Phosphorylation of transitory starch by α-glucan, water dikinase during starch turnover affects the surface properties and morphology of starch granules


  • Sebastian Mahlow,

    1. Plant Physiology, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    2. Biopolymers Analytics, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    Search for more papers by this author
    • These authors contributed equally to the work.
  • Mahdi Hejazi,

    1. Plant Physiology, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    Search for more papers by this author
    • These authors contributed equally to the work.
  • Franziska Kuhnert,

    1. Plant Physiology, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    Search for more papers by this author
  • Andreas Garz,

    1. Photonics, Institute of Physics and Astronomy, University of Potsdam, Potsdam-Golm, Germany
    Search for more papers by this author
  • Henrike Brust,

    1. Plant Physiology, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    Search for more papers by this author
  • Otto Baumann,

    1. Zoophysiology, Institute of Biochemistry and Biology, University of Potdam, Potsdam-Golm, Germany
    Search for more papers by this author
  • Joerg Fettke

    Corresponding author
    1. Plant Physiology, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    2. Biopolymers Analytics, Institute of Biochemistry and Biology, University of Potsdam, Potsdam-Golm, Germany
    Search for more papers by this author


  • Glucan, water dikinase (GWD) is a key enzyme of starch metabolism but the physico-chemical properties of starches isolated from GWD-deficient plants and their implications for starch metabolism have so far not been described.
  • Transgenic Arabidopsis thaliana plants with reduced or no GWD activity were used to investigate the properties of starch granules. In addition, using various in vitro assays, the action of recombinant GWD, β-amylase, isoamylase and starch synthase 1 on the surface of native starch granules was analysed.
  • The internal structure of granules isolated from GWD mutant plants is unaffected, as thermal stability, allomorph, chain length distribution and density of starch granules were similar to wild-type. However, short glucan chain residues located at the granule surface dominate in starches of transgenic plants and impede GWD activity. A similarly reduced rate of phosphorylation by GWD was also observed in potato tuber starch fractions that differ in the proportion of accessible glucan chain residues at the granule surface.
  • A model is proposed to explain the characteristic morphology of starch granules observed in GWD transgenic plants. The model postulates that the occupancy rate of single glucan chains at the granule surface limits accessibility to starch-related enzymes.


Starch represents the predominant carbohydrate reserve in plants. It consists of two polymers: the branched α-1,4:α-1,6 d-glucan amylopectin, which accounts for 70–90% of starch DW, and the almost unbranched amylose. The structure of amylopectin is responsible for starch being present as semi-crystalline granules. The chain length distribution and arrangement of branch points lead to formation of ordered arrays of densely packed double helices (Smith, 2001; Zeeman et al., 2002). Two crystalline forms, A and B, can be distinguished by X-ray diffraction analysis, which depend on the packing arrangement of the double helices. The B-type is the typical allomorph in the transitory starch of Arabidopsis leaves (Hejazi et al., 2009).

In addition to glucosyl residues, starch contains small amounts of proteins, lipids and phosphate (Hoover, 2001). Unlike proteins and lipids, which are noncovalently attached to the starch granule, phosphate is covalently attached via ester linkages to glucosyl residues of amylopectin (Hizukuri et al., 1970). Starch is continuously phosphorylated during both net biosynthesis and degradation, with different rates depending on the metabolic status (Ritte et al., 2004). Labelling studies using photoautotrophic cultures of Chlamydomonas reinhardtii revealed that the phosphorylation rate is increased when starch is mobilized in the dark. Furthermore, in potato leaves the phosphorylation level at the granule surface is considerably higher during starch breakdown than during starch biosynthesis (Ritte et al., 2004). These results indicate that starch phosphorylation during degradation is transient. However, it is reasonable to assume that starch phosphorylation has an important role in the entire transitory starch metabolism, rather than being functional only during degradation process.

Two plastidial starch phosphorylating enzymes have been identified: glucan, water dikinase (GWD, EC; Ritte et al., 2002, 2006) and phosphoglucan, water dikinase (PWD, EC; Baunsgaard et al., 2005; Kötting et al., 2005). GWD and PWD selectively catalyse the formation of C6 and C3 phosphoesters of glucosyl residues, respectively. The starch-related dikinases use ATP as a dual phosphate donor. GWD transfers the terminal γ-phosphate group of ATP to water forming orthophosphate. The β-phosphate group is first transferred to a conserved histidine residue located within the catalytic domain of the enzyme, and then to the C6 hydroxyl group of a glucosyl residue in amylopectin (Ritte et al., 2002, 2006). Interestingly, additional phospho-transfer reactions have been reported for GWD (Hejazi et al., 2012b). PWD follows the same mode of action as GWD.

In Arabidopsis, there are three phosphoglucan phosphatases involved in dephosphorylation of glucans: STARCH EXCESS4 (SEX4), LIKE-STARCH-EXCESS FOUR-1 (LSF1) and LIKE-STARCH-EXCESS FOUR-2 (LSF2). Two of them, SEX4 and LSF2, are able to remove C6- and/or C3-bound phosphate esters in vitro and both enzymes are required for normal starch breakdown at night (Kötting et al., 2009; Comparot-Moss et al., 2010; Hejazi et al., 2010; Santelia et al., 2011).

GWD activity is increased in response to darkness (Ritte et al., 2004). Using native starch granules and β-amylases, phosphorylation at the surface of starch granules by GWD was shown to increase the hydrolytic action of plastidial β-amylases. In addition, hydrolytic reactions on the granule surface stimulate phosphorylation by GWD (Edner et al., 2007). It was concluded that the action of GWD enables a phase transition of glucan chains at the granule surface from a high degree of order to a more soluble state. GWD is able to act significantly on native starch granules (Hejazi et al., 2008), unlike enzymes that hydrolyse α-1,4 or α-1,6 interglucose linkages. The latter enzymes preferentially act on soluble polyglucans (Lloyd et al., 2005; Edner et al., 2007; Fettke et al., 2012a). Crystalline maltodextrins (MDcryst) with a higher degree of crystallinity than native starch granules have been shown to be suitable substrates for in vitro activity assays of recombinant GWDs from potato and Arabidopsis. The rate of MDcryst phosphorylation was higher than that observed with any native starches tested so far. Therefore, it was concluded that GWD preferentially acts on highly ordered and crystalline structures (Hejazi et al., 2009).

