Potential conflict of interest: Nothing to report.
Progressive familial intrahepatic cholestasis type 2 (PFIC-2) is caused by mutations of the bile salt export pump (BSEP [ABCB11]), an ATP-binding cassette (ABC)-transporter exclusively expressed at the canalicular membrane of hepatocytes. An absence of BSEP from the canalicular membrane causes cholestasis and leads to liver cirrhosis, which may necessitate liver transplantation in early childhood. We report on the first case of a child with PFIC-2 suffering from repeated posttransplant recurrence of progressive intrahepatic cholestasis due to autoantibodies against BSEP. These antibodies occurred after transplantation and were detected in the patient's serum and at the canalicular membrane of two consecutive liver transplants. The antibodies were reactive toward the first extracellular loop of BSEP, were of high affinity, and inhibited transport activity of BSEP, thus causing severe cholestasis. The patient had three homozygous, missense changes in the BSEP gene. Their combination resulted in the complete absence of BSEP, which explains the lack of tolerance, a prerequisite of autoantibody formation toward BSEP. The findings illustrate a novel disease mechanism due to a new class of functionally relevant autoantibodies resulting in cholestasis and subsequent liver failure. (HEPATOLOGY 2009;50:510–517.)
Chronic cholestasis represents the major cause of severe liver disease in childhood which eventually necessitates liver transplantation. Biliary atresia and three types of progressive familial intrahepatic cholestasis (PFIC) are most frequently found in affected children. Progressive familial intrahepatic cholestasis type-1 (PFIC-1) develops in children with mutations in the ATP8B1 (adenosine triphosphatase, class I, type 8B, member 1) gene, which encodes the familial intrahepatic cholestasis (FIC-1) gene product, a class IV P-type adenosine triphosphatase.1 Within the liver, FIC-1 is localized at the canalicular membrane of parenchymal cells where it mediates the transport of phosphatidylserine from the outer to the inner leaflet of the canalicular membrane.2 Apart from cholestasis, PFIC-1 is associated with extrahepatic symptoms such as diarrhea, rickets, and growth retardation.3 Although similar in clinical presentation and laboratory findings, PFIC-2 is caused by mutations of the bile salt export pump (BSEP), which is encoded by the ABCB11 gene.4 BSEP belongs to the subfamily B of ATP-binding cassette (ABC) transporters. It is exclusively expressed in hepatocytes. An absence of BSEP results in the impairment of bile salt excretion into bile with consequent serum bile salt elevation, pruritus, and liver disease. On histological examination, giant cell hepatitis is often present. Extrahepatic symptoms are not features of PFIC-2. However, in milder forms of BSEP disease, the so-called benign recurrent intrahepatic cholestasis type 2, gallstones are a common finding.5 The third type, PFIC-3, is caused by mutations of the multidrug resistance protein 3 (MDR3).6 MDR3 is encoded by the ABCB4 gene and functions as a phospholipid floppase.7 An important difference between PFIC-1/PFIC-2 and PFIC-3 is an elevated gamma glutamyl transferase (γGT) level in the latter type, which has been attributed to high concentration of free bile acids in PFIC-3 bile due to reduced micelle formation.
Liver transplantation represents the ultimate therapeutic option for many end-stage liver diseases, such as progressive cholestasis caused by BSEP mutations (PFIC-2). The course after liver transplantation and the patient's prognosis depend on numerous factors, including graft rejection, biliary or vascular complications, infections, and side effects of immunosuppressive drugs. In contrast to adult liver transplantation, recurrence of the underlying disease is an uncommon event in pediatric liver transplantation.8, 9 In children, de novo autoimmune hepatitis (AIH) without a history of pretransplant AIH has been recognized as a rare cause of late graft failure.10–12De novo AIH has been correlated to the appearance of autoantibodies such as anti–smooth muscle actin and anti-nuclear13, 14 antibodies and usually responds well to standard immunosuppression.12, 15 Even without overt AIH, these antibodies occur in 26%-74% of children after liver transplantation.13, 14 Antibody-mediated rejection due to development of anti–donor-specific human leukocyte antigen (HLA)-antibodies has been recognized as a cause of graft failure.16 In posttransplant patients with a vanishing bile duct syndrome, HLA class I antibodies are found in about 50%. However, the pathophysiological relevance of these antibodies is unclear.16
A girl developed end-stage liver disease due to PFIC-24 with low γGT levels. Compared to the reference sequence NM_003742 of BSEP (ABCB11), three homozygous missense changes were found: 1331T→C (V444A),17 2453A→T (Y818F), and 2944G→A (G982R)4 (start codon numbered as “1”). The healthy parents were consanguinous and both were heterozygous for all three changes described above. The patient had two sisters who likewise suffered from PFIC-2, with both of them having the same mutations. They were successfully transplanted in early childhood.
