Potential conflict of interest: Nothing to report.
Deficiency in P-type ATP8B1 is a severe and clinically highly variable hereditary disorder that is primarily characterized by intrahepatic cholestasis. It presents either as a progressive (progressive familial intrahepatic cholestasis type 1 [PFIC1]) or intermittent (benign recurrent intrahepatic cholestasis type 1 [BRIC1]) disease. ATP8B1 deficiency is caused by autosomal recessive mutations in the gene encoding ATP8B1, a putative aminophospholipid-translocating P-type adenosine triphosphatase. The exact pathogenesis of the disease is elusive, and no effective pharmacological therapy is currently available. Here, the molecular consequences of six distinct ATP8B1 missense mutations (p.L127P, p.G308V, p.D454G, p.D554N, p.I661T, and p.G1040R) and one nonsense mutation (p.R1164X) associated with PFIC1 and/or BRIC1 were systematically characterized. Except for the p.L127P mutation, all mutations resulted in markedly reduced ATP8B1 protein expression, whereas messenger RNA expression was unaffected. Five of seven mutations resulted in (partial) retention of ATP8B1 in the endoplasmic reticulum. Reduced protein expression was partially restored by culturing the cells at 30°C and by treatment with proteasomal inhibitors, indicating protein misfolding and subsequent proteosomal degradation. Protein misfolding was corroborated by predicting the consequences of most mutations onto a homology model of ATP8B1. Treatment with 4-phenylbutyrate, a clinically approved pharmacological chaperone, partially restored defects in expression and localization of ATP8B1 substitutions G308V, D454G, D554N, and in particular I661T, which is the most frequently identified mutation in BRIC1. Conclusion: A surprisingly large proportion of ATP8B1 mutations resulted in aberrant folding and decreased expression at the plasma membrane. These effects were partially restored by treatment with 4-phenylbutyrate. We propose that treatment with pharmacological chaperones may represent an effective therapeutic strategy to ameliorate the recurrent attacks of cholestasis in patients with intermittent (BRIC1) disease. (HEPATOLOGY 2009.)
New insights into the genetic basis of liver disease have had enormous impact on our understanding of disease pathogenesis, but translation into pharmacological treatment remains a challenging task. One such hereditary liver disorder is P-type adenosine triphosphatase member 8B1 (ATP8B1) deficiency (previously called familial intrahepatic cholestasis type 1 [FIC1] disease; caused by mutations in ATP8B1).1 ATP8B1 deficiency is primarily characterized by low gamma-glutamyl transferase intrahepatic cholestasis, due to a defect in bile salt secretion.2 A severe manifestation is progressive familial intrahepatic cholestasis type 1 (PFIC1), which also comprises Greenland familial cholestasis,3 causing end-stage liver disease if untreated. A milder manifestation is called benign recurrent intrahepatic cholestasis type 1 (BRIC1), which is characterized by intermittent cholestasis. The severity, duration, and frequency of cholestatic attacks in BRIC1 are variable, and it is unknown what triggers their onset and spontaneous resolution. ATP8B1 deficiency is distinct from ABCB11 deficiency. The latter is characterized by similar cholestatic phenotypes (called PFIC2 and BRIC2) but is caused by mutations in ABCB11 (ATP-binding cassette B11), the gene encoding the bile salt export pump (BSEP).2 ATP8B1 is a member of the P4 subfamily of P-type adenosine triphosphatases (ATPases). P4-type ATPases associate with members of the CDC50 protein family, and formation of these complexes is required for P4 ATPase stability and export from the endoplasmic reticulum (ER).4, 5 Studies in yeast have suggested that these protein complexes translocate phospholipids across cellular membranes.4, 6 Although not yet unequivocally demonstrated, a role of ATP8B1 in transport of phosphatidylserine from the outer leaflet of the canalicular membrane to the inner leaflet is suggested.5, 7, 8 How loss of ATP8B1 activity secondarily causes impairment of bile salt secretion is still being investigated.
