Contribution of vasculature to stem integrity in Arabidopsis thaliana Free

The ability to bear the load of aboveground organs is a fundamental feature of land plant evolution (Serrano-Mislata and Sablowski, 2018). A key feature of land colonization was the differentiation of stiff sap channels and fibers in the so-called vascular plants. From a physiological perspective, xylem- and phloem-containing stem vascular bundles serve to transport water, nutrients, and signaling molecules (e.g. RNA, organic acids, and signal peptides), thereby allowing long-distance communication throughout the plant body (Satoh, 2006; Kehr and Buhtz, 2008). Here, we focus on the mechanical contribution of the vasculature in stem growth.

Cross-sections of plant stems usually exhibit radial symmetry. This is presumed to allow considerable adaptability to internal perturbations (e.g. growth variations) and external perturbations (e.g. wind) (Moulia et al., 2021). This radial symmetry is also present in stem histology: the various cell types that compose the inflorescence stem of Arabidopsis thaliana (hereafter Arabidopsis) exhibit radial symmetry. Because the weeping mutants of cherry species have a defect in secondary cell wall biosynthesis (Nakamura et al., 1994), the mechanical role of vascular bundles in maintaining an upright posture has been extensively explored (Moulia et al., 2019). NAC SECONDARY CELL WALL THICKENING PROMOTING FACTOR 1 (NST1) and NST3 are master transcription factors of secondary cell wall biosynthesis in the interfascicular fiber cells of Arabidopsis (Mitsuda et al., 2005, 2007). nst1-1 nst3-1 double mutants show a typical pendent stem phenotype, suggesting a role for NSTs in maintaining plant posture stability (Mitsuda et al., 2005, 2007; Sakamoto and Mitsuda, 2015; Sakamoto et al., 2016, 2018); however, loss-of-function mutants of these genes do not exhibit stem cracking. Furthermore, vascular bundles respond to weight through a process known as vertical proprioception (Alonso-Serra et al., 2020). Late bundle differentiation in wood can also exhibit asymmetric signatures; for example, when stems are oblique (e.g. tension wood in tree lateral branches; Clair et al., 2018) or when stems undergo artificial flexion (Roignant et al., 2018).

Stem integrity consists of both mechanical (material properties) and structural (topological properties) integrity. It is challenged by stress magnitude (force per unit area) up to a breaking point (defining the strength of the stem). In previous work, we showed that the mechanical and structural integrity in young stems relies on a load-bearing epidermis. Indeed, the epidermis of plant aerial organs is continuously subjected to tensile stress, leading to mechanical reinforcement in a feedback loop, whereas inner tissues are subjected to compression (Kutschera and Niklas, 2007; Trinh et al., 2021). We identified genetic contexts in which the load-bearing role of the epidermis is compromised. Additionally, we observed deep cracks in the inflorescence stems of clavata3-8 de-etiolated3-1 (clv3-8 det3-1) mutants (Maeda et al., 2014). Our analysis revealed that the det3-1 mutation leads to weakened cell walls (Asaoka et al., 2021) and reduced cell elongation (Schumacher et al., 1999; Ferjani et al., 2013), whereas the clv3-8 mutation leads to enhanced cell proliferation. These effects produce considerable mechanical heterogeneity, which ultimately leads to mechanical failure in inflorescence stems (Maeda et al., 2014; Asaoka et al., 2021).

