3.2.1. Flavonols and Anthocyanins
The flavonols detected and identified in both harvests (
Table 2) were derivatives of the Q, I, M, L and S aglycones. No flavonol derived from K was found, which is the first compound formed in the biosynthetic route of flavonols and soon transformed into other compounds [
22]. This compound was also absent in the hybrid BRS Violeta [
23] and the
Vitis vinifera Italia [
45] grapes. Free flavonols were reported in harvest 1, and this suggested that during the extraction process of the grapes, acid hydrolysis of these compounds might have occurred. Therefore, to compare the results of the individual molar ratios of flavonols obtained for each harvest, the molar proportions were recalculated per type of flavonol aglycone (M-type, Q-type, L-type, I-type and S-type).
Both harvests (
Table 2) showed higher concentrations of M-type flavonols, with 81% and 76%, respectively for WG1 and SK1, and 51% and 48%, respectively for WG2 and SK2, followed by flavonols of Q-type flavonols. In addition to the free form, for harvest 1, flavonols linked to a glucuronide (3-glcU), galactoside (3-gal) and glycoside (3-glc) were identified. A similar flavonol profile to that determined for BRS Carmem harvest 1 was described for the Bordô grape, in SK and flesh analyzed separately; however, the flavonol K-3-glc was present in Bordô grapes [
46], but not in BRS Carmem (in the present study). The Rufete (
Vitis vinifera) [
47] and BRS Violeta grapes [
23] also showed a similar flavonol profile to BRS Carmem in the present study, but with the absence of flavonols from I and the presence of flavonols from K. The coumaroylated flavonol derived from S (S-cmglc), recently found and identified by Favre et al. (2018) [
48] in
Vitis vinifera grapes and their wines, was identified in the present study for BRS Carmem of harvest 2 with concentrations of 0.8% for WG2 and 0.8% for SK2.
Liang et al. [
49] evaluated the commercial developmental patterns of the individual flavonols from three grape cultivars grown in the same region during two consecutive years and found that the edaphoclimatic conditions, in particular the solar incidence, rainfall volume during the production cycle and temperature between the annual harvests markedly influenced the flavonol profile of the grapes. The authors reported, for example, that the flavonols Q-3-glc and Q-3-gal were not detected in Merlot and Cabernet Gernischt from the first harvest, whereas they were present in the following harvest. Moreover, some cultivars were more phenologically and climatologically sensitive, which can result in a variation in qualitative and quantitative composition in the same cultivar from one season to another.
Analyzing the qualitative profile of anthocyanins for BRS Carmem, we found 26 compounds for WG1 and 31 compounds for WG2, which were derived from the five main anthocyanidins; delphinidin (dp), cy, pn, petunidin (pt) and mv were found, while diglycosylated (3,5-glc) and monoglycosylated (3-glc) derivatives in both non-acylated and acylated (acetylated, cumarylated, and some caffeylated) forms were also found (
Table 3). The major anthocyanins were derived from dp for WG1. However, for WG2, the major anthocyanins were derivates from mv. The
p-coumaryl monoglycosylated (3-cmglc) derivatives from pt, pn and mv were identified for both harvests, but the derivative from cy was present just in WG2. Acetylated diglycosydes (3-acglc-5-glc) derived from pn and mv were detected in both harvest, but in addition, the derivative of pt was identified for WG2. The acetylated monoglycosylated (3-acglc) derivatives from dp, pt and mv were reported for both harvests, and the derivative from pn was also found for WG2. The
p-caffeylated anthocyanins were found in monoglycosylated form (3-cfglc) only for WG2 and from mv, as well as the
p-caffeylated diglycosyde anthocyanin (3-cfglc-5-glc) derivatives of cy, pt, pn and mv. On the other hand, for WG1, there was only the
p-caffeylated diglycosyde derivative from dp.
