Uncoupling of reactive oxygen species accumulation and defence signalling in the metal hyperaccumulator plant Noccaea caerulescens

Authors


Summary

  • The metal hyperaccumulator plant Noccaea caerulescens is protected from disease by the accumulation of high concentrations of metals in its aerial tissues, which are toxic to many pathogens. As these metals can lead to the production of damaging reactive oxygen species (ROS), metal hyperaccumulator plants have developed highly effective ROS tolerance mechanisms, which might quench ROS-based signals. We therefore investigated whether metal accumulation alters defence signalling via ROS in this plant.
  • We studied the effect of zinc (Zn) accumulation by N. caerulescens on pathogen-induced ROS production, salicylic acid accumulation and downstream defence responses, such as callose deposition and pathogenesis-related (PR) gene expression, to the bacterial pathogen Pseudomonas syringae pv. maculicola.
  • The accumulation of Zn caused increased superoxide production in N. caerulescens, but inoculation with P. syringae did not elicit the defensive oxidative burst typical of most plants. Defences dependent on signalling through ROS (callose and PR gene expression) were also modified or absent in N. caerulescens, whereas salicylic acid production in response to infection was retained.
  • These observations suggest that metal hyperaccumulation is incompatible with defence signalling through ROS and that, as metal hyperaccumulation became effective as a form of elemental defence, normal defence responses became progressively uncoupled from ROS signalling in N. caerulescens.

Introduction

Signalling via reactive oxygen species (ROS) is widely regarded to be central to disease resistance in plants (Mehdy, 1994; Wojtaszek, 1997; Fobert & Després, 2005; Torres et al., 2006). The oxidative burst is one of the earliest plant responses to the detection of pathogen-associated molecular patterns (PAMPs; Navarro et al., 2004) and is required for downstream plant defences, such as callose deposition (Vellosillo et al., 2010), salicylic acid (SA) signalling and the expression of pathogenesis-related (PR) genes (Alvarez et al., 1998). In Arabidopsis thaliana and tobacco, a peak in H2O2 production within 6 h is associated with the detection of pathogens (Wojtaszek, 1997), whereas a later peak is seen in incompatible reactions and is involved in the signalling of pathogen-induced cell death and systemic acquired resistance (SAR; Fobert & Després, 2005). Similarly, superoxide is produced in response to pathogens by plants such as A. thaliana, tobacco and tomato (Mehdy, 1994).

ROS production may also be the toxic consequence of abiotic stresses that affect cell homeostasis (Choudry et al., 2013), creating a potential conflict between ROS functionality and suppression, whose reconciliation is not currently understood. Among many potential triggers, toxic excess of metal ions can, both directly and indirectly, lead to the production of ROS (Garnier et al., 2006; Miethke & Marahiel, 2007). In an extreme case of adaptation to stressful environmental conditions, metal hyperaccumulator plants not only thrive on metal-enriched soils, but also accumulate metals to exceptionally high concentrations in their aerial tissues without suffering toxicity (Pollard, 2000; Reeves & Baker, 2000; Pollard et al., 2002; Verbruggen et al., 2009; Krämer, 2010). Thus, these plants must be able to withstand metal-induced ROS. Comparison of the transcriptomes of these plants with those of related non-accumulators has uncovered many differences in gene expression. For example, using DNA microarrays, Becher et al. (2004) found differences in the expression of genes involved in glutathione metabolism between the hyperaccumulator Arabidopsis halleri and the congeneric model plant A. thaliana, whereas similar studies by Weber et al. (2004) and Chiang et al. (2006) with these species, and by Hammond et al. (2006), comparing Thlaspi (now Noccaea) caerulescens with the non-accumulator Thlaspi arvense, uncovered elevated expression and activity of catalase and peroxidase in the hyperaccumulator species. Distinctive biochemical features of hyperaccumulator plants include an elevated production of antioxidants, such as glutathione (Freeman et al., 2004; Jin et al., 2008; Tian et al., 2011), organic acids (Wójcik et al., 2006) or amino acids (Ingle et al., 2005; Sun et al., 2007b), and high activity of antioxidant enzymes, such as superoxide dismutase (Srivastava et al., 2005; Sharma & Dietz, 2008), catalase (Boominathan & Doran, 2003; Srivastava et al., 2005; Sharma & Dietz, 2008) and ascorbate peroxidase (Srivastava et al., 2005; Chiang et al., 2006; Sharma & Dietz, 2008; Yang et al., 2009), which are thought to provide protection from metal-induced oxidative stress. There is evidence that the hyperaccumulated metals in these plants can lead to increased disease resistance (Boyd et al., 1994; Ghaderian et al., 2000; Hanson et al., 2003), and that this occurs via the direct ‘elemental’ toxicity of metals to pathogens, with hyperaccumulator plants deprived of metals showing increased susceptibility to disease (Fones et al., 2010).

