Modeling of antiproliferative effects of Salvia officinalis L. essential oil optimized using Box–Behnken design

The objectives of this work were to optimize hydro-distillation extraction of Salvia officinalis L. essential oil (SoEO) and to evaluate the in vitro scavenging capacity of the free radicals DPPH·, NO·, ABTS+, and the ability to reduce Fe3+. The optimization of SoEO extraction by hydro-distillation was carried out using the response-surface methodology by Box–Behnken. The cytotoxicity and anti-proliferative capacities were evaluated by measuring cell viability and then modeled. Two human tumor cell lines: MCF-7 and HeLa were involved. The obtained optimal extraction yield was Y = 1.85 g/100 g d.b. The main identified fractions were camphene (23.7%), α-thujone (19.62%), 1,8-cineole (10.6%), viridiflorol (5.9%), borneol (5.72%), β-thujone (5.4%) and caryophyllene (3.83%). Also, SoEO was mostly able to scavenge DPPH· free radical, ABTS+ radical and hydrogen peroxide in an amount dependent manner (IC50 = 0.97, 0.279 and 0.05 mg/mL, respectively). All treated cell lines showed a significant reduction in cell viability in response to the increasing of oil concentration. The cytotoxicity activity against both tumor cell lines MCF-7 and HeLa was considerably important with IC50 = 3.125 and 8.920 μg/mL, respectively. The present work provides a preliminary platform for further investigation of the possible mechanism of S. officinalis essential oils and their individual compounds in cytotoxic and antitumor activities.


Introduction
Salvia officinalis L. is a perennial shrubby herb belonging to the Lamiaceae family, originally found in the East Mediterranean region (Berdowska et al. 2013). In Tunisia, some regions consider it as a sacred plant named "tree of meriam" and use it in auto-medicinal applications to increase women's fertility and/or support tumor cure. It is also widely used as a food condiment. Some studies explored the biological activities of this plant. Most of them reported antibacterial, antifungal and/or antioxidant activities of its essential oil (SoEO) (Fellah et al. 2006;Hayouni et al. 2008;Bouaziz et al. 2009;Ben Khedher et al. 2017), and more recently, as an insecticide (Ben Khedher et al. 2017). However, up to our bibliography, no works have been reported about the anti-cytotoxic and potential anti-proliferative activities of the Tunisian SoEO.
Essential oils are the secondary metabolites of medicinal plants. They have several biological effects. Qualitatively, they play a vital role as cytotoxic agents (Bakkali et al. 2008). In fact, the predominant medicinally valuable metabolites identified were monoterpenes (e.g. β-thujone, 1,8-cineole, and camphor), which were mostly present in the distilled oil and used externally to cure inflammations, infections such as stomatitis, gingivitis, and pharyngitis, and as antimicrobial agents. Also, diterpenes (e.g. carnosic acid) and triterpenes (oleanolic and ursolic acids) have shown a potent HIV-1 RT inhibitory function (Abu-Darwish et al. 2013;Stešević et al. 2014).
The first aim of this study was to optimize the extraction yield of SoEO and its chemical composition. The second was to investigate the antioxidant activity and cytotoxic effect against two human tumor cell lines: breast cancer (MCF-7), and cervix adenocarcinoma (HeLa).

Materials and methods
Samples Salvia officinalis L. samples were collected in mature phase: stage green color of leaves with flowers in April-May 2019 from the south east of Tunisia (Zarzis, 33° 36′ 12.4″ N 11° 01′ 59.8″ E). Aerial parts of plant were collected and some of them dried at 40 °C for 3 days until constant mass. The fresh and dried samples were subjected to hydro-distillation for 3 h using a Clevenger. Indeed, the hydro-distillation gives a whitish distillate presenting the emulsion. To acquire the operative essential oil, the salting out, using sodium chloride (NaCl), was applied to separate the essential oil from the reaction mixture, and then a few washings by an organic solvent (cyclohexane) were applied. The recuperated oils were measured and stored in dark glasses at 4 °C for further studies. The extraction yield is obtained by Eq. (1):

