HvAKT2 and HvHAK1 confer drought tolerance in barley through enhanced leaf mesophyll H+ homoeostasis

Summary Plant K+ uptake typically consists low—affinity mechanisms mediated by Shaker K+ channels (AKT/KAT/KC) and high‐affinity mechanisms regulated by HAK/KUP/KT transporters, which are extensively studied. However, the evolutionary and genetic roles of both K+ uptake mechanisms for drought tolerance are not fully explored in crops adapted to dryland agriculture. Here, we employed evolutionary bioinformatics, biotechnological and electrophysiological approaches to determine the role of two important K+ transporters HvAKT2 and HvHAK1 in drought tolerance in barley. HvAKT2 and HvHAK1 were cloned and functionally characterized using barley stripe mosaic virus‐induced gene silencing (BSMV‐VIGS) in drought‐tolerant wild barley XZ5 and agrobacterium‐mediated gene transfer in the barley cultivar Golden Promise. The hallmarks of the K+ selective filters of AKT2 and HAK1 are both found in homologues from strepotophyte algae, and they are evolutionarily conserved in strepotophyte algae and land plants. HvAKT2 and HvHAK1 are both localized to the plasma membrane and have high selectivity to K+ and Rb+ over other tested cations. Overexpression of HvAKT2 and HvHAK1 enhanced K+ uptake and H+ homoeostasis leading to drought tolerance in these transgenic lines. Moreover, HvAKT2‐ and HvHAK1‐overexpressing lines showed distinct response of K+, H+ and Ca2+ fluxes across plasma membrane and production of nitric oxide and hydrogen peroxide in leaves as compared to the wild type and silenced lines. High‐ and low‐affinity K+ uptake mechanisms and their coordination with H+ homoeostasis play essential roles in drought adaptation of wild barley. These findings can potentially facilitate future breeding programs for resilient cereal crops in a changing global climate.


