VOCs Are Relevant Biomarkers of Elicitor-Induced Defences in Grapevine

Grapevine is susceptible to fungal diseases generally controlled by numerous chemical fungicides. Elicitors of plant defence are a way of reducing the use of these chemicals, but still provide inconsistent efficiency. Easy-to-analyse markers of grapevine responses to elicitors are needed to determine the best conditions for their efficiency and position them in protection strategies. We previously reported that the elicitor sulphated laminarin induced the emission of volatile organic compounds (VOCs) by grapevine leaves. The present study was conducted to characterise and compare VOC emissions in response to other elicitors. Bastid® was first used to test the conditions of VOC collection and analysis. Using SBSE-GC-MS, we detected several VOCs, including the sesquiterpene α-farnesene, in a time-dependent manner. This was correlated with the induction of farnesene synthase gene expression, in parallel with stilbene synthesis (another defence response), and associated to resistance against downy mildew. The other elicitors (Redeli®, Romeo®, Bion®, chitosan, and an oligogalacturonide) induced VOC emission, but with qualitative and quantitative differences. VOC emission thus constitutes a response of grapevine to elicitors of various chemical structures. Therefore, VOC analysis is relevant for studying the impact of environmental factors on grapevine defence responses and optimising the performance of elicitors in vineyards.


Introduction
Grapevine (Vitis vinifera L.) is an economically important crop highly susceptible to fungal diseases such as downy mildew and powdery mildew, caused by the oomycete Plasmopara viticola (Pv) and the ascomycete Erysiphe necator, respectively. Controlling these diseases mainly relies on the application of chemical fungicides. However, the extensive use of these pesticides raises environmental and health issues. Elicitor-induced resistance (IR) of plants against diseases is an attractive strategy for reducing the use of chemical fungicides. Elicitors trigger plant defence reactions, resulting in antimicrobial production or/and cell-wall reinforcement that stop or limit pathogen infection [1]. Elicitors are compounds of various biochemical classes, including lipids, proteins, and carbohydrates. Some of them are microbe-or damage-associated molecular patterns (MAMPs and DAMPs),

VOC Sampling Method
We first focused on the method of VOC collection with a view to transferring it to the vineyard in the future. In previous studies [21], VOCs were collected from vines individually enclosed in bags using fibres coated with divinylbenzene (DVB), carboxen (CAR), and polydimethylsiloxane (PDMS). Solid-phase microextraction (SPME) has proved to be a useful tool to collect plant VOCs [22]. It uses a fused silica fibre coated with a polymeric material to adsorb and pre-concentrate volatile compounds. However, SPME lacks sensitivity because of the low amount of polymeric sorbent available on fibre. Moreover, fibres are fragile and can break easily, so they are difficult to use in the field. Therefore, we also assessed Twister TM , another sensor of the Stir Bar Sorptive Extraction (SBSE) type. Twisters TM are magnetic stir rods coated with a polymeric material. They have a more significant sorbent volume than SPME fibre has. Like SPME fibres, SBSE can be used to pre-concentrate and adsorb volatile compounds in gas samples. In the present study, an SPME fibre and a stir bar were both placed in the foliage of a same plant in order to compare the nature and the intensity of the collected VOCs. The experiment was performed using six plants individually wrapped in a bag and elicited with sulphated laminarin (PS3, 2.5 g/L), as previously described by Chalal, et al. [21]. VOCs were collected 3 days (D3) after elicitation.
The concentration (µg/L) of detected volatiles was determined using external calibration with monoterpene, sesquiterpene, and green leaf volatile (GLV). These compounds were identified against the NIST-14 library and the retention times of the corresponding standards. As previously reported, PS3 induced VOC emission in grapevine [21]. Targeted analysis of eight VOCS showed that SBSE led to a better signal response for hexenyl acetate, β-caryophyllene, β-ocimene, and α-farnesene, while α-pinene was better detected by SPME ( Figure 1). SPME fibre lacks sensitivity for oxygenated and linear terpenes (linalool, β-ocimene), sesquiterpenes (β-caryophyllene, α-farnesene) and GLVs (hexenyl acetate, methyl salicylate). Therefore, SBSE stir bars, even coated with a non-polar adsorbent (PDMS), could be considered a more appropriate sensor for the sampling of VOCs of large polarity and chemical nature. Figure 1. Comparison of VOC signal responses from elicitor-treated grapevine leaves, detected by SPME fibre and SBSE stir bar in bag containment. Amounts (µg/L) of eight targeted VOCs collected for 4 h on D3 after sulphated laminarin (PS3, 2.5 g/L) treatment of four plants (A to D) individually placed in a bag, using SPME fibres and SBSE stir bars (Twisters TM ) as sensors.
We also assessed another mode of bagging plants for VOC trapping. In previous experiments [21], grapevines were individually enclosed in bags. However, the inner side of the bag was quickly covered with mist due to plant transpiration. The level of the analysed VOC was also quite variable, maybe due to physical stress related to the bag and/or variability of the response among plants ( Figure 1). For these reasons, we analysed VOC emission by plants grouped under a tent and compared it with that of plants enclosed individually, following water treatment. Sampling was performed on day 1 (D1), day 3 (D3), and day 5 (D5) after water treatment. The comparison of VOC emission showed huge differences between the two collection systems, supporting our hypothesis of a stress related to individual bagging. We indeed observed a higher number and diversity of water-induced VOCs in the bag containment mode ( Figure 2). In view of these results, the tent mode was retained for the next experiments because it appeared less stressful for grapevine.

