Fenofibrate Induces a Novel Degradation Pathway for Scavenger Receptor B-I Independent of PDZK1*

Fibrate drugs improve cardiovascular health by lowering plasma triglycerides, normalize low density lipoprotein levels, and raise high density lipoprotein (HDL) levels in patients with dyslipidemias. The HDL-raising effect of fibrates has been shown to be due in part to an increase in human apolipoprotein AI gene expression. However, it has recently been shown that fibrates can affect HDL metabolism in mouse by significantly decreasing hepatic levels of the HDL receptor scavenger receptor B-I (SR-BI) and the PDZ domain containing protein PDZK1. PDZK1 is essential for maintaining hepatic SR-BI levels. Therefore, decreased SR-BI might be secondary to decreased PDZK1, but the mechanism by which fibrates lower SR-BI has not been elucidated. Here we show that feeding PDZK1-deficient mice fenofibrate resulted in the near absence of SR-BI in liver, definitively demonstrating that the effect of fenofibrate on SR-BI is PDZK1-independent. Metabolic labeling experiments in primary hepatocytes from fenofibrate-fed mice demonstrated that fenofibrate enhanced the degradation of SR-BI in a post-endoplasmic reticulum compartment. Moreover, fenofibrate-induced degradation of SR-BI was independent of the proteasome, calpain protease, or the lysosome, and antioxidants did not inhibit fenofibrate-induced degradation of SR-BI. Using metabolic labeling coupled with cell surface biotinylation assays, fenofibrate did not inhibit SR-BI trafficking to the plasma membrane. Together, the data support a model in which fenofibrate enhances the degradation of SR-BI in a post-ER, post-plasma membrane compartment. The further elucidation of this novel degradation pathway may provide new insights into the physiological and pathophysiological regulation of hepatic SR-BI.

Fibrate drugs improve cardiovascular health by lowering plasma triglycerides, normalize low density lipoprotein levels, and raise high density lipoprotein (HDL) levels in patients with dyslipidemias. The HDLraising effect of fibrates has been shown to be due in part to an increase in human apolipoprotein AI gene expression. However, it has recently been shown that fibrates can affect HDL metabolism in mouse by significantly decreasing hepatic levels of the HDL receptor scavenger receptor B-I (SR-BI) and the PDZ domain containing protein PDZK1. PDZK1 is essential for maintaining hepatic SR-BI levels. Therefore, decreased SR-BI might be secondary to decreased PDZK1, but the mechanism by which fibrates lower SR-BI has not been elucidated. Here we show that feeding PDZK1-deficient mice fenofibrate resulted in the near absence of SR-BI in liver, definitively demonstrating that the effect of fenofibrate on SR-BI is PDZK1-independent. Metabolic labeling experiments in primary hepatocytes from fenofibrate-fed mice demonstrated that fenofibrate enhanced the degradation of SR-BI in a post-endoplasmic reticulum compartment. Moreover, fenofibrate-induced degradation of SR-BI was independent of the proteasome, calpain protease, or the lysosome, and antioxidants did not inhibit fenofibrate-induced degradation of SR-BI. Using metabolic labeling coupled with cell surface biotinylation assays, fenofibrate did not inhibit SR-BI trafficking to the plasma membrane. Together, the data support a model in which fenofibrate enhances the degradation of SR-BI in a post-ER, post-plasma membrane compartment. The further elucidation of this novel degradation pathway may provide new insights into the physiological and pathophysiological regulation of hepatic SR-BI.
Multiple clinical trials have shown that activation of the nuclear hormone receptor peroxisome proliferator-activated receptor (PPAR) 1 ␣ by fibrate drugs results in the lowering of plasma triglycerides, normalizing of low density lipoprotein levels, and increased HDL in patients with atherogenic dyslipidemia (1)(2)(3)(4)(5)(6). One mechanism by which fenofibrate increases plasma HDL is by directly inducing human apoAI gene expres-sion through activation and binding of PPAR␣ to its promoter (7,8). Gemfibrozil, another fibrate in use in humans, has been shown in clinical trials to increase plasma HDL to a similar extent as fenofibrate (3,4). However, gemfibrozil, compared with fenofibrate, only weakly increased the expression of the human apoAI gene in a transgenic mouse model (8). Therefore, these findings indicate that different fibrates act by different mechanisms to raise HDL, such as through the recruitment or availability of specific co-activators to activate gene expression (8). These studies suggest that other potential mechanisms independent of enhanced apoAI gene expression by fibrates can raise HDL.
