Antioxidant evaluation-guided chemical profiling and structure-activity analysis of leaf extracts from five trees in Broussonetia and Morus (Moraceae)

Morus and Broussonetia trees are widely used as food and/or feed. Among 23 phenolics identified from leaves of five Moraceae species using UPLC–QTOF–MS/MS, 15 were screened using DPPH/ABTS-guided HPLCs, including seven weak (flavonoids with one hydroxyl on B-ring) and eight strong (four caffeoylquinic acids and four flavonoids, each with a double hydroxyl on B-ring) antioxidants. We then determined the activity and synergistic effects of individual antioxidants and a mixture of the eight strongest antioxidants using DPPH-guided HPLC. Our findings revealed that (1) flavonoid glucuronide may have a more negative effect on antioxidant activity than glucoside, and (2) other compounds in the mixture may exert a negative synergistic effect on antioxidant activity of the four flavonoids with B-ring double hydroxyls but not the four caffeoylquinic acids. In conclusion, the eight phenolics with the strongest antioxidant ability reliably represented the bioactivity of the five extracts examined in this study. Moreover, the Morus alba hybrid had more phenolic biosynthesis machinery than its cross-parent M. alba, whereas the Broussonetia papyrifera hybrid had significantly less phenolic machinery than B. papyrifera. This difference is probably the main reason for livestock preference for the hybrid of B. papyrifera over B. papyrifera in feed.

discovered various phenolic acids and flavonoids in extracts from M. alba leaves, including several caffeoylquinic acids, kaempferol, rutin, and quercetin-3-O-glucoside 17 . Phenolic extracts from M. alba leaves have been reported to prevent chronic diseases through inhibition of nuclear factor-kB (NF-kB)-mediated inflammatory responses; these pharmaceutical effects may be linked to the phenolic composition of these extracts 18 . For example, rutin has demonstrated cytoprotective effects against H 2 O 2 -induced oxidative cell destruction, as well as antioxidant activity and anti-inflammatory effects 13 . The high economic and medicinal value of M. alba and B. papyrifera has led to the development of hybrids of M. alba and B. papyrifera; these hybrids have short growth cycles, strong adaptability, and high protein content, making them suitable for large-scale cultivation and can be used as candidates for developing functional feeds.
Conventional isolation of bioactive compounds is usually performed using high-speed counter-current chromatography and bioassay-guided fractionation 19 . However, these strategies are considered time-consuming, cumbersome, inefficient, and wasteful of organic solvents, while also having low recovery rates and high cost. The antioxidant activity of bioactive compounds is generally evaluated using spectrophotometry methods such as 2,2-diphenyl-1-picrylhydrazyl (DPPH), 2,2′-azino-bis (3-ethylbenzothiazoline-6-sulphonic acid) (ABTS), and ferric reducing antioxidant power (FRAP) assays; however, these approaches are limited to assessing the total antioxidant capacity of the extract and the total antioxidant capacity of the extract is compounded by the antioxidant capacity of individual antioxidants. The ability of each antioxidant to terminate radical chain processes cannot be evaluated directly 20 . Therefore, it is of great importance to develop a rapid and accurate technique to determine the specific ability of individual active compounds in the natural extracts of complexes. High-performance liquid chromatography (HPLC) analysis combined with DPPH and/or ABTS assays are currently implemented for rapid screening of active compounds in extracts 21,22 .
In DPPH/ABTS-guided HPLC analysis, after a specific antioxidant in the extract reacts with DPPH• or ABTS• + , one or more hydrogen atoms from the former are transferred to the latter, leading to a decrease in, or extinction of, the antioxidant peak compared to that in the HPLC profile of the unreacted extract 21 . In total, 127 compounds have been identified from the so-called 'Yangxinshi Tablet' and 34 antioxidant compounds were rapidly screened using an online DPPH-guided HPLC assay 22 . In another study, an offline DPPH-guided HPLC assay was applied to a Brazilian green propolis extract to screen nine DPPH• scavengers from among 14 characteristic peaks in the HPLC profiles 23 . A combination of DPPH-guided HPLC, diode-array detection, and time of flight mass spectrometry approaches (HPLC-DAD-TOF-MS) was initially used to screen and identify three antioxidants from peanut shells 24 . However, to our knowledge, offline methods has not been used to evaluate the synergistic or antagonistic effects of multiple antioxidants in complex extract mixtures.
