Ion-structured transcriptional network reorganization is associated with tissue-specific adaptation to saline–alkaline stress in sorghum

Abstract

Saline–alkaline stress imposes complex ionic and high-pH constraints that disrupt cellular homeostasis and metabolic stability in crops. Here, we suggest that in Sorghum bicolor, ionic imbalance is associated with coordinated transcriptional network patterns. Integrating ion homeostasis measurements, antioxidant enzyme profiling, and weighted gene co-expression network analysis (WGCNA), we characterized tissue-specific responses to saline–alkaline stress induced by 50 mM Na₂CO₃ (pH ≥ 10.5) after 24 and 72 h of exposure. Stress induced pronounced Na⁺ accumulation in both leaf and root tissues, whereas K⁺ levels remained stable in leaves but declined markedly in roots, revealing divergent ionic regulation strategies. Transcriptomic analysis uncovered distinct temporal dynamics, with roots exhibiting broader and more sustained reorganization. WGCNA identified Na⁺- and K⁺-associated co-expression modules whose eigengene expression closely paralleled ion dynamics. In leaves, a Na⁺-correlated module enriched for endoplasmic reticulum protein folding components underwent coordinated repression. In roots, a K⁺-associated module enriched for plastid metabolism and ion transport declined in parallel with K⁺ depletion. Hub gene connectivity strongly aligned with ion traits, supporting structured and ion-associated network patterns. Together, these findings suggest that ionic imbalance is closely associated with tissue-specific stress responses and provide candidate modules and hub genes for future functional validation of saline–alkaline tolerance in sorghum.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12870-026-09441-6.

Introduction

Soil salinization and alkalization are major constraints on agricultural productivity, affecting over 800 million hectares of arable land worldwide [1, 2]. Ongoing climate change, irregular precipitation, and unsustainable irrigation practices continue to accelerate soil degradation, posing a persistent challenge to global food security. As cultivation increasingly expands into marginal lands, understanding the molecular mechanisms that enable crops to tolerate saline–alkaline environments has become a central objective in plant science and crop improvement.

While plant responses to neutral salt stress, primarily NaCl, have been extensively characterized, natural soils frequently impose saline–alkaline stress, which involves both neutral and alkaline salts. This form of stress is often more detrimental than neutral salinity because it combines ionic toxicity with high pH–induced nutrient limitation [3]. Elevated soil pH restricts the availability of essential minerals, disrupts root development, and enhances oxidative stress through excessive accumulation of reactive oxygen species [4, 5]. Under these conditions, precise regulation of ion homeostasis becomes critical for plant survival. Maintenance of cytosolic K⁺/Na⁺ balance, efficient sequestration of Na⁺ into vacuoles, and selective uptake of essential nutrients are fundamental processes that preserve cellular integrity and metabolic function [6, 7]. However, how ionic imbalance under high-pH conditions is integrated into tissue-specific transcriptional networks remains poorly understood.

Sorghum bicolor (L.) Moench, a C₄ cereal crop, is widely recognized for its resilience to drought, high temperature, and moderate salinity [8]. As a staple crop in arid and semi-arid regions, sorghum represents both an important food and bioenergy resource and a valuable model for studying stress adaptation in monocots [9]. Despite its relative tolerance, the genetic architecture underlying sorghum adaptation to saline–alkaline stress—particularly the coordinated regulation of ion homeostasis under high pH conditions—remains incompletely understood. Clarifying these regulatory mechanisms will inform strategies for improving stress resilience not only in sorghum but also in other cereal crops.

Conventional differential gene expression analyses have identified numerous stress-responsive genes; however, such approaches often overlook higher-order regulatory relationships among genes. Because saline–alkaline tolerance is a complex, polygenic trait, systems-level analyses are required to uncover coordinated gene networks associated with physiological adaptation [10]. Weighted Gene Co-expression Network Analysis (WGCNA) provides a robust framework for clustering genes into modules based on shared expression patterns and linking these modules to quantitative traits. By identifying highly connected hub genes within trait-associated modules, WGCNA enables the prioritization of candidate genes that may be involved in stress responses [11]. Although network-based approaches have been applied to various abiotic stresses, an integrative analysis associating ion-related physiological traits with transcriptional networks in sorghum under saline–alkaline stress is still lacking. Particularly, how ion-associated modules integrate physiological variation across tissues remains unclear.

In this study, we combined comprehensive phenotypic profiling with tissue-specific transcriptomic analysis to investigate the regulatory networks underlying saline–alkaline tolerance in S. bicolor (cv. BTx623). Seedlings were exposed to Na₂CO₃-induced alkaline stress (50 mM; pH ≥ 10.5), and dynamic changes in 11 physiological traits in leaves and ten in roots—including ion accumulation, osmolyte content, chlorophyll levels, and antioxidant enzyme activities—were quantified over 72 h. RNA sequencing data from leaf and root tissues were integrated with these phenotypic measurements using WGCNA to construct co-expression networks. This integrative approach enabled the identification of ion-associated transcriptional modules that reflect tissue-specific physiological dynamics. By associating ion-associated transcriptional modules with quantitative physiological traits, this study provides an association-based systems-level framework for understanding saline–alkaline responses and prioritizes candidate genes for future functional validation and crop improvement.

Materials and methods

Plant materials and growth conditions

Seeds of S. bicolor (L.) Moench cv. BTx623 were obtained from the National Agrobiodiversity Center (Jeonju, Republic of Korea). Prior to sowing, seeds were surface-sterilized using 2% (v/v) sodium hypochlorite for 10 min and thoroughly rinsed with distilled water. The seeds were sown in seedling trays containing a mixture of horticultural and rice seedling substrates (7:3, v/v). The experiment was conducted during the summer of 2024 in a greenhouse at Chungnam National University, Daejeon, Republic of Korea (36° 22′ 08″ N, 127° 20′ 46″ E). At the four-leaf stage, uniform seedlings were selected and transplanted into hydroponic trays containing fine decomposed granite as a support medium. Each tray was supplied with 3 L of half-strength Hoagland’s nutrient solution, which was renewed every two days until the full emergence of the fifth leaf. The greenhouse temperature was maintained at 20–30 °C throughout the experiment.