SBSE-GC-MS Analysis of VOC Emission Elicited by Bastid ®
We first assessed VOC production induced by the marketed Bastid ® elicitor. Bastid ® is a mixture of chitooligosaccharides (COS) and oligogalacturonides (OGA) as active components, in which OGA dimers formed in the presence of calcium ions are stabilised by COS. This complex exhibits higher eliciting activity in Arabidopsis cell suspensions than the individual components do [23]. It triggers plant defence in tomato [24] and potato [25], and controls diseases of a number of crops. It is marketed for control of grape powdery mildew and downy mildew.
Bastid ® systematically induced VOC emission by grapevine leaves in our experimental conditions, although the time course of emission, the nature, and the quantity of the significantly emitted compounds varied. α-farnesene was highlighted as systematically induced by Bastid ® .

Comparison of VOC Emission Elicited by Other Elicitors
In addition to Bastid ® , two other commercial elicitors-Redeli ® and Romeo ® -were tested on grapevine plants. VOC analysis was performed, and the results were compared. Redeli ® is a phosphonate-based product with anti-oomycete and elicitor activities, used to control grapevine downy mildew [29,30]. Romeo ® is made up of cerevisiane extracted from yeast cell walls [31] and is marketed to control grapevine diseases like downy mildew, powdery mildew, and grey mould.
All three elicitors induced VOC emission, although with differences in the number and nature of the VOCs. The time course of VOC emission was also variable. Elicitorinduced VOC release has been reported in a limited number of studies [16,19,20], but to our knowledge a comparative study like ours has never been performed. Moreover, the capacity of a yeast extract and of a phosphonate to enhance grapevine VOC emission is demonstrated for the first time.  In order to complete this study, VOCs were analysed in response to chitosan and an oligogalacturonide (i.e., COS-OGA, made up of such compounds), as well as Bion ® . Chitosan and oligogalacturonides are known elicitors of grapevine defence reactions [11,[32][33][34]. Their activity depends on their degree of polymerisation and degree of deacetylation for chitosan [11,32,35]. Bion ® is also a well-known elicitor of defence reactions whose active ingredient is acibenzolar-S-methyl or benzothiadiazole (BTH), a hormone-like analogue of salicylic acid. BTH has been reported as an elicitor of grapevine defence reactions [29]. These three elicitors also induced VOC emission with different profiles (Table S1). However, a lower number of VOCs was detected following chitosan treatment, and to a lesser extent following Bion ® treatment. Few reports have addressed the effect of chitosan on VOC emission by plants. It can induce the release of various VOCs, including mono-and sesquiterpenes in tomato [19] and in rice [20]. Bion ® was previously shown to induce VOC emission in bean, and this property was associated with enhanced resistance to a bacterial pathogen [36]. Four VOCs have been recurrently induced by the three elicitors, namely α-farnesene, β-caryophyllene, 2-oxo-4-phenyl-6-(4-chlorophenyl)-1,2-dihydropyrimidine, and toluene. The presence of 2-oxo-4-phenyl-6-(4-chlorophenyl)-1,2-dihydropyrimidine, induced by an elicitor prepared in water (no formulation), rules out the hypothesis of its presence in commercial mixtures. The molecule oxime-, methoxy-phenyl-was detected in the water blank experiment in the absence of plants (data not shown), suggesting that it was present in the greenhouse ambient air. This VOC is listed among the molecules increased in fermentation processes of fish and wine [37,38] or found naturally in bamboo shoots [39], Pseudomonas aeruginosa [40], and myxobacteria [41]. As for 1,3-benzenedicarboxylic acid, 5-(dimethylamino)-, its presence is hardly reported.