Such stem-cracking phenotypes are rare among Arabidopsis mutants. So far, with the exception of clv3-8 det3-1 mutants, all Arabidopsis mutants with stem-cracking phenotypes have demonstrated cell-cell adhesion defects (e.g. quasimodo mutants; Bouton et al., 2002) or altered auxin biosynthesis, although these phenotypes exhibit low penetrance (Hentrich et al., 2013). In the footsteps of D'Arcy Thompson (Thompson, 1917), patterns of mechanical reinforcement in tissues may reveal the locations of mechanical heterogeneities and provide potential targets for regulating stem integrity. Plants are presumed to mechanically reinforce the epidermis in aerial organs because this tissue is load-bearing (Hamant and Haswell, 2017; Kutschera and Niklas, 2007; Nakayama et al., 2022). Another obvious target is the vasculature, a stiff inner tissue. We thus hypothesized that vascular bundles also contribute to inflorescence stem integrity. To assess this contribution, we examined the nst1-1 nst3-1 mutant. We conducted a genetic screen of transformants that expressed an activator or repressor version of an Arabidopsis transcription factor; we found that plants expressing a repressive domain fused to INDETERMINATE DOMAIN9 (IDD9; also known as BALDIBIS, AT3G45260) displayed deep vertical cracks along the inflorescence stem axis in the nst1-1 nst3-1 mutant. Subsequent functional analysis allowed us to determine the relative contributions of deep stem tissues to mechanical integrity in the stem.

Large-scale genetic screening was carried out to identify new regulatory components of cell wall synthesis, whereby genes encoding transcription factors were fused with an activator (VP16) or repressor domain (SRDX; Hiratsu et al., 2003), then systematically transformed into nst1-1 nst3-1 double mutants under the control of the NST3 promoter. This screening method has been effective for identifying novel factors of secondary wall formation, as well as AP2/ERF transcription factors involved in primary cell wall synthesis (Sakamoto and Mitsuda, 2015; Sakamoto et al., 2018). The above seed library was used to identify plants that exhibited a cracking phenotype. After screening, we noticed spontaneous cracks in pNST3:IDD9:SRDX transformants in the nst1-1 nst3-1 background (Fig. 1A; Fig. S1A). We then generated transformants of pNST3:IDD9:SRDX in the wild-type (WT) background and found that they also displayed pendent inflorescence stems with deep cracks, identical to the phenotypes of their counterparts in the nst1-1 nst3-1 background (Fig. 1A-C). This finding suggests that stem cracking and the pendent phenotype are triggered by pNST3:IDD9:SRDX, independent of mutations in NST1 and NST3. Note that in both WT and nst1-1 nst3-1 backgrounds, stem cracks were abundant in the first node of the main inflorescence stem, i.e. in the oldest part of the stem where there is more vasculature (Fig. 1B,D; Fig. S1B-W).

Our previous work involving the clv3-8 det3-1 mutant showed that stem cracking is a dynamic process that relies on growth (Asaoka et al., 2021). We thus analyzed the time-dependent dynamics of stem cracking in pNST3:IDD9:SRDX and pNST3:IDD9:SRDX nst1-1 nst3-1 lines. pNST3:IDD9:SRDX inflorescence stems grew normally for approximately 10 days after bolting; subsequently, their stems displayed cracks. Stem growth monitoring of WT, nst1-1 nst3-1, and pNST3:IDD9:SRDX plants revealed that stem growth in all plants slowed immediately after bolting; faster growth (maximum, ∼3 cm day⁻¹) occurred 7-10 days after bolting (Fig. 1E). In contrast to WT and nst1-1 nst3-1, which continued to grow during the study period, all four pNST3:IDD9:SRDX and pNST3:IDD9:SRDX nst1-1 nst3-1 lines exhibited earlier termination of apical growth with comparable dynamics after they had reached a height of 20-25 cm (Fig. 1E). This finding confirmed that stem cracking is triggered by pNST3:IDD9:SRDX, independently of mutations in NST1 and NST3.

We also monitored the timing of the first crack (Fig. 1F). The first crack in the stem occurred primarily 9-11 days after bolting, which corresponded to the period when stems were 13-19 cm. Notably, nst1-1 nst3-1 stems began bending when their height reached 15-18 cm. This suggests that stem cracking occurs frequently during the period when stems grow normally and is not a consequence of growth arrest. Instead, the apical growth limitation in pNST3:IDD9:SRDX (in WT or nst1-1 nst3-1 backgrounds) is a secondary effect of cracks. We also observed that the timing of bending did not always occur after cracking (Fig. 1G). Therefore, we conclude that the pendent phenotype does not trigger stem cracking.

