Contrasting nutrient–disease relationships: Potassium gradients in barley leaves have opposite effects on two fungal pathogens with different sensitivities to jasmonic acid

Abstract Understanding the interactions between mineral nutrition and disease is essential for crop management. Our previous studies with Arabidopsis thaliana demonstrated that potassium (K) deprivation induced the biosynthesis of jasmonic acid (JA) and increased the plant's resistance to herbivorous insects. Here, we addressed the question of how tissue K affects the development of fungal pathogens and whether sensitivity of the pathogens to JA could play a role for the K–disease relationship in barley (Hordeum vulgare cv. Optic). We report that K‐deprived barley plants showed increased leaf concentrations of JA and other oxylipins. Furthermore, a natural tip‐to‐base K‐concentration gradient within leaves of K‐sufficient plants was quantitatively mirrored by the transcript levels of JA‐responsive genes. The local leaf tissue K concentrations affected the development of two economically important fungi in opposite ways, showing a positive correlation with powdery mildew (Blumeria graminis) and a negative correlation with leaf scald (Rhynchosporium commune) disease symptoms. B. graminis induced a JA response in the plant and was sensitive to methyl‐JA treatment whereas R. commune initiated no JA response and was JA insensitive. Our study challenges the view that high K generally improves plant health and suggests that JA sensitivity of pathogens could be an important factor in determining the exact K–disease relationship.

However, it is important to note that experimental studies proving a relationship or even a correlation between K-induced physiological changes and disease severity are lacking.
Previous work in our laboratories identified K-dependent changes in metabolites of arabidopsis (Arabidopsis thaliana [L.] Heynh.), such as increases in reducing sugars and accumulation of glucosinolates, which are potentially of relevance to pathogens and pests in K-deficient plants Troufflard et al., 2010). K-deficient arabidopsis plants were found to have greater expression of genes related to the biosynthesis of the phytohormone jasmonic acid (JA) and of genes related to defence, the latter being dependent on the function of the JA receptor COI1 (Armengaud, Breitling, & Amtmann, 2004Yan et al., 2009). AtLOX2, encoding lipoxygenase 2, which catalyses the first committed step in JA biosynthesis (Delker et al., 2006;Wasternack & Hause, 2013), responded to low K prior to any visible symptoms (e.g., senescence and growth retardation), demonstrating that the induction of the JA pathway was not a secondary effect of stress symptoms (Troufflard et al., 2010). In agreement with the transcriptional regulation of JA-biosynthesis genes, levels of JA, as well as its precursors 12-oxo-phytodienoic acid (OPDA) and hydroxyl-12-oxo-octadecadienoic acids (HODs), were elevated in Kdeficient plants (Troufflard et al., 2010). Although extensive research on JA signalling has been carried out in dicots such as arabidopsis and tomato (Kazan & Manners, 2008;Pathak, Baunthiyal, Pandey, Pandey, & Kumar, 2017;Wasternack & Hause, 2013;Yan et al., 2016), JA signalling pathways in monocots are relatively unexplored (Ding, Yang, Yang, Cao, & Zhou, 2016;Lyons, Manners, & Kazan, 2013;Shyu & Brutnell, 2015). A number of genes induced in response to JA treatment have been identified in barley, but little is known about their function. They are referred to collectively as jasmonateinduced proteins (JIPs) and known by their molecular weight (Andresen et al., 1992;Wasternack, Parthier, & Mullet, 1997;Weidhase et al., 1987).
In light of the relationship between low plant K status and JA, it is possible that some of the variations in the effects of K nutrition on plant disease evident in the literature are due to different sensitivities of pathogens to JA. Thus, high concentrations of JA or related oxylipins in K-deficient plants might positively or negatively modulate plant-inherent defence responses. It has been proposed that necrotrophic pathogens induce plant defences through JA (Dar, Uddin, Khan, Hakeem, & Jaleel, 2015;Glazebrook, 2005;Kazan & Lyons, 2014;Thaler, Humphrey, & Whiteman, 2012) whereas biotrophic pathogens induce plant defences through the JA antagonist salicylic acid (SA). However, this generalisation does not always hold true.
To test the hypothesis that JA is an important factor for the Kdisease relationship in crops, we measured K concentrations in leaves of barley (Hordeum vulgare L. cv. Optic) plants grown under different K regimes and related them to transcript levels of JA-biosynthesis and JA-responsive genes, and the development of two fungal pathogens.
On the basis of agricultural importance and different lifestyles, we selected the obligate biotroph Blumeria graminis f. sp. hordei (powdery mildew, B. graminis) and the hemi-biotroph Rhynchosporium commune (rhynchosporium, R. commune). The UK malting barley variety Optic was selected due to its susceptibility to both fungi. Infection with B. graminis initiates no hypersensitive response or lesion formation, thereby allowing the fungus to spread across the leaf and to obtain nutrients from epidermal leaf cells (Glawe, 2008). The life cycle of R. commune (scald or leaf blotch) includes an early biotrophic phase during which the fungus grows asymptomatic under the cuticle, and a necrotrophic phase during which conidia are formed normally and necrotic lesions become visible on the leaf surface (Avrova & Knogge, 2012). The results obtained suggest that jasmonate-signalling links plant K status with disease development.

