Transcriptome-based identification and expression profiling of AP2/ERF members in Caragana intermedia and functional analysis of CiDREB3

The AP2/ERF transcription factor family plays important roles in regulation of plant growth and development as well as the response of plants to stress. However, there are currently few studies focusing on the function of the AP2/ERF-type transcription factors in Caragana intermedia Kuang et H. C. Fu. Here, the expression pattern of AP2/ERF transcription factors family in different tissues and under four stress treatments were evaluated, and the function of CiDREB3 was examined. In this study, the genes encoding the AP2/ERF family of transcription factors were screened from the C. intermedia drought transcriptome database and subjected to bioinformatic analysis using the online tool and software. The expression pattern of the members of AP2/ERF transcription factors in C. intermedia were detected via quantitative real-time PCR (qRT-PCR). The function of CiDREB3 on growth, development and drought tolerance was evaluated by transgenic Arabidopsis. As a result, 22 sequences with complete ORFs were obtained and all sequences were divided into 13 sub-groups. Most of the AP2/ERF transcription factors exhibited tissue-specific expression and were induced by cold, heat, NaCl and mannitol treatments. Furthermore, heterologous expression of CiDREB3 altered the morphology of the transgenic Arabidopsisthaliana L. Heynh and improved its drought tolerance during seedlings development. Taken together, the results of the present study helped to better understand the function of the AP2/ERF family transcription factors in response to multiple abiotic stresses and uncovered the role of CiDREB3 in affecting the morphology and abiotic stress tolerance of Arabidopsis.


Introduction
Transcription factors play important roles in plant growth and development and response to various environmental stresses. The Apetala2/ethylene response factor (AP2/ERF) transcription factors are a class of important plant transcription factors that are involved in the regulation of plant growth, development and stress response [1]. The AP2/ERF transcription factors contain the conserved AP2 domains consisting of approximately 57-70 amino acid residues. The AP2 domain is composed of an amphipathic α-helix and three antiparallel β-sheets. The β-sheets are the binding region for cis-acting elements [2]. In addition, each AP2 domain contains two conserved elements, YRG and RAYD. The YRG element is located at the N-terminus of the AP2 domain and consists of approximately 19-22 amino acid residues. The RAYD element is located at the C-terminus of the AP2 domain and consists of approximately 43 amino Kun Liu, Qi Yang and Tianrui Yang have contributed equally to this work. 1 3 acids. These two elements play an important role in binding of AP2/ERF transcription factors to various cis-elements [3]. For the DREB and ERF transcription factor subfamilies, the 14th and 19th amino acids in the second β-sheet determine the binding specificity of the transcription factors to different cis-elements [2,4].
Based on sequences similarity among the AP2/ERF-type transcription factors, the number of AP2 domains and the number of other domains, the AP2/ERF transcription factor family in Arabidopsis is divided into 5 subfamilies, namely, the AP2, DREB, ERF, RAV and Soloist subfamilies [4]. The AP2 subfamily contains two conserved AP2 domains, while the RAV subfamily contains one AP2 domain and one B3 domain. The DREB and ERF subfamilies each contain only one AP2 domain. Other subfamilies also contain one AP2 domain, which, however, does not contain the WLG motif [5]. The DREB and ERF subfamilies are each divided into 6 groups. Specifically, the DREB subfamily is divided into the A1-A6 groups, while the ERF subfamily is divided into the B1-B6 groups [4]. The nuances in the sequence and structure of the AP2 domain directly affected the specificity of binding to the downstream cis-acting elements [6].
To date, a number of plant AP2/ERF transcription factors have been reported. On the basis of the research conducted by Li et al. [7], we summarized the species that expressed AP2/ERF family of transcription factors reported so far ( Table 1) [4,5].
In 1994, Jofuku et al. [8] first discovered that the AP2/ ERF transcription factors regulated floral meristem formation, floral organ formation, flower development-related gene expression and seed development in Arabidopsis. In 1995, Ohme-Takagi et al. [9] isolated certain transcription factors from tobacco and other plants, and they specifically recognized the GCC-box in the promoter region of ethylene-inducible pathogenesis-related protein genes and participated in the ethylene response process. The transcription factor genes were named EREBP-1, EREBP-2, EREBP-3 and EREBP-4 according to their encoded products. Thereafter, an increasing number of reports were published regarding the function of the AP2/ERF transcription factors. Moreover, the study of the AP2/ERF transcription factors includes an increasingly wide range of species, such as Arabidopsis [10], Nicotiana tabacum L. [11] and Glycine max Merr. [12]. There are also reports related to other plants such as Triticum aestivum L. [13], Vitis vinifera L. [14] and Malus sieversii (Ledeb.) Roem [15].
A number of reports have demonstrated that the AP2/ERF transcription factors widely participated in the regulation of growth, development and many other biological processes in plants, such as root initiation and formation [16,17], fruit maturation [18], seed development [8], flower development [19], and somatic embryo development [20]. In addition, the AP2/ERF transcription factors, especially the members of the DREB and ERF subfamilies, are involved in the regulation of responses to abiotic stresses, such as drought, high salt and alkali, high temperature, frost, senescence, oxidation and heavy metal ions, as well as the regulation of responses to biological stresses, such as pathogens and virus infection [1,[21][22][23]. Furthermore, the AP2/ERF transcription factors play roles in plant hormone-mediated signal transduction pathways, including ethylene (ETH), abscisic acid (ABA), jasmonic acid (JA), salicylic acid (SA) and gibberellin (GA) signaling pathways [1,21,24].
Caragana intermedia Kuang et H. C. Fu. is a xerophytic deciduous shrub belonging to the genus Caragana Fabr. It is mainly distributed in the arid and semi-arid desert areas of Inner Mongolia, Ningxia, Shanxi and northern Shaanxi. Plants of the genus Caragana are cold, drought, salt-alkali and barren tolerance. C. intermedia has strong vitality and adaptability, and is ideal for soil and water preservation, or as a windbreak for sand fixation. It also can be used as pasture for sheep, camel and cattle. C. intermedia is a superior xerophytic shrub species eminently suitable for artificial afforestation in arid desert steppe [25]. However, there are currently few studies focusing on the function of the AP2/ ERF-type transcription factors in C. intermedia. In this study, genes encoding the drought stress-inducible AP2/ERF transcription factors were screened out from the data set of the drought transcriptome of C. intermedia and were cloned. Twenty-two sequences with complete open reading frames (ORFs) were obtained and their phylogenetic relationship, physicochemical properties, domains, tertiary structures, conserved motifs and expression patterns in different tissues under four types of stress treatments were analyzed. Then, the morphology and abiotic stress response of CiDREB3 was further studied in transgenic Arabidopsis. The results provided a basis for exploring the molecular function of the AP2/ERF transcription factors and the molecular mechanism of stress tolerance of C. intermedia.

