Overproduction of ABA in rootstocks alleviates salinity stress in tomato shoots

To determine whether root-supplied ABA alleviates saline stress, tomato ( Solanum lycopersicum L. cv. Sugar Drop) was grafted onto two independent lines (NCED OE) overexpressing the SlNCED1 gene (9- cis -epoxycarotenoid dioxygenase) and wild type rootstocks. After 200 days of saline irrigation (EC = 3.5 dS m (cid:1) 1 ), plants with NCED OE rootstocks had 30% higher fruit yield, but decreased root biomass and lateral root development. Although NCED OE rootstocks upregulated ABA-signalling ( AREB , ATHB12 ), ethylene-related ( ACCs , ERFs ), aquaporin ( PIP s) and stress-related ( TAS14 , KIN , LEA ) genes, downregulation of PYL ABA receptors and signalling components ( WRKYs ), ethylene synthesis ( ACO s) and auxin-responsive factors occurred. Elevated SlNCED1 expression enhanced ABA levels in reproductive tissue while ABA catabolites accumulated in leaf and xylem sap suggesting homeostatic mechanisms. NCED OE also reduced xylem cytokinin transport to the shoot and stimulated foliar 2-isopentenyl adenine (iP) accumulation and phloem transport. Moreover, increased xylem GA 3 levels in growing fruit trusses were associated with enhanced reproductive growth. Improved photosynthesis without changes in stomatal conductance was consistent with reduced stress sensitivity and hormone-mediated alteration of leaf growth and mesophyll structure. Combined with increases in leaf nutrients and flavonoids, systemic changes in hormone balance could explain enhanced vigour, reproductive growth and yield under saline stress.


| INTRODUCTION
Limited water availability is a shared component of drought and salinity stresses that constrains crop growth and yield. In addition, salinity stress limits plant growth and agricultural productivity through nutritional imbalance and ion toxicity. Roots sense their environment, triggering transcriptomic and biochemical responses that allow the plant to adapt to such conditions through local and systemic responses, with hormones playing a key role in such adaptive responses (Achard et al., 2006). Root-targeted alteration of hormone metabolism and signalling has been proposed as a biotechnological strategy to overcome the effects of saline soils, and to enable this we must understand the specific adaptive roles of plant hormones (Albacete, Martínez-Andújar, & Pérez-Alfocea, 2014;. Crops dynamically regulate their root system architecture (RSA) in response to environmental stresses to fulfil their mineral and water requirements. In dry and saline soils, plants reduce lateral root initiation and elongation while promoting root hair density and the growth of the primary root to reach deeper water and nutrient sources (Brown et al., 2012;Koevoets, Venema, Elzenga, & Testerink, 2016;Li et al., 2021;Xu et al., 2013) Depending on the level of salt tolerance of the plant species or genotype, low-moderate salinity (2-8 dS m À1 ) can promote root growth while high salt levels (8-16 dS m À1 ) restrict root development (Julkowska & Testerink, 2015).
Among the different plant hormones, tissue-specific Abscisic acid (ABA) levels (and responses) change dynamically according to developmental and environmental stimuli. Although ABA is generally considered to inhibit growth of well-watered plants, low ABA concentrations (<1 μM) can stimulate root growth of Arabidopsis (Ephritikhine, Fellner, Vannini, Lapous, & Barbier-Brygoo, 1999;Fujii, Verslues, & Zhu, 2007). Phenotypic comparisons between wild-type (WT) and ABA-deficient mutants demonstrates that WT Abscisic acid (ABA) levels are necessary to sustain primary root growth in maize seedlings grown under low water potential (Sharp & LeNoble, 2002), and for leaf expansion and shoot development in tomato (Sharp, LeNoble, Else, Thorne, & Gherardi, 2000) and Arabidopsis (LeNoble,  under well-watered conditions. ABA may stimulate growth by restricting the biosynthesis of ethylene, a growth inhibitor (reviewed in Sharp et al., 2004). Within the roots, ABA alters gene expression that induces changes in RSA , increases root hydraulic conductivity , modifies nutrient and ionic transport and changes primary metabolism leading to osmotic adjustment Sharp & LeNoble, 2002). Plants growing in dry or saline soil can show stomatal closure before shoot water status (the trigger for leaf ABA accumulation) begins to decline (Dodd, 2005;Gowing, Jones, & Davies, 1993), coincident with root ABA accumulation and export to the shoot as a rootto-shoot signal (Wilkinson & Davies, 2002;Zhang & Davies, 1989).
