Chemical intervention in plant sugar signalling increases yield and resilience

The pressing global issue of food insecurity due to population growth, diminishing land and variable climate can only be addressed in agriculture by improving both maximum crop yield potential and resilience. Genetic modification is one potential solution, but has yet to achieve worldwide acceptance, particularly for crops such as wheat. Trehalose-6-phosphate (T6P), a central sugar signal in plants, regulates sucrose use and allocation, underpinning crop growth and development. Here we show that application of a chemical intervention strategy directly modulates T6P levels in planta. Plant-permeable analogues of T6P were designed and constructed based on a ‘signalling-precursor’ concept for permeability, ready uptake and sunlight-triggered release of T6P in planta. We show that chemical intervention in a potent sugar signal increases grain yield, whereas application to vegetative tissue improves recovery and resurrection from drought. This technology offers a means to combine increases in yield with crop stress resilience. Given the generality of the T6P pathway in plants and other small-molecule signals in biology, these studies suggest that suitable synthetic exogenous small-molecule signal precursors can be used to directly enhance plant performance and perhaps other organism function.

The pressing global issue of food insecurity due to population growth, diminishing land and variable climate can only be addressed in agriculture by improving both maximum crop yield potential and resilience 1,2 . Genetic modification is one potential solution, but has yet to achieve worldwide acceptance, particularly for crops such as wheat 3 . Trehalose-6-phosphate (T6P), a central sugar signal in plants, regulates sucrose use and allocation, underpinning crop growth and development 4,5 . Here we show that application of a chemical intervention strategy directly modulates T6P levels in planta. Plant-permeable analogues of T6P were designed and constructed based on a 'signalling-precursor' concept for permeability, ready uptake and sunlight-triggered release of T6P in planta. We show that chemical intervention in a potent sugar signal increases grain yield, whereas application to vegetative tissue improves recovery and resurrection from drought. This technology offers a means to combine increases in yield with crop stress resilience. Given the generality of the T6P pathway in plants and other small-molecule signals in biology, these studies suggest that suitable synthetic exogenous small-molecule signal precursors can be used to directly enhance plant performance and perhaps other organism function.
We designed a signalling-precursor strategy on the basis of release by light (Extended Data Fig. 1). Light-activated control is a potent modulation strategy in biology, allowing for temporal and spatial resolution surpassing that of standard genetic methods 6 . Additionally, such resolution can be increased when combined with small-molecule chemical control [7][8][9] . Potency is further increased when releasing a signalling molecule, as the effects of light-activated control are increased several-fold through the inherent amplification of signalling.
Hydrophilic or charged molecules do not readily enter plants unless transported. We therefore designed unnatural precursors (1)(2)(3)(4) of T6P with groups to mask charge, increase hydrophobicity and also to induce release by light (Fig. 1a). Their construction (Fig. 1b) used different phosphorus chemistries: phosphoramidite chemistry 10,11 to create P(III)-intermediates that were then oxidized to corresponding P(V)-phosphotriesters, or direct P(V)-phosphorylation chemistry (Fig. 1b). Regioselective access to the OH-6 group in trehalose exploited trimethylsilyl as a protecting group that is chemically orthogonal to the phosphotriester; 12 was prepared on a multigram scale 12 . Phosphorylation (reaction with phosphoramidites 9-11 (refs 10, 11) followed by tBuOOH, or treatment with POCl 3 (ref. 13) followed by the addition of the appropriate alcohol) gave intermediates that were deprotected under mildly acidic conditions (see Supplementary Methods). 1-4 were all inactive against SnRK1 (SNF1-related kinase 1) (Extended Data Fig. 2).
Mass spectrometry, thin-layer chromatography and nuclear magnetic resonance spectroscopy (see Supplementary Methods, Supplementary  Table 1 and Extended Data Fig. 1) showed release times (95% release, t 95 ) dependent on both light intensity and frequency under a range of conditions. Consistent with design, light-sensitive groups were differently susceptible. Precursor ortho-nitrobenzyl (oNB)-T6P 1, for example, generated T6P by light-activated release more rapidly at lower wavelengths, while compound mono-dimethoxy(orthonitro)benzyl (mono-DMNB)-T6P 4 was more reactive at higher wavelengths. Although release with higher light intensity/ photon flux (125 W/365 μ mol m −2 s −1 (with photon flux defined as the number of photons per m 2 per second) compared to 8 W/23 μ mol m −2 s −1 ) was more rapid, direct sunlight proved sufficient, in some cases resulting in t 95 as brief as 90 min (for (ortho-nitrophenyl)ethyl (oNPE)-T6P (compound 3)). Nuclear magnetic resonance spectroscopy analysis (Supplementary Methods and Extended Data Fig. 1e, f) confirmed that T6P was formed and that potent inhibitory activity against SnRK1 was induced (Extended Data Fig. 2).
