Introduction

Modern agriculture employs advanced technologies and methods to enhance farming efficiency and crop productivity, aiming to provide food security to the global population. However, various environmental factors and human activities present significant challenges, with soil salinization being a major constraint1. Soil salinization results in the uptake of harmful ions, disrupting the cellular balance in plants, which impedes plant growth and development2. Salt stress also hinders plant growth through accelerated cell death caused by reactive oxygen species (ROS)3. While strategies such as soil reclamation and fertilizer application can mitigate the harmful effects of salinity, they often require considerable time and financial investment. In contrast, enhancing salinity tolerance by introducing beneficial genes that improve ion transport, homeostasis, and stress response offers a more efficient solution4. Therefore, an in-depth understanding of the molecular mechanisms underlying adaptation to saline environment is needed to develop crop varieties that can maintain high productivity under saline conditions5.

Soybean is widely cultivated for oil and protein, contributing approximately 28% of the world’s vegetable oil and 70% of protein meal consumption as of 2019 (http://soystats.com/international-world-protein-meal-consumption/). Although soybeans exhibit moderate tolerance to saline stress, high salt concentrations in irrigation water significantly impact plant growth6,7. When salinity levels exceed the threshold in soil, it triggers genetic, physiological, and biochemical changes that adversely affect all stages of the plant’s life cycle8.

Osmotic stress, a primary consequence of salinity, occurs due to elevated salt levels and reduces water availability to plant roots resulting in wilting, growth retardation, and plant death. Additionally, high salt concentrations disrupt the transport of ions in the root. The accumulated harmful sodium ions (Na+) in the cytoplasm disrupt normal cellular functions and hinder the uptake of potassium, calcium, and magnesium2. This disruption negatively affects vital metabolic processes and overall plant development. Furthermore, increased soil salinity interferes with the functionality of photosystem II (PSII), reducing photosynthetic activity. This decline in photosystem efficiency diminishes the photosynthetic rate leading to reduced biomass accumulation9,10.

Several genes, dispersed across the genome, regulate tolerance to environmental stresses11. A transgenic approach has been used to manipulate genes and pathways related to salt tolerance; however, despite demonstrating enhanced salinity tolerance, none of these have reached farmers’ fields12,13. As a result, exploiting naturally occurring genetic variations represents a promising approach. Several mapping studies elucidated the genetic basis of adaptation to salinity in soybeans. The first significant salt tolerance gene, Ncl, was mapped in a mapping population developed from a cross between a tolerant line S-100 and a susceptible line Tokyo7. Various mapping studies have since validated this gene in both wild and cultivated soybeans14. GmSALT3/GmCHX1 (Glyma03g32900), a key salt tolerance gene in soybeans15, is a member of the Cation/Proton Antiporter (CPA2) family, which reduces the accumulation of sodium and chloride ions in the shoot upon exposure to salinity stress by preventing their export through xylem loading16. Cheng et al.17 reported a salt tolerance mechanism mediated by the gene J, an ortholog of AtELF3, in transgenic soybean by influencing the expression of many downstream genes such as GmSIN1, GmNAC, GmWRKY12, GmWRKY27, and GmWRKY54.

Advancements in next-generation sequencing (NGS) have enabled researchers to analyze the whole repertoire of genes on a genome scale. RNA sequencing (RNA-Seq) has emerged as an efficient technique for identifying salt-stress-related differentially expressed genes (DEGs). However, only a few comparative transcriptomic studies involving salt-tolerant and sensitive genotypes have been conducted in soybean18,19. Hu et al.18 highlighted the role of phytohormone signaling, mitogen-activated protein kinase pathway, phenylpropanoid biosynthesis, and ribosome metabolism as crucial for salinity adaptation. Cheng et al.19 reported cultivar-specific strategies to mitigate salt stress, with starch and sucrose metabolism, as well as phenylpropanoid biosynthesis, being important in the tolerant genotype, and the ribosome pathway in the salt-susceptible genotype. Other studies have focused on comparisons of the transcriptomic profiles of recombinant inbred lines20 and a tolerant mutant with its susceptible parent21, highlighting the activation of stress-responsive genes and pathways under salinity stress.

Although numerous genes associated with salinity tolerance have been identified, knowledge remains limited regarding signal transduction, transcriptional regulation, and the molecular genetic mechanisms underlying salt tolerance in soybeans. Given the genetic complexity of salinity tolerance, there is a critical need for comprehensive genomic, transcriptomic, and epigenetic profiling across soybean genotypes with varying levels of salt tolerance. Furthermore, the identification and validation of candidate salt-responsive genes using gene editing approaches are crucial for breeding resilient soybean cultivars.

To address these gaps, the present study investigates the temporal dynamics of gene expression in response to salt stress through genome-wide transcriptomic profiling of a newly identified salt-tolerant Glycine max accession (PI 561363) and a salt-sensitive accession (PI 601984). By analyzing gene expression changes at multiple time points following salt exposure, we aim to uncover key regulatory networks, genotype-specific pathways, and early versus late transcriptional responses associated with salt tolerance. This approach will not only enhance our understanding of the molecular basis of salinity tolerance in soybean but also accelerate the discovery of candidate genes for crop improvement.

Methods

Plant material, salt treatment, and sampling

Two cultivated soybean (Glycine max) germplasm accessions (salt-tolerant PI 561363 and salt-sensitive PI 601984) were selected based on screening of 269 germplasm accessions under saline conditions22. The above soybean germplasm seeds were obtained from the USDA Soybean Germplasm Collection (https://www.ars-grin.gov/). The genotype PI 561363 demonstrated higher salt tolerance with a low leaf scorching score, low leaf sodium content, and leaf chloride content than PI 60198422. After surface sterilizing seeds from both accessions with 70% ethanol for 1 min, the seeds were thoroughly washed and then germinated in an incubator at 27 °C. Once germinated, the seeds were transferred to sand culture. There were two treatment groups—control and salt stress—with three biological replications each consisting of five plants per genotype.

