Ethyl Acetate Fraction and Isolated Phenolics Derivatives from Mandevilla moricandiana Identified by UHPLC-DAD-ESI-MSn with Pharmacological Potential for the Improvement of Obesity-Induced Endothelial Dysfunction

Endothelial dysfunction in obesity plays a key role in the development of cardiovascular diseases, and it is characterized by increased vascular tonus and oxidative stress. Thus, this study aimed to investigate the vasodilatory and antioxidant activities of Mandevilla moricandiana ethyl acetate fraction and subfractions. Vascular effects were investigated on aorta isolated from control and monosodium glutamate (MSG) induced-obese Wistar rats, and antioxidant activity was assessed by 2,2-diphenyl-1-picrylhydrazyl (DPPH) and oxygen radical absorbance capacity (ORAC) methods. The ethyl acetate fraction (MMEAF) induced a concentration-dependent vasodilation on aortic rings through the NO pathway, with the involvement of histamine H1 and estrogen ERα receptors and showed potent antioxidant activity. In aorta of MSG obese rats, maximal relaxation to acetylcholine was increased in the presence of MMEAF (3 µg/mL), indicating that MMEAF ameliorated obesity-induced endothelial dysfunction. Quercetin and kaempferol aglycones and their correspondent glycosides, as well as caffeoylquinic acid derivatives, A-type procyanidin trimer, ursolic and oleanolic triterpenoid acids were identified in subfractions from MMEAF and seem to be the metabolites responsible for the vascular and antioxidant activities of this fraction.


Introduction
Cardiovascular diseases (CVD) are the leading cause of death worldwide, being responsible for 17.9 million deaths in 2019 [1]. Endothelium dysfunction is involved in both the development and progression of CVD and is also present in risk factors for these diseases. It plays a key role in structural and functional vascular changes observed in atherosclerosis [2,3], coronary artery disease [4], hypertension [5,6], and obesity [7,8]. The increasing incidence of obesity represents a global health problem since it is commonly

Plant Material and Extraction
The leaves of Mandevilla moricandiana were collected at the Restinga de Jurubatiba National Park (22 •  botanical identification of the vegetal species was performed by Tatiana Ungaretti Paleo Konno and a voucher specimen has been deposited at the Herbarium of the Federal University of Rio de Janeiro, under the number RFA38748. Leaves (99 g) were air-dried at 40 • C, and then exhaustively extracted with ethanol/H 2 O (7:3) by static maceration at room temperature. The hydroalcoholic crude extract of the leaves from M. moricandiana (MM) was obtained after the removal of the solvent under reduced pressure by a rotatory evaporator. The MM extract was solubilized in 350 mL of MeOH/distilled water (9:1) under agitation and was submitted to successive liquid-liquid partitions with organic solvents in the following eluotropic order: n-hexane (MMHF), dichloromethane (MMDCMF), ethyl acetate (MMEAF), and n-butanol (MMBF). After partition with n-butanol, the remaining aqueous fraction (MMAF) was obtained, and it was submitted to freezing followed by lyophilization (Freeze Dryer: L101 Liotop, São Paulo, Brazil), as previously described by our group [31].

