Plant Physiol. (1999) 120: 1105-1116

The Mechanism of Synthesis of a Mixed-Linkage (13),(14)-D-Glucan in Maize. Evidence for Multiple Sites of Glucosyl Transfer in the Synthase Complex¹

Marcos S. Buckeridge, Claudia E. Vergara, and Nicholas C. Carpita^*

Instituto de Botânica, Secão de Fisiologia e Bioquímica Plantas, Caixa Postal 4005, CEP-01061970, São Paulo, SP Brazil (M.S.B.); and Department of Botany and Plant Pathology, Purdue University, West Lafayette, Indiana 47907-1155 (C.E.V., N.C.C.)

	ABSTRACT

Top Abstract Introduction Methods Results Discussion References

We examined the mechanism of synthesis in vitro of (13),(14)-D-glucan (-glucan), a growth-specific cell wall polysaccharide found in grasses and cereals. -Glucan is composed primarily of cellotriosyl and cellotetraosyl units linked by single (13)-linkages. The ratio of cellotriosyl and cellotetraosyl units in the native polymer is strictly controlled at between 2 and 3 in all grasses, whereas the ratios of these units in -glucan formed in vitro vary from 1.5 with 5 µM UDP-glucose (Glc) to over 11 with 30 mM substrate. These results support a model in which three sites of glycosyl transfer occur within the synthase complex to produce the cellobiosyl-(13)-D-glucosyl units. We propose that failure to fill one of the sites results in the iterative addition of one or more cellobiosyl units to produce the longer cellodextrin units in the polymer. Variations in the UDP-Glc concentration in excised maize (Zea mays) coleoptiles did not result in wide variations in the ratios of cellotriosyl and cellotetraosyl units in -glucan synthesized in vivo, indicating that other factors control delivery of UDP-Glc to the synthase. In maize sucrose synthase is enriched in Golgi membranes and plasma membranes and may be involved in the control of substrate delivery to -glucan synthase and cellulose synthase.

	INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

The mixed-linked (13),(14)-D-glucan (hereafter referred to as simply -glucan) is a cell wall polysaccharide found only in grasses and cereals (Carpita, 1996). The -glucan is not synthesized in dividing cells but accumulates specifically during cell enlargement (Carpita and Gibeaut, 1993). The -glucan also accumulates in the walls of the endosperm of the developing grains and their surrounding maternal tissues (Fincher and Stone, 1986; Brown et al., 1997).

The -glucan structure was established by use of a sequence-dependent Bacillus subtilis endoglucanase (lichenase) that cleaves (14)-D-glucosyl units only if preceded by (13)-units and yields primarily a diagnostic trisaccharide, cellobiosyl-(13)-D-Glc, and a tetrasaccharide, cellotriosyl-(13)-D-Glc (Anderson and Stone, 1975). It is an unbranched glucan and over 90% of the polymer consists of these cellotriosyl and cellotetraosyl units in ratios ranging from 2 to 3 in grasses, each connected by a single (13)-linkage (Wood et al., 1991, 1994). The remainder of the polymer consists of longer runs of the cellodextrin interspersed within the polymer and connected by single (13)-linkages (Staudte et al., 1985; Wood et al., 1994). The ratio of the odd cellodextrin oligomers is generally about 2-fold greater in abundance than the next-higher even-numbered oligomer in the series (Wood et al., 1991, 1994). Although not found in any other angiosperms, a similar mixed-linkage -glucan is found in the lichen Cetraria islandica (Wood et al., 1994). In the lichen -glucan, the cellotriosyl units comprise 86% of the polysaccharide.

Synthesis of -glucan with cellular membranes was shown by digestion of the radioactive products with several enzymes, including the B. subtilis enzyme, and subsequent gel-permeation chromatography, HPAE-HPLC, or TLC of the hydrolysis products (Henry and Stone, 1982; Gibeaut and Carpita, 1993; Becker et al., 1995). Synthesis occurs strictly at the Golgi apparatus, uses UDP-Glc as substrate, and requires Mg²⁺ or Mn²⁺ as a cofactor (Gibeaut and Carpita, 1993). The proportions of (13)- and (14)-D-glucosyl linkages in total glucan products are altered substantially by reaction conditions in microsomal fractions from Lolium multiflorum (Meikle et al., 1991), and significant amounts of -glucan are made only at UDP-Glc concentrations above 100 µM. A combination of gel permeation chromatography, linkage analysis, and enzymic digestion confirmed that entire tri- and tetrasaccharide units were synthesized and that the macromolecular -glucan synthesized in vitro in the Golgi apparatus was similar to the cell wall polysaccharide (Gibeaut and Carpita, 1993).

Unlike the Golgi apparatus from nongramineous plants, the maize (Zea mays) Golgi apparatus also synthesizes a considerable proportion of callose in vitro (Gibeaut and Carpita, 1994). The Golgi membrane-associated callose synthase was not attributed to contamination with plasma membrane; this activity is inhibited slightly by Mg²⁺ and Mn²⁺ (Gibeaut and Carpita, 1993) and stimulated only 2-fold by CaCl₂ compared with 7-fold for the plasma membrane-associated synthase (Gibeaut and Carpita, 1994). Compounds that disrupt membrane integrity, such as detergents and ionophores, abolish -glucan synthase activity and increase the amount of callose synthesized. We have proposed that, like cellulose synthase, -glucan synthase may revert to callose synthase when disrupted (Gibeaut and Carpita, 1994).

