Verticillium longisporum is a soil-borne vascular pathogen causing economic loss in rape. Using the model plant Arabidopsis this study analyzed metabolic changes upon fungal infection in order to identify possible defense strategies of Brassicaceae against this fungus.
Metabolite fingerprinting identified infection-induced metabolites derived from the phenylpropanoid pathway. Targeted analysis confirmed the accumulation of sinapoyl glucosides, coniferin, syringin and lignans in leaves from early stages of infection on. At later stages, the amounts of amino acids increased.
To test the contribution of the phenylpropanoid pathway, mutants in the pathway were analyzed. The sinapate-deficient mutant fah1-2 showed stronger infection symptoms than wild-type plants, which is most likely due to the lack of sinapoyl esters. Moreover, the coniferin accumulating transgenic plant UGT72E2-OE was less susceptible. Consistently, sinapoyl glucose, coniferyl alcohol and coniferin inhibited fungal growth and melanization in vitro, whereas sinapyl alcohol and syringin did not. The amount of lignin was not significantly altered supporting the notion that soluble derivatives of the phenylpropanoid pathway contribute to defense.
These data show that soluble phenylpropanoids are important for the defense response of Arabidopsis against V. longisporum and that metabolite fingerprinting is a valuable tool to identify infection-relevant metabolic markers.
Verticillium longisporum is a soil-borne fungus which causes vascular disease primarily in Brassicaceae (Karapapa et al., 1997; Zeise & von Tiedemann, 2002; Fradin & Thomma, 2006). Triggered by plant root exudates, fungal hyphae develop from microsclerotia, penetrate the root epidermis, grow through the root cortex and enter the xylem elements from where V. longisporum spreads throughout the plant (Eynck et al., 2007). The life cycle is completed after formation of microsclerotia on senescent tissue (Wilhelm, 1955).
Verticillium longisporum differs in molecular and morphological traits from its close relatives V. dahliae and V. albo-atrum (Karapapa et al., 1997; Zeise & von Tiedemann, 2001): The microsclerotia of V. longisporum are elongated in contrast to the spherical morphology of those of V. dahliae and its conidia are nearly twice as long as those from other Verticillium species. The DNA content was estimated to be nearly 1.8 times that of V. dahliae, suggesting that V. longisporum is a ‘nearly diploid’ hybrid (Inderbitzin et al., 2011; Tran et al., 2013).
Whereas V. dahliae and V. albo-atrum are able to infect a broad variety of crop species, V. longisporum primarily colonizes members of the family of the Brassicaceae (Zeise & von Tiedemann, 2002). In contrast to V. dahliae and many other vascular pathogens, V. longisporum (isolate Vl43) does not induce wilting of the host plant (Reusche et al., 2012), which may result from only partial colonization of the vessel elements (Eynck et al., 2007). Etiology of the disease caused by V. longisporum is not fully understood; artificial infection of oilseed rape in field trials has not caused any yield loss, but it has a significant yield damage potential under conditions where the systemic spread of the pathogen is accelerated (Dunker et al., 2008). As fungicides are not effective against vascular diseases, it is important to understand the plant defense reactions (Rygulla et al., 2007; Klosterman et al., 2009).
Over the last years a number of publications have established Arabidopsis as a suitable host to study plant reactions against V. longisporum (Veronese et al., 2003; Johansson et al., 2006; Floerl et al., 2010, 2012; Haffner et al., 2010; Iven et al., 2012; Ralhan et al., 2012; Reusche et al., 2012; Tran et al., 2013). In Arabidopsis, infection leads to stunted growth, hyperplastic xylem formation and premature senescence.
Plant–pathogen interactions are generally characterized by a reprogramming in plant metabolism. In general, growth is arrested and defense mechanisms are activated. In Arabidopsis, several secondary metabolites with proven or potential antimicrobial effects are synthesized upon infection: indolic compounds including the phytoalexin camalexin, glucosinolates, phenylpropanoids, terpenes, benzenoids and fatty acid derivatives (D`Auria & Gershenzon, 2005). We have shown previously that secondary metabolites derived from the indole pathway are induced in Arabidopsis roots upon attempted entry of V. longisporum (Iven et al., 2012).
The aim of this study was to identify any metabolic patterns that are altered in Arabidopsis when V. longisporum has already colonized the xylem vessels of the aerial parts. Metabolite fingerprinting and subsequent targeted analysis identified c. 20 compounds that were derived from the phenylpropanoid pathway, as well as several amino acids. To test the contribution of the phenylpropanoid pathway on the interaction of Arabidopsis with V. longisporum, different mutants in the pathway were analyzed. Comparison of cell wall amount suggested that lignin is not a major factor in the defense response, but that soluble phenylpropanoids such as sinapoyl glucose, coniferyl alcohol and coniferin do directly inhibit fungal growth.
Materials and Methods
Plant growth and inoculation
Plant lines used were wild-type Arabidopsis thaliana (L.) Heynh ecotype Columbia 0 (Col 0), fah1-2 (Chapple et al., 1992), ref1-s (SALK_27911), sng1-1 (Lorenzen et al., 1996), C4H:F5H (Meyer et al., 1996), UGT72E1-OE, UGT72E2-OE and UGT73E1-OE (Lanot et al., 2006, 2008). The V. longisporum isolate Vl43 (Zeise & von Tiedemann, 2002) was obtained from A. von Tiedemann (Georg-August-University, Göttingen).
