Mentha x rotundifolia plants from different annuities [2014 (MR14), 2015 (MR15), 2016 (MR16) and 2017 (MR17)] were kindly provided by Dr. J. Navarro Rocha from Centro de Investigación y Tecnología Agroalimentaria de Aragón (Spain). These plants were experimentally grown in field in Ejea de los Caballeros (Zaragoza, Spain; 42°8’8.73” N, 1°12’31.50” W/346 m a.s.l.) and were provided with drip irrigation from June to August every annuity due to the low rainfall in this area. Leaves of M. rotundifolia plants, harvested at their flowering stage, were air-dried at ambient temperature and in the absence of light, stored in closed amber vials for preservation of their original composition, ground in a domestic mill (Moulinex, Barcelona, Spain) and sieved (< 500 µm) before extraction.

Caffeic acid and chlorogenic acid were purchased from Sigma-Aldrich^® (St. Louis, MO, USA), luteolin 7-O-glucoside and salvianolic acid B were acquired from Extrasynthese (Genay, France) and rosmarinic acid from Cayman Chemical Company (Michigan, USA). All standards used were of analytical grade (purity ≥ 95%).

A preliminary study was carried out to select the optimal extraction technique (UAE, MAE, or SWE). In all cases, 1 g of sample (MR16) and 12 mL of ultrapure water (Milli-Q^® System, Millipore, Burlington, MA, USA) were used as sample to solvent volume ratio (s/v). All experiments were carried out in triplicate at 50°C (otherwise below specified) for 5 min and the extracts thus obtained were frozen at –20°C until analysis. For the selection of the optimal technique, total phenolic content (TPC, section 2.5) and the antioxidant activity [1,1-diphenyl-2-picrylhydrazyl (DPPH) and 2,2’-azino-bis (3-ethylbenzthioazoline-6-sulfonic acid) (ABTS) methods, sections 2.6] of the extracts were used as response variables to be maximized.

First, the s/v ratio was optimized by evaluating three sample amounts (s), i.e., 0.5 g, 1.0 g, and 1.5 g, and measuring the resulting volume of extract recovered after each experiment. The selection of the optimal sample amount was based on the extraction yield of three phenolic compounds of different nature (luteolin 7-O-glucoside, rosmarinic acid, and salvianolic acid B), previously described in the literature as the most abundant of Mentha extracts [20–22].

Subsequent optimization of the extraction temperature (T, °C) and time (t, min) was carried out using a 3-level factorial design (Table S1) with the aim of maximizing the extraction efficiency and of providing a fast SWE extraction method. The quadratic model proposed was:

R = β₀ + β₁T + β₂t + β_1,1T² + β_2,2t² + β_1,2Tt + ε (equation 1)

In equation 1, β₀ is the intercept, β_i are the first-order coefficients, β_i,i the quadratic coefficients for ith factors, β_i,j the coefficients for the interaction of factors i and j, and ε is the error. The ranges of the experimental factors evaluated were T = 50º–150ºC and t = 5–30 min. Temperature values above the boiling point of water were considered to evaluate the benefits associated to the extraction of M. rotundifolia phenolics under the pressurized conditions provided by SWE. Potential adverse effects such as the possible degradation of thermolabile phenolics were also considered in the selection of the temperature range.

As response variables (R) in equation 1, the individual content (R_x, in mg g^–1 of dry sample) of selected bioactives (luteolin 7-O-glucoside, rosmarinic acid, and salvianolic acid B), the TPC (R_TPC) and the antioxidant activity, as determined by the DPPH method (R_DPPH) and the ABTS (R_ABTS) assay, were considered. The parameters of the model that individually maximized R_x, R_TPC, R_DPPH and R_ABTS, were estimated by multiple linear regression (MLR) using Statgraphics Centurion XV software (Statistical Graphics Corporation, Rockville, MD, USA).

A desirability function (D) that simultaneously maximized all previously mentioned individual responses was also considered to provide optimal SWE extraction. This D function takes values between 0 (completely undesirable value) and 1 (completely desirable or ideal response). Finally, the number of extraction cycles (C1–C3) that maximized the extraction yield of target compounds was also evaluated.

