Abstract
Growth at the shoot apical meristem (SAM) is essential for shoot architecture construction. The phytohormones gibberellins (GA) play a pivotal role in coordinating plant growth, but their role in the SAM remains mostly unknown. Here, we developed a ratiometric GA signalling biosensor by engineering one of the DELLA proteins, to suppress its master regulatory function in GA transcriptional responses while preserving its degradation upon GA sensing. We demonstrate that this novel degradation-based biosensor accurately reports on cellular changes in GA levels and perception during development. We used this biosensor to map GA signalling activity in the SAM. We show that high GA signalling is found primarily in cells located between organ primordia that are the precursors of internodes. By gain- and loss-of-function approaches, we further demonstrate that GAs regulate cell division plane orientation to establish the typical cellular organisation of internodes, thus contributing to internode specification in the SAM.
Introduction
The shoot apical meristem (SAM) at the tip of shoot axes comprises a stem cell niche whose activity produces lateral organs and stem segments in a modular iterative fashion during the whole plant life. Each of these repetitive units or phytomere includes an internode and lateral organs at a node and an axillary meristem at the leaf axil1. The growth and organization of phytomeres change during development. In Arabidopsis thaliana, internode growth is inhibited during the vegetative phase and axillary meristems rest dormant at the axils of rosette leaves. Upon floral transition, the SAM turns into an inflorescence meristem, producing elongated internodes and axillary buds that form branches at the axils of cauline leaves, and later flowers without leaves2. While substantial progress has been made on our understanding of the mechanisms controlling the initiation of leaves, flowers and branches, much less is known on how internodes are initiated.
Growth via cell division and expansion is essential for reiterative organogenesis at the SAM. The tetracyclic diterpenoid hormones gibberellins (GA) are key growth regulators3–5, with a crucial role in many embryonic and post-embryonic developmental processes6. Central to the GA signalling pathway are the five DELLA proteins, GIBBERELLIC ACID INSENSITIVE (GAI), REPRESSOR OF GA1-3 (RGA) and RGA-Like (RGL) 1-37. These nuclear proteins are composed of an N-terminal DELLA/TVHYNP domain and of a GRAS domain. The GRAS domain allows DELLAs to interact with diverse transcription factors and transcriptional regulators and to suppress growth by modulating their activity7. Binding of GA to the GIBBERELLIN INSENSITIVE DWARF1 (GID1) GA receptor promotes GID1 interaction with the N-terminal domain of DELLAs, triggering DELLA degradation by the ubiquitin-dependent proteasome pathway7–10, and in turn, de-repressing GA responses. Despite the relatively specialised role of the GRAS and DELLA/TVHYNP domains, residues required for DELLA degradation and partner protein interaction are widely distributed within the protein sequence.
The identification of genes encoding GA catabolic enzymes as direct targets of the class I KNOX meristem identity regulators has led to propose that GA levels are low in the SAM cells, while high GA concentrations trigger growth of lateral organs3–5. Low GA levels could then contribute to SAM maintenance3. However, more recent analysis indicate also that GAs promote the increase in SAM size during floral transition by regulating the division and expansion of inner SAM cells11, 12, in a similar manner as in the root13, 14. DELLAs were likewise found to limit meristem size by directly regulating the expression of the cell-cycle inhibitor KRP2 in the internal part of the SAM (the rib zone)12. Together, these findings support a role for GA in positively regulating cell division in the inner tissues of the SAM, and thus SAM size. At the same time, several genes encoding GA biosynthetic and catabolic enzymes are expressed specifically in lateral organs4, 11, illustrating a complex spatio-temporal GA distribution in the SAM to likely fulfil different functions.
Accessing spatio-temporal GA distribution has been instrumental to better understand functions of these hormones in different tissues and at various developmental stages. Accumulation of GA in the root endodermis and regulation of their cellular level via GA transport were discovered by using bioactive fluorescein (Fl)-tagged GA15, 16. More recently, the nlsGPS1 GA FRET sensor revealed that GA levels are correlated with cell elongation in roots, stamen filaments and dark-grown hypocotyls17. Here, building on the knowledge on the GA signalling pathway, we report on the engineering and characterization of a quantitative degradation-based GA signalling biosensor. We used this novel biosensor to map the spatio- temporal distribution of GA signalling activity and to quantitatively analyse how GAs regulate cell behaviour in the SAM epidermis. We demonstrate that GAs regulate the orientation of division planes of SAM cells located between organ primordia, therefore specifying the typical cellular organization of internodes.
Results
Modifying the RGA protein for sensor construction (Fig. 1; Extended Data Fig. 1; Supplementary Table 1)
To generate a degradation-based GA signalling biosensor, we engineered a DELLA protein to meet two criteria, (i) a specific degradation upon GA perception; and (ii) a minimal interference with GA signalling. To do so, we modified DELLAs to preserve the interaction with GID1 while abolishing interactions with partner transcription factors, by introducing mutations in the GRAS domain. Among the 5 Arabidopsis DELLAs, RGA displays one of the highest GA-dependent degradation rate18, and RGA-GFP fusions are widely used as GA signalling reporter8. Leveraging on the results of GRAS domain mutant analyses in rice19 and Arabidopsis20, 21, we generated four modified RGA versions (RGAm1 to RGAm4) and tested their ability to meet the above defined criteria (Fig. 1a, Extended Data Fig. 1a). We first used a yeast-two hybrid (Y2H) assay to test the binding capacity of these modified candidates with three well known DELLA interacting partners, JAZ1, TCP14 and IDD222. While RGAm1 had a minor effect on interactions, RGAm2, RGAm3 and RGAm4 lost their capacity to interact with these partners (Fig. 1b, Extended Data Fig. 1b). In addition, RGAm2, RGAm3 and RGAm4 were able to bind GID1 in the presence of GA, thus suggesting that these DELLA candidates are still degraded in response to GA (Fig. 1c, Extended Data Fig. 1c). To assess this possibility, we explored GA-dependent degradation of RGAm2, RGAm3 and RGAm4 fused to GFP in transient expression assays. These three candidates were degraded after GA treatment (Fig. 1d, Extended Data Fig. 1d), RGAm2-GFP having the fastest degradation kinetics although it was slightly more stable than RGA-GFP. Noteworthy, RGAm2 harbours three amino acid substitutions in the GRAS PFYRE motif (Fig. 1a), that is highly conserved in all plant DELLAs. Therefore, this variant is likely to have similar properties in plant species other than Arabidopsis.
