Ammopiptanthus nanus seeds were collected from Wuqia County (39°72′N, 75°26′E), Xinjiang Uygur Autonomous Region, China. Seeds were sown in a Murashige and Skoog (MS) solid medium and germinated in an artificial growth chamber under the following conditions: 16-h light/8-h dark at 25°C. Abiotic stress was applied to 20-day-old seedlings of different periods (0, 6, 12, 18, 24, and 48 h) with 4°C, 20% PEG 8000, and 0.25 M NaCl as our previous study (Liu et al., 2019). In addition, the seedlings were treated with different concentrations of hormones (ABA, 2 μM; IAA, 1 μM; NAA, 2 μM; MeJA, 10 μM, ETH, 50 μM; and GA₃, 50 μM), according to the previous study (Liu et al., 2019). The different tissues of roots, stem, and leaves were then frozen with liquid nitrogen and conserved at −80°C for RNA isolation.

The wild-type (WT) Arabidopsis thaliana (Arabidopsis) seeds used for transformation in this study were ecotype “Columbia” (Col-0). The seedlings were grown in a standard growth chamber (60% humidity, and 120 μE m^–2 s^–1 at 22°C) with a 16-h/8-h light/dark cycle.

For real-time quantitative PCR analysis, three A. nanus tissues, such as roots, stems, and leave materials, were collected for total RNA extraction with a TRIzol reagent (Invitrogen, Carlsbad, CA, United States) as described by Liu et al. (2019). Moloney Murine Leukemia Virus (M-MLV) reverse transcriptase (Promega, Madison, WI, United States) was used for first-strand cDNA synthesis. The genomic DNA was isolated from the young leaf (0.4 cm length) of A. nanus with a Plant Genomic DNA kit (Tiangen, Beijing, China). Total RNA and DNA quality was determined using a BioDrop μLITE+ (Harvard Bioscience Shanghai).

Gene-specific primers for AnDHN were designed with primer-blast¹ using partial sequence fragment that was obtained from transcriptome sequencing results. The partial cDNA sequence was isolated with a SMART RACE cDNA Amplification Kit (Clontech, San Jose, CA, United States). The cDNA pools for 3′ and 5′ RACE were generated with the total RNA extracted from leaves of A. nanus. The product of reverse transcription was used for outer PCR with the 5′ and 3′ RACE outer primers (gene-specific primers, GST1 and GST4, respectively), and the first-round products were further used for the inner PCR (gene-specific primers, GST2, GST3, and GST5). The detail sequences of outer and inner primers are described in Supplementary Table 4. The final PCR products were purified and cloned into pGEM-T vector (Takara, Dalian, China). The positive clones were extracted as recombinant plasmids. More than three independent positive clones were sequenced by Sangon Biotech (Shanghai, China). The full-length cDNA of AnDHN was obtained by aligning and assembling to the 5′ and 3′ sequences with an overlapping fragment using Vector NTI Advance^® 11.5 software.

For the phylogenetic tree of DHNs, we obtained multi-species DHN protein sequences from Phytozome 13² (Goodstein et al., 2012) and aligned with Clustal X version v2.1 with default parameters (Larkin et al., 2007). A maximum-likelihood (ML) phylogenetic tree was constructed with IQ-Tree v1.6 using JTT + F + G4 model as suggested by IQ-Tree model test tool (BIC criterion) with 1,000 times of bootstrap replicates, edited and visualized using iTQL³. The phylogenetic tree was carried out using a common taxonomy tree application of NCBI⁴, visualized using MEGA5 software (Tamura et al., 2011) and manually optimized for viewing according to recent studies (Puttick et al., 2018).

Real-time-qPCR and data analysis were conducted as described previously by Sun et al. (2020). Gene-specific primers were designed using primer-blast (see text footnote 1) with AnACTIN (GenBank: KJ873129.1) and AtACTIN as reference. Tissues from five seedlings with the same treatment were pooled for RNA extraction as one biological replicate. Three biological replicates were included for each treatment for one independent experiment, and each sample was amplified in three parallel reactions as technical replicates. PCR was carried out using Real Master Mix (SYBR Green) (Tiangen, Beijing, China) on QuantStudio 7 Flex (Applied Biosystems, Waltham, MA, United States) with 96-well format. The relative expression level was determined according to Sun et al. (2020). The 2^–ΔΔCt methods were used to calculate the relative expression level of AnDHN (Yin et al., 2020).