The germinated seedlings were applied with the nutrient solution following Pruthi et al.22. Salt stress (150 mM NaCl) was imposed on the seedlings at the V2 stage. Leaf samples were collected from both control (SUSC and TOLC) and salt-stressed plants at 0 h, 6 h (SUS06 and TOL06), 24 h (SUS24 and TOL24), and 48 h (SUS48 and TOL48). Twenty-four samples were collected for RNA sequencing, representing four time points across the two genotypes, with three biological replicates each.

RNA-sequencing

RNA was extracted using the Direct-zol™ RNA Miniprep kit (Zymo Research, Irvine, CA, USA). The total RNA in each sample was quantified using an ND-1000 Spectrophotometer (Thermo Fisher Scientific, Waltham, USA). RNA sample purity was checked using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, USA). RNA samples were then used for cDNA library construction. Paired-end sequencing was done on the NovaSeq X Plus Series sequencing platform at Novogene Corporation Inc. (Sacramento, CA, USA). All raw read sequences of both genotypes were deposited in the NCBI’s Sequence Read Archive (SRA) under the accession number PRJNA1254663.

Quantification of morpho-physiological parameters

Chlorophyll content (CC) was measured following Pruthi et al.‘s procedure22. After collecting and freezing leaves from both genotypes, samples were stored in an ultralow freezer (−80 °C) until further measurements of physiological parameters were made.

Peroxidase (POX) activity

Fifty milligrams of leaf samples were ground using a Fast Prep homogenizer in an extraction buffer as described by Hartmann and Asch23. The enzyme extract was used to measure POX activity after vortexing and centrifugation of the ground leaf extract for 10 min at 4 °C, as described by Acevedo et al.24. Briefly, the enzyme extract (10 µl) was mixed with 3% H2O2 (10 µL) and 3 mM guaiacol (190 µL) in a 96-well polystyrene microplate. A microplate reader (SpectraMax® ParadigmMulti-Mode, Microplate Reader) was used to record the absorbance changes at 450 nm for 3 min. POX activity was expressed as units per gram of fresh weight.

Malondialdehyde (MDA) content

The method of Liu et al.25 was used to measure MDA content. Leaf samples (100 mg) were ground using 1500 µL of 5% (w/v) trichloroacetic acid (TCA) in a Fast Prep homogenizer. After centrifuging the homogenate at 13,000 × g for 20 min, lipid peroxidation was analyzed using the supernatant. A mixture of the supernatant (400 µL) and 0.5% Thiobarbituric acid (TBA) in 5% TCA (400 µl) was boiled for 30 min, followed by incubation in an ice bath to terminate the reaction. Measurement of MDA content was done at 532 nm in a spectrophotometer, and the nonspecific turbidity value at 600 nm, obtained using a microplate reader (SpectraMax® ParadigmMulti-Mode) helped in correcting the MDA content, which was expressed as nmoles g−1, following Du and Bramlage26.

Proline content

Proline content was measured following Shabnam et al.27. One hundred milligrams of leaves were homogenized in both liquid nitrogen and then in 3% (w/v) sulfosalicylic acid (1 mL). The mixture was placed into a 1.5 mL centrifuge tube, and after vortexing and centrifugation at 4 °C, 200 µL of acidic ninhydrin solution was added to the 100 µL of the supernatant and placed at 100 °C for 1 h. Measurement of absorbance at 520 nm was done using a microplate reader. Proline content was determined using a standard curve (0–50 µg mL−1) and expressed as µg g−1 of fresh weight.

Leaf ion content

To measure sodium content (LSC), potassium content (LPC), calcium content (LCaC), and chloride content (LCC), leaves that were oven-dried at 65 °C for 7 days were ground. One hundred milligrams of ground leaf tissue was digested using 2.2 mL of deionized water and 5 mL of concentrated HNO3 at 125 °C for 2 h and 45 min. Digestion continued at 125 °C for 45 min after the addition of 3 mL of H2O2 to the digestion mixture28. The digested sample volume was adjusted to 20 mL with deionized water, and the concentrations of Na+, K+, Ca2+, and Cl− were measured using ICP-OES. The leaf Na/K ratio (LNaK) was calculated using the following formula: leaf Na+ concentration/leaf K+ concentration.

Transcriptome profiling

The read quality was examined using FASTQC (available at http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Both the low-quality reads and adapter sequences were purged using Trimommatic version 0.3929 with the following default settings: LEADING:3 TRAILING:3 SLIDINGWINDOW: 4:15 MINLEN:36. The filtered reads were mapped to the soybean reference genome (Glycine_max_Wm82.a2.v1) from Phytozome (http://phytozome.jgi.doe.gov, accessed on 15 March 2024), using the Spliced Transcripts Alignment to a Reference (STAR 2.7.11b; https://github.com/alexdobin/STAR). Genome-guided assembly was performed using Trinity version 2.15.030. The expression levels of genes were estimated using the Expectation-Maximization (RSEM) method31. DESeq2 (https://bioconductor.org/packages/release/bioc/html/DESeq2.html)32 was used to analyze the differential expression of genes. Genes were considered upregulated when the Benjamini-Hochberg adjusted p-value (FDR) was ≤ 0.001 and a log2 fold change was ≥ 2, while those with a log2 fold change ≤ 2 were considered downregulated. Functional annotation of transcripts was performed using blastx against the UniProt-SwissProt protein database (ftp://ftp.uniprot.org/pub/databases/uniprot/). Further, GO terms33 including cellular component, biological process, and molecular function were identified for the soybean transcriptome using Trinotate pipeline30. Cytoscape34 and BiNGO35 package was utilized to perform GO enrichment analysis. The Web-based tools named REVIGO36 and ShinyGO37 were used to visualize the enriched GO terms. Additionally, KEGG pathways for the soybean transcriptome were identified using KEGG database38 and bar plots were constructed using ShinyGo37.

