Transcriptomic insights into drought response in wild Arachis relatives A. dardani and A. ipaënsis

Abstract

Drought is a major environmental constraint limiting global peanut productivity. Wild peanut species, characterized by greater genetic diversity, represent valuable resources for improving drought resilience in cultivated peanut. However, the molecular mechanisms underpinning drought tolerance in wild peanut species remain largely unexplored. This study evaluated the drought tolerance of three wild-type peanut accessions from two different species, Arachis dardani GK12946, Arachis dardani V7215, and Arachis ipaënsis K30076. Physiological measurements such as fresh weight and dry weight revealed statistically non-significant differences between drought-stressed and well-watered conditions, indicating strong inherent drought tolerance. Transcriptome analysis revealed that 3272, 3648, and 1181 genes in leaf samples of A. dardani GK12946, A. dardani V7215, and A. ipaënsis K30076 were differentially expressed, respectively. In root samples, 3014, 3472, and 2033 genes were differentially expressed in the same accessions. Notably, differentially expressed genes (DEGs) and set intersection (Venn) analysis suggests A. dardani V7215 exhibited the highest number of DEGs (1155) uniquely expressed in leaves, and 899 DEGs uniquely expressed in roots, suggesting accession-specific gene expression. Gene Ontology enrichment revealed that upregulated genes were associated with abiotic stress responses, temperature stimulus, heat stress, and DNA-binding transcription factor activity. Co-expression network analysis using WGCNA identified key drought-responsive modules, enriched for GO terms like stress regulation, protein folding, as well as GST family amino acid metabolic processes. Overall, this study provides comprehensive insights into the molecular basis of drought tolerance in wild peanut accessions. Our findings establish a valuable resource for functional genomics and crop improvement under water-limited conditions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-07799-7.

Introduction

Global peanut productivity is continuously threatened by the escalating impacts of climate change, characterized by increased frequency and intensity of abiotic stresses, particularly drought [1]. These environmental stressors severely compromise crop quality and yields, thus threatening food security worldwide. Peanuts serve as a crucial protein source in arid and tropical regions of Asia and Africa, where legumes are a primary dietary protein source. The world’s leading peanut producers are China (38%), India (13%), Nigeria (9%), the United States (6%), and Senegal (3%), contributing significantly to global peanut production. The world’s peanut production in 2024 was 50.597 million tonnes. In the United States, peanut production reached 2.925 million tonnes in 2024. The USA is in 4th place with a 6% share of the worldwide production of peanuts based on harvested quantity [2]. Over the last decade, the peanut production per hectare has fluctuated widely, primarily due to abiotic stresses. In the 2025 growing season, as an estimate, 35% of peanut-producing areas in Texas, USA experienced drought conditions. Among these 35% areas, about 69% are experiencing severe drought, while 13% are experiencing exceptional drought at some point [3]. Thus, in these areas, the peanut production can be negatively affected. Modern-day cultivated peanuts (Arachis hypogea L.) are vulnerable to several abiotic stresses, including drought, which significantly impacts their yield and quality [1]. Therefore, developing drought-tolerant peanut varieties is crucial for sustainable production and growing demand in both the USA and worldwide. In the development of new drought tolerance cultivars, the wild relatives of cultivated peanuts, such as Arachis dardani, could serve as a valuable reservoir of genetic diversity for peanut crop improvement [4]. These wild species have evolved under diverse and often challenging environmental conditions, accumulating adaptive traits including enhanced drought tolerance [5]. Unlike cultivated peanut (A. hypogaea), which is an allotetraploid (2n = 4x = 40) and has undergone a significant genetic bottleneck due to the random doubling event. Wild diploid (n = 2x = 20) Arachis species, such as A. dardani and A. ipaënsis (B genome donor for A. hypogaea), have a broader genetic diversity, providing a rich source of stress tolerance genes and alleles that confer resilience to abiotic stresses, including drought stress [6]. Moreover, due to the complex taxonomy and ploidy differences within the Arachis genus, introgression of beneficial alleles into cultivated peanut requires pre-breeding strategies such as interspecific hybridization followed by chromosome doubling to restore fertility and facilitate gene transfer.

To determine the value of this investment, the initial step is identifying drought-tolerance genes in source materials. Identifying and characterizing drought-responsive genes in wild Arachis species can provide valuable insights into the mechanisms of drought tolerance and facilitate the development of improved peanut cultivars. This may be of particular importance because previous efforts at introducing drought tolerance from diploid wild species into the cultivated tetraploid species were not successful [7]. It has been observed that the combination of two or more different species’ genomes to form the polyploid has effects on gene expression that change the phenotype from what may have been expected [8–10]. Under such circumstances, having identified genes for drought tolerance may provide markers for selecting in the progeny of crosses between synthetic amphidiploids (generated to introduce drought-tolerant alleles) and cultivated parents.

Recent advancements in genomics and transcriptomics have enabled comprehensive analysis of gene expression and genome organization in plants under stress conditions. These technologies have facilitated the identification of numerous drought-responsive genes in various crops, including peanuts. For example, studies have revealed the involvement of LEA proteins, heat shock proteins, and aquaporins in drought tolerance in cultivated peanuts [1, 11, 12]. However, the genetic basis of drought tolerance in wild Arachis species remains largely unexplored, even in more closely related species such as A. ipaënsis. Transcriptomics has proven effective in pinpointing differentially expressed genes (DEGs) under stress conditions, with previous studies identifying genes related to salt stress [13, 14], pod development [15], and drought [5, 16–22]. As of now, most of the transcriptomic studies have been done on the cultivated peanuts only; therefore, understanding the molecular mechanisms underlying drought tolerance traits in wild species such as A. dardani and A. ipaënsis can contribute to developing molecular markers for marker-assisted selection and identifying candidate genes for genetic engineering. This study aims to discover drought-tolerance genes from A. dardani and A. ipaënsis. We hypothesize that these wild species harbour unique genes contributing to their enhanced drought tolerance [6, 23]. By employing the comparative transcriptomics approaches, we seek to identify these genes, providing valuable resources for peanut breeding programs to develop drought-resilient cultivars. This research will contribute to a deeper understanding of the genetic mechanisms underlying drought tolerance in Arachis species and pave the way for the development of sustainable agricultural practices in water-limited environments.

