Mutagenesis of cysteine 81 prevents dimerization of the APS1 subunit of ADP-glucose pyrophosphorylase and alters diurnal starch turnover in Arabidopsis thaliana leaves


(fax +49 331 567 8408; e-mail


Many plants, including Arabidopsis thaliana, retain a substantial portion of their photosynthate in leaves in the form of starch, which is remobilized to support metabolism and growth at night. ADP-glucose pyrophosphorylase (AGPase) catalyses the first committed step in the pathway of starch synthesis, the production of ADP-glucose. The enzyme is redox-activated in the light and in response to sucrose accumulation, via reversible breakage of an intermolecular cysteine bridge between the two small (APS1) subunits. The biological function of this regulatory mechanism was investigated by complementing an aps1 null mutant (adg1) with a series of constructs containing a full-length APS1 gene encoding either the wild-type APS1 protein or mutated forms in which one of the five cysteine residues was replaced by serine. Substitution of Cys81 by serine prevented APS1 dimerization, whereas mutation of the other cysteines had no effect. Thus, Cys81 is both necessary and sufficient for dimerization of APS1. Compared to control plants, the adg1/APS1C81S lines had higher levels of ADP-glucose and maltose, and either increased rates of starch synthesis or a starch-excess phenotype, depending on the daylength. APS1 protein levels were five- to tenfold lower in adg1/APS1C81S lines than in control plants. These results show that redox modulation of AGPase contributes to the diurnal regulation of starch turnover, with inappropriate regulation of the enzyme having an unexpected impact on starch breakdown, and that Cys81 may play an important role in the regulation of AGPase turnover.


The daily alternation of light and dark leads to large recurrent changes in the plant carbon budget. Plants buffer these diurnal changes by retaining part of their photosynthate in the leaves, often as starch, and remobilizing it at night to support respiration and export of carbon to sink organs (Geiger and Servaites, 1994; Smith et al., 2005; Smith and Stitt, 2007; Stitt et al., 2007; Zeeman et al., 2007). Starch turnover is regulated such that, over a wide range of conditions, starch reserves are almost fully remobilized but not quite exhausted at dawn (Stitt et al., 1978; Chatterton and Silvius, 1979, 1980, 1981; Mullen and Koller, 1988; Lorenzen and Ewing, 1992; Gibon et al., 2002). This maximizes investment in growth, while avoiding carbon starvation (Smith and Stitt, 2007; Stitt et al., 2007; Graf and Smith, 2011). Achieving this balance requires precise regulation of the rates of starch synthesis and degradation. In addition, inhibition of starch synthesis at night is required to avoid a futile cycle of degradation and resynthesis.

The first committed step in the pathway of starch biosynthesis involves conversion of ATP and glucose-1-phosphate (Glc1P) to adenosine-5′-diphosphoglucose (ADPGlc) and inorganic pyrophosphate (PPi), catalysed by ADP-glucose pyrophosphorylase (AGPase; EC; Preiss, 1988). In leaves, AGPase is exclusively localized to the chloroplast. Although the reaction is freely reversible in vitro, hydrolysis of PPi by plastidial alkaline pyrophosphatase renders it essentially irreversible in vivo (Weiner et al., 1987; Tiessen et al., 2002). Higher-plant AGPase is a heterotetramer (α2β2) consisting of two small subunits (approximately 50 kDa) and two large subunits (approximately 51 kDa; Morell et al., 1987; Okita et al., 1990; Ballicora et al., 2004). The small subunit (APS, also known as ADG1 or BRITTLE2) and large subunit (APL, also known as ADG2 or SHRUNKEN2) are closely related, having evolved from a common ancestral form. Both subunits influence the kinetic and regulatory properties of the AGPase holoenzyme (Cross et al., 2004; Ventriglia et al., 2008). The A. thaliana genome contains four large-subunit genes (APL1, At5g19220; APL2, At1g27680; APL3, At4g39210; APL4, At2g21590) and two small-subunit genes (APS1, At5g48300; APS2, At1g05610). Expression of the APS2 gene is barely detectable. Furthermore, the APS2 protein contains substitutions of several active-site residues, and had no detectable catalytic activity when heterologously expressed (Crevillén et al., 2003, 2005). Thus, APS1 is thought to encode the only catalytically active small subunit. Transcript and proteomic analyses have shown that APL1 is the predominant large-subunit isoform in leaves, whereas APL3 and APL4 are mostly expressed in sink tissues (Crevillén et al., 2003, 2005).

Starch synthesis is severely compromised in AGPase-deficient mutants that have lost a predominant isoform of either the small or large subunits, e.g. the adg1 and adg2 mutants in A. thaliana, the brittle2 and shrunken2 mutants in maize (Zea mays), and the rb mutant in pea (Pisum sativum; Tsai and Nelson, 1966; Dickinson and Preiss, 1969; Lin et al., 1988a,b; Smith et al., 1989; Wang et al., 1997, 1998). This provides strong genetic evidence that AGPase activity is essential for starch synthesis. Flux control analyses using heterozygotes of adg1 mutants (Lin et al., 1988a,b) have shown that AGPase exerts considerable control over the rate of starch synthesis in A. thaliana leaves (Neuhaus and Stitt, 1990; Sun et al., 1999; Stitt et al., 2010).

