A chemical switch regulates fibrate specificity for peroxisome proliferator-activated receptor alpha (PPARalpha ) versus liver X receptor.

Fenofibrate is clinically successful in treating hypertriglyceridemia and mixed hyperlipidemia presumably through peroxisome proliferator-activated receptor alpha (PPARalpha)-dependent induction of genes that control fatty acid beta-oxidation. Lipid homeostasis and cholesterol metabolism also are regulated by the nuclear oxysterol receptors, liver X receptors alpha and beta (LXRalpha and LXRbeta). Here we show that fenofibrate ester, but not fenofibric acid, functions as an LXR antagonist by directly binding to LXRs. Likewise, ester forms, but not carboxylic acid forms, of other members of the fibrate class of molecules antagonize the LXRs. The fibrate esters display greater affinity for LXRs than the corresponding fibric acids have for PPARalpha. Thus, these two nuclear receptors display a degree of conservation in their recognition of ligands; yet, the acid/ester moiety acts as a chemical switch that determines PPARalpha versus LXR specificity. Consistent with its LXR antagonistic activity, fenofibrate potently represses LXR agonist-induced transcription of hepatic lipogenic genes. Surprisingly, fenofibrate does not repress LXR-induced transcription of various ATP-binding cassette transporters either in liver or in macrophages, suggesting that fenofibrate manifests variable biocharacter in the context of differing gene promoters. These findings provide not only an unexpected mechanism by which fenofibrate inhibits lipogenesis but also the basis for examination of the pharmacology of an LXR ligand in humans.

Originally developed as hypolipidemic therapeutic agents, the fibrate class of molecules has been extensively characterized as ligands for the nuclear receptor, peroxisome proliferator-activated receptor ␣ (PPAR␣) 1 (1). In humans, the major pharmacological effects of PPAR␣ activation by fibrates are the reduction of plasma cholesterol and triglyceride levels. Mechanistically, decreases in triglyceride-rich plasma lipoprotein levels could occur via accelerated catabolism, decreased syn-thesis, or both. Indeed, activation of PPAR␣ by fibrates promotes ␤-oxidation of fatty acids in the peroxisome as well as in the mitochondria, thus reducing the fatty acid available to the liver for triglyceride synthesis (2).
Conversely, under appropriate metabolic states, such as those when starved animals are fed a high carbohydrate diet, increased lipogenesis in liver occurs in part by means of transcriptional activation of genes associated with de novo fatty acid biosynthesis, such as fatty acid synthase (FAS), via the transcription factor, sterol regulatory element-binding protein-1 (SREBP1) (3)(4)(5). Coincidentally, recent studies revealed that both oxysterols and synthetic agonists for the nuclear receptor liver X receptor (LXR) indirectly activate the lipogenic program by LXR-mediated induction of SREBP1, thereby leading to the coordinate expression of major lipogenic enzymes and profound elevation of triglyceride levels in liver (6 -8). LXR responsiveness of the FAS promoter also is attributed to a conserved LXR/RXR binding site within the 5Ј-flanking region of the FAS gene (9). Moreover, LXR target genes are pivotal to overall lipid metabolism, since potent LXR agonists activate genes involved in catabolism of cholesterol to bile acids, in regulation of several genes important for reverse cholesterol transport from peripheral tissues, in high density lipoprotein accumulation, and in cholesterol excretion into bile or intestinal lumen. The battery of LXR target genes mediating these effects includes Cyp7A1, ATP-binding cassette A1 (ABCA1), ABCG1, apolipoprotein E, lipoprotein lipase, cholesterol ester transfer protein, ABCG5, and ABCG8 (10 -18). Thus, cholesterol and fatty acid homeostasis involves integration of synthesis and degradation pathways that are controlled by the nuclear proteins PPAR␣, SREBP, and LXR.
Repression of de novo fatty acid synthesis in liver by fenofibrate has been observed, but the mechanism of repression remains unknown (19). Here we describe a mechanism by which fenofibrate ester may decrease fatty acid synthesis both in vitro and in vivo by decreasing expression of SREBP1 and FAS. Our studies reveal that fenofibrate ester represses the expression of SREBP1 and FAS mRNA by directly binding to and antagonizing LXR. Interestingly, the ester form of multiple fibrate molecules bind LXR with greater affinity than their respective carboxylic acid derivatives bind to PPAR␣. Importantly, the effects of fenofibrate on LXR target gene repression appear promoter-specific; fenofibrate does not repress LXR agonist-induced ABC transporter gene expression either in liver or in macrophages. These findings provide an unexpected mechanism by which fenofibrate ester inhibits SREBP1 transcription and lipogenesis without negating the beneficial role LXR that plays in mediating secretion of sterols from the liver and in increasing reverse cholesterol transport from macrophages.
