Regulation of ATP-binding Cassette Sterol Transporters ABCG5 and ABCG8 by the Liver X Receptors α and β

Mutations in the ATP-binding cassette (ABC) transporters ABCG5 and ABCG8 have recently been shown to cause the autosomal recessive disorder sitosterolemia. Here we demonstrate that the ABCG5 and ABCG8 genes are direct targets of the oxysterol receptors liver X receptor (LXR) α and LXRβ. Diets containing high cholesterol markedly increased the expression of ABCG5/G8 mRNA in mouse liver and intestine. This increase was also observed using synthetic ligands of LXR and its heterodimeric partner, the retinoid X receptor. In situ hybridization analyses of tissues from LXR agonist-treated mice revealed that ABCG5/G8 mRNA is located in hepatocytes and enterocytes and is increased upon LXR activation. In addition, expression of the LXR target gene ABCA1, previously implicated in the control of cholesterol absorption, was also dramatically up-regulated in jejunal enterocytes upon exposure to LXR agonists. These changes in ABC transporter gene expression were not observed in mice lacking LXRs. Furthermore, in the rat hepatoma cell line FTO2B, LXR-dependent transcription of the ABCG5/G8 genes was cycloheximide-resistant, indicating that these genes are directly regulated by LXRs. The addition of ABCG5 and ABCG8 to the growing list of LXR target genes further supports the notion that LXRs serve as sterol sensors to coordinately regulate sterol catabolism, storage, efflux, and elimination.

Cholesterol homeostasis is maintained by a series of regulatory pathways to control the synthesis of endogenous cholesterol, the absorption of dietary sterol, and the elimination of cholesterol and its catabolic end products, bile acids. Transcriptional control of many genes vital to these processes can be attributed to two families of transcription factors: the sterolregulatory element-binding proteins (SREBPs), 1 especially SREBP-2, which control the production of key enzymes in cholesterol biosynthesis (for review, see Ref. 1), and the liver X receptors LXR␣ and LXR␤, which regulate the expression of genes involved in cholesterol efflux, storage, catabolism, and elimination (for review, see Ref. 2). LXRs are ligand-activated transcription factors that are members of the nuclear hormone receptor superfamily. LXRs are bound and activated by a specific class of naturally occurring oxysterols (3)(4)(5) as well as a recently described nonsteroidal synthetic agonist, T0901317 (6,7). LXRs bind DNA as obligate heterodimers with the retinoid X receptors (RXRs) and can be activated by either LXR agonists or RXR ligands. The RXR/LXR heterodimer binds to a DNA sequence comprised of two direct repeats of the hexanucleotide motif AGGTCA separated by four bases, referred to as an LXR response element of the DR4 type (8). Upon binding ligand, LXR undergoes a conformational change that recruits coactivator proteins and enhances transcription of the target gene.
Increasing evidence suggests that the RXR/LXR heterodimer serves as a sensor that responds to high intracellular sterol concentrations by increasing the expression of genes that reduce the cellular sterol load. In rodents, high cholesterol feeding results in the production of oxysterols in the liver (9) that activate LXRs thereby increasing expression of cholesterol 7␣hydroxylase and promoting cholesterol conversion to bile acids for excretion (10). Mice lacking LXR proteins accumulate large quantities of esterified cholesterol in their livers following the consumption of dietary cholesterol (10,11). 2 LXRs also play a major role in orchestrating cholesterol efflux from cholesterolloaded macrophages by up-regulating two ATP-binding cassette transporters, ABCA1 and ABCG1 (6,(12)(13)(14)(15), and apolipoprotein E, which serves as an acceptor of effluxed cholesterol in the formation of high density lipoprotein particles (16). LXR activation also promotes the formation of cholesterol esters by increasing expression of SREBP-1c and stearoyl-CoA desaturase, resulting in increased production of oleic acid, the preferred substrate for cholesterol esterification (7,17,18).
