The application of a biostimulant based on tannins affects root architecture and improves tolerance to salinity in tomato plants

Roots have important roles for plants to withstand adverse environmental conditions, including salt stress. Biostimulant application was shown to enhance plant resilience towards abiotic stresses. Here, we studied the effect of a tannin-based biostimulant on tomato (Solanum lycopersicum L.) grown under salt stress conditions. We investigated the related changes at both root architecture (via imaging and biometric analysis) and gene expression (RNA-Seq/qPCR) levels. Moreover, in order to identify the main compounds potentially involved in the observed effects, the chemical composition of the biostimulant was evaluated by UV/Vis and HPLC-ESI-Orbitrap analysis. Sixteen compounds, known to be involved in root development and having a potential antioxidant properties were identified. Significant increase of root weight (+ 24%) and length (+ 23%) was observed when the plants were grown under salt stress and treated with the biostimulant. Moreover, transcriptome analysis revealed that the application of the biostimulant upregulated 285 genes, most of which correlated to root development and salt stress tolerance. The 171 downregulated genes were mainly involved in nutrient uptake. These data demonstrated that the biostimulant is able not only to restore root growth in salty soils, but also to provide the adequate plant nourishment by regulating the expression of essential transcription factors and stress responsive genes.

www.nature.com/scientificreports/ for farmers. Among the new generation products available on the market, biostimulants could be used for this purpose. They are defined by the new Regulation (EU) 2019/1009 as "products stimulating plant nutrition processes independently of the product's nutrient content with the sole aim of improving one or more of the following characteristics of the plant or the plant rhizosphere: nutrient use efficiency, tolerance to abiotic stress, quality traits, availability of confined nutrients in soil or rhizosphere" 9 . In general, biostimulants are not involved in a direct action against pathogens and biotic stress, but they are largely employed to enhance plant performance, especially under environmental adverse conditions. In this study, a biostimulant based on a mixture of hydrolysable and condensed tannins derived from waste industry (VIVEMA TWIN) was tested on tomato (Solanum lycopersicum L.) plants under optimal and salt stress conditions. S. lycopersicum is an established crop model system with global economical relevance 10,11 . Moreover, the importance of this crop as food source, the easy greenhouse cultivation, and the large amount of data present in literature, make tomato a good crop model system.
Tannins are water-soluble and condensed phenolic compounds of variable size widely distributed in the plant kingdom 12 . In addition to their use in leather treatment, as textile dyes and coagulant agents in rubber production 13 , tannins have been also shown to have several applications in animal and human nutrition, due to their antioxidant properties and their ability to bind and precipitate proteins 14 . For these reasons they could be considered as a good green alternative not only in food and nutraceutical industry but also in other fields 15 . On the other hand, the use of tannins in agricultural practices and in plant nutrition is still poorly explored. Few studies on the effect of tannins on plant performance have been reported in literature 16 . In particular, it has been suggested that gallic acid (GA), a compound found in high quantity in tannin rich mixtures, has an effect on root development. Its effects were shown to be mainly related to its antioxidant and auxin-like activity 17 , and were reflected in an increase in growth and antioxidant defenses in gallic acid-treated roots of plants subjected to salt stress 18 .
In this work, the chemical composition of the biostimulant VIVEMA TWIN was first investigated by UV/Vis, and then via HPLC-ESI-Orbitrap MS analysis. In addition, the reducing and radical scavenging properties of the biostimulant were characterized spectrophotometrically. Finally, the mechanism of action of the biostimulant was studied in tomato at morphological and biomolecular level. Morphological estimation included the evaluation of the effect of VIVEMA TWIN on S. lycopersicum plants during short-term and long-term salt stress, by imaging analysis and monitoring of biometric parameters. The molecular targets of this innovative biostimulant were identified by RNA-Seq analysis, and the expression level of the most representative genes was confirmed by qPCR.

Results and discussion
HPLC-ESI-FTMS revealed that the main consistituents of VIVEMA TWIN are phenolic compounds belonging to the tannin family. Due to the different origin of biostimulants, their composition displays significant variations in the chemical profile, not only from a quantitative but mainly from a qualitative point of view 19 . Indeed, biostimulant composition may range from single compounds to complex combinations of bioactive components, only partly characterized 20 . However, the biological action of biostimulants on plants depends on the complexity of these matrices, and for this reason their characterization represents a challenge that needs to be addressed.
