Syndecan-1 Expression Is Regulated in an Isoform-specific Manner by the Farnesoid-X Receptor*

Syndecan-1 ( SDC1 ), a transmembrane heparan sulfate proteoglycan that participates in the binding and internalization of extracellular ligands, was identified in a screen designed to isolate genes that are regulated by the farnesoid X-receptor (FXR, NR1H4). Treatment of human hepatocytes with either naturally occurring (chenodeoxycholic acid) or synthetic (GW4064) FXR ligands resulted in both induction of SDC1 mRNA and enhanced binding, internalization, and degradation of low density lipoprotein. Transient transfection assays, using wild-type and mutant SDC1 promoter-luciferase genes, led to the identification of a nuclear hormone receptor-binding hexad arranged as a direct repeat separated by one nucleotide (DR-1) in the proximal promoter that was necessary and sufficient for activation by FXR. The wild-type, but not a mutated DR-1 element, conferred FXR responsiveness to a heterologous thymi-dine kinase promoter-reporter gene. Four murine FXR isoforms have been identified recently that differ Lipoprotein Binding— We relied on previously published protocols to examine proteoglycan-mediated catabolism of lipoproteins in our system (24, 35). As a model ligand, we used LPL-enriched 125 I-labeled methylated LDL (LPL/ 125 I-mLDL), which binds HSPGs but does not interact with the LDL receptor. The LPL was purified from cows’ milk by heparin-agarose chromatography (28). The 125 I-mLDL was prepared from fresh human plasma by ultracentrifugation (1.019 (cid:5) d (cid:5) 1.063 g/ml), radioiodination, and finally reductive methylation of (cid:6) 35% of the lysyl residues to abolish LDL receptor binding (28, 35). Monolayers of HepG2-FXR cells were incubated at 37 °C for 24 h in MEM supplemented with 10% SS-FBS. The cells were then incubated for 24 h in MEM/ss-FBS that was supplemented with either 1.0 (cid:4) M GW4064 or the corresponding volume of Me 2 SO. Cells that received GW4064 remained exposed to this agent for the remainder of the experiment. To examine proteoglycan-mediated catabolism of lipoproteins, cells were placed into serum-free MEM, 0.2% bovine serum albumin, supplemented with 5 (cid:4) g of 125 I-mLDL protein/ml, with or without GW4064, and either 5 (cid:4) g of LPL/ml or the corresponding volume of lipase buffer. Cells were incubated at 37 °C for 5 h inthese media, after which the surface-bound (assessed by the radioactivity released at 4 °C by 10 mg of heparin/ml), intracellular (heparin-resistant), and degraded (assessed by trichloro- acetic acid-soluble, CHCl 3 -insoluble radioactivity in media) lipoproteins were quantitated (28, 36). To assess efficiency of endocytosis, we calcu- lated lipoprotein internalization as the sum of intracellular accumula-tion plus degradation, as described previously (28, 36). Results are expressed as ng 125 I-mLDL protein catabolized per mg of cellular protein.

Nuclear hormone receptors are a superfamily of ligand-activated transcription factors that play critical roles in development and adult physiology (1). Most family members function by binding to cis-acting response elements located within the promoters, introns, or enhancers of their target genes and regulate gene expression, usually in response to the binding of small lipophilic ligands (1).
A subclass of nuclear hormone receptors form obligate heterodimers with a common partner, the 9-cis-retinoic acid receptor (RXR) 1 (2). A number of these nuclear receptors have been shown to undergo a conformational change upon binding ligand that promotes the release of co-repressor proteins and the subsequent recruitment of co-activator proteins; the net result is increased transcription of the target gene (3,4). The farnesoid-X receptor (FXR, NR1H4) is a member of this subclass and binds as an FXR/RXR heterodimer to FXR-response elements (FXREs) (5,6). In 1999, three groups independently reported (7)(8)(9) that specific bile acids bind to and activate both human and rat FXR at physiological concentrations. The most potent of these natural ligands is the primary bile acid chenodeoxycholic acid (CDCA). These unexpected results extended our understanding of bile acids beyond their traditional roles in the solubilization and absorption of dietary lipids and fatsoluble vitamins to a new class of hormonal ligands that play key roles in gene expression.
