Antiatherosclerotic Effects of Small-Molecular-Weight Compounds Enhancing Endothelial Nitric-Oxide Synthase (eNOS) Expression and Preventing eNOS Uncoupling

Nitric oxide (NO) generated by endothelial NO synthase (eNOS) is physiologically important for vascular homeostasis. Blockade of NO synthesis with pharmacological inhibitors causes significant peripheral vasoconstriction and elevation of blood pressure. Likewise, mice with a disrupted eNOS gene are hypertensive, and they lack endothelium-dependent, NO-mediated vasodilatation (Huang et al., 1995). Besides its role in controlling blood pressure, NO protects the vasculature from thrombosis by inhibiting platelet aggregation and adhesion. In addition, endothelial NO possesses multiple antiatherosclerotic properties, including inhibition of leukocyte adhesion and prevention of smooth muscle proliferation (Li and Förstermann, 2000a). In agreement with this concept, pharmacological inhibition of eNOS causes accelerated atherosclerosis in rabbits (Cayatte et al., 1994) and in mice (Kauser et al., 2000), and genetic eNOS deficiency accelerates the development of atherosclerosis in apolipoprotein E-knockout (apoE-KO) mice (Kuhlencordt et al., 2001).

Although eNOS is a constitutively expressed enzyme, its expression is regulated by a number of biophysical, biochemical, and hormonal stimuli, both under physiological conditions and in pathology (Li et al., 2002a,b; Searles, 2006). Physiological stimuli up-regulating eNOS expression include shear stress produced by flowing blood, growth factors, and hormones such as estrogens (Li et al., 2002a,b; Searles, 2006). Pleiotropic effects of some cardiovascular drugs such as statins (Endres et al., 1998; Laufs et al., 1998), angiotensin-converting enzyme inhibitors (Linz et al., 1999a,b), AT₁ angiotensin receptor blockers and dihydropyridine calcium channel blockers (Ding and Vaziri, 1998) also include an up-regulation of eNOS expression.

Due to the antithrombotic, antiatherosclerotic, and antihypertensive properties of endothelial NO, the eNOS enzyme is an interesting target for the prevention or therapy of cardiovascular diseases.

A primary screening of chemical libraries for compounds increasing eNOS transcription yielded two small-molecular-weight compounds with related structures, namely, AVE9488 (earlier designation C2431; Wohlfart et al., 2002) and AVE3085. Experiments presented in the current article provide evidence that AVE9488 and AVE3085 increase endothelial NO production by a simultaneous up-regulation of eNOS expression and a reversal of eNOS uncoupling. In vitro pretreatment of marrow mononuclear progenitor cells from patients with ischemic cardiomyopathy with AVE9488 significantly increased their eNOS expression. This was associated with an enhanced migratory capacity in vitro and improved neovascularization capacity of the infused marrow mononuclear progenitor cells in vivo (Sasaki et al., 2006). In the present study, we provide evidence that the increased endothelial NO production induced by AVE9488 and AVE3085 was associated with reduced cuff-induced neointima formation and reduced formation of atherosclerotic plaques in apoE-KO mice.

Materials and Methods

Chemicals and Reagents. AVE9488 and AVE3085 were synthesized at Sanofi-Aventis (Industriepark Höchst, Frankfurt, Germany). All other biochemical reagents were of highest analytical purity, and they were purchased from Sigma-Aldrich Chemie GmbH (Taufkirchen, Deisenhofen, Germany).

Cell Culture. All cell culture experiments were performed in accordance with the German genetic engineering law and German biosafety guidelines. Before use of primary human cell cultures, a respective protocol was submitted to and approved by a local biosafety committee. Isolation of human umbilical vein endothelial cells (HUVEC) and measurements of intracellular cGMP were performed as described previously (Wohlfart et al., 1997). Cell culture medium for HUVEC consisted of Iscove's modified Dulbecco's medium containing GlutaMax (Invitrogen, Karlsruhe, Germany), and it was supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin, and 20% fetal bovine serum. HUVEC-derived EA.hy926 endothelial cells were kindly provided by Dr. Cora-Jean Edgell (University of North Carolina, Chapel Hill, NC), and they were grown under 10% CO₂ in Dulbecco's modified Eagle's medium (Sigma-Aldrich Chemie) supplemented with 10% fetal bovine serum, 2 mM l-glutamine, 1 mM sodium pyruvate, 100 U/ml penicillin, 100 μg/ml streptomycin, and 1× hypoxanthine, aminopterin, and thymidine) (Invitrogen) (Li and Förstermann, 2000b).

