The generation and basic molecular properties of OsHAK1 transgenic lines (OsHAK1p:GUS, OsHAK1p:OsHAK1[OsHAK1-Ox], and oshak1 homozygous mutants in cv. Dongjin and cv. Manan genetic backgrounds) were previous described in Chen et al. (2015b).

For hydroponic experiments, seed sterilization and basal nutrient solution composition for seedling growth were described previously (Li et al., 2006). Same size 1-week-old rice seedlings were selected and transferred to IRRI nutrient solution (Chen et al., 2015b). The hydroponic experiments were carried out in a growth room with a 16 h light (30°C)/8 h dark (22°C) photoperiod and 70% relative humidity. In all treatments, nutrient solutions were replaced every 2 days. For drought stress experiments, rice seedlings were grown with the normal IRRI solution for 2 weeks and then transferred to nutrient solution supplemented with 15% (w/v) PEG6000 (mimics drought stress) for 1 week. At harvest, roots of rice plants were washed with 0.1 mM CaSO₄ for 5 min. Roots and shoots were separated before recording their biomass and K concentrations were determined as described previously (Ding et al., 2006; Chen et al., 2015b).

Experiments with soil-grown plants were performed in a greenhouse. For assessing drought tolerance at seedling and reproductive stage, 5-week-old rice seedlings were grown in pots filled with 10 kg of air-dried loam soil until tillering or booting stage. Rice plants at tillering stage were divided into two groups for 3 weeks, while those at booting stage were divided into two groups until harvest: group 1 with full watering treatment as control and group 2 watered to approximately 40% of field capacity (Ahmad et al., 2016b). Each treatment included five biological replications.

Entire root and shoot tissues from WT and transgenic lines after control or PEG treatment under hydroponic condition and the first two leaves from plants used in the soil drought experiment were used for isolation of total RNA. qRT-PCR was performed according to the protocol described previously (Chen et al., 2015a). The Ubq gene was used as internal control to normalize all data and expression levels were calculated by using the 2^-ΔΔC_T relative quantification method (Li et al., 2014). Primers used for qRT-PCR are listed in Supplementary Table S1.

Histochemical GUS assay in different tissues from plants subjected to PEG treatment was carried out as described previously (Ai et al., 2009). Quantification of GUS activity was performed as described by Chen et al. (2015a). Intensity of 4-methylumbelliferone fluorescence was measured using a multi-mode microplate reader (SpectraMax M5, Molecular Devices, United States). Protein concentrations were determined using the Coomassie blue G-250 colorimetric assay.

The adventitious roots were scanned to record differences in the elongation of roots among the plant genotypes and treatments. The analysis was carried out as described previously (Song et al., 2011) using the WinRhizoV4.0b (Regent Instrument, Canada) root analysis system. A ruler was used to measure the length of adventitious roots. Five individual plants of each line were measured.

Relative water content (RWC) was determined according to the method described by Zhao et al. (2014). Leaves were detached and weighted to obtain the fresh weight (FW) at the end of the stress period under soil treatment. Leaves were then soaked in de-ionized water for 4 h to obtain the saturated weight (SW). Subsequently, leaves were dried at 80°C for 48 h to determine dry weight (DW). RWC were calculated according to the formula: RWC = (FW - DW)/(SW - DW) × 100%.

Analysis of electrolyte leakage was carried out following the protocol of Guo et al. (2016). Fresh leaf samples from the soil treatment were harvested and washed with deionized water. Each sample was immediately placed into a beaker containing 30 mL of deionized water. The beaker was incubated at 25°C, shaken at 120 rpm for 3 h, and then electrical conductivity (EC1) of the solution was measured with a conductivity meter (Hanna, Italy). Next, samples with the immersion solution were boiled for 20 min and the conductivity measured after cooling it to room temperature (EC2). Relative electrolyte leakage (REL) was defined as REL = EC1/EC2 × 100%.

Photosynthetic CO₂ fixation rates were measured in rice seedlings between 9.00 and 11.00 am using a Li-COR6400 portable photosynthesis system equipped with a LED leaf cuvette (Li-COR, Lincoln, NE, United States), essentially as described in Li et al. (2016). At least five individual WT and transgenic lines in each stress treatment were selected for the measurements.

