•Reserve starch is an important plant product but the actual biosynthetic process is not yet fully understood.
•Potato (Solanum tuberosum) tuber discs from various transgenic plants were used to analyse the conversion of external sugars or sugar derivatives to starch. By using in vitro assays, a direct glucosyl transfer from glucose 1-phosphate to native starch granules as mediated by recombinant plastidial phosphorylase was analysed.
•Compared with labelled glucose, glucose 6-phosphate or sucrose, tuber discs converted externally supplied [14C]glucose 1-phosphate into starch at a much higher rate. Likewise, tuber discs from transgenic lines with a strongly reduced expression of cytosolic phosphoglucomutase, phosphorylase or transglucosidase converted glucose 1-phosphate to starch with the same or even an increased rate compared with the wild-type. Similar results were obtained with transgenic potato lines possessing a strongly reduced activity of both the cytosolic and the plastidial phosphoglucomutase. Starch labelling was, however, significantly diminished in transgenic lines, with a reduced concentration of the plastidial phosphorylase isozymes.
•Two distinct paths of reserve starch biosynthesis are proposed that explain, at a biochemical level, the phenotype of several transgenic plant lines.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Almost all plants (and their heterotrophic derivatives as well) are capable of forming starch (Ball & Morell, 2003). Native starch exists as water-insoluble particles with a defined intra- and intermolecular order. Starch biosynthesis includes at least five distinct groups of reactions: the formation of a carbohydrate-like glucosyl acceptor; the synthesis of a glucosyl donor; the glucosyl transfer from the donor to the acceptor (by forming α-1,4-interglucose bonds); the branching of α-1,4-linked glucan chains (by forming α-1,6-interglucose bonds); and the physical ordering of α-glucan chains (Zeeman et al., 2002). Quantitatively most prominent are the biosynthesis of the general glucosyl donor, adenosine diphosphoglucose (ADPglucose), and the subsequent glucosyl transfer to the nonreducing end(s) of α-glucans, resulting in a massive elongation of the glucosyl acceptors.
In higher plants the entire process takes place in the stroma of chloroplasts in source organs (transitory starch) and of nongreen plastids in sink organs (reserve starch). Reserve starch biosynthesis is initiated by the import of sucrose into the sink organ but the subsequent steps are diverse. In the endosperms of graminaceous species ADPglucose is largely formed in the cytosol of starch-accumulating cells and subsequently imported into the plastid via a selective sugar nucleotide transporter (Tomlinson & Denyer, 2003). In all other heterotrophic tissues, sucrose-derived sugar compounds are thought to enter the plastid via a glucose 6-phosphate/orthophosphate antiporter (Kammerer et al., 1998). Subsequently, glucose 6-phosphate is converted, by a two-step reaction sequence, to ADPglucose, which finally acts as substrate for the glucosyl transfer to an α-glucan chain. For potato tubers this biosynthetic path is substantiated by several independently derived pieces of evidence. By using nonaqueous fractionation of wild-type tubers, most of the glucose 6-phosphate and all the detectable ADPglucose have been localized in the amyloplast compartment (Farréet al., 2001). An antisense inhibition of the plastidial isozyme of the phosphoglucomutase (pPGM) that mediates the interconversion of glucose 6-phosphate and glucose 1-phosphate (EC 184.108.40.206; PGM) significantly impedes the accumulation of reserve starch as well as the entire tuber growth (Tauberger et al., 2000). Similarly, tubers from transgenic potato lines having a reduced expression of the ADPglucose synthesizing enzyme, ADPglucose pyrophosphorylase (EC 220.127.116.11; AGPase), accumulate minute quantities of reserve starch (Müller-Röber et al., 1992). All these data underline the importance of the glucose 6-phosphate-dependent and ADPglucose-mediated path of starch biosynthesis in potato tubers. However, they do not prove that it is the only functional path by which glucose derivatives can be converted to starch. In fact, some recently published results point to a higher metabolic diversity and flexibility in heterotrophic tissues. In transgenic potato tubers that simultaneously underexpress both the cytosolic phosphoglucomutase (cPGM) and the pPGM activity, the rate of reserve starch biosynthesis approximates that of the wild-type control and is much higher than in those lines that selectively repress pPGM (Fernie et al., 2002a). Based on the path of starch biosynthesis outlined above the phenotype of the double mutants is difficult to explain. Furthermore, glucose 1-phosphate possesses an unexpected subcellular distribution when analysed by the nonaqueous fractionation technique. A significant proportion of the anomeric glucose phosphate exhibits the same distribution as markers of the vacuole. However, because of the high phosphatase activity inherent to this organelle, it has been suggested that glucose 1-phosphate is actually located in a compartment co-fractionating with the vacuole (Farréet al., 2001). Later subcellular localization studies clearly substantiated this suggestion, as a marker of the apoplastic space, crystalline cellulose, can co-distribute with vacuolar proteins (Fettke et al., 2005). Recently, a rapid uptake of glucose 1-phosphate by tuber discs has been demonstrated and the cytosolic path of the imported glucose monophosphate was found to be largely separated from that of glucose. In 14C-labelling experiments, glucose 1-phosphate was converted to starch at a rate exceeding that of neutral sugars, such as glucose or sucrose (Fettke et al., 2008).
