Deficiency of abscisic acid (ABA) in the sitiens mutant of tomato (Solanum lycopersicum) culminates in increased resistance to Botrytis cinerea through a rapid epidermal hypersensitive response (HR) and associated phenylpropanoid pathway-derived cell wall fortifications. This study focused on understanding the role of primary carbon : nitrogen (C : N) metabolism in the resistance response of sitiens to B. cinerea. How alterations in C : N metabolism are linked with the HR-mediated epidermal arrest of the pathogen has been also investigated.
Temporal alterations in the γ-aminobutyric acid (GABA) shunt, glutamine synthetase/glutamate synthase (GS/GOGAT) cycle and phenylpropanoid pathway were transcriptionally, enzymatically and metabolically monitored in both wild-type and sitiens plants. Virus-induced gene silencing, microscopic analyses and pharmacological assays were used to further confirm the data.
Our results on the sitiens–B. cinerea interaction favor a model in which cell viability in the cells surrounding the invaded tissue is maintained by a constant replenishment of the tricarboxylic acid (TCA) cycle through overactivation of the GS/GOGAT cycle and the GABA shunt, resulting in resistance through both tightly controlling the defense-associated HR and slowing down the pathogen-induced senescence.
Collectively, this study shows that maintaining cell viability via alterations in host C : N metabolism plays a vital role in the resistance response against necrotrophic pathogens.
Plants have evolved a plethora of defense strategies to persist in a hostile environment that abounds with pathogens. Besides constitutive structural and biochemical defenses, plants possess sophisticated defense machinery that is put to use against those few pathogens that succeed in overcoming the preformed barriers. The plant possesses an arsenal of these inducible defenses, including the reactive oxygen species (ROS) burst, hypersensitive (HR) cell death, cell wall fortifications and de novo synthesis of various antimicrobial compounds, which ultimately deprive a (hemi)biotrophic pathogen of the plant's nutritional resources (Pieterse et al., 2009). However, the efficacy of such ROS/HR-mediated defense strategies against necrotrophic pathogens, which favor host cell death, remains questionable (Robert-Seilaniantz et al., 2007).
Botrytis cinerea has recently been nominated as the most scientifically and economically important necrotrophic fungal plant pathogen (Dean et al., 2012). As a true necrotroph, induction of cell death in the host tissue is known as a crucial mode of pathogenicity (Van Baarlen et al., 2004; Govrin et al., 2006). To accomplish this, B. cinerea is armed with an array of virulence factors, such as phytotoxic metabolites, oxidative burst-evoking enzymes and oxalic acid (Govrin & Levine, 2000; Van Kan, 2006). On the other hand, resistance to necrotrophic pathogens such as B. cinerea depends, to some extent, on the ability of the host to control the balance between cell death and survival (Mengiste, 2012). It has been shown that expression of a senescence-associated gene of Arabidopsis (SAG12) is induced by B. cinerea infection, while transgenic nonsenescing tomatoes exhibit some degree of resistance to the pathogen (Swartzberg et al., 2008). A model has been proposed for Arabidopsis nonhost resistance to B. cinerea, according to which presence of a cellular zone around the primary infection site with a slower rate of cell death can restrict pathogen colonization (Van Baarlen et al., 2007).
We have previously demonstrated that abscisic acid (ABA) deficiency in the sitiens mutant of tomato results in a strong resistance response to B. cinerea (Audenaert et al., 2002). Additionally, it was recently shown that cuticle permeability and specific cell wall structure in the mutant lead to rapid defensive responses, including an early and localized production of hydrogen peroxide (H2O2) and HR followed by cell wall fortifications in the epidermis, which ultimately impede further spreading of the pathogen (Asselbergh et al., 2007; Curvers et al., 2010). Microarray analysis revealed that, in addition to groups of genes coding for pathogenesis-related (PR) proteins and enzymes related to cell wall structure being up-regulated, expression of a cluster of genes involved in primary amino acid metabolism significantly increased in the resistant mutant at 8 h postinoculation (hpi; Asselbergh et al., 2007). The most notable of the latter group of genes were peroxisomal/glyoxysomal aspartate transaminase (AAT), the lysine catabolic gene ketoglutarate reductase/saccharopin dehydrogenase (LKR/SDH) and the γ-aminobutyric acid (GABA) biosynthetic gene glutamate decarboxylase (GAD). However, the cause and the putative role of these amino acid-associated transcriptional alterations in the resistance mechanism of sitiens against B. cinerea remained unclear.
