An important goal in biotechnological research is to improve the yield of crop plants. Here, we genetically modified simultaneously source and sink capacities in potato (Solanum tuberosum cv. Desirée) plants to improve starch yield. Source capacity was increased by mesophyll-specific overexpression of a pyrophosphatase or, alternatively, by antisense expression of the ADP-glucose pyrophosphorylase in leaves. Both approaches make use of re-routing photoassimilates to sink organs at the expense of leaf starch accumulation. Simultaneous increase in sink capacity was accomplished by overexpression of two plastidic metabolite translocators, that is, a glucose 6-phosphate/phosphate translocator and an adenylate translocator in tubers. Employing such a ‘pull’ approach, we have previously shown that potato starch content and yield can be increased when sink strength is elevated. In the current biotechnological approach, we successfully enhanced source and sink capacities by a combination of ‘pull’ and ‘push’ approaches using two different attempts. A doubling in tuber starch yield was achieved. This successful approach might be transferable to other crop plants in the future.
Crop yield improvement is critical to feed the constantly growing world population (United Nations, 2011) especially as arable acreage becomes increasingly scarce. Compounding this problem, more acreage will be used for the production of biofuels at the expense of food and feed (FAO, 2010) because of increasing demand, shortage of fossil fuel and global warming (FAO, 2009; IPCC, 2007; OECD/FAO, 2011). Potato represents an important crop since tubers contain large amounts of nutritious storage compounds, the most important of which is starch. Besides its importance as staple food, tuber starch is also used for several industrial applications, for example, to produce adhesives or to coat high-quality paper (www.potato2008.org). Taken together, it appears a relevant biotechnological issue to increase starch content and yield of potato tubers.
Apart from its biotechnological relevance, because of their strong sink organs, potato is a suitable model plant to study source–sink interactions. Source tissues are capable of net exporting energy, whereas sink tissues depend on the net import of energy from source tissues (Turgeon, 1989) with energy being photosynthesis products, that is, sucrose. The terms ‘source’ and ‘sink’ are assigned relating to the respective plant organ rather than cell type. However, other definitions like in the study of Sweetlove et al. (1998) who assign phloem loading a sink-specific process are equally meaningful.
In a previous transgenic approach, we were able to demonstrate that potato tuber starch content and yield can be increased when sink strength is elevated (Zhang et al., 2008). To achieve this, two plastidic metabolite transporters were simultaneously overexpressed in tubers, the pea glucose 6-phosphate/phosphate translocator (GPT; Kammerer et al., 1998) and the Arabidopsis adenylate translocator (NTT; Kampfenkel et al., 1995). These transporters supply amyloplasts with carbon skeletons in form of glucose 6-phosphate (G6P) and energy as ATP, respectively. The supply of amyloplasts with G6P and ATP is essential for starch synthesis as they are precursors for ADP-glucose pyrophosphorylase (AGPase), the key enzyme for starch synthesis that catalyses the formation of ADP-glucose, the precursor of starch. GPTs form a subfamily of the phosphate translocator (PT) family (Fischer and Weber, 2002). Other subfamilies characterized to date are the triose phosphate/phosphate translocators (TPTs; Flügge, 1999; Flügge et al., 1989), the phosphoenolpyruvate/phosphate translocators (PEPs; Fischer et al., 1997) and the xylulose-5-phosphate/phosphate translocators (Eicks et al., 2002). The PT family belongs to the nucleotide sugar transporter/triose phosphate/phosphate translocator (TPT) superfamily (Knappe et al., 2003). Members of the PT family mediate an antiport of phosphorylated intermediates with inorganic phosphate (Pi) across the inner envelope membrane of plastids. In addition to G6P and Pi, triose phosphates and, to a lesser extent, 3-phosphoglycerate (3-PGA) serve as transport substrates for GPTs (Kammerer et al., 1998). The importance of GPT function in heterotrophic tissues was shown in Arabidopsis thaliana and Vicia narbonensis. In Arabidopsis, a knockout of GPT1 resulted in lethality of gametophytes, most likely caused by a restriction of the oxidative pentose phosphate pathway (OPPP) because of limited G6P supply. Pollen grains of mutants did not contain any starch granules and contained less lipid bodies than wild-type pollen (Niewiadomski et al., 2005). In seeds of V. narbonensis antisense GPT plants, starch content was reduced and storage protein biosynthesis increased (Rolletschek et al., 2007).