To compare inflorescence stem inner tissue organization between WT and pNST3:IDD9:SRDX, we prepared histological sections stained with Astra Blue and Safranin. Under these conditions, cellulose is stained blue and lignin is stained red; this double-staining approach allows rapid visual discrimination between the primary and secondary cell walls, which appear blue and red, respectively, within the histological sections.

At mature stage, observation of cross-sections indicated that lignin deposition was suppressed in fiber tissue of nst1-1 nst3-1, whereas red staining was clearly observed in the WT stem, as expected (Fig. 2; Fig. S2A). Cross-sections of pNST3:IDD9:SRDX lines were devoid of lignin staining, similar to nst1-1 nst3-1 (Fig. 2); this finding indicates that, although stem cracking in the pNST3:IDD9:SRDX lines may involve defects in lignin deposition, such defects are not the sole cause of stem cracking. We thus suspected that stem cracking also involves geometrical and topological defects.

Quantification of stem cross-sectional areas and other cellular parameters revealed that pNST3:IDD9:SRDX and pNST3:IDD9:SRDX nst1-1 nst3-1 shared striking similarities (Table 1). There were no significant differences in cortex or endodermal cell number among the lines examined (Table 1). Inflorescence stem inner cells, namely pith and interfascicular fiber cells, were counted together because of difficulties in distinguishing the boundary between these two tissues. There were 49% more inflorescence stem inner cells in nst1-1 nst3-1 than in the WT. However, this increase was much less pronounced in the pNST3:IDD9:SRDX (28%) and pNST3:IDD9:SRDX nst1-1 nst3-1 (9%) lines. These results suggest that stem cracking is not caused by an excess of inner cells, as in clv3 det3 (Asaoka et al., 2021). Indeed, stem cracking in pNST3:IDD9:SRDX was negatively correlated with inner cell number.

Excessive cell expansion may fuel mechanical heterogeneity and cracks. To explore this possibility, we assessed stem cross-sectional area and cell number. pNST3:IDD9:SRDX stems were radially expanded up to approximately twofold, compared with the WT (Table 1). A similar trend was observed between pNST3:IDD9:SRDX nst1-1 nst3-1 and nst1-1 nst3-1 (Table 1). Notably, epidermal cell numbers in all backgrounds were similar to the epidermal cell number in the WT (Table 1). These results indicate that cell expansion, rather than proliferation, is enhanced in pNST3:IDD9:SRDX lines; such enhancement contributes to stem thickening. To explore this behavior further, we next analyzed cell expansion across the stem radius.

Fiber cells are classified into two types, namely interfascicular or xylary, according to their position. Analysis of stem sections showed that both types of fiber cells were larger in pNST3:IDD9:SRDX lines (Fig. 2; Fig. S2A). We focused on interfascicular fiber and evaluated cell morphologies by quantifying cross-sectional area and shape.

First, we measured the distance between interfascicular fiber cells and the inner-side edge of the endodermis (r), as well as the area (Fig. 3B; Fig. S3A). We used this approach because of heterogeneity in fiber cell size between layers; fiber cells located on the inner side appeared larger than fiber cells located on the outer side, even in the WT. Thus, even when r was zero (i.e. at the outermost layer of interfascicular fiber), cells were significantly expanded in pNST3:IDD9:SRDX lines (Fig. 3A). The difference became more pronounced as r increased (Fig. 3B). We next calculated the cell aspect ratio, defined as the ratio of the maximal length of the radial cell axis to the maximal horizontal cell width (Fig. 3C,D; Fig. S3B). The cell aspect ratio was greater in regions distant from the endodermis in pNST3:IDD9:SRDX, although there was no difference between the WT and pNST3:IDD9:SRDX when r was zero. These observations were supported by the radially expanded interfascicular fiber cells in longitudinal sections of pNST3:IDD9:SRDX stems (Fig. 3E; Fig. S4). Because NST3 is normally expressed in fibers, abnormal cell expansion in pNST3:IDD9:SRDX fiber is likely the primary cause of the SRDX-fused IDD9 phenotype.

