GC-MS Chemical Profiling, Biological Investigation of Three Salvia Species Growing in Uzbekistan

Salvia is a potentially valuable aromatic herb that has been used since ancient times. The present work studied the chemical profile of three Salvia species essential oils (EO): S. officinalis, S. virgata and S. sclarea, as well as assessing their antioxidant and enzyme inhibitory activities. A total of 144 compounds were detected by GC-MS analysis, representing 91.1, 84.7 and 78.1% in S. officinalis, S. virgata and S. sclarea EOs, respectively. The major constituents were cis-thujone, 2,4-hexadienal and 9-octadecenoic acid, respectively. The principal component analysis (PCA) score plot revealed significant discrimination between the three species. The antioxidant activity of the EOs was evaluated using in vitro assays. Only S. virgata EO showed antioxidant activity in the 2,2-diphenyl-1-picryl-hydrazyl (DPPH) assay (26.6 ± 1.60 mg Trolox equivalent (TE)/g oil). Moreover, this oil exhibited the highest antioxidant activity in 2,2-azino bis (3-ethylbenzothiazoline-6-sulphonic acid) (ABTS), cupric-reducing antioxidant capacity (CUPRAC) and ferric-reducing power (FRAP) assays in comparison with the other two EOs (190.1 ± 2.04 vs. 275.2 ± 8.50 and 155.9 ± 1.33 mg TE/g oil, respectively). However, S. virgata oil did not show any effect in the chelating ability assay, while in the PBD assay, S. officinalis had the best antioxidant activity (26.4 ± 0.16 mmol TE/g oil). Enzyme inhibitory effect of the EOs was assessed against acetylcholinesterase (AChE) and butyrylcholinesterase (BChE), tyrosinase, α-glucosidase and α-amylase. AChE enzyme was more sensitive to S. officinalis EO (4.2 ± 0.01 mg galantamine equivalent (GALAE)/g oil), rather than S. virgata EO, which was ineffective. However, S. virgata had the highest BChE effect (12.1 ± 0.16 mg GALAE/g oil). All studied oils showed good tyrosinase inhibitory activity, ranging between 66.1 ± 0.61 and 128.4 ± 4.35 mg kojic acid equivalent (KAE)/g oil). Moreover, the EOs did not exhibit any glucosidase inhibition and were weak or inefficient on amylase enzyme. Partial least squares regression (PLS-R) models showed that there is an excellent correlation between the antioxidant activity and the volatile profile when being compared to that of enzyme inhibitory activity. Thus, the studied Salvia essential oils are interesting candidates that could be used in drug discovery for the management of Alzheimer’s and hyperpigmentation conditions.


Introduction
Salvia, a popular aromatic plant known as sage, is an evergreen perennial subshrub native to the Mediterranean region and cultivated in several parts of the world [1]. Genus

