Photosynthesis is particularly sensitive to heat stress and recent results provide important new insights into the mechanisms by which moderate heat stress reduces photosynthetic capacity. Perhaps most surprising is that there is little or no damage to photosystem II as a result of moderate heat stress even though moderate heat stress can reduce the photosynthetic rate to near zero. Moderate heat stress can stimulate dark reduction of plastoquinone and cyclic electron flow in the light. In addition, moderate heat stress may increase thylakoid leakiness. At the same time, rubisco deactivates at moderately high temperature. Relationships between effects of moderate heat on rubisco activation and thylakoid reactions are not yet clear. Reactive oxygen species such as H2O2 may also be important during moderate heat stress. Rubisco can make hydrogen peroxide as a result of oxygenase side reactions and H2O2 production by rubisco was recently shown to increase substantially with temperature. The ability to withstand moderately high temperature can be improved by altering thylakoid lipid composition or by supplying isoprene. In my opinion this indicates that thylakoid reactions are important during moderate heat stress. The deactivation of rubisco at moderately high temperature could be a parallel deleterious effect or a regulatory response to limit damage to thylakoid reactions.
Three papers in this issue of Plant, Cell and Environment describe new understandings of how moderate heat stress affects photosynthesis. Haldimann & Feller (2005) show that pea plants grown in different temperatures exhibit similar deactivation of rubisco at 38 °C but adjust so that thylakoid reactions are more tolerant of high temperature when the plants are grown at high temperature. The other two papers (Peñuelas et al. 2005; Velikova & Loreto 2005) explore thermoprotection of photosynthetic capacity by isoprene. These papers also explore the relationships among heat stress, thermotolerance, and reactive oxygen species. In this opinion I discuss the importance of moderate heat stress effects on photosynthesis and a number of new observations on the mechanisms by which heat may damage photosynthesis.
High leaf temperatures reduce plant growth and limit crop yields. Estimates range up to a 17% decrease in yield for each degree Celsius increase in average growing season temperature (Lobell & Asner 2003). The global mean temperature increased by 0.6 °C from 1990 to 2000 and is projected to increase by another 1.4 to over 5 °C by 2100 (Houghton et al. 2001; IPCC 2001). This is driven in large measure by increasing atmospheric CO2, which is up by 31% since 1750. High leaf temperature is also a consequence of drought because plants lose the ability for transpirational cooling when water availability is limited. It has been specifically shown for cotton that the greater the transpirational cooling, the greater the yield (Radin et al. 1994; Lu et al. 1997), proving that heat-stress-induced reductions of photosynthesis can limit overall yield.
Leaves of many plants are very thin and so have minimal heat capacity and as a result, leaves exposed to full sunlight can warm up substantially above air temperature. Using fine-wire thermocouples it has been shown that leaves with low transpiration rates, such as oak leaves, suffer frequent high-temperature episodes during which leaf temperature can exceed air temperature by as much as 15 °C (Singsaas & Sharkey 1998; Hanson et al. 1999; Singsaas et al. 1999). The high transpirational cooling of cotton can also lead to rapid and large variation in leaf temperature with peak temperatures frequently in the range of 35–40 °C (Wise et al. 2004). Thus, heat stress must be considered in terms of both high air temperature and fluctuations in leaf temperature related to the energy balance of leaves. Leakey, Press, & Scholes (2003) found that high temperature was more deleterious under simulated light-fleck conditions than under steady-state conditions. To understand the effects of moderately high temperature under realistic conditions, the simple ‘cook and look’ methods that have been the most common protocol for exploring effects of moderate heat on photosynthetic capacity up to now, should be supplemented by more realistic temperature regimes.
