Direct and Indirect Mechanisms for Regulation of Fatty Acid Synthase Gene Expression by Liver X Receptors*

The nuclear receptors LXRα and LXRβ have been implicated in the control of lipogenesis and cholesterol homeostasis. Ligand activation of these receptors in vivoinduces expression of the LXR target gene SREBP-1cand increases plasma triglyceride levels. Expression of fatty acid synthase (FAS), a central enzyme in de novo lipogenesis and an established target of the SREBP-1 pathway, is also induced by LXR ligands. The effects of LXR ligands on FAS expression have been proposed to be entirely secondary to the induction of SREBP-1c. We demonstrate here that LXRs regulate FAS expression through direct interaction with the FAS promoter as well as through activation of SREBP-1c expression. Induction of FAS expression in HepG2 cells by LXR ligands is reduced, but not abolished, under conditions where SREBP processing is suppressed. Moreover, LXR ligands induce FAS expression in CHO-7 cells without altering expression of SREBP-1. We demonstrate that in addition to tandem SREBP sites, the FASpromoter contains a high affinity binding site for the LXR/RXR heterodimer that is conserved in diverse animal species including birds, rodents, and humans. The LXR and SREBP binding sites independently confer LXR responsiveness on the FASpromoter, and maximal induction requires both transcription factors. Transient elevation of plasma triglyceride levels in mice treated with a synthetic LXR agonist correlates with transient induction of hepatic FAS expression. These results indicate that the LXR signaling pathway modulates FAS expression through distinct but complementary mechanisms and suggest that the FAS gene may be a critical target in the control of lipogenesis by LXRs.

A central enzyme in the pathway of de novo lipogenesis, fatty acid synthase (FAS) 1 catalyzes all of the steps in the conversion of malonyl-CoA to palmitate. Expression of the FAS gene is controlled primarily at the level of transcription and is responsive to both hormonal and nutritional signals (1,2). Previous work has shown that sterol regulatory element-binding proteins (SREBPs) play a critical role in the transcriptional regulation of a number of genes in the lipogenic pathway, including FAS, steroyl-CoA desaturase (SCD-1), and acetyl-CoA carboxylase (ACC) (3)(4)(5)(6)(7)(8). Three SREBP isoforms have been described: SREBP-1a and Ϫ1c (also called ADD1), which are derived from the same gene through alternative splicing, and SREBP-2, which is encoded by a separate gene (9,10). Although their transcriptional targets overlap significantly, studies suggest that SREBP-1 preferentially activates genes involved in lipogenesis, whereas SREBP-2 preferentially activates genes in the cholesterol biosynthetic pathway (11)(12)(13)(14). SREBPs have been shown to regulate FAS expression through direct interaction with the FAS promoter at multiple sites (7,15). Overexpression of nuclear SREBP-1 is sufficient to induce expression of the FAS gene in cultured cells as well as transgenic mice (5,8).
Recent work has also implicated the nuclear receptors LXR␣ and LXR␤ in the control of lipogenesis. Both LXRs bind to DNA and regulate transcription of target genes in a heterodimeric complex with RXR (16). Although early studies on LXRs focused on their role in cholesterol metabolism, mice carrying a targeted disruption in the LXR␣ gene were noted to be deficient in expression of FAS, SCD-1, ACC, and SREBP-1, consistent with a role in lipogenesis as well (17). Further support for this idea came with the observation that the administration of the synthetic LXR ligand T1317 to mice triggers induction of the lipogenic pathway and raises plasma triglyceride levels (18). The demonstration that the SREBP-1c promoter is a direct target for regulation by LXR/RXR heterodimers provided a straightforward explanation for the ability of LXR ligands to induce hepatic lipogenesis (19,20). Until now, the effects of LXR activation on the expression of lipogenic genes, including FAS, have been presumed to be entirely indirect.
