PAT1-type GRAS-domain proteins control regeneration by activating DOF3.4 to drive cell proliferation in Arabidopsis roots

Abstract

Plant roots possess remarkable regenerative potential owing to their ability to replenish damaged or lost stem cells. ETHYLENE RESPONSE FACTOR 115 (ERF115), one of the key molecular elements linked to this potential, plays a predominant role in the activation of regenerative cell divisions. However, the downstream operating molecular machinery driving wound-activated cell division is largely unknown. Here, we biochemically and genetically identified the GRAS-domain transcription factor SCARECROW-LIKE 5 (SCL5) as an interaction partner of ERF115 in Arabidopsis thaliana. Although nonessential under control growth conditions, SCL5 acts redundantly with the related PHYTOCHROME A SIGNAL TRANSDUCTION 1 (PAT1) and SCL21 transcription factors to activate the expression of the DNA-BINDING ONE FINGER 3.4 (DOF3.4) transcription factor gene. DOF3.4 expression is wound-inducible in an ERF115-dependent manner and, in turn, activates D3-type cyclin expression. Accordingly, ectopic DOF3.4 expression drives periclinal cell division, while its downstream D3-type cyclins are essential for the regeneration of a damaged root. Our data highlight the importance and redundant roles of the SCL5, SCL21, and PAT1 transcription factors in wound-activated regeneration processes and pinpoint DOF3.4 as a key downstream element driving regenerative cell division.

IN A NUTSHELL

Background: Plants display an unrivaled ability to regenerate. This allows them to recover from tissue damage (ranging from the death of a single cell to loss of a complete root tip) resulting from harsh weather conditions or herbivore attack. Following damage, plants activate transcriptional cascades that re-activate cell division in the surviving cells, resulting in the replacement of the damaged cell or even lost tissues. The Arabidopsis thaliana transcription factor ETHYLENE RESPONSE FACTOR 115 (ERF115) is a key regulator that initiates tissue regeneration together with SCARECROW-LIKE (SCL) proteins.

Question: Although ERF115's role in the regeneration process is well-characterized, the contribution of the SCL protein partners was less clear, probably due to gene redundancy. Moreover, the molecular mechanism by which these partner proteins aid in regeneration was unclear. We, therefore, searched for novel ERF115 binding partners and studied their contribution to regeneration.

Findings: Protein–protein interaction studies identified SCL5 as an ERF115 interaction partner, in addition to the known, related proteins SCL21 and PHYTOCHROME A SIGNAL TRANSDUCTION 1 (PAT1). Phenotypic testing of higher order mutants revealed that these 3 proteins together are dispensable for development under control growth conditions, but essential for cell renewal following wounding, as their combined knockout results in a complete inhibition of regeneration. Genome-wide expression analysis showed that the DOF3.4 transcription factor acts downstream of ERF115 and the SCL proteins and activates wound-induced cell division by directly stimulating components of the cell cycle machinery.

Next steps: Although some plant species easily and effectively regenerate damaged tissue, other species, including many economically important cereal crops, appear deficient in regeneration. At some point, we aim to translate the knowledge gained in Arabidopsis into crop species with a low regeneration capacity to boost their regenerative capacities.

Introduction

Regeneration is a self-preservation mechanism by which a tissue or organ regains its structure and function after injury. Depending on the species, the extent of regeneration is highly variable. Vertebrates can regenerate certain structures, such as limbs in salamanders or tail fins in fish. In humans, there are anecdotal reports of digit regeneration after amputation of fingertips (Illingworth 1974). Invertebrate species, including sea squirts, planarian flatworms, and cnidarians, are capable of more extensive tissue replacement (Ricci and Srivastava 2018). No clade, however, competes with plants, which are known for their unparalleled regenerative capacity, allowing them to regenerate for survival under severe stress conditions, such as herbivore attack and harsh weather conditions. This regeneration potential is exemplified in Arabidopsis thaliana (Arabidopsis), where a completely excised root tip efficiently regenerates within 3 to 4 d (Sena et al. 2009). Remarkably, plant regeneration not only replenishes tissues and restores damaged organs, but can also give rise to whole plant bodies (Ikeuchi et al. 2019; Christiaens et al. 2021). This unique feature of asexual reproduction is widely exploited in modern agricultural applications, including vegetative propagation and grafting (Sussex 2008; Duclercq et al. 2011; Xu and Huang 2014).

One critical feature of plant regeneration is the reformation and maintenance of meristems, the proliferative tissues that hold the stem cells. Like animal stem cells, plant stem cells are defined by their ability to renew themselves and to generate daughter cells that establish tissues and organs. These stem cells are maintained in specialized microenvironments, commonly referred to as stem cell niches (SCN). In Arabidopsis, the SCN holds a quiescent center (QC) that is surrounded by a single tier of stem cells (van den Berg et al. 1997; Sarkar et al. 2007). Despite their low proliferation rate, QC cells have been found to operate as reservoirs to provide new cells. When stress occurs, QC cells are triggered to divide and contribute to replace lost neighboring stem cells, a process driven by many different signals, including phytohormones like auxin, cytokinin, brassinosteroids, ethylene, and jasmonate (Aida et al. 2004; Curtis and Hays 2007; Fulcher and Sablowski 2009; Cruz-Ramírez et al. 2013; Mähönen et al. 2014; Zhou et al. 2019; Lu et al. 2021). Within the QC, WUSCHEL-RELATED HOMEOBOX 5 (WOX5) expression is required to maintain QC cell identity (Sarkar et al. 2007).

Previously, we identified ETHYLENE RESPONSE FACTOR 115 (ERF115), a transcription factor that controls the replenishment of lost cells or tissues within the Arabidopsis root meristem following wounding (Heyman et al. 2013, 2016). ERF115 expression is instantly induced in cells adjacent to those undergoing cell death induced by root tip excision, selective killing of stem cells by cytotoxic drugs, or laser-mediated cell ablation (Heyman et al. 2016; Marhava et al. 2019; Zhou et al. 2019; Hoermayer et al. 2020). Activated ERF115 subsequently facilitates the reprograming of these cells to become the new stem cells. These cells consequently undergo formative cell division, generating cells that replenish the damaged ones (Heyman et al. 2016; Canher et al. 2020; Hoermayer et al. 2020). Strikingly, even a single dying cell can activate a rapid ERF115 response in its surrounding cells. During recovery from laser-assisted root stem cell removal, ERF115 was found to interact biochemically and genetically with the RETINOBLASTOMA RELATED–SCARECROW (RBR–SCR) signaling network that controls stem cell divisions in response to environmental stress (Zhou et al. 2019).

Once activated, ERF115 acts as a transcriptional activator of the PHYTOSULFOKINE 5 (PSK5) precursor gene (Heyman et al. 2013), encoding a small peptide phytohormone known for its cell division-promoting potential in suspension cultures and QC cells (Matsubayashi and Sakagami 1996; Kong et al. 2018). In addition, ERF115 controls the expression of the WOUND INDUCED DEDIFFERENTIATION 1 (WIND1) transcription factor gene (Heyman et al. 2016), facilitating the reprograming of cells and allowing them to regain pluripotency required to initiate tissue regeneration (Iwase et al. 2011). Furthermore, ERF115 enhances auxin sensitivity in wound-adjacent cells through transcriptional activation of the MONOPTEROS/AUXIN RESPONSE FACTOR 5 (MP/ARF5) gene, thereby triggering the reacquisition of stem cell identity (Canher et al. 2020). In contrast, the molecular components that link ERF115 with the cell cycle machinery are still unknown.

Next to ERF115, GIBBERELLIN-INSENSITIVE (GAI), REPRESSOR OF GA1-3 (RGA), and SCARECROW (SCR) (GRAS)-domain transcription factors may play a role in regeneration. The GRAS-domain protein family comprises 8 subfamilies that act in a diverse set of stress and developmental processes (Tian et al. 2004). Among these, the members of the DELLA family act as repressors of the gibberellin signaling pathway (Cheng et al. 2004; Cao et al. 2005), whereas HAIRY MERISTEM (HAM) proteins are essential for indeterminate growth maintenance, with ham mutants displaying a shoot meristem arrest and differentiation phenotype (Engstrom 2012). SHORT ROOT (SHR) and SCR are involved in the control of root radial patterning and root growth (Helariutta et al. 2000). The Arabidopsis PHYTOCHROME A SIGNAL TRANSDUCTION 1 (PAT1) subfamily has 5 genes, namely PAT1, SCARECROW-LIKE 1 (SCL1), SCL5, SCL13, and SCL21 (Bolle 2004). Among these, PAT1 and its closest homolog SCL21 were originally identified as components that act positively on the phytochrome A (phyA)-dependent light signaling pathway (Bolle 2004; Torres-Galea et al. 2006, 2013). Additionally, the PAT1 protein dimerizes with ERF115 to boost its regenerative potential, because ectopic coexpression of Arabidopsis ERF115 with PAT1 activates neoplastic growth that manifests as spontaneous callus formation (Heyman et al. 2016). Correspondingly, the erf115 and pat1 single and double mutants show an equally impaired root tip regeneration phenotype, suggesting that these 2 proteins form an active protein complex rather than acting in 2 independent pathways.

