A common Dubin-Johnson syndrome mutation impairs protein maturation and transport activity of MRP2 (ABCC2)

Verena Keitel, Anne T. Nies, Manuela Brom, Johanna Hummel-Eisenbeiss, Herbert Spring, Dietrich Keppler


Absence of a functional multidrug resistance protein 2 (MRP2; symbol ABCC2) from the hepatocyte canalicular membrane is the molecular basis of Dubin- Johnson syndrome, an inherited disorder associated with conjugated hyperbilirubinemia in humans. In this work, we analyzed a relatively frequent Dubin-Johnson syndrome mutation that leads to an exchange of two hydrophobic amino acids, isoleucine 1173 to phenylalanine (MRP2I1173F), in a predicted extracellular loop of MRP2. HEK-293 cells stably transfected with MRP2I1173F cDNA synthesized a mutant protein that was mainly core-glycosylated, predominantly retained in the endoplasmic reticulum, and degraded by proteasomes. MRP2I1173F did not mediate ATP-dependent transport of leukotriene C4(LTC4) into vesicles from plasma membrane and endoplasmic reticulum preparations while normal MRP2 was functionally active. Human HepG2 cells were used to study localization of MRP2I1173F in a polarized cell system. Quantitative analysis showed that GFP-tagged MRP2I1173F was localized to the apical membrane in only 5% of transfected, polarized HepG2 cells compared with 80% for normal MRP2-GFP. Impaired protein maturation followed by proteasomal degradation of inactive MRP2I1173F explain the deficient hepatobiliary elimination observed in this group of Dubin-Johnson syndrome patients.

  • multidrug resistance protein 2
  • ATP-dependent transport
  • deficient protein maturation
  • protein trafficking

human dubin-johnson syndrome (DJS) is an autosomal recessive disorder characterized by chronic conjugated hyperbilirubinemia (5, 33, 38). Absence of a functionally active multidrug resistance protein 2 (MRP2; gene symbol ABCC2) from the canalicular membrane of hepatocytes has been identified as the molecular basis of this disorder, accounting for the impaired secretion of conjugated bilirubin and other anionic conjugates into bile (18, 21, 31, 41). MRP2 is an integral membrane glycoprotein of ∼190 kDa, localized to the apical membrane of polarized epithelia, including hepatocytes (2, 21, 23, 24,30-32, 41), where it functions as an ATP-dependent export pump for both conjugated and unconjugated amphiphilic anions (4,6, 17, 23).

Established mutations in the MRP2 gene leading to DJS are predominantly found in the 3′-proximal half of the mRNA and particularly in the exons encoding both nucleotide-binding domains (22, 31, 40-42). So far, however, no MRP2 protein was detectable in liver samples obtained from DJS patients (18, 21,31, 41). A clear relationship between the site of the mutation in the MRP2 gene and the mechanisms contributing to the absence of a functional MRP2 protein from the canalicular membrane of hepatocytes has not been established. Some DJS mutations may lead to rapid degradation of the mutant mRNA, whereas others may affect chaperone interaction, protein stability, protein maturation, apical sorting, or function of a correctly localized protein. We recently demonstrated that the absence of the MRP2 protein from the canalicular membrane in a DJS patient carrying a two-amino-acid deletion within the second nucleotide-binding domain of MRP2 (41), results from impaired protein maturation, accumulation of the mutant protein within the endoplasmic reticulum (ER), and subsequent degradation by proteasomes (19).

In extension of our previous work (19), we analyzed in the present study the consequences of two DJS-associated mutations described recently (27). On the basis of current topology predictions, the corresponding amino acid exchange is located in an extracellular loop of the third transmembrane-spanning domain of MRP2, as indicated by the transmembrane-hidden Markov model (TMHMM) program (36). An A→T substitution in nucleotide 3517 in exon 25 leads to a conservative exchange of two hydrophobic amino acids, isoleucine (I) 1173 to phenylalanine (F). The other DJS-associated mutation results in a conservative exchange of two basic amino acids, arginine (R) 1150 to histidine (H). Both mutations are more frequent in the Iranian-Jewish and Moroccan-Jewish populations and comprise the largest groups of DJS patients known to date (27, 35).

Recent work has indicated that the MRP2I1173F mutation affects membrane insertion in the nonpolarized cell line HEK-293 (27). Because apical sorting of mutant MRP2 proteins can only be analyzed in a polarized cell system, we emphasize in the present study the use of polarized human hepatoblastoma G2 (HepG2) cells to study the consequences of the MRP2I1173F mutation. We demonstrate that GFP-tagged MRP2I1173F (MRP2I1173F-GFP) was predominantly retained in the ER of polarized HepG2 cells; however, a small amount of mutant protein apparently overcame the ER quality control mechanisms and reached the apical membrane. In contrast to normal MRP2, MRP2I1173F did not mediate ATP-dependent transport of the high-affinity substrate leukotriene C4 (LTC4) into isolated ER or plasma membrane vesicle preparations. Thus even when a small amount of mutant protein reaches the canalicular membrane of hepatocytes of patients carrying the MRP2I1173F mutation, transport activity is impaired, explaining the conjugated hyperbilirubinemia observed in these patients (35).



