Inhibition of cholesterol absorption associated with a PPAR -dependent increase in ABC binding cassette transporter A1 in mice

Dietary supplementation with the peroxisome proliferator-activated receptor α (PPARα) ligand WY 14,643 gave rise to a 4- to 5-fold increase in the expression of mRNA for the ATP binding cassette transporter A1 (ABCA1) in the intestine of normal mice. There was no effect in the intestine of PPARα-null mice. Consumption of a high-cholesterol diet also increased intestinal ABCA1 expression. The effects of WY 14,643 and the high-cholesterol diet were not additive. WY 14,643 feeding reduced intestinal absorption of cholesterol in the normal mice, irrespective of the dietary cholesterol concentration, and this resulted in lower diet-derived cholesterol and cholesteryl ester concentrations in plasma and liver. At each concentration of dietary cholesterol, there was a similar significant inverse correlation between intestinal ABCA1 mRNA content and the amount of cholesterol absorbed. The fibrate-induced changes in the intestines of the normal mice were accompanied by an increased concentration of the mRNA encoding the sterol-regulatory element binding protein-1c gene (SREBP-1c), a known target gene for the oxysterol receptor liver X receptor α (LXRα). There was a correlation between intestinal ABCA1 mRNA and SREBP-1c mRNA contents, but not between SREBP-1c mRNA content and cholesterol absorption. These results suggest that PPARα influences cholesterol absorption through modulating ABCA1 activity in the intestine by a mechanism involving LXRα.

The ATP binding cassette transporter A1 (ABCA1) is a membrane-bound protein that is thought to facilitate the transport of cholesterol and phospholipids out of cells (1). Deficiency of ABCA1 gives rise to Tangier disease and familial HDL deficiency, both of which are characterized by a low concentration of HDL in plasma (2)(3)(4)(5). In Tangier disease, cholesteryl esters accumulate in macrophagederived foam cells present in tonsils, liver, and intestinal mucosa (6). Consistent with this observation, it has been shown that primary and transformed macrophages increase ABCA1 mRNA content and cholesterol efflux in response to cholesterol loading and oxysterols (5,(7)(8)(9). Thus, ABCA1 could be an important component of the apolipoprotein A1-mediated reverse cholesterol transport pathway for the removal of cholesterol from macrophages and other peripheral cells and its return to the liver as HDL.
Further evidence using ABCA1-null mice suggests that ABCA1 may have an additional role in maintaining wholebody cholesterol homeostasis by transporting dietary cholesterol out of intestinal absorptive cells and returning it to the intestinal lumen (10). However, this conclusion is somewhat controversial, because a second study has reported that ABCA1 deficiency gave rise to a decrease in cholesterol absorption (11). Nevertheless, in normal mice, administration of a synthetic ligand for retinoid X receptor (RXR) markedly stimulated intestinal ABCA1 mRNA content while abolishing net absorption of dietary cholesterol (8). The heterodimeric partner for RXR in regulating ABCA1 transcription in the gut was liver X receptor (LXR), which had previously been shown to mediate the effects of oxysterols on ABCA1 mRNA levels in macrophages (7,12).
A second partner of RXR, peroxisome proliferator-activated receptor (PPAR), has also been implicated in the control of expression of ABCA1 in macrophage foam cells (13), although the mechanism remains obscure and it is not known if it operates in other cell types. PPARs regulate a number of important genes involved in lipid and glucose metabolism and are activated by unsaturated fatty acids and their derivatives (14). The fibrate group of drugs, used extensively as lipid-lowering agents, are pharmacological activators of PPAR ␣ , the form found mainly in liver, muscle, and intestine (15). A role for PPAR ␣ in modulating intestinal cholesterol metabolism was suggested by the observation that gemfibrozil, a pharmacological ligand of PPAR ␣ , suppressed cholesterol absorption from the gut into mesenteric lymph in rats (16). Elucidation of the physiological role of PPAR ␣ in the control of lipid metabolism has been simplified by the availability of a PPAR ␣ knockout mouse in which the ligand binding domain has been disrupted (17). These animals have an increased plasma HDL concentration (18) and reduced mitochondrial ␤ -oxidation (19), with an impaired response to fasting (20,21) and a disturbed circadian rhythm of genes regulating fatty acid and cholesterol synthesis (22). In this report, we have used normal and PPAR ␣ -null mice to examine whether PPAR ␣ is involved in the regulation of cholesterol absorption through changes in intestinal ABCA1.

