Comprehensive co-expression network reveals the fine-tuning of AsHSFA2c in balancing drought tolerance and growth in oat

Introduction

Drought, which is one of the most pressing challenges in agriculture today, often alters plant physiology, biochemistry, and molecular regulation, leading to the termination of photosynthesis and metabolic disorders, and ultimately resulting in a decrease in grain yield^1,2,3,4,5. Thus, there is a critical need to identify key genes associated with the drought stress response in highly drought-resistant species or accessions and characterize their molecular regulatory mechanisms. Common oat (Avena sativa L., 2n = 6x = 42, AACCDD), which is ranked seventh among cereals in terms of production (http://www.fao.org/faostat/en/; accessed May 2021), is an economically important food and feed crop cultivated worldwide and is highly adaptable to various climatic conditions, especially drought^6,7,8. Therefore, identifying key genes that contribute to drought tolerance in oats has become increasingly important. However, due to the lack of a mature and stable genetic transformation system and omics data for drought stress, relatively few drought resistance-related genes have been identified in oats.

Under drought conditions, plant growth and development are seriously impaired. To cope with increasingly serious drought stress, there has been a steady increase in research on the balance between plant drought tolerance and growth⁹. There are two main viewpoints regarding the balance between plant growth and stress responses, including the allocation of limited energy and the enhanced expression of genes that promote growth to inhibit drought stress responses⁹. Research on the balance between plant drought tolerance and growth is of great significance to agricultural development. To date, many genes related to balancing drought tolerance and growth have been identified in rice, maize, wheat, and other important crops^10,11,12. These studies showed that hormonal signals play a crucial role in regulating the balance between drought tolerance and growth in plants. Specifically, the interactions between abscisic acid (ABA) and indole-3-acetic acid (IAA), as well as between ABA and gibberellin (GA) are essential for maintaining the balance^12,13. However, genes responsible for balancing drought tolerance and growth and their regulatory mechanisms remain largely unknown in oats.

In response to biotic and abiotic stresses, transcription factors play an important role in the regulation of plant growth and development. The heat shock factor (HSF) gene family is a kind of transcription factor, and is widely present in prokaryotes and eukaryotes¹⁴. In plants, the HSF family plays a central role in regulating various processes, including growth and development and responses to abiotic stresses, including heat, drought, salt, and cold stress^15,16,17. Previous studies have reported that OsHSFa7, OsHSFb2b, ZmHSF08, GmHSF-34, and TaHSFa2e-5d played key roles in regulating drought tolerance in crops^{18,19,20,21,22}. However, little has been reported about whether HSF family genes are involved in regulating the balance between plant drought tolerance and growth. Furthermore, the biological functions and molecular regulatory mechanisms of the HSF family in oats also remain elusive.

In this study, we constructed a comprehensive co-expression network of genes that balance drought tolerance and growth in oats. This network included 84 modules and numerous candidate genes associated with drought tolerance and growth, offering valuable resources for future analyses of drought tolerance in oats and the associated regulatory mechanisms. To verify the network’s credibility, we utilized the virus-induced gene silencing (VIGS) system in oat for functional validation of several candidate modules and genes. By integrating over-expression transgenic oat lines, we characterized the function of AsHSFA2c in regulating the balance between drought tolerance and growth in oat. Additionally, we determined that AsHSFA2c was regulated by the upstream transcription factor AsDOF25 and transcriptionally repressed the downstream gene AsAGO1. Overall, this study provides insights into the genetic basis of drought tolerance in oats by elucidating how AsHSFA2c mediates the balance between drought tolerance and growth. This resource will also facilitate further investigations on the genetic basis of strong drought tolerance in oats.

Result

Co-expression network construction for oat drought response and growth

To comprehensively characterize transcriptional dynamics related to oat plant growth and drought stress response, we generated a gene expression atlas comprising 84 samples collected from cultivated oats exposed to drought stress or treated with polyethylene glycol (PEG)-6000 and hormones (ABA and IAA) (Fig. 1a, Supplemental Fig. 1a–d, and Supplemental Data 1). The correlation analysis of the 84 transcriptome samples revealed a high degree of consistency among biological replicates, all of which clustered together (Supplemental Fig. 2a). To further validate the quality of the gene expression profiles, the effects of each treatment were assessed by examining the response of marker genes (Supplemental Fig. 2b and Supplemental Data 2). Each marker gene was either up- or down-regulated after the corresponding treatment, which is consistent with previously reported patterns. For example, AREB1 and PP2C were significantly induced by ABA^23,24, indicating that the treatment effectively altered the expression of these response marker genes. Gene expression analysis showed that a total of 62,242 differentially expressed genes (DEGs) were identified across all treatments (Fig. 1b, Supplemental Fig. 2c, and Supplemental Data 3). Furthermore, 13,301 DEGs were found to be shared among the drought, PEG6000, ABA, and IAA treatments, suggesting their potential role in both drought tolerance and growth in oat (Fig. 1b). As previously reported^{25,26,27,28,29}, our transcriptome data showed that a substantial number of genes exhibited opposite expression trends after IAA and drought treatments (Supplemental Fig. 2d). Meanwhile, we performed a quantitative analysis, using Fisher’s exact test^30,31, to evaluate the overlap of differentially expressed genes (DEGs) between hormone treatments (ABA and IAA) and drought-related treatment (drought and PEG6000 treatments) (Supplemental Fig. 2e). The results showed that the overlapping DEGs between ABA treatment and drought-related treatment were significantly higher compared to those between IAA treatment and drought-related treatment (Supplemental Fig. 2e). Then, we conducted transcriptome analysis on genes annotated as being involved in the ABA and IAA signaling pathways. As reported previously^32,33, most of the genes explored their antagonistic patterns at the transcriptomic level (Supplemental Fig. 3a–d and Supplemental Data 4 and 5). This result further supports the reliability of our transcriptome data. Subsequently, we performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses to functionally characterize DEGs and determine differences among treatment groups. Many of the DEGs were functionally associated with plant growth and defense responses (Supplemental Fig. 4a–b and Supplemental Data 6 and 7). For example, the DEGs responsive to the ABA treatment were mainly involved in the glutathione metabolic process, carbohydrate binding, response to water, transferase activity, and defense response. In contrast, DEGs following the IAA treatment were mainly associated with response to auxin, plant hormone signal transduction, and photosynthesis antenna proteins. For the PEG6000 treatment, DEGs were primarily involved in processes related to photosynthesis, glutathione metabolic process, response to water, phosphatidylinositol phospholipase C activity, and carbohydrate binding. In drought-treated samples, DEGs were significantly enriched in processes like glutathione metabolic process, photosynthesis, cell wall macromolecule catabolic process, and embryo development ending in seed dormancy (Supplemental Fig. 4a, b and Supplemental Data 6 and 7).