Recently, four murine and human FXR isoforms (FXR␣1, ␣2, ␤1, and ␤2) were identified that are derived from a single gene as a result of alternative promoter usage and alternative splicing of the mRNA (10,11). The FXR␤ isoforms have an extended amino terminus with respect to the FXR␣ isoforms. In addition, FXR␣2 and FXR␤2 contain a four-amino acid insert in the hinge region that is absent from both FXR␣1 and FXR␤1 (10,11). The four isoforms are both differentially expressed in a tissue-specific manner and differentially activate the FXR target gene encoding the intestinal bile acid-binding protein (I-BABP) (10). In contrast, all other FXR target genes studied to date (SHP, BSEP, and PLTP) are activated to similar degrees by all four FXR isoforms (10).
adrenal gland (5) and at low levels in the heart, lung, stomach, and adipose tissue (10). Targeted deletion of FXR in a murine model revealed an important role for FXR in cholesterol and bile acid homeostasis because the FXR null mouse was unable to respond normally to diets enriched in fat, cholesterol, or bile acids (12). This imbalance was presumably due to the inability of the mice to regulate the FXR target genes encoding several bile acid and organic anion transporters, including BSEP (13,14), OATP2 (13), I-BABP (14), and SHP (NR0B2). SHP, an unusual member of the nuclear receptor superfamily that lacks a DNA binding domain, functions as a transcriptional repressor and inhibits transcription of CYP7A, a regulatory gene for bile acid synthesis (15,16).
Other FXR target genes encode the secreted proteins phospholipid transfer protein (PLTP) (17,18), apoE (19), and apoCII (20), all known to be involved in the metabolism of plasma lipids and lipoproteins. The observation that plasma triglyceride levels declined when rodents were treated with either a synthetic (GW4064) (21) or a natural (CDCA) FXR ligand (20) is consistent with the hypothesis that activation of FXR modulates lipoprotein levels. The observation that administration of a cholesterol-rich diet to FXR null mice results in elevated serum levels of cholesterol and triglycerides and increased levels of proatherogenic lipoproteins provides additional support for this hypothesis (12). Taken together, these data suggest that activated FXR results in increased metabolism of plasma lipoproteins.
SDC1 is a member of the syndecan family of transmembrane heparan sulfate proteoglycans, which are widely expressed in partially overlapping patterns of expression in many cell types and tissues (22,23). Their principal function appears to be to modulate the ligand-dependent activation of primary signaling receptors at the cell surface leading to an increase in binding and/or internalization of extracellular ligands (23). SDC1 is highly expressed in the liver, where heparan sulfate proteoglycans (HSPGs) bind to lipoproteins via a number of bridging proteins that include lipoprotein lipase (LPL) (24), apoE (25), and hepatic lipase (26). Several studies (27) have detailed an important role for such binding in the hepatic clearance of lipoprotein remnants. The HSPGs on the hepatocyte cell surface are thought to bind to and sequester remnants prior to their transfer to specific receptors such as LDL receptor and LDL receptor-related protein and subsequent endocytosis. Support for this proposal comes from the following observations: (i) apoE-or LPL-enhanced internalization of remnant-like lipoproteins is abolished by either enzymatic or genetic removal of HSPGs from the cell surface (24,25); and (ii) in vivo injection of heparinase into the portal vein of mice reduces hepatic clearance of 125 I-labeled ␤-very low density lipoproteins enriched in apoE (25). A role for the independent transport of lipoprotein remnants or LDL into hepatocytes by SDC1, in the absence of other receptors, has also been described (28,29). Importantly, Fuki et al. (30,31) showed that transfection of Chinese hamster ovary cells with an expression vector for SDC1 led to a significant increase in the catabolism of lipoproteins enriched in LPL and that chimeric FcR-SDC1 receptors could internalize IgG in the absence of endogenous IgG receptors. However, the relative contribution of SDC1, LDL receptor, and LDL receptor-related protein to remnant/lipoprotein clearance in vivo awaits further investigation.
In the current study, we demonstrate that treatment of either primary human hepatocytes or cells derived from human hepatomas with FXR ligands results in increased expression of SDC1 mRNA and increased cellular binding, uptake, and degradation of both LDL and a preformed LDL⅐LPL complex. The current studies suggest that ligands for FXR increase the ex-pression of the SDC1 gene and result in enhanced processing and clearance of lipoproteins.
