Two carbon fluxes to reserve starch in potato (Solanum tuberosum L.) tuber cells are closely interconnected but differently modulated by temperature.

Parenchyma cells from tubers of Solanum tuberosum L. convert several externally supplied sugars to starch but the rates vary largely. Conversion of glucose 1-phosphate to starch is exceptionally efficient. In this communication, tuber slices were incubated with either of four solutions containing equimolar [U-¹⁴C]glucose 1-phosphate, [U-¹⁴C]sucrose, [U-¹⁴C]glucose 1-phosphate plus unlabelled equimolar sucrose or [U-¹⁴C]sucrose plus unlabelled equimolar glucose 1-phosphate. C¹⁴-incorporation into starch was monitored. In slices from freshly harvested tubers each unlabelled compound strongly enhanced ¹⁴C incorporation into starch indicating closely interacting paths of starch biosynthesis. However, enhancement disappeared when the tubers were stored. The two paths (and, consequently, the mutual enhancement effect) differ in temperature dependence. At lower temperatures, the glucose 1-phosphate-dependent path is functional, reaching maximal activity at approximately 20 °C but the flux of the sucrose-dependent route strongly increases above 20 °C. Results are confirmed by in vitro experiments using [U-¹⁴C]glucose 1-phosphate or adenosine-[U-¹⁴C]glucose and by quantitative zymograms of starch synthase or phosphorylase activity. In mutants almost completely lacking the plastidial phosphorylase isozyme(s), the glucose 1-phosphate-dependent path is largely impeded. Irrespective of the size of the granules, glucose 1-phosphate-dependent incorporation per granule surface area is essentially equal. Furthermore, within the granules no preference of distinct glucosyl acceptor sites was detectable. Thus, the path is integrated into the entire granule biosynthesis. In vitro C¹⁴C-incorporation into starch granules mediated by the recombinant plastidial phosphorylase isozyme clearly differed from the in situ results. Taken together, the data clearly demonstrate that two closely but flexibly interacting general paths of starch biosynthesis are functional in potato tuber cells.

Introduction

Starch is the almost ubiquitous storage polysaccharide of plastid-containing eukaryotes. It is deposited as water-insoluble granules that possess a highly ordered intra- and intermolecular structure and is thought to consist of two types of α-glucans, amylopectin and amylose. Except in some mutant starches, the former is the major constituent of the starch particles and, to some extent, resembles chemical features of glycogen which is the principal storage polysaccharides of animals and fungi (and of the majority of prokaryotes as well). In both glycogen and amylopectin, most of the glucosyl residues are connected via α-1,4-linkages, but approximately 5–10% of the total glucosidic bonds are α-1,6 linkages that constitute branching points. By contrast, amylose is a heterogeneous α-glucan type consisting of linear or poorly branched chains whose degree of polymerization (DP) covers a wide range but is consistently below that of amylopectin (Ball and Morell, 2003; Pérez and Bertoft, 2010). Starch biosynthesis is a complex process that largely proceeds by a close collaboration of starch synthase isozymes and starch branching/debranching isozymes (for reviews see Keeling and Myers, 2010; Zeeman ). The functional interaction between the various starch-related enzymes includes reversible protein–protein interactions (Tetlow , 2008; Hennen-Bierwagen ; Liu , 2009).

In green algae, mosses, ferns, and in higher plants starch biosynthesis is restricted to the stromal space of plastids. Typically, higher plants possess two major and metabolically distinct types of starch, transitory and reserve starch. Transitory starch is accumulated in chloroplasts during those periods of time in which the reductive pentose phosphate cycle is operating. In the subsequent dark period, transitory starch is partially or almost completely degraded and, therefore, the cellular starch content exhibits large diurnal fluctuations (Zeeman ; Stitt ). By contrast, reserve starch accumulation is only indirectly driven by photosynthesis and typically is restricted to heterotrophic organs. Reserve starch is formed over an extended period of time and has a much lower turnover rate. Its biosynthesis depends on both the photosynthetic activity in the source and the import of photosynthate into the sink organs (Keeling and Myers, 2010). Massive reserve starch accumulation is one of the major determinants of sink strength (Ball and Morell, 2003; Keeling and Myers, 2010).

Tubers from potato (Solanum tuberosum L.) represent a prominent example for the efficient production of reserve starch. In addition to its relevance for fundamental plant research, this species is one of the most important crops worldwide whose reserve starch is used for many nutritional and (bio)technological applications (Zhang ; Santelia and Zeeman, 2010). Massive reserve starch accumulation in S. tuberosum L. is due to the fact that tuber parenchyma cells are equipped with a specialized form of plastids, i.e. amyloplasts that are capable of synthesizing large starch granules and, thereby, constitute a highly active sink.

Sliced potato tuber tissue is a well-established system for short-term studies on carbon fluxes to starch. The parenchyma tissue is compact and lacks an extended intercellular space. This anatomical feature favours an efficient uptake of various exogenously supplied compounds, such as glucose and sucrose, whose metabolism can easily be followed by using tracer techniques. However, the actual apoplastic concentration of exogenously supplied carbohydrates (or their derivatives) is essentially undefined and difficult to determine.

Obviously, the results obtained in short-term experiments performed with sliced tuber tissue reflect only the biochemical potential of the system and, therefore, cannot immediately be transferred to tubers growing as an organ of the intact plant. The results do, however, permit a detailed view on biochemical paths that, in principle, are likely to be functional in planta but cannot be studied at the level of the entire organ or plant. It is reasonable to assume that, in the intact plant, most of the paths analyzed in a simpler system, i.e. tuber slices, are modulated by complex (and flexible) regulatory processes that balance reserve starch biosynthesis in various tubers with the photosynthetic activity of the source organs and the capacity of the long-distance transport. Therefore, it is impossible to compare directly carbon fluxes or biosynthetic rates determined for tuber slices with those of the long-term process of reserve starch accumulation in planta.

According to a widely accepted view, sucrose is an efficient carbohydrate driving starch biosynthesis and this feature is robust over a wide range of concentrations. By contrast, externally supplied glucose has been reported to favour both glycolysis and respiration at the expense of starch biosynthesis (Geiger ). These results concur with the fact that, in potato plants, photosynthesis-derived sucrose is the dominant carbon compound transported from source to sink organs. Following this line, sucrose is imported into the parenchyma cells and, subsequently, the disaccharide is converted to monosaccharides that are metabolized further in the cytosol. The plastidial path of starch biosynthesis is initiated by the import of sucrose-derived glucose 6-phosphate and the subsequent conversion to glucose 1-phosphate. By the action of the plastidial ADPglucose pyrophosphorylase (AGPase; EC 2.7.7.27), the anomeric glucose ester enters the plastidial pool of ADPglucose which is the general glucosyl donor used by all starch synthases (EC 2.4.1.21; Kammerer ; Tauberger ; Fernie et al., 2002c).

However, several data clearly indicate that, in tubers, the process of starch synthesis and the starch-related carbon fluxes are more complex. Transgenic potato plants possessing strongly reduced levels of either the plastidial or the cytosolic phosphoglucomutase are compromised in starch accumulation (Tauberger ; Fernie et al., 2002b). However, transgenic lines that exhibit a combined antisense repression of both the cytosolic and the plastidial phosphoglucomutase unexpectedly display partial or complete recovery of starch accumulation (Fernie et al., 2002a). The underlying biochemistry of this phenotype is difficult to reconcile with the path of the sucrose-starch conversion mentioned above.

Recently, it has been demonstrated that tuber slices convert external 14C-labelled glucose 1-phosphate to starch and, unexpectedly, the rate of conversion by far exceeds that of glucose, glucose 6-phosphate, or sucrose (Fettke ). The conversion of glucose 1-phosphate into starch is initiated by the action of a distinct transporter that selectively mediates the import of glucose 1-phosphate into the cytosol. This conclusion has been reached as the simultaneous addition of unlabelled glucose 6-phosphate or glucose does not diminish the labelling of starch (Fettke ). In transgenic lines possessing a strongly reduced activity of the cytosolic phosphoglucomutase (EC 2.7.5.1), the glucose 1-phosphate-dependent incorporation into starch is enhanced compared with the wild-type (Fettke ). Furthermore, repression of the cytosolic disproportionating isozyme 2 (DPE2; EC 2.4.1.25) does not affect the carbon flux to starch. By contrast, 14C-incorporation into starch strongly correlates with the level of the plastidial phosphorylase (Pho1; EC 2.4.1.1) isozyme (Fettke ). These data imply that the anomeric glucose phosphate ester is capable of entering the amyloplast. Therefore, a novel plastidial transporter has to be postulated that mediates the import of glucose 1-phosphate. Inside the amyloplast, a major proportion of the imported glucose 1-phosphate acts as substrate for the incorporation of glucosyl residues into starch as catalysed by the plastidial phosphorylase. However, these data do not exclude the possibility that some of the imported glucose 1-phosphate is converted into the plastidial pool of ADPglucose and, subsequently, acts as substrate of starch synthase isozymes. Results obtained with mesophyll protoplasts from leaves of Arabidopsis thaliana are fully consistent with the proposed import of glucose 1-phosphate into the cytosol and, subsequently, into the plastid. However, Arabidopsis mesophyll cells differ from tuber parenchyma cells as, essentially, all the glucose 1-phosphate that has been imported into the chloroplast enters the ADPglucose pool (Fettke ).

Thus, reserve starch biosynthesis in potato tuber amyloplasts is unexpectedly complex and seems to proceed by (at least) two paths: One is initiated by the import of glucose 6-phosphate via the respective transporter (Kammerer ). It includes the plastidial conversion of glucose 6-phosphate to glucose 1-phosphate and, subsequently, the formation of ADPglucose which acts as the principal substrate for starch biosynthesis mediated by several starch synthases. The second path starts with the import of glucose 1-phosphate into the amyloplasts via a selective transporter. Subsequently, the hexosyl moiety is immediately transferred to the non-reducing end of α-glucan chains at the surface of starch granules. This starch synthesizing reaction is catalyzed by the plastidial phosphorylase (Pho1) isozyme(s) and bypasses the ADPglucose pool (Fettke ). Both starch synthases and the plastidial phosphorylase isozyme(s) mediate a series of glucosyl transfer reactions and the kinetics of this type of reactions cannot be described by using the Michaelis–Menten theory (Kartal et al., 2011).

