Metabolic Engineering Towards the Enhancement of Photosynthesis


  • This invited paper is part of the Symposium-in-Print: Photosynthesis

*Corresponding author email: (Christoph Peterhansel)


Photosynthetic capacity is a promising target for metabolic engineering of crop plants towards higher productivity. Crop photosynthesis is limited by multiple factors dependent on the environmental conditions. This includes photosynthetic electron transport, regeneration of CO2 acceptor molecules in the reductive pentose phosphate cycle, the activity and substrate specificity of the CO2-fixing enzyme Ribulose-1,5-bisphosphate carboxylase/oxygenase, and the associated flow through the photorespiratory pathway. All these aspects of the photosynthetic network have been the subject of recently published metabolic engineering approaches in model species. Together, the novel results raise hopes that engineering of photosynthesis in crop species can significantly increase agricultural productivity.


The increasing world population and the use of agricultural products for the production of renewable energy resources in addition to human nutrition will demand a sharp rise in agricultural productivity in the next decades. It has long been questioned whether increases in photosynthetic capacity can help to enhance crop productivity at least in terms of grain yield. During the last years, a multitude of free air CO2 enrichment (FACE) studies were implemented to understand the consequences of atmospheric CO2 increase caused by the burning of fossil resources. One important result of these studies is that an increase in photosynthesis under these conditions is translated into higher biomass production and seed yield albeit to a lower extent than theoretically predicted (1,2). Moreover, modeling analyses suggest that the photosynthetic capacity of crops under the present atmospheric conditions can be enhanced by genetic manipulation (2,3).

Most of our crop plants fix CO2 according to the C3 mechanism. The photosynthetic capacity of these plants is limited by multiple factors dependent on the environment. Under conditions of high CO2 supply and sufficient illumination, the activity of Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) often limits carbon fixation. Rubisco is an exceptionally slow enzyme and is only capable of fixing a few CO2 molecules per second (4). Huge quantities of the enzyme are required to produce sufficient amounts of carbohydrates for plant growth and the energy requirements of the cell making Rubisco the most abundant protein in the world (5). Besides the low specific activity, only part of all Rubisco proteins are activated by carbamylation, the covalent addition of a CO2 molecule to a lysine side chain, and the subsequent binding of a Mg2+ ion (6,7). This further reduces the maximum CO2 fixation capacity of the plant. Parts of the already fixed and reduced carbon are lost from the plant because Rubisco also fixes molecular oxygen instead of CO2 and the recycling of the reaction products of Rubisco oxygenase activity in the process of photorespiration includes a CO2 release step (8). At atmospheric CO2 concentrations and a temperature of 25°C, approximately one out of five catalytic events is an oxygenase reaction (9). This ratio clearly increases if plants are exposed to drought and heat because the specificity of Rubisco for CO2 as well as the solubility of CO2 (relative to O2) in water decrease at higher temperatures. The closure of stomata under such conditions results in reduced CO2 supply while oxygen is constantly produced in the light reactions of photosynthesis. This further favors the oxygenase reaction of Rubisco (10).

Beside these inadequacies of Rubisco, photosynthesis is also often limited by regeneration of the CO2 acceptor molecule ribulose-1,5-bisphosphate (RuBP) in the photosynthetic carbon reduction cycle (PCRC) or “Calvin cycle.” This either relates to the activity of some enzymes of the PCRC (11) or to the provision of energy and reducing equivalents from the photosynthetic electron transport chain. The latter are required for the reduction of the fixed carbon molecule from a high oxidation state in CO2 to a low oxidation state in carbohydrates.

In this review, we summarize recent metabolic engineering approaches that address all the listed limitations of photosynthesis in C3 plants. An overview of the processes targeted in the different studies is given in Fig. 1. Background to much of this work has been excellently reviewed recently (2,12) and is therefore only briefly discussed here. Although most of the experiments have been performed in model species and not in crops, the results provide good evidence that modification of components of the photosynthetic system by gene technology can assist breeding of crops with improved biomass productivity and/or yield.

