WIND transcription factors orchestrate wound-induced callus formation, vascular reconnection and defense response in Arabidopsis.

Wounding triggers de novo organogenesis, vascular reconnection and defense response but how wound stress evoke such a diverse array of physiological responses remains unknown. We previously identified AP2/ERF transcription factors, WOUND INDUCED DEDIFFERENTIATION1 (WIND1) and its homologs, WIND2, WIND3 and WIND4, as key regulators of wound-induced cellular reprogramming in Arabidopsis. To understand how WIND transcription factors promote downstream events, we performed time-course transcriptome analyses after WIND1 induction. We observed a significant overlap between WIND1-induced genes and genes implicated in cellular reprogramming, vascular formation and pathogen response. We demonstrated that WIND transcription factors induce several reprogramming genes to promote callus formation at wound sites. We, in addition, showed that WIND transcription factors promote tracheary element formation, vascular reconnection and resistance to Pseudomonas syringae pv. tomato DC3000. These results indicate that WIND transcription factors function as key regulators of wound-induced responses by promoting dynamic transcriptional alterations. This study provides deeper mechanistic insights into how plants control multiple physiological responses after wounding.

Introduction

Wounding is a serious threat to the plant survival and it triggers multiple physiological responses to quickly heal and protect damaged tissues from pathogen invasion (Reymond et al., 2000; Cheong et al., 2002). In plants, formation of a pluripotent cell mass, called callus, at wound sites is often a key step to regenerate new organs and develop physical and chemical barriers against pathogens (Birnbaum & Alvarado, 2008; Asahina et al., 2011; Ikeuchi et al., 2013, 2016; Melnyk, 2017). Importantly, wounding and callus formation is often accompanied by vascular reformation presumably to establish the route for water and nutrient transport in developing cell mass (Fukuda, 1997; Mazur et al., 2016). Accordingly, earlier studies reported ectopic tracheary element formation in the genetic tumor of Nicotiana tabacum callus (White, 1939) and crown galls (Van Lith‐Vroom et al., 1960). It is also known that grafted plants initially form callus at wound sites, followed by vascular bundle reformation within callus (Melnyk et al., 2015; Melnyk, 2017). Surface regeneration of debarked tree trunk is another well‐characterized regeneration phenomenon after wounding where xylem and phloem reformation occur after callus formation (Stobbe et al., 2002). Although we have made considerable progress in our understanding of how plants perceive wounding signals (Toyota et al., 2018; Ikeuchi et al., 2019, 2020; Marhava et al., 2019), our knowledge on how plants initiate such a diverse array of wound‐induced responses is still very limited (Bloch, 1941; Walker‐Simmons et al., 1984; Savatin et al., 2014).

Given that these wound‐induced events require dynamic changes in gene expression, it is likely that plants possess some transcriptional mechanisms to coordinate their progression. Recent studies have indeed identified several wound‐inducible transcription factors that have critical roles in regeneration (Ikeuchi et al., 2013, 2016, 2019; Xu & Huang, 2014). We previously reported that an APETALA2/ETHYLANE RESPONSE FACTOR (AP2/ERF) transcription factor WOUND INDUCED DEDIFFERENTIATION 1 (WIND1) and its close homologs, WIND2, WIND3 and WIND4, promote wound‐induced callus formation through activating the cytokinin response (Iwase et al., 2011a,b, 2013, 2015). WIND1 also promotes shoot regeneration via direct activation of another AP2/ERF transcription factor ENHANCER OF SHOOT REGENERATION 1 (ESR1) (Iwase et al., 2017, 2018). WIND induction, in addition, leads to somatic embryogenesis on phytohormone‐free medium (Ikeuchi et al., 2013), implying that WIND1 can drive multiple developmental pathways to promote regeneration. At the cellular level, WIND1 promotes the acquisition of regenerative competence since ectopic overexpression of WIND1 can bypass wounding and early incubation steps on auxin‐rich callus inducing medium (CIM), which are the prerequisite for shoot regeneration on cytokinin‐rich shoot inducing medium (SIM) (Valvekens et al., 1988; Iwase et al., 2015). Several other AP2/ERF family transcription factors are also implicated in the control of regeneration since PLETHORA3 (PLT3), PLT5 and PLT7, are critical for wound‐induced callus formation and pluripotency acquisition under CIM/SIM condition (Kareem et al., 2015; Ikeuchi et al., 2017, 2020). Another wound‐inducible AP2/ERF protein ETHYLENE RESPONSE FACTOR 115 (ERF115), acting upstream of WIND1, is required for reformation of root stem cells and regeneration of root meristems after injury (Heyman et al., 2013, 2016; Marhava et al., 2019; Zhou et al., 2019). ERF113/RELATED TO AP2 6 LIKE (RAP2.6L), which is a close homolog of ERF115, is reported as a key regulator of tissue reconnection process (Asahina et al., 2011) as well as for shoot regeneration under CIM/SIM condition (Che et al., 2006).

Several important transcriptional regulators of plant vascular development have also been identified (Kondo, 2018) and for instance, VASCULAR‐RELATED NAC‐DOMAIN6 (VND6) and VND7 transcription factors function as master regulators for the formation of vascular vessels (Kubo et al., 2005). Overexpression of VND6 or VND7 provokes ectopic tracheary element formation in diverse cell types (Kubo et al., 2005). LATERAL ORGAN BOUNDARIES DOMAIN 30 (LBD30), a putative positive feedback regulator for VND6 and VND7, also shows similar ectopic tracheary formation when overexpressed in Arabidopsis thaliana (Arabidopsis) (Soyano et al., 2008). Other Arabidopsis NAC domain transcription factors ANAC071 and ANAC091 are required in tissue reconnection and conversion of mesophyll cell fate to cambial cells (Asahina et al., 2011; Matsuoka et al., 2021). Recent studies using the in vivo and in vitro culturing system have started to unveil further transcriptional regulatory networks driving the vascular development (Kondo et al., 2016; Miyashima et al., 2019). When Arabidopsis leaf tissues are incubated under the Vascular Cell Induction Culture System Using Arabidopsis Leaves (VISUAL), leaf mesophyll cells reprogram into vascular cells and start to express cambium cell marker genes such as TDIF RECEPTOR (TDR) and Arabidopsis thaliana HOMEOBOX GENE8 (AtHB8) (Kondo et al., 2016). This is followed by the upregulation of xylem marker genes, such as IRREGULAR XYLEM3 (IRX3), and phloem marker genes such as SIEVE‐ELEMENT‐OCCLUSION‐RELATED1 (SEOR1). Despite these progresses, whether these key regulators contribute to xylem formation after wounding and if so, how wounding activates these regulators remain unknown.

Hierarchal transcriptional networks acting from pathogen perception to the immune responses (Cui et al., 2015) are well characterized and many WRKY transcription factors are known to play major roles in defense signaling (Eulgem & Somssich, 2007). WRKY18 and WRKY53, for instance, positively regulate defense responses and these regulators induce genes for key enzymes in biosynthesis of phytoalexins, the antimicrobial secondary metabolites (Wang et al., 2006; Murray et al., 2007). Camalexin, a well‐known phytoalexin in Arabidopsis, is synthesized de novo after various biotic and abiotic stress including pathogen infection (Ahuja et al., 2012). P450 monooxygenases, Cytochrome P450 71B15 (CYP71B15/PAD3) and CYP71A13, are involved in camalexin biosynthesis and they are induced after pathogen infection in a WRKY33‐dependent manner (Qiu et al., 2008). A single knock‐out mutation of these P450 monooxygenases enhances disease susceptibility against bacterial pathogen infection (Rajniak et al., 2015).