| Plant material and growth conditions
Barley (H. vulgare L. cv. Optic) seeds were germinated on water-saturated paper towels in an environmentally controlled growth chamber with 9-hr light (270 μmol·m −2 ·s −1 ) at 22°C and 15-hr dark at 18°C and constant 70% relative humidity. After 4 days, seedlings were transferred to hydroponic solution, supported by corrugated plastic sheets, each holding 60 plants, suspended above 10 L of nutrient solution.

| Preparation of detached leaf segments
Barley seedlings were grown for 14 days in control or −K solutions.
Segments that are 40 mm long were cut from the tip, middle, and base parts of the emerged blade of the second leaf (Supporting Information Figure S1). For subsequent analysis of K content, RNA, or oxylipins, the tissue was frozen immediately after cutting. Treatment of the leaf segments with Me-JA or fungal pathogens is described below.

| Determination of tissue water, K, and oxylipin contents
Approximately 100 mg of frozen shoots, roots, or leaf segments was weighed and freeze-dried overnight. Water content was determined as the loss of weight by drying and expressed as percentage of fresh weight. To determine K content, freeze-dried tissue from shoots, roots, or leaf segments was incubated in 2 M HCl (100 μl for 1 mg of dry tissue) at room temperature for 48 hr. Tissue debris was removed by centrifugation, and the extracts were diluted 1:500 in ddH 2 O. K was detected using a flame photometer (Sherwood flame photometer 410). K concentrations in the diluted extracts were determined from a standard curve established with solutions containing 15 to 250 μM KCl in 4 mM HCl. Tissue K concentrations were then calculated by multiplication with the dilution factor and the incubated dry weights. Oxylipins were measured in triplicate 50-mg samples of lyophilized leaf tissue from leaf segments of plants grown for 14 days in control or −K media (20 plants each). Extraction and liquid chromatography-mass spectrometry analysis were carried out according to previously described procedures (Dave et al., 2011). Initial analysis showed that the variation was too large to resolve differences between leaf segments. Therefore, data from all leaf segments grown in either control or −K media were pooled for statistical analysis.

| Measurement of transcript levels using quantitative PCR
Total RNA was extracted from leaf tissue using TRIzol® Reagent (Invitrogen, Cat. 15596-026)   Detached leaf segments were blotted dry on paper towel and transferred to 0.5% agar/120 mg L −1 benzimidazole plates for subsequent inoculation.