Figure 1.

 Overview of photosynthetic carbon metabolism. The photosynthetic carbon reduction cycle (PCRC) is shown in blue, the photosynthetic electron transport chain (PETC) in green, photorespiration in red and Rubisco activase (RCA) in purple. Recent papers on the modification of specific reaction steps towards the improvement of photosynthesis are indicated. PSII, Photosystem II; PQ, Plastoquinone; Cytb6f, Cytochrome b6f; PC, Plastocyanin; PS I, Photosystem I; Fd, Ferredoxin; FNR, Ferredoxin-NADP(H)-reductase; GAP, Glycerol aldehyde phosphate; 3-PGA, 3-phosphoglycerate; RuBP, Ribulose-1,5-bisphosphate; RUBISCO, RuBP carboxylase/oxygenase; 2-PG, 2-phosphoglycolate.

Photosynthetic Electron Transport Chain

The photosynthetic electron transport chain (PETC) provides energy and reducing equivalents for the reduction of fixed CO2 to carbohydrates in the PCRC. Transport of electrons through the PETC is mostly limited by the reduction and re-oxidation of the diffusible components plastoquinone, plastocyanin and ferredoxin that interconnect the membrane-bound complexes (13). In an approach to increase electron flow through the PETC, Chida et al. overexpressed cytochrome c6 (Cytc6) in Arabidopsis (14). In many algae, Cytc6 participates in electron transfer between the cytochrome b6f complex (Cytb6f) and photosystem I (PSI), whereas this reaction is exclusively mediated by plastocyanin (PC) in higher plants (cf. Fig. 1). Sequence homologues to Cytc6 are found in higher plants, but these are probably not part of the PETC, but rather signaling molecules that record the redox state of the chloroplast especially under stress conditions (15). A Cytc6 protein from the red alga Porphyra yezoensis was chosen for the approach based on the similar redox potential compared with Arabidopsis PC. This was supposed to enable efficient electron uptake and transfer in the heterologous system. Overexpression of Cytc6 resulted in faster stem elongation as well as leaf and root growth when plants were grown at low light intensities and long-day conditions. The growth effect was accompanied by higher contents of photosynthetic metabolites and improved chlorophyll fluorescence characteristics. Particularly, the data indicated that the plastoquinone pool was more oxidized in transgenic plants suggesting that electron transfer downstream of plastoquinone was more efficient. These in vivo data suggest that red algal Cytc6 can in fact act as an electron transfer protein between Cytb6f and PSI in higher plants and that an enhancement of this transfer process can increase plant growth under specific conditions. It remains to be shown whether the same effect could be obtained simply by overproduction of Arabidopsis PC. Based on in vitro electron transfer studies, the authors argued that Cytc6 transfers electrons faster than PC to PSI and that these enhanced biophysical properties form the basis of the positive growth effects.

In an alternative approach, Rodriguez et al. overexpressed chloroplastic ferredoxin NADP(H) oxidoreductase (FNR) in tobacco (16). This enzyme catalyzes the terminal reaction of the PETC by transferring electrons from reduced ferredoxin to NADP+ (see Fig. 1). Previous experiments with antisense lines had shown that even a moderate reduction of the amounts of this enzyme has a negative impact on net photosynthesis under both light-limited and light-saturated conditions indicating a high level control of this enzyme over electron transport (17). Unexpectedly, overexpression of FNR did not increase photosynthesis or plant growth independent of whether plants were grown at high or low light. Analyses with thylakoid preparations from transgenic plants and different artificial electron donors revealed that electron transfer from PSI to NADP+ was indeed more efficient in FNR overexpressing lines. However, the data also indicated that at least one additional transfer reaction upstream of PSI constitutes a second bottleneck of the PETC. Combination of FNR and Cytc6 overexpression in one transgenic line might therefore turn out to be valuable in enhancing electron transport through the PETC. Whether this can be translated into enhanced biomass production will also depend on the capacity of the PCRC to use the increased supply of reducing equivalents. The phenotype of the Cytc6 overexpressors provides evidence that this is possible at least under some growth conditions.

