Biliary Epithelial Cell Antibodies Induce Expression ofToll-Like Receptors 2 and 3: A Mechanism for Post–

Liver Transplantation Cholangitis?Xupeng Ge,1 Mehmet Uzunel,2 Bo-Goran Ericzon,1 and

Suchitra Sumitran-Holgersson1

See Editorial on Page 878

Studies to determine the role of preformed antibodies tobiliary epithelial cells (BECs) in liver transplant rejectionshave been initiated. However, the clinical importance ofthese antibodies in the posttransplantation period stillremains to be elucidated. Reactivity to BECs isolated froma normal healthy liver was investigated in sera of 56patients before and after liver transplantation (LTX) usingflow cytometry. Functional capacity of BEC antibodieswas determined by the ability to induce expression ofToll-like receptors (TLRs) on BECs. Cytokine and che-mokine production induced by BEC antibodies was deter-mined by enzyme-linked immunosorbent assay. In all, 7patients (13%) had BEC antibodies only pre-LTX, 14(25%) only after LTX, 18 (32%) both before and afterLTX, and 17 (30%) had no detectable antibodies. Pres-ence of preformed BEC antibodies correlated with acuterejections (P < 0.03). Deposition of immunoglobulins inbile ducts was detected in biopsies of patients during rejec-tions. Significantly higher numbers of patients with post-LTX antibodies (9 of 32) developed cholangitis, comparedwith 0 of 17 without antibodies (P < 0.02). Specificitystudies indicated that these antibodies were both non–HLA- and HLA-specific. Normal BECs expressed mRNAbut not the proteins for the TLRs. However, treatmentwith F(ab�)2 fragments of BEC antibodies induced proteinexpression of TLRs 2 and 3 and significantly high produc-tion of interleukin (IL)-6, monocyte chemoattractant pro-tein (MCP)-1, macrophage inflammatory protein (MIP)-1�, epithelial neutrophil activating peptide (ENA)-78,and IL-8. In conclusion, BEC antibodies via induction ofTLR2 and TLR3 expression, as well as inflammatory cyto-kine and chemokine production may induce epithelial cellinflammatory responses to bacterial components and con-tribute to posttransplantation cholangitis. (Liver Transpl2005;11:911-921.)

Currently, hyperacute rejection is known to bemediated by preformed alloantibodies, and acute

rejection is mainly considered to be caused by alloreac-tive T cells while both humoral and cellular immunityare responsible for chronic rejection.1 The detectionand treatment of allograft rejections has been mainlyfocused on T-cell–mediated processes.2 The presenceof antibodies reactive to donor human leukocyte anti-gen (HLA) and non-HLA antigens frequently found inthe sera of recipients undergoing rejection has insti-

gated an interest in the importance of humorally medi-ated rejections.3 Furthermore, renewed interest in anti-body-mediated rejection came partly from Feucht’sdiscovery of using C4d as a marker for the detection ofhumoral rejection in kidney transplantations,4 and it isbelieved that the incidence of biopsy-confirmed acutehumoral rejections in renal transplant is about 8% to9%.5 Antibody-mediated rejection is typically unre-sponsive to conventional antirejection therapy.1 Theoutcome for acute humoral rejection is worse than foracute cellular rejection and therefore has recently beenrecognized as a major cause of allograft loss.6-8 Humoralrejection in other organ transplantations such as heart,lung, and liver is also well recognized.9-11

Liver transplantation (LTX) is considered to be thechoice of treatment for a wide variety of acute andchronic hepatic disorders. Over the years the 1-yearsurvival rate for recipients of LTX has increased toabout 90%.12 These results reflect advances in organpreservation, immunosuppression, surgical techniques,and better management of complications.12,13 How-ever, infections including cholangitis, septicemia, andpneumonia occur in more than 70% of all transplantrecipients within the first year after LTX.14 Liver allo-

Abbreviations: LTX, liver transplantation; BEC, biliary epithe-lial cell; RPTECs, renal proximal tubular epithelial cells; TLR, Toll-like receptor; HLA, human leukocyte antigen; ENA-78, epithelialneutrophil activating peptide-78; IL, interleukin; MCP-1, monocytechemoattractant protein-1; MIP-1�, macrophage inflammatory pro-tein-1�.

From the Divisions of 1Transplantation Surgery and 2ClinicalImmunology, Karolinska University Hospital-Huddinge, KarolinskaInstitute, Stockholm, Sweden.

Received January 3, 2005; accepted February 7, 2005.Supported by grants K2002-06X-14004-02B from the Medical

Research Council and Lars Erik Gelins Foundation to S.S.-H.Address reprint requests to Dr. Xupeng Ge, Department of Trans-

plantation Surgery, B56, Karolinska University Hospital-Huddinge,S-141 86 Stockholm, Sweden. Telephone: 46-8-58583988; FAX: 46-8-7743191; E-mail: xupeng.ge@cfss.ki.se

Copyright © 2005 by the American Association for the Study ofLiver Diseases

Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/lt.20420

911Liver Transplantation, Vol 11, No 8 (August), 2005: pp 911-921

graft is known to be resistant to antibody-associatedrejection,1 but less resistant to acute rejections.15

In an earlier study, we observed that preformed anti-bodies to biliary epithelial cells (BECs) are associatedwith acute rejections of liver allograft, and we discussedthe clinical significance of detecting such antibodiesprior to liver transplantation.16 As a follow-up, in thepresent study we monitored the presence of biliary epi-thelial cell reactive antibodies in the posttransplanta-tion period to evaluate their clinical relevance in liverallograft outcome.

Patients and Methods

Approval for the present study was given by the local ethicalcommittee at Huddinge Hospital, Stockholm, Sweden.