NTTs represent a group of ATP/ADP transporters existing in all higher and lower plants examined to date (Linka et al., 2003). They reside in the inner plastid envelope membrane (Neuhaus et al., 1997) and show no structural similarities to mitochondrial ADP/ATP carriers (AACs) which are located in the inner membrane of mitochondria (Winkler and Neuhaus, 1999). NTTs have been described to catalyse the counter-exchange of ATP and ADP across the inner envelope membrane of plastids (Heldt, 1969; Schünemann et al., 1993). In a more recent study, it was shown that ADP plus Pi rather than solely ADP are co-exported from plastids in exchange for imported ATP (Trentmann et al., 2008). The importance of ATP supply to amyloplasts was demonstrated in NTT antisense potato plants. Tuber morphology was altered, and tuber starch content and yield was decreased in these plants (Tjaden et al., 1998). Arabidopsis double knockout mutants of the two present NTTs generally show a retarded plant development and a reduced plant growth rate under short day conditions (Reiser et al., 2004). Moreover, Reinhold et al. (2007) showed that NTTs are important for nocturnal metabolism of chloroplasts. Knockout mutants suffer from photooxidative damage due to the accumulation of the phototoxic chlorophyll precursor protoporphyrin IX induced by shortage of stromal ATP.
It was shown that GPT and NTT co-limit starch content and yield of potato tubers as the simultaneous tuber-specific overexpression using the B33 promoter (Rocha-Sosa et al., 1989) of both transporters resulted in an increased tuber yield of up to 19%, an increased tuber starch content of up to 28% and an increased total starch content produced in tubers per plant of up to 44% in double transgenic so-called BGA plants compared to wild type (Figure 1, lower panel; Zhang et al., 2008). Thus, starch formation within amyloplasts is co-limited by the import of carbon skeletons and energy in potato, whereas an exclusive overexpression of GPT or NTT had no impact (Zhang et al., 2008).
Tuber starch content and yield of BGA plants could still be sink-limited—when sink strength had been increased to less than a level that meets the source capacity—or, conversely, could be source-limited. In the latter case, enhancing source capacity in the background of BGA plants could lead to further increased tuber starch content and yield.
To clarify whether a simultaneous boosting of source and sink capacities is suitable to further increase tuber starch content and yield, source capacity was altered additionally in the above-mentioned double-transgenic BGA line background here. Two approaches were applied to enhance source capacity, namely overexpression of an Escherichia coli pyrophosphatase (PPase) in mesophyll cells using the cytosolic fructose-1,6-bisphosphatase (cyFBPase) promoter (Ebneth, 1996; patents EP0938569, US6229067) or moderate down-regulation of AGPase activity in leaves mediated by antisense expression of the catalytic B subunit of the AGPase under control of the leaf-specific StLS1 promoter (Stockhaus et al., 1989). Both approaches make use of re-routing photoassimilates to tubers at the expense of leaf starch accumulation (Figure 1, upper panel).
Soluble PPases dissipate inorganic pyrophosphate (PPi) to Pi. Most of the PPase activity resides in plastids as has been exemplified in spinach leaves. Accordingly, plastidic PPi content is low compared with the higher PPi content in the cytosol (Weiner et al., 1987). In a recent study, virus-induced gene silencing was applied to specifically down-regulate a plastidic soluble PPase of Nicotiana benthamiana. In the resulting lines, total soluble PPase activity was reduced by 90%, once more demonstrating the majority of PPase activity to reside in plastids (George et al., 2010). Overexpression of a soluble PPase in the cytosol of leaf cells is intended to enhance sucrose synthesis (Figure 1, upper panel). In the metabolic pathway of sucrose biosynthesis, there are two reactions that yield PPi in the direction of sucrose formation, namely the reactions of PPi-dependent phosphofructokinase (PFP) and of UDP-glucose pyrophosphorylase (UGPase). Upon removal of PPi, a putative back reaction would be hampered. Moreover Pi, the product of the PPase reaction, could be used as counter-exchange substrate for the triose phosphate/phosphate translocator (TPT) and thus lead to an improved supply of the cytosol with triose phosphates, initial substrates for sucrose synthesis. Constitutive overexpression of the E. coli PPase in tobacco (Nicotiana tabacum cv. Samsun NN) led to plants with sugar-storing leaves (Jelitto et al., 1992; Sonnewald, 1992) because of a block in sucrose phloem loading brought about by inability of companion cells to break down sufficient sucrose taken up by the sucrose/H+ symporter (Riesmeier et al., 1992, 1993) to yield energy because of missing PPi for the upstream UGPase reaction (Lerchl et al., 1995). Energy is essential to drive a proton pumping ATPase establishing the proton gradient for sucrose uptake into the phloem against a steep sucrose gradient. Therefore, a mesophyll-specific promoter (Ebneth, 1996; patents EP0938569, US6229067) was used to prevent PPase expression in companion cells and to generate plants that are not impaired in sucrose phloem loading.
