented as mean ± SD or as median with minimum and maximum values where appropriate. Statistical significance was calculated with unpaired Student's t-test for comparison of two groups and one way analysis of variance for three or more groups followed by Bonferroni's multiple comparison. The statistical tests were performed using GraphPad Prism version 5.0 b (GraphPad Software, La Jolla, CA) and IBM SPSS Statistics version 21 (IBM Corporation, Armonk, NY).

Results

Murine cholangiocyte cell lines express CD1d and present lipid antigens to NKT cell hybridomas and primary NKT cells

We first explored the CD1d expression of murine small and large cholangiocyte cell lines (19). Flow cytometry of these two cell lines demonstrated that both small and large immortalized murine cholangiocytes express CD1d (Fig. 1A). CD1d expression was further confirmed by Western blotting (Fig. 1B), and the cholangiocyte phenotype ascertained by immunoblotting for the cholangiocyte marker cytokeratin 19 (CK19) (Fig. S2A).

Fig. 1.

CD1d is expressed by murine cholangiocytes. (A) Representative flow cytometry staining for CD1d expression on small and large murine cholangiocytes obtained from small and large bile ducts respectively. Grey histograms represent isotype controls. (B) CD1d expression of small and large cholangiocytes (chol.) evaluated with Western blotting. A CD1d-transfected RMA-S murine cell line was used as a positive control. GAPDH was used as a loading control.

Next, we investigated whether cholangiocytes can present lipid antigens. Interestingly, both small and large murine cholangiocytes presented exogenous lipid antigen (α-GalCer) to two out of three iNKT cell hybridomas (DN32.D3 and 24.7) as indicated by an increased release of IL-2 (Fig. 2A) and IL-13 (Fig. 2B) in the culture supernatants. In similar assays without addition of exogenous antigen, the small cholangiocyte cell line activated three out of seven iNKT and niNKT cell hybridomas (14S.6, 14S.15 and 24.8) indicating the presence of endogenous antigen(s), demonstrated by a 10-fold increase in the IL-2 production compared to the basal state. Similar results were obtained with the large cholangiocyte cell line, which activated one niNKT cell hybridoma (14S.15) (Fig. 2C). This activation was confirmed to be dependent upon CD1d by adding a monoclonal anti-CD1d antibody (19G11), which abolished IL-2 secretion (Fig. 2D).

Fig. 2.

Murine cholangiocyte cell lines present exogenous and endogenous lipid antigens to NKT cells. Antigen presentation assays with large and small cholangiocytes and invariant NKT and non-invariant NKT cell hybridomas with (A, B) and without (C) the addition of the exogenous lipid antigen α-GalCer (α-GC). Activation of hybridomas is indicated by an increased release of IL-2 and IL-13. (D) CD1d-restriction was evaluated by addition of a monoclonal anti-CD1d antibody (19G11). (E) Assays with small cholangiocytes and primary murine lymphocytes investigate CD1d-restricted antigen presentation to primary murine NKT cells extracted from the livers of mice. (F) Antigen presentation assays with small and large cholangiocytes (chol.) compared to primary dendritic cells (DCs) and CD1d-transfected cell line RMA-S and with the addition of the exogenous lipid antigen α-GalCer. Activation of hybridomas is indicated by an increased release of IL-2.

In assays using primary liver lymphocytes, murine cholangiocytes were able to present exogenous lipid antigen to the primary NKT cells, demonstrated by IL-4 secretion after 24 hours and in IFN-γ secretion after 72 hours, compared to wells with no cholangiocytes as antigen presenting cells or to wells lacking α-GalCer. To prove that this activation was CD1d-restricted, we added an anti-CD1d antibody and could again block the activation of the NKT cells and inhibit the cytokine secretion (Fig. 2E). In assays using cholangiocyte cell lines, primary dendritic cells and a CD1d transfected cell line (RMA-S), we found that the cholangiocytes activated the iNKT cell hybridomas efficiently and at a level comparable to professional antigen presenting cells (Fig. 2F).

Human cholangiocytes can activate iNKT cells

To examine the CD1d expression by human cholangiocytes we investigated the human cholangiocyte cell line H69. Flow cytometry demonstrated that H69 expressed substantial levels of CD1d (Fig. 3A). This was further confirmed by Western blots with clear bands indicating CD1d expression (Fig. 3B). The cholangiocyte phenotype of the H69 cell line was verified by expression of CK19 and EpCAM (Fig. S2B and S2C).

Fig. 3.