Determination of chlorophyll concentration was performed following previously described procedures (Li et al., 2016). Leaves of WT and transgenic lines were harvested, weighed and extracted with aqueous ethanol (95% v/v). The absorbance (A) of the extract was recorded at wavelengths of 663 and 645 nm using a spectrophotometer (Shimadzu UV2400, Japan). Total chlorophyll concentration was calculated as 8.02A₆₆₃+20.21A₆₄₅, and was expressed as mg chlorophyll g^-1 FW.

Proline content in rice leaves was determined according to the method described by Bates et al. (1973). About 0.5 g of leaf tissue was homogenized in 5 mL of 3% sulfosalicylic acid. After centrifugation at 12,000 × g for 10 min, the supernatant (1 mL) was mixed with 1 mL of ninhydrin and 1 mL of glacial acetic acid and then incubated at 100°C for 1 h. The reaction was then cooled down in an ice bath. Two milliliters of toluene was added to extract the resulting colored product and the absorbance was measured at 520 nm with a microplate reader (SpectraMax M5).

Malondialdehyde (MDA) content was determined based on the method described by Heath and Packer (1968) and described in detail in Cai et al. (2015), with minor modifications. Briefly, approximately 0.5 g of rice leaves were homogenized in 10 mL of 10% trichloroacetic acid (w/v) and centrifuged at 5,000 × g for 10 min. The supernatant (2 mL) was reacted with 2 mL of a chilled mix of 0.6% (w/v) thiobarbituric acid in 10% (w/v) trichloroacetic acid in a test tube at 100°C for 15 min. The reaction was quickly cooled on ice and then centrifuged at 5,000 × g for 10 min. Absorbance of the supernatant was measured at 450, 532, and 600 nm and the MDA content was calculated using the equation: 6.45 × (OD₅₃₂ - OD₆₀₀) - 0.559 × OD₄₅₀.

Extraction and determination of hydrogen peroxide (H₂O₂) through reaction with 0.1% TiCl₄ in 20% H₂SO₄ were performed according to the method of Mostofa and Fujita (2013).

Total protein from rice leaves was extracted with 0.05 M potassium phosphate buffer (pH 7.0). After centrifugation at 12,000 × g at 4°C for 15 min, the resulting supernatant was used to determine POX and CAT activities. POX activity was determined according to the method described by Ning et al. (2010). Briefly, each POX reaction mixture contained 0.1 mL of the enzyme extract, 2.9 mL of 0.05 M potassium phosphate buffer (pH 5.5), and 1 mL of 0.5% (v/v) H₂O₂ plus 1 mL of 0.05 M guaiacol as substrates. Absorbance readings at 470 nm were performed every 10 s to monitor the oxidation of guaiacol. CAT activity was determined following the method reported by Mostofa and Fujita (2013).

Data were analyzed by ANOVA using the statistical SPSS 10 program (SPSS Inc., Chicago, IL, United States). Statistically differences (P ≤ 0.05) between WT and OsHAK1 transgenic lines are indicated on the histograms by asterisks.

The possibility that expression of OsHAK1 could be affected by water deficit was analyzed by qRT-PCR. Results showed that expression of OsHAK1 was significantly induced in both roots and shoots of plants treated with 15% PEG (Figure 1). Overall, OsHAK1 was more induced in roots than in shoots after dehydration treatment. The expression profile of OsHAK1 in both roots and shoots followed a sigmoid curve. In roots transcript levels increased after 1 h of PEG application, reaching its peak after 6 h, to then gradually decline with prolonged treatment (12–24 h), showing two-fold to three-fold increase (Figure 1A). In shoots, expression of OsHAK1 was highly expressed at 6 h after treatment to later gradually decline after up to 12 h of stress, showing a barely two-fold increase (Figure 1B).

To further evaluate the stress-regulated expression of OsHAK1, transgenic rice plants expressing the GUS reporter gene under the control of the OsHAK1 promoter were generated. OsHAK1 expression was examined in the roots and leaf blades (Figures 2A–D). GUS activity was significantly increased in OsHAK1p: GUS transgenic plants treated with 15% PEG (Figure 2E). Quantified GUS activity was also higher in roots than in shoots (Figure 2E), which was consistent with the qRT-PCR analysis (Figure 1). These results indicate that OsHAK1 is induced by osmotic stress, and may be playing significant roles in the response of rice to drought.

