POLYPHENOL CONTENT AND ANTIOXIDANT POTENTIAL OF Allophylus edulis (A. ST.-HIL. ET AL.) HIERON. EX NIEDERL. AND Cupania vernalis CAMBESS. (SAPINDACEAE)

Allophylus edulis and Cupania vernalis (Sapindaceae) are Brazilian native trees used as medicinal plants for the treatment of respiratory, digestive, circulatory, and skin diseases. Ubiquitously distributed in the Brazilian territory, these species are indicated for mixed plantations aimed at the recovery of degraded ecosystems. In this study, the total phenolic content (TPC) and total fl avonoid content (TFC), and the antioxidant activity of extracts and fractions obtained from A. edulis and C. vernalis leaves were assessed. The TPC and TFC was determined spectrophotometrically. Antioxidant activity was evaluated through radical scavenging activity of 2,2-diphenyl-1-picrylhydrazyl (DPPH). The extracts were obtained by two methods: maceration (method 1) and Soxhlet (method 2). Solvents of increasing polarity (hexane, dichloromethane, ethyl acetate, and n-butanol) were used to obtained the fractions. The results showed that the ethyl acetate fraction from A. edulis, obtained from the maceration method, had the highest TPC (442.0 ± 18.2 mg GAE g) and TFC (58.1 ± 0.4 mg RUE g), and antioxidant activity (EC50 = 43.6 ± 2.6 μg mL). By C. vernalis, superior results were obtained with the n-butanol fraction (TPC = 126.1 ± 5.8 mg GAE g, TFC = 37.7 ± 0.6 mg RUE g). The highest antioxidant potential was found in the crude hydroalcoholic extract (EC50 = 816.1 ± 50.9 μg mL) and butanol fraction (1,156.4 ± 3.8 μg mL). The results of this study show that the fractions obtained by maceration and liquid-liquid partition with more polar solvents (ethyl acetate and n-butanol) are the richest in TPC and TFC, and presented the greater antioxidant activity. Comparing the two plants, A. edulis showed the best results, with a high content of TPC, TFC, and antioxidant potential, and therefore may be used to treat diseases related to oxidative stress.


1.INTRODUCTION
Among several Brazilian native plants of medicinal importance, Allophylus edulis and Cupania vernalis (Sapindaceae) have shown potential use in the treatment of respiratory, digestive, circulatory, and skin diseases (Rovedder et al., 2016). Commonly known as "chal-chal, vacum, fruto-de-pombo", among others, A. edulis is a tree that can reach up to 20 m in height, with a fenestrate trunk of 15 to 30 cm in diameter, and spiral leaves without stipules composed of three leafl ets with serrate margins. Flowering occurs from September to November and fruiting from December to March. The fl owers are whitish and arranged in short axillary racemes. The species is widely known for its intense fl owering and especially for its red fruits with sweet pulp. It occurs from north to south of Brazil -Amazon region, and states of Ceará, Bahia, Mato Grosso, Minas Gerais, Rio de Janeiro and Rio Grande do Sul (Backes and Irgang, 2004;Lorenzi, 2016). It is the species with the largest geographical distribution in Brazil and the main representative of the genus comprising tropical seasonal forest trees, and it also occurs in Argentina, Bolivia, Paraguay, and Uruguay (Coelho, 2014). Cupania vernalis is known as "Camboatã" or "Camboatá-vermelho", this is frequently seen in almost every forest formation and found in the states of Minas Gerais, Mato Grosso, and São Paulo, extending as far as Rio Grande do Sul. It may reach up to 22 m in height, has alternate spiral leaves and yellowish fl owers. This tree fl ourishes from March to May, and its fruit ripens from late September to November. Both A. edulis and C. vernalis are indicated for mixed plantations intended for the recovery of degraded ecosystems, and their fruits are much appreciated and eaten by birds and other animals, which are responsible for seed dispersal (Lima Jr. et al., 2006;Lima Junior et al., 2005;Lorenzi, 2016).
In the past years, free radicals and other oxidants have been held largely responsible for aging and associated degenerative diseases. In living beings, free radical formation is controlled by antioxidant compounds, which may be endogenous in origin or derived from the diet (Suleman et al., 2019). Among naturally occurring antioxidants, phenolic compounds have received much attention. The content of polyphenols in plant extracts has been used as an important antioxidant capacity parameter (Rojas and Buitrago, 2019). Spectrophotometric methods for the analysis of phytochemical content in plant species are often used because of their simplicity and reproducibility. The Folin-Ciocalteu assay is one of the most widely used for the determination of total phenolic content (TPC). Spectrophotometric analyses of total fl avonoid content (TFC) are based on aluminum chloride complexation. For evaluation of in vitro antioxidant activity of plant extracts, the 2,2-diphenyl-1-picrylhydrazyl (DPPH) free radical assay is a great alternative. DPPH is an unstable organic nitrogen radical with a purple color. In this method, antioxidants reduce DPPH radicals by donating hydrogen, converting them to DPPH-H. The color changes from purple to yellow, with a decrease in absorbance that allows the determination of antioxidant capacity (Huyut et al., 2017).
In this work, total phenolic and fl avonoid contents and antioxidant activity of Allophylus edulis and Cupania vernalis leaves were investigated. By using two extraction methods and solvents with diff erent polarities, the purpose of this study was also determined the best method and solvent for extraction of phenols, fl avonoids and antioxidant activity in the two plants.