Lysine‐derived pipecolic acid is a critical regulator for an establishment of systemic acquired resistance in Arabidopsis upon pathogen infection, and is synthesized via AGD2‐LIKE DEFENSE REPONSE PROTEIN 1 (ALD1) (Hartmann et al., 2018). ALD1 also shows WRKY33‐dependent expression manner, and importantly, plants defective in this gene show higher susceptibility to pathogens (Song et al., 2004; Návarová et al., 2012; Wang et al., 2018), indicating that the control of WRKY‐mediated phytoalexin and signal molecule production are crucial for the defense response.

Interestingly, ERF108/RELATED TO AP2 6 (RAP2.6), a close homolog of ERF115 and RAP2.6L in the subfamily X of AP2/ERF transcription factors, may function in the defense response since its expression is strongly induced after challenged with a virulent Pseudomonas syringae pv. tomato DC3000 (Pst DC3000). ERF108/RAP2.6 also promotes resistance against cyst nematode infection though the enhancement of callose deposition in Arabidopsis (He et al., 2004; Ali et al., 2013). These findings highlight the multifaceted roles of AP2/ERF transcription factors in tissue repair and defense responses (Heyman et al., 2018) but how their stress‐induced expression is regulated remains to be elucidated.

Several recent transcriptome studies revealed that the wound‐induced transcriptional changes highly overlap with those elicited by various biotic and abiotic stresses (Cheong et al., 2002; Ikeuchi et al., 2017; Melnyk et al., 2018), suggesting the existence of common regulators that function in multiple stress responses. Given that callus produced by constitutive WIND1 expression shows increased expression of some vascular genes and defense response genes (Iwase et al., 2011a), it is plausible that WIND1 and its homologs play diverse roles in response to wounding and other forms of stress. In this study we conducted the time‐course transcriptome analyses after WIND1 induction to explore how WIND1 functions in stress response. Our data show that WIND1 transcriptionally activates over 2000 genes implicated in cellular reprogramming, vascular formation and defense response. Further functional analyses confirmed that WIND transcription factors have important roles during wound‐induced cellular reprogramming, vascular regeneration and defense response. Our results, therefore, provide important molecular insights into how plants coordinately control regeneration and innate immunity through WIND‐mediated transcriptional mechanisms.

Materials and Methods

Plant materials, growth condition, and transformation

All Arabidopsis plants used in this study were in the Col‐0 background. XVE‐WIND1, WIND1pro:WIND1‐SRDX, wind1 (SALK_020767 and SALK_027272), wind2 (SALK_139727), wind3 (SALK_091212), wind4 (SALK_099481), wind1 wind2 wind3 wind4 quadruple mutant (generated by crossing wind single mutants), rap2.6l‐1 (SALK_051006), rap2.6‐1 (GK_053G11.01), rap2.6‐3 (SAIL_1225_G09), 35S:RAP2.6‐1, 35S:RAP2.6‐2, erf115 (SALK_021981), ERF115pro:ERF115‐SRDX, plt3 plt5 plt7 triple mutant, wox5‐1 (SALK_038262), and anac071 anac096 anac011 were described previously (Che et al., 2006; Sarkar et al., 2007; Iwase et al., 2011a; Ali et al., 2013; Heyman et al., 2013; Kareem et al., 2015; Ikeuchi et al., 2017; Matsuoka et al., 2018; Matsuoka et al., 2021). T‐DNA insertion lines were obtained from the Arabidopsis Biological Resource Center (ABRC). Plants were grown on 0.6% (w/v) gelzan plates containing Murashige & Skoog (MS) salt and 1% sucrose medium at 22°C with a photoperiod of 16 h white light and 8 h darkness, unless noted otherwise. For plant transformation, T‐DNA vectors carrying an appropriate construct were introduced into Agrobacterium tumefaciens strain GV3101 by electroporation, and Arabidopsis plants were transformed by the floral dip method (Clough & Bent, 1998). For WIND1 induction, 10‐d‐old wild‐type (WT) plants and XVE‐WIND1 plants were grown on MS plates, which were treated with MS liquid medium containing either mock control or 10 µM 17‐β‐estradiol (ED; Sigma‐Aldrich, St Louis, MO, USA).

Plasmid construction

To construct the Pro and Pro vectors, genomic fragments containing the 2000‐bp promoter sequence and coding sequence were amplified by PCR and cloned between SpeI and SmaI sites of pGFP_NOSG (Iwase et al., 2017) and pSRDX_NOSG vectors (Yoshida et al., 2013), respectively. The resulting Pro and Pro fragments were subcloned into the pBCKH vector (Mitsuda et al., 2006) by Gateway LR Clonase II (Thermo Fisher Scientific, Waltham, MA, USA) and used for plant transformation. To construct the Pro:L‐LUC reporter vectors, the 3000‐bp promoter sequence of each gene (At1g05100, At1g09950, At3g50260, At4g28140, At4g38400 and At1g02460) was amplified by PCR and cloned between ApaI or SacII and NotI sites of the ProESR1:LUC vector (Iwase et al., 2017). A list of primers used for PCR amplification is provided in Supporting Information Table S5.

Transient expression assay

The Pro (Iwase et al., 2011a) and Pro (Ohta et al., 2001) vectors were used as an effector and control, respectively. The Pro:L‐LUC vectors were used as reporters, and the pPTRL vector which drives the expression of a luciferase gene from Renilla (R‐LUC) by the 35S promoter (Fujimoto et al., 2000), was used as an internal control. Particle bombardment and luciferase assays were performed as described previously (Iwase et al., 2017).

RNA isolation, microarray and RT‐qPCR analyses

Total RNA was isolated with RNeasy Plant Mini Kit (Qiagen, Vemlo, the Netherlands) according to manufacturer’s instruction. Microarray experiments were performed using Agilent Arabidopsis (V3) (4x44k) microarray according to the manufactures’ instruction. Four biological replicates were tested in one‐color method. Spot signal values were calculated by feature extraction v.9.1 software supplied by Agilent Technologies (Santa Clara, CA, USA). The quality control (QC) values were defined as one when a spot passes the ‘FeatNonUnifOL’ filter and as two when the spot further passes the ‘FeatPopnOL’ filter. Detection values were defined as one when a spot passes the ‘IsPosAndSignif’ filter and as two when the spot further passes the ‘IsWellAboveBG’. Global normalization among different replicates and experiments were performed with quantile normalization method (Amaratunga & Cabrera, 2001) by using > 100 our in‐house same‐platform data followed by division of each value by the median value among spots with QC = 2. Spot‐to‐gene conversion was accomplished based on a table provided by TAIR (ftp://ftp.arabidopsis.org/home/tair/Microarrays/Agilent/agilent_array_elements‐2009‐7‐29.txt). The average values were used for the genes corresponding to two or more probes. Genes with average QC value < 1.5 in the ‘test’ sample or the ‘reference’ sample were excluded from further analyses, and only genes with average detection value ≥ 1.5 in the ‘test’ or ‘reference’ sample were analyzed when selecting for upregulated or downregulated genes. The P values of each gene were calculated by Welch’s t‐test. To estimate the false discovery rate (FDR), Q‐value was calculated from P value, using the Qvalue software (default settings; (Storey & Tibshirani, 2003)), and upregulated or downregulated (> 2‐fold or < 0.5‐fold) genes were selected with P value less than 0.05 (FDR was less than 0.05 in each case). Binomial test was performed by R (http://www.r‐project.org/). Hierarchical clustering analysis was performed using the cluster 3 software, using default settings (Eisen et al., 1998; de Hoon et al., 2004). Reverse transcription quantitative polymerase chain reaction (RT‐qPCR) was performed as described previously (Mitsuda et al., 2005). Mean expression levels were normalized against the PROTEIN PHOSPHATASE 2A SUBUNIT A3 (PP2AA3) gene. The transcriptome data were deposited in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus as GSE167174 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE167174;token=slwleqkextstbon).

