Compensatory role of P-glycoproteins in knockout mice lacking the bile salt export pump


  • Potential conflict of interest: Nothing to report.


Bile salt export pump (BSEP; ATP-binding cassette, subfamily B, member 11) mutations in humans result in progressive familial intrahepatic cholestasis type 2, a fatal liver disease with greatly reduced bile flow. However in mice, Bsep knockout leads only to mild cholestasis with substantial bile flow and up-regulated P-glycoprotein genes (multidrug resistance protein 1a [Mdr1a] and Mdr1b). To determine whether P-glycoprotein is responsible for the relatively mild phenotype observed in Bsep knockout mice, we have crossed mouse strains knocked out for Bsep and the two P-glycoprotein genes and generated a triple knockout mouse. We found that a knockout of the three genes leads to a significantly more severe phenotype with impaired bile formation, jaundice, flaccid gallbladder, and increased mortality. The triple knockout mouse is the most severe genetic model of intrahepatic cholestasis yet developed. Conclusion: P-glycoprotein functions as a critical compensatory mechanism, which reduces the severity of cholestasis in Bsep knockout mice. (HEPATOLOGY 2009.)

Bile salt export pump protein (BSEP; ATP-binding cassette, subfamily B, member 11 [ABCB11], sister of P-glycoprotein [SPGP]) functions as the major bile salt export pump, translocating bile salts across the canalicular membranes in the liver.1–3 This is the rate-limiting step in the enterohepatic circulation of bile salts and a major driving force of bile flow. Cholestasis, the condition where bile flow into the intestine is significantly reduced, may be accompanied by accumulation of retained bile components in the liver and other tissues. In humans, mutations in BSEP cause a fatal cholestatic disease, progressive familial intrahepatic cholestasis type 2 (PFIC-2). These patients display reduced secretion of bile salts across the canaliculus, to less than 1% of normal, low or normal serum γ-glutamyl transferase (γ-GT),4, 5 hepatic inflammation, and dilated canalicular lumens lacking microvilli.6, 7 In contrast, bsep inactivation in mice leads only to relatively mild symptoms. Notably, the bile flow rate in these mice was not significantly reduced.3

Mice null for the Bsep gene (bsep−/−) have sharply reduced secretion of hydrophobic biliary bile salts, such as cholate (17-fold lower than wild-type), but maintain nearly normal amounts of the more hydrophilic bile salts, such as β-muricholate and Ω-muricholate and greatly increased amounts of tetrahydroxy bile acids; which are not normally detected.3 These results indicate that Bsep is the major transporter for the more hydrophobic bile salts (e.g., cholate) and suggest that some compensatory mechanism is available to maintain biliary transport of the more hydrophilic bile salts. Because expression of both P-glycoprotein isoforms, multidrug resistance protein 1a (Mdr1a) and Mdr1b, was greatly increased in the bsep−/− mice,8, 9 we postulated that P-glycoproteins play such a compensatory role.

P-glycoprotein (ABCB1) is closely related to BSEP and is best known for its ability to confer multidrug resistance (MDR) in cancer cells.10 P-glycoprotein knockout mice (mdr1a−/−/b−/−) show no defect under normal laboratory conditions, but suffer from a failure of the blood-brain barrier when challenged with a neurotoxin, ivermectin.11, 12 We reported previously that isolated membrane vesicles from a hamster ovary cell line overexpressing P-glycoprotein displays bile salt transport activity9; however, it is not known whether P-glycoprotein transports endogenous substrates under physiological conditions.

In this study, we test the role of P-glycoproteins in bile formation in mice on a bsep−/− background. We provide the first evidence that P-glycoprotein contributes to a primary function of the liver by reducing the severity of cholestasis caused by Bsep inactivation.


ABC, ATP-binding cassette; Bsep, bile salt export pump; CA, cholic acid; γ-GT, γ-glutamyl transferase; Mdr, multidrug resistance transporter; PCR, polymerase chain reaction; PFIC, progressive familial intrahepatic cholestasis; TKO, triple knockout of bsep/mdr1a/mdr1b genes; TUNEL, terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling.

Materials and Methods


As described previously, both the bsep knockout8 (maintained in this laboratory) and mdr1a−/−/b−/− mice12 (purchased from Taconic, Hudson, NY) had been backcrossed to FVB/NJ wild-types for 22 and 12 generations, respectively. Mice were maintained in a 12-hour light and dark cycle, at 22°C, with free access to food and water. All mice used were 2-4 months old except where indicated. Experiments were performed using approved protocols of the Committee on Animal Care, University of British Columbia, according to the guidelines of the Canadian Council on Animal Care.