For WG1, the anthocyanins derived from dp were also predominant for BRS Violeta [
23]. The predominance of anthocyanins derivatives from mv, as reported for WG2, was also noted for Bordô grape skins [
46] and for the hybrid grape Maximo [
50]. Higher concentrations of mv derivatives were also reported by Silva et al. [
11] for BRS Carmem skins grown on the same rootstocks use in the present study. These authors reported the presence of mv-3,5-glc, cy-3,5-glc and 3-glc derived from mv, cy, dp, pn and pelargonidin anthocyanins. For BRS Vitória grapes, the authors reported that in the monoglycosylated anthocyanin fraction, there is a higher predominance of the derivatives from dp, while for the diglycosylated anthocyanin fraction, there is a predominance of mv derivatives [
29]. From the results of our study, it can be established that the predominance of mv or dp among the anthocyanins of BRS Carmem grape was influenced not only by the grape cultivar.
Both monoglycosylated and diglycosylated anthocyanins are seen, although the relationship between these two forms (ratio 3,5glc/3glc; 3,5glc, diglycosylated and 3glc, monoglycosylated) showed that there is a predominance of diglycosylated anthocyanins for both harvests, with values significantly higher for WG2 (5.6) than WG1 (2.2). In this context, BRS Carmem grape presented percentages of diglycosylated anthocyanins for WG1 of 66% and for WG2 of 85%, and these results are in accordance with the genealogy of this cultivar [
31].
There are reports in the literature of non-
vinifera grapes with a percentage of diglycosylated anthocyanins of approximately 90%, such as for Bordô grapes [
46] and BRS Violeta grapes [
23]. It is noteworthy that the lower the ratio, the smaller the difference between the concentrations of monoglycosylated and diglycosylated anthocyanins.
In the present study on BRS Carmem grapes, the qualitative profiles of the flavonols and anthocyanins showed that, for both harvests, there was a predominance of tri-substituted flavonols, with values of 90% (WG1), 89% (SK1), 75% (WG2) and 70% (SK2), as well as of tri-substituted anthocyanins (dp, pt and mv), both in the monoglycosylated fraction (92%) and the diglycosylated fraction (92% and 94%), respectively, for WG1 and WG2. WG1 showed a higher concentration of non-acylated diglycosylated anthocyanins (57%) than monoglycosylated anthocyanins, which were mostly acylated (54%); meanwhile, WG2 presented a higher concentration of acylated anthocyanins, both monoglycosylated (82%) and diglycosylated (85%). Silva et al. (2019) [
19], studying the influence of different rootstocks, reported that the proportion of di-substituted anthocyanins to tri-substituted anthocyanins for Cabernet Sauvignon grapes was also affected by the differences in the rootstocks.
It can be suggested that for both harvests, during grape development, a higher activity of the F3’5’H enzyme in the biosynthetic pathway of flavonols and anthocyanins occurred. That is because a higher concentration of tri-substituted compounds was observed, especially of M-type flavonols and the anthocyanins dp-3-glc, mv-3-glc and pt-3-glc. According to the general route of this biosynthetic pathway, an expressive activity of the enzyme 5-
O-glycosil transferase probably resulted in the formation of diglycosylated anthocyanins derived from dp, mv and pt. It is still not clear how the complex interactions between abiotic and biotic factors influence the activity of acyltransferase enzymes involved in the formation of acylated anthocyanins [
49]. However, the results of the present study suggest that the growing conditions of the first harvest positively enabled the anthocyanin acylation. This is a potential advantage for these grapes due to the greater color stability of acylated anthocyanins when compared to its corresponding non-acylated anthocyanins [
51].
Regarding the quantitative profile of flavonols (mg Q-3-glc•kg fruit
−1,
Table 2), harvest 2 showed significant differences between the edible parts (76.8 and 38.7 for WG2 and SK2, respectively), but this was not the case not for harvest 1 (70.5 and 68.5 for WG1 and SK1, respectively). All the samples showed lower values than reported for BRS Violeta skins (153 mg) [
23] and higher than Bordô grape (approximately 154 µmol) [
45]. In addition, the total concentration of flavonols present in SK1 represented approximately 97% of the fruit, whereas the concentration present in SK2 represents 51% of the fruit, which means that flavonols accumulated mainly in the skins of BRS Carmem grapes.