In this communication, the novel carbon flux towards starch and possible interconnections with the established path of starch biosynthesis were studied in more detail. The latter was analysed by incubating tuber discs with mixtures containing [U-14C]glucose 1-phosphate and unlabelled sucrose or [U-14C]sucrose plus unlabelled glucose 1-phosphate. The labelling of starch was determined as affected by the non-labelled sugar/sugar derivative. Slices from freshly harvested or resting tubers were also prepared and major differences in the interaction between the two paths of starch biosynthesis were observed. Furthermore, the two paths were tested for their temperature dependency. For a detailed characterization of the route of starch formation initiated by external glucose 1-phosphate, we examined whether it is a selective or a more general biosynthetic process. Given the wide distribution of the starch granule sizes and the complex internal structure of native starch granules, the question was asked whether or not this path leads to a selective 14C-incorporation into distinct sizes of granules or of distinct α-glucan chains within the granule. Therefore, native starch granules were separated according to size and the 14C-incorporation was quantified for each size class. To monitor the intraparticulate distribution of 14C, starch was solubilized and enzymatically debranched and the labelling of the side chains released was quantified. To identify the plastidial enzyme(s) mediating the conversion of [U-14C]glucose 1-phosphate to starch, transgenic potato plants were used that exhibit a strongly reduced expression of the two plastidial phosphorylase isozymes, designated as Pho1a and Pho1b (Sonnewald ). Data obtained by in situ labelling were compared with in vitro studies performed with isolated native starch granules and recombinant Pho1.

Materials and methods

Plant material

Throughout this study, wild-type potato plants (Solanum tuberosum L. cv. Desiree) and three independently generated transgenic lines were used. Due to an antisense construct, one transgenic line has a decreased level of the cytosolic phosphorylase (Pho2) activity (line 1; Fettke ). The two other transgenic lines (designated as Pho1(a+b)-1 and Pho1b(a+b)-2) contain two antisense constructs each of which is directed against one of the plastidial phosphorylase isoforms, Pho1a and Pho1b. These transgenic lines had been generated during the preparation of a PhD thesis (Duwenig, 1996) and have been used in previous studies (Fettke , 2010). Briefly, for the generation of these lines transgenic potato plants were used that possess an anti-Pho1a construct which is under the control of the 35S promoter and carries kanamycin resistance as a selection marker. Into these lines an additional construct was introduced carrying an anti-Pho1b construct that contains the 35S promoter and, for selection, hygromycin resistance. Despite the low expression of the Pho1b isozyme in potato tubers (Albrecht ), the simultaneous antisense inhibition of both plastidial phosphorylase isozymes (Pho1a and Pho1b) results in a strong inhibition of the expression of the dominant plastidial phosphorylase isozyme, Pho1a. Neither tuber yield nor tuber starch content was found to deviate significantly from the wild-type control under standard growth conditions (Duwenig, 1996).

All plants were grown under controlled conditions [16 h light (300 μE m−2 s−1), 20 °C and 8 h darkness, 17 °C]. For 30 min, both at the beginning and at the end of the light period, illumination was lowered to 150 μE m−2 s−1. Throughout the light–dark cycle, the relative humidity was kept at 50%.

Except where stated, tubers were removed from the mother plants and were immediately used for the preparation of slices.

Incubation of potato tuber discs (in situ incubation)

Discs were prepared from freshly harvested or resting tubers and were then repeatedly washed with a large volume of ice-cold water. Subsequently, the discs were incubated as indicated under continuous agitation. Throughout this study, four experimental procedures for in situ incubation were chosen.

Incubation procedure I (used for simultaneous incubation with labelled and unlabelled compounds at room temperature or at varying temperatures). Three discs (4 mm diameter, thickness around 2 mm) were incubated in 10 ml of the following mixtures which all contain 50 mM citrate-NaOH, pH 6.5, and, in addition, (a) 10 mM unlabelled glucose 1-phosphate, and 37 kBq [U-14C]glucose 1-phosphate; (b) 10 mM unlabelled glucose 1-phosphate, and 37 kBq [U-14C]glucose 1-phosphate plus 10 mM unlabelled sucrose; (c) 10 mM unlabelled sucrose, and 185 kBq [U-14C]sucrose; (d) 10 mM unlabelled sucrose, and 185 kBq [U-14C]sucrose plus 10 mM unlabelled glucose 1-phosphate.

Following 20, 30, 40, or 60 min incubation at temperatures as indicated, the discs were washed and immediately frozen in liquid nitrogen. The starch fraction was isolated according to Fettke . The discs were homogenized using an Ultra Turrax in 5 ml 20% (v/v) ethanol. Following centrifugation (10 000 g for 10 min; 4 °C) the supernatant was discarded and 1 ml water was added to the pellet, mixed, and centrifuged as above. The washing procedure was repeated twice. Finally, the starch fractions were resuspended in 1 ml water and the 14C-label was monitored using liquid scintillation counting.

Incubation procedure II (used for size-dependent starch granule fractionation). Approximately 10 discs (4 mm diameter, thickness around 2 mm) were incubated in 12 ml 50 mM citrate-NaOH, pH 6.5, 2 mM unlabelled glucose 1-phosphate, and 74 kBq [U-14C]glucose 1-phosphate (possessing a specific radioactivity of 9.21 GBq mmol−1). Following 30 or 40 min incubation at RT, the discs were washed three times with 20 ml each of 20% (v/v) ethanol and were then homogenized in 20 ml 20% (v/v) ethanol using an Ultra Turrax (2 times 3 s each). The homogenate was centrifuged (3 min, 100 g) and the pellet was washed four times with 20% (v/v) ethanol (30 ml each) followed by centrifugation (as above). The pellet was resuspended in water and was then passed through a nylon net (100 μm pore size). In the filtrate, starch granules were collected by centrifugation (as above). Finally, the starch granules were suspended in 2 ml water and were used for size-dependent fractionation.

Incubation procedure III (used for debranching of the solubilized starch). Discs (5 mm diameter, thickness approximately 2 mm) were incubated in 50 μl 100 mM citrate-NaOH, pH 6.5 and 37 kBq [U-14C]glucose 1-phosphate (specific activity: 9.21 GBq mmol−1) for 20 min at 30 °C. Subsequently, unlabelled glucose 1-phosphate was added to give a final concentration of 4 mM and incubation was continued as stated at 30 °C. Discs were then washed six times (1 ml each) with water and homogenized using a bead mill. Starch granules were pelleted by centrifugation (as above) and were washed six times with 20% (v/v) ethanol (1 ml each) followed by centrifugation. Finally, 500 μl 200 mM KOH was added to each pellet and the mixtures were kept at 95°C for 1 h. For neutralization, 2 M HCl was added.

Incubation procedure IV (used for incubation in the presence of isolated starch granules and/or buffer-soluble proteins). Discs (5 mm diameter, thickness approximately 2 mm) were incubated in 70 μl 100 mM citrate-NaOH, pH 6.5, 10 mM unlabelled glucose 1-phosphate, and 9.25 kBq [U-14C]glucose 1-phosphate (specific activity: 9.21 GBq mmol−1) for 15 min at 30 °C. As stated, some discs were incubated in a medium that contained, in addition to the compounds listed above, 75 μg buffer-soluble proteins extracted from wild-type potato tubers. Alternatively, some discs were incubated in the presence of both buffer-soluble proteins (75 μg each; as above) and 100 μg native starch granules (see text). Otherwise, all samples were treated identically. Following incubation, discs were washed six times with water (1 ml each) and were homogenized using a bead mill. Starch granules were pelleted by centrifugation (as above) and were washed six times with 20% (v/v) ethanol (1 ml each) followed by centrifugation. Subsequently, the starch was solubilized by adding 500 μl 200 mM KOH each and heating (20 min at 95 °C). Finally, all samples were centrifuged (5 min, 10 000 g) and the supernatants were used for liquid scintillation counting.

Protein-related biochemical techniques

Heterologous expression and purification of plastidial phosphorylase:

Heterologous expression of the plastidial Pho1 isozyme (Pho1) from rice in E. coli and purification of the recombinant protein were performed as described elsewhere (Fettke ).

Extraction and quantification of buffer-soluble proteins:

Throughout this study, two procedures were used for the extraction of buffer-soluble proteins designated as protein extraction procedure A and B. In some experiments, protein extraction was followed by a fractionating precipitation with ammonium sulphate (protein extraction procedure C).

Protein extraction procedure A. For the quantification of enzyme activities, native PAGE, and 14C-incorporation using ADP-[U-14C]glucose into α-glucans, tuber material was homogenized in grinding medium A [100 mM HEPES-NaOH pH 7.5, 1 mM EDTA, 5 mM dithioerythritol (DTE), 0.5 mM phenylmethylsulphonyl fluoride, 0.1% (w/v) sodium sulphite, 0.075% (w/v) sodium disulphite, and 10% (v/v) glycerol] using an Ultra Turrax (2 times, 5 s each). The homogenate was passed through a nylon filter (100 μm pore size) and the filtrate was centrifuged (12 min at 14 000 g, 4 °C). The supernatant was either directly used for protein quantification, enzyme activity measurements, and native PAGE or was frozen in liquid nitrogen and stored at –80 °C until use.

Protein extraction procedure B. For testing the effect of buffer-soluble proteins on the [U-14C]glucose 1-phosphate-dependent in situ labelling of starch (see Supplementary Fig. S1 at JXB online), buffer-soluble proteins were extracted following procedure B. Tuber material was homogenized in grinding buffer B [100 mM citrate-NaOH pH 6.5, 2 mM dithioerythritol (DTE), and 0.5 mM phenylmethylsulphonyl fluoride] using an Ultra Turrax (2 times, 5 s each). The resulting homogenate was passed through a nylon net (100 μm pore size) and the filtrate was centrifuged (20 000 g for 5 min). The supernatant was directly used.

Protein extraction procedure C (including a fractionated precipitation with ammonium sulphate). Buffer-soluble proteins were extracted from tubers harvested from the anti-Pho2-antisense line (Fettke ) following protein extraction procedure A (see above; but glycerol was omitted in the buffer). Subsequently, the extracted proteins were fractionated by precipitation with solid (NH4)2SO4 (40–55% saturation). For desalting, ultrafiltration (MWCO, 50 kDa) was applied. Glycerol [final concentration 20% (v/v)] was added to the desalted sample and the protein preparation was stored at –80 °C until use.

