Field-grown grapevine (V. vinifera L.) canes were collected after fruit harvest from a commercial vineyard of the Controlled Appellation (DOC) region of Vinhos Verdes in the northwest region of Portugal (41°48′45.3″N 8°24′36.4″W) in 2019. Lignified shoots of similar diameter (~2 cm) were cut into single bud segments of approximately 15 cm, its basal ends were inserted into wet floral foam, and the top ends were sealed with paraffin film to prevent dehydration, as previously reported (Noronha et al., 2021). Samples were then incubated at 4°C for 7 (T1) and 14 days (T2). Shoots not subjected to cold were used immediately after their collection (T0; Figure 1). Control canes were collected in the same conditions, during 2021, and incubated for the same periods at room temperature (20°C). At each incubation period, three biological replicates (four similar sized canes per replicate) were sampled. The periderm of each sample was removed with a scalpel, frozen with liquid N₂ to be ground in a IKA A11 basic analytical mill and stored at −80°C. Part of these samples was freeze-dried in a Christ-Alpha 2–4 LD lyophilizer for quantification of raffinose, stachyose, and starch.

Gamay Fréaux cultured cells were grown in liquid Gamborg B5 supplemented with 2% (w/v) sucrose and .025% (w/v) casein hydrolysate and maintained in 250 ml flasks on a rotatory shaker at 100 rpm and 24°C. Cells were subcultured weekly by transferring 10 ml aliquots into 40 ml of fresh medium. To study the effect of different temperatures on VviRafs5 expression, 20 ml aliquots of 6-days-old cells were incubated for 24 h at 24 (control), 4, and 37°C. The effect of 100 mm NaCl and 100 μm ABA was tested in cells cultivated for 24 h at 24°C. Cells were collected by vacuum filtration, ground in liquid N₂, and stored at −80°C.

The extraction of sugars was performed as reported by Conde et al. (2015) adapted from Eyéghé-Bickong et al. (2012). In short, for sucrose, glucose, and fructose quantification, 150 mg of fresh weight (FW) cane tissue was mixed with 1 ml of ultrapure H₂O, vortexed, and mixed with 1 ml of chloroform. The same procedure was done with lyophilised material for raffinose and stachyose quantification. The biphasic solvent was vortexed for 5 min and incubated at 50°C for 30 min with continuous shaking (150 rpm). Samples were then centrifuged at 15000 × g for 10 min and the supernatant was collected and filtered with a .45 μm nitrocellulose filter. Sugars were quantified by HPLC-RI with a Rezex RCM–Monosaccharide Ca²⁺ (8%) column (Phenomenex) at a flow rate of .6 ml min⁻¹ at 40°C using water as the mobile phase. Twenty μl of each sample were injected and sugar concentration was measured after the comparison of the peak area and retention time with calibration curves.

Starch was quantified in canes according to Silva et al. (2017). To remove soluble sugars, lyophilized cane tissues (50 mg DW) were extracted three times with 5 ml 80% ethanol, and starch grains were gelatinized by autoclaving. Starch was enzymatically degraded by α-amylase (AMY; 1 U, Sigma–Aldrich) and β-glucosidase (10 U, Sigma–Aldrich) in a medium containing 200 mm sodium acetate (pH 5.5) and released sugars were measured with the DNS method.

To identify grapevine RFO gene family members, Arabidopsis thaliana sequences of GolS and RafS were blasted against the genome of V. vinifera L. (12X) using the online platform Phytozome.¹ Annotated genes were named following phylogenetic analysis using amino acid sequences from V. vinifera and Arabidopsis thaliana obtained from the National Center of Biotechnology (NCBI) and Uniprot (Supplementary Figure 2). Sequence alignment was performed with PRANKSTER (Whelan and Goldman, 2001) and Genedoc (Nicholas et al., 1997). The phylogenetic tree was constructed using these alignments with PROTDIST, NEIGHBOR, and RETREE from the PHYLIP software package (Felsenstein, 1989) and Mega 4 (Tamura et al., 2007).

Total RNA from cane tissues and cultured cells was isolated as previously described (Reid et al., 2006) with some adaptations. For each condition, 500 mg of frozen tissue was mixed with 1 ml of extraction buffer containing 300 mm Tris HCl (pH 8.0), 25 mm EDTA, 2.0 M NaCl, 2% CTAB, 2% PVP (K-30), and 30 mm DTT. Samples were then incubated at 60°C for 15 min and shaken every 5 min. Then, mixtures were extracted twice with 850 μl of chloroform: isoamyl alcohol (24:1) followed by centrifugation at 15000 × g for 15 min at 4°C. Subsequently, .1 vol of 3 M NaOAc (pH 5.2) and .6 vol of isopropanol were added to the aqueous phase, followed by incubation at −80°C for 1 h. The mixture was centrifuged at 15000 × g at 4°C during 30 min and later resuspended in 100 μl of ultrapure H₂O. To purify the samples, the GRS Total RNA Kit—Plant (GriSP, Lda.) was used following the manufacturer’s instructions. RNA concentration and purity were quantified spectrophotometrically in the NanoDrop ND-1000 (Thermo Fisher Scientific Inc.) and integrity checked in a 1% (w/v) agarose gel. First strand cDNA synthesis was performed using Xpert cDNA Synthesis Mastermix protocol (GriSP, Lda.), according to the manufacturer’s instructions.