Ultra-High Performance Liquid Chromatography Coupled to Diode Array Detector and Electrospray Ionization Mass Spectrometry (UHPLC-DAD-ESI-MS n )
The UHPLC-DAD-ESI-MS n analysis was performed on a Thermo LCQ FLEET system (Thermo Fisher Scientific, Waltham, CA, USA) consisting of a FINNIGAN Surveyor MS Pump Plus, a FINNIGAN Surveyor Autosampler Plus, a dual detection system of a photodiode array detector (PDA), and a LCQ FLEET ion trap mass spectrometer equipped with an electrospray ionization (ESI) source (LCQ, Thermo Fisher Scientific, Waltham, CA, USA). Thermo Xcalibur version 2.2 software (Waltham, USA, 2011) was used for data acquisition and processing. The chromatographic condition established was: RP18 end capped column (2 µm) (Hibar HR 100 × 2.1 mm Purospher Star ® ); sample injection volume, 20 µL; the temperature of column oven, 25 • C; flow rate, 0.2 mL/min. The mobile phases adopted were formic acid 0.1% (A) and acetonitrile + formic acid 0.1% (B). A gradient elution protocol was used as follows: 0-10 min, 5-15% B; 10-20 min, 15-25% B; 20-35 min, 25-50% B; 35-50 min, 50-100% B; 50-55 min, 100% B; 55-60 min, 100-5% B; 60-65 min, 5% B. The PDA acquisition wavelength was set in the range of 190 to 600 nm. The samples were prepared through the dilution of 2 mg in 1000 µL methanol (2 mg/mL), and the chromatography runs were monitored at 190, 210, 254, and 365 nm wavelengths. The quantitative analyses were based on the peak relative area percentage. The mass spectrometer equipped with an electrospray ionization (ESI) source was operated in the negative ionization mode (ESI-(-)) with the final conditions as follows: electrospray voltage 4.5 kV and 5 kV, capillary temperature 350 • C, capillary voltage ± 24 V. Nitrogen was used as sheath and auxiliary gas. Full scan data acquisition was performed from m/z 120 to 1500 in negative MS scan mode. MS/MS experiments were performed by the collision of the precursor ions with helium gas. The collision energy value was set at 30% of the instrument maximum.

Nuclear Magnetic Resonance (NMR)
NMR spectra of the isolated substances were recorded on a VARIAN INOVA 500 MHz spectrometer (Varian Inc., Palo Alto, CA, USA) working at 499.78 ( 1 H) and 125.68 MHz ( 13 C) at the Institute of Natural Product Research (IPPN)-UFRJ/RJ; on a 400 MHz BRUKER Model: Avance II HD spectrometer equipped with a Bruker Prodigy BB0400S1 cryosonday or on a 400 MHz spectrometer at the Analytical Methodology Platform (PMA)-Farmanguinhos-Fiocruz/RJ. Bidimensional spectra (HSQC, HMBC, COSY) also proceeded. Chemical shifts (δ) were endorsed with tetramethylsilane (TMS) as internal standards (δ = 0, 1H) being expressed in ppm units and coupling constants (J) in Hertz (Hz). The pulse sequences used are all standard in the VNMRJ software, and the experiments were conducted at 25 • C. The samples were dissolved in deuterated methanol (CD 3 OD).

Animals
All experimental protocols were approved by the Animal Care and Use Committee at the Federal University of Rio de Janeiro under the license MAC019. Male Wistar rats were maintained on a 12 h light/dark cycle with controlled temperature (23 ± 2 • C) and free access to food and water.

Preparation of Aortic Rings for Isometric Tension Recording
Thoracic aortas were dissected from male Wistar rats (200-250 g). After the removal of connective tissue, the aorta was cut into 2-3 mm rings, which were suspended in organ baths filled with Krebs-Henseleit solution (118 mM NaCl; 4.7 mM KCl; 1.2 mM KH 2 PO 4 ; 1.2 mM MgSO 4 ; 2.5 mM CaCl 2 ; 25 mM NaHCO 3 , and 11 mM glucose; pH 7.4; 37 • C) continuously gassed with carborgen gas (95% O 2 , 5% CO 2 ). Aortic rings were mounted between two hooks, in which one was attached to a force transducer (MLT0201; AD Instruments, Bella Vista, New South Wales, Australia). Signals were digitalized (Power Lab 4/30; AD Instruments, Bella Vista, New South Wales, Australia) and stored on a computer for analysis using the software LabChart Pro (AD Instruments, Bella Vista, New South Wales, Australia). After an equilibrium period of 90 min under 1 g resting tension, aortic rings were contracted with phenylephrine (Phe; 10 µM), and the presence of functional endothelium was confirmed by a relaxation response to acetylcholine (ACh; 10 µM) greater than 80%. In some rings, the endothelium was mechanically removed, which was confirmed by the lack of relaxation in response to ACh [31].

