Lilium regale Wilson WRKY3 modulates an antimicrobial peptide gene, LrDef1, during response to Fusarium oxysporum

WRKY transcription factors (TFs) play vital roles in plant growth and development, secondary metabolite synthesis, and response to biotic and abiotic stresses. In a previous transcriptome sequencing analysis of Lilium regale Wilson, we identified multiple WRKY TFs that respond to exogenous methyl jasmonate treatment and lily Fusarium wilt (Fusarium oxysporum). In the present study, the WRKY TF LrWRKY3 was further analyzed to reveal its function in defense response to F. oxysporum. The LrWRKY3 protein was localized in the plant cell nucleus, and LrWRKY3 transgenic tobacco lines showed higher resistance to F. oxysporum compared with wild-type (WT) tobacco. In addition, some genes related to jasmonic acid (JA) biosynthesis, salicylic acid (SA) signal transduction, and disease resistance had higher transcriptional levels in the LrWRKY3 transgenic tobacco lines than in the WT. On the contrary, L. regale scales transiently expressing LrWRKY3 RNA interference fragments showed higher sensitivity to F. oxysporum infection. Moreover, a F. oxysporum-induced defensin gene, Def1, was isolated from L. regale, and the recombinant protein LrDef1 isolated and purified from Escherichia coli possessed antifungal activity to several phytopathogens, including F. oxysporum. Furthermore, co-expression of LrWRKY3 and the LrDef1 promoter in tobacco enhanced the LrDef1 promoter-driven expression activity. These results clearly indicate that LrWRKY3 is an important positive regulator in response to F. oxysporum infection, and one of its targets is the antimicrobial peptide gene LrDef1.


Background
WRKY transcription factors (TFs), the key regulators of plant response to biotic and abiotic stresses, constitute one of the largest TF families in plants. WRKY TFs possess a highly conserved domain comprising approximately 60 amino acid residues that contains one or two highly conserved short peptides ('WRKYGQK') at the N terminal, and a conserved zinc-finger motif of CX 7 CX 23 -HXC (C 2 HC) or CX 4-5 CX 22-23 HXH (C 2 H 2 ) at the C terminal [1,2]. The WRKY TF family can be classified into three groups based on the number of WRKY domains and the type of zinc-finger motif [1]. In general, group I contains two WRKY domains and a zinc-finger motif. Group II contains a WRKY domain and a C 2 H 2 zinc-finger motif, whereas group III includes one WRKY motif and the zinc-finger domain C 2 HC. Group I WRKY TFs can be divided into two subgroups: the Ia subgroup possessing a C 2 H 2 zinc-finger motif, and the Ib subgroup containing a C 2 HC zinc-finger motif. Group II WRKY TFs can be further subdivided into five subgroups, IIa, IIb, IIc, IId, and IIe, according to their phylogenetic relationships [3,4].
WRKY TFs have been shown to be important regulators in plant defense response against a variety of pathogens. For example, 16 CmWRKY genes exhibit distinct expression patterns in melon (Cucumis melo) during Podosphaera xanthii infection [5]. In peony (Paeonia lactiflora) infected by Alternaria tenuissima, PlWRKY65 expression is significantly enhanced, whereas reduced expression of PlWRKY65 via virus-induced gene silencing enhances sensitivity to Alternaria tenuissima infection [6]. Arabidopsis thaliana plants expressing grape (Vitis labrusca × V. vinifera) WRKY3 exhibit improved salt and drought tolerance at germination, seedling, and maturity stages. Compared with wild-type (WT) plants, VlWRKY3 transgenic Arabidopsis lines also have increased resistance to Golovinomyces cichoracearum but increased sensitivity to Botrytis cinerea [7].
WRKY TFs self-regulate and cross-regulate with other TFs to control the defense response of plants to biotic stress. In rice (Oryza sativa), OsWRKY45-2 enhances resistance to Magnaporthe oryzae by transcriptionally activating OsWRKY13 [8]. The overexpression of oilseed rape (Brassica napus) WRKY15 inhibits the transcriptional activation of BnWRKY33, which leads to increased plant susceptibility to Sclerotinia sclerotiorum [9]. CaW-RKY40b in pepper (Capsicum annuum) exhibits positive feedback regulation at the transcriptional level by directly targeting the W box element in its own promoter; this gene plays a negative regulatory role in the defense response of pepper against Ralstonia solanacearum, thus reducing plant immunity [10].
Lily (Lilium) is one of most popular fresh-cut flowers but is vulnerable to Fusarium wilt disease caused by pathogenic Fusarium species, such as F. oxysporum. Fusarium wilt affects the yield and quality of lily bulbs and cut flowers and also causes withering and death of lily plants. In our previous study, we carried out a preliminary investigation of the molecular interaction mechanism between a wild lily (L. regale Wilson) and F. oxysporum and confirmed that L. regale has strong resistance to F. oxysporum [11]. Drawing on the results of our previous L. regale transcriptome sequencing analysis, 35 WRKY genes responsive to F. oxysporum were cloned from L. regale [12]. qRT-PCR analysis revealed that the expression levels of LrWRKYs were induced by the exogenous application of four signaling molecules, such as salicylic acid (SA), methyl jasmonate (MeJA), and ethylene (ETH). Furthermore, the functional analysis confirmed that LrWRKY2 is a positive regulator that mediates L. regale defense responses to F. oxysporum infection by regulating chitinase 2 expression [13]. Compared with LrWRKY2, LrWRKY3 (GenBank no. MW125548) showed higher expression levels in root, stem, leaf, and flower tissues and upon treatment with SA, MeJA, and ETH. Moreover, LrWRKY3 exhibited more transcript accumulation in the L. regale roots during F. oxysporum infection than LrWRKY2 did. WRKY TF family members regulate a complex defenses network in plants in response to biotic stress [14], and similar observations were made in L. regale.
In this study, we investigated the regulatory mechanism of LrWRKY3 in defense response of L. regale to F. oxysporum. The onion (Allium cepa) epidermis cells without the chloroplast disturbance are large and easy to observe under microscope, thus they are commonly used for subcellular localization analysis of plant proteins. The transient expression in onion epidermal cells mediated by the Agrobacterium tumefaciens was used to analyze the subcellular localization of LrWRKY3. In addition, due to the current lack of a rapid and efficient method for the lily genetic transformation [15], LrWRKY3 was heterologously expressed in tobacco (Nicotiana tabacum) for functional verification. It is widely known that tobacco is an important model plant for rapid confirmation of gene functions, and F. oxysporum can rapidly cause Fusarium wilt in tobacco [16]. In addition, the LrWRKY3 function in defense response was explored in depth by transiently expressing its RNA interference (RNAi) fragment in L. regale scales. Finally, we isolated a F. oxysporum-responsive defensin gene (LrDef1, GenBank no. MZ872924) from L. regale and confirmed the resistance of LrDef1 to F. oxysporum. Moreover, this study further verified that LrWRKY3 regulates LrDef1 expression considering that the LrDef1 promoter had a W-box and rarely known interaction between the WRKY TFs and defensin during response to phytopathogen infection in plants.