The objective of this study was to determine and compare phenolic compositions (i.e., total phenols and flavonoids) in leaf extracts of five tree species in the two genera of the family Moraceae, and to determine their antioxidant abilities via DPPH, ABTS, oxygen radical absorbance capacity (ORAC), and cellular antioxidant activity (CAA) assays. Of the 23 phenolics identified by ultra-performance liquid chromatography quadrupole time of flight tandem mass spectrometry (UPLC-QTOF-MS/MS), 15 were screened by our modified DPPH/ ABTS-guided HPLC assays. Structure-activity relationships and possible synergistic effects of the eight strongest phenolic antioxidants were analyzed, and the antioxidant capacity of the five leaf extracts was comprehensively evaluated. The present work is expected to provide references for the chemical composition and biological activity of M. alba and B. papyrifera hybrids as functional feeds.

Results and Discussion
Total phenol and flavonoid characterization and evaluation of antioxidant activity in five extracts. To 1A). The highest phenol content in HMA was 2.73-fold that of the lowest phenol content in HBP, and all pairs of adjacent extracts were significantly different (P < 0.05). In contrast, there was a 1.50-fold difference between the highest and lowest total flavonoid content, as determined by the aluminum chloride colorimetric assay (Fig. 1B), in the following order (mg rutin/g d.w. extract): BK (313.77) > HMA (269.43) > MA (265.61) > HBP (230.78) > BP (209.62). There was no significant difference in flavonoid content between HMA and MA (P > 0.05).
We then conducted DPPH, ABTS, ORAC, and CAA assays to evaluate antioxidant activity in the five extracts. DPPH• and ABTS• + scavenging capacities were in the order BK > HMA > MA > BP > HBP (Fig. 1C,D), with significant differences between each pair of extracts. The ABTS assay yielded larger values than the DPPH assay, in all extracts except BK. These patterns were comparable to those observed for total phenol content (Fig. 1A), implying that total phenols were the main antioxidants in these two assays. For BK, however, flavonoid content was the main antioxidant in DPPH and ABTS assays. Similar orders of ORAC (BK > HMA > HBP > BP > MA; all pairs significantly different, P < 0.05) and CAA values (BK > HMA > BP > HBP > MA; all pairs significantly different except MA and HBP) were observed (Fig. 1E,F). These similarities suggested that similar mechanisms underlie the antioxidant activity examined by these two assays; the marked differences in ranking order between these assay results and those of the DPPH and ABTS assays may indicate different mechanisms between these groups of assays. Finally, the ranking order for total flavonoid content ( Fig. 1B) was similar between the two groups of assays, suggesting that total flavonoids in the five extracts may be the main contributors to antioxidant activity in these two assays.
Kim et al. 25 determined the total phenol content of 10 extracts from M. alba leaves and found that the highest content was 55.40 mg GAE/g extract, similar to our result for MA (52.71 mg GAE/g), but much lower than that for HMA (71.62 mg GAE/g). Mahmoud et al. 26 reported total phenol and total flavonoid content in M. alba leaves (67.55 mg GAE/g and 39.24 mg rutin/g, respectively), both substantially less than that of HMA in the present study (71.62 mg GAE/g for phenols and 269.43 mg rutin/g for flavonoids) and BK (63.68 for phenols and 313.77 for flavonoids); our results were 6-fold higher. This difference may be attributable to different extraction schemes; we used petroleum ether as a degreasing agent following extraction with 70% ethanol, rather than using only 90% ethanol. Lee et al. 5 investigated total phenol content (22.00 mg GAE/g) and total flavonoid content (24.00 mg rutin/g) in B. kazinoki leaf extracts with 70% ethanol; their contents were much lower than ours for BK.
