Salicylic acid dependent signaling promotes basal thermotolerance but is not essential for acquired thermotolerance in Arabidopsis thaliana


For correspondence (fax +44 1970 622350; e-mail


Salicylic acid (SA) is reported to protect plants from heat shock (HS), but insufficient is known about its role in thermotolerance or how this relates to SA signaling in pathogen resistance. We tested thermotolerance and expression of pathogenesis-related (PR) and HS proteins (HSPs) in Arabidopsis thaliana genotypes with modified SA signaling: plants with the SA hydroxylase NahG transgene, the nonexpresser of PR proteins (npr1) mutant, and the constitutive expressers of PR proteins (cpr1 and cpr5) mutants. At all growth stages from seeds to 3-week-old plants, we found evidence for SA-dependent signaling in basal thermotolerance (i.e. tolerance of HS without prior heat acclimation). Endogenous SA correlated with basal thermotolerance, with the SA-deficient NahG and SA-accumulating cpr5 genotypes having lowest and highest thermotolerance, respectively. SA promoted thermotolerance during the HS itself and subsequent recovery. Recovery from HS apparently involved an NPR1-dependent pathway but thermotolerance during HS did not. SA reduced electrolyte leakage, indicating that it induced membrane thermoprotection. PR-1 and Hsp17.6 were induced by SA or HS, indicating common factors in pathogen and HS responses. SA-induced Hsp17.6 expression had a different dose–response to PR-1 expression. HS-induced Hsp17.6 protein appeared more slowly in NahG. However, SA only partially induced HSPs. Hsp17.6 induction by HS was more substantial than by SA, and we found no SA effect on Hsp101 expression. All genotypes, including NahG and npr1, were capable of expression of HSPs and acquisition of HS tolerance by prior heat acclimation. Although SA promotes basal thermotolerance, it is not essential for acquired thermotolerance.


Capacity to survive heat shock (HS) varies with plant species and genotype (Martineau et al., 1979; Srinivasan et al., 1996; Tal and Shannon, 1983), as well as developmental stage (Hong and Vierling, 2001; Nieto-Sotelo et al., 2002). HS tolerance in the absence of pre-adaptation has been termed basal or basic thermotolerance (Hong and Vierling, 2001; Lee et al., 1995; Nieto-Sotelo et al., 2002). In addition, plants subjected to a milder heat stress can transiently acquire tolerance to previously lethal high temperatures: this phenomenon is known as acquired thermotolerance or heat acclimation (Dat et al., 1998a; Hong and Vierling, 2000, 2001; Hong et al., 2003; Howarth et al., 1997; Lee et al., 1995; Nieto-Sotelo et al., 2002), and is probably an adaptation to the gradual increases in temperature in the natural environment.

Relatively little is known about the signaling pathways that control basal and acquired thermotolerance, although hormonal, calcium, and redox signals have been implicated (Foyer et al., 1997; Larkindale and Knight, 2002; Pastori and Foyer, 2002). In recent years, salicylic acid (SA) has been found to protect potato, mustard, tobacco, tomato, bean, and Arabidopsis from heat stress (Dat et al., 1998a, 2000; Larkindale and Knight, 2002; Lopez-Delgado et al., 1998; Senaratna et al., 2000). However, major issues concerning the role of SA in thermotolerance remain unresolved, and this paper addresses the following details. First, although transient accumulation of endogenous SA was found in mustard seedlings subjected to heat acclimation treatment (Dat et al., 1998b), it has not yet been established whether SA might be an essential signal for acquired thermotolerance, or rather one component of a multifactorial regulation of plant thermotolerance. We therefore compared basal and acquired thermotolerance in plants with altered SA signaling, including plants constitutively expressing the Pseudomonas putida SA hydroxylase transgene NahG, which are unable to accumulate SA (Delaney et al., 1994).

Second, further insights into the thermoprotective cellular effects of SA are needed. SA has been found to induce changes in the antioxidant system (Dat et al., 1998a,b, 2000; Lopez-Delgado et al., 1998), and to protect against heat-induced oxidative damage (Larkindale and Knight, 2002). However, protection against oxidative damage was incompletely correlated with survival in Arabidopsis, so thermotolerance probably also involves protection against additional kinds of heat damage (Larkindale and Knight, 2002).

Much evidence supports the involvement of plant HS proteins (HSPs) in thermotolerance (Gurley, 2000; Hong and Vierling, 2000, 2001; Malik et al., 1999; Prändl et al., 1998; Queitsch et al., 2000), and we investigated whether SA affects expression of Arabidopsis HSPs from two families. First, Hsp17.6 belongs to the class I cytosolic family of plant small HSPs (sHSPs; Wehmeyer et al., 1996). Hsp17.6 assembles into 200–300 kDa oligomeric complexes and has protein-refolding activity (Sun et al., 2001). A Synechocystis sHSP has also been shown to act as a stabilizing factor for heat-stressed membranes (Török et al., 2001). Malik et al. (1999) obtained evidence for the physiological importance of carrot Hsp17.7, as modified expression of its gene could increase or decrease thermotolerance. Second, Hsp101 belongs to the Hsp100/ClpB family of ATP-dependent chaperones thought to disaggregate heat-induced protein complexes (Hong and Vierling, 2001; Mogk et al., 1999). Hsp101 has been shown to be essential in thermotolerance by mutation and antisense RNA inhibition (Hong and Vierling, 2000, 2001; Queitsch et al., 2000).

The physical state of membranes is highly sensitive to heat stress, and has attracted interest as a possible cellular temperature sensor (Carratu et al., 1996). Electrolyte leakage has been an experimental screen for heat-tolerant crop plants for a number of years (Howarth et al., 1997; Martineau et al., 1979; Srinivasan et al., 1996; Tal and Shannon, 1983). New evidence for the importance of membrane properties in thermotolerance has been obtained by Hong et al. (2003), whose hot2 Arabidopsis mutant fails to acquire thermotolerance and shows increased ion leakage, despite apparently normal HSP levels. In the present study, we found that protection of membrane integrity is an important effect of SA in heat stress.