We demonstrate here that the FAS promoter is a direct target for regulation by the LXR/RXR heterodimer as well as SREBPs. This novel mechanism for the regulation of FAS expression and lipogenesis by LXRs has implications for the development of LXR agonists as modulators of human lipid metabolism.
Cell Culture and Transfections-HepG2 cells were cultured in modified Eagle's medium containing 10% fetal bovine serum or lipoproteindeficient fetal bovine serum (LPDS). THP-1 cells were cultured in RPMI 1640 containing 10% fetal bovine serum and differentiated for 24 h with 40 ng/ml phorbol myristyl acetate (PMA). Transient transfections of HepG2 cells were performed in triplicate in 48-well plates. Cells were transfected with reporter plasmid (100 ng/well), receptor plasmids (5-50 ng/well), pCMV-␤-galactosidase (50 ng/well), and pTKCIII (to a total of 205 ng/well) using the MBS mammalian transfection kit (Stratagene). Following transfection, cells were incubated in modified Eagle's medium containing 10% LPDS and the indicated ligands or vehicle control for 24 h. Luciferase activity was normalized to ␤-galactosidase activity.
RNA Analysis-Total RNA was isolated using Trizol reagent (Invitrogen). S1 nuclease analysis was carried out as described (3). Real time quantitative PCR assays were performed using an Applied Biosystems 7700 sequence detector as in Ref. 23. Briefly, 1 g of total RNA was reverse transcribed with random hexamers using Taqman reverse transcription reagents kit (Applied Biosystems). Each amplification mixture (50 l) contained 50 ng of cDNA, 900 nM forward primer, 900 nM reverse primer, 100 nM dual-labeled fluorogenic probe (Applied Biosystems), and 25 l of Universal PCR master mix. All samples were analyzed for 36␤4 or 18 S rRNA expression in parallel in the same run. Quantitative expression values were extrapolated from separate standard curves for controls and unknowns generated with 10-fold dilutions of cDNA. All assays were performed in duplicate. Primer and probe sequences are available upon request.
Animals-C57BL/6 mice (5 animals/group) were maintained on standard mouse chow and dosed with 50 mg/kg of T0901317 or vehicle alone (0.5% methylcellulose) by oral gavage once a day for 3 or 7 days. Animals were sacrificed 4 h after the last treatment on days 3 and 7. Plasma triglyceride levels were determined on an Instrumentation Laboratories ILab600 Clinical Chemistry Analyzer.

RESULTS
Previous work has shown that expression of FAS and SREBP-1c in rodent liver is induced in response to the synthetic LXR agonist T1317 (18). Induction of FAS expression by LXR has been presumed to be secondary to the induction of SREBP-1c; however, this hypothesis has not been tested. We identified FAS as an LXR-responsive gene in macrophages using Affymetrix cDNA microarrays (data not shown). As shown in Fig. 1A, real time quantitative PCR (Taqman) analysis confirmed that FAS expression was induced in 12-O-tetradecanoylphorbol-13-acetate (TPA)-differentiated THP-1 macrophages in response to treatment with either of the nonsteroidal LXR agonists, T1317 (18) or GW3965 (23,25). The RXR agonist LG268 also induced FAS expression, and the combination of LG268 and T1317 had an additive effect. Consistent with previous results in liver (19,20), treatment of THP-1 macrophages with LXR ligands also led to a significant induction of SREBP-1 RNA (data not shown).