The structure and ratio of the cellotriosyl and cellotetraosyl units of -glucan and the iteration of this ratio in higher order odd- and even-numbered oligomers are relatively constant within a species (Wood et al., 1994). How this consistency is maintained during synthesis is not known. The catalysis of polymerization of cellulose, -glucans, and polymers containing (14)-linked mannosyl and xylosyl units must overcome a steric problem, because the (14)-linkage requires that each of these sugar units be rotated nearly 180° with respect to its neighbors. With a single site of glycosyl addition, the nonreducing acceptor sugar of the growing chain must be rotated with each addition, the active site of the substrate relative to the chain terminus must be rotated, or the sugar must be added in an activated configuration strictly at the O-4 position but then rotate into its proper 180° orientation as the chain extends.

As an alternative, we proposed that two sites of glucosyl transfer generate disaccharide units (Carpita et al., 1996; Carpita and Vergara, 1998). We proposed further that the trisaccharide unit of -glucan can be formed if an ancestral cellulose synthase complex were modified to contain a third glucosyl transferase activity in addition to the cellobiose generating activity (Carpita et al., 1996). With an odd number of glucosyl units added, the nonreducing terminal sugar of the growing chain must always be oriented to expose the O-3 hydroxyl for attachment instead of the O-4 and, therefore, cellotriosyl units would always be linked by single (13)-linkages. The experiments reported here were designed to resolve some of these possibilities. Our results indicated that cellotriosyl units are selectively enriched in abundance with respect to the higher cellodextrin series at more saturating substrate concentrations, and therefore support a three-site model of glucosyl addition.

Treatments designed to reduce UDP-Glc concentration in excised coleoptile sections did not result in marked changes in cellodextrin oligomeric ratios or in molecular size of the newly synthesized -glucan. These data indicate that UDP-Glc supply to the -glucan synthase is regulated by factors independent of cytosolic nucleotide sugar concentration. We immunodetected SuSy associated with Golgi membranes of maize but not those of soybean (Glycine max), whereas this UDP-Glc generating enzyme was found associated with the plasma membrane of both species. Amor et al. (1995) proposed such a role for SuSy in cellulose synthesis, and our data suggest that SuSy regulates delivery of UDP-Glc to the maize -glucan synthase at the Golgi apparatus by a similar mechanism.

	MATERIALS AND METHODS

Top Abstract Introduction Methods Results Discussion References

Plant Material

5

Synthesis of -Glucan in Vivo under Conditions of Depleted UDP-Glc

For determinations of nucleotide sugars, frozen coleoptile sections were homogenized in ice-cold 10% (w/v) TCA, centrifuged at 10,000g for 10 min, and the supernatant was collected. The TCA was removed by two partitionings against trioctylamine in 1,1,2-trichlorotrifluoroethane, as described previously (Kanabus et al., 1986). Nucleotides and nucleotide sugars in the neutralized aqueous fraction were separated by HPAE-HPLC and detected by UV A₂₆₀ essentially as described by Liljebjelke et al. (1995). UDP-Glc was also assayed enzymically with UDP-Glc dehydrogenase (Keppler and Decker, 1981). The UDP-sugar fraction from HPLC was collected, and the sugars were hydrolyzed by 2 M trifluoroacetic acid. The trifluoroacetic acid was then evaporated in a stream of air and the sugars were separated by HPAE-HPLC and detected by pulsed amperometry (Gorshkova et al., 1997), or alditol acetate derivatives were prepared and separated by GLC (Gibeaut and Carpita, 1991). In parallel experiments, Suc and other small molecules were extracted from the frozen coleoptile sections with 80% (v/v) ethanol at 45°C for 30 min with gentle agitation, and the extract was collected by pipette and dried under a stream of nitrogen at 35°C. Suc and Glc were quantified in samples with and without invertase with a coupled hexokinase/Glc-6-P dehydrogenase assay (Carpita and Kanabus, 1987).

The cell walls pelleted in ice-cold TCA were resuspended in 50 mM sodium acetate, 50 mM NaCl, and 30 mM ascorbate, pH 5.5, filtered on nylon mesh, and washed sequentially with 100 mM NaCl, water, acetone, 100 mM NaCl, and water. A portion of the wall preparation was saved for further analysis, and the remainder (11-15 mg) was extracted sequentially with 20 mL each of 0.5% (w/v) ammonium oxalate, pH 7.0 (90°C with occasional stirring), and 0.08 N NaOH, 0.8 N NaOH, and 4 N NaOH (ambient temperature with continuous stirring). Each alkali solution was supplemented with 3 mg mL¹ NaBH₄ to prevent reducing-end elimination, and extractions were made at atmospheric N₂. All extractions were for 1 h except the 4 N NaOH fractionation, which was for 15 h. After each extraction the unextracted wall was pelleted by centrifugation and the supernatant liquid was filtered through glass-fiber mats to remove small amounts of suspended wall material. Alkali-soluble fractions were chilled to ice temperature and acidified with glacial acetic acid to about pH 5.0. All fractions were dialyzed extensively against running deionized water and freeze-dried. The -cellulose remaining was acidified to pH 5.0 with glacial acetic acid and then washed several times with deionized water. Radioactivity in these fractions was determined by liquid scintillation spectroscopy. The freeze-dried materials were suspended in 2.5 mL of water, and 0.5 mL was assayed for radioactivity by liquid scintillation spectroscopy.

