Regulation of the angiopoietin-like protein 3 gene by LXR.

Angiopoietins are members of the vascular endothelial growth factor family. One family member, angiopoietin-like protein 3 (Angptl3), was recently shown to be predominantly expressed in the liver and to play an important role in regulating lipid metabolism. In this study, we show that the Angptl3 gene is a direct target of the liver X receptor (LXR). Mice fed a high cholesterol diet exhibited a significant increase in Angptl3 expression in the liver. Oral administration to mice of T0901317, a synthetic LXR-selective agonist, increases levels of plasma lipids and Angptl3 mRNA in the liver. Treatment of HepG2 cells with LXR selective agonists led to a dose-dependent increase of Angptl3 mRNA. Analysis of the DNA sequence just 5' of the Angptl3 transcriptional start site revealed the presence of several potential transcription factor binding sites, including that for LXR. When transfected into HepG2 cells, the promoter activity of Angptl3 was significantly induced by LXR- or retinoid X receptor-selective agonists. Mutation of the predicted LXR binding site (DR4 element) completely abolished the LXR agonist-mediated activation of the promoter. Together, these studies show that Angptl3 is transcriptionally regulated by LXR, and reveals a novel mechanism by which LXR may regulate lipid metabolism.

The angiopoietins are a family of secreted growth factors. Together with their respective endothelium-specific surface receptors, such as the receptor tyrosine kinase Tie2, the angiopoietins play important roles in angiogenesis [as reviewed in ref. (1)]. One family member, Angiopoietin-like protein 3 (Angptl3), was previously found to be expressed in a liver specific manner (2), and more recently, it was found to play an important role in the regulation of lipid metabolism (3). Mutation of Angptl3 results in low levels of plasma lipids in KK/San mice (3). These hypolipidemic mice are derived from KK obese mice, which display a multigenic syndrome of moderate obesity and a diabetic phenotype including hyperinsulinemia, hyperglycemia, and hyperlipidemia (4)(5)(6). Administration or overexpression of Angptl3 in mice elicited a rapid increase in circulating levels of plasma cholesterol, triglycerides (TG), and non-esterified fatty acids (3). While these studies revealed a role of Angptl3 in lipid metabolism, they did not address the molecular mechanism by which a hyperlipidemic response was mediated by Angptl3. Understanding the regulation of Angptl3 expression is an important step toward a better comprehension of the physiological network that comprises lipid metabolism.
Many genes involved in lipid metabolism are regulated by nuclear receptors. Liver X receptors (LXRs), including LXR ␣ and LXR ␤ , are members of the nuclear receptor superfamily [as reviewed in ref. (7)]. LXRs heterodimerize with retinoid X receptors (RXRs) and can be activated by both RXR and LXR ligands. Upon activation, LXRs regulate the expression of target genes by binding to specific promoter response elements (termed LXREs) that contain a hexameric nucleotide direct repeat separated by four bases (DR4). Recent studies have shown that these ligand-activated transcription factors play important roles in the regulation of genes that govern cholesterol homeostasis in the liver and peripheral tissues (7).
In addition to cholesterol metabolism, accumulating evidence suggests that LXRs also play important roles in regulating fatty acid metabolism. For example, the expressions of sterol regulatory element-binding protein-1c (SREBP-1c), fatty acid synthase, and lipoprotein lipase were all shown to be strongly activated by LXR agonists (8)(9)(10). It has been suggested that the regulation of SREBP-1c by LXR may contribute to increased levels of plasma TG when animals are treated with LXR agonists (11)(12). However, SREBP-1c cannot be solely responsible for the elevated TG because mice overexpressing SREBP-1c had a reduced level of circulating TG (13). To determine if Angptl3 might be such a factor, we stud-ied regulation of Angptl3. Here we report that the discovery that Angptl3 is indeed regulated by LXR. Given the profound effect of Angptl3 on circulating levels of cholesterol and TG, our findings may offer a novel mechanism for a role of LXR in regulating lipid metabolism.