Patient Population and Immunosuppression

Fifty-six patients between January 1999 and December 2000who received a primary orthotopic LTX at Huddinge Hospi-tal from whom sera were available both before and after trans-plantation were included in this study and in a previous studypublished by us.16 The median age was 52.0 years (interquar-tile range, 46.0 to 58.5 years), and 42.9% of the patients werewomen. The indications of LTX were mainly end-stagechronic liver diseases (Table 1). Recipient-donor selection wasbased upon ABO blood group compatibility and size match-ing, but not HLA compatibility. Follow-up time for everypatient was 2 years.

One pre-LTX serum sample was taken immediately priorto LTX from each patient. A total of 219 post-LTX sera werecollected from all patients during the first 4 weeks after trans-plantation and at 3 months thereafter, and also during acuterejection episodes and in case of hospitalization. The sera werestored at �20°C.

Immunosuppressive protocol for most of the patients(n � 39) was based on Tacrolimus/FK506 (Prograf, Fujisawa,Munich, Germany), steroids, with or without mycophenolatemofetil (Cellcept, Roche, Stockholm, Sweden). For the other17 patients Cyclosporine A/CsA (Neoral, Novartis, Basel,Switzerland) based triple immunosuppression includedmycophenolate or Imuran (Upjohn, Stockholm, Sweden) wasused. Acute rejection episodes were diagnosed by clinical signsand histopathologic confirmation.17 In some rejection epi-sodes, biopsy was contraindicated or not available due tologistic reasons, but all of these patients received standardrejection treatment based on clinical and biochemical dataand responded to the treatment. The standard treatment foracute rejection episode in this study was a 1000-mg bolus doseof intravenous methylprednisolone (Solu-Medrol, Pharmacia& Upjohn Co., Stockholm, Sweden) followed by a full recy-cling of prednisolone (i.e., tapering daily dose from 200 mg to20 mg). In the case of steroid-resistant rejection, treatmentwas switched to Tacrolimus/FK506 (Prograf) for cyclospor-ine patients or treated for 7 to 14 days with OKT3 (Ortho-Biotech, Inc., Raritan, NJ).

The monitoring of bilirary tract after transplantationincluded ultrasonography, scintigraphy, secondary cholan-giogram, endoscopic retrograde cholangiopancreatography,and percutaneous transhepatic cholangiogram. The diagnosisof cholangitis was based on the presence of fever (�38.0°C),right upper quadrant pain, pathological liver function tests,and histological signs of cholangitis in liver biopsy speci-men.18 Cytological evidence (granulocytes and/or intracellu-lar bacteria) of infection in the exteriorized bile and isolationof the same microorganism in bile were used as additionalcriteria in most cases.

Isolation and Culture of BECs

BECs were isolated as previously described.16 Briefly, livertissue from a normal healthy liver was mechanically disrupted

Table 1. Indications of Patients for Liver Transplantation (LTx) and the Presence of Biliary Epithelial Cell–Reactive Antibodies

Indications for LTX No. of patients

BEC antibody–positive patients (%)

Pre-LTX (�) Post-LTX (�)Pre-LTX (�)/Post-LTX

(�)

Hepatitis C cirrhosis 11 3 (27.3) 5 (45.5) 2 (18.2)PSC 10 8 (80.0) 5 (50.0) 5 (50.0)FAP 8 2 (25.0) 3 (37.5) 1 (12.5)PBC 6 4 (66.7) 5 (83.3) 3 (50.0)Alcoholic cirrhosis 6 4 (66.7) 4 (66.7) 3 (50.0)AIH 4 1 (25.0) 2 (50.0) 1 (25.0)Hepatitis B cirrhosis 3 2 (66.7) 3 (100.0) 2 (66.7)Others 8 1 (12.5) 5 (62.5) 1 (12.5)Total 56 25 (44.6) 32 (57.1) 18 (32.1)

Abbreviations: PSC, primary sclerosing cholangitis; FAP, familial amyloidotic polyneuropathy; PBC, primary biliary cirrhosis; AIH,autoimmune hepatitis.

912 Ge et al.

and enzymatically digested. The cell mixture was seeded ontissue-culture flasks coated with 0.2% gelatin until theyreached confluence. BECs were purified by immunomagneticisolation using Dynabeads conjugated with epithelial cell spe-cific antibody, BerEp4 (Dynal, Oslo, Norway) at a concen-tration of 20 �L immunomagnetic beads per 106 to 107 cells.Mixture of the cells and beads was incubated at 4°C for 30 to40 minutes with gentle agitation. Cells coupled to the beadswere harvested by applying a magnetic particle concentrator(Dynal).

Isolated BECs were cultured under conditions as previ-ously described.19 Characterization indicated that more than95% of the cells were BECs, which were positive for cytoker-atin 19, 7, and a common cytokeratin epitope. CulturedBECs were stimulated with recombinant tumor necrosis fac-tor-alpha and interferon-gamma (20 ng/mL and 200 ng/mLrespectively; R&D Systems, Abingdon, England) overnightprior to harvesting for analysis. Control cells renal proximaltubular epithelial cells (RPTECs) were purchased from Clo-netics (BioWhittaker, San Diego, CA), and cultivated accord-ing to the instructions from the suppliers.

Detection of Antibodies

BEC/RPTEC antibodies were detected using a flow cytomet-ric assay as previously described.16 In short, 5 � 105 BEC/RPTEC were incubated with 50 �L of patient serum for 1hour at 22°C, then washed three times with phosphate-buff-ered saline. Ten microliters of mixed 1:4 diluted fluoresceinisothiocyanate (FITC) F(ab�)2 fragments of goat anti-humanIgG (Fc specific) and IgM antibodies (Jackson Immuno-Research, PA) were added and incubated at 4°C on ice in thedark for 25 minutes. The cells were washed and then analyzedon a Becton Dickinson flow cytometer (FACSorter, BectonDickinson, San Jose, CA). Fluorescence signals from 10,000cells were counted and the percentage of FITC-positive cellswas recorded. A shift in the mean fluorescence of 20 channelsin the test sample as compared to negative control was con-sidered positive.20 All sera giving a positive reaction werefurther diluted (1:100, 1:500, and 1:1000) in phosphate-buffered saline to determine the titer of the antibodies.