In this work, a preliminary spectrophotometric characterization of the VIVEMA TWIN biostimulant was carried out in order to perform an initial chemical screening useful for subsequent analyses with HPLC-ESI-FTMS. The spectrophotometric determination included the quantification of the total amount of polyphenols (TPC), anthocyanins (TAnthC) and flavan-3-ols (TF3C) estimated by Folin-Ciocalteu, pH differential and DMAC methods, respectively. Results of the UV/Vis quantification are shown in Table 1.
Interestingly, VIVEMA TWIN showed a very high TPC value when compared to the plant raw materials listed in the top 100 highly enriched in polyphenol content 21 . Polyphenols are characterized by the presence in their structure of one or more phenolic groups, able to accept electrons and reduce not only reactive oxygen species, but also organic substrates and minerals 22 . The strong redox properties of polyphenols explain the considerable interest in human and plant nutrition in order to prevent conditions associated with excessive oxidative stress 23 . Among polyphenols, several classes may be identified, mainly because of the close structural similarity and only Table 1. UV/Vis spectrophotometric determination of bioactive compounds and antioxidant properties of VIVEMA TWIN. Total polyphenol content (TPC), total flavan-3-ol content (TF3C), total anthocyanin content (TAnthC), radical scavenging activity (ABTS and DPPH) and ferric reducing antioxidant power (FRAP) of VIVEMA TWIN. Values are expressed as a mean ± SD of three experiments carried out in triplicate. GAE gallic acid equivalents, PAC-A A-type proanthocyanidin equivalent, CyE cyanidin-6-glucoside equivalents, TE trolox equivalents, ABTS 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid), DPPH 2,2-diphenyl-1picrylhydrazyl, FRAP ferric reducing antioxidant power. www.nature.com/scientificreports/ few assays are able to detect and quantify these molecules selectively. In this context, the pH differential method is able to find the presence of anthocyanin compounds thanks to their characteristics to display different color depending of the pH of the environment mixture 14,24 . The 4-(Dimethylamino)-cinnamaldehyde (DMAC reagent) can selectively react with flavonol compounds with free meta-oriented hydroxyl groups in the flavonoid scaffold, and with a single bond in 2,3-position of the C-ring 14,24,25 . The spectrophotometric determination obtained through these assays showed that, even though no anthocyanins were present (LOD: 3 μg mL −1 ; LOQ: 10 μg mL −1 ), almost 17% of the mixture consisted of flavan-3-ols.
The strong antioxidant properties of GA (#2), and its derivatives, have been reported in literature, and several studies claim the possible role and application of GA in plant nutrition at different stages of growth, especially in early root and plant development 26,27 . For example, Singh and co-workers showed that the application of exogenous GA on rice seeds led to an increase of root length compared to the control 27 . Moreover, Negi and colleagues showed that GA might have an auxin-like function, directly acting on root length and development 17 . Compound #3 and the related monolactones (#7 and #15) and dilactone (#8) are also hydrolysable tannins belonging to the family of ellagitannins 28 , known to exert an antioxidant activity 29 . Concerning condensed tannins, they are strong antioxidant compounds 30 , and due to this property are largely used to supplement human and animal diet 31 . However, the knowledge about the use of tannin-based products in agriculture is limited 16 and need to be investigated. In addition, other bioactive compounds not belonging to the family of tannins were also found in the biostimulant. The compound #9, a flavone molecule, is known to be involved in root length development and in plant growth 32 . Compounds #12 and #16 correspond to natural steroids belonging to the triterpenoid family, which display interesting pharmacological effects on humans, modulating the Na + /K + -ATPase activity 33,34 . However, their potential action on plant physiology is still unknown. Finally, compounds #11 and #14, are two natural substances derived from wood hydrolysis 35 . VIVEMA TWIN has antioxidant properties. In this work, the antioxidant properties of VIVEMA TWIN were investigated in term of both reducing and radical scavenging activity via FRAP, ABTS and DPPH assays. The data of these measurements are shown in Table 1. In general, all the assays displayed very high values, when compared to those reported in Phenol-Explorer database for the top ranking fruits and vegetables 21 . However, due to the lack of antioxidant evaluations previously carried out on biostimulants, it is not possible to make a comparison with products belonging to this category. The results shown here indicate a strong antioxidant activity of this biostimulant suggesting that this product could help plants to survive different stresses related to accumulation of Reactive Oxygen Species (ROS).