Light and Transmission Electron Microscopy.

For light microscopy, mice were killed with CO2 after 2-4 hours of fasting. Livers were immediately removed and transferred into 10% neutral buffered formalin followed by paraffin sectioning and hematoxylin-eosin staining, Masson's trichrome staining, or apoptotic assay using the Promega (Madison, WI) DeadEnd Colorimetric terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) system. For transmission electron microscopy, livers were perfusion-fixed in situ using ice-cold 2.5% glutaraldehyde and kept in 2.5% glutaraldehyde. Dehydration, plastic-embedding, and sectioning were performed as described previously.13 One-micrometer-thick plastic-embedded sections were also obtained, and were examined in a Philips EM400T transmission electron microscope (Eindhoven, The Netherlands).

Liver Enzyme Tests.

Kits for bilirubin, alkaline phosphatase, aspartate aminotransferase, and alanine aminotransferase assays were purchased from Biotron Diagnostic Inc. (Hemet, CA), and the γ-GT assay kit was purchased from Diagnostic Chemical Ltd. (Charlottetown, Canada). Liver enzyme tests were performed according to the manufacturer's instructions.

Bile Acid Measurements of Mouse Samples.

Bile, plasma, and liver samples for bile acid analyses were collected from mice of different genotypes after bile duct cannulation. Bile duct cannulation was performed as reported.3, 8 Plasma and liver samples were collected after the mice were killed by an overdose of anesthetic. Bile acid concentrations were determined in plasma by a colorimetric assay of 3α-hydroxysteroid dehydrogenase activity (Diazyme, San Diego, CA).

Quantitative Reverse Transcription Polymerase Chain Reaction.

Liver samples from mice were used to prepare RNA as previously described.8 With the complementary DNA obtained, polymerase chain reaction (PCR) reactions were undertaken with the SYBR Green PCRMaster Mix (Foster City, CA) in a Prism 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA), using the “Standard Curve Method” (ABI-Prism User Bulletin 2). Primers used were, Ostα sense: 5′-tcatccctgacggcatctatg-3′, and antisense: 5′-ccagagggccagaagggtag-3′; Ostβ sense: 5′-gaagatgcggctccttgga-3′ and antisense: 5′-tttctgtttgccaggatgctc-3′. The other primers were as reported.8 For each sample, aliquots (5-10 ng) of total RNA were used for each reverse transcription PCR reaction, and the results were normalized against expression of ribosomal protein S15 (Rps15).

Statistical Analysis.

Data are presented as means ± standard deviation of the mean (SD). Student t test was used for comparison between groups in the tables. A chi-squared test was used for comparing mortality rates between genotypes before 30 days of age. The Mantel-Cox test was used for comparing survival curves after 30 days of age.


Generation of Triple Knockout Mice for P-Glycoproteins and Bsep.

We were initially surprised at the relatively mild consequences of Bsep inactivation in mice, compared to humans with the same defect. We therefore hypothesized the existence of an alternative bile salt transporter in mice which we considered was likely to be one or both of the P-glycoproteins (Mdr1a and Mdr1b in mice).9 In order to test the hypothesis that P-glycoproteins provide the alternative bile salt transport activity seen in bsep−/− mice, we generated triple knockout (TKO) mice carrying null mutations of mdr1a and mdr1b (co-orthologues of human MDR1 in rodents) and bsep genes by multistep crossing of bsep−/− mice3 with mdr1a−/−/b−/− double knockout mice12 (Supporting Fig. 1A). The null genotype and phenotype for these three genes in the TKO mice was confirmed in genomic PCR by the absence of a wild-type allele for mdr1a, mdr1b, and bsep (Supporting Fig. 1B) and western blotting (Supporting Fig. 2C).

Characterization of TKO Mice for P-Glycoproteins and Bsep.

The TKO mice display greater evidence of cholestasis compared to bsep−/− mice, manifesting as severe jaundice, greatly enlarged liver (Fig. 1A,B), bile duct blockage by 2 months of age, and flaccid gallbladders containing a greenish gel-like or viscous substance (Fig. 1C). The TKO mice also displayed growth retardation early in life (Fig. 4C) and higher mortality (Fig. 4A,B). Necropsies of euthanized animals, or those that spontaneously died, revealed that more than 80% of terminally ill TKO mice suffered internal hemorrhage in either their livers or intestines (Supporting Fig. 3). Similar internal hemorrhaging is observed occasionally in terminally ill bsep−/− mice at much older ages.

Figure 1.

