The plant materials in this study included the arbor willow ‘Suliu 795’ (Salix × jiangsuensis ‘J795’) and two genotypes of Arabidopsis, Col-0 (WT) and SmTTF30/WT (OE). All Arabidopsis lines were germinated on half-strength Murashige and Skoog media supplied with 2% sucrose (1/2 MS) for 10-14 days, and then the seedlings were planted on mixed soil (50% peat soil and 50% vermiculite). One-year-old stem cuttings (length, 8-10 cm; coarse, 2-3 cm) of ‘suliu 795’ were immersed in water, which was collected from the botany garden of Nantong University (Nantong, China). All materials were grown in an artificial climate chamber (16 h day/8 h night, 22°C day/18°C night). Five-week-old Arabidopsis plants were subjected to flooding treatment with the water surface 3 cm above the plants. After treatment for 0, 4 h, 12 h, 24 h, 48 h, 3 days, 5 days, recovery 4 h and 24 h, and leaves were collected. Two-week-old willows were incubated with 20% PEG6000, 200 mM mannitol solutions and low temperature (4°C). After treatment for 0, 4 h, 12 h, 24 h, and 48 h, adventitious roots were collected. This experiment was independently repeated three times. All samples were subsequently frozen and stored at -80°C for RNA extraction. Flooding stress transcriptome data are under 4 h, 12 h, 24 h and 48 h of flooding treatment, and salt stress transcriptome data are after 0 h, 4 h, 8 h and 12 h of 200 mM NaCl treatment.

Sequences of Arabidopsis thaliana, Oryza sativa, Populus trichocarpa, and Salix purpurea were downloaded from JGI (https://www.phytozome.net/). S. matsudana sequences were obtained from our sequencing, and assembly results were obtained by Roche/454 and Illumina/HiSeq-2000 sequencing technologies(Zhang et al., 2020). Arabidopsis trihelix family member sequences were downloaded from the Plant Transcription Factor Database PlantTFDB (http://planttfdb.gao-lab.org/) (Zhang et al., 2011). The P. trichocarpa Trihelix gene sequence was obtained from the article(Wang et al., 2016). Based on two BLASTp methods, the 34 triheix protein sequences of Arabidopsis retrieved from the Plant Transcription Factor Database (PlantTFDB) were used as queries to obtain the possible Trihelix proteins in the S. matsudana genome by BLASTp search with a cutoff E-value of 1.0×10^-10. All the trihelix candidate proteins were checked using the Conserved Domain Database (CDD) of NCBI and SMART database (Lu et al., 2020). ExPASy was used to determine the basic physical and chemical characteristics of 78 trihelix proteins, such as the molecular weight (MW), isoelectric point (pI) and amino acid sequence length (Wilkins et al., 1999). Cell-PLoc was used to predict subcellular localization (Chou and Shen, 2008). DEGs were obtained from the results of transcriptome data. DEGs were identified using the DEGseq R package with filtering criteria of false discovery rate (FDR)<0.01 and fold-change ≥2. Then we entered the gene names of TTFs to search among all DEGs.

Arabidopsis and Populus Trihelix genes were extracted from their genomes according to previous reports. The amino acid sequences of the Trihelix gene from three plants were used to construct a neighbor-joining (NJ) phylogenetic tree by MEGA X software with default parameters (Kumar et al., 2018). The bootstrap value was set to 1,000. The tree was embellished by web-based software ChiPlot(https://www.chiplot.online/tvbot.html).

The intron-exon structures of SmTTFs were obtained based on the gene annotation Gff3 files we assembled. To characterize the structure of SmTTF proteins, the conserved motifs were analyzed using the online tool MEME with the following parameters: motif 10 and width between 6-50 amino acid residues (Bailey and Elkan, 1994). Cis-elements on the promoter (-2,000 bp upstream ATG) were predicted on the PlantCARE website. Graphics of structures were drawn using TBtools software (Chen et al., 2020).