Higher-plant AGPase is regulated at several levels. The enzyme from leaves displays sigmoidal substrate kinetics for ATP and Glc1P, and is allosterically activated by 3-phosphoglycerate (3PGA) and inhibited by orthophosphate (Pi) (Sowokinos, 1981; Sowokinos and Preiss, 1982; Preiss, 1988). These properties mean that formation of ADPGlc is highly sensitive to changes in the availability of photoassimilate. They are thought to promote starch synthesis and recycling of phosphate when the rate of photosynthesis exceeds the rate of synthesis of other end products, such as sucrose (MacRae and Lunn, 2006). Substrate availability and allosteric regulation of AGPase may also contribute to re-direction of carbon into starch when sucrose accumulates in the leaf. In heterotrophic tissues, the kinetic properties of AGPase, including its sensitivity to regulation by 3PGA and Pi, may differ from those of the leaf enzyme (Ballicora et al., 2004; Boehlein et al., 2010). Crevillén et al. (2003) showed that the substrate affinities and allosteric properties of Arabidopsis AGPase are largely determined by which of the four isoforms of the large subunit is present in the holoenzyme, thus accounting for the different properties of the enzyme between leaves and sink tissues. Expression of allosterically up-regulated forms of AGPase led to enhanced rates of starch synthesis and turnover in A. thaliana (Obana et al., 2006).

AGPase is also subject to post-translational redox modulation. Studies on potato (Solanum tuberosum) AGPase, heterologously expressed in Escherichia coli, revealed that the heterotetramer contains an intermolecular disulphide bridge, linking the two small subunits via their Cys12 residues. The enzyme is activated when the disulfide bridge is reduced using dithiothreitol or reduced thioredoxin (Ballicora et al., 1998, 1999, 2000; Fu et al., 1998). The reduced (dithiol) form of the enzyme is less heat-stable than the oxidized (disulfide) form, being rapidly inactivated at temperatures above 40°C, as were mutated forms of the enzyme in which the Cys12 residue had been substituted by alanine or serine (Ballicora et al., 1999). The reduced APS monomers (50 kDa) are readily separated from the oxidized dimers (100 kDa) by non-reducing SDS–PAGE and visualized by immunoblotting to determine the redox status of the enzyme (Tiessen et al., 2002). In maize endosperm AGPase, the large subunits dimerize via formation of a disulfide bridge but the small subunits do not form dimers in this way (Lyerly Linebarger et al., 2005), presumably due to the absence of Cys residues in the N-terminal domain (Figure S1).

The potato tuber AGPase is redox-activated in response to higher sugar levels, providing a mechanism to coordinate the rate of starch synthesis in the tubers with the supply of photoassimilate from the leaves (Tiessen et al., 2002). In A. thaliana, pea and potato leaves, redox activation of AGPase increases upon illumination or in response to sugar accumulation (Hendriks et al., 2003; Gibon et al., 2004b; Lunn et al., 2006). However, it is not known which of the cysteine residues in the APS1 protein is responsible for light- and sucrose-induced changes in the redox status of the enzyme in vivo. Light-dependent activation of AGPase presumably depends on ferredoxin-dependent reduction of thioredoxins via ferredoxin–thioredoxin reductase, as is known to occur for enzymes involved in photosynthesis. The dual-function NADPH-dependent thioredoxin reductase NTRC has been implicated in sugar-induced redox activation of AGPase (Michalska et al., 2009). It has also been proposed that trehalose-6-phosphate is an intermediary in the sugar-induced redox activation of AGPase, both in leaves and non-photosynthetic tissues (Kolbe et al., 2005). In support of this proposal, the redox activation state of AGPase was found to be correlated with the levels of both sucrose and trehalose-6-phosphate in A. thaliana leaves and seedlings (Lunn et al., 2006). However, many of the molecular details of the redox regulation of AGPase and the role of trehalose-6-phosphate in vivo are unresolved.

Light and sugars lead to transcriptional activation of AGPase in leaves (Sokolov et al., 1998; Bläsing et al., 2005; Crevillén et al., 2005), and this is counteracted by nitrate and phosphate (Scheible et al., 1997; Nielsen et al., 1998). This may allow the plant to adjust its accumulation of starch in response to sustained changes in its carbon and nutritional status. Protein turnover may also contribute to the regulation of AGPase activitiy. Maximum AGPase activity and AGPase protein show marked changes during diurnal cycles, which sometimes occur independently of the changes in transcript levels (Gibon et al., 2004a,b, 2009), and modelling indicates that AGPase protein is rapidly turned over (Piques et al., 2009).

Thus, AGPase is subject to multiple levels of transcriptional and post-translational regulation. In leaves, these form part of a wider regulatory network that coordinates end-product synthesis with the rate of CO2 assimilation, and controls the partitioning of photoassimilates between sucrose and starch, balancing immediate export for growth of sink organs with the need to retain enough reserves in the leaf to last through the night. These multiple types of regulation provide considerable flexibility and robustness, but make it difficult to assess the importance of any single regulatory mechanism for the control of starch synthesis. The aims of the present experiments were to identify which of the five cysteine residues in the A. thaliana APS1 protein are involved in formation of the disulphide in vivo, and to investigate the consequences of loss of redox regulation of AGPase activity in this species. To do this, we replaced the native APS1 protein with various modified forms in which individual cysteine residues were replaced by serine. Mutant lines in which formation of the disulphide bridge was prevented by substitution of the crucial cysteine residue were analysed in detail to determine how loss of redox regulation of AGPase affected the diurnal pattern of starch synthesis and degradation.