Saline–alkaline stress treatment and sampling

Saline–alkaline stress treatment was initiated when the fifth leaf had fully emerged. Control plants were grown in half-strength Hoagland nutrient solution, while the treatment group was supplied with the same solution supplemented with 50 mM Na₂CO₃. The stress treatment was conducted for periods of 24 h and 72 h, respectively. To maintain stable stress conditions, nutrient solutions for both the control and treatment groups were renewed every 24 h. The pH of the treatment solution was continuously monitored using a pH meter (Model 8100, ETI, UK) and maintained at ≥ 10.5, while the pH of the control solution was adjusted to within a range of 6.0–8.0. The pH electrode was calibrated daily using standard buffer solutions. Leaf and root tissues were harvested at 24 h and 72 h after treatment initiation from six biological replicates. Samples intended for ion content analysis were oven-dried at 60 °C for 72 h, whereas samples for other analyses were immediately frozen in liquid nitrogen and stored at − 80 °C until use.

Determination of anthocyanin content

Fresh leaf and rood samples (0.2 g) were ground in liquid nitrogen and extracted with 1.5 mL of extraction buffer containing 45% methanol and 5% glacial acetic acid [12]. The homogenate was centrifuged at 12,000 × g for 5 min at 25 °C. The pellet was re-extracted under the same conditions, and the supernatants were pooled. Absorbance was measured at 530 and 657 nm using a PDA UV–Vis spectrophotometer (Nano-MD, Scinco, Republic of Korea). The total anthocyanin content was calculated using the following equation:

Determination of proline content

Fresh samples (50 mg) were ground in liquid nitrogen and homogenized in 1 mL of 40% (v/v) ethanol [13]. After overnight incubation at 4 °C, the homogenate was centrifuged at 14,000 × g for 5 min at 4 °C. The supernatant (500 µL) was mixed with 1 mL of reaction mixture consisting of 1% (w/v) ninhydrin, 60% (v/v) glacial acetic acid, and 20% (v/v) ethanol. The mixture was incubated at 95 °C for 20 min, cooled, and then centrifuged at 10,000 × g for 1 min. Absorbance was measured at 520 nm using a PDA UV–Vis spectrophotometer (Nano-MD, Scinco, Republic of Korea). Proline concentration was calculated based on a standard curve generated using L-proline (0.2–5 mM). The proline content was calculated using the following equations:

Determination of chlorophyll content

Fresh leaf samples (10 mg) were extracted in 700 µL of pre-heated dimethyl sulfoxide at 65 °C for 30 min [14]. The volume was adjusted to 1 mL with dimethyl sulfoxide. Absorbance of the extract was measured at 645 and 663 nm using a PDA UV–Vis spectrophotometer (Nano-MD, Scinco, Republic of Korea). Chlorophyll a, b, and total chlorophyll contents were calculated according to the following equations:

Determination of ion content

Plant samples were dried at 60 °C for three days and ground into a fine powder. Leaf and root samples (0.2 g) were digested with 10 mL of 69% (v/v) nitric acid using a microwave digestion system (MARS 6 iWAVE, CEM, Matthews, NC, USA). The digested samples were diluted to 50 mL with deionized water. The concentrations of Na, K, Mg, Fe, and Mn were determined using Inductively Coupled Plasma Mass Spectrometry (ICP-MS; iCAP PRO X Duo, Thermo Fisher Scientific, Waltham, MA, USA).

Antioxidant enzyme activities and total soluble protein

For protein extraction and enzyme assays, powdered leaf samples (0.2 g) were homogenized in 1.5 mL of 0.2 M potassium phosphate buffer (pH 7.8) containing 0.1 mM EDTA. The homogenate was centrifuged at 12,000 × g for 25 min at 4 °C, and the supernatant was used for subsequent assays.

Total soluble protein content was determined using the Bradford method [15]. The reaction mixture consisted of 2 µL of supernatant, 8 µL of distilled water, and 190 µL of Quick Start™ Bradford 1 × Dye Reagent (Bio-Rad, CA, USA). Absorbance was measured at 595 nm using a microplate spectrophotometer (Agilent Technologies, Inc., Santa Clara, CA, USA), and bovine serum albumin was used as a standard. The protein content was calculated as follows:

The activities of catalase (CAT), ascorbate peroxidase (APX), and peroxidase (POD) were determined using the protein extract. CAT activity was measured by monitoring H₂O₂ reaction at 240 nm (ε = 40 mM⁻¹·cm⁻¹) [16]. APX activity was determined by observing ascorbate oxidation at 290 nm (ε = 2.8 mM⁻¹·cm⁻¹) [17]. POD activity was assessed via pyrogallol oxidation at 420 nm (ε = 12 mM⁻¹·cm⁻¹) [18]. For all assays, changes in absorbance were monitored for 8 min at 10 s intervals using the microplate spectrophotometer. Specific enzyme activity was calculated using the following equations:

RNA sequencing and transcriptomic analysis

Total RNA was extracted from leaf and root tissues of control and saline–alkaline–treated plants (24 h and 72 h post-treatment) using TRIzol™ Reagent (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. Three independent biological replicates were used for each tissue and treatment condition. Sequencing libraries were prepared using the TruSeq Stranded mRNA Sample Prep Kit (Illumina, San Diego, CA, USA) and sequenced on an Illumina NovaSeq X platform to generate 150 bp paired-end reads (Table S1). Raw sequencing reads were assessed using FastQC (v0.12.1) and processed with Trimmomatic (v0.40) to remove adapter sequences and low-quality bases [19]. Clean reads were aligned to the S. bicolor reference genome (NCBI assembly GCF_000003195.3) using STAR (v2.7.11b) [20] with default parameters. Gene-level read counts were obtained using featureCounts (v2.0.6) [21] based on the corresponding genome annotation file. Differential expression analysis was conducted using DESeq2 (v1.46.0) in R [22]. p-values were adjusted for multiple testing using the Benjamini–Hochberg method, and genes with an adjusted p < 0.05 and absolute log₂ fold change ≥ 1 were considered differentially expressed genes (DEGs).

