Effect of fruit ripening on bioactive compounds and antioxidant capacity of apple beverages

Apples (Malus domestica Borkh) has high phenolic compounds content (Khanizadeh et al., 2008; Goulas et al., 2014). On apple juice and cider, phenolics can contribute to the sensory characteristics, such as their color, astringent and bitter taste. Moreover, phenolic compounds have demonstrated potent antioxidant properties in various systems that exhibit biological activity due to their cytoprotective properties (Babbar et al., 2015; Yassin et al., 2017). These compounds are related with reduction of the risk of incidence of various diseases such as cancer, coronary and neurodegenerative diseases (Alzheimer’s and Parkinson’s) (Hyson, 2011).


Introduction
Apples (Malus domestica Borkh) has high phenolic compounds content (Khanizadeh et al., 2008;Goulas et al., 2014). On apple juice and cider, phenolics can contribute to the sensory characteristics, such as their color, astringent and bitter taste. Moreover, phenolic compounds have demonstrated potent antioxidant properties in various systems that exhibit biological activity due to their cytoprotective properties (Babbar et al., 2015;Yassin et al., 2017). These compounds are related with reduction of the risk of incidence of various diseases such as cancer, coronary and neurodegenerative diseases (Alzheimer's and Parkinson's) (Hyson, 2011).
On the processing of apple juice or apple must to obtain fermented beverages (cider) there is significant loss of phenolic compounds. On the extraction of juice by the traditional method using pressing, about 42 to 58% of phenolic compounds are retained in the apple pomace ( Van der Sluis et al., 2002). Among those phenolic compounds which are extracted to the juice, about 20 to 60% are oxidised by the polyphenol oxidase (Zardo et al., 2013). Thus, the processing of dessert apples results in a juice or cider with only 3 to 10% of the antioxidant activity of the intact fruit (Ćetković et al., 2008).
On several countries, fruits that are not considered to be commercially marketable, or fruits that do not have acceptable physical characteristics (appearance, size and shape) for fresh consumption, are used for processing into juice. Therefore, these fruits at different ripening stages are processed without prior classification. Onformation about the influence of the ripening stage on the concentration of phenols in dessert apples for beverage processing is essential in order to obtain a product with high levels of sensory and functional characteristics. Consequently, the aim of this study was to evaluate the influence of the cultivar and the ripening stage of the raw material on the phenolic content and antioxidant activity of apple juices and ciders.

Samples
Samples of the Fuji Suprema and Lis Gala apple varieties were harvested at the Experimental Station of the Agricultural and Rural Extension (EPAGRO) in Santa Catarina, Brazil (26° 46' 31" S, 51° 00' 54" D). The Gala variety was harvested from Boutin Agricola Company in Paraná, Brazil (25° 32' 08" S, 49° 53' 33" D). The varieties were chosen based on world apple production. Gala, Fuji and their clones are among the five most produced varieties in the world. The fruits were harvested (20 kg) at three ripening stages (unripe, ripe and senescent) from different cardinal points, and from the top and the bottom of six trees. The maturation index was determined by using the Starch-iodine test according Blanpied & Silsby (1992) using an eight point scale. The unripe fruits used in the experiments had an index of 1, while the ripe and senescent apples had indices of 4 -5 and 8, respectively.

Apple juice and cider processing
Samples of apples were selected, sanitised (Gomes et al., 2014) and crushed in a microprocessor (Metvisa, Type MPA, Brazil). The grated mass was placed in plastic screens (polyethylene thread 0.27 mm) and subjected to a pressure of 294 kPa for 5 min (Eureka Hydraulic press, Brazil). The juice was depectinised with 3.0 mL.hL -1 of pectinase (Pectinex® Ultra Clear, LNF Latino Americana, Novozymes, Brazil) at 40 °C for 60 min and after sedimentation it was racked. Part of the despectinised apple juice was bottled and frozen (-18 °C) until further analysis (Figure 1).
The other part of the despectinised apple juices was fermented with the yeast Saccharomyces cerevisiae r.f. cerevisiae (inoculated at 2.10 6 cells.mL -1 ; Fermol Bouquet, AEB Group, Otaly) under anaerobiosis to obtain the ciders. Alcoholic fermentation was conducted at 20-23 °C until complete utilization of fermentable sugars (< 0.1 g.L -1 ). The ciders were then centrifuged (8000 g at 4 °C for 20 minutes) (HOMAC CR-GOO, Hitachi, Japan) and bottled. The cider samples were kept at -18 °C until further analysis. All process (juice and cider) was made in duplicate.