The cholestatic phenotype of the triple knockout (bsep−/−/mdr1a−/−/b−/− TKO) mice: (A) View of abdomens showing liver enlargement and severe jaundice on the body wall and paws of a male TKO mouse in comparison with (B) a male wild-type mouse. (C) gallbladders of TKO, bsep−/−, and wild-type mice after 2–4 hours fasting. (D) Periportal fibrosis in the liver of a 2-month-old male TKO mouse where blue staining shows collagen, a sign of fibrosis. Masson trichrome staining, 600× magnification. (E) Liver histology of a TKO female fed a normal diet. An area of inflammation showing lymphocyte infiltration is visible at left (arrow). Hematoxylin and eosin (H&E) staining, 600× magnification.

Figure 4.

Mortality in TKO and bsep−/− mice. (A) Mortality before 30 days of age. The mortality rates were significantly different between TKO and bsep−/− mice (chi-squared test, P < 0.01). (B) Mortality after 30 days of age. The survival curves of TKO (bsep−/−/mdr1a−/−/b−/−), bsep−/−, and mdr1a−/−/b−/− double knockout mice fed a normal diet. The survival curve for TKO mice is significantly different from that of bsep−/− mice (Mantel-Cox test, P < 0.005). There are no statistically significant differences in the survival rate and lifespan between males and females. (C) Comparison of body weights at ages 2, 3, and 8 weeks between TKO, bsep−/−, and wild-type mice (***P < 0.001, Student t test).

Upon histological analysis, an elevated number of hematopoietic cells are readily visible in TKO pups before weaning, which suggests that they are either under greater inflammatory stress than bsep−/− pups (Fig. 2A,C), or that primitive hematopoiesis persists in the livers of these mice longer than in the other strains. Adult TKO mice show mild, but visible, periportal fibrosis (Fig. 1D), enlarged hepatocytes, and lymphocyte infiltration (Fig. 1E). By TUNEL assay, no difference in the number of apoptotic cells was observed between the livers of TKO and bsep−/−, mdr1a−/−/b−/−, or wild-type mice (data not shown). Using western blotting, we also measured some major proapoptotic and antiapoptotic proteins, i.e., caspase 3, Bax, Bcl2 and phospho-Bad (Ser112), in liver lysates of TKO mice, in comparison with those of other genotypes. It was found that in TKO and bsep−/− samples, expression of Bcl-2 was about 1.4 times that of mdr1a−/−/b−/− and wild-type samples (Supporting Fig. 4). However, the mechanism and significance of this moderately higher Bcl 2 expression is unclear. No differential expression of the other three proteins was observed between genotypes (Supporting Fig. 4). Western blotting of liver lysates from TKO mice suffering terminal illness revealed a similar expression profile (data not shown). Taken together, these results suggest apoptosis may not play a major role in the catastrophic breakdown suffered by TKO mice.

Figure 2.

Liver histology of the triple knockout (bsep−/−/mdr1a−/−/b−/− TKO) in comparison with bsep−/−, mdr1a−/−/b−/− and wild-type mice. The TKO pups display readily visible foci of cells of hematopoietic origin (arrows). A large number of vacuoles are present in the cytoplasm of TKO pups and adults (more clearly shown in Fig. 3A), which are less prominent in bsep−/−, and mdr1a−/−/b−/− mutant animals, and absent in the wild-type. (A,C,E,G) Liver sections from 13-day-old male pups of the TKO, bsep−/−, mdr1a−/−/b−/−, or wild-type lines, respectively. (B,D,F,H) Liver sections from 2-month-old male TKO, bsep−/−, mdr1a−/−/b−/−, or wild-type adults, respectively; H&E staining, 600× magnification.

Figure 3.

Subcellular changes in the hepatocytes of TKO mice. (A) Liver section of a 14-day-old TKO male pup, showing vacuoles located predominantly at the sinusoidal surface of hepatocytes. Shown is a 1-μm plastic-embedded section, toluidine blue staining, 400× magnification. (B) Part of a hepatocyte showing a large cytoplasmic vacuole at upper left containing electron-lucent material (fluid) and several dense deposits. One deposit is shown at high magnification in the inset. The deposit is composed of layers of membranes and resembles a fingerprint, which is the typical ultrastructural appearance of phospholipids. (C) Electron micrograph showing clear vacuoles at the sinusoidal surface of neighboring hepatocytes (“H”) of a 14-day-old TKO pup. A few vacuoles contain slightly flocculent material and several contain phospholipid. The spaces of Disse (“D”) are widened and contain similar material. (D) Ultrastructure of a hepatocyte of an adult male TKO mouse. On the left can be seen abnormal mitochondria of variable size with cristae pushed to one side and small ledges of cristae not crossing the midline (arrowhead). The mitochondrial matrix is homogeneous and granules are absent. On the right-hand side are greatly increased numbers of hypertrophied Golgi vesicles filled with dense material (arrow). (E) Ultrastructural changes in hepatocytes of an adult male TKO mouse, showing a dilated canalicular lumen, loss of microvilli (arrow), and retained biliary material in the form of lamellae and lipid droplets (arrowheads). (F) Normal hepatocyte ultrastructure of a male mdr1a−/−/b−/− mouse. The arrows point to normal canaliculi. Abbreviations: pss, perisinusoidal space; s, sinusoid; ics, intercellular space; ec, endothelium.