| Treatment with pathogens
Barley leaf segments were placed on 0.5% agar/120 mg L −1 benzimidazole plates (Newton, 1989;Newton, Hackett, & Guy, 1998) and incubated in a lit incubator (LEEC) with continuous light (light intensity 200 μmol·m −2 ·s −1 at 17°C) for 24 hr before inoculation with the fungal pathogens. R. commune isolate 13-13 from the culture collection at The James Hutton Institute was grown on CZV8CM agar medium (Newton, Hackett, & Guy, 1998) at 17°C in the dark. The mycelia were scraped from 14-day-old cultures using a sterile spatula, transferred to a homogenizer containing sterile water, and homogenized for approximately 30 s. The suspension was filtered through glass wool and resuspended in sterile distilled water at a concentration of 10 6 spores ml −1 . The leaf area to be inoculated was brushed gently with a trimmed-down paint brush to disrupt the cuticle (Newton, Searle, Guy, Hackett, & Cooke, 2001). Ten microlitres of 10 6 spores ml −1 solution was dispensed on to each leaf segment.
The plates were returned to the 17°C incubator. The severity of infection was assessed by measuring the length of the lesions ( Figure S2A).
B. graminis f. sp. hordei was isolated from infected barley leaves.
Spores from individual colonies were used to inoculate detached leaf segments with a paint brush, and the fungus was allowed to grow for approximately 2 weeks. To ensure a pure culture, individual colonies were selected twice more. To inoculate the leaf segments uniformly, an inoculation tower was used ( Figure S2B). The plate containing the spores was inverted over a sheet of paper and tapped to dislodge the spores. A cone was formed from the paper, and the spores were blown into the inoculation column. The spores were allowed to settle on the leaf segments for 5 min before the lids were replaced, and plates were returned to the lit incubator at 17°C. The level of infection was assessed by counting the number of visible colonies on each leaf segment ( Figure S2C) and dividing by the leaf area (measured from photographs using ImageJ). Noninoculated leaf segments kept in the same conditions as the inoculated leaf segments showed no visible signs of deterioration ( Figure S2D).

| Statistical analysis
Statistical analysis was performed using analysis of variance with Genstat Version 15.1 and calculation of Pearson correlation between parameters measured over time and across the leaf using Minitab 15 statistical software. Correlation coefficients are shown in Table 1, and P values for all correlations tested are given in Supporting Information Table S1.

| Leaves of K-deprived barley plants reach critically low tissue K concentrations
Barley seedlings were transferred to hydroponic culture 4 days after germination and grown on a minimal nutrient solution with either 2 mM K (control) or no K added (−K). No differences in plant size or development were apparent between treatments until 10-12 days after transfer to hydroponics ( Figure 1). Subsequently, K-deprived plants displayed constantly lower shoot fresh weights (Day 12, n = 3, P = 0.005) and shoot lengths (Day 12, n = 3, P = 0.013) than control plants (Figure 1a,b). The time point at which K deprivation started to impact visually on growth coincided with the emergence of the third leaf (Figure 1c-e). At this time, seed K reserves for leaf growth will have been exhausted (White & Veneklaas, 2012). The first leaf of Kdeprived plants grew to its full length, and the second leaf showed only a minor reduction in length at the end of its growth period ( Figure 1c,d). The third leaf, however, was shorter in K-deprived plants than in control plants from the beginning of its emergence on Day 10 ( Figure 1e). The root fresh weight of K-deprived plants was also less than that of the control plants grown in full nutrient medium after 10 days ( Figure S2A), although the roots were longer than those in control medium ( Figure S3B).
The K concentration in the medium had an impact on tissue K concentrations, expressed on a dry weight basis, before a difference in fresh weight was apparent (Figure 2, Supporting Information Figure S3). Three days after transfer to hydroponics, the K-deprived plants already had lower shoot K concentrations than the control plants (1.4% compared with 2.5% dry weight). Over the following 12 days, shoot K concentrations increased in the control plants and decreased in K-deprived plants (n = 3, P = 0.012; Figure 2a). The root K concentration in K-deprived plants was also lower than that in control plants on Day 3 (n = 3, P = 0.043) and remained constant thereafter while root K concentrations of control plants increased (Supporting Information Figure S3C). On Day 12, the shoot K concentration of K-deprived plants was only 14% (n = 3, P = 0.044) and the root K concentration was 22% (n = 3, P = 0.010) of the shoot K concentration of control plants. From this time point onwards, shoot growth was no longer sustained in K-deprived plants (Figure 1a,b).
Nevertheless, the overall shoot water content was maintained (Table 1).

