Author for correspondence: Luis A. J. Mur Tel: +44 01970 622981 Email: email@example.com
•The staygreen (SGR) gene encodes a chloroplast-targeted protein which promotes chlorophyll degradation via disruption of light-harvesting complexes (LHCs).
•Over-expression of SGR in Arabidopsis (SGR-OX) in a Columbia-0 (Col-0) background caused spontaneous necrotic flecking. To relate this to the hypersensitive response (HR), Col-0, SGR-OX and RNAi SGR (SGRi) lines were challenged with Pseudomonas syringae pv tomato (Pst) encoding the avirulence gene avrRpm1. Increased and decreased SGR expression, respectively, accelerated and suppressed the kinetics of HR-cell death. In Col-0, SGR transcript increased at 6 h after inoculation (hai) when tissue electrolyte leakage indicated the initiation of cell death.
•Excitation of the chlorophyll catabolite pheophorbide (Pheide) leads to the formation of toxic singlet oxygen (1O2). Pheide was first detected at 6 hai with Pst avrRpm1 and was linked to 1O2 generation and correlated with reduced Pheide a oxygenase (PaO) protein concentrations. The maximum quantum efficiency of photosystem II (Fv/Fm), quantum yield of electron transfer at photosystem II (φPSII), and photochemical quenching (qP) decreased at 6 hai in Col-0 but not in SGRi. Disruption of photosynthetic electron flow will cause light-dependent H2O2 generation at 6 hai.
•We conclude that disruption of LHCs, possibly influenced by SGR, and absence of PaO produce phototoxic chlorophyll catabolites and oxidative stress leading to the HR.
Plant resistance to pathogenic challenge can be associated with the elicitation of localized cell death – known as a hypersensitive response (HR) – which is likely to contribute to neutralization of the infection (reviewed by Beers & McDowell, 2001; Lam et al., 2001; Mur et al., 2008). Many signals have been shown in combination to bring about the HR, including H+/K+ exchange, calcium fluxes from extracellular and intracellular stores (Lecourieux et al., 2002), nitric oxide (Delledonne et al., 2001) and ceramides (Townley et al., 2005). It remains the case, however, that reactive oxygen species (ROS) are the most intensively investigated initiatory HR signal (Lamb & Dixon, 1997; Bolwell et al., 2002). Light is now emerging as a critical factor in the HR (Zeier et al., 2004) and in this paper we identify the chloroplast-centred events that underlie the light-dependent development of oxidative stress during the HR.
Investigations of ROS have been concerned with largely apoplastic sources. Good evidence supports contributions from polyamine oxidases (Yoda et al., 2006), a pH-dependent extracellular peroxidase (Bolwell et al., 2002) or oxalate oxidase (Lane et al., 1993; Wei et al., 1998). Most work has centred on plant homologues of neutrophil NADPH oxidases encoded by the Atrboh gene family in Arabidopsis, which were shown to be major sources of apoplastic ROS during the HR (Torres et al., 1998, 2002). Intracellular sources of ROS during the HR have come to be seen as important, and the focus is increasingly on organelles as sources of oxidative stress. Ashtamker et al. (2007) demonstrated a calcium-mediated increase of ROS in the nuclei of cultured tobacco (Nicotiana tabacum) cells treated with a cell elicitor. The redox status of the mitochondrion is essential to cell viability (reviewed by Noctor et al., 2007) and mitochondrial ROS have been implicated in cell death initiated by the elicitors victorin and harpin (Yao et al., 2002; Vidal et al., 2007). Evidence for a chloroplastic source of ROS came from observations that some forms of the HR could be light-dependent (Asai et al., 2000; Zeier et al., 2004; Chandra-Shekara et al., 2006), as is the oxidative burst during a tobacco mosaic virus (TMV)-elicited HR in tobacco (Liu et al., 2007). Recently, expression of a cyanobacterial flavodoxin, which has an antioxidant function, in tobacco was found to suppress chloroplast ROS production and also an HR elicited by Xanthomonas campestris pv vesicatoria (Zurbriggen et al., 2009). One likely source for such ROS is a disruption of photosynthetic electron transport (PET). Alterations in thylakoid membrane properties suggesting diminished photosynthetic capacity were seen as little as 30 min following inoculation with harpin (Boccara et al., 2007). Examination of chlorophyll (Chl) fluorescence (Fv/Fm) indicated photosystem II (PSII) damage at c. 9 h following initiation of a bacterially induced HR in Arabidopsis and 4 h following the elicitation of an HR by TMV in tobacco (Seo et al., 2000; Almeras et al., 2003). Such disruptions in PET will lead to ROS photoproduction as a result of excess excitation energy (EEE), where photon intensity exceeds that required for CO2 fixation (Karpinski et al., 2003; Szabo et al., 2005; Asada, 2006). Cell death in the spontaneous death (SD) mutant, lesions simulating disease 1 (lsd1), has been linked to a failure in photorespiratory dissipation of EEE coupled to reduced antioxidant capacities (Mateo et al., 2004).
