Glucan, water dikinase phosphorylates crystalline maltodextrins and thereby initiates solubilization


  • Mahdi Hejazi,

    1. UP Transfer GmbH, Am Neuen Palais 10, Building 6, 14469 Potsdam, Germany,
    Search for more papers by this author
  • Joerg Fettke,

    1. Department of Plant Physiology, Institute of Biochemistry and Biology, University of Potsdam, Karl-Liebknecht-Str. 24-25, Building 20, 14476 Potsdam-Golm, Germany,
    Search for more papers by this author
  • Sophie Haebel,

    1. Interdisciplinary Center of Mass Spectrometry of Biopolymers, University of Potsdam, Karl-Liebknecht-Str. 24-25, Building 20, 14476 Potsdam-Golm, Germany,
    Search for more papers by this author
  • Christoph Edner,

    1. Department of Plant Physiology, Institute of Biochemistry and Biology, University of Potsdam, Karl-Liebknecht-Str. 24-25, Building 20, 14476 Potsdam-Golm, Germany,
    Search for more papers by this author
  • Oskar Paris,

    1. Department of Biomaterials, Max-Planck Institute of Colloids and Surfaces, Am Mühlenberg 1, 14476 Potsdam-Golm, Germany, and
    Search for more papers by this author
  • Claus Frohberg,

    1. Bayer Bioscience GmbH, Hermannswerder 20, 14473 Potsdam, Germany
    Search for more papers by this author
  • Martin Steup,

    Corresponding author
    1. Department of Plant Physiology, Institute of Biochemistry and Biology, University of Potsdam, Karl-Liebknecht-Str. 24-25, Building 20, 14476 Potsdam-Golm, Germany,
    Search for more papers by this author
  • Gerhard Ritte

    1. Department of Plant Physiology, Institute of Biochemistry and Biology, University of Potsdam, Karl-Liebknecht-Str. 24-25, Building 20, 14476 Potsdam-Golm, Germany,
    Search for more papers by this author
    • Present address: Metanomics GmbH, Tegeler Weg 33, 10589 Berlin, Germany.

*(fax +49 331 977 2512; e-mail


Starch phosphorylation by glucan, water dikinase (GWD; EC is an essential step in the breakdown of native starch particles, but the underlying mechanisms have remained obscure. In this paper, the initial reactions of starch degradation were analyzed using crystallized maltodextrins as model carbohydrates. As revealed by X-ray diffraction analysis, the crystallized maltodextrins represent the B-type starch allomorph. Recombinant GWD phosphorylated crystalline maltodextrins with a high specific activity (55–60 nmol mg−1 protein min−1), but exhibited very little activity with the same maltodextrins that had been solubilized by heat treatment. Recombinant phosphoglucan, water dikinase (PWD; EC utilized the crystalline maltodextrins only when pre-phosphorylated by GWD. Phosphorylation of crystalline maltodextrins, as catalyzed by GWD, initiated solubilization of neutral as well as phosphorylated glucans. In both the insoluble and the soluble state, mono-, di- and triphosphorylated α-glucans were observed, with wide and overlapping ranges of degree of polymerization. Thus, the substrate specificity of the GWD is defined by the physical arrangement of α-glucans rather than by structural parameters, such as the distribution of branching points or degree of polymerization. Unlike GWD and PWD, recombinant β-amylase isozyme 3 (BAM3), which has been shown to be essential for plastidial starch degradation, preferentially degraded soluble maltodextrins rather than crystallized glucans. In summary, two conclusions were reached. Firstly, carbohydrate targets of GWD are primarily defined by the molecular order of glucan helices. Secondly, GWD-catalyzed phosphorylation mediates the phase transition of glucans from a highly ordered to a less ordered and hydrated state.


Almost all plants synthesize water-insoluble starch particles that allow growth to be sustained in a changing environment (Smith and Stitt, 2007). Starch granules are thought to be composed of two types of glucose polymers, amylose and amylopectin. The latter is a branched α-1,4:α-1,6 d-glucan and accounts for 70–90% of the starch dry weight. Amylose, by contrast, is a linear α-1,4 d-glucan containing very few α-1,6 linkages (Smith et al., 2005).

Both the arrangement of the amylopectin molecules and features of the individual polyglucan determine the organization of the entire starch particle. Amylopectin molecules appear to be radially arranged, with their axes perpendicular to the granule surface (Gallant et al., 1997). Due to clustering of the branching points and the length distribution of the side chains within a single amylopectin molecule, neighboring chains are capable of double-helix formation (Smith, 2001). A liquid-crystalline model for the structure of reserve starch was presented by Waigh et al. (2000).

As revealed by X-ray diffraction analysis, several starch structures can be distinguished that differ in the arrangement of the double helices. The A-type allomorph, which is typical of wild-type cereal starches, is more compact and contains fewer water molecules between the double helices compared to the B-type structure. The latter is found in storage organs, such as potato tubers, and also in some high-amylose cereal starches (Gallant et al., 1997; Gérard et al., 2001). Legume starches are considered to represent a third allomorph (C-type), as both A- and B-type crystallites are found within the same starch particle (Bogracheva et al., 2001; Imberty et al., 1991). The three allomorphs have also been observed in crystallized maltodextrins (Gidley and Bulpin, 1987).

