PUCHI controls MYB94 and MYB96 to form the root cap cuticle of developing lateral roots

Abstract

Lateral root organogenesis underlies root branching and plays a key role in plant root system architecture. While the transcriptional and hormonal control of lateral root development has been extensively studied over the past decades, an increasing number of studies highlight the importance of specific metabolic pathways contributing to the lateral root regulatory network. The AP2/ERF transcription factor PUCHI has recently been shown to orchestrate the induction of very-long-chain fatty acid biosynthesis in developing lateral root primordia. In parallel, a root cap cuticle, reminiscent of the very-long-chain fatty acid–rich cuticle covering the emerging primary root, was shown to cover developing lateral root primordia and facilitate their emergence. Here, we demonstrate that PUCHI plays an essential role in lateral root cap cuticle formation. We identified the transcription factors MYB94 and MYB96 as key intermediates in this PUCHI-dependent, cuticle regulatory network. Similar to PUCHI, MYB94 and MYB96 induce the expression of a set of very-long-chain fatty acid- and cutin–related genes in lateral root primordia. The loss of function of either PUCHI or that of MYB94 and MYB96 compromises lateral root cuticle deposition, delays lateral root emergence, and alters the expression pattern of lateral root meristematic genes. These results shed light on a PUCHI- and MYB94 and MYB96-dependent regulatory network controlling the lateral root cap cuticle formation and influencing lateral root primordium development and emergence.

Introduction

The continuous and adaptive development of the root system architecture is crucial for plants to optimize the acquisition of resources in complex and heterogeneous soil environments (Jia et al. 2022). Lateral root (LR) formation plays a significant role in shaping the root system architecture of dicotyledonous plants, making it a subject of considerable importance for agricultural improvement. In the model plant Arabidopsis thaliana, LRs originate from a small group of xylem-pole pericycle cells inside the primary root upon local induction by the phytohormone auxin (Torres-Martínez et al. 2019). Primed pericycle cells undergo new rounds of cell division to produce a lateral root primordium (LRP). While growing through overlying cell layers, the developing LRP progressively organizes into a new root apical meristem, which ultimately emerges from the primary root to become a LR (Malamy and Benfey 1997; Torres-Martínez et al. 2019). Decades of research on Arabidopsis have revealed a core LR regulatory network, centered on auxin signal transduction and target transcription factors (Cavallari et al. 2021; Torres-Martínez et al. 2022; Zhang et al. 2023). More recently, several studies have shown that specific metabolic pathways and their products also contribute to the complex regulatory networks that control LR formation (Beckers et al. 2025). Unraveling these LR regulatory networks and understanding how metabolic pathways are involved is important for a deeper understanding of the development of the plant root system architecture.

PUCHI, which belongs to the AP2/ERF family of transcription factors, has been identified as a key regulator of LR initiation and development (Hirota et al. 2007; Kang et al. 2013). PUCHI contributes to the spacing of LR initiation (Toyokura et al. 2019), is required for the wild-type (WT) pattern of cell divisions in the developing primordium (Hirota et al. 2007; Kang et al. 2013), and influences the expression of several LR formation-related genes including the PLETHORA (PLT) transcription factors, which are involved in LR meristem formation (Du and Scheres 2017; Bellande et al. 2022). In addition, PUCHI was shown to influence LR development, at least in part, by regulating the biosynthesis of very-long-chain fatty acids (VLCFAs) (Trinh et al. 2019).

VLCFAs are fatty acids with an acyl chain of at least 20 carbon atoms and are components of various structural and functional metabolites in plant cells (Li-Beisson et al. 2013; Kosma et al. 2025). VLCFAs are synthesized by fatty acid elongase complexes through a 4-step mechanism: condensation by a β-ketoacyl-CoA synthase (KCS; 21 genes in Arabidopsis), reduction by a β-ketoacyl-CoA reductase (2 genes, KCR1 and KCR2, in Arabidopsis), dehydration catalyzed by the β-hydroxyacyl-CoA dehydratase HCD/PAS2, and reduction catalyzed by the enoyl-CoA reductase ECR/CER10 (Zhukov and Popov 2022). VLCFAs can be incorporated into membrane lipids such as phospholipids and sphingolipids; extracellular lipids such as suberin, waxes, and in some cases cutin (Berhin et al. 2019); and storage lipids such as triacylglycerols (Li-Beisson et al. 2013).

The cuticle is a multi-layered structure of lipid components that forms at the outermost surface of certain plant organs in continuity with the cell wall (Reynoud et al. 2021; González-Valenzuela et al. 2023). The cuticle is made up of a complex and reticulated cutin polyester matrix embedded in waxes. Loss or alteration of the cuticle increases the plant's susceptibility to environmental stresses such as drought or pathogens. Strikingly, it also causes numerous developmental defects, highlighting the dual roles of the cuticle in plant defense and developmental processes (Ingram and Nawrath 2017). Although the roles of cuticles covering aerial organs have been studied for decades, it was only recently discovered that cuticles also cover the emerging primary root and LR tips in Arabidopsis, forming an aliphatic layer called root cap cuticle (RCC) (Berhin et al. 2019). The RCC covering the emerging primary root was shown to display an atypical cutin composition, with a high proportion of VLCFAs compared with the cutin present in aerial organs (Berhin et al. 2019). In addition, while VLCFAs are important components of waxes covering aerial plant organs, the wax composition of the RCC remains elusive (Berhin et al. 2019). Interestingly, several RCC-defective mutants exhibited delayed LRP emergence, revealing the contribution of the RCC to this developmental process and its relevance for root branching (Berhin et al. 2019; Uemura et al. 2023). Key components involved in the formation of the RCC on LRs have been identified, including the enzymes DEFECTIVE IN CUTICULAR RIDGES (DCR) and BODYGUARD (BDG), which catalyze steps of the cutin polyester formation; and the transporters GPI-ANCHORED LIPID TRANSFER PROTEINS 1 and 2 (LTPG1 and LTPG2), which are required for the export of VLCFA components of cuticular waxes (Berhin et al. 2019; Uemura et al. 2023). Additionally, it was found that the expression of BDG and DCR, as well as the RCC formation, depends on the transcription factors ARABIDOPSIS THALIANA MERISTEM LAYER1 (ATML1) and PROTODERMAL FACTOR2 (PDF2) (Nagata et al. 2025). However, the regulatory network that mediates the formation of the RCC in developing LRs and links it to LR organogenesis still needs to be elucidated.

