An Arabidopsis thaliana gene encoding a homologue of the potato α-glucan, water dikinase GWD, previously known as R1, was identified by screening the Arabidopsis genome and named AtGWD3. The AtGWD3 cDNA was isolated, heterologously expressed and the protein was purified to apparent homogeneity to determine the enzymatic function. In contrast to the potato GWD protein, the AtGWD3 primarily catalysed phosphorylation at the C-3 position of the glucose unit of preferably pre-phosphorylated amylopectin substrate with long side chains. An Arabidopsis mutant, termed Atgwd3, with downregulated expression of the AtGWD3 gene was analysed. In Atgwd3 the amount of leaf starch was constantly higher than wild type during the diurnal cycle. Compared with wild-type leaf starch, the level of C-3 phosphorylation of the glucosyl moiety of starch in this mutant was reduced. Taken together, these data indicate that the C-3 linked phospho-ester in starch plays a so far unnoticed specific role in the degradation of transitory starch.
The only in planta substitution of starch is the phospho-esterification of the C-3 and C-6 hydroxyl groups of the highly branched starch molecules, termed amylopectin (recent review: Blennow et al., 2002). Starch phosphate levels vary considerably with the plant origin. Endosperm starches have very low levels (<0.04% w/w) and tuberous starches tend to have the highest contents of starch phosphate (0.2–0.4% w/w; Blennow et al., 2000). Leaf starch is generally less phosphorylated than tuberous starch (0.1% w/w; Yu et al., 2001).
The enzymatic activity of the glucan, water dikinase (GWD), a starch phosphorylating enzyme from potato, was recently unravelled and studied in detail (Mikkelsen et al., 2004; Ritte et al., 2002). The reaction mechanism is a dikinase-type reaction with three reactants, the glucan substrate, ATP and water. First, the enzyme binds ATP and the gamma phosphate is transferred to water. Secondly, the β-phosphate of ATP is transferred autocatalytically to a conserved histidine residue in the active site of GWD. Finally, the phosphate is transferred from the P-His-GWD intermediate to the glucan substrate (Mikkelsen et al., 2004; Ritte et al., 2002). The dikinase-type reaction is also known from a family of pyruvate phosphorylating enzymes, namely the inter-kingdom-conserved pyruvate phosphate dikinases and the prokaryotic phosphoenolpyruvate synthases (Sakai and Ohta, 1993). A phylogenetic study of the conserved nucleotide binding domain from the various dikinases suggests that the conserved GWD nucleotide binding domains diverged from other dikinases in parallel with the divergence of the plant kingdom from other kingdoms (Mikkelsen et al., 2004).
Leaf starch, in contrast to tuberous and seed starch, accumulates in the plastid during the day and is degraded during the night. This makes it an ideal physiological system for studying the dynamic situation where both synthetic and degradative enzymes are present in the cell at the same time. A battery of enzymes and enzyme isoforms involved in starch synthesis has been described during the last 30 years of starch research. A less clear picture has emerged for transitory starch degradation (recent review: Zeeman et al., 2004). This picture has recently evolved to include the starch phosphorylating enzyme, GWD. By antisense technology in potato it was determined that downregulation of this enzyme impairs starch degradation in leaves and reduces cold sweetening in tubers (Lorberth et al., 1998). In Arabidopsis it was established by map-based cloning that the well-characterized starch excess mutation sex1-1 (Caspar et al., 1991; Trethewey and ap Rees, 1994) mapped to a gene (SEX1, now termed AtGWD1), which encodes a homologue of the potato GWD (Yu et al., 2001). Analysis of the levels of phosphorylated starch in the various sex1 mutant alleles revealed that the allele-specific variation in starch accumulation correlated with a corresponding variation in starch phosphorylation at the C-6 position of the glucosyl residues (Yu et al., 2001). The AtGWD1 gene is located on chromosome 1, whereas the gene for another putative chloroplastic GWD isoform is positioned on chromosome 5 (Yu et al., 2001). The open reading frame (ORF) of this GWD homologue, termed AtGWD3 (Mikkelsen et al., 2004), has a typical chloroplast targeting signal and a conserved carbohydrate binding module sequence situated in the N-terminus. The likely plastidic location of AtGWD3 and its similarity to the known starch phosphorylating enzyme challenged us to study whether it plays a role in starch metabolism. We characterized the enzyme function of this new GWD isoform, and investigated its role in the pathway of Arabidopsis leaf starch degradation.