The organization of α-glucan chains has profound effects on the action of carbohydrate-active enzymes. When A- and B-type allomorphs were hydrolyzed using a pancreas α-amylase, B-type starches were more slowly degraded than their A-type counterparts (Gérard et al., 2001). Many enzymes that cleave α-1,4 or α-1,6 interglucose linkages act efficiently on soluble polyglucans, such as glycogen, amylopectin or heat-treated starch, but exhibit little or no activity with native starch granules (Edner et al., 2007; Lloyd et al., 2005). However, the two plastidial starch-phosphorylating enzymes, i.e. glucan, water dikinase (GWD, EC; Ritte et al., 2002, 2006) and phosphoglucan, water dikinase (PWD, EC; Kötting et al., 2005) display significant activities with granular starch. GWD and PWD selectively catalyze phosphorylation of the C6 and C3 positions, respectively, of glucosyl residues within amylopectin molecules, and are both required for the normal metabolism of transitory starch and for development of the entire plant as well (Baunsgaard et al., 2005; Kötting et al., 2005; Yu et al., 2001).

Another enzyme that is involved in the initial steps of transitory starch degradation is the β-amylase isozyme 3 (BAM3 or BMY8; At4g17090). Arabidopsis mutants with a diminished expression of BAM3 possess elevated starch levels (Kaplan and Guy, 2005), and these results have been confirmed using a null mutant of BAM3 (Fulton et al., 2008). Similar results have been reported for Solanum tuberosum L. (Scheidig et al., 2002). The in vitro breakdown of native starch particles, as catalyzed by BAM3, is stimulated by the simultaneous phosphorylation of starch by GWD (Edner et al., 2007). Remarkably, β-amylolysis of starch granules also enhances the phosphorylating activity of GWD (Edner et al., 2007). It has been proposed that the activity of GWD causes partial unwinding of double helices, thereby rendering glucan chains accessible for β-amylolysis (Edner et al., 2007).

For several reasons, this hypothesis is difficult to test with native starch granules. Firstly, any alterations of the molecular order caused by GWD are likely to be restricted to the particle surface (or distinct areas of it), and will therefore have a limited effect on the crystallinity of the entire granule. Secondly, due to the structural heterogeneity of the starch granule surface, the actual sites of GWD action are difficult to determine. Finally, if GWD favors the hydration of surface-located glucan chains, these chains will remain covalently linked to the insoluble starch particle and are therefore difficult to identify.

To study the action of GWD in more detail, we have used a structurally less complex model substrate, i.e. crystallized maltodextrins. By using this approach, we demonstrate that recombinant GWD phosphorylates crystallized maltodextrins and thereby promotes glucan solubilization. The action of recombinant PWD and BAM3 on crystalline maltodextrins has also been analyzed. Thus, the data presented here provide an explanation of the relationship between phosphorylation and degradation of starch at a mechanistic level.


Physicochemical properties of the crystallized maltodextrins

As revealed by high-performance anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD), the degree of polymerization (DP) of the dominant chains in the starting material maltodextrins ranges from 1 to 29, whereas that of the crystalline maltodextrin preparation is shifted to higher degrees of polymerization (maltohexaose, up to approximately DP 40; Figure 1a). Throughout this study, the following abbreviations are used: MD, maltodextrins used as the starting material for crystallization; MDcryst, pelletable maltodextrins derived from MD by crystallization; MDsol, soluble maltodextrins obtained from MDcryst either by heat treatment or following an enzymatic action. As linear and branched maltodextrins with the same DP usually do not co-migrate, the HPAEC-PAD analyses of both MD and MDcryst clearly indicate that the vast majority of the maltodextrins are unbranched chains.

Figure 1.

 (Physico)chemical characterization of various α-glucans.
(a) HPAEC-PAD analysis of the maltodextrin starting material (MD) and MDcryst (30 μg each). DP, degree of polymerization.
(b) X-ray diffraction patterns of potato tuber starch (1), MDcryst (2), and starch from maize kernels (3).
(c) X-ray diffraction patterns of leaf starch granules from potato (2) and the sex1–3 mutant of Arabidopsis thaliana L. (3), compared to that of potato tuber starch (1). All X-ray analyses were performed with fully hydrated samples using the NanoStar instrument.

The X-ray diffraction pattern of several independent MDcryst preparations was compared with that of starches from potato tubers, maize kernels, leaves of potato plants and leaves of the GWD-deficient Arabidopsis mutant sex1–3 (Figure 1b,c; Yu et al., 2001). Starch samples were measured in the fully hydrated state, and most of them also in a dried state. Except for wild-type maize starch which represents the A-type allomorph (Gérard et al., 2001), the patterns for MDcryst and all other starches were very similar, displaying the typical characteristics of the B-type allomorph.

The degree of crystallinity of the various samples was estimated from the X-ray diffraction patterns according to the method described by Nara and Komiya (1983). This approach has frequently been used in starch analyses, although the background and scattering from amorphous regions are not precisely defined. The relative crystallinity estimated for maize and potato tuber starches was similar to published data (Cheetham and Tao, 1998; Nara and Komiya, 1983). Interestingly, MDcryst exhibited the highest degree of crystallinity of all measured samples (Table 1).

Table 1.   Degree of crystallinity of various α-glucans as determined by X-ray diffraction
  1. Wet, fully hydrated samples; dry, samples dried under ambient humidity. –, not measured.

Potato tuberB0.660.45
Potato leafB0.69
Arabidopsis leafB0.60
Maize kernelA0.470.40

MDcryst and several native starch granules representing either the A- or B-type allomorphs were used as the substrate for recombinant GWD (Table 2). For starch particles, the rate of phosphorylation was determined using GWD assay I (see Experimental procedures). However, when using MDcryst as the substrate, the 33P label was not exclusively recovered in the pellet (see below), and assay II was therefore applied. Recombinant GWD phosphorylated B-type starches at a higher rate than the A-type allomorph (Table 2). Within the B-type allomorph, the phosphorylation rate for MDcryst was exceptionally high. Thus, MDcryst is likely to be an appropriate model carbohydrate for studying the action of GWD.