Vasodilatador Effect
Concentration-response curves for MMEAF, subfractions, and isolated compound (0.1-30 µg/mL) were obtained in aortic rings with endothelium precontracted with 10 µM Phe. In order to determine the mechanisms involved in MMEAF-induced vasodilation, concentration-response curves for MMEAF were obtained in aortic rings without endothelium, or in aortic rings with endothelium pretreated (15 min) with the NO synthase inhibitor L-NAME (100 µM), the histamine H1 receptor antagonist diphenhydramine (10 µM), or the nonselective estrogen receptor antagonist fulvestrant (10 µM). The relaxation response was expressed as the percentage of maximal contraction induced by Phe, and the concentration necessary to induce 50% of maximal response (EC 50 ) was calculated by non-linear regression analysis.

Effects of MMEAF on Vascular Reactivity of Aorta from Obese Rats
Obesity was induced in male Wistar rats through subcutaneous injections of MSG (4 g/kg body weight: MSG group) during the first 5 days of life. Control (CTL) group received hyperosmotic saline solution (1.25 g/kg body weight). Body weight and food intake were determined once a week. After euthanasia, perigonadal and retroperitoneal fat pads were collected and weighed. Obesity was confirmed by the increase in fat stores, despite the reduction in body weight, as described in our previous publications where MSG-induced obesity was used [34,35]. All MSG rats used in this study presented these characteristics. The thoracic aorta was dissected and prepared for isometric tension recording, as previously described in Section 2.8. In order to evaluate vascular reactivity, concentration-response curves for Phe (10 −9 -10 −5 M) and ACh (10 −9 -10 −5 M) were obtained in the absence and presence of 3 µg/mL of MMEAF. EC 50 was calculated by non-linear regression analysis and expressed as pEC 50 (negative log EC 50 in Molar).

2,2-Diphenyl-1-picrylhydrazyl (DPPH) Assay
The DPPH scavenging activity of MMEAF and subfractions was measured according to [36], with minor modifications. In 96-well microtiter plates, 125 µL of each methanolic solution of MMEAF and the correspondent fractions (1-200 µg/mL) was mixed with 50 µL of 300 µM DPPH methanolic solution and kept in the dark for 30 min at room temperature. Then, absorbances were measured at 517 nm using a spectrophotometer (UVM 340 spectrophotometer, Biochrom ASYS, Cambridge, UK). The blank consisted of 125 µL of samples and 50 µL of methanol, and negative control was 50 µL of DPPH and 125 µL of methanol. The standardized leaf extract of Ginkgo biloba (EGb761 ® ; 1-200 µg/mL) and Trolox (1-200 µg/mL) were used as positive controls. Three independent experiments were performed in triplicate for each sample, and final results were expressed as a percentage of antioxidant activity (AA%). EC 50 was calculated by non-linear regression.

Oxygen Radical Absorbance Capacity (ORAC) Assay
The ORAC of MMEAF and subfractions was measured according to [37], with minor modifications. To 20 µL of MMEAF or subfractions (0.25-2 µg/mL) was added 120 µL of 116.7 µM fluorescein solution in 96-well black plates. After a 10 min-period of incubation at 37 • C, 60 µL of 40 mM AAPH solution was added, and fluorescence was recorded at 480/530 nm, for 95 min at 10 min intervals by using a fluorimeter (Fluostar Optima; BMG Labtechnologies GmbH, Offenburg, Germany). Trolox (0.25-2 µg/mL) was used as positive control. The blank consisted of 80 µL of phosphate-buffered saline (PBS) and 120 µL of fluorescein, and negative control consisted of 20 µL of PBS, 60 µL of AAPH, and 120 µL of fluorescein. Three independent measures in triplicate were performed for each sample. Final results were expressed as mmol of Trolox equivalent (TE)/g of the sample by using the standard curves.

Statistical Analysis
Data are expressed as means ± S.E.M. Analyses were performed using Prism 5.0 software (GraphPad Software, La Jolla, CA, USA). Data were analyzed using a one-way analysis of variance followed by Tukey's post-hoc test. Differences between groups were considered statistically significant when p < 0.05.