Here, we demonstrate that PUCHI, in addition to orchestrating the upregulation of VLCFA production, also induces the expression of several cutin-related genes and is necessary for the formation of the RCC in the developing LRs of Arabidopsis. Observed perturbations in RCC formation in the puchi loss-of-function mutant are comparable with those described in the cutin-defective bdg mutant. We identified MYB94 and MYB96 as target transcription factors of PUCHI, controlling LR-associated induction of VLCFA and cutin biosynthesis genes as well as RCC formation. The myb94 myb96 double mutant (hereinafter referred to as myb94/96) exhibited delayed LRP development and altered expression patterns of PLT genes. Thus, our findings indicate that PUCHI controls the RCC deposition on developing LRPs through the induction of its intermediary regulators MYB94 and MYB96, ultimately promoting LR development and emergence.

Results

PUCHI induces the expression of several cutin-related genes and the formation of the RCC in lateral roots

Previous efforts to understand how PUCHI participates in the regulation of LR development and emergence unraveled its role as a master regulator of VLCFA biosynthesis genes during LRP development (Trinh et al. 2019; Bellande et al. 2022). Interestingly, a gene ontology analysis of genes differentially expressed in the puchi background compared with the WT showed that genes associated with the biosynthesis of apoplast aliphatic products such as suberin, cutin, and waxes were overrepresented among the underexpressed genes in puchi (Trinh et al. 2019; Bellande et al. 2022). To further explore whether VLCFA incorporation pathways were influenced by PUCHI during root branching, we analyzed gene expression in LR development transcriptomic time series (Bellande et al. 2022), specifically focusing on genes associated to lipid metabolism in 6 different categories: phospholipids, sphingolipids, triacylglycerol, suberin, cutin, and waxes (Li-Beisson et al. 2013). These genes were selected from the literature and code for enzymes, transporters, or regulators that participate in the formation of these different lipid structures. Then, we calculated the proportion of genes in each lipid class that exhibited differential expression (*P < 0.05; see Materials and Methods) for at least 1 time point in puchi compared with the WT, independently of their expression kinetics during LRP development (Fig. 1a). While the expression of only 4.4% of the genes was affected by the puchi mutation in the root branching transcriptomic dataset (Bellande et al. 2022), 18% of the selected 194 genes were found to be dependent on PUCHI during LRP development (Fig. 1a). The proportion of PUCHI-dependent genes was moderate in the gene classes related to membrane lipids, with 13% of phospholipid-associated genes and 8% of sphingolipid-associated genes displaying perturbed expression kinetics in puchi. In contrast, this proportion was high when focusing on genes involved in cutin, suberin, and waxes, with 52% of genes involved in cutin metabolism displaying altered expression dynamics in the puchi mutant background (13 genes out of 25) (Fig. 1a and Fig. S1). These observations suggest that PUCHI is an important regulator of cutin gene expression during LRP development.

Figure 1.

The expression of several cutin-related genes and the formation of the RCC depend on PUCHI during LR development. a) Percentage of the genes known to be involved in different lipid biosynthetic pathways that are differentially expressed in puchi LRP compared with WT. Genes differentially expressed (P < 0.05) for at least 1 time point in the RNAseq data from Bellande et al (2022) were selected. For each lipid class, n depicts the number of genes found to be differentially expressed in puchi LRP, relative to the total number of genes studied. b) Expression profile of cutin-related genes involved in RCC formation during LRP development in WT and puchi. Data are represented as the mean number of reads ± standard error of the mean (SEM) of 3 replicates per time point from the transcriptomic dataset of Bellande et al. (2022). Statistical analysis on RNAseq replicates was performed using DESeq2 package and Wald Test: *P < 0.05; **P < 0.02; ***P < 0.01. c) Fluorol yellow 088 (FY) staining of the cuticle of emerged LRs of WT, bdg and puchi, as well as transmission electron microscopy (TEM) pictures showing the cell wall with a cuticle on its surface (black arrowhead indicating the expected position of the cuticle). Numbers indicate the occurrence of the represented pattern over the total number of observations. Scale bars: 50 µm in pictures showing FY staining, 500 nm in TEM pictures. d) Penetration of the fluorescent cellular tracer fluorescein diacetate (FDA) into lateral root cells of WT, bdg, and puchi after 4 min of incubation. Relative intensity of the fluorescence is depicted by the color code. Numbers indicate the occurrence of the represented pattern over the total number of observations. Scale bars: 50 µm.

Interestingly, DCR, LTPG1, LTPG2, and BDG, 4 genes that are known to play a key role in RCC formation on LRs (Berhin et al. 2019; Uemura et al. 2023), displayed a consistent induction of expression during WT LR formation, and that induction was lost in the puchi mutant (Fig. 1b). The induction of additional genes related to cutin synthesis, such as the genes coding for the enzyme HOTHEAD (HTH), 2 cytochrome P450 oxidases (CYP86A2 and CYP86A8), and the transporters of cutin monomers across the plasma membrane ATP-BINDING CASSETTE 11 and 13 (ABCG11 and ABCG13), or genes involved in its regulation, like ATML1 and PDF2, are also dependent on PUCHI (Fig. S2). Therefore, PUCHI activity is essential for inducing the expression of a large group of cutin metabolism-related genes, including key genes for RCC establishment during LRP formation.

We then investigated whether the perturbations in expression of cutin-related genes during LR development in puchi resulted in altered RCC formation. To test this, we compared cuticle depositions on newly emerged LRs in puchi and WT and in bdg, which cannot form a proper RCC due to its inability to polymerize cutin precursors (Berhin et al. 2019), using the lipid-staining dye Fluorol Yellow (FY) (Fig. 1c). The RCC was observed at the surface of emerging WT LRs but was absent in bdg or puchi LRs (Fig. 1c). In addition, analysis by transmission electron microscopy indicated that the recently emerged LRs in puchi and bdg mutants display a permeable RCC compared with the electron-opaque RCC layer observed in WT (Fig. 1c). Thus, PUCHI activity is required for the formation of a WT cuticle over the emerging LRs of Arabidopsis.

The RCC of the LR was previously shown to provide barrier functions limiting the diffusion of molecules (Berhin et al. 2019). To determine whether the altered structure of RCCs on puchi LRs is associated with increased LR surface permeability, we performed a fluorescein diacetate (FDA) diffusion assay as described by (Barberon et al. 2016; Berhin et al. 2019) (Fig. 1d). As expected, the WT LRs barely showed any fluorescence after a 4-min incubation period in FDA, while bdg showed a clear fluorescence signal associated with FDA penetration from the medium into the LR cells (Fig. 1d) (Berhin et al. 2019). In the same conditions puchi LRs also displayed FDA fluorescence, demonstrating diminished diffusion barrier properties of their LR surface (Fig. 1d).