The formation of the starch precursor ADP-glucose from glucose 1-phosphate and ATP by AGPase exclusively occurs in plastids, and the importance of AGPase for starch synthesis has been shown by analysing potato antisense plants. Expression of an antisense construct against the gene encoding the catalytic B subunit of the AGPase under control of the constitutive 35S CaMV promoter in potato plants led to an up to 10-fold decrease in leaf starch content and to a reduction of tuber starch of 96%. Tuber sucrose and glucose contents were strongly increased, tuber number was increased and tuber size decreased (Müller-Röber et al., 1992). To produce high starch tubers, the mutated E. coli AGPase encoding gene glgC16 which is not subject to allosteric regulation, was overexpressed in potato (Stark et al., 1992) and the resulting lines showed an increase in tuber starch content. These results could not be corroborated using a different potato cultivar; an increase in starch synthesis but also degradation was described, that is, an enhanced tuber starch turnover (Sweetlove et al., 1996a,b). Transgenic potato plants expressing a leaf-specific AGPase antisense construct that led to a strong down-regulation of residual AGPase activity had been described to accumulate less leaf starch than the wild type with no effect on photosynthesis or on tuber yield. In these plants, photoassimilate export from leaves was increased during the day and decreased during the night (Leidreiter et al., 1995).
Results and discussion
To investigate whether enhancing source capacity by mesophyll-specific PPase overexpression or leaf-specific AGPase repression in the background of plants exhibiting increased sink strength has consequences on tuber yield, starch content and total tuber starch yield, triple-transgenic potato plants named AGN and PGN were generated and analysed (Figure 1). It is important to mention that aerial parts of the triple-transgenic lines lacked any particular phenotype and appeared to be indistinguishable from the wild type.
Molecular analyses of PGN and AGN plants
Potato plants (Solanum tuberosum cv. Desirée) harbouring the B33::PsGPT (line BG1) construct (Zhang et al., 2008) were co-transformed with a mixture of Agrobacterium tumefaciens cultures carrying a B33::NTT1 construct and a cyFBPase::EcPPase construct. Three out of several PGN lines were analysed for transcript levels of PsGPT and AtNTT1 using RT-PCR. Both transcripts were detected in tubers of PGN lines and were absent in wild-type tubers (Figure 2a). Furthermore, transcript levels of EcPPase were found to be abundant in leaves of PGN lines and were not in wild-type leaves (Figure 2b). Accordingly, when E. coli PPase activity was assayed using an in-gel activity technique, a band was stained in triple-transgenic extracts and E. coli extracts as positive control, and not in wild-type extracts (Figure 2c). To quantify PPase activity, a photometric assay was chosen which does not only record E. coli PPase activity, as it is the case for the in-gel assay, but also endogenous, mostly plastidic activities. Assuming similar endogenous PPase activities in PGN and wild-type plants, the observed 50%–80% increase of total PPase activity in PGN plants reflects the E. coli PPase activity (Figure 2d). As control, leaf AGPase activities were measured and found to be similar in PGN and wild-type plants (Figure 2e).