Introduction
Severe drought can affect terrestrial ecosystems at regional to global scales, and the intensity and duration of drought significantly affect plant productivity and the health of ecosystems. Hence, understanding drought tolerance mechanisms in crops is crucial (Li et al., 2017;Selvaraj et al., 2017;Umezawa et al., 2006). Potassium (K + ) is the most abundant inorganic essential cation in plants and contributes up to 10% of their dry mass (Marschner, 2012). In plants, K + is involved in enzyme function, the maintenance of turgor pressure, leaf, and stomatal movement, and cell elongation, playing important roles under abiotic stresses (Dreyer and Uozumi, 2011). Drought tolerance in plants has been linked to K + homoeostasis (Mak et al., 2014;Shabala and Pottosin, 2014;Zhang et al., 2019).
During the early evolution of life, K + was utilized by cells as the major cation to maintain electroneutrality and osmotic equilibrium. Land plants have evolved from streptophyte algae and further evolution of biochemical processes, resulting in K + to be an absolutely necessary element (Rodriguez-Navarro and Rubio, 2006;Zhao et al., 2019). The emergence of the terrestrial plants in the Cambrian era and their evolution from bryophytes to flowering plants took place in limited K + conditions and periodical drought events (Heckman et al., 2001;Qiu and Palmer, 1999;Rodriguez-Navarro and Rubio, 2006). Despite the lack of K + , terrestrial plants not only kept their K + dependence but also developed new functions for efficient utilization of K + Rodriguez-Navarro and Rubio, 2006). Therefore, the terrestrial plants have evolved high-and low-affinity K + uptake systems with multiple K + transport families for K + uptake and translocation to various plant tissues to respond to changing environmental conditions (Ahn, 2004;Shabala and Cuin, 2008;Marschner, 2012;Cai et al., 2017;Zhao et al., 2019).
The classic high-and low-affinity K + uptake mechanisms in plants (Epstein et al., 1963) have been validated through functionally characterization of key transporters mainly in model plants (Dreyer and Uozumi, 2011;Rodriguez-Navarro and Rubio, 2006). The transport of K + through the plasma membrane into the plant and its allocation within the plant are mediated by K + transporters (HAK/KUP/KTs, HKTs), K + /H + antiporters (NHXs) and Shaker-type K + channels (AKTs, KATs, KC1, GORK, SKOR) (Grabov, 2007;Dreyer and Uozumi, 2011;Chen et al., 2016;Chen et al., 2017). The role of many K + transporters has been extensively studied (Ch erel et al., 2014;V ery and Sentenac, 2003). For example, the roles of K + channels in stomata operation and abscisic acid (ABA) responses (Chen et al., 2016;Hosy et al., 2003) and increased root K + uptake in osmotic stress (Fulgenzi et al., 2008;Shabala and Lew, 2002) were demonstrated, but this link has not been fully validated at molecular level using gene silencing and overexpression in a non-model plant. Also, the function of K + transport-related genes in barley such as HvAKT2 and HvHAK1 has not yet been fully elucidated and studies using gene-silencing and gene-overexpressing mutants in barley under drought stress need to be made.
The potassium channel AKT2, a component of low-affinity K + uptake system, plays important roles in long distance phloem loading and unloading of K + . It can operate as an inwardrectifying channel that allows H + -ATPase-energized K + uptake. AKT2 is also an important contributor, along with AKT1, to the mesophyll K + permeability (Boscari et al., 2009;Lacombe et al., 2000;Sklodowski et al., 2017;V ery and Sentenac, 2003). Arabidopsis AKT2 is strongly up-regulated by ABA, implicating its potential role in drought tolerance (Lacombe et al., 2000). When expressed in Xenopus laevis oocytes, HEK293 cells and COS cells, AKT2 forms a weakly voltage-dependent channel. However, patch-clamp studies on AKT2-transformed Arabidopsis, tobacco and poplar mesophyll cells displayed characteristics of a Ca 2+ -and pH-sensitive, K + inward rectifier with pronounced differences from those found in heterologous expression systems (Ivashikina et al., 2005;Langer et al., 2002;Latz et al., 2007). Moreover, a Ca 2+ sensor CBL4 modulates K + channel activity by promoting a kinase interaction-dependent translocation of the channel to the plasma membrane .
The plant high-affinity transporter HvHAK1 was first identified in barley and has sequence similarity with fungal HAK transporters. Subsequently, the HAK1s from Arabidopsis and rice have been isolated and characterized. Functional expression of HvHAK1, AtHAK5 and OsHAK in yeast mutants revealed that these transporters are responsible for the high-affinity K + uptake (Banuelos et al., 2002;Gierth et al., 2005;Santa-Maria et al., 1997). Direct electrophysiological evidence for a K + /H + symporter with a stoichiometry of 1K + /1H + was obtained using protoplasts from Arabidopsis root cells (Maathuis and Sanders, 1994), which could mediate K + influx at submillimolar external K + . Transcript expression of HAKs is enhanced by K + starvation, paralleling the onset of the high-affinity K + uptake (Armengaud et al., 2009;Chen et al., 2015).
In plants, drought phytohormone, ABA, and downstream nitric oxide (NO) and hydrogen peroxide (H 2 O 2 ) signalling molecules regulate many key physiological processes including drought stress response (Shi et al., 2014). In Arabidopsis, K + starvation impacts nitrate reductase activity, NO 3 À uptake and nitrate transporters (NRTs) (Armengaud et al., 2009) and disruption of nitrate reductases and NO production affects K + homoeostasis and stomatal regulation (Chen et al., 2016). High cellular H 2 O 2 can generate more hydroxyl radicals, resulting in K + loss and programmed cell death (Apel and Hirt, 2004). Moreover, our previous study revealed that the drought-tolerant wild barley XZ5 has highly efficient in K + uptake and translocation mechanisms, especially under drought (Feng et al., 2016;He et al., 2015;Wang et al., 2018). Therefore, we hypothesized that HvAKT2 and HvHAK1 improve barley performance under drought. In this study, we revealed the K + selectivity of AKT2 and HAK1 is originated from strepotophyte algae and AKT2 and HAK1 are evolutionarily conserved for K + uptake in land plants. We found that overexpressing HvAKT2 and HvHAK1 improves drought tolerance in barley through the regulation of leaf H + homoeostasis and cellular signalling.

Results
K + selectivity of AKT2 and HAK1 originated from strepotophyte algae is evolutionarily conserved in plants AKT2 and HAK1 form important parts of the plant low-and highaffinity K + uptake systems in plants. However, these genes have only been studied in limited number of angiosperm species. Here, we employed bioinformatics tools to explore the evolution of AKT2 and HAK1 in plant and algae (Figure 1a; Figure S1-S3; Table S1). Analysis using 1000 Plant Transcriptome database showed that the protein sequences of AKT2 and HAK1 are found in 80% and 75% of the 1,322 plant and algal species, respectively ( Figure S1; Table S1). Among the major clades, AKT2 had a significant presence in Rhodophyta occurring in 57% of the tested species while only 2 out of 28 species of Rhodophyta contained HAK1. AKT2 and HAK1 were identified in 69% and 39% of the 117 chlorophyte algae species, respectively. Interestingly, AKT2 and HAK1 have evolved in 40 and 35 out of 54 streptophyte algae, respectively (Table S1), indicating a possible origin of the two genes of the high-and low-affinity K + uptake systems. Moreover, the hallmark of the K + selective filter TxxTxGYGD in AKT2 and a putative K + selective filter component GxxYGD was both found in significantly more species of streptophyte algae than in chlorophyte algae ( Figure 1b; Table S1). Topology prediction, gene and protein evolution analysis, and evolutionary conservation analysis showed that HvAKT2 and HvHAK1 both have typical predicted K + transport characteristics (Figures 1c, S2 and S3).