Administration of synthetic LXR or RXR agonists to mice has multiple effects on cholesterol metabolism (7,17,19). Serum high density lipoprotein cholesterol and phospholipid levels are increased, and the biosynthesis and secretion of both bile acids and cholesterol are enhanced in treated mice (6,7). 3 Cholesterol absorption is reduced in LXR agonist-treated mice, and this reduction is associated with a dramatic increase in ABCA1 expression in the intestine (6). Moreover, cholesterol absorp-tion was recently reported to be increased in abca1-knockout mice, suggesting that this ATP-binding cassette transporter may serve to efflux cholesterol back into the intestinal lumen for elimination (6,20).
Two other members of the ABC transporter family, ABCG5 and ABCG8, have also recently been implicated in mediating the secretion of sterols from the liver and efflux of dietary sterols from the gut. Mutations in either of the genes encoding these two proteins cause the autosomal recessive disorder sitosterolemia (21,22). The diagnostic finding in this disorder is markedly elevated serum levels of the plant sterol ␤-sitosterol, which is normally present in only trace amounts (23). Individuals with sitosterolemia exhibit hyperabsorption of ␤-sitosterol, as well as other sterols, and have markedly reduced secretion of sterols into the bile. The expression of ABCG5 and ABCG8 are both increased in cholesterol-fed mice, suggesting that ABCG5 and ABCG8, as well as ABCA1, are regulated by the RXR/LXR heterodimer (21).
In this report, we provide unequivocal evidence that ABCG5 and ABCG8 are direct target genes of LXRs in liver and intestine. Furthermore, using in situ hybridization, we show that the levels of ABCG5 and ABCG8, as well as ABCA1, are dramatically increased in hepatocytes and enterocytes treated with an LXR agonist. These data provide strong evidence that the maintenance of dietary sterol homeostasis is coordinately regulated by the LXR-dependent expression of a family of ABC sterol transporters in the liver and intestine.

Materials
Nuclear receptor agonists were obtained from the following sources. Fenofibrate, pregnenolone ␣-carbonitrile, and chenodeoxycholic acid were purchased from Sigma; troglitazone was a generous gift from Roger Unger (University of Texas Southwestern Medical Center at Dallas); LG268 and GW4064 were provided by Richard Heyman and Raju Mohan (X-Ceptor Therapeutics, Inc., San Diego, CA); and T0901317 was supplied by Bei Shan (Tularik, South San Francisco, CA). Sodium mevalonate, sodium compactin, and delipidated fetal calf serum were prepared as described previously (24,25).

Animal Studies
Lxr-knockout mice (6, 10) 2 were maintained on a mixed-strain background (C57BL/6:129Sv) and housed in a temperature-controlled environment with 12-h light/dark cycles. Mice were fed ad libitum a cerealbased mouse/rat diet (No. 7001, Harlan Teklad, Madison WI) that contained 0.02% cholesterol and 4% total lipid. Diets were supplemented with cholesterol (ICN Biomedicals), vehicle (0.9% carboxymethylcellulose, 9.95% polyethylene glycol, 0.05% Tween 80) added at 3.3 ml/100 g of diet, and/or nuclear receptor agonist. Nuclear receptor agonists were added to diets to provide the appropriate dose (mg/kg of body weight, mpk) on consumption of 5 g of diet by a 25-g mouse per day. There were no changes in food consumption observed by animals fed the various synthetic agonists. Following dietary treatment, mice were anesthetized with Nembutal and exsanguinated prior to organ harvest. The small intestine was removed, flushed with ice-cold saline, and opened lengthwise, and the mucosae were gently scraped and transferred to tubes for flash freezing in liquid nitrogen. In some studies the small intestine was divided into three segments of equal length designated duodenum (proximal), jejunum (medial), and ileum (distal). All experiments were approved by the Institutional Animal Care and Research Advisory Committee.