Plant sampling and cultivation
The seeds of C. intermedia were collected from Helingeer County, Hohhot and Siziwang County, Ulanqab, Inner Mongolia Autonomous Region, China. The seeds of Arabidopsis thaliana wild-type used in this work are Columbia-0 ecotype (Col-0), and mutant of DREB3(dreb3) was obtained from SIGnAL (Salk Institute Genomic Analysis Laboratory) and ABRC (Arabidopsis Biological Resource Center), and the Name/Stock Number was SALK_206788C.
Cultivation of C. intermedia: Plump C. intermedia seeds free of worm holes were selected, sown in pots containing nutrient-rich soil: vermiculite (1:3, v/v), and cultivated in a greenhouse at 25 °C under long-day conditions (16 h light/8 h dark cycle).

Plant treatments
The 25-day-old uniformly growing C. intermedia seedlings were exposed to 4 different types of stress: cold, heat, salt and mannitol. The treatment procedures were as below: (1) For cold or heat treatment: C. intermedia seedlings previously grown under normal conditions were transferred to 4 °C (for cold treatment) or 42 °C (for heat treatment) in the RUMED incubators and cultivated under 16 h light/8 h dark condition. (2) For salt or mannitol treatment: 3 to 4 days after the last watering, C. intermedia seedlings previously grown under normal conditions were irrigated with 200 mmol L −1 NaCl solution (for salt treatment) or 300 mmol L −1 mannitol solution. The seedlings were exposed to the above stress treatments for 0, 1, 3 and 12 h respectively. Samples were collected from the aerial parts of the seedlings, rapidly frozen in liquid nitrogen, and stored in a − 80 °C freezer for subsequent experiments. At each time point, samples were collected from 3 seedlings and mixed. For each treatment, three independent biological replicates were performed.
Examination of tissue-specific expression: Root, stem and leaf tissues were collected from 25-day-old C. intermedia seedlings. Each type of tissue was harvested from five seedlings, rapidly frozen in liquid nitrogen, and stored in a − 80 °C freezer. Three independent biological replicates were performed.

