Modeling the protection of photosynthesis

It is hard to overstate the importance of photosynthesis for mankind and the biosphere. It produces the oxygen we breathe and the food we eat, and images of Earth from space show the green of terrestrial vegetation and swirls of marine phytoplankton. To meet our increasing demand for food and energy, it seems inevitable that we will need to increase the efficiency of photosynthesis in plants and algae. There is therefore some urgency in our drive to better understand the operation, regulation, and limitations of photosynthesis. This ambition is made particularly challenging because of the complexity of photosynthesis; it comprises many significant subprocesses that range in scale from quantum mechanics to ecosystems. Given the complexity of photosynthesis, mathematical models have proven to be a vital tool with which to encapsulate knowledge and to describe, analyze, and simulate the operation of photosynthesis in vivo (1). In PNAS, Zaks et al. (2) describe a comprehensive mathematical model for qE, a mechanism with a somewhat odd name that is essential for protecting a component of photosynthesis, photosystem II (PSII), from photodamage. In vivo, qE is a dynamic, actively controlled process whose regulation depends on the combined effects of photosynthetic electron and proton transport, and photosynthetic metabolism. To model qE, therefore, Zaks et al. needed to produce an impressive toolkit of models that will be useful for modeling much more than just qE.

Photosynthesis uses the energy of absorbed photons to drive the otherwise endothermic reduction of CO₂, and the importance of qE arises because leaves are often unable to use for CO₂ fixation all the light absorbed by their photosynthetic pigments. Ideally, the light absorbed by the leaf would be used to fix carbon dioxide with a constant quantum efficiency across all natural light intensities, with the efficiency being determined by the chemistry of the fixation process. This would result in a linear relationship between light intensity and the rate of CO₂ fixation (Fig. 1, curve A). If real-world photosynthesis behaved like this, photosynthesis would need protection from only those minor, damaging side reactions that occur alongside the physiological chemistry of photosynthesis. Unfortunately, very few leaves even come close to this ideal (3), with most showing a light response for CO₂ fixation that is distinctly nonlinear (Fig. 1, curve B), with the degree of nonlinearity depending on the species and its growth environment. The underlying chemistry of CO₂ fixation does not change as the light–response curve becomes nonlinear so more light is being absorbed than can be used for CO₂ fixation, creating the problem of excess light. The limited capacity for CO₂ fixation limits photosynthetic electron transport, which then restricts the functioning of the reaction centers of photosystem I (PSI) and PSII. In the case of PSI, this causes few problems, but, for PSII, the result is side reactions, resulting in harmful singlet oxygen production (4, 5) and damage to the reaction center (6, 7) and membranes (4). As the light intensity increases, the limitations on center function increase (8) and the rate of the damage increases (6). To moderate the extent of damage, a range of mechanisms are found in PSII that dissipate excess excitation energy (9, 10). These are collectively known as nonphotochemcial quenching (NPQ), and they are regulated so they are active under conditions of excess light and inactive when light intensity limits the rate of photosynthesis. The most important form of NPQ in unstressed leaves is qE (11). It effectively dissipates excess excitation in PSII (12) and is highly regulated to maintain the optimal efficiency of PSII and photosynthesis (13). Although the mechanism of qE is still intensely debated—several models have been proposed (14–16)—the regulation and effects of qE are well understood (11). The model of Zaks et al. (2) encapsulates this knowledge, which is a notable achievement because although the control of qE is easy to describe in a cartoon (Fig. 2), the various processes that contribute to this control are individually complex to model.

Fig. 1.

The light responses of CO₂ fixation. Curve A shows an idealized light-response curve that would be obtained from a leaf with unlimited photosynthetic capacity. The slope of curve A, which is the quantum yield for CO₂ fixation, depends on the physics and chemistry of the photosynthetic process and involves factors such as light absorption of by the leaf, the quantum efficiency of electron transport, the ATP/electron transport ratio, and the ATP and NADPH requirements of the CO₂ fixation reactions. Curve B shows a realistic light response curve. In this case, photosynthetic capacity is limited and has a maximum, light-saturated rate. Except at low irradiances, the light-use efficiency of CO₂ fixation in this leaf is lower than that of the idealized leaf. A CO₂ fixation rate of J₁, produced by an irradiance I₂, uses only the light supplied by an irradiance of I₁, so the difference (i.e., I₂ − I₁) is in excess of the needs of CO₂ fixation.