Analysis of eNOS Promoter Activity. A stable EA.hy 926 cell line was generated by transfection of EA.hy 926 cells with pGL₃-eNOS-Hu-3500-neo, which contains a neomycin-resistance gene and a 3.5-kb promoter fragment of human eNOS driving the luciferase reporter gene (Li et al., 1998). These EA.hy926 cells were cultured in medium containing additionally 1 mg/ml Geneticin (G418; Invitrogen) as selection reagent. Cells were plated in 96-wells at a density of 40,000 cells/well. One day later, confluent cells were incubated with AVE9488 and AVE3085 for 18 h at the indicated concentrations. After washing once with phosphate-buffered saline (PBS), luciferase activity was determined using the Luciferase Assay System (Promega, Mannheim, Germany) in a Genios microplate reader (Tecan, Vienna, Austria). The luciferase activity, normalized for protein concentration of cell lysates, was used as a determinant of eNOS promoter activity (Li et al., 1998).

To localize the promoter region responsible for the transcription activation by the compounds, transient transfection experiments were performed in EA.hy 926 cells using the transfection reagent SuperFect (QIAGEN GmbH, Hilden, Germany). In addition to pGL₃-eNOS-Hu-3500 (Li et al., 1998) and pGL₃-eNOS-Hu-1600 (Li and Förstermann, 2000b), pGL₃-eNOS-Hu-954 and pGL₃-eNOS-Hu-263 were also used. The latter two constructs contained the proximal human eNOS promoter sequence of 954 bp and 263 bp, respectively, and they were derived from pGL₃-eNOS-Hu-1600 by progressive deletion of the promoter fragment. The plasmid pRL-SV40 (containing the Renilla luciferase gene driven by a simian virus 40 promoter) was cotransfected for normalization. Twenty-four hours after transfection, cells were incubated for further 18 h with AVE9488 and AVE3085. Thereafter, the luciferase and Renilla luciferase activities of the extracts were determined using the Dual-Luciferase System (Promega) as described previously (Li and Förstermann, 2000b).

Electrophoretic Mobility Shift Assay. Binding activities of the transcription factors were determined by EMSA as described previously (Kleinert et al., 1998). EA.hy 926 cells were treated with 5 μM AVE9488, and nuclear proteins were extracted. Ten micrograms of nuclear protein was incubated with ³²P-labeled double-stranded oligonucleotide containing either of the following binding motifs from the human eNOS promoter (Karantzoulis-Fegaras et al., 1999): GATA, 5′-GCTCCCACTTATCAGCCTCAGT-3′ (positions –239 to –218); Sp1/3-like, 5′-TTTAGAGCCTCCCAGCCGGG-3′ (–153 to –134); Elf-1, 5′-AGCCGGGCTTGTTCCTGTC-3′ (–140 to –122); YY1, 5′-TCCCATTGTGTATGGGATA-3′ (–123 to –105); Sp1, 5′-GGATAGGGGCGGGGCGAGG-3′ (–109 to –91); and PEA3, 5′-CTCCCTCTTCCTAAGGAAAAGGCC-3′ (–44 to –20). DNA-protein complexes were analyzed on 5% polyacrylamide gels (buffer consisting of 6.7 mM Tris-HCl, pH 7.5, 3.3 mM Na acetate, and 1 mM EDTA). The gels were dried, and then they were autoradiographed on X-ray film (Kleinert et al., 1998).

siRNA-Mediated Knockdown of Sp1. EA.hy 926 cells stably transfected with the 3.5-kb human eNOS promoter luciferase construct were seeded at a density of 50,000 cells/well in 96-well plates. After 24 h, the cells were washed twice with PBS. Lipofection of siRNA was performed according to the manufacturer's instructions using Lipofectamine 2000 and Opti-MEM I medium (Invitrogen). To knockdown Sp1, commercially available siRNA was used (siGENOME SMARTpool; Perbio-Dharmacon, Schwerte, Germany) at a final concentration of 100 nM. A control siRNA of similar length served as control. After a 6-h incubation, siRNA was removed, and cells were grown in normal growth medium for a further 48 h. Thereafter, cells were treated with AVE3085 and AVE9488 at the indicated concentrations for 18 h for the analysis for eNOS promoter activity.

Western Blotting. After washing once in ice-cold PBS, cells were lysed in Laemmli sample buffer containing a mixture of complementary inhibitors (Complete; Roche Diagnostics, Mannheim, Germany) and the endonuclease Benzonase (Merck, Darmstadt, Germany). The samples were incubated at 37°C for 15 min to reduce viscosity, and then they were denatured at 70°C for 20 min before electrophoresis.