Although there are dozens of studies corroborating the link between primary metabolism and plant defense responses (reviewed by Bolton, 2009), molecular knowledge on the role of primary amino acid metabolism in plant resistance mechanisms against pathogens is still fragmentary. We have recently reviewed the role of glutamate metabolism in plant–pathogen interactions, highlighting the importance of infection-triggered alterations in central amino acid metabolism in molding a resistance/susceptibility response in the host (Seifi et al., 2013). A causal link has been proposed between the disruption in glutamate/glutamine homeostasis and induction of cellular redox imbalances in leaf cells of Arabidopsis, leading to ROS-mediated cell death and resistance against the biotrophic fungus Erysiphe cichoracearum and the hemibiotrophic fungus Colletotrichum higginsianum (Liu et al., 2010). Conversely, drastic depletion in glutamate storage in distal, noninvaded regions of sunflower was postulated as a defensive strategy against B. cinerea. It was hypothesized that, through this mechanism, glutamate-derived N-rich amino acids are supplied to the infected area, delaying the necrotroph-induced senescence (Dulermo et al., 2009). In Arabidopsis, LKR/SDH is significantly up-regulated by abiotic stresses, but also by B. cinerea infection (Genevestigator; Zimmermann et al., 2005). It is suggested that stress-induced lysine catabolism serves as a generator of glutamate, the key precursor for important stress-related metabolites such as arginine, proline and GABA (Galili et al., 2001). Similarly, the observed increased susceptibility to B. cinerea in cytosolic aspartate transaminase (AAT1) overexpressing lines of Arabidopsis was suggested to be caused through the repression of glutamate-consuming pathways to maintain normal glutamate concentrations. In this way, less glutamate would be converted to key defense-related metabolites such as proline or GABA (Brauc et al., 2011).
γ-Aminobutyric acid is known to accumulate in plant tissues in response to various abiotic and biotic stresses, suggesting assorted functions for the molecule, ranging from involvement in central carbon : nitrogen (C : N) metabolism to functioning as a signaling molecule during plant–microbe interactions (Kinnersley & Turano, 2000; Roberts, 2007; Fait et al., 2008). Making up a considerable fraction of total free amino acid content in some plants (Roberts, 2007), GABA might be an attractive nutritional target, particularly for intercellularly growing pathogens. For instance, infection of tomato by the biotrophic fungus Cladosporium fulvum was shown to induce high concentrations of GABA in the apoplast, providing the pathogen with a rich nitrogen source (Solomon & Oliver, 2002). GABA also appears to be a nutrient source for Pseudomonas syringae pv. tomato in Arabidopsis. However, conditions promoting high concentrations of GABA can increase plant resistance by repressing the expression of the hrp genes in the bacterium (Park et al., 2010). The involvement of GABA in plant–stress interactions has been also attributed to the metabolic functions of the GABA shunt, a cytosolic-mitochondrial pathway with an important role in the plant central C : N metabolism, connecting amino acid metabolism to the tricarboxylic acid (TCA) cycle (Fait et al., 2008). The GABA shunt consists of three key enzymes: GAD, GABA transaminase (GABAT) and succinic-semialdehyde dehydrogenase (SSADH). Specifically using glutamate as the main precursor for GABA biosynthesis, GAD activity catalyzes the first step of the GABA shunt in the cytosol, while the other two steps of the shunt, GABAT and SSADH, occur in mitochondria (Shelp et al., 2012). Although there are a number of studies demonstrating involvement of the GABA shunt in response to abiotic stresses (Bouché et al., 2003; Ludewig et al., 2008), information on the molecular mechanism underpinning the role of this pathway in plant defense mechanisms against pathogens is still scarce (Wu et al., 2006).
This study aimed to unravel the role of the GABA shunt in the resistance response of the sitiens mutant to B. cinerea. In addition, we investigated how alterations in GABA metabolism may be linked to the HR-mediated epidermal arrest of the pathogen seen in the mutant. The efficacy of the ROS-fueled defense of sitiens against a ROS-favoring pathogen, such as B. cinerea, appears to be vitally dependent on a secondary anti-cell-death defense mechanism, activated in the area surrounding the invaded cells via concurrent overactivation of the GABA shunt and the cytosolic glutamine synthetase. Our sitiens model has revealed that maintaining cell viability via alterations in central C : N metabolisms is crucial in the resistance response to a necrotrophic pathogen.
Materials and Methods
Plant and pathogen materials and growth conditions
The ABA-deficient sitiens mutant of tomato (Solanum lycopersicum L.); (Taylor et al., 1988, 2000), and the corresponding wild-type ‘Moneymaker’ were grown in potting compost soil in a growth chamber under the following conditions: temperature, 20–25°C; 16 h : 8 h light : dark regime; relative humidity, 75%. Conidia of Botrytis cinerea strain R16 (Faretra & Pollastro, 1991) and codon-optimized green fluorescent protein (GFP)-expressing Botrytis cinerea strain B05.10 (Leroch et al., 2011) were maintained as previously described (Audenaert et al., 2002; Asselbergh et al., 2007).
Inoculation method and disease index evaluation
The inoculation conidial suspension was prepared according to Curvers et al. (2010), containing 5 × 105 spores ml−1 in 0.01 M KH2PO4 and 6.67 mM glucose, while, for mock inoculation, conidia were omitted from the suspension. For infection trials, leaf disks were inoculated according to Asselbergh et al. (2007) by putting 10 μl droplets of the inoculation suspension on the adaxial surface of leaf disks punched from fifth or sixth leaves of 5-wk-old plants. Inoculated leaf disks were placed in enclosed 24-well plates and incubated at 22°C under darkness. Disease index was evaluated at 5 dpi as described previously (Curvers et al., 2010), using a disease scoring scale containing four infection severity categories (Supporting Information, Fig. S1). Spray inoculation was used for enzymatic assays and transcriptional analyses. The fifth or sixth leaves of 5-wk-old plants were excised and placed in a plastic tray on four layers of absorbent paper, moisturized with 400–500 ml of demineralized water. Leaves were sprayed with the inoculation suspension using an atomizer until a homogenous coverage of 1–2 μl droplets of the leaf was reached.