AGN plants were generated by co-transforming the line BG1 (Zhang et al., 2008) with a mixture of A. tumefaciens cultures carrying a B33::AtNTT1 and an StLS1::αStAGPase construct. Three out of several AGN lines were analysed for the presence and expression of both constructs. Transcript levels of PsGPT and AtNTT1 were analysed in tubers of the resulting AGN plants using RT-PCR. Both transcripts were abundant in tubers of AGN plants and absent in tubers of wild-type plants (Figure 3a). Transcript levels of the catalytic B subunit of the StAGPase were reduced in leaves of AGN plants compared with the wild type (Figure 3b). This finding was generally corroborated using quantitative real time PCR (Figure 3c). Residual leaf AGPase activity was 60%–70% in AGN compared with wild-type plants (Figure 3d). An only moderate down-regulation of AGPase activity was intended in order not to disturb sucrose supply of tubers during night when breakdown of leaf starch fuels sucrose synthesis. A stronger reduction in AGPase activity has been shown to cause lower leaf starch contents what, in turn, affected nocturnal export of sucrose to tubers (Leidreiter et al., 1995). In general, potato leaves only partially break down starch during the night phase to fuel nocturnal sucrose supply of sinks, that is, potato leaf starch is less transitory than Arabidopsis leaf starch which is synthesized during the day and almost completely degraded during the night (Gibon et al., 2004; Schneider et al., 2002). In a recent study, Ferreira et al. (2010) showed that potato leaves contained about 30% of the starch at the onset of light compared to leaf starch content at the end of the light period.
Leaf starch content is decreased in PGN and AGN plants
To assess the impact of EcPPase overexpression or StAGPase repression in leaves on carbon metabolism, leaf starch content was analysed in triple-transgenic potato plants in comparison with the wild type and with the double-transgenic BGA31 line (Zhang et al., 2008). Mature leaves of tuberizing triple-transgenic plants contained about 50%–70% starch compared to wild-type and BGA31 leaves (Figure 4a), that is, restricting leaf starch accumulation or increasing sucrose synthesis seems to allocate photoassimilates from chloroplast stroma to the cytosol. Remarkably, in single StLS1::αAGPase transgenic plants, a residual AGPase activity of only 26% did not lead to any changes in leaf starch content (Ludewig et al., 1998) indicating an important role of the increased sink demand of triple-transgenic PGN and AGN plants in controlling leaf starch content. Furthermore, leaf glucose, fructose and sucrose contents did not differ between triple-transgenic, BGA31 and wild-type plants (Figure 4b–d), indicating surplus sucrose to be exported from leaves.
In addition, assimilation rates were analysed in AGN lines in comparison with the wild type. During a light curve experiment, assimilation increased with increasing photon flux density heading for light saturation of assimilation. However, no significant differences between triple-transgenic AGN plants and the wild type were obvious (Figure 5).
Tuber yield and starch content are strongly increased in PGN and AGN plants
To analyse the fate of photoassimilates ‘missing’ in leaves of triple-transgenic plants, that is, the difference in starch content of wild-type and PGN and AGN leaves, tuber starch content and yield of PGN, AGN and wild-type plants were analysed. As desired, tuber yield was increased. The observed increase was unambiguous with up to 56% and 57% in PGN and AGN plants, respectively, compared to wild type (Figure 6a,d), and tuber starch content was increased up to 30% and 35% (Figure 6b,e). This resulted in an overall increase in starch yield (tuber starch per plant) of up to 103% and 112% in PGN and AGN plants, respectively, compared to wild type (Figure 6c,f), demonstrating the success of both approaches (refer also to Tables S1 and S2). These findings strongly suggest that advancing the source capacity by restricting leaf starch accumulation in the background of plants exhibiting increased sink strength, further enhances tuber starch content and yield as compared to double-transgenic BGA plants with just increased sink strength. This further increase could be traced back mainly to an enhanced tuber yield rather than starch content.
A straightforward way to explain increased tuber yield and starch content (Figure 6) without an obvious increase in photosynthesis (Figure 5) would be to predict an increased photosynthate transport from source to sink organs, especially when the pool of leaf transitory starch was shown to be lower in triple-transgenic compared to wild-type plants (Figure 4a). Thus, leaf export experiments were performed (King and Zeevaart, 1974; Riesmeier et al., 1994; Wingenter et al., 2010) in which total sugar export exuded from detached leaf petioles was determined (Figure 7). Control BGA31 plants showed a (nonsignificant) trend for increased photosynthate transport to tubers, compared to wild type, most likely due to an increased sink demand induced by possible adaptive processes. This would explain why a plant without direct modifications in leaves could export more sugar from leaves when the actual sink is absent or substituted by a water-containing cup. However, irrespective of the presence of the proposed adaptive process, all analysed PGN and AGN lines clearly transported more sugar per time than wild-type and also BGA31 plants. This finding is in line with the increased tuber yield and starch content found in PGN and AGN plant lines (Figure 6).