HvAKT2 and HvHAK1 confer drought tolerance in barley
We conducted a comprehensive set of experiments to functionally analyse HvAKT2 and HvHAK1 for their potential roles in barley drought tolerance (Figures 2-5). Tissue-specific gene expression in XZ5 showed that HvAKT2 is mainly expressed in leaves, and HvHAK1 is highly expressed in both roots and leaves, with slightly higher expression in leaves ( Figure 2a). In situ PCR assay of HvAKT2 and HvHAK1 demonstrated that, in comparison with the positive (HvACT) and negative (No RT) controls (Figure 2b), the expression of HvAKT2 is predominantly within the stele, particularly in vascular bundles-phloem and xylem of roots, but the expression of HvHAK1 is in phloem and xylem cells, endodermis and epidermal cells of roots. In leaves, both HvAKT2 and HvHAK1 were mainly expressed in leaf mesophyll cells and vascular bundles particularly in phloem cells ( Figure 2b). These results led us to focus on drought tolerance mechanism of leaf mesophyll tissue, where both HvAKT2 and HvHAK1 are highly expressed. Moreover, genotypic difference of HvAKT2 and HvHAK1 expression was also found between wild and cultivated barley accessions. Compared to the controls, 5-d PEG treatment led to 4.8-and 2.6-fold up-regulation of HvAKT2 and HvHAK1 in the drought-tolerant wild barley genotype XZ5, whereas HvAKT2 and HvHAK1 were significant down-regulated in the drought-sensitive cultivar ZJU9 (Figure 3a).
We then conducted cloning and expression of the HvAKT2 and HvHAK1 in onion epidermal cells and Xenopus laevis oocytes to analyse their subcellular localization and electrophysiological properties, respectively (Figures 3 and S4). GFP fluorescence of both HvAKT2 and HvHAK1 was overlapped with the plasma membrane RFP marker, indicating that HvAKT2 and HvHAK1 are localized in the plasma membrane ( Figure 3b were detected from HvAKT2 and HvHAK1 expressed oocytes in a [K + ] dependent manner ( Figure S4). HvAKT2 and HvHAK1 showed high selectivity for K + , low selectivity to Rb + and no permeability to Na + , Cs + , Li + and Tris + (Figure 3c), validating the bioinformatics prediction (Figures 1, S1 and S3) that both HvAKT2 and HvHAK1 are highly K + selective.
Disruption of HvAKT2 and HvHAK1 resulted in significant reduction of drought tolerance in XZ5 (Figure 4). The BSMV-VIGS system was verified by inhibited expression of HvPDS by 95.3% in PDS-inoculated XZ5 plants ( Figure S5). Five days of PEG-induced drought resulted in significantly greater leaf wilting and growth inhibition in BSMV:HvAKT2-and BSMV:HvHAK1-inoculated plants compared to the mock-inoculated ones (Figure 4a). The expression of HvAKT2 and HvHAK1 was suppressed in leaves of silenced plants by 68% and 93% under drought,respectively (Figure 4b), which was accompanied by significant decreases of biomass at 53.9% and 53.8% in drought treatment, respectively ( Figure 4c). Drought treatment led to a significant increase of K + Overexpression of HvAKT2 and HvHAK1 caused a significant increase in drought tolerance in barley cultivar (wild type) Golden Promise ( Figure 5). The gene expression in the transgenic lines HvAKT2-OXs and HvHAK1-OXs was validated with significant upregulation in four independent lines, and HvAKT2-OX2 and HvHAK1-OX2 were selected for further experiments ( Figure S6). Drought-induced large growth inhibition in the wild type was significantly mitigated in the HvAKT2-OX and HvHAK1-OX plants ( Figure 5a). PEG-induced drought treatment slightly increased the transcripts of already overexpressed of HvAKT2 and HvHAK1, which were significantly decreased in Golden Promise ( Figure 5b). In contrast to the control, drought decreased biomass of Golden Promise by 44.7%, but the biomass reduction was 23.1% and 22.8% in HvAKT2-OX and HvHAK1-OX plants, respectively ( Figure 5c). Strikingly, gene overexpression led to a significant increase of leaf K + concentration by 15.5% and 16.4% in the leaves of HvAKT2-OX and HvHAK1-OX plants subjected to drought, respectively, while it was decreased by 13.5% in Golden Promise (Figure 5d).