Despite the low frequency of phosphorylation (c. 0.1% of glucosyl residues are phosphorylated in Arabidopsis leaf starch), it is essential for normal starch metabolism, as GWD-deficient Arabidopsis mutants have a starch excess phenotype (sex1; Yu et al., 2001). Leaves of GWD-deficient lines have increased starch concentrations that are at least five times higher than those of wild-type (WT) leaves. Even after prolonged dark periods the starch is not totally degraded. In addition, the plants are strongly compromised in growth (Yu et al., 2001; Ritte et al., 2002; Baunsgaard et al., 2005; Kötting et al., 2005).

So far, there is little or no available information about the properties of starch granules from GWD-deficient plants, such as size, morphology, crystallinity, thermal stability and surface properties. Such information could provide important clues to help us understand the role of GWD activity in starch metabolism and its regulation. Therefore, we have characterized isolated starch granules from WT plants and the sex1-8 null mutant, along with starches from transgenic Arabidopsis lines with different levels of GWD expression. We show that the surface, rather than the inner structure of the granules, differs to that of WT, and that there is a correlation between accessible glucans on the granule surface and glucan phosphorylation by GWD. Furthermore, the accessible glucan chains also affect enzymes involved in starch synthesis and thus are responsible for the overall starch granule morphology.

Materials and Methods

The following products were used: [β-33P]ATP (3000 Ci mmol−1) (Hartmann Analytic, Germany); isoamylase (5MU) from Pseudomonas sp. and 8-aminopyrene-1,3-6-trisulfonic acid (APTS) (Sigma-Aldrich); SuperScript II reverse transcriptase (Invitrogen); Phusion high-fidelity DNA polymerase, dNTPs, glucose 6-phosphate dehydrogenase from Leuconostoc mesenteroides, hexokinase from yeast, Taq DNA polymerase, and starch kit (Roche Diagnostics); pGEM-T Easy cloning vector (Promega); NucleoTrap clean-up kit (Macherey-Nagel, Düren, Germany). [U-14C]ADPglucose was prepared as described in Fettke et al. (2012b).

For the expression and purification of recombinant β-amylase 3 from Arabidopsis thaliana (L.) Heynh. (AtBAM3 (At4g17090)), see Edner et al. (2007). GWD from potato (Solanum tuberosum L.; StGWD) was expressed and purified according to Hejazi et al. (2012b).

Plant material and growth conditions

The SALK T-DNA mutant sex1-8 (Salk line no. 077211.26.50.x) was obtained from the Nottingham Arabidopsis Stock Centre (NASC, The transgenic plants gwd-c1 and gwd-c2 were generated by complementation of sex1-8 plants with constructs containing the AtGWD cDNA (see ‘Cloning of Arabidopsis STARCH SYNTHASE 1 (AtSS1) and StrepII-tagged-GWD’). Plants were cultivated together with wild-type (WT, Col-0) in a growth chamber under controlled conditions (12 h : 12 h, light : dark), 22°C : 17°C, and c. 100 μmol quanta m−2 s−1. After 6 wk, rosette leaves from plants were harvested and directly frozen in liquid nitrogen and stored at −80°C until use.

Starch isolation and determination of starch density

Starch granules were isolated from Arabidopsis leaves according to Kötting et al. (2005).

Potato tuber starch was isolated from the variety Désirée as published by Ritte et al. (2004). Starch density was determined according to Dengate et al. (1978) using 1 g of starch granules, 4 mg ml−1 blue dextran solution and a 5-ml volumetric flask.

Cloning of Arabidopsis STARCH SYNTHASE 1 (AtSS1) and StrepII-tagged-GWD

The cloning and expression of AtSS1 (At5g24300) were the same as described for AtSS3 in Fettke et al. (2011). First-strand cDNA was synthesized using reverse transcriptase and the 3′primer (5′-GCTGACATAGGGAGGGTCCATGAAAACCCA-3′). The resulting cDNA was amplified by PCR using the forward primer (5′-GAATTCGTCTTCTTCCTTCTCCGGTGACTC-3′) including an EcoRI restriction site and the reverse primer (5′-CTCGAGGCTGACATAGGGAGGGTCCATGAA-3′) with a XhoI restriction site.

The sex1-8 mutant was complemented with constructs encoding Arabidopsis GWD fused to a C-terminal StrepII-tag.

For cloning of the cDNA encoding GWD (At1g10760) total RNA was isolated from Arabidopsis rosette leaves as described by Kartal et al. (2011). First-strand cDNA encoding the precursor of AtGWD was synthesized without the stop codon by using a specific SEX1rev primer containing an AgeI restriction site (5′-ACCGGTCACTTGTGGTCGTGTCTGGACGAC-3′) and reverse transcriptase according to the manufacturer's instructions. The cDNA was amplified by Phusion DNA polymerase using SEX1fwd primer containing a SmaI restriction site (5′-CCCGGGATGAGTAACTCTGTAGTGCATAAC-3′) and SEX1rev primer. The resulting PCR product was blunt-ligated into HincII digested pGEM vector and checked for orientation to yield pGEM/AtGWD.

For construction of a DNA fragment encoding the StrepII-tag and a C-terminal stop codon, the following primers were used: StrepIIfwd, which contains an AgeI site (5′-ACCGGTTGGAGCCACCCGCAGTTCGAAAAATGAGTCGAC-3′) and StrepIIrev with a SalI site (5′-GTCGACTCATTTTTCGAACTGCGGGTGGCTCCAACCGGT-3′). A primer dimer was formed by incubating a mixture containing 1.5 μg of each primer in a final volume of 100 μl at 72°C for 30 s, followed by incubation with Taq DNA polymerase and dATP (72°C, 20 min) to generate 3′A-overhangs. The resulting product was purified using a NucleoTrap clean-up kit and ligated into pGEM-T Easy vector. The plasmid was digested with AgeI and NaeI and ligated to AgeI and NaeI digested pGEM/AtGWD to achieve a sequence encoding the C-terminally StrepII-tagged AtGWD. The fragment encoding the fusion protein was excised from the cloning vector with SalI and SmaI and cloned into SalI/SmaI digested binary vector pBinAR-Hyg (Kossmann et al., 1999).