Scavenger receptor B-I is the primary HDL receptor in mouse that plays an essential role in the hepatic uptake of plasma HDL-derived cholesterol and cholesteryl ester into liver for excretion into bile (9 -12). Targeted disruption of the SR-BI gene resulted in a significant increase in plasma HDL and increased atherosclerosis in multiple mouse models of atherosclerosis (12)(13)(14). The multi-PDZ domain-containing protein PDZK1 was shown to interact with SR-BI from rat liver (15). This interaction requires the ultimate four amino acids in the C terminus of SR-BI that constitute a PDZ-interacting domain (16). Deletion of the PDZ-interacting domain in SR-BI resulted in a failure of SR-BI to accumulate at the plasma membrane in mouse hepatocytes (16). Recently, gene-targeted disruption of PDZK1 resulted in the near absence of SR-BI in mouse liver, proving that PDZK1 is essential for SR-BI expression in liver (17).
In mouse lesional macrophage and human macrophage, fenofibrate and other PPAR␣ activators enhance SR-BI expression (18). Enhancement of SR-BI in lesional macrophage would be considered beneficial against atherosclerosis because the absence of SR-BI expression in macrophage appears to increase atherosclerosis in mouse (19,20). Surprisingly, in mouse liver, fibrates significantly decrease the hepatic levels of SR-BI and PDZK1 in a PPAR␣-dependent fashion without a decrease in their respective mRNAs, indicating that PPAR␣ down-regulates SR-BI through a posttranscriptional mechanism (21,22).
Mardones et al. (21) proposed a number of mechanisms by which fibrates may down-regulate both SR-BI and PDZK1, such as effects on SR-BI and PDZK1 protein synthesis, trafficking, SR-BI-PDZK1 interactions, degradation, or mechanisms acting independently on SR-BI and PDZK1 (21). Because PDZK1 was reduced by fibrates, and it has recently been shown that PDZK1 is essential for SR-BI expression in liver, SR-BI down-regulation may be secondary to a reduction in PDZK1. Here we explore further the mechanism of fibrate action on SR-BI, test whether PDZK1 is essential for mediating the effect of fibrates on down-regulating SR-BI, and determine at which post-transcriptional level fibrates down-regulate SR-BI in mouse liver.

EXPERIMENTAL PROCEDURES
Reagents-Mice were fed a chow diet supplemented with 0.2% fenofibrate (Sigma-Aldrich). Chow diet containing fenofibrate was produced by Bio-Serv (Frenchtown, NJ). For metabolic labeling studies using inhibitors, cells were chased in the presence of the following inhibitors: 100 M ALLN, 50 M chloroquine, 10 M lactacystin, 5 g/ml brefeldin A, and 20 g/ml nocodazole (purchased from Sigma-Aldrich). These concentrations were previously shown to be inhibitory (23). Vitamin E succinate (150 M) and desferrioxamine (DFX; 150 M) were purchased from Sigma-Aldrich. Anti-PDZK1 antibody was produced in rabbit using fulllength recombinant murine PDZK1 as antigen. Neutralizing anti-SR-BI antibody used for all immunoprecipitations was previously described (24).
Mice-All mice were between 8 and 12 weeks of age. Both male and female mice were utilized for all experiments, and no sex-specific effects on SR-BI levels or turnover were observed with fenofibrate. PDZK1Ϫ/Ϫ and littermates mice were of a mixed genetic background (C57BL/6J and 129Sv/Ev) produced by intercrossing PDZK1ϩ/Ϫ mice. PPAR␣deficient mice and control wild-type mice (129SvEJ background) and C57BL/6J mice were obtained from Jackson Laboratory.
