The full-length TheIF1A gene is 432 bp and encodes a 143-aa protein with a predicted molecular weight of 16.40 kDa and a theoretical pI of 5.09. Multiple sequence alignment analysis indicated that the eIF1A proteins from different plants share a highly conserved eIF1A domain (Supplementary Figure S1A) located between residues 28 and 110 but without SANT, TUDOR, FN1, CNX, or PLDc features. Phylogenetic analysis indicates that TheIF1A is most similar to eIF1A from B. vulgaris (Supplementary Figure S1B).

A 1352-bp (from -1 to -1352) promoter region was cloned using thermal asymmetric interlaced polymerase chain reaction (TAIL-PCR). Searches using the PLACE algorithm demonstrated the presence of many cis-elements, such as ABRELATERD1, CAATBOX1, DOFCOREZM, MYB1AT, TATABOX2, and WRKY71OS, in this region, which indicates that TheIF1A may be regulated by different types of transcription factors (TFs) (Supplementary Figure S2).

The results of qRT-PCR showed TheIF1A to be induced in response to NaCl and polyethylene glycol (PEG) treatments. After NaCl treatment for 6 h, expression of TheIF1A in roots was increased by 187.6-fold compared to the control. At 48 h, TheIF1A expression was down-regulated in roots compared to the control but up-regulated by 1.97-fold in stems (Figure 1A). Under PEG stress, TheIF1A transcription was considerably higher in leaves than in roots and stems, with TheIF1A transcripts in leaves being 4.22-fold higher in roots and 2.43-fold higher in stems at 12 h (Figure 1B).

To further investigate expression via the TheIF1A promoter, transgenic Arabidopsis plants expressing GUS under the control of the TheIF1A promoter (namely, pTheIF1A::GUS) were generated (Figure 2A). The results showed GUS activity throughout the entire plant, except for in immature seeds of the silique (Figures 2B–P). Furthermore, expression levels differed among tissues and developmental stages. pTheIF1A::GUS was also transiently transformed into T. hispida seedlings followed by GUS staining, with the highest GUS activity in roots and the lowest in stems (Figure 2Q–S), further confirming the results for the tissue expression profile of TheIF1A.

Twelve Dof motifs were identified in the TheIF1A promoter, suggesting that TheIF1A is likely regulated by TFs that interact with this motif. A yeast one-hybrid assay was conducted to identify upstream regulators using the reporter vector pHIS2-cis (containing triple tandem Dof motif repeats) as bait to screen a Tamarix TF cDNA library. One protein (ThDof, KF896302) was found to bind to the Dof motif but failed to bind to mutant Dof motifs (Figure 3A). In addition, ThDof could bind to a truncated TheIF1A promoter retaining the Dof motif but could not bind to promoter fragments lacking the motif or with a mutated Dof motif (Figure 3A). These findings suggest that ThDof may specifically bind to the Dof motif and may regulate TheIF1A by binding to this motif in the TheIF1A promoter.

To validate these interactions, we co-transformed reporter plasmids (the motifs, promoter fragments and serially mutated motifs followed by a 46-bp minimal promoter were independently cloned into pCAMBIA1301) and the effector construct (pROKII-ThDof, Supplementary Figure S3B and Figure 3B) into Arabidopsis leaves. Measurement of GUS activity and histochemical staining both demonstrated that activation of GUS in Arabidopsis cells when co-transformed with reporters containing complete motifs or the TheIF1A promoter-including motifs. However, when the reporter contained a mutated motif or TheIF1A promoter fragments without a motif, the Arabidopsis cells displayed less GUS staining and decreased GUS activity (Figures 3B,C). These findings clearly confirmed that ThDof may activate expression of TheIF1A by binding to the Dof motif present in its promoter.

The expression patterns of TheIF1A and ThDof in T. hispida under salt and drought stresses were analyzed by qRT-PCR, and the results indicated that they share similar expression profiles. Under salt treatment, both genes exhibited peak transcription at 6 h in roots and stems, reaching peak levels in leaves at 24 h (Figure 1A). Under drought conditions, the lowest expression of both genes was observed at 48 h in roots and stems. In leaves, both genes showed highest and lowest expression at 12 and 72 h, respectively (Figure 1B).