Even in the absence of heat stress injury, photosynthesis would be expected to decline as temperature increases because photorespiration increases with temperature faster than does photosynthesis (Schuster & Monson 1990). However, it is now well established that moderately high temperature (e.g. 35–40 °C) reduces the rate of photosynthesis more than can be explained by stimulated photorespiration. In the past, photosystem II (PSII) was considered a key weak link (Santarius 1975; Santarius & Müller 1979; Berry & Björkman 1980; Enami et al. 1994) but damage to PSII only occurs at high temperatures, often above 45 °C (Terzaghi et al. 1989; Thompson et al. 1989; Gombos et al. 1994; Çjánek et al. 1998; Yamane et al. 1998). The reversible effects of moderate heat stress on PSII likely represent regulation to match reductions in the capacity of downstream reactions. Reactions downstream of PSII are inhibited at lower temperature than that needed to damage PSII. Thus, damage to PSII cannot explain the heat-induced depression in photosynthesis seen at temperatures between 35 and 40 °C. This is a surprising conclusion that is counter to the prevailing wisdom. Nevertheless, it is now well-confirmed that moderate heat stress normally encountered by most plants does not damage PSII, but does substantially reduce the rate of photosynthesis.
When leaves that have never been heat-stressed are first exposed to moderately high temperature, photosynthetic electron flow can be stimulated (Schrader et al. 2004). However, upon returning the leaf to the pre-stress temperature, the capacity for photosynthetic electron flow can be substantially inhibited relative to the control situation (Schrader et al. 2004). Because of the frequent, wide fluctuations of individual leaf temperature, the inhibition of electron transport capacity following heat stress is physiologically relevant. As reduced electron transport capacity is sometimes seen only following heat stress, the effects of moderate heat stress on thylakoid reactions are easily missed despite their physiological relevance. Heat stress studies should include examination of the consequences of one or more heat stress episodes, not simply the rate of photosynthesis during the first heat stress episode experienced by a leaf.
Therefore, the damage to thylakoid reactions by moderate heat stress is not a result of damage to PSII, but probably involves pathways related to cyclic electron flow and perhaps the cytochrome complex. Substantial progress in understanding cyclic electron flow and the cytochrome complex has been reported recently and the relevance of these new findings to heat stress will now be described.
Cyclic electron flow
The amount of energy consumed in cyclic and pseudocyclic (a.k.a. the water–water cycle (Asada 1999)) electron flow has been debated, with most studies reporting relatively low fluxes in either pathway under normal steady-state conditions. PSI-mediated cyclic electron flow can occur by either of two routes (Joët et al. 2001), an antimycin A-sensitive route involving ferredoxin plastoquinone reductase or a second route involving the NAD(P)H dehydrogenase complex (Thomas et al. 1986; Boucher, Harnois, & Carpentier 1989; Bukhov et al. 2000; Joët et al. 2001). Joliot & Joliot (2002) hypothesized that cyclic electron flow occurs in structurally isolated electron transport chains that include PSI, cyt b6f complex, one ferredoxin and one plastocyanin. These isolated chains would account for many PSI complexes when leaves are dark-adapted. The thylakoid membrane is hypothesized to undergo a structural rearrangement during photosynthetic induction so that linear electron flow dominates during steady-state photosynthesis.
Makino, Miyake & Yokota (2002) investigated the physiological function of PSI-mediated cyclic electron flow and the water–water cycle. They found that when dark-adapted leaves were first illuminated, there was a substantial electron flow through the water–water cycle. As carbon fixation began, linear electron flow to carbon predominated but a low level of cyclic electron flow mediated by PSI was found at high light. Upon switching to low oxygen, PSI activity increased whereas PSII activity decreased indicating a switch from the water–water cycle to PSI-mediated cyclic electron flow. The Km for oxygen of the endogenous Mehler reaction (part of the water–water cycle) is equivalent to 8–15% O2 in the atmosphere so it is expected that an atmosphere with 2% O2 will effectively eliminate the water–water cycle (Furbank & Badger 1983). Makino et al. (2002) interpreted their results to indicate that during the induction phase of photosynthesis the water–water cycle and PSI-mediated cyclic electron flow play critical roles in establishing a transthylakoid energy gradient. The disadvantage of the water–water cycle is the generation of activated oxygen species, which must be quenched before damage can occur. PSI-mediated cyclic electron flow (which would generate the transthylakoid energy gradient needed to regulate PSII (Demmig-Adams & Adams 1992) could be a mechanism for limiting PSII activity and so limit the production of activated oxygen species (Heber & Walker 1992; Heber 2002).