Materials and methods

Plant material and growth conditions

The wild-species Arachis dardani GK12946, Arachis dardani V7215 were selected due to their putative drought tolerance [24]. A. dardani GK12946 was collected near the Brazilian coast, while A. dardani V7215 was collected about 650 km inland, both in the dry Northeastern tip of Brazil [6, 23]. Thus, A. dardani V7215 is expected to be the most adapted to drought, followed by A. dardani GK12946 due to its proximity to the coast. A diploid relative of these, A. ipaënsis K30076, was also selected as it is donor species of the B genome to cultivated peanut (A. hypogaea) [25] and has a published reference genome [26]. A. ipaënsis K30076 inclusion not intended as an experimental control but rather as a reference genotype served two purposes such as to provide a comparative reference for gene expression patterns in the context of peanut domestication, and to examine whether its drought response was less pronounced or fundamentally different due to its narrower adaptation. The plants were grown, drought-treated, and sample collection was performed according to previous research by Cason et. al. [6]. Briefly, seeds of A. dardani GK12946, A. dardani V7215, and A. ipaënsis K30076 were surface-sterilized with 1% sodium hypochlorite and grown in 10-inch plastic pots under controlled conditions at 28 °C, 16-hour light/8-hour dark photoperiod. The seedlings were grown in a randomized block design with four biological replicates of each accession. Drought was imposed at about day 75 and continued until at least a 10% relative water content in leaf was achieved. To determine the reduction in RWC, the fourth expanded tetrafoliate was collected from the stipule of each plant in both control and drought-stressed conditions. Samples were immediately weighed to record the fresh weight (FW), then immersed in water for 24 h and reweighed to obtain the turgid weight (TW). Subsequently, the samples were placed in a Blue M dryer (General Signal, Garland, Texas) at 37 °C for 7 days and weighed again to obtain the dry weight (DW). Relative water content was calculated using the formula %RWC = [(FW-DW)/(TW-DW)] *100 [27], where FW is the fresh weight, DW is the dry weight, and TW is the turgid weight. The collection of samples occurred after a drought had been imposed for 7 days. The samples were collected for root and leaf tissues of well-watered (W) and drought-imposed (D) plants. Tissue was collected and immediately flash-frozen in liquid nitrogen and stored at -80 °C for RNA extraction. At the same time, we measured the fresh weight and dry weight of both the large upper mainstem leaves and the smaller lateral end leaves. Fresh weight was recorded immediately after harvesting, and to determine dry weight, the leaves were dried for seven days until a constant weight was observed, indicating complete moisture loss. Statistical analyses were performed in R using two-tailed pairwise t-tests comparing drought-treated and well-watered samples within each genotype, with P-values adjusted using the BH (Benjamini-Hochberg (FDR)) method.

RNA extraction and sequencing

Total RNA was extracted from leaf samples using the Spectrum™ Plant Total RNA Kit (Sigma-Aldrich, Inc., USA) according to the manufacturer’s instructions. RNA quantity and quality were assessed using Nanodrop spectrophotometer and Agilent Bioanalyzer. RNA samples with RIN values ≥ 8 were used for library preparation. RNA sequencing libraries were prepared using the Illumina TruSeq RNA Library Prep Kit and sequenced on Illumina HiSeq platform at the Texas A&M Genomics & Bioinformatics Service.

Transcriptome data processing and analysis

Raw sequencing reads were subjected to quality control using FastQC [28]. Adapter sequences were trimmed using Trimmomatic, where required [29]. High-quality reads were then aligned to the reference genome of A. ipaënsis K30076 [26] downloaded from PeanutBase using HISAT2 [30]. Gene expression levels were quantified using HTSeq [31]. Differential gene expression analysis was performed using DESeq2 in the R environment [32]. Genes with a false discovery rate (FDR) ≤ 0.05 and absolute log2 fold change ≥ 2 were considered as DEGs. These stringent criteria ensured the selection of biologically relevant DEGs, offering robust insights into gene expression patterns. The resulting DEGs were visualized using a volcano plot using the EnhancedVolcano R package (Blighe et al., 2025) to highlight expression changes, and a Venn diagram was created using the ggvenn R package [33] to identify shared and unique genes involved in the resistance pathway.

Gene ontology (GO) enrichment analysis

Gene Ontology (GO) terms for the identified DEGs from all four comparisons were annotated using the clusterProfiler R package [34–36], utilizing the custom A. ipaënsis K30076 GO database created through the AnnotationDbi R package [37]. GO terms were considered significantly enriched if they had a p-value ≤ 0.05. The enriched GO terms were visualized as a dot plot using the dot-plot function of the clusterProfiler R package.

Gene Co-expression analysis

Variance Stabilizing Transformation (VST) normalized data were used for the co-expression analysis to further investigate gene relationships by considering all DEGs across 3 accessions with log2Fc ≥ ± 2, p ≤ 0.05. Further, the gene clustering was performed by the Weighted Gene Co-expression Network Analysis (WGCNA) [38]. The genes from selected co-expression modules were correlated with known pathway genes in the R environment to provide a more mechanistic view of how the observed transcriptional regulation relates to established drought-response pathways. Moreover, these genes were functionally annotated by orthology-based mapping. Protein sequences from Arachis ipaënsis K30076 were annotated using Mercator4 (https://www.plabipd.de/portal/mercator4) to assign MapMan BIN categories. The resulting annotation files were imported into MapMan version 3.6.0RC1 [39] to visualize regulatory pathways and functional categories enriched among upregulated genes. Pathways highlighted included transcription factors, protein modification and degradation, hormone metabolism, signal transduction, carbon metabolism, redox processes, and light signalling.

Gene regulatory network inference

Gene regulatory networks (GRNs) were constructed using the GENIE3 algorithm [40] implemented in R, employing the random forest approach with default settings. While GENIE3 can be computationally intensive, it has been recognized for its consistently high performance in benchmarking studies across various sequencing platforms [41, 42]. Transcription factors (TFs) identified in DEGs, which were obtained from the PlantTFDB database (https://planttfdb.gao-lab.org/index.php? sp=Aip) and treated as candidate regulators within the expression dataset of all DEGs. Regulatory links were ranked by their associated importance scores (≥ 0), which reflect the strength of predicted regulatory influence. From this ranked list, the top 5% of interactions were selected for visualization. To identify key regulatory hubs, the ten transcription factors with the highest cumulative importance scores were considered primary regulators under each condition. The resulting networks were visualized using the ggraph package in R.

Gene selection

To identify candidate genes associated with drought tolerance, a multi-tiered selection strategy was employed by integrating various computational analyses. Initially, DEGs were identified under drought conditions using statistical thresholds based on fold change and adjusted p-values. These DEGs served as the foundation for further analysis. Next, weighted gene co-expression network analysis was performed to cluster genes into modules with similar expression patterns, providing insights into potential functional groupings. In parallel, gene regulatory networks (GRNs) were inferred using the GENIE3 algorithm to predict transcription factor-target interactions. This allowed for the identification of key regulatory genes likely to control drought-responsive pathways. In addition, Gene Ontology (GO) enrichment analysis was conducted to functionally categorize the DEGs. This analysis helped pinpoint biological processes, molecular functions, and cellular components significantly associated with drought response, further supporting candidate gene selection. Genes were prioritized for further validation based on a combination of the following criteria: (i) strong differential expression under drought stress; (ii) membership in co-expression modules enriched in drought-related GO terms; (iii) centrality and influence within the GRNs; and (iv) functional annotation indicating a putative role in stress signalling, transcriptional regulation, or other drought-associated mechanisms. This integrative approach enabled the identification of high-confidence candidate genes for future experimental validation and functional characterization.

Results

Drought tolerance in the wild-type peanut accessions

In earlier studies, two wild accessions were recognized for their superior drought tolerance [6, 23]. Building on these findings, we conducted a detailed analysis of three wild accessions, A. dardani GK12946, A. dardani V7215, and A. ipaënsis K30076. Fresh and dry weights of the large upper mainstem leaf and the smaller lateral end leaf revealed that these accessions maintained stable performance under drought conditions, showing no statistically significant differences (adj_p >0.05) in fresh and dry weight compared to well-watered conditions (Fig. S1A-D). These findings indicate that leaf biomass accumulation was largely unaffected by the drought treatment in all accessions showing higher drought tolerance.