Verification of the coding region of the AP2/ ERF-type genes in C. intermedia
AP2/ERF gene sequences whose expression levels were significantly up-regulated or down-regulated (reads per kilobase per million mapped reads (RPKM) ≥2) were selected from the C. intermedia drought transcriptome database (NCBI Accession number: SRP121096) and subjected to sequence alignment analysis using the NCBI Blastn program. Gene sequences containing complete ORFs were verified using specifically designed primers (see supplementary  Table S1 for primer sequences). The remaining sequences that lacked a complete ORF were amplified using the rapid amplification of cDNA ends (RACE) technique to obtained the flanking sequences. The flanking sequences were then assembled with the known intermediate fragments using the Vector NTI 10.0 software (Invitrogen), generating gene sequences with complete ORFs. The ORF sequences were verified using specifically designed primers (the primer sequences are summarized in supplementary Table S1). The primers were designed using Primer Premier 5.0 software. Polymerase chain reaction (PCR) amplification was performed using high-fidelity PrimeSTAR HS DNA Polymerase. The complementary DNAs (cDNAs) or the genomic DNAs (gDNAs) of C. intermedia served as PCR templates. The melting temperature (Tm) and extension time varied for different genes in the AP2/ERF family were summarized in Table S2.

Bioinformatic analysis
The sequences in the drought transcriptome database were subjected to alignment analysis using the NCBI Blastn online alignment tool. The theoretical pI (isoelectric point) and MM (molecular mass) were calculated using the Compute pI/MW online tool (ExPASy, Switzerland). The subcellular localization of proteins was predicted using the WoLF PSORT online program [26]. Protein sequences from different species were aligned using the DNAMAN software.
The conserved protein motifs were analyzed using the MEME online tool [27]. The settings of the parameters were as follows: minimum motif width = 6; maximum motif width = 50, maximum number of motifs = 15. Protein domains were analyzed using the NCBI's Conserved Domain Database (CDD) [28], and the distribution map of the conserved protein domains was plotted using DOG 2.0 software [29]. The tertiary structures of various subfamilies of C. intermedia AP2/ERF-type proteins were subjected to homology modeling analysis using SWISS-MODEL, which allowed the establishment of tertiary structure models of AP2/ERF-type proteins in C. intermedia.
The MEGA 6.0 software was used to analyze the homology of the AP2/ERF-type proteins in C. intermedia with those in M. truncatula and A. thaliana [30]. The sequences of Medicago-derived AP2/ERF-type proteins were obtained from a study published by Shu et al. [31], whereas the sequences of the Arabidopsis-derived AP2/ERF-type proteins were acquired from a study published by Nakano et al. [5]. The obtained sequences were compared using ClusterW, and a portion of the sequences that was significantly different in amino acid composition was removed. The algorithm used in the analysis was neighbor-joining. The bootstrap value was set to 1000. The mode used in the analysis was the Poisson model, and the gaps were set to partial deletion.

Real-time fluorescence-based quantitative PCR (qRT-PCR)
The diluted templates were removed from the − 80 °C freezer, placed on ice, and thawed naturally. The expression level of various genes were analyzed at the transcriptional level by real-time fluorescence-based qPCR. The PCR instrument used in the present study was a LightCycler 480 (Roche, Basel, Switzerland), and the fluorescent dye used was SYBR Green Premix II. The CiEF1α of C. intermedia was selected as the internal reference gene (Table S1). Three technical replicates were performed for each sample, while three biological replicates were performed for each gene. The experimental results were analyzed using the 2 −ΔΔCt method.

Analysis of the expression patterns
The genes encoding the AP2/ERF family of transcription factors were screened out from the drought transcriptome database of C. intermedia. The expression pattern of these genes in various tissues (root, stem, leaves and whole plant) or under different stress conditions (cold, hot, salt or mannitol treatment) were analyzed. Briefly, RNA was extracted using the RNAsimple Total RNA Kit (DP419) (Tiangen Biotech (Beijing) Co, Ltd, China) in accordance with the manufacturer's instructions. The first-strand cDNA synthesis was performed using TransScript gDNA Removal and cDNA Synthesis SuperMix (AT311) (Beijing TransGen Biotech Co., Ltd., China) in accordance with the manufacturer's instructions. qRT-PCR was performed using the C. intermedia cDNAs that had been diluted 16-fold as template. The sequence of PCR primers are shown in Table S1. The PCR system and procedure were set up accoding to the manufacturer's instructions. A heat map was constructed using the HemI 1.0 software [32] based on the results of qRT-PCR.

CiDREB3 transgenic Arabidopsis
To generate the recombinant CiDREB3 overexpression vector, the full-length CDSs of CiDREB3 was amplified using the wild type C. intermedia cDNA and cloned into the expression vector pCanG-HA using the restriction enzymes SalI/SacI, under the control of the CaMV35S promoter. The recombinant vectors were expressed in wild-type A. thaliana using the floral dipping method, mediated by Agrobacterium tumefaciens (strain GV3101). The empty vector was used as the control.

