An hvcbp20.ab mutant was isolated using the TILLING (Targeted Induced Local Lesions In Genomes) strategy (McCallum et al., 2000). TILLING was performed in the M₂ generation of the HorTILLUS population that was developed via the chemical mutagenesis of barley cultivar “Sebastian” in the Department of Genetics, the University of Silesia (Szarejko et al., 2017). The mutated population of ca. 9,600 M₂ plants were derived from a double treatment of seeds with two mutagens—sodium azide (NaN₃, 1.5 mM/3 h) and N-methyl-N-nitrosourea (MNU, 0.7 mM/3 h) with a 6-h inter-incubation germination period (iig) between the treatments.

Conserved regions of the HvCBP20 gene were determined using the ClustalOmega (http://www.ebi.ac.uk/Tools/msa/clustalo/) and CODDLE (Codons Optimized to Discover Deleterious Lesions; http://blocks.fhcrc.org/~proweb/input/) tools. Protein functional domains were mapped in a sequence of the HvCBP20 gene using the Pfam (http://pfam.sanger.ac.uk/) tool. Based on the results obtained, three fragments of the HvCBP20 gene were selected for TILLING (Figure 1). The first fragment encodes the part of the N-terminal domain that is responsible for RNA-binding and interaction with CBP80 (53–758 bp); the second fragment includes the region that encodes the part of the N-terminal domain that is responsible for the interaction with CBP80 (2,428–3,157 bp) and the third fragment encodes the C-terminal domain that is responsible for the nuclear import of the cap-binding complex (3,435–4,221 bp). The length of three HvCBP20 gene fragments that were analyzed using TILLING were 706, 730, and 787 bp long, respectively (in total 2,223 bp). DNA pools of eight individual M₂ plants were used as the templates for DNA amplification. Mutation detection was performed according to the protocol described by Szarejko et al. (2017).

The secondary structure was predicted using ENDScript (Robert and Gouet, 2014). The 3D structure of the barley protein (NCBI Acc. No. ACL83596.1) was predicted using LOMETS (http://zhanglab.ccmb.med.umich.edu/LOMETS; Wu and Zhang, 2007) and I-TASSER (http://zhanglab.ccmb.med.umich.edu/I-TASSER; Zhang, 2008; Roy et al., 2010; Yang et al., 2015) servers, which produced similar results. The final 3D structure model was predicted using an LOMETS server based on an H. sapiens subunit of CBP20 (1h2nX) as the template. Based on the maximum C-score, the most appropriated predicted structure was chosen for further analysis. The 3D structure was visualized and labeled using UCSF Chimera software (http://www.cgl.ucsf.edu/chimera; Pettersen et al., 2004) running on a Windows system.

In addition, the impact of the substitution that was identified in the hvcbp20.ab mutant was analyzed according to the protein affection using the SIFT tool (Kumar et al., 2009). SIFT is a sequence homology-based tool that sorts intolerant amino acid substitutions from tolerant ones. According to the SIFT criteria, the amino acid substitution is predicted to be damaging when the score is ≤ 0.05, and tolerated if the score is >0.05. The median that is used to measure the diversity of the sequences that are used for prediction ranges from 0 to 4.32 (ideally the number is between 2.75 and 3.5).

Thirty seeds of the barley cv. “Sebastian” and the hvcbp20.ab mutant were planted in a Petri dish plate (Φ = 90 mm) containing four layers of “Whatman” filters that had been soaked with 5 mL of sterile distilled water with or without 100 μM ABA. The plates were chilled at 4°C in the dark for 4 days (stratified) and moved to 22°C with a 16-h-light/8-h-dark cycle. On the second day after stratification (DAS), the Whatman paper filters were replaced with fresh ones. The seeds that were used for these experiments were harvested and stored at the same time. The percentage of seed germination was scored on the 1st, 2nd, 3rd, and 4th DAS. Germination was defined as the visible emergence of the radicle through the seed coat. The analyses were performed using a Stemi 2000-C stereoscopic microscope (Carl Zeiss) with an attached camera (Canon). In order to document the results, AxioVision LE (Carl Zeiss) software was used. The experiment was performed in three biological replicates, with 90 seeds per replicate (each replicate was performed using 3 plates, each containing 30 seeds).