Wang et al. 27 determined antioxidant activity in ethanolic extracts from leaves, stems, and fruits of M. alba, and found that leaves had the strongest free radical scavenging ability. Sánchez-Salcedo et al. 17 examined antioxidant capacity in leaves from eight M. alba clones using DPPH and ABTS assays. In the current study, both HMA (351.31 for DPPH and 370.55 for ABTS) and BK (37,3.85 for DPPH and 388.71 for ABTS) exhibited stronger antioxidant capacity than the strongest detected among the eight M. alba clones. This difference may be attributed to differences in the methods and solvents used during extraction.
Radojković et al. 28 reported ORAC values for M. nigra leaves (743.10 µmol Trolox/g dry plant material) that were dramatically lower than those of HMA (1195.39 µmol Trolox/g) and BK (2305.02 µmol Trolox/g) in the present study, because our results are expressed as ratios of extract d.w. rather than of plant material. To our knowledge, this is the first study to determine total phenols and total flavonoids in leaves of HMA and HBP, as well as their antioxidant activity, using DPPH, ABTS, ORAC, and CAA assays, and the first to determine antioxidant activity in BK and BP via ORAC and CAA assays, and in MA via a CAA assay.  Table 1. According to Fu et al. 29 , compound 2 was tentatively assigned as gentisoyl hexoside; to the best of our knowledge, this is the first report of gentisoyl hexoside detection in Broussonetia species and HMA. Following comparison of ultraviolet (UV) and MS data, we unambiguously identified compound 1 as quinic acid, which is an essential component of various feruloyl-, coumaroyl-, and caffeoylquinic acid derivatives 30 . According to a previous report 31 , the base peak of 3-feruloylquinic acid and 5-feruloylquinic acid were 191, while the base peak of compound 10 was 173, and we identified compound 10 as 4-feruloylquinic acid. To our knowledge, this is the first study to detect 4-feruloylquinic acid in Broussonetia. MS data for compounds 4, 7, and 8 indicated the presence of p-coumaroylquinic acid isomers; following a comparison of column retention behavior and other data from previous reports 31, 32 , we identified these compounds as 3-, 4-, and 5-p-coumaroylquinic acid, respectively. These three p-coumaroylquinic acids in BK Mass data for compounds 3, 5, and 6 indicated the presence of monocaffeoylquinic acid isomers; following comparison of column retention behavior and MS data between the authentic standards and those previously reported 33 , these compounds were identified as 5-, 3-, and 4-caffeoylquinic acid, respectively (Table 1). These monocaffeoylquinic acids were previously identified in M. alba leaves 11,17 , but not in BK leaves; 5-and 4-caffeoylquinic acids were previously reported in BP leaves. Compound    (Table 1); this compound was previously identified in M. alba leaves 11 . Table 1, eight flavonoids (compounds 9, 11, 12, 13, 14, 17, 20, and 21) were identified from Broussonetia; the other five flavonoids (compounds 15, 16, 18, 19, and 23) were identified from Morus. Interestingly, these two genera belonging to the same family contained flavonoids with completely different structures.

Identification of the 13 flavonoids. As shown in
Based on reports by Dugo et al. 11 and Benayad et al. 34 , and the MS data for their standards, compounds 9 and 15-17 herein were determined as vicenin-2, rutin, isoquercitrin, and luteolin-7-O-glucuronide, respectively. Compound 9 has not previously been reported in HBP; in the present study, it was identified only in HBP. Compound 17 was detected for the first time in BK, BP, and HBP leaves.
Following comparison of typical MS 2 fragments in the standards with those previously reported 7,35 , compounds 11-14 and 18-21 were identified as isoorientin, orientin, vitexin, isovitexin (vitexin and isovitexin are a pair of isomers), kaempferol-3-O-rutinoside, kaempferol-3-O-glucoside, apigenin 7-glucoside, and apigenin-7-O-glucuronide, respectively. Compounds 12-14 have been reported in BP leaves 7,35 ; however, to our knowledge, this is the first study to report compounds 11-14 in BK and compounds 13 and 14 in HBP. Compounds 18 and 19 have been identified in M. alba leaves but are reported in HMA for the first time 11 . Compound 20 has been reported in BP leaves 35 , but not in BK and HBP, and compound 21 is reported in both for the first time.
