The Role of Cytosolic α-Glucan Phosphorylase in Maltose Metabolism and the Comparison of Amylomaltase in Arabidopsis and Escherichia coli

Abstract

Transitory starch of leaves is broken down hydrolytically, making maltose the predominant form of carbon exported from chloroplasts at night. Maltose metabolism in the cytoplasm of Escherichia coli requires amylomaltase (MalQ) and maltodextrin phosphorylase (MalP). Possible orthologs of MalQ and MalP in the cytosol of Arabidopsis (Arabidopsis thaliana) were proposed as disproportionating enzyme (DPE2, At2g40840) and α-glucan phosphorylase (AtPHS2, At3g46970). In this article, we measured the activities of recombinant DPE2 and AtPHS2 proteins with various substrates; we show that maltose and a highly branched, soluble heteroglycan (SHG) are excellent substrates for DPE2 and propose that a SHG is the in vivo substrate for DPE2 and AtPHS2. In E. coli, MalQ and MalP preferentially use smaller maltodextrins (G₃–G₇) and we suggest that MalQ and DPE2 have similar, but nonidentical, roles in maltose metabolism. To study this, we complemented a MalQ⁻ E. coli strain with DPE2 and found that the rescue was not complete. To investigate the role of AtPHS2 in maltose metabolism, we characterized a T-DNA insertion line of the AtPHS2 gene. The nighttime maltose level increased 4 times in the Atphs2-1 mutant. The comparison of maltose metabolism in Arabidopsis with that in E. coli and the comparison of the maltose level in plants lacking DPE2 or AtPHS2 indicate that an alternative route to metabolize the glucan residues in SHG exists. Other plant species also contain SHG, DPE2, and α-glucan phosphorylase, so this pathway for maltose metabolism may be widespread among plants.

Substantial progress has been made recently in understanding the pathway of transitory starch breakdown (Ball and Morell, 2003; Lloyd et al., 2005; Smith et al., 2005; Lu and Sharkey, 2006). It is now generally accepted that transitory starch is broken down to maltose and Glc in the chloroplast at night by β-amylase and a disproportionating enzyme (DPE) inside chloroplasts, DPE1 (Lao et al., 1999; Critchley et al., 2001; Scheidig et al., 2002). Recent data from three different groups indicated that maltose and Glc are the two major forms of carbon exported to the cytosol during starch degradation (Servaites and Geiger, 2002; Ritte and Raschke, 2003; Weise et al., 2004). This finding is supported by the discovery of a novel maltose transporter, MEX1, on the chloroplast envelope (Niittylä et al., 2004). MEX1-deficient Arabidopsis (Arabidopsis thaliana) plants showed growth retardation and a starch-excess phenotype and they accumulated a substantial amount of maltose, indicating that maltose export from chloroplasts is essential in transitory starch breakdown.

Recent studies from two different groups showed that mutations in the cytosolic amylomaltase (MalQ; or DPE2) led to a substantial increase in leaf maltose content, a starch-excess phenotype, and lower nighttime Suc, suggesting that DPE2 is involved in cytosolic maltose conversion to Suc in Arabidopsis (Chia et al., 2004; Lu and Sharkey, 2004). A significant amount of maltose was detected in the phloem exudates collected from dpe2-1 petioles. This suggests that excess maltose in the dpe2-1 leaves can be exported (Lu et al., 2006). The high affinity of DPE2 toward glycogen on native glycogen gels suggests that glycogen could be a second substrate for the disproportionating activity of DPE2 in vitro and this reaction has been demonstrated (Chia et al., 2004). One possibility is that DPE2 uses a soluble heteroglycan (SHG) as one of its substrates (Lu and Sharkey, 2004, 2006). According to this hypothesis, DPE2 should require both maltose and SHG for maximal activity. Fettke et al. (2006) showed that a subfraction of SHG increases in Glc content in the presence of maltose and DPE2 recombinant enzyme.

In the cytoplasm of Escherichia coli, maltose and maltodextrins (up to maltoheptaose, G₇) are metabolized to Glc, Glc-1-P (G1P), and Glc-6-P (G6P) by the combined action of MalQ, maltodextrin phosphorylase (MalP), maltodextrin glucosidase, and glucokinase (Boos and Shuman, 1998). MalQ transfers maltosyl and longer dextrinyl residues onto Glc, maltose, and longer maltodextrins (Palmer et al., 1976; Boos and Shuman, 1998). Maltose is a poor substrate for MalQ in the absence of any other maltodextrins, but the Glc-releasing activity with maltose as the sole substrate is autocatalytic and Glc release increases after a lag phase (Palmer et al., 1976). It has been proposed that, analogous to MalQ in E. coli, DPE2 converts maltose and SHG to Glc and an SHG that is one glucosyl unit longer (Lu and Sharkey, 2004, 2006; Fettke et al., 2006). Mutants of E. coli lacking MalQ accumulate maltose just like mutants of Arabidopsis lacking DPE2 (Szmelcman et al., 1976; Lu and Sharkey, 2004). Lloyd et al. (2004) identified potato (Solanum tuberosum) DPE2 (stDPE2) by transforming a cDNA library from potato leaves into the E. coli strain TSM90 that cannot metabolize maltose. stDPE2-transformed TSM90 cells grew on maltose, suggesting that the TSM90 strain has a lesion in MalQ (Lloyd et al., 2004). By comparing maltose metabolism in Arabidopsis with well-studied maltose/maltodextrin metabolism in E. coli, we could make predictions for genes of uncertain functions in the Arabidopsis pathway.

SHG has been isolated from the leaves of spinach (Spinacia oleracea; Yang and Steup, 1990), pea (Pisum sativum; Yang and Steup, 1990; Fettke et al., 2004), Arabidopsis (Lu and Sharkey, 2004; Fettke et al., 2005a), and potato (Fettke et al., 2005b). The major constituents of the SHG in Arabidopsis are Gal, Ara, and Glc (Lu and Sharkey, 2004; Fettke et al., 2005a). SHG has a high affinity for the cytosolic α-glucan phosphorylase (Pho2 in general, or AtPHS2 in Arabidopsis; Yang and Steup, 1990; Fettke et al., 2004, 2005a, 2005b). The total SHG isolated from Arabidopsis leaves can be separated into low (<10 kD [SHG_S]) and high (>10 kD [SHG_L]) M_r polysaccharides (Fettke et al., 2005a). Fettke et al. (2005a) reported that the pool size of a subfraction of SHG_L varied during the light-dark regime. SHG_S was localized to the cytosol, whereas SHG_L contained a cytosolic subfraction and an apoplastic subfraction (Fettke et al., 2005a). The cytosolic SHG possessed priming capacity for recombinant Pho2 from fava bean (Vicia faba). It is plausible to speculate that the glucosyl residues in the cytosolic SHG, analogous to maltodextrin in the cytoplasm of E. coli, are both the substrate and product of DPE2 and the substrate for Pho2 in the cytosol. If SHG is involved in cytosolic maltose metabolism, one would expect production of Glc in the presence of maltose and SHG by recombinant DPE2. In the absence of Pho2, one would expect an increase in the length of the outer chains of certain SHG pools.