MATERIALS AND METHODS
PPAR ␣ -null mice (17) bred onto a SV/129 genetic background were kindly provided by Dr. J. Peters and Dr. F. J. Gonzalez (National Institutes of Health, Bethesda, MD). Wild-type SV/ 129 mice were used as controls. All mice were male, and were used between the ages of 14 and 20 weeks. Animals were maintained in temperature-controlled rooms (22-24 Њ C) on a 12 h light/12 h dark cycle (22). When required, the commercial highcarbohydrate pelleted diet containing 0.02% cholesterol (Special Diet Service, Witham, Essex, UK) was supplemented with 2% cholesterol ( Ͼ 99% purity; Sigma-Aldrich, Dorset, UK) or 0.1% WY 14,643 (Chemsyn Science Laboratories, Lenexa, KS) and fed for 7 days. Food, and thus cholesterol consumption, was measured on a daily basis, and was similar for each genotype. Mice were anaesthetized with pentobarbital, and samples of blood were taken from the descending vena cava. The proximal intes-tine between approximately 5 and 15 cm from the stomach was removed, rinsed with ice-cold Dulbecco's phosphate-buffered saline, and frozen in liquid nitrogen. All procedures were conducted with the appropriate ethical approval.
The frozen tissues were ground to a powder under liquid nitrogen, and total RNA was extracted from portions of powder using the RNAzol kit (Biogenesis, Poole, UK). The amount of mRNA was assayed by reverse transcription (23) followed by realtime PCR using the ABI PRISM Sequence Detection System (PE Applied Biosystems, Foster City, CA) using the primers and probes as given previously (22) or as shown in Table 1 . Reactions were carried out in triplicate in 30 l of TaqMan Universal PCR Master Mix with ␤ -actin as internal standard (22). All values were related to a curve generated by a standard liver or intestinal preparation and were corrected for ␤ -actin mRNA content. ABCA1 protein was detected by immunoblotting the intestinal membrane fraction (24) using a specific antipeptide polyclonal antibody raised in rabbits to residues 1,201-1,211 of the human ABCA1 protein (ab7360, from Abcam, Cambridge, UK). Blots were scanned to give a semiquantitative estimate of changes in intensity.
A previously described method using the nonabsorbed plant sterol marker sitostanol (25) was adapted to measure cholesterol absorption in the present work. Trace quantities of [4-14 C]cholesterol and [5,6-3 H]sitostanol were incorporated into the appropriate diet ( ‫ف‬ 20 Ci and ‫ف‬ 25 Ci, respectively, per 100 g chow) and fed for the final 2 days of the 7 day regimen. Portions of the different types of diet were used for measurement of the exact 3 H/ 14 C sterol ratios following lipid extraction. For the chow, chow ϩ WY 14,643, high-cholesterol, and high-cholesterol ϩ WY 14,643 diets, these values were 0.84, 0.82, 0.74, and 0.75, respectively. The feces were collected after the final 2 day period, frozen in liquid nitrogen, ground to a powder, and an aliquot used for total lipid extraction (26). A part of the extract was used to determine the 3 H/ 14 C ratio by scintillation counting. The remainder was subjected to thin-layer chromatography (27) to separate the esterified and nonesterified sterol fractions.
Plasma and tissue lipid concentrations were assayed as described previously (22) following solvent extraction. Labeled esterified and nonesterified sterols were separated by thin-layer chromatography of total lipid extract (27) and their radioactivities assayed by scintillation counting using a Beckman LS 6500 scintillation counter. Tissue protein content was determined by the method of Lowry et al. (28). Statistical analysis was carried out using Student's t -test for unpaired data.