A partial biliary diversion was performed at the age of 15 months without clinical benefit and the patient required a first liver transplantation at the age of 42 months. Immunosuppression was induced by 10 mg of the interleukin-2 receptor antibody basiliximab, 30 mg of prednisolone (3 mg/kg body weight per day), and 2× 1 mg of the calcineurin inhibitor tacrolimus. The dose of prednisolone was reduced consecutively and tacrolimus was given according to blood levels. However, primary nonfunction of the transplanted liver developed with low clotting factors and serum aminotransferases above 4000 IU/L. Histopathology revealed a severe perfusion injury with massive necrosis, hypoxic fatty degeneration, and cholestasis but without signs of acute rejection. There was no evidence for an acute viral hepatitis. An early retransplantation was required after 17 days (Fig. 1A). At this occasion, immunosuppression was conducted by the use of 10 mg of basiliximab, 15 mg of prednisolone (1.5 mg/kg body weight), and 2× 0.1 mg of tacrolimus.
The patient developed a heparin-induced thrombocytopenia, but otherwise the second transplantation was successful and the patient was discharged. Within 1 year of the second transplantation, the patient again developed symptoms of cholestatic liver disease with pruritus and elevated plasma bile salts concentrations, mimicking a PFIC-2 phenotype. Chronic rejection and other causes of liver disease such as viral hepatitis or vascular or bile duct injury were excluded, and histopathology showed liver fibrosis, high-grade cholestasis, and multinuclear hepatocytes.
A third liver transplantation was necessary 17 months after the first two transplantations. At this time, γGT was normal at 11 U/L. Cholestasis with low γGT levels is typical of PFIC-1 and PFIC-2, but it is unusual in the common causes of cholestatic graft failure. Immunosuppression was achieved by 10 mg of basiliximab, 9 mg of prednisolone (1 mg/kg body weight), and 2× 1.5 mg of tacrolimus.
Again, symptoms of cholestasis and liver failure recurred within 5 months while γGT levels remained normal. Plasmapheresis (PP) followed by five courses of rituximab treatment resulted in a decrease in serum bile salt concentrations without resolution of symptoms. Thereafter, immunosuppression was changed to sirolimus and mycophenolate or sirolimus alone. Both regimens had no benefit on cholestasis, indicating that cholestasis was not a side effect of tacrolimus. Extracorporal liver support therapies (Molecular Adsorbents Recirculating System [MARS]) were carried out to alleviate symptoms of progressive cholestasis (Fig. 1A); however, pruritus rapidly reoccurred.
Immunofluorescence staining was performed as described18 using the following antibodies: K24,19 K165 and K168 (BSEP),18 P3II26 (MDR3), M2I4 (multidrug resistance–associated protein 2 [MRP2]), and K12 (rat-Bsep). The patient's posttransplant serum S3 showed a canalicular staining of control livers at dilutions from 1:10 to 1:5000. For the shown experiments, the sera of the patient (S0-S3), as well as that of eight patients with PFIC-2/PFIC-3, nine patients with AIH, and six controls were used both undiluted and at 1:25. Human immunoglobulins were detected by specific Cy-3–conjugated antibodies for human immunoglobulin A (IgA)-α-chain, immunoglobulin M (IgM)-Fc5μ-fragment, and immunoglobulin G (IgG)-Fcγ-fragment.
BSEP Sequencing, Cloning, Mutagenesis, and Transfection.
Sequencing of BSEP (NM_003742.2) was performed using genomic DNA and primers (on request), which covered all exons and exon/intron boundaries of BSEP.
Total RNA from human liver was reverse-transcribed into complementary DNA. Primers containing a BglII and KpnI restriction site upstream and downstream of the coding sequence were used for polymerase chain reaction of full-length BSEP, which was cloned into pEYFP-N1-Vector (Clontech, Palo Alto, CA). The construct contained three silent mutations (2190G→A, 3210A→G, 3516C→C) and one mutation (1331T→C), leading to an amino acid replacement (V444A). Using site-directed mutagenesis (Stratagene, La Jolla, CA), cytidine1331 was changed to thymidine with the following primer: 5′-ctaaatgacctcaacatggtcattaaaccaggggaaatg-3′. The patient's mutations were introduced using the Quickchange-Multisite-mutagenesis kit (Stratagene) and the following primers: 2453A→T (Y818F): 5′-ccaatttctacagggatttgcctttgctaaatc-3′; 2944G→A (G982R): 5′-cagaaagccaatatttacagattctgctttgcctttgc-3′. Successful mutagenesis was verified by sequencing of the whole construct.
