Perspectives on improving photosynthesis to increase crop yield

1While photosynthesis is fundamental to life on Earth, powering the growth of plants and sustaining food chains, its solar energy conversion efficiency is surprisingly low, typically below 1% in crops. This low efficiency represents a significant opportunity for improvement, which is especially relevant in the context of a growing global population and pressure on available arable land to climate changes, with the consequent increase in agricultural demand. While substantial progress has been made in enhancing other aspects of crop yield, photosynthesis itself remains a largely untapped area for improvement. This is primarily due to the complexity of the photosynthetic process, which involves a multitude of genes and biochemical pathways. Unlike traits such as disease resistance or plant height, which can often be traced to a few key genetic changes and can be targeted by traditional breeding, the optimization of photosynthesis demands a deep molecular understanding and precise genetic manipulation. Only recently are the tools and knowledge for such interventions becoming available, making the direct improvement in photosynthesis a relatively new frontier in increasing crop productivity.

2Can photosynthesis be improved? Recent proof-of-principle experiments have provided compelling evidence that it can. These studies have explored various innovative approaches, from introducing new biochemical pathways into plant systems to genetic modifications aimed at enhancing light-harvesting or carbon fixation efficiency. For instance, experiments have successfully demonstrated the feasibility of engineering plants to utilize light or assimilate carbon more efficiently or to bypass photorespiratory pathways, thereby boosting overall photosynthetic efficiency (e.g.;; ). These ground-breaking experiments show the potential to enhance photosynthesis and open new avenues for increasing crop productivity.

3Why, after billions of years of evolution, are there still so many opportunities to enhance photosynthesis? Beyond the common understanding that plants evolve mainly for reproduction and survival rather than maximizing photosynthesis and productivity, there are additional factors. Climate changes, particularly changes in atmospheric carbon dioxide (CO2) levels, have made the photosynthetic properties of current plants less suited to modern environments. Enhancing ribulose 1,5-bisphosphate (RuBP) regeneration has been shown to increase photosynthesis in C3 plants, suggesting that the process may not have been fully optimized in response to new environmental conditions. Moreover, inefficiencies in some photosynthetic enzymes and structures may be evolutionary legacies. For example, Rubisco evolved when oxygen (O2) was scarce, making differentiation between CO2 and O2 less critical. As O2 became prevalent, increasing the CO2 specificity of Rubisco meant reducing its catalytic rate, potentially trapping it in a state of low specificity and efficiency. These factors suggest that contemporary photosynthesis systems, shaped by environmental changes and evolutionary history, hold potential for optimization and efficiency gains.

4This perspective explores a range of promising strategies for improving photosynthetic efficiency. The strategies described here are at various stages of development; some have already demonstrated proof of principle, while others are still in the conceptual phase, collectively representing the state of the art in this field.

5Light is the energy source of photosynthesis. However, only visible photons in the 400 to 700 nm range, the so-called range of photosynthetic active radiation, are used to power photosynthesis in most organisms, including plants. This limits the use of solar energy to <50% of what reaches the Earth's surface. For a long time, it has been believed that photons above 700 nm would not carry enough energy to support efficient water oxidation. However, the discovery of cyanobacteria containing pigments absorbing in the far-red (FR) spectral region has shown the feasibility of this process. It has thus been proposed that expanding the spectrum of plants up to 750 nm would lead to a gain of light absorption of around 20%, which is a considerable increase in the energy available for growth. Making use of FR photons would be highly relevant for crops, as plants in the field are close together, and the light reaching the lower leaves is almost exclusively FR (Fig. 1) and currently cannot be used for photosynthesis, resulting in a close-to-zero photosynthetic rate at the bottom of a crop canopy. The idea of broadening the absorption of crops to the FR is thus considered a promising strategy to improve their productivity.

6How can this be achieved? Although this strategy has not yet been implemented in plants, its success in cyanobacteria serves as an excellent starting point for adapting it to crops. Three main factors are crucial for the efficient use of FR light in the first steps of photosynthesis: (i) the organism's capacity to harvest FR light, (ii) efficient transport of this energy to the photochemical reaction center (RC), and (iii) utilization of this energy for charge separation and stabilization of the charge-separated state. See for a recent review.

7Plants and algae use chlorophyll a (Chl a) and Chl b and carotenoids to harvest light. These pigments strongly absorb blue and red light but do not absorb in the FR. Some cyanobacteria, however, can synthesize two additional types of chlorophyll, called d and f, which absorb in the FR region of the spectrum. Both these chlorophylls differ from Chl a due to the presence of a formyl group in one of the substituents of the tetrapyrrole ring. This also means that a single enzyme is sufficient to transform the red-absorbing Chl a into the FR absorbing Chl d or f. While the enzyme for Chl d synthesis has not yet been discovered, the one for Chl f is known and has been successfully heterologously expressed in cyanobacteria strains incapable of FR acclimation, showing that it leads to the production of Chl f. These pigments then need to be coordinated to the photosynthetic proteins to form functional pigment–protein complexes. In FR light, alongside the Chl f synthase, cyanobacteria express paralogs of some/most photosynthetic proteins involved in pigment binding. These novel proteins have a higher affinity for Chl f in specific sites. However, it has already been shown that even the canonical photosynthetic proteins, such as the components of photosystem I (PSI), can bind Chl f and that the light-harvesting complexes (LHCs) of plants, which are totally different from those of cyanobacteria, can accommodate both Chl d and Chl f.

8In addition to the use of Chl d and f, FR light can also be absorbed by red-shifted Chls a. These are present in a small amount in PSI, also of plants, where they are called “red-forms”. The FR absorption spectrum of these pigments is due to the strong coupling between two or more Chls a (see for details) which is tuned by the proteins that determine the organization of the pigments. Designing FR Chl a binding sites can also be a strategy to enhance the FR absorption of plants without the need for additional Chl types.

9After light is absorbed, the energy needs to be efficiently delivered to the RC for charge separation (see for more details). This process is made possible by the presence of proteins that organize the pigments. The plant photosynthetic units, PSI and photosystem II (PSII), are modular assemblies where a set of pigment–protein complexes act as an antenna responsible for harvesting light and transferring the excitation energy to the RC. The efficiency of this process hinges on the rate of energy transfer, as faster transfer reduces energy losses. The transfer rate depends on the inter-pigment distance, their mutual orientation, the magnitude of their excited-state transition dipole moments, and their excited-state energy landscapes. With Chl d and f having larger dipole moments than Chl a, and with the organization of the pigments within the complexes remaining largely the same in the studied cyanobacteria complexes, the key factor is the excited-state landscape. The substitution of all Chls a with FR absorbing Chls, as in the case of the cyanobacterium Acaryochloris marina, does not affect the rate of excitation energy transfer. On the contrary, integrating a few FR chlorophylls into complexes mainly containing Chl a can create local energy traps, slowing down the excitation energy transfer rate and reducing efficiency, as observed in cyanobacteria. Careful planning of the number and position of low-energy pigments is therefore crucial to ensure efficient delivery of the absorbed energy to the RC. This would also require the design of modified complexes where the affinity of Chl d and f is enhanced in specific binding sites.

10The presence of low-energy chlorophylls in the RC can influence charge separation efficiency, as these Chls need to perform the same function as Chl a, but with less excited-state energy. This can affect the recombination reactions and, in principle, even increase the possibility of photodamage. The regulation of the midpoint potential of the cofactors in the electron transport chain might be an issue. This aspect needs to be carefully studied in cyanobacteria to understand how they have overcome possible problems. However, while these factors might be critical for PSII, experimental results indicate that Chl f-containing PSI still works very efficiently with Chl a in the RC, suggesting that shifting PSI absorption to the FR might be easier than in the case of PSII.

11Finally, the idea is to implement FR absorption in crops using the smart canopy concept, e.g. gradually increasing the absorption of FR photons in the lower part of the canopy. This allows for better use of light, as the upper leaves in a canopy, typically exposed to full sunlight, would have no advantage in absorbing FR light. In fact, having low-energy Chls in upper canopy leaves could decrease the overall efficiency of charge separation. On the contrary, the absence of FR pigments in the upper leaves allows the penetration of FR light to the bottom of the canopy, where leaves are light-limited and where it will be absorbed and used for photosynthesis. To create this absorption gradient across the canopy, we can again take inspiration from cyanobacteria. More and more species are being discovered that can undergo FR acclimation, meaning that these strains have a normal, Chl a-based, photosynthesis apparatus when exposed to white light and only express the Chl f synthase and paralogues of the photosynthetic proteins with enhanced ability to bind Chl f upon exposure to FR light. This process is controlled by phytochrome, a photoreceptor that senses the red/FR ratio and triggers the expression of FR-related subunits when needed. The same control should then be implemented in plants.

12In summary, while the utilization of FR light in plant photosynthesis remains to be achieved, the expanding understanding of FR photosynthesis in cyanobacteria is paving the way for this application.

13In C3 plant canopies, the photosynthetic machinery saturates at ∼25% of maximum solar flux, and this is one of the main factors that constrain productivity in these species. Plants probably overinvest in light capture and synthesize high levels of thylakoid antenna proteins and associated chlorophylls because this increases their own fitness by depriving neighboring plants of light and nutrients. However, since this trait is found in elite crops, it constitutes a major drawback for monocultures, in which all plants are expected to have the same fitness and yield capacity in order to maximize productivity in cultivated fields.

