The Contig3867, a predicted ERF1-V gene, was identified as a Bgt induced gene in H. by GeneChip microarray as described by Cao et al. (2011). The leaves of H. villosa at the two-leaf stage were inoculated with Bgt, or treated with 200 μmol/L ethephon (ET), 25 μmol/L jasmonic acid (JA), 5 mmol/L salicylic acid (SA), 50 μmol/L abscisic acid (ABA), 16% (w/v) PEG–6000, 200 mmol/L NaCl or 200 mmol/L CaCl₂ for analyzing the response of ERF1-V to different stresses using the real-time quantitative reverse transcription–PCR (qRT–PCR). The treated leaves were sampled at time points 0, 12, 24, 48, and 60 h after Bgt, ET, JA, and SA treatment and at time points as 0, 1, 6, 12, and 24 h after ABA, PEG-6000, CaCl₂, NaCl, and 4°C treatment. Three were three independent biological replicates. Total RNA (2 μg) of each sample was synthesized to first-strand cDNA using AMV reverse transcriptase. The primers of ERF1-V used for qRT–PCR were ERF1-V-QRT-F and ERF1-V-QRT-R, and the primers of the control gene Tubulin were Tubulin-F and Tubulin-R (Supplementary Table S1). The reagents were purchased from TAKARA Co. (Japan), and the reaction was performed in an ABI 7500fast (ABI, United States).

The cDNA library of H. villosa inoculated with Bgt was previously constructed, and the pooled clones were stored in the 384 well plates (Chen et al., 2007). The full length ERF1-V gene was isolated from the cDNA library by PCR analysis using the primers designed according to the probe Contig3867 (forward primer ERF1-V-F and reverse primer ERF1-V-R, Supplementary Table S1). The mixed plasmid DNA of cDNA library pool were isolated and used as templates, and the positive pool containing ERF1-V was identified using PCR screening. The mixed plasmids of the positive pool were extracted and transformed into E. coli, and the positive single clone containing ERF1-V was screened using colony PCR with the same primers.

The conserved domain of ERF1-V was predicted by Prosite¹, and the protein structure of ERF1-V was predicted by SWISS-MODEL². Forty ERF proteins from ten species, including Saccharum officinarum, Solanum lycopersicum, Triticum aestivum, Glycine max, Gossypium hirsutum, Capsicum annum, Nicotiana tabacum, Thinopyrum intermedium, Arabidopsis thaliana, Orazy sativa, were initially aligned and then used to construct the phylogenetic tree. Multiple sequence alignments were performed using ClustalX (ver.1.83), and a phylogenetic tree was constructed by the Maximum likelihood (ML) method in MEGA6 ³. The bootstrap test of phylogeny was performed with 1,000 replications.

The ERF1-V gene was amplified using the forward primer ERF1-V-EcoRI-F and the reverse primer ERF1-V-SmaI-R (Supplementary Table S1). The PCR product and plasmid pGBKT7 (CLONTECH Co., United States) were treated with EcoRI and SmaI enzymes (TAKARA Co., Japan), respectively, followed by ligation to construct the recombinant vector pGBKT7:ERF1-V. The recombinant vector and the negative control pGBKT7 were transformed into the wild yeast cells AH109 (CLONTECH Co., United States), respectively, and cultured in Trp lacking media. The clones grown in the Trp lacking media were then transferred to the Trp and His lacking media with X-gal to test the transcriptional activation activity.

The ERF1-V gene was amplified with the forward SmaI-containing primer and the reverse SacI-containing primer (ERF1-V-SmaI-F and ERF1-V-Sacl-R, Supplementary Table S1). The PCR product and the plasmid pAHC 25 (Christensen and Quail, 1996) were digested with SmaI and SacI enzymes, followed by ligation to construct the recombinant vector pAHC:Hv-ERF. The transgenic plants over-expression ERF1-V were produced by particle bombardment of callus cultured from immature embryos of Bgt-susceptible variety Yangmai158 (Xing et al., 2008). The positive transgenic plants were identified by PCR analysis of the selective gene Bar (Bar-F and Bar-R, Supplementary Table S1) and the target gene ERF1-V (forward primer Ubi-F was located in the Ubi-promoter and reverse primer ERF1-V-TR which was located in the ERF1-V gene, Supplementary Table S1) in plants regenerated from the bialaphos-containing culture media. Only when both the Bar gene and ERF1-V could be amplified, the plants were considered as positive transgenic plants. The gene expression level of ERF1-V was further analyzed by semi-quantitative RT-PCR (ERF1-V-RT-F and ERF1-V-RT-R, Supplementary Table S1), and GMY27, GMY60, GMY88 and GMY90 were identified as ERF1-V over-expressed lines when the expression of ERF1-V in these liens were compared with that in Yanamai158. The T₁ to T₆ generations of these four lines were used for powdery mildew resistance evaluation, the T₅ generation plants were used for drought and salt tolerance evaluation, and the T₅ to T₆ generation plants were used for agricultural traits comparison.