Cell culture
HepG2 cells, a human hepatocellular carcinoma cell line obtained from the American Type Culture Collection (Rockville, MD), were maintained in MEM medium with 10% heat inactivated fetal calf serum (FCS), non-essential amino acids, and sodium pyruvate in an atmosphere containing 5% CO 2 and 95% air at 37 Њ C. Before assay, cultured cells were trypsinized, washed once with PBS, and then resuspended in the assay medium (same as that of complete medium but with only 0.5% of FCS). All test compounds used in the experiments were diluted in the same assay medium.

RNA isolation
Total RNA was extracted from the cultured cells using TRIZOL reagent according to the protocol provided by the manufacturer (Life Technologies, Grand Island, NY). The RNA was treated with DNase (Ambion, Inc., Austin, TX) and was reverse transcribed using TaqMan reverse transcription reagents as described by the manufacturer (Applied Biosystems, Foster City, CA) before analysis by real-time quantitative RT-PCR.

Primers and fluorogenic probes
Oligonucleotide primers and TaqMan probes were designed using Primer Express software (Applied Biosystems) and were synthesized by Qiagen Operon (Alameda, CA). Sequences of probes and primers are listed in Table 1 . Primers and probes for 18 S rRNA were purchased from Applied Biosystems.

Real-time quantitative PCR
Real-time quantitative TaqMan PCR analysis was used to determine the relative levels of Angptl3 mRNA. TaqMan PCR reactions were performed according to the manufacturer's instructions (Applied Biosystems, TaqMan Universal PCR Master Mix). Target cDNA amplification was detected by the increased fluorescent signal of FAM (reporter dye) during the amplification cycles. Amplification of the 18 S rRNA transcript was performed in the same reactions using a different reporter dye, VIC, as an internal control for variations in RNA amounts. Levels of the target mRNAs were subsequently normalized using 18 S mRNA levels, and are presented as the ratio of treated cells to untreated cells.

Sequence analysis of the genomic region of Angptl3
BLAST searches were performed to identify genomic sequences encoding human and mouse Angptl3 (14). The genomic sequences of human and mouse Angptl3 were compared using mVISTA (15)(16) and ClustalW (17). Putative binding sites for transcription factors were identified by searching against the TRANSFAC database (18) and the positional weight matrices for transcription factors constructed internally.

Construction and mutagenesis of the human Angptl3 promoter-luciferase construct
A firefly luciferase reporter construct was generated with 953 bp of the promoter of human Angptl3 and 49 bp of the 5 Ј UTR. Briefly, human genomic DNA (CLONTECH, Palo Alto, CA) was PCR-amplified with Angptl3 specific primers and a Bgl II restriction site 8 nucleotides from the 5 Ј end of the forward primer (Dexter-10F: 5 Ј -GATCGATCAGATCTGGGAGGCCAAGGTAAAAGAA-3 Ј ) and a Hind III restriction site 8 nucleotides from the 5 Ј end of the reverse primer (Dexter-10R: 5 Ј -GATCGATCAAGCTTTTTTATC TTGATTTTCAATTTCAAG-3 Ј ). After restriction with Bgl II and Hind III, this fragment was cloned into a similarly digested pGL3 Basic vector (Promega, Inc., Madison, WI) and confirmed by sequencing the entire insert and nearby vector (ACGT, Inc.) to create the human Angptl3 promoter-Luciferase (hpAngptl3-Luc) construct.
A similar construct with a mutation of the DR4 element of the human Angptl3 promoter (hpAngptl3/DR4Mut-Luc) was created by site directed mutagenesis of the 5 Ј half-site of the hpAngptl3 DR4 element. The mutagenesis was performed using the Gene-Editor system (Promega, Inc.) and mutagenic oligonucleotide according to the manufacturer's instructions. The mutagenic oligonucleotide, 5 Ј -TCTAACTCAATGTGGAAGAGGGCCCCAT-TCGTGCAAGTTAACAC-3 Ј , was designed to introduce an Apa I site into the 5 Ј half-site of the Angptl3 DR4 element. A restriction digestion with Apa I (Life Technologies, Inc., Bethesda, MD) was used to screen for mutants and the clones were sequenced (ACGT, Inc.) to confirm the lack of any additional mutations in the Angptl3 sequence.

