Clinical, genetic and functional perspectives on ATP-binding cassette subfamily B member 4 variants in five cholestasis adults

Abstract

BACKGROUND

ATP-binding cassette subfamily B member 4 (ABCB4) deficiency is associated with cholestatic liver disease primarily because of missense mutations, and many variants remain unidentified. Here, we validate the pathogenicity and mechanism of ABCB4 variants in clinical and in vitro trials, hypothesizing that these variants are responsible for impaired biliary function and contribute to the development of cholestatic liver diseases.

AIM

To clarify the functional features and pathogenicity of ABCB4 variants.

METHODS

Clinical data were collected from five patients with cholestatic liver disease that was initially not detected by routine examinations. Later, whole-exome sequencing confirmed ABCB4 variants and the patients were treated from January 2017 to December 2023. Pathogenic mechanisms were analyzed using bioinformatics tools, and a cell model in vitro was established to investigate ABCB4 mRNA expression, multidrug resistance protein 3 (MDR3) expression, cellular localization, and phosphatidylcholine secretion. Results were compared using Student's t-tests.

RESULTS

Five missense variants (c.1757T>A, c.1865G>A, c.2362C>T, c.2777C>T and c.3250C>T), one intron variant (c.537-32G>T), and one synonymous (c.C504T) variant were identified. Three of the five patients had various degrees of cholestasis, two presented with liver cirrhosis, and all had elevated gamma-glutamyl transferase. Three of the four patients who underwent a liver biopsy had bile duct dilation, and one had gallstones. Two of the four patients had normal and reduced MDR3 immunohistochemical levels. Bioinformatic analysis indicated that these variants were likely pathogenic except c.C504T variant. None of the missense variants influenced subcellular MDR3 Localization in vitro. However, the c.1865G>A variant significantly decreased ABCB4 mRNA values, and all missense variants down-regulated phosphatidylcholine secretion.

CONCLUSION

This study uncovered new ABCB4 variants and emphasized the pathogenic potential of specific variants. The findings from five patients provided insight into the pathogenic mechanisms underlying ABCB4-related diseases.

Key Words: ATP-binding cassette subfamily B member 4; Multidrug resistance protein 3; Cholestasis; Functional analysis; Clinical; Gene mutation; Whole-exome sequencing

Core Tip: In this study, we included patients with cholestatic liver disease, and through whole exome sequencing, we identified five patients carrying ATP-binding cassette subfamily B member 4 variants. We analyzed their clinical, pathological, and prognosis. Compound heterozygosity leads to poor prognosis. We predicted the pathogenicity of the seven variants using bioinformatics tools and further investigated the pathogenicity of missense variants in vitro. We studied the mRNA, protein content, subcellular localization, and phosphatidylcholine secretion of the variants in vitro cell models. Ultimately, we found that all missense variants were pathogenic in clinical settings and in vitro.

INTRODUCTION

Intrahepatic cholestasis is a rare clinical condition with several genetic etiologies that include liver diseases associated with ATP-binding cassette subfamily B member 4 (ABCB4). The symptoms can be complex, variable, and frequently misdiagnosed[1]. The spectrum of effects of ABCB4 variants includes mild abnormal liver function, cholelithiasis, low phospholipid-associated cholelithiasis, intrahepatic cholestasis during pregnancy, drug-induced liver injury, severe progressive familial intrahepatic cholestasis type 3 (PFIC-3) and liver cirrhosis[2-4]. Severe PFIC-3 can develop during infancy or puberty with clinical complications, including liver cirrhosis, portal hypertension, and end-stage liver disease[5]. In practice, PFIC-3 can progress earlier in some children and present later depending on the pathogenicity of mutations[6]. Current treatment options are limited, as only a subset of patients respond to ursodeoxycholic acid (UDCA), and patients with a poor prognosis might require a liver transplant, or they might not survive. Correlations between the genotypes and phenotypes of ABCB4 variants remain unclear, which complicates clinical disease prediction and subsequent treatment.

Located on chromosome 7q21.1 (GRCh38/hg38), ABCB4 comprises 28 exons and spans approximately 74 kb. It encodes multidrug resistance protein 3 (MDR3), which is located on the canalicular membrane of hepatocytes and transports phosphatidylcholine using ATP. An imbalance between phosphatidylcholine and bile salts can lead to cholesterol precipitation and damage to canalicular bile ducts[7,8]. ABCB4 exhibits a characteristic architecture comprising two transmembrane domains (TMD) forming the substrate-binding cavity and two nucleotide-binding domains (NBDs) responsible for ATP hydrolysis. NBDs are characterized by the presence of highly conserved structural motifs, including Walker-A, Walker-B, LSGGQ, H-loop, Q-loop, and D-loop[9]. In recent years, numerous studies have been conducted on the potential mechanisms of PC transport[10,11]. According to the Human Gene Variants Database (HGMD Professional 2024.12), most ABCB4 variants are missense types; known variants include 311 missense and 58 splice variants, as well as 37 and 23 small deletions and insertions, respectively (http://www.hgmd.org). However, most available data are based on sequencing, and clinical data on patients harboring these mutations are scarce. The pathogenic mechanisms of ABCB4 variants remain unclear, thus posing challenges regarding treatment. We studied five patients with ABCB4 variants to clarify these issues.

