Reciprocal Grafting Reveals Differential Metabolic Responses Between Robusta Clones with Contrasting Tolerances to Drought

Faced with global warming, the surface area of coffee cultivation regions is expected to diminish significantly in the near future. As a result, new varieties or agronomical practices improving drought tolerance need to be found. The aim of this work is to characterize drought tolerance of Coffea canephora genotypes and their reciprocal grafted plants with physiological tools and biochemical analyses. Under greenhouse conditions, control plants (sensitive or tolerant) and reciprocal grafted plants submitted to 14 days of water deprivation show variations of the monitored parameters, such as soil and leaf water potential, stomatal conductance, and osmoprotectant compounds (sugars, polyols, amino acids). The variations observed confirm the differences between the phenotypes defined as drought-tolerant and drought-sensitive. Reciprocal grafting shows enhanced and contrasting situations. A sensitive clone grafted onto tolerant rootstock presents higher tolerance to drought and physiological or biochemical parameters similar to a drought-tolerant clone. The opposite is observed for tolerant clones grafted onto a sensitive one. More contrasted results are obtained with glucose, fructose, proline, and mannitol content which could be used as indicators for drought tolerance. Our finding shows strong variability for drought tolerance in our Robusta clones and demonstrates the impact of grafting on physiological and biochemical parameters linked to drought tolerance. The use of drought-tolerant rootstock leads to better regulation of water management and biochemical composition of the scion in drought-sensitive clones. This could be an approach to improving drought tolerance of Coffea canephora genotypes and to limiting the impact of global warming on coffee farming.


Introduction
Coffee is one of the most important agricultural commodities exchanged throughout the world [14]. More than 25 million people depend on this production [18]. However, due to climate change, a significant reduction of coffee cultivation area (up to 50%) is expected by 2050 [2,3,13,21]. Consequently, it is imperative to develop new tolerant varieties.
The first evidence of genetic variability for drought tolerance was identified in coffee collection [1,6,11,12,26,32], but more efficient varieties for commercial production still need to be developed.
However, breeding for drought tolerance is still empirical, and the mechanism involved in drought tolerance remains unknown [22], assumptions are made that deeper and more regularly developed root systems are related to drought tolerance. Recently, Silva et al. [30] demonstrated the benefit of reciprocal grafting of a drought-sensitive Conilon onto a resistant rootstock.
Recently, Brum et al. [10], relates modification in carbohydrate metabolism during water stress for Arabica and Santos et al. [26], describes variabilities of Raffinose oligosaccharide family (RFOs) metabolite content in sensitive and in tolerant varieties Robusta. Nevertheless, as interactions between biosynthesis and degradation occurred in the biochemical pathway, contradictory results are still reported [28].
Today, the modulation of those metabolites through grafting, is unknown, even if variations in the expression of antioxidant enzymes (SOD, CAT, APX) in grafted plants are described in coffee leaves [30].
The main purpose of this study is to evaluate the impact of grafting on physiological or biochemical parameters expressed during water stress. Using Robusta clones identified for some of them as drought-tolerant and for the others as drought-sensitive, we investigate the impact of grafting on plant response to water stress. As previously described in the literature, these concern sugars, polyols and amino acids. These results open the way for the use of tolerant rootstocks to improve drought tolerance in elite trees selected for yield and quality.