However, experiments with reciprocal grafts of ABA-deficient and WT plants showed that stomatal closure of WT scions in response to dry (Holbrook, 2002) or saline (Li, de Ollas, & Dodd, 2018) soil was rootstock independent. Instead, roots in drying soil alkalize xylem sap causing a redistribution of existing pools of ABA within the leaf that affects stomatal closure (Wilkinson, Corlett, Oger, & Davies, 1998), and other non-ABA chemical signals such as sulphate (Malcheska et al., 2017) or jasmonic acid (De Ollas, Arbona, G omez-Cadenas, & Dodd, 2018) may also be involved. ABA detected in the root system may either be synthesized locally or translocated from the shoot via the phloem (McAdam, Brodribb, & Ross, 2016), and ABA can recirculate between roots and shoots, with roots either acting as a sink for ABA or as a net exporter of ABA to the shoot, depending on plant nutrient and water status (Peuke, 2016).
Genetically increasing endogenous ABA levels is a promising strategy to improve resistance to abiotic stresses such as drought and salinity. The enzyme 9-cis-epoxycarotenoid dioxygenase (NCED) is ratelimiting for ABA biosynthesis, and over-expression of NCED genes increased ABA content of tissues, as first shown in tobacco and tomato by overexpressing the tomato gene SlNCED (Thompson et al. 2000. This work provided transgenic tomato lines with different levels of expression of SlNCED1 and ABA contents (SP12 and SP5) and offers the opportunity to study the effects of high ABA on root-to-shoot communication. In previous reciprocal grafting experiments between WT, SP12 and SP5, ABA in xylem sap collected from de-topped roots was mainly determined by the root genotype, as might be expected in the absence of the shoot. In addition, root cultures (again independent of the shoot) of SP12 and SP5 had higher ABA content that WT, thus overexpression of SlNCED1 was sufficient to increase ABA biosynthesis in the root alone , despite the much lower level of NCED substrate available in roots compared to leaves (Taylor, Sonneveld, Bugg, & Thompson, 2005). In contrast, stomatal conductance in well-watered reciprocal grafting experiments was significantly affected only by the shoot genotype . Overexpression of NCED has now been explored in many systems, and its limiting effect on stomatal conductance confers improved water use efficiency (WUE; Thompson, Andrews, et al., 2007) and resistance to terminal drought (withdrawal of irrigation in pot experiments). Lower transpiration rate and slower soil moisture depletion of these NCEDoverexpressing lines maintains turgor of tobacco (Qin & Zeevaart, 2002), grapevine (He et al., 2018) and petunia (Estrada-Melo, Ma, Reid, & Jiang, 2015) in drying soil. NCED overexpression also increased growth relative to WT under osmotic stress (NaCl, mannitol) in tobacco (Zhang, Yang, Lu, Cai, & Guo, 2008) and improved transpiration and reduced chloride accumulation in Arabidopsis grown in 'a 150 mM chloride dominant solution' (Zhang, Yang, You, Fan, & Ran, 2015). However, the effect of rootstocks overexpressing NCED on plant growth and yield responses to saline soil has not been investigated.
ABA interacts with other hormones to mediate local and systemic stress responses (Sah, Reddy, & Li, 2016): it antagonizes the growth inhibitory effects of ethylene production in tomato shoots , Arabidopsis shoots (LeNoble et al., 2004) and maize roots (Spollen, Lenoble, Samuels, Bernstein, & Sharp, 2000), and also during grain-filling in wheat (Yang, Zhang, Liu, Wang, & Liu, 2006). Moreover, root-supplied ABA from WT rootstocks was sufficient to revert xylem 1-aminocyclopropane-1-carboxylic acid (ACC) concentrations and foliar ethylene production of ABA-deficient scions, while enhancing their leaf area (Dodd, Theobald, Richer, & Davies, 2009). However, night-time maize leaf expansion of water-stressed plants did not appear to be regulated by either ABA or ethylene (Voisin et al., 2006), but probably by more complex hormone interactions.
Grafting is commonly applied to many woody and herbaceous horticultural species in commercial practice (Albacete et al., 2014).
