2.1 Reagents and materials

All organic solvents used in the extractions, chromatographic and spectroscopic processes were of analytical quality: acetonitrile (ACN), dichloromethane (CH₂Cl₂), chloroform (CHCl₃), diethyl ether, methyl tert-butyl ether (MTBE), hexane (Hex), ethyl acetate (EtOAc), anhydrous ethanol (EtOH), methanol (MeOH), in addition to deuterated solvents – chloroform (CDCl₃) and methanol (CD₃OD), all provided by TEDIA® COMPANY (Fairfield, USA). In the analysis by HPLC-PDA-MS, ultrapure water with resistivity of 18.3 MΩ/cm³ produced in a purification system of water model Scholar UV UP 900 (BIOHUMAN, Curitiba, Brazil) was used. For preparation of aqueous stationary phase that comprised the two-phase solvent system used in the separation by HSCCC, distilled water was used. For analysis by analytical thin layer chromatography (A-TLC), SiliaPlate TLC, Aluminum-Backed, Silica, 200 µm, 20 × 20 cm, F 254 (SiliCycle Inc., Quebec, Canada) chromatographic plates were used. Trolox, gallic acid, DPPH (1,1-diphenyl-2-picryl-hydrazil) and ABTS [2,2′-Azino-bis(3-ethylbenzothiazoline-6-sulfonicacid) diammonium salt were purchased from Sigma Aldrich (St Louis, MO, USA).

2.2 Sample collection and processing

The waste samples were produced from the pericarp of fruits collected directly from an orchard of G. mangostana, located in the municipality of Santa Bárbara - PA/Brazil, georeferenced coordinates 1° 15′ 3.88′’ S; 48° 16′ 36.84″ O, in April 2019, with authorization for collection and transportation of botanical material granted by the Authorization and Information System in Biodiversity - SISBIO, with registration number 65490-1, and activity record of access to genetic heritage registered in the National System for the Management of Genetic Heritage and Associated Traditional Knowledge – SisGen with the number A246F9A. After collection, the fruits were screened in order to classify those that had the best characteristics in terms of color, appearance and integrity of the pericarp, resulting in 94 selected fruits. The fruits were stored in plastic trays and transported to the laboratories where they were kept in a conditioned environment (19 ± 2 °C) for approximately 15 h before processing. The first stage followed by rinsing the fruits under running water to remove coarse dirt and microorganisms adhered to the skin. After rinsing, the fruits were kept in a bath of 0.05% commercial sodium hypochlorite solution for 20 min. Then, the fruits went through a second rinse, with sterile distilled water, to remove the residual disinfectant solution. The fruits were then pulped and the pericarp, separated for weighing and subsequent drying. The mass of fresh pericarp (MF) was 4.99 kg. After weighing, the fresh material was dehydrated in an air circulation kiln at a temperature of 45.0 ± 2.5 °C for 72 h. The dry material was subjected to grinding operation in a Willey knife mil model SL-31 (SOLAB, Piracicaba, Brazil) equipped with a 10 mesh granulometric classification and standardization sieve, then it was collected, separated into fractions, labeled, and stored in low density polyethylene plastic (LDPE) bags, colorless 45 × 300 × 0.04 mm, at room temperature of 25 °C until the extraction stage was carried out. The final mass of processed waste was 1.82 kg.

2.3 Extracts preparation

For the extraction of bioactive compounds, an ultrasonic bath with indirect contact was used, model USC-2800A (UNIQUE Ind. and Com. Ltda., Indaiatuba, Brazil), operating at 40 kHz, power of 130 W and temperature of 25.0 ± 0.5 °C. Initially, to evaluate the extracting capacity of the solvents used, as well as the selectivity of them with the analytes (α- and γ-mangostin), the microextraction scale was used, starting from 5 g of dry and ground waste to 50 mL of solvent. The following solvents have been tested: CHCl₃, EtOAc, EtOH and MeOH, with extraction time of 30 min. The obtained solutions were filtered and the solvents, removed under reduced pressure by using a rotary evaporator model Syncore (BÜCHI Labortechnik AG, Flawil, Switzerland). Then, the extracts were named according to the respective solvents used as extracting phase: chloroformic extract (CE), ethylacetic extract (EaE), ethanolic extract (EtE) and methanolic extract (ME).