In the present paper, this effect has been further studied and the glucose 1-phosphate-dependent intracellular carbon flux has been analysed by using various transgenic potato lines deficient in enzymes of the plant primary metabolism. The data provide clear evidence that in potato tubers a further path of forming reserve starch exists. This path can be initiated by external glucose 1-phosphate and allows glucosyl residues to be transferred to native starch granules even if the interconversion of the two glucose monophosphate pools is largely slowed down.
Materials and Methods
Potato wild-type plants (Solanum tuberosum cv. Desiree), transgenic lines expressing an antisense construct directed against the plastidial phosphorylase isozymes (Pho1a, Pho1b, Pho1a + b; for details see Fettke et al., 2005) or possessing an altered expression of the cytosolic phosphorylase isozyme (Pho2; Fettke et al., 2005), transgenic lines with a lowered expression of either the cPGM or of both the cPGM and the pPGM (for details see Lytovchenko et al., 2002; Fettke et al., 2008) and RNAi lines with a lower expression of DPE2 (Lloyd et al., 2004) were grown under controlled conditions as previously described (Fettke et al., 2005).
Incubation of potato tuber discs with sugars or sugar derivatives
Except where stated, growing tubers were removed from flowering plants and were then immediately used for the experiments. Discs (13 mm diameter, c. 2 mm thick) were repeatedly washed with a large volume of ice-cold water to remove free starch granules. Subsequently, the discs were incubated (eight discs per 10 ml) at room temperature in one of the following mixtures, which all contained 50 mM citrate-NaOH, pH 6.5, and, in addition, one of the following compounds or mixtures of compounds: (a) 50 mM [U-14C]glucose 1-phosphate; (b) 50 mM [U-14C]glucose; (c) 50 mM [U-14C]maltose; (d) 50 mM [U-14C]sucrose; (e) 50 mM [U-14C]glucose 6-phosphate; (f) 50 mM [U-14C]glucose 1-phosphate and 10 mM unlabelled glucose 6-phosphate; (g) 50 mM [U-14C]glucose 1-phosphate and 10 mM unlabelled Pi. For each mixture the total radioactivity applied was balanced to be 37 kBq. At intervals, three to four discs each were withdrawn and immediately frozen in liquid nitrogen. Alternatively, the concentration of the glucose 1-phosphate was varied as indicated, but the specific radioactivity remained constant. In control experiments we ensured that the incubation of tuber discs with [U-14C]glucose 1-phosphate resulted in an almost exclusive labelling of starch (Supporting Information, Fig. S1). Therefore, in later experiments, starch fractions were isolated in which > 85% of the insoluble material consisted of starch. These fractions were isolated by addition of 5 ml precooled 20% (v/v) ethanol to the discs and homogenization using an Ultra Turrax. Following centrifugation (10 000g for 12 min; 4°C) the supernatants were discarded and 1 ml water was added to each pellet, mixed and centrifuged, as described earlier. This washing procedure was repeated two times. Finally, the starch fractions were resuspended in 1 ml water and the 14C-label was monitored using a liquid scintillation counter.
Preparation of native starch granules
Native starch granules were prepared as described elsewhere (Ritte et al., 2000). Briefly, plant material was homogenized and mixed with extraction buffer (100 mM HEPES-KOH, pH 8.0, 1 mM EDTA, 5 mM dithioerythritol (DTE), 0.5 mM phenylmethylsulphonyl fluoride, and 0.05% (v/v) Triton-X-100). For leaf starch, following filtration using a nylon net the filtrate was centrifuged and the resulting pellet was washed with extraction buffer. Finally, starch was purified using a Percoll cushion and dried in a vacuum. For starch from tubers and rice endosperm, following homogenization the starch granules were allowed to settle and the sediment was washed and finally dried in a vacuum.
Extraction of buffer-soluble proteins
Tuber material (1–2 g fresh weight) was homogenized in 5 ml grinding buffer (100 mM HEPES-NaOH, pH 7.5, 1 mM EDTA, 2 mM DTE, 10% (v/v) glycerol, and 0.5 mM phenylmethylsulphonyl fluoride) using an Ultra Turrax. The resulting homogenate was centrifuged (20 000g for 12 min) and the supernatant was passed through a nylon net (60–100 μm pore size). The filtrate was used for protein quantification and for native PAGE. The entire procedure was performed at 4°C.
Native PAGE following phosphorylase activity staining and photometric phosphorylase activity assay
Native PAGE, the subsequent phosphorylase activity staining and the photometric assay of the phosphorylase activity were performed as described elsewhere (Fettke et al., 2005).