Callus formation and vascular induction assay

Callus formation assay from petioles was performed as described previously (Iwase et al., 2017). To induce tracheary element formation in cotyledons, VISUAL (Kondo et al., 2016) was employed using whole seedlings without cut treatment. The rate of tracheary element formation was evaluated by BF‐170 (Sigma‐Aldrich) staining according to a previous study (Nurani et al., 2020).

Petiole grafting assay

The middle part of the first and second leaf petioles of 16‐d‐old plants was transversely cut by razor blade (FA‐10; Feather, Osaka, Japan) and grafted together in a silicon tube with 0.5 mm diameter (As One, Osaka, Japan). After incubating the grafted plants at 22°C under continuous light for 12 d, their roots were soaked in water containing 1 mM 5‐CFDA florescent dye (Cosmo Bio, Tokyo, Japan) for 1 h. Xylem reformation at wound sites was judged by dissection microscope and the transmission of fluorescent dyes into grafted leaf vasculature was used as criteria for successful grafting. Physical reconnection at graft sites was assessed by pulling leaf blades apart after the xylem formation check. Tracheary elements in graft junction were stained with phloroglucinol reagent (1% (w/v) phloroglucinol (Fujifilm Wako, Osaka, Japan) in 20% (v/v) hydrochloric acid) for 10 min under vacuum at room temperature. Stained samples were mounted in chloral hydrate solution (8 g chloral hydrate (Fujifilm Wako), 1 ml glycerol (Fujifilm Wako) and 2 ml deionized water) before microscopy.

Microscopy

Callus and BF‐170 signals were observed under Leica M165 C stereomicroscope. Propidium iodide signals, autofluorescence and phloroglucinol signals from secondary cell walls were visualized by a Leica TCP SP5 II confocal laser microscope and BX51 microscope (Olympus).

Pathogen infection assay

The pathogenic bacterial growth assays were performed as described previously (Laohavisit et al., 2020), with slight modification. Plants for pathotest were grown in a mixture of vermiculite and soil (1 : 1) under short day conditions (8 h : 16 h, light : dark, 21°C : 22°C) for 5–6 wk. Pseudomonas syringae pv. tomato (Pst) DC3000 were grown in liquid King’s B medium at 28°C before experiment. Leaves of different Arabidopsis lines were inoculated with the virulent bacterial pathogen either by infiltration. Bacterial suspension (optical density at 600 nm (OD600nm) = 0.0002) were syringe infiltrated into the leaves of 5‐ to 6‐wk‐old plants. Plants were maintained at high humidity during the course of infection and returned to the same growth regime (8 h : 16 h, light : dark, 21°C : 22°C). Leaf discs were taken 3 d post‐inoculation (day 3) from three leaves per plant, with six plants per genotype per independent trial. Bacterial growth was assessed by colony counting.

Accession numbers

Sequence data from this article can be found in the Arabidopsis Genome Initiative under the following accession numbers: PP2AA3 (At1g13320), WIND1 (At1g78080), WIND2 (At1g22190), WIND3 (At1g36060), WIND4 (At5g65130), PLT3 (At5g10510), PLT5 (At5g57390), PLT7 (At5g65510), ERF115 (At5g07310), RAP2.6L (At5g13330), RAP2.6 (At1g43160), VND6 (At5g62380), VND7 (At1g71930), LBD30 (At4g00220), WOX5 (At3g11260), ANAC071 (At4g17980), ANAC096 (At5g46590), ANAC011 (At1g32510), WRKY18 (At4g31800), WRKY40 (At1g80840), WRKY53 (At4g23810), CYP71A13 (At2g30770), ALD1 (AT2G13810), MAPKKK18 (At1g05100), RAS1 (At1g09950), DEAR1 (At3g50260), ERF54 (At4g28140), ATEXPL2 (At4g38400) and a gene cording pectin lyase‐like superfamily protein (At1g02460).

Results

WIND1 induction causes dynamic transcriptional changes with distinct temporal expression patterns

To understand the gene expression dynamics after WIND1 induction, we performed time‐ course transcriptome analyses using XVE‐WIND1 plants where we induced WIND1 expression by the application of ED (Supporting Information Fig. S1a). We harvested WT and XVE‐WIND1 plants at 1, 3, 6, 12 and 24 h after ED treatment and confirmed the induction of WIND1 transcript level by RT‐qPCR analysis in the ED‐treated XVE‐WIND1 seedlings (Fig. S1b). We subsequently searched for differentially expressed genes (DEGs) between ED‐ and dimethyl sulfoxide (DMSO)‐treated XVE‐WIND1 seedlings by the microarray analysis and identified 2390 genes that were more than two‐fold upregulated and 2140 genes that were more than two‐fold downregulated in at least one time point after WIND1 induction (FDR < 0.05) (Tables S1 and S2). We observed an increase in the number of DEGs over the 24 h time course (Fig. S1c), implying the existence of gene regulatory networks governed by WIND1. Since WIND1 functions primarily as a transcriptional activator (Iwase et al., 2017), we decided to focus on the 2390 upregulated genes and subjected them to the k‐means clustering analysis. These upregulated genes can be grouped in four different classes according to their temporal expression patterns (Fig. 1a). Class I included 274 genes that showed early and transient induction, given their expression was upregulated at 1 and/or 3 h after induction, and reverted back to the basal level by 6 h. By contrast, 728 genes in class II showed upregulation at 1 or 3 h after WIND1 induction and their expression remained high within the 24 h time period. Classes III and IV included 721 genes which were upregulated at 6 h, and 667 genes which were upregulated at 12 h, respectively, after WIND1 induction. Since WIND1 was expressed constitutively in this experimental system (Fig. S1b; Table S1), the majority of its downstream genes remained highly expressed once they were induced. The transient nature of class I gene expression, however, inferred that their expression was tightly regulated by negative feedback mechanisms.