Chemical Analysis of MMEAF and Its Subfractions
MMEAF was analyzed by UHPLC-DAD-ESI-MS n , and the chromatogram wavelength monitored was set at 254 nm, as it is shown in Figure 1. Two constituents were observed as major compounds, such as that at 14.65 min (λ 258 and 352 nm) (41.76%) and at 16.70 min (λ 266 and 347 nm) (27.02%), with UV spectra characteristic of the benzyl (ring A) and cinnamoyl (ring B; C2-C3) portions, respectively, from flavonoid chemical structures. Minority compounds were also observed in 4.57 and 14.19-25.52 min, also with phenolic absorption characteristics, which will be discussed further in detail.

Chemical Analysis of MMEAF and Its Subfractions
MMEAF was analyzed by UHPLC-DAD-ESI-MS n , and the chromatogram wavelength monitored was set at 254 nm, as it is shown in Figure 1. Two constituents were observed as major compounds, such as that at 14.65 min (λ 258 and 352 nm) (41.76%) and at 16.70 min (λ 266 and 347 nm) (27.02%), with UV spectra characteristic of the benzyl (ring A) and cinnamoyl (ring B; C2-C3) portions, respectively, from flavonoid chemical structures. Minority compounds were also observed in 4.57 and 14.19-25.52 min, also with phenolic absorption characteristics, which will be discussed further in detail.  Figure 2). UV spectra and MS data of each main peak allowed the identification of flavonols (aglycones and glycosides), caffeoylquinic acid derivatives, procyanidin, and pentacyclic-type triterpenes, as shown in Figure 3 and Table 1.  Figure 2). UV spectra and MS data of each main peak allowed the identification of flavonols (aglycones and glycosides), caffeoylquinic acid derivatives, procyanidin, and pentacyclic-type triterpenes, as shown in Figure 3 and Table 1.         7), respectively. The glycosylation site of 3-OH in astragalin (7) was indubitably proved by NMR spectroscopic data, as will be discussed. In addition to these, it was also identified a peak at Rt = 19.  From MMEAF-F, quercetin (9) (Rt = 21.64 min) (90% relative area) was solely found, with UV max absorption at 267 and 370 nm.

NMR Data
From MMEAF-C, the same substances as observed in MMEAF-D were detected but in different amounts concerning the percentage of relative area. Kaempferol (10) was detected as the major compound in MMEAF-D.
The fractions MMEA-F-B and MMEA-F-D were purified by semi-preparative HPLC, as described in the methods section, and the isolated compounds were structurally elucidated by 1 H-NMR, 13 C-NMR, HSQC, and HMBC spectroscopic data, as it is described below.

Vasodilator Effect of MMEAF
MMEAF induced a concentration-dependent relaxation of aortic rings with endothelium, with an EC 50 of 6.91 ± 1.26 µg/mL. Maximal relaxation (30 µg/mL) was 86.56 ± 2.97%. Mechanical removal of endothelium completely inhibited the vasodilation induced by MMEAF, indicating that the effect depends on endothelium-derived factors ( Figure 5A).