14Reduction of chlorophyll content in leaves could offer a means to mitigate competition for light in monocrop stands, i.e. fields where one crop species is grown at a time, since the level and quality of the light that reaches leaves in the lower canopy should be higher and more uniform, thus boosting overall photosynthetic performance and yields. Indeed, this strategy has been validated in several studies on cyanobacteria and microalgae, although the same approach in higher plants has produced discordant results. A decrease in antenna size in tobacco (Nicotiana tabacum), for instance, led to an increase of about 25% in above-ground biomass accumulation under high-density cultivation conditions. Similarly, beneficial effects were observed in a rice genotype with pale green leaves cultivated under high-light conditions. Conversely, the few field studies that have used chlorophyll-deficient soybean (Glycine max) mutants showed a marked decrease in leaf mass accumulation and grain yield. These results underline the complexity of this trait, which is influenced by several factors including the gene/pathway involved, the degree of leaf chlorophyll reduction, the architecture of the crop canopy, as well as the environmental and growth conditions.

15Recently, we have used the chemically induced mutant happy under the sun 1 (hus1) to assess the impact of leaf chlorophyll reduction on biomass accumulation and grain yield in barley. The pale green phenotype of hus1 is due to a 50% reduction in the chlorophyll content of leaves, which enhances the efficiency of the conversion of light energy into chemical energy without increasing its susceptibility to photoinhibition (Fig. 2). The mutation introduces a premature stop codon in HvcpSRP43, which encodes the 43-kDa chloroplast signal recognition particle (cpSRP43) responsible for the delivery of antenna proteins to the thylakoid membranes, and its truncation resulted in a decrease in the size of photosystem antenna complexes. Furthermore, the HvcpSRP43 protein has been shown to efficiently chaperone and stabilize glutamyl-tRNA reductase, a rate-limiting enzyme in tetrapyrrole biosynthesis, which enables the insertion of LHCs into thylakoids to be coordinated with chlorophyll biosynthesis. The dual role of HvcpSRP43 in chloroplasts makes hus1 a rather specific mutant, since the reduced antenna size is coupled with a marked reduction in the activity of the tetrapyrrole biosynthetic pathway, which in turn suppresses the formation of highly toxic tetrapyrrole intermediates and pleiotropic photo-oxidative damage. Intriguingly, the hus1 mutant accumulated biomass and grains at levels comparable to those observed for the control cultivar Sebastian, when grown under field conditions at standard density. These findings demonstrate that, when the selective pressure imposed by competition for resources in cultivated fields is attenuated, crops can indeed decrease their investment in antenna proteins and chlorophyll biosynthesis significantly, without detrimental effects on productivity.

16Although cpsrp43 knockout mutants showing the characteristic pale green phenotype have also been reported in other species, including rice, the hus1 mutant was chosen for our investigations because a large set of functional genomics tools are available in barley, making it an optimal model crop in which to test the performance of the several strategies reported to improve photosynthesis efficiency under realistic field conditions. For instance, the Fast Identification of Nucleotide variants by droplet DigITal PCR technology developed by the Carlsberg Research Laboratory provides a novel strategy for the rapid identification and isolation of targeted genetic variants in the elite spring barley cultivar RGT Planet. Thanks to this method, mutations that confer advantageous phenotypes on old varieties with low commercial value that carry several other single-nucleotide polymorphisms (SNPs) introduced by chemical mutagenesis—as in the case of the hus1 mutant isolated within the HorTILLUS population—can be transferred to commercially competitive varieties without the need for multiple, time-consuming (several years) backcrosses. Such an approach will make it possible to translate the major gains in biomass and grain yield reported in the last two decades as a consequence of the manipulation of photosystem antenna size in model species and crops, with tests mostly conducted in greenhouses or in small-scale field trials, into yield increases on farms. To this end, collaboration with plant breeders, agronomists, and crop physiologists is needed to select the most appropriate yield-testing protocols, including plot designs that avoid edge effects (which can distort yield estimates), definition of growing plant densities, and standard parameters to define yields. Such careful field trials will also allow us to test other major advantages thought to be associated with the pale green phenotype. Independent studies have predicted, for instance, that reductions in chlorophyll content should increase the efficiency of nitrogen use. Similarly, simulations in soybean predict savings of up to 9% of leaf nitrogen upon a 50% reduction in leaf chlorophyll content. Furthermore, the development of pale green crops and the consequent increase in the fraction of reflected light, i.e. increased albedo, have been shown to mitigate the effects of heat waves triggered by global climate change and improve the efficiency of water use by reducing canopy temperature. This latter aspect is supported by the finding that certain Syrian barley landraces and a few accessions of wild barley (Hordeum vulgare spp. spontaneum) in Israel, which are adapted to stable and very dry environments, are characterized by pale green leaves.

17Overall, the time has come to translate photosynthesis research into the field, using barley as a model crop that can also exploit the availability of large collections of natural genetic diversity. Numerous publicly funded organizations, such as the “Genomes to Fields Initiative” and the Consultative Group on International Agricultural Research (CGIAR), are conducting field trials that could contribute to achieving this goal with the medium-term objective of transferring this knowledge into other cereals, including wheat, given the high degree of conservation of the photosynthetic machinery in higher plants.

18Photosynthesis needs photoprotection. Excess light can increase the lifetime of the singlet excited state of chlorophyll (1Chl*), resulting in higher yields of the longer-lived (∼ms) triplet excited state of chlorophyll (3Chl*) and photoinhibitory singlet oxygen (1O2*). 3Chl* and 1O2* can be quenched by carotenoids, but a ubiquitous first line of defense is the photosystem-scale de-excitation of 1Chl* and dissipation of excess absorbed light energy as heat—a suite of mechanisms that are measured (and referred to) as nonphotochemical quenching (NPQ) of Chl fluorescence.

19NPQ comprises several components defined by their relaxation kinetics and molecular players. In plants, the rapidly induced, energy-dependent quenching (qE) requires the PSII subunit S (PsbS), the xanthophyll zeaxanthin (Zea), and a trans-thylakoid pH gradient (ΔpH), regulating quenching that is induced and relaxed on a seconds-to-minutes timescale. A slower type of Zea-dependent but ΔpH-independent quenching (qZ) operates on the timescale of several minutes. Zea is produced by violaxanthin de-epoxidase (VDE) and removed by zeaxanthin epoxidase (ZEP) in a xanthophyll cycle (Fig. 3A). The slowest components of NPQ include photoprotective qH and photoinhibitory qI, which relax on a timescale of hours. In concert, these highly conserved NPQ mechanisms provide the flexibility to cope with light fluctuations at varying intensities and timescales to sustain plant fitness.

20Although NPQ protects the photosynthetic apparatus from excess light, it can also compete with and limit photosynthetic efficiency, especially in fluctuating light, where modeling has shown that slow relaxation of NPQ can limit CO2 assimilation by up to 30%. Light intensity fluctuates by time of day, by local shading, and, most dramatically, by the movement of sunflecks in dense crop canopies. Optimizing the light-dependent reactions of photosynthesis under dynamic, field-relevant light conditions is a rapidly expanding field with significant agronomic potential.

21A transgenic approach to accelerate the relaxation of NPQ by overexpression of VDE, PsbS, and ZEP (hereafter VPZ, Fig. 3A) increased N. tabacum biomass by ∼15% and elite soybean seed yield by ∼20% in small-scale field trials. Although large-scale, multi-location field trials are still needed, these results point to NPQ as a novel potential target for crop improvement. However, the VPZ approach did not increase biomass in Arabidopsis (Arabidopsis thaliana; ) or yield in potato (Solanum tuberosum; ) under greenhouse and simulated fluctuating light conditions.

22Here, we consider possible explanations for the varying VPZ results in different plants. Modeling suggests that the stoichiometries of the three proteins are critical, and thus, it may be necessary to optimize VPZ construct expression for each species depending on native qE and qZ capacities. Plant species with slower NPQ relaxation time constants (τ) and cultivars with higher yield potential may benefit the most from VPZ engineering efforts. Beyond species-specific factors, the literature to date suggests several key phenotypes that may be necessary to achieve greater photosynthetic efficiency by optimizing NPQ.

23In all published VPZ studies (tobacco, soybean, Arabidopsis, and potato), transformants exhibited faster NPQ relaxation than wild type (WT), when normalized to high-light-acclimated NPQ. However, only in tobacco and two soybean lines (YZ-26-1C and ND-18-44) was faster relaxation also associated with the maintenance of WT high-light NPQ capacity. In other words, maximum NPQ in high light was also increased in VPZ soybean, Arabidopsis, and potato lines, which, except for soybean, was associated with decreased biomass relative to WT. The actual magnitude of residual NPQ in low light decreased only in tobacco, YZ-26-1C soybean, and Arabidopsis lines. Importantly, tobacco, soybean, and Arabidopsis lines showed increases in the low-light effective quantum yield of PSII (ΦPSII) (Fig. 3B).

24While less residual NPQ and higher ΦPSII under fluctuating light are likely strong indicators of improved NPQ kinetics, it is apparent that these traits are not sufficient for increased biomass. Rigorous phenotyping that couple gas exchange measurements of CO2 assimilation with Chl fluorescence under field-relevant conditions is necessary to demonstrate the impact of faster NPQ kinetics, for example, using 2,000/200 µmol photons m−2 s−1 fluctuating light assays that maximize fold changes in NPQ while replicating light fluctuations experienced in crop canopies. Measuring diurnal de-epoxidation states of the xanthophyll cycle or the rate of re-epoxidation following HL-to-LL transitions may be additional useful proxies for mitigation of the more slowly relaxing qZ component of NPQ. A recent analysis of ZEP-overexpressing lines of the stramenopile alga Nannochloropsis indicates that for some species, there may also be room to reduce overall NPQ capacity while accelerating NPQ relaxation, highlighting the importance of species-specific phenotypic interrogation.

25Natural variation in NPQ may provide nontransgenic avenues to breed for faster NPQ kinetics and higher photosynthetic efficiency. Genome-wide association studies for NPQ have previously revealed diversity in NPQ across rice subspecies and Arabidopsis ecotypes. Notably, both studies identified their strongest effect quantitative trait locus within the cis-regulatory regions upstream of the PSBS gene, suggesting NPQ as a trait that is subject to selection. measured gas exchange and chlorophyll fluorescence phenotypes across 155 African rice accessions and found that shoot biomass was negatively correlated with NPQ induction capacity and positively correlated with NPQ relaxation rate, consistent with transgenic VPZ phenotypes associated with changes in biomass (Fig. 3B).

