Farnesoid X Receptor Agonist Reduces Serum Asymmetric Dimethylarginine Levels through Hepatic Dimethylarginine Dimethylaminohydrolase-1 Gene Regulation*

The farnesoid X receptor (FXR, NR1H4) is a bile acid-responsive nuclear receptor that plays critical roles in the transcriptional regulation genes involved in cholesterol, bile acid, triglyceride, and carbohydrate metabolism. By microarray analysis of hepatic genes from female Zucker diabetic fatty (ZDF) rats treated with the FXR agonist GW4064, we have identified dimethylarginine dimethylaminohydrolase-1 (DDAH1) as an FXR target gene. DDAH1 is a key catabolic enzyme of asymmetric dimethylarginine (ADMA), a major endogenous nitric-oxide synthase inhibitor. Sequence analysis of the DDAH1 gene reveals the presence of an FXR response element (FXRE) located 90 kb downstream of the transcription initiation site and within the first intron. Functional analysis of the putative FXRE demonstrated GW4064 dose-dependent transcriptional activation from the element, and we have demonstrated that the FXRE sequence binds the FXR-RXR heterodimer. In vivo administration of GW4064 to female ZDF rats promoted a dose-dependent and >6-fold increase in hepatic DDAH1 gene expression. The level of serum ADMA was reduced concomitantly. These findings provide a mechanism by which FXR may increase endothelium-derived nitric oxide levels through modulation of serum ADMA levels via direct regulation of hepatic DDAH1 gene expression. Thus, beneficial clinical outcomes of FXR agonist therapy may include prevention of atherosclerosis and improvement of the metabolic syndrome.

The global epidemic of obesity has led to an increasing prevalence of metabolic syndrome, a wide spectrum of metabolic risk factors characterized by abdominal obesity, low levels of high density lipoprotein cholesterol, high triglycerides, hypertension, glucose intolerance, and a systemic proinflammatory state. These risk factors are closely associated with incidence of cardiovascular disease, a leading cause of mortality in industrialized nations (1,2). In the last decade, several ligand-activated nuclear receptors, such as peroxisome proliferator-activated receptors, liver X receptors, and thyroid hormone receptors, have emerged as promising targets for pharmacological intervention in the treatment of metabolic syndrome (3).
Nuclear receptors are transcription factors that serve as intracellular sensors for endocrine hormones and lipid metabolites, such as bile acids, fatty acids, oxysterols, and xenobiotics. To control a variety of physiological processes, receptors bind to specific cis-acting DNA elements and regulate the expression of target genes by cofactor recruitment upon activation by ligands (4). Based on their pharmacological and ligand-binding properties, the nuclear receptor superfamily can be divided into three groups (5). The first group consists of classical steroid hormone receptors that are activated by high affinity ligands. The second group comprises low affinity receptors for metabolic intermediates. The third group corresponds to orphan receptors that have no known ligands.
The farnesoid X receptor (FXR, 2 NR1H4) belongs to the second group of the nuclear receptor superfamily and functions as a metabolic sensor for bile acids (6 -8). FXR is primarily expressed in liver, intestine, kidney, and adrenal gland. In liver, FXR plays a critical role in the negative feedback regulation of bile acid synthesis through indirect repression of the rate-limiting enzyme cholesterol-7␣-hydroxylase (CYP7A1) (9 -11). FXR also regulates the expression of genes involved in bile acid homeostasis, such as bile acid transporters, bile acid synthetic enzymes, and enzymes required for detoxification (12)(13)(14)(15). Importantly, through evaluation of FXR null mice, FXR has been shown to be involved in the regulation of lipid metabolism, as FXR-null mice display a proatherogenic lipid profile characterized by elevated cholesterol and triglyceride levels (16). Indeed, a number of genes regulating lipid metabolism, such as sterol response element binding protein, lipoprotein lipase, apolipoproteins E, AI, CII, and CIII, and scavenger receptor class B type I, are regulated by FXR (17)(18)(19)(20)(21). Furthermore, a growing body of evidence supports a role for FXR in carbohydrate metabolism (22)(23)(24)(25).
