Interindividual variability of canalicular ATP-binding-cassette (ABC)–transporter expression in human liver


  • Potential conflict of interest: Nothing to report.


Interindividual variability in hepatic canalicular transporter expression might predispose to the development of hepatic disorders such as acquired forms of intrahepatic cholestasis. We therefore investigated expression patterns of bile salt export pump (BSEP, ABCB11), multidrug resistance protein 3 (MDR3, ABCB4), multidrug resistance associated protein 2 (MRP2, ABCC2) and multidrug resistance protein 1 (MDR1, ABCB1) in healthy liver tissue of a white population. Protein expression levels were correlated with specific single nucleotide polymorphisms (SNPs) in the corresponding transporter genes. Hepatic protein expression levels from 110 individuals undergoing liver resection were assessed by Western blot analysis of liver plasma membranes enriched in canalicular marker enzymes. Each individual was genotyped for the following synonymous (s) and nonsynonymous (ns) SNPs: ABCB11: (ns:1457T>C and 2155A>G), ABCB4: (ns:3826A>G) and ABCC2 (ns:1286G>A,3600T>A and 4581G>A) and ABCB1 (ns:2677G>T/A and s:3435C>T). Transporter expression followed unimodal distribution. However, of all tested individuals 30% exhibited a high expression and 32% a low or very low expression phenotype for at least one of the four investigated transport proteins. Transporter expression levels did not correlate with age, sex, underlying liver disease, or presurgery medication. However, low BSEP expression was associated with the 1457C-allele in ABCB11 (P = .167) and high MRP2 expression was significantly correlated with the 3600A and 4581A ABCC2 variants (P = .006). In conclusion, the results demonstrate a considerable interindividual variability of canalicular transporter expression in normal liver. Furthermore, data suggest a polymorphic transporter expression pattern, which might constitute a risk factor for the development of acquired forms of cholestatic liver diseases. (HEPATOLOGY 2006;44:62–74.)

Hepatic bile formation is driven by various adenosine triphosphate (ATP)-binding cassette (ABC)-transporters localized at the canalicular membrane of hepatocytes. They include the bile salt export pump (BSEP, ABCB11), the phospholipid flippase MDR3(ABCB4), the organic anion transporter MRP2(ABCC2), and the xenobiotic transporter MDR1(ABCB1). BSEP mediates ATP-dependent secretion of bile salts across the canalicular membrane of hepatocytes and, thus, represents the main driving force for the generation of bile salt–dependent bile flow.1, 2 MDR3 acts as a primary active phospholipid flippase and translocates phosphatidylcholine from the inner to the outer leaflet of the canalicular membrane. MRP2 is a multispecific organic anion transporter that mediates biliary excretion of a broad spectrum of divalent organic anions, including bilirubin diglucuronide and reduced glutathione, thereby contributing most significantly to the generation of bile salt–independent bile flow.3–5 The exact contribution of MDR1 to hepatic bile formation remains to be established, but it is thought to contribute to the canalicular excretion of drugs and other xenobiotcs into bile. Any functional disturbance of these transport systems can therefore lead to the intracellular accumulation of potentially harmful exogenous and endogenous compounds and constitute a risk for the development of cholestatic liver disease. For instance, progressive familial intrahepatic cholestasis type2 (PFIC2) or benign recurrent intrahepatic cholestasis type2 (BRIC2) both represent BSEP deficiency syndromes.6–10 Similarly, hereditary absence of MDR3 expression leads to PFIC3,11 and lack of MRP2 deficiency is the cause of the Dubin Johnson syndrome.12 In contrast to the hereditary forms of cholestasis, the pathophysiological significance of decreased canalicular ABC transporter expression or function for acquired forms of cholestasis is not well understood. However, BSEP function is inhibited by several potentially cholestatic drugs, and it has been suggested that low BSEP expression or function might represent a risk factor for certain forms of drug-induced cholestasis.4, 13 Furthermore, disease-specific ABCB4 gene polymorphisms have been described in intrahepatic cholestasis of pregnancy, raising the possibility that decreased MDR3 expression and function, in addition to transinhibition of BSEP by estrogen conjugates,13 might represent a risk factor for the development of intrahepatic cholestasis of pregnancy.11, 14 Additionally, altered expression and function of MDR1 in human small intestine has been associated with changes in drug absorption,15 pointing toward a possible role of canalicular MDR1 expression levels in determining biliary elimination of drugs and hence hepatic exposure to xenobiotics. To investigate whether such hypothesized “low expressors” of canalicular ABC transporters occur indeed among the white population, we screened 110 individuals who were subjected to hepatectomy because of focal liver disease, for canalicular expression of BSEP, MDR3, MRP2, and MDR1.


