Stable establishment of organ polarity occurs several plastochrons before primordium outgrowth in Arabidopsis

Plant leaves can adopt a wide range of shapes and sizes depending on the species and the environment in which they develop. In many species, leaves initiate at the flanks of the shoot apical meristem as small bulges. In their most basic form, they then develop as flat, ellipsoid structures, because growth mainly occurs in the plane of the leaf blade. In combination with a restricted expansion in the dorsoventral direction, this leads to the formation of a thin leaf blade with functionally distinct abaxial and adaxial faces, pointing away from and towards the stem, respectively, a property also referred to as leaf polarity.

Several genes have been associated with leaf shape and size, which are determined from the earliest stages of development onwards (e.g. Hervieux et al., 2016; Zhao et al., 2020; Zhu et al., 2020). Key are the regulators involved in establishing leaf polarity and which are activated before or during organ outgrowth. These include, for example: abaxial determinants, such as FILAMENTOUS FLOWER (FIL) (Sawa et al., 1999; Siegfried et al., 1999) or KANADI1 (KAN1) (Kerstetter et al., 2001); adaxial determinants, such as REVOLUTA (REV) or PHABULOSA (PHB) (McConnell et al., 2001); and the so-called margin genes, including PRESSED FLOWER (PRS) and WUSCHEL RELATED HOMEOBOX1 (WOX1) (Scanlon et al., 1996; Nakata et al., 2012), expressed in the future leaf margin and in the middle domain separating the abaxial and adaxial domains. The network also involves regulation via small RNAs (Kuhlemeier and Timmermans, 2016). Some of the polarity genes are expressed very early, and REV and KAN1 in particular might define a prepattern (Caggiano et al., 2017) even before the upregulation of organ founder cell markers, such as DORNROESCHEN-LIKE (DRNL) (Chandler et al., 2011) or ARABIDOPSIS PHOSPHOTRANSFER PROTEIN 6 (AHP6) (Besnard et al., 2014). If the regulatory network controlling polarity is impaired, leaf and leaf-like organs lose their bilateral symmetry and tend to become axisymmetric (Yamaguchi et al., 2012; Kuhlemeier and Timmermans, 2016).

How is the establishment of polarity coordinated? In a classical set of experiments, Sussex used local incisions in potato meristems to investigate this further (Sussex, 1951; reviewed by Kuhlemeier and Timmermans, 2016). When an incision was made between the vegetative meristem and incipient leaf primordia, leaves did not flatten, but developed into almost axisymmetric organs. This led to the hypothesis that leaf polarity depends on a signal coming from the meristem center. This result was further consolidated in tomato (Reinhardt et al., 2005; Qi et al., 2014; Shi et al., 2017). Reinhardt et al. (2005) showed that ablation in the epidermal L1 layer was sufficient to perturb leaf polarity, suggesting that the signal requires an intact surface layer. Interestingly, this layer is essential in the control of polarized auxin transport during organ formation (Reinhardt et al., 2003; Kierzkowski et al., 2013) and the hormone has also been considered as part of the signal proposed by Sussex. Indeed, certain mutants involved in auxin transport, such as PINOID or PIN-FORMED1, develop axisymmetric lateral organs (Qi et al., 2014) and auxin transport generates differences in hormone concentrations between the periphery and the center of the shoot apical meristem (de Reuille et al., 2006). Qi et al. (2014) proposed that auxin could act as a reverse ‘Sussex signal’, flowing from the adaxial side of the leaf primordium to the adjacent meristem to consolidate leaf dorsoventrality.