Cell Culture and Stable Cell Lines-The generation and maintenance of HepG2 and stably infected HepG2-FXR cells have been described (20). HuH7 cells were maintained in modified Eagle's medium supplemented with 10% fetal bovine serum. Primary human hepatocytes were obtained from BioWhittaker (Walkersville, MD) and were cultured on Matrigel-coated 6-well plates at a density of 1.5 ϫ 10 6 cells per well. Culture media consisted of serum-free Williams' E medium supplemented with 100 nM dexamethasone, 100 units/ml penicillin G, 100 g/ml streptomycin.
RNA Isolation and Northern Blot Hybridization-Unless otherwise indicated, HepG2 and HuH7 cells were cultured in medium containing superstripped FBS for 24 h before the addition of ligands or Me 2 SO (vehicle) for an additional 8 -24 h. Primary human hepatocytes were cultured in growth medium with the addition of 1 M GW4064 or Me 2 SO (vehicle) for 8 or 12 h. Total RNA was isolated using TRIzol reagent and was resolved (5-10 g/lane) on a 1% agarose, 2.2 M formaldehyde gel, transferred to a nylon membrane (Hybond N ϩ ; Amersham Biosciences), and cross-linked to the membrane with UV light. cDNA probes were radiolabeled with [ 32 P]dCTP using the Rediprime TM II labeling kit (Amersham Biosciences). Membranes were hybridized using the QuikHyb hybridization solution (Stratagene, La Jolla, CA) according to the manufacturer's protocol. Blots were normalized for variations of RNA loading by hybridization to a control probe, either 18 S ribosomal cDNA or the ribosomal protein 36B4. The RNA levels were quantitated using a PhosphorImager (ImageQuant software; Amersham Biosciences).
Lipoprotein Binding-We relied on previously published protocols to examine proteoglycan-mediated catabolism of lipoproteins in our system (24,35). As a model ligand, we used LPL-enriched 125 I-labeled methylated LDL (LPL/ 125 I-mLDL), which binds HSPGs but does not interact with the LDL receptor. The LPL was purified from cows' milk by heparin-agarose chromatography (28). The 125 I-mLDL was prepared from fresh human plasma by ultracentrifugation (1.019 Ͻ d Ͻ 1.063 g/ml), radioiodination, and finally reductive methylation of ϳ35% of the lysyl residues to abolish LDL receptor binding (28,35). Monolayers of HepG2-FXR cells were incubated at 37°C for 24 h in MEM supplemented with 10% SS-FBS. The cells were then incubated for 24 h in MEM/ss-FBS that was supplemented with either 1.0 M GW4064 or the corresponding volume of Me 2 SO. Cells that received GW4064 remained exposed to this agent for the remainder of the experiment. To examine proteoglycan-mediated catabolism of lipoproteins, cells were placed into serum-free MEM, 0.2% bovine serum albumin, supplemented with 5 g of 125 I-mLDL protein/ml, with or without GW4064, and either 5 g of LPL/ml or the corresponding volume of lipase buffer. Cells were incubated at 37°C for 5 h in these media, after which the surface-bound (assessed by the radioactivity released at 4°C by 10 mg of heparin/ml), intracellular (heparin-resistant), and degraded (assessed by trichloroacetic acid-soluble, CHCl 3 -insoluble radioactivity in media) lipoproteins were quantitated (28,36). To assess efficiency of endocytosis, we calculated lipoprotein internalization as the sum of intracellular accumulation plus degradation, as described previously (28,36). Results are expressed as ng 125 I-mLDL protein catabolized per mg of cellular protein.