In TKO, bsep−/−, and mdr1a−/−/b−/− mouse pups, an obvious characteristic commonly found is the presence of vacuoles in the hepatocytes (Fig. 3A-C). These cytoplasmic vacuoles are located predominantly at the sinusoidal surface of hepatocytes in all lobules. Such vacuoles were not found in wild-type mice. In the TKO mice, larger vacuoles appear alongside the smaller ones found in the other mutant lines. The marked vacuolization of the cytoplasm in the TKO mice is reminiscent of transcellular and intercellular changes in Alagille's syndrome in humans, that are associated with chronic intrahepatic and extrahepatic cholestasis.14 These intracellular abnormalities together with the observed jaundice and lack of bile flow point to a transport block at the level of the canalicular membrane itself. No significant progressive changes were observed in the hepatic phenotype between pups sampled at 13 days of age and 2-month-old young adult mice. Lack of severe hepatocyte damage in these mice may reflect the generally milder cholestatic responses in rodents.15

Profound canalicular damage was observed in TKO mice, including loss of almost all microvilli, ectoplasmic thickening, and dilated canalicular spaces filled with electron-dense bile substances (Fig. 3E). There were ultrastructural changes in the cytoplasm of hepatocytes showing variable increases in smooth endoplasmic reticulum and lipid droplets (Fig. 3E), and changes in mitochondria with an increase in matrix and disorganized cristae (Fig. 3D). The Golgi apparatus displayed a great increase in vesicles, many filled with dense material of uncertain type (Fig. 3D). Distorted mitochondria, hypertrophied Golgi apparatus, increased smooth endoplasmic reticulum, excessive lipid droplets, and increased numbers of peroxisomes are cytoplasmic abnormalities typical of hepatic toxicity. Additional characteristics such as hepatic inflammation (lymphocyte infiltration) and slightly dilated canalicular spaces lacking microvilli and filled with electron-dense bile substances are consistent with what is found in human patients with PFIC-2.6

TKO pups suffer higher mortality than wild-type or even bsep−/− pups (Fig. 4A). By removing competing healthy non-TKO littermates, some TKO pups do survive to adulthood but with a median survival of about 9 weeks (Fig. 4B). The elevated mortality seen in the TKO mice after weaning appears to occur stochastically, not as the result of the mice reaching some limiting threshold or reduced lifespan per se. This is consistent with the nonprogressive nature of the underlying liver condition (see above). The adult TKO males are fertile, however, and can be used for producing TKO offspring.

Compared with bsep−/− mice, TKO mice have higher bilirubin and aspartate aminotransferase in the plasma. The level of γ-GT is low or normal (Table 1). The liver indicators of the TKO mice approximate what is found in patients with PFIC-2, who usually have very poor biliary secretion of bile salt with high bilirubin, alkaline phosphatase, and low or normal γ-GT in plasma.6, 7

Table 1. Liver Biochemical Indicators in the Plasma of bsep−/−/mdr1a−/−/b−/− Triple Knockout Mice, bsep−/−, mdr1a−/−/b−/−, and Wild-Type Female Mice Fed (A) a Normal Diet and (B) a 0.5% CA-Supplemented Diet (n = 4–7)
GenotypeBilirubin (mg/dL)ALP (U/L)γ-GT (U/L)ALT (U/L)AST (U/L)
  1. ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CA, cholic acid; γ-GT, γ-glutamyl transpeptidase; N.S., not significant. Numbers are expressed as a mean ± standard deviation (P value). Numbers in brackets indicate statistical significance determined by the Student's t test between the triple knockout and other mice in the same group.

