Heat stress in leaves under natural conditions is characterized by rapid fluctuations in temperature. These fluctuations can be on the order of 10 °C in 7 s. By using a specially modified gas-exchange chamber, these conditions were mimicked in the laboratory to analyse the biochemical response to heat spikes. The decline in ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) activity during prolonged heat stress is generally associated with an increase in ribulose 1,5-bisphosphate (RuBP) levels. However, rapid heating caused an initial decline in RuBP which was subsequently followed by a small decline in Rubisco carbamylation. The ratio of RuBP to Rubisco sites declined from a saturating concentration to a sub-saturating concentration, providing a possible mechanism for the decarbamylation of Rubisco. If RuBP is saturating (>1.8 RuBP Rubisco site−1), it acts as a cap on the catalytic site and keeps Rubisco activated. Measurements of triose-phosphate levels and NADP-malate dehydrogenase activation (a stromal redox proxy) indicated that the regeneration of RuBP by the Calvin cycle was limited by the availability of redox power.
High temperatures inhibit plant growth and reduce crop yields (Lobell & Asner 2003), with photosynthesis considered among the plant functions most sensitive to high temperature (Berry & Björkman 1980; Quinn & Williams 1985). Most temperate C3 plants exhibit a broad photosynthetic temperature optimum between 20 and 35 °C with peak CO2 assimilation often occurring below 30 °C. Increasing leaf temperature beyond this range inhibits photosynthesis. Measurements of leaf temperature using fine-wire thermocouples demonstrated that oak and cotton leaves not only experience moderate heat stress, but that this heat stress comes in the form of rapid heat spikes (Wise et al. 2004). Exploratory experiments examining the effect of this rapid heating demonstrated that ribulose 1,5-bisphosphate (RuBP) pools decline initially and then recover after sustained heating (Schrader et al. 2004). Experiments using longer, more constant heating periods missed this transient decline and consistently reported that RuBP levels increase and phosphoglyceric acid (PGA) levels decline during heat, reflecting the deactivation of ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) (Weis 1981; Law & Crafts-Brandner 1999; Schrader et al. 2004). This initial, transient decline in RuBP followed by recovery has also been mimicked by rapid changes in light (Mott et al. 1984) and by end-product-synthesis feedback on photosynthesis (Sharkey, Seemann & Berry 1986a). Mate et al. (1996) suggested that RuBP acts as a trap on the carbamate of Rubisco's catalytic site and that when the concentration of RuBP falls below 1.8 times the concentration of Rubisco sites, Rubisco decarbamylates. Rubisco carbamylation is required for catalytic activity and involves the reversible binding of CO2 to a lysine residue in the catalytic site followed by the rapid ionic bonding of Mg2+.
It has been suggested that the decline in photosynthesis at moderate temperature is caused primarily by deactivation of Rubisco because of the heat-labile nature of Rubisco activase (Crafts-Brandner & Salvucci 2000; Salvucci & Crafts-Brandner 2004a). Sharkey et al. (2001) found that in wild-type tobacco leaves, decarbamylation and inhibition of photosynthesis were rapidly reversible. However, in tobacco lacking Rubisco activase, heat caused a reduction of photosynthesis that was not reversible. Similarly, the rca Arabidopsis mutant lacking Rubisco activase suffers irreversible heat damage under conditions that cause a reversible decline in photosynthesis in wild type (Kim & Portis 2005). These observations indicate that processes other than Rubisco activation are also damaged by moderate heat stress and raise the possibility that the deactivation/decarbamylation at high temperature is an adaptive response to prevent irreversible damage.
Another potential mechanism of Rubisco deactivation at high temperature is a build-up of sugar-phosphate analogues of RuBP on Rubisco's catalytic site (Salvucci & Crafts-Brandner 2004b). However, this conclusion is based predominantly on work done in vitro and two other studies have shown that heat causes a loosening of Rubisco's catalytic site, causing these inhibitors to be released without the help of activase (Kim & Portis 2006; Schrader et al. 2006),although this may be dependent on the inhibitor formed during high temperature stress. An increase in pentodiulose bisphosphate and carboxytetritol bisphosphate during heat stress (Kim & Portis 2004) may not be as easily released from the catalytic site because of the tight binding of these RuBP analogues (Schrader et al. 2006).