VOC emission is therefore consistently induced by elicitors, with different blends despite a relevant core of molecules including α-farnesene. It is now well established that VOCs belong to the defence repertoire of plants against pathogens, especially pests such as caterpillars. In Arabidopsis, β-caryophyllene and GLVs also directly inhibit bacterial growth [42]. The release of GLV 4-hexen-ol-acetate was triggered by insect infestation on maple [27]. Together with oxime-, methoxyphenyl-production, 4-hexen-ol-acetate production can be induced by aphid attacks on apple [26]. VOCs can prime defence reac-tions like airborne signals [15][16][17] or act directly on targets thanks to their antimicrobial properties [42]. In grapevine, increased VOC emission has been reported in the course of interactions with insects. For instance, the feeding of the Japanese beetle on leaves of a Vitis labrusca cv. resulted in higher abundance of the GLV MeSA and terpenes, including β-ocimene and α-farnesene [43]. Moreover, the two latter VOCs were the only ones significantly increased following feeding and oviposition of a xylem-feeding sharpshooter on Chardonnay cultivar [44]. Three recent reports also support the role of VOCs in grapevine defence against P. viticola. Resistant cultivars released more mono-and sesquiterpenes than a susceptible one did following P. viticola inoculation [45]. Then, higher emission of more than 50 VOCs, including mono-and sesquiterpenes, was reported in similar experiments with resistant cultivars [18]. Finally, the role of VOCs in the resistance of the Georgian cultivar Mgaloblishvili and the hybrid Bianca to downy mildew was recently confirmed and suggested antisporulant activity of four terpenes (farnesene, nenolidol, ocimene, and valencene) [45]. Therefore, treating susceptible cultivars with elicitors could mimic the VOC response of resistant cultivars or hybrids to P. viticola infection.

Elicitor-Induced Resistance Against Grapevine Downy Mildew
In parallel to VOC analyses, we assessed the efficiency of the elicitors in the induction of leaf resistance against P. viticola. All of them reduced the disease (Figure 4). The highest levels of protection were obtained with chitosan (100%) and Bastid ® (98%), followed by Romeo ® (89.9%) and Redeli ® (89.5%). The oligogalacturonide (OG) turned out to induce the lowest resistance, probably because it has no direct antimicrobial effect; moreover, it was diluted in water in the absence of a surfactant to promote its penetration into the leaves, and in turn elicitor activity [46]. . Induced resistance assays. Six elicitors were tested for their efficiency in protecting grapevine against downy mildew (P. viticola). Bastid ® (0.8 mL/L, 4 assays), Romeo ® (2 g/L, 2 assays), Redeli ® (1 g/L, 2 assays), OG (2.5 g/L, 2 assays), and chitosan (1 g/L) and Bion ® (0.5 g/L, 1 assay). Protection levels (%) were calculated based on comparisons of the sporulating leaf area with the H 2 O control considered as 0% protection.

