Effects of thermal treatments on 10 major phenolics and their antioxidant contributions in Acer truncatum leaves and flowers

This study aimed to investigate effects of thermal treatments on major phenolics and their antioxidant contributions in Acer truncatum leaves and flowers (ATL and ATF, respectively). With ultra performance liquid chromatography-diode array detector-quadrupole time-of-flight-mass spectrometer/mass spectrometer, phenolic compositions of ATF were first characterized and compared with those of ATL. An optimized high performance liquid chromatography fingerprint was then established, and 10 major phenolics existing in both ATL and ATF were quantified. Gallic acid derivatives and flavonol-3-O-glycosides were found to be their dominant phenolic constituents, with the former being key constituents which was affected by thermal treatments and further influencing the variations of total phenols. Moreover, the mechanism underlining the changes of phenolics in ATL and ATF by the treatments was characterized as a thermolhydrolysis process. During thermal treatments, polymerized gallotannins were hydrolysed to 1,2,3,4,6-pentakis-O-galloyl-β-d-glucose, ethyl gallate and gallic acid, resulting in more than fivefold and twofold increase of their contents in ATL and ATF, respectively. By contrast, contents and antioxidant contributions of flavonol-3-O-glycosides gradually decreased during the process.\absbreak Overall, this is, to our knowledge, the first report on the effects of thermal treatments on phenolics and their antioxidant contributions in ATL and ATF, and the three gallic acid derivatives with potentially higher bioactivity could be efficiently achieved by thermal treatments.

Twenty Acer truncatum Bunge trees with similar tree-age and growing environment were randomly selected in Bajia Outskirts Park (GPS coordinates, 40°00'57. 21  on 20 April 2017, which was the florescence period of A. truncatum, as well as an optimal season with high phenolic contents and antioxidant activity in ATL [15].

Thermal treatments of Acer truncatum leaf and flower extracts
To prepare phenolic extracts, the ultrasonic-assisted extraction procedure previously optimized was conducted [14]. To be specific, 15.30 ml of 66.20% ethanol was added to 1.000 g air-dried and ground ATL or ATF powder, the mixture was then sonicated (KQ-300DE type, Kunshan Ultrasonic Instrument Co., Ltd., Kunshan, China) in a 270 W and 50°C water bath for 30 min. The sonicated mixture was filtered to obtain the supernatant. The extraction process was repeated two more times with the residue, and three supernatants were poured together and diluted to 45.90 ml for further experiments.
For thermal treatment, 10 ml of ATL or ATF extract solutions was placed at 100°C in a sirocco-blasting drying trunk. The treatment time was set at 0.5, 1, 2, 4, 8, 10, 12 or 24 h. After the treatments, all extract solutions were diluted to 10 ml for further experiments.

Quantification of individual phenolics by high performance liquid chromatography analysis
HPLC analyses were performed using a Shimadzu HPLC system (Shimadzu, Kyoto, Japan) equipped with two LC-10AT VP pumps: a SPDM20A ultraviolet detector and a SIL-20AC TH autosampler. The reversed phase column, column temperature, solvent system, gradient programme, as well as the flow rate were the same as those in the UPLC analysis described above. Detection wavelength was set at 280 nm to monitor phenols simultaneously.