Next, we focused on epidermal tissue where cracks appeared. Scanning electron microscopy analysis of inflorescence stem surfaces showed a depression at the junction between epidermal cells. This depression was more pronounced in cracking plants than in the WT or nst1-1 nst3-1, which displayed a smooth epidermal surface (Fig. 4A). Magnified images of vascular bundles of stem cross-sections revealed that contiguity between adjacent epidermal cells was considerably reduced in pNST3:IDD9:SRDX (Fig. S2A,B). Epidermal cells were drastically deformed, with a more elongated, horizonal direction toward the stem periphery and flatter ellipses in pNST3:IDD9:SRDX lines (Fig. 4B; Fig. S2C). In addition to epidermal cell number (Fig. 2), the epidermal cell area (viewed from above) was comparable between pNST3:IDD9:SRDX and the WT (Fig. 4C). These observations suggest that epidermal cells were stretched during stem maturation because fiber tissue expansion increased tensile stress level in the epidermis.

To confirm the above findings, we determined cell wall thickness of mature inflorescence stem epidermis by transmission electron microscopy (Fig. 4D,E). This analysis revealed that the inner periclinal wall was thinner in pNST3:IDD9:SRDX, which was 0.95±0.24 µm in the WT or 0.69±0.19 µm in pNST3:IDD9:SRDX (mean±s.d., n=9 cells, P<0.05 by paired, two-tailed Student's t-test). In contrast, the anticlinal wall was thicker in pNST3:IDD9:SRDX, which was 0.71±0.09 µm in the WT or 0.93±0.15 µm in pNST3:IDD9:SRDX (n=9 cells, P<0.05 by paired, two-tailed Student's t-test). These observations suggest that the epidermal cells in pNST3:IDD9:SRDX respond differently to the tensile stress. Note that, as shown in other organs and plant species already (Kutschera and Niklas, 2007), the epidermis cell wall, especially the outer wall, was thicker compared with that of pith cells (Fig. 4F,G).

Thus far, we showed that an increase in stem volume, enhancement of expansion in fiber cells, and flattening of the epidermis are morphological changes characteristic of cracked stems; these changes represent an additional, cell expansion-based mechanism that underlies stem integrity, along with the previously reported cell proliferation-based mechanism (Asaoka et al., 2021).

Next, to analyze whether changes in tissue and/or cellular parameters affect stem cracking, we generated pNST3:IDD9:SRDX transformants in the clv3-8 background (Fig. 5A). clv3-8 nst1-1 nst3-1 was generated for morphological comparison, as well. As mentioned above, the inflorescence stem inner volume is increased, but the number of epidermal cells remains constant in pNST3:IDD9:SRDX (Fig. 1). We hypothesized that if cell number is a limiting factor for stem integrity, the number of epidermal cells would increase in pNST3:IDD9:SRDX clv3-8; this change would reduce the tension from the inner tissue towards the epidermis, thus partially reducing stem-cracking frequency. Alternatively, if cell growth is a limiting factor for stem integrity, pNST3:IDD9:SRDX clv3-8 stems would contain more expanded cells; this would lead to a cracking frequency similar to or greater than the frequency observed in the WT background.

Cracking was not suppressed in pNST3:IDD9:SRDX clv3-8 stems (Fig. 5B). Surprisingly, the number of cracks increased in pNST3:IDD9:SRDX clv3-8 stems (Fig. 5C). In extreme cases, we observed more than ten cracks in a single inflorescence stem (Fig. 5C). In addition, pNST3:IDD9:SRDX clv3-8 exhibited earlier slowing of primary inflorescence stem elongation, compared with clv3-8 and clv3-8 nst1-1 nst3-1, as observed in pNST3:IDD9:SRDX transformants in the WT background (Fig. 1E; Fig. S5A). This result shows that both cell expansion- and proliferation-based mechanisms contribute to stem integrity in an additive manner, in combination with the previously demonstrated role of the load-bearing epidermis.