[14,15,19,20-3H]LTC4 (4.2 TBq/mmol) and 17β-d-glucuronosyl [6,7-3H]estradiol (1.5 TBq/mmol) were from New England Nuclear Life Science Products (Boston, MA). Unlabeled LTC4 was from Amersham Biosciences (Piscataway, NJ). Nitrocellulose filters (pore size 0.2 μm) were from Schleicher & Schuell (Dassel, Germany). Leupeptin, pepstatin, protein standard mixture (M r 26,000 to 180,000), agarose, and cell culture media were from Sigma (St. Louis, MO). Geneticin (G418) was from Calbiochem (San Diego, CA). Endoglycosidase H and Fugene 6 transfection reagent were from Roche Molecular Biochemicals (Indianapolis, IN). Pfu-DNA polymerase was from Stratagene (Cedar Creek, TX). Restriction endonucleasesSfiI, SacII, and T4-DNA ligase were from Promega (Madison, WI). All other chemicals were either from Sigma or Merck (Darmstadt, Germany).


Polyclonal EAG5 and MLE antisera were raised in rabbits against the carboxy and amino terminals of human MRP2, respectively (1,2, 4). The monoclonal mouse antibodies against protein disulfide isomerase (anti-PDI) and against dipeptidyl peptidase IV (DPPIV) were from Affinity Bioreagents (Golden, CO) and Ancell (Bayport, MN), respectively. The secondary goat anti-mouse and goat anti-rabbit antibodies either coupled to Cy3 or Cy2 were from Jackson Immunoresearch (West Grove, PA). The monoclonal mouse M2III-6 antibody was from Alexis Biochemicals (San Diego, CA). The monoclonal mouse anti-vimentin antibody was provided by Progen (Heidelberg, Germany), and the monoclonal mouse anti-α-tubulin antibody was from Sigma. All antibodies were applied as described recently (19).

Cloning of the human MRP2I1173F, MRP2I1173F-GFP, and MRP2R1150H-GFP constructs.

Cloning was performed as described previously (19). Briefly, the substitution of adenine 3517 to thymine was introduced into the human MRP2 cDNA sequence (GenBank/EBI data bank accession no. X96395) by PCR. To generate the mutation, three independent PCRs were performed. PCR1 primer pairs were: sense primer, 5′-CAGTGGATGCTCATGTAGG-3′ (bases 2369–2387); antisense primer, 5′-GGCACGGAAAACTGGCAAACC-3′ (bases 3525–3505, A→T substitution underlined). PCR2 primer pairs were: sense primer, 5′-GGTTTGCCAGTTTTCCGTGCC-3′ (bases 3505–3525, A→T substitution underlined). The antisense primer for the MRP2I1173F construct was 5′-CCGCGG CTAGAATTTTGTGCTGTTCAC-3′ (bases 4638–4618 of the MRP2 sequence) containing aSacII restriction site (italic) and a stop codon (bold), the antisense primer for the MRP2I1173F-GFP construct was 5′-TGGTCCGCGGGAATTTTGT-3′ (bases 4635–4627 of the MRP2 sequence) containing a SacII restriction site (italic) but lacking the stop codon. PCR3 was run with the sense primer of PCR1 and either one of the antisense primers of PCR2 using MRP2-GFP (19) as the template. Fragments were subcloned between theSfiI and SacII restriction sites of MRP2-GFP that had been subcloned into the mammalian expression vector pcDNA3.1(+) (Invitrogen, Carlsbad, CA). For transfection into HepG2 cells, MRP2-GFP chimeras were generated. GFP was inserted into constructs lacking a stop codon using two SacII restriction sites. GFP-tagged MRP2R1150H (MRP2R1150H-GFP) was generated using the QuikChange site-directed mutagenesis kit (Stratagene), MRP2-GFP as the template, and the following primer pair: sense primer, 5′-CCAGCTGAGGCATCTGGACTCTGTCAC-3′ (G→A substitution underlined); antisense primer, 5′-GTGACAGAGTCCAGATGCCTCAGCTGG-3′. Successful cloning was verified by sequencing.

Cell culture and transfection.