The raw ethanolic extract (REE), submitted to fractionation by HSCCC-Prep, was produced through the same procedure used to obtain the extracts in microscale. 100 g of dry waste were used for 300 mL of EtOH; 1 batch. The dry REE mass obtained was 9.638 g.

2.4 Evaluation of metabolic profiles by A–TLC

For the analysis, 50 × 50 mm silica gel chromatography plates and 40 mm chromatographic path were used. Individual samples of each extract were prepared at a concentration of 1 mg/mL, and aliquots of 5 µg/spot (5 µL) were inoculated in 5 mm bands. Then, the chromatoplates were eluted in a glass vat by using the combination of dichloromethane and methanol 9:1 (v/v) as mobile phase. After elution, the plates were observed under the incidence of 254 nm and 365 nm UV radiation, revealing the separate constituents in the form of dark and fluorescent bands, respectively. After observation, the appropriate chromogenic reagents for detection of classes of interest were applied to the plates (Hellmut et al., 1990; Kinghorn, 1997). The determination of the R_f (retention factor) of α and γ-mangostin was performed by the identification of both by HPLC-PDA-MS.

2.5 Selection of two-phase solvent system

The selection of the most suitable solvent system was guided by the study and evaluation of separation or accommodation times between the phases (τ), and partition coefficient values (K) and xanthones α- and γ-mangostin separation factor (α). Each system consisted of an organic upper phase (OP) and an aqueous lower phase (AP). The phases were prepared separately for each of the tested systems. A volume of 5 mL of each phase was transferred to a 13 × 100 mm glass test tube, with screw cap, and shaken vigorously for 20 s for mutual phase saturation. After stirring, the tubes were left to stand until a clear and clean separation interface was formed. Then, 100 mg of REE were added to each of the solvent systems at equilibrium followed by vigorous manual stirring for 20 s. The tubes were then left to rest and the separation time between the phases of each system was recorded. After equilibration, 1.0 µL aliquots from both phases of each system were injected separately in an HPLC-PDA system for determination of K and α values from the absolute peaks area data corresponding to analytes in the chromatograms registered. By convention, in liquid chromatography, K value is expressed as the concentration of solute in the upper phase divided by its concentration in the lower phase (K_U/L); α value is given by the ratio between the K values of each analyte in the same solvent system (α = K₂/K₁; K₂ > K₁) and must be greater than 1.5 to obtain an adequate resolution. K value was calculated by the ratio between the analyte peak area in OP and the peak area in AP.

2.6 Prep-HSCCC separation procedure

The REE fractionation and α- and γ-mangostin isolation were performed in a preparative high-speed countercurrent chromatograph (Prep-HSCCC), model Quattro ccc (AECS-Quikprep™, United Kingdom) equipped with three independent multilayer separation columns made of polytetrafluoroethylene (PTFE), and arranged in parallel. The separation was developed in column-1 (preparative, 3.7 mm I.D. and total volume of 358 mL) using the two-phase solvent system consisting of MeOH:H₂O:EtOH:Hex:MTBE 6:3:1:6:4 (v/v). For pumping of both phases, collection and monitoring of effluent during separation, Prep-HSCCC was coupled to an Isolera One flash purification system (Biotage®, Uppsala, Sweden), equipped with a quaternary solvent manager (QSM), automated fraction collector, internal detector operating in UV range (200–400 nm), and sampling loop with maximum volumetric capacity of 12 mL. The previous saturation between the system phases was made in a separation funnel by vigorous stirring repeatedly at a temperature of 25 °C. After equilibration, the two phases were separated again in different flasks and degassed by sonication for 10 min before use. AP was used as stationary, and OP, as mobile. First, column-1 was completely filled with AP at a flow of 10 mL/min in rotation at 200 rpm. The sample was dissolved in 12 mL of an AP/OP mixture (1:1, v/v) and injected into the loop to be propelled into the column along with the stationary phase. After the column was completely filled, with AP containing the sample, OP was then eluted. The elution process of mobile phase took place in tail-head direction at an OP flow of 5 mL/min. The system rotation speed was 800 rpm and the temperature was maintained at 32 °C. The effluent was monitored at wavelengths of 254 and 320 nm. The collection volume of fractions was 8 mL/tube. OP was made up of Hex/MTBE in the ratio of 6:4 (v/v), and AP, composed of MeOH/H₂O/EtOH in the ratio of 6:3:1 (v/v). After the end of separation, the rotation was stopped and the stationary phase was pushed out of the column. Its content was collected, and the volume was measured in order to calculate the retention rate of the stationary phase.