Altogether, our results demonstrate that PUCHI is required for proper expression of several cutin-related genes during LR development, including key genes involved in RCC formation. As a consequence, PUCHI is required for the formation of a hydrophobic RCC on the LR.

Loss of the RCC partially phenocopies LRP development alterations observed in the puchi mutant background

It has previously been reported that mutants with RCC formation defects exhibited delayed LR development (Berhin et al. 2019; Uemura et al. 2023). Interestingly, 9-d-old bdg and puchi seedlings displayed root branching reduction compared with WT (Fig. 2a). To analyze more precisely the LR phenotypes of puchi and bdg, we scored the number of emerged LRs and non-emerged LRPs along the primary root of 9-d-old plants (Fig. 2b, c). In contrast to puchi, the primary root length and the density of total root branching events (LR + LRP) were unaffected in the bdg mutant (Fig. S3). However, emerged LR density was significantly reduced in bdg and in puchi compared with WT (Fig. 2b) (Berhin et al. 2019; Trinh et al. 2019). In contrast, the density of non-emerged LRPs was significantly higher in bdg and puchi compared with WT (Fig. 2c). In both instances, the bdg phenotype was significantly less severe than the puchi phenotype. Hence, the loss of function of BDG results in comparable, although less severe perturbations in LR development and reduction in LR emergence compared with the loss of PUCHI function.

Figure 2.

The lack of RCC results in perturbations of LR development. a) WT, bdg, and puchi plants 9 d after germination. Scale bars: 5 mm. b, c) Density of emerged LR (b) and density of non-emerged LRP (c) of 9-day-old WT, bdg, and puchi plants. A minimum number of 30 seedlings were analyzed for each genotype. Data are represented as mean ± SEM. Significant differences to WT were determined by Student t test: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not significant. d) Distribution of developmental stages as described by Malamy and Benfey (1997), after gravistimulation-induced LRP formation in WT, bdg, and puchi roots at 18 and 48 h after gravistimulation (hpg). Data are represented as mean ± SEM for a minimum number of 30 seedlings for each genotype. Significant differences to WT were determined by Student t test: *P < 0.05; **P < 0.01; ***P < 0.001. e) Confocal images showing the expression patterns of promPLT5:PLT5-YFP, promPLT7:PLT7-YFP, promPLT1:PLT1-YFP, promPLT2:PLT2-YFP, and promPLT4:PLT4-YFP (green), respectively, in WT, puchi, and bdg LRPs of stages IV or V of development. Numbers indicate the occurrence of the represented pattern over the total number of observations. Cell walls were stained using propidium iodide (magenta). Scale bar: 100 μm.

Compliance with permission from the rights holder to display this image online prohibits further enlargement or copying.

To gain a clearer understanding of the impact of the absence of the RCC on LRP development and compare this with the puchi phenotype, a gravistimulation-based LR induction system (Lucas et al. 2008) was used to analyze the kinetics of LRP development in WT, puchi, and bdg plants. At 18 h after gravistimulation, no significant difference in LRP stage distribution was observed between WT, bdg, and puchi, suggesting that the bdg mutation does not affect early LRP development (Fig. 2d). At 48 h after gravistimulation, the majority of WT LRPs reached either stage VII of development or have emerged from the primary root (Fig. 2d; developmental stages defined according to Malamy and Benfey 1997). By contrast, most of the bdg LRPs have only reached stages IV, V, or VI, with only 10% of emerged LRs (Fig. 2d). This delay in LRP development in bdg is significant but weaker compared with that observed in puchi, which mostly displays LRPs at stages II, III, and IV and no emerged LRs at 48 h after gravistimulation (Fig. 2d). Therefore, considering the bdg mutant as a proxy to assess the relevance of the RCC for LRP development (Berhin et al. 2019), our analyses confirmed that the RCC is necessary for the late stages of LRP developmental progression up to emergence, but its absence results in less severe delays in LRP development than those observed with the puchi mutation.

During WT LR development, a sequential induction of different PLETHORA transcription factors, namely PLT5 and PLT7 during early stages and then PLT1 and PLT4 in later stages, provides a remarkable template to monitor the progression toward LRP meristem functional organization (Du and Scheres 2017; Bellande et al. 2022). Perturbations in LRP development in puchi was associated to altered PLT expression (Bellande et al. 2022). To test whether these observed PLT perturbations in puchi could be explained by its altered RCC, we investigated PLT expression patterns in developing bdg LRPs. As the RCC forms between stages III and IV of development (Ursache et al. 2021), we examined the expression of these reporters in LRPs at stages IV and V (ie just after RCC formation) (Fig. 2e). Consistently with what has been reported previously (Du and Scheres 2017; Bellande et al. 2022), PLT5-YFP, PLT7-YFP, PLT2-YFP, PLT1-YFP, and PLT4-YFP expression was detected in various and partially overlapping domains in WT LRPs at these stages. In contrast, the expression of PLT5-YFP, PLT7-YFP, and PLT2-YFP constructs was barely detectable in puchi compared with WT, while PLT1-YFP and PLT4-YFP were expressed at higher levels (Fig. 2e) (Du and Scheres 2017; Bellande et al. 2022). Similar to puchi, a very weak expression of PLT5-YFP, PLT7-YFP, and PLT2-YFP is observed in bdg LRPs. However, in contrast to puchi, the bdg mutation resulted in lower levels of PLT1-YFP, and PLT4-YFP expression compared with WT (Fig. 2e). These perturbations in PLT gene expression in the bdg background were confirmed by RT-qPCR analysis (primers used are listed in Table S1) following LR induction by 1-naphthalene acetic acid (NAA) (Fig. S4).

Hence, the RCC is necessary for LRP development and emergence, but its absence results in less severe delays than those observed with the puchi mutation. Interestingly, loss of function of either PUCHI or BDG (our proxy for the loss of LRP RCC formation) reduces PLT5, PLT7, and PLT2 expression in stage IV/V LRPs. However, the 2 mutations result in an opposite phenotype regarding the expression pattern of PLT1 and PLT4. This suggests that the absence of RCC alone may contribute to but is not sufficient to explain the perturbations of LR development kinetics and PLT gene expression dynamics in puchi.