The concentration of flavonols in grapes remains relatively constant throughout the development of the berry [
52]. Nevertheless, Silva et al. (2018) [
11] showed differences in flavonols’ total concentration for BRS Carmem skins produced in harvest 1 (11.6) and harvest 2 (7.83), due, among other things, to differences in the chemical properties of the grapes, such as pH, TA and SS.
For the quantitative anthocyanin profile (
Table 3), harvest 1 provides a significantly higher concentration of monoglycosylated anthocyanins than that of harvest 2, and the concentrations of diglycosylated anthocyanins were significantly different from each other. For the quantitative anthocyanin profile (
Table 3), harvest 1 (559 mg mv-3-glc•kg fruit
−1) provides a significantly higher concentration of monoglycosylated anthocyanins than that of harvest 2 (194 mg mv-3-glc•kg fruit
−1), and the concentration of diglycosylated anthocyanins was similar between the two harvests (1650 and 1631 mg mv-3,5-diglc•kg fruit
−1). Silva et al. [
11] reported for BRS Carmem SK produced on ‘IAC-766′ rootstock a value of 2342 mg mv-3-glc•kg fruit
−1, and Lago-Vanzela et al. (2011) [
46] reported for Bordô grape skins a value of 1360 mg mv-3,5-glc•kg fruit
−1.
The accumulation of anthocyanins in the skin of the grape berries occurs first due to the slower accumulation of anthocyanins, followed by a rapid increase and a stabilization phase. At the end of the ripening process, it is possible to observe a decrease in the concentration of this dye [
52].
The difference between the total concentrations of anthocyanin (mg mv-3,5-glc•kg fruit
−1) determined in harvests 1 (2528) and 2 (1921) may be related to the different degrees of maturation. The values found for both harvests were close for BRS Carmem SK produced on IAC-766 rootstock (2342 mg cy-3-glc•kg fruit
−1) [
11], as well as that reported for Bordô skins (1359 mg mv-3,5-glc•kg fruit
−1) [
23], despite the differences in the anthocyanins used to express the results. Previously, Oliveira et al. (2019) [
53] found, for Syrah grapes, maximum anthocyanin values (g mv-3-glc•kg fruit
−1) of 32.9 for a grape harvested in Bahia (Brazil) and 15.8 for a grape harvested in Pernambuco (Brazil), both lower than those reported in this study.
The interest in anthocyanins, besides their coloring properties, is also related to their potential health benefits, such as antioxidant action and protection against neural and cardiovascular diseases, cancers, diabetes, and inflammation, among others [
54,
55]. Flavonols are related to reducing the risk of developing pathogenic diseases, such as cancers [
56] and cardiovascular diseases [
57]. Several studies have investigated how to obtain such products (bioingredients and dyes) in solid (powder) and liquid forms or in an extracted form from fruits and vegetables, such as blueberry [
58],
jabuticaba [
59], grape [
60], purple carrot [
61] and eggplant peel [
62].
Flavonols, in addition to possibly having functional properties, can act as copigments, stabilizing anthocyanins through intermolecular copigmentation reactions and inhibiting anthocyanins’ degradation and prolonging color stability [
63]. When comparing the concentrations of anthocyanins and flavonols in BRS Carmem grapes with those available in the literature, it is noticeable that even when immature, this cultivar is still rich in these compounds, and may be used as an alternative source for the production of bioingredients, either dehydrated or as extracted anthocyanins, to be used as natural dyes.
3.2.2. Qualitative and Quantitative Profiles of Stilbenes, Flavan-3-ol Monomers and Dimers and Proanthocyanidins
The quantification of the flavan-3-ol monomers and dimers is shown individually and expressed in mg of each compound•kg fruit
−1 (
Table 4). The total concentration (sum of flavan-3-ol monomers and dimers) was also separated into concentrations of monomers and dimers. The total concentration of PA was expressed in mg of C•kg fruit
−1; the PA structural characteristics such as %galoylation, %prodelfinidines, and the percentages of flavan-3-ols with extension units and terminal units were given in percentages (
Table 4).