Here, we use the ‘Prayon’ population of N. caerulescens to investigate whether defence signalling via ROS is incompatible with metal hypertolerance and accumulation. We propose that, as the ancestors of N. caerulescens evolved to accumulate higher concentrations of metals, more antioxidant capacity would have been needed to counteract metal-induced oxidative stress. Constitutively high levels of antioxidants may thus have disrupted, and led to the eventual uncoupling of, ROS-based defence signalling, leaving the plants increasingly dependent on accumulated metals for defence. We show that the metal hyperaccumulator plant Noccaea (=Thlaspi) caerulescens has undergone a fundamental reorganization of defence signalling when compared with its non-accumulator relative Arabidopsis thaliana, with the loss of the oxidative burst in response to pathogen infection and of associated downstream defence responses, such as callose deposition and PR gene expression. These results provide a new perspective on the evolutionary origins of metal hyperaccumulation and the emergence of this complex physiological trait.

Materials and Methods

Plant growth conditions, bacterial strains and inoculations

Noccaea caerulescens (J. Presl & C. Presl) F.K. Mey. (Prayon population) was grown hydroponically on modified 0.1-strength Hoagland solution (Fones et al., 2010) containing 0.04, 10, 30 or 300 μM ZnSO4 in a glasshouse. Sodium vapour lamps provided 14 h of supplemental lighting per day. Plants were maintained at minimum temperatures of 14°C (night) and 24°C (day). Arabidopsis thaliana (L.) Heynh. Col-0 plants were sown on peat-based compost and grown under the same conditions. The zinc (Zn) concentrations used did not adversely affect the growth of N. caerulescens; plants grown on the highest Zn concentration (300 μM) had significantly greater root and shoot biomass than those grown on the lowest (0.04 μM), but neither differed significantly from the 10 μM Zn control condition (Supporting Information Fig. S1). For all experiments, 10-wk-old N. caerulescens and 6-wk-old A. thaliana plants were used. At these ages, the two species were at similar physiological stages, each showing a large rosette of leaves, but not beginning the transition to the generative phase.

For plant inoculations, Pseudomonas syringae pv. maculicola (Psm) was streaked onto Luria–Bertani (LB) agar from stocks kept at −80°C, and cultured overnight at 28°C, before resuspension in 10 mM MgCl2 at 107 colony-forming units (CFU) ml−1 (N. caerulescens) or 105 CFU ml−1 (A. thaliana).

Measurement of hydrogen peroxide (H2O2), superoxide and callose production

Leaves were inoculated with Psm M4 (Debener et al., 1991) at 107 CFU ml−1 (Noccaea caerulescens assays) or 105 CFU ml−1 (Arabidopsis thaliana assays) in 10 mM MgCl2, sterile 10 mM MgCl2, or not inoculated. For callose deposition assays, hrpS, hrcC and wild-type Pseudomonas syringae pv. tomato (Pst) DC3000 strains (Yuan & He, 1996) were used as controls. Lower inoculum densities were used for A. thaliana assays, as wild-type P. syringae strains caused leaf collapse within 24 h at higher densities. Leaves were harvested and stained after 2 h (superoxide), 6 h (H2O2) or 24 h (callose). For each treatment and time point, a minimum of three leaves was used from each of three plants.