Analyses of other SD mutants suggest another mechanism of photooxidative stress, as several mutants have been linked to a compromised Chl catabolic pathway. The Arabidopsis accelerated cell death mutants pheophorbide a oxygenase 1 (pao1) and accelerated cell death 1 (acd1-1) lack functional pheophorbide a oxygenase (PaO; Tanaka et al., 2003; Pružinskáet al., 2005), the monooxygenase that cleaves the macrocycle of the Chl catabolite pheophorbide a (Pheide), resulting in the loss of green colour (Hörtensteiner et al., 1998). The SD phenotype in acd1 leaves is light-dependent (Greenberg & Ausubel, 1993), as is the similar phenotype in the maize (Zea mays) lethal leaf spot 1 (lls1) mutant (Gray et al., 2002). The Lls1 gene is the maize homologue of Acd1, and also encodes PaO. Similarly, Spassieva & Hille (2002) produced a light-dependent SD phenotype in tomato (Solanum lycopersicum) by viral-induced gene silencing of the tomato Lls1 homologue. The acd2 mutant of Arabidopsis is deficient in red Chl catabolite reductase (RCCR), the enzyme that converts the red product of PaO action to a fluorescent but colourless intermediate (Wüthrich et al., 2000), and production of SD lesions in acd2 plants is light-dependent (Mach et al., 2001). Upon light excitation, Chl can be converted to the triplet state (3P680*), which ultimately transfers its energy to the ground state of molecular oxygen to generate 1O2 (Apel & Hirt, 2004). Release of the pigments from the light-harvesting complexes removes Chl and Chl catabolites from quenching carotenoids and may initiate considerable oxidative damage (Triantaphylides & Havaux, 2009). The Arabidopsis protein fluorescent (FLU) and its barley (Hordeum vulgare) orthologue Tigrina have been shown to play vital roles in the regulation of tetrapyrrole biosynthesis (Meskauskiene et al., 2001; Lee et al., 2003; Triantaphylides & Havaux, 2009). Overaccumulation of protochlorophyllide in flu and tigrina leads to light-induced cell death directly associated with singlet oxygen generation (Triantaphylides & Havaux, 2009). Flu-initiated cell death has been linked to two chloroplast-located proteins, EXECUTER 1 and 2, which mediate retrograde chloroplastic signalling to the nucleus (Wagner et al., 2004; Lee et al., 2007). The nuclear-encoded cell death mechanisms initiated via a chloroplast-located 1O2–EXECUTER1/2 pathway remain to be defined, but may involve products of lipid peroxidation (Triantaphylides & Havaux, 2009).
The staygreen (SGR) mutation was originally identified in Festuca pratensis (Thomas & Stoddart, 1975) and was subsequently shown to affect the Festuca homologue of Mendel’s I locus (Armstead et al., 2007). Orthologous genes have been found in rice (Oryza sativa), pea (Pisum sativum) and Arabidopsis and have been designated STAY-GREEN (SGR; Jiang et al., 2007; Park et al., 2007; Sato et al., 2007). Mutations within SGR result in a ‘cosmetic’ staygreen phenotype in which photosynthetic competence progressively declines identically to that of wild type during leaf senescence, but without the decline in Chl content that normally parallels it. While the majority of leaf proteins, including Rubisco, decline at the normal rate during senescence in SGR plants, Chl-associated plastid proteins are preferentially retained in the mutant, indicating overall failure to dismantle Chl–protein complexes in plants lacking functional SGR. Analysis of SGR mutants in pea has demonstrated delayed reduction of maximal PSII fluorescence yield (Fv/Fm) and PSII degradation during senescence, supporting a role for SGR in the disassembly of PSII (Aubry et al., 2008). The SGR protein sequence is highly conserved across all higher plants investigated, but its function is as yet uncertain.
We here show how modification of SGR expression can influence an HR elicited by Pseudomonas syringae pv tomato (Pst) avrRpm1 on Arabidopsis. The modulation of the HR was at least partially associated with Chl catabolism and particularly the formation of Pheide and hence 1O2. The formation of Pheide was correlated with a reduction in PaO protein. Disruption of PET probably leads to the production of ROS during the HR which could trigger Chl catabolism. Therefore, this adds a photosensitive step in the cell death process.
