Identification, subcellular localization and biochemical characterization of water-soluble heteroglycans (SHG) in leaves of Arabidopsis thaliana L.: distinct SHG reside in the cytosol and in the apoplast


  • Joerg Fettke,

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

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

    1. Max-Planck-Institute of Molecular Plant Physiology, Am Mühlenberg 1, D-14476 Potsdam-Golm, Germany
    Search for more papers by this author
  • Peter Geigenberger,

    1. Max-Planck-Institute of Molecular Plant Physiology, Am Mühlenberg 1, D-14476 Potsdam-Golm, Germany
    Search for more papers by this author
  • Martin Steup

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

(fax +49 331 977 2512; e-mail


Water-soluble heteroglycans (SHG) were isolated from leaves of wild-type Arabidopsis thaliana L. and from two starch-deficient mutants. Major constituents of the SHG are arabinose, galactose, rhamnose, and glucose. SHG was separated into low (<10 kDa; SHGS) and high (>10 kDa; SHGL) molecular weight compounds. SHGS was resolved into approximately 25 distinct oligoglycans by ion exchange chromatography. SHGL was further separated into two subfractions, designated as subfraction I and II, by field flow fractionation. For the intracellular localization of the various SHG compounds several approaches were chosen: first, leaf material was subjected to non-aqueous fractionation. The apolar gradient fractions were characterized by monitoring markers and were used as starting material for the SHG isolation. Subfraction I and SHGS exhibited a distribution similar to that of cytosolic markers whereas subfraction II cofractionated with crystalline cellulose. Secondly, intact organelles were isolated and used for SHG isolation. Preparations of intact organelles (mitochondria plus peroxisomes) contained no significant amount of any heteroglycan. In isolated intact microsomes a series of oligoglycans was recovered but neither subfraction I nor II. In in vitro assays using glucose 1-phosphate and recombinant cytosolic (Pho 2) phosphorylase both SHGS and subfraction I acted as glucosyl acceptor whereas subfraction II was essentially inactive. Rabbit muscle phosphorylase a did not utilize any of the plant glycans indicating a specific Pho 2–glycan interaction. As revealed by in vivo labeling experiments using 14CO2 carbon fluxes into subfraction I and II differed. Furthermore, in leaves the pool size of subfraction I varied during the light–dark regime.


The majority of the plant species possesses the capacity to synthesize and to degrade water-insoluble starch granules (Ball and Morell, 2003; Tetlow et al., 2004). Two types of polyglucan molecules form the starch particle: the highly branched amylopectin that often accounts for approximately 75% of the granule mass and, as a minor compound, the linear, largely unbranched amylose. The supramolecular arrangement of the amylopectin molecules results in a regular order of semi-crystalline and amorphous layers and is highly conserved (Pilling and Smith, 2003).

A remarkably low number of different glycosidic linkages occurs in the starch granules. Amylose possesses almost exclusively one type of intersugar linkages, namely the α-1,4-glycosidic bond which is also the predominant linkage in amylopectin. In addition, the latter contains α-1,6-interglucose bonds and, although to a far lesser extent, covalent glucose modifications such as phosphate esters that are linked to the C6 or C3 of glucosyl residues. Although phosphorylation is restricted to less than 0.3% of the glucosyl moieties, it appears to be of general importance for the degradation of transitory starch (Kötting et al., 2005; Ritte et al., 2002; Yu et al., 2001).

Despite the relatively simple chemical structure of both amylopectin and amylose, plant cells possess a wide range of enzymes that form or cleave starch-like interglucose bonds. Among these enzymes are starch synthase (EC, branching enzyme (EC, isoamylase-type debranching enzyme (EC, limit dextrinase (EC, α-amylase (EC, β-amylase (EC, glucan phosphorylase (EC, and glucanotransferase (EC Complexity of the enzymology of starch is further increased by multiplicity of most of these enzymes. Up to five isoforms of starch synthase are often expressed in the same tissue and two major groups of branching isozymes have been frequently reported to reside inside the plastid (Tetlow et al., 2004). In the genome of Arabidopsis thaliana L. three genes are annotated that encode plastidial isoamylase-like proteins (Delatte et al., 2005) and at least seven β-amylases have been identified as plastidial isoforms (Smith et al., 2005).

Unexpectedly, several starch-related enzymes are not restricted to the plastid but rather possess a dual intracellular location. These enzymes exist as distinct plastidial and extraplastidial isoforms (Lin et al., 1988a; Smith et al., 2005; Steup, 1990). Using immunocytochemistry one of the phosphorylase isozymes (designated as Pho 2) has been localized in the cytosol of mesophyll and parenchyma cells (van Berkel et al., 1991; Conrads et al., 1986; Schächtele and Steup, 1986). The Arabidopsis genome encodes for three α-amylase-like proteins, only one of which possesses a predicted transit peptide and has been observed in isolated chloroplasts (Yu et al., 2005). As revealed by cell fractionation of spinach (Spinacia oleracea L.) leaves, the glucan synthase (EC exists in the mesophyll protoplasts both as plastidial and non-plastidial isozymes (Tacke et al., 1991). The dual intracellular location of starch synthase, as described for spinach leaves, concurs with the fact that in A. thaliana L. one of the genes putatively encoding starch synthases does not possess a predicted targeting sequence (At5g65685) and, therefore, the gene product is likely to be located outside the plastid. Similarly, a transglucosidase (designated as DPE2; At2g40840) recently described for A. thaliana L. is detectable in mesophyll protoplasts but not in isolated chloroplasts and, therefore, has been attributed to the cytosolic compartment (Chia et al., 2004; Lu and Sharkey, 2004).

Under in vitro conditions all the extraplastidial isozymes act on starch-like polyglucans but in vivo they do not have access to the plastidial starch nor to any of the plastidial intermediates of starch metabolism and, therefore, their biochemical functions remained enigmatic.

However, the phenotype of the DPE2-deficient Arabidopsis mutant strongly suggests that the plastidial starch turnover is connected with a so far unknown cytosolic glycan metabolism. This mutant possesses a starch excess phenotype and, compared with wild-type leaves, an approximately 100-fold higher maltose level. Thus, the extraplastidial DPE2 isozyme appears to be essential for the cytosolic maltose turnover and, although indirectly, for normal starch metabolism (Chia et al., 2004; Lu and Sharkey, 2004; Weise et al., 2004). As a further complication, maltose has been shown to exist in plant cells both as α- and as β-anomer and the pool sizes of both anomeric forms are altered independently from each other (Weise et al., 2005).

During starch degradation plastids are thought to form maltose and to export it into the cytosol via the maltose transporter (Niityläet al., 2004). The DPE2-mediated maltose metabolism is assumed to proceed via a glucosyl transfer using maltose as donor and a cytosolic high molecular weight glucan or glycan as acceptor (Chia et al., 2004; Lu and Sharkey, 2004; Zeeman et al., 2004). Interestingly, in the DPE2-deficient Arabidopsis mutants the activity of the cytosolic phosphorylase isoform is increased three- to fourfold whereas that of the plastidial isozyme (Pho 1) is essentially unchanged (Chia et al., 2004).

Some years ago, water-soluble heteroglycans (SHG) have been identified that contain, as major constituents, arabinose, galactose and glucose and strongly interact with the cytosolic phosphorylase isozyme but are different from and more complex than starch-like glucans (Yang and Steup, 1990). By using affinity electrophoresis, the heteroglycans prepared from potato tubers were shown to interact selectively with cytosolic phosphorylase isoforms from several dicotyledon and monocotyledon species (Eckermann et al., 2004). Recently, an improved isolation protocol has been elaborated and evidence was presented that within the SHG several distinct glycans can be distinguished (Figure 1a; see also Fettke et al., 2004).

Figure 1.