To understand any role of OsHAK1 in drought stress responses, plants overexpressing this gene in cv. Nipponbare (Ox lines) and two T-DNA insertion lines of oshak1 in the Dongjin and Manan backgrounds (KO lines) were obtained. Generation and basic characterization of OsHAK1 transgenic lines have been previously described by Chen et al. (2015b).

To further assess how OsHAK1 expression affects rice growth, WT, KO, and Ox lines were grown hydroponically in control medium and solution with 15% PEG to produce an osmotic stress treatment mimicking drought stress. As shown in Figures 3A,C, no differences were detected between the growth of each genotype and its respective wild type when they were grown under control condition. Upon exposure to the dehydration medium, wilting and foliar chlorosis was observed in WT and KO lines, but not in the OsHAK1 overexpressing plants (Figures 3B,D). In addition, the stress treatment led to suppression of shoot and root growth in all three genotypes. The magnitude of reduction of shoot and root growth in Ox plants was significantly less than that in KO plants (Figures 3A–D). Plant height, dry root biomass, and shoot biomass of Ox seedlings were significantly higher than those of WT under stress condition. Meanwhile, the total dry weights of the oshak1 mutants grown with 15% PEG decreased in average to about 87% of their respective WT (Figures 3E,F).

The architecture of the root system is an important trait responsible for efficient nutrient acquisition under stress conditions. The observation that OsHAK1 expression affected rice growth under stress conditions, prompted us to evaluate whether the roots of OsHAK1 transgenic lines differed from WT in their response to the stress treatment. As shown in Figure 4, exposure of the seedlings to dehydration medium led to significant reductions in their adventitious root length, root surface area, and total root length. The reduction was greater in KO lines than in their respective WT (Figure 4), such that KO lines showed shortened length of the total root system and adventitious roots, and decreased adventitious root number and root surface area. However, these parameters maintained the opposite tendency in the Ox lines (Figures 4E–H). These results suggest that the larger root system of Ox lines under dehydration stress may facilitate nutrition acquisition, thus providing beneficial conditions for rice growth and water stress tolerance.

Given that the improved stress tolerance of OsHAK1-Ox plants may be due to increased K accumulation in root and shoot tissues, moreover, Na competes with K for uptake across the plasma membrane, we compared the effects of the osmotic stress on K and Na concentrations in roots and shoots. There were no significant differences between each genotype and its respective WT under control conditions (Figure 5). In the presence of 15% PEG, KO lines had extremely low K concentration while Ox lines contained significantly more K in both root and shoot tissue (1.26-fold in roots and 1.17-fold in shoots) than their respective WT (Figures 5A,B), whereas Na content maintained the opposite tendency (Figures 5C,D). These results suggested that sustained high K levels in OsHAK1 overexpression rice plants can contribute to the enhanced tolerance to water deficit.

To better understand the mechanisms underlying K accumulation in OsHAK1 transgenic seedlings, we analyzed the expression level of genes encoding K transport proteins. The data showed that drought enhanced the expression of OsTPKb (K selective vacuolar channel) and OsAKT1 (K inward rectifying channel) in all genotypes (Figures 6A,B). In addition, these genes were the most strongly induced under osmotic stress in overexpression seedlings compared to KO lines and WT (Figure 6). In contrast, the expression level in KO lines was evidently lower than in WT (Figure 6).

To evaluate plants’ response to drought in a more natural growing condition, rice seedlings grown hydroponically were transferred to pots to further investigate the phenotype of all three genotypes applying full watering (100% field capacity) or drought stress (40% field capacity) for a period of 3 weeks. The oshak1 mutants took up substantially less K into both their roots and shoots under both control and drought stress conditions, and the over-expressor plants accumulated significantly more K in both their root and shoot tissue than did WT plants (Supplementary Figure S1).

