Barley plants (Hordeum vulgare cultivar Golden Promise) were grown in pots filled with water-soaked vermiculite in a climate-controlled growth chamber under long-day conditions with 16 h light at 20°C and a light intensity of 120 µmol photons m^-2 s^-1 (Philips TLD 18W of alternating 830/840 light color temperature) and 8 h darkness at 18°C for five days.

Five-day-old seedlings were harvested and washed carefully to remove any remaining vermiculite prior to submersion in 10 mM H₂O₂ (Carl Roth, Germany) or ddH₂O (control) for three hours. The duration of H₂O₂ treatment was selected based on previous studies, which showed that at this time point H₂O₂ induced the strongest changes in the expression of most of the H₂O₂-responsive genes (Desikan et al., 2001; Stanley Kim et al., 2005; Hieno et al., 2019). Subsequently, seedlings were carefully rinsed with ddH₂O and dissected into roots and leaves. Samples were shock-frozen in liquid nitrogen and homogenized using a sterile, ice-cold mortar and pestle. Total RNA was extracted using the Quick-RNA miniprep Kit (Zymo Research, USA) according to the manufacturer’s instructions. The yield and purity of extracted RNA was determined with a NABI Nanodrop UV/Vis Spectrophotometer (MicroDigital, South Korea). The integrity of the extracted RNA was verified by separation of the 28S and 18S rRNA bands on a 1% agarose gel.

RNA sequencing was performed on three biological replicates for each treatment. Each replicate furthermore consisted of pooled material from three plants. Library preparation and transcriptome sequencing (3’ mRNA sequencing) were carried out at the NGS Core Facility (Medical Faculty at the University of Bonn, Germany) using a NOVASEQ 6000 (Illumina, USA) with a read length of 1x100 bases and an average sequencing depth of >10 million raw reads per sample ( Table 1 ). 3’ end sequencing libraries were prepared using the QuantSeq protocol (Moll et al., 2014). Briefly, oligo dT priming were followed by synthesis of the complementary first strand without any prior removal of ribosomal RNA. After successful introduction of Illumina specific adapter sequences, the resulting cDNA was further purified with magnetic beads. The unpaired reads were processed for quality control using fastQC and cutAdapt (Martin, 2011) in order to trim any remaining adapter sequences. They were then aligned using Tophat2 software (Trapnell et al., 2012) against a H. vulgare IBSC v2 reference genome obtained from Ensembl (http://plants.ensembl.org/info/data/ftp/index.html) using a Bowtie index (Langmead and Salzberg, 2012) created with the help of the reference genome (in FASTA format; the individual FASTA files of the chromosomes were concatenated using the “cat” command in UNIX shell). The alignment with Tophat2 was performed on an Ubuntu 18.04 LTS operating system, in a UNIX shell environment. Every step after alignment was performed using R 4.0.0 (R Core Team, 2020). Gene counts from the aligned BAM files were generated using featureCounts function in RStudio (Liao et al., 2014). Differential gene expression analyses was carried out using DESeq2 (Love et al., 2014). The p-values were corrected using the False Discovery Rate (FDR) method (Benjamini and Hochberg, 1995) and subsequently the FDR and the log₂FC cutoffs were set to 0.01 and 1, respectively. Principal Component Analyses (PCA) plots were prepared with the raw gene counts for all samples and replicates using the tidyverse and ggplot2 packages. The volcano plots and heatmaps were generated using the EnhancedVolcano and Pheatmap packages, respectively. In addition, transcript per million (TPM) values of each gene were calculated using a separate function designed in the R environment ( Supplementary Table S1 ). With common regulated DEGs, a clustering was performed with four predefined clusters based on FDR and log₂FC cutoffs of 0.01 and 0.5, respectively. The first and second cluster consisted of commonly down- and up-regulated genes, respectively, while the third and fourth cluster contained counter-regulated genes between leaves and roots of barley. The clusters were then represented as heatmaps using the pheatmap package and line plots using the ggpubr package.