Because FXR has been identified as a modulator of lipid and carbohydrate metabolism, development of a synthetic FXR agonist for treatment of dyslipidemia and diabetes, two important indicators of metabolic syndrome, appears to be an attractive therapeutic approach. The potent synthetic FXR agonist GW4064 reduces triglyceride and increases high density lipoprotein cholesterol when administrated to Fisher rats and improves insulin sensitization in genetically obese ob/ob mice and lowers blood glucose levels in diabetic db/db mice (25)(26)(27). In the present study, we have shown that GW4064 increases hepatic expression of the dimethylarginine dimethylaminohydrolase-1 (DDAH1) gene (but not DDAH2 gene) that is accompanied by a reduction in serum asymmetric dimethylarginine (ADMA), a substrate of DDAH1 enzyme (28,29). Because ADMA is a nitric-oxide synthase (NOS) inhibitor and nitric oxide (NO) plays an important role in protection against the onset and progression of cardiovascular disease, reduction of plasma ADMA levels through FXR activation may lead to overall improvement of vascular health.

EXPERIMENTAL PROCEDURES
Animal Care and Treatment-Female ZDF rats were purchased from Charles River (Wilmington, MA). Animals were acclimated for one week prior to study initiation. Rats were individually housed in polycarbonate cages with filter tops. All animals received deionized water and 5008 Diet (PMI Nutrition International, Brentwood, MO) ad libitum and were maintained in accordance with the Institutional Animal Use and Care Committee of Eli Lilly and Company and the National Institutes of Health Guide for the Use and Care of Laboratory Animals. FXR agonist GW4064 was administered twice daily for nine days by oral gavage at the indicated doses (Figs. 1, 5, and 6). Animals were sacrificed, and serum and liver were collected and snap-frozen in liquid nitrogen for later use.
Microarray Analysis of Liver Gene Expression-Total RNA was isolated from ϳ100 mg of female ZDF rat liver tissue using TRIzol methodology (Invitrogen). RNA was further purified using RNeasy TM columns (Qiagen, Valencia, CA). Equal amounts of RNA from each animal within treatment groups were pooled (n ϭ 6/group), and samples were prepared for GeneChip TM analysis according to the Affymetrix eukaryotic expression sample protocol revision 1 (Affymetrix, Santa Clara, CA). Samples were then hybridized to RAE230A oligonucleotide arrays. Data analysis and mining were performed using the Affymetrix Microarray Suite (MAS version 4.0) and data mining tool (DMT version 2.0) software.
Real-time PCR Analysis of Gene Expression-Real-time quantitative PCR was performed using the 5Ј fluorogenic nuclease assay and an ABI 7900 Prism system (PE Applied Biosystems, Foster City, CA) to determine the relative abundance of assayed mRNAs. Samples were normalized by determining the relative abundance as compared with 36B4 mRNA. Real-time quantitative PCR primer-probe sets for rat DDAH1 (Rn00574200_m1), rat SHP (Rn00589173_m1), rat PEPCK (Rn001529008_g1), rat glucose-6phosphatase (Rn00565347_m1) and rat 36B4 (Rn00821065_g1) were obtained from Taqman Assays-on-demand Gene Expression Products (PerkinElmer Life Sciences). PCR reactions were run in quadruplicate 10-l reactions in 384-well plates that contained Universal Master Mix (PE Applied Biosystems), 2 pmol of each forward and reverse primer, 1.5 pmol of probe, and cDNA. Data are shown as mean Ϯ S.E.
Reporter Constructs-A 470-bp DNA sequence (90 kb downstream from the transcription initiation site) containing the conserved FXRE in the DDAH1 intron 1 region was PCR-amplified from rat genomic DNA using primers 5Ј-CGCCGTGT-ACCCAGTTTTATGTCTA-3Ј and 5Ј-CTTAATACCTCAT-GGATCCCGACCT-3Ј. The PCR product was subsequently cloned into the pGL3-tk-LUC, which contains the minimal thymidine kinase promoter (bases Ϫ105 to ϩ51) linked to the luciferase gene. This construct was designated pGL3-tk-DDAH1 enhancer-luc. Point mutations were introduced in the FXRE using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). The primers used were 5Ј-CACACTT-ATTCCGGAATCAATGATTAATTTTAAGGGAAAC-3Ј and 5Ј-GTTTCCC TTAAAATTAATCATTGATTCCGGAATAA-GTGTG-3Ј, where the mutated bases are indicated in bold. The presence of the mutation was confirmed by automated DNA sequencing.
