†Present address: Max Planck Institut für Molekulare Pflanzenphysiologie, Karl Liebknecht Strasse 25, 14476 Golm, Germany.
Environmental stresses commonly encountered by plants lead to rapid transient elevations in cytosolic free calcium concentration ([Ca2+]cyt) (Bush 1995;Knightet al. 1991). These cellular calcium (Ca2+) signals lead ultimately to the increased expression of stress-responsive genes, including those encoding proteins of protective function (Knightet al. 1996;Knightet al. 1997). The kinetics and magnitude of the Ca2+ signal, or ‘calcium signature’, differ between different stimuli and are thought to contribute to the specificity of the end response (Dolmetschet al. 1997;McAinsh & Hetherington 1998). We measured [Ca2+]cyt changes during treatment with mannitol (to mimic drought stress) in whole intact seedlings ofArabidopsis thaliana. The responses of plants which were previously exposed to osmotic and oxidative stresses were compared to those of control plants. We show here that osmotic stress-induced Ca2+ responses can be markedly altered by previous encounters with either osmotic or oxidative stress. The nature of the alterations in Ca2+ response depends on the identity and severity of the previous stress: oxidative stress pre-treatment reduced the mannitol-induced [Ca2+]cyt response whereas osmotic stress pre-treatment increased the [Ca2+]cyt response. Therefore, our data show that different combinations of environmental stress can produce novel Ca2+ signal outputs. These alterations are accompanied by corresponding changes in the patterns of osmotic stress-induced gene expression and, in the case of osmotic stress pre-treatment, the acquisition of stress-tolerance. This suggests that altered Ca2+ responses encode a ‘memory’ of previous stress encounters and thus may perhaps be involved in acclimation to environmental stresses.
To combat the effects of drought, plants express a number of genes which encode proteins of a protective nature including, for example, proteins which increase the accumulation of compatible solutes within the cell ( Savouréet al. 1995 ). The tolerance of plants to drought and other abiotic stresses is known to increase as a result of previous exposure to such stresses. An encounter with a particular type of abiotic stress can endow the plant with greater tolerance to both the same and different abiotic stresses ( Bridger et al. 1994 ;Gilmour et al. 1988 ;Jennings & Saltveit 1994;Levitt 1986;Mäntyläet al. 1995 ;Prasad et al. 1994 ;Siminovitch & Cloutier 1982). Thus, in nature the pre-treatment of plants with stress alters their ability to manage their response to subsequent stress episodes, resulting in greater survival.
We have recently shown that osmotic stress can mediate rapid elevations in cytosolic free calcium ([Ca2+]cyt) in Arabidopsis seedlings, and that these changes in Ca2+ levels mediate increases in the expression of drought-induced genes encoding proteins which have a protective function ( Knight et al. 1997 ). We have also previously shown that, in the case of low temperature at least, pre-treatment with cold alters subsequent cold-induced [Ca2+]cyt responses in Arabidopsis ( Knight et al. 1996 ). In this paper we describe research aimed at determining whether pre-treatment with abiotic stress alters the subsequent drought-induced Ca2+ signalling of Arabidopsis seedlings, and whether this could contribute to increased survival of drought treatment through increasing the levels of Ca2+-regulated drought gene expression.