**Fig. 1: Construction of comprehensive co-expression networks that balance drought tolerance and growth in oat.**

We further integrated these 84 transcriptome datasets to perform the weighted gene co-expression network analysis (WGCNA) (Fig. 1c), and combining the evidence from the correlation analysis, module classification, and functional enrichment analysis, a total of 84 functional modules were identified (Fig. 1c, d and Supplemental Data 8). Among them, the largest and smallest modules contained 6226 and 19 genes, respectively (Fig. 2a and Supplemental Data 9). To evaluate the interaction of all modules, eigengene adjacency was calculated. The heatmap showed that the modules were distinguished well from one another (Supplemental Fig. 4c). Gene functional enrichment analysis showed that many modules covered functions such as photosynthesis, plant hormone signal transduction, arginine and proline metabolism, as well as cell wall biogenesis (Fig. 1d, Supplemental Fig. 4d, and Supplemental Data 8 and 10), indicating these modules may play an important role in regulating plant growth and responding to drought stress. For example, in our network, we detected the co-expression of PYL (PYRABACTIN RESISTANCE 1-LIKE, oat091221), SnRK2 (phosphatsenon-fermenting SNF1 related protein Kinase2s, oat067672) and PP2C (phosphatase type 2Cs, oat041125), which were the previously reported important components of a drought response-related ABA signaling pathway (Fig. 1e)^34,35, further supporting the reliability of our network. Besides, we also identified three additional genes (oat058263, oat049732, and oat079894) that were co-expressed with PYL, SnRK2, and PP2C (Fig. 1e). Our co-expression network may serve as a valuable resource for the oat research community, enabling the identification of functional genes or modules associated with growth and drought stress response. Additionally, it may contribute to studies on the regulation of the balance between plant drought tolerance and growth.

**Fig. 2: Identification of functional modules related to drought tolerance.**

Candidate modules related to drought tolerance in oat

The correlation coefficients between the 84 gene modules and various treatments were evaluated, revealing that many functional modules were strongly associated with the corresponding treatments (Fig. 2a). For instance, 52 modules were closely related to drought treatment or PEG6000 treatment, indicating their potentially essential roles in regulating drought tolerance in oat (Supplemental Data 11). Among them, we randomly selected one module (Cluster 1) to show (Fig. 2b, c and Supplemental Data 12). In this module, transcription factors with connectivity greater than or equal to 5 were identified as hub transcription factors (Fig. 2c). This transcription factor network identified hub transcription factors potentially involved in oat drought stress and revealed their associated target genes. Functional enrichment analysis showed that the genes in Cluster 1 were primarily involved in the MAPK signaling pathway, flavone and flavonol biosynthesis, and calcium signaling pathways (Fig. 2b and Supplemental Data 13), which have previously been reported to be involved in drought signaling transduction^27,36,37. Additionally, many homologous genes, such as OsWRKY10, OsWRKY72, ANAC102, ANAC032, OsbZip25, and OsDREB1C, which have been previously reported to be involved in the regulation of drought tolerance, were identified in Cluster 1 (Fig. 2c and Supplemental Data 14 and 15)^{38,39,40,41,42}. To further validate the reliability of this drought tolerance-related module, we randomly selected five genes with homologs known to be associated with drought tolerance in other species^{43,44,45,46,47}, as well as five genes with unknown function, for qRT-PCR analysis (Fig. 2d). The results showed that 9/10 genes respond to drought treatment, implying that the module was likely related to drought tolerance. To further validate the identified module, we selected the hub transcription factor AsMYC2 (oat068987), a homolog of TaMYC2 in wheat, as a candidate gene for functional validation. The transcription factor MYC2 has been reported to regulate drought tolerance in several plants, including Triticum aestivum, Arabidopsis thaliana, Brassica napus, and Solanum lycopersicum^48,49,50,51. qRT-PCR assay showed that the expression of AsMYC2 was up-regulated under drought conditions and down-regulated after rewatering (Fig. 2d). Furthermore, to investigate the effects of silencing AsMYC2 on oat phenotypes, we used the tobacco rattle virus (TRV) VIGS system to suppress the expression of AsMYC2 (Supplemental Fig. 5a, b). Following the successful infection by the TRV virus using the VIGS system, we obtained the following three oat lines in which AsMYC2 was silenced: TRV:AsMYC2-1 (V-AsMYC2-1), TRV:AsMYC2-2 (V-AsMYC2-2) and TRV:AsMYC2-3 (V-AsMYC2-3) (Supplemental Fig. 5c, d). Six-week-old oat seedlings of AsMYC2 knock-down lines (V-AsMYC2) were assessed for drought tolerance phenotype and statistical survival rate (Fig. 2e), and the V-AsMYC2 seedlings showed decreased drought-tolerance and lower survival rate after drought treatment. These results showed that malfunctioning AsMYC2 could lead to a decrease in plant drought tolerance, and we obtained a relatively reliable drought tolerance sub-network.

AsHSFA2c contributes to balancing drought tolerance and growth in oat

The activation of plant stress tolerance-related responses is often accompanied by energy consumption and changes in metabolic substances, with detrimental effects on growth and development. Especially, the continuous activation of stress tolerance-related responses may lead to limited plant growth and a decrease in crop yield. Therefore, there is a trade-off between plant stress tolerance and growth⁵². To mine the candidate genes involved in balancing drought tolerance and growth in oats, we identified the ‘balance’ modules based on the analyses of DEGs, Pearson correlation coefficient, functional enrichment, and annotated gene functions. A total of 23 modules were defined as ‘balance’ modules, indicating their potential roles in regulating the balance between drought tolerance and growth in oats (Supplemental Data 11). We selected Cluster 12 as a representative module to illustrate our findings (Fig. 3a) due to its high proportion of genes showing differential expression under drought-related stress (86%) and IAA treatment conditions (69%). Additionally, this cluster was highly correlated with both drought-related stress and IAA treatment (Supplemental Data 11). Functional enrichment analysis revealed that the genes in Cluster 12 were significantly associated with drought tolerance, as well as plant growth and development (Supplemental Fig. 6 and Supplemental Data 16). Notably, Cluster 12 included key genes that are known in other species for their roles in drought tolerance and growth regulation, including OsPIL1 (oat057108)^53,54 and OsDOG1L-2 (oat008214)⁵⁵. Furthermore, we randomly selected 10 candidate genes from this cluster for a qRT-PCR analysis (Fig. 3b). 8 genes were responsive to both drought and IAA treatments. Among the genes in this balance module, AsHSFA2c (oat069028), a member of the HSF family, was identified as an ortholog of OsHSFA2a (Supplemental Fig. 7a, b) homologs of this gene had been previously reported to participate in the regulation of plant drought tolerance and growth^56,57,58. Transcriptome and qRT-PCR data showed that the expression of AsHSFA2c was up-regulated by drought, as well as ABA and IAA treatments (Fig. 3c and Supplemental Fig. 8a–f). The ABA signaling pathway is the most important hormone signaling pathway in response to drought stress, whereas auxin is one of the most important hormones affecting plant growth and development. The qRT-PCR results showed that the expression of AsHSFA2c was significantly up-regulated after ABA and IAA treatment (Supplemental Fig. 8e, f), suggesting that the response of AsHSFA2c to drought stress may depend on the ABA signaling pathway and AsHSFA2c may affect plant growth and development via the IAA signaling pathway. Furthermore, the results of tissue-specific expression analysis showed that AsHSFA2c was relatively uniformly expressed across various oat tissues, with notably higher levels in flag leaves and floral organs (Supplemental Fig. 8g). The subcellular localization and nuclear-cytoplasmic separation results showed that AsHSFA2c was localized in the nucleus and cytoplasm (Supplemental Fig. 8h, i). These results suggest that AsHSFA2c may be a key molecular switch that balances drought tolerance and growth in oats.