MATERIALS AND METHODS

Patients and genetic analysis

This study initially assessed patients with cholestatic liver disease who attended the outpatient clinic or ward of Nanjing Second Hospital from January 2017 to December 2023. However, ABCB4 variants in only five patients were ultimately detected by whole-exome sequencing in Illumina platform after routine enquiries, laboratory tests, imaging studies, and pathological evaluations had been uninformative and when the cause of the cholestatic liver disease remained obscure. Genomic DNA extracted from peripheral blood samples was treated with EDTA to prevent enzymes from breaking it down and to prevent degradation during storage.

The reference for nucleotide substitutions was ABCB4 (NM_000443.3). The DNA was amplified using the polymerase chain reaction (PCR), and then amplicons were sequenced using the chain termination method (Sanger) at KingMed Diagnostics (Nanjing, China). The obtained data demonstrated an average sequencing depth of ≥ 90 × across the exon regions of known human genes and their 5-bp flanking sequences, with approximately 98% of target regions achieving ≥ 20 × coverage depth. The quality control of Q30 score values is 0.93. Comprehensive base calling was systematically implemented for all sequencing reads. This clinical detection platform was developed and rigorously validated by KingMed Diagnostics. Supplementary Table 1 and Supplementary Figure 1 show the oligonucleotide primers and mass spectrometry validation. We excluded the influence of other mutant genes that could cause cholestasis, such as ATP8B1, BSEP, TJP2, and MYO5B. Moreover, ABCB4 variants that were bioinformatically determined as meaningless were excluded. The study protocol adhered to the ethical guidelines enshrined in the Declaration of Helsinki (2013 amendment) and was approved by the Ethics Committee of The Second Hospital of Nanjing, Affiliated to Nanjing University of Chinese Medicine (2021-LY-kt052). Written informed consent was obtained from patients or their guardians to collect clinical samples from them and to publish clinical data generated in this study.

Liver biopsy and immunohistochemical staining

Liver biopsies were obtained under ultrasound guidance from four patients when they initially presented at our hospital, but one (patient 4) declined to undergo this procedure. The samples were fixed in formalin, dehydrated with alcohol, embedded in paraffin, stained with hematoxylin and eosin, and immunohistochemically stained with the anti-ABCB4 antibody P3II-26 (Thermo Fisher Scientific Inc., Waltham, MA, United States). The samples were incubated with fluorescent secondary antibodies (Abcam, Cambridge, United Kingdom), followed by DAB-H₂O₂ for 10 minutes, then stained and sealed. Pathological diagnoses were based on the Scheuer scores[12] and an evaluation by two clinicians and pathologists.

Bioinformatics

The detected variants were compared with data from the Exome Sequencing Project (https://evs.gs.washington.edu/EVS/), Human Gene Variants Database (http://www.hgmd.cf.ac.uk/ac/index.php), and the gnomAD database (https://gnomad.broadinstitute.org) to exclude common variants. We predicted the pathogenicity of missense variants using the bioinformatics tools Provean[13], PolyPhen-2[14], MutPred-2 (http://mutpred.mutdb.org/), and VASOR[15]. Intron variants were analyzed using NNSplice[16], and the pathogenicity of non-missense variants was assessed using MutationTaster[17]. Interspecific conservation was analyzed using DNAMAN9.0 (http://www.lynnon.com/index.html). A three-dimensional structural model of MDR3 (PDB ID: 6S7P) was generated using SWISS-MODEL and visualized using PyMOL. Interpretation of genetic variations adhered to the pathogenic variant classification guidelines established by the American Society of Medical Genetics and Genomics[18].

Mutagenesis and plasmid constructs

ABCB4 isoform A (NM_000443.3) variants were constructed using pcDNA3.1 as described[19], and ABCB4 wild-type (wt) glycerol stocks were synthesized (Nanjing Proteinbio Co., Nanjing, China). Mutant variants were generated by site-directed mutagenesis (c.1757T>A, c.1865G>A, c.2362C>T, c.2777C>T and c.3250C>T). Primers were designed to introduce specific variants as described by the manufacturer (Supplementary Table 2). The entire open reading frame of ABCB4 was sequenced to confirm mutagenesis. Plasmids were extracted using the TIANprep Mini Plasmid Kit (Tiangen Biotech Co., Ltd., Beijing, China). Supplementary Figure 2 shows a plasmid map and the gene sequence.

