Seeds of L. bicolor were collected from the saline inland environment (N37°20’; E118°36’) in the Yellow River Delta, Shandong, China. The seeds were well dried and stored at 4°C until use. The seeds were washed with sterile deionized water after surface disinfection with 70% (v/v) ethanol for 5 min and 6% (v/v) sodium hypochlorite (Sigma, United States) for 15–20 min. Surface-sterilized seeds were then sown onto Murashige and Skoog (Murashige and Skoog, 1962) medium (MS medium; adjusted to pH 5.8 with KOH before autoclaving). The plates were incubated at 28 ± 3°C/23 ± 3°C (day/night cycle) under a light intensity of 600 µmol/m²/s (15-h-light/9-h-dark photoperiod) and 70% relative humidity. Samples were collected and frozen in liquid nitrogen at the undifferentiated stage (5,000 leaves in stage A), the salt gland development stage (4,000 leaves in stage B), and the first true leaf was collected (Yuan et al., 2015). Total RNA was extracted for gene cloning.

The Arabidopsis thaliana accession Columbia-0 (Col-0) was used for heterologous overexpression of Lb1G04794. Seeds were first surface sterilized three times with 75% (v/v) ethanol for 4 min, during which a full eddy was applied, followed by 95% (v/v) ethanol for 1 min, repeated three times, with a full rinse with sterile water four times. The seeds were then sown onto half-strength MS medium (pH 5.8). After stratification at 4°C for 2 days, the plates were released at 22°C/18°C (day/night) under a light intensity of 150 mol/m²/s, relative humidity of 70%, and a light cycle of 16 h light/8 h dark (Sui et al., 2017). Seedlings were transferred into small pots (10 cm in diameter and 8 cm in height) containing mixed soil (soil:vermiculite:perlite, 3:1:1) after 1 week and were allowed to grow under the same growth conditions for transformation and treatment.

The first true leaves over the A-E period of L. bicolor leaf development were collected, frozen in liquid nitrogen, and stored at –80°C before total RNA extraction according to Yuan. (Yuan et al., 2015). A ReverTra Ace^® qPCR RT Kit (Japan TOYOBO CO, LTD) was used for reverse transcription to obtain cDNA for each developmental stage. Based on Iso-seq transcriptome data from L. bicolor, the primers Lb1G04794-S and Lb1G04794-A for Lb1G04794 were designed using Primer Premier 5.0, and the full-length coding sequence was amplified by PCR ( Supplementary Table 1 ).

DNA and protein sequences were compared using DNAman and DNAstar. After BLAST with Lb1G04794 as a query at the National Center for Biotechnology Information (NCBI), 33 related proteins were selected for phylogenetic tree construction, using MEGA and the ClustalX adjacency method. The percentage support at each node was determined from at least 1,000 bootstrap replicates. Using the ProtParam tool in ExPASy online software, the physicochemical properties of each protein were predicted. The hydrophilicity and hydrophobicity of all proteins were analyzed by prot-Scala in ExPASy. The software tools Signal4.14, ExPASy and Swiss-Model were used to predict the secondary and tertiary structures of the proteins and the presence of signal peptides.

A 2×Taq Plus Master Mix II and the primer pair Lb1G04794 1300-S and Lb1G04794 1300-A ( Supplementary Table 1 ) were used to amplify the full-length coding sequences with homologous terminal vectors. The Lb1G04794 coding sequence was cloned into the pCAMBIA 1300 vector containing the cauliflower mosaic virus (CaMV) 35S promoter, the hygromycin resistance gene, and the green fluorescent protein sequence (GFP). The resulting pCAMBIA 1300-Lb1G04794 vector was transformed into onion (Allium cepa) epidermal cells using Agrobacterium (Agrobacterium tumefaciens) strain GV3101 (Sun et al., 2007). The fluorescence signal of the GFP fusion protein was detected with a confocal microscope (TCS S8 MP two-photon confocal laser scanning microscope, Leica, Germany). Staining with 4′,6-diamidino-2-phenylindole (DAPI) was used to show the nucleus under 358-nm excitation.

