Trehalose-6-phosphate signaling regulates lateral root formation in Arabidopsis thaliana

As the degrader of T6P (number 12 in SI Appendix, Fig. S1), TPP activities are a hallmark of T6P pathway activity in relevant tissues. A complete set of promoter-β-glucuronidase protein reporter lines was used (19) to perform a comparative analysis of all AtTPP expression patterns during LR development (SI Appendix, Fig. S2). We found that TPPA, TPPB, TPPH, and TPPI were expressed in cells of LR primordium (LRP) stages while TPPD, TPPG, and TPPJ were expressed in the tissues surrounding LR. TPPB was expressed from early LR formation onward, from stage I (SI Appendix, Fig. S2 and Fig. 1A), indicative for a potential key role. TPP-focused in silico analyses using the Visual LR Transcriptome Compendium (VisuaLRTC) (20) confirmed that TPPB is differentially expressed during LR formation in contrast to the other TPPs (SI Appendix, Fig. S3).

Next, to probe the effect of the removal of the TPP T6P-degrader activity, tppb-1 and tppb-2 seedlings from knockout and knockdown TPPB T-DNA insertion lines (21) (Fig. 1 B and C) were studied. Consistent with a putative role for TPPB and T6P, these showed higher LR branching density (22) (Fig. 1 D and F). Interestingly, in TPPB mutant plants, a greater density of early stages LRP was observed (Fig. 1E). Furthermore, to investigate whether TPPB function is required locally in LR cells, we generated tissue-specific CRISPR/Cas knockout (TSKO) lines (23) based on the expression of a CRISPR/Cas construct targeting TPPB with the promoter of GATA23 (pGATA23) commanded Cas9 expression, GATA23 is commonly used as an early LRP marker (24). Tissue-specific downregulation of TPPB resulted in an enhancement of LR formation in the root branching zone (Fig. 1G) without affecting the primary root length (SI Appendix, Fig. S4). In addition, we generated TPPB overexpressor transgenic lines (TPPBOE1, TPPBOE2, and TPPBOE3: Fig. 1 B and C) and found lower LR branching density (Fig. 1 D and F). Together, these results suggest that T6P levels are relevant for early LR formation and are during this process negatively controlled by TPPB.

Next, given the possible central role of T6P, we perturbed T6P levels in planta in multiple ways. First, we used the plant-permeable, high-intensity light (HL)-inducible T6P precursor, DMNB-T6P to directly increase T6P levels in planta, designed to mask T6P’s charge and increase its hydrophobicity (25). After irradiation by HL (25), T6P generated from DMNB-T6P at levels as low as 10nM promoted the LR branching density (Fig. 2 A and B); interestingly, there are more emerged LRs in the newly formed part of the primary root (“Section III” in SI Appendix, Fig. S5A). Moreover, following the photoconversion of DMNB-T6P to T6P in planta, there is a significant rise of LR density over time (Fig. 2C and SI Appendix, Fig. S5B), indicating a general promotive effect on LR formation in the root branching zone by higher T6P levels. Interestingly, direct modulation even with T6P itself was also seen, albeit at much higher concentrations (1 mM) with again, an increase in LR branching density (Fig. 2D) and promotion of early LR developmental stages (Fig. 2E) in the primary root, specifically in the root branching zone (SI Appendix, Fig. S6) without accelerating LRP development (SI Appendix, Fig. S7). In addition, we monitored the effect of T6P treatment on the establishment of “prebranch sites” (PBS), static sites of DR5:LUC expression in the primary root that contain cells that are competent to develop into future LR (26). We found that T6P promotes the establishment of PBS (Fig. 2G) without affecting the periodicity and amplitude in DR5:LUC oscillations in the elongation zone (SI Appendix, Fig. S8 A and B). At the effective T6P concentrations, 10,000 to 100,000-fold greater than required for DMNB-T6P, T6P accumulation in roots of the treated seedlings could indeed be detected (SI Appendix, Fig. S9), although it should be noted that this, in part, could also reflect direct cell-surface adsorption under these conditions. Consistent with a T6P-specific mechanism, using an inhibitor of trehalase: validamycin A (27) or an alternatively phosphorylated trehalose-4-phosphate (T4P) instead had no measurable effect on LR density in the root branching zone upon application (Fig. 2F and SI Appendix, Fig. S10). The T6P-driven LR induction was found to be moderated by T6P-degrader TPPB. Consistent with a central role for T6P, TPPB overexpression and knockout transgenic lines inhibited DMNB-T6P-mediated LR-induction (Fig. 2H). Moreover, this modification did not affect the primary root lengths (SI Appendix, Fig. S11).