HepG2 and HEK-293 cells were maintained in RPMI and minimum essential medium, respectively, as described (19). Stably MRP2-transfected HEK-293 cells (4) were cultured with the addition of G418 (0.8 mM). Fugene 6 transfection reagent was used according to the manufacturer's instructions. After transfection withMRP2I1173F cDNA, cells were selected with G418 for 3 wk. Single G418-resistant colonies were screened for MRP2I1173F protein expression by immunoblot analysis and immunofluorescence microscopy. HepG2 cells were transiently transfected with MRP2-GFP, MRP2I1173F-GFP, or MRP2R1150H-GFP as described (28). Expression of MRP2 constructs was enhanced by the addition of sodium butyrate (4) 24 h before cell harvesting (4, 19).

Preparation of crude membranes, immunoblot analysis, and deglycosylation.

Preparation of crude membrane fractions, separation by SDS-PAGE, and immunoblotting have been described (4, 19). In some experiments, cells were treated for 24 h before being harvested with tunicamycin (1 μg/ml), which prevents formation ofN-linked oligosaccharides, thus leading to synthesis of unglycosylated proteins. Membrane preparations of MRP2- andMRP2I1173F-transfected HEK-293 cells were also subjected to endoglycosidase H digestion as described (19). Core-glycosylated, immature proteins are sensitive to endoglycosidase H digestion, whereas proteins that have passed beyond the ER are insensitive to endoglycosidase H treatment.

Immunofluorescence and confocal laser scanning microscopy.

Immunofluorescence microscopy of stably transfected HEK-293 cells and HepG2 cells transiently transfected with either GFP construct was carried out by using the following antibody dilutions: 1:150 for EAG5, anti-PDI, and MLE; 1:100 for anti-vimentin; 1:200 for anti-α-tubulin; and 1:2,000 for anti-DPPIV (19).

Quantitative analysis of the subcellular localization of MRP2-GFP, MRP2I1173F-GFP, and MRP2R1150H-GFP in polarized HepG2 cells.

HepG2 cells transiently transfected with MRP2-GFP, MRP2I1173F-GFP, or MRP2R1150H-GFP were induced with 5 mM butyrate for 24 h and fixed with methanol (−20°C, 1 min) 48 h after transfection. Immunostaining with the anti-DPPIV antibody (1:300) and quantitative analysis were carried out as described (28). At least 100 transfected (as observed by GFP fluorescence) and polarized (as observed by ring-like DPPIV fluorescence) cells were counted for each transfection. Localization of the respective MRP2-GFP protein was analyzed for each transfected and polarized cell and was classified into one of three categories:1) apical localization, in which GFP and DPPIV fluorescence merged in ring-like, microvilli-lined structures between adjacent cells, i.e., the apical membrane; in addition, GFP fluorescence may also be present in intracellular structures; 2) localization in vesicles, in which the respective MRP2-GFP protein was absent from the apical membrane but present in vesicles, sometimes additionally in reticular structures; and 3) ER localization, an exclusive reticular fluorescence with absence of GFP fluorescence from the DPPIV-stained apical membrane and from vesicles. Six independent transfections were analyzed, and the percentage of each localization was calculated for each transfection.

Preparation of plasma membrane and ER vesicles and transport studies.

Plasma membrane vesicles from control HEK-293 cells and HEK-293 cells stably expressing MRP2 and MRP2I1173F, respectively, were prepared as described (4, 20, 26). Briefly, cells (∼3 × 109) were harvested from the cell culture by centrifugation (1,200 g, 10 min, 4°C) and washed twice in ice-cold phosphate-buffered saline (150 mM NaCl, 5 mM sodium phosphate, pH 7.4). After centrifugation (1,200 g) the pellet (∼5 ml) was diluted 40-fold with hypotonic buffer (0.5 mM sodium phosphate, pH 7.0, 0.1 mM EDTA) supplemented with protease inhibitors (0.1 mM PMSF, 1 μM leupeptin, and 0.3 μM aprotinin), gently stirred on ice for 1.5 h, and subsequently centrifuged at 100,000 g (40 min at 4°C). The pellet was resuspended in 20 ml hypotonic buffer, homogenized with a Braun potter S 886 (500 rpm, 4°C, 300 strokes, 2 strokes/min), diluted with incubation buffer (250 mM sucrose, 10 mM Tris · HCl, pH 7.4) and centrifuged at 12,000g for 10 min at 4°C. The resulting postnuclear supernatant was stored on ice, and the corresponding pellet was resuspended in 20 ml incubation buffer supplemented with proteinase inhibitors, homogenized by 20 strokes, and centrifuged (12,000 g, 10 min, 4°C). Both postnuclear supernatants were combined and centrifuged (100,000 g, 40 min, 4°C). The pellet was resuspended in 20 ml incubation buffer, homogenized manually by 50 strokes with a tight-fitting Dounce B (glass/glass) homogenizer on ice, and subsequently diluted with 10 ml incubation buffer and layered over 38% sucrose in 5 mM HEPES/KOH, pH 7.4. After centrifugation at 280,000g for 2 h at 4°C in a swing-out rotor, the interphases were collected, diluted in 20 ml incubation buffer, and homogenized by 30 strokes with a tight-fitting Dounce B homogenizer on ice. The suspension was centrifuged at 100,000 g (40 min, 4°C), the pellets were diluted in 1 ml incubation buffer, and vesicles were formed by passing the suspension 20 times through a 27-gauge needle with a syringe. ER vesicles were obtained by using the same procedure as for the plasma membrane vesicles, but instead of collecting the interphases after centrifugation of the homogenized membrane fraction over 38% sucrose, we retrieved the pellet from this preparation step and diluted it in 20 ml of incubation buffer. Formation of membrane vesicles and transport of [3H]LTC4 (100 nM; 0.34 TBq/mmol) and 17β-d-glucuronosyl [3H]estradiol (1.5 μM; 0.15 TBq/mmol) into the vesicles was performed as described (4,20, 26). For transport measurements into plasma membrane vesicles and ER vesicles, 50 μg and 500 μg of protein were used, respectively, in a final volume of 110 μl.