ERF115 belongs to the EREB-DREB subfamily X of the AP2/ERF group of transcription factors, comprising 8 members, namely ERF108 to ERF115 (Heyman et al. 2018). Sequence alignment of the EREB-DREB subfamily X revealed a conserved 11-amino-acid (AA) CMX-1 motif in the N-terminal region as a common feature of all members but ERF112 (Nakano et al. 2006). This motif was found to be both essential and sufficient to drive interaction with the PAT1-related SCL21 transcription factor (Heyman et al. 2016). Here, we identified SCL5 as an interaction partner of ERF115 through a systematic interaction study between all EREB-DREB subfamily X and PAT1 subfamily members. SCL5 operates redundantly with its family members PAT1 and SCL21 in the reconstitution of the SCN following wounding. Moreover, we demonstrate that SCL5, along with PAT1 and SCL21, drives regenerative cell division through the transcriptional control of DNA-BINDING ONE FINGER 3.4 (DOF3.4, also known as OBP1), which in turn activates the expression of cyclin CYCD3;3.

Results

SCL5 interacts with ERF114 and ERF115

Although ectopic coexpression of ERF115 and PAT1 was found to dramatically boost the regenerative capacities of ERF115, the corresponding erf115 pat1 double mutant failed to mimic the complete loss-of-regeneration phenotypes that were observed for plants expressing the dominant-negative ERF115^SRDX allele, suggesting redundancy with putative additional ERF-GRAS heterodimers. To gain insight into the possible pairwise interactions between the subfamily X EREB-DREB transcription factors and the GRAS-domain protein subgroup PAT1 family, a yeast two-hybrid (Y2H) strategy was used. In this assay, SCL13 showed strong autoactivation, whereas SCL1 failed to show interaction with any of the tested ERF members (Fig. 1, A and B). PAT1 interacted with 5 ERF members (all except ERF108, ERF112, and ERF113), whereas SCL21 interacted with all members except ERF112, consistent with the absence of the N-terminally located 11-AA CMX-1 motif demonstrated before to mediate interaction with SCL21 (Heyman et al. 2016). SCL5 interacted with 4 ERF proteins, namely ERF109, ERF111, ERF114, and ERF115.

Figure 1

SCL5 is an interaction partner of ERF115. A and B) Y2H interactions between EREB-DREB subfamily X (cloned as prey in the pDEST22 vector) and PAT1 subfamily GRAS-domain (baits in the pDEST32 vector) transcription factors grown on medium holding (-LT) (A) or lacking histidine (-LTH) supplemented with 10 mM 3-AT (B). Interactions were scored by gain of histidine prototrophy. The GUS gene was taken along as a negative control. C and D) Y2H interactions between the newly annotated SCL5 (cloned as prey in the pDEST22 vector) and the N-terminal ERF A-fragments (baits in the pDEST32 vector) grown on medium holding (-LT) (C) or lacking histidine (-LTH) supplemented with 80 mM 3-AT (D). Interactions were scored by gain of histidine prototrophy. E) Confocal imaging of SCL5 and ERF115 expression in a root tip treated for 24 h with 0.6 μg/mL bleomycin. pSCL5:NLS-GFP/GUS (left), pERF115:NLS-TdTomato (middle), merge (right). All images were acquired at the same magnification; scale bar = 50 µm. Cell walls were counterstained using SCRI Renaissance 2200.

Previously, SCL5 could not be detected in a tandem affinity purification experiment using ERF115 as the protein bait (Heyman et al. 2016), possibly due to the interaction between SCL5 and ERF115 being transient or weak in nature. Therefore, an ERF115 single-step affinity purification followed by mass spectrometry was performed, increasing the chance to identify loose and/or transient interaction partners. Next to ERF113 and ERF114, both SCL21 and SCL5 could be detected among the ERF115 copurifying proteins (Table 1; Supplemental Data Set S1), confirming the interaction between SCL5 and ERF115 at the biochemical level.

Table 1.

ERF115 FAST-IP data

Protein accession	Protein description	# identified/3 experiments
AT5G07310	ERF115, Integrase-type DNA-binding superfamily protein	3
AT1G50600	SCL5 | scarecrow-like 5	3
AT1G74380	XXT5 | xyloglucan xylosyltransferase 5	3
AT2G04890	SCL21 | SCARECROW-like 21	3
AT2G44040	Dihydrodipicolinate reductase, bacterial/plant	3
AT3G59890	Dihydrodipicolinate reductase, bacterial/plant	3
AT3G62720	ATXT1, XT1, XXT1 | xylosyltransferase 1	3
AT5G13330	Rap2.6L (ERF113) | related to AP2 6l	3
AT5G61890	ERF114, Integrase-type DNA-binding superfamily protein	2

List of significantly enriched proteins identified in ERF115 pull-downs, as identified by LC-MS/MS on Q Exactive HF in GS^TEV-tagged ERF115 pull-downs. Only proteins with at least 2 peptide sequences, identified in at least 2 out of 3 replicates, and enriched by quantitative analysis versus a large data set of similar non-related bait pull-down experiments were retained.

To study putative coexpression of SCL5 and ERF115, available gene annotation data were used to construct an SCL5 transcriptional reporter line. Using 2,587 bp of the predicted sequence upstream of the SCL5 coding region, independent reporter lines driving the expression of a nuclear-targeted GFP were generated. By examining these lines, we detected a weak GFP-positive signal within cortical and endodermal meristem cells and a stronger signal in a subset of the vascular stem cells (Supplemental Fig. S1A). However, these reporter lines did not show the expected nuclear-localized GFP signal, suggesting that the predicted promoter region might be incorrect. Therefore, we remapped SCL5 RNA-sequencing (RNA-seq) reads on the genomic locus, revealing the absence of detectable reads matching the beginning of the first predicted SCL5 exon (Supplemental Fig. S1C), suggesting a misannotation of the SCL5 translation start site. The next in-frame start codon is located 569 bp downstream of the annotated start codon (Supplemental Fig. S1D), resulting in an open reading frame (ORF) of 1,581 bp, encoding a 526-AA-long protein. This prediction is supported by the Arabidopsis PeptideAtlas (www.peptideatlas.org/builds/arabidopsis) that maps all available mass spectrometry peptide data (van Wijk et al. 2021).

The newly predicted SCL5 ORF was retested in the Y2H assay. Although the newly encoded SCL5 protein could not be used as bait due to autoactivation, it could be used as a prey against ERF fragment baits harboring the N-terminal GRAS-domain interaction motif, resulting in an interaction with the same ERF members as found before (Fig. 1, C and D). In addition, a bimolecular fluorescence complementation experiment was conducted using the ERF115-headGFP and SCL5-tailGFP protein fusions. We observed a clear restoration of GFP fluorescence in the nucleus of Nicotiana benthamiana pavement cells, indicative of in vivo ERF115-SCL5 association (Supplemental Fig. S2A), a pattern similar to the previously reported ERF115 and SCL21 interaction (Heyman et al. 2016) (Supplemental Fig. S2B), opposed to its negative controls (Supplemental Fig. S2C and S2D).

Supported by these observations, new SCL5 promoter reporter lines were generated using the reannotated 2,525-bp promoter sequence, revealing SCL5 expression throughout the root meristem displaying the characteristic nuclear-localized signal, with an apparently stronger expression in the SCN, as well as the stele, pericycle, and endodermal tissue (Supplemental Fig. S1B), a pattern substantiated by available single-cell RNA-seq data (https://rootcellatlas.org). Following treatment of a pSCL5:NLS-GFP/GUS pERF115:NLS-TdTomato dual reporter line with bleomycin to induce vascular stem cell death and subsequent ERF115 expression, a substantial overlap between the SCL5 and ERF115 expression patterns could be observed (Fig. 1E, Supplemental Fig. S3), supporting the potential in planta interaction between the 2 gene products.