Several physiological parameters including RWC, REL, photosynthesis rate (Pn), chlorophyll content, proline content, and MDA content were examined in the transgenic lines and the respective WT plants to assess whether any differences can be associated with a higher tolerance to drought. Under normal growth condition, no significant differences in leaf RWC was observed between transgenic lines and WT plants (Figure 7A). When subjected to drought stress, RWC of both WT and transgenic lines was reduced; however, RWC was noticeably higher in the Ox lines and lower in the KO lines compared to the WT plants (Figure 7A). Under control conditions, electrolyte leakage, an indicator of membrane damage, was similar in transgenic and WT plants, whereas leaves of Ox lines presented significantly lower (20–25% decrease) electrolyte leakage levels compared to WT after 3 weeks of stress treatment (Figure 7B). Moreover, over 60% of the ions leaked from tissue of KO plants, while ion leakage of their respective WTs was less than 50% (Figure 7B).

Photosynthesis rate and chlorophyll content are important biochemical trait markers of stress tolerance in plants. We compared the effects of drought stress on Pn and foliar chlorophyll content of the three genotypes. Both Pn and chlorophyll content were comparable in the three genotypes under control conditions (Figures 7C,D). After drought treatment, there were significant reductions in Pn in all genotypes, but the magnitude of the decline was greater in WT than in Ox plants, leading to a significantly higher Pn in Ox lines (Figure 7C). Similarly, a significant reduction in leaf chlorophyll content was detected in WT seedlings, while Ox plants maintained relatively constant chlorophyll content when challenged by drought (Figure 7D). On the other hand, the KO lines exhibited dramatically lower Pn and chlorophyll content than their respective WT cultivars under drought conditions (Figures 7C,D).

The accumulation of compatible osmolytes such as proline (Pro), is generally, considered to be an adaptive response against environmental stresses (Ahmed et al., 2013). Meanwhile, accumulation of MDA content reflects damage to the structural integrity of cell membranes caused by oxidative stress such as that derived from drought stress (Zhao et al., 2014; Cai et al., 2015). To associate these functional attributes with the apparent drought tolerance exhibited by OsHAK1 overexpressors, Pro and MDA contents were analyzed. Accordingly, Pro and MDA contents in transgenic lines were similar to those in WT plants when grown under control condition (Figure 7E). Drought treatment resulted in an overall increase in Pro content in both transgenic and WT plants compared to those under control condition (Figure 7E). Even so, the OsHAK1 Ox lines exhibited 27% more Pro content after 3 weeks of drought stress compared to WT plants, whereas in KO lines Pro accumulation was about 25% less than that in WT plants after drought treatment (Figure 7E). In contrast, OsHAK1 overexpression caused significant reduction in the accumulation of MDA. Under drought stress, Ox plants had 23% less MDA compared to WT plants (Figure 7F), meanwhile KO plants presented 33% additional MDA content than WT (Figure 7F).

Drought usually causes injury at the cellular level via oxidative stress including the generation of ROS, such as H₂O₂ and superoxide (Mittler, 2002; Xiong and Zhu, 2002; Jiang et al., 2016). The lower level of lipid peroxidation in Ox transgenic plants, indicated by a reduced MDA accumulation, may be associated with a reduced ROS accumulation upon drought stress. Therefore, we determined whether OsHAK1 is involved in ROS detoxification. Indeed, the Ox plants showed much less drought-induced H₂O₂ accumulation compared to WT (Figure 8A). In contrast, H₂O₂ accumulation in KO lines was ∼23% higher compared to their respective WT, while both WT and transgenic plants showed similar H₂O₂ levels under control conditions (Figure 8A). The reduced stress-induced H₂O₂ accumulation in the Ox lines could be the result of changes in ROS-scavenging activities such as those of POX and CAT, enzymes involved in H₂O₂ elimination (Mittler, 2002; Apel and Hirt, 2004; Cai et al., 2015). Under normal growth conditions, levels of POX and CAT activities were not significantly different among the genotypes tested, however, after 3 weeks of drought treatment, both activities were markedly higher in Ox plants and lower in the KO lines compared to WT plants (Figures 8B,C). We further assayed the expression of genes encoding for OsPOX1, OsCATA, and OsCATB. Consistent with the increase of the enzymes activities, all three genotypes presented up-regulation of the transcript levels of these genes in response to drought, with a greater increase in Ox plants and a smaller increase in KO lines than in WT (Figures 8D–F). These results suggest that over-expression of OsHAK1 enhances the ROS-scavenging capacity, which decreases ROS damage under drought stress conditions.