Based on the above results, we selected RGAm2 for further analysis. We next showed that RGAm2 is unable to bind with the BZR1 DELLA interacting partner in yeast and a larger screen confirmed that these mutations abolish interactions with practically all known DELLA partners (Fig. 1b, Supplementary Table 1). Moreover, co-immunoprecipitation studies demonstrated that binding of RGAm2 to IDD2 is also strongly reduced in planta compared to RGA (Fig. 1e). Finally, transient expression assays confirmed that, while RGA represses BZR1 and TCP14 transcriptional activities, RGAm2 did not (Fig. 1f-g). Taken together, these data indicate that the expression of RGAm2 in plants, hereafter named RGAmPFYR, might have a limited impact on GA signalling. Thus, RGAmPFYR constitutes a suitable DELLA variant candidate for engineering a degradation-based biosensor that monitors GA signalling activity and more specifically the combinatorial effect of GA and of its complex perception machinery.
Engineering a GA signalling sensor (Fig. 2; Extended Data Fig. 2-6)
To create a ratiometric GA signalling sensor, we fused the RGAmPFYR protein to the fast maturing yellow fluorescent protein VENUS23 and co-expressed this fusion protein together with a nuclear-localized non-degradable reference protein, TagBFP-NLS, under the pUBQ1024 or pRPS5a25, 26 constitutive promoter. We used the 2A self-cleaving peptide to allow for a stoichiometric production of both fluorescent proteins, enabling quantification of GA signalling activity using fluorescence intensity ratio between RGAmPFYR-VENUS and TagBFP27, 28 (Fig. 2a). We named the sensor lines qRGAmPFYR (quantitative RGAmPFYR) and first analysed their TagBFP fluorescence pattern in vegetative and reproductive tissues (Fig. 2b and Extended Data Fig. 2a-b). TagBFP fluorescence of pUBQ10::qRGAmPFYR lines was homogeneously distributed in hypocotyl, the vegetative shoot meristem, cotyledons and roots (except in the root tip) but the TagBFP signal was unevenly distributed in the inflorescence SAM, with a stronger signal in organ boundaries (Extended Data Fig. 2a). Conversely, the TagBFP signal was strong in the root tip and the vegetative SAM in pRPS5a::RGAmPFYR lines (Fig. 2b), and showed a homogenous distribution in the inflorescence SAM (Extended Data Fig. 2b). Hence, the choice of one of these constructs depends on the analysed tissue: we used pUBQ10::RGAmPFYRin subsequent experiments in seedlings, and pRPS5a::RGAmPFYR for experiments performed in inflorescence SAM.
To test that qRGAmPFYR activity is indeed not interfering with signalling activity and thus with plant growth, we first investigated the effects of exogenous GA and paclobutrazol (PAC; an inhibitor of GA biosynthesis) treatments on the growth of qRGAmPFYR plants. We found that hypocotyl length of qRGAmPFYR seedlings was similar to that of wild-type under mock conditions and upon GA or PAC treatment, although hypocotyls were slightly longer after GA treatment (Fig. 2c). Similarly, shoot development and plant fertility were not significantly affected in qRGAmPFYR plants (Fig. 2d and Extended Data Fig. 2c). Last, we showed that while qd17RGA plants (d17RGA is a mutant version of RGA that is fully insensitive to GA29) exhibited a severe dwarf phenotype reminiscent of GA-insensitive mutants, qd17RGAmPFYR plants had similar rosette size and height than wild-type plants (Extended Data Fig. 2d-i). Altogether, these results demonstrate that qRGAmPFYR negligibly interferes with plant growth and GA responses.
Next, we assessed if qRGAmPFYR can detect changes in GA levels. Consistent with the above transient expression assays, GA treatment induced the degradation of RGAmPFYR in pRPS5a::qRGAmPFYRseedlings, although the protein tends to be slightly more stable than RGA (Fig. 2e,f). Accordingly, while the TagBFP signal was unaffected in hypocotyls of pUBQ10::qRGAmPFYR seedlings upon GA application, the sensor element RGAmPFYR-VENUS fluorescence was substantially reduced after this treatment (Fig. 2g). Similar results were observed in qRGAmPFYR roots for which GA and PAC application respectively reduced and increased RGAmPFYR-VENUS signal (Extended Data Fig. 3). In contrast, VENUS fluorescence did not change in any of the treatments in RGAm3-VENUS and RGAm4-VENUS lines, consistently with the lower GA-dependent degradation rate of these variants (Extended Data Fig. 3).
In further image analyses, to fully cover a highly variable range of values in VENUS/TagBFP fluorescence ratio, we used 3 – (RGAmPFYR-Venus/TagBFP) as a positive proxy for GA signalling activity (hereafter named “GA signalling”; Fig. 2a; see also Supplementary Method 1). This quantitative approach confirmed statistically significant changes in GA signalling in hypocotyls, with an increase and decrease respectively upon GA and PAC treatments compared to untreated seedlings (Fig. 2g,h; Extended Data Fig. 4a-d). Exogenous GA and PAC treatments induced similar responses in the SAM, although the effect of PAC was less pronounced than in hypocotyl (Extended Data Fig. 5). Furthermore, GA signalling activity increased in the SAM with both exogenous GA concentration and treatment duration (Extended Data Fig. 4e-i), showing that qRGAmPFYR is suitable to be used as a GA signalling sensor in the SAM as in all the other tissues tested.
Finally, we asked whether qRGAmPFYR is able to report for changes in endogenous GA levels using growing hypocotyls. We previously showed that nitrate promotes growth by increasing GA synthesis and in turn DELLA degradation30. Accordingly, we observed that hypocotyl length of pUBQ10::qRGAmPFYRseedlings grown on adequate nitrate supply (10 mM NO3-) was significantly longer compared to those grown on nitrate-deficient conditions (Extended Data Fig. 6a). Consistent with the growth response, GA signalling was higher in hypocotyls of seedlings grown with 10 mM NO3- compared with those grown in absence of nitrate (Extended Data Fig. 6b,c). Thus qRGAmPFYR also allows monitoring changes in GA signalling resulting from endogenous changes in GA concentration.
qRGAmPFYR fluorescence depends on GA receptor activity (Fig. 3 and Extended Data Fig. 7)
As GA signalling activity depends on both GA concentration and GA perception, we analysed the expression of the three GID1 receptors in vegetative and reproductive tissues. In seedlings, GID1-GUS reporter lines showed that GID1a and c are highly expressed in cotyledons (Fig. 3a-c). Moreover, all three receptors are expressed in leaves, lateral root primordia, root tip (excluding root cap for GID1b) and vasculature (Fig. 3a-c). In inflorescence SAM, we only detected a GUS signal for GID1b and 1c (Extended Data Fig. 7a-c). In situ hybridization confirmed these expression patterns and further demonstrated that GID1c is homogenously expressed at a low level in the SAM while GID1b shows higher expression at the SAM periphery (Extended Data Fig. 7d-l). A pGID1b::2xmTQ2-GID1b translational fusion further revealed a graded GID1b expression ranging from low or no expression in the SAM centre to high expression in organ boundaries (Extended Data Fig. 7m). Thus, GID1 receptors are unevenly distributed across and within tissues. In a subsequent experiment, we also observed that overexpressing GID1 (pUBQ10::GID1a-mCherry) enhanced qRGAmPFYR sensitivity to external GA application in hypocotyls (Fig. 3d). By contrast, the fluorescence measured from qd17RGAmPFYR in hypocotyls was insensitive to GA (Fig. 3e). Taken together, these results confirm that the qRGAmPFYR biosensor reports for the combinatorial action of GA and GA receptors and suggest that differential expression of GID1 receptors can modulate the emission ratio of the sensor.