For extraction procedures A to C, proteins were quantified using the micro version of the Bio-Rad protein assay kit according to Bradford (1976).

Native PAGE and staining for glucosyl transferase activities. A discontinuous native PAGE was used. The separation gel contained 7.5% (w/v; T) acrylamide-bisacrylamide, and, except where stated, 0.2% (w/v) glycogen from oysters (type II; Sigma, Taufkirchen, Germany). Proteins were loaded as indicated. During electrophoresis, the temperature was kept at 4 °C. For phosphorylase zymograms, separation gels were washed for 20 min with 100 mM citrate buffer (pH 6.5). Subsequently, gels were incubated in a mixture containing 20 mM glucose 1-phosphate and 100 mM citrate-NaOH pH 6.5. During incubation (30 min), the temperatures were as indicated. For soluble glucan synthase zymograms, gels were incubated in a mixture consisting of 50 mM Tricine-KOH (pH 8.0), 0.025% (w/v) bovine serum albumin, 5 mM dithioerythritol, 2 mM EDTA, and 25 mM potassium acetate. After 20 min, the incubation mixture replaced by a fresh medium and, in addition, 1 mM ADPglucose was applied. Incubation was for 1 h at the temperatures as indicated. For both phosphorylase and starch synthase activity, gels were stained with iodine and the intensity of the staining was quantified using the AIDA software.

Photometric phosphorylase activity assay. Phosphorylase activity was monitored using a continuous spectrophotometric assay according to Steup (1990). Assays were performed at 30 °C using saturated levels of soluble starch.

In vitro 14C-labelling of native starch granules and soluble carbohydrates

Throughout this study, four in vitro labelling techniques were applied.:

Labelling method I ( for in vitro labelling of native starch granules varying in size). In a total volume of 1.2 ml, the reaction mixture contained 50 mM citrate-NaOH, pH 6.5, 2 mM unlabelled glucose 1-phosphate, 74 kBq [U-14C]glucose 1-phosphate (specific activity: 9.21 GBq mmol−1), and 80 mg native starch granules isolated from potato tubers. Following pre-incubation for 3 min at 37 °C, recombinant plastidial phosphorylase (Pho1) from Oryza sativa L. was added (3 μg each). The reaction was further incubated at 37 °C and agitation continued until the reaction was terminated as indicated by adding SDS to give a final concentration of 2% (w/v). The starch granules were washed three times with water (30 ml each) and used for size-dependent fractionation.

Labelling method II (for enzymatic debranching after in vitro labelling). In a total volume of 80 μl, the reaction mixture contained 50 mM citrate-NaOH, pH 6.5, 5 mM unlabelled glucose 1-phosphate, 18.5 kBq [U-14C]glucose 1-phosphate (specific activity: 9.21 GBq mmol−1), and 1 mg native starch granules. Following pre-incubation for 3 min at 37 °C, 4.5 μg recombinant Pho1 from Oryza sativa L. was added. Incubation was continued for 5, 10 or 60 min and the reaction was then terminated by adding SDS to give a final concentration of 2% (w/v). The starch was washed three times with 20% (v/v) ethanol, three times with water, and then 105 μl of 200 mM KOH was added to the pelleted starch granules. Following incubation at 95 °C for 20 min, the samples were neutralized with 2 M HCl and processed further by using isoamylase treatment, HPEAC-PAD and liquid scintillation counting as described below.

Labelling method III (temperature-dependency of starch synthase activity). Buffer-soluble proteins (protein extraction procedure A; see above) and native starch granules were obtained from the same tuber. Pelleted starch granules were repeatedly washed with grinding medium A and were finally diluted to give a concentration of 172 mg ml−1. A series of incubation mixtures (100 μl each) was prepared consisting of either 4.3 or 8.6 mg native starch, either 40 or 80 μg buffer-soluble proteins, 0.5 M sodium citrate pH 8.0, 25 mM unlabelled ADPglucose, 1.85 kBq ADP-[U-14C]glucose. Reaction mixtures were incubated at 10, 15, 20, 25, and 30 °C. At intervals (5, 10, and 15 min), the reaction was terminated by adding ice-cold ethanol to give a final concentration of 33% (v/v). Starch granules were separated by centrifugation (1 min at 14 000 g). The supernatant was analysed by thin-layer chromatography and phosphor imaging (Hejazi ). The pelleted starch granules were washed three times with 500 μl each of ice-cold 20% (v/v) ethanol. Finally, radioactivity in the starch granules was quantified by liquid scintillation counting.

Labelling method IV (temperature-dependency of plastidial phosphorylase activity, Pho1). To minimize interference by the cytosolic phosphorylase as well as by non-selective phosphatase activity, buffer-soluble proteins were extracted from tubers of transgenic lines possessing a reduced level of the cytosolic phosphorylase isozyme, Pho2 and were subjected to fractionating precipitation by ammonium sulphate (40–55% saturation; Protein extraction procedure C; see above). Native starch granules were isolated from freshly harvested wild-type tubers (see above) and were washed five times with grinding medium A (see above). Subsequently, the washed native starch granules were diluted prior to use. The incubation mixture (total volume 90 μl each) contained 16 mg native starch, 50 mM glucose 1-phosphate, 4.4 kBq [U-14C]glucose 1-phosphate, 40 mM citrate-NaOH pH 6.5, 0.5 mM EDTA, 2.5 mM dithioerythritol (DTE), 66 μg protein, and 0.25 mM phenylmethylsulphonyl fluoride. Reaction mixtures were kept either at 10, 15, 20, 25, or 30 °C. At intervals (5, 10, and 15 min), the reaction was terminated by adding ice-cold ethanol [to give a final concentration 30% (v/v)]. Following centrifugation (1 min at 14 000 g), the supernatant was collected and subjected to thin layer chromatography. Radioactivity was quantified by phosphor imaging (Hejazi ).

The pelleted starch granules were washed three times with ice-cold 20% (w/v) ethanol (500 μl each) and, subsequently, three times with ice-cold water. Finally, 14C-content was quantified by liquid scintillation counting.

Carbohydrate-related techniques

Carbohydrate quantification: Following acid hydrolysis (2 M TFA; 2 h at 100 °C), carbohydrates were quantified according to Waffenschmidt and Jaenicke (1987).

Preparation of larger quantities of native starch granules. Potato tubers were cut into small pieces and homogenized in extraction buffer [100 mM HEPES-KOH, pH 8.0, 1 mM EDTA, 5 mM dithioerythritol (DTE), and 0.5 mM phenylmethylsulphonyl fluoride]. The starch granules were allowed to settle and washed several times with extraction buffer.

Starch fractionation and multisizer analyses. Following isolation, the native starch granules were sequentially passed through three filters possessing decreasing pore sizes (Filcon® with cone; pore size 50 μm, 30 μm, 20 μm; Süd-Laborbedarf GmbH, Gauting, Germany). Following filtration, each filter was washed ten times with 1 ml water each. The filtrates were combined, centrifuged (3000 g, 3 min) and 1 ml water was added to each pellet. The resulting suspensions were then loaded on the next filter unit. All retentates were suspended in 1 ml water each and the filters were then washed four times (1 ml water each) and were combined. Finally, the four starch granule fractions were centrifuged (3000 g, 3 min) and each pelleted starch fraction was resuspended in a small volume of water (100–1000 μl).

Both the size distribution and the volume per particle were determined using a Coulter Counter Multisizer 3 equipped with a 100 μm aperture tube (range from 2–60 μm; Beckman Coulter GmbH, Krefeld, Germany). Prior to the measurements, 100 μl of the starch suspensions were added to 70 ml of Coulter Isoton II solution (Beckman Coulter).

Scanning electron microscopy (SEM). Isolated native starch granules were coated with gold and analysed by scanning electron microscopy using a Quanta apparatus (Philips).

Starch debranching with isoamylase and HPAEC-PAD analyses. Solubilized starch samples were brought to pH 4.0 by the addition of 100 mM acetate-HCl, pH 4.0 (final concentration 10 mM) and 5 units isoamylase (Megazyme; Bray, Ireland) were added. Samples were incubated for 30 min at 40 °C and, subsequently, the reaction was terminated by heating (5 min at 95 °C). Samples were dried in vacuo and dissolved in 200 μl water each.

Ion chromatography was performed with a DX 600 system equipped with a CarboPac™-PA1 column (Dionex, Idstein, Germany). For oligosaccharide separation the column was equilibrated for 10 min with 5 mM sodium acetate dissolved in 100 mM NaOH. Following sample application, a linear gradient of sodium acetate (5–500 mM in 100 mM NaOH; 30 min) and, finally, 500 mM sodium acetate in 100 mM NaOH for 10 min were applied. Throughout the chromatography the flow rate was 1 ml min−1. Eluate fractions (833 μl each) were collected and used for liquid scintillation counting.

Isolation of soluble carbohydrates. Potato tuber discs were incubated using incubation procedure II with some modifications. Following washing, water (1 ml each; six times), and 20% (v/v) ethanol (1.4 ml each) was added and the discs were mechanically broken using a bead mill. The resulting suspensions were centrifuged (20 000 g, 10 min, 4 °C). Supernatants were collected and dried in vacuo. The dry samples were dissolved in water (200 μl each) and were then subjected to HPAEC-PAD analysis as well as to liquid scintillation counting.

Preparation of ADP-[U-. ADP-[U-14C]glucose was synthesized enzymatically using recombinant ADPglucose pyrophosphorylase (AGPase) from E. coli. In a reaction mixture (final volume of 100 μl) consisting of 5 mM TRIS-HCl pH 7.5, 1 mM MgCl2, [U-14C]glucose 1-phosphate (0.13 μmol; 1.4 MBq), and 0.15 μmol unlabelled ATP, the two recombinant enzymes from E. coli [35 μg AGPase and 5 units pyrophosphatase (Sigma, Taufkirchen, Germany)] were incubated for 3 h at 30 °C. Efficiency of the reaction was ensured by thin layer chromatographic separation of the reaction mixture. Following termination (5 min at 95 °C), the reaction mixture was centrifuged and the supernatant was collected. Subsequently, the concentration of the radioactivity was diluted to 2.1 kBq μl−1 with water and aliquots were stored frozen at –20 °C until use.