Finally, we generated two independent pGATA23:otsA lines (Fig.3 A and B) containing the promoter of GATA23, fused to the coding sequence of otsA for the Escherichia coli TPS capable of promoting T6P production in planta (28). We observed that pGATA23:otsA lines displayed higher LR branching density (Fig. 3C). pGATA23:otsA plants also exhibit perturbation in the expression of some SnRK1 target genes (SI Appendix, Fig. S12A). Moreover, both lines had a higher density of non-emerged LRP as well as emerged LRs (Fig. 3D) without affecting primary root length (SI Appendix, Fig. S12B). Together, these varying modes of targeted in planta T6P generation coherently demonstrated that higher levels of T6P are sufficient and required to promote LR formation.

It has recently been shown that SnRK1 is involved in LR formation during the acute stress of short-term energy deprivation (17) – SnRK1 is a known metabolic stress sensor (9). However, any broader fundamental role of this master kinase in the formation of an LR has not been described. In silico analyses (20) of the expression profiles of the catalytic subunits of SnRK1—KIN10 and KIN11—showed that both are regulated during LR formation (SI Appendix, Fig. S13A) and that notably these are down-regulated specifically in LRP cells (SI Appendix, Fig. S13B) based on an LR single-cell transcriptome study (29). Prompted by such an indicative link, we found that kin10 and kin11 mutant plants both displayed higher LR densities than wild-type in the root branching zone (Fig. 4 A and C and SI Appendix, Fig. S14A), while the respective overexpression lines, KINO1 and KINO2, displayed lower emerged LR densities in the root branching zone (Fig. 4 B and D and SI Appendix, Fig. S14B). This indicated that energy balance control of LR development relies on a low SnRK1 activity. This is supported by the generated TSKO-KIN10 lines (23) through the expression of a guide RNA targeting KIN10 and Cas9 expression under the control of pGATA23 (24). This TSKO-KIN10 system targeted KIN10 reduction to early LR stages including LRFCs in this way drove the formation of more LRs in the root branching zone (Fig. 4 E and F and SI Appendix, Fig. S14C). These results strongly suggested that a decrease of SnRK1 activity in early LR developmental stages (as would be caused by SnRK1 inhibitor T6P) is sufficient to achieve the necessary carbon energy balance signaling to trigger LR formation.

To evaluate the power of T6P to drive the SnRK1 signaling pathway, we again used T6P and T6P precursors directly. Additionally, we employed SnRK1-marker genes as previously identified in reference (30), whose expression profiles were shown to be mediated by KIN10 that positively correlates with sugar and carbon deprivation. Expression analysis of validated downstream SnRK1 target genes, such as UDPGDH, TPS5, and bZIP11 (repressed by SnRK1) and ASN1 and TPS8 (induced by SnRK1) (14) in roots treated with T6P (1 mM) or light-activated precursor DMNB-T6P (10 nM under HL), respectively, showed up- and downregulation in a manner consistent with T6P inhibiting SnRK1 activity in roots (Fig. 4 G and H). Interestingly, T6P-driven LR induction apparently requires only threshold reduction of levels of SnRK1 activity since kin10, kin11-1, and kin11-2 mutant seedlings exposed additionally to light-activated DMNB-T6P did not result in an even greater LR branching density (Fig. 4I). Moreover, promotion of SnRK1 activity in KINO1 and KINO2 seedlings does not allow T6P to induce LR formation in the root branching zone (Fig. 4J). All these phenotypes occurred without affecting primary root length (SI Appendix, Fig. S15)

