Berries of the wine-making cultivar ‘Beihong’ were used in this study. ‘Beihong’ is a cross between ‘Muscat Hamburg’ (V. vinifera L.) and wild V. amurensis Rupr. generated by the Institute of Botany, Chinese Academy of Sciences, and approved as a new grape cultivar in China in 2008 (Fan et al., 2009). Vines in this study were planted in the spring of 2003 in the experimental vineyard of the Institute of Botany, Beijing, China. The vineyard received local standard management practices including irrigation (kept grapevines from drought stress), fertilization, pruning, and disease control.

The experiments were divided into two parts. Part I assessed soluble solids in berry juice and ABA, anthocyanin and resveratrol accumulation in berry skins throughout berry development. Part II assessed the effect of exogenous ABA on ABA, anthocyanin and resveratrol synthesis and accumulation in berry skin. For Part I, the berries were sampled at different development stages under natural conditions. Clusters, chosen at random, were harvested at 40, 50, 60, and 70 DAA, and at 1 week intervals after veraison (70 DAA) until maturity (126 DAA). The veraison date was determined as the date when 50% of the berries had a visible change color. The maturity date was based on the seed color changing to dark brown without senescence of berry tissue and on previous records. Three replicates per cluster were sampled at each date. After the clusters were harvested, the berries were peeled rapidly, and the skins were frozen in liquid nitrogen. Samples were stored at -80°C until use. For part II, berry clusters at a similar developmental stage were chosen for ABA and fluridone (an inhibitor of ABA synthesis) application. Four treatment stages were at 20, 10 days pre-veraison, version and at 7 days post-veraison. 500 mg L^-1 (+)-ABA or fluridone solution was used to soak berry clusters for 1 min. All treatments occurred after nightfall. Purified water was used as the control. Samples were collected in the morning (6:00–7:00) at 0, 1, 3, 5, 7, 14, 21, 28 days after treatment and maturity. All treatments had three independent replicates, and each cluster was a replicate (consisting of ∼50 berries) and from different vines. The clusters were taken to the laboratory immediately after being harvested, to take photographs, and washed using running potable water, then washed using deionized water (to remove any residual exogenous ABA or ABA inhibitor solution on the berry surface). Berry skins were collected and stored as described for Part I.

Abscisic acid and its metabolites were quantified according to the method of Speirs et al. (2013). In brief, berry skins were freeze-dried and coursely ground berry skins were ground to a fine powder. A 50–100 mg sample was used for the extraction, to which a deuterated internal standard was added (400 μl, containing D3-7′, 7′, 7′-PA (phaseic acid) and -dihydrophaseic acid, D5-4, 5, 8′, 8′, 8′-ABA-GE, and D6-3′, 5′, 5′, 7′, 7′, 7′-ABA). Phenomenex SPE columns (60 mg mL^-1: 8B-S100-UAK) were used to provide an initial sample ‘clean-up’ step prior to analysis. The samples were loaded onto the columns, washed with 20% aqueous methanol, and eluted with 90% aqueous methanol. An aliquot of the eluate was dried in a vacuum centrifuge in preparation for analysis.

Abscisic acid concentration was quantified by LC-MS/MS. The dried berry skin extracts were dissolved in aqueous acetonitrile (10% with 0.05% acetic acid) and analyzed by LC-MS/MS (Agilent 6410). Separations were carried out on a Phenomenex C18 column at 40°C, using a gradient from 10% acetonitrile, 90% water and ending at 90% acetonitrile, 10% water. Compounds were identified by retention times and multiple reaction monitoring.

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 25°C in the dark (Liu et al., 2013). The suspension was centrifuged at 10000 × g for 10 min. The supernatant was separated and the resulting residue was extracted with 3 ml methanol and ethyl acetate (1/1, v/v). The organic solvent phases were pooled and vacuum-dried in a rotary evaporator (N-1001D-WD, EYELA, Tokyo Rikakikai, Japan) at 40°C. The dried samples were re-dissolved in 2 ml pure methanol and stored at -40°C for resveratrol analysis.