Results
The L. regale subgroup-IIa WRKY3 gene was found to encode a nuclear protein In our previous study, we identified a series of WRKY genes from a Fusarium wilt-resistant wild lily (L. regale) by transcriptome sequencing. According to the gene expression pattern data, LrWRKY3 was induced by F. oxysporum and responded to signaling molecule treatments [12]. In the present study, the function of LrWRKY3 was therefore further characterized to understand the transcriptional regulation of WRKY in L. regale in response to Fusarium wilt. The full-length cDNA of LrWRKY3, which was found to be 722 bp in length with a 537-bp open reading frame (ORF), was predicted to encode a subgroup-IIa WRKY protein with 178 amino acid residues. The calculated molecular mass of the deduced LrWRKY3 was 20.3 kDa. According to our sequence analysis, LrWRKY3 contained a highly conserved 'WRKYGQK' sequence and a zinc-finger motif CX 4-5 CX 22-23 HXH (C 2 H 2 ), which suggests membership in the IIa WRKY subgroup (Fig. 1a). The deduced amino acid sequence of LrWRKY3 was highly similar to that of some WRKY proteins, including Phoenix dactylifera WRKY75 (GenBank no. XP_008809548.2), Cocos nucifera WRKY75 (KAG1370131.1), Elaeis guineensis WRKY75 (XP_010913024.1), and Durio zibethinus WRKY75 (XP_022740913.1) (Fig. 1b).
The result of PSORT prediction suggested that the protein encoded by LrWRKY3 is localized in the plant cell nucleus. To confirm the subcellular location of LrWRKY3, a fusion expression vector of LrWRKY3 and green fluorescent protein (GFP) was constructed and expressed in onion epidermal cells through Agrobacterium tumefaciens-mediated transformation. The green fluorescence signal of LrWRKY3-GFP fusion protein was exclusively distributed in the nuclei of onion epidermal cells and colocalized with the nuclear localization marker PI (Fig. 1c). In contrast, green fluorescence was detected throughout the entire onion epidermal cell in control. This result indicates that LrWRKY3 is localized in plant cell nucleus.
Overexpression of LrWRKY3 in tobacco increased resistance to F. oxysporum and enhanced transcription levels of some JA/SA signaling pathway genes, PRs, and SODs To further understand the biological function of LrWRKY3, the plant overexpression vector pCAM-BIA2300s-LrWRKY3 was constructed and transformed into WT tobacco leaf disks. A total of 50 T 0 transgenic tobacco plants were obtained according to PCR analysis with LrWRKW3-specific primers. qRT-PCR was used to analyze the expression levels of LrWRKY3 in leaves of 11 T 2 generation transgenic lines   (Fig. 2a). LrWRKY3 transcription was highest in line W3-6, with a relative expression value of 10.25. LrWRKY3 relative expression levels in W3-20, W3-35 and W3-44 were also high, and the expression value was 7.53, 8.40, and 7.25, respectively. These data indicate that LrWRKY3 was stably expressed in T 2 generation transgenic tobacco.
Four transgenic lines (W3-6/20/35/44) with high LrWRKY3 expression and WT plants were inoculated with a spore suspension of F. oxysporum to evaluate their resistance to the pathogen. The roots of WT tobacco turned black and rotted, and whole leaves had wilted 7 days after inoculation (Fig. 2b). In contrast, the roots of T 2 generation transgenic tobacco lines showed no evident disease symptoms, and the leaves were healthy, bright green, and non-wilted. The results of leaf inoculation experiments were consistent with these observations. The leaves of transgenic lines exhibited only slight deterioration and local yellowing near the inoculation wound, whereas leaves of WT were obviously more yellowed and decayed, with a larger damaged area, 7 days after inoculation (Fig. 2c-d). These results indicate that the overexpression of LrWRKY3 in tobacco improves disease resistance to F. oxysporum.
Transcription levels of some JA/SA signaling and defense-related genes in transgenic tobacco lines and WT tobacco were analyzed by qRT-PCR. As shown in Fig. 3, transcription levels of JA biosynthetic pathwayrelated genes, including NtLOX, NtAOC, NtOPR, NtAOS, NtJMT, NtKAT, and NtPACX, were obviously elevated in LrWRKY3 transgenic lines compared with WT. The expression level of NtJMT was highest in line W3-44: 16.89-fold higher than in WT tobacco. Compared with their expressions in WT, NtLOX and NtAOS were upregulated in W3-44 by 6.49-and 2.98-fold, respectively. Relative expression levels of the SA signaling pathwayrelated genes NtPR1 and NtNPR1 were increased in transgenic tobacco lines. The expression of NtPR1 in line W3-44 was 24-fold that of WT, and the expression of NtNPR1 in line W3-35 was 7-fold that of WT. In W3-44, the expression levels of pathogenesis-related protein (PR) genes NtGlu2 and NtCHI were respectively 24-and 8.5-fold higher than that of WT. In addition, the expression level of PR gene Ntosmotin in line W3-20 was 3.4fold higher than that in WT. Furthermore, antioxidant stress-related superoxide dismutase (SOD) genes, including NtSOD, NtCu-ZnSOD, and MnSOD, were all upregulated in transgenic tobacco compared with WT. In conclusion, the overexpression of LrWRKY3 in tobacco up-regulated the expressions of JA/SA signaling pathway-related genes, PRs and SODs.