Finally, compound 23 showed an [M-H] − ion at m/z 489; its MS 2 yielded a fragment at m/z 285 due to the loss of a sugar and an acetyl ( Table 1). Following comparison with data from the MassBank MS database, we tentatively identified it as kaempferol-3-O-6″-O-acetyl-β-D-glucopyranoside, which has not been identified previously in M. alba leaves.

Screening of active antioxidants by DPPH/ABTS-guided HPLC analysis. The above results indicate
strong antioxidant activity among extracts from the five Moraceae species or hybrids, and 23 phenolic substances were identified via UPLC-QTOF-MS/MS. We next used DPPH/ABTS-guided HPLC assays, a rapid method for screening active antioxidant compounds from complex mixture extracts 36 , to screen effective antioxidants and evaluate the contribution of individual compounds to total antioxidant activity within a complex mixture extract 21 . When an antioxidant compound reacts with a radical such as DPPH• or ABTS• + , the result is a redox reaction; i.e., the molecular structure of the antioxidant changes 37 , with the peak area (PA) becoming significantly smaller in the chromatogram 38 . Among the antioxidant compounds in the mixture extract, a greater change in PA due to reaction with the free radical indicates a greater contribution of the compound to the antioxidant activity of the extract.

DPPH-guided HPLC analysis.
To visually evaluate the free radical scavenging ability of an antioxidant, we modified the DPPH-guided HPLC assay on the basis of a previous study 39 ; we used three different concentrations of DPPH•, and the free radical scavenging ability of each antioxidant was determined based on the extent to which its PA decreased as the DPPH• concentration increased. Figure 2A-E shows the HPLC profiles of HMA, BK, MA, BP, and HBP leaf extracts before (upright) and after (reversed) reaction with 1.0, 5.0, and 10 mM DPPH•. FRSRs of individual antioxidants in an extract after reaction with DPPH• (Table 2) were calculated as described in materials and methods.
For the three tested DPPH• levels, FRSRs of compounds 3, 5, 6, and 18 from HMA, and compounds 3, 5, 6, and 14 from BK, were significantly enhanced in a dose-dependent manner; however, no significant differences were observed in the FRSRs of compound 21 from BK, compounds 3, 13, 14, and 21 from BP, and compounds 13, 14, 17, and 21 from HBP, among which compound 21 from BK and compounds 17 and 21 from HBP had the highest FRSRs (69, 100, and ~65%, respectively) at the lowest DPPH• level of 1 mM ( Table 2). FRSRs of compounds 15, 16, 19, and 22 from HMA, compounds 12 and 17 from BK, compounds 3 and 5 from MA, and compounds 6, 12, and 17 from BP had significantly enhanced FRSRs at only the two lowest DPPH• levels, among which FRSRs of compound 16 from HMA, compounds 12 and 17 from BK, and compounds 12 and 17 from BP approached or attained 100% at the medium DPPH• level of 5 mM. Finally, compound 6 from MA and compound 9 from HBP showed significantly different FRSRs only at DPPH• levels between 1 and 10 mM, and compound 20 from HBP did so only between 5 and 10 mM. The compounds were ordered as follows in terms of FRSRs (highest to lowest): 17, 12, 16, 15, 22, 3, 5, 6, 19, 14, 13, 18, 21, 20, and 9. Several compounds identified from multiple extracts showed identical FRSRs including compounds 3, 5, and 6 from HMA, BK, MA, and BP (compound 5 was not detected from BP), compounds 14, 17, and 21 from BK, BP, and HBP, and compound 13 from BP and HBP, confirming that antioxidant activity was relatively stable in these compounds in mixed extracts from