C, EC, GC, EGC and ECG were present in the SK1 sample, while their respective SE1 also contained CG in addition to these compounds. For harvest 2, the seeds presented C, EC, CG, ECG and two glycosylated monomers (MG1 and MG2), while in their respective skins, C, EC, GC, EGC and ECG were found. In BRS Carmem, the C was the major flavan-3-ol monomer and the seeds showed higher total concentration than the skins. These results were significantly higher for SE1 (179 mg C•kg fruit−1) than for SE2 (127 mg C•kg fruit−1). There was no significant difference between the skins (1.17 and 1.04 mg C•kg fruit−1 for SK1 and SK2, respectively).
For flavan-3-ol dimers, all samples presented the three type B dimers (PB1, PB2 and PB4); PB4 was predominant for both samples of SE1 and SK1, and for harvest 2, PB2 was the compound with the highest concentration for SE2, while PB1 was the highest for SK2. The total concentration of flavan-3-ols (TCF) dimers, within the same harvest, was higher in seeds than in skins, whereas harvest 2 showed higher concentrations than those of harvest 1, both for the seeds and for the skins. The TCF monomers dimers and sum were also significantly higher in the seeds than the skins. There were significant differences between the total concentrations of the sum of flavan-3-ols of the skins, with higher concentrations for harvest 2 than harvest 1.
The total concentration of PA was significantly higher in the seeds. However, in this case, the harvest 1 seeds showed a higher concentration, while in relation to the skins, harvest 2 showed a higher concentration than harvest 1. As expected, the % galloylation was also higher in the seeds, with values of 9 and 7%, respectively, for SE1 and SE2, as opposed to values of 2 and 4%, respectively, for SK1 and SK2, with only the skins showing a significant difference between the harvests. The skins had higher % prodelphinidins than the seeds, but, in this cultivar, these percentages were significantly higher in harvest 1. As usually found in grapes, the mDP was statistically higher in skins than in seeds.
BRS Violeta grape skins reported a higher proportion of C (49%) than the flesh (83%) and seeds (87%), and a TCF with values of 346 mg C•kg fruit
−1 in the seeds and 8.6 mg C•kg fruit
−1 in the skins [
23]. Those values of TCF were higher than the reported for BRS Carmem grape.
The recorded mDP values in the skin samples of this study exhibited lower measurements in comparison to the findings reported by Lago-Vanzela et al. (2011) [
45] for Bordô grape skins (12), and the %galloylation found in the skins of the grapes in the present study was lower than that for BRS Violeta SK (3%), except for the SK2 sample, which had a value of 4% (still quite near to 3%). For the seeds, the values found for PA and mDP in the present study were lower than those reported for the seeds of the BRS Violeta grape (12%) [
23].
The percentage of prodelphinidins in all of the samples of skins in the present study was higher than that reported for the Bordô grape, which had an approximate value of 14.2% in its skins [
46]. However, all of these values were lower than that reported for BRS Violeta skins (58%); similarly, in the same work, the authors report a value of 2% for BRS Violeta seeds [
23], higher than found here. Regarding the structural characterization of PA, for terminal units, “% C-term” and “% EC-term” showed higher percentages for all samples. For extension units, “% EC-ext” occupied a higher percentage in both harvests, with a higher concentration in the seeds.
The different phenolic compositions between the harvests are supported by other studies. Rajha et al. (2017) [
64] reported that the maximum concentration of polyphenols in the grapes was induced by higher temperatures (up to 25 °C) and lower rainfall. In addition, they described that thermal and water stresses have also been shown to increase polyphenolic production. When comparing the productive cycles of the two harvests studied, it can be highlighted that in the second harvest, the stages of veraison and complete maturation occurred with higher temperature values, and lower minimum humidity values. These facts may have enhanced the concentration of flavan-3-ol and proanthocyanidins in skin and seeds of BRS Carmem from the second harvest, even with incomplete maturation. These results show that the immature BRS Carmem grape, produced under these stress conditions, can be a relevant source of such compounds.