Antioxidant Effect of the Essential Oils of Salvia Species
Six in vitro assays were employed to evaluate the antioxidant activity of the three Salvia EOs. These were radical scavenging activity using 2,2-diphenyl-1-picryl-hydrazyl (DPPH), 2,2 -azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) radical cation-based assay, total antioxidant capacity using cupric-reducing antioxidant capacity assay (CUPRAC), ferric-reducing antioxidant power assay (FRAP), EDTA chelating activity and phosphomolybdenum (PBD) assay. Only S. virgata oil showed antioxidant activity in DPPH assay (26.6 ± 1.60 mg TE/g oil, IC 50 : 1.98 ± 0.23 mg/mL). Moreover, the same oil exhibited the highest antioxidant activity in ABTS, CUPRAC and FRAP assays with the lowest IC 50 values (0.75 ± 0.02, 0.39 ± 0.02 and 0.28 ± 0.01 mg/mL, respectively) in comparison with the others. In addition, the essential oil was more active than Trolox (IC 50 : 0.44 ± 0.02 mg/mL) in CUPRAC. However, S. virgata oil did not show any effect in the metal chelating assay, while in PBD assay, S. officinalis had the best antioxidant activity (26.4 ± 0.16 mmol TE/g oil, IC 50 : 0.10 ± 0.01 mg/mL) ( Table 2). All tested essential oils exhibited stronger abilities in PBD assay compared to Trolox (IC 50 : 0.68 ± 0.01 mg/mL).  It has been noted that natural products with antioxidant potential represent promising therapies for various diseases since excessive production of free radicals and lipid peroxidation of cell membranes are involved in the mechanistic pathophysiology of certain ailments, especially cardiovascular diseases, diabetes, Alzheimer's, various types of cancers and others [26]. It is always recommended to assess the antioxidant activity of natural products by different methods with different mechanisms due to the complex nature of natural compounds [27]. Antioxidant activity of Salvia essential oils may be attributed to their volatile components. In the present study, it was found that S. officinalis oil is rich in oxygenated monoterpenes, which have been proven to possess the strongest antioxidant capacity relative to other classes of volatile compounds [28]. The major identified compound in this oil, α-thujone, showed good to moderate antioxidant capacity in a concentrationdependent manner in various assays such as DPPH, FRAP and hydroxyl, superoxide and nitric oxide radical scavenging activity [29]. A study showed that the antioxidant capacity of S. officinalis oil (with major compounds camphor and 1,8-cineole) was influenced by the time of hydro-distillation. The highest DPPH radical scavenging activity was observed for oil distilled in 2 h, while the highest activity in the TBARS assay was for oil distilled in 30 min. [30]. Regarding S. virgata, its flower oil showed better DPPH radical scavenging activity than its leaf oil, with activity equal to the standard butylated hydroxyanisole (BHA) [31]. Moreover, it was observed that oil isolated from aerial parts of S. virgata had better antioxidant activity in DPPH and FRAP assays when using the oil of full flowering rather than pre-flowering stage [32]. To the best of our knowledge, no previous extensive evaluation of the antioxidant activity of S. sclarea oil has been performed. However, its antioxidant capacity may also be attributed to some of its volatile constituents such as linalyl acetate, which has previously proven antioxidant potential in different assays, either in pure form or in oils where it is found as a major compound [33]. In addition to the different levels of different chemical components in the tested essential oils, the interactions between these components, namely antagonistic and synergetic, could affect the observed antioxidant properties [34][35][36].

Enzyme Inhibitory Effects of the Essential Oils of Salvia Species
The enzyme inhibitory effect of the oils was assessed against five enzymes which play a crucial step in certain medical conditions. Highest AChE inhibitory activity was recorded for S. officinalis (4.3 ± 0.01 mg galantamine equivalent (GALAE)/g oil; IC 50 : 0.68 ± 0.01 mg/mL), while S. virgata showed no effect at all. However, S. virgata had the highest BChE effect (12.1 ± 0.16 mg GALAE/g oil; IC 50 : 0.60 ± 0.01 mg/mL). All studied oils showed good tyrosinase inhibitory activity ranging between 66.1 ± 0.61 using S. sclarea EO to 128.4 ± 4.35 mg kojic acid equivalent ((KAE)/g oil) with S. officinalis EO. In addition, S. officinalis (IC 50 : 0.73 ± 0.01 mg/mL) exhibited stronger tyrosinase ability than standard inhibitor, kojic acid (IC 50 : 0.75 ± 0.01 mg/mL). Moreover, the oils did not exhibit any glucosidase inhibition, and exhibited weak or no activity as amylase inhibitors (Table 3).  Inhibition of AChE leads to the accumulation of acetylcholine, leading to better communication between nerve cells, and thus eases the symptoms in Alzheimer's patients [37]. BChE is also a co-regulator of acetylcholine. Therefore, its inhibition leads to better symptoms and prognosis in Alzheimer's [38]. Previous clinical studies showed that administration of sage oil and herbal teas improved mental and cognitive function in Alzheimer's individuals [39]. Alcoholic extracts of S. officinalis exhibited in vitro inhibition of AChE and BChE, with higher inhibition observed against BChE [40], which is in accordance with the present results, but regarding the essential oil.
Tyrosinase is a rate-limiting enzyme in melanin biosynthesis, as it oxidizes the amino acid tyrosine into melanin [41]. Its inhibitors, such as kojic acid, ellagic acid and hydroquinone, are used in the treatment of hyperpigmentation conditions and in skin-whitening cosmetics. A study on 19 essential oils showed that S. officinalis oil had moderate tyrosinase inhibitory activity with IC 50 99.8 ± 1.750 µg/mL relative to kojic acid with IC 50 2.3 ± 0.054 µg/mL [42]. Regarding S. virgata and S. sclarea oils, no previous data on their tyrosinase inhibitory activity were reported.
Both α-glucosidase and α-amylase digest carbohydrates, which leads to increasing levels of postprandial blood glucose, and their inhibition would lead to controlling postprandial hyperglycemia in diabetic patients, as well as reducing the risk for developing diabetes [43]. Although the studied Salvia oils showed no α-glucosidase inhibition and weak or no activity as α-amylase inhibitors, however, previous reports regarding their alcoholic and aqueous extracts recorded inhibitory activity for those enzymes [44]. Thus, their antidiabetic activity may be attributed to other active constituents not present in their essential oils, such as phenolic compounds.
Taken together, the observed enzyme inhibitory effects of the Salvia essential oils could have great potential for further pharmaceutical, nutraceutical and cosmeceutical applications. However, due to the complex nature of essential oils, interactions between chemical components should not be forgotten [45][46][47].

