Potential conflict of interest: Nothing to report.
The marked deficiency of peroxisomal organelle assembly in the PEX2−/− mouse model for Zellweger syndrome provides a unique opportunity to developmentally and biochemically characterize hepatic disease progression and bile acid products. The postnatal survival of homozygous mutants enabled us to evaluate the response to bile acid replenishment in this disease state. PEX2 mutant liver has severe but transient intrahepatic cholestasis that abates in the early postnatal period and progresses to steatohepatitis by postnatal day 36. We confirmed the expected reduction of mature C24 bile acids, accumulation of C27–bile acid intermediates, and low total bile acid level in liver and bile from these mutant mice. Treating the PEX2−/− mice with bile acids prolonged postnatal survival, alleviated intrahepatic cholestasis and intestinal malabsorption, reduced C27–bile acid intermediate production, and prevented older mutants from developing severe steatohepatitis. However, this therapy exacerbated the degree of hepatic steatosis and worsened the already severe mitochondrial and cellular damage in peroxisome-deficient liver. Both untreated and bile acid–fed PEX2−/− mice accumulated high levels of predominantly unconjugated bile acids in plasma because of altered expression of hepatocyte bile acid transporters. Significant amounts of unconjugated bile acids were also found in the liver and bile of PEX2 mutants, indicating a generalized defect in bile acid conjugation. Conclusion: Peroxisome deficiency widely disturbs bile acid homeostasis and hepatic functioning in mice, and the high sensitivity of the peroxisome-deficient liver to bile acid toxicity limits the effectiveness of bile acid therapy for preventing hepatic disease. (HEPATOLOGY 2007;45:982–997.)
Bile acids are synthesized from cholesterol in the liver and are important for the digestion and absorption of lipid nutrients. The final steps of bile acid synthesis occur in peroxisomes, where the 2-methyl branched side chain of the C27–bile acid intermediates dihydroxycholestanoic acid (DHCA) and trihydroxycholestanoic acid (THCA) is shortened by β-oxidation, leading to formation of the mature C24 bile acids cholic acid (CA) and chenodeoxycholic acid (CDCA).1 In peroxisome biogenesis disorders, the defective assembly of the peroxisomal organelle leads to accumulation of C27 bile acids and a paucity of C24 bile acids.2 The most severe of these disorders is Zellweger syndrome, which is a recessively inherited genetic disease in infants associated with neonatal hypotonia, cerebral neuronal migration defects, and severe hepatic dysfunction with cholestasis, fatty liver, and eventually cirrhosis.3 As peroxisomes contain many enzymes involved in lipid metabolism,2 absence of peroxisome assembly leads to extensive cellular metabolic disarray. In this disorder the role of C27 bile acids and/or biochemical heterogeneity in the pathogenesis of liver disease is still largely undetermined.
To gain further insight into the progression of hepatic disease and the role of bile acid defects in peroxisomal disorders, we used the PEX2−/− mouse model for Zellweger syndrome, in which homozygous mutants survive in the early postnatal period.4, 5 Clinical studies in a limited number of Zellweger patients have shown some improvement in hepatic functioning with oral bile acid treatment.6, 7 Because bile acid synthesis is feedback down-regulated by exogenous bile acids,1 bile acid feeding may reduce endogenous hepatic production of potentially toxic C27–bile acid intermediates and improve intestinal lipid absorption. In this study, we examined the progression of hepatic disease, characterized the bile acid products, and determined the effect of bile acid treatment on these parameters in postnatal PEX2−/− mice.
PEX2 control and mutant mice were analyzed on a Swiss Webster × 129SvEv genetic background as described.5 Control mice had either PEX2+/+ or +/− genotypes as biochemical and morphologic measures did not differ.8, 9 All protocols for animal use and experiments were reviewed and approved by the Institutional Animal Care and Use Committee of Columbia University.
Bile Acid Feeding.