The positive transgenic plants of T₁ lines generated from the identified T₀ individuals over-expression ERF1-V were again screened by PCR analysis of both the Bar gene and the Hv-ERF gene. The positive transgenic plants were used to evaluate resistance to powdery mildew. The T₂ generation individuals derived from the transgenic T₁ plants were molecular characterized again and re-evaluated of their responses to the powdery mildew to reveal the contribution of the ERF1-V to the powdery mildew resistance. At the seedling stage, the powdery mildew resistance levels were evaluated using the detached leaves inoculated with a mixture Bgt collected from Eastern China. At the adult stage, the resistance levels were also recorded for the corresponding adult plants in the greenhouse. The resistance levels of the transgenic plants were continuously observed each year in the following T₃ to T₆ generation. The level of resistance at the seedling stage was classified as grades 0 to 4 according to the standard of Sheng (1988), and the level of resistance in the adult stage was classified as grades 0 to 9.

It was induced from the molecular analysis and resistance evaluation in T₁ and T₂ generation that the transgenic plants showed higher level of the powdery mildew resistance. Then, T₃ and T₄ generation individuals derived from the transgenic T₂ plants with improved resistance were evaluated of the powdery mildew resistance, and the homozygous lines without resistance separation in T₅ generation were selected for drought and salt tolerance evaluation. Healthy and plump seeds of ERF1-V transgenic lines and the recipient control Yangmai158 were immersed in water for 10 h at 25°C. The seeds were then transferred to Petri dishes layered with two damp sheets of filter papers. Three days later, the uniform seedlings were further transferred to a 100-well box for culturing with 1/2 Hoagland nutrition solution. This was replaced with fresh solution every 2 days. The seedlings were cultured using 16 h light (25°C)/8 h dark (22°C) cycle with 80% humidity. For drought tolerance evaluation, the seedlings were treated with 20% PEG-6000 containing 1/2 Hoagland solution for 72 h at the three-leaf stage, and then were re-watered with 1/2 Hoagland solution. Ten seedlings were sampled for fresh weight and dry weight before and after re-watering, and the mean values were calculated based on three repeats. For salt tolerance evaluation, the seedlings were treated with 150 mmol/L NaCl-containing 1/2 Hoagland solution at the three-leaf stage. This was replaced with fresh solution every 2 days. At the five-leaf stage, ten seedlings were sampled for fresh and dry weight of the root and the leaf, and the mean values were calculated based on three repeats. All data were statistically analyzed using SPSS 16.0 software (Norusis, 2008).

The physiological changes and the gene expression analysis were compared between the transgenic plants and Yangmai158. Chlorophyll and malondialdehyde (MDA) content, and superoxide dismutase (SOD) and peroxidase (POD) activity were measured when phenotypic differences were observed between the transgenic plants and Yangmai158 after salt treatment (Rong et al., 2014; Zhu et al., 2014). Gene expression of TaGSK (TaGSK-F and TaGSK-R, Supplementary Table S1), TaP5CR (TaP5CR-F and TaP5CR-R, Supplementary Table S1), TaHKT (TaHKT-F and TaHKT-R, Supplementary Table S1), TaNHX (TaNHX1-F and TaNHX1-R, Supplementary Table S1) and TaOAT (TaOAT-F and TaOAT-R, Supplementary Table S1) involved in the salt or drought tolerance pathway were measured 72 h after PEG-6000 or NaCl treatment in the transgenic plants and Yangmai58. All data were statistically analyzed using SPSS 16.0 software (Norusis, 2008).

The T₅ generation plants used for drought and salt tolerance evaluation and the derived T₆ generation plants without resistance separation were used for agricultural traits comparison. The plants were grown in 1.5 m rows spaced 25 cm apart, with 30 seeds planted per row. Agronomic traits, including plant height, flowering time, tiller number, spikelet number and grain size, were compared between the four transgenic lines and the recipient Yangmai158 at different developmental stages. The experiments were conducted in 2 years using T₅ and T₆ generation plants, respectively. In each year, ten plants were sampled and measured for one replicate, the experiments were each repeated three times. All phenotypic data were statistically analyzed using SPSS 16.0 software (Norusis, 2008).

Haynaldia villosa, which contains the Pm21 gene, confers broad spectrum resistance to powdery mildew (Chen et al., 1995). To identify the resistance related genes and to study the resistance mechanism, the transcription profiles of H. villosa were obtained by GeneChip microarray, and the data were compared between the Bgt inoculated sample and the uninoculated sample. Many differentially expressed genes were isolated with a signal ratio of more than 2. The ethylene-responsive element-binding factor gene ERF1-V, corresponding to the probe of Contig3867, was characterized as a Bgt induced gene. The following qRT-PCR was conducted and the results showed that ERF1-V was a real Bgt responsive gene (Figure 1).