Quantitative PCR-Total RNA was isolated from cells or tissue by the Trizol method (Invitrogen). cDNA was synthesized using random hexamer primers and Omniscript (Qiagen, Valencia, CA). TAQMAN TM real time PCR analysis was applied using prepared reagents and protocols from Applied Biosystems (Foster City, CA). The primer sequences for human SREBP1 were as follows: forward, acagcccacaacgccatt; reverse, TGCCGAAGACAGCAGATTTATT; and probe, cgctaccgctcctccatcaatgaca. The primer sequences for mouse SREBP1 were as follows: forward, catcgactacatccgcttcttg; reverse, ttgcttttgtgtgcacttcgt; and probe, cacagcaaccagaagctcaagcagga. For mouse ABCA1, they were as follows: forward, GGTTTGGAGATGGTTATACAATAGTTGT; reverse, TTCCCGGAAACGCAAGTC; and probe, CGAATAGCAGGCTCCAAC-CCTGACC. For mouse ABCG5, they were as follows: forward, ttgcgatacacagcgatgct; reverse, tgactgcctctaccttcttgttgt; and probe, ccctctgccgcagctccg. For mouse ABCG8, they were as follows: forward, gctgcccgggatgatagag; reverse, ccggaagtcattggaaatctg; and probe, ttttccaccctgatccgtcg. All PCRs were performed in triplicate using an Applied Biosystems Prism 7900HT Sequence Detection System (Applied Biosystems). Relative levels of mRNA are expressed as a ratio of the target gene mRNA values to the values obtained using the 18 S rRNA probe set (Applied Biosystems).
Scintillation Proximity Assay (SPA)-The scintillation proximity radioligand binding assay for LXR was previously described (21). Briefly, we utilized 800 ng of baculovirus-expressed His-tagged LXR␣LBD protein (amino acids 162-447) or 600 ng of LXR␤LBD protein (amino acids 202-461), 25 nM [ 3 H]25-Hydroxycholesterol (Amersham Biosciences), 0.05 mg of yttrium silicate polylysine-coated SPA beads (Amersham Biosciences), and varying concentrations of competitor per well of a 96-well OptiPlate (Packard Instrument Co.). Protein, radioligand, and competitor were added to the plate. SPA beads were then added to the assay plate followed by 10 min of gentle shaking at room temperature protected from light. The plates were incubated in the dark at room temperature for 2 h prior to reading in a TopCount plate reader (Pack-ard Instrument Co.). PPAR␣ binding assays were performed using the ABCD assay (22) as described previously (23).
Animal Studies-To determine fenofibrate ester and acid concentrations in liver, mice (129Sv) were dosed for 7 days with 300 mg/kg/day fenofibrate, and following a final dose, perfused livers were collected at various time points. Individual liver samples (1.0 g) were homogenized in acetonitrile. Debris was pelleted by centrifugation. Supernatant (50 l) was diluted with 150 l of 0.2% formic acid. Fenofibrate ester and acid concentrations were measured on a Sciex API 3000 mass spectrophotometer. To measure SREBP1, FAS, ABCA1, ABCG5, and ABCG8 gene expression, mice (129Sv) were dosed for 7 days with 50 mg/kg/day T0901317, 100 mg/kg/day fenofibrate, or a combination of both. Liver expression of the various mRNAs was measured by quantitative PCR.

Fenofibrate Represses LXR-induced Expression of SREBP-
To determine whether fenofibrate can repress de novo fatty acid synthesis at the level of gene transcription, we studied the regulation of SREBP1-mediated gene transcription by fenofibrate. A stable HepG2 cell line harboring an integrated luciferase reporter gene driven by the minimal thymidine kinase promoter preceded by three tandem repeats of a sterol response element was generated (HepG2 3XSRE). HepG2 3XSRE cells were treated with the LXR agonist, T0901317, in the absence or presence of increasing concentrations of fenofibrate (Fig. 1A). The LXR agonist effectively caused a 5-fold increase in SREBPinduced transcription of the reporter gene; fenofibrate alone did not significantly affect reporter activity. When added together, fenofibrate inhibited LXR-induced SREBP activity in a dose-dependent manner (IC 50 ϳ25 M).
