Salvia elegans, Salvia greggii and Salvia officinalis Decoctions: Antioxidant Activities and Inhibition of Carbohydrate and Lipid Metabolic Enzymes

Salvia elegans Vahl., Salvia greggii A. Gray, and Salvia officinalis L. decoctions were investigated for their health-benefit properties, in particular with respect to antioxidant activity and inhibitory ability towards key enzymes with impact in diabetes and obesity (α-glucosidase, α-amylase and pancreatic lipase). Additionally, the phenolic profiles of the three decoctions were determined and correlated with the beneficial properties. The S. elegans decoction was the most promising in regard to the antioxidant effects, namely in the scavenging capacity of the free radicals DPPH•, NO• and O2•–, and the ability to reduce Fe3+, as well as the most effective inhibitor of α-glucosidase (EC50 = 36.0 ± 2.7 μg/mL vs. EC50 = 345.3 ± 6.4 μg/mL and 71.2 ± 5.0 μg/mL for S. greggii and S. officinalis, respectively). This superior activity of the S. elegans decoction over those of S. greggii and S. officinalis was, overall, highly correlated with its richness in caffeic acid and derivatives. In turn, the S. officinalis decoction exhibited good inhibitory capacity against xanthine oxidase activity, a fact that could be associated with its high content of flavones, in particular the glycosidic forms of apigenin, scutellarein and luteolin.


Introduction
Salvia genus (Salvia spp.), belonging to the Lamiaceae family, comprises more than 900 species that are used for distinct purposes, including the culinary and cosmetic industries or in traditional medicines due to their claimed health benefits [1,2]. Among them, Salvia officinalis L., i.e., "common sage" or "Dalmatian sage", is widely cultivated. These plants usually grow 30-70 cm tall, with a woody stem, whitish beneath and grayish-green above, and with purple-blue flowers up to 3 cm long appearing from early summer to early autumn [1]. Due to its worldwide spread, S. officinalis has been the most monitored species in relation to the biological potential of the whole plant as well as of its essential oils and polar extracts. For example, promising results were obtained in clinical studies with aqueous or ethanolic extracts of this Salvia species when focused on memory and cognitive functions, pain, and the biochemical profile of glucose and lipids [3]. In addition, in vivo assays in an ear edema induced by croton oil model pointed out the good anti-inflammatory activity of hydroethanolic extracts [4]. Indeed, in vitro experiments demonstrated that its ability to

Phytochemical Composition
The decoction yields of the three Salvia species were approximately 20%, with slightly higher levels observed for S. elegans and S. greggii in comparison to S. officinalis (22.1 ± 2.2% and 22.2 ± 1.5% vs. 19.3 ± 2.3%, respectively). Consistent with their prevalence in Salvia plants [1,2,29], caffeic acid derivatives (particularly rosmarinic acid) were dominant compounds in S. officinalis, S. elegans, and S. greggii decoctions, accounting for about one third of the global identified phenolic species (Table 1, Figure 1, Figure S1). Nevertheless, significant differences could be found between extracts. S. elegans was distinguished by its richness in caffeic acid derivatives, namely rosmarinic acid (peak 36 13.4 ± 0.6 mg/g extract), in addition to rosmarinic acid (28.3 ± 0.6 mg/g extract). Note that the predominance of apigenin-O-glucuronide and rosmarinic acid in this decoction is coherent with the abundance of these two constituents previously reported for polar extracts of this species (e.g., ethanol, methanol, aqueous, hydroalcoholic) [2,30]; however, it is worth noting that this is the first time that scutellarein-O-glucuronide was detected in S. officinalis extracts, while the previous studies only described its aglycone form.    Table 1.
Furthermore, all the extracts were screened for their potency in inhibiting xanthine oxidase. Globally, the S. elegans decoction was more promising than S. greggii in regard to its ability to scavenge free radicals and to reduce Fe 3+ (Table 2). It presented EC 50 values about 1.8-2.5 lower than the latter in DPPH • , NO • , O 2 •-, and reducing power tests, and a tendentially higher ability to capture RO 2 • . Notably, the S. elegans extract also presented tendentially better antioxidant potential than S. officinalis, with tendentially reduced EC 50 values being registered for NO • , O 2 •-, and reducing power tests, and even three times lower for the DPPH • assay. In addition, one must highlight that the potency of S. elegans decoction to counteract DPPH • and NO • was 0.6-and 2.3-fold that of the ascorbic acid, respectively. The only exception was observed for the oxygen radical absorbance capacity (ORAC) assay for which the result observed for S. officinalis was better than that for S. elegans, although it was not statistically significant. Table 2. Antioxidant properties of S. officinalis, S. elegans, and S. greggii decoctions. (4) 404.4 ± 1.80 a 373.1 ± 28.1 a 335.6 ± 69.6 a -Xanthine oxidase (EC 50 µg/mL) (5) 55.1 ± 10.6 a 71.8 ± 3.8 b 70.1 ± 4.0 a,b 0.09 ± 0.01 c (1) Ascorbic acid was used as the reference compound. (2) Amount of extract able to provide 0.5 of absorbance by reducing 3.5 µM Fe 3+ to Fe 2+ . Butylated hydroxyanisole (BHA) was used as a reference compound. (3) Gallic acid was used as the reference compound. (4) TE-Trolox Equivalent. (5) Allopurinol was used as the reference compound. Mean values ± SD; statistical analysis was performed by one-way ANOVA followed by Tukey's test.