Cells (human embryonic kidney 293 [HEK293], HepG2, and HeLA) were seeded onto coverslips and transiently transfected with the BSEP–yellow fluorescent protein (YFP) plasmids using LipofectAMINE (Invitrogen). Cells were fixed (100% methanol) for immunofluorescence 48 hours after transfection. Treatment with MG-132 (2 μM) was started 24 hours after transfection for 24 hours.
Western and Dot Blot Analysis.
BSEP-YFP–transfected HEK293 cells were harvested in hypotonic buffer (0.1 mM ethylene diamine tetraacetic acid, 0.05 mM sodium phosphate, pH 7.0) supplemented with protease inhibitor cocktail (Roche, Mannheim, Germany). After homogenization using a tight-fitting homogenizer (1000 rpm/minute; 10 strokes), crude membrane pellets were obtained by centrifugation (20,000g, 1 hour, 4°C). Pellets were resuspended and equal protein amounts were separated by soium dodecyl sulfate polyacrylamide gel electrophoresis and blotted onto nitrocellulose membranes. The patient's serum was diluted 1:5000 and detected with a peroxidase-conjugated anti-human-IgG–specific antibody. BSEP-YFP was detected using the anti–green fluorescent protein (GFP) antibody (Santa Cruz Biotechnology, Santa Cruz, CA) or the anti-BSEP antibodies K165/K168, respectively.
For dot blot analysis, 2.5 μL peptide solutions (1 mg/mL) were dotted onto nitrocellulose membranes. The patient's serum S3 was used as primary antibody at 1:10000 and detected with a peroxidase-conjugated secondary antibody and chemiluminescence.
Rat hepatocytes were isolated by collagenase perfusion, plated onto collagen-coated Matek dishes and maintained in William's medium. After 24 hours, cells were incubated with the patient's serum (S0, S3) and control serum (each 1:10) for 1 hour. Cholyl-lysyl-fluorescein (CLF, 1 μM) was added for 10 minutes. Cells were washed and analyzed on a confocal microscope. Images of 10 adjacent visual fields were taken with the same instrument settings for all conditions.
The two-tailed Mann-Whitney U test (Wilcoxon) was used, and the alpha level was 0.01.
Needle biopsies from the second and third liver transplant were analyzed by immunofluorescence. The canalicular transporters BSEP, MDR3, and MRP2 were distributed in a normal canalicular pattern (Supporting Fig. 1). It is known that immunosuppressants can induce vesicular retrieval of canalicular transporter proteins as a cause of cholestasis.20 However, in the patient's liver there was no apparent change of the distribution of BSEP immunoreactivity between the canalicular membrane and the cytoplasm as compared to control (Supporting Fig. 2).
Because the occurrence of de novo autoantibodies is a common phenomenon after pediatric liver transplantation,13 the presence of autoantibodies in the transplanted livers was tested by the use of antibodies directed against human immunoglobulins. Remarkably, an intense canalicular staining with a specific anti–human-immunoglobulin antibody was found, indicating the deposition of immunoglobulins in the canalicular space (Fig. 1B). This was a patient-specific finding, because there were no immunoglobulins detectable in the canaliculi in three healthy human livers (from organ donors), eight transplant biopsies from children with posttransplantation cholestasis (treated with the same immunosuppressive regimen), and four livers from patients with primary biliary cirrhosis or primary sclerosing cholangitis, two common cholestatic liver diseases of autoimmune pathophysiology (representative example, Fig. 1B).
The presence of autoantibodies in the patient's sera from different time points (S0/S1 = 10 months/1 day before the first liver transplantation; S2/S3 = after the second/third transplantation) was evaluated by staining of normal human liver slices. Notably, both posttransplant sera S2 and S3, but not the pretransplant sera S0 and S1 contained IgG antibodies with immunoreactivity to a canalicular epitope (Fig. 1C). The canalicular staining pattern was apparently a unique feature of this patient's serum, because no canalicular immunoreactivity was obtained with serum samples from six healthy persons, nine patients with autoimmune hepatitis (AIH), or with serum samples from other children with inherited PFIC-2 (five before and three after transplant) and from children with defective MDR3 (PFIC-3)18 (three before and two after transplant).