Fig. 1

WIND1 induces gene expression with distinct temporal patterns. (a) Heat‐map representation of WIND1‐induced transcriptional changes. Gene expression at 1, 3, 6, 12, 24 h after 17β‐estradiol (ED) treatment is shown for XVE‐WIND1 and wild‐type (WT) seedlings. Values are fold change relative to dimethyl sulfoxide (DMSO)‐treated seedlings. WIND1‐induced genes can be grouped into four classes by k‐means clustering analysis. Class I contains 274 genes that are transiently expressed within 1 or 3 h after WIND1 induction. Class II includes 728 genes that are constitutively expressed after WIND1 activation. The 721 genes in Class III and 667 genes in Class IV are expressed from 12 and 24 h, respectively, after WIND1 induction. (b) Gene ontology (GO) analysis for classes I–IV genes. GO categories such as response to chitin and transcription factor activity are over‐represented among class I genes. GO categories such as transcription factor activity and response to wounding are over‐represented among class II genes. GO categories such as extracellular region and secondary cell wall biogenesis are over‐represented among classes III and IV genes. The P‐values are shown in color codes for each GO category. (c) Venn diagram showing the overlap between WIND1‐induced genes in XVE‐WIND1 plants and genes induced by wounding in Arabidopsis hypocotyls (Ikeuchi et al., 2017). The significance of overlap between a pair of gene sets is evaluated by Fisher’s exact test, and P‐values and odds ratio are shown. (d) Transient activation of putative WIND1 target genes in Arabidopsis culture cells. Upper panel shows the reporter construct, GENEp:L‐LUC, and effector constructs, control and 35S:WIND1, used in the transient activation analysis. The black bar in the reporter construct represents the 3000‐bp promoter sequence of putative downstream genes. The gray box represents the coding sequence of L‐LUC, encoding a firefly luciferase gene, and +1 ATG indicates the translational start site. For the effector constructs, gray arrows mark 35SΩ, the cauliflower mosaic virus 35S promoter with the tobacco mosaic virus omega translation amplification sequence, and gray boxes mark NOS, the Agrobacterium tumefaciens nopaline synthase transcriptional terminator. The white box marks the WIND1 coding sequence. Bottom panel shows the promoter activity of six putative WIND1 target genes as judged by the L‐LUC activity relative to R‐LUC, Renilla luciferase. Cobombardment of 35S:WIND1 and GENEp:L‐LUC shows the transactivation of target promoter by WIND1. Data are mean ± SE (n = 6, technical replicates). Statistical significance is determined by Student t‐test (***, P < 0.001).

Gene ontology (GO) analysis for class I genes revealed that GO categories such as ‘response to chitin’ (P value = 1.66E‐08, odds ratio = 10.29) were over‐represented (Fig. 1b). Since chitin is implicated in defense signaling (Kaku et al., 2006), this enrichment suggests that WIND1 activates defense responses. The GO term ‘transcription factor activity’ (P value = 5.63E‐06, odds ratio = 2.36) was also highly ranked (Fig. 1b), strongly suggesting that WIND1 primarily targets transcriptional regulators. The GO category ‘transcription factor activity’ (P value = 7.44E‐14, odds ratio = 2.40) was also ranked as the top category among class II genes, reinforcing the idea that WIND1 functions as a key regulator of transcriptional network. As expected, we also found ‘response to wounding’ (P value = 8.08 E‐09, odds ratio = 4.77) in class II as well as in class III (P value = 1.26 E‐05, odds ratio = 3.67), confirming that majority of wound‐responsive genes are WIND1‐inducible (Cheong et al., 2002; Delessert et al., 2004; Iwase et al., 2011a). In addition, ‘Response to water deprivation’ was over‐represented among class II (P value = 7.98 E‐11, odds ratio = 4.55) and class III (P value = 1.09E‐07, odds ratio = 3.77) genes, supporting the notion that water deprivation may contribute to the expression of wound‐induced genes (Reymond et al., 2000). Furthermore, GO terms such as ‘extracellular region’ (P value = 2.20E‐16, odds ratio = 2.20), ‘secondary cell wall biogenesis’ (P value = 8.46E‐10, odds ratio = 11.89) and ‘xylan biosynthetic process’ (P value = 2.56E‐06, odds ratio = 12.08) were over‐represented among class III genes, suggesting that cell wall remodeling may be initiated within 12 h after WIND1 induction. Enrichment of a GO term ‘glucuronosyltransferase activity’ (P value = 2.18E‐05, odds ratio = 15.53) from class III genes also supported such an idea since genes, such as GLUCURONIC ACID SUBSTITUTION OF XYLAN1 (GUX1) and GUX2 (Table S1) which encode xylan glucuronosyltransferases required for the remodeling of cell wall polysaccharides, hemicelluloses (Rennie et al., 2012), were included in this category. It is also worth noting that other GO terms such as ‘response to jasmonic acid (JA)’ (P value = 4.54E‐09, odds ratio = 5.49) and ‘regulation of JA‐mediated signaling pathway’ (P value = 3.11E‐07, odds ratio = 12.85) were also over‐represented among class III genes, implying that WIND1 activates long‐term JA responses. Among class IV genes, ‘extracellular region’ was the most highly ranked GO term and several other terms such as ‘cell wall’ and ‘plant‐type cell wall organization’ were also over‐represented, suggesting that WIND1 promotes cell wall reorganization beyond 24 h. Among 11 genes representing the ‘response to water’ category (P value = 3.41E‐06, odds ratio = 22.89), 5 out of 10 dehydrin genes in Arabidopsis (Hanin et al., 2011) were upregulated (Table S1), predicting that WIND1 may participate in the dehydration tolerance. Enrichment of ‘unknown molecular functions’ (P value = 2.25E‐08, odds ratio = 1.43) is intriguing and could reflect unclarified physiological responses driven by WIND1. These data together suggest that WIND1 activation triggers a series of cellular and physiological processes ranging from the early defense response to the later, more prolonged dehydration response and rearrangement of cell wall polysaccharides.

We have previously shown that wound stress upregulates 5595 genes within 24 h in Arabidopsis hypocotyls (Ikeuchi et al., 2017). Comparison between these wound‐induced genes and 2390 WIND1‐induced genes revealed that 32.1% of WIND1‐induced genes (766 genes out of 2390 genes, P value = 2.2E‐16, odds ratio = 1.92) were also induced by wounding (Fig. 1c), further substantiating an important role of WIND1 as the key transcriptional activator in response to wounding. To further confirm that WIND1 acts as a transcriptional activator for wound‐induced genes, we randomly selected the following six genes from the list of 766 genes, including MITOGEN ACTIVATED PROTEIN KINASE KINASE KINASE 18 (MAPKKK18), RESPONSE TO ABA AND SALT 1 (RAS1), DREB AND EAR MOTIF PROTEIN 1 (DEAR1), ERF54, AARABIDOPSIS THALIANA EXPANSIN LIKE 2 (ATEXPL2) and At1g02460/pectin lyase‐like superfamily protein, that were induced by both WIND1 induction and wounding (Fig. 1c; Table S1) and tested whether WIND1 can induce their expression in a transient transactivation assay. We constructed the luciferase reporter plasmids to drive the expression of luciferase proteins by 3 kb upstream sequence of the candidate genes and co‐bombarded these plasmids with the WIND1 effector plasmids into Arabidopsis suspension cells. WIND1 overexpression significantly increased the luciferase activity in five out of the six genes tested, when compared to the control plasmids (Fig. 1d). This indicates that WIND1 can activate the promoter in five of the tested wound‐induced genes either directly or indirectly.