Following successful in vitro release, uptake in planta was examined. Compounds 1-4 (at a final concentration of 1 mM) were fed to roots of plantlets of Arabidopsis thaliana and the aerial parts were analysed over time and with increasing dose (Fig. 2 and Supplementary Tables 2-7). High-performance liquid chromatography (HPLC) followed by quantitative mass spectrometry 14 (HPLC-MS) of extracts of the aboveground biomass (shoot and leaves) showed increasing uptake of the compounds over time ( Fig. 2a and Extended Data Fig. 3) and with increasing dose. Consistent with design, structural variation showed that altered hydrophobicity modulated permeability 15 and transport 16 . Notably, systematic variation of group type and copy number identified compound 3 (log[P] of the compound is 0.11 ± 0.60, where P is the partition coefficient; see also Supplementary Methods and Supplementary Table 17) as the most readily taken up (Fig. 2a), with absorption of approximately 20% after 72 h. Compounds 1, 2 (dimethoxy(ortho-nitro)benzyl (DMNB)-T6P) and 4 (log[P] of the compounds ranged from − 2.35 to − 0.17), on the other hand, were less readily taken up.
Next, we investigated the light-activated release in planta. Plants were treated with compounds dissolved in medium, grown for a further three days, irradiated, and the shoots were harvested and extracted 17 . T6P release was confirmed by tandem mass spectrometry (Extended Data Fig. 4) and determined by quantitative HPLC-MS (with 2-deoxyglucose-6-phosphate as an internal standard 14 ; Fig. 2b, Extended Data Fig. 4 and Supplementary Table 10). Release in planta could be controlled and modulated by the choice of light source and signalling precursor ( Fig. 2b and Supplementary Table 10).
Most transgenic approaches only alter T6P over a 2-3-fold range 4,5 . Using compounds 1-4, levels of up to 900 nmol g −1 fresh weight were attainable, which was 100-fold higher than endogenous levels and 75-fold higher than can be achieved with genetic methods. Consistent with this strategy, maximal T6P was released when precursor-treated plants were irradiated with the highest flux (100 W/292 μ mol m −2 s −1 UV) in all cases. Notably, and with relevance to application in the field, with sunlight only, all treated plants released significantly increased amounts of T6P (39-296 nmol g −1 fresh weight) (Supplementary Table  10), some approximately 4-30-fold above endogenous levels. There was no significant reduction in the fresh weight of plantlets treated with 1 mM of compound (Supplementary Methods and Extended Data Fig. 5), suggesting low toxicity. Accumulation of T6P after treatment was analysed with mass spectrometry imaging 18 (Fig. 3) using signature-ion markers 19 in treated leaves of A. thaliana seedlings (Fig. 3b) after 2 h of irradiation; the different distributions from compounds 2 and 3 appeared consistent with their measured release rates. Notably, increased trehalose was also observed 20 in the same regions (Fig. 3c), suggesting metabolism. Moreover, mass spectrometry imaging using treatment-specific ions corroborated the uptake of precursors into leaves (Fig. 3d, e).
The dynamics of this enhanced in planta T6P release, and possible consequent metabolic products, were determined not only through both quantitative HPLC-MS and/or enzymatic quantification, but also through the use of unnaturally enriched isotopic labelling of the signalling precursors, allowing for unambiguous delineation of their fate (Fig. 4, Extended Data Fig. 6 and Supplementary Methods). Thus, in 7-day-old A. thaliana seedlings 21 , treatment with 1 mM of compound 2 or 3, fed for 24 h before exposure to light (UV 8 W/23 μ mol m −2 s −1 ), led to peak T6P after 60 min (229 and 159 nmol g −1 fresh weight, respectively), which declined over the following 2 days (Fig. 4a). Corresponding trehalose levels were also elevated, with peaks at around 2 h (up to a maximum of 134 nmol g −1 fresh weight compared to control levels of 20 nmol g −1 fresh weight; Fig. 4b), confirming the metabolism indicated by mass spectrometry imaging. Glucose, the next sugar in the pathway, was also increased but to a smaller degree, peaking at around 2-4 h (Fig. 4d). These levels are consistent with known low metabolic fluxes 22 . Given known interrelationships 5 , sucrose levels were also determined. Notably, these increased 2-3-fold over the first 2 h of irradiation ( Fig. 4c) and correlated positively with T6P for both compounds 2 and 3 (Fig. 4)  ANOVA showed significant differences (P < 0.001) between treatments (water or precursor) for each regime. All treatments with precursor and UV showed significance (P < 0.001, least significant difference (LSD)) compared to water and UV. For growth light irradiance * P < 0.05; * * P < 0.01 (LSD, data shown on a linear scale). See also Supplementary Table 10.