ABC, adenosine triphosphate–binding cassette; BSEP, bile salt export pump; MDR, multidrug resistance protein; PFIC, progressive familial intrahepatic cholestasis; ALT, alanine aminotransferase; AP, alkaline phosphatase; cBili, conjugated bilirubin; cMF, membrane fraction enriched in canalicular marker enzymes; AUC, area under the curve; SNP, single nucleotide polymorphism.

Patients and Methods

Human Liver Tissue.

Human liver tissue and blood was obtained from 110 individuals of Caucasian origin undergoing liver resection (Table 1). For all patients, information was available regarding age, sex, type of hepatobiliary disease, and drug intake before liver surgery. Available presurgery liver serum parameters included alanine aminotransferase (ALT), aspartate aminotransferase, alkaline phosphatase (AP), gamma-glutamyltranspeptidase, conjugated bilirubin (cBili), albumin, and prothrombin time (Table 1). Liver tissue was shock-frozen immediately after resection and stored at −80°C. Resected tissue samples were morphologically examined, and only histologically normal liver tissue was used for further studies. The study was approved by the local ethics committee following the guidelines of the Declaration of Helsinki.

Table 1. Characteristics of Patients With Partial Hepatectomy
Indication of HepatectomyLiver Parameters
nALT*AST*AP*γ-GT*cBili*Albumin*Prothrombine time*
  • *

    Liver serum parameters from 92 patients. Values, as fold values of upper limit of normal, are given as median (Q1;Q3), except for prothrombine time which is given as percent of normal value (=100%).

Liver metastasis630.5 (0.3,1.0)0.6 (0.5,0.9)0.7 (0.5,1.3)1.0 (0.4,2.5)0.7 (0.3, 1.0)1.2 (1.1,1.3)105 (100,116)
Hepatocellular carcinoma240.8 (0.5,1.4)0.7 (0.5,1.2)0.9 (0.7,1.6)2.0 (0.9,2.6)0.7 (0.7,1.0)1.1 (1.0,1.3)102 (89,113)
Klatskin tumor91.4 (0.4;1.8)1.1 (0.7,1.6)1.8 (1.1,2.5)4.0 (2.1,6.2)7.2 (0.7, 9.0)1.1 (1.0,1.3)106 (91,108)
Cholangiocellular carcinoma40.8 (0.5,1.8)0.8 (0.7,1.3)1.3 (1.0,1.5)0.7 (0.2,3.4)0.5 (0.3,0.7)1.0 (0.2,1.2)92 (88,101)
Caroli syndrome30.4 (0.2,0.4)0.5 (0.5,0.6)0.6 (0.5,0.7)1.9 (0.8,2.5)0.7 (0.4, 0.7)1.3 (1.3,1.3)102 (95,122)
Haemangioma30.4 (0.3,4.0)0.5 (0.4,2.5)0.5 (0.4,0.8)0.5 (0.4,3.8)1.0 (0.8,1.0)1.3 (2.3,1.3)111 (86,115)
Liver trauma24. 1.391
Gallbladder carcinoma13.
Liver cyste10.
Total1100.6 (0.4,1.2)0.7 (0.5,1.1)0.9 (0.6,1.4)1.5 (0.6,3.2)0.7 (0.3,1.0)1.2 (1.3,1.0)104 (96,113)

Antibodies and Chemicals.

Antibodies for immunological detection of transporter proteins: The previously characterized antiserum against human BSEP1 was affinity purified with the oligopeptide used for immunization using the AminoLink Kit(Pierce Biotechnology,Boston MA). Commercial antibodies P3II-26, M2III-6 (Alexis, Lausen, Switzerland) and anti-P-Glycoprotein clone F4 MC-208 (Kamiya, Seattle, MA) were used to detect MDR3, MRP2, and MDR1, respectively. Horseradish peroxidase conjugated secondary antibodies were purchased from BioRad (Hercules, CA) and Amersham Biosciences (Freiburg, Germany). All chemicals were obtained commercially at the highest degree of purification.

Processing of Human Liver Tissue.