Cell culture and transfection

We cultured HEK293 cells (Xiehe Cell Resource Centre, Beijing, China) in Dulbecco’s modified Eagle medium supplemented with 10% fetal bovine serum. Its mycoplasma test was negative (PCR method), then we seeded them into 6-well plates at a density of 3 × 10⁵/well. The cells were transiently transfected for 24 hours with a plasmid vector containing either the ABCB4 wt or mutant sequences at a plasmid-to-transfection reagent ratio of 2:1. The cells were transfected using the JetPRIME reagent[20] (Polyplus, Illkirch, France) as described by the manufacturer. The control plasmid was an empty vector plasmid[21-23], and here we used pcDNA3.1.

Immunofluorescence staining

We incubated HEK293 cells (0.5 × 10⁵/well) in 24-well plates for 24 hours, then transfected and incubated them for 48 hours. The cells were fixed in 4% paraformaldehyde, permeabilized for 20 minutes at room temperature, and washed with phosphate-buffered saline (PBS). Non-specific binding was blocked with 1% bovine serum albumin at room temperature. The cells were then incubated overnight with MDR3 antibody and rabbit polyclonal anti-calnexin antibody (Abcam, Cambridge, United Kingdom) at 4 °C. The cells were washed with PBS with Tween 20 and incubated with 1:200-diluted red (anti-mouse IgG H&L 568) and green (goat anti-rabbit IgG H&L 488) fluorescent secondary antibodies (Abcam). Images were captured using a TCS SP8 Laser scanning confocal microscope (Leica Microsystems GmbH, Wetzlar, Germany). ABCB4 wt was the negative control and ABCB4-I541F[24] was the positive control.

Quantitative reverse transcription PCR

The cells were transfected for 48 hours, and then the total cellular RNA was isolated using FastPure^® Cell/Tissue Total RNA Isolation Kit V2 (Vazyme, Nanjing, China). Complementary DNA synthesized from 1 µg of total RNA and reverse transcribed using HiScript^®II Q RT SuperMix for qPCR (+gDNA wiper) served as the template for Quantitative reverse transcription PCR (RT-qPCR), with GAPDH as the reference gene. Amplification was carried out using ChamQ SYBR qPCR Master Mix (Low ROX Premixed) and an Applied Biosystems 7500 fast real-time PCR system for 40 cycles. The forward and reverse (5′ to 3′) specific primers were: ABCB4: AACCCCAAGATCCTTCTGCT and GGACCGTAGACAGTCGGTGT GAPDH: GAAGGTGAAGGTCGGAGTCA and GACAAGCTTCCCGTTCTCAG.

Relative expression was calculated using the 2^−ΔΔCt method.

Western blotting

Total cellular proteins were extracted from HEK293 cells in ice-cold RIPA lysis buffer after transfection for 48 hours. Protein concentrations were determined using bicinchoninic assays, and then proteins were separated by 6% (w/v) polyacrylamide gel electrophoresis. Proteins and the GAPDH (ABclonal, Wuhan, China) control were transferred onto polyvinylidene difluoride membranes and incubated with anti-MDR3 antibody overnight at 4 °C, followed by the secondary antibody, horseradish peroxidase goat anti-mouse IgG (ABclonal, Wuhan, China). Fluorescence emission was detected using ECL kits, a ProteinSimple FluorChem M imaging system, and ImageJ.

Measurement of phosphatidylcholine secretion

We incubated HEK293 cells (4 × 10⁴/well) for 24 hours before transfection for 24 hours, then equal volumes of cell supernatants containing wt and mutant variants were collected. Phosphatidylcholine was assayed in cell supernatants using a Kit (Abcam, Cambridge, United Kingdom) that uses an enzyme-coupled reaction to hydrolyze phosphatidylcholine and release choline, which oxidizes an OxiRed probe to generate fluorescence (excitation/emission 535/587 nm) or absorbance. Fluorescence was measured at 570 nm using a Microplate Reader (Thermo Fisher Scientific Inc.). Phosphatidylcholine content in the supernatants was determined using a standard curve.

Statistical analysis

Means between the two groups were compared using Student's t-tests. P values < 0.05 were considered statistically significant. All data were analyzed using SPSS version 25 (IBM Corp., Armonk, NY, United States) and GraphPad Prism version 9.0 (GraphPad Software, San Diego, CA, United States). The statistical review of the study was performed by a biomedical statistician.

RESULTS

Clinical and genetic findings

We characterized five patients (male, n = 4; female, n = 1; average age, 26.6 years). All patients had clinical signs of cholestasis at various degrees of severity. Patients 2 and 3 had chronic cholestasis, patient 4 had hepatic cysts and splenomegaly, and patients 1 and 5 had more complex chronic cholestasis, liver cirrhosis, upper gastrointestinal bleeding, and portal hypertension. All patients were evaluated to exclude conditions such as viral hepatitis, cholangitis, autoimmune hepatitis, or drug-induced hepatitis. Table 1 summarizes the clinical and histological findings.