Osmotic stress-induced [Ca2+]cyt responses ( Knight et al. 1997 ) were measured in Arabidopsis expressing recombinant aequorin, a Ca2+-activated luminescent protein ( Knight & Knight 1995;Knight et al. 1996 ). When seedlings were pre-treated with either hydrogen peroxide (to administer oxidative stress ( Price et al. 1994 )) or mannitol (to administer osmotic stress ( Knight et al. 1997 )) and then allowed to recover overnight, subsequent stimulation with mannitol elicited a modified form of the [Ca2+]cyt elevation; a restyled Ca2+ signal ( McAinsh et al. 1997 ). Figure 1 shows the effect of pre-treatments with oxidative stress upon subsequent osmotic stress-induced [Ca2+]cyt responses. Oxidative stress pre-treatments drastically reduced the magnitude of the osmotic stress-mediated [Ca2+]cyt response ( Fig. 1a). The magnitude of the reduction appeared to increase with increasing concentration of hydrogen peroxide used in the pre-treatment (1, 3 and 10 m m). Peak [Ca2+]cyt values for these data are shown in Fig. 1(c). One-way anova statistical analysis showed that the mean peak [Ca2+]cyt values observed in response to 0.44 m mannitol in plants pre-treated with 1, 3 or 10 m m hydrogen peroxide were all significantly different to those of the non-pre-treated control (P < 0.05). Mean peak values observed in plants pre-treated with 1 m m hydrogen peroxide were significantly different to those observed after pre-treatment with 10 m m hydrogen peroxide (P = 0.019), but not statistically significantly different to those of plants pre-treated with 3 m m hydrogen peroxide (P = 0.065). Similarly, mean peak [Ca2+]cyt values observed after pre-treatment with 3 m m hydrogen peroxide were not significantly different to those obtained after pre-treatment with 10 m m hydrogen peroxide (P = 0.94). This analysis suggests, therefore, that all treatments with hydrogen peroxide caused a statistically significant decrease in subsequent osmotic stress-induced [Ca2+]cyt responses. Furthermore, the severity of this effect seemed to be dependent on the concentration of the hydrogen peroxide applied (as determined by the significant difference between mean peak [Ca2+]cyt values in response to 1 m m and 10 m m hydrogen peroxide). The reductions in osmotic stress-induced [Ca2+]cyt responses were not due to significant tissue damage or general toxic effects of the hydrogen peroxide treatment. This was shown by the fact that the response of seedlings pre-treated at the highest concentration of hydrogen peroxide (10 m m) to subsequent cold shock appeared virtually unchanged ( Fig. 2a), with cold-induced [Ca2+]cyt peak values showing no significant difference to controls (P > 0.6). Also, when seedlings which had been pre-treated with 10 m m hydrogen peroxide (the highest concentration we used) were returned to normal conditions, they continued to grow, developing normal root and aerial tissues (data not shown). Pre-treatments at concentrations of hydrogen peroxide higher than 10 m m had toxic effects (data not shown), and therefore the 10 m m hydrogen peroxide pre-treatment regime was chosen for subsequently examining the effects on osmotic stress-induced gene expression (see below). To ascertain whether the pre-treatment with hydrogen peroxide had itself any direct effect on Ca2+ signalling, we measured [Ca2+]cyt responses upon addition of 10 m m hydrogen peroxide. Figure 2(b) shows that when 10 m m hydrogen peroxide was added to Arabidopsis seedlings this provoked a rapid increase in [Ca2+]cyt, as previously described in tobacco ( Price et al. 1994 ). This same treatment was also sufficient to induce the expression of genes encoding proteins which play a detoxifying role during oxidative stress. We show here that the expression of glutathione-S-transferase (gst) was greatly induced by this treatment after 1 h ( Fig. 2c).
When seedlings pre-treated with hydrogen peroxide were allowed a further 24 h after the end of the pre-treatment and before investigating the effects on osmotic stress-induced [Ca2+]cyt responses, a marked recovery phenomenon was observed ( Fig. 1b,d). All three pre-treatments (1, 3 and 10 m m mannitol) produced [Ca2+]cyt responses that were relatively higher that those obtained 24 h previously (compare Fig. 1a,c with Fig. 1b,d). This suggests that the inhibitory effect of hydrogen peroxide pre-treatment upon subsequent osmotic stress-mediated Ca2+ signalling is reversible. One-way anova statistical analysis showed that pre-treatment with 1 m m hydrogen peroxide showed total reversibility after 40 h (average peak [Ca2+]cyt value of 1 m m hydrogen peroxide-pretreated plants showing no significant difference to control (P = 0.97)). Pre-treatments with 3 and 10 m m hydrogen peroxide also showed reversibility after 48 h, but this was only partial, with the average peak [Ca2+]cyt values being significantly lower than at 1 m m (P = 0.006 and 0.005, respectively). There was no significant difference between the average peak [Ca2+]cyt values observed in plants pre-treated with 3 m m and 10 m m hydrogen peroxide after 40 h (P = 1). Figure 1(d) shows the peak values for the subsequent osmotic stress-induced [Ca2+]cyt responses. As 1 m m hydrogen peroxide showed significant inhibition of the mannitol-induced [Ca2+]cyt elevation after 16 h ( Fig. 1a), the fact that 1 m m hydrogen peroxide shows no significant difference to the control after 40 h ( Fig. 1b) suggests that the inhibitory effect of hydrogen peroxide is totally reversible.