26Several studies in soybean and maize (Zea mays) have identified potential loci for breeding for improved photosynthetic efficiency, using 41 nested association mapping soybean population parents, 751 diverse maize accessions, and 320 multi-parent advanced generation inter-cross maize lines across multiple years. However, in soybean, there appeared to be little quantitative variation in NPQ relaxation, with photosynthetic differences across lines primarily attributed to differences in Rubisco activation. Similarly, both maize studies did not resolve robust quantitative variation in residual low-light NPQ specifically, although differences across putative quantitative trait loci in NPQ capacity and maintenance of ΦPSII suggest that there may be diverse avenues for fine-tuning light-harvesting efficiency. However, crop germplasm screens have not yet revealed differences in NPQ relaxation that are comparable to those in transgenic VPZ plants. Such results could suggest that there are tradeoffs associated with faster NPQ, or more simply, that faster NPQ relaxation has not been selected for in these crops.

27Further efforts to improve NPQ kinetics should include accelerating recovery from the slowly reversible types of NPQ, qH, and qI. Nuclear expression of psbA, encoding the D1 subunit of PSII, to minimize qI associated with heat stress is one such approach with demonstrated success across several species. Additionally, we see tremendous potential in the use of gene editing of endogenous VPZ and other NPQ-related genes in crops to achieve desired NPQ phenotypes. Fine-tuning of VPZ expression and testing diverse VPZ orthologues may be other avenues to bring NPQ-based improvements in photosynthesis closer to their theoretical potential.

28Chloroplast electron transport rate is a primary factor limiting photosynthesis. Classical steady-state quantitative models predict the rate of leaf photosynthesis over a range of intercellular CO2 partial pressures by considering which biochemical reactions are limiting in certain conditions. In the Farquhar–von Caemmerer–Berry model of C3 photosynthesis, carbon metabolism (CO2 availability, Rubisco function) limits assimilation at nonsaturating CO2 and electron transport limits assimilation under saturating light and saturating CO2 (intercellular CO2 partial pressure above 500 µbar; ). Based on the model, at current atmospheric CO2 partial pressure (420 ppm, roughly corresponding to 200 µbar intercellular CO2) C3 photosynthesis is primarily limited by the CO2 supply to Rubisco, but this limitation is anticipated to transition to electron transport when atmospheric CO2 exceeds 600 ppm. In the biochemical model of C4 photosynthesis, electron transport, Rubisco function, and regeneration of RuBP and phosphoenolpyruvate potentially co-limit assimilation under saturating light and saturating CO2 (intercellular CO2 partial pressure above 150 µbar;; ). The vast difference in the saturating levels of CO2 between the two photosynthetic pathways is due to the metabolic C4 cycle of C4 photosynthesis acting as a carbon-concentrating mechanism. The C4 cycle operates across two cell types, mesophyll and bundle sheath cells (BSCs), and increases CO2 partial pressure in BSCs, where Rubisco resides, allowing Rubisco to work at maximum carboxylation rate. Crop models based on these metabolic models of photosynthesis predict yield increases for crops engineered with enhanced electron transport capacity up to 5.2% for wheat and 14.3% for maize.

29Early experiments on plants with genetically reduced electron transport components identified the thylakoid cytochrome (Cyt) b6f complex as the key regulator of the electron transport rate. The complex catalyzes plastoquinone oxidation, considered the slowest reaction of electron transport, and combines it with the translocation of protons across the thylakoid membrane, establishing the proton motive force for ATP generation. A low luminal pH makes proton translocation more difficult and therefore slows down Cyt b6f activity—a phenomenon known as photosynthetic control. Photosynthetic control via Cyt b6f plays an important role in photoprotection by limiting electron flow to PSI and thereby matching the output of the light reactions with the rates of ATP or NADPH consumption by carbon metabolism. Photosynthetic control therefore represents one molecular mechanism coordinating electron transport and carbon metabolism-related limitations of photosynthesis.

30Since Cyt b6f was identified as a bottleneck of electron transport, efforts have focused on alleviating this limitation. In higher plants, the complex is comprised of seven essential subunits, of which Rieske FeS and PetM are nuclear-encoded. Enhancing the content of the Rieske FeS subunit through overexpression of the petC gene has been established as a method for increasing the abundance and activity of Cyt b6f in both C3 and C4 plants. The increased abundance of the complex in transgenic plants has been confirmed through immunodetection of multiple subunits of the complex in leaves and of the whole complex in isolated thylakoids. Elevated Cyt f activity has also been detected in thylakoid membranes isolated from leaves of plants overexpressing Rieske FeS.

31Increasing abundance of Cyt b6f in the model C3 and C4 plants, A. thaliana and Setaria viridis, results in faster electron transport rates through PSI and PSII. Furthermore, in line with the models’ predictions, a boost of electron transport capacity enables elevated CO2 assimilation rates in both model species (Fig. 4). These results confirm that Cyt b6f is a bottleneck of electron transport and that electron transport limits the rate of C3 and C4 photosynthesis at saturating light and saturating CO2. After genetically engineering plants with increased Cyt b6f content was identified as a promising route to improve photosynthesis, overexpression of Rieske FeS has been tested in model C3 and C4 crops, tobacco (N. tabacum cv. Petit Havana) and sorghum (Sorghum bicolor Tx430) (Fig. 4). No differences in steady-state CO2 assimilation rates and only transient increases of electron transport rate (seen as a faster re-oxidation of plastoquinone in the QA binding site of PSII upon increases of irradiance) are detected in tobacco with enhanced Cyt b6f abundance. While in sorghum, increasing Cyt b6f content does not affect the steady-state rates of electron transport and CO2 assimilation, it does speed up the induction of photosynthesis, a process of light-induced activation of photosynthesis after a long period of darkness, and stimulates biomass and grain yield in glasshouse conditions. Overall, results of Rieske FeS overexpression in model crops suggest that the steady-state electron transport rate is no longer limited primarily by Cyt b6f.

32The photosynthesis rate has undergone significant improvement during crop domestication and early breeding. Therefore, it is conceivable that artificial selection of the biggest plants with the greatest yield may have unknowingly increased Rieske FeS content and improved the steady-state electron transport capacity in crops. However, overexpression of Rieske FeS in sorghum has still offered higher yield though transient increases of CO2 assimilation, highlighting the necessity of studying limitations of crop photosynthesis in dynamic light environments. Looking forward, it is also essential to gain a better understanding of electron transport processes, especially in C4 plants, where electron transport chains of mesophyll and bundle sheath chloroplasts, distinct in their composition and outputs, cooperatively support seamless operation of biochemical pathways across the two cell types. Cyt b6f is the first major bottleneck of electron transport to be successfully alleviated using genetic engineering, demonstrating the viability of this approach for crop improvement. Developing models with more detailed and mechanistic description of electron transport will be instrumental for identifying new targets for optimizing electron transport to boost productivity of crops.

33The Calvin–Benson–Bassham (CBB) cycle evolved over 2 billion years ago and is arguably the most important pathway on earth, capturing CO2 from the atmosphere and converting it into organic molecules that are used directly for the synthesis of isoprenoids, sucrose, starch, phenylpropanoids, thiamine, and nucleotides providing the basis for life on our planet. The CBB cycle is the primary biochemical pathway for the fixation of atmospheric CO2 in over 85% of plants, named C3 species, as the first stable product of this cycle is a three-carbon compound, glycerate 3-phosphate. In the 70 years since the CBB cycle was elucidated, it has been shown to be highly conserved across nature from cyanobacteria to the largest of land plants. The CBB cycle involves 11 enzymes and has three stages: carboxylation carried out by Rubisco, reduction, and RuBP regeneration (Fig. 5). Under light-saturating and CO2-limiting conditions, Rubisco activity is the major determinant of the efficiency of carbon fixation via the CBB cycle. However, as atmospheric CO2 levels rise and the light intensity decreases, this balance shifts, such that both the reductive and regenerative phases of the CBB cycle that catalyze the synthesis of the CO2 acceptor molecule RuBP become limiting. Improving photosynthesis has been identified as a target to increase crop yield based on theory, modeling, and empirical studies.

34The complexity of the genetics and the biochemistry has made the Rubisco enzyme a challenging target for manipulation. Nevertheless, major efforts to improve photosynthesis have focused on improving Rubisco activity through both direct and indirect approaches. Direct approaches include protein engineering, directed evolution, natural variation screening, and manipulation of expression in transgenic plants. The introduction of CO2-concentrating mechanisms (CCMs) from algae and cyanobacteria and engineered synthetic pathways to bypass photosynthesis are indirect approaches being taken and are showing some promise (e.g.; ). This section will focus on reactions in the CBB cycle other than Rubisco; see the section below by Carmo-Silva for a discussion of Rubisco.

35In the 1990s, antisense technology demonstrated that Rubisco did not have total control over CO2 assimilation under all conditions and identified sedoheptulose 1,7-bisphosphatase (SBPase), fructose 1,6-bisphosphate aldolase (FBPA), and transketolase (TK) as promising targets for the improvement in photosynthesis. Based on these studies, a transgenic overexpression approach has shown that increasing the levels of SBPase can improve photosynthesis and growth in algae and a number of plant species including: tobacco (in the field and greenhouse), wheat, and Arabidopsis; in contrast, no positive effect was observed in rice. Furthermore, tomato plants with increased SBPase activity were found to be more chilling tolerant with increased photosynthetic capacity. Overexpression of FBPA in tobacco also resulted in positive effects on photosynthesis and biomass, and in tomato, an increase in seed weight in both optimal and suboptimal temperatures was observed. Introduction of the bifunctional cyanobacterial CBB cycle enzyme SBPase/FBPase into tobacco plants, lettuce, and soybean (in elevated CO2) has also resulted in improved CO2 assimilation and growth.

