A novel gene LbHLH from the halophyte Limonium bicolor enhances salt tolerance via alleviating osmotic stress and reducing root hair development

Identifying genes involved in salt tolerance in the recretohalophyte Limonium bicolor could facilitate the breeding of crops with enhanced salt tolerance. Here we cloned the previously uncharacterized gene LbHLH and explored its role in salt tolerance. The 2,067-bp open reading frame of LbHLH encodes a 688-amino-acid protein with a typical helix-loop-helix (HLH) domain. In situ hybridization showed that LbHLH is expressed in salt glands of L. bicolor. LbHLH localizes to the nucleus, and LbHLH is highly expressed during salt gland development and in response to NaCl treatment. To further explore its function, we heterologously expressed LbHLH in Arabidopsis thaliana under the 35S promoter. The overexpression lines showed signicantly increased trichome number and reduced root hair number. LbHLH might interact with GLABRA1 to inuence trichome and root hair development, as revealed by yeast two-hybrid analysis. The transgenic lines showed higher germination percentages and longer roots than the wild type under NaCl treatment. Analysis of mM NaCl LiCl ionic that overexpressing LbHLH relieved osmotic results

L. bicolor is a typical recretohalophyte, with salt glands on its stems and leaves that excrete excess salt ions [16,17]. L. bicolor is considered to be a pioneer plant for improving saline soils. It is easy to observe salt glands in this plant under a uorescence microscope, as they exhibit blue auto uorescence [18]. The development of the rst true leaves of L. bicolor can be divided into ve stages: the undifferentiated stage (A), the salt gland differentiation stage (B), the stomata differentiation stage (C), the pavement cell differentiation stage (D), and the mature stage (E) [19].
Transcriptome pro ling of leaves at these stages has uncovered various candidate genes involved in salt gland differentiation [13,19,20]. Some of these genes are highly homologous to genes related to trichome development in other plants, such as GLABRA1 (GL1), TRANSPARENT TESTA GLABRA1 (TTG1), GLABRA3 (GL3), ENHANCER OF GLABRA 3 (EGL3), SUPER SENSITIVE TO ABA AND DROUGHT2 (SAD2), TRIPTYCHON (TRY), and CAPRICE (CPC) [19]. The functions of some of these genes have been demonstrated via heterologous expression in Arabidopsis thaliana. For example, the heterologous expression of LbTTG1 or LbSAD2 increased trichome development and salt resistance in Arabidopsis [21,22], whereas expression of LbTRY increased root hair development and salt sensitivity [23]. These ndings suggest that salt glands and trichomes could be homologous organs arising from the same ancestor.
These transcriptome studies also uncovered another group of candidate genes of uncharacterized/unknown function that are highly expressed during different stages of salt gland development. Given that no plants with publicly available genome sequences have salt glands, these unannotated genes are thought to be unique to salt glands and may play important roles in salt gland development.
In the current study, we investigated the role of Lb1G04899, a gene of unknown function that is highly expressed in L. bicolor during early salt gland development, as determined by transcriptome analysis [19].
No other family members of this gene were detected, and it only produces one type of transcript. Since Lb1G04899 encodes a protein with a typical helix-loop-helix domain, we named this gene LbHLH. Arabidopsis plants heterologously expressing this gene showed increased trichome development, reduced root hair development, and enhanced salt tolerance. LbHLH might interact with AtGL1, as revealed in a yeast two-hybrid assay. We propose a possible mechanism for the roles of this previously uncharacterized gene in the differentiation of both trichomes and root hairs as well as in salt resistance.

Plant materials and growth conditions
Limonium bicolor seeds were obtained from plants grown in saline soil (N37°20'; E118°36') in the Yellow River Delta (Shandong Province, China) with the permission of the Dongying government. The author Baoshan Wang had formally identi ed Limonium bicolor, and the seeds harvesting process is in full compliance with relevant government guidelines. Unfortunately, we were unable to nd a voucher specimen of Limonium bicolor stored in any publicly available herbarium. The dried seeds were stored at 4°C for at least six months [24]. Before use, the seeds were surface-disinfected with 70% ethanol (5 min),  [25]. To facilitate infection and transformation by Agrobacterium tumefaciens, seedlings were cultivated for one week on 1/2MS medium and transplanted into pots (9 cm height × 9 cm diameter) lled with nutrient-rich soil (soil:vermiculite:perlite, 3:1:1).