Heterologous expression and purification of Pho1 from Oryza sativa
The Pho1 clone from rice (Oryza sativa; Satoh et al., 2008) was transformed into E. coli BL21 (DE3) cells. Subsequently, the transformed E. coli cells were grown for 24 h at 25°C in 600 ml Luria-Bertani medium containing both chloramphenicol (34 μg ml−1) and ampicillin (100 μg ml−1). Induction by isopropylthio-β-galactoside was omitted as it resulted in a significant degradation of the expressed protein. Bacterial cells were collected by centrifugation (5 000g for 10 min; 4°C) and were resuspended in15 ml extraction buffer (20 mM sodium phosphate buffer, pH 7.4, 500 mM NaCl, 20 mM imidazole, 2.5 mM DTE, and protease inhibitor cocktail I (Calbiochem, Bad Soden, Germany). Cells were broken by ultrasonification and the homogenate was cleared by centrifugation (13 000g for 15 min; 4°C). The supernatant was loaded on a HisTrap HP column (product no. 17-5319-01; GE Healthcare, Freiburg, Germany). Subsequently, the column was washed with 15 ml extraction buffer and the His-tagged Pho1 protein was eluted by stepwise increasing concentrations of imidazole (50–500 mM; pH 7.4) dissolved in extraction buffer. Pho1-containing fractions were identified by Western blotting using an Anti-His-Antibody (Quiagen), pooled and concentrated by ultrafiltration (MWC 10 kDa; Amicon Ultra, Millipore, Schwalbach, Germany). Subsequently, the purified Pho1 protein preparation was equilibrated with a buffer containing 100 mM HEPES-NaOH, pH 7.5, 1 mM EDTA and 2 mM DTE, and was stored frozen at −80°C.
In vitro14C-labelling of native starch granules using recombinant Pho1
In a total volume of 120 μl, the reaction mixture contained 100 mM citrate-NaOH, pH 6.5, 20 mM [U-14C]glucose 1-phosphate (1.23 kBq) and 175 μg native starch granules. Following preincubation at 37°C for 5 min, recombinant Pho1 from rice (15 μg each) was added. Alternatively, buffer-soluble proteins (40 μg each) extracted from potato tubers from wild-type plants or from transgenic lines containing an antisense construct directed against the Pho1 phosphorylase isozyme(s) were applied. In another experiment 30 μg soluble maltodextrins (product no. EC 232-940-4, Sigma-Aldrich) were added to the reaction mixture. The reaction mixtures were further incubated at 37°C as indicated. The reaction was terminated by adding SDS to give a final concentration of 2% (w/v). Following centrifugation (10 000g for 10 min; 4°C) the pelleted starch granules were washed twice with water (1 ml each), resuspended in 1 ml water and the 14C-content was monitored using a liquid scintillation counter.
Quantification of protein
Buffer-soluble protein was quantified using the microassay of Bradford (1976) with bovine serum albumin serving as standard.
Potato tuber discs efficiently convert external glucose 1-phosphate into starch
In a first series of 14C-labelling experiments, discs were prepared from freshly harvested wild-type potato tubers and were incubated with equimolar concentrations of [U-14C]-labelled neutral mono- or disaccharides or with [U-14C]glucose phosphates. In addition, two incubation solutions were prepared that contained a mixture of [U-14C]glucose 1-phosphate plus unlabelled glucose 6-phosphate or labelled glucose 1-phosphate and unlabelled orthophosphate. For each mixture, radioactivity was balanced to the same specific radioactivity for all sugars or sugar derivatives tested. Following incubation, starch was isolated from the tuber discs and the 14C-content was monitored (Fig. 1a). Following 20 or 40 min incubation the highest labelling of starch by far was observed when the tuber discs were incubated with [U-14C]glucose 1-phosphate. External [U-14C]glucose was less effective and the lowest labelling of starch was obtained following incubation with either sucrose, maltose or glucose 6-phosphate. The glucose 1-phosphate-dependent incorporation into starch was not affected by the presence of unlabelled glucose 6-phosphate. By contrast, external orthophosphate diminished the glucose 1-phosphate-dependent incorporation into starch.
It should be noted that the different rates of starch labelling, as shown in Fig. 1(a), do not necessarily indicate different rates of starch biosynthesis, as internal parameters, such as pool sizes of intermediates, affect the labelling kinetics. However, the data clearly indicate that in parenchyma cells from freshly harvested tubers, a path of starch biosynthesis is functional that is selectively initiated by the uptake of the anomeric glucose phosphate ester. The initiation of this path is separated against external glucose 6-phosphate. However, the inhibitory effect of external orthophosphate suggests that the putative glucose 1-phosphate transporter of the cell membrane does also transport orthophosphate and, possibly, mediates an exchange of both anionic compounds.
In another series of experiments, we tested whether this glucose 1-phosphate-dependent path of starch biosynthesis is constitutively functional in potato tubers or, alternatively, whether it is restricted to distinct metabolic or developmental stages. Therefore, discs were prepared from wild-type potato tubers that had been stored for different periods of time or under different conditions and, consequently, differed in their physiological states. Discs were then incubated with [U-14C]glucose 1-phosphate and incorporation of 14C into starch was monitored (Fig. 1b). The glucose 1-phosphate-dependent path of starch biosynthesis is functional both in sprouting and in resting tubers. It is not significantly altered if the tubers were stored at 4°C. Thus, at least in short-term experiments, the conversion of external glucose 1-phosphate to reserve starch is constitutively functional in potato tubers. However, discs from very small tubers deviate from all other samples tested as they incorporate 14C into starch at a high rate only during the first 30 min of incubation, and labelling of starch subsequently ceases, for unknown reasons.