Transient expression of the LrWRKY3 RNAi vector in L. regale scales increased susceptibility to F. oxysporum
To further analyze the function of LrWRKY3, a LrWRKY3 RNAi construct was transiently expressed in L. regale scales via Agrobacterium tumefaciens mediation, and the scales were then inoculated with F. oxysporum spore suspension. Obvious blackening and decay appeared on the infected scales expressing the LrWRKY3 RNAi construct, but disease symptoms were much less evident on L. regale scales transformed with Agrobacterium tumefaciens containing an empty RNAi vector (Fig. 4a). Calculation of the damage area on F. oxysporum-infected L. regale scales confirmed that L. regale expressing the LrWRKY3 RNAi fragment was less resistant to F. oxysporum infection than the control (RNAi empty vector) (Fig. 4b). In addition, qRT-PCR analysis revealed that the expression level of LrWRKY3 in RNAi fragmentexpressing L. regale was evidently lower than that in control scales, whether inoculated with F. oxysporum or not (Fig. 4c). These data clearly indicate that the decreased expression of LrWRKY3 in L. regale scales enhanced sensitivity to F. oxysporum.

The L. regale defensin gene Def1 was revealed to be a F. oxysporum resistance gene
To explore the regulation of PR gene expression by LrWRKY3, a L. regale defensin gene, LrDef1, was cloned on the basis of transcriptome sequencing data (unpublished). The full-length cDNA of LrDef1 was 501 bp, with a 225-bp coding region, and was predicted to encode a protein containing 74 amino acid residues. The predicted molecular weight of the deduced protein LrDef1 was approximately 8.06 kDa. The LrDef1 promoter sequence (850 bp) was obtained by genome walking. The cDNA sequence of LrDef1 and the sequence of its promoter fragment were given in the supplementary information (Additional file 1). Using the PlantCARE prediction program, we identified many cis-acting elements in the LrDef1 promoter, including the TGACG motif (MeJA response element), W box (ET, SA, and MeJA response element), and ABRE (ABA response, hypersaline, and dark induction element), and a series of cis-elements involved in response to plant hormone signals and abiotic and biotic stresses. The prediction results of cis-acting elements in the promoter sequence of LrDef1 were given in supplementary information (Additional file 2, Table S1).
In the qRT-PCR analysis, LrDef1 transcripts were detected in L. regale roots, stems, leaves, flowers, and scales (Fig. 5a). In particular, LrDef1 was strongly expressed in scales. After F. oxysporum inoculation, the expression of LrDef1 in L. regale roots was rapidly induced, with a peak at 24 h ( Fig. 5b) when the expression level was approximately 4.1-fold of the control. According to these results, LrDef1 is a F. oxysporum infectioninduced gene and is dominantly expressed in L. regale scales. An N-terminal signal peptide was detected in the deduced protein LrDef1, which was predicted to localize in plant cell wall, thus indicating that LrDef1 may be a secretory protein. Moreover, the subcellular localization of LrDef1 also examined by fusion expression with GFP in onion epidermal cells. The green fluorescence signal of fusion gene of LrDef1 and GFP was distributed in the cell wall of onion epidermal cells (Fig. 5c). This result demonstrates that LrDef1 is located in plant cell wall as an extracellular protein.
The recombinant plasmid pET-32(a)-LrDef1-His containing a His-tag was transformed into Escherichia coli strain BL21 (DE3) for heterologous expression. As revealed by SDS-PAGE, the molecular mass of the induced His-LrDef1 fusion protein was consistent with its predicted size, 25 kDa (Fig. 6a). The recombinant protein was purified by Ni-NTA-sepharose (Sangon Biotech, China) column affinity chromatography and imidazole elution buffer (Fig. 6b) and then used in antifungal experiments. Original image of Fig. 6a and Fig. 6b were given in supplementary information (Additional file 3). As shown in Fig. 6c-e, the LrDef1 recombinant protein had different inhibitory effects on the mycelial growth of three fungi: F. oxysporum, F. solani, and Alternaria alternata. The recombinant protein had the strongest inhibitory effect on F. oxysporum, followed by Alternaria alternata and F. solani. Moreover, the antifungal activity increased along with the mass of LrDef1 recombinant protein (Fig. 6f ).
In contrast, the roots of transgenic tobacco lines were only slightly darkened, and the leaves were still fully extended (Fig. 7b). Overexpression of LrDef1 in tobacco thus enhanced resistance to F. oxysporum infection.