Table 2.   Specific GWD activity using various native starches and MDcryst as substrate
 AllomorphGWD activity
  1. The allomorph of the respective glucans was determined using the NanoStar instrument or was taken from aGérard et al. (2001) and bHizukuri (1985).

  2. cae, amylose extender mutant.

  3. GWD activity is given as nmol phosphate esterified per min mg−1 GWD protein. The standard deviation is given for 3–6 independent experiments.

WT wheat, kernelAa0.07 ± 0.02
WT maize, kernelA0.11 ± 0.04
Waxy maize, kernelAa0.08 ± 0.01
ae maize, kernelcBb1.2 ± 0.2
Potato, tuberB2.3 ± 0.25
Arabidopsis sex1–3, leafB11.2 ± 1.4
MDcrystB58.3 ± 4

GWD preferentially phosphorylates crystalline maltodextrins

To study the action of recombinant GWD on soluble or crystalline maltodextrins, MDcryst was added to the reaction mixture either without or following complete solubilization by heat treatment (which completely converted MDcryst to MDsol). Thus, the α-glucans added to the assays were identical in both chain length distribution and amount but differed in their molecular order. As controls, either maltodextrins were omitted or GWD was inactivated.

After 30 min, reactions were terminated by boiling (which also completely solubilized MDcryst), and equal aliquots of the various mixtures were chromatographed on polyethylene imine (PEI)-cellulose plates. As a reference, 33P-labeled glucose-6-phosphate (G-6-P) was subjected to TLC. Radioactivity was quantified using phosphor imaging. Throughout this study, we noticed that commercial preparations of [β-33P]ATP contain several 33P-labeled compounds that co-migrate with Pi, ADP and G-6-P (Figure 2a, lanes 2 and 3).

Figure 2.

 Analysis of the GWD- and PWD-catalyzed phosphorylation of maltodextrins.
(a) GWD. Reaction mixtures (containing 100 ng recombinant GWD, 10 mg ml−1 MDcryst except where stated, 25 μm unlabeled ATP and 2 μCi [β-33P]ATP) were incubated for 30 min at 30°C under continuous agitation. Reactions were terminated by heating (5 min at 95°C). Aliquots (2 μl each) of various reaction mixtures were analyzed by TLC (lanes 2–6). Subsequently, 33P-labeled compounds were detected by phosphor imaging. As a reference, 33P-labeled G-6-P that had been synthesized by using [γ-33P]ATP, glucose and hexokinase was chromatographed. During heat inactivation, [33P]-orthophosphate is released from the [γ-33P]ATP (lane 1). As controls, MDcryst was omitted (lane 2) or GWD was inactivated by heat treatment prior to addition to the reaction mixture (lane 3). MDcryst was added either without heat treatment (lanes 3, 5 and 6) or following heat treatment (lane 4). In lane 6, the reaction mixture (as in lane 5) was subjected to acid hydrolysis in 2 m TFA prior to TLC.
(b) PWD. Reaction mixtures (containing 4 mg ml−1 maltodextrins each) were incubated as in (a). Lane 1, MDcryst with PWD inactivated by heat treatment; lane 2, PWD with MDcryst; lane 3, PWD with MDsol; lane 4, PWD with pre-phosphorylated MDsol; lane 5, PWD with pre-phosphorylated MDcryst.

In PEI-TLC, negatively charged compounds interact with immobilized ionic groups, and therefore analytes are separated mainly according to their mass/charge ratio. Consequently, 33P-ATP has a very low mobility whereas 33P-ADP, 33Pi and 33PG-6-P move faster (Figure 2a). Phosphoglucans (and uncharged α-glucans also; see below), which have a high mass/charge ratio, display an even higher mobility and migrate as a relatively wide zone at or slightly behind the front (lanes 4 and 5). The formation of this fast moving 33P-labeled spot was most pronounced when MDcryst was added to the reaction mixture. By contrast, much weaker phosphorylation was observed when MDcryst had been converted to MDsol prior to GWD action (lanes 4 and 5). No phosphoglucans were detectable in either control (lanes 2 and 3).

Using MDcryst as the acceptor of the β-phosphate, the specific GWD activity ranged between 55 and 60 nmol phosphate incorporated per mg GWD protein per minute (six independent experiments). Up to 1 h, the rates of phosphorylation were linear with respect to enzyme concentration and time (data not shown). When MDcryst was replaced by MDsol, the specific activity was consistently below 1 nmol mg−1 GWD protein min−1.

The fast-moving compounds were further characterized by acid hydrolysis and subsequent TLC of the hydrolysates (Figure 2a, lane 6). Following hydrolysis, the fast-moving spot was undetectable and 33P-labeling was restricted to two spots that co-migrated with Pi and G-6-P. Acid hydrolysis of labeled ATP and ADP (which both are undetectable following hydrolysis; see lanes 5 and 6) liberated 33Pi, and G-6-P was released from phosphorylated glucans.

These results were confirmed by MS analysis of the fast-moving compounds that had been eluted from the thin-layer plate. A series of neutral and phosphorylated glucans was observed with a minimum DP of 6, but the mass spectra were dominated by signals derived from the neutral oligosaccharides (data not shown).

PWD phosphorylates maltodextrins that have been pre-phosphorylated by GWD

Using native starch granules as the carbohydrate substrate, PWD activity was detectable only if the starch particles had been pre-phosphorylated by GWD (Kötting et al., 2005). In this study, the 33P labeling of MDcryst by recombinant PWD was studied using a similar approach to that described above (Figure 2a). However, we also included MDcryst that had been pre-phosphorylated by GWD using non-labeled ATP. Following pre-phosphorylation, GWD was removed by treatment with SDS at room temperature and repeated washing steps. A reaction mixture containing heat-inactivated PWD served as control.