Optimization of extraction of essential oil
A Box-Behnken design was chosen to optimize the essential oils extraction yield. Three factors were tested: drying (x 1 ), number of washings (x 2 ) and salt concentration (x 3 ). Table 1 presents the low, central and high levels for each factor.
Fourteen experimental conditions were defined by Box-Behnken design ( Table 2). The obtained experimental results defined as the extraction yield are also presented in the same Table 2.
The proposed mathematical model presenting the dependence of the response to the different factors is presented as a second-degree polynomial (Eq. 2): where β 0 , β i , β ij and β ii are the regression coefficients, x i is the centered reduced level of the ith factor (value is equal to − 1, 0 or + 1) and n is the number of factors. The determination of model coefficients was carried out using the least squares method.

GC/MS analyses
The SoEO was analyzed using an Agilent-Technologies 6890 N Network GC system using the protocol described by Zarai et al. (2011). A sample of 1.0 μL was injected, using split mode (split ratio 1:100). The composition was reported as a relative percentage of the total peak area. The identification and authentication of the SoEO compounds was realized using a comparison of their retention times to n-alkanes, and their mass spectra compared to the published data and spectra of authentic compounds (Wiley and NIST Library).

Antioxidant capacity assays
2,2-Diphenyl-1-picrylhydrazyl (DPPH · ) free radical scavenging activity assay The essential oil antioxidant activity was evaluated by its ability in quenching the stable free radical DPPH · . Based on the Osawa and Namiki (1981) method, the radical scavenging activity against DPPH · free radicals was determined, and all experiment details were previously described by Kallel et al. (2019). Then, The DPPH · radical scavenging activity was carried out according to the following Eq. (3): (3) I(%) = A 517 of blank − A 517 of sample A 517 of blank × 100.

Hydrogen peroxide radical scavenging assay (H 2 O 2 )
Scavenging activity of hydrogen peroxide was resolute using the reference method (Ruch et al. 1989). Also, all precision experiment details were described in our previous work ). The percentage of H 2 O 2 scavenging was determined using Eq. (5):

Nitric oxide radical scavenging assay (NO · )
Nitric oxide generated from aqueous sodium nitroprusside solution interacts with oxygen to produce nitrite ions at physiological pH, which may be quantified and determined according to Griess-Ilosvay reaction (Marcocci et al. 1994). A solution of sodium nitroprusside (20 mM) was prepared in phosphate buffer (0.5 M, pH 7.4). The reaction mixture was composed of 2 mL of sodium nitroprusside and 250 μL of each sample, and incubated at 25 °C for 150 min. After incubation, a volume of 1 mL of each solution was taken and diluted with 1 mL of Griess reagent (1% sulfanilamide, 2% H 3 PO and 0.1% N-1-anphthyl ethylenediamine). The mixture was incubated again for 30 min at room temperature (25 °C), and then the absorbance was measured at 546 nm against the blank. Ascorbic acid was used as standard. The pink chromophore generated during diazotization of nitrite ions with sulfanilamide and subsequent coupling with α-naphtylethylenediamine was measured using a spectrophotometer at 546 nm. The percentage of NO · trapping of the extract was calculated according to the following formula (6):

Ferric reducing-antioxidant power (FRAP)
This method is based on the capacity of the plant to reduce ferric iron (Fe 3+ ) to ferrous iron (Fe 2+ ). The mechanism is known to be a marker of electron donor activity (Karagözler et al. 2008). To 1 mL of the sample at different concentrations (0.0625, 0.125, 0.25, 0.5 and 1 mg/mL), 2.5 mL of a buffer solution phosphate (0.2 M, pH 6.6) and 2.5 mL of 1% potassium ferricyanide (K 3 Fe(CN) 6 ) solution were added. The mixture was incubated at 50 °C for 20 min and then cooled to room temperature. Then, 2.5 mL of 10% trichloroacetic acid (TCA) was added to stop the reaction. The tubes were centrifuged at 3000 rpm for 10 min. A volume of 2.5 mL of the supernatant were then added to 2.5 mL of distilled water and 500 μL of 0.1% solution of iron trichloride (FeCl 3 , 6H 2 O) (Karagözler et al, 2008). Absorbance reading was done against a blank at 700 nm using a spectrophotometer. Ascorbic acid was used as a positive control. The increase in absorbency in the constituents indicates the raise in iron reduction. The percentage of iron reducing power was calculated by the following Eq. (7):