Root length measurement
For root length measurement under normal condition, wildtype, the mutant and the transgenic Arabidopsis seedlings germinated for 48 h were transferred to 1/2 MS, and hold on upright for 6 days under normal condition. At least 40 seedlings per genotype were used for root length measurement. This experiment was repeated three times.

Fresh weight measurement
The aerial part of the transgenic Arabidopsis, wild-type and the mutant seedlings grown for 3 weeks under normal condition was chose to measure the fresh weight. it was calculated using four plants per parallel group and three parallel group per genotype for each experiment. This experiment was repeated three times.

Rosette leaf diameter and leaf number measurement
The 7th to 9th rosette leaf diameter for three-week-old transgenic Arabidopsis, wild-type and the mutant was detected, and using 30 plants per genotype for each experiment. This experiment was repeated three times.

Drought tolerance test
To evaluate the potential drought tolerance, surface-sterilized seeds from wild-type and the CiDREB3 transgenic Arabidopsis were planted on the plate. The plates were incubated at 4 °C for 3 days in dark before being placed at 22 °C under 16-h light/8-h dark conditions. Seedlings grown for 3 weeks under a normal watering regime were subjected to drought stress by withholding watering for 14 days and were then re-watered for 2 days. Survival rate and chlorophyll content were calculated and assayed using 28 plants per genotype for each experiment. This experiment was repeated three times.

Transcriptome-based identification of the AP2/ERF family genes in C. intermedia
After screening the drought transcriptome database and performing the Blastn sequence alignment analysis, a total of 37 gene sequences with intact AP2 domains were identified, and these included 20 gene sequences containing the complete ORFs. Sequences lacking the complete ORF were amplified using the RACE technique to obtain the flanking sequences. After assembly of the sequences, two more gene sequences with complete ORFs were obtained, namely, CiERF008 (comp92811_c1) and CiERF004 (comp123668_c0). Eventually, 22 sequences with complete ORFs were obtained (Table S3). The 22 gene sequences were compared with the sequences in the TAIR database. The results showed that 3 of the 22 genes belonged to the DREB A1 group, 2 belonged to the DREB A2 group, 2 belonged to the DREB A4 group, 4 belonged to the DREB A5 group, 1 belonged to the DREB A6 group, while 2 belonged to the ERF B1 group, 7 belonged to the ERF B3 group, and 1 belonged to the RAV group.

Phylogenetic evolution analysis of the AP2/ERF transcription factors in C. intermedia
To investigate the phylogenetic relationship among the C. intermedia AP2/ERF transcription factors, MEGA 6.0 software was used to construct a phylogenetic tree from all 22 AP2/ERF members. The results are shown in Fig.  S1. Different background colors represent the different subgroups. The DREB subgroups were clustered in one branch, and those belonging to the ERF B3 subgroup were clustered in the second branch, whereas the 2 members of the ERF B1 subgroup (CiERF004 and CiERF009) and CiRAV1 were clustered in the third branch.
To further investigate the evolutionary relationship between the AP2/ERF transcription factors in C. intermedia and their homologs in other plants, phylogenetic evolution analysis was performed based on the 22 AP2/ ERF transcription factor sequences from C. intermedia (22), M. truncatula (101), and A. thaliana (128). Proteins from M. truncatula and A. thaliana consisted of DREBtype, ERF-type and RAV-type. None of these sequences belong to the AP2 or Solist subfamilies. As shown in Fig. 1, the sequences were divided into 13 groups, including the DREB A1-A6, ERF B1-B6 and RAV groups. Members of the DREB A1 group were clustered with M. truncatula MtERF021, MtERF023 and MtERF024 (Fig.  S2). CiDREB2C and CiDREB2D were clustered with MtERF048 and AtERF045 in one branch (Fig. S3). The members of the A4 group of C. intermedia were clustered with MtERF025 and MtERF029 (Fig. S4). Some scholars merged all of the above M. truncatula sequences into one group, which was named group III. CiERF008, CiRAP2-1, CiERF020 and CiERF017, which are members of the DREB A5 group, were clustered with MtERF018, MtERF014, MtERF017 and MtERF012, respectively (Fig.  S4). CiERF061 had a relatively close genetic relationship with AtERF061 (Fig. S5). CiERF004 and CiERF009 were clustered together with MtERF073 (Fig. S6). All members of the ERF B3 group except CiERF109 were derived from the same ancestor (Fig. S7). In addition, the RAV subfamily member CiRAV1 was closely related to MtERF120 and AtRAV2 (Fig. S8).