Northern Analysis
Total RNA was extracted from tissues of individual mice or from cultured cells using RNA STAT-60 (Tel-Test, Inc.) according to the manufacturer's instructions. Equal amounts of total RNA from the four to six animals per treatment were pooled, and mRNA was isolated using oligo-dT cellulose columns (Pharmacia). 5 g of poly(A ϩ ) RNA or 10 g of total RNA were size-fractionated on 1% agarose, formaldehyde gels and transferred to nylon membranes (Zetaprobe, BioRad). Hybridizations were performed using 32 P-labeled cDNA probes for mouse or rat ABCG5 and ABCG8 (21), mouse SREBP-1 (26), and ␤-actin or glycer-aldehyde-3-phosphate dehydrogenase. The amount of radioactivity in each band was quantified by phosphorimaging and normalized to the signal of ␤-actin or glyceraldehyde-3-phosphate dehydrogenase.

Real-time PCR Analysis
Total RNA was prepared from FTO2B hepatoma cells using RNA STAT-60 (Tel-Test, Inc.), treated with DNase I (RNase-free, Roche Molecular Biochemicals), and reverse-transcribed with random hexamers using the SuperScript II First-Strand Synthesis System (Invitrogen) to generate cDNA. Primers for each gene were designed using Primer Express Software (PerkinElmer Life Sciences) based on sequence data available through GenBank TM (accession numbers in parentheses): rat glyceraldehyde-3-phosphate dehydrogenase (AF106860): forward 5Ј-GAGGTGACCGCATCTTCTTG, reverse 5Ј-CC-GACCTTCACCATCTTGTC; rat ABCG5 (AF312714): forward 5Ј-CCC-TGTCCTGAACATTCCAA, reverse 5Ј-TTGGGTGTCCACCGATGTCA; and rat ABCG8 (AF351785): forward 5Ј-CGGCTGGCTTCATGATAA-AC, reverse 5Ј-GGAACGACATCTTGGAAATCC. The real-time PCR contained, in a final volume of 10 l, 25 ng of reverse-transcribed total RNA, a 150 nM concentration of each primer, and 5 l of 2ϫ SYBR Green PCR Master Mix (Applied Biosystems). All PCRs were performed in triplicate on an Applied Biosystems Prism 7900HT Sequence Detection System, and relative mRNA levels were calculated using the comparative C T method. 4 Primers were validated by analysis of a standard curve and dissociation curve for each primer pair in a template titration assay. In addition, real-time PCR results were confirmed by comparison to a Northern analysis for a single experiment.

In Situ Hybridization
Probes-cDNA fragments encompassing the 3Ј-untranslated regions of mouse ABCG5 and ABCG8 were amplified by PCR and subcloned into pGEM in both orientations to allow the synthesis of RNA riboprobes using the T7 promoter. PCR primers for mouse ABCG5 were 5Ј-ATCCAACACCTCTATGCTAAATCAC (forward) and 5Ј-TACATTAT-TGGACCAGTTCAGTCAC (reverse) to generate a 411-bp fragment. Primers for mouse ABCG8 were 5Ј-ACAAGGCTCACACAGATCTCTCA (forward) and 5Ј-TATAATTGGTTCCCATTCCATACTG (reverse) to produce a 212-bp cDNA. A 413-bp cDNA fragment of mouse ABCA1 containing nucleotides 390 -802 (GenBank TM accession number NM_013454) and flanked by BamHI and XhoI sites was generated by PCR using the primers 5Ј-CGGGATCCTCTCGCCTGTTCTCAGACGC (forward) and 5Ј-CCGCTCGAGACCACTGGCTTCAGGATGTCC (reverse). Ligating this fragment into pBS(KSϪ) allowed production of both antisense (T3/BamHI) and sense (T7/XhoI) in situ probes. RNA riboprobes were generated by in vitro transcription with the Ambion Maxiscript kit (Ambion, Austin, TX) in the presence of 35 S-UTP (Amersham Biosciences, Ͼ1000 Ci/mmol). Probes were purified over G50 spin columns and evaluated by diagnostic polyacrylamide gel electrophoresis.
Tissues-Tissues were harvested from A129 strain male mice following a 16-h dietary treatment (see "Animal Studies" for diet information). Organs were removed from anesthetized mice after transcardial perfusion with cold heparinized diethyl pyrocarbonate-saline followed by chilled 4% paraformaldehyde, diethyl pyrocarbonate-phosphatebuffered saline. Tissues were fixed for 16 h in paraformaldehyde, transferred to diethyl pyrocarbonate-saline, then dehydrated, paraffin-embedded, and sectioned.