Inhibition of proteasome activity.

Proteasome function was inhibited in HEK-293 cells by the addition of MG132 at a final concentration of 2 μM 12 h before immunofluorescence microscopy (19).


Immunoblot detection and immunolocalization of MRP2 and MRP2I1173F in HEK-293 cells.

Synthesis of the MRP2 and MRP2I1173F proteins was studied in HEK-293 cells by immunoblot analysis using the MLE antiserum and the M2III-6 antibody, which recognize the amino terminus and the carboxy terminus of human MRP2, respectively. Both antibodies detected MRP2 as a mature glycoprotein of 190 kDa together with a less glycosylated form of ∼175 kDa (Fig. 1, A andD). The same molecular masses were observed for the MRP2I1173F protein (Fig. 1 B); however, the intensities of these bands differed from those seen for MRP2. Whereas MRP2 was predominantly present in the fully glycosylated 190-kDa form, the MRP2I1173F protein was mainly detected at 175 kDa, representing a core-glycosylated form of the protein. Tunicamycin prevented the formation of N-linked oligosaccharide residues leading to unglycosylated MRP2I1173F visualized at 170 kDa (Fig. 1 B). The MRP2I1173F-GFP chimera showed only one band at 205 kDa, characteristic of an immature, partially glycosylated form of MRP2-GFP, which has an apparent molecular mass of 220 kDa (Fig. 1 C). In the presence of tunicamycin, the chimeric protein had a mass of ∼200 kDa. Endoglycosidase H digestion decreased the molecular mass of the 175-kDa form of both MRP2 and MRP2I1173F, whereas the bands at 190 kDa remained unaffected (Fig. 1 D). The detection signal in all MRP2I1173F crude membrane preparations was less intense than in MRP2-containing membranes, although the same amount of total protein (50 μg) was used. We chose longer film exposure times for MRP2I1173F-containing membranes to achieve a similar intense detection signal for MRP2 and MRP2I1173F. These results suggest that a small fraction of mutant MRP2I1173F protein matures to the fully glycosylated 190-kDa form, whereas the larger amounts of MRP2I1173F and MRP2I1173F-GFP remain in full length, yet immature, core-glycosylated, ER-resident forms.

Fig. 1.

Immunoblot and immunofluorescence analysis of multidrug resistance protein 2 (MRP2) and the Dubin-Johnson syndrome (DJS) mutation MRP2I1173F in HEK-293 cells. Cells were transfected with vector pcDNA3.1(+) (Control), human MRP2 (MRP2),MRP2-GFP (MRP2-GFP), or the DJS mutation constructsMRP2I1173F and MRP2I1173F-GFP. Tunicamycin (Tunic.) was added at 1 μg/ml 24 h before cell harvesting. Crude membrane fractions (50 μg) were separated on SDS-PAGE, transferred to nitrocellulose membranes, and analyzed for MRP2 using either the MLE or the M2III-6 antibody. A: MRP2 was mainly detected as a mature glycoprotein of ∼190 kDa (arrow) and a less glycosylated form of ∼175 kDa. B: mutant MRP2I1173F protein showed one band at 190 kDa (arrow) and a more intense band at 175 kDa. To obtain a staining for MRP2I1173F of similar intensity to that seen for MRP2, we used longer film exposure times. C: MRP2I1173F-GFP was detected at ∼205 kDa, representing a partially glycosylated form of the MRP2-GFP protein that has a molecular mass of ∼220 kDa (arrow). D: endoglycosidase H (EndoH) digestion of MRP2 and MRP2I1173F led to a shift in mobility of the 175-kDa form, whereas the fully glycosylated 190-kDa forms remained unaffected.E and F: Confocal laser scanning micrographs of HEK-293 cells transfected with MRP2I1173F cDNA. Double-label experiments using the EAG5 antibody and the anti-PDI antibody showed that the MRP2I1173F protein accumulated predominantly in the endoplasmic reticulum (ER) and, in addition, in some vesicular structures near the nucleus. Bars, 10 μm.