Results

Carbon fluxes towards starch from externally supplied glucose 1-phosphate and sucrose are closely interconnected

Using discs prepared from freshly harvested potato tubers, carbon fluxes towards starch were initiated by externally supplied glucose 1-phosphate and sucrose. To determine a possible interconnection of the two fluxes, discs prepared from freshly harvested tubers were incubated at room temperature in a medium containing [U-14C]glucose 1-phosphate, [U-14C]glucose 1-phosphate plus unlabelled sucrose, [U-14C]sucrose, or [U-14C]sucrose plus unlabelled glucose 1-phosphate (incubation procedure I; see the Materials and methods). After 20 min or 40 min incubation reaction was terminated and the incorporation into starch was quantified.

When applying equimolar concentrations of either [U-14C]glucose 1-phosphate or [U-14C]sucrose, 14C-labelling of starch was more efficient with glucose 1-phosphate compared with that obtained with sucrose. These data concur with previously published data (Fettke ). In the presence of unlabelled sucrose, the conversion of [U-14C]glucose 1-phosphate to starch was even faster. When unlabelled glucose 1-phosphate was added to [U-14C]suc-rose, an approximately 5 fold enhancement of the incorporation into starch was observed. The stimulation exerted by the respective unlabelled carbohydrate was essentially the same for both incubation periods (Fig. 1).

Fig. 1.

In situ 14C-incorporation into native starch using various externally supplied carbohydrates. Slices prepared from freshly harvested tubers were incubated (Procedure I; see Materials and methods) in [U-14C]sucrose (14C-Suc), [U-14C]sucrose plus unlabelled glucose 1-phosphate (14C-Suc+G1P), [U-14C]glucose 1-phosphate (14C-G1P), or [U-14C]glucose 1-phosphate plus unlabelled sucrose (14C-G1P+Suc). Final concentration of each carbohydrate was 10 mM. Following 20 or 40 min, native starch granules were isolated and the 14C contents were quantified. Average values from three independent experiments ±SD are given.

The stimulatory effect is unexpected as the addition of unlabelled sugars or sugar derivatives is expected to lower the specific radioactivity of those intermediates that are commonly used by starch synthesizing paths. If so, the rates of the 14C incorporation are diminished even if the actual fluxes remain unchanged. Thus, the experimental design may even lead to an underestimation of potential synergistic effects. However, the experimental design clearly shows that the stimulatory effects observed cannot be attributed to any unidirectional effect, such as a selective inhibition of the AGPase which would favour the glucose 1-phosphate-dependent path. The addition of both unlabelled glucose 1-phosphate (in the presence of [U-14C]sucrose) and sucrose (in the presence of [U-14C]glucose 1-phosphate) enhances the 14C-incorporation into starch and, therefore, each route benefits from the addition of the unlabelled compound.

In summary, the data presented in Fig. 1 clearly demonstrate a mutual stimulation of the two paths of starch biosynthesis that are initiated by externally supplied glucose 1-phosphate or sucrose.

Interconnection of the two fluxes depends on the physiological state of the tubers

As evidenced by two results, the stimulation of either carbon flux strongly depends on the physiological state of the tuber. First, enhancement is large in freshly harvested tubers but strongly declines when, following harvest, the tubers have been stored at room temperature. When tubers were sliced after 14 d of storing, the 14C-carbon fluxes towards starch were almost unaffected by the respective unlabelled additive (incubation procedure I; Table 1). Thus, following harvest, the metabolic state of the tubers rapidly changes resulting in a complete loss of the synergistic interaction. Second, when tuber slices were used from several freshly harvested tubers the enhancement of the 14C-carbon fluxes towards starch varied strongly. By contrast, 14C-incorporation into starch was less variable when tuber slices from freshly harvested tubers were incubated with only [U-14C]glucose 1-phosphate or [U-14C]sucrose. This strongly suggests that the interconnection of the two paths is more flexible than the rate of each of the two paths itself. It should be noted that tubers freshly harvested from the plants are likely to exhibit some heterogeneity in their physiological state (such as sink activity and/or the actual rate of starch biosynthesis). Growth processes of the various tubers attached to a single plant are far from being synchronized but difficult to assess. When tubers had been stored at room temperature for several days, the simultaneous incubation with both glucose 1-phosphate and sucrose results in a considerably lower variation of the 14C-incorporation.

Table 1.

Stimulation of the carbon fluxes depending on the physiological state of the tubers

	14C-Suc+G1P/14C-Suc	14C-G1P+Suc/14C-G1P
Freshly harvested tubers	2.0–4.1	1.5–1.7
Tubers stored for 1 week	1.5–2.0	1.1–1.3
Tubers stored for 2 weeks	0.9–1.1	1.1–1.2

For each physiological state, 2–4 independent experiments were performed. The incubation time was 30 or 60 min and the incubation temperature was 20 °C. The lowest and the highest ratios are given. The tubers were stored at room temperature.

The two carbon fluxes and their interconnection vary depending on the temperature

The temperature dependence of the two carbon fluxes towards starch was studied by using two approaches. When using the first approach, slices from freshly harvested potato tubers were incubated (Incubation procedure I) for 30 min with either [U-14C]glucose 1-phosphate or [U-14C]sucrose at varying temperatures ranging from 10 °C to 30 °C and the 14C-incorporation into starch was quantified. Values obtained are given in relative units and are plotted against the respective incubation temperature (Fig. 2A). For both 14C-labelled externally applied sugar (derivative), 14C-incorporation increases with temperature but the slopes are different. The sucrose-dependent flux depends more strongly on the temperature compared with that of glucose 1-phosphate.

Fig. 2.

In situ 14C-incorporation into native starch as affected by the incubation temperature. (A) Slices prepared from freshly harvested tubers were incubated (incubation procedure I) for 30 min at different temperatures in a mixture containing either [U-14C]sucrose (10× Suc) or [U-14C]glucose 1-phosphate. The final concentration of each of the two compounds was 10 mM. Please note that, because of the far lower sucrose-dependent incorporation into starch, the values from [U-14C]sucrose have been multiplied by a factor of 10. Average values from three independent experiments ±SD are given. (B) Slices prepared from freshly harvested potato tubers were incubated for 30 min at different temperatures in a mixture containing either [U-14C]glucose 1-phosphate (14C-G1P), [U-14C]glucose 1-phosphate plus unlabelled sucrose (14C-G1P+Suc), [U-14C]sucrose (14C-Suc) or [U-14C]sucrose plus unlabelled glucose 1-phosphate (14C-Suc+G1P). The final concentration of each carbohydrate was 10 mM. Following incubation, native starch granules were isolated and the 14C-contents were monitored. Values were normalized by setting the 14C incorporation without unlabelled compounds to unity. Average values from four independent experiments ±SD are given.

In the second approach, aliquots of slices prepared from freshly harvested tubers were incubated (incubation procedure I) with either of four incubation mixtures containing either [U-14C]glucose 1- phosphate, [U-14C]glucose 1-phosphate plus equimolar unlabelled sucrose, [U-14C]sucrose or [U-14C]sucrose plus equimolar unlabelled glucose 1-phosphate. Each aliquot was incubated at temperatures ranging from 10 °C to 30 °C. Subsequently, 14C-incorporation into starch was quantified. The ratio of 14C-incorporated into starch in the presence of the unlabelled compound to that observed with the respective 14C-labelled compound alone is plotted against the temperature. Data were normalized by setting the 14C-incorporation into starch, as obtained with either [U-14C]glucose 1-phosphate or [U-14C]sucrose alone at 20 °C to unity (Fig. 2B). Therefore, enhancement of 14C-incorporation into starch that is observed in the presence of the respective unlabelled compound results in a value higher than 1, whereas any inhibitory effect would be indicated by values lower than 1 (for details see Legend of Fig. 2). By using this approach, the temperature-dependency of the synergy between the two paths towards starch can be determined.

At low temperatures (ranging from 10–15 °C), 14C-labelling of starch by [U-14C]sucrose is strongly enhanced by the addition of unlabelled glucose 1-phosphate. However, at increasing temperatures, the enhancing effect decreases. The temperature dependency of the stimulation of 14C-incorporation from [U-14C]glucose 1-phosphate by unlabelled sucrose is largely different. Unlabelled sucrose enhances the 14C-incorporation at higher temperatures but is essentially undetectable at 15 °C or 10 °C (Fig. 2B).

In vitro activities of the starch synthases and of the plastidial phosphorylase (Pho1) differ in temperature dependence

The entire process of 14C-incorporation into starch that is initiated by externally applied sucrose or the anomeric glucose phosphate is a complex process that includes several enzyme-catalysed reactions and transporter-mediated uptake into both the cytosol as well as in the plastidial stroma (Fettke , 2011). In principle, all these steps may contribute to the largely difference temperature dependency but most of them are difficult to analyse separately. This limitation does not apply for the terminal step, the actual glucan synthesizing reaction that is mediated either by the plastidial phosphorylase isozyme (Pho1 or, in Arabidopsis thaliana, PHS1; EC 2.4.1.1) or the various buffer-soluble starch synthase isozymes (EC 2.4.1.21). Both types of glucosyl transferases differ in the glucosyl donor used: phosphorylases strictly rely on glucose 1-phosphate whereas all starch synthases from higher plants utilize ADPglucose.

In order to identify distinct reactions within the complex process of starch biosynthesis which largely differ in temperature dependence, two approaches were chosen. First, the in vitro 14C-incorporation, mediated either by starch synthase or phosphorylase activity, was tested at different temperatures. Secondly, quantitative zymograms of the starch synthase or phosphorylase activity were obtained following the incubation of the separation gels at different temperatures.

To monitor the temperature dependence of 14C-transfer in vitro, as mediated by the starch synthases, buffer-soluble proteins were extracted from freshly harvested tubers and incubated with native starch granules isolated from the same tuber. ADP-[U-14C]glucose served as the glucosyl donor. The experiments were designed to detect 14C-incorporation as comprehensively as possible. Therefore, the reaction mixtures contained a high concentration of citrate.

Furthermore, both the 14C-incorporation into the soluble phase and into the starch granules was monitored. The latter can be due to soluble starch synthases acting on the granules and also to the granule-bound starch synthase (GBSS) activity. 14C-content of the starch granules was quantified by liquid scintillation counting. Aliquots of the supernatants were subjected to thin layer chromatography and, subsequently, radioactivity was measured by Phosphor imaging. At intervals (5 min each) the reactions were terminated. For each temperature, reaction mixtures were prepared that contained two concentrations of buffer-soluble proteins and two amounts of native starch but the ratio between the protein and starch levels remained unchanged. Under otherwise unchanged conditions, essentially the same glucosyl incorporation per protein was observed, indicating that throughout the entire incubation the glucosyl transfer reaction to be monitored proceeds without any obvious limitation by either ADP-[U-14C]glucose or starch.