Measurement of total phenols by Folin-Ciocalteu assay
Total phenols were determined according to the Folin-Ciocalteu method [17] with slight modifications. In brief, 40 µl of 25% Folin-Ciocalteu solution was added to a 96-well plate, followed by addition of a 20 µl standard (10-400 mg l −1 gallic acid, R 2 = 0.999), sample or blank (MilliQ water) to designated wells. After blending, 140 µl of 700 mM Na 2 CO 3 solution was added to each well and the plate was shaken for 5 min at 250 r.p.m. The microplate was then incubated in the dark at 40°C for 30 min, followed by absorbance measurement at 765 nm with a microplate reader (Bio-Rad xMark™ Microplate Absorbance Spectrophotometer, USA). Results were expressed as mg gallic acid equivalent (GAE) 100 g d.w. −1 ATL or ATF powder.
2.6. Assessment of antioxidant capacities 2.6.1. 2, 2 -diphenyl-1-picrylhydrazyl• scavenging activity 2,2 -diphenyl-1-picrylhydrazyl (DPPH)• scavenging activity was determined by Brand-Williams et al. [18] with slight modifications. In brief, 10 µl standard, sample or blank was added to designated wells of a 96-well microplate, followed by addition of 40 µl of 1 mM freshly prepared DPPH solution to each well. Subsequently, 190 µl methanol was added to each well, and the plate was then stood in an orbital shaker setting at 200 r.p.m. for 1 min. After 30 min incubation in the dark at room temperature, absorbance was recorded at 517 nm using the microplate reader. A standard calibration curve of Trolox (0-400 mg l −1 , R 2 = 0.999) was plotted, and results were expressed as μmol Trolox equivalent (TE) 100 g d.w. −1 ATL or ATF powder.
2.6.2. 2, 2 -azobis-(3-ethylbenzothiazoline-6-sulfonic acid) + • scavenging activity ABTS + • scavenging capacities were evaluated by using the method of Re et al. [19] with slight modifications. ABTS + • was generated by the reaction of a 7 mM ABTS aqueous solution with a 2.4 mM K 2 S 2 O 8 aqueous solution in equivalents, and the mixture was incubated in the dark at room temperature for 12-16 h. Subsequently, the ABTS solution was diluted with methanol to an absorbance of 0.70 ± 0.02 at 734 nm measured using the microplate reader to generate an ABTS working solution. After the working solution was prepared, a 5 µl standard, sample or blank was added to corresponding wells in a 96-well microplate, followed by addition of 200 µl ABTS working solution to each well. After 5 min incubation in the dark at 30°C, the absorbance was measured also at 734 nm. A standard calibration curve of Trolox (0-800 mg l −1 , R 2 = 0.998) was plotted, and the results were expressed as μmol Trolox equivalent (TE) 100 g d.w. −1 ATL or ATF powder.

Oxygen radical absorbance capacity
Oxygen radical absorbance capacity (ORAC) assay was performed with all reagents prepared in a 75 mM phosphate buffer (pH = 7.4) according to a report by Sun et al. [20], and direct light was not allowed during the whole process. Briefly, a 25 µl standard (Trolox 5-50 µM, R 2 = 0.997), sample or blank (75 mM phosphate buffer) was mixed with 75 µl fluorescein (0.20 µM) in wells of a 96-well microplate. After 5 min shaking at 250 r.p.m., the plate was incubated in a 37°C-prewarmed oven for 15 min. Subsequently, 100 µl of 37°C-prewarmed AAPH was immediately added to each well with a 12-channel multipipet to initiate the reaction. The fluorescence was recorded every 1.5 min with a multi-functional fluorescence detector (Tecan infinite M200, Swiss) at 37°C for 75 min with an excitation at 530 nm and emission at 485 nm. The net areas under curve (AUC) of samples and standards were calculated by subtracting the AUC of the blank. Results were calculated by comparing the net AUC of each sample with that of the standard. ORAC values were calculated as μmol Trolox equivalents (TE) 100 g d.w. −1 ATL or ATF powder.

Data analysis
Each experiment was carried out in triplicate and data were expressed as mean ± standard deviation. The statistical significance was determined by Tukey's HSD test in one-way ANOVA using SPSS software (v. 22.0, SPSS Inc., Chicago, IL, USA). p < 0.05 was set to be significant.