Silencing and overexpressing of HvAKT2 and HvHAK1 modulate HvHA1 expression, H + -ATPase activity and H + flux in leaves
Five putative HvHAs genes including HvHA1 encode the H + -ATPases in barley in the wild barley genome (Dai et al., 2018). Surprisingly, we found that, in response to drought, HvHA1 expression in leaves was significantly down-regulated by 57.4% and 32.9% in the BSMV:HvAKT2-and BSMV:HvHAK1-inoculated plants, respectively ( Figure 6a). On the contrary, drought significantly up-regulated HvHA1 by 40.8% and 45.2% in HvAKT2-OX and HvHAK1-OX plants, respectively, without affecting the HvHA1 transcripts in leaves of Golden Promise ( Figure 6b). Consistently, in response to drought, H + -ATPase activity was significantly decreased by 31.2% and 32.6% in leaves of BSMV: HvAKT2-and BSMV:HvHAK1-inoculated plants, respectively (Figure 6c). H + -ATPase activity was increased in leaves of HvAKT2-OX, and HvHAK1-OX plants by 34.7% and 49.5% in response to drought, while there was a little change in Golden Promise ( Figure 6d). Similar results were found in roots of these transgenic lines ( Figure S8). Moreover, drought (1-h in 20% PEG) led to a reverse of H + fluxes from efflux (net H + release) to influx (net H + uptake) in leaf mesophyll of drought-sensitive BSMV:HvAKT2- In leaf mesophyll of plants overexpressing HvAKT2 and HvHAK1, PEG-induced drought stress resulted in 2.6-and 1.8fold higher K + influx in HvAKT2-OX and HvHAK1-OX plants than that of Golden Promise (Figure 7a). The K + influx decreased after 24 h of PEG treatment among all plants, but remained significantly higher K + uptake (70-80 nmol m À2 s À1 ) in the overexpressing lines than that of the control (Figure 7a  Holding potential was À20 mV, and voltage was clamped from À160 to 0 mV for 10 cycles. Data are mean AE SE (10-15 oocytes from at least three independent experiments). NO and H 2 O 2 are two key secondary messengers for drought response in plant cells (Chen et al., 2016;Zhao et al., 2019). Addition of the NO donor sodium nitroprusside (SNP) and H 2 O 2 scavenger dimethyl thiourea (DMTU) significantly alleviated drought symptoms in HvAKT2 and HvHAK1 silencing plants (Figure 8a,b). Drought-induced biomass reduction in plants with silenced HvAKT2 and HvHAK1 was 67.7% and 60% under PEG treatment, but the biomass decrease was significantly less at 40.6% and 41.2% with SNP and 40.7% and 30.8% with DMTU, respectively (Figure 8d,e). In contrast, exogenous NO scavenger c-PTIO and H 2 O 2 in addition to the PEG treatment decreased drought tolerance of XZ5 (Figure 8c), where significantly larger drought-induced decrease of biomass in XZ5 was 60.1% and 50% after addition of c-PTIO and H 2 O 2 to PEG treatment, respectively (Figure 8f).
Leaf NO content decreased significantly in BSMV:HvAKT2-and BSMV:HvHAK1-inoculated plants in response to PEG treatment, in contrast to a significant 22.3% increase of NO content in leaves of mock-inoculated plants (Figure 7d). Drought significant significantly increased NO content in the leaves of HvAKT2-OX, and HvHAK1-OX transgenic plants as compared to that of Golden Promise (Figure 8h). Moreover, drought-induced increase of leaf H 2 O 2 contents was significantly larger in HvAKT2-and HvHAK1silenced lines compared to the mock plants (Figure 8i). In contrast, drought treatment resulted in a highly significant increase in H 2 O 2 content in Golden Promise, but a smaller increase was observed in the HvAKT2-OX and HvHAK1-OX lines (Figure 8j).