| Leaf K concentration displays a gradient across the emerged blade
Potassium is mobile in the plant and is preferentially allocated to growing and metabolically active tissues (White & Karley, 2010).  To investigate spatial differences of tissue K concentrations within the leaf area that is most accessible to airborne pathogens, we measured K concentrations in three zones of the emerged part of the second leaf (base, middle, and tip as shown in Supporting Information Figure S1). In control plants, the K concentration decreased significantly from the base to the tip of the leaf blade (n = 3, P = 0.012), with the K concentration at the tip being 70% of the K concentration at the base ( Figure 3). This is consistent with the observations of Fricke, Leigh, and Deri Tomos (1994b). A decreasing base-to-tip leaf K concentration trend was also apparent in K-deprived plants, although the differences were not statistically significant ( Figure 3a).
In accordance with the function of K as a major osmoticum, K-deprived plants showed a significant decrease in water content (expressed as percentage of fresh weight) from the base to the tip of the leaf (n = 3, P = 0.004; Table 1), and the tip of the leaf was the first part of the plant to show chlorosis and necrosis ( Figure   S3F). Pearson correlation analysis of the data confirmed a positive correlation between K and water content within the second leaf (n = 9, R = 0.507, P = 0.032; Table 2). In summary, the experimental system allowed us not only to manipulate leaf K concentrations by varying external K supply but also to take advantage of natural differences between local leaf K concentrations within leaves of K-sufficient plants.  Figure S4 shows that it is difficult to identify the most likely functional homologue of arabidopsis LOX2 among the three barley genes on the basis of sequence similarity alone. In a preliminary expression analysis with all three genes, we found that LOX2.A displayed a more consistent response to −K than did the other HvLOX2 genes identified, and therefore, we selected it for further study. No VSP homologue was found in the available barley nucleotide or protein sequence databases, but a number of Me-JA-induced genes ("JA-induced proteins", JIPs) have been identified (Andresen et al., 1992; Weidhase et al., 1987). HvJIP60 (BM815987), used here, encodes a ribosome-inactivating protein with glycosidase activity (Chaudhry et al., 1994;Dunaeva, Goebel, Wasternack, Parthier, & Goerschen, 1999;Reinbothe et al., 1994). Three barley genes, (R = 0.548, P = 0.019) as reliable reporters of the overall shoot K concentration, and of local K and water concentrations within the leaf (Table 2). Note. Significant positive (green) and negative (red) correlations are shaded according to P value (<0.05 light, <0.01 medium, <0.005 dark). For exact P values, see Supporting Information Table S1. D: day after inoculation.

| Transcript levels of JA-related genes are inversely related to leaf K concentration
To test whether the increase in gene expression observed in response to K deficiency was associated with an increase in the con-

| Low tissue K has contrasting effects on powdery mildew and rhynchosporium
Typical disease symptoms from B. graminis and R. commune infection on barley leaves are shown in Supporting Information Figure S2.  Table 2). Thus, a low tissue K concentration in the leaves seems to protect barley against powdery mildew. Correlation analysis also revealed a significant negative correlation between B. graminis and transcript levels of JA-related genes or oxylipin concentrations (Table 2).  (Table 2). Thus, a low tissue K concentration in barley leaves seems to promote the development of R. commune.

| B. graminis, but not R. commune, is sensitive to Me-JA and induces JA-related genes
The preceding results suggest that induction of the JA signalling pathway by low K nutritional status may protect barley plants against powdery mildew but not against R. commune. This hypothesis is consistent with reports that external application of Me-JA or other oxylipins to barley inhibited powdery mildew development both locally and systemically (Cowley & Walters, 2005;Schweizer, Gees, & Mosinger, 1993;Walters, Cowley, & Mitchell, 2002) but had variable effects on infection by R. commune (Steiner-Lange et al., 2003;Walters et al., 2014;Weiskorn, Kramer, Ordon, & Friedt, 2002). These previous studies used different barley varieties and growth conditions; therefore, we compared JA sensitivity of the two fungal pathogens in our experimental system directly (Figure 6a