Ectopic ERF-GRAS expression results in neoplasm in both root and shoot

To support a genetic interaction between SCL5 and ERF115, as well as its closest relative ERF114, we phenotypically compared single versus ERF-GRAS co-overexpressing plants. Within these experiments, PAT1 and SCL21 were used as positive controls. Young seedlings with ectopic overexpression of ERF115 and SCL21 displayed a significant reduction in root length compared with the wild type, in accordance with previous observations (Heyman et al. 2016; Lu et al. 2021) (Fig. 2A, Supplemental Fig. S4). At the microscopic level, only overexpression of SCL21 appeared to cause a significant reduction in the number of meristematic cortex cells compared with the wild type (Supplemental Fig. S5).

Figure 2

ERF115-GRAS ectopic co-overexpression activates hyperproliferation. A) Twelve-day-old wild-type, parental plants and F1 progeny plants resulting from a cross between ERF115^OE and respective GRAS^OE lines. All images were acquired at the same magnification; scale bar = 1 cm. B) Confocal images of 7-d-old root meristems of the ERF115^OE-GRAS^OE lines. Scale bars = 20 µm. Cell walls were counterstained using PI. C) One-month-old control (Col-0) and ERF115^OE-GRAS^OE F1 plants. Scale bars = 1 cm. D) PSK5 relative expression levels in control (Col-0) and ERF115 and GRAS single and double overexpressors. Error bars indicate ±SEM (**P < 0.01, ANOVA followed by Tukey's multiple comparison test, n = 3 biological repeats each holding >3 seedlings).

Subsequently, we phenotypically analyzed the offspring obtained from crosses between these different lines. Co-overexpression of ERF115 with SCL5, SCL21, or PAT1 resulted in a strong inhibition of root growth compared with the single overexpression lines (Fig. 2A; Supplemental Fig. S4), which coincided with a disorganized SCN (Fig. 2B), callus formation at the root–hypocotyl junction, and a disorganized shoot (Fig. 2C), as was reported before for ERF115-PAT1 co-overexpressing plants (Heyman et al. 2016). When testing the genetic interaction with ERF114, effects on primary root growth were less severe compared with ERF115 interaction (Supplemental Figs. S4 and S6A). Likewise, a disorganized root SCN was detected for all respective combinations (Supplemental Fig. S6B), although less pronounced when compared with the ERF115 combinations (Fig. 2B), whereas the disorganized shoot phenotypes were much alike (Supplemental Fig. S6C). Following 25 d of growth, ERF114^OE plants displayed the previously reported increased bud outgrowth (Mehrnia et al. 2013), whereas SCL21^OE plants showed a reduction in rosette size (Supplemental Fig. S7). For the different ERF-GRAS co-overexpressing combinations, distinct phenotypic responses were observed (Supplemental Fig. S7), of which the severity varied. Overall, these data demonstrate that ERF114 and ERF115 in combination with SCL5, PAT1, or SCL21 trigger phenotypes that affect meristem activity. The synergetic effect of both transcription factor classes was confirmed for the ERF115-related crosses at the level of PSK5 expression, a previously reported ERF115 target gene (Fig. 2D).

The GRAS-domain scl5 pat1 scl21 triple mutant is regeneration deficient

Single mutants of pat1 were reported to show only a mild defect in root regeneration post wounding (Heyman et al. 2016), probably owing to gene redundancy. Therefore, we generated scl5 scl21, scl5 pat1, and pat1 scl21 double mutants and the scl5 scl21 pat1 triple mutant. Under control conditions, even the scl5 scl21 pat1 triple mutant did not show any clear macroscopic difference compared with wild-type (Col-0) plants (Supplemental Fig. S8). Subsequently, these mutant combinations were tested for a putative wound-responsive phenotype. After a 24-h bleomycin treatment that results in vascular cell death, the roots were transferred to bleomycin-free medium, allowing them to replenish the dead cells through regenerative cell divisions (Heyman et al. 2013; Canher et al. 2020). Following 5 d of recovery, the regeneration rate was slightly, but significantly, reduced for the pat1 scl5 and pat1 scl21 double mutant combinations, whereas the triple scl5 scl21 pat1 mutant displayed a strong impairment in regeneration capacity (Fig. 3A). The triple mutant phenotype could be complemented by the introduction of a pSCL5:SCL5-GFP or pPAT1:PAT1-GFP translational reporter (Fig. 3B), confirming that the combinational absence of the 3 GRAS proteins accounted for the defective regeneration phenotype.

Figure 3

Triple mutant scl5 scl21 pat1 plants are regeneration deficient. A) Recovery percentage following 5 d after a 24-h 0.6-μg/mL bleomycin treatment of the wild type (Col-0) and indicated higher order mutants. Recovery was scored defective by arrested root growth and visual confirmation of a collapsed root meristem by confocal microscopy (*P < 0.05; **P < 0.001, Fisher's exact test, n > 80 root tips evenly distributed over 3 biological repeats). B) Complementation of the scl5 scl21 pat1 triple mutant using translational reporters of C-terminally GFP-tagged SCL5 and PAT1 proteins (*P < 0.05; **P < 0.001, Fisher's exact test, n > 80 root tips evenly distributed over 3 biological repeats). C) Regeneration percentage following 72-h post root tip excision of wild-type (Col-0) and scl5 scl21 pat1 triple mutant plants (**P < 0.001, χ² test, n > 75 root tips evenly distributed over 3 biological repeats). D) Quantification of periclinal divisions in cortex and endodermis following ablation of epidermal and cortical cells in wild-type (Col-0) and scl5 scl21 pat1 triple mutant plants (**P < 0.01; *P < 0.05; Student's t-test, n > 30 ablation events executed in at least 18 roots). E and F) pWOX5:NLS-GFP/GUS marker expression in wild-type (Col-0) (E) and scl5 scl21 pat1 triple mutant (F) root meristems before treatment (Mock), 24-h post-treatment (hpt) with 0.6 μg/mL bleomycin, and 24-, 48- and 72-h post-recovery (hpr) on bleomycin-free medium. Scale bars = 20 µm. Cell walls were counterstained using PI. The number of identical independent observations is shown within the respective representative images.

The regeneration-deficient phenotype seen following bleomycin treatment was studied at the cellular level using the WOX5 QC cell identity marker, demonstrated before to accumulate in an ERF115-dependent manner in the endodermal and vascular cells neighboring the dead cells (Heyman et al. 2013; Canher et al. 2020). To this end, the pWOX5:NLS-GFP/GUS construct was introduced in the scl5 scl21 pat1 triple mutant by transformation. Under control conditions and following 24 h of bleomycin treatment, no clear differences in the WOX5-GFP expression pattern between wild-type and scl5 scl21 pat1 triple mutant plants could be observed (Fig. 3, E and F). In Col-0 plants, upon retransfer of the bleomycin-treated plants to drug-free medium, the WOX5-GFP expression domain increased shootwards and infiltrated the vascular tissue that surrounded the damaged cells (Fig. 3E). Three days post transfer, the WOX5-GFP signal was again confined to the SCN, which coincided with the recovery of the root tip visualized by the strong reduction in cell death, demonstrated before to be in part due to periclinal endodermal cell divisions (Canher et al. 2020). By contrast, within the scl5 scl21 pat1 mutant roots, the WOX5-GFP accumulation pattern appeared to diminish during the 24- to 48-h post-recovery timeframe, followed by a collapse of the vascular tissue at 72 h (Fig. 3F), probably accounting for the arrested root growth (Fig. 3A). These data indicate that these 3 GRAS transcription factors are required for root meristem re-establishment following vascular stem cell death, likely by participating in the periclinal divisions demonstrated before to replenish the vascular cells.

Next, we tested the regenerative response of the scl5 scl21 pat1 mutant following root tip excision. Whereas 55% of the control plants were able to generate a new root tip within 3 d following removal of the 200-μm distal portion, the regeneration rate dropped to less than 2% in the triple mutant (Fig. 3C). Likewise, the regenerative response of the triple mutant to laser-mediated cell ablation was investigated, where single cells were killed in a cell type-specific manner. Single-cell wounding typically involves a change in division plane from anticlinal (perpendicular to the growth axis) to periclinal (parallel to the growth axis) in cells located adjacent to the damaged cell in the inward located cell file, allowing for efficient replacement of the dead cells (Marhava et al. 2019). Compared with the wild type, a significant decrease in periclinal division potential in the adjacent cortex or endodermal cells could be observed within the scl5 scl21 pat1 mutant following ablation of an epidermal or cortical cell, respectively (Fig. 3D).