A Bacillus subtilis endoglucanase preparation was added to portions of wall and fractionation material in water to digest material into cellodextraoyl-(13)-D-Glc oligosaccharides. These oligosaccharides were separated by HPAE-HPLC in a NaOH/sodium acetate gradient designed to optimize separation of oligosaccharides to DP 10 (Gibeaut and Carpita, 1993). One-half-milliliter fractions were collected in 1.0 mL of 2 N acetic acid, and radiolabeled oligosaccharides were determined by liquid scintillation spectroscopy. The remainder of the neutralized alkali extracts were dialyzed and freeze-dried for gel permeation chromatography.

Preparation of Membranes

About 15 mL of the homogenate was pipetted into a 38.5-mL centrifuge tube (Ultraclear, Beckman) onto a cushion of 50% (w/v) Suc in a gradient buffer containing 20 mM HEPES[Bis-Tris-propane], pH 7.6, and overlaid with 8 mL each of 34% and 25% (w/v) Suc, 7 mL of 18% (w/v) Suc in gradient buffer, and the remaining volume was made up with 9.5% (w/v) Suc in the same buffer. After Suc gradient centrifugation at 140,000g in a rotor (model SW28, Beckman) for 45 min, the interfaces containing the enriched Golgi apparatus were collected with a Pasteur pipette or by a displacement fractionator. These membranes were used directly for -glucan synthase reactions at low substrate concentration or were diluted with 1.08 M Suc in gradient buffer and pelleted by centrifugation at 140,000g for 30 min in the rotor. The membrane pellet was gently resuspended with a camel-hair paintbrush into 2 mL of 1.08 M Suc in gradient buffer for -glucan synthase reactions at high substrate concentration.

Samples of the other layers were also taken for protein determination and enzyme marker assays. For immunolocalization of SuSy, portions of the mitochondrial- (47%/34% interface), Golgi- (34%/25% interface), and ER/tonoplast-enriched membranes (25%/9.5% interface) were diluted with 1.08 M Suc in gradient buffer, pelleted at 140,000g, and the pellets were dissolved by heating in a PAGE sample buffer containing 10% (w/v) SDS and 100 mM DTT. For preparation of enriched plasma membrane, coleoptile and soybean tissues were gently mashed with a mortar and pestle in homogenization buffer containing 9.5% (w/v) Suc, the homogenate was cleared of cell debris by centrifugation at 3,000g for 10 min, and the total membranes were then pelleted at 140,000g in the rotor.

A portion of the supernatant was saved for immunolocalization of soluble SuSy, and the membrane pellets were gently resuspended in 2 mL of a two-phase buffer consisting of 6.5% (w/v) PEG 3350 and 6.5% (w/v) dextran 500T in 5 mM potassium phosphate, pH 7.2, 3 mM KCl, 10 mM DTT, and 0.2 mM PMSF. The suspension was frozen, thawed, and gently sheared by six strokes in a smooth glass tube with a Teflon piston. Additional two-phase buffer was added, and the plasma membrane was prepared by two rounds of partitioning (100 inversions followed by centrifugation at 400g for 5 min at 4°C). The membranes were washed twice in 20 mM HEPES[KOH], pH 7.2, containing 1.08 M Suc and 10 mM KCl, and pelleted by centrifugation at 140,000g. Marker assays were performed after resuspension in the dilution buffer. Soluble and membrane proteins were dissolved by heating in a PAGE sample buffer containing 10% (w/v) SDS and 100 mM DTT.

Protein was determined by bicinchoninic acid assay (Pierce), and assays for IDPase (Golgi), NADH oxidase (mitochondria), vanadate-sensitive ATPase (plasma membrane), NADP reductase (ER), and NO₃-sensitive ATPase (vacuole) were performed as optimized for maize from published methods described in Gibeaut and Carpita (1994).

Immunolocalization of SuSy

1

The developed blots were scanned (QuickScan, Epson, Torrance, CA) at 254 grayscale units and 600 dpi, and the band intensities were quantified with an image analysis program (Scanalytics, Vienna, VA). SuSy was also detected in a standard series of 0.1 to 10 µg of total soluble protein to determine the linearity of signal with protein content.

Glucan Synthase Reactions

Digestion with B. subtilis Endoglucanase as a Specific Assay for (13),(14)-D-Glucan

HPAE-HPLC

Gel Permeation Chromatography

	RESULTS

Top Abstract Introduction Methods Results Discussion References

-Glucan Synthesis in Vitro

1

1

1

1

1

1

1

1

View larger version (35K): [in this window] [in a new window]   Figure 1. HPAE-HPLC separation of labeled cellodextrin-(13)-D-Glc oligosaccharides from -glucan synthesized in vitro. A, Separation of the -glucan cellodextrin products synthesized at low substrate concentrations. Radioactivity from 0.5-mL fractions was determined by liquid scintillation spectroscopy after acidification of the NaOH with acetic acid. Upper trace, Pulsed amperometric detection of cellodextrin-(13)-D-Glc oligosaccharides from -glucan. B, Separation of the -glucan cellodextrin products synthesized at high substrate concentrations. The cellodextrin-(13)-D-Glc oligosaccharides from -glucan made in vitro were quantified by pulsed amperometric detection (PAD).