RESULTS
Induction of SDC1 by Bile Acids and the Synthetic FXR Ligand GW4064 -In an effort to identify target genes that are regulated by the bile acid receptor FXR, HepG2 cells were infected with retroviral vectors that express either rat FXR␣2 and the neomycin-resistant gene or the neomycin-resistant gene alone (20). G418-resistant cells were isolated, and pooled cell populations were propagated that harbored either the vector alone (HepG2-Neo) or overexpressed FXR (HepG2-FXR). Total RNA was isolated from HepG2-Neo or HepG2-FXR cells that had been treated for 24 h with either vehicle (Me 2 SO), the FXR ligand CDCA (100 M), or the synthetic FXR ligand GW4064 (1 M). These RNA samples were then used to prepare biotinylated cRNAs that were hybridized to high density microarrays containing ϳ6,000 cDNAs/ESTs (Affymetrix HuFL Gene Chip). Analysis of the microarray data identified a number of genes, including SDC1, whose mRNAs were induced by treatment of the cells with either natural or synthetic FXR ligands. We chose to explore FXR-mediated regulation of SDC1 based on the fact that it has been reported to be involved in the binding and internalization of plasma lipoproteins (37).
In order to confirm that SDC1 mRNA levels were induced in response to FXR ligands, HepG2-FXR cells were treated with CDCA or GW4064 for 24 h prior to RNA isolation. Northern blot assays showed that both FXR ligands resulted in a dosedependent increase in the levels of two SDC1 mRNAs (Fig. 1). As expected, the mRNA levels for SHP, a previously characterized FXR target gene (15,16), were highly induced in response to both ligands (Fig. 1). Induction of SDC1 and SHP mRNAs was also observed when the human hepatoma HuH7 cell line was treated with CDCA or GW4064 ( Fig. 2A).
In order to extend our results to a more physiologically relevant system, we treated primary cultured human hepatocytes with GW4064. As shown by Northern blot analysis, SDC1 mRNA levels were induced 3.5-fold after 8 h of treatment (Fig.  2B). A significant, albeit less robust, induction (2.4-fold) was observed after 12 h (Fig. 2B). Induction of SHP mRNA paralleled those of SDC1 (Fig. 2B).
The pregnane-X receptor (PXR) is known to be activated by pathophysiogical concentrations of bile acids (38). To confirm that the induction of SDC1 mRNA levels by bile acids was the result of the activation of FXR, and not PXR, we treated HuH7 cells with rifampicin, a potent PXR ligand. The data of Fig. 2A show that neither SDC1 nor SHP mRNAs were induced by rifampicin, although this treatment resulted in induction of the PXR target Cyp3A11 (data not shown). Taken together, the results of Figs. 1 and 2 indicate that induction of human SDC1 mRNA levels is dependent upon activation of hepatic FXR.
Induction of SDC1 mRNA Levels by Activated FXR Is a Primary Response-In order to determine whether protein synthesis is required for the induction of SDC1 mRNA by FXR ligands, HepG2-FXR cells were treated with cycloheximide prior to the addition of CDCA or GW4064. Northern blot analysis illustrates that the induction of SDC1 mRNA levels by FXR ligands was not blocked by cycloheximide (Fig. 3A). A similar pattern was observed for SHP, a known FXR target gene (Fig. 3A). Consequently, we conclude that the induction of SDC1 mRNA by FXR ligands is likely a primary response, as it occurs in the absence of protein synthesis.
Induction of SDC1 mRNA Levels by FXR Ligands Is Attenuated by Actinomycin D-The data of Fig. 3B demonstrate that incubation of HepG2-FXR cells for 8 h with CDCA resulted in induction of SDC1 and SHP mRNA levels by a process that was attenuated by actinomycin D, an inhibitor of RNA polymerase II.
The experiment shown in Fig. 3C was performed in order to rule out the possibility that FXR ligands induce SDC1 mRNA levels through a mechanism involving stabilization of the mRNA. HepG2-FXR cells were pretreated for 24 h with either vehicle or the FXR agonist GW4064 to induce SDC1 mRNA levels. Actinomycin D was then added to all cells to inhibit polymerase II-dependent transcription. Subsequently, RNA was isolated at the time points indicated, and the relative SDC1 mRNA levels were quantitated and the data plotted (Fig.  3C, left panel). The results demonstrate that under these conditions the half-life of SDC1 is long (Ͼ Ͼ8 h) in both vehicle-and GW4064-treated cells. Consequently, the rapid induction of SDC1 mRNA in response to FXR ligands (Figs. 1 and 2) cannot result from increased stability of the SDC1 mRNA. Under the same conditions, we determined that SHP mRNA half-life is ϳ4 h and is also unaffected by GW4064 treatment (Fig. 3C,  right panel). Based on the results of Figs. 1-3, we conclude that induction of SDC1 mRNA levels is dependent on increased transcription of the gene in response to activated FXR.