The LrWRKY3 recombinant protein specifically bound to the LrDef1 promoter fragment with a W box
The recombinant vector pET-32(a)-LrWRKY3 was constructed and transformed into E. coli strain BL21 (DE3) for heterologous expression to obtain a LrWRKY3 recombinant protein. An electrophoretic mobility shift assay (EMSA) was used to analyze the binding between the LrWRKY3 recombinant protein and probes to reveal whether the specific binding site of LrWRKY3 is W box. As shown in Fig. 8a, lanes 1 and 2 on the EMSA gel both contained biotin-labeled probes designed from the LeDef1 promoter fragment with a W box. Original image of Fig. 8a was shown in the supplementary information (Additional file 3). The band in lane 1, corresponding to biotin-labeled probe with no LrWRKY3 recombinant protein added, was not retarded on the gel. The presence of retarded bands in lanes 2 and 3 after the addition of LrWRKY3 indicates that LrWRKY3 was able to bind to the probes containing W box, which slowed their migration rates on the gel. The band in lane 3 was less delayed than the band in lane 2 because of competition due to the large number of unlabeled probes, which were 50 times more abundant than the biotin-labeled ones. Moreover, the LrWRKY3 protein was unable to bind to the mutant probe in lane 4 and thus no mobility shift was observed. These data fully illustrate that LrWRKY3 specifically binds to the W box.

LrWRKY3 transcriptionally activated LrDef1 in yeast cells
To determine whether LrWRKY3 has transcriptional activation activity, we integrated the ORF of LrWRKY3 into the prey vector pGADT7 AD in a Y1H system followed by co-transformation with the recombinant bait vector pAbAi-pLrDef1 in Y1HGold yeast cells, with the pAbAi-p53 plasmid used as positive control and the pAbAi empty vector as negative control. Yeast cells co-transformed with pGADT7 AD-LrWRKY3 and pAbAi-pLrDef1 vectors were able to grow normally on auxotrophic SD/−Leu/AbA solid medium (Fig. 8b). In contrast, yeast cells co-transformed with pGADT7 AD-LrWRKY3 and bait empty vectors could not grow on the solid medium. These results indicate that pGADT7 AD-LrWRKY3 could be integrated into the yeast genome containing bait plasmid to produce a fusion protein, Gal4-LrWRKY3, capable of recognizing and activating the W box of pLrDef1 in yeast cells. The Gal4-LrWRKY3 protein activated the expression of UR1-C in the recombinant bait plasmid pAbAi-pLr-Def1, which enabled yeast cells to grow normally on SD/−Leu/AbA medium. This outcome suggests that the LrWRKY3 protein can specifically bind to the LrDef1 promoter in yeast and has trans-activation activity.  (Fig. 8c). At the same time, GUS activity in LrWRKY3/pLrDef1 transgenic tobacco was significantly higher than that in pLrDef1 transgenic tobacco. The average GUS activity of four pLrDef1 transgenic tobacco lines (PD-2/3/7/10) was approximately 24, 26, 33, and 31 pM 4-MU min − 1 μg − 1 , respectively. The average GUS activity of four LrWRKY3/pLrDef1 co-expressing transgenic tobacco lines (Co-1/7/9/12) was higher than 40 pM 4-MU min − 1 μg − 1 , namely, approximately 52, 48, 46, and 42 pM 4-MU min − 1 μg − 1 , respectively. Noteworthily, the average activity of Co-1 was approximately 1.7-fold higher than that of PD-7. These results indicate that the specific bindings of LrWRKY3 with W box in LrDef1 promoter activated the expression of downstream reporter gene GUS driven by LrDef1 promoter, thereby enhancing GUS activity.