Visualization of defense responses
H2O2 accumulation was visualized using 3,3′-diaminobenzidine (DAB) staining according to the protocol of Thordal-Christensen et al. (1997). Samples were floated on an aqueous solution of DAB-HCL (1 mg ml−1, pH 4) for 3 h and then cleared and fixed in 100% ethanol at the desired time point as previously described (Azami-Sardooei et al., 2010). To visualize cell wall modifications, safranin-O staining was performed according to Lucena et al. (2003). Leaf disks were incubated in 0.01% safranin-O in 50% ethanol for 3 min. Fungal structures were stained using a slightly modified trypan blue staining technique (Asselbergh et al., 2007), with 0.1% trypan blue in 10% acetic acid for 15 s. All the staining protocols were followed by extensive rinsing steps in demineralized water and the samples were then mounted in 50% glycerol before microscopic observation. Accumulation of autofluorescing phenolic compounds was detected using an Olympus (Tokyo, Japan) BX51 epifluorescence microscope (U-MWB2 GPF filter set, excitation filter 450–480 nm, dichroic beam splitter 500 nm, barrier filter BA515) according to De Vleesschauwer et al. (2010). Live cell imaging was done using a U-MWG2 filter cube (excitation filter 510–550 nm, dichroic beam splitter DM 570, barrier filter BA590). Images were digitally acquired using an ‘Olympus Color View II’ camera, and further processed with the Olympus analysis ‘cell^F’ software.
Enzymatic activity assays and total protein content measurement
Soluble proteins were extracted by resuspending the crushed tissue (150 mg FW) in 0.8 ml of potassium phosphate buffer 0.1 M, pH 8, containing 2% (w/v) polyvinylpyrrolidone, 0.1% (v/v) Triton X-100, 1 mM dithiothreitol (DTT) and 1 mM phenylmethylsulfonyl fluoride (PMSF). The extracts were then vortexed for 1 min and centrifuged at 10 000 g and 4°C for 10 min. The Bradford method (Bradford, 1976) was used to measure protein abundance in extracts used in enzymatic assays.
Phenylalanine ammonia-lyase (PAL) enzyme activity was assayed according to Sreelakshmi & Sharma (2008) in a reaction mixture containing an aliquot of the protein extract (100 μl) in 0.9 ml of 0.1 M borate buffer, pH 8.8, and 10 mM l-phenylalanine (total volume of 1 ml). The reaction was carried out for 60 min at 37°C and the increase in absorbance at 290 nm was recorded at every 10 min interval. The enzyme activity was determined kinetically, based on the rate of formation of trans-cinnamic acid by calculating the slope of the linear part of the plot as ‘ΔAbs290 min–1 mg–1 protein’.
Glutamine synthetase (GS) enzyme activity was measured by the synthetase reaction (glutamate + ammonia + ATP → glutamine + ADP), according to Machado et al. (2001) with minor modifications. The assay mixture was composed of 500 mM glutamate, 100 mM hydroxylamine, 200 mM MgSO4, 80 mM ATP and 50 mM imidazole buffer, pH 7.4. The reaction was started by adding an aliquot of the protein extract (0.2 ml) to a final volume of 2 ml. One millilitre of the mixture was immediately removed (at 0 min) and added to 1 ml of Ferguson & Sims (1971) reagent, consisting of 0.67 N HCl, 0.20 M TCA and 0.37 M FeCl3. Another 1 ml of the assay was incubated at 32°C for 15 min, at which time the reaction was stopped by the addition of 1 ml of Ferguson & Sims reagent. The resulting precipitate in both tubes (0 and 15 min) was removed by centrifugation at 1500 g for 3 min and the absorbance of the supernatant was recorded at 535 nm. The enzyme activity was calculated as ‘ΔAbs535 (15 min)–1 mg–1 protein’.
Succinic-semialdehyde dehydrogenase enzyme activity was assayed as described by Busch & Fromm (1999). The reaction mixture contained 900 μl of 0.1 M sodium phosphate buffer, pH 9.0, 1 mM DTT, 0.1 mM succinic-semialdehyde (SSA), 0.5 mM NAD+ and 100 μl of SSADH-containing protein extract. Enzyme activities were monitored for 5 min, and the increase in absorbance at 340 nm was recorded at every 0.5 min interval at 24°C. The enzyme activity was determined by calculating the slope of the linear part of the absorbance plot as ‘ΔAbs340 min–1 mg–1 protein’.
The leaf disk/24-well plate method, as described by Asselbergh et al. (2007), was used for all inhibitor treatments. Before inoculation and microscopic observations, leaf disks of sitiens and wild-type plants were floated for 12 h on aqueous solutions of the respective inhibitors at the indicated concentrations, while control leaf disks were placed on sterile demineralized water.