Materials and Methods
Plant materials and chemicals
Arabidopsis plants of various genotypes were grown at 20 ± 2°C under an 8-h light period and used at c. 4 wk following germination. Unless otherwise stated, all mutants were obtained from the Nottingham Arabidopsis Stock Centre (NASC; Nottingham, UK; http://arabidopsis.info/). The acd1-20 mutant used in this study exhibited an essentially identical phenotype to that of acd1-1 (Greenberg & Ausubel, 1993); the latter mutant is not available from public stock centres. The derivation of Landsberg erecta (Ler) transgenic line PR1-GUS has been described previously (Clarke et al., 2004) while Ler transformed with the senescence responsive SAG12 promoter-GUS fusion was derived by Richard Amasino (University of Wisconsin, Madison, WI, USA). For the derivation of the staygreen over-expression line (SGR-OX), the production of SGR-OX, full-length AtSGR (Aubry et al., 2008) was cloned into pC-TAPa (Rubio et al., 2005) using Gateway technology (Invitrogen). The 9× myc, 6× His and 2× IgG-binding-domain tags were added at the C-terminus of SGR with the intention to use them for tandem affinity purification experiments. Such experiments will be described elsewhere. As a control, GFP was used, which was tagged with the same domains and was directed to the chloroplast via N-terminal fusion with the chloroplast transit peptide of the small subunit of Rubisco (pNTAP-RBCS-GFP; U. Eckhardt, pers. comm.). Columbia-0 (Col-0) was transformed with the floral dip method and homozygous plants of the T2 generation were selected.
Plants were illuminated with 55-W (Osram, Sylvania, Munich, Germany) high-frequency lighting tubes (4580 lumen output), supplemented with 2 × 30-W clear tube cooled lighting. Light fluence rates at the top of the plants varied according to the requirements of the experiments.
All chemicals were purchased from Sigma Pharmaceuticals Ltd (Poole, UK).
Pathogen inoculation procedures and estimations of cell death
Arabidopsis plants of various genotypes were inoculated with avirulent Pseudomonas syringae pv tomato (Pst) strain DC3000 avrRpm1 or nonpathogenic/non-HR-eliciting Pst strain hrpA as described previously (Mur et al., 2000) using inocula of 2 × 106 cells ml−1.
Cell death was estimated by electrolyte leakage in 1 cm diameter cores as described in Mur et al. (2000).
Visualization of H2O2 and singlet oxygen sensor green (SOSG)
After sampling, the excised leaves were immersed in an aqueous solution of 1 mg ml−1 3,3′-diaminobenzidine (DAB), pH 3.8, and incubated at room temperature for 4 h. The leaves were removed from the DAB solution and fixed and cleared in absolute ethanol. The samples were scanned using a flat-bed scanner and the intensity of staining was quantified using Image-Quant TL software (GE Healthcare Life Sciences, Little Chalfont, UK). SOSG dye (Invitrogen, Paisley, UK) was used at 10 μM (in deionized water) and infiltrated into areas previously inoculated with Pst avrRpm1, Pst hrpA or 10 mM MgCl2 at 6 h after inoculation (hai). Infiltrated areas were viewed following excitation with 488-nm light and emission with 520-nm light using an Olympus BH2 microscope (Southend-on-Sea, UK).
Analysis of Chl and its derivatives
Chorophylls and their immediate catabolites were extracted from frozen leaf material using 80% (v/v) acetone and separated by reversed-phase high-performance liquid chromatography (HPLC) as described previously (Roca et al., 2004). Pheide levels were expressed in terms of HLPC trace peak area normalized to leaf fresh weight.
Chl fluorescence parameters
Chl fluorescence from plant leaf areas infiltrated with bacteria was measured in air with a PAM 2000 portable fluori-meter (Waltz, Effeltrich, Germany) as described previously (Kingston-Smith et al., 1997).
Histochemical assays of GUS activity involved placing samples into 1 mM 5-bromo-4-chloro-3-indolyl glucuronide (first dissolved in dimethyl formamide at 1 mg ml−1) in 50 mM NaPO4, pH 7.0, and 0.1% Triton X-100. For infiltration, the samples were placed under vacuum twice for 2 min each time and then incubated at 37°C overnight. The Chl was removed by sequential extraction with 30%, 75% and 95% ethanol.
RNA was extracted using RNAEasy kits (Qiagen, Crawley, UK) following the manufacturer’s instructions. Northern blotting and hybridization were undertaken as described in Draper et al. (1988). A probe for PR1 (At2g19990) was obtained from The Arabidopsis Information Resource (TAIR) (http://www.arabidopsis.org). SAG12 (At5g45890) was the gift of Dr Vicky Buchanan Wollaston (Warwick University, Coventry, UK).