In the supernatant of all the samples, the only compound containing detectable amounts of 14C is ADP-[U-14C]glucose (data not shown). Thus, none of the conditions tested results in complete consumption of the glucosyl donor, ADPglucose. Furthermore, under the conditions used, any interfering hydrolytic enzyme activity does not noticeably affect the 14C-incorporation into starch as monitored in the pellet fraction. This conclusion is safe as, otherwise, 14C-labelled glucans (or even 14C-glucose) would have been detectable by thin layer chromatography and phosphor imaging.

By contrast, the pellet fractions did contain 14C-label and the amount increases with time. Thus, under the reaction conditions used, starch synthases mediate glucosyl transfer reactions (utilizing ADPglucose as the glucosyl donor) to the surface or the interior of the pelletable fraction, i.e. native starch granules. 14C-incorporation can be attributed to the activity of GBSS and/or soluble starch synthases acting on insoluble α-glucans. More importantly, 14C-incorporation is low at 10 °C or 15 °C but increases at higher temperatures without reaching a plateau (Table 2). Thus, the temperature dependence of the starch synthase activity as monitored in vitro is similar to that of the sucrose-starch conversion observed in situ (Fig. 2).

Table 2.

In vitro 14C-incorporation into α-glucans mediated by the starch synthase activity at varying temperatures

Time (min)	Incubation temperature (°C)
	10	15	20	25	30
5	0.59±0.09	0.67±0.11	0.94±0.15	1.17±0.10	1.46±0.14
10	0.87±0.12	1.03±0.17	1.47±0.15	2.10±0.21	2.79±0.21
15	0.89±0.10	1.34±0.22	2.24±0.44	3.21±0.25	4.10±0.72

Buffer-soluble proteins were extracted from growing potato tubers. From the same tubers, native starch granules were isolated. In a total volume of 100 μl, buffer-soluble proteins (40 or 80 μg) were incubated with native starch granules (4.3 or 8.6 mg). As indicated, incubation mixtures were kept at five temperatures (10, 15, 20, 25, and 30 °C). At intervals (5 min each), reaction was terminated and the mixtures were separated into native starch and supernatant by centrifugation. 14C-incorporation into starch was quantified by scintillation counting. 14C-incorporation into the pelleted starch granules is given as nmol glucosyl residues μg−1 soluble protein (n=6; SD).

When similar in vitro 14C-labelling experiments are performed to monitor the Pho1-mediated glucosyl transfer, two aspects should be considered. First, the buffer-soluble proteins extracted from potato tubers contain both the plastidial and the cytosolic phosphorylase isozymes. Although the phosphorylase activity pattern is dominated by the plastidial phosphorylase activity, the cytosolic counterpart may differ in temperature dependence and, consequently, for some of the temperatures tested a significant interference cannot be excluded. Therefore, proteins were extracted from tubers of a transgenic line possessing a strongly reduced level of the cytosolic phosphorylase, Pho2. Secondly, tubers contain a significant amount of non-selective phosphatase activity which could act on glucose 1-phosphate. In order to minimize this possible interference, a high concentration of glucose 1-phosphate (50 mM) was applied. Furthermore, the extracted proteins were fractionated by ammonium sulphate precipitation and only those proteins insoluble between 40–55% (NH4)2SO4 saturation were used for incubation with native starch granules that had been isolated from freshly harvested wild-type tubers. Throughout the incubation period, most of the glucose1-phosphate was retained in the reaction mixtures (data not shown). For all reaction periods and all temperatures tested, the supernatant fractions did contain detectable amounts of 14C only as glucose 1-phosphate but not as free glucose or maltodextrins (data not shown). Therefore, in the soluble phase, no action of the phosphorylase on α-glucans is detectable. Furthermore, no release of 14C-labelled glucosyl residues from the starch granules by any hydrolases is detectable. By contrast, a time-dependent 14C-incorporation into the pellet fraction is clearly detectable (Table 3). However, the maximal rate of 14C-incorporation is observed at approximately 20 °C and further increase in the temperature results in essentially the same 14C-incorporation into starch (Table 3). Thus, in vitro labelling of native starch granules, as mediated by the plastidial phosphorylase, Pho1, exhibits a very similar temperature dependence as the in situ labelling studies (Fig. 2; see below).

Table 3.

In vitro 14C-incorporation into α-glucans mediated by the phosphorylase activity at varying temperatures

Time (min)	Incubation temperature (°C)
	10	15	20	25	30
5	29.8±1.4	56.8±7.2	78.8±1.2	74.5±3.2	81.9±3.6
10	41.2±3.0	63.8±4.1	92.3±4.9	100.8±3.9	105.7±5.8
15	47.8±2.2	92.0±1.2	114.7±3.41	112.8±8.1	120.0±2.3

Buffer-soluble proteins were extracted from tubers of a transgenic line that, due to an antisense construct, possesses a strongly reduced expression of the cytosolic phosphorylase, Pho2. Buffer-soluble proteins were subjected to precipitation with (NH4)2SO4 (40–55% saturation; protein extraction procedure C). Native starch granules were isolated from freshly harvested wild-type tubers. Labelling method IV was used. Reaction mixtures were kept at different temperatures (10, 15, 20, 25, and 30 °C). At intervals (5 min each), reaction was terminated and the reaction mixtures were separated into native starch and supernatant by centrifugation. 14C-incorporation into starch was quantified by scintillation counting. 14C-incorporation into the pelleted starch granules is given as nmol glucosyl residues (n=3; SD)

Soluble starch synthase and phosphorylase activities differ in the temperature dependence as revealed by quantitative zymograms

Both starch synthase and phosphorylase activities can easily be detected by a native PAGE followed by the enzymatic elongation of α-glucans and iodine staining. Sites of elongation can easily be visualized as dark bluish or brownish bands at an essentially unstained background provided the non-elongated glucosyl acceptor molecules poorly interact with iodine. Due to its frequent and regular branching, native glycogen contains only short linear α-glucan chains and, therefore, fulfils this requirement. In principle, following electrophoresis, the separation gel can be incubated in a glycogen-containing mixture. Alternatively, glycogen can be added to the polymerization mixture during preparation of the polyacrylamide gel. The latter procedure offers the advantage that the elongating enzymes have immediate access to the glucosyl acceptors. If the temperature dependence of the catalytic activities is to be measured by using the zymogram technique, the intensity of the stained bands must be linearly related to the enzyme activity applied (quantitative zymograms).

Soluble starch synthase isozymes

Compared with the phosphorylase zymograms, larger amounts of buffer-soluble proteins were applied per lane in order to obtain a clearly visible staining of the products of the soluble starch synthase isozymes. Except at high temperatures, linearity was given in a more narrow range of protein quantities (100–150 μg protein per lane). Following native PAGE, separation gels were incubated with ADPglucose at temperatures ranging from 5–35 °C and were then stained by iodine. Staining intensities were normalized to the respective staining of the bands obtained at 20 °C. Normalized intensities were plotted against the respective temperature during incubation (Fig. 3). As the two bands of starch synthase activity possess the same temperature dependence (data not shown), both intensities were summed up. As opposed to the Pho1 activity, the total starch synthase activity (as well as that of each band) doubles in the range of 20–35 °C (Fig. 3). Minor starch synthase isozymes undetectable in the separation gel may also be involved in the in situ process of starch biosynthesis. Nevertheless, these data are consistent with the in situ labelling experiments shown in Fig. 2.

Fig. 3.

Activities of glucosyl transferases from potato tuber crude extracts as affected by varying temperatures. The activity of the plastidial phosphorylase (Pho1) and that of soluble starch synthases (SS) was quantified by native PAGE using a glycogen-containing separation gel. Following electrophoresis, separation gels were incubated at temperatures ranging from 5 °C to 35 °C. The intensity of the iodine-stained bands was quantified. The relative intensity (based on the intensity of the respective enzyme activity at 20 °C) was plotted against the temperature. (Insert) The linearity of the quantitative PAGE is exemplified by two incubation conditions. For phosphorylase activity, 2 μg and 4 μg were applied. For soluble starch synthase activity, the amounts applied to a lane were 100 μg and 150 μg.

Phosphorylase isozymes

Potato tubers contain an exceptionally high specific phosphorylase activity and, therefore, low amounts of buffer-soluble proteins (2 or 4 μg) were applied per lane. Following native PAGE, separation gels were incubated at temperatures ranging from 5–35 °C. Potato tubers contain two phosphorylase isozymes, designated as Pho1 and Pho2. Pho1 is the dominant plastidial isozyme whereas the cytosolic isozyme, Pho2, contributes relatively little to the total enzyme activity. The two isozymes differ largely in glucan specificity and, therefore, can easily be distinguished. When subjected to native PAGE in a glycogen-containing separation gel, Pho2 is essentially immobile whereas the mobility of Pho1 remains almost unchanged (Fettke ). For both isozymes, linearity between the amount of enzyme activity and the intensity of the stained band of the zymograms was observed in a relatively wide range. The two phosphorylase isozymes differ in the temperature dependence of the activity. The activity of the plastidial isozyme, Pho1, increases with temperature until approximately 20 °C and remains essentially unchanged at higher temperatures (Fig. 3). By contrast, the activity of the cytosolic isozyme, Pho2, increases essentially linearly with the temperature (see Supplementary Fig. S2 at JXB online). The effects are observed both in glycogen-containing and glycogen-free separation gels (data not shown). This result is important as it is unlikely that in both types of separation gels any interfering enzyme activities exactly comigrate with Pho1. Furthermore, the diffusion of glycogen into the separation does not noticeably affect the temperature dependence observed.

The in situ flux into starch initiated with glucose 1-phosphate does not discriminate between granule sizes

The data presented above strongly suggest that the path of starch biosynthesis initiated by externally applied glucose 1-phosphate is closely associated with the route that converts external sucrose to starch. If so, the flux of 14C into starch from labelled glucose 1-phosphate is expected to be non-selective. Incorporation into starch should be unaffected by the starch granule size and, furthermore, within a single starch granule a wide range a glucosyl acceptors is expected to be functional. Therefore, labelling of the starch granules and the intraplastidial acceptor sites of the glucosyl residues were studied in more detail.