Site-directed mutagenesis of cysteine residues in A. thaliana APS1 and expression in the adg1 background

Studies of the potato tuber AGPase, heterologously expressed in E. coli, indicated that a disulphide bridge formed between the Cys12 residues of the two small subunits within the heterotetrameric holoenzyme (Fu et al., 1998). Although this suggests that Cys12 is the residue most likely to be involved in redox regulation of the enzyme in planta, we cannot exclude the possibility that other cysteine residues, alone or in combination with Cys12, may also be involved. The A. thaliana small subunit (APS1) contains five cysteine residues (Figure S1). Cys131 is located in the nucleotidyl transferase domain and is conserved across plants and bacteria. Cys81 (equivalent to Cys12 in the mature potato small subunit protein after cleavage of the 70 amino acid plastid transit peptide), Cys406, Cys411 and Cys423 are mostly conserved in higher plants and oxygenic photosynthetic bacteria (cyanobacteria), but are absent in facultative anaerobic bacteria (enterobacteria).

The APS1 gene (At5g48300) was amplified by PCR from genomic DNA extracted from A. thaliana leaves. The 3085 bp amplicon contained the entire intergenic region upstream of the APS1 translation initiation codon, up to and including part of the 3′ UTR of the adjacent upstream gene (At5g48290), the APS1 protein coding region, and 280 bp downstream of the translation stop codon, including the 3′ UTR and polyadenylation site of the pre-mRNA (Figure S2). Site-directed mutagenesis was performed to replace individual cysteine residues in the encoded APS1 protein with serine. The native (APS1WT) and mutated APS1 genes (APS1C81S, APS1C131S, APS1C406S, APS1C411S and APS1C423S) were cloned into the pGreen binary vector and introduced into the starch-deficient adg1 mutant, which is a recessive EMS mutant with a single base change in the APS1 coding region (resulting in a Gly92→Arg92 substitution; G92R) and has no measurable AGPase activity in the leaf (Lin et al., 1988a; Figure 1a). Prior to transformation, all constructs were sequenced to confirm that no mutations other than the desired site-directed change had been introduced. After initial selection for resistance to phosphinothricin, the primary transformants were screened by harvesting leaves at the end of the day for iodine staining to detect the presence of starch, which would indicate complementation of the adg1 mutant. Wild-type Col-0 plants and the parental adg1 mutant were used as positive and negative controls, respectively. For each construct, 30–35 primary transformants were identified that showed restoration of starch synthesis, indicating the presence of a functional APS1 transgene (see Figure 1b).

Figure 1.

 Cys81 is required for formation of the intermolecular disulfide bridge in APS1.
(a) APS1 (At5g48300) gene constructs used for transformation of the AGPase-deficient adg1 mutant. The Cys residues (red) in wild-type APS1 (top) were individually changed to Ser (green).
(b) Iodine staining showing the presence of starch in the APS1C81S-complemented adg1 mutant.
(c) Monomerization of APS1 protein determined by non-reducing SDS–PAGE and immunoblotting. Leaves were harvested from wild-type and mutant plants 2 h into the night when all the APS1 protein in wild-type plants is dimerized.

Cys81 is necessary and sufficient for dimerization of APS1 in the dark

Leaves were harvested from wild-type controls and each of the APS1-complemented adg1 mutants 2 h into the night, when AGPase is expected to be fully oxidized and the APS1 protein is expected to exist in the dimeric form (Tiessen et al., 2002; Hendriks et al., 2003). Protein extracts were analysed by immunoblotting after non-reducing SDS–PAGE to determine the redox state of AGPase (Figure 1c). A major immunoreactive band at approximately 100 kDa, corresponding to the APS1 dimer, was present in wild-type extracts, with little or no signal at 50 kDa, representing monomeric APS1 protein. A similar result was obtained for the complemented adg1 lines, with the exception of adg1/APS1C81S. Darkened leaves from adg1/APS1C81S contained only the monomeric (50 kDa) form of APS1, as expected if the mutagenized small subunit protein is unable to dimerize. These findings provide compelling evidence that Cys81 is both necessary and sufficent for the APS1 protein to form a disulphide bridge.

For further physiological characterization of the transformants, plants were propagated and screened for two more generations to obtain one adg1/APS1WT line (line 21) and three independent lines for adg1/APS1C81S [line A (35-237), line B (11-2) and line C (13-8)] that were homozygous for the respective transgenes. As the activity of AGPase is known to be dependent on gene dosage (Neuhaus and Stitt, 1990), we checked that these lines contained a single transgenic locus by Southern hybridization (Figure S3).

Involvement of the large subunits in formation of the disulfide bridge in vivo

Previous in vitro studies on heterologously expressed potato tuber AGPase indicated that the disulphide bridge is formed between the small subunits of the enzyme, but we cannot exclude the possibility that the APS1 protein can also form a disulphide bridge with one of the large (APL) subunits in vivo. To investigate this possibility, we harvested leaves from wild-type Col-0 plants in the dark, separated the proteins by non-reducing SDS–PAGE, and identified the AGPase subunits in the 100 kDa region of the gel by peptide mass spectrometry. In experiment 1, all 11 of the identified AGPase peptides were derived from the small subunit (APS1), and in experiment 2, ten of the 11 peptides were also from APS1, with only a single peptide derived from APL1 (Table 1). We would expect to find approximately equal numbers of APS1 and APL peptides if APS1 formed a dimer with any of the large subunit isoforms, so the observed ratio of APS1:APL1 peptides of 21:1 is inconsistent with the presence of APS1–APL1 dimers. The single APL1 peptide detected may be attributable to contamination of the sample, which is difficult to avoid with such a sensitive analytical technique as peptide mass spectrometry. Therefore, we conclude that the redox regulation of AGPase in A. thaliana leaves in vivo occurs via formation of a disulfide bridge between the Cys81 residues of the two small subunits.

