Some 4-5-year-old Allium senescens plants were obtained from the BaShang grasslands, Zhangjiakou, Hebei Province, China (115.833307′N, 41.649034′E). The region has an average elevation of about 1,700 m, with the highest elevation of about 2,400 m. The plants were further transferred to pots in a greenhouse and subjected to drought stress by 20% PEG-6000.

Arabidopsis (ecotype Columbia, Col-0) seeds are planted directly in vermiculite, and then cultivated in a greenhouse, under conditions of 4500 lux supplementary light, 60% humidity, 16 h of light at 25°C, and 8 h dark. The development/growth period of Arabidopsis was maintained with regular watering using Hoagland nutrient solutions at two to three-day intervals.

During the vegetative growth period of Allium senescens, it germinated and grew for about 20 days, and was subjected to conditions of drought (by 20% PEG-6000) and non-drought (normal condition) for 15 days. For transcriptome analysis, the seedlings with the same growth were screened and the middle and upper parts of leaves were cut to 1.0 g, with three repeats. The RNA library construction, checking, and all sequencing were performed using Majorbio of Shanghai with HiSeq 2500. For our project without reference genomes, clean reads were obtained and spliced with Trinity software to obtain reference sequences for subsequent analysis. The primary sequencings were further analyzed to obtain clean reads. In addition, the contig was acquired to use as a control sequence. The unigene of all transcript gene was regarded as the longest transcript. The gene-related functional annotation was analyzed by using the Shanghai Majorbio Cloud.

All genes’ transcripts acquired from the transcriptome sequencing of Allium senescens were analyzed with NR (Non-redundant protein database), SwissProt (SwissProt protein database), Pfam (Protein families database), Cluster of Orthologous Groups (COG), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Gene Ontology (GO). The functional annotation information from all six databases was obtained and counted.

The AsTCPs genes were screened from the whole genome database of Allium senescens to obtain candidate sequences containing TCP structural domains, and all candidate sequences were then submitted to http://pfam.xfam.org/search (Pfam) and NCBI to validate the conserved structural domains. After eliminating the false-positive sequences without TCP special structural domains, the remaining candidate sequences were all TCP genes of Allium senescenss with the identification on ExPASy (http://www.expasy.org/tools/).

The WOLF PSORT (https://wolfpsort.hgc.jp/) software was also used to assess the subcellular sites of the AsTCP family. We identified the conserved domain of AsTCP family according to the Conserved Domain Search Service software in NCBI. The amino acid sequences of the Allium senescens TCP members were analyzed by MEME software to clarify the sequence characteristics and distribution features of the conserved motifs.

The sequence homology comparison and conserved structural domain analysis of the Allium senescens TCP genes were further performed using the DNAMAN8.0 software. We compared multiple sequences of Arabidopsis and Allium senescens AsTCP proteins after downloading the sequences of all Arabidopsis TCP proteins isolated from the Arabidopsis Information Resource (TAIR). We performed a phylogenetic analysis by the maximum likelihood method and embellished the phylogenetic tree with the iTOL tool to help in the classification of members of the AsTCP protein family.

The levels of expression of the AsTCPs genes were identified by the qRT-PCR method. The Allium senescenss were maintained by hydroponics and divided into five portions in hydroponic vials and incubated at 22°C in a greenhouse after the roots were washed and fixed in hydroponic baskets. And then the five portions of Allium senescenss were treated with water (control), 200 mmol/L NaCl, 20% PEG6000, 100 μmol/L ABA and 200 μmol/L MeJA after 5–6 weeks of growth, respectively. Samples were collected from each of the five portions at 0, 6, 12, and 48 h after treatment, and then RNA was extracted by the trizol method, and reverse transcribed into cDNA. The primers of qRT-PCR for the AsTCPs were made by Primer-BLAST software, and the actin gene from Allium senescens was used as a reference gene (Suppl. Table S1), for which 3 biological repeats were set up and repeated three times. The RNA extraction reagent is Trizol, the reverse transcription kit is 2×Fast Super EvaGreen qPCR Master Mix kit. The temperature and time were set as 96°C for 5°C min, and then by 45 cycles with 95°C for about 15 s, and 57°C for 30 s. The levels of expression of the AsTCPs gene were calculated according to the 2^−ΔΔCt method. All primers of AsTCPs and ABA-related genes are listed in Suppl. Table S1.