The three Arabidopsis GWD genes belong to different GWD subgroups
Three genes encoding GWDs exist in the Arabidopsis genome (Mikkelsen et al., 2004; Yu et al., 2001). At1g10760 encodes the chloroplastic SEX1/AtGWD1 protein (Yu et al., 2001). The At4g24450 was predicted by ChloroP to encode a non-chloroplastic protein (AtGWD2), whereas the At5g26570 gene product (AtGWD3) was predicted to be chloroplastic. We constructed a phylogenetic domain tree on the basis of a Basic Local Alignment Search Tool (blast) (Altschul et al., 1990) of both expressed sequence tag (EST) sequences and genomic sequence from higher plants available in GenBank using the putative ORFs of the three AtGWD genes as query sequences. As the number of GWD sequences including N-termini was <10 in the database, we chose to construct a neighbour-joining phylogenetic tree on the basis of the highly conserved C-terminal ATP-binding domain of higher plant GWD sequences. This domain tree shows that the GWD protein family splits up into three different subgroups (Figure 1). One subgroup comprises AtGWD1/SEX1 and several putative orthologues from other higher plants, while the second group was only represented by one Arabidopsis GWD sequence (At4g24450), termed AtGWD2. Notably, ESTs – aligning to other regions of AtGWD2 – were found in other higher plants (M.A. Glaring, unpublished data). The third subgroup comprises the At5g26570 gene product, AtGWD3, and putative orthologues from other higher plants.
Molecular characterization of AtGWD3
An Arabidopsis EST from ABRC (BE527207) was found to harbour the 3′-end of the AtGWD3 cDNA by a blast search of Arabidopsis ESTs using the gene-predicted mRNA sequence as query sequence. This 435 bp EST was used as a probe in a screen of a cDNA library obtained from Arabidopsis leaf mRNA (λC4-16). A total of 50 000 plaque-forming units from the amplified Arabidopsis seedling hypocotyl library were screened using high stringency washes and eight positive plaques were isolated. Three of these clones were in vivo excised as phagemids and sequenced. The sequences all aligned to the At5g26570 gene and the longest insert lacked 509 bp of the 5′-end of the putative ORF. A fragment of 720 bp of the 5′-end of the ORF with SacI and SpeI restriction sites at each end was produced by RT-PCR from leaf mRNA and ligated into the corresponding unique restriction sites in the phagemide harbouring the longest cDNA clone, thereby obtaining a full-length cDNA (GenBank accession number AY747068). The predicted translation product (Figure 2) was 1196 amino acids and predicted to contain a 52 amino acid chloroplastic transit peptide by ChloroP (Emanuelsson et al., 1999). A region close to the N-terminus of the mature protein aligned with the carbon binding module 20 (CBM20) of the CAZy database (Coutinho and Henrissat (1999); http://afmb.cnrs-mrs.fr/CAZY/index.html).
Purification of AtGWD3
AtGWD3 was heterologously expressed in Saccharomyces cerevisiae as a fusion protein with a V5 epitope and polyhistidine tag (6xHis) fused to the C-terminus and then purified to apparent homogeneity. To this end, an affinity-tag chromatographic step proved to be sufficient. The resulting AtGWD3 preparation visibly lacked contaminating proteins as shown by SDS-PAGE and the purification of the fusion protein was verified by immuno blot using anti-V5 antibody (Invitrogen, Taastrup, Denmark). Further confirmation of the identity of the purified protein was achieved by incubating AtGWD3 with [β-33P]ATP before running an SDS-PAGE and visualizing with autoradiography (Figure 3). The purified protein was significantly labelled in this procedure, which clearly demonstrated that AtGWD3 is able to autophosphorylate and thereby perform the first step of the dikinase reaction mechanism (Figure 3). AtGWD3 is able to go through this part of the reaction mechanism in the absence of glucan substrate as was observed with potato GWD (Mikkelsen et al., 2004; Ritte et al., 2002).
AtGWD3 substrate specificity
Leaf starch from mature Arabidopsis rosette leaves from wild-type Col-0 was purified according to Zeeman et al. (1998b) to investigate the physiologically relevant substrates for AtGWD3. The ability of the AtGWD3 protein to transfer 33P from [β-33P]ATP to gelatinized leaf starch was tested according to Mikkelsen et al. (2004), but no GWD activity could be detected above background level (Table 1).
Table 1. Specific activity of AtGWD3 with various glucan substrates
Mean chain length (DP)
Phosphate content (nmol Glc-6-P mg−1 starch)
Specific activity (nmol incorporated P mg−1 protein)
Mean chain length and Glc-6-P contents of substrates were determined from HPAEC analyses. Activity of the AtGWD3 was determined using the dikinase activity assay as described in Experimental procedures. All substrates were gelatinized at 100°C for 5 min. The data represent one of two independent experiments with similar results.