the five different tree species. In contrast, compounds 12 and 17 from BK and BP showed identical FRSRs at 5 and 10 mM (99-100%), but significantly different FRSRs at the lowest DPPH• level of 1 mM (72-73% in BK and 98-99% in BP), indicating a synergistic effect among phenolics. www.nature.com/scientificreports www.nature.com/scientificreports/ ABTS-guided HPLC analysis. The ABTS-guided HPLC assay is commonly used to screen bioactive antioxidants from complex mixed extracts. Chromatograms of the five extracts before and after reaction with 7 mM ABTS• + are shown in Fig. 2F-J; most peaks significantly decreased in size, or disappeared, after the reaction. As shown in Table 2, FRSRs for spiking ABTS• + were much lower than those for DPPH• scavenging, and clearly reflected differences in analytes among the five extracts. The compounds were ordered as follows in terms of FRSRs (highest to lowest): 12, 6, 17, 16, 3, 13, 14, 22, 5, 21, 15, 20, 9, 19, and 18; this pattern was similar to that observed for the DPPH• peaks. www.nature.com/scientificreports www.nature.com/scientificreports/ Compounds identified from multiple extracts exhibited different ABTS• + scavenging abilities: FRSRs of compounds 3, 5, and 6 from HMA and BK were similar, and were much higher and lower than those from MA and BP, respectively; compound 5 was not detected from BP. Compounds 14, 17, and 21 were identified from BK, BP, and HBP, with significantly higher FRSRs in BP than in the other two extracts. Finally, the FRSRs of compounds 12 and 13 from BP were much higher than those from BK and HBP, respectively. Together, these differences indicate synergistic effects among phenolic compounds.
Antioxidant ability and structure-activity relationships among the 15 phenolics. DPPH• and ABTS• + antioxidant ability. To evaluate the reliability of the antioxidant activity results for each phenolic compound in the extract mixture obtained using DPPH/ABTS-guided HPLC assays, we determined the antioxidant efficacy of each compound via traditional DPPH and ABTS spectrophotometric assays (Fig. 3).
Scientific RepoRtS | (2020) 10:4808 | https://doi.org/10.1038/s41598-020-61709-5 www.nature.com/scientificreports www.nature.com/scientificreports/ According to the results of the ABTS assay, the scavenging activity pattern of these 15 antioxidants (Fig. 3B) was roughly similar to that determined by the DPPH method (Fig. 3A). Interestingly, the order of the 11 flavonoids in terms of antioxidant activity was exactly the same with both assays. The highest ABTS value was exhibited by compound 22 (2. Overall, among the 15 phenolic compounds, 8 had clearly higher free radical scavenging ability than the remaining 7 (Fig. 3). The eight phenolic compounds with the strongest scavenging ability included all four caffeoylquinic acids: compounds 22, 6, 5, and 3. All flavonoids among these compounds had double hydroxyls on the B-ring: compounds 12, 17, 16, and 15. The remaining seven phenolic compounds with weak scavenging ability were flavonoids with a single hydroxyl on the B-ring; there were almost no significant differences in free radical scavenging ability among these compounds (P > 0.01).
Structure-activity relationships. The above results (Fig. 3) suggest that the position at which caffeoyl groups attach to quinic acid, and the presence of two hydroxyl groups instead of one on the flavonoid B-ring, greatly enhanced scavenging ability in both DPPH• and ABTS• + . In contrast, the attachment of glucosides and glucuronides to flavonoids decreased free radical scavenging ability to some extent.