Monomer pattern of water-soluble heteroglycans (SHG) from leaves of Arabidopsis thaliana L.
(a) Scheme of the SHG fractionation procedure.
(b) Equal amounts (6 μg each) of the hydrolyzed SHGT prepared from wild type (wt) or from two starch-deficient mutants (AGPase: deficient in the plastidial ADP-glucose pyrophosphorylase; p-PGM: deficient in the plastidial phosphoglucomutase) were applied to the HPAEC column.
(c) Equal amounts (6 μg each) of SHGL and SHGS (see text), both derived from SHGT that has been prepared from the leaves of wild-type Arabidopsis were applied to the HPAEC column.
Fuc: fucose, Ara: arabinose, Rha: rhamnose, Gal: galactose, Glc: glucose, Xyl: xylose, Man: mannose, wt: wild type.

The SHG were deprived of low molecular weight compounds (<1 kDa) and the resulting SHGT was then separated into two fractions, SHGL and SHGS. SHGS contains oligoglycans with an apparent molecular weight ranging from 1 to 10 kDa. By using field flow fractionation (FFF; Wyatt, 1993) SHGL was further resolved into subfraction I and II. In leaflets from Pisum sativum L. these two subfractions possess some similarities in their monomer patterns and glycosidic linkages but differ in several important features: Unlike subfraction II, subfraction I does not react with the β-glucosyl Yariv reagent (Gaspar et al., 2001; Yariv et al., 1962) and in in vitro assays it acts as glucosyl acceptor for the cytosolic phosphorylase (Pho 2) whereas subfraction II does not. Unlike subfraction II, subfraction I was recovered in isolated mesophyll protoplasts but it was not observed in isolated intact chloroplasts. Therefore, the conclusion was reached that the SHG-derived subfraction I resides in the cytosol of mesophyll cells (Fettke et al., 2004).

In fully expanded plant cells, the cytosol represents a relatively small cellular compartment. In mesophyll cells, it accounts for 3.8–7% of the total cell volume whereas the relative size of the vacuole is at least one order of magnitude higher ranging from 73 to 78% (Heineke et al., 1997). Furthermore, the cytosol is closely associated with various organelles (such as mitochondria, peroxisomes, endoplasmic reticulum or the Golgi apparatus) that are mechanically fragile and comprise only a very small proportion of the cellular volume but are highly divers in their biochemical functions and constituents. The relative cellular volume of the leaf mitochondria is reported to be, at most, 1% (Heineke et al., 1997) and that of the other compartments is even lower not permitting any estimation. However, for a precise intracellular localization of a given extraplastidial compound these minor compartments have to be taken into consideration. In particular, this is relevant if extraplastidial oligo- or polyglycans are to be localized. Whilst cellulose biosynthesis proceeds by an enzyme complex integrated into the plasmalemma non-cellulose cell wall-related glycans or glycoconjugates are formed and exported via the endomembrane system (Reiter, 2002; Scheible and Pauly, 2004). Unless transport is completed these compounds reside inside the endomembrane system and are difficult to distinguish from cytosolic constituents.

In this communication the subcellular location of soluble heteroglycans from leaves of A. thaliana L. has been studied using three complementary approaches: first, leaf material from A. thaliana L. was lyophilized, separated by the non-aqueous fractionation technique (Farréet al., 2001; Stitt et al., 1989) and in the resulting fractions various marker proteins were quantified. As a non-proteinogenic marker, crystalline cellulose was isolated from all non-aqueous fractions and was quantified. The distribution of SHG was determined by using the individual fractions separately as starting material for the SHG preparation. Subsequently, the glycan preparations were separated into different subfractions followed by quantification. Secondly, we isolated an intact intracellular microsomal fraction and used it as starting material for the glycan isolation. Subsequently, the glycan preparation obtained was quantified and analyzed. Thirdly, we prepared a fraction containing intact mitochondria and peroxisomes and analyzed whether or not glycans reside inside these organelles. By using these approaches clear evidence is obtained that the SHG-derived subfraction I (but not subfraction II) represents a pool of cytosolic heteroglycans that contain, as main constituents, arabinose and galactose.

As revealed by glycosidic linkage analysis, subfraction I and II share a relatively high percentage of arabinogalactan-like linkages. However, the glycosidic linkage pattern of subfraction I is more complex than that of subfraction II.

For a biochemical characterization of the SHG, we determined the interaction between recombinant cytosolic phosphorylase (Pho 2) and SHG. The photosynthesis-dependent carbon flux into SHG was analyzed by 14CO2 pulse labeling experiments performed with intact Arabidopsis plants. Furthermore, the SHG pool sizes in Arabidopsis leaves were monitored during the day–night cycle.


Complex glycans are formed independently of the presence of plastidial starch

Water-soluble heteroglycans were isolated from leaves of wild type and of two starch-deficient mutants of A. thaliana L. Leaves of non-flowering plants were harvested during the middle of the light period. Under these conditions, wild-type leaves possess transitory starch but the mutants are starch deficient as either the plastidial ADP-glucose pyrophosphorylase (AGPase) (EC; Lin et al., 1988b) or the plastidial phosphoglucomutase (p-PGM) (EC; Caspar et al., 1985; Gibon et al., 2004) is not functional. The SHG preparations from wild type and the two mutants were deprived of low molecular weight compounds (<1 kDa) by dialysis (Figure 1a; for details see Fettke et al., 2004) and the resulting total heteroglycan preparations (SHGT) were subjected to acid hydrolysis followed by the quantification of the monomer contents. For all three preparations, the amount of SHGT was approximately 150 μg glucose equivalents g−1 fresh weight (FW). Thus, the transitory starch levels of the starting material do not noticeably affect the yield of the SHGT preparations nor does the formation of SHG require a functional starch metabolism.

Heterogeneity and monomer patterns of the water-soluble heteroglycan preparation

The monosaccharide patterns of the three SHGT preparations were determined by HPAEC-PAD (Figure 1b). Galactose and arabinose were consistently found to be the most prominent sugars. Regarding the minor sugar constituents, the SHGT preparations of the two mutants contain more mannose than the wild type. Furthermore, the glucose content is relatively high in the phosphoglucomutase-deficient mutant but it is low in the other starch-deficient mutant and, therefore, does not reflect the starch level of the starting material.

The SHGT preparation from the leaves of wild-type Arabidopsis was separated into high (>10 kDa; SHGL) and low (<10 kDa; SHGS) molecular weight glycans by membrane filtration. SHGL and SHGS account for approximately 75 and 25% of SHGT. The two glycan samples differ in their monomer composition: SHGS contains almost equal amounts of arabinose, galactose and glucose whereas in SHGL galactose is clearly the most prominent sugar and glucose a very minor compound. Compared with SHGL, SHGS possesses relatively more glucose, rhamnose, xylose, and mannose (Figure 1c).

The SHGL fraction can be further separated into two subfractions either by treatment with the β-glucosyl Yariv reagent or by FFF (Figure 1a; see also Fettke et al., 2004). When separated by FFF, the first eluting subfraction (Figure 2a; elution volume 5–11 ml; subfraction I) represents a minor proportion of SHGL accounting for approximately 14% of SHGL. As determined by FFF-MALLS, the size of this subfraction ranges from 2.5 to 7 × 104 Da (Figure 2a). The molecular masses of subfraction II (eluted from 12 to 21 ml) are higher ranging from 9 × 104 to 2 × 105 Da. In the two starch-deficient mutants both subfraction I and II were present and the ratio was essentially unchanged (data not shown).

Figure 2.

Analysis of SHGT-derived glycans from leaves of Arabidopsis thaliana L. (wild type).
(a) Separation of SHGL into subfraction I (Sub I) and subfraction II (Sub II) by FFF. The molar mass distribution determined by MALLS-DRI (dots) and the refractive index values (solid lines) that reflect the abundance of the respective polymer are plotted against the elution volume.
(b) Monomer pattern of subfraction I and subfraction II as revealed by HPAEC-PAD. SHGT was separated into subfraction I (Sub I; elution volume 5–11 ml; see Figure 2a) and subfraction II (Sub II; elution volume 12–21 ml). Both subfractions were hydrolyzed prior to HPAEC-PAD analysis. Signals are normalized at arabinose. Abbreviations as in Figure 2.
(c) Native SHGS (20 μg) analyzed by HPAEC-PAD. A mixture of linear maltodextrins served as standard. DP: degree of polymerization.

In both subfraction I and II, arabinose and galactose are the two most prominent monomers. However, subfraction I contains relatively more rhamnose, glucose, xylose, and mannose (Figure 2b).