Starting on postnatal day 1 (P1), mice were hand-fed a solution containing cholic acid and ursodeoxycholic acid (Sigma, St. Louis, MO), each at 3 mg/ml in sterile 1.5% sodium carbonate. A 15 mg/kg daily dose of each bile acid was administered by orogastric lavage using a fine polyurethane catheter tube (L-cath catheter, 28 ga, O.D. 0.4 mm, #384512, Becton Dickinson, Bedford, MA) connected to a calibrated syringe. Higher doses (20–25 mg/kg of each bile acid) caused growth retardation and a significant number of deaths in early postnatal mice of either genotype (data not shown). In older mice (>10 g), a larger polyethylene tube was used for the gavage (Intramedic, 0.024″ O.D., #204–008, Curtin Matheson Scientific, Houston, TX). Between P12 and P15, control mice were fostered to an alternate dam until weaning on P21, and young pups (∼P3–P6) were added to the litter with the PEX2−/− mice to keep the dam nursing. By P36, PEX2−/− mice were not completely weaned but were observed to be eating solid food, added in dispersed form to the cage bottom as early as P18.
Livers from P0-P36 mice were fixed in Bouin's fixative and rinsed in 70% ethanol, or were fixed in formalin. Tissues were paraffin-embedded, sectioned at 5–7 μm, and stained with hematoxylin-eosin. Bouin's fixation retains hepatic bile deposits during paraffin embedding, but these are largely extracted with formalin fixation. The number of hepatic bile deposits was quantified in 25–100 randomly chosen 40× objective microscopic fields. Bright-field images were obtained with a Zeiss Axioplan 2 microscope and an Axiocam digital camera (Carl Zeiss, Thornwood, NY).
For histologic detection of lipid and bile, mice were cardiac-perfused with 4% paraformaldehyde-PBS. The livers were postfixed overnight at 4°C in the same fixative and cryoprotected in 30% sucrose-PBS, and cryostat sections (10–12 μm) were mounted on Superfrost Plus slides (Fisher Scientific). Lipids and bile were demonstrated by Oil red O in propylene glycol and Hall's stain, respectively.10
For immunohistochemistry, cryostat sections (10 μm) were prepared from fresh-frozen or paraformaldehyde-perfused liver from P0-P36 mice. Fresh-frozen liver sections were fixed for 5–10 minutes in 4% paraformaldehyde-PBS at room temperature (for peroxisomal enzymes) or 100% methanol at −20°C (for bile acid transporters). Because fresh-frozen liver sections from P36 untreated PEX2 mutants were limited in number, antigen retrieval by microwave heating11 was used for P36 bile acid transporter studies, except for Bsep. Antibodies are listed in Supplementary Table 1, and immunostains were performed as described.5, 8 Images were obtained by confocal microscopy (Zeiss LSM 510), with similar acquisition settings for control versus mutant samples.
Mice were cardiac-perfused with 2.5% glutaraldehyde in 100 mM cacodylate (pH 7.4) and postfixed in the same buffer. Tissue processing was performed as described8 and examined on a JEOL 1200EX electron microsope.
Protein and RNA Analysis.
Western blotting of total liver homogenates (20 μg) or microsomal pellets (50 μg) was performed as described,12, 13 with minor modifications. Blocking and detection reagents for an Odyssey infrared imaging system (LI-COR Biosciences, Lincoln, NE) were used according to manufacturer's protocols. Protein loading was evaluated by immunodetection of actin (total homogenates) or sodium potassium ATPase (microsomes). Antibodies are listed in Supplementary Table 1. Signal intensities were analyzed using ImageQuant software.
RNA was extracted from livers (RNeasy kit and Oligotex mRNA kit; Qiagen, Valencia, CA) and analyzed by Northern blotting as described.9 Blots were consecutively hybridized with 32P-labeled cDNA probes and analyzed via phosphorimaging (Molecular Dynamics, Sunnyvale, CA). The probes were isolated from mouse liver total RNA by reverse transcriptase-PCR (Invitrogen) using primers based on mouse sequences available through GenBank databases (Supplementary Table 2). The amount of radioactivity in each band was quantified (Image Quant, Molecular Dynamics) and normalized to the signal generated by β-actin.
Bile Acid Analysis.