3.2.3. Qualitative and Quantitative Profiles of Hydroxycinnamic Acid Derivatives and Stilbenes
For both harvests, qualitative profiles and the total concentration of HCAD are shown in
Table 5.
Trans-caftaric acid,
trans-coutaric acid and
trans-fertaric acid were identified, as well as the isomers of glycosylated coumaric acid called 1”-glucoside-coumaric ester, 2”-glucoside-coumaric ester and 3”-glucoside-coumaric esters. These esters and hexoses of hydroxycinnamic acid have been reported for other grape cultivars that are cultivated and imported for the Brazilian market, such as BRS Violeta grape [
23] and Bordô grape [
46]. The predominant HCAD for BRS Carmem was mostly
trans-caffeic acid for all edible parts and harvests analyzed. This acid was also the major compound for the Isabel grape, followed only by
trans-coutaric and
cis-coutaric acids [
65].
For harvest 1 and harvest 2, the second major compound was
trans-coutaric acid and 3-glc-coumaric acid, respectively. There are reports that indicated that the HCAD profile may vary for the same grape cultivar, as was the case for BRS Violeta grape, which had variance in the HCAD profile: in most studies,
trans-caftaric acid was the major compound, such as for BRS Violeta and BRS Lorena grapes produced in São Roque (Brazil) in 2011 and 2012 [
66], but in this same study, the major compound in the skin was
p-1-glc-cumaric acid; a 2013 harvest, produced in Jales (Brazil), showed a higher concentration of
trans-caftaric acid in the whole fruit, followed by the fertaric acid [
30].
In the present study, the HCAD total concentration (all expressed in mg caftaric acid•kg fruit
−1) was significantly higher in WG when compared with their respective skins (
Table 5). It was expected that these compounds accumulate at the skins and flesh [
23]. For harvest 1, the concentration of these compounds in the skins represented 45% of the total, and for harvest 2, the concentration was 32% of the total. For WG, only WG1 showed a higher HCAD concentration than that reported for BRS Violeta grape (134 mg, the sum of skin and flesh) [
23], for Bordô grape (483 µmol) [
45] and also for Garnacha Tintorera (689–799 µmol) [
22]; converting the results from WG1 to µmol, we obtain a value of 1999 µmol. The HCAD results obtained for all of the examined skin samples in this investigation demonstrated higher values compared to those observed for BRS Violeta skins (14 mg) [
23].
Silva et al. (2018) [
11] also studied BRS Carmem grapes and did not report any significant differences between the results for the phenolic acids present. Costa et al. [
67], studying the influence of rootstocks in two distinct productive cycles of the Chenin Blanc grape, verified that the same rootstock can present different results between the cycles. In the present study, harvest 1 provided a higher concentration of HCAD. A similar result was disclosed for one of the productive cycles of the Chenin Blanc grape [
67]. In view of these results, it is suggested that harvest 1 provided a greater quantity of 4CL and C3H enzymes [
68] when compared to harvest 2 for BRS Carmem grape.
The stilbene profile and the total concentration (3-glc-resv•kg fruit
−1) are shown in
Table 6. The
cis- and
trans-piceid isomers were identified in both harvests. The
cis-resv compound was only identified in harvest 2. Stilbenes were seen in higher concentrations in the seeds than in the skins. The total concentrations of stilbenes for harvest 1 were 48.7 µg and 21.8 µg for seeds and skins, respectively, and for harvest 2 they were 896 µg and 3.69 µg for seeds and skins, respectively. For Bordô grapes, higher values were reported than for the samples studied here, at 10.9 mg [
46].
The results suggest that harvest 1 induced a higher concentration of piceid derivatives than harvest 2, probably due to the higher activity of the GT enzyme. On the other hand, harvest 2 significantly induced the formation and concentration of resv derivatives, probably due to greater activity of the STS enzyme [
69], which may be further investigated in future studies.