Comprehensive analysis of TGA genes in barley: structure, evolution, and stress-responsive expression Authors: Bahman Panahi, Rasmieh Hamid, Zahra Ghorbanzadeh, Komal G Lakhani, Feba Jacob Published: 2026 ABSTRACT BackgroundTGA transcription factors (bZIP family) are central to hormone signalling and stress responses in plants. However, their genomic organization and regulatory potential in barley (Hordeum vulgare L.) remain insufficiently characterised.ResultsA total of 49 HvTGA genes were identified, which are unevenly distributed across the barley genome, with the most prominent clusters located on chromosomes 3 H and 5 H. Phylogenetic reconstruction divided the family into four clades supported by conserved bZIP domains, clade-specific motifs, and distinct exon–intron structures, suggesting both evolutionary conservation and functional diversification. Gene duplication analysis revealed that segmental and dispersed duplications are the dominant drivers of expansion, while Ka/Ks ratios < 1 indicate strong purifying selection maintaining functional stability. Promoter analysis revealed numerous cis-regulatory elements associated with ABA, SA, JA, osmotic stress, and light responses, with HvTGA-18, HvTGA-24, and HvTGA-35 carrying particularly dense clusters. Post-transcriptional regulation was supported by the prediction of 31 miRNAs targeting 29 HvTGAs, with hvu-miR6192 emerging as a key regulator of eight members. Protein–protein interaction networks based on Arabidopsis orthologues identified HvTGA-13, HvTGA-26, and HvTGA-28 as putative components of predicted regulatory interaction networks, interacting with NPR and GRX proteins, along with other stress-responsive transcription factors. Functional enrichment analysis further indicated that HvTGAs are involved in transcriptional regulation, hormone-mediated signalling cascades, MAPK signalling pathways, and carotenoid biosynthesis. Expression analysis of HvTGA genes across multiple abiotic stress conditions and tissues revealed distinct and dynamic transcriptional responses to heat, salt, and osmotic stress. Under heat stress, several HvTGA genes, including HvTGA-29, HvTGA-43, HvTGA-41, and HvTGA-38, were significantly up-regulated in both shoot and root tissues. In response to salt stress, HvTGA-29, HvTGA-12, HvTGA-14, HvTGA-21, and HvTGA-33 showed strong induction in root tissue, particularly at later time points. Notably, HvTGA-29, HvTGA-43, and HvTGA-41 exhibited consistent and robust induction across multiple abiotic stress conditions and tissues, supported by functional enrichment analysis, indicating their potential roles as central regulators within stress-responsive transcriptional networks in barley.ConclusionsThis first genome-wide characterisation of HvTGAs uncovers their structural diversity, evolutionary origins, and regulatory complexity, while highlighting HvTGA-29, HvTGA-43, and HvTGA-41 as key candidate regulators likely contributing to stress- and hormone-responsive networks in barley.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12870-026-08997-7. Introduction Plants are constantly exposed to a variety of biotic and abiotic stress factors. They have evolved sophisticated defense mechanisms and complex signalling networks that enable them to sense these signals and reprogram gene expression to support survival and adaptation [1, 2]. Transcription factors (TFs) play a key role in plant signalling and are essential for the regulation of metabolic balance, growth, development and defense [3, 4]. By binding to cis-regulatory elements in the promoters of target genes, TFs can activate or repress transcription in a coordinated and context-dependent manner [5]. The bZIP (basic region-leucine zipper) family is one of the most important TF families in plants and is of particular importance due to its size, structural diversity and involvement in numerous physiological processes [6]. Based on the sequence similarity and structural features of the bZIP domain, the bZIP family in Arabidopsis thaliana has been categorised into ten major groups (A, B, C, D, E, F, G, H, I and S) [7]. Within this classification, the TGACG motif-binding factors (TGA) transcription factors form group D of the bZIP family [8]. TGAs contain a conserved bZIP domain (basic region-leucine zipper) and a relatively conserved C-terminal region, as well as a highly variable N-terminal region [9]. The basic region acts as a DNA recognition motif and as a signal for nuclear localisation, while the N-terminus shows considerable sequence and length variation between the paralogues. The C-terminus usually contains two glutamine-rich regions (Q1 and Q2) and may also contain a DOG1 (Delay of Germination 1) domain that facilitates protein-protein interactions rather than direct DNA binding [10, 11]. TGA1a was historically the first TGA gene to be cloned in plants (from tobacco), which was an important milestone in the characterization of this gene family [12]. Since then, TGA transcription factors (TFs) have been reported in various species [13–16]. In Arabidopsis, ten TGA members were categorised into five subgroups: Group I (TGA1 and TGA4); Group II (TGA2, TGA5, and TGA6); Group III (TGA3 and TGA7); Group IV (TGA9 and TGA10); and Group V (TGA8, also known as PERIANTHIA) [11]. Beyond this structural classification, functional studies have shown the involvement of TGAs in pathogen defense, developmental programmes and responses to abiotic stress [14, 17, 18]. TGAs act as critical nodes in hormone-mediated signalling pathways, particularly in the salicylic acid (SA) and jasmonic acid (JA) pathways. Here they interact with co-regulators such as NPR1 and are modulated by redox status and post-translational modifications [19]. Recent studies also suggest that TGAs play a role in abiotic stress tolerance. For example, genome-wide analyses in kiwifruit and Brassica napus have revealed TGA members that respond to drought stress [14, 16], while recent studies in cereals such as rice suggest that certain TGAs contribute to antioxidant defense and hormonally-driven stress signalling [20]. However, in cereals, these functions are largely inferred from expression profiling and functional validation remains limited. Barley (Hordeum vulgare L.) is one of the earliest cereals to be domesticated. Its origins are traced back to Hordeum spontaneum in the Fertile Crescent about 10,000 years ago. Today, it is one of the most important cereals worldwide and is used as a staple food, animal feed and malt [21, 22]. Nevertheless, barley productivity is increasingly threatened by environmental changes. Drought and heat stress can severely affect yield stability, while fungal pathogens such as Blumeria graminis f. sp. hordei (powdery mildew), Pyrenophora teres (net blotch) and Fusarium spp. (late blight) can lead to significant losses in grain yield and quality [23, 24]. These stress factors disrupt cellular homeostasis, redox balance and hormone signalling and ultimately impair important physiological processes, including photosynthesis and osmotic adjustment [25]. Improving the resilience of barley, therefore requires a detailed understanding of stress-responsive regulatory networks and their transcriptional control. Although TGA transcription factors (TFs) have been investigated in model species and, more recently, in crops such as wheat [26] and Brassica napus [16], they remain largely uncharacterized in barley. Comprehensive identification and annotation of TGAs in this species are therefore essential to clarify their structural diversity, evolutionary history and potential regulatory functions. In this study, we conducted a genome-wide analysis of the barley TGA gene family (hereafter designated HvTGAs). We investigated their chromosomal distribution, phylogenetic relationships, gene structures, conserved motifs and cis-regulatory elements, and further assessed their expression under drought stress using transcriptomic data. To the best of our knowledge, this work represents the first systematic characterization of TGAs in barley, thereby filling a critical gap in Triticeae transcription factor research and providing a foundation for functional genomics and targeted crop improvement. Materials and methods Identification and characterization of the TGA gene family in barley The reference genome assembly of Hordeum vulgare cv. Morex V3 and its corresponding gene annotation (Ensembl Plants, release 56, 2023) were retrieved from the Ensembl Plants database (https://plants.ensembl.org). For comparative purposes, all annotated protein sequences of TGA transcription factors from Arabidopsis thaliana were obtained from The Arabidopsis Information Resource (TAIR; https://www.arabidopsis.org) [27]. Putative barley TGA genes (HvTGAs) were identified using a dual strategy combining Hidden Markov Model (HMM) searches and BLASTP similarity searches. The HMM profile corresponding to the conserved basic leucine zipper (bZIP) domain (Pfam ID: PF00170; Pfam v35.0) [28] was used to query the barley proteome using HMMER v3.0 [29], with an E-value threshold of 1 × 10⁻⁵. In parallel, BLASTP searches were conducted using Arabidopsis TGA protein sequences as queries with an E-value cut-off of 1 × 10⁻⁵. Candidate sequences obtained from both approaches were merged, and redundancy was removed using CD-HIT v4.8.1 [30] at 100% sequence identity. Partial sequences lacking a complete bZIP domain were excluded. To discriminate TGA transcription factors from the broader bZIP superfamily, a hierarchical multi-criterion filtering framework was implemented, integrating domain integrity, structural features, sequence composition, and phylogenetic placement. (i) Core domain requirement: All candidate proteins were required to contain a complete and intact bZIP domain, including both the basic DNA-binding region and the leucine zipper dimerisation motif, as confirmed by Pfam, SMART, and the Conserved Domain Database (CDD; https://www.ncbi.nlm.nih.gov/Structure/cdd/cdd.shtml) [31]. Proteins lacking a continuous heptad leucine repeat or showing truncated domain architecture were excluded. (ii) Structural filtering: Sequences shorter than 150 amino acids were removed to eliminate truncated or non-functional bZIP proteins. This threshold reflects the minimal structural requirement for functional TGA transcription factors reported in previous studies [16]. (iii) TGA-specific auxiliary features: Candidate proteins were further evaluated for the presence of glutamine-rich (Q-rich) regions, which are characteristic of TGA transcription factors and are associated with transcriptional activation capacity. Q-rich regions were operationally defined as segments of ≥ 30 amino acids containing ≥ 30% glutamine residues and were identified through amino acid composition analysis combined with low-complexity detection (SEG) and motif profiling. (iv) DOG1-like domain assessment: The presence of DOG1-like domains was examined using Pfam and CDD annotations, with a detection threshold of E-value ≤ 1 × 10⁻³ and minimum domain coverage of 50%. Given that DOG1-like domains are not universally conserved among TGAs, this feature was treated as supportive rather than diagnostic. (v) Phylogenetic validation: To ensure accurate classification, candidate HvTGA proteins were subjected to phylogenetic analysis alongside experimentally characterised TGA proteins from Arabidopsis thaliana and representative monocot species. Multiple sequence alignment was performed using MUSCLE, followed by maximum likelihood tree construction with 1,000 bootstrap replicates under the Jones–Taylor–Thornton (JTT) substitution model. Only sequences that clustered within well-supported TGA clades (bootstrap ≥ 70%) were retained, reflecting orthology-based classification and evolutionary conservation of TGA subfamily members. (vi) Exclusion criteria: Proteins containing the generic bZIP domain but lacking TGA-associated auxiliary features and failing to group within TGA-specific phylogenetic clades were excluded, ensuring removal of non-TGA bZIP transcription factors. This integrative and hierarchical framework ensured stringent discrimination of HvTGA proteins from other bZIP family members, improving both classification accuracy and reproducibility. Physicochemical characterization and subcellular localization prediction To characterize the basic biochemical properties of the identified barley TGA transcription factors (HvTGAs), their amino acid sequence lengths, molecular weights, and isoelectric points (pI) were calculated using the ExPASy ProtParam tool (https://web.expasy.org/protparam/) [32]. These parameters provide insight into the molecular characteristics of HvTGA proteins, including potential solubility, stability, and susceptibility to post-translational modifications. Subcellular localisation of HvTGA proteins was predicted using three complementary in silico tools: Plant-mPLoc (http://www.csbio.sjtu.edu.cn/bioinf/plant-multi/), TargetP v2.0 (http://www.cbs.dtu.dk/services/TargetP/), and WoLF PSORT (https://wolfpsort.hgc.jp/) [33]. Each protein sequence was analyzed independently with all three programs, and a consensus localization was assigned when at least two tools predicted the same primary compartment. This majority-rule approach improves prediction confidence by mitigating tool-specific biases. Given the established role of TGA proteins as transcriptional regulators, nuclear localisation was considered the predominant functional compartment, whereas additional predicted localisations may indicate the presence of overlapping targeting signals