| The K-JA relationship: Possible signals and physiological functions
Previous work by our groups had discovered a strong effect of K deficiency on the JA biosynthesis and signalling pathways in arabidopsis (Armengaud, Breitling, & Amtmann, 2004;Armengaud, Breitling, & Amtmann, 2010;Troufflard et al., 2010). Many of the downstream targets of JA signalling (e.g., production of glucosinolates) are particularly prominent in Brassicaceae, and it was therefore conceivable that the JA response to K deprivation was limited to species of this angiosperm family. The results presented here show that this is not the case.
Transcript levels of HvLOX2 and HvAOC, encoding JA-biosynthetic enzymes that underlie positive feedback regulation by JA in arabidopsis (Delker et al., 2006), and of HvJIP60, previously identified in a screen for Me-JA inducible genes in barley (Andresen et al., 1992;Wasternack, Parthier, & Mullet, 1997;Weidhase et al., 1987), were consistently increased in K-deprived barley plants ( Figure 2). More strikingly, the relative levels of these three transcripts increased from the base to the tip of the emerged blade of the second leaf and thus displayed a gradient that was the inverse of the tissue K concentration gradient, even in plants that were grown in K-sufficient conditions ( Figure 3). We conclude that the expression of the genes is quantitatively determined by variation in tissue K concentration, whether the latter is the result of external supply or of endogenous tissue allocation. At this stage, we cannot distinguish whether the local K signal for JA metabolism is apoplastic or intracellular, and we can only speculate about the downstream events. A number of early signals in wounding and pathogen responses, for example, change in membrane potential, rise of cytoplasmic calcium, and H 2 O 2 production (Maffei, Mithöfer, & Boland, 2007;Thordal-Christensen, Zhang, Wei, & Collinge, 1997;Yang, Shah, & Klessig, 1997), also occur in response to reduced apoplastic K (Allen et al., 2001;Amtmann, Troufflard, & Armengaud, 2008;Armengaud et al., 2009;Shin & Schachtman, 2004). However, whether these signals can be quantitative and can persist long enough to explain a continuous dose-response gradient within the leaf is uncertain. More intriguing is the observation that constitutively high activity of the vacuolar cation channel TPC1 in the arabidopsis fou2 mutant results in high LOX2 activities (Bonaventure et al., 2007). The vacuole plays an essential role in cellular K homeostasis because it is used as a reversible K reservoir to maintain stable cytoplasmic K over a wide range of external K concentrations (Carden, Walker, Flowers, & Miller, 2003;Walker, Leigh, & Miller, 1996;White & Karley, 2010). Transtonoplast K fluxes through vacuolar channels will therefore reflect tissue K status in a quantitative manner. Indeed, TPC1 is permeable to K and has been implicated in K homeostasis (Amtmann & Armengaud, 2007;Beyhl et al., 2009;Peiter et al., 2005;Ranf et al., 2007), although it is not clear whether the link is direct (K transport through TPC1) or indirect.
Another good candidate for mediating between cellular K status and defence responses would be calcium. Single-cell measurements of ion concentrations in different parts of barley leaves have shown a negative correlation between vacuolar concentrations of K and Ca (Fricke, Hinde, Leigh, & Tomos, 1995). It has also been shown before for arabidopsis leaves that a decrease of tissue K under K starvation is compensated by a rise of Ca . Although it is unlikely that a change of the vacuolar Ca concentration directly impacts on the development of fungal pathogens, it could alter the signature of intracellular Ca signals in response to pathogens and thus impact on defence responses.