Transcriptome analysis highlights the accession-specific gene expression patterns

Leaf and root tissues were collected for RNA sequencing to investigate drought-responsive gene expression in the wild-type peanut accessions A. dardani GK12946, A. dardani V7215, and A. ipaënsis K30076. After rigorous quality control, including the removal of low-quality reads, adapter sequences, and rRNA contaminants, the RNA-Seq analysis generated 3006.5 million reads across 48 samples (3 peanut accessions X 2 water treatments X 2 tissues X 4 reps), averaging 62.64 million reads per sample. The raw sequencing data have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject PRJNA1261117. High-quality reads were aligned to the A. ipaënsis K30076 reference genome using HiSAT2, achieving alignment rates up to 94.89%. Transcript abundance was quantified by mapping uniquely aligned reads to the reference annotation file (araip.K30076.gnm1.ann1.J37m.gene_models_main.gff3) using HTSeq-count. Of the 41,839 annotated genes, 18,782 genes were detected in leaf samples and 20,234 genes in root samples. To maximize the detection power of potential candidate genes [43], we filtered out genes that did not have at least 10 counts in a minimum set of 8 samples (Table S1) [44]. Two drought-stress samples (K30076_D_L_1 and K30076_D_R_1) were excluded from further analyses due to evidence of cross-contamination with well-watered samples. Principal component analysis (PCA) on VST normalized data, performed using the DESeq2 package, confirmed that replicates of each sample type clustered tightly, indicating low variability within groups (Fig. 1A-B).

Additionally, PCA revealed that A. dardani GK12946 and A. dardani V7215 clustered closely, suggesting a high degree of genetic similarity between them, while A. ipaënsis K30076 formed a distinct cluster. Treatment effects were evident, as drought and well-watered samples separated distinctly along the principal components, highlighting treatment-specific shifts in gene expression (Fig. 1A-B). Additionally, the hierarchical clustering of all samples based on global gene expression suggests that all biological replicates cluster closely within their respective sample groups, and clear separation is observed between treatment conditions. Leaf samples from both A. dardani GK12946 and A. dardani V7215 cluster together, while samples from A. ipaënsis K30076 form a distinct group (Fig. S2A). A similar clustering pattern was also observed in root samples (Fig. S2B). These results indicated that both A. dardani accessions are closer to each other; only their adaptation in different environments may affect some changes in gene expression.

Fig. 1.

Principal component analysis (PCA) and differential gene expression analysis. PCA biplot of (A) leaf samples and (B) root samples, illustrating the separation of samples based on treatment and cultivar. Each point represents a biological replicate (in = 4 per group), and samples are labeled by accession and treatment group. The analysis includes two cultivars of A. dardani GK12946 and A. dardani V7215 and one cultivar of A. ipaënsis K30076. Differentially expressed genes (DEGs) identified in (C) leaf samples and (D) root samples. Differential expressions were determined by separately comparing drought-treated versus watered samples within each cultivar and tissue type. Genes with a log2 fold change ≥ ±2 and FDR ≤ 0.05 were considered significantly up- or downregulated in drought vs well-watered conditions

Differential gene expression highlights drought-responsive genes

Differential gene expression analysis identified a large number of drought-responsive genes across all three wild-type peanut accessions. Genes with a log2 fold change ≥ ± 2 and a FDR ≤ 0.05 were considered significantly up- or downregulated between drought and well-watered conditions (Fig. S3). These stringent criteria help to identify genes that are not only statistically significant but also show a substantial change in expression, indicating a more meaningful biological difference.

In leaf samples, A. dardani GK12946, a total of 3,272 DEGs were identified, with 1,283 upregulated and 1,989 downregulated under drought stress. Similarly, in A. dardani V7215 exhibited 3,648 DEGs, of which 1,272 were upregulated and 2,376 were downregulated in drought stress. In A. ipaënsis K30076 leaves, 1,181 DEGs were detected, including 442 upregulated and 739 downregulated genes under drought stress (Fig. 1C, Table S2). In root samples, A. dardani GK12946 showed 3,014 DEGs, with 693 genes upregulated and 2,321 downregulated during drought. In A. dardani V7215 roots, 3,472 genes were differentially expressed, including 946 upregulated and 2,526 downregulated. A. ipaënsis K30076 roots exhibited 2,033 DEGs, with 534 genes upregulated and 1,499 downregulated under drought stress (Fig. 1D, Table S2). We found that the number of DEGs varied widely among the genotypes analyzed. This variation suggests that genotypes respond differently at the transcriptional level to the treatment. Such differences could arise from inherent variation in baseline gene expression or distinct sensitivities to the applied stress. These results highlight that the extent of transcriptomic remodelling is genotype dependent.

To further understand the functional categories of drought-induced genes, we conducted Gene Ontology (GO) enrichment analysis on the upregulated genes from both tissues. In A. dardani GK12946 leaves, enriched GO terms were associated with responses to abiotic stress (GO:0009628), response to temperature stimulus (GO:0009266), response to heat (GO:0009408), cell recognition (GO:0008037), recognition of pollen (GO:0048544), as well as homeostasis, like cellular homeostasis (GO:0019725), homeostatic process (GO:0042592), cell redox homeostasis (GO:0045454) were significantly enriched in upregulated DEGs (Fig. 2A, Table S3). Similarly, in A. dardani V7215 leaves, GO terms like response to abiotic stimulus (GO:0009628), response to temperature stimulus (GO:0009266), response to heat (GO:0009408), cell recognition (GO:0008037), as well as cellular homeostasis (GO:0019725) were significantly enriched (Fig. 2B, Table S3). In A. ipaënsis K30076 leaves, GO enrichment analysis highlighted terms associated with response to abiotic stimulus (GO:0009628), response to temperature stimulus (GO:0009266), and response to heat (GO:0009408) were significantly enriched (Fig. 2C, Table S3). A similar GO enrichment analysis was performed on root tissues revealed common enrichment of terms such as response to abiotic stimulus (GO:0009628), response to temperature stimulus (GO:0009266), response to heat (GO:0009408), protein maturation (GO:0051604), protein folding (GO:0006457) which were significantly enriched in the A. dardani GK12946 and A. dardani V7215 (Fig. 2DE, Table S4). While in A. ipaënsis K30076 the GO terms such as glutamine family amino acid metabolic process (GO:0009064), response to oxygen-containing compound (GO:1901700), glutamine family amino acid (GO:0009084) were significantly enriched along with response to abiotic stimulus (GO:0009628), response to temperature stimulus (GO:0009266), response to heat (GO:0009408) (Fig. 2F, Table S4). Thus, along with a common response to abiotic stimuli, temperature stimuli, and heat, A. ipaënsis K30076 harbours different mechanisms for drought tolerance, which include redox balancing rather than protein maturation (GO:0051604) and protein folding (GO:0006457) in A. dardani GK12946 and A. dardani V7215.

Fig. 2.

Gene Ontology (GO) enrichment analysis of differentially upregulated genes. GO enrichment in (A) Arachis dardani GK12946 leaf, (B) Arachis dardani V7215 leaf, (C) Arachis ipaënsis K30076 leaf, (D) Arachis dardani GK12946 root, (E) Arachis dardani V7215 root, and (F) Arachis ipaënsis K30076 root. The top ten significantly enriched GO terms from all categories of biological processes, molecular functions, and cellular components are presented. Enrichment analysis was performed using the Cluster Profiler R package to identify GO terms overrepresented among upregulated genes compared to the background gene set, with significance determined by a false discovery rate (FDR) cutoff of < 0.05. The point color corresponds to the significance level, and the size corresponds to the gene count of a particular enriched GO term

Additionally, GO enrichment analysis of the downregulated genes in leaves indicated a suppression of processes associated with regular cellular metabolism, such as GO terms like microtubule-based process and movement, cell cycle, and DNA replication significantly enriched across all accessions (Fig. S4A-C, Table S4). Similarly, in root samples the GO terms like cell wall organization or biogenesis, cell wall organization, external encapsulating structure organization were significantly enriched in downregulated DEGs across all accessions (Fig. S4D-F, Table S4).