Plant material
Allophylus edulis leaves were collected in July 2017, in Casca, state of Rio Grande do Sul, Brazil (28º33'20.9"S, 51º57'44.2"W), and Cupania vernalis leaves were collected in May 2016, in Rio da Várzea, municipality of Passo Fundo, state of Rio Grande do Sul,Brazil (28º13'24.4"S,52º29'37.4"W). The plants were identifi ed and deposited in the herbarium of the Universidade de Passo Fundo (voucher numbers RSPF 14385 and RSPF 14166, respectively). The plant material was air-dried at 35 °C for about 48 h and crushed into a fi ne powder.

Extraction
The extracts were obtained by two diff erent methods: maceration with subsequent liquid-liquid partition at room temperature (method 1) and exhaustive extraction with a Soxhlet extractor, using a heat source (method 2).
Method 1: The crushed leaves were macerated (ratio of plant material to solvent of 1:10) for eight days with methanol:water (1:1) to obtain the crude hydroalcoholic extract. In a rotary evaporator, the extract was concentrated until methanol elimination and successive liquid-liquid partitions were carried out with solvents of increasing polarity: hexane, dichloromethane, ethyl acetate, and n-butanol. The obtained fractions were concentrated and lyophilized. A portion of the crude hydroalcoholic extract was also dried and lyophilized. The samples obtained were thus coded: ECH (crude hydroalcoholic extract); FH1 (hexane fraction); FEA1 (ethyl acetate fraction); and FB1 (n-butanol fraction).
Method 2: The dried and ground plant material from both plants was packed in a Soxhlet extractor and subjected to exhaustive extraction with hexane, dichloromethane, ethyl acetate, and n-butanol, followed by concentration on a rotary evaporator. Each extraction lasted approximately 10 h. The following samples were then obtained: FH2 (hexane fraction); FD2 (dichloromethane fraction); FEA2 (ethyl acetate fraction); and FB2 (n-butanol fraction).

Phytochemical screening
General characterization reactions of fl avonoids and tannins (polyphenolic compounds) were performed in the samples obtained from the two extraction methods. The presence of tannins was investigated by reaction with 1 % (w/v) ferric chloride solution in ethanol and with 10 % (w/v) aqueous lead acetate solution. The bluish-green color and the reddish-brown precipitate indicate positivity in the assays, respectively. For the detection of fl avonoids, the oxalo-boric reaction was used, which is performed with 3 % (w/v) boric acid in ethanol and 10 % (w/v) oxalic acid in ethanol. The observation of greenyellow fl uorescence at 365 nm is a positive reaction for fl avonoids (Harborne, 1980).