Plants were grown under short day conditions (8 h : 16 h, light : dark) at 22°C, 60% humidity and exposure to 130–150 μmol photons m−1 s−1.
For root cut inoculation, plants were grown on ½ Murashige and Skoog medium for 3 wk (Murashige & Skoog, 1962). Before inoculation, roots were cut 1 cm above the root tip. Plants were potted into soil cavities together with 10 ml V. longisporum-spore solution (1 × 106 ml−1) or with tap water (controls). Infection symptoms were documented up to 35 d post inoculation (dpi; Floerl et al., 2012).
For mutant analyses by root dip inoculation, plants were grown in sand : soil mixture (1 : 1) for 20 d. Roots were washed and incubated in spore suspension (4 × 105 ml−1) for 35 min or in tap water (controls) and then transferred into soil. Infection symptoms were documented for 21 dpi according to (Ralhan et al., 2012).
V. longisporum was grown in potato dextrose broth medium (Sigma-Aldrich, Deisenhofen, Germany) supplemented with streptomycin. Cultures were inoculated with spores and cultivated for 2 wk on a shaker. Three to four days before inoculation, the fungal biomass was transferred into Czapek Dox medium to induce spore production. Spores were harvested by draining the culture through miracloth followed by centrifugation and washed twice with sterile tap water.
Leaf area analysis
Pictures of each plant were taken from above and the projected leaf area was determined by the custom made software Bildanalyseprogramm 18.104.22.168 (datInf, Tübingen, Germany).
V. longisporum DNA analysis
Preparation and densitometric analysis of the DNA standard was performed according to (Brandfass & Karlovsky, 2006) and real-time PCR analysis was performed according to (Eynck et al., 2007).
Extraction with chloroform/methanol
For metabolite fingerprinting plant material was extracted using a two-phase extraction with chloroform and methanol. Each 80 mg sample of homogenized leaf material was mixed with 1 ml methanol and shaken for 10 min at 70°C. After centrifugation 1 ml of the supernatant was transferred into a glass vial, 1 ml ddH2O was added and the samples were stored at 4°C. The pellet was extracted with 500 μl chloroform for 10 min at 37°C. The mixture was filtered with a glass syringe through a filter (2-μm PTFE, Whatman 4 mm) and combined with the methanol/H2O extract. The mixed extracts were stored over night at 4°C to separate the different phases. Samples were centrifuged and 1.8 ml of the upper phase (methanol/H2O) was transferred into a new tube, dried and dissolved in 200 μl methanol. The samples were shaken for 5 min and again dried. The residue was solved in 10 μl methanol, 10 μl acetonitrile and 180 μl ddH2O.
The lower nonpolar phase (chloroform) of the two-phase solvent system was transferred into a reaction tube and dried under a stream of N2. The dried residue was solved in 30 μl methanol, 10 μl acetonitrile and 75 μl ddH2O.
Metabolite fingerprinting was performed with modifications as described in König et al. (2012). The analysis was performed twice for each sample by Ultra Performance Liquid Chromatography (UPLC) coupled with a photo diode array detector (PDA) and an orthogonal time-of-flight mass spectrometer (TOF-MS). For LC an ACQUITY UPLC BEH SHIELD RP18 column was used at a temperature of 40°C, a flow rate of 0.2 ml min−1 and with a binary gradient of solvent A (water/formic acid (100 : 0.1 (v/v)) and solvent B (acetonitrile/formic acid (100 : 0.1 (v/v)). The following gradient was applied for the analysis of the samples of the polar extraction phase: 0–0.5 min 10% solvent B, 0.5–3 min from 10% to 28% solvent B, 3–8 min from 28% up to 95% solvent B, 8–10 min 95% solvent B and 10–14 min 10% solvent B and for the analysis of the samples of the nonpolar extraction phase: 0–0.5 min 46% solvent B, 0.5–5.5 min 46–99% solvent B, 5.5–10 min 100% solvent B and 10–13 min 46% solvent B.
After electrospray ionization (ESI) in negative as well as positive modes, TOF-MS was operated in W optics to ensure a mass resolution larger than 10 000. Data were acquired by MassLynx 4.1 (Waters Corp., Milford, MA, USA) software in centroided format over a mass range of m/z 50–1200 (negative ionization mode) and m/z 85–1200 (positive ionization mode) with the scan duration of 0.5 s and an interscan delay of 0.1 s. The capillary and the cone voltage were maintained at 2700 V and 30 V, respectively, and the desolvation and source temperature at 350 and 80°C, respectively. Nitrogen was used as cone (30 l h−1) and desolvation gas (800 l h−1). The dynamic range enhancement (DRE) mode was used for data recording. All analyses were monitored by using Leucine-enkephaline ([M-H]− 554.2615 or [M+H]+ 556.2771 as well as its 13C isotopomer [M-H]− 555.2615 or its double 13C isotopomer [M+H]+ 558.2836 as lock spray reference compound.