When plants experience a low metabolic sugar status, SnRK1 is activated. On the other hand, in the high metabolic sugar status, TOR is induced and promotes growth (11). Interestingly, it is known that aspects of root architecture can be modified when TOR signaling is repressed (31), and recently, the TOR complex has been recognized as a gatekeeper for postembryonic root branching involving auxin-dependent pathways (18). Moreover, we found that expression of genes encoding for the TOR subunits RAPTOR1A and RAPTOR1B (20) is transcriptionally regulated during LR formation (SI Appendix, Fig. S13A) and, indeed, more strongly expressed in LRP cells as based on a recently reported LR single-cell transcriptome study (SI Appendix, Fig. S13B) (23). Second, using a similar tissue-specific strategy that we developed for TPPB and KIN10 (see above), we generated TSKO-TOR lines and found that they displayed lower LR branching densities (Fig. 5A and SI Appendix, Fig. S16 A and B); these results are consistent with a need for a high TOR activity locally in early LR stages including LRFCs to promote LR formation. Next, to assess whether T6P has any effect on TOR, we observed that addition of T6P slightly promoted TOR phosphorylation (Fig. 5B) and TOR transcription in roots (Fig. 5C DMNB-T6P in Fig. 5D). Finally, we also found that boosting T6P levels in roots was not sufficient to promote new LR biogenesis in raptor1b mutants nor in TOROE plants (Fig. 5E), again suggesting TOR interaction without affecting primary root lengths (SI Appendix, Fig. S17).

In addition, we evaluated the role of auxin, as the main regulator, in the T6P-LR induction. VisuaLRTC data suggest that TPPB might be repressed by auxin during early LR development in an INDOLE-3-ACETIC ACID 14 (IAA14)/SOLITARY ROOT 1 (SLR1)—AUXIN-RESPONSIVE FACTOR (ARF)7/19-dependent manner (20). This prediction was further confirmed through the analysis of TPPB expression in auxin-treated seedlings showing that TPPB was strongly down-regulated by auxin (SI Appendix, Fig. S18B), and this repression occurs in LR cells (SI Appendix, Fig. S18A). TPPB has an auxin-responsive element (TGTCTC, 15 bp upstream of the start codon) in its promoter, while the translational inhibitor cycloheximide did not block TPPB downregulation by auxin (SI Appendix, Fig. S18D), meaning that it might represent a direct target of auxin signaling. Moreover, this downregulation was partially reduced and abolished, respectively, in slr1 and arf7arf19 mutant plants (SI Appendix, Fig. S18C), thereby establishing that TPPB repression by auxin seems to be dependent on the IAA14/ARF7ARF19 auxin response module. This also proposes a temporally auxin-induced promotion of T6P levels at the onset of LR formation.

Together, these data suggest that T6P-driven LR induction makes use of TOR as a positive energy regulator and that the boosting effect of T6P on LR formation relies on both an inhibitory relationship with SnRK1 and a positive effect of TOR, thereby revealing the existence of an energy balancing mechanism from T6P to TOR in the process of LR formation that could be regulated by auxin (Fig. 6).

Seeds from different lines were sterilized by a chlorine-gas method (45), sown, and germinated in square plates that contain 0.5× Murashige and Skoog (MS) salt mixture medium with MES (0.5 g/L), Myo-inositol (0.1 g/L), and solidified with Phytoagar (8 g/L) or Phytagel (6 g/L) in the case of DMNB-T6P treatment. Seeds were stratified in a dark room at 4 °C for 48 h, and then, plates were put vertically in the in vitro culture room under continuous light (100 µmol/m²s⁻¹) at 22 °C until investigation.

For detailed analysis of TPP expression during LR formation, the available developmental stages in roots at six DAG of pTPP::GUS seedlings were analyzed. β-glucuronidase(GUS) staining assays were performed as described in ref. 46. After 24 h, the reaction was stopped, and the seedlings were mounted on glass microscopy slides in 90% lactic acid (Merck) to clear the tissue for inspection of primordia. Detailed staging of LRs was done as described in ref. 1. Samples were examined by differential interference contrast microscopy (BX53; Olympus).