Stimulation of cyclic electron flow by heat stress
Heating appears to specifically engage cyclic electron flow around PSI (Havaux 1996; Bukhov, Boucher, & Carpentier 1998; Bukhov et al. 2000). The dark reduction of PSI was found to undergo ‘spectacular acceleration’ with the half-life of P700+ falling from over 500 ms to less than 50 ms between 34 and 40 °C (Havaux 1996). Havaux (1996) also reported that following heating, plastoquinone was reduced in darkness by stromal reductants, and could be oxidized by brief illumination of far red light (preferentially exciting PSI). A flow of electrons from the stroma to the plastoquinone pool in the dark at 36 °C was also reported by Yamane et al. (2000). In these studies a specific protein was hypothesized that might catalyse electron donation to plastoquinone and that would only be active above 35 °C. Yamane et al. (2000) found little effect of antimycin A, the inhibitor of ferredoxin-plastoquinone reductase-type cyclic electron flow. They also saw only minor inhibition of the high temperature electron flow to plastoquinone by feeding inhibitors of Ndh1, which catalyses the other known cyclic electron transport path. Haldimann and Feller report in this issue that the dark reduction of plastoquinone following high temperature stress is one of the processes adjusted by growing pea plants at high temperature. Low-temperature-grown pea plants exhibited dark reduction of plastoquinone following a 38 °C leaf temperature treatment but plants grown at 35 °C did not. Apparently, in the dark, plastoquinone is normally not in redox equilibrium with the stroma, but heat stress opens a path for electrons from the stroma to plastoquinone. This results in reduction of plastoquinone in the dark and stimulation of cyclic electron flow in the light.
The increase in cyclic electron flow could require movement of cyt b6/f complexes from appressed regions of the thylakoid to non-appressed regions (Joliot & Joliot 2002). In addition, recent crystallography studies indicate that cyt b6/f complexes normally exist as dimers with a lipid-rich cavity between the two monomers (Kurisu et al. 2003; Cramer et al. 2004). This cavity is the site of electron arrival from plastoquinone. The nature of the lipid in the cavity, and the effects of heat on the overall structure is not known. It has been suggested that the cytochrome complex can exist in different states (Schreiber, Heimann, & Klughammer 1998), but the effect of heat on the cyt b6 /f complex location or structure has not been examined.
During a 39 °C heat pulse, a stimulation of PSI with little concurrent effect on PSII was observed (Schrader et al. 2004) confirming that cyclic electron flow is dramatically accelerated during heat stress. We also found a significant decline in stromal oxidation status in moderately high temperature (Schrader et al. 2004) consistent with a heat-stress-induced pathway for electrons to flow from the stroma to plastoquinone. Lowered stromal redox status can lead to significant stimulation of PSI-mediated cyclic electron flow (Joët et al. 2002), which raises the question – is cyclic electron flow stimulated by high temperature directly or does stromal oxidation cause the stimulation of cyclic electron flow?
The involvement of protein phosphorylation with cyclic electron flow and heat stress
One of the mechanisms involved in stimulating cyclic electron flow is phosphorylation of light-harvesting chlorophyll complex of PSII (LHCII). Phosphorylated LHCII moves from the appressed thylakoid regions, where PSII is located, to the unappressed regions, where PSI is located (Chow, Miller, & Anderson 1991). The phosphorylated LHCII becomes energetically disconnected from PSII core complex (slowing its turnover rate) and energetically coupled to a PSI core (increasing its turnover rate). This process requires a specific polypeptide within PSI (Lunde et al. 2000). This is the well-known state transition (Allen 1992) and is known to accompany an increase in cyclic electron flow around PSI, including in response to heat (Wise et al. 2004). However, heat has been reported to stimulate dephosphorylation of a number of PSII core proteins namely D1, D2 and CP43 (Rokka et al. 2000; Vener et al. 2001). Therefore, the regulation of phosphorylation of LHCII may well be quite different from the regulation of phosphorylation of the PSII core proteins (Harrison & Allen 1991; Pursiheimo et al. 2003). The regulation of phosphorylation of thylakoid proteins interacts with redox status (Vener et al. 1995). Likewise, dephosphorylation of LHCII appears to be catalysed by a different phosphatase than dephosphorylation of other thylakoid-associated proteins (Hammer, Markwell, & Sarath 1997; Vener et al. 1999). There are several thylakoid associated kinases (TAKs) (Snyders & Kohorn 1999, 2001) but also another kinase that appears unrelated to TAKs that is necessary for state transitions and phosphorylation of LHCII (Depège, Bellafiore, & Rochaix 2003).