Generation of PDZK1 Gene-targeted Mice-A 16-kb mouse genomic DNA fragment was cloned from a mouse 129Sv/Ev genomic library. This genomic fragment contains exon 1 with the ATG codon for the start of translation. A targeting vector was constructed by using a 930-bp PCR-generated DNA fragment. This PCR fragment was generated using the primer pair sequences 5Ј-TCCTTAACACTTAGAACCT-GAGAG-3Ј located 930 bp upstream from the ATG start codon and 5Ј-TTTCTGTGATTAAAAACAAATGGATAAGTC-3Ј located 1 bp upstream of the ATG start codon. This 930-bp fragment was designated the "short arm." This short arm was inserted 5Ј of the neomycin resistance cassette (Neo) using a MluI site. The "long arm" was a 10-kb KpnI genomic fragment obtained from the 16-kb clone. The 10-kb long arm begins within intron 1. The resulting targeted PDZK1 allele will have exon 1 and part of intron 1 replaced by the Neo cassette. 10 g of targeting vector was linearized by a unique NotI site and then electroporated into iTL1 (129Sv/Ev) embryonic stem cells (Columbia University Transgenic Facility). After selection in G418, surviving colonies were expanded, and PCR analysis was performed to identify clones that had undergone homologous recombination. PCR was performed using primer pair 1 and 3. Primer 1 is located outside the short arm, 28 bp upstream of the end of the short arm, with a sequence of 5Ј-CATTC-TACCAAGTTTGAGAGTCAG-3Ј. Primer 3 is located in the 5Ј promoter region of the Neo cassette with the sequence 5Ј-TGCGAGGCCAGAG-GCCACTTGTGTAGC-3Ј. Two positive clones gave rise to a 1.1-kb PCR fragment. The correctly targeted embryonic stem cell lines were microinjected into C57BL/6J blastocysts. Chimeric mice were generated and gave germ-line transmission of the disrupted PDZK1 gene after backcrossing of chimeric mice to C57BL/6J mice. To identify the wild-type allele, primer pair 1 and 2 was used in PCR reactions from genomic DNA isolated from tails. Primer 2 is located in exon 1, which was replaced by the Neo cassette. Primer 2 begins 16 bp downstream of the ATG start codon, with the sequence 5Ј-CAGGTGACCATCAGTGTCCT-TCTC-3Ј. The PCR fragment generated from this reaction will be 1050 bp.
Primary Hepatocyte Culture and Pulse-chase Assays-Primary hepatocytes were isolated according to our previously reported method (25). Freshly isolated primary hepatocytes were allowed to attach to collagen-coated, 6-well plates for 1.5 h before the start of experiments. For pulse or pulse-chase experiments, hepatocytes were incubated with Dulbecco's modified Eagle's medium without cysteine or methionine (Invitrogen) for 60 min to deplete cells of methionine and cysteine. 200 Ci of 35 S-labeled methionine/cysteine mix (Expres 35 S 35 S-protein labeling mix; NEG-072 from PerkinElmer Life Sciences) was added per milliliter of cell culture media (Dulbecco's modified Eagle's medium without cysteine or methionine) for the times indicated. For pulse-chase assays, cells were incubated with labeling mix for 30 min, followed by a chase period for the times indicated in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, supplemented with 1.5 mg/ml L-methionine and 0.5 mg/ml L-cysteine. Hepatocytes were washed and lysed with radioimmune precipitation assay buffer (10 mM Tris, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA Complete EDTA-Free Protease Inhibitor Mixture (Roche)). SR-BI was immunoprecipitated using our neutralizing polyclonal antibody described under "Reagents." Immunoprecipitates were separated on 4 -15% Tris-glycine gels and transferred to nitrocellulose membranes. Membranes were initially analyzed by PhosphorImager analysis followed by Western blot analysis to determine total SR-BI levels in immunoprecipitates.
Cell Surface Biotinylation-Primary hepatocytes were metabolically labeled for the times indicated in Fig. 7. At the end of each time period, cells were washed and incubated with 10 mg/ml EZ-Link Sulfo-NHS-SS-Biotin (Pierce) for 30 min on ice. Following quenching (Biotinylation Quenching Buffer; Pierce) and washing, cells were lysed in radioimmune precipitation assay buffer. SR-BI was first immunoprecipitated as described above, except that an antibody we generated to the last C-terminal 15 amino acids was used for immunoprecipitation because biotinylation of the extracellular loop of SR-BI would likely inhibit immunoprecipitation using our SR-BI neutralizing antibody. Immunoprecipitates were boiled in 4% SDS, and an aliquot was removed that represented total cellular SR-BI that was newly synthesized. The remaining boiled immunoprecipitate was incubated with streptavidinagarose beads (Pierce) to pull down biotinylated SR-BI (i.e. newly synthesized SR-BI that is at the plasma membrane). Both fractions, total and cell surface SR-BI, were separated on SDS-PAGE gels and analyzed by PhosphorImager analysis, followed by Western blot analysis using either streptavidin-horseradish peroxidase (Santa Cruz Biotechnology) or anti-SR-BI antibody.
Statistical Analysis-All experiments were repeated between two and four times. All data were analyzed by Student's t test for statistical significance and are represented as the mean Ϯ S.D.