Real-time polymerase chain reactions (qPCRs)
cDNA was synthesized from purified RNA by reverse transcription using ImProm II reverse transcriptase (Promega) and oligo dT primers as per the manufacturer’s protocol. Similar reactions containing no reverse transcriptase were used to test for genomic contamination. qPCR analysis was carried out using an Applied Biosystem 7500 Real Time PCR System and SYBR green master mix in accordance with the manufacturer’s instructions. Initial reactions with actin primers (Act F1 (5′-TCTTGTTCCAGCCCTCGTTT) and Act R1 (5′-TCTCGTGGATTCCAGCAGCT) created using Primer Express software (Applied Biosystems)) were used to standardize template concentration. Relative transcript abundance was determined using the 2−ΔΔCT method (Livak & Schmittgen 2001) normalized to actin, following qPCR of the cDNA template using gene-specific primers; PaO F1 (5′-TTCAAGGTGGAGTCAAGTGGG), PaO R1 (5′-GTCATCATTCGCACCTTGGA), SGR F1 (5′-GGGAAGCTGTTTCGCCTGA) and SGR R1 (5′-GAGCCTCGGGAAGAGTCTCA) primers were created using Primer Express software. Each cDNA reaction was analysed in triplicate per run, and mean values were obtained from three separate runs.
Salicylic acid (SA) and jasmonic acid (JA) measurements
Western transfer of proteins to nitrocellulose membranes was undertaken as described in Bi et al. (1995). PaO antibodies (Aubry et al., 2008) were detected using assays for horseradish peroxidase on the secondary antibody with an ECL chemiluminescence kit (GE Healthcare UK Ltd, Little Chalfont, UK).
Repetition and statistical analysis
Each experiment was performed at least three times, generating similar data to those presented in this paper. The exception was the northern hybridization experiment, which was repeated only once, and gave similar results. All other data were tested for significance by ANOVA using Minitab v 13 (Minitab Ltd, Coventry, UK).
We previously described an Arabidopsis homologue of Mendel’s I gene (At4g22920) which, when suppressed by RNA interference in line SGRi, resulted in a staygreen phenotype (Armstead et al., 2007). To explore further the roles of SGR in plant physiology, over-expression lines (SGR-OX) in Arabidopsis were generated (Supporting Information Fig. S1a). These transgenic lines formed discrete necrotic-like flecks when grown under long-day conditions (Fig. 1a) and tended to be smaller than wild-type plants (Fig. S1b,c). Staining with trypan blue established that the flecks in SGR-OX lines represent areas of cell death (Fig. 1b). The H2O2 stain DAB revealed sites of ROS generation in apparently asymptomatic leaves (Fig. 1c), which were likely to be contributing to the formation of necrotic flecks. Lesions did not occur in control plants expressing chloroplast-directed GFP (data not shown), indicating that SGR expression was responsible for the phenotype.
Modulating the expression of SGR influences the HR
Given the apparent parallels between flecking and the formation of an HR, SGR-OX plants were challenged with avirulent Pst avrRpm1 bacteria. No apparent macroscopic change in the HR was observed (Fig. 2a), but electrolyte leakage indicated that cell death was accelerated compared with controls (Fig. 2b).
To assess how SGR could contribute to the Pst avrRpm1-elicited HR in wild-type plants, SGR transcript accumulation was determined. Compared with inoculations with the non-HR-eliciting strain Pst hrpA, qPCR indicated significant (P <0.001) increases in SGR transcript by 6 h after inoculation of Arabidopsis (Fig. 2c).
The HR was investigated in RNAi SGR (SGRi) and wild-type lines with Pst avrRpm1 in the light (160 μmol m−2 s−1; Fig. 3a). A delay in HR cell death (Fig. 3b) and electrolyte leakage (Fig. 3b) in SGRi plants compared with Col-0 was observed. To determine whether these effects were specific to a Pst avrRpm1-elicited HR, SGRi lines were inoculated with Pst avrPphB or Pst avrRps4. In both cases, electrolyte leakage suggested that the HR was delayed (Fig. S2). As the kinetics of cell death are significantly influenced by ROS (Levine et al., 1994), Col-0 and SGRi were stained with DAB at 6 h after inoculation with Pst avrRpm1 or a 10 mM MgCl2 infection control. In SGRi, DAB staining was reduced but not abolished compared with Col-0 (Fig. 3d). There was a significant but not dramatic increase in Pst avrRpm1 bacterial numbers in SGRi compared with Col-0. This indicates that the delayed formation of an HR does not greatly compromise plant resistance (Fig. 3e).