Tomato is one of the most important economic crops in the world and is commonly propagated by grafting high productivity scions onto vigorous rootstocks to alleviate soilborne diseases and abiotic stress effects (Bletsos & Olympios, 2008;. Cultivated tomato is moderately tolerant to salinity with a threshold of tolerance of 2.5 dS m À1 but there is a subsequent yield loss of 10% for each unit of salinity increase (François & Maas, 1994), which means that 30-40% yield losses due to salinity are quite common in many horticultural areas such as the tomatoproducing region of Southeast Spain. Root-specific traits such as RSA, sensing of edaphic stress and root-to-shoot communication can be exploited to improve resource (water and nutrients) capture and plant development under resource-limited conditions. Root system engineering and rootstock breeding provides new opportunities to maintain sustainable crop production under changing environmental conditions. We hypothesize that grafting a commercial tomato cultivar scion onto ABA over-producing tomato rootstocks would enhance growth and yield under saline conditions, potentially through multiple local and systemic mechanisms.

| Plant culture
Two independent tomato transgenic lines, SP5 and SP12, in the genetic background of the WT cultivar Ailsa Craig (AC;  were used in this study as rootstocks of the commercial cherry variety Sugar Drop (SD, Unigenia Semillas, Murcia, Spain). SP5 and SP12 transgenic rootstocks constitutively overexpress the SlNCED1 gene (Thompson et al., 2000), under the control of the Gelvin superpromoter (SP) and contain elevated ABA levels compared to WT, with SP5 accumulating more ABA than SP12 . Since germination rates differed between genotypes, different sowing dates were used to synchronize development of the three genotypes: SP12 and SP5 seeds were sown one and two weeks before the WT, respectively, as described previously . Seeds of the scion SD were sown 5 days earlier than AC seeds (7 days earlier than SP12 and 14 days earlier than SP5) to ensure equal stem diameters at grafting.
For all genotypes, seeds were sown in commercial vermiculite, watered with deionized water and kept at 26-28 C and 80-90% relative humidity in the dark until germination. Grafting was performed using the splicing method at the two to three true leaf stages (3-4 weeks after sowing) where the scion was attached at the first node of the rootstock (Savvas et al., 2011). Grafting with the two transformants and the WT AC resulted in three graft combinations: SD/SP5, SD/SP12 and SD/AC ( Figure S1).
One month later, when the grafted plants were well established, they were cultivated under commercial-like conventional plastic greenhouse conditions using a sand substrate during an autumnwinter season, in Almería area (Spain). Fertilizers and water were supplied by a drip fertigation. From 10 days after transplanting, a low salinity treatment with an electrical conductivity (EC) of 3.5 dS m À1 was applied for a period of 200 days ( Figure S1). Six plants per graft combination were randomly cultivated and distributed in blocks.

| Plant phenotyping
Throughout the experiment (after 130, 163 and 180 days of salt treatment, DST), photosynthesis (A N ), stomatal conductance ( g s ) and substomatal CO 2 (Ci) were measured in the youngest fully expanded leaves (one leaf per plant) using a CIRAS-2 (PP Systems, Massachusetts, USA) between 09.00 and 12.00 hr (lights were turned on at 08.00 hr). CO 2 was set at ambient levels (400 ppm) and radiation matched the chamber conditions (1,500 μmol m À2 s À1 PPFD). Intrinsic water-use efficiency (WUE i ) was calculated as the ratio between the values of A N and g S .
After 130 DST, the second fully expanded mature leaf over the fourth truss (with actively growing fruits) of six plants per graft combination was assayed for various physiological parameters (described above), then detached to weigh and determine leaf area using an LI-3100 AC area meter (LI-Cor, Lincoln, NE, USA). Plant stem diameter was also measured at the second node level using an electronic liquidcrystal display (LCD) digital vernier caliper (0-150 mm). At the end of the experiment (200 DST), the shoot and root were detached and weighed to determine biomass.
Young fully expanded leaves and young roots were immediately frozen in liquid nitrogen and stored at À80 C for hormonal and gene expression analysis. Leaf, root and truss xylem sap were obtained by applying a pneumatic pressure (between 0.6 and 0.7 MPa) to excised organs. Sap was collected with a pipette, immediately frozen in liquid nitrogen and stored at À80 C for hormonal analysis. Phloem exudate was collected using the method described by Pérez-Alfocea, Balibrea, Alarc on, and Bolarín (2000). The distal stem with the shoot apex and the two youngest expanded leaves were excised and the basal 2-3 cm immediately immersed in a 150 ml glass containing 30 ml of 20 mM Ethylenediamine tetraacetic acid (EDTA) (pH 6, adjusted with LiOH to avoid interactions with cation measurements). Each container with the plant material was placed in a plastic bag and hermetically sealed. The exudate was obtained by incubating the plant material for 20 hr in the dark at room temperature.