The AnDHN coding sequence (CDS) was fused with green fluorescent protein (GFP) driven by CaMV 35S promoter. The recombinational fragment was inserted into the pCAMBIA1302 vector for subcellular localization. After sequenced, the 35S: AnDHN-GFP constructions were transformed into Agrobacterium tumefaciens GV3101 strain. Agrobacterium tumefaciens transformant strains were grown overnight at 28°C in 20 ml yeast extract mannitol broth (YEB) medium plus selective antibiotics, collected by centrifugation, and resuspended in infiltration medium (2 M MgCl₂, 0.2 M MES, and 100 mM acetyleugenone, PH = 5.7) until the optical density (OD) value was 0.5–1.0. About 5-week-old Nicotiana benthamiana leaves were infiltrated with the bacterial cell suspensions and the plasma marker PAD62. GFP fluorescence signals were observed and documented under a confocal laser-scanning microscope (Olympus Fluoview Ver. 2.0c Viewer) after 48 h in the dark. This experiment was repeated for three biologicals, and each assay was performed with three N. benthamiana leaves.

The AnDHN promoter was isolated from A. nanus genomic DNA using genome walking as described by Liu et al. (2019). For PCR reaction, the AnDHN gene-specific primers were designed according to the sequence of AnDHN full-length cDNA. Putative functional cis-acting elements (CRE) of the AnDHN promoter were identified by the PlantCARE database⁵, and the transcription factor-binding sites (TFBSs) were predicted by JASPAR⁶ (Wasserman and Sandelin, 2004).

The CDS was driven by the CaMV 35S promoter. The recombinational fragment was inserted into the pCAMBIA3301 vector for genetic transformation. Arabidopsis transformation was performed by the floral dip procedure (Clough and Bent, 1998). The seeds were collected from the infiltrated plants and selected on a half-strength MS medium containing 50 μg/ml of hygromycin. Hygromycin-resistant plants were transferred to soil 10 days after germination. The progeny of three lines with 3:1 segregation was further treated with hygromycin reagents to screen for the homozygous single insertion. Finally, 10 independent overexpressed T₃ lines were obtained. Three representative T₃ independent lines (AnDHN#1, AnDHN#2, and AnDHN#3) were used for subsequent scoring phenotypes.

For seeds germination assay, seeds of WT and AnDHN transgenic lines were germinated under vernalization conditions (4°C) for 3 days. Then, seeds were on a half-strength MS solid medium containing different concentrations of mannitol (0, 200, 300, and 400 mM) and NaCl (0, 100, 150, and 200 mM). To assess cold tolerance, seeds were sown in a normal half-strength MS solid medium and vernalize treatment for 2 days. Then, the samples were transferred to a low-temperature artificial growth chamber at 4°C for 4 days. And then the seeds were transferred to another growth chamber at 22°C for 5 days. Germination rates were counted for different abiotic stress for three biological replicated assays. For every biological replicate, we tested at least 30 seeds of each line from the same batch three times as one technical replicate.

For the root growth experiment, seeds of WT and AnDHN transgenic lines were transferred to a constant temperature incubator at 22°C for 4 days after vernalization. Then, these seeds further on a half-strength MS solid medium containing mannitol (0, 300 mM) and NaCl (0, 150 mM) were grown in a chamber with a condition (16/8 h d/night at 22°C) for 7 days. Meanwhile, another part of the seeds was also sown on a normal half-strength MS solid medium has grown in a chamber with a condition (16/8 h d/night at 4°C) for 7 days. The root length of seedings of WT and AnDHN transgenic lines was collected and measured. Average values were calculated from three biological replicates. And for every biological replicate, we tested six seedlings from the same batch.