Starch granules isolated from growing tubers of potato (Solanum tuberosum L.) possess a relatively complex morphology and cover a wide range of sizes (Fig. 4A). Therefore, the question was asked whether 14C-incorporation from externally supplied glucose 1-phosphate is a general phenomenon that results in incorporation irrespective of granule size. To answer this question empirically, a procedure was established allowing the fractionation of isolated starch granules according to size. Despite their relatively complex morphology native starch granules were separated according to size by sequentially passing through a series of commercially available filters with decreasing pore sizes (50, 30, and 20 μm). Subsequently, the size distribution of the four starch granule fractions was analysed by using a multisizer. Due to the compactness of native starch, the volume and the number of the individual particles can easily be monitored using the Coulter Counter technique. In a first approximation, particles were considered to be spherical and both the resulting diameter (as a measure of the particle size; Fig. 4B) and the surface area was calculated. The values obtained by this calculation are obviously not precise as native starch granules deviate from a sphere (Fig. 4A) and, presumably, granule morphology impedes a more efficient physical separation of particles differing in size. Nevertheless, sequential filtration results in the resolution of distinct yet overlapping size classes (Fig. 4B). The smallest and the largest particle fractions possess a mean diameter of 18 μm and 33 μm, respectively.

Fig. 4.

Potato tuber starch fractionation according to size. (A) Scanning electron micrograph of native starch granules isolated from growing wild-type potato tubers. Bar equivalent to 100 μm. (B) Size distribution of four starch granules fractions from wild-type tubers as revealed by multisizer. The starch granules were separated by successive membrane filtrations. Black curve: retentate of the 50 μm filter; dark grey: retentate of the 30 μm filter; medium grey: retentate of the 20 μm filter; light grey: filtrate of the 20 μm filter unit.

In the next series of experiments, this separation procedure was applied to analyse native starch granules isolated from potato tuber discs that had been incubated with [U-14C]glucose 1-phosphate. After incubation for either 30 min or 40 min at RT (Incubation procedure II), discs were washed and the 14C-containing starch granules were isolated. Subsequently, the isolated starch particles were subjected to a series of filtration steps as described for Fig. 4B. An aliquot from each of the various fractions was used for multisizer analyses. In the residual suspensions of the size-fractionated starch granules the 14C-contents were quantified by liquid scintillation counting (Fig. 5).

Fig. 5.

In situ incorporation of glucosyl-residues derived from [U-14C]glucose 1-phosphate into starch granules differing in size. Tuber discs prepared from wild-type potato tubers were incubated with [U-14C]glucose 1-phosphate for 30 min and 40 min, respectively (in situ experiment; incubation procedure II). Following incubation, discs were washed and the native starch granules were isolated. Granules were then fractionated using a series of filteration steps (see Fig. 4B). Aliquots of the four fractions were analysed using a multisizer. Both the size and the surface of the particles were calculated assuming a spherical shape. In addition, the 14C-contents of the four size classes were determined using liquid scintillation counting. For comparison, native starch granules were isolated from wild-type potato tubers and were then incubated with recombinant Pho1 and [U-14C]glucose 1-phosphate for 15 min or 25 min (in vitro experiment). Following incubation, starch granules were repeatedly washed and processed as in the in situ experiment. In each plot, the mean of three independently performed experiments and the SD are given.

As a control, the in vitro Pho1-mediated incorporation into starch granules was monitored. In these experiments, starch granules were isolated from potato tuber discs and were then incubated with both [U-14C]glucose 1-phosphate and recombinant Pho1 at 37 °C (Labelling method II). After 15 min or 25 min incubation, the reaction was terminated and the starch granules were used for both physical separation according to size and quantification of the 14C-content (Fig. 5). Incubation of tuber discs with [U-14C]glucose 1-phosphate and the subsequent fractionation according to particle size resulted in a even distribution of 14C-incorporation per surface area. By contrast, in vitro labelling of starch by recombinant plastidial phosphorylase revealed an uneven distribution of the 14C-label with a preferential incorporation into small starch particles (Fig. 5). In experiments in which the ratio between native starch particles and recombinant plastidial phosphorylase was altered, the same conclusion was reached (data not shown).

Incubation of tuber slices with labelled glucose 1-phosphate results in non-selective 14C-incorporation into the α-glucan chains of the starch granules

The data shown above clearly demonstrate that glucosyl residues derived from externally supplied glucose 1-phosphate are incorporated into starch granules irrespective of their size (Fig. 5). As the 14C-incorporation into starch is largely diminished in transgenic potato lines possessing strongly reduced levels of the plastidial phosphorylase isozyme (Pho1; Fettke ) labelling of starch is likely to be due to the action of Pho1. However, it remains unclear whether Pho1 exerts a strong selectivity for the glucosyl acceptors used. As an example, glucosyl residues could preferentially or exclusively be transferred by Pho1 to the non-reducing ends of α-glucan chains possessing a distinct length. If so, the Pho1-mediated glucosyl transfer is expect to represent a minor process within the entire starch synthesizing process. To test this possibility, a more detailed analysis of the labelled starch granules was performed. Following incubation of potato tuber discs with [U-14C]glucose 1-phosphate, starch granules were isolated, solubilized, and debranched by using an isoamylase. Subsequently, the linear glucan chains were separated by high performance anion exchange chromatography (HPAEC) and their 14C-contents were monitored. The HPAEC pattern of the individual α-glucans obtained by debranching has a wide range of degree of polymerization (DP) from 1 to approximately 30 (Fig. 6A). However, even longer chains (having a retention time from 26–32 min) appear to exist but these compounds were poorly resolved. It should also be noted that pulsed amperometry detects longer chains with an inferior sensitivity compared with equimolar shorter chains. Therefore, the pattern of linear glucans that is obtained by debranching is actually dominated by longer chains. Following 20 min in situ incorporation, distribution of the 14C-content is similar to that observed for the glucan chains by HPAEC (Fig. 6B). When the potato tuber discs are incubated with [U-14C]glucose 1-phosphate for a longer period of time (30 min), the total 14C incorporation into the glucan chains increases but the distribution of the label is unaltered (Fig. 6B).

Fig. 6.

Distribution of 14C-label in glucan chains released from starch granules by debranching. Wild-type tuber discs were incubated with [U-14C]glucose 1-phosphate (incubation procedure III). Following isolation of the starch granules, solubilization, and debranching by isoamylase, the linear glucan chains were analysed by HPAEC-PAD. In (A) a typical result out of 20 replicates is shown. Eluted fractions were collected and the 14C-contents were monitored using liquid scintillation counting. In (B) the distribution of 14C-label is shown after in situ incubation for 20 or 30 min. The means of three independently performed incubations and the SD are given. In another series of in situ incubations, discs generated from either freshly harvested wild-type or Pho1-repressed transgenic plants were incubated for 30 min with [U-14C]glucose 1-phosphate. Following isolation, solubilization, and debranching, the linear glucans were separation by HPAEC and the 14C-contents of the eluted fractions were monitored using liquid scintillation counting (C). The means of three independently performed incubations and the SD are given.

As a control, potato tuber discs from transgenic lines with a strongly repressed expression of the Pho1 isoforms were incubated with [U-14C]glucose 1-phosphate. In Solanum tuberosum L. two plastidial (Pho1-type) phosphorylase genes exist that have been designated as Pho1a and Pho1b (Albrecht ). Two independently generated transgenic potato lines were used which contain both an anti-Pho1a and an anti-Pho1b antisense construct (Duwenig, 1996; Fettke ). The two lines are designated as Pho1(a+b)-1 and Pho1(a+b)-2. The total extractable phosphorylase activities from tubers of the two transgenic potato lines and the wild-type are given in Table 4. In potato tubers, the cytosolic phosphorylase, Pho2, represents a very minor proportion of the total enzyme activity and, therefore, in the two transgenic lines most of the measureable Pho1 activity is repressed. As previously shown, in both lines expression of the cytosolic phosphorylase, i.e. Pho2, remains largely unaltered (Fettke , 2010).

Table 4.

Total phosphorylase activity of transgenic lines and wild-type potato tubers

Line	Specific activity (nmol glucose 1-phosphate min−1 mg−1 protein)
WT	113.45±4.22
Pho1(a+b)-1	7.62±4.32
Pho1(a+b)-2	9.97±1.41

Buffer-soluble proteins were extracted from growing potato tubers and were used for the photometric phosphorylase assay. Tubers from wild-type plants (wt) and two independently generated transgenic lines were used that carry both an antisense construct directed against Pho1a and Pho1b [Pho1(a+b)-1 and (Pho1(a+b)-2]. Enzyme activities are given as specific activities (nmol glucose 1-phosphate formed per min and mg protein; n=6; SD).

Tuber discs from the two transgenic lines and from the wild-type control were incubated with [U-14C]glucose 1-phosphate. Subsequently, native starch was isolated and equal amounts were identically processed including solubilization, debranching with isoamylase, separation of the chains released by HPAEC, and monitoring of the 14C-content (Fig. 6C). Compared with the wild-type control, the total 14C incorporation into starch from both transgenic lines is strongly reduced. In line Pho1(a+b)-1 even less 14C is incorporated into starch compared with line Pho1(a+b)-2, thus reflecting the lower residual Pho1 activity of line Pho1(a+b)-1.

Despite the lower 14C-incorporation into starch, the patterns of the 14C-labelled glucan chains were similar in the two transgenic lines and the wild type. It should be noted that, in this experiment, slices were prepared from freshly harvested potato tubers (Fig. 6C). By contrast, data shown in Fig. 6B were obtained from slices of resting potato tubers and, in addition, the incubation time was different. Despite these differences, the 14C-labelling patterns are similar and, therefore, neither the developmental state nor the incubation time significantly alters 14C-incorporation within the native starch granule. This result concurs with previous analyses of the glucose 1-phosphate-starch conversion (Fettke ).