or context-dependent intracellular redistribution observed in certain plant regulatory proteins. Predicted localisations were then interpreted in the context of known TGA TF compartmentalisation in other plant species, providing insights into potential functional roles of HvTGA proteins. Phylogenetic analysis of TGA proteins To investigate the evolutionary relationships of TGA transcription factors in barley (Hordeum vulgare), full-length HvTGA protein sequences were aligned together with representative TGA sequences from Arabidopsis thaliana and Oryza sativa. Multiple sequence alignment was performed using the MAFFT v7 algorithm with the L-INS-i strategy, which is recommended for protein families with variable sequence length and conserved domains, providing improved alignment accuracy compared to earlier methods. To minimise the impact of poorly aligned or divergent regions, the resulting alignments were subjected to trimming using trimAl v1.4 with default settings, thereby retaining only phylogenetically informative positions for downstream analysis. Phylogenetic trees were constructed using the Maximum Likelihood (ML) method implemented in MEGA v12, employing 1,000 bootstrap replicates to assess branch reliability. The optimal amino acid substitution model was determined using the built-in model selection function in MEGA v12 based on the lowest Bayesian Information Criterion (BIC) score, and the LG + G+I model was selected as the best-fitting model for tree construction. Two trees were generated: one including only HvTGA proteins to explore intra-species divergence, and another incorporating all three species to assess inter-species evolutionary relationships. The resulting phylogenies were visualized using the Interactive Tree of Life (iTOL) platform [34, 35]. To evaluate domain-level conservation within the HvTGA family, sequence logos were generated from the aligned protein sequences using WebLogo (https://weblogo.berkeley.edu), highlighting conserved amino acid motifs and putative functional residues. Chromosomal distribution, synteny analysis, gene duplication, and Ka/Ks ratio estimation The chromosomal positions of the identified barley TGA genes (HvTGAs) were obtained from the reference genome annotation and their physical distribution across chromosomes was visualized using TBtools v1.108 [36]. To investigate gene duplications and syntenic relationships, collinearity analysis was performed using the MCScanX toolkit [37] integrated within TBtools. All-against-all BLASTP analysis was first conducted using an E-value threshold of 1e⁻⁵ to identify homologous gene pairs, and collinear blocks were defined using MCScanX with default parameters, including a minimum of five collinear gene pairs per block. Duplication types, including whole-genome/segmental, tandem, proximal, dispersed, and singleton duplications, were classified according to MCScanX standard criteria based on gene location and collinearity relationships. Intra-genomic duplications and syntenic links were visualized using a circular map generated by the Basic CIRCOS function, highlighting the physical and evolutionary relationships among HvTGA genes. To assess selection pressures acting on duplicated gene pairs, coding sequences (CDS) of HvTGA genes were aligned using MUSCLE at the amino acid level, followed by back-translation to codon alignments using a codon-aware approach implemented within TBtools (PAL2NAL-based pipeline) and analyzed with the “Simple Ka/Ks Calculator” module in TBtools v1.108 [36]. Ka and Ks values were estimated using the Nei–Gojobori (NG) method implemented in the calculator module. Non-synonymous (Ka) and synonymous (Ks) substitution rates were calculated, and the Ka/Ks ratio was used to infer selective constraints: Ka/Ks < 1 indicates purifying selection, Ka/Ks > 1 indicates positive selection, and Ka/Ks ≈ 1 indicates neutral evolution [38]. Motif, domain, and gene structure analysis Conserved motifs within the HvTGA protein sequences were identified using MEME Suite v5.4.1 (http://meme-suite.org/) [39], with the following parameters: maximum number of motifs = 10; motif width set between 6 and 50 residues; distribution of motifs = zero or one occurrence per sequence (zoops); and an E-value threshold of < 1 × 10⁻⁵. Motif discovery was performed on the full-length protein sequences to capture both the conserved bZIP core and variable C-terminal regions, including low-complexity segments associated with transcriptional activation. Identified motifs were subsequently compared with annotated domains to assess correspondence with the canonical bZIP region and to evaluate the distribution of auxiliary sequence features, including glutamine-rich (Q-rich) regions. Functional domain annotation was conducted using the SMART database (http://smart.embl-heidelberg.de/) [40] and the Pfam database (https://pfam.xfam.org/) [41] to assess the presence of conserved bZIP and DOG1-like domains characteristic of TGA transcription factors. DOG1-like domain predictions were cross-validated between databases to minimise annotation artefacts and were interpreted as supportive features rather than definitive classification criteria. Gene structure analysis, including exon–intron organisation, was performed using TBtools v1.108 [36], based on the GFF3 annotation file of the barley reference genome. The results of the motif, domain, and structural analyses were subsequently visualized alongside the phylogenetic tree using the integrated visualization functions of TBtools. This integrative framework enabled the assessment of structural conservation within phylogenetic clades and provided complementary evidence supporting the classification of HvTGA family members. Cis-acting element analysis in TGA gene promoters Promoter sequences (2000 bp upstream of the translation start site) of the HvTGA genes were extracted from the barley reference genome (IBSC_v2, Ensembl Plants) and analyzed using the PlantCARE database (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) [42] to predict cis-acting regulatory elements. The predicted elements were subsequently classified into functional categories (e.g., stress-responsive, hormone-responsive, and light-responsive). Their distribution patterns were visualized using the motif heatmap and cis-element visualization tools implemented in TBtools v1.108. Prediction of HvTGA protein-protein interactions To investigate potential functional associations of HvTGA transcription factors, protein–protein interaction (PPI) networks were predicted using the STRING database v12.0 (https://string-db.org) [43]. Full-length HvTGA protein sequences were retrieved from the barley reference proteome (Ensembl Plants, IBSC_v3) and queried against the STRING database. Interactions were inferred using STRING’s default pipeline, which integrates multiple evidence sources, including curated experimental datasets, co-expression, gene neighbourhood, gene fusion events, phylogenetic co-occurrence, and interolog-based inference from orthologous proteins. The following parameters were applied: minimum required interaction score = 0.7 (high confidence); network type = full STRING network; active interaction sources = all; and maximum number of interactors = 10 (1st shell only). Prediction of putative mirnas targeting HvTGA genes and functional enrichment analysis To identify potential microRNAs (miRNAs) regulating barley TGA genes (HvTGAs), mature miRNA sequences of Hordeum vulgare were retrieved from miRBase v22.1 (http://www.mirbase.org) [44] and supplemented with experimentally validated barley miRNAs reported in the literature. Coding sequences (CDS) of HvTGA genes were obtained from the Ensembl Plants reference proteome (IBSC_v3) and submitted to the psRNATarget server (psRNATarget 2017 release; Schema V2) (https://www.zhaolab.org/psRNATarget/) for miRNA–mRNA interaction prediction. The analysis was performed using the following parameters: expectation score threshold ≤ 3.0; maximum energy to unpair the target site (UPE) = 25; flanking length around target site for accessibility analysis = 17 bp upstream and 13 bp downstream; central mismatch range for translational inhibition = positions 9–11; and seed region requirement = no more than one mismatch. Only interaction pairs fulfilling all parameter constraints were retained for downstream analyses. Candidate miRNAs were additionally filtered based on annotation support in miRBase and previously reported barley miRNA datasets. To investigate the potential functional roles of miRNA-regulated HvTGAs, enrichment analyses were performed using the subset of genes predicted as miRNA targets. Gene Ontology (GO) enrichment was carried out with AgriGO v2.0 (http://systemsbiology.cau.edu.cn/agriGOv2/) [45], applying Fisher’s exact test with Benjamini–Hochberg false discovery rate (FDR) correction and using the barley genome as background. Kyoto Encyclopaedia of Genes and Genomes (KEGG) pathway analysis was performed using KEGG Mapper (https://www.kegg.jp/) to assign targeted HvTGAs to relevant signalling and metabolic pathways. This approach provides a framework for identifying putative post-transcriptional regulatory modules and associated biological processes. Expression analysis of TGA genes under different abiotic stress conditions To investigate the expression profiles of HvTGA genes under abiotic stress, we used publicly available RNA-Seq datasets from the NCBI Sequence Read Archive (SRA). Barley faces challenges from environmental factors such as heat, salinity, and drought-related osmotic stress, which can reduce yield and quality. These stresses were prioritised to capture primary transcriptomic responses relevant to barley resilience. The dataset PRJNA324116 was selected for analysing HvTGA transcriptional dynamics in root tissues exposed to heat stress, which affects water and nutrient uptake. In the heat experiment, barley plants cultivar Rolap were grown under conditions as previously described [46], and the heat stress conditions were established also as previously described [47], except that the soil water content was maintained at 70% to prevent drought stress. The soil/sand mixture (7:2) was amended with 35 mL of a medium containing: 28 mM NH₄NO₃, 20 mM KH₂PO₄, 4 mM K₂SO₄, 16 mM MgSO₄·7 H₂O, 53 µM H₃BO₃, 8 µM CuSO₄, 4 µM MnSO₄·H₂O, and 120 µM FeCl₃·6 H₂O. Two-week-old barley plants were exposed to heat stress for 48 h. For ionic stress, PRJNA489775 was used to assess root responses under salinity stress, a critical constraint in barley-growing regions where soil salinisation limits productivity. In this experiment, transcriptome profiles of salt-treated roots of Hordeum vulgare cultivar Morex were analysed. The seedlings were transferred to plastic pots containing sand culture medium (particle size 50–200 μm) under greenhouse conditions with a photoperiod of 12 h light/12 h dark, a photosynthetic photon flux density (PPFD) of approximately 400 µmol m⁻² s⁻¹, day/night temperatures of 25 ± 3 °C/17 ± 3 °C, and relative humidity of approximately 50%. Plants were irrigated every two days and fertilised once per week using half-strength Hoagland’s solution. In addition, PRJNA679445 was used to evaluate HvTGA expression in shoot tissues under osmotic stress, simulating drought-like conditions affecting photosynthetic tissues. In this experiment, seeds of cultivated barley (Hordeum vulgare ssp. vulgare) varieties Barke, Morex, and Golden Promise were stratified at 4 °C for one week. Wild accessions were identified according to their country of origin. Seeds were germinated in darkness at 25 °C on moist filter paper. After three days, seedlings were transferred to an aerated hydroponic system containing half-strength Hoagland solution and grown in a climate-controlled chamber under long-day conditions (16 h light/8 h dark), with day/night temperatures of 23/20°C and relative humidity ranging from 50% to 65%. At six days of age, seedlings were subjected to either osmotic stress or control treatment for an additional six days. By day 12, plants typically exhibited two leaves and five to six seminal roots [48]. All seed material was obtained from the seed bank of the University of Tabriz, Iran. The plant material comprised cultivated barley (H. vulgare L.) varieties and associated germplasm maintained within an institutional collection; no specific permits were required for their use. All procedures were conducted in accordance with relevant institutional and national regulations governing the use of plant genetic resources. For each dataset, the experimental design included biological replicates for both control and stress-treated conditions, typically comprising three independent biological replicates per condition as reported in the original studies (PRJNA324116, PRJNA489775, and PRJNA679445, three replicates for each). These replicate-level count data were incorporated into downstream statistical modelling to enable robust estimation of dispersion and biological variability within the DESeq2 framework. The integration of root and shoot datasets provided a tissue-specific comparative framework, enabling the identification of stress-responsive patterns across different organs. These datasets offered a robust foundation for comprehensive evaluation of HvTGA expression dynamics and their potential roles in barley adaptation to abiotic stress. Raw RNA-Seq reads were first subjected to quality control using FastQC v0.11.5 [49]. Low-quality reads (Phred score < 30) and adapter sequences were removed using Trimmomatic v0.32 [50], with the following parameters: