Introduction

Drought is the most common abiotic stress that inhibits plant growth and development and poses a serious threat to crop production sustainability1,2. Every year, numerous cases of reduced crop productivity due to droughts result in substantial economic losses, presenting a significant challenge for agricultural production3,4. Furthermore, with global climate change and the continued rise in greenhouse gas concentrations, droughts have become more frequent, severe, and prolonged than in recent decades5,6,7. Enhancing plant resilience is imperative for agricultural production and environmental sustainability8. Therefore, gaining a comprehensive understanding of the molecular mechanisms governing plant responses and adaptations to drought stress and breeding superior varieties with high drought resistance are promising approaches for ensuring global food security and promoting ecological restoration worldwide9,10,11.

In general, virtually every aspect of plant physiology, biochemistry, and cellular metabolism is affected by drought stress12,13,14,15,16. Drought results in hyperosmotic stress in plant cells, inhibiting root system water absorption, leading to abnormal growth, an imbalance in water relations, and decreased water-use efficiency (WUE)4,8,17,18,19. In addition, drought disrupts ionic homeostasis, osmotic balance, the cell membrane system, and photosynthesis13,14,20,21. Drought also leads to a reduction in stomatal conductance, leaf CO2 assimilation rate, and enzyme activity13,15,20. Moreover, the dry environment induced the production of reactive oxygen species (ROS) in biochemical reactions, including superoxide anion (\({\text{O}}_{2}^{{\cdot}-}\)), hydrogen peroxide (H2O2), hydroxyl radical (HO·), and singlet oxygen (1O2)13,22. ROS production and consumption are strictly controlled by the antioxidant systems15,18,20. Under drought stress, the activities of many oxidative protective enzymes, such as superoxide dismutase (SOD), catalase (CAT), ascorbate peroxidase (APX) and glutathione reductase (GR), are altered3. Ascorbate/Ascorbic acid (ASA) and glutathione (GSH) are crucial for maintaining ROS levels within plant cells, as they prevent oxidative damage by scavenging these free radicals13,23. Dehydroascorbate (DHA) is the oxidized form of ASA and can be converted back into ASA through reduction processes in plant cells, thereby maintaining the levels of ASA in the plant24. During the process of plant adaptation to drought stress, the ASA/DHA system protects cellular structures and maintains normal physiological functions through its antioxidant capabilities25. However, as the drought progressively worsens, ROS production overwhelms the scavenging function of the antioxidant system, leading to severe cell damage and death26.

H2O2 has been identified as a key signaling molecule that regulates stomatal closure in plants27,28. Stomatal closure is the first defense mechanism by which plants respond to drought by preventing transpiration loss, but at the cost of reducing photosynthetic rates13. This reduction in photosynthesis can have significant consequences on plant growth under drought conditions. In addition to the ROS pathway, plants regulate stomatal closure through various pathways, including the abscisic acid (ABA) and Ca2+ signaling pathways. ABA-mediated stomatal closure is a crucial drought resistance mechanism in plants29. Moreover, the cuticle, which is composed of insoluble cutin and waxes, is another factor that limits transpiration under drought stress and serves as an important barrier for plants against various biotic and abiotic stresses30,31. Cutin plays a vital role in the prevention of moisture loss32. Leaf epidermal wax increases as soil moisture decreases, and the wax layer reduces stomatal conductance, transpiration, and photosynthesis, thereby facilitating plant adaptation to drought conditions33. Nonetheless, a recent study in Populus revealed no alteration in the overall amount of leaf waxes, but rather an elevation in their transpiration rate due to modifications in their composition34.

In addition to ROS scavenging and stomatal closure, plants undergo significant alterations in physiological and biochemical metabolism, and regulate root growth and morphology as adaptive responses to drought environments. These adaptations are intricately regulated by multiple signal networks, including signal perception, signal transduction, transcription, translation, and post-translational modifications2,35,36,37. It has been identified that several key components within these signaling networks play pivotal roles in regulating drought responses in various plant species, such as Arabidopsis, rice, and maize38,39,40. These components include receptors, phosphatases, kinases, and a wide array of transcription factors (TFs). These TFs encompass those involved in phytohormone signaling, such as PYR/PYL/RCAR (pyrabactin resistance 1/PYR1-like/regulatory component of ABA receptors), PP2C (clade A type 2C protein phosphatases), SnRK2s/SAPKs (SNF1/AMPK-related protein kinases), calcium-dependent protein kinases (CDPKs), as well as abscisic acid responsive elements binding factor (ABF), abscisic acid responsive elements (ABRE), myeloblastosis (MYB), WRKY, DREB, NAC, and BHLH TFs8,36,41,42,43,44,45. However, our knowledge of the genome-wide molecular mechanisms underlying drought tolerance in plants, especially in gramineous crops and forage grasses, remains limited.

Many native grass species grown in arid and semi-arid areas possess special stress tolerance mechanisms and abundant gene resources. Research on these grasses can potentially significantly contribute to the genetic improvement of crop stress resistance, thereby benefiting agriculture and animal husbandry. Siberian wildrye (Elymus sibiricus), a typical perennial meso-xerophytic forage grass widely distributed at high altitudes worldwide, offers good forage quality and environmental adaptability. It is extensively used in natural grasslands, cultivated pastures, and the ecological restoration of degraded grasslands46,47,48,49. Furthermore, E. sibiricus exhibits remarkable resistance to barren conditions, drought, cold, salinity, and diseases48,50. However, despite its impressive resilience, the mechanisms underlying its drought tolerance remain poorly understood.