Cloning and bioinformatic analysis of LbHLH
The rst true leaves of L. bicolor were collected at different stages of leaf development, including the undifferentiated stage (stage A; 4-5 days after sowing, 5000 leaves), salt gland differentiation stage (stage B; 6-7 days after sowing, 4000 leaves), stomata differentiation stage (stage C; 8-10 days after sowing, 3000 leaves), pavement cell differentiation stage (stage D; 11-16 days after sowing, 1000 leaves), and mature stage (stage E; more than 17 days after sowing, 1000 leaves) according to Yuan [19].
The total RNA was extracted from pooled leaves of each stage using a FastPure Plant Total RNA Isolation kit (RC401-01; Vazyme Biotech Co., Ltd.). cDNA was reverse transcribed from the RNA with a ReverTra Ace quantitative PCR (qPCR) RT kit (TOYOBO Co., Ltd, Japan) according to the manufacturer's instructions. The reference sequence of LbHLH was obtained from the assembled sequence from a previously reported transcriptome [19]. Full-length LbHLH was cloned using the primers LbHLH-S and LbHLH-A (Table S1), which were designed with Primer Premier 5.0. The conserved domain of LbHLH was predicted using the online tool SMART (http://smart.embl-heidelberg.de/).

Subcellular localization of LbHLH
The subcellular localization of LbHLH was determined using transformed onion epidermal cells harboring a GFP expression vector [26]. The pCAMBIA1300 vector was digested with SalI to form a linear vector. To obtain LbHLH cDNA, primers LbHLH OE-S and LbHLH OE-A were designed with SalI digestion sites (Table  S1). The full-length coding sequence (CDS) of LbHLH carrying a SalI digestion site was introduced into the pCAMBIA1300 vector under the control of the CaMV 35S promoter by homologous recombination using a ClonExpress II One Step Cloning Kit (Vazyme Biotech Co., Ltd., China). Agrobacterium tumefaciens GV3101 was used to transform the pCAMBIA1300-LbHLH recombinant vector into onion epidermal cells [27]. After two days of cultivation in the light, uorescent signals of GFP-labeled LbHLH were detected under a TCS S8 MP two-photon laser-scanning confocal microscope (Leica, Germany). DAPI was used to locate the nucleus and was observed under excitation at 358 nm [28]. FM4 − 64 (N-(3-Triethylammoniumpropyl)-4-(6-(4-(Diethylamino)Phenyl) Hexatrienyl) Pyridinium Dibromide, Invitrogen) was used to locate the plasma membrane and observed under excitation at 559 nm [29].
Expression analysis and in situ hybridization of L. bicolor A-E stage leaves, stems, roots, and aged leaves of L. bicolor grown on MS medium were collected for RNA extraction. Seedlings grown under different treatments (100 mM NaCl, 0.04 mg/L 6-BA and 0.1 mg/L ABA added in MS medium) were also collected for RNA extraction. Quantitative RT-PCR primers LbHLH-RT-S and LbHLH-RT-A were designed using Beacon Designer software (version 7.8) (Table S1). RT-PCR was performed in a 20 µl reaction system including 10 µl SYBR qPCR Master Mix (Vazyme Biotech Co., Ltd.), 0.2 µM primers, and 300 ng cDNA in a uorometric thermal cycler (Bio-Rad CFX96 ™ Realtime PCR System) under the following conditions: 95°C for 30 s, 40 cycles (95°C for 5 s, 60°C for 30 s). Lbtubulin (primers Lbtubulin-RT-S and Lbtubulin-RT-A, Table S1) was used as an internal control [18]. The expression level of LbHLH in different tissues was calculated relative to the expression level in roots (which was set to 1). Three biological replicates (separate experiments) were performed. Relative expression levels were calculated using the formula 2 −ΔΔC(T) .
To further explore the expression patterns of LbHLH in L. bicolor, developing leaves (the rst true leaf at 5-8 days of germination) were isolated from L. bicolor for in situ hybridization. Brie y, the leaves were xed in 4% paraformaldehyde, embedded in para n, and dehydrated through an alcohol series. Thin sections (8 µm) of tissue were treated with proteinase K and hybridized in 6 ng/µL hybridization solution at 37°C overnight. Digoxin-labeled LbHLH probe (5'-DIG-CUCCCUAACAUUAACCUUCAGAUCCAGCCC-3', puri ed by HPLC) appeared blue-violet.