Results and discussion
3.1. Characterization and classification of phenolics in Acer truncatum leaves and flowers with ultra performance liquid chromatography-diode array detector-quadrupole time-of-flight-mass spectrometer/mass spectrometer and high performance liquid chromatography fingerprint  [21]; 'Di' represents Díaz-de-Cerio et al. [22]; 'A' represents Aguirre-Hernandez et al. [23]; 'P' represents Petropoulos et al. [24]; 'Do' represents Dorta et al. [25]; 'ST' represents those confirmed by the authentic standards while those without a superscript letter was described in Yang et al. [14].) peak no.  ATL and ATF in this study has been described in our previous study [14]. As for those phenolics first found in A. truncatum, the identification was conducted by comparing their ultraviolet spectra, mass spectra and fragment ions to those in the literature and the METLIN database (https://metlin.scripps. edu). For instance, as shown in table 1, compound 20 in ATL with the parent ion at m/z 577.1586 was identified as kaempferol-3,7-O-dirhamnoside in that it fragmented at 431.0899 with the loss of a rhamnoside moiety (146) and at 285.0413 (corresponding to kaempferol aglycone) with the loss of another rhamnoside moiety (146) [23]. As shown in table 1 and figure 1a, 27 phenolics were identified in unprocessed ATL. Among which, 22 of them had been identified in ATL previously [14] and five of them, namely, cis-4-caffeoylquinic acid (9 in ATL profile; similarly hereinafter), trans-4-caffeoylquinic acid (10), quercetin-galloylhexoside (19), kaempferol-3,7-O-dirhamnoside (20) and kaempferol-3-O-glucuronide (27), were identified in A. truncatum for the first time (table 1 and the green curve of figure 1a). On the other hand, 19 phenolics were identified in ATF (table 1 and the yellow curve of figure 1a). As this is the first report on phenolics composition of ATF, all these 19 phenolics were considered to be identified in it for the first time. Nevertheless, 14 of them had been previously identified in ATL, implying that phenolics composition of ATF was similar to that of ATL. In other words, the other five phenolics, namely, galloyl glucose (2 in ATF profile; similarly hereinafter), quercetin-3-O-pentosylhexoside (7), quercetin-3-O-rutinoside (9), kaempferol-3-O-(6-p-coumaroyl)-glucoside (13) and kaempferol-3-O-glucoside (15), were first identified in A. truncatum. Furthermore, all those 10 phenolics identified for the first time in A. truncatum in the current study still belonged to gallates, gallotannins, chlorogenic acids and flavonoids [14], which further indicated that phenolic compositions of ATF and ATL consisted of mainly these four groups.  1). Therefore, they were further confirmed by comparing their MS/MS data and retention time with those of the authentic standards (figure 1b), and subsequently used as the markers for both ATL and ATF in the following HPLC quantification during the thermal treatments.
As shown in figure 2a,b, HPLC spectra of ATL and ATF clearly showed that the 10 phenolics exhibited higher intensities than other compounds during the whole thermal treatment. Therefore, we preliminary considered them as major phenolics in ATL and ATF. Meanwhile, there was a set of less intensive peaks (other than compound 10) eluted out between 75 and 100 min in both profiles, which were difficult to be separated further. For those 10 major phenolics, they could be divided into four categories, i.e. gallates (1 and 3), caffeoylquinic acids (2), flavonol-3-O-glycosides (4, 5, 6, 7, 9 and 10) and gallotannins (8) based on their chemical structures (figure 2c). As both gallates and gallotannins shared gallic acid as a common subunit, they were regrouped as gallic acid derivatives. In addition, caffeoylquinic acids were recombined with flavonol-3-O-glycosides as neochlorogenic acid was the only one in the caffeoylquinic acids category with relatively low intensity. Therefore, we further divided the 10 major phenolics into two main groups, namely, gallic acid derivatives and flavonol-3-O-glycosides.  (table 2). During thermal treatment, three flavonol-3-O-rhamnosides kept relatively steady in the beginning, and dramatically decreased after certain treatment duration. For instance, kaempferol-3-O-rhamnoside slightly increased to 109.63% at 2 h, but dramatically decreased to 34.18% at 8 h, and continuously declined to 20.7% at 24 h compared with those in unprocessed ATL (table 2).
On the contrary, the three gallic acid derivatives considerably increased during the first 8 h. There was a 7.14-and 5.09-fold increase for 1,2,3,4,6-pentakis-O-galloyl-β-D-glucose and ethyl gallate at 8 h, respectively. After that, they both gradually decreased but remained to be the top among the 10 phenolics (table 2). It should be noted that, among the 10 major phenolics, gallic acid was the only one that kept increasing during the entire thermal treatment, giving a 6.31-fold increase within 24 h. As gallic acid is the common basic unit of 1,2,3,4,6-pentakis-O-galloyl-β-D-glucose and ethyl gallate, and both of them stopped increasing after 8 h, we hypothesized that the rise of gallic acid in 8 h might be caused by their depolymerization in the prolonged thermal treatment. It can also be speculated that the increased generation of 1,2,3,4,6-pentakis-O-galloyl-β-D-glucose and ethyl gallate within the first 8 h might be caused by the reduction of other unidentified compounds with even higher molecular weight.
As also shown in table 2, total phenols determined by HPLC (TPLC) and Folin-Ciocalteu assay (TPFC) of ATL both increased from 0 to 8 h, and then decreased. Therefore, gallic acid derivatives might be the dominant constituents making up total phenols and influenced by the thermal treatment, which can be validated by the fact that there existed extreme significance (R 2 > 0.842, p < 0.01) among 1,2,3,4,6pentakis-O-galloyl-β-D-glucose, ethyl gallate and TPLC. In addition, the proportion of TPLC over TPFC gradually increased from 55.18% in unprocessed ATL to approximately 100% from 4 to 10 h (100.82% at 4 h, 96.83% at 8 h and 95.96% at 10 h), indicating that nearly all phenolics in ATL were converted to the 10 major phenolics during the thermal treatment.