A GA signalling map in the shoot apical meristem (Fig. 4; Extended Data Fig. 8-10)
Distribution of GA signalling within the SAM has remained elusive so far. Thus, we used plants expressing qRGAmPFYR together with the pCLV3::mCherry-NLS stem cell reporter to compute a high-resolution quantitative map of GA signalling activity, focusing on the L1 layer (epidermis; Fig. 4a-b, see Methods and Supplementary Method 1) given the key role of L1 in controlling growth at the SAM31. Here, expression of pCLV3::mCherry-NLS provided a fixed geometric reference for analysing the spatio-temporal distribution of GA signalling activity32. Although GAs were proposed to be required for lateral organ development4, we observed that GA signalling was lower in flower primordia (P) from the P3 stage onward (Fig. 4a-b), while young P1 and P2 primordia had intermediate activity similar to the one found in the central zone (Fig. 4a-b). Higher GA signalling activity was found in the boundaries of organ primordia, starting from P1/P2 (on the lateral sides of the boundary) and culminating from P4, and in all peripheral zone cells located between primordia (Fig. 4a-b; Extended Data Fig. 8a-b). This higher GA signalling activity was not only monitored in the epidermis, but also in the L2 and upper L3 layers (Extended Data Fig. 8b). This reveals a GA signalling pattern mostly mirroring primordia distribution. This inter-primordia-region (IPR) distribution results from the progressive establishment of a high GA signalling activity between developing primordia and the central zone, while in parallel GA signalling activity decreases in primordia (Fig. 4c-d).
The distribution of GID1b and GID1c receptors (Extended Data Fig. 7b-c, g-m) suggests that differential expression of GA receptors contributes to shape the GA signalling activity pattern in the SAM. We wondered if differential accumulation of GA could also be involved. To investigate this possibility, we used the nlsGPS1 GA FRET sensor17. An increased emission ratio was detected in nlsGPS1 SAMs treated with 10 µM GA4+7 for 100 min (Extended Data Fig. 9), indicating that nlsGPS1 responds to changes in GA concentration in the SAM as it does in roots17. The spatial distribution of the nlsGPS1 emission ratio indicates that GA levels are relatively low in the SAM external layers, but it shows that they are elevated in the centre of the SAM and in the boundaries (Fig. 4e, Extended Data Fig. 9a,c). We also treated SAMs with fluorescent GAs (GA3-, GA4-, GA7-Fl) or with Fl alone as a negative control. The Fl signal was distributed in the whole SAM, including the central zone and primordia, although with lower intensity (Fig. 4f, Extended Data Fig. 10d). In contrast, all the three GA-Fls specifically accumulated in primordia boundaries and, to different degrees, in part of the rest of the IPR, with GA7-Fl accumulating in the largest domain in the IPR (Fig. 4g, Extended Data Fig. 10). Fluorescence intensity quantification demonstrated higher IPR to non-IPR intensity ratio in the GA-Fl treated SAMs, compared to Fl-treated SAMs (Fig. 4h, Extended Data Fig. 10c). Taken together, these results suggest that GAs are present at higher levels in the IPR cells closest to organ boundaries. This indicates that the SAM GA signalling activity pattern results from both differential expression of the GA receptors and from the differential accumulation of GA in IPR cells closest to organ boundaries. Our analysis thus identifies an unexpected spatio-temporal GA signalling pattern, with a lower activity in the centre of the SAM and in primordia, while activity is elevated in the IPR of the peripheral zone.
Correlation analyses suggest a role for GA signalling in cell division plane orientation in the shoot apical meristem (Fig. 5)
To understand the role of differential GA signalling activity at the SAM, we analysed the correlation between GA signalling activity, cell expansion and cell division using time-lapse live imaging of qRGAmPFYR pCLV3::mCherry-NLS SAMs. Given the role of GA in growth regulation, a positive correlation was expected with cell expansion parameters. We thus first compared maps of GA signalling activity to those of cell surfacic growth rate (as a proxy for cell expansion intensity for a given cell and daughter cells if it divides) and growth anisotropy, which measures the directionality of cell expansion (here also for a given cell and daughter cells if it divides; Fig. 5a-b, see Methods and Supplementary Method 1). Our cell surfacic growth intensity map of the SAM was consistent with previous observations33, 34, with a minimal growth rate in boundaries and a maximal rate in developing flowers (Fig. 5a). A principal component analysis (PCA) showed an anti-correlation between GA signalling activity and cell surfacic growth intensity (Fig. 5c). It further showed that the main axis of variability, encompassing GA signalling input and growth intensity, was orthogonal to the direction defined by high expression of CLV3, which argues in favour of excluding cells from the centre of the SAM in the rest of the analysis. Spearman correlation analyses confirmed the PCA results (Fig. 5d), suggesting that higher GA signalling in the IPR does not lead to higher cell expansion. However, the correlation analysis demonstrated a mild positive association between GA signalling activity and growth anisotropy (Fig. 5c,d), suggesting that higher GA signalling in the IPR acts on cell growth orientation and possibly cell division plane positioning.
Thus, we next studied the correlation between GA signalling and cell division activity, by identifying newly formed cell walls during a time course analysis (Fig. 5e). This methodology allows us to measure both the frequency and orientation of cell divisions. Strikingly, we found that cell division frequency was similar in the IPR and the rest of the SAM (non-IPR, Fig. 5f), showing that differences in GA signalling between IPR and non-IPR cells do not have a major effect on cell division. This, together with the positive correlation between GA signalling and growth anisotropy, led us to ask whether GA signalling activity could act on cell division plane orientation. We measured the orientation of new cell walls as the acute angle relative to the radial axis connecting the centre of the meristem to the centre of new cell walls (Fig. 5g), and observed that cells had a clear tendency to divide at angles closer to 90° relative to the radial axis, with the highest frequency observed at 70-80° (23.28%) and 80-90° (22.62%) (Fig. 5h) i.e. corresponding to cell divisions oriented in the circumferential/transverse direction. To explore the contribution of GA signalling to this cell division behaviour, we analysed separately the cell division parameters in the IPR and non-IPR (Fig. 5h). We observed that the angle of cell division was different in the IPR compared to non-IPR or the entire SAM, with IPR cells showing a higher rate of transverse cell divisions, i.e. 70-80° and 80-90° (33.86% and 30.71% respectively) (Fig. 5h). Thus, our observations reveal a link between high GA signalling and cell division plane orientation that parallels the correlation between GA signalling activity and growth anisotropy (Fig. 5c,d). To further establish spatial conservation of this link, we measured the orientation of division planes in IPR cells above primordia starting at stage P3, given that the highest GA signalling activity is detected in this region from stage P4 (Fig. 4). The division angles of the IPR above P3 and P4 did not show statistically significant differences, although an increase in the frequency of transverse cell divisions was observed in the IPR above P4 (Fig. 5i). Differences in cell division plane orientation were however statistically significant in IPR cells above P5, in which the frequency of transverse cell divisions was drastically increased (Fig. 5i). Taken together, these results suggest that GA signalling could control orientation of cell division in the SAM coherently with previous reports35, 36, with high GA signalling likely inducing a transverse orientation of cell divisions in the IPR.