Our detailed observations of stem cracks revealed several common structural features. In addition to apparent stem cracks, we found that small cracks at the cell junction on the stem surface were often located in the vicinity of a large crack (Fig. 6A). Within a fully collapsed stem (i.e. displaying a large axial crack), some tissues were broken by twisting; in some cases, inner cells were not completely split (Fig. 6B,C; see also Fig. S1A). These observations suggest that cracking begins at several distinct spots at the stem surface; tensile stress then pulls cells further apart and triggers tissue collapse on the inner side of the stem. Therefore, the inner tissue may be heterogeneous in terms of mechanical stiffness; in addition to the slight twisting nature of stems, there may be mechanical hotspots where tissue collapse is less likely to occur.

More than 100 histological sections have been carefully examined in this study and in previous studies (Asaoka et al., 2021); thus far, none has exhibited disrupted vascular bundles. Based on these assessments, cracking can be classified into three different modes (Fig. 6D-F): cracks that occur within interfascicular fiber issue (Fig. 6D); cracks that occur on both sides of a vascular bundle, avoiding the vascular bundle and reaching the pith tissue (Fig. 6E); and cracks that occur on the outer side and cannot deepen because they are blocked by a vascular bundle (Fig. 6F). These observations suggest that although the stiffness of vascular bundles contributes to stem integrity, it also generates weak points in its vicinity.

This relationship is even more apparent when considering the pNST3:IDD9:SRDX clv3-8 background. Although the number of cracks increased in pNST3:IDD9:SRDX clv3-8, each crack was smaller and shallower than the cracks in pNST3:IDD9:SRDX, where inner tissues were deeply split lengthwise at the cracking sites (Fig. S6). Observations of the stem inner tissue revealed that cross-sectional areas of pNST3:IDD9:SRDX clv3-8 and pNST3:IDD9:SRDX clv3-8 nst1-1 nst3-1 stems were 150-170% larger than the cross-sectional areas of clv3-8, although the numbers of inner cells, epidermal cells and vascular bundles were similar among pNST3:IDD9:SRDX clv3-8, pNST3:IDD9:SRDX clv3-8 nst1-1 nst3-1, and clv3-8 (Fig. 5D; Table 2). In addition, interfascicular fiber cell size increased throughout the tissue in pNST3:IDD9:SRDX clv3-8 and the sizes of individual pith cells were nearly equal to the sizes of such cells in the WT and clv3-8 (Table 2; Fig. S5B,C). Altogether, these findings show that the local strength, number and arrangement of stem vascular bundles may fundamentally change stem mechanical integrity, resulting in the cracking pattern observed in pNST3:IDD9:SRDX clv3-8.

As described above, in all cracking genotypes primary stems grew normally after emergence and then slowed when they reached ∼20 cm, whereas primary stems in the WT, nst1-1 nst3-1, and clv3 continued to grow until they reached ∼30 cm (Fig. 1E; Fig. S5A). Moreover, wilting and withering frequently occurred in the vicinity of inflorescence stem apices at advanced growth stages (≥40 days after seed sowing) in the pNST3:IDD9:SRDX lines (Fig. 7), although there was no obvious difference relative to the WT in younger plants (Fig. S7). Cracking, wilting and withering also occurred in lateral branches (data not shown). In wilted stems, although interfascicular fiber cells were highly compressed, vascular bundles maintained their structure, indicating that interfascicular fiber cells are less stiff than vascular tissues (Fig. 7G-J). Seed formation in pNST3:IDD9:SRDX young fruits properly progressed, allowing the harvest of next-generation seeds; however, their fruits displayed an indehiscent phenotype. This indehiscent phenotype was also observed in nst1-1 nst3-1 mutants and was attributed to a loss of lignification at the conjunction of the two carpels (Mitsuda and Ohme-Takagi, 2008). Taken together, these observations indicate that stem cracking does not represent an instantaneous threat for plant survival, highlighting the remarkable robustness of plant development. Nonetheless, the irreversible loss of mechanical and structural integrity has a long-term effect on the physiological properties of the stem.

