Global warming and climate change have become a primary concern worldwide (Intergovernmental Panel on Climate Change [IPCC], 2012). Since the middle of the 20^th century, there have been considerable changes in the nature of droughts, extreme weather events, and floods in many regions of the world, which caused marked damage to crop production and great threat to global food security (Wheeler and Von Braun, 2013; Lesk et al., 2016). It has been estimated that, during 1964–2007, droughts significantly reduced cereal production by 10% on average, and this percentage was increasing annually, due to the rising drought severity, increasing vulnerability, and exposure to drought (Lesk et al., 2016). In recent two decades, the effect of drought stress on crop growth, yield, and quality was increasingly becoming a major issue of scientific concerns (reviewed by Kang et al., 2009; Alqudah et al., 2010). And, breeding programs with the aim to enhance crop productivity under drought stress are a top priority in the era of climate change (Jagadish et al., 2015). The occurrence of water stress at the reproductive stage is the most critical, as it strongly impacts yield and seed quality. Understanding the response mechanism of drought stress at crop reproductive stage will help to partially address some of concerns for improving crop tolerance to drought stress and to minimize consequent impacts.

Barley is the fourth most important cereal crop worldwide in terms of planting area, and is primarily used for food, brewing, and animal feed (Druka et al., 2011). In last few decades, malt quality of barley grains was heavily addressed and found to be associated with many chemical constituents and enzymes, such as β-amylase activity and β-glucan content (Wei et al., 2009a). β-Amylase, which is a key factor affecting the capacity of starch degradation during grain germination (Beck and Ziegler, 1989), is closely related to diastatic power (DP) (Liu et al., 2005). There have two genes encoding β-amylase been identified: BMY1 and BMY2. Of them, BMY1 is mainly expressed during the gain filling stages, and plays a vital role in regulating the gain β-amylase activity and malting quality (Li et al., 2002). And, low β-glucan content is required for brewing as it favors high wort filtration rate and malt extract. In barley grains, the β-glucan content was found to be largely regulated by three genes, GSL1, GSL4, and GSL7 (Schober et al., 2009). Furthermore, it has been documented that both β-glucan content and β-amylase activity are not only genetically controlled, but also greatly affected by environmental factors (Zhang et al., 2001, 2006; Djukić and Knežević, 2014; Rakszegi et al., 2014). It was reported that β-glucan content in barley grains was highly reduced when plants suffered from heat or water stress during grain-filling process (Macnicol et al., 1993; Christensen and Scheller, 2012). On the other hand, total protein content in barley grains was generally increased under the drought or heat stress condition, consequently resulting in higher β-amylase activity (Todaka et al., 2000; Yin et al., 2002; Qi et al., 2006). Such reduction of β-glucan content and increase of β-amylase activity were also observed in our previous study on water stress at the grain-filling stage, and they were closely associated with the reduction of grain yield and increase of protein content, which would consequently cause deterioration and instability of grain yield and malt quality (Wu et al., 2015). Furthermore, our previous study also found such effect of water stress on the β-glucan content and β-amylase activity varied dramatically among barley genotypes, especially between the Tibetan wild barley and cultivated ones (Wu et al., 2015). Therefore, to achieve the stability of grain yield and malt quality under water stress condition, it is imperative to reveal the mechanisms in the adverse impact of water stress on malt quality, including β-glucan content and β-amylase activity, and the difference among barley genotypes in the response to water stress.

Metabolite profiling may provide a comprehensive approach to assess a broader spectrum of constituent analysis and has been proven to be a suitable tool for investigating the metabolites changes caused by genetic modification and environmental condition (Röhlig et al., 2009; Frank et al., 2011; Cañas et al., 2015). Currently, metabolomic analysis has been extensively conducted on the changes of metabolite profiles under water stress in many species, such as tomato (Rivero et al., 2015), soybean (Silvente et al., 2012), maize (Sun et al., 2015), and rice (Li et al., 2015). In barley, several metabolite profiling studies have also been performed to determine the influence of water stress on free amino acids (Lanzinger et al., 2015), abscisic acid, and the oxidative status (Thameur et al., 2014). However, to our knowledge, no such report has been addressed to the effect of water stress on malt quality. In the present study, four wild and cultivated barley genotypes contrasting in drought tolerance were used to compare metabolic changes in the response to water stress using gas chromatography-mass spectrometry (GC-MS). The objectives of the current work are to determine the mechanisms in the influence of water stress on β-amylase activity and β-glucan content, and to reveal the reasons why barley genotypes differ in their responses to water stress.