GA signalling activity positively regulates transverse cell divisions in the shoot apical meristem (Fig. 6; Extended Data Fig. 11-13)
Cells in the IPR are expected not to be incorporated into primordia but rather in the internodes2, 37, 38. A transverse orientation of cell divisions in the IPR could generate the typical organization in parallel longitudinal cell files of the internode epidermis. Our observations above indicate that GA signalling is likely to act in this process by regulating cell division orientation.
Loss-of-function of multiple DELLA genes leads to constitutive GA response, and thus della mutants could be used to test this hypothesis39. We first analysed the expression patterns of the five DELLA genes in the SAM. Transcriptional fusion GUS lines40 showed that GAI, RGA, RGL1, and, to a much lesser extent, RGL2 are expressed in the SAM (Extended Data Fig. 11a-d). In situ hybridization further showed that GAI mRNA specifically accumulates in primordia and developing flowers (Extended Data Fig. 11e). RGL1 and RGL3 mRNA were detected throughout the SAM dome and in older flowers, while RGL2 mRNA was more abundant in the boundary regions (Extended Data Fig. 11f-h). Confocal imaging of pRGL3::RGL3-GFP SAM confirmed the expression observed with in situ hybridization and showed that the RGL3 protein accumulates in the central part of the SAM (Extended Data Fig. 11i). Using a pRGA::GFP-RGA line, we also found that the RGA protein accumulates in the SAM, but its abundance is reduced in boundaries starting from P4 (Extended Data Fig. 11j). Notably, the expression patterns of RGL3 and RGA are compatible with a higher GA signalling activity in the IPR, as detected with qRGAmPFYR (Fig. 4). In addition, these data indicate that all DELLAs are expressed in the SAM and that collectively, their expressions cover the entire SAM.
We next analysed the cell division parameters in wild-type (Ler, control) and quintuple (global) gai-t6 rga-t2 rgl1-1 rgl2-1 rgl3-4 della mutants (Fig. 6a-b). Interestingly, we observed a statistically significant change in frequency distribution of cell division angles in the SAM of global della mutant compared to wild-type (Fig. 6c). This change in the global della mutant resulted from an increase in frequency of 80-90° angles (34.71 % vs 24.55 %) and to a lesser extent of 70-80° angles (23.78 % vs 20.18 %), i.e. corresponding to transverse cell divisions (Fig. 6c). The frequency of non-transverse divisions (0-60°) was also lower in global della mutant (Fig. 6c). The increased occurrence of transverse cell divisions was easily visible in the SAM of global della mutant (Fig. 6b). The frequency of transverse cell divisions in the IPR was also higher in global della mutant compared to wild-type (Fig. 6d). Outside of the IPR region, distribution of the angle of cell division was more homogeneous in wild-type, while in the global della mutant it was skewed toward tangential divisions, as in the IPR (Fig. 6e). We thus show that constitutive activation of GA signalling in the SAM induces transverse cell division both in the IPR and the rest of the SAM.
We then tested the effect of inhibiting GA signalling specifically in the IPR. To do so, we used the CUP-SHAPED COTYLEDON 2 (CUC2) promoter to drive expression of the dominant negative gai-1 protein fused to VENUS (in a pCUC2::gai-1-VENUS line). CUC2 promoter drives expression in a large part of the IPR (including the boundary cells) in the SAM starting from P4 (Extended Data Fig. 12k). The distribution of cell division angles in the entire SAM or in the IPR of pCUC2::gai-1-VENUS plants did not show statistically significant differences with respect to the wild-type, although unexpectedly, we found in these plants a higher frequency of 80-90° divisions in non-IPR cells (Fig. 6f-j).
The orientation of cell division has been proposed to be influenced by the geometry of the SAM and notably by tensile stresses prescribed by the curvature of the tissue41. We thus asked whether the SAM shape of the global della mutant and pCUC2::gai-1-VENUS plants was changed. As previously shown12, the size of the global della mutant SAM was bigger than wild-type (Extended Data Fig. 12a,b,d). However, the SAM curvature was identical in the two genotypes (Extended Data Fig. 12b,e,f,g,h,j). We observed a comparable increase in size in the quadruple gai-t6 rga-t2 rgl1-1 rgl2-1 della mutant, again without modification of the curvature compared to wild-type (Extended Data Fig. 12c,d,f,i,j). Frequency of cell division orientation was also affected in the quadruple della mutant, but to a lesser extent than in the global della mutant (Extended Data Fig. 13). This dosage effect, along with the absence of effects on curvature, suggests that the remaining RGL3 activity in quadruple della mutants limits the changes in cell division orientation caused by the loss of DELLA activity, and that changes in the occurrence of transverse cell division depend on changes in GA signalling activity rather than in SAM geometry. By contrast, the size of pCUC2::gai-1-VENUS SAMs was reduced, whereas it had a significantly higher curvature (Extended Data Fig. 12l-q). This change in the pCUC2::gai-1-VENUS SAM curvature could generate a mechanical stress distribution whereby high circumferential stress starts at a short distance from the SAM centre42. This could counteract in part the effect of GA signalling changes by increasing the probability of cell division with circumferential/transverse orientation, thus explaining our observations.
Taken together, our data support a positive role for higher GA signalling in transverse orientation of the cell division plane in the IPR. They further suggest that the curvature of the meristem can also influence the orientation of the cell division plane in the IPR.
High GA signalling initiates internode specification in the shoot apical meristem (Fig. 7; Extended Data Fig. 14)
Transverse orientation of division planes in the IPR as a result of higher GA signalling activity opens the possibility that GAs pre-organize radial cell files in the epidermis within the SAM to specify the cellular organization later found in the internode epidermis. Indeed, such cell files are often visible in the SAM images of the global della mutant (Fig. 6b). Therefore, to further understand the developmental function of the GA signalling spatial pattern in the SAM, we used time-lapse imaging to analyse the cell spatial organization in the IPR in wild- type (Ler and Col-0), global della mutant and pCUC2::gai-1-VENUS transgenic plants.