Total phenolic and fl avonoid contents
The TPC of crude extract and fractions was determined by the Folin-Ciocalteu assay (Sousa et al., 2007). The lyophilized samples were diluted in methanol at an accurately weighed fi nal concentration of about 200 µg mL -1 . An aliquot of 0.5 mL was transferred to a 10 mL volumetric fl ask, adding 0.5 mL of Folin-Ciocalteu reagent, 3 to 4 mL of distilled water, and 2 mL of 14 % aqueous sodium carbonate solution, completing the volume with distilled water. After incubation for 2 h at room temperature, absorbance was measured at 750 nm against a blank, prepared with all reagents, except the sample. The total phenolic content was expressed in milligram of gallic acid equivalents (GAE) per gram of dry extract and obtained from a calibration curve of standard gallic acid (5 to 150 µg mL -1 ). The assay was performed in triplicate.
A spectrophotometric assay based on aluminum complex formation (Schmidt and González-Ortega, 1993) was used for TFC analysis. The lyophilized samples were diluted in 70 % methanol at concentrations around 350 µg mL -1 (FBH, FEA1, and FB1), 500 µg mL -1 (FH1), and 1,500 µg mL -1 (FD1). Aliquots of 1 mL of samples were transferred to a 10-mL volumetric fl ask and mixed with 1 mL of 5 % aluminum trichloride (AlCl 3 ) in methanol and completed with 70 % methanol. A blank without AlCl 3 was prepared for all samples. After 10 min, absorbance was read at 425 nm. The TFC in the samples was determined based on the calibration curve of rutin, diluted in methanol 70 % (from 20 to 150 µg mL -1 ). The results were expressed in milligram of rutin equivalents (RUE) per gram of dry extracts. The assay was performed in triplicate.

Free-radical scavenging activity
Antioxidant activity was evaluated by the 2,2-diphenyl-1-picrylhydrazyl (DPPH) free radical assay, following the methodology described by Sousa et al. (2007), with modifi cations. A stock solution of DPPH was initially prepared in methanol at the concentration of 40 µg mL -1 (0.1 mM). Dilutions of 35,30,25,20,15,10,5, and 1 µg mL -1 were obtained from the solution, and absorbances were read at 516 nm, using methanol as blank. The calibration curve of DPPH was built using the absorbance results. The assay was carried out in triplicate.

Methanol
solutions containing diff erent concentrations of lyophilized samples of A. edulis and C. vernalis were tested using DPPH. Aliquots of 0.3 mL of the samples and 2.7 mL of the DPPH stock solution (0.1 mM) were used. The solutions obtained were protected from the light, and absorbances were read at 516 nm, 30 min after addition of DPPH. The mixture of methanol (2.7 mL) and of the highly concentrated methanol solution of each sample (0.3 mL) was used as blank. Methanol (0.3 mL) and DPPH 0.1 mM (2.7 mL) were mixed and used as control. The assays were carried out in triplicate.
The equation obtained from the DPPH calibration curve and the absorbance data read within 30 min were used to determine the concentration of DPPH after reaction with the samples, denoted by [DPPH] 30 . By also using the initial concentration of DPPH, denoted as [DPPH] o , the percentage of remaining DPPH was estimated by the following equation: Based on an exponential curve graph between the percentage of remaining DPPH and the concentration of the sample, the eff ective concentration (in µg mL -1 ), which is necessary for reduction of the initial concentration of DPPH by 50 % (EC50), within 30 min, was obtained.

Statistical analysis
The results for FD1, FEA1, and FB1 (method 1) were compared with those obtained for FD2, FEA2, and FB2 (method 2) by Student's t test. Student's t test was also used to compare the two plants in samples obtained by the same extraction procedure and solvent regarding TPC, TFC and antioxidant activity. The results for samples obtained from the same plant were compared by one-way analysis of variance (ANOVA), followed by Tukey's test. Pearson's correlation coeffi cient was used to check the correlation among: TPC and TFC; TPC and EC50; TFC and EC50. The signifi cance level was set at 5 %. Data were obtained in triplicate and the results were expressed as mean (n = 3) ± standard deviation (SD).

Phytochemical screening
General tests for the identifi cation of secondary metabolites revealed presence of fl avonoids and tannins in the crude extract and more polar fractions of both plants, obtained from extractions performed with ethyl acetate and n-butanol (ECH, FEA1, FEA2, FB1, and FB2 fractions). There was more intense positive reaction to fl avonoids in FB1 fraction of C. vernalis. This was the most polar fraction obtained, with extraction at room temperature. The nonpolar fractions, hexane (FH1, FH2) and dichloromethane (FD1, FD2), did not yield positive results in these tests (Table 1).