The role of Pho2 in conversion of transitory starch to Suc is not clear. It was recently reported that the expression of DPE2 and AtPHS2 was coordinated (Smith et al., 2004; Lu et al., 2005). In long days, their transcript levels increased between 1 and 9 h of the day and declined during the rest of the day and throughout the night (Lu et al., 2005). Based on the above results, it is reasonable to propose that AtPHS2 is involved in maltose metabolism. Interestingly, the activity of AtPHS2 was increased 3- to 4-fold in DPE2-deficient Arabidopsis mutants (Chia et al., 2004).

The involvement of AtPHS2 in cytosolic glycan metabolism has been proposed by three different groups (Chia et al., 2004; Lu and Sharkey, 2004; Schupp and Ziegler, 2004). However, antisense inhibition of Pho2 in potato plants has little impact on carbohydrate metabolism and it was proposed that Pho2 is not involved in transitory starch degradation (Duwenig et al., 1997). If AtPHS2 is involved in maltose metabolism in the cytosol, we expect to observe an increase in maltose in AtPHS2-deficient mutants. Because the maltose content in the leaves of Pho2 RNAi potato plants was not measured (Duwenig et al., 1997), we cannot rule out the possibility that Pho2 is involved in cytosolic glycan metabolism.

In this article, we looked at substrate preferences of recombinant DPE2 and AtPHS2 expressed in E. coli. To investigate the similarity and difference of MalQ and DPE2 functions in maltose metabolism, we studied a MalQ⁻ E. coli strain rescued with DPE2 from Arabidopsis. Using AtPHS2-deficent Arabidopsis plants and transgenic plants constitutively expressing AtPHS2, we investigated the effect of AtPHS2 on maltose metabolism and hypothesize why plants lacking DPE2 exhibit a more severe phenotype than plants lacking AtPHS2. Finally, we compare hypothetical pathways for maltose metabolism in E. coli and Arabidopsis.

RESULTS

SHG and Maltose Are the Preferred Substrates for DPE2

We found that maltose plus glycogen or SHG gave the greatest rate of Glc production by DPE2 (Fig. 1). DPE2 activity with maltose and dextrin was approximately one-sixth of that with maltose and glycogen. There was some DPE2 activity with maltose plus linear maltooligosaccharides (G₃–G₇) and the rate of Glc production increased as the degree of polymerization of the other substrate (G₃–G₇) increased. When no other substrate was present except maltose, DPE2 released very little Glc. When glycogen alone was present as the substrate, DPE2 released 0.0038 μmol Glc mg⁻¹ protein min⁻¹, even less than the rate of Glc production by DPE2 when maltose alone is present as the substrate (Fig. 1). DPE2 activity was not observed in boiled recombinant DPE2 protein or in protein purified from an E. coli culture containing empty-vector pET28a.

Figure 1.

Recombinant DPE2 activity with various substrates. G₂, Maltose; G₃, maltotriose; G₄, maltotetrose; G₅, maltopentaose; G₆, maltohexaose; G₇, maltoheptaose. All substrates were supplied in excess amounts so that the rate of Glc production was not limited by the concentration of the substrates but by the specificity of the substrates. SHG used was isolated from Arabidopsis leaves. Values are mean ± se (n = 3).

Having established that DPE2 prefers highly branched glycans as one of its substrates, the activity of recombinant DPE2 with linear maltooligosaccharides (G₂–G₇) was assayed in the presence of glycogen (Fig. 1). The combination of maltose and glycogen as the two substrates gave the greatest rate of Glc production by DPE2, indicating maltose is the preferred substrate for DPE2 in the presence of glycogen.

Recombinant AtPHS2 Uses SHG to Produce G1P

The AtPHS2 gene encodes a 95-kD functional cytosolic α-glucan phosphorylase containing 841 amino acid residues. To study the substrate specificity of AtPHS2, we assayed the activity of recombinant AtPHS2 with dextrin, amylopectin, glycogen, and total SHG isolated from Arabidopsis leaves. AtPHS2 converted glucosyl residues in SHG to G1P in the presence of excess inorganic phosphate (Pi; Fig. 2). The rates of G1P production of AtPHS2 with amylopectin, glycogen, and SHG were approximately 3 times as much as that with dextrin.

Figure 2.

Recombinant AtPHS2 activity with various substrates. Substrates were supplied in excess amounts in the assay buffer: 1 mg/mL for dextrin, amylopectin, SHG, and glycogen. Dextrin is less branched relative to amylopectin, glycogen, or SHG. SHG used was isolated from Arabidopsis leaves. Values are mean ± se (n = 3).

DPE2 from Arabidopsis Partially Complements the MalQ⁻ E. coli Strain

To study whether DPE2 in Arabidopsis is analogous to MalQ in E. coli, we observed the effect of DPE2 expression in a MalQ⁻ E. coli strain, MH70 (http://cgsc.biology.yale.edu). MH70 can grow on Glc, but not maltose, as the carbon source. The parent strain MC4100 can grow on either Glc or maltose. The expression of DPE2 allowed the growth of the MalQ⁻ strain MH70 on maltose, whereas introduction of the empty vector did not (Fig. 3).

Figure 3.

Partial complementation of MalQ⁻ E. coli strain MH70 with DPE2. A, E. coli on M63-maltose plates. B, E. coli on M63-Glc plates. Plates were incubated at 37°C for 24 h before photographing. C, E. coli stained with iodine on M63-maltose plates. D, E. coli stained with iodine on M63-Glc plates. Plates were incubated at 37°C for 24 h before staining and photographing. E, Growth curves in liquid M63-maltose medium. F, DPE2-transformed MH70 cells in liquid M63-maltose medium. White arrow, E. coli cells of regular size; black arrow, E. coli cells of large size. 1, pQE30 in MH70 (MalQ⁻); 2, T5:DPE2 in MH70; 3, pQE30 in MC4100 (the parent strain of MH70); 4, T5:DPE2 in MC4100.

DPE2 from Arabidopsis can only partially complement the MalQ⁻ E. coli strain MH70. MH70 transformed with DPE2 did not grow as fast as MC4100 transformed with the empty vector in liquid M63-maltose medium during the first 12 h (Fig. 3). In the same medium, MC4100 transformed with DPE2 did not grow as fast as MC4100 transformed with the empty vector during the first 12 h. After 24 h, MC4100 transformed with DPE2 overgrew MC4100 transformed with the empty vector. Approximately 50% of DPE2-transformed MH70 cells became very long and large in liquid M63-maltose medium, indicating that long glucan chains were accumulated in the cytoplasm (Schwartz, 1965, 1967; Boos and Shuman, 1998). The rest of DPE2-transformed MH70 cells were of normal size. After staining with iodine, DPE2-transformed MH70 cells turned dark purple on M63-maltose plates, whereas MH70 cells carrying the empty vector did not. This confirmed that DPE2-transformed MalQ⁻ E. coli cells tended to accumulate long glucans.