Fig. 3: Identification and verification of the balance modules and the key candidate gene, *AsHSFA2c.*

To further investigate the function of AsHSFA2c in oats, we used the TRV VIGS system to suppress its expression (Fig. 3d and Supplemental Fig. 9a, b). One to three-week-old oat seedlings of AsHSFA2c (V-AsHSFA2c) knock-down lines were observed for growth phenotype and six-week-old V-AsHSFA2c seedlings were observed for drought-tolerance phenotype (Fig. 3e). Compared with the control, the knock-down lines of AsHSFA2c have been shown to be more conducive to plant growth and development, with increased chlorophyll content in leaves, heavier fresh weights, and higher plant height (Fig. 3e and Supplemental Fig. 10a–c). At the cellular level, the total number of pavement cells decreased in V-AsHSFA2c plants (Supplemental Fig. 10d), indicating that it may influence leaf development in oats by modulating cell division and expansion. In addition, AsHSFA2c knock-down plants exhibited increased sensitivity to drought (Fig. 3e and Supplemental Figs. 10a and 11a, b). The chlorophyll content, survival rate, and water loss rate were used as indicators of drought tolerance. Compared with the control, under drought conditions, V-AsHSFA2c had lower chlorophyll content, lower survival rate, and faster water loss rate (Supplemental Figs. 10a and 11a, b). Additionally, decreased AsHSFA2c expression increased the number of stomata in the lower epidermis of the leaf blade (Supplemental Fig. 10d). AsHSFA2c overexpression in Arabidopsis thaliana resulted in dwarfism and increased drought tolerance (Supplemental Fig. 12a–c), implying that in response to drought, AsHSFA2c expression may have regulatory effects on stomatal density. To confirm the function of AsHSFA2c, we used an oat genetic transformation system to generate transgenic AsHSFA2c-overexpressing lines (Fig. 3f–h), after which 2-week-old AsHSFA2c-overexpressing oat seedlings were examined in terms of growth and 4-week-old AsHSFA2c-overexpressing oat seedlings were observed for drought-tolerance phenotype (Fig. 3h). Compared with the control, AsHSFA2c-overexpressing lines have shown to be not conducive to plant growth and development while higher drought tolerance, with lower plant height (Supplemental Fig. 13a) and higher survival rate under drought treatment (Supplemental Fig. 13b). These results suggest that AsHSFA2c may inhibit plant growth and positively regulate drought tolerance. Collectively, we identified a module and a key candidate gene, AsHSFA2c, that balances drought tolerance and growth in oats.

AsDOF25 positively regulates AsHSFA2c as a transcription activator

To explore the upstream regulators of AsHSFA2c and understand how AsHSFA2c responds to drought stress, the nuclear run-on assay was performed. The result showed that the transcription rate of AsHSFA2c increased under drought conditions (Supplemental Fig. 14a). Notably, we found that a DOF transcription factor core recognition site (AAAG or CTTT) in the promoter of AsHSFA2c at 12-bp upstream of its translation start site (Supplemental Fig. 14b)⁵⁹. Previous studies have shown that DOF family members participated in the regulation of plant growth and drought tolerance^60,61,62. The transcriptomic analysis confirmed that most DOF family members were drought-responsive (Supplemental Fig. 14c). Therefore, we hypothesized that AsHSFA2c might be regulated by a DOF transcription factor. Interestingly, our co-expression network analysis revealed that 43 DOF family members were closely associated and co-expressed with AsHSFA2c (Fig. 4a). Among them, AsDOF25, a gene with close connectivity to AsHSFA2c, was induced by drought, ABA, and IAA treatments (Supplemental Fig. 14d–h). According to the AsDOF25 network diagram, AsDOF25 was correlated with HSF family genes (Supplemental Fig. 14i). Moreover, the AsDOF25 promoter contains an ABRE binding site, which is associated with ABA signaling, and an AUXRR-core site, related to auxin response (Supplemental Fig. 14j). Thus, we hypothesized that AsDOF25 protein might bind to the AsHSFA2c promoter and regulate its expression. To test this hypothesis, we performed an electrophoretic mobility shift assay (EMSA) using GST-tagged AsDOF25 protein, with GST protein as a negative control (Fig. 4b and Supplemental Fig. 14k, l). The result showed that a single shifted complex was formed between the AsDOF25 protein and AsHSFA2c promoter, and the complex intensity decreased with the addition of the unlabeled AsHSFA2c promoter, but not with the mutated promoter fragment (Fig. 4b). These results indicated that AsDOF25 protein could directly bind to the promoter of AsHSFA2c. Subsequent experiments demonstrated that co-expressing pAsHSFA2c::LUC with p35s::AsDOF25 in Nicotiana benthamiana leaves led to a significant increase in LUC transcript level in the co-expressed tissues compared to controls (Fig. 4c). In addition, using the TRV VIGS system, we effectively suppressed AsDOF25 expression (Supplemental Fig. 15a, b), achieving over 50% silencing efficiency as confirmed by qRT-PCR (Supplemental Fig. 15c). The AsHSFA2c transcript level in AsDOF25 mutants was significantly lower than that in control (Fig. 4d). Altogether, these results indicate that AsDOF25 protein could promote the transcription of AsHSFA2c as a transcription activator.

**Fig. 4: AsDOF25 positively regulates *AsHSFA2c* via transcription activation.**

Furthermore, we also examined the roles of AsDOF25 in plant growth and drought tolerance. Growth phenotypes of two-week-old AsDOF25 knockdown oat seedlings (V-AsDOF25) were observed, and six-week-old seedlings were evaluated for drought tolerance (Fig. 4e–h). Results showed that knock-down lines for both AsDOF25 and AsHSFA2c exhibited similar trends in all observed phenotypes. For example, two-week-old V-AsDOF25 plants were taller than controls (Fig. 4e, f). Moreover, plants with AsDOF25 malfunction showed heightened drought sensitivity (Fig. 4g), characterized by a lower survival rate and faster leaf water loss rate (Fig. 4h, i). These results indicated that AsDOF25 inhibited plant growth and positively regulated plant drought tolerance. Therefore, we inferred that the regulation of the balance between plant growth and drought tolerance by AsHSFA2c partially relied on the regulation of AsDOF25.

AsHSFA2c negatively regulates AsAGO1 in balancing drought tolerance and growth in oat

A DNA affinity purification sequencing (DAP-seq) experiment was conducted to investigate the underlying mechanism by which AsHSFA2c balances drought tolerance and growth in oats. Following a credibility assessment of the DAP-seq peaks, we identified 1219 highly reliable peaks (Fig. 5a and Supplemental Fig. 16a). Among these peaks, 145 peaks were located within the 3 kb regions of genes, and 88 candidate genes were identified (Supplemental Data 17). To further investigate AsHSFA2c recognition motifs, the ±100 bp sequences flanking peaks were submitted to MEME-ChIP to search for enriched motifs (Fig. 5b). Intriguingly, among these candidate genes, the gene oat023722 (AsAGO1) had a significant binding signal for AsHSFA2c in its promoter region (Fig. 5c). The gene, AGO1, is a core component of the RNA-induced silencing complex (RISC) and plays a central role in the RNA silencing pathway. AGO1 has been previously reported to regulate plant growth, development, and drought tolerance through sRNA in various species, such as Arabidopsis thaliana and Oryza sativa^63,64,65,66. The electrophoretic mobility shift assay (EMSA) with competitive probes and mutation probes further supported that AsHSFA2c protein could bind to AsAGO1 promoter (Fig. 5d and Supplemental Fig. 16b). Moreover, when pAsAGO1::LUC was co-expressed with p35s::AsHSFA2c in Nicotiana benthamiana leaves, the LUC transcript level was significantly lower than that in control tissues (Fig. 5e), implying that AsHSFA2c negatively regulated AsAGO1. The qRT-PCR results showed that the expression level of AsAGO1 was up-regulated in AsHSFA2c knock-down mutants (Fig. 5f). These results indicated that AsAGO1 was a downstream target gene of AsHSFA2c and was negatively regulated by AsHSFA2c.