The in vitro substrate specificity of purified AtGWD3 was characterized further using a small series of enzymatically obtained model substrates derived from the waxy maize endosperm amylopectin which has a very low native phosphorylation level (Blennow et al., 2000; Yun and Matheson, 1993). The following substrates were generated: (i) waxy maize amylopectin elongated by phosphorylase a, (ii) waxy maize pre-phosphorylated with potato GWD according to Mikkelsen et al. (2004) and (iii) elongated, pre-phosphorylated waxy maize amylopectin. As regards the contamination of small amounts of phosphoglucomutase in the phosphorylase a extension mixture, the background of free (not glucan bound) glucose-6-phosphate (Glc-6-P) in the elongation mixture was subtracted from the Glc-6-P value determined after glucan hydrolysis. AtGWD3 was incubated with [β-33P]ATP and each respective maize glucan substrate to determine the efficiency of glucan phosphorylation. AtGWD3 was neither able to phosphorylate the native waxy amylopectin, nor the elongated waxy amylopectin which contained a very low amount of phosphate-esters, nor the pre-phosphorylated amylopectin with fivefold more C-6 linked phosphate groups. Only the waxy maize amylopectin, which was both pre-elongated and pre-phosphorylated, was a suitable substrate for AtGWD3 phosphorylation (Table 1). Thereby, AtGWD3 demonstrated much higher substrate specificity than potato GWD as it required an amylopectin substrate with a phosphorylation level above a certain level, as well as a high availability of long chains which has also been reported for potato GWD (Mikkelsen et al., 2004).
AtGWD3 phosphorylates primarily the C-3 position of phospho-α-glucans
Ritte et al. (2002) showed that potato GWD phosphorylates the C-6 and C-3 positions in the starch glucose units in a ratio similar to the one found in native potato amylopectin, which is about 7:3 (Blennow et al., 2000). To clarify the site preference of AtGWD3, a dikinase assay using the optimal glucan substrate (elongated, pre-phosphorylated waxy maize amylopectin) was performed. A high concentration of [β-33P]ATP (2 × 107 Bq) was employed to ensure a final distinct labelling of the Glc-6-P and glucose-3-phosphate (Glc-3-P) peaks. Following incubation and an extensive washing procedure the reaction product was acid-hydrolysed, separated by HPAEC-PAD and the incorporated radioactivity was quantified. In five experiments, the phosphorylation of C-3 was at least 40-fold greater than that of C-6, showing that AtGWD3 specifically phosphorylates the C-3 hydroxyl group of the glucose residue (Figure 4).
AtGWD3 and AtGWD1 transcript levels are co-regulated
Total RNA was isolated at regular intervals during the diurnal cycle. Specific primers for each ORF of AtGWD1 and AtGWD3 were designed (Table 2) and real-time RT-PCR was conducted. The transcript level of the two genes increased throughout the short-day 8 h light period to reach a maximum 1 h after the light period and decreased during the dark period (Figure 5a). These data indicate that the transcription of these two genes is under the same diurnal control. The expression of the two genes was further examined by real-time RT-PCR under different growth conditions of abiotic stress: drought, cold and light-stress, which revealed a co-regulation of the transcription of these genes regardless of the growth conditions (data not shown). A leaf RNA blot with gene-specific probes made with the above-mentioned primer sets affirmed that AtGWD1 and AtGWD3 expression is co-regulated (Figure 5b). The transcript levels of the AtGWD1 and AtGWD3 genes were significantly upregulated in the sex4 mutant which is a starch excess mutant with wild-type levels of starch phosphorylation (Zeeman et al., 1998a; A. Blennow, unpublished data). There were wild-type levels of expression both in the starch-less mutant pgm and in the dbe1-1 mutant producing the highly branched phytoglycogen and very little starch (Caspar et al., 1985; Zeeman et al., 1998b). A reduction of the AtGWD1-transcript level, but no decrease in AtGWD3, was observed in sex1-3, which carries a mutated allele of AtGWD1 and has been shown at the protein level to be a null mutation in this gene (Yu et al., 2001). The blot was washed with high stringency to avoid cross-hybridization between the probes and the other AtGWD isoforms.
Table 2. Primers used for the verification of the Atgwd3 mutant, for RT-PCR experiments and molecular biology applications
AtGWD3 is expressed in all starch-containing plant tissues
To determine whether the AtGWD3 gene is expressed in all starch-synthesizing tissues, we examined the level of AtGWD3 mRNA in various plant tissues isolated at the end of the light period. RT-PCR – using primer AtGWD3-F and AtGWD3-R (Table 2) – showed AtGWD3 expression in both roots, rosette leaves, stems, inflorescence and siliques (Figure 6a).