Analysis of the physicochemical characteristics of the C. intermedia AP2/ERF transcription factor family
The physicochemical properties of the amino acid sequences that were encoded by the 22 genes with complete ORFs were predicted using the online prediction tool Protparam. As shown in Table S3, the number of the amino acids encoded by the 22 genes varied greatly (ranging from 151 to 453 amino acids). The molecular mass of the proteins also varied significantly, ranging from 16.5 kD to 50.9 kD. Of the 22 transcription factors, 6 had isoelectric points greater than 7. Among these 6 transcription factors, CiERF009 had the highest isoelectric point (9.51). The other 16 transcription factors had isoelectric points lower than 7, among which, CiDREB2C had the lowest isoelectric point (4.84). Except for CiDREB3 and CiERF020, the other 20 transcription factors were all predicted to locate in the nucleus.

Domains distribution in the AP2/ERF family of transcription factors from C. intermedia
Through protein sequence alignment and analysis, the length and position of the conserved domains in the 22 C. intermedia AP2/ERF-type transcription factors were determined. The shortest domain was 57 amino acids in length (CiRAV1), while the longest domain was 70 amino acids (CiDREB2D), which conformed to the length of the typical AP2 domain (Table S4). In addition to a conserved AP2 domain, the CiRAV protein contained a B3 domain consisting of 107 amino acids (Fig. S9).

Multiple sequence alignment of the AP2/ERF family transcription factors from C. intermedia
The AP2 domains of the 22 C. intermedia AP2/ERF transcription factors were subjected to multiple sequence alignment using the DNAMAN software. As shown in Fig. S10, all AP2/ERF transcription factors contained the typical elements of the AP2 domain: YRG and RAYD. The YRG element was composed of YRGVRxRxxxGKWVCE-VREPNKK, while the RAYD element was composed of RIWLGTFxxxxMAAxAxDVAAxAxRGxxACLNFxxx-AxxLxxx. The two elements bind to various cis-acting elements, thereby playing important roles in a variety of signaling pathways. In addition, the YRG element contains β1 sheet (VRxR) and β2 sheet (KWVCEVRE) structure, while the RAYD element contains a β3 sheet (TRIWLGTF) and an amphipathic α-helix (TAEMAARAHDVAALALRG). The above findings indicate that all 22 sequences are typical AP2/ERF transcription factors.

The distribution of the conserved motifs in the AP2/ ERF family transcription factors from C. intermedia
The MEME online tool was used to predict the conserved motifs in the of 22 C. intermedia AP2/ERF transcription factors. The results are shown in Fig. S11. All sequences contain the highly conserved motif 1, which contains the classical element of the AP2 domain-RAYD. In addition, 21 of the 22 sequences contains the highly conserved motif 2, which contains the YRG element. The RAYD and YRG elements are the key components of the AP2 domain. Members of the DREB A1 group contains motif 3. CiDREB1C and CiDREB1F also contain motif 14. This motif is short in length and contains the conserved sequence VQQRD(H) M(Q). However, the action mechanism of motif 14 remains unclear. Members of the DREB A2 group contain motif 5, motif 12 and motif 13, among which motif 5 and motif 12 are highly conserved. Two members of the ERF B3 group, CiERF1B and CiERF109, contain motif 15. It is short in length and highly conserved.

Prediction of the tertiary structures of the AP2/ERF family of transcription factors from C. intermedia
Protein tertiary structure directly determines protein function. To intuitively understand the tertiary structures of the 22 C. intermedia AP2/ERF transcription factors, in this study, one transcription factor was selected from each subgroup and subjected to homology-based protein tertiary structure modeling. The results are shown in Fig. S12. Since there was little difference in the AP2 domain among the various subgroups, members of the A1, A4, A5, A6, B1, and B3 groups were modeled using the same template (5wx9.1.A). For the members of the A2 and RAV1 groups, the reference template was 3gcc.1.A. It was predicted that each subgroup contained an amphipathic α-helix, as well as β1, β2 and β3 sheets. These secondary structures are important components of the AP2 domain, among which the α-helix is essential for the stability of plant cell membranes. The 3 β-sheets are arranged in an anti-parallel fashion and located in front of the α-helix. The β-sheet region is responsible for DNA binding. For the DREB and the ERF subfamilies, the 14th and 19th amino acids in the second β-sheet determined the binding specificity of the transcription factors to various cisacting elements [4].