For several diseases, including cystic fibrosis (CF) and alpha-1 antitrypsin deficiency, elucidation of the deleterious consequences of genetic defects on protein folding has opened avenues to develop effective treatment.9, 10 A recent example is the pharmacological chaperone 4-phenylbutyrate (4-PBA), which has turned into a promising tool to ameliorate the plasma membrane expression of a number of proteins affected by genetic diseases.9, 10 These diseases have in common that the gene mutations result in defects in protein folding. Importantly, the molecular consequences of ATP8B1 mutations on the folding, expression, and localization of the ATP8B1 protein have not been identified. Here, we selected seven distinct mutations in ATP8B1, previously identified in PFIC1 and/or BRIC1 patients (Fig. 1A), and systematically assessed the cellular mechanisms explaining the defects due to these specific mutations. This detailed characterization then permitted attempts to rescue ATP8B1 expression at the plasma membrane using the pharmacological chaperone 4-PBA.
ATP8B1, P-type ATPase member 8B1; BRIC, benign recurrent intrahepatic cholestasis; BSEP, bile salt export pump; cDNA, complementary DNA; CF, cystic fibrosis; ER, endoplasmic reticulum; FIC, familial intrahepatic cholestasis; NT, not transfected, 4-PBA, 4-phenyl butyric acid; PCR, polymerase chain reaction; PFIC, progressive familial intrahepatic cholestasis; TfR, transferrin receptor; WT, wild-type.
Materials and Methods
Rabbit anti-ATP8B1 antibody was characterized previously.7 Other antibodies used were rabbit anti-calreticulin (Alexis Biochemicals, San Diego, CA), mouse anti-transferrin receptor (Zymed, San Francisco, CA), rabbit anti-VSV-G (vesicular stomatitis virus glycoprotein G) (Abcam, Cambridge, UK), mouse anti-V5, mouse anti-V5, fluorescein isothiocyanate (FITC)-conjugated, goat anti-rabbit Cy3-conjugated (Invitrogen, San Diego, CA), goat anti-rabbit, horseradish peroxidase–conjugated (DAKO, Carpinteria, CA), goat anti-mouse, horseradish peroxidase–conjugated (Pierce, Rockford, IL). Rabbit anti-Na/K ATPase was a generous gift from Dr. J. Koenderink (Nijmegen, The Netherlands).
ATP8B1 constructs with a VSV-G epitope tag at the amino-terminus were constructed by polymerase chain reaction (PCR) using human ATP8B1 complementary DNA (cDNA) as template and cloned into the AscI and NheI sites of pCB7-VSV. This construct was used as template for site-directed mutagenesis according to the manufacturer's protocol (Stratagene), to create seven mutations previously identified in patients with ATP8B1 deficiency.11 Human CDC50A was cloned into pCDNA3-V5 by PCR or into pmKate2-N to create a carboxyl-terminal tagged constructs (see Supporting Table 1 for primer sequences.)
Cell Culture, Transfections and Drug Treatment.
Human bone osteosarcoma epithelial cells (U2OS)12 and human embryonic kidney 293T (HEK293T)13 cells were cultured in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum and antibiotics. Transient transfections were performed using calcium phosphate or polyethylenimine using standard procedures, and cells were harvested after 1–3 days. Cells were incubated at 27°C, 30°C, or 40°C, treated with 3 μM MG132 (Calbiochem, San Diego, CA), 100 nM epoxomycin (Sigma, St. Louis, MO), or 5 mM 4-phenyl butyric acid (Sigma, St. Louis, MO) for 40, 16, and 40 hours, respectively.
Coimmunoprecipitation and Western Blot Analysis.