Total phenolic compounds, total flavonoids and total flavanols
The analysis of phenolic compounds was performed by the Folin-Ciocalteu method (Singleton & Rossi, 1965). Results were expressed as milligrams of 5-caffeoylquinic acid equivalents per litre of apple juice or cider using a calibration curve. The total flavonoid content (TFC) of the juices and ciders was performed according to method described by Zhishen et al. (1999). The results were expressed as milligrams of catechin equivalents (CTE) using a calibration curve. The vanillin-HCl method was used for the determination of flavanols (Broadhurst & Jones, 1978); results were expressed in catechin equivalents.

Antioxidant capacity
The antioxidant capacity was evaluated in the apple juices and ciders using the FRAP (Benzie & Strain, 1996) and DPPH (Brand-Williams et al., 1995) methods. To estimate the antioxidant capacity, the variation in absorbance between the FRAP reagent and the mixture (FRAP reagent and sample) was correlated with a calibration curve of Trolox (0.1 to 1.0 mmol.L -1 ). The results were expressed in Trolox equivalents per litre of apple juice or cider (μmol TE.L -1 ). The assay is based on the reducing power of antioxidants present in extracts, in which a potential antioxidant reduces the ferric ion (Fe 3+ ) to ferrous ion (Fe 2+ ); the latter forms a blue complex (Fe 2+ /TPTZ).
The antiradical capacity determined by the DPPH method was defined as the amount of apple necessary to decrease the DPPH concentration by 50% or the EC 50 ; the lower the EC 50 , the higher the antioxidant power. This method determines the hydrogen donating capacity of molecules and does not produce oxidative chain reactions or react with free radical intermediates.

HPLC analysis of phenolic compounds
Samples (8 mL) of each apple juice and cider were lyophilized (LS 3000 Terroni, BR) and reconstituted with a solution (4 mL) of 2.5% acetic acid and methanol (3I:1, v/v) for the analysis of the individual phenolic compounds and their classes.
The chromatographic analysis was performed according to Alberti et al. (2014) in a chromatographic system, Waters 2695 Alliance (USA), which was equipped with a Symmetry C 18 (4.6 × 150 mm, 3.5 µm) column (Waters, USA) and a photodiode array detector PDA 2998 (Waters, USA). The identification and quantification of the compounds was performed by comparison of the retention times and spectra of the calibration curve of standards. For the compounds where commercial standards were not available, the quantification was performed from compounds belonging to the same class of phenolic compounds, as verified by fractionation (Picinelli-Lobo et al., 2009).
The fractionation was performed according to the method proposed by Dszmianski et al. (1988). The total peak area, which was obtained by HPLC analysis, was summed for each phenolic class and quantified as 5-caffeoylquinic acid, quercetin 3-glucoside and phloridzin equivalents for phenolic acids, flavonols and dihydrochalcones, respectively.

Statistical analysis
The data were presented as mean and standard deviation (SD). The Shapiro-Wilk and Hartley tests were used to check for normality of distribution and homogeneity of variances, respectively. Differences between samples were assessed by one-way ANDVA, followed by Fisher's LSD test. Pearson products (r) were used to evaluate the strength of correlation among the parameters that were evaluated. A p-value below 0.05 was considered to be significant.
Principal component analysis (PCA) was applied to separate the samples according to their values of responses. Before applying PCA, all the variables were auto scaled to standardize their statistical significance.
Multivariate regression by partial least squares (PLS) was used to predict the antioxidant activity of the juices and ciders. The antioxidant activity was used as the dependent variable (Yi), and the total phenolic compounds, phenolic acids, flavonoids, flavanols, flavonols and dihydrochalcones were used as the independent variables (X n ). Linear models were constructed, as can be observed in Equation 1.
where Y is the predicted response and β 0 , and β i , are the regression coefficients. Onitially, all the variables were autoescaled and the models were established.
The statistical significance of the equations was examined by ANDVA and the adequacy and quality of the models were evaluated by regression coefficients (R 2 ), adjusted R 2 and calculation of the root mean square error of calibration (RMSEC) (Zielinski et al., 2014b). The statistical analysis was performed using Statistica 7.0 software (StatSoft Onc., USA).