(Extended Data Fig. 6d). We confirmed that inhibited growth was not an explanation for sucrose accumulation and found instead that growth was stimulated by T6P (Extended Data Fig. 6e). Creation of a 13 C-isotopically-labelled variant 2* (Extended Data Fig. 7a) allowed direct tracking via 'mass-shifts' of the corresponding ions using mass spectrometry. Treatment with 2* led to release of 13 C-T6P and consequent sequential metabolism (to 13 C-trehalose and 13 C-glucose) following essentially the same dynamics (Extended Data Fig. 7). Notably, using this method, mass labelling also showed that the compounds not only released, but also induced T6P. This accounted for approximately half of T6P measured at 30 min, thereby providing direct evidence of induction of de novo T6P synthesis and this induction continued, giving rise to increased T6P accumulation over time (Extended Data Fig. 7). This could be due to the large increase in sucrose observed, as sucrose induces T6P 5,21 . Together these data suggest that perturbation of T6P levels occurs via two modes of action: direct release from the signalling precursor and simultaneously induced biosynthesis of T6P by the plant.
Our data indicate that compound 3 is the plant-permeable signalling precursor with the greatest tissue uptake coupled with the greatest temporal control (consistent with its tuned permeability and its fastest release rates), allowing minimal application amounts (0.1 mM), while still able to enhance T6P levels around 1.5-6.5-fold above endogenous levels without potentially detrimental disruption of metabolism 23 . Plants were treated with precursor 3 for 72 h and were then subjected to a single 8-h period under growth lights supplemented with 8 W UV (Supplementary Table 10, generating around 21 nmol g −1 fresh weight T6P) and harvested a day later.   thaliana seedlings grown in liquid culture were treated with 1 mM of either 3 or 2; control seedlings were treated with water. Seedlings were left under growth lights to take up the signalling precursors for 24 h, after which the plants were exposed to 23 μ mol.m 2 s −1 UV for 2 h. Measurements were taken 1 d after uptake (pre-UV), 30, 60 and 120 min after the initiation of UV treatment (23 μ mol m −2 s −1 ), and 1 and 2 days after the initiation of UV treatment. a, T6P content. b, Trehalose content. c, Sucrose content. d, Glucose content. In all cases n = 3; * P < 0.05; * * P < 0.01; Student's t-test; data are shown as mean ± s.e.m.   The mean starch level (63.2 μ mol g −1 fresh weight) determined 24 was significantly higher (F (1,14) = 13.59; P = 0.002) than for water-treated plants (40.7 μ mol g −1 fresh weight; Extended Data Fig. 5d).
The production of T6P from compounds 1-4 involves fragmentation with concomitant release of side products. Although considered 25 to be non-toxic, we nevertheless tested for any unexpected phenotypic changes. Glucose-6-phosphate (G6P) analogues 14-17 of compounds 1-4 were synthesized (Supplementary Methods and Extended Data Fig. 8) and compared for their activity; G6P-methyl-glycoside itself is inactive in planta and in all interactions with SnRK1 (Extended Data Fig. 8) and so its light-activated release from analogues 14-17 provided a useful control. Analogues 14-17 showed similar lightactivated release parameters to compounds 1-4 (Supplementary Table 11) and relative uptake performance was similarly dependent on the identity of the light-sensitive moiety (Extended Data Fig. 8 and Supplementary . No toxicity was observed in any of the plants treated with up to 0.5 mM of analogues 14-17 (Supplementary Methods and Extended Data Fig. 5a-c), suggesting that the lightreleased moiety is benign. Critically, starch was not affected in controls treated with compound 16, the G6P analogue of T6P precursor 3.