The raw mass spectrometry data of all samples of one experiment were processed using the MarkerLynx Application Manager for MassLynx 4.1 software, which resulted in four data matrixes (one each for the polar extraction phase positively or negatively ionized, and for the nonpolar extraction phase positively or negatively ionized). The toolbox MarVis (MarkerVisualization, http://marvis.gobics.de) was used for further data processing including ranking and filtering, adduct identification and correction of the raw masses, and combining of the data matrixes, as well as for clustering and visualization. The MarVis-Suite toolbox includes the MarVis Filter and MarVis Cluster interfaces. First, a Kruskal–Wallis test in combination with false discovery rate (FDR) control was performed to select high-quality marker candidates with a FDR < 10−4. Next, an automated adduct correction has been applied according to the following rules: [M+H+]+, [M+Na+]+, [M+NH4+]+ for the positive and [M-H+]−, [M+CH2O2-H+]−, [M+CH2O2+Na+-2H−]− for the negative ionisation mode (Kaever et al., 2012). Afterwards, datasets of both ionisation modes could be combined and used for cluster analysis (Kaever et al., 2009). The exact corrected masses of the marker candidates were used for automated database search (mass window: 5 mDa) against the following databases: KEGG, LipidMaps, Aracyc, Knapsack and in-house-databases. The identity of selected markers was confirmed by co-elution with identical standards, quantitative RP-HPLC-DAD analysis and/or by MS2 experiments. For that, markers were analyzed by LC 1290 Infinity (Agilent Technologies, Santa Clara, CA, USA) coupled with a 6540 UHD Accurate-Mass Q-TOF-MS instrument with Agilent Jet Stream Technology as ESI source as described (Floerl et al., 2012). The compounds (+)-pinoresinol, pinoresinol diglucoside, (−)-matairesinol, (+)-lariciresinol, (−)-secoisolaricoresinol and sesamolin were obtained from PhytoLab (Vestenbergsgreuth, Germany).
Sinapate ester analysis
Plant material was two-times extracted with 80% methanol for 30 min at 60°C. Fluoroindole carboxyaldehyde was used as internal standard. Combined supernatants were dried and dissolved in 50% methanol for HPLC analysis.
Samples were separated on a Nucleosil 120-5 C-18 column (EC250/2) using the following gradient: solvent A: 0.1% acetic acid, solvent B 98% acetonitrile and 0.1% acetic acid; 0 min 100% A, 4 min 80% A, 8 min 76% A, 44 min 37% A, 46 min 0% A. Sinapate esters were detected at λ = 320 nm. Peaks were identified and quantified by commercially available standard substances.
Lignan and monolignol analysis
Homogenized plant tissue was two-times extracted with hot methanol and deoxyrhapontin was added as internal standard. Combined supernatants were dried under a stream of N2. β-glucosidase from almonds (1 mg ml−1) in Na-acetate buffer was added and the samples incubated at 37°C for 24 h. The mixture was extracted three times with ethylacetate, the supernatants dried and dissolved in 50% methanol for HPLC analysis. HPLC separation was done as described for sinapate ester analysis without the second step in the gradient. Light absorption of the eluate at 280 nm was monitored. Peaks assignment and analyte quantification was carried out with the help of commercial standards (Phytolab).
Cell wall isolation
Freeze-dried petioles of control and infected plants were milled to a fine powder in a ball mill (MM2; Retsch, Hanover, Germany) and incubated for 1 h with 10 ml 80% methanol. The pellet was washed once with 1 M NaCl/0.5% Triton X-100, two times with water, two times with methanol and two times with acetone. Each washing step lasted 30 min. The material was dried and used for lignin analysis.
Cell walls were extracted in acetyl bromide for the spectrophotometric lignin determination as described (Brinkmann et al., 2002).
Lignin composition was determined according to (Foster et al., 2010) with minor modifications: 15-mg cell wall material was incubated with 10% ethanethiol and 2.5% boron trifluoride diethyl etherate in dioxane at 100°C for 4 h. The reaction mixture was cooled down to room temperature and 600 μl 0.4 M bicarbonate, 2 ml water and 1 ml ethyl acetate were added. The upper phase was separated, the solvent was removed and the residue dissolved in ethyl acetate. For the derivatisation 20 μl sample was mixed with 5 μl BSTFA and 25 μl acetone. GC-MS analysis was performed as described in Foster et al. (2010).
Hypocotyls and petioles from control and inoculated plants were stored in FAE (3.7% formaldehyde, 5% acetic acid and 50% ethanol). For paraffin embedding the material was stepwise-dehydrated in a gradient of ethanol up to 100%. The ethanol was exchanged by Roti-Histol by stepwise increasing Roti-Histol up to 100%. For the embedding, the samples were placed into Roti-Histol saturated with paraplast at 42°C for 1 h and afterwards incubated in 100% paraplast at 60°C overnight. Samples were casted into forms of aluminum foil and thin hand sections were made. The cuts were dressed on a glass slide (Polysine) at 42°C and the remaining paraffin was removed by exposing the cuts to 100% Roti-Histol. The Roti-Histol was removed with ethanol and the staining was performed with 1% phloroglucin in 90% ethanol. Microscopy was performed in 10% hydrochloric acid with a BX51 microscope (Olympus, Hamburg, Germany).
RNA extraction and quantitative reverse transcription (RT)-PCR analysis was performed as described (Fode et al., 2008). Calculations were done according to the method (Livak & Schmittgen, 2001). UBQ5 served as a reference (Kesarwani et al., 2007). Primers used to amplify and quantify the cDNA are indicated in Supporting Information Table S1.