Snap-frozen mice tissues were homogenized in liquid nitrogen using a 6750 Spex-Freezer-mill (SPEX CertiPrep, Metuchen, NJ), and then they were extracted for 1 h on ice with a Tris-SDS lysis buffer (10 mM Tris-HCl, pH 7.4, 1% SDS, and Complete protease inhibitors). After centrifugation at 4°C for 30 min at 100,000g, supernatants were mixed with 5× Laemmli sample buffer, and they were denatured at 70°C for 20 min before electrophoresis.

Electrophoresis and transfer to nitrocellulose membranes were carried out according the manufacturer's instructions using precast Criterion-SDS-polyacrylamide gels (4–15% gradient gels; Bio-Rad, Munich, Germany). Membranes were blocked overnight in TBST (10 mM Tris-HCl, 150 mM NaCl, and 0.05% Tween 20, pH 7.6) containing 5% dry milk powder. After incubation with primary antibodies for 18 h at 4°C in TBST containing 1% bovine serum albumin and three washing steps in TBST (each 10 min), samples were incubated with an alkaline phosphatase-conjugated secondary antibody (sheep anti-rabbit IgG; Zymed/Invitrogen) in TBST and 0.2% bovine serum albumin for 1 h at room temperature. After three additional washing steps, bound antibodies were detected using a chemifluorescence substrate (catalog no. RPN-5785; GE Healthcare, Munich, Germany) on a Fluorimager FI-595 (GE Healthcare).

In some cases (e.g., Sp1 protein expression), detection of secondary antibodies conjugated with near-infrared fluorescent dyes (LI-COR Biosciences, Lincoln, NE) was performed on an ODYSSEE infrared-imaging system according to manufacturer's instructions (LI-COR Biosciences). We observed in general no major differences in quantification using the different fluorescence detection systems. However, the ODYSSEE system allows for the simultaneous detection of two antigens (Sp1 and GAPDH) on the same blot with two secondary antibodies linked to different excitation wavelengths of λ = 680 and 800 nm.

The following primary antibodies were used: a polyclonal rabbit antibody against eNOS (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), a polyclonal rabbit antibody against Sp1 (Abcam plc, Cambridge, UK), a polyclonal rabbit antibody against β-tubulin (Santa Cruz Biotechnology) and a monoclonal mouse antibody against GAPDH (Millipore GmbH, Schwalbach, Germany).

RNA Isolation and Real-Time RT-PCR. RNA was isolated from HUVEC using the RNeasy Mini kit (QIAGEN GmbH). Quantitative real-time RT-PCR analysis for eNOS mRNA expression was performed using an iCycler iQ System (Bio-Rad, Munich, Germany) and the QuantiTect SYBR Green PCR kit (QIAGEN GmbH). Primers for human eNOS were 5′-CTGCACTATGG-AGTCTGCTC-3′ (sense) and 5′-AGCCCTTTGCTCTCAATG-3′ (antisense). eNOS mRNA expression was normalized to house-keeping gene GAPDH [primers 5′-CAACGGATTTGGTCGTATT-3′ (sense) and 5′-ATATTGGAAC-ATGTAAACCATGTA-3′ (antisense)].

mRNA expression of other genes was analyzed with the QuantiTect Probe RT-PCR kit (QIAGEN GmbH). TaqMan gene expression assays (predesigned probe and primer sets) were obtained from Applied Biosystems (Foster City, CA): Nox1 (assay ID Mm00549170_m1), Nox2 (assay ID Mm00432775_m1), Nox4 (assay ID Mm00479246_m1), p22phox (assay ID Mm00514478_m1), CuZn superoxide dismutase (SOD1; assay ID Mm01344233_g1), mitochondrial superoxide dismutase (SOD2; assay ID Mm00449726_m1), extracellular SOD (SOD3; Mm00448831_s1), GTP-cyclohydrolase I (GCH1; assay ID Mm00514993_m1), α1- and β1-subunit of the soluble guanylate cyclase (sGC; assay ID Hs00168325_m1 and Hs00168336_m1, respectively), and TATA box binding protein (for normalization; assay ID Hs00427620_m1).

Cuff-Induced Neointima Formation. All animal experiments were performed in accordance with the German animal protection law and the guidelines for the use of experimental animals as given by the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.

Only male animals were included into the study. Neointima was induced by placing a nonocclusive polyethylene cuff around the left femoral artery in apoE-KO mice and eNOS-knockout (eNOS-KO) mice as described previously (Moroi et al., 1998). Both of these mice strains have been backcrossed to a C57BL/6J background to ensure genetic homogeneity. Although neointima formation can be induced in normal mice, apoE-KO mice were used in these experiments to ensure comparability with the subsequent atherosclerosis study. Three days before surgery, the animals were randomized into a placebo group and a group receiving AVE9488 via gavage (10 mg/kg/day b.i.d.) over the whole study period (17 days). Before the surgery, mice were anesthetized with an i.p. injection of 60 mg/kg pentobarbital followed by an i.m. injection of 2 mg/kg xylazine. Two weeks after surgery, neointima formation was analyzed in the femoral artery (hematoxylin and eosin staining) and eNOS protein expression in the aorta (Western blotting).

