1 an Integrated Approach for the Valorization of Mango Seed Kernel: Efficient Extraction

1 An integrated approach for the valorization of mango seed kernel: efficient extraction

2 solvent selection, phytochemical profiling and antiproliferative activity assessment.

4 Diego Ballesteros-Vivas1,2a, Gerardo Alvarez-Rivera2a, Sandra Johanna Morantes Medina3

5 Andrea del Pilar Sánchez Camargo1, Elena Ibánez2, Fabián Parada-Alfonso1, Alejandro

6 Cifuentes2*

8 1 High Pressure Laboratory, Department of Chemistry, Faculty of Science, Universidad

9 Nacional de Colombia, Carrera 30 #45-03, Bogotá D.C., 111321, Colombia.

10 2 Laboratory of Foodomics, Institute of Food Science Research, CIAL, CSIC, Nicolás Cabrera

11 9, 28049 Madrid, Spain.

12 3 Unit of Basic Oral Investigation (UIBO), School of Dentistry, Universidad El Bosque, Av.

13 Carrera 9 #131 A-02, Bogotá D.C., 110121, Colombia.

14 15 a These two authors contributed equally to this work.

17 *Corresponding author:

18 Prof. Dr. Alejandro Cifuentes, Laboratory of Foodomics, Institute of Food Science Research,

19 CIAL (CSIC), Nicolás Cabrera 9, 28049 Madrid, Spain, e-mail: [email protected], Tel.: +34

20 910017955; fax: +34 910017905.

22 Keywords:

23 Mangifera indica L.; fruit by-products; Hansen solubility parameters; Pressurized-liquid

24 extraction; LC-Q-TOF; GC-Q-TOF; High-resolution mass spectrometry; antiproliferative

25 activity; HT-29 cell line; CCD-18Co cell line.

27 ABSTRACT

28 A novel valorization strategy is proposed in this work for the sustainable utilization of a major

29 mango processing waste (i.e. mango seed kernel, MSK), integrating green pressurized-liquid

30 extraction (PLE), bioactive assays and comprehensive HRMS-based phytochemical

31 characterization to obtain bioactive-rich fractions with high antioxidant capacity and

32 antiproliferative activity against human colon cancer cells. Thus, a two steps PLE procedure

33 was proposed to recover first the non-polar fraction (fatty acids and lipids) and second the polar

34 fraction (polyphenols). Efficient selection of the most suitable solvent for the second PLE step

35 (ethanol/ethyl acetate mixture) was based on the Hansen solubility parameters (HSP) approach.

36 A comprehensive GC- and LC-Q-TOF-MS/MS profiling analysis allowed the complete

37 characterization of the lipidic and phenolic fractions obtained under optimal condition (100%

38 EtOH at 150°C), demonstrating the abundance of oleic and stearic acids, as well as bioactive

39 xanthones, phenolic acids, flavonoids, gallate derivatives and gallotannins. The obtained MSK-

40 extract exhibited higher antiproliferative activity against human colon adenocarcinoma cell line

41 HT-29 compared to traditional extraction procedures described in literature for MSK utilization

42 (e.g. Soxhlet), demonstrating the great potential of the proposed valorization strategy as a

43 valuable opportunity for mango processing industry to deliver a value-added product to the

44 market with health promoting properties.

52 1. INTRODUCTION

53 Mango (Mangifera indica L.) is one of the most important tropical fruit crops, with an annual

54 production of more than 38 million tonnes (Mitra, 2016). The commercial importance of mango

55 fruit is due, among other reasons, to its sensorial quality attributes, high nutritional value and

56 functional compounds content (Ediriweera, Tennekoon, & Samarakoon, 2017; Gentile et al.,

57 2019; Ribeiro & Schieber, 2010). Colombia plays an increasing role in world mango

58 production with cultivars such as ‘Sugar mango’, recognized by its sensorial qualities

59 (Corrales-Bernal, Maldonado, Urango, Franco, & Rojano, 2014). The industrial mango

60 processing generates about 40–60% of fruit wastes (12–15% of peels and 15–20% of kernels

61 seeds); none of them currently used for commercial purposes (Nawab, Alam, Haq, & Hasnain,

62 2016). Recently, several researches about the chemical composition and bioactive potential of

63 mango seed kernel (MSK) have been reviewed (Jahurul et al., 2015; Torres-León et al., 2016).

64 MSK contains important families of health-promoting compounds including fatty acids and

65 triacylglycerols (Lieb et al., 2018), gallotanins (Luo et al., 2014), xanthones (e.g. mangiferin)

66 (Barreto et al., 2008), flavonoids and phenolic acids, among others (Dorta, González, Lobo,

67 Sánchez-Moreno, & de Ancos, 2014; Lopez-Cobo et al., 2017). Polyphenolic compounds from

68 mango have been reported to have a strong antioxidant activity (Barreto et al., 2008; Soong &

69 Barlow, 2006; Sultana, Hussain, Asif, & Munir, 2012), and exhibit bioactivity in cancer cell

70 line models, including breast, liver, leukemia, cervix, prostate, lung and colon (Abdullah,

71 Mohammed, Rasedee, & Mirghani, 2015; Abdullah, Mohammed, Rasedee, Mirghani, & Al-

72 Qubaisi, 2015; Luo et al., 2014; Timsina & Nadumane, 2015). In particular, mangiferin (2-β-

73 D-glucopyranosyl-1,3,6,7-tetrahydroxy-9H-xanthen-9-one) has been reported as one of the

74 most bioactive phytochemicals in mango; in both in vitro and in vivo models (Imran et al.,

75 2017).

76 Considering the bioactive potential of MSK, the development of green valorization strategies

77 to obtain polyphenolic-rich extracts from this valuable biowaste, pose a great challenge and a

78 unique opportunity for mango processing industry to deliver a value-added product to the

79 market with health promoting properties. Thus, strategies based on efficient extraction solvent

80 selection and use of new green extraction processes can help fulfilling the goals of the green

81 extraction of natural products (Chemat, Vian, & Cravotto, 2012). Hansen solubility parameters

82 (HSP) was shown to be a useful predictive model to ascertain the solubility of solutes, such as

83 secondary metabolites, in different solvents through their affinity and miscibility estimation.

84 In terms of new green extraction processes, Pressurized Liquid Extraction (PLE) is a

85 recognized environmentally friendly technique due to its higher extraction efficiency, lower

86 solvent consumption, short extraction time and the possibility of using green solvents (Ameer,

87 Shahbaz, & Kwon, 2017; Herrero, Castro-Puyana, Mendiola, & Ibañez, 2013). Several

88 research works have been conducted employing the joint strategy involving HSP+PLE to

89 target bioactive compounds recovery from natural sources (Ballesteros-Vivas et al., 2019;

90 Damergi et al., 2017; Sánchez-Camargo et al., 2017; Srinivas, King, Monrad, Howard, &

91 Hansen, 2009).

92 In this context, the present research aimed to develop an integrated valorization strategy,

93 involving HSP approach and sequential PLE procedure, in vitro antioxidant assays and

94 comprehensive characterization with advanced analytical techniques (liquid chromatography

95 and gas chromatography coupled to high resolution mass spectrometry) to obtain mangiferin

96 and other phenolic compounds from ‘sugar MSK’ with selective antiproliferative activity

97 against human colon adenocarcinoma cell line HT-29. An integrated process scheme of the

98 proposed MSK valorization strategy is shown in Figure 1.

100

101 2. MATERIAL AND METHODS

102 2.1 Samples and reagents

103 Sugar mango fruits were purchased from a local market in Bogotá D.C., Colombia in February

104 2018. Mango fruit by-products were obtained after mechanical pulping process. Seeds were

105 split into coat and kernel (endosperm). ‘Sugar MSK’ (5.3% moisture content) was dried at

106 room temperature in the darkness during 48 h, subsequently ground to fine powder and stored

107 at -20 °C until its use.

108 Gallic acid, quercetin, trolox, 2,2'-azino-bis(3-ethylbenzothiazoline-6-sulphonic acid) (ABTS),

109 2,2-diphenyl-1-picrylhydrazyl (DPPH), 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium

110 bromide (MTT), RPMI-1640 cell culture medium, streptomycin (0.1 mg/mL), penicillin (100

111 U/mL) potassium acetate, ammonium acetate, sodium carbonate, formic acid, potassium

112 persulfate, aluminum chloride, were purchased from Sigma-Aldrich (Madrid, Spain). Fetal

113 bovine serum (Gibco) and 0.05% trypsin-EDTA (Gibco) were purchased from Thermo Fisher

114 Scientific (Rockford, IL). Merck (Darmstadt, Germany) provided the Folin-Ciocalteu phenol

115 reagent. Solvents employed were HPLC-grade. Acetonitrile, chloroform, ethanol and methanol

116 were acquired from VWR Chemicals (Barcelona, Spain), whereas ethyl acetate by Scharlau

117 (Barcelona, Spain). Ultrapure water was obtained from a Millipore system (Billerica, MA,

118 USA). For the UPLC-q-TOF-MS analyses, MS grade ACN and water from LabScan (Dublin,

119 Ireland) were employed.

120

121 2.2 Hansen Solubility Parameters estimation

122 HSP for mangiferin and green solvents, including ethanol, ethyl acetate, ethyl lactate and (+)-

123 limonene, were estimated using HSPiP® software v 5.0 at normal conditions, following the

124 methodology previously reported by Sánchez-Camargo et al (Sánchez-Camargo et al., 2017).

125 Briefly, the SMILES (Simplified molecular input line syntax) of mangiferin

126 [C1=C2C(=CC(=C1O)O)OC3=CC(=C(C(=C3C2=O)O)C4C(C(C(C(O4)CO)O)O)O)O] was

127 break into corresponding functional groups using Yamamoto-molecular break (Y-MB) method

128 and then HSP parameters were estimated by “Do It Yourself” tool. Subsequently, the affinity

129 between the mangiferin and the green solvents was measured by Ra or “distance” term using

130 their HSPs values through “Solvent optimizer” tool (the smaller Ra corresponding to the greater

131 affinity between solvent and solute). The variation of Ra at different temperatures (25-150 °C)

132 was also studied. For this purpose, the temperature dependence of mangiferin solubility

133 parameters was estimated by Jayasri and Yaseen method (Jayasri & Yaseen, 1980), employing

134 the critical data obtained by Marrero & Gani group contribution method (Marrero & Gani,

135 2001). The temperature effect on HSPs of green solvents was evaluated by the Gunn-Yamada

136 (Pereira, Silva, & Rodrigues, 2011) and Williams et al. (Williams, Rubin, & Edwars, 2004)

137 methods. Finally, Ra between mangiferin and green solvents were estimated at different

138 temperatures.

139

140 2.3 Pressurized liquid extraction (PLE)

141 A commercial ASE 200 device (11 mL stainless steel cells) was used for PLE process in two

142 steps. For each extraction, ‘sugar MSK’ samples and sea sand were mixed in a 1:2 w/w

143 proportion. The mixture was extracted on static mode at 100 bar. After the extraction, the

144 solvent was removed by evaporation with continuous stream of gaseous nitrogen. Extraction

145 yield was expressed as g of extract/100 g dry weight basis of sample (mean of duplicate).