Opposite to the effect observed with hydrogen peroxide pre-treatments ( Fig. 1), pre-treatments with mannitol caused an increase in the magnitude of the subsequent osmotic stress-induced [Ca2+]cyt response ( Figs 3 and 4). After pre-treatment with 0.22 m mannitol, a significant increase in the magnitude of the mannitol-induced [Ca2+]cyt elevation was seen ( Fig. 3a). Peak [Ca2+]cyt values are shown in Fig. 3(b); one-way anova analysis shows that the peak [Ca2+]cyt values from mannitol pre-treated and control samples are significantly different to each other (P = 0.01). When a lower concentration was used (0.11 m mannitol) for pre-treatment, no significant effect on peak [Ca2+]cyt values was seen (P = 0.3) 24 h after the end of the pre-treatment ( Fig. 4a). However, if the plants were left for an additional 24 h after the end of the pre-treatment, a significant (P = 0.01) increase in mannitol-induced [Ca2+]cyt elevation was observed ( Fig. 4b). Peak [Ca2+]cyt values are shown in Fig. 4(c). This observation implies that the concentration of mannitol in the pre-treatment and the time elapsed since the pre-treatment are both important in mediating the increase in subsequent mannitol-induced [Ca2+]cyt response, i.e. a lower concentration of mannitol for pre-treatment may produce the same effect if given long enough. Concentrations higher than 0.22 m mannitol for pre-treatment produced toxic effects (e.g. see Fig. 6b). Therefore, 0.22 m mannitol was used as our standard pre-treatment to subsequently examine the effect on osmotic stress-induced gene expression (see below).
During exposure to environmental stress, the expression of a number of genes is induced in plants in order to drive the cellular changes necessary for survival ( Vierling & Kimpel 1992). It has been shown that the expression of proteins is highly sensitive to changes in calcium signature ( Dolmetsch et al. 1997 ). Therefore, we investigated whether the alterations in Ca2+ response mediated by stress history ( Figs 1, 3 and 4) also correlated with modified Ca2+-regulated osmotic stress-induced gene expression. The mannitol-induced expression of two Arabidopsis genes, p5cs ( Savouréet al. 1995 ) and rab18 (La°ng et al. 1994), which are regulated by the osmotic stress-induced [Ca2+]cyt increases shown in Figs 3 and 4 ( Knight et al. 1997 ), was monitored in previously stressed and control seedlings ( Figs 5 and 6). Oxidative stress pre-treatment substantially inhibited the subsequent osmotic stress induction of both rab18 and p5cs ( Fig. 5a,b), correlating with the reduction in osmotic stress-induced [Ca2+]cyt response seen in plants treated in the same way ( Fig. 1a). Conversely, osmotic stress pre-treatment promoted subsequent osmotic stress-induced p5cs expression ( Fig. 6a), again mirroring the altered Ca2+ signal output ( Fig. 3a). Overall, the correlations observed indicate that stress history mediated alterations in the Ca2+ response might be used in nature to mediate altered patterns of stress gene expression. Tolerance of plants to low temperatures is known to increase after previous exposure to cold ( Gilmour et al. 1988 ), drought ( Jennings & Saltveit 1994;Mäntyläet al. 1995 ) or oxidative stress ( Prasad et al. 1994 ). Similarly, cold pre-treatment subsequently renders plants less susceptible to oxidative stress ( Bridger et al. 1994 ). Thus, in nature the pre-treatment of plants with stress alters their ability to manage subsequent stress episodes, resulting in greater survival. This has been shown to be the case with drought, where a drought pre-treatment mediates changes in the plant which increase its ability to survive subsequent episodes of drought ( Levitt 1986;Siminovitch & Cloutier 1982). This phenomenon was observed in our experimental system with seedlings which had been pre-treated with mannitol ( Fig. 6b). In this case, reduced occurrence of chlorosis in mannitol pre-treated plants indicated that they were better able to survive during prolonged osmotic stress. Figure 6(b) shows that the same mannitol pre-treatment that we used prior to the [Ca2+]cyt measurement ( Fig. 3) and measurement of gene expression ( Fig. 6a), endowed plants with the ability to survive more effectively during extended periods in the presence of 0.3 and 0.4 m mannitol.