Atherosclerosis Studies. Male apoE-KO mice at the age of 10 weeks mice were randomized to receive either standard rodent chow (Altromin, Lage, Germany) or chow supplemented with AVE9488 (10 and 30 mg/kg/day) for 12 weeks. At the end of the treatment, mice were anesthetized with an i.p. injection of 60 mg/kg pentobarbital followed by an i.m. injection of 2 mg/kg xylazine. Femoral arteries were taken for analysis of eNOS protein expression. Heart and aorta were fixed in 4% formaldehyde overnight. Thereafter, heart was cut into two halves. The lower half of the heart was discarded, and the upper half was embedded in paraffin. Cross-sections were cut and discarded until the three-valve cusps at the junction of the aorta to the heart became visible. Once this section was located, 18 cross-sections (1 μm) were continuously cut, and they were subjected to standard hematoxylin and eosin staining to determine the average lesion size per section. To determine the lesion area covering aortic surface, aortas were opened longitudinally along a lateral margin, stained with oil-red-O, and mounted on slides with endothelium side up. The stained area was measured in relation to the total aortic surface by an observer who was blinded regarding the different treatment groups using an image analysis computer program (LeicaQWin; Leica, Wetzlar, Germany).

In additional experiments, atherosclerosis studies were performed in male apoE-KO mice and apoE/eNOS-double knockout (DKO) mice. apoE/eNOS-DKO mice were generated by crossing apoE-KO with eNOS-KO mice and by subsequent intercrossing of heterozygous F1 animals. Male mice at the age of 10 weeks were treated with AVE9488 or AVE3085 (30 mg/kg/days, each) applied in Western-type diet (20% fat and 0.5% cholesterol; Altromin) for 12 weeks.

Measurement of Reactive Oxygen Species Production byl-012 Chemiluminescence. Vascular ROS production was determined using the luminol derivate l-012 (Li et al., 2006). In brief, apoE-KO mice were treated for 2 weeks with AVE9488 (30 mg/kg/day pressed in chow). Aortas were isolated and dissected free from remaining connective and fat tissue, and then they were cut into 2-mm rings. These rings were incubated for 30 min at 37°C on 96-well plates in Hanks' buffered salt solution containing 500 μM l-012, with or without l-NAME. l-012-derived chemiluminescence was measured using a microplate luminometer (Berthold Technologies, Bad Wildbad, Germany). The photon counts were normalized for the respective dry weight of aortic rings.

Measurement of Vascular Content of (6R)-5,6,7,8-Tetrahydro-l-biopterin. After treatment of male apoE-KO mice with AVE9488 (30 mg/kg/day) for 2 weeks, aortas were isolated and homogenized in ice-cold lysis buffer (0.1 M Tris-HCl, pH 7.8, containing 5 mM EDTA, 0.3 M KCl, 5 mM 1,4-dithioerythritol, 0.5 mM Pefabloc, and 0.01% saponin). Samples were oxidized under either acidic conditions (with 0.2 M HCl containing 50 mM I₂) or alkaline conditions (with 0.2 M NaOH containing 50 mM I₂). Biopterin content was assessed using high-performance liquid chromatography with fluorescence detection (350-nm excitation, 450-nm emission). BH₄ concentration was calculated as femtomoles per microgram of protein by subtracting the biopterin peak resulting from alkaline oxidation (accounting for 7,8-dihydrobiopterin) from the biopterin peak resulting from acidic oxidation (accounting for both 7,8-dihydrobiopterin and BH₄).

Statistics. Data are expressed as mean ± S.E.M. Data were first analyzed for distribution of values. In case of a Gaussian distribution, statistical differences were measured by one-way analysis of variance followed by Newman-Keuls post-hoc test. In case of non-Gaussian distribution, a nonparametric Kruskal-Wallis test was used. For all statistical analyses, p < 0.05 was considered significant.

Results

Effects of AVE9488 and AVE3085 on eNOS Transcription. In the human endothelial cell line EA.hy 926 stably transfected with a 3.5-kb human eNOS promoter fragment, two small-molecular-weight compounds with related structures, AVE9488 and AVE3085, enhanced eNOS promoter activity in a concentration-dependent manner (Fig. 1A). Simvastatin is an HMG CoA reductase inhibitor that up-regulates eNOS by stabilizing eNOS mRNA, and it has little effect on eNOS promoter activity (Laufs et al., 1998). In our eNOS promoter study, simvastatin was used as a negative control (Fig. 1A). Consistent with previous findings, simvastatin did not change eNOS promoter activity up to 3 μM, and it showed an effect on eNOS promoter activity only at the high concentration (10 μM) (Fig. 1A).