Plant transformation

The AtGWD-StrepII construct in the pBINAR-Hyg binary vector was introduced into competent GV3101 (pMP90) Agrobacterium tumefaciens cells and transformed into the sex1-8 mutant (Col-0 background) by floral dip method (Clough & Bent, 1998). T0 transformants were selected on solid MS media (1x Murashige & Skoog salts, 1% (w/v) sucrose, 0.5 g MES (w/v), 0.8% (w/v) agar, pH adjusted to 5.7 with NaOH) supplemented with 30 mg l−1 hygromycin and 100 mg l−1 carbenicillin. Resistant plants were transferred to soil and cultivated in the green house. The transformants were selfed and zygosity was determined by segregation analysis in the T1 and T2 generations. Two homozygous T3 lines (gwd-c1 and gwd-c2) that expressed the GWD fusion protein, as shown by western blot analysis, were used in the described experiments.

Protein extraction, PAGE and immunoblotting

For native affinity-electrophoresis, 400 mg of frozen leaf material were extracted with 0.8 ml extraction buffer containing: 50 mM HEPES/KOH (pH 7.0), 1 mM EDTA, 5 mM dithioerythritol (DTE), 0.5 mM phenylmethanesulfonylfluoride (PMSF) and 10% (w/v) glycerol. Following centrifugation (20 000 g for 10 min at 4°C), aliquots of the supernatant were assayed for total protein content using BSA as standard (Bradford, 1976). Aliquots containing 7 μg (for phosphorylase activity) and 60 μg (for starch synthase activity) of soluble protein were separated by native PAGE as described by Steup (1990). For analysis of AtPHS1, AtPHS2, AtSS1 and AtSS3 activities, the gels were stained and quantified according to Fettke et al. (2005) and Brust et al. (2013), respectively.

Samples for SDS-PAGE were prepared as described above with a different extraction buffer (50 mM HEPES/KOH, pH 7.5, 1 mM EDTA, 5 mM DTE, 2 mM 6-aminocaproic acid, 2 mM benzamidine and 1 mM PMSF). Soluble proteins were mixed with 3× concentrated SDS-sample buffer and incubated at 95°C for 5 min.

SDS-PAGE and immunoblotting were carried out as described by Fettke et al. (2004). AIDA software was used for quantification of proteins after immunoblotting.


Polyclonal antibodies used for the detection of AtGWD, AtPWD and AtSEX4 are described in Ritte et al. (2000), Kötting et al. (2005) and Niittylä et al. (2006), respectively. The polyclonal rabbit antibody directed against AtSS4 (At4g18240) was a generous gift from Professor Ángel Meridá (Universidad de Sevilla, Spain). The polyclonal rabbit antibody used for the detection of AtSS2 (At3g01180) was a generous gift from Professors Michael J. Emes and Ian Tetlow (University of Guelph, Canada). The monoclonal mouse antibody directed against AtBE2 (At5g03650) and the polyclonal rabbit antibody directed against AtBE3 (At2g36390), respectively, were generous gifts from Professor Christophe D'Hulst (Université des Sciences et Technologies de Lille 1, France).

Enzyme assays

α-Glucan, water dikinase

For all GWD assays, the 33P incorporation into starch was proportional to enzyme concentration. Unless otherwise stated, 30 mg dried starch were resuspended in 50 mM HEPES/KOH, pH 7.4, 1 mM EDTA, 6 mM MgCl2, 2 mM DTE, 0.4 mg ml−1 BSA and 500 ng recombinant StGWD. The reaction was started by addition of 25 μM ATP and 2 μCi [β-33P]ATP (300 μl final volume). Samples were incubated at 30°C under continuous agitation for 9 min. Every 3 min, 100-μl samples were removed and mixed with 30 μl of 10% (w/v) SDS to stop the reaction. The starch granules were washed seven times with water (0.7 ml each) to remove unbound radioactivity. Washed starch granules were resuspended in 200 μl water and mixed with 3 ml scintillation liquid. The incorporation of 33P was determined by scintillation counting.

β-Amylolysis of starch granules

In a final volume of 1.3 ml, 60 mg of native starch granules were resuspended in 50 mM HEPES/KOH, pH 7.4, 0.4 mg ml−1 BSA, 2 mM DTE, 2 mM CaCl2 and 2 μg recombinant AtBAM3. The reaction mixtures were incubated at 30°C for 44 h with continuous agitation. At intervals, released maltose was separated by centrifugation for 10 min at 14 000 g at room temperature. The supernatants were collected and starch pellets were resuspended in fresh buffer containing AtBAM3. The maltose content in the supernatants was determined using the reducing end assay according to Waffenschmidt & Jaenicke (1987). Finally, starch pellets were treated with SDS (1% (w/v) final concentration) to remove AtBAM3 and washed seven times with water (1 ml each). Subsequently, aliquots (30 mg each) of treated starch were used for phosphorylation with StGWD (see earlier).

Isoamylase digestion of starch granules

Isoamylase digestion was performed using either native starch granules or solubilized starches generated by heat treatment (5 min at 95°C). One milligram of heat-treated starch, or 10 mg of native starch granules were incubated with 10 mM Na-acetate buffer, pH 5.5, 2 mM DTE and 12 U of isoamylase (100 μl final volume) overnight at 40°C with continuous agitation. The samples were centrifuged for 10 min at 14 000 g. The supernatant was decanted, heated to 95 °C for 5 min and then passed through a 10 kDa filter unit to remove the inactivated enzyme. The chain length distribution was analysed using a capillary electrophoresis apparatus equipped with laser induced fluorescence detection (CE-LIF). Following assay of reducing ends, 10 nmol of released glucan chains were labelled with 8-aminopyrene-1,3-6-trisulfonic acid (APTS), then separated and quantified as described by Malinova et al. (2014). In other experiments, 20 mg of WT or sex1-8 starch were incubated with isoamylase as already described. The supernatants containing free glucan chains were concentrated using a centrifugal vacuum concentrator. The pellets from each starch digestion were treated with SDS (2% (w/v) final concentration) and freed from SDS by washing with water (seven times). After washing, the WT and sex1-8 starches were mixed with the free glucan chains derived from sex1-8 and WT starch, respectively, and used for phosphorylation assays with recombinant StGWD.