Potato dextrose agar was supplemented with indicated substances in different concentrations. The plates were spot-inoculated with 2500 spores and incubated at 20°C in the dark for 28 d. Photos of colonies were taken under a binocular microscope and the colony area was measured using the leaf area program, Bildanalyseprogramm 22.214.171.124 (datInf).
Phenylpropanoid treatment of V. longisporum in liquid culture
V. longisporum was grown for 7 d in simulated xylem fluid medium (SXM)-medium (Neumann & Dobinson, 2003) at 20°C in the dark. Then the 7-d-old cultures were treated with coniferyl alcohol or sinapyl alcohol at 100 μM for 24 h.
Phenylpropanoids are increased in Arabidopsis after V. longisporum infection
Metabolite fingerprinting was used as a nontargeted approach to identify metabolic changes in Arabidopsis leaves after inoculating the roots with V. longisporum. Infected plants were compared with control plants at 10, 21 and 35 dpi. Disease symptoms (stunted growth, vein clearing, early senescence) were as described previously (Ralhan et al., 2012). A set of 862 metabolite marker candidates was derived from the polar extraction phase with a false discovery rate (FDR) ≤ 10−4 (Table S2). Using the MarVis clustering tool (Kaever et al., 2009), the metabolite marker candidates were grouped according to their relative intensities in the six different treatments in a one-dimensional self-organizing map (1D-SOM) using 20 clusters (Fig. 1a). All metabolite markers that were grouped into clusters 10–20 were increased after infection (Fig. 1a, blue box). Clusters 11–17 include infection markers that had already accumulated at 10 dpi (Fig. 1a, red box). These early infection markers belong to different branches of the phenylpropanoid biosynthesis pathway (Fig. 2). The identities of all markers described below were confirmed by MS2 fragmentation experiments via UHPLC-QTOF-MS, co-elution with authentic standards (Table S3, Fig. S1), and/or RP-HPLC-DAD analysis (Fig. 3). In addition, metabolite markers of so far unknown identity accumulated during infection (Table S2).
Infection-induced metabolites of the phenylpropanoid pathway were the sinapate esters sinapoyl and bissinapoyl glucose (Figs 1b, 2, sinapoyl glc and bissinapoyl glc), as well as coniferin and syringin (Figs 1b, 2). Coniferin and syringin are the glucosylated forms of coniferyl and sinapyl alcohol, respectively. Both alcohols belong to the group of monolignols (Fig. 2, yellow box, bold black substances) and act as precursors for the lignan and lignin biosynthesis. Lignans are dimeric or oligomeric products of monolignols, whereas lignin is the polymeric product (Fig. 2, green, blue and brown boxes). Another V. longisporum-induced metabolite was a sinapyl alcohol-derived lignan, the syringaresinol glucoside (Fig. 2, brown box). The third group of V. longisporum-induced metabolites consisted of derivatives of the lignan pinoresinol, which is derived from coniferyl alcohol (Fig. 2, blue box). Pinoresinol and lariciresinol again were not enriched, but a strong increase of their glucosides was observed (Fig. 1b). Furthermore, secoisolaricoresinol is more abundant after infection. Finally, the glucosides of the sesamin biosynthesis pathway, which are derived from pinoresinol, were enriched (Fig. 2, green box).
Whereas these metabolites hasstarted to accumulate already at 10 dpi, increased concentrations of the amino acids valine, arginine, glutamine, asparagine, leucine and the cyclic pyroglutamate were detected only after 21 dpi (Table S2, Fig. S2). This might be due to the senescence-related induction of protein mobilization.
Quantification of metabolite markers for V. longisporum infection
In order to quantify the amount of the identified infection markers, metabolites were extracted with methanol and quantified by RP-HPLC-DAD analyses. As already observed by metabolite fingerprinting, both sinapoyl glucose esters accumulated upon infection while the amount of sinapoyl malate, the main sinapate ester in leaves of Arabidopsis, was not affected (Figs 1b, 3a). In the cases of the lignans and monolignols, extracts were treated with β-glucosidase before detection. A strong accumulation of coniferin (quantified as coniferyl alcohol), pinoresinol and lariciresinol glucosides after infection was recorded (Figs 1b, 3b). Of these metabolites, lariciresinol glucosides, which accumulated up to 100 nmol g−1 FW, were the most abundant.
Comparison of HPLC-chromatograms of leaf samples from inoculated plants (21 dpi) with those from control plants revealed the accumulation of additional UV-absorbing substances (Fig. 4). These metabolites were collected and analyzed with ESI TOF-MS to obtain their exact masses and to derive the sum formula of the compounds. This led to the identification of coniferyl alcohol, secoisolaricoresinol, lariciresinol, and three unknown substances: peak 1 (406.1628 Da, C21H26O8), peak 2 (358.1416 Da, C20H22O6), peak 3 (372.1209 Da, C20H20O7).
V. longisporum infection leads to an increased formation of vascular bundle cells
As recently described, new vessel elements are generated in the vascular tissue in infected plants (Ralhan et al., 2012; Reusche et al., 2012). In agreement with this finding, cross-sections of hypocotyls and petioles from our experiments stained by Wiesner stain showed more lignified cells in the vascular bundle (Fig. S3). This observation suggested that the observed increase in glycosylated precursors of lignin may be a consequence of enhanced lignification in the infected plants.
