Stem integrity in Arabidopsis thaliana requires a load-bearing epidermis

Growth coordination among cells and tissues within a plant organ is considered crucial for the shape and integrity of plant structures. Biochemical cues, such as hormones and microRNA gradients, contribute to such supracellular coordination (Finet and Jaillais, 2012; Skopelitis et al., 2017). Because cells within plant tissues are glued together via their walls, collective growth generates significant mechanical stress (Sassi and Traas, 2015). Such stress can act as an additional cue to coordinate growth. For example, supracellular alignment of cortical microtubules guides cellulose deposition and reinforces epidermal cell walls, enabling them to withstand tensile stress in plant stems (Verger et al., 2018). Two major questions arise: What is the nature of the interplay between biochemical and biomechanical signaling? What is the relative contribution of inner and outer tissues in coordinating growth (both mechanically and biochemically)?

In our previous study, we reported that Arabidopsis thaliana clv3-8 det3-1 double mutants, with molecular lesions in both CLAVATA3 (CLV3) and DE-ETIOLATED 3 (DET3) genes, exhibit deep longitudinal cracks within their inflorescence stems (Maeda et al., 2014). The CLV3 peptide signal is perceived directly by CLV1 and CLV2, suppressing WUSCHEL expression (Miwa et al., 2008, 2009; Müller et al., 2008; Betsuyaku et al., 2011; Kinoshita et al., 2010). In the absence of CLV3, the central zone of the shoot apical meristem is enlarged owing to a delay in transition to the peripheral zone in which new organs are generated (Laufs et al., 1998). Together with a long-distance impact on cell proliferation (Reddy and Meyerowitz, 2005), possibly through a scaling mechanism involving a geometrical feedback (Gruel et al., 2016), this enlarged central zone promotes the development of thicker stems. DET3 encodes subunit C of vacuolar H⁺-ATPase and is ubiquitously expressed in both roots and shoots throughout the plant life cycle. Vacuolar ATPases are often abundantly expressed in cells involved in active vesicular trafficking and secretion, and several vacuolar ATPase mutants exhibit growth defects (Padmanaban et al., 2004). These vacuolar ATPase mutants may show reductions in cell-wall stiffness and in ATPase activity. In this respect, DET3 is an essential gene, as a complete loss of function results in lethality (Schumacher et al., 1999). A knockdown allele, det3-1, has been generated. It bears a point mutation (T→A) 32 bp upstream of the putative 3′ splice site junction, and the mutant still holds ∼50% of mRNA, protein and V-ATPase activity (Schumacher et al., 1999; Fukao et al., 2011). DET3 is particularly important for lignin and carbohydrate biosynthesis, as well as for proton transmembrane transport (Schumacher et al., 1999; Caño-Delgado et al., 2000; Newman et al., 2004; Rogers et al., 2005). The det3-1 mutant exhibits reduced levels of cellulose synthesis due to impaired secretion and recycling of the cellulose synthase complexes to the plasma membrane (Caño-Delgado et al., 2003; Luo et al., 2015). We hypothesized that the cracked stem phenotype in cvl3-8 det3-1 results from excessive growth driven by the clv3-8 mutation and weaker mechanical properties resulting from the det3-1 mutation. Here, we tested this hypothesis through growth analysis, mechanical assessments and genetic complementation.