Cloning of the LbHLH promoter and histochemical analysis
Genomic DNA was extracted from L. bicolor using a FastPure Plant DNA Isolation Mini Kit (Vazyme Biotech Co., Ltd.) to obtain the full-length LbHLH promoter. The reference sequence of the LbHLH promoter was obtained from the L. bicolor genome (unpublished). The promoter sequence was cloned using primers LbHLH-P-S and LbHLH-P-A (Table S1). Elements in the promoter were predicted using PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/), and maps were drawn using CSDS 2.0 ( http://gsds.gao-lab.org/).
To replace the 35S promoter with the LbHLH promoter in pCAMBIA3301, HindIII and NcoI were used to excise the CaMV 35S promoter from pCAMBIA3301 and to obtain a linear vector. The promoter was cloned into the vector using primers 3301-LbHLH-P-S and 3301-LbHLH-P-A (Table S1) to add HindIII and NcoI digestion sites in advance. The linear vector pCAMBIA3301 and the inserted LbHLH promoter were ligated together using an In-Fusion HD Cloning Kit (Takara) to construct the recombinant vector.
Agrobacterium tumefaciens GV3101 cells were transformed with the recombinant plasmid and used to infect Arabidopsis thaliana Col-0 to generate Col::pLbHLH-GUS. The transgenic seedlings were continuously screened with herbicides (0.1%, v/v), and homozygous plants of the T3 generation were subjected to histochemical staining. Ten-day-old seedlings were immersed in GUS staining solution and incubated at 37°C overnight with shaking. The stained plant materials were decolorized by incubating in 70% ethanol 2-3 times and observed under a dissecting microscope (Nikon, Japan) [30].