AtGWD3 is localized in the chloroplast
One explanation for the synchronized presence of multiple isoforms of GWD in the same tissues of Arabidopsis could be that the proteins have different subcellular localizations. Yu et al. (2001) showed AtGWD1 to be primarily plastidic by chloroplast protein immunoblot. Transgenic Arabidopsis lines expressing an AtGWD3-eGFP fusion protein under the control of the 35S promoter were generated as described in Experimental procedures. The subcellular localization of AtGWD3 in rosette leaves was checked by confocal laser scanning microscopy and the AtGWD3-eGFP fusion protein was localized in the chloroplasts. This was clearly visible in guard cells (Figure 6b) and the signal was somewhat weaker in mesophyll chloroplasts (data not shown).
Characterization of Atgwd3
Here, we describe the detailed characterization of the phenotype of a homozygous T-DNA insertion mutant, SALK_110814, found in the SALK collection (Alonso et al., 2003), verified by PCR on genomic DNA to contain an insertion in the AtGWD3 gene and thereafter called Atgwd3. AtGWD3 cDNA-specific primers (Table 2) on each side of the T-DNA insert were used in RT-PCR, which revealed a residual AtGWD3 expression in the Atgwd3 mutant (Figure 7a). Although we observed slower Atgwd3 growth than wild type, the growth difference was not significant under standard growth conditions.
Atgwd3 leaves have elevated levels of starch
The sex1-1 mutant accumulates starch gradually during consecutive photoperiods with a low starch degradation rate during the dark (Zeeman and ap Rees, 1999). To investigate whether AtGWD3 is necessary for starch metabolism, leaves were harvested at different time points during the diurnal cycle, decolourized and then stained with iodine solution (Figure 7b). Leaves of Atgwd3 plants stained with approximately the same intensity as wild-type leaves. However, whereas wild-type leaves were unstained at the end of the dark period – indicating that starch had been degraded –Atgwd3 rosette leaves still stained darkly. The higher-starch phenotype was observed in leaves of young (14 days old) Atgwd3 plants and mature rosettes. Leaves of Atgwd3 plants contained starch even after plants were placed in the dark for periods up to 96 h (data not shown).
To provide more detailed information about the impact of downregulated AtGWD3 we measured the starch and sugar content in wild type and Atgwd3 leaves over a short-day diurnal cycle (Figure 7c). The amount of starch present at the beginning of the day was threefold higher in Atgwd3 than in wild type. During the light period this difference decreased, but at the end of day there was still 15–20% more starch in the Atgwd3 leaves. Thus the mutant synthesized less starch during the light and degraded less starch during the dark, than did the wild type. So the mutant starch pool was oscillating diurnally to a lower extent than that observed in the wild type. The lower degree of starch synthesis in Atgwd3 during the day correlated with a small increase in free sugars compared with wild type, whereas at night the levels were similar (Figure 7c).
Other enzymes of starch metabolism are unaffected in Atgwd3
As the starch synthesis and degradation rates were affected in Atgwd3, we measured the activities of other enzymes involved in starch metabolism in Atgwd3 and wild type (Table 3). The activities of the tested enzymes did not differ significantly. Native PAGE of four representative plants of Atgwd3 and wild type revealed no changes in the isoform pattern of starch hydrolysing enzymes or starch phosphorylytic enzymes (data not shown).
Table 3. Activities of starch-metabolizing enzymes in wild type and Atgwd3 plants
(nmol min−1 g−1 fresh weight) Wild type
Measurements are the mean ± SE of values from four samples, each from an individual plant.
30.0 ± 2.8
25.0 ± 3.0
1904 ± 194
1979 ± 274
Soluble starch synthase
45.2 ± 1.9
47.6 ± 3.5
218 ± 10
234 ± 19
Amylopectin structure is not affected by downregulation of AtGWD3
The indication from our in vitro data that AtGWD3 activity could depend on both the phosphorylation status and the chain length distribution of the amylopectin substrate, led us to analyse the amylopectin structure of starch from leaves of wild type and Atgwd3 plants. Starch from leaves at the end of the photoperiod was debranched using isoamylase and subjected to HPAEC-PAD. The chain length distributions of the mutant and wild-type leaf starch at the end of the day (8 h light) were similar (Figure 8).
