Potential conflict of interest: Nothing to report.
Interindividual variation exists in response to statin therapy. It has been hypothesized that subjects with higher baseline cholesterol synthesis rates are more sensitive to statins. To directly test this hypothesis, mice overexpressing the heterodimeric ATP-binding cassette (ABC) transporter G5/G8 (G5G8Tg mice) were treated with lovastatin because they have a compensatory increase in cholesterol biosynthesis as a result of increased cholesterol excretion into bile and feces. As expected, lovastatin treatment did not alter plasma and hepatic cholesterol levels in wild-type mice. Interestingly, this treatment significantly reduced plasma concentration and hepatic content of cholesterol by 42% and 17.3%, respectively, in the statin-treated versus untreated G5G8Tg mice despite a greater feedback upregulation of genes in the pathway of cholesterol biosynthesis in the lovastatin-treated G5G8Tg mice. The reduced plasma cholesterol concentration is unlikely to be attributed to LDL and HDL receptors because the protein levels of both receptors remained unchanged. Surprisingly, statin treatment resulted in an increase in biliary cholesterol concentration, which was associated with an upregulation in hepatic mRNA and protein levels of ABCG5 and ABCG8, and in hepatic mRNA levels of Niemann-Pick C1-Like 1 (NPC1L1), a gene that is required for intestinal cholesterol absorption. In conclusion, mice with higher endogenous cholesterol synthesis rates are more sensitive to statin. A synergistic hypocholesterolemic effect could be potentially achieved in humans by simultaneously inhibiting cholesterol biosynthesis and promoting ABCG5/ABCG8-mediated cholesterol excretion. (HEPATOLOGY 2006;44:1259–1266.)
High blood cholesterol concentration is an independent risk factor for coronary heart disease, the leading cause of death in developed countries. Statins inhibit 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase,1 a rate-limiting enzyme in the pathway of cholesterol biosynthesis. Inhibition of cholesterol biosynthesis lowers plasma cholesterol by raising the amount of low-density lipoprotein (LDL) receptor (LDLR)2 and by reducing the secretion of apolipoprotein B (apoB)-containing lipoproteins from the liver,3 thereby reducing morbidity and mortality from coronary heart disease in both primary and secondary prevention trials.4–10 In spite of these beneficial effects, considerable interindividual variation exists in response to statin treatment.11 Miettinen's group reported that the subjects with higher ratios of serum cholestanol and plant sterols to cholesterol (surrogate marker of cholesterol absorption) at baseline were less responsive to statin than those with lower respective ratios, and baseline cholesterol metabolism was speculated to regulate statin responsiveness.12, 13
Genetic basis plays an important role in drug responses.14 Genotypes and polymorphisms of many genes are involved in interindividual variation in response to statins.11 It was recently reported that in patients with hypercholesterolemia, ATP-binding cassette (ABC) transporter G8 (ABCG8) D19H variant is associated with greater LDL-cholesterol lowering response to atorvastatin therapy.15 ABCG8 heterodimerizes with ABCG5 to transport cholesterol from the hepatocytes into the bile and probably from enterocytes into intestinal lumen.16–18 Mutations in either ABCG5 or ABCG8 cause sitosterolemia,19, 20 an autosomal recessive sterol disorder characterized by the increased plasma and tissue levels of sitosterol and campesterol, the two major plant sterols in diet, as a result of increased dietary absorption and impaired biliary secretion of these sterols.21–23 Mice lacking abcg5/abcg8 genes recapitulate the major phenotypes of human sitosterolemia.16 On the other hand, transgenic mice overexpressing human ABCG5/ABCG8 in the liver and small intestine (G5G8Tg mice) retain much less plant sterols in the body and have a compensatory upregulation of hepatic cholesterol synthesis rates due to increased biliary secretion and reduced intestinal absorption of cholesterol.17 In human subjects, carriers of the ABCG8 D19H variant have a significantly lower ratio of plasma cholestanol/cholesterol when compared to noncarriers,24 suggesting that D19H may be a “gain of function” variant of ABCG8 gene. We hypothesized that stimulation of ABCG5/ABCG8 function would increase statin sensitivity, and that animals with high baseline cholesterol synthesis rates would be more responsive to statins. To directly test these hypotheses, the relationship between cholesterol synthesis and statin responsiveness of plasma cholesterol was examined in G5G8Tg mice.