For NAA treatments, 6 DAG of pTPPB::GUS or five DAG Col0, slr-1 and/or arf7arf19 seedlings were transferred to 0.5XMS solid medium without sucrose supplemented with 10 µM NAA and 0.1% DMSO in control plates. For Cycloheximide (CHX) treatments, 5 DAG Col0 seedlings were treated for 2 h with 10 µM NAA or control (0 µM NAA, DMSO) and pretreated 30 min with 50 µM CHX. Trehalose (Tre) and Validamycin A (Val) treatments performed by transferring 2 DAG Col0 seedlings to 0.5XMS solid medium without sucrose supplemented with or without Trehalose dihydrate (10 µM, 100 µM, 1 mM, or 10 mM from a 130 mM stock solution diluted in water, Sigma-Aldrich) and/or Val (10 µM from a 10 mM stock solution diluted in water, Sigma-Aldrich). For the T6P or T4P treatments, 2 DAG Col0 seedlings were transferred to plates with 0.5XMS solid medium without sucrose supplemented with or without trehalose 6(or 4)-phosphate dipotassium salt (1 mM from 100 mM stock solution in water) synthesized in the Department of Biotechnology, Centre for Synthetic Biology, Ghent University, Ghent, Belgium.

DMNB-T6P was synthesized according to previous methods (25). 2DAG Col0, Ler, TPPB-1, TPPBOE1, kin10, kin11-1, kin11-2, KINO1, KINO2, raptor1b-1, raptor1b-2, TOROE, and pCYCB1;1::GUS seedlings were transferred to 0.5XMS solid medium without sucrose containing 10 nM of DMNB-T6P dissolved in DMSO and with 0.1%DMSO used as control. The seedlings were grown for 3 d under these treatments in standard growth conditions. After treatment, half of the plates from both conditions were exposed to HL treatments (600 µmol/m²s⁻¹) for 7 h. Then, the plates were returned to normal growth conditions until analysis. DAHL was used to identify the effect of the DMNB-T6P to T6P conversion on root growth.

To quantify the prebranching sites, we used 2DAG pDR5:LUC seedlings that were transferred to T6P 1 mM in 0.5XMS solid medium for 6 d. The seedlings were sprayed with 1 mM D-Luciferin (Duchefa Biochemie) solution [D-Luciferin dissolved in 0.01% (v/v) Tween80 and 0.1% (v/v) DMSO]. Then, they were kept in darkness for 10 min to allow the appropriate absorption of D-Luciferin by the roots (47). Emitted luminescence was captured using ANDOR iKon-M 934 charge-coupled device (CCD) camera (Oxford Instruments) paired with a fixed lens (Spacecom 43F2409M-MP C 4/3” 24 mm F0.9) over a 20-min exposure time. Bright-field images of the roots were taken in each condition to be able to measure the primary roots and quantify the density of DR5:LUC sites. DR5:LUC oscillation frequencies and amplitudes were determined via 22-h time-lapse assays. 3DAG seedlings were transferred to plates with or without 1mM T6P, sprayed with 1 mM D-luciferin, and incubated in the dark for 1 h. Next, emitted luminescence was captured every 15 min with a 10-min exposure time using a NightSHADE LB985 in vivo plant imaging system (Berthold technologies) equipped with a deep-cooled slow scan CCD camera and the accompanying lens (Andor Instruments). Based on these time-lapses, the amplitude of and period between oscillation peaks in the oscillation zone were determined using ImageJ

To generate the pGATA23::otsA transgenic line, Gateway Technology (Invitrogen) was used. Genomic DNA was extracted from E.coli (strain K12) using the alkaline lysis method. Q5® High-Fidelity DNA Polymerase (NEB) and the primers indicated in SI Appendix, Table S3 were used to amplify the otsA gene and then cloned in the pDONR221 donor vector. pGATA23 was available already in the entry vector pEN-L4-pGATA23-R1 generated in ref. 17. Gateway LR recombination reaction was performed using pK7m24GW2 as a destination vector, obtained from the VIB Gateway vectors database (https://gatewayvectors.vib.be/). Then, Col0 plants were transformed with the final expression clones by Agrobacterium tumefaciens via floral dip method (48). T1 transformant seeds were selected by sowing them in 0.5XMS solid medium supplemented with 50µg/mL Kanamycin. Then, the resistant seeds were transferred to Jiffy-7 pellets® and grown in a greenhouse until the T2 generation.