Cells transiently transfected with pCB7-ATP8B1 and pcDNA3-CDC50A were lysed in 20 mM Tris-HCl; pH 7.4, 5 mM Na-ethylene diamine tetraacetic acid, 135 mM NaCl, 1.0% (vol/vol), Nonidet P-40, and 10% (wt/vol) sucrose and centrifuged at 16,000g for 15 minutes. Supernatants were subjected to western blot analysis or incubated with mouse anti-V5 antibodies immobilized on protein A–agarose beads (Sigma, St. Louis, MO) for 2 hours at 4°C, followed by western blot analysis.
HEK293T cells were transiently cotransfected with pCB7-ATP8B1 and pcDNA3-CDC50A using calcium phosphate. After 3 days, RNA was isolated using TRIZOL (Invitrogen), and residual DNA in the samples was degraded by deoxyribonuclease I treatment according to the manufacturer's protocol (Invitrogen). After reverse transcription using oligo-deoxythymidine primers (Roche) and Superscript II reverse transcriptase (Invitrogen), quantitative PCR was performed using a MyIQ real-time PCR cycler (Bio-Rad Laboratories). Results were presented as fold induction, normalized to hypoxanthine-guanine phosphoribosyltransferase, which was selected as the most stable reference gene as described.14 Hygromycin phosphotransferase was used as a transfection marker, encoded within the pCB7-ATP8B1 construct.
U2OS cells were grown on coverslips and co-transfected with pCB7-ATP8B1 and pcDNA3-CDC50A using polyethylenimine. After 2 days, cells were fixed using paraformaldehyde and ATP8B1 and the ER-marker protein disulfide isomerase (PDI) or CDC50A were visualized using rabbit anti-VSV-G and Cy3 coupled secondary antibody together with mouse anti-PDI and AlexaFluor 488–coupled secondary antibody or FITC-conjugated mouse anti-V5. Images were acquired using a LSM710 Meta confocal microscope (Carl Zeiss, Jena, Germany).
Cell Surface Biotinylation.
Two days after transfection, U2OS cells were washed with phosphate-buffered saline supplemented with 0.5 mM CaCl2 and 1.0 mM MgCl2 (PBS-CM), and proteins present at the cell surface were biotinylated using sulfo-NHS-SS-biotin and solubilized as described.15 Biotinylated proteins were precipitated for 2 hours using neutravidin-coupled beads (Pierce) and analyzed by immunoblot analyses. Cytosolic proteins were undetectable in the precipitated fraction, and no precipitated protein was detected when sulfo-NHS-SS-biotin was omitted, demonstrating the specificity of the procedure.
All figures represent at least three independent experiments. Protein expression was measured by densitometry using ImageJ (http://rsbweb.nih.gov/ij/). Background intensity was subtracted and values were compared using Mann-Whitney after testing for overall significance using the Kruskal-Wallis test (P < 0.05 was considered significant). Data are provided as mean ± standard deviation (SD).
Low ATP8B1 Mutant Expression Due to Protein Misfolding and Proteasomal Degradation.
To study the effect of cholestasis-associated mutations in ATP8B1 on protein expression, HEK293T cells were cotransfected with CDC50A and ATP8B1 wild-type (WT) and mutants. ATP8B1 WT protein was readily detectable at approximately 140 kDa, and endogenous expression in HEK293T cells was very low. The protein expression of ATP8B1 mutations G308V, D454G, D554N, I661T, G1040R, and R1164X was significantly reduced, whereas the p.L127P mutation did not affect ATP8B1 expression levels (Fig. 1B). Identical results were obtained using U2OS cells, strongly suggesting cell type independence, and data of both cell types are averaged in Fig. 1C. ATP8B1 R1164X migrated faster than ATP8B1 WT, in agreement with the absence of the carboxyl-terminus (Fig. 1B), and also exhibited reduced expression. In contrast, the messenger RNA (mRNA) expression of all mutants with reduced protein expression was unaffected (Fig. 1D). Protein expression of ATP8B1 WT, I661T, and G1040R was increased 1.1-fold to 2-fold upon treatment with the proteasomal inhibitors MG132 or epoxomycin (Fig. 2A). ATP8B1 mutants with lowest protein expression in control conditions, i.e., p.G308V, p. D454G, p.D554N, and p.R1164X, were increased 2-fold to 11-fold upon treatment with MG132 or epoxomycin. This suggested that reduced expression of ATP8B1 mutant proteins is due to proteasomal protein degradation. Proteasomal degradation can be triggered by protein misfolding.16 Therefore, cells were cultured at reduced temperature, because this can stimulate expression of otherwise misfolded proteins. Protein expression levels of ATP8B1 mutations G308V, D454G, D554N, I661T, and R1164X were increased approximately 2-fold, and ATP8B1 mutation G1040R was increased approximately 1.4-fold, when cells were cultured at 27°C (not shown) or 30°C (Fig. 2B).