In transient transfection experiments with eNOS promoter fragments progressively deleted from –3500 bp to –263 bp, AVE9488 and AVE3085 displayed an identical activation pattern (Fig. 1B). Even with the smallest eNOS promoter fragment of 263 bp, both compounds remained active (Fig. 1B).

To identify the responsible transcription factor binding sites, we used EMSA analysis with oligonucleotides derived from this 263-bp promoter region of human eNOS: GATA (positions –239 to –218), Sp1/3-like (–153 to –134), Elf-1 (–140 to –122), YY1 (–123 to –105), Sp1 (–109 to –91), and PEA3 (–44 to –20). Significant DNA-protein binding could be observed with all the above-mentioned oligonucleotides. AVE9488 did not change the binding activity of nuclear proteins to Sp1 site or GATA site (Fig. 1, C and D). Binding activity to binding sites of Sp1/3-like, Elf-1, YY1, or PEA3 was not changed by AVE9488 either (data not shown).

Treatment with Sp1 siRNA resulted in a complete knockdown of this transcription factor in EA.hy 926 cells (Fig. 1E), which was associated with a significant decrease of baseline eNOS promoter activity (Fig. 1F). However, the inducing effect by AVE9488 and AVE3085 on eNOS transcription was still present (Fig. 1F), indicating that Sp1 does not contribute to the eNOS-stimulating effect of both compounds. In addition, siRNA-mediated knockdown of transcription factors ETS and GATA2 (with putative binding sites to PEA3 and GATA within eNOS promoter) could not prevent eNOS promoter activation by AVE9488 and AVE3085 (data not shown).

AVE9488 and AVE3085 Did Not Change eNOS mRNA Stability. To determine the effect of AVE9488 and AVE3085 on the stability of eNOS mRNA, EA.hy 926 cells were pretreated with both compounds for 24 h. Then, 60 μM RNA polymerase II inhibitor 5,6-dichlorobenzimidazole riboside (DRB) was added to stop gene transcription. eNOS mRNA was analyzed at 6 or 24 h after DRB. eNOS mRNA showed a half-life of approximately 24 h, which is consistent with previous findings (Li et al., 2004). Neither AVE9488 nor AVE3085 had an effect on eNOS mRNA stability (Fig. 2A). When gene transcription was first stopped by DRB pretreatment and AVE9488 or AVE3085 was added 30 min after DRB, both compounds could no longer increase eNOS mRNA expression (Fig. 2B).

Effects of AVE9488 on eNOS Expression and NO Production in HUVEC. When primary HUVEC were incubated with AVE9488 for 18 h, eNOS mRNA expression was increased, an effect comparable with simvastatin (Fig. 3A), which increases eNOS expression by stabilizing eNOS mRNA (Laufs et al., 1998). AVE9488 increases eNOS protein expression in a concentration-dependent manner (Fig. 3B).

Finally, to examine whether the enhanced eNOS expression resulted in enhanced release of NO, HUVEC pretreated with AVE9488 for 18 h were stimulated acutely with bradykinin. Intracellular cGMP content was measured as an indicator of NO production. Whereas the baseline levels of cGMP were not significantly affected by AVE9488, in bradykinin-stimulated cells, however, AVE9488 treatment significantly enhanced NO production (Fig. 3C). AVE9488 had no effect on expression of sGC subunits (Fig. 3D).

Subchronic Effects of AVE9488 in Mice. AVE9488 was administered for 17 consecutive days to adult C57BL/6J mice at a dose of 30 mg/kg/days pressed in the chow. Then, eNOS protein expression was analyzed by Western blotting in different tissues (Fig. 4). Significantly higher amounts of eNOS protein were detected in aortas and femoral arteries of AVE9488-treated animals compared with controls (Fig. 4).

We then investigated the subchronic effects of AVE9488 in a mouse model of vascular neointima formation, in which the endothelium is left undamaged (Moroi et al., 1998). We applied cuff placements in three different mouse strains, wild-type C57BL/6 and apoE-KO or eNOS-KO mice. Vascular pathologies were most strongly induced in the apoE-KO animals in which neointima became clearly visible consisting of up to five cell layers overlaying the lamina elastica interna (Fig. 5A). Treatment with AVE9488 (10 mg/kg b.i.d., starting 3 days before the cuff placement) for 17 days significantly reduced neointima formation (Fig. 5, B and C). In parallel, an increased eNOS protein expression was detected in the aorta of AVE9488-treated animals (Fig. 5D). When eNOS-knockout mice were subjected to identical conditions of cuff placement, no effect of AVE9488 on neointima formation could be observed (Fig. 5C), indicating that the inhibiting effect of AVE9488 on neointima formation was mediated by eNOS.