To generate the recombinant AsTCP17 overexpression vector, the obtained open reading frame (ORF) of AsTCP17 was cloned into the expression pCam-AsTCP17 vector driven by the cauliflower mosaic virus 35S promoter (Suppl. Fig. S1A). Then, the recombinant vectors were transferred into wild-type Arabidopsis thaliana (Ecotype Columbia) using Agrobacterium-mediated transformation method to generate the construct, AsTCP17-OE. Transformation of the empty vector plant was used as the wild-type plant. After infection by Agrobacterium containing AsTCP17, T1, T2, and T3 generation plants were successfully obtained based on the Hygromycin medium (Suppl. Fig. S1B), and the T3 generation was used to detect the drought resistance index in our study.

The 4-5-week-old Arabidopsis with drought stress for 48 h were rapidly stored at −80°C for stabilizing and measuring the enzymes related to the drought-resistance physiological index. All detection of defense enzyme kits were provided by the Nanjing Jiancheng Bioengineering Institute. China, and were used for SOD detection (No. A001-1-1), POD detection (No. A048-3-1), CAT detection (No. A007-1-1), MDA detection (No. A003-1-2), and PRO detection (No. A107-1-1). To further measure ROS in transgenic Arabidopsis, histochemical staining was used according to a previously reported method (Fryer, 2002). The leaves of the transformed Arabidopsis and wild type were respectively stained by Diaminobenzidine (DAB) buffer (pH 3.8, 1 mg/ml) at 37°C in the dark condition. The leaves with the DAB-stained were further set in 75% alcohol to delete chlorophyll, and then photographed for the stained performance.

The related drought-resistance hormones, such as ABA, SA, IAA, and JA were detected by Zoonbio Biotechnology. The leaves extract was vibrated at 4°C for 40 min (12,000 rpm), with further adding dichloromethane (10 mL) in samples, and vibrated at 13,000 rpm for 15 min at 4°C. The supernatant was filtered using Chromosep C18 columns and then fractions were eluted with 10 ml 100% (v/v) methanol. It was then dried by N2 and set in 100 μL 0.1% methane acid by a 0.22 μm filter membrane. The product of samples was analyzed with a ZORBAX SB-C18 column (Agilent Technologies) of the high-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS) method. The solvents for the mobile phase were ultrapure water and 0.1% methanoic acid, methanol and 0.1% methanoic acid, and the injected volumes were 2.5 μL. Mass spectrometer (MS) treatment conditions were the voltage at 4000 V, the pressure of nebulizer, air curtain, aux gas of respectively 60, 20, and 75 psi, and the temperature of atomizing was set at 350°C.

With analyses of the ABA biosynthetic metabolic pathway, we can infer that AtSVP (short vegetative phase), AtLOS5 (molybdenum cofactor sulfurase), AtNCED3 (nine-cis-epoxycarotenoid dioxygenase 3) and AtAO (aldehyde oxidase) are the main ABA biosynthesis-related genes. Therefore, AtSVP, AtLOS5, AtNCED3 and AtAO sequences were derived based on the Arabidopsis genome database (accession number SRP121096). Four genes were detected after overexpression of the AsTCP17 gene in Arabidopsis by qRT-PCR. The primers for the genes were designed as presented in Table 1. The method is the same as the for AsTCPs gene expression analysis given earlier.

The physiological index and relative expression level of key genes were considered as mean±SD from three independent repeats. Data analysis was carried out using Origin software (version 8.5). Data were considered to be significantly different at **p < 0.01 or *p < 0.05.

We found that a total of 175,936,701 clean paired-end reads were assembled into 97,195,096 unigenes containing an N50 length of 1299 bp and a GC base share of 37.83% based on transcriptome data (Table 1). Of these, the percentage of transcripts with sequence lengths of 0–500, 501–1000, and 1001–1500 bp were 38%, 31%, and 14%, respectively. The percentage of single gene clusters with sequence lengths of 0–500, 501–1000, and 1001–1500 bp were 46%, 29%, and 11%, respectively (Table 2). The results showed that our transcriptome data is credible.

