Resistance of barley to Fusarium graminearum was studied using a pair each of resistant and susceptible black and yellow barley lines. The spikelets were inoculated with a trichothecene-producing isolate, a trichothecene-nonproducing isolate (tri5−), or a mock solution. Spikelets were collected 72 h after inoculation and metabolites were analysed using a LC-hybrid MS system. Metabolite abundances were used to identify the constitutive (RRC) and induced resistance-related metabolites (RRI). The pathogen virulence factor, DON, and its plant detoxification product, DON-3-O-glucoside (D3G), were also identified and designated as resistance-indicator (RI) metabolites. The RRC, RRI and RI metabolites were putatively identified. Jasmonic acid was significantly induced in barley following inoculation with a trichothecene-producing isolate, but not with a tri5− isolate. The former isolate reduced the induction of both the number and amount of RR metabolites. The metabolites cinnamic acid, sinapoyl alcohol, coniferin, catechin and naringin were identified only in response to the inoculation with a tri5− mutant. The abundances of p-coumaric acid, coniferaldehyde and sinapaldehyde increased more in response to the tri5− mutant than to the trichothecene-producing isolate. The total amount of DON synthesized and its conversion to D3G varied greatly between the resistant and susceptible black barley, but not in yellow barley. Interestingly, an increase in the amount of total DON produced was associated with a decrease in the conversion of DON to D3G. The roles of RRC, RRI and RI metabolites in plant defence and their further use as potential biomarkers in screening are discussed.
Fusarium head blight (FHB), caused by Fusarium graminearum (teleomorph: Gibberella zeae) is a serious threat to wheat and barley production. It reduces not only yield but also grain quality by producing mycotoxins that are hazardous to livestock and humans (Gilbert & Tekauz, 2000). Toxins produced by Fusarium spp. are a family of sesquiterpenoid trichothecenes, the major ones being deoxynivalenol (DON), 3-acetyl-4-deoxynivalenol (3ADON) and nivalenol (NIV). DON is known to be a virulence factor (Proctor et al., 1995).
The tri5 gene of F. graminearum codes for the enzyme trichodiene synthase, which catalyses the first step in the trichothecene biosynthetic pathway (Proctor et al., 1995). The tri5− mutant GZT40 was developed by disrupting the tri5 gene of the trichothecene-producing F. graminearum isolate GZ3639 (Proctor et al., 1995). Because F. graminearum has a haploid genome, disruption of the tri5 gene blocks the first step of the trichothecene biosynthetic pathway, leading to complete inhibition of trichothecene production. The tri5− mutant failed to spread within the inoculated spike in wheat (Jansen et al., 2005).
Resistance to FHB in wheat and barley is evaluated mainly based on type I, resistance to initial infection, and type II, resistance to disease spread within the spike (Schroeder & Christensen, 1963). Other types are: type III, resistance to kernel infection; type IV, tolerance; and type V, resistance to toxins by detoxification (Mesterhazy, 1995). Type II resistance is very high in barley, and thus, the FHB research in barley is mainly focused on type I resistance (Bai & Shaner, 2004). Coloured barley is considered to have more resistance than yellow barley, but among coloured barley, genotypes range from highly resistant to highly susceptible to FHB (Choo, 2006). The ranking of genotypes based on types of resistance has been inconsistent over locations and years (Kolb et al., 2001).
The resistance in wheat and barley to FHB is quantitative (Bai & Shaner, 2004). More than 100 FHB resistance quantitative trait loci (QTL) have been identified in wheat (Buerstmayr et al., 2009) and barley (Choo, 2006), but only the function of the QTL at chromosome 3BS is partially known, which involves the enzymatic detoxification of DON by the plant to D3G (Lemmens et al., 2005). The resistance mechanisms in barley and wheat have been explored through transcriptomics (Boddu et al., 2006; Golkari et al., 2007), proteomics (Zhou et al., 2006; Geddes et al., 2008) and metabolomics (Hamzehzarghani et al., 2008; Bollina et al., 2010; Kumaraswamy et al., 2011). Previous metabolomics studies, on type I resistance in barley, detected several constitutive resistance-related metabolites, but only a few induced resistance-related metabolites, following pathogen inoculation (Bollina et al., 2010; Kumaraswamy et al., 2011). However, these studies used trichothecene-producing F. graminearum isolates. DON, a trichothecene, is known to inhibit protein synthesis in eukaryotic cells (Rocha et al., 2005). Thus, it is hypothesized that it might also inhibit host enzymes that induce resistance-related metabolites. In addition, the previous studies involved only one resistant genotype (Bollina et al., 2010; Kumaraswamy et al., 2011), and an association of these resistance-related metabolites with several resistant genotypes can help in the selection of potential biomarkers for resistance. Accordingly, the objective of this study was to compare the metabolic profiles of barley genotypes varying in colour and resistance to FHB, following inoculation with trichothecene-producing and -nonproducing isolates of F. graminearum. Mechanisms of resistance to trichothecene-nonproducing isolates are useful since the evolution of different chemotypes has been reported under commercial wheat and barley production conditions (Carter et al., 2002).