Bile acids were analyzed in plasma, liver, and bile by HPLC negative ion electrospray tandem mass spectrometry as described14 with minor modifications. A C8-HPLC column (Phenomenex Luna, 50 × 1 mm) was used with a gradient from 100% ammonium formate (pH 8.1)/ methanol (50:40 vol/vol) to 100% acetonitrile/water (90:10 vol/vol). This approach did not separate chenodeoxycholic acid and deoxycholic acid. Alternatively, bile acids were not HPLC-separated but directly scanned by total negative electrospray mass spectrometry (MS). Selection of daughter ions at mass-to-charge ratios (m/z) of 74 and 80 identifies the glycine and taurine conjugates, respectively. Plasma sample preparation was performed as described.14 Bile was diluted 1000-fold for analysis. For liver, 15–20 mg (wet weight) was sonicated in 150 μl water, followed by the addition of 150 μl methanol and 100 μl internal standard solution. The sample was resonicated, deproteinized by the addition of 750 μl acetonitrile and centrifuged at 20,000g at 4°C for 10 minutes. The supernatant was removed and stored briefly, and the pellet sonicated in 300 μl water/methanol (1:1 vol/vol) and deproteinized as described above. Combined supernatants were evaporated under a nitrogen stream and then dissolved in 250 μl water/methanol (2:1 vol/vol). Five microliters were injected.
Stool lipid analysis was performed as described.15 Fecal content was obtained from the large intestine at the time mice were killed. Stool samples were pooled from several mice of the same genotype and treatment group at each age.
For determination of total lipid content in liver, P10 livers were extracted with chloroform-methanol (2:1) and filtered through #1 paper.16 A 0.2 volume of water was added, and the organic and aqueous phases were separated by centrifugation (833g). The organic phase was re-extracted with chloroform-methanol-water (75:1200:1175) and centrifuged again. The final organic phase was placed in a preweighed tube, dried under a nitrogen stream, and total lipid weight determined (per gram of liver wet weight).
Data are expressed as means ± SD. Statistical significance was evaluated by ANOVA followed by Fisher's least-squares difference post hoc test to compare the 4 sample group means at each time. For P0 and P3 mice, an unpaired Student t test was used. Results were considered significant at P ≤ 0.05.
Bile Acid Treatment Alters Hepatic Disease Progression in PEX2−/− Mice.
Control and PEX2−/− mice were fed a mixture of cholic acid (CA) and ursodeoxycholic acid (UDCA) starting on postnatal day 1 (P1). In control mice, bile acid (BA) feeding slightly decreased survival in the first few days of life, perhaps because of difficulties in tube feeding and/or a mild toxicity of bile acids in very young mice (Fig. 1A). In PEX2−/− mice, BA therapy improved survival after P6, with a 1.5-fold increase at P10 (45% of BA-fed versus 30% of untreated mutants) and a 9-fold to at least 20-fold increase in survival of mutant mice P15 or older (Fig. 1A). The survival rate of BA-fed PEX2−/− mice remained stable after about P21. Although most mutant mice still died in the early postnatal period, the survival to P36 of approximately 9% of the BA-fed PEX2 mutants was particularly striking given the extreme rarity of untreated mutants living to this age (only 2 obtained, survival of rate < 0.5%). The full survival advantage of BA therapy has not yet been determined, because P36 mutants appeared healthy and were killed only for analysis. Newborn mice of either genotype did not tolerate CA monotherapy, and survival and growth of PEX2−/− mice was slightly better with CA and UDCA than with CA and CDCA (data not shown). Bile acid–fed mutants had increased tonus and activity compared to untreated mutants but neurologic defects such as hind-limb clasping when hung from the tail5 and abnormal righting reflex persisted (Supplementary Movies 1 and 2).
BA feeding did not affect the growth of control mice and only mildly improved the growth of PEX2−/− mice, first evident by P4 (Fig. 1B,C). However, BA-treated mutants had increased body fat (not shown) and near complete normalization of stool fat content by P9, demonstrating significant resolution of the severe steatorrhea seen in untreated mutants (Fig. 1D). Stool fat content was normal in P36 BA-fed PEX2−/− mice, but the rate of decline was slower than in control mice, likely reflecting the prolonged nursing needed to maintain these mice (see Materials and Methods section). There was also a slow decrease in stool fat content in untreated PEX2−/− mice after P9, and the stool was of normal appearance in both P36 untreated mutants (sufficient material was not available for biochemical analysis).