Antiatherosclerotic Effect of AVE9488 and AVE3085 in apoE-KO Mice. apoE-KO mice were randomized at the age of 10 weeks to receive either standard rodent chow or chow supplemented with AVE9488 at doses of 10 or 30 mg/kg/day for 12 weeks. In control animals, approximately 13.3% of the aortic surface was covered by atherosclerotic plaques, which is comparable with previous studies (Kauser et al., 2000; Davis et al., 2001). At 30 mg/kg/day, AVE9488 reduced plaque formation to 39% of control (Fig. 6, A and B). The plaque area in the aortic root region close to the aortic valves was 32% lower in AVE9488-treated animals compared with placebo-treated mice (Fig. 6, C and D). Accordingly, eNOS protein expression remained enhanced by AVE9488 treatment to the end of the study, as analyzed by Western blot using protein samples from femoral arteries (data not shown). The 12-week long-term treatment with AVE9488 did not affect plasma lipids or heart rate (Table 1). In our experimental setup, blood pressure tended to be lower in animals treated with AVE9488 (Table 1). Likewise, treatment of apoE-KO mice on Western-type diet with AVE9488 or AVE3085 (30 mg/kg/day, each, 12 weeks) resulted in a reduction (36 and 45%, respectively) of plaque formation (Fig. 6E). In contrast, the antiatherosclerotic effect of both compounds could not be observed in apoE/eNOS-DKO mice (Fig. 6F). The aortic lesion area in placebo-treated apoE/eNOS-DKO mice was 149.4 ± 15.4% (p < 0.05) of placebo-treated apoE-KO mice (set 100%).

Effects of AVE9488 on Vascular ROS Formation and eNOS Functionality in apoE-KO Mice. apoE-KO mice were treated for 2 weeks with the antiatherosclerotic dose of AVE9488, 30 mg/kg/day. Aortas from AVE9488-treated animals showed higher levels of the eNOS cofactor BH₄ in the aorta (Fig. 7A). ROS production in aortic rings was measured with l-012, a compound specifically reacting with superoxide and peroxynitrite, but not with NO (Sohn et al., 1999; Mollnau et al., 2002; Daiber et al., 2004). Aorta of apoE-KO mice showed significant basal ROS formation. This could be partially inhibited by the NOS-inhibitor l-NAME (Fig. 7B), indicating that part of ROS was produced by an uncoupled eNOS. Treatment of apoE-KO mice for two weeks with AVE9488 resulted in a significant reduction in aortic ROS production (Fig. 7B). Under these conditions, l-NAME did not further reduce vascular ROS production in AVE9488-treated animals (Fig. 7B), indicating a recoupling of oxygen reduction and NO synthesis in eNOS.

Effect of AVE9488 on Expression of Genes Important for Vascular BH₄ Content. To determine how AVE9488 increase the vascular content of BH₄, we investigated its influence on the expression of various genes important for anabolism and catabolism of this cofactor. apoE-KO mice were treated orally for 2 weeks with AVE9488 (30 mg/kg/day). Thereafter, RNA from the aortas was isolated, and mRNA gene expression was determined by real-time RT-PCR (Table 2). The rate-limiting step of BH₄ biosynthesis is initial GTP modification catalyzed by the enzyme GCH1. Aortic expression of GCH1 mRNA was not changed significantly by AVE9488. BH₄ is susceptible to oxidation by reactive oxygen species. Among the NADPH-oxidase subunits, we detected significant levels of Nox2, Nox4, and p22phox mRNA in the aorta (Nox1 was undetectable). Aortic expression of these NADPH subunits was not modulated by AVE9488. All three superoxide-dismutating enzyme isoforms, SOD1, SOD2, and SOD3, could be detected in the aortas. Again, none of these factor enzymes was changed by AVE9488 at mRNA level.

Discussion

In the present study, we have identified AVE9488 and AVE3085, two novel small-molecular-weight compounds with oral bioavailability, as eNOS up-regulating agents. In human endothelial cells, the two compounds increased the activity of human eNOS promoter both in stable (Fig. 1A) and in transient transfection (Fig. 1B) experiments. The shortest promoter fragment used (263 bp) was still responsive to AVE9488 and AVE3085 (Fig. 1B), indicating that the cis-elements responsible for the transcription activation by the two compounds are located in the proximal 263-bp promoter region.

