Resveratrols in Grape Berry Skins and Leaves in Vitis Germplasm

Background Resveratrol is an important stilbene that benefits human health. However, it is only distributed in a few species including grape and is very expensive. At present, grape has been an important source resveratrol. However, the details are scarce on resveratrol distribution in different Vitis species or cultivars. Methodology/Principal Finding The composition and content of resveratrols were investigated by HPLC for assessing genotypic variation in berry skins and leaves of 75 grape cultivars, belonging to 3 species and 7 interspecific hybrids. Trans-resveratrol, cis-piceid and trans-piceid were detected in berry skins and leaves, but cis-resveratrol was not. Resveratrol content largely varied with genetic background as well as usage. In most cultivars, total resveratrol including the above three compounds was higher in berry skins than leaves. In berry skins of most cultivars and leaves of almost all cultivars, cis-piceid was the most abundant resveratrol; trans-resveratrol and trans-piceid were minor components. Some specific cultivars were found with extremely high levels of trans-resveratrol, cis- piceid, trans-piceid or total resveratrols in berry skins or leaves. In skins and leaves, rootstock cultivars had a higher content of total resveratrols, and the cultivated European type cultivars and their hybrids with V. labrusca had relatively low totals. There were no significant correlations of the amounts of total resveratrols or any individual resveratrol between berry skins and leaves. All 75 cultivars can be divided into four groups based on the composition of resveratrols and their concentration by principal component analysis. Conclusion Resveratrol content of grape berries and leaves varied largely with their genetic background and usage. Rootstock cultivars had a higher content of total resveratrols than the other germplasm. Total resveratrols were lower in leaves than berry skins in most cultivars. Cis-piceid was the most abundant resveratrol in most cultivars, and trans-res and trans-pd were minor components.


Introduction
The polyphenol trans-resveratrol (trans-3,5,49-trihydroxystilbene), a member of the stilbene family, was first isolated from white hellebore (Veratrum grandiflorum O.) [1]. It has become the focus of a number of studies in medicine and plant physiology, and has also emerged as a molecule that may affect human health. In leaves and berries of grape (Vitis), trans-resveratrol is a phytoalexin that is produced in response to stresses, such as wounding or pathogen attack [2]. Due to its daily consumption in the form of red wine [3][4], trans-resveratrol has been associated with the ''French paradox'' for having beneficial effects [5], particularly protection against coronary diseases [6]. In addition, trans-resveratrol may have anticancer properties, as suggested by its ability to suppress proliferation of a wide variety of tumor cells, including lymphoid and myeloid cancers; multiple myeloma; cancers of the breast, prostate, stomach, colon, pancreas, and thyroid; melanoma; head and neck squamous cell carcinoma; ovarian carcinoma and cervical carcinoma [7][8]. Nowadays, the main market for trans-resveratrol is in nutraceuticals, and an increasing awareness of the possible health benefits of the chemical is likely to further increase consumer demand for the product. Consequently, it may be necessary to develop new sources for obtaining the chemical.
The various resveratrols include transor cis-resveratrol and the corresponding piceids. Trans-resveratrol is formed via the phenylalanine/polymalonate route, where stilbene synthase plays a critical step. Trans-resveratrol may also be transformed into cisresveratrol. Transand cis-resveratrols may also be transformed into transand cis-piceids by glycosyltransferases. Glycosylation of polyphenolic compounds commonly occurs in plants to protect the plant cell from their potential toxic effects. At the same time, glycosylation may protect resveratrol from oxidation and enzymatic degradation, thus enhancing its stability [9]. Piceids are especially interesting for human health because they have also been reported to inhibit DNA synthesis in Lewis lung carcinoma cells and the formation of capillary-like tube networks (angiogenesis) of human umbilical vein endothelial cells [10]. They also inhibit thromboxane B2 in the human body [9].
Grape (Vitis) is one of the most important fruit crops in the world. Viticulture and enology play an important role in the economy of many developed and emerging countries. Gross world production of grape berries was estimated at more than 68.35 million metric tons in 2010 (FAO STAT Database at www.fao. org). About 27% of the grapes are consumed as fresh fruit (table grapes) and 2% as dried fruit, whereas 71% of the crop is processed, especially for winemaking.
The grapes cultivated throughout the world today mainly belong to three types, the European type (V. vinifera L.), the American bunch type (V. labrusca L.), and their derivatives, especially the hybrids obtained from V. labrusca and V. vinifera. Given the potential importance of the resveratrols, it is important to evaluate the content of resveratrols in grape germplasm in order to better utilize germplasm resources directly or in breeding programs to obtain new grape cultivars rich in resveratrols in berries and other tissues. Previous evaluations have only focused on trans-resveratrol in berries [11][12][13][14], and none have evaluated all forms of resveratrol, i.e., transresveratrol, cis-resveratrol, trans-piceid and cis-piceid. Although grape leaves are also rich in resveratrols [7], there are no evaluations of foliar resveratrols in grape cultivars. The objective of the present study was to evaluate resveratrol distribution in cultivars, hybrids, and species, focusing on content of all four forms in berries and leaves in order to provide information for direct use in the wine and resveratrol industries and to assist in the choice of parents in breeding programs.