V. longisporum infection affects gene expression in the phenylpropanoid pathway
In order to analyze if the accumulation of infection-induced metabolites of the phenylpropanoid pathway coincided with the expression of genes involved in the formation of these substances (shown in Fig. 2), quantitative RT-PCR analysis was performed at 21 dpi (Fig. 5). Phenylalanine ammonia lyase (PAL) catalyzes the first committed step into the phenylpropanoid biosynthesis pathway (Fig. 2). At least, PAL1 was induced upon infection confirming an induction of the pathway (Fig. 5). The transcript level of ferulate-5-hydroxylase (F5H, Fig. 5), which catalyzes the conversion from coniferyl to sinapyl derivatives, was not affected by V. longisporum infection. This may be explained by the accumulation of both coniferyl as well as sinapyl derivatives upon infection. By contrast, the transcript of a gene putatively involved in glycosylation of sinapic acid (UDP-glycosyl transferase 84A3 (UGT84A3)), was induced. An infection-dependent increase in transcripts that may be involved in the formation of coniferyl- and sinapyl alcohol was detected for two cinnamyl alcohol dehydrogenase (CAD) genes: CAD5 and especially CAD8 (corresponding to ELI3-2: Elicitor-activated gene 3-2; Fig. 5). The putative glucosyl transferase involved in sinapic glucoside synthesis UGT84A3 as well as CAD5 and CAD8 co-localize with a QTL for resistance to Verticillium systemic colonization on chromosome 4 (Haffner et al., 2010). Furthermore, UGT72E2, a glucosyl transferase which is involved in the synthesis of coniferin/syringin from coniferyl alcohol/sinapyl alcohol in leaves, was slightly induced during infection. By contrast, UGT72E1 and UGT72E3, whose gene products also catalyze glucosylation of monolignols, were not induced (Lanot et al., 2006, 2008). Finally, the transcript level encoding the dirigent protein DIR6, which is responsible for stereo-specific coupling of the monolignol coniferyl alcohol to the lignan pinoresinol (Fig. 2), was induced and therefore may be involved in the formation of pinoresinol glucosides (Pickel et al., 2012).
Loss of sinapates enhances the susceptibility to V. longisporum
In order to examine whether the accumulation of sinapate esters (Fig. 2, yellow box) has an effect on symptoms of the V. longisporum infection, mutants impaired in different enzymes of the phenylpropanoid biosynthesis were analyzed. The fah1-2 mutant, defective in F5H, contains no sinapate esters as well as no syringyl lignin (Figs 2, 6a; Chapple et al., 1992). Indeed, this mutant showed increased susceptibility to V. longisporum (Fig. 6b–d). The leaf area was more strongly reduced (30%) than in the wild-type (WT). Moreover, 3.6 times more V. longisporum DNA was found in these plants. The fah1 gene co-localizes as well with the QTL for resistance to Verticillium mentioned above (Haffner et al., 2010). This result points at a protective role of sinapate esters, although other symptoms including vein clearing and water soaking of the leaves at the beginning of the light period were as pronounced as in the WT. In order to differentiate whether the inducible levels of sinapoyl glucose or the constitutive levels of sinapoyl malate are important, two further mutants were analyzed. The ref1-s mutant (SALK_027911) is defective in the aldehyde dehydrogenase (ALDH, Fig. 2) resulting in strongly reduced amounts of sinapoyl glucose with only two-fold-reduced amounts of the malate ester (Fig. 6a). This mutant was indistinguishable from WT with regard to symptom development and fungal proliferation, suggesting that sinapoyl malate can contribute to the defense (Fig. 6b–d). The sng1-1 mutant, which is defective in the sinapoyl glucose:malate sinapoyl transferase (SMT) thus accumulating high amounts of sinapoyl glucose and no sinapoyl malate (Lorenzen et al., 1996; Fig. 6a) was also indistinguishable from WT suggesting that sinapoyl glucose alone is as effective as sinapoyl malate (Fig. 6b–d). Neither ref1 nor sng1 co-localize with the QTL for resistance to Verticillium, which comprises UGT84A3, CAD5, CAD8 and fah1 (Haffner et al., 2010).
In addition to these knock-out lines, an overexpressing line of the F5H gene was analyzed (C4H:F5H). This transgenic line is described to contain high amounts of syringyl instead of guaiacyl lignin (Meyer et al., 1998). However, the reduction of the leaf area in these plants post inoculation was only in the same range as in WT plants (Fig. S4). In agreement with the results obtained with the fah1-2 mutant, the amount of fungal DNA was 6.5 times lower compared to WT plants for this plant.