Third, the signal transduction pathway(s) of the thermotolerance effects of SA remain unknown, as does the relationship with signaling in the better known effects of SA in systemic acquired resistance (SAR) and the hypersensitive response (Delaney et al., 1994; Mur et al., 1997). There are common pathways and components in different biotic and abiotic stress responses (Pastori and Foyer, 2002), so interactions between SA signaling in thermotolerance and SAR could occur. To investigate this, we utilized Arabidopsis mutants with defective or constitutive SAR, and modified SA signaling, which have been discovered in disease resistance research but not previously tested for thermotolerance. We used mutants defined in relation to pathogenesis-related (PR) proteins, which are SA-inducible markers of SAR. The nonexpresser of PR proteins (npr1) mutant has enhanced susceptibility to pathogens and lacks SA-responsive expression of PR genes, despite wild-type SA levels on infection (Cao et al., 1994). The recessive mutation suggests a positive regulator of the SA signal pathway. NPR1 is a nuclear-localized protein, which may interact with basic Leu zipper transcription factors that bind the PR-1 promoter (Després et al., 2000; Fan and Dong, 2002; Kinkema et al., 2000; Zhou et al., 2000). We also tested recessive mutants that are constitutive expressers of PR proteins (cpr). The cpr1 and cpr5 mutants have elevated levels of SA and constitutive expression of SAR; both gene products seem to act upstream of SA, in contrast to NPR1 (Bowling et al., 1994, 1997; Clarke et al., 2000). The cpr1 gene maps to a Resistance gene cluster on chromosome 4 (Stokes and Richards, 2002). CPR5 encodes a novel protein with a potential nuclear localization motif in its amino-terminal region, and five putative transmembrane domains in its carboxyl-terminal region (Yoshida et al., 2002).

Fourth, we sought more evidence on the stage(s) of heat stress in which SA acts. In addition to the stress during the period of high temperature, thermotolerance studies have recognized events during a distinct recovery period when plants return to non-stressful temperatures. Recovery after heat stress is characterized by restoration of cellular protein activity and/or synthesis (Hong et al., 2003), and transient elevations of cytosolic calcium ions (Larkindale and Knight, 2002).

Fifth, we tested the effects of SA at several plant growth stages, as thermotolerance can show developmental variation (Nieto-Sotelo et al., 2002). Thus, four hot mutant loci were identified by defective hypocotyl thermotolerance, but only the hot1 mutant has defective seed thermotolerance, and only hot1 and hot2 show defective thermotolerance as 10-day-old seedlings (Hong et al., 2003).

This paper thus provides new evidence to clarify the role of SA in thermotolerance and in the interaction between the thermotolerance and SAR signal pathways.


SA promotes basal thermotolerance in 3-week-old plants

To establish that exogenous SA increases thermotolerance, 3-week-old wild-type (Columbia (Col-0)) plants were subjected to a 38°C heat treatment ± SA. This was achieved by pipetting SA solution onto the medium 24 h prior to heat treatment. Following the heat treatment, electrolyte leakage was measured. Col-0 treated with either 500 or 1000 µm SA showed reduced electrolyte leakage compared to the control plants (Figure 1a). Electrolyte leakage was also measured during heat stress following an SA pre-treatment to the medium. A reduced level of electrolyte leakage was observed when Col-0 plants were pre-treated with SA (Figure 1b). To confirm that the application of SA by pipetting the solution onto the medium induced the SA-signaling pathway in the aerial parts of the plants, expression of PR-1 was investigated. A correlation between the level of PR-1 transcription and the concentration of SA applied was detected in the aerial part of the plant (Figure 1c).

Figure 1.

Salicylic acid treatment protects 3-week-old Col-0 from heat.

(a) During post-HS recovery. H2O (0 µm SA; ▪), 500 µm SA (●), or 1000 µm SA (▴) was applied to the MS medium 24 h prior to HS (38°C, 16 h). Following HS, aerial parts were placed in H2O at 22°C, and the bathing solution conductivity was monitored. Data are means from three experiments each with five plants per treatment. Bars are SE (n = 15).

(b) During HS treatment. H2O (0 µm SA; ▪), 500 µm SA (●), or 1000 µm SA (▴) was applied to the MS medium 24 h prior to HS. The aerial parts were incubated at 38°C in H2O, and the bathing solution conductivity monitored. Data are means from two experiments each with five plants per treatment. Bars are SE (n = 10).

(c) Induction of PR-1 in Col-0 aerial parts by SA application to the MS medium. RNA was extracted 24 h after H2O (0 µm SA) or the indicated SA concentration was applied. Northern blots of separated RNA (10 µg) were hybridized with 32P-labeled PR-1 cDNA. The ethidium bromide-stained gel is below.

The preceding results show that exogenous SA protects Col-0 plants from heat stress, but do not establish an in vivo role for SA. NahG plants were therefore used to test the effects of endogenous SA during each phase of basal thermotolerance. Necrotic lesions were observed on NahG plants following a 15-h heat treatment at 38°C (Figure 2a). Image analysis detected 13-fold more heat-induced necrosis on NahG leaves compared to Col-0 (Figure 2b).

Figure 2.

Reduced basal thermotolerance in 3-week-old NahG plants.

(a) Susceptibility of NahG to HS-induced necrotic lesions. Col-0 (left) and NahG (right) plants on MS medium were exposed to HS (38°C, 15 h), returned to 22°C, and photographed 3 days later. Bar = 1 cm.

(b) Quantification of HS-induced necrotic lesions on Col-0 and NahG plants by image analysis. Necrosis was measured as percentage of total leaf area of each plant. Values are means of 10 plants, 3 days after HS as above (gray bars) or kept at 22°C (black bars). Bars are SE (n = 10).

(c) Endogenous glucosylated SA in aerial parts of Col-0 and NahG plants exposed to HS (38°C) for the indicated time. Data are means of at least five experiments. Bars are SE (n ≥ 5).

(d) Endogenous free SA in aerial parts of Col-0 and NahG plants exposed to HS (38°C) for the indicated time. Data are means of at least five experiments. Bars are SE (n ≥ 5).

(e) Electrolyte leakage during post-HS recovery. Following HS (38°C, 16 h), aerial parts of Col-0 (□) and NahG (●) plants were placed in H2O at 22°C, and the bathing solution conductivity was monitored. Data are means from seven experiments each with five plants per genotype. Bars are SE (n = 35).

(f) Electrolyte leakage during HS. Aerial parts of Col-0 (□) and NahG (●) plants were incubated in H2O at 38°C, and the bathing solution conductivity was monitored. Data are means from seven experiments each with five plants per genotype. Bars are SE (n = 35).