ILLUMINACLIP:2:30:10; LEADING:3; TRAILING:3; SLIDINGWINDOW:4:15; MINLEN:36. The filtered high-quality reads were then aligned to the barley reference genome using HISAT2 v2.2.1 [51], employing the default end-to-end alignment mode with the --dta option enabled for downstream transcript assembly, allowing up to two mismatches per read and a maximum intron length of 5,000 bp. Transcript abundance was quantified using HTSeq-count v0.13.5 [52], using the “union” mode with parameters set to –stranded = no and –type=exon to ensure consistent gene-level read summarization. Differential gene expression analysis was performed using DESeq2 v1.38.0 in R, based on raw count data generated from HTSeq-count. Expression values were normalised using the median-of-ratios method implemented in DESeq2. A generalised linear modelling framework based on the negative binomial distribution was applied to assess differential expression between control and stress-treated samples within each dataset. The design formula incorporated treatment condition as the primary factor, and dispersion estimates were derived from biological replicates to account for within-group variability. Differentially expressed genes (DEGs) were identified using the thresholds of absolute log₂ fold change (|log₂FC|) ≥ 1 and false discovery rate (FDR)-adjusted p-value < 0.05, following Benjamini–Hochberg correction for multiple testing. Adjusted p-values were used to control for multiple testing and to ensure robust identification of statistically significant transcriptional changes. Dispersion estimation and model fitting were conducted using DESeq2 default parameters, with log2 fold change shrinkage applied using the apeglm method to improve effect size estimation for low-count genes. These criteria were applied to define statistically significant transcriptional responses under heat, salt, and osmotic stress conditions. For visualisation purposes, fragments per kilobase of transcript per million mapped reads (FPKM) values were calculated and used for heatmap generation in TBtools v1.108 [36], allowing comparative representation of HvTGA expression profiles across tissues and stress treatments. FPKM-normalised values were used exclusively for visualisation and clustering analyses, whereas all statistical inferences regarding differential expression were derived from count-based modelling within the DESeq2 framework. Hierarchical clustering was performed using Euclidean distance and complete linkage to group genes with similar expression patterns across conditions. Assessment of differential expression robustness of HvTGA Genes Differentially expressed HvTGA genes identified from RNA-seq analysis were further assessed using leave-one-out cross-validation (LOOCV) to evaluate the functional importance HvTGA genes based on expression values across stress and normal growth conditions. The RNA-seq datasets used in this study were obtained from previously published experiments deposited in public repositories and therefore represent experimentally generated transcriptomic data under defined abiotic stress conditions. LOOCV was implemented as a resampling-based statistical approach to assess the consistency of expression-derived classifications within each dataset. In this procedure, individual samples were sequentially excluded, while the remaining samples were used to train the model, which was then tested on the omitted sample. This iterative framework allows each replicate to serve as both training and testing data, thereby providing an estimate of model generalisation and reducing overfitting. The application of LOOCV enabled the evaluation of the robustness and stability of HvTGA expression patterns within the analysed datasets. In addition, comparative analysis across independent RNA-seq datasets derived from different experimental studies was performed to examine the reproducibility of stress-associated expression patterns across tissues and conditions. These complementary analyses support the identification of HvTGA genes exhibiting consistent stress-associated expression patterns. LOOCV was used solely as a statistical approach to assess internal consistency of the expression data. Results Identification and characterization of the TAG gene family in barley A total of 49 HvTGA genes were identified in the barley genome by BLASTp and HMMER3 searches using Arabidopsis thaliana and Oryza sativa TGA protein sequences as queries (Table S1). These genes were defined based on a hierarchical classification framework integrating bZIP domain integrity, TGA-associated auxiliary features, and phylogenetic clustering within well-supported TGA clades, enabling robust discrimination from other bZIP transcription factors. The HvTGA gene models showed substantial variation in genome organisation and transcript structure. Transcript lengths, including untranslated regions (UTRs) and coding sequences (CDS), ranged from 585 to 2,397 bp, whereas CDS lengths ranged from 482 to 2,298 bp. Most genes encoded a single transcript isoform, while HvTGA-17, HvTGA-22, HvTGA-31, HvTGA-34 and HvTGA-41 produced two isoforms and HvTGA-19 produced three isoforms, indicating the occurrence of alternative splicing events within the family. The predicted HvTGA proteins ranged from 154 to 645 amino acids in length, with molecular weights between 16.9 and 68.3 kDa. The theoretical isoelectric points (pI) ranged from 4.73 to 9.96, reflecting substantial variation in charge properties. GRAVY values ranged from − 1.451 to 0, indicating that most HvTGA proteins are hydrophilic; HvTGA-41 exhibited the lowest GRAVY value and was therefore predicted to be the most hydrophilic family member. The aliphatic index ranged from 46.9 to 82.3, whereas the instability index ranged from 41.8 (stable) to 99.0 (unstable), suggesting diversity in predicted thermostability and structural stability across the family. Subcellular localisation was predicted using Plant-mPLoc, TargetP v2.0 and WoLF PSORT, and consensus assignments were obtained using majority agreement among the three platforms (Table S2). Most HvTGA proteins were predicted to localise to the nucleus, which is consistent with the canonical function of TGA proteins as bZIP transcription factors involved in transcriptional regulation. In addition, several members were predicted to localise to the chloroplast, cytoplasm or mitochondria, whereas fewer proteins were associated with the endoplasmic reticulum, Golgi apparatus, vacuoles or peroxisomes. These non-nuclear predictions may reflect overlapping targeting sequences, partial similarity to organellar transit peptides, or transient intracellular redistribution patterns reported for some plant regulatory proteins. Dual localisation patterns were also predicted for several HvTGAs, including nuclear–cytoplasmic, chloroplast–mitochondrial, nuclear–plasma membrane and endoplasmic reticulum–plasma membrane combinations, indicating possible subcellular distribution flexibility among specific HvTGA members. Since these localisation patterns were inferred exclusively from computational prediction tools, they should be considered putative and require experimental confirmation through approaches such as GFP-fusion assays, transient expression analysis, or subcellular fractionation before definitive functional assignment (Fig. 1A). Fig. 1Subcellular localisation and chromosomal distribution of HvTGA genes in barley. A Predicted subcellular localisation of HvTGA proteins obtained using Plant-mPLoc, TargetP v2.0, and WoLF PSORT, with majority consensus applied. B Chromosomal positions of HvTGA genes mapped across the seven barley chromosomes using TBtools. Gene names correspond to chromosomal order Subcellular localisation and chromosomal distribution of HvTGA genes in barley. A Predicted subcellular localisation of HvTGA proteins obtained using Plant-mPLoc, TargetP v2.0, and WoLF PSORT, with majority consensus applied. B Chromosomal positions of HvTGA genes mapped across the seven barley chromosomes using TBtools. Gene names correspond to chromosomal order Chromosome mapping showed that the 49 HvTGA genes were unevenly distributed across the seven barley chromosomes (Fig. 1B). The largest clusters were identified on chromosomes 4 H and 5 H, each containing 10 genes, followed by chromosome 3 H with 9 genes and chromosome 1 H with 7 genes. Chromosomes 2 H and 6 H contained 6 and 5 genes, respectively, whereas chromosome 7 H harboured the fewest HvTGA members, with only 2 genes. This uneven chromosomal distribution suggests a non-random expansion pattern of the HvTGA gene family during barley genome evolution. Evolutionary relationships and phylogenetic classification of TGA Genes in H. vulgare To investigate the evolutionary relationships of the TGA gene family in H. vulgare, two phylogenetic trees were constructed using full-length protein sequences. The first tree (Fig. 2A) included HvTGA proteins together with those of O. sativa (Os) and A. thaliana (At). Multiple sequence alignment was performed using MAFFT, followed by trimming of poorly aligned regions to improve phylogenetic accuracy, and the tree was generated in MEGA v12 using the maximum likelihood (ML) method with 1,000 bootstrap replicates under the LG + G+I substitution model. Based on the phylogenetic topology, TGA proteins were resolved into three major clades (Clades I–III), as indicated by the colour-coded sectors in Fig. 2A, enabling direct correspondence between clade annotation and tree structure. Fig. 2Phylogenetic analysis of TGA proteins. A Maximum likelihood phylogenetic tree of TGA proteins from Hordeum vulgare, Arabidopsis thaliana, and Oryza sativa, constructed using full-length amino acid sequences. Multiple sequence alignment was performed using MAFFT, followed by trimming of poorly aligned regions using trimAl. The tree was generated using the maximum likelihood method in MEGA v12 under the LG + G+I substitution model with 1,000 bootstrap replicates. B Maximum likelihood phylogenetic tree of HvTGA proteins showing subgroup classifications. Phylogenetic reconstruction was performed using the same alignment, trimming procedure, and model parameters as described in (A). Bootstrap support values are indicated by coloured circle Phylogenetic analysis of TGA proteins. A Maximum likelihood phylogenetic tree of TGA proteins from Hordeum vulgare, Arabidopsis thaliana, and Oryza sativa, constructed using full-length amino acid sequences. Multiple sequence alignment was performed using MAFFT, followed by trimming of poorly aligned regions using trimAl. The tree was generated using the maximum likelihood method in MEGA v12 under the LG + G+I substitution model with 1,000 bootstrap replicates. B Maximum likelihood phylogenetic tree of HvTGA proteins showing subgroup classifications. Phylogenetic reconstruction was performed using the same alignment, trimming procedure, and model parameters as described in (A). Bootstrap support values are indicated by coloured circle Clade I (blue sector) comprised a relatively small subgroup of HvTGA (HvTGA-39 and HvTGA-28) proteins that clustered tightly with a limited number of OsTGA homologues (OsTGA-15, OsTGA-19, OsTGA-50), forming a compact branch with short evolutionary distances, indicative of strong sequence conservation and potential retention of ancestral regulatory functions. Clade II (red sector) represented a distinct and well-supported lineage containing multiple HvTGA members that grouped closely with OsTGA homologues, forming several sub-branches. Notably, this clade consisted exclusively of monocot TGA proteins (barley and rice), with no direct inclusion of Arabidopsis members, suggesting lineage-specific conservation and possible functional specialisation within cereal species. Clade III (green sector) constituted the largest and most diverse group, encompassing the majority of HvTGA proteins along with numerous OsTGA members. This clade displayed extensive branching and greater phylogenetic dispersion, indicating higher sequence divergence and suggesting that it has undergone substantial evolutionary expansion. In contrast to Clades I and II, Arabidopsis TGAs were present within Clade III but were positioned on peripheral branches rather than forming tight clusters with HvTGAs, reflecting partial conservation of ancestral features alongside divergence between monocot and dicot lineages. The overall topology demonstrates that HvTGAs consistently cluster more closely with OsTGAs than with AtTGAs across all clades, reflecting the shared evolutionary origin of monocot species within the Poaceae. This pattern indicates that core TGA functions were established prior to monocot–dicot divergence, whereas subsequent lineage-specific diversification has likely contributed to functional specialisation, particularly within stress-responsive regulatory networks. To explore intra-species relationships, a second tree was generated using only the 49 HvTGA protein sequences (Fig. 2B). These were divided into four clades (Clades I–IV), with Clade IV containing the largest number of members, followed by Clades II, I, and III. The increased resolution observed in the barley-specific phylogeny reflects finer-scale diversification following species-specific gene duplication events, resulting in the subdivision of broader clades identified in the multispecies tree. Several paralogous pairs and groups were evident, including HvTGA-27/HvTGA-14, HvTGA-22/HvTGA-6, and HvTGA-23/HvTGA-5, all of which were supported by strong bootstrap values. These results suggest that gene duplication, particularly