Quantitative real-time PCRs were performed with Xpert Fast SYBR Blue (GriSP, Lda.) along with the conditions previously optimized in a CBX96 Real-Time Detection System (Bio-Rad). The amplification protocol included an initial denaturation step at 95°C for 3 min, followed by additional 40 cycles of denaturation for 3 s at 95°C, annealing for 20 s at 55°C, and 20 s at 72°C. Experiments were done in biological replicates and then interpreted with the software Bio-Rad CFX Manager (Bio-Rad), while VviGAPDH (glyceraldehyde-3-phosphate dehydrogenase) and VviACT1 were used as internal control. The primers used in this study are listed in Supplementary Table 1.

Due to experimental difficulties in using bacterial hosts, gap repair was used to clone VviRafs5 into the pYES-DEST52 vector (V5-epitope; Invitrogen) using Saccharomyces cerevisiae INVSc1 (Bessa et al., 2012). VviRafS5 CDS, without the stop codon (allowing the production of a VviRafS5-V5-epitope fusion protein), was amplified using primers listed on Supplementary Table 1 (VviRafS5-Gap-FW and VviRafS5-Gap-RV). Fragment amplification was done using NZYTaq II DNA polymerase (NZYTech, Lda), according to the manufacturer’s instructions. pYES-DEST52 (Invitrogen) vector was digested with EcoRI (Thermo Scientific) and SmaI (Thermo Scientific) and purified using GRISP PCR & Gel Band Purification (GriSP, Lda.). The transformation of INVSc1 yeast was carried out using the LiAc/SS-DNA/PEG method and linearized pYES-DEST52 and VviRafS5 PCR product (Gietz and Woods, 2002). As control, yeast cells were also transformed with the empty vector. Transformants were selected on solid synthetic medium: .7% YNB (Alfa Aesar), 1.3% dropout (US Biological; without uracil), .5% ammonium sulfate, 2% glucose, and 2% agar. Positive clones, confirmed by PCR, were cultivated overnight at 30°C in liquid synthetic medium with 2% (w/v) glucose and protein synthesis was induced by diluting cells to DO_640nm .4 in synthetic medium supplemented with 2% galactose and allowed to grow overnight at 30°C. Cells were then collected, washed twice with deionized water, and resuspended in extraction buffer (pH 7.5, 50 mm HEPES, 25 mm NaCl, 1 mm PMSF, 1 mm DTT, and 1 mm EDTA). Acid-washed glass beads were added and cell lysis was accomplished in the cell disruptor (Scientific) with 3 cycles of 30 s homogenization followed by 1 min on ice. The mixture was centrifuged at 16000 × g for 20 min at 4°C, the supernatant was collected, and sugars present in the crude protein extracts were removed using a Amicon Ultra-4 10,000 NMWL (Merck), following the manufacturer’s instructions. Protein extracts were used to study VviRafS5 activity.

Protein concentration was assessed with the Bradford assay. To functionally characterize VviRafS5 protein, galactinol and sucrose were used as substrates in saturating concentrations according to Li et al. (2020). The reaction was conducted in a 200 μl reaction system containing reaction buffer (50 mm HEPES; pH 7.5), 30 mm sucrose, 25 mm galactinol, and 50 μl of crude yeast extract. Reactions proceeded for 2 h at 37°C. Produced raffinose was measured by HPLC as described above.

Protein samples obtained as described above were separated on 10% acrylamide gels as described by Laemmli (1970). Proteins were transferred to a nitrocellulose membrane during 1.5 h at 100 V and were blocked during 1 h in TRIS-buffered saline containing .1% (v/v) Tween-20 (TBS-T) with 5% (w/v) skimmed milk powder. The membranes were probed against V5-tag (Anti-V5 antibody, 1:3000 dilution, Sigma) during 1 h at room temperature in blocking solution, followed by an incubation with an anti-rabbit peroxidase-conjugated antibody (Sigma) at 1:10000 dilution in TBS-T, for 45 min. The immunodetection was accomplished with the chemiluminescent ECL detection substrate (Bio-Rad) and observed in the ChemiDoc system (Bio-Rad).