DNA transfection and reporter gene assays
Transfections of HepG2 cells were performed in 96-well plates with FuGENE 6 reagent according to the manufacturer's instructions (Roche Diagnostics Corp., Indianapolis, IN). Immediately prior to transfection, cells were trypsinized and the complete media was replaced by assay medium to a concentration of 5 ϫ 10 5 cells/ml. One hundred microliters of cells were added per well of the 96-well plates. The constructs, hpAngptl3-Luc or hpAngptl3/ DR4mut-Luc, were co-transfected with phRL-null ( Renilla, Promega, Inc.) at a total of 37.5 ng of reporter DNA and 12.5 ng of the Renilla internal control per well to maintain a FuGENE to DNA ratio of 3:1. Six hours after transfection, test compounds suspended in assay medium were added, and 42 h later the cells were harvested using Passive Lysis Buffer and assayed for firefly and Renilla luciferase activities with the Dual-Luciferase Reporter Assay System purchased from Promega. All transfections and subsequent steps were performed in triplicate.  Animal procedures C57BL/6J mice were purchased from Taconic (Germantown, NY). Animals were maintained in normal light cycle, and were provided with water ad libitum. Mice were fed with normal chow diet either with or without 4% of cholesterol. Compound-treated mice were dosed once a day with either vehicle (0.5% of methylcellulose solution) or 10 mpk (mg compound/kg of animal body weight/day) of T0901317 by oral gavage. After 7 days of feeding or gavage dosing, mice were euthanized and blood samples were taken through cardiac puncture. Liver samples were collected for RNA analyses. Plasma total cholesterol, HDL cholesterol (HDL-C), and triglyceride concentrations were determined with assay kits from Sigma Chemical Co. (St. Louis, MO). All procedures were approved by the Institutional Animal Care and Research Advisory Committee at Merck.

Cholesterol feeding increases expression of Angptl3 mRNA in mouse liver
To determine if expression of Angptl3 mRNA may be subject to regulation in response to perturbations of cholesterol homeostasis, mice were fed for 7 days with a normal chow diet or with the same supplemented with 4% of cholesterol. Quantitative PCR analysis showed, as expected, that mice fed a high-cholesterol diet had an increased expression of CYP7A ( Fig. 1 ), which encodes cholesterol 7 ␣ -hydroxylase, the rate-limiting enzyme in bile acid synthesis (19). The high-cholesterol diet also resulted in a 3.4 Ϯ 0.3-fold higher expression of Angptl3 in the liver compared with mice fed a normal chow diet ( P Ͻ 0.01, n ϭ 5) (Fig. 1). Thus, liver Angptl3 expression is responsive to alterations in dietary cholesterol intake.

Administration of LXR-selective agonist increases expression of Angptl3 mRNA in mouse liver
Recent studies reported that feeding mice with a highcholesterol diet or loading cells with cholesterol may result in activation of the nuclear receptor LXR or increased expression of its target genes including CYP7A (8,(19)(20). We thus asked if LXR could play a role in regulating the expression of Angptl3 by treating mice with T0901317, a synthetic LXR-selective agonist (11). Consistent with a previous report (11), oral treatment of mice with T0901317 resulted in significant elevations of plasma total cholesterol, HDL-C, and triglycerides ( Fig.  2A ). Analysis of liver mRNA levels revealed a significant increase of Angptl3 mRNA (Fig. 2B). Oral treatment of mice with 10 mpk of T0901317 for 7 days resulted in a 5.0 Ϯ 1.1-fold increase of Angptl3 mRNA in the liver ( P Ͻ 0.01, n ϭ 5) (Fig. 2B). Thus, liver Angptl3 expression is responsive to the activation of LXR in vivo.