36Improvements in RuBP regeneration have also been realized through the introduction of additional proteins that function outside of the CBB cycle. Examples of this approach include combining expression of SBPase and FBPA with either ictB (a cyanobacterial membrane protein of unknown function previously shown to improve CO2 assimilation) or the H subunit of glycine decarboxylase (GDC) system (shown to increase CO2 fixation possibly through stimulating the photorespiratory cycle and reducing the negative impact of intermediates on the CBB cycle) in tobacco, which resulted in a further improvement in photosynthesis and growth over single-gene manipulations. Overexpression of either the endogenous SBPase enzyme or the bifunctional cyanobacterial SBPase/FBPase, together with the algal Cyt C6 protein, not only improved photosynthesis and yield but also increased water use efficiency when grown under field conditions. A more recent example is the co-overexpression of SBPase with cytosolic FBPase in tobacco plants, resulting in improvements in biomass, plant height, stem diameter, and pod weight, but in contrast co-overexpression of Rubisco with SBPase in rice did not result in an improvement in photosynthesis. Enhanced photosynthetic capacity was observed in plants in which the expression of a group of CBB cycle genes (FBA1, RCA1, FBP5, and PGK1) in response to higher levels of the Brassinazole resistant 1 transcription factor was increased. This result suggests that simultaneous overexpression of these enzymes may stimulate the CBB cycle.

37Importantly as atmospheric CO2 rises, theoretical models predict that the limitation of carbon assimilation shifts from Rubisco to RuBP regeneration. Therefore, modifications that improve RuBP regeneration are predicted to stimulate photosynthesis and yield under elevated atmospheric CO2. This is supported by experimental evidence using plants grown in free-air CO2 enrichment (FACE) facilities; when grown at 585 ppm CO2, transgenic tobacco plants overexpressing SBPase have greater yield increases at elevated CO2. When both CO2 and temperature are manipulated, transgenic overexpression of cyanobacterial bifunctional FBPA/SBPase in soybean protects against temperature-induced yield loss under elevated CO2. These results indicate that improving RuBP regeneration is one approach that could be used to mitigate the effects of climate change on yield, and also demonstrate the importance of testing manipulations in food crops under future climate conditions.

38Advances in kinetic flux and multi-scale modeling have provided novel predictions on how to further enhance the CBB cycle, and the application of rapid high-throughput and iterative approaches will be essential to identify the best candidates to achieve improvements to photosynthesis. Synthetic biology may provide a route to build a completely synthetic, more efficient CO2 fixation pathway to operate in parallel with the endogenous cycle or to introduce improved enzymes to operate within the existing cycle. At the same time, new approaches enabling the identification of genetic factors and mechanisms involved in regulating the expression of CBB cycle genes will underpin the application of gene editing technologies to modify this pathway.

39Rubisco is imperfect yet unique and remarkable. It catalyzes a rather complex set of reactions during the carboxylation of RuBP and can react with O2 rather than CO2, leading to oxygenation of the sugar phosphate substrate. It is the only carboxylase that uses substrates and products that are part of central plant metabolism and that is compatible with the CBB cycle (reviewed by ). Consequently, Rubisco is responsible for assimilating carbon from atmospheric CO2 into sugars, thereby enabling life on Earth. Rubisco's imperfections have been a research subject and a target for improvement since its discovery. Because the enzyme carboxylates RuBP at relatively slow rates, plants invest large amounts of resources (especially nitrogen and carbon skeletons) into making Rubisco in sufficiently high abundance in the leaves of crop plants to support adequate rates of photosynthetic CO2 assimilation and plant growth. The biogenesis of Rubisco, consisting of synthesis and assembly into the hexadecameric L8S8 form present in plants, is itself dependent on several ancillary proteins, adding further demand on agricultural inputs and impacting resource use efficiency. Thus, while adequate abundance of Rubisco is a necessary consideration in strategies aimed at improving photosynthesis, it needs to be balanced with ensuring efficient use of resources to be compatible with sustainable agricultural practices.

40Rubisco catalytic diversity investigations led to the suggestion that tradeoffs associated with the complex reaction mechanism limit the scope for Rubisco improvement. A recent phylogenetic analysis demonstrated that, despite being among the 1% slowest evolving enzymes, the continual evolution of Rubisco was associated with improved catalytic efficiency and with greater leaf-level CO2 assimilation and photosynthetic nitrogen use efficiency (NUE) in C3 plants. Excitingly, a deep mutational scan using a Rubisco-dependent Escherichia coli strain revealed that several highly conserved residues of the loop 6, which folds over the catalytic site and is involved in substrate binding and catalysis, can be mutated without impacting catalysis, suggesting it is possible to target combinations of these residues for future Rubisco engineering efforts.

41Rubisco catalytic sites are located at the interface between two large subunits (RbcL), and thus, RbcL has been considered the main source of catalytic diversity and the primary target for engineering efforts. However, despite being far from the catalytic site, just a few changes in the Rubisco small subunit (RbcS) sequence can alter Rubisco catalytic properties. The role of RbcS in determining the abundance of Rubisco in leaves and adjusting catalytic efficiency has been recently reviewed by who highlighted the rise of RbcS as a target for improvement that is increasingly tractable as engineering approaches become more robust and widely applicable. Replacement of a crop plant Rubisco by a superior version such as that found in rhodophytes is exciting yet challenging given the need to express adequate levels of RbcS, RbcL, and compatible ancillary proteins. A more straightforward solution might be the use of fast-developing gene editing technologies to mutate specific residues required for enhanced catalysis, when we know what these are.

42While the abundance and catalytic properties of Rubisco determine the maximum rate of carboxylation for a given leaf, the activity of Rubisco in crops is regulated by interaction with Rubisco activase (Rca), posttranslational modifications, and the chloroplast stroma environment, which can change rapidly in response to dynamic environmental conditions surrounding the leaf. Under fluctuating light conditions, for example, photochemical, biochemical (including Rubisco regulation), and/or diffusional limitations can affect the efficiency of photosynthesis depending on the duration, frequency, and intensity of shade and full sun periods.

43Rca is a molecular chaperone that couples ATP hydrolysis with conformational remodeling of inhibited Rubisco catalytic sites, releasing sugar phosphate derivatives that occur naturally and bind tightly and unproductively to the enzyme. Rca is itself regulated by the redox status, ATP, and Mg2+ availability in the chloroplast stroma, and some Rca isoforms activate Rubisco more efficiently under fluctuating light. The chaperone is thermolabile, and its ability to restore Rubisco activity is impacted at moderately high temperatures. Progress in understanding the mechanism of Rubisco activation by Rca and identification of more thermostable and more efficient Rca isoforms suggest that improvement in Rubisco regulation to maximize carboxylation in current and future warmer climates is possible. Significant unknowns exist in our understanding of Rubisco regulation by Rca, PTMs, and the chloroplast environment and the role of sugar phosphate derivatives and their phosphatases. Addressing these will aid identification of successful strategies for improving Rubisco, for sustainably increasing crop productivity and climate resilience.

44Importantly, to maximize impact, any approach to improve photosynthesis should be considered holistically, and stacking of improvements in various reactions and sub-processes of photosynthesis will be required to ensure that other processes do not become limiting. For example, chloroplast electron transport is also sensitive to heat stress and has been shown to co-limit photosynthesis alongside Rubisco activation. Optimizing Rubisco activity in crop plants requires consideration of Rubisco abundance, Rubisco catalytic properties, and regulation of its activity (Fig. 6A), as well as how Rubisco activity interacts and is coordinated with other plant processes (Fig. 6B). The interaction with the regeneration of RuBP in the CBB cycle is most obvious, but abundance and properties of Rubisco should take into consideration integration with central and specialized metabolism, as well as the specific crop canopy architecture, the needs of the leaves at various canopy layers throughout crop development, and the remobilization of N and C skeletons stored in Rubisco into the crop product to be harvested.

45The growth of many photosynthetic organisms, including most crops, is limited by the slow rate of CO2 assimilation by Rubisco and competition with O2 at the active site resulting in energetically wasteful photorespiration. In response to the shortcomings of Rubisco, nearly every photosynthetic clade has evolved CCMs to supply Rubisco with concentrated CO2 and preferentially drive CO2 assimilation over photorespiration. CCMs can be broadly divided into two categories based on biophysical or biochemical processes. Biochemical CCMs initially capture CO2 as an organic metabolite before re-conversion to CO2 near Rubisco and include eukaryotic species that perform C4, C2, and crassulacean acid metabolism (CAM) photosynthesis. Biophysical CCMs channel or actively pump in inorganic carbon (Ci, i.e. CO2 and HCO3−) to increase the intracellular Ci pool and include prokaryotic autotrophs, eukaryotic algae, hornworts, and seagrasses. All cyanobacteria and some chemoautotrophic bacteria sequester Rubisco within proteinaceous shells called carboxysomes, while almost all algae and some hornworts condense Rubisco into a micro-compartment called the pyrenoid. Both carboxysomes and pyrenoids are supplied with HCO3− that is rapidly dehydrated by localized carbonic anhydrase (CA) activity to facilitate CO2 enrichment at the active sites of Rubisco.

46The introduction of biochemical or biophysical CCMs into C3 crops is considered a high-risk high-gain engineering strategy to enhance crop yields and resilience (Fig. 7). Models have estimated that the theoretical gains in source leaf photosynthetic efficiency range from 30% to 60%, some of the largest improvements predicted for an engineering approach. Although the levels of translation of such improvements into productivity are debated, for example, the importance of also considering sink demand to take full advantage of source enhancements, a recent conservative model has predicted an 8% increase in wheat yields, specifically for successful introduction of the cyanobacterial CCM. Nevertheless, re-constituting any functional CCM in a C3 plant remains a complex endeavor, and engineering each CCM type is beset by specific challenges. In general, advances in our mechanistic understanding of different CCMs have proceeded in parallel with ongoing engineering efforts in a design build test learn (DBTL)-like cycle.