To investigate whether LXR activation induced FAS expression in other peripheral cells we examined CHO-7 cells. S1 nuclease analysis demonstrated that T1317 strongly induced expression of FAS (Fig. 1B). However, unlike THP-1 cells and HepG2 cells, this induction occurred in the absence of significant induction of either SREBP-1 or the SREBP target gene HMG-CoA reductase. This observation suggests that the ability of LXR to regulate FAS expression is not entirely dependent on the ability to induce SREBP-1 expression. We further explored the regulation of the FAS promoter in HepG2 cells under different cellular sterol conditions. As shown in Fig. 1C, basal FAS expression and induction by T1317 were highest in steroldepleted cells (cultured in LPDS, 5 M simvastatin, 100 M mevalonic acid) in which levels of nuclear SREBPs are expected to be high. However, a significant induction of FAS expression was also observed in the presence of 2.5 g/ml 25-hydroxycholesterol (25-HC) and 5 g/ml cholesterol, conditions under

FIG. 1. FAS expression is regulated by LXR in multiple cell types. A,
THP-1 cells were differentiated with TPA for 24 h prior to the addition of T1317 or GW3965 (5 M). RNA was isolated after 48 h of ligand treatment, and FAS transcript levels were quantitated by real time quantitative PCR (Taqman). B, CHO-7 cells were treated with T1317 (5 M) or 2 g/ml of the indicated oxysterols for 24 h. Transcript levels were measured by S1 nuclease protection analysis (FAS and HMG-CoA reductase) or Northern analysis (SREBP-1). C, HepG2 cells were cultured in media containing either 10% LPDS, LPDSϩ5 mM simvastatinϩ100 mM mevalonate, or LPDSϩ2.5 g/ml 25-HCϩ5 g/ml cholesterol and treated for 24 h with the indicated concentration of T1317. FAS transcript levels were quantitated by real time quantitative PCR. which SREBP cleavage is suppressed. Again, these observations suggest that LXR can regulate FAS expression, at least in part, by a mechanism that is independent of SREBP-1c.
The regulatory regions that mediate induction of the FAS gene by LXR ligands have not been defined. In an effort to map these sequences, we analyzed the ability of LXRs and synthetic LXR ligands to regulate the FAS promoter in transient transfection assays. HepG2 cells were cotransfected with a luciferase reporter containing sequences from Ϫ1594 to ϩ67 bp of the rat FAS promoter and cytomegalovirus promoter-driven expression vectors for LXR␣ and RXR␣ or LXR␤ and RXR␣. Neither the LXR/RXR expression vectors nor T1317 had any effect on the control pGL2 luciferase reporter plasmid lacking FAS promoter sequences (Fig. 2). In contrast, treatment with T1317 (1 M) led to a significant induction in FAS reporter activity, even in the absence of transfected receptors. This induction is presumably mediated by the endogenous LXR␣ and LXR␤ expressed by HepG2 cells. Promoter induction by T1317 was strongly enhanced by cotransfection of either LXR␣Ј2fRXR␣ or LXR␤Ј2fRXR␣, consistent with an LXR-mediated effect.
The above observations indicated that the region between Ϫ1594 and ϩ67 of the FAS promoter contains sequences that mediate induction by LXR. To further define the cis-acting elements involved in this regulation we analyzed a series of deletion constructs. As shown in Fig. 3B, constructs extending to Ϫ1594 or Ϫ700 bp were equivalently activated by LXR. Deletion of the sequence between Ϫ700 and Ϫ150 bp significantly reduced, but did not abolish, the response. Further deletion from Ϫ150 to Ϫ135 bp had no effect. Together, these observations indicate that sequences located between Ϫ700 and Ϫ150 bp as well as between Ϫ135 and ϩ1 mediate the response of the FAS promoter to LXR/RXR. The previously identified SRE present at Ϫ150 bp (7) is apparently not involved in promoter induction by LXR.
The transfection studies described above were performed in cells that were cultured with lipoprotein-depleted serum, where the low levels of sterols resulted in a significant accumulation of nuclear SREBP protein. To prevent nuclear accu-mulation of SREBP, cells were cultured in the presence of high levels of cholesterol and 25-HC, conditions known to suppress SREBP cleavage and reduce the concentration of nuclear SREBPs to undetectable levels (26). Under these conditions the vast majority of the LXR response is conferred by sequences between Ϫ700 and Ϫ150 bp (Fig. 3C). The Ϫ150 and Ϫ135 FAS constructs were not significantly induced by LXR or LXR ligand in the presence of cholesterol and 25-HC. These results are consistent with previous work localizing the sterol response sequences of the FAS promoter to two tandem SREBP binding sites between Ϫ71 and Ϫ54 bp ( Fig. 3A and Ref. 15). Thus, LXR can regulate FAS expression through an SREBP-dependent pathway involving sequences between Ϫ135 and ϩ67 bp as well as an SREBP-independent pathway involving sequences present between Ϫ700 and Ϫ150 bp.