WIND1 induces multiple reprogramming regulators to promote callus formation

WIND1 promotes diverse modes of regeneration including callus formation, shoot and root regeneration, and somatic embryogenesis (Iwase et al., 2011a, 2013, 2015; Ikeuchi et al., 2013). WIND1 induces ESR1 expression to promote callus formation and shoot regeneration (Iwase et al., 2017) but it could also affect expression of other reprogramming regulators. Indeed, our Venn diagram analysis showed that 30 out of 91 reprogramming regulators in Arabidopsis (Ikeuchi et al., 2019) were highly upregulated in XVE‐WIND1 seedlings and/or 35S:WIND1 callus (Iwase et al., 2011a) (Fig. 2a; Table 1). As expected, this list of WIND1‐induced reprogramming regulators included ESR1, and we found many other key transcriptional regulators, such as LBD18, OBF BINDING PROTEIN 4 (OBP4), ESR2, RAP2.6L, ERF115, WUSCHEL (WUS), PLT5, WUSCHEL‐RELATED HOMEOBOX 5 (WOX5), LEAFY COTYLDEON 2 (LEC2) and Arabidopsis NAC DOMAIN CONTAINING PROTEIN 71 (ANAC071), which were expressed within 24 h after WIND1 induction (Table 1). We have previously shown that WIND1 enhances the cytokinin response near wound sites (Iwase et al., 2011a) and consistently, cytokinin biosynthesis genes, LONELY GUY 1 (LOG1) and LOG4, were induced by WIND1 overexpression within 6 h (Table 1). WIND1 may, in addition, regulate auxin homeostasis during regeneration since an auxin biosynthesis gene, YUCCA4 (YUC4), and the transport gene, PIN‐FORMED 1 (PIN1), were also activated by 6 h after WIND1 induction. Finally, consistent with the upregulation of ERF115 and ANAC071 by WIND1, their downstream target genes such as the wound‐responsible small peptide PHYTOSULFOKINE 5 (PSK5; Loivamäki et al., 2010; Heyman et al., 2013), XYLOGLUCAN ENDOTRANSGLUCOSYLASE/HYDROLASE 19 (XTH19) and XTH20 (Pitaksaringkarn et al., 2014), were induced within 24 h after WIND1 induction (Table 1). These results strongly suggest that WIND1 activates a multifaceted set of reprogramming regulators to promote cell proliferation and cell wall remodeling during regeneration.

Fig. 2

WIND transcription factor‐induced RAP2.6L and ERF115 promote callus formation at wound sites. (a) Venn diagram showing the overlap between WIND1‐induced genes in XVE‐WIND1 seedlings, WIND1‐induced genes in 35S:WIND1 callus (Iwase et al., 2011a) and genes implicated in reprogramming in Arabidopsis (Ikeuchi et al., 2019). The significance of overlap between a pair of gene sets is evaluated by Fisher’s exact test, and P‐values and odds ratio are shown. (b) Quantitative analysis of wound‐induced callus formation in wild‐type (WT), wind1/2/3/4, Pro (WIND1‐SRDX), rap2.6l‐1, Pro (RAP2.6L‐SRDX), erf115, Pro (ERF115‐SRDX), plt3/5/7, wox5‐1 and anac071/096/011. Leaf explants were cultured on phytohormone‐free Murashige & Skoog (MS) medium and callus phenotypes were scored at 8 d after wounding. Box plots represent the distribution of projected callus area (n = 33 to 352 per genotype); horizontal line shows median, the lower and upper bounds of each box plot denote the first and third quartiles, and whiskers above and below the box plot indicate 1.5 times the interquartile range. Outliers are shown as dots. Letters indicate statistical significance determined by ANOVA and Tukey's multi‐comparison test (P < 0.05). (c) Representative images of wound‐induced callus generated in WT, wind1/2/3/4, Pro (WIND1‐SRDX), rap2.6l‐1, Pro (RAP2.6L‐SRDX), erf115, Pro (ERF115‐SRDX), plt3/5/7, wox5‐1 and anac071/096/011 leaf explants. Photographs are taken at 8 d after cutting leaf petioles. Bars, 1 mm. (d) Reverse transcription quantitative polymerase chain reaction (RT‐qPCR) analysis of RAP2.6L, ERF115, PLT5 and ANAC071 expression after wounding. First and second leaves of 14‐d‐old WT wind1/2/3/4 and WIND1‐SRDX seedlings were cut and leaf explants were cultured on phytohormone‐free MS medium. Expression levels are normalized against those of the PP2AA3 gene. Data are mean ± SE (n = 3, biological replicates).

Table 1

Heat‐map representation of WIND1‐induced transcriptional changes for genes implicated in regeneration.

Since WIND1 is required for callus induction at wound sites (Iwase et al., 2011a), we next asked whether WIND1‐targetted reprogramming regulators participate in wound‐induced callus formation. As previously reported (Iwase et al., 2011a, 2017), WT leaf explants generated a large mass of callus cells at wound sites while the WIND1‐SRDX plants in which WIND1 function is dominantly repressed by the WIND1‐SRDX chimeric proteins showed smaller callus at wound site (Fig. 2b,c). We also confirmed that the wind1 wind2 wind3 wind4 quadruple mutant (hereafter referred to as wind1/2/3/4) was not defective in the wound‐induced callus (Iwase et al., 2011a). Given that this quadruple mutant line showed clear reduction in expression levels of each WIND genes (Fig. S2a), this suggests there are functionally redundant factors which enhance callus formation other than WIND transcription factors in wounded petiole. Strikingly, callus formation was significantly compromised in leaf explants of rap2.6l‐1/erf113, RAP2.6L‐SRDX, erf115, ERF115‐SRDX and plt3 plt5 plt7 triple mutant (hereafter referred to as plt3/5/7) loss‐of‐function mutants (Fig. 2b,c), clearly demonstrating the requirement of RAP2.6L, ERF115, and PLT proteins in wound‐induced callus formation. By contrast, wox5‐1 and anac071 anac096 anac011 triple mutant (hereafter referred to as anac071/096/011) loss‐of‐function mutants showed no significant difference in callus formation compared to WT (Fig. 2b,c), indicating that these transcriptional regulators are dispensable for wound‐induced callus formation in Arabidopsis leaf petioles. We should note that both ERF115 and PLTs are required for callus formation in wounded hypocotyls but RAP2.6L is unessential (Ikeuchi et al., 2017). This indicates that tissue specific regulatory mechanisms exist for wound‐induced callus formation.

To investigate whether these WIND1‐targetted genes are induced by wounding, we quantified their expression levels in wounded leaf petioles. Our RT‐qPCR analysis showed that several tested genes, including RAP2.6L, ERF115 and PLT5, were induced within 1 h in response to wounding and declined after 3 h (Fig. 2d). By contrast, the expression of ANAC071 was induced more slowly after wounding and its transcript level continued to increase up to 12 h (Fig. 2d). We did observe clear downregulation of RAP2.6L and ERF115 expression in the WIND1‐SRDX and wind1/2/3/4 plants (Fig. 2d), indicating that WIND transcription factors are required for their upregulation upon wounding. Overall, these data suggest that wound stress induces multiple WIND1 downstream genes through robust transcriptional mechanisms and loss of WIND transcription factor function is insufficient to block their wound‐induced expression. Interestingly, RAP2.6, a close homolog of RAP2.6L/ERF113 and ERF115 within the ERF subfamily X, was strongly induced by wounding and suppressed in the WIND1‐SRDX but neither its loss‐of‐function nor gain‐of‐function mutants displayed any defects in wound‐induced callus formation (Fig. S2b,c). These data thus demonstrate that WIND transcription factors selectively utilize RAP2.6L and ERF115 among this subfamily for callus formation.

WIND transcription factors promote xylem reconnection

Strong enrichment of cell wall‐related genes among WIND1‐induced genes suggests that WIND1 participates in the remodeling of cell walls after wounding. The formation of vasculatures in tissues near wound sites is the post‐wound process that involves dynamic deposition and modification of cell wall polysaccharides (Lipetz, 1970). Our Venn diagram analysis indeed revealed that genes upregulated in either XVE‐WIND1 or 35S:WIND1 plants significantly overlapped with those induced during vascular formation in the VISUAL system (Fig. 3a; Tables 2, S3). Importantly, WIND1 induces several key transcriptional regulators implicated in vascular formation including VND6, VND7 and LBD30 (Kubo et al., 2005; Soyano et al., 2008) (Table 2), strongly suggesting that WIND1 promotes vascular development. Consistent with this transcriptional activation, we found that overexpression of WIND1 induced ectopic formation of tracheary elements in both XVE‐WIND1 and 35S:WIND1 plants (Figs 3b, S3a). Importantly, tracheary elements can be formed from differentiated cells such as leaf mesophyll cells and root hair cells (Figs 3b, S3a), confirming the strong reprogramming capability of WIND1.