A membrane fraction enriched in canalicular marker enzymes (cMF) was isolated by a modification of a method previously described.16 All subcellular fractionation steps were performed at 4°C. A 20% homogenate (wt/vol) was prepared from a 0.7- to 1.9-g piece of thawed liver tissue sample in 0.25 mol/L sucrose supplemented with proteinase inhibitors antipain(1 μg/mL), leupeptin (1 μg/mL), and phenylmethanesulfonylfluoride (1 mmol/L), homogenized with a Polytron homogenizer (12 mm; Kinematica, Luzern, Switzerland) at 6000 rpm for 30 seconds, followed by 10 strokes in a loose-fitting Dounce homogenizer. The homogenate was filtered through two layers of prewetted cheese cloth, loaded on top of a 40% (3 mL)/20% (5 mL) sucrose step-gradient and centrifuged at 90,000gav in a TST-28.17 rotor (Kontron, Zürich, Switzerland) for 3 hours. After centrifugation, the cMF was recovered from the 40%/20% interface and subjected to enzymatic characterization (Supplemental Data; Supplementary material for this article can be found on the HEPATOLOGY website ( or stored in liquid nitrogen. Protein concentration was determined by a modified Lowry procedure.17

Immunoblot Analysis.

cMF (128 μg for BSEP, MDR3, and MDR1, 32 μg for MRP2) were denaturated for 20 minutes at 37°C using the Laemmli-system18 in the presence of dithiothreitol (3 mg/mL), antipain (1 μg/mL), leupeptin(1 μg/mL), and phenylmethanesulfonylfluoride (1 mmol/L). Proteins were separated by 7% to 15% SDS or 7.5% SDS gel electrophoresis for BSEP/MDR3/MDR1 and MRP2, respectively. Western blots were blocked for 1 hour with 5% (wt/vol) skimmed milk in Tris-buffered saline (140 mmol/L NaCl, 20 mmol/L Tris/HCl, pH 7.6) and 0.1% (wt/vol) Tween20 as blocking buffer. They were incubated for 2 hours at room temperature in the presence of appropriate dilutions of corresponding primary antibodies in Tris-buffered saline containing 0.4%(wt/vol) bovine serum albumin, followed by corresponding secondary antibodies in blocking buffer and developed with ECL-plus (Amersham Biosciences, Vienna, Austria). For MDR3 detection, BSEP blots were stripped with 100 mmol/L NaCl, 100 mmol/L glycine/HCl, pH 3.0 (30 minutes, room temperature). Transport protein expression levels were analyzed semiquantitatively by densitometry with a CAMAG TLC-ScannerII (CAMAG, Muttenz, Switzerland) and expressed individually as area under the curve (AUCsample).

Real-Time Polymerase Chain Reaction and Allelic Discrimination.

With the exception of the synonymous 3435C>T site in ABCB1, only non-synonymous single nucleotide polymorphism (SNPs) were analyzed, as they are expected to have the greatest impact on protein expression levels. The ABCB1 3435C>T was included, as it has been associated with decreased MDR1 expression in human small intestine.15 All included genetic variants are given in Table 2. DNA was isolated from blood using the QIAamp DNA Blood Minikit (Qiagen, Hilden, Germany). Genotyping of the SNPs (except for ABCB1) was performed with the Custom TaqMan SNP Genotyping Assays (Applied Biosystems, Foster City, CA). For ABCB1, primers were synthesized by MWG (MWG-Biotech AG, Ebersberg, Germany), and probes were synthesized by Applied Biosystems. For primer and probe sequences of SNPs, see Table 2. 0.625μL Probe solution (0.625 μL) and 12.5 μL 2× Universal PCR Master-Mix (Applied Bisosystems) were brought to 25 μL with 20 ng genomic DNA. PCR reaction and allelic discrimination was processed with an ABI PRISM7700 Sequence Detector.

Table 2. Primers and Probes of RealTime PCR for Allelic Discrimination of Single Nucleotide Polymorphisms (SNPs) in Whites
GeneExoncDNA PositionASNPBGenBank ReferenceAmino Acid ExchangeSense–Antisense PrimerProbesc
  • A

    cDNA sequence from GenBank accession numbers: ABCB11(NM_003742);ABCB4(NM_000443);ABCC2(NM_000392);ABCB1(NM_000927).

  • B

    SNPs are taken from GenBank accession number (*) or the literature (**)

  • c

    SNP probes were labeled with the fluorescent reporters VIC (allele 1) and FAM (allele 2).



Correlations between two transporter expression pattern were calculated with Spearman rank order correlation. Correlations between individual genotypes and normalized transporter protein expression levels were tested for significance with the Mann-Whitney U test. Differences in allelic frequencies between two phenotypes were calculated by using the 2 × 2 Fisher exact test. P less than .05 was considered significant.