We hypothesized that the FAS promoter might be a target for direct regulation by LXR/RXR heterodimers as well as SREBPs. The preferred binding site for LXR/RXR is a DR-4 (direct repeat with a 4-nucleotide spacing) hormone response element (16). An alignment of the FAS promoter (-699 to ϩ1 bp relative to the transcriptional start site) from human, rat, and chicken is shown in Fig. 4. Previous work has demonstrated that the actions of SREBPs on the FAS promoter are mediated primarily by a pair of non-canonical binding sites located between Ϫ71 and Ϫ54 bp (rat) (15). These regulatory elements, as well as binding sites for SP-1 and NF-Y, are highly conserved. However, the SREBP site at Ϫ150, which is a perfect match to the human LDL receptor SRE in the human and rat FAS promoters, is not completely conserved in the chicken promoter. Surprisingly, although the sequence upstream of Ϫ150 bp is quite divergent in these three species, a DR-4 element present between Ϫ669 and Ϫ655 bp in the rat promoter is highly conserved. The striking conservation of this element suggested that it was likely to be an important regulatory sequence. Interestingly, the sequence of this DR-4 element in rodents and humans is identical to that of the LXR response element (LXRE) identified previously in the murine SREBP-1c promoter (Fig. 5A) (19,20). Gel mobility shift analysis using in vitro translated LXR␣ and RXR␣ proteins and radiolabeled oligonucleotides confirmed that FAS LXRE binds LXR␣/RXR heterodimers (Fig. 5B). Competition assays using unlabeled oligonucleotides revealed that the affinity of this site for LXR/ RXR was slightly greater than that of the previously identified LXRE from the human apoE macrophage enhancer (27) (Fig. 5C).
The presence of binding sites for both LXR/RXR and SREBP in the FAS promoter suggests that these two classes of transcription factors both contribute to the regulation of FAS expression. In support of this idea, we found that cotransfection of expression vectors for LXR␣ and RXR␣ along with expression vectors for the nuclear forms of SREBPs had a dramatic effect on the induction of the Ϫ700 FAS promoter in transient transfection assays. At the highest amount of expression vector tested, the combination of nuclear SREBP-1a expression vector and LXR activation led to a greater than 50-fold increase in FAS promoter activity (Fig. 6A). Expression of SREBP-1c also had an additive effect (Fig. 6B), although ϳ10-fold more expression vector was needed to achieve the same level of promoter activity as with SREBP-1a. This is consistent with the known difference in activity between the SREBP-1a and -1c isoforms (14,22). We further analyzed the dose response of the Ϫ700 FAS promoter to the synthetic LXR ligand T1317 in the presence and absence of SREBP. Fig. 6C illustrates that the maximal response of FAS promoter to the LXR ligand is observed in the presence of expression vectors for both LXR␣/RXR and nuclear SREBP-1c.
To definitively demonstrate that the effects of LXR on the FAS promoter are mediated by the combined action of the LXRE and tandem SREBP sites we analyzed FAS promoter constructs carrying specific mutations in these elements (Fig.  7A). Consistent with the results of the FAS promoter deletion analysis (Fig. 3), mutation of either the LXRE or the tandem SREBP sites between Ϫ71 and ϩ54 reduced the activity of the promoter in the presence of cotransfected LXR expression vector and synthetic LXR ligand (Fig. 7B). Simultaneous mutation of both sites virtually abolished promoter activity. We further examined the effect of these mutations on activation of the FAS promoter by SREBP-1a. In agreement with previous work (15), mutation of the tandem SREBP sites resulted in a complete loss of induction by the SREBP-1a expression vector (Fig. 7C), despite the presence of the SRE element at Ϫ150 bp. The activity of the SREBP mutant FAS promoter in the presence of both LXR and SREBP was not different from that with LXR alone. As expected, mutation of the LXRE had no effect on the ability of SREBP to activate the FAS promoter, but eliminated the additive effect of LXR and SREBP. Taken together, these results indicate that LXR/RXR heterodimers and SREBPs additively regulate the FAS promoter and that this regulation requires the combined action of both the LXR and SREBP binding sites.