Total phenolic and fl avonoid contents
TPC in A. edulis samples obtained by maceration and later liquid-liquid partition (method 1) were compared with those of samples obtained by Soxhlet extraction (method 2). Method 1 proved to be more effi cient, with signifi cant diff erences among samples (Table 2).
Based on these results, we decided not to proceed with the tests in samples obtained by Soxhlet extraction (method 2) , but only in those obtained by maceration (method 1). From A. edulis samples obtained by method 1, the ethyl acetate fraction (FEA1) had the highest TPC (442.0 ± 18.2 mg GAE g -1 ), followed by the FB1 fraction (194.9 ± 1.3), the crude extract ECH (99.1 ± 4.4), and the FD1 fraction (47.1 ± 1.2). The hexane fraction (FH1) revealed undetectable phenolic content. Cupania vernalis had higher TPC in FB1 (126.1 ± 5.8 mg GAE g -1 ), followed by FEA1 (77.6 ± 2.6), and ECH (55.0 ± 2.3). Phenolic content was also undetectable in FH1 and FD1 for this plant. All samples of both plants showed signifi cant diff erences. TPC was also compared between the two plants in samples obtained with the same solvent. Allophylus edulis revealed a signifi cantly higher TPC in all tested samples (Figure 1).
vernalis leaves. Tabela 1 -Resultados da prospecção fi toquímica para identifi cação de fl avonoides e taninos no extrato bruto e frações das folhas de A. edulis e C. vernalis.  did not diff er signifi cantly, all the other C. vernalis samples revealed diff erences in the test results. TFC was also compared between the species and A. edulis showed signifi cant higher results than C. vernalis in all tested samples (Figure 2).
results. EC50 values were quite high for C. vernalis. ECH and FB1 also did not show any statistically signifi cant diff erences and had the highest antioxidant potential, with an EC50 of 816.1 ± 50.9 and 1,156.4 ± 3.8 µg mL -1 , respectively. Following, the results of FEA1 (1,232.3 ± 142.1) and FD1 (1,382.4 ± 285.7) did not show diff erences between them. There was signifi cant diff erence in FH1, with an EC50 of 2,325.8 ± 7.7 µg mL -1 (Figure 3).

Correlation between assays
A positive Pearson's correlation was observed between TPC and TFC, while negative correlations were found between TPC and EC50, TFC and EC50. A statistically signifi cant strong positive correlation was observed between TPC and TFC for C. vernalis (Person's value = 0.985; p = 0.01). The Pearson's correlation for this plant was of -0.585 (p = 0.300) between TPC and EC50, and of -0.500 (p = 0.391) between TFC and EC50. In A. edulis samples, Person's value was of 0.761 for TPC and TFC (p = 0.135), -0.755 for TPC and EC50 (p = 0.140), and -0.855 for TFC and EC50 (p = 0.065).