146

147 2.3.1. PLE- first step evaluation

148 Due to ‘sugar MSK’ fat content, a first defatting step was required in order to recover the fat

149 while cleaning the sample for polyphenolics’ extraction. Three “alternative and usable”

150 solvents were tested to avoid the use of n-hexane: n-heptane, cyclohexane and (+)-limonene.

151 n-Hexane was used as reference nonpolar solvent. In order to achieve the maximum defatting

152 of ‘sugar MSK’, kinetics extraction curves for each nonpolar solvent were studied at 100 °C

153 and 100 bar for 90 min.

154

155 2.3.2. PLE-second step optimization

156 The polyphenolic compounds recovery, including mangiferin, from ‘sugar MSK’ after the

157 defatting process was optimized using a three-level face-centered central composite design

158 (CCD). The effect of temperature (50-150 °C) and green solvent composition (according to

159 HSP results) were investigated on mangiferin content, extraction yield, total phenolic content,

160 total flavonoid content and antioxidant activity. Experimental data was fitted with the following

161 second order polynomial equation:

162 , , , Eq. (1)

163 where is the response variable, and , , , , , , and , are regression

164 coefficients of variables for intercept, linear, quadratic, and interaction terms, respectively, and

165 and are the independent variables, representing solvent composition and temperature,

166 respectively. The adequacy of the model was determined by coefficient of regression (R2) and

167 the F-test value obtained from the analysis of variance (ANOVA) by the statistical software

168 STATISTICA 12 (Stat Soft, Inc., Tulsa, OK 74104, USA). Pareto charts for the standardized

169 effects of independent variables on response factors were also generated. A multiple response

170 optimization was carried out by combining the experimental factors, looking for maximizing

171 the desirability function (Ballesteros-Vivas et al., 2019).

172

173 2.4 Conventional extractions

174 Conventional solvent extractions Bligh & Dyer (Bligh & Dyer, 1959; Breil, Abert Vian, Zemb,

175 Kunz, & Chemat, 2017) (B&D) and dynamic maceration (DM) were used for comparison as

176 standards to determine the efficiency of the PLE-first and -second steps, respectively. For the

177 B&D method, a ‘sugar MSK’ sample and 2:1 chloroform:methanol (v/v) mixture were

178 homogenized for 2 min. Then, chloroform and water were added to get a final ratio of 2:2:1.8

179 chloroform:methanol:water (v/v/v). The mixture was shaken vigorously for 2 min and the final

180 biphasic system was then separated by centrifugation (10 min at 2000 rpm). The lower

181 chloroform phase layer was collected and the solvent was evaporated under a stream of

182 nitrogen, in order to recover the lipid content. Moreover, DM was performed using green

183 solvents according to HSPs results. Samples of ‘sugar MSK’ and solvent were mixed in a

184 proportion of 1:5 (w/v) and kept in agitation at 750 rpm, 25 ºC for 24 h. Subsequently, the

185 extract was separated from the sample by centrifugation (20 min at 5000 rpm) and the solvent

186 was evaporated under a stream of nitrogen.

187

188 2.5 Determination of total phenolic content (TPC)

189 TPC was determined by the Folin–Ciocalteu (Hosu, Cristea, & Cimpoiu, 2014) method with

190 slight modifications. Gallic acid (0–100 µg/mL) was used for calibration of a standard curve.

191 10 µL of extract solution, 600 µL of water and 50 µL of Folin-Ciocalteu reagent (0.2 M), were

192 mixed. After 5 min, 150 μL of Na2CO3 (20% w/v) and 190 µL of water were added. After 120

193 min for allowing the reaction to take place, the absorbance was measured at 760 nm using a

194 microplate spectrophotometer reader (Synergy HT, BioTek Instruments, Winooski, VT, USA).

195 The results were expressed as milligrams of gallic acid equivalents per gram of dry weight

196 basis (mg GAE/g Db) as mean of three replicates.

197

198 2.6 Determination of total flavonoid content (TFC)

199 TFC was estimated by Aluminium chloride colorimetric method (Hosu et al., 2014) with slight

200 modifications. Quercetin (0–100 µg/mL) was used for calibration of a standard curve. 100 µL

201 of extract solution, 30 µL of AlCl3 (10% w/v), 30 µL of potassium acetate (1 M), 300 µL of

202 EtOH and 540 µL of water were mixed and incubated for 30 min. Then, the absorbance was

203 measured at 415 nm using a microplate spectrophotometer reader. The results were expressed

204 as milligrams of quercetin equivalents per gram of dry weight basis (mg QE/g Db) as mean of

205 three replicates.

206

207 2.7 Antioxidant capacity assays

208 2.7.1 DPPH assay

209 DPPH scavenging activity was performed according to procedure previously described by

210 (Brand-Williams, Cuvelier, & Berset, 1995) with some modifications. The EC50 value was

211 defined as the concentration of the extract sufficient to reduce to 50% the maximum absorption

212 value estimated in the blank DPPH. To this end, 25 µL of different methanolic solutions of

213 extracts and 975 µL DPPH solution (60 µM) were mixed and incubated for 4 h. After, the

214 absorbance was measured at 516 nm in a microplate spectrophotometer reader. EC50 was

215 expressed as µg/mL of extract solution (mean of three replicates).

216

217 2.7.2 TEAC assay

218 The TEAC assay was performed following Re et al. procedure (Re et al., 1999). Trolox (0.25-

219 2.0 mM) was used for calibration of a standard curve. The ABTS•+ radical cation was produced

220 by reacting ABTS solution (7.00 mM) with K2S2O8 solution (2.45 mM) in dark for 16 h. The

221 ABTS•+ radical was diluted to an absorbance of 0.7 at 734 nm. Next, 10 µL of different

222 solutions of extracts were added to 990 µL of ABTS•+ solution. Absorbance of mixture was

223 recorded at 734 nm every 5 min for 45 min in a microplate spectrophotometer reader. The

224 extracts were analysed in triplicate and results expressed as TEAC values (mM trolox/g

225 extract).

226 2.8 Cell lines and cell culture

227 HT-29 (human colon adenocarcinoma) and CCD-18Co (normal human colon fibroblast) cells

228 were purchased from the American Type Culture Collection. Cell lines were cultured in RPMI

229 1640 medium, supplemented with Hepes 25 mM, L-glutamine 2.05 mM, 10% fetal bovine

230 serum and 25 μg/mL gentamicin, and incubated at 37 °C under 5% CO2 in a humidified

231 atmosphere. When the cell achieved 80%–90% confluent, it was detached by trypsin-EDTA

232 and sub-cultured into new sterile culture flasks for further propagation.

233

234 2.9 Antiproliferative activity assay

235 The antiproliferative activity of ‘sugar MSK’ extracts were evaluated by MTT assay. Cells in

236 exponential growth phase (70-80% confluence) were trypsinized, counted and seeded in 96-

237 well plates at a density of 1.0 ×104 (HT-29) and 4.5 ×103 (CCD-18Co) cells/well. The plates

238 were incubated for 24 hours at 37°C to allow the cell adhesion. Cells were treated with the

239 vehicle (DMSO 0.1% v/v) regarded as untreated controls or with different concentrations of

240 extracts (6.25 – 100 μg/mL) and incubated at three different time points 24, 48, and 72 h. After

241 the incubation, the medium was removed and 100 μL of MTT solution (0.25 mg/mL RPMI

242 1640 medium) was added to each well and the plate was incubated for 4 h. After, medium was

243 discarded and cells were washed with 200 μL of phosphate buffer saline (PBS). 100 μL of

244 DMSO were added to each well to dissolve the formazan crystals. The absorbance was

245 measured at 570 nm using a microplate reader (Tecan, Infinite® 200 PRO). Triton X-100

246 (1.0%) was used as a positive control. The cell viability was expressed as percentage of live

247 cells relative to controls. The IC50 values (concentration of extract that causes 50% inhibition

248 or cell death) were determined based on the dose-dependent response curves of extract using

249 GraphPad Prism 7.0 software (GraphPAD Corp., San Diego, CA, USA). Each experiment was

250 performed as three independent test with minimum three replicated.

251

252 2.10 Phytochemical profiling of ‘sugar MSK’ extracts and mangiferin quantification

253 2.10.1 Gas chromatography-mass spectrometry (GC-q-TOF-MS)

254 Fat composition of ‘sugar MSK’ extract obtained in the PLE- first step was studied using GC-

255 q-TOF-MS after derivatization according to Fiehn method, with some variations (Ibáñez, Simó,

256 Palazoglu, & Cifuentes, 2017). For this purpose, fat samples were subjected a two-step

257 derivatization process by methoxyamination and silylation reactions. 5 µL of fat solution (20

258 mg/mL n-heptane) were dried in SpeedVac Concentrator (SC200, Savant Instrument, Inc.,

259 Farmingdale, NY, USA). Subsequently, 10 µL of methoxyamine (CH3ONH2•HCl) solution (40

260 mg/mL pyridine) were added to dried sample and the mixture was shaken at 750 rpm for 60

261 min and 30 °C. Then 90 µL of MSTFA (N-methyl-N-(trimethylsilyl) trifluoroacetamide) with

262 1% TMCS (trimethylchlorosilane) and 2 µL of d27-myristic acid were added to mixture and

263 shaken again (750 rpm for 30 min and 37 °C). Derivatized samples were analysed employing

264 a 7890B Agilent system (Agilent Technologies, Santa Clara, CA, USA) coupled to a

265 quadrupole time-of-fight (q-TOF) 7200 (Agilent Technologies, Santa Clara, CA, USA)

266 equipped with an electronic ionization (EI) interface. An Agilent Zorbax DB5- MS + 10 m

267 Duragard Capillary Column (30 m × 250 μm x 0.25 μm) was used for chromatographic

268 separation. Sample injection volume was 1 µL. The injector operated in split mode (ratio of

269 10:1 and a split flow of 8.4 mL/min) at 250 °C. Helium was used as carrier gas at a constant

270 flow (0.8 mL/min). The oven temperature was programmed to start at 60 °C, heated to 325 °C

271 at 10 °C/min and held at this temperature for 10 min. MS parameters were the following:

272 electron impact ionization at 70 eV, filament source temperature of 250 °C, quadrupole

273 temperature of 150 °C, m/z scan range 50–600 amu at a rate of 5 spectra per second. Systematic

274 mass spectra deconvolution of chromatographic signals and tentative identification of

275 unknowns was performed using the Agilent Mass Hunter Unknown Analysis tool and mass

276 spectral databases (i.e. NIST MS Search v.2.0 and Fiehn Lib).