Identification of candidate transcription factors regulating the expression of the VLCFA biosynthesis genes downstream of PUCHI during LRP development

We previously showed that while PUCHI regulates a set of genes of the VLCFA elongation pathway, these genes were not primary targets of PUCHI (Trinh et al. 2019). To identify genes acting downstream of PUCHI in the gene regulatory network regulating VLCFA biosynthesis during LR development, we mined published databases. We cross-referenced 3 lists of candidate genes that might participate in the regulation of VLCFA biosynthesis in developing roots by PUCHI (Fig. 3a). The first list (“DEG RNAseq puchi vs WT”) contains the PUCHI-dependent genes identified by RNAseq profiling of branching root segments (Bellande et al. 2022). The second candidate gene list was generated using a custom-designed algorithm named REGINE (Resampling-based Engine for Gene-set Inference and Network Enrichment). REGINE systematically interrogates DAP-seq interaction networks (O’Malley et al. 2016) to identify transcription factors exhibiting statistically significant co-binding enrichment across a given set of target genes. By iteratively sampling gene subsets and comparing observed shared binding frequencies to those obtained from randomized background sets, REGINE ranks transcription factors based on signal-to-noise enrichment, thereby highlighting candidate regulators with coordinated regulatory potential (see Materials and Methods). Here we used PUCHI-dependent genes associated with VLCFA, cutin, and wax biosynthesis as target genes for this analysis. The third list (“Regulators of VLCFA”) is a list of Arabidopsis transcription factors that were shown to influence the expression of VLCFA biosynthesis genes in the literature. These 3 lists can be found in Table S2. Comparison of those lists yielded 1 single gene coding for the transcription factor MYB96 (Fig. 3a). The transcription factor MYB94 is closely related to MYB96 according to its phylogeny (Abdullah-Zawawi et al. 2021). Moreover, MYB94 and MYB96 exert an additive effect on the expression of VLCFA biosynthesis genes in the aerial parts of plants in response to drought (Lee et al. 2016). Interestingly, MYB96 was also shown to be expressed in developing LRP and LR meristems, and an overexpressing MYB96-ox allele displayed reduced LR density and LR growth (Seo et al. 2009). Although activity of a MYB94 4.2-Kb or 3-Kb promoter—GUS fusion could not be clearly detected in roots of young Arabidopsis seedlings (Lee and Suh 2015a; Dai et al. 2020), low expression of MYB94 was repeatedly detected in independent root transcriptomics datasets associated with lateral root formation (Brady et al. 2007; Voß et al. 2015; Bellande et al. 2022), as well as in several cell populations of the primary root tip, including the pericycle (Denyer et al. 2019). Furthermore, MYB94 and MYB96 were shown to repress root pericycle-derived callus induction (Dai et al. 2020), a phenotype reminiscent both of puchi and root VLCFA-depleted mutants (Shang et al. 2016; Trinh et al. 2019). The expression dynamics of MYB94 and MYB96 in the time course transcriptomic dataset of WT and puchi branching root segments (Bellande et al. 2022) showed that the 2 genes are induced during LRP development in WT and that this induction is compromised in the puchi mutant (Fig. 3b). Conversely, PUCHI expression is unaffected in a loss-of-function myb94/96 background (Dai et al. 2020) (Fig. S5). Thus, we conclude that PUCHI induces the expression of MYB94 and MYB96 during LRP development, and these 2 MYBs are plausible intermediate transcription factors mediating PUCHI control over VLCFA biosynthesis during LR formation.

Figure 3.

The expression of MYB94 and MYB96 is induced by PUCHI during LRP development. a) Venn diagram showing the comparison of 3 gene databases in relation to PUCHI, LR development, and VLCFA biosynthesis to identify candidates for PUCHI effectors. The first database (“DEG RNAseq puchi vs WT”) contains the PUCHI-dependent genes identified by RNAseq profiling of branching root segments (Bellande et al. 2022). The second named “REGINE” is an algorithm that aims to detect, from DAP-seq-type datasets (O'Malley et al. 2016), transcription factors whose probability of interacting with a set of target genes (here the PUCHI-dependent VLCFA and cutin genes) is significantly higher than that expected by chance. The third that we named “Regulators of VLCFA” is a list of genes extracted from the literature that are known as regulators of VLCFA biosynthesis genes. b) MYB94 and MYB96 expression in WT and puchi background during the formation of LRP. Data are represented as the mean number of reads ± SEM of 3 replicates per time point from the transcriptomic dataset of Bellande et al. (2022). Statistical analysis on RNAseq replicates was performed using DESeq2 package and Wald Test: *P < 0.05; **P < 0.02; ***P < 0.01.

MYB94 and MYB96 promote the expression of VLCFA- and cutin-related genes and the formation of the root cap cuticle on developing lateral roots

MYB94 and MYB96 were reported as regulators of wax deposition in aerial organs where they were shown to directly induce the expression of VLCFA biosynthesis genes, including KCS1, KCR1, and ECR (Lee and Suh 2015b, 2022; Lee et al. 2016). To test the hypothesis that MYB94 and MYB96 are intermediate regulators of VLCFA biosynthesis genes downstream of PUCHI during LRP development, we analyzed whether the PUCHI-dependent expression dynamics of VLCFA-elongation genes were also dependent on MYB94 and MYB96 during LR development. We used RT-qPCR to compare the expression levels of VLCFA biosynthesis genes KCS1, KCS20, KCR1, PAS2, and ECR/CER10 in the roots of WT, puchi, myb94, myb96, and the double null mutant myb94/96 during LR formation (Fig. 4a). LR formation is inhibited in control plants treated with N-1-naphthylphthalamic acid (NPA), while LR initiation is induced along the whole primary root on medium supplemented with NAA (modified from Himanen et al. 2004; see Materials and Methods). RT-qPCR experiments revealed that 18 h after LR induction by NAA treatment, transcript levels of KCS1, KCS20, KCR1, PAS2, and ECR/CER10 increased in a similar fashion in WT and single mutants myb94 and myb96 (Fig. 4a). However, this induction of expression was disrupted in the puchi and myb94/96 double mutants (Fig. 4a), indicating that MYB94 and MYB96 are acting redundantly to regulate VLCFA biosynthesis gene expression during LR development.

Figure 4.

The expression of several VLCFA- and cutin-related genes, as well as the formation of the RCC, depend on the activity of MYB94 and MYB96. a, b) VLCFA- (a) and cutin-related (b) gene expression by RT-qPCR in WT, myb94, myb96, myb94 myb96 (myb94/96) and puchi roots. LR formation is inhibited in control plants treated with N-1-naphthylphthalamic acid (NPA), while on 1-naphthalene acetic acid (NAA) LR initiation is induced synchronously along the primary root. Roots without the root apical meristem were harvested after 18 h (for VLCFA biosynthesis genes) and 24 h (for cutin biosynthesis genes) treatment with 10 µM NPA or 10 µM NAA. Normalization was achieved with CYCLIN-DEPENDENT KINASE A;1 (CDKA;1) and TUBULIN BETA CHAIN 3 (TUB3) genes. The calibrator cDNA is WT under NPA treatment. Data are represented as mean ± SEM of 3 biological replicates. Significant differences to WT were determined by Student t test: *P < 0.05; **P < 0.01; ***P < 0.001; ns: not significant. c) Fluorol yellow 088 (FY) staining of the cuticle of WT, myb94/96 and puchi emerged LR. Numbers indicate the occurrence of the represented pattern over the total number of observations. Scale bars: 50 µm. d) Penetration of the fluorescent cellular tracer fluorescein diacetate (FDA) into lateral root cells of WT, myb94/96 and puchi after 4 min of incubation. Relative intensity of fluorescence is depicted by the color code. Numbers indicate the occurrence of the represented pattern over the total number of observations. Scale bars: 50 µm.