Table 1 Clinical features of five patients and the corresponding ATP-binding cassette subfamily B member 4 mutations.

No.	Gender	Age at first presentation	Clinical manifestation	Biochemical examination, (maximal value)	Abdominal enhanced, CT scan	MRCP	Pathological features1	Pathological score	CK7	IHC5	Therapy	Outcome on 2024
1	M	24	Haematemesis, Hematochezia, Cholelithiasis	TB 175, ALT 512, ALP 883, GGT 713, TBA 54	Liver cirrhosis, Megalosplenia, ascites	Cholecystitis, Cholelithiasis, Choledocholithiasis	Ductopenia, cholestasis, hepatic fibrosis	G1S4	(+)	Normal	UDCA, transmetil, glycyrrhein	Died
2	M	21	Scleral icterus, Hepatic dysfunction	TB 29, ALT 159, ALP 346, GGT 1016, TBA 59	Cholecystitis, splenomegaly	Intrahepatic bile duct dilatation	Bile duct hyperplasia	G2S2	(-)	Normal	UDCA, transmetil, Fenofibrate, glycyrrhein	Alive
3	M	18	Hepatic dysfunction, Pruritus, Cholestasis	TB 26, ALT 117, ALP 163, GGT 321, TBA 48	Cholangiectasis	Cholangiectasis	Bridging fibrosis, Ductopenia	G1S3	(+)	Decrease	UDCA, transmetil, glycyrrhein	Alive
4	M	31	Hepatic dysfunction, abdominal distention, jaundice	TB 7, ALT 80, ALP 139, GGT 309	Hepatic cyst, splenomegaly	/	/	/	/	/	UDCA, transmetil, glycyrrhein	Alive
5	F	38	Hepatic dysfunction, abdominal distention, edema	TB 72, ALT 52, ALP 528, GGT 233, TBA 257	Splenectomy, Esophageal and gastric varices, liver cirrhosis	Mild dilation of bile duct	Ductopenia, portal inflammation	G2S4	(+)	Decrease	UDCA, Transmetil, glycyrrhein	Alive

¹It represents the hematoxylin-eosin staining characteristics of liver biopsy.

Reference value: Total bilirubin 3-19 μmol/L, glutamic pyruvic transaminase: 5-40 U/L, aspartate aminotransferase: 5-40 U/L, alkaline phosphatase: 35-135 U/L, GGT: 0-50 U/L, TBA: 0-13 μmol/L. The pathological diagnosis was based on the Scheuer score. ALP: Alkaline phosphatase; ALT: Glutamic pyruvic transaminase; GGT: Glutamyl transpeptidase; TB: Total bilirubin; TBA: Total bile acids; CK7: Cytokeratin 7; IHC: Immunohistochemistry; UDCA: Ursodeoxycholic acid; CT: Computed tomography; MRCP: Magnetic resonance cholangiopancreatography; M: Male; F: Female; /: No data.

Genomic DNA sequencing identified five new ABCB4 missense variants among the patients as follows: Patient 1 carried both c.2362C>T and c.2777C>T, patient 2 had c.2362C>T, along with the intronic variant, c.537-32G>T, patient 3 had c.1865G>A, patient 4 had c.1757T>A, and patient 5 had c.3250C>T and c.C504T. These were distributed across various regions of the MDR3 protein, including first nucleotide-binding domain (V586E), third intracellular loop (G622E), fourth intracellular loop (R788W), fifth intracellular loop (P926 L), and second transmembrane domains (R1084W) (Figure 1). Table 2 shows details of the variant profiles.

Figure 1 Schematic representation of ATP-binding cassette subfamily B member 4. A: ATP-binding cassette subfamily B member 4 (ABCB4) gene structure with mutations representing exons 1 to 28. Each mutation found in our study is marked; B: The secondary structure of multidrug resistance protein 3 protein and the distribution of missense mutations found by our team. Black circle marked ABCB4 variants were detected in the patients cohort in our research. Grey pentagram marked ABCB4 variants reported in the literature; A: Walker A; B: Walker B; C: The signature motif C; COOH: Carboxyl terminus of polypeptide chain; H₂N: Amino terminus of polypeptide chain; In: Inside the cell membrane; Out: Outside the cell membrane; ABCB4: ATP-binding cassette subfamily B member 4; MDR3: Multidrug resistance protein 3.

Table 2 Analysis of ATP-binding cassette subfamily B member 4 variants and corresponding pathogenicity in five patients.