For a first characterization of the conversion of the external glucose 1-phosphate to starch, media containing varying concentrations of the anomeric glucose phosphate were prepared and the incubation of the tuber discs was restricted to 5 min. Subsequently, starch granules were isolated and their 14C-content was quantified. Glucose 1-phosphate-dependent starch labelling was detectable at an external concentration of 1 mM and saturated at c. 20-fold higher concentrations (Fig. 1c). It should, however, be noted that the experimental system used does not permit a more detailed analysis of this process. In tuber discs most of the parenchyma cells do not have immediate access to the externally supplied glucose ester but rather rely on the glucose 1-phosphate pool that has entered the apoplastic space. It is reasonable to assume that the apoplastic glucose 1-phosphate concentration somehow increases with that of the glucose ester in the medium, but any quantification is difficult to achieve. Furthermore, it is impossible to determine when the apoplastic glucose 1-phosphate pool has reached a steady state. Because of these limitations, a further shortening of the incubation time is not meaningful.
In the cytosol neither the PGM nor the glucosyl transferases mediate indispensable steps within the conversion of external glucose 1-phosphate into starch
Following uptake by the tuber parenchyma cells, [U-14C]glucose 1-phosphate is accessible to the cytosolic glucose monophosphate metabolism. Two major cytosolic glucose monophosphate pools exist, namely glucose 1-phosphate and glucose 6-phosphate. Both pools are interconnected by the action of cPGM, that is, the cytosolic isozyme(s) of the PGM (EC 18.104.22.168). The cPGM-mediated reaction is considered to be crucial for the entire tuber development, as transgenic potato plants possessing a lower cPGM activity form fewer and smaller tubers compared with the wild-type control. Tuber yield can be as low as 20% (or, occasionally, close to 0%) of that of the control plants. At the cellular level, the tuber parenchyma cells from these lines deviate from the wild-type control by possessing a lower glucose 6-phosphate concentration (Fernie et al., 2002b).
Because of this phenotype of the cPGM-reduced lines, we asked the question whether or not the conversion of external glucose 1-phosphate to starch includes, as a rate-limiting step, the cPGM-mediated reaction. Therefore, two transgenic lines were applied that, under the control of the 35S promoter, express a cPGM antisense construct. For both lines, the specific (protein-based) PGM activity was < 70% of that of the wild-type. As revealed by zymograms following native PAGE, tubers from line D have a lower residual cPGM activity than those from line C (for details see Fettke et al., 2008). Discs prepared from freshly harvested tubers were incubated with [U-14C]glucose 1-phosphate. Tuber discs from wild-type plants grown alongside the transgenic lines served as control. At intervals, discs were removed from the incubation medium and the 14C-incorporation into starch was quantified. In all samples the labelling of starch (based on the fresh weight of the tuber discs) increased with time (Fig. 2a). Compared with the wild-type control, labelling of starch was not consistently diminished in the two transgenic lines. In line C incorporation into starch was less than that of the control, but discs from line D in which the antisense inhibition of cPGM is most effective incorporated significantly more label into starch compared with the wild-type (t-test, P =0.05). A considerable phenotypical variation between different transgenic lines has also been observed previously (Fernie et al., 2002b). Possibly, the increased flux into starch, as measured in line D, is the result of a diminished cytosolic interconversion of glucose 1-phosphate and glucose 6-phosphate, which, in turn, may favour the import of glucose 1-phosphate into the amyloplasts (see below). In summary, the data shown in Fig. 2a do not point to an essential function of the cPGM-mediated reaction within the flux from external glucose 1-phosphate to starch.
The results shown in Fig. 2(a) are consistent with the labelling data obtained with transgenic potato lines that possess altered levels of either of two glucosyl transferase activities, the cytosolic phosphorylase isozyme Pho2 or the cytosolic transglucosidase, DPE2. The latter enzyme has been shown to be essential for starch degradation, as mutants lacking functional DPE2 possess a starch-excess phenotype (Chia et al., 2004; Lloyd et al., 2004). Under in vitro conditions DPE2 catalyses the reversible transfer of a glucosyl residue from maltose to the nonreducing ends of highly branched polyglucans, such as glycogen, that act as nonphysiological substitutes of cytosolic heteroglycans (Fettke et al., 2009). DPE2-deficient Arabidopsis mutants possess largely altered cytosolic heteroglycans (Fettke et al., 2006). Tubers from two transgenic potato lines were applied that, because of an antisense construct, strongly repress the expression of Pho2 (Fettke et al., 2005). In tuber discs from both lines the carbon flux from external glucose 1-phosphate to starch was essentially unchanged as compared with the wild-type (Fig. 2b). Whilst Pho2 directly interacts with the cytosolic glucose 1-phosphate pool, the transglucosidase, DPE2, indirectly interacts with the glucose monophosphates (Fettke et al., 2008). In Arabidopsis, DPE2 has been consistently observed in the cytosol (Chia et al., 2004; Lu and Sharkey 2004). In S. tuberosum L. originally the enzyme has been attributed to the plastidial compartment (Lloyd et al., 2004). However, nonaqueous fractionation clearly indicated a cytosolic location (Fettke et al., 2005).
Tuber discs derived from RNAi lines with a strongly reduced expression of DPE2 were incubated with [U-14C]glucose 1-phosphate, and labelling of the reserve starch granules was followed (Fig. 2b). In both RNAi lines, labelling of starch did not differ from that obtained with the wild-type control.