Transient Transfections and Luciferase Assays-Human epithelial kidney cell line HEK293T was maintained in high glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. For transient transfection of HEK293 cells, 2 ϫ 10 4 cells were seeded into 96-well plates and transfected using the FuGENE 6 transfection reagent (Roche Applied Science) according to the manufacturer's instructions. Each transfection contained 20 ng of luciferase reporter and 20 ng of each cytomegalovirus (CMV)-driven expression plasmid where indicated (pCMV6-FXR, pCMV6-RXR). FXR agonist GW4064 was added to the medium 24 h after transfection and incubated for another 24 h. Firefly luciferase activity in the cell extract was then measured using standard luciferase substrate reagents (BD Biosciences). All experiments were performed in triplicate and repeated at least three times. Data are shown as mean Ϯ S.E.

ADMA/NOS Pathway Metabolite Analysis-N
Arginine, homoarginine, citrulline, and ortho-phthaldehyde were purchased from Sigma. Methanol, ammonia, and acetonitrile were of high pressure liquid chromatography grade. ADMA was analyzed as previously described by Teerlink et al. (32) with modifications. The method was cross-validated by direct quantitation using liquid chromatography-tandem mass spectrometry analysis. Briefly 100 -200 l of EDTA plasma or serum was diluted in 1ϫ phosphate-buffered saline (Invitrogen), treated with 8 M MMR, mixed, and loaded onto 30-mg OASIS 30-mg MCX cation exchange columns in 96-well plate format without pretreatment of solid phase. Samples were washed one time with 1 ml of 100 mM HCl followed by a second wash with 1 ml of methanol. ADMA and NOS pathway analytes were eluted twice with 0.7 ml of ammonia:water:methanol (10: 40:50). The combined eluent was dried under N 2 at 60°C in a 96-well plate evaporator. Dried samples were reconstituted in 100 l of water and transferred to a clean plate or vials for OPA derivatization. Cross-validation samples were reconstituted in 90% acetonitrile and injected onto a 2.1 ϫ 150-mm 5-Atlantis HILIC column with a 20-mm guard column (Waters, Bedford, MA). Citrulline, Arg, homoarginine, SDMA, and ADMA were eluted upon a gradient from 90% acetonitrile in 10 mM ammonium formate, 0.4% formic acid to 50% acetonitrile in 10 mM ammonium formate, and 0.4% formic acid. Multiple reaction monitoring transitions were as follows: 203.13172.1 SDMA, 203.13115 ADMA, 189.13158.1 MMR, 175.13116 Arg, and 176.13159.1 citrulline. Study samples were derivatized with ortho-phthaldehyde as previously described (32), injected onto a 3.0 ϫ 150-mm Luna C18 (2) column equipped with a 4-mm precolumn (Phenomenex, Torrance, CA), and eluted under isocratic conditions (8.7% acetonitrile in 50 mM potassium phosphate, pH 6.5). Detection was at excitation and emission wavelengths of 340 and 455 nm, respectively. Response factors generated from peak areas for ADMA, SDMA, homoarginine and citrulline relative to MMR were used to interpolate concentration from 8-point standard curves between 50 nM and 10 M ADMA, ADMA, citrulline, and homoarginine and between 0.5 nM and 300 M arginine. ADMA was expressed alone and also normalized to either SDMA or arginine. In the case of the latter, this was referred to as the NOS impairment index.

RESULTS
In an attempt to study the hepatic pharmacology of FXR agonist GW4064, liver gene expression profiles from insulinresistant female ZDF rats treated for 9 days with doses of GW4064 ranging from 10 -100 mg/kg were analyzed. Pools of total RNA isolated from liver were hybridized to Affymetrix RAE230A oligonucleotide arrays representing 10,011 rat genes.
An increase or decrease of 1.5-fold in mRNA levels was used as a threshold to define significant gene regulation. A total of 452 genes on the arrays exhibited significant expression changes in GW4064-treated rat livers, with the expression of 239 genes being increased and 213 genes being decreased. Identified genes displaying dose-dependent expression changes upon treatment include previously well characterized FXR target genes, such as small heterodimer partner (SHP) (10,11), bile salt export pump (12), phosphoenolpyruvate carboxykinase (PEPCK) (22,33), and glucose-6-phosphatase (25). In addition, expression of a novel FXR target gene, DDAH1, was elevated in a dose-dependent manner. Using quantitative real-time PCR (Q-RT-PCR), we confirmed GW4064 dose-dependent elevations of DDAH1 mRNA with a peak induction of Ͼ6-fold at the highest dose of 100 mg/kg (Fig. 1A). Similarly, the changes of gene expression in SHP, PEPCK, and glucose-6-phosphatase were also confirmed by Q-RT-PCR (Fig. 1, B-D).