In in vitro experiments, starch granules isolated from wild-type potato tubers were incubated with [U-14C]glucose 1-phosphate and recombinant Pho1. Starch levels were adjusted as closely as possible to the in situ content of tuber discs but even relatively large amounts of the recombinant Pho1 resulted into an in vitro incorporation rate of less than 10% of that observed in situ (data not shown; see also below). Following 5, 10, or 60 min incubation, aliquots of the incubation mixture were withdrawn and the reaction was terminated by adding SDS. Subsequently, the native starch granules were repeatedly washed, solubilized, and debranched using isoamylase. The linear glucan chains released were separated by HPAEC and the distribution of the 14C-label was monitored (Fig. 7A). A short incubation time (5 min) almost exclusively resulted in a labelling of glucans having a DP of 3–4. During prolonged incubation, most of the label is shifted towards larger glucan chains thus yielding a wider distribution (Fig. 7A). When the patterns of the 14C-label obtained in situ and in vitro were compared, major differences are obvious: Under in vitro conditions, the 14C-label is incorporated into longer linear glucan chains compared with in situ incorporation (Fig. 7B). The difference in the 14C-patterns between in situ and in vitro incorporation reflects a difference in the degree of polymerization (DP) of approximately 20. Noticeably, this shift is observed despite the fact that the in vitro incorporation rate is far lower than that of in situ labelling.

Fig. 7.

The distribution of 14C-label in glucan chains derived from starch granules during in vitro labelling with [U-14C]glucose 1-phosphate and recombinant plastidial phosphorylase. (A) Native starch granules isolated from wild-type potato tubers were incubated with recombinant plastidial phosphorylase and [U-14C]glucose 1-phosphate (incubation procedure IV) for 5, 10 or 60 min. The reaction was terminated by addition of SDS. Starch granules were repeatedly washed and the 14C-incorporation was monitored using liquid scintillation counting. The means of three incubation experiments and the SD are given. (B) Comparison of the 14C-label distribution between in situ and in vitro incorporation. The in situ incubation experiments were performed using tuber discs generated from wild-type tubers. In in vitro experiments, native starch granules isolated from wild-type or Pho1-repressed transgenic potato plants were used. The solubilized starches were debranched. The glucan chains released were separated by HPEAC and the distribution of the 14C-label was monitored using liquid scintillation counting. For comparison the incorporation is given as relative units. Note the difference in the total incorporation in starch between the in situ and in vitro experiments (see text).

When in vitro labelling was performed with tuber starch granules isolated from the Pho1-reduced transgenic line, Pho1(a+b)-2, the labelling pattern was indiscernible from that of the wild-type control (Fig. 7B). Thus, in vivo starch biosynthesis at a strongly reduced Pho1 activity does not diminish the acceptor efficiency of non-reducing ends in the starch granules that, in vitro, are used for the Pho1-mediated glucosyl transfer.

Both the low incorporation rates and the different 14C distribution (Fig. 7B) observed in the in vitro experiments indicate that the results obtained by in situ labelling do reflect intracellular processes. This conclusion was confirmed by control experiments in which buffer-soluble proteins were added to the tuber slices. In some samples, isolated starch granules were also included. Buffer-soluble proteins strongly diminished 14C-labelling of starch granules isolated from the tuber slices and the addition of both starch granules and proteins was even more effective (see Supplementary Fig. S1 at JXB online). Therefore, it is highly likely that the glucose 1-phosphate/starch conversion, as analysed in the in situ experiments, does reflect biochemical processes inside the intact parenchyma cells.

Uptake of [14C]glucose 1-phosphate by tuber discs results in a Pho1-mediated incorporation into the soluble maltodextrin pool

The plastidial and cytosolic phosphorylase isoforms differ in their preferences for α-glucans. The cytosolic phosphorylase (Pho2) has high affinity for large and branched glycans (Fettke , 2005, 2008) and lower affinity for small linear maltodextrins. By contrast, the plastidial isoform has opposing α-glucan selectivities (Steup and Schächtele, 1981; Mori ). Thus, the kinetic properties of Pho1 may point to an involvement in the plastidial maltodextrin metabolism but any direct evidence is still lacking (Zeeman ; for discussion see Zeeman ).

Because of these uncertainties, the following experiments were performed. Potato tuber discs from both wild-type and the two Pho1-repressed transgenic lines were incubated with [U-14C]glucose 1-phosphate. Subsequently, soluble compounds, including maltodextrins, were extracted and resolved by HPAEC and in the eluate fractions the 14C-content was quantified.

The HPAEC-PAD pattern is dominated by monosaccharides and small maltodextrins having a DP of up to DP5 but larger maltodextrins are minor constituents. In addition, other carbohydrates are also detectable, which appear to be branched oligoglucans or compounds with an altered monomer composition. As revealed by PAD, the soluble fractions from the two Pho1-repressed lines did not significantly differ from that of the wild-type (Fig. 8A).

Fig. 8.

Uptake of [U-14C]glucose 1-phosphate by tuber discs results in a Pho1-mediated 14C-labelling of maltodextrins. Tuber discs derived from wild-type and from Pho1-repressed transgenic potato plants (as in Table 4) were incubated for 20 min with [U-14C]glucose 1-phosphate. The soluble fractions were separated and analysed by HPEAC-PAD. A typical elution profile for each sample is shown (A). The elution fractions were collected and the 14C-contents were monitored using liquid scintillation counting (B). The means of three incubation experiments and the SD are given.

By contrast, 14C-labelling of the maltodextrins is clearly different: In wild-type tubers, total incorporation into soluble components was three to five times higher compared with the tubers of the transgenic plants (Table 5). In wild-type samples, 14C-labelling of monosaccharides, maltose, and maltodextrins having a degree of polymerization of up to15 was observed. In addition to these quantitative differences, the distribution of the 14C-label within the soluble fraction differed between the wild-type and the transgenic lines (Fig. 8B). In tuber discs from transgenic potato plants small oligosaccharides were most heavily labelled whereas, in the wild-type tubers, longer maltodextrins (DP9) contained a significantly higher proportion of 14C. In addition, larger oligosaccharides were also labelled. By contrast, in tuber slices derived from the transgenic lines, 14C-label of these maltodextrins was below the limit of detection (Fig. 8B).

Table 5.

Total 14C-content of the soluble fraction isolated from transgenic lines and wild-type potato tubers

Line	Total incorporation (Bq)
WT	700±23.5
Pho1(a+b)-1	175±18.3
Pho1(a+b)-2	233±12.1

Designation of the tubers from the wild type and two transgenic lines as in Table 4.

These data provide direct evidence for a Pho1-mediated flux from externally supplied glucose 1-phosphate both to starch and to soluble components.

Discussion

Compared with other externally applied sugars or sugar derivatives (such as glucose, sucrose, maltose, or glucose 6-phosphate), potato tuber discs convert glucose 1-phosphate to starch with an unexpected efficiency (Fettke ). Obviously, the short-term 14C-labelling rates are affected by several parameters, including the specific radioactivity of the various pools and, therefore, they do not necessarily reflect rates of starch biosynthesis under in situ conditions. Nevertheless, the data presented here clearly demonstrate the existence of a general path of starch biosynthesis initiated by externally supplied glucose 1-phosphate that closely but flexibly interacts with the established glucose 6-phosphate-dependent route. In previous publications, a conversion of glucose 1-phosphate to reserve starch has repeatedly been reported but the contribution of this flux to the entire starch biosynthesis has not yet been determined (Tyson and ap Rees, 1988; Kosegarten and Mengel, 1994; Tetlow ).

In the present study, it is demonstrated how closely this flux to starch is integrated into the entire starch synthesizing process.

First, both the conversion of [U-14C] glucose 1-phosphate to starch, as affected by the addition of unlabelled sucrose, and that of [U-14C]sucrose in the presence or absence of unlabelled glucose 1-phosphate (Fig. 1), was analysed. 14C-incorporation into starch is stimulated when unlabelled glucose 1-phosphate and sucrose were added to [U-14C]sucrose and [U-14C]glucose 1-phosphate, respectively (Fig. 1). The uptake of [U-14C]glucose 1-phosphate by cells or protoplasts is not affected by the presence of unlabelled sugars or sugar derivatives (Fettke , 2011). The stimulatory effect as shown in Fig. 1 is unexpected. If (assuming unchanged fluxes) inside the cells both the 14C-labelled sucrose and the unlabelled glucose 1-phosphate (or vice versa) enter the same intermediate pools, their specific radioactivities are diminished and, thereby, the rate of 14C-incorporation into starch decreases. Based on the results obtained, it is proposed that the two paths initiated by externally supplied glucose 1-phosphate or sucrose are mainly separated and interact in the late stages of the entire starch biosynthesis.

Second, the enhancement of 14C-incorporation into starch by the unlabelled compounds completely disappears over a few days if the detached tubers are stored at room temperature and are then sliced. This decline strongly resembles the labelling experiments that were performed by injecting [U-14C]sucrose or [U-14C]glucose into intact potato tubers. In these experiments, detachment of the tuber resulted in a severe inhibition of the sucrose–starch interconversion within one day that was paralleled by a strong decrease in the level of ADPglucose (Geigenberger ). These data strongly suggest that, by a largely unknown mechanism, the mother plant triggers starch accumulation in the growing tuber.

Third, under in situ conditions the two paths of starch biosynthesis (that are initiated by either sucrose or glucose 1-phosphate) differ in their temperature dependence (Fig. 2A). Similarly, the enhancement exerted by the respective unlabelled compound is differently affected at varying temperatures (Fig. 2B). The glucose 1-phosphate-dependent path (as well as the glucose 1-phosphate-dependent enhancement of the sucrose–starch conversion) is efficient at temperatures at or below 20 °C. By contrast, the sucrose-dependent path (as well as the sucrose-dependent enhancement of the conversion of glucose 1-phosphate to starch) is more effective at higher temperatures.

Fourth, the in vitro 14C-labelling studies (Tables 2, 3) are fully consistent with the results obtained in the in situ experiments. The reactions mediated by starch synthases (including GBSSI) consist of a glucosyl transfer from ADP-[U-14C]glucose to native starch, and the rate of these reactions increases with temperature (Table 2). By contrast, the rate of the Pho1-mediated glucosyl transfer from [U-14C]glucose 1-phosphate to starch is essentially maximal at 20 °C (Table 3).

Fifth, quantitative zymograms (Fig. 3) reveal a similar temperature dependence of the terminal glucosyl transfer reactions to glucosyl acceptors mediated by the plastidial phosphorylase (Pho1) and by the soluble starch synthases. Activity of the soluble starch synthases increases strongly at temperatures higher than 20 °C but that of the plastidial phosphorylase remains essentially unaltered at temperatures above 20 °C suggesting a temperature-dependent change in the ratio of the two biosynthetic routes under in situ conditions.