However, the concept of a single meristem-centered signal regulating leaf polarity has also been under debate. Although they were able to do so in Epilobium hirsutum, Snow and Snow could not confirm the ablation results in potato (Snow and Snow, 1954b). Traditional surgical experiments on meristems mostly rely on the morphology of young leaf primordia as the only marker to define the right position for incision. Therefore, Snow and Snow suggested that cuts leading to a loss in polarity inadvertently caused a reduction in size of the primordia in the radial direction, which then led to abnormal development. More recently, several, more elaborated hypotheses have been brought forward to explain the processes leading to dorsoventrality (Husbands et al., 2009; Caggiano et al., 2017), although all of them require some form of meristem-based signal or information. From a conceptual point of view, Koch and Meinhardt (1994) proposed a model that combined the existence of an organ-inhibiting gradient around the meristem center, with more local, self-organizing patterning processes within the initium itself as soon as the cells leave the inhibiting field. Although this has remained purely hypothetical, findings of Caggiano et al. (2017) in Arabidopsis suggest a more complex network of interactions involving auxin transport and pre-existing gradients of several transcription factors, such as KAN1 and REV. The same authors performed a number of laser ablations between the meristem and incipient primordia and showed that the wounds inflicted by cutting can locally induce the expression of certain polarity genes, such as KAN1. This result potentially questions the interpretation made by Sussex, because the loss of polarity observed in potato might be due to a wounding reaction involving the ectopic activation of abaxial determinants. However, in Arabidopsis, the organs were only followed in small vegetative meristems over short time frames after ablation and no organs with altered polarity were observed. Therefore, we revisited the ablation experiments in Arabidopsis thaliana, using very early primordium and polarity markers. We were unable to induce the formation of radialized leaves and sepals by performing ablations along or perpendicular to the radii of vegetative and floral meristems. This was even the case when the ablations were made two or three plastochrons (i.e. the time between successive leaf initiation) before organ outgrowth. Therefore, our results support a model in which organ polarity does not require any meristem-based signal from the earliest stages of primordium specification onwards.

To identify the cell populations that would be ablated, we used a set of fluorescent markers for early organ development. For this purpose, we chose pDRNL:erCERULEAN as an organ founder cell marker (Chandler et al., 2011). In addition, we used pFIL:erGFP. This promoter is active in the abaxial domain of young outgrowing primordia and shows an extended expression in abaxialized mutants (Nakata et al., 2012; Tameshige et al., 2013). We also used pPRS:SV40-3×GFP to label the middle domain between both sides of the primordia (Scanlon et al., 1996; Nakata et al., 2012; Xu et al., 2015). Given that the precise relative expression dynamics of these markers has not been described, we constructed a set of lines co-expressing pDRNL with either pFIL or pPRS. To monitor cell growth and lineage, the membrane marker pUBQ10:Lti6b-tdTomato was included. In some experiments, propidium iodide (PI) was used for this purpose.

In the vegetative meristem, we followed organ initiation from I₃ (I₁ being the first incipient primordium that is ready to bulge out, and I_1+n the subsequent younger incipient primordia) to P₆ (P₁ designates the first visible outgrowing primordium and P_1+n the subsequent older outgrowing primordia). In complement to previous results (Chandler et al., 2011), pDRNL is active throughout the meristematic dome, with a patchy pattern in the L2 and L3 layers (Fig. S1A,B). However, at the I₃ position, pDRNL activity significantly increases and it also becomes active in the L1 layer (Fig. 1A; Fig. S1A, Fig. S2). Lineage analysis showed that, from I₃ to I₁, this activity covers both the future organ boundary and adaxial domain of the future primordium (Fig. 1; Fig. S2) (i.e. a small group of cells, approximately five or six cells in diameter). Slightly later than pDRNL-, pFIL- and pPRS-driven signals become visible at I₁. They largely overlap with the pDRNL maximum, and no apparent polarization of these markers was distinguishable before P₁, when the primordia start to bulge out. At the P₁ stage, the activity of pDRNL and pPRS is more concentrated at the adaxial domain, whereas pFIL expression shifts to the abaxial domain and also partially overlaps with the middle domains. Slightly later on, at P₃, pPRS activity becomes restricted to the middle domain only. The polarized expression of the markers is fully resolved when the primordium is clearly growing out and is approximately three cells high (P₄ in our experimental set-up). Note that pPRS signals always extended more towards the lateral organ boundaries compared with the other two markers (Fig. 1; Figs S1, S3).