Weighted gene co-expression network analysis

Weighted gene co-expression network analysis (WGCNA) was performed for the leaf and root datasets, respectively, using the WGCNA package (v1.73) in R (v4.4.2) [11]. The soft-thresholding power (β) was selected to satisfy scale-free topology fit of R² > 0.75 (β = 11 for leaf; β = 15 for root), followed by the construction of a signed co-expression network. The relationship between modules and phenotypic traits was assessed through Pearson correlation analysis. Modules were considered significantly correlated with a trait if | r |> 0.6 and p < 0.05.

Identification of hub genes and network visualization

Candidate hub genes within key modules were identified based on module membership (MM) > 0.8 and absolute gene significance (|GS|) > 0.8. Genes were ranked according to intramodular connectivity (kWithin), and the top five genes with the highest kWithin values were defined as hub genes. Co-expression networks were visualized using Cytoscape (v3.10.4) [23]. For visualization, only edges with an adjacency value > 0.5 were retained.

Functional enrichment analysis

To investigate biological functions, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed using the Database for Annotation, Visualization and Integrated Discovery (DAVID) v2023q4 (https://davidbioinformatics.nih.gov/) [24]. The analysis was applied to the following gene sets: (i) time-specific and overlapping DEGs, and (ii) key modules identified via WGCNA. For all comparisons, GO terms and KEGG pathways with a false discovery rate (FDR) < 0.1 were considered significantly enriched.

Statistical analysis

Statistical analysis was performed using GraphPad Prism (v8.4.2). All experimental data are presented as mean ± standard error of the mean (SEM) (n = 6 biological replicates). Differences between treatment groups were evaluated using one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test. A value of p < 0.05 was considered statistically significant.

Results

Contrasting phenotypic and metabolic responses of leaves and roots to saline–alkaline stress

Exposure to saline–alkaline stress induced visible morphological deterioration in S. bicolor seedlings (Fig. 1a). Leaves exhibited curling, wilting, and pronounced chlorosis by 72 h, whereas roots displayed browning and reduced lateral root formation. Consistently, total chlorophyll content declined significantly from 1.32 ± 0.12 (control) to 0.37 ± 0.05 at 72 h (Fig. 1b).

Fig. 1.

Phenotypic and physiological responses of S. bicolor seedlings to saline–alkaline stress. Seedlings were exposed to 50 mM Na₂CO₃ treatment. a Representative photographs of seedlings under control (CT) conditions and at 24 h and 72 h post-treatment. Scale bar = 10 cm. b Total chlorophyll content in leaves. Proline content in leaves c and roots d. Anthocyanin content in leaves e and roots f. Data are presented as mean ± SEM. Different letters indicate statistically significant differences between time points (p < 0.05, Tukey’s test)

Proline and anthocyanin, major biomarkers of plant abiotic stress, showed contrasting tissue-specific responses (Fig. 1c–f). In leaves, both metabolites accumulated markedly: proline increased approximately 18-fold, from 1.74 ± 0.69 to 31.35 ± 2.73, and anthocyanin rose from 0.48 ± 0.12 to 2.27 ± 0.17 at 72 h. In contrast, both metabolites declined significantly in roots, with anthocyanin decreasing from 0.15 ± 0.01 to 0.07 ± 0.01 and proline from 2.30 ± 0.04 to 0.50 ± 0.20 over the same period.

Tissue-specific changes in measured ion traits under saline–alkaline stress

Under saline–alkaline stress, Na⁺ and K⁺ displayed distinct tissue-specific accumulation patterns (Fig. 2a, b, f, g). Na⁺ content increased substantially in both tissues. In leaves, Na⁺ rose from 0.04 ± 0.01 to 9.97 ± 0.07 at 72 h, representing an approximately 250-fold increase (Fig. 2a). In roots, Na⁺ increased from 1.64 ± 0.10 to 7.30 ± 0.20 over the same period (Fig. 2f). In contrast, K⁺ levels remained relatively stable in leaves (Fig. 2b), whereas root K⁺ declined significantly from 14.74 ± 0.22 to 4.49 ± 0.18 at 72 h (Fig. 2g).

Fig. 2.

Ion homeostasis responses of S. bicolor seedlings to saline–alkaline stress. Seedlings were exposed to 50 mM Na₂CO₃ treatment. Na⁺ (a, f), K⁺ (b, g), Mg²⁺ (c, h), total Fe (d, i), and total Mn (e, j) contents were measured in leaves (a–e) and roots (f–j) under control (CT) conditions and at 24 h and 72 h post-treatment. Data are presented as mean ± SEM. Different letters indicate statistically significant differences between time points (p < 0.05, Tukey’s test)

Saline–alkaline stress also influenced other mineral nutrients, including Mg²⁺, Fe, and Mn (Fig. 2c–e, h–j). Mg²⁺ content decreased slightly but significantly in both tissues (Fig. 2c, h). Total Fe content did not change significantly across time points in either tissue (Fig. 2d, i). In contrast, Mn exhibited tissue-specific responses, increasing in leaves at 72 h (Fig. 2e) but decreasing significantly in roots (Fig. 2j).

These Mg²⁺ and Mn responses indicate that Na₂CO₃ exposure disturbed mineral homeostasis beyond Na⁺/K⁺ balance. Because alkaline pH can reduce nutrient availability and alter ion uptake, these changes may reflect the combined effects of ionic toxicity and high-pH–mediated nutrient disturbance.

Tissue-specific activation of antioxidant defense systems under saline–alkaline stress

Antioxidant enzyme activities exhibited pronounced tissue-specific patterns (Fig. 3). Basal activities were consistently higher in roots than in leaves. CAT activity in leaves remained relatively stable throughout the treatment period (Fig. 3a), whereas root CAT activity increased from 3.62 ± 0.25 in the control to 10.84 ± 1.71 at 72 h (Fig. 3d), representing an approximately threefold increase.

Fig. 3.