Chl catabolism occurs during the HR
One feature of SGRi lines is a greatly reduced accumulation of phytotoxic Chl catabolites (Armstead et al., 2007). We hypothesized that a build-up of catabolites could contribute to oxidative events. To test this hypothesis, the kinetics of HR cell death were investigated in acd1-20 and acd2-2 mutants, where breakdown of Chl catabolites would be perturbed. acd1-20 exhibited augmented electrolyte leakage at 6 hai, while acd2-2 did not significantly differ from wild-type plants in this respect (Fig. 4a). These data imply that Chl catabolites were rapidly generated during the HR. To verify this, Chl and its photolabile catabolites were extracted from dark-incubated Col-0, acd1-20 and acd2-2 following inoculations with Pst avrRpm1 and analysed by HPLC. The dark incubation was essential, otherwise the Chl catabolites would immediately photobleach in the light and would therefore be undetectable. On the basis of retention times and absorption spectra, the major compounds present were identified as Chla, Chlb and Pheide a (Fig. S3 shows a typical trace for Col-0, in which these three compounds are all represented). No accumulation of Pheide was detected in any genotype either before inoculation with Pst avrRpm1 or at any time-point in tissue inoculated with Pst hrpA (data not shown). Following inoculation with Pst avrRpm1, Pheide was first detected in acd1-20 at 6 hai (Fig. 4b) and in wild-type plants at 12 hai. Given the results obtained with acd1-20, it may be presumed that Pheide is produced earlier than 12 hai in wild-type plants, but is further metabolized by PaO. Interestingly, no Pheide was observed in acd2-2 before 24 hai. When Pheide accumulation was assessed in SGRi up to 3 d following inoculation of SGRi and Col-0 with Pst avrRpm1, none was detected (Table 1). Thus, the altered kinetics of the HR in SGRi, Col-0 and acd1-20 correlated with differing Pheide accumulation.
Table 1. Pheophorbide accumulation in Arabidopsis thaliana Columbia (Col-0) and RNAi staygreen (SGRi) following inoculation with the Hypersensitive Response eliciting Pseudomonas syringae pv tomato DC3000 avrRpm1
Pheide levels represent high-performance liquid chromatography (HPLC) peak areas normalized with respect to mg fresh weight. Each value represents the average ratio of three samples (± SE). nd, not detected.
To correlate these observations with 1O2 generation, in planta visualization was carried out using the sensitive and specific SOSG dye (Flors & Nonell, 2006). 1O2 generation was detected at 6 hai with Pst avrRpm1, that is, before plant cell collapse and when Pheide concentrations would be increasing during the HR. SOSG staining indicated that 1O2 concentrations in Col-0 were increased in acd1-20 and SGR-OX but reduced in SGRi (Fig. 4c). No SOSG staining was observed in uninoculated control plants of any genotype (data not shown).
Levels of PaO protein were determined in wild-type plants following whole-leaf inoculation with Pst avrRpm1, and also with Pst hrpA and 10 mM MgCl2 as controls (Fig. 4d). The control inoculations caused a transient increase in amounts of PaO between 3 and 12 hai, probably as a consequence of the wounding associated with the inoculation procedure (Yang et al., 2004). With Pst avrRpm1, this increase was not observed and, indeed, by 12 hai PaO protein was not detected, which agreed with the timescale of Pheide appearance in wild-type plants (Fig. 4a). Parallel assays for Rubisco large subunit (LSU) protein using monoclonal antibody demonstrated equally loading (data not shown) and also that PaO protein loss preceded the general proteolysis known to occur during the HR (Mur et al., 2008). Conversely, increased PaO transcript levels following inoculations of both Pst avrRpm1 and Pst hrpA after 9 hai (Fig. 4e) suggested that reduced PaO protein concentrations seen with the Pst avrRpm1 inoculations arose from a control mechanism operating at the post-transcriptional and/or post-translational level.
Cell death associated with the Pst avrRpm1-elicited HR are influenced by light
One prediction of accumulation of Chl catabolites during the HR is that they will contribute to a light-mediated effect. Arabidopsis thaliana Col-0 was inoculated with Pst avrRpm1 and incubated under different light fluence rates, leading to distinctive phenotypes (Fig. 5a). As light fluence rates were progressively reduced from c. 160 μmol m−2 s−1, the inoculated areas became correspondingly less necrotic, appearing increasingly green, while the surrounding areas became more chlorotic. Measurements of electrolyte leakage indicated that diminishing light fluence rates also reduced cell death (Fig. S4).