**Fig. 5: AsHSFA2c negatively regulates *AsAGO1* through transcriptional inhibition.**

To elucidate how AsAGO1 balances growth and drought tolerance in oats, we collected plant samples exposed to drought stress and treated them with IAA for qRT-PCR analyses. The results indicated that the expression of AsAGO1 was down-regulated by drought stress and IAA treatment in root, but up-regulated by the IAA treatment in plant aerial part (Fig. 5g, h). We subsequently used a TRV VIGS system to suppress the expression of AsAGO1 (Fig. 5i and Supplemental Fig. 16c–e). One to three-week-old knock-down lines of AsAGO1 (V-AsAGO1) were observed for growth phenotype, and six-week-old V-AsAGO1 were observed for the drought-tolerance phenotype (Fig. 5j). Plant height, chlorophyll content, survival rate, and water loss rate were measured as phenotypic indicators (Fig. 5k-l). Under normal growth conditions, V-AsAGO1 plants were smaller and had less chlorophyll than control plants (Fig. 5i-j). However, under drought conditions, V-AsAGO1 lines exhibited enhanced drought tolerance, characterized by increased chlorophyll content, higher survival rate, and slower leaf water loss rate (Fig. 5j–l). These results indicated that AsAGO1 was a downstream target gene of AsHSFA2c, and AsAGO1 could promote plant growth and negatively regulate drought tolerance in oats.

Discussion

Drought stress is intensifying globally, posing a serious threat to food security⁶⁷. More efforts are needed to improve the drought resistance of our major crops by identifying excellent genes related to drought tolerance. Common oat is an economically important crop worldwide, and is highly adaptable to various climatic conditions, especially drought stress^6,7. Therefore, exploring the molecular regulatory mechanisms of the strong drought resistance in oats has become increasingly important. In this study, we constructed a genome-wide co-expression network to investigate the balance between plant growth and drought tolerance in oats, utilizing 84 transcriptome datasets from the PEG6000, drought stress, ABA, and IAA treatments (Supplemental Fig. 17a). Through in-depth data mining, we integrated networks covering various biological aspects with functional analysis tools (Supplemental Fig. 17b), and identified 23 balance modules which were significantly associated with drought tolerance, as well as plant growth and development. Many genes in these modules are involved in plant hormone signal transduction, arginine and proline metabolism, and MAPK signaling pathway (Fig. 1d), which were previously reported to be associated with plant growth and drought stress responses^{68,69,70,71,72}. Our co-expression network and the tightly linked genes within one module may serve as a valuable resource for identifying functional genes or modules associated with growth and drought stress response, which can be used for genetic improvement and in-depth study of molecular regulatory mechanisms in oat.

HSF transcription factor family was reported to be involved in the regulation of plant growth and development and drought stress responses^{16,73,74,75,76,77}. However, there are few reports about the regulation of HSF family members to balance between drought tolerance and plant growth. To date, the ‘balancing’ function of HSFA6B has only been reported in dryland cotton, though its molecular mechanisms and functional aspects remain insufficiently understood⁷⁸. In this study, we uncovered the molecular mechanism of AsHSFA2c in balancing drought tolerance and growth in oats. Under normal conditions, as a component of the plant stress response, AsHSFA2c was maintained at a low expression level, which was conducive to plant growth and development. Upon drought stress, the expression of AsHSFA2c was upregulated, leading to an increase in plant drought tolerance. Our findings revealed that ABA and auxin reciprocally regulated the AsDOF25-AsHSFA2c-AsAGO1 module to fine-tune the balance between drought tolerance and growth in oat, highlighting potential targets for breeding drought-tolerant oat lines. The upstream gene AsDOF25 acts as a transcriptional activator, enhancing AsHSFA2c expression, whereas AsHSFA2c, acting as a transcriptional repressor, suppresses the downstream target gene AsAGO1 (Supplemental Fig. 17c). We found that 2273 DEGs were co-expressed with the AsDOF25-AsHSFA2c-AsAGO1 module in response to drought and PEG6000 treatments. Additionally, we found that under IAA treatments, the expression of AsAGO1 showed opposite expression trends in different tissues (Fig. 5h), which is similar to previous reports in rice that under IAA treatment, the expression trends of OsAGO1b (Os04g0566500) showed inconsistency in buds and roots⁷⁹. So far, the underlying causes of this phenomenon remain unclear. AGO protein is a core member of the ISC complex in RNA silencing signaling pathway⁸⁰, which regulates the function of target genes by loading sRNA. When AsAGO1 is induced, the expression of many sRNAs and its target genes will change. Therefore, we proposed that the response of AsAGO1 to IAA, along with its influence on growth and development, was likely a complex process that may not be regulated by a single factor.

When drought stress comes, specific receptors on the plasma membrane of plant cells perceive the drought signal, including G protein-coupled receptors (GPCRs)⁸¹, receptor-like kinases (RLKs)⁸², histidine kinases (HKs)⁸³, abscisic acid (ABA) receptors⁸⁴ and calcium-sensing receptors(CAS)⁸⁵. Subsequently, several secondary messengers are generated within the cell, which in turn activate downstream signal transduction pathways, including the ABA-mediated signaling pathway, Ca²⁺-activated CBL-CIPK signaling pathway, and MAPK cascade signaling pathway^86,87,88,89. Among them, the ABA-mediated signal transduction pathway is considered to be the main pathway mediating plants’ response to drought stress⁹⁰. Specifically, the signaling receptor PYR/PYL/RCAR protein complex binds to PP2Cs (phosphatase type 2Cs) and inhibits the activity of PP2Cs, thereby activating the phosphorylation activity of SnRK2s (phosphatsenon-fermenting SNF1-related protein Kinase2s). Finally, a series of downstream ABA signaling pathway response factors are activated to strengthen plant resistance to drought stress¹. Previous studies have shown that DOF family genes are responsive to the ABA signaling pathway^91,92,93,94. However, the role of DOF family genes in drought response via ABA signaling has not been reported. In this study, we found that the expression of AsDOF25 was up-regulated after ABA treatment (Supplemental Fig. 14e, f), and AsDOF25 was correlated with the ABA receptor PYR/PYL/RCAR protein complex in the co-expression network (Supplemental Fig. 14i). Furthermore, we also identified an ABRE motif in the AsDOF25 promoter region (Supplemental Fig. 14j). ABA generally regulates target gene expression through transcription factors known as ABRE-binding protein/ABRE-binding factors (AREB/ABF)^95,96,97. These results suggest that AsDOF25 may respond to drought stress via the ABA signaling pathway. However, more evidence is needed to support this hypothesis.