For the measurement of H2O2 or superoxide, harvested leaves were submerged in 0.1% (w/v) 3,3′-diaminobenzidine (DAB) (Sang et al., 2001) or nitroblue tetrazolium (NBT) (Love et al., 2005) overnight, before clearing by boiling in methanol. Cleared, stained leaves were photographed using a Leica MZFL III (Leica Ltd, Milton Keynes, UK) macroscope with a Coolsnap Camera (RS Photometrics, Surrey, BC, Canada) and Q-Capture Pro software (MediaCybernetics, Marlow, UK). Images were exported as .bit files using ImageProPlus (MediaCybernetics, Marlow, Bucking-hamshire, UK) for analysis by custom image-analysis software developed in Fortran (script available on request; Fig. S2, Methods S1).

For callose assays, inoculated leaves were first cleared in methanol and then stained by submerging in 0.01% (w/v) aniline blue (Ton et al., 2005). Leaves were then visualized using epifluorescence microscopy. Callose deposits were counted in three randomly selected fields of view per leaf.

Measurements of SA

Leaves were inoculated with either 10 mM MgCl2 or Psm suspended in 10 mM MgCl2 at 107 CFU ml−1 (N. caerulescens) or 105 CFU ml−1 (A. thaliana). For LC/MS/MS, leaves were harvested by snap freezing in liquid nitrogen, together with uninoculated controls, at 24 h after inoculation, and lyophilized. Lyophilized tissue was prepared according to Forcat et al. (2008).

For assays of apoplastic SA, the SA biosensor Acinetobacter ADPWH-lux (Huang et al., 2006) was used. Apoplastic fluid was extracted from leaves 24 h after inoculation, and the degree of apoplast dilution was estimated as described by Rico & Preston (2008). Apoplast extract was lyophilized and resuspended in an appropriate volume of distilled water to return it to its estimated concentration in planta. The SA biosensor Acinetobacter ADPWH-lux was grown overnight on LB agar, subcultured into LB broth and grown overnight at 28°C. Five millilitres of the resulting culture were added to 100 ml of LB broth and shaken at 37°C until the culture reached an optical density at 600 nm (OD600) of 0.4. One hundred microlitre aliquots of this culture were added to 96-well microplates containing 100-μl samples of apoplast extract diluted 1 : 10 (eight technical replicates per treatment) and calibration ladders consisting of a range of SA concentrations from 0 to 1 μM in 100 μl of 10 mM MgCl2. Luminescence was measured after incubation of the biosensor for 1 h at room temperature using an Infinite 200 plate reader (Tecan Group Ltd, Männedorf, Switzerland). Three technical replicates of each treatment were analysed. The entire experiment was performed three times with independent sets of plants.

Before these analyses, dilution series of leaf apoplast extracts from A. thaliana, and from N. caerulescens plants grown on the highest (300 μM) and lowest (0.04 μM) Zn treatments used, were added to SA calibration ladders and compared with control ladders to determine the maximum possible concentration of apoplast extract that could be used without risk of biosensor quenching by Zn or other extract constituents. A 1 : 10 dilution of apoplast with 10 mM MgCl2 was found to be appropriate.

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) of SA pathway genes in N. caerulescens and A. thaliana

For RNA extraction, leaves were harvested by snap freezing in liquid nitrogen. Samples were stored at −80°C for a maximum of 1 wk before homogenization in liquid nitrogen for RNA extraction from 50 μg samples of tissue using the RNeasy RNA extraction kit (Qiagen, UK). Extracted RNA was further purified by precipitation in 100 μl of 8 M LiCl (Sigma, UK) overnight. Pellets were washed twice in 70% (v/v) ethanol and resuspended in 30 μl of ultrapure water. RNA concentration and purity were measured using a Nanodrop-1000 spectrophotometer (Thermo Scientific, Loughborough, UK) and integrity was checked by electrophoresis on an ethidium bromide-stained agarose gel. RNA was then treated with DNAse I (Fermentas, Loughborough, UK) according to the manufacturer's specifications. cDNA was prepared from 0.5 μg of RNA using the Bioline cDNA synthesis kit (Bioline, London, UK) with oligo-dT primer. cDNA integrity was checked by gel electrophoresis. cDNA was stored at −20°C.