We explored whether fenofibrate represses LXR-mediated induction of endogenous SREBP1 mRNA expression in HepG2 cells by quantitative PCR analysis (Fig. 1B). The LXR agonist increased SREBP1 mRNA levels 4-fold as shown previously, and fenofibrate alone had no effect on basal levels of SREBP1 mRNA (6). The addition of fenofibrate to T0901317-treated cells resulted in a dose-dependent reduction of induced SREBP1 mRNA levels, which is consistent with the decrease observed in HepG2 3XSRE cells. These data indicate that fenofibrate may repress genes required for de novo fatty acid synthesis by disrupting LXR-mediated SREBP1 gene activation.
Fenofibrate Binds to the LXR Ligand Binding Domain-Whether fenofibrate does indeed alter LXR-mediated transactivation was tested by assessing the ability of fenofibrate to inhibit an LXRE-luciferase reporter gene in cells expressing exogenous human LXR␣. Treatment of transfected HEK293 cells with T0901317 caused powerful induction of the reporter gene. Fenofibrate inhibited T0901317-induced reporter activity in a dose-dependent manner with an IC 50 of ϳ25 M ( Fig. 2A). Similarly, the natural LXR agonist (22R)-hydroxycholesterol (22RHC) induced LXR␣ activity that was potently inhibited by fenofibrate (data not shown). Interestingly, basal LXR␣ activity also was inhibited, suggesting either the presence of an endogenous ligand or low constitutive activity of the receptor that can be inhibited by an antagonist (24). Similar results were obtained with LXR␤ (data not shown). Hence, fenofibrate can inhibit LXR-dependent transcription, irrespective of the nature of the LXR ligand.
To test the possibility that fenofibrate competes directly for agonist binding to the ligand binding domain (LBD) of LXR, we used chimeric receptors encoding the LXR-LBD fused to the Gal4 DNA binding domain that were co-transfected with a Gal4-responsive luciferase reporter into HEK293 cells. The LXR agonist elicited a robust transcriptional response from Gal4-LXR␣-LBD. Whereas fenofibrate by itself did not induce transcription (Fig. 2B), it caused potent inhibition of LXR ag-onist-induced LBD-mediated transcriptional activation. Thus, it became apparent that fenofibrate action might be due to direct modulation of LXR. If true, fenofibrate should bind to recombinant LXR-LBD proteins in vitro; therefore, an SPA was developed using radiolabeled 25-hydroxycholesterol as ligand (Fig. 2C) (21). Fenofibrate bound to LXR␣, potently displacing the tritiated 25-hydroxycholesterol. These findings suggest that in addition to increasing ␤-oxidation of fatty acids via activation of PPAR␣, fenofibrate also may modulate LXR-mediated transcription of genes involved in the lipogenic signaling pathway by direct antagonism of LXR.
Fibrate Esters, but Not Fibric Acids, Bind to LXR and Inhibit LXR Transactivation-Several fibrate derivatives have been used clinically in the treatment of hyperlipidemia, namely bezafibrate, gemfibrozil (Lopid), and fenofibrate (Tricor). In contrast to fenofibrate, which is formulated in its ester form, the other fibrates are utilized as carboxylic acids. We examined both the ester and acid derivatives of each fibrate molecule and determined whether other fibrate esters or fibric acid derivatives could modulate LXR-dependent transcription. HEK293 cells transiently transfected either with LXR␣ and LXRE-lu-  6). C, fenofibrate binds LXR-LBD directly. The scintillation proximity radioligand binding assay for LXR was previously described (21). Receptor protein (His 6 -hLXR␣LBD, amino acids 162-447, 800 ng) immobilized on SPA beads was incubated with 25 nM 3 H-labeled 25-hydroxycholesterol in the presence of increasing amounts of fenofibrate ester or 22RHC. These results are representative of at least two separate studies. ciferase plasmids or with Gal4-LXR␣ and 5XUAS-luciferase plasmids were treated with T0901317 in the absence or presence of increasing concentrations of the fibrate esters and acids. The LXR agonist induced transcriptional activity, which was sustained in the presence of all carboxylic acid PPAR␣ ligands (Fig. 3). However, each of the corresponding fibrate esters repressed agonist-induced transcription with varying potencies. Consistent with this observation, the fibrate esters bound only to the LXRs, whereas the fibric acids were specific for PPAR␣ (Table I). These data expose a degree of promiscuity and similarity in recognition of ligands by these two nuclear receptors. Clearly, the ester/acid moiety acts as a switch that determines LXR versus PPAR␣ affinity, yet the remaining portion of the ligand seems to have relatively little impact on this selectivity. That this relationship is not restricted to a single structural class of ligands (fibrates versus Wy14643) indicates that ligand sharing by PPAR␣ and the LXRs may be common. Indeed, the primary amino acid sequences of human PPAR␣ and human LXR␣ ligand binding domains are 30% identical and share 50% similarity. Since PPAR␣ and the LXRs are essential regulators of lipid metabolism, related natural ligands that regulate both receptors may exist.