Ion leakage (IL) was according to Sakuraba et al. (2014) with minor modifications. The leaves were placed in a tube with 25 ml of double-distilled water for 2 h, and then the initial conductivity (C1) was detected. After heated in boiling water for 15 min and cooled until to room temperature, the final electrolyte conductivity (C2) was detected. Hence, the relative IL (%) = C1/C2 × 100% was obtained. At least three independent biological replicates were analyzed for each line, and no less than 10 seedlings were collected as one sample for each biological replicate.

Relative water content was calculated according to the previous study described in Vysotskaya et al. (2010) with minor modifications. Fresh weight (FW) was measured, and the leaves were in double-distilled water for 4 h in darkness at room temperature. The turgid weight (TW) further was recorded. For dry weight (DW), the leaves were dried at 80°C for 24 h. Relative water content (RWC) equation, RWC (%) = [(FW–DW)/(TW–DW)] × 100%. Three biological replicates were used for each line, and at least 10 seedlings were collected as one sample detected for each biological replicate.

Malondialdehyde (MDA) content was detected using a maleic dialdehyde assay kit (A003-3, Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Three biological replicates for each sample and at least 10 seedlings were collected as one sample for each biological replicate.

The ROS scavenging enzymes activities of PRO content, catalase (CAT) activity, superoxide dismutase (SOD) activity, peroxidase (POD) activity, and glutathione (GSH) content were detected with kits also produced by Nanjing Jiancheng Bioengineering Institute (Nanjing, China). Three biological replicates for each sample and at least 10 seedlings were collected as one sample as one biological replicate.

For histochemical staining assays, the leaves were collected from 2 to 3-week-old seeding of WT and AnDHN transgenic lines treated with mannitol (300 mM), NaCl (150 mM) and 4°C. The histochemical staining assays were conducted with 0.2% nitro-blue tetrazolium (NBT) and 1 mg/ml 3.3′-diaminobenzidine (DAB) solutions with vacuum infiltration for three times. The leaves were kept for 12 h in NBT and DAB solutions. Then, the leaves were decolorized by boiling in 75% ethanol. Images were visualized using stereomicroscopy (SZX16, Olympus).

These experiments were repeated at least three times. All of the statistical analyses were performed using IBM SPSS statistics 19. Significant variation was estimated with Student’s t-test, *p < 0.05; ^**p < 0.01; ^***p < 0.001. The data were presented as the mean ± SD of three independent experiments.

To identify the key regulator of drought resistance in A. nanus, a significantly upregulated candidate gene (c195333_g1_i1) (NCBI GenBank: AFH89648) was obtained from an RNA-seq library prepared from drought-induced A. nanus seedlings (Figure 1A and Supplementary Table 1). Lacking genomic information, the full-length sequence of this gene was isolated using 5′ and 3′ RACE. The gene-specific primers were designed from the partial sequence obtained from the RNA-seq data for degenerate RT-PCR (Figure 1C). As a result, the final 5′- and 3′-fragments were amplified through three and two rounds, respectively (Figure 1B and Supplementary Figure 1A), and the final PCR products were sequenced. Both the final 5′- and 3′- fragments (540 and 345 bp, respectively) were aligned, and the coding sequence was determined to be 543 bp in length, encoding a putative protein of 181 amino acids (Figure 1D). The final obtained 5′ UTR and 3′ UTR sequences were 73 and 386 bp, respectively. And there is no intron and only one exon exists (Figure 1C). The protein had a predicted molecular weight of 18,125.87 MW and a theoretical pI of 5.74. Using the online servers NCBI-CDD and SMART, two conserved motifs, called Y- (DEYGNPV) and K- (KKGIMNKIKEKLPGY) segments, were identified (Figure 1D), which might play key roles in AnDHN protein. These results demonstrate that c195333_g1_i1 is a new drought-responsive gene from A. nanus.

Because of its adaptation to extreme environmental conditions, A. nanus is an excellent model species in which to study the drought and cold tolerance in legumes and other species. To further understand the evolutionary history of c195333_g1_i1, we constructed a phylogenetic tree. The ML phylogenetic tree was derived from an alignment of 189 DHN sequences from 23 species of both monocots and eudicots (Figure 2A and Supplementary Figure 2). Based on our results, the gene from A. nanus encoding a protein could be orthologous of the proteins encoded by AT3G50980 (XERO1) and AT5G66400 (RAB18) (Figures 2A,B and Supplementary Figure 2). XERO1 and RAB18 are known as DHN proteins from Arabidopsis. Among them, RAB18 is the protein extensively studied, which is widely reported involved in cold tolerance and drought tolerance of plants (Lång and Palva, 1992; Zou et al., 2021).