qRT primers were designed based on the sequences of the A. thaliana genes ICS1 (AT1G74710.1), PR-1 (AT2G14610.1) and PR-2 (AT3G57260.1). qRT primers were designed so that the product would encompass an intron for the detection of contaminating gDNA. Primers were tested on gDNA of N. caerulescens and A. thaliana before use. PCRs were run with these primers using cDNA templates to check for the amplification of larger, gDNA fragments, indicating gDNA contamination. Primers were manufactured by MWG (Ebersberg, Germany; sequences are given in Table S1). qRT-PCR was performed in clear 96-well plates (Ambion, Huntingdon, UK) using 12.5 μl of SYBR green (Applied Biosystems, Warrington, UK), 7.5 μl of template cDNA and 2.5 μl of each primer at 3 μM in a 7300 Realtime PCR machine (Applied Biosystems), according to the thermal cycle in Table S2. Analysis of gene expression was performed by calibration with a standard curve of template concentration (Larionov et al., 2005). Standard curves for gene expression were produced using serial dilutions of pooled cDNA from all treatments. For each gene, it was possible to obtain standard curves with R2 values of 0.98 or more (Fig. S3), which were used to calculate the relative expression of the gene of interest and control genes. For each point on the standard curve, three technical replicates were used. For each gene of interest, three technical replicates of each cDNA sample plus one control, in which reverse transcriptase had been excluded from the cDNA synthesis process, were used. Four independent reference genes (ELF-1α, β-TUBULIN, ORNITHINE TRANSCARBAMYLASE and UBIQUITIN) were assessed, and the expression of the target genes was normalized to the geometric mean expression of the reference genes. No template controls were used for each primer pair. Dissociation curves were inspected for each primer–sample combination and found to show no evidence of primer dimers or mispriming.

Statistics

The P values given relate to analysis of variance (ANOVA) tests carried out with Bonferroni simultaneous tests. ANOVA analyses, as well as simultaneous Bonferroni comparisons, were carried out in Minitab 12.21 (Minitab Inc., Coventry, UK). In the figures, the data points whose means were not found to be significantly different in Bonferroni comparisons are marked with the same letter. No letters are appended to groups of data points whose means were not found to differ in the initial ANOVA tests, as Bonferroni simultaneous comparisons are not meaningful in these cases. Error bars representing standard error (SE) are shown in the figures; the data presented are means. Histograms and normal plots of the residuals, as well as residual vs fitted plots, were examined in Minitab to ensure that the assumptions of ANOVA were met.

Results

A model system for the study of N. caerulescens defence responses

In this work, we use the model system developed in our previous study (Fones et al., 2010), consisting of the Prayon population of N. caerulescens and the bacterial pathogen Psm. The Prayon population of N. caerulescens was chosen because it is a widely studied and well-characterized population that is native to Zn-contaminated soil and shows typical hyperaccumulation behaviour (Roosens et al., 2003; Milner & Kochian, 2008; Fones et al., 2010). Arabidopsis thaliana was chosen as a reference non-accumulator from the family Brassicaceae because, as one of the best-studied plants worldwide, extensive knowledge exists concerning its typical defence responses, including this plant's status as a model system for gene expression studies.

We have shown previously that Psm is a pathogen of N. caerulescens (Fones et al., 2010). In addition, Psm is a pathogen of the model plant A. thaliana (Debener et al., 1991). When inoculated into low-Zn (0.04 μM)-grown leaves of N. caerulescens at 106 CFU ml−1, Psm shows a 2000-fold increase in CFU ml−1 over 5 d, similar to two naturally occurring P. syringae strains isolated from N. caerulescens (Fones et al., 2010), which showed 2000- and 500-fold increases in population density over the same time period. The taxonomic identity of these strains was confirmed by sequence analysis of gyrB, rpoD and 16S rDNA (Yamamoto & Harayama, 1998; Yamamoto et al., 2000). In studies of both the endophytes of hyperaccumulators and the bacteria found in the metalliferous rhizosphere of these plants, metal-tolerant bacteria, including Pseudomonas strains, have also been recovered (Idris et al., 2004; Barzanti et al., 2007). Therefore, P. syringae can be considered to be an ecologically relevant pathogen of N. caerulescens.