HMG-CoA, 3-hydroxy-3-methylglutaryl coenzyme A; LDLR, low-density lipoprotein receptor; apoB, apolipoprotein B; ABC, adenosine triphosphate-binding cassette transporter; G5G8Tg, adenosine triphosphate-binding cassette transporters G5/G8 transgenic; SREBP, sterol regulatory element binding protein; HDL, high-density lipoprotein; SR-BI, scavenger receptor class B type I; NPC1L1, Niemann-Pick C1-Like 1; qPCR, quantitative real-time polymerase chain reaction; LXR, liver X receptor; and PCSK9, proprotein convertase subtilisin/kexin type 9a
Materials and Methods
A rabbit polyclonal antiserum against mouse ABCG5, a monoclonal antibody recognizing both mouse and human ABCG8, and a rabbit polyclonal antipeptide antibody (Q820-6) against the last 14 amino acids of murine scavenger receptor class B type I (SR-BI) were reported previously.16, 18, 25 Rabbit anti-mouse LDLR (3413) and anti-rat receptor-associated protein polyclonal antibodies (692)26 were kindly provided by Joachim Herz (University of Texas Southwestern Medical Center at Dallas).
Animals and Diets.
G5G8Tg mice (line 14-2) were generated as previously described.17 These animals have about 14 copies of a 90-kb human ABCG5/ABCG8 genomic DNA fragment integrated into their genome and consequently overexpress human ABCG5/ABCG8 in the liver and small intestine. The animals were housed in plastic cages in a temperature-controlled room (22°C) with a daylight cycle from 6 AM to 6 PM. The mice were fed ad libitum a cereal-based rodent chow diet (Diet 7001, Harlan Teklad, Madison, WI) containing 0.02% cholesterol and 4% fat. All animal procedures were performed with approval of the Institutional Animal Care and Research Advisory Committee at the University of Texas Southwestern Medical Center at Dallas.
Diets containing 0.2% (wt/wt) of lovastatin (Merck & Co.) were made by mixing powdered chow diet (Diet 7001, Harlan Teklad) with lovastatin, and stored at 4°C before use. Female mice were housed individually 1 week before initiation of the statin diet and then fed for 3 days with either the statin diet or the chow diet dispensed from a feeder jar.
Plasma, Bile, and Tissue Collection.
Mice were anesthetized and killed by drawing blood from vena cava after lovastatin treatment for 3 days. Plasma was separated from blood cells by centrifuging blood at 1,500 × g for 10 minutes in a bench-top microcentrifuge. Bile was collected from the gallbladders of anesthetized mice using a 30-gauge needle. Livers were excised and frozen in liquid nitrogen for the analysis of lipid, RNA, and protein as described below.
Sterol levels in bile, plasma, and liver were measured by gas chromatography (GC) as described.16 Briefly, bile lipids were extracted by methanol. Plasma and tissues were saponified in 3% potassium hydroxide/ethanol at 65°C for 3 hours after addition of 5α-cholestane as an internal standard, and the lipids were extracted by using petroleum ether. Extracted lipids in organic solvents were dried under nitrogen, and residual lipids were redissolved in trimethylsilyl (TMS)-reagent (pyridine:hexamethyldisilasane:chlorotrimethylsilane; 9:3:1 vol/vol/vol) (product no. 48999, Pierce Biotechnology Inc., Rockford, IL) for analysis by GC.
Quantitative Real-Time PCR (qPCR).
Total RNAs were extracted from livers using RNA Stat-60 kit (Tel-Test Inc., Friendswood, TX), and qPCR was performed as described.17 Cyclophilin was used as an internal control for these studies, and the cycle number of cyclophilin was subtracted from the cycle number of each gene in the same sample. The values of mRNA levels for each gene represent the amount relative to the amount in the chow-fed wild-type mice, which was arbitrarily standardized to 1. Primer sequences of genes used for quantification of mRNA levels by qPCR are listed in Supplementary Table 1 (available at the Hepatology website: http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html).
Preparation of Liver Membrane Proteins.