Validation of gene expression via qRT-PCR

Nine genes showing differential expression levels were selected and primers were designed based on the coding sequences using Primer3Plus software. The internal standard gene was represented by the elongation factor 1 alpha (GmEF1α; Glyma.02G276600). qRT-PCR was performed on a QuantStudio 3 Real-Time PCR system (Applied Biosystems, Waltham, MA, USA) using PerfeCTa™ SYBR® Green FastMix™, Low ROX™ (Quanta BioSciences, Gaithersburg, MD, USA) following Karan et al.39. Each reaction was performed in two technical replicates, with three independent biological replicates. The 2–∆∆CT method40 was followed to calculate the gene expression. The TMM (Trimmed Mean of M-values) values from the transcriptomic data and the relative gene expression from qRT-PCR were used to calculate the Pearson correlation coefficient for each gene. The individual correlation coefficients were then averaged to determine the overall correlation between the transcriptomic and qRT-PCR results.

Statistical analysis

Physiological and biochemical data were analyzed using R version 4.3 (https://www.r-project.org/). To evaluate the interaction effects between genotype and time point, a two-way factorial ANOVA was conducted with both factors treated as fixed effects; post hoc comparisons of interaction differences were assessed using Tukey’s HSD test and are denoted by letter groupings. Histograms were generated using Microsoft Office Excel 2019, and the standard errors of the mean (± SE) were represented by error bars. Intra-sample and inter-sample variability were analyzed using the Pearson correlation coefficient and principal component analysis (PCA)41.

Results

Phenotypic response of soybean genotypes to salt stress

There was no change in malondialdehyde (MDA) production among the samples collected at different time points for the tolerant genotype. However, the susceptible genotype showed a significant increase in MDA production from 0 h to 48 h (Fig. 1a), rising from 23.7 nmol g−1 at 0 h to 60.0 nmol g−1 at 48 h. Additionally, proline level was significantly higher in the susceptible genotype at 48 h compared to the tolerant genotype (Fig. 1b), suggesting that the susceptible genotype experienced higher degree of osmotic stress.

Peroxidase activity remained significantly higher at the 6 h time point in both genotypes. However, it stabilized in the tolerant genotype as exposure to saline conditions continued, and it decreased from 6 h to 24 h and 48 h in the susceptible genotype. However, activity remained significantly higher than at the initial 0 h condition (Fig. 1c). Chlorophyll content in the tolerant genotype remained consistently higher across all time points than in the susceptible genotype (Fig. 1d).

Fig. 1
figure 1

Variability in physiological parameters between tolerant (Pl 561363) and susceptible genotype (PI 601984) under saline conditions. (a) MDA content, (b) Proline content, (c) POX activity, (d) Chlorophyll content (CC), (e) Leaf calcium content (g/kg), (f) Leaf chloride content (g/kg), (g) Leaf potassium content (g/kg), and (h) Leaf sodium content (g/kg). Error bars represent the standard error based on three independent samples. Bars marked with different letters indicate statistically significant differences (p<0.05).

Full size image

Calcium uptake increased at the 6 h time point but returned to baseline levels in both-genotypes. In the tolerant genotype, chloride uptake decreased at 6 h but increased significantly thereafter (Fig. 1e). In contrast, the susceptible genotype exhibited continuous chloride accumulation from 0 h to 48 h. However, the two genotypes did not differ in chloride accumulation (Fig. 1f).

Potassium uptake decreased significantly at 6 h in the tolerant genotype but increased at 24 h and stabilized at 48 h. The susceptible genotype showed a similar trend, although the decrease at 6 h was not statistically significant compared to the 0 h time point (Fig. 1g). The tolerant genotype effectively reduced sodium uptake at both the 6 h and 24 h time points, although sodium accumulation sharply increased by 48 h (Fig. 1h). In contrast, the susceptible genotype showed a continuous and significant increase in sodium accumulation from 0 h to 48 h, with sodium uptake at 48 h significantly higher than in the tolerant genotype (Fig. 1h).

RNA‑seq data analysis

For transcriptomic profiling under salt stress, we performed mRNA sequencing on 24 leaf samples collected at four different time points from two genotypes after treatment with 150 mM NaCl. After filtering, the susceptible genotype samples retained an average of 32 million reads, while the tolerant genotype samples retained an average of 28 million reads (Table 1). The soybean reference genome (Glycine_max_v2.1, INSDC Assembly GCA_000004515.4), was used to map the high-quality reads, with 91–94% mapping efficiency (Table 1). The percentage of guanine and cytosine bases is 38%. The subsequent assembly identified 315,386 transcripts and 141,933 genes using the Trinity package. The assembly produced an N50 of 2,981 bp, indicating that half of the total assembled sequence is contained in contigs of this length or longer, signifying strong assembly quality (Table 2). Additionally, the overall size of the soybean transcriptome was approximately 590 Mb.