Previous studies have focused only on the structural types of DHN proteins, with little consideration given to the evolutionary relationships between the DHN family members. Our results show that the DHN family is widely distributed in angiosperms (Figure 2A). Among these, all of the DHN family members from legume species were clustered in subfamilies I, III, V, and VIII, with most found in subfamily VIII (Figures 2A,B). We identified AnDHN in A. nanus as being a protein closely related to Lup004542 and Ca_12999 in legumes (Figure 2B). A multiple sequence alignment showed that AnDHN harbors conserved K- and Y-segment domains similar to AT3G50980, AT5G66400, Lup004542, and Ca_12999 (Supplementary Figure 3), consistent with our bioinformatic prediction (Figure 1D). The sequences of predicted DHN proteins from Zea mays, Sorghum bicolor, Setaria italica, Brachypodium sylvaticum, Oryza sativa, A. nanus, Lupinus angustifolius, Vigna angularis, Pisum sativum, Medicago truncatula, Cicer arietinum, Arachis hypogaea, Malus domestica, and Arabidopsis were further used in phylogenetic reconstruction (Figure 2C). Interestingly, all of the monocot species had more than six DHN proteins, and the genomes of the eudicot species Arabidopsis and apple contain at least 10 DHN proteins of different coding for each. However, the number of DHN genes in legume species ranged from three to six, except for A. nanus, which lacks genomic information. Our findings suggest that DHNs are evolutionarily conserved in multiple species, which is consistent with previously reported (Close, 1997; Svensson et al., 2002; Rorat et al., 2006).

We next conducted a comprehensive set of experiments to functionally analyze AnDHN for its potential roles in simulated drought stress. The temporal expression patterns of AnDHN were analyzed in roots, stems, and leaves from 0 to 48 h of PEG 8000 treatment. Our results further confirmed that AnDHN is a drought-responsive gene (Figure 3A). Our data show that AnDHN expression in roots was significantly upregulated after 6 h of drought treatment and that expression peaked at 24 h of PEG 8000 treatment. Unlike roots, AnDHN expression was increased by 2.8-fold and 3.1-fold at 48 h compared to the 0 h control treatment in stems and leaves, respectively (Figure 3A). To determine why AnDHN expression responds to drought stress, we obtained the AnDHN promoter region, which was 830 bp in length, by genomic walking (Supplementary Table 2). Several types of drought-induced cis-acting elements were identified from the PlantCARE and JASPAR databases as expected (Supplementary Figure 4 and Supplementary Table 3). Unexpectedly, we also identified numerous cold and salt-induced elements in the promoter of AnDHN (Supplementary Figure 4 and Supplementary Table 3). To test our predictions, the expression pattern of AnDHN was analyzed after following cold and salt stress treatments. AnDHN expression was markedly upregulated at 12 h in the roots, 24 h in stems, and within 24 h in leaves exposed to salt stress (Supplementary Figure 5). In response to cold (4°C) treatments, AnDHN expression was dramatically different from the 0-h control at most times during the 48-h experiment in the roots, stems, and leaves (Supplementary Figure 5).

To determine the subcellular localization of the AnDHN protein, the AnDHN-GFP fusion was transiently expressed in N. benthamiana leaf epidermal cells. The result showed that GFP fluorescence overlaps with the plasma membrane (PM) marker, indicating the AnDHN protein is localized at the PM (Figure 3B).

To further verify whether AnDHN is related to drought stress, AnDHN was driven by the 35S promoter, and 10 overexpressions of AnDHN transgenic lines were generated in Arabidopsis. Among these lines, we chose three representative independent lines for further analysis (abbreviated as AnDHN#1, AnDHN#2, and AnDHN#3), and AnDHN had a relatively high expression level in these lines (Supplementary Figure 6). Germination rates in the transgenic plants showed no differences from the WT plants under normal conditions (Figures 4A,B). However, we observed higher germination rates in seeds from transgenic lines compared to seeds from WT plants at three different concentrations of mannitol (Figures 4A,B). As expected, the germination rate of WT seeds was decreased in the presence of mannitol compared to that of WT seeds without mannitol (Figures 4A,B). This may indicate that the germination rate was related to drought stress.