S. officinalis S. elegans S. greggii Standard
In each line, different letters mean significant differences (p < 0.05). Table 3 summarizes the correlation coefficients between the amounts of classes of phenolic components found in the Salvia decoctions (caffeic acid and derivatives, coumaric acid derivatives, flavones and flavonols) and the data from the distinct biological experiments. According to these results, it is possible to suggest that the superior antioxidant activity of the S. elegans decoction is strongly associated with its richness in caffeic acid and derivatives, since correlation factors in DPPH • , reducing power, NO • , and O 2 •assays were 0.801, 0.948, 0.986, and 0.844, respectively.
The comparison of the herein gathered data with that previously reported for other solventextracts or other Salvia species is not an easy task, since methodologic adaptations (e.g., radical precursor concentrations and their producing conditions) cause inevitable changes in EC 50 values. This difficulty can be partly overcome by the comparison of the extract's potencies with that of reference compounds. Unfortunately, this approach is often not addressed by the authors. Moreover, there is no universal reference compound for a specific antioxidant assay, and variations in the selected standards are frequent within literature. Regardless of that, one must note that S. officinalis polar extracts have been commonly used as a reference for the assessment of antioxidant properties of other less-investigated plants [14,15], showing EC 50 values in the range of 2.0 to 233.0 µg/mL for the DPPH • assay [5,7,[13][14][15]31]. Other ethanolic, methanolic, or aqueous extracts of Salvia origin, including those obtained from Salvia amplexicaulis [32], Salvia ringens [33], Salvia verbenaca, S. sclarea [34], Salvia argentea [15], and Salvia nemorosa [35], have been claimed to be good DPPH • scavengers as well, with antioxidant potentials that, in some cases, equal those of the standard compounds (ascorbic acid, butylated hydroxytoluene-BHT, or butylated hydroxyanisole-BHA). Hence, one might conclude that, in agreement with other studies reported for polar extracts of several Salvia species, S. officinalis, S. elegans, and S. greggii decoctions have a high ability to scavenge DPPH • , with S. elegans showing the most promising activity, followed by S. greggii and S. officinalis. Table 3. Correlation coefficients between the amounts of phenolic components found in the Salvia decoctions (caffeic acid and derivatives, coumaric acid derivatives, flavones and flavonols) and the data from the distinct biological experiments.
Polar extracts obtained from Salvia plants have also been previously screened for antioxidant abilities through other assays, although not as frequent as for DPPH • . In this context, Hamrouni-Sellami et al. [36] reported that the Fe 3+ reducing ability of S. officinalis methanolic extracts was 6.5-fold less that of ascorbic acid, being in agreement with our results which also pointed to good effectiveness for decoctions of the same species. In general, our results also indicate that the three sage species herein studied possess promising NO • scavenging capacities, as all the extracts had a lower EC 50 compared to ascorbic acid. Moreover, their activity seems to be superior to that described by Chen and Kang [37] for the methanolic extracts of Salvia plebeia (EC 50 = 216 ± 2.9 µg/mL), albeit that the absence of a reference compound in that study hampered solid conclusions. Furthermore, in our study, S. elegans and S. officinalis decoctions showed good O 2 •scavenging capacity, also suggesting that these extracts might be more active than the methanolic extracts of Salvia splendens (EC 50 = 527 µg/mL) [38]. Likewise, decoctions of S. officinalis, S. elegans, and S. greggii showed high capacity to scavenge RO 2 • (336-404 µM TE/mg), which was significantly superior to those previously reported for the aqueous and ethanolic extracts of S. officinalis (1143 and 2535 µM TE/g) and that other Salvia species (279-4735 µM TE/g). Phenolic compounds have been previously reported to counteract the activity of xanthine oxidase (XO) [39,40], i.e., the enzyme that catalyzes the oxidation of hypoxanthine to xanthine, and further catalyzes xanthine to uric acid with a concomitant production of O 2 •-, thus contributing to increment of oxidative stress events in cells. As can be observed in Table 2, the decoctions of the three Salvia species could effectively inhibit the activity of XO, albeit being less potent than the commercial drug allopurinol (EC 50 = 55.1-71.8 µg/mL for Salvia extracts vs. 0.09 ± 0.01 µg/mL for allopurinol, respectively). Among the extracts, the most powerful was that from S. officinalis, a fact that could be related to its richness in apigenin glucuronide or in other flavones (i.e., scutellarein and luteolin glycosides), since these compounds have been described as strong inhibitors of this enzyme [41][42][43][44]. Although the individual effect of the compounds has not been tested by us, the correlation coefficients between the antioxidant assays and the main compounds of each aqueous extract are in good agreement (Table 3). In XO inhibitory assay, the flavones content of the S. officinalis decoction was highly correlated (0.901) with its XO inhibition capacity.