View larger version (18K): [in this window] [in a new window]   Figure 2. A, Rates of -glucan synthase activity dependent on UDP-Glc concentration. Product made was determined from total radioactivity and specific activity of substrate or from calibration of the pulsed amperometric signal. Values in the range of 5 to 255 µM represent the averages of three experiments, whereas those in the range of 5 to 30 mM are from two experiments. B, Molar ratios of cellotriose to cellotetraose and cellopentaose to cellotetraose are dependent on the UDP-Glc concentration. The molar ratios based on incorporation of radioactivity were corrected based on the number of glucosyl units per oligosaccharide, whereas the mass detected by pulsed amperometry was calibrated by direct assay of eluted sugars.

The values for total incorporation per microgram of membrane protein were proportional to the ratios of both cellotriose to cellotetraose and cellopentaose to cellotetraose. The relative amounts of cellotriose formed increased with increasing concentrations of UDP-Glc provided, but exhibited an apparent saturation at concentrations of UDP-Glc below 5 mM (Fig. 2B). At 5 µM substrate, the ratio of the cellotriose and cellotetraose units was only 1.5 but increased to 3.7 at 255 µM UDP-Glc. The ratio of cellotriose to the less-abundant cellopentaose also rose with increasing substrate concentration, from about 8.0 to 26, and the cellopentaose to cellotetraose ratio was nearly constant at 0.15 (Fig. 2B). The cellopentaose to cellohexaose rose only slightly, from 2.5 to 4.0, over the same concentration range. However, at substrate concentrations between 5 and 30 mM, the ratio of cellotriose to cellotetraose units increased markedly again, from about 5.0 with 15 to 20 mM UDP-Glc to over 11 with 30 mM UDP-Glc (Fig. 2B). The ratio of cellotriose to cellopentaose units was about 15 with 10 mM UDP-Glc, but then fell to only about 5.0 with 30 mM substrate. Concomitant with the marked increase in cellotriose units between 15 and 30 mM UDP-Glc, the ratio of cellopentaose to cellotetraose units increased even more strongly, from about 0.5 at 5 mM UDP-Glc to 2.0 at 30 mM. The cellopentaose units are a small fraction of the total at suboptimal UDP-Glc but become twice as abundant as cellotetraose units at high substrate concentrations.

Separation of the radiolabeled -glucan synthesized in vitro by gel permeation chromatography revealed that polymers less than 50 kD were formed at substrate concentrations of 15 µM and less. The product made at 80 µM was about 100 kD, and that formed at 255 µM was about 250 kD, which is about the size of native maize -glucan (Fig. 3).

View larger version (35K): [in this window] [in a new window]   Figure 3. Gel-permeation chromatography showing the molecular mass distribution of native maize -glucan compared with -glucan synthesized in vitro at various low substrate concentrations. The native -glucan and the products of the in vitro reactions were dissolved in McIlvaine's buffer, pH 5.5, and loaded onto a 2.5- × 40-cm column of Sepharose 4B equilibrated in the same buffer. Four-milliliter fractions were collected, and a portion was digested with the B. subtilis endoglucanase enzyme for HPAE-HPLC. The -glucan was detected either by total sugar (Dubois et al., 1956) or by radioactivity as assayed by liquid scintillation spectroscopy. The distribution of native -glucan observed here was essentially identical for -glucans extracted from excised coleoptile sections incubated with labeled Glc as described in Tables II and III. Two experiments were run with essentially identical results. Fractions from in vivo labeling experiments were pooled (designated I, II, III, and Incl. [included]), dialyzed against deionized water, and assayed for cellodextrin distribution in -glucan as described in Table III. Peak fractions of dextran standards for molecular mass (in kD) are marked.

Synthesis of -Glucans in Vivo

View this table: [in this window] [in a new window]   Table I. Changes in Glc, Suc, and UDP-Glc content and relative concentration during growth of excised coleoptile sections Sections exactly 11 mm long were sliced with a razor blade 1 to 2 mm below the tip of 3-d-old coleoptiles about 2 to 2.5 cm long. Sections were floated on a medium containing 5 mM potassium phosphate, pH 5.5, 5 mM KCl, 0.01% tetracycline, and 2 × 105 M IAA, and, in some instances, supplemented with 100 mM Glc or 10 mM Gal. The sections were transferred either immediately or after 4.5 or 9.5 h of incubation at 30°C. Matching sets of samples were transferred either immediately or after 3 or 8 h of incubation at 30°C to the same respective media containing radiolabeled Glc and incubated an additional 1.5 h at 30°C. After incubations the sections were rinsed with water and frozen in liquid nitrogen. Length measurements are from 20 coleoptile sections. Values are typical of three independent experiments.