Drought stress leads to water deficit in plants, while osmotic stress leads to water absorption difficulties, Therefore, many studies use osmotic stress to simulate drought stress. Polyethylene Glycol 6000 (PEG-6000), a nonionic osmotic agent, effectively simulates drought conditions by reducing water potential and is a valuable tool for studying plant responses to drought51. Drought, resulting from insufficient precipitation or soil moisture under natural conditions, leads to reduced water availability for plants. Similarly, PEG-6000-induced osmotic stress decreases soil water potential, limiting plant water absorption52. While these two phenomena differ in mechanism, physiological impact, adaptation strategies, and research applications, both focus on how plants adapt to water scarcity53. In our previous study, we screened drought-tolerant (DT, I-1-1-46) and drought-sensitive (DS, I-1-1-2) genotypes of E. sibiricus, and revealed that the Casparian strip and suberin lamellae in the root endodermis played a pivotal role in the difference in drought resistance between the DT and DS genotypes54. Here, we performed a comparative analysis of the physiological, biochemical and transcriptomic changes in the leaves of DT and DS genotypes in response to drought stress to clarify two major issues: (1) What are the differences in the physiological responses to drought stress in the leaves of different genotypes? (2) What is the molecular basis underlying the distinct drought tolerances of the two genotypes? The results of this study are expected to shed new light on the understanding of plant drought resistance, offering potential pathways for the development of more drought-tolerant forage grass and crop cultivars through molecular breeding methods.

Results

Effects of drought on leaf water status in DT and DS

Under the control condition (75% field water content [FWC]), there was no significant differences in leaf water potential (Ψw), osmotic potential (Ψs) and turgor pressure (Ψt) between DT and DS (Table 1). With increasing drought intensity, both DT and DS leaves exhibited significant decreases in Ψw and Ψs. Under mild (60% FWC) and moderate (45% FWC) drought stress, although there was no significant difference in Ψw between the two genotypes, Ψs of DT was significantly lower than that of DS (Table 1). Under severe drought stress (30% FWC), Ψw and Ψs of DT leaves were significantly lower than those of DS leaves (Table 1). With the exacerbation of drought stress, Ψt in both genotypes initially increased and then decreased, and it was significantly lower in DT than in DS under severe drought stress (Table 1). These results indicate that E. sibiricus can absorb water from drought soils by reducing the osmotic potential and, consequently, water potential, with DT exhibiting a more advantageous capacity in this regard.

Table 1 Effects of drought on leaf water status of two E. sibiricus genotypes.
Full size table

Effects of drought stress on photosynthesis of two E. sibiricus genotypes

Under the control conditions, there were no significant differences in net photosynthetic rate (Pn), stomatal conductance (Gs), and water use efficiency (WUE) between DT and DS, whereas the transpiration rate (Tr), intercellular carbon dioxide concentration (Ci), and limiting value of stomata (Ls) in DT were significantly higher than those in DS (Fig. 1). With decreasing soil water content, the Ci values of both DT and DS gradually increased, while the other five photosynthetic parameters showed a decreasing trend. The increase in Ci was as pronounced as the reduction in Tr and Ls, with Pn exhibiting the largest magnitude of change (Fig. 1). Overall, under drought stress, DT consistently exhibited higher values for all six photosynthetic indices compared to DS. The most pronounced differences were observed under severe drought stress, where DT had a 46, 30, 13, 17, and 19% higher Pn, Gs, Ci, WUE, and Ls than DS. Notably, at this point, the Tr of the DT leaves was was not significantly higher than that of the DS leaves (Fig. 1).

Fig. 1
figure 1

Responses of leaf photosynthetic characteristics in two E. sibiricus genotypes to drought stress. (A) Net photosynthetic rate (Pn); (B) transpiration rate (Tr); (C) stomatal conductance (Gs); (D) intercellular carbon dioxide concentration (Ci); (E) water use efficiency (WUE); (F) limiting value of stomata (Ls). The data are presented as the means of three replicates (± SD) based on Tukey's post hoc test. Means denoted with the different letters indicate significant differences at P < 0.05. The horizontal coordinates indicate different levels of field water content (FWC). DT drought-tolerant genotype, DS drought-sensitive genotype.

Full size image

Differences in leaf ROS accumulation and ascorbate–glutathione cycling in drought responses

The superoxide anion radical (\({\text{O}}_{2}^{\cdot -}\)) content did not significantly differ between DT and DS in either the control or mild to moderate drought stress (≥ 45% FWC). However, under severe drought stress, the \({\text{O}}_{2}^{\cdot -}\) content in DS was notably higher (38%) than that in DT (Fig. 2A). Hydrogen peroxide (H2O2) levels in both DT and DS leaves increased with decreasing soil water content. Under mild drought stress (60% FWC), there was no significant difference in H2O2 levels between DT and DS; however, under other drought conditions, H2O2 levels in DT were significantly lower than those in DS (Fig. 2B). GSH and ASA are important antioxidants involved in ROS scavenging. There was little difference in GSH content between DT and DS across the 30–75% drought gradient (Fig. 2C). However, ASA content increased with increasing drought intensity and was significantly higher in DT than in DS under mild to severe drought stress (Fig. 2D).

Fig. 2
figure 2

Responses of leaf ascorbic acid-glutathione cycle in two E. sibiricus genotypes to drought stress. (A) Superoxide anion radical (\({\text{O}}_{2}^{\cdot -}\)); (B) hydrogen peroxide (H2O2); (C) glutathione (GSH); (D) ascorbic acid (ASA); (E) mono-dehydroascorbic acid reductase (MDHAR); (F) dehydroascorbate reductase (DHAR); (G) ascorbate peroxidase (APX); (H) ascorbic acid/dehydroascorbic acid (ASA/DHA). The data are presented as the means of three replicates (± SD) based on Tukey's post hoc test. Means denoted with the different letters indicate significant differences at P < 0.05. The horizontal coordinates indicate different levels of field water content (FWC). DT drought-tolerant genotype, DS drought-sensitive genotype.