T-DNA Insertion in the AtPHS2 Gene Causes the Absence of Cytosolic α-Glucan Phosphorylase Activity

Cytosolic localization of AtPHS2 was formerly demonstrated by cell fractionation experiments (Delvalle et al., 2005). In Atphs2-1, a T-DNA was inserted in the 12th intron (Fig. 4). Two forms of AtPHS2 activity, H1 and H2 (H stands for cytosolic forms), were absent in the native glycogen gel of soluble leaf protein from Atphs2-1 plants (Fig. 4). Both H1 and H2 forms were increased in 35S:AtPHS2 lines (see below), indicating both H1 and H2 forms are encoded by the AtPHS2 gene. The major form of AtPHS2 activity, H1, had high affinity for glycogen as indicated by its lack of migration into the native glycogen gel. The minor form of AtPHS2 activity, H2, had intermediate affinity toward glycogen. The activity of the plastidial α-glucan phosphorylase (AtPHS1), the L1 form (L stands for plastidial forms), was not affected in Atphs2-1 leaves. No visible phenotype was observed, except that the rosette of Atphs2-1 plants was slightly bigger than that of wild-type plants at the same age (Fig. 4). Lesions seen in Arabidopsis plants lacking AtPHS1 (Zeeman et al., 2004) were not observed in Atphs2-1 plants.

Figure 4.

Analysis of the AtPHS2 gene and protein. A, The AtPHS2 locus in Arabidopsis. The gene structure is depicted from 0 (translation start) to 4,487 (translation stop) bp. Exons are depicted as white boxes and the T-DNA insertion site in Atphs2-1 is shown. B, The phenotype of Atphs2-1 plants. Plants were grown in 16-h light/8-h dark conditions and photographed at the same scale when they were 20 d old. C, Glycogen-containing native PAGE gel of soluble proteins from Atphs2-1 or wild-type Ler leaves. A total of 12 μg of soluble protein was loaded per lane. Gels were stained with 0.67% (w/v) I₂ and 3.33% (w/v) KI solution. Two forms of AtPHS2 activity, H1 and H2, were present in wild-type leaves and were absent in Atphs2-1 leaves. The activity of AtPHS1 was not affected in Atphs2-1 leaves. H, Cytosolic forms; L, plastidial forms.

We quantitatively measured the total activity of α-glucan phosphorylase (AtPHS1 and AtPHS2) with glycogen and maltoheptaose (G₇). AtPHS1 prefers to use maltooligosaccharides, such as maltoheptaose, and AtPHS2 prefers to use branched polysaccharides, such as glycogen (Smith et al., 2005). Whereas AtPHS2 contributes most of the activity with glycogen, AtPHS1 may also contribute, to a lesser extent, to the activity with glycogen. During the day, the total activity of α-glucan phosphorylase with glycogen was reduced by 82% in Atphs2-1 relative to the wild type and the total activity of α-glucan phosphorylase with maltoheptaose was reduced by 67% (Table I). During the night, the total activity of α-glucan phosphorylase with glycogen was reduced by 97% and the total activity of α-glucan phosphorylase with maltoheptaose was reduced by 84% (Table II).

Table I.

Midday activities of maltose-metabolizing enzymes in Atphs2-1 and Ler wild-type leaves

Leaf tissues were harvested 8 h into the light period. Total activity of α-glucan phosphorylase (AtPHS2 plus AtPHS1) is shown as the rate of G1P production from either glycogen or maltoheptaose. AtPHS1 has weak activity with glycogen and contributes to the residual α-glucan phosphorylase activity we observed in Atphs2-1 leaves when glycogen is the substrate. Activity of DPE1 is shown as the rate of Glc production from maltotriose. Activity of DPE2 is shown as the rate of Glc production from maltose plus glycogen. Activity of β-amylase is shown as the rate of maltose production from maltoheptaose. Activity of PGM is shown as the rate of G6P production from G1P. Activity of HXK is shown as the rate of G6P production from Glc. Values are mean ± se (n = 5).

Enzyme	Substrate	Activity
				Ler Wild Type	Atphs2-1
		μmol product g−1 protein min−1
α-Glucan phosphorylase	Glycogen	9.62 ± 0.46	1.69 ± 0.21
α-Glucan phosphorylase	Maltoheptaose	17.49 ± 0.68	5.81 ± 0.41
DPE1	Maltotriose	11.97 ± 0.30	11.60 ± 0.47
DPE2	Maltose + glycogen	16.69 ± 1.06	15.70 ± 1.02
β-Amylase	Maltoheptaose	368.3 ± 25.9	379.9 ± 32.9
PGM	G1P	225.6 ± 8.63	218.2 ± 10.1
HXK	Glc	12.58 ± 1.17	15.15 ± 0.77

Table II.

Midnight activities of maltose-metabolizing enzymes in Atphs2-1 and Ler wild-type leaves

Leaf tissues were harvested 4 h into the dark period. Enzymatic activities are shown the same way as in Table I. AtPHS1 has weak activity with glycogen and contributes to the residual α-glucan phosphorylase activity we observed in Atphs2-1 leaves when glycogen is the substrate. Values are mean ± se (n = 5).

Enzyme	Substrate	Activity
				Wild Type	Atphs2-1
		μmol product g−1 protein min−1
α-Glucan phosphorylase	Glycogen	9.23 ± 0.76	0.28 ± 0.17
α-Glucan phosphorylase	Maltoheptaose	13.56 ± 0.88	2.18 ± 0.55
DPE1	Maltotriose	7.88 ± 0.61	7.12 ± 0.62
DPE2	Maltose + glycogen	11.08 ± 1.02	9.84 ± 0.42
β-Amylase	Maltoheptaose	73.05 ± 2.83	70.34 ± 3.79
PGM	G1P	209.7 ± 8.5	193.1 ± 5.5
HXK	Glc	16.53 ± 1.28	15.93 ± 2.16

Absence of AtPHS2 Causes an Increase in Nighttime Maltose

The nighttime maltose content in Atphs2-1 leaves was about 4 times as much as that in Landsberg erecta (Ler) wild-type leaves (Fig. 5). This suggests that AtPHS2 is involved in maltose metabolism in the cytosol. However, the daytime maltose content in the mutant leaves was about the same as that in wild-type leaves. Atphs2-1 plants had normal amounts of G1P, Suc, and starch in leaves. Atphs2-1 plants had relatively high amounts of Glc, Fru, G6P, and F6P during the day in leaves. During the night, the Glc content in the total SHG extracted from Atphs2-1 leaves was not significantly different from that in corresponding Ler wild-type leaves (Fig. 6). We separated the total SHG into SHG_S (<10 kD) and SHG_L (>10 kD), and measured the Glc content. We found that the midnight SHG_L Glc content in Atphs2-1 leaves was approximately 40% higher than that in Ler wild-type leaves. The midnight SHG_S Glc content in Atphs2-1 leaves was not different from that in Ler wild-type leaves. It is worth mentioning that SHG Glc levels are presented as absolute amounts on a fresh weight basis, not as percentages of the total monosaccharides in SHG.

Figure 5.

Diurnal carbohydrate content in Atphs2-1 and Ler wild-type leaves. A, Maltose. B, Starch (μmol Glc equivalents g⁻¹ fresh weight). C, G1P. D, Suc. E, Glc. F, Fru. G, G6P. H, F6P. The last data point is a repeat of the first data point. Squares, Ler wild type; circles, Atphs2-1. White bars, Light period; black bars, dark period. Values are mean ± se (n = 5).

Figure 6.