RNA extraction, cDNA synthesis, and quantitative RT-PCR analysis
RNA extraction, cDNA synthesis, and quantitative reverse transcription polymerase chain reaction (RT-PCR) analysis were performed following the procedure described by De Vleesschauwer et al. (2010). Briefly, total RNA was isolated from frozen leaf tissue using the Spectrum Plant Total RNA kit (Sigma). First-strand cDNA synthesis was done using the High Capacity cDNA Reverse Transcription kit and random primers (Applied Bioscience, Foster City, CA, USA), according to the manufacturer's instructions. Nucleotide sequences of all primers used in this study are listed in Table S1. Maxima Sybr Green qPCR master mix (Fermentas, Waltham, MA, USA) and the Mx3005P real-time PCR detection system (Stratagene, La Jolla, CA, USA) were used for quantitative PCR amplifications. A dissociation curve analysis was included to verify the target specificity of the primers used. Two different reference genes, actin (accession no. AB199316.1) and elongation factor 1α (accession no. X14449.1; Løvdal & Lillo, 2009), were used as internal controls to normalize the amount of plant RNA in each sample. Samples collected from mock-inoculated plants at 0 hpi were selected as a calibrator and the generated data were analyzed with the Mx3005P software (Stratagene).
Virus-induced gene silencing
Virus-induced gene silencing (VIGS) was carried out on tomato seedlings according to Velásquez et al. (2009). Briefly, fragments of the tomato, herein called PAL7 (AK320988.1), GS1 (AF200360.1) and GABAT1 (NM_001247407.1), were amplified by PCR from a tomato cDNA library with gene-specific primers (Table S2). Adaptor-PCR was used to attach the GATEWAY recombinatorial cloning (Invitrogen) attB1 (5′-GGGGACAAGTTTGTACAAAAAAGCAGGCT-3′) and attB2 (5′-GGGGACCACTTTGTACAAGAAAGCTGGGT-3′) sites to the fragments. The resulting fragments were cloned into the pTRV2-GW vector (Liu et al., 2002a). Constructed plasmids and the helper plasmid pTRV1 were isolated and transferred into Agrobacterium tumefaciens strain GV3101 by electroporation. A mixture of A.tumefaciens containing the pTRV1 and the pTRV2:GUS (or the pTRV2 containing the genes of interest) in a 1 : 1 ratio was used for agroinoculation. Three to four weeks after inoculation, fully expanded leaves of both silenced plants and vector control plants were used for infection with B. cinerea.
GABA concentration measurement
γ-Aminobutyric acid was extracted from 50 mg of freeze-dried tissue powder with 400 μl of methanol, 200 μl of chloroform and 400 μl of water at room temperature. Aliquots of 200 μl of the upper phase were dried under vacuum and residues were resuspended in 50 μl of ultrapure water. GABA was derivatized using the AccQ™ Tag Ultra derivitization kit (Waters Corporation, Milford, MA, USA) according to the manufacturer's instructions. Derivatized GABA was analyzed using an Acquity™ UPLC system (Waters Corporation) as described by Jubault et al. (2008). For each sample, GABA was reliably identified by comparison of sample chromatograms with standard, and then quantified after normalization against internal standards and plant material fresh weight.
The GABA shunt is involved in resistance of sitiens to B. cinerea
Previous microarray analysis has shown an up-regulation of the GABA anabolic gene, glutamate decarboxylase 1 (GAD1), in inoculated sitiens (Asselbergh et al., 2007; Brauc et al., 2011). Additionally, a link between AAT1 and LKR/SDH up-regulations by B. cinerea infection and GABA metabolism was previously proposed (Brauc et al., 2011). These observations prompted us to investigate whether GABA metabolism might play a role in the defense machinery of sitiens against B. cinerea. Quantitative reverse transcription (qRT)-PCR analyses revealed that all the key genes comprising the GABA shunt (i.e. GAD (GAD1 and GAD2), mitochondrial GABAT (GABAT1) and SSADH) were induced during early stages of the sitiens–B. cinerea interaction, whereas these alterations could not be seen in the wild-type (Fig. 1a). An SSADH enzymatic assay confirmed the rapid increase in the activity of the last step of the GABA shunt upon inoculation of the mutant (Fig. 1b).
In order to further evaluate whether the GABA shunt plays a role in the resistance of tomato to B. cinerea, GABAT1 (the transcript that showed the highest up-regulation in response to B. cinerea) was functionally tested using the tobacco rattle virus (TRV)-based VIGS technique (Liu et al., 2002a,b; Anand et al., 2007). The GABAT1 gene was cloned and a TRV construct was developed for VIGS in 10-d-old wild-type and sitiens seedlings. Three weeks after agroinfiltration, plants were assayed for resistance against B. cinerea. As a control, a TRV construct harboring the β-glucuronidase (GUS) gene sequence (with no homology with plant DNA) was used. Furthermore, the phytoene desaturase (PDS) gene was used to visually detect the proper time for pathogen inoculation (Fig. S2). VIGS of GABAT1 increased susceptibility to B. cinerea in sitiens but not in the wild-type (Fig. 1c). However, silencing of GABAT1 imposed severe physiological disturbances in the mutant, causing the death of some sitiens seedlings (c. 30%) and a hampered growth phenotype in the remaining plantlets, whereas such a negative effect was not observed in GABAT1-silenced wild-type plants (Fig. 1d). To avoid the severe physiological effect of GABAT1 silencing, we tried to inhibit the shunt in sitiens plants pharmacologically. Application of a potent GABA-shunt inhibitor, 4-hydroxybenzaldehyde (HBA; Tao et al., 2006), resulted in enhanced susceptibility in the mutant (Fig. 1e). Both silencing of GABAT1 and HBA treatment ultimately subdued the mutant's capability to contain the pathogen, while the early epidermal HR-like symptoms and cell wall fortifications appeared intact, resulting in an uncommon susceptible phenotype. The symptoms observed were quite similar to the resistance responses in the control plants up to 2–3 d postinoculation (dpi), and then spreading mycelial tissue colonization outside of the inoculation site occurred afterwards (Fig. 2a (HBA, TRV2:GABAT1), c; the phenotype of the HBA-treated wild-type is shown in Fig. S3). Conversely, resistant control plants could effectively arrest pathogen growth within the inoculation site by forming a mesophyllic HR-like ring, differing from the epidermal HR, which appeared after 3 dpi as a ring encircling the inoculation area (Fig. 2a (Ctrl), b).