Discussion
A large amount of evidence indicates that WRKY TFs play important roles in regulating plant resistance to pathogenic fungal infection. To explore the function of WRKY TFs in L. regale, we isolated the WRKY family member gene LrWRKY3 from L. regale in the present study. Structural analysis revealed that LrWRKY3 contains a highly conserved 'WRKYGQK' short peptide and a C 2 H 2 zinc finger, thus indicating that this gene belongs to the IIa WRKY TF group. Nuclear localization is essential for TFs to activate or inhibit expression of their target genes under stress conditions. In subcellular localization experiments, the green fluorescence signal of the LrWRKY3-GFP fusion protein was detected in the nuclei of onion epidermal cells. These results indicate that LrWRKY3 is a nuclear-localized protein that possibly regulates cellular transcriptional reprogramming under biotic stress.
Gene overexpression and gene silencing are commonly used to investigate the role of WRKY TFs in plant defense. Transgenic Arabidopsis plants overexpressing grapevine (V. davidii) WRKY53 show strong disease resistance to Coniella diplodiella, Pseudomonas syringae, and G. cichoracearum [17]. Overexpression of SpWRKY3 in currant tomato (Solanum pimpinellifolium) positively modulates defense response to Phytophthora infestans, as indicated by a reduced number of necrotic cells, lesion size, and disease index; however, the resistance is weakened by SpWRKY3 silencing [18]. The transient silencing of strawberry (Fragaria × ananassa) WRKY1 reduces sensitivity to Colletotrichum acutatum and has indicated that FaWRKY1 plays a negative regulatory role in strawberry resistance to P. scutellum. Even more interesting, FaWRKY1 positively regulates resistance to P. syringae in Arabidopsis [19]. In the present study, reverse genetics was used to verify the function of LrWRKY3. Overexpression of LrWRKY3 in tobacco enhanced resistance to F. oxysporum. At the same time, the transient expression of LrWRKY3 RNAi vector in L. regale scales increased decay and disease area, which indicates that the decreased expression of LrWRKY3 in L. regale scales increased susceptibility to F. oxysporum. These data clearly indicate that LrWRKY3 is an important positive regulator in defense responses of L. regale.
Plant defense systems rely on a complex signal regulatory network. Phytohormones play a crucial role, with SA and JA, in particular, serving as important immune response regulators [20]. WRKY TFs are important regulatory nodes in SA, JA, and ETH signaling pathways [21]. OsWRKY4 enhances rice resistance to the necrotrophic pathogen Rhizoctonia solani and participates in JAmediated pathway [22]. Overexpression of NtWRKY50 in tobacco alters SA and JA contents, increases expression of defense-related genes, and enhances resistance to R. solanacearum [23]. In our previous study, SA, MeJA, Fig. 8 Analysis of the interaction between LrWRKY3 and LrDef1 promoter. a EMSA revealed that the LrWRKY3 specifically bound to W box. 1, reaction solution containing only free probes; 2, LrWRKY3 can interact with biotin labelled probes containing the W box sequence; 3, LrWRKY3 can both interact with biotin labelled probes containing the W box sequence and unlabeled probes containing the W box sequence; 4, LrWRKY3 can't interact with mutant probes; Complex: the combination of LrWRKY3 recombinant protein and probes; Free probe: the unbound probes. b Analysis of the transactivation effect of LrWRKY3 on LrDef1 promoter. LrWRKY3 + pLrDef1: transcriptional activity analysis of pGADT7-LrWRKY3 interaction with pLrDef1; Positive control: pGADT7-p53 can interact with pAbAi-p53 as a positive control; Negative control: pAbAi empty vector was used a negative control. c GUS activity was higher in transgenic tobacco lines co-expressing of LrWRKY3 and LrDef1 promoter than in the transgenic tobacco lines expressing of LrDef1 promoter. The results were shown as average values calculated from three replicates and calculated by the 2 -ΔΔCt method and analyzed by the Student's t test. pBI121: pBI121-GUS transgenic tobacco; PD-2/3/7/10: pLrDef1 transgenic tobacco lines. Co-1/7/9/12: LrWRKY3/pLrDef1 transgenic tobacco lines H 2 O 2 , ETH, and F. oxysporum treatment induced the expression of LrWRKY3 [12]. In the SA-mediated signaling pathway, PR1, which is downstream of NPR1, is positively regulated by NPR1 to acquire resistance [24]. In the present study, transcription levels of NtNPR1 and NtPR1 were significantly increased in LrWRKY3 transgenic tobacco lines. In addition, expression levels of JA biosynthetic pathway-related genes (NtLOX, NtAOC, NtOPR, NtAOS, NtKAT, NtPACX, and NtJMT) were significantly higher in LrWRKY3 transgenic tobacco lines compared with WT tobacco. These results suggest that LrWRKY3 participates in JA-and SA-mediated signal transduction pathways, which are essential in the immune response of L. regale against Fusarium wilt.
The overexpression of OsWRKY80 in rice significantly enhances resistance to sheath blight and up-regulates transcription levels of some disease resistance-related genes, including PR1a, PR1b, PR5, and PR10 in transgenic lines compared with WT plants. The RNAi of OsWRKY80 significantly reduces sheath blight resistance, however, and the above-mentioned disease resistancerelated genes are depressed in OsWRKY80 RNAi lines [25]. In the present study, we also analyzed the expression levels of three PRs (NtGlu2, NtCHI, and Ntosmotin) and three SODs (NtSOD, NtCu-ZnSOD, and MnSOD). These genes were significantly up-regulated in the LrWRKY3 transgenic tobacco lines compared with WT, which suggests that LrWRKY3 modulates the expression of defense-related genes during response to F. oxysporum.
Antimicrobial peptides, including defensins, are the main components of plant innate immune system. Plant defensins, a class of small, cysteine-rich alkaline cationic peptides, are the first line of defense against pathogen invasion [26]. These compounds can bind to fungusspecific membrane components, leading to invagination of the cell membrane of pathogenic fungi; they can also be intercalated or transported into the cell membrane to give rise to a series of biochemical changes [27]. Various plant defensins have been introduced into plants to enhance their resistance. For instance, overexpression of a Panax notoginseng defensin-like gene (PnDEFL1) in tobacco enhanced resistance to F. solani; moreover, the PnDEFL1 recombinant protein showed antifungal activity against F. solani, F. oxysporum, Botrosphaeria dothidea, and S. sclerotiorum in vitro [28]. As another example, overexpression of the NmDef02 reduced the susceptibility of soybean (Glycine max) to Phakopsora pachyrhizi and C. truncatum and improved resistance against diseases [29]. In the present study, a F. oxysporum-responsive defensin gene, LrDef1, was isolated from L. regale. Bioinformatics analysis revealed that LrDef1 possesses a signal peptide corresponding to the characteristics of plant defensins, and subcellular localization experiments confirmed that LrDef1 encodes a cell wall protein. Further analyses revealed that LrDef1 is an antifungal protein, with the recombinant protein LrDef1 able to inhibit the mycelial growth of F. oxysporum, F. solani, and Alternaria alternata. In addition, the overexpression of LrDef1 in tobacco enhanced the resistance of tobacco to F. oxysporum. We thus conclude that LrDef1 is a F. oxysporum-resistance gene in L. regale.
The DNA binding domain can bind to the W box element [(C/T) TGAC (T/C)] in promoter of target genes of WRKYs, with WRKYs in turn activating or inhibiting the expression of downstream genes [30]. The currant tomato WRKY1 activates the non-coding RNA lncrna33732 through sequential-specific interaction with the W box element in promoter; moreover, lncrna33732 can induce the expression of respiratory oxidase and increase the accumulation of H 2 O 2 , thereby enhancing currant tomato resistance to P. infestans [31]. In the present study, we obtained the LrDef1 promoter sequence (850 bp) by TAIL-PCR, and bioinformatics analysis indicated that the LrDef1 promoter sequence had a binding site specific to WRKYs, the W box. The results of subsequent EMSA and yeast one-hybrid experiments leave little doubt that LrWRKY3 can specifically bind to the W box of LrDef1 promoter and has transcriptional activation activity in yeast cells. In addition, GUS activity was significantly enhanced in pLrDef1-GUS/LrWRKY3 co-expressed tobacco compared with transgenic tobacco expressing pLrDef1-GUS alone. All these data strongly support the hypothesis that LrWRKY3 has a transcriptional regulatory effect on the defensin gene LrDef1, an important resistance gene in response to F. oxysporum infection.
The present work will be subsequently explored for further research. Firstly, the LrWRKY3 transgenic tobacco lines will be subjected to transcriptome sequencing and determination of hormone content in order to further reveal the molecular net regulated by LrWRKY3 and verify the crosstalk between LrWRKY3 and hormonemediated signaling pathways. In addition, it is urgent to establish a rapid genetic transformation method of L. regale, and we will be able to thoroughly understanding the regulatory mechanism of WRKY TFs in L. regale during response to Fusarium wilt in the near future.