The SDC1 Promoter Contains a Novel FXR-response Element-Previous studies (6, 32) have shown that both IR-1 and ER-8 motifs can function as FXREs. Computer-assisted analysis of 10 kb of the published nucleotide sequence upstream of the transcriptional start site of the human SDC1 gene failed to identify any sequences corresponding to either an IR-1 or ER-8. Consequently, in order to identify a cis-acting element responsible for the observed induction of SDC1 mRNA levels by FXR ligands, we cloned ϳ1.3 kb of the proximal promoter of the SDC1 gene and a series of 5Ј deletions into the luciferase reporter pGL3 (Promega). These reporters were co-transfected into HepG2 cells in the presence or absence of plasmids encoding RXR and either rat FXR or the constitutively active VP16-FXR fusion protein. The cells were subsequently treated for 24 h with FXR-specific ligands. In the current studies with SDC1-reporter constructs, as in previous transient transfection studies from this laboratory (20,32), the co-transfected plasmid encoding FXR corresponded to the rat FXR␣2 isoform. This isoform lacks both the amino-terminal extension that is present in the ␤ isoform and the four-amino acid insert in the hinge region (10).
Co-transfection of a reporter construct under the control of the SDC1 promoter (nucleotides Ϫ1298 to ϩ53 relative to the transcription start site) with FXR and RXR led to a 20-fold increase in luciferase activity following the addition of GW4064 (Fig. 4A). Co-transfection of this reporter with a plasmid encoding the constitutively active VP16-FXR showed an ϳ13-fold ligand-independent induction of luciferase activity (Fig. 4A). Similar results were observed with longer reporter constructs (Ͼ5 kb, data not shown), suggesting that the FXR-responsive element in the SDC1 promoter was contained within the 1246 bps flanking the 5Ј end of the transcriptional start site of the SDC1 gene.
Studies with reporter genes containing 5Ј deletions demonstrated that FXR-dependent induction was seen when the promoter contained nucleotides Ϫ937 to ϩ53 (Fig. 4A) but not when the promoter contained nucleotides Ϫ888 to ϩ53 of the SDC1 gene (data not shown). Analysis of the region between Ϫ937 and Ϫ888 bps identified a sequence (AGAG-CAnAGGGGA, at Ϫ921 to Ϫ908 bps) that shows weak homology with an idealized DR-1 element. Whereas DR-1 elements with the consensus hexad sequence AGGTCA have been shown to form an in vitro complex with the FXR/RXR heterodimer (17), no DR-1 element has been shown to support regulation by FXR in vivo.
We next investigated the importance of this DR-1 element in FXR-dependent reporter activation by introducing 4 single nucleotide mutations to produce pGL3SDC1 (Ϫ937)Mut. Fig. 4A shows that induction of this mutant reporter in response to either FXR and GW4064 or VP16-FXR was greatly attenuated compared with the wild-type control reporter. The empty pGL3 control vector was unaffected by GW4064 or VP16-FXR (Fig. 4A).
The DR-1 Functions as an FXRE-To confirm that the DR-1 element identified in the SDC1 promoter is able to act as an FXRE, we constructed a luciferase promoter gene under the control of two copies of either the wild-type [pTK-2x(DR-1)WT] or mutant [pTK-2x(DR-1)Mut] element. The data show that pTK-2x(DR-1)WT-luciferase reporter was activated over 50fold in an FXR and FXR ligand-dependent manner (Fig. 4B). Co-transfection of the same reporter with the plasmid encoding VP16-FXR led to a similar increase in luciferase activity, by a process that was ligand-independent (Fig. 4B). In contrast, induction was completely abrogated when the reporter gene contained two copies of the mutant DR-1 element (Fig. 4B). As expected, co-transfection of FXR/RXR or VP16-FXR/RXR had no effect on luciferase activity when the empty pTK control vector was used (Fig. 4B).
A number of nuclear receptors, including PPARs, HNF4␣, and TR4, have been shown to bind to sequences that correspond to consensus DR-1 elements and to subsequently activate transcription. However, the data of Fig. 4C demonstrate that the pTK-2x(DR-1)WT reporter gene was specifically induced by activated FXR and was unresponsive to HNF4␣, TR4, or SHP, or to ligand-activated PPAR␣ or PPAR␥. Thus, of the seven nuclear receptors examined, only FXR/RXR was able to activate the reporter gene via the non-consensus DR-1 element.