Both DOF and HSF family genes are involved in regulating plant growth and development in response to IAA signals^{98,99,100,101}, while also regulating plant drought tolerance through ABA signaling pathways^{94,102,103,104}. Our results further demonstrated the role of the AsDOF25-AsHSFA2c-AsAGO1 module in fine-tuning the balance between plant growth and drought tolerance. Unexpectedly, the expression levels of AsDOF25 and AsHSFA2c were up-regulated after IAA treatment, whereas a plant-enlarged phenotype was observed after the malfunction of AsDOF25 and AsHSFA2c (Figs. 3c, e, and 4e and Supplemental Fig. 14g, h). We hypothesized that this might be due to the expression of other plant growth-promoting genes downstream of the AsDOF25–AsHSFA2c module (Supplemental Fig. 18a–d). For example, the candidate downstream genes targeted by AsHSFA2c, namely oat043386 and oat123997, which were identified on DAP-seq, may be transcriptional repression by AsHSFA2c (Supplemental Fig. 18c, d), have homologs in rice (OsGAE1 and OsVTE5, respectively). Previous studies have shown that they play a positive regulatory role in promoting plant growth and development^105,106. Meanwhile, we evaluated the impact of the IAA concentration used in the experiment on oat growth and found that its growth was notably suppressed at 2 days after IAA treatment (Supplemental Fig. 19). Therefore, we hypothesize that this also may be associated with the high concentration of IAA treatment.

Furthermore, it is generally expected that IAA (a growth-promoting hormone) and drought stress would induce opposite gene expression patterns while our results showed that the expression levels of AsDOF25 and AsHSFA2c were up-regulated after both IAA and drought treatments. Previous studies have reported that there exists complex crosstalk among various signaling pathways that regulate growth and development, as well as responses to biotic and abiotic stresses in plant^{107,108,109,110,111}. Our transcriptome data revealed that most of genes exhibited opposite expression trends under IAA and drought treatments (Supplemental Fig. 2d). Conversely, a portion of genes exhibited similar expression patterns following drought and IAA treatments (Supplemental Fig. 2d). This may be because that plants also can actively respond drought stress by up-regulating the expression of growth-related genes⁹, such as the previously reported genes TaIAA15-1A, RRS1, TrIAA27, SOT17^{112,113,114,115}, or that exogenous IAA treatment significantly increased ABA and jasmonic acid (JA) content, leading to the up-regulated expression of drought stress-responsive genes, such as bZIP11, DREB2, MYB14, MYB48, WRKY2, WRKY56, WRKY108, and RD22¹¹⁶.

Methods

Plant materials, growth conditions, and treatments

Avena sativa (A. sativa; oat) plants were grown in a greenhouse at 25 °C/18 °C under a 16-h light/8-h dark photoperiod. N. benthamiana and A. thaliana plants were grown at 22 °C under a 12-h light/12-h dark photoperiod. ‘cv. Mengyan No.1’ was obtained from Inner Mongolia Agricultural University. To generate transgenic Arabidopsis thaliana plants overexpressing AsHSFA2c, the p35S::oat069028 (AsHSFA2c) was transformed into Col-0. To generate transgenic Avena sativa plants overexpressing AsHSFA2c, the pUBI::oat069028 (AsHSFA2c)-GFP was transformed into ‘cv. Bayou No.18’.

To initiate the experiment, seeds of ‘cv. Mengyan No.1’ and ‘cv. Bayou No.3’ were immersed in 75% ethyl alcohol for 3 min, and then in a 10% (w/v) sodium hypochlorite (NaClO) solution for 20 min. The seeds were washed four times with sterilized deionized water and then vernalized for 3 days at 4 °C under shaded light conditions. Seeds of N. benthamiana and A. thaliana, using a 75% ethyl alcohol for 2 min, a 3% (w/v) sodium hypochlorite (NaClO) solution for 10 min, were washed four times with sterilized deionized water and then vernalized for 3 days at 4 °C under shaded light conditions. Seeds of N. benthamiana and A. thaliana were immersed in 75% ethyl alcohol for 2 min, and then in 3% (w/v) sodium hypochlorite (NaClO) solution for 10 min. The seeds were washed four times with sterilized deionized water and then vernalized for 3 days at 4 °C under shaded light conditions.

Sterilized seeds were carefully placed on 1/2 MS medium supplemented with 1% (w/v) agar and adjusted to pH = 5.8. Germinated seeds were transferred to pots filled with 1/2 Hoagland solution, which was refreshed every 2 days. When the seedlings reached the trifoliate stage, they were subjected to the following four treatments: drought, ABA (spraying and soaking), IAA (spraying and soaking), and 20% PEG6000 solution (simulated drought). Whole plants were collected at specific time points after initiating the drought treatment [70% soil moisture content (watered, 0 day after treatment), 35% soil moisture content (moderate drought, 4 days after treatment), 0% soil moisture content (severe drought, 10 days after treatment), and recovered-24h]. Treatments were completed using solutions containing 20% PEG6000 to simulate drought conditions. Plant aerial parts and roots were collected at specific time points (0 h, 6 h, and 12 h) after initiating the PEG6000 treatment. Hormone treatments involved the following solutions: 250 μmol L⁻¹ (spraying)/125 μmol L⁻¹ (soaking) ABA and 250 μmol L⁻¹ (spraying)/125 μmol L⁻¹ (soaking) IAA. Plant aerial parts and roots were collected at specific time points (0 h, 1 h, 3 h, 6 h, and 9 h) after initiating the ABA and IAA treatments.

For plant organ-specific expression analysis, ‘cv. Mengyan No. 1’ plants were grown in pots and then various tissue samples were collected from various plant parts, including roots, stems, leaves, flag leaves, flowers, pods, and seeds.

RNA sequencing

Tissues from the roots and the plant aerial parts of plants treated with ABA, IAA, and PEG6000, as well as whole plants that underwent the drought treatment were collected for transcriptome sequencing. The tissue samples were frozen in liquid nitrogen and stored at −80 °C. Three biological replicates were collected for each tissue sample. Total RNA was extracted from the tissue samples using an RNeasy Plant Mini Kit (Qiagen). cDNA library was constructed by VAHTS Universal V6 RNA-seq Library Prep kit for Illumina (Vazyme, NR604-01). The mRNA of eukaryotic cells is enriched using Oligo (dT) magnetic beads and then fragmented by adding Fragmentation Buffer. Using mRNA as a template, a first-strand cDNA is synthesized with hexamer random primers. Then, with the addition of dNTPs, DNA polymerase I, and RNase H, the second-strand cDNA is synthesized. Purified double-stranded cDNA undergoes end repair, addition of an A-base, ligation of sequencing adapters, and fragment selection to recover cDNA fragments of approximately 350 bp. Finally, PCR enrichment to obtain the cDNA library, and paired-end sequencing was performed on a NovaSeq sequencing platform (Illumina) (Annoroad Gene Technology, Beijing, China).

Transcriptome analysis

The sequenced reads were trimmed by Fastp (Version: 0.20.1). Clean RNA-seq reads from the 84 samples were mapped to the ‘cv. Pinyan No.6’ genome by HISAT2 (version: 2.2.1)¹¹⁷. Gene expression levels were calculated in terms of the fragments per kilobase of exon per million mapped fragments (FPKM) value using StringTie (version: 2.1.6)¹¹⁸ with the parameter “-G -e -A”. Differentially expressed genes were identified using the DESeq2 R package (version:1.38.3), with default parameters. We filtered the DEGs with a minimum of two-fold differential expression (|fold change| ≥ 2; false discovery rate (FDR) ≤ 0.05;) and with the FPKM value higher than 1.

Identify genes involved in the ABA and IAA signaling pathways in oat

To identify genes involved in the ABA and IAA signaling pathways in oats, we first utilized the Rice Annotation Project Database (RAP-DB, http://rapdb.dna.affrc.go.jp/) to gather genes that have been previously identified as key components of the ABA and IAA signaling pathways. Subsequently, we utilized JCVI (version: 1.2.1) to construct the gene collinearity between rice and oat, identifying orthologous genes of rice in the oat genome. Based on annotations from RAP-DB, we then determined the members of the ABA and IAA signaling pathways in oats.