The rate of starch synthesis in A. thaliana over a 12-h period (Extended Data Fig. 5f, g) indicated a flux (0.037 μ mol min −1 g −1 fresh weight) nearly three times that of water-treated controls (0.013 μ mol min −1 g −1 fresh weight). T6P is proposed to stimulate starch synthesis through redox activation 26 of ADP-glucose pyrophosphorylase (AGPase), a rate-limiting enzyme. While not necessarily causal, consistent with this hypothesis, plants treated with compound 3 had significantly higher AGPase activity (increased by 35%, Extended Data Fig. 5e). AGPase has been previously shown to affect starch turnover 27 .
We also measured the number of transcripts of genes known to be associated with T6P. First, SnRK1 is a proposed target of T6P 30 . SnRK1-induced (TPS5, bZIP11 (also known as GBF6) and UDPGDH (At3g29360)) and -repressed (TPS8 and ASN1) markers responded synchronously to the activation of the precursors in a manner consistent with known effects of T6P on SnRK1 activity (Extended Data Fig. 9). However, other markers (for example, UDPGDH (At3g29360) and bGAL4) showed clear temporal delay (Extended Data Fig. 9b); the observed synchronization of these 'secondary markers' only occurred after a day, which suggests that these could be later, downstream targets of T6P. Second, starch is also a proposed target of T6P 26,31 . As starch levels were increased as a result of treatment, expression of starch biosynthetic genes was also analysed. Transcripts of APL3, SS3, BE1 (also known as EMB2729) and GBSS1 were increased up to fivefold (Extended Data Fig. 9c).
These data from A. thaliana raise the noteworthy possibility of enhanced starch synthesis in crops, potentially providing increased yield. Signalling precursors 2 and 3 were applied (0.1, 1 and 10 mM) to spring wheat (Triticum aestivum Cadenza), which was grown in a controlled environment representative of summer in northern Europe. Spraying occurred either to ears only or to the whole plant during the grain-filling period (5, 10, 15 and 20 days post-anthesis (the flowering period, DPA)) at mid-photoperiod. This increased grain yield per plant due to the formation of larger grain, particularly in plants treated with 1 mM of compounds 2 or 3 (Fig. 5a, b). In these grains, starch content increased 13-20% (Fig. 5c). A trend towards higher levels of starch and protein, when expressed as a percentage of component content per gram of grain, was also observed (Supplementary Table 16). Dose-response analysis showed that yield peaked at a precursor concentration of 1 mM (Extended Data Fig. 10f). Minimal spray amounts at only 10 DPA increased yield substantially at 1 and 10 mM doses (Extended Data Fig. 10g). Plants treated with compounds 2 and 3 stayed greener for longer than plants treated with water only, consistent with chlorophyll content (Extended Data Fig. 10a, b) and previous observations for genetically-enhanced T6P content 28  in the wheat grains treated with compounds 2 and 3 was enhanced at 5 DPA (128 nmol g −1 fresh weight and 81 nmol g −1 fresh weight, respectively) and further at 10 DPA (378 nmol g −1 fresh weight and 300 nmol g −1 fresh weight, respectively) (Extended Data Fig. 10c-e). Trehalose levels were also higher (30-70 nmol g −1 fresh weight compared to endogenous levels of 13 nmol g −1 fresh weight), consistent with the metabolism observed in A. thaliana.
Next, the effects of signalling precursors on plant resilience and recovery were analysed. Drought is still the biggest global factor limiting crop yields, even in developed countries 29 . When 4-week-old wheat plants were sprayed with compounds 2 or 3 (30 ml, 1 mM, once) after 9 days of drought, the regrowth effects following resumption of watering 1 day after treatment were substantial (Fig. 6a, b). Regrowth of new tissue from plants cut back after drought was also higher in precursortreated plants (Fig. 6c, d). This demonstrated both growth of new tissue (resurrection response) and salvage and growth of existing tissue (recovery response). T6P solution alone gave identical results to water (Fig. 6), consistent with the inability of T6P to enter directly into plants, further highlighting the design principles of signalling precursors.
In conclusion, we have shown here that a chemical strategy can directly control amounts of an important sugar signalling molecule in vivo. The collected data are consistent with the signalling action of released T6P. For example, the mass balance of added signalling precursor appears insufficient to simply act as a carbon source. That said, we do not discount other possible mechanisms behind the noteworthy traits that we have observed here. The apparent result of 'biosynthetic amplification' observed from signalling precursors is, we believe, a promising concept; we calculate here up to 50-fold 'molecular amplification' of the plant sugar product compared to the precursor. It may therefore be possible to design a self-sustaining production strategy in which a fraction of the additional starch generated by this amplification is used as a feedstock chemical for eventual synthesis of the signalling precursors themselves (Supplementary Discussion and  Supplementary Table 18).