A human cholangiocyte cell line express CD1d and present exogenous lipid antigens to iNKT cell clones. (A) Representative flow cytometry staining for CD1d expression on the human cholangiocyte cell line H69. Grey histogram represent isotype control. (B) CD1d expression of H69 cholangiocytes evaluated with Western blotting. An EBV cell line ectopically expressing human CD1d was used as a positive control. GAPDH was used as a loading control. (C) Antigen presentation assays with H69 cholangiocytes and human iNKT cell clones J3N.5 and JC2.7 +/− exogenous antigen α-GalCer (α-GC). Activation of iNKT cell clones is indicated by increased IL-4 secretion after 24 hours and IFN-γ after 72 hours and (D) IL-13 secretion after 72 hours. (E) CD1d restriction was evaluated by addition of a monoclonal anti-CD1d antibody (51.1). (A,B,C,D) All panels are representative of at least three independent experiments.

In co-cultures we demonstrated that human cholangiocytes can present lipid antigens, and observed that H69 cholangiocytes presented exogenous lipid antigen α-GalCer to two iNKT cell clones (JC2.7 and J3N.5), indicated by increased IL-4 secretion after 24 hours and IFN-γ secretion, as well as IL-13 secretion, after 72 hours in the culture supernatants (Fig. 3C, 3D). To verify CD1d-restriction in the human assays, we added a monoclonal anti-CD1d antibody (51.1), demonstrating markedly diminished cytokine secretion (Fig. 3E).

To examine the relevance of CD1d in human disease, we first confirmed CD1d expression in human primary cholangiocytes from PSC and non-alcoholic steatohepatitis livers by Western blotting (Fig. 4A). The cholangiocyte phenotype of these primary cholangiocytes was verified by both CK19 and EpCAM expression (Fig. S2B and S2C). To assess the antigen presenting potential of the primary human cholangiocytes to NKT cells, we performed co-culture experiments with the iNKT cell clones. The primary cholangiocytes presented exogenous antigen (α-GalCer) and activated the iNKT cell clones, as indicated by increased IL-4 and IFN-γ secretion (Fig. 4B).

Fig. 4.

Primary human cholangiocytes express CD1d and present exogenous lipid antigen. (A) CD1d expression of primary cholangiocytes (chol.) from a primary sclerosing cholangitis (PSC) and a non-alcoholic steato-hepatitis (NASH) patient evaluated with Western blotting. GAPDH was used as a loading control. (B) Antigen presentation assays with primary cholangiocytes, human iNKT cell clone JC2.7 and α-GalCer (α-GC). Activation of the iNKT cell clone is indicated by increased IL-4 secretion after 24 hours and IFN-γ after 72 hours.

Human cholangiocarcinoma cell lines express CD1d and present antigens to iNKT cells

In line with the experiments performed in the murine and human cholangiocyte cell lines, we evaluated CD1d expression in eight cholangiocarcinoma cell lines by Western blotting which demonstrated that all cell lines expressed CD1d (Fig. 5A). The carcinoma cell lines were also co-cultured with two human iNKT cell clones, which demonstrated that the cholangiocarcinoma cell lines possessed the ability to present lipid antigens (Fig. 5B).

Fig. 5.

Cholangiocarcionoma cell lines express CD1d and present lipid antigens. (A) CD1d expression of eight cholangiocarcinoma cell lines evaluated with Western blotting. GAPDH was used as a loading control. (B) Antigen presentation assays with cholangiocarcinoma cell lines EGI-1 and MzChA1 and human iNKT cell clones J3N.5 and JC2.7 +/− α-GalCer (α-GC). Activation of iNKT cell lines is indicated by increased IL-4 secretion after 24 hours and IFN-γ after 72 hours.

CD1d is down-regulated in human biliary epithelium of diseased livers

We next explored whether biliary epithelium in patients with PSC, PBC and alcoholic cirrhosis and controls without known biliary disease expressed CD1d (Clinical information in Table 1). Paired immunohistochemical detection of CD1d and CK19 revealed a clear signal for CD1d in the bile ducts of all healthy control livers tested (Fig. 6), and a weaker signal in the biliary epithelium of patients with PSC, PBC and alcoholic cirrhosis (Fig. 6). Of note, CD1d was predominantly expressed on the basolateral side of the cholangiocytes. Quantitative image analysis showed that the small bile ducts of the healthy controls expressed significantly more CD1d than the small bile ducts of PSC-, PBC- and alcoholic cirrhosis patients (Fig. 7A), and the same tendency was seen in the large bile ducts (Fig. 7B). A tendency to a stronger CD1d signal was seen in the large bile ducts of PSC patients compared to PBC patients (Fig. 7B), while there were no significant differences in CD1d expression in the small bile ducts between the different disease groups (Fig. 7A) and no correlation to biochemical parameters (data not shown).