Activation of LXR increases expression of Angptl3 in HepG2 cells
To determine if the T0901317 treatment-induced changes of Angptl3 in mice could be mediated by direct effects on hepatocytes, we studied the impact of LXR agonists on Angptl3 expression in HepG2 cells. Cultured HepG2 cells were incubated for ‫ف‬ 16 h with various oxy- Fig. 4. Comparison of the genomic regions upstream of Angptl3 in human and mouse. Genomic regions spanning Angptl3 in human and mouse were compared using mVISTA (http://www.gsd.lbl.gov/vista). Conserved regions with an identity greater than or equal to 50% between human and mouse sequences are shown as peaks of similarity (vertical axes) relative to the positions along the mouse genomic sequence (horizontal axes). Peaks representing highly conserved regions (identity Ͼ75% over Ͼ100 bp) are shaded if they overlap with non-coding regions, or marked by slanted lines if they overlap with exons. There are two highly conserved non-coding regions upstream of human and mouse Angptl3: the 500 bp region immediately upstream and the 150 bp region 2.5 kb upstream of the gene.  Fig. 3A and 3B, treatment of HepG2 cells with 22( R )-hydroxycholesterol (OH Ch) or T0901317, both of which activate LXR (7,11), resulted in a dose-dependent increase of Angptl3 mRNA. Similar results were observed when cells were treated with 25-OH Ch, which also activates LXR (data not shown). In contrast, addition of 22( S )-OH Ch, which does not activate LXR (7), failed to significantly change the expression of Angptl3. Together, these results suggest that LXR can activate Angptl3 transcription in hepatocytes.

Identification of the Angptl3 promoter
To determine if the LXR-agonist induced effects in HepG2 cells could be due to direct action of LXR on the Angptl3 gene, we studied the Angptl3 promoter. We located the putative transcriptional start site by identifying the full 5 Ј UTR by 5Ј RACE PCR using total RNA from C57Bl/6J mouse liver as the template. The reaction produced essentially a single PCR product, and sequence analysis identified the putative transcription start site as 51 bp upstream of the initiation methionine codon for mouse Angptl3. Comparison of the 5Ј RACE product generated from the mouse with the human Angptl3 genomic sequence suggests that the putative human transcription start site is located 49 bp upstream of the initiation methionine codon.
Next we compared the genomic sequence 5Ј of the mouse and human Angptl3 open reading frames. The comparison revealed that the 500 bp region adjacent to, and a 150 bp region ‫5.2ف‬ kb upstream of, the initiation methionine codon are highly conserved with Ͼ75% sequence identity (Fig. 4). These regions may play a role in the regulation of the Angptl3 gene, thus we searched for putative binding sites of transcription factors in these conserved sequences. Figure 5 shows the putative binding sites identified in the 400 bp region adjacent to the transcription start sites. Several putative binding sites for transcription factors, including HNF-1, LXR, NFB, and C/EBP, are conserved between human and mouse, which merit further experimental verification. Similarly, putative binding sites for HNF-3, HNF-4, and C/EBP were also predicted in the conserved 150 bp region 2.5 kb upstream of the transcription start site (data not shown).