The oxidative burst and callose deposition in response to P. syringae are altered or absent in N. caerulescens

DAB and NBT staining were used to visualize the production of H2O2 and superoxide, respectively, in leaves in response to the virulent bacterial pathogen Psm (Fones et al., 2010). In A. thaliana, a clear increase in H2O2 was observed after 6 h in leaves inoculated with Psm (Fig. 1a,b); superoxide levels also increased significantly in response to Psm by 2 h post-inoculation (Fig. 1c,d). By contrast, in N. caerulescens, uniformly low levels of DAB staining were observed with all inocula, even when 100-fold elevated cell densities were used for inoculation (Fig. 1a,b); NBT staining in N. caerulescens also showed no response to pathogen inoculation. Thus, in contrast with A. thaliana, N. caerulescens shows no detectable H2O2 or superoxide response to Psm over several hours post-infection. Nevertheless, molecular mechanisms leading to ROS accumulation (or at least superoxide production) are operative in N. caerulescens, as seen by the progressive increase in NBT staining as a response to increasing Zn concentration (Fig. 1d).

Figure 1.

Reactive oxygen species (ROS) production and callose deposition responses to Pseudomonas syringae are lacking in Noccaea caerulescens. Plants were grown in hydroponic medium supplemented with 0.04, 10, 30 or 300 μM zinc (Zn) and inoculated with either P. syringae pv. maculicola (Psm), 10 mM MgCl2 or left untreated. For callose assays, leaves were additionally inoculated with P. syringae pv. tomato DC3000 (Pst DC3000), and with insertion mutants of DC3000 lacking the ability to construct or express the type III secretion system used by P. syringae to suppress plant defences (hrcC and hrpS; Yuan & He, 1996). (a) 3,3′-Diaminobenzidine (DAB) staining (red–brown speckling) for H2O2. (c) Nitroblue tetrazolium (NBT) staining (blue speckling) for superoxide. (e) Aniline blue staining for callose, visible as punctate pale blue or white spots. Representative images are shown. Quantitative data, obtained by measuring the percentage stained area or by counting callose deposits, are shown in (b) (H2O2), (d) (superoxide) and (f) (callose). Data in (b, d, f) are averages from three independent experiments. In each of these experiments, three to five leaves (technical replicates) per inoculum were used from each of three plants (biological replicates). Mean values were calculated from technical replicates to give one value per biological replicate. ANOVAs with Bonferroni simultaneous tests were used to test for differences in percentage staining between different inocula within each Zn treatment and within Arabidopsis thaliana. Means differing from each other significantly (α = 0.05) are marked with different letters. Error bars indicate ± SE.

Another common defence response of A. thaliana to PAMPs is the deposition of callose papillae (Aist, 1976). Psm is able to suppress callose deposition by means of the type III secretion system (T3SS). Thus, mutants of the closely related pathogen Pst DC3000 lacking a functional T3SS (Yuan & He, 1996) were used as positive controls for callose elicitation (Hauck et al., 2003). These hrcC and hrpS mutants elicited substantial callose deposition in A. thaliana within 24 h. However, when tested in N. caerulescens, none of the inocula resulted in callose deposition during this period (Fig. 1e,f).