Table 1.

Clinical information on controls, primary sclerosing cholangitis-(PSC), primary biliary cirrhosis-(PBC) and alcoholic cirrhosis patients. Biochemistry values were measured at the time of liver transplantation/sampling and are presented as mean ± SD, except bilirubin that is presented as median with minimum and maximum values.

	Controls* (n=5)	PSC (n=12)	PBC (n=12)	Alcoholiccirrhosis (n=10)
Gender (M/F)	3/2	10/2	3/9	10/0
Age at sampling	57.0 (±16.7)	48.4 (±14.4)	59 (±11)	58 (±5)
IBD fraction	0/4	10/12	0/12	0/10
MELD score	6 (±1)	11 (±5)	16 (±9)	23 (±13)
Creatinin (umol/L)	67 (±4)	62 (±12)	70 (±23)	155 (±88)
Bilirubin (umol/L)	7.5 (5, 9)	32.5 (9, 326)	95 (8, 794)	37 (16, 836)
INR	1.0 (±0.1)	1.1 (±0.3)	1.3 (±0.3)	1.9 (±0.7)
ALT (U/L)	23 (±5)	150 (±122)	103 (±65)	43 (±39)
AST (U/L)	31 (±6)	135 (±82)	128 (±107)	71 (±44)
ALP (U/L)	94 (±47)	435 (±385)	331 (±207)	117 (±39)
GGT (U/L)	37 (±25)	459 (±450)	197 (±154)	94 (±75)
Albumin (g/L)	35 (±5)	37 (±7)	34 (±6)	33 (±4)
Thrombocytes (10^9/L)	261 (±34)	260 (±188)	172 (±118)	129 (±69)

^*The group of controls consists of two colorectal carcinoma patients, two FAP patients and one healthy donor. Clinical information on one of the patients in the control group was missing.

Fig. 6.

CD1d is clearly expressed basolaterally in healthy biliary epithelium and less clearly expressed in diseased biliary epithelium. Sections of liver tissue from controls without biliary disease and primary sclerosing cholangitis (PSC), primary biliary cirrhosis (PBC) and alcoholic cirrhosis (alc.cirrhosis) patients were double-stained with anti-CD1d/isotype control and anti-CK19. Merged images of sections stained with anti-CD1d and anti-CK19 at the bottom row.

Fig. 7.

CD1d is down-regulated in the biliary epithelium of patients with various liver diseases compared to controls without biliary disease. CD1d expression of (A) small bile ducts and (B) large bile ducts from controls, primary sclerosing cholangitis-(PSC), primary biliary cirrhosis-(PBC) and alcoholic cirrhosis (alc. cirrhosis) patients. CD1d expression is given as the mean pixel intensity of CD1d staining in CK19 positive cells after subtraction of the mean pixel intensity in the corresponding isotype control (arbitrary units (arb.units)).

Discussion

In the present study we demonstrate for the first time that the biliary epithelium can present lipid antigens to, and activate NKT cells via the non-polymorphic MHC homologue CD1d. We demonstrate that both murine and human cholangiocytes express the antigen-presenting molecule CD1d, and that CD1d is down-regulated in biliary epithelium in different liver diseases.

It has been unclear whether cholangiocytes can function as antigen presenting cells. In functional assays using murine and human cells it has previously not been possible to prove that these epithelial cells present peptide antigens despite the fact that they can express MHC II molecules (4,8,9). It has been suggested that the main interaction between cholangiocytes and T cells is through other molecules, such as leukocyte factor antigen-3/CD2 and CD40/CD40L (2). In the present study, we investigated whether cholangiocytes can present antigens, but in contrast to previous studies we focused on the ability to present lipid antigens to NKT cells.

We demonstrate that both murine and human cholangiocytes can present the exogenous antigen α-GalCer to iNKT cells. Although α-GalCer is a non-natural CD1d ligand activating iNKT cells, it has similar structure and properties to some bacterial lipids and is a well documented CD1d-restricted model antigen (11,32). Interestingly Bacteroides fragilis has specifically been shown to produce a novel α-galactosylceramide molecule, distinct from that derived from marine sponge (11,33) and a range of other lipid antigens have been found in bacteria that can activate iNKT cells in a similar manner as this model glycosphingolipid (34–36). We also found that murine cholangiocytes not only could present exogenous antigen, but could also present endogenous antigens in a CD1d-dependent manner, which lead to a clear, but weaker stimulation of the NKT cells. The activation of NKT cells by self antigens may be important in the case of tumor immunity and autoimmunity, but can also be important during infections, as human iNKT cells can be activated by a stimulation from weak responses to self antigens amplified by interleukin-12 (IL-12) induced by microbial pathogens (27). This implies that the antigen presentation by cholangiocytes could potentially play a role in the context of NKT cell autoreactivity and tuning of peripheral NKT cells as well as microbial infections such as bacterial cholangitis.