From a biophysical perspective, the inflorescence stem is often considered a ‘pressurized vessel’ (Kutschera and Niklas, 2007; Baskin and Jensen, 2013; Hamant and Haswell, 2017). Stem structure is classically viewed as a two-component system, with the outermost component, the epidermis, under tension and the remaining inner components under compression (Hejnowicz and Sievers, 1996; Niklas and Paolillo, 1997). The balance between tension and compression implies that the epidermis is a load-bearing layer, whereas the inner tissue is held in place by the epidermis (Peters and Tomos, 2000; Galletti et al., 2016). For a cylindrical structure such as a stem, tensile stress is predicted to be twice as high in the transverse direction as in the longitudinal direction. This is consistent with the longitudinal gaping wounds that appear after incisions (Kutschera and Niklas, 2007) and in cell–cell adhesion mutants (Verger et al., 2018). More generally, this is also consistent with the observation that dwarf mutant phenotypes can be complemented by expressing the wild-type (WT) gene in the epidermis (Savaldi-Goldstein et al., 2007; Vaseva et al., 2018). Therefore, spontaneous longitudinal cracking of clv3-8 det3-1 mutant stems may also be due to high transverse tensile stress in the epidermis. However, this remains debated. In roots, the endodermis, rather than the epidermis, is generally considered the load-bearing layer (Vermeer et al., 2014). Furthermore, tensile stress direction throughout inner tissues plays an important role in generating flat leaves, whereas the cellular response to stress in the epidermis would have to be inhibited for flat leaves to form (Zhao et al., 2020). Here, we have studied the clv3 det3 double mutant because it provides a unique opportunity to test the epidermal growth theory in a context where wounds are not induced mechanically or genetically but emerge solely as a result of growth.

In previous work, we found a correlation between the extent of cell proliferation and stem cracking frequencies in the det3-1 mutant background, with clv1 and clv2 exhibiting the weakest phenotypes and clv3 the strongest. Indeed, the frequency of cracks in clv1-4 det3-1 and clv2-1 det3-1 stems were 37% and 35%, respectively, significantly less than that in clv3-8 det3-1 stems, in which the cracking frequency was almost 100% (Maeda et al., 2014). To gain further insight into this correlation, we prepared histological cross-sections of inflorescence stems, quantified their cellular parameters and investigated whether there was a correlation between the frequency of stem cracks and inner tissue phenotypes (Fig. 1). Tissue cross-sections showed that the pith cells in all the mutants in the det3-1 background were distorted, as we reported previously for det3-1 and clv3-8 det3-1 (Maeda et al., 2014), although the pith cell morphological defects were relatively mild in clv2-1 det3-1 (Fig. 1A). Stem section perimeter measurements correlated positively with the number of vascular bundles (R²=0.57; Fig. 1B). Similarly, cross-sectional areas correlated positively with epidermal cell numbers (R²=0.88; Fig. 1C). These results indicate that clv mutations independently enhance stem thickening and increase the number of epidermal cells and vascular bundles, even in the det3-1 mutant background. However, there was little correlation between pith cell numbers and cross-sectional areas (R²=0.08; Fig. 1D), reflecting an excessive increase in pith cell numbers relative to stem thickening. Importantly, clv3-8 det3-1 stems contained approximately twice as many pith cells as WT stems, despite their similar stem cross-sectional areas (Fig. 1D). Taken together, these results indicate that an excessive increase in the number of stem inner-tissue cells occurs in all clv det3-1 mutants. However, the low correlation between pith cell numbers and stem thickness suggests that pith cell features may also be a secondary effect of the mutations. In other words, altered cell differentiation may be triggered by both the det3-1 mutation and in response to cracks in the stem.

We used kinetic analyses to disentangle the different contributions of the clv3 and det3 mutations to stem cracks, growth and pith cell differentiation. First, we focused on the temporal relationship between stem cracks and growth. Compared with WT plants, stem axial growth in the clv3-8 det3-1 mutant was slow, and this became obvious three days after bolting (Fig. 2A,B). Interestingly, the det3-1 mutant exhibited the same stem growth defects as clv3-8 det3-1, whereas clv3-8 exhibited an almost normal growth pattern (Fig. 2A,B). Therefore, the det3 mutation is epistatic to clv3.

Next, we monitored the timing of crack occurrence on a daily basis (Fig. 2C). Our time course analyses revealed that cracks in the stem occurred most frequently when clv3-8 det3-1 stems were ∼2 cm long, and all plants examined had cracks before their stems reached 3 cm in length (Fig. 2C). We also noticed that stem elongation varied among individual plants: stem elongation was either slow or totally arrested after bolting. Nevertheless, even in relatively short stems, the first cracks appeared at later stages. Together, these results show that slow inflorescence stem growth precedes nascent cracks in the clv3-8 det3-1 mutant. This suggests that cracks are a secondary effect of growth, which is consistent with the epidermis resisting and limiting growth before it yields and cracks.