Generation of Col-35 S::LbHLH plants
The full-length CDS of LbHLH was cloned using primers LbHLH OEAt-S and LbHLH OEAt-A (Table S1) containing NcoI digestion sites for cloning into different vectors. A ClonExpress II One Step Cloning Kit (Vazyme Biotech Co., Ltd.) was used to generate p35S::LbHLH via homologous recombination. p35S::LbHLH was introduced into Agrobacterium tumefaciens GV3101 cells, which were used to transform Arabidopsis. After screening for three generations with herbicides (0.1%, v/v), the Col-35S::LbHLH overexpression lines were subjected to physiological measurements.
Three Col-35S::LbHLH overexpression lines were selected for physiological characterization based on LbHLH expression (low, medium, and high). Speci cally, positive transgenic plants were rst identi ed using primers LbHLH OEAt-S and LbHLH OEAt-A based on the genomic sequences of the transgenic lines. mRNA was then extracted from different Col-35S::LbHLH lines using a FastPure Plant Total RNA Isolation kit (Vazyme Biotech Co., Ltd.) according to the manufacturer's instructions. LbHLH expression levels in different Col-35S::LbHLH lines were analyzed by qRT-PCR using primers LbHLH RT-S and LbHLH RT-A (Table S1). Given that no homologs of LbHLH were detected in Arabidopsis, the line with the lowest LbHLH expression level (OE35) was used as a control (relative expression level set to 1) to calculate the expression levels of LbHLH in the Col-35S::LbHLH lines. Three biological replicates were performed for each group. Lines with high (OE40), medium (OE26), and low expression (OE4) levels were retained for analysis.
Phenotypic observation and expression analysis of trichome/root hair-related genes in Col-35S::LbHLH Trichome and root hair development were measured in three overexpression lines (OE4, OE26, and OE40) and the wild type. The trichomes on the rst true leaves of one-week-old seedlings were counted under a dissecting microscope (Nikon, Japan). The root hairs 0.5 cm-1.5 cm from the tip of the roots of 5-day-old seedlings were counted. Trichome and root hair numbers were calculated with ImageJ software. Twenty seedlings were examined per line.
RNA was extracted from seedlings grown on 1/2MS medium for one week. The expression levels of ten genes involved in trichome differentiation and root hair fate determination were identi ed by qRT-PCR, including AtTTG1, AtTRY, AtCPC, AtEGL3, AtGL1, AtGL3, AtSAD2, GLABRA 2 (AtGL2), MYB DOMAIN PROTEIN 23 (AtMYB23), and ZINC FINGER PROTEIN 5 (AtZFP5). The expression levels of genes related to root hair development, including the root hair initiation genes ROOT HAIR DEFECTIVE 6 (AtRHD6) and RING FINGER OF SEED LONGEVITY 1 (AtRSL1) and the root hair elongation gene LJRHL1-LIKE 1 (AtLRL1) were also measured by qRT-PCR. All primers used in qRT-PCR are listed in Table S1 as gene name-RT-Sense (gene name-RT-S) and gene name-RT-Antisense (gene name-RT-A) (e.g., AtEGL3-RT-S and AtEGL3-RT-A). AtActin (ampli ed with primers Atactin-RT-S and Atactin-RT-A) was used as an internal control [23]; three replicate biological experiments were performed.
Yeast two-hybrid assay to examine self-activation of LbHLH and identify candidate LbHLH-interacting proteins The full-length CDS of LbHLH was cloned into pGBKT7 (BD) to generate BD-LbHLH using an NdeI digestion site by homologous recombination using a ClonExpress® II One Step Cloning Kit (Vazyme Biotech Co., Ltd.) and the primers BD-LbHLH-S and BD-LbHLH-A (Table S1). The same method was used to construct AD-AtGL1 and AD-AtGL3 using pGADT7 (AD) and primers AD-AtGL1-S, AD-AtGL1-A, AD-AtGL3-S, and AD-AtGL3-A with NdeI digestion sites (Table S1).  (Table S1) to analyze the expression pattern of LbHLH over a time course of 100 mM NaCl treatment.
To investigate salt resistance among different transgenic lines, OE4, OE26, OE40, and wild-type (WT) seeds were germinated on 1/2MS medium (containing 1% agar) with different concentrations of NaCl (0, 50, 100, or 150 mM). The germinated seeds were counted each day for ve days: a seed containing a radicle > 1 mm long that had emerged from the seed coat was considered to be germinated. The germination percentage (%) was calculated as the number of germinated seeds / total number of seeds × 100%. Thirty seeds were sown per line for each treatment, and three biological replicates were performed.
The emergence of green cotyledons was used as an indicator of cotyledon growth. The cotyledon growth rate of each line was measured after three days of germination. Cotyledon growth rate (%) = (number of seeds with cotyledons / number of all tested seeds) × 100%. After continuous cultivation for ve days on different media, the root lengths of different lines were measured using ImageJ software. Thirty replicates were performed per treatment.
Five-day-old seedlings on 1/2MS medium were transplanted into a matrix irrigated with different concentrations of NaCl (0 or 100 mM NaCl dissolved in Hoagland solution, pH 6.2). The leaves (0.5 g) of two-week-old seedlings subjected to control or 100 mM NaCl conditions were harvested separately. The Na + , K + , proline, and MDA contents were measured as described previously [30,31]. Ion concentrations were measured with a ame photometer (Cole-Parmer, USA). Five replicates per measurement were performed for each line.
To verify the effect of LbHLH on alleviating salt stress, all lines were cultured in 180 mM mannitol (causing the same osmotic pressure as 100 mM NaCl) and on 10 mM LiCl medium (causing the same ionic effect as 100 mM NaCl). After ve days of culture, the germination rate and root length were measured to compare the effects of ionic stress and osmotic stress on LbHLH expression.

qRT-PCR of marker genes related to salt stress in transgenic Arabidopsis
To investigate the expression of genes under salt stress, all lines were cultured for ~ 10 days in 1/2MS medium containing 0 or 100 mM NaCl and subjected to RNA extraction using a FastPure Plant Total RNA Isolation kit (RC401-01; Vazyme Biotech Co., Ltd.). The RNA was reverse-transcribed into cDNA and used for qRT-PCR.

Statistical analysis
Statistical signi cance at P = 0.05 (Duncan's multiple range tests) was determined using SPSS. ANOVA with orthogonal contrasts and mean comparison procedures was used to detect signi cant differences between treatments.