To achieve near-field conditions, AGN and wild-type plants were grown in beds, that is, without possible pot limitation, in a greenhouse without additional lightening lacking climate control. The results obtained were similar to the results described earlier for controlled conditions indicating that enhanced yield seems not to limit a flexible response to sub-optimal environmental conditions. Tuber yield was increased up to 89% (Figure 8a), tuber starch content up to 27% (Figure 8b) and tuber starch per plant up to 150% (Figure 8c).
By analysing plants that solely overexpress an E. coli PPase under control of the cyFBPase promoter (Ebneth, 1996; patents EP0938569, US6229067) without simultaneously increased sink capacity, we ruled out the possibility that this sole overexpression could be sufficient to increase tuber starch content and yield. Five lines of so-called ME plants grown in greenhouse and field experiments were analysed. PPase activity was increased in the same range as in PGN plants (Figure 9a), and leaf starch content was slightly decreased (Figure 9b). In contrast to PGN plants, leaf soluble sugar contents were increased (Figure 9c) indicating limited stimulation of sucrose export to tubers in these plants. Accordingly, the ratio of leaf soluble sugars to starch was increased in ME plants (Figure 9d). Consistently, tuber yield as well as starch content, expressed as % tuber dry weight of fresh weight, was not increased in ME plants (Figure 9e,f).
In a similar ‘push’ approach, single-transgenic plants with antisense expression of the catalytic B subunit of AGPase under control of the leaf-specific StLS1 promoter (Stockhaus et al., 1989) had been analysed. However, despite leaf starch content was lower than in wild type, no effect on photosynthesis or on tuber yield could be observed (Leidreiter et al., 1995) indicating that ‘push’ approaches alone are not appropriate to increase tuber starch content and yield of potato.
Additionally, tubers of AGN plants grown under field-like conditions in a bed (compare Figure 8) were analysed for their tuber starch composition. Amylose proportion was significantly increased in tubers of AGN plants in comparison with the wild type, reaching approximately 30% amylose content versus approximately 21% of wild-type tuber starch (Figure 10). This increased amylose content confirms previous data of tuber starch of double-transgenic BGA plants. Amylose contents were increased but to a lesser extent than in tuber starch of AGN plants (Zhang et al., 2008). This could be explained with an improved availability of ADP-glucose for starch synthases, in particular, granule-bound starch synthases (GBSSs) for amylose synthesis (Visser et al., 1991). GBSS isoforms have a lower affinity for ADP-glucose (Smith et al., 1997) than soluble starch synthases (which are involved in amylopectin synthesis), that is, an increase in the amylose-to-amylopectin ratio can be explained by an increased ADP-glucose availability (Tjaden et al., 1998), presumably due to an increased flux into ADP-glucose and further into starch. ADP-glucose contents of BGA tubers, however, were shown to be similar to that of wild-type tubers (Zhang et al., 2008).
To our knowledge, the increases in tuber yield and tuber starch we report here together with the ones obtained by altering tuber adenylate pools (Regierer et al., 2002) are the highest published to date although other approaches to enhance tuber starch yield in transgenic potato plants were also successful, for example, by stimulating the pyrimidine salvage pathway by inhibiting the de novo synthesis of pyrimidines (Geigenberger et al., 2005) or by overexpression of a sucrose synthase in tubers (Baroja-Fernandez et al., 2009).
Taken together, using potato as an example, the simultaneous increase in source and sink capacities led to a substantial increase in crop yield. This general approach to modify carbon allocation within a plant might be transferred to other crop plants to enhance yield in the future.