Analysis of Stilbenes in Bastid ® -and Chitosan-Treated Grapevine Plants
Stilbene phytoalexins were analysed as a reference, as markers of defence elicitation and resistance to diseases [47]. The test was performed with Bastid ® , used as a reference elicitor in this study, and also with chitosan because it showed the highest level of protection against downy mildew (Figure 4). Four stilbenes, i.e., piceid (trans-and cisisomers) and resveratrol (trans-and cisisomers) were analysed. Both elicitors induced the accumulation of these compounds ( Figure 5). In the water control, the concentrations of resveratrol (0.39 to 1.60 µg/g DW) and piceid (3.40 to 14.38 µg/g DW) were low throughout the 8 days of the experiment compared to those measured after the two elicitor treatments (up to 7.39 and 75.29 µg/g DW for the resveratrol and piceid isomers, respectively). After chitosan treatment, the piceid and resveratrol isomers were only significantly accumulated on D8-3and 13-fold more than in the H 2 O control, respectively. Bastid ® promoted the accumulation of both stilbenes from D2 till D8. It induced higher levels of stilbenes than chitosan did, particularly on D6. Both elicitors triggered the accumulation of phytoalexins. Bastid ® induced earlier and higher phytoalexin accumulation than chitosan did, and also higher VOC emission. This observation suggests that the protection of grapevine against downy mildew provided by Bastid ® and chitosan involves different modes of action. For Bastid ® , the induction of phytoalexin synthesis probably plays a key role. Chitosan has long been studied as an elicitor of plant defence [48]. It is very effective to reduce grapevine downy mildew severity in controlled conditions probably thanks to both its direct antimicrobial and indirect elicitor effects. The present study shows that it also induces the emission of terpenes. It may be tempting to conclude that chitosan is a less effective VOC inducer than Bastid ® (Table S1). However, it should be noted that the tent experiment was only performed twice.

Conclusions
This study shows that VOC emission can be considered as a universal response of grapevine to elicitors of plant defence, even though no specific marker can as yet be highlighted. The elicitors assessed in the present study induced the emission of a common set of VOCs encompassing a dozen chemically different compounds, including the sesquiterpenes α-farnesene and β-caryophyllene. Such a response is concomitant with the induction of stilbene phytoalexins; however, it cannot be directly linked to the level of protection against downy mildew. It rather suggests a combined action of several responses to build an effective defence system against pathogens. Those first results open onto prospects for VOC analysis in field experiments. This could be relatively tricky because of the atmospheric dilution effect and the impact of multiple factors (i.e., plant development and climate) likely to occur, but would also help to identify the critical factors that influence elicitor-induced defence. Such information will be crucial to determine the optimal conditions for efficient elicitor application. Compared to the analysis of other defence responses, VOC analysis has the advantage of being a non-destructive method that could be applied in situ (portable GC-MS apparatus). VOC measurements in the field will also be of interest for chemical ecologists and plant biologists. VOC monitoring indeed contributes to evaluate the impact of environmental stresses and could be useful for improving crop productivity and sustainable defence strategies.

Elicitor Treatments
Six elicitors were used: Bastid ® , an oligogalacturonide (OG), chitosan, Redeli ® , Romeo ® , and Bion ® (Table 3). On day 0 (D0), the elicitor solutions were applied as a foliar spray onto the whole plants using a hand-held sprayer ( Figure 6A). The products and the doses are detailed in Table 3. Ultrapure H 2 O treatment was systematically used as a negative control in each experiment. The experiments were replicated once for Bion ® ; twice for chitosan; three times for Romeo ® , Redeli ® , and OG; and five times for Bastid ® . Table 3. Elicitors and doses used in the study. DP: degree of polymerisation.

Induced Resistance Assays
Plants were inoculated with Pv 2 days post treatment (dpt) (D2) by spraying a suspension of sporangia at a concentration of 10 4 /mL onto the lower leaf surfaces, as previously described by Trouvelot, et al. [49] (Figure 4). Five days after inoculation (D7), foliar disks were punched out from two leaves of the treated plants and transferred onto humid Whatman paper in translucid Perspex boxes to induce sporulation. The percentage of sporulating area of each disc was determined using image analysis (Visilog), as previously described by Kim Khiook, et al. [50]. Elicitor efficacy was calculated by comparing the sporulating area obtained after treatment versus the H 2 O control considered as 0% of protection efficacy.