Bioinformatics analysis and in situ hybridization of LbHLH
We cloned LbHLH based on its full-length sequence in the transcriptome data. LbHLH contains a 2,067bp open reading frame and encodes a 688-amino-acid protein ( Figure S1A). LbHLH harbors a typical helix-loop-helix (HLH) domain between amino acids 480 and 526 and three low-complexity regions ( Figure S1B). No gene sharing more than 30% similarity with LbHLH was detected by NCBI-BLAST analysis ( Figure S1C).
To determine the subcellular localization of LbHLH, we transformed onion epidermal cells with Agrobacterium tumefaciens carrying p35S::LbHLH-GFP. As shown in Fig. 1A, compared to the positions of DAPI and FM4-64 staining, LbHLH-GFP was located only in the nucleus, while unfused GFP was located in the nucleus and plasma membrane. We also analyzed the expression pattern of LbHLH in L. bicolor at different developmental stages and under different treatments (Fig. 1B). LbHLH was expressed at the highest level in stage A leaves and at the lowest level in roots. LbHLH expression was highly induced by 6-BA treatment.
Given that LbHLH was highly expressed during early salt gland development, we performed in situ hybridization to determine whether LbHLH localizes to salt glands in L. bicolor. Hybridization signals were detected in salt glands (Fig. 1C), suggesting that LbHLH functions in salt gland development and differentiation.

Analysis of the LbHLH promoter and histochemical localization of LbHLH
We identi ed the 2,055 bp promoter sequence of LbHLH based on its sequence in the L. bicolor genome. As shown in Fig. 2A, the LbHLH promoter is enriched in typical TATA box and CAAT box elements and harbors various stress-responsive elements such as ABRE and ARE. The identi cation of MYB binding sites suggests that LbHLH might be regulated by MYB-type transcription factors. To verify the site of LbHLH expression, we generated pLbHLH::GUS and transformed Arabidopsis with this construct. Analysis of GUS staining patterns revealed that LbHLH is expressed in leaf veins in Arabidopsis (Fig. 2B).

LbHLH participates in trichome and root hair development in Arabidopsis
To explore the role of LbHLH in trichome and root hair formation, we heterologously expressed LbHLH in Arabidopsis ecotype Col-0. Eight lines harboring Col-35S::LbHLH (Fig. 3A) were identi ed and their gene expression levels analyzed by qRT-PCR (Fig. 3B). Lines OE4, OE26, and OE40, with low, medium, and high LbHLH expression levels, respectively, were selected for phenotypic observation.
We compared the number of trichomes on the rst true leaves of wild type (WT), OE4, OE26, and OE40, and found that the Col-35S::LbHLH overexpression lines contained signi cantly more trichomes than the WT (Fig. 4A). In addition, the expression level of LbHLH had a dosage effect on the number of trichomes (Fig. 4B), that is, the higher the expression level of LbHLH, the more trichomes that were produced. These results indicate that LbHLH promotes trichome development.
The same genes are involved in the initiation of trichome and root hair development, but they play opposite roles in these processes. Therefore, we counted the root hairs in each line. The Col-35S::LbHLH overexpression lines produced fewer roots hairs than the WT (Fig. 5A), and this phenotype also showed a dosage effect, as the number of root hairs decreased with increasing LbHLH expression (Fig. 5B). These results indicate that LbHLH has an inhibitory effect on root hair development.
LbHLH interacts with AtGL1 to interfere with root hair development Given that the heterologous expression of LbHLH increased the number of trichomes and reduced the number of root hairs in Arabidopsis, we analyzed the expression levels of genes related to trichome and root hair initiation and development in these lines, including AtTTG1, AtTRY, AtCPC, AtEGL3, AtGL1, AtGL3, AtSAD2, AtGL2, AtMYB23, AtLRL1, AtRHD6, AtRSL1, and AtZFP5 (Fig. 6). Most genes were expressed at the highest levels in OE40, whereas no signi cant difference in expression was detected between the WT and OE4. Among these genes, AtGL1 and AtGL3 were the most highly induced in OE26 and OE40 vs. the WT. Therefore, LbHLH most likely interacts with AtGL1 or AtGL3 to in uence trichome and root hair development.
To test for interactions between LbHLH and AtGL1 or LbHLH and AtGL3 in vitro, we performed a yeast two-hybrid assay (Fig. 7A). All colonies grew normally on SD/-Leu/-Trp medium, indicating that the yeast transformation was successful (Fig. 7B). When grown on SD/-Leu/-Trp/X-a-gal/Aba and SD/-Ade/-His/-Leu/-Trp/X-a-gal/Aba media, only the positive control and BD-LbHLH&AD-AtGL1 turned blue and showed normal growth (Fig. 7C), whereas BD-LbHLH&AD-AtGL3 produced no signal. No self-activation of LbHLH was detected. These results indicate that LbHLH strongly interacts with AtGL1, an MYB-like protein required for root hair development, which could explain why root hair formation was signi cantly inhibited in the transgenic lines.
LbHLH is induced by salt stress in L. bicolor and signi cantly increases the salt resistance of Arabidopsis during germination Since the heterologous expression of LbHLH reduced the number of root hairs in Arabidopsis, we wondered whether this phenotype is related to salt tolerance. We therefore examined the expression pattern of LbHLH in L. bicolor over a time course of a 100 mM NaCl treatment (Fig. 8A). Compared to the level of LbHLH expression the start of the NaCl treatment, expression was signi cantly increased after 12 h of NaCl treatment but returned to a lower level after 24 h. These results indicate that LbHLH expression is regulated by NaCl, perhaps due to the osmotic effects of NaCl.
We next examined the salt tolerance of the transgenic Arabidopsis lines during germination and early plant growth by sowing seeds of each line on medium containing a gradient of NaCl concentrations. As expected, the transgenic lines showed better growth than the wild type under high-salt conditions (Fig. 8B), and OE40 (with the highest LbHLH expression level) grew the best. Compared with WT, all three Col-35S::LbHLH lines showed a higher germination percentage at 48 h after sowing (Fig. 8C), especially under the 100 mM NaCl treatment, where the germination rates of the OE lines were more than twice that of the WT. The effect was even more apparent under the 150 mM NaCl treatment, as only a few OE26 seeds germinated, compared to none for the other lines. After 5 days of culture, there were no obvious differences among the lines under the 0 and 50 mM NaCl treatments; however, under the 100 and 150 mM NaCl treatments, the germination percentage was signi cantly higher in the transgenic lines than in the WT (Fig. 8C).
The growth rates of cotyledons showed a similar trend (Fig. 8C). Under the 50 and 100 mM NaCl treatments, cotyledon growth was much better in the Col-35S::LbHLH lines than in the WT. Root growth was inhibited with increasing NaCl concentration, but at each NaCl concentration, the roots of the Col-35S::LbHLH lines were longer than WT roots.
To determine the possible reason for the enhanced salt resistance of the transgenic lines, we measured various physiological indicators in plants under the 0 and 100 mM NaCl treatments (Fig. 9). OE40 accumulated the least amount of Na + and the most proline and K + under 100 mM NaCl conditions, while the opposite results were obtained for the WT. MDA contents were lower in the transgenic lines than in the WT under NaCl treatment, indicating that the overexpression lines suffered less injury than the WT under salt treatment.