Human cell line (MCF-7and HeLa)
Two tested tumor cell lines were used in this study: MCF-7 and HeLa. The first tumor cell line, MCF-7, is a stable epithelioid cell line originally obtained from the pleural effusion of a female patient with metastatic breast cancer whose disease responded to hormone therapy ). The second one, HeLa, is a transformed line expressing the HPV18 virus (Human Papilloma Virus). This adherent line was obtained from tumor cells from cancer of the cervix of a 31-year-old woman (Gey et al. 1952). These cells have substantial amounts of estrogen receptor (Brooks et al., 1973).

Culture medium
The cell lines were grown in RPMI 1640 (Rosewell Park Memorial Institute) (Gibco) medium. The medium was added with 2 g/L sodium bicarbonate (HCO 3 Na) and adjusting the pH to 7.2 with 1 N HCl, the mixture was filtered through a filter of 0.22 µm and then supplemented with 10% fetal bovine serum (FBS) (Gibco), gentamycin 1%, and l-glutamine 2 mM.
Cells were grown at 37 °C in a humidified atmosphere of 5% CO 2 .

MTT test
The proliferation rates of MCF-7 and HeLa tumor cells after treatment with essential oils were determined by the colorimetric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay. The yellow compound MTT was reduced by mitochondrial dehydrogenases to the waterinsoluble blue compound formazan, depending on the viability of cells. MCF-7 and HeLa cells (240 × 10 5 in each well) were incubated in 96-well plates for 48 h in the presence or absence of essential oil. Twenty microliters of MTT solution (Sigma) (5 mg/mL in PBS) were added to each well. The plate was incubated for 4 h at 37 °C in a CO 2 -incubator. One hundred and eighty microliters of medium were removed and added to the same volume with DMSO-methanol. This reaction was monitored quantitatively by spectrophotometry. The DO at 570 nm reflects the activity of mitochondrial cytochromes. This activity can be considered as an index of cell proliferation (Mosmann 1983).

Statistical analyses
All biological measurements were conducted on the basis of three independent experiments. Statistical significance difference was evaluated using independent Student test. The difference between individual means was deemed to be significant at p < 0.05. To test the significant effects of factors and their interactions in model (Eq. 2), the analysis of variance test (ANOVA) was applied (p < 0.05).
The experimental design and all statistical analyses were performed with STATISTICA 12.0 Software, StatSoft, Inc.

Yield optimization of essential oil
The experimental results (Table 2) were used to determine the model coefficients (Eq. 8) and to identify the different influences of factors (via ANOVA test in Table 3) on the extraction yield of SoEO. Ferhat et al. (2008) showed an oil yield of aerial parts ranging from 1.1 to 1.2% based on dry weight, which is higher than previously investigated samples cultivated in Tunisia (Hayouni et al. 2008). Therefore, all these yield values and those obtained in this work are included in the range of 0.4-2.2% of sage (Salvia officinalis L.) essential oil yields from a variety of European sources (Perry et al. 1999;Pinto et al. 2007): Equation (8) presents all model coefficients. Drying in linear term ( x 1 ) and concentration of saline solution in quadratic term ( x 2 3 ) have the only significant effects on yield of essential oil extraction (Eq. 8 and Table 3; p < 0.05). Moreover, drying of the matter before extraction has a negative effect on the extraction yield (p < 0.01). This result was observed previously for other matters' essential oil (Dehghani Mashkani et al. 2018;Elaguel et al. 2019). Then, to produce a higher extraction yield of essential oil, it is important to establish it with fresh matter. There are no interactions between factors (Eq. 8, Table 3-p > 0.05). The quality of the determined model was carried out principally with three coefficients: coefficient of determination R 2 = 91%; adjusted coefficient of determination R 2 A = 71% and the root mean squared error RMSE = 0.29 g/100 g d.b. Moreover, the model presents a non-significant lack-of-fit criterion (p > 0.05). Therefore, the model is considered as suitable for all tested experimental data and it presents an important validity overall chosen experimental fields. So, the established model can be adopted for this study. Figure 1 presents the extraction yield of essential oil's dependence to different factors: drying and washings (a), drying and concentration of saline solution (b) and washings and saline solution (c). The higher linear influence of the drying on the extraction yield is shown in Fig. 1a, b. (8) Also, the quadratic saline solution concentration influence is shown in Fig. 1b, c. The presented graphic results confirm those obtained previously. The maximum extraction yield was obtained (Y = 1.85 g/100 g d.b.) with a desirability of 90%. The corresponding extraction must be without drying; five washings must be applied to the distillate using a saline solution concentration with 238.3 g/L.