Antioxidant enzyme activities in S. bicolor seedlings under saline–alkaline stress. Seedlings were exposed to 50 mM Na₂CO₃ treatment. CAT (a, d), APX (b, e), and POD (c, f) activities were measured in leaves (a–c) and roots (d–f) under control (CT) conditions and at 24 h and 72 h post-treatment. Data are presented as mean ± SEM. Different letters indicate statistically significant differences between time points (p < 0.05, Tukey’s test)

Both APX and POD activities increased in response to saline–alkaline stress in both tissues. Leaf APX activity reached 176.90 ± 19.60 at 72 h (Fig. 3b), while root APX activity increased to 765.40 ± 82.30 (Fig. 3e). Similarly, POD activity increased in both tissues, with leaf POD reaching 98.53 ± 2.61 (Fig. 3c) and root POD rising to 286.40 ± 26.44 at 72 h compared with the control (Fig. 3f).

Tissue-specific transcriptomic reorganization under saline–alkaline stress

Saline–alkaline stress triggered extensive and tissue-specific transcriptional reorganization in both leaves and roots of S. bicolor (Fig. 4). Stress-treated samples were clearly separated from controls along PC1 in both tissues (Fig. 4a, b). PC1 explained 43% of the variance in leaves and 55% in roots, indicating pronounced transcriptomic alterations under stress.

Fig. 4.

Global transcriptome landscape and identification of differentially expressed genes (DEGs) in sorghum leaves and roots under saline–alkaline stress. Principal component analysis (PCA) showing sample clustering in a leaves and b roots under control (CT), 24 h, and 72 h after Na₂CO₃ treatment. c Numbers of total (gray), up-regulated (red), and down-regulated (blue) DEGs at 24 h and 72 h compared with CT in leaves and roots. d, e Venn diagrams showing the overlap of up-regulated (upper panels) and down-regulated (lower panels) DEGs between 24 and 72 h in leaves (d) and roots (e). DEGs were defined as |log₂FoldChange|≥ 1 with adjusted p < 0.05

Differential expression analysis revealed distinct temporal response patterns between tissues (Fig. 4c). In leaves, 3,137 differentially expressed genes (DEGs) (1,328 up- and 1,809 down-regulated) were detected at 24 h, whereas only 1,352 DEGs (423 up- and 929 down-regulated) were identified at 72 h. In contrast, roots exhibited 4,270 and 4,279 DEGs at 24 h and 72 h, respectively, with consistently high numbers of up-regulated genes (> 2,300) at both time points.

Venn diagram analysis further highlighted tissue-specific temporal dynamics (Fig. 4d, e). In leaves, only 23.0% of up-regulated and 35.3% of down-regulated genes were shared between 24 and 72 h, whereas roots displayed greater overlap (46.1% of up-regulated genes shared between time points). Unsupervised hierarchical clustering supported these findings, clearly separating control and stress-treated samples in both tissues (Fig. S1a, b). Pearson correlation analysis further confirmed sample-to-sample consistency among RNA-seq samples in leaves and roots (Fig. S1c, d). Volcano plot analysis showed the distribution of up- and down-regulated DEGs across tissues and time points (Fig. S2).

Saline–alkaline stress triggers redox activation but suppresses photosynthesis and protein folding in leaves

To explore biological functions associated with transcriptional changes, GO and KEGG enrichment analyses were performed (Fig. 5; Fig. S3 and S4). Among DEGs commonly up-regulated at both 24 h and 72 h, redox-related terms, including glutathione metabolic process (GO:0006749) and oxidoreductase activity (GO:0016705), were significantly enriched (Fig. 5a). In contrast, 24 h-specific up-regulated DEGs were enriched in signaling and protein synthesis–related processes, such as receptor-mediated signaling and ribosome-associated functions (Fig. 5b). KEGG pathway analysis further identified glutathione metabolism and MAPK signaling pathway – plant as significantly enriched among up-regulated genes (Fig. S4a). Enrichment analysis of down-regulated DEGs revealed repression of photosynthesis-related processes (Fig. S3). DEGs commonly down-regulated at both time points were enriched in terms such as photosynthesis, light harvesting (GO:0009765), and starch biosynthetic process (GO:0019252). Notably, 72 h-specific down-regulated DEGs were enriched in protein folding and chaperone-related processes, including protein folding (GO:0006457) and ATP-dependent protein folding chaperone activity (GO:0140662). KEGG analysis similarly highlighted photosynthesis, photosynthesis – antenna proteins, carbon fixation by the Calvin cycle, and protein processing in the endoplasmic reticulum among down-regulated genes (Fig. S4b).

Fig. 5.

GO enrichment analysis of up-regulated DEGs in leaves under saline–alkaline stress. Bubble plots showing significantly enriched GO terms for a commonly up-regulated DEGs in both time points and b 24 h-specific up-regulated DEGs. No significantly enriched terms (FDR < 0.1) were detected for 72 h-specific up-regulated DEGs. The top 20 terms ranked by fold enrichment are shown. Bubble size represents gene count, and color indicates statistical significance (− log10 FDR)

Chromatin remodeling and autophagy characterize temporal stress responses in roots

To elucidate functional characteristics of transcriptional changes in roots, GO and KEGG enrichment analyses were performed (Fig. S5–S7). Among DEGs commonly up-regulated at both 24 h and 72 h, terms associated with oxidative stress and defense responses, including response to hydrogen peroxide (GO:0070301), chitin catabolic process (GO:0006032), and chitinase activity (GO:0004568), were significantly enriched (Fig. S5a).

The 24 h-specific up-regulated DEGs were enriched in histone kinase activities, including histone H2, H3, and H4 kinase activity (Fig. S5b), consistent with chromatin-associated regulatory processes. In contrast, 72 h-specific up-regulated DEGs were enriched in autophagy-related terms, such as mitochondrial autophagy (GO:0000422) and phagophore assembly site membrane (GO:0034045) (Fig. S5c).