RESULTS
Fenofibrates Decrease SR-BI before PDZK1-To determine whether we could reproduce the effect of fenofibrate on lowering SR-BI and PDZK1 in liver and whether PPAR␣ was required, wild-type mice (C57BL6/J mice) were fed a diet of 0.2% fenofibrate for 7 days, based on the study by Mardones et al. (21). To our surprise, fenofibrate reduced the expression of SR-BI at day 2 without a decrease in PDZK1 (Fig. 1). PDZK1 was found to be reduced at day 7. These data indicate that PDZK1 may not be essential for mediating the fenofibrate effect on SR-BI. In further support of this hypothesis, wild-type mice of a different genetic background (129SvEJ), but not PPAR␣-deficient mice of the same genetic background, fed fenofibrate for 7 days showed the expected decrease in SR-BI, but without a decrease in PDZK1 (Fig. 2, A and B). PPAR␣ knock-out mice had a small but significant increase in SR-BI protein levels compared with wild-type control mice (Fig. 2, A  and B). No decrease in SR-BI mRNA was observed, indicating a post-transcriptional process (Fig. 2C). SR-BI is highly expressed in adrenal gland (9) but was not decreased by fenofibrate (Fig. 2D). Taken together, the data indicate that PDZK1 down-regulation by fenofibrate is dependent on the genetic background of the mouse, whereas down-regulation of SR-BI is not, pointing strongly to the possibility that PDZK1 is not essential for the effect of fenofibrate on SR-BI levels.
FIG. 1. SR-BI decreased in response to fenofibrate feeding before PDZK1 was decreased. C57BL/6J mice were fed 0.2% fenofibrate (FF) for the times indicated. Decreased levels of SR-BI protein but not PDZK1 were detected at 2 days of fenofibrate feeding, as indicated by arrows. The graph below the Western blot shows the changes in both SR-BI and PDZK1 during the time course. PDZK1 levels are decreased by 50% after 7 days on diet. Each time point is the mean normalized to ␤-actin levels from two mice.

PDZK1 Is Not Required for Down-regulation of SR-BI by
Fenofibrate-To definitively prove that PDZK1 is not essential for the effect of fenofibrate on SR-BI levels, we tested the effect of fenofibrate on SR-BI levels in PDZK1 gene-targeted mice. Fig. 3, A and B, describes our gene targeting strategy to generate PDZK1-deficient mice using standard homologous recombination techniques. The targeting vector used to mutate the PDZK1 gene contained a 930-bp fragment located 1 bp upstream from the ATG start codon and a 10-kb KpnI genomic fragment starting at the beginning of intron 1 flanking the neomycin resistance gene (Neo). As a result of homologous recombination, exon 1 and part of intron 1 were replaced by the Neo gene, trapping gene transcription inside the Neo gene. In addition, the absence of an ATG would abolish translation of the mutated PDZK1 mRNA if it were to be transcribed. Two independently derived embryonic stem cell clones were identified as having a single homologous recombination event of the targeting vector with the wild-type gene. Both embryonic stem cell clones were injected into C57BL/6J blastocysts, and chimeric offspring were backcrossed to C57BL/6J mice, and germline transmission was confirmed for one of the embryonic stem cell clones. Mice heterozygous for the PDZK1 gene-targeted locus (PDZK1ϩ/Ϫ) were intercrossed to produce mice with a homozygous disruption of the PDZK1 locus (PDZK1Ϫ/Ϫ). The frequency of wild-type (PDZK1ϩ/ϩ), heterozygous (PDZK1ϩ/ Ϫ), and homozygous (PDZK1Ϫ/Ϫ) mice in intercrosses of heterozygotes adhered to a Mendelian ratio, and male and female PDZK1Ϫ/Ϫ mice had normal fertility (data not shown), in agreement with an earlier report describing PDZK1-deficient mice (26).
PDZK1-deficient mice showed a significant decrease (ϳ85-95%) in hepatic SR-BI levels (Fig. 3, C and D), similar to the PDZK1-deficient mouse model previously generated and described by Kocher et al. (17). The levels of SR-BI in PDZK1 heterozygous mice were not previously described (17). In this regard, it is important to note that PDZK1 heterozygous mice, while having 50% less PDZK1, do not have decreased SR-BI (Fig. 3, C and D), indicating that PDZK1 is not rate-limiting in the liver for maintaining SR-BI levels. Mice heterozygous for PDZK1 have ϳ50% less PDZK1 compared with wild-type mice, similar to the levels obtained with fenofibrate treatment of wild-type C57BL/6J mice (Fig. 1A), but without a decrease in SR-BI, which indicates that PDZK1 is likely not required to mediate the effects of fenofibrate on SR-BI. To directly prove this hypothesis, wild-type (mixed genetic background of 129Sv/Ev and C57BL/6J) and PDZK1-deficient littermates were fed fenofibrate for 7 days, and SR-BI and PDZK1 levels were examined. Fig. 4, A and B, shows that in the absence of PDZK1, fenofibrate significantly reduced SR-BI to nearly undetectable levels. Taken together, the data indicate that fenofibrate acts through a mechanism that directly effects SR-BI levels and does not require PDZK1.