Similar expression patterns of pDRNL and pFIL were found in the inflorescence meristem: pDRNL expression is visible from I₃ onwards, whereas pFIL initiates at I₁ (Fig. 2A; Heisler et al., 2005; Refahi et al., 2021). During early stage 2 of flower formation, pDRNL and pFIL expression is restricted to the area where the cryptic bract bulges out slightly, but does not define a clear polarity. Both genes are switched off in the bract during late stage 2 of floral development. Until then, pPRS is only activated in a small number of individual nuclei, scattered randomly over the floral bud as well as over the entire inflorescence meristem (Fig. 2; Fig. S3).

Soon after the inactivation of pDRNL and pFIL in the cryptic bract, the sepal primordia are specified. This is marked by a local activation of pDRNL, first in the outer (abaxial) primordium and two lateral primordia (Zhu et al., 2020). At late stage 2, pDRNL is activated in all four sepal primordia. Similar to the vegetative meristem, this activation also covers the future boundary with the floral meristem. At floral stages 3 and 4, pDRNL activity becomes restricted to the adaxial side of the sepal primordia and the promoter is also activated in a ring-like domain in one of the future inner whorls, probably corresponding to part of the third whorl (Fig. 2A-C). The activation of pFIL and pPRS is slightly later than that of pDRNL but still during late floral stage 2. In the leaf, pFIL is initially activated at the sepal primordia in a pattern that overlaps with that of pDRNL. At late floral stage 2, only the outer sepal is marked, followed by the inner and lateral sepals (see also Zhu et al., 2020) during stage 3. Similar to the leaf primordia (Fig. 1C-E,H), pFIL expression shifts peripherally during stage 3 and becomes limited to the abaxial and middle domains of the sepals at stage 4 (Fig. 2C-E,H; Fig. S3). pPRS is first activated in the outer sepal at late stage 2, partially overlapping with the pDRNL expression domain. Subsequently, pPRS is mainly expressed in the adaxial side of the sepal primordia. Gradually, pPRS becomes active in a ring covering the middle domain of all sepals as well as the lateral boundaries between them during stage 4 (Fig. 2B,F-H; Fig. S3).

In summary, although the staging of primordia can vary a little depending on the size of meristems, the relative expression dynamics of the three different markers allows us to identify the precise moments when the organ founder cells and the different domains that characterize organ polarity are established. In this context, the upregulation of pDRNL marks the future organ boundary and adaxial domain, whereas, in later stages, pFIL marks the abaxial domain and the middle domain, the latter being characterized in addition by pPRS expression. We next used these patterns to characterize the precise locations of the ablations. We distinguished two categories: ablations that did not or only marginally touched the expression domains; and ablations within specific domains of the initia or primordia. These will be discussed separately.

We first sought to repeat experiments in which young primordia were physically separated from the central meristem through incision, which, in potato and tomato, resulted in the formation of abaxialized, axisymmetric organs. To this end, we performed circular ablations surrounding the meristem center (i.e. perpendicular to the meristem radius) using pDRNL as an early leaf initium marker (Fig. 3A). To exclude short-distance signal transmission between the meristem and the young leaves, the incisions were set as close as possible to the incipient primordia. In 12 meristems, 22 I₃-I₁ primordia were left intact or only a few proximal cells were eliminated. In addition, we were able to follow 22 intact P₁-P₂ primordia in these meristems. Under these conditions, none of the leaves that grew out showed radial symmetry (Fig. 3; Figs S4, S5, Fig. S6A,B). Polarized leaves grew out even following ablations performed at I₃, when pDRNL upregulation was only starting and pFIL was not expressed at all (Fig. 1A; Fig. S1C). However, we could not determine whether there was any transient change in polarity. Nevertheless, the patterning process did not require a meristem-based signal.

In tomato, not only incisions isolating the primordium from the central meristem cause a loss of dorsoventrality. In addition, incisions parallel to the meristem radius (i.e. at the lateral boundaries between organs) are sufficient to induce abaxialized organs (Shi et al., 2017).