Location and chromosome length information of every SmTTF gene was obtained from the gene annotation Gff3 file. TBools software was used to draw the chromosomal distribution image of SmTTF genes. The detection and identification of gene duplication events in SmTTF genes were performed using multiple collinear scanning toolkits (MCScanX) with default parameters. MCScanX was also used with default settings to analyze the syntenic relationship of the trihelix family genes among the S. matsudana, A. thaliana, O. sativa, P. trichocarpa, and S. purpurea genomes. After obtaining duplicated gene pairs, the synonymous substitution rate (K_s) and nonsynonymous substitution rate (K_a) of gene pairs were calculated using the ‘Simple Ka/Ks Calculator’ of TBtools.

RNA from the plant tissue of all stress-treated materials was extracted using the RNAprep Pure Plant Plus Kit (TIANGEN, DP441, Beijing, China). The RNA was used to synthesize first-strand cDNA with the PrimeScript RT Reagent Kit (Takara, RR037Q, Beijing, China). The primers used for qRT-PCR are listed in Supplementary Table 2 . The program was performed using ABI7500 according to the manufacturer’s instructions. PCR mixes were made following the protocols of the UltraSYBR Mixture(CWBIO, CW2601S, Taizhou, China). The expression levels were calculated using 2^-ΔΔCt and compared to the internal control and CK sample (Livak and Schmittgen, 2001).

Leaf RNA of S. matsudana and first-strand cDNA were obtained using the methods described above. SmTTF30 was amplified by 2×Es Taq MasterMix (CWBIO, CW0690L, Taizhou, China) using the primers listed in Supplementary Table 2 . The PCR products were cloned into a PMD18-T vector (Takara, 6011, Beijing, China) according to the manufacturer’s protocols. pWM101-35S:SmTTF30 was constructed using an infusion strategy (ClonExpress II One Step Cloning Kit, C112-01, Vazyme, Nanjing, China). The pWM101-35S:SmTTF30 construct was first transformed into Arabidopsis WT (Col-0) through the agrobacterium-mediated (GV3101) floral dip method reported previously (Clough and Bent, 1998). The positive SmTTF30/WT(OE) T1 plants were screened using half strength (1/2) MS with 20 mg/L hygromycin and genomic PCR with SmTTF30-specific primers ( Supplementary Table 2 ).

Leaves of WT and three independent SmTTF30 lines were immersed in 0.25% Evans Blue solution, rinsed with water and placed in anhydrous ethanol for decolorization (Jacyn Baker and Mock, 1994). The leaves were laid flat and scanned with a scanner. All leaves were homogenized using liquid nitrogen. The activity of peroxidase (POD, A084-3-1, NJJCBIO, Nanjing, China) and the content of malondialdehyde (MDA, A003-3-1, NJJCBIO, Nanjing, China) were detected using the respective kits according to the manufacturer’s instructions. All treatments were applied to at least three replicates.

Origin 2018 (Originlab, MA, USA) was used to construct graphs of the data. SPSS version 25 statistical software (SPSS, NY, USA) was used to detect statistically significant differences.

In this study, 78 Trihelix family members were identified in the S. matsudana genome with reference to known Trihelix protein sequences in Arabidopsis, and every gene was confirmed to contain the Trihelix structural domain. The 78 Trihelix genes were named SmTTF01-SmTTF78 according to their position on the chromosome. Among the 78 members identified, the SmTTF gene encoded proteins with lengths ranging from 206 aa to 769 aa, molecular weights ranging from 24.48 kDa to 83.64 kDa, and isoelectric points ranging from 4.55 to 10.1, with the smallest being SmTTF34 and the largest being SmTTF70. The subcellular localization prediction results showed that all proteins could localize to the nucleus, and five proteins localized to chloroplasts, SmTTF05, SmTTF39, SmTTF40, SmTTF55 and SmTTF78, which may be related to photosynthesis. SmTTF40 and SmTTF68 were also localized to the peroxisome, and we speculated that those proteins are related to photosynthesis, metabolism and developmental processes. Supplementary Table 1 listed basic information about all Trihelix members in S. matsudana.