Glc content in SHG from Atphs2-1, 35S:AtPHS2, and wild-type Arabidopsis leaves. A, Glucosyl residues in SHG from Atphs2-1 (bars with no lines) and corresponding Ler wild-type Arabidopsis (bars with hatched lines) leaves. B, Glucosyl residues in SHG from 35S:AtPHS2 (bars with crosshatched lines) and corresponding Ws wild-type Arabidopsis (bars with hatched lines) leaves. Leaves were harvested 4 h into the dark period of a 16-h light/8-h dark photoperiod. Plants used in A and B were grown at two different growth chambers. Asterisks over bars indicate Atphs2-1 and 35S:AtPHS2 samples that are significantly different from corresponding wild-type samples (Student's t test; P = 0.05). Values are mean ± se (n = 4). Values from 35S:AtPHS2 lines 4 and 12 were pooled.

Activities of Other Maltose-Metabolizing Enzymes in Atphs2-1 Plants

We measured the activities of other enzymes involved in maltose metabolism to test for any pleiotropic changes. The midday and midnight activities of DPE1, DPE2, β-amylase, and phosphoglucomutase (PGM) were not changed in the Atphs2-1 leaves relative to wild-type leaves (Tables I and II). The enzyme that could possibly interfere with measurement of DPE2 activity in crude Arabidopsis leaf extracts is maltase (Critchley et al., 2001), which hydrolyzes maltose to Glc. However, maltase activity in crude leaf extracts is so low that its contribution to Glc production in DPE2 activity assay is negligible. (1) In wild-type Arabidopsis leaves, the Glc-forming activity of maltase is approximately 7.3% of that of DPE1 (Chia et al., 2004). (2) Based on the relative activity of DPE1 and DPE2 in wild-type leaves (Table II), at night, the Glc-forming activity of maltase is about 5.2% of that of DPE2, well within the range of se for DPE2 activity (Table II). We also measured nighttime DPE2 activity on native glycogen gels. No significant changes of DPE2 activity were observed (data not shown). The average midday activity of hexokinase (HXK) was 20% higher in the Atphs2-1 leaves (Table I), statistically significant at the P = 0.1 level of confidence.

Overexpression of AtPHS2 Causes an Increase in Cytosolic α-Glucan Phosphorylase Activity and G1P Levels

To study the effects of overexpression of AtPHS2 in Arabidopsis, we measured the transcript levels of DPE2 and AtPHS2, and the activities of AtPHS2 and AtPHS1 in the leaves of 13 independent 35S:AtPHS2 lines (Fig. 7). Overexpression of AtPHS2 did not cause an increase in DPE2 expression, but caused an increase in AtPHS2 activity in Arabidopsis leaves. Both H1 and H2 forms of AtPHS2 activity were increased in most 35S:AtPHS2 lines, whereas the activity of AtPHS1, L1 form, was not increased. We also complemented Atphs2-1 plants with the same 35S:AtPHS2 construct. Both H1 and H2 forms of AtPHS2 activity reappeared in Atphs2-1 plants complemented with 35S:AtPHS2. Both the transcript level and the activity of AtPHS2 in 35S:AtPHS2 lines 4 and 12 were substantially increased. The nighttime G1P level in 35S:AtPHS2 lines 4 and 12 was about twice as much as that in Wassilewskija (Ws) wild-type plants (Fig. 7). During the night, the Glc levels in the total SHG (SHG_T) and low-M_r SHG (SHG_S) from 35S:AtPHS2 lines 4 and 12 were not significantly different from those in Ws wild-type plants. However, the midnight SHG_L Glc content in 35S:AtPHS2 lines 4 and 12 was reduced by approximately 40% relative to that in Ws wild-type plants. It should be noted that SHG Glc levels in the two ecotypes of wild-type Arabidopsis were different (Fig. 6).

Figure 7.

Analysis of the 35S:AtPHS2 lines. A, Transcript levels of AtPHS2 and DPE2 in individual 35S:AtPHS2 lines. Each line was a result of independent insertion events. The values have been normalized to the transcript levels of actin 2. The values are the average of three PCR replicates. Standard errors were not shown because these are instrumental replicates, not biological replicates. Na, Not analyzed. B, AtPHS2 and AtPHS1 activities in the same 35S:AtPHS2 lines, resolved by native glycogen gels. Both forms of AtPHS2 activity, H1 and H2, were elevated in 35S:AtPHS2 lines. H, Cytosolic forms; L, plastidial forms. C, Midnight G1P levels in Ws wild-type and 35S:AtPHS2 lines 4 and 12. Values are mean ± se (n = 5).

DISCUSSION

Maltose Is the Preferred Substrate for DPE2

The increased amount of maltose in the DPE2-deficient Arabidopsis and potato plants suggested that maltose is metabolized by DPE2 (Chia et al., 2004; Lloyd et al., 2004; Lu and Sharkey, 2004). The chloroplast DPE, DPE1, does not use maltose (Takaha et al., 1993) and DPE2 will not use maltose alone. Here we show that recombinant DPE2 uses maltose in preference to other maltooligosaccharides when SHG or glycogen is present. It was proposed that DPE2 may use the cytosolic SHG as primers to metabolize maltose (Chia et al., 2004; Lu and Sharkey, 2004). Here we confirm that recombinant DPE2 catalyzes the reaction with maltose plus SHG in vitro (Fig. 1; Fettke et al., 2006). The highest activity of DPE2 was found with glycogen or SHG together with maltose. The substrate preference is ideally suited to the role of DPE2 as proposed. Fettke et al. (2006) show that DPE2 will also catalyze the reverse reaction using Glc and other pyranoses as acceptors of Glc from SHG, yielding disaccharides. The normal physiological direction of the reaction catalyzed by DPE2 must be consumption of maltose given the very high accumulation of maltose in dpe2 mutants.

Both DPE2 and AtPHS2 Preferred Highly Branched Glycans

DPE2 has a high affinity for oyster glycogen on native glycogen gels, suggesting that glycogen could be the other substrate for DPE2 (Chia et al., 2004). When assayed with various substrates, including SHG isolated from Arabidopsis leaves, recombinant DPE2 was found to prefer branched glycans, such as glycogen and SHG as the other substrate, besides maltose (Fig. 1), consistent with its hypothesized role in cytosolic metabolism (Lu and Sharkey, 2004).

Pho2 from spinach and pea leaves had a high affinity for highly branched glycans such as glycogen and SHG (Preiss et al., 1980; Conrads et al., 1986). Based on the migration rate of AtPHS2 on native glycogen gels (Zeeman et al., 2004), AtPHS2 also had a high affinity for glycogen. Recombinant Pho2 from fava bean was shown to transfer the glucosyl residues from G1P to SHG from Arabidopsis (Fettke et al., 2005a). However, we were interested to know whether recombinant AtPHS2 from Arabidopsis could convert the glucosyl residues in Arabidopsis SHG to G1P. By incubating recombinant AtPHS2 protein with Pi and various other substrates, we found that AtPHS2 does use SHG and Pi to make G1P (Fig. 2). Moreover, AtPHS2 prefers highly branched glycans such as glycogen and SHG over long, but less branched, dextrin. The activity of DPE2 and AtPHS2 with SHG isolated from Arabidopsis indicates that SHG could be the substrate for both DPE2 and AtPHS2 in vivo. Fettke et al. (2006) also presented evidence that DPE2 from Arabidopsis and Pho2 from fava bean act on a subfraction of SHG from Arabidopsis and that they prefer the same sites on SHG.