ClonExpress^® II was used to clone the coding sequence of Lb1G04794 into the vector pGBKT7/BD via the NdeI restriction site ( Supplementary Table 1 ). The three vector pairs pGADT7-T+pGBKT7-Lb1G04794 (experimental group), pGADT7-T+pGBKT7-lam (negative control), and pGADT7-T+pGBKT7-53 (positive control) were introduced into Y2H Gold yeast cells through the yeast Maker transformation system. Yeast colonies were selected on synthetic defined (SD) medium lacking Trp (SD –Trp) for 3 days. Transcriptional activity was evaluated according to yeast growth on SD –Trp –Leu medium at 30°C for 2 days (Guo et al., 2013). β-Galactosidase activity was determined by growth on SD –Trp –Leu –Ade –His medium containing X-α-gal (Han et al., 2019).

According to transcriptome deep sequencing (RNA-seq) results of L. bicolor samples at different developmental stages, Lb1G04794 was expressed at different levels across different developmental stages (Yuan et al., 2015). To validate these results, samples were collected from the first true leaf at the A and B stages (undifferentiated stage, 4–5 days after sowing; salt gland differentiation stage, 6–7 days after sowing, using 4,000 leaves), the C and D stages (stomatal differentiation, 8–10 days; epidermal cell differentiation stage, 11–13 days, 2,000 leaves), the E stage (mature young stage, 14 days, 500 leaves), old leaves (20 days) and the E stage petioles and roots. Meanwhile, we also collected true leaf materials treated with 300 mM NaCl for 14 days. Total RNA was extracted from the above materials. A separate set of seedlings was treated with 25 mg/L salicylic acid, 0.1 mg/L methyl jasmonate, or 300 mM NaCl, then sampled at 0, 6, 12, 24, 48, and 72 h. Beacon Designer Free Edition software (Version 7.8) was used to design the primers for quantitative PCR (qPCR) of Lb1G04794, using LbTUBULIN as internal control ( Supplementary Table 1 ). PCR thermal cycling conditions were as follows: denaturation at 95°C for 5 min, followed by 40 cycles of denaturation at 94°C for 20 s, annealing at 58°C for 15 s, extension at 65°C for 15 s. Three biological replicates were analyzed. Relative expression levels were calculated according to the formula 2^−ΔΔC(T).

The first true leaves of L. bicolor seedlings were collected, and genomic DNA was extracted by the cetyltrimethylammonium bromide (CTAB) method. The Lb1G04794 promoter sequence was identified in the L. bicolor genome sequence (Lescot et al., 2002), and specific primers (Lb1G04794-P-S and Lb1G04794-P-A) were designed using Primer Premier 5.0 ( Supplementary Table 1 ). The promoter fragment was amplified by PCR with a 2×Taq Plus Master Mix and cloned, as detailed below.

The CaMV 35S promoter in pCAMBIA3301-35S-GUS vector was excised by digestion with HindШ/NcoI to linearize the vector. The Lb1G04794pro:GUS reporter construct was obtained by linking the linearized vector and the Lb1G04794 promoter with ClonExpress^® II. The resulting construct was transformed into wild-type Arabidopsis plants by Agrobacterium-mediated transformation (strain GV3101). Transgenic plants were obtained by selection on basta. A GUS staining kit (Zhongkelitai Biological Technology Co, LTD; Cat No : RTU4032)was used for staining: a 50× X-Gluc concentrated solution was diluted 50 times with GUS staining buffer. The prepared materials were soaked in GUS staining solution overnight at 25–37°C. All materials were then transferred into anhydrous ethanol for chlorophyll clearing two to three times until the negative control material turned white. GUS staining was observed with the naked eye or under a microscope.

After linearization of pCAMBIA3301-35S-GUS vector by digestion with NcoI to remove the 35S promoter coding sequence, the Lb1G04794 coding sequence was amplified by PCR with the primers Lb1G04794 3301-S and Lb1G04794 3301-A ( Supplementary Table 1 ) and cloned into the linearized vector with the ClonExpress^® II recombination reaction system to generate 35S: Lb1G04794. The resulting construct was introduced into Agrobacterium strain GV3101 and transformed into Arabidopsis Col-0 by Agrobacterium-mediated floral dipping (Clough and Bent, 1998). After three generations of selection for herbicide resistance, homozygous 35S:Lb1G04794 lines were identified. Genomic DNA was extracted for PCR with the primers pCAMBIA-S and Lb1G04794-A to confirm transgenic plants harboring the overexpression construct ( Supplementary Table 1 ). Total RNA of 35S:Lb1G04794 transgenic lines was extracted with a FastPure Plant Total RNA Isolation Kit (Vazyme, China). The expression levels of Lb1G04794 were measured in the transgenic lines by RT-qPCR using Lb1G04794 RT-S and Lb1G04794 RT-A primers ( Supplementary Table 1 ), using Arabidopsis ACTIN2 as internal control (primers ACTIN2 sense and ACTIN2 anti). The expression of each transgenic line was repeated three times. The line with the lowest Lb1G04794 expression level (line OE14) was used as the control (with relative expression level set to 1) to calculate the relative expression level of Lb1G04794 in the other overexpression lines (Leng et al., 2021).