(g) Lack of effects of catechol on Col-0 thermotolerance. H2O (0 µm; ▪), 25 µm (◆), 50 µm (▴), 100 µm (□), or 1000 µm catechol (●) was applied to the MS medium 24 h prior to HS (38°C, 16 h). Following HS, aerial parts were placed in H2O at 22°C and the bathing solution conductivity monitored. Data are means from two experiments each with five plants per treatment. Bars are SE (n = 10).

Analysis of endogenous SA in NahG and Col-0 plants confirmed the effectiveness of the bacterial hydroxylase transgene, as levels were greatly reduced in NahG plants (Figure 2c,d). During heat treatment, glucosylated SA levels in Col-0 showed a moderate (more than twofold; P < 0.01) increase at 1.5 h (Figure 2c). Changes in the level of free SA were not significant during heat treatment (Figure 2d).

Electrolyte leakage was measured to investigate ability to recover from and to cope during heat stress. Figure 2(e) shows that the recovery from heat treatment was greatly affected in NahG, which showed more than twofold greater electrolyte leakage than Col-0 after 7-h recovery. NahG plants were also less tolerant during the heat stress, where a faster rate of electrolyte leakage than Col-0 was found (Figure 2f). The NahG salicylate hydroxylase converts SA to catechol. To investigate the possibility that the NahG phenotype is in part because of this degradation product, 3-week-old Col-0 plants were subjected to a heat treatment ± catechol. Figure 2(g) shows that catechol did not increase the level of electrolyte leakage following heat treatment. In addition, catechol was not detected in the same samples used for SA analysis (not shown).

Heat induces PR-1 expression

We employed plants transformed with GUS driven by the PR-1 promoter to establish if heat stress induces PR-1 transcription. Heat-induced GUS expression was observed throughout the plant after a 5-h heat treatment at 38°C (Figure 3a). The localized intense staining at some leaf margins may represent incipient necrosis. Northern analysis showed that during heat stress, PR-1 was expressed within 30 min, and then decreased to near background levels by 7 h (Figure 3b).

Figure 3.

Heat induces PR-1 expression.

(a) PR-1–GUS induction during HS. Three-week-old PR-1–GUS plants were incubated at 38°C for 0, 3, 5, or 16 h, and then kept at 22°C for 24 h before staining.

(b) Induction of PR-1 transcripts in Col-0 during HS. RNA was extracted from 3-week-old Col-0 aerial parts after the indicated period at 38°C. Northern blots of separated RNA (10 µg) were hybridized with 32P-labeled PR-1 cDNA. The ethidium bromide-stained gel is below.

Basal thermotolerance is altered in SA-signaling mutants

A range of plant growth stages was used to test thermotolerance in genotypes with modified SA signaling. Seed germination was tested after an HS (2 h, 47°C) at intervals after plating on MS medium. All genotypes showed about 95% germination when heat shocked 0 or 24 h post-plating, although this was delayed by 5 days compared to unstressed seeds (data not shown). However, when seeds were heat shocked 30 h post-plating, cpr1 and cpr5 were least sensitive and NahG was most sensitive (Figure 4a): after 7 days, 65 and 57% of cpr1 and cpr5 seeds, respectively, had germinated to principal growth stage 1.0, compared to 14% of Col-0 and 0% of NahG seeds. HS did not significantly affect npr1 seeds compared to Col-0 (Figure 4a). Ten days post-plating, 68 and 61% of cpr1 and cpr5 seeds had germinated to principal growth stage 1.02, compared to 27% of Col-0, 29% of npr1, and only 3% of NahG seeds (data not shown). HS 48 h post-plating was lethal to all genotypes (data not shown).

Figure 4.

Basal thermotolerance is affected in seeds and 5-day-old seedlings of genotypes with altered SA signaling.

(a) Seeds were subjected to HS (47°C, 2 h) at 30 h post-plating, and at 7 days post-plating, were scored as ungerminated (no germ), 0.5, 0.7, or 1.0 principal growth stages (Boyes et al., 2001). Data are means from three experiments each with at least 20 seeds per genotype, and are expressed as percentage of seeds at each stage. Bars are SE (n = 3).

(b) Five-day-old seedlings were exposed to 47°C for the specified period. Percentage survival was recorded 2, 5, and 7 days post-HS. Data are means from three experiments each with at least 20 seedlings per genotype. Bars are SE (n = 3).

The basal thermotolerance of 5-day-old seedlings to a 47°C HS was also examined (Figure 4b). Two days post-HS, all genotypes survived HS of up to 60 min. When the HS was extended to 90–120 min, fewer seedlings survived. In agreement with the germination data, NahG was most sensitive with 0% survival 2 days after a 120-min HS, while cpr5 was least sensitive with 55% survival (Figure 4b). HS for 90–120 min was ultimately lethal and none of the seedlings survived 5 days post-heat treatment. In contrast, after a 45-min HS, 47% of Col-0 survived after 5 days, compared to 63% of cpr5 and only 3% of NahG (Figure 4b). The npr1 genotype was intermediate between Col-0 and NahG. Most seedlings that had survived 5 days after a 45-min treatment continued to develop.

To test further the role of SA-signaling pathways in basal thermotolerance, electrolyte leakage was measured in 3-week-old npr1, cpr1, and cpr5 mutants in the recovery period immediately after a 16-h heat treatment at 38°C (Figure 5a). Greater leakage was observed for npr1 compared to Col-0 (Figure 5a), although to an extent intermediate to NahG leakage (cf. Figure 2e). In contrast, cpr5 electrolyte leakage was substantially less than that of Col-0. No differences were detected between cpr1 and Col-0. The electrolyte leakage during the recovery phase in Figure 5(a) was measured in the light. No significant difference was found between electrolyte leakage measurements in the dark or light for any of the genotypes (data not shown). Electrolyte leakage was also measured during the period at 38°C, in which a lower rate of electrolyte leakage was again observed for cpr5 compared to Col-0 (Figure 5b). In this period, however, leakage for npr1 was similar to Col-0. Electrolyte leakage for cpr1 during heat stress was again similar to that for Col-0.

Figure 5.

Basal thermotolerance is affected in 3-week-old SA-signaling mutants.

(a) During post-HS recovery. Following HS (38°C, 16 h), aerial parts of Col-0 (□), npr1 (●), cpr1 (▴), and cpr5 (◆) plants were placed in H2O at 22°C, and the bathing solution conductivity was monitored. Data are means from seven experiments each with five plants per genotype. Bars are SE (n = 35).