Six databases were further used as the references for the functional annotation of the Allium senescens unigenes, including the NR database for 35,269 single gene clusters, the SwissProt database for 26,284 single gene clusters, the Pfam database for 26,617 single gene clusters, the COG database for 31,490 single gene clusters, the GO database to 28,201 single gene clusters, and the KEGG database to 16,201 single gene clusters (Fig. 1A). Of these, 10,608 unigenes were annotated in all six databases, showing that the sequencing data quality is sufficient for further subsequent experimental purpose. In addition, the annotation results of the three major databases, COG, GO and KEGG, were developed for the in-depth understanding of the unigenes results of Allium senescens (Fig. 1B).

We further obtained unigenes based on the KEGG database as a basis for functional classification. A total of 16,201 unigenes were categorized into six pathways: Cellular Processes, Genetic Function Processing, Metabolism Function, Human Diseases and Organismal Systems, and Environmental Information Processing, respectively, which were further refined into 27 metabolic pathways. The highest number of differentially expressed genes was 1467 for carbohydrate metabolism, followed by 1279 for amino acid metabolism (Fig. 1C).

Transcription factors are involved in the complex process of regulating transcription initiation in plants. A total of 8 transcription factor families were selected in the Allium senescens. The families were as follows: the bHLH (77), bZIP (34), NAC (55), WRKY (48), GRAS (34), MADS (34), TCP (14) and GRF (12). All the transcription factor families possibly played a critical role in the drought-stress response. Of all the transcription factor families, the gene expression difference indicated that the TCP family was almost all upregulated with drought induction, with the percentage of upregulated genes being up to 92.9% (Table 3). Therefore, the TCP transcription factor family has become the focus of our research.

The gene members of the TCP family of Allium senescens were further identified, and 14 TCP family genes were obtained and named according to the TCP gene family reported in the model plant Arabidopsis thaliana. Based on the results of the functional, structural domain and sequence Blast comparison analysis, seven belonged to the class I PCF subfamily, two to the class II CYC/TBI subfamily, and five to the class II CIN subfamily (Table 4).

A subcellular localization prediction analysis revealed that the AsTCPs protein is predicted to be located in the nucleus (Fig. 2A). Ten conserved motifs of the AsTCPs proteins were identified, and most of them contained motif 5, which is the most important motif of TCP (Fig. 2B). To further analyze the homologous evolution of the AsTCP family genes with other species and construct a phylogenetic tree, we downloaded the TCP transcription factor family protein sequences of Arabidopsis, Zea mays and Allium senescens. The evolutionary tree diagram reveals that the 14 TCP transcription factors of the Allium senescens are distributed in three subfamilies, CYC/TB1, CIN and PCF. There are seven in the PCF subfamily of Class I, two in the CYC/TB1 subfamily of Class II, and five in the CIN subfamily of Class II. The AsTCP showed significant clustering with the AtTCP and ZmTCP family genes, from which it can be inferred that the Allium senescens TCP genes may have similar biological characteristics as in Arabidopsis and Zea mays (Fig. 2C).

To further identify the function of AsTCPs, we tested the level of expression under drought stress. We found that the level of expression of AsTCP12-1/20-2/23 was not different before or after drought treatment of the plants based on the transcriptome data analysis. In contrast, the other TCP genes showed obvious changes in expression after drought stress, and the transcript level of AsTCP2/8/13/17 genes was significantly increased than that of WT plants (Table 5). The results showed that AsTCP2/8/13/17 genes may be the essential genes that confer drought resistance in Allium senescens.

Based on the results of the RNA-seq analysis, four essential genes, AsTCP2/8/13/17 showed upregulated expression, and were further analyzed by Real-time PCR analysis under PEG, NaCl, ABA and MeJA treatments. The results for the four genes indicated that AsTCPs were upregulated under four treatment conditions, among which the AsTCP17 gene was most significantly and rapidly upregulated after 6 h following treatment. The expression of the AsTCP2/8/13 genes frequently fluctuated without any apparent pattern (Fig. 3), which indicated that the AsTCP17 is likely to be the key gene in Allium senescens plants resistance to drought stress.