47Research efforts to introduce biochemical CCMs into C3 plants have largely concentrated on engineering the C4 pathway. The C4 pathway is estimated to increase CO2 levels around Rubisco up to 10-fold above ambient, which provides C4 crops (e.g. maize and sugarcane) with increased photosynthetic efficiencies, growth rates, and in some cases higher water use efficiencies compared to C3 crops. Much of this work has been driven by the C4 rice project, which has focused on introducing the specialized anatomical and biochemical traits of the model C4 crop maize into rice. Thus, C4 engineers have faced two complex challenges: construction of a “Kranz-like” C4 leaf anatomy with reduced vein spacing and increased organization between mesophyll and BSCs, and biochemical re-regulation for appropriate levels of cell-specific protein expression, for example, sequestration of Rubisco in the bundle sheath chloroplasts. Although the factors involved in the development of the Kranz anatomy are still not fully understood, advances in engineering biology have led to exciting progress in reconstituting C4 biochemistry in rice. This includes the generation and application of leaf cell-specific expression systems, such that several key C4 proteins can now be co-expressed and appropriately localized in rice from a single construct. Attention is now also being expanded to the engineering potential of CAM and C2 pathways. The CAM pathway could facilitate increased drought tolerance, while C2 engineering offers some of the benefits of the C4 pathway with fewer anatomical modifications.

48The most efficient biophysical CCMs are carboxysome-based CCMs (cCCMs) and pyrenoid-based CCMs (pCCMs), which can enhance CO2 concentrations around Rubisco by up to 1,000-fold and 40-fold, respectively. Biophysical CCMs also function within a single cell, so transfer to C3 leaf mesophyll cells may be potentially simpler compared to introducing the C4 pathway. cCCMs have been consistently predicted to increase yield gains in C3 crops, in part because cyanobacteria have the fastest known form I Rubiscos, with carboxylation turnover rates (kcat) up to five times higher than C3 species. To date, good progress has been made in understanding the components involved in carboxysome assembly and in reconstructing α- and β-carboxysomes in plants. Recently, reconstructed the α-carboxysome from the chemoautotroph Halothiobacillus neapolitanus in tobacco chloroplasts, which contained active heterologous Rubisco and CA enzymes. However, functional cCCMs in bacteria still require active uptake of HCO3− and CA activity restricted to carboxysome. Thus, testing whether carboxysomes can enhance growth in plants will require the removal of native chloroplastic CA activity and the introduction of functional HCO3− transporters on the chloroplast envelope. Although the latter remains a long-standing challenge, new screening tools to test the functionality of active HCO3− transporters and channels in planta could help to make progress in this area.

49pCCMs are exclusively found in eukaryotes and are likely the most globally abundant CCM type. Pyrenoids are much larger than carboxysomes and characterized by highly diverse morphologies, suggesting that there are many ways to achieve pyrenoid formation. The pCCM is best understood in the model alga Chlamydomonas reinhardtii, whose pyrenoid is characterized by three architectural features: a liquid-like phase-separated matrix of condensed Rubisco, pyrenoid tubules derived from thylakoid membranes that traverse the matrix to supply Ci, and a sheath of starch around the matrix that acts as a CO2 diffusion barrier. Modeling predicts that building a functional Chlamydomonas pCCM in a C3 plant chloroplast could increase CO2 assimilation rates by up to 3-fold. pCCMs do not offer the high carboxylation efficiencies of cCCMs but might be more readily plant-compatible based on the appropriate localization of pCCM components thus far expressed. Furthermore, pCCMs do not require the removal of chloroplastic CA activity and do not need active HCO3− transport, at least at ambient CO2 levels. To date, Rubisco condensation into a “proto-pyrenoid” matrix has been achieved in Arabidopsis, with work ongoing to reconstitute the two remaining architectural features. Excitingly, recent work in diatoms and hornworts shows that a variety of solutions may exist to achieve a functional pCCM in plants, for example, by employing a pyrenoid protein shell instead of a starch sheath. Future work could involve creative synthetic strategies that draw from aspects of all CCMs, such as prokaryotic HCO3− transporters coupled to pyrenoids, or a biochemical C4 pathway localized within a single mesophyll cell (MC) where decarboxylation occurs in a modified chloroplastic carboxysome rather than the BS.

50In this section, we aim to provide a concise overview of photorespiration as a limitation on photosynthetic efficiency, the evolution of C4 photosynthesis via C3–C4 intermediates as an adaptation to low atmospheric CO2 conditions, and efforts to introduce C4 photosynthesis and bypasses of photorespiration into C3 plants as a means to increase photosynthetic efficiency. The efficiency of light energy conversion to biomass in C3 plant photosynthesis is limited by the rate of the oxygenation reaction of Rubisco. In the oxygenation reaction, the acceptor molecule of the CBB cycle, RuBP, is oxidized, not carboxylated, and one of the resulting metabolites, 2-phosphoglycolate (2PG), must be rapidly removed and recycled to 3-phosphoglycerate (3PGA) in a complex pathway called photorespiration. The recycling of two molecules of 2PG to 3PGA releases one molecule each of ammonium and CO2. It also consumes ATP and redox energy (Fig. 8). estimated that photorespiration reduces the yield of major C3 crops by 30% or more. Thus, suppression of photorespiration has great potential to increase crop yield.

51Over the past 20 My, more than 65 dicotyledonous and monocotyledonous plant lineages have independently and convergently evolved a more efficient form of photosynthesis called C4 photosynthesis. The frequent evolution of C4 is thought to have been triggered by a decline in atmospheric CO2 concentrations from >1,000 ppm to about 350 ppm. The decrease in CO2 was accompanied by a drying of the atmosphere and consequently increasing aridity in many parts of the world. Warm temperatures, reduced water availability, and lower atmospheric CO2 are conditions under which the oxygenation reaction of Rubisco is promoted and photosynthetic efficiency decreases.

52C4 plants have a biochemical CO2 concentration mechanism that increases the CO2 concentration at the Rubisco site, thereby reducing the rate of photorespiration. The biochemical CO2 pump is supported by a characteristic leaf anatomy called the Kranz anatomy. Kranz anatomy is characterized by the formation of two concentric layers of photosynthetic cell types around the leaf vasculature. The inner ring, adjacent to the vasculature, consists of BSCs, which have a large cross-sectional area and are densely filled with chloroplasts containing Rubisco and operating the CBB cycle. BSCs usually perform predominantly cyclic photosynthetic electron transport and have low activity of PSII and hence linear electron transport. The outer ring consists of mesophyll cells, which are always associated with BSCs and face the leaf air space. Mesophyll cells typically contain fewer chloroplasts than BSCs, and they contain little or no Rubisco. MC chloroplasts are capable of linear photosynthetic electron transport. The vascular-bundle sheath-mesophyll (V-BSC-M) cell cluster is the basic unit of a C4 leaf and is repeated, resulting in a repeating V-BSC-M-M-BSC-V pattern throughout the leaf.

53Mesophyll cells serve as the first carbon-fixing cells. CO2 from the leaf airspace enters the MC and is converted to HCO3− by CA. HCO3− reacts with phosphoenolpyruvate, catalyzed by phosphoenolpyruvate carboxylase (PEPC), to form the C4 acid oxaloacetate (OAA). OAA is further converted to malate and/or aspartate, which diffuses along their concentration gradient to the BSCs. In the BSCs, malate and/or aspartate is decarboxylated by one of three decarboxylating enzymes (or a combination thereof; Fig. 1B). These are mitochondrial NAD-malic enzyme (NAD-ME), chloroplastic NADP-malic enzyme (NADP-ME), and PEP carboxykinase (PEPCK). The predominant pathway of decarboxylation defines the C4 subtype (NAD-ME, NADP-ME, or PEPCK), although it is controversial whether PEPCK is a distinct subtype or a complementary pathway to either NAD-ME or NADP-ME. The C4 carbon pump raises the CO2 concentration in BSCs to >1,000 ppm, which suppresses the oxygenation reaction of Rubisco to low levels. However, we emphasize that photorespiration is essential in C4 plants, as indicated by the lethal phenotype of mutants in the C4 photorespiratory pathway.

54Based on physiological, anatomical, and biochemical data, as well as computational modeling, it has been proposed that C4 photosynthesis has gradually evolved from the ancestral C3 state through C3–C4 intermediate states. The basic concept is that the photorespiratory pathway loses cell autonomy and is instead split between mesophyll and BSCs. The CO2 (and ammonia) liberation step of photorespiration, catalyzed by GDC in the mitochondria, ceases to function in mesophyll cells and is restricted to BSCs. Glycine decarboxylation in BSCs locally increases the CO2 concentration in this cell type, thereby increasing the efficiency of Rubisco in BSC chloroplasts.

55It has been estimated that under conditions that promote high rates of photorespiration (high temperature, low leaf-internal CO2 concentration due to, e.g. low stomatal conductance), CO2 accumulates in BSCs at two to three times the level in the mesophyll. Glycine decarboxylation by GDC also releases ammonia, which must be refixed to avoid ammonia toxicity and returned to mesophyll cells to maintain nitrogen balance. It has been proposed that carbon skeletons for ammonia shuttling are provided by mesophyll cells in the form of malate, forming a low-level C4 metabolic cycle. In this scenario, malate is oxidized to OAA by NADP-malate dehydrogenase. OAA is subsequently transaminated to aspartate, and aspartate is then transported back to the mesophyll cells. Additionally, a portion of the malate can undergo decarboxylation, forming pyruvate, which can be transaminated into alanine, providing an alternative nitrogen transport mechanism to the mesophyll cells. Both malate oxidation via NADP-MDH and oxidative decarboxylation through NAD(P)-ME are facilitated by a predominantly oxidized NAD(P) pool. If PSII activity is lost from BSCs, it would lead to an increase in the oxidized plastid NADP+ pool. This, in turn, would further promote the evolution toward a full C4 carbon fixation.