Full size image

Therefore, we analyzed ASA-related indices, including mono-dehydroascorbic acid reductase (MDHAR), dehydroascorbate reductase (DHAR), APX, and ASA/DHA. The results showed that the MDHAR, DHAR contents, and ASA/DHA continuously increased with increasing drought intensity (Fig. 2E,F,H), whereas the APX content initially decreased drastically and then slightly increased (Fig. 2G). In the control treatment (75% FWC), the contents of MDHAR, DHAR, and ASA/DHA were not significantly different between DT and DS, whereas the APX content was significantly higher in DT than in DS (Fig. 2E–H). The MDHAR content was significantly higher in DT than in DS under severe drought stress, but there was no significant difference between them in the remaining treatments (Fig. 2E). Under mild drought stress (60% FWC), the DHAR content was not significantly different between DT and DS, while under moderate (45% FWC) and severe (30% FWC) drought stress, the DHAR content in DT was significantly higher than in DS (Fig. 2F). The APX content and ASA/DHA of DT were significantly higher than those of DS under all drought treatments (Fig. 2G,H). These findings highlight the importance of ASA metabolism in the response of DT to drought stress.

Transcriptome analysis of DT and DS in response to osmotic stress

Libraries were constructed and sequenced on an Illumina HiSeq™ 4000 deep sequencing platform, and detailed RNA sequencing results are shown in Supplementary Table S1. Clean reads were assembled to obtain 136,699 unigenes; detailed information on the evaluation indicators can be found in Supplementary Table S2. These unigenes were annotated in four protein databases: 80,610 in Nr, 67,301 in KEGG, 31,763 in COG, and 39,932 in SwissProt (Supplementary Table S3 and Supplementary Fig. S1). The RNA-seq data were credible and reproducible using PCA and Pearson’s correlation coefficient analyses (Supplementary Figs. S2, and S3).

Volcano plots identified significantly upregulated or downregulated genes during osmotic stress at each time point (Fig. 3A–D). In DT, there were 3002 differentially expressed genes (DEGs) (1664 upregulated and 1338 downregulated) and 2464 DEGs (986 upregulated and 1478 downregulated) were observed at 3 and 24 h, respectively, compared with the control (Fig. 3E). In DS, there were 1555 DEGs (481 upregulated and 1074 down-regulated) at 3 h and 1691 DEGs (873 upregulated and 818 down-regulated) at 24 h compared with the control (Fig. 3E). Venn diagrams showed in Supplementary Fig. S4 illustrated the number of DEGs that were common and unique to two genotypes at the same treatment time (Supplementary Fig. S4). At 3 h, 2572 and 1125 DEGs were uniquely expressed in DT and DS, respectively, when compared to the control, 430 DEGs were common expressed in both DT and DS (Supplementary Fig. S4A). At 24 h, 1735 and 962 DEGs were uniquely expressed in DT and DS, respectively, when compared to the control, 729 DEGs were common expressed in both DT and DS (Supplementary Fig. S4B). Of which 197 DEGs were co-expressed at two treatment times (Supplementary Fig. S4C). Another common form of analysis prioritizes the number of common and uniquely expressed DEGs in the same genotype at different treatment times (Fig. 3F–G), 2234 and 1696 DEGs were unique in DT at 3 and 24 h, respectively, compared to the control (Fig. 3F), whereas 900 and 1036 DEGs were uniquely expressed in DS at 3 and 24 h, respectively, compared with the control (Fig. 3G). In addition, 768 and 655 DEGs were common at both time points for both DT and DS, respectively (Fig. 3F,G), indicating that a higher number of unigenes were regulated in DT than in DS. Further analysis revealed 197 DEGs were co-expressed in DS and DT (Fig. 3H).

Fig. 3
figure 3

Analysis of differentially expressed genes in E. sibiricus genotypes in response to osmotic stress. (A–D) Volcano plots of all expressed genes in drought-tolerant (DT) and drought-sensitive (DS) genotypes at different time points compared to the control (CK), with log2FC values drawn against -log10 (FDR) adjusted P-values. Orange and blue dots represent upregulated and downregulated differentially expressed genes (DEGs) based on |log2(foldchange)|≥ 2 and FDR ≤ 0.001; green dots represent genes that are not differentially expressed; (E) number of upregulated and downregulated DEGs identified under different osmotic-stress treatments compared to the control DT and DS. Orange indicates upregulated DEGs, while blue indicates downregulated DEGs; (F) common and unique DEGs under 3 and 24 h osmotic-stress treatments in DT; (G) common and unique DEGs under 3 and 24 h osmotic-stress treatments in DS; (H) common and unique DEGs under both 3 and 24 h osmotic-stress treatments in DT and DS.

Full size image

GO enrichment analysis of the upregulated and downregulated genes revealed key GO term associated with osmotic stress. GO enrichment for DEGs was higher in DS than in DT, the number of genes annotated to GO was higher in DS (fg_num = 1380) than in DT (fg_num = 978) (Supplementary Fig. S5 and Supplementary Table S4). DS and DT are mainly enriched in the two GO terms associated with hydrolase activity in molecular function, hydrolase activity, acting on glycosyl bonds (GO:0016798) and hydrolase activity, hydrolyzing O-glycosyl compounds (GO:0004553), respectively (Supplementary Fig. S5 and Supplementary Table S4). Using the KEGG database to investigate the key genes involved in the KEGG pathways, these DEGs were found to be significantly enriched in plant hormone signal transduction (ko04075), biosynthesis of secondary metabolites (ko01110), and the MAPK signaling pathway (ko04016) (Supplementary Fig. S6 and Supplementary Table S5).