Tissue-specific expression of the genes encoding AP2/ERF transcription factors in C. intermedia
The tissue-specific expression of the AP2/ERF-type transcription factors in C. intermedia were examined. The results are shown in Fig. S13. Except for CiDREB3, CiT-INY2 and CiERF110, the other genes were expressed at 1 3 relatively low level in stem. The expression level of these genes was all within two times of that of the reference gene. The CiDREB3, CiTINY and CiTINY2, which belonged to DREB A4 group, showed relatively high expression level in stem tissue and suggested that these group members might play important roles in stem development. Most of the AP2/ ERF-type genes were highly expressed in root, especially for CiRAP2.11, CiERF013-1 and CiERF112, and the expression level of CiDREB3 was almost three folds in root. In leaves, 6 genes were expressed at relatively high level and showed a fold change greater than 8, while DREB A4 group, including CiDREB3, had almost no expression. In addition, CiRAP2.11 was highly expressed in both root and leaves. Tissue-specific expression indicates that these genes might play roles in the development of these tissues. Except for CiDREB2C, members of the DREB A1, DREB A2, DREB A4 and ERF B4 groups expressed at low level in the leaves. Low level of expression of the CiRAP2. 12 and CiERF1A were observed in all tissues examined. The above results demonstrated that the genes expression of the AP2/ERF family in C. intermedia exhibited tissue specificity.

The expression level of the AP2/ERF-type transcription factors encoding genes in C. intermedia under four types of stress treatments
qRT-PCR was employed to examine the expression level of 37 genes encoding the AP2/ERF-type transcription factors in C. intermedia under cold, heat, salt and mannitol stress treatments. The results are shown in Fig. S14. Overall, the AP2/ERF gene family in C. intermedia was inducible by abiotic stresses. Exposure to cold treatment for 12 h significantly increased the expression levels of most AP2/ ERF-type genes. Among these genes, 5 showed a more than 20-fold change in expression level. CiERF017-1 and CiDREB2C had the highest expression level (fold increase of 68.08 and 57.14, respectively). Eighteen genes responded rapidly to heat treatment and showed a more than 2-fold change in expression level after exposure to heat treatment for 1 h. Among the 18 genes, CiDREB1C exhibited the most significant response (fold-change of 228). After 12 h of heat treatment, the expression level of 7 genes continued to rise. Among these 7 genes, CiDREB1C showed the highest fold increase (more than 800 folds). In addition, the expression level of the AP2/ERF family genes showed an increasing trend after NaCl and mannitol treatment, indicating that these genes might be important in abiotic stress responses.

Heterologous expression of CiDREB3 decreased root length and fresh weight in transgenic Arabidopsis
According to expression profiles of AP2/ERF family genes, CiDREB3, which exhibited relatively high expression in stem and root, and low expression in leaf and under stress treatments, was selected for further analysis. Three independent T3 homozygous CiDREB3 overexpression lines (OE1, OE8 and OE29 ) showing relatively high expression level were selected for further study (Fig. S15). CiDREB3 showed the highest homology to AT5G25810.1 and hence the corresponding mutant, SALK_206788C, was obtained from ABRC (more details in Materials and methods). After germinated for 48 h, the transgenic Arabidopsis seedlings were transferred to 1/2 MS and hold for 6 days, then their morphology was observed and compared (Fig. 2a). The OE lines showed significantly shortening root length compared with the wild-type and the dreb3 mutant under normal growth condition. No significant differences in root length were found between the dreb3 mutant and wild-type under normal growth condition (Fig. 2b).
Then, we measured the fresh weight of the three-weekold plants, the OE8 and OE29 lines showed significantly lower fresh weight than wild-type, while OE1 line showed a similar fresh weight as the wild-type. The reason may be due to the expression level of CiDREB3 in OE8 and OE29 lines were higher than that in OE1 line, which has a dose effect on transgenic Arabidopsis. By contrast, the dreb3 mutant exhibited a significantly higher fresh weight than wildtype (Fig. 2c). These results suggested that overexpression of CiDREB3 could cause a dwarf phenotype in transgenic Arabidopsis.

Heterologous expression of CiDREB3 decreased rosette diameter and number of leaves in transgenic Arabidopsis
We observed the morphological differences of three-weekold transgenic Arabidopsis under normal growth condition, and the growth of OE8 line was significantly delayed and the leaves were smaller and curled up compared with wildtype. While the dreb3 mutant grew faster and larger than the wild-type (Fig. 3a). The rosette diameter of OE8 was also significantly lower than that of the wild-type and the dreb3 mutant. The dreb3 mutant showed significantly longer rosette diameter than wild-type (Fig. 3b). In addition, the number of leaves in OE8 line was obviously less than that in other lines (Fig. 3c). These results suggested that overexpression of CiDREB3 could cause a dwarf and delayed growth in transgenic Arabidopsis.