Conserved drought-responsive genes identified by venn diagram analysis

To further identify key genes contributing to drought tolerance, we performed Venn diagram analyses to examine overlapping DEGs among the three accessions (Fig. 3). The analysis revealed 774 overlapping common genes in leaf (Fig. 3A); among them 282 genes were upregulated (Fig. 3B) while 490 were downregulated (Fig. 3C). Similarly in the root samples 1117 genes were overlapping and found commonly across all accession (Fig. 3D); among them 289 genes were upregulated (Fig. 3E) while 819 genes were downregulated (Fig. 3F). There were 1155, 785 and 237 unique genes that differentially expressed in the A. dardani V7215, A. dardani GK12946, and A. ipaënsis K30076 leaf tissue. Among them, 408, 404, and 109 genes were differentially upregulated, while 760, 394, and 152 genes were downregulated, respectively, in the A. dardani V7215, A. dardani GK12946, and A. ipaënsis K30076 leaf tissue. The number of DEGs uniquely expressed in the root was 899, 540, and 533, among them 337, 116, and 165 genes were upregulated, while 580, 441, and 408 genes were downregulated in the A. dardani V7215, A. dardani GK12946, and A. ipaënsis K30076 root tissue. Gene Ontology (GO) enrichment analysis of the commonly upregulated genes indicated significant enrichment in terms related to response to abiotic stimuli (GO:0009628), response to temperature stimuli (GO:0009266), response to heat (GO:0009408), in both leaf and root tissues (Fig. 4G-H). The uniquely differentially upregulated genes might show the accession-specific mechanism; therefore, we performed GO enrichment of uniquely upregulated DEGs to find the unique mechanism. Although not significant GO term enrichment the uniquely upregulated genes of A. dardani GK12946 were associated with GO terms like monoatomic ion transport (GO:0006811), monoatomic cation transport (GO:0006812), and metal ion transport (GO:0030001) (Fig. S5A). While in A. dardani V7215, the GO terms like cell recognition (GO:0008037), recognition of pollen (GO:0048544), phenylpropanoid metabolic process (GO:0009698), lignin metabolic process (GO:0009808), phenylpropanoid catabolic process (GO:0046271), lignin catabolic process (GO:0046274) and secondary metabolic process (GO:0019748) were significantly enriched (Fig. S5B). Similarly, in A. ipaënsis K30076, the upregulated genes were associated with response to biotic stimulus (GO:0009607), defense response (GO:0006952) (Fig. S5C).

Fig. 3.

Venn diagrams and Gene Ontology (GO) enrichment analysis of core differentially expressed genes (DEGs) under drought versus watered conditions. (A-C) leaf samples and (D-F) root samples. (A, D) All DEGs, (B, E) upregulated DEGs, and (C, F) downregulated DEGs, with comparisons performed between drought-treated and well-watered plants within each cultivar and tissue type. Overlapping regions in the Venn diagrams indicate genes commonly differentially expressed across all three cultivars. GO enrichment analysis for genes commonly upregulated in (G) leaf and (H) root samples. The top ten significantly enriched GO terms were identified across the categories of biological process, molecular function, and cellular component. Enrichment analysis was conducted using the clusterProfiler R package with an FDR threshold of < 0.05. Dot size reflects the number of genes associated with each GO term, and color indicates the adjusted p-value

Fig. 4.

Gene co-expression and Gene Ontology (GO) enrichment analysis of drought-induced gene modules. Heatmap of drought-induced modules of A leaf and B root samples. The GO enrichment analysis of the modules C leaf and D root samples. The top ten significantly enriched GO terms from biological processes are presented. Enrichment analysis was performed using the Cluster Profiler R package to identify GO terms overrepresented among upregulated genes compared to the background gene set, with significance determined by a false discovery rate (FDR) cutoff of ≤ 0.05. The point color corresponds to the significance level, and the size corresponds to the gene count of a particular enriched GO term

At the same time, in A. dardani GK12946 root samples, genes associated with GO terms like carbohydrate biosynthetic process (GO:0016051), protein folding (GO:0006457), protein maturation (GO:0051604), and organophosphate metabolic process (GO:0019637) (Fig. S5D). In A. dardani V7215, GO terms like cellular component organization (GO:0016043), protein folding (GO:0006457), and protein maturation (GO:0051604) were enriched, which suggests the unique response of A. dardani V7215 under drought stress (Fig. S5E, Table S5). Similarly, in A. ipaënsis K30076, terms like vesicle-mediated transport (GO:0016192), exocytosis (GO:0006887), secretion by cell (GO:0032940), and secretion complex (GO:0046903) were significantly enriched (Fig. S5F, Table S5). These overlapping gene sets represent a conserved core drought response shared across the peanut accessions, likely reflecting essential mechanisms such as stress sensing, signalling, and heat tolerance that are critical for maintaining cellular homeostasis under water deficit conditions. The GO enrichment of commonly upregulated genes in abiotic and temperature stress responses supports this notion, indicating that these core pathways are universally activated during drought. Conversely, the uniquely expressed DEGs reveal genotype-specific adaptations. For example, enhanced ion transport in A. dardani GK12946 may contribute to osmotic balance, while phenylpropanoid and lignin metabolism in A. dardani V7215 suggest structural reinforcement as a drought coping strategy. These findings illustrate how both shared and distinct molecular programs work together to shape drought tolerance in different peanut genotypes, providing a comprehensive view of the genetic basis underlying phenotypic variation in stress resilience.

Key co-expression modules reveal biological processes underlying drought tolerance

To explore gene Co-expression patterns under drought stress, we analyzed gene expression profiles based on their coordinated expression behaviour. We normalized filtered counts using VST from the fitted dispersion-mean relations and identified 4,752 genes in leaf samples, while 4,687 genes in root samples were significantly differentially expressed in all three accessions under drought conditions. Hierarchical clustering of these genes revealed that the samples are clustered in two major groups. The 1 st major group suggests that the drought-stressed samples of A. dardani GK12946 and A. dardani V7215 were clustered together. While the 2nd major cluster was divided into 2 groups, 1 of which contained well-watered samples of A. dardani GK12946 and A. dardani V7215, and another with samples of A. ipaënsis K30076 drought and well-watered grouped closely together, suggesting that drought stress expressed a comparatively smaller number of DEGs (Fig. S6A). Similarly, in root samples, the cluster was separated into two groups by treatment, showing a more pronounced impact of drought on the root system. In the root system accession A. dardani GK12946 and A. dardani V7215 were clustered together in both treatments, suggesting the species-specific response, with both A. dardani species performing similarly in the root transcriptome (Fig. S6B). To further investigate gene networks associated with drought tolerance, we performed weighted gene co-expression network analysis (WGCNA) using the topological overlap matrix (TOM) approach with the WGCNA R package [38]. This analysis was performed using a scale-free topology criterion, with a soft-thresholding power of 12. Eight distinct co-expression gene modules were identified, each representing a group of genes with highly correlated expression patterns. These modules, designated by color, varied in size such as black (265 genes), blue (845 genes), brown (704 genes), green (408 genes), pink (12 genes), red (349 genes), turquoise (1,707genes) and yellow (462 genes) (Fig. S7A, Table S6). Among these, the black, brown, and yellow modules were particularly noteworthy, exhibiting substantially higher expression under drought stress than well-watered conditions (Fig. 4A). In contrast, the blue, red, and turquoise modules were predominantly associated with downregulated expression in drought-stressed leaf samples.