2.7 Analysis by HPLC-PDA

For studies of REE and fractions obtained in separation by HSCCC chemical profile, it was used a modular high-performance liquid chromatography system: Proeminence Ultra fast liquid cromatograph (UFLC), model LC-20A (Shimadzu, Tokyo, Japan), consisting of binary pump, DGU-20A online degasser, SIL-20A auto-sampler and auto-injector, CTO-20A thermostatic column compartment, and photodiod array detection system (PDA) SPD-M20A (Shimadzu, Tokyo, Japan). Separations occurred in a RP-18, 5 µm, 2.0 × 150 mm, model Shim-pack VP - ODS (Shimadzu, Tokyo, Japan) column, kept in an oven at 28 °C. The absorption spectra were recorded in 200–800 nm range, and the chromatograms, processed at 254 nm. The mobile phase consisted of 0.1% formic acid in water (solution A), and ACN (solvent B). The separation occurred in gradient mode, with variation of 55–85% of the solvent B amount, in the range of 0–45 min and flow of 0.189 mL/min. The Volume injected per sample was 5 μL.

2.8 Analysis by HPLC-ESI-MS

The determination of molecular masses and identification of compounds present in the fractions samples collected in the separation was carried out in a mass spectrometer coupled to HPLC system, model LCMS-2010 EV (Shimadzu, Tokyo, Japan), with single-stage quadrupole analyzer and electrospray ionization source (ESI). The m/z analyzer operated in the acquisition mode of full scan of positive and negative ions, in the range of m/z 200 – 800, with speed of 1000 amu/s. The ESI source operated under the following conditions: flow of nebulizer gas (N₂) of 1.5 L/min; capillary voltage, (+) 4.5 kV, (−) 3.5 kV; heater block temperature, 200 °C; curved desolvation line temperature (CDL), 230 °C. The chromatographic operating conditions of HPLC were the same as described in section 2.7 for all samples. The acquisition and processing of data generated in the analysis by HPLC-PDA-MS was done by the workstation and data management LCMSsolution (Shimadzu, Tokyo, Japan).

2.9 Analysis by NMR

The ¹H and ¹³C Nuclear Magnetic Resonance (NMR) spectra were acquired on a Mercury 300 spectrometer (Varian, Inc., Palo Alto, USA), operating at 300 (¹H) and 75 MHz (¹³C). The samples were dissolved in deuterated chloroform (CDCl₃) or deuterated methanol (MeOD). Chemical shifts (δ) were expressed having as reference the solvent internal signal CHCl₃ (¹H, δ 7.25; ¹³C 77.0) or CH₃OH (¹H, δ 3.30; ¹³C, 49.0).

2.10 Evaluation of in vitro antioxidant capacity

2.11 EC₅₀ determination

EC₅₀ values of samples and standards were estimated by using a nonlinear regression model built by graphical plotting of inhibition percentage results in function of samples concentration logarithm. Results were subjected to statistical analysis by using GraphPad Prism® software version 8.0.1 (GraphPad Software Inc. – San Diego, CA, USA). Data were treated by non-linear regression by adopting a logistic equation with 5 parameters, and the result, plotted according to dose-response inhibition of antioxidant capacity of patterns and fractions. Data were analyzed for statistical differences between treatment groups and estimated by using the Spearman (r) correlation. All statistical tests were performed with level significance with p < 0.05.