Plants accumulating coniferin showed higher resistance to V. longisporum
In order to further characterize the biochemical basis of the enhanced susceptibility to V. longisporum by the loss of sinapate esters, transgenic plant lines were analyzed in which the flux through the phenylpropanoid pathway was redirected by glucosylation of soluble intermediates. The three glycosyl transferases UGT72E1, E2 and E3 are described to be involved in the synthesis of coniferin and syringin, which are the transport and storage forms coniferyl and sinapyl alcohol, respectively (Lanot et al., 2006, 2008). The corresponding overexpressor (OE) lines preferentially accumulate coniferin (UGT72E2) or both coniferin and syringin (UGT72E3) in leaves. In addition, these plants (UGT72E2-OE and E3-OE) contain lower amounts of sinapoyl malate and higher amounts of ferulic acid-4-O-glucoside (E2, E3) and sinapic acid-4-O-glucoside (E3) compared to WT plants (Lanot et al., 2006, 2008). The three UGT72E OE lines were analyzed with respect to the accumulation of coniferin and/or syringin under the growth conditions applied in this study. While UGT72E1-OE accumulated negligible amounts of the two compounds in the leaves, UGT72E2-OE preferentially accumulated coniferin and UGT72E3-OE preferentially accumulated syringin (Fig. 7a). The amount of coniferin increased approximately five-fold in infected UGT72E2-OE lines, while the infected UGT72E3-OE reached only the level of uninfected UGT72E2-OE lines for coniferin (Fig. S5). After inoculation with V. longisporum, only UGT72E2-OE plants exhibited differences to the WT plants. As shown in Fig. 7(b–d), the leaves of UGT72E2-OE plants, which preferentially accumulate coniferin, were less stunted in response to the infection (50%). Quantification of fungal DNA revealed that there was also 10 times less fungal material in these plants as compared to WT (Fig. 7d). UGT72E1-OE plants, which accumulate negligible amounts of coniferin or syringin, and UGT72E3-OE plants, which preferentially accumulate syringin, were indistinguishable from WT (Fig. 7).
Lignin analysis of infected plant lines
Because the two genotypes fah1-2 and UGT72E2-OE show alterations in a metabolite pathway that leads to lignin synthesis (Fig. 2), petioles of control and inoculated WT, fah1-2 and UGT72E2-OE plants were analyzed for lignin composition and amount at 21 dpi (Table 1). After inoculation, no significant enrichment in total lignin was detected in all three lines. In case of lignin composition, only a slight increase in p-hydroxyphenyl lignin (H-lignin, Fig. 2) was found in UGT72E2-OE plants, whereas S-lignin was barely detectable in the fah1-2 mutant (Table 1).
Table 1. Lignin content and monomer composition in Arabidopsis petioles of control and inoculated plant lines at 21 d post inoculation (dpi)
Lignin content (mg g−1 DW)
Relative monomeric composition (%)
H-units derive from p-hydroxyphenyl (or p-coumaryl) lignin, G-units from guaiacyl (or coniferyl) lignin and S-units from syringyl lignin. Four samples from two independent experiments were analyzed.
34.8 ± 8.5
38.7 ± 3.6
34.9 ± 6.6
39.5 ± 5.6
34.8 ± 8.8
40.4 ± 9.7
Fungal growth is specifically inhibited by coniferyl alcohol and sinapoyl glucose whereas melanization is specifically affected by coniferin
Our metabolite analysis showed that the more susceptible fah1-2 plants and the more resistant UGT72E2-OE plants differed from WT plants in the accumulation of sinapate esters or coniferin, respectively (Fig. S5). Therefore, the impact of sinapoyl glucose, coniferin and coniferyl alcohol on fungal growth was analyzed (Fig. 8a–c). Sinapoyl malate could not be tested, because it was not available in sufficient amounts for this analysis. V. longisporum was grown on potato dextrose agar (PDA) plates supplemented with different concentrations of these substances, and the colony area of 17-d-old cultures was determined. Interestingly, fungal growth was specifically inhibited by coniferyl alcohol up to 20% (10 μM) and up to 30% (100 μM) as well as by sinapoyl glucose up to 35% (100 μM). Coniferin, sinapyl alcohol and syringin had no effect.
V. longisporum grown for > 21 d on PDA plates starts to melanize. However, when grown in the presence of 100 μM coniferin, melanization was delayed and only minor black areas were detected at 28 d of growth (Fig. 8d). This effect was not observed for the other four metabolites tested.
The growth-inhibiting coniferyl alcohol differs from the noneffective sinapyl alcohol only by one methoxy group (Fig. 9). Because we rationalized that this difference might not account for the different degrees of toxicity, we hypothesized that both compounds might be converted to different metabolites by V. longisporum-derived enzymes. To challenge this hypothesis, the fungus was grown in liquid cultures in the presence of each of these monolignols for 1 d and the liquid medium was analyzed by metabolite fingerprinting (Fig. 9). During this time, the coniferyl alcohol was oxidized to ferulic acid, which also inhibited V. longisporum growth on PDB plates (Fig. S6). However, sinapyl alcohol was mainly dimerized into syringaresinol whereas very little sinapic acid accumulated. The latter had no inhibitory effect on V. longisporum growth (Fig. S6).
The aim of this study was to identify metabolic pathways that are involved in the defense of Arabidopsis against V. longisporum in aboveground organs. We analyzed different time points during infection, starting with 10 dpi where the fungus just starts to colonize the xylem vessels. Later, it causes disease symptoms such as induction of xylem vessel formation, stunted growth, vein clearing and early senescence, and we aimed to correlate these changes with changes in the metabolite profile. At early and late stages of the infection, most of the infection-related metabolites (namely sinapoyl glucose, bissinapoyl glucose, monolignols and lignans) were derived from the phenylpropanoid pathway (Figs 1, 2, Table S2). Only trace amounts of indolic glucosinolates and other phytoalexins such as camalexin were identified, which contrasts to the situation in roots, where indolic compounds were strongly induced upon attempted fungal entry (Iven et al., 2012). This difference is not unexpected, considering that the fungus shows invasive growth in roots, whereas growth in the xylem might not be accompanied by attempts of the fungus to colonize living tissue. Thus, the elicited plant defense reactions are likely to be completely different, as has been shown before for Arabidopsis inoculated with Pseudomonas syringae pv tomato and Pythium sylvaticum (Tan et al., 2004).