Fig. 3

WIND transcription factors induce key regulators of xylem differentiation and promotes leaf petiole grafting. (a) Venn diagram showing the overlap between WIND1‐induced genes in XVE‐WIND1 seedlings, WIND1‐induced genes in 35S:WIND1 callus (Iwase et al., 2011a) and genes induced during vascular development under the Vascular Cell Induction Culture System Using Arabidopsis Leaves (VISUAL) condition (Kondo et al., 2015, 2016). The significance of overlap between a pair of gene sets is evaluated by Fisher’s exact test, and P‐values and odds ratio are shown. (b) Confocal optical sections of propidium iodide (PI)‐labeled mesophyll cells in XVE‐WIND1 cotyledons. XVE‐WIND1 seedlings were germinated in the presence of dimethyl sulfoxide (DMSO) or 10 μM 17β‐estradiol (ED) and tracheary element formation was visualized by PI staining in 7‐d‐old seedlings. Bars, 50 µm. (c) Fluorescent images of wild‐type (WT) and WIND1‐SRDX cotyledon cells cultured under the VISUAL condition. Tracheary element formation visualized by BF‐170 staining were marked by arrowheads. Bars, 500 µm. (d) Quantitative analysis of xylem differentiation in WT, WIND1‐SRDX, wind1/2/3/4 cotyledon cells cultured under the VISUAL condition. Differentiation rate is shown as relative fluorescence intensity of BF‐170 normalized by leaf area. Box plots represent the distribution of projected callus area (n = 66 for WT, 63 for WIND1‐SRDX and 75 for wind1/2/3/4). Letters indicate statistical significance determined by ANOVA and Tukey's multi‐comparison test (P < 0.05). (e) Experimental setup for the petiole grafting assay. Incised petioles were grafted in a silicon tube and tissue reconnection was examined after 12 d. Vascular reconnection was assessed by successful transmission of a fluorescent dye CFDA into grafted leaves. Bars, 1 mm. (f) Quantitative analysis of tissue reconnection (left) and vascular reconnection (right) in WT and WIND1‐SRDX seedlings. Statistical significance is determined by a proportion test (n = 161 for WT, 112 for WIND1‐SRDX, 47 for wind1 and 64 for wind1/2/3/4; *, P < 0.05; ***, P < 0.001). (g) Reverse transcription quantitative polymerase chain reaction (RT‐qPCR) analysis of VND6, VND7 and LBD30 expression after wounding. First and second leaves of 14‐d‐old WT, WIND1‐SRDX and wind1/2/3/4 seedlings were cut and leaf explants were cultured on phytohormone‐free Murashige & Skoog (MS) medium. Expression levels are normalized against those of the PP2AA3 gene. Data are mean ± SE (n = 3, biological replicates).

Table 2

Heat‐map representation of WIND1‐induced transcriptional changes for genes implicated in vascular development.

Our time‐course RT‐qPCR analysis further showed that WIND1 expression itself is strongly induced within 1 h after seedlings were transferred to the VISUAL condition (Fig. S3b), implying that WIND1 functions at an early step during vascular formation in this experimental setup. We, therefore, compared the tracheary element formation between WT, WIND1‐SRDX and wind1/2/3/4 plants under the VISUAL condition and observed significant reduction in leaf mesophyll cells of both WIND1‐SRDX and wind1/2/3/4 plants (Figs 3c,d, S3c). We also found that the expression of several key genes in xylem formation, for example IRX3, XCP1, VND7 and LBD15, was compromised in WIND1‐SRDX plants (Fig. S3d), further corroborating that WIND transcription factors and functional redundant factors activate key transcriptional regulators in vascular formation. By contrast, expression of genes implicated in cambium development, that is, TDR and AtHB8, or phloem development, that is, SEOR1, APL and SMXL5, was comparable between WT and WIND1‐SRDX plants (Fig. S3d), supporting the idea that WIND1 selectively promotes xylem formation in the VISUAL system.

To examine whether WIND transcription factors contribute to vascular formation after wounding, we developed an experimental system which enabled the assessment of vascular reconnection (Fig. 3e). In this system, cut leaf petioles were forced to graft within a silicon tube. Using this assay, c. 80% of WT, wind1 single knockout and wind1/2/3/4 quadruple mutant leaf petioles were physically reconnected, while only 62% of WIND1‐SRDX petioles managed to reconnect (Fig. 3f). We also observed xylem reconnection at the graft junction by examining the transport of a florescent dye from roots to the grafted leaves (Fig. 3e). We found that 39% of WT petioles transported the dye into grafted leaves. Phloroglucinol staining confirmed xylem reconnection is achieved by bridge of tracheary elements between stock and leaf petiole scission (Fig. S3e). Importantly, only 24% of WIND1‐SRDX petioles succeeded. Moreover, wind1/2/3/4 and wind1 mutant significantly reduced vascular reconnection rate to 5% and 19%, respectively (Fig. 3f), clearly indicating an involvement of WIND1 and other WIND homologs in xylem reconnection in wound‐induced petiole callus (Fig. 3f). We also found that wound‐induced expression of VND6, VND7 and LBD30 were significantly reduced in WIND1‐SRDX plants but not in wind1/2/3/4, suggesting upregulation of these key transcription factors soon after wounding by functionary redundant factors other than WIND transcription factors.

WIND1 induces defense response regulators to convey resistance against Pseudomonas syringae DC3000

Since defense response genes are enriched in WIND‐induced genes (Fig. 1), we sought to explore whether WIND1 also plays a role in the resistance against pathogen infection. Defense‐related genes were activated in XVE‐WIND1 or 35S:WIND1 plants (Fig. 4a; Tables 3, S4), which were significantly overlapped with those induced in response to the pathogen infection (Sato et al., 2007). Importantly, WIND1 itself was induced by pathogens (Table 3), supporting a possible role for WIND1 in the defense response. Consistently, we found that several pathogen‐responsive WRKY genes, including WRKY18 and WRKY53 (Pandey et al., 2010; Abeysinghe et al., 2019), were induced by WIND1 (Table 3). To test whether WIND1 is involved in the immune responses, we challenged WT, WIND1‐SRDX, wind1/2/3/4, and wind1 plants with a virulent bacterial pathogen Pst DC3000. WIND1‐SRDX and wind1/2/3/4 plants consistently were more susceptible to the Pst DC3000 infection (Figs 4b, S4a), demonstrating that WIND1 and other WIND transcription factors are required for the resistance against Pst DC3000 infection.