Soluble starch synthase 1

The incorporation of glucosyl residues into starch granules was proportional to protein concentration. The reaction mixture contained 10 or 14 mg native starch granules, 100 ng AtSS1, 1 mM ADPglucose, 0.08 μCi [glucosyl-14C]ADPglucose, 200 mM Na-citrate, 0.025% (w/v) BSA, 10 mM Tricine/NaOH, final pH 8.0. Samples were incubated at 30°C with continuous agitation. Aliquots were taken at intervals and the reaction was stopped by adding SDS (final concentration 2% (w/v)). Following washing of starch granules, incorporation of 14C-glucosyl residues was quantified using scintillation counting. In another assay, starch granules were prephosphorylated with GWD and ATP overnight (see earlier) before incubation with AtSS1 and [glucosyl-14C]ADPglucose.

Estimation of starch content and glucose 6-phosphate

For enzymatic determination of starch and glucose 6-phosphate content, isolated starch granules were hydrolysed in 0.7 N HCl for 2 h at 95°C with continuous agitation. Samples were neutralized with 0.7 N NaOH and starch content (glucose equivalents) was quantified using the starch kit. Glucose 6-phosphate was measured as described by Haebel et al. (2008). Leaf starch content was determined according to Fettke et al. (2011).

Scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray diffraction measurements and light scattering experiments

Dried starch granules were sprinkled onto double-sided, carbonated, sticky stills attached to SEM stubs and imaged at 1 kV in an SEM Ultra Plus instrument (Zeiss, Germany). Data were collected using the SE2 detector. Measurements of granule radii were made with SmartTiff software (Zeiss).

For TEM specimens see Malinova et al. (2014). X-ray diffraction was done as described in Hejazi et al. (2008). The relative changes in the light scattering intensities of starch suspensions were monitored at an angle of 90° using a Fluorolog-3 fluorescence spectrometer (Horiba Jobin Yvon GmbH, Munich, Germany), as described in Hejazi et al. (2009).

Preparation and fixation of leaf material

Leaf material was prepared using the ‘Tape-Arabidopsis Sandwich’ method described by Wu et al. (2009) for nonlinear microscopy using the second harmonic generation on starch as signal source. Leaf material was harvested from 6-wk-old plants at the end of the light period. After removing the lower epidermis, the peeled leaves were decolourized in 70% (v/v) ethanol until endogenous pigments were fully extracted. Leaf material was rehydrated by incubation in water before analysis.

Second harmonic generation measurements and image analysis

For the nonlinear microscope set-up, see Garz et al. (2012). Pixel dwell time was 2.64 μs for sex1-8 and 12.72 μs for WT. Linear correlation between two measurement conditions was verified. In every slice the starch granules were identified: first, individual granules with a diameter of at least 2 μm were labelled; second, all signals were summed over the cell. Adjacent slices were then stacked to build a 3D image. Starch density was determined using the volume and second harmonic generation signal (starch content).


Absence of GWD results in altered starch granule morphology but no change in the physicochemical properties of starch granules

The internal structure of starch granules isolated from Arabidopsis WT and sex1-8 were analysed by small angle X-ray scattering (NanoStar) using fully hydrated starch granules (Supporting Information Fig. S1). The diffraction patterns of WT and sex1-8 starch were very similar, displaying typical characteristics of a B-type allomorph. Both starch granules possessed a similar crystallite size, which is comparable to previous studies on WT (data not shown; Hejazi et al., 2009).

In addition, the thermal stability of starch granules from both genotypes was studied by monitoring temperature-dependent light scattering. There were no significant differences between the two starch types in Tonset, Tpeak and Tend (Table S1).

Starch granules isolated from WT and sex1-8 plants were subjected to morphological analysis by uncoated scanning (SEM) and transmission (TEM) electron microscopy. Modern sample coating techniques for imaging generate thin surface coats of c. 10 nm thickness. Such coats, however, are particulate, leading to a potential problem of misinterpretation of results due to the presence and size of metal grains. High resolution imaging with uncoated starch granules was performed using low-voltage SEM. Starch granules were isolated from 6-wk-old rosettes of WT and sex1-8 harvested at the end of the light period. Each starch type displayed a relatively homogeneous population in shape but the size of the granules from sex1-8 appeared to be more heterogeneous (Fig. 1c). Starch granules from WT exhibited a regular rounded form and a flat discoid structure that is typical for Arabidopsis leaf starch (Fig. 1a,b; Zeeman et al., 2002). However, the starch granules from sex1-8 plants were deformed, thin and uneven (Fig. 1d). The maximal diameters of WT and sex1-8 starch granules were 1.9 ± 0.5 μm and 6.9 ± 1.4 μm (mean ± SD; n = 250, each), respectively. Due to their complex geometry and uneven surface properties the resulting volumes of the starch granules were afflicted with substantial errors.

Figure 1.

Morphology of starch granules isolated from Arabidopsis wild-type (WT) and sex 1-8 plants. Scanning electron micrographs of native starch granules isolated from WT (a, b) and sex1-8 (c, d) plants. Leaf material was harvested at the end of the light period and starch granules were isolated, lyophilized and imaged by scanning electron microscopy. Bars, 10 μm or 5 μm.

Analysis by TEM confirmed that the size of WT and sex1-8 starch granules varied, and that on average the chloroplasts from sex1-8 plants had twice as many granules as those from WT plants (c. 4 and 8 granules, respectively; number of analysed chloroplasts: WT, 260; sex1-8, 153). Furthermore, cell and chloroplast size differed, as sex1-8 plants had a smaller average mesophyll cell size but bigger chloroplasts (Fig. S2).

Additional information on starch content per leaf volume and starch density was obtained by second harmonic generation measurements of ethanol-decolourized rosettes from 6-wk-old plants, which reflects amylopectin content. The signal intensity per leaf volume was five times higher in sex1-8 than in WT (Fig. S3a). These results are in agreement with previous measurements of average leaf starch content in GWD-deficient Arabidopsis plants (Yu et al., 2001; Ritte et al., 2002; Baunsgaard et al., 2005; Kötting et al., 2005). The density of both starches was determined to be similar (Fig. S3b). Starch granule density was confirmed by the tracer dilution method (Dengate et al., 1978) and estimated to be 1.27 ± 0.02 and 1.28 ± 0.01 g cm−3 (n = 3) for WT and sex1-8, respectively.