Heterologous expression of CiDREB3 improved drought tolerance in transgenic Arabidopsis
Three-week-old seedlings were exposed to drought stress by withholding watering for 14 days and were then rewatered for 2 days. The OE lines were insensitive to drought, had significantly increased survival rate and less wilting and yellowing compared with wild-type (Fig. 4a, b). In addition, OE8 and OE29 lines showed a higher chlorophyll level than the wild-type under drought stress. After rewatering for 2 days, the majority of wild-type plants never recovered, while the OE lines exhibited a significantly higher survival ratio, except for the line OE1 (Fig. 4c). These results confirmed that CiDREB3 enhanced the tolerance of Arabidopsis to drought stress, and CiDREB3 acted as a positive regulator in plant response to drought.

The differences between the AP2/ERF-type transcription factors of the same species
Different scholars have obtained different statistical results when investigating the AP2/ERF transcription factor genes from the same species. For example, Nakano et al. reported Significance level was analyzed using t test. *Significant difference(P < 0.05); **Extremely significant difference (P < 0.01) that there were a total of 157 AP2/ERF transcription factor genes in O. sativa [5], while Sharnoi et al. reported 163 AP2/ERF transcription factor genes [33]. Chen et al. reported that B. distachyon had 149 genes of the AP2/ERF transcription factor family, which could be divided into four subfamilies [34]. Cui et al. also carried out statistical analysis on the AP2/ERF transcription factor genes from B. distachyon. A total of 141 genes of the AP2/ERF transcription factor family were identified, which were also divided into four subfamilies [35]. Wu et al. reported that Phyllostachys edulis (Carrière) J. Houz. had a total of 121 AP2/ERF transcription factors [36]. Later, Huang et al. reported that there were a total of 142 AP2/ERF transcription factors in P. edulis [37]. Song et al. reported that there was a total of 291 AP2/ERF transcription factors in B. rapa [38]. In the same year, Liu et al. found 281 members of the AP2/ERF transcription factor family in B. rapa [39]. We have summarized the reasons behind such phenomenon, and the possible reasons are as follows: (1) Certain genes in the AP2/ERF family undergo alternative splicing. Some scholars believe that each splice variant is a gene, while others believe that all splice variants derived from the same gene should be counted as one gene. As a result, there was a discrepancy in the number of AP2/ERF genes [35,40]. (2) Some scholars believe that all genes containing the AP2 domain belong to the AP2/ERF transcription factor family, while other scholars do not count the genes that contain other domains besides the AP2 domain as members of the AP2/ERF transcription factor family, which results in a different number of AP2/ERF genes [38,39]. (3) When analyzing the AP2/ERF gene family in P. edulis, some scholars have included genes with an AP2/ ERF domain integrity of less than 70% in the AP2/ERF family [37]. However, other scholars did not include such genes and only counted the genes with intact AP2/ERF domains. Again, this caused a discrepancy in the number of AP2/ERF genes [36].

Analysis of the AP2/ERF transcription factor family in leguminous plants
Currently, genome-wide identification and analysis of the AP2/ERF transcription factor family has been performed in the sequenced leguminous plants. For example, Zhang et al. identified 148 members of the AP2/ERF transcription factor family in tetraploid G. max [41]. Genes encoding one AP2 domain were classified into one group, which was named the ERF family. The ERF family consists of 120 genes. Among the 120 genes, 98 genes that were capable of encoding one complete AP2 domain were called ERF family genes. The 98 ERF family genes (GmERF001 to GmERF098) were further divided into the DREB subfamily (36 genes) and the ERF subfamily (62 genes). The remaining 22 genes were unable to encode the complete AP2 domain, and the authors did not further group the 22 genes. Agarwal et al. counted the number of AP2/ERF transcription factors in 5 types of sequenced leguminous plants [42] ( Table 1). The results showed that there were 147 AP2/ERF-type transcription factors in C.