Materials and methods
Two pairs of FHB-resistant and FHB-susceptible recombinant inbred lines of black and yellow barley, derived from a cross between AC Legend (yellow) and CH9403-2 (black), were used in this study (Choo, 2006). The lines used were: black = B32-27 (resistant) and B32-11 (susceptible), and yellow = Y32-57 (resistant) and Y32-7 (susceptible). Plants were grown in the greenhouse at 22 ± 3°C. Three plants, each with one additional tiller, were maintained per pot. Plants were irrigated daily and fertilized at 2-week intervals with 200 mL 0·3% PlantProd (20-20-20 NPK + trace elements).
Pathogen production, inoculation and incubation
A trichothecene-producing F. graminearum isolate, 15–35, and a trichothecene-nonproducing tri5− mutant, GZT40 (obtained from Dr R. H. Proctor; Proctor et al., 1995), were cultured on rye agar media. Seven-day-old cultures were used to collect macroconidia. Ten barley spikelets in the middle of the spike at 50% anthesis to early milky stage (GS = 65–73; Zadoks et al., 1974), were individually inoculated with 10 μL spore suspension, containing 1 × 105 macroconidia mL−1 of either the trichothecene-producing isolate or the tri5− mutant, or a mock solution containing 0·02% aqueous solution of Tween80, using a syringe equipped with a dispenser (Hamilton, model: PB600-1). The inoculated plants were placed inside plastic bags, the inner surfaces of which had been sprayed with sterile water, to facilitate initial establishment of the infection. The bags were removed after 48 h.
Experimental design and statistical analyses
The experiment was designed as a randomized complete block with 12 treatments, consisting of four genotypes and three inoculations of the trichothecene-producing isolate, the trichothecene-nonproducing isolate, or the mock solution, with five replicates in time, conducted at about weekly intervals. Each experimental unit consisted of a pooled sample of 60 inoculated spikelets collected from six spikes.
Disease severity assessment
For the disease severity analysis, spikelets were spray-inoculated with a spore suspension of the trichothecene-producing isolate of F. graminearum, using an air-brush sprayer (Badger Air-Brush Co., model: 200·3™, deluxe set). Disease severity was assessed as the number of diseased spikelets among the middle 10 in a spike at 21 days post-inoculation, from which the proportion of diseased spikelets (PSD) was determined (Hamzehzarghani et al., 2008). This experiment was designed as a randomized complete block with three replicates in time. The experimental units consisted of 30 spikes per replicate.
The individually inoculated spikelets were collected 72 h post-inoculation. The rachilla node was cut off and the remainder was frozen immediately in liquid nitrogen and stored at −80°C. Metabolites were extracted as previously reported (Kumaraswamy et al., 2011). In brief, 100 mg (fresh weight) of homogenized sample in liquid nitrogen was placed in a 2·2-mL microcentrifuge tube to which 400 μL ice-cold methanol (HPLC grade 99·96% pure, Fisher Scientific), 200 pg genistein, 200 pg ribitol and 330 μL water were added. After vortex mixing for 10 s, the samples were sonicated for 15 min at a frequency of 40 kHz in a water bath at room temperature. The tubes were then centrifuged for 10 min at 20 000 g. The supernatants were filtered through a 0·1-μm PVDF membrane filter (Millipore Corporation) and the filtrate was collected and stored at −20°C until analysis.