Using PWD and non-phosphorylated MDcryst, no PWD activity was detectable (Figure 2b, lane 2). However, PWD did phosphorylate MDcryst following pre-phosphorylation by GWD (lane 5). Solubilization of MDcryst abolished any noticeable PWD activity, irrespective of whether or not the maltodextrins had been pre-phosphorylated by GWD (lanes 3 and 4).

Taken together, two conclusions can be drawn from the data shown in Figure 2: firstly, both GWD and PWD preferentially act on insoluble maltodextrins, and secondly, the catalytic action of PWD requires preceding phosphorylation of MDcryst.

Phosphorylation of MDcryst by GWD initiates solubilization of both neutral and phosphorylated maltodextrins

In another series of labeling experiments, a low amount (0.1 μg) or a high amount (3 μg) of GWD was incubated with the same amount of MDcryst. At intervals, the reaction was terminated by the addition of EDTA (final concentration 20 mm), and the 33P content of MDcryst and MDsol was separately monitored. Using 0.1 μg of GWD, the majority of 33P was recovered in MDcryst, and labeling of MDsol was considerably delayed: it was undetectable during the first 30 min of incubation and later occurred at a lower rate than that for MDcryst (Figure 3a). After 2 h incubation, approximately 85% and 15% of the total radioactivity was found in MDcryst and MDsol, respectively. When a 30-fold higher amount of GWD was used, both the labeling kinetics and the ratio of 33P contents of MDcryst and MDsol were altered (Figure 3b). Throughout the incubation, the total rate of glucan esterification was more than one order of magnitude higher than that observed in Figure 3(a), and the majority of the 33P content was recovered in the supernatant.

Figure 3.

 Glucan phosphorylation catalyzed by GWD initiates solubilization of MDcryst.
(a, b) Pulse labeling. In a final volume of 50 μl, MDcryst (10 mg ml−1) and 25 μm ATP (plus 2 μCi [β-33P]ATP) were incubated with 0.1 μg (a) or 3 μg (b) recombinant GWD. At intervals, the reaction was terminated by addition of EDTA (final concentration 20 mm), and the reaction mixture was centrifuged for 10 min at 10 000 g. The supernatant was collected, and the pellet was resuspended in an equal volume of water. Following heat treatments aliquots of both the supernatant and the resuspension were spotted onto a thin-layer plate. Following development, radioactivity in the fast-moving spot (see Figure 2a) was quantified by phosphor imaging. In each case, a typical experiment out of four is shown.
(c) Pulse-chase labeling. Following pre-labeling, MDcryst was incubated with GWD and unlabeled ATP. At intervals, aliquots of the soluble and insoluble glucans were used for liquid scintillation counting. As a control, GWD was heat-inactivated prior to addition to MDcryst. A typical experiment out of three is shown.
(bsl00001) MDcryst, (•) soluble glucans, (bsl00000) MDcryst control, (bsl00043) soluble glucans in the control. Radioactivity is given as cpm per 100 μl supernatant or suspension of MDcryst.

The data shown in Figure 3(a,b) clearly support the assumption that esterification and solubilization of α-glucans follow a two-step mode. First GWD catalyzes the phosphorylation of glucosyl residues at the C6 position. This reaction is essentially restricted to MDcryst. Then phosphorylated glucans are released into the soluble phase. The rate of this transition increases with the extent of phosphorylation of MDcryst.

This conclusion was confirmed by a pulse-chase experiment (Figure 3c). Using both [β-33P]ATP and recombinant GWD, MDcryst was labeled for 1 h. Under these conditions, the majority of the label was retained in MDcryst. Subsequently, MDcryst was separated from MDsol, and was resuspended in a mixture containing non-labeled ATP (final concentration 25 μm) and recombinant GWD. As a control, heat-inactivated GWD was used. At intervals, aliquots of both the soluble and insoluble fractions were withdrawn, and the 33P content was monitored (Figure 3c). Following the chase, the distribution of the label changed: the 33P content of MDcryst decreased and that of the soluble fraction increased by more than 50% (Figure 3c). In the control, far less radioactivity was recovered in the soluble phase.

A more detailed analysis of the (phospho)glucans released from MDcryst was performed using unlabeled ATP and a prolonged incubation period. Subsequently, MDcryst and MDsol were separated, hydrolyzed (see Experimental procedures) and the glucose and G-6-P contents were determined enzymatically. Two controls were included in which either ATP or GWD were omitted (Table 3). Following 16 h incubation, 47% of the ATP had been utilized by GWD to yield G-6-P residues. Of the total amount of G-6-P residues formed, 20% and 80% were recovered in MDcryst and in MDsol, respectively. In the absence of either ATP or GWD, neither the pellet nor the supernatant contained detectable amounts of G-6-P. During the GWD-catalyzed phosphorylation, approximately 37% of the total glucosyl residues were solubilized. In MDsol, 1.8% of the glucosyl residues were phosphorylated, whereas the glucose-based degree of phosphorylation was below 0.3% in MDcryst. This difference is expected as the action of GWD is restricted to the surface of MDcryst.

Table 3.   Solubilization of neutral and phosphorylated α-glucans during the GWD-catalyzed esterification of MDcryst
 Glucose (μmol)G-6-P (nmol)
  1. In a final volume of 500 μl, recombinant GWD (1.5 μg) was incubated with 5 mg MDcryst and 1 mm ATP for 16 h at 30°C under continuous agitation. As controls, either ATP or GWD were omitted. Samples were centrifuged (10 min at 10.000 g) at the end of the incubation time. Following acid hydrolysis, the contents of glucose and G-6-P were determined enzymatically both in the pellet (insoluble fraction) and the supernatant (soluble fraction). ND, not detectable.