A truncated Angptl3 promoter is functional and responsive to LXR activation
To study the function of the Angptl3 promoter, we transfected HepG2 cells with a human Angptl3 promoterluciferase construct (953 bp of the promoter and 49 bp of 5Ј UTR fused to the luciferase coding sequence). To test if such a truncated promoter is functional and responsive to LXR activation, cells were treated for 42 h with various oxysterols, 9-cis retinoic acid (9-RA), or T0901317. As shown in Fig. 6A, treatment of cells with the LXR agonists, 22(R)-OH Ch, or 25-OH Ch, resulted in a 4-and 3-fold increase in promoter activity of Angptl3, respectively. In contrast, addition of 22(S)-OH Ch, an oxysterol that does not activate LXR, failed to significantly alter the Angptl3promoter activity. In addition, a RXR agonist, 9-RA, in- Fig. 5. Analysis of the promoter region of human and mouse Angptl3. The sequence segment overlapping the 5Ј RACE product is shown in italic, and its 5Ј end is marked by arrow as the putative transcription start sites (i.e., position ϩ1 in the sequences). The translational start codon (i.e., ATG) is also marked by an arrow. A number of putative binding sites of transcription factors are predicted in this region. Those conserved between human and mouse are boxed, and those found in only one species are underlined.
by guest, on July 20, 2018 www.jlr.org Downloaded from creased the promoter activity by about 5-fold. When 22(R)-OH Ch and 9-RA were combined for a single treatment, there was a synergistic ‫-51ف‬fold induction of the Angptl3 promoter activity. Similar but more profound inductions were observed when cells were treated with T0901317, a synthetic LXR-selective agonist (Fig. 6B). Treatment of cells with 1 M of T0901317 increased the Angptl3 promoter activity by 24-fold. Cotreatment of 1 M of T0901317 with 10 M of 9-RA increased the Angptl3 promoter activity by 45-fold. These results suggest that transcriptional regulation of Angptl3 is transcriptionally regulated by LXR and RXR.

The DR4 sequence is a functional LXRE in the Angptl3 promoter
Earlier studies have shown that transcriptional regulation of several genes including ABCA1 (21)(22), SREBP-1c (8), and LPL (10) by LXR requires the presence of the DR4 element. Analysis of the Angptl3 promoter identified a putative DR4 element (5Ј-AGGTTACATTCGTGCA-3Ј) 123 bp upstream of the transcription start site. To test if the DR4 element is indeed necessary for transcriptional activation of the Angptl3 gene by LXR, we introduced a mutation into the 5Ј half site of the DR4 element within the Angptl3 promoter (5Ј-GGGCCCCATTCGTGCA-3Ј). As shown in Fig. 6A, this mutation completely abolished the inducible response of the human Angptl3 promoter to 22(R)-OH Ch, 25-OH Ch, 9-RA, or the combination of 22(R)-OH Ch and 9-RA. Therefore, these results strongly suggest that the identified DR4 element is necessary for LXR-responsiveness.