4.DISCUSSION
This work demonstrated that A. edulis and C. vernalis contain phenolic compounds in their leaves and that their extracts have antioxidant properties. Phenolic compounds extracted from plants have been intensively investigated due to their benefi ts for human health (Tungmunnithum et al., 2018). Several studies have showed that these compounds possess antioxidant activity (Aryal et al., 2019;Dong et al., 2019;Kakouri et al., 2019;Sarikurkcu et al., 2020). Therefore, it is important to fi nd suitable methods of extraction of these compounds. In the present study, it was observed that the extraction of phenolics by the Soxhlet method, which involves exposing the sample to intense heat, does not provide eff ective results. The samples obtained by maceration and liquid-liquid partition with ethyl acetate (in A. edulis) and n-butanol (in C. vernalis) are the richest in TPC and TFC, presenting greater antioxidant activity. Some studies evaluated the infl uence of extraction conditions and solvents on the phenolic content and antioxidant activity in plant extracts. Narimane et al. (2017) reported the impact of using various solvents in the extraction of potentially active compounds from Algerian propolis. The authors conclude that the use of ethyl acetate and n-butanol result in samples with the strongest antioxidant activity and the highest amount of TPC and TFC. Ismail et al. (2019) analyzed baobab (Adansonia digitata) fruit pulp, and concluded that, in this case, 80 % acetone is the best solvent. Sarikurkcu et al. (2020) worked with ethyl acetate, methanolic, and aqueous extracts from Micromeria nervosa aerial parts, and concluded that the diff erent extracts possess diff erent mechanisms responsible for the antioxidant activity. Array et al. (2018) worked with Curcuma longa and showed that the alcoholic extracts, obtained at room temperature, exhibited the highest phenolic content and antioxidant activities. Even with vernalis samples needed for the initial concentration of DPPH to be reduced by 50% (EC50)  variations in the employed methods of extraction, the studies indicate the effi cacy of extraction of phenolic compounds at room temperature and with more polar solvents. The results of the phytochemical screening in the present work also revealed the presence of phenolic compounds (fl avonoids and tannins) in the fractions obtained with more polar solvents (Table 1).
Phenolic compounds have been described in A. edulis extracts, with isolation of glycosyl fl avones and fl avonols, among other substances (Hoff mann-Bohm et al., 1992) (Arisawa et al., 1989). TPC was measured in ethanol crude extract (176 mg GAE g -1 ) and aqueous extract (90 mg GAE g -1 ) obtained from the leaves of A. edulis (Tirloni et al., 2015). In the present study, the results showed a concentration of 99.1 ± 4.4 mg GAE g -1 in ECH (hydroalcoholic crude extract). This extract was used for partitioning with solvents in increasing order of polarity, and fractions with a higher phenolic content were obtained (FEA1 = 442.0 ± 18.2; FB1 = 194.9 ± 1.3 mg GAE g -1 ) (Figure 1). By using quercetin as standard, Tirloni et al. (2015) obtained 20 mg g -1 of fl avonoids for the ethanol crude extract, and 10 mg g -1 for the aqueous extract of A. edulis leaves. The fl avonoid content obtained in the present study was much higher: 42.8 ± 0.0 mg RUE g -1 in the crude extract (ECH) and 58.1 ± 0.4 mg RUE g -1 in the ethyl acetate fraction (FEA1) ( Figure  2). Despite small variations in the technique used, it could be observed, once again, that the choice of the appropriate solvent yielded better results.
Allophylus edulis has also been investigated using the DPPH method. Umeo et al. (2011) found an EC50 of 46.7 µg mL -1 in alcoholic extracts of A. edulis fruits. Trevizan et al. (2016) obtained an EC50 of 82.9 µg mL -1 for essential oil of A. edulis, and 74.7 µg mL -1 for viridifl orol, the major constituent of this oil. Tirloni et al. (2015) found an EC50 of 17.7 and 45.8 µg mL -1 in ethanolic and aqueous crude extracts, respectively, from A. edulis leaves. Our fi ndings for FEA1, with an EC50 of 43.6 ± 2.6 µg mL-1, were similar to those described for this species (Figure 3).
In A. edulis, TPC, TFC, and antioxidant activity were signifi cantly higher in all tested samples when compared to C. vernalis. In the samples obtained from C. vernalis, FB1 showed a signifi cantly higher content of total phenols and fl avonoids, followed by ECH (Figure 1 and 2). The highest antioxidant activity in C. vernalis was detected in ECH and FB1, and the lowest one in FH1. The statistical analysis did not show signifi cant diff erence between ECH and FB1. Also, there was no signifi cant diff erence between FD1, FEA1, and FB1, but only between FH1 and the other samples. By looking at the result for FH1, the infl uence of polarity of the solvent used for the extraction of compounds of interest is quite clear. Based on the fact that the mechanism of DPPH reduction is correlated with the presence of hydroxyl groups in the antioxidant molecule, it may be inferred that higher antioxidant capacity is encountered in polar extracts owing to the presence of substances with an available hydroxyl group (Mensor et al., 2001). A nonpolar solvent was used to obtain FH1, which exhibited lower antioxidant capacity ( Figure  3).
The correlation between TPC and TFC, TPC and EC50, and TFC and EC50 was assessed. A signifi cant Person's correlation was obtained between TPC and TFC for C. vernalis. Despite the lack of statistical signifi cance, probably due to the small number of repetitions in each sample, the Pearson's value in all A. edulis samples is considerable. The results indicate the relationship between phenolic content and antioxidant potential of these plant extracts.

5.CONCLUSION
The results of this study show that extraction solvents of diff erent polarities vary signifi cantly in their extraction capacity and selectivity for phenolic and fl avonoid contents, as well as in antioxidant activity. The use of maceration with posterior liquidliquid partition with solvents of increasing polarity (a method of extraction without heat use) is essential for obtaining samples with higher content of these phytochemicals and, consequently, greater antioxidant capacity. The ethyl acetate and n-butanol fractions are the richest in TPC and TFC, and presented the greater antioxidant activity. The comparison between the results obtained with the two plant species showed the antioxidant potential of A. edulis, given its high content of total phenolics and fl avonoids. Therefore, these results suggest that A. edulis may be used to treat diseases related to oxidative stress.

AUTHOR CONTRIBUTIONS
Andréa M. Sobottka: conceived the ideas, carried out the research supervision, contributed to the study conception and design, contributed to the writing, and analyzed the data.
Elisandra Tessaro and Suelen M. da Silva: carried out the experimental work, contributed to the study conception and design, contributed to the writing, and analyzed the data.
Marina Pedron and Lara T. Seff rin: carried out the experimental work.