Next, we analyzed whether the expression of cutin-related genes involved in RCC formation was also dependent on MYB94 and MYB96. RT-qPCR was performed 24 h after NPA or NAA treatment (ie just after RCC formation) to quantify the expression of DCR, LTPG1, LTPG2, and BDG, 4 key genes involved in RCC formation whose expression is induced in a PUCHI-dependent manner (Bellande et al. 2022). The expression of these 4 genes was increased following LR-inducing NAA treatment in WT and in the 2 single myb94 and myb96 mutants, whereas no induction of their expression was detected in the puchi and myb94/96 mutants (Fig. 4b). Hence, MYB94 and MYB96 redundantly control the induction of genes involved in VLCFA and cutin metabolism in the same way as PUCHI during the auxin-induced root branching process.

We then wanted to know whether the disruption of VLCFA- and cutin-related gene expression in myb94/96 would result in similar perturbations of RCC formation as observed in puchi. FY staining of developing lateral roots as well as FDA diffusion assays in these backgrounds revealed that myb94/96 mutant LRPs lack functional RCCs, thus phenocopying puchi's phenotype (Fig. 4c, d).

Altogether, these results revealed that MYB94 and MYB96, whose expressions are induced downstream of PUCHI during LRP development, act redundantly to stimulate the expression of several VLCFA- and cutin-related genes in LRPs and are required for the formation of a functional RCC covering the LRP. Thus, MYB94 and MYB96 are plausible downstream targets of PUCHI in the LR regulatory network, mediating its influence on VLCFA production and RCC formation and, possibly, on LRP development.

MYB94 and MYB96 redundantly promote LR development progression

To determine whether MYB94 and MYB96 are also involved in LR development, we analyzed the LR phenotype of their knockout lines. The root system architecture of 9-d-old myb94 and myb96 single mutant seedlings was similar to that of the WT, whereas the myb94/96 double mutant displayed a reduced root branching density, comparable, although less severe, to that observed in the puchi background (Fig. 5a–c). While the primary root length and the total LR initiation density were unaffected in either the single or double mutants (Fig. S6), the density of emerged LRs in myb94/96 was significantly reduced compared with WT, and the density of non-emerged LRPs was significantly higher in myb94/96 compared with WT (Fig. 5b, c). In addition, the single mutation puchi was found to be epistatic to myb94/96 in relation with LRP and LR density, as the triple mutant puchi myb94 myb96 (puchi/myb94/96) displayed a statistically similar density of emerged LRs and non-emerged LRPs as the puchi single mutant (Fig. S7a–c). These results are consistent with MYB94 and MYB96 being intermediate factors mediating PUCHI function in LRP development and emergence.

Figure 5.

The roots of myb94 myb96 double mutants exhibit reduced LRP development and PLETHORA gene expression. a) WT, myb94, myb96, myb94 myb96 (myb94/96), and puchi plants 9 d after germination. Scale bars: 5 mm. b, c) Density of emerged LR (b), and density of non-emerged LRP (c) of 9-d-old WT, myb94, myb96, myb94/96, and puchi. Data are represented as mean ± SEM for a minimum number of 30 seedlings for each genotype. Significant differences to WT were determined by Student t test: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns: not significant. d) Distribution of developmental stages as described by Malamy and Benfey (1997), achieved by gravistimulation-induced LRP formation in WT, myb94, myb96, myb94/96, and puchi roots at 18 and 48 h after gravistimulation (hpg). Data are represented as mean ± SEM for a minimum number of 30 seedlings for each genotype. Significant differences to WT were determined by Student t test: *P < 0.05; **P < 0.01; ***P < 0.001. e) PLETHORA gene expression measured by RT-qPCR in WT, myb94/96 and puchi roots. LR formation is inhibited in control plants treated with N-1-naphthylphthalamic acid (NPA), while on 1-naphthalene acetic acid (NAA) LR initiation is induced synchronously along the primary root. Roots without the root apical meristem were harvested after 24 h treatment with 10 µM NPA or 10 µM NAA. Normalization was achieved with CYCLIN-DEPENDENT KINASE A;1 (CDKA;1) and TUBULIN BETA CHAIN 3 (TUB3) genes. The calibrator cDNA is WT under NPA treatment. Data are represented as mean ± SEM of 3 biological replicates. Significant differences to WT were determined by Student t test: *P < 0.05; **P < 0.01; ***P < 0.001.

Compliance with permission from the rights holder to display this image online prohibits further enlargement or copying.

To analyze the impact of myb94 and myb96 mutations on the LRP developmental kinetics, we induced LRP formation in WT, myb94, myb96, myb94/96, and puchi seedlings using gravistimulation (Lucas et al. 2008) and scored the developmental stages that the LRP had reached at 18 and 48 h after induction (Fig. 5d). After 18 h, no significant differences were observed between the different genetic backgrounds, suggesting that the myb94 and myb96 mutations do not affect early LRP development. After 48 h of development the distribution of the observed LRPs stages was similar in the WT, myb94, and myb96 single mutant backgrounds, with most of their LRPs being at stage VII of development or having emerged from the primary root. By contrast, myb94/96 LRP developmental stages were delayed compared with WT, with the majority of myb94/96 LRPs only reaching stages III, IV, and V of development. Hence, the loss of MYB94 and MYB96 functions impairs LRP developmental progression, although to a less severe extent compared with puchi. Again, the analysis of the triple mutant puchi/myb94/96 showed that the puchi mutation was epistatic to myb94/96 (Fig. S7d). These observations are consistent with PUCHI, MYB94, and MYB96 being involved in a shared regulatory network that controls LR development, with MYB94 and MYB96 acting downstream of PUCHI.