Intestinal ABCA1 and cholesterol absorption
Wild-type and PPAR ␣ -null mice were fed for 7 days with diets containing high and low concentrations of cholesterol, either alone or in combination with the potent fibrate, WY 14,643. Changes in the abundance of intestinal ABCA1 mRNA were then compared with changes in cholesterol absorption from the gut. In these experiments, all animals were killed during the light phase of the diurnal cycle, when stimulation of ABCA1 mRNA expression by the PPAR ␣ ligand was greatest. The animals were apparently unaffected by the dietary supplements, as evidenced by a similarity in food intake and overall weight. The highcholesterol diet produced a slight increase in hepatic triacylglycerol content in the normal animals (29.6 Ϯ 6.1 mol/g to 41.3 Ϯ 4.3 mol/g; P Ͻ 0.05), but not of the already raised level in the PPAR ␣ -null animals (50.5 Ϯ 3.0 mol/g to 58.7 Ϯ 12.8 mol/g). No other evidence of steatosis was observed.
Feeding the high-cholesterol diet alone increased the intestinal expression of ABCA1 mRNA to the same extent in both the normal and the PPAR ␣ -null mice ( Fig. 1A ). In-clusion of WY 14,643 in the low-cholesterol diet increased ABCA1 mRNA expression several-fold in the wild-type animals, but there was no response in the knockouts. Immunoblotting confirmed that these changes in mRNA level gave rise to similar changes in ABCA1 protein (Fig. 1B). In normal mice, inclusion of WY 14,643 in the high-cholesterol diet further increased the expression of intestinal ABCA1 mRNA, but only to a value similar to that with WY 14,643 alone (Fig. 1A). In PPAR ␣ -null mice, inclusion of WY 14,643 in the high-cholesterol diet did not further increase intestinal ABCA1 mRNA expression (Fig. 1A).
To measure cholesterol absorption, both the high-and low-cholesterol diets were labeled with trace amounts of [ 14 C]cholesterol and [ 3 H]sitostanol. Because sitostanol is not absorbed from the intestine and is excreted quantitatively with the feces, comparison of the 3 H/ 14 C ratio of the fecal nonesterified sterols with that of the dietary sterols provided a measure of the proportion of the dietary cholesterol that was absorbed (25). Addition of WY 14,643 to both the low-cholesterol and high-cholesterol diets significantly reduced the proportion of dietary cholesterol absorbed by the wild-type mice but had no effect on the PPAR ␣ -null mice ( Table 2 ). When expressed for each individual mouse, there was a significant inverse correlation between cholesterol absorption and the intestinal ABCA1 mRNA content at both high and low dietary cholesterol concentrations ( Fig. 2A ). A given increase in ABCA1 mRNA content was associated with a similar fall in absorption, irrespective of the amount of cholesterol in the diet (Fig. 2B).

Cholesterol concentrations in liver and plasma
To confirm the observed changes in cholesterol absorption, the amount of total and dietary-derived [ 14 C]cholesterol was determined in both liver and blood plasma. Under no conditions could we detect any tritium in the plasma, confirming that none of the dietary [ 3 H]sitostanol had been absorbed. Thin-layer chromatography of the extracted lipids showed that in each case, the [ 14 C] label was associated almost entirely with esterified and nonesterified cholesterol.
Changes in dietary [ 14 C]cholesterol absorption were reflected by parallel changes in the concentration of [ 14 C]labeled sterols in the liver. Feeding WY 14,643 to the normal mice, but not the PPAR ␣ -null mice, was accompanied by a 50% decrease in the amount of hepatic free [ 14 C]cholesterol and a 65% decrease in [ 14 C]cholesteryl ester ( Table 3 ). Feeding cholesterol increased the proportion of [ 14 C]cholesterol that became esterified. Inclusion of WY 14,643 in the high-cholesterol diet decreased the amount of diet-derived free and, particularly, esterified [ 14 C]cholesterol in the livers of normal mice, so that there was a significant reduction in the proportion that was esterified. These changes in supply of dietary cholesterol in the normal mice gave rise to significant changes in the total hepatic cholesterol content (Table 3). Cholesterol feeding doubled the amount of cholesterol in the liver, and inclusion of fibrate in the low-and high-cholesterol diets re-  duced it by 30% and 50%, respectively. The hepatic cholesterol content of the PPAR␣-null mice was higher than that of the wild type on the low-cholesterol diet, and was not affected by WY 14,643. The proportion of the liver cholesterol derived from the diet was about 5% for animals on the low-cholesterol diet and 80% for those on the high-cholesterol diet (Table 3). These proportions were reduced by fibrate in the wild-type but not in the PPAR␣null animals.