We speculate that this chemical approach also offers more temporal and strategic flexibility than genetic methods (for example, a 'pulse' to circumvent adaptation effects or in manipulating more genetically complex crops) as well as the prospect of providing an immediate boost to productivity at critical times in the plant life cycle (for example, to allow synchronicity with the sun or to rescue drought-stricken regions, Supplementary Discussion)-the potential to contribute towards global food security seems notable and immediate. Given the widespread importance of cell signalling and of carbohydrates in biology, this system may also have even wider utility.
Online Content Methods, along with any additional Extended Data display items and Source Data, are available in the online version of the paper; references unique to these sections appear only in the online paper.

Synthesis of signalling-precursor compounds 1-4.
1H-tetrazole solution (0.45 M in CH 3 CN) (0.6 ml, 0.24 mmol, 2.0 equiv.) was added into a stirred solution of compound 12 (100 mg, 0.12 mmol, 1 equiv.) and bis-(2-nitrobenzyl)-N, N-diisopropylphosphoramidite (compound 9; 78.3 mg, 0.18 mmol, 1.5 equiv.) in anhydrous CH 2 Cl 2 (5 ml) under an argon atmosphere at 0 °C. The resulting reaction mixture was stirred at 0-5 °C and progress of the reaction was monitored by thinlayer chromatography (petroleum ether:ether, 8:2) and mass spectrometry. After complete disappearance of starting material (1 h), tBuOOH (0.1 ml) was added at 0 °C and stirring was continued for another 30 min. After 30 min the reaction mixture was concentrated in vacuo and the residue was suspended in methanol (2 ml) and stirred in the presence of 30 mg of Dowex-H + resin for 1 h at room temperature to globally remove trimethylsilyl groups. Dowex-H + was removed through filtration and the filtrate was concentrated, which on flash chromatography (water:isopropanol:ethyl acetate, 1:2:8) purification yielded compound 1 (70 mg) in 87% isolable yield. Similar reaction protocols were used for the synthesis of compounds 2 and 3. Compound 4 was obtained when a stirred solution of 12 (100 mg, 0.12 mmol) in pyridine (2 ml) at room temperature was treated with POCl 3 (0.012 ml, 0.132 mmol) for 10 min followed by addition of 4,5-dimethoxy-2-nitrobenzyl alcohol (76.7 mg, 0.36 mmol) and continuous stirring for 1 h. The resulting reaction mixture was concentrated in vacuo to yield a crude product mixture, which was treated with Dowex-H + (30 mg) in methanol (2 ml). After filtration, concentration in vacuo and flash chromatography purification yielded compound 4 (45 mg, 62%) as a pure sticky solid. For additional details see Supplementary Methods. In planta uptake of signalling-precursor compounds and release of trehalose-6-phosphate and metabolites. In planta uptake was carried out using A. thaliana plantlets. A. thaliana (Columbia 0) seeds were surface-sterilized for 10 min in 10% sodium hypochlorite, 0.01% Triton X-100 and then copiously washed with sterile water and stratified for 3 d at 4 °C. Seeds were sown onto 0.5 ml solid medium (0.5× Murashige and Skoog medium with Gamborg's vitamins (Sigma P0404), 0.5% sucrose and 0.5% agar) in 0.5 ml Eppendorf tubes, pierced in the bottom with a tiny hole. The tubes were arrayed in hand-cut polystyrene racks in Phyta trays (Sigma) and floated on liquid medium (same as solid medium but lacking sucrose and agar). Plantlets were grown under the following conditions: 12 h day under Philips master TL-D 840/58W fluorescent lights outputting 250 μ mol m −2 s −1 , with 23 °C day and 18 °C night temperatures. At 18 days after sowing the liquid medium was removed and the tubes were sealed with electrician's tape. All plants were topped up with 0.5× Murashige and Skoog medium with no sucrose.