The dissected plant tissues were homogenized in a sterile, ice-cold mortar containing 500 μl of an RLT buffer (RNeasy Plant Mini kit; Qiagen, Hilden, Germany). After homogenization, total RNA was extracted using an RNeasy Plant Mini kit according to the manufacturer's instructions. For the microarray analyses, RNA was additionally purified using precipitation in 1 M lithium chloride, and each RNA precipitate was then dissolved in 15 μl of nuclease-free H₂O. The yield and purity of the RNA was determined using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, USA). The integrity of the RNA was checked using an Agilent 2100 Bioanalyzer equipped with an RNA 6000 Nano chip (Agilent Technologies, Santa Clara, USA).

The functional analysis of HvCBP20 (Gene Bank Acc. No. FJ548567.1; NCBI) using TILLING was performed on DNA that was extracted from 5,376 plants of the M₂ HorTILLUS population. In total 11,950,848 bp were scanned and the estimated density of the mutations for HvCBP20 was 1/373 kb. Three fragments of the HvCBP20 gene were selected for analysis based on their highest conservation with the orthologous sequences and their engagement in the processes that are crucial for the CBP20 function (Figure 1). A substitution of nucleotide G to A was identified in the hvcbp20.ab allele within the gene fragment encoding N-terminal domain responsible for its interaction with the CBP80 subunit. This mutation caused the change of glycine to serine (G141S) in the amino acid sequence.

In order to predict whether an amino acid substitution in a protein affects its functionality, analysis using SIFT (Sort Intolerant from Tolerant; http://sift.jcvi.org/) was performed. According to the analysis, the G141S was predicted to affect the protein function with a score of 0.01. In regards to the SIFT criteria, this substitution is predicted to be damaging. The analysis of the secondary structure of the HvCBP20 that is based on the alignment of orthologs showed a significant difference between barley and human or animal CBP protein and its simultaneous similarity to Arabidopsis and other plants (Supplementary Material S1A). It is worth noting that CBP20 is highly conserved across the compared species when domains responsible for RNA binding-RND (RNA-binding domain) composed of one RNP2 and one RNP1 motif are considered. However, the main difference observed between plant CBP20 and its animal orthologs is the presence of a long C-terminal tail containing nuclear localization signals (Supplementary Material S1A). In Arabidopsis, NLS localized in the C-terminal tail of AtCBP20 is responsible for the import of the whole CBC complex into the nucleus, since the AtCBP80 does not contain NLS (Kierzkowski et al. (2009). Taking into account other highly conserved motifs within the C-terminal tail of plant CBP20 (depicted as green boxes in Supplementary Material S1A), apart from NLSs, it can be hypothesized that the function of this part of CBP20 may be important for activities other than RNA binding and interacting with CBP80 or activities that are specific to plants. The function of these conserved domains is not known yet and needs further investigation.

Then, we attempted to computationally model the HvCBP20 3D structure. The 3D modeling confirmed the C-terminal tail and showed that it protrudes from the CBC complex (Figure 3A). Based on the computational analysis, we could visualize the barley CBP20 protein and identify the amino acids that are crucial for RNA binding and interacting with CBP80 (Figure 3). Therefore, we could formulate the assumption that their function might be conserved in barley, Arabidopsis and human homologs. Our comparative analysis allowed us to pinpoint the amino acids that are responsible for binding the cap, interacting with CBP80 and folding the C-terminal domain in barley (Figure 3B). Among these amino acids, which are highly conserved across the species (Supplementary Material S1A), there is F43 (Figures 3C,E), which corresponds to F49 in human CBP20. This amino acid reorients upon cap binding in order to pack between the backbones of the human residues R146-G147 and G81-C82 (Calero et al., 2002; Mazza et al., 2002), which are respectively R140-G141 and G75-C76 in barley CBP20. In the hvcbp20.ab mutant, glycine in the 141 position is substituted by serine (G141S). The first observed structure change that was revealed by comparative modeling was the division of the groove that was visible in the wild-type into two parts in the hvcbp20.ab protein (Figures 3D,F). Another change resulting from the G141S mutation was an additional hydrogen bond between the mutated S141 and R140, absent in WT (Supplementary Material S1B). Taking into account the importance of packing the F43 between R140-G141 and G75-C76 for the proper folding of CBP20 protein in humans, it can be presumed that the changes that were observed in the mutated protein may lead to the imperfect folding and thus to its partial dysfunctionality. One can draw a scenario in which the additional hydrogen bond between R140 and S141 interferes with the proper packing of F43 between them and G81-C82, and further affects the proper folding of the CBP20. Taking into account the high conservation of the CBP20 function and the position of the substituted amino acid, the described changes in the hvcbp20.ab structure may be significant for CBP20 functioning. However, these assumptions are based only on computational predictions, and therefore they should be considered to be a hypothesis that needs further verification.