NKT cells have been shown to make both protective and pathogenic contributions to disease in different models of IBD (37–39), which represents the most common co-morbidity in PSC (15). Gut activated T cells in the context of IBD may contribute to biliary inflammation in PSC, and as up to 80% of PSC patients also have IBD, it is plausible that PSC shares pathogenic mechanisms with IBD (15). In addition to the expression on B cells, dendritic cells, and thymocytes, CD1d is also expressed by hepatocytes and epithelial cells such as intestinal epithelial cells (IECs) (40), and the IECs ability to present lipid antigens to NKT cells is well-established (41). IECs and their CD1d expression play an important role in intestinal inflammation, as intestinal epithelial CD1d appears to protect from intestinal inflammation through its ability to induce secretion of IL-10 when ligated by iNKT cells, while iNKT cells activated by bone-marrow-derived antigen presenting cells (APCs) contribute to inflammation (42). Thus, it is possible that similar mechanisms could also be involved in the biliary epithelium and biliary inflammation.

We found that CD1d expression is a characteristic of healthy cholangiocytes. The expression of CD1d by hepatocytes and other liver-resident cells such as Kupffer cells is well known (40,43), but the biliary expression has so far not been well-characterized. The localization of CD1d seemed to be most pronounced basolaterally on cholangiocytes. IECs have been found to present lipid antigens primarily on the basolateral surface of these cells (41), and it is likely that this is also the case for cholangiocytes. On the apical side, cholangiocytes are exposed to the lipid constituents of bile and it is possible that lipid antigens can be taken up by the cholangiocytes, presented by basolaterally localized CD1d and then potentially trigger immune reactions by NKT cells in the surrounding liver tissue. Compared to controls with normal liver parenchyma CD1d was down-regulated in biliary epithelium in PSC, PBC and alcoholic cirrhosis, which suggests that CD1d-dependent antigen presentation is affected in disease. In a study by Tsuneyama et al. investigating CD1d expression in PBC CD1d expression was only found in the biliary epithelium of early PBC patients and not in late PBC, and was not seen in healthy controls (17). Our finding of reduced CD1d expression in end-stage liver disease is overall consistent with this observation. In our study we used fluorescent antibodies for immunostaining and quantification by imaging software and technical factors could explain some of the discrepancy between the two studies.

Cholangiocyte cell lines and NKT cell hybridomas represent excellent tools to investigate interactions between APCs and NKT cells. In an attempt to model a more physiological system, we also used primary murine NKT cells and primary human cholangiocytes from explanted livers. As we have demonstrated consistent results using primary cells, as well as several different cell lines, we consider the present results to be relevant in a physiological setting. Interestingly, there was some variation in CD1d presentation activity among cholangiocytes derived from small and large bile ducts, independent of actual CD1d expression level. Of course, multiple factors that can influence antigen presentation go into all types of T cell stimulation, including processing machinery and adhesion molecule as well as cytokine expression (11,27,42). Similarly, different NKT cell hybridomas reacted differentially with different cholangiocytes tested. Such findings have previously been attributed to differences in endogenous antigens (11,13,25–27), which could also impact relative α-GalCer exchange and consequent presentation. To further explore the role of NKT cells in biliary inflammation, and to evaluate the potential for medical intervention in NKT cell-related pathways, studies in in vivo systems are required.

In conclusion, we have demonstrated a novel function of cholangiocytes with CD1d-restricted activation of NKT cells that can potentially play a role in the immune system during various disease processes in the liver.

Abbreviations

α-GC

α-GalCer, α-galactosylceramide

APC

antigen presenting cell

EBV

Epstein-Barr Virus

FAP

familial amyloid polyneuropathy

IFN-γ

interferon-γ

IL-2

interleukin-2

IL-4

interleukin-4

IL-10

interleukin-10

IL-12

interleukin-12

IL-13

interleukin-13

MHC

major histocompatibility complex

NASH

non-alcoholic steato-hepatitis

NKT

natural killer T

PBC

primary biliary cirrhosis

PSC

primary sclerosing cholangitis