We marked Ler cells above and on the side of P4 according to their developmental fates (analysed 34 hrs after the first observation, i.e. more than two plastochrones) using three different colours: yellow for those incorporated into the primordia in the vicinity of P4, green for those located in the IPR and magenta for those contributing to both (Fig. 7a-c). At t0 (0 h), 1-2 layers of IPR cells were visible in front of P4 (Fig. 7a). As expected, when these cells divide, they mostly do it with a transverse division plane (Fig. 7a-c). Similar results were obtained with Col-0 SAMs (focusing on P3 that had a comparable folding at the boundary than P4 in Ler), although in this genotype the formation of the crease at the flower boundary hides the IPR cells more rapidly (Fig. 7g-i). Thus, their division pattern and localization in the IPR support the idea that these cells correspond to internode precursors.
Compared to Ler, 1-2 extra layers of IPR cells were observed in front of P4 at t0 (0 h) in the SAM of global della mutants. These cells divided several times in 34 hrs (Fig. 7d-f, compared to 7a-c), and in consequence, the mostly transverse divisions of IPR cells led to a higher population of cells organized in radial cell files (Fig. 7d-f, compared to 7a-c). This indicates that the higher GA signalling activity in global della mutant SAMs promotes internode specification. We conducted a similar analysis in pCUC2::gai-1-VENUS plants. Since expression of this transgene causes changes in SAM geometry (Extended Data Fig. 12m,n,p,q), we analysed IPR cells above the first primordium showing a comparable folding at the boundary as Col-0 P3. In these plants, opposite to global della mutants, much less cell divisions occurred in the IPR and there was no clear sign of an organization in radial cell files (Fig. 7j-l), thus showing that inhibition of GA signalling in the IPR perturbs the specification of the cellular organisation of internodes in the SAM. In lines with these results, we were able to detect the appearance of internodes in the global della mutant just below the SAM using electron microscopy, while flowers remained compacted in Ler (Fig. 7m-n). By contrast, organs were much more compacted in the SAM of pCUC2::gai-1-VENUS than in Col-0 (Fig. 7o-p), consistent with the taller and shorter inflorescence stem of the global della mutant39, 43 and pCUC2::gai-1-VENUS plants, respectively (Extended Data Fig. 14a-b). Our results thus support the hypothesis that higher GA signalling activity in the IPR specifies the cellular organization of internodes in the SAM, through a regulation of the orientation of cell division planes (Extended Data Fig. 15).
Discussion
Here, we developed a new ratiometric GA signalling biosensor, qRGAmPFYR, that provides information on GA function at the cellular level by allowing quantitative mapping of GA signalling activity that results from the combinatorial action of GA and GA receptors concentrations, with minimal interference with the endogenous signalling pathway. To this end, we have engineered a modified DELLA protein, RGAmPFYR, that has lost its capacity to bind DELLA interacting partners but that remains sensitive to GA-induced proteolysis. qRGAmPFYR responds to both exogenous and endogenous changes in GA levels and its dynamic sensing properties allow the assessment of spatio-temporal changes in GA signalling activity during developmental processes. qRGAmPFYR is also a highly flexible tool as it can be adapted to a variety of tissues simply by changing, if needed, the promoter used for its expression and is very likely transferable to other species given the conserved nature of the GA signalling pathway and of the PFYRE motif in angiosperms.
Internode development is a key trait for plant architecture and crop improvement. qRGAmPFYR revealed a higher GA signalling activity specifically in cells of the IPR, that are the precursors of internodes. By combining quantitative image analysis and genetics, we show that the GA signalling pattern imposes circumferential/transverse cell division planes in the SAM epidermis, shaping the cell division organization required for internode development. Few developmental regulators of the orientation of the cell division plane during development have been identified44, 45. Our work provides a striking example where GA signalling activity regulates this cellular parameter. DELLA can interact with the prefoldin complex36 and GA signalling could thus regulate the orientation of the cell division plane through a direct effect on cortical microtubule orientation35, 36, 46, 47. The fact that we show that unexpectedly not cell elongation nor cell division but only growth anisotropy correlates in the SAM with higher GA signalling activity is coherent with a direct effect of GAs on cell division orientation in the IPR. However, we cannot eliminate the possibility that this effect could also be indirect, e.g. mediated by GA-induced softening of the cell wall48. Changes in cell wall properties induces mechanical stress49, 50 that could also affect cell division plane orientation by acting on cortical microtubule orientation34, 41, 51. A combined effect of GA-induced mechanical stress and direct regulation by GA of microtubule orientation could then participate to create the specific pattern of cell division orientation in the IPR to specify the internode and further work is needed to test this idea. Likewise, previous works have highlighted the importance of the DELLA interacting proteins TCP14 and 15 in the control of internode patterning52, 53 and these factors could convey GA action, together with BREVIPEDICELLUS (BP) and PENNYWISE (PNY), which regulate internode development and have been shown to affect GA signalling2, 54. Early cytological studies showed that both the inner tissues and the peripheral zone of the SAM are required for internode development in Arabidopsis2, 37. The fact that GAs positively regulate cell division in the inner tissues12 supports a dual function of GAs in regulating meristem size and internode at the SAM. Patterns of oriented cell division are also highly regulated in the inner SAM tissues, and this regulation is essential to stem growth44. It will be interesting to explore whether GAs also play a role in orienting cell division planes in the inner tissues of the SAM and thus synchronize internode specification and development within the SAM.
Methods
Growth conditions and Plant material
Plants were grown on soil or in vitro on 1x Murashige-Skoog (MS) medium (Duchefa) supplemented with 1% sucrose and 1% agar (Sigma) under standard conditions (16h photoperiod at 22°C), except for hypocotyl and root growth experiments, for which the seedlings were grown on vertical plates under continuous light and 22°C. For experiments with nitrate, plants were grown on MS modified medium without nitrogen (bioWORLD plant media) supplemented with an adequate nitrate concentration (0 or 10 mM KNO3), 0.5 mM NH4-succinate, 1% sucrose and 1% type-A agar (Sigma) under long-day photoperiod.
We used the following Arabidopsis mutants and transgenic lines in the Ler background: gai- t6 rga-t2 rgl1-1 rgl2-1 quadruple della39, gai-t6 rga-t2 rgl1-1 rgl2-1 rgl3-4 global della55 or Col-0: pGID1a::GID1a-GUS, pGID1b::GID1b-GUS, pGID1c::GID1c-GUS, pGAI::GUS, pRGA::GUS, pRGL1::GUS, rgl2-5 (a promoter trap GUS line)56, pRGA::GFP-RGA8, pRGL3::RGL3-GFP, pCLV3::mCherry-NLS57, nlsGPS117.