It is well documented that metabolic changes happen in the plants exposed to water stress, resulting in yield loss (Boyer, 1982; Bray et al., 2000). In this study, water stress induced a significant reduction of grain weight for all genotypes, compared with the control (Figure 1), being consistent with the results obtained in our previous work (Wu et al., 2015). However, such reduction of grain weight differed greatly between genotypes, with XZ147 being the least and XZ5 being the most. It was reported that carbohydrates and starch account for 78–83 (MacGregor and Fincher, 1993) and 50–70% (Henry, 1988) of barley grain weight, respectively. Photosynthetic product (glucose) is transported to grains in the form of sucrose, and used for synthesis of starch as well as β-glucan. The inhibited synthesis of starch by water stress may be a major reason for the reduction of grain weight (Chaves, 1991). In this study, the available sucrose transported to grains was dramatically increased under water stress for all barley genotypes, with XZ147 having the most increase (Figures 2, 5). Thanks to its much quicker grain-filling process under water stress (Wu et al., 2017), XZ147 remained relatively smaller change in grain weight than other genotypes (Figure 1A).

In the present study, the PCA revealed that the clustering between barley genotypes strongly depended on fraction II (amino acids and amines) and fraction III (sugars and sugar alcohols) (Figures 4B,C). So, it can be assumed that these low molecular compounds (amino acids, amines, sugars, and sugar alcohols) played an important role in barley’s tolerance to water stress. Indeed, both our and numerous previous studies have found that the content of amino acids in plant vegetative tissues and reproductive grains changed remarkably when subjected to water stress (Widodo et al., 2009; Sicher et al., 2012; Nam et al., 2014; Table 1). It is well known that osmotic adjustment by the accumulation of compatible solutes is a key mechanism for maintaining cell turgor under water stress (Serraj and Sinclair, 2002; Hummel et al., 2010). Proline is one of the main osmoprotectant in plants when subjected to osmotic stress, such as drought or salinity (Delauney and Verma, 1993; Liu and Zhu, 1997). It has been also reported that the accumulation of mannitol and inositol increased in plants when exposed to osmotic stress like salinity (Abebe et al., 2003; Sanchez et al., 2008). In this study, the contents of both proline and mannitol were increased by water stress in all used barley genotypes, whereas the drought tolerant wild barley XZ147 showed the least increase among all genotypes (Table 1). Moreover, XZ147 showed the reduction in inositol content under water stress relative to the control, while other three genotypes showed the increase of this metabolite. Very similar results were also obtained in the study on the metabolic responses to salt stress of barley (Widodo et al., 2009). These results suggested that the increasing accumulation of osmoprotectants like proline and mannitol is an important adaptive strategy of barley to survive under the terminate drought stress, especially for the sensitive genotypes. Likewise, synthesis of some amino acids and organic acids was significantly enhanced under water stress (Table 1), which could be in favor of osmotic adjustment and membrane stability (Ueda et al., 2004; Chen et al., 2007; Widodo et al., 2009). It has been previously documented that poly-amines (PAs) associated with some related amino acids play important roles in the adaptation to abiotic and biotic stresses by regulating carbon/nitrogen homeostasis or acting as signaling molecules and compatible solutes under drought stress (Moschou et al., 2008; Liu et al., 2010; Hussain et al., 2011; Zeier, 2013). However, the biosynthesis of these compatible solutes by plants is always at the high cost of photo-assimilates and energy, which consequently sacrifice the grain yield and quality (Bolton, 2009). In this study, we found a much higher increased accumulation of osmoprotectants (such as proline, mannitol, fructose, sucrose, and citric acid) in drought-sensitive wild XZ5 than drought-tolerant wild XZ147 (Table 1), which consequently severely affected the yield formation and the two malt quality traits (β-amylase activity and β-glucan content) (Figure 1), completely coinciding with the findings of Serraj and Sinclair (2002).