Influence of ripening stage on phenolic compounds of the juices
A significant variation (p < 0.05) in the total phenolic compound content was observed between varietal apple juices (326 to 583 mg.L -1 ) at different ripening stages (Table 1), as previously reported in other studies (Valavanidis et al., 2009;Zardo et al., 2013;Le Bourvellec et al., 2015). These values are typical of apple juices that were processed with dessert apples and obtained by the traditional method by pressing, without antioxidants (Guo et al., 2013).
No significant difference was observed (p > 0.05) between the total phenolic content of the juices at different ripening stages for the Fuji Suprema and Lis Gala varieties. However, in the Gala apple juice (highest phenolic compounds content) showed a decrease from the unripe to ripe stages, followed by an increase until the senescent stage. This could also be observed in the analysis of the phenolic classes, except for the dihydrochalcones. Alberti et al. (2017) also found reduction of dihydrochalcones with senescence in fruits of Gala variety.
Flavonoids were the main phenolic compounds class quantified in the apple juices (32%) and within this class, flavanols represented 70% of the total, followed by dihydrochalcones (15%) and flavonols (11%). On the Gala and Lis Gala varieties, the flavonoid and flavonol contents evolved in the same manner as the total phenols in the Gala variety; in other words, there was a decrease in flavonoid content from the unripe to the ripe stage, followed by an increase until the senescent stage. The increase observed between the ripe and senescent stages may have been related to higher activity of the pectinolytic enzymes of the fruits (Goulão et al., 2007), which facilitates the extraction during juice processing (Zielinski et al., 2014a) and not due to a synthesis process.
The monomeric flavonoid identified in the largest quantities in the apple juice was phloridzin (from 8.7 mg.L -1 in Senescent Lis Gala to 19.5 mg.L -1 in unripe Fuji Suprema juice). Apples, and a few other species of the Rosaceae and Ericaceae families, have phloridzin in their composition, however, only apples contain high levels of this compound (Gosch et al., 2010;Turner et al., 2005). The highest concentrations of phloridzin were observed in unripe juices, with 13.6; 10.3 and 19.5 mg.L -1 in samples of the Gala, Lis Gala and Fuji Suprema varieties, respectively, which decreased about 30% with senescence. Despite the reduction in the phloridzin content with fruit ripening, dihydrochalcones levels remained stable in juices from the Gala and Lis Gala varieties.
Hydroxycinnamic acids are the second largest class of phenolic compounds found in apples and within this group 5-caffeoylquinic acid is the compound with the highest concentration in juices and ciders (Goulas et al., 2014;Khanizadeh et al., 2008). The highest levels of 5-caffeoylquinic acid were found in the juices from unripe apples, and a reduction with ripening was observed in all cultivars. The content ranged from 85.69 to 52.18 mg.L -1 in Gala apple juice and from 40.8 to 25.3 mg.L -1 in Fuji Suprema, a variation of about 40%, while in Lis Gala this variation was 15% (66.1 to 56.1 mg.L -1 ).
The Gala juices from unripe and senescent apples had the highest levels for antioxidant activities by the FRAP method; while in the Fuji Suprema juices the highest values were found in the unripe stage, and these decreased with maturation. The samples of Lis Gala juices showed no significant difference in terms of antioxidant activity (FRAP) between the samples that were analysed. The results of antioxidant activity obtained by the FRAP method were correlated to those found by DPPH (r = -0.82, p < 0.001).
The flavonoid content was highly correlated with the antioxidant activity of the samples (FRAP, r = 0.92, p < 0.001; DPPH, r = -0.88, p < 0.01) and in this class the flavanols had the highest significant correlation (FRAP, r = 0.91, p < 0.01; DPPH r = -0.91, p < 0.01). Polymers of catechin and epicatechin (procyanidins) are the major flavanols found in apples and they are recognised for high antioxidant activity and health benefits (Guyot et al., 2002).