The number of emerged LR was quantified in each seedling using an M165C binocular microscope (Leica). We determined the LR densities in the root branching zone according to ref. 22. To perform LR staging, the protocol indicated in ref. 1 was followed. Then, LR stages were identified and counted using an Olympus BX53 DIC microscope with a x400 magnification. During the microscopic analysis, the start of the LR formation and branching zones were marked on the slides. To measure the root lengths of the root branching zone and LR formation zone, the plugin NeuronJ (49) of Fiji software was used to analyze digital images obtained from scanning the square plates with EPSON expression 11000 XL flatbed scanner. To determine the effect of T6P on the speed of LRP development, a bending assay was performed in which LR development was initiated via gravistimulation. 5DAG seedlings were transferred to medium with or without 1mM T6P and gravistimulated (90° counterclockwise rotation). At different time points after gravistimulation, 30 treated and untreated seedlings were collected and cleared for 30 min using a chloral hydrate solution (5M chloral hydrate, 3% HCl). Next, the developmental stage of the LRP in each root bend was determined using an Olympus BX53 DIC microscope.

Total RNA was extracted from root tissue using the ReliaPrep™ RNA Tissue Miniprep System (Promega) according to the manufacturer’s instructions. cDNA was prepared from 1 µg of total RNA with the qScript^® cDNA SuperMix (Quantabio). Gene expression was quantified using SYBR^® Green Mix (Roche) mediating a RT-qPCR in a LightCycler^® 480 (Roche). cDKA1;1 and EEF1α4 were used as housekeeping genes to analyze TPPB expression in transgenic lines and treatments. Relative expression levels in T6P/DMNB-T6P treatments, TOR, pGATA23:otsA#1, pGATA23:otsA#2, pGATA23:Empty, TSKO-KIN10#1, TSKO-TOR#1, and TSKO-TOR#2 transgenic lines were normalized using UBI10 and PP2A as reference genes. All reactions were performed in triplicates, and a total of three biological replicates (two for TSKO lines) were analyzed per sample. Primers used are listed in SI Appendix, Table S4.

The gRNAs were designed using the web tool CRISPOR.org (50) to target TPPB, KIN10, and TOR kinase catalytic sites (Primers are enlisted in SI Appendix, Table S4). Then, a single gRNA was inserted in the TSKO destination vector pFASTR-pGATA23-Cas9-P2A-mCherry-G7T-AtU6-BsaI-CmR-ccdB-BsaI-gRNA scaffold generated as in ref. 23. For primers, see SI Appendix, Table S5. Following this, Arabidopsis plants were transformed by floral dip using A. tumefaciens with the destination vector including the designed gRNA (48). Next, the FASTR-positive seeds from T1 were selected by red fluorescence detection under a Leica M165FC fluorescence stereomicroscope. They were sown in 0.5XMS solid medium without sucrose and checked under the Leica SP8X confocal microscope for the presence of Cas9-mCherry in early stages of LR. The positives for Cas9-mCherry were transferred to jiffy and grown in a greenhouse. Lines that showed a 3:1 segregation FASTR positive:negative were selected for further propagation to generate a T3 population and to analyze their phenotype.