Metabolic Enzyme Activity
α-Glucosidase, α-amylase and pancreatic lipase are key digestive enzymes involved in the metabolism of carbohydrates and lipids, which make them important targets for therapeutic control of diabetes and obesity. α-Amylase and α-glucosidase catalyze the hydrolysis of carbohydrates into simple sugars, thus their inhibition retards the digestion of starch and oligosaccharides contributing to the reduction of postprandial increase in plasma glucose levels. In turn, lipase inhibition decreases the digestion of dietary triglycerides, hence reducing the levels of free fatty acids and monoacylglycerols in the intestinal lumen [40,45,46]. In this study, the ability of S. officinalis, S. elegans, and S. greggii decoctions to inhibit the activity of these three digestive enzymes were assessed through in chemico models.
Notably, the inhibitory activities of the three Salvia extracts against α-glucosidase were very promising, especially for S. elegans and S. officinalis (EC 50 = 36.0 ± 2.7 µg/mL and 71.2 ± 5.0 µg/mL, respectively), demonstrating activities of 9-and 4-times that of the antidiabetic pharmaceutical drug, acarbose, respectively (Table 4). Moreover, despite being less effective than S. elegans and S. officinalis, S. greggii decoction was as effective as acarbose. Hence, our results suggest that the decoctions of S. elegans, S. officinalis, and S. greggii could serve as natural antidiabetic and anti-obesity agents to help in the control of glucose levels through the control of α-glucosidase activity. This hypothesis is also consistent with previous studies that reported identical results for polar extracts of Salvia against this enzyme, e.g., hydroethanolic extracts of S. officinalis (EC 50 value of 69.7 µg/mL) [47] and methanolic extracts of Salvia acetabulosa, S. nemorosa, and Salvia chloroleuca (EC 50 = 76.9 µg/mL, EC 50 = 19 µg/mL and EC 50 = 13.3 µg/mL, respectively) [35,48,49]. In vivo experiments have even demonstrated that the administration of a daily dose of S. officinalis methanolic extracts (500 mg/kg body weight) to alloxan-induced diabetic rats caused the inhibition of α-glucosidase activity comparable to that of the administration of acarbose (20 mg/kg bw) [45]. --6.5 ± 3.0 0.7 ± 0.2 Pancreatic lipase (3) 4.6 ± 3.6 a 8.2 ± 0.3 a 14.4 ± 7.4 a 1.8 ± 0.4 (1) Acarbose was used as standard. (2) Results are expressed as percentage (%) inhibition at the concentration of 0.5 mg/mL (Salvia decoctions) or as EC 50 (µg/mL), for the reference compound acarbose. (3) Results are expressed as percentage (%) inhibition at the concentration of 0.2 mg/mL (Salvia decoctions) or as EC 50 (ng/mL), for the reference compound orlistat. In each line, different letters mean significant differences (p < 0.05).