Early postnatal PEX2−/− mice had severe intrahepatic cholestasis, with numerous brownish-yellow deposits in bile canaliculi and small bile ducts (Fig. 2B,C,F). Cholestatic deposits were observed in newborn PEX2−/− livers, and their density rapidly increased until about P9. In untreated mutants this was followed by a decrease between P12 and P18 (Figs. 1E and 2F,H). This cholestatic pattern was observed in untreated mutants regardless of animal size, viability, or rearing with control mice, suggesting an evolution in the disease process and that the less severe phenotype in older mutants cannot be entirely explained by a less penetrant phenotype. The ductular cholestasis was often associated with cholangitis, suggesting severely reduced bile flow (Fig. 2C). Electron microscopy further identified characteristic cholestatic alterations in early postnatal PEX2 livers (Supplementary Fig. 1). BA therapy markedly reduced the number of cholestatic deposits (∼10-fold at P9) and alleviated the cholangitis in PEX2−/− mice (Figs. 1E and. 2B,D,F,G), but some canalicular damage persisted, and hepatocytes with severe cytosolic degeneration were then frequently seen (Supplementary Fig. 1F,G).
The decreased hepatic cholestasis after P9 paralleled the reduction in stool fat content in untreated mutants (Fig. 1D,E), suggesting that the bile acid pool slowly increases in early postnatal PEX2−/− mice. This raises the possibility of some residual peroxisomal import and/or bile acids obtained via coprophagy in PEX2 mutants. Immunostaining with the peroxisomal matrix marker catalase in newborn PEX2−/− livers revealed a few import-competent peroxisomes, confirmed by Western blot analysis of acyl-CoA oxidase (ACOXI) and 3-ketoacyl-CoA thiolase (Fig. 3A,B). However, in P9-P36 mutants, it was difficult to confidently identify catalase-immunolabeled, thiolase-immunolabeled, or ACOXI-immunolabeled particles (not shown). The imported protein product for ACOXI was greatly decreased and that for thiolase no longer detectable in P9 and P36 mutants, independent of BA feeding (Fig. 3C). Thus, only early neonatal PEX2 mutants synthesized mature bile acids through peroxisomal pathways, and coprophagy and/or BA feeding best explains the later accumulation.
Oil red O staining demonstrated mild hepatic steatosis in P9 untreated mutants, which was further exacerbated by BA therapy (Fig. 2I–K; Supplementary Fig. 1F). Total hepatic lipids analysis demonstrated a 1.75-fold increase in P10 untreated PEX2−/− livers versus those of the controls and an additional 2-fold exacerbation in P10 BA-fed mutants (Fig. 1F). The hepatic steatosis increased further in BA-fed mice between P9 and P18 diffusely involving the hepatic lobule (Fig. 2D,E), again independent of the viability or size of the mutant mouse.
By P36, variable degrees of steatohepatitis had developed in untreated PEX2−/− mice, with persistently fatty livers (Fig. 4H), scattered dying hepatocytes (Fig. 4B, arrow; Supplementary Fig. 2A), and a mixed neutrophilic/lymphocytic inflammatory infiltrate (Fig. 4C,E). Hyperplasia of the hepatocytes, with enlarged nuclei and abundant glassy eosinophilic cytoplasm, was evident in P36 untreated mutants (Fig. 4B,D). Bile deposits were less frequently observed and were found predominantly in hepatic sinusoids and occasionally in bile canaliculi (Fig. 4B,K–L). P36 BA-fed mutants did not have inflammatory infiltrates in the liver but still had increased hepatic steatosis relative to that in untreated mutants (Fig. 4G–I). There was significant clearing of fat in the pericentral zone in all P36 PEX2−/− mice.
Electron microscopy revealed additional hepatic degeneration, which was modified by BA feeding in older PEX2−/− mice (Fig. 5). In a P36 untreated mutant, there were numerous large vacuolated cells that predominantly clustered around central veins (Fig. 4B). These were not macrophages and did not label with cleaved caspase-3 antibody, and their vacuoles were largely devoid of neutral lipids (Supplementary Fig. 2B,D,E). Rather, these were degenerating hepatocytes with enlarged cytolysosomes containing flocculent material or circular membranes (Supplementary Fig. 2F) but also associated with tightly whorled myelin figure membranes (Fig. 5B,C). These myelin figures appeared to be bile products as they were observed around hepatocytes, in bile canaliculi, and in hepatocyte cytosol (Fig. 5A,D). In BA-fed mutants, vacuolated hepatocytes were associated with smaller autophagic vacuoles and a profound increase in smooth endoplasmic reticulum (Fig. 5F; Supplementary Fig. 3).