To investigate the phylogenetic relationships among the Trihelix proteins in Arabidopsis, poplar and salix, a neighbor-joining phylogenetic tree was constructed using 168 Tihelix sequences, including 34, 56 and 78 sequences from Arabidopsis, poplar and salix, respectively ( Figure 1 ). Among these species, GT-2, with 28 SmTTF family members, was the largest subfamily, whereas SH4 was the smallest clade, with 13 SmTTF family members. The SIP1 and GT-1 clusters contained 23 and 14 SmTTF members, respectively. There was no GTγ family member in S. matsudana; therefore, we hypothesized that S. matsudana is conservative in evolution, which also indicated evolutionary differences between S. matsudana and other species. The phylogenetic tree showed that the evolutionary relationship between GTγ, GT1 and GT2 was relatively close and intermediate, indicating that GTγ evolved from GT1 and GT2, and the functions of the three subfamily genes were similar.

The conserved motifs of the Trihelix gene of S. matsudana were analyzed by the MEME Suite tool, and a total of 10 conserved motifs were identified. The visualization results were shown in Figures 2A, B , where the conserved structural domains of the same subfamilies were similarly distributed. All genes contained Motif1, which was annotated as a Myb/SANT-like DNA-binding domain because of its overlap and similarity to the helix-turned-angle helix structure of the MYB transcription factor. GT-1 and SIP1 subfamily genes all contained Motif1 and Motif2, located at the 5’ end. The GT-1 subfamily of genes also included motifs 3, 5, 6, and 10. In the SIP1 subfamily, motifs 1, 2 and 10 were located at the 5’ end, and motif 8 is located at the 3’ end. The SH4 subfamily genes also contained motifs 3, 6, 7, and 9, with motif 7 located at the 5’ end and motif 9 at an intermediate position. The GT-2 subfamily genes also included motifs 2, 3, 4, 5, 6, 7, and 9, with motifs 2 and 4 concentrated at the 5’ end, motifs 3 and 7 concentrated at the 3’ end, and motifs 5, 6, and 9 concentrated at the intermediate position.

The structure of these Trihelix genes was further analyzed to characterize the S. matsudana Trihelix gene family ( Figure 2C ). The number of exons of the Trihelix gene in S. matsudana was distributed between 1 and 6, with the highest average number of exons in the GT-1 and GT-2 subfamilies and the lowest in the SIP1 subfamily. We found that the SIP1 gene had only 1-2 exons, 5 of the 13 genes in the SH4 subfamily had 3 exons, and 3 genes had more than 3 exons. Nine and 19 genes in the GT-1 and GT-2 subfamilies, respectively, contained two exons, five of the GT-1 genes had 5-6 exons, seven of the GT-2 genes contained three exons, and one had four more exons. Members of the same subfamily were observed to share a similar genetic structure.

Chromosomal localization of the identified Trihelix genes was performed ( Figure 3 ), and 68 of the 78 SmTTF genes were distributed on 29 of the 38 chromosomes of S. matsudana. Chromosomes 11 and 21 contained the highest number of genes with six genes, and chromosomes 2, 8, 9, 12, 13, 14, 15, 19, 23, 24, 25, 30, 35, 36, 37, and 38 all had only one gene distribution. Trihelix genes from different subfamilies were mostly randomly distributed; for example, 11 of the 13 SH4 subfamily genes were located on Chr10, Chr11, Chr12, Chr18, Chr21, Chr30, Chr33, Chr35, and Chr36. The analysis identified 111 pairs of duplicated genes, including three pairs of tandem repeats, SmTTF07 and SmTTF10, SmTTF48 and SmTTF50, and SmTTF62 and SmTTF63 tandem repeats. Intragenomic covariance analysis identified the remaining 108 pairs as segmental repeats ( Figure 4 ). The Ka/Ks of all duplicated gene pairs were less than 1 ( Supplementary Table 3 ), indicating that these genes may have experienced strong purifying selection pressure during evolution.