In two I₂ and five I₁ initia, the ablations were four-to-six cells apart and either left the initium (five or six cells wide) intact or only slightly reduced its width. In all seven cases, the resulting primordia showed a polarized distribution of pDRNL and pFIL signals 2 or 3 days after incision (Fig. 4A,B). Finally, a characteristic leaf with dorsoventral identity formed 5 days after incision (Fig. 4C,D). Once the primordia had already started to bulge out, lateral ablations of three-to-eight cells wide did not affect growth and leaf polarity (13/13 of P₁ and P₂ primordia examined) (Fig. S7).

Within the limits of the markers that we used, we did not find any indication that lateral ablations at the boundaries and ablations between the vegetative meristem and initiating organs perturbed polarity from I₃ onwards (i.e. before pFIL and pPRS are activated).

We next assessed the effect of laser ablation on sepal shape and polarity at the floral meristem. As performed for the vegetative meristem, circular ablations were used to isolate sepal initia from the central meristem. In 7/19 stage-2 floral buds, this led to an increase in sepal number, the remaining 12 buds had four sepals, as in wild type (WT) (Fig. 5A-D). None of the sepals was radially symmetric. This was even the case when early stage-2 flowers were ablated (i.e. before pFIL expression was activated) (8/8 meristems, Fig. 5A,B,D).

We then confined sepal initia by two parallel ablations at the lateral boundaries in stage-2 floral buds. All 14 sepal initia established normal polarity and no radial sepals were formed (Fig. 5E,F).

In summary, in the floral meristems, neither parallel nor perpendicular ablations close to the initia perturbed dorsoventrality.

As mentioned earlier, Snow and Snow (1954a) proposed that the incisions made by Sussex could have led to a reduction in size of the initia, thus causing the formation of radialized leaves. We further sought to confirm this with a set of ablations in which either the adaxial or abaxial domain of the initia/primordia was eliminated or initium/primordium width was reduced to only one to four cells.

We also performed more restricted ablations, limited to an arc of the circumference, targeting single initia. These were sufficiently large to ablate up to the entire adaxial half of the pFIL expression domain. After ablation at I₁ or P₁, polarized leaves were formed (n=5; Fig. 6A-A″; Fig. S6C). However, this was correlated with an initial inactivation of pFIL, followed by a more distal reactivation of pFIL toward the meristem periphery and adaxial cell regeneration from the abaxial side of the wound (Fig. 6A-A″). Notably, this also involved a redefinition of the organ boundary with the meristem (Fig. 6A′,A″).

Twelve of the circular ablations described above went through I₃-I₁ initia, eliminating the abaxial part of the pDRNL-labeled domain (see Fig. S5 I₃ for an example of precise positions). In all cases, growth arrest was observed. Some outgrowths occurred next to the wound, but these did not develop further and did not express pDRNL (Fig. 3A,B; Fig. S4). We also performed six more restricted ablations, limited to an arc of the circumference, targeting single initia. This resulted in an elimination of 60-80% of the entire pFIL domain, leaving only a few cells adaxially. In these cases, the outgrowth of leaf primordia was also halted, showing limited cell division of the remaining adaxial cells and loss of pFIL expression (Fig. 6B′,B″; Fig. S6C).

As we have discussed, lateral ablations leaving most of the leaf primordia or initia intact did not affect development. We next analyzed a set of lateral ablations at an early stage of primordia development, eliminating a significant part of the founder cells. In four cases (at two I₁ and two I₂), the incisions were very close (i.e. one or two cells apart). These initia did not further develop and the signal of pDRNL or pFIL was lost (Fig. 7A-B′). Thirteen ablations at I₁/I₂ were three or four cells apart. Seven of the resulting primordia then developed normally (Fig. S6C). However, in six cases, growth was dramatically reduced, leading to the formation of small bulges. Polarity in these primordia was maintained for at least 6 days. Later on, the abaxial pFIL marker extended adaxially (Fig. 7C-E″) in these small, non-growing primordia. This apparent loss of polarity could be due to indirect effects, possibly related to their limited growth.