By using HPAEC-PAD the native low molecular weight glycans (SHGS) can be resolved into more than 25 distinct oligosaccharides (Figure 2c) that elute during 20–90 min. When compared with a chromatogram of a mixture of linear maltooligosaccharides, SHGS appears to cover a degree of polymerization up to approximately 30. However, chromatographic characteristics are also affected by features such as degree of branching, charge or monomeric composition and, therefore, the DPs of SHGS may deviate to some extent from those of the maltodextrin standard.

Localization of SHG by non-aqueous fractionation of Arabidopsis leaf material

Leaves from A. thaliana L. wild-type plants were harvested during the light period and were subjected to non-aqueous fractionation. The marker enzyme activities that were quantified using photometric assays are listed in Table 1. In addition, some marker proteins were quantified following polyacrylamide gel electrophoresis (PAGE) performed either under native or denaturing conditions: the plastidial glucan, water dikinase (EC and the cytosolic phosphoenol pyruvate carboxylase (EC were quantified following SDS-PAGE and Western blotting. The cytosolic DPE2 (EC and the phosphorylase (EC isozymes were monitored by native PAGE using glycogen-containing separation gels (affinity electrophoresis) and activity staining (Figure 3a).

Table 1.  Distribution of the SHGT-derived heteroglycans and of marker enzyme activities during non-aqueous fractionation of leaf tissue. As a cell wall marker, crystalline cellulose has been included
Compound/markerDistribution (%)
  1. UGPase, UDP-glucose pyrophosphorylase; AGPase, ADP-glucose pyrophosphorylase; GAPDH, glyceraldehyde 3-phosphate dehydrogenase.

AGPase (plastidial)68172743
GAPDH (plastidial)68172743
UGPase (cytosolic)3621161611
Subfraction I27.718.723.615.414.6
Glucosidase (vacuolar)541811107
Galactosidase (vacuolar)521911108
Crystalline cellulose (cell wall)52.426.
Subfraction II49.621.814.28.56
14C-glucosyl incorporation by Pho 2 into SHGL24.724.122.614.514.1
Figure 3.

Analysis of marker proteins or marker enzyme activities following non-aqueous fractionation of leaf tissue from Arabidopsis thaliana L. (a). In addition, the phosphorylase isoforms in an aqueous leaf extract were resolved (b). Equal volumes of buffer-soluble proteins of each of the five apolar fractions (P, 0–3) were subjected to native PAGE and activity staining (line a and b) or to Western blotting following SDS-PAGE (line c and d). Open arrows: plastidial markers; closed symbols: cytosolic markers. (b) 100 μg of buffer-soluble proteins were applied per 0.5 cm gel width. Following native PAGE, lanes were incubated with or without glucose 1-phosphate followed by iodine staining. Phosphorylase bands having an high, medium, and low apparent affinity toward glycogen are marked with I, II, and III, respectively. (a) Line a: DPE2; line b: phosphorylase isoforms; line c: PEPCO; line d: glucan, water dikinase; (b) zymogram of the phosphorylase activity from an aqueous leaf extract.

DPE2 exhibits a high affinity toward the immobilized glycogen and is essentially immobile during affinity electrophoresis. The highest proportion of the enzyme activity was recovered in the pellet fraction (P). Activity declined moderately with increasing distance from the pellet (Figure 3a, line a). A similar distribution was revealed by Western blotting for another cytosolic marker protein, the phosphoenol pyruvate carboxylase (line c) and by the photometric assay for the UDP-glucose pyrophosphorylase (UGPase) activity (Table 1). In glycogen-containing separation gels the phosphorylase pattern consisted of three forms of activity. The immobile form and that having an intermediate mobility are equally distributed in the apolar gradient (Figure 3a, line b; for identification of the three isoforms see below). Distribution of the mobile highly phosphorylase isoform is similar to that of plastidial marker enzymes/proteins: in fractions P and 0 these markers were below the limit of detection whereas in fraction 2 the highest amounts were recovered (Figure 3a, line b and d; Table 1).

The phosphorylase pattern (Figure 3a) consists of three bands of activity all of which are strictly depending upon glucose 1-phosphate (Figure 3b). In the Arabidopsis genome only two genes (At3g46970 and At3g29320) are known to encode for the cytosolic and the plastidic phosphorylase isoform, respectively, and, therefore, the zymogram was unexpectedly complex. The same zymogram was also obtained with freshly prepared aqueous extracts of Arabidopsis leaves (Figure 3b) and, consequently, an artifact due to the non-aqueous fractionation can be excluded. For identification, the three bands containing phosphorylase activity (designated as I, II, and III according to their increasing electrophoretic mobilities) were excised from the separation gel. Following denaturation and SDS-PAGE, proteins derived from each of the three gel slices were identified by trypsination, MALDI-MS analysis and database search. In both band I and II the cytosolic phosphorylase (Pho 2) was identified (sequence coverage at least 40%) which concurs with the distribution in the apolar gradient. The plastidial (Pho1) phosphorylase was detectable only in the highly mobile form, band III. In the latter gel slice, Pho 2 was undetectable. Pho 2 derived from band I and II had exactly the same electrophoretic mobility. Thus, the Pho 2 isozyme appears to exist in two states that both are located in the cytosol but differ in their apparent affinity toward glycogen. A similar observation has been recently described for the phosphorylase isoforms from Triticum aestivum L. (Schupp and Ziegler, 2004). The molecular basis of the heterogeneity of Pho 2, as revealed by affinity electrophoresis, is currently under investigation.

For the localization of the SHG-derived glycans each of the five fractions of the apolar gradient was used as starting material for the isolation and subsequent fractionation of the heteroglycans. In addition, from each fraction crystalline cellulose was isolated. For two reasons the β-1,4-polyglucan was included: first, it is an unambiguous marker for the extracellular compartment which, until now, has not been analyzed by the non-aqueous fractionation technique. Secondly, the use of a high molecular weight polysaccharide marker was desired because no information is available as to whether or not hydrophilic polymers, such as polysaccharides, are similarly distributed in the apolar gradient, as do amphiphilic polymers, such as proteins.

Crystalline cellulose prepared from each of the five apolar fractions was quantified following acid hydrolysis. Each SHGL preparation derived from each apolar fraction was resolved into subfraction I and II by using field flow fractionation (Figure 4a). Subsequently, each of the two subfractions was quantified following exhaustive acid hydrolysis. Likewise, an aliquot of each SHGS preparation was quantified and another aliquot was analyzed by HPAEC-PAD to determine the oligosaccharide patterns (Figure 4b; Table 1). All low molecular weight glycan (SHGS) preparations comprise a complex oligosaccharide pattern.

Figure 4.

Analysis of SHG-derived glycans isolated from each of the five apolar fractions.
(a) FFF analysis of SHGL. All SHGL preparations were balanced to equal content of monomers. The refractive index is plotted against the elution volume. The apolar fractions used as starting material for the SHGL isolation are labeled P (pellet) and 0–3.
(b) Native SHGS preparations from each of the five apolar fractions analyzed by HPAEC-PAD.

As shown in Figure 4(a) (see also Table 1), the SHG-derived subfractions I and II clearly do not coincide during the apolar gradient centrifugation. The ratio between subfraction I and subfraction II is high in the top fraction (fraction 3) and decreases gradually in the four other fractions of the gradient. In the pellet fraction (P) subfraction II is clearly dominant and subfraction I is recovered as a shoulder of the major peak (i.e. subfraction II). It should be noted that FFF analyses were performed by using equal amounts of SHGL from each preparation. The ratio between subfraction I and II varies depending upon the fraction of the apolar gradient that has been used as starting material. However, for the determination of the subcellular distribution of the two subfractions the total amount of each subfraction is relevant (and not the ratio between both).