A decrease in starch-linked Glc-3-P in Atgwd3
To examine the difference in C-6 and C-3 hydroxyl group phosphorylation between Atgwd3 and wild-type starch, we used the same isolated starch as above in acid hydrolysis and subjected the hydrolysate to HPAEC-PAD on a PA-1 column (Dionex, Rødovre, Denmark) according to Blennow et al. (1998). The concentration of Glc-6-P and Glc-3-P linked to starch was 5.2 ± 0.8 nmol mg−1 starch and 0.20 ± 0.02 nmol mg−1 starch for wild type and 5.8 ± 0.4 nmol mg−1 starch and 0.13 ± 0.01 nmol mg−1 starch for Atgwd3 respectively. Thus, a two-fold decrease in C-3 positioned phospho-esters was observed in the Atgwd3 starch compared with wild-type starch.
In vivo starch phosphorylation results in phospho-esters linked to both the C-6 and C-3 positions of glucosyl moieties (Blennow et al., 2000; Lim et al. 1994). Ritte et al. (2002) showed the potato GWD to catalyse the phosphorylation of both the C-6 and the C-3 hydroxyl groups in vitro with approximately the same ratio as measured for potato tuber amylopectin. Our in vitro results establish that AtGWD3 is able to phosphorylate pre-phosphorylated glucans specifically at the C-3 hydroxyl of the starch glucosyl unit. This observation is supported by our measurements of a reduced amount of Glc-3-P linked to transitory starch from Arabidopsis with downregulated levels of AtGWD3 but normal protein levels of AtGWD1; the latter was examined by immunoblot analysis using anti-R1 antibody (data not shown). Yu et al. (2001) failed to detect Glc-3-P linked to starch in sex1-3, but we found that AtGWD3 is expressed in wild-type levels in that mutant. In summary, this indicates that starch phosphorylation activity of AtGWD1 could be a prerequisite for AtGWD3 activity.
Iodine staining revealed a significantly higher level of starch in Atgwd3 seedlings compared with wild type, and this difference proceeded with age (data not shown). This altered carbohydrate metabolism in Atgwd3 was not reflected in significant changes of growth rate or plant morphology as is evident with the sex1 mutants. In leaves of Atgwd3 the rates of both starch accumulation and degradation were slightly reduced compared with wild type, but the minimum level of starch was elevated leaving the starch content in Atgwd3 at a constantly higher level during the diurnal cycle. The starch level of Atgwd3 was lower than in sex1 leaves where up to five times the starch level of wild type have been measured (Yu et al., 2001). Ritte et al. (2004) showed that potato GWD phosphorylates starch most actively in the dark, suggesting a dual role for GWD. The role of phosphorylation during starch biosynthesis is unknown so far, but phosphorylation during the dark appears to be necessary for starch degradation to proceed. We propose from our study that during starch synthesis and/or degradation in Arabidopsis leaves AtGWD3 phosphorylates the C-3 position of starch glucose units, and that this activity is involved in the starch–phosphate connected regulation of starch degradation established by Yu et al. (2001). The necessity for two sequential phosphorylating activities suggests that the C-3, and not the C-6 phospho-ester, linkage might be the starch modification which triggers the initial attack of starch-degrading enzymes. This functional specificity could relate to differences in starch structure exerted by a phosphate group at either the C-3 or C-6 position. Molecular models of a double-helical starch motif with phosphate groups indicate a difference in the way C-3 linked respectively C-6 linked phosphate protrudes from the double helix (Blennow et al., 2002; Engelsen et al., 2003; S.B. Engelsen, The Royal Veterinary and Agricultural University, Copenhagen, Denmark, unpublished data). This might affect double-helical interactions and glucan packing and finally how the outer layer of the starch granule is presented to starch-degrading enzymes.
Independently of plant growth conditions or genetic background, we observed a tight, parallel transcriptional control of the AtGWD1 and AtGWD3 transcripts, which suggests the possibility of a heteromeric enzyme complex between these two isoforms. The presence of high molecular mass enzymatic complexes is known from other starch metabolic enzyme activities in plants. The existence of a multimeric enzyme was recently proved by coprecipitation of different isoforms of isoamylases in potato (Hussain et al., 2003). Furthermore, in wheat, a multiprotein complex of starch branching enzyme isoforms and phosphorylase, dependent on protein phosphorylation, was recently discovered (Tetlow et al., 2004). Evidence for a similar multifunctional heteromeric complex of GWD isoforms, which might also include starch-degrading enzymes, awaits further investigations. In vitro AtGWD3 activity studies indicate that there is no direct requirement of a protein–protein interaction between the two GWD isoforms, thus the phospho-ester linkage generated by AtGWD1 could in itself be a recognition tag for AtGWD3 to bind to the phospho-glucan. Alternatively, the phosphate group could generate the substrate conformation, which is necessary to interact with the active site of AtGWD3. The latter possibility implies the likely output of doubly phosphorylated glucan chains, which have also been detected in recent in vivo studies of starch phosphorylation patterns of the outer layer of potato leaf granules during degradation (Ritte et al., 2004).