Senthil Kumar and Kumaresan et al. 40 reported that the presence of an ortho-(3′,4′-) dihydroxy structure on the flavonoid B-ring strongly influences free radical scavenging ability, probably because the B-ring acts as a hydrogen atom donor and the relatively stable flavonoid radical is formed through electron delocalization 41 . As shown in Fig. 3, all four flavonoids with an ortho-dihydroxy structure on the B-ring showed significantly higher antioxidant activity than the remaining seven flavonoids, which did not possess this dihydroxy structure. The most prominent differences in antioxidant activity were observed, respectively, between compounds 12, 17, 16, and 15, which have the ortho-dihydroxy structure on the B-ring, and compounds 13, 20, 19, and 18, which have a single hydroxyl group on the B-ring (Fig. 4). These four pairs of flavonoids have the same structure except for this single difference in the B-ring; however, the resulting difference in antioxidant activity was at least 10-fold (Fig. 4), providing strong and direct evidence that the ortho-dihydroxy structure on the B-ring contributes greatly to flavonoid antioxidant capacity. Results are means ± SD of three independent experiments (n = 3) and are expressed as μmol Trolox eq/μmol standard phenolic compound. Different letters indicate significant differences (P < 0.01). The order of phenolics from left to right along the x-axis corresponds to DPPH values from highest to lowest.

Scientific RepoRtS |
(2020) 10:4808 | https://doi.org/10.1038/s41598-020-61709-5 www.nature.com/scientificreports www.nature.com/scientificreports/ Previous studies have shown that glycosylation of the hydroxyl groups in flavonoids weakens their antioxidant activity, possibly due to the destruction of ortho-hydroxyl structures and/or a reduction in the number of hydroxyl groups 42,43 . The attachment of sugars may prevent the entry of free radical scavengers into the radical center of ABTS• + and DPPH•; thus, sugar moieties may negatively affect antioxidant capacity in various flavonoids 44 . Our comparison of antioxidant ability and structure between compounds 12 and 16 (Figs. 3 and 4) indicated that the sugar moiety negatively affected flavonoid antioxidant capacity to a greater extent on the C-ring than on the A-ring. Our comparison between compounds 16 and 15 (Figs. 3 and 4) clearly showed that the presence of two sugar moieties attached to the C-ring had a greater negative impact than a single sugar moiety. Similarly, comparing compounds 20 and 21 (Figs. 3 and 4) revealed that compound 21 possesses one more hydroxyl group on the A-ring, which had a positive effect on its antioxidant ability, which was similar to that of compound 20. Thus, the glucuronide moiety attached to 7-C on the A-ring weakened the antioxidant ability to a much greater extent than the attachment of a glucoside moiety at exactly the same position. To our knowledge, this represents the first evidence of the effects of glucuronide and glucoside moieties on antioxidant activity. We therefore hypothesize that the glucuronide moiety has a greater effect than the glucoside group in reducing flavonoid antioxidant activity. showed the weakest antioxidant activity; differences in antioxidant activity among vitexin, kaempferol-3-Oglucoside, kaempferol-3-O-rutinoside, isovitexin, and vicenin-2 were caused by the position and number of glycosyl groups on the A-and C-rings. The last row contains four caffeoylquinic acids; the first of these is dicaffeoylquinic acid, which showed the strongest antioxidant activity, followed by three monocaffeoylquinic acids, in which antioxidant activity was affected by the position of the caffeoyl group on the quinic acid. The number above the compound name represents the compound number consistenting with that in Table 1, and numbers in parentheses under each compound represents DPPH/ABTS antioxidant capacity.

Synergistic effects and antioxidant activity of the five extracts. Synergistic effects of eight
high-antioxidant activity phenolics. To verify whether synergistic effects are present in the eight phenolics exhibiting the strongest antioxidant activity, we performed a DPPH-guided HPLC assay to evaluate the free radical scavenging ability of these compounds. As shown in Fig. 5A, when each compound reacted with DPPH•, the four flavonoids showed significantly higher FRSRs than the four caffeoylquinic acids, with no significant differences among three of the four flavonoids (P > 0.01), except compound 17, which showed significant differences with compounds 12 and 16; moreover, there were no significant differences among three of the four caffeoylquinic acids, except compound 5, which exhibited significant differences to the other three. However, in a mixture of the compounds, FRSRs of the four flavonoids were significantly lower than those of the four phenolic acids, and there were no significant differences (P > 0.01) among three of the four flavonoids, except compound 17, or among the four caffeoylquinic acids. The FRSRs of the four flavonoids in the compound mixture were approximately 40% lower (with the exception of ~20% for compound 17) than those of the individual compounds. However,  Table S1,  Table 2, and Fig. 5A. Results are means ± SD of three independent experiments (n = 3) and are expressed as percentages. Different letters indicate significant differences (P < 0.01).