Screening of differential candidate genes in DT and DS under osmotic stress

We performed hierarchical clustering for 1226 DEGs (log2FC value) in response to osmotic stress in the two E. sibiricus genotypes (Fig. 3H). The heat map identified two clustering modules for the DEGs. Cluster1 consists of 833 DEGs that exhibited high expression levels in DT and DS under normal conditions, but low expression levels under osmotic stress. In contrast, cluster2 includes 393 DEGs that showed the opposite expression trend compared to cluster1 (Fig. 4A). Further analysis of cluster1 identified five notable trend blocks linked to profiles 3, 10, 2, 6, and 0, with 212, 167, 153, 134, and 95 unigenes, respectively. Additionally, cluster2 exhibited four significant trend blocks attributed to profiles 9, 16, 17, and 13, comprising 118, 115, 55, and 54 unigenes, respectively (Fig. 4B,C).

Fig. 4
figure 4

Clustering and trend analysis of differentially expressed genes (DEGs) in DT and DS genotypes under osmotic stress. (A) Hierarchical clustering analysis indicates DEG expression patterns in DT and DS after 0, 3, and 24 h of osmotic stress treatment; (B,C) trend analysis shows that the lines reflect expression patterns for each DEG at osmotic stress time points. Figures (B) and (C) respectively demonstrate the expression trends of DEGs in Cluster 1 and Cluster 2. The horizontal axes of Figures (B) and (C) are the same as that of Figure (A). Black lines in each trend blocks indicate mean changes in DEG expression. The different profiles represent the numbers of the different trend blocks, and the numbers within the trend blocks indicate the number of enriched genes. Colored trend blocks indicate significant enrichment (p < 0.05), while uncolored trend blocks do not show significant enrichment. Each inflection point represents a set of sample data.

Full size image

To efficiently screen candidate genes associated with DT osmotic tolerance, we focused on profile3 within cluster1 and profile 16 within cluster2. In profile3 of cluster1, 212 DEGs exhibit high expression levels in both DT and DS genotypes in the absence of stress treatment. Expression levels decreased significantly in both DT and DS by osmotic stress, with a faster reduction in DT (from 3 to 24 h) compared to DS (24 h) under treatment. At 24 h, both exhibited similar low expression levels (Figs. 4B and 5A,B). Heatmap analysis showed that the expression of 212 DEGs in profile 3 was lower in the DT genotype compared to the DS genotype under osmotic stress (Fig. 5B). In profile16 of cluster1, 115 DEGs exhibit lower expression levels in both DT and DS genotypes without treatment. However, DT was highly induced compared to DS in the early stages of the 3-h osmotic stress treatment, and both were highly expressed in the 24-h stress treatment (Figs. 4C and 5C). Heatmap analysis showed that the expression of 115 DEGs in profile 16 was higher in the DT genotype than in the DS genotype under osmotic stress (Fig. 5D). Immediately after that, we adopted the following strategy to further screen the candidate unigenes: by analyzing the expression differences of DEGs between two genotypes under osmotic stress treatments and combining the information from related literature, we screened the unigenes that were specifically low-expressed in DT under osmotic stress in profile3 of Cluster1, and also screened the unigenes that were specifically high-expressed in DT under osmotic stress in profile16 of Cluster2. Following this strategy, eight candidate unigenes that were highly expressed specifically in DT under osmotic stress were screened from 327 DEGs: Unigene0005863 (EsSnRK2), Unigene0047636 (EsCER1), Unigene0005889 (EsGOLS1), Unigene0066379 (EsCAD8D), Unigene0005868 (EsSAHH), Unigene0031985 (EsCIPK5), Unigene0041137 (EsAZF2), and Unigene0053902 (EsLRK10) (Fig. 5E). Interestingly, all 8 candidate unigenes are from profile 16 of cluster 2. The expression of these eight candidate genes in DT was significantly upregulated under osmotic stress, whereas they remained unchanged in DS in response to osmotic stress. These candidate unigenes were enriched in cutin, suberin, and wax biosynthesis, galactose metabolism, cysteine and methionine metabolism, MAPK signaling pathway, plant hormone signal transduction, phenylpropanoid biosynthesis, biosynthesis of secondary metabolites and metabolic pathways (Fig. 5F). Among these, Unigene0005863 (EsSnRK2) encodes a key protein kinase in the MAPK and plant hormone (ABA) signal transduction pathways (Supplementary Figs. S8 and S9). In addition, our study found that unigenes associated with ROS scavenging in the ASA-GSH system were differentially expressed in DT and DS (Supplementary Fig. S10).

Fig. 5
figure 5

The trend analysis plots, expression heatmap, and enrichment bubble plots of DEGs in DT and DS genotypes at different time points. (A) Trend analysis graph for DT and DS in profile 3 trend block from cluster1; (B) heatmap of DT and DS expression in the profile 3 trend block; (C) Trend analysis graph for DT and DS in profile 16 trend block from cluster2; (D) heatmap of DT and DS expression in the profile 16 trend block; (E) heatmap expression of candidate genes; (F) enrichment bubble map of candidate genes.

Full size image

Effects of osmotic stress on leaf cuticular wax and cutin monomer contents

Among the DEGs presented in Fig. 5E, we noted that Unigene0047636, a homologous gene of CER1 in Arabidopsis through sequence alignment which may encode an enzyme involved in forming alkane monomers during wax biosynthesis, was significantly induced by osmotic stress in DT, but not in DS (Supplementary Fig. S7). Therefore, we analyzed the cuticular wax and cutin monomer contents. Results showed that there was little difference in the wax and cutin monomer contents between the two genotypes without osmotic stress (Fig. 6A,B). However, osmotic stress (20% PEG-6000) significantly increased the contents of alkanes and alcohols aldehydes in the wax monomers of DT, without increasing any monomers in DS, resulting in the two major wax monomers being higher pronouncedly in DT than in DS under osmotic stress (Fig. 6A). However, only one DEG, Unigene0047636 (CER1), was detected in the wax biosynthesis pathway (Supplementary Fig. S7), suggesting that CER1 may be a key factor in wax biosynthesis in the DT response to osmotic stress. In addition, although osmotic stress slightly increased the contents of some cutin monomers in DT, such as alcohols, ω-OH fatty acids, and polyhydroxy fatty acids (PFAs), which were not induced by osmotic stress in DS, transcriptome sequencing did not detect any DEGs between the two genotypes in the cutin biosynthesis pathway (Fig. 6B, Supplementary Fig. S7).