Samples of histologically normal liver tissue were obtained from 110 individuals (49 females, 61 males) undergoing partial hepatectomy. The median age of the collective was 59 years (range,15-85 years). No age difference was found between women and men. Indications for liver resections are given in Table 1. Presurgery serum liver parameters were available from 92 patients (Table 1). They were normal in most patients (86% had normal ALT or AP). ALT and AST were increased twofold above the upper limit of normal range (≥2N) in 15 (16%) and 12 (13%) patients, respectively. Eighteen patients had either AP ≥ 2N or cBili ≥ 1.5N, both reflecting moderate cholestasis. Four patients had a positive serology for hepatitis B, and two patients were tested positive for hepatitis C. Presurgery medication included antihypertensives (36 patients), analgesics (16), anti-ulcer drugs (13), psychopharmaceuticals (7), estrogens (6), glucocorticoides (4), statins (3), antibiotics (3), ursodeoxycholic acid (2), and cytostatics (2).

Characterization of the Membrane Fraction Enriched in Canalicular Marker Enzymes.

The subcellular fractionation procedure yielded a cMF with enriched canalicular marker enzymes. A detailed characterization of the fractionation and the analysis of the cMF is given in the Supplementary Materials.

Western Blot Analysis.

A representative example of a Western blot of 20 individuals is given in Fig. 1. The visual inspection indicates clear differences in the band intensities, indicating interindividual differences in expression levels of the four canalicular ABC transporters (Fig. 1A). This visual impression was confirmed by densitometric scanning of the blots. The scanning data are given as the area under the curves (AUC) of the measured densitometric absorption units (Fig. 1B), whereas in Fig. 1C the measured densitometric scanning data were transformed into “normalized” AUCsample/AUCmeanper blot values (see Supplementary Materials). Comparison of the two data sets demonstrates a similar interindividual variability in transporter expression levels, supporting that the data normalization procedure does not prevent the adequate detection of variable transporter expression levels (Fig.1B-C).

Figure 1.

Expression patterns of BSEP, MDR3, MRP2, and MDR1 in cMF of 20 individuals. (A) Immunoblot detection of transporter in cMF samples of 20 different individuals. (B) Protein expression levels are expressed as area under the densitometric absorption curve (AUC). The AUCs (mean ± SD) amounted to 4088 ± 1944(48%) for BSEP, 10208 ± 2982(29%) for MDR3,5141 ± 2511(49%) for MRP2, 8004 ± 3213(40%) for MDR1. (C) Transformation of absolute AUC data into “normalized” AUCsample/AUCmean per blot values. The normalized data showed the same interindividual variability of transporter expression as the absolute data (Fig. 1B) and show mean ± 1 SD values of 1.00 ± 0.46 (46%) for BSEP; 1.00 ± 0.29 (29%) for MDR3; 1.00 ± 0.48 (48%) for MRP2, and 1.00 ± 0.47(47%) for MDR1. Numbers represent patient ID#.

Distribution of BSEP, MDR3, MRP2, and MDR1 Expression Levels in Human Liver Samples of 110 Individuals.

Before analysis of the 110 individuals, control experiments indicated acceptable reproducibility and interblotting variability (see Supplementary Materials). Normalized transporter protein expression levels, calculated as AUCsample/AUCmean per blot (Supplementary Materials), ranged between 0.13 and 2.41 for BSEP, 0.07 and 2.74 for MDR3, 0.01 and 3.65 for MRP2, and 0.15 and 3.08 for MDR1 in all 110 individuals (Fig. 2A). The corresponding mean ± SD values amounted to 1.0 ± 0.52, 1.0 ± 0.46, 1.0 ± 0.62 and 1.00 ± 0.63 for BSEP, MDR3, MRP2, and MDR1, respectively (Fig. 2A). To correct for the skewness of distribution of canalicular transporter expression levels, the normalized expression data were ln-transformed. As illustrated in Fig. 2B, this ln-transformation converged the data to a unimodal distribution. In addition, the ln-transformed data clearly indicate that some individuals exhibited very low expression of individual transporters (Fig. 2B). In the study population, the ln-transformed expression levels ranged from −1.84 to 1.17 for BSEP, from −2.28 to 1.34 for MDR3, from −3.79 to 1.78 for MRP2, and −1.70 to 1.41 for MDR1. Based on these distributions, all individuals exhibiting expression levels within the mean ±1 SD (i.e., 0.0 ± 0.57 for BSEP, 0.0 ± 0.55 for MDR3, 0.0 ± 0.79 for MRP2, 0.0 ± 0.62 for MDR1; Fig. 2B) were defined as normal expressors. Correspondingly, high expressors are defined as having expression levels above the mean + 1 SD, low expressors exhibit expression levels below the mean − 1 SD, and very low expressors below the mean − 2 SD. As a consequence, the following frequencies were found for high, low, and very low expressors of the individual transporters: BSEP, 17 (15%) high, 14 (13%) low, and 4 (4%) very low expressors; MDR3, 11 (10%) high and 13 (12%) low and 5 (5%) very low expressors; MRP2, 9 (8%) high, 11 (10%) low, and 4 (4%) very low expressors; MDR1, 16 (15%) high, 17 (15%) low, and 1 (1%) very low expressors. Thus, among the 110 individuals, 30% exhibited a high expression and 32% a low or very low expression phenotype for at least one of the four investigated transport proteins.