Finally, to investigate the potential contribution of direct LXR activation of the FAS promoter to the control of lipogenesis in vivo, we analyzed the ability of the synthetic LXR ligand T1317 to regulate FAS expression and influence plasma triglyceride and HDL levels in mice. C57Bl/6 mice (5 animals per group) were treated for 3 or 7 days with either vehicle or 50 mg/kg T1317. After 3 days of treatment, plasma triglycerides increased ϳ200% in response to T1317 (Fig. 8A), consistent with previous work (18). After 7 days, however, triglyceride levels in these mice had nearly normalized. Treatment with T1317 also led to a significant elevation of plasma HDL cholesterol (Fig. 8B, HDL-C). Unlike the effect on triglycerides, the effect on HDL persisted after 7 days. Next, we endeavored to correlate changes in plasma lipid levels induced by T1317 with changes in hepatic gene expression. After 3 days of treatment, expression of FAS was induced ϳ15-fold by T1317 but had largely normalized by day 7, an effect that mirrored the normalization of triglyceride levels (Fig. 8C). In contrast, expression of SREBP-1c, SCD-1, and ABCG1 was induced at day 3 and remained elevated after 7 days. Thus, alterations in plasma triglyceride levels correlated closely with temporal changes in FAS expression. These data are consistent with the hypothesis that direct action of LXR on the FAS promoter as well as induction of SREBP-1c expression are likely to contribute to regulation of lipogenesis by LXR ligands in vivo.

DISCUSSION
The nuclear receptors LXR␣ and LXR␤ are emerging as key regulators of lipid homeostasis. The physiologic ligands for these receptors are likely to be specific intermediates in the cholesterol biosynthetic pathway such as (24S,25)-epoxycholesterol. LXR␣ is expressed primarily in liver, intestine, adipose tissue, and macrophages, whereas LXR␤ is ubiquitously expressed (16,28). In peripheral cells such as macrophages LXRs play an important role in the regulation of reverse cholesterol transport and the induction of genes in response to cellular lipid loading. Multiple genes involved in the cholesterol efflux pathway, including those encoding the putative cholesterol/

FIG. 5. LXR/RXR heterodimers bind to the FAS DR-4 element with high affinity.
A, alignment of LXR response elements from the chicken, rat, and human FAS and murine SREBP-1c promoters. B, electrophoretic mobility shift assays were performed using labeled oligonucleotides corresponding to either the FAS LXRE or the LXRE from the apoE ME enhancer (27) and in vitro translated hLXR␣ and hRXR␣. Unlabeled oligonucleotides were included as competitors as indicated. In the liver, LXRs appear to regulate both cholesterol and fatty acid metabolism. Mice carrying a targeted disruption of the Lxr␣ gene fail to induce transcription of the gene encoding cholesterol 7␣-hydroxylase (CYP7A1), the rate-limiting enzyme in bile acid synthesis, in response to dietary cholesterol (17). In addition, mice lacking LXR␣ are deficient in the expression of several genes involved in lipogenesis, including FAS, SCD-1, ACC, and SREBP-1. Further evidence for the involvement of LXRs in lipogenesis came with the observation that treatment of mice with the synthetic LXR ligand T1317 induces expression of lipogenic genes and raises plasma triglyceride levels (18). The recent demonstration that the SREBP-1c promoter is a direct target for regulation by LXR/RXR heterodimers provided a potential explanation for the ability of LXR ligands to induce hepatic lipogenesis (19,20). Until now, the effects of LXR activation on the expression of lipogenic genes, including FAS, have been presumed to be entirely indirect.