As TheIF1A was induced by NaCl and PEG stresses, we investigated whether overexpression of TheIF1A in plants enhances salt and osmosis tolerance using 35S::TheIF1A-transgenic tobacco plants of the T₃ generation. PCR analysis of the transgenic lines confirmed the successful transformation of TheIF1A into tobacco (Supplementary Figure S4A). The three lines of transgenic tobacco showing the highest expression were Lines 1, 3, and 9 (439.58-, 252.48-, and 508.46-fold compared with tubulin, respectively) (Supplementary Figure S4B), and these lines were selected for further analysis.

Germination rates were compared between the transgenic lines and wild type (WT) plants. Without treatment, there was no observable difference between WT and transgenic plants (Figure 4A). However, germination in the transgenic plants was significantly increased compared with WT under mannitol stress (Figures 4A,B) (p < 0.05). The water loss rates of the transgenic and WT plants (Figure 4C) were compared after leaves were exposed to air for 10 min, and the rate of water loss in the transgenic lines was 83% of that in WT; after 70 min, the water loss rate of the transgenic lines was also significantly decreased (80%) compared with WT plants. Although the chlorophyll content (tcc) of the transgenic plants was different from that in WT plants prior to salt or mannitol stress, tcc in all plants decreased strongly after 6 days of salt or mannitol stress. Nonetheless, the transgenic lines showed significantly higher tcc than WT plants (the difference was, on average, 1.20–1.26-fold) (Figure 4D), which suggests that overexpression of TheIF1A prevents chlorophyll loss under salt or mannitol stress conditions.

To understand the regulation of physiological responses, physiological parameters involved in stresses were compared, including the ROS content, the activity of protective enzymes, such as SOD, POD, GST and GPX, EL rates and the MDA content. Staining with diaminobenzidine (DAB) and nitroblue tetrazolium (NBT) demonstrated that H₂O₂ and O^2- levels, respectively, in leaves were similar in WT and transgenic lines under no treatment. However, WT plants exhibited increased H₂O₂ and O^2- accumulation compared with transgenic lines, particularly at 1 h after stress treatment, at which point the degree of staining in the transgenic lines was notably decreased compared with the WT plants (Figures 5A,B). ROS levels in intact guard cells and the main roots, as determined by stained with H₂DCF, were similar to the levels indicated by DAB and NBT staining. That is, with the exception of 0 h, WT plants exhibited increased ROS accumulation in both guard cells and main roots compared with transgenic lines (Figures 5D,E). Additionally, the H₂O₂ content according to the degree of staining showed significantly less H₂O₂ in TheIF1A-transgenic plants than in WT plants (Figure 5F, p < 0.05). These findings indicate that TheIF1A overexpression leads to a significant decrease in ROS accumulation in plant cells under salt and mannitol stress conditions.

Superoxide dismutase, POD, GST, and GPX are key ROS scavengers, and our results demonstrated similar changes in the activities of these enzymes. Under no treatment, no significant differences in the activities of these enzymes were observed between transgenic and WT plants. However, the activities of these enzymes in the transgenic plants were significantly (p < 0.05) increased compared with the WT plants under salt and mannitol stresses. For example, the SOD activity of the transgenic lines was, on average, approximately 1.4-fold that of WT under both stresses, and POD activity in the former was 1.41-fold that of the latter under salt stress. GST activity of 1A#3 was 2.16-fold that of WT exposed to NaCl, and GPX activity of 1A#3 was 1.63- and 1.62-fold of that of WT in response to NaCl and mannitol, respectively. Although the MDA content did not differ between transgenic lines and WT plants under no stress, the transgenic plants accumulated significantly less MDA compared with the WT plants after treatment. The average MDA content of WT was 1.40- and 1.23-fold of the transgenic lines under NaCl and mannitol stresses, respectively (Figure 6). These findings indicated that overexpression of TheIF1A increased enzyme activities, subsequently decreasing MDA accumulation.

In addition, the plasma membrane system was analyzed using Evans blue staining and EL rate determination. The tested lines did not exhibit apparent differences under control conditions. However, when treated with salt or mannitol, the transgenic plants exhibited weaker Evans blue staining and a lower EL rate compared with WT plants (the EL of WT was, on average, 1.33-fold higher than that of transgenic lines exposed to NaCl and 1.47-fold following mannitol), and these differences were significant (Figures 5C, 6).

Taken together, these findings suggested that TheIF1A overexpression improves salt and osmotic stress tolerance by increasing ROS scavenging and preventing cell damage to maintain growth.