Fenofibrate Enhances SR-BI Degradation by a Post-ER, Postplasma Membrane Mechanism-Because our data indicate that fenofibrate decreases SR-BI levels post-transcriptionally and independently of PDZK1, we determined whether fenofibrate acted to decrease SR-BI synthesis or increase SR-BI degradation. To first determine whether SR-BI synthesis is decreased in liver of fenofibrate fed mice, we performed [ 35 S]methionine/cysteine metabolic labeling assays using freshly isolated primary hepatocytes from fenofibrate-fed mice. Fig. 5A shows that during a time course of up to 2 h there was no significant difference in the synthetic rate of SR-BI in fenofibrate-fed mice compared with controls. The concentration of antibody used for immunoprecipitation of SR-BI was limiting, allowing equal amounts of SR-BI to be immunoprecipitated in order to more accurately quantify levels of SR-BI during its synthesis (Fig. 5A). Fig. 5B shows that steady-state levels of SR-BI in hepatocytes from fenofibrate-fed mice remained significantly lower than those in controls during the course of the experiment. PDZK1 levels did not decrease in hepatocytes from mice fed fenofibrate. Because fenofibrate does not affect SR-BI synthesis, the alternative mechanism is that fenofibrate enhances SR-BI degradation. To test this idea, we performed [ 35 S]methionine/cysteine pulse-chase assays with hepatocytes isolated from fenofibrate-fed mice and controls. In addition, we sought to determine whether degradation of SR-BI occurred in the ER or is mediated by a post-ER process, and whether the FIG. 2. PDZK1 is not down-regulated in 129SvEJ mice. 129SvEJ wild-type and PPAR␣Ϫ/Ϫ mice were fed a diet containing 0.2% fenofibrate (ϩFF) or standard chow diet (ϪFF) for 7 days. A, fenofibrate reduces SR-BI levels in livers from wild-type mice but not PPAR␣Ϫ/Ϫ mice. PDZK1 levels are unchanged by fenofibrate diet in the 129SvEJ strain. B, quantification of SR-BI and PDZK1 levels from the Western blot in A; *, p Ͻ 0.001, wild-type ϪFF versus wild-type ϩFF or PPAR␣Ϫ/Ϫ. Data represent the mean Ϯ S.D. (n ϭ 5) for each genotype and treatment. C, SR-BI down-regulation by fenofibrate is post-transcriptional. Northern blot analysis from liver of wild-type ϩFF and PPAR␣Ϫ/Ϫ ϩFF used in A showed no decrease in Sr-bi mRNA levels by fenofibrate feeding. D, SR-BI levels in adrenal gland from wild-type mice fed fenofibrate (ϩFF) were unchanged compared with wild-type mice not fed the fenofibrate diet (ϪFF); n ϭ 2 mice for each group. proteasome, calpain protease, or the lysosome is required. Fig.  6A shows that fenofibrate significantly enhanced the degradation of SR-BI compared with controls. Of note is the observation that from 40% to 50% of newly synthesized SR-BI (Fig. 6, A and  B) is degraded within 4 h, whereas the remainder appears stable up to 10 h in hepatocytes from fenofibrate-fed mice.
To test whether SR-BI degradation enhanced by fenofibrate occurred through an ER-mediated pathway or is a post-ER event, [ 35 S]methionine/cysteine-labeled hepatocytes were treated with brefeldin A (BFA) to disrupt the Golgi apparatus or with brefeldin A plus nocodazole to disrupt the Golgi appa-ratus and prevent retrograde transport of Golgi-derived vesicles back to the ER. Fig. 6B shows that BFA and BFA plus nocodazole both inhibited SR-BI degradation in hepatocytes from fenofibrate-fed mice, indicating that degradation occurred in a post-ER compartment. Moreover, neither of the inhibitors for the proteasome (lactacystin and ALLN), calpain protease (ALLN), nor lysosomal function (chloroquine) blocked the enhanced degradation of SR-BI in hepatocytes from fenofibratefed mice. No significant effects of these inhibitors were seen on SR-BI turnover in control hepatocytes. Fig. 6C shows that steady-state levels of SR-BI during the course of the experi-

FIG. 4. The reduction of SR-BI in fenofibrate-fed mice is independent of PDZK1.