56Many staple crops, such as rice and wheat, use C3 photosynthesis. Given the high efficiency of C4 photosynthesis, it has been proposed that conversion of these crops to C4 would result in large yield gains. Indeed, the analysis of a rice new plant type with large panicles and a low number of tillers showed a grain formation efficiency of only 42%, i.e. a substantial number of juvenile spikelets were not converted into mature, filled spikelets. This finding indicated that rice yield is not limited by sink size, but rather by source strength, i.e. the capacity of leaves to produce and export photo-assimilates. Based on these data, it was estimated that rice yield could be doubled if source strength was increased appropriately. To enhance the source strength, the conversion of rice into a C4 plant was proposed as a strategic approach, with the potential to increase yield by 50%. This proposition has gained support through FACE experiments. When elite rice varieties were cultivated in FACE conditions with 200 ppm CO2 concentrations higher than the ambient levels, they produced 13% more grain yield. This suggests that improving photosynthesis and thus source strength directly contributes to higher yields in rice.

57Significant progress has been made in advancing C4 photosynthesis in rice. Activation of photosynthetic organelles in the rice BS was achieved by constitutive expression of the maize transcription factors GOLDEN2 or GOLDEN2-LIKE. Furthermore, five minimal C4 cycle genes (NADP-ME, PEPC, PPDK, MDH, and CA) were successfully expressed in rice from a single construct. The respective gene products were found to be expressed in the correct cell types and subcellular compartments. [13C]–CO2 labeling showed that some of the labeled carbon is directed through PEPC to malate and aspartate. Rapid labeling of citrate indicated movement of the introduced label toward the tricarboxylic acid (TCA) cycle, while no label was detectable in CBB cycle intermediates. The absence of [13C] label in the CBB intermediates indicates that either labeled malate does not enter the plastid stroma or that there is little malate decarboxylation via plastidic NADP-ME. These results suggest that expression of the core C4 metabolic enzymes alone is unlikely to be sufficient to establish a functional C4 cycle. Further development of C4 rice will likely require the co-expression of organellar metabolite transporters, some of which are unknown. Additionally, it is crucial to prevent the diversion of recently fixed carbon into the TCA cycle, and it may be necessary to inhibit photosynthetic linear electron transport in BSCs to achieve the oxidized NADP+ pool required for decarboxylation by NADP-ME. Recently, a single promoter TALE system has been reported for tissue-specific expression of multiple transgenes in rice. In this system, a single cell-specific promoter drives the expression of a synthetic designer transcription activator-like effector that can bind to synthetic TALE-activated promoters. This means that multiple genes can be expressed from a single cell-specific promoter in the desired cell type. This technological advance overcomes previous limitations associated with the limited choice of cell-specific promoters for transgene expression in rice and will facilitate transgene stacking, which is critical given the substantial number of transgenes required to implement the trait.

58Efforts to establish C4 photosynthesis (or any other C3 crop) are ongoing and showing significant progress. In the meantime, other approaches to increase photosynthetic carbon gain in rice and other crops have been explored. The introduction of synthetic bypasses to photorespiration in rice has been associated with yield increases of up to 15%, which is close to what has been observed in FACE experiments. Furthermore, it may be sufficient to install a basic C3–C4 intermediate photorespiratory carbon pump in rice, as metabolic modeling suggests that C3–C4 intermediate photosynthesis increases carbon gain over a wider range of environmental conditions than C4, which is most advantageous under high-light and high-temperature conditions.

59Gaseous exchange between the leaf interior and the external atmosphere is determined by stomatal conductance (gs), and therefore, stomatal regulatory mechanisms play a pivotal role in determining the rates of photosynthetic carbon assimilation (A) and the loss of water through transpiration. Both of these physiological processes hold paramount significance for overall plant performance, productivity, and yield. The importance of CO2 uptake for photosynthesis is self-evident; however, water loss is equally vital for evaporative cooling and the maintenance of optimal leaf temperature to facilitate photosynthesis, as well as transpiration driving uptake of essential nutrients from the soil to the aerial parts of the plant. The regulation of stomatal aperture, which balances CO2 uptake and water loss, is thus fundamental in determining plant water use efficiency, determined as A/E or intrinsic water use efficiency (Wi) when assessed directly as a function of gs, (A/gs). The significant impact stomata have on photosynthetic processes has led to increasing recognition of their potential as valuable targets for manipulation, to enhance crop performance, as well as to develop future crops that can tolerate the challenges posed by climate change.

60Stomatal conductance is the product of both anatomical features and biochemical factors, providing several research avenues for exploitation and manipulation for improved performance. Manipulation of stomatal density (SD) through changes in expression of key genes in the stomatal development and/or patterning pathways has clearly illustrated their potential to either increase the rate of carbon assimilation through removing stomatal limitation or enhance the water use efficiency (Wi; ). Although there are various approaches to manipulating stomatal numbers, genes within the epidermal patterning factor (EPF) and epidermal patterning factor-like (EPFL) family have been a particular focus. Overexpression of EPF1 and 2 has demonstrated that a reduction in SD (and conductance) resulted in improved drought tolerance and increased water use efficiency. However, decreased stomatal conductance usually lowers the rate of carbon assimilation; therefore, it is intriguing that reducing SD (by between 46% and 58%) in the key C3 crops, rice and wheat did not impose any diffusional constraints on carbon assimilation. Overexpressing EPFL9/STOMAGEN increases SD leading to a greater conductance associated with enhanced A, highlighting the impact of diffusional constraints on carbon assimilation. However, it is important to note that while augmenting SD to increase stomatal conductance may appear promising for removing diffusional constraints, increased water loss erodes water use efficiency. Furthermore, high SD has been linked to stomatal clustering and impaired stomatal kinetics, resulting in reduced stomatal conductance and rates of carbon assimilation. These effects are attributed to reductions in critical ion channels necessary for guard cell (GC) osmoregulation and pore opening, such as reduced K+ channel activity. Therefore, while the manipulation of SD presents a promising avenue for modifying stomata conductance, it is becoming evident that future endeavors should adopt a more holistic approach that considers both anatomical aspects and functional attributes such as stomatal kinetics (Fig. 9).

61The close correlation between the rate of carbon assimilation and stomatal conductance (under steady-state conditions) is well established; however, it is not always constant, and under dynamic conditions, stomatal responses to changing environmental conditions are an order of magnitude slower than the rate of carbon assimilation, leading to a temporal disconnect between these two processes. Slow stomatal opening has been reported to reduce carbon assimilation by ca. 10% across a range of species, while slow closure leads to unnecessary water loss that can erode water use efficiency by up to 50%. Significant species and cultivar differences in the kinetic responses of stomata that depend on anatomy and biochemistry have been demonstrated. This research has led to increasing interest in the rapidity of stomatal conductance and GC regulation as novel targets for improving carbon assimilation and water use efficiency. Recently, used optogenetics to enhance solute fluxes in GCs, accelerating stomatal opening and closing, and verifying that enhancing stomatal kinetics improved water use efficiency with no carbon penalty. Engineering ion channels involved in K+ fluxes in the GCs has further demonstrated the potential to increase the kinetic responses of stomatal conductance and improve water use efficiency. GC metabolism also holds the promise of offering numerous innovative targets for manipulating both the rapidity of stomatal responses and the coordination between carbon assimilation and stomatal conductance (see review by ).

62Smaller stomata have been proposed to have faster kinetics, most likely facilitated by the greater surface area to volume ratio enabling more rapid solute and water exchange between GCs and the surrounding cells. However, to date little is known about the genetic regulation of stomatal size, and no key targets have been identified for manipulation. Stomatal size has been linked to stomata density, with smaller stomata associated with higher density; therefore, it may not be possible to manipulate size independently of density. The shape of the GCs and cells that surround them (called subsidiary cells) also influence the speeds at which stomata open and close. confirmed the importance of subsidiary cells in Brachypodium distachyon mutants lacking these cells, which resulted in slower kinetics and reduced stomatal conductance. These findings have opened up subsidiary cell metabolism and transport as a potential new avenue of research to explore novel targets for increasing stomata kinetics (Fig. 9).

63While the scope of this perspective may not encompass all recent advancements, it is worth noting that numerous exciting developments in stomatal research could provide new opportunities to improve crop performance in a changing climate. Nonfoliar stomatal behavior is gaining increasing attention due to the significant impact on photosynthesis and crop yield (especially during periods of plant stress). Additionally, current work has demonstrated considerable water loss associated with these nonfoliar tissues, providing possible new targets to improve whole-plant water use. While most research primarily concentrates on leaf-level stomatal responses, recent studies have unveiled differences in stomatal kinetics between the adaxial and abaxial stomata. This discovery opens up the potential to independently manipulate these two surfaces, if the genes responsible for the stomatal patterning on each surface can be elucidated. However, if we are to exploit the potential targets highlighted in this perspective we need a greater understanding of GC signal transduction pathways, metabolism, and the mechanisms that coordinate carbon assimilation and stomatal conductance.

64The model of Z-scheme oxygenic photosynthesis, with the CBB cycle fixing CO2, is the dominant photosynthetic process of the biosphere, and as such, it provides most of the energy and biomass that supports life on Earth. The consistent use of a Z-scheme CBB cycle photosynthetic engine means that there is little variation in the basic physiological and biochemical mechanisms of photosynthesis. However, since the colonization of land by the embryophytes in the late Ordovician, this basic mechanism has been adapted by evolution to face the challenges of photosynthesizing optimally in the diverse environmental niches found in the terrestrial biome.