(b) During HS treatment. Aerial parts of Col-0 (□), npr1 (●), cpr1 (▴), and cpr5 (◆) plants were incubated in H2O at 38°C, and the bathing solution conductivity was monitored. Data are means from seven experiments each with five plants per genotype. Bars are SE (n = 35).

(c) Endogenous glucosylated SA in aerial parts of Col-0 and mutant plants exposed to HS (38°C) for the indicated time. Data are means of at least four experiments. Bars are SE (n ≥ 4).

(d) Endogenous free SA in aerial parts of Col-0 and mutant plants exposed to HS (38°C) for the indicated time. Data are means of at least four experiments. Bars are SE (n ≥ 4).

Measurements of SA in 3-week-old mutants confirmed that by far the highest levels occurred in the most thermotolerant genotype cpr5 (Figure 5c,d). Basal levels (i.e. in unstressed plants) of glucosylated and free SA in cpr5 were about 14-fold those in Col-0, although these showed some decline during heat treatment. SA levels in cpr1 and npr1 were much lower than that in cpr5, although greater than the SA level in Col-0, and did not show clear trends in heat treatment. Basal glucosylated SA levels in cpr1 were 2.7-fold (P < 0.01) those in Col-0 (Figure 5c), and free SA was similarly higher (Figure 5d). Basal levels of SA in npr1 were nearly twice those in Col-0 (Figure 5c,d).

SA induces Hsp17.6 expression

We explored how HSPs were expressed in SA-induced thermotolerance, and in genotypes with modified SA signaling, and compared with the widely studied expression of PR-1. Application of 1 mm SA did not appear to induce Hsp101 expression in Col-0 (not shown). In contrast, Figure 6(a) shows that incubation of 3-week-old Col-0 in SA at 22°C induced expression of Hsp17.6 with an optimal concentration of 100 µm. This expression pattern differed from PR-1 where a positive correlation with SA up to 500 µm was observed. In cpr1 and cpr5, a lower optimum of 10 µm SA was observed for Hsp17.6 expression (Figure 6a). PR-1 expression in cpr5 was constitutive and abundant, while cpr1 appeared to be intermediate with Col-0 (Figure 6a). Further support for SA-induced sHSP transcription was provided by plants in which the soybean Hsp17.3B promoter (Prändl et al., 1995) was linked to GUS. We not only confirmed the expected HS-induced GUS expression in these plants (not shown), but also found SA-induced GUS expression in these plants at an optimal 100 µm, as in the Northern blots (Figure 6a,b). SA-induced GUS activity in plants with a tobacco PR-1a promoter (Uknes et al., 1993) linked to GUS was also in agreement with the Northern data, as a positive correlation between SA concentration and both PR-1 expression and GUS activity was observed (Figure 6a,b). Expression of Hsp17.6 in response to the optimal concentration of SA was, however, less than that induced in HS (Figure 6c).

Figure 6.

Hsp17.6 and PR-1 are induced by SA and heat in 3-week-old plants.

(a) SA-induced Hsp17.6 and PR-1 transcripts in Col-0, cpr1, and cpr5. RNA was extracted from aerial parts after 24 h in SA solutions at the indicated concentrations at 22°C. Northern blots of separated RNA (10 µ g) were hybridized with 32P-labeled Hsp17.6 or PR-1 cDNA. The ethidium bromide-stained gel is below.

(b) SA induction of GUS under control of the Hsp17.3B or PR-1a promoters. Plants were incubated at 22°C in SA solutions at the indicated concentrations for 24 h before GUS activity staining.

(c) Hsp17.6 transcripts in SA-treated (100 µm, 24 h), control (22°C without SA), or HS-treated (38°C, 30 min) Col-0 plants. Northern blots of separated RNA (10 µg) from aerial parts were hybridized with 32P-labeled Hsp17.6 cDNA. The ethidium bromide-stained gel is below.

(d) Hsp17.6 transcripts in untreated genotypes at 22°C, compared with SA-treated Col-0 (10 µm, 24 h).

(e) Induction of Hsp17.6 in Col-0, NahG, and npr1 plants subjected to 38°C for the indicated periods.

(f) Induction of HSP17.6 and HSP101 proteins in Col-0, NahG, and npr1 plants subjected to 38°C for the indicated periods. Western blots of electrophoretically separated proteins from aerial parts were probed with antibodies against HSP17.6 or HSP101.

(g) Induction of HSP17.6 and HSP101 proteins in Col-0, cpr1, and cpr5 plants subjected to 38°C for the indicated periods.

The relationship between endogenous SA and Hsp17.6 was investigated in the npr1, cpr1, and cpr5 mutants, all of which had basal SA levels higher than in Col-0 (see above). A low level of constitutive expression of Hsp17.6 was detected in cpr5 plants (Figure 6d). To ascertain how the absence of SA or disruption of SA signal transduction pathways affects heat-induced transcription of Hsp17.6, Northern analysis of Hsp17.6 transcripts in heat-treated NahG and npr1 was compared with Col-0. Figure 6(e) shows that induction of Hsp17.6 in response to heat was similar in Col-0, NahG, and npr1. Levels of HSP17.6 protein also appeared similar in Col-0, NahG, and npr1 by 5 h at 38°C, but induction was delayed in NahG (Figure 6f). Induction of HSP17.6 in cpr1 and cpr5 appeared similar to Col-0 (Figure 6g). Expression of HSP101 in all of the mutants was similar to that in Col-0 (Figure 6f,g).

SA signaling is not essential for acquired thermotolerance

To investigate the role of SA in acquired thermotolerance, we applied the hypocotyl elongation assay described by Queitsch et al. (2000) to Col-0, NahG, and npr1. When non-acclimated 2.5-day-old seedlings were heat shocked for 2 h at 45°C, none of the genotypes showed subsequent hypocotyl elongation. In contrast, when seedlings received a mild pre-treatment for 90 min at 38°C before the 45°C HS, all genotypes displayed subsequent hypocotyl elongation, confirming that both NahG and npr1 can be acclimated (Figure 7a).

Figure 7.

Salicylic acid is not essential for acquired thermotolerance.

(a) Acquired thermotolerance in Col-0, NahG, and npr1 hypocotyls. Seedlings (2.5 days old) were maintained at 22°C, or exposed to HS (45°C, 2 h) with or without a prior acclimation (38°C, 90 min). The photo 2.5 days after HS is representative of three experiments.