Tobacco plants overexpressing TheIF1A showed resistance to NaCl and mannitol stresses, and the activities of the corresponding protective enzymes were increased in transgenic lines compared with WT plants, suggesting that expression of key potential stress-related genes may be altered by exogenous expression of TheIF1A. Stress response-related genes were selected for analysis, and seven tobacco genes, including GST, MnSOD, NtMPK9, poxN1, CDPK15, and TOBLTP, exhibited increased levels of expression in TheIF1A-overexpressing lines compared with WT plants. In particular, levels of GST and MnSOD expression were 5.63- and 10.60-fold higher, respectively, than those in WT plants (Figure 7). These findings indicated that TheIF1A improves ROS scavenging via regulation of antioxidant genes.

To further confirm the role of TheIF1A in the T. hispida response to abiotic stress, 35S::TheIF1A and RNAi::TheIF1A were transiently transformed into T. hispida, and TheIF1A expression in two lines was determined by qRT-PCR. The results showed the highest TheIF1A expression in the 35S::TheIF1A plants and the lowest in the RNAi::TheIF1A plants. The level of TheIF1A expression in the overexpression line was 27.9-fold higher than that in the control plants and was 50.01-fold that in the RNAi::TheIF1A plants under no stress. These findings indicate the successful overexpression of TheIF1A in the 35S::TheIF1A plants and the RNAi-mediated silencing of TheIF1A in the RNAi::TheIF1A plants (Figure 8A). Interestingly, the expression patterns of ThSOD, ThPOD, ThGSTZ1, and ThGPX were similar to that of TheIF1A. Expression of these genes was highest in 35S::TheIF1A lines and lowest in RNAi::TheIF1A lines (Figure 8A), which suggests that the abiotic stress response of TheIF1A may correlate with the level of ThSOD, ThPOD, ThGSTZ1, and ThGPX expression.

The related physiological indexes were determined. Under NaCl treatment, SOD, POD, GST, and GPX activities in the control plants were 49∼77% of the activities in the 35S::TheIF1A plants, and the activities of these enzymes in the RNAi::TheIF1A plants were 36∼63% of the activities in the 35S::TheIF1A lines. After mannitol treatment, SOD, POD, GST, and GPX activities in the control plants were 56∼77% of the activities in the 35S::TheIF1A plants; the activities of these enzymes in the RNAi::TheIF1A lines were 32∼67% of the activities in the 35S::TheIF1A seedlings. The MDA content in the 35S::TheIF1A plants was 71% of that in the control plants and 59% of that in the RNAi::TheIF1A plants under NaCl stress. After treatment with mannitol, the MDA content in the 35S::TheIF1A plants was 70% of that in the control plants and 58% of that in the RNAi::TheIF1A lines. The EL rate in the 35S::TheIF1A plants was 61% of that in the control plants and 48% of that in the RNAi::TheIF1A lines under NaCl stress. However, after exposure to mannitol, the EL rate in the 35S::TheIF1A plants was 75% of that in the control plants and 61% of that in the RNAi::TheIF1A plants (Figure 8C), which indicates that TheIF1A effectively increased enzymatic activity and decreased cell damage.

Furthermore, DAB, NBT and Evans blue staining also demonstrated that 35S::TheIF1A T. hispida seedlings accumulated less ROS and suffered less cell damage and that RNAi::TheIF1A plants accumulated more ROS and suffered more cell damage compared with control plants after treatment with 100 mM NaCl or 200 mM mannitol (Figure 8B). Taken together, these findings indicate that TheIF1A improved plant tolerance via regulation of other stress-related genes, the management of ROS homeostasis, and the prevention of cell damage and death.

Tamarix hispida and Arabidopsis thaliana (Col, WT) seedlings were cultivated in pots containing a mixture of turf peat and sand (2:1 v/v) in a greenhouse for 2 months under 14 h light/10 h dark, 70–75% relative humidity, an average temperature of 24°C and well-watered conditions. Two-month-old T. hispida seedlings were exposed to the following treatments: 400 mM NaCl or 20% (w/v) PEG₆₀₀₀ for 6, 12, 24, 48, or 72 h. The roots, stems, and leaves from at least 20 seedlings were then independently harvested at the indicated time points after each treatment and pooled for qRT-PCR analyses. Seedlings irrigated with water alone were used as the control; every treatment was carried out three times as biological replicates, with every replicate containing at least 20 seedlings.