Metabolite analysis using LC-ESI-LTQ-Orbitrap
Metabolites were analysed using an LC-hybrid MS system (LC-ESI-LTQ-Orbitrap, Thermo Fisher). A relatively polar C18 Kinetex column (210 μm internal diameter × 10 cm, with a particle size of 2·6 μm, Phenomenex) was used. The column was maintained at 25°C and the mobile phase was adjusted to a flow rate of 150 μL min−1 and eluted with 2·5 mm ammonium acetate in water (A) and methanol (B). A gradient elution profile was used starting with 98% mobile phase A for 4 min then shifting to 10% mobile phase A for 26 min and subsequently ramped to 90% mobile phase B for 5 min, followed by a 5-min ramp to 100% mobile phase B and a rapid return to 100% mobile phase A. All samples were first analysed in MS1 mode, and later, two samples were separately analysed in MS/MS mode, using a normalized collision induced dissociation (CID) energy of 35 eV, to determine the fragmentation patterns. A blank sample of only solvents was run after each sample.
LC/MS output analysis
The LC-hybrid MS output for all experimental units within black or yellow genotypes were separately aligned using XCMS 1·22·1 (Kumaraswamy et al., 2011). Data were processed using a signal to noise threshold (snthresh) of 5:1 and a band width (bw) of 10 s. The bioinformatics tool, CAMERA, linked to XCMS, was used to annotate peaks with isotopes, neutral losses, adducts and dimers, which represented multiple peaks of the same compound (Tautenhahn et al., 2007).
The abundances (ion counts) of peaks from different treatments were subjected to pairwise comparison based on Student’s t-test (sas v. 9·2; SAS Institute Inc.), to identify metabolites with significant treatment effects. The tests of hypotheses included four basic pairs of comparisons: RM vs. SM, RP vs. RM, SP vs. SM, RP vs. SP, where R = resistant, S = susceptible, M = mock and P = pathogen, either the trichothecene-producing or the trichothecene-nonproducing isolate. The black and yellow genotypes were separately analysed.
Peaks with significant treatment effects, based on a t-test, were used to identify the constitutive resistance-related (RRC) or induced resistance-related (RRI) metabolites (Hamzehzarghani et al., 2008). A metabolite whose abundance was greater in a resistant genotype than in a susceptible genotype was considered an RR metabolite. An RR metabolite found following mock inoculation was considered a constitutive metabolite (RRC = RM > SM, where R = resistant, S = susceptible and M = mock). A metabolite whose abundance in a resistant genotype inoculated with pathogen was significantly greater following both inoculation with mock solution and also in a susceptible genotype following inoculation with a pathogen was considered an induced RR metabolite (RRI = RP > RM and RP > SP, where P = pathogen, either the trichothecene-producing or -nonproducing isolate of F. graminearum).
The RR metabolites were assigned with putative names, based on accurate mass and fragmentation pattern, either matching with the spectra in databases or with those obtained by spiking authentic standards (Bollina et al., 2010; Kumaraswamy et al., 2011). Attempts were also made to manually verify the fragmentation patterns using ChemSketch of IntelliXtract and to calculate the number of carbons in the formula based on isotope ratio. The change in abundance in the FHB-resistant genotype, relative to its corresponding FHB-susceptible genotype, was calculated as RM/SM and RP/SP (if significantly induced in a resistant genotype, RP > RM) for constitutive and induced metabolites, respectively. DON is a virulence factor, and accordingly the amounts of DON and its detoxification product, D3G, in a resistant genotype relative to a susceptible one were considered as resistance-indicator (RI) metabolites. DON and D3G (Sigma-Aldrich) were spiked at different concentrations, following the above LC-hybrid MS protocol, and regression models were developed to predict the concentrations from abundances of their pure forms, rather than their acetate adducts. The amount of total DON produced was calculated as: TDP = DON + D3G; the proportion of TDP converted to D3G was calculated as: PDC = D3G/TDP.
The FHB severity significantly varied among genotypes, and PSD was 0·12 and 0·44 in the resistant (B32-27) and susceptible (B32-11) black barley, respectively. PSD were 0·15 and 0·53 in the resistant (Y32-57) and susceptible (Y32-7) yellow barley, respectively. PSD under field conditions was 0·14 for B32-27 and 0·28 for B32-11 for black barley, and 0·12 for Y32-57 and 0·66 for Y32-7 for yellow barley.
Metabolic profiles of black barley
In treatment combinations involving inoculations with the trichothecene-producing isolate or the mock solution, a total of 735 metabolites had significant treatment effects, of which 154 were RRC metabolites and 49 were RRI metabolites (Table 1). In treatment combinations involving inoculations with the trichothecene-nonproducing isolate or the mock solution, a total of 618 metabolites had significant treatment effects, of which 154 metabolites were RRC, and 73 were RRI.