GO and KEGG enrichment analysis

Genes annotated with GO terms in the ‘cv. Pinyan No. 6’ genome annotated to 124,604 genes was considered as the background. The GO terms and KEGG pathways are assigned to the genes in ‘cv. Pinyan No.6’ were retrieved from the corresponding InterPro entry. GO and KEGG enrichment analyses were performed using the R package clusterProfiler. The resulting p values were corrected according to Benjamini and Hochberg’s method. GO terms and KEGG pathways with an adjusted p values < 0.05 were considered to be significantly enriched.

Construction of co-expression networks

To comprehensively evaluate the co-expression relationships among genes during drought response and growth in oats, gene expression levels were first calculated for the 84 transcriptome data for the different treatments associated with growth and drought tolerance in oats to obtain the fragments per kilobase of exon per million mapped fragments (FPKM) using StringTie (version 2.1.6)¹¹⁸ with the parameters -G -e -A. Then, differentially expressed genes (DEGs) were identified using the R package of DESeq2 (version 1.38.3) with the parameters of a minimum two-fold change in expression (FPKM > 0.1; |fold-change| ≥ 2; false discovery rate (FDR) ≤ 0.05)⁴. Next, we applied the R package of WGCNA to obtain the genes that were expressed in at least three samples (goodSamplesGenes (datExpr, verbose = 3))^119,120, and the co-expression network was constructed using 33,611 DEGs that met the criteria of FPKM > 0.1, a two-fold expression change between control and treatment, and a topological overlap value > 0.3. Cytoscape (version: 3.10.1) and Gephi (version: 0.10) were used to visualize the network and to extract network subsets.

Definition of the ‘balance’ module

To identify candidate modules involved in balancing drought tolerance and growth in oats, four conditions were required. First, more than 5% of the genes in the module were differentially expressed both in IAA-treated and drought-treated conditions. Second, based on the correlation analysis between different treatments and the modules, we identified the modules that were highly correlated with both of IAA and drought treatment, and the Pearson correlation coefficient is >0.8 for both IAA treatment and drought treatment. Third, gene functional enrichment analysis for each module was performed, and the modules with the functional enrichment items covering both of drought-related and growth-related items were defined as the candidate ‘balance’ modules. Additionally, the module, harboring key genes that have been previously reported to play important roles in plant development and drought tolerance, was also defined as the candidate ‘balance’ module. Finally, the modules that meet all the bove four conditions were defined as the ‘balance’ modules.

Correlation analysis

The Pearson correlation analysis was conducted to assess the correlation between gene modules and treatments^121,122. According to a Spearman correlation analysis, the degree of correlation between gene modules and treatments was determined on the basis of the correlation coefficient “r” (|r| > 0.8).

Gene family identification

HSF gene sequences were extracted from the ‘cv Pinyan No.6’ genome and then compared with HSF gene sequences from the O. sativa genome by BLAST and retrieved by HMMER (version: 3.4).

Phylogenetic analysis

To explore the phylogenetic relationships of the HSF gene family in A. sativa, we constructed phylogenetic trees comprising HSF genes from A. sativa, T. aestivum, and O. sativa. Three subclasses were assigned according to the relationships of HSF gene family members in O. sativa (https://ricedata.cn/). The phylogenetic tree of the HSF gene families was constructed using iqtree (version: 2.2.2.7)¹²³. The website iTOL was used for the visualization of the phylogenetic tree¹²⁴.

DAP-seq

DAP-seq was performed at Bluescape Hebei Biotech^{125,126,127,128} to purify gDNA from the leaves of Avena sativa. Genomic DNA (gDNA, 5 μg in 130 μL TE buffer) was extracted using the CTAB (Sangon Biotech, A600108) and fragmented to an average of 200 bp using a Covaris M220 (Woburn, MA, USA). A genomic DNA library was prepared using a DNA Affinity Purification Sequencing Kit (Bluescape Hebei Biotech Co., Ltd.). AsHSFA2c was fused to the Halo affinity tag and expressed using the TNT SP6 Coupled Wheat Germ Protein Expression System (Bluescape, Hebei, China) for expression in a 50 μL reaction with a 2 h incubation at 37 °C. AsHSFA2c protein and the gDNA library were incubated in vitro and DNA bound to AsHSFA2c protein was isolated by the HaloTag beads (Promega)^71,129.

To significantly decrease or eliminate false peak signals, the input DNA was used as a negative control sample for background reference. Two biological replicates were conducted for the experiment. DNA obtained through affinity purification and elution was subjected to paired-end sequencing using an Illumina HiSeq platform, with quality-filtered reads aligned to the ‘cv. Pinyan No.6’ genome sequence using BWA. In addition, peak calling was performed by MACS2¹³⁰, whereas the irreproducibility discovery rate (IDR) method was used to obtain highly reproducible peaks (IDR < 0.05) (https://github.com/nboley/idr)¹³¹. Then we retained the peaks overlapping the 3 kb regions upstream and downstream of genes for the subsequent analysis. Finally, motifs were identified using the Simple MEME Wrapper function model of TBtools¹³².

Synteny analysis

Genome-wide synteny analysis between A. sativa and O. sativa. was conducted using JCVI (Version: 1.2.1) with the default parameters.

Virus-induced gene silencing mediated by TRV

Agrobacterium tumefaciens strain GV3101 with the tobacco rattle virus (TRV) full-length infectious clone TRV was cultured at 28 °C in LB medium containing 25 μg mL⁻¹ rifampicin, and 25 μg mL⁻¹ kanamycin.

To investigate the potential role of AsMYC2, AsHSFA2c, AsDOF25, and AsAGO1 in drought tolerance and growth of oats, we employed the Tobacco Rattle Virus (TRV) virus-induced gene silencing (VIGS) system to knock down the expression of these genes. Three fragments of different lengths within these gene ORFs, namely V-AsMYC2-1, V-AsMYC2-2, V-AsMYC2-3, V-AsHSFA2c-1, V-AsHSFA2c-2, V-AsHSFA2c-3, V-AsDOF25-1, V-AsDOF25-2, V-AsDOF25-3, V-AsAGO1-1 were selected for these purposes. These fragments were subcloned into the TRV2 infection plasmid vector using an NC clone (NC Biotech). To initiate the virus-induced gene silencing assay, A. tumefaciens strains GV3101 carrying pTRV1 and different pTRV2 derived vectors (TRV2, V-AsMYC2, V-AsHSFA2c, V-AsDOF25, and V-AsAGO1) in a 1:1 ratio were mixed in medium supplemented with acetosyringone (AS) (19.62 mg L⁻¹), cysteine (Cys) (400 mg L⁻¹), and Tween-20 (5 mL L⁻¹). The infection experiment conditions were set to 20 kPa vacuum infiltration for 5 min¹³³. And then, co-cultured overnight at 28 °C with shaking at 180 rpm. After co-cultivation, the Agrobacterium-infected (germinated) seeds were washed with sterile water to remove surface-adhered Agrobacterium. Finally, these seeds were planted in the soil.