Plastidial reactions of the glucose 1-phosphate-dependent path of starch biosynthesis
The carbon flux from external glucose 1-phosphate to starch does not require a normal level of the cPGM activity (Fig. 2a). Therefore, it is conceivable that, following the uptake into the cytosol, at least a major proportion of the glucose monophosphate remains unchanged until it is imported into the stroma of the amyloplast and enters the plastidial pool of glucose 1-phosphate. In principle, the latter can fulfil two biochemical functions: first, it serves as a substrate for the AGPase-dependent formation of ADPglucose, which is subsequently used for starch biosynthesis by various starch synthases; and second, it functions as a glucosyl donor for a transfer reaction mediated by the plastidial (Pho1-type) phosphorylase isozyme(s) whose expression is exceptionally high in potato tubers. If the Pho1-catalysed glucosyl transfer reactions are relevant for the synthesis of reserve starch, transgenic potato lines having lower Pho1-type isozyme concentrations are expected to convert external glucose 1-phosphate less efficiently to starch. In order to test this possibility, tuber discs derived from several transgenic Pho1-antisense lines were incubated with [U-14C]glucose 1-phosphate and labelling of starch was monitored.
In S. tuberosum L., two Pho1-type phosphorylases exist that have been designated as Pho1a and Pho1b. Both genes are highly homologous except the large insertion (Albrecht et al., 1998). The two gene products form homodimeric and heterodimeric phosphorylases, but in tubers the predominant native Pho1 by far is the homodimeric Pho1a, whereas the homodimeric Pho1b is usually below the limit of detection (Albrecht et al., 1998). Six independently generated antisense lines were used, two of which contain an antisense construct directed against Pho1a or Pho1b, or against both Pho1a and Pho1b (Duwenig, 1996; Fettke et al., 2005). For all six lines, and for the wild-type control as well, the total buffer-soluble phosphorylase activity was determined and the specific phosphorylase activities are given (Table 1). In transgenic lines containing an antisense construct directed against Pho1b, a relatively high specific phosphorylase activity was observed. This implies that the activity level of the Pho1a gene product is largely unaffected by the expression of the anti-Pho1b antisense construct. By contrast, the total phosphorylase activity of the four other transgenic lines significantly is much lower than that of the wild-type control. These data were confirmed by a native PAGE performed with extracts from tubers of the six transgenic lines and from the wild-type tubers as well (Fig. 3a). Following electrophoresis, the gel was incubated with glucose 1-phosphate and, subsequently, sites of phosphorylase-dependent glucan synthesis were detected by iodine staining. In this experiment, a glycogen-containing separation gel was used. Owing to the high affinity of the cytosolic (Pho2) phosphorylase isozyme, during electrophoresis this enzyme strongly binds to the immobilized polyglucan and is therefore retained at the top of the separation gel. The plastidial (Pho1) phosphorylase isozymes have an extremely low affinity towards glycogen, and electrophoretic mobility is therefore retained. During the subsequent incubation with glucose 1-phosphate they form a distinct band stained by iodine that is located in the middle of the gel. In four of the transgenic lines, this mobile band was strongly reduced or even below the limit of detection, but in the two anti-Pho1b lines no inhibition or only a moderate reduction of this band was observed (Fig. 3a).
Table 1. Total phosphorylase activity of various transgenic potato (Solanum tuberosum) plants
Relative specific activity (percentage of wt)
Specific activity (nmol glucose 1-phosphate min−1 mg−1 protein)
Pho1(a)-1 and Pho1(a)-2, two transgenic lines containing an anti-Pho1a antisense construct; Pho1(b)-1 and Pho1(b)-2, two lines containing an anti-Pho1b construct; Pho1(a + b)-1 and Pho1(a + b)-2, two lines containing both an anti-Pho1a and an anti-Pho1b construct.
Buffer-soluble proteins were extracted from potato tubers. Enzyme activities are given both as specific activities (nmol glucose 1-phosphate min–1 mg–1 protein) and as percentage of the specific activities relative to that of the wild-type control (100%). The means of 12 biological replicas (derived from two independently grown batches of plants) and the SD are given.
123.52 ± 3.94
15.48 ± 5.60
9.36 ± 1.66
116.37 ± 3.30
60.93 ± 8.47
Pho1(a + b)-1
10.10 ± 6.71
Pho1(a + b)-2
10.86 ± 0.51
Discs prepared from tubers of the transgenic potato lines, and also of wild-type controls, were incubated with [U-14C]glucose 1-phosphate. Subsequently, the labelling of the starch was quantified (Fig. 3b). For the wild-type control, incorporation into starch was higher than in all transgenic lines. In the four lines with the most efficient inhibition of the Pho1-type isozyme (two anti-Pho1a lines and two double antisense lines), labelling of starch was more strongly inhibited that in the two anti-Pho1b lines. However, even in transgenic lines possessing < 10% of the total phosphorylase activity (Pho1(a)-2 and both double antisense lines), the incorporation into starch accounted for c. 30% of that of the wild-type control.