LbHLH enhances salt tolerance by alleviating osmotic stress
To further explore how LbHLH improves salt tolerance, we cultured the three transgenic Arabidopsis lines and the WT in medium containing 180 mM mannitol, which has the same osmotic potential as 100 mM NaCl, and in medium containing 10 mM LiCl, which induces the same ionic stress as 100 mM NaCl (Fig. 10A). Under the 10 mM LiCl treatment, all lines showed similar levels of growth inhibition, and no growth advantage was detected in the transgenic lines. However, under isotonic mannitol treatment, the overexpression lines showed the same trends in growth as they did under 100 mM NaCl treatment, i.e., OE40 had the highest germination rate (Fig. 10B) and the longest root (Fig. 10C). These results indicate that WT plants suffered both osmotic and ionic stress under NaCl treatment, whereas the heterologous expression of LbHLH signi cantly improved resistance to osmotic stress, allowing the transgenic lines to perform better than WT at the germination stage.
Finally, to explore why transformation with LbHLH signi cantly improved salt tolerance in Arabidopsis at the molecular level, we examined the expression of marker genes in these plants under salt stress.
AtP5CS1 and AtP5CS2 were expressed at much higher levels in the Col-35S::LbHLH lines than in the WT, while the expression of AtSOS1 and AtSOS3 declined in the transgenic lines (Fig. 11); these expression levels corresponded to the levels of proline accumulation (Fig. 9) and osmotic stress resistance in these plants (Fig. 10).