Rubisco Activity and Specificity

As mentioned in the introduction, Rubisco catalyzes both the carboxylation (CO2 fixation) and the oxygenation (O2 fixation) of RuBP. The products of oxygenation have to be recycled at high energy costs in the process of photorespiration and parts of the fixed carbon are lost (see also chapter 4). Increasing the specificity of Rubisco for CO2 would therefore augment crop photosynthesis especially under conditions where light and thus energy are limiting. However, under high light conditions, the maximum catalytic efficiency of Rubisco becomes more important and is often limiting photosynthesis. During drought, specificity is again more important as CO2 supply is limited by the closure of stomata. Unfortunately, there is an inverse relationship between specificity and catalytic efficiency of Rubisco. A recent theoretical analysis of the molecular interactions underlying both parameters indicate that evolution provided Rubisco enzymes already with an optimized trade-off between both features (18). However, this trade-off might have been established under atmospheric CO2 concentrations of only 200 μmol mol−1 (the average levels during the last 450 thousand years) and the recent increase in CO2 caused by the burning of fossil resources might now rather favor a Rubisco with higher catalytic efficiency and reduced specificity for CO2 (3).

The general inverse relationship between both parameters is not an absolute rule as Rubisco enzymes from some red algae show high specificity values at acceptable catalytic rates (19). A recent survey of Rubiscos from higher plants with hot and arid habitats also identified promising candidates. Namely, the Rubisco of the sea lavender Limonium gibertii showed 30% higher catalytic rates than the enzyme from tobacco with similar specificity values (20,21). This enzyme is already a promising candidate for transfer to crop plants. More importantly, the results suggest that a further survey will identify additional Rubisco isoforms with both high activity and specificity.

The alternative of improving Rubisco’s enzymatic properties by mutagenesis has had little success so far (22). In an elegant metabolic engineering approach, Parikh et al. have constructed an Escherichia coli cell that is dependent on Rubisco activity in order to grow (23). The authors used this system to select mutant variants of a cyanobacterial Rubisco that showed increased CO2 fixation activity compared with the starting situation. Bacteria overexpressing this enzyme grew significantly faster, but still required a CO2-enriched atmosphere. Optimization of this strategy might allow screening for improved catalytic features of Rubisco at very high throughput. However, currently it is not possible to use this system to study higher plant Rubiscos because of their improper folding and assembly in bacteria (24,25).

Over and above trying to identify the most efficient Rubisco enzyme, an additional major problem is posed in trying to transfer a foreign Rubisco to plants. The functional enzyme of higher plants is made up from each 8 molecules of a plastome-encoded large subunit (LSU) and a nuclear-encoded small subunit (SSU). Most plants contain multiple copies of SSU genes that are very highly expressed to keep up with LSU synthesis from the approximately 10 000 plastome copies in a plant cell (26). Homologous recombination during plastid transformation has been used to replace the LSU gene in tobacco with other LSU and/or SSU genes but the results from earlier studies indicated major difficulties. The transfer of a bacterial Rubisco that forms an enzyme as a LSU dimer and does not require SSUs resulted in the assembly of a functional enzyme (27). However, this enzyme did not promote growth under normal CO2 concentrations because its specificity for and affinity to CO2 is very low compared with higher plant enzymes. Conversely, LSUs and SSUs from algal enzymes with suitable catalytic properties did not assemble into functional complexes when they were simultaneously expressed from the plastid genome (28). Replacement of tobacco LSU with the homologous polypeptide from sunflower did not result in transplastomic plants that could be grown autotrophically (29).