Figure 2.

Distribution of BSEP, MDR3, MRP2, and MDR1 expression in cMF of 110 individuals. Distribution of (A) normalized transporter expression and (B) ln-transformed normalized transporter expression (both x-axis) in liver tissue of 110 individuals (y-axis).

Differential Expression Pattern of Canalicular Transport Proteins.

Because low and very low expression levels of BSEP, MDR3, and MRP2 have the greatest functional impact for decreased (i.e., cholestatic) canalicular bile formation, we analyzed low and very low expression phenotypes of BSEP, MDR3, and MRP2 in bivariate scattergrams (Fig. 3). Only two low expressors of BSEP were also associated with low expression of MDR3 and MRP2 (#183, 198) and one very low MDR3 expressor showed also low expression of BSEP and very low expression of MRP2 (#141) (Fig. 3). Nevertheless a pronounced and statistically significant correlation was seen between the relative levels of BSEP and MDR1 (not shown) (rS= 0.51; P < .0001), BSEP and MDR3 (rS= 0.49; P < .0001) as well as between MDR3 and MDR1 (not shown) (rS= 0.47; P < .0001), whereas the correlation coefficients for MRP2-pairs were less pronounced (rS< 0.34; P < .0001) but still significant.

Figure 3.

Scattergrams of canalicular transporter expression levels in isolated cMF of 110 individuals. ln-transformed normalized transporter expression levels are plotted against each other: lnBSEPvslnMRP2 (white circle), lnBSEPvslnMDR3 (white triangle), and lnMDR3vslnMRP2 (white square). Numbers represent ID-numbers of individuals with at least one transporter with expression level outside the mean ± 1 SD range.

Correlation of Canalicular Transporter Protein Expression With Age, Sex, Cause of Liver Resection, Cholestasis, and Presurgery Medication.

No differences in transporter expression were found between young individuals (≤52) and individuals 67 years of age or older (data not shown). Overall the age of low and very low expressors ranged between 15 and 85 years and that of high expressors between 26 and 85 years. Similarly, no significant differences in expression levels were found between males and females: mean BSEP expression for males −0.02 ± 0.56 and for females 0.03 ± 0.59, for MDR3 −0.03 ± 0.66 and 0.05 ± 0.38, for MRP2 −0.07 ± 0.90 and 0.08 ± 0.62, and for MDR1 0.07 ± 0.63 and −0.08 ± 0.58, respectively. However, for MDR3 10 of 13 low and very low expressors and 8 of 11 high expressors were male, suggesting a higher, albeit statistically insignificant, variability of MDR3 expression in males than in females.

Also, comparison of transporter expression levels with the different indications for liver resection showed no significant correlations (Fig. 4A). Among 33 individuals with low expression of at least one transporter, 21 (64 %) had liver resection because of liver metastasis, 5 (15 %) because of hepatocellular carcinoma, and 2 (6 %) because of Klatskin tumor. When BSEP, MDR3, MRP2, and MDR1 expression levels were plotted against serum parameters of cholestasis (e.g., AP, cBili), no correlations between cholestasis and transporter expression levels were observed (data not shown). Eighteen cholestatic individuals with either AP ≥ 2N or cBili ≥ 1.5N demonstrated no significant differences in mean transporter expression levels as compared with individuals with no signs of cholestasis (Fig. 4B). In fact, with the exception of MDR3 and MDR1, the lowest transporter expression levels were found in non-cholestatic individuals (Fig. 4B). And finally, presurgery medication also did not correlate with transporter expression levels, because patients pretreated with either estrogens, glucocorticoides, statins, or ursodeoxycholic acid exhibited normal transporter expression with the exception of one patient treated with ursodeoxycholic acid that showed high expression of all three canalicular transporters investigated.

Figure 4.

Correlation of canalicular transporter expression levels with underlying liver disease and serum parameters of cholestasis. Individual ln-transformed normalized transport protein expression levels are plotted against (A) indication of liver resection and (B) against 13 individuals with AP ≥ 2N or cBili ≥ 1.5 (cholest.) and not cholestatic (normal).