The inhibitory capacity of polar extracts of Salvia species towards α-glucosidase have been mostly correlated with their phenolic constituents. In fact, Chen and Kang [37] reported that the inhibition of this enzyme by S. plebeia methanolic extracts increased proportionally to their total phenolic content. Moreover, Kocak et al. [50] reported that aqueous and methanolic extracts of S. cadmica, both rich in rosmarinic acid, luteolin, and apigenin, had high inhibitory effects towards α-glucosidase and α-amylase. Moreover, several phenolic compounds isolated from S. miltiorrhiza, namely, tanshinone IIA, rosmarinic acid, rosmarinic acid methyl ester and salvianolic acid C methyl ester, were reported to be stronger inhibitors of α-glucosidase than acarbose (EC 50 = 0.042-0.23 µM and EC 50 = 5.8 µM, respectively) [51]. The flavonoid compounds luteolin-7-O-glucoside, luteolin-7-O-glucuronide, and diosmetin-7-O-glucuronide, isolated from the aerial parts of S. chloroleuca, also showed potent α-glucosidase inhibitory effects with EC 50 values of 18.3, 14.7, and 17.1 µM, respectively, exhibiting an inhibitory effect close to that of acarbose (EC 50 = 16.1 µM) [49]. Note that rosmarinic acid and caffeoyl rosmarinic acid are two major phenolic components in S. elegans decoctions and, based on the mentioned bibliographic data, it is feasible to hypothesize that they might be important contributors for the higher inhibitory ability of this extract compared to the other two. In fact, correlation coefficients determined for the different assays have shown a strong correlation between the results obtained for inhibitory activities on α-glucosidase and the content in caffeic acid and derivatives of the extracts (0.995). Interestingly, high correlation coefficients were also observed between the α-glucosidase and the antioxidant assays (0.976, 0.998, and 0.996 for reducing power, nitric oxide scavenging, and superoxide anion scavenging, respectively; Table 3), suggesting that metabolic and antioxidant effects might possibly be related.
However, regardless the great α-glucosidase inhibitory capacity and the fact that some authors have previously found potential inhibitory capacities in polar extracts of Salvia, namely for aqueous and methanolic extracts of S. cadmica, as well as for some individual phenolic compounds from Salvia origin [49], our results showed no substantial inhibition towards α-amylase up to the concentration of 0.5 mg/mL. Moreover, at 0.2 mg/mL, only S. greggii showed an anti-lipase activity higher than 10%. This could possibly be owed to its main phenolic component, i.e., luteolin-7-O-glucoside, since its aglycone has been reported to be a good lipase inhibitor [52,53], a hypothesis also supported by the high correlation found between the content of this flavone and the anti-lipase activity (0.930, data not shown). Interestingly, the anti-lipase activity of polar extracts of Salvia species has been previously described, namely for the methanolic extract of the leaves of S. officinalis (EC 50 = 94 µg/mL) [54], the methanol extract of Salvia spinosa (EC 50 = 156.2 µg/mL) [55], and methanol extracts of Salvia triloba (EC 50 = 100.8 µg/mL) [56]. Hence, despite data from literature that seems to suggest that at least some polar extracts from Salvia species might be promising with respect to their abilities to control the activity of α-amylase and pancreatic lipase, this was not the observed for S. officinalis, S. elegans and S. greggii decoctions herein studied.