Within this promoter region, several binding sites have been shown to play an important role in controlling eNOS promoter activity. These include Sp1, Sp1/3-like, GATA, PEA3, Elf-1, and YY1 (Zhang et al., 1995; Karantzoulis-Fegaras et al., 1999). Mutation of Sp1, GATA, or PEA3 sites, for example, resulted in a significant reduction in eNOS promoter activity (Zhang et al., 1995; Cieslik et al., 1998; Karantzoulis-Fegaras et al., 1999). However, in EMSA experiments, binding activity of nuclear proteins to any of these binding sites remained unchanged by AVE9488 (Fig. 1, C and D). siRNA-mediated knockdown of Sp1 in EA.hy 926 cells resulted in reduction of basal eNOS promoter activity, but this could not prevent eNOS promoter activation by AVE9488 or AVE3085 (Fig. 1, E and F). Knockdown of GATA2 or ETS (which bind to GATA and PEA3 sites, respectively) could not prevent eNOS transcription activation by the two compounds either (data not shown). These results indicate that Sp1, GATA2, and ETS are not required for the effect of AVE9488 and AVE3085 on eNOS expression. The responsible transcription factors/cis-elements still remain to be identified.

As shown above, AVE9488 and AVE3085 stimulate eNOS transcription. These compounds are unlikely to be nonspecific, broad-band gene regulators. In a microarray experiment performed in EA.hy 926 cells with the human genome U133 set (which contains approximately 33,000 genes/ESTs; Affymetrix, Santa Clara, CA), only 24 genes/ESTs were up-regulated (>2-fold) and 31 genes/ESTs were down-regulated (>50%) by 10 μM AVE9488 for 18 h (data not shown). In a series of cellular transcription factor assays, AVE9488 and AVE3085 did not activate liver-X-receptor isoforms α and β), retinoid acid receptor isoforms α and β, or retinoid-X-receptor isoforms α and β (data not shown).

AVE9488 and AVE3085 had no effect on eNOS mRNA stability (Fig. 2). Thus, both compounds seem to be pure eNOS transcription enhancers. Activation of eNOS transcription by AVE9488 resulted in enhanced eNOS mRNA (Fig. 3A) and protein (Fig. 3B) expression. The efficacy of AVE9488 was comparable with that of simvastatin (Fig. 3A), which increases eNOS mRNA by stabilizing eNOS mRNA, and it has little effect on eNOS transcription (Fig. 1A) (Laufs et al., 1998). Such an increase in eNOS protein levels may well have biological significance. Treatment of mice with simvastatin, for example, resulted in a stroke protection which was mediated by eNOS up-regulation, because no such protection was seen in eNOS-KO mice (Endres et al., 1998).

AVE9488 increased eNOS protein expression also in the mouse in vivo (Figs. 4 and 5). Treatment via chow (30 mg/kg/day; Fig. 4) or via gavage (10 mg/kg b.i.d.; Fig. 5) resulted in similar plasma concentrations (data not shown) and a similar extent of eNOS up-regulation.

In mice, in vivo treatment with AVE9488 resulted in inhibition of neointima formation (Fig. 5) and reduced atherosclerosis (Fig. 6). These protective effects observed in vivo are probably mediated by eNOS up-regulation, because they were not seen in eNOS-KO mice (Fig. 5C) or apoE/eNOS-DKO mice (Fig. 6F). AVE9488 had no effects on plasma lipid levels (Table 1). Thus, the effects of AVE9488 on neointima and atherosclerotic plaque formation are unlikely to be related to hyperlipidemia.

Cuff-induced neointima formation is a nonocclusive vessel injury model. In these animals, the endothelium remains intact and eNOS-derived NO has been demonstrated to play a protective role (Moroi et al., 1998). Consistent with this concept, our results show that pharmacological up-regulation of eNOS expression by AVE9488 reduced neointima formation in this model (Fig. 5). apoE-KO mice have been widely used as a model of hyperlipidemia and atherosclerosis (Meir and Leitersdorf, 2004). Treatment with AVE9488 or AVE3085 reduced atherosclerosis in apoE-KO mice (Fig. 6, A–E) but not in apoE/eNOS-DKO mice (Fig. 6F). These results clearly indicate that the antiatherosclerotic effect of both compounds is eNOS-dependent.

An antiatherosclerotic role of endogenous eNOS has been demonstrated in apoE-KO mice (Chen et al., 2001; Kuhlencordt et al., 2001). apoE/eNOS-DKO mice displayed accelerated atherosclerosis, and they developed abdominal aortic aneurysm formation and ischemic heart disease compared with apoE-KO mice. Likewise, pharmacological inhibition of eNOS causes accelerated atherosclerosis in rabbits (Cayatte et al., 1994) and in mice (Kauser et al., 2000). Based on these data, one would expected that overexpression of eNOS protects against atherosclerosis. However, transgenic mice highly overexpressing eNOS on an apoE-KO background developed larger atherosclerotic lesions than apoE-KO mice alone (Ozaki et al., 2002). These data can be easily explained because the apoE-KO mouse is a model of vascular oxidative stress and eNOS tends to become dysfunctional under such pathophysiological conditions (Förstermann and Munzel, 2006). This has been shown in animal models of hypertension or diabetes [where eNOS was also up-regulated (Hink et al., 2001; Mollnau et al., 2002)]. Under pre-existing oxidative stress, oxygen reduction by eNOS uncouples from NO synthesis, and the enzyme becomes a source of superoxide. This has been referred to as eNOS “uncoupling” (Förstermann and Munzel, 2006).