ROS play different roles in plants 36 , however their overproduction may result in undesirable consequence for the plants 37 . Indeed, once formed, ROS must be detoxified as efficiently as possible to minimize potential damage. Plant cells are protected by a complex antioxidant system, including both non-enzymatic and enzymatic defenses 38 . Plant polyphenolic compounds stored by plants in their fruits, flowers and leaves belong to the first category 39 . In this context, plant-based biostimulants originating from industrial wastes might be rich in polyphenols that, if applied to plants, may promote beneficial effects by reducing potential oxidative threats.
Currently, the evaluation of the antioxidant properties of fruit and vegetable extracts is a crucial point to better understand the action mechanism of plant-based supplements to be used for human and animal nutrition 40 . Conversely, although a large part of biostimulants are derived from plants, this approach is still not used for plant nutrition. On the other hand, the measurements of antioxidant parameters of plant biostimulants will help to understand their mechanism of action, and contribute for their optimal use in agriculture. www.nature.com/scientificreports/ VIVEMA TWIN and gallic acid are able to modify the root architecture in tomato. In order to understand the root development in the early growth phases both in the presence and in the absence of biostimulant or salt stress, imaging analyses were performed using Root System Analyzer software. Through the use of this software, we were able to convert a 2D picture in a black&white skeleton, in which roots of different orders are dissimilarly colored 41 . Consequently, we could visualize not only the root structure, but also compare the total root number of the different plant samples. Figure 42 . In this case, tomato plants subjected to different concentrations of NaCl (0, 50, 100 mM) showed a decrease of biometric parameters such as shoot fresh weight, plant height and number of leaves, in correlation with the higher salt concentration. Finally, when VIVEMA TWIN was applied on plants stressed with 100 mM NaCl, an increase of the root number (from 7.0 ± 1.1 to 10.4 ± 0.8), root length (from 2.8 ± 0.3 to 5.1 ± 0.6 cm) and root fresh weight (from 23.3 ± 3.1 to 39.3 ± 1.2 mg) was recorded. Similar effects were also displayed after the treatment with 75 μM GA, in which the number of roots increased up to 9.6 ± 1.7, their length to 4.2 ± 0.8 cm and their fresh weight to 40.3 ± 1.5 mg. Moreover, a comparable trend, although in lesser extent, was also observed in plants stressed with 200 mM NaCl and treated with the same dosage of biostimulant or Gallic Acid. Finally, the observed effects displayed after the application of the biostimulant, may not only be the result of a synergic action of the different chemical compounds present in the mixture, but it can also be originated from a protective antioxidant effects supplied by the application of VIVEMA TWIN. Moreover, our data show that this biostimulant can be used to improve the salt stress resilience in tomato, and perhaps in other crops as well. www.nature.com/scientificreports/ VIVEMA TWIN and gallic acid are able to modify the root architecture also in long-term salt stress treatment. The root development of tomato plants was also studied during the long-term stress treatment, by evaluating biometric parameters, such as root length, root fresh weight and NDVI index. NDVI is a unit designed to measure both red and near infrared reflectance on vegetation, two parameters useful to determine plant health. NDVI measurements can range from − 1 to 1, with higher values indicating better plant health 43 . The roots used for biometric data measurements were collected 24 h after the second (Supplementary Fig. 1) and the fourth (Fig. 4) biostimulant treatment. A parallel test, under the same experimental conditions, was also performed in order to compare the effects shown after the application of VIVEMA TWIN with those resulting from the application of 75 μM GA or water only. As showed in Supplementary Fig. 1, 24 h after the second plant treatment, 100 mM NaCl did not significantly affected root length and fresh weight, but strongly influenced NDVI. Indeed, in the second sampling time point, this value decreased from 0.72 ± 0.06 to 0.61 ± 008. On the other hand, the treatment with the biostimulant or with gallic acid was able to completely recover the NDVI index. Different effect was observed 24 h after the fourth treatment (Fig. 4). In this case, we did not observed any NDVI change between plants watered with 100 mM NaCl or with water only (Fig. 4C), although a strong reduction in the fresh weight was recorded upon salt treatment (Fig. 4B). The lack in the NDVI changes may be linked to a possible plant adaptation to a prolonged salt