To determine whether DDAH1 is a direct target gene of FXR, we aligned rat and human genome sequences using the online whole-genome navigation tool ECR Browser and identified an evolutionarily conserved region (ECR) between rat and human DDAH1 genes. Because critical regulatory elements are most likely located in the regions conserved between rodent and human, the ECRs were then scanned by the rVISTA program for FXRE sequences that are typified by an inverted repeat of the AG(G/T)TCA hexad containing a single nucleotide spacer or IR-1 element. The human DDAH1 gene region that was scanned spanned Ͼ100 kb upstream of the transcription initiation site. Three putative FXREs in the 5Ј-flanking region and one putative FXRE in the 108-kb intron 1 region were identified. Within the 5Ј-flanking region, the sequence of the putative FXRE located ϳϪ108 kb upstream of the DDAH1 coding region is AGGTTAtAAATTC, the putative FXRE located approximately Ϫ81 kb upstream is AAGTCAaCAACCC, and the putative FXRE located approximately Ϫ19 kb upstream is ATGTTAtAAACCC. The half-sites within these putative elements resembled nuclear receptor binding sites, but they deviated from consensus core sequences. On the other hand, the intronic FXRE located 90 kb downstream of the transcription initiation site showed higher similarity to the consensus IR-1 element GGGTCAaTGACCA (Fig. 2). To generate a chimeric rat DDAH1-luciferase reporter construct, a 470-bp fragment encompassing the rat intron 1 FXRE was inserted upstream of the luciferase reporter gene (Fig. 3). The ability of FXR to transactivate luciferase expression was examined by transient transfection of this reporter construct into human epithelial kidney cell line HEK293T. In the presence of exogenously co-expressed FXR protein, the DDAH1-enhancer-luciferase reporter responded positively to GW4064 treatment in a dose-dependent manner, with a peak induction of 7-fold at the 1 M concentration (Fig. 3A). When co-transfected with both FXR and RXR expression vectors, basal activation of the reporter was elevated by 4-fold, as compared with co-transfection with the FXR expression vector alone. GW4064 caused a dose-dependent elevation in reporter activity, resulting in 5.4-fold induction at the 1 M concentration (Fig. 3A). This level of induction by GW4064 is consistent with the transcription induction potential that is promoted by a reporter construct containing two copies of the IR-1 sequence from the PLTP promoter linked to luciferase. This construct was activated by both FXR and RXR in a similar manner, although the maximal fold induction was 14-fold over basal (Fig. 3B). Mutation of the putative FXRE from the rat DDAH1 intron 1 reporter construct significantly reduced basal transcription levels and ablated the ability of GW4064 to induce reporter expression (Fig. 3C). Taken together, these data demonstrate that intron 1 of the rat DDAH1 genomic sequence contains a functional FXRE that has the potential to promote a transcriptional response to levels comparable with well characterized hepatic FXR target genes in the presence of an FXR agonist.
To determine whether FXR/RXR heterodimers bind directly to the FXRE identified in the first intron of rat DDAH1 gene, EMSAs were performed using 32 P-radiolabeled oligonucleotides containing the putative FXRE. Neither FXR nor RXR alone were capable of binding to the element. When both FXR and RXR proteins were incubated together with the DNA probe, a robust FXR-RXR complex was formed with the DDAH1-FXRE (Fig. 4A). The DNA binding was effectively competed by an excess amount of either unlabeled cognate DDAH1-FXRE or the PLTP-IR-1 oligonucleotide. The DDAH1-FXRE oligonucleotide that corresponds to the reporter construct mutation could not compete for FXR-RXR binding, suggesting that the FXR/RXR heterodimer interacts specifically with the FXRE consensus site in the DDAH1 promoter. A similar result was also obtained when DDAH1-FXRE-radiolabeled oligonucleotide was incubated with HepG2 nuclear extract (Fig. 4B).
To determine whether induction of hepatic DDAH1 gene expression by FXR agonist results in modulation of circulating ADMA levels, we measured ADMA in GW4064-treated female ZDF rats using high pressure liquid chromatography with fluorescence detection methodology. Average relative recovery of serum ADMA, as measured by a modification of the method of Teerlink et al. (32), was 99.03 Ϯ 0.09% across the dynamic range of the experiment. Coefficients of variation were 3.5% at the lower limit of quantitation (LLOQ), 0.05 M LLOQ, and 0.53% at the upper limit of quantitation (ULOQ), 10 M ULOQ. Precision of control samples was Ϫ5.6% at the LLOQ and 1.59% at the ULOQ. Calibration slopes averaged 0.2368 with an intercept at 0.00125 response factor, r 2 ϭ 0.9999. The method was cross-validated using HILIC high pressure liquid chromatography tandem mass spectrometry of underivatized NOS pathway metabolites.