(A) Normal diet     
TKO18.2 ± 5.45409 ± 1311.52 ± 1.4756.9 ± 35.9150 ± 59.7
mdr1a−/−/b−/−1.25 ± 0.40 (<0.001)64.8 ± 9.46 (<0.001)4.15 ± 2.71 (N.S.)63.1 ± 64.6 (N.S.)168 ± 102 (N.S.)
bsep−/−0.94 ± 0.20 (<0.001)277 ± 117 (N.S.)2.03 ± 1.01 (N.S.)58.7 ± 50.1 (N.S.)71.5 ± 29.5 (<0.05)
WT1.35 ± 0.25 (<0.001)185 ± 22.0 (<0.001)3.42 ± 2.22 (N.S.)24.2 ± 8.40 (N.S.)126 ± 82.9 (N.S.)
(B) CA diet     
TKO19.4 ± 3.34630 ± 1211.11 ± 0.52300 ± 127850 ± 296
mdr1a−/−/b−/−0.51 ± 0.13 (<0.001)180 ± 31.3 (<0.001)0.47 ± 0.31 (<0.05)105 ± 21.0 (N.S.)179 ± 66.1 (<0.01.)
bsep−/−1.47 ± 0.38 (<0.001)697 ± 216 (N.S.)0.82 ± 0.75 (N.S.)405 ± 86.5 (N.S.)525 ± 139 (N.S.)
WT1.07 ± 0.64 (<0.001)205 ± 19.7 (<0.001)1.81 ± 1.70 (N.S.)157 ± 128 (N.S.)344 ± 343 (<0.05)

Bile flow and bile acid distribution by tissue and genotype are given in Table 2. Biliary bile acid output in all the genotypes displayed an order of mdr1a−/−/b−/− > wild-type > bsep−/− > triple knockout (TKO); significant differences were present between all groups except for the comparison between bsep−/− and TKO mice (Table 2). Bile flow in TKO mice cannot be measured by gallbladder cannulation except in young TKO mice at 40 days of age where the gallbladder was not as flaccid. In these cases, the average bile flow rate observed was approximately one-quarter that produced by the wild-type but not significantly different from the bsep−/− mice which average about 50% of the wild-type (Table 2). Surprisingly, higher biliary bile acid output and lower plasma bile salt levels were also observed in mdr1a−/−/b−/− double knockout mice compared to the other genotypes, which has not been previously reported (Table 2).

Table 2. Bile Flow and Bile Acid Distribution by Tissue and Genotype. Bile Flow Rates (BFR), Bile Acid Output (BAO) Per Gram of Liver in the Equilibrated Bile, Bile Acid Concentrations in Plasma and Liver of bsep−/−/mdr1a−/−/b−/− Triple Knockout (TKO), mdr1a−/−/b−/−, bsep−/−, and Wild-Type (WT) Female Mice Fed a Normal Diet. The Bile of TKO Mice Was Obtained from Young Animals (40 Days Old)
 (A) BFR and BAO in Bile, and Bile Acid Concentration in Plasma and Liver of Mice
BFR (μL/min/g LvW)BAO (nmol/min/g LvW)Plasma (μmol/L)Liver (μmol/kg BW)
TKO0.44 ± 0.07 (4)0.289 ± 0.128 (4)13.1 ± 1.70 (4)172.1 ± 39.5 (4)
mdr1a−/−/b−/−1.58 ± 0.41 (5)26.8 ± 5.54 (4)3.06 ± 0.833 (4)115 ± 25.1 (4)
bsep−/−0.93 ± 0.40 (6)0.785 ± 0.360 (6)17.3 ± 9.01 (6)177 ± 237 (5)
WT1.67 ± 0.32 (26)16.8 ± 7.75 (14)15.1 ± 7.34 (6)115 ± 42.4 (6)
All numbers are expressed as a mean ± standard deviation (n). BW, body weight; LvW, liver weight.
(B) StudenttTests of Bile Salt Concentrations in the Bile, Plasma, and Liver Between Different Genotypes
chemical structure image

The TKO mice have a reduced tolerance for cholestatic stress. This was shown by feeding female TKO mice a 0.5% cholic acid (CA) diet, a condition that is well-tolerated by bsep−/− females8 and mdr1a−/−/b−/− double knockout mice. The TKO mice became terminally ill or died after 3–7 days of feeding (Fig. 5). These mice displayed (focal areas of) lobular disarray and hepatocyte swelling that likely result from cholestatic stress induced by CA feeding (Fig. 5C,D). The histological features resemble what was reported previously in the liver biopsies of patients with PFIC.16 In terminally ill CA-fed TKO mice, liver hemorrhage was often observed, associated with a massive area of cell death in the liver (Fig. 5E). This is in sharp contrast with the bsep−/− mice, the females of which could endure more than 100 days of the same feeding conditions without showing any terminal illness.8 Interestingly, no dramatic change in liver biochemical indicators accompanies the lethal CA-induced stress in TKO mice (Table 1B), which suggests that in the TKO mouse, these indicators are not good predictors for the severity of the underlying pathology, where increased mortality appears to result from increased likelihood of catastrophic bleeding.