In Arabidopsis seedlings, pre-treatment with osmotic stress or oxidative stress caused modifications of the Ca2+ responses elicited by subsequent challenge with osmotic stress ( Figs 1, 3 and 4). This suggests that information relating to environmental signals may be transduced differently depending on the stress history of the plant. The nature of the modification of the osmotic stress Ca2+ response varied with the type of pre-treatment (decreasing after hydrogen peroxide pre-treatment and increasing after mannitol pre-treatment). Furthermore, a given pre-treatment does not globally alter Ca2+ signalling as the responses to different subsequent stimuli are affected in different ways. For instance, after hydrogen peroxide pre-treatment (which reduces the osmotic stress-mediated [Ca2+]cyt response ( Fig. 1a)), the magnitude of the [Ca2+]cyt response to cold is the same ( Fig. 2a). Therefore, the changes in signalling we observe are clearly not a global event and may reflect a more subtle and specific resetting of sensitivity to particular signals. Information relating to previous encounters with stress could thus be regarded as being encoded via a kind of chemical ‘memory’. This information is then retrieved upon subsequent stimulation, resulting in an altered Ca2+ signal. Subtle alterations in Ca2+ responses (producing altered ‘calcium signatures’) have been shown to have profound effects on the downstream events mediated by Ca2+ ( Dolmetsch et al. 1997 ). The altered Ca2+ responses we describe here may encode the necessary information to allow a stress response appropriate to the context of the life history of the plant. The concept of signal information storage in plants has been proposed previously ( Verdus et al. 1996 ) and Ca2+ has been implicated ( Verdus et al. 1997 ). The work presented here suggests that such stored information can be retrieved and manifested as an altered Ca2+ signal.
When Arabidopsis seedlings were pre-treated with hydrogen peroxide, they subsequently showed both a reduced osmotic stress-mediated [Ca2+]cyt response ( Fig. 1a,c) and reduced levels of osmotic stress-mediated gene expression ( Fig. 5). Why this inhibition of osmotic stress-mediated gene expression occurs is not clear. It is clear, however, that oxidative stress pre-treatment seems to desensitise the pathway leading to osmotic stress-mediated induction of the genes p5cs and rab18. However, hydrogen peroxide pre-treatment does not impair the ability of Arabidopsis to survive osmotic stress in our experimental system (data not shown), even though osmotic stress-mediated gene expression is inhibited ( Fig. 5). As drought itself causes oxidative stress ( Moran et al. 1994 ) and antioxidant systems are important in drought tolerance ( Van Rensburg & Kruger 1994), it may be that the pathways leading to the induction of genes protecting plants from either oxidative stress or drought actually interact with each other. It may be that the production of active oxygen species (AOS) in response to drought is itself an intracellular signal to induce drought-regulated genes. If this is the case it can be envisaged that prolonged oxidative stress might lead to the desensitisation of AOS-mediated drought-gene expression in such a way that further induction by drought is inhibited. Alternatively, it may be that oxidative stress-induced signalling pathways act antagonistically to osmotic stress signalling pathways, and hence flux through the former inhibits the latter through a cross-talk mechanism. Further work, however, is required to discriminate between all these possibilities.