The hvcbp20.ab mutant displayed inhibited seed germination in the presence of 100 μM of ABA. The WT seeds that were with this ABA concentration germinated at a level of 60% of the control (Figure 4). These data support the conservation of the CBP20 function, since the elevated sensitivity to ABA during seed germination of hvcbp20.ab is in concordance with studies performed in Arabidopsis cbp20 and cbp80 mutants (Hugouvieux et al., 2001, 2002; Daszkowska-Golec et al., 2013). This clearly indicates the engagement of HvCBP20 in ABA signaling.

The relative water content (RWC) was measured in the second leaf of hvcbp20.ab and its WT parent after 10 days of exposure to severe water deficiency stress (25 DAS; Figure 2). There were no significant differences in the RWC between the mutant and the WT under control conditions. After 10 days of drought (25 DAS), the RWC decreased in both of the genotypes studied. However, the mutant was able to store 30% more water than its parent (Figure 5A). The phenotype of drought-stressed plants allowed to distinguish the mutant and the WT genotype. The mutant exhibited leaf rolling and leaves with higher turgid (Figure 5B; Supplementary Material S2A). Another physiological parameter that was highly responsive to drought stress was the electrolyte leakage. After 10 days of drought stress, the WT exhibited a 5.4-fold higher electrolyte leakage than in the control conditions, while the hvcbp20.ab did not show any significant differences between the values obtained under a normal water supply and drought (Figure 5C). Taken together, these results indicate an enhanced ability of the hvcbp20.ab to store more water and electrolytes within the leaves under water deficiency conditions.

Taking into account that the fastest physiological response of plants exposed to drought stress is stomatal closure (Sirichandra et al., 2009; Tuberosa, 2012), the stomatal conductance (g_s) during the drought stress was monitored in the hvcbp20.ab mutant and its parent cultivar “Sebastian.” Neither genotype differed when stomatal conductance was measured under non-stressed conditions (Figure 5D). At the onset of drought stress (15 DAS), the response of both genotypes was rapid; however, the stomatal closure in the hvcbp20.ab was significantly stronger than in the WT. The stomatal conductance in the WT reached the value of the mutant after 3 days of severe drought (18 DAS). No significant changes in g_s between the mutant and “Sebastian” were recorded on the 18 DAS or the 22 DAS. On the last day of drought (25 DAS), both genotypes were characterized by a low stomatal conductance but the mutant exhibited a significantly lower g_s than the WT (Figure 5D). It should be noted that the drought tolerance of the cbp80 and cbp20 mutants in Arabidopsis was associated with the closure of their stomata (Hugouvieux et al., 2002; Papp et al., 2004). All of these results suggest that the rapid stomatal closure preventing water loss during stress, that was observed in the drought-tolerant CBC mutants, is conserved across species and this universal mechanism is regulated by CBP20/CBP80.