Gene ontology (GO) and enrichment analyses were carried out using shinyGO (Ge et al., 2020). Categories were chosen as significant if the FDR was less than 0.05 (Benjamini and Hochberg, 1995). Homology searches against the A. thaliana genome were carried out using the BaRT (Barley Reference Transcript) tool available on www.ics.hutton.ac.uk (Mascher et al., 2017) based on a E-value cutoff of 1e^-30.

qRT-PCR was performed with three replicates for each sample. Each replicate consisted of the pooled RNA material from three different plants. Synthesis of first strand cDNA for qRT-PCR was carried out from at least 1 µg of total RNA using the RevertAid first strand cDNA synthesis kit (Thermo Fisher Scientific, USA) with oligo-dT₁₈ primers following the manufacturer’s instructions. The quality of cDNA was assessed using a NABI UV/Vis Nanodrop Spectrophotometer. Gene expression was quantified in 48-well plates using a BioRad CFX 96 real-time PCR detection system (BioRad, Germany) and a SYBR Green PCR master mix (Thermo Fisher Scientific, USA). All forward and reverse primers used for qRT-PCR are listed in Supplementary Table S2 . Data were quantified using the BioRad CFX Maestro software, and the expression was estimated using the 2^–δδCt method (Livak and Schmittgen, 2001) after normalization against the two reference genes HvACTIN and HvGAPDH, as the C_q values of both genes were unchanged upon H₂O₂ treatment. Data were analyzed statistically with one-way analysis of variance (ANOVA) and Tukey’ Post-Hoc HSD test using the agricolae and tidyverse packages, respectively. Graphs were prepared using the ggpubr package.

Staining of hydrogen peroxide in barley leaves and roots was performed with 2’,7’-dichlorodihydrofluorescein diacetate (H₂-DCFDA; Thermo Fisher Scientific, USA) based on a modified protocol (Kaur et al., 2016). Briefly, five-day-old barley seedlings were treated with either 10 mM H₂O₂ or ddH₂O (control) for 3 hours. Afterwards, the seedlings were briefly rinsed and treated with 10 µM H₂-DCFDA prepared from a 4 mM stock dissolved in DMSO for 1 hour in the dark. After staining, seedlings were washed, and roots and leaves were mounted separately on a microscopy slide. 2’,7’-Dichlorfluorescein (DCF) fluorescence was analyzed using a Leica SP8 Lightning confocal laser scanning microscope (Leica Microsystems, Germany). For excitation, an argon laser with a wavelength of 488 nm was used, and emission of 517-527 nm was detected using a HyD Detector. Fluorescence intensity was quantified in regions of interest (ROI) using the integrated LASX software.

To investigate the transcriptomic modulation in barley (Hordeum vulgare cv. Golden Promise) in response to oxidative stress, five-day-old plants were exposed for three hours to 10 mM H₂O₂ or to ddH₂O as control ( Figure 1A ). H₂-DCFDA staining confirmed that H₂O₂ penetrated both roots and leaves ( Figures 1B, C and Supplementary Figure 1 ). RNA was then extracted separately from roots and leaves, and RNA-seq analysis was carried out on three biological replicates per tissue and treatment, each comprising the pooled RNA from three different plants ( Supplementary Table S1 ). On average approximately 13 million total reads were obtained per sample. About 75-85% of these reads could be aligned to the barley reference genome ( Table 1 ). To assess the main variances within the dataset, a principal component analysis (PCA) was performed. The result showed that PC1 (X-axis), which separates the samples by tissue, represents the largest variation in our dataset compared to PC2 (Y-axis), which separates the samples by treatment ( Figure 2A ). Consequently, the differential gene expression analysis was separately performed for the leaf and root samples.

Differentially expressed genes (DEGs) between H₂O₂-treated and control samples were identified based on fold change (FC) │Log₂FC ≥ 1│ and FDR < 0.01 ( Supplementary Table S3 ). A total number of 2884 DEGs were detected across both tissues. H₂O₂ application clearly resulted in stronger transcriptional changes in roots compared to leaves ( Figure 2B ). Of the 1883 DEGs detected in roots, 701 were up- and 1182 were down-regulated, while in leaves 1001 DEGs were identified with 546 up- and 455 down-regulated ( Figure 2C ). Among all DEGs only 75 and 134 were commonly up- and down-regulated, respectively, in both tissues, while 37 were counter-regulated.

GO classification was used to identify the 20 most significant biological process categories within the DEGs. The results show that not only the number of genes, but also the biological processes affected by H₂O₂ were clearly different between leaves and roots ( Figure 3 ). In leaves, GO terms associated with genes that showed the highest fold change were related to protein complex oligomerization, response to H₂O₂ and jasmonate. Further categories with lower fold change but often higher number of genes comprised quite global stress effects associated with different, mostly abiotic stimuli, but also wounding ( Figure 3A ). In roots, many of the enriched GOs were associated with response to oxygenic stress including H₂O₂ catabolism, glutathione and ROS metabolism, or cellular oxidant detoxification as well as with cell wall modulation ( Figure 3B ).