In situ hybridization was performed as described previously (27). Optimal autoradiographic exposure of the slides was achieved at 14 days for ABCG5 and ABCG8 probes and 21 days for the ABCA1 antisense riboprobe. Control slides using sense riboprobes were evaluated at those times and gave no significant signals.

Cell Culture
The rat hepatoma cell line FTO2B (28) was maintained in Dulbecco's modified Eagle's medium/F12 containing glutamine and 5% fetal calf serum at 37°C in an 8% CO 2 incubator. Cells were plated at 700,000/ 10-cm dish and allowed to adhere overnight. The following morning, the cells were washed, and the sterol depletion medium was applied as described by DeBose-Boyd et al. (

Statistical Analysis
GraphPad Prism computer software was used to perform all statistical analyses (GraphPad, San Diego, CA). Experimental results were tested by a two-way analysis of variance with genotype and diet as factors. In all cases, a significant interaction of these factors was observed and a post-hoc Newman-Keuls test was performed to evaluate differences between the means of the various treatment groups. Statistical significance was declared it the calculated p value was less than 0.05. All values are represented as means Ϯ S.E., and different letters above the graphed means indicate significantly different values.

RESULTS
Dietary cholesterol feeding was previously shown to increase duodenal, jejunal, and hepatic expression levels of ABCG5 and ABCG8 mRNA in wild-type mice (21). To determine the role of LXRs in the up-regulation of these genes, we challenged mice with various diets under conditions in which either LXR␣, LXR␤, or both LXR genes had been ablated. ABCG5 and ABCG8 exhibited a statistically significant ϳ2.5-4-fold increase in both transcripts in the liver of wild-type mice after 7-10 days of feeding a 2% cholesterol diet (Fig. 1, A and B). No increase in ABCG5 or ABCG8 was seen in mice lacking either LXR␣ or LXR␣ plus LXR␤. In fact, in the lxr␣/␤-knockout mice the expression levels were significantly reduced by ϳ50% upon cholesterol feeding (Figs. 1 and 2). Interestingly sterol regula-tion was maintained in the lxr␤-knockout mice demonstrating that sterol up-regulation of ABCG5/G8 requires LXR␣. The apparent LXR␣-specific response may be due to the greater abundance of LXR␣ than LXR␤ in the liver (31). Alternatively LXR␣ may exhibit a higher activity than LXR␤ in activating these genes (10). Similar trends in hepatic gene expression have been observed for other LXR target genes such as cholesterol 7␣-hydroxylase, which is dependent on LXR␣, but not LXR␤, expression (10).
An LXR-dependent increase in ABCG5 and ABCG8 mRNA expression was also observed in the duodenum, jejunum, and ileum of cholesterol-fed wild-type mice (Fig. 2). Interestingly the expression levels of ABCG5 and ABCG8 were higher in tissues of lxr␣/␤Ϫ/Ϫ mice than wild-type mice fed control diets ( Figs. 1 and 2). This increase in expression in lxr␣/␤-knockout mice has also been observed for other ABC transporters identified as LXR target genes such as ABCA1 and ABCG1 (6,15).