Sinapate esters may contribute to the defense of Arabidopsis against V. longisporum
Sinapoyl glucose was one of the prominent markers in Arabidopsis leaves enriched already at early time points (10 dpi) of infection (Fig. 1b). The sinapate esters are specifically found in members of the Brassicaceae with an important function in protection against UV-B light (Landry et al., 1995). Sinapoyl glucose is described to be the high-energy sinapate donor for sinapoyl choline in the seeds and sinapoyl malate in the leaves (Mock & Strack, 1993; Milkowski & Strack, 2010). Sinapate ester accumulations were detected to in Fusarium oxysporum-infected Brassica rapa where the accumulation of sinapoyl-, feruloyl- and 5-hydroxy feruloyl malate was shown by nontargeted metabolite profiling. Still no function was attributed to these metabolites (Abdel-Farid et al., 2009). To further elucidate the role of sinapoyl glucose accumulation for V. longisporum infection, different mutants of the sinapate biosynthesis pathway were examined. The fah1-2 mutant, that is devoid of all sinapate esters, was more susceptible than the WT. Because lignin composition was not significantly changed (Table 1) we suggest that either the sinapoyl malate or the sinapoyl glucosides lacking in the mutant have anti-fungal activities. Along these lines, the WT-like susceptibility of the ref1-s and sng1-1 mutants, which differ from the fah1-2 mutant by higher amounts of either sinapoyl malate or sinapoyl glucose, can be explained by the protective activity of each substance alone (Fig. 6). Indeed, at least sinapoyl glucose inhibited fungal growth in vitro (Fig. 8c).
High concentrations of coniferin may contribute to defense responses against V. longisporum
In order to address the functional relevance of the elevated amounts of coniferin and syringin in infected plants, plant lines overexpressing glycosyl transferases that are involved in the synthesis of these compounds were analyzed (UGT72E1-3, Fig. 7). Only the UGT72E2-OE line with an up to 600-fold enriched content of coniferin showed higher resistance to the fungus when compared to WT (Figs 7, S5). This is in agreement with the observation that only the transcript of this glycosyl transferase was induced upon infection (Fig. 5). Coniferin is discussed as a storage form of coniferyl alcohol that may be synthesized either by CAD5 or CAD8 (Fig. 5) and the question arises of how this compound might exhibit antifungal activity (Boerjan et al., 2003; Wang et al., 2013). An indirect effect through altering the cell wall can be excluded, because no significant changes in lignin content and only a slight increase in H-lignin were found compared to the WT (Table 1). This is also consistent with the notion that coniferin is not a direct precursor of lignin (Boerjan et al., 2003; Wang et al., 2013). Notably, coniferyl alcohol inhibited the growth of V. longisporum in vitro and coniferin inhibited melanization (Fig. 8). Because coniferyl alcohol is converted to ferulic acid by V. longisporum (Fig. 9a) and because ferulic acid has the potential to inhibit V. longisporum growth, we can imagine that the toxicity of ferulic acid could be a reason for the resistance in plants overaccumulating coniferin (Fig. S6). Moreover, ferulic acid is known to be incorporated into the cell wall, where it enhances structural barriers against pathogens by cross-linking cell wall polysaccharides (Boutigny et al., 2009; Lloyd et al., 2011). Thus, both, the immediate toxic effects of ferulic acid and the stronger cross-linking of cell wall polysaccharides might lead to the observed restriction of fungal proliferation in Arabidopsis. In addition, coniferin might interfere with the final stages of fungal development in planta, as suggested by the inhibition of melanization in vitro.
Consistently, we find that plants with elevated concentrations of syringin are not more resistant and that sinapyl alcohol is not toxic. In the presence of the fungus, sinapyl alcohol is only slowly oxidized to the corresponding acid and dimerizes to syringaresinol (Fig. 9b). These derivatives do not seem to inhibit fungal growth in vitro (Fig. S6).
Therefore, the higher resistance of the UGT72E2-OE line as compared to the WT may be explained by its high content of coniferin and thus a faster production of coniferyl alcohol and ferulic acid. As these intermediates do not increase during infection, we have to postulate that only the flux through this pathway is increased in the presence of higher amounts of coniferin. However, we cannot exclude the possibility that increased contents of other soluble phenylpropanoids in this line may contribute to its higher resistance (Lanot et al., 2006, 2008). Notably, in the Botryis cinerea–Arabidopsis interaction, ferulate and monolignol metabolism appeared to be a major source for resistance as well (Lloyd et al., 2011; Demkura & Ballare, 2012).