Similarly, co-expression analysis of root samples, based on 4,687 significantly expressed genes, identified six modules after merging, ivory (2,897 genes), lightcyan (510 genes), lightcyan1 (1,183 genes), mediumpurple3 (53 genes), palevioletred3 (3 genes), and plum1 (41 genes) (Fig. S7B, Table S6). Notably, the lightcyan1 modules showed strong upregulation under drought stress (Fig. 4B). In contrast, the ivory and lightcyan modules were predominantly associated with downregulated expression in drought-stressed samples (Fig. S7 B).

To better understand the biological roles of these modules, we conducted Gene Ontology (GO) enrichment analyses. In leaf samples, the selected modules, i.e., black, brown, and yellow, were significantly enriched for terms such as cell recognition (GO:0008037), recognition of pollen (GO:0048544), response to abiotic stimulus (GO:0009628), and response to heat (GO:0009408) (Fig. 4C). Similarly, in root samples, GO enrichment analysis of the lightcyan1 module revealed significant enrichment for response to abiotic stimuli (GO:0009628), cell recognition (GO:0008037), response to temperature stimulus (GO:0009266), and response to heat (GO:0009408) were significantly enriched, while protein folding (GO:0006457) showed nonsignificant enrichment but contained a large number of genes (Fig. 4D).

To better understand how gene regulation under drought stress is connected to known hormonal signalling pathways, we integrated co-expression network (WGCNA) with genes involved in abscisic acid (ABA) and ethylene signalling (Fig. S10). Many of these signalling-related genes showed strong differential expression in both roots and shoots under drought conditions. Correlation analyses confirmed that these genes clustered within specific drought-responsive co-expression modules, highlighting their coordinated regulation. In shoot tissues, several ethylene-responsive transcription factors including Araip.T3D3V, Araip.F3QBW, Araip.6J28H, Araip.82L4V, Araip.6ZJ3N, and Araip.T94IA were positively correlated with key drought-response modules. This pattern suggests an important role for ethylene signalling in leaf responses to water deficit, potentially contributing to stress-induced senescence and stomatal regulation [45, 46]. Additionally, ABA-responsive regulators such as the SNF1-related protein kinase (Araip.DH7JW) and a GRAM domain-containing protein (Araip.FHM8U) were also strongly associated with drought modules in shoots. In roots, we observed a somewhat different pattern. ABA-related kinases (Araip.TVQ9L, Araip.I4I55) and the GRAM domain-containing protein (Araip.FHM8U) were positively correlated with drought-responsive modules, consistent with ABA’s established role in root adaptation to drought [47, 48]. Interestingly, certain ethylene-responsive transcription factors (Araip.RA8PB) were also positively correlated, while AP2-like ethylene-responsive factors (Araip.BB9QB, Araip.MYZ56, Araip.E0UEG) were negatively correlated with drought modules. This pattern suggests that ethylene signalling may exert both activating and repressing effects on drought responses in roots, reflecting known antagonistic interactions between ethylene and ABA pathways [49]. Altogether, these findings highlight a clear network connecting ABA signalling components and central transcription factors with broader gene expression changes under drought conditions.

Further MapMan visualization revealed that drought stress induced the expression of multiple regulatory pathways in both leaf and root tissues. In leaves (Fig. S11A), upregulated genes were predominantly associated with redox regulation, transcription factors, hormone signalling (particularly abscisic acid and ethylene), and carbon fixation processes. In roots (Fig. S11B), additional upregulation was observed in categories related to redox regulation (e.g., thioredoxin, glutaredoxin, heme, and catalase), and carbohydrate metabolism. Notably, distinct patterns emerged between tissues, indicating tissue-specific regulatory responses to drought. The color scale indicated strong induction of several gene modules, especially those related to stress hormone metabolism and reactive oxygen species detoxification pathways.

Gene regulatory network (GRN) inference under drought stress in peanut

To unravel the transcriptional regulatory mechanisms underlying drought stress responses in peanut, we constructed a gene regulatory network (GRN) using commonly upregulated DEGs and genes obtained by WGCNA modules. Expression data were processed as VST normalized values after log2 + 1 transformation. The transcription factors were identified from the Plant Transcription Factor Database (PlantTFDB v5.0) [50–52]. The gene regulatory network (GRN) was inferred using the GENIE3 algorithm, applying random forest-based machine learning to predict regulatory relationships with 1000 trees constructed [40]. Out of the 1,377 selected DEGs, which were selected by expressed genes in selected WGCNA modules, a subset of 95 TFs was found to overlap with the expression matrix and used as candidate regulators in leaf samples. The network inference yielded 688 high-confidence regulatory interactions, forming a directed network of 470 genes with 10 TFs, which emerged as central regulators of the drought response, potentially orchestrating large-scale transcriptional reprogramming in the leaf (Fig. 5A, Table S7). Similarly, in root samples, out of the 1,183 selected DEGs, a subset of 69 TFs was found to overlap with the expression matrix and used as candidate regulators in root samples. The network inference yielded 591 high-confidence regulatory interactions, forming a directed network of 285 genes with 10 TFs, which emerged as central regulators of the drought response, potentially orchestrating large-scale transcriptional reprogramming in the root (Fig. 5B, Table S7). These regulatory networks provide a foundational step toward identifying key transcriptional modules and candidate genes for further functional validation in drought tolerance breeding programs. To further contextualize the inferred regulatory networks within established drought-response mechanisms, we examined the functional annotations of key transcription factors identified as central regulators in both leaves and roots. In leaves, several transcription factors were identified, including abscisic acid-responsive element-binding factors (Araip.G3UI0, Araip.ID7EL) and ethylene-responsive element-binding factors (Araip.82L4V, Araip.Z2VYZ, Araip.T3D3V, Araip.F3QBW), which showed strong associations with drought-responsive gene modules (Fig. 6). In root tissues, Araip.ID7EL, Araip.T3D3V, and Araip.3W5AL were also detected as central regulators within the network. Beyond these, additional transcription factors belonged to well-characterized bZIP, NAC, and MYB families, which have been previously implicated in drought tolerance and ABA signalling pathways (Fig. 6). Together, these transcription factors formed hubs in the regulatory networks predicted to coordinate large sets of drought-responsive genes, suggesting coordinated regulation through hormone-mediated pathways.

Fig. 5.

Gene regulatory network (GRN) inference and Sanky plot of selected genes under drought stress in peanut. GRNs were constructed from DEGs in (A) leaf and (B) root samples. Expression data were normalized by VST normalization, and regulatory relationships were inferred using the GENIE3 algorithm, which applies a random forest-based ensemble method to predict gene-gene interactions. From the full DEG set, a subset of transcription factors (TFs) overlapped with the expression matrix and were used as candidate regulators. The analysis yielded high-confidence regulatory interactions (edge weight >0.045), forming a directed network of genes. Network visualization was performed using the ggraph package with a Fruchterman-Reingold force-directed layout to highlight network structure and connectivity. Transcription factors are shown in red and are often central nodes, regulating multiple downstream targets. (C) The Sanky flow diagram revealed the key genes potentially involved in drought stress responses (from left to right gene family, genes, accessions, tissue)

Fig. 6.