Table 1 Summary of mapping statistics for susceptible (PI 601984) and tolerant (PI 561363) soybean genotypes under saline stress at 150 mM NaCl.
Full size table
Table 2 Overview of genome-guided assembly for the soybean leaf transcriptome.
Full size table

Principal component analysis (PCA) revealed clear distinctions among different time points (Figure S1). This was further supported by the Pearson correlation coefficient, which showed high correlations among the biological replicates from each time point (Figure S2).

Differentially expressed genes under salt stress

We analyzed DEGs in two genotypes across four time points to investigate the molecular mechanisms underlying soybean adaptation to saline conditions. At 6 h, we identified 1,465 DEGs in the susceptible genotype, with 506 upregulated and 959 downregulated (Table S1). In the tolerant genotype, 1,807 DEGs were found, with 521 upregulated and 1,286 downregulated (Table S1). At 24 h, the number of DEGs decreased for both genotypes. The susceptible genotype had 681 DEGs (402 upregulated and 279 downregulated), while the tolerant genotype had 786 DEGs (547 upregulated and 239 downregulated) (Table S1). At 48 h, there was a substantial increase in DEGs for both genotypes. The susceptible genotype exhibited 5,479 DEGs, with 2,553 upregulated and 2,926 downregulated, while the tolerant genotype had 4,561 DEGs, with 2,214 upregulated and 2,348 downregulated (Table S1).

Venn diagrams illustrate unique and common DEGs across the two soybean genotypes at the four time points (Fig. 2). We made six comparisons: SUSC-SUS06, SUSC-SUS24, SUSC-SUS48, TOLC-TOL06, TOLC-TOL24, and TOLC-TOL48. The SUSC-SUS06 pair showed 538 unique DEGs, while the TOLC-TOL06 pair had 880 unique DEGs (Fig. 2a). At 24 h, the number of unique DEGs decreased to 471 in the SUSC-SUS24 pair and 576 in the TOLC-TOL24 pair (Fig. 2b). At 48 h, the number of unique DEGs increased, with 2,379 in the SUSC-SUS48 pair and 1,461 in the TOLC-TOL48 pair (Fig. 2c). One hundred eighteen genes were consistently differentially expressed within the susceptible genotype at the 6 h, 24 h, and 48 h time points (Fig. 2d). For the tolerant genotype, 189 unique DEGs were consistently differentially expressed at all three time points (Fig. 2e). Across all comparisons (SUSC-SUS06, SUSC-SUS24, SUSC-SUS48, TOLC-TOL06, TOLC-TOL24, and TOLC-TOL48), 33 genes exhibited consistent differential expression at each time point in both genotypes.

Transcript abundance of selected genes

Based on qRT-PCR analysis (Table S2) of nine selected DEGs, gene expression patterns observed in the RNA-seq were confirmed (Fig. 3). For instance, GmCHX20a (Glyma.03G171500) and GmHAK5 (Glyma.03G264100) showed the highest log-fold changes at 48 h in the susceptible genotype. Similarly, GmTDT (Glyma.09G010800) and GmLTI (Glyma.09G139600), exhibited the highest relative expression changes at the same time point. The reliability of the transcriptome results was further supported by a significant correlation coefficient (r = 0.85) between the TMM values from the transcriptomics analysis and the relative gene expression of nine genes obtained from qRT-PCR, reinforcing the validity of the transcriptomic findings.

Fig. 2
figure 2

Venn diagrams illustrating the distribution of differentially expressed genes (DEGs) across genotypes and time points under salt stress conditions. The diagrams highlight both common and unique DEGs identified in pairwise comparisons. (a) SUSC-SUS6 vs. TOLC-TOL6, (b) SUSC-SUS24 vs. TOLC-TOL24, (c) SUSC-SUS48 vs. TOLC-TOL48, (d) Temporal comparison within the SUSC genotype (SUS6 vs. SUS24 vs. SUS48), (e) Temporal comparison within the TOLC genotype (TOL6 vs. TOL24 vs. TOL48).

Full size image
Fig. 3
figure 3

Validation of DEGs identified in RNA-Seq data using qRT-PCR. The y-axis represents the relative change in mRNA expression level, while the x-axis depicts the different time points and genotypes. (a) GmCLC-C, (b) GmCHX20a, (c) GmHAk5, (d) GmCAX3, (e) GmLT1, (f) GmKUP6, (g) GmTDT, (h) GmGST19, (i) GmOST1.

Full size image

Gene ontology analysis in susceptible genotype

GO enrichment analysis was conducted for each genotype at different time points of stress treatment to evaluate the functional relevance of genes in both genotypes (Fig. 4).

For the SUSC-SUS06 pair, we identified 161 biological process (BP) terms, 85 molecular functions (MF) terms, and 12 cellular components (CC) terms (Table S3). The BP terms were categorized into sub-groups such as responses to stress, regulatory processes, biosynthetic processes, developmental processes, hormone and signaling pathways, cellular localization, photosynthesis, cell wall-related processes, and ion transport (Table S4). Notable terms in the stress response sub-group included GO:0009415 (response to water deprivation) and GO:0009628 (response to abiotic stimulus). For cell wall-related processes, key terms were GO:0042545 (cell wall modification), GO:0071555 (cell wall organization), and GO:0009827 (plant-type cell wall modification and loosening). In the ion transport sub-group, significant terms included GO:0015822 (ornithine transport) and GO:0006810 (transport). Additionally, terms related to biosynthetic processes, such as suberin (GO:0010143), cutin (GO:0010345), and wax (GO:0010025), were also enriched.