Thus, there may be two different regulatory systems that control thylakoid protein phosphorylation/dephosphorylation; one that controls phosphorylation/dephosphorylation of LHCII and state transitions and a second system that controls phosphorylation/dephosphorylation of other thylakoid proteins. A large number of thylakoid proteins undergo reversible phosphorylation (Hansson & Vener 2003) and heat stress is one of the most effective ways of modulating the phosphorylation status of many of them (Vener et al. 2001). For some thylakoid proteins whose phosphorylation status varies, the function of the protein is not known (Carlberg et al. 2003; Hansson & Vener 2003).
In summary, the physical state and structure of the thylakoid membrane could be critically important for shifting between linear and cyclic electron flow. It is possible that structures found at high temperature (e.g. unstacked thylakoids) optimize its function at high temperature. However, it may be that the structural rearrangements that allow photosynthesis to tolerate moderate heat stress leads to damage when the leaves cool back to non-stressful temperatures. Repeated episodes of moderately high temperature may result in very different types of damage than occurs during the first heating episode.
In my opinion there is no doubt that rubisco deactivates at moderately high temperature and three hypotheses for this behaviour of plants can be put forward.
1Rubisco activase is inherently heat sensitive and cannot keep up with the demands on its activity at high temperature.
2Natural selection has favoured plants in which rubisco deactivates at moderately high temperature because of the unfavourable ratio of photorespiration to photosynthesis at these temperatures. If the cost of metabolizing phosphoglycolate is high and the carbon gain is low at moderately high temperature it could be advantageous to deactivate rubisco to avoid the generation of phosphoglycolate and forgo the carbon that could be fixed.
3Natural selection has favoured plants in which rubisco deactivates at moderately high temperature because this prevents more severe damage, for example to thylakoid structure or function, that might otherwise occur in the absence of deactivation.
The first hypothesis is the easiest to understand. If the observation is that rubisco deactivates, and in some cases rubisco activase becomes less soluble as a result of moderately high temperature (Feller et al. 1998; Haldimann & Feller 2005), it is logical to assume that rubisco deactivation is the lesion caused by heat stress.
The second hypothesis is best illustrated with a graph. Figure 1 shows how net CO2 assimilation declines with temperature according to models of photosynthesis. This decline comes about because the ratio of photorespiration to photosynthesis increases with temperature. The rate of oxygenation, the first step of photorespiration, is shown in Fig. 1 as circles. In today's atmosphere net CO2 assimilation is high, even at 40 °C. However, if the model is run for 190 p.p.m. CO2 (open symbols), as atmospheric modellers say was probably the CO2 level for many thousands of years, then CO2 assimilation is much reduced and the velocity of oxygenation is much higher. The cost of photorespiratory metabolism relative to the benefit in terms of CO2 assimilation may be unfavourable to plants when the temperature is above 35 °C, or perhaps this was the case when the CO2 concentration was 190 p.p.m.