Specificity of BEC Antibodies

Paramagnetic microbeads coated with pooled HLA class I orII antigens (in some cases donor-specific HLA class I– orII–coated microbeads) were used for the removal of HLAantibodies from patient sera.21 The absorbed sera wereretested for binding to BECs. RPTECs were used as controltarget cells to test for tissue specificity.

Purification of IgG Antibodies from Sera ofPatients with BEC Antibodies and Preparationof F(ab�)2 Fragments

IgG fractions were isolated from the sera of all patients withcholangitis and BEC antibodies using goat antihuman IgG

agarose beads (Sigma, Munich, Germany) according to stan-dard procedure. IgG fractions were also purified from 5 LTXpatients without BEC antibodies and rejections.

For the isolation of anti-BEC IgG F(ab�)2 fragments, anImmunoPure F(ab�)2 preparation kit was used according tothe instructions of the manufacturer (PIERCE; Rockford,IL). The purity of antibody and antibody fragment prepara-tions were verified by sodium dodecyl sulfate/polyacrylamidegel electrophoresis following silver staining of the gels. Thepurified IgG F(ab�)2 fragments from patients and controlswere concentrated individually and each fraction was used ata concentration of 4 mg/mL for the following experiments.

Reverse Transcriptase-Polymerase ChainReaction

Since posttransplantation cholangitis are considered to becaused by bacterial infections, we investigated whether theIgG F(ab�)2 fractions from patients with cholangitis couldinduce the expression of Toll-like receptors (TLRs) on BECs,which may result in development of an innate immuneresponse. We therefore initially tested gene expression forTLRs 1, 2, 3, 4, and 9. For this purpose, RNA was extractedfrom approximately 1 � 106 untreated BECs, cytokine-treated BECs, and BECs treated with IgG F(ab�)2 fragmentsfrom patients with cholangitis according to the instructions ofthe Qiagen RNA Blood Mini-kit (Qiagen, Hilden, Ger-many). RNA was eluted with 30 �L of RNAse-free water. ThecDNA synthesis was performed in a total volume of 50 �Lcontaining 30 �L RNA, 1� first strand buffer (50 mmol/LTris-HCl pH 8.3, 75 mmol/L KCl, 3 mmol/L MgCl2), 105�g/mL pdN6 (Pharmacia Biosciences, Uppsala, Sweden), 1mmol/L of each dNTP (Pharmacia Biosciences), 1 mmol/LDTT, 0.48 U/�L RNasin (Promega, Madison, WI) and 4.8U/�L M-MLV RT (Invitrogen, Paisley, Scotland), at 37°Cfor 1.5 hours. The reaction was stopped by heating at 68°C for15 minutes.

PCR: 2.5 �L of cDNA was used in a 25 �L PCR-reactioncontaining 1� PCR buffer (10 mmol/L Tris-HCl, pH 8.3,50 mmol/L KCl, 1.5 mmol/L MgCl2, 0.001% gelatin), 200mol/L of each dNTP (Applied Biosystems/Roche, Branch-burg, NJ), 5% glycerol (Sigma), 100 ng/�L cresol red(Sigma), 0.03 units/�L AmpliTaq polymerase (Applied Bio-systems) and 0.5 �mol/L of each primer. Primer sequenceswere: ABL-F: 5�-CGG CTC TCG GAG GAG ACG ATGA-3�; ABL-R: 5�-CCC AAC CTT TTC GTT GCA CTGT-3�; TLR1-F: 5�-GCA ATG CTG CTG TTC AGC TC- 3�;TLR1-R: 5�-GTC AGA AGT CCA AAG CTC AGA-3�;(389 bp), TLR2-F: 5�-GTG ACC GCA ATG GTA TCTGCA-3�; TLR2-R: 5�-TGT AGA TCT GAA GCA TCAATC TC-3�; (472 bp), TLR3-F: 5�-GGA GCC AGA ATTGTG CCA GA -3�; TLR3-R: 5�- GAA TTG GCA AAGATA TCC AGT TC -3�; (321 bp), TLR4-F: 5�-CTC TCCTGC GTG AGA CCA- 3�; TLR4-R: 5�-AGA GGT GGCTTA GGC TCT GA-3�; (262 bp), TLR9-F: 5�-CGG CCT

913Significance of Biliary Epithelial Cell Antibodies

GGT GAA CTG CAA CT-3�; TLR9-R: 5�-GCA GCG CAGGCA CAG TCA T-3� (305 bp).

PCR conditions were 94°C for 4 minutes followed by 32PCR amplification cycles. The first 10 cycles were done in atwo-segment step at 94°C for 30 seconds and at 61°C for 1minute. The next 22 cycles were done in a three segment stepat 94°C for 15 seconds, 59°C for 50 seconds, and 72°C for 30seconds. Five microliters of the PCR products were run in aready-to-use PAGE system (Amersham Biosciences). Nonde-naturing, 12.5% polyacrylamide gels were run for 1.5 hoursand visualized after an automated silver staining procedure(Amersham Biosciences).

Protein Expression of TLRs on BECs afterPretreatment with BEC Antibodies

We next tested whether BECs expressed the proteins for thevarious TLRs tested. For this purpose, BECs in the fourthpassage were grown to confluence in 6-well culture plates(Falcon, Becton Dickinson). The cells were incubated over-night in the presence of medium alone or stimulated withrecombinant tumor necrosis factor-� and interferon-� at afinal volume of 2 mL.