pCAMBIA3301-35S-GUS and pCAMBIA3301-35S-Lb1G04794-GUS was transferred into Agrobacterium EHA105 and used to infect L. bicolor referring to Yuan et al. (2104). After the shoot regeneration, the regenerated leaves were transferred to the root regeneration medium. After screened in hygromycin for two weeks, the regenerated seedlings were used for qRT-PCR to verify the expression level in overexpression line. Then, the leaves were fixed in Carnoy (ethanol: acetic acid, 3:1) for 12 h and decolorized with 70% ethanol for 12 h, and finally decolorized with Hoyers’ solution (chloral hydrate saturated with lactic acid solution). The structure and morphology of the salt glands were observed under the excitation light of 330–380 nm under the fluorescence microscope, and then the number of salt glands in a single leaf was counted.

The phenotypes of one-week-old T₃ homozygous seedlings were observed using an anatomical microscope (Nikon, Japan). After the first pair of true leaves had fully expanded, the total number of trichomes was counted on the first pair of rosette leaves, with 20 seedlings analyzed for each line. The number and length of root hairs from five OE-Lb1G04794 seedlings were scored, with 20 seedlings per line.

Three Arabidopsis transgenic lines overexpressing Lb1G04794 at high, medium, or low expression levels, together with Arabidopsis wild-type Col-0, were treated with NaCl. All seeds were sown on half-strength MS medium alone or containing different concentrations of NaCl (50, 100, and 150 mM). After stratification for 2–3 days, the plates were released into tissue culture chambers. After 24 h, the germination percentage was scored as the emergence of the radicle through the seed coat. Thegermination percentage was calculated as follows: germination percentage (%) = number of germinated seeds/total seeds × 100%. Since the germination percentage of overexpressing lines was significantly lower than that of Col-0, the relative germination percentage of impermeable salt treatment was calculated as follows: (germination percentage under control − germination percentage under NaCl treatment)/germination percentage under control × 100% on half-strength MS medium.

All seeds were evenly sown on medium with different NaCl concentrations (0, 50, 100, and 150 mM) as three replicates. Seedlings were photographed after 5 days with an anatomical microscope, and root length was measured in ImageJ. As the relative inhibition rate of root length elongation was significantly lower than that of the WT, the relative shortening rate was also calculated. The relative inhibition rate of root length elongation was calculated as root length under control – root length under NaCl treatment)/root length under control × 100% on half-strength MS medium.

According to Han et al. (Han et al., 2019) and Guo et al. (Guo, 2017), 7-day-old Arabidopsis seedlings were transferred to nutrient soil and then treated with different concentrations of salt for 1 week. The fresh weight and various physiological indicators were determined after salt treatment for 1 week. The relative decrease in fresh weight was calculated as (fresh weight under control – fresh weight under NaCl treatment)/fresh weight under control × 100%. The relative reduction in dry weight was calculated as (dry weight under control – the dry weight under NaCl treatment)/dry weight under control × 100%. To determine the contents of various small molecules, 0.5 g of seedlings growing under different NaCl concentrations was collected and Na⁺, K⁺, malondialdehyde (MDA), and proline contents were measured. Ion concentrations were determined using a flame photometer (M410, Sherwood, United Kingdom). Five replicates were performed for each line.

Endogenous phytohormones were extracted from 7-day-old Arabidopsis seedlings by the isopropanol-water-hydrochloric acid method. Endogenous phytohormone contents were determined with an Agilent 1290 high-performance liquid chromatography (HPLC) tandem AB Sciex QTRAP 6500⁺ mass spectrometer, and internal standard substances were added during extraction. Since the growth and development of overexpression lines were clearly weaker than those of the WT, the contents of indole-3-acetic acid (IAA) and ABA were specifically targeted for determination.