Table 1. Metabolic profiles (numbers of metabolites) of black and yellow barley genotypes following inoculation with a mock solution, a trichothecene-producing (tri5+) isolate or a trichothecene-nonproducing (tri5−) isolate of Fusarium graminearum
Black barley (R = B32-27; S = B32-11)
Yellow barley (R = Y32-57; S = Y32-7)
Treatment-significant metabolites: RM vs. SM, RP vs. RM, SP vs. SM, RP vs. SP (note: since the data for the trichothecene-producing isolate and the tri5− mutant were separately aligned, the mock inoculation is common for both; the RRC and RRI metabolites were derived from these combinations); RRC: constitutive resistance-related metabolite = RM > SM; RRI: induced resistance-related metabolite = RP > RM and also RP > SP, where R: resistant, S: susceptible, M: mock and P: pathogen, either a trichothecene-producing or -nonproducing isolate (tri5−) of F. graminearum.
Metabolic profiles of yellow barley
In treatment combinations consisting of inoculation with the trichothecene-producing isolate or the mock solution, a total of 608 metabolites had significant treatment effects, of which 416 were RRC metabolites and 39 were RRI metabolites (Table 1). In treatment combinations consisting of inoculation with the trichothecene-nonproducing isolate or the mock solution, a total of 789 metabolites had significant treatment effects, of which 416 were RRC and 102 were RRI metabolites.
Constitutive resistance-related metabolites (RRC)
RRC metabolites in black barley
Among the 154 RRC metabolites selected in resistant black barley (B32-27), 35 were putatively identified (Tables 2 & S1). The metabolites with a large change in abundance in the resistant vs. the susceptible genotype were: catechol, oxyquinoline, decadienoic acid, coniferaldehyde, cyano-4-hydroxycinnamic acid, naringenin, heptadecenedioic acid, naringenin-7-O-glucoside and eucommin A.
Table 2. Resistance-related constitutive (RRC) metabolites, and their fold change in abundancea in fusarium head blight (FHB)-resistant black and yellow barley genotypes relative to susceptible ones
Observed mass + Hb
aThe fold change in abundance in the resistant genotype relative to the susceptible one (RM/SM, where R: resistant, S: susceptible and M: mock); fold changes ≥4 are indicated in bold; RRC: resistance-related constitutive metabolite; metabolites not significantly affected are left blank; the significance levels of RM > SM: *P < 0·05; **P < 0·01; ***P < 0·001.
bObserved mass: observed monoisotopic mass (M) in negative mode (M–H)− with the mass of hydrogen (H) = +1·007276 added, to facilitate comparison with the theoretical mass of the compound. Details on compound identification such as observed mass and accurate mass error (AME) are given in Table S1.
Among the 416 RRC metabolites selected in resistant yellow barley (Y32-57), only 21 were putatively identified (Table 2). The metabolites with a large change in abundance in the resistant genotype relative to the susceptible one were: guaiacol, phenylalanine, myo-inositol, decenedioic acid and 5-methoxyfuranocoumarin. Among 21 metabolites identified, only five were common to both black and yellow barley.
RRC metabolites discriminating black and yellow barley
Among the 35 RRC metabolites identified, only five were common to both black and yellow barley. Ten metabolites were more than fourfold higher in abundance in the FHB-resistant black vs. the susceptible black genotype (Table 2, shown in bold). Of these, the abundances of coniferaldehyde, vitexin and 8E-heptadecenedioic acid were sixfold greater in the resistant black genotype than in the susceptible black one. There were no metabolites with greater than fourfold differences in abundance between the resistant and susceptible yellow barley genotypes.
Induced resistance-related metabolites following inoculation with a trichothecene-producing isolate (RRITP)
RRITP metabolites in black barley
Among 49 selected RR metabolites induced in response to the trichothecene-producing isolate in resistant black barley (B32-27), only 12 were putatively identified (Tables 3 & S2). A signal molecule, jasmonic acid, was induced 3·1-fold more in the resistant genotype than in the susceptible one.
Table 3. Resistance-related induced (RRI) metabolites and their fold change in abundancesa in fusarium head blight (FHB)-resistant black and yellow barley genotypes relative to susceptible ones, inoculated with a trichothecene-producing isolate or a tri5− mutant of Fusarium graminearum
Observed mass + Hb
aThe fold change in abundance in the resistant genotypes relative to the respective susceptible ones was calculated based on RP/SP, only when the metabolite was pathogenesis-related in a resistant genotype (PRr) = RP > RM, where R: resistant, S: susceptible, M: mock and P: pathogen, either a trichothecene-producing or -nonproducing isolate of F. graminearum; the significance levels of RP > SP were based on: *P < 0·05; **P < 0·01; ***P < 0·001; fold changes ≥4 are shown in bold; metabolites not significantly affected are left blank.
bObserved mass: observed monoisotopic mass (M) in negative mode (M–H)− with the mass of hydrogen (H) = +1·007276 added, to facilitate comparison with the theoretical mass of the compound. Details on compound identification such as observed mass and accurate mass error (AME) are given in Table S1.