The Glc content of SHG in Arabidopsis lacking DPE2 was increased (Fettke et al., 2006). It is not clear why the SHG Glc content of dpe2 plants is increased, but the results of Fettke et al. (2006) do not support the earlier conclusion of Lu and Sharkey (2004), who reported that the increased Glc was from phytoglycogen coprecipitating with SHG. Fettke et al. (2006) hypothesized that the excess Glc associated with SHG in dpe2 plants is attached to alternate locations on the SHG. These locations were still accessible to amylase and AtPHS2, but presumably the catalytic efficiency was reduced, accounting for the accumulation of Glc on SHG in these plants. SHG isolated from dpe2 mutants showed very high activity with DPE2 in vitro, suggesting that the normal sites glucosidated by DPE2 are different from those glucosidated in dpe2 mutants and that the preferred sites are saturable. Because the maltose level in dpe2 plants is about 100 times higher than in wild-type plants, it is possible that there are nonphysiological reactions adding Glc from maltose to SHG in linkages that are less favorable for AtPHS2 activity. Further linkage analysis and characterization of the specific SHG accepting Glc during maltose metabolism is needed for all plants used to analyze this pathway. Fettke et al. (2006) have made important progress in this direction by showing that AtPHS2 and DPE2 use the same sites on SHG.

The free pool of Suc in Ler wild type was 0.5 to 3 μmol/g fresh weight (Fig. 5), whereas the glucosyl residues of SHG_L in Ler wild type were around 0.015 μmol/g fresh weight (Fig. 6). Considering these amounts and the relatively stable and constant nature of the SHG pools, one would postulate very rapid turnover of Glc units and tight metabolic control. Although SHG may not be directly involved in starch degradation and Suc synthesis, SHG is important to both pathways. SHG is involved in metabolizing a starch degradation product—maltose—and in producing precursors for Suc synthesis. Disruption of maltose and SHG metabolism in dpe2 mutants resulted in an increase in the starch and Glc content of SHG and a decrease in nighttime Suc (Chia et al., 2004; Lu and Sharkey, 2004; Fettke et al., 2006). The reactions involving SHG could be the bridge between starch degradation and Suc synthesis.

AtPHS2 Is Involved in Maltose Metabolism in the Cytosol

Pho2 was proposed to phosphorylate the glucosyl residues in SHG and yield G1P, analogous to MalP in E. coli (Lu and Sharkey, 2004, 2006). We observed a 4-fold increase in nighttime maltose levels in Atphs2-1 leaves (Fig. 5). To confirm whether this was caused by T-DNA insertion in the AtPHS2 gene, we measured the activity of AtPHS1 and AtPHS2 on native glycogen gels. We also measured the total activity of α-glucan phosphorylase with glycogen and maltoheptaose using NADP(H)-linked assays. Because AtPHS1 and AtPHS2 have different affinities for glycogen, their activities could be distinguished on native glycogen gels. No residual activity of AtPHS2 was observed on the gels (Fig. 4). NADP(H)-linked enzymatic assays on a filterphotometer are quantitative. But the activity of AtPHS1 and AtPHS2 could not be easily distinguished from each other if the substrates used are not absolutely specific. In fact, we observed about 18% of residual midday activity of α-glucan phosphorylase in Atphs2-1 when the substrate was glycogen (Table I). This residual activity decreased from 1.69 ± 0.21 μmol G1P g⁻¹ protein min⁻¹ during the day to 0.28 ± 0.17 μmol G1P g⁻¹ protein min⁻¹ at night as the activity with maltoheptaose (G₇) decreased from 5.81 ± 0.41 μmol G1P g⁻¹ protein min⁻¹ during the day to 2.18 ± 0.55 μmol G1P g⁻¹ protein min⁻¹ at night. This suggests that the residual activity with glycogen in the Atphs2-1 mutant is indeed contributed by AtPHS1, the plastidial α-glucan phosphorylase, and that the plastidial α-glucan phosphorylase is more active during the day than at night.

The T-DNA insertion in AtPHS2 did not cause any pleiotropic increase of β-amylase activity or decease of DPE2 activity (Tables I and II). We conclude that AtPHS2 is involved in maltose metabolism in the cytosol just as Fettke et al. (2006) concluded for the Vicia homolog. However, the increase in maltose content in Atphs2-1 leaves is much less compared with that in dpe2 leaves (Chia et al., 2004; Lu and Sharkey, 2004). This may partially explain why the starch content did not increase in Atphs2-1 mutants. The amount of Glc in total SHG was not increased in Atphs2-1 plants (Fig. 6). However, the glucosyl residues in SHG_L on a fresh weight basis was increased by 40% in Atphs2-1 leaves and decreased by 40% in 35S:AtPHS2 lines (Fig. 6). The quantitative changes in SHG_L Glc content in AtPHS2-deficent and AtPHS2-overexpression lines may reflect changes in the length of the outer chains of the SHG_L subfraction. This is consistent with the proposed role of AtPHS2 in removing Glc units from SHG.

It was previously stated that a T-DNA line of the AtPHS2 gene had an embryo-lethal phenotype in the homozygous condition (Sharkey et al., 2004). However, the embryo-lethal phenotype could not be complemented with AtPHS2 under the control of the endogenous AtPHS2 promoter, the embryo-specific AGL15 promoter (Heck et al., 1995), or the constitutive 35S promoter (Y. Lu and T.D. Sharkey, unpublished data). On the other hand, wild-type Arabidopsis plants transformed with the same constructs had increased AtPHS2 transcript level and activity. The segregation pattern of Basta resistance (as an indication of T-DNA insertion) in the F2 population of the embryo-lethal mutant suggested that a second T-DNA insertion existed (Lu and Sharkey, unpublished data). Together with data reported here, we conclude that embryo lethality was unrelated to AtPHS2.

There are two AtPHS2 activity forms resolved by native glycogen gels (Fig. 4). Both forms were absent in Atphs2-1 plants and were increased in 35S:AtPHS2 lines. Whereas it is possible that one of the observed activities results from pleiotropic effects, the loss of both forms with the loss of the gene and the increase of both forms with the overexpression of the gene are a strong indication that both activities result from the same gene. Two activity forms of Pho2 were also observed in wheat (Triticum aestivum; Schupp and Ziegler, 2004). The glycogen- or SHG-binding sites in a certain population of AtPHS2 enzyme molecules may be sterically blocked (Schupp and Ziegler, 2004). Thus, some AtPHS2 molecules showed reduced affinity toward glycogen on native gels and two bands of AtPHS2 were observed (Fig. 3). Nevertheless, it is clear that AtPHS2 is involved in cytosolic maltose metabolism.

Direction of the Reaction Catalyzed by AtPHS2 in Vivo

The reaction catalyzed by Pho2 (e.g. AtPHS2) is bidirectional; the ratio of Pi/G1P as well as the concentration of SHG substrates in the cytosol determines the direction of the reaction. In the presence of Pi, SHG is used by Pho2 as a glucosyl donor. In the presence of G1P, SHG is used by Pho2 as a glucosyl acceptor (Fettke et al., 2005a). Recombinant Pho2 from fava bean converted the glucosyl group in G1P to the glucosyl groups in SHG in vitro (Fettke et al., 2006). In this article, we show that recombinant AtPHS2 converted the glucosyl residues in SHG to G1P in the presence of excess Pi in vitro. We are convinced that the reaction catalyzed by AtPHS2 is in the direction of making G1P in vivo, as is normally the case for glucan phosphorylases. (1) In the dark, the Pi concentration in the leaf cytosol is about 700 nmol mg⁻¹ chlorophyll (Chl) (Sharkey and Vanderveer, 1989). (2) During the day, the G6P concentration in the leaf cytosol is 54 nmol mg⁻¹ Chl (Sharkey and Vassey, 1989). (3) Because the G1P level is one-fifth or less of the G6P level in the cytosol, and because the average nighttime G1P level is not significantly different from the average daytime G1P level (Lu et al., 2005), we estimate that there are 10 nmol mg⁻¹ Chl or less G1P in the cytosol at night, on average. This indicates that, in vivo, AtPHS2 would convert the glucosyl residues in SHG to G1P rather than the reverse. This is further supported by the fact that the nighttime G1P level was increased and the SHG Glc content was decreased in the 35S:AtPHS2 lines (Fig. 7).