After incubation for 9.5 h without sugar, the levels of all metabolites were diluted, but the UDP-Glc concentrations were substantially lowered in sections incubated without sugar or in the presence of growth-inhibiting Gal. UDP-Glc was the major UDP-sugar in the presence or absence of Glc, representing 85% of the fraction, whereas incubation with Gal resulted in UDP-Gal concentrations representing 60% of the total UDP-sugar (Fig. 4). Relative Glc and Suc concentrations also decreased during incubations without sugar or in the presence of Gal but to a lesser extent than the marked decreases in nucleotide sugar. The relative concentrations of Glc and Suc could be maintained to a certain extent by incubation of the coleoptile sections with 100 mM Glc, but UDP-Glc levels fell to about one-third that of the freshly isolated coleoptiles.

View larger version (21K): [in this window] [in a new window]   Figure 4. Separation of nucleotide sugars by HPLC. Nucleotide sugars were extracted with ice-cold 10% (w/v) TCA, and the TCA was partitioned into a mixture of trioctylamine in 1,1,2-trichlorofluoroethane. The neutralized aqueous extract was passed through a 0.22-µm membrane filter and injected onto a column equilibrated in 20 mM ammonium formate, pH 6.0. The nucleotides and nucleotide sugars eluted in a nonlinear gradient to 1 M ammonium formate, pH 6.0, and were detected and quantified from UV absorbance based on standards. Profiles are typical of three experiments.

After incubation in radiolabeled sugars, the coleoptile walls were separated into pectic and several alkali-soluble fractions. Most of the label was incorporated into polysaccharides of the alkali-soluble fractions, and the -glucans were found primarily in the 0.8 and 4 N NaOH fractions (Fig. 5). The ratios of the cellotriosyl to cellotetraosyl units in these two fractions were similar in the individual samples depending on the incubation conditions (Table II). When the 4 N NaOH fractions were separated by gel permeation chromatography, the molecular mass of -glucan and the distribution of radioactivity in the newly synthesized glucan were each centered at about 250 kD regardless of the incubation conditions (as in Fig. 3). The ratio of newly synthesized cellotriosyl to cellotetraosyl units in each of three fractions, representing the larger third to the smaller third of the included polymers and the total included volume, were similar to that expected in the total -glucan digested from the fraction, although slightly lower ratios were obtained with the smallest of polysaccharides (Table III). Based on pulsed amperometric detection, the molar ratios of cellotriosyl to cellotetraosyl units in -glucan mass digested from the coleoptile was about 2.9 in most fractions from gel permeation chromatography.

View larger version (22K): [in this window] [in a new window]   Figure 5. Relative amounts of -glucan in the alkali extracts of depectinated maize coleoptile walls determined by HPAE-HPLC of the cellodextrin-(13)-D-Glc oligosaccharides. The -glucan oligomers were separated and quantified as described in Figure 1B. PAD, Pulsed amperometric detection.

View this table: [in this window] [in a new window]   Table II. Ratio of cellodextrin oligomeric products synthesized in maize coleoptile cell wall -glucans synthesized in vivo from [14C]Glc -Glucan was digested with the B. subtilis enzyme, and the oligomeric products were separated by HPAE-HPLC and quantified by liquid scintillation spectroscopy. Treatment of coleoptile sections was as described in Table I. Ratios of treated samples are based on radioactivity incorporated after adjustment for molar ratio; ratio for maize -glucan is made by pulsed amperometry, with empirical calibration of the signal from sugar determinations (Dubois et al., 1956).

View this table: [in this window] [in a new window]   Table III. Molar ratio of the cellotriosyl and cellotetraosyl units based on incorporation of radioactivity from [14C]Glc into several size fractions of -glucans synthesized by excised coleoptile sections Treatments are as described in Table I. Ratios of treated samples are based on radioactivity incorporated after adjustment for molar ratio. Peak fraction designations and incubation treatments are described in Figure 3 and Table I, respectively.

Immunolocalization of SuSy in Maize Coleoptiles versus Soybean Hypocotyls

View this table: [in this window] [in a new window]   Table IV. Comparison of marker assays of membrane purity with quantitation of intensity of the immunogel blot signal with SuSy antisera The soluble protein of the homogenate and the Golgi membranes were prepared from flotation centrifugation as adapted from Gibeaut and Carpita (1994), whereas the plasma membrane was obtained by two-phase aqueous partitioning of total membranes. The vanadate-sensitive ATPase for soybean membranes represented 70% and 23% of the total phosphatase activity of the plasma membranes and Golgi membranes, respectively. This activity for maize membranes represented 66% and 41% of the total phosphatase activity of the plasma membranes and Golgi membranes, respectively. The blot intensities of the SuSy bands revealed by Western analysis (Fig. 6) were digitally scanned and quantified after subtraction of background. Values are typical of four independent experiments.