Two recent publications raise new hopes that genetic engineering of Rubisco is now feasible. Sharwood et al. analyzed the tobacco lines expressing the sunflower LSU in more detail and showed that functional Rubisco enzymes were in principle assembled from the transgenic LSU and the endogenous SSU (30). The analyses revealed that the lack of autotrophic growth of these lines was exclusively caused by insufficient amounts of Rubisco during the seedling stage. This was either due to low expression levels of the sunflower LSU in the tobacco chloroplast or an inefficient assembly of the subunits. Both factors might actually be connected to each other as low efficiency of Rubisco assembly negatively feeds back on LSU protein translation (31). After CO2 supplementation of young plants, transplastomic lines could be grown to maturity without further support. In the second study, LSU and SSU subunits from the cyanobacterium Synechococcus connected with flexible linkers assembled into functional complexes at least when expressed in bacteria (32). If such an synthetic enzyme would be used to replace the endogenous LSU in the chloroplast by homologous recombination and the nuclear SSU genes would be suppressed by antisense technology (33) or RNA interference, a complete exchange of Rubisco for an improved variant might be possible at least in the model plant tobacco. For this species, even a master line prepared for highly efficient recombination of foreign Rubisco genes into the plastid genome is now available (34). A major future challenge will be to establish the chloroplast transformation technology for more species including crops.

Rubisco Activation State

Rubisco is activated by carbamylation of a lysine residue and subsequent binding of a Mg2+ ion. This process is facilitated by the enzyme Rubisco activase (RCA) most probably because it removes RuBP or other inhibitory sugar phosphates from the active site of noncarbamylated Rubisco molecules (35). This prevents the formation of noncatalytic substrate–enzyme complexes. When temperature increases, Rubisco becomes progressively inactivated and this is counteracted for by RCA. However, RCA is also inactivated at higher temperatures and associates with thylakoid membranes at temperatures above 40°C (36). This membrane association might be reversible because rice plants expressing a RCA antisense construct can partially compensate for reduced RCA levels by recruiting more of the thylakoid-bound RCA to the stroma (37). Still, transgenic Arabidopsis lines with reduced RCA levels are much more sensitive to heat stress (38) and RCA activity limits the photosynthetic potential of cotton and tobacco at high temperatures (39).

In a recent publication, Kurek et al. complemented an Arabidopsis RCA mutant with either wild-type RCA or more thermostable variants that had been generated by gene shuffling (40). This strategy is advantageous over the use of thermostable enzymes from foreign species because interactions of RCA and Rubisco show species-dependent variations (41). Plants were grown at 22°C with an increment to 30°C for 4 h in the middle of the day to induce moderate heat stress that does not directly impact on photosynthetic electron transport. Under these conditions, the plants expressing thermostable variants of RCA showed 30% higher photosynthesis, enhanced vegetative growth and produced up to four-fold more siliques and seeds. Also at a constant elevated temperature (26°C), seed production was clearly higher and the produced seeds germinated much better than those produced from plants expressing the wild-type RCA. The improvements reported might be strongest for reproductive tissues because plants were grown under long-day conditions with high light intensities where the vegetative growth phase of Arabidopsis is very short. The results indicate that engineering of RCA for increased stability is an efficient strategy to augment photosynthesis at high temperatures.


The products of the oxygenase activity of Rubisco are regenerated in the photorespiratory pathway. This pathway includes the mitochondrial conversion of two molecules of glycine to one molecule of serine, where inorganic CO2 and NH3 are released from organic compounds (see black pathway in Fig. 2). This represents a net loss of fixed compounds for the plant.

Figure 2.