Discussion
AKT2 and HAK1 are evolutionarily conserved for potassium uptake and stress response in plants Land plants have evolved to thrive in the terrestrial environment, where K + availability is dependent on the soil type, rainfall and many environmental factors. High-and low-affinity K + uptake systems have enabled plants to adapt to the low K + land lifestyle (Cao et al., 1995;Rodriguez-Navarro and Rubio, 2006;Wang and Wu, 2013). The evolution of high-and low-affinity K + uptake mechanisms was a prerequisite for the colonization of land by plants (Grabov, 2007). AKT2s and HAK1s have been found in evolutionarily diverse organisms ranging from green algae to angiosperms (Figure 1; Figure S1-S3; Table S1). The presence of AKT2s and HAK1s in plant genomes  implies that they play an important role in K + acquisition, allowing plants to survive in potassium-poor environments. Importantly, our data showed that the presence and structure of AKT2s and HAK1s are highly conserved in green plants and are likely to originate from streptophyte algae (Figure 1; Figure S1-S3; Table S1). Thus, it might be expected that high-affinity and low-affinity K + uptake systems that land plants possess for acclimation and adaptation to the variable and harsh terrestrial environments existed in their ancestors.
Plant inward-rectifying K + channels harbour regulatory domains comprising a putative cyclic nucleotide-binding site, a KHA region and an ankyrin domain (Jan and Jan, 1997;V ery and Sentenac, 2003). A unique fingerprint of K + channels is the highly conserved selectivity filter motif TxxTxGYGD, allowing the highly selective passage of K + ions across the membrane (Doyle et al., 1998;Jegla et al., 2018;Riedelsberger et al., 2015). Mutations in the selectivity filter can fundamentally alter the permeation properties of the channels (Dreyer and Uozumi, 2011). The KAT1-like channel from melon with unusual K + ⁄Na + permeation/blocking properties (Zhang et al., 2011) indicates that the channel pores have evolved to balance selectivity over a wide spectrum of potential competing ions. Given the evolutionary importance of plant K + channels, changes in ion selectivity may affect the evolution of K + transporters for plant adaptation to land. In our study, we found that the full TxxTxGYGD motif of AKT2s appeared in streptophyte algae but not in chlorophyte algae (Figure 1; Figure S1-S3; Table S1). Although there are many extant land-dwelling chlorophyte algae, a group of freshwater streptophyte algae represents the lineage that is sister to embryophytes. Environmental stress tolerance features of streptophyte algae may have facilitated their transition to land (Becker and Marin, 2009; Zhao et al., . Meanwhile, the current study demonstrated that heterologous expression of HvAKT2 and HvHAK1 in Xenopus laevis oocytes shows higher selectivity to K + over other cations (Figures  3d and S3). These data indicated that AKT2 and HAK1 have become conserved for potassium uptake and stress response in plants.
Overexpressing HvAKT2/HvHAK1 enhances H + homoeostasis for efficient K + uptake under drought Plant K + uptake is mediated by low-and high-affinity transport systems taking advantage of the electrical gradient and the proton motive force established by H + -ATPase (Dreyer and Uozumi, 2011;Epstein et al., 1963;Palmgren, 2001;Shabala et al., 2016;Wang et al., 2014). However, little evidence for the interaction of K + transport and H + pumping for drought tolerance has been obtained using transgenic plants. Here, we present solid experimental results that the expression of HvHA1, H + -ATPase activity and H + efflux is highly induced in the barley lines overexpressing HvAKT2 and HvHAK1 ( Figure 6).
AKT2, which encodes a K + channel subunit, shows gating similar to AKT1/KAT1 channels (Mode 1) and gating of little voltage sensitivity (Mode 2) (Dreyer et al., 2001;Xicluna et al., 2007). Switching between the two gating modes is under posttranslational control (Dreyer and Uozumi, 2011;Michard et al., 2005). Interestingly, shifting AKT2 from Mode 1 to the voltageindependent Mode 2 efficiently assists the plasma membrane H + -ATPase in energizing transmembrane transport processes . Here, we showed that overexpression of HvAKT2 in barley leaves increases HvHA1 expression, enzyme activity of H + -ATPase and high H + efflux (Figures 6 and  7) in response to drought, leading to drought tolerance ( Figure 5). It is also likely that increasing the expression of HvAKT2 provides more chance for Ca 2+ sensing and phosphorylation in barley plants, especially under stress conditions, similar to other types of ion channels Shabala et al., 2019). Indeed, an Arabidopsis phosphatase of type 2C interacts with AKT2, which links the AKT2 phosphorylation status to the ABA signalling pathway and drought (Ch erel et al., 2002), and an Arabidopsis plasma membranelocalized receptor-like kinase, MRH1, post-translationally regulates AKT2 (Sklodowski et al., 2017). Plant ion channels are usually low abundant proteins due to their high transport capacity. Therefore, plants only need to express between 10 and 1,000 ion channels as compared to 10 5 -10 6 pumps per cell to control ion homoeostasis (Hills et al., 2012;Latz et al., 2007;Shabala et al., 2019). Channels and pumps often colocalize in a cell membrane, indicating potential interactions. For instance, the Arabidopsis slow anion channel, SLAC1, interacts with most of the Shaker channels present in guard cells, including AKT2 . Therefore, a physical interaction between HvAKT2 and HvHAs may occur given the overexpression of HvAKT2 and huge number of the HvHAs at plasma membrane in transgenic barley lines. This is the first report showing this potential interaction of AKT2 with HvHA1 to actively regulate plant drought tolerance. However, this physical interaction requires future investigation.