Coordinated ABA and ethylene signalling networks drive tissue-specific drought responses in peanut leaves and roots.The figure presents an integrated view of how drought-responsive gene expression in peanut is shaped by hormonal signalling and regulatory networks in leaf (left) and root (right) tissues. Differentially expressed genes (DEGs) responsive to abscisic acid (ABA) and ethylene were identified and visualized through clustered heatmaps, highlighting genotype-specific transcriptional patterns. Gene regulatory networks (GRNs) reveal hormone-driven regulatory hubs that orchestrate downstream responses. These include enhanced expression of genes involved in reactive oxygen species (ROS) detoxification and other key stress-adaptive functions. In leaves, the networks align with physiological processes such as stomatal regulation and senescence, while in roots they support structural and metabolic adaptations for drought resilience. Together, this model underscores the complexity and tissue-specific nature of hormonal crosstalk during drought stress. Abbreviations: DEG, differentially expressed gene; ABA, abscisic acid; GRN, gene regulatory network; ROS, reactive oxygen species

Selected candidate genes may have potential role in drought stress tolerance

To identify key genes involved in drought stress tolerance, we integrated hub genes from co-expression networks, commonly upregulated genes from Venn diagram analysis, and highly connected nodes from the gene regulatory network. Transcription factors were identified based on their central role in the regulatory network across both tissue types. Cross-referencing the selected genes with published literature enhanced the refinement of the gene list, reinforcing their biological significance in drought response across species (Fig. S8, Table S8). The selection process and relationship among these datasets are illustrated using a Sankey diagram (Fig. 5C, Fig. S8), clearly depicting the flow from the gene family to the final set of selected potential candidate genes for drought tolerance.

Discussion

Drought stress is a major limiting factor for crop productivity, and understanding the underlying molecular mechanisms is essential for developing tolerant varieties [12]. In this study, we performed a comprehensive comparative transcriptomic analysis in three Arachis wild species accessions, such as A. dardani GK12946, A. dardani V7215, and A. ipaënsis K30076, under drought and well-watered conditions. This study provides novel insights into drought-responsive transcriptional dynamics across three wild peanut accessions, highlighting both conserved and divergent gene expression patterns that contribute to drought stress adaptation. Through rigorous analytical filtering, we captured robust transcriptional responses in both leaf and root tissues, emphasizing the interplay between genetic background and environmental stress in shaping gene expression [43]. The strong alignment and clustering patterns observed among biological replicates reinforce the quality and consistency of the dataset, while the distinct separation of drought and control samples in the PCA plots illustrates the profound impact of drought on transcriptional activity. These findings are consistent with the previous drought-induced transcriptome reprogramming observed in other peanut species under abiotic stress [53, 54]. Interestingly, the clustering behaviour revealed clear transcriptional similarities between the A. dardani GK12946 and A. dardani V7215. Despite their geographic or ecological differences, their gene expression profiles under drought stress remain closely aligned, suggesting the conservation of core drought response pathways. Such convergence might reflect shared evolutionary pressures or genomic synteny, especially considering their taxonomic proximity. However, subtle expression differences that may emerge from ecotype-specific adaptations cannot be overlooked and warrant further investigation through targeted gene-level analyses. In contrast, A. ipaënsis K30076 displayed a markedly distinct transcriptional profile, separating from the A. dardani accessions under both drought and well-watered conditions. This divergence may point to a fundamentally different regulatory architecture, potentially involving alternate transcription factors, hormone signalling networks, or epigenetic modifications that shape drought perception and response. Given that A. ipaënsis K30076 represents one of the diploid progenitors of cultivated peanut [26], these distinctions offer a valuable opportunity to trace functional genes that could be leveraged in improving drought tolerance.

The identification of a high number of expressed genes in leaf (18,782 genes) and root (20,234 genes) tissues reflects not only the comprehensive sequencing depth achieved but also highlights tissue-specific gene regulation under drought stress. Root tissues, acting as the primary interface for water uptake and stress signalling, understandably exhibited a broader transcriptional repertoire. This observation is consistent with prior reports that emphasize the role of roots in early drought sensing and signalling, which trigger systemic responses in above-ground organs [55, 56].

Moreover, our stringent filtering criteria, guided by best practices in RNA-seq data analysis, ensured that only biologically meaningful gene expression signals were retained for interpretation. By excluding low-count transcripts, we minimized noise and maximized detection power, facilitating clearer distinctions between treatment effects and genotype-specific expression [43, 44]. The differential expression analysis revealed distinct and accession-specific transcriptional responses to drought stress, underscoring the genetic complexity underlying adaptive stress responses. While all accessions demonstrated substantial gene regulation in both leaf and root tissues under drought, the number and identity of DEGs varied significantly, indicating lineage-specific strategies of drought tolerance. The substantial differences in DEG counts observed between genotypes likely reflect underlying biological diversity in their stress response mechanisms. Genotypes with higher DEG numbers may be mounting a more dynamic or complex transcriptional response, while those with fewer DEGs could rely on more constitutive or targeted regulatory pathways. Variation in baseline gene expression and transcriptome complexity also likely contributes to these patterns. Both A. dardani GK12946 and A. dardani V7215 exhibited a high number of DEGs in leaves and roots, with a predominance of downregulated genes. This trend suggests a widespread suppression of growth and primary metabolism under drought conditions, likely reflecting a resource-conservation strategy commonly observed in drought-avoiding species [57]. On the other hand, A. ipaënsis K30076 displayed a comparatively moderate DEG count, implying a potentially more stable transcriptomic response or a narrower but more specialized drought adaptation mechanism. This genotype-specific transcriptional plasticity aligns with known phenotypic differences in drought tolerance and emphasizes the need to consider genetic background when interpreting transcriptomic data. Overall, these findings provide insight into the diverse strategies plants employ to cope with environmental challenges.

Further, functional enrichment analyses of the upregulated genes revealed a conserved activation of core drought-responsive processes across all accessions. These included responses to abiotic stimuli (GO:0009628), temperature (GO:0009266), and heat stress (GO:0009408), which are classical hallmarks of drought-induced signalling in plants [58]. However, the enrichment of additional unique functional categories in each accession provides evidence of divergent adaptation pathways. For instance, the upregulation of genes involved in protein maturation (GO:0051604) and folding (GO:0006457) in A. dardani roots indicates the activation of chaperone-mediated stress tolerance, a mechanism often associated with maintaining proteome integrity during cellular dehydration [59, 60]. Conversely, A. ipaënsis K30076 root-specific DEGs were significantly enriched for terms related to redox regulation and glutamine family amino acid metabolism (GO:0009064, GO:0009084). These categories suggest an alternate physiological adjustment involving nitrogen assimilation and oxidative stress mitigation, aligning with previous studies that highlight glutamine metabolism and redox homeostasis as pivotal components of drought adaptation in plants [60, 61]. Such differences may reflect the ecological histories of these accessions, where A. ipaënsis may have evolved under more oxidative or nutrient-restrictive environments. In addition to the upregulated genes, downregulated DEGs provided insight into the cellular processes that are repressed during drought stress. Across all accessions, a consistent suppression of genes associated with cell division, DNA replication, and microtubule-dependent processes in leaf tissues was observed. This suggests a general arrest of cell proliferation and developmental activity, an adaptive feature to limit energy expenditure during stress [62]. Furthermore, root samples displayed a repression of cell wall-related processes, including cell wall organization and external encapsulating structure biogenesis. The suppression of these pathways might reduce metabolic costs and alter root architecture (Fig. S9), possibly enhancing water uptake efficiency under drought conditions [63]. Overall, these results indicated that while A. dardani and A. ipaënsis share fundamental drought-responsive pathways; they differ markedly in their reliance on protein stabilization versus redox management and amino acid metabolism. These findings emphasize the value of wild Arachis species as reservoirs of diverse and potentially complementary drought-resilience mechanisms.