277

278 2.10.2 Liquid chromatography-tandem mass spectrometry (UHPLC-q-TOF-MS/MS)

279 The mangiferin content determination and phytochemical profiling of ‘sugar MSK’ extracts

280 obtained during the PLE- second step were studied using an Agilent 1290 UHPLC system

281 (Agilent Technologies, Santa Clara, CA, USA) coupled to an Agilent 6540 quadrupole-time-

282 of-flight mass spectrometer (q-TOF MS). A Zorbax Eclipse Plus C18 column (2.1 × 100 mm,

283 1.8 µm particle diameter, Agilent Technologies, Santa Clara, CA) was used for

284 chromatographic separation at 30 °C. The mobile phases were as follows: eluent A, H2O

285 (0.01% v/v formic acid), and eluent B, acetonitrile (0.01% v/v formic acid). The linear gradient

286 program was 0-30% B in 0-7 min, 30-80% B in 7-9 min, 80-100% B in 9-11 min, 100% B in

287 11-13 min and 0% B in 13-14 min at a flow rate of 0.5 ml/min and the sample injection volume

288 was 5 µL. The MS and MS/MS analyses were obtained in the negative ion mode using an

289 orthogonal ESI source (Agilent Jet Stream, AJS, Santa Clara, CA, USA). MS parameters were

290 the following: capillary voltage, 4000 V; nebulizer pressure, 40 psi; drying gas flow rate, 10

291 L/min; gas temperature, 350 ºC; skimmer voltage, 45 V; fragmentor voltage, 110 V. The MS

292 and Auto MS/MS modes were set to acquire m/z values ranging between 50-1100 and 50-800,

293 respectively, at a scan rate of 5 spectra per second. Agilent Mass Hunter Qualitative analysis

294 software (B.07.00) was used for post-acquisition data processing. The accurate mass data,

295 isotopic patterns, ion source fragmentation, MS/MS fragmentation patterns, MS databases (i.e.,

296 HMDB, Metlin, MassBank) and bibliographic search were employed for tentative

297 identification of ‘sugar MSK’ phytochemicals. Quantitative data for mangiferin were obtained

298 by calibration curve constructed with the standard compound in the range of 0.1-100 µg/mL.

299

300 3. RESULTS AND DISCUSSION

301 3.1 Theoretical selection of green solvents for phenolics recovery by HSP approach:

302 mangiferin as target compound

303 Using Y-MB method (HSPiP®), HPSs of mangiferin were obtained at room conditions (25 °C-

304 1.01 bar) as can be seen in Table 1. The dispersive interaction parameter, D, for mangiferin

305 (20.3 MPa1/2) showed a higher influence on solubility parameters due to geometry (very

306 flattened boat conformation) and aromatic character of three-ring system (benzenoid-pyraoind-

307 benzenoid) (Gales & Damas, 2005). On the other hand, polar (P), and hydrogen-bonding (H)

308 parameters were affected by high hydroxylation degree of the xanthone and the -D-

309 glucopyranosyl moiety. These HSP data were used to predict the mangiferin solubility in green

310 solvents calculating the “distance” or Ra value (Table 1). Ra scores showed greater miscibility

311 of mangiferin with ethyl lactate (9.1 MPa1/2) in comparison with ethanol (11.5 MPa1/2), ethyl

312 acetate (11.8 MPa1/2) and (+)-limonene (13.6 MPa1/2). However, the physicochemical

313 properties of ethyl lactate (boiling point 154 °C at 1.01 bar) make difficult its evaporation to

314 obtaining dry extracts, thus limiting its application. For this reason, the HSPs of ethanol-ethyl

315 acetate mixtures were calculated in order to obtain a similar ethyl lactate affinity (Table 1). The

316 ethanol:ethyl acetate 50:50 v/v mixture showed a close distance to ethyl lactate respect to

1/2 317 mangiferin (Ra = 9.7 MPa ). This can be explained by the decrease of polar and hydrogen-

318 bonding parameters of ethanol due to ethyl acetate addition. Therefore, ethanol, ethyl acetate

319 and their mixture (50:50 v/v) were preferred as extraction solvents for the PLE-second step

320 optimization process. The temperature effect on Ra is also presented in Table 1. As can be seen,

321 Ra value increases with the temperature showing a lower affinity for mangiferin, however this

322 can be different in practice, because HSP approach is based on thermodynamic data and the

323 kinetic phenomena, such as mass transfer and solubility increase due to temperature, are not

324 considered. These effects have been previously demonstrated for the extraction of natural

325 compounds from algae (Sánchez-Camargo, Montero, Cifuentes, Herrero, & Ibáñez, 2016).

326

327 3.2 Selection of solvent for the first step of the PLE procedure: defatting step

328 Kinetic experiments were performed to select the most suitable non-polar solvent for fat

329 extraction from ‘sugar MSK’. Four kinetic curves using n-heptane, cyclohexane, (+)-limonene

330 and n-hexane (reference solvent in PLE process) were obtained considering times from 10 to

331 90 min with sample collection every 10 min. In addition, B & D was employed for total lipid

332 recovery as standard method. Figure 2 shows the comparison of the B & D results and the

333 kinetic curves. As can been seen, B & D method and (+)-limonene presented close extraction

334 efficiency (16.01% and 15.32%, respectively). However, (+)-limonene exhibited a slow

335 extraction rate with a maximum accumulated recovery at 90 min, requiring large solvent- and

336 time-consuming, consequently (+)-limonene was discarded as solvent for this PLE-step. The

337 profiles of the n-heptane, n-hexane and cyclohexane kinetic curves showed that the extraction

338 yields increased rapidly in the first 10 min, reaching the equilibrium after 20 min of extraction.

339 The rapid increase of extraction yields in the initial stage is usually attributed to washing of

340 components located on the external surface of the matrix particles (Okiyama et al., 2018), in

341 this case the cellular lipid matrix from MSK. After this stage, the extraction rate decreases and

342 the lipophilic compounds are principally recovery from plastids in broken cells by a diffusion

343 process (Xi, Yan, & He, 2014). The n-heptane provided the extraction yield (13.82%) nearest

344 to B & D method followed by n-hexane (13.19%) and cyclohexane (8.83%). The performance

345 differences among n-heptane and B & D method can be explained by the high selectivity of

346 chloroform/methanol/water system for lipids recovery due to the partial miscibility of the

347 chloroform in the aqueous and organic phases, which ensures that neutral and polar lipids

348 independent of their molecular volume, are solubilized in the coexisting phases (Breil et al.,

349 2017). In this sense, n-heptane can be limited to medium polar and non-polar lipids recovery.

350 In addition, n-heptane is considered as usable and alternative to more toxic non-polar solvents

351 such as n-hexane and petroleum ether (Calvo-Flores, Monteagudo-Arrebola, Dobado, & Isac-

352 Garcia, 2018). For these reasons, n-heptane was selected as solvent for the PLE-first step and

353 20 min were considered as appropriated extraction time.

354

355 3.3 Optimization of the second step of the PLE procedure

356 The effects of solvent composition (percentage of EtOH in the mixture EtOH/EtOAc: 0, 50 and

357 100 % v/v) and temperature (50, 100 and 150 ºC) were investigated on mangiferin content,

358 extraction yield, TPC, TFC and antioxidant activity (EC50 and TEAC) to optimize the PLE-

359 second step. CCD was used to study the best possible combination of independent extraction

360 parameters for each response. Table 2 shows the codified and real levels of independent

361 variables and the resulting responses. The experimental data were fitted to linear, interaction

362 and quadratic regression models. Polynomial equation coefficients were calculated through

363 response surface methodology (RSM) and are provided in Table S1 (Supplementary material).

364 The analysis of variance (ANOVA) was performed to confirm the adequacy, significance level

365 and predictive value of regression models (Table S1). The values from the lack-of-fit test were

366 not significant (p > 0.05), consequently the models fit to experimental data. Most of terms

367 showed high F-values and low p-values (<0.0001) indicating that the regression models were

368 significant. The coefficients of determination (R2) were close to 1 (0.756-0.933), indicating a

369 high degree of correlation between the experimental and predicted values of the models

370 (Briones-Labarca, Giovagnoli-Vicuna, & Canas-Sarazua, 2019). Pareto charts and surface

371 responses were plotted to display the influence of the extraction parameters on response

372 variables (See Figure 3). Mangiferin content was principally influenced by negative effect of

373 solvent composition followed by positive effect of temperature (Figure 3A). In this sense, the

374 highest mangiferin content (13.27 ± 1.67 mg mangiferin/g Db) was observed using 100% EtOH

375 at 150 °C, however this amount is equivalent to that obtained by DM using 50% EtOH (13.60

376 ± 2.22 mg mangiferin/g Db) and 100% EtOH (12.34 ± 1.67 mg mangiferin/g Db) at 25 °C,

377 since no statistical significant differences between these values were observed (p > 0.05). These

378 results are consistent with HSP data prediction, given that mangiferin showed affinity by

379 EtOH:EtOAc mixture (50:50 v/v) and EtOH solvents. The mangiferin content from M. indica

380 organs, including seed kernel, has been an object of study in several investigations, due to the

381 high bioactivity of this compound. Mangiferin content in MSK varies depending on the

382 cultivars’ nature and its geographical origin, with values ranging between 0.22 and 8.98 mg

383 mangiferin/g Db (Barreto et al., 2008; Lopez-Cobo et al., 2017; Luo et al., 2012; Ruales et al.,

384 2018). Solid-liquid extraction at room conditions or ultrasound-assisted extraction using

385 methanol and methanol-water mixtures have been the most employed extraction techniques for

386 mangiferin recovery from MSK.

387 On the other hand, extraction yield was mainly affected by both solvent composition and

388 temperature, with a minor contribution of the interaction and the quadratic effects (Figure 3B).

389 Thus, the highest extraction yield (12.15 ± 0.90%) was obtained employing 150 ºC and 100 %

390 EtOH. The TPC and TFC responses showed a very similar behaviour according to Pareto charts

391 (Figures 3C and 3D, respectively). In both cases, TPC and TFC were influenced by linear effect

392 of temperature and by negative and quadratic effect of solvent composition. As for TPC and

393 TFC, the best results (143.79 ± 2.09 mg GAE/g Db and 1.21 ± 0.16 mg QE/g Db and,

394 respectively) were obtained using 100% EtOH and 150 °C. Comparatively, TPC values were

395 within the ranges reported in previous studies (2.19-740 mg GAE/g Db), in contrast TFC results

396 were higher than those previously reported (0.72-1.31 g (+)-catechin 100/g Db) for MSK

397 (Torres-León et al., 2016). TFC obtained by DM at 25 °C and 50 % EtOH was higher (3.72 ±

398 0.08 mg QE/g Db) than the obtained by PLE extractions. This difference can be explained by

399 the low temperatures employed in DM that ensure the thermolabile flavonoids recovery, as

400 well as by the longer extraction time (24 h), which allows higher contact between the sample

401 and the solvent.

402 EC50 response was influenced mainly by negative and quadratic effect of solvent composition

403 and by the negative and linear effect of temperature (Figure 3E), obtaining the highest free

404 radical scavenging capacity (14.34 ± 0.19 µg/mL) employing 100% EtOH and 100 °C. As for

405 TEAC the principal effect was the temperature followed by solvent composition (Figure 3F)

406 and the best value (2.31 mM trolox/g) was observed using 100% EtOH at 150 ºC.

407 Comparatively, the ‘sugar MSK’ extracts had a moderate antioxidant capacity respect to the

408 one reported previously for other MSK cultivars: 0.56-14.00 µg/mL and 0.44-1.03 mmol

409 trolox/g, for EC50 and TEAC, respectively (Luo et al., 2014; Maisuthisakul & Gordon, 2009).

410 Due to the notably high mangiferin and phenolic contents as well as the moderate antioxidant

411 power of ‘sugar MSK’, PLE extraction was optimized for all responses studied. To attain this,

412 desirability function combining mangiferin content, extraction yield, TPC, TFC, EC50 and

413 TEAC responses was calculated. Profiles for predicted values estimated by desirability

414 function are shown in Figure 4. Optimal conditions were 100% EtOH v/v and 150 °C at 0.92

415 desirability value, corresponding to experimental run number 10 of the CCD. The desirability

416 value was very close to 1, indicating a high maximisation degree for multi-response

417 optimization. Predicted response values obtained by global desirability function under

418 optimum conditions were checked with those of experiment 10. The observed and predicted

419 data were within the confidence intervals (Figure 4). Despite the optimum conditions were at

420 the experimental region limit, the proximity between predictive and experimental data

421 confirmed that selected RSM model was successfully applied for PLE of ‘sugar MSK’ to obtain

422 extracts with maximum mangiferin content, phenolic content and antioxidant activity.