As described above, we identified a total of 246 common DEGs between leaves and roots of barley when using a │log₂FC ≥ 1│cutoff ( Supplementary Table S3 , Figure 2C ). For several genes, we noticed that they were differentially regulated in both tissues, however, in one tissue they showed an expression with a FC>2 (│log₂FC ≥ 1│) while in the other tissue a FC less than 2 but higher as 1.5 Thus, for (│log₂FC between 1 and 0.5│) was detected. determination of commonly regulated genes in leaves and roots we used a cutoff of Log₂FC≥0.5 and listed these genes separately in Supplementary Table S3 . Using this cut-off, a total 349 common DEGs were identified between roots and leaves of barley ( Supplementary Figure S2 ; Supplementary Table S3 ). Of these, 116 and 176 genes were up- and down-regulated, respectively, while 58 genes showed counter-regulation. These common DEGs were organized in four clearly distinguishable clusters ( Figure 4A ), with either commonly down- (cluster 1) and up-regulated (cluster 2) genes or genes up-regulated in leaves but down-regulated in roots (cluster 3) and vice versa (cluster 4). Heat maps and line plots were constructed to visualize the changes in gene expression pattern for each cluster ( Figures 4A, B ).

In order to confirm the results obtained from RNA-seq analyses, we performed quantitative RT-PCRs (qRT-PCR) on some of the identified DEGs. For these, we selected several DEGs that showed common regulation in leaves and roots in our dataset and which, based on their functional annotation, could be related to oxidative stress ( Supplementary Table S5 ). Orthologs to some of them had already been shown to play an important role in H₂O₂ and ROS-related signaling not only in Arabidopsis but also in important crops like wheat, maize, and rice (Polidoros et al., 2005; Mylona et al., 2007; Steffens, 2014; Dudziak et al., 2019). They also represent different levels of regulation, some being among the most highly up- or down-regulated genes and other showing a much more subtle response. These genes represent different gene ontologies, and encode for a catalase, a peroxidase, a glutathione S-transferase, several TFs, a MAPKKK, and a xyloglucan endotransglucosyalase, a protein involved in cell wall modification. As shown in Figure 7 and in Supplementary Table S5 , the log₂FC changes observed with the different techniques were often quite close and, in all cases, the results of the qRT-PCR matched the trend observed in the RNA-seq data.

Our data showed that several genes encoding for small heat shock proteins (SHSPs) were up-regulated by H₂O₂ in barley leaves ( Table 2 ). In barley, the roles of several HSPs in response to a diverse range of abiotic stimuli have been characterized (Hlaváčková et al., 2013; Chaudhary et al., 2019; Landi et al., 2019). HSPs have also been shown to play crucial roles during abiotic stresses such as cold and heat in other important crop genera, like rice, maize, and wheat (ul Haq et al., 2019). SHSPs are a subgroup of HSPs defined by their size and a conserved α-crystalline C-terminal domain. They are known to form oligomeric complexes and prevent denatured proteins from aggregation until they can be refolded by other HSPs. They have been speculated to interact with transcription factors of the HEAT SHOCK FACTOR (HSF) family to create the HSP-HSF complex, alteration of which can drive essential reactions in response to ROS (Driedonks et al., 2015). The SHSPs in our data set belong to subfamilies with close orthologs in Arabidopsis, i.e. HSP17.6, 15.4, 15.7, and 17.4 (Li and Liu, 2019). HSP17.6 and HSP15.7 have been shown to be localized in the peroxisomes in Arabidopsis (Ma et al., 2006; Li et al., 2017). Peroxisomes are one of the main subcellular compartments in which ROS are produced by processes such as ß-oxidation and photorespiration, and which are crucial for antioxidant defense (Sandalio et al., 2013; del Río and López-Huertas, 2016). Additionally, HSP17.4 and 17.6 have been shown to exhibit increased transcript levels during periods of abiotic stress in Arabidopsis (Swindell et al., 2007). Thus, the induction of these HSPs points to a potential role of these proteins in increasing the tolerance to oxidative stress also in barley leaves. The single down-regulated SHSP is an ortholog to AtHSP15.4, for which this contrary behavior upon stress was already described (Siddique et al., 2008).

Not surprising, considering the well-established juxtaposition between ROS production and photosynthesis, the application of H₂O₂ negatively affected several photosynthetic components ( Table 2 ). The most affected group represents chlorophyll a/b binding proteins orthologous to various light-harvesting complex proteins of the LHCb-type and to a component of the light-harvesting complex I, LHCa1, of Arabidopsis. Down-regulation of LHCb-type proteins upon oxidative stress has been previously described (Staneloni et al., 2008). It is likely part of an established photoprotection mechanism to alleviate increased ROS levels generated when the photosynthesis reaction becomes unbalanced, e.g., under high light conditions.