To assess the influence of MYB94 and MYB96 on LRP functional organization, we tested whether the expression of PLT transcription factors was perturbed in the myb94/96 LRs, as observed in the puchi mutant. As expected, RT-qPCR quantification at 24 h after NAA-mediated LR induction showed that the expression of PLT5, PLT7, and PLT2 genes was stimulated in the WT, and this expression was lost in puchi (Fig. 5e). Interestingly, their induction was also severely compromised in the myb94/96 background. Conversely, while PLT1 and PLT4 were expressed at higher levels in puchi than in WT during LRP development (Bellande et al. 2022), their expression was reduced in the myb94/96 background compared with WT (Fig. 5d). Thus, MYB94 and MYB96 are necessary to observe the induction of several PLT genes in developing LRPs. However, the loss of MYB94 and MYB96 function, or that of BDG, does not fully recapitulate the puchi phenotype in terms of PLT expression dynamics.

Discussion

Here, we further explored the function of the transcription factor PUCHI and its downstream regulatory network during LR development. We showed that PUCHI was necessary for RCC deposition over the developing LRPs and that this could at least partly explain the delay in LRP development in the puchi mutant background. In addition, our results show that MYB94 and MYB96 are induced in a PUCHI-dependent-manner during root branching (Fig. 3b), are redundantly required for the induction of genes involved in VLCFA and cutin biosynthesis during LR formation as well as for WT RCC deposition on the developing LRP (Fig. 4), and promote LRP development and emergence (Fig. 5). Interestingly, out of the 2 MYB genes, either the functional MYB94 or MYB96 gene is sufficient to observe the WT dynamics in VLCFA gene expression and LRP development. The fact that the LR phenotype of the puchi/myb94/96 triple mutant is similar to that of puchi (Fig. S7) further supports a model in which MYB94 and MYB96 (hereinafter referred to as MYB94/MYB96) act as redundant intermediates, downstream of PUCHI, to influence RCC deposition and favor LRP developmental progression (Fig. 6). These results are consistent with the previously described roles of MYB94 and MYB96 in the promotion of VLCFA biosynthesis and cuticular wax deposition in shoot organs (Lee et al. 2016; Lee and Suh 2022), although the specific contribution of wax components to the LR RCC remains to be explored (Berhin et al. 2019). Experimental data support a direct binding of MYB94 and MYB96 to the promoters of VLCFA biosynthesis genes such as KCS1 and KCR1 (Lee et al. 2016), or KCS20 and PAS2 (Bartlett et al. 2017). This suggests that MYB94 and MYB96 could be direct regulators of VLCFA biosynthesis genes in the VLCFA and cuticle regulation network controlled by PUCHI.

Figure 6.

Proposed model of the PUCHI-dependent, LR regulatory network controlling LR RCC formation. Schematic representation of a current working model of the PUCHI-dependent, LR regulatory network. In this model, PUCHI induces the expression of MYB94 and MYB96, which then induce genes involved in VLCFA and cutin biosynthesis, resulting in RCC formation on developing LRs. This RCC formed under the control of PUCHI, MYB94, and MYB96, participates in the progression of LR development and the regulation of PLT meristematic gene expression. Finally, PUCHI inhibits the expression of some PLT genes independently of the regulatory pathway involved in RCC formation. The transcription factors are represented by the grey circles. The straight lines represent direct regulation, and the dotted lines represent possibly indirect regulation.

Interestingly, the VLCFA molecules whose synthesis is modulated by MYB94/MYB96 and PUCHI could be relevant for various structural or functional traits in roots. Indeed, VLCFAs can incorporate into various classes of molecules impacting highly diverse cellular and tissular processes such as cell membrane and cell division dynamics (Molino et al. 2014), membrane functional organization (Babić et al. 2025), lipid signaling (Nagata et al. 2021), or the deposition of apoplastic barriers (Grünhofer and Schreiber 2023), with multiple impacts on the mechanisms of plant development. Here, we show that PUCHI and MYB94/MYB96 orchestrate not only the induction of VLCFA biosynthesis genes but also that of cutin-associated factors and the deposition of the RCC on the developing LRP, impacting its development.

Plant cuticles appeared during land colonization and are key to prevent water and nutrient loss from the aerial tissues, to protect the plant against biotic and abiotic stresses, as well as to reduce communication or adhesion between organs or tissues developing in close contact (González-Valenzuela et al. 2023). Although cuticles were identified on the aerial parts of plants in the early 19th century, their presence on the roots was only described in 2019 in Arabidopsis (Berhin et al. 2019). Alterations in the RCC covering the LRP were shown to hinder their development and emergence, possibly because of increased adhesion between the developing primordium and the overlying root tissues (Berhin et al. 2019; Uemura et al. 2023). Similarly, defects in cuticle deposition in developing shoot organs were shown to cause organ adhesions resulting in organ deformations (eg during the rapid outgrowth of petals from the floral bud) (Ingram and Nawrath 2017). Alternatively, the RCC may influence the movement of water or solutes between cell layers or impact mechanical or chemical communication between the developing LRP and its overlying tissues (Ingram and Nawrath 2017; Doll and Ingram 2022). Interestingly, in developing Arabidopsis seeds, such a specialized cuticle layer was shown to modulate the movement of signaling peptides, thereby influencing embryo and endosperm communication and coordination of development (Moussu et al. 2013; San-Bento et al. 2014). Developing LRPs and their overlying cells were shown to form distinct symplastic domains as early as stage IV, which is concomitant with the formation of the RCC (Benitez-Alfonso et al. 2013; Sager et al. 2020). Hence, the diffusion rate of molecules through the apoplasm layer separating the developing LRP and the overlying parent root tissues is key for the communication between these tissues.

Interestingly, although PUCHI, MYB96, and possibly MYB94 are predominantly, if not exclusively, expressed in LRPs and root apical meristems in the Arabidopsis root system (Hirota et al. 2007; Seo et al. 2009; Lee and Suh 2015a), these 3 transcription factors are also expressed in developing shoot organs. During the transition from vegetative to reproductive development, PUCHI is expressed in the adaxial domain of the lateral organs produced by the inflorescence meristem, where it promotes flower meristem establishment and represses bract outgrowth (Karim et al. 2009). On the other hand, MYB94 and MYB96 were shown to promote cuticular wax deposition on aerial plant organs in response to abscisic acid-mediated drought stress signaling (Seo et al. 2011). Whether PUCHI may regulate MYB94/MYB96-mediated regulation of VLCFA biosynthesis in developing shoot organs remains to be explored.