Each sample was filtered through a 0.22 μm PTFE membrane filter. Resveratrol was analyzed in a Dionex P680 HPLC system (Dionex Corporation, Sunnyvale, CA, USA) coupled to a Dionex PDA-100 detector. Separation was achieved using a reverse-phase C18 column of Atlantis^® T3 (4.6 mm × 250 mm, 5.0 μm particle size, Waters, USA) and a guard column (Atlantis T3, 4.6 mm × 20 mm, 5.0 μm cartridge, Waters, USA) maintained at 30°C with a Dionex TCC-100 thermostat column. The injection volume was 10 μl. Separation was performed at a flow rate of 1.0 ml min^-1 with the mobile phase consisting of H₂O (A) and acetonitrile (B). The solvent gradient was performed in the following manner, 0–5 min with 10–17% solvent B; 5–12 min with 17–18% solvent B; 12–22 min with 18–22% solvent B; 22–30 min with 22–33% solvent B; 30–45 min with 33–38% solvent B; 45–50 min with 38–80% solvent B; 50–53 min with 80–10% solvent B; and 53–60 min with 10% solvent B. For fluorimetric detection, the maximum absorption wavelength of two trans-isomers (i.e., trans-Res, and trans-Pd) was 306 nm and of two cis-isomers (i.e., cis-Res and cis-Pd) was 288 nm. Each sample was 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 two trans-isomer standards were partly converted to the two cis-isomers after UV-C irradiation for 30 min at 6 W m^-2 and a distance of 15 cm. Conversion coefficients were computed from the two trans-isomers; standard curves of the four isomers were generated. However, we only detected trans-res, trans-pd, and cis-pd in the samples. Therefore, the total content of resveratrol was obtained from the sum of these three isomers.

Anthocyanin relative concentration and composition were determined by HPLC as in Guan et al. (2012). The frozen samples were ground to a fine powder in liquid nitrogen using an A11 Basic Analytical Mill (IKA, Staufen, Germany). One gram of skin powder was extracted twice in 10 mL methanol (2% formic acid v/v) with shaking in darkness for 24 h at 4°C. Extracts were then centrifuged at 12000 × g for 15 min. The supernatant was concentrated under vacuum at 39°C using a rotary evaporator (RE-52AA; YaRong, Shanghai, China) until dryness. The dry residue was dissolved in 4 mL deionized water, and then 1 mL was passed through a 0.22 μm Millipore filter for HPLC analysis. Samples were analyzed using an HPLC fitted with a PDA detector (1290 series; Agilent, Palo Alto, CA, USA) with a reversed phase C18 column Kromasil-100 (5 μm particle size, 250 mm × 4.6 mm i.d.; Tracer Analitica, Barcelona, Spain) protected by a C18 Nova Pack guard precolumn (Waters, Milford, MA, USA). The separation used aqueous 2.5% formic acid (solvent A) and acetonitrile containing 2.5% formic acid (solvent B). The gradient was from 10 to 15.8% B for 7 min; from 15.8 to 17.3% B for 12 min; from 17.3 to 20% B for 3 min; from 20 to 22.3% B for 9 min; from 22.3 to 23% B for 9 min; isocratic 23% B for 20 min; and then returned to initial conditions for 4 min at a flow rate of 0.3 ml min^-1. Injection volumes were 5 μl and the column temperature was set at 30°C. Anthocyanins were detected by UV absorbance at 280 and 520 nm. The identification of anthocyanins was carried out by LC-MS/MS (Guan et al., 2012), supported by spectral analysis and comparison of the elution time and absorbance spectra with a commercial standard of malvidin-3-O-glucoside chloride (PhytoLab Gmbh, KG, Dutendorfer, Germany). The concentration of individual anthocyanins was expressed as malvidin-3-O-glucoside chloride equivalents calculated from the standard.

Total RNA was extracted using a General Plant Total RNA Extraction Kit (Bioteke, Beijing, China). Grape berry cDNA was prepared using the Reverse Transcription Kit (Fast quant RT with gDNase KR106, Tiangen, Beijing, China). Briefly, 2 μl of 5 × DNA Buffer was mixed with 13 μl of total RNA and incubated for 3 min at 42°C. This solution (15 μl) was mixed with 2 μl of 10× Fast RT Buffer, 1 μl RT Enzyme Mix, and 2 μl FQ-RT Primer Mix and incubated for 15 min at 42°C and for 3 min at 95°C to produce cDNA.