Plasmids and cloning procedures

To generate entry constructs, the different lengths ORF fragment regions of AsMYC2, AsHSFA2c, AsDOF25, and AsAGO1 were amplified from ‘cv. Mengyan No.1’ cDNA. These fragments were then cloned into the TRV2 infection plasmid vector using NC clone (NC Biotech) to generate the following constructs: pTRV2::AsMYC2-1, pTRV2::AsMYC2-2, pTRV2::AsMYC2-3, pTRV2::AsHSFA2c-1, pTRV2::AsHSFA2c-2, pTRV2::AsHSFA2c-3, pTRV2::AsDOF25-1, pTRV2::AsDOF25-2, pTRV2::AsDOF25-3, and pTRV2::AsAGO1.

A 3,218-bp genomic DNA fragment containing the AsHSFA2c promoter and genic sequence was cloned into pCambia1305-GFP using the ClonExpressII One Step Cloning Kits (Vazyme, Nanjing, China) to generate the pAsHSFA2c::AsHSFA2c-GFP construct. The full-length CDS region of AsHSFA2c was cloned into pCambia3300-GFP using the ClonExpressII One Step Cloning Kits to generate the pUBI::AsHSFA2c-GFP construct. A 2,000-bp genomic DNA fragment containing the AsHSFA2c promoter sequence was cloned into pNC-Green-LUC using the NC clone mix (NC Biotech) to generate the pAsHSFA2c::LUC construct. A 3,000-bp genomic DNA fragment containing the AsAGO1 promoter sequence was cloned into pNC-Green-LUC using the NC clone mix (NC Biotech) to generate the pAsAGO1::LUC construct.

The full-length CDS region of AsDOF25 and a 936-bp DNA fragment containing 1-312 amino acids from AsDOF25 were amplified from ‘cv. Mengyan No.1’ cDNA, these fragments were then cloned into pCambia3304 (containing GFP tag, not fused with the target gene) with NC clone mix (NC Biotech) to generate p35s::AsDOF25 and p35s::AsDOF25-D₃₁₂ constructs. The full-length CDS region of AsHSFA2c was amplified from ‘cv. Mengyan No.1’ cDNA, these fragments were cloned into pCambia3304 with NC clone mix (NC Biotech) to generate p35s::AsHSFA2c constructs. A 438-bp DNA fragment containing 1-146 amino acids from AsHSFA2c was amplified from ‘cv Mengyan No.1’ cDNA, the fragments were then cloned into pCambia3304 with NC clone mix (NC Biotech) to generate p35s::AsHSFA2c-bd constructs.

For prokaryotic expression, the pGEX4T-1-GST-AsDOF25 construct was generated by inserting AsDOF25 full-length CDS into pGEX4T-1 digested with BamHI and EcoRI. pGEX4T-1-GST-AsHSFA2c construct was generated by inserting AsHSFA2c full-length CDS into pGEX4T-1 digested with BamHI and EcoRI. All primers are listed in Supplemental Data 18.

Luciferase assay

For AsDOF25 activitied AsHSFA2c promoter activity assay, A. tumefaciens GV3101 strain carrying p35s::AsDOF25 and pAsHSFA2c::LUC constructs were co-infiltrated into the leaves of 4-week-old N. benthamiana plants. Two days after infiltration, plant samples were collected to detect fluorescence due to luciferase activity and conduct a qRT-PCR analysis. The p35s::AsDOF25-D₃₁₂ construct was used as a negative control. For AsHSFA2c inhibited AsAGO1 promoter activity assay, A. tumefaciens GV3101 strain carrying p35s::AsHSFA2c and pAsAGO1::LUC constructs were co-infiltrated into leaves of 4-week-old N. benthamiana plants. Two days after infiltration, plant samples were collected to detect fluorescence due to luciferase activity and conduct a qRT-PCR analysis. The p35s::AsHSFA2c-bd construct was used as a negative control.

Reverse transcription PCR analysis

Total RNA was extracted with TRIzol reagent (Thermo Fisher Scientific). 1 μg total RNA was converted to cDNA with the PrimeScript RT Reagent Kit (Takara, RR047A), and the RT-qPCR was performed with a TB Green Premix Ex Taq kit (Tli RNase H Plus) (Takara, R420A). AsACT2 was used as an internal control. All primers used in the RT-qPCR assays are listed in Supplemental Data 18.

Genetic transformation in Avena sativa

Mature embryos of healthy ‘cv. Bayou No.18’ plants grown in a well-conditioned greenhouse were collected and cultured on L3-M medium (4.6 g L⁻¹ L3 base salts with vitamins, 30 g L⁻¹ maltose, 4 g L⁻¹ phytagel, 2 g L⁻¹ 2,4-D, 1 g L⁻¹ dicamba) until embryonic calli were produced¹³⁴. Agrobacterium tumefaciens strain GV3101 with pUBI::AsHSFA2c-GFP was cultured at 28 °C in YEP medium overnight. After centrifugation at 25 °C and 5,000 rpm for 10 min, the precipitate was resuspended in WLS solution (pH 5.8) consisting of 4.30 g Linsmaise & Skoog Base Salts, 100 μL 1000 × MS vitamins, 10 g glucose, and 0.5 g MES (per liter of H₂O) for an optical density (OD) of 0.5¹³⁵. Embryonic calli were immersed in a mixture comprising equal proportions of A. tumefaciens GV3101 cells containing pUBI::AsHSFA2c-GFP and A. tumefaciens GV3101 cells containing pUBI::TaWOX5 for 30 min. Then, the embryogenic callus was removed, and the residual bacterial liquid was absorbed using filter paper. Embryonic calli were collected, after which the residual bacterial mixture was absorbed using filter paper and the embryonic calli were cultured on filter paper containing 75 µmol acetosyringone in darkness for 3 days. Next, embryonic calli were cultured on WLS-RES medium (4.6 g L⁻¹ L3 base salts with vitamins, 2.2 mg L⁻¹ Picloram, 0.5 g L⁻¹ glutamic acid, 0.1 g L⁻¹ Casein acid Hydrolysate, 0.75 g L⁻¹ MgCl₂‧6H₂O, 40 g L⁻¹ Maltose, 1.95 g L⁻¹ MES, 4 g L⁻¹ phytagel, 0.5 mg L⁻¹ 2,4-D, 0.85 mg L⁻¹ AgNO₃, 0.1 mg L⁻¹ Ascorbic Acid, 0.2 g L⁻¹ Timentin, PH 5.8) for 5 days. WLS-P5 medium (the wls-res medium containing 0.5–5 mg L⁻¹ Basta) was used for gradient screening, with surviving calli transferred to regeneration medium (pH 5.8) comprising 4.6 g L3 base salts with vitamins, 5 mg zeatin, 20 g sucrose, 0.5 g MES, 200 μL 12.5 g L⁻¹ CuSO₄‧5H₂O, and 4 g phytagel (per liter of H₂O)^136,137. The calli were grown until they reached a size of 3–5 cm. Finally, the roots were cultured in a rooting medium (pH 5.8) comprising 4.6 g L3 base salts with vitamins, 0.2 mg L⁻¹ IBA, 15 g sucrose, 0.5 g MES, and 4 g phytagel (per liter of H₂O). AsHSFA2c expression was analyzed by qRT-PCR.

Drought phenotype analyses

For drought phenotype analyses of oats at the seedling stage, ten seedlings were planted in one pot with a soil mixture. The seedlings were fully watered at the trifoliate stage and then stopped water supply until the gene knock-down lines/ the gene overexpression lines had different leaf wilting phenotypes compared with the negative control line (TRV1TRV2/ pUBI::GFP transgenic lines). Photographs were taken after a 2-day period of recovery with full irrigation post-drought treatment. Survival rates were recorded after rewatering for 2 days.