SA concentration, but not PR gene expression, is elevated in response to Psm in N. caerulescens

The plant hormone SA is known to be an important signal in local and systemic defences against pathogens (Loake & Grant, 2007). In both A. thaliana and N. caerulescens, inoculation with Psm at densities that induced chlorosis within 24 h resulted in significant increases in the SA content of whole-leaf tissue, as measured by LC/MS/MS (Fig. 2a). SA concentrations in N. caerulescens were, however, unaffected by Zn treatments sufficient to permit Zn hyperaccumulation. SA concentrations in apoplastic fluid also increased in response to Psm in both species (Fig. 2b), indicating that SA was present as a mobile, systemic signal. qRT-PCR confirmed that ISOCHORISMATE SYNTHASE (ICS1), a gene important in the synthesis of SA, was upregulated in both species in response to Psm inoculation (Fig. 3a). However, the genes PR-1 and PR-2, commonly used as markers of SA-signalled defence responses and upregulated in response to Psm in A. thaliana (Fig. 3b,c), were expressed at very low levels and showed no responsiveness to Psm inoculation in N. caerulescens (Fig. 3b,c).

Figure 2.

Salicylic acid (SA) is produced in response to pathogen inoculation, but not to zinc (Zn), in Noccaea caerulescens. SA was measured in whole-leaf tissue by liquid chromatography tandem mass spectrometry (LC/MS/MS) (a) and in apoplast using an Acinetobacter biosensor ADPWH-lux (Huang et al., 2006) (b). Leaves were inoculated with Pseudomonas syringae pv. maculicola (Psm) (dark blue), MgCl2 (mid-blue) or not treated (light blue). Data shown represent the average of three biologically independent samples (LC/MS/MS) or three independent experiments, each with eight technical replicates (biosensor). ANOVAs with Bonferroni simultaneous tests were used to test for differences in SA concentration between inocula within each Zn treatment and within Arabidopsis. Means differing from each other significantly (α = 0.05) are marked with different letters. Error bars indicate ± SE.

Figure 3.

Salicylic acid (SA)-responsive genes are not upregulated in response to Pseudomonas syringae pv. maculicola (Psm) inoculation in Noccaea caerulescens. RNA for quantitative reverse transcription-polymerase chain reaction (qRT-PCR) analysis was isolated from leaves of 10-wk-old N. caerulescens grown on either 0.04 μM (low) or 300 μM (high) zinc (Zn), and leaves of 6-wk-old Arabidopsis thaliana. Plants were subjected to no treatment (light blue), 10 mM MgCl2 mock inoculation (mid-blue) or inoculation with Psm in 10 mM MgCl2 (dark blue), 24 h before RNA extraction. The expression of target genes ICS1 (SA synthesis), PR-1 and PR-2 (SA responsive) is shown relative to that of the geometric mean expression of four reference genes (UBQ-10, OTC, ELF-1α and β-tubulin). For PR-1 and PR-2, the insets show the values obtained for N. caerulescens on expanded y-axes. Representative data from one of two or three biologically independent experiments showing similar patterns of gene expression are shown. In each of these experiments, three biological replicates, consisting of leaves from three different plants, were used. Error bars indicate ± SE.

Discussion

Metal hyperaccumulation in the crucifer N. caerulescens is known to be able to protect the plant from bacterial and fungal disease (Fones et al., 2010). In a previous study, we showed that the two highest Zn concentrations (30 and 300 μM) on which N. caerulescens was grown in this work produced plants containing sufficient Zn to be directly toxic to Psm. Here, however, we have focused on the plant's short-term responses to the detection of the pathogen. Our results demonstrate that N. caerulescens lacks some of the typical defence responses to pathogen invasion seen in A. thaliana and other plants, including the H2O2 and superoxide components of the ROS burst, callose deposition and PR gene expression. In this work, we studied the Prayon population of N. caerulescens, a native of Zn-contaminated soil, whose metal tolerance and hyperaccumulation capacities are typical for this species (Roosens et al., 2003; Fones et al., 2010). Further investigations, using a broader range of N. caerulescens populations or Noccaea species, would also be of great interest.

One possible explanation for the observations reported here is that, as hyperaccumulator plants acquired the ability to use metals for defence, they became less dependent on defence responses associated with the oxidative burst and downstream signalling pathways normally induced by pathogens. However, this explanation poses two major questions. First, does N. caerulescens experience an advantageous reduction in defence costs as a result of this trade-off between ‘conventional’ and metal-based defences? Second, is the loss of conventional pathogen-induced defences in N. caerulescens also a consequence of their incompatibility with other biochemical processes required for metal hyperaccumulation within the cell?