the FRSRs of the four individual caffeoylquinic acids were similar to those measured in the compound mixture. Therefore, we suspect that other compounds in the mixture may have exerted a negative synergistic or antagonistic effect on the antioxidant ability of the four flavonoids, but exerted no effect on the caffeoylquinic acids.
Comprehensive evaluation of the antioxidant capacity of the five extracts. Figure 5B shows the comprehensive free radical scavenging ability of the five extracts, in terms of the eight strongest phenolics in each extract. On average, free radical scavenging ability was 8-fold higher in BK and HMA than in BP, MA, and HBP; the overall ranking in terms of free radical scavenging ability was generally consistent with that of total phenols, total flavonoids, and antioxidant activity among the five extracts (see table in Fig. 5B), suggesting that the DPPH/ ABTS-guided HPLC assays reliably screened the eight strongest bioactive phenolics. These results also indicate that hybridization allowed HMA to gain considerably more phenolic biosynthesis machinery compared to its cross-parent MA, whereas HBP lost a significant amount of phenolic biosynthesis machinery compared to BP. This process is likely the main reason for the greater palatability of HBP compared to BP, and explains the preference for leaves of this species among pigs, cattle, and sheep.

Methods
Chemicals and leaf extract preparation. Folin-Ciocalteu reagent, Trolox, DPPH, ABTS, ultra-pure water, acetonitrile, formic acid for HPLC, and other chemicals were obtained following Yang et al. 20  Leaves were washed with purified water and placed in a ventilation oven at 105 °C for 15 min for de-enzyming, and then dried at 60 °C for 24 h in the same oven. Dried leaves were ground, and the powder was sifted through a 1-mm mesh and stored at −20 °C for further extraction. To prepare each extract, 150 mL 70% ethanol was added to 10 g dry leaf powder, and the mixture was sonicated for 30 min at room temperature (power, 500 W; frequency, 40 kHz; KQ-300DE NC ultrasonic cleaner; Kunshan Ultrasonic Instrument Co., Ltd., Jiangsu, China). The sonicated mixture was then filtered using a 0.22-µm filter, and the supernatant was collected. This extraction procedure was repeated twice; the three resulting supernatants were combined and evaporated by rotation at 45 °C until one third of the volume remained. The remaining solution was degreased with petroleum ether 35 , and the lower layer was collected. The collected solutions were dried in a ventilation oven at 60 °C, and the five extracts were collected and stored at −20 °C for further experiments.
Determination of total phenols, total flavonoids, and antioxidant activity. Total phenols and total flavonoids in the five extracts were measured following the Folin-Ciocalteu and aluminum chloride colorimetric methods 20 . DPPH• and ABTS• + scavenging activities were assessed using our modified method 46 . An ORAC assay was performed by modifying a method described previously 47 . A CAA assay was performed as described previously 20 , with minor modifications. All assays were performed in triplicate.
HPLC and UPLC-QTOF-MS/MS analyses of phenolic composition. HPLC analyses were performed using a Shimadzu (Kyoto, Japan) HPLC system as described previously with slight modifications 17,48 . Briefly, an extract or the mixture of 15 standards was dissolved in 70% ethanol at 0.005 and 0.5 mg/mL, respectively, and then filtered through a 0.22-µm filter before injection. A reversed-phase column (Diamonsil C 18 5 µm 250 × 4.6 mm i.d.; Dikma, Beijing, China) was used for separation; the column temperature was set at 30 °C. Two solvents were applied for elution: water containing 0.4% (v/v) formic acid (A) and acetonitrile (B). The gradient started with 8% solvent B, reaching 15% at 25 min, 22% at 55 min, and 40% at 60 min. The flow rate was set at 1.0 mL/min, and the injection volume was 20 µL. The detection wavelength was set at 320 nm to monitor all polyphenols simultaneously. Before each run, the column was equilibrated for 15 min under the initial conditions.