segmental duplication, was a major factor in the expansion of the HvTGA family. Importantly, members within the same clades frequently shared similar structural features, motif compositions, and expression patterns (as described in subsequent analyses), indicating that clade-specific grouping reflects not only evolutionary relatedness but also potential functional coherence. In particular, clades enriched with stress-responsive HvTGA genes may represent specialised regulatory modules involved in hormone signalling, redox regulation, and transcriptional reprogramming under abiotic stress conditions. The clade distribution in the barley-specific tree was largely consistent with the multispecies phylogeny and provided a robust framework for HvTGA classification. Together, these results indicate that the HvTGA family has evolved through a combination of conserved ancestral functions and clade-specific diversification, enabling both functional stability and adaptive flexibility in response to environmental challenges. Evolutionary origins and divergence of HvTGA gene members To investigate the evolutionary origins of the HvTGA gene family, an interspecific synteny analysis was performed using MCScan (JCVI Toolkit). A total of 29 syntenic gene pairs were identified between HvTGA and OsTGA, while only 2 pairs were found between HvTGA and AtTGA (Fig. 3A; Supplementary Table S3). The much higher number of syntenic connections with rice is consistent with the closer phylogenetic relationship between barley and other Poaceae members, while the sparse connections with Arabidopsis reflect the divergence between monocots and dicots. These results indicate that the HvTGA family shows extensive syntenic conservation within grasses but limited conservation across broader angiosperm phyla. Fig. 3Evolutionary analysis ofHvTGAgene family members inHordeum vulgare. A Interspecific synteny relationships of TGA genes between H. vulgare, Oryza sativa, and Arabidopsis thaliana. Red lines connect syntenic gene pairs, with 18 identified between barley and rice and 2 between barley and Arabidopsis. B Collinearity analysis of the HvTGA gene family within the barley genome, showing intra-chromosomal and inter-chromosomal relationships. C Proportion of different duplication types contributing to the expansion of the HvTGA family, including dispersed, whole-genome/segmental, tandem, and singleton duplications. D Distribution of Ka/Ks ratios for HvTGA gene pairs, illustrating strong purifying selection with most values < 1 and only a few > 1, suggesting limited positive selection Evolutionary analysis ofHvTGAgene family members inHordeum vulgare. A Interspecific synteny relationships of TGA genes between H. vulgare, Oryza sativa, and Arabidopsis thaliana. Red lines connect syntenic gene pairs, with 18 identified between barley and rice and 2 between barley and Arabidopsis. B Collinearity analysis of the HvTGA gene family within the barley genome, showing intra-chromosomal and inter-chromosomal relationships. C Proportion of different duplication types contributing to the expansion of the HvTGA family, including dispersed, whole-genome/segmental, tandem, and singleton duplications. D Distribution of Ka/Ks ratios for HvTGA gene pairs, illustrating strong purifying selection with most values < 1 and only a few > 1, suggesting limited positive selection In barley, analysis of intragenomic collinearity indicate that HvTGA genes are not evenly distributed across chromosomes, but frequently occur in collinear blocks (Fig. 3B). Several HvTGA clusters were linked by strong collinearity relationships, suggesting that they arose from large-scale segmental duplications. For example, HvTGA-9 and HvTGA-12 on chromosome 2 H showed distinct collinear relationships with HvTGA-36 and HvTGA-37 on chromosome 5 H, suggesting common ancestry through block-level duplication events. The retention of multiple HvTGA paralogues within such conserved regions indicates that large-scale genome rearrangements and polyploidization have played a central role in the formation of the HvTGA family. In contrast, several other HvTGA members were located outside collinear blocks, suggesting additional contributions from scattered duplications and rearrangements independent of the whole genome duplication. The HvTGA genes were further categorised by duplication type (Fig. 3C). Dispersed duplication accounted for the largest proportion (51.1%), followed by whole genome/segmental duplication (41.1%). Tandem (4.4%) and singleton duplications (3.4%) were rare, suggesting that HvTGA expansion was primarily driven by dispersed and large-scale duplications, with minimal contributions from localized tandem or single-gene events. The selective pressure acting on duplicated HvTGA gene pairs in barley was assessed by calculating Ka/Ks ratios (Fig. 3D; Supplementary Table S3). The majority of duplicated pairs exhibited Ka/Ks < 1, indicating strong purifying selection that maintains their functional integrity. Furthermore, a substantial proportion of gene pairs displayed Ka/Ks values considerably lower than 1, including many below 0.5, reflecting strong evolutionary constraints and functional conservation across duplicated HvTGA genes. Only a limited number of gene pairs exhibited Ka/Ks values greater than 1, suggesting potential instances of adaptive divergence; however, such cases were infrequent and likely represent lineage-specific functional differentiation rather than a dominant evolutionary pattern within the HvTGA family. Exploring the conserved motifs, protein domains and gene structures of HvTGA family members in H. vulgare To characterize the structural features of the HvTGA gene family, exon–intron organisation, conserved protein domains, and motif distribution were examined in the context of their clustering patterns (Fig. 4A–E). Gene structure analysis (Fig. 4B) revealed exon numbers ranging from one to eleven. Members within the same clade generally exhibited similar exon–intron architectures, indicating conservation of gene structure at the subfamily level; in contrast, Clade I displayed the greatest heterogeneity, Clades II and III were comparatively conserved, and Clade IV showed intermediate variability. These intra- and interclade patterns support the phylogenetic subdivisions. Fig. 4Phylogenetic relationships, gene structures, conserved domains, and motif composition of HvTGA family members inH. vulgare. A Phylogenetic tree of HvTGA proteins constructed using maximum likelihood analysis. B Exon–intron structures of HvTGA genes, with exons represented by green boxes, untranslated regions (UTRs) by yellow boxes, and introns by black lines. C Conserved protein domains of HvTGA proteins predicted by CDD, with domain types indicated by coloured boxes. D Conserved motifs of HvTGA proteins identified by MEME, with each motif represented by a coloured box. E Sequence logos of the ten conserved motifs generated by MEME, with corresponding motif numbers, widths, E-values, and occurrence data Phylogenetic relationships, gene structures, conserved domains, and motif composition of HvTGA family members inH. vulgare. A Phylogenetic tree of HvTGA proteins constructed using maximum likelihood analysis. B Exon–intron structures of HvTGA genes, with exons represented by green boxes, untranslated regions (UTRs) by yellow boxes, and introns by black lines. C Conserved protein domains of HvTGA proteins predicted by CDD, with domain types indicated by coloured boxes. D Conserved motifs of HvTGA proteins identified by MEME, with each motif represented by a coloured box. E Sequence logos of the ten conserved motifs generated by MEME, with corresponding motif numbers, widths, E-values, and occurrence data Protein domain annotation (Fig. 4C) confirmed the presence of the canonical bZIP core in all HvTGA sequences, supported by conserved domain models (bZIP_1, bZIP_2, and bZIP_superfamily/bZIP_2_superfamily). A minority of sequences also matched non-bZIP domains (such as DOG1-like or Jnk-SapK_ap_N), but these occurrences were sporadic and not clade-specific. The inconsistent presence of DOG1-like domains among HvTGA members is consistent with previous reports indicating that this feature is not universally conserved within the TGA subfamily and therefore represents a lineage-specific or auxiliary characteristic rather than a defining domain. Thus, HvTGA proteins are primarily defined by a conserved bZIP scaffold with limited additional signatures. Motif analysis identified ten conserved motifs across the HvTGA family (Fig. 4D). Motifs 1–6 and 8 were broadly distributed, whereas motifs 7, 9, and 10 were confined to specific subgroups. Motif 1 was the most ubiquitous and statistically significant (E-value 1.5 × 10⁻¹⁶⁹⁷; width 50 aa), overlapping the bZIP/basic leucine zipper region in the aligned sequences. Several motifs located in the C-terminal regions corresponded to low-complexity, glutamine-enriched segments, consistent with the presence of Q-rich regions previously associated with transcriptional activation domains in TGA transcription factors. Although the precise boundaries of these regions varied among members, their recurrent detection indicates conserved compositional features characteristic of TGA proteins rather than strictly conserved domain architecture. Motif composition and arrangement were highly conserved within clades, consistent with clustering patterns, while subgroup-specific motifs likely reflect functional diversification. Sequence logo analysis (Fig. 4E) further highlighted conserved residues within the bZIP region, consistent with essential roles in DNA binding and dimerization. The concordance between motif architecture, domain composition, and phylogenetic grouping provides integrated structural support for the classification of HvTGA proteins. Together, these findings demonstrate that HvTGA proteins share a conserved bZIP domain and common motif repertoire, while clade-specific exon–intron structures and selective motif variation underpin subgroup-level diversification in barley. Cis-acting regulatory element analysis of HvTGA genes The analysis of cis-acting regulatory elements in promoter regions provides critical insights into transcriptional regulation and the functional diversification of gene families. To characterize HvTGA promoters, the 2 kb upstream sequences were examined using PlantCARE and cross-validated against JASPAR, PlantPAN, and PlantRegMap. This integrated approach confirmed most motifs detected by PlantCARE and identified additional transcription factor binding sites (TFBSs) supported by at least two platforms (Supplementary Table S4). The identified cis-elements were grouped into four functional categories: development-related, hormone-responsive, stress-responsive, and light-responsive (Fig. 5A, B; Supplementary Table S4). Fig. 5Cis-regulatory element analysis of HvTGA promoter regions. A Heatmap representation of cis-acting regulatory elements identified in the 2 kb upstream promoter regions of HvTGA genes. Elements are grouped by functional categories, and the phylogenetic relationships of HvTGA genes are shown on the left. B Bar plot summarising the distribution of cis-elements across HvTGA promoters, classified into four major categories: cell cycle and development, hormone responsiveness, light responsiveness, and stress responsiveness Cis-regulatory element analysis of HvTGA promoter regions. A Heatmap representation of cis-acting regulatory elements identified in the 2 kb upstream promoter regions of HvTGA genes. Elements are grouped by functional categories, and the phylogenetic relationships of HvTGA genes are shown on the left. B Bar plot summarising the distribution of cis-elements across HvTGA promoters, classified into four major categories: cell cycle and development, hormone responsiveness, light responsiveness, and stress responsiveness Within the development-related group, motifs such as the CAT-box, GCN4 motif, AAGAA motif, and O2-site were frequently identified, indicating potential roles in meristem activity, differentiation, and seed development. Certain members, including HvTGA-2 and HvTGA-9, contained higher densities of these elements. Hormone-responsive motifs were broadly distributed among HvTGA promoters. ABA-responsive elements (ABRE variants) were notably enriched in HvTGA-21 and HvTGA-28, whereas jasmonate-responsive motifs (TGACG, CGTCA) were most abundant in HvTGA-18 and HvTGA-24. Auxin-responsive motifs (TGA-element, AuxRR-core) were identified in HvTGA1 and HvTGA7, while gibberellin (P-box, GARE, TATC-box), salicylic acid (TCA), and ethylene (ERE) elements were present in multiple promoters. Stress-responsive motifs constituted the most abundant category overall (Fig. 5B). Nearly all HvTGA promoters harboured multiple stress-related cis-elements, including drought-responsive (MBS), cold-responsive (LTR), and defense-associated (W-box) motifs. Promoters such as those of HvTGA-8, HvTGA-18, HvTGA-24, and HvTGA-35 contained particularly dense clusters of stress-related motifs, with MBS and W-box being the most frequent. Light-responsive motifs were also widespread, though generally less abundant than stress-related motifs. Box 4 was nearly ubiquitous, often occurring in multiple copies, followed by the G-box, GT1-motif, and GATA-motif. Promoters of HvTGA-2, HvTGA-27, and HvTGA-46 exhibited relatively high densities of light-responsive motifs, suggesting regulatory roles linked to light signalling. Overall, stress-related cis-elements predominated across HvTGA promoters, followed by