Conclusion
A nuclear-localized IIa WRKY TF gene, WRKY3, was isolated from L. regale. Overexpression of LrWRKY3 in tobacco improved resistance to F. oxysporum; however, RNAi of LrWRKY3 in L. regale scales conferred sensitivity to F. oxysporum infection. Moreover, LrWRKY3 activated JA and SA signal pathways and up-regulated the expression levels of PRs and SODs in the transgenic tobacco lines. In addition, LrWRKY3 showed transactivation on a F. oxysporum-resistance gene, LrDef1. We thus conclude that LrWRKY3 modulates the antimicrobial peptide gene LrDef1 and JA/SA signal pathways in L. regale during response to Fusarium wilt.

Fungal and plant materials
Wild L. regale was collected in Wenchuan County, Sichuan Province, China, and was identified by our research group through referring to a L. regale specimen (Kun 222,503) preserved in the herbarium of Kunming Institute of Botany, Chinese Academy of Sciences, and then was further confirmed by Prof. Zhineng Li from Southwest University. The L. regale was planted in a greenhouse (25 °C, 60% relative humidity) at Kunming University of Science and Technology. Wild type tobacco (N. tabacum) seeds were provided by Dr. Guanze Liu from Yunnan Agricultural University, Kunming, China. Sterile tobacco seedlings were cultured in a climatic cabinet (25 °C, 16 h light/8 h dark cycle) and used for genetic transformations. A F. oxysporum strain from a diseased lily with typical symptoms of root rot was characterized and preserved by our research group. F. oxysporum, F. solani, and A. alternata were conserved at 4 °C and activated through culturing on potato dextrose agar medium plate for one week at 28 °C before use.

Cloning of LrWRKY3 and LrDef1
Total RNA was extracted from L. regale roots, and cDNA was synthesized with a reverse transcription kit (Promega, USA) to isolate LrWRKY3 and LrDef1. The methods used for RNA extraction and cDNA synthesis were the same as those detailed in Liu et al. [32]. The ORFs of LrWRKY3 and LrDef1 were cloned by PCR with specific primers (Table S1), and the PCR products were cloned into pGEM-T vector (Promega). The gene-specific primer sequences were given in the supplementary information (Additional file 2, Table S2). Positive E. coli clones containing pGEM-T-LrWRKY3 and pGEM-T-LrDef1 were confirmed by PCR.

Quantitative reverse transcription PCR (qRT-PCR)
The expression patterns of LrDef1 in L. regale were analyzed by qRT-PCR. The roots of healthy L. regale plants were injured and then infected with F. oxysporum (2 × 10 6 spores/mL) for 30 min. The roots were collected 24, 48, and 72 h after inoculation. The mimic inoculation with sterile distilled water was used as a control. Roots, leaves, stems, flowers, and scales of L. regale under normal growth conditions were collected to analyze the expression pattern of LrDef1. Total RNA was extracted from each sample and used for cDNA synthesis. qRT-PCR amplification conditions and instrumentation were the same as in Liu et al. [32]. The expression level of L. regale glyceraldehyde-3-phosphate dehydrogenase (LrGAPDH, GenBank no. JZ391059) was used as an internal control to standardize different RNA samples. The qRT-PCR gene-specific primers sequences were given in the supplementary information (Additional file 2, Table S3). Three biological replicates were performed for each qRT-PCR assay, and the 2 −ΔΔCt method was used to calculate relative gene expression levels.