HvAKT2 and HvHAK1 modulate ion homoeostasis and signalling for drought tolerance in barley
An important response of plants to drought stress is the uptake of K + (Shabala and Pottosin, 2014). Inward-rectifying K + channels such as AKT1 and AKT2 in plants provide major pathways for low-affinity K + uptake (V ery and Sentenac, 2003). In Arabidopsis, AKT2 is up-regulated by salinity and ABA, suggesting that AKT2 is involved in the recirculation of K + through the phloem (Shabala and Cuin, 2008) and in drought tolerance (Lacombe et al., 2000). Also, AKT2 currents are reduced in the presence of the phosphatase AtPP2CA (Ch erel et al., 2002) and enhanced by CBL4/SOS3 and CIPK6 (Held et al., 2011). Moreover, High-affinity K + uptake is coupled with the transport of H + through the H + /K + symporters (Banuelos et al., 2002;Rodriguez-Navarro and Rubio, 2006), but the molecular regulation of HAK1s in plant drought tolerance is less studied. Here, we show that drought stress causes a high K + influx in plants overexpressing HvAKT2 and HvHAK1, but it was significantly reduced in the gene-silenced plants (Figure 7), which may be due to the activation of low-and high-affinity K + uptake systems.
Plant apoplastic pH is sensitive to soil moisture and drought stress, and the alteration of pH could be a chemical signal that affects ion transport (Bacon et al., 1998). Low apoplastic pH dramatically down-regulates AKT2 activity, and even small physiological cytosolic pH variations substantially affect the rectification properties of AKT2 (Lacombe et al., 2000). Here, we demonstrated a dramatic H + influx in leaf mesophyll of plants silencing HvAKT2 and HvHAK1 under drought stress, while control plants displayed less H + influx (Figure 7b,e). It suggests that leaf apoplastic alkalization caused by reduced H + efflux may be a key reason for drought sensitivity in the HvAKT2-and HvHAK1-silenced plants. However, further study is necessary to demonstrate whether this apoplastic alkalization is accompanied by membrane depolarization. NO is a key cellular signal tightly associated with K + nutrition (Armengaud et al., 2009;Chen et al., 2016), and NO deactivates guard cell inward K + currents through a process that involves Ca 2+ signalling (Garcia-Mata et al., 2003). Here, NO content decreased in plants silencing HvAKT2 and HvHAK1 and increased in plants overexpressing HvAKT2 and HvHAK1 under drought stress (Figure 8). Hydrogen peroxide (H 2 O 2 ) is an important secondary messenger for plant stress signalling and regulation of K + homoeostasis (Apel and Hirt, 2004;Wang et al., 2017). H 2 O 2 is an early response of K + deficiency and externally added H 2 O 2 was sufficient for the induction of the high-affinity K + uptake (Shin and Schachtman, 2004;Wang et al., 2017). Although there was no report on the correlation between AKT2 and HAK1 regulated K + homoeostasis and H 2 O 2 accumulation in plant drought tolerance, positive relationships between H 2 O 2 production and K + accumulation under salt stress were found for Arabidopsis, cucumber (Cucumis sativus) and pumpkin (Cucurbita moschata) (Huang et al., 2019;Ma et al., 2012;Redwan et al., 2016). For instance, it was demonstrated that higher salt tolerance in pumpkin is related to its higher K + uptake and H 2 O 2 accumulation in the root apex. This was associated with salt-induced higher expression of key differentially expressed genes (DEGs) such as respiratory burst oxidase homolog D (RBOHD), 14-3-3 protein (GRF12), plasma membrane H + -ATPase (AHA1) and HAK5 in pumpkin than those in cucumber. Knocking out of RBOHD in pumpkin via genome editing led to a saltsensitive phenotype: lower root apex H 2 O 2 and K + content and GRF12, AHA1 and HAK5 expression. Therefore, salt tolerance in pumpkin is regulated by RBOHD-dependent transcriptional and post-translational activation of AHAs operating upstream of HAK5 (Huang et al., 2019). In this study, H 2 O 2 content increased significantly in plants silencing HvAKT2 and HvHAK1, but significantly reduced in overexpression plants under drought stress, which were also linked to altered K + and H + homoeostasis and drought tolerance in barley (Figures 4-8). In summary, these results indicated that NO and H 2 O 2 may regulate low-and highaffinity K + uptake systems through enhanced H + homoeostasis in HvAKT2 and HvHAK1 overexpressing barley lines to enhance drought tolerance.