Sixth, the glucose 1-phosphate-dependent carbon flux to starch reflects a general rather than a selective starch synthesizing route. Physical separation of the native starch granules that had been isolated from potato tubers (Fig. 4) enabled us to monitor the 14C-incorporation into differently sized particles. Based on the surface area, the in situ 14C-incorporation was unaffected by particle size (Fig. 5).

Seventh, non-selectivity of the novel route was also demonstrated when the intragranular glucosyl acceptor sites used by the plastidial phosphorylase (Pho1) were determined. These experiments included 14C-incorporation and, subsequently, solubilization of the native starch followed by debranching. The glucan chains released were separated by HPAEC-PAD and the 14C-content of the eluate fractions was quantified (Figs 6, 7). The 14C-label was widely distributed over the DP range and chains with a degree of polymerization (DP) of more than 30 were most heavily labelled. The wide distribution of label remained unchanged when the incubation period was elongated (Fig. 6).

The conclusions drawn before are strongly supported by two other types of experiments. The glucose 1-phosphate incorporation into starch was significantly diminished in transgenic potato lines that, due to an antisense construct, repress the plastidial phosphorylase. Tuber slices from these lines incubated with [U-14C]glucose 1-phosphate incorporated far less 14C into starch (Fettke ) and 14C-labelling of glucans released by debranching is strongly reduced but the labelling patterns remained essentially unchanged (Fig. 6C). These data are fully consistent with the view that, under in situ conditions, the conversion of the externally supplied [U-14C]glucose 1-phosphate into starch requires the activity of the plastidial phosphorylase isozymes which, however, lacks any obvious acceptor specificity when acting under in situ conditions. In the other type of experiments native starch granules were isolated from potato tubers and incubated with the recombinant plastidial phosphorylase, Pho1, from Oryza sativa L. Under in vitro conditions, the action of recombinant Pho1 differs significantly from that observed in situ: when [U-14C]glucose 1-phosphate is added, the recombinant protein transfers the glucosyl residues preferentially to small-sized starch granules (Fig. 5). Possibly, the size preference as observed under in vitro conditions reflects subtle yet currently unknown structural alterations at the granule surfaces of different starches. In a previous study, starch granules isolated from leaves of Arabidopsis thaliana or Solanum tuberosum L. were more efficiently labelled by recombinant Pho 1 compared with potato tuber starch (Fettke ). The tuber starch is characterized by a dominance of longer linear glucan chains compared with leaf starch (data not shown). However, it is uncertain whether or to what extent these differences affect the structural properties of the granule surface. Under in vitro conditions, the recombinant Pho1 exhibits a higher selectivity even within the starch granule. In the initial transfer reactions, short glucans appear to be preferred as acceptor sites and only with time, 14C-labelling proceeds to larger chains (Fig. 7A). When using a prolonged in vitro action of the recombinant phosphorylase isozyme, most of the 14C-label is recovered in glucan chains whose length considerably exceeds that observed during in situ 14C-incorporation (Fig. 7B). This difference between the in situ and in vitro labelling of starch exists despite the fact that under in vitro conditions the rate of the transfer of glucosyl residues was far lower than under in situ conditions. It clearly indicates that, in situ, the action of the plastidial phosphorylase is integrated into reaction sequences that limit the size of the glucan chains elongated by the Pho1-mediated glucosyl transfer. Such a size limitation could occur by a closely associated action of other carbohydrate-active enzymes, such as branching isozymes.

Based on the data presented above, the existence was postulated of two distinct (yet closely interconnected) plastidial paths of reserve starch biosynthesis that both originate from the plastidial pool of glucose 1-phosphate. One of the paths utilizes the plastidial pool of ADPglucose as the glucosyl donor and includes the subsequent action of various starch synthases. The other path consists of the Pho1-mediated glucosyl transfer to native starch granules using the plastidial glucose 1-phosphate as donor (Fettke ). Both biosynthetic routes are likely to be functional throughout starch granule development (Fig. 5). Unfortunately, no structural information is available that describes the starch granule surface at a sufficiently high resolution (Pérez and Bertoft, 2010). This lack of knowledge prevents any reasonable explanation of the interaction of the two paths at a molecular level. It is unknown (and, currently, difficult to analyse) whether both biosynthetic routes actually lead to identical or non-identical surface elements. One may speculate that the simultaneous action of both routes results in a higher number of acceptor sites that are available to both glucosyl transfer reactions and, thereby, favours both fluxes towards starch. In any case, the dual path of starch biosynthesis, as proposed here, is fully consistent with the strong expression of the plastidial phosphorylase isozyme(s), Pho1 in organs exhibiting a massive starch biosynthesis, such as seeds of spinach (Spinacia oleracea L.). Furthermore, in spinach leaf discs Pho1 is strongly induced by the addition of exogeneous sugars and, under these condition, induction of Pho1 is similar to that of the large subunit of the AGPase (Duwenig ).

It is widely accepted that the in vivo function of soluble starch synthases is the elongation of side chains during amylopectin biosynthesis (Hennen-Bierwagen ; Keeling and Myers, 2010; Zeeman ; Szydlowski et al., 2011). In heterotrophic tissues from cereals, a close interaction of phosphorylase and branching isozymes has been reported that, depending on the metabolic state of the tissue, includes the formation of heteromeric protein complexes (Tetlow , 2008; Liu ). The size limitation of the 14C-labelled glucan chains as observed under in situ conditions (Figs 6, 7) may point to such an interaction between the plastidial phosphorylase and (at least) one branching isozyme but, until now, functional protein complexes have not yet been reported to exist in potato tubers. If so, the molecular target of both routes would be the amylopectin molecule itself. In any case, the two paths are capable of strongly interacting and largely stimulating each other (Fig. 1). However, the enhancement strictly depends on the temperature as both the established and the novel route of starch biosynthesis differ in their temperature dependency (Fig. 2). The glucose 1-phosphate-dependent route is functional at relatively low temperatures and, at higher temperatures, remains essentially unchanged. By contrast, the efficiency of the sucrose-dependent route increases at temperatures above 20 °C. These data do not exclude that, within the sucrose-dependent path of starch biosynthesis, other reactions and/or processes, such as the regulation of the AGPase and/or the activity of the plastidial phosphoglucomutase, may also contribute to the control of the fluxes towards reserve starch at different temperatures.

Obviously, the results presented in this communication resemble the phenotype of recently described rice mutants. In these lines, deficiency of the plastidial phosphorylase isozyme results in a strongly compromised seed formation at low temperatures but this phenotype disappears at higher temperatures. Thus, within starch biosynthesis, biochemical redundancies exist but they are restricted to a relatively narrow range of temperatures (Satoh ). It implies that the phenotypical characterization of a given mutant may indicate a loss of function but this phenotype may be restricted to distinct external conditions and, consequently, an alteration of the external conditions may lead to a seemingly contradicting phenotype. When growing under natural conditions, wild-type plants (such as rice or potato) are facing highly variable physical parameters. Possibly, the interaction between the two starch biosynthetic routes described above permits a better performance under a wide range of external conditions.

Another uncertainty related to the compartment in which the glucose 1-phosphate-dependent path of starch biosynthesis described here is initiated. Glucose 1-phosphate could be generated inside the cytosol or, alternatively, an apoplastic pool of the anomeric glucose phosphate could exist. In this case, the import of glucose 1-phosphate into the parenchyma cells would initiate the biosynthetic route. It has recently been shown that the apoplastic space hosts compounds, such as ATP, and enzymes acting on these compounds, such as the extracellular apyrase (Riewe ). In transgenic potato lines lacking the apoplastic apyrase, tuber growth and morphology is massively altered. However, the underlying biochemical processes are largely unknown (Riewe ). Similarly, it is not yet known whether, in potato tubers, the apoplastic space contains glucose 1-phosphate. Currently, a detailed analysis of the metabolites residing in the apoplast is being performed in this laboratory.

When comparing the starch labelling kinetics observed in tuber slices with the intact potato plants a fundamental difference is obvious: Tubers are highly active sinks that are connected with the source organs by the long-distance transport of assimilates, i.e. by the phloem. Based on electron microscopical studies, a special relevance of the symplastic phloem unloading has been postulated (Oparka, 1986). In tuber slices incubated with carbohydrates, a direct import into parenchyma cells is likely to be dominant.

Despite this difference, evidence has recently accumulated that glucose 1-phosphate plays an important and distinct role in reserve starch biosynthesis. Transgenic potato plants possessing strongly reduced activities of both the plastidial and the cytosolic phosphoglucomutase are capable of accumulating reserve starch with almost wild-type rates (Fernie et al., 2002a). The biosynthetic route functional in these transgenic lines appears to be independent of both the cytosolic and plastidial generation of glucose 1-phosphate. Furthermore, transgenic potato tubers in which the amyloplastic glucose 6-phosphate/orthophosphate transporter is diminished by approximately 85% accumulated only slightly reduced starch levels (Ludewig, personal communication). These data imply that, in the intact potato plant, the well-established import of glucose 6-phosphate into the amylopast, the subsequent conversion to glucose 1-phosphate and, finally, to ADPglucose can be efficiently bypassed by other reactions. By contrast, a significantly altered phenotype has been reported for potato plants with a reduced activity of the ADPglucose pyrophosphorylase (Müller-Röber ) or of the plastidial ATP/ADP transporter (Tjaden ). In addition to a reduced starch content, phenotypical alterations include tuber size and/or morphology, amylopectin/amylose ratio, expression of the major storage protein and the levels of both sucrose and glucose. Obviously, these lines exhibit a complex deviation from the wild-type control.

Slices from potato tubers that perform a massive 14C-incorporation into starch do also incorporate the 14C-label into soluble glycans (Fig. 8). Because of their chemical diversity and their heterogeneous subcellular distribution, these compounds are difficult to analyse. As observed with native starch granules, incorporation into soluble glycans is far lower when slices from two transgenic potato lines are analysed that repress expression of the plastidial phosphorylase, Pho1 (Fig. 8B). Therefore, Pho1 appears to exert a dual function by directly transferring glucosyl residues to both the starch granules surface and soluble maltodextrins. The function of the soluble maltodextrin pool within the plastidial starch metabolism is largely unknown.

Supplementary data

Supplementary data can be found at JXB online.

. [U-14C]Glucose 1-phosphate-dependent in situ-labelling of starch as affected by the addition of soluble proteins and native starch.

. Activities of the cytosolic phosphorylase from potato tuber crude extracts as affected by varying temperatures.