To gain a closer understanding of the replication mechanism of TriHelix gene family members in S. matsudana, we constructed covariance maps of S. matsudana and A. thaliana, O. sativa, P. trichocarpa and S. purpurea ( Figure 5 ). The results of covariance analysis showed that 37 SmTTF genes in S. matsudana showed covariance with 16 genes in Arabidopsis ( Figure 5A ). Twenty SmTTF genes showed covariance with 9 genes in rice ( Figure 5B ). Sixty-four SmTTF genes showed collinearity with 40 genes in poplar and 41 genes in S. purpurea with 65 SmTTF genes ( Figures 5C, D ). The homologous gene pairs of S. matsudana with A. thaliana, O. sativa, P. trichocarpa and S. purpurea were 45, 35, 142 and 147, respectively, indicating a high species affinity between S. matsudana and P. trichocarpa and S. purpurea.

To more closely analyze the functions of the Trihelix gene in S. matsudana that have diverged during evolution, especially the plant response to abiotic stresses. We submitted a 2000 bp sequence upstream of each SmTTF translation start site to the PlantCARE database to search for specific cis-acting elements. All Trihelix genes in S. matsudana contained at least 10 cis-acting elements related to the light response, which was also consistent with the light response properties of the Trihelix transcription factor family ( Figure 6 )(Kaplan-Levy et al., 2012). We also identified cis-elements associated with abiotic stress. The hypothermia response element, drought response element, and hypoxia response element were present in 30, 27, and 69 gene promoter regions, respectively. Hormone-related regulatory elements were also identified, including those related to auxin, gibberellin, abscisic acid and salicylic acid, of which 77 genes contained at least one hormone-related gene.

To investigate the role of Trihelix genes in abiotic stress, we analyzed the expression pattern of SmTTFs from submergence stress and salt stress transcriptome data. A total of 25 and 32 SmTTF genes were identified as differentially expressed genes under flooding stress and salt stress, respectively, using fold change ≥ 2 and FDR < 0.01 as screening criteria. FPKM values of these genes were entered using TBtools software to construct heatmaps showing gene expression patterns under flooding stress ( Figures 7A, B ). The A-plot showed the expression pattern of 78 SmTTF genes under flooding stress, with 25 differentially expressed genes (* labeled). Fourteen genes in the GT-2 subfamily were differentially expressed genes, seven differentially expressed genes in the GT-1 subfamily, and three and one in the SIP1 and SH4 subfamilies, respectively. The GT-1 subfamily genes SmTTF02, 06, 19, 30, 37, 47, and 59 had the highest gene expression levels after 4 h of submergence. Five of the GT-2 subfamily genes were downregulated after flooding (SmTTF03, 05, 48, 62, 63), and differentially expressed genes in the SIP1 and SH4 subfamilies were also downregulated after flooding and upregulated again at 48 h of flooding.

Panel B showed the expression pattern of SmTTF genes under salt stress, with 17 differentially expressed genes in the GT-2 subfamily, seven DEGs in both the GT-1 and SIP1 subfamilies, and only one DEG in the SH4 subfamily. The differentially expressed gene SmTTF38 in the SH4 subfamily had the highest expression at 4 h of stress, followed by a gradual decrease in expression. The differentially expressed genes SmTTF15, 21, 35, 53, and 70 in the SIP1 subfamily had downregulated gene expression after salt stress, whereas SmTTF16 and 36 had upregulated expression at 4 h and subsequently downregulated expression. The differentially expressed genes in the GT-1 subfamily were upregulated after salt stress and remained relatively stable from 4-12 h. SmTTF01, 03, 31, 42 and 74 in the GT-2 subfamily were upregulated in the gene table at 8 h of stress, whereas the remaining differentially expressed genes were, in contrast, downregulated in expression after salt stress. The expression changes in these genes at different times may had direct or indirect regulatory relationships with other genes. Transcriptomes suggested that most Trihelix family genes play different roles at specific times during the response of willow to submergence stress and salt stress.