In studies of inflorescence stems involving forward genetics, the descriptions of stem morphological changes in mutants have mostly been limited to reduced stem elongation or pendent phenotypes because of defects in secondary cell wall formation in fiber cells. In this context, we recently demonstrated the importance of the epidermis in maintaining stem mechanical integrity through analysis of the stem-cracking phenotype of clv3-8 det3-1 (Asaoka et al., 2021). The cracking phenotype alone represents emerging concepts in stem development: the presence of substantial differences in elastic strain between the epidermis and inner tissues (Baskin and Jensen, 2013); the visualization of mechanical conflicts reflected by tissue breakage; and the regulation of cell growth, followed by maintenance of stem structure, as a prerequisite for proper stem development and integrity.

Thus far, the contributions of inner tissues to stem integrity have not been experimentally assessed. Furthermore, although stem cracks in clv3-8 det3-1 can be regarded as an extreme manifestation of mechanical integrity failure in living tissues, the severe dwarfism of this mutant may conceal the nature of stem development. Importantly, organ cracking was observed only in inflorescence stems, not in leaves or individual floral organs, although clv3-8 det3-1 mutations apparently affect the entire plant (Maeda et al., 2014). These findings may reflect the limited growth flexibility in cylindrical stems compared with flat leaves, where the stress pattern is generally isotropic (Zhang et al., 2011; Verger et al., 2018); the shape diversity of leaves emphasizes their capacity to grow in multiple directions.

In this study, we revealed the contribution of inner cell expansion to stem integrity, as well as its relationship with cell proliferation. Initially, stem cracking appeared to involve drastic morphological changes in inner tissue, as we previously reported (Maeda et al., 2014; Asaoka et al., 2021). However, our current findings reveal that stem cracking can be primarily attributed to excessive cell expansion with the expression of pNST3:IDD9:SRDX. There were no drastic changes in cellular parameters between pNST3:IDD9:SRDX lines and the WT, with the exception of increases in pith cell and stem cross-sectional areas (Fig. 2, Table 1). In addition, most morphological changes observed in pNST3:IDD9:SRDX could be regarded as a consequence of excessive expansion of interfascicular fiber cells. For instance, epidermal cells are likely deformed because of the high tension generated by the volumetric increase in inner tissues (Fig. 4). Increases in pith cell size may also contribute to increase tensile stress in the epidermis in pNST3:IDD9:SRDX; however, this was not required to trigger stem cracking. Because pith cell size was comparable between the WT and pNST3:IDD9:SRDX clv3-8 (Fig. 5D; Table 2), increase in pith area seems to contribute to yield a higher tension on the epidermis.

Importantly, enhanced interfascicular fiber cell expansion was observed in both pNST3:IDD9:SRDX and pNST3:IDD9:SRDX clv3-8 (Fig. 3; Fig. S5B,C). In pNST3:IDD9:SRDX, interfascicular fiber cells were larger and demonstrated more expansion in the stem radial direction. Such conflicts in growth magnitude and direction between the cortex and inner tissues may permit the disruption of stem integrity. However, this observation deserves a more refined assessment.

In particular, local mechanical conflicts between vascular bundles and interfascicular fibers can modulate cracking phenotypes. The finding that cells at the vascular bundle did not break (Fig. 6) implies that intercellular strength is higher in the vascular bundle than in parenchymatous tissue. This may be a result of cell geometry in vascular tissue. A combination of stacking pattern of fiber cells and stiff cell walls in vessels may add strength to vasculature, guarding against organ collapse. In the simplest scenario, inner tissues would experience compressive and tensile stress in contrast to the epidermis, which is under high tensile stress only. This finding also implies the existence of mechanical heterogeneities within the stele. Indeed, the presence of more numerous, but smaller, cracks in pNST3:IDD9:SRDX clv3-8 suggests that the number and distribution of vascular bundles may compensate for defects in fiber cell expansion. In this context, there would be no opportunity for deeper cracks to form because vascular bundles act as vertical pillars that sustain stem structure; an increase in the number of vascular bundles would confer additional mechanical strength by distributing the load.