Table 1.   AGPase subunits identified in the 100 kDa region from Arabidopsis thaliana leaf extracts
Protein identifierNumber of peptides
Sample 1Sample 2
  1. Extracts from wild-type Col-0 leaves were separated by non-reducing SDS–PAGE. The gel region adjacent to the 100 kDa marker protein was excised, and proteins were identified by peptide mass spectrometry after tryptic digestion. The number of peptides identified for each APS and APL isoform is shown for two independent replicates.

At5g48300.1 (APS1)1011
At1g05610.1 (APS2)00
At5g19220.1 (APL1)10
At1g27680.1 (APL2)00
At4g39210.1 (APL3)00
At2g21590.1 (APL4)00

Characterization of adg1/APS1C81S mutants under a 12 h light/12 h dark diurnal cycle

Wild-type Col-0, adg1/APS1WT and three independent adg1/APS1C81S lines, each containing a single transgenic locus, were grown under a 12 h light/12 h dark diurnal cycle in low light (150 μE m−2 sec−1), and sampled every 4 h to investigate APS1 transcript levels, AGPase activity, AGPase activation state and starch content.

Compared to the wild-type Col-0, APS1 transcript levels were almost exactly twofold higher in the adg1/APS1WT controls and adg1/APS1C81S mutants throughout the diurnal cycle (Figure 2a). The wild-type Col-0 and adg1/APS1WT plants showed similar maximum catalytic activities of AGPase (measured in the presence of 3PGA) (Figure 2b). In contrast, all three of the adg1/APS1C81S lines contained much lower AGPase activity (five- to tenfold lower than the controls, depending on the time of day), despite having comparable APS1 mRNA expression levels to adg1/APS1WT. The extraction buffer used to extract the leaf tissue contained DTT, and DTT was included in the AGPase assay; the measured AGPase activities therefore represent the fully reduced form of the enzyme.

Figure 2.

 Diurnal changes in APS1 transcripts, AGPase activity, starch content and post-translational modification of APS1 in the wild-type and adg1 mutants complemented with APS1WT or APS1C81S.
Wild-type Col-0, adg1/APS1WT and adg1/APS1C81S plants were grown under a 12 h light/12 h dark diurnal cycle with 150 μE m−2 sec−1 irradiance.
(a) Transcripts of APS1 (At5g48300).
(b) Maximum catalytic activity of AGPase.
(c) Oligomeric state of APS1.
(d) Leaf starch content.
For (a), (b) and (d), the values are means ± SD (= 5 separate samples, each comprising three individual rosettes). Significant differences from wild-type Col-0 according to Student’s t test are indicated by asterisks (*< 0.05, **< 0.05, ***< 0.001).

The APS1 protein is almost immediately oxidized in extracts prepared from A. thaliana leaves (Hendriks et al., 2003). Therefore, to detect differences in dimerization of the enzyme, extracts were prepared in trichloroacetic acid (TCA). TCA immediately denatures the AGPase holoenzyme and prevents formation of APS1 dimers during extraction and analysis (Hendriks et al., 2003). In wild-type Col-0 plants (and adg1/APS1WT, data not shown) APS1 was totally dimerized at the end of the night and became progressively reduced to the monomeric form during the day (Figure 2c). In contrast, only the monomeric (50 kDa) form of the APS1 protein was observed at any time point in the adg1/APS1C81S mutants, either during the day or at night. In agreement with the large decrease in total AGPase activity (Figure 2b), the total amount of APS1 protein detected by immunoblotting was noticeably lower in adg1/APS1C81S plants than in wild-type controls (Figure 2c).

Despite the five- to tenfold decrease in maximal AGPase activity, the starch content of leaves from all three adg1/APS1C81S lines was consistently higher than in wild-type Col-0 and adg1/APS1WT, although the diurnal pattern of starch accumulation and degradation was retained (Figure 2d). The difference was especially marked at dawn, when the adg1/APS1C81S plants contained twice as much starch as wild-type Col-0 or adg1/APS1WT. Sugar levels showed a slight tendency to be lower during the light period in adg1/APS1C81S plants than in control plants (data not shown).

In summary, replacement of Cys81 by serine not only prevents dimerization of APS1, but also leads to the plants having less APS1 protein and five- to tenfold lower maximum catalytic activity of AGPase activity than wild-type plants. Despite the lower levels of AGPase protein, constitutive activation of APS1 allows the adg1/APS1C81S plants to maintain similar or even higher rates of starch synthesis than control plants.

ADPGlc levels

We used LC-MS/MS to investigate whether ADPGlc levels were altered in adg1/APS1C81S plants (Figure 3). In both wild-type Col-0 and adg1/APS1WT plants, leaves harvested in the light at the end of the day contained approximately 1.5 nmol of ADPGlc per gram fresh weight. The level of ADPGlc was 70–110% higher in the adg1/APS1C81S plants (Figure 3a), indicating that constitutive activation of APS1 allows a higher level of ADPGlc to be maintained despite the much lower maximal AGPase activity in these lines. In leaves from both wild-type and mutant plants harvested in the dark, the level of ADPGlc was 35–45 times lower than in the light. Both of the adg1/APS1C81S lines showed a tendency to have higher absolute levels of ADPGlc than the wild-type Col-0 and adg1/APS1WT plants (Figure 3b), although the differences were not statistically significant. The much lower levels of ADPGlc in the adg1/APS1C81S lines in the dark, compared to the same plants in the light, indicate that AGPase activity was restricted at night, despite the inability of the mutated enzyme to be redox-regulated. This suggests that other mechanisms, such as substrate limitation and allosteric regulation, are operating to limit ADPGlc synthesis at night, which would otherwise allow futile cycling of ATP and hexose phosphates. Nevertheless, the tendency of the adg1/APS1C81S lines to have slightly higher night-time levels of ADPGlc than the control plants suggests that there may be a very low rate of ADPGlc synthesis in the dark in the redox-insensitive mutants.