β-Amylase activity is an important malt quality parameter, which is positively correlated with DP (Ovesná et al., 2012). Furthermore, a number of studies have demonstrated that β-amylase induction could help plants cope with unfavorable growing conditions (Dreier et al., 1995; Nielsen et al., 1997; Todaka et al., 2000; Kaplan and Guy, 2004), possibly by hydrolyzing more starch to maltose to function as a compatible-solute stabilizing factor (Lu and Sharkey, 2004; Kaplan and Guy, 2005). Hejgaard and Boisen (1980) found that the genotypes rich in lysine had higher β-amylase activity because of more serine protease inhibitor (Z protein) and chymotrypsin inhibitors (CI-1 and CI2). In this study, the grain lysine content was dramatically increased by water stress especially for wild genotypes (Figures 2, 5), being consistent with the results reported previously (Macnicol et al., 1993; MacGregor et al., 1994; Wei et al., 2009b). So, it could be assumed that the increased grain lysine content is an important reason for the increased β-amylase activity under water stress. On the other hand, during grain filling stage of cereal crops or ripening stage of fruits, the increased levels of auxins could be observed along with the inhibition of β-amylase activity for starch accumulation (Eeuwens and Schwabe, 1975; Bangerth et al., 1985; Aufhammer et al., 1986; Purgatto et al., 2001). In the present study, we found that the content of L-Tryptophan, the precursor for auxin biosynthesis (Mano and Nemoto, 2012) was extensively increased by water stress for all used barley genotypes except XZ147. This might be one other explanation for the higher β-amylase activity in XZ147 than in the other genotypes under water stress. Abiotic stress like drought stress always causes excess reactive oxygen species (ROS) accumulation, which will result in cellular damage and consequently yield losses. Fortunately, plants have obtained numerous antioxidants to scavenge the excess ROS, for instance GSH and AsA, which play a key role in intracellular ROS level regulation (Apel and Hirt, 2004; Sairam and Tyagi, 2004; Møller et al., 2007). On the other hand, Wei et al. (2009b) reported that exogenous H₂O₂ treatment could elevate barley β-amylase activity, suggesting that the enhanced grain H₂O₂ concentration would increase β-amylase activity under osmotic stress. However, the increase in GSH concentration might extinguish this effect. In the present study, we found that the precursor of GSH biosynthesis, L-glutamine, was increased by water stress in all used barley genotypes, with XZ5 being significantly higher than the other genotypes. Such over accumulation of GSH in XZ5 would be helpful for the scavenging of ROS, but not benefit for the induction of β-amylase activity (Figure 1C). All these results indicated that the induction of barley grain β-amylase activity was highly correlated with the lysine content, indole-3-acetic acid (IAA) biosynthesis, and H₂O₂ homeostasis. Furthermore, our results revealed that the expression level of BMY1, which encodes β-amylase in barley grains, was dramatically induced in XZ147 after 7 days of water stress being accompanied by higher β-amylase activity (Figures 1C, 6A). Therefore, it can be hypothesized that the reason for the higher grain β-amylase activity in XZ147 than the other genotypes under water stress might be attributed to its higher lysine content, less IAA biosynthesis, more stable H₂O₂ homeostasis, and also the higher up-regulation of BMY1 gene.

β-Glucan content is a critical parameter for malt brewing, with high β-glucan content always resulting in slower wort filtration rate and beer haze (Bamforth, 1982). It has been reported that β-glucan accumulated in barley grains throughout the whole filling stage (Christensen and Scheller, 2012). Thus, it may be understandable that the faster grain filling process causes the lower β-glucan content. Indeed, our results revealed that the grain β-glucan content in wild barley (XZ147 and XZ5) was decreased much more than that in cultivated barley (Tr and TL43) when subjected to water stress (Figure 1B), completely coinciding with our previous results that the grain filling process of wild barley (XZ147 and XZ5) was much faster than the cultivated barley (Tr and TL43) after onset of drought stress (Wu et al., 2017). Furthermore, it has been also reported that the synthesis of β-glucan in barley grains was largely controlled by the genes GSL1, GSL4, and GSL7 (Schober et al., 2009). In the present study, we found that, in comparison with the cultivated barley, the expression level of GSL7 in wild barley was much lower and it was reduced much more with the prolongation of water stress (Figure 6D), which could account for the lower β-glucan content in the grains of the wild barley (Figure 1B).

In summary, water stress remarkably altered grain weight, β-amylase activity, and β-glucan content of the four genotypes used in this study. When subjected to water stress, the wild barley XZ147 showed the most increase of β-amylase activity among the four genotypes which might be attributed to its higher lysine content, less IAA biosynthesis, more stable H₂O₂ homeostasis, and more up-regulation of BMY1 gene. XZ147 also had the most reduction of β-glucan content under water stress than the other genotypes, which could be explained by the faster grain filling process and the less expression of β-glucan synthase gene GSL7. Obviously, XZ147 is interesting for malt barley breeders to develop the new cultivars with high water stress tolerance and stable malt quality.

FZ and GZ designed the project; XW and KC carried out the experimental work; XW and FZ wrote the paper with contributions from GZ; and all authors read and approved the paper.

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.