Total yield was calculated using all the fruits collected from each plant during the harvest period. Fully ripe fruits were harvested weekly for 2 months. The truss length and fruit weight were also recorded in the third truss. Fruit at green and mature stages were also harvested for hormonal analysis.

| Nutritional, hormonal and flavonoid analysis
For ionome composition, leaves were dried for 48 hr at 80 C, milled to a powder and 200 mg dry tissue was digested with an HNO 3 :HClO System (Bio-Rad). Three biological and two technical replicates were performed per genotype and treatment. The thermal cycling programme started with a step of 30 s at 95 C, followed by 40 cycles (5 s at 95 C, 10 s at 55 C and 20 s at 72 C) and a melt curve (from 65 to 95 C, with increments of 1 C every 5 s). Dissociation kinetic analyses and agarose gel loading and sequencing of the PCR product were used to confirm its specificity.
Primer pair validation and relative quantification of gene expression levels were performed using the comparative Ct method (Schmittgen & Livak, 2008). Data were represented as the relative gene expression normalized to the Ct value for the tomato housekeeping gene SlACTIN2 (Solyc04g011500) as previously described (Ferr andez-Ayela et al., 2016). In each gene, mean fold-change values relative to the expression levels of WT were used for graphic representation. ΔCt values were analysed using SPSS 21.0.0 (SPSS, Inc.) by applying the Mann-Whitney U test for determining statistical differences between samples (p-value ≤.05).

| Microarray hybridization and data analysis
Four biological replicates per genotype were used for RNA extraction using the method described above. RNA (200 ng Arrays were imaged using an MS200 microarray scanner using only the 480 nm laser using the autogain feature of the NimbleScan software (Roche NimbleGen, Madison, WI, USA). Image (tiff) files were imported into the Agilent Feature Extraction software for quality control assessment, grid alignment and expression value extraction at the probe and transcript level with the RMA algorithm (Irizarry et al., 2003) used to carry out background subtraction, quantile normalization and summarization via median polish and output log2 normalized gene expression levels (GEO record GSE79307; Ferr andez-Ayela et al., 2016). Linear Models for Microarray Data (package LIMMA in R) was then used to fit linear models to pairs of samples, identifying genes that contrasted the most between the experimental pairs (Smyth, 2004). Transcripts were deemed to be differentially expressed if they showed a Benjamini-Hochberg adjusted p ≤ .05 when comparing rootstocks genotypes.
The molecular pathways where differentially expressed genes were involved in the biosynthesis of plant hormones ( Figure S2) and hormone signal transduction ( Figure S3) were marked in the relevant Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (Kanehisa & Goto, 2000).

| ABA sensitivity
Surface-sterilized (washed in 5% NaOCl) WT and SP12 seeds were germinated in Petri dishes containing 1/5 Hoagland nutrient solution supplemented with 10 g L À1 agar and 1% sucrose. Seedlings were transferred to culture medium supplied with 0, 1.5, 3 and 5 μM (+)-cis, trans-ABA (Sigma-Aldrich) when the two cotyledons were developed (6 days for WT and 9 days for SP12). After 30 days of ABA treatment, main total root length was measured using WinRHIZO software (Pro 2016, Regent, Canada).

| Statistical analysis
Data were subjected to analysis of variance (ANOVA) to test the main effects of genotype. Genotypic means were compared using Tukey's test at 0.05 of confidence level. All analyses were performed using SPSS for Windows (Version 22.0, SPSS, Inc., Chicago, IL).