Figure 5.

Survival rate (A) and body weight changes (B) of TKO (bsep−/−/mdr1a−/−/b−/−), bsep−/−, and wild-type (wt) female mice fed a 0.5% cholic acid (CA) diet. (C) Liver section of a TKO female fed 0.5% CA. Focal swelling of eosinophilic hepatocytes (arrowheads) suggests cholestatic stress caused by CA-feeding in TKO mice. Hematoxylin & eosin (H&E) staining, 400× magnification. (D) A liver section of a TKO female fed 0.5% CA. Focal swelling of eosinophilic hepatocytes (arrowheads) is shown. H&E staining, 100× magnification. (E) Liver section of a TKO female fed 0.5% CA. Hemorrhage with a large necrotic area is visible on the right (arrow). H&E staining, 200× magnification.

To explore potential compensatory changes in the hepatocytes of TKO mice, we measured gene expression profiles, using real-time PCR, in TKO, wild-type, bsep−/−, and mdr1a−/−/b−/− mice. TKO mice displayed molecular changes consistent with cholestasis, similar to the bsep−/− mice (Table 3). Multidrug resistance-associated protein 3 (Mrp3)17 and Mrp4,18 basolateral bile salt transporters for clearance of bile salt into the sinusoidal circulation, are both up-regulated, as is a gene likely to function as the major bile salt hydroxylase,19 cytochrome P450 3a11 (Cyp3a11). Decreased expression of Cyp3a41 and Cyp3a44 in TKO mice was also noted, but the biological significance of these is unknown. The expression profiles of the TKO and bsep−/− mice diverge from those of the wild-type and mdr1a−/−/b−/− mice. For example, hepatic 3-hydroxy-3-methylglutaryl coenzyme A reductase (hmgcr), the rate-limiting enzyme for cholesterol biosynthesis, was greatly reduced in mdr1a−/−/b−/− mice, whereas in bsep−/− and TKO mice, the same transcript was up-regulated, indicating higher cholesterol synthesis in bsep−/− and TKO mice. TKO mice also displayed molecular changes different from both bsep−/− and mdr1a−/−/b−/− mice. For example, Ost-a was down-regulated in the livers of both parental mutant lines, while Ost-b messenger RNAs were found in greater abundance, while neither change is pronounced in the TKO mice. The significance of this is not presently understood, but the range of changes observed in these mice will continue to be of interest for future investigation.

Table 3. Relative mRNA Expression of Some Major Liver-Expressed Genes in Wild-Type, bsep−/−, mdr1a−/−/b−/− and bsep−/−/mdr1a−/−/b−/− Triple (TKO) Knockout Mice Female Mice at 60 Days of Age as Determined by Real-Time PCR
  1. The mRNA levels were normalized against those of ribosomal protein S15. All numbers are expressed as a ratio to female wild-type mRNA, mean ± standard deviation (n = 4). The Mdr1b transcripts apparent in the TKO mouse are prematurely terminated gene products that do not encode functional protein (Supporting Fig. 2A,B).