Mislocalization of ATP8B1 G308V, D454G, D554N, I661T, and R1164X.
To test for possible defects in protein trafficking, the localization of all ATP8B1 mutants was compared with ATP8B1 WT by immunocytochemistry. When ATP8B1 and CDC50A, the heterodimer partner of ATP8B1 in the liver, were expressed individually, both proteins exclusively localized to the ER (Fig. 3A). When coexpressed, ATP8B1 WT colocalized with CDC50A at the plasma membrane (Fig. 3B). Similarly, ATP8B1 L127P and G1040R localized to the plasma membrane. However, ATP8B1 G308V, D454G, D554N, and to a lesser extent ATP8B1 I661T displayed predominant intracellular localization, with little signal outside the ER (Fig. 3B,C). Interestingly, in all cases, complete colocalization between CDC50A and ATP8B1 was observed. ATP8B1 plasma membrane localization was subsequently determined by cell surface biotinylation in U2OS cells. ATP8B1 WT was detected in the biotinylated fraction when CDC50A was coexpressed, but not when CDC50A or biotin was omitted (Fig. 4A). In addition, ATP8B1 L127P, I661T, and G1040R were present at the plasma membrane (Fig. 4B). Although clearly detectable in the total cell lysate, only minute amounts of ATP8B1 G308V, D454G, and D554N were detected in the biotinylated fraction. ATP8B1 R1164X signal was completely absent from the plasma membrane (Fig. 4B). Together, these data demonstrate that ATP8B1 WT, L127P, and G1040R are efficiently targeted to the plasma membrane in the presence of CDC50A. ATP8B1 I661T is distributed between the ER and the plasma membrane and all other mutants are virtually exclusively localized in the ER.
ATP8B1 Mutations Do Not Abolish the Interaction with CDC50A.
Because CDC50A interaction is required for plasma membrane localization of ATP8B1, we investigated whether this association is impaired due to any ATP8B1 mutation. ATP8B1 WT and all mutants were coimmunoprecipitated with CDC50A (Fig. 5). Although the difference in expression levels of the ATP8B1 mutants precluded quantitative assessment of the interaction, this finding showed that none of the ATP8B1 mutations abolishes CDC50A binding. The ER localization of most mutants can therefore not be explained by an inability to interact with CDC50A.
Homology Modeling Points to Structural Changes in ATP8B1.
In order to get more detailed insight into the molecular consequences of the selected mutations, we constructed a homology model of WT ATP8B1 based on the crystal structure of the related P-type Ca2+-ATPase and modeled the effects of individual patient-associated mutations on protein structure (Fig. 6A and Supporting Information). Substitutions p.L127P and p.G1040R are predicted to result in structural changes within the transmembrane domains of ATP8B1 (Fig. 6B,F). The p.G308V mutation causes a destabilizing rearrangement in the ATP8B1 Actuator domain (Fig. 6C), which likely influences the association between the Actuator and Phosphorylation domains (indicated by A and P in Fig. 6A), two ATP8B1 structural domains that are highly conserved in all P-type ATPases. The residues D454 and D554 are close together in the cytosolic core of the protein, and are critical for the catalytic cycle of P-type ATPases (Fig. 6D). I661 is a fully exposed residue, located in the Nucleotide-binding domain (N-domain) (Fig. 6E). The I661T mutation does not seem to result in major structural changes within ATP8B1, in line with the relatively mild clinical consequences of this mutation.11 ATP8B1 R1164X lacks three helical turns of the last transmembrane helix (shown green in Fig. 6A) and 80 C-terminal residues, whose structure could not be reliably predicted. Together, these modeling data support the hypothesis that most of the studied mutations result in significant structural alterations.