FXR agonist treatment caused a dose-dependent decrease in serum ADMA following nine days of oral dosing. Average serum ADMA levels of vehicle-treated rats were 0.74 Ϯ 0.017 M. This was ϳ40% higher than other rat lineages studied, including Dahl salt-sensitive, Fisher 344, Sprague-Dawley, ZSF, and male diabetic ZDF rats (data not shown). Decreases in ADMA in response to the FXR agonist GW4064 ranged from 11.0% at the lowest dose to a maximum reduction of serum FIGURE 1. Dose-dependent modulation of FXR target genes by FXR agonist GW4064. Target gene transcripts were quantified by Q-RT-PCR using RNA isolated from the liver tissue of female ZDF rats treated with either vehicle or GW4064 at the indicated doses for nine days (n ϭ 6/group). A, expression of a novel FXR target gene, DDAH1, is elevated in rat liver by GW4064 administration. The genes known to be regulated by FXR responded dose-dependently to GW4064 (B, small heterodimer partner (SHP)); C, glucose-6-phosphatase (G6Pase); D, phosphoenolpyruvate kinase (PEPCK)). Data are normalized to the 36B4 gene and are presented as the mean Ϯ S.E. Statistical analyses using one way analysis of variance followed by comparison with vehicle by Dunnett's method were performed. ADMA of 30.7% at the highest dose (Fig. 5A). Serum arginine levels in vehicle-treated animal averaged 78.3 Ϯ 37.5 M and were variable as might be predicted from animals consuming diet ad libitum to the point of sacrifice (Fig. 5B). Levels of arginine trended upward dosedependently in response to FXR agonist, reaching a maximum at high dose of 173 Ϯ 38, but the elevation was not statistically significant. To establish a ratio to describe an index of potential endothelial nitricoxide synthase (eNOS) impairment, ADMA was expressed as a function of arginine in the form of the ADMA/arginine molar ratio. Fig. 5C illustrates that the eNOS impairment ratio was significantly and dose-dependently decreased by FXR agonist by Ͼ90% at all doses. Despite the high variability within treatment groups, this reached significance at middle and high dose treatments of GW4064. Thus, FXR agonist treatment causes a dramatic decrease in the ratio of NOS inhibitor to NOS substrate. Through scatter plot display of individual animal hepatic DDAH1 and DDAH2 gene expression levels and individual serum ADMA levels, a clear correlation between serum ADMA reduction and elevated DDAH1 (but not DDAH2) gene expression resulted (Fig. 6). Collectively, these data indicate that FXR-mediated regulation of hepatic DDAH1 levels translate into lower circulating ADMA levels.

DISCUSSION
In this study, we have identified DDAH1 as a direct target gene of FXR in liver. Two isoforms of DDAH, DDAH1 and DDAH2, share similar gene structure and enzyme activity but display distinct tissue distributions. DDAH1 is found predominantly in tissues expressing neuronal NOS and in peripheral tissues such as liver, adrenal gland, testis, and kidney, whereas DDAH2 is found predominantly in tissues expressing endothelial NOS and inducible NOS, such as heart and aorta (34,35).
Dimethylarginines originate from the degradation of methylated proteins. Methylation of arginine residues is catalyzed by protein arginine methyltransferases, giving rise to three forms of methylarginines, N-monomethyl-arginine, asymmetric dimethylarginine (ADMA), and symmetric dimethylarginine (SDMA) (36). Proteolytic catabolism of arginine-methylated proteins releases free methylarginine. N-Monomethyl-arginine and ADMA (but not SDMA) act as inhibitors of NOS by com-  peting with the substrate of the enzyme arginine (37,38). In the cardiovascular system, nitric oxide that is generated from endothelium acts as an endogenous anti-atherosclerotic molecule by inducing vasodilatation, inhibiting the adhesion of platelets and white blood cells, suppressing vascular smooth muscle cell proliferation, and reducing vascular production of superoxide radicals (37).