A, mice fed 0.2% fenofibrate (ϩFF) or standard chow diet (ϪFF) for 7 days showed decreased SR-BI regardless of the presence or absence of PDZK1. Again, PDZK1 levels were not reduced in wild-type mice of mixed genetic background (129SvEv/C57BL/6J). B, quantification of SR-BI and PDZK1 from the Western blot shown in A. SR-BI levels were significantly reduced in both PDZK1ϩ/ϩ and PDZK1Ϫ/Ϫ mice fed fenofibrate; *, p Ͻ 0.001, PDZK1ϩ/ϩ, ϪFF versus ϩFF and PDZK1Ϫ/Ϫ, ϪFF versus ϩFF. No significant change in PDZK1 was measured in PDZK1ϩ/ϩ mice, ϪFF versus ϩFF. ment in hepatocytes from fenofibrate-fed mice remained significantly lower than those in controls, and once again, PDZK1 levels did not decrease in hepatocytes from mice fed fenofibrate.
Because enhanced SR-BI degradation did not occur in the ER, we next determined whether fenofibrate enhanced the degrada-tion of SR-BI before or after it reached the plasma membrane. To test this idea, primary hepatocytes isolated from control mice and mice fed fenofibrate were metabolically labeled with [ 35 S]methionine/cysteine for the times indicated in Fig. 7A, followed by cell surface biotinylation to label cell surface SR-BI. If the ratio of newly synthesized SR-BI at the cell surface to total cellular newly synthesized SR-BI in hepatocytes from fenofibrate-fed mice is similar to the same ratio from control hepatocytes, then SR-BI degradation occurs in a post-plasma membrane compartment. However, if the ratio of newly synthesized SR-BI at the cell surface to total cellular newly synthesized SR-BI in hepatocytes from fenofibrate-fed mice is lower than the same ratio from control hepatocytes, then SR-BI degradation occurs before it reaches the plasma membrane. Fig. 7A shows that the rate of SR-BI synthesis was similar in hepatocytes from fenofibratetreated mice compared with controls, as was previously demonstrated in Fig. 5A. In addition, the rate of newly synthesized SR-BI arriving at the plasma membrane (i.e. radiolabeled and biotinylated SR-BI) was similar in both groups. Fig. 7B confirms that steady-state levels of SR-BI in total cell lysates remained decreased in hepatocytes from fenofibrate-fed mice relative to controls during the course of the experiment. Taken together, the data indicate that the amount of SR-BI reaching the plasma membrane is not decreased as a result of fenofibrate and that enhanced degradation of SR-BI likely occurs in a post-ER, postplasma membrane compartment.
Based on similarities in the pathway for degradation of SR-BI induced by fibrates with two recent reports of a novel degradation pathway for apoB that is induced by oxidant stress in a post-ER, pre-secretory proteolysis pathway (27,28), we determined whether antioxidants could reverse fenofibrateinduced degradation of SR-BI. Fig. 8 shows that the antioxidant vitamin E succinate given to hepatocytes at a concentration shown to effectively normalize cellular lipid peroxide FIG. 5. Fenofibrate does not decrease SR-BI translation. A, hepatocytes from mice fed 0.2% fenofibrate (ϩFF) or standard chow diet (ϪFF) for 7 days were metabolically labeled with [ 35 S]methionine/cysteine for the times indicated. SR-BI was immunoprecipitated, and newly synthesized SR-BI was quantified. The fold increase in SR-BI levels is a measurement of the SR-BI signal relative to the 10 min time point. Total SR-BI levels in the immunoprecipitates (IP) were measured by probing the above blot with anti-SR-BI antibody. Similar levels of SR-BI were immunoprecipitated in all samples due to the use of limiting amounts of SR-BI antibody. B, total cell lysates taken from the experiment in A before immunoprecipitation were used to measure steady-state levels of SR-BI, PDZK1, and ␤-actin as a loading control. SR-BI, but not PDZK1, in hepatocytes from fenofibrate-fed mice was significantly reduced compared with controls during the course of the experiment.

FIG. 6. Fenofibrate increases SR-BI degradation in a post-ER compartment.