The expression levels of Actin. STS. 3-O-GT (3-O-β-glycosyltransferases), Myb14. CHS (chalcone synthase), UFGT (UDP-glucose: flavonoid 3-O-glucosyltransferase), MybA1. NCED1, and NCED2 were analyzed. The primers (Supplementary Table S1) were designed based on previous studies (Wang et al., 2013a; Zheng et al., 2013). The qRT-PCR system consisted of a Rotor-Gene 3000 Amp PCR system (Agilent Technologies); a real-time PCR Kit (Real master mix with SYBR Green FP202, Tiangen, Beijing, China) was used. The amplification of the actin rRNA gene was used as an internal control and for data normalization (Reid et al., 2006; Jarošová and Kundu, 2010; Wang et al., 2014).

Data were analyzed by ANOVA, Student–Newman–Keuls’s multiple range tests (at P < 0.05), and paired-sampled T-test (at P < 0.05 and P < 0.01) using the SPSS 13.0 program (SPSS, USA).

The total soluble solid of ‘Beihong’ berry increased from 5.8°Brix at veraison (70 DAA), to 25.5°Brix at 112 DAA, with no further increase to maturity at 126 DAA (Figure 1A). Free ABA concentration was low prior to veraison (Figure 1B), but increased to approximately 400 ng g^-1 fresh weight (FW) at veraison, and continued to increase to approximately 1400 ng g^-1 FW at 91 DAA. After this point, ABA concentration decreased to maturity, but remained at a higher level than pre-veraison. The concentration of anthocyanins was very low before veraison, increased rapidly from 10 μg g^-1 FW at veraison to 3000 μg g^-1 FW at 98 DAA, and then continued to increase to maturity at a very slow rate, reaching 3470 μg g^-1 FW (Figure 1C). Resveratrol concentration was low until 91 DAA (2 weeks after veraison), ranging from 5 to 8 μg g^-1 FW, but increased sharply to 35 μg g^-1 FW at maturity (Figure 1D).

We applied exogenous ABA to grape berries at 20, 10 days before veraison, at veraison and 7 days after veraison. The concentration of ABA in berry skin increased in each stage of development (Figures 2A1–A4), reaching as high as 16000 ng g^-1 FW (about 26-fold higher than the control) at 20 days pre-veraison application (Figure 2A1). It was notable that in all four treatments, peak ABA concentration was not reached until approximately 5–7 days after ABA was applied. After this time ABA concentration in berry skin declined irrespective of the timing of application, but remained higher than controls for at least 20 days or, in three of the four applications, until maturity. The concentration of ABA-GE was also increased after applying ABA in all four treatments, while the peak time was latter than that of ABA (Figures 2B1–B4). However, the change trend of PA (Figures 2C1–C4) and dehydro-phaseic acid (DPA; Figures 2D1–D4) concentration was very similar. In addition, their peak time was almost accordance to that of ABA concentration.

We also treated grape berries with fluridone at the same developmental stages as the exogenous ABA applications. This resulted in lower berry skin ABA concentration than control in each case, although the timing and duration of the difference varied according to the developmental stage (Figures 2A1–A4). When the fluridone treatment occurred at 20 days pre-veraison, where the ABA concentration of controls was very low, there was no significant difference until 28 days after application, coinciding with the increase in ABA associated with veraison, but the difference was maintained until maturity (Figure 2A1). The fluridone treatment at 10 days pre-veraison had a much more immediate effect on ABA compared with control, but was applied at a time when ABA in the control was already increasing (Figure 2A2). In contrast, when applied at veraison or 7 days post-veraison, the inhibitor had no significant influence on the concentration of ABA (Figures 2A3,A4).