Under drought stress, a deep root system architecture is beneficial and allows plants to adapt to adverse living conditions (Wei et al., 2020; Yang et al., 2020; Zhou et al., 2020). To confirm this, we compared the root length of WT and overexpressed plants under normal conditions and 300 mM mannitol treatment. As the result, an insignificant difference can be visualized between WT plants and overexpressing plants grown under normal conditions except for AnDHN#1, and there is a significant difference between the WT plants and AnDHN#1 (Figures 5A,B). However, we found that all three transgenic lines displayed markedly increased root length when grown on a medium containing 300 mM mannitol, especially for AnDHN#1 (Figures 5C,D). Thus, our data suggest that overexpression of AnDHN enhances drought tolerance by increasing the germination rate and the deep root system architecture.

Under drought stress, plants that are capable of physiological change will be able to adapt to changeable environments. We therefore investigated two physiological indexes, IL and relative water content (RWC), to investigate water retention in the plants. We found a slight decline in electrical conductivity under drought stress (Figure 6A). However, all three independent AnDHN-OX transgenic lines retained much more water than did the WT plants under drought stress (Figure 6B). Our results suggest that overexpression of AnDHN in Arabidopsis can enhance drought tolerance by increasing water retention.

Proline is a proteinogenic amino acid that contributes to osmotic adjustment in drought tolerance. Our assays confirmed that, when treated with 300 mM mannitol, proline accumulated to higher levels in the AnDHN-OX transgenic plants than in WT plants, indicating that AnDHN-OX plants have increased drought tolerance (Figure 6C). We next measured the MDA contents, because MDA is an important indicator of damage of cellular membranes (lipid peroxidation) caused by abiotic stresses. The MDA contents in AnDHN-OX lines were significantly lower than in WT plants under drought treatment (Figure 6D) showing that increased expression of AnDHN relieves damage to the cellular membranes.

Excess ROS (H₂O₂ and O₂^–) production can damage cellular membranes (Feng et al., 2020a), and the ROS levels are strictly controlled by antioxidant enzymes. Therefore, were also determined the activities of POD, CAT, SOD, and GSH that are responsible for ROS scavenging. As shown in Figures 6E–H, the activities of these indicator enzymes, except for SOD, were increased in the transgenic lines in response to drought stress (Figures 6E–H). Nitro blue tetrazolium (NBT) and DAB staining were performed to detect H₂O₂ and O₂^–, respectively. We observed only slight staining in the leaves of the WT and AnDHN-OX plants grown under normal conditions, with no visible differences between them (Figures 7A,B). Under drought stress, however, weaker staining was found in the leaves of the AnDHN-OX lines compared with the WT plants (Figure 7). These results indicate that AnDHN-OX plants had increased ROS scavenging capability in response to drought stress.

In our previous study, we found that AnDHN expression could also be induced by both salt and cold stress (Supplementary Figure 5). Moreover, the promoter of AnDHN contains salt and cold stress-induced cis-acting elements (Supplementary Figure 4 and Supplementary Table 3). We next performed experiments to determine whether AnDHN affects salt and cold stress by assaying seed germination rate, root length, IL content, RWC, and the activities of crucial antioxidant enzymes. As expected, our results indeed confirmed our hypothesis that AnDHN simultaneously improves plant tolerance to both salt and cold stress. The AnDHN-OX lines had relatively higher germination rates and longer root lengths in comparison to the WT under salt and cold stress conditions (Figures 4, 5). Furthermore, the RWC and proline contents and POD and GSH activities were significantly higher in AnDHN-OX line plants under salt stress (Figures 6B,C,E–H). The concentrations of IL, proline, and the POD, CAT, and SOD activities in the AnDHN-OX lines were higher than in the controls in response to cold stress (Figures 6A,C,E–G). In addition, there was less DAB and NBT staining in the overexpression lines relative to that in the WT plants (Figure 7). These results show that AnDHN also has the potential to enhance salt and cold tolerance in plants.