Total RNA was extracted from Arabidopsis seedlings grown on half-strength MS medium for 5 days and transplanted to soil for 2 weeks. The expression of six stress-related genes, SALT OVERLY SENSITIVE1 (At2G01980, AtSOS1), SOS2 (At1G01140), SOS3 (At5G35410), HIGH-AFFINITY K⁺ TRANSPORTER1 (At4G1030, AtHKT1), Na⁺/H⁺ EXCHANGER1 (At5G27150, AtNHX1), and GST CLASS TAU5 (At2G29450, AtGSTU5), was determined by RT-qPCR ( Supplementary Table 1 ). Three biological replicates were performed. The formula 2^−ΔΔC(T) was used to calculate relative expression, with ACTIN2 used as internal reference.

Statistical analysis was performed using SPSS at P = 0.05 (Duncan’s multiple range tests). Analysis of variance (ANOVA) with orthogonal contrasts and mean comparison procedures was used to detect differences between the treatments.

Based on the full-length sequence of Lb1G04794, we cloned a 1,074-bp open reading frame encoding a 357–amino acid protein with a predicted molecular weight of 39,928.25 Da and an isoelectric point (PI) of 6.36 ( Supplementary Figure 1A ). The gene sequence was verified to be consistent with the genomic data by sequencing. Of these 357 amino acids, 110 were hydrophobic, accounting for 30.8% of the total amino acid number, and 247 were hydrophilic (or 69.2%). The predicted protein had an aliphatic index of 65.01, indicating that the protein encoded by Lb1G04794 is a hydrophilic protein without a transmembrane helical structure ( Supplementary Figure 1B ). We identified no homologous protein for Lb1G04794 in Arabidopsis. We thus used a BLAST search at NCBI with the predicted Lb1G04794 protein sequence as query ( Supplementary Figure 2 ). We determined that the protein encoded by Lb1G04794 is a hypothetical protein with no transmembrane domain ( Supplementary Figure 1C ), no signal peptide ( Supplementary Figure 1D ), and no conserved domain ( Supplementary Figure 1E ), making its function completely unknown. An analysis of the Lb1G04794 promoter (2,000 bp upstream of the ATG) identified a core promoter element and a typical cis-element involved in light responses, a cis-element associated with methyl jasmonate (MeJA) responses (CGTCA-motif), a cis-acting element involved in salicylic acid (SA) reaction (TCA-element), and a cis-acting element involved in defense and stress response (TC-rich repeats) ( Figure 1F ). We generated transgenic Arabidopsis lines harboring the Lb1G04794 promoter driving the transcription of the β-GLUCURONIDASE (GUS) reporter gene; we observed GUS staining in cotyledons and roots, especially in the root tip of these transgenic reporter lines ( Figure 1E ).

To determine the subcellular localization of the protein encoded by Lb1G04794, we transiently transformed onion epidermal cells with Agrobacterium harboring the vector p1300-Lb1G04794, encoding a fusion protein between GFP and the protein encoded by Lb1G04794. Observations with a two-photon fluorescence inverted microscope showed that GFP-Lb1G04794 localizes specifically in the nucleus ( Figure 1A ), while the empty vector overexpressing free GFP resulted in green fluorescence in both the nucleus and the cytoplasm. We also studied the expression of Lb1G04794 in L. bicolor at different developmental stages ( Figure 1B ) and under different treatments ( Figure 1C ). Lb1G04794 was highly expressed during stages A and B, responded to salt treatment, and was most highly expressed in leaves compared to other tissues. We also observed that Lb1G04794 expression responded to NaCl, SA, and MeJA treatments, with a peak in Lb1G04794 transcript levels 6–12 h into the treatment, followed by a gradual decrease back to normal levels ( Figure 1C ). Finally, we determined that the protein encoded by Lb1G04794 displays self-activation activity in yeast when fused to the GAL4 DNA-binding domain ( Figure 1D ).

The Lb1G04794 was overexpressed in L. bicolor ( Figure 2 ), after expression level verification ( Figure 2A ), compared with the control transformed with the empty vector, overexpression of Lb1G04794 can significantly enhanced salt gland development ( Figures 2B, C ). This indicated that Lb1G04794 may participate in promoting salt gland differentiation.