(b) Acquired thermotolerance in 3-week-old Col-0 (□), NahG (●), and npr1 (◆) plants, subjected to HS (45°C, 2 h) with (dashed lines) or without (solid lines) a prior acclimation (38°C, 90 min). After a 24-h recovery period at 22°C, aerial parts were placed in H2O, and the conductivity was monitored. Data are means from two experiments each with five plants per treatment. Bars are SE (n = 10).

Similar conclusions were drawn when 3-week-old plants were subjected to a 2-h HS at 45°C, with (acclimated) or without (non-acclimated) a 90-min pre-treatment at 38°C, and returned to 22°C for 24 h before electrolyte leakage was measured. The level of electrolyte leakage from acclimated Col-0 plants was more than threefold less than that from the non-acclimated Col-0 plants (Figure 7b). Moreover, the level of electrolyte leakage from acclimated NahG and npr1 plants was similar to the level observed for acclimated Col-0 plants (Figure 7b). Both cpr1 and cpr5 mutant plants were able to acquire thermotolerance in a manner similar to that observed for Col-0 (data not shown).


This study substantiated the thermoprotective effects of SA in Arabidopsis using multiple criteria. Extending demonstrations of enhanced survival and protection against oxidative stress in various species (Dat et al., 1998a, 2000; Larkindale and Knight, 2002; Lopez-Delgado et al., 1998; Senaratna et al., 2000), we showed that membrane properties are stabilized in heat-stressed plants by SA, whether applied externally or as an endogenous metabolite that is depleted in plants expressing the NahG transgene. Hong et al. (2003) have shown genetically that thermotolerance mechanisms can differ during seedling growth, but we found endogenous SA-promoted tolerance of HS at all growth stages tested. Thus, the NahG genotype showed greater propensity to leaf necrosis and electrolyte leakage in 3-week-old plants, and poorer survival in 5-day-old seedlings and imbibed seeds. Additionally, 10-day-old NahG seedlings show poorer survival and greater oxidative damage after HS (Larkindale and Knight, 2002).

Electrolyte leakage, as a rapidly measured parameter, enabled us to demonstrate that SA protected membranes in both the HS and subsequent recovery periods. A 24-h pre-treatment with SA substantially reduced leakage during both HS and recovery, while NahG plants showed greater leakage in both periods. While the increased leakage in NahG plants during HS may seem less dramatic (Figure 2f), these plants would have had no prior period for induction of protective mechanisms like those recovering from HS or pre-treated with SA.

We could find no evidence that these effects of heat on NahG Arabidopsis were because of catechol production by the transgene product, as recently suggested for non-host resistance to P. syringae (Van Wees and Glazebrook, 2003). We found no effects of exogenous catechol on thermotolerance, unlike their non-host resistance system, and we detected no accumulation of catechol in NahG plants. Another difference from their system is that our findings with SAR mutants were also consistent with the involvement of SA.

While NahG was the least thermotolerant genotype and 3-week-old NahG plants had nearly 90% less total (free + glucosylated) SA than Col-0, cpr5 as the most thermotolerant genotype had about 14-fold more SA than Col-0. It is interesting that the electrolyte leakage for Col-0 treated with 1 mm SA (Figure 1b) was similar to that for untreated cpr5 (Figure 5b). Thermotolerance intermediate between these extremes was displayed by cpr1 and npr1 mutants. Three-week-old cpr1 mutants accumulated total SA to 2.7-fold the Col-0 levels, which was apparently insufficient to enhance thermotolerance in these plants, although cpr1 seeds were highly thermotolerant. In npr1 mutants at 3 weeks, total SA level was nearly twice that of the Col-0 levels, but the ability to recover from HS was impaired, and 5-day-old npr1 seedlings were also less thermotolerant: this would be explained if the npr1 mutation compromises a thermotolerance response downstream of SA accumulation, as in SAR (Cao et al., 1994). The greater amounts of SA in npr1 mutants have been noted by others, and it may be a function of NPR1 to downregulate SA accumulation (Clarke et al., 2000).

We thus obtained a range of evidence that SA-dependent signaling promotes tolerance of HS in plants not given a prior heat acclimation treatment, i.e. basal thermotolerance. This indicates that SA content and the state of SA-inducible thermotolerance mechanisms could be important determinants of basal thermotolerance in plants. In a similar way, we found clearly that plants are also able to acquire thermotolerance by mechanisms other than SA signaling. All tested genotypes, including NahG and npr1 with deficient SA content or signaling, when given a mild heat treatment were subsequently able to tolerate a previously lethal HS, whether tested as 3-week-old plants or as 2.5-day-old hypocotyls. SA content remained low in NahG during heat treatment, so its ability to acclimate was not because of heat-induced loss of activity of the transgene product.

It may be that SA is dispensable because heat acclimation is always initiated by some other key factor(s), or because SA is only one of multiple alternative acclimation signals. If SA were a potential mediator of a heat-induced acclimation response, SA levels should be heat inducible. As in the study by Dat et al. (1998b) on mustard, we found a moderate and transient but statistically significant increase in glucosylated SA during heat treatment in Col-0 plants. Given the significant basal levels of SA in Arabidopsis and mustard, the extent to which such changes in SA metabolism provide additional thermotolerance is uncertain, but they suggest that despite the metabolic stress of heat treatment, the plant actively maintains biosynthesis of this hormone. In plants, SA is subject to glucosylation (Dat et al., 1998b, 2000; Dean et al., 2003), which might be involved in transport (Seo et al., 1995) or vacuolar localization (Dean et al., 2003). Heat-induced SA increases were not apparent in the other Arabidopsis genotypes, although this might be because of their altered SA metabolism or signaling.

Our mutant studies indicate that SA-signaling pathways involved in SAR overlap with those promoting basal thermotolerance. In SAR, a key regulator of SA-inducible gene expression is the NPR1 protein, which interacts with members of the TGA subclass of basic Leu zipper transcription factors (Després et al., 2000; Zhou et al., 2000). SA promotes formation of the TGA–NPR1 complex, and both SA and NPR1 are required for the DNA binding of the transcriptional factor (Fan and Dong, 2002). There are seven known TGA factors in Arabidopsis, with differing affinities for NPR1 (Després et al., 2000; Zhou et al., 2000), and binding sites for the TGA factors occur in gene promoters responding to diverse stress signals (Chen and Singh, 1999; Pascuzzi et al., 1998). In view of this versatility of the transcription factors with potential to interact with NPR1, it is not surprising that evidence is emerging that the npr1 mutation affects diverse physiological processes in which SA is implicated, including senescence (Morris et al., 2000) as well as our new results on thermotolerance.