PR1 gene expression, which was modified in SGR modulated lines, has been reported to be light-dependent (Zeier et al., 2004). We have tested this using northern hybridization (Ougham et al., 2008), in conjunction with assays of PR1-GUS transgenic plants (Fig. 5bi). Both approaches indicated that PR1 expression at the periphery of the HR zone was influenced by light fluence rates. To assess whether the chlorosis observed around the HR at low fluence rates could represent a senescence-like symptom, we investigated expression of the senescence marker gene SAG12 (Pontier et al., 1999). Both SAG12 northern hybridization (Ougham et al., 2008) and SAG12-GUS promoter-fusion activity showed increasing up-regulation with lowering light fluence rates (Fig. 5bii), supporting the view that a form of senescence is initiated at low light fluence rates.
PR1 is a marker for SA, and a compromised HR in the dark has been associated with reduced accumulation of this stress hormone (Zeier et al., 2004). We also observed that SA concentrations were significantly (P <0.05) reduced in dark incubated HR (Fig. 5c). To investigate whether this reduced concentration of SA was the source of the altered HR in the dark, Arabidopsis constitutive pathogenesis related mutants cpr1-1, cpr5-2 and cpr6-1 (Fig. 5d), each of which exhibits elevated concentrations of SA, were inoculated with Pst avrRpm1. In these mutants the dark-incubation HR was not restored to that seen in the light. This indicated that the dark HR phenotype did not arise as a consequence of reduced concentrations of SA.
Photosynthetic light-harvesting efficiencies are perturbed during the HR leading to ROS production
Although we have focused on the accumulation of Chl catabolites and 1O2 generation, SGR is also likely to disrupt PET, leading to the light-dependent production of O2− and H2O2. To establish the role of PET, the intrinsic efficiency of electron transport through photosystem II (Fv/Fm) was examined in Col-0 following inoculation with Pst avrRpm1 and Pst hrpA (Fig. 6a). Fv/Fm was compromised by 5.5 hai in Pst avrRpm1- but not in Pst hrpA-inoculated leaves in the light. This decrease in Fv/Fm was coincident with increased electrolyte leakage and was associated with a decrease in the quantum yield of electron transfer at photosystem II (φPSII) and photochemical quenching (qP), indicating that infection resulted in a decrease in the number of available reaction centres, limiting flux through PET. As consumption of electrons in photochemistry declined, nonphotochemical quenching (qN) increased. Significantly, the impact of a Pst avrRpm1-induced HR on Fv/Fm, φPSII, qP and qN was reduced in SGRi (Fig. S5).
To correlate PET disruption in Col-0 with ROS generation, the extent of DAB staining was examined under different fluence rates at 6 h after inoculation with Pst avrRpm1, before there was any visible sign of leaf collapse. More intense staining was seen with higher fluence rates, while in the dark very little staining was observed (Fig. 6b). Expression of the H2O2-responsive marker gene glutathione S-transferase (GST-6), determined at 6 h after inoculation with Pst avrRpm1, also proved to be light-dependent (Fig. 6c).
The SGR protein plays a key role in the initiation of Chl degradation. Mutations in SGR genes have been associated with a cosmetic, Chl-retaining phenotype during senescence in a Lolium/Festuca introgression (Armstead et al., 2006), rice (Jiang et al., 2007; Park et al., 2007), pea (Sato et al., 2007), tomato and pepper (Capsicum annuum) (Barry et al., 2008). The SGR mode of action remains to be clearly resolved but the protein is known to be chloroplast targeted and to bind to the light-harvesting complex (LHC) in vivo (Park et al., 2007). In a pea SGR mutant line, senescence-associated proteolysis of the LHCs, reduction of Fv/Fm and the appearance of Chl catabolites were delayed (Aubry et al., 2008). The weight of evidence suggests that SGR is required for LHC disassembly with concomitant effects on photosynthetic efficiency and Chl catabolism.
There have been many reports demonstrating that light/shading influences either resistance or the progression of disease (reviewed by Roberts & Paul, 2006). Light can influence plant cell death, for example in Arabidopsis protoplast cultures treated with the toxin fuminosin B1 (Asai et al., 2000) or in whole plants inoculated with either the avirulent P. s. pv maculicola (Zeier et al., 2004) or turnip crinkle virus (TCV; Chandra-Shekara et al., 2006). Reduced resistance to pathogens in the dark has been noted under experimental conditions (Zeier et al., 2004) and, given the importance of the HR to resistance, may explain why certain pathogens infect at night. Indeed, light can suppress phytopathogenic fungal conidial germination or germ tube growth (Beyer et al., 2004). The present study shows that over-expression of SGR is sufficient to elicit cell death in the absence of infection (Fig. 1a), while suppression of SGR may delay the Pst avrRpm1-elicited HR (Fig. 3a). In the light of these observations, we suggest that SGR has a function in triggering certain forms of HR and that under these conditions Chl catabolites accumulate and play a role in the cell death process, thereby contributing to the often-reported light sensitivity of certain types of HR. It should be noted that, as in a recent study where suppression of chloroplast-generated ROS reduced an HR (Zurbriggen et al., 2009), our results suggest that the light-mediated step in the HR contributes little to plant resistance (Fig. 3e).