Elevated plasma ADMA levels are associated with reduced nitric oxide synthesis and are detected in the human plasma of several clinical conditions, including hypercholesterolemia, hypertriglyceridemia, hypertension, type II diabetes, chronic renal failure, and chronic heart failure, implicating ADMA as a marker of cardiovascular risk (37, 39 -43). Because ADMA is generated constantly in the course of normal protein turnover, two known mechanisms prevent its accumulation in the body and eliminate potential detrimental effects on vascular tone and structure. One mechanism occurs through renal excretion of ADMA and the other mechanism is through conversion to citrulline and methyamines by the action of DDAH (42). Indeed, overexpression of human DDAH1 in transgenic mice results in a 2-fold increase in NOS activity and nitric oxide production that is associated with 2-fold reduction in plasma ADMA (44,45).
Here, we have discovered that administration of the FXR agonist GW4064 to female ZDF rats leads to dose-dependent activation of DDAH1 (but not DDAH2) gene expression in liver. The induction of DDAH1 gene expression is highly correlated with the reduction of serum ADMA levels, indicating that liver may be a major site for elimination of ADMA through the action of DDAH1. In agreement, previous reports suggest that the liver likely plays an important role in the metabolism of ADMA by taking up large amounts of ADMA from the systemic circulation (46 -48). Other nuclear receptor agonists, such as the PPAR␣ agonist fenofibrate, influence serum arginine levels without modulation of ADMA levels; whereas, both arginine and ADMA are influenced in opposite directions by rosiglitazone and all-trans-retinoic acid (41,49,50). As the endogenous eNOS inhibitor ADMA is lowered by FXR agonist treatment, we show a corresponding trend toward increasing arginine levels in a dose-dependent manner (Fig. 5). In fact, the benefit of increasing arginine alone in rescuing metabolic dysfunction phenotypes has been shown previously (51). Thus, it is plausible that fenofibrate, glitazones, and FXR-mediated trends in elevating serum arginine alone could play a role in improving metabolic dysfunction. On the other hand, increases in arginine that are accompanied by decreases in ADMA levels can be expressed as an index of potential NOS impairment. Accordingly, as arginine levels increase and ADMA levels decrease, the index may be reflective of the FXR influence on the metabolic rate of eNOS. The decrease in ADMA in response to FXR correlates with the relative hepatic gene expression of DDAH1 but not with that of DDAH2 (Fig. 6), suggesting that the primary clearance of ADMA from the vascular compartment may be FIGURE 5. FXR regulates serum ADMA and arginine levels. A, serum ADMA levels in female ZDF rats treated orally with GW4064 for nine days. Average ADMA levels for each group are shown (n ϭ 6, ϮS.E.). Analysis of variance by Student-Newman-Keuls showed all comparisons were significant (p Ͻ 0.02). B, serum arginine levels in female ZDF rats treated orally with GW4064 for nine days. Average levels for each group are shown (n ϭ 6, ϮS.E.). None of the comparisons were statistically significant. C, expression of the eNOS impairment index is described as ADMA/arginine molar ratio, which reflects the eNOS inhibitor to substrate ratio. Treatment group averages (shown ϮS.E.) varied as a function of arginine variability; however, significance was observed based on Mann-Whitney rank sum comparison (p Ͻ 0.05). through FXR-mediated hepatic gene expression of DDAH1 and consequent metabolism of ADMA.
An FXRE was identified in the intron I region of the DDAH1 gene that is conserved among human, dog, rat, and mouse. A luciferase reporter construct driven by a DDAH1 gene intronic enhancer fragment containing the FXRE site linked to the thymidine kinase basal promoter responded to GW4064 in a dosedependent manner in the presence of co-transfection of FXR and RXR expression plasmid. This response was completely abolished when the FXRE site was mutated by site-directed mutagenesis, suggesting that the activation was through the FXRE site. Furthermore, the EMSA results confirmed that FXR/RXR heterodimer bind directly to the DDAH1-FXRE site in a similar manner as a previously well characterized PLTP-IR-1 site.
Although a putative FXRE site was also identified in the proximal 5Ј upstream region of the rat DDAH2 gene, no agonistresponsive transcriptional activation was observed with a DDAH2 promoter-reporter construct, consistent with Q-RT-PCR analysis of DDAH2 gene expression from the in vivo study (data not shown and Fig. 6). DDAH2 expression has been reported to be induced by all-trans-retinoic acid treatment with increased nitric oxide synthesis in endothelial cells (49), suggesting that DDAH1 and DDAH2 may be regulated by differential tissue-specific factors.
A recent study showed that FXR is expressed in the vascular smooth muscle cells both in vitro and in vivo (52). Our study provides evidence for the first time that FXR may play an important role in the preservation of vascular health, not only by regulation of lipid metabolism, but also by modulation of ADMA and consequently nitric oxide levels through the regulation of liver DDAH1 expression.