All other transgenic lines were generated in the Col-0 background. To generate these lines and to perform transient assays in Nicotiana benthamiana leaves, plasmids were constructed in the following way. RGA cDNA (AGI code AT2G01570) and RGAm1, RGAm2, RGAm3 and RGAm4 mutant variants were obtained by PCR using specific primers numbered 1 to 8 in Supplementary Table 2, and inserted into pDONR221 (Thermo Fisher Scientific) by Gateway cloning and recombined with pB7FWG258 to generate p35S::RGA-GFP and p35S::RGAm1/m2/m3/m4-GFP. To generate pRPS5a::qRGA, pUBQ10:qRGA, pRPS5a::qRGAmPFYR, pUBQ10:qRGAmPFYR, pRPS5a::RGAm1/m3/m4-VENUS-2A-BFP, pUBQ10::RGAm1/m3/m4-VENUS-2A-BFP, and pCUC2::gai-1-VENUS, the promoter region of pRPS5a (1.7 kb fragment), pUBQ10 (2.5 kb fragment) or pCUC2 (3.2 kb fragment) inserted into pDONR P4-P1R (Thermo Fisher Scientific), RGA-VENUS, RGAmPFYR -VENUS, RGAm1/m3/m4 -VENUS or gai-1 cDNA inserted into pDONR221, and 2A-TagBFP or VENUS (for pCUC2::gai-1-VENUS) inserted into pDONR P2R-P3 (Thermo Fisher Scientific), were recombined into pB7m34GW58.
d17RGA (RGA deleted of the 17 amino acids DELLAVLGYKVRSSEMA composing the DELLA domain29) and d17RGAmPFYR mutant variant obtained by PCR using primers 1, 2, 5, and 6 (Supplementary Table 2) were inserted into pDONR221 and recombined into pB7m34GW with p35S (inserted into pDONR P4-P1R) and VENUS (inserted into pDONR P2R-P3) to generate p35S::d17RGA-VENUS and p35S::d17RGAmPFYR-VENUS. d17RGA-VENUS and d17RGAmPFYR-VENUS were then amplified by PCR using primers 1 and 12, inserted into pDONR221 and recombined as previously with pRPS5a or pUBQ10 and 2A- TagBFP to generate pRPS5a::qd17RGA, pUBQ10:qd17RGA, pRPS5a::qd17RGAmPFYR, pUBQ10:qd17RGAmPFYR.
To obtain pUBQ10::GID1a-mCherry, GID1a cDNA inserted into pDONR221 was recombined with pDONR P4-P1R-pUBQ10 and pDONR P2R-P3-mCherry into pB7m34GW. p35S:IDD2-RFP was obtained by recombining IDD2 cDNA inserted in pDONR221 into pB7RWG258. To get pGID1b::2xmTQ2-GID1b, a 3.9-kb fragment upstream of the coding region of GID1b and a 4.7-kb fragment including GID1b cDNA (1.3 kb) and terminator (3.4 kb) were first amplified using primers in Supplementary Table 2, then inserted into pDONR P4-P1R (Thermo Fisher Scientific) and pDONR P2R-P3 (Thermo Fisher Scientific), respectively, and finally recombined into pGreen 012559 destination vector with pDONR221 2xmTQ260 by Gateway cloning. To make pCUC2::LSSmOrange, the promoter sequence of CUC2 (3229 bp upstream the ATG), followed by the coding sequence for the large stokes shift mOrange (LSSmOrange)61 with a N7 nuclear localization signal, and the NOS transcription terminator, was assembled into the pGreen Kanamycin destination vector using 3 fragment Gateway recombination system (Invitrogen). For transactivation assays, pTCP- LUC and pBRE-LUC reporter constructs (containing synthetic promoters) and p35S::3xHA-VP16-TCP14 and p35S::3xHA-VP16-BZR1 encoding effector proteins were used as previously described62, 63. Plant binary vectors were inserted into Agrobacterium tumefaciens GV3101 strain and respectively introduced into Nicotiana benthamiana leaves by agro- infiltration and Arabidopsis Col-0 by floral dip. pUBQ10::qRGAmPFYR pUBQ10::GID1a- mCherry and pCLV3::mCherry-NLS qRGAmPFYR were isolated from F3 and F1 progeny of the appropriate crosses, respectively.
Pharmacological treatments
Chemical treatments with GA (GA3, Sigma or Duchefa) and paclobutrazol (PAC, Duchefa) were performed at the concentrations and time indicated in figures. For Y2H assays, 100 µM GA3 were added in selective media to promote interaction between RGA and GID1. For RGA and RGAm1/m2/m3/m4 degradation kinetics, N. benthamiana agro-infiltrated leaf discs were incubated with 100 µM GA3 and 100 mM cycloheximide (Sigma) over 300 min. For hypocotyl length measurements and fluorescence analyses of qRGAmPFYR hypocotyls and roots, seedlings were grown for 5 days on MS agar medium and then transferred for 4 days on MS agar plates supplemented with 5 µM PAC or 10 µM GA3. For GA3 treatment on qRGAmPFYR, GA4+7 treatment on nlsGPS1, and GA-Fl treatment on Ler, dissected shoot apices were inserted into Apex Culture Medium (ACM, 1/2x MS medium (Duchefa), 1% sucrose, 1% agarose, 2 mM MES (Sigma), 1x vitamin solution (myo-Inositol 100 mg/L, nicotinic acid 1 mg/L, pyridoxine hydrochloride 1 mg/L, thiamine hydrochloride 10 mg/L, glycine 2 mg/L), 200 nM N6-Benzyladenine) with indicated concentrations of GA/GA-Fl, and also immersed under 200 µl of GA/GA-Fl solution of indicated concentrations for a indicated period of time. For PAC treatment on qRGAmPFYR, 50 µM PAC were sprayed on the whole inflorescence every two days for a period of 5 days before observation.
Yeast two-hybrid assays
For Y2H assays, both the full-length and the C-terminal part of RGA and RGAm1/m2/m3/m4 (named M5 version, amino acids 199 to 587; the N-terminal part is subject to self-activation in yeast64) inserted into pDONR221 were recombined into pGBKT7 (Clontech) to obtain BD- RGA, BD-RGAm1/m2/m3/m4, BD-M5RGA and BD-M5RGAm1/m2/m3/m4. On the other hand, JAZ1, TCP14, IDD2, BZR1, GID1a, GID1b and GID1c cDNAs inserted into pDONR221 were fused to the activation domain GAL4 (AD) after recombination into pGADT7 (Clontech).
Direct interaction assays were carried out following the Clontech procedures. Briefly, BD- M5DELLA and AD-JAZ1, AD-TCP14, AD-IDD2, AD-BZR1 and, on the other hand, BD- DELLA (full length) and AD-GID1 constructs were co-transformed in the yeast strain AH109 and interactions tests were surveyed on selective medium lacking tryptophan, leucine and histidine. In some cases, the medium was supplemented with 3-amino-1, 2, 4 triazole (3AT, Sigma) or with GA to promote interaction between DELLA and GID1.