Overall, stem cross-sections were almost round in pNST3:IDD9:SRDX and pNST3:IDD9:SRDX clv3-8, whereas they were nearly square in clv3-8 (Fig. 5D). Considering that the initial stem shape was comparable before the expression of pNST3:IDD9:SRDX and that the stem exhibits a distorted dome-shape in the clv3-8 mutant (Diévart et al., 2003), the pNST3:IDD9:SRDX clv3-8 stem became cylindrical in shape during stem growth. Because a circle is the maximal volume state for a specific peripheral length, excessive expansion of interfascicular fiber cells resulted in continuous generation of tensile stress at the epidermis until it reached the highest possible tension. Eventually, stem cracks could be triggered when pressure from inner tissue exceeds the load-bearing limit of the epidermis (Galletti et al., 2016; Kutschera and Niklas, 2007; Savaldi-Goldstein et al., 2007).

Earlier termination of stem elongation and stem apex wilting suggests that the effects of stem integrity loss extend beyond stem cracking (Figs 1F and 7; Fig. S5A). Earlier termination of stem elongation is a potential consequence of improper development because stem growth in pNST3:IDD9:SRDX was initially comparable with stem growth in the WT (Fig. 1E). The length of pith cells in this background was also similar to that in the WT, suggesting that shorter stems are caused by a loss of cell proliferation at the shoot apex, rather than a defect in cell elongation in pNST3:IDD9:SRDX (Fig. 3E). The direct connection with this phenotype remains unclear, but a change in vessel cell morphology, likely related to the expansion of surrounding fiber cells, was observed in cracking plants (Fig. S2A). Overall, we have demonstrated the importance of cell growth regulation and coordination between layers, particularly involving interfascicular fiber cells, the epidermis and the vascular bundle, for inflorescence stem integrity.

The WT used in this study was Columbia-0 (Col-0). The pNST3:IDD9:SRDX lines were obtained during screening as previously described (Sakamoto and Mitsuda, 2015; Sakamoto et al., 2018). The chimeric repressor construct (CRES-T lines) contains a transcription factor fused to SRDX, a plant-specific repression domain (Hiratsu et al., 2003). Stem cracks were found in three plants from 138 pNST3:TF:SRDX nst1-1 nst3-1 transgenics, which is a mixture of plants harboring 28 chimeric repressor constructs. Then, the coding sequence of IDD9 was amplified with the primer pair shown in Table S1. Individual pNST3:TF:SRDX constructs were prepared by the Gateway LR reaction from TF gateway entry clones (Mitsuda et al., 2010) into the pDEST_NST3p_SRDX_HSP_GWB5 plasmid (Sakamoto and Mitsuda, 2015). The resultant binary plasmid was transformed into Arabidopsis (WT) and nst1-1 nst3-1 plants (Mitsuda et al., 2007) to express pNST3:IDD9:SRDX using the floral dip method (Clough and Bent, 1998). At least 24 positive T1 seedlings were isolated on selective media for each transgenic line. The lines with a single insertion site were selected at the T2 generation according to the segregation ratio of seedlings on selective media. Two representative lines were confirmed to be homozygous for a single transfer DNA insertion at T3 and used for in-depth analyses.

Seeds were sown on rockwool, watered daily with 0.5 g L⁻¹ Hyponex solution (Hyponex, Tokyo, Japan), and grown in a growth room with a 16-h/8-h light/dark cycle and white light fluorescent lamps at 50 µmol m⁻² s⁻¹.

clv3-8 nst1-1 nst3-1 mutants were generated in this study by crossing nst1-1 nst3-1 and clv3-8 (Diévart et al., 2003; Maeda et al., 2014). nst1-1 and nst3-1 mutations were genotyped using the oligonucleotide primers listed in Table S1. pNST3:IDD9:SRDX clv3-8 was generated by crossing pNST3:IDD9:SRDX and the clv3-8 mutant. pNST3:IDD9:SRDX clv3-8 nst1-1 nst3-1 was generated by crossing pNST3:IDD9:SRDX nst1-1 nst3-1 and clv3-8 nst1-1 nst3-1.

Images of plant gross phenotypes were acquired with a digital camera (Nikon D5000 Nikkor lens AF-S Micro Nikkor 60 mm; Nikon or WG-6; RICOH). Images of stem cracks were acquired with a stereoscopic microscope (Leica M165 FC; Leica Microsystems) connected to a charge-coupled device (CCD) camera (DFC 7000T; Leica Microsystems). Cracks were monitored with the naked eye or using an ordinary loupe.