The total amounts of the various carbohydrates (subfraction I and II, SHGS and crystalline cellulose) were quantified and compared with the activities of the marker enzymes (Table 1). From these data, the subcellular distribution of the various glycans was calculated according to Riens et al. (1991) (Table 2). All SHG-derived glycans are essentially located outside the plastids. The vast majority of both SHGS and subfraction I reside in the cytosol whereas subfraction II is recovered in the vacuolar or apoplastic compartment. These data concur with results recently obtained by an aqueous fractionation of leaflets of P. sativum L. Isolated pea mesophyll protoplasts contained both SHGS and subfraction I but not subfraction II whereas in the isolated intact chloroplasts both SHGL and SHGS were essentially undetectable (Fettke et al., 2004). In Arabidopsis leaves, subfraction I closely coseparates with the cytosolic marker, UGPase, Pho 2 and PEPCase, and does clearly not coseparate with subfraction II which exhibits a distribution similar to that of vacuolar or apoplastic markers. As the distribution of cellulose coincides with that of both α-galactosidase and α-glucosidase the non-aqueous fractionation does not permit any distinction between these subcellular sites. As subfraction II was not detected in the pea protoplasts and vacuolar arabinose/galactose-containing glycans have never been described we propose an extracellular location of subfraction II. This assumption is also supported by the similarities of both pea and Arabidopsis subfraction II in the monomeric composition, the glycosidic linkages and the Yariv reagent reactivity (see below and Fettke et al., 2004).

Table 2.  Subcellular distribution of the glycans in Arabidopsis leaves. The subcellular distribution was calculated by comparing the glycan and marker enzyme distributions using a three-compartment calculation program
 Distribution (%)
SHGSSub ISub IICrystalline cellulose

It should be noted that the quantitative determination of the subcellular distribution of the various glycans (Table 2) is affected by some uncertainties that are inherent to the methods applied:

  • 1Using FFF no baseline separation of subfraction I and II is achieved. Due to some overlapping, the accuracy of the quantification of both glycans is limited.
  • 2All glycans were quantified using acid hydrolysis and, subsequently, a standard assay that estimates the concentration of reducing groups. However, the molar absorbance coefficient varies significantly between various monosaccharides (Waffenschmidt and Jaenicke, 1987) and, therefore, differences in the monosaccharide patterns affect the accuracy of the glycan quantification.
  • 3For calculation, a glucose standard was used and all data are given as ‘μg glucose equivalents’. Consequently, glycans containing a significant proportion of pentoses/pentuloses are, to some extent, overestimated.
  • 4No information is available as to whether hydrophilic polysaccharides and amphiphilic proteins do possess exactly the same distribution when subjected to centrifugation in an apolar gradient.

Despite these limitations, the data presented here clearly show that the vast majority (if not all) of both SHGS and subfraction I reside(s) in the cytosol whereas most (if not all) of subfraction II is located outside this compartment.

Localization of SHG by isolation of intact microsomes and organelles followed by heteroglycan isolation

The non-aqueous fractionation offers the advantage of rapid quenching of the tissue and avoiding redistribution of mobile compounds during sample processing. However, some minor compartments – such as mitochondria, peroxisomes, endoplasmic reticulum or the Golgi apparatus – cannot be analyzed by this technique. As many cell wall-related glycosyl transfer reactions are associated with the endomembrane system this limitation could be crucial if an extraplastidial glycan metabolism is to be analyzed. In order to identify the precise intracellular location of the soluble heteroglycans, intact microsomes were prepared from leaves of wild-type plants of A. thaliana L. and used as starting material for the isolation of target glycans. Intactness of the isolated vesicles derived from the endoplasmic reticulum and the Golgi apparatus was determined by using an antibody directed against the luminal BIP protein and by measuring the latent UDPase activity, respectively. Cytosolic contaminations were estimated by using an antibody raised against PEPCO (Figure 5a). Signal intensities were quantified using the AIDA program (Table 3). Approximately 90% of the latent UDPase activity was recovered inside the microsomes. A similar distribution was observed for BIP whereas most of the PEPCO activity was observed outside the microsomes. Thus, the microsomal preparation used is reasonable intact and lacks significant amounts of cytosolic contaminants. The pelleted microsomal preparation and the microsomal supernatant were used as starting material for the SHG isolation. The SHG preparation was separated into compounds having an apparent size <10 and >10 kDa and both glycan preparations were quantified separately. For both samples, the vast majority of the glycans was obtained from the supernatant (Table 3). The >10 kDa glycans isolated from either the microsomal preparation or the supernatant exhibit essentially the same monomer pattern (Figure 6a). This strongly suggests that the low amount of the >10 kDa glycans obtained from the microsomal fraction are due to cytosolic and/or apoplastic contaminations but are not indicative for a distinct pool of microsomal glycans. As revealed by FFF the >10 kDa glycans derived from the supernatant contain both subfraction I and II (Figure 6b) whereas the small amount of the >10 kDa glycans obtained from the microsomal fraction is below the limit of detection when analyzed by FFF (Figure 6b).

Figure 5.

Western blotting analysis of the microsome and organelle preparations.
(a) Buffer-soluble proteins from a pelleted microsome preparation (M) and from the corresponding supernatant (S) were separated by SDS-PAGE and transferred to nitrocellulose. For detection, a polyclonal antibody directed against a marker protein of the lumen of the endoplasmic reticulum, BIP (open arrow) or against a cytosolic marker, PEPCO (closed arrow) was used.
(b) Buffer-soluble proteins from a pelleted preparation of mitochondria and peroxisomes (O) and the corresponding supernatant (S) were analyzed using an antibody against a cytosolic marker protein, PEPCO (closed arrow).

Table 3.  Characterization of microsome and organelle preparations. Distributions of the marker enzymes are given in percentage. Carbohydrate amounts are given as percentage of the sum of the glycans derived from the pellet and the supernatant
Microsome preparation
 Latent UDPase928
  1. UDPase, UDP phosphatase; BIP, binding protein; PEPCO, phosphoenol pyruvate carboxylase; n.d., not detected.

Organelle preparation
 Citrate synthase82.417.5
Figure 6.

Analysis of glycans prepared from isolated microsomes.
(a) Monomer composition of soluble glycan fractions obtained from the pelleted microsome preparation (M) and the corresponding supernatant (S). Each glycan preparation was separated into a >10 and <10 kDa fraction and hydrolyzed. From each hydrolysate 5 μg carbohydrates were applied to the HPAEC column. Abbreviations as in Figure 1.
(b) FFF analysis of the >10 kDa glycan fractions prepared from the pelleted microsomes (M) and the corresponding supernatant (S). The refractive index values (abundance) are plotted against the elution volume.
(c) HPAEC-PAD analysis of the native <10 kDa glycan fractions obtained from the pelleted microsome preparation (M) and the corresponding supernatant (S). From each fraction, 20 μg carbohydrates were applied to the HPAEC column.

Unlike the higher molecular weight glycans, the <10 kDa compounds derived from the microsomal preparation differ significantly from those of the supernatant. The monomer patterns of the two glycan preparations are different (Figure 6a) and resolution of the native oligosaccharides by HPAEC-PAD results in two entirely different patterns (Figure 6c). In the <10 kDa glycans derived from the microsomal preparations, a few oligosaccharides are prominent. These compounds elute at 44.5 min and between 60 and 80 min. They are, however, minor compounds in the supernatant-derived sample. These data indicate that the isolated microsomal vesicles are reasonably intact. More importantly, not all of the constituents of SHGS reside in the same compartment. Only the earlier eluting oligosaccharides can be attributed to the cytosol. In contrast, the last eluting compounds occur inside the isolated membrane vesicles. These compounds do, however, represent only a small proportion of SHGS.

From isolated mitochondria and peroxisomes which intactness was analyzed by determining citrate synthase (EC and catalase (EC activity, only traces of SHGT were isolated (Figure 5b; Table 3). Therefore, we exclude that in vivo SHGT or any SHGT-derived glycans reside inside these organelles.

Glycosidic linkages of the heteroglycans

The SHGL-derived subfractions I and II were resolved by FFF and used for the analysis of glycosidic linkages. The data are compiled in Table 4. In both subfraction I and II, 3,6-Galp, 6-Galp, 3-Galp, t-Galp and t-Ara belong to the most prominent linkages. However, compared to subfraction II, subfraction I possesses more minor linkages, such as various glucosyl, mannosyl and fucosyl linkages and, therefore, the linkage pattern is more complex. Compared to subfraction I, subfraction II possesses a higher degree of branching. Similar results have been observed for SHG-derived subfraction I and II from leaflets of P. sativum L. (Fettke et al., 2004).