View larger version (30K): [in this window] [in a new window]   Figure 6. Immunodetection of SuSy in soluble (Sol), and plasma membrane (PM)- and Golgi membrane-enriched fractions of elongating maize coleoptiles and soybean hypocotyls. The Golgi membranes and soluble fraction were obtained after flotation centrifugation as described by Gibeaut and Carpita (1994), whereas the plasma membrane was obtained by two-phase aqueous partitioning. Ten micrograms of protein per lane was separated by SDS-PAGE in 10% (w/v) gels, and electroblotted onto cellulose nitrate by semidry blotting. SuSy was detected with polyclonal antisera at a 1:500 dilution for maize proteins and a 1:200 dilution for soybean proteins. Values are typical of four determinations.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

The Synthesis of -Glucans in Vitro

At the highest UDP-Glc concentrations tested, the cellotriose comprised nearly 80% of the total polymer synthesized. The increase in the proportion of cellotriose occurs relative to all other cellodextrins at low substrate concentrations, but the proportion of cellopentaosyl units also increases concomitantly at high substrate concentrations. While the cellotetraosyl units exceed the cellopentosyl units by about 7:1 at concentrations of UDP-Glc below 1 mM, the cellopentaosyl units exceed the cellotetraose units by 2:1 when the substrate concentration is 30 mM (Fig. 2B). The distribution of cellodextrin units in -glucan synthesized in vitro at 30 mM UDP-Glc becomes similar to that of the -glucan from the lichen Cetraria islandica, in which the cellotriose unit is 86% of the polysaccharide, and the cellopentaose is second in abundancetwice that of cellotetraose (Wood et al., 1994).

These data indicate that the mechanism of synthesis occurs in such a way as to favor the formation of cellotriosyl and odd-numbered cellodextrin units at saturating substrate concentrations. Thus, the synthesis of -glucan resembles that of cellulose synthesis, but cellotriosyl units are linked by single (13)-linkages instead of two cellobiosyl units linked by a (14)-linkage. Regardless of the UDP-Glc-dependent variation in the ratio of cellodextrins, the single (13)-linkages between them is a constant, and a biochemical mechanism of synthesis to explain our data must also account for these two features.

The orientation of the two Glc units in a (13)- or (14)-linkage are so different that it may seem difficult to imagine how both can be made interchangeably in a single synthase. However, Glc units in a linear (14)-linked chain are slightly angled such that the O-4 or O-3 hydroxyl of the nonreducing terminal sugar will fill an acceptor position merely by inverting the chain (Fig. 7A). If the positioning of the terminal sugar is controlled at the chain-holding portion of the synthase, then the introduction of (14)- or (13)-linkages can be strictly regulated. For cellulose synthase, we and others have proposed that cellobiosyl units are added to the nonreducing end of a growing chain in which the O-4 hydroxyl is in the acceptor position (Carpita et al., 1996; Koyama et al., 1997; Carpita and Vergara, 1998). When cellobiose units are added, the chain is never reoriented but instead glides through the chain-holding portion of the synthase until the new nonreducing terminal O-4 hydroxyl is positioned for the next round of synthesis. This model of synthesis also provides a simple explanation for how a callose synthesis can "default" by damage to the cellobiose generating synthase. If the second glucosyl transferase fails to function, then only a single glucosyl unit will glide into the nonreducing end and the O-3 hydroxyl will be positioned in the acceptor position of the synthase (Carpita et al., 1996). Because the (13)-linkage does not invert one sugar with respect to its neighboring sugars, every nonreducing sugar afterward will have its O-3 hydroxyl in the acceptor position and callose will be the only product.

View larger version (20K): [in this window] [in a new window]   Figure 7. A, Two possible positions of the nonreducing terminal acceptor glucosyl unit of a growing chain can be made by rotating the chain over or by processively adding a single glucosyl unit from a fixed UDP-Glc site. Both (14)- and (13)-linkages are possible, so positioning of the nonreducing terminal sugar to expose either the O-3 or O-4 must be tightly controlled in glucans with a defined linkage structure. B, A single fixed site of glucosyl transfer requires the growing chain to be "read" by the enzyme to determine when the chain is rotated to direct linkage structure at the O-3 or O-4 position, as defined in A. C, Three glucosyl transferases working processively would always link a cellotriosyl unit to the O-3 of the nonreducing acceptor sugar. The third site is of lower affinity than the cellobiosyl generating glucosyl transferases, and failure to fill it produces the longer cellodextrins. Formation of cellotriosyl units should form almost exclusively at saturating UDP-Glc concentrations. D, Five glucosyl transferases work processively to form mostly cellotriosyl units, but occasionally cellotetraosyl and cellopentaosyl units, because the fourth and fifth sites are of lower affinity for substrate binding. Longer cellodextrins would be favored at saturating UDP-Glc concentrations. In B through D, the arrows indicate the reducing end and continuation of the synthesized polymer.

How does the mechanism of synthesis of -glucan parallel that of cellulose synthesis? We formulated three possible mechanisms of -glucan synthesis. In all models, the synthase is considered to have two domains: one or more active sites for substrate binding and a single chain-holding domain that positions the nonreducing terminal sugar to form the proper glycosidic linkage. To synthesize (14)-D-glucosyl linkages with a single UDP-Glc site, either the substrate-binding domain or the chain-holding domain must rotate or toggle with respect to each other with each round of glycosyl transfer. Alternatively, the glucosyl unit is guided into the proper orientation by a steric mechanism within the chain-holding domain (Fig. 7B). However, to synthesize a mixed-linkage -glucan, the chain-holding domain must also direct the proper ratio of cellotriosyl and cellotetraosyl units and the placement of the (13)--D-glucosyl units. Thus, the chain-holding domain functions as a "chain-reading" domain.