No temporal, conditional or spatial differences were observed between the expression of AtGWD1 and AtGWD3. However, evolution has preserved the starch phosphorylating activity of two GWD isoforms and this suggests the substrate selectivity of each isoform to be of major importance to the plant. We propose a model, where initially AtGWD1 phosphorylates starch primarily at the C-6 position of the glucosyl residues, and, subsequently, AtGWD3 phosphorylates the C-3 hydroxyl group. This modification might refine the starch structure for amylolytic enzymes so as to degrade the leaf starch granule in the dark.
Plant material and growth conditions
The mutant Atgwd3 of Arabidopsis thaliana ecotype Columbia was obtained from the SALK T-DNA mutant population (Alonso et al., 2003) provided by the Nottingham Arabidopsis Stock Centre (NASC, http://nasc.nott.ac.uk/home.html) and homozygous lines were selected by PCR. sex1-3, sex4, dbe1-1 and pgm mutant lines were kindly donated by Sam Zeeman and Alison Smith. Arabidopsis thaliana ecotype Columbia and mutants were grown from seed in growth chambers at 20°C at 70% humidity with an 8-h photoperiod at a photon flux density of 120 μmol photons m−2 sec−1.
Sequence manipulations and phylogenetic analyses
Sequence searches of the non-redundant and unfinished genome or EST databases at NCBI were conducted using the tblastn program with the entire AtGWD1, AtGWD2 and AtGWD3 ORFs as query sequences. From each selected protein sequence, we used exclusively the sequence aligning to the amino acid number 1006-1194 of AtGWD3 (the nucleotide binding domain) for construction of the unrooted tree. The ClustalX package (Thompson et al., 1997) was used to create an alignment of the sequences that was then submitted to a neighbour-joining analysis to generate a branching pattern. Bootstrap analysis with 1000 replicates was performed to assess the statistical reliability of the tree topology (data not shown). The phylogenetic tree was displayed using the treeview program (Page, 1996).
Isolation of cDNA encoding AtGWD3
EST BE527207 – harbouring 800 bp from the 3′-end of the AtGWD3 ORF – was obtained from NASC (NASC on-line catalogue). This EST provided a probe for screening an Arabidopsis 3-day-old seedling hypocotyl cDNA library (CD4-16 acquired from ABRC, originally donated by Joe Ecker) using high-stringency washes (0.5 × SSC, 0.1% SDS at 68°C). The screen resulted in eight positive clones, and the longest clone was sequenced and shown to lack 500 bp in the 5′end of the ORF. ORF-specific primers (GWD3Nt- F and GWD3Nt- R, Table 2) bordering the lacking fragment including SacI and SpeI sites, respectively, were designed according to the genomic sequence for fusion with the cloned 3′-part of the ORF. The 5′-end cDNA fragment was generated by RT-PCR in triplicate using a proofreading DNA polymerase (Pwo; Roche, Copenhagen, Denmark) from Arabidopsis leaf mRNA (purified using MicroPoly(A)pure; Ambion, Huntingdon, UK) and sequenced in triplicate to assign the correct sequence. Finally, the two partial AtGWD3-cDNA fragments were fused by ligation to obtain a full-length AtGWD3-cDNA which was verified by sequencing.
Expression and purification of AtGWD3
The plasmid for AtGWD3 expression in S. cerevisiae YWO 0046 (MATα ura3, leu2-3,112, his, pra1-1, prb1-1, prc1-1, cps1-3; Heinemeyer et al., 1991) was constructed as in-frame fusions of the mature dikinase protein to the V5 epitope and 6xHis-tag in the vector pYES2.1/V5-His-TOPO (Invitrogen). The DNA fragment encoding the mature protein without the predicted transit peptide was generated with A-overhang by PCR using specific primers (GWD3pYES-F and GWD3pYES-R; Table 2) by Easy-A Polymerase (Stratagene, A.H. Diagnostics, Århus, Denmark) and inserted into the vector pYES2.1/V5-His-TOPO by a topoisomerase reaction according to the manufacturer's protocol (Invitrogen). The resulting plasmid was transformed into S. cerevisiae by the Li-Ac transformation method (Ito et al., 1983). Transformed cells were inoculated in 10 ml YPD (1% yeast extract, 2% peptone, 2%d-glucose) and incubated overnight at 30°C with agitation at 220 rpm. A 2.5-ml aliquot was transferred into 250 ml medium with 2% (w/v) glucose plus 0.7% yeast nitrogen base without amino acids (Invitrogen) supplemented with leucine (0.01% w/v), and l-histidine (0.005% w/v), and incubated overnight at 30°C with shaking at 220 rpm. Cells were collected by centrifugation at 1500 g for 5 min