Parallel experiments with cytokine stimulated BECs incu-bated with IgG F(ab�)2 fractions (4 mg/mL) from patientswith posttransplantation cholangitis or control patients at thesame final volume were performed. After incubation, the cellswere detached by trypsinization, washed and incubated withthe following unconjugated monoclonal antibodies anti-TLR1, -TLR2, -TLR3, -TLR4, and -TLR9 (all from eBio-science, Taby, Sweden), followed by labeling with FITC-conjugated secondary goat-antimouse antibodies (JacksonImmunoResearch Laboratories, Stockholm, Sweden). Wetested for surface as well as cytoplasmic expression of theTLRs using 0.5% saponin. Isotype control antibodies wereused as negative controls. After incubation on ice at 4°C thecells were washed and analyzed by flow cytometry.

BECs treated in the same manner as above with IgGF(ab�)2 fractions were also double stained by immunofluores-cence as previously described.22 Staining with the above-men-tioned TLR antibodies was followed by labeling with Texasred–conjugated secondary goat-anti-mouse antibodies andFITC-conjugated CK-19 (DAKO, Glostrup, Denmark).

Cytokine Production by Cultured BECsTreated with BEC Antibodies

Last, we investigated whether BEC antibodies upon bindingto BECs could induce production of various inflammatorycytokines and chemokines. For this purpose, purified IgGF(ab�)2 fractions from patients and controls diluted inmedium as stated above were added to BECs (1 � 106 cells)and the culture supernatants were collected after 72 hours,sterile filtered and kept frozen at �70°C until assayed. Thecytokines and chemokines were measured by standard sand-wich ELISA techniques using Quantikine sandwich enzyme

immunoassay from R&D Systems. Assays were performedaccording to the manufacturer’s instructions.

Liver Biopsy Staining

Liver biopsies were obtained from 6 patients with rejectionsand BEC antibodies before and after transplantation, 5patients with rejections but without BEC antibodies, 5patients with post-LTX cholangitis, and 2 healthy liver trans-plant donors. Paraffin-embedded specimens were obtainedfrom the Department of Pathology at Huddinge Universityhospital. Deparaffinized sections were enzymatically stained.After blocking with normal bovine serum, sections werestained with F(ab�)2 fragments of goat antihuman IgG andIgM antibodies (Jackson ImmunoResearch, West Grove,PA). The immunoperoxidase procedure was carried out usingVectastain Elite ABC kit (ImmunKemi, Stockholm, Sweden)as described by the manufacturers. The DAB- nickel (givesbrown/black color staining) or the Vector NovaRed kits wereused as color developers. The sections were lastly counter-stained with Mayer’s hematoxylin (Sigma) and analyzedunder a light microscope.

Statistical Analysis

Continuous outcome measures are presented in terms ofmedian and interquartile range, and differences betweengroups were assessed by Mann-Whitney test. Chi-square testwas used to compare categorical data. Fisher’s exact test wasapplied when appropriate. Log-rank test was used to comparepatient survival rates between different groups. All statisticalanalyses were done by using software STATISTICA version6.0 (StatSoft Scandinavia AB, Uppsala, Sweden), and P valuesof less than 0.05 were considered to be statistically significant.

Results

Patients

The 2-year patient survival in this selected group of 56patients was 91.1%. Thirty-five patients (62.5%) devel-oped acute rejection episodes after LTX, in which 31(88.6%) were confirmed by biopsy result. Acute rejec-tion in 5 of the patients was steroid resistant. There wasno difference in acute rejection episodes betweenFK506-based (60.0%) and cyclosporine A–based(70.6%) immunosuppression regimens (P � not signif-icant [ns]). Nine patients developed cholangitis within2 years after transplantation (times vary from 20 days toseveral months after transplantation).

Presence of BEC-Reactive Antibodies in Sera ofLTX Patients and its Clinical Correlations

In the present study, 7 patients (13%) had BEC anti-bodies only before LTX, 14 (25%) only after LTX, 18

914 Ge et al.

(32%) both before and after LTX, and 17 (30%) had nodetectable antibodies. There was no significant differ-ence in the distribution of BEC antibodies among dif-ferent diseases (Table 1).

A higher numbers of patients with BEC antibodiesprior to LTX developed acute rejections (20 of 25,80.0%) compared with 15 (48.4%) of 31 without anti-bodies (P � 0.03) (Table 2). However, no significantdifference was observed in acute rejection episodesbetween post-LTX BEC antibody–positive and –neg-ative patients (P � ns). Interestingly, 9 (9 of 32, 28.1%)patients with post-LTX BEC antibodies developedcholangitis, while BEC antibodies negative patients didnot (P � 0.02, Table 2). In all 9 patients with cholan-gitis these antibodies were already detected prior todevelopment of cholangitis as well as during the activephase of the clinical episode. Three of the 9 cholangitispatients had a bile duct anastomosis of choledocho-jejunostomy. Choledocho-choledochostomy was per-

formed in the other 6 patients (Table 3). The occur-rence of acute rejection episodes and cholangitis isshown in Figure 1 A-B.

We found that lower numbers of patients (5 of 56,8.9%) had antibodies against RPTECs compared withBECs. The presence of RPTEC antibodies did not cor-relate with acute rejections, and BEC antibodies fromfive patients crossreacted with RPTEC.

Specificities of BEC Antibodies

The majority of the immunoglobulins in BEC anti-body–positive sera were a mixture of IgG and IgM.The titer of BEC antibodies varied from lower than1:100 to higher than 1:1000. A higher BEC antibodytiter was found in most patients with rejection episodescompared with those without rejection episodes, butthe difference was not statistically significant.

We tested whether the antibodies were HLA specific

Table 2. Correlation Between BEC Antibodies and Acute Rejections and Cholangitis Among Different Patient Groups

Pre-LTX (�)/Post-LTX (�)

Pre-LTX (�)/Post-LTX (�)

Pre-LTX (�)/Post-LTX (�)

Pre-LTX (�)/Post-LTX (�)

No. of patients 7 14 18 17Acute rejections (%) 7 (100.0) 8 (57.1) 13 (72.2)* 7 (41.2)Cholangitis (%) 0 (0.0) 2 (14.3) 7 (38.9)* 0 (0.0)

* 6 patients developed both acute rejection and cholangitis after LTx.