Water loss rate analyses

The seedlings were grown until the second leaf totally expanded under well-watered conditions. The leaves were taken from six plants in the same state, which were cut off and weighed immediately. Placed on filter paper at room temperature for 0 h, 1 h, 2 h, 3 h, 4 h, 5 h, 6 h, 7 h, 8 h, and 9 h, and the leaves were weighed at each time point. Three biological replicates were performed for one experiment.

Protein purification

Recombinant proteins were expressed in E. coli transseta cells. For AsDOF25 protein purification, the protein expression was induced with 0.2 M IPTG, and the E. coli bacteria was cultured at 16 °C overnight. For AsHSFA2c protein purification, the protein expression was induced with 0.5 mM IPTG, and the E. coli bacteria was cultured at 28 °C overnight. Bacterial cells collected were resuspended in lysis buffer [1% Triton X-100, 50 mM Tris-HCl (pH 8.0), 200 mM NaCl, 1 mM DTT, and 1 pellet per 50 mL of complete EDTA-free protease inhibitor (Roche)] and lysis by sonication. After centrifugation, supernatants were affinity-extracted with Glutathione Sepharose 4B¹³⁸.

Electrophoretic mobility shift assay

Purified GST-AsDOF25 protein or GST-AsHSFA2c protein were incubated with biotinylated or non-biotinylated double-stranded DNA in the binding buffer (Thermo) at room temperature for 25 min. The binding reaction mixture was resolved with 6% native PAGE at 4 °C. The interaction between proteins and DNA probes was detected by LightShift™ EMSA Optimization and Control Kit (Thermo)¹³⁹.

Leaf cell observation

The third leaf of trifoliate stage ‘cv. Mengyan No.1’ was treated with ethanol, which was soaked in 75% ethanol overnight. Then leaves were soaked in 2% methanol containing 0.24 M HCl for 25 min at 37 °C. Followed, at room temperature, soaking in 60% ethanol containing and 7% NaOH for 25 min¹³⁹. Leaves were observed with a stereoscope.

Protein extraction and western blot

Total protein was extracted by the extraction buffer [0.2 M NaCl, 5 mM MgCl₂, 5 mM DTT, 20 mM Tris-HCl (pH 7.5), 0.03% Tween-20 (Ameresco), and 0.5 tablets of protease inhibitor (Roche)]. The supernatant was collected by centrifuging at 12,000 rpm for 15 min. Total proteins were examined by western blot analysis using α-tubulin (1:5000; EASYBIO, BE0031) as a loading control. Proteins in the study were also probed with α-GFP (1:2000; EASYBIO, BE2001), α-H3 (1:2000; EASYBIO, BE7004), α-PEPC (1:2000; Agrisera, As09458). Secondary antibodies were goat anti-rabbit IgG (1:5000; EASYBIO, BE0101) and goat anti-mouse IgG (1:5000; EASYBIO, BE0102). The protein Marker (product #26616) purchased from Thermo Scientific was used in all the western blot assays in this manuscript. The instrument (BIO-01, O1900) was used to obtain images.

Protein colocalization in plants

To determine the subcellular localization of AsHSFA2c, A. tumefaciens containing pAsHSFA2c::AsHSFA2c-GFP was inoculated in N. benthamiana, respectively. The images were taken under confocal fluorescence microscopy (LSM900).

Isolation and purification of nucleus and cytoplasm

3 g aerial parts of 4-week-old N. benthamiana under 12-h light/12-h dark photoperiod were ground to a fine powder in liquid nitrogen and were homogenized with 10 mL nuclei isolation buffer (20 mM Tris-HCl (pH 7.4), 5 mM MgCl₂, 25% glycerol, and 5 mM DTT) at 4 °C. The homogenate was filtered through a double layer of Miracloth, and the resulting flow-through was centrifuged at 200 g for 5 min at 4 °C. The supernatant consisting of the cytoplasmic fraction was collected by 12,000 g centrifugation for 10 min at 4 °C and the pellet was washed five times with nuclear wash buffer (20 mM Tris-HCl (pH 7.4), 5 mM MgCl₂, 25% glycerol, 5 mM DTT, and 0.2% Triton X-100) at 1500 g for 3 min. α-H3 (1:2000; EASYBIO, BE7004) and α-PEPC (1:2000; Agrisera, As09458) were used as nuclear and cytoplasmic protein marker, respectively¹⁴⁰.

Nuclear run-on assay

To determine the transcription after drought treatment, 3 g trifoliate stage oat leaves were collected. Briefly, plant samples were ground in liquid nitrogen and were suspended in 30 mL lysis buffer (20 mM Tris-HCl (pH 7.5), 20 mM KCl, 2 mM EDTA (pH 8.0), 2.5 mM MgCl₂, 25% Glycerol, 250 mM Sucrose, 5 mM DTT, 1 pellet per 50 mL of cOmplete EDTA-free protease inhibitor)^139,140. After centrifugation at 4 °C, the precipitate was washed four times with NRBT buffer (20 mM Tris-HCl (pH 7.4), 25% Glycerol, 2.5 mM MgCl₂, 0.2% Triton X-100, 4 mM DTT). Then, precipitates were resuspended with nuclei storage buffer (50 mM Tris-HCl (pH 7.8), 1 mM DTT, 20% Glycerol, 5 mM MgCl₂, 0.44 M Sucrose) and added to the transcription system:10 μL 10× Transcription buffer (50 mM Tris-HCl (pH 7.5), 5 mM MgCl₂, 150 mM KCl, 0.2% Sarkosyl, 20 U/mL RNase inhibitor, 1 mM DTT), 5 μL NTP mixture (100 mM ATP, 100 mM CTP, 100 mM GTP, 100 mM BrUTP) and 35 μL RNase-free H₂O. The run-on reaction was performed at 37 °C for 45 min. Total RNA was extracted from the transcription system using TRIzol reagent followed by DNase I (NEB) treatment. The purified RNAs were incubated with 60 μL anti-BrdU beads (Santa Cruz) at 4 °C for 2 h. After IP, beads were washed twice with low-salt buffer (0.2 × SSPE (0.03 M NaCl, 2 mM NaH₂PO₄, 0.2 mM EDTA, pH 7.4), 1 mM EDTA (pH 8.0), 0.05% Tween-20) and then twice with high-salt buffer (0.5 × SSPE (0.075 M NaCl, 5 mM NaH₂PO₄, 0.5 mM EDTA, pH 7.4), 0.05% Tween-20, 37.5 mM NaCl, 1 mM EDTA (pH 8.0)) at 4 °C. The precipitated RNA was extracted using TRIzol reagent and used for qRT-PCR analysis.

Chlorophyll measurement

The leaves were incubated in 95% (v/v) ethanol for 5 days in darkness to obtain a solution containing chlorophyll. The absorbances were measured at 665 and 649 nm. The chlorophyll contents were calculated according to the following ratio: (6.63A₆₆₅ + 18.08A₆₄₉)/g fresh weight¹⁴¹.

Statistics and reproducibility

The analysis was performed using GraphPad Prism version 6.01 (San Diego, California, USA). Sample sizes were chosen based on previous publications^{12,13,111,142,143,144}, the detailed specifics are presented in the corresponding methods section. Each value shown for each group is the mean ± SD (standard deviation). The p value was calculated using Student’s t-test (two-tailed). p values < 0.05 were considered significant, while p values > 0.05 were considered non-significant. Each images represent at least three independent biological repeats.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.