| Plant growth, gas exchange, leaf nutrients and yield
To determine whether rootstock ABA overproduction can alleviate salt stress, two independent tomato transgenic lines, SP5 and SP12, in the genetic background of the WT cultivar AC, as previously reported (Thompson et al., 2000), were used as rootstocks of the commercial cherry variety SD. At the end of the growing cycle (up to 200 days of irrigation with saline water), plants grafted onto NCED OE rootstocks had almost twice the leaf area, leaf and shoot biomass (shoot fresh weight; SFW) and stem diameter of plants grafted onto WT rootstocks (Figure 1a,b). However, the root biomass of SP12 and SP5 rootstocks was 30% and 60% smaller than WT rootstocks, respectively ( Figure 1b). Visually, these NCED OE grafts had less a complex RSA (the spatial configuration of a root system in the soil), than the WT ( Figure 1a). Moreover, plants grafted onto NCED OE rootstocks had up to 20-30% increases in length and weight of the third fruiting truss, fruit number, fruit weight and total fruit yield ( Figure 1b). Thus, NCED OE rootstocks promoted shoot (and fruit) growth but reduced the root system growth.  (Table 2). Thus, NCED OE rootstocks affected leaf structure, nutritional status and function.

| Hormone accumulation
Since hormones mediate many physiological changes (Albacete et al., 2008a;Ghanem et al., 2008), we measured hormone levels of several root and shoot tissues and xylem and phloem exudates of grafted plants (Figures 3 and 4; Table S1).
Generally, NCED OE grafts produced few significant effects on ABA concentrations in tissues and transport pathways compared to the WT rootstock (Figures 3a and 4). Interestingly, the NCED OE rootstocks significantly increased ABA concentrations in the xylem  Plants grafted onto NCED OE rootstocks had lower total CKs (t-Z and iP type) in the xylem sap of roots and flowering truss, as well as in leaf tissue and green fruits mainly due to lower t-Z levels ( Figure 4; Table S1). The different graft combinations had similar t-Z and iP concentrations in leaf xylem sap and root tissues. However, iP type CK concentrations on leaf tissue (130 DST) and leaf phloem exudate were 5-14-fold higher in plants grafted on NCED OE rootstocks than on WT rootstocks, with iP the only hormone increasing in leaf phloem exudate ( Figure 4; Table S1). Thus, rootstock NCED OE significantly affected CK concentrations in root xylem sap and shoot tissues.
Rootstock genotype also significantly affected auxin (IAA) and ethylene precursor (ACC) measurements. Leaf phloem exudate and root tissue ACC concentrations were 3-25 times lower in plants grafted on NCED OE rootstocks, while they had a higher ACC concentration in xylem sap of a mature fruit truss ( Figure 4; Table S1).
Leaf phloem exudate and xylem of mature fruit truss had up to sixfold lower IAA concentrations when grafted on the SP5 rootstock ( Figure 4; Table S1), otherwise there were no significant rootstock impacts on IAA levels. Similar to ABA, xylem sap of trusses at flowering and green-fruited stages had 7.5 to fourfold more GA 3 when grafted on NCED OE rootstocks, with these differences disappearing at fruit maturity ( Figure 4). However, leaf xylem GA 3 concentration of plants grafted on NCED OE rootstocks was 65-80% lower than when grafted on WT rootstock. Furthermore, root xylem JA concentration of plants grafted on SP5 was lower, even though plants grafted on NCED OE rootstocks had leaf JA concentrations that were more than twice that of plants grafted on WT rootstocks at 80 DST (Table S1) Table S1). Thus, NCED OE rootstocks also occasionally affected tissue and transport fluid concentrations of other acidic hormones.

| Gene expression
To determine the molecular basis of the physiological changes, the  (Table 3).
To highlight any classes of genes that are over-represented in the differentially expressed genes, GO terms were searched for higher difference in the frequency between the differentially expressed transcripts and all the transcripts included in the microarray (Figure 5c).
When comparing SP rootstocks to WT, differentially expressed genes were enriched in several classes, including serine-type endopeptidases, defence response genes, oxygen binding, snoRNA binding, chlorophyll binding and glucuronosyltransferase activity (Figure 5c).
To interpret the gene expression data in a physiological context, we analysed DEGs related to hormone metabolism ( Figure S2) and signalling ( Figure S3) pathways, initially focusing on ABA-related genes because of the known role of NCED. Both PCR and transcriptomic data showed that SlNCED1 gene expression was higher in SP5 than SP12 (Figure 6a; Table S2 and    Both NCED rootstocks upregulated the ACC synthase genes (ACC2, Solyc01g095080; ACS1a, Solyc08g081540) and most ethylene response factors (ERFs; Figure 9a; Table 3). SP12 and SP5 rootstocks upregulated 2 and 1 ACC oxidase genes, respectively, but    Note: The 25 most upregulated genes (largest logFC values) and the 25 most downregulated genes (smallest, most negative logFC values) are given with their mean relative expression (AveExpr) level and the adjusted p-value (Adj.P.val).