AP2/ERF is widely involved not only in the biotic stress response but also in the abiotic stress response, and H. villosa shows excellent tolerance to abiotic stress, such as salt and drought (Blanco et al., 1983). So, it is interesting to see whether the ERF1-V takes part in the responses of H. villosa to abiotic stresses. The qRT-PCR analysis indicated that ERF1-V was induced by NaCl, PEG-6000 and cold treatment (Figure 1).

The SA, JA, and ethylene (ET) are signal molecules commonly involved in biotic stress response (Yang et al., 2015), and the ABA is a key signal molecule of abiotic stress response (Yoshida et al., 2014). Gene expression analysis showed that ERF1-V was up-regulated by ET and ABA, but down-regulated by SA and JA (Figure 1). Furthermore, calcium, another signaling molecule, induced the expression of ERF1-V (Figure 1). These results indicated that the ERF1-V might be involved in different signal pathways.

Among the subfamilies of AP2/ERF, the RAV subfamily has both B3 domain and AP2/ERF domain. The AP2 subfamily possess two repeated domains, while the members of the DREB and ERF subfamilies have a single AP2/ERF domain (Yamasaki et al., 2013). To characterize which subfamily ERF1-V belongs to, the full length of ERF1-V was cloned by screening the cDNA libraries of H. villosa, and the gene structure and phylogenetic analysis of ERF1-V was conducted. ERF1-V is a 1558 bp gene which contains an 1185 bp ORF corresponding to a 394 amino acid protein (Accession Number: ACN58181). Using the ExPASy software, it was found that ERF1-V contains only a single AP2/ERF conserved domain from 129R to 189P (Supplementary Figure S1a). Analysis by SWISS-MODEL showed that there were three β-sheets and one α-helix in the conserved domain (Supplementary Figures S1a,b). Forty AP2/ERF genes from different subfamilies of 10 species were used for construction of the phylogenetic tree, and the results showed that the ERF1-V belonged to the B2 ERF subfamily (Supplementary Figure S2). Multiple sequence alignment using different B2 type genes indicated that the AP2/ERF domain was conserved among the B2 family. Furthermore, conserved ‘A’ and ‘D’ specific to the ERF subfamily were presented in the ERF1-V. However, these two amino acids were ‘V’ and ‘E,’ respectively, in most of the DREB subfamily members (Supplementary Figure S1a). The phylogenetic tree showed that TaERF1 and TaERF2 are B2 type, TaERF3 are B3 type, and TaERF4 are B1 type (Supplementary Figure S2).

The transcription factor GAL4, which containing the DNA BD and the transcription activation domain (AD), can bind to the promoter of the downstream genes by BD domain and then activate the downstream gene by AD domain. In the yeast two-hybrid system, if the AD domain of GAL4 was replaced by another protein with transcriptional activation activity, then the transcription of the downstream reporter genes can also be activated. This system is often used to test whether a target gene has transcriptional activation activity. In this study, the ERF1-V was inserted into the plasmid pGBKT7 to replace the AD domain of the GAL4. The recombinant plasmid pGBKT7:ERF1-V was then transformed to the yeast cell AH107, and the positive transformed cell grew in the Trp and His lacking culture media, indicating the reporter gene His synthesis gene was activated. The positive cells were stained with X-gal, and the result showed that the reporter gene LacZ was activated (Figure 2). This indicated that ERF1-V had transcriptional activation activity.

Since ERF1-V was induced by Bgt in H. villosa, the role of the ERF1-V in powdery mildew resistance was then evaluated using transgenic plants overexpression ERF1-V. Six T₀ positive transgenic plants were identified as containing both the selective gene Bar and the target gene ERF1-V. Increased expression of ERF1-V was identified in four lines, including GMY27, GMY60, GMY88, and GMY90 (Supplementary Figure S3). The positive T₁ individuals and the following T₂ to T₆ generation of the four T₀ transgenic lines were used for powdery mildew resistance evaluation both at the seedling stage and the adult stage. At the seedling stage evaluation, the detached leaves of the recipient control Yangmai158 showed high susceptibility with grade 4 infection type after inoculation with Bgt on which were covered with high density of Bgt spore colony. The detached leaves of T₁ generation positive individuals showed increased resistance to Bgt, on which were covered with reduced density of spore colony. The leaves of transgenic lines GMY88 and GMY90 showed high resistance to Bgt with grade 1-2 infection type (Figure 3A). At the adult stage evaluation, resistance was evaluated from generations T₂ to T₆, and the resistance level of the four transgenic plants ranged from grade 1–5, while the resistance level of the recipient control Yangmai158 ranged from grades 7–8 (Figure 3B). Over-expression of the ERF1-V gene could enhance the resistance of transgenic plants and their progenies to Bgt.