Insoluble fraction26.6026.9017.5NDND47.5
Soluble fraction0.830.8210.2NDND187.4

During the prolonged incubation, some glucans (<3%) were solubilized in the two controls, i.e. in the absence of either ATP or GWD (Table 3). Presumably, this transition is due to continuous agitation of the samples.

MS analysis of the solubilized phosphoglucans

HPAEC-PAD of MDsol (procedure B; see Experimental procedures) resulted in a complex pattern (Figure 4a) that consisted of a series of maltodextrins with a DP ranging from 6 to more than 30 and several additional PAD signals. The peaks designated 1 to 3 (Figure 4a) contained phosphoglucans (see below). These peaks were detectable only with the complete reaction mixture (see Figure 4b). Furthermore, they contained 33P-label when [β-33P]ATP was included in the complete mixture (data not shown). Other peaks were identified by systematical analyses of both control mixtures and defined constituents of the reaction mixtures (for details, see Figure 4a).

Figure 4.

 Neutral and phosphorylated glucans in MDsol.
(a) In a final volume of 500 μl, 3 μg recombinant GWD was incubated with 500 μg MDcryst, BSA (400 μg ml−1), 1 mm ATP, 6 mm MgCl2, 1 mm EDTA and 50 mm HEPES-KOH, pH 7.5.
(b) As control, GWD was inactivated prior to the addition to the assay mixture.
Following incubation for 16 h at 30°C and centrifugation, 100 μl of each supernatant was passed through a 10 kDa membrane filter and the resulting filtrate was loaded onto the HPAEC column. Eluting compounds were detected by pulsed amperometry using the same sensitivity for both samples. Peaks 1 to 3 represent phosphoglucans (see Figure 5). Peaks labeled with an asterisk (*) represent compounds of the reaction mixture: x, HEPES; y, nucleotides; z, DTT.
(c, d) Separation of neutral and phosphorylated α-glucans by chromatography on graphitized carbon. Samples were loaded onto a Carbograph column, and neutral glucans were eluted using 30% v/v acetonitrile (c) and subsequently phosphoglucans with 30% v/v acetonitrile containing 0.2% v/v TFA (d).

Physical separation of neutral and phosphorylated α-glucans was also achieved by chromatography on carbon black columns using a two-step elution (see Experimental procedures). This technique has the advantage of being more easily compatible with subsequent MALDI MS analyses. Using this procedure, the series of maltodextrins was selectively eluted with 30% v/v acetonitrile (Figure 4c), and peaks 1–3 were recovered using the TFA-containing eluent (Figure 4d).

For analysis by MALDI MS, peaks 1–3 were collected separately. All three peaks cover relatively wide ranges of DPs, but differ in the degree of phosphorylation (single, dual and triple phosphorylation; Figure 5). The minimum size of monophosphorylated maltodextrins was maltohexaose, but the signal intensity of this compound was low (data not shown). Whilst the more abundant singly and doubly phosphorylated maltodextrins differed very little with respect to their DPs (Figure 5a,b), the most prominent triply phosphorylated α-glucans were shifted towards higher DPs (Figure 5c).

Figure 5.

 MALDI MS analysis of purified phosphoglucans.
Following HPAEC-PAD, peaks 1–3 (see Figure 4d) were separately collected and passed through a miniaturized Carbograph column (see Experimental procedures). Peak 1 represents monophosphorylated glucans (a), peak 2 doubly phosphorylated glucans (b), and peak 3 triply phosphorylated glucans (c). All analytes were recovered as sodium adduct ions. Theoretical masses of the phosphorylated analytes were calculated as follows: singly phosphorylated glucans, DP × (Glc − H2O) + H2O + 2Na+ + HPO3; doubly phosphorylated glucans, DP × (Glc − H2O) + H2O + 3Na+ + 2HPO3; triply phosphorylated glucans, DP × (Glc − H2O) + H2O + 4Na+ + 3HPO3.

Singly, doubly and triply phosphorylated maltodextrins were also observed in MDcryst (data not shown). Thus, in summary, the data shown in Figures 2–4 clearly indicate that the GWD-catalyzed phosphorylation takes place at the surface of the crystalline maltodextrins and results in formation of a wide range of phosphoglucans. Subsequently, both the phosphoglucans formed and the neutral glucans are released into the soluble phase.


Ten years ago, Lorberth et al. (1998) described a novel protein (provisionally designated R1) that was essential for starch phosphorylation in Solanum tuberosum L. Down-regulation of R1 expression resulted in a significantly decreased phosphate content of starch, and, in addition, to altered starch turnover. In subsequent studies, a similar phenotype was described for Arabidopsis mutants deficient in a R1 orthologue (Yu et al., 2001). Finally, R1 was identified as a glucan, water dikinase (GWD) that selectively phosphorylates amylopectin-related glucosyl residues in the C6 position (Ritte et al., 2002, 2006). A second dikinase (designated PWD) has also been described that catalyzes the phosphorylation of glucosyl residues in the C3 position (Baunsgaard et al., 2005; Kötting et al., 2005; Ritte et al., 2006). However, at a mechanistic level, it remained unclear how phosphorylation affects starch turnover.

Under in vivo conditions, phosphorylation of native starch particles occurs both during net biosynthesis (Nielsen et al., 1994) and net degradation, but, at least in chloroplasts, the rate of phosphorylation is higher during net starch mobilization (Ritte et al., 2004). More recently, we observed that β-amylolysis is enhanced by simultaneously occurring phosphorylation of transitory starch granules (Edner et al., 2007). Based on these in vitro results, it has been proposed that GWD-mediated phosphorylation reduces the molecular order of glucan helices at the surface of the starch particle, and thereby renders glucan chains accessible to hydrolytic enzymes (Edner et al., 2007).