SNPs of Canalicular Transporter Genes and Their Relationship to Protein Expression Levels.

As indicated in Table 3, the overall distribution of the allelic variants in our study population was found to be similar to previously published data in whites,14, 19 with the exception of ABCC2 variants 3600T>A(V1188E) and 4581G>A(C1515Y), which occurred with allelic frequencies of 6% in our study population as compared with the 1% recently published for a Finish collective.20 However, correlation of expression levels of transporter proteins with allelic variants of the corresponding transporter genes indicated a deviation from this distribution pattern. Specifically, for ABCB11 1457T>C(V444A), corresponding BSEP mean expression levels were 1.15 ± 0.52 for the 1457TT genotype, 0.90 ± 0.46 for 1457CC, and 1.00 ± 0.54 for 1457TC (Fig. 5A). Furthermore, 13 of 14 low BSEP-expressors exhibited a T>C variant in at least one allele (6 × CC; 7 × TC), resulting in an allelic frequency of 68% for 1457C in low expressors whereas it was 53% for normal or high expressors (P = .094; Table 3). Moreover, all very low expressors had at least one C-allele (3 × CC; 1TC), resulting in an allelic frequency of 88%. Although these polymorphic BSEP protein expression levels were statistically not different (TT vs. CC, P = .167, Fig. 5A), the presence of a 1457C-allele tended to be associated with the phenotype of low BSEP expression (Table 3, 1A). In contrast, no correlation between the second polymorphism 2155A>G (M677V) and BSEP protein expression level was found, because the M677M variant was equally distributed between individuals with low and high BSEP protein expression phenotypes (Table 3). For the ABCB4 3826A>G(R652G) polymorphism, there was a trend of lower MDR3 expression levels for the 3826AG(R652G) variant (1.03 ± 0.43) compared with 3826AA (R652R) (0.90 ± 0.56) (P = .086) (Fig. 5B), but low and high MDR3 expression phenotypes were similarly distributed between the two variants (Table 3). Among the three ABCC2 polymorphisms, the 1286G>A (V417I) was not associated with different MRP2 expression phenotypes (Table 3). However, for 3600T>A (V1188E), a significant difference (P = .006) was seen between the mean protein expression levels in the 3600TT variant (0.93 ± 0.56) and the 3600TA variant(1.36 ± 0.83) (Fig. 5C). Among nine individuals exhibiting high MRP2 expression phenotypes, four individuals had at least one 3600A-allele (1 × AA; 3 × TA), resulting in an allelic frequency of 28% for 3600T>A(V1188E) compared with 5% of normal or low expressors (P = .005), further supporting and indicating that the 3600A-allele is associated with increased MRP2 expression (Table 3). A similar distribution and association with MRP2 expression was found for the 4581G>A(C1515Y) (Table 3; Fig. 5C), indicating that the two ABCC2 SNPs 3600T>A and 4581G>A might be closely linked. Analysis of the 2 ABCB1 polymorphisms showed no correlation with protein expression levels nor with expressor phenotypes (Table 3).

Table 3. Distribution of Genotypes and Allelic Frequencies of Investigated SNPs in Individuals With Low, Normal and High Transporter Expression Phenotypes
SNPStudy populationLow expressorsANormal expressorsBHigh expressionC
  • A

    Individuals with phenotype low expressors (<mean-1SD) and very low (<mean-2SD) transporter expression levels.

  • B

    Individuals with transporter expression within mean±1SD.

  • C

    Individuals with transporter expression >mean±1SD.