Blood plasma showed changes similar to although lesspronounced than those in liver, with significant reductions in [ 14 C]cholesterol content after feeding WY 14,643 ( Table 4). Cholesteryl ester content was particularly resistant to change, although small significant effects of fibrate were observed in the normal animals on the high-cholesterol diet. In the absence of cholesterol or fibrate, the PPAR␣-null animals retained significantly more of the dietary-derived cholesterol in the plasma than did the wild type ( Table 4).

Involvement of intestinal LXR␣
If PPAR␣ acts via LXR␣ in the intestine, as it does in macrophages (13), it might be expected that activation by WY 14,643 would enhance the mRNA concentration of other LXR␣-responsive genes, such as sterol-regulatory element binding protein-1c (SREBP-1c) (29). To examine this, we assayed the intestinal mRNA content of SREBP-1c and compared it to that of the other SREBP isoforms. Inclusion of WY 14,643 in the diet produced a small increase in the mRNA content of SREBP-1a and SREBP-2, but had a much larger effect on SREBP-1c (Fig. 3). This response to the drug was absent in the intestines of the PPAR␣-null mice. The patterns of SREBP mRNA expression were reflected by changes in the mRNA content of their own target genes, fatty acid synthase and acetyl-CoA carboxylase for SREBP-1c, and HMG-CoA reductase for SREBP-2 (Fig. 3). PPAR, peroxisome proliferator-activated receptor. Normal and PPAR␣-null mice were fed a low-(0.02% w/w) or high-(2.0% w/w) cholesterol diet supplemented (ϩ) or not supplemented (Ϫ) with WY 14,643 (0.1% w/w) for 7 days. Values are expressed as the mean (n ϭ 4 in each group) Ϯ SEM.
a Significantly different (P Ͻ 0.05) from those obtained from mice fed the corresponding diet without WY 14,643.
b Significantly different (P Ͻ 0.05) from the wild-type animals fed the corresponding diet. c Significantly different (P Ͻ 0.01) from the wild-type animals fed the corresponding diet. Feeding a high-cholesterol diet increased intestinal SREBP-1c mRNA content 3-to 4-fold in both wild-type and PPAR␣-null mice (Fig. 4B). It had no effect on the mRNA levels for SREBP-1a (mean values Ϯ SEM, for wildtype mice, of 0.863 Ϯ 0.102 arbitrary units, without and 1.014 Ϯ 0.118 with cholesterol) or SREBP-2 (0.945 Ϯ 0.091 without and 0.919 Ϯ 0.196 with cholesterol). As for ABCA1, the effects of cholesterol and fibrate on SREBP-1c mRNA levels in wild-type animals were not additive (Fig.  4B). Over the whole range of diets and genotypes tested, there was a strong correlation between the intestinal content of SREBP-1c mRNA and that of ABCA1 (Fig. 5A). This suggests that the mechanism by which these compounds affect SREBP-1c may be the same as that by which they influence ABCA1. However, there was no indication that changes in SREBP-1c mRNA levels had any direct effect on cholesterol absorption. There was no significant correlation, with either the high-or low-cholesterol diet, between SREBP-1c mRNA content and the proportion of dietary cholesterol absorbed (Fig. 5B).
Intestinal LXR␣ mRNA content was the same in the normal and the PPAR␣-deficient mice, and was not affected by dietary manipulation (Fig. 4A).