Exogenous application of GABA decreases susceptibility to B. cinerea in wild-type tomato
To further ascertain whether GABA metabolism plays a role in the resistance mechanism against B. cinerea, the effect of exogenous GABA application on the defense responses of both sitiens and wild-type tomato plants was analyzed. Interestingly, wild-type leaf disks incubated overnight in solutions of 1, 5 and 10 mM GABA showed partial resistance to B. cinerea, while no significant change could be seen in GABA-treated sitiens compared with the untreated sitiens plants. However, simultaneous application of GABA and the GABA-shunt inhibitor HBA (1 mM) led to increased disease severity in both the wild-type and the mutant (Fig. 3a; only data of 5 mM GABA treatment are shown). Further microscopic analysis on the GABA-induced resistance in wild-type leaves revealed that pathogen growth was suppressed in the cell layer beneath the epidermis, forming a mesophyllic HR-like ring that surrounded the slightly spreading primary lesion (Fig. 3b), quite similar to what was observed in sitiens during the later stages of the infection (Fig. 2b). To examine the direct effect of GABA on the growth of B. cinerea, the fungus was grown on salt base medium with either sucrose (as carbon source) or GABA (as potential carbon and nitrogen source) or sucrose + GABA. The fungus grew and sporulated on the sucrose + GABA medium, indicating that GABA could serve as the sole nitrogen source and had no direct negative impact on fungal growth (Fig. 3c).
Metabolic analysis of GABA fluctuations during the B. cinerea–tomato interaction
To investigate how GABA metabolism is modulated upon pathogen invasion in tomato, alterations in the concentrations of GABA were measured at different time points after inoculation using a specific ‘ultra performance liquid chromatography’ (UPLC) method. As shown in Fig. 3(d), GABA concentrations showed a sharp rise at 8 hpi in sitiens, which was totally consumed after 16 hpi, followed by a gradual increase until 48 hpi. Conversely, GABA concentrations in wild-type plants showed a marginal increase at 24 hpi followed by a sudden drop at 48 hpi (Fig. 3d). Absolute GABA concentrations, however, were higher in wild-type than in sitiens plants (Table S3).
GS contributes to the defense mechanism against B. cinerea in sitiens
The GS/GOGAT cycle constitutes an important metabolic node with a central position in plant amino acid metabolism via which the amino acid glutamate, the main precursor for GABA biosynthesis, is continuously metabolized (Lam et al., 1996). Therefore, we next focused on GS, the key enzyme of the GS/GOGAT cycle that carries out ammonium reassimilation (Cren & Hirel, 1999), and potentially contributes to plant defense (Tavernier et al., 2007). Total GS enzymatic activity in sitiens started to increase from 8 hpi onwards and peaked, at 24 hpi, at approximately six times the levels found in noninoculated controls (Fig. 4a). Application of methionine sulfoximine (MSO), a selective inhibitor of GS (Takeba, 1984), increased susceptibility in sitiens (Fig. 4b), resulting in promoted cell death, measured by the amount of electrolyte leaked from the inoculated tissue (Kawasaki et al., 2005; Fig. S4).
Interestingly, the rapid increase in the expression of the GABA shunt comprising genes in sitiens (Fig. 1a) was paralleled by an early up-regulation of cytosolic GS (GS1). Save for an 8 h delay, GS1 expression in the wild-type mirrored the profile observed for sitiens, while expression of the chloroplastic isoform (GS2) rapidly decreased in both sitiens and the wild-type (Fig. 4c). To determine whether GS1 was involved in the sitiens defense response against B. cinerea, GS1-silenced sitiens plants were infected and found to exhibit severe susceptibility to B. cinerea. (Fig. 4d). Both silencing of GS1 and MSO treatment impaired the ability of the mutant to arrest the pathogen, with symptoms very similar to those resulting from GABAT1 silencing and GABA-shunt inhibition (Fig. 2a (MSO, TRV2:GS1)).