Plant Sampling and Preparation of Extracts
S. officinallis, S. elegans¸and S. greggii were purchased from Ervital (Viseu, Portugal) as a mixture of flowers and leaves, and stems where were cultivated under an organic regime. After collection, the aerial parts were dried in a ventilated incubator at 20 to 35 • C for 3 to 5 days.
Phenolic compounds were extracted by decoction according the method described by Ferreira et al. [57], with adaptations. A volume of 100 mL of distilled water was added to 5 g of plant material (0.5 mm mesh powder) and the mixture was heated and then boiled for 15 min and filtered under reduced pressure through a G3 sintered plate filters. The resulting filtrated solution was concentrated in a rotary evaporator at 37 • C, followed by defatting with n-hexane (1:1 v/v). The resulting fraction was frozen, freeze-dried, and kept under vacuum in a desiccator in the dark for subsequent use [58].

Identification and Quantification of Phenolic Compounds
UHPLC-DAD-ESI/MS n analyses of phenolic profiles from S. officinalis, S. elegans, and S. greggii decoctions (5 mg/mL) were carried out on an Ultimate 3000 (Dionex Co., San Jose, CA, USA) apparatus equipped with an ultimate 3000 Diode Array Detector (Dionex Co., San Jose, CA" USA) and coupled to a Thermo LTQ XL mass spectrometer (Thermo Scientific, San Jose, CA, USA), an ion trap MS equipped with an electrospray ionization (ESI) source, following a method previous described [58]. Control and data acquisition were carried out with the Thermo Xcalibur Qual Browser data system (Thermo Scientific, San Jose, CA, USA). Nitrogen above 99% purity was used, and the gas pressure was 520 kPa (75 psi). The instrument was operated in negative-ion mode with the ESI needle voltage set at 5.00 kV and an ESI capillary temperature of 275 • C. The full scan covered the mass range from m/z 100 to 2000. CID-MS/MS and MS n experiments were simultaneously acquired for precursor ions using helium as the collision gas with a collision energy of 25-35 arbitrary units.
Gradient elution was carried out with a mixture of two solvents. Solvent A consisted of 0.1% (v/v) of formic acid in water, and solvent B consisted of acetonitrile, which was degassed and filtrated, using a 0.2 µm Nylon filter (Whatman International, Ltd., Maidstone, England) before use. The solvent gradient used consisted of a series of linear gradients starting from 5% of solvent B and increasing to 23% at 14.8 min, to 35% at 18 min, and to 100% at 21 min over three minutes, followed by a return to the initial conditions.
For quantitative determinations, the parameters of calibration curves, obtained by injection of known concentrations of the exact or structurally-related standard compounds, allowed the calculation of the limits of detection (LOD) and quantification (LOQ) [58].

DPPH • Scavenging Assay
Extracts capacity for scavenging DPPH • were evaluated following the procedure previously described by Catarino et al. [59]. Ascorbic acid was used as positive control. The concentration of the extract/standard able to scavenge 50% of DPPH • (EC 50 ) was determined using linear regression by plotting the percentage of inhibition against the concentration of the extracts.

Ferric Reducing Antioxidant Power (FRAP) Assay
For the reducing power assay, five different concentrations of each extract were prepared (0.05-0.25 mg/mL), and the assay was carried out according to a procedure described previously [59]. BHA was used as the positive control. A linear regression analysis was carried out by plotting the mean absorbance against the concentrations, and the EC 50 value was determined considering the extract/standard concentration that provided 0.5 of absorbance.