light-, hormone-, and development-related motifs. The heatmap and bar chart (Fig. 5) clearly illustrate these distributions, underscoring the complexity and diversity of regulatory inputs shaping HvTGA gene expression in barley. Secondary, tertiary structural and PPI interaction analysis of TGA proteins in H. vulgare Proteins exert their regulatory functions through defined structural organisation; therefore, secondary and tertiary structure prediction can provide valuable preliminary insight into their potential functional properties. Analysis of HvTGA proteins revealed that α-helices and random coils were the predominant secondary structural elements, together accounting for more than 75% of residues in most family members (Supplementary Table S5). The proportion of α-helical content showed considerable variation, ranging from 17.83% (HvTGA-2) to 79.17% (HvTGA-9), whereas random coil content displayed an inverse trend (14.60–72.56%). Extended strands (0.39–7.96%) and β-turns (0–2.65%) contributed only marginally across the family. Although α-helical enrichment formed the dominant structural scaffold, several members such as HvTGA-2, HvTGA-11, HvTGA-36, and HvTGA-47 exhibited comparatively high random coil proportions, suggesting structural heterogeneity within the family. The predominance of α-helical structures is consistent with the conserved architecture of basic leucine zipper (bZIP) transcription factors, in which the leucine zipper domain forms a coiled-coil α-helical structure that mediates homo- or heterodimerisation, while the adjacent basic region is responsible for sequence-specific DNA binding. In the predicted HvTGA structures, this organisation is reflected by a conserved central α-helical segment corresponding to the bZIP domain, which integrates both the DNA-binding basic region and the dimerisation interface. In this context, variation in α-helical content among HvTGA proteins may reflect differences in the stability or extent of the coiled-coil dimerisation interface, potentially influencing DNA-binding affinity and partner selectivity. Conversely, regions enriched in random coils are often associated with intrinsically disordered segments that contribute to regulatory flexibility, facilitating protein–protein interactions and dynamic responses to environmental cues. The elevated coil content observed in specific HvTGA members may therefore indicate expanded regulatory regions outside the conserved bZIP core, potentially enabling interaction with diverse signalling components involved in stress-responsive transcriptional regulation. Such intrinsically disordered regions are frequently implicated in modulating transcriptional activation potential and interaction with NPR1-like co-regulators, which are central to TGA-mediated signalling pathways. Tertiary structures generated using AlphaFold2 indicated that HvTGA proteins are predominantly composed of helical bundles, consistent with the conserved architecture of bZIP transcription factors (Fig. 6A). Members belonging to the same phylogenetic clades generally displayed similar three-dimensional conformations, supporting the relationship between structural conservation and evolutionary grouping. Notably, the predicted structures revealed a conserved central α-helical region corresponding to the bZIP domain across all HvTGA proteins, whereas greater structural variability was observed in the N-terminal and C-terminal regions, which are typically implicated in transcriptional activation, repression, or interaction with co-regulatory proteins. This structural partitioning between a conserved DNA-binding/dimerisation core and variable regulatory termini is consistent with the modular organisation of plant bZIP transcription factors and may underlie functional specialisation among HvTGA members. Fig. 6Predicted tertiary structures and protein–protein interaction (PPI) network of HvTGA proteins. A Representative tertiary structures of HvTGA proteins predicted using AlphaFold2, aligned with their respective phylogenetic clades. B Predicted PPI network of HvTGA proteins generated using the STRING database, based on orthology with Arabidopsis thaliana. Nodes represent proteins, with filled nodes indicating known or predicted 3D structures. Edges denote protein–protein associations derived from multiple evidence sources, including co-expression, database annotations, and experimental datasets Predicted tertiary structures and protein–protein interaction (PPI) network of HvTGA proteins. A Representative tertiary structures of HvTGA proteins predicted using AlphaFold2, aligned with their respective phylogenetic clades. B Predicted PPI network of HvTGA proteins generated using the STRING database, based on orthology with Arabidopsis thaliana. Nodes represent proteins, with filled nodes indicating known or predicted 3D structures. Edges denote protein–protein associations derived from multiple evidence sources, including co-expression, database annotations, and experimental datasets In contrast, several proteins, including HvTGA-2, HvTGA-9, and HvTGA-12, showed altered helical-to-coil ratios, indicating local divergence within otherwise conserved folds. Such structural variation, particularly outside the core DNA-binding and dimerisation domains, may contribute to functional diversification by modulating interaction specificity, transcriptional regulatory capacity, or responsiveness to abiotic stress signals. In particular, variation in non-core regions may influence the recruitment of co-factors and integration into stress-responsive transcriptional networks. As these structural models were derived from computational prediction rather than experimentally resolved protein structures, they should be interpreted as putative conformational frameworks requiring further validation through biochemical or structural approaches. Protein–protein interaction (PPI) analysis using STRING, based primarily on Arabidopsis orthologues and integrated database evidence, provided a predicted framework of conserved regulatory associations evidence of physical interaction in barley (Fig. 6B). Several HvTGA members, including HvTGA-13, HvTGA-26, HvTGA-28, HvTGA-30, HvTGA-41, and HvTGA-48, occupied central positions within the inferred network. Predicted associations included GRX proteins (GRXC7, GRXC8) and NPR regulators (NPR5), which are known to participate in redox-regulated DNA binding and defence signalling. Additional candidate interactions involved bZIP proteins (BZIP1, BZIP8), stress-responsive transcription factors (DREB1A, NAC096), and kinases (SRK2D, MUK11.16). Within this predicted network, BZIP8 appeared as a highly connected hub linking multiple HvTGA proteins with external regulators. Similarly, HvTGA-26 and HvTGA-28 were associated with both NPR5 and GRX proteins, suggesting possible functional redundancy or cooperative regulation. These interaction patterns provide useful hypotheses for functional investigation; however, direct barley-specific validation using approaches such as yeast two-hybrid analysis, co-immunoprecipitation, or bimolecular fluorescence complementation will be necessary to confirm these proposed regulatory relationships. The combined structural and interaction analyses therefore suggest both conserved regulatory modules and subgroup-specific diversification within the HvTGA family. Interaction between HvTGA genes and miRNAs MicroRNAs (miRNAs) are important post-transcriptional regulators that modulate gene expression through transcript cleavage or translational inhibition. To investigate their potential regulatory relationships with HvTGA genes, a genome-wide prediction of miRNA–target interactions was performed. In total, 31 distinct barley miRNAs were predicted to target 29 HvTGA genes, resulting in 54 unique interaction pairs (Supplementary Table S6; Fig. 7). Most predicted interactions were associated with cleavage, whereas a smaller proportion were associated with translational inhibition. Fig. 7Predicted interactions between barley miRNAs and HvTGA genes. A Sankey diagram illustrating the predicted regulatory network between 31 H. vulgare miRNAs (left) and 29 HvTGA genes (right). Interactions were identified using psRNATarget with default parameters, integrating information on cleavage- and translation-based inhibition modes. Distinct colours denote individual miRNA–target relationships Predicted interactions between barley miRNAs and HvTGA genes. A Sankey diagram illustrating the predicted regulatory network between 31 H. vulgare miRNAs (left) and 29 HvTGA genes (right). Interactions were identified using psRNATarget with default parameters, integrating information on cleavage- and translation-based inhibition modes. Distinct colours denote individual miRNA–target relationships Among the predicted regulators, hvu-miR6192 exhibited the broadest targeting range, with predicted interactions involving at least eight HvTGA members, including HvTGA-22, HvTGA-25, HvTGA-29, HvTGA-30, HvTGA-34, HvTGA-37, HvTGA-39, and HvTGA-42/47. Additional broadly acting miRNAs included hvu-miR6184 (HvTGA-1, HvTGA-4, HvTGA-25, HvTGA-36), hvu-miR5053 (HvTGA-1, HvTGA-19, HvTGA-20, HvTGA-28), and members of the hvu-miR5049 family (b/e/f), collectively targeting HvTGA-32, HvTGA-35, HvTGA-39, and HvTGA-43. hvu-miR6214 was also predicted to interact with HvTGA-29, HvTGA-34, and HvTGA-35. Several of these miRNA families have previously been associated with stress-responsive signalling and transcriptional regulation in cereals, suggesting potential involvement in HvTGA-mediated adaptive pathways [53]. These miRNAs may represent important candidate regulators within the HvTGA regulatory network, although confirmation of these interactions will require experimental approaches such as degradome sequencing, 5′ RLM-RACE, or transient reporter assays. Several miRNAs displayed relatively narrow target specificity. For example, hvu-miR159a/b targeted HvTGA-16, hvu-miR171-5p was predicted only for HvTGA-2, and hvu-miR1436 was associated with HvTGA-39. The hvu-miR444 family showed lineage-specific targeting of HvTGA-12 and HvTGA-24, consistent with its reported role in transcription factor regulation in grasses. In addition, miRNAs such as hvu-miR6176, hvu-miR6178, hvu-miR6180, hvu-miR6181, hvu-miR6190, and hvu-miR6210 were each predicted to regulate only one or two HvTGA members. The predicted regulatory modes varied across interaction pairs. Cleavage was predominant, as observed for hvu-miR6192 targeting HvTGA-25, HvTGA-30, and HvTGA-37, whereas translational inhibition was predicted for specific interactions such as hvu-miR5049b/e with HvTGA-43, hvu-miR5053 with HvTGA-1, HvTGA-19, HvTGA-20, and HvTGA-28, and hvu-miR6181 with HvTGA-4 and HvTGA-22. These results suggest the coexistence of multiple candidate post-transcriptional regulatory mechanisms across the HvTGA family. Expression dynamics of HvTGA genes under heat stress To investigate the expression profiles of HvTGA genes under heat stress, RNA-seq datasets generated from barley shoot and root tissues exposed to elevated temperature were reanalysed. Differentially expressed genes were defined using the thresholds of absolute log2 fold change (|log2FC|) ≥ 1 and false discovery rate (FDR)-adjusted p-value < 0.05. Among the identified differentially expressed genes, 27 HvTGA family members showed distinct and tissue-specific transcriptional responses to heat stress (Fig. 8A). In the shoot tissues, several HvTGA genes showed consistent induction after heat exposure. In particular, HvTGA-1, HvTGA-10, HvTGA-29, HvTGA-39, HvTGA-43 and HvTGA-48 showed significant upregulation under stress conditions compared to control samples. Among these, HvTGA-29 showed the highest and most consistent transcript accumulation in heat-treated samples. This gene was therefore prioritised based on the magnitude and consistency of its differential expression across analysed conditions. In contrast, genes such as HvTGA-28 and HvTGA-31 were significantly down-regulated under heat stress, suggesting that their repression may contribute to adaptive stress signalling. Several other genes, including HvTGA-16, HvTGA-17, HvTGA-30, HvTGA-34, HvTGA-37, and HvTGA-47, exhibited stable transcript accumulation across treatments, indicative of constitutive expression patterns with limited sensitivity to heat perturbation (Fig. 8A). Fig. 8Expression pattern of HvTGA gene family across different stress and tissues. A, B and C represent the expression profile of HvTGA gene family under normal and heat stress condition for root, normal and salt stress condition in root samples, and normal and osmotic stress conditions for shoot tissues, respectively Expression pattern of HvTGA gene family across different stress and tissues. A, B and C represent the expression profile of HvTGA gene family under normal and heat stress condition for root, normal and salt stress condition in root samples, and normal and osmotic stress conditions for shoot tissues, respectively In root tissues, the transcriptional reprogramming of HvTGA genes was more pronounced, emphasizing the importance of belowground organs in coordinating adaptive responses to thermal stress. Strong induction was observed for HvTGA-1, HvTGA-10, HvTGA-29, HvTGA-33, HvTGA-39, HvTGA-43, and HvTGA-48, with fold changes ranging from two- to fourfold relative to controls. HvTGA-29 again showed strong and sustained induction under heat stress, together with HvTGA-33 and HvTGA-43, indicating stable transcriptional responsiveness across both tissues. Conversely, HvTGA-12, which was highly expressed under