The role of increased lignan-glucosides after infection is yet unexplored
Next to sinapate esters, coniferin and syringin, lignan-glucosides were strongly enriched upon infection (Figs 1b, 3b). Plant lignans have been attracting the attention of medical professionals because of their health-promoting and anti-cancerogenic effects. These metabolites possess antimicrobial, antifungal and antiviral properties (Lewis et al., 1995) and act as defense compounds against insects (Schroeder et al., 2006). In addition, antifungal activity of pinoresinol against phytopathogenic fungi such as Fusarium verticilloides was reported (Carpinella et al., 2003). In Arabidopsis, only lariciresinol has so far been found. It is abundant in the roots but does not occur in the leaves (Nakatsubo et al., 2008). We found that numerous lignans such as pinoresinol-, lariciresinol-, syringaresinol- and sesamins, for example sesamolinol glucoside, are abundant in the leaves of Arabidopsis. Furthermore, these lignans accumulate in high amounts (up to 100 nmol g−1 FW) upon infection with V. longisporum (Figs 1, 3). In agreement with these results, the transcript levels of Dirigent Protein 6 (DIR6), a protein that catalyzes formation of lignans, was induced upon infection (Fig. 5). The induction of the expression of dirigent proteins was shown to be involved in resistance of conifers against insects (Ralph et al., 2007). Our data suggest that it is probably also important for the defense against fungal pathogens. DIR6 co-localizes with the same QTL for resistance to Verticillium on chromosome 4 as UGT84A3, which is putatively involved in glycosylation of sinapic acid, cinnamyl alcohol dehydrogenases CAD5 and CAD8 and ferrulic acid hydroxylase gene fah1 (Haffner et al., 2010).
Due to the lack of mutants, the impact of the lignans on infection was difficult to examine. A pinoresinol-reductase double mutant (Nakatsubo et al., 2008) was tested, but in this mutant merely the ratio of lariciresinol to pinoresinol is changed (Table S4). Therefore, it was not surprising that no differences in the interaction with Verticillium were detected compared to the WT plants (data not shown).
In addition, we could not detect any effect of pinoresinol, lariciresinol and pinoresinol glucoside on the growth of V. longisporum (data not shown). Another potential function is to block fungal spreading as structural elements of the cell wall, as some lignans are also present in the extracellular compartment (Floerl et al., 2012). In woody plants, lignans, which are highly abundant in heartwood, act as antioxidants. In addition, they shut off the nonproductive water and nutrient transport to protect the sapwood against wood-rotting fungi (Naoumkina et al., 2010).
Total amount in lignin was not increased upon Verticillium infection
One of the end-products of the phenylpropanoid biosynthesis pathway is lignin. Thus, increased concentrations of soluble metabolites after V. longisporum infection prompted us to analyze the lignin content after infection. However, the increase in lignin content was not significant as the overall ratio of G, H to S-lignin did not change after infection (Table 1). On the one hand, G-lignin is more condensed as it contains more resistant linkages in comparison to S-lignin due to only one methoxy group (Boerjan et al., 2003). On the other, S-lignin is more linear and supposed to better protect cell wall polysaccharides from degradation by hydrolytic enzymes (Jung & Deetz, 1993). There are examples of pathosystems where S-lignin might be involved in preformed or induced resistance (Wuyts et al., 2006; Menden et al., 2007; Lloyd et al., 2011; Eynck et al., 2012), but in our pathosystem we do not find any evidence for such a mechanism. Our analysis also excludes the possibility that differences in lignin composition in the more susceptible fah1-2 mutant and the more resistant UGT72E2-OE line contribute to differences in resistance.
In summary, we show that nontargeted metabolite fingerprinting is a valuable tool to identify metabolites or metabolic pathways involved in plant–pathogen interactions. Our data show that sinapoyl glucosides, coniferin, syringin and lignans from phenylpropanoid metabolism, as well as amino acid metabolism, were affected by the infection of Arabidopsis with V. longisporum. Using mutant analysis and in vitro growth assays, we have identified sinapate glucose and coniferin to be involved in restriction of the pathogen. Functional analysis of the lignans and their glucosides might add these compounds to the list of anti-fungal defense compounds. The importance of the phenylpropanoid pathway in the defense of the aerial plant parts against V. longisporum is also supported by studies on Brassica napus (Eynck et al., 2009; Obermeier et al., 2013). Here, soluble and cell wall-bound phenolics seem to be involved in the early defense against the fungus, whereas lignin might be important at later infection stages (Eynck et al., 2009). Five genes of lignol and lignan metabolism, that our data showed to be involved in resistance to Verticillium, co-localize with a QTL for Verticillium resistance on chromosome 4 of Arabidopsis (Haffner et al., 2010). In line with this, Obermeier et al. (2013) found that quantitative trait loci (QTL) for some phenylpropanoids co-localized with a QTL for V. longisporum resistance in B. napus. Concentrations of phenylpropanoids were correlated with V. longisporum resistance, making genes of the phenylpropanoid pathway interesting candidates for resistance breeding in oilseed rape (Obermeier et al., 2013).
We are grateful to Drs D. Bowles, C. Chapple and T. Umezawa for providing mutant plants and Gurneet Braich, Pia Meyer, Sabine Freitag, Larissa Kunz and Gabriele Lehmann for excellent technical assistance, as well as the members of the Verticillium research group for fruitful co-operation and discussions. This work was supported by the DFG Research Unit FOR546 Fe 446/6, Ka 1209/3, Po 362/15 and the excellence initiative FL3 INST 186/822-1.