Influence of fermentation and ripening stage on phenolics of the ciders
The phenolic composition was changed during the fermentation (Table 1). The reduction in total phenols ranged from 17% (senescent Lis Gala) to 50% (unripe Gala). The Gala variety had the highest content of total phenols, however, the greatest reduction was also observed in this cultivar, resulting in ciders with levels similar to the Lis Gala variety.
The average of total phenols indicated that the transformation of the juices (426 mg.L -1 ) to ciders (289 mg.L -1 ) was responsible for a 32% reduction in phenolic compounds. However, the reduction of the phenolic classes was lower (14%), which was consistent with a reduction in antioxidant activity as measured by FRAP (10%). The greater reduction observed by the analysis of total phenols may have been related to interfering compounds in the juice, such as sugar and amino acids, which can react with the Folin Ciocalteu reagent (Everette et al., 2010;Prior et al., 2005). Apple juice has sugars (fructose, glucose and sucrose) and amino acids that are consumed during fermentation by the yeast (Santos et al., 2016).
The most affected classes were the phenolic acids in the Gala variety and the flavanols in Fuji Suprema. According to Caridi et al. (2004), the effect of yeast can change the phenolic profile of wines and this can be explained by its ability to adsorb phenols (Márquez et al., 2009); the cell wall polarity of the yeast defines the ability of the yeast to adsorb various molecules such as volatile compounds, fatty acids, and pigments. On addition, the yeast Saccharomyces cerevisiae is capable of converting the phenolic acids of the apple must to vinyl phenols by decarboxylation. The activity of the enzyme cinnamate carboxy-lyase (SCD) only occurs during alcoholic fermentation and the intensity depends on the strain that is used (Chatonnet et al., 1993).
The antioxidant activity did not change uniformly due to changes in the phenolic compounds. The correlation with flavonoids remained high (FRAP, r = 0.90, p = 0.01; DPPH, r = -0.98, p < 0.001) because the fraction of these compounds, which represented 32% of total phenols of the juices, increased to 45% in the ciders. The flavanols showed a higher correlation with the antioxidant activity of the ciders (FRAP, r = 0.96, p < 0.001; DPPH, r = -0.97, p < 0.001). The change in the content of phenolic acids due to fermentation caused a decrease in the correlation with antioxidant activity measured by FRAP (r = 0.69, p = 0.04), while that obtained by the DPPH method increased (r = -0.85, p = 0.004). The same behaviour was observed for 5-caffeoylquinic acid (FRAP, r = 0.80, p = 0.01; DPPH r = -0.86, p = 0.003), which was the major representative of the hydroxycinnamic acids.

Multivariate analysis of data
Principal component analysis (PCA) was performed to evaluate the data related to the phenolic composition and the antioxidant capacity of the juices and ciders obtained from fruits at different ripening stages.
The scatter plot (Figure 2) suggested the formation of three groups. The group on the right side of the scatter plot was formed by the beverages produced with Fuji Suprema apples that had high DPPH values (lower antioxidant capacity). The two other groups were formed by the beverages produced from the Gala and Lis Gala cultivars, which had higher levels of phenolic compounds and antioxidant capacity. The sample of unripe Fuji Suprema juice in the Gala and Lis Gala group showed the influence of the ripening stage on the content of phenolic compounds in apples.
On order to predict the antioxidant capacity using the FRAP and DPPH methods in a quick and easy manner, linear models were built for the apple beverages using multivariate PLS regression as a function of total phenols and classes analysed by HPLC. Multiple regression analysis of FRAP values showed that the model (Equation 2) was significant (p < 0.0001) and could explain 84.15% (R 2 adj = 83.10%) of all variance in the data. The RMSEC value for the model was 90.81 μmol.L -1 (11.22% error). The model for the DPPH assay (Equation 3) was also significant (p < 0.0001) and could explain 79.72% (R 2 adj = 78.45%) of the total variance. The RMSEC of the model was 77.66 mg.L -1 (14.00% error). Although it was only used a limited number of samples, PLS appears to be a suitable approach to predict the antioxidant capacity of apple beverages.

Conclusion
The phenolic content in apple juices at different ripening stages is related to variety. The contents of the main constituents of the Gala and Lis Gala apple juices decreased from the unripe to ripe stages and increased from the ripe to senescent stages, while in the Fuji Suprema variety these values decreased with senescence. The fermentation of the juice for cider production reduced the phenolic content and mainly affected the phenolic acids and flavanols (> ripening stage < phenolic losses).