Subcellular localization analysis
The subcellular locations of LrWRKY3 and LrDef1 were predicted using the PSORT online program (https:// www. gensc ript. com/ psort. html) and verified by transient expression of a GFP-tagged fusion protein in onion epidermal cells. The ORFs of LrWRKY3 and LrDef1 without stop codons were digested by restriction enzymes and then ligated to a pBIN M-gfp5-ER vector using T4 ligase for fusion with GFP. The pBIN m-gfp5-ER-LrWRKY3 and pBIN m-gfp5-ER-LrDef1 recombinant plasmids were individually transferred into Agrobacterium tumefaciens EHA105 via the freeze-thaw method. An empty pBIN m-gfp5-ER vector served as a control. The transformation protocols into onion epidermis are detailed in Qiu et al. [33]. Confirmation of LrWRKY3 and LrDef1 subcellular localizations was performed with a confocal laser scanning microscope (Nikon, Japan). The GFP fluorescence was detected using FITC channel with excitation wavelength of 488 nm, and the red fluorescence of PI was detected using Cy5 channel with excitation wavelength of 655 nm. The localizations of LrWRKY3 and LrDef1 were determined according to the GFP fluorescence distribution of two fusion proteins in the onion cells.

Construction of LrWRKY3 and LrDef1 plant overexpression vectors and genetic transformation of tobacco
The pCAMBIA2300s vector was used to construct plant overexpression vectors for LrWRKY3 and LrDef1. The ORFs of LrWRKY3 and LrDef1 were respectively obtained from pGEM-T-LrWRKY3 and pGEM-T-LrDef1 by double enzyme digestion. The enzyme digestion products were then ligated with the pCAMBIA2300s vector, which was digested with the same restriction enzyme. The ligation products were transferred into E. coli DH5α competent cells, and positive clones were screened by PCR. The pCAMBIA2300s-LrWRKY3 and pCAMBIA2300s-LrDef1 plasmids were separately transferred into A. tumefaciens LBA4404. Positive clones were identified by PCR and used for genetic transformation of tobacco according to Horsch et al. [34]. The positive transgenic tobacco seedlings were confirmed by PCR with two pairs of gene-specific primers, with the WT included in the PCR analysis as a negative control. In addition, some transgenic tobacco plants were randomly selected for selfbreeding cultivation T 2 generation in a greenhouse.

Gene expression and disease resistance analyses of transgenic tobacco
Transcription levels of LrWRKY3 and LrDef1 in young leaves of T 2 generation transgenic tobacco were analyzed by qRT-PCR, with the tobacco actin gene (NtACT , GenBank no. AB158612.1) used as an internal reference. Four lines of T 2 generation transgenic tobacco were selected to study their resistance to F. oxysporum infection. The leaves of WT tobacco and T 2 generation transgenic tobacco seedlings were injured with sandpaper and inoculated with 100 μL F. oxysporum spore suspension (2 × 10 6 spores/mL). The infected leaves were laid flat on wet filter paper to maintain humidity and placed in an illumination incubator. The injured roots of T 2 generation LrWRKY3 transgenic tobacco and WT tobacco seedlings were immersed in F. oxysporum spore suspension (2 × 10 6 spores/mL) for 30 min and then transferred to half-strength Murashige-Skoog liquid culture medium. The inoculated tobacco samples were collected after 7 days, and disease symptoms were recorded with a digital camera (Nikon). Lesion areas were measured with Photoshop. The leaves of LrDef1 T 2 generation transgenic tobacco were inoculated with F. oxysporum for disease resistance analysis using the above-mentioned approaches.

Transient expression of LrWRKY3 RNAi construct in L. regale scales
Gene-specific primers (Table S1) of LrWRKY3 with an attB linker were designed to amplify a RNAi fragment (410 bp). The LrWRKY3 RNAi PCR products were recombined with pHellsgate2 vectors using Gateway BP Clonase II enzyme mixture (Invitrogen, USA). The BP reaction mixture was then transformed into E. coli DH10B competent cells. The pHellsgate2-LrWRKY3 recombinant plasmids were further transformed into A. tumefaciens EHA105, and the positive clones were identified by PCR. The scales of healthy L. regale plants were wounded with sandpaper, and A. tumefaciens suspension (50 μL) containing pHellsgate2-LrWRKY3 or pHellsgate2 empty vector was then added. The infected scales were placed in a climate chamber at 28 °C for 24 h for the transient expression of RNAi vectors in L. regale scales, with subsequent F. oxysporum inoculation with spore suspension (2 × 10 6 spores/mL) at the wound sites. Symptoms and lesion areas of lily scales were recorded and measured after inoculation for 72 h. In addition, the expression levels of LrWRKY3 in scales were measured by qRT-PCR.

Prokaryotic expression and antifungal activity analysis of LrDef1 recombinant protein
The SignalP-5.0 program (https:// www. cbs. dtu. dk/ servi ces/ Signa lP/) was used to predict the signal peptide in LrDef1. The PCR-amplified LrDef1 ORF without the signal peptide-encoding sequence but containing the restriction sites of EcoRV and EcoRI was subcloned into the prokaryotic expression vector pET-32(a) to produce 6 × His-labeled fusion protein. The pET-32(a)-LrDef1 construct was then transformed into E. coli BL21 (DE3) for heterogenic expression. The LrDef1 recombinant protein was purified by Ni-NTA-sepharose affinity chromatography. F. oxysporum, F. solani, and A. alternata were used to analyze the antifungal effects of the LrDef1 recombinant protein using the method described by Taif et al. [35].