We treated 3-week-old cuttings of Suliu 795 with 200 mM mannitol, 20% PEG6000 and 4°C low temperature stress and selected 20 genes of different subfamilies for analysis ( Figure 8 ). This result indicated that the expression levels of response members differ under different abiotic stresses. SmTTF genes showed upregulated expression at 4-24 h under mannitol and low temperature treatment, while SmTTF59 of the GT-1 subfamily and SmTTF69 of the SH4 subfamily showed the highest expression at 48 h of treatment under mannitol treatment. The expression of SmTTF14 and 28 of the GT-2 subfamily and SmTTF18, 20 and 39 of the SH4 subfamily were gradually reduced under low temperature treatment. Under PEG treatment, all of the SmTTF genes, except SmTTF59 and SmTTF42, were determined to show the highest expression at 4 h of submergence, followed by a gradual decrease in expression. The results showed that SmTTF genes can respond to submergence stress, salt stress, drought stress and low temperature stress, and the potential functions of this gene family in responding to and regulating abiotic stresses were valuable for further study.

Transcriptome and qRT-PCR data showed that SmTTF30 responded significantly to a variety of abiotic stresses. Our previous study showed that SmTTF30 of the S. matsudana Trihelix gene family was identified as a flooding tolerance hub gene, and analysis of transcriptomic data revealed that the gene was significantly induced at 4 h of flooding (Chen et al., 2022b). Evolutionary tree analysis revealed that this gene is closely related to Arabidopsis HRA1 (At3g10040), and the study demonstrated that HRA1 can be involved in Arabidopsis in response to flooding stress(Giuntoli et al., 2014). To investigate whether SmTTF was involved in flooding stress, we constructed the structure of 35S:SmTTF30 and transformed it into Arabidopsis WT (Col-0). Overexpression lines were successfully obtained by genotypic identification and expression analysis of the transformed lines. Five-week-old plants of wild-type and three transgenic lines were subjected to 5 days of submergence stress. After flooding stress, the gene expression of the overexpression lines was significantly upregulated, and the gene expression pattern showed a gradual upregulation after flooding, reaching the highest level at 12 h, followed by a gradual downregulation and then a significant upregulation after reoxygenation ( Figure 9C ). Under normal conditions, there was no difference in growth between the transgenic lines and the wild-type plants. On Day 5 of submergence, wild-type Arabidopsis leaves showed significant yellowing and hyalinization with curled leaf edges, whereas the overexpression lines grew well ( Figure 9A ). Immediately after reoxygenation, the overexpression lines grew in a better state than the wild type after 4 h and 24 h of reoxygenation. The results of Evans blue staining showed that at 5 days of flooding, the blue area of wild-type Arabidopsis leaves was larger than the blue area of the overexpression lines, and the leaf cell activity was reduced and cell death occurred, while the overexpression lines had less cell damage in the leaves ( Figure 9B ). In addition, the malondialdehyde content increased after flooding stress, and the MDA content was significantly higher in the WT than in the OE lines ( Figure 9D ). The POD content also increased after flooding and was significantly higher in the OE than in the WT lines ( Figure 9E ). In addition, we determined the expression levels of hypoxia-responsive genes. The expression levels of ADH1, PCO2, SUS1, and SUS4 in the overexpression strain were significantly higher than those of the wild type after 5 days of inundation ( Figure S1 ). Taken together, these results suggested that SmTTF30 overexpression in Arabidopsis can improve plant tolerance to flooding stress.