Although DHNs are predicted to be stress-related proteins (Nylander et al., 2001; Kosová et al., 2007; Peng et al., 2008; Kim and Nam, 2010; Shekhawat et al., 2011), the study of DHN functions in many species is still in its infancy. Previous studies have been focused on analyses of the putative promoters of DHN genes (Zhu et al., 2014; Liu et al., 2015; Aguayo et al., 2016). All of these studies showed that the expression of DHN genes is upregulated in response to multiple hormones and abiotic stresses (Zhu et al., 2014; Liu et al., 2015; Aguayo et al., 2016). Our results show that AnDHN positively improves drought tolerance through morphological and physiological changes in A. nanus.

Abscisic acid can improve stress tolerance in plants by allowing them to adapt to changeable environments (Zhu, 2016). Previous studies have shown that ABA regulates the expression of many downstream genes in response to drought stress with morphological and physiological changes, such as deep root system architecture, stomatal closure, and enhanced ROS scavenging (Ullah et al., 2017; Xing et al., 2019; Gupta et al., 2020). Also, ABA-dependent signaling pathways play a critical role in the response to drought stress (Ullah et al., 2017). Our data demonstrate that AnDHN is a membrane protein (Figure 3B), suggesting that AnDHN could be an upstream gene involved in drought tolerance. Moreover, previous studies in rice showed that overexpression of OsDhn-Rab16D modulated the expression patterns of ABA signaling genes to alter endogenous ABA concentrations (Tiwari et al., 2019). Thus, we can also hypothesize that AnDHN indirectly changes the expression of ABA biosynthesis genes or ABA signaling genes to alter physiological responses and the expression of numerous downstream tolerance genes under drought stress in A. nanus. To test this hypothesis, we also detected the expression of ABA biosynthetic and catabolic genes in the WT plants and AnDHN overexpressed lines. The results showed that the expression levels of AtNECD3 and AtNECD5 (two ABA biosynthetic genes) were significantly upregulated, while the expression levels of AtCYP707A1 and AtCYP707A3 (two ABA catabolic genes) were downregulated (Supplementary Figure 7). Interestingly, our results also show that AnDHN expression is induced by ABA (Supplementary Figures 4, 8 and Supplementary Table 3), indicating that there might be an AnDHN-ABA loop to improve drought tolerance in A. nanus.

Previous studies have shown that DHN proteins are localized in the cytosol, nucleus, mitochondria, vacuole, and vicinity of the PM (Houde et al., 1995; Danyluk et al., 1998; Rorat et al., 2004). The localization of the DHNs in cells may depend on the existence or deficiency of the Y-, S-, and K- segments (Graether and Boddington, 2014). YnKn, YSK, and Kn types of DHN proteins have been found to localize to the nucleus and cytoplasm (Wisniewski et al., 1999; Lin et al., 2012). SKn type of DHNs is found to localize in the vicinity of the PM (Danyluk et al., 1998; Hara et al., 2003). Y-segment has shown no connection with the localization of DHN proteins. In our research, we found that the AnDHN protein was localized at the PM (Figure 3B). According to the role in this study and the previous reports (Hara et al., 2003; Yang et al., 2014), we guessed that it could play a key role in protecting lipid membranes against peroxidation under drought and freezing stress.