NPR1-related signal transduction could therefore be a target for improvement of thermotolerance, but did not exclusively determine thermotolerance. While the ability of npr1 mutants to recover from HS was impaired (although less severely than NahG), wild-type tolerance during HS was found in this mutant. It may be that an NPR1-dependent pathway is partially involved in recovery from HS, but an NPR1-independent pathway predominates during HS. There is evidence for SA-mediated, NPR1-independent responses in disease resistance and senescence (Clarke et al., 2000; Morris et al., 2000).

In the case of cpr5, it is interesting to find a gene whose mutation enhances thermotolerance. The cpr mutants have pleiotropic effects in pathogen defence and plant development, although it is still unproven that CPR proteins are normal components of pathogen resistance signaling (Clarke et al., 2000). Clarke et al. (2000) obtained evidence that elevated SA content is the primary determinant of pathogen resistance in cpr mutants. Yoshida et al. (2002) selected early senescence hys1 mutants allelic to cpr5, and isolated the gene by map-based cloning. Although the deduced HYS1/CPR5 sequence revealed no homology to known sequences, it has features of a signal transduction protein with a possible nuclear localization signal and five putative transmembrane domains. One interpretation of the recessive mutant phenotypes is that HYS1/CPR5 plays a repressive role in early processes for pathogen defence and senescence (Yoshida et al., 2002). Our data are consistent with a further repressive role for CPR5, in thermotolerance signaling, possibly mediated by SA. It remains to be seen whether CPR5 is an integral component of thermotolerance signaling, or whether thermotolerance in cpr5 is an abnormal pleiotropic effect, e.g. because of excessive SA levels.

A further uncertainty in thermotolerance signaling is the extent of interaction between different hormonal signaling pathways. Such cross-talk might be expected by analogy with pathogen defence responses, in which a complex of signaling pathways mediated by SA, jasmonic acid (JA), and ethylene is now recognized (Clarke et al., 2000). These pathways do not function independently, but influence each other through positive and negative regulatory interactions (Kunkel and Brooks, 2002). Recent studies with Arabidopsis mutants implicate ethylene and abscisic acid alongside SA in protection against HS (Larkindale and Knight, 2002), while induction of HSPs by methyl-JA has been reported (Ding et al., 2001; Hamilton and Coleman, 2001). It is therefore likely that the chemical and genetic effects we report do not involve SA action in isolation.

Our analyses of gene expression support our mutant studies in suggesting that thermotolerance and SAR share common features. Thus, genes associated with these two defense responses had overlapping expression patterns. We found either heat- or SA-induced expression of certain PR protein and HSP genes and their promoters as mentioned below.

HS induced PR-1 expression

We tested PR-1 expression because this group of pathogen- and SA-induced PR proteins are common markers for SAR (Malamy et al., 1992; Uknes et al., 1992). Although now there is evidence for an antifungal role, however, their biochemical action remains uncertain (Rauscher et al., 1999). We found that HS induced PR-1 expression. Although in certain cases, PR proteins are induced by abiotic stresses (Van Loon and Van Strien, 1999), there are relatively few reports of heat induction of PR genes. Margis-Pinheiro et al. (1994) reported that a bean acidic class IV chitinase is induced by heat and SA, although class III chitinase, β-1,3-glucanase, and thaumatin-like protein were not heat inducible. In Arabidopsis, the loss-of-thermotolerance hot2 mutation (Hong et al., 2003) maps close to the chitinase-like AtCTL1 gene (Zhong et al., 2002). PR-1 proteins, however, have been reported not to be heat inducible in tobacco (Ohashi and Matsuoka, 1985; Pfitzner et al., 1988). This may somehow reflect the temperature sensitivity of the disease resistance response in tobacco (Malamy et al., 1992), because in Arabidopsis, we found that not only the native PR-1 transcript but also the introduced tobacco PR-1a promoter was heat inducible.

SA-induced sHSP expression

Salicylic acid treatment of Arabidopsis induced expression of the native Hsp17.6 gene and an introduced soybean Hsp17.3B promoter, while Hsp17.6 was constitutively expressed in the SA-accumulating cpr5 mutant. The SA-concentration dependence of Hsp17.6 induction differed from that of PR-1 and indeed appeared more sensitive. Expression of Hsp17.6 was optimal at lower applied SA concentrations in the cpr mutants, possibly because of their higher endogenous SA levels. We also compared heat induction of this sHSP in genotypes with altered SA signaling. Interestingly, although heat induction of Hsp17.6 appeared similar, Hsp17.6 protein was reproducibly slower to appear during HS in NahG. However, Hsp17.6 had accumulated to similar levels in all genotypes by 5 h. SA treatment only partially induced the HSP response, however. SA-induced levels of Hsp17.6 were lower than those rapidly accumulating in HS. Moreover, SA did not induce Hsp101 expression, and expression of Hsp101 protein was unaltered in the mutants.

Few other plant studies have looked at SA effects on HSP expression. Cronjé and Bornman (1999) reported that SA enhanced Hsp70 expression in tomato cells, although effects were relatively small as this HSP is constitutively expressed. Methyl-SA increased accumulation of HSP mRNAs in tomato fruit (Ding et al., 2001). Interestingly, the human HS response can also be partially induced by SA, which induces a DNA-binding state of the human HS transcription factor, but apparently via a different phosphorylation pattern to heat, and SA alone does not induce HS gene expression (Jurivich et al., 1995).

The fact that both NahG and npr1 were capable of heat-induced Hsp17.6 and Hsp101 expression indicates that factors other than SA can induce HSPs. This is consistent with our conclusion that SA is not essential for acquired thermotolerance, in which Hsp101 in particular has been proven to play a key role (Gurley, 2000; Hong and Vierling, 2000; Queitsch et al., 2000). On the other hand, our observations that NahG and npr1 show poor basal thermotolerance, despite being capable of Hsp17.6 and Hsp101 expression, indicate that HSPs are not the sole determinants of basal thermotolerance. This accords with the recent genetic evidence provided by Hong et al. (2003), who found that the hot2 and hot4 mutants produced normal levels of Hsp101 and sHSPs, despite their failure to develop thermotolerance, and moreover that the map positions of these mutations fall outside intervals containing HSP genes.