Such a model of synthesis must be independent of UDP-Glc concentration because the information needed to direct the synthesis of the proper ratios of cellodextrin units and their proper spacing by (13)-linkages must come from polymer already synthesized (Fig. 7B). Although our data seem to rule out this model, variations of the single-site model excluding the chain-reading function are possible. Because there is considerable freedom of rotation of the nonreducing terminal glucosyl unit around the glycosidic bond, one could argue that there is a statistical probability of forming a (13)-linkage that is dependent on substrate concentration. However, the strict unit structure, the lack of cellobiosyl units, and the lack of contiguously linked (13)-units strongly argue against any model of random introduction of (14)- and (13)-linkages.

In our second model of synthesis, several glucosyl transferase activities are associated with the synthase; the chain-holding domain positions the nonreducing end of the polymer but exhibits no chain-reading activity. Three UDP-Glc transferase activities would result in cellotriose units and positioning of the nonreducing end of the chain in a conformation that always places the O-3 hydroxyl in the acceptor position (Fig. 7C). In such a model, the higher cellodextrins would be made by occasional failure to fill the third site of substrate binding, resulting in the formation of a cellobiose unit instead of cellotriose. In some instances, the cellobiosyl unit slides through the chain-holding domain and a subsequent complete cellotriosyl unit formation in the next round would result instead in the formation of a cellopentaosyl unit. However, the formation of one cellobiosyl unit must greatly favor the formation of a second cellobiosyl unit to form the more abundant cellotetraosyl unit (Fig. 7C). As the third site is occasionally unfilled, its affinity for UDP-Glc must be lower than that of the cellobiosyl generating sites. In contrast to the single-site model, the ratio of cellodextrin units is dependent on the substrate concentration and not on chain reading. Suboptimal UDP-Glc should result in decreased proportions of cellotriose because of more frequent failure to fill the third site. Our data support this three-site model of glycosyl transfer.

A third model for -glucan synthesis places additional substrate binding sites and transferases to account for the longer cellodextraoyl units (Fig. 7D). These sites would possess affinities for UDP-Glc much lower than the first three sites to account for the smaller proportions of the higher cellodextrin units. In contrast to the three-site model, suboptimal UDP-Glc should result in increased proportions of cellotriose because of more frequent failure to fill the fourth and fifth sites. Our results show that the opposite occurs.

A three-site model of substrate attachment is attractive because a -glucan can be made in vitro almost exclusively with cellotriose units, which are the most abundant units in cereal -glucans (Fig. 1B and 2B). The Poales are the only taxonomic order among angiosperms to make these mixed-linkage -glucans (Smith and Harris, 1999), but the lichen Cetraria islandica also synthesizes a -glucan that is almost exclusively cellotriose units (Wood et al., 1994). Like the maize -glucan made in vitro at very high substrate concentrations, the lichen -glucan contains cellopentaosyl units in greater abundance than the cellotetraosyl units.

The mechanism also includes an apparently absolute requirement for the spacing of the units with a single (13)-linkage. -Glucans with alternating (13)- and (14)-linked glucosyl units are not observed under any conditions. Both of these features demand a cellodextrin unit spacing mechanism, which is provided by the three-site model. This model also predicts that the cellopentaosyl units will also be favored over cellotetraosyl units at near-saturating UDP-Glc concentrations, and this is exactly what we observed (Figs. 1B and 2B).

The three-site model also requires that the formation of the (13)-linkage be restricted to the attachment of the cellodextrin unit to the nonreducing end of the growing chain. If the (13)-linkage were specified at the outer site, then failure to fill the sites would result in a minimum of a cellopentaosyl unit and no possibility to form a cellotetraosyl unit. If the (13)-linkage were specified at the middle site, then failure to fill the outer site would result in a cellobiosyl unit, and these units are never observed under any conditions. In all multiple-site models of synthesis for cellulose and -glucan, loss of all but one of the glycosyl transferase activities would result in the nonreducing end of the chain always being oriented with the O-3 in the acceptor position, which would form callose (Carpita et al., 1996).

Although a "three-site" mechanism of glucosyl transfer is indicated by our data, we certainly have not defined how the active site and the transferase activities are organized. The discovery of a higher plant cellulose synthase gene (CelA) by Pear et al. (1996) has made it possible to compare the deduced amino acid structure of the active site with rational models of catalysis. First, the CelA gene family is extensive and consists of several subclasses (Cutler and Somerville, 1997). A "true" CelA gene encodes a large polypeptide with the presumed active site comprising a large cytosolic domain with four highly conserved domains. Each of the first three contain an essential aspartyl residue and the fourth possesses a QxxRW motif. In addition, there are two plant-specific sequences, one of which exhibits extensive amino acid variability (Pear et al., 1996).