Isolation of the LrDef1 promoter fragment
The LrDef1 promoter fragment was isolated using a Genome Walking kit (Takara, Japan) according to the kit instructions. Two nested gene-specific primers and a degenerate primer were used in two successive PCR rounds to amplify the LrDef1 promoter region. The secondary PCR products were cloned into pGEM-T vector and then transformed into E. coli DH5α. Positive clones were selected for sequencing. In addition, the cis-elements in promoter were predicted using the PlantCARE program (http:// bioin forma tics. psb. ugent. be/ webto ols/ plant care/ html/).

Prokaryotic expression of the LrWRKY3 recombinant protein and EMSA
A pair of gene-specific primers (Table S1) were designed to amplify the LrWRKY3 ORF, which was then subcloned into the pET-32(a) vector. The recombinant pET-32(a)-LrWRKY3 plasmids were further transformed into E. coli BL21 (DE3) to express the recombinant protein containing a 6 × His tag. Detailed protocols on the recombinant protein denaturation and renaturation of inclusion body protein are given in Zhao et al. [36].
The EMSA was used to confirm the specific binding of LrWRKY3 recombinant protein with the LrDef1 promoter fragment. A probe containing the W box sequence of LrDef1 promoter was designed and labeled with biotin (Sangon Biotech) and is hence referred to as the biotin-labeled probe (Table S1). In addition, a W box sequence containing a mutation was labeled with biotin and used as a mutant probe. Moreover, a probe without biotin labeling, designated the unlabeled probe, was used as the competitor. The biotin-labeled and mutant probes were diluted to a concentration of 0.01 μM, and the final concentration of unlabeled probe was 0.5 μM. Equivalent amounts of sense and antisense fragments of these probes were annealed to synthesize double-stranded DNA. After addition of the biotinlabeled probes, the EMSA reaction mixture including 0.5 μg purified LrWRKY3 protein was incubated for 20 min at 25 °C and then subjected to gel electrophoresis. For the competition experiment, the unlabeled probe was added as a competitor in the binding reaction and incubated for 10 min before addition of the biotin-labeled probe. The reaction mixture was incubated at room temperature for another 15 min prior to loading onto the gel. EMSA was performed according to the standard procedures of a Light Shift Chemiluminescent EMSA kit (Pierce, USA).

Yeast one-hybrid assay
A yeast one-hybrid system (Takara) was used to analyze the transactivation of LrDef1 by LrWRKY3. The 500bp LrDef1 promoter fragment containing a W box was ligated to the bait vector pAbAi, and the ligation product pAbAi-pLrDef1 was then transformed into E. coli DH5α. Positive clones containing the pAbAi-pLrDef1 recombinant plasmids were selected by PCR, and the recombinant plasmids were transformed into yeast cell Y1Hgold to obtain the yeast bait strains according to the method described in Li et al. [12]. In addition, the LrWRKY3 ORF was ligated to the prey vector pGADT7 AD and transferred into E. coli DH5α. The yeast prey recombinant vector pGADT7 AD-LrWRKY3 was then transferred into bait strain pAbAi-pLrDef1-Y1HGold competent cells. Moreover, the yeast prey empty vector pGADT7 AD was transferred into bait strain pAbAi-pLrDef1-Y1HGold competent cells as a negative control, and the pGADT7 AD-P53 vector (kit provided, Takara) was transferred into bait vector pAbAi-p53-Y1HGOLD competent cells as a positive control. The appropriate inhibitory concentration of aureobasidin A (AbA) against the bait reporter strains was determined according to the manufacturer's instructions (Takara). The resulting three groups of yeast cells were streaked onto SD/−Leu/AbA solid medium, and the plates were placed in a 30 °C incubator for 2-3 days to observe whether LrWRKY3 activated the expression of LrDef1 promoter.

Co-expression of the LrDef1 promoter and LrWRKY3 in tobacco
To verify whether the LrWRKY3 TF has a regulatory effect on LrDef1 in plants, the LrDef1 promoter was transformed into LrWRKY3 transgenic tobacco to obtain the co-expression tobacco. The CaMV 35 s promoter in pBI121 vector was replaced by LrDef1 promoter to construct a plant expression vector that included a GUS gene driven by LrDef1 promoter. The pBI121-pLrDef1-GUS plasmids were transformed into A. tumefaciens LBA4404 competent cells and then transformed into WT or LrWRKY3 T 2 generation transgenic tobacco leaf discs. In addition, the WT tobacco leaf disks were transformed with the empty pBI121-GUS vector as a control. Positive transgenic tobacco plants were identified by PCR with GUS-specific primers. A fluorescence spectrophotometer (Hitachi f-4600, Japan) was used to determine the GUS enzyme activity (pM 4-MU min − 1 μg − 1 protein) of positive transgenic tobacco as described by Chen et al. [37].

Data analyses
Gene expression levels, lesion sizes, fungal growth inhibition zones, and GUS activities were calculated as means plus standard deviations. Student's t-test was performed using SPSS software (version 17.0) to reveal statistical differences between treatments, inoculations, and controls.