abca11.00 ± 0.251.29 ± 0.460.788 ± 0.4090.94 ± 0.32
abcg11.00 ± 0.121.44 ± 0.421.45 ± 0.5751.37 ± 0.09 (<0.01)
abcg21.00 ± 0.091.53 ± 0.13 (<0.001)1.16 ± 0.2782.40 ± 0.25 (<0.001)
abcg51.00 ± 0.120.66 ± 0.18 (<0.05)0.883 ± 0.2110.77 ± 0.16
abcg81.00 ± 0.120.41 ± 0.15 (<0.001)0.923 ± 0.1730.38 ± 0.05 (<0.001)
acact1.00 ± 0.581.33 ± 0.561.31 ± 0.6341.09 ± 0.20
ae21.00 ± 0.130.86 ± 0.181.34 ± 0.2560.87 ± 0.09
apoa11.00 ± 0.360.60 ± 0.070.77 ± 0.1430.49 ± 0.10 (<0.05)
cyp2b101.00 ± 0.320.27 ± 0.04 (<0.01)0.762 ± 0.3860.52 ± 0.10 (<0.05)
cyp3a111.00 ± 0.432.55 ± 0.66 (<0.01)0.733 ± 0.0504 (<0.05)3.83 ± 0.45 (<0.001)
cyp3a131.00 ± 0.281.19 ± 0.261.04 ± 0.2451.37 ± 0.20
cyp3a161.00 ± 0.342.27 ± 0.32 (<0.01)0.765 ± 0.093.08 ± 0.27 (<0.001)
cyp3a251.00 ± 0.091.24 ± 0.211.18 ± 0.2051.26 ± 0.08 (<0.01)
cyp3a411.00 ± 0.290.83 ± 0.762.17 ± 0.695 (<0.05)0.19 ± 0.07 (<0.01)
cyp3a441.00 ± 0.590.07 ± 0.02 (<0.05)2.13 ± 1.140.05 ± 0.00 (<0.05)
cyp7a11.00 ± 0.482.62 ± 1.323.47 ± 1.04 (<0.01)1.06 ± 0.26
cyp8b11.00 ± 0.360.43 ± 0.23 (<0.05)2.03 ± 1.160.45 ± 0.22 (<0.05)
cyp271.00 ± 0.120.75 ± 0.250.981 ± 0.2570.56 ± 0.09 (<0.01)
fic11.00 ± 0.212.17 ± 0.90 (<0.05)0.752 ± 0.4191.45 ± 0.42
hmgcr1.00 ± 0.055.91 ± 0.59 (<0.001)0.356 ± 0.124 (<0.001)2.88 ± 0.90 (<0.01)
ldlr1.00 ± 0.441.46 ± 0.481.69 ± 0.283 (<0.05)1.07 ± 0.14
lrh11.00 ± 0.181.56 ± 0.620.964 ± 0.2270.57 ± 0.03 (<0.01)
lxr-a1.00 ± 0.120.85 ± 0.101.5 ± 0.3941.04 ± 0.08
lxr-b1.00 ± 0.130.82 ± 0.111.3 ± 0.3070.72 ± 0.08 (<0.05)
mdr21.00 ± 0.101.06 ± 0.271.09 ± 0.04570.90 ± 0.15
mrp11.00 ± 0.481.29 ± 0.421.08 ± 0.411.11 ± 0.14
mrp21.00 ± 0.131.19 ± 0.351.04 ± 0.2220.82 ± 0.09
mrp31.00 ± 0.241.35 ± 0.13 (<0.05)1.12 ± 0.06421.88 ± 0.19 (<0.01)
mrp41.00 ± 0.1414.12 ± 2.09 (<0.001)4.32 ± 1.27 (<0.01)9.60 ± 1.77 (<0.001)
ntcp1.00 ± 0.061.01 ± 0.341.24 ± 0.2630.73 ± 0.16 (<0.05)
oatp11.00 ± 0.040.18 ± 0.04 (<0.001)0.943 ± 0.2620.22 ± 0.04 (<0.001)
oatp21.00 ± 0.161.45 ± 0.341.33 ± 0.3311.29 ± 0.29
oatp41.00 ± 0.210.82 ± 0.161.84 ± 0.389 (<0.05)1.34 ± 0.30
fxr1.00 ± 0.080.87 ± 0.111.23 ± 0.2450.70 ± 0.03 (<0.001)
ppar-a1.00 ± 0.140.55 ± 0.16 (<0.01)1.25 ± 0.131 (<0.05)0.51 ± 0.10 (<0.01)
pxr1.00 ± 0.161.89 ± 0.58 (<0.05)0.632 ± 0.124 (<0.05)1.04 ± 0.08
shp1.00 ± 0.511.15 ± 0.600.877 ± 0.1391.09 ± 0.17
bsep1.00 ± 0.11201.3 ± 0.227 (0.063)0
mdr1a1.00 ± 0.204.03 ± 1.66 (<0.05)0.094 ± 0.0189 (<0.001)0.11 ± 0.03 (<0.001)
mdr1b1.00 ± 0.234.22 ± 2.00 (<0.05)2.37 ± 0.572 (<0.01)2.78 ± 0.55 (<0.01)
ost-a1 ± 0.5120.185 ± 0.103 (<0.05)0.0992 ± 0.076 (<0.05)0.851 ± 0.398
ost-b1 ± 0.48536.9 ± 18.3 (<0.001)20.5 ± 4.61 (<0.01)3.02 ± 1.57 (<0.05)