Support for the third hypothesis comes from studies of tobacco plants in which rubisco activase was reduced by antisense technology leading to a constant 20% activation state of rubisco and a lack of response of activation state to temperature. Photosynthesis of these plants was just as sensitive to temperature as was photosynthesis of plants with functional activase in which rubisco deactivated at high temperature. However, the heat-induced reduction of photosynthesis in plants unable to deactivate rubisco was irreversible while that in plants able to deactivate rubisco was reversible (Sharkey et al. 2001a). Some of the dangers at high temperature which could be avoided by deactivation of rubisco include build-up of products of photorespiration and H2O2. Kim & Portis (2004) reported that H2O2 is made as a side reaction of oxygenation of RuBP and that this side reaction occurs more frequently at higher temperature, both because it is inherently temperature sensitive and because oxygenation goes faster at high temperature. Another danger at high temperature could be deleterious effects of changes in thylakoid structure and function as described above.
If either hypothesis 1 or 2 are correct, then engineering rubisco to stay more active at high temperature should improve thermotolerance. However, if hypothesis 3 is correct, engineering a rubisco that does not deactivate could result in a plant that is less able to tolerate heat. My opinion is influenced by the fact that rubisco deactivation is regulatory in several other circumstances. These include low light (Perchorowicz, Raynes, & Jensen 1981) and when starch and sucrose synthesis limit photosynthesis (feedback-limited photosynthesis, as can be induced in low oxygen) (Sharkey 1985, 1989). Rubisco deactivation in low light or under feedback conditions appears to be regulatory; is the same true for deactivation at moderately high temperature?
THERMOTOLERANCE INDUCED BY ISOPRENE
An important reason for studying how moderate heat reduces the photosynthetic capacity of leaves is to determine how photosynthesis can be made more tolerant of heat. Studies of isoprene emission resulted in the hypothesis that isoprene inside leaves can improve thermotolerance of photosynthesis (Sharkey & Singsaas 1995). The ultimate proof of isoprene providing thermotolerance requires knowing the mechanism. However, given the uncertainty of how heat damages photosynthesis, it is difficult to hypothesize how isoprene protects against it. On the other hand, if isoprene provides substantial thermotolerance, then the damage caused by heat must have some relationship to things that isoprene can influence.
Direct tests of the thermotolerance hypothesis became possible when it was discovered that fosmidomycin, an inhibitor of the isoprene synthesis pathway (Zeidler et al. 1998) was highly specific and did not inhibit photosynthesis (Sharkey, Chen, & Yeh 2001b). Following three heat spikes (2 min at 42 °C), photosynthesis at 30 °C was inhibited by one-third when leaves were able to make isoprene but by two-thirds when isoprene production was inhibited. Exogenous isoprene given to fosmidomycin-fed leaves could restore thermotolerance (Sharkey et al. 2001b).
Peñulas et al. report in this issue that isoprene fumigation of Quercus ilex, which normally does not emit isoprene, resulted in substantial thermotolerance between 35 and 45 °C. Photosynthesis at 50 °C was depressed to the same degree in both treatments (Peñuelas et al. 2005). Photosystem II chlorophyll fluorescence was not affected by heat below 50 °C. Isoprene fumigation interacted with some antioxidant enzymes and suppressed the endogenous monoterpene emission. Also in this issue, Velikova and Loreto report that exposing leaves of Phragmites australis to 38 °C for 90 min caused a greater reduction in photosynthesis when isoprene synthesis was inhibited by fosmidomycin, than when it was not (Velikova & Loreto 2005). In addition, photosynthesis recovered more when the leaf temperature was returned to the pre-stress level of 30 °C when isoprene was present than when it was inhibited. The effect was substantial, with about twice as much photosynthetic capacity found in leaves making isoprene than in leaves not making isoprene. Thermotolerance is much easier to see during recovery from heat stress than during the initial heat stress episode (Sharkey et al. 2001b; Velikova & Loreto 2005). Heat damage to photosynthetic electron flow was apparent after heat stress but not during an initial heat stress episode (Schrader et al. 2004), and so in this way, thermotolerance induced by isoprene appears similar to heat effects on thylakoid reactions. Specifically, much of the protection occurs during recovery of leaf temperature to non-stressful levels rather than during the heat stress.