stress condition 44 . On the other hand, even thought NDVI was not negatively affected by 100 mM NaCl treatment, a significantly increased of NDVI was recorded after the application of VIVEMA TWIN or GA. In this case, the treatments led to even higher values in comparison to control plants grown in optimal conditions. Concerning root weight, after the application of the biostimulant or gallic acid, we observed a recovery of the negative effects caused by salt stress. Finally, no   Table 3 the significantly upregulated genes (FC > 1.7) are shown. Most of these genes are involved in abiotic stress response (63%), root growth (18.5%) and other metabolic functions (18.5%) based on GO analysis. The most significantly downregulated genes (1/FC > 1.7, FC < 0.6) are reported in Table 4. Table 3. List of the most significantly upregulated genes (FC > 1.7; p < 0.05; n = 3) from RNA-Seq analysis on tomato roots. The fold change expression (FC) of genes from roots treated with 100 mM NaCl and 1 mL L −1 VIVEMA TWIN compared to untreated plants grown in the same conditions is shown.  The upregulation of these genes is related to the increase of ABA activity, correlated to an increase of stress tolerance 46 . ABA, also called "stress hormone", is fundamental in plant development and plays a key role both in the integration of stress signals and in the control of the stress response 47 . In particular, based on the plant status, WRKY transcription factors act as activators or repressors of ABA signaling, and are also involved in plant adaptation to salt stress 48,49 . Trehalose 6-phosphate phosphatase (T6PP) catalyzes the conversion of trehalose-6-P (T6P) to trehalose, a disaccharide involved in stress tolerance increase 50,51 . T6P, an highly soluble and low molecular weight compound, works as osmoprotectant by enhancing the resistance against salt stress 52 . Moreover, T6PP expression and trehalose content are increased in response to ABA and the synergistic action between the disaccharide and ABA also leads to a positive effect on root elongation in Arabidopsis 53 . ABA 8′-hydrolase, is a cytochrome P450 enzyme, involved in ABA catabolism, and importantly in maintaining the hormone balance 54 . The Protein phosphatase 2C was shown to play a key role in ABA signal transduction in Arabidopsis as well 55 , whereas HVA22-like protein in cereals is an ABA/stress induced protein, whose upregulation inhibits the formation of gibberellin GA-induced large vacuoles 56 . Our data suggest that in tomato similar ABA salt stress related signal transduction is activated for the survival of the plant due to the application of the biostimulant as compared to Arabidopsis and other species.
Late embryogenesis abundant (LEA) proteins. This group of genes is represented by two different Late embryogenesis abundant protein (LEA) family proteins (FC = 2.50 and FC = 1.97). Even if the mechanism of action of these proteins is not completely known in tomato, the upregulation of these genes in plants is strongly associated to the increment of stress tolerance, in particular in response to water limitation 57 . Drought stress presents many similarities with salt stress, since, in presence of salt, water is less available to the plants. The stress response involves the activation of common mechanisms, including the synthesis of LEA proteins 58 . NHE8 is known to be involved in salt and drought stress in tomato plants, and in particular encodes for a plasma membrane antiporter, essential to maintain Na + /K + homeostasis 59 . Its upregulation usually leads to the increase of salt stress tolerance in tomato plants 60 . C 2 H 2 ZnFP and DRZnFP are genes coding for transcription factors activated by stress conditions. With regard to glutathione (GST), this enzyme catalyzes the conjugation between reduced glutathione (GSH) and electrophilic substrates. In particular, these enzymes have been largely studied for their capacity to bind toxic exogenous compounds, thus protecting plants from multiple types of stresses 61,62 , including abiotic stress responses 62,63 , and their upregulation may suggest a protective antioxidant effect of VIVEMA TWIN application.
Root development related genes. A number of the genes upregulated by the biostimulant application were shown to be involved in the root developmental process. These are R2R3MYB transcription factor 41 (FC = 2.34), . Different studies showed that these genes are involved in primary and lateral root formation, by acting on cellulose deposition (xyloglucan endotransglucosylase-hydrolase 5) 64 , auxin transport (Major facilitator superfamily protein) 65 in Arabidopsis and in different other aspects like root hair development, root elongation and root architecture (R2R3MYB transcription factor 41) 66 . Finally, as reported above, also T6PP seems to have a role in root elongation.