Construction of B33::AtNTT1, B33::PsGPT, cyFBPase::EcPPase and StLS1::αStAGPase
Generation of the B33::AtNTT1 and B33::PsGPT constructs had been described earlier (Zhang et al., 2008). To generate the cyFBPase::EcPPase construct, base pairs 291–822 of the 1195 bp-long E. coli inorganic pyrophosphatase (EcPPase) gene, flanked by an NcoI and a SalI restriction site, were cloned behind the 5′-untranslated region of the tobacco mosaic virus gene U1 (Gallie et al., 1987) flanked with an Asp718I and an NcoI restriction site between the 1723 bp-long potato FBP3 gene and the 190 bp-long ocs terminator. The mesophyll-specific cyFBPase promoter is covered by patents (EP0938569, US6229067). To generate the StLS1::αStAGPase construct, a 535 bp-long part of the catalytic subunit of the S. tuberosum ADP-glucose pyrophosphorylase (StAGPase) gene, flanked by BamHI restricition sites, was cloned in antisense direction into the pBin19-StLS1, which had been linearized with BamHI. Resulting constructs were transformed into A. tumefaciens (GV2260) which were used to transform plants.
Generation of PGN and AGN plants
Previously described (Zhang et al., 2008) hygromycin-resistant plants (S. tuberosum cv. Desirée) over expressing the pea GPT under control of the tuber-specific B33 promoter (Rocha-Sosa et al., 1989) were co-transformed with a mixture of A. tumefaciens cultures either carrying a B33::NTT1 construct conferring kanamycin resistance and tuber-specific expression of Arabidopsis NTT1 or a second construct also conferring kanamycin resistance. The respective second constructs were used (i) to overexpress an E. coli PPase only in mesophyll cells and, most importantly, not in companion cells to prevent impaired sucrose phloem loading mediated by the cyFBPase promoter (patents EP0938569, US6229067) or (ii) to down-regulate the expression of AGPase in leaves mediated by antisense expression of the catalytic B subunit under control of the leaf-specific StLS1 promoter (Stockhaus et al., 1989). Resulting triple-transgenic plants were termed PGN and AGN, respectively.
Potato plants were transferred from sterile culture to soil and grown in a greenhouse from spring to autumn under constant long-day conditions with additional lightening (16 h light/8 h dark) in 3.5 L pots for the indicated time periods and were watered daily. The temperature was kept at 21 °C during the light period and 18 °C during the dark period. Potato plants were also grown under field-like conditions in a bed without pot limitation in a greenhouse without additional lightening that lacks humidity and temperature control. For that shoots from sterile culture were transferred to soil, pre-cultivated in a greenhouse under the above-mentioned conditions for 3 weeks and then transferred to the bed where they were grown for the indicated time and watered every second day.
Molecular, biochemical and physiological analyses
Isolation of total RNA was performed as described (Logemann et al., 1987). DNaseI restriction of the samples was performed using the DNAase-free kit™ by Ambion (Life Technologies GmbH, Darmstadt, Germany) and reverse transcription using BioScript™ reverse transcriptase by Bioline (Luckenwalde, Germany) according to instructions of manufacturers. Expression levels of genes were determined by RT-PCR or quantitative real time PCR, the latter using the SensiMix™ SYBR green kit by Bioline. Starch and soluble sugar contents were determined as described (Müller-Röber et al., 1992). AGPase activity measurements were taken using slightly modified conditions as described (Müller-Röber et al., 1992; Tiessen et al., 2002). Activity of the inorganic pyrophosphatase was detected using a zymogram (Sonnewald, 1992) or was quantified photometrically (Jelitto et al., 1992). The amylose-to-amylopectin ratio of tuber starch was determined as described before (Hovenkamp-Hermelink et al., 1988). Assimilation rates with increasing photon flux density were determined using an infrared gas analysis system (GFS-3000; Walz, Effeltrich, Germany). Harvesting phloem exudates of potato petioles were carried out according to a slightly modified (mostly because of morphological differences of potato versus Arabidopsis leaves) protocol described by Wingenter et al. (2010). In brief, leaves were cut from the plant at the petioles using a scalpel with water attached to it, and re-cut under water. Leaves with their petioles were then put into 15 mL tubes containing 8 mL 15 mm EDTA (pH 7.25). Sugar export occurred during 4 h in the light in the greenhouse, and total sugars were determined as indicated above.
Statistical relevance of differences in data sets was analysed. For that P values were produced by Student's t-test (Welch test).
We thank Iris Schmitz and Raphael Wemhöner for excellent technical assistance and Dr. Jens-Otto Giese for his contribution analysing ME plants. This work was supported by the German Federal Ministry of Education and Research (BMBF; project 0315059). All authors have no conflict of interest to declare.