Fate of Glucosyl Residues in SHG in the Absence of AtPHS2

Maltose increased up to 100-fold in dpe2 mutants (Chia et al., 2004; Lu and Sharkey, 2004), whereas in Atphs2 mutants, maltose only increased up to 4 times (Fig. 5). From native glycogen gels and NADP(H)-linked activity assays, we are convinced that there was no other cytosolic α-glucan phosphorylase activity in Atphs2 mutants. The substantial difference in maltose levels in dpe2 and Atphs2 leaves indicated that an alternative route for Suc synthesis from maltose must exist. The high level of maltose and significant growth effect of loss of DPE2 activity indicate that the pathway up to and including DPE2 is the major pathway for carbon conversion from starch to Suc. The metabolism that bypasses AtPHS2 requires a product of DPE2. The most likely candidate is the SHG. Possibly some enzyme can release Glc from SHG, perhaps by hydrolysis. Elevated HXK activity may be required to convert extra Glc to G6P. Indeed, we observed a 20% increase of daytime HXK activity in Atphs2-1 plants (Table I). Although the 20% increase in HXK activity does not justify the proposal of an alternative cytosolic path, the substantial difference in maltose levels between dpe2 and Atphs2 leaves does. Increased Glc levels in the Atphs2 mutant (Fig. 5E) also indicate an alternative way to liberate free hexoses. Higher hexoses and HXK activity also lead to higher G6P levels (Fig. 5G). Compared with the route via AtPHS2, a hydrolytic route and HXK will cost one extra ATP per Suc unit.

Comparison of Maltose Metabolism in E. coli to the Arabidopsis Pathway

MalQ in E. coli uses maltose and linear maltodextrins as substrates (Boos and Shuman, 1998). MalQ-deficient E. coli mutants are not only Mal⁻, but also their growth is inhibited by maltose (Hofnung et al., 1971). In the presence or absence of external maltose, malQ mutants accumulate large amounts of free maltose inside the cell (Szmelcman et al., 1976). If DPE2 is analogous to MalQ (Lu and Sharkey, 2004), then DPE2 should partially complement the Mal⁻ phenotype of the malQ mutant, MH70. We found that the DPE2-transformed MH70 strain grew on maltose and that the growth was not as fast as the parent strain MC4100 carrying the empty vector. We proposed that DPE2 is an ortholog to MalQ, but there are differences in their functions in maltose metabolism.

It was reported that malP⁻ malQ⁺ mutants growing on maltose become very long and large, are filled with a long and linear glucan—amylose—and stain blue with iodine (Schwartz, 1965, 1967; Boos and Shuman, 1998). It should be noted that a proportion of malP⁻ malQ⁺ cells were of normal sizes and these cells coexisted with giant cells in the same medium (Schwartz, 1967). These mutants cannot efficiently degrade maltodextrins produced by MalQ because they lack MalP. We observed that some DPE2-transformed MalQ⁻ cells were very long and large and accumulated long glucan chains when growing on maltose (Fig. 3). These cells grew slower on maltose than MC4100 cells carrying the empty vector (Fig. 3). We also observed that normal-size cells coexisted with giant cells in DPE2-transformed MalQ⁻ culture growing in liquid M63-maltose medium (Fig. 3). We speculate that the normal-size cells may survive by obtaining Glc produced by neighboring giant cells.

The above findings may reflect the difference in substrate preference for DPE2 and MalQ proposed by Lu et al. (2006). We speculate that endogenous MalP cannot metabolize long glucan chains produced by DPE2 as fast as short maltodextrins produced by MalQ. This may explain the slower growth and accumulation of long glucan chains in DPE2-transformed MalQ⁻ cells. Although DPE2 and MalQ both use maltose as one of their substrates, DPE2 from Arabidopsis can only partially complement the MalQ⁻ E. coli strain. We compared the similarities and differences of maltose metabolism in Arabidopsis and E. coli (Fig. 8). In E. coli, MalP is not as important as MalQ for growth on maltose: MalQ⁻ E. coli strains cannot grow on maltose (Hofnung et al., 1971; Szmelcman et al., 1976), whereas MalP⁻ and MalQ⁺ E. coli strains grow on maltose (Schwartz, 1965, 1967). In Arabidopsis, AtPHS2 is not as essential as DPE2 for maltose metabolism: dpe2 mutants accumulate more maltose than Atphs2 mutants (Chia et al., 2004; Lu and Sharkey, 2004). As we discussed earlier, an alternative route may exist to convert the glucosyl residues in SHG to precursors for Suc synthesis (Fig. 8).

Figure 8.

Schemes for maltose metabolism in Arabidopsis and E. coli. Top, Maltose conversion to Suc in Arabidopsis; bottom, maltose metabolism in E. coli. A hypothetical hydrolysis reaction to release Glc from SHG (or heteroglycan) is shown in gray with a question mark.

DPE2 has also been found in potato (AAR99599) and rice (Oryza sativa; BAC22431); Pho2 has also been found in potato (A40995), sweet potato (Ipomoea batabas; AAK01137), fava bean (T12091), and wheat (AAF82787); SHG has also been found in pea (Yang and Steup, 1990; Fettke et al., 2004), spinach (Yang and Steup, 1990), and potato (Fettke et al., 2005b) leaves. Besides DPE2 in Arabidopsis, DPE2 in potato, stDPE2, has also been characterized (Lloyd et al., 2004). Repression of stDPE2 leads to inhibition of starch degradation and an increase in maltose in leaves. It would be interesting to know whether repression of DPE2 in rice will also result in an increase in maltose and starch. Nevertheless, we speculate that the pathway for maltose metabolism in Arabidopsis may be widespread among higher plants. Bacteria, such as E. coli, have analogous enzymes for these pathways, except that the short maltodextrins are preferred to the heteroglycan of higher plants (Fig. 8).

In summary, by complementing a MalQ⁻ E. coli strain with Arabidopsis DPE2, we showed that DPE2 is an ortholog of MalQ and functions in the direction of maltose breakdown, not synthesis. However, MalQ prefers to use short maltooligosaccharides as the other substrate, besides maltose, and DPE2 prefers to use branched glycans. The in vitro assays of recombinant DPE2 and AtPHS2 indicated that SHG is the common substrate for both DPE2 and AtPHS2 in vivo. The increased amounts of nighttime maltose in Atphs2 leaves further proved that AtPHS2 is involved in maltose metabolism in Arabidopsis. However, the substantial difference in the maltose levels in dpe2 and Atphs2 mutants suggested that an alternative route to convert the glucosyl residues in SHG to the precursors for Suc synthesis must exist.