The UPLC-QTOF-MS/MS system comprised an Acquity UPLC system (Waters, Milford, MA, USA) and a QTOF-MS mass spectrometer (Xevo G2-XS; Waters). The C 18 used for separation was described above, with a column temperature of 30 °C and a mass range of 100-1,000 m/z. MS experiments were performed in both positive and negative ionization modes under the following conditions: nitrogen drying gas flow, 10.0 L/min; nebulizer pressure, 45 psi; gas drying temperature, 370 °C; capillary and fragment voltage, 2.500 kV; and MS/MS collision energy, 20 V. DPPH/ABTS-guided HPLC assays for active antioxidant screening. DPPH/ABTS-guided HPLC assays were performed as described previously with slight modifications 24,36,39 . Briefly, each of the five extracts was mixed with a different concentration of DPPH• (1, 5, or 10 mM) or ABTS• + (7 mM) solution at a ratio of 1:1 (v:v) to react at room temperature in the dark for 30 min. The extract solution (5 mg/mL) without DPPH• or ABTS• + was used as a control; Vc was used as a radical scavenger reference compound 49 . All evaluated solutions were filtered through a 0.22-μm microporous organic membrane filter prior to HPLC analysis.
The scavenging capacity of DPPH• and ABTS• + for each compound in each mixture extract was expressed as the free radical scavenging rate (FRSR), calculated as follows: FRSR (%) = (PA 0 -PA 1 )/PA 0 × 100 (Equation 1), where PA 0 is the area of an individual peak in the profile prior to reaction of the extract with DPPH• or ABTS• + , and PA 1 is the area after the reaction.
The peak area (PA), representing the contribution rate of each of the eight strongest phenolics to an extract, was calculated as follows: PA = PA 0 × FRSR, where PA 0 (Equation 2) is the area of an individual peak in the profile of the extract prior to reaction with DPPH•.

Conclusion
We investigated the phenolic composition and antioxidant activities of B. papyrifera, B. kazinoki, M. alba, and commercialized hybrid of B. papyrifera and that of M. alba. Among the five leaves extracts, the hybrid of M. alba had the highest contents of total phenols and flavonoids, and its four tested antioxidant activities were also the strongest. On the other hand, the hybrid of B. papyrifera exhibited the lowest contents of total phenols and flavonoids as well as the four antioxidant activities. Subsequently, through UPLC-QTOF-MS/MS analysis of the five extracts, 23 compounds were identified, of which nine and two compounds were detected for the first time in Broussonetia and Morus leaves, respectively. Then, 15 phenolics, including seven weak antioxidants and eight strong antioxidants, were screened out by the DPPH/ABTS-guided HPLC method, with compounds 22, 16, 15 being found only in the hybrid of M. alba. Combined with structure-activity relationships analysis, we speculated that this may be the major reason for the stronger antioxidant activity of the hybrid of M. alba. In order to explore the synergistic or antagonistic effects between the compounds, the interactions of eight strong antioxidants to scavenge DPPH• were further analysed by DPPH-guided HPLC method. Interestingly, the antioxidant activity of flavonoids was inhibited in the mixture, while phenolic acid compounds showed no significant changes. We hypothesised that the stronger antioxidant activity of the hybrid of M. alba was mainly contributed by phenolic acid compounds, while phenolic acids have not been identified in the hybrid of B. papyrifera, and thus the antioxidant activities were relatively weaker. Nevertheless, the palatability of the hybrids of B. papyrifera was better, which may be related to the synthesis mechanism of phenolic compounds. In conclusion, our results contribute greatly to a comprehensive understanding of the potential of the hybrids of M. alba and B. papyrifera as a source of natural phenolics and antioxidants. This study could also provide useful phytochemical information for them as raw materials for developing functional feeds.