Vasodilatory Activity of MMEAF Subfractions and Isolated Compound
All subfractions obtained from the column chromatography of MMEAF were evaluated in aortas with endothelium. MMEAF-A and MMEAF-B showed no vasodilator effect at all concentrations tested. Instead, they produced a small increase in vascular tonus. Subfractions C-H induced intense vasodilation of aortic rings ( Figure 6A). MMEAF-D was less effective and potent when compared to the other subfractions ( Table 2).
The vasodilator effect of the A-type procyanidin trimer identified in subfractions C, D, and E was also evaluated. This compound induced a concentration-dependent relaxation of aorta with endothelium, with maximal relaxation of 75.77 ± 4.39% ( Figure 6B; Table  2). As can be seen in Table 2, the isolated compound presented a greater maximal effect than MMEAF-D, but not statistically different from MMEAF-C and E. In contrast, the trimer was less potent than these two subfractions, suggesting that quercetin especially contributes to the higher potency of MMEAF-E, possibly through additive/synergistic actions We have previously described that the hydroalcoholic extract of leaves of emphM. moricandiana induces endothelium-dependent vasodilation through the NO pathway, with the participation of histamine H1 and estrogen receptors [31]. Thus, the investigation of the mechanism of action of MMEAF was focused on this.
In the presence of L-NAME, the vasodilator effect of MMEAF was completely inhibited, indicating that the effect is dependent on NO release ( Figure 5A). The involvement of histamine H1 and estrogen receptors in the activation of NO production was investigated through the pretreatment of aortic rings with diphenhydramine and fulvestrant, respectively. In the presence of diphenhydramine, the maximal effect of MMEAF was reduced to 51.33 ± 4.87% relaxation, and EC 50 was increased to 26.41 ± 4.45 µg/mL (p < 0.05; Figure 5B). Pretreatment with fulvestrant did not alter maximal vasodilator response but induced a rightward displacement of the concentration-response curve of MMEAF, increasing EC 50 to 13.94 ± 3.44 µg/mL (p < 0.05; Figure 5B).
These results indicate that the vasodilation induced by MMEAF is mediated, at least in part, through the activation of endothelial histamine H1 and estrogen receptors. Agonists of histamine H1 receptors lead to eNOS activation through the increase in intracellular calcium, while estrogen receptors signaling involves protein kinase B (Akt)-mediated Ser 1177 phosphorylation of eNOS [43].
Estrogen receptors are involved in the vascular effects of some polyphenols, such as resveratrol and delphinidin, found in red wine and black currant. These compounds induce the endothelial production of NO through the activation of estrogen ERα receptors [44][45][46]. Histamine H1 receptors have been implicated in the mechanism of vasodilation of the lignan eudesmin [47] and the extracts of Cirsium japonicum [48] and black currant [49].

Vasodilatory Activity of MMEAF Subfractions and Isolated Compound
All subfractions obtained from the column chromatography of MMEAF were evaluated in aortas with endothelium. MMEAF-A and MMEAF-B showed no vasodilator effect at all concentrations tested. Instead, they produced a small increase in vascular tonus. Subfractions C-H induced intense vasodilation of aortic rings ( Figure 6A). MMEAF-D was less effective and potent when compared to the other subfractions (Table 2).  MMEAF-A, in which was identified chlorogenic acid, did not show vasodilator effect. Chlorogenic acid was shown to induce endothelium-dependent vasodilation of rat aorta, with the involvement of NO, prostacyclin, and endothelium-derived hyperpolarizing factor [54]. The lack of vasodilator effect of MMEAF-A may be related to the low con-  The vasodilator effect of the A-type procyanidin trimer identified in subfractions C, D, and E was also evaluated. This compound induced a concentration-dependent relaxation of aorta with endothelium, with maximal relaxation of 75.77 ± 4.39% ( Figure 6B; Table 2). As can be seen in Table 2, the isolated compound presented a greater maximal effect than MMEAF-D, but not statistically different from MMEAF-C and E. In contrast, the trimer was less potent than these two subfractions, suggesting that quercetin especially contributes to the higher potency of MMEAF-E, possibly through additive/synergistic actions with the trimer and other compounds.
As far as we know, this is the first report describing the vasodilator effect of an A-type procyanidin trimer. Procyanidins are polyphenols widely distributed in fruits, vegetables, nuts, and grains, where they are found as a mixture of oligomers [50]. Procyanidins have been shown to induce endothelium-dependent vasodilation, mainly via NO release [51,52], and the B-type procyanidin trimer C1 was shown to induce NO production in rat endothelial cells [53]. The biological effects of procyanidins vary according to the oligomeric, dimeric, and monomeric forms, highlighting the importance of the study of the different procyanidin compounds [50,53].
MMEAF-A, in which was identified chlorogenic acid, did not show vasodilator effect. Chlorogenic acid was shown to induce endothelium-dependent vasodilation of rat aorta, with the involvement of NO, prostacyclin, and endothelium-derived hyperpolarizing factor [54]. The lack of vasodilator effect of MMEAF-A may be related to the low concentration of this polyphenol in this subfraction.
Kaempferol was identified in subfractions C, D, and E, and quercetin was identified in subfractions E, F, and G. These flavonols are associated with cardiovascular protective effects due to many biological activities, including vasodilatory, antioxidant, and antiinflammatory [58]. Quercetin has been shown to induce vasodilation in isolated vessels through endothelium-dependent [59] and -independent mechanisms [60], while the vasodilatory activity of kaempferol is partially dependent on endothelium [61,62]. Quercetin was the main compound found in subfraction F, which produced potent and intense relaxation of aortic rings with endothelium.
Oleanolic acid, which was identified in many subfractions of MMEAF with vasodilator effect, induces NO-dependent relaxation of rat superior mesenteric arteries [63] and aorta isolated from Wistar [64] and spontaneously hypertensive rats [65]. Similar effects are reported for ursolic acid in rat aorta [25,66].