Next, we analyzed the temporal relationship between stem cracks and pith cell differentiation. To do so, we analyzed pith cell differentiation relative to the size and age of cracks in clv3-8 det3-1 stems. Our previous work showed that the stem cracks extend longitudinally. Therefore, cracks that appear early can be distinguished from those that appear later (Maeda et al., 2014). Longitudinal sections of the clv3-8 det3-1 stem showed that pith cells exhibited random shapes and growth directions (Fig. 3A). Whereas pith cells in WT plants were typically piled up in columns, those in clv3-8 det3-1 were transversely arranged in distorted shapes (Fig. 3A). Although deformed pith cells appeared to be spread across the entire stem, a series of cross-sections revealed that pith cells were not distorted before stem cracking (Fig. 3B). In contrast, pith cells were severely distorted at crack sites, further suggesting that altered differentiation is also a consequence of stem cracking (Fig. 3B).

Next, we prepared cross-sections on the day that a crack occurred from two different parts of the same inflorescence stem: at the cracked region (basal part) and uncracked region (upper part). Pith cells from the upper part of the stem generally showed mild deformation, whereas those from the crack sites were severely distorted (Fig. S1). Note that cracks consistently occurred at interfascicular regions and never through vascular bundles, probably because the former are mechanically weak (Fig. 3).

To confirm these results, we compared stem morphology at four key growth stages, which were selected based on inflorescence stem length. Stem portions were dissected at 5 mm from the base when the main inflorescence stems reached 1, 2, 3 or 4 cm in length, and histological cross-sections were prepared (Fig. 4, Fig. S2). We found that pith cells in the 1- and 2-cm stems of det3-1 and clv3-8 det3-1 mutants were round, whereas those in stems that were at least 3 cm in length exhibited severely distorted shapes (Fig. 4). Furthermore, ultraviolet illumination revealed ectopic lignification in det3-1 mutants after pith cells were deformed, whereas in WT plants, lignification was restricted to the xylem (Fig. S3). Therefore, we concluded that, although pith cell distortion basically correlates with stem aging, it is further enhanced following the cracks in clv3-8 det3-1 mutant stems.

Based on the observations described above, stem cracks may slow growth further by promoting pith cell lignification or increase growth by freeing the inner tissues from a mechanically resistant outer layer. To test these scenarios, we compared real and theoretical cross-sectional areas, calculated from stem section perimeter measurements. To estimate the theoretical cross-sectional area, we assume that stem cross-section is circular. Therefore, ratios between real and theoretical cross-sectional areas will be greater in cracked stems than in uncracked stems if inner tissue in cracked stems has expanded. A ratio of 100% means that there is no deviation between real and theoretical cross-sectional area. The ratios calculated for cracked clv3-8 det3-1 stems were significantly higher (121.15±14.51; mean±s.d.) than those calculated for uncracked stems (88.37±0.79; Fig. 3C). In addition, ratios calculated for stems with deeper or wider cracks were further increased, suggesting that substantial inner tissue growth occurred after the stems had cracked. Our kinetic analysis of stem development (Fig. 4) also showed that clv3-8 det3-1 stem cross-sectional areas increased significantly, even after pith cell deformation. These findings suggest a scenario in which stem cracks unleash inner tissue growth by weakening epidermal growth constraints.

Consequently, a positive feedback loop occurs in clv3 det3 stems: excess of growth fuels stem cracking, and cracking promotes excess of growth. Conversely, when the epidermis has not cracked yet, the growth of inner tissues is inhibited. These observations highlight the important role played by the epidermis in controlling growth and stem integrity and suggest that further analyses should focus on epidermal cells in the clv3 det3 mutant.