Analysis of the expression patterns of the AP2/ ERF-type transcription factors in C. intermedia
The AP2/ERF transcription factor genes display tissuespecific expression in many plants. There is a total of 291 AP2/ERF transcription factor genes in B. rapa. Among the 291 genes, 31.95% are specifically expressed in the roots, 22.9% are specifically expressed in the seeds, and 20.06% are highly expressed in the leaves. In contrast, only a small number of genes are specifically expressed in flowers, pods and buds [38]. In this study, qRT-PCR was performed to examine the expression levels of the 37 AP2/ERF transcription factors in different tissues of C. intermedia (Fig. S13). It was found that the expression levels of these genes varied greatly in the roots, stems and leaves and displayed a certain degree of tissue specificity. The CiDREB3 showed relatively high expression level in stem and root tissue, suggested that these subgroup members might play important roles in stem and root development.
In addition, many studies show that all AP2/ERF-type transcription factors are inducible by a variety of abiotic stresses. For example, among the 106 AP2/ERF transcription factors examined in Brachypodium distachyon (L.)., 69 were significantly induced by cold stress and 16 were induced by drought stress [35]. The expression levels of these transcription factors were increased by more than two folds after exposure to stress. In O. sativa, 70 genes in the AP2/ERF transcription factor family are inducible by cold, drought, flooding, osmotic, salt and hormonal stresses [33]. In this study, qRT-PCR was employed to examine the expression patterns of 37 AP2/ERF transcription factor genes under four types of stress (Fig. S14). The results showed that these genes were all inducible by cold, heat, NaCl and mannitol. After cold treatment, 17 genes showed a more than two-fold increase in expression level, while 8 genes showed a more than 10-fold increase in expression level. After heat treatment, the expressions levels of 18 genes showed a more than two-fold increase. After NaCl treatment, 14 genes showed a more than two-fold increase in expression level. After mannitol treatment, there were 12 genes with a more than two-fold increase in expression. Interestingly, CiDREB3 was induced negatively by cold and heat stress. The expression level of CiDREB3 showed a trend of decreasing first and then increasing in salt stress, while it was not induced by mannitol, and these results were accordant with that in C. intermedia drought transcriptome database. In A. thaliana, AtTINY2 was inducible by SA, ABA, mechanical damage, cold, drought and NaCl [43]. Another gene that belonged to the same group (the A4 group) as AtTINY2 and showed a high sequence similarity was AtTINY. Studies have shown that AtTINY is induced by drought, cold, ethylene and methyl jasmonate. The above results demonstrated that the AP2/ ERF transcription factors are inducible by a variety of abiotic stresses.

Functional analysis of CiDREB3 in transgenic Arabidopsis
A phylogenetic tree was constructed from the 22 C. intermedia AP2/ERF transcription factor sequences with complete ORFs, 101 M. truncatula AP2/ERF sequences and 128 A. thaliana AP2/ERF sequences (Fig. 1). The higher the similarity between genes, the closer their functions are. Among the A. thaliana genes, AtERF041 (AtTINY2) showed the highest similarity to CiDREB3. It has been reported that overexpression of AtTINY causes dwarfism and severely delayed growth and development in plants [44]. In addition, AtTINY participates in the ethylene-responsive signaling pathways and induces apparent triple response in the absence of ethylene (shortening and thickening of the hypocotyls, shortening of the roots, and exaggeration of the curvature of the apical hook) [45]. The expression levels of the stress-related genes such as COR6.6, COR15A and COR78 were upregulated in AtTINY transgenic A. thaliana [44]. Our results showed that overexpression of CiDREB3 in Arabidopsis decreased root length and fresh weight, decreased rosette leaf diameter and number of leaf (Figs. 2 and 3), and these results are consistance with the function of AtTINY2. The dwarfism and severely delayed growth phenotype in CiDREB3 overexpression lines was possibly due to the high expression level in stem and root tissue, and CiDREB3 was a negatively regulator in growth and development and negatively regulates stem and root growth in Arabidopsis thaliana.
What' s more, transgenic Arabidopsis lines overexpressing CiDREB3 improved drought tolerance during seedlings development (Fig. 4), and the result is similar to StDREB1, also belongs to DREB A4 group, in potato. Drought and salt tolerance were increased by homologous expression of StDREB1 in potato during seedlings growth, and free proline content and expression level of P5CS1 were significantly higher than those in Col-0 under salt stress. Due to CiDREB3 could stunt plant growth and have a negatively impact on plant growth, it is deduced that the CiDREB3 participated in regulating resopnse to drought stress through slowing plant growth and development.
In summary, 22 genes with complete ORFs were identified and divided into 13 groups, multiple sequence alignment 1 3 (MSA), domain distribution, conservative motif, tertiary structure prediction of these genes were systematically analyzed and verified that these genes were typical AP2/ERF transcription factors. What's more, the expression profile of AP2/ERF transcription factors in C. intermedia were tissuespecific and involved in different stress such as cold, heat, salt and mannitol. Additionally, overexpression of CiDREB3 in A. thaliana resulted in dwarfism and severely delayed growth and development, and drought tolerance phenotype in plants compared with the wild-type. This study will provide a further understanding of AP2/ERF transcription factor in plant development and stress resistance, and will provide the basis for future functional studies on AP2/ERFs.