Leaf metabolomic data of eight sunflower lines and their sixteen hybrids under water deficit

Thierry Berton, Stéphane Bernillon , Olivier Fernandez, Harold Duruflé , Amélie Flandin, Cédric Cassan, Daniel Jacob , Nicolas B. Langlade , Yves Gibon and Annick Moing 1 INRAE, Univ. Bordeaux, Biologie du fruit et pathologie, UMR 1332, Centre INRAE de Nouvelle Aquitaine–Bordeaux, 33140 Villenave d’Ornon, France 2 Bordeaux Metabolome, MetaboHUB, PHENOME, IBVM, Centre INRAE de Nouvelle Aquitaine–Bordeaux, 33140 Villenave-d’Ornon, France 3 LIPME, Université de Toulouse, INRAE, CNRS, Castanet-Tolosan, France


Highlights
Leaf metabolomic data were produced on sunflower plants of inbred lines and their hybrids subjected to water deficit at the vegetative stage.
They differentiate both plant water status and different genotype groups.
They constitute a valuable resource to be combined with other omics data and study the adaptation to drought and the bases of heterosis.  (Debaeke et al., 2017). In the present experiment, plants were subjected to two treatments (Well-Watered or Water-Deficit) during the vegetative stage. This experiment was performed in the outdoor high-throughput, semi-automated phenotyping facility Heliaphen (https://www6.inrae.fr/phenotoul_eng/WHO-weare/PhenoToul/HeliaPhen).
Heterosis is an outstanding phenomenon involved in natural selection and used in crop breeding to adapt plants to environmental constraints. Twenty-four genotypes of cultivated sunflower consisting in four maintainer lines, four restorer lines and their 16 corresponding hybrids are included in this experiment which allows studying heterosis effect on metabolism.
This dataset provides metabolomic data of sunflower leaves of lines and hybrids under control and water deficit conditions. These data consist in unique untargeted and targeted metabolomic profiles of sunflower responses to drought based on a large genetic variability.

Data
Climate change is affecting plant biodiversity, and crop choice and yields. A better knowledge of plant adaptation mechanisms to this recent phenomenon is, therefore, of major interest for crop science, agriculture and for feed and food security (Porter et al., 2019). Helianthus annuus L., the domesticated sunflower, is the fourth most important oilseed crop in the world (USDA, 2019). It seems promising for agriculture adaptation to global change because it can maintain stable yields across a range of environmental conditions, especially during stress induced by water limitation (Debaeke et al., 2017). It can be considered as an archetypical systems biology model with large drought stress response which involves many molecular pathways (Moschen et al., 2017) and subsequent metabolic and physiological processes.
In this data article, we are sharing the metabolomic data of 24 sunflower genotypes grown in two environmental conditions in an outdoor phenotyping facility. This dataset is part of a larger project that integrates other omics data (Blanchet et al., 2018;Gody et al., 2020;Balliau et al., 2021).
The LC-MS data and metadata associated with this article were deposited in the Data INRAE repository. The targeted analyses data were deposited in the Data INRAE repository.

Experimental design, plant material and growth conditions
The experiment was performed from May to July 2013 on the outdoor Heliaphen phenotyping facility at the Institut National de Recherche pour l'Agriculture, l'Alimentation et l'Environnement (INRAE) station, Auzeville, France (43°31'41.8"N, 1°29'58.6"E) as previously described (Blanchet et al., 2018;Gosseau et al., 2019). Briefly, germinated plantlets were transplanted into individual pots filled with 15-l potting soil and covered with a 3-mm-thick polystyrene sheet to prevent soil water evaporation. Plants were fertilized with Peters Professional fertilizer (17-07-27, 500 mL, 0.6 g/L) and an oligo-element mixture solution (Hortilon, 0.46 g/L) at 17 days after germination (DAG), and treated with Polyaxe (5 mg/L applied on foliage) against thrips at 21 DAG.  (Blanchet et al., 2018;Gody et al., 2020;Balliau et al., 2021) In total, 144 plants, corresponding to 24 genotypes, four maintainer (SF009, SF092, SF109 and SF193) and four restorer (SF279, SF317, SF326 and SF342) lines and their corresponding hybrids obtained by crossing, were grown in two conditions: well-watered (WW) and water-deficit (WD) with three biological replicates (Blanchet et al., 2018;Gody et al., 2020). Before the beginning of the water deficit application at 35 DAG, pots were saturated with water and excessive water was drained. Pots were weighed to obtain the full soil water retention mass. At 38 DAG, irrigation was stopped (approximately 20-leaf stage) for WD plants as described previously (Gosseau et al., 2019). Plants were weighed by the Heliaphen robot to estimate transpiration (Gosseau et al., 2019). WW plants were re-watered at each weighing to reach soil water full retention capacity. Pairs of WD and WW plants were harvested when the fraction of transpirable soil water of the stressed plant reached 0.1 (occurring between 42 and 47 DAG). Two out of the three SF342 line plants died under the WW condition. The corresponding plant samples could not be harvested and data could not be obtained.
At harvest, leaves for metabolome analyses were cut without their petiole and immediately frozen in liquid nitrogen from 11 a.m. to 1 p.m. The harvested leaf was the leaf above the leaf that had reached its maximum size the most recently, as for the proteomic and transcriptomic studies (Blanchet et al., 2018;Gody et al., 2020;Balliau et al., 2021).