Table 3. Some Demographic and Clinical Characteristics of Patients With Post–Liver Transplantation Cholangitis

No. IndicationsAge,yrs Sex

BEC Antibodies

Bile DuctAnastomosis

AcuteRejections

Expression ofTLR2/TLR3

Microbial InfectionsCausing Cholangitis

After LTXGraftLoss

Pre-LTX(titers)

Post-LTX(titers)

1 PSC 30 F � (1:50) � (�1:500) CJ � �/� Enterococcus faecium, Citrobacter species,Candida albicans

No

2 PSC 36 M � (1:100) � (�1:100) CJ � �/� Enterococcus faecium, Pseudomonasaeruginosa

No

3 PSC 61 M � (1:100) � (�1:1000) CJ � �/� Candida albicans, Pseudomonas aeroginosa Yes*4 PBC 58 F � (1:50) � (�1:500) CC � �/� Enterococcus species No5 Alc 46 M � (1:50) � (�1:100) CC � �/� Enterococcus faecium, Candida albicans,

Saccharomyces cerevisiae, LactobacillusYes*

6 Alc 60 M � (1:50) � (�1:100) CC � �/� Enterococcus species, Candida albicans,Candida glabrata

No

7 FAP 30 F � (1:20) � (�1:500) CC � �/� Enterococcus faecium, coagulase-negativestaphylococci

Yes*

8 HepC 56 M � � (�1:1000) CC � �/� Not available Yes*9 Acute LF 54 F � � (�1:500) CC � �/� Candida albicans, Stenotrophomonas

maltophiliaNo

Abbreviations: PSC, primary sclerosing cholangitis; PBC, primary biliary cirrhosis; Alc, alcoholic cirrhosis; FAP, familial amyloidotic polyneuropathy;HepC, hepatitis C cirrhosis; Acute LF, acute liver failure; CJ, choledocho-jejunostomy; CC, choledocho-choledochostomy.*Graft loss is due to severe cholangitis (1), arterial thrombosis (1), arterial pseudoaneurysm (1), and sepsis and renal failure (1).

915Significance of Biliary Epithelial Cell Antibodies

by using absorption assays with HLA class I and class IIantigen-coated magnetic beads. Our results demon-strated that in 8 patients reactivity against BECs wascaused by HLA antibodies, and 4 (50%) developedcholangitis. In 7 patients reactivity against BECs waspartially absorbed by HLA antigen-coated beads, indi-cating presence of HLA and non-HLA antibodies. In24 patients no demonstrable HLA antibodies weredetected indicating presence of only non-HLA antibod-ies (Fig. 2).

Gene and Protein Expression of TLRs on BECs

We first tested whether BECs expressed mRNAs forTLRs 1, 2, 3, 4, and 9 and found that unstimulated andcytokine-stimulated BECs expressed mRNAs for all theTLRs tested. (Fig. 3A). We next tested for the proteinexpression of the various TLRs on untreated and cyto-kine-treated BECs and found no constitutive or cyto-kine-induced surface or intracellular expression of thesemolecules (Table 4, Fig. 3B). However, treatment ofBECs with anti-BEC IgG F(ab�)2 fragments frompatients with post-LTX cholangitis induced surface

expression of TLR-2 and -3 on cytokine-stimulated butnot unstimulated BECs (Table 4, Fig. 3B). No expres-sion of any of the TLRs was observed with IgG F(ab�)2

fractions from control patients (Table 4, Fig. 3B).

BEC- Reactive Antibodies Induced Productionof Inflammatory Cytokines and Chemokines

Anti-BEC IgG F(ab�)2 from cholangitis patientsinduced BECs to produce significantly higher levelsof IL-6 and the chemokines IL-8, MCP-1 (monocytechemoattractant protein-1), MIP-1� (macrophageinflammatory protein-1 alpha), and ENA78 (epithe-lial neutrophil activating peptide-78) at 72 hourscompared with F(ab�)2 from controls (P � 0.005)(Table 5).

In Vivo Deposition of Immunoglobulins onBECs during Rejection Episodes andCholangitis

No IgG or IgM deposition in bile ducts of normal liversor patients without BEC antibodies was observed (Fig.

Figure 1. Number of acute rejection and cholangitis episodes in LTX patients. (A) Significantly higher numbers ofpatients with BEC-reactive antibodies present before LTX had acute rejections compared with those without anydemonstrable antibodies (P < 0.03). (B) Significantly higher numbers of patients with BEC-reactive antibodies presentafter liver transplantation had posttransplantation cholangitis compared with those without any demonstrable antibodies(P < 0.02).

Figure 2. Specificities of BEC-reactive antibodies. No correlation of the presence of HLA or non-HLA antibodies reactivewith BECs was observed with acute rejections (P � ns) or cholangitis (P � ns). *Some patients had both cholangitis andacute rejections.

916 Ge et al.

4A-C). However, intense binding of IgG and IgM toBECs (black/brown staining) during rejections wasobserved in livers of patients with BEC antibodies andrejections (Fig. 4D-F). Binding of immunoglobulins(red staining) to BECs was also observed in patientsduring episodes of cholangitis (Fig. 4G-I).