The dietary cholesterol-induced increase in ABCG5 and ABCG8 expression observed only in mice harboring LXR suggested these genes were primary targets of the RXR/LXR heterodimer. To determine the specificity of this effect, wild-type mice were treated for 12 h with a series of synthetic agonists for a variety of nuclear hormone receptors (Fig. 3). This treatment protocol was previously shown to elicit increases in RNA levels of target genes for each of these nuclear receptors (6). Both the RXR-specific agonist LG268 (32) and the LXR-specific agonist T0901317 (7) caused coordinate up-regulation of ABCG5 and ABCG8 mRNA expression in liver and intestine (Figs. 3 and 4). A reproducible increase in expression of these genes was also 3-month-old male mixed-strain mice of various LXR genotypes were fed ad libitum diets containing 0.02% cholesterol (Teklad 7001 powdered rodent chow) or that diet supplemented with 2% cholesterol. A, following a 7-day dietary treatment, hepatic mRNA was pooled from six mice per group, and 5 g of RNA/lane was analyzed by Northern analysis using 32 P-labeled cDNA probes. B, hepatic mRNA was prepared from individual mice (n ϭ 8 per group) after a 10-day feeding study and analyzed by Northern analysis. RNA levels were quantified by phosphorimaging and standardized against ␤-actin (results for the lower bands of ABCG5 and ABCG8 are shown; upper bands gave very similar results). The relative mRNA levels (bar height) reflect expression compared with wild-type mice on 0.02% cholesterol diet, and the numeric values above the bars in A reveal the relative change in hepatic RNA level due to cholesterol feeding for a given genotype. In B, the relative mRNA levels are represented as means Ϯ S.E., and bars with different letters are significantly different (p value Ͻ0.05). WT, wild type; Chol, cholesterol.

FIG. 2. ABCG5 and ABCG8 mRNA expression in liver and small intestine of wild-type and lxr␣/␤؊/؊ mice fed dietary cholesterol.
3-month-old male mixed-strain (A129/C57Bl6) mice were fed diets as described in the legend for Fig. 1 for 10 days. RNA was extracted from liver and intestinal mucosa of the duodenum (D), jejunum (J), and ileum (I). Poly(A ϩ ) RNA was prepared from pooled RNA representing equal amounts from the eight mice per group, and 5 g mRNA/lane was analyzed by Northern analysis. RNA was quantified by phosphorimaging and standardized against cyclophilin to reveal the relative mRNA levels (results for the smaller transcripts of ABCG5 and ABCG8 are shown; larger transcripts gave very similar relative expression changes). White bars, wild-type mice; black bars, lxr␣/␤Ϫ/Ϫ mice; hatched bars, high cholesterol diet. WT, wild type; Chol, cholesterol. seen in the liver, but not the intestine, of mice treated with the bile acids chenodeoxycholic acid (Fig. 3) and cholic acid (see Fig.  6C below), which are known to be ligands for the nuclear bile acid receptor FXR. In these experiments all mice received diets supplemented with 0.2% cholesterol. The simultaneous feeding of cholesterol and bile acids results in increased hepatic sterol levels (33), raising the possibility that the up-regulation of ABCG5 and ABCG8 by bile acids under these conditions is mediated by LXR rather than being a direct effect of FXR. This mechanism, although possible, seems unlikely since other LXR target genes in the liver were not increased in mice fed bile acids under identical conditions (6,17). As discussed below, regulation of ABCG5 and ABCG8 by FXR agonists appears to be an indirect effect. There was also a decrease in hepatic expression of ABCG5/G8 when mice were treated with fenofibrate, a strong PPAR␣ agonist. The basis of this liver-specific repression by PPAR␣ is at present unknown.
Oral administration of the synthetic LXR agonist for 12 h (Fig. 4A) or 10 days (Fig. 4B) resulted in increased expression of ABCG5 and ABCG8 mRNA in liver and intestine from wildtype but not from lxr␣/␤-knockout mice. Under these conditions, the increase in intestinal mRNA expression was modest, yet reproducible (five independent experiments), suggesting that further experimentation will be necessary to statistically establish LXR-mediated regulation in this tissue. Treatment for 12 h or 10 days (Fig. 4, A and B) with the RXR ligand LG268 was also associated with an increased level of expression of ABCG5/G8 in these tissues. No increase in ABCG5/G8 mRNA was seen in the intestine of lxr-double knockout mice, consistent with the effects of the rexinoid being mediated by the RXR/LXR heterodimer. In contrast, rexinoids continued to stimulate ABCG5/G8 expression in the liver of lxr-double knockout mice. This response was detectable within 12 h (Fig.  4A) and was still maintained after 10 days (Fig 4B). These results suggest that an additional RXR heterodimer partner, such as FXR, may also regulate expression of these genes in the liver. We note that FXR ligands can also stimulate expression of ABCG5 and ABCG8 in the liver (see Figs. 3 and 6).