4-PBA-Mediated Up-Regulation of ATP8B1 Protein Expression and Cell Surface Abundance.
We investigated whether treatment with the pharmacological chaperone 4-PBA ameliorated the low expression of ATP8B1 mutants. ATP8B1 G308V protein expression was significantly increased by 4-PBA treatment in a dose-dependent manner (Fig. 7A). Total cellular expression of ATP8B1 G308V, D454G, D554N, and R1164X was induced two-fold to five-fold by 4-PBA treatment (Fig. 7B). Interestingly, protein expression of ATP8B1 I661T and G1040R showing only mildly reduced expression levels in control conditions, also poorly responded to 4-PBA treatment. ATP8B1 WT expression was not stimulated by 4-PBA, suggesting specific up-regulation of otherwise misfolded proteins.
Subsequently, cell surface biotinylation was performed to determine whether 4-PBA stimulated the trafficking of ATP8B1 mutants to the cell surface. Neither ATP8B1 nor the transferrin receptor (used as a loading control) was detected when biotin was omitted, indicating the specificity of the signal for cell surface resident proteins. ATP8B1 G308V, D454G, and D554N showed a 1.5-fold to 2-fold increase in plasma membrane expression upon 4-PBA treatment (Fig. 8B). Despite increased protein expression upon 4-PBA treatment, no ATP8B1 R1164X signal was detectable at the cell surface in either condition. Interestingly, ATP8B1 I661T abundance in the biotinylated fraction was strongly enhanced (5-fold to 10-fold) upon 4-PBA treatment, suggesting markedly improved trafficking to the plasma membrane (Fig. 8A,B). The reverse occurred when cells were cultured at 40°C. This temperature increase resulted in a significant decrease in the amount of ATP8B1 I661T, but not WT protein at the cell surface (Fig. 8C). Neither expression of ATP8B1 I661T in the lysate nor the amount of transferrin receptor at the plasma membrane changed as a result of incubation at 40°C.
ATP8B1 deficiency constitutes a potentially lethal form of intrahepatic cholestasis. We and others have previously identified many distinct mutations in ATP8B1.1, 3, 11, 17–19 Correlations between missense mutations and phenotypes of individual patients remained limited, because most mutations are confined to only few patients and because of the high variation in penetrance and clinical presentation. Furthermore, the molecular consequence of ATP8B1 mutations remained largely unexplored. Here, we combined protein expression and localization studies with homology modeling to demonstrate the effects of selected ATP8B1 mutations on the protein level. These studies have high relevance for the patient population affected with ATP8B1 deficiency, because three of the selected mutations—p.G308V detected in Amish families, p.D554N in Greenland Inuit, and p.I661T in most European BRIC1 patients—are the most frequently identified mutations, together affecting the vast majority of all patients characterized with ATP8B1 deficiency.