(a) (b) (c) F I G U R E 6 ABA related genes differentially expressed in root tissues comparing plants of SD/SP12 and SD/SP5 against SD/AC in response to 3.5 dS m À1 (equivalent to 35 mM NaCl) for 200 days in greenhouse conditions. Real time PCR quantification (RT-qPCR) of some ABA-related selected genes is also given (a). Root xylem sap ABA concentration (as a percentage with respect to control conditionsno salt, data in the embedded table-for each genotype) as a function of salt concentration in the medium (35, 70 and 100 mM NaCl) of tomato cv Ailsa Craig selfgrafted (AC/AC, open circles) and grafted onto the NCED OE line SP12 (AC/SP12, closed circles) during 27 days. Each point represents the mean value of four replicates. Different letters indicate significant differences between treatments within each graft combination (p ≤ .05). * and ** indicate significant difference between graft combinations at p ≤ .05 and p ≤ .01, respectively (b). The relationship between main root total length (RL) and ABA concentration in the culture medium (0, 1.5, 3 and 5 μM ABA) in tomato cv AC (open circles) and the transgenic line SP12 (SP12, closed circles) grown in vitro during 30 days. Each point represents the mean value of four replicates along with its standard error. Different letters indicate significant differences between treatments within each graft combination (p ≤ .05). * and ** indicate significant difference between graft combinations at p ≤ .05 and p ≤ .01, respectively (c)

| NCED OE rootstocks have reduced gene expression for ABA receptors and signalling components
Rootstock SlNCED1 overexpression (Figure 6a) was consistent with transgene expression level in own-rooted plants (Martínez-Andújar, , implying that shoot-to-root signalling has little effect on constitutive (root-specific in grafted plants) SlNCED expression. Although bulk root ABA status did not increase in fruiting plants (Figure 3a), previously ABA in root exudates from approximately 7-week old de-topped plants (Thompson, Andrews, et al., 2007), in root cultures  and in bulk root tissue and xylem sap of younger ungrafted plants  was elevated. Moreover, bulk root ABA concentration of grafted plants was determined by the root genotype and increased in SP5 and SP12 , as in the root xylem sap prior to stress (Figure 6b). Therefore, the lack of bulk root ABA accumulation in this study is consistent with increased root export (Figures 3a and 6b) and catabolism of ABA (Figure 3b).
NCED OE rootstocks showed differential gene expression com-   (Figure 2a,b) without changing g s , thereby increasing intrinsic WUE (Figure 2b). Similarly, reciprocal grafting experiments under non-stressed conditions indicated that only NCED OE scions decreased g s with only modest effects on A N , while NCED OE rootstocks had no effect on g s .
Irrespective of environmental stresses, elevated ABA tissue concentrations can promote developmental changes in stomata and leaf anatomy that mimic the effects of water deficit (Franks & Farquhar, 2001;Galmés et al., 2011;Quarrie & Jones, 1977).
Enhanced cuticular wax deposition and changes in its composition can protect photosynthesis (Ziv, Zhao, Gao, & Xia, 2018). In this study, grafting scions onto NCED OE rootstocks increased elongation of leaf epidermal cells and reduced the number of cuticular wax crystals on leaf adaxial and abaxial surfaces ( Figure 2e; Table 1). Similarly, scions grafted onto autotetraploid Rangpur lime rootstocks with high ABA levels had higher expression of the wax synthesis WAX2 gene than scions grafted onto the diploid equivalent with lower ABA levels (Allario et al., 2013). In contrast, there was a positive relationship between ABA level and wax deposition in ABA-deficient tomato mutants and following exogenous ABA application (Martin, Romero, Fich, Domozych, & Rose, 2017). NCED OE rootstocks may diminish wax deposition by directly downregulating wax synthesis pathways, or indirectly by alleviating salinity stress, thereby allowing greater leaf expansion and consequently diluting wax deposition or attenuating stress-induced wax synthesis. Furthermore, rootstocks can improve photosynthesis by affecting leaf structure to enhance mesophyll conductance to CO 2 ( g m ; Fullana-Pericàs, Conesa, Pérez-Alfocea, & Galmés, 2020), with g m negatively correlated to sub-stomatal and/or ambient CO 2 concentration under long-term stress (Flexas et al., 2012(Flexas et al., , 2013. Here, grafting onto NCED OE rootstocks disorganized laminar mesophyll structure (Figure 2c), possibly explaining decreased Ci (Figure 2d) by enhancing CO 2 diffusion to the cells (Flexas et al., 2012(Flexas et al., , 2013.