Figure 3.

 ADP glucose content in wild-type and APS1-complemented adg1 mutants.
ADPGlc content was measured by LC-MS/MS in leaves harvested at the end of the day (ED) (a) and at the end of the night (EN) (b). The measurements were performed using material from the experiment shown in Figure 2. Values are means ± SD (= 3). Significant differences from wild-type Col-0 according to Student’s t test are indicated by asterisks (*< 0.01).

Characterization of diurnal starch turnover in adg1/APS1C81S mutants under various photoperiod conditions

Starch levels in the adg1/APS1C81S lines were moderately, but consistently, higher than in wild-type plants, when grown under a 12 h light/12 h dark diurnal cycle (Figure 2d). Further experiments were performed to investigate the influence of photoperiod and irradiance on diurnal starch turnover. Wild-type Col-0, adg1/APS1WT and adg1/APS1C81S mutant plants were grown under short-day (8 h light/16 h dark) and long-day (16 h light/8 h dark) conditions with low irradiance (150 μE m−2 sec−1, comparable to the experiment in Figure 2), or under a 12 h light/12 h dark cycle with high irradiance (450 μE m−2 sec−1), and samples were taken every 4 h for determination of starch content. As previously observed in A. thaliana (see Introduction), the wild-type Col-0 and adg1/APS1WT plants adjusted their rates of starch synthesis and degradation according to the day length. Only a small amount of starch was left at the end of the night, indicating that the rates of starch accumulation in the light and utilization during the night are tightly coordinated. This pattern was modified in the adg1/APS1C81S plants. Under low-light and short-day conditions, starch was synthesized more rapidly during the 8 h light period, leading to a 40% higher starch content at the end of the day, and degraded more rapidly at night (Figure 4a). As already mentioned, under low-light and equinoctial (12 h light/12 h dark) conditions, the rates of starch synthesis and degradation resembled wild-type controls, but more starch was left at the end of the night (Figure 2d). This pattern was even more marked under long-day conditions (16 h light/8 h dark; Figure 4b), under which the mutant showed a mild starch excess phenotype. Finally, under high-light equinoctial (12 h light/12 h dark) conditions, starch synthesis rates were marginally increased, and even more starch remained at the end of the night (Figure 4c). Under all growth conditions, the wild-type Col-0 and adg1/APS1WT lines had almost identical patterns of diurnal starch turnover to each other, and the two independent adg1/APS1C81S lines consistently showed similar differences in behaviour compared to the control plants.

Figure 4.

 Influence of photoperiod and irradiance on starch turnover in wild-type and adg1/APS1C81S mutant plants during the diurnal cycle.
Starch content was determined in leaf extracts of wild-type Col-0, adg1/APS1WT and adg1/APS1C81S plants grown under (a) 8 h light/16 h dark or (b) 16 h light/8 h dark diurnal cycles in low light (150 μE m−2 sec−1), or (c) a 12 h light/12 h dark diurnal cycle in high light (450 μE m−2 sec−1). The values are means ± SD (= 3 separate samples, each comprising three individual rosettes). In (c) the samples harvested at 24 h were lost during analysis and the values shown at this time point are the same as the 0 h samples. Significant differences from wild-type Col-0 according to Student’s t test are indicated by asterisks (*< 0.05, **< 0.01, ***< 0.001).

Maltose levels in wild-type and adg1/APS1C81S plants

The results in Figures 2 and 4 show that the adg1/APS1C81S mutants appear to have lost the ability to match their rates of starch degradation and synthesis, except when grown under short-day conditions with low light. Under these conditions, the mutants were able to increase the rate of starch degradation to remobilize all of their starch during the night, even though they had more starch than the control plants at the start of the night (Figure 4a). However, under long-day or high-light conditions, the adg1/APS1C81S mutants did not degrade all of their starch during the night, suggesting either that they had insufficient starch degrading capacity to cope with the moderately elevated amount of starch, or that starch degradation was somehow impaired in these mutants.

To investigate these possibilities, we measured the levels of maltose, which is the major product of starch degradation in leaves (Zeeman et al., 2007a). Maltose levels were measured in plants grown under equinoctial (12 h light/12 h dark) conditions with an irradiance of 150 μE m−2 sec−1 (in the same samples as those shown in Figure 2). In the wild-type Col-0 and adg1/APS1WT plants, maltose levels increased in the first 4 h of the night and decreased through the rest of the night, and maltose was present at only very low levels in the light (Figure 5). In the two adg1/APS1C81S lines, maltose levels were consistently 50–100% higher than in wild-type controls in the dark. The redox-insensitive APS1 mutants also contained more maltotriose and maltotetraose (data not shown). Unexpectedly, maltose levels were also consistently twofold higher in these mutants in the light than in wild-type controls (Figure 5). These data suggest that there may be a higher rate of starch degradation in adg1/APS1C81S plants than in the wild-type, and that starch degradation may also occur in the light.

Figure 5.