Fig. 4

WIND transcription factors induce a set of defense response regulators and promotes resistance against Pseudomonas syringae pv. tomato (Pst) DC3000. (a) Venn diagram showing the overlap between WIND1‐induced genes in XVE‐WIND1 seedlings, WIND1‐induced genes in 35S:WIND1 callus (Iwase et al., 2011a) and genes induced in response to pathogen infection (Sato et al., 2007). The significance of overlap between a pair of gene sets is evaluated by Fisher’s exact test, and P‐values and odds ratio are shown. (b) WIND1‐SRDX and wind1/2/3/4 are susceptible to Pst DC3000 infection. Bacterial growth was assessed by colony counting after 3 d of inoculation. Leaf discs were taken from three leaves per plant, with six plants per genotype per independent trial. Left, representative figures from infiltrated plants at 3 d post‐inoculum. White asterisks represent Pst DC3000‐infiltrated leaves. Right, box plots represent the colony‐forming units (CFU) per cm2 which were determined at 3 d post‐inoculum; horizontal line shows median, the lower and upper bounds of each box plot denote the first and third quartiles, and whiskers above and below the box plot indicate 1.5 times the interquartile range. Individual data points are shown with mean ± SD, n = 18 plants from three separate trials. Letters indicate statistical significance determined by ANOVA and Tukey's multi‐comparison test (P < 0.1). Bar, 2 cm. (c) Reverse transcription quantitative polymerase chain reaction (RT‐qPCR) analysis of WRKY18, WRKY53, CYP71A13 and ALD1 expression after wounding. First and second leaves of 14‐d‐old wild‐type (WT), wind1/2/3/4 and WIND1‐SRDX seedlings were cut and leaf explants were cultured on phytohormone‐free Murashige & Skoog (MS) medium. Expression levels are normalized against those of the PP2AA3 gene. Data are mean ± SE. (n = 3, biological replicates).

Table 3

Heat‐map representation of WIND1‐induced transcriptional changes for genes implicated in pathogen response.

Time‐course RT‐qPCR analysis further showed that WIND1 was also induced by the application of bacterial‐derived flagellin peptide (flg22), a well‐established pathogen‐associated molecular pattern (PAMP) (Fig. S4b). As reported previously (Sato et al., 2007), flg22 also induced the expression of WRKY18 and WRKY53 in WT plants but this induction was not significant in WIND1‐SRDX plants (Fig. S4c), implying that gene expression could be regulated by factors, other than WIND transcription factors in flg22 treatment. Strikingly we observed similar upregulation manner of these WRKY genes after wounding (Fig. 4c), and WRKY18 upregulation is significantly compromised in wind1/2/3/4, inferring that the WINDs‐mediated transcriptional cascade is activated during post‐wound defense signaling. Furthermore, we also found that CYP71A13 and ALD1 were wound‐inducible and their expression during the early time point was impaired in WIND1‐SRDX and wind1/2/3/4 plants (Fig. 4c), suggesting that WIND1 may promotes a rapid production of camalexin and/or pipecolic acid at wound site. Since WIND1‐inducible RAP2.6 was also induced by pathogen or flg22 (Fig. S4c; Table 3), we tested whether it is required for the defense response against Pst DC3000 infection. We, however, did not observe clear defects in rap2.6 or its close homolog mutants RAP2.6L (Fig. S4a). Overall, these results suggest that WIND transcription factors positively affected defense responses, presumably via the regulation of downstream defense‐related genes.

Discussion

In this study we demonstrated that WIND1 is sufficient to activate several transcriptional cascades to coordinately drive cellular reprogramming, xylem formation and defense responses against pathogen infection. Our further experiments by WIND loss‐of‐function mutants revealed WIND transcription factors orchestrate these physiological responses when plants are exposed to wounding or other stress conditions (Fig. 5).

Fig. 5

A schematic model describing how WIND1 promotes multiple physiological responses. Our transcriptome‐based analysis in this study reveals that WIND1 has a potential to activate reprogramming regulators such as RAP2.6L, ESR1, ERF115, PLT3, PLT5 and PLT7 to promote callus formation. WIND1 is also able to upregulate master regulators of xylem differentiation, including VND6, VND7, LBD18 and LBD30, to promote tracheary element formation. WIND1, in addition, promotes defense response against pathogens by activating a set of key regulators such as RAP2.6, CYP71A13, WRKY18, WRKY53 and ALD1.

Roles of WIND1 in wound‐induced regeneration

We have previously shown that wounding rapidly increases WIND1 expression to promote cellular reprogramming near wound sites (Iwase et al., 2011a). In this study, we further corroborate WIND1 as an important driver of wound‐induced transcription, with over 30% of WIND1‐induced downstream genes are also wound‐inducible (Fig. 1a,b). It is also striking that WIND1 can induce c. 30% of genes implicated in regeneration in Arabidopsis (Fig. 2; Table 1). This agrees with our observation that WIND1 promotes several regenerative events such as callus formation, shoot regeneration and somatic embryogenesis, and confirms the central roles that WIND1 plays in regeneration. In addition to ESR1 (Iwase et al., 2017), we showed that RAP2.6L functions downstream of WIND transcription factors in wound‐induced callus formation (Fig. 2) although we do not know at this point whether WIND1 and/or other WINDs directly activates RAP2.6L expression. Similar to WIND1 and ESR1, RAP2.6L is required for both callus formation at wound sites and shoot regeneration under the tissue culture condition (Che et al., 2006). These observations suggest that these two regenerative processes are somehow linked at the molecular level and further elucidation of how WIND1‐ESR1/RAP2.6L pathway promotes these processes should clarify the underlying mechanisms. WIND1 also induces RAP2.6, a close homolog of RAP2.6L but neither knock‐out nor overexpression caused any significant changes in callus formation (Fig. S2). A previous study has shown that RAP2.6 functions in protecting plants against nematode infection (Ali et al., 2013). It is thus likely that the activation of RAP2.6L and RAP2.6, which are both WIND1‐driven, transcriptionally coordinates regeneration and defense responses (Fig. 5). How these closely related homologs target different sets of genes to promote distinct physiological processes is an interesting question that should be investigated in future studies.

In addition to RAP2.6L, wounding induces ERF115 and PLT3, PLT5, PLT7 to promote callus formation at wound sites (Fig. 2; Table 1). This suggests that this group of AP2/ERF proteins plays a key role in decoding wounding signals to acquire a new developmental trajectory. Heyman et al. (2016) has shown that ERF115 is required to induce WIND1 after bleomycin treatment, thus it appears that ERF115 and WIND1 form a positive feedback loop to enhance this transcriptional pathway. It is interesting that repression of WIND functions does not block wound‐induced expression of PLT5 (Fig. 2d). We predict that wound‐induced transcriptional changes are regulated by robust, highly overlapping mechanisms and loss of WIND1 function can probably be compensated by activating alternative pathways to induce downstream gene expression. Although WIND1‐SRDX should dominantly suppress transcriptional regulators that share the same cis elements within target promoter, there might be other transcriptional regulators that can utilize different cis elements to modulate target genes. Investigating the genome‐wide, wound‐induced transcriptome in WT, WIND1‐SRDX and wind1/2/3/4 will be useful to comprehensively elucidate the WIND‐dependent and ‐independent transcriptional pathways.

Given that genes involved in JA response are highly represented among WIND1‐induced genes (Fig. 1a), WIND1 may also promote JA‐dependent regeneration. Recent studies unveiled an involvement of JA‐mediated pathways in root tip regeneration after laser ablation and de novo root formation from leaf cuttings (Zhang et al., 2019; Zhou et al., 2019). Our previous time‐course transcriptome analysis using Arabidopsis hypocotyls also showed that JA responses are activated after wounding although JA pathways are not required for callus formation in wounded hypocotyls (Ikeuchi et al., 2017). These results thus imply the existence of context‐dependent regeneration mechanisms (Ikeuchi et al., 2020) and an involvement of WIND1‐induced JA pathways should be tested in various regeneration contexts.