Conclusions
In this study, significant correlations between the key traits ( accumulation, H + flux and H + -ATPase activity, gene expression, and NO and H 2 O 2 signalling are key mechanisms for drought tolerance of barley overexpressing HvAKT2 and HvHAK1. Therefore, we propose that HvAKT2 and HvHAK1 contribute to drought stress tolerance in barley, implying that both high and low-affinity K + uptake mechanisms play essential roles in plant drought tolerance. Future research should focus on the manipulation of both K + uptake systems using genome editing technology to not only improve the nutrient use efficiency for sustainable agriculture but also enhance drought tolerance of crops to better adapt in a changing climate.

Plant materials
A range of barley lines were used in this study including the drought-tolerant Tibetan wild barley genotype XZ5 (Hordeum vulgare L. ssp. spontaneum), the drought-tolerant barley (Hordeum vulgare L.) cultivar Tadmor, the drought-sensitive cultivar ZJU9, the wild-type Golden promise for barley transformation, and barley stripe mosaic virus-induced gene silencing (BSMV-VIGS) and overexpression lines.

Evolutionary bioinformatics
Evolutionary bioinformatics were conducted as described in Zhao et al., (2019

Quantitative real-time PCR
Quantitative real-time PCR (qRT-PCR) assays were conducted as described by Chen et al. (2016). Seven-day-old uniform seedlings were exposed to 1, 3 and 5 days of 20% PEG, and leaves were sampled from 2-leaf-stage plants after drought treatment. Total RNA was isolated with the TaKaRa MiniBEST Plant RNA Extraction Kit, and first-strand cDNA synthesis was carried out using the PrimeScipt TM RT reagent Kit (Takara, Otsu, Japan). qRT-PCR was performed on a CFX96 system machine using the SYBR Green Supermix (Bio-Rad, Hercules, CA). The fold changes were expressed as 2 ÀDDCt relative to the control. HvACT was used as a reference gene. Three biological and two technical replicates were conducted. Primers are listed in Table S2.

In situ PCR
The in situ PCR analysis of HvAKT2 and HvHAK1 was performed as described by Ye et al. (2019) and Zhao et al. (2019). Briefly, barley leaves and roots were immersed in ice-cold buffer. The samples were then embedded in 5% (w/v) agarose and sectioned to 60 lm (leaves) or 50 lm (roots). Ten sections were collected in a 200 µL tube to perform the DNase treatment, and RT-qPCR was performed. Samples were then washed, incubated in alkaline phosphatase-conjugated anti-digoxigenin Fab fragments and developed in the dark with BM Purple AP substrate (Roche, Penzberg, Germany). The images were then taken with a Leica microscope. The positive control was carried out with HvACT, and no RT was used as a negative control. Primers are listed in Table S3.