Antioxidant activity of S. officinalis essential oil
Numerous methods have been used to evaluate antioxidant activity of the studied essential oil; and IC 50 values obtained for all test are shown in Table 6. Comparing the obtained results and their positive controls, S. officinalis essential oil exhibited a main antioxidant effect to scavenging DPPH · radical (Fig. 2a), ABTS + (Fig. 2b), FRAP (Fig. 3), and NO · (Fig. 3),but relatively lower than the synthetic antioxidant BHT, TROLOX, and vitamin C used as positive controls.
Moreover, it shows a comparable effect as standard vitamin C in H 2 O 2 test (Fig. 3). At all tested SoEO concentrations, showed a considerable antioxidant activity. The inhibiting percentage of free radical DPPH increased with elevated concentration of essential oil, and also with BHT (Fig. 2a). In fact, this essential oil has a substantial anti-radicular effect attending 50%. Also, it accounts for 85% compared to TROLOX as positive control with ABTS + essay (Fig. 2b). Using a Student t-test, we can validate a significant difference between means of DPPH · and ABTS + with essential oil The major components of the essential oils, as camphor, 1,8-cineol and α-pinene have been reported to have high anti-oxidative activity (Ruberto and Baratta 2000).
Also, the scavenging ability of sage essential oil on hydrogen peroxide is shown on Fig. 3. SoEO was capable of scavenging hydrogen peroxide in an amount dependent No.  manner. 0.05 mg of essential oil exhibited 50% scavenging activity on hydrogen peroxide. Wang et al. (2008) and more recently Tamil Selvi et al. (2015) supposed the antioxidant activity of sage oil due to presence of monoterpenes; and several sesquiterpenes compound; It is assumed that the contribution of minor and major compounds exhibited this activity and not only one or few active molecules (Wang et al. 2008).