The cellular distribution of ABCG5 and ABCG8 mRNA in the liver and intestine and their regulation by LXR were investigated further by in situ hybridization analyses. Adult male mice were fed diets containing vehicle or 50 mpk T0901317 for 16 h, and tissues were obtained and processed (see "Experimental Procedures"). No evidence of cellular hypertrophy or hyperplasia due to agonist exposure was seen on routine histology (data not shown). The antisense riboprobes gave a substantially higher signal than background (Fig. 5), and no signal was seen with sense control probes (not shown). In the liver, ABCG5 and ABCG8 mRNAs were localized to hepatocytes with a uniform distribution across the hepatic lobule and no apparent enrichment in endothelium lining the portal vein or central vein or in the biliary epithelium. The intensity of the signals for both ABCG5 and ABCG8 was increased in the agonist-treated mouse livers, although the signal was also clearly evident in the vehicle-treated control livers. In the jejunal sections, ABCG5 and ABCG8 mRNAs were detected exclusively in enterocytes lining the villi. No signal was evident in cells of the lamina propria or tunica muscularis. LXR agonist treatment caused a substantial increase in the expression levels of ABCG5 and ABCG8, but this expression remained confined to the villus enterocytes with a modest gradient suggesting slightly lower levels in the crypt epithelium. In comparison, examination of ABCA1 expression and distribution revealed that this transporter is present at very low levels in the vehicletreated jejunum and is found predominately in cells present in the lamina propria and occasionally in enterocytes as reported previously for the primate (34). However, upon treatment with T0901317, ABCA1 expression was dramatically increased in enterocytes lining the villi. Histologic examination revealed no pathologic changes in villus length or tissue architecture associated with treatment. Taken together, these data provide strong evidence that ABCA1, -G5, and -G8 are expressed in absorptive enterocytes, further supporting a role for all three ABC transporters in regulating cholesterol flux in the intestine.
To determine whether RXR/LXR-mediated transactivation of the ABCG5/G8 genes was a direct transcriptional response or whether it involved synthesis of another transcription factor or protein, we examined LXR agonist-induced expression in the presence of the protein synthesis inhibitor cycloheximide. The inability of cycloheximide to inhibit ABCG5/G8 transcription by LXR agonists would indicate a direct mechanism of LXR action as was demonstrated previously for the LXR target gene ABCG1 in macrophage (15). Testing an array of liver cell lines revealed that two rat hepatoma cells lines (FTO2B and FAO) exhibited an increase in ABCG5 and ABCG8 mRNA levels following administration of LXR agonists such as 22(R)-hydroxycholesterol, 24(S),25-epoxycholesterol, or the synthetic ligand T0901317 (Fig. 6). Treatment of FTO2B cells with T0901317 in the presence of cycloheximide demonstrated that ABCG5 and ABCG8 mRNA levels, like those of the target gene SREBP-1c, were still increased and firmly established that these genes are directly regulated by RXR/LXR (Fig. 6A). Consistent with a direct transcriptional response, induction of ABCG5 and ABCG8 was rapid with an increase in mRNA levels apparent as early as 3 h after adding T0901317 to cells (not shown). Cycloheximide treatment also appeared to stabilize the mRNA of these genes relative to all internal controls tested (␤-actin, glyceraldehyde-3-phosphate dehydrogenase, and cyclophilin) as transcript levels are clearly increased at 18 h of exposure to cycloheximide even in the absence of all transcription-inducing agents (Fig. 6A).