Among the 39 selected RR metabolites induced in response to the trichothecene-producing isolate in the resistant yellow barley (Y32-57), 17 were putatively identified (Table 3). Out of these, four were significant only in the resistant yellow barley: oxyquinoline, p-coumaryl alcohol, phenyl beta-d-glucopyranoside and naringenin-7-O-glucoside. Jasmonic acid was induced 4·1-fold more in the resistant than in the susceptible genotype.
RRITP metabolites discriminating black and yellow barley
Three metabolites were induced more than fourfold more in the resistant black and yellow barley genotypes than in the respective susceptible genotypes following inoculation with the trichothecene-producing isolate (Table 3, shown in bold).
Induced resistance-related metabolites following inoculation with a trichothecene-nonproducing (tri5−) isolate (RRITN)
RRITN metabolites in black barley
Among the 73 RR metabolites induced in response to the tri5− mutant in the resistant black barley (B32-27), 26 were putatively identified, of which 15 were induced in response to inoculation with the tri5− mutant, but not in response to the trichothecene-producing isolate (Table 3): trans-2-hexenal, cinnamic acid, 9-oxo-2E-decenoic acid, sinapoyl alcohol, 5,8-heptadecadiynoic acid, 2′,4′-dihydroxy-6′-methoxychalcone, 4-hydroxy palmitic acid, catechin, quercetin, dihydroquercetin, geranylchalconaringenin, keto stearic acid, dihydroxylinoleic acid, coniferin and naringin. Interestingly, the signal molecule jasmonic acid was not significantly induced in that case, but was significantly induced in response to inoculation with the trichothecene-producing isolate.
RRITN metabolites in yellow barley
Among the 102 RR metabolites induced in response to the tri5− mutant in the resistant yellow barley (Y32-57), 30 metabolites were putatively identified, of which 18 were uniquely induced in response to inoculation with the tri5− mutant (Table 3): trans-2-hexenal, leucine, adenine, octenoic acid, glutamine, cinnamic acid, ribulose, decadienoic acid, mannitol, oxo-decanoic acid, methoxyindoleacetate, sinapoyl alcohol, tetradecenoic acid, macrophylline, heptadecenedioic acid, keto stearic acid, dihydroxylinoleic acid and coniferin. In addition, similar to black barley, jasmonic acid was not significantly induced in yellow barley following inoculation with the tri5− mutant.
RRITN metabolites discriminating black and yellow barley
Following inoculation with the trichothecene-nonproducing isolate, nine metabolites were induced fourfold more in the resistant vs. the susceptible black barley, compared with six metabolites in the resistant vs. the susceptible yellow barley (Table 3, in bold).
Differentially induced resistance-related metabolites by trichothecene-nonproducing (tri5−) and trichothecene-producing isolates (RRI)
Among the 27 induced RR metabolites identified in black barley, 15 were significantly induced only following inoculation with the tri5− mutant but not with the trichothecene-producing isolate (Table 3). Metabolites with large changes in abundance were: cinnamic acid (3·5×), sinapoyl alcohol (3·2×), dihydroxylinoleic acid (12·7×), geranylchalconaringenin (8·3×), dihydroquercetin (3·7×), heptadecatrienoic acid (8·7×) and naringin (6·2×). Only jasmonic acid was induced (3·1×) following inoculation with the trichothecene-producing isolate. In the resistant yellow barley, among the 35 RRI metabolites identified, 26 were either induced only following inoculation with the tri5− mutant, or induced in higher amounts following inoculation with the tri5− mutant than with the trichothecene-producing isolate. Metabolites with large changes in yellow barley were: p-coumaric acid (4·1×), coniferaldehyde (9·7×), 3-indoleacetic acid (7·7×), 5-hydroxyindoleacetic acid (4·0×) and methoxyindoleacetate (5·8×). Even though 10 metabolites were induced more by the trichothecene-producing isolate than by the tri5− mutant, the magnitude of induction was greater than fourfold only for coumarin (7·2×), coniferaldehyde (4·6×) and jasmonic acid (4·1×) (Table 3).