A decrease in the RWC has been known to induce stomatal closure and thus a parallel decrease in photosynthetic efficiency (Cornic, 2000). To check whether differences between hvcbp20.ab and “Sebastian” regarding the state of photosynthetic processes occur under drought stress, the parameters of the JIP-test based on chlorophyll a fluorescence were analyzed (Strasser et al., 2004; Zlatev and Yordanov, 2004; Kalaji et al., 2014). The differences between the genotypes studied were revealed during the analysis of the fluorescence transients based on the calculation of the differences in the variable fluorescence curves (ΔV_t) (Figure 6). In the case of the WT, the curve recorded at the onset of drought stress (15 DAS; depicted as an open blue circles on the Figure 6) did not differ significantly from the control conditions, whereas after the exposure to prolonged drought (25 DAS; depicted as filled blue circles on the Figure 6), it showed a clearly defined ΔK-band (300 μs), which was related to the impairment of the oxygen-evolving complex (OEC; Guissé et al., 1995). Moreover, the presence of a high ΔJ-band (1–2 ms) and ΔI-band (10–30 ms) was observed in WT, which were presumed to be associated with an accumulation of QA- and the inactivation of ferredoxin-NADP⁺ oxidoreductase (FNR), respectively (Guissé et al., 1995; Schansker et al., 2005). The variable fluorescence curve that was obtained for the hvcbp20.ab mutant at the onset of drought (15 DAS; depicted as open yellow triangles on the Figure 6) was characterized by the presence of a low value of each described for WT bands. After 10 days of severe drought stress (25 DAS; depicted as filled yellow triangles on the Figure 6), all of the bands mentioned above were also noted in the mutant, but their values in the mutant were significantly lower for ΔK and ΔJ than the values recorded for the WT under the same conditions.

The OJIP transients were translated into biophysical parameters according to Strasser et al. (2004)—specific activities per reaction center (RC) and performance index (PI_ABS; Table 1). Changes in parameters of photosynthesis in the mutant were observed already at the onset of drought stress, while WT exhibited changed photosynthetic parameters only after prolonged drought (Table 1). Interestingly, most of the parameters measured in hvcbp20.ab mutant at the onset of drought stress were not significantly different from the WT-values recorded under prolonged drought. Moreover, the values of ABS/RC, TR₀/RC, ET₀/RC, and DI₀/RC were not significantly changed in the period between 15 DAS until the 25 DAS in the mutant, except for RC/CS₀ parameter (Table 1). At the onset of drought (15 DAS), the increased value of absorbed energy (ABS/RC) was observed in hvcbp20.ab. This sort of reaction can be related to either (i) the inactivation of a part of the RC pool or (ii) increased size of the antenna in response to stress conditions (Kalaji et al., 2014). The analysis of the number of active RCs per excited cross section (RC/CS₀) revealed that the cause of changes in ABS/RC in the mutant was a significant reduction of reaction centers at the onset of drought. The increase of trapping energy per active reaction center (TR₀/RC) according to literature (Kalaji et al., 2014) is associated with changes in ABS/RC and RC/CS. The same relation was observed at the onset of water deficiency in the mutant and after the prolonged drought in the WT. It is worth noting that this parameter (TR₀/RC) is tightly linked to the appearance of the ΔK band, since it is directly related to the impairment of the oxygen-evolving complex (OEC). Taking these results together, the photosynthetic machinery of the mutant adapted to the water deficit faster than the WT.

Our results indicated also that hvcbp20.ab was able to shift the excess of light energy into dissipation more effectively and much earlier that its WT parent. At the onset of drought, the hvcbp20.ab mutant had already achieved 120% of the control DI₀/RC-value and interestingly, it still displayed the same dissipated energy value after prolonged water deficiency (Table 1). Conversely, cv. “Sebastian” did not dissipate more energy at the onset of drought compared to the control conditions but after 10 days of drought stress, the DI₀/RC increased by 30% compared to the control value. These results together indicated a faster adjustment of the action of the photosynthetic apparatus in the hvcbp20.ab mutant.

An analysis of growth reduction under drought stress revealed that the mutant did not react as drastically to a water deficiency as its WT (Table 2). A significant reduction of hvcbp20.ab growth (93% of the height of the control plants grown under optimal conditions) was observed only on the last day of drought treatment (25 DAS), whereas a reduced growth rate of Sebastian (67% of the growth of the control) was observed already after 5 days of severe drought (22 DAS; Table 2). The prolonged drought stress did not lead to such a rapid reduction in growth rate in hvcbp20.ab as it did in WT. This may be related to the strong increase in the ABA level in the mutant at the onset of drought stress (about 100% more than the control conditions), whereas the highest peak of the ABA level in the WT occurred only after prolonged drought treatment (Supplementary Materials S3A,B). These results provide additional evidence of a delayed response to drought stress in the WT. Non-destructive imaging of plants during the drought experiment showed that the mutant produced a significantly larger (20% over the wild-type) projected shoot area than its wild-type at the beginning of drought stress (15 DAS; Supplementary Material S2B). However, in accordance with the expectations, the projected shoot area in both genotypes decreased during the drought treatment, and hvcbp20.ab exhibited a significantly larger projected shoot area (7% more than the wild type; Supplementary Materials S2B,C) at the end of the treatment. Taking into account that the traits that were observed in hvcbp20.ab such as leaf rolling and thus its later turgid loss were also reported to be associated with a later onset of the senescence process (Neilson et al., 2015), it can be concluded that the larger shoot area may be a common outcome of these events.