A total of 100-200 mg liver was homogenized by a polytron in 1.2 mL Buffer A (250 mmol/L sucrose, 2 mmol/L MgCl2, 20 mmol/L Tris-HCl, pH 7.5) containing protease inhibitors (Complete Protease Inhibitor cocktail, Roche Diagnostics). The crude preparation was centrifuged at 2000 × g for 10 minutes at 4°C. The supernatant was collected and recentrifuged at 120,000 × g for 45 minutes at 4°C. The membrane pellet was resuspended in a solution containing 80 mmol/L NaCl, 2 mmol/L CaCl2, 1% Triton X-100, 50 mmol/L Tris-HCl, pH 8, and protease inhibitors as above.
The protein concentrations of liver membranes were determined using the BCA Kit (Pierce Biotechnology). Membrane proteins (50 μg) were fractionated on 8% polyacrylamide gel in the presence of sodium dodecyl sulfate and transferred to Hybond-C Extra nitrocellulose filters (Amersham Biosciences, Piscataway, NJ). The filters were immunoblotted with antibodies in phosphate-buffered saline, pH 7.4 with Tween 20 (catalog# P-3563, Sigma-Aldrich, St. Louis, MO), 5% powdered milk, and 5% newborn calf serum, and then incubated with horseradish peroxidase-conjugated donkey anti-rabbit IgG (Jackson Immunoresearch Laboratories, West Grove, PA) or donkey anti-mouse IgG (Jackson Immunoresearch Laboratories). Immunodetection was performed using the SuperSignal Substrate System (Pierce Biotechnology). Filters were exposed to Kodak X-Omat XLS-1 films at room temperature.
All data are reported as the mean ± standard error of the mean (SEM). The difference between the mean values of control and treated groups were tested for statistical significance by the two-tailed Student t tests.
Lovastatin Reduces Plasma and Liver Cholesterol Concentration in G5G8Tg Mice.
The baseline cholesterol synthesis rate is higher in G5G8Tg mice than in wild-type mice due to increased biliary and fecal loss of cholesterol in G5G8Tg mice.17 To test the hypothesis that animals with higher endogenous cholesterol synthesis rate are more sensitive to statin, wild-type and G5G8Tg female mice were fed a chow diet or a chow diet containing 0.2% lovastatin for 3 days. Plasma cholesterol concentrations were determined. Consistent with a previous report,27 statin had no effect on plasma cholesterol levels in wild-type mice (Fig. 1A). Interestingly, lovastatin treatment significantly reduced plasma cholesterol concentrations by 41.6% in G5G8Tg mice (40.14 mg/dL on chow+lovastatin diet versus 68.74 mg/dL on chow diet) (Fig. 1A).
Liver is the major organ controlling whole-body cholesterol homeostasis. To probe alterations in liver cholesterol metabolism in G5G8Tg mice treated with statin, hepatic cholesterol content was measured. Lovastatin treatment did not affect hepatic total cholesterol content in wild-type mice but significantly decreased that in G5G8Tg mice (2.06 mg/g wet liver weight on chow+lovastatin diet versus 2.49 mg/g wet liver weight on chow diet) (Fig. 1B).
Feedback Upregulation of Cholesterol Synthetic Genes Can Not Compensate for the Decreased Plasma Cholesterol in G5G8Tg Mice on Statin.
It was reported decades ago that statins increase HMG-CoA reductase activity in rodents.27, 28 To determine the expression level of genes in the pathway of cholesterol biosynthesis in G5G8Tg mice treated with statin, liver total RNAs were extracted and qPCR was performed. As expected, the mRNA levels of genes in the pathway of cholesterol synthesis including HMG-CoA reductase, HMG-CoA synthesis, sterol regulatory element binding protein 2 (SREBP2), farnesyl diphosphate synthase, and squalene synthase were all upregulated by statin treatment in wild-type mice (Fig. 2). Intriguingly, this upregulation was much greater in lovastatin-treated G5G8Tg mice than that in all other groups. As qPCR controls, genes whose expression levels were not affected by lovastatin were also shown in Fig. 2. These included ABCA1, a gene that is important in high-density lipoprotein (HDL) lipidation; cholesterol 7α hydroxylase, a gene that governs bile acid synthesis; and apolipoprotein B, a gene that is essential for very low density lipoprotein secretion.