To further understand the mechanism underlying the increased drought tolerance in OsHAK1-overexpressing plants, we performed gene expression profiling of selected genes known to respond to multiple abiotic and biotic stresses. These were OsDREB2A (Dubouzet et al., 2003), SNAC2 (Nakashima et al., 2007; Hu et al., 2008), OsP5CS1 (Sripinyowanich et al., 2013), OsbZIP23 (Zong et al., 2016), OsMYB2 (Yang et al., 2012), and OsAP37 (Oh et al., 2009). Under normal growth conditions, most genes displayed relatively weak expression and insignificant variations of transcription levels in both WT and transgenic plants except for OsDREB2A and SNAC2, which were significantly down-regulated in KO lines (Figures 9A–F). After 3 weeks of stress treatment, the relative expression of the genes was notably increased (Figures 9A–F). Stress-related transcription factors genes such as OsDREB2A, SNAC2, OsbZIP23, OsMYB2, and OsAP37 were up-regulated in Ox plants (0.25- to 1-fold over WT plants) after drought treatment (Figures 9A,B,D–F). Expression of the delta-pyrroline-5-carboxylate synthetase gene, OsP5CS1, was strongly induced in Ox lines under drought stress compared to control plants (Figure 9C). The higher transcript level of OsP5CS1 was consistent with the higher Pro content in Ox plants (Figure 7E). By contrast, expression of all of these genes showed opposite trends in oshak1 mutants, with noticeably lower values when compared to WT plants subjected to drought (Figures 9A–F). These data indicate that over-expression of OsHAK1 in rice enhances the expression of a set of stress-response genes that ultimately can support an increased tolerance to drought.

Reports have demonstrated that rice yields are affected by drought stress at the booting stage (Guan et al., 2010; Zhang et al., 2013). To evaluate whether OsHAK1 can function in improving rice yield under stress conditions at this developmental stage, we compared several agronomic traits between OsHAK1-Ox and WT plants. T₅ transgenic Ox lines and WT plants were subjected to 40% field capacity treatment at the booting stage until harvest. Results showed that all OsHAK1 Ox plants exhibited more effective tiller number (more than 10–15%) (Figure 10A), higher spikelet fertility (15–20% higher) (Figure 10B), increased 1000-grain weight (more than 7–10%) (Figure 10C), and higher grain yield per plant (25–35% higher) (Figure 10D) compared to WT under drought conditions. Meanwhile, under well irrigation, all Ox lines and WT plants displayed similar performance for these agronomic traits (Figures 10A–D). These results indicate that over-expression of OsHAK1 did not affect growth or yield productivity of rice grown under normal conditions, but improved the degree of drought tolerance under water-limiting conditions.

Potassium nutrition is closely related to plant water homeostasis and water use efficiency (Kuchenbuch et al., 1986; Tanguilig et al., 1987; Ahmad et al., 2016a). Enhanced K uptake is a key response of plants suffering from drought (Andersen et al., 1992; Wang et al., 2004; Ahmad et al., 2016b) as limiting K deficiency improves water retention, ensures appropriate stomatal regulation and helps to maintain photosynthetic activity via photoassimilate translocation (Römheld and Kirkby, 2010; Zörb et al., 2014). A causal link between OsHAK1 overexpression and improved tolerance to water deficit derives from the observation of a higher K accumulation in both roots and shoots of Ox plants compared to control plants (Figures 5A,B and Supplementary Figure S1). Improved K retention lowers cellular water potential and prevents further water loss, for example from roots to soil. As previously reported, OsTPKb can alter K concentration in small vacuoles, which determines the overall cellular K homeostasis which, in turn, influences stress tolerance (Ahmad et al., 2016a). It has been reported that overexpression of OsAKT1 in rice improves osmotic and drought stress tolerance by boosting tissue K levels, particularly in roots (Ahmad et al., 2016b). To better comprehend the mechanisms of drought tolerance conferred by the relatively higher K levels in Ox lines, the expression of two reported genes were investigated. Significant up-regulation of the transcript levels of OsTPKb and OsAKT1 was observed in OsHAK1-Ox plants compared to WT plants under osmotic stress conditions (Figures 6A,B), indicating that OsHAK1 affected the expression of genes encoding K transporters or channels, such as OsTPKb and OsAKT1, that can stimulate drought tolerance in rice.