To analyze epidermal cell features, we compared four key growth stages based on clv3-8 det3-1 inflorescence stem length. As described above, stem portions were dissected at 5 mm from the base when the main inflorescence stems reached 1, 2, 3 or 4 cm in length, and histological cross-sections were prepared (Fig. 4, Fig. S2). Strikingly, we found an intriguing relationship between epidermal cell shape and stem growth. Despite the increase in inner stem volume and in the number of pith cells, the number of epidermal cells in the stem remained constant throughout development in all genotypes (Fig. 5A,B). In addition, there were no marked changes in the cross-sectional areas of epidermal cells in any of the genotypes or growth stages examined (Fig. 5C). However, the shapes of cells varied dramatically during clv3-8 det3-1 stem development, most notably between stems that were 1 and 3 cm in length. We quantified this morphological trait by calculating epidermal cell aspect ratios (Fig. S4), and the results clearly indicated that epidermal cell flattening only occurred in clv3-8 det3-1 mutants (Fig. 5D,E). These results suggest that epidermal cells respond to the increase in inner tissue volume during stem growth by halting their expansion, which produces flatter epidermal cells in clv3 det3. This observation is consistent with the epidermis restricting growth in the stem. To gain insight into the biophysical mechanism underlying the role of the epidermis in clv3 det3, we quantified its mechanical properties.

Anatomically, epidermal cells are distinct from inner plant cells because of their thickened cell walls. For example, in light-grown sunflower hypocotyls, epidermal cell walls are approximately 2 µm thick, whereas typical cortex cell walls are only approximately 0.2 µm thick and the walls of pith cells are even thinner (∼0.1 µm; Kutschera, 1992). However, cell walls do more than simply maintain cell shape, they also play a central role in controlling growth patterns via mechanical feedback. Consequently, cell wall properties must be measured to understand whether they actively resist, via stiffening, or yield to stress (Milani et al., 2013; Sassi and Traas, 2015). In the simplest scenario, crack occurrence could reflect decreased epidermal strength. However, cracks are not always associated with weaker cell walls. For example, mature fruits crack due to cell–cell adhesion defects, and their epidermal cells exhibit increased outer wall stiffness instead (Szymańska-Chargot et al., 2016). The apparent stretching of epidermal cells in the clv3 det3 mutant before cracks appear tends to support the former scenario. To test this hypothesis, we measured the average stiffness of epidermal cell walls in our lines.

More specifically, we measured the apparent elastic modulus of epidermal cells at the stem surface using atomic force microscopy (AFM; Milani et al., 2011, 2014). Indentations were made in single epidermal cells at two depths: <200 nm to measure wall stiffness without involving other parameters; and 0.5-1 µm to obtain more integrated, large-scale measurements of wall stiffness, albeit at depths that are slightly more sensitive to parameters such as turgor pressure and wall geometry (Milani et al., 2013). Note that neither of these indentation depths were as thick as the epidermal cell wall. We found that, at both depths, the mean values of cell wall stiffness in clv3-8 det3-1 mutants were approximately half those of WT values (Fig. 6).

In principle, this effect may be due to clv3-8, det3-1, or a synergistic interaction involving both mutations. Therefore, we also analyzed the single mutants. This analysis demonstrated that the reduction in wall stiffness was entirely attributable to the det3-1 mutation, consistent with previous results that showed the det3 mutation is epistatic to clv3 and also with the original biochemical function of DET3 in promoting etiolation (Schumacher et al., 1999). Altogether, these observations suggest that DET3 is necessary to enable the epidermis to mechanically reinforce its cell walls and constrain the growth of inner tissues.

To test the hypothesis that DET3 restricts growth in the epidermis, we sought to complement the det3 mutation by generating lines that expressed DET3 specifically in the epidermis. We used the PROTODERMAL FACTOR 1 (PDF1) promoter (pPDF1:DET3 clv3-8 det3-1 hereafter), which is exclusively expressed in the epidermis (Abe et al., 2001; Kawade et al., 2013), in the clv3-8 det3-1 background. Using these new transgenic lines, we tested whether WT DET3 activity expressed in the epidermis was sufficient to suppress the cracked stem phenotype of the clv3 det3 mutant.