KEGG pathway analysis supported these patterns (Fig. S7a). Propanoate metabolism, plant–pathogen interaction, and glutathione metabolism were enriched among commonly up-regulated genes, whereas autophagy – other and fatty acid degradation were specifically enriched at 72 h. Enrichment analysis of down-regulated DEGs revealed suppression of photosynthesis-related processes (Fig. S6). Genes commonly down-regulated at both time points were enriched in photosynthesis-associated terms. The 24 h-specific down-regulated DEGs were enriched in microtubule-based movement (GO:0007018) and microtubule motor activity (GO:0003777), whereas 72 h-specific down-regulated genes were associated with chromatin-related terms, including nucleosome (GO:0000786) and structural constituent of chromatin (GO:0030527) (Fig. S6). KEGG pathway analysis further highlighted enrichment of the motor proteins pathway at 24 h and photosynthesis-related pathways among commonly down-regulated genes (Fig. S7b).

Na⁺-associated endoplasmic reticulum protein folding module characterizes leaf stress responses

To identify coordinated transcriptional responses to saline–alkaline stress, WGCNA was performed separately for leaves and roots. In leaves, 18,019 filtered genes from nine samples were used to construct the network. A soft-thresholding power of β = 11 was selected to achieve scale-free topology (signed R² > 0.75; Fig. 6a). Hierarchical clustering followed by dynamic tree cutting identified 37 distinct co-expression modules (Fig. 6b).

Fig. 6.

Construction of weighted gene co-expression networks and module–trait relationships in leaves. a Analysis of scale-free topology fit index (signed R²) and mean connectivity across soft-thresholding powers (β). The red horizontal line indicates the scale-free topology threshold (R² = 0.75). b Hierarchical clustering dendrogram of genes based on topological overlap dissimilarity (1 − TOM). Modules are indicated by color bars below the dendrogram. c Heatmap showing Pearson correlation coefficients (r) between module eigengenes and physiological traits. The numbers in parentheses indicate the number of genes in each module. Only modules significantly correlated with at least one trait (|r|> 0.6, p < 0.05) are shown. Asterisks denote statistical significance (*p < 0.05, **p < 0.01, ***p < 0.001)

Module–trait relationship analysis revealed that the salmon module showed the strongest association with Na⁺ accumulation (Fig. 6c). This module exhibited a significant negative correlation with Na⁺ content (r = − 0.84, p = 0.005). The eigengene expression of the salmon module progressively decreased from control to 72 h, inversely mirroring the temporal increase in Na⁺ levels.

To characterize the biological functions represented in the salmon module, GO and KEGG enrichment analyses were performed (Fig. 7a, b). Genes within this module were significantly enriched in processes related to protein folding and endoplasmic reticulum (ER) function, including protein refolding, ATP-dependent protein folding chaperone activity, unfolded protein binding, and ER lumen localization. KEGG pathway analysis similarly identified protein processing in the endoplasmic reticulum as a significantly enriched pathway (Fig. 7b). Consistent with the decline in module eigengene expression (Fig. 6c), these results indicate reduced expression of ER-associated protein folding components under stress conditions.

Fig. 7.

Functional enrichment and hub gene characterization of the leaf salmon module. a GO enrichment analysis of genes in the salmon module. Bubble size represents gene count and color indicates statistical significance (− log10 FDR). b KEGG pathway enrichment analysis of the salmon module. c Relationship between module membership (MM) and gene significance (GS) for Na⁺ content within the salmon module. d Co-expression network of the top 50 candidate hub genes (MM > 0.8 and |GS|> 0.8). Nodes represent genes and edges indicate adjacency weights > 0.5; node size reflects intramodular connectivity. e Expression profiles of the top 20 candidate hub genes identified in the leaf salmon module. Heatmap colors represent log₂ fold-change values relative to the control

Hub gene analysis further strengthened this association. A strong positive correlation between module membership and gene significance for Na⁺ content was observed (cor = 0.72, p = 1.3 × 10⁻⁵⁸; Fig. 7c), indicating that genes most central to the module are also tightly linked to Na⁺ variation. The top 50 hub genes formed a densely interconnected network (Fig. 7d), including ER-associated chaperones such as Luminal-binding protein 2 (BiP2) and Activator of HSP90 ATPase homolog 1 (Table S2 and S3).

K⁺-associated plastid-related metabolic module characterizes root stress responses

In roots, 20,269 genes were subjected to WGCNA. A soft-thresholding power of β = 15 satisfied the scale-free topology criterion (signed R² > 0.75; Fig. 8a), resulting in the identification of 28 co-expression modules (Fig. 8b). Module–trait relationship analysis identified the brown module as the module most strongly correlated with K⁺ content (Fig. 8c). This module exhibited a robust positive correlation with K⁺ levels (r = 0.95, p = 0.0001). The eigengene expression of the brown module progressively decreased under stress, paralleling the observed reduction in root K⁺.

Fig. 8.

Construction of weighted gene co-expression networks and module–trait relationships in roots. a Scale-free topology fit index (signed R²) and mean connectivity across soft-thresholding powers (β). The red line indicates the threshold for scale-free topology (R² = 0.75). (b) Hierarchical clustering dendrogram of genes based on topological overlap dissimilarity (1 − TOM). Modules are indicated by the color bar below the dendrogram. (c) Heatmap showing Pearson correlation coefficients (r) between module eigengenes and physiological traits. Numbers in parentheses indicate the number of genes in each module. Asterisks denote statistical significance (*p < 0.05, **p < 0.01, ***p < 0.001)

Functional enrichment analysis revealed that genes within the brown module were significantly enriched in plastid- and metabolism-related categories (Fig. 9a). GO terms included photosynthesis, light harvesting, chloroplast thylakoid membrane, plastoglobule, and heme binding. KEGG pathway analysis further identified enrichment in photosynthesis–antenna proteins and broader metabolic pathways (Fig. 9b).

Fig. 9.

Functional enrichment and hub gene characterization of the root brown module. a GO enrichment analysis of genes in the brown module. Bubble size represents gene count and color indicates statistical significance (− log10 FDR). b KEGG pathway enrichment analysis of the brown module. c Relationship between module membership (MM) and gene significance (GS) for K⁺ content within the brown module. d Co-expression network of the top 50 candidate hub genes (MM > 0.8 and |GS|> 0.8). Nodes represent genes and edges indicate adjacency weights > 0.5; node size reflects intramodular connectivity. e Expression profiles of the top 20 candidate hub genes identified in the root brown module. Heatmap colors represent log₂ fold-change values relative to the control

Because roots are non-photosynthetic tissues, enrichment of terms such as “photosynthesis” and “photosynthesis–antenna proteins” should not be interpreted as evidence of photosynthetic activity in roots. Rather, these annotations indicate plastid-associated and light-harvesting complex–related genes that are categorized under photosynthesis-related pathways in public annotation databases.