3.1 Selection of extract rich in α- and γ-mangostin

The de-replication of extracts, CE, EaE, EtE and Me, was performed by A-TLC and HPLC-PDA-MS to select the one that had the highest relative concentration of the main bioactive xanthones, α- and γ-mangostin, targeting fractionation to the isolation of both with maximum yield without the need for pre-treatment or previous fractionation of the sample, reducing the number of steps and making the procedure simpler, faster, cleaner, and cheaper.

The extracts samples were then subjected to analysis by A-TLC and to phytochemical detection tests for the characterization and comparison of chromatographic profiles (Hellmut et al., 1990; Kinghorn, 1997). The rapid identification and structural characterization of constituents of the extract selected for fractionation was carried out by HPLC-PDA-MS and it is described in Sections 3.5 and 3.6. Fig. 1 illustrates the image of a chromatographic plate with profiles of the extracts samples, revealed under the exposure of 254 nm radiation, after the chromatography development.

Close

Fig. 1 Chromatographic profiles of extracts samples on A-TLC plate exposed to 254 nm radiation; CE, cloroformic extract; EaE, ethylacetic extract; EtE, ethanolic extract; ME, methanolic extract; Retention factor (R_f): 0.66, α-mangostin; 0.54, γ-mangostin.

Export to PPT

The different patterns of bands produced and detected on the plate surface (Fig. 1) reveal the differences in composition between the extracts due to the polarity degree of the solvent used as extracting phase (Suttirak and Manurakchinakorn, 2014). As can be seen in the figure (Fig. 1), the CE and ME profiles differ significantly from the other extracts. It is possible to observe in the figure (Fig. 1) that there was no migration of the ME sample applied to the plate, that is, the sample remained fully retained in the base. The lack of migration can be explained by the sample being constituted, predominantly, of molecules of high polarity and molecular weight, such as tannins and their glycosides, characterized by reactions with the appropriate detection reagents; the pattern and the number of bands observed in the CE profile shows that its composition is more complex, however, being constituted by a major substance, identified as α-mangostin (Rf 0.66), indicating that chloroform has a high selectivity for the removal of this component. EtOAc and EtOH solvents, according to the observed profiles of EaE and EtE (Fig. 1), can extract substances from a wide range of polarities, such as the main xanthones and other present phenolic compounds. According to Fig. 1, EtE presented the highest metabolic diversity with a higher concentration of analytes, α- (R_f 0.66) and γ-mangostin (R_f 0.54), being the extract chosen for fractionation. The choice of the appropriate solvent and the extraction method employed are crucial for obtaining natural products comprehensively and in maximized quantities, that is, matrices that are representative of the totality of metabolites present in the natural source (Hubert et al., 2017). The preliminary study of the metabolic profile carried out by A-TLC was fundamental in deciding on which of the solvents should be used in the preparation of an extract that would offer a greater metabolic diversity and a greater concentration of analytes, enabling their isolation with maximum yield.