 Overview of photorespiration and transgenic approaches aiming to reduce photorespiratory losses. Co-factors are only shown for transgenic pathways. The chloroplastic photorespiratory bypass proposed by Kebeish et al. (50) is shown in red, the peroxisomal pathway by Parry et al. (21) in green, and the complete oxidation of glycolate in the chloroplast (V. Maurino and U.-I. Flügge, personal communication) is shown in blue. RuBP, Ribulose-1,5-bisphosphate; RUBISCO, RuBP carboxylase/oxygenase; PGP, Phosphogycolate phosphatase; GOX, Glycolate oxidase; GGAT, Glyoxylate glutamate aminotransferase; SGAT: Serine glyoxylate aminotransferase; GDC/SHMT: Glycine decarboxylase/serine hydroxymethyl transferase; HPR, Hydroxypyruvate reductase; GK, Glycerate kinase. GS, Glutamine synthetase; GOGAT, Glutamate oxoglutarate aminotransferase; GDH, Glycolate dehydrogenase; GCL, Glyoxylate carboxyligase; TSR, Tartronic semialdehyde reductase; CAT, Catalase; MS, Malate synthase; ME, Malic enzyme; PDH, Pyruvate dehydrogenase; AcCoA, Acetylated coenzyme A; CoA, Coenzyme A; HYI, Hydroxypyruvate isomerase.

C4 plants reduced these losses by evolving a biochemical pump that concentrates CO2 in the vicinity of Rubisco and by this strongly suppresses the oxygenation of RuBP. This requires both anatomical and physiological adaptations. With few exceptions (42), C4 plants separate primary and secondary CO2 fixation in two leaf tissues, the mesophyll and the bundle sheath. Primary CO2 fixation is catalyzed by the oxygen-insensitive enzyme phosphoenolpyruvate carboxylase (PEPC) in the mesophyll. The resulting organic acids (containing 4 C atoms, thus C4 plants) diffuse into the bundle sheath where they are decarboxylated. Rubisco is restricted to the bundle sheath and thus active in a high CO2 environment (43). Attempts to reduce photorespiration by overexpression of C4 enzymes in C3 plants without spatial separation into two tissues showed limited success so far. Overexpression of maize PEPC in rice resulted in a slight reduction in the oxygen inhibition of photosynthesis (44). This effect was later assigned to reduced apparent photosynthesis at low oxygen rather than increased photosynthesis at ambient oxygen concentrations (45). Combination of PEPC with the decarboxylase malic enzyme in potato decreased the electron requirement for CO2 assimilation at higher temperatures (46). This parameter is considered as an indicator for a partial reduction of photorespiration, but transgenic plants did not show enhanced growth or higher photosynthetic activity. It therefore seems that an efficient transfer of C4 photosynthesis to C3 plants will require the cellular separation of primary and secondary CO2 fixation. This question is now being re-addressed by the International Rice Research Institute and a consortium of researchers has been assembled with the aim to introduce a true C4 cycle into the important C3 crop rice (47,48). Success of this approach will strongly depend on whether we understand how the establishment C4-type leaf anatomy is controlled in maize and other C4 plants. On the other hand, the single cell system has never been finally tested by establishment of the complete C4 biochemistry including enzymes and metabolite transporters in a C3 mesophyll cell. Such a probably less effective CO2 pump might still provide selective advantages under conditions of drought and heat where leaf stomata are closed and little CO2 is available for photosynthesis (49).

Recently, an alternative strategy for the reduction of photorespiratory losses has been established by us and colleagues (50). The strategy is not based on the reduction of Rubisco’s oxygenase activity, but rather on a more efficient recycling of the products of this reaction by installation of the E. coli“glycerate pathway” in the chloroplast (see red pathway in Fig. 2). Both photorespiration and the glycerate pathway convert glycolate to glycerate (51), but the latter pathway does not include a NH3 release step and CO2 release is shifted from mitochondria to the chloroplast. The products of oxygenation thus enhance the plastidal CO2 concentration and reduce the probability of a second oxygenation reaction. Compared with a C4 cycle, such a mechanism does not require any intercellular or subcellular metabolite transport and does not consume energy but instead produces reducing equivalents in the chloroplast. Interestingly, a pathway homologous to the E. coli glycerate pathway represents one of at least two photorespiratory pathways in the cyanobacterium Synechocystis (52). Biochemical and biophysical assays of transgenic Arabidopsis lines overexpressing all necessary enzymes revealed reduced metabolite flow through photorespiration, enhanced carbon assimilation and faster growth of transgenic lines under short-day growth conditions (22°C, 100 μE light). The growth effect was reproducible at higher temperatures or higher light intensities.