Table 4.  Glycosidic linkage analysis of subfraction I and II from SHGL. All sugars and linkages are given in mol.%
Sugar and Linkage(s)Sub ISub II
  1. p, Pyranose; f, furanose.

 5-Araf (4-Arap)4.34.2
 4,6-Glcp 0.3

Biochemical characteristics of the water-soluble heteroglycans

As the vast majority (or all) of the SHG-derived subfraction I and many constituents of SHGS reside inside the cytosol (in a strict sense) they are potential carbohydrate substrates of the cytosolic phosphorylase isoform (Pho 2). Protein–carbohydrate interactions were analyzed by in vitro tests using recombinant Pho 2 from Vicia faba L. Two different approaches were chosen:

SHGS was incubated with recombinant Pho 2 (or, for comparison, with the rabbit muscle phosphorylase a) and glucose 1-phosphate. Following incubation, the oligosaccharides were analyzed by HPAEC-PAD. A phosphorylase-catalyzed glucosyl transfer that utilizes any compound of SHGS is expected to cause the appearance of additional peaks and a diminished peak size of the oligosaccharide(s) that act(s) as glucosyl acceptor(s). Following incubation of SHGS with Pho 2 a series of additional peaks (elution times between 65 and 82 min, Figure 7a) was observed whereas other earlier eluting peaks decreased (Figure 7b). It should be noted that none of the late eluting glycans were affected by the Pho 2-catalyzed glucosyl transfer. This is not unexpected as oligoglycans from this part of the chromatogram are located inside the microsomal fraction and, under in vivo conditions, are probably not accessible to Pho 2.

Figure 7.

Pho 2-catalyzed glucosyl transfer into SHGS.
(a) SHGS (25 μg) were incubated with glucose 1-phosphate and phosphorylases. The later eluting underlined oligosaccharides are the result of the glucosyltransfer catalyzed by Pho 2.
(b) Zoom in on the SHGS-oligosaccharides, which decrease during Pho 2 incubation.
(c) Activity control of the Pho a (equal amount as in a) by incubation with glucose 1-phosphate and maltooligosaccharides (DP 6 and 7).

In control assays, either SHGS or Pho 2 was omitted and the reaction mixtures were processed identically. When the oligoglycan patterns from the control assays are compared with that of the complete reaction mixture the Pho 2 preparation does not contain any endogenous primer (data not shown). Similarly, the assay mixture does not contain, to any noticeable extent, an endogenous carbohydrate active enzyme activity.

The rabbit muscle phosphorylase is incapable to react with SHGS (Figure 7a). A significant glucosyl transfer was, however, observed when SHGS was replaced by a mixture of maltoheptaose and hexaose (Figure 7c).

The interactions between the recombinant Pho 2 and SHGL, subfraction I and subfraction II were monitored using 14C-labeled glucose 1-phosphate as glucosyl donor. Following incubation, the heteroglycans were separated from the residual sugar phosphate by membrane filtration and the incorporation of 14C-label into the glycans was quantified. In addition, assays were performed in which glycogen was applied as glucosyl acceptor or rabbit muscle phosphorylase a replaced the recombinant Pho 2 (Figure 8a). Pho 2 transfers glucosyl residues to both SHGL and subfraction I. Incorporation increases with time. The Pho 2-catalyzed incorporation into subfraction II is very low and probably due to an incomplete separation of the two subfractions. When the plant-derived glycans were replaced by glycogen (using equal amounts of glucose equivalents) the rate of the Pho 2-catalyzed glucosyl transfer increased. This difference is expected as a single glycogen molecule possesses far more glucosyl acceptor sites than the plant-derived glycans.

Figure 8.

Pho 2-catalyzed 14C-glucosyl incorporation into SHG.
(a) 10 μg SHGT and derived subfractions (I and II) were incubated with glucose 1-phosphate and Pho 2 or Pho a for 25 and 35 min, respectively. As controls the enzymes were incubated with glucose 1-phosphate and either 10 μg glycogen or without any glycans. Please note that for glycogen the scale is 10 times larger.
(b) Specific Pho 2-dependent priming capacity of SHGL derived from the five apolar fractions. Specific priming capacity was calculated based either on the carbohydrate content of subfraction I (Sub I) or of subfraction II (Sub II).

When the recombinant Pho 2 was replaced by a threefold higher activity of the rabbit muscle phosphorylase a, no incorporation into SHGL or subfraction I was observed (Figure 8a). Thus, interaction between the plant-derived SHG and the cytosolic phosphorylase appears to be highly selective. Similarly, the rabbit muscle phosphorylase was unable to utilize SHG preparations from P. sativum L. (Fettke et al., 2004). Furthermore, the selectivity of the Pho 2–SHG interactions implies that the heteroglycan preparations used do not contain, to any noticeable extent, starch-derived homoglucans. This conclusion is further supported by the fact that both SHGL and subfraction I prepared from the two starch-deficient Arabidopsis mutants (Figure 1b) clearly act as glucosyl acceptors for Pho 2 but not for the rabbit muscle phosphorylase (data not shown).

The data presented in Figure 8a indicate that subfraction I contains molecules that selectively act as acceptor for the Pho 2-catalyzed glucosyl transfer. However, in a strict sense these data do not exclude the possibility that the actual acceptor is a minor compound within subfraction I which, in this case, would be a complex mixture of structurally and/or functionally heterogeneous glycans. In order to test this possibility, we determined the Pho 2-dependent priming capacity of the various glycan preparations derived from the non-aqueous fractionation (see Figure 4a). Equal amounts of SHGL prepared from the five non-aqueous fractions were incubated for 45 min with recombinant Pho 2 and 14C-labeled glucose 1-phosphate. The priming capacity of the various glycans was estimated by monitoring the 14C incorporation. The relative distribution of the total Pho 2-dependent priming capacity was determined and compared with that of subfraction I, subfraction II, and the crystalline cellulose (Table 1). The Pho 2-dependent priming capacity clearly codistributes with subfraction I and the cytosolic marker enzyme activities but not with subfraction II. In addition, the specific priming capacity of each of the glycan preparations was calculated. The specific priming capacity is defined as the Pho 2-dependent 14C incorporation based on the carbohydrate content of the respective subfraction I or II (nmol 14C-glucosyl residues incorporated per μg glucose equivalents; Figure 8b). When the specific priming capacity was calculated based on the carbohydrate content of subfraction I for the five apolar fractions essentially the same value (7 nmol 14C glucosyl residues per μg carbohydrate) was obtained. Calculation on the basis of subfraction II resulted in values ranging from 1 to 5. The lowest value was observed in the pellet of the apolar gradient and the values increased with increasing distance to the pellet. Thus, it is highly likely that the Pho 2-dependent priming is an intrinsic property of the entire subfraction I.

Photosynthesis-dependent carbon flux toward heteroglycans

Photosynthesis-dependent carbon fluxes toward the SHG were analyzed by in vivo labeling using intact Arabidopsis plants supplied with 14CO2. In each experiment, four plants were placed into a sealed glass box and were continuously illuminated. Following 15, 30, 60 or 120 min of photosynthesis, the plants were withdrawn and the leaf material was used for the SHG isolation. SHGL was separated into β-glucosyl Yariv reagent reactive and non-reactive compounds which are essentially equivalent to subfraction II and I, respectively (Fettke et al., 2004). For both subfractions, the total monomer content and the 14C labeling were monitored. In addition, leaf starch was isolated and both the starch-derived glucose content and the radioactivity were quantified (Table 5). The fluxes into both SHGL-derived carbohydrates differ: based on FW incorporation of carbon is equal during a 15-min labeling period. During the conditions of prolonged labeling periods flux into subfraction II exceeds that into subfraction I. When the labeling is based upon the total monosaccharide content of the respective carbohydrate incorporation of 14C into subfraction I is at first higher than that into subfraction II. Later the incorporation into subfraction II exceeds that into subfraction I. This suggests that subfraction I represents a faster labeled, smaller but metabolically more active pool of heteroglycans.