The comparison of LRP development phenotype of puchi, myb94/96, and bdg loss-of-function mutants also suggests complex relationships in the LR regulatory network, and because the bdg mutation does not fully recapitulate the puchi LR phenotype, only some of the multiple roles of PUCHI in LRP initiation, development, and emergence are dependent on the RCC (Fig. 6). First, while the bdg RCC mutant displays changes in LRP shape suggesting perturbed mechanical relationships between the LRPs and the overlying tissues, the puchi mutant LRPs are consistently flattened due to ectopic cell divisions in their flanking domains (Hirota et al. 2007; Berhin et al. 2019). Second, the bdg mutant is unaffected in terms of LR initiation, while PUCHI promotes the spacing of LRP initiation sites (the mutant displays increased LR initiation density) through a mobile peptide signaling pathway (Toyokura et al. 2019). Last, the loss of PUCHI function results in enhanced PLT1 and PLT4 expression in developing LRPs, while on the contrary the expression of these root meristem-associated PLT genes is reduced in the bdg and myb94/96 background. Hence, while this regulatory module depends on PUCHI, it does not fully explain PUCHI's role in LR formation. Further studies are therefore needed to decipher the other facets of this complex PUCHI-driven regulatory network controlling LR organogenesis.

In conclusion, this work contributed to elucidating the mechanisms by which PUCHI influences LR formation and has shed light on the pivotal role that this transcription factor plays in the production of the LRP RCC. Two related MYB transcription factors, MYB94 and MYB96, were identified as effectors of PUCHI's influence over VLCFA- and cutin-related genes, controlling LR RCC deposition and thereby promoting LRP developmental progression. These findings also provide deeper insights into the regulatory mechanisms of cuticle formation in LRs and confirm their importance for plant root system development.

Materials and methods

Plant materials and growth conditions

Arabidopsis thaliana ecotype Col-0 plants were used for all experiments and as the background of transgenic lines in this study. For all the experiments presented here, Arabidopsis seeds were surface-sterilized and sown on half-strength Murashige and Skoog (½ strength MS) solid medium with 0.7% plant agar supplemented with B5 vitamins. Plates were kept at 4 °C for 2 d and then placed in long-day conditions (16-h-light/8-h-dark cycle) in a vertical position. The puchi mutant line has previously been described (Hirota et al. 2007). The bdg (SAIL_1286_E08) mutant was obtained from the Nottingham Arabidopsis Stock Centre (www.arabidopsis.info/). The myb94, myb96, and myb94 myb96 (myb94/96) lines have been previously described (Dai et al. 2020). The promPLT:PLT-YFP lines were reported in (Galinha et al. 2007; Hofhuis et al. 2013).

Analysis of LR transcriptomic time series

A list of genes known to be involved in 6 different classes of lipids was compiled from the literature. This list contains 194 genes belonging to the metabolism of phospholipids (62 genes), sphingolipids (35 genes), cutin (25 genes), waxes (22 genes), suberin (31 genes), and triacylglycerols (19 genes).

Among these, differentially expressed genes in the Bellande et al. 2022 dataset were identified using the Bioconductor (Gentleman et al. 2004) package (http://www.bioconductor.org) DESeq2 1.26.0 (Love et al. 2014) (R version 3.6.1). Data were normalized using the DESeq2 normalization method. Genes with an adjusted P < 5% (according to the false discovery rate method of Benjamini and Hochberg) for at least 1 time point were declared differentially expressed.

Database mining for PUCHI target genes involved in the regulation of VLCFA-related genes

To identify candidate genes involved in the PUCHI-dependent regulatory network controlling VLCFA biosynthesis gene expression, we compared 3 lists of genes extracted from already published experimental datasets.

The “DEG RNAseq puchi vs WT” gene list contains all the PUCHI-dependent genes identified by RNAseq profiling of branching root segments (Bellande et al. 2022). Genes with an adjusted P < 5% (according to the false discovery rate method of Benjamini and Hochberg) for at least 1 time point were declared differentially expressed. To identify candidate regulators of, specifically, PUCHI-dependent VLCFA and cutin-related genes, we took advantage of the DAPseq cistrome libraries published for a set of 349 transcription factors (O’Malley et al. 2016) using the REGINE algorithm fully described below. Last, a third list of genes of interest named “Regulators of VLCFA” was extracted from the literature based on their reported role in regulating VLCFA biosynthesis genes. These 3 lists can be found in Table S2. List contents were compared using the Venn diagram creator jvenn from INRAE Toulouse (https://jvenn.toulouse.inrae.fr/app/index.html) (Bardou et al. 2014).

Description of REGINE algorithm

REGINE is a resampling-based statistical framework developed to identify transcription factors (TFs) whose binding profiles are significantly enriched across a predefined gene set. The algorithm operates on a genome-scale TF→ target interaction network derived from DAP-seq datasets, represented as a bipartite graph G = (T,G,E), where T denotes transcription factors, G genes, and E experimentally supported binding interactions (O’Malley et al. 2016). For a user-defined gene set S⊂G, REGINE evaluates the enrichment of shared TF regulators by iteratively sampling subsets (“kernels”) of size k from S. For each sampled subset Sk, the algorithm computes the intersection of TFs binding all genes in SkS, thereby quantifying higher-order co-binding events. To estimate the null distribution, equivalent-sized subsets are sampled from the global gene universe represented in the DAP-seq network, and the same intersection procedure is applied. This resampling procedure is repeated n times to generate empirical distributions of shared TF counts under both signal (gene set) and background conditions.

For each TF, REGINE computes 2 frequencies: (1) the number of iterations in which the TF appears as a shared regulator within sampled subsets of S (Real score); and (2) the corresponding frequency in background subsets (Noise score). A signal-to-noise enrichment statistic is then defined as Δ=Real−Noise, which serves as the primary ranking criterion. This framework inherently controls for network density and TF connectivity biases by explicitly incorporating background sampling. The kernel size parameter k is optimized empirically by evaluating the separation between signal and background distributions across a range of values. REGINE therefore identifies TFs exhibiting coordinated binding enrichment across subsets of the gene set beyond random expectation, prioritizing candidate regulators with statistically supported co-regulatory potential.