Sitiens does not undergo B. cinerea-induced senescence
Next, the expression pattern of the mitochondrial glutamate dehydrogenase (GDH1), which links the GS/GOGAT and the TCA cycle and is known as a molecular marker for stress-induced senescence (Masclaux-Daubresse et al., 2002; Pageau et al., 2006), was monitored. Despite an up-regulation in sitiens at 8 hpi, the relative gene expression level did not alter significantly at later time points (16 and 24 hpi), whereas there was a large rise in GDH transcriptional level in the infected wild-type (Fig. 5a). To further study the difference in the process of senescence in the two genotypes, protein degradation during the course of the infection was also monitored. In sitiens, no clear change was detected in the total protein content of infected leaves compared with the control until 3 dpi. However, the total protein content of the wild-type markedly decreased (Fig. 5b). Furthermore, sitiens plants did not exhibit any visually detectable symptoms of senescence until 7 dpi (Fig. 5c). From this time onward, symptoms of Chl degradation (pale yellowish color) were uniformly seen around the infection site, while the inoculation site (leaf area underneath the inoculation droplet), surrounded by the aforementioned HR ring, remained green. The green area resembled a phenomenon typically associated with biotrophic interactions, the so-called ‘green bionissia’ (Walters et al., 2008; Fig. 5d). The ongoing photosynthesis in the green region at 21 dpi was confirmed by live-cell imaging when gas bubbles were observed being released through the stomata after light stimulation (Video S1). Interestingly, the pathogen contained within the boundaries was still viable as it was successfully recultured from beneath the inoculation droplet at 21 dpi.
The phenylpropanoid pathway plays a pivotal role in the resistance response of sitiens against B. cinerea
We have previously shown a strong induction of PALactivity in sitiens (at 16 hpi) in response to B. cinerea (Audenaert et al., 2002). However, the importance of the phenylpropanoid pathway in the resistance response observed in the mutant remained unclear. The function of phenylpropanoid metabolism in plant resistance mechanisms is diverse, including the strengthening of cell walls through deposition of phenolic compounds (Hartley & Ford, 1989; Dixon et al., 2002). PAL activity is also known as a major secondary source of ammonium production in plant cells, entailing an influx of ammonium into the cytosol as the inevitable byproduct of the enzyme activity (Bernard & Habash, 2009). This premise led us to hypothesize that the ammonium generated by the strong overactivation of PAL might be reassimilated by the GS enzyme, linking the rapid epidermal cell wall fortification studied previously (Asselbergh et al., 2007) with the observed alterations in the GS/GOGAT-GABA shunt in the mutant. Therefore, the involvement of the phenylpropanoid pathway in the sitiens–B. cinerea interaction was further analyzed.
The relative expression of tomato PAL5 (accession no. M90692), the main isoform reported to be induced in response to biotic stresses in tomato (Chang et al., 2008), was next monitored in both sitiens and wild-type plants. The relative expression level of PAL5 was higher in B. cinerea-infected sitiens plants (3.3 at 16 hpi and 4.8 at 24 hpi) than in infected wild-type plants (0.73 at 16 hpi and 2.14 at 24 hpi; Fig. 6a). Blast analysis revealed a GenBank tomato cDNA entry (accession no. AK320988), which had 81 and 80% identity at the protein level (E value = 0) with tomato PAL1 (accession no. P26600) and PAL5 (accession no. P35511) proteins, respectively, suggesting that it may encode a new tomato PAL isoform (hereafter referred to as PAL7; see Fig. S5 for more information). In comparison with PAL5, the presumable PAL7 is expressed faster and more strongly in the inoculated mutant. Interestingly, PAL7 showed a clear parallel expression pattern with GS1 and the GABA-shunt genes (Fig. 6a). Furthermore, silencing of PAL7 in sitiens increased susceptibility to B. cinerea (Fig. 6b). However, the actual existence and functionality of PAL7 protein remain to be confirmed.
Further enzymatic analysis of PAL activity in sitiens demonstrated a pronounced increase in the overall activity of the enzyme at 8 hpi that was still present at 16 and 24 hpi (Fig. 6c). In wild-type plants, PAL activity increased during infection to only a fraction of the levels in sitiens, confirming the importance of the phenylpropanoid pathway in regulating resistance of tomato to B. cinerea-sitiens. Accordingly, epifluorescence (EF) microscopy revealed either epidermal (12 hpi) or mesophyllic deposition of phenolic compounds (24 hpi; Fig. 6d). Furthermore, independent application of two different PAL inhibitors, α-aminooxyacetic acid (AOA) and α-aminooxy-β-phenylpropanoic acid (AOPP; Amrhein & Gödeke, 1977), gave similar results and led to extreme susceptibility in both sitiens and the wild-type (Fig. 7a – only AOPP data are shown). EF microscopy confirmed the attenuating effect of PAL inhibition on the deposition of phenolic compounds in epidermal cell walls of the mutant (Fig. 7b). However, using DAB to visualize H2O2 production at the inoculation site revealed that PAL inhibition did not affect the early epidermal accumulation of H2O2 typically seen in Botrytis-infected sitiens (Asselbergh et al., 2007), suggesting an upstream role for ROS spiking in the sitiens–Botrytis interaction (Fig. 7c).
In this study, the sitiens–B. cinerea pathosystem was used as a model to understand the mechanisms needed to effectively resist a necrotrophic pathogen. The focus was mainly directed toward the role of GABA metabolism in the sitiens defense response, inspired by previous studies (Asselbergh et al., 2007; Brauc et al., 2011). The results indicate that a timely overactivation of the GS/GOGAT cycle and the GABA shunt maintains cell viability, slowing down senescence in the site of primary invasion. Induction of such a cell death-alleviating mechanism in the cells surrounding the epidermal sites penetrated by B. cinerea may also explain how the HR-mediated defense response observed in sitiens (Asselbergh et al., 2007) can effectively suppress a necrotrophic pathogen.