 Diurnal changes in maltose levels.
Wild-type Col-0 plants (red circle), adg1/APS1WT plants (pink circle), and plants of adg1/APS1C81S line A (green circle) and adg1/APS1C81S line B (green triangle) were grown under a 12 h light/12 h dark diurnal cycle with an irradiance of 150 μE m−2 sec−1. Maltose levels were determined by HPAEC-PAD using material from the experiment shown in Figure 2. Values are means ± SD (= 3).

Response of purified AGPase to redox changes

As already mentioned, APS1 is rapidly oxidized to form a dimer in A. thaliana leaf extracts (Hendriks et al., 2003), making it difficult to investigate the effect of redox changes on the activity of the AGPase from the adg1/APS1C81S lines in crude extracts. To overcome this problem, we purified AGPase from both wild-type Col-0 plants and the adg1/APS1C81S mutants by dye affinity chromatography on a Mimetic Orange 1 column followed by size exclusion chromatography on a Superdex 200 column (for details, see Appendix S1). Purification of the mutated C81S enzyme was complicated by the low concentration of the protein in the adg1/APS1C81S mutant, and by the poor recovery of activity from the Orange1 and Superdex200 columns during the purification, although a substantial degree of purification (356-fold) was achieved (Table S1). The wild-type enzyme was purified 184-fold with a yield of 6%. The final specific activity of the wild-type enzyme was approximately tenfold higher than that of the mutated C81S form. This was at least partly due to a greater degree of contamination by other proteins in the final C81S preparation (Figure S4 and Table S2), and possibly also to partial inactivation during purification. Peptide analysis after tryptic digestion indicated that APS1 was the only small subunit isoform present in the purified enzymes, and that APL1 was by far the predominant large subunit isoform (Table S2). The activity of the wild-type AGPase was significantly decreased by incubation with oxidized dithiothreitol, but this treatment had no effect on the activity of the mutated C81S form (Figure S5). The low yield of the purified enzymes prevented further analysis of their kinetic properties.


The A. thaliana leaf AGPase is redox-activated by light and in response to sucrose accumulation in the leaf (Hendriks et al., 2003; Lunn et al., 2006). This undoubtedly forms part of the wider network of regulatory mechanisms that control carbon partitioning between starch and sucrose, and coordinate end-product synthesis with CO2 assimilation, but its precise physiological significance is uncertain. We have established that redox regulation of the A. thaliana AGPase in vivo ocurs via formation of a disulphide bridge between the Cys81 residues of the two small subunits within the heterotetrameric holoenzyme. This leads to APS1–APS1 dimerization and decreased maximal activity. The chloroplast transit peptide of the A. thaliana APS1 protein is predicted to be 69 amino acids long (Hädrich et al., 2011), thus Cys81 is predicted to become Cys12 in the mature protein after cleavage of the transit peptide. This residue is equivalent to Cys12 in the mature potato enzyme, which forms a disulfide bridge in the heterologously expressed enzyme (Fu et al., 1998; Ballicora et al., 1999, 2000). Individual replacement of the other four cysteine residues by serine did not affect the ability of the A. thaliana APS1 protein to form a dimer when oxidized. Thus, Cys81 is both necessary and sufficient for formation of the intermolecular disulphide bridge between the two APS1 subunits. Interestingly, all five of the mutated forms of APS1 containing Cys/Ser substitutions were able to complement the starchless phenotype of the adg1 mutant, and thus none of the cysteine residues in APS1 appear to be essential for catalytic activity of the holoenzyme. This was true even for the mutated C131S form. This lacks the only cysteine residue found in the nucleotidyl transferase domain of the APS1 protein, which is the only cysteine residue that is conserved in all plant and bacterial AGPases (Figure S1).

The full-length APS1 gene, including all of the upstream intergenic region and part of the downstream region beyond the polyadenylation site, was used to produce constructs for complementation of the adg1 mutant. The aim was to include the whole promoter and all other regulatory elements, so that the level and pattern of expression of the transgene would be as close as possible to that in wild-type plants. Expression of the APS1 gene was measured by quantitative RT-PCR using primers that did not distinguish between transcripts derived from the introduced APS1 transgene and non-functional APS1 transcripts from the endogenous mutated APS1 gene in the adg1 mutant. APS1 transcript levels in the complemented adg1 mutants were almost exactly twofold higher than in wild-type Col-0 plants at all points during the diurnal cycle (Figure 2a). This doubling of APS1 transcript abundance in the complemented adg1 lines, but similar diurnal patterns of expression to wild-type Col-0, indicates that the APS1 transgenes were subject to the same transcriptional regulation as the endogenous APS1 gene. Despite the differences in APS1 transcript levels, the AGPase activity and starch content of the adg1/APS1WT line were remarkably similar to those of wild-type Col-0 plants. This suggests that either the non-functional APS1 transcripts derived from the endogenous mutated APS1G92R gene in the adg1 mutant were not translated, or that the inactive APS1G92R protein is unstable and rapidly degraded. It is worth noting that the expression level of the APL1 transcript, which encodes the predominant large subunit isoform in leaves, was essentially the same in both wild-type Col-0 and the complemented adg1 mutant lines (data not shown).

It has previously been observed that APS1 and APL1 transcript levels are not always correlated with maximal AGPase activity and amounts of AGPase protein in A. thaliana leaves (Gibon et al., 2004a,b; Piques et al., 2009), which suggests that post-transcriptional regulation can also be important in determining the amount and activity of the enzyme. Despite having almost identical APS1 transcript levels to the adg1/APS1WT line, the adg1/APS1C81S lines had noticeably lower APS1 protein levels (Figure 2c) and five- to tenfold lower AGPase activity (Figure 2b) than both the wild-type Col-0 and adg1/APS1WT controls. This suggests that post-transcriptional regulation of the enzyme’s abundance is perturbed in the adg1/APS1C81S mutants.