The data shown in Fig. 3(b) strongly suggest that the Pho1-type isozymes are involved in the conversion of external glucose 1-phosphate into starch. However, they do not indicate if the Pho1-type isozymes act directly on native starch granules. To test this possibility, native starch granules were prepared from various sources (leaves from wild-type Arabidopsis, potato or rice plants and also from wild-type tubers ). Equal amounts of each starch granule preparation were incubated with recombinant Pho1 from rice and [U-14C]glucose 1-phosphate. At intervals, aliquots of the reaction mixtures were withdrawn and the 14C-content of the starch was determined. Under the in vitro conditions used, the recombinant phosphorylase isozyme transferred glucosyl residues from glucose 1-phosphate to all native starch granules tested (Fig. 4a). However, quantitatively the rate of incorporation varied largely depending on the source of the granules. Incorporation into transitory starches from both Arabidopsis and potato was equally high and exceeded that into reserve starch from tubers. At a molecular level, these difficulties are hard to explain. It is reasonable to assume that they reflect selective structural properties at the surface of the various types of the starch granules, such as the ratio between highly and less ordered regions or even different arrangements of glucan chains within the less ordered regions, but these features are difficult to analyse. A large variation has also been observed when a variety of native starch granules was in vitro phosphorylated by a recombinant glucan, water dikinase. Depending on the botanical source, the maximum rate of phosphorylation varied by more than one order of magnitude (Hejazi et al., 2008). A common feature of the labelling kinetics shown in Fig. 4b is nonlinearity. With all the starches tested, the rate of the Pho1-mediated glucosyl transfer rapidly decreases with time. Although the phosphorylase-catalysed glucosyl transfer does not diminish the total number of glucosyl acceptor sites, it appears that the repetitive glucosyl transfer to a given number of sites decreases the transfer efficiency. The same observation was detectable under conditions in which a tenth of the amount of the recombinant Pho1 was used.
In another series of experiments, we tested whether the plastidial (Pho1-type) phosphorylase isozymes from S. tuberosum L. are capable of transferring glucosyl residues from glucose 1-phosphate to potato tuber starch granules. Therefore, buffer-soluble proteins were extracted from tubers from two transgenic lines, containing double antisense constructs directed against both Pho1a and Pho1b (Pho1(a +b)-1 and Pho1(a + b)-2), and also from wild-type tubers. Native starch granules isolated from wild-type tubers were incubated with the three protein extracts and [U-14C]glucose 1-phosphate. At intervals, in aliquots of the mixtures, reaction was terminated and the 14C- content of the starch granules was quantified. Extracts from tubers of the two transgenic lines were less efficient in labelling the starch granules than that from the wild-type control (Fig. 4b). This result concurs with that obtained with the recombinant Pho1 (Fig. 4a) and clearly indicates that the plastidial phosphorylase isozymes from S. tuberosum L. is capable of using native starch granules as glucosyl acceptors. Buffer-soluble proteins extracted from the various potato tubers contain several glucan-hydrolysing enzyme activities that may interfere with the phosphorylase-mediated glucosyl transfer. Because of these complications, the data shown in Fig. 4b may, to some extent, underestimate the rate of the Pho1-catalysed glucosyl transfer.
As indicated by the starch-deficient control, the glucosyl transfer that is observed in the various assays (Fig. 4b) appears to be restricted to native starch granules. Furthermore, incorporation into starch is strongly diminished if the reaction mixture contained soluble maltodextrins in addition to starch granules. This inhibitory effect is not unexpected as the plastidial phosphorylase isozymes have a high affinity towards small maltodextrins (Steup & Schächtele, 1981). It is reasonable to assume that both native starch granules and soluble maltodextrins act as glucosyl acceptors for the Pho1-mediated glucosyl transfer and thereby compete for each other.
Transgenic potato lines with reduced levels of both the cytosolic and the plastidial PGM activity efficiently convert external glucose 1-phosphate into starch
The results shown in Fig. 4(b) concur with 14C-labelling experiments performed with tuber discs from two transgenic potato lines that exhibit a combined antisense repression of both cPGM and pPGM. These lines (line 23 and 32) possess < 10% of the wild-type level of both compartment-specific PGM activities (Fernie et al., 2002a). Discs derived from freshly harvested tubers from lines 23 and 32, and also from the wild-type, were incubated with 25 mM [U-14C]glucose 1-phosphate. At intervals, discs were removed, starch was isolated and the content of radioactivity was determined (Table 2). After 20 or 40 min incubation, labelling of starch was not significantly altered in either of the transgenic lines compared with the wild-type control (t-test, P =0.05). Therefore, we conclude that neither cPGM nor pPGM exerts any efficient control over the conversion of external glucose 1-phosphate to starch, as observed in short-term experiments with tuber discs.
Table 2. 14C-content of starch following the incubation of potato (Solanum tuberosum) tuber discs in [U-14C]glucose 1-phosphate
nmol glucose g−1 FW
Freshly harvested tubers from two transgenic lines that underexpress both the cytosolic and the plastidial phosphoglucomutase isozymes (c/pPGM 23 and c/pPGM 32) and wild-type tubers were used to prepare discs. Subsequently the discs were incubated in a medium containing 25 mM [U-14C]glucose 1-phosphate and 50 mM citrate-NaOH (pH 6.5) at room temperature. Following 20 and 40 min, discs were removed and starch was isolated. Finally, the 14C-content of the starch was quantified. The average of three independently performed experiments and SD are shown.