DOF3.4 operates downstream of the PAT1-branch GRAS-domain transcription factors

To explore the underlying mechanism of the root regeneration defect of the scl5 scl21 pat1 triple mutant, an RNA-seq experiment was performed, comparing the transcriptome of 2-mm root tips, both under control conditions and following a 24-h bleomycin treatment, using Col-0 plants as a negative control. When comparing Col-0 and the scl5 scl21 pat1 triple mutant grown under control conditions, no significant differential gene expression was observed (false discovery rate (FDR) < 0.05), except for the 3 targeted genes, SCL5, SCL21, and PAT1 (Supplemental Data Sets S2–S3), supporting the absence of any phenotype under control conditions. Under the bleomycin-treated condition, 1,461 genes in Col-0 and 1,746 genes in scl5 scl21 pat1 were differentially expressed compared with their respective mocks (Supplemental Data Sets S2–S3).

To pinpoint the genes responsible for the impaired root regeneration phenotype of the scl5 scl21 pat1 mutant, we divided these genes into 3 classes. The first class of genes were those that were upregulated in both genotypes upon bleomycin treatment (Supplemental Data Set S4). The second class holds genes only upregulated in Col-0 following bleomycin treatment (Supplemental Data Set S5), whereas the third group includes genes only upregulated in the scl5 scl21 pat1 triple mutant (Supplemental Data Set S6). Using this subgrouping, 1,022 genes were found, among which 468 genes were upregulated in both genotypes, 317 specifically in Col-0, and 237 specifically in scl5 scl21 pat1. In the resulting gene list, the DOF3.4 transcription factor gene was identified as one of the genes showing a high fold induction in Col-0 following bleomycin treatment, but not in the scl5 scl21 pat1 triple mutant (Table 2). DOF3.4 belongs to a gene family of which members have been characterized in the context of periclinal cell division during vascular development (Miyashima et al. 2019; Smet et al. 2019). Considering the need for periclinal cell divisions in the recovery of bleomycin-treated root tips (Heyman et al. 2016; Canher et al. 2020), the DOF3.4 gene appeared as a good candidate gene for driving cell divisions in the root regeneration process.

Table 2.

Top-5 upregulated genes in Col-0 only, post 24-h bleomycin treatment, identified following RNA-seq

Gene accession	Gene function	Fold change
AT4G24260	Endoglucanase	21/KOR3 202
AT1G15580	Auxin-responsive protein	IAA5 133
AT3G50410	Dof zinc finger protein	DOF3.4 79
AT1G20120	GDSL esterase/lipase	33
AT4G18910	Aquaporin	NIP1-2 14

FC (fold change) indicates the ratio between Col-0 under treated (bleomycin) vs. untreated condition.

To study the putative role of DOF3.4 in tissue regeneration, a pDOF3.4:NLS-GFP/GUS reporter line was constructed, fusing its promoter (1,642 kb) to NLS-GFP/GUS. Under control conditions, DOF3.4 expression within the root meristem appeared variable. During the initial days of growth, weak expression could be seen in the entire root meristem, including columella cells (Supplemental Fig. S9A). However, at later growth stages, its expression was only sporadically observed in SCN cells, including cell initials of cortex/endodermis (Supplemental Fig. S9B). DOF3.4 expression mostly appeared in paired neighboring cells, suggesting a correlation between expression and recent cell division. Following treatment with bleomycin for 24 h, its expression was strongly induced in the SCN and columella cells, confirming the RNA-seq data (Fig. 4A).

Figure 4

DOF3.4 is a wound-responsive gene operating downstream of SCL5, SCL21, and PAT1. A and B) Expression pattern of pDOF3.4:NLS-GFP/GUS in control (Col-0) (A) and scl5 scl21 pat1 triple mutant plants (B) before treatment (Mock), 24-h post-treatment (hpt) with 0.6 μg/mL bleomycin, and 24-, 48-, and 72-h post-recovery (hpr) on bleomycin-free medium. C) pDOF3.4:NLS-GFP/GUS (Col-0 background) during root tip regeneration at indicated h post cut (hpc). A–C) Cell walls were counterstained using PI. All images were acquired at the same magnification; scale bars = 20 µm.

To determine whether the GRAS transcription factors are required for DOF3.4 expression, the pDOF3.4:NLS-GFP/GUS reporter line was introduced in the scl5 scl21 pat1 triple mutant background. Our results showed that the DOF3.4 expression pattern was variable in the triple mutant (Fig. 4B), being largely identical to that observed in the wild-type background under non-wounded conditions. However, upon stem cell death following bleomycin application, DOF3.4 expression was strongly reduced in the scl5 scl21 pat1 triple mutant and remained notably weaker compared with the wild type during the 24-, 48- and 72-h recovery time points (Fig. 4B). Because of the observed regeneration defects of the scl5 scl21 pat1 triple mutant following root tip excision, we also mapped DOF3.4 expression in this regeneration setup. In wild-type plants, DOF3.4 expression was strongly activated in endodermis and cortex cell layers of the remaining stump after tip removal and subsequently expanded to inner cell layers during root tip re-establishment (Fig. 4C).

Since the GRAS proteins interact with ERF115 and the triple GRAS mutant shows a regeneration phenotype resembling that of the dominant-negative version of ERF115, we tested whether DOF3.4 expression was affected in the dominant-negative ERF115^SRDX background by the introduction of the pDOF3.4:NLS-GFP/GUS reporter construct through crossing. The resulting F1 seedlings were treated with bleomycin and recovery was followed on a bleomycin-free medium. In contrast to the strong transcriptional activation observed in wild-type plants, DOF3.4 expression was strongly attenuated in ERF115^SRDX root tips (Fig. 5A), reminiscent of the response observed for the scl5 scl21 pat1 triple mutant (Fig. 4B). This observation was confirmed through reverse transcription-quantitative PCR (RT-qPCR, Fig. 5B), suggesting direct control of DOF3.4 promoter activity by ERF115. Accordingly, previously published data generated from a tandem chromatin affinity purification experiment using ERF115 as bait, followed by sequencing, revealed enrichment of DOF3.4 promoter sequences (Supplemental Fig. S10; Heyman et al. 2013).

Figure 5

Bleomycin-induced DOF3.4 expression is ERF115 dependent. A) pDOF3.4:NLS-GFP/GUS expression in the ERF115^SRDX mutant root meristem during the recovery process following bleomycin-induced stem cell death. All images were acquired at the same magnification; scale bar = 20 µm. Cell walls were counterstained using PI. B) Relative expression levels of DOF3.4 measured by RT-qPCR in wild-type (Col-0) versus ERF115^SRDX root tips being either mock-treated or treated for 24 h with 0.6 μg/mL bleomycin. Data indicate mean ± SEM (**P < 0.01, ANOVA followed by Tukey's multiple comparison test compared with control, n = 3 with each biological repeat holding > 100 root tips).

DOF3.4 drives periclinal divisions in the cortex and epidermis

A dof3.4 T-DNA insertion mutant showed only a mild phenotype upon bleomycin recovery, where only 20% of the seedlings showed impaired stem cell recovery, compared with 13% in the wild-type plants (Supplemental Fig. S11A), probably due to gene redundancy in the large DOF-type transcription factor family (Zhang et al. 2022). Regeneration after root tip excision also did not show any significant differences between both genotypes (Supplemental Fig. S11B). Therefore, to gain functional insight into the role of DOF3.4, we adopted a chimeric repressor silencing technology to convert DOF3.4 into a dominant-negative regulator by fusing it with the ETHYLENE-RESPONSIVE ELEMENT BINDING FACTOR-ASSOCIATED AMPHIPHILIC REPRESSION (EAR)-motif repression domain. Two independent DOF3.4^SRDX expression lines were generated (Supplemental Fig. S11D), which displayed a statistically significant reduction in root tip regeneration following excision (Supplemental Fig. S11E), despite the lack of an obvious root length or meristem phenotype (Supplemental Fig. S11, C and F).