Oxygen Radical Absorbance Capacity (ORAC) Assay
The ORAC assay was performed according to the method previously described by Catarino et al. [60]. solution was added to each well to reach a final reaction volume of 200 µL. The plate was immediately placed in the plate reader (Biotek, Austria), and fluorescence was monitored every minute over 60 min. The measurement was carried out at 37 • C with automatic agitation for 5 s prior to each reading. Excitation was conducted at 485 nm with a 20 nm bandpass, and emission was measured at 528 nm with a 20 nm bandpass. Six concentration-dependent kinetic curves were obtained for each sample and for trolox as well. The area under the curve (AUC) of the fluorescence decay and Net AUC were calculated according to the following equations (1-3): where R 0 is the fluorescence reading at the initiation of the reaction and R i is the fluorescence reading at the time i. Antioxidant activities (ORAC values) of the extracts were calculated by using the following ratio: where m e is the slope of the curve of Net AUC vs. extract concentrations, and m T is the slope of the curve of Net AUC vs. trolox concentrations. The final results were expressed in µM of trolox equivalents (µM TE) per µg of sample extract.

NO • Scavenging Assay
The NO • scavenging method was adapted from Catarino et al. [60]. Briefly, 100 µL of six different extract concentrations (0-1 mg/mL) were mixed with 100 µL of sodium nitroprusside (3.33 mM in 100 mM sodium phosphate buffer pH 7.4) and incubated for 15 min under a fluorescent lamp (Tryun 26 W). Afterwards, 100 µL of Griess reagent (0.5% sulphanilamide and 0.05% naphthyletylenediamine dihydrochloride in 2.5% H 3 PO 4 ) were added to the mixture, which was allowed to react for another 10 min in the dark. The absorbance was then measured at 562 nm, and the percentage of NO • scavenging was calculated using the equation described by Yen and Der Duh [61] as follows: where A c is the absorbance of the control (without extract addition) and A e is the absorbance of the extract. Ascorbic acid was used as the reference compound. The concentration of the extract/standard able to scavenge 50% of NO • (EC 50 ) was then calculated by plotting the percentage of inhibition against the extract concentrations.

Superoxide Anion (O 2 •-) Scavenging Assay
The O 2 •scavenging method was carried out according to the method described by Catarino et al. [60]. Briefly, in a 96-well plate, 75 µL of six different sample concentrations (0.0-250 µg/mL) were mixed with 100 µL of β-NADH (300 µM), 75 µL of NBT (200 µM), and 50 µL of PMS (15 µM). After 5 min, the absorbances at 560 nm were recorded and the scavenging activity of superoxide radicals was calculated according to Equation (4). Gallic acid was used as the reference compound. The concentration of the extract/standard able to scavenge 50% of O 2 •-(EC 50 ) was determined using linear regression by plotting the percentage of inhibition against the concentration of the extracts.

Inhibition of Xanthine Oxidase Activity
Inhibition of xanthine oxidase activity was carried out following the method described by Filha et al. [62], with slight modifications. Briefly, in a 96-well plate, 40 µL of extract concentrations (0-2 mg/mL) were mixed with 45 µL of sodium dihydrogen phosphate buffer (100 mM, pH 7.5) and 40 µL of enzyme (5 mU/mL). After 5 min incubation at 25 • C, the reaction was started with the addition of 125 µL of xanthine (0.1 mM dissolved in buffer) and the absorbance at 295 nm was measured every 45 s over 10 min at 25 • C. The inhibitory effects towards xanthine oxidase activity was calculated as follows: where m c is the slope of the straight-line portion of the curve generated by the control (no inhibitor) and m e is the slope of the straight-line portion of the curve generated by each extract. Allopurinol was used as a positive control of inhibition. The concentration of the extract/standard able to inhibit 50% (EC 50 ) of the activity of the enzyme was determined using linear regression by plotting the percentage of inhibition against the concentration of the extracts.