423

424 3.4 UHPLC-q-TOF-MS/MS profiling analysis of ‘sugar MSK’ extracts

425 As a result of the phytochemical profiling of MSK polar fractions obtained by PLE, a total of

426 71 compounds were tentatively identified on the basis of their accurate mass, MS/MS

427 fragmentation patterns, MS databases (i.e., HMDB, Metlin, MassBank) and previously

428 reported data in literature. Table 3 summarizes the phytoconstituents identified by ESI-q-TOF-

429 MS/MS analysis in negative ionization mode, including the retention time (min), molecular

430 formula, experimental deprotonated molecular ions ([M-H]-), calculated mass error (ppm), and

431 MS/MS product ions.

432 Although the composition of the edible part of mango and some by-products such as peel, seed

433 husk and seed kernel has been described in literature for several cultivars, information about

434 the composition of mango seed kernel for ‘sugar mango’ cultivar is limited. In this regard,

435 typical phenolic acid and flavonoids, along with characteristic xanthones and benzophenone

436 derivatives were identified in sugar MSK, as reported in literature for other mango cultivars.

437 The intense peak observed in the TIC at 2.0 min (see Figure 5A), confidently identified as

438 gallic acid (m/z 169.0142 [M−H]−), was one of the main compounds in the phytochemical

439 profile of the polar PLE extracts. The widespread presence of this phenolic acid in the seed

440 kernel of sugar mango is evidenced by the broad variety of gallic acid derivatives identified in

441 sugar MSK extracts (see Table 3). The presence of other relevant phenolic acids such as quinic

442 acid (compound 1, m/z 191.0561), protocatechuic acid (compound 11, m/z 153.0193), p-

443 hydroxybenzoic acid (compound 19, m/z 137.0244), ferulic acid (compound 47, m/z 193.0506)

444 and ellagic acid (compound 52, m/z 300.9990) could also be confirmed by commercial

445 standards.

446 Gallates and gallotannins were the main family of compounds identified in the polar MSK

447 extracts (see Table 3 and Figure 5E). The presence of gallic acid derivatives can be identified

448 by MS/MS product ions at m/z 169.0142 and m/z 125.0239 or 124.0160. According to this

449 fragmentation pattern, the methyl and ethyl esters of gallic acid were identified at m/z 183.0299

450 (compound 24) and m/z 197.0455 (compound 43), respectively, being ethylgallate the most

451 abundant compound in the analysed extract. Digallic acid (compound 29), galloyl

452 methylgallate (compound 59), galloyl ethylgallate (compound 67), and ethyl trigallate

453 (compound 70) could be identify due to the loss of the galloyl moiety (C7H4O4, 152.0110 Da).

454 Two minor galloyl derivatives (compound 7 and 9, m/z 343.0671) were assigned as

455 galloylquinic acid isomers, showing fragment ions at m/z 169 and 127, as representative

456 diagnostic ions of both galloyl and quinic acids, respectively.

457 Gallotannins represent the main family of compounds identified in polar MSK extracts. Since

458 the composition of these hydrolysable tannins is based on a core structure of glucose esterified

459 with gallic acid residues, the MS/MS fragmentation pattern of these polyphenolic biopolimers

460 is mainly characterized by the successive loss of galloyl (Gall, 152.0110 Da) and glucose (Glu,

461 162.0529 Da) subunits. Thus, a set of gallotannin isomers containing up to 6 galloyl subunits

462 were identified. Galloyl glucose (compound 3), the simplest isomer, shows [M-H]- ion at m/z

463 331.0671, exhibiting m/z 169.0142 [M−H−Glu]− as the major product ion. The identity of

464 deprotonated molecular ions m/z 483.0780 (compounds 13, 17, 26, 27), m/z 635.0890

465 (compounds 23, 34, 35, 38, 39, 42, 44), m/z 787.1000 (compounds 45, 51, 56, 60, 63), m/z

466 939.1109 (compound 64) and m/z 1091.1219 (compound 71) could be confidently assigned to

467 digalloyl-, trigalloyl-, tetragalloy-, pentagalloyl- and hexagalloyl glucose isomers,

468 respectively, with Δm/z below 5 ppm, as determined by HRMS.

469 Gallotannins identified at m/z 493.1199 (Compounds 2, 4, 6 and 10) and m/z 645.1309

470 (Compounds 8, 14, 15, 18, 20, 22, 28, 32) share product ions at m/z 331 [493−Glu]− and m/z

471 169 [331−Glu]−, suggesting the presence of two glucose subunits. These compounds were

472 tentatively assigned as galloyl diglucoside and digalloyl diglucose isomers, respectively. Other

473 group of gallotanins with [M-H] − ion at m/z 493.0991 (compounds 49, 62, 55, 65, 66, 68) were

474 tentatively identified as feruloyl galloyl glucose isomers. This assignation is supported by

− 475 characteristic product ions at m/z 295.0450 [M–H–Gall–CH2O2] and m/z 169.0142 [M–H–

476 Glu–Feruloyl]−.

477 Unlike in mango peel, where a wide variety of quercetin and rhamnetin derivatives have been

478 reported (Gomez-Caravaca, Lopez-Cobo, Verardo, Segura-Carretero, & Fernandez-Gutierrez,

479 2016), the content of flavonoids in sugar MSK is mainly based on catechin and epicatechin

480 (compounds 31 and 37, m/z 289.0718 [M−H]−), as well as on (epi)catechin gallate (compound

481 57, m/z 289.0718). In addition, quercetin and quercetin glucoside (compounds 69, m/z

482 289.0718) and 54, m/z 289.0718) were also present, although in minor extent. The identity of

483 common flavonoids such as (epi)catechin and quercetin was confirmed by commercial

484 standard. The glycosylated and galloylated derivatives were identified based on the 162.0528

485 (C6H10O5) and 152.0110 (C7H4O4) Da neutral loss in the MS/MS spectrum, respectively.

486 Seven xanthone-like structures were detected in MSK extracts. Mangiferin (m/z 421.0776,

487 compound 36) was the most abundant xanthone, being one of the most relevant phytochemicals

488 reported in mango. The presence of this xanthone C-glycoside in MSK was confirmed by

489 commercial standards and by typical MS/MS fragment ions at m/z 301.0361 and m/z 331.0468.

490 The same diagnostic ions were observed for compounds 50 and 61 (m/z 421.0776 [M−H]−),

491 tentatively assigned as mangiferin isomers. Similar fragmentation pattern was shown for

492 compounds 46 (m/z 573.0886 [M−H]−), readily identified as mangiferin gallate with a major

493 product ion at m/z 403 [M−H−170]−, indicating the neutral loss of gallic acid. An extra methyl

494 group was observed in MS/MS spectra of compound 41, tentatively identified as O-methyl

495 mangiferin, also called as homomangiferin.

496 The presence of benzophenones, major intermediates in the biosynthetic pathway of xanthones,

497 could also be confirmed in MSK polar extracts. Compounds 16, 30 and 40, were tentatively

498 identified as maclurin-C-glucoside derivatives. These phytochemicals contain a 2,3',4,4',6-

499 pentahydroxybenzophenone as core structure, sharing the typical fragmentation pattern

500 characterized by the loss of 120 and 90 Da neutral fragments corresponding to C-glycosilated

501 derivatives. Similar neutral loses were observed for compound 24, tentatively identified as

502 iriflophenone glucoside, a C-glycosylated tetrahydroxybenzophenone. Typical fragment ions

503 at m/z 333, 303, and 193 were consistent with data reported in literature (Dorta et al., 2014).

504 Hexahydroxybenzophenone isomers (peaks 21, 48 and 53) were also detected at m/z 421.0776

− − 505 [M−H] , exhibiting m/z 125.0239 [C6H5O3] as the main product ion after the loss of a the

506 galloyl moiety (152 Da).

507

508 3.5 GC-q-TOF-MS profiling analysis of ‘sugar MSK’ extracts

509 The non-polar PLE extracts obtained from MSK were analysed by GC-q-TOF-MS to

510 characterize the main lipidic components. A derivatization procedure described section 2.10.1

511 was applied in order to improve detectability of fatty acids (FAs) and other lipids (e.g.

512 phytosterols), leading to the identification of the corresponding trimethylsilyl derivatives.

513 Table 4 summarizes the tentatively identified metabolites, including their corresponding

514 characteristic GC-HRMS parameters (e.g. retention time, match factor values given by NIST

515 database, monoisotopic mass, calculated mass error (∆m/z) and main HR-MS/MS fragments),

516 that confirm their unambiguous identification. As shown in Table 4, five major fatty acids

517 including palmitic acid, linoleic acid, oleic acid, stearic acid, and eicosapentaenoic acid were

518 positively identified as trimethylsilyl derivatives, showing the characteristic [M-H+TMS-

519 CH3]+ ions GC-(EI+)-MS spectra. Unlike fatty acids, the terpenoid β-sitosterol could be

520 clearly detected by the [M-H+TMS]+ ion of the terpenoid-TMS derivatives.

521 Figure 6 illustrates the lipidic profile of MSK extracts, showing palmitic acid, oleic acid and

522 stearic as the most abundant compounds. The presence of stearic acid and oleic acid as major

523 fatty acids in non-polar MSK extracts is consistent with data reported in literature (Lieb et al.,

524 2018). The predominance of these fatty acids provides more stability compare to other oils rich

525 in polyunsaturated fatty acids. Several studies report mango seed kernel fat with typical

526 characteristics of a vegetable butter, and these oils are suitable for mixing together with

527 vegetable oils for use in the confectionery industry. The absence of “trans” fatty acids is another

528 advantage of the lipids from mango seed, as they are responsible for the development of various

529 diseases and adverse effects on human health (Torres-León et al., 2016).

530

531 3.6 Antiproliferative activity

532 HT-29 cell line is considered as paradigm of the colon carcinogenesis and is one the most

533 refractory colon cancer line against the antiproliferative activity of natural compounds (Castro-

534 Puyana et al., 2017; Fearon & Vogelstein, 1990; Valdes et al., 2013). For these reasons and

535 considering the mangiferin content, phenolic content and antioxidant capacity of the PLE-

536 extract obtained from ‘sugar MSK’ under optimum conditions (100% EtOH-150 °C), its

537 antiproliferative activity was also tested on HT-29. Thus, HT-29 cells were incubated with

538 different concentrations of the optimum-PLE extract (from 6.25 to 100 μg/mL) during 24, 48

539 and 72 h, and cell proliferation was measured by the MTT assay. Comparatively,

540 antiproliferative activity of the DM extract (100% EtOH at 25 °C) was studied under the same

541 conditions. In addition, the PLE-extract was also tested on CCD-18Co cell line to determine

542 its potential toxicity in non-cancer colon cells. Antiproliferative activity was expressed as IC50

543 value and the results are shown in Figure 7. As can be seen, the viability of HT-29 cells was

544 reduced in response to treatment with PLE-extract after 48 and 72 h of exposition (IC50 = 56.37

545 ± 3.45 and 28.67 ± 5.35 μg/mL, respectively). PLE-extract was most active at 72 h of treatment,

546 decreasing the cell viability in a dose-dependent manner. In contrast, DM extract did not induce

547 any cytotoxic effect at the concentrations and times tested. On the other hand, PLE-extract did

548 not exert inhibition on CCD-18Co cell proliferation at 48 h, but it showed an antiproliferative

549 effect at 72 h (IC50 = 85.19 ± 5.26 μg/mL). According to these results, the selectivity index (SI)

550 of PLE-extract on HT-29 respect to CCD-18Co cells was calculated, revealing a value of 2.97

551 (Figure 7). According to (Badisa et al., 2009) a compound with SI value > 2 exhibits selective

552 toxicity toward cancer cells but gives minimal toxicity or no harm to normal cells, while a

553 compound with SI value < 2 is considered toxic even to normal cells. This approach has also

554 been applied to establish the selectivity degree of extracts from vegetable sources towards

555 cancer cells (Asif et al., 2017; Asif et al., 2016); following this criteria, it is possible to state

556 that PLE-extract has selectivity toward HT-29 cells.