To assess the possibility of DOF3.4 driving the regeneration process through the activation of cell divisions, we made use of DOF3.4 overexpression lines. Within independent segregating T2 populations, we observed a Mendelian inheritance (1:1:2) of wild-type plants with a long root phenotype, homozygous (Hmz) plants with a short root, and hemizygous (Htz) plants with an intermediate phenotype (Fig. 6A), suggesting a dose-dependent effect of DOF3.4 on root growth. The Hmz DOF3.4^OE plants were marked by a greater root diameter (Fig. 6B). To investigate in detail the root phenotype of these overexpression lines, radial sections of DOF3.4^OE roots were examined. A significant increase in the epidermis and vascular cell numbers was observed in both the Hmz and Htz DOF3.4^OE lines; those in the cortex and endodermis were only significantly increased in Hmz plants, although an increase was also observed in the endodermis cell number in Htz plants (Fig. 6, C and D). Division plane orientations indicated that the increase in cell number was in part due to periclinal divisions in the endodermal, cortical, and epidermal cell layers (Fig. 6C).

Figure 6

Ectopic DOF3.4 expression activates periclinal cell division. A) Phenotypes of control (Col-0) and DOF3.4 overexpression (DOF3.4^OE) homozygous (Hmz) and hemizygous (Htz) seedlings. Scale bar = 1 cm. B) Confocal images of control (Col-0) and DOF3.4^OE Hmz and Htz root meristems. Scale bars = 20 µm. Cells were counterstained using PI. C) Control (Col-0) and DOF3.4^OE Hmz and Htz root meristems showing periclinal divisions in epidermal, cortical, and endodermal cell layers indicated by *. Root sectioning was performed on 5-d-old plants. Scale bars = 50 µm. D) Quantification of cell numbers in different cell layers in Col-0 and DOF3.4^OE plants (**P < 0.01, *P < 0.05, ANOVA followed by Tukey's multiple comparison test, n = 4 sections).

Since constitutive overexpression of DOF3.4 resulted in a severe growth phenotype, inducible lines were generated by fusing DOF3.4 to the glucocorticoid receptor (GR) domain. Following 24 h of dexamethasone (DEX) treatment, confocal imaging of root sections revealed a statistically significant increase in the number of epidermal, cortical, endodermal, and vascular cells in DOF3.4^GR roots compared with mock-treated roots (Supplemental Fig. S12), reminiscent of DOF3.4^OE seedlings (Fig. 6D). When grown for 55 h on DEX-supplemented medium, these DOF3.4^GR inducible lines phenocopied (Fig. 7A) the drastic root radial expansion trait of the constitutive DOF3.4^OE Hmz plants (Fig. 6B). This root thickening was accompanied by an increase in stem cell activity, as exemplified by an increased number of columella stem cell layers visualized by pseudo-Schiff staining (Fig. 7B). A similar increase in columella stem cell layers has been observed upon increased activity of the D-type cyclin CYCD3;3 (Forzani et al. 2014), matching the previously reported observation that DOF3.4 acts as a transcriptional activator of the CYCD3;3 gene (Skirycz et al. 2008). Therefore, we analyzed the expression of a CYCD3;3 reporter construct (pCYCD3;3:GFP/GUS) in the DOF3.4^GR lines, revealing an expansion of the CYCD3;3 expression domain from the distal columella to the columella stem cells already after 24 h and further increasing after 48 h of DEX treatment (Fig. 7C).

Figure 7

DOF3.4 activates CYCD3;3 expression required for root tip regeneration. A) Confocal images of DOF3.4^GR root meristems in the absence (control) or presence (+) of 10 μM DEX (55-h treatment). Scale bars = 20 µm. Cell walls were counterstained using PI. B) Schiff staining of DOF3.4^GR meristems in the absence or presence of 10 μM DEX (55-h treatment). Scale bars = 50 µm. C) pCYCD3;3:GFP/GUS expression in DOF3.4^GR meristems in the absence or presence of 10 μM DEX (48-h treatment). Scale bars = 50 µm. Cell walls were counterstained using PI. Ratios at the right bottom indicate the number of identical observations. D) Regeneration percentage 72-h post root tip excision of the wild type (Col-0), and cycd3;3 single and cycd3;1 cycd3;2 cycd3;3 triple mutants (cycd3) (*P < 0.01; ***P < 0.001, Fisher's exact test, n > 75 evenly distributed over 3 biological repeats).

We further reasoned that DOF3.4-targeted D-type cyclins might be involved in the regeneration process following root tip removal. Interestingly, already the single cycd3;3 mutant showed a significant reduction in root tip regeneration after excision (Fig. 7D). This defect was enhanced in the cycd3 triple mutant (cycd3;1 cycd3;2 cycd3;3) that showed a drastic reduction in regeneration 72 h post root tip excision. These data suggest a putative role of D-type cyclins in root regeneration.

Discussion

PAT1-branch transcription factors are required for wound regeneration

The ERF115 transcription factor is indispensable for the regeneration of a damaged root meristem (Heyman et al. 2013), with its cell division capacity being dramatically enhanced through heterodimerization with PAT1 (Heyman et al. 2016). Although both erf115 and pat1 mutants display a regeneration defect following root tip excision, this phenotype is not as penetrant as that of the dominant-negative ERF115^SRDX line, suggesting gene redundancy at the level of both ERF115 and PAT1. Here, through Y2H and protein affinity purification experiments, we illustrated that both ERF114 and ERF115 display redundancy in their interaction with different members of the PAT1-branch of GRAS transcription factors. Moreover, we demonstrated the combined activity of at least SCL5, SCL21, and PAT1 to be an absolute requirement for regeneration following wounding. Additionally, their ectopic coexpression with either ERF114 or ERF115 resulted in hyperproliferation phenotypes that were distinct depending on the combination, with phenotypes for heterodimers with ERF115 being stronger than those with ERF114. Likewise, combinations with SCL21 yielded a more pronounced hypocotyl-derived callus phenotype, compared with combinations with SCL5 or PAT1. Although it cannot be ruled out that these distinct phenotypes originate from differences in ectopic expression levels or protein stability, a part of it might be attributed to the recognition of different target genes. It has been reported before that the related PSK2 and PSK5 genes are both a target of ERF115, but only the latter is controlled by ERF114 (Kong et al. 2018). Additionally, we have previously reported tissue-specific expression of ERF114 and ERF115, with both showing a similar response following wounding, but with ERF114 and ERF115 being expressed in the lateral root founder cells and underlying maturing protoxylem cells, respectively (Canher et al. 2022), suggesting some subfunctionalization between different ERF-GRAS heterodimers.

Among the GRAS-domain transcription factors, Arabidopsis HAM proteins are important for the maintenance of indeterminate growth (Engstrom 2012; Geng and Zhou 2021), whereas SHR and SCR are indispensable for the specification of the root SCN as well as the radial patterning of the root (Helariutta et al. 2000; Nakajima et al. 2001), with the corresponding mutants displaying outspoken developmental phenotypes. In contrast, the scl5 scl21 pat1 triple mutant appeared identical to the wild type under unwounded control conditions in every examined aspect, consistent with the lack of any differential gene expression between wild-type and mutant plants under control conditions, at least within the root meristem. Contrary, the triple mutant plants were hypersensitive to bleomycin that triggers vascular cell death, resulting in a collapse of the root meristem. This phenotype is most likely due to a regeneration defect rather than an impaired DNA damage response, as both genotypes share a similar transcriptional activation of genes involved in DNA double-strand break repair, oxidative stress, and DNA biosynthesis following bleomycin treatment. Moreover, a similar regeneration failure was observed following root tip excision, as well as a reduced potential to induce periclinal divisions following laser-mediated single-cell ablation. Interestingly, the list of genes being specifically upregulated in the scl5 scl21 pat1 mutant background following bleomycin treatment relates to cell wall organization and biogenesis (Supplemental Data Set S7). Such enrichment likely reflects the loss of division competence following wounding, resulting in cell enlargement rather than cell division. These data thus suggest the specific involvement of these 3 GRAS-domain proteins in maintaining stem cell activity following wounding, although it is still possible that contributions to normal developmental processes in the absence of wounding are masked by the other 2 remaining members of the PAT1 subfamily, SCL1 and SCL13.