Another proposed mechanism for the heat-induced decline in photosynthesis is the redox status of the stroma (Schrader et al. 2004). It is fairly clear now that heat stress induces cyclic electron transport and this induction is rapid (Bukhov et al. 1999; Bukhov, Samson & Carpentier 2000; Bukhov, Samson & Carpentier 2001; Schrader et al. 2004). However, neither oxidation of the stroma nor a decline in NADPH has been shown to impact Rubisco activity during heat stress. An oxidation of the stroma should inhibit many of the redox-sensitive enzymes within the stroma, including several in the Calvin cycle and sometimes affects the activity of Rubisco activase (Zhang, Schürmann & Portis 2001). A decline in NADPH availability should reduce the triose-phosphate pools and cause a build-up of PGA or phosphoglycolate. However, this seems contradictory to previous research which showed a decline in PGA pools with heat stress.
In this study, we examined the effects of rapid heating from 30 to 40 °C in approximately 17 s on the biochemistry of photosynthesis. From this we have tried to identify several of the early effects of heat stress on photosynthesis and how the biochemistry of photosynthesis adjusts to the early events.
MATERIALS AND METHODS
Tobacco (Nicotiana tabacum L. cv. W38) was germinated in Metro-Mix 366p growth medium (Grace-Sierra, Marysville, OH, USA) mixed with Osmocote Vegetable and Bedding Plant Food (Scotts-Sierra Horticultural Products, Marysville, OH, USA) in a Conviron growth chamber (Winnipeg, Manitoba, Canada) at the University of Wisconsin–Madison. Plants were watered daily with a solution of Miracle Gro's Excel 15-5-15 Cal-Mag (Miracle Gro, Marysville, OH, USA). Plant growth conditions were 25/18 °C day/night temperatures with an 18 h photoperiod. An equal mix of high-pressure sodium vapour lamps and metal halide high-intensity discharge lamps provided a photosynthetic photon flux of 500 µmol m−2 s−1.
Gas-exchange parameters were recorded using a modified fast-kill leaf cuvette equipped with a laboratory-made infrared heating element (Schrader et al. 2004). The heating element was constructed of 22 AWG nickel-chromium wire (60% nickel, 26% chrome and 14% iron; Arcor, Northbrook, IL, USA) woven into a Teflon insulator. The heating element was connected to a variable, step-down transformer (Powerstat, Bristol, CT, USA). Leaf temperature could be elevated by more than 12 °C in less than 10 s, which was sufficient to mimic naturally occurring rapid increases in leaf temperature. Once the desired temperature was reached, it could be maintained within ±0.5 °C to simulate sustained heating episodes. The heating element was mounted on an air cylinder that was used to rapidly pull the heating element away from the leaf cuvette when the fast-kill blocks were clamped. All gas-exchange parameters were recorded every 1 s by computer. The first or second fully expanded leaf from the top of a cotton plant was clamped in the fast-kill cuvette with circular disposable windows on top (Glad Cling Wrap; Best Brands Corporation, Danbury, CT, USA) and bottom (clear Teflon), exposing 11.46 cm2 of leaf tissue for gas exchange. Leaf temperature was monitored with a thermocouple pressed to the abaxial leaf surface. Gas entering the cuvette was controlled with Datametrics (Edwards High Vacuum, Wilmington, MA, USA) type 825 mass-flow controllers. The entering gas was humidified by bubbling a mixture of N2 and O2 gas through water, condensing the excess water in an ice-cooled copper coil, and then mixing 5% CO2 in air into the gas stream to achieve the desired CO2 partial pressure. Water and CO2 partial pressures in the air stream entering and exiting the cuvette were analysed with a Li-Cor 6262 infrared gas analyser (Li-Cor Inc., Lincoln, NE, USA). Light was provided by a KL1500 quartz halogen lamp (Schott Glas, Mainz, Germany) at 1500 µmol m−2 s−1 photosynthetic photon flux density, as determined by a Li-Cor quantum sensor.