Further, the identification of conserved and accession-specific drought-responsive genes among three wild Arachis accessions underscores the complexity and diversity of drought adaptation mechanisms in peanut. The differences we observed in drought-responsive gene overlaps reflect a mix of common and unique strategies among the peanut genotypes. This comparative transcriptomic analysis revealed a substantial number of DEGs shared across accessions, particularly those enriched in stress-related biological processes, pointing to a core genetic toolkit that may underlie fundamental drought response pathways. The enrichment of GO terms such as response to abiotic stimulus (GO:0009628), temperature stimulus (GO:0009266), and heat (GO:0009408) in commonly upregulated genes across both leaf and root tissues suggests a conserved activation of heat shock proteins, transcriptional regulators, and protective cellular processes essential for maintaining homeostasis under drought [64, 65]. The core set of genes activated across all accessions highlights fundamental drought survival mechanisms, such as heat shock response, osmotic regulation, and cellular homeostasis, that are consistently employed regardless of genetic background. These conserved responses represent the essential physiological adjustments plants make to cope with water deficit. In contrast, individual genotypes exhibited distinct expression patterns such as differential regulation of ion transporters, secondary metabolism, or cell wall modification pathways which may underlie variation in drought resilience. This interplay between shared core pathways and genotype-specific responses underscores the complexity of drought tolerance as a trait shaped by both universally conserved mechanisms and tailored adaptive strategies. Similar patterns have been reported in other crop species, where drought tolerance emerges from the combined action of broadly conserved stress signalling and genotype-dependent regulatory flexibility [66, 67].

Interestingly, although all accessions activate similar overarching stress responses, the unique gene expression profiles observed in each genotype reflect divergent strategies evolved to counter water deficit. For instance, A. dardani GK12946 exhibited upregulation of genes associated with ion transport (GO:0006811) and ion homeostasis (GO:0050801), indicating potential involvement of osmoregulatory processes in drought mitigation. Metal ions such as potassium and calcium are known to play critical roles in osmotic adjustment and signal transduction under drought stress [68]. On the other hand, A. dardani V7215 showed enrichment in genes related to the phenylpropanoid pathway and lignin biosynthesis, suggesting a cell wall modification strategy that might enhance mechanical strength and reduce water loss, and contribute to improved drought stress endurance [69, 70]. A. ipaënsis K30076 displayed upregulation of genes associated with defence responses and vesicle-mediated transport, implying a more complex network involving crosstalk between biotic and abiotic stress signalling pathways. The involvement of vesicle trafficking and secretion pathways may facilitate the mobilization of stress-signalling molecules or protective proteins [71]. Notably, the defence-associated genes in A. ipaënsis may reflect its ecological niche and evolutionary history, potentially conferring enhanced adaptability under environmental fluctuations.

Root-specific responses also mirrored this diversity (Fig. S9). For instance, upregulated genes involved in protein folding (GO:0006457) and maturation in both A. dardani accessions indicate a role for molecular chaperones in maintaining proteostasis under water-limited conditions. Such responses are consistent with prior findings that heat shock proteins protect cells by preventing protein misfolding and aggregation during drought [72, 73]. Conversely, the enriched expression of genes involved in vesicle trafficking in A. ipaënsis roots may represent an alternative protective mechanism centered on efficient cellular communication and transport [74]. Together, our results highlight a dual strategy in wild Arachis species such as a conserved core drought response framework complemented by genotype-specific enhancements. These unique transcriptomic signatures likely reflect evolutionary adaptations to distinct environments and hold valuable clues for crop improvement. Incorporating genes from both conserved and genotype-specific categories into breeding programs could offer a holistic approach to developing drought-resilient peanut cultivars.

Gene co-expression network analysis revealed key insights into the modular structure of the transcriptomic response to drought stress in wild Arachis accessions. By leveraging variance-stabilized normalized gene expression data, we identified a number of genes that were differentially expressed across leaf and root tissues under drought conditions. The distinct clustering patterns observed through hierarchical analysis reflect both treatment and genotype-specific responses, with notable grouping of the two A. dardani accessions and a distinct expression profile for A. ipaënsis. Particularly in leaf tissue, A. ipaënsis showed minimal transcriptomic shifts under drought, suggesting either a constitutive tolerance mechanism or a delayed response, consistent with findings in other leguminous species where drought responses vary by ecotype and adaptive strategy [60]. Weighted Gene Co-expression Network Analysis (WGCNA) further partitioned these transcriptomes into distinct regulatory modules, each representing tightly co-expressed gene sets. In leaves, drought-responsive modules such as black, brown, and yellow were enriched in biological processes including cell recognition, response to heat, and response to abiotic stimuli. These modules likely represent critical nodes in drought-induced reprogramming mechanisms.

In contrast, modules associated with downregulated genes, such as blue, turquoise, and red, reflect suppression of growth-related or energy-expensive functions, a common adaptation to conserve resources during drought [62]. Such repression is consistent with a metabolic shift favouring stress mitigation over plant growth, particularly in prolonged drought scenarios.

Similarly, root-specific modules revealed a different regulatory landscape, where the lightcyan1 module was strongly upregulated in response to drought. This module was enriched in GO terms related to abiotic stimulus, heat response, and temperature perception key traits for root survival and function under water deficit. Interestingly, while protein folding (GO:0006457) was not significantly enriched, its prevalence among genes in the lightcyan1 module suggests that chaperone activity and proteostasis maintenance may be vital but underexplored aspects of the root stress response. The divergence in module composition between leaf and root tissues underscores the organ-specific regulation of drought response pathways.

By integrating known hormonal signalling pathways with gene co-expression and regulatory networks, we gained a clearer view of how peanut plants orchestrate their transcriptional responses to drought stress. Our results highlight the central importance of ABA signalling, particularly in roots, where ABA-related kinases and regulatory proteins showed strong associations with drought-responsive gene modules. This finding aligns with extensive evidence in Arabidopsis and crop species showing that ABA accumulation and SnRK2 kinase activation drive adaptive responses such as osmotic adjustment, LEA protein induction, and root growth modulation under water deficit [47, 75]. In shoots, ethylene-responsive transcription factors were prominently linked to drought-regulated modules, suggesting ethylene plays a major role in modulating leaf responses. Ethylene has been shown to promote stress-induced senescence and to interact with ABA in regulating stomatal closure and photosynthetic adjustments [45]. The observation that some ethylene-related transcription factors were negatively correlated with drought modules in roots highlights the complex cross-talk between these hormonal pathways, which can have both synergistic and antagonistic effects depending on tissue type and stress intensity [76]. The distinct regulatory patterns we observed between roots and shoots underscore that drought tolerance is not driven by a single mechanism but instead by coordinated, tissue-specific networks integrating multiple hormones signalling pathways. Moreover, the MapMan visualization revealed that drought stress in wild Arachis species activates a broad array of regulatory pathways, prominently including transcription factors (TFs), hormone signalling and redox regulation, reflecting the complex molecular networks underlying drought adaptation. In leaves, the upregulation of TF families such as NAC, MYB, and bZIP aligns with previous findings that these regulators orchestrate stress-responsive gene expression, osmotic adjustment, and stomatal control under water deficit [77, 78]. Hormonal pathways, particularly those involving abscisic acid (ABA) and ethylene, were strongly induced, supporting their well-established roles in drought sensing and signalling cascades that promote adaptive responses [79]. Notably, in roots, our analysis highlighted enhanced expression of redox-related proteins, including thioredoxins and glutaredoxins, consistent with reports showing that root tissues prioritize ROS detoxification and signalling processes to mitigate cellular damage during drought [80].Further, the integration of WGCNA module genes with differentially expressed transcription factors enabled the construction of regulatory networks using the GENIE3 machine-learning algorithm. This approach identified high-confidence transcriptional hubs in both leaf and root networks. Despite the relatively small number of TFs (10 in each tissue) acting as central regulators, they were linked to several target genes, indicating a hierarchical control structure consistent with stress-induced reprogramming models in plants [81]. These key TFs likely represent master regulators orchestrating gene expression cascades during drought, making them prime candidates for targeted functional studies and transgenic manipulation. Collectively, the co-expression and GRN analyses emphasize a dual-layered architecture of drought responsiveness in wild peanut: modular transcriptomic coordination of gene sets under environmental stress and centralized regulatory control through specific transcription factors.