DOF3.4 activates periclinal cell division upon wounding

RNA-seq identified the DOF3.4 transcription factor as a molecular component controlling regeneration in an SCL5/SCL21/PAT-dependent manner. Under control conditions, DOF3.4 transcription is hardly detectable in the young root meristem, most frequently occurring in paired cells, indicative of expression correlating with stem cell division. Although a uniform pDOF3.4:NLS-GFP expression pattern could be detected in 2-d-old root meristems, expression dramatically declined in the 5-d-old root, suggesting that the early expression pattern represents a remnant of expression during embryogenesis. Indeed, DOF3.4 is strongly expressed in the embryo proper and suspensor cells (Skirycz et al. 2008; Babu et al. 2013). Therefore, under control conditions, DOF3.4 activity might be restricted to early embryonic development. In contrast, upon wounding, we observed a strong transcriptional activation of DOF3.4. Calling upon the embryonic role of DOF3.4 during regeneration supports the model put forward based on single-cell transcriptome analysis, demonstrating that stem cell reformation following root tip excision is preceded by rapid cell identity transitions and developmental events that closely resemble that of embryonic root formation (Efroni et al. 2016).

Differently from the DOF3.4^SRDX line, no clear developmental or regeneration-related phenotypes could be observed for an available T-DNA insertion line, suggesting putative redundancy with other DOF transcription factors. Accordingly, DOF family members have been linked with proliferation and regeneration before, with DOF5.4 promoting callus formation (Ramirez-Parra et al. 2017), whereas HIGH CAMBIAL ACTIVITY2 controls grafting dynamics, likely by promoting cell division and cambium formation to enhance wound healing (Konishi and Yanagisawa 2007; Guo et al. 2009; Gardiner et al. 2010; Melnyk et al. 2018). The latter has recently been demonstrated to operate redundantly with the related TARGET OF MONOPTEROS 6 (TMO6) and DOF2.1 transcription factors, whereas DOF3.4 expression was found as well to be rapidly and strongly induced following grafting (Zhang et al. 2022). We, therefore, speculate that the DOF3.4^SRDX protein targets other DOF transcription factors that might work together with DOF3.4, although no such other DOF gene was found to be significantly differentially expressed in control versus scl5 scl21 pat1 mutant plants following bleomycin treatment.

DOF3.4 holds the potential to drive periclinal cell divisions, as illustrated by the cell division phenotypes in the vascular bundle, ground tissues, and epidermis following its overexpression. This potential is shared with different DOF factors that have been demonstrated to redundantly control periclinal cell divisions in vascular tissues (Papi et al. 2002; Ward et al. 2005; Fornara et al. 2009; Rueda-Romero et al. 2012; Miyashima et al. 2019; Smet et al. 2019). However, differently from these DOF genes, we failed to detect any DOF3.4 expression in the vascular cells of seedlings under unstressed conditions. It indicates that DOF3.4's post-embryonic function might be restricted to regeneration.

Previously, the CYCD3;3 gene was demonstrated to be a direct DOF3.4 target (Skirycz et al. 2008). Accordingly, excised cycd3;3 root tips displayed a regeneration phenotype, being aggravated in the cycd3;1 cycd3;2 cycd3;3 triple mutant. Likewise, the mutant is less efficient in wound-induced callus formation following a hypocotyl cut, compared with wild-type plants (Ikeuchi et al. 2017). These data suggest that D3-type cyclins might be among the most important targets of DOF3.4 in the regeneration process. In wild-type plants, CYCD3;3 expression can be found in the vascular initials and most distal columella cells. The latter might again be a remnant of CYCD3;3 requirement for division activity during the establishment of the embryonic columella (Forzani et al. 2014). Upon inducible DOF3.4^GR expression, the CYCD3;3 expression domain expanded to the columella stem cells within 48 h of DEX induction, cells in which DOF3.4 expression could also be observed following bleomycin treatment. Thus, activation of CYCD3;3 by DOF3.4 might explain the extra layers of columella stem cells observed in the DOF3.4^GR lines.

An enhanced stem cell division rate is a feature shared with RBR1-deficient plants (Wildwater et al. 2005; Wachsman et al. 2011; Weimer et al. 2012; Cruz-Ramírez et al. 2013), indicating that DOF3.4 might operate through inactivation of RBR1, facilitated by phosphorylation by D-type cyclin–CDK complexes (Nakagami et al. 1999). Seeing the phenotypic resemblance between DOF3.4^GR and RBR1 RNAi lines, it is likely that DOF3.4 converges on the RBR1 pathway through the transcriptional activation of CYCD3;3 (Wildwater et al. 2005). Another interesting feature of CYCD3s is that they act downstream of cytokinin signaling (Dewitte et al. 2007). Correspondingly, endogenous cytokinin levels have been found to rise after wounding of etiolated hypocotyls (Ikeuchi et al. 2017). However, whether the DOF3.4-activated CYCD3 response following wounding in the root meristem is achieved through the cytokinin pathway, rather than being downstream of a transcriptional cascade instigated by ERF115, still needs to be demonstrated.

In addition to DOF3.4, the auxin-responsive INDOLE-3-ACETIC ACID INDUCIBLE 5 (IAA5) gene was found among the strongest SCL5/SLC21/PAT1-responsive genes. Accordingly, IAA5 expression is wound responsive (Asahina et al. 2011; Matsuoka et al. 2016; Ikeuchi et al. 2017). IAA5 is a canonical AUX/IAA protein that inhibits distal stem cell differentiation by repressing the transcriptional auxin response (Lv et al. 2020), indicative of a role in acquiring or maintaining stem cell regeneration during the regeneration process. However, further experiments are needed to verify this claim.

ERF-GRAS heterodimerization drives cell division via DOF3.4

Next to a strong reduction in the transcriptional activation of DOF3.4 following bleomycin treatment of the scl5 scl21 pat1 triple mutant, its expression was strongly repressed in the ERF115^SRDX background. These data suggest that it is most likely a dimer between ERF115 and the PAT1-type GRAS-domain transcription factors that controls DOF3.4 expression. However, the relative contribution of both subunits in their transcriptional activation still needs to be clarified. Whereas ERF proteins directly bind to DNA using their AP2 domain (Ohme-Takagi and Shinshi 1995), GRAS-domain proteins have been suggested to be transcriptional cofactors that interact with other transcription factors (Hirano et al. 2017) or histone deacetylases (Gao et al. 2004, 2015) to indirectly bind to DNA. Accordingly, we failed to detect direct binding of GRAS proteins to the DOF3.4 promoter. Contrary, previously published tandem chromatin affinity purification data revealed a putative binding site for ERF115 on the DOF3.4 promoter region (Heyman et al. 2013). It is, therefore, tempting to speculate that the ERF transcription factor provides target gene specificity, whereas the GRAS interaction partner controls the regulation of its transcription by recruiting histone-modifying enzymes. However, indirect control of expression by ERF115 cannot be excluded, since DOF3.4 has been recently described as an auxin-induced gene being under putative transcriptional control of MP/ARF5 (Larrieu et al. 2022). On the other hand, MP/ARF5 was reported to be a bona fide direct ERF115 target gene (Canher et al. 2020), in that way still placing DOF3.4 downstream of the wound-induced ERF115-dependent transcriptional pathway.

Taken together, our data pinpoint DOF3.4 as an important target operating downstream of at least ERF115 in complex with SCL5, SCL21, or PAT1. Whereas the previously identified ERF115 targets PSK2/PSK5, WIND1, and MP/ARF5 play a role in promoting cell proliferation, cellular reprograming, and inducting stem cell fate, respectively, DOF3.4 appears to play an important role in driving periclinal divisions needed for tissue repair through transcriptional induction of D3-type cyclins (Fig. 8). Additionally, our work suggests the presence of a multitude of putative ERF-GRAS heterodimeric combinations that may play different roles in initiating the regeneration program, probably underscoring the plasticity of tissue recovery following damage.

Figure 8

Model for ERF114/ERF115-mediated regeneration. Wound-induced activation of ERF114 and ERF115 would result in the heterodimerization with the SCL5/SCL21/PAT1 transcription factors to activate the expression of PSK2/5, WIND1, ARF5/MP, and DOF3.4 that promote cell proliferation, cellular reprograming, stem cell identity, and periclinal cell division, respectively. DOF3.4 activates cell division through transcriptional activation of CYDD3;3.