Confocal laser scanning microscopy of MRP2I1173F-transfected HEK-293 cells, using the EAG5 antibody to detect MRP2 (Fig.1 E) and the anti-PDI antibody to visualize the ER (Fig.1 F), showed colocalization of the reticular pattern obtained from the EAG5 fluorescence with the ER staining, suggesting that a large amount of MRP2I1173F was retained in the ER of HEK-293 cells. In addition, the MRP2I1173F protein was also detected in some vesicular structures near the nucleus. In contrast, MRP2 was mainly localized to intracellular membraneous structures and to the plasma membrane of transfected HEK-293 cells as described previously (19).

Immunofluorescence and confocal laser scanning microscopy of MRP2-GFP, MRP2R1150H-GFP, and MRP2I1173F-GFP in HepG2 cells.

To analyze localization of MRP2I1173F-GFP in a polarized cell system we used human HepG2 cells, which retain hepatic polarity and form apical vacuoles corresponding to the bile canaliculus (37). To distinguish the recombinant proteins from endogenous MRP2 (3, 13,25), GFP chimeras were used. Imaging for GFP fluorescence inMRP2-GFP transfectants showed ring-like structures, representing the circumference of apical vacuoles, in addition to some vesicular structures. Apical localization of MRP2-GFP was confirmed by overlay of the GFP fluorescence with the DPPIV-immunoreactive fluorescence as a marker protein for the apical membrane of hepatocytes (10) (Fig. 2,AC). MRP2I1173F-GFP was predominantly localized in reticular structures within the cytoplasm (Fig. 2, DF) that were identified as cisternae of the ER using the anti-PDI antibody as described (19). In addition to its ER accumulation, MRP2I1173F-GFP was occasionally found in vesicles, or, rarely, the mutant protein was visualized within the apical membrane, where it colocalized with the DPPIV-immunoreactive fluorescence (Fig. 2,GI). In contrast, MRP2R1150H-GFP was predominantly localized to the apical membrane of polarized HepG2 cells (Fig. 2, JL). We then quantified the distribution of the mutant proteins among the different cellular compartments (Table 1). Normal MRP2-GFP reached the apical membrane in 80% of polarized HepG2 cells, whereas it remained in intracellular structures in 20% of polarized cells. In contrast, mutant MRP2I1173F-GFP localized to the ER in 83% of transfected and polarized cells without reaching the apical membrane (representative image in Fig. 2 F). Apical localization of MRP2I1173F-GFP, in addition to its ER accumulation, was found in only 5% of polarized HepG2 cells (representative image in Fig.2 I). Mutant MRP2R1150H-GFP reached the apical membrane in 85% of polarized and transfected HepG2 cells (representative image in Fig. 2 L).

Fig. 2.

Immunolocalization of MRP2-GFP, MRP2R1150H-GFP, and MRP2I1173F-GFP in polarized HepG2 cells. HepG2 cells were transfected with MRP2-GFP, MRP2I1173F-GFP, orMRP2R1150H-GFP and were examined for green fluorescent protein fluorescence by confocal laser scanning microscopy. Depeptidyl peptidase IV (DPPIV) served as marker protein for the apical membrane (red fluorescence in C, F, I,L). MRP2-GFP was mainly detected in the apical membrane (A), where it colocalized with the DPPIV-immunoreactive fluorescence (C), as well as in some intracellular vesicular structures (arrows). Imaging for MRP2I1173F-GFP always showed an intracellular, reticular staining, which did not colocalize with the DPPIV-immunoreactive fluorescence (F). Vesicles containing MRP2I1173F-GFP protein (arrows) were detected occasionally (G and I). Rarely was MRP2I1173F-GFP found in the apical membrane where it coincided with the DPPIV-immunoreactive fluorescence (I), in addition to its ER localization. MRP2R1150H-GFP was present in the apical membrane (J) colocalizing with the DPPIV fluorescence (L). Asterisks mark apical vacuoles. N, nuclei. Bars, 10 μm.

View this table:
Table 1.

Quantitative analysis of the subcellular localization of MRP2-GFP, MRP2I1173F-GFP, and MRP2R1150H-GFP in polarized HepG2 cells

Immunodetection of wild-type and mutant MRP2I1173F in plasma membrane and ER vesicle preparations.