The situation regarding the mannitol pre-treatments is somewhat clearer. Mannitol pre-treatments resulted in an altered [Ca2+]cyt response to osmotic stress with a higher peak value than in non-pre-treated plants ( Fig. 3a). This pre-treatment also results in greater expression of genes proven to help in drought survivability ( Fig. 6). Overexpression of p5cs in transgenic plants has been previously shown to endow plants with improved drought tolerance ( Kishor et al. 1995 ). Pre-treatment with mannitol endows seedlings with an enhanced ability to induce p5cs and rab18 gene expression in response to subsequent osmotic stress ( Fig. 6a). The same pre-treatment does also promote greater survivability during extended periods in mannitol, as shown in Fig. 6(b). Hence it seems possible that alterations in Ca2+ response may be, at least in part, responsible for the phenomenon of acclimation to abiotic stresses in plants, at least in the case of osmotic stress.
Taken together, the results presented in this paper suggest that another tier of information can be encoded by a stress-induced Ca2+ response, i.e. information relating to previous encounters with stress. Episodes of stress may thus be recorded in a ‘memory’ which redefines subsequent Ca2+ signalling pathways at the level of the [Ca2+]cyt changes themselves. The manner in which the subsequent stress-induced [Ca2+]cyt response is affected seems to depend on the nature of the stimulus used in the pre-treatment ( Figs 1 and 3). In terms of mechanism, it will be interesting in the future to determine whether the effects described in this paper occur as a result of an effect on external or internal Ca2+ stores specifically. It has been suggested that as expression of phospholipase C (PLC) is induced by drought, this may be a mechanism by which to amplify the drought signal, possibly by enhancing the release of Ca2+ from internal stores by the IP3 produced by PLC ( Hirayama et al. 1995 ). It would also be interesting to ascertain whether these effects require de novo gene expression in order to occur. It has been proposed that stress-activated Ca2+ channels most probably constitute the primary sensing mechanism for at least one environmental stress ( Ding & Pickard 1993;Minorsky 1989). If this is the case, our data could indicate that previous stress experiences determine the initial perception of future stimuli. These observations also have implications for experimental design and procedure as they highlight the importance of reproducible pre-experimental growth conditions when studying signalling in plants. Similar conditionality on growth conditions has been observed in guard cell Ca2+ signalling ( Allan et al. 1994 ). Much of the ‘variability’ encountered between individual organisms or cells in studies of signal transduction may thus be as a result of differing histories of stress encounters. As Ca2+ is a ubiquitous second messenger transducing the effects of myriad of signals in plant, animal and fungal cells, it will be interesting to determine whether the phenomenon of stress history altered Ca2+ signalling described here is also ubiquitous.
Plant material and chemicals
All experiments were performed using seedlings of transgenic Arabidopsis expressing recombinant aequorin, a Ca2+-activated photoprotein ( Knight et al. 1991 ). Seedlings were grown on agar plates containing Murashige and Skoog nutrient medium as described previously ( Knight et al. 1996 ) and were 6 to 7 days old at the beginning of subsequent stimulation with mannitol.
Hydrogen peroxide (30% solution) was obtained from Sigma (Poole, UK) and mannitol (AnalaR grade) from BDH (Poole, UK).
In vivo [Ca2+]cyt measurements using luminometry
Changes in [Ca2+]cyt were measured non-invasively by luminometry of intact seedlings of transgenic Arabidopsis expressing recombinant aequorin. Reconstitution of aequorin was performed in vivo essentially as described previously ( Knight et al. 1991 ) by floating seedlings on water containing 2.5 μm coelenterazine in the dark, overnight at 20°C. Experiments were performed by putting a single seedling in a plastic luminometer cuvette which was placed inside a digital chemiluminometer with a discriminator ( Knight et al. 1991 ;Knight et al. 1996 ). Luminescence counts were integrated every 0.1 sec and subsequently calibrated ( Knight et al. 1996 ) to obtain [Ca2+]cyt values. Statistical significance of differences in average peak [Ca2+]cyt values between treatments in each experiment were assessed by performing a standard one-way anova statistical analysis (significance taken as P < 0.05). After 5 sec of counting, cold water, mannitol or hydrogen peroxide solutions were injected via a light tight port. 0.4 ml of mannitol (to mimic the effect of drought ( Knight et al. 1997 )), or hydrogen peroxide ( Price et al. 1994 ) were added to a seedling floating in 0.2 ml water resulting in a final concentration of 0.44 m mannitol or 10 m m hydrogen peroxide in the cuvette. Cold shock was performed by injecting 1 ml cold water at 10°C into a cuvette containing only the seedling. All solutions were injected slowly to minimise the effect of mechanical stimulation on [Ca2+]cyt ( Knight et al. 1997 ). All traces presented are means of five or six individual seedling traces and all experiments were performed at least four times.