Sensors
Two types of sensors were tested: i-fibres coated with divinylbenzene (DVB), carboxen (CAR), and polydimethylsiloxane (PDMS); StableFlex 1 cm-50/30 µm, (Supelco Inc., Bellefonte, PA, USA), and ii-2-cm-long Stir Bar Sorptive Extraction (SBSE) (Gerstel GmbH, Co.KG, Mülheim an der Ruhr, Germany) coated with 1-mm-thick polydimethylsiloxane (PDMS). The stability of the collected volatiles on the stir bar was previously tested under various transportation temperatures and storage conditions until analysis (data not shown). Finally, stir bars were stored at room temperature, sent for analysis 2 weeks after VOC collection at most, and immediately analysed by GC-MS. Before sampling, each stir bar was conditioned in the injector of a gas chromatographer under a stream of helium according to the manufacturer's recommendations.

Containments
Two types of containments were assessed for VOC collection: i-with plants enclosed individually in a commercial Teflon ® oven bag (Albal ® , Cofresco Frischhalteprodukte GmbH, Co.KG, Minden, Germany), or ii-grouped in a tent built with the same material on all six cube faces, as illustrated in Figure 6B. When the individual bag setup was employed, each plant was placed in the bag together with a single stir bar, and four plants were used per condition. In the tent device, six plants under the same treatment were gathered with two or three stir bars. The stir bar was magnetically held within the upper part of the foliage using a metal paper clip. The plants were bagged or placed under a tent together with the stir bars for 4 h, from 10:00 a.m. to 2:00 p.m., 1, 3, and 5 dpt (from D1 to D5) ( Figure 6A). At the end of the experiment, the leaf areas were measured with an area meter  to normalise VOC abundance measured by GC-MS with the leaf emission surface.
The blank experiment using water and Bastid ® sprayed on aluminium foil was performed with plant-less pots, wood stakes, and metal paper clips (data not shown).

Chemicals
α-pinene, camphene, β-pinene, D-limonene, cis-β-ocimene, trans-β-ocimene, γ-terpinene, α-terpinolene, (Z)-3-hexenylacetate, (E)-3-hexenylacetate, linalool, methyl salicylate, longicyclene, β-caryophyllene, α-humulene, valencene, (E-E)-α-farnesene, and α-sabinene were purchased from Merck (Isle d'Abeau Chesnes, Saint Quentin Fallavier, France). Stock solutions of each standard were prepared at 10 g/L in ethanol (ethanol absolute, VWR, Fontenay sous Bois, France) and kept at 4 • C. A stock solution of a mixture of all standards was prepared at 100 mg/L in ethanol and stored at 4 • C up to 6 months. For analysis, the stock solution was diluted successively to 1 mg/L in water (LC-MS grade water, Fisher Scientific, Illkirch, France) and then to 50 µg/L. An external calibration was performed with monoterpenes, sesquiterpenes, and green leaf volatile standards prepared in water in the 12.5 µg/L to 1000 µg/L concentration range. The final standard mixture solution was placed into a 20-mL headspace vial, and a stir bar was exposed in the headspace of the vial for 1 h at 50 • C. The VOCs collected on the stir bar were desorbed and analysed according to the method described below. Standards and parameters are listed in Supplemental Table S2.

VOC Analysis by GC-MS
Analyses were carried on an Agilent 7890 Chromatograph hyphenated to a triple quadrupole mass spectrometer (7000C, Agilent Technologies, Les Ulis, France) equipped with an MPS robotic, thermal desorption unit (TDU 2) and a programmed temperature vaporiser inlet (PTV) from Gerstel (Gerstel GmbH, Co.KG, Mülheim an der Ruhr, Germany). After VOC sampling, the stir bars were kept at room temperature in their storage vials until they were placed in the thermal desorption unit (Gerstel, TDU 2). The stir bars were desorbed in splitless mode under a 50 mL/min helium flow at 270 • C for 3 min. Analytes were cryogenically trapped in the CIS 4 (Gerstel) inlet at 150 • C, on a baffled liner. Once the desorption step was completed, analytes were transferred on a 30-m length, 0.250-mm internal diameter, 0.25-µm film thickness GC column, (RTX-5MS Restek, Bellefontaine, PA, USA) in splitless mode by heating the CIS rapidly to 270 • C, with a hold time of 2 min. Analyses were performed under a constant helium flow of 1 mL/min. The GC oven temperature was held at 30 • C, then ramped up to 250 • C at a rate of 5 • C/min. The transfer line temperature was maintained at 280 • C. The analytes were ionised with an electron ionisation source at 70 eV maintained at 230 • C. Data were acquired on the MS1 quadrupole held at 150 • C using the scan mode.