Plant Materials
Coffea canephora plants used were selected inside clonal hybrid progeny created in Nestlé breeding program. They are coming from two principal crosses of different Coffea canephora groups defined by Merot-L'anthoene et al. [19] (FRT133 = Group A 9 Group E and FRT140 = Group A 9 Group D) and propagated through cuttings as described by Priyono et al. [24] to obtain homogeneous plant material. Two years old rooted cuttings were cultivated in 14 L plastic pots in sandy-based substrate (v/v: clay 15%, brick chippings 10%, sand 30%, coco fiber 20%, worm compost 2% plus fertilizer: 1 kg/m 3 (N 12/ P12/ K17 ? oligo-elements; pH 6.2; Ec 1.0mS/cm). Soil is covered with coco mulch disc to prevent evaporation. Plants were irrigated daily with 0.5L and supplemented with 1% fertilizer (v/v) once a week with plant-prodÒ Fertil (N20-P20-K20) and grown under greenhouse conditions (day/night temperature 30/26°C; 80% relative humidity). Additional lighting with 12 h photoperiod given by HPS PhilipsÒ Sodium-vapor 400 W lamp, to maintain constant PAR 350 lmol/m 2 /s.

Experimental Design
Plants with similar size and number of leaves were used for water stress experiments, randomized on greenhouse bench. Water stress experiments started by stopping irrigation over the 14 days and physiological parameters were recorded. Five individual plants were used as control for the two clones FRT133 and FRT140, but only three plants for each grafting combination (FRT133/ FRT140 and FRT140/FRT133) were available limiting the number of repetitions.

Physiological Variables
Physiological measurements were taken at 12 am (UTC time). Soil water potential Ws was measured using a WatermarkÒ (Agroressources) sensor (maximum sensor resistance value 200 cbar). Leaf water potential Wl was measured using Scholander pressure chambers [27] 1505D, PMS instrument (Sol et Mesures). Stomatal measurements with an SC1 porometer from Decagon DevicesÒ Company. Biochemical sampling was performed on the same day (12 am) on the 3 rd leaves, starting from the apex of plagiotropic branches. Leaves were frozen and ground in liquid nitrogen, the powder sifted at 500 lm and stored frozen at -20°C until use.

Sugars and Polyols
Extractions were carried out using 70 mg of ground powder re-watered with 7 mL milliQÒ water and incubated for 30 min at 70°C. After cooling to ambient temperature, the sample volume was adjusted to 10 mL with milliQÒ water then sugars were analyzed through high-pressure chromatography (HPLC). Soluble sugars or polyols were separated through anion exchange chromatography using Dionex TM column CarboPac PA100 (2 9 250 mm) and Dionex TM pre-column CarboPac PA100 (2 9 50 mm), Flow: 0,25 mL/min, injection volume 5 lL, duration 53 min, heating temperature 25°C., mobile phase milliQÒ water/ NaOH, amperometric detection. Quantification was performed with standard and peak area calculations.

Proline Content
Leaf powder (200 mg) were re-watered in 10 mL milliQÒ water, vortexed and extraction were done in an ultrasonic bath for 1 h at 15°C, centrifuged at 15,000 g for 15 min at 25°C. One mL of supernatant is filtered at 0.20 lm and derived using ACCQ reagent (Waters) comprised in the Waters kit AccQ-FluorTM for UV detection. One mL of acetonitril reagent was added, heated at 75°C for 10 min, then 50lL of samples was added to 50 lL reagent AccQ completed with 350 lL borate buffer, vortexed, heated, and injected into the system. The UPLC proline quantification done using AccQ-Tag Ultra Column BEH C18 2,1 9 100 mm, 1.7 lm, heating temperature 48°C mobile phase (Eluent A: 20% AccQ-Tag Ultra Eluent A concentrate and 80% milliQÒ water; Eluent B 100% AccQ-Tag Ultra Eluent B (Waters), output 0.65 mL/min, injection volume 2 lL, UV detection at 2600 nm.

Statistical Analysis
Statistical analyses were performed using MinitabÒ18 software. Data normal distribution was checked, and ANOVA was performed using the general linear model followed by the Tukey pairwise comparison test (p \ 0.05).