To confirm that the RGAm2 mutation impaired interaction with all RGA protein partners, we used the C-terminal part of RGA and RGAm2 (amino acids 199 to 587), to probe the Arabidopsis transcription factors REGIA + REGULATORS (RR) arrayed library, following the protocol described in Castrillo et al.65. TFs in the RR library were fused to GAL4 activation domain of the pDEST22 vector and independently transformed into the yeast strain YM4271 in 96-well plates. The RGAm2 protein fused to the GAL4 BD in the pDEST32 vector was transformed into the pJ694 yeast strain and used as bait. Replicates of the library were grown overnight on SD-Trp solid media and inoculated together with 100 μL of an overnight RGAm2 culture grown on SD-Leu on microtiter plates containing 100 μL of YPDA per well. Plates were incubated for 2 days at 30°C for mating, and diploid colonies were selected in new 96-well plates containing 200 μL of SD-Leu/Trp. 5 μL of the diploid cell cultures were lastly tested for protein interaction by placing them on solid SD medium lacking both Leu and Trp (positive growth control), and on SD medium lacking Leu, Trp and His, in the presence of 1 mM 3-aminotriazol (3-AT) (Sigma-Aldrich). Results were expressed in the form of a heat map for the strength of interaction according to the colony growth after five days of incubation at 30°C.
Co-immunoprecipitation assays
CoIP assays were performed on N. benthamiana agro-infiltrated leaves with p35S::IDD2- RFP, p35S::RGA-GFP or p35S::RGAm2-GFP. Three days after infiltration, total proteins were extracted with the native extraction buffer [Tris-HCl (pH 7.5) 50 mM, glycerol 10%, Nonidet P-40 0.1% supplemented with Complete Protease Inhibitors 1X (Roche)], and then incubated for 2h at 4°C with 50 µl of anti-GFP antibody conjugated with paramagnetic beads (Miltenyi Biotec). After incubation, samples were loaded onto a magnetic column system (µ columns ; Miltenyi Biotec) to recover the immunoprotein complexes according to manufacturer’s protocol. The immunoprecipitated (RGA-GFP and RGAm2-GFP) and co-immunoprecipitated (IDD2-RFP) proteins were detected by western-blot with anti-GFP (JL8; Clontech) and anti- RFP (6G6; Chromotek), respectively.
Immunodetection analyses
Plant materials were ground in 2x SDS-PAGE buffer followed by heating at 95°C for 5 min. After centrifugation at 13000 g for 5 min, total proteins were separated on 8.5% SDS-PAGE gel and transferred to an immobilon-P (PVDF) membrane (Millipore). Membranes are then saturated with blocking buffer (TBS 1x, Tween-20 0.1%; milk 5%) and incubated with a 2000-fold dilution of anti-GFP (JL8; Clontech), anti-RGA (Agrisera), anti-HA (Sigma) or anti-actin (Agrisera) antibodies, and a 5000-fold dilution of peroxidase-conjugated goat anti- rabbit or mouse IgG (Invitrogen). Signals were detected using the Luminata Forte Western HRP Substrate (Millipore).
Transactivation assays
Transactivation assays were performed using the Dual-Glo Luciferase Assay System (Promega) according to manufacturer’s instructions. Briefly, N. benthamiana leaves were agro-infiltrated with pTCP-LUC and pBRE-LUC reporter constructs, p35S::3xHA-VP16- TCP14 or p35S::3xHA-VP16-BZR1 encoding effector proteins, and p35S::RGA-GFP or p35S::RGAm2-GFP. Three days after infiltration, total proteins were extracted in lysis buffer (Promega), then Firefly and control Renilla LUC activities were quantified with FLUOstar Omega luminometer (BMG Labtech) using OMEGA2 software version 5.50 R4. For loading control, protein levels were analysed by immunodetection.
GUS staining
For GUS expression detection in inflorescence apex, 28-d-old plants grown under LD conditions were used. After fixation in 90% cold acetone at room temperature for 20 min, shoot apexes were transferred into GUS staining solution containing 1 mM potassium ferrocyanide, 1 mM potassium ferricyanide and 1 mg/ml 5-bromo-4-chloro-3-indolyl-ß-D- glucuronide (X-Gluc), vacuum-infiltrated on ice for 15 min, and incubated overnight at 37 °C in dark before washing with ethanol series and microscope detection. For the experiment with seedlings, 7-d-old plantlets grown in vitro under LD conditions were vacuum-infiltrated for 15 min in GUS staining solution containing 2 mM potassium ferrocyanide, 2 mM potassium ferricyanide and 0.25 mg/ml X-Gluc, and incubated at 37°C for 24 h. Then GUS solution was replaced with ethanol 70% and seedlings were observed using optical microscope.
In situ hybridizations
RNA in situ hybridizations were performed according to Vernoux et al.66. with some minor modifications on fixation. Briefly, shoot apices with ∼1cm stem were collected and immediately fixed in FAA solution (3.7% Formaldehyde, 5% Acetic Acid, 50% Ethanol) precooled at 4 °C. After vacuum treatments for 2x 15 min, the fixative was changed and the samples were incubated overnight. Antisense probes of the cDNA and 3’-UTR of GID1a, GID1b, GID1c, GAI, RGL1, RGL2 and RGL3 were synthesized as described in Rozier et al.67. using primers indicated in Supplementary Table 2. Immunodetection of the digoxigenin- labelled probes was performed using an anti-digoxigenin antibody, and sections were stained with 5-Bromo-4-chloro-3-indoryl Phosphate (BCIP)/Nitroblue Tetrazolium (NBT) solutions.
Microscopy
For confocal microscopy observations of hypocotyl, complete seedlings were put on slides and images were obtained with a Zeiss LSM 780 confocal microscope. SAM live imaging was performed as described32. Briefly, dissected SAMs were let to recover overnight after dissection. Confocal images were taken with a Zeiss LSM 710 or 700 confocal laser scanning microscope equipped with a water-dipping lens (W Plan-Apochromat 40x/1.0 DIC). The confocal settings were as previously described32 and detailed below. To observe GFP or GFP with propidium iodide (PI), a laser of 488 nm was used to excite, and the emission for GFP was 500-520 nm (in some cases it was 500-540 nm), for PI was 610-650 nm. For imaging of qRGAmPFYR or qRGAmPFYR with mCherry and PI, lasers of 514 nm, 405 nm, 561 nm and 488 nm were used for the excitation of VENUS, TagBFP, mCherry and PI, respectively, and the corresponding emission wavelength was 520-560 nm, 430-460 nm, 580-615 nm and 620-660 nm. mTURQUOISE2 (mTQ2) was excited by 445 nm laser and the emission range was 470-510 nm. For FRET detection of nlsGPS1, an acquisition mode of spectral imaging (λ-scan) with emission wavelength from 461 nm to 597 nm was used with excitation at 458 nm.