Hub gene analysis demonstrated a strong correlation between module membership and gene significance for K⁺ content (cor = 0.89, p < 1 × 10⁻²⁰⁰; Fig. 9c), indicating that genes central to the module are tightly associated with K⁺ variation. The top 50 hub genes formed a highly interconnected network (Fig. 9d), including genes encoding membrane transporters and ion flux regulators such as Sodium/calcium exchanger NCL1 and Aquaporin PIP2-6 (Table S4 and S5).

Expression patterns of hub genes support coordinated module repression

Expression patterns of hub genes across tissues mirrored module-level dynamics under saline–alkaline stress (Figs. 7e, 9e). In leaves, down-regulated hub genes were primarily associated with ER-related protein processing, whereas in roots, down-regulated hub genes were enriched for plastid-associated metabolic and ion transport functions. These expression patterns were consistent with the corresponding module eigengene trajectories, indicating tissue-specific differences in transcriptional responses.

Discussion

Saline–alkaline stress imposes complex ionic, osmotic, and high-pH constraints that profoundly affect plant growth and metabolism [25]. In S. bicolor, stress exposure induced marked phenotypic deterioration, including leaf curling, chlorosis, and root browning (Fig. 1a), accompanied by progressive chlorophyll decline (Fig. 1b). These visible symptoms were paralleled by pronounced ionic imbalance (Fig. 2) and antioxidant reconfiguration (Fig. 3), establishing a physiological foundation for subsequent transcriptional reorganization. These coordinated physiological changes suggest that Na₂CO₃ stress does not produce a uniform stress syndrome but instead redirects resource allocation in a tissue-dependent manner.

Divergent ionic regulation strategies between leaves and roots under saline–alkaline stress

The accumulation of Na⁺ in both tissues (Fig. 2a, f), together with tissue-specific K⁺ dynamics, highlights differential ionic regulation strategies. While leaves maintained relatively stable K⁺ levels despite dramatic Na⁺ accumulation (Fig. 2b), roots exhibited substantial K⁺ depletion (Fig. 2g). Given the essential role of K⁺ in enzyme activation, osmotic balance, and membrane potential stabilization [6], its decline in roots likely represents a critical destabilizing factor contributing to metabolic restructuring. In contrast, the ability of leaves to preserve K⁺ homeostasis may partially buffer metabolic collapse despite Na⁺ influx. Thus, the central physiological contrast is not only Na⁺ accumulation but the ability or inability to maintain K⁺-dependent cellular functions under alkaline-salt exposure.

Metabolite responses further underscored tissue divergence

Proline and anthocyanin accumulated strongly in leaves but declined in roots (Fig. 1c–f). Proline functions as an osmoprotectant and redox buffer [26], and anthocyanins contribute to reactive oxygen species (ROS) scavenging and photoprotection [27]. Their accumulation in leaves likely reflects reinforcement of osmotic stabilization and oxidative buffering in photosynthetically active tissues facing Na⁺ influx. In contrast, the reduction of these metabolites in roots may indicate a strategic shift away from osmolyte accumulation toward ion transport and membrane stabilization processes, consistent with the pronounced K⁺ depletion observed in roots. Together, these findings suggest that leaves and roots deploy distinct metabolic priorities to cope with saline–alkaline stress. This opposite directionality suggests that leaves preferentially invest in compatible solute and pigment-based protection, whereas roots may prioritize maintenance of ion transport and cellular integrity over accumulation of these metabolites.

Antioxidant enzyme activities further reinforced this divergence

Basal CAT, APX, and POD activities were consistently higher in roots (Fig. 3d–f), and stress-induced increases were more pronounced compared with leaves (Fig. 3a–c). The sustained elevation of root APX and POD activities suggests that roots act as a major oxidative stress buffering site during prolonged saline–alkaline exposure [28]. In contrast, leaf CAT activity remained relatively stable, while APX and POD exhibited moderate increases, indicative of a more dynamic and possibly transient oxidative adjustment. Together, these patterns support the notion that roots maintain a sustained antioxidant defense program, whereas leaves rely on more flexible metabolic modulation under stress conditions. This difference is biologically meaningful because roots directly encounter the high-pH solution and therefore require sustained enzymatic detoxification, whereas leaves appear to combine redox adjustment with broader metabolic remodeling.

Ion-associated transcriptional reorganization under saline–alkaline conditions

Building upon the physiological divergence observed between leaves and roots, transcriptomic network analysis indicated that ionic imbalance is associated with coordinated co-expression patterns. Saline–alkaline stress induced pronounced and tissue-specific transcriptional reorganization in S. bicolor, with roots exhibiting broader and more sustained activation than leaves (Fig. 4a–e). Integration of ion measurements with co-expression network analysis indicated that ionic imbalance is associated with module-level transcriptional variation (Figs. 6c and 8c).

In leaves, a Na⁺-associated module enriched for ER-related protein folding components showed coordinated repression as Na⁺ accumulated (Figs. 6c and 7a, b). In roots, a K⁺-associated module enriched for plastid metabolic functions and ion transport processes declined in parallel with K⁺ depletion (Figs. 8c and 9a, b). The close alignment between module eigengene dynamics and physiological ion trends (Figs. 2a, g; 6c and 8c) supports a structured relationship between ionic status and transcriptional network behavior. Consistent with established systems-level stress responses, saline–alkaline stress was associated with trait-linked reconfiguration of transcriptional modules. Biologically, these modules connect the major ionic disturbances to cellular systems that determine stress tolerance: protein quality control in leaves and plastid-associated metabolism in roots.