control conditions, exhibited a dramatic downregulation following heat stress, suggesting its potential role as a negative regulator of stress adaptation. Similarly, HvTGA-26 and HvTGA-28 were down-regulated under heat treatment compared to control roots, indicating selective suppression of specific TGA-mediated pathways (Fig. 8A). Expression profiles of HvTGA genes in root tissue under salt stress To investigate the expression profiles of HvTGA genes under salt stress, RNA-seq datasets generated from barley root tissues exposed to salinity were reanalysed. Genes showing |log2FC| ≥ 1 with FDR-adjusted p-values < 0.05 were considered significantly differentially expressed. Among the identified differentially expressed genes, 40 HvTGA genes out of the 49 identified family members showed distinct transcriptional responses to salt stress (Fig. 8B). Analysis of HvTGA transcription factors in barley root tissues exposed to salt stress revealed distinct temporal expression patterns, suggesting diverse roles in early signaling, sustained regulation, and long-term adaptation to ionic stress. Several HvTGA genes were significantly induced in root tissue, with transcript abundance increasing as the duration of stress extended. Of which, HvTGA-10, HvTGA-12, HvTGA-14, HvTGA-21, HvTGA-29, HvTGA-33, and HvTGA-39 exhibited marked upregulation compared to the controls. HvTGA-29 showed the strongest and most sustained induction, with elevated expression observed from 1 h and maintained through 12 h of salt exposure. Similarly, HvTGA-12 and HvTGA-14 displayed robust induction, with expression increasing sharply within the first hour and reaching their highest levels at 12 h. HvTGA-33 and HvTGA-21 also showed strong late-stage induction, reaching maximum expression levels at 12 h, indicating involvement in prolonged regulatory responses in the root system. A distinct group of HvTGA genes exhibited transient activation at early stages of salt stress (Fig. 8B). HvTGA-17 and HvTGA-31 were rapidly upregulated within 1 h but declined sharply thereafter, with expression nearly abolished by 12 h. This pattern suggests these genes likely participate in early perception and rapid signaling mechanisms within the root but are subsequently downregulated as the stress response stabilizes. HvTGA-41 displayed a similar early induction but was suppressed at later time points. In contrast, several HvTGA genes were progressively downregulated under salt stress. HvTGA-24 and HvTGA-48, both moderately expressed under control conditions, exhibited a consistent decline across all time points, suggesting functional repression during root adaptation to high salinity. Likewise, HvTGA-19, HvTGA-34, and HvTGA-38 demonstrated reduced expression at later time points, indicating selective silencing of certain transcriptional pathways under prolonged stress. Interestingly, a subset of genes, including HvTGA-26 and HvTGA-44, displayed biphasic expression profiles (Fig. 8B). Both genes showed modest activity at early stages of stress but underwent strong induction at 12 h (up to approximately 101 and 47 transcripts, respectively), implying a role in late-stage regulatory processes associated with long-term root adaptation to ionic and osmotic imbalances. These results demonstrate that HvTGA genes in barley roots exhibit diverse and temporally distinct expression patterns under salt stress. Genes showing sustained induction, particularly HvTGA-29, HvTGA-12, HvTGA-14, and HvTGA-33, may contribute to prolonged transcriptional regulation during salt adaptation, whereas early transiently induced genes such as HvTGA-17 and HvTGA-31 may participate in initial stress perception and signalling events. Expression profiles of HvTGA genes in shoot tissue under osmotic stress To investigate the expression profiles of HvTGA genes under osmotic stress, RNA-seq datasets generated from barley shoot tissues exposed to osmotic stress were reanalysed. Differential expression significance was determined using the criteria of |log2FC| ≥ 1 and FDR-adjusted p-value < 0.05. Among the identified differentially expressed genes, 24 HvTGA genes out of the 49 identified family members showed distinct transcriptional responses under osmotic stress conditions (Fig. 8C). Expression profiling of HvTGA genes under osmotic stress in shoot tissues revealed different and gene-specific transcriptional responses. A subset of HvTGA members showed pronounced induction, indicating their active involvement in osmotic stress signalling pathways. Among these, HvTGA-29 showed the strongest upregulation. Similarly, HvTGA-43 showed an almost twofold increase under stress conditions compared to the control condition. HvTGA-41 and HvTGA-38 were also strongly upregulated and showed approximately twofold higher expression in stressed shoots compared with controls. These genes were prioritised based on the magnitude of differential expression and consistent directional changes across analysed conditions. In particular, HvTGA-29 displayed strong and consistent upregulation under heat, salt, and osmotic stress, supporting its identification as a high-confidence candidate for further functional investigation. Moderate down-regulation was also observed for HvTGA-12, HvTGA-16, HvTGA-26, HvTGA-34, HvTGA-48, HvTGA-5 and HvTGA-7 (Fig. 8C). These genes may contribute to fine-tuning transcriptional reprogramming and improving stress tolerance through complementary regulatory mechanisms. In contrast, the expression of HvTGA-25 was reduced in treated shoots compared to controls, suggesting a possible role in the negative regulation or redistribution of resources during stress adaptation. Interestingly, several genes with relatively low basal expression levels, including HvTGA-21, HvTGA-28, HvTGA-33, HvTGA-37, and HvTGA-44, showed modest but reproducible induction under stress conditions (Fig. 8C). Although their absolute expression levels remained low, the consistent trend towards upregulation suggests that these members may function as conditionally activated regulators or contribute to secondary layers of stress-responsive signalling. Taken together, these results indicate coordinated transcriptional reprogramming of HvTGA genes in response to osmotic stress in shoot tissue. Strongly induced genes such as HvTGA-29, HvTGA-43, HvTGA-41, and HvTGA-38 represent suitable candidates for further functional characterisation, while moderate and low responses among other HvTGA members suggest the presence of both primary and auxiliary regulatory components within the broader stress-response network. GO enrichment analysis of members of the HvTGA family To investigate the potential functional roles of HvTGA genes in H. vulgare, Gene Ontology (GO) enrichment analysis was performed across biological process (BP), molecular function (MF), and cellular component (CC) categories (Fig. 9A; Table S7). In the BP category, the most significant enrichments were associated with transcriptional regulation, including “regulation of transcription, DNA-templated” (GO:0006355; 33 genes) and “transcription, DNA-templated” (GO:0006351; 11 genes). Stress- and hormone-related terms were also prominently enriched, particularly “abscisic acid-activated signalling pathway” (GO:0009738; 9 genes), “response to abscisic acid” (GO:0009737; 11 genes), and “response to salt stress” (GO:0009651; 5 genes), indicating the likely involvement of HvTGA members in hormone-mediated stress adaptation. Within the MF category, enrichment was dominated by canonical transcription factor functions, including “DNA binding” (GO:0003677; 37 genes), “DNA-binding transcription factor activity” (GO:0003700; 31 genes), and “sequence-specific DNA binding” (GO:0043565; 27 genes). Terms associated with dimerisation and protein interactions, such as “protein dimerisation activity” (GO:0046983; 8 genes), were also significantly enriched, consistent with the functional characteristics of bZIP transcription factors. In the CC category, the strongest enrichment was observed for “nucleus” (GO:0005634; 34 genes), supporting the predominant nuclear localisation expected for transcriptional regulators. Fig. 9Gene Ontology (GO) and KEGG pathway enrichment analysis ofHvTGAfamily members inHordeum vulgare. A Bubble plot of enriched GO terms. Bubble size indicates the number of genes per term, bubble colour represents significance level (–log₁₀ p-value), and bubble shape corresponds to GO category (circle = biological process [BP], square = molecular function [MF], triangle = cellular component [CC]). B KEGG pathway enrichment analysis of HvTGA genes. Bubble size reflects the number of mapped genes per pathway, and bubble colour indicates the false discovery rate (FDR) Gene Ontology (GO) and KEGG pathway enrichment analysis ofHvTGAfamily members inHordeum vulgare. A Bubble plot of enriched GO terms. Bubble size indicates the number of genes per term, bubble colour represents significance level (–log₁₀ p-value), and bubble shape corresponds to GO category (circle = biological process [BP], square = molecular function [MF], triangle = cellular component [CC]). B KEGG pathway enrichment analysis of HvTGA genes. Bubble size reflects the number of mapped genes per pathway, and bubble colour indicates the false discovery rate (FDR) KEGG pathway analysis (Fig. 9B) further supported these findings, with significant enrichment in “plant hormone signal transduction” (ko04075), “MAPK signalling pathway – plant” (ko04016), and “carotenoid biosynthesis” (ko00906). These pathways suggest that HvTGA genes are closely associated with hormone signalling, stress-responsive phosphorylation cascades, and cellular protection mechanisms under adverse environmental conditions. Several HvTGA genes showing strong transcriptional induction under abiotic stress treatments, including HvTGA-13, HvTGA-29, HvTGA-28, HvTGA-30, HvTGA-41, HvTGA-38 and HvTGA-43 were also associated with stress-related GO terms and KEGG pathways, providing functional support for their potential regulatory roles. These genes were linked to “response to abscisic acid” (GO:0009737), “abscisic acid-activated signalling pathway” (GO:0009738), and “response to salt stress” (GO:0009651), suggesting their involvement in ABA-dependent osmotic stress adaptation under PEG-, NaCl-, and ABA-induced stress conditions. Similarly, enrichment in “plant hormone signal transduction” and “MAPK signalling pathway – plant” indicates that these stress-responsive HvTGAs may function at the interface between signal perception and downstream transcriptional reprogramming. This functional convergence between expression profiling and enrichment analysis supports the presence of coordinated HvTGA regulatory modules involved in stress perception, hormonal signalling, redox homeostasis, and adaptive transcriptional regulation during abiotic stress responses in barley. These candidate genes therefore represent strong targets for future functional validation and for the development of stress-resilient barley germplasm. Assessment of differential expression robustness of HvTGA genes The discriminatory capacity of differentially expressed HvTGA genes under abiotic stress conditions was evaluated using leave-one-out cross-validation (LOOCV) applied to their normalised transcript abundance profiles. In this approach, each sample was sequentially excluded and used for testing while the model was trained on the remaining data, enabling an assessment of classification performance in datasets with limited biological replicates. LOOCV analysis yielded an average classification accuracy of 89.4%, indicating that HvTGA expression profiles reliably distinguish between heat-, salt-, and osmotic stress conditions and their respective controls. Several HvTGA genes, including HvTGA-29, HvTGA-43, HvTGA-41, and HvTGA-38, showed relatively higher contributions to classification performance, reflecting consistent variation in their expression across stress treatments. In contrast, other members such as HvTGA-17 and HvTGA-31 displayed more variable or tissue-specific expression patterns, corresponding to lower contributions to classification outcomes. These results indicate the presence of stable and stress-associated transcriptional signatures among specific HvTGA members. Discussion In plants, transcriptional regulation is central to growth, development and adaptation to environmental challenges [3, 54]. These processes are based on complex regulatory networks in which transcription factors (TFs) serve as key nodes [55, 56]. TGA proteins, a subgroup of the bZIP superfamily, are known regulators of developmental and stress-related processes, and their structural organisation and functional roles have been well characterised in model species [57, 58]. Nevertheless, the TGA family has not been extensively studied in barley. In this study, we identified 49 HvTGA genes in the barley genome. Remarkably, the HvTGA repertoire is larger than that of rice (~ 30 members [20]) but smaller than that of hexaploid wheat (> 60 members [26]), reflecting the intermediate genomic complexity of barley within the Triticeae. Chromosome mapping showed that the HvTGAs are unevenly distributed across the seven barley chromosomes, with large clusters on chromosomes 4 H, 5 H and 3 H. This non-random distribution suggests subgenome divergence and lineage-specific retention, as has been observed in other transcription factor families in cotton [59, 60]. Phylogenetic analysis grouped the HvTGAs, OsTGAs and AtTGAs into three major clades, each containing members from all three species, suggesting that