Fig. 6
figure 6

Difference in leaf cuticle between DT and DS genotypes under osmotic stress. (A) Wax monomers content; (B) Cutin monomers content. Wax and cutin data are presented as the means ± SD based on Tukey's post hoc test. Means denoted with the different letters indicate significant differences at P < 0.05. DT drought-tolerant genotype, DS drought-sensitive genotype.

Full size image

Discussion

The drought tolerance of plants is strongly related to their ability to use’ water17. The difference in the water potential gradient between the inside and outside of plant cells is the key to water transport across the membrane55. Under drought stress, plants can reduce cellular water potential through osmoregulation to maintain the water potential difference between the root system and the soil environment, and to absorb water from dry soils with low water potential56. Under drought stress, DT had a lower leaf Ψw and Ψs than DS (Table 1), which indicated the higher drought resistance of DT. Furthermore, our results showed that the Ψt of DT was significantly lower than that of DS under severe drought stress (Table 1). Cell turgor pressure plays a pivotal role in governing the stomatal aperture, with reduced turgor pressure predominantly leading to stomatal closure57. However, the Gs of DT was significantly higher than that of DS, corresponding to a higher Pn of DT than that of DS (Fig. 1A,C), indicating a higher photosynthetic capacity of DT. In addition, Ci and WUE of DT were higher than those of DS under different drought stress treatments (Fig. 1D,E). The factors causing reduced photosynthesis under drought stress can be divided into two categories: stomatal limitation and non-stomatal limitation; the former is due to a limitation of Gs, causing Ci to fail to meet photosynthetic demand, and the latter is due to reduced chloroplast and photosynthetic enzyme activities58,59. In the present study, the Gs of both genotypes gradually decreased with increasing drought intensity, whereas the Ci gradually increased and Ls decreased (Fig. 1C,D). This result indicated that the stomatal limitation of E. sibiricus gradually decreased with increasing drought intensity, and the factors leading to the inhibition of photosynthesis gradually changed to non-stomatal limitation.

In response to osmotic stress, plants control water balance through stomatal movement60. Plants employ multiple pathways to close stomata under drought stress, with the ABA signaling pathway being one of the most crucial mechanisms29. In addition to the ABA signaling pathway, other signaling pathways, such as ROS, can also regulate stomatal opening and closure during drought stress61. These mechanisms work synergistically to reduce water transpiration and enhance drought tolerance62. Our transcriptome study screened eight DEGs involved in the response of E. sibiricus to osmotic stress, which were significantly upregulated in DT, but not in DS, under osmotic stress (Fig. 5E). Among these eight genes, Unigene0005863 (EsSnRK2), Unigene0053902 (EsLRK10) and Unigene0031985 (EsCIPK5) may be involved in stomatal closure induced by the ABA signaling pathway. SnRK2 is a key protein kinase involved in the ABA signal transduction pathway, which regulates stomatal closure to prevent leaf water loss in Arabidopsis63. The wheat SnRK2s family member TaSnRK2.1 is Involved in stomatal closure, osmotic regulator biosynthesis, cellular ROS scavenging, homeostasis, and other stress-related biological processes during drought stress and ABA induction64. Unigene0053902 (EsLRK10) is homologous to LPK10L1.2 in Arabidopsis. Arabidopsis mutants lacking LPK10L1.2 exhibit ABA-insensitive and drought-sensitive phenotypes65. CdtCIPK5 from Cynodon dactylon (L.), the homologous gene of Unigene0031985 (EsCIPK5), is induced by multiple abiotic stresses such as ABA, dehydration, low temperature, and salt stress66. The high expression of these eight candidate genes in DT under osmotic stress may be the key to its drought resistance, but how they regulate drought resistance at the molecular level in E. sibiricus DT genotypes requires further study.

Stomata and cuticle are the two pathways that regulate transpiration water loss67,68. Plant cuticular waxes are the key components that reduce cuticle water loss through the cuticle69, acting as a crucial barrier to protect plants against environmental stresses30,70. These waxes, located in the leaf epidermis, are believed to contribute to the reduction of stomatal conductance, transpiration rates, and photosynthesis, thereby enhancing drought resistance of plants32,33. However, in the current research, we found that the wax content of DT was significantly higher than that of DS. Its stomatal conductance, transpiration rates, and photosynthetic activity were also significantly higher than those of DS. This may be because, under drought stress, the lower wax content of DS necessitates a drastic reduction in transpiration water loss by decreasing Gs, while DT limits cuticle transpiration due to its higher wax content, which in turn helps to maintain a higher Gs. This promotes CO2 assimilation and balances plant growth under drought stress. In addition, in our previous studies, DT has stronger apoplastic barrieres in the root endodermis, which can more effectively limit the return of water to the medium under drought stress. This is also one of the important strategies for maintaining photosynthesis54.

Waxes are primarily composed of very-long-chain (VLC) aliphatic lipids, such as alkanes, aldehydes, primary and secondary alcohols, ketones, and esters30,70. This protective function is particularly important in xerophytes. Shellakkutti et al.71 discovered that osmotic stress in both cultivated and wild barley led to an increase in wax crystals accumulation, total wax amounts, and wax gene expression, while cuticular conductance remained unaffected. Similarly, a recent study in Populus revealed that alterations in the compositions of wax, despite without altering the total amounts, led to elevated transpiration rates34. Alkanes, the predominant constituents of plant cuticle waxes72, are thought to play a crucial role in conferring drought tolerance73. Grünhofer et al.34 pinpointed a rise in the relative proportion of esters in the Populus × canescens cer6_1,2 mutant as the primary factor to the heightened transpiration rate, in fact, alkanes in the mutant decreased significantly. Notably, our results showed that the content of alkane, alcohol, and aldehyde monomer in DT was significantly higher than that in DS (Fig. 6A), suggesting that DT may enhance drought tolerance by accumulating more wax alkanes.