3.2 Selection and optimization of separation conditions by Prep-HSCCC

Proper separation by Prep-HSCCC depends mainly on the correct choice of a two-phase solvent system that produces values of 0.5 < K < 2.0 and α > 1.5 between analytes, with K value being the most important parameter for determining the resolution of a CCS (Friesen et al., 2015; Ito, 2005; Luo et al., 2016). Initially, two two-phase solvent systems from the “Arizona” family were tested, consisting of Hex, EtOAc, MeOH and water (HEMWat), suggested by Ito (2005) and Friesen et al. (2015) as the initial choice for a preliminary study on the polarity and solubility of analytes. Six different versions of this system, with different proportions between solvents, were tested with unsatisfactory results regarding K values for the analytes. So, four new two-phase systems, very similar in composition, were created and tested, showing better results compared to the initial system. The compositions of the systems and the respective K and α values of the analytes are described in Table 1. The flow rate of the mobile phase and the speed of revolution of the coil set are two important parameters for obtaining peaks with good resolution. The flow rate delimits the separation time and the amount of stationary phase retained in the column; low speeds of rotation decrease the retention rate and very high speeds produce a widening of the peaks due to system vibration as a result of violent pulsation in the column generated by high pressure (Ito, 2005). The flow of mobile phase and the speed of revolution adopted in this work were adjusted within the range of values suggested by Ito (2005) as the most usual practiced in separations in multilayer coil systems with separation preparatory columns: 5–6 mL/min; 600–800 rpm. Another important factor in obtaining a good resolution and achieving an efficient separation is the retention rate of the stationary phase in the column, which will be greater the shorter the separation or accommodation time between the two phases in the test tube. For all systems shown in Table 1, the separation time between the phases was between 10 and 20 s, that is, within the recommended limit for a satisfactory retention rate, above 50% (Ito, 2005). However, still in the testing phase in test tubes, in systems 1 and 2, the formation of precipitates was observed. This phenomenon can be explained by the high volatility of diethyl ether, which is lost continuously by vaporization at room temperature (25 ± 2 °C). The formation of precipitates promotes a constant change in the composition of these systems due to the loss of some components that become insoluble and leave the solution due to precipitation, causing compromised separation efficiency due to loss of resolution, decrease in retention rate and lack of repeatability (Ito, 2005; Luo et al., 2016; Shan and Zhang, 2010). Diethyl ether was then replaced by MTBE, thus, solving the problem of volatility and making the system more stable. According to results shown in Table 1, system 4 presented satisfactory K and α values for production of efficient separation, being the system chosen for isolation of α- and γ-mangostin by HSCCC. The effect of different amounts of sample loading on the separation efficiency was also evaluated. The introduction of large amounts of sample in the system can lead to large losses in the stationary phase and compromise the resolution of peaks related to analytes, especially those with small K values. However, the injection of small amounts of sample may reduce the yield of products (Friesen et al., 2015; Ito, 2005; Li et al., 2017). Samples were injected with 250, 450 and 650 mg of REE, dissolved in 12 mL of a two phases mixture (1:1, v/v), and results produced were evaluated. The chromatographic separation of sample with 650 mg of REE presented a resolution of peaks, referring to analytes, very similar to those obtained for other two quantities of samples investigated, however, its yield was significantly higher with a small decrease in the retention rate of stationary phase, about 5%.

Results obtained in the study of selection and optimization of experimental conditions of operational parameters of separation by HSCCC were used in the fractionation of REE for analytes isolation. Therefore, experimental conditions practiced in the separation were as follows: solvent system composed of MeOH:H₂O:EtOH:Hex:MTBE 6:3:1:6:4 (v/v); tail-head elution direction (normal phase); revolution speed of 800 rpm; mobile phase flow of 5 mL/min; loading of 650 mg of REE sample.

3.3 Fractionation of REE by Prep-HSCCC

Phenolic constituents of REE were separated by Prep-HSCCC, in a single step, in 35 min range. The fractionation of 650 mg of REE initially produced 195 fractions. The fractions were monitored by A-TLC for the identification and assembly of those with similar chemical profiles. The assembly procedure resulted in only four fractions, referred to as F1, F2, F3 and F4. The fractionation chromatogram of REE by Prep-HSCCC is shown in Fig. 2. The blue trace observed in the chromatogram (Fig. 2), which delimits the area between 12 and 70 min, indicates the elution time of mobile phase. F1 and F2 fractions were produced between 12 and 20 min elution range. Peaks identified as F3, pink region; and F4, green region, were registered between 20 and 47 min and correspond to the 3 and 4 resulting fractions, respectively. It was observed that the mass quantity obtained for each fraction in this separation was 55.2 mg of F1; 56.3 mg of F2; 222.6 mg of F3; and 65.1 mg of F4. The method achieved a recovery above 61%, and the total separation yield was 399.2 mg.