For histological sectioning, portions of the main inflorescence stem were dissected and fixed overnight in formalin–acetic acid–alcohol (4% formalin, 5% acetic acid, and 50% ethanol) at room temperature. The stem portions collected were 2-3 cm above the base of the main stem. Fixed specimens were dehydrated using a graded ethanol series [50%, 60%, 70%, 80%, 90% and 95% (v/v); 30 min per step] and stored overnight in 99.5% (v/v) ethanol at room temperature. Next, specimens were embedded in Technovit resin (Kulzer and Co.), in accordance with the manufacturer's instructions, and sectioned with a microtome (RM2125 RTS; Leica Microsystems). Sections were stained with 1% (w/v) Safranin and 0.5% (w/v) Astra Blue (Gärtner and Schweingruber, 2013) for rapid visual discrimination between the primary cell wall (stained blue) and secondary cell wall (stained red), then imaged under a microscope (Leica DM6 B; Leica Microsystems) equipped with a CCD camera (DFC 7000T; Leica Microsystems). Cross-sectional area and cellular parameters of stems were measured from images of histological sections. Cells were traced for area analysis or marked for counting; measurements were obtained using Fiji/ImageJ software v.2.0.0 (http://imagej.nih.gov/ij/).

For scanning electron microscopy, the basal section or crack of main inflorescence stems were dissected from the indicated growth stage. Samples were fixed overnight in formalin–acetic acid–alcohol at room temperature. The fixed specimens were dehydrated in a graded ethanol series [50%, 60%, 70%, 80%, 90%, 95%, 99.5% and 100% (v/v); 60 min per step] and stored overnight in 100% (v/v) ethanol at room temperature. The ethanol was replaced with 3-methylbutyl acetate, and the samples were dried in a critical-point dryer (JCPD-5; JEOL); they were then sputter-coated with gold–palladium using an anion sputter (JFC-1100; JEOL) and examined under an S-3400N scanning electron microscope (Hitachi), as described previously (Maeda et al., 2014; Gunji et al., 2020).

Basal parts of main stem were dissected and pre-fixed with 2% (w/v) paraformaldehyde and 2% (v/v) glutaraldehyde in 0.05 M cacodylate buffer (pH 7.4) overnight at 4°C. Stem portions were washed with 0.05 mM sodium cacodylate buffer, then they were post-fixed with 2% (w/v) osmium tetroxide in 0.05 mM cacodylate buffer for 3 h at 4°C. Samples were dehydrated in a graded ethanol series and stored in 100% (v/v) ethanol overnight at room temperature. After replacing with 100% propylene oxide, samples were infiltrated with Quetol-651 resin (Nisshin EM) and embedded. Ultrathin sections (80 nm) were obtained using a diamond knife on an ultramicrotome (Ultracut UCT; Leica Microsystems). Ultrathin sections were stained with a 2% (w/v) uranyl acetate and a lead stain solution (Sigma-Aldrich). Images were photographed under a transmission electron microscope (JEM-1400Plus; JEOL) at 100 kV equipped with a CCD camera (EM-14830RUBY2; JEOL). The thicknesses of cell walls were measured from transmission electron microscopy images using Fiji/ImageJ software. The thickness was averaged from three points within a cell.

Statistical analysis was performed using R software (R version 3.1.2; http://www.R-project.org/). For interfascicular fiber cell size, non-parametric tests (Wilcoxon rank sum test) were performed because the datasets did not exhibit a normal distribution, according to the Shapiro–Wilk test. The statistical significance of normally distributed data was determined using paired, two-tailed Student's t-test or one-way analysis of variance followed by post-hoc analysis with Tukey's honestly significant difference (HSD) test.

We thank A. Hosaka (AIST), F. Tobe (AIST), M. Yamada (AIST), Y. Sugimoto (AIST) and Y. Takiguchi (AIST) for preparing transgenic seed libraries.