Partially complemented GWD transgenic lines exhibit an intermediate phenotype

In order to examine the effect of SEX1 on starch degradation in more detail, we transformed the sex1-8 plants with cDNA-derived constructs for expression of AtGWD under the control of the constitutive CaMV 35S-promotor. Two independent homozygous lines, gwd-c1 and gwd-c2, were established for further analysis. A strong immunosignal of the expected size (160 kDa) for Arabidopsis GWD was detected in soluble protein extracts of WT plants, whereas no signal was observed in sex1-8 plants (Fig. 2a). In gwd-c1 and gwd-c2, the GWD immunosignal was reduced to 80% and 93% compared to WT, respectively. As expected, the GWD knockout mutant sex1-8 has strongly retarded growth, whereas the low-moderate expression of GWD in the gwd-c1 and gwd-c2 lines led to partial complementation of the sex1-8 growth phenotype (Fig. 2b). Thus, GWD expression correlates with growth of the lines. Like sex1-8, both of the partially complemented lines, gwd-c1 and gwd-c2, exhibit a starch excess phenotype. However, gwd-c1 and gwd-c2 had 28% and 15% less starch than sex1-8 plants, respectively (Fig. 2c). The starch content in each line correlates well with the starch-derived glucose 6-phosphate (G6P) content (Fig. 2d). The G6P content in gwd-c1 and gwd-c2 was 30% and 7% of the amounts in WT starch, respectively. In summary, the results indicate that GWD expression level correlates with starch G6P content and this is associated with increased leaf starch content.

Figure 2.

Phenotype of Arabidopsis mutant sex1-8 and partially complemented sex1-8 plants (gwd-c1 and gwd-c2). (a) Detection of glucan, water dikinase (GWD) in crude extracts of wild-type (WT, Col-0), gwd-c1, gwd-c2 and sex1-8 using antibody raised against GWD. Proteins were extracted at the end of the light period. Forty micrograms of protein for each sample were subjected to SDS-PAGE followed by immunodetection. The sizes of the molecular mass markers are depicted in kDa. (b) Growth phenotype of 4-wk-old plants of WT, gwd-c1 and gwd-c2 and sex1-8. (c) Leaf starch content of WT and GWD mutants. Rosettes were harvested at the end of the light period. (d) Glucose 6-phosphate (G6P) content of starches determined after acid hydrolysis. (c, d) Glc, glucose. Values are mean ± SD of measurements made on three biological replicates. Significant differences in (c) and (d) determined by Student's t-test: **, < 0.05; ***, P < 0.01.

The activities and protein amounts of various other starch-related enzymes were determined semi-quantitatively and the results are summarized in Fig. 3. Soluble proteins were extracted from each genotype as indicated and equal protein amounts were separated by native PAGE (zymogram) or SDS-PAGE for immunoblotting (Fig. 3). The results display a reduced activity pattern for soluble starch synthases 1 and 3 (AtSS1, AtSS3) at both time points compared to WT, while the activities of plastidial (AtPHS1) and cytosolic (AtPHS2) phosphorylases remained almost unchanged (Fig. 3). The AtPWD amount was similar in all lines in the light, but was significantly higher in the mutant lines than in WT at the beginning of the night. Irrespective of the photoperiod, the AtSEX4 and AtSS4 protein levels were dramatically increased in all of the mutant lines compared to WT. By contrast, AtSS2 and AtBE2 immunosignals were decreased in the mutants, while the protein amount of AtBE3 was unchanged (Fig. 3). The alterations in protein abundances and enzyme activities in the mutant lines emphasize the intermediate characteristics of the two partially complemented lines gwd-c1 and gwd-c2.

Figure 3.

Starch-related enzymes in the four Arabidopsis lines. Activities (AtSS1, AtSS2, AtPHS1, AtPHS2) and protein amounts (AtGWD, AtPWD, AtSEX4, AtBE2, AtBE3, AtSS2, AtSS4) of starch-related enzymes were determined by activity staining following gel electrophoresis under nondenaturing conditions or by immunoblot analysis following SDS-PAGE, respectively. Band intensities were quantified using AIDA software and normalized to wild-type (WT, Col-0) values from samples harvested after 10 h in the light. Values are the mean ± SD of three independent biological replicates. Asterisks denote significant differences (Student's t-test) from WT 10 h light (*, P < 0.05). Values are also depicted in the form of a heatmap prepared using the heatmap.2 module.

The morphology of starch granules from gwd-c1 and gwd-c2 was examined by SEM. Starch granules from both lines were heterogeneous in size, bigger and more uneven than the WT granules, and similar to those from sex1-8 (Fig. 4; cf. Fig. 1). Furthermore, TEM revealed an increased granule number per chloroplast as reported for sex1-8 (see Fig. S2).

Figure 4.

Morphology of starch granules isolated from partially complemented Arabidopsis sex1-8 plants. Scanning electron micrographs of starches from gwd-c1 (a, b) and gwd-c2 (c, d). Leaf material was harvested at the end of the light phase for starch granule isolation. Bars, 10 μm or 5 μm.

In order to examine the relationship between starch granule size and glucosyl residues, the average glucose content per starch granule was determined for granules from mutant and WT plants. The number of isolated starch granules was estimated using a Thoma counting chamber. An equal number of starch granules from each starch suspension was hydrolysed and the amount of released glucose was determined enzymatically. As shown in Fig. S4, starch granules from sex1-8 contained approximately four to five times more glucose per granule than those from WT plants, while the gwd-c1 and gwd-c2 granules were intermediate (cf. Fig. 2c,d). No alteration in starch density was observed for either of the lines compared to WT and sex1-8 determined by the tracer dilution method (data not shown). Thus, the increased starch granule size observed in sex1-8 and the partially complemented lines correlates with the higher glucose content per starch granule compared to WT. Furthermore, taking into account the same density, a higher glucose content per granule in the mutants, and the observed increase in granule size (SEM), an increase in starch granule volume is expected.