Antioxidant Acitivity of MMEAF and Its Subfractions
Antioxidant activity of MMEAF and its subfractions was evaluated by using DPPH and ORAC assays, where Ginkgo biloba extract EGb761 ® and quercetin were used as positive controls. All subfractions showed high antioxidant potential (Table 3).
For the antioxidant activity, it is known that the presence of hydroxyl groups in the aromatic rings A and B are extremely important for the antioxidant activity [70]. However, in addition to the presence of the hydroxyl groups, their position also influences the antioxidant activity, which can be attributed to the presence of the hydroxyl groups at the C3, C3', and C4' positions [71]. In most plants, flavonoid glycosides occur at the C3 position, such as 3-O-glycosides [71]. The substitution of the hydroxyl group in C3 by sugars can reduces the antioxidant activity of flavonoids [72]. These data are in agree-ment with the results presented by the MMEAF-B fraction, which contains isoquercitrin and astragalin as major compounds, totaling 65.27% percentage of relative area, in addition to other glycosides, and presented lower antioxidant activity by the ORAC method (0.85 ± 0.05 mmol Eq Trolox/g sample). Among the fractions, by this same method, MMEAF-A presented the best activity, being statistically equal to the control quercetin. This fraction is also mainly composed of glycosides, such as kaempferol-di-3-glycoside and kaempferol-7-O-glycoside, but also by chlorogenic acid. By the DPPH method, the most active fraction was MMEAF-F, which is mainly composed of quercetin (90%), which justifies the activity being quite similar (no statistical difference) to the commercial control quercetin, followed by MMEAF-E, composed mainly of kaempferol (29%). The aglycone quercetin has a catechol group in the B ring, that is, OH groups in 3,4 positions in the molecule of the phenolic compound. In this case, the hydroxyl groups in the ortho position could allow the formation of two hydrogen bonds with the two oxygen atoms of the peroxyl radical, and it led to a compact reactant complex, enhancing antioxidant activity. In addition to the presence of hydroxyls, another important characteristic for the antioxidant activity is the presence of the C2-C3 conjugated double bond with a carbonyl group in C4 [71]. It is important to highlight that all flavonoids identified were from the group of the flavonols, so they were constituted by the conjugated double bond in C2-C3, and this characteristic cannot be correlated to the different activity observed.