Correct metabolic processing of the Chl macrocycle is essential for plant homeostasis. Tetrapyrrole pigments are photosensitizers, transferring energy from a short-lived singlet excited state to molecular oxygen, thereby generating 1O2. Within the LHCs, 1O2 is quenched by carotenoids (Triantaphylides & Havaux, 2009). Unsurprisingly, several mutants exhibiting SD have been mapped to genes involved in Chl metabolism; for example, the flu mutation which leads to protochlorophyllide accumulation (Meskauskiene et al., 2001). Lesions in acd1 and acd2 correspond to enzymes involved in the catabolism of Pheide a and RCC, respectively (Fig. 7; Hörtensteiner, 2006). Our observations also imply a role for Chl catabolites in the Pst avrRpm1-elicited HR. Pheide a could be detected in acd1-20 as early as 6 hai (Fig. 4b), indicating an early, rapid initiation of Chl catabolism during the HR. Crucially, where Pheide concentrations were either increased or decreased in the acd1-20 mutant or SGRi line, respectively, there was a corresponding modification in the rate of HR-associated cell death. Further, examination of PaO/ACD1 protein concentrations during the HR showed that these decreased after 6 hai, following which increases in Pheide were observed. This suggests that the HR has features in common with that of the acd1-20 mutant, where the accumulation of Pheide in the absence of PaO contributes directly to spontaneous cell death. We observed that 1O2 generation could be modulated by manipulating SGR transcript levels (Fig. 4c), thereby linking SGR effects on the HR to 1O2 generation. Further, although defence-associated phenalenone phytoalexins can generate 1O2 (Flors & Nonell, 2006), the increases seen with acd1-20 would suggest that Chl catabolites, perhaps especially Pheide, are the source of 1O2. Based on this, it may be hypothesized that Chl catabolites could be triggering the HR; but we consider this to be unlikely, for the following reasons. Pheide accumulation occurred after 6 hai, some time after other HR initiatory events, for example nitric oxide (NO) generation and the initiation of an oxidative burst, are known to occur (Mur et al., 2008). It is more probable that Chl catabolite accumulation is associated with cellular collapse which is first indicated by electrolyte leakage, which also increases at c. 6 hai (Fig. 2b). Indirect evidence that supports this later role for Chl catabolites and 1O2 is our observation that there was no major disruption in the HR when the EXECUTER mutants exe1 and exe2 were inoculated (Fig. S6).
Interference with PET leading to ROS production has been suggested to initiate tetrapyrrole release and thence lead to 1O2 formation (Triantaphylides & Havaux, 2009; Zurbriggen et al., 2009). Thus, SGR-triggered Chl catabolism could be initiated by disruption of PET. The modulation of photosynthesis following pathogenic challenge has been noted by many workers (e.g. Chou et al., 2000; Berger et al., 2004). Bonfig et al. (2006) examined photosynthetic parameters following inoculation of Arabidopsis with Pst avrRpm1, with virulent Pst or with 10 mM MgCl2 and showed that Fv/Fm decreased 3 h after inoculation with Pst avrRpm1 but not the virulent Pst. In our experiment, reduced Fv/Fm was observed between 3 (Fig. S5) and 5 (Fig. 5a) hai, the latter being coincident with the first detection of Pheide in acd1-20 (Fig. 4b). Interestingly, the reduction of Fv/Fm was not observed in SGRi lines (Fig. S5) which also failed to exhibit Pheide accumulation (Table 1). The behaviour of SGRi plants is consistent with a triggering role for PET disruption. In Col-0 the decrease in Fv/Fm was coupled with declining PSII efficiency (φ PSII), and plastoquinone pools (qP) becoming progressively reduced. The resulting over-reduction of the electron transport components would contribute to death through electron donation to molecular oxygen to produce reactive oxygen and H2O2, as visualized by DAB staining (Fig. 6b).