65Photosynthesis is a complex process that depends on the cooperative activity of many processes so it can be tuned or optimized along many dimensions, for example, resource use efficiencies such as for light, water, and mineral nutrients, responses to temperature, leaf architecture, and longevity, tolerance to abiotic stress, recovery from stress, etc. These axes for adaptation are individually complex; for example, light can vary in myriad ways (e.g. intensity, spectrum, periodicity, fluctuations over differing time scales) during the life of a leaf or plant. This has resulted in the evolution of variation in photosynthetic traits (e.g.;;; ). Taking the maximum rate of photosynthetic CO2 fixation per unit leaf area (Pmax, units μmol m−2 s−1) as an example, for a plant using the C3 photosynthesis mechanism it has been estimated that Pmax should be about 55 μmol m−2 s−1 at an irradiance of 2,000 μmol m−2 s−1. Some C3 plants (e.g. desert winter annuals) achieve a slightly higher Pmax [60 to 65 μmol m−2 s−1 for Chylismia (formerly Camissonia) claviformis and Palifoxia linearis]. In contrast, a typical C3 crop plant will have a Pmax of 20 to 30 μmol m−2 s−1, deciduous forest tree species a Pmax of about 10 μmol m−2 s−1, while for permanently shaded rainforest understory species, Pmax is 1 to 5 μmol m−2 s−1.

66Despite the variation for photosynthesis observed in natural systems, it has been argued that photosynthesis in crop plants cannot be improved because it has already been optimized by natural selection and evolution. This argument is, however, only partly correct because plants in nature do not experience the same pressures as plants in agriculture. The wide diversity of photosynthetic properties found in nature, including among the wild ancestors of crop plants, in broad terms is expected to represent different optima for photosynthesis adapted to different naturally occurring niches. These niches are, however, often limiting in various ways, for example, in water and nutrients (such as nitrogen and phosphate), and plants in nature usually experience strong competition from their neighbors. In agriculture, however, competition with noncrop plant species is usually eliminated (or largely so), nutrients and water are often supplied, and attack from pests and diseases are managed, so far as possible, especially in intensive agriculture and protected horticulture. These agricultural environments do not have any perfectly natural analogs, so evolution will not have optimized photosynthesis for agriculture, and it can therefore still be improved in crop plants. In principle, therefore, photosynthesis is no different to those other plant properties that have been improved by domestication and breeding, despite their having been previously optimized by evolution in the wild ancestors of our crop plants. In the past, however, breeding for improved photosynthesis has occurred only to a limited extent, owing to the difficulty of phenotyping for photosynthesis alongside the complexity of the process.

67Variation in photosynthetic properties of land plants is a valuable resource in terms of understanding the operation and limitations of photosynthesis and finding photosynthetic syndromes or traits that could potentially be used to improve crop plant photosynthesis. The contribution that natural variation can make to understanding and improving photosynthesis can be divided into the following five broad categories:

68The limits to the adaptability of plant photosynthesis to extreme habitats and environments.

69Trait variation within or across species that facilitates the physiological analysis of traits and the identification of their underlying mechanisms.

70The co-occurrence of subtraits that reveal syndromes of evolutionary adaptation of photosynthesis.

71Physiological models or templates that serve as options for improving photosynthesis.

72Sources of genetic (or allelic) variation that can be used to breed (including by means of novel plant breeding techniques) for improved photosynthesis.

73In addition to providing a way to extend our knowledge of the operation and regulation of photosynthesis and the limits reached by evolutionary refinement of the process, the occurrence of natural variation of photosynthesis implies there must be an underlying genetic underpinning. If we can unravel this genetic basis of variation for photosynthetic traits we can begin to systematically breed for improved photosynthetic traits. The importance of linking variation in photosynthesis (i.e. phenotypic variation) to genetics is paramount whether or not novel plant breeding techniques or conventional breeding is used to genetically improve a plant. The problem is that while the genes for the headline components of photosynthesis (e.g. the subunits of PSII or PSI) are well known, the genetic basis of variation in many key photosynthetic properties is poorly understood, even when the physiology of that variation is well known. For example, the maximum rate of photosynthesis is correlated with numerous changes in the protein, lipid, and cofactor composition of the leaf, alongside anatomical differences (e.g.;;;; ). Higher rates of photosynthesis are, for example, correlated with greater activities of Cyt b6f complex or Rubisco, but this increase in activity is largely achieved by there being more of these components per unit area and not by there being super Cyt b6f or Rubisco (i.e. complexes with substantially higher specific activity; e.g.;;; ). The genetic basis for how an increased amount of these complexes per unit area of leaf is achieved is poorly understood despite phenotypic variation in Pmax being widely studied and encountered. The same can be said of other photosynthetic traits (or subtraits). This does not mean, however, that no natural variation exists for genes coding for the headline components of photosynthesis—such variation does exist (e.g.; ) and can potentially be exploited.

74If there is variation for a trait within a species, or a pool of species, that can be hybridized to give rise to genetically segregating offspring, then it is possible to correlate phenotypic variation with genomic variation using a mapping population and by this means identify genomic regions—QTL—within which a gene or genes that are causal for phenotypic variation are located (e.g.; ). Mapping populations can be constructed in various ways, but critically they need to contain variation for the trait under investigation. The individuals comprising the population (typically 100 and often many more, depending on the nature of the population) must also be genomically mapped using markers (commonly SNPs) that essentially serve to describe the variation of the genome. If there is both genomic and phenotypic variation, these can be correlated and regions of genomic variation that are associated with phenotypic variation identified. A critical requirement is that of phenotyping; photosynthesis is strongly affected by the environment and is, in any case, difficult to measure rapidly and on the large scale needed to adequately phenotype a large population. This can be achieved by the large-scale use of portable gas analyzers or by using robotic chlorophyll fluorescence-based imaging systems (e.g. ). Comparing marker-assisted breeding and genomic selection on the one hand with genetic modification and gene editing on the other hand, the latter requires the identification of genes whose variation results in phenotypic difference, while the former can be carried out knowing only the association between phenotypic variation and variation in genomic markers (e.g. single-nucleotide markers or SNPs)—an approach that is widely used in commercial plant breeding [see for a summary of QTLs and their use].

75Improving photosynthesis is still a largely unexplored strategy for enhancing crop productivity, offering a means to expand crop improvement possibilities while sustaining yields. Even considering only the land plants (the embryophytes), photosynthesis has a wide range of properties many of which vary not only from species to species, but within a species. In some cases, these properties and their variation are not well explored. This natural variation for photosynthesis is, nonetheless, a valuable resource for improving crop photosynthetic properties, providing us with physiological templates and genetic resources with which we can improve crop photosynthesis using either conventional or novel plant breeding techniques. To make best use of this resource, we need to better understand the limitations on photosynthesis in the field and improve the tools needed for identifying phenotypic variation for photosynthesis. We also need better tools for identifying the causal genes underlying photosynthetic variation and, finally, develop strategies for applying these discoveries in crop improvement programs.

76Engineering canopy photosynthesis, instead of leaf photosynthesis, is required to improve crop yield potential. Canopy photosynthesis is an integral of photosynthetic CO2 uptake rates for all leaves in a canopy, including those at the top, which usually receive high irradiance, and those in the lower layers, which usually receive low irradiance. Earlier wheat breeding programs suggest that cultivars with a higher light-saturated rate of leaf photosynthesis are usually associated with a lower leaf area index, which consequently negates the positive impact of enhancing leaf photosynthetic rate. Therefore, identifying engineering targets that can increase canopy photosynthesis becomes crucial to achieving the desired goal of increasing photosynthesis for greater yields.

77Canopy photosynthesis is controlled by both the microclimate parameters inside a canopy, such as CO2, light, temperature, and humidity, and the photosynthetic properties of all leaves in a canopy. Although canopy photosynthesis can be measured with canopy chambers, the development of accurate canopy photosynthesis models is needed for the precise dissection of the main players and regulatory factors. Various models have been developed, often based on 3D structural modeling and energy balance approaches and which differ in their levels of detail in canopy architecture and microclimate heterogeneity. These models have been used to identify key architectural parameters, such as optimal leaf area index and ideal leaf angle to enhance canopy photosynthesis.

78Besides microclimate, another factor controlling canopy photosynthesis is leaf photosynthetic rate, which shows large variation among species and between cultivars of the same species. This variation is mainly attributed to differences in photosynthetic properties, e.g. Rubisco content, Rubisco activation state, CBB cycle enzyme activities, the abundance of electron transfer chain components, mesophyll and stomatal conductance, and leaf anatomical features. To enable the precise dissection of factors controlling leaf photosynthetic rate, 3D reaction–diffusion models of leaf photosynthesis have been developed that effectively couple leaf anatomical features, CO2 diffusion processes, and light distribution inside a leaf. These models indicate that leaf biophysical and biochemical properties associated with photosynthesis play a dominant role in determining photosynthetic rates, while leaf anatomy makes a relatively minor contribution.

79There is a long history of modeling photosynthetic systems to identify critical proteins and enzymes and their biochemical and biophysical properties that control photosynthetic efficiency. Several modeling approaches have been used, e.g. systems of ordinary differential equations to model photosynthesis-related metabolic processes, reaction–diffusion models to simulate the gas diffusion and the coupled reaction processes in a MC or leaf, and ray-tracing algorithms to simulate light distribution inside a leaf (Fig. 10). Systems models of photosynthesis for C3, C4, and CAM leaves have been developed, as well as models simulating stomatal conductance and NPQ dynamics.