Chemometric Analysis
The GC-MS-based chemical profile of essential oils included both qualitative and quantitative discrepancies among different Salvia species; chemometric analysis was applied using principal component analysis (PCA) and hierarchal cluster analysis (HCA) to segregate closely related species, as well as to recognize any significant association between them [48]. A matrix of the total number of samples and their replicates (9 samples) multiplied by 144 variables (GC-MS peak area %) was constructed in MS Excel ® , then subjected to chemometric analysis (PCA and HCA). Due to the large number of variables, PCA was first used to reduce the dimensionality of the multiple dataset, followed by removing the redundancy in the variables and utilizing raw data (peak area % for each compound as in Table 1). The PCA score plot accounting for 90% of the variation in the dataset (Figure 1a) highlights that the first principal component (48%) discriminates between S. virgata (Sv) (PC1 negative values on the lower quadrant) and the other two species (PC1 positive values), while the second principal component (42%) discriminates between S. sclarea (Ss) (positive loading along PC2) and the others (negative loading along the same axis). profiles. The species sited near different metabolites are patterned in the score plot on the bases of these metabolites. The biplot shows that there is no specific marker (compound) accounting for the discrimination between Salvia species, proving the significant importance of the whole chemical profile of the essential oils in the discrimination between different species, not solely the compounds existing in high percentage. Additionally, HCA was applied as an unsupervised pattern recognition method to support results obtained by PCA. Figure 2 shows the HCA dendrogram, which displays segregation of different Salvia species in three main clusters. Cluster I, II and III present S. virgata (Sv), S. officinalis (So) and S. sclarea (Ss), respectively. The HCA dendrogram reveals the closeness of S. officinalis (So) and S. sclarea (Ss). HCA results endorse that of PCA.  The species sited near different metabolites are patterned in the score plot on the bases of these metabolites. The biplot shows that there is no specific marker (compound) accounting for the discrimination between Salvia species, proving the significant importance of the whole chemical profile of the essential oils in the discrimination between different species, not solely the compounds existing in high percentage.
Additionally, HCA was applied as an unsupervised pattern recognition method to support results obtained by PCA. Figure 2 shows the HCA dendrogram, which displays segregation of different Salvia species in three main clusters. Cluster I, II and III present S. virgata (Sv), S. officinalis (So) and S. sclarea (Ss), respectively. The HCA dendrogram reveals the closeness of S. officinalis (So) and S. sclarea (Ss). HCA results endorse that of PCA. Partial least squares (PLS) was applied to find a correlation between the volatile com pounds and their antioxidant and enzyme inhibitory activities. PLS-R1 and PLS-R2 mod els were constricted by the data matrix X containing the peak area of the GC/MS and the response y vectors containing the antioxidant and enzyme inhibitory activities data, re spectively. The model performance was estimated by the parameters of root mean square error of calibration (RMSEC), root mean square error of validation (RMSEV) and correla tion (R2). PLS-R1 model parameters, including slope, offset, RMSEC, RMSEV and R 2 , are shown in Table 4, indicating the strong prediction ability of the PLS regression model PLS-R1 models showed excellent linearity and accuracy, with R 2 > 0.99 and slope close to 1 (a value close to 1 means the predicted values are close to the reference), with low dif ferences between RMSEC and root mean square error of validation (RMSEV) revealing the robustness of the model. It was observed that both DPPH and PBD data displayed the lowest RMSEV values (0.5325 and 0.6550), respectively, suggesting that they are more rep resentative than other techniques to measure the antioxidant activity. The prediction per formance for the developed models is shown in Table 5. The results show that the antiox idant activity is correctly predicted with ±5% accuracy.  Partial least squares (PLS) was applied to find a correlation between the volatile compounds and their antioxidant and enzyme inhibitory activities. PLS-R1 and PLS-R2 models were constricted by the data matrix X containing the peak area of the GC/MS and the response y vectors containing the antioxidant and enzyme inhibitory activities data, respectively. The model performance was estimated by the parameters of root mean square error of calibration (RMSEC), root mean square error of validation (RMSEV) and correlation (R2). PLS-R1 model parameters, including slope, offset, RMSEC, RMSEV and R 2 , are shown in Table 4, indicating the strong prediction ability of the PLS regression model. PLS-R1 models showed excellent linearity and accuracy, with R 2 > 0.99 and slope close to 1 (a value close to 1 means the predicted values are close to the reference), with low differences between RMSEC and root mean square error of validation (RMSEV) revealing the robustness of the model. It was observed that both DPPH and PBD data displayed the lowest RMSEV values (0.5325 and 0.6550), respectively, suggesting that they are more representative than other techniques to measure the antioxidant activity. The prediction performance for the developed models is shown in Table 5. The results show that the antioxidant activity is correctly predicted with ±5% accuracy.  Concerning PLS-R2, model parameters, including slope, offset, RMSEC, RMSEV and R2, are shown in Table 6, indicating the moderate prediction ability of the PLS regression model. PLS-R2 models showed good linearity and accuracy with R 2 > 0.97, except for BChE inhibition, which exhibited much lower values. The prediction performance for the developed models is shown in Table 7.