In summary, these results indicate that GWD expression level correlates with the G6P and glucose contents per starch granule, and thereby leads to increased averaged leaf starch content. The partially complemented gwd-c1 and gwd-c2 lines show phenotypes that are intermediate between the sex1-8 and WT plants. Together, this series of plants opens up the possibility of investigating the impact of starch phosphorylation by GWD on starch metabolism with greater resolution.

Altered GWD expression levels influence the surface properties of starch granules

We analysed the chain length distribution profile for each line using heat solubilized and native starch granules. After treatment with isoamylase, the soluble glucans released from the starches were separated and quantified using capillary electrophoresis with laser-induced fluorescence detection (CE-LIF). Solubilized starches are completely debranched by isoamylase. No significant differences in chain length pattern were observed between the lines (Figs S5, S6a–c), which in case of sex1-8 is in agreement with a previous report (Yu et al., 2001).

By contrast, the degree of polymerization (DP) of glucan chains obtained from the surface of the starch granules differed between the various lines (Fig. 5). The starches from sex1-8 and the two partially complemented lines had a higher proportion of short glucan chains (DP 6-8) and fewer long chains (DP 10-17). Within each line, no obvious differences in the pattern of chain length distributions were detected during the diurnal cycle (Figs 5, S6d–f). A closer view revealed that chain length distribution correlates with GWD expression level, as DP 6-8 decreases and DP 10-14 increases with higher GWD expression (Fig. S7). Native starch granules treated with isoamylase were also analysed by SEM. Results indicate no alteration in size and shape between treated and untreated granules (Fig. S8).

Figure 5.

Properties of starch granule surface from Arabidopsis plants with altered glucan, water dikinase (GWD) activity. Chain length distribution pattern of surface near glucan residues of wild-type (Col-0) (a), gwd-c1 (b), gwd-c2 (c) and sex1-8 (d). Native starch granules isolated at different time points of the photoperiod were incubated with isoamylase (white bars, 8 h light; light grey bars, 12 h light; dark grey bars, 2 h dark). The released glucan chains were labelled by APTS and quantified by capillary electrophoresis equipped with laser-induced fluorescence detection (CE-LIF). Bars show mean ± SD of three biological replicates.

Starch granules were incubated with recombinant GWD and [33P]ATP to test if the rate of phosphorylation by GWD was affected by the differences between the mutants and WT in glucan chains at the starch granule surface. As shown in Fig. 6(a), the initial phosphorylation rate of WT starch was at least 30% higher than for starches from the mutants. Within each genotype, no changes in 33P incorporation rate were observed during the diurnal cycle, with the exception of a 30% increase in WT starch at night. Such an increase at night has previously been reported (Ritte et al., 2004). Results also indicate that the rate of starch phosphorylation by GWD depends on the properties of the granule surface rather than the available granule surface area, as initial phosphorylation rates under substrate excess were determined. In previous studies it has been shown that GWD preferentially phosphorylates crystalline structures and is able to bind to soluble glucans (Hejazi et al., 2008).

Figure 6.

In vitro phosphorylation by glucan, water dikinase (GWD) of native Arabidopsis starch granules (Col-0) and granules treated with β-amylases. (a) Native starch granules isolated at different time points within the diurnal cycle (light grey bars, 8 h light; mid grey bars, 12 h light; dark grey bars, 2 h dark) were incubated with recombinant GWD and [β-33P]ATP. The 33P incorporation into starch was determined by scintillation counting. (b) Native starch granules were incubated with AtBAM3 and the released maltose content was quantified by measuring reducing ends. (c) In vitro phosphorylation of starch granules following β-amylolysis. Native starch granules were incubated with AtBAM3. The reaction was stopped, granules were washed and phosphorylated with recombinant GWD and [β-33P]ATP, and 33P incorporation determined. (a–c) Results are mean values ± SD of three independent biological replicates. Asterisks and letters indicate statistically significant differences of values of the respective time point of the diurnal cycle (a, 8 h light; b, 12 h light; c, 2 h dark) as determined by Student's t-test: *, P < 0.1; **, P < 0.05; ***, P < 0.01.

Native granules were treated with AtBAM3 in order to remove free and accessible glucan chains located at the starch granule surface, and the release of maltose was monitored (Fig. 6b). The reduction or loss of GWD in gwd-c1 and sex1-8, respectively, favoured the release of maltose compared to WT starch. The highest rate of maltose release was observed at the end of the light phase of the diurnal cycle. To determine whether starch phosphorylation is affected by the preceding β-amylolysis, 33P incorporation by GWD was determined for starch granules that had been pre-treated with AtBAM3 (Fig. 6c). For starches from all four genotypes, the rate of 33P incorporation was two- to four-fold higher with AtBAM3-treated than nontreated granules (see Fig. 6a,c). Similar results were published by Edner et al. (2007) using native starch granules from WT and sex1-3. Taken together, the increased maltose release and the lower average chain length for starches from the GWD-deficient mutants suggest that there is a higher frequency of free and accessible glucans on the surface of starch granules from the mutants. Thus, these results indicate that the frequency of small glucan chains on the starch granule surface affects phosphorylation by GWD.

It has been shown previously that soluble glucans (maltodextrins DP 6-30) are able to limit glucan phosphorylation by GWD (Hejazi et al., 2009). In light of this observation, we performed two further experiments. First, we incubated WT and sex1-8 starches with isoamylase to release glucan chains from the granule surface. Isoamylase-treated WT and sex1-8 starches were mixed with purified sex1-8 and WT glucan chains, respectively, and then incubated with GWD and [33P]ATP. The rate of 33P incorporation into WT starch that had been pretreated with isoamylase was reduced to 55% when released glucan chains of sex1-8 were added (Fig. 7a). In the parallel assays with sex1-8 starch, there was a smaller decrease (maximum 18%) in 33P incorporation when glucan chains from WT starch were included in the reaction.

Figure 7.