Oxidative stress pre-treatment was performed by floating seedlings on solutions of 1, 3 or 10 m m hydrogen peroxide for 6 h ending 16 h before subsequent stimulation. Mannitol pre-treatment was performed by floating seedlings on a solution of 0.11 m or 0.22 m mannitol for 24 h ending 24 h before subsequent stimulation. Control treatments were made by floating seedlings on water for 6 or 24 h. At the end of the pre-treatment period, the seedlings were rinsed in water and then floated in fresh water for 16 or 24 h (hydrogen peroxide and mannitol pretreatments, respectively) to allow recovery. At this stage, 2.5 μm coelenterazine was added to allow reconstitution of aequorin ( Knight et al. 1991 ) in those seedlings to be used for [Ca2+]cyt. measurements. In some experiments, [Ca2+]cyt. measurements were made a further 24 h after the first set of measurements. During this subsequent 24 h period, the seedlings remained floating in water in Petri dishes and were returned to normal growth room conditions.
The pre-treatment conditions were selected by testing the effect of range of concentrations of hydrogen peroxide or mannitol for differing periods of time, on the viability and physical appearance of the seedlings (data not shown). The highest concentrations of mannitol or hydrogen peroxide used for pre-treatments (0.22 m mannitol or 10 m m hydrogen peroxide) were the highest concentrations out of those tested not to cause detrimental effects. Neither 0.22 m mannitol nor 10 m m hydrogen peroxide caused visible damage to the seedlings over the course of the experiment and plants which were returned to MS-agar plates under normal growth conditions and observed for 2 weeks continued to grow after these pre-treatments (data not shown).
Stress gene transcript analysis by reverse-transcriptase-PCR
Approximately 25 seedlings per treatment were floated on water, mannitol or hydrogen peroxide solutions for the appropriate amount of time. Mannitol and hydrogen peroxide pre-treatments were performed exactly as above, but without the addition of coelenterazine and subsequently treated with water or 0.44 m mannitol for 0.5 to 2 h. The effect of hydrogen peroxide on gst expression was measured by treating seedlings with 10 m m hydrogen peroxide for 1 h. Total RNA was extracted from seedling tissue using RNeasy plant minipreps (Qiagen, Dorking, UK) and cDNA synthesised as described previously ( Knight et al. 1997 ). p5cs, rab18, gst and aeq (control) transcript levels were visualised by reverse-transcriptase PCR (RT–PCR) exactly as described previously ( Knight et al. 1997 ) using 10 μl of a 1:100 dilution of cDNA for 31 (p5cs, rab18 and gst) or 24 (aeq) amplification cycles. Aequorin is expressed constitutively in the plants and aeq transcript levels were used as an internal control for the starting amount of RNA used. Oligonucleotide primers were obtained from Genosys (Cambridge, UK) and sequences for primer pairs used to amplify aeq, rab18 and p5cs transcripts have been published previously ( Knight et al. 1996 ;Knight et al. 1997 ). The sequences of the primers used to amplify gst transcripts were as follows:
gst-forward (5′-3′): TTGCTTCTTGCTCTTAACCC
gst-reverse (5′-3′): CTCAACCTTCTCCAAATTCC
PCR products were visualised on ethidium bromide agarose gels as described previously ( Knight et al. 1997 ).
We would like to thank the Gatsby Foundation for the award of a summer studentship to S.B. and the BBSRC for financial support. M.R.K. is a Royal Society University Research Fellow.