Discussion
L. bicolor is a typical recretohalophyte with salt glands. Increasing numbers of genes in L. bicolor have been shown to participate in salt resistance, but all of these genes are homologs of genes of known function, such as LbTTG1, LbTRY, and LbSAD2. No genome sequences of plant species with salt glands are currently available. Therefore, identifying the activities of genes of unknown function that are expressed during salt gland development may be crucial for understanding salt gland development and salt resistance. Here, we demonstrated that LbHLH, which was annotated to be a protein of unknown function, induces the expression of genes related to trichome and root hair development by interacting with AtGL1 in Arabidopsis. Furthermore, LbHLH improves salt tolerance, primarily by relieving osmotic stress due to high NaCl levels.
Bioinformatic analysis showed that LbHLH contains an HLH (helix-loop-helix) domain. HLH domains are primarily detected in transcription factors such as OrbHLH2O in Arabidopsis and SbHLH148 in rice [32,33]. Surprisingly, in the current study, we determined that LbHLH has no transcriptional activation activity ( Fig. 7). In situ hybridization showed that LbHLH is expressed in the salt glands of L. bicolor. Therefore, we suggest that LbHLH interacts with other transcription factors or functional proteins in the halophyte L. bicolor to regulate salt gland development. In addition, the heterologous expression of LbHLH led to an increase in trichome formation in Arabidopsis. Based on previously reported transcriptomes of L. bicolor at different stages of development , salt gland and trichome development might involve homologous genes with similar developmental patterns of expression. These ndings strongly suggest that LbHLH is directly related to salt gland development, and they explain the why of heterologous expression of LbHLH affects trichome development in Arabidopsis.
The interaction between LbHLH and AtGL1 was detected in vitro in a yeast two-hybrid assay. AtGL1 positively regulates trichome development in Arabidopsis [34,35], and the gl1 mutant lacks trichomes on its leaf surfaces. These ndings suggest that LbHLH regulates trichome and root hair development by directly interacting with AtGL1. Interestingly, this gene has opposite effects on the development of aboveground trichomes and underground root hairs; that is, LbHLH promotes trichome development and inhibits root hair growth via the same interaction with AtGL1. Why LbHLH plays opposite roles in trichome and root hair development is still unknown. Nevertheless, other genes have been shown to play opposite roles in trichome and root hair initiation; the underlying mechanism would be worth investigating in the future.
It is easy to see why reduced root hair development would affect salt tolerance because root hairs directly participate in the absorption of ions. As expected, the Col-35S::LbHLH lines showed much better germination than the WT. The reduced number of root hairs makes the Col-35S::LbHLH lines absorb less Na + , suffer less ionic stress, and accumulate less MDA than the WT, phenotypes that were also observed in Arabidopsis expressing LbTTG1 [22]. Moreover, a typical dosage effect was observed in the transgenic lines: the stronger the expression of LbHLH, the greater the salt tolerance of the lines. Mannitol and LiCl are widely used to simulate osmotic and ionic stress, respectively, and the effects of these substances can easily be determined based on phenotype [30]. The Col-35S::LbHLH lines showed enhanced tolerance to mannitol but not LiCl, indicating that the increased tolerance to NaCl stress in the transgenic lines was due to increased tolerance of osmotic stress. In L. bicolor, LbHLH expression was highly induced after 12 h of NaCl treatment, suggesting that this gene responds to short-term osmotic stress and may contribute to salt tolerance in this halophyte.
The current study examined the function of LbHLH mainly by heterologous expression of the gene in the model plant Arabidopsis. However, in situ hybridization and expression analysis in L. bicolor during development strongly suggested that LbHLH functions in salt gland development. Given that a transformation system has been developed for L. bicolor [18], it should be possible to use CRISPRmediated gene editing to further investigate the function of LbHLH in L. bicolor and its role in salt gland development. Here, we demonstrated that the function of an unknown gene in L. bicolor could be successfully studied, laying the foundation for studying the roles of salt glands in salt resistance and the utilization of saline soils in the future.

Conclusion
LbHLH of L. bicolor, which was annotated to be a protein of unknown function positioned in salt gland by in situ hybridization, had highest expression at early leaf development stage. Overexpression of LbHLH induces the expression of genes related to trichome and root hair development by interacting with AtGL1 in Arabidopsis. Furthermore, LbHLH improves salt tolerance, primarily by relieving osmotic stress due to high NaCl levels.

Declarations
Ethics approval and consent to participate All experiments in the manuscript were performed at Shandong Provincial Key Laboratory of Plant Stress, and they are in compliance with relevant laws in China.

Consent for publication
Not applicable.
Availability of data and material The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
This work was supported by the National Natural Science Research Foundation of China (NSFC, project nos. 31600200 and 31770288).
Authors' contributions F.Y. designed the research; X.W. and Y.X. performed the research; X.W. and Y.Z. analyzed the data; X.W. wrote the paper; F.Y. and B.W. revised the paper. All authors have read and approved the manuscript.