The presence of ethylene-responsive transcription factors prominently connected to drought-responsive modules in leaves suggests that ethylene signalling may play an important role in coordinating leaf-specific responses, possibly related to stomatal regulation and senescence under drought stress [45]. In contrast, the stronger associations of ABA-responsive factors in roots are consistent with the central role of ABA in mediating root adaptation to water deficit, including maintenance of root growth and induction of protective proteins [48, 75]. Overall, these results indicate that drought tolerance in peanut involves a complex transcriptional reprogramming process in which ABA and ethylene pathways converge through key transcription factors to coordinate tissue-specific responses. This integrative network perspective offers a valuable framework for future functional studies and breeding strategies aimed at improving drought resilience. Further, the visualization using MapMan suggests a greater involvement of the redox regulatory pathway. The Fig. 6 showing the possible relation between ABA and ethylene signalling pathways genes.

The drought tolerance in plants is a complex trait regulated by dynamic interactions between gene expression, transcriptional regulation, and physiological adaptation. By integrating DEGs, co-expression network hubs, and gene regulatory networks (GRNs), our study identified a robust set of candidate genes potentially contributing to drought stress adaptation in peanut (Arachis spp.). This multi-dimensional approach allowed us to dissect the transcriptional landscape associated with drought responses and highlight key regulators that are consistent across tissues and accessions.

The integration of commonly upregulated DEGs from Venn diagram analysis with hub genes from WGCNA modules provided a foundational set of genes consistently responsive to drought across diverse genotypes. Such convergence highlights conserved genetic components of drought response, aligning with previous findings in other crop species, including rice and maize, where core stress-responsive genes were similarly conserved across genotypes and developmental stages [82]. The use of GRN analysis based on GENIE3 enabled us to infer directional regulatory interactions between TFs and target genes, offering a mechanistic framework for transcriptional control under stress. In both leaf and root networks, several TFs exhibited high connectivity, suggesting their role as potential master regulators. This observation is consistent with prior studies showing that network centrality often correlates with functional importance under stress conditions [83]. Moreover, the overlap of these central TFs across tissues implies their systemic role in orchestrating drought responses, rather than a compartment-specific adaptation. The Sankey diagram visualization of gene integration further provided a clear pathway from initial gene family identification to the final candidate gene selection, facilitating a systems-level understanding of gene prioritization. This visualization strengthens the interpretability of our integrative approach. Importantly, several of the identified candidate genes are associated with well-documented drought-responsive processes. For instance, genes involved in osmoprotectant biosynthesis, ROS scavenging, protein folding (GO:0006457) (e.g., HSP70), and ion transport (GO:0006811) were enriched among the selected candidates. These genes have been widely recognized for their roles in cellular homeostasis and signalling during water-deficit stress [84, 85]. We identified crucial putative genes for drought stress tolerance, such as Araip.1SK7V (uncharacterized protein), Araip.DLE10 (heat shock protein 70), Araip.KJK12 (ADP, ATP carrier protein 3), Araip.M8LQV (NADH-ubiquinone oxidoreductase), Araip.PK218 (heat shock protein 21), Araip.U5NBJ (outer membrane tryptophan-rich sensory protein), and Araip.TR541 (Dehydrin family protein), which was highly upregulated in both leaf and root tissues, and involved in drought stress tolerance [73, 86–89]. We also selected crucial transcription factors such as Araip.W5H5G and Araip.Q811U (MYB transcription factor), Araip.2404G (heat shock transcription factor A2), which are important for drought tolerance [90–92]. We also identified well-known proteins like Araip.J65RE (expansin-like B1), Araip.N419E, Araip. 84L6B, Araip. 9A6T0 (late embryogenesis abundant protein, LEA), Araip.L3NWH (galactinol synthase 1), and Araip.NW3CH (Sugar transporter SWEET), which plays an important role in drought tolerance in other plants [93–95]. The identification of such genes in our datasets not only validates the robustness of our analytical framework but also provides immediate targets for functional characterization and marker-assisted breeding. In summary, the integrative strategy adopted in this study allowed for the identification of a biologically meaningful set of candidate genes that are conserved, centrally connected, and differentially regulated under drought stress. These genes serve as promising targets for genetic engineering and breeding programs aimed at enhancing drought tolerance in peanut. Future work will involve functional validation of these candidates through reverse genetics and transcriptome-assisted trait dissection in larger populations.

Conclusion

In this study, we conducted a comprehensive transcriptomic and co-expression network analysis to elucidate the molecular mechanisms underlying drought stress responses in three Arachis accessions. Our findings reveal that drought stress triggers extensive transcriptional reprogramming, involving both conserved core responses, such as responses to abiotic stimuli, heat, and transcriptional regulation, and accession-specific regulatory mechanisms. The observed tissue- and genotype-specific expression patterns suggest divergent adaptive strategies among the accessions. By integrating DEGs, gene co-expression modules, and gene regulatory networks, we identified key candidate genes and transcription factors potentially involved in drought tolerance. The overlap of these genes across accessions and tissues, along with supporting functional annotations, highlights their significance as central regulators of drought response. These insights not only deepen our understanding of drought adaptation in peanuts but also provide a valuable genomic resource for crop improvement. The identified genes and modules represent promising targets for molecular breeding and biotechnological interventions aimed at developing drought-resilient peanut cultivars. Future functional validation of these candidate genes will be essential to harness their potential in breeding programs for climate-smart agriculture.

Abbreviations

W

well-watered

D

drought

DAP

days after planting

RNA

Ribonucleic acid

PCA

Principal Component Analysis

FDR

false discovery rate

DEGs

Differentially Expressed Genes

GO

Gene Ontology

VST

variance stabilizing transformation

WGCNA

Weighted Gene Co-expression Network Analysis

GRN

Gene regulatory network

TF

Transcription factor

SRA

Sequence Read Archive

Authors’ contributions

PV and MJ conducted formal analysis, investigation, and data visualization. JC and MJ contributed to conceptualization, funding acquisition, and methodology design. PV drafted the original manuscript. All authors participated in the review and editing of the manuscript and approved the final version.

Funding

This research was supported by salary funds from the Faculty Texas University Funds (TUF) Start-Up Program, administered by the Operations Division of Texas Tech University (TTU).

Data availability

The sequenced raw reads generated in this study have been submitted to the National Center for Biotechnology Information (NCBI) with BioProject ID: PRJNA1261117 [https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA1261117%20](https:/www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA1261117%20). Additional analytical data during this study are included in the supplemental information.

Declarations

Ethics approval and consent to participate

The plant materials (Arachis dardani and Arachis ipaënsis) used in this study were obtained from the Texas A&M AgriLife Research wild Arachis germplasm collection maintained at Texas A&M AgriLife Research, Stephenville, Texas. All collections were performed with appropriate institutional permissions, and no field collection from wild populations was undertaken for this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.