Resistance-indicator metabolites produced by a trichothecene-producing isolate (RITP)
The resistance-indicator metabolites, such as DON and D3G, were not detected in spikelets inoculated with the tri5− mutant or the mock solution. In contrast, DON and its detoxification product, D3G, were detected, both in their pure forms and also as their acetate adducts, in all four genotypes inoculated with the trichothecene-producing isolate. The concentrations of both DON and D3G were derived from the previously established regression models by substituting their abundances.
Total DON produced (TDP)
The resistant black barley had significantly lower amounts of TDP (34 mg kg−1) than the susceptible genotype, while the resistant yellow barley, in spite of its resistance to FHB, had greater amounts of TDP (72 mg kg−1) than the susceptible genotype (Fig. 1).
Proportion of DON converted to D3G (PDC)
All four genotypes studied here converted DON to D3G. The PDC was higher in the resistant (0·75) than in the susceptible (0·66) black barley, but this was reversed in yellow barley (resistant = 0·73; susceptible = 0·76) (Fig. 2). Interestingly, across genotypes, a decrease in the amount of TDP was associated with an increase in the PDC, and a linear regression model explained 48% of the variance (Fig. 3).
In both black and yellow barley genotypes not only was a greater number of RR metabolites induced, but also their abundance was greater following inoculation with the tri5− mutant than with the trichothecene-producing isolate. Inoculation with the tri5− mutant induced greater abundances of lignin precursor metabolites, such as p-coumaric acid (which was increased 10·5-fold in black barley), than inoculation with the trichothecene-producing isolate. Interestingly, in black barley the magnitude of increase was higher for syringyl lignin precursor metabolites such as sinapaldehyde (7·2×) and sinapoyl alcohol (3·2×), while in yellow barley it was greater for guaiacyl lignin precursors such as coniferaldehyde (9·7×) (Fig. 3). In black barley, several flavonoids, including naringenin and geranylchalconaringenin, which are precursors of kaempferol, were greatly induced following inoculation with the tri5− mutant. These findings clearly indicate that the trichothecene-producing isolate inhibits the induction of RR metabolites in the host plant. Furthermore, lower total DON synthesis was associated with a greater proportion of conversion of DON to D3G, meaning the enzymatic conversion of DON to D3G by the plant was inhibited by trichothecene/DON. DON is a virulence factor and is known to inhibit protein synthesis in eukaryotic cells (Rocha et al., 2005). This warrants the evaluation of wheat and barley genotypes using a trichothecene-nonproducing isolate, and also other mutants lacking specific toxins, in order not to lose any good genetic material during screening. Chemotype occurrence in commercial fields has been reported (Carter et al., 2002). Contrarily, inoculation with DON also induced several phenylpropanoids and fatty acids (Paranidharan et al., 2008). It is possible that DON can induce or suppress induction of RR metabolites in plants depending on its concentration. In the present study, the signal molecule jasmonic acid was significantly induced by the trichothecene-producing but not by the -nonproducing isolate of F. graminearum. Even though a previous study reported induction of jasmonic acid following pathogen inoculation (Kumaraswamy et al., 2011), this is the first study to report its possible induction by trichothecenes or DON.
In black barley, two anthocyanins, pelargonidin 3-O-rutinoside and pelargonidin 3-O-sophoroside, were identified as RRC metabolites, with abundances up to fivefold greater in the resistant than in the susceptible genotype. Pelargonidin has been previously reported from black barley (Kim et al., 2007). In black-, blue- and purple-coloured barley, myricetin was the main flavonoid, followed by catechin and quercetin (Kim et al., 2007), but the present study, using spikelets, detected only the latter two, not myricetin. Thus, the metabolites related to colour were also among several of the RR metabolites associated with resistance to FHB.