Discussion

In our previous study we reported that preformed BECantibodies are associated with acute rejections.12 In thepresent study we observed that the presence of HLA ornon-HLA reactive BEC antibodies in the posttrans-

Figure 3. Gene and protein expression of TLRs in BECs. (A) BECs expressed mRNAs for TLRs 1, 2, 3, 4, and 9. ABL wasused as housekeeping gene. No obvious difference in amounts of gene expression was observed between unstimulated(lane 1) BECs, cytokine-stimulated (lane 2) BECs, or BECs treated with IgG F(ab�)2 fragments from LTX patients withcholangitis (lanes 3-5). (B) Flow cytometric analysis demonstrated that unstimulated BECs, cytokine-stimulated BECs,and BECs treated with IgG F(ab�)2 fragments from LTX patients without BEC antibodies did not express the proteins forthe tested TLRs. However, BECs treated with IgG F(ab�)2 fragments from LTX patients with cholangitis and BECantibodies showed induced expression of surface proteins for TLRs 2 and 3. (C) Immunofluorescent-positive staining forTLR2 (red, Texas red) and 3 (red, Texas red) on BECs treated with IgG F(ab�)2 fragments from two different liver-transplanted patients with cholangitis. The cells were double stained for cytokeratin 19 (green, FITC). The nucleus isstained blue (DAPI). In some areas the TLRs and CK19 were colocalized (yellow).

Table 4. Expression of Toll-Like Receptors on Biliary Epithelial Cells as Determined by Flow Cytometric Analysis

Antibodies toUnstimulated

BECsTNF-alpha & IFN-gamma

stimulated BECs

BECs � F(ab�)2 Ig fractionfrom cholangitis patients

n � 9

BECs � F(ab�)2 Ig fractionfrom control patients

n � 5

TLR1 � � � �TLR2 � � �� �TLR3 � � �� �TLR4 � � � �TLR9 � � � �

917Significance of Biliary Epithelial Cell Antibodies

plantation period was significantly associated withoccurrence of cholangitis. Furthermore, binding ofthese antibodies to BECs was found to induce expres-sion of TLR2 and 3 and production of inflammatory

cytokines and chemokines such as IL-6, IL-8, MCP-1,MIP-1� and ENA78, all of which are important inrecruitment of inflammatory cells such as neutrophils,monocytes, macrophages, and cytotoxic T cells. The in

Table 5. Cytokine Production by Biliary Epithelial Cells After IgG F(ab�)2 Fraction Treatment

BEC supernatantIL-1�

pg/mLIL-6

pg/mLIL-8

pg/mLIL-10pg/mL

IL-12pg/mL

IFN-�pg/mL

TNF-�pg/mL

ENA-78pg/mL

MIP-1�

pg/mLMCP-1pg/mL

Untreated (n � 7) �30 �30 �30 �30 �30 �30 �30 �30 �30 �30IgG F(ab�)2 (from cholangitis patients)

treated (n � 9)�30 651 (614–659) 10659 (7370–11013) �30 �30 �30 �30 292 (150–371) 159 (123–348) 5002 (4375–9786)

IgG F(ab�)2 (from control patients)treated (n � 5)

�30 �30 �30 �30 �30 �30 �30 �30 �30 �30

P value NS �0.005 �0.005 NS NS NS NS �0.005 �0.005 �0.005

NOTE. Values are shown as median (interquartile range).

Figure 4. Immunohistochemical detection of immunoglobulin deposition in liver biopsy sections. No IgG or IgMdeposition in bile ducts of normal livers (A) or patients without BEC antibodies (B-C) was observed. (D-F)However,intense binding of IgG and IgM to BECs (black/brown staining) during rejections was observed in livers of patients withBEC antibodies and rejections. Some positive staining of sinusoidal cells was also observed. Results from three differentpatients are shown. (G-I), Binding of immunoglobulins (red staining) to BECs was also observed in patients duringepisodes of cholangitis. Results from three different patients are shown. The sections were counterstained with hematox-ylin. Original magnifications �60 (A-D), and �40 (E-I).

918 Ge et al.

vivo deposition of immunoglobulins in bile ducts dur-ing acute rejections and cholangitis further suggests arole for BEC antibodies in facilitating these processes.Unfortunately, detection of TLRs in bile ducts duringepisodes of cholangitis could not be demonstrated,since the TLR antibodies were not functional in paraf-fin embedded biopsy sections. In our earlier study16 wereported that HLA antibodies were significantly associ-ated with cholangitis. In the present study, no differ-ence between the association of either HLA or non-HLA BEC antibodies with cholangitis was observed.The disparity may be explained by the fact that theHLA antibodies are not donor specific. Thus, the exactrole of donor-specific HLA antibodies in cholangitisremains to be elucidated.

Cholangitis after transplantation is believed to becaused by ascending bacterial infection often in combi-nation with an obstructed biliary tree. The pathogenicmechanism is unclear, and systemic sepsis may not nec-essarily occur. Prerequisite conditions are the presenceof microorganisms in the bile and increased biliary pres-sure. Bacteria that commonly cause cholangitis areEscherichia coli, Klebsiella, Enterococcus, Enterobacter,Pseudomonas, and anaerobes.23 Bacterial pathogens arewidely recognized as inflammatory stimulants.24,25 Theinnate immune system recognizes bacterial pathogensthrough a family of receptors called TLRs. TLRs pro-vide the elusive link between recognition of invadingorganisms and development of the innate immuneresponse.

In the present study, we demonstrate that BEC anti-bodies induce expression of two TLRs on BECs,namely, TLR2 and TLR3. TLR2 responds to a numberof bacterial products from both gram-positive andgram-negative organisms, including peptidoglycans,lipoteichoic acid, and mycoplasmal proteins. Fungi arealso ligands for TLR2.26 TLR3 has shown to mediatethe response to single- and double-stranded viralRNA.24 TLR3 activation leads to an induction of anti-viral and proinflammatory cytokines and dendritic cellmaturation.27 Bile culture results from the cholangitispatients showed that the majority of microorganisms inthe bile during cholangitis episodes were gram-positivebacteria and fungi. In a few cases gram-negative bacteriawere also found. This spectrum fits well with the ligandsfor TLR2. The ligand for TLR3 is double-stranded viralRNA. Since it is rather tedious and difficult to isolatevirus from bile, the role of virus in causing posttrans-plantation cholangitis still needs to be further studiedeven though some correlations have been reported.28

On the other hand, TLR4 and TLR9 protein expres-sion were not induced on BECs by BEC antibodies in

this study. The ligands for TLR4 are gram-negativelipopolysaccharides and TLR9 binds to bacterialDNA.29 The lack of TLR4 and TLR9 expression onBECs in our study might indicate lack of the appropri-ate stimulating ligands or alternatively a limited role forthese rerceptors in post-LTX cholangitis or involve-ment of mechanisms other than BEC antibodies.