In the present paper, this assumption has been tested using crystalline maltodextrins as model carbohydrates. For several reasons, the MDcryst preparations used appear to be a suitable substrate if the initial steps of starch degradation are to be analyzed. Firstly, MDcryst, transitory starch and potato tuber starch all represent the B-type allomorph (Figure 1). Secondly, a high specific GWD activity was observed when using MDcryst, ranging from 55 to 60 nmol phosphate esterified per mg protein per minute (Table 2 and Figure 2). These values are one or two orders of magnitude higher than those reported for amylopectins or for high-DP amyloses (Mikkelsen et al., 2004, 2006; Ritte et al., 2002), and also exceed those observed with native leaf starch granules (Edner et al., 2007). Thirdly, using MDcryst as the phosphate acceptor, GWD formed a range of singly, doubly and triply phosphorylated maltodextrins (Figure 5). These data strongly resemble the pattern of phosphoglucans obtained in a previous study in which phosphoglucans were released from potato leaf starch granules by treatment with isoamylase (Ritte et al., 2004). In this study, the native starch granules had been isolated from leaves performing net starch degradation, and the starch-derived phosphoglucans reflect the in vivo phosphorylation of the granule surface. Finally, as observed with native starch granules, PWD acts mainly downstream of GWD (Figure 2; Kötting et al., 2005). In summary, both the high rates of phosphorylation and the phosphorylation pattern, as observed in the current study, strongly suggest that crystalline maltodextrins (at least if they represent the B-type allomorph) mimic important features of the physiological carbohydrate substrate of GWD, i.e. target structures at the surface of native starch.

This conclusion is supported by the almost complete loss of GWD activity when MDcryst was solubilized by heat treatment (Figure 2a). Similarly, no PWD activity was detectable when, following pre-phosphorylation by GWD, the insoluble phosphorylated maltodextrins were converted to a soluble state (Figure 2b). Thus, the activity of GWD (and, to some extent, that of PWD) depends more strongly on the molecular order of α-glucans than on structural parameters of the individual glucan molecule, such as DP or the presence of branching points.

The glucan specificity of GWD (and that of PWD as well) differs strikingly from that of BAM3 (At4g17090). The activity of this plastidial β-amylase isozyme is almost two orders of magnitude higher with MDsol compared to that observed with MDcryst (Table 4). Similarly, BAM3 posseses a strong preference for heat-solubilized starch over native starch granules (Edner et al., 2007).

Table 4.   Specific activity of GWD, PWD and BAM3 with MDcryst or MDsol as substrates
  1. PWD was assayed using MDcryst that had been pre-phosphorylated with GWD and unlabeled ATP. GWD and PWD activities are given as nmol phosphate esterified per min per mg protein. BAM3 activity is given as μmol maltose released per min per mg protein. ND, not detectable; GWD, glucan, water dikinase; PWD, phosphoglucan, water dikinase.


In in vitro experiments, BAM3 is capable of releasing maltose from the surface of native starch granules. BAM3 and GWD display mutual stimulation of their activities: the rate of β-amylolysis is enhanced by simultaneously occurring phosphorylation, and the action of BAM3 increases the rate of phosphorylation (Edner et al., 2007). It is therefore unlikely that the action of BAM3 is restricted to soluble glucans in vivo. Rather, it appears that BAM3 utilizes targets at the starch granule surface (such as extended single chains) that lack crystallinity and are not present on MDcryst. The high degree of crystallinity observed for MDcryst (Table 1) is consistent with this assumption. When the single glucan chains located at the starch granule surface are degraded by β-amylolysis, the accessibility of GWD to its crystalline substrate is increased (Edner et al., 2007).

Phosphorylation of MDcryst, as catalyzed by GWD, initiates the transition of both phosphorylated and neutral α-glucans from the crystalline to the soluble state. The amount of solubilized (phospho)glucans increases with the relative amount of phosphate esters introduced by GWD. With a high rate of phosphorylation, approximately 2% of the glucosyl residues of the soluble maltodextrins were esterified (Figure 3b), but the minimum degree of phosphorylation required for solubilization is difficult to estimate. Solubilization may be caused by the introduction of hydrophilic phosphate groups, leading to complete hydration of glucan helices. However, under in vivo conditions, the involvement of further proteins cannot be excluded. Nevertheless, the data presented here provide an explanation of how, at a mechanistic level, the phosphorylation of starch enhances the hydrolytic degradation of native starch granules.

An important difference between MDcryst and the surface of native starch granules should be noted. With the latter, complete hydration of glucan chains does not result in solubilization, as they remain covalently bound to the entire starch granule irrespective of their molecular order. Therefore, phosphorylation leads to local alterations of the starch particle surface, such as an increased accessibility for plastidial β-amylases (Edner et al., 2007) and possibly for other starch-degrading enzymes.

Based on the data shown in Table 2, recombinant GWD appears to preferentially act on the B-type allomorph. Of special interest is the kernel starch from the maize amylose extender (ae) mutant, which forms the B-type allomorph whereas wild-type maize (and other maize mutants) synthesize the A-type allomorph. The rate of GWD-catalyzed phosphorylation of ae starch is approximately one order of magnitude higher than that of other maize starches. However, crystallized maltodextrins representing the A-type allomorph have not been tested in this study. Furthermore, even within the B-type allomorph starch, the rates of phosphorylation vary by more than one order of magnitude depending upon the plant source. Therefore, other features of the starch granule also significantly affect GWD activity. Presumably, at the surface of the various starches, more subtle structural differences exist that cannot be identified by X-ray diffraction analysis.