No.	Zygosity of variant	Nucleotide change	Amino acid change	Reference sequence	Location	Domin	Minor allele frequency	SIFT	PolyPhen-2 (HumDiv)	MutPred-2	NNSplice	Mutation Taster	VASOR	Classification according to ACMG
1	Composite heterozygosity	c.2362C>T, c.2777C>T	p.R788W, p.P926 L	NM_000443.3	Exon20, Exon22	IC4, IC5	2.48e-6, -	0.00, 0.01	1, 0.961	0.904, 0.617	-, -	-, -	0.851, 0.876	Uncertain significance (PS3 + PP3), Uncertain significance (PM2 + PP3)
2	Heterozygosity	c.2362C>T, c.537-32G>T	p.R788W, -	NM_000443.3	Exon20, Intron6	IC4, NA	2.48e-6, -	0.00, -	1, -	0.904, -	-, splicing donor loss	-, Pathogenic	0.851, -	Uncertain significance (PS3 + PP3), Uncertain significance (PM2 + PP3)
3	Heterozygosity	c.1865G>A	p.G622E	NM_000443.3	Exon15	IC3	1.24e-6	0.00	0.921	0.83	-	-	0.915	Uncertain significance (PP3)
4	Heterozygosity	c.1757T>A	p.V586E	NM_000443.3	Exon15	NBD1	-	0.02	0.999	0.873	-	-	0.757	Uncertain significance (PM2 + PP3)
5	Heterozygosity	c.3250C>T, c.C504T	p.R1084W, p.N168N	NM_000443.3	Exon25, Exon6	NBD2, Cytoplasmic	3.10e-6, 0.475	0.00, -	1, -	0.807, -	-, -	-, polymorphism	0.857, -	Uncertain significance (PM2 + PP3), Benign (BA1 + BP4 + BP7)

Minor allele frequency is sourced from gnomAD Database, with samples from at least 25 different countries/regions, mainly from Europe and South Asia. SIFT: A predicted score of less than or equal to 0.05 indicates that the mutation will affect protein function, while a score greater than 0.05 indicates that the effect on the protein is tolerable. PolyPhen-2: The range of predicted values is from 0 to 1. If > 0.5, the mutation is predicted to be harmful, and if < 0.5, it is predicted to be benign. MutPred-2: The general score indicates the pathogenicity of a given variant, ranging from 0 to 1. The higher the score, the greater the pathogenicity tendency. Generally, a score greater than 0.5 is considered a pathogenic mutation. “-”does not affect amino acids or the software is not suitable for this type of mutation. Vasor: It gives out a probability of pathogenicity, which can range from 0 (likely benign) to 1 (likely pathogenic). NBD1: Nucleotide binding domain; TMD: Transmembrane domain; IC: Intracytoplasmic.

Except for synonymous variants, which are considered gene polymorphisms (minor allele frequency: 0.475), the identified variants were rare. Multiple sequence alignment analyses revealed that the five missense variants in MDR3 were evolutionarily conserved (Supplementary Figure 3). Structural analysis revealed that these missense variants disrupted local hydrogen bonds within MDR3. Notably, the G622E variant in patient 3 altered the local micropotential on the MDR3 surface (Figure 2). The NNSplice findings suggested that the c.537-32G>T intronic variant in patient 2 affected mRNA splicing. The SIFT scores for these five missense variants were all < 0.05, whereas PolyPhen-2 and MutPred-2 scores were > 0.5. The VASOR scores were also > 0.5, indicating a high likelihood of pathogenicity.

Figure 2 Structural analysis of missense variants in multidrug resistance protein 3. Prediction of three-dimensional structure of ATP-binding cassette subfamily B member 4 (ABCB4) mutation sites by Swiss Model. A: P.V586E mutation forms new hydrogen bond with T424; B: P.G622E mutation affects steric hindrance; C: P.R788W mutation breaks hydrogen bond with Q825 and L817; D: P.P926 L mutation affects steric hindrance; E: P.R1084W mutation affects steric hindrance; F: The left image represents ABCB4-WT, while the right image illustrates the G622E variant. The yellow circle indicates that the neutral glycine is replaced by the negatively charged glutamic acid, with the red region representing negative charges, the white region indicating neutrality, and the blue region denoting positive charges. ABCB4: ATP-binding cassette subfamily B member 4.

Pathological findings

All biopsied liver tissue samples obtained from four of the five patients revealed varying degrees of inflammation and fibrosis. The predominant pathological alteration was bile duct damage characterized by infiltration, cholestasis, and ductopenia (Table 1). Under normal conditions, MDR3 is uniformly expressed in liver biopsy tissues[17], and it was distributed evenly and expressed normally in patients 1 and 2. However, MDR3 was unevenly distributed, with decreased expression in patient 3 and significantly reduced expression in patient 5 (Figure 3).