Plants were treated with compounds by adding 10 μ l of a 50 mM stock prepared in water or 1% DMSO to the agar medium, avoiding contact with aerial parts. The final concentration of the precursor compound in the agar medium was 1 mM. After a certain period of time (after 24 h, 48 h and 72 h) the aerial part was harvested carefully, weighed and extracted in H 2 O:MeOH (1:1) under liquid nitrogen. The crude fresh plant extract thus obtained was analysed by mass spectrometry and HPLC.
For in planta T6P release experiments, compound-treated plants were exposed to UV-light treatment after 72 h. UV treatments consisted of: (a) 8-h exposure to natural daylight; (b) 8-h exposure to a 100 W UV spotlight (BlackRay B-100AP) at a distance of 18 cm; (c) 8-h exposure to an 8 W UV bulb (365 nm, Gelman transilluminator Model 51438) at a distance of 6 cm; or (d) exposure for two 8-h periods to 8 W. UV treatments were in addition to normal growth lights. Control plants (except for daylight treatment) were treated under the same conditions but without UV light. At the end of the day of exposure to UV or visible light, the aerial parts of the plants were quickly harvested, weighed and frozen in liquid nitrogen. For starch extractions, a moderate light regime was selected of a single 8 h exposure to 8 W UV light. After irradiation plants were returned to the growth room for a further day (day 5) to recover after light treatment and to respond to altered T6P levels before being harvested as above. Frozen tissue was stored at − 80 °C until extracted.
Harvested plant material was extracted by liquid/liquid extraction (LLE) followed by solid phase extraction (SPE) for T6P analysis 17 . For LLE/SPE extractions around 25 mg plant tissue was used, pooled from several plants. Samples were reconstituted in 50 μ l of H 2 O:MeOH (1:1) and 10 μ l was used for T6P analysis; T6P release was determined using quantitative HPLC-MS (Quattro, Waters), with 2-deoxy-glucose-6-phosphate as a calibration internal standard.
Liquid chromatography tandem mass spectrometry (LC-MS/MS) was used to confirm the identity of disaccharide monophosphates via fragmentation pattern analysis performed on a Waters Xevo G2-S QTof (quadrupole timeof-flight) mass spectrometer coupled to a Waters Acquity Ultra Performance Liquid Chromatography (UPLC) system, and a Waters Micromass Quattro micro API Mass Spectrometer coupled to a Waters 1525μ binary HPLC pump and a Waters 2777 auto sampler using a SIELC Primesep SB column. Solvent A (0.1% formic acid in H 2 O) and solvent B (1.0% formic acid in H 2 O:CH 3 CN (75:25)), were used as the mobile phase at a flow rate of 0.4 ml min −1 . For the Xevo G2-S QTof MS, the electrospray source was operated with a capillary voltage of 2.0 kV and a cone voltage of 30 V. Nitrogen was used as the desolvation gas at a total flow of 800 l h −1 . The intact molecular ion of T6P was detected as m/z 421.0759 (C 12 H 22 O 14 P, calculated as 421.0753) in a negative ion mode. The time of flight (ToF) tandem mass spectrum of the parent ion 421.00 was then obtained in a negative ion mode for the m/z range from 50 to 500 using optimized collision energy of 20 eV. For the Quattro micro API MS, the electrospray source was operated with a capillary voltage of 3.0 kV and a cone voltage of 40 V. The quadrupole tandem mass spectra of the parent ion 421.0 of T6P and S6P were obtained in a negative ion-mode for the m/z range from 50 to 500 using collision energy of 20 eV. With the reference to standard fragmentation patterns, tracking of the fragment ions of T6P in the plant sample was also performed by quadrupole tandem mass spectra. The five most intense m/z peaks recorded in the MS/MS spectrum of T6P; m/z 78.3, 96.4, 138.6, 240.9 and 421.0 (unfragmented) were also selected for multiple reaction monitoring (MRM) and cross-referenced with selected ion recordings (SIR) for the intact molecular ion m/z 421.0.
For seedling liquid culture, seeds of A. thaliana were grown in liquid culture as described previously 21 . Once the seedlings were 7 days old, oNPE-T6P (compound 3) or DMNB-T6P (compound 2) were added to the growth medium to a final concentration of 1 mM. Plants were left under growth lights to take up the compounds for 24 h. To facilitate precursor release, plants were placed under 23 μ mol m −2 s −1 UV for 2 h, after which they were returned to previous environmental conditions. Samples were taken for analysis before addition of the compound, 1 day after addition, after 30, 60 and 120 min during UV treatment, and sampled again at 1 and 2 d after UV treatment. Samples were weighed, snap-frozen and stored at − 80 °C.