The wax biosynthetic pathway involves multiple steps and pathways, among which the modification of aldehydes, alkanes, secondary alcohols, and ketones by very-long-chain fatty acids (VLCFAs) through the alkane formation pathway is an extremely critical part74. This process is regulated by multiple genes that encode enzymes and transporters necessary for the deposition of cuticle constituents75. Among these genes, CER1 plays a pivotal role in the drought stress-induced biosynthesis of cuticular wax76,77. In a study on alkane synthesis in wheat epidermal waxes, researchers found that TaCER1-1A, a TaCER1 homolog, increased the alkane contents in both transgenic rice and Arabidopsis, leading to an improved epidermal barrier and enhanced drought tolerance in the leaf epidermis of transgenic lines78. Additionally, TaCER1-6A is a crucial gene in wheat, significantly involed in alkane biosynthesis. Knockout mutants of TaCER1-6A exhibited reduced alkane contents in their leaves, whereas overexpression of TaCER1-6A increased alkane contents, thereby improving drought tolerance in wheat73. Furthermore, in tomato and Poa pratensis, the SlCER1-1 and PpCER1-2, respectively, positively regulate the synthesis of wax alkanes, reducing cuticle permeability and enhancing drought tolerance79,80. Wen et al.81 found that the downregulation of two TaCER1 genes led to decreased long-chain alkanes synthesis in low-wax wheat mutants. Drought stress caused a significant increase in very-long-chain alkanes in the wax components of Brachypodium distachyon. BdCER1-8, homologous gene of CER1, is highly expressed in leaves82. In Arabidopsis, CER1 encodes an alkane-forming enzyme that catalyses the decarbonylation of aliphatic aldehydes to alkanes83. Our results showed that the expression of Unigene0047636, homologous to CER1, was upregulated in DT but remained unchanged in DS under drought stress (Fig. 5E). Harb et al.84 found that in barley, HvCER1 was specifically induced under drought stress in the drought-tolerant genotype, but not in the drought-sensitive genotype, which is consistent with our results. EsCER1 positively regulates the conversion of long-chain aldehyde monomers to long-chain alkane monomers during wax biosynthesis, which consequently leads to a significant increase in the contents of alkanes monomers in wax within DT (Fig. 6A). Furthermore, in the wax biosynthesis pathway, long-chain secondary alcohols, as downstream products of long-chain alkanes (Supplementary Fig. S7), increased correspondingly with the increased contents of long-chain alkanes (Fig. 6A), although no DEGs regulating the synthesis of alcohols from alkanes were detected (Supplementary Fig. S7). Therefore, we speculate that EsCER1 is involved in the alkanes synthesis of the wax biosynthesis pathway and plays a certain regulatory role in the synthesis of alcohols, which may be an important reason for the difference in drought resistance among the different genotypes of E. sibiricus.

The reactive oxide species (ROS) are essential signaling molecules in various physiological processes, playing a crucial role in regulating plant growth and development85. However, excessive ROS accumulation can be toxic to plants and can lead to oxidative damage86,87. To counteract ROS toxicity, plants possess a complex detoxification system, that includes a non-enzymatic ASA-GSH pathway88. The ASA-GSH pathway serves as a vital component in reducing oxidative damage by scavenging ROS through the detoxification mechanisms of ASA, GSH, and their associated enzymes89,90. The ASA-GSH cycle is comprised of four key enzymes: ascorbate peroxidase (APX), mono-dehydroascorbate reductase (MDHAR), dehydroascorbate reductase (DHAR), and glutathione reductase (GR)90. In the ASA-GSH cycle, ASA provides electrons for the APX-catalyzed reaction to complete the effective scavenging of H2O291. In this study, we found that under drought stress, the contents of ASA and APX in DT were significantly higher than those in DS (Fig. 2D,G), but its H2O2 content was significantly lower than that in DS(Fig. 2B), indicating that DT has a stronger H2O2 scavenging capacity. In the ASA-GSH cycle, GSH provides electrons for DHAR and promotes the reduction of DHA to regenerate ASA92. In this study, we found that the DHAR content in DT was significantly higher than that in DS under drought stress (Fig. 2F), suggesting stronger ASA regeneration in DT. Meanwhile, as GSH is oxidized to Glutathione Disulfide (GSSG) during ASA regeneration, the degree of GSH oxidation in DT might be greater than in DS, which could also lead to no significant difference in GSH content between DT and DS (Fig. 2C). The above studies suggest that the greater drought tolerance of DT compared to DS may be due to its ability to maintain lower ROS levels through higher ASA, DHAR, and APX concentrations in the ASA-GSH cycle.

APX1 and APX2 are cytoplasmic enzymes that play crucial roles in ROS scavenging93,94. Previous studies have demonstrated their significance in mitigating the ROS levels in plants. For instance, in a study on ROS accumulation in Arabidopsis, it was found that the overexpression of APX1 and APX2 resulted in a significant decrease in ROS levels95. Research on Stevia rebaudiana has demonstrated that the plant's response to abiotic stress involves the upregulation of key genes encoding ROS-scavenging enzymes such as glutathione S-transferase (GST) and APX, contributing to the modulation of its stress response96. Under osmotic stress, RNA-seq results revealed that genes relevant to the ASA-GSH cycle (APX, GR, DHAR, and GST) were differentially expressed in DT and DS (Supplementary Fig. S10). In addition, the overall expression of Unigene0106502 (EsAPX1), Unigene0005132 (EsAPX2), Unigene0013814 (EsGR1) and Unigene0023285 (EsGST4) were higher in DT than in DS (Supplementary Fig. S10). These findings may suggest a potential role for these unigenes in the regulation of ROS accumulation in DT and DS.