Immunoblot analysis of the different membrane preparations (Fig.3 A,inset) using the EAG5 antiserum demonstrated the large amount of MRP2 in the plasma membrane fractions ofMRP2-transfected HEK-293 cells and a low level of MRP2 in the ER fractions from these cells. MRP2I1173F was mainly detected in the ER fractions. For the chosen film exposure time, the fully glycosylated form of MRP2I1173F was not visible in the plasma membrane preparations. No signal was obtained in vector-transfected HEK-293 cells.

Fig. 3.

ATP-dependent transport of [3H]leukotriene C4([3H]LTC4) (A and B) and 17β-d-glucuronosyl [3H]estradiol (C) into plasma membrane (A and C) and ER vesicles (B) from MRP2-transfected (filled circles), MRP2I1173F-transfected (filled triangles), and vector-transfected (open circles, HEK-Co) HEK-293 cells. A: plasma membrane vesicles from MRP2-transfected,MRP2I1173F-transfected, and vector-transfected HEK-293 cells (50 μg protein) were incubated with 100 nM [3H]LTC4 in the presence of 4 mmol/l ATP or 4 mmol/l 5′-AMP-PCP. B: transport of [3H]LTC4 mediated by ER fractions (500 μg protein) from MRP2-transfected,MRP2I1173F-transfected, and vector-transfected HEK-293 cells. The net ATP-dependent transport was calculated by subtracting the values measured in the presence of 5′-AMP-PCP from those obtained in the presence of ATP. Results are means ± SD from at least 6 incubations with membranes from 2 different preparations. A,inset: immunoblot of the different membrane preparations using equal amounts of protein and the EAG5 antiserum to detect MRP2. PM, plasma membrane; C: ATP-dependent transport of 17β-d-glucuronosyl [3H]estradiol (1.5 μM) mediated by plasma membrane fractions (50 μg protein) fromMRP2-transfected, MRP2I1173F-transfected, and vector-transfected HEK-293 cells. Results are means ± SD from 3 incubations.

Transport of [3H]LTC4 and 17β-d-glucuronosyl [3H]estradiol by MRP2 and MRP2I1173F.

Transport of [3H]LTC4 into ER and plasma membrane vesicles from HEK-293 cells, stably transfected withMRP2I1173F cDNA, was studied over a 3-min period and compared with transport rates measured with vesicles from MRP2 transfectants and vesicles from vector-transfected control cells. ATP-dependent transport of [3H]LTC4 by plasma membrane vesicles from MRP2I1173F-transfected and vector-transfected control cells was 13.8 ± 2.7 pmol/mg protein over 3 min and 18.7 ± 1.7 pmol/mg protein over 3 min, respectively (mean ± SD, n = 7) (Fig.3 A). The plasma membrane vesicles fromMRP2-transfected cells showed a more than fivefold higher ATP-dependent transport with 95.7 ± 13.5 pmol/mg protein over 3 min (n = 9) (Fig. 3 A).

ATP-dependent [3H]LTC4 transport by ER vesicle preparations from MRP2 transfectants was only 4.6 ± 0.8 pmol/mg protein over 3 min (Fig. 3 B) and may result from a contamination with plasma membrane vesicles. ATP-dependent transport rates obtained with ER vesicles from MRP2I1173F-transfected HEK-293 cells (0.8 ± 0.2 pmol/mg protein over 3 min) did not differ significantly from rates calculated for controls (0.84 ± 0.24 pmol/mg protein at 3 min), despite the fact that more MRP2I1173F was detected in the ER fraction of MRP2I1173F transfectants than in the ER preparation from the HEK–MRP2 cells (Fig. 3 A, inset).

Because enrichment of MRP2I1173F in plasma membrane preparations was low compared with that of MRP2, the difference observed in [3H]LTC4 transport may underestimate the activity of MRP2I1173F. Therefore, plasma membrane vesicles fromMRP2-transfected cells were mixed with those from vector-transfected cells in a ratio of 1:9, thus approximately reaching a MRP2 protein content comparable to that of plasma membrane vesicles from MRP2I1173F-transfected cells. With this experimental setup, an ATP-dependent transport rate of [3H]LTC4 of 23.5 ± 2.2, 18.3 ± 3.2, and 11.8 ± 0.6 pmol/mg protein over 3 min (n= 6) was measured in plasma membrane vesicles of MRP2-, vector-, and MRP2I1173F-transfected cells, respectively, indicating that if MRP2I1173F were to transport [3H]LTC4 it should be measurable with this experimental setup. Plasma membrane vesicles of vector-transfected cells also showed ATP-dependent [3H]LTC4transport mediated by the endogenous canine MRP2 (4).