The overall morphology of pPDF1:DET3 clv3-8 det3-1 plants indicated that normal vegetative and stem growth were largely rescued compared with clv3-8 det3-1 (Fig. 7A, Fig. S5). Although pPDF1:DET3 clv3-8 det3-1 plants remained small, plant height and size were notably superior to their clv3-8 det3-1 counterparts.

Next, we quantified the frequency of stem cracks in pPDF1:DET3 clv3-8 det3-1 complementation lines. The frequency of stem cracking in the pPDF1:DET3#1 clv3-8 det3-1 and pPDF1:DET3#14 clv3-8 det3-1 transgenic lines was restored to 20% and 40% of clv3-8 det3-1, respectively (Fig. 7B). Furthermore, the cracks in the stems of pPDF1:DET3 clv3-8 det3-1 plants appeared to be narrower and shallower than those observed in clv3-8 det3-1 plants, where large regions of pith tissue were exposed (Fig. 7C; Maeda et al., 2014).

Next, we investigated the relationship between cracks and stem growth in the pPDF1:DET3 clv3-8 det3-1 lines. Whereas the first cracks in clv3-8 det3-1 stems occurred when plants were young or their stems were still short, cracks in pPDF1:DET3 clv3-8 det3-1 stems occurred during later stages of growth or when the stems were longer (Fig. 7D). These results indicated that pPDF1:DET3 clv3-8 det3-1 transgenic lines exhibited significant suppression in overall cracking frequencies and a significant delay in the appearance of the first stem crack.

Based on these observations, we prepared stem cross-sections to study the inner morphology of pPDF1:DET3 clv3-8 det3-1 plants (Fig. 8). The cross-sectional areas of the pPDF1:DET3 clv3-8 det3-1 lines had increased by up to 150-200% compared with clv3-8 det3-1 plants (Fig. 8A,B). The pith cell distortion phenotype remained in the pPDF1:DET3 clv3-8 det3-1 lines but was less marked than in clv3-8 det3-1 plants, and it was restricted to small clusters of cells (Fig. 8A). Importantly, the number of epidermal and pith cells had increased significantly, correlating with the increase in cross-sectional area (Fig. 8C,D). Taken together, these findings strongly suggest that epidermis-targeted expression of DET3 effectively restores the ability of the epidermis to resist mechanical stress from the inner stem tissue and that the epidermis acts as a load-bearing layer to maintain stem integrity.

Based on our results, we can now provide a scenario explaining the appearance of cracks in clv3-8 det3-1 stems. The det3-1 mutation leads to weaker walls, probably owing to defects in exocytosis and cell wall synthesis. The clv3-8 mutation generates a large pool of growing cells, amplifying the mechanical conflict between inner cells and the epidermis. When combined, the det3 and clv3 mutations increase tension in the epidermis. The det3-1 mutation also weakens the walls of cells in the epidermis, and these cells become stretched to the point of mechanical failure, leading to cracks. In inner tissues, lignified pith cells appear in the det3-1 and clv3-8 det3-1 mutants because weakening of the cell walls triggers the wall integrity pathway (Hématy et al., 2007). Our results also suggest that the cracks relieve the constraints on growth imposed by the epidermis and allow inner cells to expand further, in a positive feedback loop. Because the expression of DET3 in the epidermis is sufficient to rescue most of these defects, our results are consistent with the epidermal growth theory and suggest that DET3 plays a key role in maintaining stem integrity.