Analysis of the epidermal cell pattern in hvcbp20.ab revealed that the mutant differed from the WT parent under control conditions (Table 3; Figure 7). This morphological trait may be one of the phenotypic modifications that ensure the mutant's adaptation to drought. In non-stressed conditions, hvcbp20.ab exhibited 9% fewer stomata and 26% more trichome-like structures on the adaxial surface of the epidermis than the WT parent. It should be noted that the guard cells of mutant were smaller, more precisely shorter (80% of the WT) and wider (124% of the WT). These results are in accordance with previously reported cbp80 mutants of A. thaliana and S. tuberosum, which exhibited fewer stomata (respectively, 75 and 80% of the WT) and more trichomes (respectively, 140% and 130% of the WT) on the adaxial side of the leaves (Daszkowska-Golec et al., 2013; Pieczynski et al., 2013).

The abaxial surface of the hvcbp20.ab leaves had significantly more (10%) guard cells than the WT, and the guard cells were smaller, similar to those that were observed on the adaxial side (Table 3). Additionally, the abaxial surface of the mutant leaf was coated with 20% more trichome-like structures than the WT (Table 3; Figure 7). Our observations were similar to those of a study that was performed in a cbp20 mutant of A. thaliana in which more guard cells that were smaller in size together with more trichomes than in the wild-type were observed (Jäger et al., 2011). Taken together, the epidermal pattern of hvcbp20.ab under the control conditions indicates its potential adaptation to stress conditions—the lower number of stomata on the adaxial side, which is more exposed to adverse environmental conditions and the increased number of stomata on the abaxial side of the leaf together with an increased number of trichome-like structures. All of these traits help to balance the transpiration rate.

However, the most striking observation was the plasticity of the epidermis pattern in response to drought conditions in the hvcbp20.ab mutant. It has already been shown that changes in the number of guard cells are induced in response to changes in the environmental conditions (Yoo et al., 2010). The most common response is a reduction in the number of guard cells. The total number of stomata was also reduced on the adaxial side of the leaf in both genotypes studied in response to drought (Table 3). It is worth noting that after drought treatment, the length of the cells separating the guard cells was increased by 30% in the WT, whereas no change was observed in hvcbp20.ab (Table 3; Figure 7). Interestingly, the abaxial side of the leaf in the mutant after drought exposure was characterized by an increased number of stomata (132% of the WT) that were smaller. In addition, the cells separating the guard cells were shorter than in the wild type under the same conditions (60% of the wild type; Table 3; Figure 7). Additionally, the abaxial side of the epidermis in the mutant was coated with an increased number of trichomes (240% of the wild type; Table 3; Figure 7). Taken together, our observations allowed us to propose the model of the hvcbp20.ab leaf that describes changes in epidermal pattern in order to adapt to drought. We speculate that when the mutant leaf was rolled, it formed a humid environment and thus enabled minimal transpiration even under drought stress conditions. On the other hand, the abaxial side of the leaf in the mutant after drought exposure was enriched with increased number of smaller stomata, coated with an increased number of trichomes and also the cells separating the guard cells were shorter than in the wild type under the same conditions (Supplementary Material S4).

With the aim to examine the molecular mechanism of hvcbp20.ab response to water deficiency, the genome-wide transcriptome analysis was performed. We have already applied Agilent microarray with success for evaluation of transcriptome changes induced by drought in a root hairless mutant and its parent cultivar (Kwasniewski et al., 2016). With the aim of testing the reliability of the results that were obtained with the microarray, the expression of differentially expressed genes (DEGs) was validated using sensitive RT-qPCR (see Supplementary Material S5).