The production of specific cytokines influences localimmunity, including neutrophil influx and transmigra-tion across the epithelial cell barrier, which is necessaryfor bacterial clearance.30 Thus, BEC antibody-medi-ated enhancement of IL-6, IL-8, ENA-78, MIP-1� andMCP-1 production, as well as TLR2 and TLR3 expres-sion on BECs, suggest that BECs may play an impor-tant role in the local immune response, which in turnmay be an important determinant of clinical outcome.Since BEC antibodies induce TLR2 and TLR3 expres-sion, it is possible that the induced responsiveness ofepithelial cells to bacterial products is the result ofenhanced or sustained signaling via newly expressedTLR2 and TLR3. This is supported by the finding thatin all 9 patients with cholangitis BEC antibodies wereconsistently found prior to and persisted long aftertreatment for cholangitis (as shown by samples takenbefore, during and after cholangitis episode). The factthat 7 of the 9 patients had recurrent cholangitis (daysor months after the first cholangitis episode) also sup-ports the existence of a sustained inflammatoryresponse. In addition, these patients may be at a risk forsubsequent tissue scarring or chronic rejections viathese sustained inflammatory responses. It is wellknown that bile duct anastomosis by choledocho-jeju-nostomy is a risk factor for cholangitis after transplan-tation.31 However, this anastomosis was performedonly in 3 of the 9 cholangitis patients with positive BECantibodies. Arterial thrombosis, which is another riskfactor for cholangitis, occurs only in one patient. Thus,BEC antibodies may be an additional risk factor fordevelopment of cholangitis after transplantation. Theresponse to bacterial pathogens via pattern recognitionreceptors on epithelial cells is important for innateimmune defense but their amplified activation may leadto persistent epithelial inflammation.

To our knowledge, this is the first study that dem-onstrates the ability of antibodies to upregulate TLR2and TLR3 expression and modulate chemokine secre-tion in BECs. The induction of TLR expression byantibodies may provide a critical role for antibodies inepithelial cell response to microbial pathogens and acti-vation of inflammatory responses. Not all patients withBEC antibodies in the posttransplantation period hadcholangitis. This may be explained by the fact that

919Significance of Biliary Epithelial Cell Antibodies

antibodies are heterogenous with respect to specificities,affinity, titers, and isotypes. In addition, we found somepatients with detectable antibodies only in the pre- butnot posttransplantation period. These antibodies werefound to be significantly associated with acute rejec-tions but not cholangitis. The lack of antibody detec-tion in the posttransplantation serum may not neces-sarily mean the loss/disappearance of antibodies. Thepossibility remains that the antibodies may be bound toBECs in the graft during rejections (as demonstrated inthe present study) and therefore may not be found incirculation. Furthermore, the specificities of the anti-bodies associated acute rejections may be different fromthose facilitating cholangitis.

The results presented here unravel one of the mech-anisms by which BEC antibodies induce epithelial cellsensitivity to TLR2 and TLR3 ligands leading tocholangitis. Therefore, BEC antibodies may play a rolein the pathogenesis and clinical course of bacterialcholangitis. Even though we here have, presentedresults using BECs isolated from one liver, we have overthe years prepared a panel of primary BEC lines fromdifferent healthy liver transplant donors. We are able toconfirm the present results with all the primary BEClines (n � 5) at our laboratory. The results of thepresent study warrant further investigations, whichinclude larger number of patients to evaluate the role ofBEC antibodies in post-LTX cholangitis.

In conclusion, the results suggest that patients whohave BEC antibodies in the posttransplantation periodmay be at a risk for persistent and amplified inflamma-tory responses to bacterial pathogens. Plasmapheresisand other B-cell– targeting therapies, which have beenshown to be effective in blood group ABO-incompati-ble kidney transplant,32 may be considered for BECantibody–positive liver transplant patients.

Acknowledgments

The authors wish to thank Silvia Nava for help with immu-nohistochemical staining and Dr Grzegorz Nowak for theliver biopsy specimens.

References1. Michaels PJ, Fishbein MC, Colvin RB. Humoral rejection of

human organ transplants. Springer Semin Immunopathol 2003;25:119-140.

2. Mor E, Solomon H, Gibbs JF, Holman MJ, Goldstein RM,Husberg BS, et al. Acute cellular rejection following liver trans-plantation: clinical pathologic features and effect on outcome.Semin Liver Dis 1992;12:28-40.

3. Terasaki PI. Humoral theory of transplantation. Am J Transpl2003;3:665-673.

4. Feucht HE, Schneeberger H, Hillebrand G, Burkhardt K, WeissM, Riethmuller G, et al. Capillary deposition of C4d comple-ment fragment and early renal graft loss. Kidney Int 1993;43:1333-1338.

5. Crespo M, Pascual M, Tolkoff-Rubin N, Mauiyyedi S, CollinsAB, Fitzpatrick D, et al. Acute humoral rejection in renal allo-graft recipients: I. Incidence, serology and clinical characteristics.Transplantation 2001;71:652-658.

6. Halloran PF, Schlaut J, Solez K, Srinivasa NS. The significanceof the anti-class I response. II. Clinical and pathologic features ofrenal transplants with anti-class I-like antibody. Transplantation1992;53:550-555.