The results shown in Fig. 3 suggested that FXR may also play a role in regulating ABCG5/G8 expression in liver. To further address this possibility, wild-type and cyp7a1Ϫ/Ϫ mice were fed a basal diet with or without added cholic acid for 10 days (Fig. 6B). Similar to other FXR-regulated genes such as shp (35), there was a tendency toward reduced expression of ABCG5/G8 in the cyp7a1Ϫ/Ϫ mice due to diminished bile acid ligand availability. Consistent with this notion, dietary administration of cholic acid to restore the bile acid pool in cyp7a1Ϫ/Ϫ mice resulted in a significant increase in hepatic ABCG5 and ABCG8 mRNA levels. In addition, bile acid feeding of wild-type mice resulted in increased hepatic ABCG5/G8 mRNA levels that were not observed in fxr-null mice. 5 To explore whether this bile acid-mediated change in ABCG5/G8 expression was due to direct transcriptional activation by FXR, cycloheximide experiments were performed in the FTO2B cell line using the synthetic FXR agonist GW4064 (36) or chenodeoxycholic acid (Fig. 6C). Both FXR ligands induced an increase in ABCG5 and ABCG8 mRNA levels in the absence of cycloheximide as expected, similar to that observed with LXR agonists. However, in contrast to the LXR agonists, cycloheximide treatment of FXR agonist-dosed cells eliminated this increase in ABCG5/G8 mRNA. These results demonstrate that FXR-mediated increases in ABCG5/G8 are indirect and that these agents require the expression of other protein(s) to transcriptionally regulate these genes. DISCUSSION The studies detailed in this report demonstrate that ABCG5 and ABCG8 are target genes of the RXR/LXR heterodimer. Under conditions of cellular sterol loading or LXR/RXR agonist administration, liver cells exhibit an LXR-dependent increase in the transcription of ABCG5/G8 mRNA. Furthermore we show that ABCG5/G8 genes are direct targets of LXR action since transcriptional up-regulation of these genes does not require new protein synthesis. The ABCG5 and -G8 genes sit adjacent to one another, and both genes are transcribed in opposite directions from independent transcription start sites that are separated by only ϳ150 bp of intergenic sequence that does not contain an LXR response element (21). Similar to other paired genes, it is likely that the regulatory elements for these two genes are in adjacent introns or are located far from the coding regions of the genes (37)(38)(39). Alternatively it is possible that LXR may be affecting transcription in a novel manner through protein-protein interactions with another transcription factor. Future studies will be required to characterize these regulatory regions and possibly identify an LXR response element.
Our studies show that other nuclear receptors are also capable of regulating liver-specific expression of ABCG5/G8. Of notable interest is the apparent up-regulation by the bile acid receptor FXR. However, in contrast to the LXR response, the transcriptional response mediated by FXR is indirect, indicating a secondary mechanism that requires synthesis of other regulatory factors. This bile acid-mediated response was observed only in the liver and not in the intestine, although FXR is present in both tissues. This suggests that the factor induced by FXR is produced or functional only in hepatocytes.
The inclusion of ABCG5 and ABCG8 to the growing list of LXR target genes provides further evidence that the RXR/LXR heterodimer is an intracellular sterol sensor that up-regulates the expression of genes involved in cholesterol elimination and reverse cholesterol transport. In peripheral cells, particularly macrophages, LXR-mediated up-regulation of ABCA1, ABCG1, and apolipoprotein E promotes the efflux of free cholesterol and phospholipids to produce high density lipoprotein particles. High density lipoprotein cholesterol is then delivered to the liver where an LXR-dependent increase in cholesterol 7␣-hydroxylase expression results in enhanced bile acid synthesis and elimination. Increased biliary excretion and decreased in-  5. Localization of ABCG5, ABCG8, and ABCA1 mRNA in the mouse liver and small intestine by in situ hybridization. Male A129 strain mice were fed for 16 h diets containing vehicle or the LXR agonist T0901317 (50 mpk). The expression patterns of ABCG5, ABCG8, and ABCA1 mRNA were visualized by dark field microscopy following hybridization with antisense 35 S-labeled riboprobes. The right panels display the hematoxylin staining, revealed by bright field microscopy, of the LXR agonist-treated samples shown in adjacent panels. All tissues were negative with the corresponding sense riboprobes (data not shown). LP, lamina propria. Solid arrows indicate red blood cells that refract light by dark field illumination but do not show silver grains indicative of specific riboprobe hybridization. Magnification of view is ϫ10 for jejunum and ϫ20 for liver. testinal uptake through the LXR-mediated regulation of ABC sterol transporters allow for the ultimate elimination of excess body sterol.