Fenofibrate Ester Accumulates in Liver and Represses LXRmediated Induction of SREBP1 and FAS-
The fenofibrate ester is rapidly converted into the carboxylic acid form by esterases in the liver and plasma (25). It has been assumed that the pharmacologically relevant form of fenofibrate is the acid form due to its specificity for PPAR␣ and the lack of detectable ester in the plasma. However, since the primary target tissue for the hypolipidemic effects of the fenofibric acid (PPAR␣) and ester (LXR) is the liver, we compared the relative levels of these two compounds in the liver in order to determine the relevance of the LXR antagonist component of fenofibrate action. Mice were treated with fenofibrate for duration of 7 days followed by collection of plasma and liver at multiple time points after the final dose administration. Consistent with previous reports, we observed that fenofibrate ester was undetectable in mouse serum, whereas the fenofibric acid metabolite was easily detected (data not shown). In contrast, fenofibrate ester was readily detectable in the liver (Fig. 4A). The highest concentration of fenofibrate ester, 3.7 g/g, was present at 30 min. After 1 h, fenofibrate ester reached steady-state levels in the liver that persisted for 24 h. Although the ester levels were lower than the acid levels in the liver, the relative amount is consistent

Fibrate Esters Modulate LXR
with the potential for affecting LXR activity, based on the greater affinity of the ester for the LXRs relative to acid affinity for PPAR␣. Furthermore, the concentration of fenofibrate that is required to inhibit LXR transactivation might be achievable in human liver, given commonly prescribed doses of fenofibrate ester (160 mg/day). In order to determine whether indeed there is an LXR-mediated effect of fenofibrate treatment in vivo, we examined the ability of this drug to antagonize the expression of lipogenic LXR target genes in T0901317 treated mice. After 7 days of treatment either with vehicle, fenofibrate, T0901317, or a combination of T0901317 and fenofibrate, the level of SREBP1 and FAS mRNA expression in liver was measured by quantitative PCR. Mice treated with T0901317 expressed 6-fold higher levels of SREBP1 mRNA and 12-fold higher levels of FAS mRNA, as compared with the vehicle-treated mice (Fig.  4B). Administration of fenofibrate ester alone promoted a decrease in the basal level of SREBP1 and FAS expression. This may be due to antagonism of endogenous LXR agonists in the liver. In agreement with the ability of fenofibrate to antagonize LXR-induced expression of SREBP1 mRNA in vitro, we observed an ϳ3-fold reduction in T0901317-induced SREBP1 and FAS mRNA levels in vivo. Thus, the novel LXR antagonist action of the fenofibrate ester may be relevant in vivo.