DISCUSSION
The experiments presented here establish the existence of a strong inverse relationship between the expression of intestinal ABCA1 mRNA and the transfer of dietary cholesterol from the intestinal lumen into the body (Fig. 2). This observation provides strong support for the concept of ABCA1 acting as an intestinal gatekeeper to prevent excessive entry of exogenous cholesterol (30). Additionally,  it provides firm evidence that activation of intestinal PPAR␣ decreases dietary cholesterol absorption by a mechanism that involves an increase in intestinal ABCA1 mRNA and protein levels. This conclusion is supported by the antiparallel effects of the PPAR␣ ligand WY 14,643 on cholesterol absorption and ABCA1 mRNA levels in mice in which the PPAR␣ gene was intact, but a complete lack of either effect in mice in which the PPAR␣ gene was deleted. The physiological consequences of an inability to upregulate intestinal ABCA1 mRNA expression in a PPAR␣-dependent manner include a continuing high rate of dietary cholesterol flux to the liver in PPAR␣-null mice on a high-cholesterol diet, resulting in an almost 10-fold increase in hepatic cholesteryl ester compared with the wild type fed WY 14,643, and a significant increase in dietderived and total cholesteryl ester in the blood plasma. It is unlikely that PPAR␣ affects ABCA1 gene transcription directly, because the sequences of the ABCA1 promoter region that are conserved between human and mouse do not contain a typical PPAR␣ response element (31). However, the present results provide some clues as to the mechanism of the linkage between PPAR␣ activation and intestinal ABCA1 mRNA expression. It is known that activation of the oxysterol receptor LXR␣ in the intestine gives rise to an increase in ABCA1 expression and a reduction in cholesterol absorption (8). The most potent natural LXR␣ ligands in vitro are monohydroxy-cholesterol derivatives with the hydroxyl group at position 22 or 24 on the side chain (32). The physiological importance of these ligands in vivo is yet to be established (33). However, it has been shown that the concentration of potential oxysterol ligands, in particular 24S-hydroxycholesterol, can be increased in the liver by increased dietary cholesterol (34). Thus, it seems probable that it was oxysterols derived from dietary cholesterol, acting through LXR␣, that were responsible for the increased ABCA1 mRNA levels in the intestines of mice fed the high-cholesterol diet. The effects of cholesterol and WY 14,643 on ABCA1 mRNA levels in the intestine were not additive, suggesting that activation of PPAR␣ also increased ABCA1 mRNA by a mechanism that involved LXR␣. To examine this possibility, we used SREBP-1c, a known LXR␣ target gene (29,35,36), as a reporter for LXR␣-mediated responses. The results showed that activation of PPAR␣ by WY 14,643 increased expression of mRNA for SREBP-1c and its target genes in the intestines of the normal mice. This was not related to any general changes in cholesterol flux, because SREBP-2 mRNA levels were not similarly increased and were not affected by cholesterol feeding. WY 14,643 had no effect on SREBP-1c mRNA when the PPAR␣ gene was deleted, consistent with the idea that LXR␣ had not been activated in these animals. The closeness of the rela- tionship between the expression of ABCA1 and SREBP-1c mRNA also argues for the operation of a common mechanism. Despite this relationship, there was no correlation between SREBP-1c mRNA content and cholesterol absorption, indicating that SREBP-1c was probably not directly involved in any regulation of absorption by ABCA1.
In the intestine, we could find no evidence for a direct interaction between PPAR␣ and LXR␣ that inhibited transcription, as has been observed on promoter reporter constructs in vitro (37,38), or for any apparent antagonism of LXR␣ activation (39) or function (40,41) by PPAR␣ ligands, as observed in hepatoma cells. Furthermore, in contrast to the reported increase in LXR␣ mRNA in response to WY 14,643 and other fatty acid-derived PPAR␣ ligands in human macrophages (13) and liver cells (42) in vitro, we could detect no change in LXR␣ mRNA expression in response to dietary WY 14,643 in the intestine. Thus, if LXR␣ is involved in transducing the PPAR␣-mediated signal in the intestine, it must result from activation of existing LXR␣. This conclusion is supported by the observation that cholesterol was unable to potentiate further the increase in expression of the ABCA1 or SREBP-1c genes after PPAR␣ activation by WY 14,643 in the intestines of the normal animals. Such a nonadditive response is characteristic of a mechanism that is saturated, consistent with the presence of limiting amounts of LXR␣ in the intestine.