In angiosperms, there are two distinct GS isoforms, cytosolic (GS1) and chloroplastic (GS2), which feed ammonium in the GS/GOGAT cycle. GS1 is commonly believed to be involved in ammonium reassimilation during natural and stress-induced senescence, as the GS1 transcript is known as a putative senescence-associated gene (SAG)/marker (Pageau et al., 2006). The chloroplastic isoform (GS2), though, plays a crucial role in the assimilation of ammonium obtained from nitrate reduction and photorespiration (Perez-Garcia et al., 1998; Bernard & Habash, 2009). Coinduction of SAGs and defense-related genes has previously been observed in many cases, highlighting the necessity for efficient protection against opportunistic pathogens during the critical process of senescence (Quirino et al., 2000). For instance, overlapping expression of GS1 and PAL has been reported in a number of studies on plant tissues undergoing some form of senescence, such as lignin depositing sclerenchyma cells in naturally maturing leaf blades of rice (Sakurai et al., 2001), and senescing leaves of Phaseolus vulgaris in response to Colletotrichum lindemuthianum (Tavernier et al., 2007). Accordingly, the concurrent high levels of GS1/PAL7 expression and GS/PAL activities observed in inoculated sitiens suggest a metabolic link between the phenylpropanoid pathway and the GS/GOGAT cycle, presumably through reassimilation of PAL-derived ammonium. However, according to the results presented here, the rapid up-regulation of GS1 in sitiens seems not to be associated with senescence, as senescence is delayed in the mutant, particularly during the early stages of the interaction. Under the pathogen challenge, and considering the down-regulation of GS2, the cytosolic isoform might play a compensatory role to ensure the critical functionality of the GS-GOGAT cycle. Therefore, it could be hypothesized that overactivation of the GS-GOGAT cycle is not merely involved in scavenging the unwanted byproduct of a defensive pathway, but, more importantly, might be an influential part of an activated anti-cell death mechanism in sitiens. This was further confirmed when either VIGS of GS1 or inhibition of GS by MSO promoted cell death and abolished the capacity of sitiens to contain the pathogen during the later stages of the interaction (Fig. 2a). However, a late up-regulation of GS1 was observed in wild-type plants, which may be linked to the putative senescence-related role of the gene in agreement with the strong senescence occurring in these plants (Fig. 5c).
Besides functioning as the basic step in forming different amino acids, glutamate generation via the GS-GOGAT cycle serves as a key source for various compounds with possible roles in plant–pathogen interaction, such as GABA (Galili et al., 2001; Brauc et al., 2011). Cytosolic decarboxylation of glutamate by GAD is the main route for GABA biosynthesis (Baum et al., 1996), as well as a controlling point in the GABA shunt (Busch & Fromm, 1999). It is also known that this route has a regulatory role in central C : N metabolism by providing a major link between amino acid metabolism and the TCA cycle (Fait et al., 2008). The genes encoding GABA-shunt enzymes exhibited in sitiens an early up-regulation in response to the pathogen, concomitant to the PAL7 and GS1 expression pattern, suggesting a concerted interplay between defense response and primary metabolism in the mutant. Further evidence of this interplay is provided by the UPLC analysis, showing an increase in GABA concentrations in the resistant sitiens mutant at 8 hpi, followed by a considerable consumption of the amino acid. Since absolute GABA concentrations are higher in wild-type plants than in the sitiens mutant, it seems that, rather than the concentration of GABA per se, temporal activation of the GABA shunt is important in defense.
Under oxidative stress conditions, the TCA enzymes aconitase, succinyl-CoA ligase and the NADH-generating α-ketoglutarate dehydrogenase are deactivated, resulting in accelerated cell death in the tissue undergoing stress (Sweetlove et al., 2002). Although involvement of the GABA shunt in maintaining redox equilibrium during plant responses to abiotic stresses or natural senescence has gained some attention (Ansari et al., 2005; Fait et al., 2005; Shelp et al., 2012), the literature relating to the cell death-alleviating role of the GABA shunt in plant responses to pathogen-induced oxidative stress is scarce. According to the data presented here, overactivation of the GABA shunt in sitiens plays a vital role in resistance to B. cinerea. Likewise, exogenous application of GABA to the wild-type could effectively restrict the pathogen progress by forming an HR-like ring around the spreading lesion (Fig. 3b), mimicking the mesophyllic ring observed as the secondary wave of defense in sitiens (Fig. 2a). Taken together, it could be hypothesized that timely activation of the GABA shunt in sitiens might restrict the extent of cell death caused by the H2O2-mediated defense response to B. cinerea, particularly in cells in the vicinity of the pathogen penetration sites. The redox regulating effect of the GABA shunt may be explained by generation of NADH via SSADH and/or by ensuring the functionality of the TCA cycle, under oxidative stress, through bypassing the ROS-sensitive enzymes of the cycle. This hypothesis is in agreement with the report of Wu et al. (2006), suggesting that the GABA shunt may play a role in restricting cell death in both the incompatible and compatible interaction of rice with the blast fungus Magnaporte oryzae. Likewise, we have recently proposed a model describing how overactivation of metabolic pathways that can maintain the functionality of the GS/GOGAT and TCA cycle in invaded cells may function as anti-cell death defense strategy, termed as ‘endurance’, resulting in resistance against necrotrophic pathogens (Seifi et al., 2013).