557 In a recent study, the antiproliferative potential of a methanolic extract from ‘sugar MSK’

558 obtained using Soxhlet was evaluated against a panel of human cancer cell lines that included

559 MDA-MB-231 (breast adenocarcinoma), PC-3 (prostate adenocarcinoma), A-549 (lung

560 adenocarcinoma) and HT-29 (Castro-Vargas et al., 2019). Results showed a decrease of HT-

561 29 cells viability (~75%) at 125 µg/mL of methanolic extract and the authors related the

562 phenolic composition with its antiproliferative activity. This antiproliferative potential was

563 comparatively lower than the present study and this difference can be explained by the

564 extraction technique employed in each case, since unlike the Soxhlet method, the PLE can

565 more efficiently concentrate the polyphenolic compounds responsible for the antiproliferative

566 properties of the extract.

567 The antioxidant, antiproliferative and chemopreventive properties of polyphenolics and other

568 compounds identified in ‘sugar MSK’ PLE-extract have been described previously in the

569 scientific literature. Mangiferin is one of the most important compounds from Anacardeciae

570 and its bioactivity has been reported in several review works (Adam, Piotr, Edyta, & Dorota,

571 2013; Imran et al., 2017; Khurana, Kaur, Lohan, Singh, & Singh, 2016; Rajendran, Rengarajan,

572 Nandakumar, Divya, & Nishigaki, 2015). According to (Gold-Smith, Fernandez, & Bishop,

573 2016) mangiferin is involved in different molecular mechanisms, including cell protection

574 against oxidative stress and DNA damage, as well as down-regulation of inflammation, cell

575 cycle arrest, apoptosis promotion and proliferation reduction of malignant cells. Overall,

576 previous studies showed that the anticancer effect of mangiferin is more pronounced when used

577 as a chemopreventive agent against induced colon carcinogenesis (Khurana et al., 2016).

578 However, the well-known antioxidant power of mangiferin and its capacity to induce apoptosis

579 through inhibition of NF-κB activation in different cancer cell lines (Gold-Smith et al., 2016),

580 allows thinking that this compound contributes to the antiproliferative activity of PLE-extract

581 observed on HT-29 cells.

582 Gallic acid was also identified in the PLE-extract and its antiproliferative activity has been

583 previously reported both in vitro and in vivo models (Verma, Singh, & Mishra, 2013). The

584 antiproliferative effects of gallic acid are mediated via gene modulation of cell cycle,

585 metastasis, angiogenesis and apoptosis, as well as by the inhibition of NF-κB and Akt

586 activation. In the cancer colon cell lines HT-29 (Verma et al., 2013), LS180 (Velderrain-

587 Rodriguez et al., 2018) and Caco-2 (Salucci, Stivala, Bugianesi, & Vannini, 2002) the

588 antiproliferative activity of gallic acid has been related with apoptosis and antioxidant

589 mechanisms.

590 Gallic acid structurally related compounds such as gallates and galloyl glycosides are also

591 present in the PLE-extract. Ethyl gallate is the most abundant compound in the extract and its

592 anticancer activity against different cancer cell lines through induction of apoptosis has been

593 established (Kim et al., 2012; Mohan, Thiagarajan, & Chandrasekaran, 2014). Likewise,

594 galloyl glycosides may regulate the production of reactive oxygen species by altering the redox

595 balance in the cell, which activates the intrinsic mitochondrial apoptosis pathway (Banerjee,

596 Kim, Krenek, Talcott, & Mertens-Talcott, 2015). Ellagic acid was another bioactive compound

597 found in the optimum PLE-extract. The antiproliferative activity of ellagic acid has been

598 studied in the cancer colon cell lines HTC116 and Caco-2, whose cell viability was altered by

599 cell cycle modulation, Bax translocation, caspase-8 activation and PCNA expression reduction

600 (Yousef, El-Masry, & Yassin, 2016).

601 The antiproliferative activity observed for PLE-extract could not be attributed to a single

602 component but to the possible synergistic effect of some of the compounds present in the

603 extract. In this respect, previous studies have shown the synergic behavior of various

604 polyphenolics for antioxidant and antiproliferative activities. García-Rivera et al. (García-

605 Rivera, Delgado, Bougarne, Haegeman, & Berghe, 2011) observed significant antiproliferative

606 effects with the Vimang® (a standardized extract derived from mango bark) compounds

607 mangiferin and gallic acid, against the MDA-MB-231, HT-1080 and Caco-2 cancer cells and

608 upon measuring cytotoxic activities, the authors found that Vimang® and gallic acid, but not

609 mangiferin, were able to kill MDA-MB-231 cells, suggesting that minor amounts of gallic acid

610 in Vimang® are sufficient to trigger significant antiproliferative effects. Likewise,

611 hydroxybenzoic acids and hydroxycinnamic acids have a potential inhibitory effect on cancer

612 cells proliferation by synergistic interactions arresting cell cycle and inducing apoptosis

613 (Rocha, Monteiro, & Teodoro, 2012). In this way, the potentiated effects by polyphenolic

614 compounds’ combination in the reduction of cancer cell viability and apoptosis induction has

615 been confirmed by isobolographic analysis of cell proliferation data for ellagic acid and

616 quercetin (Mertens-Talcott, Talcott, & Percival, 2003).

617

618 4. CONCLUSIONS

619 In this work, an integrated valorization strategy was proposed through an optimized PLE

620 procedure in two sequential steps to obtain bioactive-rich fractions with high antioxidant

621 capacity and demonstrated antiproliferative activity. The lipidic fraction was firstly recovered

622 with n-heptane, whereas a mixture of ethanol/ethyl acetate was selected as a suitable green

623 extraction solvent for the subsequent recovery of mangiferin and other phenolic compounds,

624 on the basis of preliminary studies applying the HSP approach. Phenolic extracts obtained

625 under optimal PLE conditions after RSM optimization showed satisfactory extraction yield and

626 good antioxidant activity with notably high mangiferin and phenolics concentration levels. The

627 profiling analysis of the lipidic and phenolic MSK fractions by GC and UHPLC coupled to q-

628 TOF-MS/MS revealed the presence of abundant oleic and stearic acids, as well as typical

629 phenolic acids, flavonoids, characteristic xanthones, as well as a broad family of gallate

630 derivatives and gallotannins with demonstrated in vitro bioactivity, as evidenced by the

631 selective antiproliferative activity exhibited against human colon adenocarcinoma cell line HT-

632 29. The proposed valorization strategy represents a powerful multiplatform of integrated

633 analytical technologies to improve the sustainability of mango processing industry.

634

635 Acknowledgements

636 This research was supported by COOPA20145, project from CSIC (Programa de Cooperación

637 Científica para el Desarrollo “i-COOP+”). G.A.-R. would like to acknowledge Ministerio de

638 Ciencia Investigacion y Universidades (MICINN) for a “Juan de la Cierva” postdoctoral grant.

639 The authors also thank the support from the AGL2017-89417-R project (MICINN).

640 References

641 Abdullah, A. S., Mohammed, A. S., Rasedee, A., & Mirghani, M. E. (2015). Oxidative stress-

642 mediated apoptosis induced by ethanolic mango seed extract in cultured estrogen

643 receptor positive breast cancer MCF-7 cells. International Journal of Molecular

644 Sciences 16(2), 3528-3536.

645 Abdullah, A. S., Mohammed, A. S., Rasedee, A., Mirghani, M. E., & Al-Qubaisi, M. S. (2015).

646 Induction of apoptosis and oxidative stress in estrogen receptor-negative breast cancer,

647 MDA-MB231 cells, by ethanolic mango seed extract. BMC Complementary and

648 Alternative Medicine, 15, 45.

649 Adam, M., Piotr, K., Edyta, G., & Dorota, W. (2013). Mangiferin – a Bioactive Xanthonoid,

650 not only from Mango and not just Antioxidant. Mini-Reviews in Medicinal Chemistry,

651 13(3), 439-455.

652 Ameer, K., Shahbaz, H. M., & Kwon, J.-H. (2017). Green Extraction Methods for Polyphenols

653 from Plant Matrices and Their Byproducts: A Review. Comprehensive Reviews in Food

654 Science and Food Safety, 16(2), 295-315.

655 Asif, M., Shafaei, A., Abdul Majid, A. S., Ezzat, M. O., Dahham, S. S., Ahamed, M. B. K.,

656 Oon, C. E., & Abdul Majid, A. M. S. (2017). Mesua ferrea stem bark extract induces

657 apoptosis and inhibits metastasis in human colorectal carcinoma HCT 116 cells,

658 through modulation of multiple cell signalling pathways. Chinese Journal of Natural

659 Medicines, 15(7), 505-514.

660 Asif, M., Yehya, A. H. S., Al-Mansoub, M. A., Revadigar, V., Ezzat, M. O., Khadeer Ahamed,

661 M. B., Oon, C. E., Murugaiyah, V., Abdul Majid, A. S., & Abdul Majid, A. M. S.

662 (2016). Anticancer attributes of Illicium verum essential oils against colon cancer.

663 South African Journal of Botany, 103, 156-161.

664 Badisa, R. B., Darling-Reed, S. F., Joseph, P., Cooperwood, J. S., Latinwo, L. M., & Goodman,

665 C. B. (2009). Selective Cytotoxic Activities of Two Novel Synthetic Drugs on Human

666 Breast Carcinoma MCF-7 Cells. Anticancer Research, 29(8), 2993-2996.

667 Ballesteros-Vivas, D., Álvarez-Rivera, G., del Pilar Sánchez-Camargo, A., Ibáñez, E., Parada-

668 Alfonso, F., & Cifuentes, A. (2019). A multi-analytical platform based on pressurized-

669 liquid extraction, in vitro assays and liquid chromatography/gas chromatography

670 coupled to high resolution mass spectrometry for food by-products valorisation. Part 1:

671 Withanolides-rich fractions from goldenberry (Physalis peruviana L.) calyces obtained

672 after extraction optimization as case study. Journal of Chromatography A, 1584, 155-

673 164.