6DAG Col0 seedlings were used. The seedlings were pretreated for 1h in 0.5XMS liquid medium without sucrose. Then, 1 mM T6P or 15 mM sucrose was added to the medium for 20 min. Next, the seedlings were collected in liquid nitrogen. Plant material was ground and extracted as indicated in ref. 51. Protein total amount in the samples was quantified using the Qubit protein assay kit (ThermoFisher). Thirty micrograms of total protein per sample was loaded, and phospho-mTOR (Ser2448) antibody (#2971, 1:1,000, Cell Signaling Technology) was used to detect TOR phosphorylation. Then, the membranes were stripped using a 1:1 (v/v) 10% SDS and 100mM glycine–HCl (pH 2.5) solution. Next, the membranes were reblotted with mTOR antibody (#2,972, 1:1,000, Cell Signaling Technology). Horseradish peroxidase–conjugated anti-rabbit was used as secondary antibody and visualized using Western Lightning Plus ECL (PerkinElmer). The ChemiDoc XRS+ imaging system (Bio-Rad) was used to visualize the blots. To quantify the pTOR signal, the band intensity was measured with ImageLab software (v.6.0.0, Bio-Rad) and then normalized by the mTOR signal in each treatment.

Triethylamine (11.6 mL, 83.1 mmoL) was added to a stirred solution of PCl₃ (2.27 mL, 26 mmoL) in diethyl ether (160 mL) at 0 °C. A solution of compound 4 (2.6 g, 2.6 mmoL) in diethyl ether (100 mL) was added. The resulting reaction mixture was stirred for 30 min at 0 °C and then at room temperature. After 24 h, it was cooled to 0 °C, and benzyl alcohol (16.2 mL, 156 mmoL) was added. Stirring was continued for 30 min at 0 °C and then for 4 d at room temperature. Pyridine (25.1 mL, 312 mmoL) and acetic anhydride (14.7 mL, 156 mmoL) were added, after which the reaction mixture was stirred for 24 h. Subsequently, H₂O₂ (30% aq, 26 mL) was added, followed by stirring for 24 h. The reaction mixture was then cooled to 0 °C, and Na₂S₂O₃ (10% aq, 260 mL) was added. After stirring for 30 min at 0 °C, the mixture was transferred to a separation funnel containing H₂O (250 mL). The organic phase was separated, after which the aqueous layer was extracted with CH₂Cl₂ (2 × 250 mL). The combined organic layers were dried over Na₂SO₄. After filtration, concentration in vacuo, and partial purification by flash column chromatography (gradient elution: hexane/EtOAc 100/0 to 60/40), compound 5 was yielded as a mixture with tribenzyl phosphate. This mixture was dissolved in THF (26 mL). Tetra-n-butylammonium fluoride (1M in THF, 7.8 mL, 7.8 mmol) was added, followed by stirring overnight. The reaction mixture was transferred to a separation funnel containing KHSO₄ (1M in H₂O, 250 mL). The aqueous layer was extracted with EtOAc (3 × 250 mL). The combined organic layers were dried over anhydrous Na₂SO₄. After filtration, concentration in vacuo, and flash column chromatography (gradient elution: hexane/EtOAc 60/40 to 40/60), compound 6 was obtained (1.3 g, 1.2 mmol, 44%). To a solution of 6 (574 mg, 0.502 mmol) in MeOH (5 mL) and EtOAc (2 mL), palladium on carbon (10%, 53 mg, 0.050 mmol) was added. A balloon containing H₂ gas was attached to the flask. The reaction mixture was stirred at room temperature for 2 d under H₂ atmosphere, and subsequently filtered over celite and concentrated in vacuo, affording trehalose-6-phosphate as a white powder in a quantitative yield (212 mg, 0.502 mmol). To a solution of trehalose-6-phosphate (212 mg, 0.502 mmol) in H₂O (20 mL), KOH (0.05 M in H₂O, 20.1 mL) was added. The mixture was subjected to lyophilization, giving compound 7 (trehalose-6-phosphate dipotassium salt trihydrate) as a white powder in quantitative yield (278 mg). Compound 11 was obtained from 10 in a similar reaction procedure as for compound 5. Compound 12 was afforded via the general procedure described for the synthesis of compound 7. For additional details, see SI Appendix, Supplementary Materials and Methods.