The TheIF1A gene (KF801668) was cloned from the T. hispida leaf cDNA library (Gao et al., 2008) The phylogenetic tree of eIF1A proteins from different plant species was constructed using the Neighbor-Joining method (Saitou and Nei, 1987). The TheIF1A promoter was PCR amplified from T. hispida genomic DNA using a genome walking kit. cis-elements in the TheIF1A promoter were identified using the PLACE database¹ (Higo et al., 1999). T₃ seedlings of TheIF1A promoter transgenic line were used for temporal and spatial expression analyses of TheIF1A via GUS staining (Clough and Bent, 1998). Moreover, pCAMBIA1301-TheIF1A was transiently expressed in T. hispida seedlings to verify the TheIF1A expression results obtained in A. thaliana.

Total RNA was isolated from seedlings exposed to different treatments using the CTAB method (Wang et al., 2002); 0.5 μg of RNA was used for reverse transcription with PrimeScript^TM RT reagent Kit (Takara). The resulting cDNA product was diluted 10-fold and used for RT-PCR analyses. qRT-PCR was applied using an MJ Opticon^TM2 machine (Bio-Rad, Hercules, CA, USA); the reaction system and procedures were performed according to Gao et al. (2011). β-Actin (FJ618517), α-tubulin (FJ618518), and β-tubulin (FJ618519) were used as reference genes. The primer sequences are listed in Supplementary Table S1. Relative expression levels were calculated using the delta-delta Ct method (Livak and Schmittgen, 2001). The RT-PCR data for the transgenic tobacco lines were normalized to tubulin (AJ421412) (Ban et al., 2011). Three independent experiments were performed, and at least 20 seedlings were used in each treatment.

Twelve Dof motifs (core base sequence “AAAG”) were identified in the TheIF1A promoter (Supplementary Figure S2). A yeast one-hybrid assay (Clontech, Palo Alto, CA, USA) was used to investigate specific genes recognizing these motifs and activating TheIF1A expression. The three tandem copies of the promoter fragment containing the Dof motif core (“AAAAGT”) sequence were cloned into the pHIS2 vector (Supplementary Figures S3A,B). TFs from different families were identified, PCR amplified and subsequently cloned into pGADT7-Rec2 (Clontech) to produce a cDNA library used in the one-hybrid assay (Zheng et al., 2013).

To determine interactions between the Dof motif and related positive clones, a mutated Dof core motif “AAAG” with “CCCT” (Dof-M) was constructed. Furthermore, fragments of the TheIF1A promoter containing the Dof motif (Dof-S), excluding the Dof motif (Dof-S-M1) or containing a mutated Dof motif (Dof-S-M2) were cloned into pHIS2 (Supplementary Figure S3B). Cells transformed with the empty p53HIS2 plasmid (pHIS2 contains three copies of the p53 DNA element) were used as a negative control in the yeast one-hybrid assay. All primers are presented in Supplementary Table S2.

To confirm these interactions, all constructs were fused with a CaMV35S-46 minimal promoter and cloned into pCAMBIA1301 for expression of the GUS gene. The full-length cDNA of ThDof was cloned into pROKII under the control of the 35S promoter (referred to as pROKII-ThDof) to create effecter vectors (Supplementary Figure S3C), which were transformed into Arabidopsis using the floral dip method (Clough and Bent, 1998). A T₃ generation line of ThDof-transgenic Arabidopsis was then used for transient expression of all Dof reporters using Agrobacterium-mediated transformation. All co-transformed Arabidopsis leaves were stained to measure GUS activity (Jefferson et al., 1987; Jefferson, 1989). Furthermore, the expression patterns of TheIF1A and ThDof under salt and drought conditions in T. hispida were analyzed by qRT-PCR.

The full-length cDNA of TheIF1A was amplified and cloned into pROKII (referred to as 35S::TheIF1A). The primer sequences are presented in Supplementary Table S2. 35S::TheIF1A was transferred into tobacco (Nicotiana tabacum) using Agrobacterium-mediated transformation. Sixty-one lines were generated in the T₀ generation, which were further selected until the T₃ generation; 12 different lines were detected by PCR, and the expression levels were analyzed by qRT-PCR. Three transgenic lines (Lines 1, 3, and 9, referred to as 1A#1, 1A#3, and 1A#9, respectively) with the highest expression levels were selected for further analysis.