Effect of glucan chain residues located at the surface of Arabidopsis starch granules on glucan, water dikinase (GWD) activity in vitro. (a) Native starch granules isolated at the end of the light phase from wild-type (WT, Col-0) and sex1-8 were debranched by isoamylase. The released soluble glucans were separated by centrifugation. The remaining starch granules of the sex1-8 mutant were mixed with soluble glucans from WT and vice versa. Mixtures were incubated with [β-33P]ATP and recombinant GWD and 33P incorporation into starch was determined. Starch granules treated with isoamylase served as controls. (b) Mixtures of WT and sex1-8 native starch granules were incubated with [β-33P]ATP and recombinant GWD. After washing of starch granules, 33P incorporation was monitored. Different amounts of WT starch were used as controls. (a, b) Bars mean values ± SD of three biological replicates. Asterisks denote values that are significantly different as determined by Student's t-test: *, < 0.1; ***, P < 0.01.

In a second experiment, WT and sex1-8 starches were mixed in ratios of 2 : 1 and 1 : 1, respectively, and incubated with GWD. As controls, WT starch was incubated separately (Fig. 7b). The rate of 33P incorporation was up to 35% lower when sex1-8 starch was included in the reaction mixture, indicating that small glucan chains, even if bound to the granule surface, influence GWD activity.

The activity of GWD correlates with the amount of accessible glucan chains on the starch granule surface

We confirmed our conclusions from the above experiments using potato reserve starch isolated from the variety Désirée. Isolated starch granules were separated using nylon nets of 60-, 30- and 20-μm mesh size. The size distribution, number and volume of granules were analysed using a Coulter counter (data not shown). A fixed amount of each starch population was incubated with AtBAM3 and the amount of maltose was measured. Maltose content per starch granule surface area was calculated using the volumes determined by Coulter counter. The rate of incorporation by GWD was also quantified and the data are expressed in pmol per μm2 of starch granule surface area. A clear correlation between maltose release and 33P incorporation via GWD was observed in the various starch fractions (Fig. 8).

Figure 8.

Correlation between maltose release by β-amylase and 33P incorporation by glucan, water dikinase (GWD). Starch granules isolated from potato tubers were separated by size using nylon nets and an aliquot of each granule population was incubated with recombinant AtBAM3. The release of maltose was determined by reducing end assay. In addition, untreated starch granules were incubated with GWD and [β-33P]ATP and 33P incorporation was quantified. Light grey bars represent released maltose values (mean ± SD, n = 3) and black dots show 33P incorporation rates. Statistically significant differences of released maltose values were determined by Student's t-test: *, P < 0.1; ***, P < 0.01.

GWD action on the starch granule surface affects starch synthesis by AtSS1

We investigated whether differences in the surface properties of starch granules from mutants with altered GWD expression also affect starch synthesizing enzymes. This was done by testing the ability of soluble starch synthase 1 (AtSS1) to incorporate glucosyl residues into starch. Native starch granules from all four genotypes were incubated with 14C-labelled ADPglucose and recombinant AtSS1. Incorporation of 14C-labelled glucosyl residues was determined as shown in Fig. 9(a). The rate of incorporation of glucosyl residues was 30% lower with starch granules from the mutants compared to WT. The accessibility of suitable glucan chains as glucosyl acceptors at the granule surface for AtSS1 was two-fold higher when starch granules were prephosphorylated by GWD (Fig. 9b). Thus, the observed surface properties of starch granules influence both enzymes related to starch degradation and synthesis.

Figure 9.

Incorporation of glucosyl residues into starch granules from Arabidopsis catalysed by soluble starch synthase 1 in vitro. (a) Starch isolated from wild-type (Col-0), gwd-c1, gwd-c2 and sex1-8 plants (8 h light) were incubated with AtSS1 and [glucosyl-14C]ADPglucose. Following incubation and washing of starch granules, incorporated 14C-labelled glucosyl residues were quantified by scintillation counting. (b) Isolated native starch granules (12 h light) were phosphorylated by glucan, water dikinase (GWD) and ATP overnight and then incubated with AtSS1 and [glucosyl-14C]ADPglucose. Data are mean ± SD of three independent technical replicates. Student's t-test was applied to determine statistically significant differences: *, P < 0.1; ***, P < 0.01.


This study was initiated by the observation that starch granule morphology is different in Arabidopsis plants when GWD expression is prevented. Therefore, we generated partially complemented lines in the sex1-8 background, giving intermediate levels of GWD expression. We showed that the amount and phosphorylation level of starch in these Arabidopsis lines correlated with GWD activity and, finally, summarized the reported dependence of C6 phosphorylation on starch degradation (Fig. 2, and Hejazi et al., 2012a). In the partially complemented lines, the activities of various other enzymes involved in starch metabolism were also intermediate between those in wild-type (WT) plants and sex1-8 (Fig. 3). Thus, in plants with decreased GWD activity, we saw changes not only in the amount of starch, but also in the activities of starch-metabolizing enzymes. So far, it is unclear at which level these alterations are perceived and finally implemented in planta. In most cases the protein amounts or activities of starch-synthesizing enzymes were downregulated (AtSS1, AtSS2 and AtSS3; see Fig. 3), but an interesting exception is AtSS4, whose protein abundance was increased in the GWD-deficient plants. Enhanced AtSS4 activity might explain the increased number of starch granules per chloroplast in the GWD-deficient mutants, based on previous observations that point to an involvement of AtSS4 in starch granule initiation (Roldán et al., 2007; Szydlowski et al., 2009; Crumpton-Taylor et al., 2013).

The protein abundances of enzymes involved in the starch phosphorylation cycle at the surface of starch granules – for example, SEX4 and PWD – were also increased in the GWD-deficient lines (Fig. 3). Assuming that their activities increased in parallel with their abundance, this could indicate a compensatory response by the plant to the decrease or absence of C6 phosphorylation. Furthermore, reduced changes in diurnal starch amount, which occur in some other plants with a disrupted phosphorylation/dephosphorylation cycle at the granule surface, for example, the lsf1 mutant (Comparot-Moss et al., 2010), are likely a result of both starch degradation and synthesis.

Like the starch-related phenotypes, the growth phenotypes of the partially complemented lines, gwd-c1 and gwd-c2, were also intermediate between WT and sex1-8. Detailed analysis revealed that the cell and chloroplast size, as well as the number of starch granules per chloroplast, are affected (Fig. S2). Similar increases in chloroplast size (to be accurate the volume) and number of starch granules have been reported for some other Arabidopsis mutants (Crumpton-Taylor et al., 2012).