Molecular biology
Subcellular localization of expressed proteins was as described in Chen et al. (2011). The coding regions of HvAKT2 and HvHAK1 were amplified and cloned into pCAMBIA1300 with a CaMV 35S promoter:green fluorescent protein (35S:GFP) cassette to create fusion proteins. The final 35S:HvAKT2-GFP and 35S:HvHAK1-GFP fusion constructs were transiently expressed in onion epidermal cells using microprojectile bombardment. After 13-18 h, GFP was imaged after cell plasmolysis in 30% sucrose solution using confocal microscopy (Carl Zeiss Meditec AG, Jena, Germany). A plasma membrane RFP marker (pm-rb CD3-1008) was also used. Barley stripe mosaic virus (BSMV) virus-induced gene silencing (VIGS) was used according to He et al. (2015). The cDNA fragments of HvPDS, HvAKT2 and HvHAK1 were amplified using primers containing NheI sites and inserted reversely into the RNAc of BSMV to create cDNA clones of BSMV:HvPDS, BSMV:HvAKT2 and BSMV:HvHAK1 (primers are in Table S4). Sterilized seeds of XZ5 were germinated in an incubator (22/18°C, day/night) for 7 days, and then, uniform seedlings were transplanted to 1-L containers filled with a modified Hoagland's solution ( 8.42. After inoculation, the seedlings were sprayed with diethyl pyrocarbonate (DEPC)-treated water and covered with a transparent plastic to keep the high humidity for 3 days. HvAKT2-and HvHAK1-silenced seedlings were subjected to mock-inoculation with BSMV:c, BSMV:c+20% PEG, BSMV:gene-inoculated, and BSMV:gene+20% PEG treatments for 5 days. Six replicates were conducted.
Barley transformation was described in Bartlett et al. (2008). Open reading frames of HvAKT2 and HvHAK1 were PCR amplified, cloned into pDONR-Zeo (Invitrogen). Using Gateway technology (Invitrogen,Carlsbad, CA), the genes in the pDONR-Zeo vector were mobilized into the binary vector pBract214, which contains the Ubiquitin promoter. Subsequently, pBract214 and pSoup were introduced into the Agrobacterium strain AGL1 and infect embryo callus of Golden Promise. Regenerated seedlings were obtained 12 weeks after transformation. Evaluation of HvAKT2-OX and HvHAK1-OX transgenic barley lines was determined by real-time PCR. Sterilized T 3 seeds of the transgenic lines were germinated in an incubator (22/18°C, day/night) for 7 days, and then, uniform seedlings were transplanted to 1-L containers filled with half-strength Hoagland's solution. The transgenic plants were grown in BNS for 10 days, and then, plants were subjected to BNS containing 20% PEG to examine the phenotype.
Net K + , H + and Ca 2+ fluxes were measured from leaf mesophyll cells using the noninvasive ion-selective microelectrode (MIFE) technique (Feng et al., 2016;Newman, 2001). For transient ion flux measurements, excised leaf and root segments were promptly immersed in a chamber for 2 h before measurements (Zeng et al., 2014). Transient ion fluxes were measured for 10 min to ensure a steady flux, and then, 20% PEG were added for another 30 min. For steady-state ion flux measurements, excised leaf and root segments were pretreated with 20% PEG for 0, 1, 12 and 24 h. Net ion fluxes were measured for 10-15 min. Net ion fluxes were measured from at least eight biological replicates.

K measurement
The barley leaves and roots from hydroponic experiments were harvested and rinsed with deionized water. The samples were then dried at 80°C for 72 h to a constant weight and were digested with HNO 3 (Wu et al., 2014). The levels of K were measured with an inductive-coupled plasma (ICP)-emission spectrophotometer (Optima 2100DV; PerkinElmer, Wellesley, MA).

H + -ATPase activity
H + -ATPase activity was measured essentially as described by Regenberg et al. (1995) using an enzyme assay (Jiancheng Bio Co., Nanjing, China). The assay medium included 3 mM ATP and 0.02% Brij-58. The barley leaf and root samples were preincubated for 10 min with 0.02% Brij-58. The reaction was initiated by the addition of 2 mg of samples to the assay medium.

Measurement of NO and H 2 O 2
Fresh leaves (0.3 g) were homogenized in 8 mL of 50 mM phosphate-buffered saline (PBS) at pH 7.8 using a pre-chilled mortar and pestle. Then, the homogenates were centrifuged at 10 000 g for 15 min and the supernatants were used for NO and H 2 O 2 measurement. The NO content was measured with the nitrate reductase method using an NO assay kit, and the H 2 O 2 content was determined with a H 2 O 2 assay kit (Jiancheng Bio Co., Nanjing, China). For the experiments with the treatments of NO and H 2 O 2 , 200 µM NO scavenger 2-(4-Carboxyphenyl)-4,4,5,5tetramethylimidazoline-1-oxyl-3-oxide (c-PTIO), 200 µM NO donor SNP, 5 mM H 2 O 2 scavenger N,N-dimethylthiourea (DMTU) and 5 mM H 2 O 2 were dissolved in distilled water and added to the nutrient solutions. Then, seedlings were subjected to 20% PEG treatments for 5 days in the hydroponic experiments.

Statistical analysis
Statistical analyses were performed with a Processing System statistical software package using ANOVA followed by the Duncan's multiple range test.

Figure S1
Phylogenetic analysis of AKT2 and HAK1 in different plant and algal species selected from the 1KP database. Figure S2 Protein evolutionary analysis of AKT2 and HAK1 using PIECE with an inbuilt GLOOME and an inbuilt Exalign function in plant and algal species. Figure S3 Functional domain of HvAKT2 and HvHAK1. Figure S4 Electrophysiology of HvAKT2 and HvHAK1 in Xenopus laevis oocytes. Figure S5 BSMV-VIGS used in wild barley XZ5. Figure S6 Analysis of the expression of HvAKT2 and HvHAK1 in transgenic overexpression lines using qRT-PCR. Figure S7 The K + concentration in roots of silenced and overexpression lines. Figure S8 The activity of H + -ATPase in roots of silenced and overexpression lines. Figure S9 Transient and steady-state changes in K + , Ca 2+ and H + fluxes from root epidermal cells of inoculated plants via BSMV-VIGS subjected to 20% PEG treatments.