ABA and ABA catabolites detection and quantification were performed as reported in Diretto et al. (2020) and Barja et al. (2021). Briefly, 50 mg freeze-dried grounded cane samples were extracted using unbuffered Tris-ethyl acetate as reported before (Welsch et al., 2008). LC-HRMS was carried out using an Ultimate UHPLC system with a photodiode array detector (Dionex), and a Q-exactive quadrupole Orbitrap mass spectrometry system (Thermo Fisher Scientific; LC-HRMS) equipped with an electrospray ionization (HESI) source, operating in negative ion mode, as previously described (Di Meo et al., 2019) with the following modifications: with nitrogen as sheath and auxiliary gas set at 35 and 25 units, respectively. The vaporizer and capillary temperatures were set at 280 and 320°C, respectively. The discharge current was set to 4.0 μA and S-lens RF level set at 50. Internal standard-based quantification was carried out using the MS data, while retention times and MS2 fragmentation patterns were used for ABA identification by using authentic reference standards (trans-ABA from OlChemIm and ()-ABA from Sigma, St. Louis, MO, United States).

The results were statistically analyzed by ANOVA tests (one-way and two-way ANOVA) using Prism vs. 7 (GraphPad Software, Inc.). Statistical differences between mean values are marked with letters for the different conditions or were marked with asterisks (^*p ≤ .05; ^**p < .01; ^***p < .001; and ^****p < .0001).

Results showed a steadily reduction in the amount of stored starch in woody canes after cold treatment, from 62.74 ± 5.90 at T0 to 48.74 ± 3.87 mg g DW⁻¹ at T2 (Figure 2A). In parallel to starch degradation, an increase in sucrose levels was observed, from 4.52 ± .40 to 7.44 ± .80 mg g FW⁻¹ at T2 (Figure 2B). Glucose and fructose also increased from 8.31 ± .51 and 5.73 ± .26, respectively, at T0 to 13.65 ± .83 and 9.99 ± .62 mg g FW⁻¹ at T2 (Figure 2A). Regarding RFOs, raffinose at T2 almost doubled its amount from T0, increasing from 2.70 ± .44 to 5.09 ± .80 mg g DW⁻¹, while a 5-fold increase in stachyose was observed, from .65 ± .02 at T0 to 3.40 ± .53 mg g DW⁻¹ T2 (Figure 2C). Canes incubated at room temperature showed no changes in raffinose and stachyose content (Figure 2, inset).

ABA and ABA-related metabolites were quantified in grapevine woody tissues by LC-HRMS. Results showed that ABA increased from T0 to T2 together with ABA-glucoside and dihydrophaseic acid, while phaseic acid decreased (Figure 3A). Accordingly, the expression of VviNCED2 and VviNCED3, which code for 9-cis-epoxycarotenoid dioxygenases and are considered major control points of ABA synthesis, was upregulated following cold treatment (Figure 3B).

Real-Time PCR analysis was performed to determine the expression of genes coding for enzymes responsible for the synthesis of RFOs during cold adaptation of grapevine woody tissues (Figure 4). Eight raffinose synthase genes, VviRafS1-8, and one stachyose synthase, VviStaS1, were identified (Supplementary Figure 3) and their expression in grapevine woody tissues following incubation at 4°C was analyzed (Figure 4B). VviRafS1, −3, and − 4 were not transcriptionally modified in response to cold and VviStaS1 and VviRafS2 and − 7 were not detected by qPCR in these tissues. On the contrary, the steady-state transcript levels of VviRafS6 increased by up to 6-fold compared to T0 the expression of VviRafS5 and − 8 was strongly upregulated up to 40-fold. Furthermore, results showed that cold treatment strongly upregulated the expression of VviGolS1 and VviGolS3, which are key enzymes in galactinol synthesis (Figure 4B).

VviRafS5 was cloned into S. cerevisiae strain INVSc1 and its ability to synthesize raffinose in vitro was studied by HPLC (Figure 5). Results showed that yeast protein extracts containing VviRafS5-V5-tag are able to synthesize raffinose in the presence of saturating galactinol and sucrose (Figures 5A,B). Also, the presence of VviRafS5-V5-tag in yeast protein extracts was confirmed by Western blot using an anti-V5 antibody, with a molecular weight of ~87 kDa (Figure 5C).

Because VviRafS5 was the most expressed raffinose synthase in woody tissues during cold treatment, its transcription was studied in different grapevine organs (canes, leaves, flowers, and roots) and berries at two development stages (green and mature). Results showed that VviRafS5 is mainly expressed in mature berries, but also in flowers and roots (Figure 6A). Following previous results showing that RafS is responsive to different abiotic stresses (Egert et al., 2013) and the identification of several cis-regulatory elements involved in abiotic stresses and ABA responses in VviRafS5 promotor (Figure 6B), cultured cells of cv. Gamay Fréaux were subjected to different temperatures and elicited with salt and ABA to check their effect on the expression of VviRafS5. In agreement with previous data, the steady-state transcript levels of VviRafS5 resulted higher after cold (2.4-fold) and ABA treatments (1.7-fold), and lower (.2-fold) after heat treatment. On the contrary, salt stress did not affect significantly the expression of VviRafS5 (Figure 6C).