The role of phytohormones like ABA and jasmonate in response to several biotic and abiotic stimuli has been extensively studied in plants (Verma et al., 2016). In our data, several genes related to jasmonate signaling were found to be down-regulated ( Table 2 ), including an ortholog of Arabidopsis 12-OXOPHYTODIENOATE REDUCTASE (OPR). The OPR3 protein of Arabidopsis has been denoted as one of the most crucial enzymes in jasmonate synthesis, which converts 12-oxophytodieonic acid (cis-OPDA) to OPC8:0 in peroxisomes (Bittner et al., 2022). However, recent studies highlighted the role of an OPR3-independent pathway for jasmonic acid (JA) biosynthesis, involving an OPR2-mediated alternative bypass via dinor-OPDA (dnOPDA) and 4,5-didehydro-JA, which is then converted to JA (Chini et al., 2018). Interestingly, we found a down-regulation of the barley ortholog of OPR2 in leaves, the consequence of which remains speculative due to the unclear role of the OPR3-independent bypass pathway. By contrast, genes coding for ALLENE OXIDE CYCLASE (AOC) and ALLENE OXIDE SYNTHASE (AOS) were up-regulated in leaves. These enzymes catalyze the generation of both cis-OPDA and dnOPDA, which in turn would increase OPDA production for both pathways. This is interesting, because OPDA is believed to have an independent regulatory function both on transcription (similar to JA-Ile), but also on protein activity by OPDadylation. Moreover, OPDA-mediated signaling seems closely associated with thiol metabolism and redox-mediated processes (Böttcher and Weiler, 2007; Ohkama-Ohtsu et al., 2011; Bittner et al., 2022). Also related to jasmonate signaling are two TIFY domain-containing proteins that were induced in response to H₂O₂ ( Table 2 ). The TIFY domain is found in members of the JASMONATE ZIM DOMAIN (JAZ)-type transcriptional repressors involved in jasmonate signaling (Chung and Howe, 2009; Pauwels and Goossens, 2011). However, no regulation of TFs associated with jasmonate signaling was detected in our data set.

By contrast, many of the genes associated with other phytohormones, e.g. auxins and ABA, encode TFs or other proteins involved in transcription regulation ( Table 2 ). Several of these genes belong to the large family of AP2/ERF-type TFs, members of which have been associated with environmental stresses including hypoxia and oxidative stress. While mostly associated with ethylene, AP2/ERF function is also connected to ABA, gibberellic acid, cytokinin, and brassinosteroids (Xie et al., 2019). The largest group of genes associated with hormones relates to auxin ( Table 2 ), the role of which is mostly associated with development and growth. However, experimental evidence linked auxin also to oxidative stress, especially auxin-mediated stress-dependent cell proliferation including the RSL-type TF ROOT HAIR DEFECTIVE SIX-LIKE4 (RSL4) that targets NADPH oxidases also known as respiratory burst oxidase homologs (RBOHs) and secreted plant-specific type III peroxidases that impact apoplastic ROS homeostasis and in turn stimulate root hair cell elongation (Pasternak et al., 2005; Iglesias et al., 2010; Mangano et al., 2017).

In roots, many DEGs were found to be associated with the detoxification of H₂O₂ ( Table 3 ), especially peroxidases and genes related to glutathione metabolism. GLUTATHIONE TRANSFERASES (GSTs) and GLUTATHIONE PEROXIDASES (GTPs) have both been shown to be involved in plant stress responses (Bela et al., 2015; Nianiou-Obeidat et al., 2017). However, somewhat surprisingly, our data showed clear down-regulation of several GSTs and GTPs along with other key players associated with H₂O₂ detoxification such as orthologs of Arabidopsis ASCORBATE PEROXIDASE 1 (APX1) and CATALASE 1(CAT1). Moreover, two putative DETOXIFICATION EFFLUX CARRIERS/MULTIDRUG AND TOXIC COMPOUND EXTRUSION (DXT/MATE) proteins were strongly up-regulated in roots. The MATE family proteins facilitate the efflux of various compounds including substances, such as hormones or flavonoids, that improve adaptation to stress (Ku et al., 2022).