Gene validation by qPCR analysis. RNA-Seq data were validated through qPCR analysis on selected genes significantly regulated by the application of the biostimulant (Supplementary Table 2). The qPCR analysis was performed on samples derived from VIVEMA TWIN-treated roots of plants grown under salt stress conditions (100 mM NaCl) and compared to the roots of stressed untreated plants. The roots were collected 24 h after the fourth treatment, at the same time point for which RNA-Seq analysis was carried out. Moreover, in order to compare the gene expression data related to the application of the biostimulant with those of the pure compound, qPCR was also performed on roots of stressed plants treated with 75 μM GA. Roots were collected at the same time point as previously described. The expression data are reported in Table 5.
In general, the up/down regulation of selected genes observed in the RNA-Seq experiments was confirmed by qPCR analysis on the samples treated with the biostimulant. Differently, the gene expression of plants treated with 75 μM GA seems not to be significantly modulated by the treatment. Only WRKY transcription factor 61, GST and GSyT showed a significant (p ≤ 0.05) upregulation, similarly to plants treated with VIVEMA TWIN. Based on these results, the biostimulant seems to be effective on the expression of genes involved in stress response, root growth and nutrient uptake. Interestingly, the application of the pure GA showed a lower effect on gene expression, suggesting a possible action in synergy with other components present in VIVEMA TWIN mixture. The study of synergistic effect of compounds in a mixture is a current research topic in the biostimulant field. Indeed, not only the role of single components needs to be investigated, but also the effect resulting from their interactions in a complex mixture.

Conclusions
In conclusion, our study presents novel insights on the modifications of root architecture in salt-stressed tomato plants after the treatment with the biostimulant VIVEMA TWIN. The development of root architecture with the concomitant modulation of genes involved in root growth, abiotic-stress responses and nutrient uptake, suggest that the biostimulant may have important role not only in plant nutrition but also in changing plant morphological traits for better plant salt stress survival and adaptation (Fig. 5).
Moreover, the comparison with GA treatments, showed how the synergistic activity of the biostimulant compounds led to more evident results on root development and plant resilience to salt stress than the application of a pure compound alone. Indeed, it has been shown that the effect of multiple bioactive molecules together is often higher than the sum of the effects of individual molecules 67 . Finally, it would be interesting to evaluate the effects of VIVEMA TWIN on other abiotic stresses and on different crops in order to verify its specificity and demonstrate its application potentially useful in multiple adverse conditions.

Materials and methods
Chemical characterization of VIVEMA TWIN. Spectrophotometric determination of bioactive compounds. The Total polyphenol content (TPC), Total Anthocyanin Content (TAnthC) and Total Flavan-3-ol content (TF3C) were determined spectrophotometrically using 1 mL L -1 of VIVEMA TWIN (Green Has Italia S.p.A., Canale, Italy) dissolved in water. Data are shown as mean of three biological replicates ± standard devia-  www.nature.com/scientificreports/ tion (SD). TPC was determined by the reduction of phosphotungstic-phosphomolybdic acids (Folin-Ciocalteu's reagent) to blue pigments in alkaline solution. The assays were carried out as previously described 24 , and the results were expressed as mg Gallic Acid Equivalents (GAE) per mL of biostimulant. TAnthC was determined via pH differential method, by reading the absorbance developed at 520 nm and 720 nm both at pH 1 and 4.5 68 . Data were expressed as mg of Cyanidine Equivalent (CyE) per mL of biostimulant. TF3C was determined by the formation of green-coloured complexes after the reaction of the flavan-3-ols with the BL-DMAC reagent. The assays were carried out as previously described 69 , and the results were expressed as PAC-A type equivalent per mL of biostimulant. Limit of detection (LOD) and Limit of Quantification (LOQ) were calculated for each assay using calibration curves of pure standards (Extrasynthese, France), ranging between 1-20 μg mL −1 . LOD and LOQ were calculated as previously reported 69,70 .