Phenotypic Responses of Sensitive FRT133, Tolerant FRT140 and Their Reciprocal Grafted Plants
Under our greenhouse conditions, the symptoms of water deprivation follow mainly the same chronological steps for drought-sensitive and drought-tolerant trees. Starting with turgidity and leaf loss followed by extreme leaf dehydration (Fig. 1a). After 14 days of water stress, leaves start burning with severe plasmolysis and leaf drying. Additionally, leaves from the main stem present crispier phenotypes on FRT133 than leaves from FRT140.
Two weeks after rehydration, symptoms are enhanced as shown in Fig. 1b: the sensitive tree (FRT133) shows brown dead leaves prone to fall, whereas the tolerant tree (FRT140) shows turgid green leaves with moderate damage. The differences between tolerance and sensitivity in our conditions are more strongly expressed after rehydration to plant recovery.
Finally, the sensitive trees have lost more leaves and exhibit darker brown stem color (indicating higher mortality of the branches) than tolerant FRT140 trees. On the contrary, FRT140 trees have lost a limited number of leaves, and new buds have developed on plagiotropic branches which are still green after re-watering.
The impact of water stress on grafted plants (reciprocal grafting) 14 days after re-watering is shown in Fig. 2.
The phenotypes recorded on grafted plants two weeks after re-watering are similar to those observed in Fig. 1b (leaf loss or leaf burning) but, in the opposite way according to grafting order. The FRT140 scion grafted onto sensitive rootstock FRT133 presents darker brown and crispy leaves compared to the FRT133 scion grafted onto the tolerant FRT140 rootstock presenting green leaves and moderate burning.

Soil and Leaf Water Potentials
To follow the evolution of water consumption and transpiration, leaf water potentials are expressed according to soil water potentials in Fig. 3.
Water consumption presents different behavior with tolerant and sensitive genotypes (Fig. 3a) starting with the same consumption up to Ws (-1.0 bar), and then, the Wl of sensitive FRT133 decreases more rapidly than tolerant FRT140. At Ws (-1.9 bar) the Wl of FRT 140 reach (-55 bar) compared to Wl (-75 bar) for FRT 133. This reflects a better modulation of water uptake and transpiration for tolerant FRT140. We assume better regulation of transpiration or water uptake for FRT 140 for the same soil water potential, this could explain the better recovery of FRT140 after re-watering (Fig. 1b).
For grafted plants (Fig. 3b) the tolerant scion FRT140 grafted onto the sensitive rootstock FRT133 shows a trend of water consumption like the control sensitive FRT 133. On the contrary, higher water preservation is recorded for sensitive scion FRT133 when FRT140 is used as rootstock.

Stomatal Conductance
At the starting point in fully irrigated conditions (Fig. 4a), the level of stomatal conductance is similar (145 mmol/m 2 / s) for both sensitive FRT133 or tolerant FRT140 clones. The stomatal conductance decreases rapidly on Day 4 (T4) for Sensitive clone FRT133. Stomata were closed at T7 (68.2 mmol/m 2 /s). Inversely, for Tolerant clone FRT140, the stomatal conductance is maintained at the same level between 150 and 130 mmol/m 2 /s until day 7 (T7), then as the water stress increases, the stomata start to close. Reciprocal grafting shows contrasting results according to the rootstock used (Fig. 4b). Stomatal regulation seems to be more efficient for FRT133 grafted onto FRT140, like the FRT140 scion. On the contrary, FRT140 grafted onto sensitive FR133 shows stomatal closure like sensitive FRT133.

Raffinose Oligosaccharides Family
In most cases, the evolution of sugar content mainly shows differences on Days T7 when the plants start to express water stress close to Wl -4 MPa (Fig. 5). The content of myo-inositol is similar in control plants (Fig. 5a) and a strong effect of rootstock impact is observed when FRT133 is grafted onto FRT140 (Fig. 5b). More variations are observed for Galactinol in control plants (Fig. 5c) with higher levels for FRT140. Grafted plants (Fig. 5d) show similar concentrations to their control plants when it is used as rootstock (FRT133 vs. FRT140 grafted onto FRT133 and FRT140 vs. FRT133 grafted onto FRT140). Similar results are obtained with the impact of grafting for Raffinose ( Fig. 5f) with an inversion of the curves between T4 and T7 for control plants and grafted plants (Fig. 5e). For Stachyose, no difference is observed in control plants ( Fig. 5g). For grafted plants (Fig. 5h) the variations are like those observed with Raffinose.