Mutations in the MRP2 protein may cause a change of specificity for a single substrate without affecting transport characteristics of other substrates (11, 12). Therefore, transport of 17β-d-glucuronosyl [3H]estradiol by MRP2I1173F was studied, in addition to transport of [3H]LTC4. ATP-dependent transport of 17β-d-glucuronosyl [3H]estradiol by plasma membrane vesicles from MRP2-transfected cells was 219.3 ± 44.4 pmol/mg protein over 10 min (n = 3). In contrast, plasma membrane vesicles fromMRP2I1173F-transfected and vector-transfected control cells showed about fivefold lower ATP-dependent transport with 33.6 ± 17.2 and 44.0 ± 6.22 pmol/mg protein over 10 min (n = 3), respectively (Fig. 3 C).

Inhibition of proteasome function leads to the formation of aggresomes.

It has been described that inhibition of proteasomal activity may result in a paranuclear accumulation of proteins that are bound for the ubiquitin-proteasome degradation pathway (7, 15, 43). To investigate the degradation of the MRP2I1173F protein in HEK-293 cells, we inhibited proteasome function with 2 μM MG132 added 12 h before analysis by immunofluorescence microscopy (29). We detected MRP2I1173F in paranuclear, aggresome-like structures with both the EAG5 antiserum directed against the carboxy terminus of MRP2 (Fig.4 A), as well as with the MLE antiserum (Fig. 4 B), recognizing the amino terminus of MRP2, indicating that degradation of the mutant protein had not yet occurred. The normal organization of microtubules, radiating from the central microtubule organization center was distorted by the aggresome formation (Fig. 4 C). Microtubule distribution in the periphery remained unaffected. Redistribution of the intermediate filament vimentin, surrounding the aggresome structure, was also seen after MG132 treatment (Fig. 4 D).

Fig. 4.

HEK-293 cells were transfected with MRP2I1173F and incubated with 2 μM MG132 as inhibitor of proteasome activity. The mutant MRP2 protein was detected in aggresome-like structures (Ag) with both the EAG5 (green in A, C, D) and the MLE antiserum (green in B). Microtubules are distorted by the accumulation of mutant proteins, leading to a paranuclear hole where the aggresome forms. In the periphery, microtubules show normal distribution (red in C). Treatment with MG132 also changed the distribution of the intermediate filament vimentin that surrounds the aggresome structure (red in D). Nuclei were stained with propidium iodide (red in A and B). Bars, 10 μm.


Impaired secretion of conjugated bilirubin and other amphiphilic anions across the hepatocyte canalicular membrane into bile in patients with DJS is caused by the absence of functionally active MRP2 protein from this apical membrane (for reviews see Refs. 22 and23). The lack of MRP2 in DJS may be the consequence of rapid degradation of the mutant MRP2 mRNA or a defect in synthesis, stability, or sorting of the mutant MPR2 protein.

Established DJS mutations have been summarized (22, 23), and five of the eight reported mutations were either found within or in splice sites adjacent to both nucleotide-binding domains. Currently known DJS mutations include a nonsense mutation in exon 23 leading to a premature termination codon (31, 41), a missense mutation within the ABC family signature of the first nucleotide-binding domain (40, 42), and mutations affecting splice donor sites (16, 40, 42). In addition, a two-amino-acid deletion in the second nucleotide-binding domain (41) and two missense mutations in exon 25 (27) have been described. Rapid degradation of mutant MRP2 mRNA by a process termed “nonsense-mediated decay” (39) may explain the absence of MRP2 protein in DJS mutations that lead to premature termination codons (18, 21, 31, 41). Other mutations in theMRP2 gene may affect protein stability, trafficking, or transport function of the protein. Which of these mechanisms contribute to the DJS phenotype can only be determined by expression of mutatedMRP2 cDNA in a polarized human cell line. We previously analyzed the six-nucleotide deletion in exon 30 (19, 41). Impaired maturation, ER retention, and subsequent degradation of the mutant protein by proteasomes account for the absence of MRP2 from the patient's liver (19).

The aim of our present study was to identify the consequences of the DJS mutation MRP2I1173F on a molecular and cellular level using a polarized cell system. This mutation was identified in 22 patients of Iranian-Jewish origin and results from an A→T substitution at nucleotide position 3517 in exon 25 (27). Computational topology analysis locates this mutation to an extracellular loop between the transmembrane helices 14 and 15 of the third transmembrane-spanning domain of MRP2 (36). The same region contains the MRP2R1150H mutation that has been associated with DJS in five patients of Moroccan-Jewish origin (27).