Leaf metabolite analysis
Leaf metabolism was rapidly stopped during gas-exchange measurements by clamping the leaf between two 8.2 cm2 copper blocks that had been cooled in liquid nitrogen. Frozen leaf discs were immediately cut in half. One half of the sample was used for metabolite analysis and the other half for enzyme analysis. After the first half was ground to a powder, 3.5% HClO4 (v/v) was added and further ground. After thawing and centrifugation, the supernatant was neutralized with 2 N KOH, 0.15 M Hepes and 0.01 M KCl, and the precipitate was discarded after further centrifugation. The final solution was snap frozen and stored at −80 °C until analysis. Solutions were analysed by enzyme-linked photometric assays (Lowry & Passonneau 1972) using a dual-wavelength filter photometer (Sigma ZFP22; Sigma Instrumente, Berlin, Germany) measuring the absorbance difference between 334 and 405 nm. For analysis of RuBP and PGA, a 20 µL extract was buffered to pH 8.0 with 800 µL 50 mM bicine, 15 mM MgCl2, 1 mM ethylenediaminetetraacetic acid (EDTA), 20 mM NaCl, 15 mM NaHCO3, 5 mM phosphocreatine, 10 mM dithiothreitol (DTT), 1 mM ATP and 7 mM NADH. Reactions were started by adding 5 µL creatine kinase (1 U µL−1) and 5 µL glyceraldehyde-3-phosphate dehydrogenase (0.2 U µL−1). Upon completion of these reactions, 5 µL phosphoglycerate kinase (0.2 U µL−1) was added to determine PGA levels. Upon completion of this reaction, 5 µL of activated Rubisco was added to determine RuBP. RuBP results were divided by two to account for the stoichiometry of the Rubisco-catalysed reaction. For analysis of G6P, F6P and ATP, 20 µL extract was buffered to pH 8.0 with 800 µL 50 mM bicine, 5 mM MgCl2, 1 mM EDTA, 500 mM NADP+ and 0.25 mM glucose. Reactions were started by adding 5 µL G6P dehydrogenase to determine G6P. Upon completion of this reaction, 5 µL hexose isomerase was added to determine F6P. Upon completion of this reaction, 5 µL hexokinase was added to determine ATP. For analysis of ADP, 50 µL extract was buffered to pH 7.0 with 800 µL of 50 mM Hepes, 5 mM MgCl2, 90 mM KCl, 7 mM NADH and 0.3 mM phosphoenolpyruvate (PEP). Reactions were started by adding lactic acid dehydrogenase. Upon completion of this reaction, 5 µL pyruvate kinase was added to determine ADP. Triose-phosphate levels were determined as described in Sharkey & Loreto (1993). In brief, aliquots from leaf samples isolated for metabolites were buffered to pH 6.1 in 50 mM EPPS-KOH, 5 mM MgCl2, 20 mM KCl, 1 mM EDTA, 0.1 mM NADH, 5 mM ATP and 20 U of triose-phosphate isomerase. The reaction was started with the addition of 1.25 U α-glyceraldehyde-3-phosphate dehydrogenase.
Leaf Rubisco activity
Leaf samples stored at −80 °C in 1.5 mL tubes were briefly ground, then 800 µL extraction buffer containing 50 mM bicine pH 8.0, 30 mM NaCl, 5 mM MgCl2, 2 mM EDTA, 0.5% trion X-100 (Sigma Chemical Company, St. Louis, MO, USA), 10 mM mannitol, 0.5% polyvinylpolypyrollidone (PVPP), 1% bovine serum albumin (BSA), 5 mM DTT and 1 mM phenylmethanesulphonylfluoride. The extract was used for Rubisco activation and redox analysis. Extracts were briefly centrifuged for 10 s. Reactions to determine initial Rubisco activity were started immediately by adding 15 µL extract to a scintillation vial containing 200 µL of 50 mM bicine buffered at pH 8.0, 20 mM MgCl2, 0.2 mM EDTA, 15 mM NaH14CO3 (1 mCi mmol−1) and 0.5 mM RuBP, and stopped after 30 s with 200 µL formic acid. The RuBP was made in the lab from R5P and ATP. The reaction was carefully monitored to prevent high pH that can lead to impurities in the RuBP. Reactions to determine total Rubisco activity were conducted equivalently, with the exception that 177 µL extract was incubated for 5 min with 23 µL of 20 mM MgCl2, 15 mM NaHCO3 and 0.05 mM 6-phosphogluconate. Reactions to determine background radioactivity were conducted equivalent to initial Rubisco reactions except that formic acid was added before the leaf extract. All samples were evaporated in a sand bath overnight, rehydrated with 100 µL H2O and 3 mL Biosafe II liquid scintillation cocktail (Research Products International Corp., Mt. Prospect, IL, USA), and radioactivity was determined by scintillation counting for 2 min.