Figure 3 Immunohistochemical manifestations of the liver in a patient with ATP-binding cassette subfamily B member 4 variants. Except for patient 4, all other patients underwent liver biopsy. Immunohistochemical staining was performed using anti- ATP-binding cassette subfamily B member 4 (P3II-26, Thermo, United States), with the white arrows indicating the labeling of multidrug resistance protein 3 (MDR3) tubules. Bars = 2.5 μm. A: MDR3 is evenly distributed in liver tissue (the white arrow indicates MDR3 is normal, case 1); B: MDR3 is evenly distributed in liver tissue (the white arrow indicates MDR3 is normal, 400 ×, case 2); C: Decreased distribution of MDR3 in liver tissue (the white arrow indicates MDR3 is decreasing, 400 ×, case 3); D: Decreased distribution of MDR3 in liver tissue (the white arrow indicates MDR3 is decreasing, 400 ×, case 5). ABCB4: ATP-binding cassette subfamily B member 4; MDR3: Multidrug resistance protein 3.

Impact of missense variants on subcellular localization

The ABCB4 wt or mutant alleles were transiently transfected into HEK293 cells that were incubated with anti-MDR3 and anti-calnexin antibodies. Thereafter, MDR3 and the type I integral membrane protein calnexin in the endoplasmic reticulum (ER) that served as the ER marker were labeled with red and green fluorescence, respectively. Transfection of the wt allele resulted in the uniform distribution of red fluorescence throughout entire cells, whereas green fluorescence was not emitted from the nuclear region. However, fluorophores at the periphery overlapped, indicating that MDR3 Localizes specifically to the plasma membrane of HEK293 cells. The five missense variants were all similarly localized (Figure 4).

Figure 4 Confocal microscopy of cellular localization of multidrug resistance protein 3. HEK293 cells expressing the wild-type and mutant ATP-binding cassette subfamily B member 4 alleles were labeled with anti- multidrug resistance protein 3 (red) and anti-calnexin (green) antibodies, the yellow color indicates the coexistence of the two proteins. Followed by fluorescence-conjugated secondary antibodies, and observed using laser scanning confocal microscopy. Bars = 10.5 μm. MDR3: Multidrug resistance protein 3; WT: Wild-type.

Impact of missense variants on ABCB4-mRNA and MDR3 expression

The expression of ABCB4 mRNA in HEK293 cells was measured using RT-qPCR, with GAPDH as the internal reference. The mRNA expression of ABCB4 was reduced relative to the wt due to the c.1865G>A variant, whereas that of the other variants surpassed that of the wt (Figure 5A). Wt and mutant MDR3 proteins in HEK293 cells transfected with the corresponding expression plasmids were quantified using western blotting. We identified mature (160 kDa) and immature (140 kDa) forms of MDR3 that were consistent with previous findings[21,25]. Cellular levels of MDR3 protein with the G622E variant were significantly decreased compared with the wt, whereas those of the other mutant proteins were similar to those of the wt (Figure 5B and C).

Figure 5 Expression analysis and determination of transportation capability. A: MRNA expression analysis of ATP-binding cassette subfamily B member 4 (ABCB4)-wild-type (WT) and variants by qPCR. Using GAPDH as a reference for relative quantity analysis, Values were analyzed by GraphPad Prism 8.0. In addition, the results are the mean ± SD of four independent experiments. ^aP < 0.01, ^dP < 0.0001; B: Immunbloting of ABCB4 and variants. Representative western blot of the expression levels of the mutants with respect to ABCB4-WT. The expression and the processing of ABCB4-WT and the mutants was examined by western blot analysis of whole-cell lysates from transfected HEK293 cells. ABCB4 expression was detected following SDS-PAGE and immunoblotting with the anti- multidrug resistance protein 3 (MDR3) antibody. GAPDH was used as a loading control. Molecular masses are indicated on the left (in kDa). Mature MDR3 was show in black while immature MDR3 show in white; C: Western blot analysis of ABCB4-WT and variants. The values have been normalized to ABCB4-WT. And they were measured by ImageJ and analyzed by GraphPad Prism 8.0. The dates show the means ± SD of three separate experiments. ^bP < 0.01, ^cP < 0.001; D: Determination of transportation capability of MDR3 and its variants. The phosphatidylcholine levels in the supernatant of HEK293 cells were measured, which were respectively transfected with ABCB4-WT gene and mutant alleles. The secretion of phosphatidylcholine was standardized to wild type group. The data were analyzed by GraphPad Prism 8.0. And they represent the means ± SD of four independent experiments. ^bP < 0.01, ^dP < 0.0001. MDR3: Multidrug resistance protein 3; WT: ABCB4 wild type.