A, hepatocytes from mice fed 0.2% fenofibrate (ϩFF) or standard chow diet (ϪFF) for 7 days were metabolically labeled with [ 35 S]methionine/cysteine for 30 min, followed by a chase period in the absence of radiolabeled amino acids for the times indicated. SR-BI was immunoprecipitated, and the turnover of newly synthesized SR-BI was measured and is represented as the percentage remaining relative to time point zero. Total SR-BI levels in the immunoprecipitates (IP) were measured by probing the above blot with anti-SR-BI antibody. Similar levels of SR-BI were immunoprecipitated in all samples due to the use of limiting amounts of SR-BI antibody. B, hepatocytes were treated as described in A with or without the following inhibitors during a 3.5-h chase: BFA (100 M), nocodazole (N; 20 g/ml), ALLN (100 M), lactacystin (LAC; 10 M), or chloroquine (CQ; 50 M); BFA and BFA plus nocodazole both inhibited the turnover of SR-BI in hepatocytes from fenofibrate-fed mice. C, total cell lysates taken from the experiment in A before immunoprecipitation were used to measure steady-state levels of SR-BI, PDZK1, and ␤-actin as a loading control. SR-BI, but not PDZK1, in hepatocytes from fenofibrate-fed mice was significantly reduced compared with controls during the course of the experiment. content (27) did not inhibit the fenofibrate-induced degradation of SR-BI. In addition, DFX, which is a specific iron chelator that inhibits intracellular iron-dependent lipid peroxidation (28), failed to inhibit SR-BI degradation. DISCUSSION In this study we have shown that fenofibrate fed to mice causes enhanced hepatic SR-BI degradation in a post-ER, postplasma membrane compartment independent of PDZK1. Ini-tial studies using transgenic mice overexpressing a small PDZK1-associated protein, MAP17, in liver led to the enhanced turnover of PDZK1, resulting in the complete absence of PDZK1 and an 85-90% reduction in SR-BI (29). In addition, mutation of the PDZK1-interacting domain of SR-BI that abolished SR-BI binding to PDZK1 blocked the localization of SR-BI to the cell surface of hepatocytes (16). However, in nonpolarized hepatocytes isolated from MAP17 transgenic mice, the degradation of PDZK1, but not of SR-BI, was enhanced, indicating that PDZK1 plays an essential role in cell surface expression of SR-BI only in polarized hepatocytes (29). Indeed, it has long been reported that in other non-polarized cell models such as COS or HEK293 cells, the PDZK1-interacting domain of SR-BI does not play a role in SR-BI cell surface expression (30 -32), further supporting our conclusions that PDZK1 plays a role in SR-BI expression in polarized hepatocytes. Definitive evidence that PDZK1 plays an essential role in SR-BI expression in liver was shown in gene-targeted mice lacking PDZK1 (17). In the study by Kocher et al. (17), it was shown that SR-BI levels were reduced by 85% compared with control levels in PDZK1-deficient mice. However, the SR-BI levels in PDZK1 heterozygous mice were not shown in that study. Here we show that in our PDZK1-deficient mice, SR-BI levels are similarly reduced to levels previously reported by Kocher et al. (17). However, mice heterozygous for PDZK1 showed no decrease in SR-BI, indicating that PDZK1 is essential but not limiting in the liver for SR-BI expression. This finding is relevant to the observation that in C57BL/6J mice fed fenofibrate, PDZK1 levels were reduced by 50% compared with controls, whereas SR-BI was reduced by 85% compared with controls. Moreover, we showed that in PDZK1-deficient mice still exhibiting 5-15% of SR-BI found in controls, fenofibrate feeding further reduced SR-BI to almost undetectable levels. Mice with a 129 genetic background fed fenofibrate showed no decrease in PDZK1 but showed an 85% decrease in SR-BI. Taken together, these results indicate that fenofibrate acts to reduce SR-BI independently of PDZK1 by a mechanism that likely acts directly on SR-BI in a PPAR␣-dependent fashion. In addition, we can conclude that decreased PDZK1 by fenofibrate is dependent on genetic background. Of note is the decrease in PDZK1 that Mardones et al. (21) reported was greater than the 50% decrease in PDZK1 that we consistently observed, a difference likely related to the method of fenofibrate delivery and effective dose given to mice.