5DAG Col0 seedlings were transferred to liquid 0.5XMS without sucrose as control or supplemented with 1 mM T6P for 3 d incubated in normal growth conditions and at harvest washed in distilled water three times for 15 min each time. Then, the root tissue was collected, weighed, stored at −70 °C, and ground to a fine powder in liquid nitrogen. T6P was extracted as in ref. 52. Ten microliters of each sample was loaded into a Dionex UltiMate 3,000 LC System (Thermo Scientific) equipped with a C-18 column (Acquity UPLC -HSS T3 1.8 µm; 2.1 × 150 mm, Waters) coupled to a Q Exactive Orbitrap mass spectrometer (Thermo Scientific) operating in negative ion mode. A step gradient was carried out using solvent A (10 mM TBA and 15 mM acetic acid) and solvent B (100% methanol). The gradient started with 5% of solvent B and 95% of solvent A and remained at 5% B until 2 min after injection. A linear gradient to 37% B was carried out until 7 min and increased to 41% until 14 min. Between 14 and 26 min, the gradient increased to 95% of B and remained at 95% B for 4 min. At 30 min, the gradient returned to 5% B. The chromatography stopped at 40 min. The flow was kept constant at 0.25 mL/min, and the column was placed at 40 °C throughout the analysis. The MS operated in full scan mode [m/z range: (70.0000 to 1050.0000)] using a spray voltage of 4.80 kV, capillary temperature of 300 °C, sheath gas at 40.0, and auxiliary gas at 10.0. The AGC target was set at 3.0E + 006 using a resolution of 140,000, with a maximum IT fill time of 512 ms. Data collection was performed using the Xcalibur software (Thermo Scientific). Data analysis was performed by integrating the peak areas (El-Maven—Polly—Elucidata). The T6P concentration was normalized with the fresh weight of each sample.

Statistical analyses were performed in R (53) or SAS (version 9.4 of the SAS System for Windows). For LR branching and DR5:LUC site density data, a Poisson model (or a quasi-Poisson model in case of overdispersion) was fitted to the LR or DR5:LUC site counts, with genotype and/or treatment (and if applicable their interaction) as fixed effects. A log-link function was applied, and log-transformed primary root lengths were used as an offset variable. LR numbers were analyzed in the same way without the use of an offset variable. Statistical analyses of primary root lengths were performed via an ANOVA or a Kruskal–Wallis rank-sum test as a nonparametric alternative, with genotype and/or treatment (and if applicable their interaction) as fixed effects. Where applicable, post hoc Tukey and Dunnett’s tests were set up using the “emmeans” package (54). In those cases where root lengths were only compared between two conditions, as well as for the comparison of DR5:LUC oscillation frequencies and amplitudes, two-tailed Student’s t tests or Mann–Whitney tests (nonparametric alternative) were performed. For the analysis of LR developmental stages, a GEE model was fitted to the primordium count rate with genotype and developmental stage, as well as their interaction, as fixed effects using a log-link function and selecting a Poisson distribution. Log-transformed root lengths were used as an offset. The correlations between the counts were modeled as exchangeable correlations. At each stage, we tested whether there was an equal primordium count rate in the mutant or treatment compared with the appropriate control. The analysis was done with the genmod procedure (in SAS). Contrast statements were set up with the plm procedure using the lsm estimate statement. To correct for multiple testing, the maxT procedure was used as implemented in the plm procedure. To determine statistical differences between the proportions of LRP at certain developmental stages in gravistimulation-induced root bends in control or T6P-treated seedlings, Chi-square tests were performed.

S.M.-H., L.L.-G., P.V.D., and T.B. designed research; S.M.-H., J.J., F.C., L.L.-G., A.S., and M.N. performed research; T.D.S., C.A.G., J.D.S., J.S.O.M., X.X., B.G.D., J.V.d.E., and M.J.P. contributed new reagents/analytic tools; S.M.-H, J.J., F.C., and L.L.-G. analyzed data; and S.M.-H., B.G.D., M.J.P., and T.B. wrote the paper.

C.A.G., B.G.D., and M.J.P. are founders, shareholders and directors of SugaROx, which holds rights to the commercial exploitation of DMNB-T6P; no part of this study has involved SugaROx. Other than that, the authors declare no competing interests.