Fenofibrate Ester Does Not Inhibit LXR Transactivation of ABC Transporters in Liver and Macrophages-Increasing evidence supports the notion that LXRs serve as sensors that respond to elevated intracellular sterol concentrations by increasing the expression of genes that reduce the sterol burden. ApoA1-mediated cholesterol efflux from macrophages to high density lipoprotein requires the expression of the ABCA1 transporter, which is controlled, in part, by LXR/RXR heterodimers (12, 26 -33). Likewise, LXR regulates the expression of ABCG5 and ABCG8 genes that promote secretion of sterols from the liver and decrease intestinal sterol retention, thereby LXR␣ binding assay was performed as described in the legend to Fig. 2 using increasing concentrations of fenofibrate ester or acid, gemfibrozil ester or acid, bezafibrate ester or acid, clofibrate ester or acid, and Wy14643 ester or acid. PPAR␣ binding assays were performed using the ABCD assay (22) as described previously (23). K i values were determined for each. facilitating coordinate regulation of sterol efflux and elimination. We explored the possibility that fenofibrate ester could adversely repress LXR-induced ABC gene expression both in liver and in THP-1 macrophages. The same liver samples from mice treated with T0901317 and fenofibrate in Fig. 4 were evaluated for ABCA1, ABCG5, and ABCG8 gene expression levels. Both ABCA1 and ABCG5 mRNA levels were increased ϳ5-fold, and ABCG8 increased 3-fold upon treatment with T0901317 as compared with the vehicle-treated mice (Fig. 5A). Administration of fenofibrate ester alone caused slight yet significant increases in ABCA1, ABCG5, and ABCG8 mRNA levels (p Ͻ 0.05). In contrast to the ability of fenofibrate to antagonize LXR-induced expression of SREBP1 and FAS mRNA, fenofibrate did not reduce LXR-induced expression of the ABC transporters in liver. Likewise, the addition of fenofibrate either to 22RHC-or to T0901317-stimulated THP-1 macrophage cells did not diminish LXR-induced expression of endogenous ABCA1 mRNA (Fig. 5B). The observed increase in ABC transporter expression could be due to PPAR␣ activation of the endogenous LXR␣ promoter causing increased in vivo levels of LXR␣ (34,35), or alternatively, the fibrate ester may be displaying promotor/tissue selective partial agonism/antagonism that has been reported for ligands of other nuclear receptors such as the selective estrogen receptor modulators. Unlike the endogenous LXR antagonist geranylgeranyl pyrophosphate, fenofibrate does not negate the beneficial role LXR plays in reverse cholesterol transport in macrophages (36,37). Rather, fenofibrate exhibits LXR antagonist activity on SREBP1 and FAS gene transcription and either neutral or partial agonist activity on ABC transporter expression in liver and macrophage cells. DISCUSSION As a transcriptional regulator of genes, such as the ABC transport proteins, apolipoprotein E, lipoprotein lipase, and cholesterol 7␣-hydroxylase, LXR has emerged as an attrac-tive target for the development of drugs for cardiovascular disease therapy (11, 12, 14 -16, 18). However, in addition to the desirable pharmacologic effects of increasing reverse cholesterol transport from the periphery and hepatic cholesterol catabolism upon activating LXR target gene transcription, the undesirable side effect of increasing hepatic lipogenesis occurs in the presence of LXR agonists. Here, we have described a novel mechanism by which the clinically utilized hypolipidemic compound, fenofibrate, antagonizes LXR and represses lipogenic gene expression in liver, but not members of the ABC transporter gene family (ABCA1, ABCG5, and ABCG8).
By directly binding to the LXR ligand binding domain, various fibrate ester compounds displace a naturally occurring LXR ligand in vitro, and they selectively repress LXR agonistinduced liver target gene transcription in vivo. We propose that fenofibrate could induce distinct structural changes in LXR that influence its ability to interact with other proteins, such as transcription factors residing at selective target gene promoters or with coactivators or corepressors that are critical for target gene regulation. For other nuclear receptors such as the estrogen and progesterone receptors, the relative expression of various cofactors in target tissues and the promoter context have been demonstrated to be essential for the determination of the agonist/antagonist character of a particular ligand (38 -41). Similarly, the action of fibrate esters on LXR may be determined by the exact sequence of the LXRE, the contribution of additional transcription factors, and the influence of local chromatin structure of selected LXR target genes. Whether these specific promoters differentially utilize coactivators and corepressors, as is the case for estrogen-, raloxifene-, and tamoxifen-stimulated gene expression, also remains to be investigated (39).
Although obscured by mixed PPAR␣/LXR activity, fenofibrate ester is the first clinical compound known to target LXR,

FIG. 4. LXR activation of SREBP1 and FAS gene transcription is antagonized by fenofibrate in vivo.