We next performed ablations on the floral meristem, restricted to an arc of the circumference, eliminating part of the sepal before it started to grow out. As a control, restricted ablations next to the initium did not perturb development, as was also observed for circular ablations (n=2, Fig. 8A-A′; compare Fig. 5A-D). This was also the case when the adaxial part of the pFIL domain itself was killed (n=8, Fig. 8B-B′).

When we performed lateral ablations eliminating part of the initia themselves and reducing their width, we could not observe any effect on polarity. Notably, even leaving one radial row of initium cells did not perturb sepal polarity, although the sepals were narrower (n=4, Fig. 8C-C′).

We finally eliminated all founder cells of the outer sepal primordia. In 3/3 successful ablations, no sepal regenerated along the dorsal-ventral axis. However, to our surprise, flat sepals were re-initiated from the lateral boundaries, which was characterized by the ectopic activation of pDRNL in these domains (Fig. S8A). To test whether the reactivation of pDRNL is induced by signals from the neighboring sepal primordia, we eliminated the founder cells of the outer and lateral sepal primordia in five floral buds. In all cases, pDRNL was activated in the lateral boundaries and new flat sepals developed (Fig. S8B). As shown above, the margin promoter pPRS is active in these lateral boundaries (Fig. 2). PRS has a role in cell proliferation and is important for the lateral extension of sepals (e.g. Zhao et al., 2020). Therefore, we investigated whether sepals would also regenerate from boundaries in mutants with a reduced margin activity. This was not the case, and no regeneration was observed from the boundary when the outer and lateral sepal primordia in wox1 prs double mutants were ablated (n=5, Fig. S8C).

In this study, we combined in vivo imaging with laser ablation to study the establishment of organ polarity in both vegetative and floral meristems (for a summary of the different types of ablations, see Fig. 9A-C). In particular, the use of larger vegetative meristems grown in short days allowed us to monitor leaf development over long periods. This is an important advantage over previous work, in which more precise lineage tracking was not possible.

We found that organ polarity does not require any meristem-based signal at least two or three plastochrons before the organs grow out, at a point when pDRNL is upregulated and organ founder cells are selected. Likewise, sepal polarity becomes autonomous at stage 2 of flower development (i.e. when sepals are initiated, but have not yet started to grow) (Fig. 9A).

Husbands et al. (2009) proposed that leaf polarity is determined in three phases: establishment, resolution and maintenance. The main players would be activated during the first phase, but not necessarily in the correct pattern, which would be defined during the second phase and maintained thereafter. Although the first two phases would, in principle, require external information, local molecular regulation within and around the leaf primordium itself should be sufficient for maintenance once it is fully deployed. Although results in tomato and potato suggested that the dorsoventral patterning was not very robust during the initiation process itself (Sussex, 1951; Reinhardt et al., 2005), we show here that, even at a stage when major regulators, such as FIL or PRS, are not yet activated, polarity is already robustly established and maintained. Indeed, although the ablations cause major changes in local signaling and even can affect gene expression (Heisler et al., 2010; Caggiano et al., 2017; Zhao et al., 2019), we did not see any effect on leaf polarity, suggesting that an external signaling gradient does not have a role once pDRNL is activated. Further work is now required using additional markers to show that there are no more-subtle effects on ad/abaxial polarity.

To understand this early establishment of polarity, the results reported by Caggiano et al. (2017) are particularly relevant. They showed the existence of concentric expression zones around the meristem center of genes involved in organ polarity, a domain weakly expressing the adaxial determinant REV in the meristem center, surrounded by cells expressing the abaxial gene KAN1 at the meristem periphery. Both transcription factors mutually inhibit each other and the boundary between the two domains covers cells that have not yet been incorporated in new primordia (i.e. this boundary precedes organ formation). Therefore, both genes appear to define some type of ‘prepattern’ for polarity. At sites where auxin accumulates, REV is further upregulated, at least from I₂ onwards (Caggiano et al., 2017). Our data suggest that this upregulation is also accompanied by increased DRNL expression, and there is evidence that both transcription factors directly interact (Chandler et al., 2007; Zhang et al., 2018). Therefore, we deduce that, initially, both REV and DRNL cover the future organ boundary and the adaxial domain. At a later stage, when tissue folding starts at the boundary, auxin levels decrease. Combining our results with those of Caggiano et al. (2017), we propose that this is followed by the inactivation of both REV and DRNL and the activation of KAN1 in the forming organ boundary.