The autoantibodies derived from the patient's sera were only detectable by a secondary anti-human IgG antibody, but not by anti-human IgM or IgA antibodies. This suggests that in our patient, transplantation had triggered chronic autoantibody production against a canalicular target protein, which could be identified as BSEP, as described below.
Full-length human BSEP was cloned from liver, tagged to a yellow fluorescent protein (BSEP-YFP) and transiently expressed in HepG2, HEK293, and HeLa cells, where it was correctly targeted to the apical membrane (HepG2) or the plasma membrane (HEK293/HeLa), respectively. The patient's sera S2 and S3 detected transfected, but not untransfected cells (Fig. 2A). A band of 180 kDa in BSEP-YFP–transfected but not in untransfected cells (Fig. 2B) was detected by the serum, confirming BSEP as the antibody target.
The patient's antibody cross-reacted with rat Bsep: in permeabilized rat hepatocytes, co-immunoreactivity of the serum and the Bsep antibody K12,21 which is directed against the intracellular C-terminus of rat BSEP, was observed (Fig. 2C). However, in nonpermeabilized cells, immunoreactivity to S2/S3 but not toward K12 was found (Fig. 2G), suggesting that the autoantibody binds an extracellular epitope of Bsep/BSEP.
Full-length BSEP was subcloned into its two homologous halves (amino acids 1-659 and 659-1321) and transfected into HepG2 cells. The sera S2/S3 detected the first but not the second half of BSEP (Fig. 2I,J). The first half contains three putative extracellular loops from amino acid 84-144, 237-240, and 342-361. Ten partially overlapping oligopeptides, each containing 15 amino acids of these three loops, were (Supporting Table 1) used for dot blot analysis. S3 bound solely to the peptide-2 (Fig. 2G) covering amino acids 91-105.
Employing standard settings in MODELLER version 9.1,22 a homology model of BSEP was constructed based on the crystal structure of the multidrug transporter Sav1866 from Staphylococcus aureus (Protein Data Bank entry: 2HYD).23 In this model, the amino acids 91-105 are localized close to the extracellular part of the channel pore of BSEP (Fig. 3E).
In Vitro BSEP Transport Assay.
Isolated rat hepatocytes were grown to confluence to form pseudocanaliculi, which were sealed by tight junctions. They were incubated with the patient's serum, fixed, and stained by anti-human IgG antibodies. Within 10 minutes, antibodies from the posttransplant sera S2/S3 (Fig. 3A-C) but not from control sera, including S0 (Fig. 3D), reached the canaliculi. These data show that the anti-BSEP IgG antibody may reach the canalicular lumen of hepatocytes.
In order to determine a potential steric effect of the autoantibodies on BSEP activity, rat hepatocytes were incubated with the patient's sera or human control serum for 60 minutes. Thereafter, the bile acid CLF24 (generous gift of C. Mills, Birmingham, UK) was added for 10 minutes (1 μmol/L). In hepatocytes incubated with control serum, CLF was rapidly secreted into the pseudocanaliculi (Fig. 3G). In cells preincubated with the patient's serum, the number of CLF-positive pseudocanaliculi was significantly reduced to 28.7% ± 8.1% (n = 6, P < 0.01) as compared to controls, and fluorescence intensity in remaining canaliculi was diminished. When bile acids were removed from the serum by dialysis (cutoff size of 14 kDa), the patient's dialyzed serum still reduced CLF secretion by 57% compared to controls. Bile salt concentrations were 150 μmol/L in the pretransplant serum S0 and 202 μmol/L in the posttransplant serum S3. Despite a comparable level of bile salt concentration, serum S3 reduced CLF secretion to 32.6% ± 18.8% (n = 2) compared to the pretransplant serum S0 (Fig. 3F). On the basis of these results, it is concluded that inhibition of CLF secretion is not due to substrate competition but due to the presence of posttransplant BSEP autoantibodies, which trans-inhibit BSEP function in vivo.
Effects of BSEP Mutations on BSEP Expression and Localization.
Introduction of the mutations V444A or Y818F separately or together into BSEP-YFP had no apparent effect on the localization of BSEPV444A, BSEPY818F, or BSEPV444A/Y818F as compared to wild-type BSEP-YFP (Fig. 2A, 4A). However, introduction of G982R alone or in combination with either V444A or Y818F (Fig. 4B) resulted in the retention of BSEPG982R, BSEPV444A/G982R, and BSEPY818F/G982R in the endoplasmic reticulum (ER) in colocalization with the ER marker protein disulfide isomerase. Combining all three mutations resulted in nondetectability of BSEPV444A/Y818F/G982R-YFP (Fig. 4C) independent of the cell type used (Supporting Fig. 3).