There were fewer enriched terms at the 24-hour time point for the SUSC-SUS24 pair, with 22 BP, 42 MF, and 13 CC terms (Table S5). The BP terms were primarily categorized into ‘response to stress’ and ‘cell wall-related processes’ (Fig. 4b; Table S6). Key terms in the stress response cluster included GO:0050896 (response to stimulus), GO:0006950 (stress response), and GO:0009611 (response to wounding). Cell wall-related terms included GO:0042545 (cell wall modification) and GO:0071555 (cell wall organization).

There was a substantial increase in enriched terms for the SUSC-SUS48 pair at the 48-h time point, with 218 BP, 94 MF, and 64 CC terms (Table S7). The stress response sub-group included GO:0009614 (response to water deprivation), GO:006979 (response to oxidative stress), and GO:0009628 (response to abiotic stimulus) (Fig. 4c; Table S8). The cell wall-related processes sub-group featured the following GO terms: GO:0042545 (cell wall modification), GO:0071555 (cell wall organization), and GO:0009827 (cell wall loosening). In the ion transport sub-group, important terms included GO:0006811 (ion transport), GO:0006820 (anion transport), and GO:0030004 (monovalent inorganic cation homeostasis). Additional enriched terms included GO:0009738 (abscisic acid-mediated signaling pathway), GO:0009742 (brassinosteroid-mediated signaling pathway), GO:0016051 (carbohydrate biosynthetic process), GO:0005975 (carbohydrate metabolic process), and GO:0015979 (photosynthesis).

Fig. 4
figure 4

Gene ontology (GO) enrichment analysis highlighting enriched biological processes (BP) in susceptible (SUS) and tolerant (TOL) genotypes at different time points. Subfigures detail specific enriched BP categories at different comparisons. (a) SUSC-SUS6, (b) SUSC-SUS24, (c) SUSC-SUS48, (d) TOLC-TOL6, (e) TOLC-TOL24, and (f) TOLC-TOL48.

Full size image

GO analysis in tolerant genotype

The TOLC-TOL06 pair at the 6-h time point identified 209 BP terms, 103 MF terms, and 22 CC terms (Table S9). The BP terms were categorized into the ‘ion transport’ sub-group (Fig. 4d; Table S10). Enriched terms included GO:0055074 (calcium ion homeostasis), GO:0055080 (cation homeostasis), GO:0030003 (cellular cation homeostasis), GO:0006873 (cellular ion homeostasis), and GO:0055065 (metal ion homeostasis). DEGs in this genotype also showed enrichment in lipid (GO:0008610, GO:0006629) and phosphorylation processes (GO:0016310). Notably, the TOLC-TOL06 pair did not include cell wall modification terms, which were present in the SUSC-SUS06 pair. However, standard GO terms between the two genotypes included GO:0009415 (response to water deprivation), GO:0009628 (response to abiotic stimulus), GO:0010345 (cutin biosynthesis), and GO:0010025 (wax biosynthesis).

The TOLC-TOL24 pair was associated with 140 GO terms, including 62 BP, 68 MF, and 10 CC terms (Table S11). The BP terms were sub-categorized into ‘response to stress,’ phospholipid transport,’ ‘regulation processes in plants,’ ‘developmental processes,’ ‘ion transport,’ and ‘hormone and signaling pathways’ (Fig. 4e; Table S12). Notable terms included GO:0009651 (response to salt stress), GO:0010200 (response to chitin), GO:0009723 (response to ethylene stimulus), GO:000925 (response to hormone stimulus), GO:0051179 (localization), and GO:0015914 (phospholipid transport).

For the TOLC-TOL48 pair, DEGs were associated with 224 BP terms, 93 MF terms, and 52 CC terms (Table S13). Enriched terms specific to saline stress in TOLC-TOL48 pair included GO:0055075 (potassium ion homeostasis), GO:0006813 (potassium ion transport), GO:0006873 (cellular ion homeostasis), GO:0055080 (cation homeostasis), GO:0009873 (ethylene-mediated signaling pathway), GO:0010345 (suberin biosynthetic process), GO:0008610 (lipid biosynthetic process), and GO:0006629 (lipid metabolic process) (Fig. 4f; Table S14).

The GO enrichment results revealed that the susceptible genotype was primarily enriched in BP terms related to biological regulation and response to stimuli. In contrast, the tolerant genotype showed enrichment in terms associated with light response and circadian rhythms (Figure S3). At 24 h, defense responses and cell wall modifications were prominent in the susceptible genotype, whereas metabolic processes such as carbohydrate and RNA metabolism were important in the tolerant genotype. By 48 h, the susceptible genotype continued to respond to biological and chemical stimuli, whereas the tolerant genotype exhibited a shift toward transcription regulation and biosynthetic processes.

KEGG pathway analysis

DEGs from both genotypes were subjected to KEGG pathway analysis to gain a better understanding of the metabolic pathways related to salt stress (Figures S4 and S5) For the SUSC-SUS06 pair, significant enrichment was observed in pathways related to circadian rhythm, phytohormone signaling, and signal transduction (Figure S4a). However, KEGG pathways identified at 24 h were not significantly enriched (Figure S4b). In the SUSC-SUS48 pair, the identified KEGG pathways were significantly enriched, including zeatin biosynthesis, photosynthesis, glycerolipid metabolism, glycerophospholipid metabolism, phytohormone signaling, and signal transduction (Figure S4c). For the tolerant genotype, significant enrichment in the fatty acid elongation pathway was observed at 6 h between TOLC and TOL06 (Figure S5a). However, no significant pathway enrichment was identified for DEGs in the TOLC-TOL24 pair (Figure S5b). At the 48 h time point, the KEGG pathways were the same as those observed in the SUSC-SUS48 pair (Figure S5c).