Table 5. In vivo14C incorporation into SHG-derived subfraction I and II and into starch. Intact plants were labeled in the light for various periods of time by applying 14CO2. Subfraction I was determined as SHGL-derived glycans that are non-reactive with the β-glucosyl Yariv reagent. Subfraction II was quantified as the Yariv-reactive SHGL-derived glycans. The total 14C incorporation into the various carbohydrates was determined. Values are given based upon the fresh weight (FW) and upon the total monosaccharide content of the respective carbohydrate
 15 min30 min60 min120 min
Sub I
 nmol 14C g−1 FW  0.6491.3503.7468.082
 μmol 14C g−1 carbohydrate 19.3230.98109.87205.31
Sub II
 nmol 14C g−1 FW  0.6592.31812.13126.413
 μmol 14C g−1 carbohydrate  7.7423.63130.37295.56
 nmol 14C g−1 FW785156329408136
 μmol 14C g−1 carbohydrate41685816814216

For a more detailed analysis the hydrolysates of the two SHGL-derived subfractions were resolved by HPAEC-PAD and the specific labeling of the various monosaccharides was determined. Specific labeling is defined as the amount of label in a given monosaccharide based upon the amount of that monosaccharide. After 15 min of photosynthesis labeling is found in all monomers of subfraction I as well as of subfraction II. In Figure 9(a) for all monomers which are constituents of subfraction I and II the ratios of the specific activities of both subfractions (II versus I) are given.

Figure 9.

Specific labeling of the monosaccharyl constituents of subfraction I, II and of starch following in vivo labeling using 14CO2.
(a) Subfraction I and II were hydrolyzed and the resulting hydrolysates were separated by HPAEC-PAD. Subsequently, for each of the monosaccharides the total amount (given as glucose equivalents) and the content of 14C label were determined. For each monosaccharide the ratio of the specific radioactivity in subfraction II versus that of subfraction I is plotted against the labeling period.
(b) Specific labeling of the glucosyl moieties derived from starch, subfraction I or II following various periods of photosynthesis.

Because of the higher incorporation of 14C into the glucose of subfraction I during all time points, for this constituent the ratio of the specific activities of subfraction II to I declines. In contrast for galactose, arabinose and fucose the labeling in subfraction II exceeds that of the respective sugars of subfraction I and, therefore, the ratios of the specific activities increase over the time. The strongest increase is observed for xylose. Taken together these results clearly demonstrate that the carbon fluxes into and through the various monomers of subfraction I and II differ.

In addition the specific labeling of starch was determined and compared with that of the subfraction I and II (Figure 9b). Specific labeling of the starch-derived glucosyl moieties is consistently higher than that of subfraction I and II with the two latter differing from each other. This clearly shows that the 14C labeling of the glucose derived from subfraction I and/or II does not originate from any coisolation of labeled starch-derived compounds. In this case, essentially the same specific labeling is expected.

Diurnal alterations of the heteroglycans

In the leaves of wild-type Arabidopsis, the level of subfraction I changes depending upon the metabolic status. Leaves from A. thaliana L. grown under controlled conditions were harvested either at the middle of the light period (8 h after the onset of illumination) or of the dark period (5 h after darkening). Following isolation of SHGT, subfractions I and II were separated by FFF. During dark, the amount of subfraction I increased but the size distribution was not altered significantly (Figure 10). This effect can also be determined at other time points of the dark period and is significant as it was consistently observed in at least seven independently performed experiments. The increase in subfraction I during the dark period concurs with the assumption that this heteroglycan pool possesses a metabolic activity related to the diurnal alterations in the intracellular carbon fluxes (Smith et al., 2005).

Figure 10.

Diurnal changes in leaf SHGL-derived subfractions.
SHGL was prepared from leaves that had been harvested during the middle of the dark (dark) or the light period (light). Equal amounts of the two SHGL preparations were resolved by FFF into subfraction I (Sub I) and subfraction II (Sub II). The refractive index signals reflecting the polymer abundance were normalized at subfraction II. Data collected from a single experiment are shown. In seven independently performed experiments the same results were obtained.


The phenotype of Arabidopsis mutants that are deficient in either the chloroplast maltose exporter (Niityläet al., 2004) or the extraplastidial DPE2 (Chia et al., 2004; Lu and Sharkey, 2004) strongly suggests that the turnover of transitory starch is accompanied by a cytosolic glycan metabolism whose biochemical and physiological features are largely unknown. Until now, very little information on the existence and the precise subcellular location of extraplastidial glycans was available. For methodological reasons this aspect of the primary metabolism is difficult to analyze. The polysaccharide content of a plant cell is dominated by two distinct fractions, the plastidial starch granule (including starch-related α-glucans such as maltodextrins or phytoglycogen) and the extremely complex cell wall. Because of both the quantity and complexity of these two fractions the occurrence of starch or cell wall-derived contaminants has to be vigorously excluded if any cytosolic glycans/glycoconjugates are to be identified.

In this communication, we provide several lines of independent evidence for the existence of the postulated cytosolic carbohydrates in leaves of A. thaliana L. The isolation procedure recently elaborated (Fettke et al., 2004) results in a glycan preparation that lacks, to any noticeable extent, starch-derived contaminants. In this study, we have included two Arabidopsis mutants that are deficient in starch biosynthesis. Leaves were harvested during the light period and the SHG preparations were compared with that obtained from starch accumulating wild-type leaves. The controls did not yield more SHGT than the two mutants and the relative glucose content of the wild-type SHGT did not exceed that of the two glycan preparations derived from the mutants. Thus, the starch level of the starting material does not affect the glucose content of the SHGT preparation. The same conclusions were reached when SHGT preparations were analyzed that had been prepared from either leaves or tubers of Solanum tuberosum L. (data not shown). Finally, the data obtained concur with the fact that both SHGS and subfraction I act as glucosyl acceptor selectively for the cytosolic plant-derived phosphorylase isozyme but not with the otherwise kinetically similar rabbit muscle phosphorylase. The latter is very active with starch-like glucans (Figures 7 and 8).

It should be noted that the identification of SHG in the AGPase-deficient Arabidopsis plants (Figure 1) clearly deviates from a previous characterization of this mutant in which SHG were not detected (Schneider et al., 2002). However, the isolation procedure that had been used differed significantly from that applied in the present study.

For two reasons possible cell wall-derived contaminations have to be tested with great care if cytosolic glycans are to be analyzed: first, the monomer and linkage patterns of the SHG preparations are similar to those of some apoplastic glycans/proteoglycans (Fettke et al., 2004; see Table 4). Secondly, during biogenesis constituents of the cell wall matrix are retained inside the cell and, at these stages, are difficult to distinguish from compounds that reside in the cytosol.

In this study, we have combined the established technique of the non-aqueous fractionation with the isolation and quantification of various polysaccharides. In this context crystalline cellulose was isolated from each of the fractions of the apolar gradient, hydrolyzed and quantified. The distribution of the various SHG pools was compared with that of cellulose and of other marker proteins or enzyme activities. By using this approach the SHGL-derived subfraction II from Arabidopsis leaves codistributes with the marker of the apoplast (or of the vacuole, see above) whereas distribution of subfraction I is very similar to that of several cytosolic markers (Table 1).

These data extend and confirm results that have been recently obtained for P. sativum L. By using aqueous cell and tissue fractionation techniques, subfraction I and II were attributed to the cytosol and the apoplast, respectively (Fettke et al., 2004). However, in a strict sense these data do not prove that subfraction I is exclusively located in the cytosol and not inside compartments such as the endoplasmic reticulum, Golgi vesicles or other non-plastid organelles. Although no evidence for such a subcellular location is available, in the present study this possibility was tested by preparing intact microsomes and mitochondria/peroxisomes that were then used as starting material for heteroglycan preparation. The data clearly show that SHGS and subfraction I do not reside in peroxisomes or mitochondria. Furthermore, subfraction I is not associated with the endomembrane system which, however, contains some constituents of SHGS.