When the Bsep gene was “knocked out” in mice, they displayed a relatively mild phenotype.3 We hypothesized that the bsep−/− mice are able to “detoxify” their normal complement of bile salts by compensatory mechanisms of hydroxylation (e.g., greatly increased proportion of hydrophilic bile salts in the bile) and by up-regulating an alternative bile salt transport process (e.g., increased expression of canalicular P-glycoprotein).8, 9 To test the latter possibility, we crossed knockout mice for P-glycoproteins (mdr1a and mdr1b) to the bsep−/− mice and generated TKO mice. These mice show histological changes typical of cholestatic stress, such as jaundice (Fig. 1A,B), blockage of bile flow in early adulthood (Fig. 1C), signs of hepatic inflammation (Fig. 1D,E), and inflated canalicular spaces lacking microvilli and filled with condensed bile-like substances (Fig. 3E). They also show elevated plasma bilirubin (Table 1), higher mortality (Fig. 4A,B), and greatly reduced ability to tolerate cholestatic stress (Fig. 5). To our knowledge, the TKO mouse model developed in this article is the most severe genetic model of intrahepatic cholestasis generated to date. In other mouse models such as ones carrying mutant Fxr20 or Fic121 genes, cholestasis can only be induced by cholate feeding.

BSEP has been considered to be the sole bile salt transporter in humans, because little bile salt is detected in the bile of patients with PFIC-222 and the patients have a severe clinical presentation. However, the comparatively mild phenotype of bsep−/− mice, the mild abnormalities in bile and liver found in mdr1a−/−/b−/− mice (Tables 1–3), and the severe consequences of the loss of P-glycoproteins on the bsep−/− background, are all consistent with P-glycoprotein being involved in bile formation and biliary secretion. We find it puzzling therefore that, although the TKO mice do have a more severe histological presentation and increased mortality, they did not show significant differences in terms of bile acid profiles (Table 2) and liver enzymes (Table 1) compared to the bsep−/− mice. However, factors affecting bile flow are likely to be complex and are not fully understood. For example, even though bile salt output in bsep−/− mice is reduced to ∼5% of that of the wild-type, the mice were able to maintain a substantial bile flow rate of about 50% of that of the wild-type (Table 2). They were also able to cope with cholestatic stress imposed by CA feeding (Fig. 5). It is possible that the ability to transport bile salt represented by bsep−/− mice is so low that any further reduction in bile salt transport ability (even relatively minor) could translate into a major reduction in bile flow rate or hepatic injury. We speculate that a situation exists in TKO mice where small, intermittent, reduction in their ability to transport bile salts leads to stagnation of bile flow and severe cholestatic injury by early adulthood, presenting as flaccid gallbladders and ultrastructural abnormalities. Further studies will be required to test this hypothesis more rigorously, but such an explanation appears to account for all the observations made on the TKO mice thus far.

The cause of early mortality is not understood, although we observe a high proportion of TKO mice with internal hemorrhages (Supporting Fig. 3). This leads us to believe that the bsep−/− mice may be operating close to a critical level of bile salt overload with a reduced tolerance to cholestatic stress. The ablation of P-glycoproteins removes a critical compensatory mechanism, and is sufficient to predispose the TKO mice to development of a severe phenotype and increased mortality. The TUNEL assay and western blotting of caspase 3 and related proteins, performed on liver specimens of TKO mice, however, did not reveal any greater apoptotic activity. Liver indicator profiles and lobular disarray in the TKO mice suggest cell death was more likely necrotic in nature. We believe that in TKO mice, portal hypertension caused by cholestatic stress and necrosis lead to an earlier catastrophic breakdown in the form of hemorrhage. This conclusion is consistent with the lack of apoptosis and severe fibrosis in TKO mice. This may also explain why we did not observe progressive pathological changes in older TKO mice and why CA-fed TKO mice developed terminal illness so rapidly.

The changes observed in mdr1a−/−/b−/− mice, showing a higher bile acid output and lower plasma bile acid concentration, indicate that P-glycoprotein and Bsep act as mutually compensatory mechanisms in bile salt transport (Table 2). This codependence in the maintenance of bile homeostasis suggests the possibility that P-glycoprotein could be considered as a possible target for therapeutic manipulation in cases of liver disease. This possibility is particularly intriguing because, as noted above, at a critical level small changes in bile salt output could result in major changes in bile flow. This set of three knockout mouse models should be useful for genetically dissecting the separate influences of these loci on bile flow and bile acid homeostasis. Further experimental work will be necessary in order to uncover the molecular mechanism by which these canalicular transporters modulate liver physiology.


The authors thank Drs. David Owen and Isabella Tai for helpful comments and Chris Low for technical assistance.