DISCUSSION
In light of the potential importance of Angptl3 in lipid metabolism (3), we studied the transcriptional regulation of Angptl3. We showed that expression of Angptl3 mRNA is upregulated in mice by cholesterol-feeding and by administration of the LXR agonist T0901317. Furthermore, treatment of HepG2 cells with natural or synthetic agents that selectively activate LXR increases Angptl3 expression. We cloned and analyzed the promoter of the human Angptl3 gene and identified an LXR response element, the DR4 element. We developed an assay that measures the activity of a truncated version of the human Angptl3 promoter. This truncated Angptl3 promoter is active in HepG2 cells and is induced by treatment of agents that activate LXR or RXR. Mutation of the DR4 element completely abolished the inducible response of the Angptl3 promoter to selective LXR or RXR agonists. Our data indicate that Angptl3 is a direct target gene of LXR. The DR4 element is conserved in the human and mouse Angptl3 promoters, suggesting that LXR responsiveness is also conserved.
In addition to the LXR response element, analysis of the Angptl3 promoter identified several potential binding sites for other transcription factors including HNF-1, HNF-4, NFB, and C/EBP. Some of these potential binding sites were also conserved in both the mouse and human Angptl3 promoters, suggesting that those sites may be biologically relevant. HNF-1, HNF-4, and C/EBP are liver-restricted transcription factors, which are involved in live-specific gene expression (23)(24)(25). The identification of HNF-1 and c/EBP sites that are conserved in the mouse and human Angptl3 promoter may account for a restricted expression of Angptl3 in the liver.
Angiopoietins are members of the vascular endothelial growth factor family (26)(27). Members of this family share the following characteristics: a signal peptide, an extended helical domain, a short linker peptide, and a globular fibrinogen-like domain. Several features distinguish Angptl3 from other angiopoietins, such as lack of a cysteine-based motif within the fibrinogen-like domain, lack of calcium binding motif, and almost exclusive expression in the liver (2).
Interestingly, it was recently found that mutation of Angptl3 is responsible for low plasma lipid levels in KK/ Fig. 6. Effect of LXR agonists on activity of a truncated Angptl3 promoter in HepG2 cells. A fragment of the human Angptl3 promoter (from Ϫ953 to ϩ49 bp) with or without mutations at the DR4 element was linked to the firefly luciferase reporter gene. The resulting plasmid was cotransfected with a control reporter plasmid expressing Renilla luciferase into HepG2 cells. Cells were treated with (A) buffer, 10 M of 22(R)-OH Ch, 25-OH Ch, 22(S )-OH Ch, 9-cis-retinoic acid (9-RA), or both 9-RA and 22(R)-OH Ch; or (B) buffer or increasing concentrations of T0901317 or combination of T0901317 (1 M) and 9-RA (10 M). After 42 h, the cells were lysed and relative luciferase activities were measured. Results are expressed as a ratio between the firefly and Renilla luciferase activities, and data are shown as the means Ϯ SD of triplicates of a representative experiment. Three to four independent experiments were per formed with similar results.
San mice, whereas administration of Angptl3 elicited a rapid increase in circulating levels of cholesterol, TG, and free fatty acid (3). Although these studies revealed a potential role for Angptl3 in lipid metabolism, they did not address the cellular and molecular mechanism(s) by which a hyperlipidemic response is mediated by Angptl3. Very recently, it was reported that Angptl3 may regulate levels of VLDL-TG via direct inhibition of lipoprotein lipase activity (28). Here we report for the first time a transcriptional regulation of Angptl3 by LXR, a critical lipid sensing transcription factor. Our findings are thus consistent with a role of Angptl3 in lipid metabolism, and will be useful for further elucidation of mechanism of the hyperlipidemic responses of Angptl3.
Administration of agents that selectively activate LXR results in an increase of total cholesterol (mainly HDL-C) and TG in plasma (11)(12). ABCA1, a mediator of cholesterol efflux to apolipoprotein A-I, is a direct target gene of LXR (21)(22). Thus, induction of ABCA1 by the treatment of LXR agonists may be responsible for an increase of HDL-C. However, a mechanism(s) for the LXR agonistmediated increase of circulating TG remains to be elucidated. In addition to ABCA1, activation of LXR may also increase the expression of a number of genes involved in fatty acid metabolism, including SREBP-1c, fatty acid synthase, and lipoprotein lipase (8)(9)(10). Effects of LXR agonists on such lipogenesis genes are thus believed to be responsible for an increase of circulating TG. However, this interpretation is complicated by the fact that although overexpression of SREBP-1c in mice increases TG level in the liver, TG level in the plasma was reduced (13).
The exact role of Angptl3 in LXR-mediated hyperlipidemic response remains to be tested. Given the profound effect of Angptl3 on plasma lipid levels (3), it is possible that Angptl3 may play an important role in LXR-mediated increases of plasma lipids. It is interesting to note that administration of LXR agonists (11,29) or Angptl3 (28) both result in increase of TG levels largely in the VLDL fraction. Administration of an LXR-agonist to the KK/San mouse, which has an Angptl3 mutation, could be used to test definitively an involvement of Angplt3 in LXR-mediated increases of plasma lipids.