1) BSEPn = 110 (100%)n = 14 (100%)n = 79 (100%)n = 17 (100%)
 Alleles (2n)220 (100%)28 (100%)158 (100%)34 (100%)
 a) ABCB111457T>C (V444A):    
  TT19 (17%)1 (7%)14 (18%)4 (24%)
  CC29 (26%)6 (43%)19 (24%)4 (24%
  TC62 (56%)7 (50%)46 (58%)9 (53%)
  Allelic frequency:    
  C-allele120 (55%)19 (68%)84 (53%)17 (50%)
 b) ABCB112155A>G (M677V):    
  AA102 (93%)14 (100%)72 (91%)16 (94%)
  AG8 (7%) 7 (9%)1 (6%)
  Allelic frequency:    
  G-allele8 (4%) 7 (7%)1 (3%)
2) MDR3n = 110 (100%)n = 13 (100%)n = 86 (100%)n = 11 (100%)
 Alleles (2n)220 (100%)26 (100%)172 (100%)22 (100%)
 ABCB43826A>G (R652G):    
  AA87 (89%)8 (62%)71 (83%)8 (73%)
  AG23 (21%)5 (38%)15 (17%)3 (27%)
  Allelic frequency:    
  G-allele23 (10%)5 (19%)15 (9%)3 (14%)
3) MRP2n = 110 (100%)n = 11 (100%)n = 90 (100%)n = 9 (100%)
 Alleles (2n)220 (100%)22 (100%)180 (100%)18 (100%)
 a) ABCC21286G>A (V417I):    
  GG64 (58%)7 (64%)51 (57%)6 (67%)
  AA1 (1%) 1 (1%) 
  GA45 (41%)4 (36%)38 (42%)3 (33%)
  Allelic frequency:    
  A-allele47 (26%)4 (18%)40 (22%)3 (17%)
 b) ABCC23600T>A (V1188E):    
  TT95 (86%)10 (91%)80 (89%)5 (56%)
  AA1 (1%)  1 (11%)
  TA14 (13%)1 (9%)10 (11%)3 (33%)
  Allelic frequency:    
  A-allele16 (6%)1 (5%)10 (5%)5 (28%)
c) ABCC2 4581G>A (C1515Y):    
  GG95 (86%)10 (91%)80 (89%)5 (56%)
  AA1 (1%)  1 (11%)
  GA14 (13%)1 (9%)10 (11%)3 (33%)
  Allelic frequency:    
  A-allele16 (6%)1 (5%)10 (5%)5 (28%)
4) MDR1n = 110 (100%)n = 17 (100%)n = 77 (100%)n = 16 (100%)
 Alleles (2n)220 (100%)34 (100%)154 (100%)32 (100%)
 a) ABCB13435C>T:    
  CC23 (21%)3 (18%)16 (21%)4 (25%)
  TT28 (25%)4 (24%)20 (26%)4 (25%)
  CT59 (54%)10 (58%)41 (53%)8 (50%)
  Allelic frequency:    
  T-allele115 (52%)18 (53%)81 (53%)16 (50%)
 b) ABCB12677G>T/A (A893S/T):    
  GG31 (28%)6 (35%)20 (26%)5 (31%)
  TT21 (19%)5 (29%)15 (20%)1 (6%)
  AA1 (1%)1 (6%)  
  GT48 (44%)4 (24%)35 (45%)9 (56%)
  GA4 (4%) 3 (4%)1 (6%)
  TA5 (5%)1 (6%)4 (5%) 
  Allelic frequency:    
  T/A-allele95/11 (43%/5%)15/3 (44%/9%)69/7 (45%/5%)11/1 (34%/3%)
Figure 5.

Relationship of BSEP, MDR3, MRP2, and MDR1 expression levels in cMF of 110 individuals and allelic variants of frequent polymorphisms in whites. Box-plot analysis of (A) normalized BSEP-expression against genetic variants of 1457T>C(V444A) and 2155A > G(M677V); (B) MDR3-expression against 3826A > G (R652G); (C) MRP2-expression against 1286G > A(V417I), 3600T > A(V1188E), and 4581G > A(C1515Y) and MDR1 against 3435C > T and 2677G > T/A(A893S/T). Mean expression of two genotypes were compared with the Mann-Whitney U-test, with P < .05 for significance.


We investigated the extent of interindividual variability in the expression of the canalicular ABC-transporters BSEP, MDR3, MRP2, and MDR1 in normal human liver tissue and correlated individual expression levels with patients' demographic data, laboratory parameters, underlying liver disease, and selected SNPs in the corresponding transporter genes. The results demonstrate a marked degree of interindividual variability of canalicular transporter expression in human liver. Furthermore, low expression of BSEP appears to associate with a specific ABCB11 gene polymorphism, indicating that SNPs of canalicular transporter genes might indeed represent a risk factor for low BSEP expression, which in turn might predispose to acquired cholestatic syndromes.