Impact of missense variants on the transport function of MDR3

Phosphatidylcholine is transported from hepatocytes to bile under physiological conditions by MDR3, encoded by ABCB4, which reduces the toxic effects of bile salts on the canalicular membrane. Phosphatidylcholine levels in supernatants of HEK293 cells expressing different MDR3 variants were evaluated to determine the function of the mutant MDR3 proteins. Phosphatidylcholine secretion was decreased in all variants compared with MDR3 wt, and the content was significantly reduced in c.1757T>A, c.1865G>A and c.3250C>T (Figure 5D). This suggests that these mutations have pathogenic potential in vitro.

DISCUSSION

This study analyzed the clinical and genetic profiles of five Chinese patients with cholestasis and ABCB4 mutations. Whole-exome sequencing identified five missense variants of ABCB4 (c.1757T>A, c.1865G>A, c.2362C>T, c.2777C>T and c.3250C>T), two of which (c.1757T>A, c.2777C>T) had not been previously reported in literature or database. We also found that the mechanisms underlying these five missense variants were pathogenic in clinical settings and in vitro.

The prognosis of patient 1 was predictable. He initially presented with hematemesis, melena, liver cirrhosis and splenomegaly. Test results indicated decreased peripheral blood cell counts, blood coagulation abnormalities, significantly elevated glutamyl transpeptidase (GGT), and increased bile acid levels. Microscopic examination of the liver revealed bile duct loss and a poor response to UDCA. He harbored the compound heterozygous variants c.2362C>T and c.2777C>T. His parents refused full-exon analysis; thus, we were unable to determine whether the variants were inherited or de novo. However, the condition of this patient was seriously evident. Compound heterozygous variants have been linked to increased pathogenicity[26], suggests that patients with compound heterozygous and homozygous mutations could develop more severe clinical phenotypes. It can be understood that patient 1 passed away because of decompensated cirrhosis during follow-up. Patient 2 in the present study carried the c.2362C>T heterozygous mutation but did not have liver cirrhosis and responded well to UDCA. Cholestatic liver disease associated with ABCB4 is related to a wide range of clinical and genetic variations. One report describes a patient with a homozygous ABCB4 c.2362C>T variant who developed cholestatic liver disease at age 25 and died 18 years later[27]. Our findings are consistent with the fact that the R788W variant impairs phospholipid secretion in vitro. R788 is located in the intracellular loop domain of the MDR3 protein, which is highly conserved and critical for coupling of the nucleotide-binding domain, and it plays a vital role in the conformational change induced by ATP hydrolysis. This study also revealed that the c.2777C>T variant caused a decrease in phosphatidylcholine levels that might contribute to further disease progression.

Patient 5 was the oldest patient and initially presented with liver cirrhosis. Immunohistochemical findings uncovered a significant decrease in MDR3. She harbored the c.3250C>T heterozygous variant. This variant was found to downregulate phosphatidylcholine secretion in vitro. The c.3250C>T variant has been identified in patients with cholelithiasis that is associated with low phospholipid levels, indicating a correlation between this mutation and ABCB4 expression[28]. This variant is located in the key nucleotide-binding domain 2. The nucleotide-binding domain serves as a site for ATP binding, supplying the energy necessary for transmembrane transport and facilitating phospholipid movement against concentration gradients[29,30]. Variants affecting this region might compromise ATP binding and consequently affect phospholipid transport[31]. Age is another important factor to be considered. Theoretically, long-term exposure of the bile duct to bile salts might lead to further progressive disease. Specifically, mutations can affect phosphatidylcholine secretion, leading to a lack of phospholipid neutralization by bile salts, which exerts a toxic effect on the bile duct wall and causes further damage. This might result in loss of the bile duct over time. The absence of MDR3 targeting sites after bile duct loss further decreases MDR3 function, creating a vicious cycle. This has been found in ABCB4 carriers, particularly those with early disease onset and poor responses to UDCA, who might require liver transplantation or are unlikely to survive[32,33]. Patients with low phospholipid-associated cholelithiasis, including those with recurrent cholelithiasis, might develop bile duct cancer at a later stage[34].

The c.1865G>A variant was found to significantly downregulate mRNA expression in vitro. Patient 3 presented with abnormal liver function, pruritus, and significantly elevated GGT levels. Magnetic resonance imaging suggested localized intrahepatic bile duct dilation. Immunohistochemical staining of a pathological liver biopsy specimen revealed bridging fibrosis, bile duct loss in the portal area, and reduced MDR3, which led us to consider ABCB4-related disease. A large-scale sequencing study in Iceland identified the c.1865G>A variant in patients with cholelithiasis. PyMOL was used to visualize the charge distribution of the c.1865G>A variant in a three-dimensional structural model generated by SWISS-MODEL. The uncharged glycine was replaced by negatively charged glutamate, which altered the local hydrogen bonding and charge distribution on the surface of MDR3 and might have affected the conformation and stability of the protein. Residues S666 and T667, located in the same flexible loop, exhibited significantly downregulated differential phosphorylation, which might interfere with ATP binding or hydrolysis, which are critical for phosphatidylcholine translocation[35]. The variant that significantly downregulated mRNA expression changed the codon from GGG to GAG, which did not create a new stop codon, indicating that it did not result in a nonsense mutation or lead to mRNA degradation. Various reasons for mRNA downregulation have been described, such as negative regulation by microRNAs (miRNA)[36] that require further investigation.