Defense responses in plants are known to be highly energy-demanding processes (Heil & Bostock, 2002; Berger et al., 2007), heavily draining the TCA cycle-generated energy and intermediates (Kinnersley & Turano, 2000; Bolton, 2009). The TCA cycle plays a crucial anabolic role in supporting the costly defense-related metabolic pathways, such as stress-induced phenylpropanoid metabolism, which may consume up to 20% of the total photosynthetic carbon in the plant (Dennis & Blakeley, 1995). This huge demand highlights the necessity of anaplerotic reactions (‘filling-up reactions’; Kornberg, 1965) to replenish the cycle and ensure its constant functionality during such circumstances (Kinnersley & Turano, 2000). Under pathogen invasion pressure, the GABA shunt might also operate as an important anaplerotic route to the TCA cycle, providing carbon skeleton in the form of succinate (Michaeli et al., 2011). Accordingly, the race-nonspecific Lr34-mediated resistance response against the wheat leaf rust pathogen Puccinia triticina has been shown to be considerably energy-intensive, entailing concurrent up-regulation of the TCA cycle and the GABA shunt (Bolton et al., 2008). The results presented here may also support this hypothesis, suggesting concurrent overactivation of the energy-intensive phenylpropanoid pathway and the GABA shunt during the early stages of the sitiens–B. cinerea interaction. In this context, and regarding the observed delayed senescence in the mutant, the early and moderate transcriptional up-regulation of GDH1 in sitiens might be associated with the PAL-GS1 pattern and the deaminating activity of the enzyme, forming 2-oxoglutarate from glutamate, functioning as another anaplerotic entry point into the TCA cycle. However, the late and substantial increase of GDH in the wild-type is seemingly associated with the aminating activity of the enzyme, exhausting the TCA cycle and facilitating cell death, as suggested previously (Pageau et al., 2006).
It seems that the lower concentrations of ABA in sitiens lead to a metabolic state in which the process of senescence is slowed down (Fig. 5c, Sit-Mock). Accordingly, ABA was shown to play a major role in the regulation of leaf senescence in Lilium leaves (Arrom & Munné-Bosch, 2012). Interestingly, in the inoculated sitiens, cell viability and photosynthesis are retained within the pathogen-infested area (beneath the inoculation droplet) until at least 21 dpi, while still containing viable B. cinerea mycelia. The term ‘green bionissia’ was first defined by Walters et al. (2008) as a ‘localized green area of host tissue, in which both host and pathogen cells are alive, surrounded by yellow, senescent host tissue’. This is typically seen in plants invaded by obligate biotrophs, such as rusts and mildews, benefiting the pathogen by prolonging availability of a living nutrient source. Here, however, we report that ABA deficiency in the sitiens mutant of tomato results in a rare type of green bionissia, encircled within a mesophyllic HR-like ring, functioning as a resistance mechanism to retard the necrotrophic pathogen-induced senescence in the invaded area. In wild-type tomato plants, none of the anti-senescence defense mechanisms discussed here were observed at early time points, suggesting that they are suppressed by basal ABA concentrations in the susceptible interaction. It should be noted that exogenous application of ABA suppresses all defense responses in sitiens (Asselbergh et al., 2007; Curvers et al., 2010). However, how exactly ABA may negatively control such defense mechanisms still remains unanswered.
In summary, through our model sitiens–B. cinerea pathosystem, molecular features of a biphasic resistance mechanism in response to a true necrotrophic invasion strategy were eventually unraveled. We propose that, in addition to the epidermal HR-mediated defense response (Asselbergh et al., 2007), timely reconfiguration of the host central C : N metabolism plays a vital role in shaping a secondary line of defense response in the resistant mutant (Fig. 8). Initially, the ‘fast pathogen arrest’ phase, consisting of a rapid epidermal ROS accumulation followed by cell wall fortifications, in which the phenylpropanoid pathway plays a vital role, arrests the penetrating pathogen in the epidermis. Concurrently, a ‘maintenance’ phase of basic metabolism begins presumably in the mesophyll cells surrounding the infection site, tightly controlling the extent of the epidermal HR-mediated defense response, and slowing down the infection-induced senescence. The ammonium produced by PAL activity is reassimilated by the cytosolic GS, reutilizing the byproduct of the primary phase to reduce the damage to the GS/GOGAT cycle in the absence of the chloroplastic isoform. Subsequently, the GS-GOGAT-generated carbon skeleton is supplied into the TCA cycle via the overactivated GABA shunt, replenishing the redox-regulating and anabolically critical Krebs cycle under the constant pressure of infectious mycelia of B. cinerea. Ultimately, activation of such a survival mechanism in the vicinity of the penetration site culminates in forming a rarely seen antinecrotrophic type of ‘green bionissia’ in sitiens, enabling an HR-mediated defense strategy to be ultimately effective against an HR-favoring necrotrophic pathogen.
This work was supported by grants from the Fund for Scientific Research Flanders (FWO grants 3G052607 and 3G000210) and by a postdoctoral fellowship of the Research Foundation–Flanders (to D.D.V.). We are very grateful to Sandra Villaume for technical assistance, and to Dr Mansour Karimi and Pooneh Kalhorzadeh for their support and invaluable remarks during different stages of the research. We also thank Prof. Matthias Hahn and Dr Michaela Leroch for providing GFP-expressing B. cinerea. Our heartfelt appreciation also goes out to Gareth Hill for proofreading of the manuscript.