Dehydrin proteins are reported to be involved in responses to various abiotic stresses in plants (Hundertmark and Hincha, 2008; Xu et al., 2014; Zhu et al., 2014). In our study, we also showed that overexpression of AnDHN in Arabidopsis mediates the effects of salt and cold stress (Figures 4–7 and Supplementary Figure 5), in addition to its effect on drought stress. Some hormones play critical roles in abiotic stress tolerance, and various genes that are involved in regulating abiotic stress tolerance are induced by multiple hormones (Zhu et al., 2014; Liu et al., 2019; Xing et al., 2019). Changes in AnDHN expression following treatment with ABA, Indole-3-Acetic acid (IAA), 1-Naphthyl acetic acid (NAA), Jasmine (keto) acid methyl ester (MeJA), ethylene (ETH), and GA₃ are shown in Supplementary Figure 8, consistent with previous results in A. nanus, Cucumis melo var. makuwa Makino, and Triticum aestivum (Zhu et al., 2014; Liu et al., 2019; Xing et al., 2019). This suggests that AnDHN may play an active role in abiotic stress responses in A. nanus. Cis-acting elements in promoters that affect gene expression in response to abiotic stress have been widely studied (Yamaguchi-Shinozaki and Shinozaki, 2005; Zhu et al., 2014; Liu et al., 2019). We identified many hormone-inducible elements in the AnDHN promoter that respond to MeJA, IAA, and ABA in addition to drought, cold, and salt (Supplementary Figure 4 and Supplementary Table 3). Moreover, the promoter harbored a lot of TFBSs of abiotic stress and hormones, simultaneously (Supplementary Table 3). Thus, our results imply that AnDHN expression depends upon the presence of these promoter elements to respond to abiotic stresses.

To the best of our knowledge, the DHN family has multiple members in most species; for example, there are four and 10 DHN genes in Eucalyptus globulus and Arabidopsis, respectively (Fernández et al., 2012; Aguayo et al., 2016). Phylogenetic analysis of DHN proteins from dicots and monocots showed that there are at least three members in each of the species examined, although only a single gene was identified in A. nanus (Figure 2 and Supplementary Figure 2). Furthermore, AnDHN was isolated from a drought-induced RNA-seq library of A. nanus. Hence, we speculate that there may be more than one DHN gene in A. nanus, which would be consistent with other species. DHN acts as the pivotal regulator in the response to various abiotic stresses in Eriobotrya japonica, Eucalyptus globulus, Oryza sativa, and Arabidopsis (Fernández et al., 2012; Kumar et al., 2014; Xu et al., 2014; Aguayo et al., 2016). We further demonstrated that AnDHN positively regulates drought stress tolerance and tolerance to salt and cold (Figures 3–7). Phylogenetic analysis and sequence alignment revealed that DHN proteins are highly conserved in angiosperms (Figure 2 and Supplementary Figure 3). Taken together, these results indicate that DHNs appear to define a class of proteins that regulate abiotic stress tolerance in all angiosperms.

Apart from the function of stress, LEA family proteins may also play key roles in the growth and development of plants. Unfortunately, research on LEA function is limited by their proteins with structural flexibility and lacking similarity with other known proteins. It will be a hotpot to explore the association between their structural types with different levels of water deficiency based on the previous studies of LEA proteins (Battaglia and Covarrubias, 2013). Moreover, numerous LEAs are detected in developmental root hairs, suggesting that they have the potential for water absorption and nitrogen fixation in symbiotic association rhizobium-legumes (Battaglia and Covarrubias, 2013). The research on the function of LEA proteins is not enough except for the stress response, and it will be worthwhile to investigate in the future.

Ammopiptanthus nanus is a non-model plant that grows in arid desert habitats in central Asia. Plants can experience conditions, such as very low rainfall, extremely high evaporation, and abnormally high and low temperatures (Liu et al., 2016, 2019). Because it is able to survive in such extreme environments, A. nanus has great research potential for elucidating drought and cold tolerance mechanisms that could be applied to other species. Our findings shed light on the control of drought resistance by DHN in A. nanus. However, the mechanisms underlying drought resistance are still largely unknown in A. nanus and also in other legume species.

In the future, reduced crop yields and quality loss due to drought stress will be more serious (Ullah et al., 2017). Hence the drought tolerance mechanism remaining needs to gain better exploitation. (1) From both model and non-model species, more genes responsible for drought tolerance need to be identified and characterized by genetic variation, transcriptomic, gene editing, and so on. Abiotic stress is a complicated trait. As drought often couples with others stress, pleiotropic genes are excellent. (2) Stomatal closure often with carbohydrate synthesis is reduced, how to coordinate with growth and drought stress is important. (3) Research studies of plant responses to drought are mostly studied from plants grown in the laboratory greenhouse, not natural. We need to accelerate the application of our results into natural water-deficit conditions for agriculture.