For enzymatic sugar analysis, sugars were extracted from 5-10 mg of A. thaliana ground under liquid nitrogen, 1 ml of 80% was added and the sample was heated at 100 °C for 1 h, samples were centrifuged for 10 min at 13,000g to remove debris. The samples were added to assay buffer 24 . Enzymatic reactions were performed as described previously 32 using hexokinase, glucose-6-phosphate dehydrogenase, phosphoglucose isomerase and invertase from Sigma-Aldrich (H4502, G8404, P5381 and I9274, respectively). Two technical replicates were carried out for each sample, a total of three biological replicates were analysed. See Supplementary Methods for further details. Extraction and measurement of starch in planta. Three or four chemical-and UV-light-treated plantlets were pooled and weighed (fresh weight, 70-100 mg) for each biological replicate. Extraction was based on literature methods 24 . Samples were ground in liquid nitrogen to a fine powder in a mortar. The powder was rapidly extracted with 1 ml 80% ethanol at 80 °C, followed by 2 × 0.5 ml to rinse, samples were then transferred to a 2-ml eppendorf at 100 °C and heated for 2-3 min until just boiling. Tubes were transferred to a water bath at 80 °C while other samples were accumulated. Samples were centrifuged at 13,000g for 10 min to collect all solid material. The pellet was extracted twice more with 2 ml, 80% hot ethanol. The pellet was washed with 1 ml water, the supernatant removed and 100 μ l water added. The pellet was homogenized to a smooth consistency with an Eppendorf micropestle before being made up to final volume of 500 μ l with water. Samples were heated at 100 °C for 10 min to gelatinize starch granules. Duplicate aliquots (100 μ l) were removed and digested with α -amylase (2 U) and amyloglucosidase (6 U) in 0.05 M sodium acetate pH 4.8 for 4 h at 37 °C. Control digests without enzyme were also set up. Glucose released from digested starch was measured using an enzymatic assay coupled to the reduction of NADP to NADPH 33 and adapted for a microtitre plate reader. 10-20 μ l of digest was assayed in triplicate. Starch content is expressed as hexose equivalents per gram fresh weight. See Supplementary Methods for further details. SnRK1 activity. Kinase activities were determined by measuring the incorporation of radiolabelled phosphate into the ' AMARA' peptide (AMARAASAAALARRR; Biomol International) substrate and were carried out as described previously 30 . ADP-glucose pyrophosphorylase activity. Enzyme activity was measured as described previously 34  content analysis, grain was harvested at maturity for analysis. Chlorophyll content of leaves was measured by methanol extraction and spectrophotometry 35 . Starch content of grain was measured enzymatically 24 and protein content was measured by Bradford's assay 36 . For the drought treatment, vegetative Cadenza wheat plants were grown in the same compost and environments as above. Once the plants had reached Feekes stage 4, water was withheld for 10 days. On day 9, 30-ml, 1 mM solutions of oNPE-T6P (compound 3) and DMNB-T6P (compound 2) were applied to all above-ground biomass, on day 10 the watering schedule was reinstated. Plants were harvested to measure biomass production every 5 days for 30 days after rewatering. Both experiments were completed in biological replicates of six.
For quantification of T6P, trehalose and sucrose in wheat samples, the harvested wheat grains were weighed, snap-frozen and stored at − 80 °C. Wheat grain was ground to a fine powder in liquid nitrogen and the sugars were extracted by LLE for T6P, trehalose and sucrose analysis using the same LC-MS quantification method as for A. thaliana.