Initial and total Rubisco carbamylation levels were determined as described in Ruuska et al. (1998). In brief, aliquots of Rubisco isolated during extraction for activity were incubated with [14C] carboxyarabinitol bisphosphate (CABP) (initials) or with NaHCO3 and MgCl2 (totals) for 15 min at room temperature. An excess of cold CABP was added to initials and [14C]CABP was added to totals, and the mixtures were incubated for a further 30 min at room temperature and then stored on ice until assayed. Samples were then separated by gel filtration on Bio-Rad (Bio-Rad Laboratories, Hercules, CA, USA) Columns (27 × 0.7 cm) loaded with Sephadex G-50 Fine (Sigma Chemical Company, St Louis, MO, USA) that was pre-equilibrated with 20 mM EPPS-NaOH pH 8.0 and 75 mM NaCl, and eluted with the same buffer. The CABP-bound Rubisco fractions were collected and the total [14C]CABP was determined by liquid scintillation.
Leaf NADP-MDH activity
After analysing for initial Rubisco activity, analysis for initial NADP-MDH activity was initiated with 10 µL extract buffered to pH 8.0 with 50 mM bicine, 1 mM EDTA, 0.05 mM NADPH, 1 mM DTT and 0.01% BSA. Reactions were started by adding 5 µL of 200 mM oxaloacetic acid, 1.2 mM final concentration. Analysis for total NADP-MDH activity was conducted equivalently except that 100 µL extract was incubated for 5 min with 25 µL of 2 M bicine buffered to pH 9.0 and 200 mM DTT, 0.4 M bicine and 40 mM DTT final concentration. Analysis for NAD-MDH activity was initiated with 10 µL extract that had been diluted 50-fold and buffered to pH 8.0 with 50 mM bicine, 1 mM EDTA, 5 mM MgCl2 and 0.05 mM NADH. Reactions were started by adding 5 µL of 200 mM oxaloacetic acid. Solutions were analysed using a dual-wavelength filter photometer (Sigma ZFP22; Sigma Instrumente) measuring the absorbance difference between 334 and 405 nm.
Glycolate was determined using the Calkins method (Calkins 1943). In brief, aliquots from leaf samples isolated for metabolites were acidified in 36 N H2SO4 and 0.01% 2,7-dihydroxynaphthalene. Samples were homogenized on ice, boiled in water for 20 min and cooled to room temperature. H2SO4 (2 N) was added to bring the final concentration to 13.4 N H2SO4, and absorbance was measured at 530 nm on a Beckman DU-640 spectrophotometer (Beckman Coulter Inc., Fullerton, CA, USA). Glycolate concentrations were determined using a standard curve.
Gas exchange and rapid heating
When leaf temperature was rapidly increased from 30 to 40 °C (temperature transition completed in approximately 17 s), net photosynthesis declined instantaneously with the temperature increase by 20% under 370 μl l−1CO2/2% O2 (Fig. 1, top panel) and by 40% under 370 μl l−1 CO2/21% O2 (Fig. 1, bottom panel). The progressive decay of photosynthesis when held at a constant temperature of 42 °C was not apparent in this data because of the short measurement time. While photosynthesis declined, the CO2 concentration inside the leaves (Ci) increased during the heating event, indicating that stomatal closure is not responsible for the decline in CO2 assimilation. The initial decline in photosynthesis is consistent with an increased rate of photorespiration under ambient CO2/high O2 (Ogren 1984; Brooks & Farquhar 1985), but an increase in photorespiration cannot explain the decline under ambient CO2/low O2. Thus, other heat-labile components of photosynthesis must also be limiting CO2 uptake during these heating events.