Lovastatin Treatment Increases Biliary Cholesterol Concentrations and Hepatic mRNA and Protein Levels of ABCG5/ABCG8 in Mice.
Biliary cholesterol secretion is one of the major pathways by which the body eliminates excess cholesterol. To examine the effect of statin on this important pathway of cholesterol trafficking in mice, the gallbladder cholesterol concentration was determined. Lovastatin treatment significantly increased the gallbladder cholesterol concentration in wild-type mice (7.53 ± 1.6 μmol/mL on chow+lovastatin diet versus 3.85 ± 0.45 μmol/mL on chow diet) (Fig. 3A). The gallbladder cholesterol concentration was much higher in G5G8Tg mice than in wild-type mice on chow diet, consistent with our previous report.17 Lovastatin treatment failed to further increase gallbladder cholesterol concentration in G5G8Tg mice (Fig. 3A). ABCG5/ABCG8 plays a crucial role in mediating biliary cholesterol secretion.16, 17 To examine whether increased biliary cholesterol concentration was coupled to increased expression of ABCG5/ABCG8, the mRNA and protein levels of ABCG5/ABCG8 were determined by qPCR using primers specific to mouse ABCG5/ABCG8, and by Western blotting using an antibody specific to mouse ABCG516 and an antibody to both mouse and human ABCG8.18 Both mRNA and protein levels of ABCG5/ABCG8 were upregulated by lovastatin treatment in wild-type mice (Fig. 3B-C). In the chow-fed G5G8Tg mice, the mRNA levels of mouse endogenous ABCG5/ABCG8 was higher than that in the chow-fed wild-type mice and statin failed to further increase mouse endogenous ABCG5/ABCG8 mRNAs in G5G8Tg mice (Fig. 3B). However, there was an obvious increase in mouse ABCG5 proteins (Fig. 3C). The anti-ABCG8 antibody did not distinguish mouse ABCG8 from human ABCG8 (Fig. 3C) and the size of mouse and human ABCG8 proteins are also indistinguishable, thus the protein level of mouse endogenous ABCG8 remained unknown in the statin-treated G5G8Tg mice. Phosphatidylcholine transporter ABCB4 and bile acid transporter ABCB11, which are two other canalicular ABC transporters affecting biliary lipid secretion,29, 30 were found to be similar to the mRNA levels in mice treated with lovastatin (Fig. 3B).
In the statin-fed mice, increased biliary cholesterol concentration seems uncoupled to the increased amount of ABCG5/ACG8 proteins (Fig. 3), suggesting mechanisms other than ABCG5/ABCG8 may also be involved in modulating biliary cholesterol secretion. Niemann-Pick C1-Like 1 (NPC1L1), a gene that is required for intestinal cholesterol absorption,31 is expressed not only in the intestine but also abundantly in human liver.31–33 Mouse NPC1L1 can be detected by reverse transcriptase coupled PCR (Fig. 4A), and these PCR products were verified to be NPC1L1 by sequencing (data not shown). Our recent study demonstrates that NPC1L1 protein localizes to the canalicular membrane of hepatocytes in the monkey liver.34 We speculate that NPC1L1, as a cholesterol uptake transporter,34 may counterbalance biliary cholesterol secretion mediated by ABCG5/ABCG8. Therefore, hepatic NPC1L1 mRNA level was determined by qPCR. Intriguingly, after lovastatin treatment, the mRNA level of NPC1L1 was increased moderately in the wild-type mouse liver and robustly in the G5G8Tg mouse liver (Fig. 4B).
Lovastatin Reduces Plasma Cholesterol Without Significantly Affecting the Protein Level of LDLR and SR-BI.