Caggiano et al. (2017) also suggested that the loss of polarity observed by Sussex (1951) after incision between the meristem and the young primordia could be the result of perturbation of this prepattern as a wound reaction. As we have shown here, ablations substantially modify the meristem structure, causing local outgrowth. In addition, it has been shown that ablation can alter auxin fluxes, microtubule organization and gene expression patterns (Heisler et al., 2010; Caggiano et al., 2017; Zhao et al., 2019). In particular, the polarity gene KAN1 is induced by wounding. Caggiano et al. (2017) suggested that this ectopic expression might create a zone of abaxial identity and organs formed in this zone are then abaxialized. This might explain the loss of polarity observed in tomato and potato. We did not obtain any clear evidence for this scenario, because we never obtained abaxialized leaves or sepals after ablation. However, we did observe changes in local patterning. When a large part of the future adaxial domain was eliminated, pFIL activity was initially lost, a new boundary was respecified, and a polarized primordium was formed a short distance from the wound after ablation (Fig. 6, Fig. 9B). This could indicate that the ablation leads to some type of dedifferentiation followed by local repatterning. No meristem-based signal would be required for this process, because the initium was cut off by the ablation. It is possible that ectopically induced KAN1 expression would be sufficient to guide such a reprogramming.

Nevertheless, the results from Arabidopsis are very different from those reported for Solanaceae. A possible explanation for this discrepancy is that the expression dynamics and roles of polarity determinants differ between species. For example, this is illustrated by small RNAs targeting auxin response factors (ARFs), which are implicated instead in the maintenance phase in the developing leaves of Arabidopsis, whereas they are indispensable for the establishment of polarity in several monocot species [see Kuhlemeier and Timmermans (2016) for discussion].

A final issue concerns the lateral organ boundaries (Fig. 9C). Working on tomato, Shi et al. (2017) found that lateral ablations at the level of organ boundaries are already sufficient to perturb polarity, suggesting a role of these boundaries in organ formation. Although we found changes in pFIL expression after some of the lateral ablations, these changes were always correlated with the compromised outgrowth of leaves. The FIL expression domain was reduced in width and even a loss of pFIL signals was observed in cases where very few cells were left (Fig. 7A,A′). In these cases, we could not unambiguously uncouple growth arrest and the apparent loss of polarity, because both processes occurred simultaneously. When the ablations were slightly further apart, some growth occurred and pFIL could also extend adaxially. However, this happened long after growth arrest and could be largely indirect. The link between growth and FIL has remained unclear. Eshed et al. (2004) and Goldshmidt et al. (2008) suggest that FIL could be growth promoting, whereas Refahi et al. (2021) did not find a clear correlation between growth rates and FIL expression in the Arabidopsis flower.

Interestingly, ablations of the entire sepal in the Arabidopsis flower affected boundary zones outside the domain of pDRNL expression, but where the margin promoter pPRS was still expressed (Fig. 2B). We have previously shown that margin genes are involved in extending the outgrowing sepal primordium as an important step in maintaining organ flatness (Zhao et al., 2020). We show here that the lateral boundary cells initiate new sepals precisely when the entire organ is ablated. This regeneration process requires PRS, suggesting that, under normal conditions, the gene maintains the competence of the lateral boundary cells to respecify into a polarized sepal.

In conclusion, we find that, in Arabidopsis, the local information within and around the primordium is able to withstand major perturbations in local shape and signaling, long before polarity is fully resolved. In this study, we used three different markers involved in organ initiation and polarity. Extension of this approach to other polarity markers should provide further insight into how the molecular regulatory network functions and reacts to perturbation.