This entirely cytosolic domain is sandwiched between two membrane-spanning domains on the N-terminal side, and four to six membrane-spanning domains on the C-terminal side, which may fold together to form a channel to facilitate extrusion of the glycan chain. The four U motifs of the CelA gene, i.e. the three aspartyl residues and the QxxRW motif, are iterated in the genes encoding several processive glycosyl transferases in which repeating -glycosyl unit structures are synthesized (Saxena et al., 1995). For example, the U motifs are found in hyaluronan synthases (DeAngelis et al., 1993), chitin synthases (Bulawa et al., 1986), and the NodC synthase (Geremia et al., 1994), all of which make consecutive -glycosidic linkages in which one sugar is oriented nearly 180° with respect to each neighboring sugar. Hyaluronan is a repeating disaccharide unit of -D-GlcUA and GlcNAc, whereas chitin and the Nod factor are composed of (14)-D-GlcNAc units. Hyaluronate synthase is a bifunctional transferase, with both glycosyl transferase activities contained within the single polypeptide (Spicer et al., 1997); it remains to be demonstrated if the active site consists of two distinct glycosyl transferases or if a single transferase toggles between two conformation states to produce the two distinct kinds of glycosyl units. By analogy, the -glucan synthase could have a single glucosyl transferase that toggles between three conformational modes to produce the cellotriosyl units at saturating substrate levels.

Other nonprocessive glycosyl transferases contain the first two U motifs but lack the QxxRW motif. One example of this kind of enzyme is the Escherichia coli K5 glycosyl transferase, which makes the polysaccharide N-acetylheparosan with the repeating disaccharide unit (14)-D-GlcA-(14)-D-GlcNAc. Using site-directed mutagenesis, Griffiths et al. (1998) showed that the aspartyl residues of the first two U motifs were essential for the -glycosyl transfer but independent from the UDP-GlcNAc -transferase activity. Experiments with truncated polypeptides showed that the -transferase activity is within the first 300 residues of the N terminus and upstream from the U-1 motif.

This raises the possibility that processive -glycosyl transferases with the D, D, D, and QxxRW motifs may accommodate as many as three glycosyl transferase activities, as was predicted by our kinetic data for the -glucan synthase. However, one important exception is the Agrobacterium tumefaciens (13)-glucan synthase gene. This synthase contains the four U motifs, even though the Glc units are not inverted during synthesis (Stasinopoulos et al., 1999). Such a gene may have derived from an ancestral cellulose synthase, but the encoded synthase may have lost the function of one of the two glycosyl transfer activities. The hypothesis that cellulose synthase and callose synthase are one and the same enzyme (Delmer, 1977) and our hypothesis explaining mechanistically how callose can be formed when one of the two glycosyl transferase activities is lost (Carpita et al., 1996) are both consistent with the finding that cellulose and callose synthases are from related genes.

The diversity of polysaccharide structures encoded by the four U motif structures and the size of the CelA gene family suggest that in plants some members of the CelA gene family synthesize polysaccharides other than cellulose. Virtually all cross-linking glycans of plants possess linear chain structures of -glycosyl units with one sugar inverted almost 180° with respect to each neighbor. These include xyloglucans, glucomannans, arabinoxylans, and the mixed-linked -glucan. The enormity of the CelA gene family suggests that some of the members are involved in synthesis of these noncellulose -glycans, but identification of mutants lacking these genes or experiments to define the biochemical functions of individual CelA genes are just now being undertaken.

The Synthesis of -Glucans in Vivo

Subsequent confirmation of membrane association was made immunocytochemically (Amor et al., 1995). Delmer and Amor (1995) proposed that the association of SuSy represented a UDP-Glc delivery mechanism to cellulose synthase. -Glucan microfibrils were synthesized from Suc and UDP on immobilized tobacco plasma membrane sheets (Hirai et al., 1998). These results strengthen the idea of an association of SuSy directly with plasma membrane synthases. Furthermore, the products were callose, not cellulose, additional evidence that callose synthase is a default activity from cellulose synthesis.

Conditions that greatly lowered UDP-Glc concentration in maize coleoptiles lowered Suc and Glc concentrations to a much lesser extent (Table I). Regardless, concentrations of these sugars were 3 orders of magnitude higher than that of UDP-Glc. In maize, SuSy was detected immunocytochemically in both plasma membrane-enriched and Golgi membrane-enriched fractions, whereas in soybean, the detection at the Golgi membranes was more equivocal (Fig. 6; Table IV). Xyloglucan synthase, the major synthase in dicots such as soybean, does not require SuSy for substrate delivery because the active site of the synthase is fully contained within the lumen of the Golgi apparatus and is dependent on a UDP-Glc translocator for its nucleotide-sugar substrates (Muñoz et al., 1996).

Our results are consistent with the hypothesis that -glucan synthase is derived from an ancestral cellulose synthase and therefore shares features of the mechanism of synthesis. Part of this mechanism may include the association of SuSy to control the supply of UDP-Glc to the active site(s) of glucan synthase. Although we could detect SuSy at the Golgi membranes, it appears not to function in vitro because -glucan synthesis was absolutely dependent on UDP-Glc. In preliminary experiments, we were unable to observe incorporation of label from [U-¹⁴C]Suc into -glucan in vitro in the presence or absence of UDP. We need to explore ways to preserve the SuSy:-glucan synthase interaction in vitro before we can begin to understand how the substrate delivery controls precise ratio determination of the cellodextrin oligomer units in the native -glucan structure.

	FOOTNOTES