Changes of the 10 major phenolics in Acer truncatum flowers
As for ATF, total content of the three gallic acid derivatives (3064.63 mg 100 g d.w. −1 ) were higher than the sum of all six flavonol-3-O-glycosides (2222.24) before treatment, indicating that dominant phenolic constituents in unprocessed ATF were different from those in unprocessed ATL. Moreover, flavonol-3-O-rhamnosides were no longer the leading compounds of the six flavonol-3-O-glycosides as quercetin-3-O-galactoside (562.14) and quercetin-3-O-arabinopyranoside (326.33) both exhibited considerable levels (table 2). Flavonol-3-O-glycosides kept relative steady at the beginning, then significantly decreased to their lowest levels at 24 h, but their declines were much more rapid than those in ATL. As for gallic acid derivatives, 1,2,3,4,6-pentakis-O-galloyl-β-D-glucose and ethyl gallate increased from 0 to 2 h in ATF instead of from 0 to 8 h in ATL, giving rise to a 2.18-and 2.87-fold rise, respectively. While 1,2,3,4,6pentakis-O-galloyl-β-D-glucose kept decreasing after 2 h following the pattern in ATL, there was no significant difference among ethyl gallate at 2 h and those with longer thermal treatment. On the other hand, gallic acid still kept increasing during the entire thermal treatment, bringing about a 4.90-fold increase within 24 h. Overall, the dynamic patterns of the 10 major phenolics in ATF were similar to those in ATL, with the reactions in the former being more drastic than those in the latter.
In ATF, TPLC were the highest at 2 h while TPFC were the highest at 1 and 2 h with no significant difference (table 2). The increase of TPLC in ATF was also mainly affected by 1,2,3,4,6-pentakis-Ogalloyl-β-D-glucose and ethyl gallate as indicated by significant correlations among them (R 2 > 0.686, p < 0.05). Meanwhile, the proportion of TPLC over TPFC in ATF also gradually increased from 59.53% to higher than 89% at 2-12 h.