Our results are consistent with a simple scenario in which the epidermis constrains growth. This hypothesis was developed to explain the results of peeling experiments (Baskin and Jensen, 2013; Peters and Tomos, 2000) and is compatible with Hofmeister's tissue tension theory (Hofmeister, 1859). Here, we not only confirm this biophysical view of the stem but also provide evidence that such mechanical balance requires the regulation of cell number (CLV3) and wall stiffness (DET3). The spontaneous expansion of pith tissue cells after stem cracking (Fig. 3C) illustrates this point. Our findings show that a fundamental process in stem development can resist mechanical conflict between the inner tissues and the epidermis (Galletti et al., 2016). However, there is a threshold beyond which the tissue becomes unable to resist.

The stretched and flattened epidermal cells found in the clv3 det3 mutant are particularly interesting because cell wall defects often lead to responses that rescue the original defect and may even overcompensate for it. This may explain the lignification that occurs following wall weakening. The det3-1 mutant has weak walls and the cells break when the level of tension becomes too great. However, because this does not occur in the epidermis, DET3 may play a different role in the wall integrity pathway there. However, lignification is triggered in other tissues in the det3-1 mutant, as it is in the theseus1 mutant, which is deficient in a protein involved in sensing cell wall integrity. Importantly, even if the cell wall integrity pathway is triggered, it may be unable to produce a response capable of withstanding the formidable mechanical stress built up in the inner tissue. Here, the positive feedback that stimulates growth following crack emergence may further explain why stems cannot withstand the excessive mechanical stress from inner tissues in clv3-8 det3-1 because when a local defect occurs in the epidermis, the feedback loop would rapidly amplify any damage already present.

Because the biochemical role of DET3 is primarily related to vesicle trafficking, it is unlikely to affect the mechanical anisotropy of cell walls, at least directly. For example, the export of cellulose synthase to the plasma membrane may be impaired, but any defects in cellulose microfibril orientation would rather be the consequence of defective wall remodeling. Yet, DET3 may still have a role of directional growth, notably assuming that exocytosis can be polarized and that walls can be mechanically heterogeneous. Such mechanical polarities have been observed in hypocotyls (Peaucelle et al., 2015) and cotyledons (Majda et al., 2017). The presence of longitudinal cracks in clv3 det3 is consistent with a failure to resist tensile stress magnitude and/or direction. Whether DET3 function also contributes to anisotropic wall reinforcement remains to be explored.

The clv3 det3 mutant demonstrates that epidermal cells weaken and become flatter when the growth rate of inner tissues becomes too high. Coordinating the necessary responses requires a perception mechanism, which may involve mechanotransduction. This echoes the concept of proprioception, the ability of an organism to sense its own shape and growth (Hamant and Moulia, 2016), and the presence of cracks may be viewed as a defect in that pathway too. The role of DET3 in proprioception may be an interesting avenue for future research. Our histological studies have provided new insight into why plant stems may crack during development. However, further molecular analyses are needed to unravel the role of mechanoperception, identify signaling components and understand the transduction mechanisms that are important in plant stem development. Finally, our findings may be relevant for biomechanical applications wherein active materials, amenable to auto-repair or self-reinforcement in response to stress, may be used constructively.

The WT A. thaliana ecotype used in this study was Columbia-0 (Col-0). All other mutants were in the Col-0 background, except clv1-4 and clv2-1, which were both in the Landsberg erecta (Ler) background. The clv1-4, clv2-1, clv3-8 and det3-1 single mutants, and all corresponding double mutants with det3-1, have been described previously (Maeda et al., 2014). The clv3-101 allele is a novel allele of clv3 that we isolated in the det3-1 background, as described previously (Ferjani et al., 2015; Maeda et al., 2014). pPDF1:DET3 lines were generated for this study, as described below. Seeds were sown on rockwool (Nitto Boseki), watered daily with 0.5 g l⁻¹ Hyponex solution (Hyponex) and grown at 22°C under a 16 h light/8 h dark cycle with white light fluorescent lamps at approximately 50 µmol m⁻² s⁻¹, as described previously (Ferjani et al., 2013).

Images showing overall plant phenotypes were recorded using a Nikon D5000 NIKKOR lens and AF-S Micro NIKKOR 60-mm digital camera (Nikon). Images showing cracked stems were recorded using a Leica M165 FC stereoscopic microscope connected to a CCD camera (DFC 7000T; Leica Microsystems). The appearance of cracks was monitored with the naked eye and by using an ordinary loupe.