To provide additional evidence that the DR-1 element in the SDC1 promoter can function as an FXRE, we performed EMSA competition assays as shown in Fig. 5. As expected, FXR/RXR bound to a radiolabeled probe that contains the IR-1 element from the PLTP promoter (Fig. 5). Formation of this FXR/ RXR⅐DNA complex was competed away by unlabeled PLTP IR-1 or wild-type SDC1 DR-1 but not by DNA containing the mutant DR-1 (Fig. 5). These in vitro data provide additional support for the hypothesis that the DR-1 element functions as an FXRE.
Induction of the SDC1 by FXR Is Isoform-specific-Alternate splicing and promoter usage produces four FXR isoforms from the sole FXR gene in both mouse and humans (10, 11) (data not shown). Some genes (SHP, BSEP, and PLTP) are transcriptionally activated to similar levels by all four isoforms (10). However, I-BABP is, to date, unique because it is activated in an FXR isoform-specific manner; the endogenous gene is highly induced by FXR␣2 and FXR␤2, which lack the four-amino acid insert in the hinge region but poorly activated by FXR␣1 and FXR␤1, which contain the 4 additional amino acids (10).
To investigate the possibility that SDC1 levels are induced by FXR in an isoform-specific manner, we co-transfected cells with the pGL3-SDC(-937)WT reporter, RXR, and individual murine FXR isoforms, and we then treated the cells with the FXR-specific ligand GW4064. As shown in Fig. 6A, luciferase activity was increased exclusively in response to the two FXR isoforms (FXR␣2 and FXR␤2) that do not contain the fouramino acid insert adjacent to the DNA binding domain of the receptor. Co-transfection of plasmids that encode the isoforms that contain the four-amino acid insertion (FXR␣1 and FXR␤1) failed to induce reporter activity in response to GW4064 (Fig. 6A).
Similar results were obtained when the pTK-2x(DR-1)WT construct was used as a reporter; luciferase activity was induced 20-fold in the presence of FXR␣2 or FXR␤2 and GW4064 but was unaffected in the presence of FXR␣1 or FXR␤1 (Fig.  6B). In contrast, consistent with our recent report (10), a reporter construct under the control of two copies of the IR-1 element from the PLTP promoter was activated by all four FXR isoforms (Fig. 6C). The empty pGL3 vector was unaffected by these treatments (Fig. 6D). Taken together, the transfection data indicate that a novel DR-1 element in the proximal promoter of the human SDC1 promoter functions as an FXRE and controls transcriptional induction of the gene. Moreover, this induction is absolutely FXR isoform-specific.
FXR Ligands Cause Increased Endocytosis of Lipoproteins-SDC1 is abundantly expressed in the adult liver (25), where HSPGs are involved in the binding and internalization of lipoproteins (24,25). We hypothesized that the increase in SDC1 mRNA levels observed upon treatment of HepG2-FXR cells with FXR ligands would correlate with changes in the lipoprotein binding capacity of the cells. To test this hypothesis, we assayed the effect of FXR ligands on the surface binding, internalization, and degradation of 125 I-labeled methylated LDL FIG. 3. Induction of SDC1 mRNA by FXR ligands is independent of protein synthesis but is attenuated by an inhibitor of transcription. A, HepG2-FXR cells were treated for 8 h with 100 M CDCA with or without 100 nM LG100153 (LG) or 1 M GW4064 in the presence or absence of cycloheximide (CX) (10 g/ml), as indicated. RNA isolation and Northern analysis were performed as described in the legend to Fig. 1. B, actinomycin D (ActD) prevents the FXR ligand-dependent induction of SDC1 mRNA. HepG2-FXR cells were cultured in the presence of actinomycin D (5 g/ml) for 20 min before the addition (time 0 h) of vehicle (Me 2 SO) or 100 M CDCA with or without 100 nM LG100153, as indicated. After 8 h total RNA was isolated, and Northern analysis was performed as described in the legend to Fig. 1. C, the half-life of SDC1 mRNA is unaffected by FXR ligands. Cells were cultured for 24 h in the presence of vehicle (Me 2 SO) or GW4064 (1 M). Actinomycin D (5 g/ml) was then added (0 h) to all dishes, and RNA was isolated after the indicated time. The values for both SDC1 and SHP mRNA levels were analyzed by Northern blot analysis and normalized for loading differences. Normalized values were plotted for vehicle (E) or GW4064 (q)-treated cells.