7. Yard B, Spruyt-Gerritse M, Claas F, Thorogood J, Bruijn JA,Paape ME, et al. The clinical significance of allospecific antibod-ies against endothelial cells detected with an antibody-dependentcellular cytotoxicity assay for vascular rejection and graft lossafter renal transplantation. Transplantation 1993;55:1287-1293.

8. Lobo PI, Spencer CE, Stevenson WC, Pruett TL. Evidence dem-onstrating poor kidney graft survival when acute rejections areassociated with IgG donor-specific lymphocytotoxin. Transplan-tation 1995;59:357-360.

9. Hammond EH, Yowell RL, Nunoda S, Menlove RL, RenlundDG, Bristow MR, et al. Vascular (humoral) rejection in hearttransplantation: pathologic observations and clinical implica-tions. J Heart Transpl 1989;8:430-443.

10. Magro CM, Klinger DM, Adams PW, Orosz CG, Pope-HarmanAL, Waldman WJ, et al. Evidence that humoral allograft rejec-tion in lung transplant patients is not histocompatibility antigen-related. Am J Transpl 2003;3:1264-1272.

11. Adams DH, Hubscher SG, Burnett D, Elias E. Immunoglobu-lins in liver allograft rejection: evidence for deposition and secre-tion within the liver. Transpl Proc 1990;22:1834-1835.

12. Pirenne J, Koshiba T. Present status and future prospects in livertransplantation. Int Surg 1999;84:297-304.

13. Goddard S, Adams DH. New approaches to immunosuppres-sion in liver transplantation. J Gastroenterol Hepatol 2002;17:116-126.

14. Kibbler CC. Infections in liver transplantation: risk factors andstrategies for prevention. J Hosp Infect 1995;30(Suppl):209-217.

15. Tippner C, Nashan B, Hoshino K, Schmidt-Sandte E, AkimaruK, Boker KH, Schlitt HJ. Clinical and subclinical acute rejectionearly after liver transplantation: contributing factors and rele-vance for the long-term course. Transplantation 2001;72:1122-1128.

16. Ge X, Ericzon BG, Nowak G, oHrstrom H, Broome U, Sumit-ran-Holgersson S. Are preformed antibodies to biliary epithelialcells of clinical importance in liver transplantation? Liver Transpl2003;9:1191-1198.

17. Klintmalm GB, Nery JR, Husberg BS, Gonwa TA, Tillery GW.Rejection in liver transplantation. Hepatology 1989;10:978-985.

18. Barkholt L, Ericzon BG, Tollemar J, Malmborg AS, Ehrnst A,Wilczek H, Andersson J. Infections in human liver recipients:different patterns early and late after transplantation. Transpl Int1993;6:77-84.

19. Bo X, Broome U, Remberger M, Sumitran-Holgersson S.Tumour necrosis factor alpha impairs function of liver derived Tlymphocytes and natural killer cells in patients with primarysclerosing cholangitis. Gut 2001;49:131-141.

20. Xu B, Broome U, Ericzon BG, Sumitran-Holgersson S. Highfrequency of autoantibodies in patients with primary sclerosing

920 Ge et al.

cholangitis that bind biliary epithelial cells and induce expressionof CD44 and production of interleukin 6. Gut 2002;51:120-127.

21. Sumitran-Karuppan S, Moller E. The use of magnetic beadscoated with soluble HLA class I or class II proteins in antibodyscreening and for specificity determinations of donor-reactiveantibodies. Transplantation 1996;61:1539-1545.

22. Holmen C, Stjarne P, Sumitran-Holgersson S. Heterogeneity ofhuman nasal vascular and sinusoidal endothelial cells from theinferior turbinate. Am J Respir Cell Mol Biol 2005;32:18-27.

23. Lum DF, Leung JW. Bacterial Cholangitis. Curr Treat OptionsGastroenterol 2001;4:139-146.

24. Janssens S, Beyaert R. Role of Toll-like receptors in pathogenrecognition. Clin Microbiol Rev 2003;16:637-646.

25. Han DC, Huang GT, Lin LM, Warner NA, Gim JS, Jewett A.Expression of MHC Class II, CD70, CD80, CD86 and pro-inflammatory cytokines is differentially regulated in oral epithe-lial cells following bacterial challenge. Oral Microbiol Immunol2003;18:350-358.

26. Wetzler LM. The role of Toll-like receptor 2 in microbial diseaseand immunity. Vaccine 2003;21 Suppl 2:S55-60.

27. Matsumoto M, Funami K, Oshiumi H, Seya T. Toll-like recep-

tor 3: a link between toll-like receptor, interferon and viruses.Microbiol Immunol 2004;48:147-154.

28. Martelius T, Krogerus L, Hockerstedt K, Bruggeman C, Laut-enschlager I. Cytomegalovirus infection is associated withincreased inflammation and severe bile duct damage in rat liverallografts. Hepatology 1998;27:996-1002.

29. Abreu MT, Arditi M. Innate immunity and toll-like receptors:clinical implications of basic science research. J Pediatr 2004;144:421-429.

30. Schaefer TM, Desouza K, Fahey JV, Beagley KW, Wira CR.Toll-like receptor (TLR) expression and TLR-mediated cyto-kine/chemokine production by human uterine epithelial cells.Immunology 2004;112:428-436.

31. Takano K, Iwafuchi M, Uchiyama M, Yagi M, Matsukawa T,Ueno A. Studies on intestinal motility and mechanism of cholan-gitis after biliary reconstruction. J Pediatr Surg 1989;24:1225-1231.

32. Reed EF, Hong B, Ho E, Harris PE, Weinberger J, Suciu-Foca N.Monitoring of soluble HLA alloantigens and anti-HLA antibodiesidentifies heart allograft recipients at risk of transplant-associatedcoronary artery disease. Transplantation 1996;61:566-572.

921Significance of Biliary Epithelial Cell Antibodies