Our recent studies have uncovered that plants possess an epigenetic mechanism to suppress the expression of reprogramming regulators in somatic tissues and that cell reprogramming after wounding accompanies dynamic alterations in the histone modification status (Ikeuchi et al., 2015; Rymen et al., 2019). It is thus interesting to note that many of WIND1 downstream target genes we identified in this study, including RAP2.6L, have repressive histone marks, that is, H3K27me3, before wounding and quickly gain permissive acetylation marks, that is, H3K9/14ac and H3K27ac, within a few hours after wounding (Rymen et al., 2019). Future studies should thus clarify which downstream genes are directly regulated by WIND1 and WIND1 homologs and whether these transcriptional regulations involve epigenetic mechanisms. A recent study in the acoel Hofstenia miamia revealed that a master regulator of wound‐induced regeneration is quickly expressed along wound sites and functions as a pioneering factor to open the chromatin around target loci (Gehrke et al., 2019). Testing whether WIND transcription factors perform similar roles as a pioneering factor is an exciting future direction, allowing us to deepen our molecular understanding of how plants regulate wound‐induced transcription.

Newly uncovered roles of WIND1 in wound‐triggered responses

Alterations of cell wall composition and structure are one of the key features that accompany cell differentiation and morphogenesis in plants (Tucker et al., 2018). Recent studies have begun to reveal that changes in this status can modulate various extracellular signaling pathways to modify plant growth and stress responses accordingly (De Lorenzo et al., 2019). It is therefore conceivable that cells near wound sites change the cell wall status to transduce various wound‐induced signaling. In support of this idea, several mutants with defects in cell wall biogenesis and/or modification show spontaneous callus formation on phytohormone free medium (Krupková & Schmülling, 2009; Ikeuchi et al., 2013), suggesting that the cell wall status do influence the regenerative processes. Given that the cell wall‐related genes are significantly enriched among WIND1‐induced gene sets (Fig. 1b), we speculate that WIND transcription factors may also regulate regeneration by modifying the cell wall status. Consistently, we observed failure in the tissue reconnection in the petioles of WIND1‐SRDX plants (Fig. 3e,f), demonstrating the requirement of WIND transcription factors and other functional redundant factors to reconnect two existing cell walls. Putative downstream targets of WIND1 also include XTH19 and XTH20, which are members of xyloglucan endotransglucosylase/hydrolases (Table 1). Since their loss‐of‐function mutants are impaired in fusing incised Arabidopsis stems (Pitaksaringkarn et al., 2014), WIND1 may modulate their expression to remodel the xyloglucan structure and thus interweave closely located cell walls. Further investigation of how WIND transcription factors change the cell wall structure and composition should uncover previously unknown, cell wall‐based mechanisms that underlie wound‐induced regenerative processes.

Another important role of WIND1 and other homologs uncovered in this study is its involvement in wound‐induced vascular reformation (Fig. 3f). This is interesting because wounding was thought to be one of the triggers for tracheary element formation (Fukuda, 1997) but underlying molecular mechanisms remained elusive. Our data show that wounding stress induces expression of VND6, VND7 and LBD30 and their expression level is retarded in WIND1‐SRDX line but not in wind1/2/3/4 (Fig. 3g), suggesting that the wounding‐triggered VND/LBD pathway promotes xylem formation via functional redundant factors of WIND transcription factors. Our data based on the VISUAL system implies that this WIND‐related regulation specifically works on xylem cell formation, rather than cambium stem cell and phloem cell formation (Fig. S3). Intriguingly, tracheary element formation does not occur in every cell in 35S:WIND1 or XVE:WIND1 plants, and in fact, we often observe both callus cells and cells committing to the tracheary element formation in a mosaic fashion (Fig. S3a; Ikeuchi et al., 2015). How WIND genes overexpression promotes different cell fate is a key question that should be addressed in future studies. Of note, we observed that even detached leaf explants that failed vascular reconnection often showed tight tissue reconnection in our petiole grafting system. Moreover, the wind1/2/3/4 mutant displayed strong defects in vascular reconnection but not in tissue reconnection (Fig. 3). We thus believe that tissue connectivity is also regulated by other factors such as callus qualities (e.g. size of the callus, type of callus cells, composition of cell walls) and not necessarily by vascular connectivity. In support of this idea, the WIND1‐SRDX plants showed smaller callus at wound petiole (Fig. 2b,c) so this might be one of the reasons why they showed reduced tissue reconnection.

We also demonstrated that WIND1 induces a large set of pathogen‐responsive genes and WIND transcription factors modulate resistance against Pst DC3000 infection (Tables 3, S4; Figs 4a,b, S4a). Although we could not test whether the WIND transcription factors‐mediated pathways participate in the wound‐induced defense responses, we anticipate this is very likely, given that wounding strongly induces expression of WIND genes and their downstream pathogen response regulators (Fig. 4c). Of note that there are clear differences of gene expression patterns after flg22 treatment or wounding, for instance CYP71A13, ALD1 and RAP2.6 (Figs 4c, S2b, S4c). This indicates that the degrees of WINDs involvement to trigger the immunity‐related gene expression differ between flg22 treatment and wounding.

Altogether, this study establishes WIND1 as a positive regulator of wound‐induced responses since WIND1 can regulate both cellular reprogramming and immune‐related responses. Since injury involves the risk of infection by pathogens from the wounded site, it makes physiological sense to simultaneously enhance de novo tissue/organ regeneration and the immune response at the site of injury. For example, a mammalian transcription factor Hypoxia‐inducible factor (HIF)‐1 is upregulated in response to injury and is known to have a wide range of functions, including regeneration of the wound site, angiogenesis and activation of humoral immunity (Hellwig‐Burgel et al., 2005). We predict having factors such as WINDs (for plants) or HIFs (for mammalian) that can quickly direct multiple pathways was advantageous for the survival of organisms in evolution. One molecular mechanism of how WIND1 induces a specific, yet diverse, set of genes in a given context might be that WIND1 acts in multi‐protein complexes. This could, for instance, be with other transcriptional regulators and coordinately determine the target specificity. It is also interesting that WIND1 can be induced in conditions other than wounding, such as incubation under the VISUAL and flg22 treatment (Figs S3b, S4b). Further elucidation of its activation mechanisms, including whether WIND transcription factors regulate downstream key factors directly or indirectly, and how much expression levels of WIND genes/proteins are spatiotemporally required for the downstream phenomena, should help uncover how plants transduce multiple stress signals to regeneration and defense responses through the WINDs‐mediated pathways.

Author contributions

AI and KSugimoto conceived the research and designed the experiments. AI performed plasmid construction, transient expression assay and microscopy analyses. YK, AI and HF performed VISUAL assay. AL and KShirasu performed pathogen infection assay. AI and AT performed callus formation assay. MI and AI established and performed grafting experiments. NM and AI performed gene expression analyses. KM and MA provided unpublished materials. AI and KSugimoto wrote the manuscript with input from all coauthors.

Fig. S1 Transcriptional changes after WIND1 induction.

Fig. S2 RAP2.6 does not regulate callus formation at wound sites.

Fig. S3 WIND1 promotes xylem differentiation.

Fig. S4 Flagellin treatment upregulates WIND1 expression.

Click here for additional data file.

Table S1 Genes upregulated in XVE:WIND1 seedlings.

Table S2 Genes downregulated in XVE:WIND1 seedlings.

Table S3 Heat‐map representation of WIND1‐induced transcriptional changes for genes, other than transcriptional regulators, implicated in vascular development.

Table S4 Heat‐map representation of WIND1‐induced transcriptional changes for genes, other than transcriptional regulators, implicated in pathogen response.

Table S5 Oligonucleotides used in this study.

Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Click here for additional data file.