It should be noted that coordinated transcriptional regulation is an established feature of plant responses to salinity and other abiotic stresses. Therefore, our findings should not be interpreted as evidence for a unique or entirely new regulatory principle specific to sorghum. Rather, this study provides additional systems-level evidence that, under Na₂CO₃-induced saline–alkaline stress, quantitative Na⁺ and K⁺ traits are associated with tissue-specific co-expression modules in leaves and roots. Thus, the main contribution of this study is the identification of ion-associated, tissue-specific modules in sorghum under saline–alkaline stress, rather than the demonstration that stress responses are network-based per se.

Importantly, saline–alkaline stress differs from neutral NaCl stress by imposing high-pH–mediated constraints on nutrient availability and mineral balance. Elevated soil pH reduces the availability of essential nutrients and alters ion equilibrium, thereby intensifying metabolic perturbation [29]. Although total Fe levels remained stable (Fig. 2d, i), Mg²⁺ reduction (Fig. 2c, h) and tissue-specific Mn redistribution (Fig. 2e, j) indicate mineral adjustments that may affect plastid function and redox homeostasis, given the central role of Mg²⁺ in chlorophyll structure and Rubisco activation [30] and Mn in photosystem II function [31]. These observations suggest that high-pH conditions amplify metabolic stress beyond sodium toxicity alone.

Accordingly, the Na⁺- and K⁺-associated modules identified here should be interpreted within this combined saline–alkaline context, in which ion accumulation, nutrient availability, and high-pH constraints may jointly shape transcriptional network behavior.

ER-associated repression and redox prioritization in leaves

The progressive decline in chlorophyll content (Fig. 1b) and sustained repression of photosynthesis-related genes (Fig. 5; Fig. S3, S4) suggest impairment of photosynthetic machinery. Concurrent enrichment of glutathione metabolism and oxidoreductase activity (Fig. 5a) indicates enhanced redox defense [32]. Notably, the Na⁺-associated co-expression module identified by WGCNA (Fig. 6c) was enriched for ER protein folding and chaperone-related genes (Fig. 7c, d), including luminal-binding proteins and HSP90-associated factors, whose coordinated down-regulation paralleled Na⁺ accumulation (Fig. 2a).

Salt stress commonly perturbs protein folding and activates unfolded protein response pathways [33]. However, the observed repression of ER-associated components suggests a transition from adaptive activation to energy-conserving suppression during prolonged stress. Given the high ATP demand of protein quality control systems [34], coordinated reduction of ER folding capacity may reflect metabolic prioritization toward redox maintenance and survival rather than growth. Thus, leaf responses appear to involve structured repression of biosynthetic machinery coupled with oxidative stabilization.

Under Na₂CO₃ treatment, high pH may further exacerbate ER- and redox-related stress by restricting nutrient availability and disrupting cellular homeostasis. Therefore, the leaf ER-associated module is unlikely to reflect Na⁺ accumulation alone but rather an integrated response to the combined ionic and alkaline environment. This interpretation helps explain why photosynthesis- and ER-related processes were simultaneously affected: reduced protein-folding capacity may constrain repair of photosynthetic and membrane-associated proteins under prolonged stress.

Plastid metabolic restructuring and K⁺ imbalance in roots

In roots, K⁺ depletion (Fig. 2g) coincided with repression of a plastid-associated metabolic module (Fig. 8c) enriched for light-harvesting and membrane-related components (Fig. 9a, b). Although roots are non-photosynthetic, plastids in heterotrophic tissues play essential roles in amino acid biosynthesis, fatty acid metabolism, and redox regulation [35, 36]. The strong positive correlation between module connectivity and K⁺ content (Fig. 9c) suggests that plastid metabolic stability is closely linked to ionic homeostasis.

In this context, the enrichment of photosynthesis-related terms in the root brown module likely reflects changes in plastid-associated metabolic programs rather than canonical photosynthesis. Root plastids, including proplastids and other non-green plastid types, retain important functions in amino acid biosynthesis, fatty acid metabolism, redox regulation, and ion-associated metabolic adjustment.

Likewise, in roots, high-pH–mediated mineral limitation may intensify K⁺ depletion and plastid metabolic instability. Thus, repression of the root plastid-associated module is best interpreted as a network response associated with combined K⁺ imbalance and alkaline-pH constraints. This provides a possible biological connection between K⁺ loss and root metabolic limitation, because non-green plastids support biosynthetic and redox functions required for root growth under stress.

K⁺ deficiency may impair enzymatic function and disrupt osmotic equilibrium [37], potentially destabilizing plastid metabolism. Given that plastid membrane integrity and multiple stromal enzymes are highly dependent on ionic balance and K⁺-mediated activation, a decline in cytosolic K⁺ may be associated with reduced plastid metabolic stability. The densely interconnected structure of the root module (Fig. 9d), including ion transporters such as NCL1 and aquaporin PIP2-6, supports the interpretation that K⁺ imbalance is associated with coordinated metabolic adjustment. Moreover, late-stage enrichment of autophagy-related processes (Fig. S5c; Fig. S7a) suggests activation of organelle quality control mechanisms to recycle damaged components during sustained stress [38]. Together with autophagy enrichment, this pattern suggests that prolonged root stress may shift from growth-related metabolism toward maintenance and recycling of damaged cellular components.

Coordinated repression as a systems-level survival strategy

Across tissues, a recurring theme was coordinated module-level repression associated with ionic imbalance. Hub gene expression trends closely mirrored eigengene dynamics (Figs. 7e and 9e), indicating robust network organization. Suppression of ER-associated processes in leaves and plastid-linked metabolic functions in roots is consistent with an energy allocation response under conditions of impaired photosynthesis and ionic disequilibrium [39]. Rather than relying solely on gene induction, saline–alkaline stress was associated with reshaping cellular priorities through structured down-regulation of energetically costly processes. Such repression should therefore be viewed not simply as transcriptional failure, but as a regulated reduction of biosynthetic demand under conditions where ion imbalance limits energy and metabolic stability.