In addition, we measured the expression of ABA biosynthesis genes (NCED1 and NCED2) in berry skins when exogenous ABA treatments were conducted at 20 days pre-veraison. At 5 days after ABA application, the expression of NCED2 in the treatment was higher than in the control; while at 7 days after ABA application, the expression of NCED1 in the treatment was higher than in the control. The expression of both genes then declined and remained at a very low level until maturity (Figure 3).

We further investigated the change in total anthocyanins concentration of the berry skins after ABA application. The concentration of total anthocyanins in berry skins exhibited a large increase in each case (Figure 4). Following the 20 days pre-veraison application, there was no significant difference in total anthocyanins concentration between the treatment and control from 1 to 14 days after the treatment. From 14 days after the treatment, total anthocyanins concentration in the treated skin increased until maturity. At 21 and 28 days after the treatment, total anthocyanins concentration in the treated skin was higher than in the control (Figure 4A). When exogenous ABA applications were conducted at 10 days pre-veraison, veraison and 7 days post-veraison, the skins of ABA-treated berries had significantly higher anthocyanin concentration than the control from 7 days (3 days for the application at version) to maturity after the treatment (Figures 4B–D).

Fluridone application at 20 days pre-veraison had no significant effect on anthocyanin concentration (Figure 4A). In contrast, fluridone delayed anthocyanin synthesis and accumulation in berry skin after application at 10 days pre-veraison, veraison or 7 days post-veraison (Figures 4B–D).

The results from the anthocyanin measurements were matched by visual assessment of veraison (Supplementary Figure S1). The photographs in Supplementary Figure S1 demonstrate that the application of ABA advanced coloring, and fluridone delayed coloring. Consequently, the onset of veraison of ‘Beihong’ berries was advanced by applying ABA at 20 and 10 days before veraison, while it was delayed by applying fluridone. The expression of genes related to anthocyanin biosynthesis was further investigated when exogenous ABA application or fluridone was conducted at 20 or 10 days pre-veraison. The expression levels of CHS. UFGT, and MybA1 related to anthocyanin biosynthesis in the skin of ABA-treated berries were higher than in the control. Expression of the above three genes was reduced by application of the inhibitor when the treatment was conducted at 10 days pre-veraison, but was not changed when the treatment was conducted at 20 days pre-veraison (Figure 5).

When exogenous ABA or fluridone treatments were conducted at 20 days pre-veraison or 7 days post-veraison, there were no significant differences in resveratrol concentration between the treatments and the controls (Figures 6A1–D1,A4–D4). However, significant effects of both ABA and fluridone application were observed when the application occurred 10 days pre-veraison, and a significant effect of ABA application was observed when the treatment occurred at veraison (Figures 6A2,A3).

In the case of ABA application, the increase in resveratrol occurred approximately 10 days after application, whether applied 10 days pre-veraison or at veraison, the difference remaining stable until maturity, i.e., the rate of resveratrol accumulation after this point was almost the same in ABA-treated or control berries (Figures 6A2,A3). Furthermore, the increase in total resveratrol was mainly due to the two piceids (Figures 6B2,B3,C2,C3), while trans-res remained at a low level, not significantly different from the control (Figures 6D2,D3).

Total resveratrol concentration in the skins of berries treated with fluridone at 10 days pre-veraison was not affected by the treatment on the same timescale as the effect of ABA application. The difference appeared to be in the rate of accumulation post-veraison, being significant from 28 days after the treatment onward (Figure 6A2).

We then investigated the expression of the genes related to resveratrol biosynthesis for the treated skins at 10 days pre-veraison (Figure 7). For the key gene of resveratrol biosynthesis, STS, the expression level in the ABA-treated berries reached a peak at 7 days after ABA application and was significantly higher than the control, then decreased to the level of the control at 21 days, remaining similar to controls until maturity. The expression level of STS in skins from ABA synthesis inhibitor treated berries was significantly lower than the control 7 days after application, but increased to the level of the control at 21 days (Figure 7). The expression of Myb14, an important regulatory gene of STS, was similar to STS following application of ABA or the inhibitor (Figure 7). The expression level of 3-O-GT in the berry skin, the key gene related to the biosynthesis of two piceids, was higher than the control after applying ABA, but was lower than the control after applying the inhibitor (Figure 7).