At the onset of drought stress the up-regulation of genes involved in lignin biosynthesis was observed. Interestingly, the expression of plant laccases, which correlates with the lignification process, is induced in response to biotic and abiotic stresses (Moura et al., 2010; Zhao et al., 2013; Eudes et al., 2014). In tomato and maize roots, the transcription of a laccase gene was induced by salt stress (Wei et al., 2000; Liang et al., 2006). Moreover, the overexpression of a putative laccase gene from rice, OsChI1, increased the tolerance of transgenic Arabidopsis to drought and salinity (Cho et al., 2014). Apart from laccases, peroxidases are another class of enzymes involved in the oxidation of lignin precursors (Zhao et al., 2013) that were upregulated specifically in hvcbp20.ab mutant. In monocots, highly lignified sclerenchyma cells participate in the process of rolling the leaf inwardly in response to drought stress that enhances the tolerance to water deficit stress (Kadioglu et al., 2012). The up-regulation of lignin-forming enzymes can partially explain the leaf rolling phenotype that was exhibited by hvcbp20.ab at the onset of the drought treatment. Leaf rolling in response to a water deficit protects plants from the hazardous effects of stress by decreasing water loss (Kadioglu et al., 2012; De Souza et al., 2013; Terzi et al., 2013). It can be assumed that the ability of hvcbc20.ab to store a higher content of water within the leaf after drought stress can also be an effect of leaf rolling that occurred in hvcbp20.ab much faster in response to a water deficit than in the WT. The first visible signs of leaf rolling were documented on 18 DAS (Supplementary Material S2A). Therefore, at the early stage of drought treatment, the mutant had already the reduced evaporation area, whereas in the WT the leaf rolling occurred after a couple of days of severe water deficit. It was reported that various tolerance mechanisms such as the antioxidant system and osmolyte accumulation were induced in the rolled leaves of various plant species under stress conditions (Kadioglu and Terzi, 2007).

Considering the involvement of HvCBP20 in abscisic acid signaling, it was intriguing to check whether changes in the transcriptome revealed a differential regulation of ABA synthesis. Accordingly, the up-regulation of the gene-encoding enzyme that is engaged in ABA synthesis, 9-cis-epoxycarotenoid dioxygenase (NCED, MLOC_43893), was observed. NCED performs reaction that is considered to be rate limiting in ABA biosynthesis (Tan et al., 2003). Arabidopsis and rice NCED3 expression is highly regulated by abiotic stresses, especially a water deficit (Qin and Zeevaart, 1999; Ye et al., 2011). Simultaneously, the upregulation of genes involved in GA/ABA homeostasis was observed: gibberellin 2-beta-dioxygenase (MLOC_38462) encoding the enzyme responsible for gibberellins deactivation (Richards et al., 2001) and B3 domain-containing protein LFL1 (MLOC_15725; ortholog of Arabidopsis AT3G26790; FUSCA 3). It has already been shown that the genes encoding the dioxygenases are the main sites of the regulation of the GA biosynthetic pathway by environmental signals such as drought stress (Thomas et al., 1999; Magome et al., 2008; O'Neill et al., 2010). FUS3 represses GA biosynthesis through its direct interaction with the promoter of the GA biosynthetic genes, whereas its transient activation during vegetative development increases the ABA levels Nambara et al., 2000; Curaba et al., 2004; Gazzarrini et al., 2004; Lu et al., 2010; Zawaski and Busov, 2014). Another gene involved in phytohormone pathway that was up-regulated specifically in hvcbp20.ab mutant was BAK1 encoding the leucine-rich protein kinase interacting with the brassinosteroid receptor (BRI1; Li et al., 2002; Nam and Li, 2002). Recently, it was shown that BAK1 is a positive regulator of stomatal closure under drought conditions through the formation of a complex with OST1 (Open Stomata 1). OST1 displays a dominant kinase activity during the response to drought stress when the ABA signal is relayed to the guard cells. Moreover, the complex formation is enhanced by the increased ABA level (Shang et al., 2015).

The up-regulation of genes involved in trichome differentiation at the onset of drought stress in the hvcbp20.ab mutant suggested a potential role of trichomes in the mutant adaptation to dehydration conditions. Interestingly, a significantly higher number of trichomes on both sides of the leaf surface was observed in the mutant compared to the WT in response to drought stress. It should be stressed that the increased number of trichomes was reported to enhance the drought tolerance (Huttunen et al., 2010; Hauser, 2014).