Effects of MMEAF on Vascular Reactivity of Aorta from Obese Rats
Since MMEAF presented both vasodilatory and antioxidant activities, we hypothesized that this dual effect could improve endothelial dysfunction. Thus, we evaluated the effects of MMEAF on vascular reactivity of aorta isolated from obese rats. Obesity was induced by early postnatal treatment with MSG, when the blood-brain barrier is not formed, which results in the development of hypothalamic obesity. This model presents common characteristics of clinical obesity, such as increased visceral adiposity, insulin resistance [73], systemic oxidative stress, reduced NO endothelial production, and endothelium dysfunction [74,75].
Aortic rings from MSG group showed reduced endothelium-dependent vasodilation to ACh compared to the control group, indicating the presence of endothelium dysfunction, as we have previously shown [34,35]. Maximal relaxation in response to ACh was 76.91 ± 2.94% and 31.60 ± 1.85% in control and MSG aorta, respectively (p < 0.05). Although maximal contraction induced by Phe was not altered, pEC 50 was significantly altered in aorta from MSG rats (Figure 7; Table 4). aorta from MSG rats (Figure 7; Table 4).
Incubation of aortic rings with 3 µg/mL MMEAF for 15 min did not change vascular reactivity of aorta from control rats or contractile response to Phe of aorta from MSG rats (Figure 7; Table 4). On the other hand, it significantly increased ACh-induced vasodilation in aorta from obese rats to 63.34 ± 5.36%, indicating that MMEAF improved obesity-induced endothelium dysfunction ( Figure 7B; Table 4).   Incubation of aortic rings with 3 µg/mL MMEAF for 15 min did not change vascular reactivity of aorta from control rats or contractile response to Phe of aorta from MSG rats ( Figure 7; Table 4). On the other hand, it significantly increased ACh-induced vasodilation in aorta from obese rats to 63.34 ± 5.36%, indicating that MMEAF improved obesityinduced endothelium dysfunction ( Figure 7B; Table 4).
The effect of MMEAF was only observed in aorta from obese rats, where NO bioavailability was reduced. In an ambient of overproduction of reactive oxygen species, NO reacts with superoxide anion to form peroxynitrite, which oxides the eNOS cofactor tetrahydrobiopterin, leading to eNOS uncoupling and reduced NO bioavailability [13]. Thus, the improvement of ACh-induced vasodilation in aortic rings from obese rats may be related to the free radical scavenging activity of MMEAF. In addition, the activation of NO production in aorta could contribute to the effect of MMEAF. Considering the important role of the interaction between oxidative stress and inflammation in obesity-induced endothelial dysfunction, further studies are necessary to investigate if MMEAF could also reduce vascular inflammation.
The phytochemical analysis of MMEAF revealed the presence of different phenolic compounds that have protective vascular effects, which could additively and/or synergistically contribute to the effects of this fraction on aorta from obese rats. Studies have shown that polyphenolic mixtures present synergistic and additive antioxidant and vascular effects [76][77][78][79]. In this respect, quercetin and kaempferol synergistically acted in the reduction in H 2 O 2 -induced oxidative stress [80].
Kaempferol was shown to protect against paraquat-induced vascular injury by reducing oxidative stress and inflammation [81], and to reduce atherosclerotic formation and endothelial dysfunction in apolipoprotein E-deficient mice [82]. However, there are no reports regarding the effects of the kaempferol glycosides identified in MMEAF on endothelial dysfunction.
Quercetin and kaempferol presented anti-inflammatory activity in cytokine-stimulated human umbilical vein endothelial cells (HUVECs), where both flavonols inhibited the expression of adhesion molecules and cyclooxygenase-2 [83]. Quercetin reduced NO and anion superoxide imbalance induced by angiotensin II in HUVECs [84] and rat aorta [85], as well as restored endothelial function in spontaneously hypertensive rats [86] and in deoxycorticosterone acetate-salt hypertensive rats treated with this flavonol [87]. Pretreatment of aortic rings isolated from streptozotocin-induced diabetic rats with quercetin improved the relaxation response to ACh, which was mediated by its free radical scavenging activity [88]. The quercetin glycoside isoquercetrin was shown to inhibit high glucose-induced endothelial apoptosis in HUVECs [89].
Oleanolic acid improved endothelium-dependent vasodilation in high glucose treatedaorta, as well as restored NO production in HUVECs exposed to high glucose [90]. Ursolic acid treatment increased serum NO and prostacyclin levels and improved morphological parameters of aorta isolated from high choline-fed mice [91].

Conclusions
In this work, we investigated the phytochemical and pharmacological properties of MMEAF in order to evaluate its potential as a source of phenolic compounds with vascular protective actions. Our results showed that MMEAF is a polyphenol-enriched fraction, specially constituted by the flavonols kaempferol and quercetin and their glycosides, as well as phenylpropanoid derivatives, with potent antioxidant activity and NO-mediated vasodilator effect. Moreover, MMEAF restores endothelium-dependent vasodilation in aortas from MSG-induced obese rats, indicating its pharmacological potential for conditions associated with endothelium dysfunction. This effect seems to be mediated, at least in part, by the antioxidant effect of MMEAF and the consequent reduction in vascular oxidative stress. We also described, for the first time in this vegetal species, an A-type (epi)catechin trimer with vasodilatory activity, that should be investigated on deep concerning the effect observed.