HPLC-ESI-FTMS identification and quantification. VIVEMA TWIN was analyzed by using an Orbitrap Fourier Transformed Mass Spectrometer (FTMS; Thermo Fisher) hybrid system. The biostimulant was firstly 1:1000 (v/v) diluted in water, and then mixed 1:2 (v/v) with 50% (v/v) methanol containing 1% (v/v) formic acid. Samples were vortexed (XH-D lab Vortex Mixer, Scientific Instrument) and sonicated (Ultrasonic Bath P120H, Elma) for 10 min, and then centrifuged (Sorvall Primo, Thermo Fisher) for 15 min at 10.000 g. The obtained supernatant was transferred to a new microcentrifuge tube, and used for further analysis. The chromatographic separation was carried out using a LUNA 3 μm C18(2), 150 × 2.00 mm column (Phenomenex, USA) and with a linear gradient from 5 to 35% B in 45 min, using a flow rate of 0.20 mL min −1 , as previously described 71 . The FTMS was set at a mass resolution of 60,000 HWHM, and with a mass range of m/z 140-2000. Electrospray ionization (ESI) in negative mode was employed for the ionization of compounds. Identification and quantification was based on retention times (RT) and accurate masses (MW) compared to the pure standards (5-20 µg mL −1 ) of ellagic acid, gallic acid and tannic acid. Data analysis was performed following the methods previously described 71,72 . Radical scavenging activity (ABTS and DPPH). The radical scavenging property of VIVEMA TWIN was evaluated by ABTS (2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) and DPPH (2,2-diphenyl-1-picrylhydrazyl) assay. ABTS radical cation decolorization assay was performed as previously described 73 . The assays are based on monitoring the colorization decay of the radical forms (ABTS ·+ or DPPH) respectively at 515 or 735 nm. For both assays, samples were analyzed at five different dilutions, within the linearity range of the assay. Gallic acid was used as a reference compound, and the reducing activity was expressed as mmol GAE per mL of biostimulant. All measurements were repeated three times.

Evaluation
Plant material and treatment with biostimulant. Tomato (S. lycopersicum L. Heinz 1706) seeds were sown in plate on a wet filter paper. Plates were incubated in a growth chamber (25 °C, 16/8 h light/dark, PPFD 100 μmol m −2 s −1 ) for 7 days. Seedlings were then transferred in the greenhouse in pots containing 100% sand. The pots were watered three times a week with 1 g L −1 nutrition solution (Hyponex, Japan). After the first leaf emergence (BBCH 11), plants were treated by application of water (untreated control), 1 mL L −1 VIVEMA TWIN (Green Has Italia S.p.A., Canale (CN), Piedmont, Italy) (treated samples) or 75 μM GA. The different treatments were also performed under standard or salt stress conditions. For each growth condition (unstressed/ untreated, unstressed/treated, stressed/untreated, stressed/treated), twenty plants were used, randomly distributed, considering each plant as a biological replicate using a fully randomized experimental design. For "shortterm" test ( Fig. 6), plants were treated every 4 days, starting 2 days after transplantation in the greenhouse. In this case, salt stress was induced by watering plants with 100 mM or 200 mM NaCl solutions in two parallel experiments. Because of the shortness of the experiment, the saline solution was given at the same time of each treatment. Roots were collected 24 h after the second treatment. For "long-term" tests ( Fig. 7), plants were treated once per week, for 4 weeks and the salt stress was induced after the first treatment (priming treatment) by watering the plants three times a week with 100 mM NaCl added to the nutrition solution. For the stressed plants, no further irrigations were carried out, other than those with salt. Leaves and roots were collected 24 h after the second treatment and 24 h after the fourth one.
Evaluation of morphological parameters. Short-term test. After washing and drying the roots, length of the primary root was manually measured, and the fresh weight was determined using an analytical scale. Moreover, the picture of each root was processed with Root System Analyzer. This is an automated approach employed to measure root architectural parameters from two-dimensional root images 41,74 . The data are obtained in the form of original photos, which are consequently converted into a black&white skeleton. In the skeleton each root order is indicated with a different color. The total number of roots, measured according to the procedure described above, was also taken into account.
Long-term test. Normalized Difference Vegetation Index (NDVI) of both untreated and treated plants was monitored on leaves before the plant material collection, using Pigment Analyzer PA110 (Control in Applied Physiology, Germany). Plants were then collected, and root length was manually measured, and the root fresh  Genome expression analysis. Total RNA isolation. Total RNA was isolated from powdered roots using TRIzol reagent (Thermo Fisher Scientific, MA, USA) and following the manufacturer's instructions. RNA quality was evaluated by 1% (w/v) agarose gel electrophoresis, whereas its purity was measured by NanoPhotometer (IMPLEN, USA).