A high performance liquid chromatography (LC; HPLC)-ultraviolet (UV)/mass spectrometry (MS) apparatus consisting of a 1200 Series HPLC system (Agilent Technologies, Santa Clara, CA, USA) was used in this study for characterization of SWE extracts. This equipment was provided with an in-line degasser, a binary pump, a Rheodyne^® 7125 injection valve, and a column oven (Kariba Instruments, UK), coupled via an electrospray ionization (ESI) interface to a single quadrupole MS detector (MSD) 1100 (Agilent Technologies). The operating parameters of the electrospray source were as follows: spray voltage, 4 kV; drying gas (N₂, 99.5% purity) temperature, 300°C; drying gas flow, 12 L min^–1; nebulizer (N₂, 99.5% purity) pressure, 276 kPa; and fragmentor voltage, 80–100 V. Quasimolecular ions for target compounds were recorded in the selected ion monitoring (SIM) mode under negative polarity [(M-H)^– 179, 353, 359, 447, and 717 for caffeic acid, chlorogenic acid, rosmarinic acid, luteolin 7-O-glucoside, and salvianolic acid B, respectively]. Data acquisition and processing were performed using HP ChemStation Rev. A.07.01 software.

Chromatographic separations were carried out on a reverse-phase Luna C18 analytical column (Phenomenex, Cheshire, UK; 100 mm × 2.0 mm i.d., 3 µm) operating at a flow rate of 0.4 mL min^–1. The mobile phase was a binary mixture of solvent A (water with 0.1% acetic acid) and B (acetonitrile with 0.1% acetic acid), according to the following gradient: 0 min 5% B; 20–25 min 36% B; 35 min 90% B; 36–50 min 5% B. Injection volume was 5 µL and the column temperature was set at 25°C.

Quantitation of target phenolics was performed in triplicate using external standard calibration curves within the range 0.01–100 µg mL^–1. Goodness of fitting for these calibration curves and reproducibility of the method were previously assured. Results were expressed in milligrams per gram of dry sample.

TPC of M. rotundifolia extracts was determined using the Folin-Ciocalteu reagent (2N) and gallic acid as standard (both from Sigma-Aldrich^®), according to the method of Soria et al. [23] with slight modifications. An aliquot (100 μL) of extracts previously diluted [1:5–1:175, v/v], 100 μL of MeOH (Sigma-Aldrich^®) and 100 μL of Folin-Ciocalteu reagent were vortexed in a 1.5 mL eppendorf tube. After 5 min, 700 μL of 75 g L^–1 Na₂CO₃ (Panreac, Barcelona, Spain) were added and the samples were vortexed briefly. The eppendorfs were then allowed to stand in the dark for 20 min at room temperature. Following this, the samples were centrifuged at 4,400 g for 3 min and their absorbance was measured [number of replicates (n) = 3] at 750 nm using a BioTek ELx800^TM microplate reader (BioTek Instruments, Inc., USA). The same procedure was repeated with aqueous solutions of gallic acid in the 10–100 mg L^–1 concentration range to build up a calibration curve. Results were expressed as gallic acid equivalents (GAE; mg mL^–1 or mg g^–1 dry sample).

Data were subjected to statistical analysis by using Statistica 7.0 software (StatSoft, Inc., Tulsa, OK, USA). Significance (P < 0.05) of differences was assessed by the analysis of variance (ANOVA, Tukey test).

The efficiency of UAE, MAE and SWE in terms of TPC and antioxidant activity of Mentha x rotundifolia MR16 extracts obtained under similar extraction conditions (50ºC and 5 min) was evaluated (Table 1).

For UAE, a US bath and two sonoprobes with different internal diameter (3 mm and 12 mm) were used. Although no significant differences were observed in the antioxidant activity of the extracts obtained using the 12 mm probe and the bath, a higher content of phenolic compounds was found for the last treatment. As regards the different sonoprobes, significantly lower TPC and antioxidant activity values were achieved using the 3 mm probe.

As compared with MAE and SWE treatments, UAE bath provided the highest TPC. However, the antioxidant activity measured by the DPPH method of UAE extracts was significantly lower than those provided by MAE and SWE, ruling out UAE for further investigation.

Although no significant differences were found in the antioxidant activity (ABTS method) of extracts obtained by MAE and SWE, this last technique was selected for subsequent studies and for in-depth optimization on the basis of its higher TPC yield and greater antioxidant activity (DPPH assay).