Using isolated hepatocytes from fenofibrate-fed mice, we found that SR-BI degradation was consistently and significantly enhanced relative to control cells, despite variations in the absolute levels of degradation between experiments (Figs. 6 and 8). Because these experiments utilized freshly isolated primary hepatocytes, such variation was likely due to inherent differences between mice. These findings are consistent with a model in which fenofibrate acts to directly up-regulate a degradation pathway for SR-BI independent of PDZK1. The degradation pathway enhanced by fenofibrate occurred in a post-ER, post-plasma membrane compartment because brefeldin A blocked degradation, and the amount of newly synthesized SR-BI reaching the plasma membrane was unchanged compared with controls. The machinery that mediates fenofibrateenhanced degradation of SR-BI does not appear to involve the proteasome because lactacystin or ALLN failed to inhibit the degradation of SR-BI. Therefore, we can rule out the ER/proteasome-associated degradation pathway that involves the proteasome and is a major pathway for the quality control of proteins that enter the ER during translation (33,34). Moreover, we found that neither ALLN, an inhibitor of both the proteasome and calpain protease, nor chloroquine, an inhibitor FIG. 7. Fenofibrate does not inhibit the trafficking of SR-BI to the plasma membrane. A, hepatocytes from mice fed 0.2% fenofibrate (ϩFF) or standard chow diet (ϪFF) for 7 days were metabolically labeled with [ 35 S]methionine/cysteine for the times indicated, followed by biotinylation of the cell surface at each time point. Total SR-BI was immunoprecipitated (designated by the arrow) using limiting amounts of an antibody that recognizes the C terminus of SR-BI, and 15% of total immunoprecipitated SR-BI (total) was saved. The C-terminal SR-BI antibody frequently immunoprecipitates a second higher molecular weight band that does not correspond to SR-BI. Biotinylated SR-BI (surface) in the remaining immunoprecipitated SR-BI was isolated using streptavidin-agarose beads. Both total and surface 35 S-SR-BI (newly synthesized) are shown, and the average fold increase in synthesis is shown below each lane. The ratio of newly synthesized cell surface to newly synthesized total SR-BI is shown below each lane. Blots were probed with an anti-SR-BI antibody showing total levels of SR-BI in immunoprecipitates (IP) were similar between both groups. Note that the small decrease in mobility of the cell surface SR-BI was due to its modification by biotin. This size shift is also noticeable in the 35 S-labeled SR-BI. B, total cell lysates from the experiment in A taken before immunoprecipitation were used to measure steady-state levels of SR-BI, PDZK1, and ␤-actin as a loading control. SR-BI, but not PDZK1, in hepatocytes from fenofibrate-fed mice was significantly decreased compared with controls throughout the course of the experiment.

FIG. 8. Antioxidants do not inhibit SR-BI degradation induced by fenofibrate.
Hepatocytes from mice fed 0.2% fenofibrate (ϩFF) or standard chow diet (ϪFF) for 7 days were metabolically labeled with [ 35 S]methionine/cysteine for 30 min, followed by a chase period in the absence of radiolabeled amino acids for 3.5 h. 0 h represents the time immediately after the 30-min labeling period. During the chase period, vitamin E succinate (VitE; 150 M) or DFX (150 M) was added to media. SR-BI was immunoprecipitated, and the turnover of newly synthesized SR-BI was measured and is represented as the average percentage remaining relative to time point zero of two independent samples. Total SR-BI levels in the immunoprecipitates (IP) were measured by probing the above blot with anti-SR-BI antibody. Similar levels of SR-BI were immunoprecipitated in all samples due to the use of limiting amounts of SR-BI antibody. of lysosomal function, blocked SR-BI degradation. Interestingly, PPAR␣ knock-out mice had small but significantly higher levels of hepatic SR-BI without increased SR-BI mRNA, suggesting that endogenous activity of PPAR␣ may cause a basal level of SR-BI turnover.
Interestingly, some parallels between the results presented here and recent findings showing that the degradation of apoB in hepatocytes is highly sensitive to oxidant stress and occurs via a novel pathway involving post-ER, pre-secretory proteolysis (27,28). We tested the idea that fenofibrate induces the degradation of SR-BI as a result of increased reactive oxygen species produced as a product of ␤-oxidation. However, unlike the inhibition of oxidant stress-induced degradation of apoB by vitamin E or DFX (27), SR-BI degradation does not seem to be the result of oxidant stress because neither vitamin E nor DFX blocked its degradation. The mediators of the post-ER, presecretory proteolysis pathway and the pathway governing fibrate-induced degradation of SR-BI remain to be elucidated.
The physiological function of decreased SR-BI by fenofibrate may be related to the observation that fibrates significantly down-regulate the activity of cholesterol 7␣-hydroxylase, the rate-limiting enzyme in bile acid synthesis, in humans (35,36) and rodents (37)(38)(39), resulting in decreased bile acid secretion and a marked increase in biliary cholesterol secretion. More recently, it has been demonstrated that fibrates down-regulate both cholesterol 7␣-hydroxylase and sterol 27-hydroxylase gene expression, an alternative bile acid synthesis pathway, in a PPAR␣-dependent fashion (37). SR-BI expression positively correlates with biliary cholesterol secretion in mouse because biliary cholesterol secretion is reduced by ϳ45% in SR-BIdeficient mice relative to wild-type mice (11,12). Therefore, it may be possible that the down-regulation of SR-BI by fenofibrate is a compensatory mechanism for the hepatocyte to maintain cholesterol homeostasis.
Additional studies into the pathway leading to enhanced degradation of SR-BI by fenofibrate may reveal important mechanisms that control the hepatic turnover of SR-BI. Given the wide use of fibrate drugs in treating humans with atherosclerosis and dyslipidemia, it will be important to determine whether this pathway regulating SR-BI turnover occurs in human liver.