(mLDL). Methylation is known to block the interaction of LDL with the LDL receptor (39).
Pre-treatment of HepG2-FXR cells with 1 M GW4064 for 24 h led to robust increases in the binding, internalization, and degradation of the LPL⅐ 125 I-mLDL complex (Fig. 7, A-C), consistent with an increase in SDC1 expression. Interestingly, there was also a significant increase in the binding, internal-ization, and catabolism of mLDL in the absence of exogenous LPL (Fig. 7). This latter observation may result from multiple transcriptional effects of FXR, which include increased expression of both SDC1 (Fig. 1) and apoE (19). These results, coupled with previous reports (12, 17, 18, 20), suggest that FXR plays an important role in multiple steps in lipoprotein metabolism by regulating the expression of several genes (PLTP, apoCII, and apoE) involved in lipoprotein catabolism and by increasing the lipoprotein/remnant clearance potential of the hepatocyte. This increase in lipoprotein clearance correlates with the changes in SDC1 mRNA levels reported here. In contrast, neither LDL receptor-related protein nor LDL receptor mRNA levels changed following treatment with FXR ligands, as determined by either Northern blot assays or analysis of the data derived from the microarrays, respectively (data not shown). DISCUSSION The current study identifies SDC1 as a novel FXR target gene. SDC1 is a trans-membrane heparan sulfate proteoglycan that participates in the binding and internalization of a wide variety of extracellular ligands (23). It has been proposed that SDC1 functions as a co-receptor, as it binds a variety of ligands prior to their binding to co-localized receptors (23). Because expression of heparan sulfate proteoglycans on the sinusoidal membranes of hepatocytes has been shown to be important in mediating lipoprotein remnant uptake from the space of Disse (27), our data suggest that the activation of FXR is likely to enhance clearance of such remnants from the circulation.
The current studies demonstrate that SDC1 mRNA levels are highly induced in HepG2 cells, HuH7 cells, and primary human hepatocytes in response to natural and synthetic FXR ligands. Interestingly, the FXR-dependent induction of SDC1 mRNA may be specific to humans/primates, because similar studies using multiple rodent hepatic cell lines and primary murine hepatocytes failed to demonstrate regulation of Sdc1 levels by ligands for FXR (data not shown). Characterization of the induction of SDC1 mRNA levels in response to FXR ligands suggests a primary transcriptional mechanism because treatment of HepG2-FXR cells with FXR ligands and cycloheximide did not block induction of SDC1 mRNA, whereas induction was completely blocked by actinomycin D (Fig. 3). The finding that FXR ligands do not affect the half-life of the SDC1 mRNA provides additional support for the hypothesis that SDC1 is a primary FXR target gene.
Initial characterization of response elements in the first few identified FXR target genes suggested the requirement for an IR-1 element containing two half-sites of the traditional nuclear hormone receptor hexad AGGTCA. This strict definition was challenged by the subsequent finding that a novel ER-8 arrangement in the proximal promoter of the multidrug resistance protein 2 gene (ABCC2) could also function as a potent FXRE (32). Computer-assisted scanning of the sequence flanking the 5Ј region of the SDC1 gene failed to identify any such traditional FXRE sequences (IR-1 and ER-8) in the 10 kb proximal to the SDC1 transcriptional start site. However, a non-biased method to determine potential FXR/RXR-binding sites (17) suggested that many arrangements of the hexanucleotide repeat can function as binding sites for FXR/RXR heterodimers in vitro.