There are several possible explanations for the low levels of AGPase protein in the adg1/APS1C81S lines. One is that AGPase containing the mutated C81S form of the small subunit is inherently less stable than the wild-type enzyme. Ballicora et al. (1999) showed that the wild-type potato enzyme was heat-stable up to 60°C when oxidized (i.e. when APS1 is dimerized), but was heat-stable up to only 40°C when reduced (i.e. when APS1 is monomerized) or when Cys12 was replaced by alanine or serine preventing formation of the intermolecular disulphide bridge. However, below 40°C, there were no significant differences in the stability of the various forms of the enzyme. As our plants were grown at 18–20°C and AGPase activity was measured at 25°C, temperatures that are well below the 40°C threshold, it seems most unlikely that differences in the thermal stability of the mutated enzyme account for the lower APS1 protein abundance and AGPase activity in the adg1/APS1C81S plants. Furthermore, it is possible that the thermal stability of the AGPase in vivo is greater than the in vitro experiments on the isolated potato enzyme may suggest, as the very high concentration of proteins in the chloroplast stroma may be expected to have a stabilizing effect on the enzyme.

It is known that AGPase is subject to rapid turnover in leaves, and that reduction of the enzyme in the light is often accompanied by a decrease in the enzyme’s abundance (Gibon et al., 2004a,b; Piques et al., 2009). These observations suggest that the reduced (monomeric APS1) form of the enzyme may be more susceptible to proteolytic turnover than the oxidized (dimeric APS1) form, which may account for the low abundance of the enzyme in the adg1/APS1C81S plants (in which the APS1 cannot dimerize). It is unclear whether the monomeric APS1 form of the enzyme simply has a conformation that makes it more accessible to chloroplastic proteases than the dimeric APS1 form, or whether proteolytic turnover of the enzyme is more precisely regulated.

The level and diurnal turnover of starch were identical in wild-type and adg1/APS1WT plants. In contrast, adg1/APS1C81S lines, despite their lower maximal activity of AGPase, had increased rates of starch accumulation and/or higher levels of starch than the control plants. This suggests that the in vivo activity of the C81S form of AGPase is the same as, or even higher than, the in vivo activity of the more abundant wild-type enzyme in the control plants, or that AGPase was not limiting for starch synthesis under our growth conditions. The differences in the diurnal pattern of starch turnover depended on the photoperiod and irradiance (Figures 2d and 4). Rates of both starch synthesis and degradation were increased under short days and low light, whereas the plants grown under long days or high-light conditions had increased levels of starch at the end of the light period but incomplete remobilization of starch at night. It is well established that plants respond to changes in photoperiod by adjusting their rates of starch synthesis and degradation, although very little is known about the mechanisms involved (Smith and Stitt, 2007). Under 12 and 16 h days, the rates of starch synthesis in the adg1/APS1C81S and wild-type control plants were remarkably similar (Figure 2d and 4). This suggests that redox regulation of AGPase is not required to adjust the rate of starch synthesis to longer day lengths, or that other regulatory mechanisms compensateBfor the loss of redox regulation.

The adg1/APS1C81S lines contained 75–110% more ADPGlc than wild-type Col-0 or adg1/APS1WT plants in the light (Figure 3a). This is consistent with the in vivo flux catalysed by AGPase being higher in the adg1/APS1C81S lines. The increase in ADPGlc is greater than the increase in the net rate of starch accumulation (Figure 2d). This may indicate that the starch synthases are already near-saturated by the wild-type levels of ADPGlc, or that there is increased turnover of starch in the mutants during the day. In all genotypes, the level of ADPGlc at night was 35–45 times lower than in the day (Figure 3b). The low night-time levels of ADPGlc indicate that the in vivo activity of the mutated enzyme, like that of the wild-type enzyme, is much lower at night than during the day, despite the loss of redox regulation. Limited substrate availability (ATP and Glc1P) and a low 3PGA:Pi ratio in the chloroplast stroma may therefore be important factors in restricting AGPase activity at night in both wild-type and mutant plants.

Transitory starch degradation involves an initial cycle of glucan phosphorylation and dephosphorylation, followed by β-amylolysis to release maltose (Zeeman et al., 2007a). Maltose levels increased in wild-type Col-0 plants in the dark (Figure 5), as previously observed (Niittyla et al., 2004; Delatte et al., 2005; Lu et al., 2005; 2006), and very similar changes were seen in the adg1/APS1WT plants (Figure 5). The adg1/APS1C81S lines contained higher maltose levels than wild-type Col-0 plants or the control adg1/APS1WT line, not only in the dark but also in the light (Figure 5). The incomplete degradation of starch at night, and the increased levels of maltose, indicate that the presence of the APS1C81S form of AGPase somehow disrupts starch breakdown and/or maltose metabolism.