For the histological cross-sections and longitudinal sections, portions of the first internode from the main flowering stem were dissected and fixed overnight in formalin–acetic acid–alcohol (4% formalin, 5% acetic acid, 50% ethanol) at room temperature. Then, samples were dehydrated using a graded series of ethanol washes [50%, 60%, 70%, 80%, 90% and 95% (v/v); 30 min each] and stored overnight in 99.5% (v/v) ethanol at room temperature. Next, fixed specimens were embedded in Technovit resin (Kulzer), in accordance with the manufacturer's instructions, and sectioned using a microtome (RM2125 RTS; Leica Microsystems). Sections were stained with Toluidine Blue or Safranin or double stained with Astra Blue–Safranin and photographed under a microscope (Leica DM6 B) connected to a CCD camera (DFC 7000T; Leica Microsystems). Images were analyzed using ImageJ software (LOCI; University of Wisconsin, Madison, WI, USA). Statistical analyses were performed using R software (ver. 3.6.1; R Development Core Team).

A Catalyst Bioscope (Bruker) mounted under a MacroFluo optical epifluorescence macroscope (Leica Microsystems) was used for the AFM indentation experiments. A 2× Plano objective was used to observe the apices (Leica Microsystems). To record surface topology and stiffness modulus maps, we used PeakForce QNM AFM mode (Bruker/Veeco), a NanoScope V controller and NanoScope software (ver. 8.1; Bruker). All quantitative measurements were recorded using SD-Sphere-NCH 0.8-µm-diameter spherical probes (NANOSENSORS). The spring constant of cantilevers (42 N m⁻¹) was measured using the thermal tuning method (Hutter and Bechhoefer, 1993; Lévy and Maaloum, 2009). All measurements were recorded under water at room temperature, and the standard cantilever holder for operating in liquid environments was used. The 30 mm Petri dishes containing each sample (∼2-3 cm of stem) were placed on an XY stage in a customized sample holder. Then, the AFM head was mounted on the stage and positioned with the cantilever over relatively flat areas that were free from trichomes. To extract elastic moduli from the force–indentation retract curves, we used the Hertz model, which applies to the adhesive contact between two elastic, thick isotropic solids (Derjaguin et al., 1975). Images were obtained using the following parameters: scan size, 50 µm; force, 423 nN. For the point and shoot mode, the following parameters were used: size, 50×15 µm, 20×20 points; trig threshold, 247 nm; ramp size, 4 µm.

The full-length cDNA of the DET3 gene coding region was amplified by PCR using a pair of oligonucleotides: DET3-B1-FW 5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTATGACTTCGAGATATTGGGTG-3′ and DET3-B2-RV 5′-GGGGACCACTTTGTACAAGAAAGCTGGGTTTAAGCAAGGTTGATAGTGAAG-3′.

The amplified fragments were fused into pDONR201 using the BP recombination system (Invitrogen). The DNA sequences of all clones were confirmed by sequencing. The LR reaction was used to convert the resulting vector (pDONR201–DET3), pENTP4P1R–pPDF1 (Kawade et al., 2013) and the R4 gateway binary vector R4pGWB501 (Nakagawa et al., 2008) into R4pGWB501–pPDF1–DET3. The final constructs were used to transform WT plants via the floral dip method (Clough and Bent, 1998). Several independent homozygous T₃ lines expressing the DET3 gene driven by PDF1 promoters from a single T-DNA insertion locus were identified in the WT background. Representative lines (pPDF1:DET3#1 and pPDF1:DET3#14) were then crossed to clv3-8 det3-1 to obtain the final complementation lines, and these were used in the study.

We thank T. Nakagawa (Shimane University) for providing the R4pGWB501 binary vector, and K. Kawade (National Institute for Basic Biology) for providing the pENTP4P1R-pPDF1 vector.