RuBP and PGA pool sizes
Under atmospheric CO2 and O2 levels, RuBP levels declined rapidly upon heating (Fig. 2). Continued sampling past the initial heating event showed that RuBP remained low for at least 15 s after leaf temperatures reached a constant 40 °C. The ratio of RuBP to RuBP binding sites was below 1.8 at high temperatures when O2 was present (Fig. 3). PGA levels remained constant during the initial phases of heating but then dropped after approximately 15 s of constant heating. Under low O2, RuBP levels were initially higher than under atmospheric conditions but during heating RuBP levels did not decline as they did when O2 was present, remaining high at all measured stages of the heating event. PGA levels were lower under 2% O2 than under 21% O2, but did not decline during the heating event even after 15 s of constant heating.
Rubisco activity and carbamylation status
In the normal O2 concentration, Rubisco carbamylation was high (0.8) and remained high during the initial rise in temperature (Fig. 2). However, after 5 s of constant 40 °C, carbamylation declined slightly and remained down. Rubisco activity declined throughout the heating event, including during the initial rise in temperature. Under low O2, Rubisco carbamylation did not decline during heating while Rubisco activity did not decline until after 15 s of constant heating.
Under atmospheric conditions, MDH activity exhibited a steep initial decline in activity during the initial heating phase and remained low (Fig. 4). Under low O2, MDH activity remained high during the initial heating event, but then steadily declined during constant heating for 15 s.
Triose-phosphate pool size
Triose-phosphate pool size reflected MDH activity, declining rapidly during heating and remaining low under 370 μl l−1 CO2/21% O2 (Fig. 4). Under 370 μl l−1 CO2/2% O2, triose-phosphate pools remained high during heating then declined as temperature remained constant.
Glycolate pool size
The Calkins method for glycolate determination cross-reacts with many leaf metabolites including carbohydrates and thus does not solely reflect glycolate pools. However, some general observations can be made. Under photorespiratory conditions, estimates of glycolate increased as heating continued, while under non-photorespiratory conditions, estimates remained constant (Fig. 5). The increase in glycolate pools is consistent with an increase in photorespiration and its products during heat stress.
Rapidly imposed heat stress caused RuBP levels to fall and PGA levels to rise before any changes in Rubisco activation state were seen. The concentration of RuBP fell from 2.6 RuBP Rubisco site−1 to 1.3, a value low enough to allow deactivation (Mate et al. 1996). Schrader et al. (2004) first reported a decline in RuBP during a rapid heating event in cotton (Gossypium barbadense L. cv. Pima S-6). They suggested that this initial decline and subsequent recovery may be responsible, in part, for the observed decline in photosynthesis during moderate heat stress because of a decarbamylation of Rubisco. The supply of CO2 inside the leaf was higher during heating and so could not account for the decarbamylation observed.
Increasing temperature increases the rate of RuBP consumption by Rubisco (Eckardt & Portis 1997; Crafts-Brandner & Salvucci 2000; Salvucci et al. 2001) and if the rate of RuBP regeneration is not increased, RuBP levels will decline. This trend is reversed over longer heating periods because of a deactivation of Rubisco (Law & Crafts-Brandner 1999; Schrader et al. 2004). The decline in RuBP seen immediately after heating (Fig. 2) is also closely correlated with a rapid decline in NADP-MDH activity (Fig. 4). NADP-MDH activity is correlated with the redox status of the stroma and NADPH availability (Scheibe & Stitt 1988). Oxidation of the stroma could lower the activity of Rubisco activase because the larger isoform of Rubisco activase has cysteine residues that are redox regulated (Zhang & Portis 1999). If this were the initial and primary cause of the reduction in photosynthetic rate, we would expect an increase in RuBP and a decrease in PGA because of an inhibition of Rubisco. Because we see the opposite, a decrease in RuBP and no change or slight increase in PGA, we hypothesize that, because of an increase in cyclic electron transport, less NADPH is available to regenerate RuBP and that this is the first mechanism of inhibition of photosynthesis when heat stress is imposed. Triose-phosphate pools and RuBP pools declined in response to rapid heat stress, similar to NADP-MDH activity consistent with a limited availability of reducing power for the Calvin cycle. Thus, the limitation in RuBP regeneration during rapid heating could be caused by a decrease in reducing power in the stroma.