at 4°C and resolubilized in 2% (w/v) galactose plus 0.7% yeast nitrogen base without amino acids (Invitrogen) supplemented with leucine (0.01% w/v), and l-histidine (0.005% w/v) and incubated for 20 h at 30°C with shaking at 220 rpm. Cells were collected by centrifugation at 1500 g for 5 min at 4°C and washed in water and frozen as pellets in liquid N2 to be stored at −80°C. Cells were resolubilized and washed in protein extraction buffer [50 mm Tris–HCl pH 7.5; 5% glycerol; 1 mm PMSF; 10 μg ml−1 Leupeptin; 10 μg ml−1 Antipain, 2 mm Benzamidin; 2 mm DTT; 1× Complete Protease Inhibitor Cocktail (Roche)]. Cells were lysed and extracted by 6 × 30 sec vortexing with an equal volume of glass beads (0.4–0.6 mm size, Sigma, Brøndby, Denmark). Cell debris were removed by centrifugation at 10 000 g for 10 min at 4°C. The soluble protein fraction was loaded onto a HisTrap HP column (Amersham Biosciences, Hillerød, Denmark) equilibrated with buffer A (20 mm Na3PO4/0.5 m NaCl pH 7.4/40 mm imidazole). Following sample application, the column was washed with 50 ml buffer A and, subsequently, with 7 ml buffer A supplemented with 0–500 mm imidazole at a flow rate of 0.5 ml min−1. The fractions containing AtGWD3-V5-6xHis were combined. The buffer in the AtGWD3-V5-6xHis preparation was changed by passage through a HiTrap desalting column (Amersham Biosciences) equilibrated with 50 mm Tris/HCl pH 7.5, 3 mm EDTA, 10% (v/v) glycerol, 5 mm DTT, 2 mm Benzamidin, 10 μg ml−1 Antipain, 10 μg ml−1 Leupeptin, 1 mm PMSF at 8 ml min−1. The desalted fractions containing AtGWD3-V5-6xHis were concentrated through an Amicon Ultra Centrifugal Filter Device (Millipore, Glostrup, Denmark). Aliquots of the protein preparation were stored at −80°C.
Protein gel electrophoresis and immunoblot analysis
Protein was quantified according to Bradford (1976), using BSA as a standard. SDS-PAGE and immuno blot analysis with Anti-V5 antibody (Invitrogen) was performed according to the manufacturer's protocol. Native, amylopectin-containing PAGE was performed and incubated for hydrolytic and phosphorylytic activities according to Zeeman et al. (1998b).
waxy maize amylopectin, kindly provided by AKV-Cerestar (Vodskov, Denmark), was elongated according to the protocol used for elongation of potato amylopectin (Mikkelsen et al., 2004). Phosphorylation of both native and elongated waxy maize amylopectin was performed by adding 100 μg partly purified potato GWD/R1 to 5 mg ml−1 gelatinized, amylopectin in 50 mm Hepes pH 7.0, incubated for 16 h at 30°C with agitation and terminated by boiling for 2 min. The phosphorylated polyglucan was precipitated with 75% (v/v) ethanol/1% (w/v) KCl. Following centrifugation, the supernatant was discarded, and the pellet was resuspended in 1 ml water, mixed briefly and precipitated as before. This procedure was repeated four times and the final pellet was dried under vacuum.
Dikinase activity assay
Activity assay for GWD activity was performed as previously described (Mikkelsen et al., 2004). One microgram AtGWD3 protein was used in incubation with 0.2 μm [β-33P]ATP and 5 mg ml−1 glucan substrate for 1 h at 30°C, if not indicated otherwise. The incorporation of labelled 33P was determined by adding 3 ml scintillation liquid to the mixture. Radioactivity in the samples was determined with a liquid scintillation counter (Microbeta, Wallac, Perkin-Elmer, Hvidovre, Denmark).
Gene expression studies
The reaction for real-time RT-PCR was performed on the ICycler Instrument (Bio-Rad, Herlev, Denmark) with the iQ SYBR Green Supermix kit for PCR (Bio-Rad) according to the manufacturer's protocol. Each reaction was performed with 5 μl of 1:10 to 1:100 dilutions of the first-strand cDNA in a total volume of 20 μl. The reactions were incubated at 96°C for 2 min to activate the hot start recombinant Taq DNA polymerase, followed by 50 cycles of 30 sec at 96°C, 30 sec at 60°C and 1 min at 72°C. Gene-specific primers (GWD1-F, GWD1-R, GWD3-F, GWD3-R, ACTIN-F and ACTIN-R, listed in Table 2) were used. The specificity of the PCR amplification was checked with a heat dissociation protocol (from 65 to 95°C) following the final cycle of the PCR. The efficiency of the primer sets was calculated by performing real-time PCR on several dilutions of first strands. The efficiency of the different primer sets was almost similar. The results obtained for the different conditions analysed were standardized to the ACTIN1 RT-PCR product level.