We observed that overexpression of OsHAK1 driven by its native promoter significantly increased the growth of roots and aerial parts, whereas OsHAK1 knockout lines displayed impaired growth compared to WT cultivars under stress (Figures 3, 4). In plants, K participates in maintaining both root and shoot growth, including regulation of cell cycle (Sano et al., 2007) and in the completion of cell death programs (Peters and Chin, 2007). We found that K concentration was markedly higher in OsHAK1-Ox plants and lower in the KO mutants compared to WT after water-limiting treatments under both hydroponic (Figures 5A,B) and soil culture conditions (Supplementary Figures S1A,B).

Characteristics of the root system are important for efficient K acquisition in plants growing under K-deficient conditions. In an Arabidopsis mutant of the HAK-homolog TRH1 (AtKUP4/AtKT3), both Rb⁺ uptake and root hair growth were inhibited, implying that TRH1-mediated K uptake is required for elongation of root hair cells (Rigas et al., 2001). athak5 mutants presented significant shorter roots than those of WT when grown under low K levels or in the absence of the nutrient, however, such difference in growth was slim under a high external K medium (Qi et al., 2008). The observation that OsHAK1 KO seedlings had a lower K concentration in their roots and shoots than WT plants under stress conditions, prompted us to evaluate whether their root systems differed in response to the treatments. The oshak1 mutants showed shorter roots, dwarf size, less adventitious root number, and lower root surface area compared to WT under stress conditions at the seedling stage (Figure 4). In contrast, Ox lines showed an opposite trend with significantly improved root growth when plants were exposed to 15% PEG (Figure 4). These results suggest that the more extensive root system of Ox seedlings may facilitate K acquisition, thus equipping them to better overcome the K deficiency associated with drought stress.

Improved growth and yield productivity were observed in Ox lines at the harvest stage in soil water deficit experiments (Figure 10). There are two possible explanations for the phenotype of Ox plants. First is the enhanced vegetative and reproductive growth by increased K acquisition in the Ox lines. Overexpression of OsHAK1 increased root surface and total root length which would lead to an improved absorption surface for K, hence, an increase of K concentrations in all the tissues when grown under stress conditions (Figures 4G,H, 5A,B and Supplementary Figure S1). The second is the enhanced chlorophyll content and photosynthetic rates in OsHAK1 overexpressors under stress conditions. Chlorophyll content and photosynthetic rates in Ox lines were noticeably higher than in WT plants under drought stress (Figures 7C,D). OsHAK1 led to decreased chlorophyll loss which benefits in sustaining photosynthetic rates, therefore helping rice adaptation to drought. These findings highlight that maintenance of a higher K concentration and a better growth performance, particularly with a larger root system, are influential strategies for tolerance to dehydration stress in rice.

Reactive oxygen species accumulation associated with abiotic and biotic stress responses is often considered a manifestation of stress-induced damage, but it can also have a role as signal to trigger stress adaptation (Dat et al., 2000). The experiments performed to evaluate oxidative damage, ROS accumulation and detoxification activities in OsHAK1 transgenic lines and WT plants support two conclusions: (1) OsHAK1-Ox plants display fewer indications of oxidative stress upon challenge by dehydration and (2) OsHAK1-Ox plants activate detoxification mechanisms more readily when challenged by drought stress.

The first indication of oxidation damage was the lower MDA level detected in Ox lines under drought conditions compared to WT (Figure 7F). A similar trend was reported by Zhao et al. (2014), where MDA accumulation was significantly reduced in Zmhdz10 overexpression plants (less sensitive to drought stress). Another sign of oxidative stress is H₂O₂ buildup. Accumulation of H₂O₂ as result of dehydration damage has been documented in rice (Cai et al., 2015; Jiang et al., 2016). In accordance with the lower MDA accumulation in OsHAK1 Ox lines (Figure 7F), lower levels of H₂O₂ was also detected in these plants compared to WT under drought stress (Figure 8A).