Close

Fig. 2 Chromatograms for Raw Ethanolic Extract (REE) fractionation by HSCCC: red line (254 nm); black line: (320 nm); solvent system: MeOH:H₂O:EtOH:Hex:MTBE (6:3:1:6:4, v/v); rotation speed: 800 rpm; flow: 5 mL/min; sample mass: 650 mg; stationary phase retention: 80%; F3: α-mangostin; F4: γ-mangostin.

Export to PPT

3.4 Prep-HSCCC system accuracy

The accuracy of chromatographic system for REE fractionation and xanthones isolation was tested by successive injection of sample on five consecutive days. The relative standard deviation (RSD) for mass values of fractions F1 to F4, obtained in the five injections were 5.2%, 6.5%, 3.8% and 4.5%, respectively.

3.5 Characterization of REE and fractions chromatographic profile

The characterization of REE and fractions metabolic profile was done through samples analysis by HPLC-PDA-MS. Compounds referring to peaks, numbered from 1 to 9, in REE chromatogram (Fig. 3, A), are listed in Table 2 with their respective retention times (t_R), spectral data of maximum absorption in UV–Vis (λ_max) and protonated ([M+H]⁺) and deprotonated ([M−H]^-) molecules.

Close

Fig.3 Chromatograms by HPLC-PDA processed at 254 nm: (A) raw ethanolic extract (REE); (B) fraction F3, α-mangostin with purity above 96%; (C) fraction F4, γ-mangostin with purity above 93%.

Export to PPT

The identity of each detected compound was determined by comparing instrumental data generated in the analysis with data already reported in literature for GML xanthones (Fang et al., 2011; Shan and Zhang, 2010; Walker, 2007; Wittenauer et al., 2012). The elution order of detected species, recorded by the peaks in REE chromatogram (Fig. 3, A), also contributed to the identities confirmation. The elution order was exactly the same as that observed in other studies involving studies of characterization and separation of xanthones from GML by HPLC that used reverse phase column (RP-18) and mobile phase composed of acidified water/ACN (Aisha et al., 2012; Fang et al., 2011; Muchtaridi et al., 2017; Shan and Zhang, 2010; Walker, 2007; Wittenauer et al., 2012). The structures of xanthones identified in REE are shown in Fig. 4.

Close

Fig. 4 Structures of xanthones identified in the Garcinia mangostana L. pericarp raw ethanolic extract by LC-PDA-MS: xanthone nucleus; Mangostanol, 1; 1,7- dihydroxy-3-methoxy-2-prenylxanthone, 2; γ-mangostin, 3; 8-Deoxigartanin, 4; Gartanin, 5; α-mangostin, 6; Garcinone E, 7; 9- Hydroxycalabaxanthone, 8; β-mangostin, 9.

Export to PPT

Results obtained and summarized in Table 2 confirmed the identities of α-mangostin as the most abundant species present in F3, referring to peak 6 (Fig. 3, B), with 96.3% purity; and γ-mangostin, present in F4, referring to peak 3 (Fig. 3, C), with 93.2% purity. Purities were calculated by normalization of peak areas corresponding to substances in the chromatograms and are expressed as percentages of relative area. These results demonstrate the high resolution power of HSCCC and the efficiency of the method that was able to promote the isolation of analytes in just under 35 min, reaching levels of purity very close to those obtained in other works of CCS reported in the most recent literature (Fang et al., 2011; Michel et al., 2012; Shan and Zhang, 2010). The chromatograms of F1 and F2 fractions are shown in Fig. S1A (section S1, supplementary material). Results of F1 analysis (Fig. S1A, D) showed that 9-hydroxycabalaxantone and β-mangostin, peaks 8 and 9, respectively, are present in predominant quantities, corresponding to 38.7 and 25.0% of the total in mass, respectively. These molecules have very similar structures and low degrees of polarity, which is why they were the first to be eluted by the organic mobile phase in tail-head elution mode (Shan and Zhang, 2010). The profile of F2 (Fig. S1A, E) shows that it is mainly composed of a mixture of two major constituents, gartanine and α-mangostin, corresponding to 13.8% and 55.1% by mass, peaks 5 and 6, respectively. In REE chromatogram, peaks 3 and 6, referring to α- and γ-mangostin, have relative percentage areas of 12.3 and 48.9%, respectively.