The increase in dietary phytosterol and cholesterol absorption observed in sitosterolemics strongly supports a role for ABCG5 and ABCG8 in net sterol uptake by the intestine (21,40). The most likely mechanism for mediating this effect would be that ABCG5 and ABCG8 participate in the movement of cholesterol from the enterocyte back into the intestinal lumen, resulting in a decrease in net sterol absorption. In situ hybridization revealed that these genes are expressed in enterocytes lining the villi of intestinal segments involved in cholesterol absorption (Fig. 5). We previously showed that ABCA1 expression was increased in the duodenum and jejunum of mice treated with an LXR agonist and suggested that this up-regulation may be responsible for the associated decrease in cholesterol absorption (6). In this report we show that ABCA1 is also abundantly expressed in the absorptive cells of LXR agonisttreated animals.
Although the relative roles of ABCA1 and ABCG5/G8 in cholesterol efflux from the gut have yet to be clearly defined, our work provides unequivocal data that demonstrate ABCA1, -G5, and -G8 are expressed in enterocytes and are strongly up-regulated by LXR agonists. Under basal conditions (i.e. in the absence of dietary cholesterol or RXR/LXR agonists) ABCA1 expression is relatively weak in the absorptive cells of mice (Fig. 5) and primates (34). An initial report showing increased cholesterol absorption efficiency by abca1Ϫ/Ϫ mice (20) has recently been contradicted by analysis of a second abca1-knockout mouse strain fed a low cholesterol diet (41), leaving the question open on the role of ABCA1 in regulating cholesterol absorption in chow-fed animals. However, the strong up-regulation of ABCA1 in enterocytes, accompanied by a decrease in the efficiency of cholesterol absorption in the cholesterol-fed and/or RXR/LXR agonist-treated mice, suggests that ABCA1 may play a role in cholesterol absorption under high cholesterol or pharmacologic conditions (Fig. 5 and Ref. 6). Further studies will be required in abcg5/8-knockout mice and abca1-knockout mice treated with LXR agonists to firmly establish the relative roles of these three ABC transporters in cholesterol absorption.
The recent identification of genes in which mutations result in human genetic disorders involving cholesterol homeostasis has opened the exciting opportunity to study the mechanisms of these gene products and their modes of regulation in normal, pathologic, and pharmacologic states. Concurrent up-regulation of ABCG5 and ABCG8 promotes biliary sterol secretion, and in the intestine the LXR-mediated up-regulation of ABCA1, ABCG5, and ABCG8 prohibits the absorption of lumenal cholesterol thereby resulting in net cholesterol loss from the body. These actions point to the therapeutic potential of LXR agonists in the treatment of atherosclerosis and other disorders of hypercholesterolemia.
FIG. 6. Differential regulation of ABCG5/G8 transcription by LXR and FXR agonists (35). A, FTO2B rat hepatoma cells were exposed to vehicle or the LXR agonist T0901317 (T1317, 10 M) for 18 h in the presence or absence of cycloheximide (CHX, 500 M). 10 g of total RNA/lane was used for Northern blot assay. B, mixed-strain wild-type or cyp7a1Ϫ/Ϫ mice (35,42) were fed diets supplemented with 0.2% cholesterol and cholic acid (at 0 or 0.1%) for 10 days. Hepatic mRNA was prepared from individual mice (n ϭ 4 per group) and analyzed by Northern analysis. ABCG5 and ABCG8 mRNA levels are expressed as means Ϯ S.E. Bars with different letters are significantly different (p value Ͻ0.05). C, FTO2B cells were treated with vehicle, chenodeoxycholic acid (CDCA, 100 M), synthetic FXR agonist (GW4064, 10 M), or LXR agonist (T1317, 10 M) in the absence or presence of cycloheximide (CHX, 500 M) for 12 h. Real-time PCR was performed as described under "Experimental Procedures" with the average (and S.D.) shown for results from two independent experiments, each analyzed in triplicate. Asterisks denote values significantly different from the vehicle group of the same cycloheximide treatment as determined by Student's t test. Veh, vehicle; WT, wild type; CA, cholic acid.