Fig. 2.

The thylakoid processes simulated by the model of Zaks et al. (2). The linear electron flow (blue arrows) passes through PSII, the cytochrome b6/f complex, and PSI, resulting in the transfer of electrons from water to ferredoxin (a strong reducing agent). This electron transport process is coupled with a proton transport flux (brown arrows) that results in the deposition of protons in the thylakoid lumen, making it positively charged and more acidic than the surrounding stroma. Proton transport is paralleled by a flux of counterions (Cl⁻, Mg²⁺, and K⁺) across the thylakoid membrane. The resulting transthylakoid voltage and pH differences generates a transthylakoid proton potential that drives ATP synthesis as protons pass from the lumen to the stroma via the ATPase. The pH of the lumen activates violaxanthin de-epoxidase (VDE) and protonates subunit S of PSII (psbS), stimulating the formation of qE. Lumen pH also acts to restrict linear electron transport.

The pH of the thylakoid lumen plays a pivotal role in photosynthetic physiology and is a focus of the model (2). Protons are pumped into the lumen by the action of electron transport, forming a proton motive force (Δμ_H+) across the thylakoid membrane (Fig. 2). This Δμ_H+ drives ATP synthesis as protons exit the lumen via the ATPase. Lumen pH is therefore determined by the balance between a supply process, electron transport, and a demand process, ATP synthesis. It also plays a regulatory role: decreased lumen pH will limit electron transport by slowing the activity of the cytochrome b6/f complex (8, 13), and it will induce qE, with the protonation of subunit S of PSII and activation of the enzyme violaxanthin de-epoxidase being important for the full development of qE (11). The significant consequence of qE, and the phenomenon that Zaks et al. seek to describe (2), is the regulated dissipation (as heat) of excited states of chlorophyll a in PSII. These dissipated (or quenched) excited states are no longer available to cause damaging side reactions, but a consequence of their dissipation is a reduction in the intrinsic photosynthetic efficiency of PSII, which is why qE needs to be carefully regulated. If it is too active, it will restrict photosynthesis; if it is too inactive, photosynthesis will be damaged. Each of the processes that are involved in the development of qE—the formation and dissipation of lumen pH, the formation of qE in response to lumen pH, and the impact qE on the operation of PSII—are included in the model (2). Despite the complexity of the physiology, the results of tests of the model are convincing and impressive. The model is also Matlab-compatible, which should make it relatively easy to modify, extend, and apply.

A model is a tool, and predicting how any tool will be used and developed is a risky affair, but I expect this model will make a significant impact on plant modeling and physiology. Some elements of the model of Zaks et al. (2) have been previously described, but the level of integration achieved by them is exceptional. If there are weaknesses in the model in its current form, they lie at its limits. At one extreme, the qE process is not explicitly modeled—only its effects—but, given the lack of any consensus about the mechanism of qE, this is not surprising. The model also only deals with the qE component of NPQ. Although this is the most significant form of NPQ in unstressed plants, other more slowly reversible forms exist (11, 17) and will ultimately need to be accounted for. I expect that the model will be extended over the coming years to include these. At the other extreme, the authors deliberately exclude from the model modulation of ATPase activity in response to metabolic changes (2). Comprehensive metabolic models of photosynthetic metabolism exist, and it should be possible to combine them with the model of Zaks et al. (2). The addition of a metabolic model would also make it possible to include other electron transport fluxes, such as the Mehler reaction (18) and cyclic electron transport (19), that appear to be subject to metabolic control. A more complete photosynthesis model of this kind would make possible the development of a more detailed and mechanistic whole-plant or crop-canopy productivity model by linking together models for (at least) radiation input, gaseous diffusion, photosynthesis, and plant growth and structure. This kind of model could be used to explore options for improving the productivity of crop plants (20). More directly, the ability to model qE will make it possible to more quantitatively analyze qE in vivo and in vitro, a process that will almost certainly drive improvements in the model as well as revealing just how complete our understanding of qE regulation is. qE is also finely controlled, and the basis of this control in relation to the parameters of regulation could be easily explored in silico by using the model. It has been estimated that crop yields could be improved by 15% if changes in qE could be accelerated (20), and exploring in silico how qE regulation could be improved in terms of its extent and dynamics would be an obvious and useful application of the model.