When the experiments were carried out in 21% oxygen, RuBP declined below the level needed to saturate Rubisco during the rapid heating and activation state was reduced, soon to be followed by reduced carbamylation. However, in low oxygen, RuBP levels did not decline below saturation, and Rubisco showed little or no deactivation or decarbamylation over 40 s. Heating in low oxygen also resulted in less oxidation of the stroma and this too may have allowed Rubisco to stay activated. The role of redox regulation is also seen in recent data of Salvucci, DeRidder & Portis (2006). Arabidopsis with 50% of the amount of activase as found in wild type (line rwt46), but all of in the redox-sensitive form, had a lower activation state during heating than plants with just 12% of wild-type activase that was all the shorter, insensitive form (line Δ43). In other words, deactivation at high temperature was enhanced by the redox sensitivity of activase. In data reported earlier, photosynthesis rwt46 plants showed significant recovery from heat stress although the full potential for recovery was not shown (Kim & Portis 2005). In any case, these observations suggest that deactivation and decarbamylation initially result from changes in metabolite levels and/or changes in stromal redox status rather than from an inherent sensitivity of activase to heat, although we fully accept that activase is inherently sensitive to heat (Salvucci et al. 2001).
Heating in 21% oxygen specifically stimulates photorespiration because the ratio of oxygenation to carboxylation increases with temperature (Farquhar, von Caemmerer & Berry 1980). An increase in photorespiration would lead to an increase in photorespiratory intermediates. The intermediates P-glycolate (Somerville & Ogren 1979) and glyxoxylate (Flügge, Freisl & Heldt 1980; Chastain & Ogren 1985, 1989) are known to cause a deactivation of Rubisco and thus a build-up of these intermediates during rapid heating may also influence photosynthesis. Estimates of glycolate showed an increasing trend during heating under ambient CO2/O2 while no increase was observed under ambient CO2/low O2 (Fig. 5). Low oxygen often results in high concentration of RuBP (Badger, Sharkey & von Caemmerer 1984), and over short time periods, the high level of RuBP, low level of glyoxylate and relatively higher stromal redox status in low oxygen appeared to protect photosynthesis during heating under ambient CO2/low O2. Overall, the photosynthetic parameters measured indicated that photosynthesis was initially stable to heat stress when O2 was low during rapid heating.
The physiological role of Rubisco deactivation has been hypothesized to be a mechanism for matching Rubisco activity with the capacity for RuBP regeneration. Thus, Rubisco deactivates in low light and, in Arabidopsis, this depends on the presence of the longer, redox-sensitive form of the enzyme (Zhang et al. 2002). When low light is imposed rapidly, RuBP levels first fall and then recover as Rubisco deactivates (Mott et al. 1984). When RuBP regeneration is limited by triose-phosphate use, ATP levels fall (Sharkey et al. 1986b) and Rubisco deactivates (Sharkey et al. 1986a). Again, RuBP levels first fall and then recover as Rubisco deactivates (Sharkey et al. 1986b). Is heat stress a third example where Rubisco deactivation is adaptive? If this were true, we expect that plants lacking activase would be more heat sensitive and this has been reported for both tobacco (Sharkey et al. 2001) and Arabidopsis (Kim et al. 2005). Data presented here show that RuBP first falls in response to heating and then recovers as Rubisco deactivates, although this happens much faster than in the case of low light or limited triose-phosphate use.
How might deactivation of Rubisco benefit the plant? One possibility is that the high rate of photorespiration at high temperature is deleterious. Glycolate doubled in leaves during the first 20 s at high temperature. Deactivation of Rubisco would reduce the rate of photorespiration and the benefits of this may outweigh the lost photosynthesis. A second possibility is that thylakoid membranes are more thermotolerant when there is a significant energy gradient across them. Thylakoid function is more damaged by heat stress given in the dark that when the same heat stress is given in the light (Weis 1982; Havaux, Greppin & Strasser 1991; Schrader et al. 2004), indicating that there could be a thermoprotective effect of a transthylakoid energy gradient. Deactivation of Rubisco would help keep a high thylakoid energy gradient at high temperature.
In summary, rapid heating, similar to heat stress that occurs naturally, indicates that there are significant changes in RuBP and other metabolites that precede changes in Rubisco activation. Rubisco activation may thus be a response to early events that occur when a leaf is heat stressed, and engineering plants to tolerate heat flecks may require understanding the primary effects of heat on the photosynthetic apparatus.
This work was supported by grants from Monsanto (T.D.S.) and by USDA grant no. 2002-35100-12057 (T.D.S.).