This study detected several RRI metabolites following pathogen inoculation, whereas previous studies detected only a few induced RR metabolites (Bollina et al., 2010; Kumaraswamy et al., 2011). This is mainly because the previous studies used spray inoculation (as opposed to individual spikelet inoculation), which resulted in high variation among replicates in the number of infected spikelets. In addition, the use of a Kinetex column enabled detection of trichothecenes, which were not well detected in the previous study using the Jupiter column (Bollina et al., 2010). The use of trichothecene-producing and -nonproducing isolates of F. graminearum has enabled the inducer of jasmonic acid, possibly DON, to be discovered. Even though a trichothecene-producing isolate, 15-35, was used to allow the identification of RR metabolites as before, the use of the wildtype (GZ3639) from which the tri5− mutant (GZT40) was derived (Proctor et al., 1995) would better reveal the metabolites differentially induced by the isolates. Furthermore, inoculation with a mutant lacking only in DON production or inoculation of DON alone is needed to prove the role of DON in jasmonic acid induction. The use of F. graminearum mutants lacking different specific pathogenicity factors can lead to the identification of other mechanisms of resistance. Similarly, the use of plant mutants lacking specific genes responsible for the production of different RR metabolites identified here would also help to determine the mechanisms of resistance. However, since the effect of a single metabolite on disease severity is generally small, proving its effect would be very challenging. This study used recombinant inbred populations, so not all the RR metabolites detected in the parents will be present in high abundances in the resistant recombinant inbred lines selected from the population. Comparative metabolomics of different cultivars varying in resistance should detect more common RR metabolites that can be used as biomarkers.
Several of the RR metabolites reported in previous studies were also identified here as RR metabolites, demonstrating consistency in the occurrence of these metabolites across FHB-resistant genotypes. When some were not selected again as RR metabolites, those in fact were related to each other in metabolic pathways. Accordingly, the biomarkers are discussed here as groups of metabolites belonging to specific metabolic pathways. The following RRC and RRI metabolites were selected as biomarkers (Tables 4 & S3) based on five criteria: (i) putative identification of metabolite; (ii) statistical significance; (iii) relatively high change in abundance in the FHB-resistant genotype relative to the susceptible one; (iv) occurrence in different FHB-resistant genotypes; and (v) with known plant defence mechanisms. Several other metabolites are also potential candidates, and further studies are needed to prove their relevance.
Table 4. Potential biomarker metabolitesa and their fold changes in abundance in resistant black and yellow barley genotypes relative to susceptible ones following inoculation with a mock solution, a trichothecene-producing isolate or a trichothecene-nonproducing isolate of Fusarium graminearum
Black barley (R = B32-27)
Yellow barley (R = Y32-57)
aMass details used to identity these metabolites are given in Table S1. Relative fold changes ≥4 are indicated in bold; the fold change of induced resistance-related (RRI) metabolites is based on RP/SP, when RP > RM; significance levels for RP > SP are: *P < 0·05; **P < 0·01; ***P < 0·001.
bAll the biomarkers presented here were identified as RRI metabolites, except for naringenin in B32-27 and phenylalanine in Y32-57, which were RRC (resistance-related constitutive) metabolites.
Jasmonic acid was induced in both the resistant black (4·1×) and yellow (3·1×) barley following inoculation with the trichothecene-producing isolate of F. graminearum. Jasmonic acid was not only detected as a pathogenesis-related metabolite, but also significantly reduced F. graminearum biomass in vitro (Kumaraswamy et al., 2011). Jasmonic acid was also induced against another necrotrophic pathogen, Alternaria brassicicola, in Arabidopsis (Glazebrook, 2005). The present results show that trichothecenes, possibly DON, induce jasmonic acid in barley, which in turn activates the DON-detoxification enzymes, glucosyltransferases, converting DON to less toxic D3G. External applications of DON and jasmonic acid have been shown to weakly induce glucosyltransferase in Arabidopsis thaliana (Poppenberger et al., 2003).
Phenylalanine was identified here as an RRC metabolite in yellow barley and as an RRI metabolite in black barley. Phenylalanine was also reported as an RRC metabolite in another study with six barley genotypes, against FHB (Kushalappa et al., 2010). Phenylalanine is a major precursor of the phenylpropanoid-flavonoid metabolic pathway. Two transcript-derived fragments, TDF83 and TDF17, that code for phenylalanine ammonia-lyase (PAL), were strongly up-regulated in FHB-resistant genotypes following Fusarium inoculation (Steiner et al., 2009).
p-Coumaric acid was induced several-fold more in the resistant black and yellow barley genotypes than in the respective susceptible genotypes following inoculation with either of the isolates, but the induction was greater with the tri5− mutant (10·5×). p-Coumaric acid was identified as a pathogenesis-related metabolite in a resistant genotype (PRr) and also as an RRC metabolite in previous studies, and it also significantly reduced the biomass of F. graminearum (Bollina et al., 2010; Kumaraswamy et al., 2011). Being an antioxidant, it can also act as potential inhibitor of DON synthesis (Boutigny et al., 2009).