We isolated Arabidopsis transgenic lines overexpressing Lb1G04794 from the cauliflower mosaic virus (CaMV) 35S promoter. We confirmed the presence of the transgene by PCR and its expression by RT-qPCR ( Supplementary Figure 3 ). We selected the overexpression lines OE-21, OE-1, and OE-4 as lines with high, medium, and low expression levels of the transgene for subsequent experiments ( Supplementary Figure 3 ). As Lb1G04794 was highly expressed during the development of salt glands in L. bicolor, we first characterized trichome and root hair development.

We counted the number of trichomes of the first fully extended true leaf of the Arabidopsis transgenic lines, which indicated that the overexpression lines have significantly more trichomes than the WT ( Figures 3A, C ). We identified Lb1G04794 by screening for genes with high expression during salt gland development, underscoring the correlation between salt gland and trichome development. Similarly, we counted the number of root hairs over a 1-cm region from the root tip and observed that overexpression of Lb1G04794 results in fewer root hairs, together with a shorter root, relative to the WT ( Figures 3B, C ).

Considering that Lb1G04794 expression was induced by NaCl treatment ( Figure 1C ) and that fewer root hairs developed in Lb1G04794 overexpression lines ( Figure 3B ), we explored salt tolerance–related indices at the germination stage in the transgenic lines. To this end, we sowed seeds from Arabidopsis and three transgenic lines onto medium containing different NaCl concentrations to determine the effect of NaCl on seed germination ( Figure 4A ). Notably, the germination of all 35S:Lb1G04794 overexpression lines was significantly lower under all conditions than in the WT of 24, 48, and 72 h after sowing, even when sown on control medium lacking added NaCl.

Although the germination percentage of the transgenic lines was lower than that of the WT ( Figure 4B ), the relative inhibition of root elongation ( Figure 4C ) was less affected by NaCl treatment than in the WT. These results indicated that germination and growth are inhibited upon overexpression of Lb1G04794, but the transgenic lines showed a greater salt tolerance than the WT under equivalent NaCl conditions.

The germination percentage of the transgenic lines was clearly inhibited under control (0 NaCl) conditions, which prompted us to measure the contents of endogenous phytohormones, with a focus on the three plant hormones IAA and ABA. We established that IAA contents in the overexpression lines are significantly lower than in the non-transgenic WT ( Supplementary Figure 4 ), a result that was in agreement with the shorter roots of these lines, while the contents for ABA were similar between all genotypes. The lower IAA levels explained the shorter roots seen in the overexpression lines, indicating that Lb1G04794 may negatively regulated IAA biosynthesis. These changes in plant hormone levels in the overexpression lines may reflect internal physiological changes of the overexpressing lines.

We also investigated the long-term effect of NaCl treatment on the growth of Arabidopsis transgenic lines after a 2-week NaCl treatment in soil ( Figure 5A ). Although the biomass of the transgenic lines was lower than that of the WT, their relative dry and fresh weights appeared less affected by increasing NaCl concentrations than the WT ( Figures 5B, C ).

We also measured several physiological indices (Na⁺, K⁺, proline, and MDA contents) to explore the relationship between Lb1G04794 and salt stress. Under 100 mM NaCl treatment, the proline contents of 35S:Lb1G04794 lines were lower than in WT. However, MDA contents in the overexpression lines under control conditions were higher than in the WT, suggesting that Lb1G04794 overexpression imposes a degree of cellular stress in Arabidopsis. Na⁺ and K⁺ contents showed no significant differences between genotypes when treated with 100 mM NaCl ( Figure 5D ). MDA contents in the 35S:Lb1G04794 lines increased to a lesser extent than in the WT when exposed to 100 mM NaCl ( Figure 5E ).

We determined the expression levels of the six salt tolerance marker genes AtSOS1, AtSOS2, AtSOS3, AtHKT1, AtNHX1, and AtGSTU5 by RT-qPCR. Notably, all of these genes displayed the similar trends ( Figure 6 ). Under control conditions, all six genes showed the same expression levels in the WT and the overexpression lines, with little significant difference between genotypes. Upon treatment with 100 mM NaCl, however, the transcript levels of all salt stress–related marker genes increased significantly in all genotypes, although the expression levels in the overexpression lines were significantly lower than those in the WT.