Genetic manipulation of vacuolar K and Ca transporters in barley
needs now to be undertaken to investigate whether it is possible to uncouple cellular K and/or Ca homeostasis from JA signalling and whether fluxes of K and/or Ca across the tonoplast underpin the effect of K on pathogen development.
The highest expression of HvLOX2, HvAOC, and HvJIP60 was measured in the tips of leaves of K-deprived plants, which not only had the lowest K concentration but also were the first parts of plants to show chlorosis and a significant drop in water content. It has been shown for arabidopsis that induction of two senescence-associated genes, AtSAG12 and AtSAG13, by K deprivation no longer occurred when JA antagonists SA and acetyl SA were applied (Cao, Su, & Fang, 2006). These findings raise the possibility that JA-related genes inform the plant about local tissue concentrations of the most important cellular osmoticum, K + , and induce senescence when tissue K concentration falls below a critical threshold.

| What underlies the differential effect of leaf K on B. graminis and R. commune?
The The protocols used here for inoculation and disease scoring followed established techniques in the pathogen field (Newton, 1989), but potential problems for combined nutrient-pathogen studies should be discussed. The extended incubation of the leaf segments did not lead to any visible deterioration of the tissues apart from chlorosis in a small area adjacent to the cut (see uninoculated segments after 15 days on plates shown in Supporting Information Figure   S1D). However, it is possible that the segments lose some K during the incubation period. Therefore, our K-disease results strictly relate to the differences of K/JA status before inoculation. which has constitutively high endogenous JA levels (Ellis, Karafyllidis, & Turner, 2002;Ellis & Turner, 2001). External application of jasmonate has also been shown before to reduce B. graminis infection in barley both directly and systemically, under controlled conditions (Schweizer, Gees, & Mosinger, 1993;Walters, Cowley, & Mitchell, 2002). Although this could be due to a direct beneficial role of K as an essential nutrient, it is difficult to conceive that the small differences of K  (Evans, Gottlieb, & Bach, 2003;Yan et al., 2015), it is also possible that an initial rise in JA leads to redistribution of K and/or Ca between cellular compartments and cell types. Monitoring ion concentrations and pathogen development at a much higher spatial resolution would be a good way forward to test these hypotheses. The experimental protocols developed here to score K-disease interaction provide a basis for such studies.

| A working model for the K-JA-disease interaction
The results from this study can be summarized in a simple working model ( Figure S7) in which a low K concentration in leaf tissue induces JA signalling, which in turn enhances the inducible defence response of the plant against B. graminis. In this case, the effect seems to be strong enough to overcome any other effects of low K status that may increase plant susceptibility. By contrast, R. commune does not induce a JA-based defence response, and this pathogen is not sensitive to JA. Induction of JA signalling by low K has therefore no consequence on pathogen development. The observed effect of K on R. commune is in accordance with the conventional view that K deficiency promotes disease, but the exact cause still remains to be identified. Our finding that the effect is local and continuous over a range of K tissue concentrations narrows the spectrum of potential causes.
For example, levels of sugars increased in −K conditions but were not correlated with K concentrations in the leaf segments ( Figure S8).
Interestingly, it has been reported that soil-grown barley plants FIGURE 7 Gradients in tissue K concentration, transcript levels, and disease symptoms across leaf zones and K treatments. Semiquantitative representation of tissue K concentrations (K), transcript levels of JA-related genes (HvLOX2, HvAOC, HvJIP60) and disease symptoms of Blumeria graminis (Bgh) and Rhynchosporium commune (R. c.) in whole shoots (left) as well as base (B), middle (M), and tip (T) regions of the second leaves of plants grown in control (grey background) or −K (white background) media. To build the profiles, measured values were classified into two levels (+, −) for whole shoots or into seven levels for leaf segments, ranging from much lower (−−−) to much higher (+++) than the median (0) across all samples (see scale bar). If amounts differed between adjacent segments, a continuous gradient within the segments was assumed

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article.      inhibits the fungus (data in Fig. 6). JA-induction by low tissue K therefore protects barley against B. graminis (data in Fig. 5). Other physiological factors accompanying low K may weaken the plant's resistance against B. graminis (white arrow, literature) but the protective effect of JA outweighs these factors. B: JA is not induced by R. commune, and R. commune is insensitive to JA (data in Fig.6). The low-K induced rise of JA has therefore no effect on R.commune.