However, eNOS up-regulated by AVE9488 remains functional, and the enhanced eNOS expression is associated with an elevation of bradykinin-stimulated cGMP generation in HUVEC (Fig. 3C). Because AVE9488 had no effect on sGC expression (Fig. 3D), the elevated cGMP content probably indicates increased production of bioactive NO.

Untreated atherosclerotic apoE-KO mice showed a significant ROS production in their aortas, part of which was inhibited by NOS inhibitor l-NAME (Fig. 7B). This is consistent with previous findings (Alp et al., 2004), which demonstrated that eNOS is in an uncoupled state and that it is producing ROS in this pathological model. Treatment with AVE9488 resulted in a marked reduction in aortic ROS production to a level that could not be lowered any further by l-NAME (Fig. 7B). This suggests that eNOS was no longer producing ROS in AVE9488-treated apoE-KO mice, i.e., AVE9488 was able to reverse eNOS uncoupling.

The main reason for eNOS uncoupling is a deficiency of the essential eNOS cofactor BH₄ (Channon, 2004). Supplementation with BH₄ is capable of correcting eNOS dysfunction in several types of pathophysiology (Förstermann and Munzel, 2006). In isolated aortas from prehypertensive spontaneously hypertensive rats, BH₄ supplementation diminished the NOS-dependent generation of superoxide. Administration of BH₄ restored endothelial function in animal models of atherosclerosis, diabetes and insulin resistance, as well as in patients with hypercholesterolemia, diabetes mellitus, essential hypertension, and in chronic smokers (for reviews, see Channon, 2004; Förstermann and Munzel, 2006; Schmidt and Alp, 2007). Oral administration of BH₄ also slowed the progression of atherosclerosis in apoE-KO mice (Hattori et al., 2007).

In the present study, treatment with AVE9488 significantly enhanced vascular BH₄ content (Fig. 7A). This may be an important mechanism for the reversal of eNOS uncoupling in apoE-KO mice (Fig. 7B).

Intracellular BH₄ levels depend on the balance of its de novo synthesis and its oxidation/degradation. BH₄ is synthesized from GTP with GCH1 being the rate-limiting enzyme (Alp et al., 2004; Channon, 2004). BH₄ is one of the most potent naturally occurring reducing agents and susceptible to oxidation by ROS such as peroxynitrite (Laursen et al., 2001). Oxidation of BH₄ due to NADPH oxidase-mediated vascular oxidative stress may represent a major cause of BH₄ deficiency in many cases (Landmesser et al., 2003; Förstermann and Munzel, 2006). Suppression of oxidative stress by down-regulating the expression or activity of vascular NADPH oxidase has been shown to increase vascular BH₄ levels (Li et al., 2006) and restore eNOS functionality (Hink et al., 2001; Mollnau et al., 2002; Li et al., 2006).

Treatment of AVE9488 resulted in elevated levels of vascular BH₄ (Fig. 7A). However, mRNA expression of the BH₄-generating enzyme GCH1 was not increased in response to AVE9488 treatment (Table 2). The expression of the major superoxide-producing enzyme, NADPH oxidase, was also not changed. In addition, the superoxide-depredating enzymes (SOD1, SOD2, and SOD3) were not changed (Table 2). Thus, the mechanisms for the increase in BH₄ by AVE9488 still remain unknown. In cultured EA.hy 926 cells, AVE9488 and AVE3085 had no effect on BH₄ content, whereas sepiapterin (a precursor of BH₄ synthesis via the salvage pathway) significantly increased BH₄ content (data not shown). The fact that AVE9488 increased vascular BH₄ content when administered in vivo, but not in endothelial cells in vitro, suggests that the elevation of BH₄ in vivo may be indirect (e.g., by an action of AVE9488 on cell types other than endothelial cells).

In conclusion, we have identified AVE9488 and AVE3085 as two novel small-molecular-weight compounds with vasoprotective properties in experimental atherosclerosis. The beneficial vascular effects of these compounds probably result from a combination of eNOS up-regulation and a reversal of eNOS uncoupling. Such compounds have therapeutic potentials for the treatment of cardiovascular diseases.