ERF1-V was rapidly up-regulated after the PEG-6000 treatment, therefore T₅ generation of the four transgenic lines were tested for their tolerance to drought stress. At the three-leaf stage, the treated plants of the four transgenic lines and Yangmai158 were transferred to 1/2 Hoagland culture solution supplied with 20% PEG-6000. The control plants from the same lines were transferred to the fresh 1/2 Hoagland culture solution. After 3 days, there was no difference between the non-treated Yangmai158 and transgenic plants, and all the plants developed well (Figure 4a). While, large differences were observed between the PEG-6000 treated plants Yangmai158 and four transgenic lines. The control Yangmai158 was highly wilted and dry. However, the GMY88 transgenic plants developed normally and the GMY90, GMY60, and GMY27 transgenic lines were only slightly wilted and remained green (Figure 4b and Supplementary Figure S4a). All the PEG treated plants were then transferred to fresh 1/2 Hoagland culture solution for re-watering to test the ability of recovery After 1 day, even greater differences were observed between the Yangmai158 and the transgenic plants. The Yangmai158 did not recover and remained dry, while the four transgenic lines completely recovered (Figure 4c and Supplementary Figure S4b).

Biochemical data were collected to study the role of ERF1-V in drought tolerance. Three days after PEG-6000 treatment, the leaf and reeo fresh weight of Yangmai158 and the transgenic plants were significantly difference (Figure 5). One day after re-watering, the four transgenic plants recovered well and there was a dramatic increase in fresh weight, while the fresh weight of Yangmai158 did not significantly change (Figure 5). Drought tolerance and recovery of the transgenic plants were improved by ERF1-V.

ERF1-V was rapidly up-regulated after the NaCl treatment. Therefore, the transgenic lines GMY88 and GMY90, which showed the best drought stress tolerance among the four lines, were further used to evaluate tolerance to salt stress. There was no significant difference between the Yangmai158 and the transgenic lines before NaCl treatment. However, there were significant differences between the Yangmai158 and the transgenic plants after NaCl treatment for plant height, root length, fresh weight and dry weight of leaf, and dry weight of root (Figure 6 and Supplementary Figure S5). These results indicated that ERF1-V could enhance the tolerance of wheat to salinity.

Transcription factors can regulate many downstream genes participating in osmosis adjustment and detoxification. To elucidate pathways activated by ERF1-V, the biochemical data and gene expression pattern of transgenic line GMY90, which showed the best tolerance to salinity and drought, were collected and compared with that of Yangmai158.

Abiotic stresses can lead to the generation of reactive oxygen species (ROSs) which induce secondary oxidative stress. Abiotic stress tolerant species are usually efficient at ROSs scavenging (Mittler, 2002). In this study, the activity of the superoxide dismutase (SOD) and peroxidase (POD), two crucial antioxidant enzymes, significantly increased in GMY90 when treated with NaCl compared with the untreated control plants. No significant difference in SOD and POD activity was observed between the treated and untreated Yangmai158 plants (Figure 7).

Higher MDA content reflected greater membrane damage. In Yangmai158, MDA content increased dramatically after NaCl treatment, while there was no such increase in the NaCl-treated transgenic plants GMY90 (Figure 7). This result indicated that the abiotic stress tolerance was increased by over-expression of ERF1-V which then reduced the membrane damaging. Lower chlorophyll content can be an indicator of extensive damage. Chlorophyll content was higher in the GMY90 plants compared with Yangmai158 following salt stress (Figure 7).

The expression of six stress responsive genes, including P5CR and OAT involved in the Proline (Pro) synthesis related to osmosis adjustment, GSK (glycogen synthase kinase 3) involved in salt signaling, HKT (Na⁺ transporter gene) and NHX (vacuolar membrane transporters) involved in the cation transport, were analyzed by qRT-PCR. The results showed that GSK, P5CR, HKT, and NHX were induced by NaCl in the Yangmai158 and GMY90, but in GMY90 the expression levels of four genes were higher than that in the Yangmai158 (Figure 8). P5CR and OAT were induced by PEG-6000 in both materials, but in GM90, the expression levels were higher than in the Yangmai158 (Figure 8).

The over-expression of TFs usually has negative effects on plant development. In this study, to find out whether the over-expression of ERF1-V could affect the growth of wheat, several main agronomic traits, including plant height, flowering time, tiller number, spikelet number and grain size, were compared between the transgenic plants and the wild recipient Yangmai158. The transformation of the ERF1-V did not impact flowering time, spikelet number, tiller number and grain size (Figure 9 and Supplementary Figure S6). Yet, plant height of GMY88 and GMY90 were slightly increased, and spike length of the GMY27 was slightly reduced.