Our findings indicated that the c.1757T>A variant significantly affected phosphatidylcholine secretion and resulted in the transformation of amino acid 586, which is located near the H-ring of the first nucleotide-binding domain of MDR3. The nucleotide-binding domain is key for ATP hydrolysis, and it includes a P ring (or Walker-A motif) to bind α and β phosphates of ATP, an A-ring that provides aromatic side chains closely packed with the purine ring of adenine, a Walker-B motif containing the catalytic glutamic acid, a characteristic LSGGQ motif that immobilizes and orients ATP, and an H-ring that acts as a switch histidine during hydrolysis to stabilize the transition geometry. The Q-ring provides contact with the TMD and the dimerization or D-ring functions in coupling hydrolysis with transport[29]. Variants in this region, such as H589Y[25], are pathogenic. However, the phenotype of patient 4 did not appear to be as severe as that indicated by the results of the phospholipid tests in vitro. This patient declined liver assessment by computed tomography, magnetic resonance imaging, and histology. Thus, his status could not be comprehensively evaluated. Additional assessments are needed to further determine correlations between genotypes and phenotypes.

Considering that the five missense variants did not affect MDR3 localization and that all downregulated phosphatidylcholine expression, they can be classified as type III mutations[21]. The effects of ABCB4 missense mutations that affect critical motifs of NBDs and significantly impair phosphatidylcholine transport activity can be rescued by treatment with the clinically approved cystic fibrosis transmembrane regulator synergist ivacaftor[25]. This and its derivatives might be potential therapeutic agents in the future, offering hope to patients with type III mutations.

These findings expand the spectrum of ABCB4 mutations. Understanding the molecular mechanisms underlying ABCB4-related cholestasis was enhanced using clinical, genetic, and functional analyses of five missense variants in Chinese patients, and a relationship was established between genotypes and phenotypes. Although our study sample size is small, ABCB4 variants are still dominated by missense variants, as confirmed by Wang et al[37]. Compound heterozygous and homozygous mutations can lead to cumulative pathogenicity, resulting in severe clinical phenotypes[12]. Age is an important factor, as long-term chronic bile duct damage can lead to disease progression and deterioration. Furthermore, the c.1865G>A variant downregulated ABCB4 mRNA expression, which warrants further investigation. Mutations affecting highly conserved key motifs within the nucleotide-binding domain, particularly those in the Walker-A, Walker-B, LSGGQ, H-loop, Q-loop, and D-loop motifs, might contribute to the disease. According to the ABCB4 pathogenicity classification, the five missense mutations described above can be classified as type III mutations, which might be rescued by ivacaftor.

The hydrophilic bile acid UDCA has several mechanisms of action that can be applied to manage ABCB4-related diseases. It improves liver function in some patients and might delay the progression of liver disease. However, its effectiveness might be reduced in advanced disease[4], which our findings also reflected. Furthermore, some chemical chaperones and small molecules could improve MDR3 Localization and function, thereby alleviating the symptoms of biliary disease[38,39]. RNA interference-related studies have proceeded in mice[40]. Xue et al[41] demonstrated inAbcb4-/- mouse models that hyodeoxycholic acid exhibits therapeutic potential for cholestatic liver fibrosis, highlighting its prospective applicability in patients with ABCB4 variants complicated by hepatic fibrosis. In addition, there is also research on ABCB4 transcription factors, which may serve as potential drug targets in the future[42]. Ongoing research and clinical trials are essential to further improve the treatment outcomes and quality of life of patients with ABCB4-related diseases. Although our study provides valuable insights into the functional effects of missense mutations in ABCB4, it has some limitations. For example, the cellular model might not fully replicate the complex physiological environment of the human body. This study focused on a single mutation, whereas patients might carry several mutations. Furthermore, this study did not explore the effects of c.537-32G>T in detail. Therefore, future studies should investigate other cell types and animal models to validate our findings and identify other potential pathogenic mutations.

CONCLUSION

This study underscores the significance of missense mutations in ABCB4-related diseases and paves the way for future studies. A deeper understanding of the pathogenic mechanisms underlying these mutations will help to provide patients with more accurate diagnoses and better treatment strategies.

ACKNOWLEDGEMENTS

We would like to express our sincere gratitude to the Science and Technology Experiment Center of Nanjing University of Chinese Medicine for the laser scanning confocal microscope.