Our observation that most of the MRP2I1173F and MRP2I1173F-GFP exhibit an electrophoretic mobility corresponding to core-glycosylated, immature proteins remaining sensitive to endoglycosidase H digestion (Fig. 1) indicates that most of the mutant proteins fail to advance from the ER through the Golgi apparatus. Immunofluorescence microscopy confirmed that most of the mutant MRP2I1173F protein accumulated in the ER of polarized HepG2 cells without reaching the apical membrane (Fig.2). Similarly, MRP2I1173F was present in the ER of nonpolarized HEK-293 cells (Fig. 1 and Ref. 27). However, a small amount of the mutant MRP2I1173F protein apparently progressed through the Golgi complex and reached full glycosylation (Fig. 1), which is consistent with our finding that MRP2I1173F-GFP was present in the apical membrane of a small fraction of polarized HepG2 cells (Fig. 2, Table 1). Degradation of cytoplasmic proteins, as well as of mutant membrane proteins, is carried out by proteasomes (8, 14). Inhibition of this pathway leads to an accumulation of misfolded mutant proteins in a structure near the nucleus and centriole that has been termed the aggresome (15). After inhibition of proteasomal activity with MG132, we detected full-length MRP2I1173F protein within aggresome-like structures indicating that degradation of the protein had not yet occurred (Fig. 4). A similar proteasome-mediated degradation was observed for the DJS mutation lacking two amino acids from the second nucleotide-binding domain (19).

Because neither plasma membrane nor ER vesicle preparations containing MRP2I1173F mediated ATP-dependent transport of LTC4 and plasma membrane vesicles mediated that of 17β-d-glucuronosyl estradiol, which represent high-affinity substrates for MRP2 (4), MRP2I1173F is most likely functionally not active even when it reaches the apical membrane. Despite the low amount of normal MRP2 protein in the ER vesicle preparation, a low rate of ATP-dependent transport of LTC4 was detected in the ER preparations from normal MRP2-expressing HEK-293 cells (Fig. 3). Thus the transport assay was sensitive enough to measure even low transport rates. Impaired transport activity of MRP2I1173F has been indicated by using a carboxyfluorescein efflux assay in whole cells (27). However, this assay requires a protein being present in the plasma membrane and cannot monitor transport activity of ER-resident forms of a protein. By contrast, with transport measurements using ER vesicle preparations as described in this study, transport activity of ER-resident forms of a protein can be assayed.

Interestingly, the mutant MRP2R1150H-GFP protein reached the apical membrane of polarized HepG2 cells to the same extent as did MRP2-GFP (Table 1). So far, localization studies have not been carried out in liver biopsies of patients carrying this mutation (27). If canalicular localization were observed, MRP2R1150H would be the first DJS mutation described leading to a MRP2 protein correctly localized in human hepatocytes but deficient in transport function. Impaired function of MRP2R1150H has been indicated by the use of a carboxyfluorescein efflux assay in nonpolarized HEK-cells (27).

Substrate specificity and transport characteristics of rat and human MRP2 may be altered when amino acid residue changes are introduced within or close to transmembrane segments by site-directed mutagenesis (11, 12, 34). MRP2R1150H and MRP2I1173F are the first DJS-associated mutations that have been tested to be functionally inactive (Fig. 3 and Ref. 27), and are predicted to be located in an extracellular loop of MRP2 when using the TMHMM program for topology analysis (36). Although extracellular loops of export pumps are most likely not involved in substrate binding, amino acid residues in extracellular loops may be critical for overall stability and conformation of the protein, thus contributing to proper function. Disease-associated mutations located to extracellular loops of the CFTR protein (gene symbol ABCC7), another ABC transporter that shares 27% amino acid identity with MRP2, were correctly sorted to the plasma membrane but seriously compromised chloride channel activity (9).

In conclusion, a mutant MRP2I1173F protein is synthesized in HEK-293 and in polarized HepG2 cells. Most of the mutant protein is retained within the ER and subsequently degraded by proteasomes. However, a small amount of mutant yet functionally inactive protein apparently escapes the ER quality control machinery and reaches the apical membrane. Impaired maturation of the inactive MRP2I1173F protein explains the deficient hepatobiliary elimination of amphiphilic anions resulting in the DJS phenotype of patients carrying this relatively frequent mutation.


The authors thank Drs. Gabriele Jedlitschky, Jörg König, Yunhai Cui, and Jürgen Kartenbeck, Heidelberg, for their advice and support during this research project.


  • This work was supported, in part, by the Deutsche Forschungsgemeinschaft through Grant SFB352.

  • Address for reprint requests and other correspondence: A. T. Nies, Division of Tumor Biochemistry, Deutsches Krebsforschungszentrum, Im Neuenheimer Feld 280, D-69120 Heidelberg (E-mail: a.nies{at}dkfz.de).

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • October 2, 2002;10.1152/ajpgi.00362.2002


View Abstract