3.6 Structural identification of α- and γ-mangostin

The structural determination of the isolated substance in F3 was performed from ESI-MS, ¹H and ¹³C NMR spectral data generated in the analysis and by comparison with data registered in the literature (Fang et al., 2011; Michel et al., 2012). The ESI-MS data for the substance referring to peak 6, in the F3 chromatogram, are m/z 411 [M + H]⁺(100), 409 [M − H]⁻(100); NMR ¹H (CDCl₃, 300 MHz) δ: 1.62, 1.63 (3H each, s, H-4″ e H-4′), 1.75 (3H, s, H-5′), 1.83 (3H, s, H-5″), 3.44 (2H, d, J = 7.2 Hz, H-1′), 3.80 (3H, s, OCH₃), 4.09 (2H, d, J = 6.0 Hz, H-1″), 5.28 (2H, m, H-2′e H-2″), 6.27 (1H, s, H-4), 6.80 (1H, s, H-5), 13.75 (1H, s, C-1-OH); ¹³C NMR (CDCl₃, 75 MHz) δ: 17.90 (C-5″), 18.20 (C-5′), 21.42 (C-1′), 25.80 (C-4″), 25.83 (C-4′), 26.55 (C-1″), 62.03 (OCH₃), 93.27 (C-4), 101.54 (C-5), 103.60 (C-9a), 108.43 (C-2), 112.16 (C-8a), 121.43 (C-2′), 123.12 (C-2″), 132.15 (C-3″), 135.75 (C-3′), 137,02 (C-8), 142.53 (C-7), 154.52 (C-4a), 155.05 (C-6), 155.76 (C-10a), 160.59 (C-1), 161,61 (C-3), 182.02 (C-9). ¹H NMR and ¹³C NMR spectral data coincided with data in the literature, confirming that the substance isolated in F3 was α-mangostin. The identity of the compound isolated in F4 was determined from its spectral data from ESI-MS, ¹H NMR and ¹³C NMR. ESI-MS data are m/z 397 [M+H]⁺(100), 395 [M−H]⁻(100); ¹H NMR (Methanol-D4, 300 MHz) δ: 1.64, 1.66 (3H each, s, H-4″ e H-4′), 1.77 (3H, s, H-5′), 1.82 (3H, s, H-5″), 3.32 (2H, d, J = 7.2 Hz, H-1′), 4.16 (2H, d, J = 6.0 Hz, H-1″), 5.27 (2H, m, H-2′ e H-2″), 6.38 (1H, s, H-4), 6.80 (1H, s, H-5), 13.92 (1H, s, C-1-OH). ¹³C NMR (Methanol-D4, 75 MHz) δ: 17.20 (C-5″), 17.92 (C-5′), 21.30 (C-1′), 25.26 (C-4″), 25.40 (C-4′), 26.20 (C-1″), 92.20 (C-4), 100.30 (C-5), 102.60 (C-9a), 110.2 (C-2), 111.3 (C-8a), 122.90 (C-2′), 123.90 (C-2″), 130.6 (C-3′), 130.8 (C-3″), 143.5 (C-7), 153.60 (C-4a), 154.80 (C-6), 155.70 (C-10a), 161.0 (C-1), 162.0 (C-3), 182.4 (C-9). NMR data are in agreement with data registered in other works in the literature (Fang et al., 2011; Michel et al., 2012), confirming that the compound isolated in F4 is γ-mangostin.