Lignin is synthesized in the phenylpropanoid metabolic pathway (Naoumkina et al., 2010). Sinapoyl alcohol and sinapaldehyde, both precursors of syringyl lignin (Fig. 4), were induced in the resistant black barley by both the isolates, but the induction was greater with the tri5− mutant (7·2×). Caffeyl aldehyde and coniferaldehyde, the precursors of guaiacyl lignin, were induced in resistant yellow barley by both the isolates, the induction being greater following inoculation with the tri5− mutant (3·7- and 9·7-fold, respectively). Deposition of lignin has been observed following F. graminearum inoculation in resistant wheat genotypes (Siranidou et al., 2002).
Catechin is one of the most abundant flavonoids present in barley, and in this study it was induced in the resistant black barley by the tri5− mutant. Catechin was induced in resistant naked barley seeds against F. graminearum (Eggert et al., 2010). Naringenin and several of its conjugates were identified in the present study as both RRC and RRI metabolites. Interestingly, two metabolites: geranylchalconaringenin (8·3×), a conjugate of naringenin, and naringin (6·2×), an advanced synthesis product of naringenin, were greatly induced in resistant genotypes following inoculation with the tri5− mutant. Both naringenin (Bollina et al., 2010; Kumaraswamy et al., 2011) and catechin have been identified as RRC metabolites in the FHB-resistant barley genotypes (Kumaraswamy et al., 2011). Naringenin significantly reduced the biomass of F. graminearum in vitro (Bollina et al., 2010).
Barley can resist DON in two ways: (i) by reducing the amount of DON synthesis by producing antioxidants (Boutigny et al., 2009); or (ii) by detoxifying the already synthesized DON to D3G (Bollina et al., 2010), through the enzyme DON-3-O-glucosyltransferase (Lemmens et al., 2005). In the present study, only the resistant black barley genotype, not the resistant yellow barley genotype, inhibited the synthesis of DON more that the susceptible genotypes. Inhibition of trichothecene/DON synthesis by phenolics, through their antioxidant properties, has been demonstrated (Boutigny et al., 2009). In the present study the black barley produced several potential antioxidants, especially the constitutive flavonoids, which can inhibit trichothecene synthesis. The reason why the lowest amount of DON synthesis was observed in the susceptible yellow barley has yet to be explored. The detoxification of DON to D3G was greatest in the resistant black barley, but was only moderate in the resistant yellow barley. Interestingly, across genotypes, a decrease in TDP was associated with an increase in the amount of DON detoxification to D3G. It is possible that in the resistant yellow barley genotype the high levels of DON synthesized may have reduced enzymatic activity, leading to a moderate level of DON conversion to D3G. Accordingly, these two metabolites should be considered together in assessing the reduced accumulation of DON in resistant genotypes.
This study reports several novel RRC, RRI and RI metabolites. More metabolites were induced following inoculation with the trichothecene-nonproducing isolate than with the trichothecene-producing isolate. Considering the occurrence of chemotypes of this pathogen under commercial conditions (Carter et al., 2002), it is crucial that screening programmes based on metabolic profiling also include RR metabolites induced against different chemotypes of the pathogen. Numerous studies have shown that FHB resistance is quantitatively inherited and more than 100 QTLs are associated with resistance to FHB in wheat and barley (Choo, 2006; Buerstmayr et al., 2009). Often, these potential RR metabolite biomarkers are lost during breeding because of a lack of tools to detect the metabolites responsible for the quantitative resistance. Metabolomics technology, as proposed here, can help in the incorporation of several mechanisms of resistance, thus increasing genetic diversity in crop production. Though the extraction and analytical protocols used here did not detect all metabolites present in the genotypes tested, several phenylpropanoids and lipids that are related to resistance were identified, some of which can be used as biomarkers. More similar studies, and also on proteomics, would further elucidate the role of biochemicals in resistance to FHB in the Triticeae. Plant defence mechanisms involving most of these biochemicals are well documented in the literature (Vidhyasekaran, 2008). These metabolites, and also proteins, can be exploited for barley crop improvement, through metabolic marker-assisted selection or through metabolic pathway engineering to enhance their abundances.
This project was funded by the Ministère de l’Agriculture, des Pêcheries et de l’Alimentation du Québec (MAPAQ), Centre de Recherche sur les Grains Inc. (CEROM), and the Fédération des Producteurs de Porc du Québec (FPPQ), Québec, Canada. We thank Dr R. H. Proctor, Mycotoxin Research Unit, USDA, ARS, USA, for providing F. graminearum isolate GZT40.