Other rootstock-derived metabolites may also protect root and leaf function. Two genes involved in flavonoid synthesis, a flavanone 3-hydroxylase-like protein and a flavonoid oxidoreductase, were among the most upregulated genes in NCED OE rootstocks (Table 3; Figure 7c). Flavonoid accumulation leads to chilling and salt stress tolerance in tomato and Arabidopsis by reducing reactive oxygen species (ROS) accumulation and sensitivity to ABA (Li, Liu, & Yao, 2017;Mahajan & Yadav, 2014;Meng, Zhang, Deng, Wang, & Kong, 2015), which is supported by the down-regulation of several peroxidase genes in the NCED OE rootstocks (Table 3). Furthermore, rootstockderived flavonoids were xylem-transported to the leaves (Albacete et al., 2015). bundles (Veselov et al., 2018). By facilitating CO 2 diffusion to carboxylation sites (Flexas et al., 2012(Flexas et al., , 2013, iP/ABA-mediated mesophyll alteration favoured CO 2 assimilation. Indeed, both ABA and iP have been proposed as signalling components of the reticulate leaf phenotype in Arabidopsis, which has altered mesophyll structure and reduced CO 2 fixation capacity (Lundquist, Rosar, Bräutigam, & Weber, 2014). Interestingly, a phosphoglycerate mutase gene (Solyc04g072800), whose function is reduced in reticulate mutants (Lundquist et al., 2014), was 2 and 1.4-fold upregulated in SP12 and SP5 rootstocks, compared to the WT (Table 3). This enzyme is key in ATP production and reducing power from glycolysis (Zhao & (a) (b) F I G U R E 9 Ethylene (a) and auxin (b) related genes differentially expressed in root tissues comparing plants of SD/SP12 and SD/SP5 against SD/AC in response to 3.5 dS m À1 (equivalent to 35 mM NaCl) for 200 days under greenhouse conditions. Real time PCR quantification (RT-qPCR) of some ABA-related selected genes is also given Assmann, 2011) and could contribute to active transport and root assimilatory processes such as nutrient uptake and Na + exclusion (Malagoli, Britto, Schulze, & Kronzucker, 2008;Munns, Passioura, Colmer, & Byrt, 2020) and nitrate or sulphate reduction (Wang et al., 2004), thereby enhancing leaf nutrient status. Moreover, iP-type CKs were related with greater xylem development and plant growth, vigour and yield in tomato (Qi et al., 2020). Since root-to-shoot CKmediated plant vigour under salinity (Albacete et al., 2009(Albacete et al., , 2014Albacete, Ghanem, et al., 2008a; was associated with decreased ABA levels, ABA-CK interactions in rootstock-mediated improvement of the scion physiology require further investigation, 4.4 | Ethylene and gibberellin related responses in NCED OE grafted plants ABA signalling maintains shoot and root growth in both well-watered and droughted tomato (Dodd et al., 2009;Sharp et al., 2000Sharp et al., , 2004 and Arabidopsis (LeNoble et al., 2004) plants by suppressing ethylene production (LeNoble et al., 2004;Sharp et al., 2000;Spollen et al., 2000). Surprisingly, NCED OE rootstocks upregulated genes for biosynthesis of the ethylene precursor ACC (ACC2, Solyc01g095080; ACS1a, Solyc08g081540) and ethylene signalling (several ERFs), while most genes responsible for the final step in ethylene biosynthetic genes (e.g., ACCO, Solyc07g049550; ACCO-like protein, Solyc12g F I G U R E 1 0 Proposed model to explain how ABA overproducing rootstocks improve growth and yield under saline conditions, by affecting local (root) and systemic (scion) responses mediated by root-to-shoot communication. (a) In the roots, ABA overproduction seems to interfere with stress mediated response by decreasing root expression of ABA receptors and signalling components, thus altering sensitivity to ABA. Decreased ABA sensitivity in the roots appears to diminish auxin activity (ARFs, auxin transport from the shoot) and increases ethylene-related processes (ERFs, ACCs) leading to reduced RSA (mainly lateral roots). Lower IPT gene expression diminishes rootstock CK synthesis and t-Z transport to the shoot. (b) In the scion, increased ABA catabolites in fruiting plants and ABA accumulation in young plants indicates that a root-toshoot ABA signal cannot be ruled out. Increased