For RNA gel blot analysis 20 μg of total RNA was blotted onto nitrocellulose filters, as described by Sambrook et al. (1989). AtGWD1, AtGWD3 and ACTIN1 cDNA were employed to make gene-specific probes by PCR using the same primers as used in real time RT-PCR.
The entire ORF of GWD3 was fused to enhanced GFP behind the constitutive 35S promoter in the binary vector pK7FWG2 (Karimi et al., 2002) using the GATEWAYTM cloning technology (Invitrogen). This construct was transformed into Agrobacterium and introduced into A. thaliana ecotype Columbia by the floral dip method (Clough and Bent, 1998). Transformants were selected on kanamycin and transferred to soil before analysis. Leaf pieces were cut from mature plants and mounted on glass microscope slides in distilled water under a glass cover slip. A confocal laser scanning microscope (TCS SP2; Leica Microsystems, Wetzlar, Germany) equipped with a 40× plan apo/1.25–0.75 oil objective was used for the detection of fluorescence. A 488-nm laser line was used for excitation and emission was detected between 510 and 535 nm for GFP fluorescence, and between 650 and 750 nm for chlorophyll autofluorescence.
Identification of Atgwd3
DNA prepared from the SALK_110814 line was subjected to PCR using the primers GWD3-RP and GWD3-LP (see Figure 7a; for primer sequences see Table 2). PCR amplification conditions were: 94°C for 5 min; 30 cycles of 94°C for 30 sec, 50°C for 45 sec and 72°C for 2 min. The obtained PCR products were subcloned and sequenced. To check for AtGWD3 expression, mature leaves were cut from intact Atgwd3 plants after 8 h light and were immediately frozen in liquid nitrogen and stored at −80°C until use. For RT-PCR, total RNA was extracted from 100 mg frozen tissue using the RNeasy Plant Kit (Qiagen, Venlo, The Netherlands), and 2.5 μg RNA was reverse-transcribed using the ThermoScript protocol (Invitrogen). RT-PCR (using cDNA usually corresponding to 250 ng of total RNA per 50 μl reaction) was carried out for 40 cycles. Primers GWD3-RP and AtGWD3-LP used for amplification are listed in Table 2.
Enzyme measurements and native PAGE
α-Amylase and β-amylase activities in leaf extracts were determined using assay kits from Megazyme (Bray, Ireland). Soluble starch synthase activity in leaf extracts was determined according to Zeeman et al. (1998b). AGPase activity in leaf extracts was determined according to Sowokinos (1976). Native PAGE on amylopectin-containing gels and incubations to visualize hydrolytic and phosphorylytic activities was performed according to Hill et al. (1996).
Analysis of amylopectin structure
The in vitro synthesized polyglucans were analysed as described in Mikkelsen et al. (2004). Native Arabidopsis leaf starch was isolated from mature rosette leaves by the method of Ritte et al. (2000). Leaf starch (5 mg ml−1) in 50 mm sodium acetate was debranched by isoamylase (Megazyme) and analysed by high-performance anion exchange chromatography with pulsed amperometric detection (Dionex) as described by Blennow et al. (1998). Starch and free sugar contents were determined by enzymatic analysis as described by Nielsen et al. (1991).
Determination of glucan-bound Glc-6-P and Glc-3-P
Glc-6-P and Glc-3-P contents were determined by acid hydrolysis and high-performance anion exchange chromatography with pulsed amperometric detection (Dionex) as described (Blennow et al., 1998).
We thank Lis B. Møller, Per Lassen Nielsen and Ann Jensen for excellent technical assistance and Tom H. Nielsen for valuable advice. We thank Dieter H. Wolf for his kind gift of the YWO yeast strain and Sam Zeeman and Alison Smith for Arabidopsis mutants. We gratefully acknowledge the donation of EST BE527207 by the Arabidopsis Biological Resource Center, the Salk Institute Genomic Analysis Laboratory for providing the sequence-indexed Arabidopsis T-DNA insertion mutant and the donation of seed by the Nottingham Arabidopsis Stock Centre. This project was supported by The National STVF frame program Exploring the Biosynthetic Potential of Potato, The Danish National Research Foundation, The Danish Biotechnology Program, the Danish Directorate for Development (Centre for Development of Improved Food Starches) and The Committee for Research and Development of the Öresund Region (Öforsk).