Hepatic transporter expression followed unimodal distribution in 110 investigated individuals of white origin. Eight percent to 15% of all individuals could be classified as high BSEP, MDR3, MRP2, or MDR1 expressors, 10% to 15% as low expressers, and 1% to 5% as very low expressers. Thus, baseline expression levels of canalicular ABC transporters exhibit significant interindividual variability. Interestingly, expression levels of all four transporter were significantly linked among each other, indicating that under physiological conditions these transporters may share some common factor(s) involved in regulating transcription or posttranscriptional events. This is in agreement with in vitro experiments that have identified binding sites for the transcription factor farnesoid X receptor in the BSEP, MDR3, and MRP2 (MDR1) transporter genes.21–23 This suggests that co-regulation of canalicular transporters by farnesoid X receptor ligands such as chenodeoxycholate may be of some importance during baseline physiological conditions.24, 25

Transporter expression levels were significantly influenced neither by age nor by sex, cause of liver resection, cholestasis, or presurgery medication. Therefore, the observation that gender is a risk factor for the development of pathological conditions associated with impaired bile formation such as cholelithiasis, which shows a higher incidence in women compared with men,26 cannot be related to mere differences in expression levels of canalicular ABC transporters. Surprisingly, an association also could not be found between canalicular transporter expression and cholestatic parameters in serum, although cholestasis has been shown to have an influenece on hepatocellular transporter expression. While mRNA and protein levels were preserved in primary biliary cirrhosis and chronic hepatitis C infection, MDR1 and MDR3 expression levels were increased in both conditions.27, 28 In more severe cholestasis, immunostaining for BSEP and MRP2 proteins was reduced, whereas canalicular MDR1 and MDR3 expression remained unchanged.29 However, cholestasis had to be of severe degree until downregulation of canalicular transporter expression became evident in rat and mouse liver.30, 31 Thus, the moderate cholestasis in our study population likely was not severe enough to induce downregulation of transporter expression in our study population. The fact that none of the investigated parameters could be identified as a cause of the observed interindividual variability for canalicular transporter expression suggests that genetic parameters might be more important in determining individual transporter expression levels than demographic and environmental factors.

This hypothesis is supported by the findings that polymorphisms in the corresponding transporter genes might indeed be an important determinant for BSEP and MRP2 expression, whereas polymorphic expression seems to be less important for MDR3. Specifically, a nonsynonymous BSEP polymorphism in exon13 of ABCB11(V444A) was more frequent in low and very low BSEP expressors than in individuals with normal or high BSEP expression. Whereas valine and alanine have similar chemical properties, the valine at position 444 is highly conserved in different mammalian species as well as in the little skate (Raja erinacea),32, 33 indicating that a valine at position 444 is an important prerequisite for normal BSEP function. Interestingly, two alanine-containing ABCB11 haplotypes were encountered more frequently in patients with pri mary biliary cirrhosis or associated with higher Mayo Risk Scores in patients with primary sclerosing cholangitis.34 The 444AA phenotype therefore also could be a risk factor to develop acquired cholestasis under certain challenges, such as inhibition of BSEP function by certain drugs such as cyclosporine, troglitazone, or bosentan.1, 2, 13, 35, 36 Furthermore, MRP2 expression levels were significantly higher in the presence of the two linked polymorphisms in exon 28 (V1188E) and in exon 32 (C1515Y) compared with carriers of the reference alleles. Because MRP2 mediates the canalicular secretion of estrogen and progesterone metabolites,13, 37 and because estrogen as well as progesteroneglucuronides have been shown to transinhibit BSEP,37 individuals with a combination of a 444AA low expression phenotype for BSEP and at least heterozygous 1188E and 1515Y high expression phenotypes for MRP2 could be at increased risks for estrogen- or progesterone-associated cholestatic syndromes such as intrahepatic cholestasis of pregnancy. Because the latter has been also associated with specific ABCB4 (MDR3) polymorphisms,11 decreased MDR3 and BSEP expression together with increased MRP2 levels could represent cumulative risks for the development of intrahepatic cholestasis of pregnancy. This assumption, however, remains to be validated by appropriate genetic analysis in patients with repetitive episodes of intrahepatic cholestasis during several consecutive pregnancies.

No correlation was found between hepatic MDR1 expression and ABCB1 genotypes. Although the variant alleles in position 2677 and 3435 of ABCB1 have been found to modify intestinal MDR1 expression,15, 19 the absence of such an influence in liver might be accounted for by the lack of intraindividual correlation between hepatic and intestinal MDR1 protein levels.38 Furthermore, because of the low hepatic MDR1 content compared with human small intestine,38 a genetic effect on protein expression might be mitigated by confounding factors such as underlying disease, drug treatment, or by our semiquantitative approach.

In conclusion, our data in healthy liver samples of 110 white individuals demonstrate that expression levels of four key transporters for canalicular bile formation and xenobiotic secretion follow a unimodal distribution mode. For all four transporters, individuals with high, low, and very low transporter expression levels were identified, which in the case of BSEP and MRP2 could be correlated with the presence of distinct genetic polymorphisms in the corresponding transporter genes. Genetic determinants of canalicular transporter expression therefore might contribute to the individual susceptibility to develop acquired forms of cholestatic liver diseases.