GC-MS Data Processing and Statistical Analysis
The acquired data were integrated and deconvoluted with Unknowns Analysis software, version B.08.00 (Agilent). The peaks detected after deconvolution were identified against two libraries-the NIST 14 library and a home-made one. The retention times of the peaks were checked against those of a set of standards analysed under the same conditions. The peaks due to PDMS coating were removed. The peaks were aligned according to their retention times using RStudio software before statistical treatment. A principal component analysis (PCA) of the log2-transformed mean values (RStudio) was performed. Significant compounds were determined by ANOVA multivariate statistical analysis (p < 0.05) using Perseus software (https://maxquant.net/perseus/, accessed on 6 April 2021). In addition to significance criteria, the compounds emitted by the elicited plants with a fold-change (FC) > 1.5 compared with the H 2 O treatment were also retained when induction was observed in at least two of the three replicates. For the elicitor-VOC list comparisons, only the FC > 1.5 obtained at least twice for one criteria (one elicitor or one time-point) was used ( Table 2). For commonly used elicitor-VOC determination (Venn diagram, Figure 3), only induced VOCs (FC > 1.5) observed twice per elicitor treatment were compared.

Analysis of Farnesene Synthase (VvFAR) Gene Expression
Total RNAs were extracted from leaf samples from three plants collected 1 and 4 dpt (on D1 and D4). Samples were ground in liquid nitrogen, and extraction was performed with PureLink Plant RNA Reagent Extraction Buffer (Invitrogen, Thermofisher, Carlsbad, CA, USA) according to the manufacturer's instructions. In order to eliminate residual genomic DNA, the samples were treated with a DNA-free DNAse removal kit (Invitrogen, Thermofisher, Carlsbad, CA, USA). Complementary DNA was synthesized from total RNAs with a Superscript III Reverse Transcriptase kit (Invitrogen™, Life Technologies, Saint Aubin, France). Real-time quantitative PCR was performed using an ABsolute SYBR-Green Low ROX qPCR Mix Kit (Thermo Scientific, Waltham, MA, USA) with an Applied Biosystems ® ViiA 7 apparatus (Life Technologies, Inc., Applied Biosystems, Foster City, CA, USA). The wells were loaded with the Tecan Freedom EVO 8 capillary robot on 384-well plates. Relative expression of the genes, with EF1-and 60S as internal reference genes, was determined using the 2 −∆∆Ct method [51], with the water-treated sample as the negative control. Primers of the VvFAR gene coding for the (E,E)-α-farnesene synthase family (NCBI entry: GSVIVT01006399001 XM_02264098.4) were used, with the following sequences: AATATGGAGATGGGCTTGGA (forward primer) and CATGGACTTCCCAAGTACCC (reverse primer).

Stilbene Analysis
Leaves were harvested at five times points from D1 to D8 ( Figure 6A). They were ground in liquid nitrogen to a fine powder that was freeze-dried, and 100 mg were extracted with 8 mL of methanol overnight at 4 • C. Six mL of supernatant were collected, evaporated, and the pellet was dissolved in 1 mL of methanol/water (70/30, v/v). The leaf extracts were purified to remove chlorophylls on a Supelclean LC-18 solid phase extraction column (Supelco Inc., Bellefonte, PA, USA). Eluates obtained with methanol/water (90/10, v/v) were evaporated and resuspended in 600 µL of methanol/water (50/50, v/v). The phenolic extracts were diluted, centrifuged at 13,500 rpm for 10 min, and stored at −20 • C until use.
Supplementary Materials: The following are available online. Figure S1: Time-dependent VOC emission. Table S1: VOCs induced by Bastid ® , chitosan, an oligogalacturonide (OG), and Bion ® . Table S2: Parameters of the VOC calibration standards. Table S3: Parameters of the stilbene analysis method.