Although our experiments therefore suggest a multiplicity of factors determining HSP expression and thermotolerance, it is possible that SA-induced sHSPs do contribute to thermotolerance. While the effects of SA on Arabidopsis Hsp17.6 expression were considerably less than those of HS, minor levels of sHSP induction can enhance basal thermotolerance. In plants whose HSP genes were de-repressed by a modified HS transcription factor, constitutive sHSP levels, only 20% of those reached in HS, were correlated with increased basal thermotolerance, although acquired thermotolerance was unchanged (Lee et al., 1995). sHSPs may preserve both protein and membrane integrity in HS. Arabidopsis Hsp17.6 has been shown to have protein chaperone activity (Sun et al., 2001), while Synechocystis sHSP may regulate membrane fluidity to preserve membrane structure during HS (Török et al., 2001) and a soybean sHSP appears to become membrane associated in HS (Lin et al., 1985). Therefore, expression of Hsp17.6 in SA-treated plants and the cpr5 mutant might at least partially contribute to their decreased electrolyte leakage during HS.

Decreased electrolyte leakage indicates a maintenance of physical order of cell membranes. It is well established that many physiological responses to environmental stress are caused by modifications of membrane structures (Carratu et al., 1996). Electrolyte leakage is a sensitive indicator of membrane damage and has been widely used to assess thermotolerance in plants (Howarth et al., 1997; Martineau et al., 1979; Srinivasan et al., 1996; Tal and Shannon, 1983). The phenotype of the hot2 mutant has recently confirmed that membrane properties are an essential aspect of thermotolerance (Hong et al., 2003). As apparently normal HSP levels failed to protect hot2 from increased ion leakage in HS, Hong et al. (2003) proposed that sHSPs cannot be the sole factor protecting membranes during HS. SA-induced changes in antioxidants and reduced oxidative damage (Dat et al., 1998a,b, 2000; Larkindale and Knight, 2002; Lopez-Delgado et al., 1998) may also protect cell membranes, but mutations such as hot2 could reveal further membrane-protective mechanisms. What seems clear from the present study is that maintenance of membrane properties is a major factor in the thermoprotective effects of SA.

In conclusion, SA promotes basal thermotolerance in Arabidopsis plants, hypocotyls and seeds, although SA is not essential for acquired thermotolerance. SA protects during HS and in the subsequent recovery. Mutations known to affect SA signaling in pathogen defences also affect thermotolerance. Common features of these defences are further suggested by induction of certain PR and HSP genes by both SA and HS. SA induces thermoprotective effects, possibly including certain HSPs in protection against membrane dysfunction and other damage. It is likely that SA represents one of the many factors regulating heat stress physiology in plants. As SA affects basal thermotolerance, its associated signaling pathways and thermoprotective mechanisms could be targets for genetic improvement of crop thermotolerance.

Experimental procedures

All chemicals were obtained from Sigma unless stated otherwise.

Plant materials

Arabidopsis thaliana (L.) Heynh. ecotype Col-0 was used as the wild-type control. The npr1, cpr1, and cpr5 mutant lines, from Xinnian Dong (Duke University), and the 35S-NahG line 10, from Scott Uknes (Ciba-Geigy Corp. USA; currently Cropsolution Inc., NC, USA), were in this ecotype background. A binary vector containing a soybean Hsp17.3B promoter–GUS fusion (Prändl et al., 1995) was donated by Ralf Prändl (University of Tübingen, Germany), and Col-0 Hsp17.3B–GUS transgenic lines were generated using the floral dip method as described by Clough and Bent (1998). A binary vector containing a tobacco PR-1a promoter–GUS fusion was donated by Uknes et al. (1993), and Landsberg erecta (La-er) PR-1a–GUS transgenic lines were derived from cultured root calli co-cultured with Agrobacterium tumefaciens following the protocol developed by Clarke et al. (1992).

Plant growth conditions

Arabidopsis plants were grown in 9-cm Petri dishes on Murashige and Skoog (MS; Duchefa, Haarlem, the Netherlands) medium (0.8% (w/v) agar; 3.0% (w/v) sucrose; pH 5.8) at 22°C with a 16-h photoperiod, 65 µmol m−2 sec−1. Vernalized seeds were surface sterilized in 20% (v/v) bleach for 5 min, rinsed twice in sterile H2O (Milli-Q water purification system; Millipore, Watford, UK) before plating onto MS agar plates and grown for 3 weeks to principal growth stage 1.08 (Boyes et al., 2001).

Heat treatment of plants

All heat treatments were carried out in the dark to ensure that cell death was a result of increased temperature and not photooxidative stress. Petri dishes containing 3-week-old plants on MS medium were heated at the appropriate temperature and specified time in a Sanyo MLR-350HT environmental test chamber, 70% relative humidity, without light. When electrolyte leakage was measured during the heat treatment, the aerial parts of 3-week-old plants were placed individually in glass scintillation vials containing 5 ml of H2O and placed in the environmental chamber without light (<1 µmol m−2 sec−1) at the appropriate temperature.

For the acquired thermotolerance assay, Petri dishes containing 3-week-old plants grown on MS medium were exposed to a 2 h, 45°C HS with (acclimated) or without (non-acclimated) a pre-treatment. The pre-treatment included 90 min, 38°C followed by a 2-h recovery period at 22°C prior to the HS. Following the 45°C HS, the plants were returned to 22°C for 24 h before the electrolyte leakage was measured.

To analyze the effect of heat stress on germination, vernalized seeds of each genotype were plated on MS medium in sectors of the same Petri dish and incubated for 2 h at 47°C, 0, 24, 30, and 48 h post-plating. Plates were then returned to 22°C and seed development was recorded daily until 10 days post-plating. Seeds were scored as either no germination, or principal growth stage (Boyes et al., 2001).

For heat treatment of 5-day-old seedlings, vernalized seeds of each genotype were plated in sections on MS medium in the same Petri dish and grown as described above for 5 days, when the cotyledons opened. Seedlings were then incubated at 47°C for 30, 45, 60, 90, or 120 min and returned to 22°C. The survival of the seedlings was scored daily until 7 days post-HS. Seedlings with green cotyledons were scored as viable.

The hypocotyl elongation assay was carried out as described by Queitsch et al. (2000). Vernalized seeds were surface sterilized and plated in rows on MS medium. Plates were covered with foil and placed in a vertical position at 22°C for 2.5 days. The plates were then either kept at 22°C or heat shocked at 45°C with (acclimated) or without (non-acclimated) a 90-min pre-treatment at 38°C. Hypocotyl elongation was examined after an additional 2.5 days.