foliar iP accumulation and phloem transport (in response to reduced t-Z transport from the roots) along with transient foliar ABA and JA accumulation seems to modify leaf growth and mesophyll structure leading to improved photosynthesis (A N ) activity. Increased transport of nutrients and flavonoids to the leaves also protects leaf function. Moreover, increased xylem GA 3 in growing fruits seems to enhance reproductive growth. Improved photosynthesis and reduced root growth optimize source-sink relations to benefit scion development and yield. Arrow and bar heads indicate positive and negative regulation, respectively [Colour figure can be viewed at wileyonlinelibrary.com] 006380) were down-regulated (Figure 9a). Root and leaf phloem ACC concentrations were significantly reduced, as in own-rooted NCED OE plants . Since diminished (lateral) root development in the NCED OE rootstocks is consistent with the phenotype of the ethylene overproducing mutant epinastic under control (Negi, Sukumar, Liu, Cohen, & Muday, 2010) and saline (Ortiz, 2017) conditions, higher up-regulation of ERFs may be involved (Figure 9a). ERFs induce GA2 oxidases to inactivate GAs and root growth by stabilizing DELLA proteins (Hetherington, Kakkar, Topping, & Lindsey, 2021;Julkowska & Testerink, 2015). Whether these local changes in ethylene and GA responses are involved in systemic signalling is less clear, as reproductive tissues of scions grafted on NCED OE rootstocks had increased ACC and GA 3 levels ( Figure 4; Table S1). These enhanced GA 3 levels are consistent with the elongated truss phenotype (Figure 1). Overall, ABA-ethylene-GA interactions seem involved in regulating root growth, while long-distance ACC and GA signalling cannot be ruled out.
Finally, regulation of pathogenesis-related proteins and subtilin-like proteases genes seems highly sensitive to elevated natural (Zhang, Cao, Li, Chen, & Xu, 2019) or transgenic (this study) constitutive ABA production, which deserves further investigation.

| CONCLUSION
Grafting WT scions onto constitutively ABA-overproducing rootstocks produced local (root) and systemic (scion) responses mediated by root-shoot communication. Evidence that rootstock SlNCED1 overexpression changed root-to-shoot ABA signalling included increased ABA concentrations in scion reproductive tissues and increased ABA catabolites in leaves, but lower leaf phloem ABA concentrations.
ABA overproduction altered stress-mediated responses associated with: decreasing root expression of PYL ABA receptors; reduced auxin signalling (lower auxin concentration in leaf phloem and decreased root expression of auxin-responsive factors); enhanced root expression of most ethylene signalling gene (ERFs); and decreased lateral root development. Moreover, rootstock NCED overexpression downregulated root expression of CK biosynthesis genes and reduced t-Z in root xylem sap and leaf, suggesting reduced CK transport from root to shoot. However, iP increased in the leaf and leaf phloem, potentially as part of feedback loop to restore CK homeostasis. Increased root glycolytic activity may mediate increased nutrient uptake and flavonoid synthesis and transport for stress protection in the scion.
Rootstock NCED overexpression modified leaf growth and anatomy and enhanced photosynthesis, possibly due to iP, JA and ABA accumulation in the leaf and leaf phloem. Enhanced GA 3 in truss xylem sap was consistent with increased truss length, weight and overall yield.
Considering whole plant source-sink relationships, the stimulation of leaf photosynthesis and reduction in root assimilate requirements for biomass could explain the more productive scion phenotypes (vegetative vigour, truss length, fruit number and yield) when grafted on NCED OE rootstocks. Overall, NCED OE rootstocks may be of great value in generating plants with higher yields under abiotic stresses ( Figure 10).

ACKNOWLEDGMENTS
The authors are very grateful to María del Puerto S anchez-Iglesias for her technical assistance on hormonal analysis. Research was also supported by the Spanish MINECO-FEDER (project RTI201