Proline accumulation, generally considered an osmoprotection response associated with membrane and protein stability (Xiang et al., 2007; Zhao et al., 2014), was also measured. Although still debatable, several studies in rice support the idea that Pro functions in the osmotic adjustment under drought stress (Cai et al., 2015; Hong et al., 2016; Jiang et al., 2016). In our research, OsHAK1-Ox plants under drought conditions accumulated significantly more Pro compared to WT (Figure 7E). P5CS1, responsible for catalyzing Pro biosynthesis, is critical for increasing abiotic stress tolerance (Zhu et al., 1998). Drought, salt, and abscisic acid induce the expression of OsP5CS1, resulting in increased Pro content, and overexpression of OsP5CS1 improves osmotolerance (Igarashi et al., 1997; Zhu et al., 1998). The distinctly elevated transcript levels of OsP5CS1 were consistent with the high Pro content detected in Ox plants after drought treatment (Figures 7E, 9C). This increased Pro content could account for higher osmolarity, thus leading to a lower water potential and making plant tissues more efficient at retaining water (Song et al., 2012). Consistently, the leaves of Ox lines presented superior RWC than WT plants (Figure 7A), suggesting that OsHAK1 overexpression resulted in an increased capacity of water conservation under drought stress. Since Pro has also been suggested to act as an antioxidant to reduce oxidative damage (Székely et al., 2008), the higher Pro in Ox plants may also contribute to some extent, to the lower REL and lower MDA content during drought stress (Figures 7B,E,F).

In addition, enhanced activities of antioxidant enzymes, such as SOD, POD, and CAT, are considered a coping strategy for ROS scavenging and reducing programmed cell death in plants (Mittler, 2002; Apel and Hirt, 2004; Farooq et al., 2009). Overexpression of genes involved in ROS detoxification results in lower cellular damage and improved abiotic stress tolerance (Zhang et al., 2013). In OsHAK1-Ox plants, increased POX and CAT activity levels (Figures 8B,C) and elevated transcription of genes encoding those antioxidant enzymes (Figures 8D–F) were detected after drought treatment, indicating that those plants had enhanced capability to scavenge ROS and were better protected from oxidative damage. Consequently, OsHAK1 overexpression could alleviate oxidative damage in transgenic plants by enhancing Pro accumulation and antioxidant defense.

Several reports have established that overexpression of transcription factors leads to induced expression of stress- and ABA-responsive genes, which in turn contributes to improved tolerance to various stresses (Xiong et al., 2014; Zhao et al., 2014; Hong et al., 2016; Jiang et al., 2016; Shen et al., 2017). In our study, expression of six stress-related genes was significantly induced in drought-treated OsHAK1-Ox transgenic plants when compared to WT (Figure 9). Overexpression of the transcription factor OsDREB2A confers salt and dehydration stress tolerance in rice (Cui et al., 2011; Mallikarjuna et al., 2011). SNAC2 is a stress responsive NAC transcription factor that improves cold, salinity, and osmotic stress tolerance of rice (Hu et al., 2008). As discussed, P5CS1 encodes a key enzyme responsible for Pro synthesis and participates in plant stress tolerance (Yoshiba et al., 1999). OsbZIP23 is a central regulator in ABA signaling and biosynthesis (Zong et al., 2016) and transgenic rice overexpressing OsbZIP23 displayed improved tolerance to drought and salinity (Xiang et al., 2008). OsMYB2, encodes a stress-responsive MYB transcription factor, is also engaged in rice tolerance to salt, cold, and dehydration stresses (Yang et al., 2012). Overexpression of AP37 in rice has been reported to enhance tolerance to drought, salinity and low temperature at the vegetative stage and significantly increases grain yield under severe drought conditions (Oh et al.,2009). Therefore, we postulate that the enhanced tolerance to drought stress shown by OsHAK1 overexpression plants may be explained, in certain degree, by the up-regulation of these genes.

A constant shortcoming caused by constitutive gene over-expression is the abnormal development and often reduced crop productivity (Kasuga et al., 1999; Hsieh et al., 2002; Nakashima et al., 2007; Morran et al., 2011). Therefore, current goals to improve stress tolerance in crops include efforts to diminish any negative impacts on plant growth and yield (Cattivelli et al., 2008). No significant phenotypic differences were observed between OsHAK1-Ox lines and control plants under normal conditions (Figures 3, 4, 10). Most notably, Ox plants showed significant enhanced drought tolerance in field conditions, with a 35% increase in grain yield per plant over controls (Figure 10D). Our data suggest that the OsHAK1 gene constitute a target for genetic engineering or breeding approaches aiming to the generation of rice cultivars with enhanced drought tolerance.