The timing of light-associated ROS generation coincided with the persistent rise in ROS that forms part of the HR-associated oxidative burst (Lamb & Dixon, 1997; Mur et al., 2005), perhaps suggesting that this could be of chloroplastic origin. The source of this oxidative burst has been most often linked to an NAPDH oxidase (Torres et al., 1998, 2002) and Zeier et al. (2004) also suggested that ROS generation was light-independent. However, in Zeier et al. (2004), oxidative events were assessed at 24 hai rather than 6 hai as in this study, which is coincident with the oxidative burst (Mur et al., 2005). In this context, it is relevant that Zeier et al. (2004) demonstrated light-dependent expression of the ROS-responsive gst1 gene at 4 hai. However, PET perturbation is not a feature of all forms of cell death. Garmier et al. (2007) investigated cell death in Nicotiana sylvestris leaves infiltrated with harpin, which forms ion-conducting pores in lipid bilayers and is likely to be a virulence factor involved in mobilizing nutrients from the host (Lee et al., 2001). Unlike the Pst avrRpm1-elicited HR, harpin-induced cell death was only marginally affected by light intensities, and programmed cell death-linked nuclear fragmentation was observed in both the light and the dark. As in our study, analyses of Chl parameters indicated considerable PSII damage but DAB staining was only trivially affected in dark-adapted leaves. It may be relevant that dark-maintained harpin-infiltrated leaves were simply delayed in dehydration, but did not remain green as we observed with Pst avrRpm1 (Fig. 5a).
It has been suggested that light dependence during the HR is attributable to reduced SA concentrations (Asai et al., 2000; Zeier et al., 2004). For example, in the P. s. pv maculicola-elicited HR, SA concentrations and expression of SA-marker genes were increased in the light (Zeier et al., 2004); features shared with the Pst avrRpm1-elicited HR (Fig. 5). Phytochrome A and B also influence cell death, at least in part, through SA-mediated effects (Genoud et al., 2002). This could reflect the chloroplastic location of the SA biosynthetic gene isochorismate synthase (Wildermuth et al., 2001). However, as avrRpm1-elicited cell death was not restored in Arabidopsis mutants where SA concentrations are increased (Fig. 5d), our data suggest that light-dependent SA biosynthesis does not explain the light-modulated character of certain types of HR. Similarly, Chandra-Shekara et al. (2006) could only partially restore resistance to TCV in the dark when SA was applied. A lack of SA has been shown to alter but not abolish the HR even in the light, through a compromised potentiation mechanism (Mur et al., 1996, 2000). It may be that some workers are similarly observing a reduced potentiating effect when the HR is incubated in the dark. The early stages of jasmonate biosynthesis are also chloroplast-located (Weber, 2002) and Zeier et al. (2004) suggested that concentrations of jasmonate were increased in a dark-incubated HR. However, in our experiments, jasmonate concentrations within the HR were actually reduced in the dark (Fig. S7), which was in accord with the light-dependent expression of the jasmonate-marker gene thionin2.1 (thi2.1) reported by Zeier et al. (2004). However, as jasmonates are elevated in the Arabidopsis mutant cpr5 (Boch et al., 1998) and this did not display a normal HR in the dark (Fig. 5d), we suggest that the reduced concentrations of jasmonate are not a major cause of light effects on the HR.
Taking all of our data together, we suggest a sequence of chloroplast-centred events occurring in the elaboration of an HR (Fig. 7). During an initial phase (up to 6 hai), well-established triggering events take place: for example, Ca2+ influxes and NO, are likely to initiate SGR-mediated LHC disassembly resulting in PET disruption and light-dependent H2O2 generation. LHC disassembly also liberates Chl, leading to the HR-associated production of Pheide, mediated by the loss of PaO enzyme; Pheide then becomes a significant source of singlet oxygen. In the light, the olive-green Pheide pigmentation is lost from inoculated areas through photobleaching. This model is currently being tested in our laboratories.
We greatly appreciate the gifts that made this study possible: the strains of P. s. pv tomato DC3000 avrRpm1 (John Mansfield, Imperial College, London, UK), the Ler SAG12-GUS transgenic line (Richard Amasino, University of Wisconsin–Madison, Madison, WI, USA), pNTAP-RBCS-GFP (Ulrich Eckhardt, Humboldt University, Berlin, Germany), the cpr mutants (Xinnian Dong, Duke University, Durham, NC, USA) and an SAG12 gene probe (Vicky Buchanan-Wollaston, University of Warwick, Coventry, UK). This work was supported by the UK Biotechnology and Biological Sciences Research Council (BBSRC; Grant P10096). The pathogen work was carried out under UK license PHF 123A/3624. Work by H.O., A.K-S., J.G., E.T-T. and C.J. is supported by BBSRC Strategic Programme Grant funding. H.T. gratefully acknowledges support from the Leverhulme Trust. S.H. is supported by the Swiss National Science Foundation.