674 Banerjee, N., Kim, H., Krenek, K., Talcott, S. T., & Mertens-Talcott, S. U. (2015). Mango

675 polyphenolics suppressed tumor growth in breast cancer xenografts in mice: role of the

676 PI3K/AKT pathway and associated microRNAs. Nutrition Research, 35(8), 744-751.

677 Barreto, J. C., Trevisan, M. T. S., Hull, W. E., Erben, G., de Brito, E. S., Pfundstein, B.,

678 Würtele, G., Spiegelhalder, B., & Owen, R. W. (2008). Characterization and

679 quantitation of polyphenolic compounds in bark, kernel, leaves, and peel of mango

680 (Mangifera indica L.). Journal of Agricultural and Food Chemistry, 56(14), 5599-5610.

681 Bligh, E. G., & Dyer, W. J. (1959). A rapid method of total lipid extraction and purification.

682 Canadian Journal of Biochemistry and Physiology, 37(8), 911-917.

683 Brand-Williams, W., Cuvelier, M. E., & Berset, C. (1995). Use of a free radical method to

684 evaluate antioxidant activity. LWT - Food Science and Technology, 28(1), 25-30.

685 Breil, C., Abert Vian, M., Zemb, T., Kunz, W., & Chemat, F. (2017). "Bligh and Dyer" and

686 Folch Methods for Solid-Liquid-Liquid Extraction of Lipids from Microorganisms.

687 Comprehension of Solvatation Mechanisms and towards Substitution with Alternative

688 Solvents. International Journal of Molecular Sciences, 18(4).

689 Briones-Labarca, V., Giovagnoli-Vicuna, C., & Canas-Sarazua, R. (2019). Optimization of

690 extraction yield, flavonoids and lycopene from tomato pulp by high hydrostatic

691 pressure-assisted extraction. Food Chemistry, 278, 751-759.

692 Calvo-Flores, F. G., Monteagudo-Arrebola, M. J., Dobado, J. A., & Isac-Garcia, J. (2018).

693 Green and Bio-Based Solvents. Topics in Current Chemistry, 376(3), 18.

694 Castro-Puyana, M., Pérez-Sánchez, A., Valdés, A., Ibrahim, O. H. M., Suarez-Álvarez, S.,

695 Ferragut, J. A., Micol, V., Cifuentes, A., Ibáñez, E., & García-Cañas, V. (2017).

696 Pressurized liquid extraction of Neochloris oleoabundans for the recovery of bioactive

697 carotenoids with anti-proliferative activity against human colon cancer cells. Food

698 Research International, 99, 1048-1055.

699 Castro-Vargas, H. I., Ballesteros Vivas, D., Ortega Barbosa, J., Morantes Medina, S. J.,

700 Aristizabal Gutierrez, F., & Parada-Alfonso, F. (2019). Bioactive Phenolic Compounds

701 from the Agroindustrial Waste of Colombian Mango Cultivars 'Sugar Mango' and

702 'Tommy Atkins'-An Alternative for Their Use and Valorization. Antioxidants (Basel),

703 8(2).

704 Chemat, F., Vian, M. A., & Cravotto, G. (2012). Green extraction of natural products: concept

705 and principles. International journal of molecular sciences, 13(7), 8615-8627.

706 Corrales-Bernal, A., Maldonado, M. E., Urango, L. A., Franco, M. C., & Rojano, B. A. (2014).

707 Sugar mango (Mangifera indica), variety from Colombia: antioxidant, nutritional and

708 sensorial characteristics. Revista Chilena de Nutrición, 41(3), 312-318.

709 Damergi, E., Schwitzguébel, J. P., Refardt, D., Sharma, S., Holliger, C., & Ludwig, C. (2017).

710 Extraction of carotenoids from Chlorella vulgaris using green solvents and syngas

711 production from residual biomass. Algal Research, 25(May), 488-495.

712 Dorta, E., González, M., Lobo, M. G., Sánchez-Moreno, C., & de Ancos, B. (2014). Screening

713 of phenolic compounds in by-product extracts from mangoes (Mangifera indica L.) by

714 HPLC-ESI-QTOF-MS and multivariate analysis for use as a food ingredient. Food

715 Research International, 57, 51-60.

716 Ediriweera, M. K., Tennekoon, K. H., & Samarakoon, S. R. (2017). A Review on

717 Ethnopharmacological Applications, Pharmacological Activities, and Bioactive

718 Compounds of Mangifera indica (Mango). Evidence-Based Complementary and

719 Alternative Medicine, 2017, 6949835.

720 Fearon, E. R., & Vogelstein, B. (1990). A genetic model for colorectal tumorigenesis. Cell, 61,

721 759–767.

722 Gales, L., & Damas, A. M. (2005). Xanthones–A Structural Perspective. Current Medicinal

723 Chemistry, 12(21), 2499-2515.

724 García-Rivera, D., Delgado, R., Bougarne, N., Haegeman, G., & Berghe, W. V. (2011). Gallic

725 acid indanone and mangiferin xanthone are strong determinants of immunosuppressive

726 anti-tumour effects of Mangifera indica L. bark in MDA-MB231 breast cancer cells.

727 Cancer letters, 305(1), 21-31.

728 Gentile, C., Di Gregorio, E., Di Stefano, V., Mannino, G., Perrone, A., Avellone, G., Sortino,

729 G., Inglese, P., & Farina, V. (2019). Food quality and nutraceutical value of nine

730 cultivars of mango (Mangifera indica L.) fruits grown in Mediterranean subtropical

731 environment. Food Chemistry, 277, 471-479.

732 Gold-Smith, F., Fernandez, A., & Bishop, K. (2016). Mangiferin and Cancer: Mechanisms of

733 Action. Nutrients, 8(396), 1-25.

734 Gomez-Caravaca, A. M., Lopez-Cobo, A., Verardo, V., Segura-Carretero, A., & Fernandez-

735 Gutierrez, A. (2016). HPLC-DAD-q-TOF-MS as a powerful platform for the

736 determination of phenolic and other polar compounds in the edible part of mango and

737 its by-products (peel, seed, and seed husk). Electrophoresis, 37(7-8), 1072-1084.

738 Herrero, M., Castro-Puyana, M., Mendiola, J. A., & Ibañez, E. (2013). Compressed fluids for

739 the extraction of bioactive compounds. TrAC Trends in Analytical Chemistry, 43, 67-

740 83.

741 Hosu, A., Cristea, V.-M., & Cimpoiu, C. (2014). Analysis of total phenolic, flavonoids,

742 anthocyanins and tannins content in Romanian red wines: prediction of antioxidant

743 activities and classification of wines using artificial neural networks. Food chemistry,

744 150, 113-118.

745 Ibáñez, C., Simó, C., Palazoglu, M., & Cifuentes, A. (2017). GC-MS based metabolomics of

746 colon cancer cells using different extraction solvents. Analytica Chimica Acta, 986, 48-

747 56.

748 Imran, M., Arshad, M. S., Butt, M. S., Kwon, J. H., Arshad, M. U., & Sultan, M. T. (2017).

749 Mangiferin: a natural miracle bioactive compound against lifestyle related disorders.

750 Lipids in Health and Disease, 16(1), 84.

751 Jahurul, M. H., Zaidul, I. S., Ghafoor, K., Al-Juhaimi, F. Y., Nyam, K. L., Norulaini, N. A.,

752 Sahena, F., & Mohd Omar, A. K. (2015). Mango (Mangifera indica L.) by-products and

753 their valuable components: a review. Food Chemistry, 183, 173-180.

754 Jayasri, A., & Yaseen, M. (1980). Nomograms for Solubility Parameter. Journal of Coatings

755 Technology, 52(667), 41-45.

756 Khurana, R. K., Kaur, R., Lohan, S., Singh, K. K., & Singh, B. (2016). Mangiferin: a promising

757 anticancer bioactive. Pharmaceutical patent analyst, 5(3), 169-181.

758 Kim, W. H., Song, H. O., Choi, H. J., Bang, H. I., Choi, D. Y., & Park, H. (2012). Ethyl gallate

759 induces apoptosis of HL-60 cells by promoting the expression of caspases-8, -9, -3,

760 apoptosis-inducing factor and endonuclease G. International Journal of Molecular

761 Sciences, 13(9), 11912-11922.

762 Lieb, V. M., Schuster, L. K., Kronmüller, A., Schmarr, H.-G., Carle, R., & Steingass, C. B.

763 (2018). Fatty acids, triacylglycerols, and thermal behaviour of various mango

764 (Mangifera indica L.) kernel fats. Food Research International.

765 Lopez-Cobo, A., Verardo, V., Diaz-de-Cerio, E., Segura-Carretero, A., Fernandez-Gutierrez,

766 A., & Gomez-Caravaca, A. M. (2017). Use of HPLC- and GC-QTOF to determine

767 hydrophilic and lipophilic phenols in mango fruit (Mangifera indica L.) and its by-

768 products. Food Research International, 100(Pt 3), 423-434.

769 Luo, F., Fu, Y., Xiang, Y., Yan, S., Hu, G., Huang, X., Huang, G., Sun, C., Li, X., & Chen, K.

770 (2014). Identification and quantification of gallotannins in mango (Mangifera indica

771 L.) kernel and peel and their antiproliferative activities. Journal of Functional Foods,

772 8, 282-291.

773 Luo, F., Lv, Q., Zhao, Y., Hu, G., Huang, G., Zhang, J., Sun, C., Li, X., & Chen, K. (2012).

774 Quantification and purification of mangiferin from Chinese mango (Mangifera indica

775 L.) cultivars and its protective effect on human umbilical vein endothelial cells under

776 H2O2-induced stress. International Journal of Molecular Sciences.

777 Maisuthisakul, P., & Gordon, M. H. (2009). Antioxidant and tyrosinase inhibitory activity of

778 mango seed kernel by product. Food Chemistry, 117(2), 332-341.

779 Marrero, J., & Gani, R. (2001). Group-contribution based estimation of pure component

780 properties. Fluid Phase Equilibria, 183-184, 183-208.

781 Mertens-Talcott, S. U., Talcott, S. T., & Percival, S. S. (2003). Low Concentrations of

782 Quercetin and Ellagic Acid Synergistically Influence Proliferation, Cytotoxicity and

783 Apoptosis in MOLT-4 Human Leukemia Cells. The Journal of Nutrition, 133(8), 2669–

784 2674.

785 Mitra, S. K. (2016). Mango production in the world – present situation and future prospect.

786 Acta Horticulturae, 91(1111), 287-296.

787 Mohan, S., Thiagarajan, K., & Chandrasekaran, R. (2014). In vitroevaluation of

788 antiproliferative effect of ethyl gallate against human oral squamous carcinoma cell line

789 KB. Natural Product Research, 29(4), 366-369.

790 Nawab, A., Alam, F., Haq, M. A., & Hasnain, A. (2016). Effect of guar and xanthan gums on

791 functional properties of mango (Mangifera indica) kernel starch. International Journal

792 of Biological Macromolecules, 93(Pt A), 630-635.

793 Okiyama, D. C. G., Soares, I. D., Cuevas, M. S., Crevelin, E. J., Moraes, L. A. B., Melo, M.

794 P., Oliveira, A. L., & Rodrigues, C. E. C. (2018). Pressurized liquid extraction of

795 flavanols and alkaloids from cocoa bean shell using ethanol as solvent. Food Research

796 International, 114(May), 20-29.

797 Pereira, C. S. M., Silva, V. M. T. M., & Rodrigues, A. E. (2011). Ethyl lactate as a solvent:

798 Properties, applications and production processes – a review. Green Chemistry, 13(10),

799 2658-2658.