The subcellular location of the SHG-derived glycans perfectly correlates with the priming capacity for Pho 2. Some compounds of SHGS and subfraction I are selectively used as glucosyl acceptors by Pho 2 but the oligoglycans located in the microsomal fraction and the extracellular subfraction II do not act as primers. Furthermore, Pho 2-dependent priming capacity and subfraction I codistribute during non-aqueous gradient centrifugation.

The glucosyl acceptor function of SHGS, as observed in Arabidopsis (Figure 7a) was also found with Solanum tuberosum L. (JF, S. Poeste, NE, MS, University of Potsdam, Potsdam-Golm, unpublished data) but it contrasts the data recently obtained for P. sativum L. (Fettke et al., 2004). In the latter case most of the oligoglycans of SHGS were recovered inside isolated protoplasts but outside intact chloroplasts and, therefore, are attributed to the cytosol. Possibly, the SHGS from pea leaflets contains glycans that are chemically closely related to the actual glucosyl acceptors used by Pho 2.

The in vivo14C labeling experiments (Table 5 and Figure 9) clearly show that subfraction I and II are metabolically different heteroglycan pools. Based upon the total monosaccharide content labeling of subfraction II is delayed compared with subfraction I. This concurs with the extracellular location of subfraction II. As the specific radioactivities of the various monosaccharyl residues strongly deviate between subfraction I and II it is highly likely that their biosynthetic pathways are separated and/or utilize different pools of glycosyl donors.

The characteristics of the 14C incorporation into starch and into subfraction I and II (Figure 9b) exclude any contamination of the heteroglycan polymers with starch-derived carbohydrates. This conclusion is further supported by the distribution of the priming capacity of SHGL obtained by the non-aqueous fractionation. The priming capacity did not cofractionate with plastidial markers.

Leaf cells are capable to form the SHGT (as well as all derived (sub)fractions, data not shown) even if starch biosynthesis is not functional (Figure 1). Furthermore, photosynthesis-dependent carbon fluxes into subfraction I and II are essentially unaffected if starch biosynthesis is impeded. This conclusion is confirmed by labeling experiments performed with a starch-deficient mutant (pPGM) of A. thaliana L. In this mutant, 14CO2-dependent carbon fluxes into subfraction I and II were essentially unchanged compared with the wild type (data not shown). Thus, whilst the cytosolic heteroglycans appear to be involved in the metabolism of starch-derived mono- or disaccharides (Chia et al., 2004; Lu and Sharkey, 2004), their biosynthesis is not strictly linked to a functional starch metabolism. Because of the complexity of the monosaccharide and the linkage patterns of subfraction I it is expected that its metabolism requires a wide range of cytosolic carbohydrate active enzymes most of which remain to be identified. The interaction of DPE2 (Chia et al., 2004) with SHGS and/or subfraction I remains to be demonstrated. Until now the cytosolic phosphorylase isoform (Pho 2) is the only enzyme whose interaction with the cytosolic heteroglycans is unequivocally documented (Fettke et al., 2004). However, the Pho 2-catalyzed glucosyl transfer from or to the cytosolic heteroglycans is probably not a rate-limiting step within the intracellular carbon fluxes. In transgenic potato plants that underexpress or overexpress Pho 2 starch levels are not significantly altered (Duwenig et al., 1997; Poeste, 2004) despite some changes in the amount and monomer composition of the cytosolic heteroglycans. However, we observe a significant change in these heteroglycans when the expression of the cytosolic phosphoglucomutase is altered (JF, A. Fernie, MS, University of Potsdam, Potsdam-Golm, unpublished data).

Experimental procedures

Plant material

Arabidopsis thaliana L. (ecotype Columbia) and the two mutants were grown under controlled conditions with 15 h light period (200 μE m−2 sec−1 except were stated, 22 °C) and 9 h dark period (18 °C). Throughout the light–dark cycle the relative humidity was kept at 50%.

Isolation of the soluble heteroglycan

The SHG was isolated from leaves as recently described (Fettke et al., 2004). The SHG preparation was deprived from low molecular weight compounds by exhaustive dialysis (MWCO 1 kDa) and the resulting SHGT was separated by membrane filtration into two fractions containing glycans having an apparent size below (SHGS) or above (SHGL) 10 kDa. From SHGL subfraction I and II were obtained and characterized by using FFF-MALLS essentially as described elsewhere (Fettke et al., 2004). However, the cross flow was 3.5 ml min−1 for 0–5 min and then declined linearly to 0.5 ml during the next 20 min.

Carbohydrate quantification and ion chromatography

Following acid hydrolysis carbohydrates were quantified according to Waffenschmidt and Jaenicke (1987). Ion chromatography (HPAEC-PAD) was performed as described elsewhere (Fettke et al., 2004).

In vitro14C-glucosylation

In a final volume of 50 μl, polysaccharides (10 μg) were preincubated for 5 min at 37 ° C in 100 mm citrate-NaOH (pH 6.5) and 20 mm sodium glucose 1-phosphate (including 0.5 or 1 μCi [U-14C]-Glucose 1 phosphate). Following the addition of 5 nkat recombinant Pho 2 or, alternatively, 16.7 nkat rabbit muscle phosphorylase a (Sigma, Taufkirchen, Germany) incubation was continued. Reaction was terminated by heating (3 min at 90 °C) and the denatured reaction mixture was washed using a 10 kDa filter. The filtrate was collected and radioactivity was monitored. Washing was continued until the radioactivity of the filtrate was below 100 dpm. The retentate was dissolved in 200 μl water and mixed with 3 ml Rotiszint Mini (Roth, Karlsruhe, Germany). Radioactivity was monitored using a Beckman liquid scintillation counter (Beckman Coulter, Fullterton, CA, USA).

Non-radioactive glycan elongation

In a final volume of 50 μl, SHGS (25 μg) was preincubated for 5 min at 37 ° C in a mixture containing 100 mm citrate-NaOH (pH 6.5) and 20 mm sodium glucose 1-phosphate. Subsequently, 0.5 nkat recombinant Pho 2 or rabbit muscle phosphorylase a was added and incubation was continued for 60 min. As a control, SHGS was replaced by 25 μg of a commercial preparation containing maltohexaose and maltoheptaose. Subsequently, reaction was terminated by heating (2 min at 90 °C) and the mixture was passed through a 10 kDa filter. Following washing with 50 μl highly purified water the combined filtrates were analyzed by HPAEC-PAD.

Native PAGE

Non-denaturing discontinuous PAGE (Steup, 1990) was performed using a separation gel containing 7.5% T (w/v) monomer concentration and 0.2% (w/v) glycogen from oyster (Sigma) that had been dialyzed against deionized water prior to use. Electrophoresis was performed at 4 °C under voltage limitation (250 V). Following electrophoresis, slab gels were equilibrated for approximately 10 min in 100 mm citrate-NaOH (pH 6.5). For phosphorylase activity staining the slab gels were then incubated in a mixture of 100 mm citrate-NaOH, 0.1% (w/v) soluble starch, and 20 mm sodium glucose 1-phosphate (pH 6.5) at 37 °C. For DPE2 activity staining the incubation buffer contained 100 mm citrate-NaOH (pH 6.5), and 25 mm maltose (37 °C). Following incubation slab gels were stained with iodine.

SDS-PAGE and Western blotting

Denatured proteins were subjected to SDS-PAGE using a separation gel with 7.5% (w/v) total monomer concentration and were then transferred to nitrocellulose (0.2 μm pore size) followed by immunochemical detection (Albrecht et al., 1998). The following polyclonal primary antibodies were used: Anti-BiP Ab (Biomol, Hamburg, Germany), Anti-PEPCO, Anti-GWD Ab (gift from Dr Ritte, this laboratory). For quantification the AIDA program (V.2.3.1; Raytest Isotopenmeßgeräte, Straubenhardt, Germany) was used.

Protein identification by tryptic digestion and MALDI-MS

Sequence-based protein identification was performed essentially as described elsewhere (Haebel and Kehr, 2001).