Sucrose-Soluble Sugar Family
The trend of evolution of soluble sugars presents stronger differences with a change of concentration on Days T7 close to Wl -4 MPa when the plants start to express water stress (Fig. 6). For control plants (Fig. 6a and c) all of the sugars present the same concentration in FRT133 and FRT140 samples. On the opposite side, reciprocal grafting shows higher variations, especially for glucose (Fig. 6b) and fructose (Fig. 6d). The content of each sugar for FRT140 grafted onto FRT133 is mainly weaker, whereas the content for FRT133 grafted onto FRT140 is mainly higher, showing a major impact of the rootstocks on these carbohydrate contents. FRT133 plants grafted onto FRT140 rootstocks show significantly higher glucose (2.5 times more) and fructose (5 times more) content versus FRT140 grafted onto FRT133.

Proline and Mannitol
Among the free amino acids present in the sap, proline is one of the main amino acids expressed during abiotic stress. For both genotypes (Fig. 7a), the basal level of proline is low for well-watered plants and starts to increase after 7 days (T7) of water stress. After 14 days of water stress, the proline content is three times higher (704.2 mg/ 100 g) in the sensitive versus tolerant control (222.3 mg/ 100 g). The inverse situation is observed in grafting experiments (Fig. 7b) where less proline content is recorded on FRT133 grafted onto FRT140. The biosynthesis of mannitol content during water stress starts 4 days (T4) after the beginning of water stress and reaches a maximum of 3 to 6 days later. The evolution of mannitol content constitutes an early response to water stress. Both genotypes (sensitive FRT133 and tolerant FRT140) react in the same manner (Fig. 7c) with a rapid increase in mannitol content soon after 4 days of water deprivation. A higher level of mannitol content is observed in sensitive genotype FRT133 compared to tolerant genotype FRT140. Inversely, FRT133 grafted onto FRT140 accumulates less mannitol content (Fig. 7d).