The control of starch degradation in leaves is poorly understood, so we can only speculate on possible mechanisms for the effect of the mutated AGPase on starch turnover. We observed no significant differences between wild-type and mutant plants in terms of the levels of gene transcripts encoding the main enzymes of starch degradation (glucan:water dikinase, phosphoglucan:water dikinase, phosphoglucan phosphatase, β-amylase, debranching enzymes and disproportionating enzymes) or the MEX1 maltose transporter (data not shown). Therefore, we can probably exclude transcript level regulation as a possible mechanism. Some of the enzymes of starch degradation (e.g. glucan:water dikinase and some isoforms of β-amylase) are known to be redox-sensitive, although the significance of this is uncertain. However, it is conceivable that thioredoxins that bind to the docking domains of the APS1C81S form of AGPase, but are unable to complete the redox transfer reaction, remain bound to the enzyme. The partial sequestration of these redox carriers may perturb redox regulation of other enzymes, including the redox-sensitive starch-degrading enzymes. It is known that the circadian clock has a major input into regulation of starch degradation at night (Graf et al., 2010; Graf and Smith, 2011). Current models for understanding the role of the clock involve a starch-sensing mechanism of some kind that, together with the clock, allows the plant to set the appropriate rate of starch degradation for the starch reserves to last through the night. Therefore, a further possibility is that some aspect of circadian regulation or the perception of starch is perturbed in the adg1/APS1C81S mutant, leading to lower rates of starch degradation and/or enhanced turnover of starch in the light.

This uncertainty in explaining the high-maltose and starch-excess phenotypes of the adg1/APS1C81S mutant highlights how little we know about the control of diurnal starch turnover in leaves, despite the importance of this process for optimal plant growth (Smith and Stitt, 2007).

Experimental procedures

Plant growth

Arabidopsis thaliana plants were grown as described by Thimm et al. (2004) under a 12 h light/12 h dark (20°C/18°C) cycle and 150 μE m−2 sec−1 irradiance unless stated otherwise, and harvested at 35–40 days after germination. Harvested material was immediately frozen in liquid nitrogen, ground to a fine powder at −70°C using a cryogenic grinding robot (Labman Automation Ltd, and stored at −80°C until analysis.

Cloning and site-directed mutagenesis

Standard molecular biology techniques were used as described by Sambrook and Russell (2001). The APS1 gene (At5g48300) was amplified from genomic DNA by PCR using KOD HiFi DNA polymerase (Merck, using forward primer 5′-TAGTCATATGAATAAAGCTCTGAGG-3′ and reverse primer 5′-AGCGCTTTAAAACGAATAATGTTGAACTAC-3′. The PCR product was cloned into the pGEM®-T Easy vector (Promega, and sequenced on both strands to check amplification fidelity.

Point mutations were introduced into the APS1 coding sequence to change individual TGT(Cys) or TGC(Cys) codons into TCT(Ser) or TCA(Ser) codons, respectively, using a QuickChange® site-directed mutagenesis kit (Stratagene, according to the manufacturer’s instructions. Primers are listed in Table S3. Mutations were confirmed by sequencing. The mutated APS1 genes were subcloned into the pGreen plant expression vector (Hellens et al., 2000), and introduced into the adg1 mutant via Agrobacterium-mediated transformation using the floral-dip method(Clough and Bent, 1998). Seeds were collected from transformed plants, germinated on soil and selected by spraying with 0.02% w/v phosphinothricin at 5, 7, 9, 11 and 13 days after germination.

AGPase activity

Frozen tissue powder (20 mg fresh weight) was extracted using extraction buffer containing 1 mm DTT and AGPase assayed in the direction of ADPGlc pyrophosphorolysis, in the presence of 1 mm DTT and 5 mm 3PGA at 25°C as described by Gibon et al. (2004b).

AGPase protein and activation state

Extraction of AGPase, separation by non-reducing SDS–PAGE and electroblotting onto nitrocellulose membrane were performed as described by Hendriks et al. (2003). After blocking with 5% (w/v) non-fat milk powder in 20 mM Tris-HCl, 0.5 M NaCl, pH 7.5 (1×TBS) for 1 h at 20°C, the membrane was incubated with anti-potato AGPB antiserum (1:10 000 dilution in 1×TBS containing 2.5 % (w/v) non-fat milk powder; Tiessen et al., 2002) for 1 h at 20°C. After washing with three changes of 1×TBS (5 min per wash, 20 °C), the membrane was incubated with IRDye800-conjugated goat anti-rabbit IgG F(c) antibody (1:7500 dilution in 1×TBS; Biomol, After washing with three changes of 1xTBS, as above, the fluorescent infrared signal from the dye-labeled secondary antibody was detected using an Odyssey infrared imaging system (LI-COR,

Determination of metabolites

Starch was measured in the insoluble fraction after ethanolic extraction as described by Hendriks et al. (2003). Protein was determined by a modification of the dye-binding method (Bradford, 1976) as described by Gibon et al. (2009).

Maltose and malto-oligosaccharides were measured in perchloric acid extracts as described by Fulton et al. (2008). Anionic and cationic compounds were removed from the extract by sequential passage through 2 ml columns of Dowex 50Wx4-100 and Dowex 1×8-200 (Sigma-Aldrich, Neutral compounds were eluted using 5 ml water, lyophilized, redissolved in 100 μl water and analysed by HPAEC-PAD using a CarboPac PA20 column (Dionex, (Fulton et al., 2008). Sugars were identified by co-elution with authentic standards, and quantified using Chromeleon analysis software (Dionex).

ADPGlc was measured in chloroform:methanol extracts by anion exchange chromatography coupled to triple quadrupole mass spectrometry (HPAEC-MS/MS; Lunn et al., 2006).


We thank Uschi Krause and Melanie Höhne for expert technical assistance. This research was supported by the German Ministry for Education and Research (GABI-TILL 0313123D) and the European Commission (FP7 Collaborative Project TiMet).