SWE parameters (s, t, T and the number of cycles) were optimized to enhance the extraction of M. rotundifolia bioactives. As previously mentioned, the volume of solvent is not a fixed parameter in SWE, since it depends on the capacity of the extraction cell and on the amount of sample and sea sand used. Thus, a number of experiments using different sample amounts (s: 0.5 g, 1 g, and 1.5 g of MR16 leaves) were carried out under the same operating conditions (100°C for 18 min).

It shows in Table 2 the concentrations of selected bioactives (luteolin 7-O-glucoside, rosmarinic acid, and salvianolic acid B) from M. rotundofolia extracted using the different sample amounts (s) evaluated. Higher concentrations of rosmarinic and salvianolic acid B were recovered with 0.5 g and 1 g, whereas this last amount provided the most efficient recovery of luteolin 7-O-glucoside. Therefore, 1 g of sample (corresponding to an s/v of 0.08 g mL^–1) was selected for further experiments.

Optimization of SWE temperature and time was carried out following a Box-Behnken experimental design, considering as response variables the TPC (R_TPC), the antioxidant activity determined by the DPPH (R_DPPH) and ABTS (R_ABTS) methods and the concentrations (mg g^–1 of dry sample) of previously selected phenolic compounds (R_x, Table S1).

In general, the highest TPC and antioxidant activity were obtained at high temperatures and intermediate times (e.g., R_TPC 6.0 mg GAE mL^–1, R_ABTS 14.02 mg TE mL^–1, R_DPPH 12.2 mg TE mL^–1 at 150ºC, 18 min). However, the highest concentrations of rosmarinic acid (3.8 mg g^–1 dry sample), salvianolic acid B (0.72 mg g^–1 dry sample) and luteolin 7-O-glucoside (0.068 mg g^–1 dry sample) were extracted at lower temperatures (100ºC, 18 min).

Response surface methodology was used to calculate the coefficients of the quadratic models proposed and to estimate the statistical significance of the regression coefficients. The equations of each model and the optimal conditions for the different responses considered (R_x, R_TPC, R_DPPH, and R_ABTS) are shown in Table 3. In general, extractions at temperatures above 100°C maximized all responses, with T and T² being the most significant (P < 0.05) variables. Optimal value of TPC was achieved at the highest temperature and the longest time (142ºC, 20 min), while the extraction of rosmarinic acid, salvianolic acid B, and luteolin 7-O-glucoside, which shared a similar trend, was maximal between 98ºC and 109ºC, and 15 min and 18 min. The optimal conditions that provided the highest antioxidant activity, measured by both the DPPH and ABTS methods, were intermediate between the previous cases, requiring 120ºC, 5 min, and 128ºC, 18 min, respectively.

Finally, when a multiple response was considered to simultaneously maximize all the response variables above mentioned, the optimal extraction parameters were found to be 120°C and 5 min (D = 0.67).

Evaluation of the number of extraction cycles required to achieve an exhaustive extraction of the phenolic compounds present in this matrix was also considered. Whereas salvianolic acid B was completely recovered after the first extraction cycle, 67% of rosmarinic acid and 39% of luteolin 7-O-glucoside were extracted. Therefore, a second extraction cycle was required to maximize the recovery of all target phenolic compounds (84% luteolin 7-O-glucoside and 90% rosmarinic acid). In summary, the optimal SWE conditions were 120ºC, 5 min, 0.08 g dry sample per 1 mL of water and 2 extraction cycles.

Concentration of bioactives in the SWE extracts obtained from MR14–MR17 leaf samples was determined by LC-UV/MS. Apart from the phenolics considered for the optimization of the method, caffeic and chlorogenic acids were detected and quantified in these extracts. As it is shown in Table 4, no significant differences in the concentration of bioactives were found among the samples in study, except for rosmarinic acid, whose concentration decreased with the collection annuity down to half of its initial value (MR14: 4.16 mg g^–1 of dry sample, MR17: 2.5 mg g^–1 of dry sample). TPC and antioxidant activity of MR14–MR17 extracts was also stable irrespective of the harvesting year, with only sample collected in 2015 (MR15) providing an unexpectedly higher TPC.