Plant Material
The

Extraction of Essential Oils of Salvia Species
Aerial parts of Salvia samples were air-dried in the shade. Essential oils were hydrodistilled (400 g dry powder in 1 L distilled water) using a Clevenger-type apparatus for 3 h. The yields were 0.8% w/w for S. officinalis, 0.2% w/w for S. virgata and 0.3% w/w for S. sclarea. The recovered oils were dried over anhydrous sodium sulphate and kept in sealed dark vials at 4 • C until analysis.

GC-MS Analysis of Essential Oils of Salvia Species
GC-MS of Salvia essential oils was carried out using an Agilent 7890 B gas chromatograph (Agilent Technologies, Rotterdam, The Netherlands). The column used was a VF-Wax CP 9205 fused silica (30 m × 0.25 mm, ID 0.25 µm). Helium was used as carrier gas at a flow rate of 0.9 mL/min. An Agilent 5977A mass selective detector was used, with a scan range of 45-950 atomic mass units with a detector temperature of 270 • C and split mode injection at a split ratio of 1:20. An autosampler was used for sample injection (0.5 µL) with an injector temperature of 250 • C. The interface temperature was 280 • C, the source temperature was 230 • C, and the ionization energy was 70 eV. The initial oven temperature was 50 • C for 5 min., which was then raised to 280 • C at a rate of 5 • C/min, then kept isothermal at 280 • C for 15 min. Standard alkanes (C7-C40) obtained from Sigma-Aldrich (Darmstadt, Germany) were used to calculate the Kovats index (KI). Chromatograms were generated using enhanced ChemStation software (Agilent Technologies, Waldbronn, Germany). Volatile compounds were identified by comparing their mass spectra and KI was calculated with the 9th edition of Wiley Registry of mass spectral data and NIST library.