Down-regulated genes specific for hvcbp20.ab mutant at the onset of drought stress were categorized as being involved in carbohydrate metabolic process. The stomatal closure in response to water stress is maintained by ion transport but recent evidence also suggests the potential role of sugar metabolism in the regulation of guard cells (Lawson et al., 2008). One of the first steps during guard cell opening is the conversion of starch into disaccharide maltose (reviewed in Daszkowska-Golec and Szarejko, 2013). This reaction is catalyzed by β-amylases (BAMs; Weise et al., 2005). It is worth noting that down-regulation of genes encoding β-amylases was observed in hvcbp20.ab mutant. Moreover, the down-regulation of β-amylases transcripts was proved to enhanced drought tolerance through stomatal closure (Prasch et al., 2015). The genes encoding sucrose synthases (MLOC_11770 and MLOC_58487; Figure 10) downregulated in hvcbp20.ab mutant were also classified into GO category: carbohydrate metabolic process. The link between guard cell-specific sucrose hydrolysis and stomatal conductance was shown using transgenic potato plants with a silenced SUCROSE SYNTHASE 3 gene (Antunes et al., 2012). The potato sus3 mutants exhibited a reduced stomatal conductance and an increased water use efficiency that was related to a decrease in sucrolytic activity. Therefore, the down-regulation of above-mentioned genes can be linked to reduced stomatal conductance observed in hvcbp20.ab mutant.

Among the over-represented GO categories when up-regulated genes in mutant were considered, the most attention was attracted by the BP—protein folding since its p-value was the highest (P = 0.000231). Protein folding plays a pivotal role in the regulation of metabolic processes and stress responses. In plants, the key rate-limiting enzymes and misfolded/damaged proteins are regulated by different strategies and are strongly dependent on environmental cues. The correct folding and subsequent protein assembly is required to ensure the proper functionality of a protein. It was reported that the level of TCP-1/cpn60 chaperonin proteins, which are essential for the correct folding and assembly of polypeptides, increased in response to oxygen radicals (Gatenby, 1992). TCP-1/cpn60 chaperonins (MLOC_56130, MLOC_63773, MLOC_38451, MLOC_57369) were up-regulated in the hvcbp20.ab mutant. The correct folding and proper assembly of the TCP-1/cpn60 complex is related to the action of the HSP60 chaperone family (Prasad and Stewart, 1992). Interestingly, two barley genes MLOC_66031 and MLOC_18416 encoding the HSP60 proteins were up-regulated in hvcbp20.ab in response to prolonged drought stress. The up-regulation of chaperonin complexes specifically in hvcbp20.ab suggests the potential role of CBP20 as a molecular negative regulator of chaperones during drought stress, but this requires further studies.

Oxygen free radicals are responsible for most of the oxidative damage in biological systems and their deleterious effects on biological structures were described in detail (Asada, 1999; Johnson et al., 2003). Although plants evolved the machinery to cope with oxidation stress, under intense stressful conditions, a cascade of molecular events often results in cell death due to severely damaged target molecules. It should be noted that turning on the genes encoding the enzymes responsible for ROS detoxification at a high level in the hvcbp20.ab mutant [MLOC_46664 (ascorbate transporter LPE1), MLOC_68517 (Glutathione peroxidase) and MLOC_22278 (Superoxide dismutase)] but not in its WT supports our hypothesis on the better adaptation of hvcbp20.ab to drought conditions. These results clearly showed that the mutant developed the ability to protect itself from the deleterious effects of the decreased photosynthesis efficiency under drought. Interestingly, one of the mentioned genes, MLOC_22278, encodes the barley ortholog gene of Arabidopsis CSD1 (copper/zinc superoxide dismutase 1) that is under miR398 regulation (Sunkar et al., 2006). It was shown that miR398 under drought stress was down-regulated in Arabidopsis and thus CSD1 was up-regulated in order to perform its protective function (Sunkar et al., 2006). miR398 and its target sites on CSD1 and CSD2 mRNA are conserved in dicotyledonous and monocotyledonous plants (Bonnet et al., 2004; Jones-Rhoades and Bartel, 2004; Sunkar et al., 2005). The most striking observation was the fact that miR398 is one of miRNAs that is related to CBP20/CBP80 (CBC) action during miRNA biogenesis (Laubinger et al., 2008). It can be hypothesized that this process is conserved across species as our observations have indicated the indirect involvement of CBP20 in the mechanisms that are connected with ROS detoxification under drought stress.