MATERIALS AND METHODS

Expression of DPE2 and AtPHS2 in Escherichia coli

Total RNA was extracted from wild-type Arabidopsis (Arabidopsis thaliana) leaves and reverse transcribed as described in Lu et al. (2005). DPE2 cDNA was amplified using Pfu DNA polymerase (Promega) with a forward primer 5′-ACACAGGATCCATGAATCTAGGATCTCTTTC-3′ (BamHI site underlined) and a reverse primer 5′-ACACACTCGAGTTATGGGTTTGGCTTAGTCG-3′ (XhoI site underlined). AtPHS2 cDNA was amplified with a forward primer 5′-ACACAGGATCCGCAAACGCCAATGGAAAAG-3′ (BamHI site underlined) and a reverse primer 5′-ACACAGTCGACTTAGGGAACAGGACAAGC-3′ (SalI site underlined). The resulting 2,887-bp (DPE2) and 2,545-bp (AtPHS2) PCR fragments were AT-cloned into a pGEM-T vector (Promega) and sequenced to check errors. BamHI/XhoI-digested DPE2 and BamHI/SalI-digested AtPHS2 were further subcloned into a pET28a expression vector (Novagen) and expressed in E. coli strain BL21 (DE3). Recombinant DPE2 and AtPHS2 proteins were purified using nickel nitrilotriacetic acid agarose columns (Qiagen). Glycerol and dithiothreitol (DTT) were added to the eluates to final concentrations of 30% (v/v) and 1 mm, respectively. Proteins from empty-vector pET28a were also purified using nickel nitrilotriacetic acid agarose columns as controls. The presence of recombinant proteins in the eluates was confirmed with western blots and the concentration was determined using the Bradford technique (Bio-Rad).

Substrate Specificities of Recombinant DPE2 and AtPHS2 Proteins

The activities of recombinant DPE2 and AtPHS2 proteins were measured using NADP(H)-linked assays in a Sigma ZFP 22 dual-wavelength filterphotometer (Sigma Instruments). To assay the activity of DPE2, phosphate buffer (pH 7.4) containing 50 mm KH₂PO₄, 20 mm KCl, 10 mm MgCl₂, 2 mm EDTA, 0.25% (v/v) Triton X-100, 0.5 mm NADP, 0.5 mm ATP, 1.25 units/mL G6P dehydrogenase (G6PDH; Sigma), and 1.25 units/mL HXK (Sigma) was used. To determine the substrate specificity for DPE2, various combinations of 2 mm maltodextrins G₂ to G₇, or 2 mg/mL dextrin, amylopectin, glycogen, or total SHG were used. The total SHG used was isolated from Arabidopsis leaves using a method described by Fettke et al. (2004).

To assay the activity of AtPHS2, a method modified from Zeeman et al. (2004) was used. The assay buffer (pH 7.4) contains 50 mm KH₂PO₄, 20 mm KCl, 10 mm MgCl₂, 2.5 μm Glc-1,6-bisP, 2 mm EDTA, 0.25% (v/v) Triton X-100, 0.5 mm NADP, 1.25 unit/mL G6PDH, and 5 units/mL PGM (Sigma). To determine the substrate specificity of AtPHS2, 1 mg/mL of dextrin, amylopectin, glycogen, or SHG isolated from Arabidopsis leaves was used. Among these substrates, dextrin is considered to have less branching than amylopectin, glycogen, or SHG.

Complementation of MalQ⁻ E. coli with DPE2 from Arabidopsis

E. coli strain MH70 (MalQ⁻) and its parent strain MC4100 were obtained from the E. coli Genetic Stock Center at Yale University (Casadaban, 1976; Peters et al., 2003). The pET28a vector containing DPE2 cDNA from Arabidopsis was digested with BamHI and XhoI. The released DPE2 fragment was cloned into BamHI/SalI-digested pQE30 (Qiagen). The resulting plasmids were cotransformed with pREP4 (Qiagen) into E. coli strain DH5α. A mixture of plasmids pQE30-DPE2 and pREP4 was isolated from positive colonies and transformed into E. coli strain MH70 and MC4100 to select for kanamycin/ampicilin resistance. A mixture of empty-vector pQE30 and pREP4 was cotransformed into MH70 and MC4100 as controls. Positive colonies were streaked on M63 plates with appropriate additives (25 μg/mL kanamycin, 50 μg/mL ampicilin, 0.5 μg/mL thiamin hydrochloride, 0.25 μg/mL casamino acids, 1 μm isopropylthio-β-galactoside, and 10 mm carbon source). The plates were incubated at 37°C for 24 h before photographing.

To test the accumulation of long glucans in E. coli, overnight Luria-Bertani cultures were harvested, washed, and resuspended with M63-maltose medium. M63-maltose plates inoculated with 2 μL of resuspended cells were incubated at 37°C for 18 h and were stained with 0.1% (v/v) I₂ and 1% (v/v) KI for 1 min before photographing. Growth of E. coli strains was compared at 37°C in liquid M63 medium containing 10 mm maltose and other appropriate additives as above. Cultures were inoculated in triplets to an initial cell density of 1 × 10⁵ colony forming units/mL. Bacterial growth was measured over time as turbidity using a spectrophotometer at 600 nm.

Growth of Plants and Isolation of Atphs2-1 Mutants

Wild-type Arabidopsis of ecotypes Ler and Ws and mutant lines were grown in a 16-h light/8-h dark photoperiod as described (Lu et al., 2006). The Atphs2-1 mutant (N169185) of Arabidopsis ecotype Ler was ordered from the Arabidopsis Biological Resource Center (Columbus, OH). T-DNA was inserted in the 12th intron of the AtPHS2 gene (At3g46970). Candidate mutant lines were genotyped using primers specific for the AtPHS2 gene (5′-GCAGTTCCCATGTTCTCTGTAAGGTCAGA-3′ and 5′-CCAAACAGGAAATCAGAAGGCTTATTGCT-3′) and the T-DNA (Ds3-1 primer, 5′-ACCCGACCGGATCGTATCGGT-3′; Sundaresan et al., 1995). The genotype of candidate homozygotes was confirmed by glycogen-containing native PAGE gels.

Protein Extraction and Glycogen-Containing Native PAGE Gels

Total soluble protein was extracted using a method modified from Häusler et al. (2000). The protein concentration was determined using the Bradford technique as modified by Bio-Rad Laboratories. A total of 12 μg of soluble protein per lane was separated on glycogen-containing native PAGE gels. Constituents and procedures for the preparation of the gels were as described (Häusler et al., 2000). The gels for α-glucan phosphorylases were washed in 100 mm sodium succinate and 0.05% (w/v) soluble potato (Solanum tuberosum) starch (pH 6.0) for 15 min and were incubated in 100 mm sodium succinate, 0.05% (w/v) soluble potato starch, and 20 mm G1P (pH 6.0) overnight at 25°C (Lu et al., 2006). The gels for DPE2 were washed in 100 mm Tris-HCl, 1 mm MgCl₂, 1 mm EDTA, and 1 mm DTT (pH 7.0) for 15 min and were incubated in 100 mm Tris-HCl, 1 mm MgCl₂, 1 mm EDTA, 1 mm DTT, and 5 mm maltose (pH 7.0) at 37°C for 2 h (Lu et al., 2006). The gels were stained with 0.67% (w/v) I₂ and 3.33% (w/v) KI solution.