For minimal spray application, spring wheat (T. aestivum Cadenza) seeds were sown in Rothamsted standard compost mix and grown in controlled environment conditions with a photoperiod of 16 h light, 8 h dark, day/night temperatures of 20 °C/16 °C, photon-flux density of 600 μ mol m −2 s −1 and ambient relative humidity. Once the plants had reached anthesis, solutions of compounds 2 or 3 (1 mM and 10 mM) and a water control, were made up in distilled water with 0.1% TWEEN-20. At 10 days after anthesis, the top 20 cm of above ground biomass encompassing ears and flag leaves were sprayed individually with 25 ml of the compounds. Grain from individual ears was harvested at maturity for analysis. All wheat experiments were repeated twice, with 3 technical and 6 biological replicates completed at each stage of analysis. RNA extraction, cDNA synthesis and qRT-PCR. Total RNA was extracted from 50 mg snap-frozen leaf tissue from A. thaliana Columbia using the Ribopure Kit (Ambion) according to the manufacturer's instructions. RNA was quantified using a Nanodrop spectrophotometer and integrity of RNA was visualized using denaturing agarose gel electrophoresis 37 . DNA was removed using RQ1 RNase-free DNase (Promega). cDNA was synthesized using SuperScript III First-Strand Synthesis System (ThermoFisher Scientific) using 2 μ g of total RNA and oligo-dT primers according to the manufacturer's instructions. Gene expression was quantified using SYBR Green chemistry on a Real-Time PCR system 7500 (Applied Biosystems). Total reaction size was 20 μ l, containing 10 μ l SYBR Green Jumpstart Taq ReadyMix (Sigma Aldrich), 2 μ l cDNA and 0.5 mM primers. PCR used an initial denaturation stage of 95 °C for 2 min, followed by 40 cycles of 95 °C for 15 s, 60 °C for 1 min. The specificity of products was confirmed by performing a temperature gradient analysis of products at temperatures ranging from 55 °C to 95 °C with 0.5 °C increments. Two technical replicates were completed for each sample, a total of three biological replicates were analysed. Relative quantification of gene expression was performed using the Livak method using ubiquitin-transferase family protein as the reference gene. Primers used for SnRK1 marker gene expression, and starch gene expression are listed in Supplementary Tables 19 and 20, respectively. Mass spectometry imaging methods. A. thaliana used in ToF-SIMS (timeof-flight secondary ion mass spectrometry) imaging were grown in Petri dishes on 0.5× Murashige and Skoog medium with 0.8% agar for 10 d, with a photoperiod of 16 h light, 8 h dark, day/night temperatures of 23 °C/18 °C and photon-flux density of 250 μ mol m −2 s −1 . Plants were then transferred to Petri dishes containing the same media supplemented with 1 mM of either compound 2 or 3 for 24 h, during which they remained under the previously stated growth conditions. After 24 h, the plants were exposed to UV light at 23 μ mol m −2 s −1 for 2 h to facilitate T6P release. Plants were left for 2 h to recover, frozen and dehydrated in a vacuum chamber before mass spectometry imaging analysis. Reference materials were drop-dried on clean substrates. The ToF-SIMS mass spectrometry imaging analysis was performed with the ToF-SIMS IV mass spectrometer (IONTOF, Muenster, Germany) from three leaves, a control, and one leaf each treated with compounds 2 or 3. A pulsed 25-keV Bi 3 + primary ion source was used as the analysis beam (pulse width, 23 ns, mass resolution, (m/Δ m), 5,000). Mass spectra of the reference material were obtained in positive and negative ion mode at a primary ion dose of 1.1 × 10 11 ions cm −2 . The leaf ion images were also collected in both polarities with a dose of 5.4 × 10 10 ions cm −2 . An electron flood gun was employed for charge compensation during the data acquisition. Mass spectrometry data was analysed in ION-TOF SurfaceLab 6.4 software and further processed in MATLAB and Origin Pro. Known melissic acid markers 19  MALDIMS imaging data were acquired using identically prepared leaf samples with a modified QSTAR XL Qq-ToF instrument (Sciex, Ontario, Canada) fitted with a Nd:YAG laser (Elforlight Ltd, Daventry, UK) operated at 1,000 kHz in positive ion mode with a fluence of approximately 205 J m −2 and a pixel size of 200 μ m × 200 μ m. The QSTAR was operated in continuous raster sampling mode. The sample was affixed to a stainless steel target plate with double-sided tape and sprayed with CHCA (5 mg ml −1 CHCA in 80% methanol, 0.1% TFA) using automated spray deposition (TM Sprayer, HTX Technologies, Carrboro, USA). Data were converted from the proprietary .wiff format into .mzML using AB MS Data Converter version 1.3 (Sciex). These mzML files were then converted to .imzML using imzMLConverter 38 and processed in custom-made MATLAB software (version R2014b, Math Works Inc, USA). Images are created by summing across the full-width, half-maximum of the peak of interest to give the intensity within each corresponding pixel. See Supplementary Methods for further details. Statistical methods. ANOVA was applied to data to test for differences between treatments. A natural log transformation was used where necessary to ensure constant variance. The GENSTAT statistical system was used for this analysis. Data availability. Primary data for Figs 2, 4, 5 and 6 are provided as spreadsheets. All other data are available on request.