Both LDLR and the HDL receptor SR-BI play critical roles in the uptake of plasma cholesterol by the liver. To examine whether the reduced plasma cholesterol was a result of alterations in the expression of these receptors, the hepatic mRNA and protein levels of LDLR and SR-BI were determined. Lovastatin treatment did not change SR-BI expression at both mRNA and protein levels in both wild-type and G5G8Tg mice (Fig. 5A-B). It also had little effect on LDLR mRNA and protein in wild-type mice. There were two bands in the immunoblots of LDLR. The upper band represents the mature form of the receptor that has undergone O-linked glycosylation in the Golgi apparatus, and the lower band represents the LDLR precursor in the endoplasmic reticulum.35, 36 In G5G8Tg mice, lovastatin feeding upregulated the hepatic mRNA level of LDLR, but this increase was not associated with an increase in the amount of mature LDLR protein instead of an increase in the amount of LDLR precursor (Fig. 5A-B).
The finding from this study is consistent with a notion that animals with higher baseline cholesterol synthesis rates are more sensitive to statins. The feedback upregulation of cholesterol synthetic genes after statin treatment is insufficient to maintain a normal level of plasma cholesterol when ABCG5/ABCG8 function is stimulated. The cholesterol lowering effect of statin in G5G8Tg mice is not coupled to the protein levels of LDLR and SR-BI, the two major cholesterol uptake receptors. Another interesting finding is that lovastatin treatment increases the hepatic expression of ABCG5/ABCG8 and NPC1L1 in mice, which is associated with an increase in biliary cholesterol concentration. These findings are clinically important because they suggest that manipulating ABCG5/ACBG8 function may have potential in altering efficacy of statins.
Wild-type rodents are insensitive to statins, believed to be partly due to the feedback regulation of HMG-CoA reductase activity in the liver.27, 28 We found in this study that lovastatin decreased plasma cholesterol in the G5G8Tg mice in spite of a much greater feedback upregulation of genes in the pathway of cholesterol synthesis, suggesting that the functional level of ABCG5/ABCG8 may also contribute to statin responsiveness, which is consistent with a clinical observation that ABCG8 D19H subjects are more responsive to atorvastatin therapy.15 Recent studies demonstrate that a SREBP-regulated gene, the proprotein convertase subtilisin/kexin type 9a (PCSK9), plays a critical role in modulating statin response in mice.37 Lovastatin administration to wild-type mice increases PCSK9 proteins in the liver through SREBP pathway, which in turn facilitates LDLR degradation.37 Disruption of PCSK9 in mice decreases plasma cholesterol and confers mice with statin hypersensitivity by raising LDLR levels.37 The molecular mechanism underlying statin hypersensitivity in G5G8Tg mice has yet to be elucidated. It is well known that plasma cholesterol is predominantly associated with HDL in chow-fed wild-type mice. Plasma cholesterol is also mainly associated with HDL in G5G8Tg mice.17 Observed reduction in total plasma cholesterol in G5G8Tg mice must be proportional to the decrease in HDL-cholesterol levels. Given that the protein level of HDL receptor SR-BI did not differ between treated and untreated animals (Fig. 5B), it is unlikely that SR-BI is responsible for the lower plasma cholesterol level observed in the lovastatin-treated G5G8Tg mice. Mouse HDL particles contain apoE,37 a ligand for LDLR. It remains possible that the clearance of HDL cholesterol through LDLR pathway is promoted by statin in G5G8Tg mice. However, the protein level of LDLR was unaffected by lovastatin in these animals, suggesting that an LDLR-independent mechanism may have been involved.
The liver of lovastatin-treated G5G8Tg mice is deprived of cholesterol (Fig. 1B). It is tempting to speculate that the availability of cholesterol for hepatic secretion into the circulation is limited, thereby resulting in lower plasma cholesterol level. Reduced production rate of apoB-containing lipoproteins following statin treatments has been reported in many independent studies.38–42 Mice are HDL animals. Inhibition of apoB-containing lipoprotein production may not cause a significant reduction in plasma HDL cholesterol in mice. Hepatic ABCA1 plays an important role in regulating plasma HDL levels in mice.43 It is possible that the deprivation of hepatic cholesterol in the statin-treated G5G8Tg mice cause a reduction in ABCA1-dependent lipidation of HDL, resulting in lower plasma cholesterol in these animals.