800 Rajendran, P., Rengarajan, T., Nandakumar, N., Divya, H., & Nishigaki, I. (2015). Mangiferin

801 in cancer chemoprevention and treatment: pharmacokinetics and molecular targets.

802 Journal of Receptors and Signal Transduction, 35(1), 76-84.

803 Re, R., Pellegrini, N., Proteggente, A., Pannala, A., Yang, M., & Rice-Evans, C. (1999).

804 Antioxidant Activity Applying an Improved ABTS Radical Cation Decolorization

805 Assay. Free Radical Biology & Medicine, 26(98), 1231-1237.

806 Ribeiro, S. M. R., & Schieber, A. (2010). Bioactive Compounds in Mango (Mangifera indica

807 L.). In W. Ronald Ross & R. P. Victor (Eds.), Bioactive Foods in Promoting Health.

808 Fruits and Vegetables (pp. 507-523). San Diego, USA: Elsevier Inc.

809 Rocha, L. D., Monteiro, M. C., & Teodoro, A. J. (2012). Anticancer Properties of

810 Hydroxycinnamic Acids -A Review. Cancer and Clinical Oncology, 1(2).

811 Ruales, J., Baenas, N., Moreno, D. A., Stinco, C. M., Meléndez-Martínez, A. J., & García-

812 Ruiz, A. (2018). Biological Active Ecuadorian Mango ‘Tommy Atkins’ Ingredients—

813 An Opportunity to Reduce Agrowaste. Nutrients, 10(9).

814 Salucci, M., Stivala, L. A., Bugianesi, R., & Vannini, V. (2002). Flavonoids uptake and their

815 effect on cell cycle of human colon adenocarcinoma cells (Caco2). British Journal of

816 Cancer, 86, 1645 – 1651.

817 Sánchez-Camargo, A. d. P., Montero, L., Cifuentes, A., Herrero, M., & Ibáñez, E. (2016).

818 Application of Hansen solubility approach for the subcritical and supercritical selective

819 extraction of phlorotannins from Cystoseira abies-marina. RSC Advances, 6(97),

820 94884-94895.

821 Sánchez-Camargo, A. d. P., Pleite, N., Herrero, M., Cifuentes, A., Ibáñez, E., & Gilbert-López,

822 B. (2017). New approaches for the selective extraction of bioactive compounds

823 employing bio-based solvents and pressurized green processes. The Journal of

824 Supercritical Fluids, 128, 112-120.

825 Soong, Y., & Barlow, P. (2006). Quantification of gallic acid and ellagic acid from longan

826 (Dimocarpus longan Lour.) seed and mango (Mangifera indica L.) kernel and their

827 effects on antioxidant activity. Food Chemistry, 97(3), 524-530.

828 Srinivas, K., King, J. W., Monrad, J. K., Howard, L. R., & Hansen, C. M. (2009). Optimization

829 of subcritical fluid extraction of bioactive compounds using Hansen solubility

830 parameters. Journal of food science, 74(6), E342-354.

831 Sultana, B., Hussain, Z., Asif, M., & Munir, A. (2012). Investigation on the antioxidant activity

832 of leaves, peels, stems bark, and kernel of mango (Mangifera indica L.). Journal of

833 Food Science, 77(8), C849-852.

834 Timsina, B., & Nadumane, V. K. (2015). Mango seeds : a potential source for the isolation of

835 bioactive compounds with anti-cancer activity. International Journal of Pharmacy and

836 Pharmaceutical Sciences, 7(3), 89-95.

837 Torres-León, C., Rojas, R., Contreras-Esquivel, J. C., Serna-Cock, L., Belmares-Cerda, R. E.,

838 & Aguilar, C. N. (2016). Mango seed: Functional and nutritional properties. Trends in

839 Food Science & Technology, 55, 109-117.

840 Valdes, A., Garcia-Canas, V., Rocamora-Reverte, L., Gomez-Martinez, A., Ferragut, J. A., &

841 Cifuentes, A. (2013). Effect of rosemary polyphenols on human colon cancer cells:

842 transcriptomic profiling and functional enrichment analysis. Genes & Nutrition, 8(1),

843 43-60.

844 Velderrain-Rodriguez, G. R., Torres-Moreno, H., Villegas-Ochoa, M. A., Ayala-Zavala, J. F.,

845 Robles-Zepeda, R. E., Wall-Medrano, A., & Gonzalez-Aguilar, G. A. (2018). Gallic

846 Acid Content and an Antioxidant Mechanism Are Responsible for the Antiproliferative

847 Activity of 'Ataulfo' Mango Peel on LS180 Cells. Molecules, 23(3).

848 Verma, S., Singh, A., & Mishra, A. (2013). Gallic acid: Molecular rival of cancer.

849 Environmental Toxicology and Pharmacology, 35(3), 473-485.

850 Williams, L. L., Rubin, J. B., & Edwars, H. W. (2004). Calculation of Hansen Solubility

851 Parameter Values for a Range of Pressure and Temperature Conditions Including the

852 Supercritical Fluid Region. Industrial & Engineering Chemistry Research, 43, 4967-

853 4972.

854 Xi, J., Yan, L., & He, L. (2014). Pressure-dependent kinetic modeling of solid-liquid extraction

855 of the major green tea constituents. Separation and Purification Technology, 133, 155-

856 159.

857 Yousef, A. I., El-Masry, O. S., & Yassin, E. H. (2016). The anti-oncogenic influence of ellagic

858 acid on colon cancer cells in leptin-enriched microenvironment. Tumor Biology,

859 37(10), 13345-13353.

860

861

862

863

864

865

866

867

868

869

870

871

872

873

874

875

876

877

878

879

880

881

882 Figure captions

883 Figure 1. Workflow of the proposed mango seed kernel (MSK) valorization strategy.

884 Figure 2. Kinetic behaviour of the extraction yield employing different solvents during the

885 defatting step. PLE extractions performed at 100 °C and 100 bar.

886 Figure 3. Standardized Pareto charts for the response variables studied and their

887 corresponding response surfaces.

888 Figure 4. Desirability value and predicted response variables in multi-response optimization.

889 Figure 5. TIC (A) and HREICs (B-E), corresponding to the phenolic fraction of MSK extracts

890 analysed by UHPLC-ESI(-)-q-TOF-MS/MS. (B) Mangiferin isomers; (C-D) Other phenolic

891 compounds; (E) Gallates and gallotannins.

892 Figure 6. GC-q-TOF(MS) profile of the non-polar fraction of MSK extracts obtained by

893 developed PLE procedure.

894 Figure 7. IC50 values and percentage of growth of HT-29 and CCD-18Co cells incubated for

895 (A) 48 and (B) 72 h, with different concentrations of ‘sugar MSK’ DM- and PLE-extracts. SI

896 of PLE-extract was expressed as ratio. (*) indicates significant differences

897 between the treated and control samples, p < 0.05. Error bars represent standard error of the

898 mean.

899

900

901

902

903

904

905

906

Table 1. Hansen Solubility Parameters and distance for mangiferin and green solvents at different temperatures

   Ra* Compound/Solvent T (°C) (MPa1/2) (MPa1/2) (MPa1/2) (MPa1/2) 25 20.3 10.7 12.5 0 50 20.2 10.6 12.4 0 Mangiferin 100 19.9 10.5 12.2 0 150 19.6 10.3 12 0 25 16 7.6 12.5 9.1 50 15 7.4 11.8 10.9 Ethyl lactate 100 13 7 10.4 14.9 150 11.2 6.6 9.2 18.6 25 15.8 8.8 19.4 11.5 50 14.7 8.5 18.2 12.6 Ethanol 100 12.6 8 16 15.2 150 10.5 7.5 14 18.4 25 15.8 5.3 7.2 11.8 50 14.9 5.2 6.8 13.1 Ethyl acetate 100 13.1 4.9 6.1 15.9 150 11.3 4.6 5.3 18.8 25 17.2 1.8 4.3 13.6 50 16.4 1.8 4.1 14.3 (+)-Limonene 100 14.9 1.7 3.7 15.7 150 13.5 1.6 3.3 17.2 25 15.8 7.1 13.3 9.7 Ethanol:ethyl 50 14.8 6.9 12.5 11.4 acetate 50:50 v/v 100 12.8 6.5 11 14.6 150 10.9 6.1 9.7 18 *Distance respect to mangiferin

Table 2. Experimental design conditions (experiments 1 to 10) and results for each response variable studied for the optimization of the PLE-second step and macerations (experiments I, II and III) of ‘sugar MSK’.

Mangiferin Extraction Codified variables Real variables TPC TFC EC TEAC content yield 50 Experiment Temperature Temperature %EtOH %EtOH (mg/g) (%) (mg GAE/g) (mg QE/g) (g/mL) (mM trolox/g) (°C) (°C) 1 -1 -1 0 50 0.48 ± 0.05 0.33 ± 0.01 5.05 ± 0.75 0.08 ± 0.01 42.51 ± 1.90 0.03 ± 0.01

2 -1 0 0 100 1.41 ± 0.33 0.89 ± 0.11 11.17 ± 0.72 0.14 ± 0.01 29.10 ± 0.16 0.10 ± 0.02

3 -1 +1 0 150 0.79 ± 0.16 2.87 ± 0.36 23.98 ± 1.78 0.60 ± 0.06 31.61 ± 1.55 1.25 ± 0.10

4 0 -1 50 50 3.44 ± 0.43 2.81 ± 0.23 39.40 ± 0.58 0.35 ± 0.04 48.56 ± 1.51 0.06 ± 0.02

5, 6 (CPa) 0 0 50 100 10.51 ± 0.55 8.19 ± 0.22 94.86 ± 3.38 1.10 ± 0.11 40.53 ± 2.59 1.74 ± 0.21

7 0 +1 50 150 10.60 ± 0.40 9.54 ± 0.50 93.42 ± 0.04 1.40 ± 0.13 30.01 ± 0.30 1.23 ± 0.31

8 +1 -1 100 50 3.00 ± 0.26 2.31 ± 0.01 31.12 ± 0.98 0.20 ± 0.01 15.89 ± 1.07 0.75 ± 0.15

9 +1 0 100 100 10.06 ± 0.63 8.93 ± 0.10 88.64 ± 1.70 0.74 ± 0.03 14.34 ± 0.19 1.41 ± 0.37

10 +1 +1 100 150 13.27 ± 1.67 12.15 ± 0.90 143.79 ± 2.09 1.21 ± 0.16 17.59 ± 1.19 2.31 ± 0.26

I 0 25 0.13 ± 0.06 9.24 ± 1.06 3.69 ± 0.30 2.75 ± 0.22 78.06 ± 2.65 0.21 ± 0.24

II 50 25 13.60 ± 2.22 11.96 ± 0.27 14.06 ± 0.32 3.72 ± 0.08 61.07 ± 0.72 0.05 ± 0.01

III 100 25 12.34 ± 0.37 6.82 ± 0.12 26.00 ± 1.75 1.61 ± 0.08 66.99 ± 1.02 0.08 ± 0.02 a Central point of the experimental design

Values presented are mean ± sd

Table 3. Tentatively identified compounds from ‘sugar MSK’ extract by LC-q-TOF-MS/MS analysis.