We have shown here that LXRs regulate FAS expression through direct interaction with the FAS promoter as well as through indirect effects on SREBP-1c. The observation that synthetic LXR ligands induce FAS expression in certain cell types in the absence of changes in SREBP-1 expression led us to map the promoter sequences involved in LXR induction. We found that in addition to binding sites for SREBPs, the FAS promoter contains a high affinity binding site for the LXR/RXR heterodimer that is conserved in birds, rodents, and humans. The LXR and SREBP binding sites independently confer LXRresponsiveness on the FAS promoter, and maximal induction requires the binding of both transcription factors. Finally, plasma triglyceride levels in mice treated with a synthetic LXR agonist correlated with increased FAS expression more closely than with SREBP-1c expression. Taken together our results strongly suggest that direct actions of LXR on the FAS promoter contribute to regulation of lipogenesis by LXR ligands in vivo.
The identification of the FAS promoter as a direct target for LXR/RXR heterodimers fits well with the previously hypothesized role for LXR as a cholesterol sensor (19). The LXR pathway provides a mechanism whereby cholesterol and fatty acid metabolism can be coupled. Under conditions where cellular cholesterol levels are high, it is appropriate that the substrate for cholesterol esterification, fatty acids, be readily available.
Our results indicate that LXRs have the ability to control FAS expression in peripheral cells such as macrophages as well as hepatocytes. The ability to up-regulate FAS expression in response to cholesterol may be particularly important for peripheral cells because, unlike hepatocytes, they do not express CYP7A and are unable to synthesize bile acids. It is also interesting to note that the SREBP-1c promoter is itself a target for both SREBP and LXR (33). In previous studies of both the SREBP-1c and FAS promoters, deletion of the regions now known to contain the LXREs resulted in an increase in steroldependent regulation (3,33). Taken together, the current and previous observations suggest that coregulation of these genes by LXR and SREBP may serve to balance their expression under fluctuating sterol conditions. The possibility that this regulatory arrangement is conserved in the control regions of other lipogenic genes is under investigation.
The ability of synthetic LXR ligands to promote cellular cholesterol efflux makes them potentially attractive agents for the modulation of human lipid metabolism. Their lipogenic activity, however, is a major limitation. Although the ability of LXR ligands to raise HDL levels is promising, the transient hypertriglyceridemia induced by the currently available agonists is an undesirable side effect. Clearly, a detailed understanding of the mechanism whereby LXR ligands raise triglyceride levels will be required before LXR can be optimized as a drug target. The observation that the FAS gene is under direct as well as indirect control of LXR suggests that the relative ability of synthetic LXR agonists to the FAS promoter may be a key determinant of their effects on hepatic lipogenesis. In our study, hepatic FAS expression and plasma triglyceride levels were acutely elevated but then declined during chronic administration of an LXR ligand over 7 days. These observations suggest that compensatory mechanisms may counter LXR effects on the FAS promoter in response to chronic stimulation. The mechanistic basis for this effect is not yet clear.
Ultimately, it may be possible to identify LXR agonists that have selective activity on certain LXR target genes. The most desirable compound would be one that was a strong inducer of ABCA1 and apoE expression yet lacked activity on the FAS and SREBP-1c promoters. Expression of LXR␣ is more prominent than LXR␤ in liver, and studies with knockout mice suggest that LXR␣ is the dominant receptor involved in the control of hepatic lipogenesis (17,18). Moreover, given the wider tissue distribution of LXR␤, this receptor is well positioned to control expression of ABCA1 and rates of cholesterol efflux in many peripheral tissues. Together, these observations suggest that LXR␤-selective agonists might be particularly useful for the modulation of human lipid metabolism.