Plant material
Seventy-five cultivars were evaluated in this study in 2009 (Table 1). These cultivars included: 34 table grapes from hybrids between V. vinifera and V. labrusca; 15 table grapes of V. vinifera; 15 wine grapes of V. vinifera; 3 interspecific juice grapes: Honey Juice (V. labrusca 6 V. vinifera), Beixiang (V. thunbergii 6 V. vinifera), Russian Concord (V. labrusca 6 V. amurensis) and 1 V. labrusca juice grape; 4 interspecific rootstock cultivars: Beta (V. labrusca 6V. riparia), Shanhe 2 (V. amurensis 6V. riparia), Zhi 168 (V. monticola 6V. riparia) and Dog Ridge (V. rupestris 6 V. berlandieri), and one V. berlandieri rootstock cultivar, respectively. All samples were collected from the experimental vineyard of the grape germplasm repository in the Institute of Botany, the Chinese Academy of Sciences located in Beijing. All of these cultivars were planted in the spring of 1993. The grapevines were trained to cordons, spaced 1.5 m apart within the row and 2.5 m apart between rows with a north-south row orientation. They were subjected to the same management practices, such as irrigation, fertilization, soil management, pruning, and disease control.
Grape berries were sampled at ripening according to the former years' ripening date and as judged from seed color change to dark brown without senescence of berry tissue. Berries were sampled from three clusters, randomly chosen in three vines of each cultivar as three replicates. The leaf samples were taken from the seventh to ninth leaves of a new shoot on each grapevine with three replicates. The skins were separated by hand. All the samples of fruit skins and leaves were immediately frozen in liquid nitrogen, and then stored at 240uC.

Extraction of resveratrols from berry skins and leaves
The frozen materials were ground to a powder in liquid nitrogen using a grinding machine (A11-b-s25, IKA, China). Each sample was extracted for 24 h in methanol and ethyl acetate (1/1, v/v) (1000 mg per 10 mL of organic solvent) at 25uC in darkness according to Liu et al. [15]. The suspension was centrifuged at 10,0006 g for 10 min. The supernatant liquid was separated for collection. The remaining residue was extracted with 3 mL of methanol and ethyl acetate (1/1, v/v) and centrifuged. The two supernatants were combined and vacuum-dried with a rotary evaporator (N-1001D-WD, EYELA, Tokyo Rikakikai, Japan) at 40uC. The dried samples were then re-dissolved in 2 mL of pure methanol and stored at 240uC for resveratrol analysis.

Measurement of resveratrols
Each liquid sample was filtered through a 0.22 mm PTFE membrane filter, then the analyses were carried out on a Dionex P680 HPLC system (Dionex Corporation, CA, USA) with a Dionex PDA-100 detector. Separation was achieved using a reverse-phase C18 column (AtlantisH T3; 4.6 mm6250 mm, 5.0 mm particle size, Waters, USA) and a guard column (Atlantis T3, 4.6 mm620 mm, 5.0 mm cartridge, Waters, USA) maintained at 30uC with a Dionex TCC-100 thermostated column compartment, and an injection volume was 10 mL. Separation was performed at a flow rate of 1.0 mL?min 21 with the mobile phase consisting of H 2 O (A) and acetonitrile (B). The solvent gradient was as follows: 0-5 min from 10% to 17% solvent B; 5-12 min from 17% to 18% solvent B; 12-22 min from 18% to 22% solvent B; 22-30 min from 22% to 33% solvent B; 30-45 min from 33% to 38% solvent B; 45-50 min from 38% to 80% solvent B; 50-53 min from 80% to 10% solvent B; and, 53-60 min 10% solvent B. For fluorimetric detection, the maximum absorption wavelength of the two trans-isomers (trans-resveratrol and trans-piceid) was 306 nm, and the two cis-isomers (cis-resveratrol and cis-piceid) was 288 nm. Three replicates of each sample were also scanned from 240 to 600 nm. Trans-resveratrol and trans-piceid standards were purchased from Sigma-Aldrich (St. Louis, MO, USA) and the Chinese Standards Research Institute, respectively. The mixed solution of the trans-resveratrol (trans-res) and trans-piceid (trans-pd) standards were partly converted to the cis-resveratrol (cis-res) and cis-piceid (cis-pd) after UV-C irradiation at 6 W m 22 and a distance of 30 cm for 30 min. Conversion coefficients were computed from the two trans-isomers, respectively, and standard curves of the four isomers were made. The total resveratrol content was obtained from the sum of the four isomers.