80Analyses with these systems models have generated tremendous insight into strategies for engineering photosynthesis. For example, a dynamic system model of canopy photosynthesis showed that faster recovery from photoprotective states improves canopy photosynthesis, which in turn results in increased biomass production in both soybean and tobacco. Systems models of photorespiratory bypass suggest that decreased expression of PLGG1, a glycolate/glycerate transporter, results in an increased benefit of photorespiratory bypass and further increased photosynthetic CO2 uptake rate, which was later experimentally confirmed in tobacco and rice. Analysis with a systems model of C3 photosynthesis suggested that simultaneous overexpression of SBPase and FBPase results in increased photosynthesis, which was again later demonstrated in wheat. Modeling together with field experiments has established a suite of engineering options that can be used to overcome limitations on photosynthesis. These options are discussed individually in the earlier sections. Here, we summarize them from the angle of access to substrates of photosynthesis, i.e. CO2 and light, and limitations due to the inefficiency of proteins or enzymes involved in photosynthesis (Fig. 10B).

81Although there are successes of current models in guiding photosynthesis engineering, the suggested strategies do not always deliver the predicted increase in photosynthesis and biomass, as in the case of increasing the NPQ relaxation in Arabidopsis and potato, or increasing the expression of SBPase in rice. One potential reason is that factors limiting photosynthesis might shift to different processes under different environments or in different species. For example, under high light, Rubisco has higher control over photosynthetic CO2 uptake, while under low light, components of the electron transfer chain exert higher control; as a result, engineering a particular enzyme may not necessarily increase photosynthesis for a particular crop under a particular condition. However, this does not nullify the possibility that a particular step limits photosynthetic rates under other conditions, which therefore calls for field studies to test whether a particular engineering option will work for a particular plant or a particular cultivar. In some sense, once the original concept that a particular protein is limiting photosynthesis in model plant species is established, private enterprises might be better positioned to take the lead in systematically testing its application in different crops under different conditions through large-scale field testing.

82It is worth pointing out that while the use of the systems model of photosynthesis has shown promise in guiding photosynthesis engineering for higher efficiency, this area of study is in its infancy. Great opportunities lay ahead to develop more advanced models to guide future engineering and design for higher photosynthetic efficiency. First, current models need to include a description of the acclimation of photosynthesis (e.g. expression, assembly, and degradation of proteins) to various environmental factors such as light, humidity, and temperature as well as internal factors like sink capacities. For example, a recent study shows that the ATP required for PSII repair processes is on average 4.6% of that used for photosynthetic carbon assimilation. Moreover, studies show that increasing the synthesis of the D1 protein, which has a high turnover rate, enhances high photosynthetic rates. Second, the complete redesign of photosynthetic CO2 fixation and carbon metabolism has the potential to dramatically improve photosynthetic efficiency. Third, systems modeling of excitation energy transfer and electron transfer processes need to incorporate the spatial organization of photosynthetic proteins and pigment–protein complexes in the thylakoid membrane. This information is crucial not only for understanding the principles underlying the high photosynthetic efficiency but also for identifying targets to optimize plants for even higher efficiency. Finally, the enzymes or proteins predicted as limiting factors by systems modeling can be directly correlated with sequence variations and molecular dynamics simulations to identify optimal genomic editing strategies for photosynthesis-related genes. With the rapid increase in computational power and the expanding capacity of photosynthesis models, we are entering an era of rational design not only for designing new pathways but also new proteins to significantly enhance the efficiency of crops.

83In the dense monoculture system of current row crop agriculture, low-light use efficiency at the top of the crop canopy and limited light availability within the canopy conspire to limit carbon gain. The notion of a “smart canopy” concept was proposed for optimizing canopy photosynthesis. Proposed strategies for improving photosynthesis at the canopy level could benefit from considering a smart canopy concept in the context of optimization of nitrogen (N) distribution within the canopy and to provide promising target genes and enabling technologies.

84Studies in optimization theory suggested that maximizing canopy photosynthesis requires N distribution proportional to the light availability within the canopy. Leaf N decreases gradually from the top to the bottom of the canopy, which is a crucial adaptation to declining light availability within the canopy. The vertical distributions of light and N within canopies are described using the extinction coefficients for light (KL) and N (KN). Canopy photosynthesis can be maximized by the optimal gradient of N and light (i.e. KN/KL = 1) within canopies. A meta-analysis of canopy N distribution has revealed that the KN/KL for most plant species is approximately 0.5, implying that improving canopy photosynthesis can be achieved by either reducing KL (more uniform light distribution) or increasing KN (less uniform N distribution) [see for more details]. More than half of the leaf N is invested in the photosynthetic apparatus, suggesting that “smart” regulation of photosynthetic apparatus within a canopy will mediate optimizing N distribution.

85Various strategies have been proposed to enhance leaf and canopy photosynthesis, with many of them having been tested in field experiments. assessed the N costs of these strategies. For example, engineering Rubisco and ATP synthase was expected to have a high N cost, as they account for a significant portion of the N budget of the leaf (20% and 8%, respectively). On the other hand, overexpressing enzymes such as Cyt b6f, Psbs-VDE-ZEP, SBPase, and FBP aldolase incurred a medium N cost. The concept of a “smart canopy” can minimize the N cost by overexpressing rate-limiting enzymes only at the top of the canopy where sufficient light is available while repressing N investments in the enzymes where light is limited. This smart canopy strategy aims to maximize canopy-level photosynthesis while optimizing N distribution within the canopy. Interestingly, pointed out that one of the strategies for improving photosynthesis can save N, rather than incur additional N costs: reducing the light-capturing machinery such as chlorophyll and antenna proteins. proposed the smart canopy has more RCs and fewer antenna proteins at the top of the canopy, with the reverse arrangement in the lower canopy.

86For agricultural purposes, crop plants overinvest in light capture while underinvesting in light utilization to optimize canopy carbon gain. Studies on low-chlorophyll crops have shown that many plants invest excessively in the production of chlorophyll and its associated LHC. One significant evolutionary advantage of overinvesting N in light capture is that it confers a selective advantage by shading potential competitors. Even when a leaf is light-saturated and cannot utilize additional light, intercepting more light prevents potential competitors from receiving and benefiting from it; plants even produce more leaves than necessary to capture light. Another benefit may be that many crops evolved under conditions of limited N, and thus, it is adaptive to sequester N whenever it is available, storing it in proteins like Rubisco to conserve this typically scarce resource. However, this N investment strategy is suboptimal for densely planted agricultural monocultures, where the main goal is to maximize net primary productivity in the field. A decrease of light capture, rather than an increase, to achieve improved light distribution in the canopy would benefit NUE and perhaps overall canopy photosynthesis.

87Theory suggests an increase in total canopy photosynthesis if N saved by reducing leaf chlorophyll content was optimally reallocated to photosynthetic capacity that matched increased light levels within the reduced chlorophyll canopy. Experimental verification is needed to assess the impacts of these modifications on leaf and canopy photosynthesis. Theoretically, a 50% reduction in leaf chlorophyll and LHC could result in 7% to 9% savings in leaf N without negatively affecting canopy photosynthesis. The saved N from the reduction of chlorophyll could be utilized for the overexpression of rate-limiting enzymes in the CBB cycle, such as Rubisco, FBP aldolase, and SBPase. Proposed strategies include different N investment approaches at various canopy levels, allocating N from lower to upper canopy leaves. Beyond the predicted advantages of re-investing saved N in increased photosynthetic capacity, demonstrated that reducing canopy chlorophyll levels in tobacco by up to 50% did not compromise carbon assimilation but increased seed N concentration by 7%, indicating that saved N from reducing chlorophyll and LHCs may be redirected toward seed N.

88Multigene constructs, overexpressing a suite of rate-limiting enzymes such as Rubisco, SBPase, and FBP aldolase at the top of the canopy while downregulating overinvested enzymes in the lower canopy, will better optimize N distribution and improve canopy photosynthesis. These constructs should be regulated by distinct promoters customized for specific canopy heights. Perhaps the ratio of red/far-red (R/FR) light serves as a reasonable inducer, especially as the R/FR ratio decreases with canopy height. With the advent of optogenetics, blue light- and FR light-inducible gene expression systems have been tested in microorganisms (; Liu et al. 2023). However, these systems have not yet been tested in plants. The red light-inducible promoter has been implemented in a plant system, but it was found to turn off gene expression under white light, rendering it unsuitable for the intended purpose. Therefore, a canopy height-specific gene expression system should be further investigated under actual canopy conditions, preferably in field experiments.

89The redistribution of the investment of N from light capture to photosynthetic capacity is a legitimate opportunity to improve light use efficiency and canopy carbon gain for the dense canopies of current row crop agriculture. Candidate targets for N savings and reinvestment have been identified and highlighted in this brief overview. The next critical steps involve adapting gene regulation and the validation of these predictions in target crop species in field experiments and ultimately in multi-location field trials to assess impacts on crop yield.

90In this perspective, we have discussed key targets and methodologies designed to enhance photosynthetic efficiency, highlighting their significant potential for crop yield improvement. The next critical step is the integration of these approaches. By combining strategies such as optimizing Rubisco efficiency, boosting electron transport, introducing novel pigments and refining canopy structures, and improving stomatal regulation and photosynthetic responses to environmental fluctuations, we can harness their collective potential. Importantly, this integration must take into account the complex interplay of synergistic and antagonistic effects among these modifications to maximize their benefits for agricultural productivity. The combination of these strategies should then aim not only at stacking improvements but also at ensuring real benefits in terms of plant resilience and productivity in diverse farming conditions. This means not just mixing different improvements but also understanding how they work together in field conditions—how they affect plant growth, sink–source relationships, and the response to environmental stresses, and when and how photosynthesis limits crop productivity. This requires extending the focus beyond enhancing photosynthetic efficiency, incorporating traits that increase resilience to abiotic and biotic stresses, improve water use efficiency and NUE, and decrease the yield gap under fluctuating environmental conditions. Importantly, improving primary production creates flexibility in crop plant design. The increased carbon could be used to increase yield or, for example, it could be allocated, without a decrease in yield, to increased root biomass to improve agricultural sustainability by improving nutrient or water capture or by increasing soil organic carbon. The goal is to create crops that are more productive but also better suited to the diverse environments they grow in, directly addressing the pressing demands of global food security.