Mitochondrial Alterations Persist in Bile Acid–Fed PEX2 Mutants.
Previous studies demonstrated that mitochondria are abnormal in livers of Zellweger patients and mouse Zellweger models17–23 and that bile acids may mediate mitochondrial toxicity.24 Electron microscopy demonstrated mitochondrial abnormalities typically seen in peroxisomal disorders in both early postnatal and P36 untreated PEX2 mutant livers, and these defects persisted in all BA-fed PEX2−/− mice (Fig. 6). However, BA feeding was associated with significantly increased mitochondrial autophagy and increased matrix density, particularly in older mutants (Fig. 6C,H; Supplementary Fig. 3). Although the density of cleaved caspase-3 immunopositive hepatocytes was somewhat increased in P36 untreated mutant livers, caspase-3 activation was strongly increased, about 3- to 7-fold, by BA feeding in both P9 and P36 PEX2−/− livers (Table 1; Supplementary Fig. 2C). These results suggest that BA feeding in PEX2−/− mice worsens cellular damage.
Table 1. Apoptotic Cell Density in Control and PEX2 Mutant Liver
Liver sections were immunostained with a cleaved caspase-3 antibody and the total number of positive cells quantified in 3 to 6 sections per mouse. The section area was determined using Open Lab software (Improvision, Lexington, MA). The number of mice examined per group is indicated in parentheses.
One-way ANOVA was used to determine statistical significance of changes at each age followed by Fisher's post-hoc test to evaluate multiple group comparisons.
significant difference vs. untreated control mice
significant difference vs. BA fed control mice
significant difference vs. untreated PEX2−/− mice.
Bile Acid Feeding Improves Bile Acid Deficiency in Bile and Liver of PEX2 Mutants.
Bile acids were analyzed in bile and liver from untreated and BA-fed control and PEX2−/− mice that ranged in age from P12 to P36. In untreated control mice (age < P18), the major bile acids in the bile were tauro-CA, tauro-muricholic acid (muriCA), and tauro-hydroxycholic acid (OH-CA; Fig. 7). The mass spectra of bile from BA-fed control mice on P12 and P36 were similar to those of untreated controls except for increased amounts of tauro-UDCA (or t-DCA or t-CDCA, but less likely because mice were treated with UDCA; Fig. 7A,C,E,G). In bile acid–fed controls, the C24–bile acid concentration increased 50%, whereas no alteration in C27 bile acids was observed (Fig. 8A–C). In contrast, 60% of all the bile acids in the bile of untreated PEX2−/− mice (age < P18) were unconjugated C27-bile acid intermediates, largely OH-THCA and THCA, and the total bile acid concentration was only 2.6% of that in the untreated controls (Figs. 7B and 8A–C). The remaining major bile acid species were tauro-CA, tauro-muriCA, and tauro-OH-CA, as seen in the control mice. Note that the taurine-conjugated C27 bile acids were extremely low in concentration or undetectable in the bile of mutant mice (detected at m/z = 540, 556, and 572). In P36 untreated PEX2−/− mice, there was only a slight increase in total biliary bile acid concentration (10.8 ± 0.5 mmol/l) versus that in the P14-P16 untreated mutants (5.2 ± 3.0 mmol/l), largely because of a greater amount of C24 bile acids at P36 (making up 71% of total bile acids; Fig 8A–C).
BA therapy in PEX2−/− mice markedly reduced the relative contribution of unconjugated C27–bile acid intermediates (4.2% at both ages) compared to that in the untreated mutants (60%–29%) (Fig. 7B,D,F,H) but did not significantly reduce their concentration in bile (Fig. 8C). BA feeding of PEX2 mutants caused an 8-fold to 20-fold increase in C24 bile acids on P14-P16 and P36, respectively (Figs. 7D,H and 8B). However, the total biliary bile acid concentration in P36 BA-fed PEX2−/− mice was still only 20%–30% of that in untreated control mice. In addition, unconjugated CA/muriCA (detected at m/z = 407.3) was now the predominant species in bile from BA-fed mutants, whereas in controls it was a very minor species (Fig. 7E–H). These unconjugated (free) C24 bile acids increase dramatically from P14 to P36 in BA-fed PEX2 mutants (Fig. 8D). There was a consistent decrease in the degree of C24–bile acid conjugation in the bile of untreated PEX2−/− mice, which was further reduced with BA feeding (Fig. 6E), suggesting that conjugation capacity is limiting. Thus, there was a persistent defect in hepatic conjugation of both C24 and C27 bile acids in PEX2−/− mice.