3.7 Antioxidant potential of the fractions obtained

EC₅₀ results that represent the antioxidant potential of fractions and patterns in the capture of DPPH^• and ABTS^•+ radicals are shown in Table 3. When evaluating the ability to sequester DPPH^• radicals, the results obtained for F2 and F4 stand out, with EC₅₀ of 8.42 and 8.43 µg/mL, respectively. By comparing the results, it is possible to conclude that F2 and F4 have the capacity to scavenge DPPH^• radicals almost nine times greater than the measured capacity for F3, and three times greater than F1. The variation in the antioxidant capacity of fractions can be attributed to the profile of mangostin present in majority quantities, corroborating the existence of a high correlation between the chemical structure and the activity of these xanthones (Scherer and Godoy, 2009; Suttirak and Manurakchinakorn, 2014; Thong et al., 2015). As for the low antioxidant capacity exhibited by F3 compared to DPPH^• radical, Tjahjani et al. (2014) reported a similar result of EC₅₀ for α-mangostin. In comparison with EC₅₀ values obtained for different pericarp extracts, prepared by using different solvents as extracting phase, α-mangostin exhibited a low antioxidant capacity, however, when measured by means of other activity assays, such as Superoxide Dismutase (SOD) and Total Antioxidant Status (TAS), it exhibited greater activity than the extracts studied. This effect suggests that α-mangostin can act as an antioxidant through another mechanism besides the transfer of hydrogen atoms (HAT) (Tjahjani et al., 2014). The reduction of a hydroxyl group (OH) in α-mangostin is the main factor responsible for decreasing the capacity of HAT in relation to the activities of γ-mangostin and gartanine, which have the same number of OH groups in the two rings, however, in the gartanin, the prenyl groups are together in ring B (Fig. 4), decreasing the steric impediment on the OH groups in ring A, in addition to favoring HAT, since these groups are p-substituted in this ring, which decreases the strength of the O–H bond and contributes to the formation of a stable intermediate radical by displacement of the unpaired electron (Suttirak and Manurakchinakorn, 2014). Due to the lack of standardization in procedures for evaluating the antioxidant capacity of extracts and pure compounds by using DPPH^• method, and due to the dependence of inhibition activity and EC₅₀ values on the final concentration of the radical, the activities expressed in AAI values allow an result evaluation more accurate, in terms of comparison (Scherer and Godoy, 2009; Sridhar and Charles, 2019). The final concentration of DPPH solution after adding the samples was 45.0 µg/mL. Calculated AAI values for gallic acid and fractions, F1, F2, F3 and F4, were 26.47, 1.75, 5.34, 0.63 and 5.33, respectively. The AAI value calculated for gallic acid coincided with the value obtained by Scherer and Godoy (2009). In this work, authors proposed a scale for classifying the antioxidant power of samples tested by DPPH^• method based on the AAI value. The AAI value for F3 fraction indicates that it exhibited moderate activity; F1 fraction exhibited strong activity; and fractions F2 and F4 have very strong activity.

Figure S2 (Section S1, supplementary material) presents graphs of the inhibition activity (%) of DPPH^• (a) and ABTS^•+ (b) radicals in function of samples concentration (µg/mL).

The fractions collected and the tested patterns showed greater antioxidant efficiency in the ABTS^•+ radical sequestration, with lower EC₅₀ values, when compared with values found for capture of DPPH^• radical. (Table 3) However, it is observed that the relative differences between the activity values for samples and standards remained constant. Again, the sequestering capacity of F2 and F4 stands out, with EC₅₀ values of 4.54 and 3.31 µg/mL, respectively. The discrepancy observed between the results can be explained in part by the use of different methods to estimate the antioxidant capacity of the fraction samples, by the differences in the performance of the procedure and the use of different reagents in each method (Villaño et al., 2006). However, although both assays are based on electron transfer mechanisms between antioxidant compounds and radicals, according to Sridhar and Charles (2019), the differences in the stereochemistry of ABTS^•+ and DPPH^• radicals, as well as the solubility, polarity and metal-chelating capacity of these compounds may affect the results and generate the observed differences. The antioxidant activity tests demonstrated the pharmacological, economic and technological potential of fractions obtained from GML waste. Patterns of α- and γ-mangostin, with purity levels close to those obtained in this work are sold on market at average prices of USD 30.0/mg. The use of mangosteen waste for the extraction and obtaining of organic molecules with useful properties in combating effects of oxidative stress and related diseases, such as premature aging, diabetes, cancer and other associated diseases, fits into new technologies for minimizing environmental impact with prospects for generation of new molecules with noble applications.