Unexpectedly, overexpression of glycolate dehydrogenase (GDH)—the first enzyme of the glycerate pathway—alone was sufficient to obtain some of the observed positive effects suggesting that chloroplasts also contain a natural system for the further conversion of the reaction product glyoxylate. This hypothesis is supported by earlier studies indicating that chloroplasts are capable of oxidizing glyoxylate to CO2 (53,54). Feeding photorespiratory metabolites into this endogenous pathway by GDH overexpression can simplify the metabolic engineering approach. This is important especially for crop plants where transfer of several genes into one line is still difficult and unfavored by breeders. Still, E. coli GDH is a heterotrimeric protein complex and, thus, three transgenes would have to be overexpressed. The approach might be further facilitated by the use of single-subunit GDH enzymes that have been recently identified in both algae (52,55) and higher plants (56,57). Plastidal glycolate oxdidation therefore provides a promising strategy to enhance crop productivity.

However, it remained unclear from this study whether the reduced NH3 release, the shift of CO2 release from the mitochondrium to the chloroplast, or the improved energy balance compared with photorespiration causes the positive growth effects. Two related approaches that also manipulate plant glycolate metabolism might help to answer this open question. Parry et al. proposed a short-circuit of photorespiration in the peroxisome [(21), green pathway in Fig. 2]. In such an approach, exclusively mitochondrial NH3 release would be reduced, but CO2 release would still take place outside the chloroplast. Unfortunately, tobacco plants overexpressing the relevant enzymes developed chlorotic lesions on their leaves when being grown at ambient CO2. The reasons for this phenotype are unclear and further investigation is necessary to understand whether this observation is related to interference with photorespiration.

An alternative approach was tested by the group of Veronica Maurino and Ulf-Ingo Flügge (Cologne University, Germany, personal communication): In this approach, glycolate is completely oxidized to CO2 in the chloroplast (blue pathway in Fig. 2). Instead of the bacterial glycolate dehydrogenase using organic co-factors, the authors used the plant glycolate oxidase that produces H2O2 and co-expressed a catalase to detoxify this compound. The resulting glyoxylate was then converted to malate by malate synthase. Endogenous enzymes of the chloroplast (NADP-malic enzyme and pyruvate dehydrogenase) might metabolize malate releasing CO2 inside the chloroplasts. Some of the transgenic Arabidopsis plants overexpressing plastidal glycolate oxidase together with catalase and malate synthase showed positive effects on photosynthesis and growth comparable to those obtained by overexpression of the glycerate pathway. This provides further evidence that diverting some of the glycolate from photorespiration inside the chloroplast may enhance plant productivity.


Flux control analyses in antisense plants revealed a high control coefficient for the activity of sedoheptulose-1,7-bisphosphatase (SBPase) - one of the two enzymes in the PCRC generating C5 precursors for the synthesis of the CO2 acceptor molecule RuBP - on PCRC activity and CO2 fixation in tobacco (58,59). Consequently, overexpression of a combined SBPase/fructose-1,6-bisphosphatase (FBPase) enzyme from cyanobacteria in tobacco resulted in improved biomass production and higher leaf sugar contents (60). Effects of this manipulation on photosynthesis were most pronounced at high light. These results have now been reproduced in plastome-transformed tobacco plants overexpressing the same enzyme (61). Again, plants grew faster and showed enhanced sugar contents. Interestingly, phenotypic and biochemical data were almost identical for lines showing 2-3-fold and 50-fold overexpression, respectively. This indicates that photosynthesis is saturated with slightly higher SBPase/FBPase levels and that another restriction limits photosynthesis and growth in these transgenic lines.