diversification occurred before the separation of monocotyledons and dicotyledons. Within barley, HvTGAs were further subdivided into four clades, reflecting both conserved and lineage-specific expansions following species divergence. The closer grouping of the HvTGAs with the OsTGAs than with the AtTGAs emphasises the evolutionary proximity of Poaceae members, while the separation of the Arabidopsis TGAs highlights the divergence between monocotyledons and dicotyledons [61]. Notably, the distribution of HvTGA genes across distinct clades was not random but corresponded with differences in gene structure, conserved motif composition, and stress-responsive expression patterns observed in subsequent analyses. Members within the same clades frequently shared similar exon–intron organisation and conserved motifs associated with bZIP-mediated DNA binding and dimerisation, indicating that clade classification reflects underlying structural conservation. At the same time, clade-specific motif variation suggests regulatory divergence that may influence target gene specificity and interaction with signalling pathways. Functionally, these clade-specific patterns are indicative of transcription factor specialisation within the HvTGA family. In particular, clades enriched with stress-inducible HvTGA members appear to be associated with hormone-mediated signalling pathways, including abscisic acid (ABA) and MAPK-dependent responses, as supported by integrated expression and enrichment analyses. Such organisation is consistent with previous studies demonstrating that bZIP/TGA transcription factors form modular regulatory networks that coordinate redox balance, hormonal signalling, and transcriptional reprogramming under abiotic stress conditions [62, 63]. Furthermore, the presence of monocot-restricted clustering patterns, particularly within specific clades lacking Arabidopsis homologues, suggests that certain HvTGA subgroups may have undergone lineage-specific functional adaptation following the divergence of cereal species. This adaptive diversification likely reflects the evolutionary pressure to fine-tune transcriptional regulation in response to distinct environmental conditions encountered by grasses. This phylogenetic framework suggests strong purifying selection that maintains key regulatory functions, with clade-specific diversification contributing to functional specialisation within barley [64]. Gene duplication represents a major evolutionary force in plant genomes and drives diversification and functional innovation [65]. In barley, our analysis revealed that the HvTGA family (49 members) expanded mainly through segmental (41.1%) and dispersed duplications (51.1%), with only minor contributions from tandem and singleton events. This pattern is consistent with the TGAs of wheat [26], sugarcane [66] and the bZIP families of rice [67], suggesting that dispersed duplications and block-level duplications are the major drivers of TGA evolution in grasses. Multiple synteny analysis also revealed extensive conservation between the TGAs of barley and rice, but only limited links to Arabidopsis, emphasising the evolutionary cohesion of cereals and their divergence from dicotyledons [68]. Within barley, segmentally duplicated blocks have retained multiple paralogues, indicating that large-scale chromosomal rearrangements were central to shaping family architecture [69]. Analysis of selection pressure showed that HvTGA paralogues are predominantly constrained by purifying selection (Ka/Ks < 1), indicating strong functional conservation of their regulatory functions [70]. Most pairs had ratios well below 0.5, indicating exceptional evolutionary stability and reflecting the conservation reported for TGAs in Brassica napus [16] and sunflower [71] TGAs. Only a few gene pairs deviated from this pattern and had Ka/Ks values greater than 1, suggesting positive selection that may have promoted subfunctionalisation or occasional neofunctionalisation within particular clades. Taken together, these results reveal a family that largely retains essential functions while allowing limited adaptive divergence, a trend that has also been observed in soybean TGAs [13]. Exon–intron organisation, conserved domains, and motif composition are central determinants of transcription factor evolution and functional diversification [72]. The HvTGA genes display substantial variability in exon–intron architecture, ranging from single-exon to multi-exon structures. Nevertheless, members within the same phylogenetic clades generally exhibit highly similar arrangements, indicating that gene structure has been maintained under evolutionary constraints. The presence of both intron-rich and intron-poor genes suggests a history of intron gain and loss, a phenomenon widely documented among plant transcription factor families [73–75]. Such structural variation may have functional consequences, influence transcriptional efficiency, alternative splicing, and regulatory flexibility, and thereby enhance the adaptive capacity of HvTGAs during development and in response to stress [76–78]. Domain analysis corroborated these observations. All HvTGA proteins contain the conserved bZIP domain, consistent with their role as transcriptional regulators and in agreement with previous reports in Arabidopsis and Brassica species [8, 16]. Only a small number of sequences displayed weak similarity to additional domains, such as DOG1 or Ty3_capsid, without strong support. This scarcity of auxiliary domains suggests that functional diversification within the HvTGA family is more likely driven by variation in exon–intron organisation and regulatory motifs rather than by domain acquisition, a trend similarly reported in other transcription factor families [79, 80]. Motif analyses further highlight this interplay between conservation and divergence. Core motifs associated with the bZIP domain are broadly conserved, reflecting their essential roles in DNA binding and dimerisation. By contrast, motifs 7, 9, and 10 are restricted to specific phylogenetic clades, indicative of subgroup-specific evolution. Comparable clade-restricted motifs have been identified in Brassica napus and sunflower TGAs, where they contribute to adaptive responses to abiotic stress [16, 71, 81]. The presence of such motifs in barley likely underpins functional diversification, enabling subsets of HvTGA genes to modulate transcriptional responses in a lineage- and context-dependent manner. Protein structure forms the molecular foundation for transcription factor function, with conserved folding patterns stabilising DNA binding and protein–protein interactions [82]. HvTGA proteins are predominantly composed of α-helices and random coils, consistent with the helical-rich bZIP domains observed in Arabidopsis thaliana, which are essential for dimerisation and promoter recognition [8, 83]. While most HvTGAs conform to this canonical structural scaffold, several members, including HvTGA-2, HvTGA-9, and HvTGA-12, exhibit altered helix-to-coil ratios, indicating local structural divergence that may contribute to subgroup-specific regulatory properties. AlphaFold2-based tertiary structure modelling further indicated that members within the same phylogenetic clade generally share comparable three-dimensional conformations, supporting the relationship between structural conservation and evolutionary grouping. As these models were generated computationally rather than derived from experimentally resolved protein structures, they are best considered putative conformational frameworks that provide functional hypotheses requiring validation through biochemical, biophysical, or crystallographic approaches. Protein–protein interaction (PPI) analysis using STRING, based primarily on Arabidopsis orthologues and integrated functional association datasets, provided a predicted framework of conserved regulatory relationships rather than direct evidence of physical interaction in barley [84]. Several HvTGAs, including HvTGA-13, HvTGA-26, HvTGA-28, HvTGA-30, HvTGA-41, and HvTGA-48, occupied central positions within the inferred interaction network and showed predicted associations with GRX proteins (GRXC7, GRXC8), NPR regulators (NPR5), and stress-responsive transcription factors (DREB1A, NAC096). These predicted associations are consistent with the well-characterised Arabidopsis TGA–GRX and TGA–NPR1 regulatory modules that integrate redox and hormonal signalling during stress responses [85–88]. Additional candidate interactions involving kinases (SRK2D, MUK11.16) and other bZIP proteins (BZIP8) further suggest that HvTGAs may participate in broader transcriptional networks integrating phosphorylation-mediated signalling with cooperative transcriptional regulation, as reported for bZIP factors in rice and wheat, where kinase interactions influence DNA-binding activity and stress-responsive gene expression [89]. Given that the STRING network is inferred largely through orthology-based evidence and computational integration, these associations should be regarded as candidate functional relationships for further investigation. Direct barley-specific validation using yeast two-hybrid assays, co-immunoprecipitation, or bimolecular fluorescence complementation will be necessary to confirm these proposed regulatory interactions. The integration of structural variation and PPI connectivity therefore suggests both evolutionary conservation and subgroup-specific diversification within the HvTGA family. Members with subtle structural deviations yet occupying central network positions may provide lineage-specific regulatory flexibility, enabling fine-tuned modulation of developmental and stress-responsive pathways in barley. This dual pattern of conservation and divergence is consistent with observations in other crop bZIP families, highlighting the close relationship between protein structure, interaction networks, and functional plasticity [90, 91]. MicroRNAs (miRNAs) constitute a critical post-transcriptional regulatory layer controlling transcription factor activity, where conserved families maintain developmental stability and lineage-specific families support adaptive innovation [92–94]. In barley, 31 distinct miRNAs were predicted to target 29 HvTGA genes, forming 54 unique interaction pairs, with cleavage representing the predominant regulatory mode. These results suggest that HvTGAs are embedded within a complex miRNA regulatory network integrating both conserved and lineage-specific signals. Conserved miRNAs, including hvu-miR156, hvu-miR166, and hvu-miR396, emerged as prominent candidate regulators of HvTGAs, consistent with their established roles in plant development and abiotic stress responses [95–97]. Several HvTGAs, such as HvTGA-25, HvTGA-30, and HvTGA-37, were associated with multiple conserved miRNAs, indicative of combinatorial regulation characteristic of regulatory hubs, which may enhance both robustness and flexibility under environmental stress [98]. By contrast, lineage-specific miRNAs, including members of the hvu-miR444 family, were predicted to target HvTGA-12 and HvTGA-24, reflecting grass-specific regulatory circuits. Additional narrowly acting miRNAs, such as hvu-miR159, hvu-miR171-5p, and hvu-miR1436, targeted individual HvTGAs, highlighting substantial variation in post-transcriptional specificity across the family [99]. Together, these predicted interactions indicate that HvTGAs may function as important nodes within barley regulatory networks. Nevertheless, miRNA–target relationships and their regulatory modes remain computationally inferred and require experimental confirmation using degradome sequencing, 5′ RLM-RACE, transient reporter assays, or related functional approaches before definitive regulatory assignments can be established. Cis-regulatory motifs shape the capacity of transcription factors to integrate environmental, hormonal, and developmental signals [100, 101]. Analysis of HvTGA promoters revealed widespread enrichment of ABA-responsive elements (ABREs), particularly in HvTGA-21 and HvTGA28, underscoring the potential role of ABA signalling in drought and osmotic stress responses. Jasmonate-responsive motifs (TGACG, CGTCA) were also prevalent, especially in HvTGA-18 and HvTGA-24, indicating possible ABA–JA crosstalk, a mechanism well documented in Arabidopsis and increasingly recognised in monocots for modulating stress resilience [102–105]. Stress-responsive elements constituted the largest category overall. Nearly all HvTGA promoters contained multiple drought-responsive (MBS), cold-responsive (LTR), and defence-associated (W-box) motifs, with HvTGA-8, HvTGA-18, HvTGA-24, and HvTGA-35 displaying particularly dense clusters. The frequent W-box occurrence suggests potential co-regulation with WRKY transcription factors, which are central to stress adaptation in cereals [106, 107]. Light-responsive motifs were also broadly distributed, including Box 4, G-box, GT1-motif, and GATA-motifs, indicating additional layers of environmental regulation, with HvTGA-2, HvTGA-27, and HvTGA-46 showing comparatively higher motif densities. Development-related motifs, such as CAT-box, GCN4, AAGAA, and O2-site, were detected in multiple promoters, notably HvTGA-2 and HvTGA-9, suggesting that subsets of HvTGAs may integrate developmental and stress-related regulatory signals [108]. These dual regulatory potential parallels the multifunctional roles of Arabidopsis TGAs in coordinating defence and development [7]. However, promoter architecture alone does not establish transcriptional