Contribution of alkaline pH to the observed saline–alkaline stress responses

A key distinction between Na₂CO₃-induced saline–alkaline stress and neutral NaCl stress is that plants are simultaneously exposed to Na⁺ accumulation and alkaline-pH constraints [3, 25]. Some responses observed in this study, including Na⁺ accumulation, oxidative stress activation, repression of photosynthesis-related genes, and changes in K⁺ homeostasis, are likely shared with neutral salt stress [1, 6, 10]. Therefore, these responses should not be interpreted as exclusively alkaline-specific. However, the marked changes in Mg²⁺ and Mn contents, together with severe root K⁺ depletion, chlorophyll loss, and repression of plastid- and ER-associated modules, suggest that alkaline pH may have intensified the stress response by disturbing mineral availability, nutrient balance [5, 29], and cellular homeostasis beyond Na⁺ toxicity alone. In particular, the reduction of Mg²⁺ and the tissue-specific redistribution of Mn are consistent with high-pH–mediated nutrient disturbance, which may contribute to impaired photosynthetic capacity, redox imbalance, and plastid metabolic instability [30, 31]. Thus, the physiological and transcriptional responses observed here are best interpreted as integrated outcomes of Na⁺ toxicity and alkaline-pH stress rather than as responses to either component alone. Because the present experimental design did not include NaCl-only or pH-adjusted non-saline controls, the relative contribution of alkalinization cannot be quantitatively separated. Future comparative experiments using neutral salt, alkaline pH, and combined saline–alkaline treatments will be required to determine the extent to which the observed responses are specific to alkaline stress.

Implications and future directions

As sorghum is frequently cultivated in marginal lands characterized by saline–alkaline soils, elucidating the transcriptional architecture underlying ion homeostasis is essential for improving stress resilience. The identification of Na⁺- and K⁺-associated co-expression modules (Figs. 6, 7, 8 and 9) suggests that ion traits are associated with coordinated co-expression modules in this saline–alkaline stress context. Because WGCNA, enrichment, and co-expression network analyses are inherently correlative, these modules and hub genes should be interpreted as candidate ion-associated components rather than validated causal regulators. Although causal relationships remain to be experimentally validated, the strong module–trait correlations (Figs. 6c and 8c) together with consistent physiological alignment (Figs. 1, 2 and 3) underscore the biological relevance of these ion-linked transcriptional modules. Future integration of targeted genetic perturbation, comparative genomics, and natural variation analyses will help clarify how network-level ion regulation contributes to adaptive stress tolerance and may inform breeding strategies for saline–alkaline environments. Although cross-tissue DEG analysis could provide additional insight into constitutive and stress-modulated expression divergence between leaves and roots, such analysis would substantially expand the scope of the present study. Therefore, we focused primarily on within-tissue stress-responsive transcriptional dynamics and ion-associated co-expression modules, while considering comprehensive cross-tissue expression divergence as an important direction for future investigation. From an applied perspective, modules that integrate physiological ion traits with coordinated gene expression may provide more robust targets for stress-resilience screening than single differentially expressed genes.

Because the present design did not include NaCl-only or pH-only control treatments, the independent effects of ionic toxicity and high pH cannot be fully separated. We therefore interpret the physiological and transcriptional patterns as integrated outcomes of Na₂CO₃-induced saline–alkaline stress, and future experiments using NaCl-only, pH-adjusted non-saline, and combined treatments will be required to disentangle these components.

Conclusion

Saline–alkaline stress imposes a multidimensional challenge that extends beyond sodium toxicity, integrating ionic disequilibrium, high-pH–mediated mineral perturbation, and oxidative constraint (Fig. 10). Our integrative physiological and transcriptomic network analyses suggest that stress responses in Sorghum bicolor involve coordinated, ion-associated transcriptional modules operating in a tissue-dependent manner. Leaves and roots exhibited distinct response patterns. Leaves maintained relative K⁺ stability while undergoing coordinated repression of ER-associated protein folding machinery, accompanied by reinforcement of redox buffering systems. In contrast, roots experienced pronounced K⁺ depletion linked to plastid metabolic restructuring, sustained antioxidant activation, and activation of organelle quality control pathways. The strong alignment between module eigengene dynamics and Na⁺/K⁺ traits indicates that ionic status is closely associated with transcriptional network behavior under alkaline saline conditions. Rather than representing stochastic gene-level responses, saline–alkaline stress was accompanied by structured module-level reorganization, consistent with possible reallocation of metabolic resources and suppression of energetically costly biosynthetic processes. This coordinated repression may represent a systems-level stress response under conditions of impaired photosynthesis and ionic imbalance. Although functional validation of individual hub genes remains necessary, the identification of Na⁺- and K⁺-associated co-expression modules provides an association-based framework for linking ion homeostasis to transcriptional organization. By integrating ionic traits with network topology, this work advances our understanding of how crops coordinate tissue-specific metabolic prioritization in saline–alkaline environments and offers a conceptual basis for network-informed breeding strategies aimed at improving stress resilience.

Fig. 10.

Conceptual model of ion-associated transcriptional reorganization under saline–alkaline stress in S. bicolor

Abbreviations

ANOVA

Analysis of variance

APX

Ascorbate peroxidase

BiP

Binding immunoglobulin protein

CAT

Catalase

CT

Control

DAVID

Database for Annotation, Visualization and Integrated Discovery

DEG

Differentially expressed gene

ER

Endoplasmic reticulum

FDR

False discovery rate

FW

Fresh weight

GO

Gene Ontology

GS

Gene significance

ICP-MS

Inductively coupled plasma mass spectrometry

KEGG

Kyoto Encyclopedia of Genes and Genomes

MM

Module membership

PCA

Principal component analysis

POD

Peroxidase

ROS

Reactive oxygen species

SEM

Standard error of the mean

TOM

Topological overlap matrix

WGCNA

Weighted gene co-expression network analysis

Authors’ contributions

S Kim, D Jeon and C Kim conceptualized the study. S Kim performed the analysis and S Kim, and D Jeon wrote the original manuscript. C Kim provided supervision and reviewed and edited the manuscript for improvement.

Funding

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Ministry of Science and ICT (MSIT), Republic of Korea (RS-2025–24535664).

Data availability

The RNA-seq datasets generated and analyzed during the current study have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject accession number PRJNA1429075.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.