Discussion

Salinity tolerance in every crop, including soybean, is influenced by various cellular, metabolic, and physiological processes. Recent advancements in genomics have enabled the exploration of the intricate molecular genetic mechanisms underlying salt tolerance in soybean. We conducted transcriptome profiling on two soybean genotypes with contrasting responses to salt stress at four different time points to investigate the molecular genetic changes during the initial stages of salt stress exposure. Our study aims to elucidate the molecular genetic basis of salt tolerance mechanisms, thereby developing soybean varieties that can maintain productivity in salt-affected farmland. By identifying key regulatory genes and metabolic pathways, the development of salt-tolerant soybean varieties can be accelerated, thereby enhancing agricultural sustainability and food security.

Physiological response of soybean seedlings to salt stress

Excessive reactive oxygen species (ROS) generated due to salinity stress can damage cellular processes. One indicator of damage to cell membranes is lipid peroxidation, measured by malondialdehyde (MDA) levels3,42. In our study, MDA levels increased sharply from 0 h to 48 h in the sensitive genotype, whereas the tolerant genotype showed no significant change. This finding is consistent with earlier reports in salt-sensitive rice seedlings43, suggesting that salt-tolerant genotypes exhibit better physiological regulation of lipid peroxidation.

Peroxidase (POX) is a key enzyme in plants’ defense systems, essential for scavenging ROS, especially under saline conditions44,45. Both genotypes exhibited increased POX activity at 6 h; however, the tolerant genotype returned to control levels by 24 h and 48 h, while the susceptible genotype maintained elevated POX activity. This suggests consistently higher ROS production in the susceptible genotype. Proline, a well-known osmolyte produced in response to abiotic stresses, helps to reduce proteolysis, scavenge free radicals, stabilize cellular structures, and enhance redox potential46. Increased proline helps prevent cellular dehydration and maintains stability under salt stress. Both genotypes showed significant proline accumulation at 48 h. However, the susceptible genotype displayed a significantly higher accumulation than the tolerant genotype, indicating greater osmotic stress in the susceptible genotype at the same time.

Calcium ions (Ca2+) are vital in driving cellular responses in stressful environments. Excess extracellular salt triggers a rise in cytosolic Ca2+47. We observed increased leaf calcium uptake at 6 h for both genotypes; however, calcium uptake plateaued in the tolerant genotype after this time, while it decreased below control levels in the susceptible genotype. This suggests that calcium uptake is disrupted under salt stress. Salt stress also affects water potential and ionic balance, influencing potassium ion uptake48. Although potassium accumulation was reduced at 6 h, it stabilized in both genotypes by 24 h and 48 h, indicating an adaptive mechanism to maintain K+ levels despite ongoing stress. Sodium and chloride ions are major contributors to impaired plant growth in saline environments49. We observed differences in the accumulation patterns of Na+ and Cl− between the genotypes. Due to effective ion exclusion mechanisms, the tolerant genotype initially showed lower Na+ and Cl− levels at 6 h. However, Na+ and Cl− uptake in the tolerant genotype increased at 48 h due to limitations in exclusion mechanisms or a shift towards ion compartmentalization50,51. In contrast, the susceptible genotype continuously accumulated higher Na+ and Cl− levels from 0 to 48 h, indicating poor exclusion or sequestration and greater ionic toxicity and damage2. This inefficiency in managing Na+ and Cl− influx exacerbated osmotic stress and disrupted crucial physiological processes.

Enriched biological processes in susceptible and tolerant genotype

Gene Ontology (GO) analysis is a valuable tool for annotating and comparing gene products, offering insights into gene functions across various biological categories52. The TOLC-TOL06 group showed significant enrichment in the lipid metabolism process, which is crucial for mitigating lipid peroxidation in plant cells. Additionally, we observed unique enrichment in the categories of ion homeostasis, phosphorylation, and carbohydrate metabolism, all of which are essential for managing Na+ and Cl− ions and alleviating osmotic stress. The TOLC-TOL24 pair exhibited a significantly higher number of enriched GO terms compared to the SUSC-SUS24 pair. Enriched terms included responses to salt stress, chitin, ethylene, hormones, localization, and phospholipid transport. This suggests that the tolerant genotype activated a broader range of mechanisms early on to counteract salt stress compared to the susceptible genotype53,54,55. Notably, the tolerant genotype enriched lipid metabolism processes at 6 h and 48 h. These processes are essential for maintaining membrane stability and reducing permeability56. This hypothesis is further supported by the fact that the tolerant genotype exhibited stable lipid peroxidation levels, in contrast to the enhanced levels observed in the susceptible genotype.

GmSALT3 expression under salt stress

Salinity tolerance in soybeans is primarily associated with the dominant gene Ncl, which controls chloride exclusion in the cultivar ‘Lee’57. Since ‘Lee’ originated from a cross between ‘CNS’ and ‘S-100’58, the salt tolerance QTL might be the Ncl gene7. This gene has been mapped to chromosome 3 across various mapping populations16. Glyma03g32900, also known as GmSALT3 (Glycine_max_Wm82.a1.v1.1)11 or GmCHX116, is a Cation/Proton Antiporter (CPA2), which plays a key role in excluding sodium and chloride ions from roots and maintaining K+ homeostasis16,59.