For investigating whether AsTCP17 is involved in the drought tolerance process in Allium senescenss, we constructed an AsTCP17 gene overexpression vector and successfully obtained the T0 generation plants after infesting Arabidopsis with Agrobacterium tumefaciens. We determined the expression of AsTCP17 in the OE1, OE6, and OE8 lines and observed that the transcript level of the AsTCP17 gene was obviously enhanced in the transgenic Arabidopsis (OE1, OE6, and OE8), indicating that the trans-AsTCP17 Arabidopsis plants were successfully obtained (Fig. 4A).

We further identified the drought tolerance of OEs and wild-type plants. The water deficit treatment of OEs and WT plants revealed that rosette leaf wilt was more serious in the wild-type Arabidopsis than in transgenic plants, from which it can be deduced that the transgenic plants showed better adaptation under water deficit conditions (Fig. 4B). The results confirmed that transgenic plants can significantly increase the drought tolerance of Arabidopsis.

In addition, the CAT content of the transgenic lines all increased significantly 0.5-fold, up to the highest by application of PEG for 12 h. (Fig. 4C). H₂O₂ content further revealed that the H₂O₂ content of the transformed lines was much lower than those of WT (Fig. 4D). When leaves of the transgenic lines were observed by DAB staining, the reddish-brown color of the leaves deepened as the treatment time increased, while the leaves of the transformed lines were remarkably lighter (Fig. 4E). Our results indicated that the overexpression of AsTCP17 in Arabidopsis increased the CAT content and improved the drought tolerance of Arabidopsis, while it reduced the content of reactive oxygen species.

The plant scavenging system helps to maintain low levels of reactive oxygen species, which can inhibit lipid peroxidation of membranes in adverse conditions and allows plants to withstand drought stress. SOD, POD, and MDA are the necessary antioxidant enzymes in the ROS system. As presented in Fig. 5, the differences in MDA, SOD, and POD activities, and chlorophyll contents between WT and transgenic Arabidopsis were not significant under normal treatment conditions, whereas significant differences in each index were observed after 48 h water deficit treatment. Among them, the peak values of POD and SOD in transgenic lines were increased by 0.8 and 0.4 times (Figs. 5A and B), while MDA was reduced by approximately 0.5-fold in the OEs (Fig. 5C), and chlorophyll content was much higher in the OEs lines than in the wild-type under drought stress (Fig. 5D). Taken together, our findings indicated that the drought-related defense enzymes were more active in the transgenic Arabidopsis which confirmed that AsTCP17 is a positive regulator of drought resistance.

To further analyze the drought-resist mechanism of the transformed Arabidopsis, we compared the drought-related hormones in leaves of the transformed and non-transformed plants under PEG treatment. The results showed that IAA, SA, JA, and ABA in the overexpression lines were a little higher than those of Col-0. However, no significant differences were found under normal conditions (Fig. 6). ABA content in each transgenic line increased by 57.56%, 140.57% and 106.60%, and the SA content increased by 35.56%, 36.29%, and 8.13% compared to Col-0 under drought treatment (Fig. 6). Therefore, overexpression of the AsTCP17 gene affected the synthesis and accumulation of drought-related hormones in plants, especially the synthesis of ABA.

AtSVP, AtLOS5, AtNCED3 and AtAO are important genes for the ABA synthesis. The expression level of key ABA synthesis genes confirmed that AtSVP and AtNCED3 expression was obviously increased in the overexpression lines, with the OE6 line showing a 20.6-times and 7.0-times increase, respectively, while AtLOS5 and AtAO showed a non-significant increase (Fig. 7). Our findings demonstrated that overexpression of the AsTCP17 gene led to increased levels of expression of AtSVP and AtNCED3, especially AtSVP, which further proved that AsTCP17 is involved in ABA synthesis pathway by regulating the transcription level of AtSVP and AtNCED3.