Analyses of the internal structure of starch granules did not show any significant alterations in chain length distribution, density, thermal stability, crystallinity or allomorph type. Thus, we can conclude that either GWD has no direct effect on the internal structure of the granule or that there are compensatory responses to the decrease or loss of GWD. The latter is more likely. In contrast to the inner structure, we found that there are substantial differences in the surface properties of starch granules in plants with altered GWD activity compared to WT, with both the amount of smaller chains and accessible glucosyl residues on the granule surface being increased in the mutant lines (Figs 5, 6b). Despite these differences, it should be noted that the starch granules from the mutants still do have some longer glucan chains on the granule surface. Interestingly, the observed alterations occur in the presence of starch degrading enzymes. We observed that the increased amount of short glucan chains at the granule surface reduces GWD activity, and that this could be overcome by removal of the glucan chains by β-amylolysis. This observation corresponds with recently published results that GWD prefers highly ordered crystalline structures as substrate (Edner et al., 2007; Hejazi et al., 2008). The positive relationship between increasing glucosyl residues (determined by maltose release) and measurable phosphorylating action of GWD at the granule surface was confirmed in experiments on various potato starch fractions, which showed that GWD activity decreases when β-amylolysis is enhanced (Fig. 8).

The most obvious difference between mutants and WT plants is the morphology of starch granules, as the granules from the former have an uneven shape, are bigger and have a broader size distribution than WT granules. Based on phenotypical and biochemical characteristics, we conclude that the uneven shape of the mutants' starch granules is a result of an uneven distribution of the glucan chain residues at the granule surface. A model based on our data regarding granule morphology is presented in Fig. 10. It explains how the starch granule morphology arises in plants with decreased or no GWD activity. The model recognizes that even in the absence of GWD, there is still some starch phosphorylation, which can be ascribed, for example, to PWD activity (Ritte et al., 2006). The incorporation of phosphate into the starch granules by GWD is enhanced in regions where less accessible glucan chains are located at the granule surface, resulting in a partial degradation at night. Other regions of the starch granule surface are densely populated with (short) glucan chain residues leading to an inhibition of GWD activity and finally to a reduced degradation. Furthermore, during starch synthesis, areas of the granule surface are used differently by enzymes and over several photoperiods the starch granules become bigger and more uneven (Figs 9, 10b). This is also observed using SEM, as small starch granules had a more even appearance than bigger ones (see Figs 1, 4). In addition, incorporation experiments with starch synthase 1, a dominant starch synthase activity in Arabidopsis leaves (Delvallé et al., 2005), revealed the same tendency as observed for GWD, as the activity reduces with a higher number of short chains at the starch granule surface (Fig. 9a). Glucosyl elongation mediated by AtSS1 was two-fold higher when starch granules were phosphorylated by GWD before incubation with AtSS1 (Fig. 9b). The ability of GWD to alter crystalline structure at the starch granule surface favours starch synthase action by generating a more accessible and suitable glucan substrate for chain elongation. Thus, if both processes, (partial) degradation and (partial) synthesis, are involved in morphology formation, alterations of the granule morphology are likely to be more pronounced (and easier to observe).

Figure 10.

Proposed model of starch granule morphology formation in Arabidopsis plants with decreased glucan, water dikinase (GWD) activity. Schematic presentation of starch granule morphology of wild-type (WT) (a) and lines with decreased or no GWD activity (b). In the upper panel, barrels represent double helices in highly ordered, i.e. crystalline regions. Twisted black lines depict less ordered glucan chain residues located at the granule surface. Note that there is an uneven distribution of soluble glucan chain residues, differing in length, at the granule surface. In WT, fewer soluble glucan chain residues are observed compared to mutants with decreased GWD activity. In addition, the chain length distribution of these soluble glucan chain residues is shifted towards shorter DPs in the GWD mutants (Fig. 5). In (b), as a consequence of a more densely populated surface by short glucan chain residues, the solvent exposed crystalline regions are limited. Therefore, as GWD preferentially acts on crystalline structures, its activity is reduced (Fig. 6). By contrast, starch synthase 1, which preferentially acts on soluble glucan chain residues, seems to be more active on longer than on shorter glucan chain residues, and glucan elongation is favoured by the phosphorylating activity of GWD (Figs 5, 6b, 9). In the lower panel, formation of starch granule morphology over several photoperiods is depicted. Note, starch granules are depicted with their maximal diameter in the plane of the diagram. The interplay between phosphorylating enzymes and synthesizing/degrading enzymes is responsible for the homogeneous synthesis and degradation (depicted as large triangles) of starch granules. If phosphorylating activity is reduced, this interplay is restricted to a limited number of granule surface regions (small triangles). This leads to differential turnover in different regions of the granule surface resulting in the observed irregular shape of the granules.

Interestingly, the altered starch granule morphology of GWD mutants has not been observed in other mutants, in which the starch phosphorylation cycle is perturbed, for example, lsf1, sex4 and lsf1/sex4 (Comparot-Moss et al., 2010). Irrespective of whether total granule bound phosphate content was increased or decreased in these other mutants, the granules were rounded in shape and bigger than WT. As in the lines with altered GWD activity described here, the G6P content of starch is also decreased in these other mutants. However, the other mutants accumulate phosphoglucans as an indication of ongoing starch turnover (Comparot-Moss et al., 2010). As the phosphoglucan phosphatases catalyse reactions downstream of GWD, and assuming GWD to be active in sex4, lsf1 and lsf1/sex4, phase transition at the granule surface will still occur in these mutants enabling starch-metabolizing enzymes to act.


Financial support by the Deutsche Forschungsgemeinschaft (DFG-Az. FE 1030/1-1) is gratefully acknowledged. The authors gratefully thank Ms Ingrid Zenke for X-ray diffraction measurements, and Ms Xenia Arngold and Julia Wienke for excellent technical assistance. The authors are indebted to Dr Klaus Gast for help with the light scattering measurements and Dr Frank Jaiser for SEM imaging. The authors thank Dr John Lunn for help during preparation of the manuscript.