Experimental procedures

Chemicals and enzymes

The following commercially available products were used: [γ-33P]ATP (10 mCi ml−1; 3000 Ci mmol−1) and [β-33P]ATP (10 mCi ml−1; 3000 mCi mmol−1) from Hartmann Analytic (, malto-oligosaccharides (EC 232-940-4) and starch from maize kernels (EC 232-679-6) from Sigma-Aldrich (, potato tuber starch from ICN (; catalogue no. 102954), glucose-6-phosphate dehydrogenase from Leuconostoc mesenteroides, hexokinase from yeast, and the starch kit from Roche Diagnostics (, and PEI-cellulose F plates from Merck (

Starch from wheat (wild-type Florida) and maize (genotype A188) were provided by Bayer Bioscience ( Extender starch was obtained from National Starch ( Waxy maize (PaW64A) was obtained from the Maize Genetics Cooperation Stock Center ( For the expression and purification of recombinant GWD from Solanum tuberosum L., and PWD and BAM3 from Arabidopsis thaliana L. (At4g17090), see Ritte et al. (2002), Kötting et al. (2005) and Edner et al. (2007).

[33P]glucose-6-phosphate was synthesized enzymatically (30 min at 30°C) in a reaction mixture (final volume 45 μl) containing 100 mm imidazole/HCl (pH 6.9), 3 units of hexokinase, 6 mm MgCl2, 1 mm d-glucose, 250 μm unlabeled ATP and 2 μCi [γ-33P]ATP. The reaction was terminated by heating for 5 min at 95°C.

Thin-layer chromatography

TLC was performed according to the method described by Ziegler et al. (1998).

Protein estimation

For protein quantification, the microversion of the Bio-Rad protein assay (Bradford, 1976; was used. BSA was used as a standard.

Crystallization of maltodextrins

The procedure was similar to that described by Gidley and Bulpin (1987). Maltodextrins from maize were dissolved in water (30% w/v), heated for 10 min at 95°C, cooled to room temperature, and then kept overnight at 4°C. Crystalline material was collected by centrifugation (10 min, 800 g), washed five times with ice-cold water (30 ml each; centrifugation as above), and then lyophilized. For quantification, the starch kit from R-Biopharm (; starch quantification kit, 10207748034) was used following the instructions of the manufacturer.

X-ray diffraction analysis

Glass capillaries were filled with samples suspended in water. Diffraction measurements were performed using a NanoStar instrument (Bruker AXS, using Cu–Kα radiation (λ = 0.154 nm) at a sample detector distance of 7.5 mm. Data were collected using a 2D position-sensitive detector covering a range of diffraction angles (2.5°<2θ < 27°), and were azimuthally averaged to obtain 1D diffraction profiles of scattered intensity versus 2θ. Background scattering arising from water and the capillary was measured separately and subtracted from each dataset. Selected specimens were also measured in the dry state at ambient humidity using a standard powder diffractometer (D8, Bruker AXS).

Enzymatic assays and labeling experiments

GWD activity was monitored using either native starch granules (assay I) or maltodextrins (assay II). In both assays, GWD activities increased linearly with the concentration of the recombinant enzyme.

GWD assay I.  In a final volume of 250 μl, starch granules (as specified; 20 mg ml−1) were resuspended in reaction buffer (50 mm HEPES/KOH pH 7.5, 6 mm MgCl2, 2 mm EDTA, 2 mm DTT) containing 25 μm unlabeled ATP, 1 μCi [β-33P]ATP and 0.1 μg recombinant potato GWD. Following incubation (30 min at 30°C) under continuous agitation, SDS (2% w/v final concentration) was added, and the starch granules were pelleted by centrifugation (10 min at 10 000 g). After washing five times (2% SDS containing 2 mm ATP each; 10 min for each wash including centrifugation), granules were resuspended in 100 μl water, mixed with 3 ml scintillation cocktail, and radioactivity was monitored using a liquid scintillation counter.

GWD assay II.  The standard reaction mixture (50 μl) contained reaction buffer (as above), MDcryst (10 mg ml−1) or, where stated, MDsol, 25 μm ATP plus 2 μCi [β-33P]ATP, 0.1 μg (except where stated) recombinant GWD and BSA (400 μg ml−1). After incubation (30 min at 30°C except where stated) under continuous agitation, the reaction was terminated by heating for 5 min at 95°C, and the reaction mixture was centrifuged for 10 min at 10 000 g. Subsequently, 2 μl of the supernatants were spotted onto PEI-cellulose plates. In assay II, GWD activity was monitored by quantifying the 33P content of the phosphoglucan-containing spot obtained by TLC using a phosphor imager (FLA-3000, Fujifilm, The relative 33P content of the spot was converted into absolute amounts of phosphate esters using the specific radioactivity of the ATP in the reaction mixture and the total amount of radioactivity applied to each lane.

PWD activity assay.  PWD activity was monitored using both MDcryst and pre-phosphorylated MDcryst as substrate (see GWD assay II). For pre-phosphorylation of MDcryst, a reaction mixture (500 μl) containing reaction buffer (see above), 10 mg ml−1 MDcryst, 25 μm unlabeled ATP and 0.3 μg GWD was incubated for 60 min at 30°C under continuous agitation (see above). Subsequently, GWD was removed by adding SDS to a final concentration of 2% w/v, and the reaction mixture was centrifuged for 10 min at 10 000 g. The insoluble pre-phosphorylated MDcryst was freed from SDS by washing eight times (10 min each) with water. Finally, the pellet was resuspended in 500 μl water and used as a substrate for PWD. In a final volume of 50 μl, the reaction mixture contained reaction buffer (as above), 4 mg ml−1 pre-phosphorylated (but unlabeled) MDcryst, 25 μm unlabeled ATP and 10 μCi [β-33P]ATP. All further steps were performed as in GWD assay II.