955.01 ± 6.52
1686.28 ± 209.74
993.14 ± 35.69
1930.25 ± 188.18
1242.10 ± 61.51
2127.85 ± 174.05
Potato tuber discs are a suitable system to study carbon fluxes in heterotrophic plant tissues as they rapidly take up and metabolise a variety of sugars or sugar derivatives. Therefore, short-time experiments can be performed without any of the complications that are related to the sink–source interactions within the intact plant (Geiger et al., 1998). Nevertheless, tuber discs are, to some extent, an artificial system in which the metabolism tends to change with time. As an example, the rate of respiration is increased several-fold during 5 h incubation (Geiger et al., 1998). In order to minimize these complications, all the experiments presented in this study have been restricted to, at the most, 1 h incubation.
We used the potato tuber discs in order to analyse the path by which external glucose 1-phosphate is converted to reserve starch. Unexpectedly, starch labelling occurs with a higher rate than the well studied conversion of external glucose or sucrose to starch (Fig. 1a). As discussed earlier, the labelling kinetics do not necessarily mirror the underlying carbon fluxes, as the uptake rates for the various labelled compounds may differ and, probably more importantly, differences in pool sizes may greatly affect the resulting specific radioactivities of the respective intermediates, which in turn strongly determine the labelling kinetics observed. The samples analysed in this study were usually derived from at least two periods of incubation, and therefore effects that are consistently observed are expected to reflect essentially a steady-state situation.
Despite the limitations of the tuber discs, two approaches can be chosen to unequivocally characterize metabolic paths. First, if both a labelled compound and a second, chemically different but unlabelled metabolite are simultaneously applied, paths can be characterized as being separated from each other or fused. Second, incubation experiments can be performed using transgenic lines that possess a largely altered expression of a target gene. By comparing the resulting labelling kinetics with those obtained with the wild-type control, essential enzymatic steps within the flux of label can be defined.
Tuber discs rapidly take up glucose 1-phosphate if applied in millimolar concentrations (Fig. 1c). The actual apoplastic glucose 1-phosphate concentrations that are relevant for the uptake process are unknown. However, it is reasonable to assume that they are considerably lower than those of the incubation medium, and therefore uptake of glucose 1-phosphate appears not to be restricted to an artificially high range of concentrations. For several reasons, the uptake seems to be largely selective: the presence of unlabelled glucose 6-phosphate did not affect the glucose 1-phosphate-dependent carbon flux to starch but the addition of orthophosphate decreased the flux (Fig. 1a). Labelled, externally supplied glucose 6-phosphate did not functionally replace glucose 1-phosphate (Fig. 1a). Finally, it should be mentioned that the selective conversion of externally supplied glucose 1-phosphate to starch cannot be explained by an apoplastic dephosphorylation that could be mediated by extracellular phosphatases. If so, various glucose monophosphates are expected to result in similar carbon fluxes and the rates obtained should be lower than (or, at best, approximate to) those obtained when labelled glucose is added to the incubation mixture. The experimental data (Fig. 1a) are not consistent with this assumption. At a molecular level, the import of glucose 1-phosphate into the cytosol has not yet been characterized.
Following the uptake into the parenchyma cells of the potato tuber, a major proportion of the 14C-labelled glucose 1-phosphate appears to be directed to the amyloplast where it enters the plastidial pool of the anomeric glucose monophosphate. This conclusion is based on several independent lines of evidence: First, transgenic potato plants that, owing to an antisense construct, possess a lower activity level of the cPGM do not consistently exhibit a lower rate of the 14C-incorporation into starch. In fact, in one antisense line, the flux to starch was significantly increased compared with the wild-type control (Fig. 2a). By contrast, the phenotype of these transgenic potato lines differs significantly from the wild-type control as tuber growth is strongly retarded and the tuber starch content is lowered (Fernie et al., 2002b). Thus, the in planta utilization of the photosynthesis-derived sucrose appears to include, as an essential step, the cytosolic interconversion of the two glucose monophosphates, but the short-term conversion of external glucose 1-phosphate to starch does not strictly rely on this reaction. Second, neither the repression of the cytosolic phosphorylase (Pho2) nor that of the transglucosidase (DPE2) diminished the conversion of the external glucose 1-phosphate to starch (Fig. 2b). Third, transgenic plants that have significantly reduced concentrations of the plastidial phosphorylase (Pho1-type) isozymes incorporated less carbon into starch than the wild-type control (Fig. 3a,b). As revealed by in vitro assays, the plastidial Pho1-type phosphorylase isozymes are capable of transferring glucosyl residues to native starch granules (Fig. 4). This has been demonstrated by in vitro assays in which native starch granules from potato tubers were applied as glucosyl acceptor and [U-14C]glucose 1-phosphate as donor. The glucosyl transfer was catalysed by either recombinant Pho1 from rice (Fig. 4a) or the phosphorylase activity extracted from potato tubers. In the latter case, the rate of the starch labelling was lower when tubers from transgenic Pho1 antisense lines were extracted (Fig. 4b).
The transporter that mediates the import into the plastid has not yet been characterized at a molecular level. Recent studies indicate that the genome of Arabidopsis encodes c. 140 putative metabolite transporters, all of which are thought to be located in the inner plastid envelope membrane (Ferro et al., 2002; Schwacke et al., 2003), and therefore the current knowledge of plastidial metabolite transporters seems to be far from complete. It should be noted, however, that recently some indications for a functional glucose 1-phosphate import into nongreen plastids have been published (for potato tubers: Kosegarten & Mengel, 1994; Naeem et al., 1997; for wheat: Tetlow et al., 1996; for soy bean: Coates & ap Rees, 1994). However, in some of these reports, characteristics of the transport (such as the pH dependency) were unexpected and difficult to explain.