Mechanism underlining the changes of phenolics in Acer truncatum leaves and flowers
According to previous studies, thermal treatment could liberate oligomers from their native polymers which are represented as repeating subunits of the oligomers [26]. Moreover, Kim et al. [27] reported      Figure 3. HPLC profiles of ATL and ATF with and without thermal treatment. All chromatographs of ATL (a) and ATF (b) were obtained under the same HPLC condition described. Insets on the upper right corner of (a) and (b) are the enlargement of (a) and (b) between 75 and 100 min, respectively, together with corresponding chromatographs of tannic acid. 'L0' and 'L8' represents ATL for thermal treatment at 0 and 8 h; 'F0' and 'F2' represents ATF for thermal treatment at 0 and 2 h. the enhancement of antioxidant ability of tannic acid after the thermal process and concluded that it was mainly contributed by the increase of pyrogallol and gallic acid during the thermal hydrolysis [28]. Therefore, the large increase of the three gallic acid derivatives in ATL and ATF upon thermal treatment in the present study might indicate drastic depolymerization of gallotannins with higher molecular weight.
As shown in figure 3, peaks eluted out between 75 and 100 min in the profiles of unprocessed ATL and ATF totally vanished after the thermal treatment at 8 h for ATL (see dark green curve in figure 3a) and at 2 h for ATF (see dark yellow curve in figure 3b) except for kaempferol-3-O-rhamnoside (compound 10). After comparing the HPLC chromatograph of tannic acid (see black curves in the insets in figure 3a,b) with those of unprocessed ATL and ATF, peaks eluted out between 75 and 100 min and their HPLC profiles largely coincided. As reported by Rodríguez et al. [29], commercial tannic acid was usually a gallotannins mixture extracted from plants and tended to undergo autodegradation even at room temperature, thus was composed of a set of depolymerized gallotannins and gallates with similar structure and polarity. Therefore, it could be concluded that compounds eluted out between 75 and 100 min in unprocessed ATL and ATF were high polymerization gallotannins which were eventually converted to 1,2,3,4,6-pentakis-O-galloyl-β-D-glucose, ethyl gallate and gallic acid. Meanwhile, the main chemical changes during the thermal treatment could be identified as the thermal hydrolysis of gallotannins.
It was reported that the higher the molecule weight and/or polymerization of tannins, the stronger the anti-nutritional effects and the lower the biological activities [30]. Therefore, thermal treatment is a simple and effective way to hydrolyse harmful gallotannins with high molecule weight to improve health benefits of ATL and ATF. Moreover, thermal hydrolysis led to exponential increases of 1,2,3,4,6pentakis-O-galloyl-β-D-glucose, ethyl gallate, and gallic acid, especially in ATL, indicating that it would be an effective preparation method to obtain these three valuable compounds.    For gallic acid derivatives, polymerization was an antioxidant increasing step based on molar concentration as 1,2,3,4,6-pentakis-O-galloyl-β-D-glucose (8) showed a significant higher antioxidant ability (3.59-fold for DPPH, and 3.38-fold for ABTS) than that of gallic acid (1). However, a contrary result was obtained based on mass unit as one 1,2,3,4,6-pentakis-O-galloyl-β-D-glucose consisted of five gallic acid units while it exhibited less than fivefold higher DPPH or ABTS value comparing with gallic acid. As for flavonol-3-O-rhamnosides, their DPPH and ABTS values were ranked as myricetin-3-O-rhamnoside (4) > quercetin-3-O-rhamnoside (9) > kaempferol-3-O-rhamnoside (10), exhibiting that the more hydroxyl groups on their ring B, the higher DPPH• and ABTS + • scavenging activity they possessed. The same pattern was observed in the study reported by Tabart et al. [34], as they determined the DPPH value of myricetin-3-O-rhamnoside was higher than that of quercetin-3-O-rhamnoside, and both of them were higher than that of kaempferol-3-O-glycosie. Among quercetin-3-O-glycosides, quercetin-3-O-glucoside (6)  exhibited the highest ABTS value. They both showed a higher scavenging ability than quercetin-3-Ogalactoside (5), while the quercetin-3-O-arabinopyranoside (7) exhibited the lowest in both assays.
Gallic acid derivatives, which were significantly higher in DPPH and ABTS activities, exhibited lower ORAC values than flavonol-3-O-glycosides and caffeoylquinic acids. Tabart et al. [34] reported that gallic acid possesses a relative similar activity in DPPH and ORAC assays while all flavonols, flavan-3-ols and flavanons exhibited a much higher ORAC value than DPPH value, which supported the results of this study. In addition, myricetin-3-O-rhamnoside (4), with three hydroxyl groups on its ring B, exhibited the lowest ORAC value while kaempferol-3-O-rhamnoside (10), with only one hydroxyl group on its ring B, exhibited the highest value, which was very different from the patterns in DPPH and ABTS assays. A few studies investigated the ORAC values of kaempferol, quercetin and myricetin, giving a concordant rank order as kaempferol > quercetin > myricetin [34][35][36]. Thus, the conclusion could be drawn that the number of hydroxyl groups on the ring B exhibited a negative impact on ORAC values for myricetin, quercetin and kaempferol, as well as their 3-O-glycosides. On the other hand, the ORAC order of quercetin-3-O-glycosides was quercetin-3-O-rhamnoside (9) > quercetin-3-O-glucoside (6) > quercetin-3-O-galactoside (5) > quercetin-3-O-arabinopyranoside (7), which was the same with that in ABTS and similar with that in DPPH, indicates that these glycosides exerted diverse impacts on their quercetin aglycone in the same order in all three antioxidant assays.
It is known that radicals are mainly quenched by two chemical principles, transfer of either a hydrogen atom or a single electron to convert the radicals to stable forms [37][38][39]. DPPH and ABTS assays are based on the transfer of a single electron, and measure the electron-donating capacity of antioxidants to reduce the radicals (DPPH• or ABTS + •). As for the ORAC method, antioxidants slow the loss of fluorescence by quenching the peroxyl radicals produced by AAPH via transfer of hydrogen atom or radical addition. The difference between these two mechanisms might elucidate the huge variation between the rank order of the 10 major phenolics in DPPH and ABTS and ORAC assays. Furthermore, it revealed that the 10 major phenolics, as main phenolic composition in both ATL and ATF, exhibited not only a high electron-donating but also a high chain-breaking capacity.