Antioxidant Assays
In vitro assays were employed to evaluate the antioxidant activity of the three Salvia EOs using the 2,2-diphenyl-1-picryl-hydrazyl (DPPH), 2,2 -azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) radical-cation-based assay, total antioxidant capacity using cupricreducing antioxidant capacity assay (CUPRAC), ferric-reducing antioxidant power assay (FRAP), EDTA chelating activity and phosphomolybdenum (PBD) assay. These assays were performed according to previously described standard procedures, and values are expressed as Trolox or EDTA equivalent [49,50]. The experimental procedures are given in supplemental materials. To provide a comparison with standard compounds, IC 50 values (the half inhibitory concentration) were also calculated for DPPH, ABTS and metal chelating assays. IC 50 values for other assays (reducing power and phosphomolybdenum) reflect that the concentration at which absorbance occurs is 0.5.

Enzyme Inhibitory Assays
The enzyme inhibitory effect of the oils was assessed against five enzymes which play a crucial step in certain medical conditions. These included AChE, BChE, tyrosinase, α-glucosidase and α-amylase. Assays were carried out according to standard procedures, with values expressed as galantamine, kojic acid and acarbose equivalent for cholinesterase, tyrosinase and α-glucosidase/α-amylase inhibitory activities, respectively [50,51]. The experimental procedures are given in supplemental materials. IC 50 values (the half inhibitory concentration) for each oil and standard inhibitors were also calculated for enzyme inhibitory assays.

Statistical Analysis
All analyses were conducted in triplicate. Values are expressed as means ± SD. Statistical significance was determined using one-way ANOVA followed by Tukey's post hoc test (significance level at p < 0.05).

Chemometric Analysis
The data obtained from GC-MS were subjected to chemometric analysis. Principal component analysis (PCA) was applied as an initial step for data investigation to present an overview of all species divergences and to recognize markers responsible for this dissimilarity [52]. Hierarchal cluster analysis (HCA) was then applied to allow the clustering of different species. The clustering pattern was constructed by the single linkage method. PCA and HCA were accomplished using the SIMCA-P version 13.0 software package (Umetrics, Umeå, Sweden). A quantitative calibration model, partial least squares (PLS), was designed to find a correlation between the volatile compounds (GC/MS peak areas) (X) matrix and their antioxidant, enzyme inhibitory activities (Y) matrices. In this state, there was no division of data into model and test set, as only nine samples for each model were assessed (small dataset). PLS was performed using CAMO's Unscrambler ® X 10.4 software (Computer-Aided Modeling, AS, Oslo, Norway).

Conclusions
Salvia species are aromatic plants that have been widely used in various cultures since ancient times. In the present work, the chemical profile of three Salvia species essential oils was investigated. The studied species were S. officinalis, S. virgata and S. sclarea. Their major identified compounds were cis-thujone, 2,4-hexadienal and 9-octadecenoic acid in S. officinalis, S. virgata and S. sclarea EOs, respectively. The PCA score plot revealed significant discrimination of the three species even though its biplot was unable to identify the compounds responsible for these differences. The three Salvia species EOs exhibited moderate antioxidant activities. Highest AChE inhibitory activity was recorded for S. officinalis, while S. virgata had the highest BChE effect. All studied oils showed good tyrosinase inhibitory activity. Moreover, the oils did not exhibit any glucosidase inhibition, and exhibited weak or no activity as amylase inhibitors. Thus, the studied Salvia essential oils are interesting candidates that could be used in drug discovery for the management of Alzheimer's and hyperpigmentation conditions.