RNA-Seq analysis. RNA-Seq analysis was performed by Novogene (Hong Kong, China) both on roots untreated or treated with VIVEMA TWIN, grown under salt stress conditions (100 mM NaCl). Before proceeding with library construction the total RNA integrity was checked by using Nano 6000 Assay Kit and Agilent 2100 Bioanalyzer 2100 (Agilent Technologies, USA) and the concentration measured using Qubit RNA Assay Kit in Qubit 2.0 Fluorometer (Life Technologies, USA). Three µg of total RNA was used as input material for the generation of the sequence libraries. Transcriptome assembly, coding potential analysis, expression analysis, GO enrichment and KEGG pathway analysis, were done using Novogene standardized protocols as already shown before in other similar studies 62,75-77 . Raw counts produced by HTSeq-count were normalized based on the DESeq2 normalization method 78 . DESeq2 linear models in R environment were implemented in order to identify statistically significant differentially expressed genes (DEGs) between treated samples and controls. Genes with an adjusted (Benjamini-Hochberg correction for multiple hypothesis testing) p value < 0.05 were considered as differentially expressed. Moreover, Log2FC cut-off was set to 1.7. The list of all DEGs obtained from this experimentation is reported in Supplementary  Table 3 Finally, RNA-Seq data were submitted to SRA (Sequence Read Archive) database with the accession number PRJNA645803.
cDNA synthesis and qPCR analysis. RNA-Seq data were validated through qPCR analysis carried out on the most significantly regulated genes after the application of VIVEMA TWIN (WRKY transcription factor 61, Trehalose 6-phosphate phosphatase, Na + /H + exchanger 8, C 2 H 2 zinc finger protein, Glutathione S-transferase, Major facilitator superfamily protein, ERD, Glycosyltransferase, Xyloglucan-endotransglucosylase-hydrolase 5, TPSI1). The same genes were also analyzed in GA-treated roots in order to evaluate gene expression variations between the biostimulant and the pure compound. Briefly, half a µg of RNA was reverse transcribed by using the Maxima H Minus First Strand cDNA Synthesis kit (Thermo Fisher Scientific, USA), following the manufacturer's instructions. For each qPCR reaction 0.3 μM primers, 4.1 μL of nuclease-free H 2 O and 5 μL of SYBR-Green I (Maxima SYBR Green/ROX qPCR Master Mix 2X, Thermo Fisher Scientific, USA) were added. The qPCR reactions were performed using a QuantStudio 1 Real-Time PCR System (Thermo Fisher Scientific, USA). All the conditions used for the different gene amplification are reported in Supplementary Table 1. For each treatment, three biological replicates (obtained by combining the twenty biological replicates in three pools) and three technical replicates were analyzed. To normalize the data, four different reference genes (Tubulin, Ubiquitin, Elongation Factor 1 and a catalytic subunit of Protein Phosphatase 2A) were also analyzed. Among these, a reference was selected using the Normfinder software 79 ; the most stable gene selected was the Elongation Factor 1. The primers of both reference and selected target genes, were designed using Primer3 software 80 and are reported in Supplementary  Table 2. The relative gene expression levels were calculated by using the Pfaffl method 62,81 , meanwhile the variations in gene expression were calculated as relative quantification of the target genes in relation to the reference gene, Elongation Factor 1 as shown before 24,62 .

Figure 7.
Schematization of Long-term test. Seeds were sown in plate, then seedlings transferred to the green house and watered with Hyponex as nutrient solution. After the first true leaf appearance, the plants were treated, under optimal or salt stress conditions. After 7, 14 and 21 days, the treatment was repeated following the same experimental conditions. Salt stressed plants were watered with 100 mM NaCl solution three times a week. www.nature.com/scientificreports/ Statistical analysis. Data, if not differently specified, were analyzed through one-way analysis of the variance (ANOVA) using "Systat Version 10" (Systat Software, San Jose, CA). Tukey-Kramer's post-hoc test (p ≤ 0.05) was used to analyze the means. For each experiment at least twenty biological replicates were used. When necessary, L1-norm exclusion test was performed with the aim to remove outliers from the data set. In this case, data of at least sixteen biological replicates were used for further statistical analysis.