Metabolite extraction
Leaf sample grinding was performed using a ZM200 grinder (Retsch, Haan, Germany) as described for transcriptome analysis (Gody et al., 2020). Fresh-frozen powdered samples were then lyophilized. Aliquots of about 10 ± 2 mg of dry powder were weighed in 1.1 mL Micro-nicTM tubes (Lelystad, The Netherlands) and extracted with a robotized Star/Starlet platform (Hamilton, Villebon sur Yvette, France) using ethanol/water (80:20, v/v) added with 0.1% formic acid as solvent at room temperature. Methyl vanillate was used as internal standard to check for the quality of injection for LC-MS. Two successive extractions consisting in 1 min vigorous shaking followed by 15 min ultrasonication were performed with 300 mL of extraction solvent. The two supernatants were combined and filtered with 0.22 mm hydrophilic Durapore filtering microplates (Merck Millipore, Carrightwohill, Ireland). Nine blank extracts from the same procedure, but without sample powder, were also prepared. A QC sample was produced for LC-MS by pooling 10 mL of each sample extract.

Targeted analyses of major compounds
The targeted analyses of major compound in all samples were performed as done previously for the parents only (Fernandez et al., 2019) and as previously described (Biais et al., 2014) using enzymatic analyses and colorimetric assays performed using a robotic Star/Starlet platform (Hamilton, Villebon-sur-Yvette, France) and spectrophotometers.
Glucose, fructose and sucrose were determined in the ethanolic supernatant obtained as described above (Stitt et al., 1989) and expressed in mmol per g dry weight (DW). Total free amino acids were determined in the supernatant with a fluorescaminebased assay (Bantan-Polak et al., 2001) and expressed as glutamate equivalents. Protein content was determined (Bradford, 1976) on the pellet re-suspended in 100 mM NaOH and heated at 95°C for 20 min and expressed as mg bovine serum albumine equivalents per g DW. After neutralisation of the suspended pellet, starch was determined and expressed in glucose equivalents per g DW (Hendriks et al., 2003). Absorbencies were read at 340 or 595 nm using an MP96 microplate reader (SAFAS, Monaco). For fluorescence, 405 nm excitation and 485 nm emission were used with a Xenius multifunction microplate reader (SAFAS, Monaco). All chemicals and substrates for targeted analyses were purchased from Sigma-Aldrich Ltd. (Gillingham, United Kingdom). All enzymes were purchased from Roche Applied Science (Meylan, France).

LC-MS based metabolomic profiling
LC-MS-based metabolomic profiling of extracts was performed using the same extracts as for targeted analyses. The sample injection order was randomized. The QC sample was injected every 12 samples to correct for mass spectrometer signal drift. The extracts were analysed using LC-MS (Ultimate 3000 À LTQ-Orbitrap Elite, ThermoScientific, Bremen, Germany), using a C18 chromatographic column (C18-Gemini 2.0 Â 150 mm, 3 mm, 110 Å, Phenomenex, Torrance, CA, USA), a 18 min acetonitrile gradient in acidified water (solvent A: ultrapure water þ 0.1% formic acid, solvent B: LC-MS grade acetonitrile) with a 300 mL.min À1 flow rate and the following elution gradient: 0-0.5 min, 3% B; 0.5-1 min, 3-10% B; 1-9 min, 10-50% B; 9-13 min, 50-100% B; 13-14 min, 100% B; 14-14.5 min 100-3% B; 14.5-18 min, 3% B. The column temperature was 30°C. The injection volume was 5 mL. The LC-MS instrument was equipped with an HESI source operated in the positive-ion mode. Source parameters were the following: source voltage, 3.2 kV; sheath gas, 45 arbitrary units (a.u.); auxiliary gas, 15 a.u.; sweep gas, 0 a.u.; capillary temperature, 350°C; heater temperature, 350°C. Full Scan MS spectra were acquired at 240k resolution power with a 50-1000 mass range. Data dependent MS/MS spectra were acquired at 60k resolution power. The selected ions were fragmented in CID mode at a 35% normalized collision energy. The MS data were processed using R (R Core Team, 2018) with XCMS (Smith et al., 2006) and MetNormalizer (Shen et al., 2016) packages. Briefly, the corresponding MS-based variables were named using their nominal masses in Da and retention time in s (MxxxTyyy). Variables detected in blank extracts were filtered out. Variables with m/z values varying by more than 0.005 Da or with retention time varying by more than 20 s between different samples were also filtered out. Variables with intensity coefficients of variation in QC greater than 20% were also removed. This resulted in a data matrix of 4843 variables. Intensity drift was corrected using support vector regression. Finally, intensities were normalized according to the sample powder mass used for extraction. Annotation of intense ions Finally, due to plant death or the lack of leaf material for several plants, 121 and 125 samples out of the 144 initial ones were analysed by the targeted (6 variables) and LC-MS based metabolomic (4843 variables) approaches, respectively. To get an overview of each data set, a principal component analysis (PCA) was performed using BioStat-Flow web tool (Jacob et al., 2020) on data mean-centred and scaled to unit variance. The two treatments tended to separate along PC2 explaining about 28% of total variability for the targeted analyses (Fig. 1A) and about 9% for the LC-MS based data (Fig. 1B). The lines and hybrids tended to separate along PC1 explaining about 10% of total variability for the LC-MS based data (Fig. 1B). These metabolome data can be combined with other omic and phenotypic data of the same samples (Blanchet et al., 2018;Gody et al., 2020;Balliau et al., 2021) to get deeper insights into drought effects and heterosis.

Supplementary material
DATA-TargetedAnalyses-SunflowerLeaf.txt: This file contains targeted measurements of major compounds for each genotype and their three biological replicates (in columns) for WW and WD conditions. DATA-LCMS-SunflowerLeaf.txt: This file contains the intensities of LC-MS-Orbitrap metabolite signatures for each genotype and their three biological replicates (in columns) for WW and WD conditions. The Supplementary Material is available at http://www.ocljournal.org//10.1051/ocl/2021029/olm.