Materials and methods

Plant materials and treatments

Two genotypes (DT and DS) of E. sibiricus were used in this study54. Plant materials used to determine physiological phenotypes were grown in soil, materials for cutin, wax and transcriptome sequencing were cultured by hydroponics.

For soil culture: Seeds were vernalized for 3 days in the dark at 4 °C, then germinated on wet filter paper. After 2 days, the seedlings were transferred to pots (diameter, 10 cm; depth, 8.5 cm; 9 seedlings/pot) containing peat soil, and incubated at 23 and 19 °C during the day and night, respectively, with 16 h of light, and 40–60% relative humidity. A compound fertilizer with a 1:1:1 N:P:K ratio was applied every 2 days. After 10 days, when the plants reached the third-leaf stage, they were divided into four experimental groups: (1) control: watered with distilled water until water content was maintained at 75% of the field water content (FWC) by weight; (2) mild drought treatment: stopped watering until water content of the medium decreased to 60% FWC, and distilled water was used to maintain this content; (3) moderate drought treatment: maintained at 45% FWC; (4) severe drought treatment: maintained at 30% FWC. The FWC was calculated by assessing the moisture level of the peat soil using a cutting ring technique. Correlation indices were measured after 12 days of treatment, and six biological replicates were used for each treatment (three plants/replicate).

In hydroponics, seed germination was the same as that observed in the soil culture experiment. Two days after germination, the seedlings were transferred to hydroponic boxes filled with half-strength Hoagland solution (H353, Phyto TechLabs, Lenexa, KS, USA), which was aerated using an air pump. The environmental conditions in the climatic chamber were consistent with those of the soil culture. Subsequently, 28-day-old seedlings were treated as follows: (1) Seedlings were transferred to half-strength Hoagland solution containing 0 (control) or 20% PEG-6000 (BioFroxx, Einhausen, Germany) for 3 days. Cutin and wax of the leaf tissue were then measured after harvest. Each treatment consisted of six biological replicates. (2) The seedlings were transferred to half-strength Hoagland solution containing 20% PEG-6000 for treatment. Fresh leaves were collected at 0, 3, and 24 h after treatment and immediately frozen in liquid nitrogen, then stored at – 80 °C for transcriptome sequencing. Each treatment had three biological replicates.

Measurement of leaf water potential, osmotic potential and turgor pressure

The leaf water potential (Ψw) was measured using a water potential meter (PSYPRO; DIANJIANG TECH, USA). For the osmotic potential (Ψs), leaf samples (0.5 g) were washed with deionized water, dried with filter paper, and immediately frozen in liquid nitrogen. Subsequently, the samples were placed in a syringe the sap was squeezed out. The sap was then centrifuged, and 50 μL of the supernatant was used to measure the ic value with a freezing point osmometer (OSMOMAT 3000, Gonotec GmbH, Germany). Ψs was calculated according to the formula:

$${\Psi_s} = \, - icRT.$$

where R is the gas constant 0.008314 and T is the measured thermodynamic temperature (298.8 K). The turgor pressure (Ψt) is calculated according to the formula:

$${\Psi_t} = \, {\Psi_w} - {\Psi_s}.$$

Measurement of photosynthetic parameters

Photosynthetic parameters were measured using a portable photosynthesis system instrument (LI-6800, LI-COR Biosciences, USA). To ensure that the plants received sufficient light before measurement. The measurements were conducted between 8:00 and 11:30 a.m. The photosynthetic system parameters were set as described below: leaf chamber area of 2 cm2, photosynthetic effective radiation of 1000 lx, and temperature of 25 °C. The following photosynthetic indicators were measured: the net photosynthetic rate (Pn), stomatal conductance (Gs), transpiration rate (Tr), and intercellular carbon dioxide concentration (Ci). Based on the above measured indices, the water use efficiency was calculated according to the formula: WUE = Pn/Gs97 and the limiting value of stomata (Ls) was calculated using the formula: Ls = (Ca − Ci)/Ca98, where Ca is the atmospheric carbon dioxide concentration.

Determination of ROS and antioxidant contents

Plant leaves (0.1 g) were used to measure the content of ROS (\({\text{O}}_{2}^{\cdot -}\) and H2O2) in the control and drought treatment groups using \({\text{O}}_{2}^{\cdot -}\) and H2O2 kits, respectively. Ascorbic acid (ASA), glutathione (GSH) and dehydroascorbic acid (DHA) levels were determined using ASA, GSH and DHA kits, respectively. The activity levels of circulatory metabolism-related enzymes, including dehydroascorbate reductase (DHAR), monodehydroascorbic acid reductase (MDHAR) and ascorbate peroxidase (APX), were measured using DHAR, MDHAR and APX kits, respectively. All the kits were obtained from Suzhou Comin Biotechnology Co., Ltd. (Suzhou, China). The measurement of these indices were performed according to the manufacturer’s instructions.

Determination of cutin and wax contents

Leaves were used to determine the cutin and wax content. Cutin monomers were extracted via tissue delipidation, ester depolymerization, and silylation, as described by Jenkin and Molina99. Waxes were determined as follows: leaf tissues were extracted by placing leaves in 10 mL of chromatographically pure chloroform for 30 s; 25 µL C24 alkane (1 μg/µL) was added as an internal standard, then the solutions were blow-dried with nitrogen, silylated with 100 µL of N, O Bis-trimethylsilyl-trifluoroacetamid (BSTFA) and 100 µL of pyridine for 30 min at 100 °C, and blow-dried with nitrogen again, and finally dissolved in 100 µL of hexane.