Graphs and data analysis
The variations of extractable amounts of the different forms of resveratrol in berry skins and leaves were analyzed, as was the distribution of the resveratrols in the samples, by S-Plus 2000 (MathSoft Inc., Cambridge, MA). Boxplots were used to display range, median, and distribution density of variables in the samples [16]. The lower and upper quartile values were indicated by the height of the box, or the interquartile range (IQR). Whiskers indicated the range of the data and were represented as vertical lines ending in a small horizontal line. Extreme values outside these whiskers were indicated by asterisks. In all the Tables and Figures, experimental data were subjected to analysis of variance using the SPSS 13.0 program (SPSS, USA.). Means were separated by Student-Newman-Keuls multiple range tests at P,0.05.

Resveratrol in grape berry skins
The frequency distribution and median of total resveratrols in grape berry skins are shown in Fig. 1. Among all the genotypes, rootstock cultivars had the highest amounts of resveratrols in skins, ranging from 20.28 to 365.98 mg g 21 fresh weight (FW) with a median of 235.32 mg g 21 FW. However, total resveratrols in berry skins of the rootstocks varied significantly with their genetic backgrounds. The highest amounts in skin were found in 'Berlandier Resseguier No.2' (74) at 365.98 mg g 21 FW, followed by 'Zhi168' (73) with 344.68 mg g 21 FW, and 'Shanhe 2' (72) with 235.32 mg g 21 FW. Among the rootstock cultivars, 'Dog Ridge' (75) had the lowest amount of resveratrols with only 20.28 mg g 21 FW.
Juice grapes and two interspecific hybrid wine grapes from V. vinifera and V. amurensis had relatively high resveratrol content. Their medians were higher than those of table grapes. In the above juice and wine groups, 'Concord' (68) had very high amounts of resveratrols with 146.60 mg g 21 FW, while resveratrols in 'Russian Concord'(67) were very low with only 0.5 mg g 21 FW.
Among most of the grape germplasm groups studied, berry skin resveratrol content was lower in the wine and table grapes of V. vinifera, and the interspecific hybrids between V. labrusca and V. vinifera, as compared with the previous groups. Moreover, the distribution of their totals largely skewed towards a low content with a median much lower than the average (Fig. 1)

Resveratrols in grape leaves
The content of total resveratrols were much lower in leaves than in fruit skins (Fig. 2). Similar to fruit skins, the rootstock cultivars had the highest resveratrol content in leaves among all the grape genotypes studied, and juice and hybrid wine grapes from V. vinifera and V. amurensis had a relatively high content. All cultivars of V. vinifera and the hybrid cultivars of V. labrusca and V. vinifera had low total resveratrol content in the leaves.