The largest set of genes whose expression was affected in response to H₂O₂ belongs to class III plant type peroxidases ( Table 3 ), whose role in plant defense mechanisms in response to a wide variety of biotic and abiotic stresses is well established. They play an important role in the cellular redox homeostasis upon stress. In addition, they also catalyze the oxidation of a variety of substrates and have been linked to processes involved in cell wall stability, including lignin and suberin polymerization in response to stress (Kidwai et al., 2020). Thus, the up-regulation of these peroxidases in roots upon H₂O₂ treatment is in line with the up-regulation of genes involved in cell wall metabolism observed in this study. Some components of the cell wall architecture, particularly the xyloglucans, have been shown to play an important role in imparting abiotic stress tolerance by coordinating with hormonal and other signaling cascades. For example, a xyloglucan galactosyl transferase from Arabidopsis, SHORT ROOT IN SALT MEDIUM 3 (RSA3), was shown to play a crucial role under salt stress by assembling actin microfilaments and thus preventing ROS accumulation induced by disruption of actin microfilaments (Cho et al., 2006; Li et al., 2013). Also the role of xyloglucan modifying enzymes along with expansins in loosening and expanding the cell wall network upon abiotic stresses has already been described (Tenhaken, 2015).

Overall, leaves and roots showed very unique transcriptional responses upon H₂O₂ treatment. Not only the number of DEGs was much higher in roots compared to leaves, the change in transcription also affected a quite different set of genes ( Figures 2 , 3 ). Nevertheless, there are DEGs that were found in both plant parts ( Figure 4 ). These 349 DEGs were further divided into four clusters, depending on their expression pattern. Looking at the two larger clusters, the commonly up- or down-regulated DEGs ( Figure 5 , Supplementary Table S3 and S4), certain patterns in the functional categories can be observed. Both clusters include TFs from different families. This is not unexpected and highlights their versatility in differentially regulating genes as an important part of all stress responses (Javed et al., 2020). However, of the TFs identified in this study, only few have previously been associated with oxidative stress, such as an Arabidopsis ortholog to HORVU2Hr1G066080 and HORVU3Hr1G016320, the LOB DOMAIN CONTAINING PROTEIN 41 (LBD41), that was previously identified in relation with low-oxygen endurance or high-light-induced increase in H₂O₂ (Mustroph et al., 2009; Vanderauwera et al., 2011). However, some were found associated with stresses, such as herbivory, that include ROS-mediated signaling or mutations that cause increased levels of ROS (Paudel et al., 2013; Garcia et al., 2016).

Several transporters were found commonly down-regulated ( Supplementary Table S4 and Figure 5A ). The aquaporin encoded by HORVU4Hr1G085250 is orthologous to the TONOPLAST INTRINSIC PROTEIN 4;1 (TIP4;1) of Arabidopsis and rice. Aquaporins not only transport water but also other molecules including H₂O₂. TIP4;1 from barley was shown to be up-regulated by ABA in roots and gibberellic acid in shoots (Ligaba et al., 2011). Moreover, its expression was also up-regulated upon drought (Kurowska et al., 2019). Also sugar transporters of the SWEET-type and PHT1.7 phosphate transporters have been demonstrated to play a role in abiotic stress tolerance and showed variable expression patterns under stress conditions (Cao et al., 2020; Gautam et al., 2022).

We also found common down-regulation of orthologs to RECEPTOR-LIKE PROTEIN KINASES(RLKs) from different subfamilies, i.e., WAK, LLR, CRK and RLCK ( Supplementary Table S4 and Figure 5A ). Experimental evidence suggests that RKLs are a vital part of the growth-defense trade-off, i.e. by facilitating the cross-talk between different phytohormones (Zhu et al., 2023). However, of the specific RLKs found commonly down-regulated in barley leaves and roots, only the pepper ortholog of WAKL20 was described in relation to stress (Zhu et al., 2023). DEGs connected to various facets of primary metabolism were found commonly up-regulated ( Supplementary Table S4 and Figure 5B ). While several of them are involved in pathways that play a role in stress responses, an obvious connection between these specific DEGs is lacking. Overall, even if no clear connection to oxidative stress exists, many of the commonly regulated DEGs have been described or postulated previously to be involved in stress tolerance mechanisms.

A very small number of DEGs was found counter-regulated upon treatment with H₂O₂ ( Supplementary Table S4 and Figure 6 ), the majority of which showing up-regulation in leaves and down-regulation in roots. Several of those genes are connected to aspects of metabolism and hormone signaling, and some orthologous genes of other plant species, such as SERAT1, OSM34, and UGT74D1 of tomato, grapevine and Arabidopsis have been previously connected to stress, ABA signaling, or auxin (Tavares et al., 2015; Jin et al., 2021; Park and Kim, 2021; Liu et al., 2022). Remarkably, this cluster also includes a group of nine HSPs, and this different expression in leaves and roots raises questions about their specific role in stress response in the different tissues.