FIG. 5. The DR-1 FXRE in the SDC1 promoter competes with an IR-1 FXRE for FXR/RXR binding. In vitro translated FXR and RXR were incubated with a radiolabeled oligonucleotide corresponding to the IR-1 FXRE from the hPLTP promoter in the presence or absence of oligonucleotides containing either wild-type (DR-1WT) or mutant (DR-1Mut) DR-1 sequences or wild-type IR-1 sequences at 50, 100, or 500 molar excess, as described under "Experimental Procedures." The shifted DNA-protein complexes were identified by autoradiography. To identify the cis-acting element responsible for induction of the SDC1 gene in response to FXR, we cloned the 5-kb region flanking the 5Ј end of the SDC1 gene into a luciferase reporter construct. Analysis of this construct, as well as constructs with sequential 5Ј deletions and mutations, localized the cis-element responsible for induction to a non-consensus DR-1 arrangement at Ϫ921 bp relative to the transcriptional start site. Mutation of this element greatly attenuated the FXR-dependent induction of the luciferase reporter gene (Fig. 4). Furthermore, a two copy tandem repeat of the wild-type, but not the mutant DR-1, element highly activated a minimal promoterreporter construct (Fig. 4). We conclude from these results that the DR-1 element identified here in the SDC1 promoter is a potent FXRE. This hypothesis is supported by competition studies using EMSAs shown in Fig. 5. However, to date, we have been unable to demonstrate direct binding of the FXR/ RXR heterodimer to this DR-1 element using EMSAs. It is possible that in vitro a stable complex is not formed because of a high off-rate of FXR/RXR from the DR-1. Alternatively, it is possible that another factor, present in hepatocytes, is required to stabilize the FXR/RXR⅐DR-1 complex.
The recent identification of four murine and human FXR transcripts and isoforms and the demonstration that the four isoforms are expressed in a tissue-specific manner provide a potential mechanism to control gene expression in different tissues. The finding that both I-BABP (10) and SDC1 are differentially responsive to the four FXR isoforms provides support for this hypothesis. However, whereas I-BABP is poorly activated by FXR␣1 and FXR␤1, SDC1 is completely unresponsive to these same two isoforms. Consequently, we conclude that activation of SDC1 by FXR is the first example of an FXR target gene with absolute isoform specificity. Based on these studies, it seems likely that there will be a subset of genes, yet to be identified, that are more responsive to FXR␣1 and FXR␤1 (the two isoforms that contain the four-amino acid insert) than FXR␣2 and FXR␤2. The identification of such genes would support the concept that FXR isoforms provide an additional level of tissue and target gene specificity.
Over the last few years, identification of FXR target genes has begun to provide insight into the network of physiological functions that are governed by FXR-dependent regulation. As expected from the identification of bile acids as natural FXR ligands, a number of these target genes are involved in cholesterol and bile acid homeostasis (6). A second group of target genes, including PLTP, apoCII, and apoE, are known to be involved in the metabolism of triglyceride-rich plasma lipoproteins. Such processing results in the formation of triglyceridepoor, cholesterol ester-rich lipoprotein remnants that are rapidly removed from the circulation. The current finding that ligand-activated FXR induces SDC1 expression and increases the potential of hepatocytes to participate in lipoprotein uptake provides evidence for a mechanism by which the metabolism and clearance of lipoproteins are coordinately regulated in response to bile acids. The finding 30 years ago that treatment of patients with gallstones with CDCA led to an unexplained decrease in plasma triglycerides (33) is consistent with this hypothesis. Thus, pharmacological activation of FXR may prove to be an alternative treatment for hypertriglyceridemia. FIG. 7. Treatment of HepG2-FXR cells with GW4064 leads to increased uptake of 125 I-mLDL. A, GW4064 treatment increases surface binding of 125 I-mLDL. HepG2-FXR cells were pretreated with 1 M GW4064 or vehicle for 24 h before exposure to 125 I-mLDL (5 g/ml) and/or LPL (5 g/ml), as described under "Experimental Procedures." After 5 h, surface-bound lipoprotein was assessed by treatment with heparin, quantitated, and normalized to total cellular protein. B, GW4064 treatment increases surface binding of 125 I-mLDL. Cells were treated as described above, and internalized lipoprotein was assessed as heparin-resistant radioactivity that remained cell-associated. The values were normalized to total cellular protein as described under "Experimental Procedures." C, GW4064 treatment increases catabolism of 125 I-mLDL. Cells were treated as described above, and degraded lipoprotein was quantitated after extraction and normalization to total cellular protein as described under "Experimental Procedures." The results are representative of three separate experiments.