RT-qPCR analysis

For RT-qPCR analyses, plant material was collected following the lateral root induction system described in (Himanen et al. 2004). Briefly, seeds were grown on ½ strength MS supplemented with 10 μM naphthylphthalamic acid (NPA) for 12 d to inhibit LR, then the seedlings were transferred to ½ strength MS supplemented with 10 μM NPA or 10 μM naphthaleneacetic acid (NAA) to synchronously induce LR formation. Three biological replicates of root segments without the root apical meristem were harvested after 24 h of treatment and immediately frozen in liquid nitrogen for each condition. Total RNA was extracted using a Qiagen RNeasy plant mini kit with an on-column DNase treatment following the manufacturer's recommended protocol (RNase-free DNase Set, Qiagen). cDNAs generated from the 3 biological replicates were synthesized following the manufacturer's recommended protocol (Omniscript RT Kit, Qiagen): 1,000 ng of RNA in RNase-free water were heated 10 min at 70 °C, and added to the manufacturer's mix (up to 20 µl) for 1 h at 37 °C. Generated cDNAs diluted 1/10 were added to the Takyon No Rox SYBR MasterMix blue dTTP (Eurogentec) reaction mix and used to perform qPCR analysis as recommended by the manufacturer. All qPCRs were carried out and analyzed using the Roche LC480 Lightcycler system and software (qPCR analysis program, Cq determination, curve calibration). Complete thermocycling parameters were as follow: 95 °C, 5 min; 95 °C, 10 s; 60 °C, 20 s; 72 °C, 40 s × 40 cycles; 95 °C, 1 min; 60 °C, 30 s; 95 °C, 30 s. Normalization was achieved with the tubulin TUB3 gene (AT5G62700) and the CDKA;1 gene (AT3G48750) (Lavenus et al. 2015; Trinh et al. 2019). The calibrator cDNA for relative quantification of the effect of each treatment was the control WT treated with NPA. Data are represented as the mean ± SEM. Significant differences between NAA-treated samples were determined using a Student t test (*P < 0.05; **P < 0.01; ***P < 0.001). Primer efficiency and specificity were confirmed by qPCR. Primers used are listed in Table S1.

Cuticle staining

For visualization of cell wall polyesters of lateral roots just emerged from the primary root, the Fluorol Yellow 088 protocol from Berhin et al. (2019) was used. Briefly, 10-d-old plants were incubated in Fluorol Yellow 088 (0.01% (w/v) in methanol) for 3 d at 4 °C. The plants were then counterstained with aniline blue (0.5% in water) for at least 1 h at room temperature, after which they were rinsed in water. They were then mounted in a solution of water and glycerol (1:1) and the newly emerged LRs were observed with a Leica stellaris confocal system. The following excitation/emission settings were used: 488/500 to 555 nm.

Permeability assays

For studying the permeability of the cuticle of lateral roots just emerged from the primary root, the fluorescein diacetate (FDA) protocol of Berhin et al. (2019) was used. Roots were incubated in FDA (5 μg/mL in ½ strength MS) for 4 min and rinsed in water before being mounted in water/glycerol (1:1) and observed with a Leica stellaris confocal system. The following excitation/emission settings were used: 488/500 to 555 nm.

Transmission electron microscopy

Plants were grown 5 d on Petri plates and then gravistimulated for 60 h for the WT, 96 h for bdg and 120 h for puchi to produce newly emerged LR in our different backgrounds. Roots were fixed in a paraformaldehyde EM-grade (2.5%) solution in 0.1 M phosphate buffer pH 6.8 for 2 h at room temperature and then washed in phosphate buffer. Samples were then included in low melting agarose (1.5%) in phosphate buffer for sections with emerging LR. Sections in agarose were then immersed in a second fixative solution with paraformaldehyde EM-grade (2.5%) and glutaraldehyde (2.5%) in 0.1 M cacodylate buffer pH 7.4 for 1 h at room temperature under gentle agitation and overnight at 4 °C.

Samples were then rinsed in PHEM buffer (Schliwa and van Blerkom 1981) and post-fixed in a 0.5% osmic acid + 0.8% potassium hexacyanoferrate trihydrate for 2 h in the dark at room temperature. After 2 rinses in PHEM buffer, the sections were dehydrated in a graded series of ethanol solutions (30 to 100%). The cells were embedded in EmBed 812 using an Automated Microwave Tissue Processor for Electronic Microscopy, Leica EM AMW. Thin sections (70 nm; Leica-Reichert Ultracut E) were collected at different levels of each block. These sections were counterstained with uranyl acetate 1.5% in 70% Ethanol and lead citrate and observed using a Tecnai F20 transmission electron microscope at 120KV in the “Institut des Neurosciences de Montpellier—Electronic Microscopy facilities”. The ultrastructure of the RCC of just emerging LR was investigated for 3 independent plants per genotype.

Confocal microscopy

Confocal microscopy images were obtained with a Leica SP8 and a Leica stellaris confocal system using a × 40 HCX corr Plan Apochromat CS 1.1 NA water objective. Samples were mounted in water or in 15 µM propidium iodide for 1 to 2 h or in 15 µM propidium iodide supplemented with 0.004% Triton X-100 for 20 to 40 min as previously described (Du and Scheres 2017). All combinatorial fluorescence analyses were run as sequential scans. The following excitation/emission settings were used to obtain specific fluorescence signals: for propidium iodide (PI), 568/573 to 740 nm; for yellow fluorescent protein (YFP), 514/520 to 550 nm. All post-acquisition images, such as channel merging, were obtained using the Leica Application Suite X software.

Root phenotyping

To perform LR and LRP density phenotyping, 9-d-old seedlings were first imaged using a digital camera to measure primary root length. Seedlings were then mounted in water, and the number of non-emerged LRPs and emerging LRs was counted. Primary root length was measured with Fiji (Schindelin et al. 2012). For the gravistimulation assay, seedlings were first grown in square Petri dishes for 5 d, then dishes were turned by 90° to induce the formation of a new LRP. After 18 or 48 h, seedlings were mounted in water and their LRPs were visualized under DIC microscopy with a Zeiss AX10 microscope. LRPs were categorized into developmental stages as described in (Malamy and Benfey 1997).

Statistical analysis

All the statistical analyses used in this study were performed with R. Except for the RNAseq analyses, significance was determined using a Student t test, normality and equality of variance were tested and the star sign was used to denote the P-value of the analysis: *P < 0.05; **P < 0.01; ***P < 0.001 for RT-qPCR and LRP stage proportion following gravistimulation assay, and *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 for the measurements of primary root and LR traits of 9-d-old seedlings. Statistical analysis on RNAseq replicates were performed using DESeq2 package and Wald Test: *P < 0.05; **P < 0.02; ***P < 0.01.

Acknowledgments

The authors thank the University of Montpellier for doctoral funding provided via the competitive fellowship program of the GAIA Doctoral School. We acknowledge the imaging facility MRI, member of the national infrastructure France-BioImaging (https://ror.org/01y7vt929) supported by the French National Research Agency (ANR-24-INBS-0005 FBI BIOGEN).

Author contributions

SG designed the research. AB, OV, CC, SG, and HA performed the research experiments. AB, OV, SG, and HA analyzed the data. GK contributed a new computational tool. AB, GK, PG, LL, and SG wrote the manuscript. SG, LL, PG, and GK obtained funding.

Supplementary material

Supplementary material is available at Plant Physiology online.

Funding

This work was supported by the Institut de Recherche pour le Développement, the Université deMontpellier, and theAgence Nationale de la Recherche (fatROOT, ANR-21-CE13-0019-01).

Data availability

Data available on request.