It is my opinion that the thermotolerance hypothesis for isoprene function now has substantial support, given three independent reports using different species and methods (Sharkey et al. 2001b; Peñuelas et al. 2005; Velikova & Loreto 2005). The effect is strongest between 25 and 45 °C and is not related to photosystem II damage. At this time, the effect of isoprene on heat-induced dark reduction of plastoquinone has not been tested. Isoprene has been shown to protect against reactive oxygen species (Loreto et al. 2001; Loreto & Velikova 2001; Affek & Yakir 2002). Both Peñuelas et al. and Velikova and Loreto report in this issue isoprene protection against both high temperature and reactive oxygen species, or increased levels of antioxidant enzymes. However, in the experiments of Velokova and Loreto, reactive oxygen species were reduced by isoprene at all temperatures and the increase in reactive oxygen species with an increase in temperature was similar whether or not isoprene was present. It is not clear whether the antioxidant activity of isoprene is parallel with the effect on thermotolerance or indicates how temperature in the range of 35–45 °C inhibits photosynthesis. Knowing the mode of action of isoprene requires knowing the mode of action of moderate heat on photosynthesis. Isoprene induced thermotolerance is very fast (Singsaas & Sharkey 1998; Singsaas et al. 1999) compared with changes in xanthophylls (Havaux et al. 1996) or fatty acid composition of thylakoid membranes (Murakami et al. 2000).
What of an alternative explanation for isoprene emission, that it is a way of freeing phosphate that has inadvertently become stuck on DMAPP (Logan, Monson, & Potosnak 2000)? The need for isoprene production to release phosphate from DMAPP depends on there being no mechanism for regulating the early steps of DMAPP synthesis. If the MEP pathway has significant feedback control so that DMAPP levels can be regulated, then a phosphate release mechanism is not needed and is a futile cycle. If there is little feedback control, then the amount of DMAPP could increase and become a significant sink for phosphate, and isoprene production could release this phosphate for recycling back into ATP. Wolfertz et al. (2004) tested the regulation of the early steps of the MEP pathway by feeding deuterium-labelled deoxyxylulose (DOX) to leaves. This compound is taken up by leaves and quickly converted to deoxyxylulose 5-phosphate, and then isoprene. The isoprene coming from the deuterated DOX is deuterated and can be distinguished from undeuterated isoprene using laser photo-acoustics (Kühnemann et al. 2002). There was essentially a one-for-one reduction in unlabelled isoprene as labelled isoprene started to be detected during feeding of deuterated DOX. In other words, exogenous sources of carbon fed into the MEP pathway past deoxyxylulose-5-phosphate synthase caused the synthase activity to be reduced so that the flux through the pathway (total isoprene emission) stayed remarkably constant (Wolfertz et al. 2004). This indicates that there is no need to invoke a futile cycle to get rid of DMAPP inadvertently made by the MEP pathway, the pathway is sufficiently regulated to prevent this problem.
Substantial progress has been made recently in understanding how photosynthesis is affected by heat. There are substantial effects of moderate heat on photosystem I and cytochrome complex reactions. There is no effect of these moderately stressful heat treatments on photosystem II. In addition, rubisco normally deactivates at these moderately stressful temperatures. The rubisco deactivation can occur in parallel with the effects on thylakoid reactions or can mask the damage to thylakoid reactions, especially during the first exposure of a leaf to heat stress. Because many of the things that improve thermotolerance are known or presumed to work on thylakoid function, my opinion is that the rubisco deactivation is a regulatory response and is not the underlying problem that needs to be fixed to make plants more thermotolerant. Three areas remain as important areas of research: (1) what is the specific nature of the effects of moderate heat stress on thylakoid reactions other than PSII? (2) what is the role of rubisco deactivation, is it regulatory or simply a deleterious response to elevated temperature? and (3) what are the roles of reactive oxygen species at high temperature?
Research in my laboratory on heat stress effects on photosynthesis is supported by the National Research Initiative, CSREES, USDA grant 2004-35100-14860. Research on isoprene is supported by the National Science Foundation grant IBN-0212204. I thank Mike Salvucci for comments on the manuscript. I first heard of the idea of rubisco deactivation protecting against high rates of photorespiration at high temperature from T. John Andrews.