A, time course of fenofibrate ester and fenofibric acid accumulation in liver tissue. Mice (129Sv) were dosed for 7 days with 300 mg/kg/day fenofibrate, and following a final dose, perfused livers were collected at the various time points (n ϭ 3 for each time point). Individual liver samples (1.0 g) were homogenized in acetonitrile, and debris was pelleted by centrifugation. Supernatant (50 l) was diluted with 150 l of 0.2% formic acid. Fenofibrate ester and acid concentrations were measured on a Sciex API 3000 mass spectrophotometer. B, LXR-induced SREBP1 and FAS mRNA expression is repressed in liver upon fenofibrate administration. Mice (129Sv) were dosed for 7 days either with vehicle, 50 mg/kg/day T0901317, 100 mg/kg/day fenofibrate, or a combination (n ϭ 3 in each group). Liver expression of the mRNAs was measured in triplicate by quantitative PCR (see "Experimental Procedures"). The results are expressed as a ratio of target gene mRNA transcripts to 18 S rRNA transcripts (mean Ϯ S.D.). These results are representative of two separate studies.

Fibrate Esters Modulate LXR
and it may be a useful tool to explore LXR pharmacology. Notably, fibrate esters bound to LXR with higher affinity than was anticipated, based on fibric acid potencies in PPAR␣-based assays (Table I) (42). Combined with further identification of molecules that interact with LXR, these studies may contribute to the design of new LXR modulators that can meet the clinical need of decreasing atherosclerotic lesions without promoting hypertriglyceridemia.
Fenofibrate ester is subject to esterases and is rapidly converted to a carboxylic acid in the plasma and liver. Indeed, we readily detected fenofibric acid in mouse plasma samples but were unable to detect fenofibrate ester in plasma, yet fenofibrate ester reached steady-state levels in liver that persisted for at least 24 h (Fig. 4A). Therefore, since fenofibrate ester did reach significant steady-state levels in the liver, it is reasonable to propose that fibrate esters (i.e. Tricor and Lopid) may decrease hepatic triglyceride levels, at least in part, by decreasing the expression of SREBP1 and FAS through modulation of LXR.
The observation that other acidic PPAR␣ agonists can be transformed into LXR antagonists by simple modification to an ester indicates similarity in ligand recognition. This suggests that there may be structurally similar natural ligands for these two receptors that underlie the evolutionary need for this degree of conservation of ligand recognition between PPAR␣ and the LXRs. In fact, certain fatty acids already have been shown to function both as PPAR␣ agonists and as LXR antagonists. Arachidonate (C20:4) binds to and activates human PPAR␣ at micromolar concentrations (IC 50 ϭ 1.2 M), and at approximately the same concentration, arachidonate antagonizes synthetic ligand binding to LXR and transcription of the LXR target gene, SREBP1 (43). Interestingly, antagonism of LXR by polyunsaturated fatty acids such as C20:4,n6 or C22:6,n3 occurs in an isoform-selective manner; LXR␣ is antagonized, whereas LXR␤ is not (44). Such observations of isoform-selective modulation further imply the molecular possibility of regulating certain LXR target genes while having no effect on others through the use of distinct, isoform-selective LXR ligands.
In summary, we have demonstrated that not only does the fibrate class of hypolipidemic compounds function through activation of PPAR␣, leading to induction of genes that control fatty acid ␤-oxidation; they also decrease SREBP1 and FAS gene expression by antagonizing LXR-mediated transcription. Like the activities of some selective estrogen receptor modulators, fibrate esters display LXR partial agonist/antagonist activity that is dependent on the target gene context. These data ascribe a novel regulatory function to the clinically utilized fibrate drugs and have therapeutic implications for identification of compounds that increase cellular cholesterol efflux through LXR yet counteract the accumulation of triglycerides by utilizing a promoter-selective mode of action. A, mice (129Sv) were dosed for 7 days either with vehicle, 50 mg/kg/day T0901317, 100 mg/kg/day fenofibrate, or a combination (n ϭ 3 in each group, same mice as in Fig. 4). Liver expression of ABCA1, ABCG5, and ABCG8 mRNAs was measured in triplicate by quantitative PCR (see "Experimental Procedures"). The results are expressed as a ratio of target gene mRNA transcripts to 18 S rRNA transcripts (mean Ϯ S.D.). B, differentiated THP-1 macrophage cells were treated with either 22RHC (10 M) or T0901317 (10 M), with a dose response of fenofibrate (as indicated) or with T0901317 (10 M) or 22RHC (10 M) in combination with a dose response of fenofibrate, as indicated. 20 g of total RNA was subjected to Northern analysis using an ABCA1 cDNA fragment as probe; 36B4 served as a loading control. Gene expression was quantitated by scanning densitometry and is shown in the histogram below.