activity, and these cis-elements should therefore be interpreted as indicators of regulatory potential pending functional promoter validation. The HvTGA gene family in barley exhibited distinct tissue-specific expression patterns under heat, salt, and osmotic stress conditions, indicating substantial functional divergence among family members during abiotic stress adaptation. Under heat stress, HvTGA-29, HvTGA-43, HvTGA-41, and HvTGA-38 showed pronounced upregulation in both shoot and root tissues, indicating their likely involvement in heat-responsive transcriptional regulation. This pattern is consistent with the findings of Kiełbowicz-Matuk et al. [109], who reported that heat stress promotes chromatin accessibility and extensive transcriptional reprogramming in barley. In contrast, HvTGA-28 and HvTGA-31 displayed reduced expression, implying potential roles in transcriptional fine-tuning or selective repression during heat adaptation. Under salt stress, HvTGA-29, HvTGA-12, HvTGA-14, HvTGA-21, and HvTGA-33 were strongly induced in root tissues, particularly at later treatment stages, indicating potential functions in ionic homeostasis and osmotic adjustment. This expression pattern agrees with the observations of Ueda et al. [110], who demonstrated that barley exhibits distinct transcriptional responses to salt and osmotic stress, reflecting the complexity of stress-specific signalling pathways. Conversely, HvTGA-24 and HvTGA-48 showed progressive downregulation, which may reflect selective repression of specific regulatory pathways during salinity adaptation. Notably, HvTGA-29 and HvTGA-12 were functionally annotated within GO categories including “response to salt stress” (GO:0009651), “response to abscisic acid” (GO:0009737), and “abscisic acid-activated signalling pathway” (GO:0009738), supporting their likely involvement in ABA-mediated osmotic regulation and stress-responsive transcriptional control. Under osmotic stress, HvTGA-29, HvTGA-43, HvTGA-41, and HvTGA-38 were again prominently induced in shoot tissues, indicating stable responsiveness across multiple abiotic stress conditions. Similar systemic transcriptional responses were reported by Habte et al. [111], who showed that osmotic stress influences shoot gene expression through root-to-shoot signalling mechanisms. These genes were also represented within KEGG pathways such as “plant hormone signal transduction” (ko04075) and “MAPK signalling pathway – plant” (ko04016), supporting their potential roles in linking upstream stress perception with downstream transcriptional reprogramming. The repeated induction of HvTGA-29 across heat, salt, and osmotic stress, together with its association with ABA- and MAPK-related pathways, indicates that this gene may function as an important regulatory component of abiotic stress adaptation in barley. Likewise, HvTGA-43, HvTGA-41, and HvTGA-38 may participate in conserved hormone-responsive regulatory modules involved in stress tolerance. Comparative analysis with phylogenetically related rice TGA homologues provides additional insight into the evolutionary conservation of these stress-responsive patterns. HvTGA-29, which clusters with OsTGA-1, exhibited consistent induction across multiple stress conditions, suggesting functional conservation of core regulatory roles in monocot stress adaptation. Similarly, HvTGA-12, grouped with OsTGA-5, displayed strong responsiveness under salinity stress, supporting the involvement of this clade in ABA-associated osmotic regulation [112]. HvTGA-43 and HvTGA-41, clustering with OsTGA-20 and OsTGA-34 respectively, also showed stable induction across stress treatments, indicating that these orthologous groups may represent conserved transcriptional modules linking hormone signalling and stress-responsive gene expression in cereals [13, 113]. Previous studies in rice have demonstrated that bZIP/TGA-related transcription factors participate in ABA signalling, redox regulation, and stress-responsive transcriptional networks, supporting the functional conservation observed here [20, 114]. Despite this overall conservation, differences in expression intensity and tissue specificity among HvTGA members suggest lineage-specific regulatory divergence following gene duplication. For instance, the pronounced and recurrent induction of HvTGA-29 across all tested stress conditions contrasts with more condition-dependent responses of other paralogues, indicating potential subfunctionalisation within the barley lineage. Such divergence is consistent with the expansion patterns observed in cereal genomes, where duplicated transcription factors often acquire specialised regulatory roles while retaining core functional attributes [115, 116]. The consistent induction of these genes across multiple stress conditions indicates that specific HvTGA members function as important components of abiotic stress-responsive regulatory networks in barley. In contrast, genes showing more variable or tissue-specific expression patterns may contribute to specialised regulatory processes associated with particular environmental conditions or developmental contexts. These findings highlight the functional diversity of the HvTGA gene family and identify HvTGA-29, HvTGA-43, HvTGA-41, and HvTGA-38 as strong candidates for further investigation of TGA-mediated stress response pathways and the development of stress-resilient barley genotypes. The robustness of these expression patterns, observed consistently across independent RNA-seq datasets and multiple stress conditions, further supports their potential regulatory relevance. However, as the present study is based on publicly available transcriptomic datasets and computational analyses, these findings should be interpreted as hypothesis-generating rather than confirmatory. Experimental validation, including qRT-PCR-based expression analysis, and functional characterisation under controlled stress conditions, will be required to confirm the transcriptional responses and biological roles of these candidate genes. Taken together, the integration of phylogenetic classification, gene duplication patterns, structural features, regulatory element composition, and stress-responsive expression profiles supports a unified and hierarchical model of HvTGA function in barley. In this framework, segmental and dispersed duplication events have expanded the HvTGA repertoire, generating paralogous groups that are subsequently partitioned into distinct phylogenetic clades with conserved structural cores. These clades retain fundamental bZIP-mediated DNA-binding and dimerisation capacity, while divergence in cis-regulatory elements, miRNA targeting profiles, and protein–protein interaction networks introduce regulatory specificity at multiple levels. Such multilayered regulation enables HvTGA transcription factors to function as integrative nodes linking upstream environmental cues, including heat, salinity, and osmotic stress, to downstream transcriptional outputs. In particular, enrichment of ABRE, MBS, and JA-responsive elements, together with predicted interactions with GRX, NPR, and kinase-associated signalling components, suggests that HvTGAs operate at the intersection of ABA-dependent, MAPK-mediated, and redox-regulated pathways. Several candidate regulatory genes, notably HvTGA-29, HvTGA-43, and HvTGA-41, were consistently identified based on their strong and reproducible induction across multiple abiotic stress conditions and tissues, further supported by functional enrichment and network-based analyses. These genes appear to integrate diverse signalling inputs and coordinate transcriptional reprogramming across tissues, thereby contributing to stress adaptation. Concurrently, lineage-specific divergence in expression amplitude and tissue specificity among paralogues indicates subfunctionalisation following duplication, enabling fine-tuned and context-dependent regulatory responses. Based on these observations, a conceptual regulatory framework is proposed (Fig. 10), in which HvTGAs act as central transcriptional regulators that integrate hormonal signalling, redox homeostasis, and environmental stress perception into coordinated gene expression programmes. This systems-level model provides a foundation for future experimental validation and offers a mechanistic perspective on TGA-mediated regulatory networks underlying abiotic stress tolerance in barley. Fig. 10Systems-level model of HvTGA-mediated regulatory networks under abiotic stress in barley. Abiotic stress signals (heat, salinity, and osmotic/drought stress; Panel A) are perceived through ABA-dependent signalling, MAPK cascades, and redox-mediated pathways (Panel B), which converge on HvTGA transcription factors acting as central regulatory hubs (Panel C). Key candidates (HvTGA-29, HvTGA-43, and HvTGA-41) are highlighted based on consistent stress-responsive expression. HvTGAs integrate cis-regulatory inputs, miRNA-mediated regulation, and protein–protein interactions (e.g., NPR and GRX proteins) to control downstream gene expression. This regulation modulates stress-responsive genes involved in ROS detoxification, osmotic adjustment, and heat protection, leading to physiological adaptation, including enhanced stress tolerance and homeostasis (Panel D). Arrows indicate activation or signal flow, dashed lines represent regulatory modulation or repression (including miRNA-mediated inhibition), and highlighted nodes denote key hub regulators. This hierarchical model illustrates the integration of signal perception, transduction, transcriptional regulation, and physiological adaptation mediated by HvTGA transcription factors Systems-level model of HvTGA-mediated regulatory networks under abiotic stress in barley. Abiotic stress signals (heat, salinity, and osmotic/drought stress; Panel A) are perceived through ABA-dependent signalling, MAPK cascades, and redox-mediated pathways (Panel B), which converge on HvTGA transcription factors acting as central regulatory hubs (Panel C). Key candidates (HvTGA-29, HvTGA-43, and HvTGA-41) are highlighted based on consistent stress-responsive expression. HvTGAs integrate cis-regulatory inputs, miRNA-mediated regulation, and protein–protein interactions (e.g., NPR and GRX proteins) to control downstream gene expression. This regulation modulates stress-responsive genes involved in ROS detoxification, osmotic adjustment, and heat protection, leading to physiological adaptation, including enhanced stress tolerance and homeostasis (Panel D). Arrows indicate activation or signal flow, dashed lines represent regulatory modulation or repression (including miRNA-mediated inhibition), and highlighted nodes denote key hub regulators. This hierarchical model illustrates the integration of signal perception, transduction, transcriptional regulation, and physiological adaptation mediated by HvTGA transcription factors Conclusion TGA transcription factors are pivotal regulators of hormone-mediated signalling, redox balance, and stress adaptation, yet their genomic organisation and functional diversification in barley have remained insufficiently characterised. This study presents the first genome-wide identification and comprehensive characterisation of the HvTGA gene family in barley, revealing 49 members distributed across the seven chromosomes. Family expansion was driven predominantly by segmental and dispersed duplications, while strong purifying selection contributed to substantial evolutionary conservation across the family. The universal presence of conserved bZIP domains, together with clade-specific exon–intron architectures and motif divergence, indicates a balance between structural conservation and functional diversification among HvTGA genes. Promoter profiling and transcriptomic expression analyses suggest that HvTGAs are closely associated with abscisic acid-mediated signalling, redox regulation, and multiple abiotic stress responses. Predicted miRNA interactions and protein–protein interaction networks further identified HvTGA-13, HvTGA-26, and HvTGA-28 as prominent candidate regulatory hubs within stress-responsive pathways. Likewise, the consistent induction of HvTGA-29, HvTGA-43, HvTGA-41, and HvTGA-38 under heat, salt, and osmotic stress highlights these genes as strong candidates for further investigation of TGA-mediated stress adaptation in barley. Subcellular localisation, tertiary protein structure, miRNA–target interactions, and protein–protein interaction networks collectively provide a preliminary regulatory framework for understanding HvTGA function; however, these analyses were derived primarily from computational prediction and orthology-based inference and should therefore be interpreted as putative rather than experimentally validated evidence. These findings establish a comprehensive foundation for understanding the evolutionary and regulatory landscape of TGA transcription factors in barley and provide a prioritised set of candidate genes for future functional studies. Further validation through molecular, genetic, and physiological approaches including expression verification, interaction assays, and functional characterisation under controlled stress conditions will be essential to define the precise biological roles and regulatory mechanisms of these HvTGA genes. Integrating such functional evidence with breeding and biotechnological strategies may contribute to the development of barley cultivars with improved resilience to environmental stress. Supplementary Information Supplementary Material 1. Supplementary Material 1.