Isolation of intact organelles

Mitochondria and peroxisomes were isolated according to Turano et al. (1996) using minor modifications. Arabidopsis leaves (5 g) were cut into small pieces in 30 ml chilled grinding buffer A [30 mm MOPS-NaOH (pH 7.5), 1 mm EDTA, 0.1% (w/v) BSA, 300 mm mannitol, and 0.02% (w/v) PVP 40] using a scalpel and were subsequently carefully homogenized with a mortar and pistil. The homogenate was passed through Miracloth and the resulting filtrate was centrifuged for 5 min at 500 g. The pellet was discharged and the supernatant was centrifuged for 10 min at 12 000 g to pellet the organelles. The supernatant was collected and the pelleted organelles were carefully resuspended in extraction buffer and centrifuged for 5 min at 12 000 g. The two supernatants were combined. The pelleted organelles were resuspended in grinding buffer A and brought to the same volume as the combined supernatant fraction. Throughout the entire procedure samples were kept at 4 °C. Both the organelle fraction and the supernatant fraction served as starting material for the SHG isolation.

Isolation of microsomes

Cellular membranes were isolated according to Lu and Hrabak (2002) using minor modifications. Arabidopsis leaves (7 g) were homogenized in 15 ml chilled grinding buffer B consisting of 50 mm Tris–HCl (pH 8.2), 1 mm PMSF, 1 mm DTE, and 20% (v/v) glycerol using a mortar and a pistil. The homogenate was passed through Miracloth and the resulting filtrate was cleared by centrifugation (5 min at 5000 g). Subsequently, the supernatant was centrifuged for 35 min at 125 000 g (Ultracentrifuge XL-90; Beckman GmbH, Krefeld, Germany). The supernatant and the pelleted microsomes were collected. The microsomal fraction was brought to the same volume as the supernatant fraction by adding grinding buffer B. During the entire isolation procedure temperature was kept at 4 °C.

Marker enzyme activities

Catalase activity was assayed by following the consumption of H2O2 according to Rao et al. (1996). Citrate synthase activity was monitored according to Stitt et al. (1989). Latent UDPase was determined according to Schaller and DeWitt (1995) using the Malachite Green method to estimate the release of orthophosphate. For non-aqueous fractionation the following marker enzyme activities were quantified using photometric assays: NADP-dependent glycerinaldehyde 3-phosphate dehydrogenase (EC; plastidial marker; Stitt et al., 1989), AGPase (EC; plastidial marker; Tiessen et al., 2003), UGPase (EC; cytosolic marker; Zrenner et al., 1993), α-glucosidase (EC; vacuolar marker; according to Stitt et al., 1989 but using PNP-N-acetyl-β-d-glucosaminidine as substrate), and α-galactosidase (EC; vacuolar marker; according to Stitt et al., 1989 but using PNP-N-acetyl-β-d-galactosaminidine as substrate). In addition, some marker proteins were quantified following PAGE (see above).

Non-aqueous fractionation of plant material

The method was adapted from the fractionation procedures described previously (Farréet al., 2001; Tiessen et al., 2002). Arabidopsis leaves (11 g FW) were harvested and immediately frozen in liquid nitrogen. The frozen material was homogenized using a metal ball mill precooled with liquid nitrogen and the resulting frozen powder was dried by lyophilization for at least 48 h. Subsequently, the dried powder was resuspended in 20 ml of a mixture of tetrachlorethylene-heptane [TCH; 66:34 (v/v)] that had been made water-free using molecular sieve beads. The suspension was sonified for 2 min under cooling and was then subjected to a two-step filtration through nylon nets of pore size 30 and 15 μm, respectively. The nylon nets were flushed with 20 ml heptane and the filtrate obtained was centrifuged for 10 min at 4000 g. The supernatant was discharged and the sediment was resuspended in 5 ml of TCH. Three aliquots (200 μl each) of the suspension were removed for the determination of enzyme activities in the unfractionated sample. The residual volume of the suspension was underlayed by a linear gradient [5% (v/v) to 85% (v/v) tetrachlorethylene in TCH; total volume 25 ml] and centrifuged for 50 min at 4000 g. Following centrifugation the entire gradient was separated into five fractions designated as P (pellet), and 0–3 according to their increasing distance from the pellet. Each fraction was brought to a volume of 20 ml by adding heptane and three aliquots (1800 μl each) were removed for the determination of marker enzymes. Subsequently, the five fractions were dried in vacuo and the residual material of each fraction, dissolved in 20 ml 20% (v/v) ethanol, served as starting material for the preparation of SHG and of crystalline cellulose.

Isolation and quantification of crystalline cellulose

Following the extraction of the five fractions obtained by the TCH gradient centrifugation with 20% (v/v) ethanol the insoluble material was collected resuspended in 70% (v/v) ethanol and was then centrifuged for 5 min at 3200 g. The pellet was deprived from proteins by treatment with 10 ml of a mixture of equal volumes of chloroform and methanol for 20 min under continuous stirring followed by centrifugation (as before). The resulting pellets were completely destained by washing with acetone. The pellets were subjected to hydrolysis in 0.5 ml Updegraf reagent (8:1:2 of concentrated acetic acid:concentrated nitric acid:water) for 30 min at 100 °C. Following centrifugation for 10 min at 20 000 g pure crystalline cellulose was recovered as pellet. The cellulose was washed twice with water and was then reacted for 1 h at room temperature with 72% (w/v) sulfuric acid. Subsequently, the samples were diluted to give a final concentration 1 m sulfuric acid and hydrolysis was completed by incubation at 100 °C for 2 h. Finally, samples were neutralized and the glucose content was monitored according to Waffenschmidt and Jaenicke (1987).

Glycosidic linkage analysis

The SHGL-derived subfraction I and II were subjected to glycosidic linkage analysis as previously described (Fettke et al., 2004).

In vivo14CO2 pulse labeling experiments

Four plants were placed in a pot and were grown under controlled conditions (170 μmol photon m−2 sec−1 for 15 h; 9 h dark) until a total shoot mass of approximately 10 g FW was reached. For in vivo labeling four intact plants were transferred into a sealed exsiccator (6.8 l) and were exposed to 14CO2 (specific radioactivity 0.98–2.4 MBq mmol−1). The total CO2 concentration in the gas phase varied depending upon the period of the pulse labeling (15 and 30 min: 500 μl l−1; 60 min: 900 μl l−1; 120 min: 1,800 μl l−1). During pulse labeling plants were illuminated with white light (approximately 110 μmol photons m−2 sec−1). Labeling was terminated by transferring the total leaf material from four plants into liquid nitrogen. The material was stored frozen until use. SHG was prepared essentially as described above and treated with β-glucosyl Yariv reagent to separate the Yariv reactive and the non-reactive compounds (equivalent to subfraction II and I, respectively) as previously described (Fettke et al., 2004). However, samples were desalted by ultrafiltration. Following acid hydrolysis carbohydrates were quantified according to Waffenschmidt and Jaenicke (1987).

Starch contents were determined photometrically after amyloglucosidase digest as described by Kötting et al. (2005). For quantifying the radioactivity an aliquot of the amyloglucosidase digest was filtrated (MWCO 10 kDa).

The filtrate as well as the hydrolyzed heteroglycan samples were separated by HPAEC-PAD and desalted by an anion self-regenerating neutralizer (ASRN-I 4 mm; Dionex, Idstein, Germany). Eluted monosaccharides were collected separately and used for monitoring the radioactivity (see above).


Financial support by the Deutsche Forschungsgemeinschaft (Sonderforschungsbereich 429, ‘Molecular Physiology, Energetics, and Regulation of Primary Metabolism’ TP B2 and B7) is gratefully acknowledged. The authors are indebted to Ms Silke Gopp for competent technical assistance, Ms Andrea Mohrenweiser for the analysis of the phosphorylase pattern, Dr Simon Poeste and Ms Andrea Mohrenweiser for providing the recombinant cytosolic phosphorylase from Vicia faba L., and Dr Gerhard Ritte for providing the antibody directed against the glucan, water dikinase. The authors thank Ms Julia Schönfeld (AG Pauly, Max-Planck-Institute of Molecular Plant Physiology, Potsdam-Golm, Germany) for kindly performing the glycosidic linkage analyses and Dr Sophie Haebel (Interdisciplinary Center for Mass Spectrometry of Biomolecules, University of Potsdam) for performing the MALDI-MS analyses.