In the livers of untreated PEX2−/− mice, the accumulation of C27–bile acid intermediates was also accompanied by a severe deficiency in C24 bile acids (Fig. 8F–H; Supplementary Table 3). On P12, C24 bile acids in untreated control livers account for 99.6% of all the bile acids. In contrast, C24 bile acid level in the livers of the P12 untreated PEX2−/− mice was only 6% of that in the controls and accounted for only 20% of all bile acids, most of which was C27 bile acid. Total bile acid concentration in the livers of P12 untreated PEX2−/− mice was only 30% of that in the controls. Both total bile acid and C27–bile acid concentrations were similar in untreated PEX2−/− livers on P12 and P36 (Fig. 8F,H). However, C24 bile acid level was 3.3-fold greater on P36 than on P12. This increased percentage of C24 bile acids in the P36 untreated mutant livers (now 51% of total bile acids) mirrored the increase in C24 bile acids in the bile of untreated mutants between P12 and P36.
BA feeding of the control mice increased the hepatic C24–bile acid concentration 2.1-fold on P12 versus that of the untreated controls, but C24 levels did not differ significantly on P36 (Fig. 8G). In P12 PEX2−/− livers, BA feeding normalized total and C24 bile acid concentrations to untreated control levels and reduced C27 bile acid level 2-fold. A similar pattern was observed in P36 BA-fed PEX2−/− mice (Fig. 8F–H, Supplementary Table 3). Despite the normalization of hepatic C24 bile acid levels in BA-fed PEX2−/− mice, reduction in the biliary bile acid concentration persisted (Fig. 8A,B). In addition, approximately 65% of all C24 bile acids were unconjugated in the livers of P36 BA-fed mice (Fig. 8I–J), which also was observed in the bile of these mutants (Figs. 7H and 8D). It is not understood why analysis showed C24 conjugation in the liver and bile only concurred in P36 BA-fed mutants. This may reflect worsening hepatic disease and limits to conjugation capacity by increased bile acid content in older treated mice or erroneous contamination of liver samples with bile in cholestatic livers.
Bile Acids Accumulate in Plasma of PEX2−/− Mice.
Bile acids were analyzed in plasma from untreated and BA-fed control and PEX2−/− mice, at ages ranging from P0 to P36. BA treatment of control mice initially causes a significant increase in total- and C24-bile acid levels in plasma between P9 (not shown) to P12 (Fig. 9A,B), but plasma bile acid level began to decrease by P14-P18 in all control mice reaching low levels by P36. On P0 (Fig. 9) and P3 (not shown), the plasma of untreated PEX2−/− mice showed the expected pattern, with the accumulation of C27 bile acids and a deficiency of C24 bile acids. Again, most of the plasma bile acid species in the untreated PEX2−/− mice were C27–bile acid intermediates (∼77% of total) that were nearly all unconjugated. BA feeding of P6 PEX2 mutants initially produced an approximately 4.5-fold reduction in the average plasma concentration of C27 bile acids (Fig. 9C, Supplementary Table 4). However, the total plasma bile acid concentration in P6 untreated PEX2−/− mice became similar to that in the untreated control mice (Fig. 9A) and continued to increase in both the untreated and BA-fed PEX2 mutants reaching high levels by P18 (∼60–80 μmol/l). C27–bile acid levels were also no longer significantly reduced in P9-P15 BA-fed mice versus in untreated mutants (Fig. 9C, Supplementary Table 4). Nonetheless, BA feeding did markedly alter the profile of accumulating plasma bile acids, which were predominantly unconjugated C27 species in untreated mutants (77% of total bile acids) and free-C24 bile acids in BA-fed PEX2 mutants (∼60% of total bile acids; Supplementary Table 4; Fig. 9D).
Altered Expression of Hepatic Bile Acid Transporters in PEX2−/− Mice.
The reduced BA content in the liver and bile of PEX2−/− mice was consistent with the low capacity of the PEX2−/− livers for BA synthesis. This led us to examine whether the hypercholanemia was caused by defective hepatic excretion of bile acids.