Quantification of glucosyl-6-phosphate residues.  Maltodextrins were hydrolyzed using either HCl or TFA. For enzymatic G-6-P quantification, the GWD assay mixtures were hydrolyzed in 0.7 N HCl (4 h at 95°C). Following neutralization with 0.7 N NaOH, G-6-P was quantified enzymatically (Nielsen et al., 1994). For TLC analyses, samples were hydrolyzed in 2 m TFA (2 h at 95°C), dried under vacuum, dissolved in water and dried again.

Pulse-chase labeling experiments.  In a final volume of 500 μl, MDcryst (10 mg ml−1) was pre-labeled for 60 min at 30°C under continuous agitation. The mixture contained reaction buffer (see above), 25 μm unlabeled ATP plus 20 μCi [β-33P]ATP, 0.3 μg recombinant GWD and 0.4 mg ml−1 BSA. As a control, a reaction mixture containing heat-inactivated GWD was treated identically. The reaction was terminated by adding 2% w/v SDS (final concentration). Following centrifugation (as above), the pellet was washed eight times (10 min each) with water, and the 33P-labeled MDcryst was resuspended in 100 μl water. Subsequently, aliquots of MDcryst (20 μl each) were incubated in reaction buffer (as above), 25 μm unlabeled ATP and 1 μg GWD at 30°C (100 μl final volume). As a control, heat-inactivated GWD was used. At intervals EDTA was added to aliquots of the reaction mixture to give a final concentration of 20 mM, and soluble oligosaccharides were separated from insoluble components by centrifugation (10 min at 10 000 g). The pellet was resuspended in 100 μl water. Finally, aliquots of soluble and insoluble glucans (10 μl each) were subjected to scintillation counting.

β-Amylolysis of soluble and crystalline maltodextrins.  MDcryst or MDsol (4 mg each) were resuspended or dissolved in a mixture (600 μl) containing 30 mm HEPES-KOH (pH 7.5), 5 mm MgCl2, 5 mm CaCl2, and 1 mg ml−1 BSA. β-Amylolysis was started by adding either 150 ng or 4 μg of recombinant β-amylase 3. Reaction mixtures were incubated at 25°C under continuous agitation. Reactions were stopped at intervals either by heating at 95°C for 5 min (MDsol) or by centrifugation (MDcryst) and heating (as above) of the supernatant. Reducing ends were quantified according to the method described by Waffenschmidt and Jaenicke (1987) using maltose as the standard.

Analysis of (phospho)glucans by HPAEC or chromatography on carbon columns

HPAEC-PAD analyses.  Neutral and phosphorylated maltodextrins were analyzed by HPAEC-PAD (Dionex ICS-3000 system containing a CarboPac PA100 or PA1 column; Throughout this study, two modes of sample preparation and HPAEC were used. For procedure A, α-glucans (32 μg glucose equivalents) were dissolved in water without any further purification and were loaded onto a column that had been equilibrated for 10 min with 5 mm sodium acetate, dissolved in 100 mm NaOH. For elution, a linear sodium acetate gradient (5–500 mm, dissolved in 100 mm NaOH; 110 min) was applied. Throughout HPAEC, the flow rate was 1 ml min−1. Alternatively (procedure B), samples were centrifuged for 10 min at 10 000 g, and the supernatants were freed from proteins using centrifugal filter units (Ultrafree-MC, Millipore, Equal volumes of the filtrates were applied to the column (equilibrated as above). Oligosaccharides were eluted using a linear sodium acetate gradient (5–500 mm; 30 min; 1 ml min−1), followed by elution with 500 mm sodium acetate dissolved in 100 mm NaOH (10 min; 1 ml min−1).

Chromatography on graphitized carbon black columns.  For separation of neutral and phosphorylated oligoglucans at a preparative scale, pre-packed non-porous graphitized carbon black columns (Carbograph SPE, Alltech, were used (Packer et al., 1998). Prior to sample application, the columns were washed with a mixture of 80% v/v acetonitrile and 0.2% v/v TFA, and were then equilibrated with water. Subsequently, 100 μl of the filtrate (see procedure B) were diluted with 900 μl water and loaded onto a carbon column. A two-step elution was used. First neutral oligosacchrides were eluted using 3 × 700 μl 30% v/v acetonitrile. Then phosphorylated oligosaccharides were eluted using 3 × 700 μl 30% v/v acetonitrile containing 0.2% v/v TFA. The eluates were dried under vacuum. The pellets were dissolved in water and were then analyzed by HPAEC-PAD (procedure B) using an online desalting device.

MS analysis of (phospho)glucans

Samples separated by HPAEC were desalted and concentrated using a miniaturized version of the chromatography on non-porous graphitized carbon black (Carbograph SPE, Alltech; see above) according to the method described by Gobom et al. (1999). A small quantity of the column material was placed into GELoader tips (Eppendorf, Following elution, samples were used directly for MS analysis.

Aliquots (0.5 μl each) were mixed on the target with 0.5 μl of DHB matrix solution (2,5-dihydroxybenzoic acid; 15 mg ml−1, dissolved in 30% v/v aqueous methanol), and the analyte–matrix mixtures were dried under a gentle stream of air.

MALDI-TOF mass spectra were recorded on a Bruker Reflex II spectrometer (Bruker Daltonik, in positive-ion mode. All spectra were measured in reflector mode using external calibration (angiotensin II).


This study was supported by the Deutsche Forschungsgemeinschaft (SFB 429 ‘Molecular Physiology, Energetics and Regulation of Primary Plant Metabolism’ TP B2 to M.S. and J.F. and B7 to G.R. ). The authors are indebted to Dr Jing Li and Professor Steven M. Smith (University of Western Australia, Crawley, Australia) for kindly providing the BAM3 expression vector. The authors thank Ms Ingrid Zenke (Max-Planck Institute of Colloids and Interfaces, Department of Biomaterials) for excellent technical assistance.