Better growth and improved photosynthesis were also observed in tobacco overexpressing an Arabidopsis SBPase (62) suggesting that the SBPase activity of the combined cyanobacterial enzyme is the main reason for the positive growth effects reported in the earlier study. Moreover, the Arabidopsis enzyme shows post-translational regulation in response to product accumulation and the redox state of the chloroplast. The lack of these regulatory features in the cyanobacterial enzyme is thus not responsible for enhanced growth, but rather the higher amount of enzyme produced in transgenic lines.

These data are in line with the recently published direct comparison of a Synechococcus FBPase (without the SBPase function) and a Chlamydomonas SBPase in tobacco (63). The overexpression of both enzymes positively influenced RuBP levels, growth, and photoynthesis. However, assessment of individual transgenic lines with different expression levels indicated that SBPase exerts the main control over RuBP regeneration whereas FBPase is primarily limiting the partitioning of carbon into starch.

Interestingly, several of these studies reported a higher in vivo activation state of Rubisco in SBPase overexpressing lines. It was speculated that this is due to enhanced supply of RuBP inducing the formation of more complexes with noncarbamylated Rubisco that are substrate for RCA (60). This effect has now been re-addressed by Feng et al. in a study with rice plants overexpressing the endogenous SBPase enzyme (64). The authors show that sensitivity of photosynthesis to heat stress is reduced and that recovery from heat stress is enhanced in transgenic lines. This correlates with a reduced aggregation (and thus inactivation) of RCA at high temperatures, although the molecular mechanisms underlying this phenomenon are unclear. Accordingly, the rice SBPase overexpressors showed higher biomass productivity under stress conditions. It should be noted, however, that transgenic lines grew comparable to control lines under standard conditions although SBPase levels were in the range of those reported for transgenic tobacco lines that also showed enhanced growth in the absence of heat stress. This points to species-specific peculiarities in the importance of SBPase levels for RuBP regeneration.


It has long been questioned whether transgenic approaches towards the enhancement of photosynthesis would ever result in increased productivity of crop plants. Together, the papers reviewed here provide strong evidence that we acquired sufficient knowledge about photosynthesis and carbon metabolism to identify and target important bottlenecks. Some of the described approaches resulted in increased biomass production of transgenic lines or even augmented seed yields. However, it is now important to test such methods as soon as possible in true crops since we are ignorant about the transferability of such knowledge from model systems in phytochambers to crops under conditions of agricultural production. Rapeseed and rice are important first candidates because these species are amenable for efficient genetic manipulation and represent major groups of crop plants.

Metabolite flux through a pathway is often limited by more than one bottleneck. By overcoming one restriction in the complex photosynthetic network, productivity can only be enhanced up to the point where the next limitation applies. Some of the approaches described in this review target different aspects of photosynthesis and might therefore show additive or even synergistic effects when combined in one plant. To our knowledge, this has not been experimentally tested so far. However, “gene pyramiding” in crops will definitely require the development of more efficient transformation technologies that allow the simultaneous transfer of many (probably dozens of) genes in a reasonable time frame. As particle bombardment allows the transfer of big DNA molecules [e.g. (65,66)], assembly of the required sequences on one DNA molecule might be the main limiting step. The rapid advance of gene synthesis technologies (67) or techniques such as MultiRound Gateway (68) will facilitate the generation of such large artificial DNA sequences and thus open the door for more ambitious genetic engineering approaches towards crop improvement.

Acknowledgements— Metabolic engineering projects in our laboratory are supported by the German Ministry for Education and Research and Bayer Bioscience. Thanks to Veronica Maurino and Ulf-Ingo Flügge for making data available prior to publication.