The major hepatocellular transport system for conjugated bile acids uses the sodium-dependent sodium-taurocholate cotransporting polypeptide Ntcp (SLC10A1) for basolateral uptake and the bile salt export protein Bsep (ABC11) for canalicular excretion.25 Although Ntcp expression was normal in the livers of P0 PEX2−/− mice (not shown), protein expression was significantly reduced in P10 untreated mutants and was further decreased with BA feeding (Fig. 10A). Ntcp protein was similarly reduced in P36 PEX2 mutants (Fig. 10B). Immunohistochemistry demonstrated an uneven distribution of Ntcp reduction across the hepatic lobule, with strong decreases around cholestatic deposits (Fig. 10C,D). For Bsep, protein expression and canalicular membrane staining were strongly increased, and expression was normalized with BA feeding in early postnatal PEX2 mutants (Fig. 10A,E). By P36, Bsep protein increased when control mice were fed bile acids but remained low in PEX2 mutants (Fig. 10B,F). Bsep immunohistochemistry also demonstrated highly irregular canalicular membranes in all PEX2 mutants.
Bile acid conjugates are actively excreted from the hepatocytes at the basolateral surface by multidrug-resistance-associated proteins (e.g., Mrp3, ABCC3)26, 27 and at canalicular membranes by Mrp2 (ABCC2).28 Immunohistochemistry revealed persistently increased Mrp3 expression in all PEX2 mutants, regardless of age or BA treatment (Fig. 11A,B), which was confirmed by Northern blot analysis (not shown). The changes in Mrp2 expression in P12 PEX2 livers (Fig. 11C) were similar to those for expression of Bsep (Fig. 10E), whereas on P36, Mrp2 levels were normal in PEX2−/− livers (Fig. 11D). As with Bsep, Mrp2 immunostains displayed canalicular membrane irregularity in all mutant livers.
Basolateral uptake of unconjugated bile acids in the liver is mediated by organic anion transporting polypeptides (OATP/SLCO) in sodium-independent pathways.25 Northern blot analysis revealed that by P36 the major Oatp1 transporter was expressed in control mice but not in BA-fed PEX2 mutants (Fig. 12). Oatp2 mRNA was strongly down-regulated in all PEX2 mice, independent of age or treatment, but there was greater preservation of Oatp4 mRNA expression.
This study demonstrated that peroxisome-deficient mice develop severe postnatal liver disease, with transient intrahepatic cholestasis in the juvenile period that progresses to steatohepatitis. BA treatment of PEX2−/− mice alleviated the cholestasis, enabling a subset of PEX2 mutants to survive past the early postnatal period and preventing the onset of steatohepatitis by P36. Although there was a favorable reduction in C27–bile acid intermediates with BA feeding, hepatic steatosis was exacerbated, and treated mutants still accumulated bile acids in their plasma. The hypercholanemia in PEX2 mice was caused by changes in bile acid transporter expression that limited hepatic excretion of bile acids. Both hepatic transport defects and cellular damage worsened in BA-treated mutants and was potentiated by defective bile acid conjugation. Thus, a high sensitivity of peroxisome-deficient liver to bile acids along with their defective conjugation limits the effectiveness of BA therapy for preventing hepatic disease.
Despite a relatively mild BA feeding regimen, all PEX2−/− mice accumulated predominantly unconjugated bile acids in plasma in the early postnatal period (Fig. 9). This could not be explained by biosynthesis because: (1) expression of the classic pathway CYP7A1 gene is reduced 0.5-fold in early postnatal untreated mutants9 and remains strongly suppressed in all P36 mutants, and (2) bile acid synthesis genes are down-regulated by bile acid treatment in PEX2−/− mice (P. Faust, manuscript in preparation). Rather, our study showed that changes in basolateral bile acid transporter expressions, including decreased Ntcp, increased Mrp3, and decreased/absent Oatps, acted to exclude bile acids and their conjugates from PEX2−/− hepatocytes, requiring excretion from alternate renal pathways29 that may also be limiting. The inhibition of basolateral uptake even occurred in untreated mutants and despite a bile acid synthesis defect, suggesting that the peroxisome-deficient liver is highly sensitive to bile acid toxicity.