Inhibition of α-Glucosidase Activity
Inhibition of α-glucosidase activity was measured following the method described by Neto et al. [63], with slight modifications. In short, 50 µL of different extract concentrations (0-2 mg/mL, in 50 mM phosphate buffer pH 6.8) were mixed with 50 µL of 6 mM 4-nitrophenyl α-D-glucopyranoside (pNPG), dissolved in deionized water. The reaction was started with the addition of 100 µL of α-glucosidase solution, and the absorbance was monitored at 405 nm every 60 s for 20 min at 37 • C. The inhibitory effects towards α-glucosidase activity was calculated as in Equation (5). Acarbose was used as a positive control of inhibition. The concentration of the extract/standard able to inhibit 50% (EC 50 ) of the activity of the enzyme was determined using linear regression by plotting the percentage of inhibition against the concentration of the extracts.

Inhibition of α-Amylase Activity
Inhibition of α-amylase activity was measured according to Wickramaratne et al. [64], with slight modifications. Briefly, 200 µL of extract six different extract concentrations (0-2 mg/mL) dissolved in 20 mM phosphate buffer (pH 6.9, containing 6 mM of NaCl) were added to 400 µL of a 0.8% (w/v) starch solution in the same phosphate buffer, and the mixture was incubated for 5 min at 37 • C. The reaction was then started with the addition of 200 µL of α-amylase solution, and after 5 min of incubation, 200 µL of the reaction mixture was collected and immediately mixed with 600 µL of DNS reagent (10 g/L of 3,5-dinitrosalicylic acid, 300 g/L of potassium and sodium tartrate tetrahydrate, and 0.4 M NaOH) to stop the reaction. A second aliquot of 200 µL was further collected 15 min later and mixed with DNS reagent as well. Samples were then boiled for 10 min, and, once they had cooled, 250 µL were transferred to each well in a 96-well microplate for absorbance reading at 450 nm. Blank readings (no enzyme) were then subtracted from each well and the inhibitory effects towards α-amylase activity was calculated as follows: where ∆Abs c is the variation in the absorbance of the negative control and ∆Abs e is the variation in the absorbance of the extract. Acarbose was used as a positive control of inhibition.

Inhibition of Pancreatic Lipase Activity
The lipase activity was measured according to the procedure described by Neto et al. [63], with slight modifications. The reaction mixture was prepared in a microtube by mixing 55 µL of five different concentrations of extract (0-2 mg/mL) dissolved in tris buffer 100 mM (pH 7.0) with 467.5 µL of tris-HCl (100 mM, pH 7.0, containing 5 mM of CaCl 2 ) and 16.5 µL of enzyme. The reaction was started by adding 11 µL of 20 mM 4-nitrophenyl butyrate diluted in DMSO. Final DMSO concentration in the reaction mixture did not exceed 2%. The reaction mixture was then quickly transferred to a 96-well plate and incubated for 35 min at 37 • C while the absorbance was being measured every 60 s at 410 nm. The inhibitory effects towards pancreatic lipase activity was calculated as in Equation (5). Orlistat was used as a positive control of inhibition.

Statistical Analysis
All data are presented as mean ± standard deviations from three independent assays performed at least in duplicate. One-way analysis of variance (ANOVA) followed by Tukey´s test was used to detect any significant differences among different means. Correlation analyses were performed using a two-tailed Pearson's correlation test. A p-value less than 0.05 was assumed as significant. The results were analyzed using GraphPad Prism 6 (GraphPad Software, La Jolla, CA, USA) and SPSS v 23.0 (Statistical Package for the Social Sciences).

Conclusions
This work clarifies the antioxidant properties of S. elegans, S. greggii, and S. officinalis decoctions as well as their inhibition towards the activity of carbohydrate and lipid metabolic enzymes, highlighting possible correlations with their phenolic components. It was shown that among the three plants, S. elegans decoctions were the most promising regarding antioxidant activity and inhibitory potential against α-glucosidase, a fact that might be related to its richness in caffeic acid and its derivatives. In turn, despite all the three decoctions of Salvia species could effectively inhibit the activity of xanthine oxidase, one should highlight the superior inhibitory capacity of S. officinalis, which is possibly associated with the presence of flavones. In conclusion, similarly to the well-known S. officinalis species, S. elegans and S. greggii are a valuable source of natural metabolites and could be used for commercial applications in novel functional foods or pharmaceutical ingredients targeting diabetes and obesity prevention.