Although our data do not fully explain the large variability in clinical presentation, they demonstrate that ATP8B1 deficiency presents as a protein folding disease, for a surprisingly large majority of the selected mutations. This conclusion is supported by several lines of evidence. First, with the exception of ATP8B1 L127P, all ATP8B1 mutants displayed significantly reduced protein expression, whereas mRNA expression was unaffected. Second, the recovery of ATP8B1 mutant expression upon MG132 and epoxomycin treatment indicates that increased proteasomal degradation is a common consequence of these ATP8B1 mutations. Third, incubation at reduced temperature has been demonstrated to restore proper folding of mutated proteins, and increased ATP8B1 mutant expression was observed when cells were cultured at 30°C. Fourth, most ATP8B1 mutants showed minimal plasma membrane localization. Instead, they were retained in the ER. Fifth, homology modeling predicted significant changes in the ATP8B1 structure due to the various mutations. In conclusion, in most cases, ATP8B1 deficiency is a consequence of protein misfolding, resulting in reduced expression at the plasma membrane. In vitro, a further reduction of ATP8B1 I661T protein abundance at the cell surface occurs when cells are cultured at 40°C. This may suggest that temporary decrease in ATP8B1 abundance at the plasma membrane could trigger the onset of a cholestatic episode in BRIC1 patients afflicted with the p.I661T mutation, because patients report that episodes are sometimes preceded by fever.
Current treatment of ATP8B1 deficiency has major obstacles. Reduction of the (hydrophobic) bile salt pool using ursodeoxycholate or cholestyramine is only rarely effective.2 Surgical and/or endoscopic drainage of bile salts is more successful, but involves invasive procedures.20, 21 The characterization of ATP8B1 deficiency as a folding disorder permitted to investigate the efficacy of pharmacological chaperones to restore expression of ATP8B1 at the plasma membrane, aiming to design novel therapy for patients with ATP8B1 mutations. We selected 4-PBA for these studies, because it is a clinically approved drug for treatment of a variety of human disorders originating or manifesting in the liver. These include urea cycle disorders, where 4-PBA is used short-term in dosages exceeding the concentration used in this study. Importantly, this drug also ameliorates cell surface abundance of a number of misfolded and mislocalized membrane proteins with relevance to human liver disease. These include ABCB11 p.E297G and p.D482G, identified in PFIC2 patients, and CF transmembrane conductance regulator ΔF508, the most common mutation associated with CF.9, 22 Clinical trials demonstrated that 4-PBA treatment resulted in increased chloride conductance in patients with CF.23 In our study, 4-PBA treatment stimulated the protein expression and/or cell surface abundance of ATP8B1 G308V, D454G, D554N, and I661T in vitro. Amelioration of cell surface expression was most prominent for the latter mutant, with a dramatic increase in ATP8B1 I661T accessible for biotinylation. However, using multiple assays, we could not detect ATP8B1-mediated internalization of fluorescently labeled phosphatidylserine upon coexpression of ATP8B1 WT and CDC50A (data not shown). This prohibited verification whether ATP8B1 I661T at the cell surface is functional. Furthermore, it is currently unclear to what level ATP8B1 cell surface abundance needs to be restored, and how much ATP8B1 catalytic activity is required, to relieve patients from an ATP8B1 deficiency phenotype. Given the episodic character of cholestatic attacks in BRIC1 patients, and the predominant absence of clinical symptoms in heterozygous carriers, we propose that partial restoration of cell surface expression by 4-PBA might already provide clinical improvement in BRIC1. A substantial proportion of BRIC1 patients carry at least one p.I661T allele, and many are homozygous for p.I661T. Whether the episodic character of BRIC1 even precludes the need for long-term administration of 4-PBA in these patients, needs to be investigated in clinical trials once the in vivo efficacy has been ascertained.
In conclusion, the results of this study investigating six missense and one nonsense mutation, indicate that future therapy aiming to restore ATP8B1 expression at the plasma membrane probably needs to be tailored to specific genetic defects, and emphasize the need for continued detailed and systematic analysis of ATP8B1 mutations that possibly result in folding defects. Treatment with pharmacological chaperones like 4-PBA, might present a clinically useful approach to increase the amount of ATP8B1 mutant protein at the cell surface, especially in patients with episodic presentation of ATP8B1 deficiency. The large proportion of folding mutations detected in this study, in a parallel analysis of mutations in ATP7B associated with Wilson Disease (Van den Berghe et al., unpublished results), and in previous studies10, 22 indicate the relevance of this strategy to hereditary liver disease in general.
We thank Rina Wichers and Gözde Isik for assistance.