Discussion
Large differences in phenotyping were noticed in our experimental conditions by submitting plants to 14 days of extreme water deprivation leading to an extremely low level of Wl (close to -9 MPa) compared with experiments conducted by Santos et al. [26] (-4.2 MPa). Thus, extreme phenotypes were recorded for sensitive trees with leaf loss and browning leaves and clear differences in their recovery after re-watering. These extreme experimental conditions were chosen to show the real behavior of plant resilience after extreme water stress. We assume that our tolerant FRT140 is less prone to embolism, as branch damage is less important than in sensitive FRT133. Unfortunately, no significant data supports this statement as it was not possible to evaluate the cavitation rate due to too long xylem vessels in Coffea canephora (Cochard and Delzon, pers. comm.). Grafting experiments support the benefit of grafting sensitive scion onto tolerant genotypes as the phenotypes express less burning and leaf loss for sensitive FRT133 grafted onto FRT140. The evolution of soil and leaf water potentials plus stomatal conductance measurements under water stress could explain the more efficient water management in tolerant FRT140 versus sensitive FRT133. Tolerant genotypes seem to present better regulation in stomata closure compared to sensitive trees with fast closure leading to more damage in the photosynthesis apparatus as observed with faster leaf browning. Their rapid stomata closure during water stress probably leads to a higher accumulation of active oxygen compounds (ROS) which are toxic for the plant cell apparatus. This aspect was highlighted by Ramahlo et al. [25] in the follow-up of antioxidative enzymes in the Coffea species. The variation in ROF content during water stress in control plants shows a limited variation on average, the distribution of data being too broad; it makes these results non-significant and therefore difficult to compare with Santos et al. [26] data. This could be explained by the differences in experimental design (pot size, number of replicates, water stress conditions) and turnover in the ROF biosynthesis pathways as described by Sengupta et al. [28].
However, more contrasted results are obtained with grafted plants, especially for myo-inositol and galactinol content with a higher level in FRT133 grafted onto FRT140. These results could be linked to the impact of FRT140 rootstocks submitted to water stress generating signals for the synthesis of ROF antioxidant compounds already described for Coffea canephora [16] and in other species [8,15], especially for the FRT133 scion which is more susceptible to oxidative damage. In the same way, our results in soluble sugar content show limited contrasting results in control plants but higher variations in grafted experiments. As for soluble sugars, higher contrasting glucose and fructose contents are observed in control plants showing differences between FRT133 (higher) and FRT140 (lower) from the beginning of water stress up to day 7 (T7) when the plant reach -4 MPa. According to the literature [23] on soluble sugar analysis (sucrose, glucose and fructose) the metabolism of these sugars is too variable and only slight variation in their contents is observed during water stress. However, for grafted plants, the synthesis of glucose and fructose reaches a level of up to five times greater from T4 to T7 in FRT133 grafted onto FRT140. We assume that signals coming from rootstocks such as ABA which are already described for drought acclimation with coffee [30] may participate in the increase of soluble sugars. For total soluble sugars, sensitive FRT133 presents a slightly higher content as opposed to tolerant FRT140 and this is clearly enhanced when sensitive FRT133 is grafted onto tolerant FRT140. On the other hand, differences are more significant for two other important osmotic compounds: proline and mannitol. For both of these molecules, the content increases only after T7 with higher levels in the sensitive FRT133 used as a control or as rootstock. All of these osmotic compounds are in higher concentrations in the control FRT133 compared to the control FRT140. Interestingly, the evolution of proline and mannitol content in grafted plants is the opposite of what was observed with soluble sugars. Indeed, the concentrations of glucose and fructose increase from the beginning of water stress up to T7 (close to -4.2 MPa), and on the contrary, proline and mannitol only start accumulating at T7.

Conclusions
Even if plant adaptation to drought is complex and combines varying phenomena, the main biochemical compounds expressed during water stress can be followed and used as markers. Osmoprotectant molecules are differentially expressed during water stress and some could be used to characterize genotypes for their drought tolerance. Under greenhouse conditions, the phenotyping for drought tolerance leads to the characterization of drought-tolerant and drought-sensitive genotypes. Physiological measurements were in line with phenotypes. As observed by Marguerit et al. and Silva et al. [17,30]; under water stress, the impact of the rootstock's characteristics on the grafted scion material was revealed by reciprocal grafting between tolerant and sensitive trees. Physiological parameters confirmed phenotyping observation for the characterization of two contrasting genotypes for drought tolerance. Our results from the biochemical analysis developed on genotypes already identified as drought-sensitive FRT133 and drought-tolerant FRT140 agreed with phenotypic observations and physiological measurements. During water stress, grafted plants show enhanced responses in the synthesis of sugars, mannitol or proline. Reciprocal b Fig. 6  grafting highlights the impact of such observations as we mainly observe the signature of drought-tolerant rootstock on grafted plants. The results obtained need to be confirmed with other genotypes but are promising for facilitating the selection process. Ongoing work is being developed to identify specific key gene sequences expressed during water stress in contrasting genotypes and this can be promising for breeding drought-tolerant coffee varieties. Identifying drought-tolerant clones in Coffea canephora should lead to their use as rootstock for improving drought tolerance with elite Coffea arabica varieties. Therefore, biochemical and physiological markers would facilitate the selection of elite clones to be used as rootstocks for future arabica plantations.
Acknowledgements Thanks to Mathieu Simon and Jean-François Bouquet for their help in greenhouse management, to Charles Lambot for helpful advice during the project and to Juan-Carlos Herrera for fruitful reading comments.

Declarations
Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.