Extraction and Measurements of Starch, SHG, and Other Carbohydrates

Leaf samples from Atphs2-1 and Ler wild-type plants were taken at different time points throughout one 16-h light/8-h dark cycle. Starch and soluble carbohydrates were extracted and the concentration of the carbohydrates was determined using NADP(H)-linked assays (Lu and Sharkey, 2004).

About 100 mg of leaf tissues were harvested from Atphs2-1, Ler wild-type, 35S:AtPHS2, and Ws wild-type Arabidopsis plants at 4 h into the dark period. Total SHG was extracted using a method modified from Fettke et al. (2004). Leaf samples were ground on dry ice, suspended in 5 μL/mg of ice-cold 20% (v/v) ethanol, and centrifuged. Supernatants were incubated at 95°C for 10 min and centrifuged to remove proteins. In Fettke et al. (2004), small molecules of carbohydrates were removed from SHG by dialysis against water (molecular weight cutoff 1 kD). In this work, small molecules of carbohydrates were removed by precipitation of SHG in 70% (v/v) ethanol and 1% (w/v) KCl (Yang and Steup, 1990). The solubility of SHG in different concentrations of ethanol and KCl was tested and pelleted SHG was not redissolved in 75% (v/v) ethanol. Pellets were washed twice with 75% (v/v) ethanol to remove residual ethanol-soluble carbohydrates. Pellets were reconstituted with 400 μL deionized water by incubation at 45°C for 30 min. After centrifugation, the supernatant was applied onto Centricon YM-10 columns (10,000 nominal molecular weight limit; Millipore), and centrifuged at 5,000g for 25 min to separate SHG_S (<10 kD) and SHG_L (>10 kD). Filtrates (SHG_S) and retentates (SHG_L) were hydrolyzed in 2 n HCl for 90 min at 100°C. We are interested in understanding metabolism of the glucosyl residues in SHG and the Glc contents in the hydrolysates were determined after neutralization. The total SHG for DPE2 and AtPHS2 activity assay was extracted from Ws wild-type Arabidopsis leaves on a larger scale and Econo-Pac 10 DG columns (Bio-Rad) were used to remove small compounds.

NADP(H)-Linked Activity Assays of Maltose-Metabolizing Enzymes

Total soluble proteins from Atphs2-1 and wild-type leaves were extracted using a method modified from Häusler et al. (2000). To assay the activities of maltose-metabolizing enzymes, phosphate buffer (pH 7.4) containing 50 mm KH₂PO₄, 20 mm KCl, 10 mm MgCl₂, 2 mm EDTA, 0.25% (v/v) Triton X-100, 0.5 mm NADP, 1.25 units/mL G6PDH (Sigma) was used. The ingredients were preincubated at room temperature for 5 min and the reactions proceeded at room temperature for 10 min after the addition of soluble protein extracts. Boiled soluble protein extracts were used as controls.

To assay the activity of β-amylase, 0.5 mm ATP, 1 mm maltoheptaose (as the carbohydrate substrate), 1.25 units/mL HXK (Sigma), 5 units/mL maltose phosphorylase (Kikkoman), and 5 units/mL maltose epimerase (Kikkoman) were also included in the assay buffer. The activity of β-amylase was calculated as the difference in the rate of maltose production before and after addition of the protein extracts.

To assay the activity of DPE1, 0.5 mm ATP, 2 mm maltotriose (as the carbohydrate substrate; Takaha et al., 1993; Zeeman et al., 1998), and 1.25 units/mL HXK were also included in the assay buffer. To assay the activity of DPE2, 0.5 mm ATP, 2 mm maltose, 2 mg/mL glycogen, and 1.25 units/mL HXK were also included in the assay buffer. The activities of DPE1 and DPE2 were calculated as the difference in the rate of Glc production before and after addition of the protein extracts.

To assay the activity of α-glucan phosphorylase, 1 mm maltoheptaose (or 1 mg/mL glycogen, as the carbohydrate substrate), 2.5 μm Glc-1,6-bisP, and 5 units/mL PGM (Sigma) were also included in the assay buffer. The activity of α-glucan phosphorylase was calculated as the difference in the rate of G1P production before and after addition of the protein extracts.

To assay the activity of HXK, 0.5 mm ATP and 2 mm Glc (as the carbohydrate substrate) were also included in the assay buffer. The activity of HXK was calculated as the difference in the rate of G6P production from Glc before and after addition of the protein extracts.

To assay the activity of PGM, 2 mm G1P (as the substrate) and 2.5 μm Glc-1,6-bisP was also included in the assay buffer. The activity of PGM was calculated as the difference in the rate of G6P production from G1P before and after addition of the protein extracts.

Overexpression of the AtPHS2 Gene in Arabidopsis

A full-length AtPHS2 gene containing a 3′-untranslated region was amplified using Pfu DNA polymerase (Promega) with a forward primer 5′-ACACATCTAGAAGTGCAAACGCCAATGGAAA-3′ (XbaI site underlined) and a reverse primer 5′-ACACAGGATCCAATCACTAACCCAAATTCAT-3′ (BamHI site underlined). Bacterial artificial chromosome clone F13I12 (Arabidopsis Biological Resource Center; Mozo et al., 1998) was used as the template. The resulting PCR product was AT cloned into pGEM-T and sequenced to check for errors. An XbaI/BamHI-digested AtPHS2 fragment was subcloned into a binary vector derived from pPZP221 (Hajdukiewicz et al., 1994). This vector contains an 800-bp cauliflower mosaic virus 35S promoter and a 260-bp polyadenylation signal from the nopaline synthase gene (NOS-ter) and both DNA fragments are derived from the PBI121 plasmid (Fang and Fernandez, 2002). The binary vector containing the AtPHS2 gene was mobilized into Argobacterium tumefaciens and was transformed into wild-type Ws Arabidopsis by the floral-dip method (Clough and Bent, 1998). Gentamycin-resistant plants were selected at the T1 generation and genotyped to verify transformation. Segregation of phenotypes was scored in the T2 generation; quantitative reverse transcription (RT)-PCR and native glycogen gels were used to check the overexpression level. RNA and protein samples were taken 1 h during the day.

Preparation of RNA and Quantitative RT-PCR

RNA samples from AtPHS2 overexpression lines were prepared and reverse transcribed as described in Lu et al. (2006). One RNA extraction per overexpression line was performed. Quantitative RT-PCR was performed on a Stratagene Mx3000P QPCR system with Brilliant SYBR Green master mix. The sequences of gene-specific primers are: 5′-TACGTCAACTGGAGCACCTC-3′ and 5′-TCATAGCATGAGCTGGAAGC-3′ for DPE2; 5′-CGCCAAGTACAGTCCACATT-3′ and 5′-CAAGCTCATAACCCAGCG TA-3′ for AtPHS2; 5′-CATCCAAGCTGTTCTCTCCT-3′ and 5′-CTTACAATTTCCCGCTCTGC-3′ for ACT2. Each PCR reaction was repeated three times. Threshold cycle values for DPE2 and AtPHS2 were normalized to those for ACT2.

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers: AtPHS2, At3g46970, NM_114564; DPE2, At2g40840, NM_129647; ACT2, At3g18780, NM_112764; malP, X06791; malQ, M32793.