Inhibition of proteasomes by MG-132 induced the formation of aggresomes25 in cells expressing wild-type BSEP-YFP (Fig. 4D), where YFP fluorescence colocalized with K165 immunoreactivity (Fig. 4D). Unexpectedly, K165 and K168 immunoreactivity, but no YFP fluorescence, was present in aggresomes of BSEPV444A/Y818F/G982R-YFP–transfected cells after MG-132 treatment (Fig. 4G,J). It can be concluded that the combination of the three mutations results in the complete degradation of BSEPV444A/Y818F/G982R-YFP within the ER-associated degradation pathway. Furthermore, the apparent loss of YFP fluorescence is suggestive of an incomplete expression or a major folding defect of BSEPV444A/Y818F/G982R-YFP.
In conclusion, only the combination of all three BSEP mutations caused a complete disappearance of BSEP from the child's original liver, which is the basis of the lack of tolerance for BSEP after liver transplantation.
The case presented here unravels a novel mechanism of graft failure due to autoantibodies, which inhibit the bile salt export pump BSEP. These antibodies were directed against the first extracellular loop of BSEP and were detected in large amounts at the canalicular membranes of hepatocytes. They obviously reach the apical membrane. This may be triggered by receptor-mediated transcytosis26 and subsequent binding of antibodies to BSEP within the canaliculi. Alternatively, antibody binding may even occur at the sinusoidal membrane, because BSEP may be targeted from the Golgi to canalicular membrane through an indirect route including a short stay at the sinusoidal membrane.27 The latter possibility is supported by the observation that BSEP antibodies from posttransplant sera but not antibodies from normal sera were enriched in the canaliculi of isolated rat hepatocytes, favoring the view of a specific instead of a general mechanism of biliary antibody deposition. The inhibitory effect of BSEP autoantibodies may be due to cross-linking of BSEP molecules, which may interfere with the assembly of microdomains and may disturb the coordinated action of canalicular transporters. Likewise, direct mechanical occlusion of the BSEP pore may contribute to the antibodies' inhibitory effect; this is supported by the epitope topology within the hypothetical BSEP model.
Serum antibodies can be reduced by PP28 or by rituximab, which binds to CD20 and thereby inhibits B cells.29 In our patient, the combination of PP and rituximab treatment temporarily reduced bile acid levels, most likely by diminishing BSEP-reactive antibodies, thus underlining the inhibitory property of these autoantibodies.
It is of note, that in addition to the G982R and Y818F mutations, the common mutation/polymorphism V444A is necessary for the complete disappearance of BSEP. V444A has a high prevalence in the Caucasian population,17 has been linked to the development of drug-induced liver injury,30 and may aggravate cholestatic liver diseases such as intrahepatic cholestasis of pregnancy31 or benign recurrent intrahepatic cholestasis.32
Expression of BSEPV444A/Y818F/G982R in hepatic and nonhepatic cell lines was below detectability. Therefore, BSEP expression and processing of T cell epitopes in medullary thymic epithelial cells (mTECs), which are responsible for the induction of self-tolerance, may be completely absent.33 Furthermore, the mutations themselves may directly alter processing and presentation of epitopes by mTECs.
The severe damage of the first transplant almost certainly led to the release of vast amounts of wild-type BSEP, which may have triggered the expression of BSEP epitopes by antigen-presenting cells. In combination with the missing tolerance to BSEP, an immune response with the induction of autoantibodies was promoted. Both self-tolerance and epitope presentation by antigen-presenting cells are HLA-type–dependent.
Absence of BSEP expression has been observed in many cases of PFIC-218, 34; however, recurrent cholestasis due to antibody formation after liver transplantation in these patients is not a common finding. Therefore, in our patient absent primary tolerance for BSEP epitopes and intense epitope presentation in the course of graft injury after the first transplantation in the combination with the patient's individual HLA type may explain BSEP antibody formation.
In summary, this is the first illustration of a specific, functionally relevant autoantibody directed against an ABC-transporter induced by liver transplantation.
Inspiring discussions with Prof. Dr. Irmgard Förster, IUF Düsseldorf, and expert manuscript editing by Prof. R. Thompson, King's College, London, are thankfully acknowledged. Provision of clinical documentation by Prof. Dr. R. Ganschow, University Clinic of Hamburg, is highly appreciated. The study was performed according to the guidelines of the declaration of Helsinki and informed written consent was obtained from patients or parents.