Full-length GmSALT3 is linked to salt tolerance, while its truncated form is associated with susceptibility to salt exposure14,16. Recent soybean genome data reveal three genes - Glyma.03G171500, Glyma.03G171600, and Glyma.03G171700 - within this region, all showing homology to the CPA2 family (Glycine_max_Wm82.a2.v1; https://phytozome-next.jgi.doe.gov/). Our study found that expression of Glyma.03G171500 was elevated in the leaves of the tolerant genotype at both 24 h (4-fold increase) and 48 h (7-fold increase). In contrast, the susceptible genotype showed a significant increase (8-fold) only at 48 h. Notably, overexpression of the GmCHX20a gene in Arabidopsis has been linked to enhanced Na+ uptake in soybean roots 60,61. The upregulation of Glyma.03G171500 correlates with increased Na+ accumulation in both genotypes at 48 h, suggesting its crucial role in sodium regulation under salt stress. However, significantly lower expression of GmCHX20a, as observed in quantitative reverse transcription PCR (qRT-PCR) analysis, supports lower Na+ accumulation in the tolerant genotype at 48 h compared to the susceptible genotype.

Putative candidate genes imparting salt tolerance

At the 6 h time point, we observed an 11.5-fold increase in the expression of the caffeic acid O-methyltransferase gene (Glyma.12G109800) in the salt-tolerant genotype. This result aligns with the findings of Hu et al.18 and Cheng et al.19. A few other β-Amyrin synthase genes (Glyma.15G101800, Glyma.15G065600, and Glyma.03G121300) were significantly upregulated. At the same time point, the vacuolar cation/proton exchanger GmCAX3 was significantly downregulated at 6 h in both the susceptible and tolerant genotypes, though to varying degrees. The susceptible genotype showed a 14-fold reduction, while the tolerant genotype exhibited a 3.2-fold decrease. A few anion transport genes, including NRT1 (Glyma.20G247400; Glyma.18G021700), SLAH1 (Glyma.06G307500), and boron transporter 1 (Glyma.03G222300), were significantly downregulated in the tolerant genotype (Fig. 5). This downregulation of these genes may help reduce chloride uptake, improving salt stress management. In contrast, the upregulation of GsCLC-c2 in the tolerant genotype suggests its role in ion homeostasis by reducing chloride and sodium accumulation in shoots. This is supported by previous studies59,62 in which over-expression of the chloride channel gene GsCLC-c2 (Glyma.16G208400) from the wild soybean contributed to Cl− and NO3− homeostasis and enhanced salt tolerance by increasing accumulation of Cl− in the roots and limiting its export to the shoots.

At the 24 h mark, the tolerant genotype exhibited a 7-fold increase in the expression of HAK5/HAK1/POT5 (Glyma.03G264100) (Fig. 6), a gene essential for potassium uptake under potassium-deficient conditions63. We also identified a gene analogous to Arabidopsis AtERF98 (Glyma.15G159100), which is linked to salt tolerance64, showing upregulation of 7-fold in the tolerant genotype. Additionally, transporters for borate (Glyma.15G011200), ammonium (Glyma.07G153800), and phosphate (Glyma.03G162800) were upregulated, suggesting enhanced nutrient management under salt stress. The gene Glyma.11G198500, homologous to AtGSTU19 (glutathione S-transferase TAU 19), showed a 15-fold increase in expression in the tolerant genotype and a 28-fold increase in the susceptible genotype. Notably, the overexpression of AtGSTU19 in Arabidopsis resulted in improved survival under salt stress65.

Fig. 5
figure 5

Heatmap illustrating the transcriptomic landscape of DEGs in the tolerant rice genotype at 6 h under salt stress and control conditions (TOLC-TOL06). DEGs were functionally categorized into major biological process categories including transporters, stress response, cell wall modification, carbohydrates, photosystems, lipid metabolism, and phytohormones & signaling. Expression values were calculated as TMM-nonnalized FPKM, and log2-transfonned for visualization of expression changes across conditions. DEGs with a false discovery rate (FDR) < 0.001 and a log2 fold change (log2FC) ≥ 2 were included. Please refer to the color scale bar for expression.

Full size image
Fig. 6
figure 6

Heatmap illustrating the transcriptomic landscape of DEGs in the tolerant rice genotype al 24 h under salt stress and control conditions (TOLC-TOL24). DEGs were functionally categorized into major biological process categories including transporters, stress response, cell wall modification, carbohydrates, photosystems, lipid metabolism, and phytohormones & signaling. Expression values were calculated as TMM-normalized FPKM, and log2-transformed for visualization of expression changes across conditions. DEGs with a false discovery rate (FDR) ≤ 0.001 and a log2 fold change (log2FC) ≥ 2 were included. Please refer to the color scale bar for expression.

Full size image

At the 48 h time point, both genotypes displayed significant upregulation of ion transporter genes, including GmCHX20a (Glyma.03G171500), GmHAK5/HAK1/POT5 (Glyma.03G264100), GmHAK11 (Glyma.15G159000), GmKEA6 (Glyma.08G029300), GmCHX18 (Glyma.01G223600), and GmCLC-C (Glyma.09G157900)63,66. Notable differences in gene expression were observed between the susceptible and tolerant genotypes for these genes (Fig. 7). We observed a significant enrichment of genes associated with the abscisic acid (ABA) and ethylene pathways. The core ABA signaling pathway, which regulates stomatal function, includes PYR/PYL/RCAR, PP2C phosphatases, the SnRK2 kinase OST1, the SLAC1 and QUAC1 anion channels67. In our study, we detected upregulation of PYL family gene GmPYL5 (Glyma.14G056300) and several GmPP2C phosphatase family genes, such as GmHAI3 (Glyma.01G225100), GmHAI2 (Glyma.19G069200), GmPP2CA (Glyma.18G035000), along with the SnRK2 kinase GmOST1/GmSnRK2.6 (Glyma.17G148800) (Fig. 8). Additionally, we observed the upregulation of downstream genes regulated by GmOST1/SnRK2.6, including GmKUP6 (Glyma.01G031800) and GmSLAC1 (Glyma.06G307500). The SnRK2.6 kinase phosphorylates these ion channel genes, promoting closure of stomata under saline conditions68, suggesting a role in salt avoidance in soybeans (Fig. 8).