Ret. [M-H]- [M-H]- Peak Error MS2 product ions (-) Time Family Tentative identif. Formula (m/z) (m/z) No (ppm) (m/z) (min) (measured) (theoretical)

1 0.652 Phenolic acid Quinic acid C7H12O6 191.0570 191.0561 -4.6 127, 93, 85 2 0.825 Gallotannin Galloyl diglucoside isomer 1 C19H26O15 493.1200 493.1199 -0.2 313, 169, 125

3 1.737 Gallotannin Galloyl glucose isomer C13H16O10 331.0680 331.0671 -2.8 271, 241, 169, 125 4 1.997 Gallotannin Galloyl diglucoside isomer 2 C19H26O15 493.1200 493.1199 -0.2 313, 169, 125

5 2.090 Phenolic acid Gallic acid * C7H6O5 169.0139 169.0142 2.1 125, 79 6 2.214 Gallotannin Galloyl diglucoside isomer 3 C19H26O15 493.1200 493.1199 -0.2 313, 169, 125

7 2.301 Gallate Galloylquinic acid isomer 1 C14H16O10 343.0680 343.0671 -2.7 127, 169

8 2.691 Gallotannin Digalloyl diglucoside isomer 1 C26H30O19 645.1315 645.1309 -1.0 465, 313, 169

9 2.765 Gallate Galloylquinic acid isomer 2 C14H16O10 343.0680 343.0671 -2.7 127, 169

10 2.865 Gallotannin Galloyl diglucoside isomer 4 C19H26O15 493.1200 493.1199 -0.2 313, 169, 125

11 3.219 Phenolic acid Protocatechuic acid * C7H6O4 153.0197 153.0193 -2.4 109, 91 13 3.299 Gallotannin Digalloyl glucose isomer 1 C20H20O14 483.0790 483.0780 -2.0 331, 313, 169, 125

14 3.299 Gallotannin Digalloyl diglucoside isomer 2 C26H30O19 645.1315 645.1309 -1.0 465, 313, 169

15 3.559 Gallotannin Digalloyl diglucoside isomer 3 C26H30O19 645.1315 645.1309 -1.0 465, 313, 169

16 3.776 Benzophenone Maclurin-C-glucoside C19H20O11 423.0937 423.0933 -1.0 193, 303

17 3.907 Gallotannin Digalloyl glucose isomer 2 C20H20O14 483.0790 483.0780 -2.0 331, 313, 169, 125

18 3.907 Gallotannin Digalloyl diglucoside isomer 4 C26H30O19 645.1315 645.1309 -1.0 465, 313, 169

19 4.173 Phenolic acid p-Hydroxybenzoic acid * C7H6O3 137.0246 137.0244 -1.3 93

20 4.254 Gallotannin Digalloyl diglucoside isomer 5 C26H30O19 645.1315 645.1309 -1.0 465, 313, 169 Hexahydroxylated benzophenone 21 4.297 Benzophenone C H O 277.0356 277.0354 -0.8 125 isomer 1 13 10 7 22 4.514 Gallotannin Digalloyl diglucoside isomer 6 C26H30O19 645.1315 645.1309 -1.0 465, 313, 169 483, 465, 423, 313, 295, 23 4.557 Gallotannin Trigalloyl glucose isomer 1 C H O 635.0907 635.0890 -2.7 27 24 18 169, 125 24 4.644 Benzophenone Iriflophenone glucoside C19H20O10 407.0991 407.0984 -1.8 317, 287 25 4.688 Gallate Methylgallate C8H8O5 183.0300 183.0299 -0.6 168, 124

26 4.688 Gallotannin Digalloyl glucose isomer 3 C20H20O14 483.0790 483.0780 -2.0 331, 313, 169, 125

27 4.861 Gallotannin Digalloyl glucose isomer 4 C20H20O14 483.0790 483.0780 -2.0 331, 313, 169, 125 40

28 4.861 Gallotannin Digalloyl diglucoside isomer 7 C26H30O19 645.1315 645.1309 -1.0 465, 313, 169

29 4.948 Gallate Digallic acid C14H10O9 321.0259 321.0252 -2.2 169, 125

30 4.948 Benzophenone Maclurin-C-(O-galloyl)-glucoside C26H24O15 575.1047 575.1042 -0.8 193, 303, 333

31 4.999 Flavonoid Catechin * C15H14O6 289.0720 289.0718 -0.8 245, 203, 109 32 5.035 Gallotannin Digalloyl diglucoside isomer 8 C26H30O19 645.1315 645.1309 -1.0 465, 313, 169 483, 465, 423, 313, 295, 34 5.686 Gallotannin Trigalloyl glucose isomer 2 C H O 635.0907 635.0890 -2.7 27 24 18 169, 125 483, 465, 423, 313, 295, 35 5.859 Gallotannin Trigalloyl glucose isomer 3 C H O 635.0907 635.0890 -2.7 27 24 18 169, 125 36 3.000 Xanthone Mangiferin* C19H18O11 421.0787 421.0776 -2.5 403, 331, 301, 259

37 6.033 Flavonoid Epicatechin C15H14O6 289.0720 289.0718 -0.8 245, 203, 109 C H O 483, 465, 423, 313, 295, 38 6.120 Gallotannin Trigalloyl glucose isomer 4 27 24 18 635.0907 635.0890 -2.7 169, 125 483, 465, 423, 313, 295, 39 6.337 Gallotannin Trigalloyl glucose isomer 5 C H O 635.0907 635.0890 -2.7 27 24 18 169, 125 40 6.337 Benzophenone Maclurin-C-(O-digalloyl)-glucoside C33H28O19 727.1157 727.1152 -0.7 169, 303, 575

41 6.423 Xanthone Homomangiferin C20H20O11 435.0940 435.0933 -1.6 345, 315, 285, 272 483, 465, 423, 313, 295, 42 6.467 Gallotannin Trigalloyl glucose isomer 6 C H O 635.0907 635.0890 -2.7 27 24 18 169, 125 43 6.684 Gallate Ethyl gallate C9H10O5 197.0471 197.0455 -7.9 169, 124 483, 465, 423, 313, 295, 44 6.771 Gallotannin Trigalloyl glucose isomer 7 C H O 635.0907 635.0890 -2.7 27 24 18 169, 125 45 6.901 Gallotannin Tetragalloyl glucose isomer 1 C34H28O22 787.1019 787.1000 -2.5 617, 465, 295, 169, 125 46 6.944 Xanthone Mangiferin gallate C26H22O15 573.0888 573.0886 -0.4 403, 331, 301

47 7.344 Phenolic acid Ferulic acid * C10H10O4 193.0506 193.0506 0.2 178, 134 Hexahydroxylated benzophenone 48 7.378 Benzophenone C H O 277.0356 277.0354 -0.8 125 isomer 2 13 10 7 49 7.465 Gallotannin Feruloyl galloyl glucose isomer 1 C22H22O13 493.0991 493.0988 -0.7 323, 295, 169, 125

50 7.508 Xanthone Mangiferin isomer 1 C19H18O11 421.0787 421.0776 -2.5 403, 331, 301, 259

51 7.638 Gallotannin Tetragalloyl glucose isomer 2 C34H28O22 787.1019 787.1000 -2.5 617, 465, 295, 169, 125

52 7.725 Phenolic acid Ellagic acid C14H6O8 300.9990 300.9990 0.0 229, 145 Hexahydroxylated benzophenone 53 7.812 Benzophenone C H O 277.0356 277.0354 -0.8 168, 124 isomer 3 13 10 7 54 7.812 Flavonoid Quercetin glucoside C21H20O12 463.0884 463.0882 -0.4 301 41

55 7.899 Gallotannin Feruloyl galloyl glucose isomer 3 C22H22O13 493.0991 493.0988 -0.7 323, 295, 169, 125

56 7.942 Gallotannin Tetragalloyl glucose isomer 3 C34H28O22 787.1019 787.1000 -2.5 617, 465, 295, 169, 125

57 8.029 Flavonoid (Epi)catechin gallate C22H18O10 441.0830 441.0827 -0.6 289, 169

59 8.159 Gallate Galloyl methylgallate C15H12O9 335.0410 335.0409 -0.4 183, 124

60 8.246 Gallotannin Tetragalloyl glucose isomer 4 C34H28O22 787.1019 787.1000 -2.5 617, 465, 295, 169, 125

61 8.333 Xanthone Mangiferin isomer 2 C19H18O11 421.0787 421.0776 -2.5 403, 331, 301, 259

62 8.376 Gallotannin Feruloyl galloyl glucose isomer 2 C22H22O13 493.0991 493.0988 -0.7 323, 295, 169, 125

63 8.463 Gallotannin Tetragalloyl glucose isomer 5 C34H28O22 787.1019 787.1000 -2.5 617, 465, 295, 169, 125

64 8.506 Gallotannin Pentagalloyl glucose C41H32O26 939.1120 939.1109 -1.2 787, 768, 617, 465, 169

65 8.767 Gallotannin Feruloyl galloyl glucose isomer 4 C22H22O13 493.0991 493.0988 -0.7 323, 295, 169, 125

66 8.940 Gallotannin Feruloyl galloyl glucose isomer 5 C22H22O13 493.0991 493.0988 -0.7 323, 295, 169, 125

67 9.070 Gallate Galloyl ethylgallate C16H14O9 349.0573 349.0565 -2.3 197, 169, 124

68 10.290 Gallotannin Feruloyl galloyl glucose isomer 6 C22H22O13 493.0991 493.0988 -0.7 323, 295, 169, 125

69 11.126 Flavonoid Quercetin * C15H10O7 301.0353 301.0354 0.3 178, 151

70 11.500 Gallate Ethyl trigallate C23H18O13 501.0681 501.0675 -1.3 359, 197

71 12.010 Gallotannin Hexagalloyl glucose C48H36O30 1091.1230 1091.1219 -1.0 787, 768, 617, 465, 169 * Identification confirmed by commercial standard

Table 4. Tentatively identified compounds from ‘sugar MSK’ extract by GC-TOF-MS analysis.

Ret. m/z [M+R]+ m/z [M+R]+ Peak Tentative Match Monoisotopic Error Main fragments Time Family Formula (measured) (calculated) No identif. Factor mass (ppm) (m/z) (min) * * 1 15.978 Tricarboxylic acid Citric acid* 79 C6H8O7 192.0270 - - - 396, 357, 145, 105, 74 a a 2 18.286 FA (16:0) Palmitic acid* 86 C16H32O2 256.2402 313.2567 313.2557 -3.1 129, 117, 75 a a 3 19.803 FA (C18:2Δ9,12) ω-6 Linoleic acid * 79 C18H32O2 280.2402 337.2574 337.2557 -4.9 262, 149, 117, 95 a a 4 19.858 FA (C18:1Δ9) ω-9 Oleic acid* 89 C18H34O2 282.2559 339.2720 339.2714 -1.8 145, 129, 117, 75 a a 5 20.090 FA (C18:0) Stearic acid * 82 C18H36O2 284.2715 341.2877 341.2870 -1.9 145, 129, 117, 75

6 23.234 FA (C20:5Δ5,8,11,14,17) ω-3 Eicosapentaenoi 69 C20H30O2 302.2246 - - - 361, 217, 169, c acid* 147, 73 b b 7 28.133 Terpenoid β-Sitosterol * 77 C29H50O 414.3862 486.4264 486.4251 -2.6 396, 357, 145, 105, 73

* Trimethylsilyl (TMS) derivative: a R = (-H+TMS-CH3); b R = (-H+TMS)