Here, we propose an evolutionary scenario, consistent with our experimental observations, that has the potential to address both of these points (Fig. 4). Initially, N. caerulescens or its ancestor, growing on metal-rich soils, began to accumulate metals, such as Zn. The accumulated metal may have led to oxidative stress, including the production of H2O2 and superoxide (Hall, 2001). Although Zn itself is not redox active under physiological conditions, it can cause oxidative stress, for example by displacement of redox-active iron from binding sites in proteins (Arrivault et al., 2006). As a result of elevated ROS, a number of pathogen defence responses, normally signalled by ROS, may have become constitutively activated (Mehdy, 1994; Wojtaszek, 1997; Fobert & Després, 2005). In combination with accumulated metal, this could have provided an effective but costly (van Hulten et al., 2006) defence against pathogens. Either this cost, or redundancy between metal-dependent and metal-independent defences, may have led to an uncoupling of the signal (ROS) and the response (anti-pathogen defences). Alternatively, ROS production may have been suppressed by the development of antioxidant-based metal tolerance mechanisms, such as increased production of glutathione (GSH) and ascorbate (Hall, 2001; Freeman et al., 2004), and increased activity of antioxidant enzymes (Boominathan & Doran, 2003). With accumulated metal providing an alternative defence against many pathogens (Fones et al., 2010), the uncoupling of the oxidative ROS burst and SA from defence signalling could have occurred, and downstream responses, such as callose deposition and PR gene expression, would also have been affected. It has also been reported that metal hyperaccumulation in these plants is associated with reduced investment in glucosinolates as feeding deterrents against herbivores (Tolrà et al., 2001; Noret et al., 2007). Such responses would leave the plants increasingly dependent on accumulated metals for defence. From this point forwards, an evolutionary arms race between plant metal accumulation and pathogen metal tolerance may have driven the emergence of the hyperaccumulation trait, in a manner similar to that suggested by Boyd (2012) in the theory he termed the ‘defensive enhancement’ hypothesis.

Figure 4.

Model for the evolution of reactive oxygen species (ROS) signalling and metal hyperaccumulation in Noccaea caerulescens. Schematic representation of the changes taking place during the evolution of hyperaccumulation, from an ancestral state similar to modern Arabidopsis thaliana, which does not accumulate metals in its normal environment (1). These plants show archetypal defence signalling through ROS. As the plants colonize and adapt to zinc-rich soils, accumulation of the metal begins to occur. This leads to elevated ROS in these plants (2). As metal tolerance mechanisms, including those based on antioxidants, develop, this elevated ROS is counteracted (3). Constitutively elevated ROS may constitutively induce defences normally only deployed on the detection of pathogens. The cost of this may confer a selective advantage to the suppression of ROS (4), preventing spurious defence signalling, or may select for the uncoupling of ROS from defence signalling (5). Additionally, the development of antioxidant-based metal tolerance mechanisms may impair defence signalling through ROS (6), leading to increased reliance on the protection provided by accumulated metals for defence (7). Reliance on metal toxicity for defence may have led to an evolutionary arms race between pathogen metal tolerance and plant metal accumulation, giving rise to the extreme hyperaccumulator phenotype seen in present-day species, such as N. caerulescens.