For optical microscopy, photographs of plants were taken with a LEICA MZ12 stereoscopic microscope equipped with a ZEISS AxioCam ICc5 camera head or Canon camera.
For scanning electronic microscopy of fresh shoot apex, a Hirox SH-3000 table-top microscope equipped with Coolstage (-20 °C to -30 °C) was used.
Image processing
Confocal stacks were processed in Fiji (fiji.sc) to get max projection or orthogonal views. Older primordia were also removed in Fiji. For 3D visualization of confocal stacks, we used the Zeiss ZEN2 software (Fig. 7) and a rendering using the VTK library68 (Fig. 5e, Fig. 6). In both cases, parameters were adjusted accordingly to show mainly the L1 cells.
Quantification and statistical analysis GA sensor quantification
For qRGAmPFYR quantification and visualization, images of hypocotyls were analyzed using a Python script that performs nuclei detection and signal quantification in 3D. The details of the algorithms used in the analysis are provided in Supplementary Method 1 (Section 2 – Nuclei detection and signal quantification).
For nlsGPS1 quantification and visualization, a series of Fiji macros were used. Briefly, a “seg-auto.ijm” macro was run for segmentation of nuclei based on the sum of signals from all the 14 channels of the spectral imaging. After removing little objects corresponding to signal noise using a “object-screening.ijm” macro, the ratio of DxAm (channel 8) to DxDm (channel 3) of each segmented nucleus was calculated by running a “3D-ratio.ijm”, and from here, the 3D construction of the SAM, with every nucleus showing its ratio value, was also achieved and printed. Statistical analysis was done in R.
SAM image sequence analysis
To analyze time-lapse confocal acquisitions of SAMs and obtain quantitative measures of GA signalling as well as cellular growth parameters, we developed a computational pipeline. It consists of 3D watershed segmentation for cell identification from the PI staining, nuclei detection and qRGA quantification, temporal registration using the PI image channel and surfacic growth estimation based on expert cell lineages. Individual SAM sequences were then aligned in order to perform population-scale statistics. Extensive details on the algorithms used in this pipeline are provided in Supplementary Method 1.
Hypocotyl length measurement
To measure hypocotyl length, seedlings were grown in vertical MS medium (Duchefa) with 0.8% (w/v) phytoagar and 1% sucrose for 5 days and transferred to MS, MS supplemented with 10 μM GA3 or 5 μM Paclobutrazol for 4 days in continuous light at 22°C. Plates were scanned and hypocotyl length was measured from the images using Fiji.
Cell division orientation quantification
New cell walls were identified by comparing images obtained at 0 h and 10 h, and masked with a manually drawn line in Fiji. Then, a macro was used to skeletonize and then to measure the angles of the drawn cell walls with an expert-defined centre of the SAM. Further angle frequency distribution was done in Excel using Pivot table. Statistical analyses of Kolmogorov-Smirnov tests were performed using an online tool: http://www.physics.csbsju.edu/stats/KS-test.n.plot_form.html.
Curvature quantification
The MorphoGraphX software was used to quantify cellular curvature of L1 cells of SAMs of different genetic backgrounds. Statistical analyses were done in R.
Meristem size measurement
To measure meristem size, a method described before was used69. Briefly, the SAM radius was determined by drawing a circle that covers I1 and I2 and that the center of which is roughly overlapping with the geometrical center of the SAM surface using Fiji software. Statistical analysis was done in R.
Synthesis and characterization of GA-Fluorescein (GA-Fl)
GA-Fluorescein (GA-Fl) were synthesized using a previously described protocol15. Extensive details on synthesis and characterization are provided in Supplementary Method 2.
Data and software availability
All experimental data and quantified data that support the findings of this study are available from the corresponding authors upon request. Quantitative image and geometry analysis algorithms are provided in Python libraries timagetk, cellcomplex, tissue_nukem_3d and sam_atlas (https://gitlab.inria.fr/mosaic/) made publicly available under the CECILL-C license. The script used to process hypocotyl images is publicly available in the qrga_nuclei_quantification project (https://gitlab.inria.fr/gcerutti/). The pipelines used to analyze SAM image sequences and to produce map visualizations are provided in a separate project (https://gitlab.inria.fr/mosaic/publications/sam_spaghetti) as Python scripts.
Supplementary Table 1. RGAm2 loses its ability to bind with RGA-interacting partners. Extended pairwise Y2H interaction assays between RGA, RGAmPFYR and RGA-interacting transcription factors. The colour code and numbers (0 to 5) indicate the strength of the interactions based on the size of the yeast colonies on selective media without leucine, tryptophan and adenine (-LWH) without or with 1 mM 3-aminotriazole (3AT). Empty pGBKT7 and pGADT7 vectors were included as negative controls.
Author contribution
P.A. and T.V. designed the study and supervised the work; B.S., A.F.-B., P.A. and T.V. designed the experiments. B.S., A.W. and A.M. performed the FRET analysis; A.N.-G. and S.P. conducted the large-scale two-hybrid analysis; S.L. and R.W. designed the synthesis strategy and synthesized the GA-Fl fluorescent probes. G.C. developed the pipeline for quantitative image analysis with the help of J.L.; B.S., A.F.-B., J.M. and G.C. performed the image analysis. B.S., A.F.-B., C.G.-A., L.J., G.B., E.V., L.S.-A. and J.-M.D. performed all the other experiments; all authors were involved in data analysis; B.S., A.F.-B., P.A. and T.V. wrote the manuscript with inputs from all authors.
Competing interests statement
The authors declare no competing interests.
Corresponding authors
Correspondence and request for materials should be addressed to T.V. (teva.vernoux{at}ens-lyon.fr) or P.A. (patrick.achard{at}ibmp-cnrs.unistra.fr).
Acknowledgments
We thank the colleagues of the Laboratoire Reproduction et Développement des Plantes for fruitful discussions on this work and for sharing material and advice. We also thank Miguel Perez-Amador, Jan Lohmann, Yvon Jaillais, Tai-ping Sun, Miguel Blazquez, David Alabadi and the NASC for seeds and plasmids. We acknowledge the contribution of SFR Biosciences (UMS3444/CNRS, US8/Inserm, ENS de Lyon, UCBL) facility PLATIM for assistance with microscopy and image analysis. This work was supported by the ANR-16-CE13-0014 (GrowthDynamics) grant to T.V. and P.A.; a European Research Council grant (GAtransport) to R.W.; a Guangdong Laboratory for Lingnan Modern Agriculture Grant NG2021001 to B.S..