Cytotoxicity assay
To evaluate cytotoxicity of the essential oils against MCF-7 and HeLa, MTT-based cytotoxicity assay was performed.  The results are expressed as percentages of cytotoxicity versus the concentrations of the essential oil. Our results showed a dissimilar cytotoxic effect of SoEO against both tumor cell line MCF-7and HeLa (Fig. 4). It can be clearly noted that MCF-7cells were the most affected cells followed by HeLa after 48 h. The proliferation of HeLa cells was considerably reduced by 50% exposure with 8.92 μg/mL, when IC 50 was obtained with 3.12 μg/mL against MCF-7 (Fig. 5). Moreover, the effect was proportional to the applied concentration.
To model the observed cytotoxicity behaviour of SoEO against two cells lines, a power model type (Cytotoxicity = a × concentration n ) was chosen and a Matlab algorithm was carried out and applied to the corresponding cytotoxicity versus concentration evolution in order to identify the model parameters using the fitting procedure. The used version of Matlab was version 2017b; © 1984-2017 The MathWorks, Inc. The results of power model parameters forMCF-7 and HeLa cells lines are presented in Eqs. (9) and (10), respectively: All the statistical parameters show a very interesting quality of fitting: coefficient of determination (73.44 ≤ R 2 ≤ 97.77%); adjusted coefficient of determination (68.13 ≤ R 2 Adj ≤ 97.33) and the root mean squared error (1.84 ≤ RMSE ≤ 4.13%). The modelling results of the cytotoxicity effect of SoEO against the used cells lines are shown in Fig. 5.  Bakkali et al. (2008) suggest that cytotoxic activity of the essential oil on the tumor cell lines may be ascribed to its antioxidant activity, since antioxidants are believed to be anti-mutagenic and anti-carcinogenic due to their radical scavenging properties.
For all tested cell lines, an increase of the cytotoxicity with the concentration of the essential oil was observed indicating a dose dependent effect. Examinations of the obtained results showed that cytotoxic effects SoEO were different between tumor line MCF-7 and HeLa. These differences could be attributed to the diversity of molecules of essential oil and affinity effect on one type of tumor more than others. In fact, this plant has been widely investigated for its chemical composition and pharmacological profile. Some diterpenoid quinines (royleanone-SAR 3, horminone-SAR 26, and acetyl horminone-SAR 43) isolated from the roots of S. officinalis are able to exert cytotoxic and DNA-damaging activity in human colon carcinoma Caco-2 cells and human hepatoma HepG2 cells cultured (Slamenová et al. 2004;Loizzo et al. 2007;Jiang et al. 2017). As well, a difference effect regarding various complex molecules compositions of essential oils and several terpenoid compounds belonging mainly to sesquiterpene and to monoterpene chemical groups may well be associated to diverse dotted functional groups as alcohols, phenols, ketones, etc. (Bakkali et al. 2008). These compounds might operate separately or in a synergic way.
Furthermore, some studies prove the cytotoxicity effect of SoEO against A549 and NCI-H226 cells which are treated with various concentrations and with a combination of two and three of its main constituents (1,8-cineole, α-thujone and camphor) (Privitera et al. 2014). Recently, Privitera et al. (2019) mentioned that the whole SoEO and a mixture of α-thujone, 1,8-cineole and camphor induced a significant reduction of cell viability at doses of 100 μg/mL and 200 μg/mL, after a 48-h incubation on breast (MCF-7), cervical (HeLa) and prostate (LNCaP) cancer cells. At the same concept, our Tunisian SoEO was reach in α-thujone (19.62%) and 1,8-cineole (10.6%), which could enhance the cytotoxicity of whole oil against MCF-7 and HeLa tumor cell lines. In addition, Siveen and Kuttan (2011) showed that α-thujone improved the cytotoxicity of colon cancer cells and inhibited lung metastasis of B16F-10 cells by inhibiting tumor cell proliferation, adhesion and invasion; while Biswas et al. (2011) revealed an anti-proliferative activity of α-thujone against melanoma. Also, Nikolić et al. (2015) reported the cytotoxic activity to the presence of camphor, 1,8-cineole and α-thujone.
Generally, the activity of an essential oil is accredited to and influenced by the major compounds. All the same, the minor compounds might also be of crucial importance (Guaouguaou et al. 2018).

Conclusion
The results obtained show that optimum yield of SoEO from Tunisian south east with RSM was1.85 g/100 g d.b. with a reach variation chemical compound. The in vitro experimental study clearly reveals the considerable antioxidant activity and the cytotoxic effect of SoEO against tumor cell line; suggesting the possibility of using this essential oil as a potential source of antioxidant ingredients for the food, cosmetic, drug and pharmaceutical industry. The modelling of the cytotoxicity effect displays that the Tunisian S. officinalis could be considered as an important Mediterranean medicinal plant and a highly promising natural agent to supplement drug tumor therapy. Therefore, the present work provides a preliminary platform for further investigation of the possible mechanism of S. officinalis essential oils and their individual compounds in cytotoxic and antitumor activity. More studies are necessary to explore the molecular mechanisms of this anticancer potential.