Is metal hyperaccumulation incompatible with defence signalling through ROS? One of the key differences between hyperaccumulator plants and their non-accumulating relatives, alongside changes such as improved transport, compartmentalization and the sequestration of metals, is a greater capacity for the neutralization of ROS. This increased capacity for ROS turnover may interfere with ROS signalling in defence, preventing the accumulation of H2O2. High superoxide concentrations were detected in Zn-treated N. caerulescens in the absence of pathogen inoculation, yet these plants showed no stress or growth impairment, and so must possess effective mechanisms for superoxide tolerance. Much of the toxicity of superoxide is a result of the production, via superoxide dismutase, of H2O2 (Scott et al., 1989). High superoxide levels might therefore require high catalase activity, dampening signalling via H2O2. Further, reduced GSH is known to have an important antioxidant role in metal tolerance in N. caerulescens (Freeman et al., 2004). Evidence suggests that depletion of the GSH pool for phytochelatin production, which is important for basal metal tolerance in A. thaliana, may be detrimental to the high degree of metal tolerance required in N. caerulescens (Sun et al., 2007a; Ebbs et al., 2008). In A. thaliana, GSH is also used in defence signalling via conjugation to breakdown products of indole glucosinolates by phytochelatin synthase (Clemens & Peršoh, 2009). High metal and ROS tolerance via GSH in N. caerulescens may be incompatible with this form of defence signalling. It is notable that the changes to pathogen-induced defences are not dependent on Zn availability, as N. caerulescens plants grown on low Zn nevertheless lack defensive signalling in response to Psm. Indeed, many biochemical components of cell homeostasis mechanisms are constitutively expressed in metal hyperaccumulator species, as these plants are typically endemic to metalliferous soils (Pence et al., 2000; Assunção et al., 2001; Freeman et al., 2005).

As pathogen-induced SA accumulation in N. caerulescens did not elicit PR gene expression, it is interesting that SA accumulation is retained (Fig. 2). Under a model of progressive loss of defence signalling mechanisms, it would be parsimonious to assert that pathogen-induced SA accumulation has simply not yet been lost, but it seems equally possible that SA retains a functional role in these plants. It has been proposed previously that SA has a role in nickel tolerance in hyperaccumulator species of Noccaea, being used as a signal to increase GSH biosynthesis (Boominathan & Doran, 2003; Jakab et al., 2005). However, Freeman et al. (2005) reported that SA levels were constitutively elevated in nickel hyperaccumulator plants, in contrast with the pathogen-inducible SA accumulation observed in this study, suggesting that this may be a feature specific to nickel-hyperaccumulating species and populations. Nevertheless, it remains possible that SA is performing non-defensive signalling roles in N. caerulescens. SA is known to be involved in, for example, SO2, salt and thermotolerance (Clarke et al., 2004; Jakab et al., 2005; Hao et al., 2011), with cross-talk occurring between these responses and defensive signalling (Clarke et al., 2004; Jakab et al., 2005; Hao et al., 2011); therefore, the loss of SA accumulation in defence signalling may be detrimental to its other functions.

It is possible that SA has a defensive function not tested here: we have demonstrated previously that the T3SS of Psm is essential for the infection of N. caerulescens (Fones et al., 2010), indicating that some T3SS-suppressed defences, which may be induced by SA, are retained. Interestingly, a recent publication by Llugany et al. (2013) found that concentrations of SA were increased in the cadmium hyperaccumulator Noccaea (=Thlaspi) praecox in response to metal exposure. Moreover, this hyperaccumulator showed a cadmium-dependent reduction in the induction of SA in response to biotic stress. Instead, plants treated with high cadmium showed induction of the jasmonic acid pathway. These findings support the concept that defence signalling in Noccaea hyperaccumulators has diverged from the typical responses known for A. thaliana. This work also raises the possibility that alternative signalling pathways may be used by Noccaea species in defence against pathogens, suggesting an important direction for future research.

In conclusion, we have shown that the metal hyperaccumulator species N. caerulescens responds to pathogen invasion in a manner very different from its non-accumulating relatives, constituting a notable departure from the current understanding of plant defence responses. Physiological incompatibility between metal hyperaccumulation and ROS-signalled plant defence responses, in conjunction with the protective effect of Zn hyperaccumulation against disease, provides a plausible explanation for the loss of ROS-based pathogen-induced defences in N. caerulescens.

Acknowledgements

We thank Joe McKenna for help with sample preparation for LC/MS/MS, Dr Harriet McWatters for discussions of qRT-PCR techniques and Astrid Woollard for the kind gift of reference gene primers.

Ancillary