Among genes that were down-regulated specifically in hvcbp20.ab after prolonged drought treatment was MLOC_43989 encoding SCARECROW-LIKE PROTEIN 27 (SCL27, AT4G00150). Recently, the link between SCL27, miR171 and chlorophyll a biosynthesis was reported in Arabidopsis (Ma et al., 2014). It was shown that SCLs are down-regulated by miR171 and that it is also able to repress the expression of the gene encoding protochlorophyllide oxidoreductase (POR; Ma et al., 2014). Interestingly, Laubinger et al. (2008) determined that the biogenesis of miR171 was under the control of CBC.

Another highly down-regulated gene in hvcbp20.ab mutant after 10 days of drought stress was the putative ortholog of Arabidopsis FAMA (AT3G24140), whose activity is required to promote the differentiation of the stomatal guard cells and to halt the proliferative divisions in their immediate precursors (Ohashi-Ito and Bergmann, 2006). It is worth noting that the function of FAMA is conserved across species including both di- and monocotyledons plants (Liu et al., 2009). Interestingly, a detailed analysis of the changes in MLOC_60291 expression revealed its higher level in control conditions compared to the WT (Supplementary Materials S7, S8) and at the onset of drought in hvcbp20.ab (Figure 10). Its expression decreased in response to prolonged drought stress in the mutant. These data provide a partial explanation of the epidermal pattern that was observed in hvcbp20.ab. The higher number of guard cells in the mutant compared to WT may be associated with the higher level of FAMA expression during development under an optimal water supply. On the other hand, after drought treatment, clusters of guard cells that were not spaced properly were detected and this could be interpreted as the effect of a strong inhibition of FAMA expression during a prolonged drought. Both types of regulation seemed to help hvcbp20.ab in its more efficient adaptation to the changed water status and enabled its proper response to stress.

There was also a group of DEGs that showed the same pattern of expression at the beginning and the end of drought treatment within the set of genes that might imply the phenotypic traits of hvcbp20.ab that enable a better adaptation to stress. It strongly support our hypothesis about the involvement of CBP20 in regulation of processes related to (i) lignification (MLOC_22174) and thus possibility of inward leaf rolling; (ii) growth inhibition via hormonal crosstalk (MLOC_38462, MLOC_15725); (iii) stomatal closure (MLOC_19405; MLOC_58487); (iv) changed epidermal pattern (MLOC_64670); (v) correct protein folding (MLOC_56130, MLOC_63773; Figures 10, 11).

The response to drought stress at the onset of the water deficit was much faster in the hvcbp20.ab mutant than in the WT. The mutant exhibited a better fitness to stress conditions by its much more efficient activation of stress-preventing mechanisms that further enabled its better performance under drought stress.

In summary, compared to the WT, the hvcbp20.ab displayed:

All of the above-mentioned traits allowed hvcbp20.ab to optimize growth in response to a water shortage and further to withstand a prolonged drought. At present it is not possible to point the direct interactors of CBP20 based on our study, and to define the exact role of cap binding proteins in drought stress response. Instead, we can rather propose the network hubs involved in the adjustment to the drought conditions of hvcbp20.ab mutant based on the integration of transcriptomic and physiological data (Figure 10).

It is worth noting that transcriptomic data from Arabidopsis cbp80 mutant where the significant changes in expression was observed only after ABA treatment (Kuhn et al., 2008) clearly support our results and raise questions about the exact role of CBP20/80 in ABA-related drought response. The most convincing mechanism of action is that expression of genes involved in ABA signaling and ABA-related drought response is regulated by the action of CBP20/80 only in the presence of stress. It is also becoming clear that the mechanism of CBP20 action is selective and does not result in severe phenotype change. Simultaneously, it should be noted that the observed pleiotropy in phenotype is probably due to the regulation of RNA-metabolism.

We believe that results presented here are a key in progress of understanding the involvement of CBP20 in drought stress in barley, we underline that there is still an open question about the exact way of CBP20 action. Further studies will allow to point the genes, that are under direct control of CBP20 regulation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.