Although the evidence suggests that cholesterol and fibrate ultimately influence ABCA1 expression through activation of LXR␣, the activators need not be the same. In- deed, the different origins of the responses would imply that they are not. The increase in ABCA1 mRNA content on cholesterol feeding is likely to be an adaptive response mediated by oxysterols derived from cholesterol in the diet. It occurs in PPAR␣-deficient animals, and any increase in the return of cholesterol to the lumen does not prevent a net increase in cholesterol absorption. WY 14,643, on the other hand, is likely to be a causative agent, acting through a PPAR␣-dependent pathway to increase ABCA1 expression and so affect absorption. The nature of the putative LXR␣ ligands produced by PPAR␣ activation is not known, although the evidence suggests that they could be particularly potent. Generally, the origins and mechanisms of formation of physiological oxysterols that could act as ligands for LXR␣ remain obscure, but members of the cytochrome P-450 superfamily are involved in the natural production of such compounds (43)(44)(45), and some, such as members of the CYP4A group, are known to be PPAR␣ target genes (14).
There was little, if any, difference in the basal levels of intestinal ABCA1 mRNA expression between the wild-type and PPAR␣-null mice consuming the low-cholesterol diet alone. There was also no difference in intestinal cholesterol absorption. Therefore, differential cholesterol absorption cannot explain the net increase in diet-derived cholesterol and cholesteryl ester observed in the plasma and liver of the PPAR␣-null mice compared with the normal mice fed the low-cholesterol diet (Tables 3 and 4). Al-though PPAR␣ deficiency had little direct effect in intestine, it is known to lead to a number of changes in other tissues and in whole-body cholesterol metabolism that could explain the accumulation of cholesterol in these animals. These include a significantly decreased expression of ABCA1 mRNA in liver (data not shown), a redistribution of hepatic cholesterol between the esterified and nonesterified forms (22), and an impaired regulation of de novo cholesterol synthesis (22,46).
In the present work, there was a decrease in plasma triacylglycerol concentration on feeding the wild-type mice with fibrate (from 646 Ϯ 93 mol/l to 324 Ϯ 37 mol/l). It is known that PPAR␣ reduces CYP7A1 transcription in vitro (47) and that cipofibrate decreases bile acid synthesis in rats, leading to a decline in fecal bile acid output (48). Thus, although it is probable that the decrease in triacylglycerol content mainly reflected an increase in hepatic fatty acid oxidation, it could, at least in part, have resulted from a decrease in absorption. On the basis of these results, a reduced intestinal bile acid content cannot be completely excluded as a factor contributing to the observed decrease in cholesterol absorption.
LXR␣ activation has no effect on intestinal ACAT and microsomal triglyceride transfer protein expression (49), suggesting that there was little change in the enzymes and other proteins involved in chylomicron assembly. Expression of ABCG5/G8 can also be regulated by LXR␣. (50), although, because the exact specificity of this plant sterol transporter is not known, it is not possible to predict the contribution that it could make to the effects presented here. Likewise, we cannot rule out a role for the unknown targets of the cholesterol absorption inhibitor ezetimibe, which has recently been shown not to act through a stimulation of ABCA1 or ABCG5/G8 (51).
In conclusion, these results provide strong evidence for a physiological role for ABCA1 in regulating the intestinal absorption of dietary cholesterol. This "gatekeeper" role is enhanced by PPAR␣ activation, which results in a decreased absorption of dietary cholesterol, a decreased delivery of cholesterol to the liver, and a lower plasma concentration of total and diet-derived cholesterol. The evidence indicates that the effects of PPAR␣ activation on ABCA1 transcription are mediated by changes in the availability of ligands that activate LXR␣, but not by changes in LXR␣ concentration per se. Thus, it is possible that an increase in intestinal PPAR␣ activation by physiological ligands such as dietary unsaturated fatty acids, particularly those of the -3 class (52), could contribute to the decreased intestinal uptake of cholesterol from diets rich in fish oil or oleic acid observed in nonhuman primates (53), and that this is mediated via an increased expression of ABCA1. The recent report that a specific activator of PPAR␦ was a potent stimulator of ABCA1 transcription in macrophages (54) indicates that ligands for PPAR␦ could also play a role in regulating cholesterol absorption. Prevention of dietary cholesterol absorption makes a significant contribution to strategies designed to reduce plasma cholesterol levels (55), so pharmacological regulation of ABCA1 activity could be important not only for reducing