The mutant [wox1 prs (Nakata et al., 2012)] and marker lines [pFIL:erGFP (Watanabe and Okada, 2003), pDRNL:erGFP (Cole et al., 2009), pDRNL:erCERULEAN (Cole et al., 2013), pPRS:SV40-3×GFP (Xu et al., 2015) and pUBQ10:Lti6b-tdTomato (Segonzac et al., 2012)] used in this study have been previously described. The double or triple fluorescent labeling marker lines were obtained by crossing and subsequently confirmed by antibiotic screening and checking of the fluorescence signals. To obtain vegetative shoot apical meristems, plants were grown under short-day conditions on soil (8 h light/16 h dark period) for 33-37 days until further dissections. To obtain inflorescence meristems, plants were first cultured in short-day conditions for 3 weeks and then transferred to long-day conditions (16 h light/8 h dark period) for an additional 2 weeks. Under these conditions, the shoot apical meristems were big enough to facilitate both dissection and laser ablations. Plants grown on soil were cultured with 60% humidity, using light-emitting diodes (150 mEm⁻²s⁻¹) and at a daytime temperature of 20-22°C. For live imaging and long-term culture, the meristems were dissected and cultured in vitro on apex culture medium (ACM) (Hamant et al., 2019) supplemented with 200 nM N6-benzyladenine and 0.1% plant preservative mixture (PPM; Plant Cell Tech).

To prepare the vegetative meristems, 33- to 37-day-old rosettes were carefully removed from the soil to preserve the root as much as possible. The rosettes were then washed with water and dried with tissue paper. Whole plants were moved to dissection plates (3% agarose in Petri dishes) and immobilized as previously described (Shi et al., 2020). To expose the vegetative meristem, the old leaves were cut off as cleanly as possible using injection needles (0.45 mm, BD). The young leaf primordia covering the meristem were dissected away carefully using tweezers under a binocular microscope (Leica M125) at high magnification. To prepare inflorescence meristems, whole-shoot apices were inserted in the ACM and dissected with tweezers. After dissection, the inflorescence shoot apices were transferred to a new ACM plate and cultured in vitro until further experiments. We usually left the dissected vegetative and inflorescence meristems to recover for ∼4 h before imaging and ablations.

The laser ablation experiments were performed as described by Zhao et al. (2019), using a Zeiss LSM 700 laser-scanning confocal microscope, equipped with an Andor MicroPoint laser ablation system driven by iQ3 software (Oxford Instruments), which delivered a 6-Hz paused laser at 356 nm. Pre-staining of the meristems with PI (Sigma-Aldrich, 100 mM) for 5 min was carried out to increase the efficiency of laser ablation. To determine the precise site of ablation, we first made an acquisition using Zeiss ZEN software (black edition). We then adjusted the focal plane and chose the circular or freeline tool in iQ3 to draw the trajectory for the ablation according to the position shown in the confocal image. We set the laser power at level eight or nine, with five repetitions for each point on the trajectory in order to make a deep wound. After the ablation, we made another acquisition using the confocal microscope. Under such conditions, our ablations were usually approximately two or three cells deep. In total, 60 ablations in vegetative meristems were carried out: 38 were two or three cells deep, 19 were three or four cells deep, and three were one or two cells deep.

For live imaging, the dissected meristems were immersed in distilled water supplemented with 0.1% PPM and examined in a Zeiss LSM 700 laser-scanning confocal microscope equipped with water immersion objectives (W Plan-Apochromat 40×/1.0 differential interference contrast). For three-dimensional (3D) image processing, we used the Zeiss ZEN software (black edition), choosing the transparency rendering mode in order to visualize the signals on the meristem/organ primordia surface. Longitudinal optical sections of the stacks were carried out using Fiji freeware.

The authors thank Olivier Hamant and Marcus Heisler for kind suggestions on this paper. We also thank Miyuki Nakata and Kiyotaka Okada for providing pFIL: erGFP and wox1 prs seeds, Wolfgang Werr and John Chandler for providing pDRNL:erGFP and pDRNL:erCERULEAN seeds, and Yuling Jiao and Chuming Liu for providing pPRS:SV40-3×GFP seeds. We thank Bihai Shi and Sana Dieudonné for instructions on vegetative meristem dissection, Claire Lionnet and PLATIM for supporting the laser ablation experiments.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.198820