Materials and methods

Plant materials and growth conditions

A. thaliana Col-0 seeds were sterilized in 70% (v/v) ethanol for 10 to 15 min and subsequently washed with 100% ethanol, after which they were left to dry in sterile conditions. For all experiments, the seeds were stratified in the dark for 2 d at 4 °C before being placed in the growth room. Plants were grown in vitro under long-day conditions (16-h light/8-h dark, Lumilux Cool White lm, 50 to 70 µmol/m² s¹) at 21 °C on solidified half-strength Murashige and Skoog (MS) medium (2.151 g/L, 10 g/L sucrose, and 0.5 g/L 2-(N-morpholino) ethanesulfonic acid (MES), adjusted to pH 5.7 with 1 M KOH and 8 or 10 g/L plant agar). DEX (Sigma) treatments were performed by either germinating seeds on 10-μM DEX-supplemented medium or by transferring plants from ½ MS to 10-μM DEX-supplemented medium.

Constructs and isolation of transgenic lines

Primers used for cloning and genotyping can be found in Supplemental Table S1. ERF114^OE, ERF115^OE, PAT1^OE, ERF115^SRDX, pat1, erf115, pWOX5:NLS-GFP/GUS, pERF115:NLS-TdTomato, and pERF114:NLS-GFP/GUS have been described previously (Heyman et al. 2016; Canher et al. 2022). The scl21 (SALK_146085), scl5 (SALK_152973), and dof3.4 (SALK_049540) mutants were obtained from the Salk Institute T-DNA Express database (Alonso et al. 2003). The cycd3;3 single and cycd3;1 cycd3;2 cycd3;3 triple mutants were described previously (Dewitte et al. 2007). DOF3.4^OE, SCL5^OE, and SCL21^OE constructs were generated by cloning their wild-type ORF including the stop codon into pDONR221 (Invitrogen) via a BP reaction and subsequently recombining it via an LR reaction behind the strong Cauliflower Mosaic Virus (CaMV) 35S promoter in the pB7WG2 vector for DOF3.4 and SCL5 and the pH7WG2 vector for SCL21 (Karimi et al. 2002). For GFP reporter constructs, the promoter region immediately upstream of the start codon of DOF3.4 (1,642 bp) and SCL5 (2,525 bp) was inserted into the pMK7S*NFm14GW plasmid (Karimi et al. 2002). The translational reporter lines pSCL5:SCL5-GFP and pPAT1:PAT1-GFP were obtained by cloning the SCL5 and PAT1 genomic fragments containing the promoter and coding region without stop codons into pDONRP4P1r (4,626 and 5,267 bp, respectively), followed by fusion of the GFP reporter without the nuclear localization signal (NLS) to the C terminus of each coding sequence; finally, the SCL5 and PAT1 genomic fragment–GFP fusions were recombined into the destination vectors pHm42GW and pBm42GW (Karimi et al. 2002), respectively. The pCYCD3;3:GFP/GUS reporter line was generated through Gateway cloning of the 2,199-bp sequence preceding the translation initiation codon into pKGWFS7 as described previously (Benhamed et al. 2008). The pDOF3.4:NLS-GFP/GUS reporter was introduced in the ERF115^SRDX background via crossing.

GRAS triple mutants harboring the pDOF3.4:NLS-GFP/GUS reporter were generated by transforming the pFASTRK-II-m43GW destination vector carrying the DOF3.4 promoter upstream of the NLS-GFP and GUS reporter sequences into the scl5 scl21 pat1 triple mutant. Similarly, the pWOX5:NLS-GFP/GUS reporter construct (Heyman et al. 2016) was introduced in the GRAS triple mutant and transformants were selected based on the presence of WOX5-GFP-derived fluorescence. DOF3.4^GR was generated by recombining the DOF3.4 coding sequence without stop codon before the GLUCOCORTICOID RECEPTOR coding sequence under the control of the CaMV 35S promoter in the pFASTHm43GW destination vector.

The DOF3.4^SRDX construct was generated by fusing the DOF3.4 coding region without stop codon with the SRDX domain (LDLDLELRLGFA) and recombined into the pH7GW2 expression vector under the control of the CaMV 35S promoter via Gateway recombination.

All constructs were transferred into the Agrobacterium tumefaciens C58C1RifR strain harboring the pMP90 plasmid. The obtained Agrobacterium strains were used to generate stably transformed Arabidopsis Col-0 by the floral dip transformation method (Clough and Bent 1998). Higher order mutants were generated by crossing.

Yeast two-hybrid assay

The respective ERF and GRAS cDNA clones were recombined into the pDEST32 and pDEST22 plasmids using Gateway recombination, generating the yeast bait and prey constructs, respectively. Primer sequences used for cDNA clone and N-terminal ERF fragment isolation can be found in Supplemental Table S1 or have been described previously (Heyman et al. 2016). Plasmids encoding the bait and prey were transformed into the yeast strain PJ69-4α (MATα; trp1-901, leu2-3,112, ura3-52, his3-200, gal4Δ, gal80Δ, LYS2::GAL1-HIS3, GAL2-ADE2, met2GAL7-lacZ) and PJ69-4a (MATa; trp1-901, leu2-3,112, ura3-52, his3-200, gal4Δ, gal80Δ, LYS2TGAL1-HIS3, GAL2-ADE2, met2TGAL7-lacZ), respectively, by the lithium acetate (LiAc) method (Gietz et al. 1992). Transformed yeast cells were selected on synthetic dextrose plates without Leu (pDEST32) or without Trp (pDEST22). Interactions between proteins were assayed by the mating method. All pDEST32 yeast cultures were inoculated in 200 μL synthetic dextrose without Trp in a 96-well microtiter plate (Falcon; BD Biosciences), whereas one pDEST22 yeast culture was inoculated in 50 mL synthetic dextrose medium without Leu. To scale up the yeast cultures, 20 μL of each culture grown for 2 d at 30 °C was added to a microtiter plate containing 125 μL of Glc-containing rich medium (10 g/L bacto-yeast extract, 10 g/L bacto-peptone, and 20% dextrose) and again grown for 24 h at 30 °C. The Glc-containing rich medium was replaced by synthetic dextrose medium without Leu and Trp. Diploid strains grown in a 96-well microtiter plate (NUNC) for 2 d at 30 °C were diluted 1/10 prior to spotting on agar-solidified synthetic dextrose medium without Leu and Trp but with His (as a control), or synthetic dextrose medium without Leu, Trp, or His, in the presence of 3-amino-1,2,4-triazole (3-AT). Growth was observed after 3 d of incubation at 30 °C. For the interaction test between the newly annotated SCL5 and the ERF A-fragments, the respective pDEST22 and pDEST32 plasmids were cotransformed in the PJ69-4a strain and plated on agar-solidified synthetic dextrose medium as described above.

Root sectioning

For Technovit sectioning, 6-d-old seedlings were fixed in a paraformaldehyde (4%)-glutaraldehyde (1%) solution, dehydrated through a graded ethanol series, and embedded in Technovit 7100 (Heraeus Kulzer). Sections of 4-μm thickness were prepared and counterstained as described in De Smet et al. (2004).

Confocal and light microscopy

Arabidopsis roots were stained using propidium iodide (PI) by incubation in a 10-μM solution for 2 min before imaging. For confocal microscopy, Zeiss LSM5 and LSM710 confocal microscopes were used. GFP fluorescence was observed after excitation using a 488-nm laser. PI fluorescence was observed after excitation using a 543-nm laser. GFP and tdTomato emissions were collected at 500 to 530 nm and 570 to 630 nm, respectively. For light and differential interference contrast microscopy, an Olympus BX51 microscope was used. Pseudo-Schiff PI staining was performed as described (Canher et al. 2020).

Root cell death and recovery experiments

For bleomycin treatment, 5-d-old seedlings were transferred to vertical plates supplemented with 0.6 μg/mL bleomycin. For recovery, seedlings were retransferred to control medium after 24 h of growth on bleomycin-containing medium and allowed to recover for 5 d. For the bleomycin recovery percentage, unless stated otherwise, plants recovered for 5 d from the bleomycin treatment were imaged under the confocal microscope and scored as regenerated versus non-regenerated based on meristem organization, after staining the roots with PI solution. Root tip excisions were performed as described previously (Sena et al. 2009). Root tip regeneration 72 h after excision was scored negative when a collapse of the root meristem was observed. For laser ablation, an LSM710 confocal microscope with a 405-nm laser was used at 70% laser power for 5 to 10 s. For periclinal division quantification, epidermal and cortical cells were ablated and periclinal divisions were assessed in the cortex and endodermis, respectively, 16-h post-ablation. The percentage of periclinal divisions was scored as previously described (Marhava et al. 2019).