Based on the labelling kinetics obtained with potato tuber discs, the following paths of carbon are proposed (Fig. 5). Sucrose-derived glucose monophosphates form the cytosolic pools of glucose 6-phosphate and glucose 1-phosphate. In addition, glucose 1-phosphate is imported from the apoplastic space into the cytosol and thereby enters the two interconvertible pools. Both glucose monophosphates can be imported into the amyloplast and join the two plastidial pools that are interconnected by the action of cPGM. However, for the conversion of apoplastic glucose 1-phosphate to starch, neither the cPGM nor the pPGM reaction is an indispensable step. Inside the amypoplast glucose 1-phosphate fulfils a dual function as it can enter two distinct starch-synthesizing paths, one of which consists of a single reaction, that is, the Pho1-mediates glucosyl transfer to the surface of native starch granules. The other biosynthetic path comprises several reactions, The first step is the conversion of glucose 1-phosphate to ADPglucose catalysed by the AGPase. Subsequently, ADPglucose acts as the general glucosyl donor used by several starch synthases for a series of glucan elongation reactions (Stensballe et al., 2008).
A direct import of glucose 1-phosphate into the plastid has also been postulated for the transgenic potato lines possessing an antisense repression of both cPGM and pPGM (Fernie et al., 2002a) and for the Chlamydomonas mutant sta5-1 lacking a pPGM activity but accumulating 4–20% of the wild-type starch amounts (Van den Koornhuyse et al., 1996; Ball, 2000).
The starch biosynthesis paths proposed in Fig. 5 probably explain the phenotype of seeds of Vicia narbonensis that possess a largely decreased activity level of the plastidial AGPase. In cotyledons of these transgenic lines, the activity of the ADPglucose-forming enzyme was reduced by up to 95%. Similarly, during the mid- to late-maturation phase, the ADPglucose concentrations were below 10% of those of the wild-type control. However, the starch content was only moderately decreased, and therefore the AGPase exerts a low control on starch biosynthesis. In fact, the flux control coefficient of the AGPase has been estimated to be as low as 0.08 (Weber et al., 2000; Rolletschek et al., 2002). Similarly, seeds of pea (Pisum sativum) plants deficient in a functional AGPase are capable of forming significant amounts of starch that, based on fresh weight, account for c. 50% of the wild-type control. However, the actual starch accumulation of the insertional mutants is even higher, as their seeds tend to possess a higher fresh weight compared with the wild-type control (Weigelt et al., 2009). Finally, in leaves of field-grown potato plants, transcript, protein and activity levels of the plastidial (Pho1 type) phosphorylase isozyme(s) are low at the beginning of the day but then significantly increase and reach a peak during the light period (Albrecht et al., 2001). All these data suggest that the entire process of starch biosynthesis (or essential steps within this process) does not exclusively rely on the AGPase-dependent path.
It should be noted, however, that the scheme (Fig. 5) does not imply that (or, alternatively, under which conditions) each of the two paths results in the formation of an entire starch granule. In the labelling experiments presented in this study, it remains unclear whether the entire starch granule population acts as acceptor for the Pho1-mediated glucosyl transfer. Alternatively, this function could be restricted to starch particles having a distinct size, such as small granules. Furthermore, it is unknown whether only distinct regions of the granule surface (such as semicrystalline or less ordered areas) are utilized for the glucosyl transfer. In the endosperm of Pho1-deficient rice seeds, starch biosynthesis was unaffected at elevated temperatures, whereas seeds developing at 10°C clearly accumulated less starch (Satoh et al., 2008). The phenotype of the Pho1-deficient rice mutants strongly suggests that, depending upon the external conditions, the Pho1-dependent and the Pho1-independent paths exert redundant or nonredundant in vivo functions. This conclusion concurs with the fact that tubers from transgenic potato lines strongly underexpressing the plastidial (Pho1-type) phosphorylase isozymes possess essentially unchanged starch concentrations when grown at 17–20°C (Fettke et al., 2005). Therefore, under these conditions the plastidial glucose 1-phosphate appears to be converted to starch almost exclusively via the AGPase-dependent path. However, the ratio between both fluxes may be different at lower temperatures. Furthermore, it is unknown if glucose 1-phosphate is also involved in the signalling processes that regulate intracellular carbon fluxes.
In summary, the dual starch biosynthetic paths as outlined in Fig. 5 contribute to a remarkable metabolic flexibility from which plants benefit under a wide range of external conditions.
This work was supported by the Deutsche Forschungsgemeinschaft – SFB 429 TP B2 (JF and MS) and GoFORSYS of the BMBF (MS). The authors thank Professor Alisdair R. Fernie and Dr Anna Lytovchenko (Max-Planck-Institut of Molecular Plant Physiology, Potsdam-Golm, Germany) for kindly providing all the transgenic PGM lines, and Professor Jens Kossmann and Dr James R. Lloyd (Stellenbosch University, Stellenbosch, South Africa) for generously giving us tubers of stDPE2 RNAi plants. The authors are indebted to Ms Jessica Alpers and Ms Irina Malinova for excellent assistance.