Antioxidant contributions of the 10 major phenolics to total antioxidant activity in Acer truncatum leaves and flowers
Apart from contents of the 10 major phenolics and their individual antioxidant activities, contributions of them to the total antioxidant activities were also evaluated. Within a single sample, the calculated contribution of each phenolic compound was obtained by multiplying its content measured by HPLC and the corresponding DPPH, ABTS or ORAC value (Trolox equivalent). The sum of the contributions of all 10 phenolics in the single sample was defined as the total calculated antioxidant activity. As shown in figure 5, contributions of the 10 major phenolics (colour columns) to the total measured DPPH ( figure 5a,b), ABTS (figure 5c,e) and ORAC (figure 5d,f ) values (dark lines) ranged from 35.15% (DPPH for ATL at 0 h) to 123.93% (ORAC for ATL at 4 h). Within most of the treatments, the proportion was higher than 50%, further confirming the dominant contribution of the 10 major phenolics. In addition, all contribution proportions (total calculated antioxidant activity/total measured antioxidant activity) for DPPH, ABTS and ORAC values were lower at the beginning and end period within the 24 h treatment for both ATL and ATF. The HPLC-undetectable gallotannins with high polymerization was responsible for the low contribution proportion at the beginning of the treatment while the HPLCundetectable hydrolysis products other than the 10 major phenolics was responsible for that at the end period of the treatment.
During the entire treatment, gallic acid derivatives always made an important contribution to the total activities of both measured and calculated, accounting for more than 50% of the total calculated activity in ATL, and more than 90% of that in ATF with both DPPH and ABTS assay (see blue colour columns in figure 5a-d). On the contrary, they made a relatively lower contribution to ORAC values as they accounted for 15.08%-59.99% in ATL and 34.79%-73.39% in ATF from 0 to 24 h (see cold colour columns in figure 5e,f )  and gradually decreased to 35.20% and 20.86% at 24 h, respectively (see warm colour and green columns in figure 5e,f ). As shown in figure 5a-c,e, there is a certain period for DPPH in ATF and for all three assays in ATL when the contribution proportions were found to be even larger than 100%. It might reveal that there is specific compound(s) inhibiting the antioxidant capacity of phenolics in ATL and ATF extracts and thus limiting the total antioxidant activity, or certain phenolics in ATL and ATF showed an antagonistic effect in total antioxidant activity. Palafox-Carlos et al. [40] reported that isolated gallic, protocatechuic, chlorogenic and vanillic acids from mango cultivar 'Ataulfo' showed high antioxidant activities against DPPH•. However, the antagonistic effect was observed if gallic acid was combined with vanillic acid, or in the case that protocatechuic was combined with chlorogenic and vanillic acids. Moreover, Heo et al. [41] found that the addition of chlorogenic acid to cyanidin, peonidin-3-O-glucoside, quercetin and quercetin-3-O-galactoside caused a decrease of DPPH values compared with their total calculated antioxidant capacity. A number of phenolics mentioned above, including gallic and neochlorogenic acid, and quercetin-3-O-galactoside were important phenolic constituents of ATL and ATF. Therefore, we assume that certain phenolics in ATL and ATF exhibited an antagonistic effect to the total antioxidant activity, resulting in that there are several ATL and ATF samples whose total calculated activity were higher than their total measured activity as shown in figure 5a-c,e.
Overall, the thermal treatment possessed a positive effect on gallic acid derivatives with aspect to both their contents and antioxidant contributions (see cold colour columns in figure 5). However, total measured antioxidant activity did not remarkably increase (see dark lines in figure 5) partially owing to the fact that flavonol-3-O-glycosieds, caffeoylquinic acids and their antioxidant contributions gradually decreased under the treatment (see warm colour and green columns in figure 5). Based on this conjecture,