Electrolyte leakage measurements

Heat-induced changes were quantified with a Horiba Twin Cond B-173 conductivity meter. For analysis of electrolyte leakage following a 16 h heat treatment, aerial parts of 3-week-old plants were placed in glass scintillation vials containing 5 ml of H2O and were incubated at 22°C in the light (65 µmol m−2 sec−1), unless otherwise stated. The conductance of the H2O was measured at intervals and calculated per milligram FW. The genotypes showed no significant difference in electrolyte content (data not shown). For experiments where electrolyte leakage was measured during a heat treatment, the aerial part of the plants were placed in glass scintillation vials containing 5 ml of H2O and were incubated in the environmental chamber at the appropriate temperature in the dark (<1 µmol m−2 sec−1). The conductance of the H2O was measured at intervals during the heat treatment and calculated per milligram FW.

Chemical treatment of plants

For experiments where SA or catechol was applied to the medium, 3 ml at the appropriate concentration in H2O was filter sterilized and pipetted around the plants 24 h prior to heat treatment. When plants were incubated in SA, individual plants were placed in a glass scintillation vial containing 5 ml of solution.

Leaf necrosis measurements

Plants grown as described above were incubated for 15 h at 22 or 38°C in darkness, and then returned to 22°C in light. After 3 days, all rosette leaves from each plant were detached and digital photographs were taken. The percentage of total plant leaf area showing necrosis was measured by analysis of these images using pc_image version 2.2 software (Foster Findlay Associates, Aberdeen, UK).

Histochemical GUS staining

Plants were placed into staining solution containing 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. The samples were then placed under vacuum twice for 2 min each time to infiltrate the samples and then incubated at 37°C overnight. The chlorophyll was removed by sequential changes in 30, 75, and 95% ethanol. For histochemical staining of heat-induced GUS activity, plants were returned to 22°C following the heat treatment for 24 h prior to staining.

Northern blot analysis

For RNA extraction, the aerial parts of the plants were harvested following the appropriate treatment and immediately frozen in liquid nitrogen. Total RNA was prepared from samples kept at −80°C, separated on a denaturing formaldehyde agarose gel, and transferred to Hybond N membrane (Amersham International Ltd.) as described by Draper et al. (1988). Hybridization was carried out as described by Warner et al. (1992) and the cDNA probe was labeled using the Life Technologies Random Primers DNA Labeling System. The AtHsp17.6 clone GenBank number X16076 (stock number CD3-5) was obtained from the Arabidopsis Biological Resource Center, and an AtHsp101 clone (Hong and Vierling, 2000) from Elizabeth Vierling (University of Arizona). The Arabidopsis PR-1 clone was from Uknes et al. (1992).

Western blotting

Proteins were extracted from the aerial part of the plant following the appropriate treatment and immediately frozen in liquid nitrogen. Frozen plant material was ground in buffer (100 mm Tris–HCl, pH 8; 5 mm EDTA; 5 mm DTT; 4 mm phenylmethylsulphonyl fluoride). Insoluble debris was removed by centrifugation at 10 000 g for 5 min. Western blotting was carried out as previously described by Warner et al. (1993). Protein concentrations were estimated using the Bradford protein assay. Ten micrograms of protein was separated electrophoretically on 12% SDS–polyacrylamide gels using a Bio-Rad Mini Protein II gel kit, and after separation transferred to Hybond-P membrane (Amersham International Ltd.) using a Millipore MilliBlot Graphite Electroblotter. Equal loading was confirmed by Ponceau S staining of the membrane. The antibodies, anti-Hsp17.6 (AZ 44I) and anti-Hsp101, were a gift from Prof. Elizabeth Vierling and used at a 1 : 500 and 1 : 1000 dilution, respectively (Hong and Vierling, 2001; Wehmeyer et al., 1996). Antigens were detected using goat antirabbit IgG alkaline phosphatase conjugate from DAKO, Denmark.

Analysis of SA and catechol

Plants grown in Petri dishes as described above were incubated at 38°C in darkness for the specified time. Plant shoots (up to 0.5 g FW) were frozen in liquid nitrogen and stored at −80°C prior to analysis. Tissue was ground in liquid nitrogen and then extracted for at least 3 h at 5°C in 20 ml of 80% methanol with addition of an internal standard of d6-SA (98 at.%; C/D/N Isotopes Inc., Quebec, Canada). After filtration, methanol was removed by rotary evaporation at 25°C and samples were centrifuged at 13 400 g for 3 min. Half of each supernatant was incubated overnight at 37°C with an equal volume of pH 4.5 buffer (0.2 m sodium acetate) in the presence of 10 U almond β-glucosidase EC (NBS Biologicals Ltd., Huntingdon, UK). The two sample fractions were each partitioned at pH 2 against an equal volume of ethylacetate, which was back washed against H2O and then reduced to dryness by rotary evaporation at 25°C. Samples were analyzed on a Waters Alliance 2690 liquid chromatograph, using a Waters Nova-Pak C18 cartridge (3.9 mm × 50 mm) eluted at 30°C at over 15 min with a 10–95% gradient of methanol in 2 mm formic acid at a flow rate of 0.5 ml min−1. One-tenth of the eluate was introduced into a Micromass LCT electrospray ionization mass spectrometer operating in negative ion mode at a sample cone voltage of 30 V, capillary voltage of 2.0 kV, and an extraction voltage of 5 V. SA was quantified by calibration of the molar ratio between the [M − H] ions at m/z 137 (SA) and m/z 141 (internal standard). The presence of catechol was investigated in the same analyses by monitoring for the m/z 109 [M − H] ion. Calibration against d6-SA indicated that the analysis was 5.6-fold less sensitive for catechol, giving a detection limit of c. 0.4 µg g−1 plant tissue. SA glucoside was estimated as the extra SA liberated in the glucosidase-treated fraction.


We are extremely grateful for the materials donated by Xinnian Dong, Scott Uknes, Elizabeth Vierling and Ralf Prändl. We thank Tallulah Crow for technical assistance and Jim Heald for operation of the LC–MS. Valuable contributions from Catherine Howarth (Institute of Grassland and Environmental Research, UK), and Steve Neill, John Hancock, and Radhika Desikan (University of the West of England, UK) are also gratefully acknowledged. This work was supported by the Biotechnology and Biological Sciences Research Council, and by an NZ Science & Technology Post-Doctoral Fellowship to S.M.C.