An interesting finding in this study is that the mRNA and protein levels of ABCG5/ABCG8 were upregulated by statin treatment, which was associated with an increase in biliary cholesterol concentrations (Fig. 3). Expression of ABCG5/ABCG8 is coordinately upregulated by cholesterol feeding through activation of liver X receptor (LXR) although the binding sites of LXRs to these genes have not been mapped.19, 44 In this study, the hepatic cholesterol content was lower in lovastatin-treated G5G8Tg mice and paradoxically the expression level of ABCG5/ABCG8 was higher, suggesting that the increased expression of ABCG5/ABCG8 may not be a result of LXR activation. It was reported that the mRNA level of ABCG5/ABCG8 was higher in the liver and small intestine of mice lacking LXRα/LXRβ,44 indicating that the baseline expression of ABCG5/ABCG8 is also controlled by mechanisms other than LXRα/LXRβ. What are these mechanisms? In the statin-treated G5G8Tg mice, SREBP-2 pathway has been activated as evidenced by the dramatic upregulation of SREBP-2 target genes (Fig. 2). ABCG5 and ABCG8 are oppositely oriented in the genome and are separated by a short bidirectional promoter.19 Two putative SREBP sites were found by MatInspector (Genomatix Software GmbH, Munich, Germany) within the 2510-base pair sequence upstream of the mouse ABCG8 start codon. We speculate that mouse ABCG5/ABCG8 may also be a target gene of SREBP-2, and the cross-talk between LXR and SREBP-2 may determine the basal expression level of ABCG5/ABCG8. Consistent with this hypothesis, high cholesterol feeding reduced the hepatic mRNA level of ABCG5/ABCG8 in the absence of LXRs.44 When liver is depleted of cholesterol, the SREBP pathway is activated to promote de novo cholesterol biosynthesis. Why is ABCG5/ACBG8, which eliminate cellular cholesterol, upregulated at the same time? One explanation could be that as an efflux transporter of cholesterol, ABCG5/ACBG8 is simultaneously increased to protect cells from overaccumulation of newly synthesized free cholesterol. Another example for this paradoxical regulation is that PSCK9 and LDLR are simultaneously upregulated through SREBP pathway when liver is depleted of cholesterol; PCSK9 upregulation causes LDLR degradation to block more cholesterol entering the liver.36, 37 An alternative explanation for this paradoxical upregulation might be that newly synthesized free cholesterol, or some of its intermediates, activates LXR pathway thereby leading to upregulation of ABCG5/ABCG8, and possibly PCSK9. This speculation can be tested in mice lacking LXRα/LXRβ.
Increased gallbladder cholesterol concentration could be a result of increased expression of ABCG5/ACBG8. However, a 2-fold increase in biliary cholesterol concentration is not proportional to a >3-fold increase in the amount of ABCG5/ABCG8 proteins in the lovastatin-treated versus untreated wild-type mice (Fig. 3). In addition, an obvious increase in ABCG5/ACBG8 proteins is not coupled to a further increase in biliary cholesterol concentration in the lovastatin-treated G5G8Tg mice. Published evidence in studies of human patients and rodents supports a notion that mechanisms other than ABCG5/ABCG8 may modulate biliary cholesterol secretion under some conditions.45, 46 Intriguingly, in the lovastatin-treated mice, especially in G5G8Tg mice, the hepatic mRNA level of NPC1L1 was increased dramatically (Fig. 4). The function of NPC1L1 in the liver remains unknown but we found recently that the NPC1L1 protein localizes to the canalicular membrane of hepatocytes in the monkey liver. Because NPC1L1 is in the cholesterol absorption pathway, we speculate that hepatic NPC1L1 may modulate biliary cholesterol excretion by counterbalancing biliary cholesterol secretion mediated by ABCG5/ABCG8. Increased NPC1L1 may reclaim a portion of biliary cholesterol back into hepatocytes in the lovastatin-treated mice, thereby explaining the uncoupling of biliary cholesterol concentration and the amount of ABCG5/ABCG8 proteins in these animals.
In summary, statin treatment significantly reduced plasma cholesterol level in G5G8Tg mice, indicating that statins may be more effective in individuals with higher ABCG5/ABCG8 activity and/or higher baseline endogenous cholesterol synthesis rate.
We thank Helen H. Hobbs and Jonathan C. Cohen at University of Texas Southwestern Medical Center at Dallas for allowing us to perform animal studies and GC analysis in their facility.