Reverse transcription-quantitative PCR

RNA was isolated from seedlings with the RNeasy Plant Mini kit (Qiagen). DNase treatment was performed using the RQ1 RNase-Free DNase (Promega) prior to cDNA synthesis with the iScript cDNA Synthesis kit (Bio-Rad). Expression levels were determined by RT-qPCR with the LightCycler 480 Real-Time SYBR Green PCR System (Roche). EMBRYO DEFECTIVE 2386 (EMB2386) and 20S PROTEASOME ALPHA SUBUNIT C1 (PAC1) were used as reference genes for normalization. Primer sequences used for RT-qPCR are listed in Supplemental Table S1.

RNA-sequencing

Five days after germination, seedlings were transferred on mesh to either fresh control medium or medium supplemented with 0.9 μg/mL bleomycin for 24 h. Root tips (approximately 2 mm) were collected in liquid nitrogen in 3 biological repeats. At least 100 roots were used per biological repeat. RNA was extracted from samples using the RNeasy Plant Mini kit (QIAGEN) according to the manufacturer's protocols. A total of 1 μg of RNA per sample was used as input material for the RNA sample preparations. Sequencing libraries were generated using NEBNext UltraTM RNALibrary Prep Kit for Illumina (NEB, USA) following the manufacturer's recommendations and index codes were added to attribute sequences to each sample. Samples were sequenced on an Illumina NovaSeq6000 instrument and 150-bp paired-end reads were generated. Transcript quantification was performed by Salmon version 0.9.1 (Patro et al. 2017), using the Araport11 list of all coding sequences (Cheng et al. 2017). The option to correct for a sequence-specific bias was turned off and all parameters were set to their default values. Transcript quantification results generated by Salmon were corrected for gene length variations across the samples using the tximport 1.6.0 R package (Soneson et al. 2015) and were further analyzed using the edgeR package (Robinson et al. 2010). Transcripts with a read count of less than 10 in all samples combined were removed from further analyses. Normalized read counts were log2-transformed, and a quasi-likelihood negative binomial generalized linear model was fitted to the data using genotype, treatment, and the genotype × treatment interaction term as factors to identify differential gene expression. Gene Ontology (GO) term enrichment analysis was performed using PANTHER (http://pantherdb.org).

Single-step immunoprecipitation and mass spectrometry

Three pull-down experiments on Arabidopsis cell suspension cultures expressing N-terminally GS^TEV-tagged ERF115 were performed as described (Van Leene et al. 2019). On-bead digested samples were analyzed on a Q Exactive HF (ThermoFisher Scientific) and co-purified proteins were identified using standard procedures (Van Leene et al. 2015). After identification, the obtained protein list was filtered against a list of nonspecific proteins, assembled similarly as described (Van Leene et al. 2015). True interactors that might have been filtered out because of their presence in the list of nonspecific proteins were retained by means of quantitative analysis using the average normalized spectral abundance factors (NSAF) of the identified proteins in the ERF115 pull-downs. Proteins identified with at least 2 peptides in at least 2 experiments, showing high (at least tenfold) and significant [−log₁₀(P-value(T-test)) ≥ 10] enrichment compared with calculated average NSAF values from a large data set of pull-downs with non-related bait proteins, were retained.

Bimolecular fluorescence complementation

The newly annotated SCL5 ORF was tagged at the C-terminal end with the GFP tail fragment and inserted into the pK7m34GW vector under the control of the CaMV 35S constitutive promoter using Gateway cloning. The ERF115 and SCL21 constructs were described previously (Heyman et al. 2016). For N. benthamiana leaf blade infiltration, equal concentrations (final A₆₀₀ = 0.5) of a 2 d-old liquid-grown Agrobacterium tumefaciens culture containing the respective constructs were incubated in infiltration buffer (10 mM MgCl₂, 10 mM MES pH 5.6, 0.1 mM acetosyringone) for 2 h before infiltration of 6-wk-old plants. Plants were allowed to grow for another 2 d before observing the interaction by means of confocal microscopy. As a negative control, single constructs were infiltrated.

Statistical analysis

Statistical analyses were performed as indicated in the figure legends. Statistical data can be found in Supplemental Data Set S8.

Accession numbers

RNA-seq raw data from this study were deposited in Gene Expression Omnibus (accession number: GSE200500). Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: ERF114 (At5g61890); ERF115 (At5g07310); PAT1 (At5g48150); SCL5 (At1g50600); SCL21 (At2g04890); DOF3.4 (At3g50410); CYCD3; 3 (At5g50050), EMB2386 (At1g02780); PAC1 (At3g22110).

Acknowledgments

The authors thank Annick Bleys for critical reading of the manuscript, and Lukas Hoermayer and Jiří Friml for training in single-cell ablation. This work was supported by grants from the Research Foundation Flanders (G007218N and G010820N). R.L. was supported by the China Scholarship Council (201704910848).

Author contributions

A.B., J.H., and L.D.V. conceived and designed the research. A.B., J.H., B.C., R.L., and I.V. performed the experiments. A.B., J.H., T.E, G.D.J., and L.D.V analyzed data. A.B., J.H., and L.D.V. wrote the article. All authors read, revised, and approved the article.

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Corrected annotation of the SCL5 promoter and coding sequence.

Supplemental Figure S2. ERF115 associates with SCL5 in planta.

Supplemental Figure S3. Fluorescence intensity overlap between the ERF115 and SCL5 transcriptional reporters following bleomycin treatment.

Supplemental Figure S4. Co-overexpression of ERF114/ERF115 and SCL5/SCL21/PAT1 inhibits root growth. Root length of 12 d-old Col-0, ERF114, ERF115, SCL5, SCL21, and PAT1 single and ERF114-SCL5, ERF114-SCL21, ERF114-PAT1 ERF115-SCL5, ERF115-SCL21, and ERF115-PAT1 co-overexpressing seedlings.

Supplemental Figure S5. Root length of ERF115, PAT1, SCL5, SCL21 and respective co-overexpressing lines.

Supplemental Figure S6. Growth phenotypes of ERF114, PAT1, SCL5, SCL21 and respective co-overexpressing lines.

Supplemental Figure S7. Rosettes of 25 d-old Col-0, ERF114, ERF115, SCL5, SCL21, PAT1 overexpressing lines and their respective combinations.

Supplemental Figure S8. GRAS higher order mutants do not show a meristem phenotype under control conditions.

Supplemental Figure S9. DOF3.4 expression in the root meristem.

Supplemental Figure S10. Binding of ERF115 to the DOF3.4 promoter.

Supplemental Figure S11. DOF3.4^SRDX seedlings display a reduced regeneration efficiency following root tip excision.

Supplemental Figure S12. Activation of DOF3.4^GR induces periclinal cell divisions.

Supplemental Table S1. Primer sequences.

Supplemental Data Set S1. Protein identification details obtained with the Q Exactive HF (Thermo Fisher Scientific) and Mascot Distiller software (version 2.5.0.0, Matrix Science) combined with the Mascot search engine (version 2.5.1, Matrix Science) using the Mascot Daemon interface and database TAIRplus (Van Leene et al. 2015).

Supplemental Data Set S2. List of genes differentially expressed (FDR < 0.05) in wild-type root tips (2 mm) following bleomycin treatment.

Supplemental Data Set S3. List of genes differentially expressed (FDR < 0.05) in scl5 scl21 pat1 root tips (2 mm) following bleomycin treatment.

Supplemental Data Set S4. List of genes differentially upregulated (FDR < 0.05) in both wild-type and scl5 scl21 pat1 root tips (2 mm) following bleomycin treatment.

Supplemental Data Set S5. List of genes differentially upregulated (FDR < 0.05) in wild-type root tips only following bleomycin treatment.

Supplemental Data Set S6. List of genes differentially upregulated (FDR < 0.05) in scl5 scl21 pat1 root tips only following bleomycin treatment.

Supplemental Data Set S7. GO enrichment analysis of the genes activated following bleomycin treatment in scl5 scl21 pat1 root tips only.

Supplemental Data Set S8. Statistical data.

Data availability

All raw data are available upon request.

Author notes

Senior author.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plcell/) is: Lieven De Veylder ([email protected]).

Conflict of interest statement. The authors declare that they have no conflicts of interest.