Canalicular bile acid transport was also abnormal in PEX2 mutants. In early postnatal untreated mutants, expression of Bsep and Mrp2 canalicular transporters was increased but biliary bile salt concentration was low (2.6% of normal). This suggests that Bsep function is limited by damage to canaliculi and low bile acid concentration in hepatocytes. A reduced level of biliary bile salt persisted in BA-fed mutants (∼20%–30% of normal) despite normalization of hepatic C24 bile acid levels, reflecting reduced Bsep expression, persistent canalicular damage, and a limit to maximal conjugation capacity with BA feeding in PEX2−/− mice. Recent studies demonstrated that Bsep will also transport unconjugated bile acids,30 which was supported by our findings.
The bile acid transporter alterations clearly failed to prevent the toxicity from bile acids in PEX2−/− livers. Electron microscopy in untreated mutants identified cytosolic membranous material, suggestive of biliary products, that is associated with cellular degeneration (Supplementary Fig. 1D; Fig. 5D). In the P36 mutants, these had a distinct myelin-figure-like morphology and most likely represent phospholipid crystals31 that precipitate because of the extremely low bile salt concentrations. In early postnatal BA-fed mutants, hepatocytes with marked cytosolic necrosis were frequently seen (Supplementary Fig. 1G), and mitochondrial autophagy was increased. Autophagy was particularly pronounced in older treated mutants, where it was associated with profound proliferation of smooth endoplasmic reticulum. BA feeding increased hepatocyte caspase-3 activation, suggesting that it worsened the already severe mitochondrial and cellular damage in peroxisome-deficient livers.
The bile acid analyses clearly demonstrated defects in the conjugation of C24 and C27 bile acids to taurine in PEX2−/− mice at all ages. In mice lacking the peroxisomal D-bifunctional protein (DBP), C27 conjugation was similarly defective in the juvenile period but normal in adult mice.32 The persistence of conjugation defects in P36 PEX2 mutants may reflect delayed maturation of the peroxisomal/cytosolic-located bile acid-CoA:amino acid N-acyltransferase enzyme (BAAT), which conjugates bile acids to taurine or glycine and increases around weaning in control mice,32 but other conjugation enzymes may also have been affected.33, 34 Interestingly, patients with BAAT gene mutations also accumulated unconjugated bile acids in plasma.35 Initial studies in which PEX2 mutants were fed conjugated bile acids, rather than unconjugated ones, have not shown any difference in growth or survival (P. Faust, unpublished data). Future investigations will further explore the efficacy of conjugated bile acid therapy.
Newborn PEX2 mutants could synthesize some mature C24 bile acids because of residual peroxisomal import but this largely disappeared by P9. This functional peroxisomal loss correlated with the rapid postnatal demise of PEX2 mutants and the onset of fatty liver and other potential metabolic disturbances that may have affected peroxisome assembly.36 Although the mechanism for this peroxisomal import in vivo in the absence of PEX2 protein is not understood, in vitro transfection with the peroxisomal transporters ABCD3 (PMP70) or ABCD2 (ALDR) restores peroxisome biogenesis in PEX2−/− cells, indicating functional complementation by these proteins.37, 38 The accumulation of C24 bile acids in older untreated PEX2 mutants is best explained by coprophagy, which normally begins about P1339 but may be delayed in PEX2 mutants. The large increase in C24-unconjugated bile acids in older BA-fed PEX2 mutants was also treatment related, compounded by BA transporter down-regulation and limitations to maximal conjugation capacity.
Mice deficient in enzymes of the peroxisomal bile acid synthesis pathway, including DBP32 and α-methylacyl-CoA racemase (Amacr),40 also accumulate C27 bile acids to a similar degree and have C27-conjugation defects, but fail to show the marked hepatic disease seen in the PEX2−/− mice. These findings suggest that the more widespread biochemical disturbances associated with peroxisome assembly defects exacerbate the toxicity of bile acids and potentiate hepatic disease. The PEX2−/− model system offers a unique opportunity to evaluate the pathogenesis of peroxisomal liver disease and to understand the complex regulation of genes that control the synthesis, conjugation, and transport of bile acids.
We thank A. Moser for assistance in performing liver Folch extractions; H. Shio for technical assistance with electron microscopy; P. Lazarow, R. Rachubinski, J. Reddy, M. Anathanarayan, and J. Boyer for providing critical antibody reagents; and S. Krisans, P. Watkins, and S. Mihalik for helpful discussions.