Fig. 7
figure 7

Heatmap illustrating the transcriptomic landscape of DEGs in the tolerant rice genotype at 48 h under salt stress and control conditions (TOLC-TOL048). DEGs were functionally categorized into major biological process categories including transporters, stress response, cell wall modification, carbohydrates, photosystems, lipid metabolism, and phytohormones & signaling. Expression values were calculated as TMM-normalized FPKM, and log2-transformed for visualization of expression changes across conditions. The colorgradient reflects expression magnitude, with red indicating upregulation and green denoting downregulation. DEGs with a false discovery rate (FDR) ≤ 0.001 and a log2 fold change (log2FC) ≥ 2 were included. Please refer to the color scale bar for expression.

Full size image
Fig. 8
figure 8

Key differentially expressed genes (DEGs) contributing to salinity tolerance in the tolerant genotype (Pl 561363) after 48 h of salt stress exposure. This figure is based on the functional roles of these DEGs based on the literature on Arabidopsis and soybean. Key genes include CHX20a (Cation/H+ Antiporter 20) for enhanced Na+ uptake, PM H+-ATPase for establishing an H+ gradient across the plasma membrane to promote K+ uptake via HAK5 to maintain ionic homeostasis, and genes in the ABA signaling pathway such as PYL (Pyrabactin Resistance-Like), PP2C (Protein Phosphatsae 2 C) family, and SnRK2 kinase OSTl/SnRK2.6, that regulate stomatal closure by regulating KUP6 (K+ Uptake Permease 6) and SLAC1 (Slow Anion Channel-Associated 1). Additionally, ethylene signaling pathway genes, including ERF1 (Ethylene-responsive transcription factor 1B), RD29B  (Desiccation-responsive protein 29B), ATOSM34 (Osmotin 34), and SRO4, were significantly upregulated, along with ERF98 (Ethylene-responsive transcription factor 098), which enhances salinity tolerance through increased production of the antioxidant ascorbinc acid. Red arrows indicate upregulation of the respective gene.

Full size image

Various biotic and abiotic stressors trigger the synthesis of ethylene69,70. In our study, we observed the upregulation of key ethylene signaling genes, including GmERF1 (Glyma.20G203700) and GmERF98 (Glyma.15G159100) (Fig. 8). ERF1 activates a specific set of genes under saline stress, such as GmRD29B (Glyma.09G139600), GmATOSM34 (Glyma.01G217600), and GmSRO5 (Glyma.17G089400), which contribute to salinity tolerance71. Consistent with this, we observed the upregulation of these genes alongside GmERF1 in our study. Additionally, ERF98 enhanced tolerance to salt stress in Arabidopsis by accumulating the antioxidant ascorbic acid, which was also upregulated in our findings (Fig. 8)64.

Beyond stress signaling pathways, we observed the upregulation of key genes associated with adaptation to low potassium (K+) conditions. Notably, GmAHA2 (Glyma.17G061800) and GmHAK5 (Glyma.03G264100) are crucial under saline stress, which often get induced under low K+ conditions (Fig. 8). AHA2 (H+-ATPase 2) is essential for maintaining membrane potential and facilitating the uptake of potassium ions under potassium-deficient conditions, while HAK5 (High-Affinity K+ Transporter 5) is vital for potassium uptake under K+ scarcity 72,73. Their upregulation suggests that potassium homeostasis regulated by this gene may be essential for plants’ adaptation to saline environments.

We observed upregulation of the GmCLC-c gene (Glyma.09G157900), which is a key player in regulating chloride (Cl⁻) transport and anion homeostasis in soybean (Fig. 8)59. In addition to GmAHA2 and GmHAK5, there was consistent upregulation of GmCHX20a (Fig. 8). This gene has been shown to increase sodium (Na+) uptake under saline conditions, potentially disrupting ion homeostasis60,61. GmCHX20a has been shown to enhance Na⁺ uptake when ectopically expressed in Arabidopsis and soybean hairy roots, resulting in increased salt sensitivity 60,61. While GmCHX20a promotes Na⁺ accumulation, it may also play a role in osmotic adjustment during the early phase of salt exposure by facilitating Na⁺ influx to maintain turgor pressure. However, sustained expression without efficient Na⁺ exclusion mechanisms can lead to ionic toxicity. In our study, GmCHX20a was strongly upregulated at the 48-hour time point under salt stress. Notably, this time point also corresponded with the highest Na⁺ accumulation in the leaves of both tolerant and sensitive genotypes, with greater accumulation observed in the sensitive genotype. In contrast, GmHAK5, a high-affinity K⁺ transporter, with shared homology with AtHAK5 in Arabidopsis thaliana is induced under low potassium conditions and functions to restore K⁺ levels and maintain ionic balance. The temporal co-induction of these genes suggests a coordinated response in which GmCHX20a may facilitate osmotic adjustment via Na⁺ influx, while GmHAK5 counteracts ionic toxicity by promoting K⁺ uptake. However, the downstream targets of GmCHX20a remain unidentified, and its interaction with GmHAK5 requires further investigation. This dual response underscores the complexity of the plant’s regulatory mechanisms under saline stress, where increased sodium uptake may counteract efforts to maintain stable potassium levels.