The cutin and wax monomer contents were determined using a gas chromatograph-mass spectrometer (GC–MS; 8890-7000D, Agilent Technologies, USA) fitted with an HP-5MS capillary column (length, 30 m; i.d., 0.25 mm; film thickness, 0.25 mm). For cutin, the injector was set at 250 °C, and the injected split ratio was 1:10 and helium was the carrier gas at a constant flow rate of 1.5 mL min−1. The oven was initially set at 50 °C for 2 min, increased by 40 °C min−1 increments to 200 °C, maintained for 5 min, then increased by 5 °C min−1 increments to 310 °C, and maintained for 20 min. The temperature of the MS detector was 325 °C, and the MS was set to scan mode > 40–600 amu (electron impact ionization). For wax, the injector was set at 250 °C, the oven was initially set at 80 °C for 1 min, increased by 15 °C min−1 increments to 260 °C, maintained for 10 min, then increased by 5 °C min−1 increments to 320 °C, and maintained for 24 min. The temperature of the MS detector was 320 °C, and the MS was set to scan mode > 40–600 amu. Six biological replicates were assessed in each experiment.

RNA extraction, library construction and sequencing

Total RNA was extracted using a TRIzol reagent kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. The RNA quality was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA) and verified using RNase-free agarose gel electrophoresis. Conventional kits were used to remove rRNA and enrich mRNA. The enriched mRNA was then cut into short fragments using a fragmentation buffer and reverse transcribed into cDNA using random primers. The cDNA fragments were purified using the QiaQuick PCR extraction kit (Qiagen, Venlo, Netherlands), end-repaired, poly(A) added, and ligated to Illumina sequencing adapters. The ligation products were size-selected by agarose gel electrophoresis, PCR-amplified, and sequenced using an Illumina HiSeqTM 4000 by Gene Denovo Biotechnology Co. (Guangzhou, China). Each treatment had three biological replicates.

RNA-seq data filtering of clean reads, denovo assembly and basic annotation of unigenes

Raw reads obtained from sequencing machines, including raw reads containing adapters or low-quality bases, affect subsequent assembly and analysis. Thus, to obtain high-quality clean reads, the reads were further filtered using Fastp software (version 0.18.0)100. The parameters were as follows: (1) reads containing adapters wer removed; (2) reads containing more than 10% of unknown nucleotides (N) were reomved; (3) low-quality reads containing more than 50% of low quality (Q-value ≤ 20) bases were removed. Transcriptome denovo assembly was conducted using the short reads assembly program, Trinity101. To annotate the unigenes, we used the BLASTx program (http://www.ncbi.nlm.nih.gov/BLAST/) with an E-value threshold of 1e-5 in the NCBI non-redundant protein (Nr) (http://www.ncbi.nlm.nih.gov), Swiss-Prot protein (http://www.expasy.ch/sprot), Kyoto Encyclopedia of Genes and Genomes (KEGG) (http://www.genome.jp/kegg), and COG/KOG databases (http://www.ncbi.nlm.nih.gov/COG). The protein with the highest sequence similarity to the given unigene was identified and the protein function annotation information of the unigene was obtained.

Differential expression and target genes analysis

The Omicsmart online data analysis platform (https://www.omicsmart.com/home.html#/) was used to calculate the differentially expressed genes (DEGs) between DT and DS genotypes under the control condition and 20% PEG-6000 treatment. Significant DEGs between the osmotic stress and control conditions were identified at |log2(foldchange) |> 2 and false discovery rate (FDR) adjusted P < 0.001. Subsequently, GO and KEGG pathway enrichment analysis was performed on the Omicsmart online data analysis platform (https://www.omicsmart.com/home.html#/) for differential basic genes in DT and DS under PEG-induced osmotic stress for 0, 3, and 24 h. Furthermore, a cluster heat map and trend analysis of the co-expressed genes were performed. Candidate genes with similar expression patterns were grouped together using the Omicsmart heatmap analysis tool (https://www.omicsmart.com/home.html#/). The clustering results were then divided into different trend blocks. Analysis of candidate genes trends using the Omicsmart trend analysis tool (https://www.omicsmart.com/home.html#/). Significant trend blocks were identified when |log2(foldchange)|> 2 and P < 0.05. The different profiles represent the numbers of the different trend blocks. Finally, pathway enrichment analysis was performed for candidate genes under osmotic stress.

Statistical analysis

Statistical analyses were performed using Microsoft Excel 2016. Analysis of physiological and phenotypic characterizations data was performed using IBM SPSS Statistics (version 26.0). Physiological and phenotypic characterizations values are shown as the mean ± standard deviation (SD) of biological replicates. Tukey's post hoc test was performed with a P-value < 0.05, considered significantly different. Origin 2021 was used for data visualization. The Omicsmart online data analysis platform (https://www.omicsmart.com/home.html#/) was used to visualize transcriptome sequencing data.

Conclusions

This study combined physiological and transcriptomic analyses to reveal differences in drought tolerance between two genotypes of E. sibiricus, DT and DS. The physiological results confirmed that DT had greater drought tolerance than DS. Transcriptome analysis identified multiple drought tolerance genes. By combining the physiological and biochemical results, we found that these genes may participate in several drought tolerance pathways in E. sibiricus, including ABA signaling pathway-induced stomatal closure, leaf cuticular wax deposition, ASA metabolism, and ROS scavenging. Unigene0005863 (EsSnRK2), Unigene0053902 (EsLRK10) and Unigene0031985 (EsCIPK5) may be involved in stomatal closure induced by abscisic acid (ABA) signaling pathway. Unigene0047636 (EsCER1) may positively regulates the synthesis of very-long-chain (VLC) alkanes in cuticular wax biosynthesis, influencing plant responses to abiotic stresses. This study provides some ideas for research on the mechanisms of drought tolerance in meso-xerophytic forage grasses and gives some clues for genetic improvement of drought tolerance in forages and crops.