Composition of resveratrols in berry skin and leaves
In this study, cis-res was not detected in any germplasm, and was thus omitted in Table 2. As shown in Table 2, the composition of berry skin resveratrols in the germplasm differed from that in leaves. In the berry skins, there are 10 cultivars which had only trans-res and cis-pd, 13 cultivars with only trans-res and trans-pd, and 5 cultivars with only trans-res. But, in leaves, almost all cultivars had three resveratrols, except 'Himrod (7)' and 'Bujiesuli (52)'.
The content of the different resveratrols varied with the genetic background. In berry skins, trans-res, trans-pd and cis-pd contents ranged from 0.01 to 84.63 mg g 21 FW, from 0 to 24.49 mg g 21 FW and from 0 to 360.12 mg g 21 FW, respectively.
As regard the percentages of the different resveratrol components, cis-pd accounted for more than 90% of the total resveratrols in berry skins of rootstocks while less than 56% in the other germplasm groups (Table 2). Moreover, both cis-pd and trans-pd were not detected in 'White Olympia' (1), 'Red Queen' (10), 'Ziguang' (33), 'Jingzaojing' (44) and 'Merlot' (54). Interestingly, the total resveratrol content of rootstock berry skin were much higher than that in the other groups, while 5 cultivars without trans-pd and cis-pd had very low total resveratrol. Moreover, the correlation among different resveratrol types (trans-res, trans-pd and cis-pd) was very low in berry skins and leaves; the only high correlation coefficients (0.973) were between total resveratrols and cis-pd in skins (Table 3). Those results signified that the total resveratrols in berry skins were mainly from the accumulation of cis-pd. This was likely due to the stability of cis-pd in the tissues [17]. It also suggested that glycosylation, especially glycosylation of cis-pd, was a critical step in accumulation of total resveratrols in berry skins. In addition, cis-res was not detected in berry skins even though cis-pd was a main component, indicating that cis-res was rapidly glycosylated to cis-pd after cis-res was converted from transres.
The percentage of cis-pd was about 60% of total resveratrols in leaves of rootstock grapevines, much lower than that in berry skins (Table 2), and was even lower than that in leaves from the other groups. Moreover, almost all germplasm had all three resveratrols in leaves at low levels. There was no correlation in resveratrol content between leaves and berry skins ( Table 3, diagonal). In contrast to berry skin, there were significant correlations between all three resveratrols in leaves (Table 3). From the differences in content and the proportion of each resveratrol, we suggest that the mechanism regulating the accumulation of resveratrols in leaves differs from that in berry skins. As in skins, transand cis-pd were also the main contributors to the total resveratrols in leaves. Gatto et al. (2008) also showed that the grape cultivars with higher resveratrol accumulate resveratrol preferentially in the glycosylated forms, transand cis-pd [18]. Three possible roles for resveratrol glucosides in berry skins and leaves include storage of resveratrol, transport from cytoplasm to apoplast, and protection of transresveratrol from peroxidative degradation [17].

Principal component analysis (PCA)
PCA was used to analyze the data for trans-res, trans-pd and cispd from berry skins and leaves of 75 grape cultivars. Fig. 3 shows the relationships between the cultivars (scores) and their resveratrols (loadings). The two PCs carried a large amount of important information and accounted for 73.6% and 95.3% of the total variance for resveratrols in berry skins and in leaves, respectively. The PCA scatter plots of berry skins and leaves are shown in Fig. 3A and Fig. 3B, respectively, and the corresponding loading plots, which establish the relative importance of the variables, are shown in Fig. 3E and Fig. 3F, respectively.
With regard to resveratrols in berry skin, the grape cultivars can be divided into four groups based on the positions in the PCA scatter plots (Fig. 3A). . This group was characterized by a very high content of trans-res. 'Venus' (29), which is a hybrid between V. vinifera and V. labrusca, was in the top right quadrant which indicates that it is associated with an extremely high positive value for PC1 and PC2. This cultivar was characterized by an extremely high trans-res. Group 4: The other Vitis cultivars were found near the PC1 axis. Their trans-res, transpd and trans-pd contents were very low.
As shown in Fig. 3B for leaves, the grape cultivars can be divided into four groups based on positions in the scores scatter  Table 2. Content (C) of different forms of resveratrols and their percentage (P) accounting for the total resveratrols in berry skins and leaves of 75 grape cultivars.   Table 3. Correlation of resveratrols between and in grape berry skins and leaves.    73)]. This group was characterized by a higher trans-res content. Group 3: This group was situated in the right side of the fourth quadrant and was comprised of the rootstock grape V. Berlandieri 'Dog Ridge'(75) with higher trans-res, trans-pd and cis-pd. Group 4: The other Vitis cultivars were found near the PC1 axis.
Their trans-res, trans-pd and cis-pd contents were very low.

Conclusions
Resveratrol content varied significantly with genetic background as well as their usage. Trans-res, cis-pd and trans-pd were detected in berry skin and leaves, but cis-res was not. Generally, resveratrol totals were lower in leaves than berry skins in most cultivars. Rootstock cultivars had a higher content of total resveratrols than the other cultivars. Cis-piceid was the most abundant resveratrol in most cultivars, while trans-res and trans-pd were minor components. Some cultivars were found to have extremely high or low trans-res, cis-pd, transpd or total resveratrols in berry skins or leaves. There were no significant correlations of total resveratrols between berry skins and leaves.