For the seed germination assay, seeds from WT and transgenic lines were sown onto 1/2 MS agar medium containing either 100 mM NaCl or 200 mM mannitol. Germination was recorded after 12 days using germination rate = germinated plantlets number/total seeds sown^∗100%. For stress tolerance assays, 28-days-old WT and transgenic plants were treated with 100 mM NaCl or 200 mM mannitol for 6 days. Leaves were used to determine tcc, SOD activity, POD activity, GPX activity, GST activity and the MDA content using a corresponding commercially available kit from Nanjing Jiancheng Bioengineering Institute (Nanjing, China). SOD activity was assessed based on the auto-oxidation of hydroxylamine. POD activity was tested using a peroxidase assay kit, with absorbance at 340 and 420 nm. GST activity was determined using a glutathione S-transferase (GSH-ST) assay kit (Colorimetric method). GPX (GSH-Px) activity was determined by the velocity method using a GSH-Px kit; the reaction was initiated by the addition of H₂O₂, and the change in absorbance during the conversion of GSH to GSSG was recorded spectrophotometrically at 412 nm. The MDA content was determined based on thiobarbituric acid (TBA) reactivity; the developed red color of the resulting reaction was measured at 532 nm using a spectrophotometer. The H₂O₂ content was assessed on the basis of a hydrogen peroxide assay kit (Minotti et al., 1994; Kwak et al., 2005; Ban et al., 2011; Liu et al., 2013). The EL rate was evaluated per the report of Yang et al. (2014). For water-holding ability, the aerial parts of seedlings were placed onto clean filter paper and dried for 70 min. The fresh weights of the samples were measured every 10 min to determine the rate of water loss relative to each time interval.

In addition, leaves from seedlings treated with 100 mM NaCl or 200 mM mannitol for 0 (control), 1, 2, and 3 h were detached and immediately used for histochemical staining analysis. O^2- accumulation, H₂O₂ accumulation, and cell death in the leaves were assayed through NBT, DAB, and Evans blue staining, respectively. ROS production in intact guard cells and roots was assessed by staining with 5 μM H₂DCF-DA (Fluka) (Fryer et al., 2002; Zhang et al., 2011). ROS accumulation was visualized using an LSM410 confocal laser scanning microscope (CLSM) (Zeiss, Jena, Germany) with excitation at 488 nm and an emission at 525 nm, and images were acquired using ZEN 2009 light edition imaging software (Zhang et al., 2011). The corresponding H₂O₂ content under the same treatments as staining assays was also examined. Furthermore, stress-related genes (TOBLTP, GST, MnSOD, NtMPK9, poxN1, and CDPK15) in TheIF1A-overexpressing tobacco plants were investigated by RT-PCR using the primers presented in Supplementary Table S3. All experiments were performed three times.

To better analyze the salt and mannitol response of TheIF1A, 295-bp of the TheIF1A sequence was cloned into a reconstructive ProKII vector to construct the RNAi-suppression vector RNAi::TheIF1A (Supplementary Figure S5) (Zhang et al., 2012; Li et al., 2014). The forward and reverse primers are presented in Supplementary Table S2. A single colony of EHA105 harboring 35S::TheIF1A, RNAi::TheIF1A and the empty ProKII vector (used as a control, labeled ck) was cultivated at 28°C to OD₆₀₀ = 0.6, after which the cells were harvested by centrifugation and adjusted to OD₆₀₀ = 0.2 using 1/2 MS solution (pH = 5.8) containing 3% (w/v) sucrose, as the transformation solution supplied with 150 μM acetosyringone. Two-month-old T. hispida seedlings were soaked in the transformation solution for 48 h at 25°C and 30–40 rpm, and the culture solution was changed every 4 h (Zheng et al., 2012). The transformed seedlings were treated with 100 mM NaCl or 200 mM mannitol for 1 h. DAB, NBT and Evans blue staining was subsequently used to visualize the O^2- and H₂O₂ contents and cell damage in the leaves; POD, SOD, GST, GPX activities and the MDA content were also examined. In addition, the expression levels of stress-related genes and ThSOD, ThPOD, ThGSTZ1, and ThGPX in seedlings of these three transgenic lines were examined by qRT-PCR. The primers are presented in Supplementary Table S1. All experiments were performed three times. Every independent replicate included at least 12 seedlings per line.

All data were analyzed using Statistical Package for Social Science (SPSS) (SPSS, Chicago, IL, USA). Differences among the tested lines were calculated using Tukey’s multiple comparison tests. The significance level was set as p < 0.05, and the standard deviation (SD) was calculated to analyze sample variability.