INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Pectic polysaccharides are major components of primary plant cell walls and middle lamella. Three main types of pectic polysaccharides have been identified, homogalacturonan, rhamnogalacturonan I (RG I), and rhamnogalacturonan II (RG II). RG I is composed of a backbone of alternating [4--D-GalA-(12)--L-Rha-(1] in which some of the rhamnose residues are substituted with complex side chains containing Ara and Gal residues. The composition and size of these side chains varies depending on the plant species, tissue type, and stage of development (O'Neill et al., 1990; Carpita and Gibeaut, 1993). In general, the arabinan side chains consist of a linear chain of (15)-linked -L-arabinofuranosyl residues that may be substituted at the O-3 and/or O-2 with additional arabinofuranosyl residues (Carpita and Gibeaut, 1993; Schols and Voragen, 2002).

The study of the biosynthesis of plant cell wall polysaccharides in plants is a growing field, but to date, very few genes encoding glycosyltransferases have been characterized (Edwards et al., 1999; Perrin et al., 1999; Faik et al., 2002). Notably, many different transferases must be involved in pectin biosynthesis, but until very recently, none had been isolated or cloned, most likely due to the structural complexity of pectic polysaccharides. The transfer of Ara onto cell wall polysaccharides is a prime example of complexity. Ara residues are present in arabinoxylans and a number of different pectic polysaccharides such as RG II and the arabinan and arabinogalactan side chains of RG I (Ridley et al., 2001; Schols and Voragen, 2002). Furthermore, Ara is also abundant in glycoproteins such as arabinogalactan proteins and extensins (Gaspar et al., 2001). Focusing on pectins only, it has been predicted that 15 arabinosyltransferases may be involved in the synthesis of pectin based on the one linkage-one enzyme assumption (Mohnen, 1999, 2002). Therefore, it is not surprising that the characterization of arabinosyltransferases has been such a slow and laborious process. Functional genomics has now enabled significant progress in identifying important genes. Thus, a gene that appears to encode a glucuronosyltransferase involved in the biosynthesis of RG II has recently been reported (Iwai et al., 2002). Furthermore, a glycosyltransferase with undetermined specificity but apparently involved in pectin biogenesis was also reported recently (Bouton et al., 2002). Although advances such as these serve to greatly increase our understanding of pectin biosynthesis, the limited number of methods for determining specific transferase activities in vitro prevents the ultimate assignment of function to the identified glycosyltransferases.

NDP-sugars appear to be the glycosyl residue substrates for the biosynthesis of cell wall polysaccharides (Mohnen, 1999, 2002). The first report of an arabinosyltransferase activity possibly involved in pectin biosynthesis was in mung bean (Vigna radiata) shoots and used UDP-Ara as the substrate (Odzuck and Kauss, 1972). Later work using membrane preparations from French bean (Phaseolus vulgaris) callus and hypocotyls and radiolabeled UDP-Ara yielded a synthesized product that was degraded by pectinase, suggesting that Ara was being incorporated into a pectic polysaccharide (Bolwell and Northcote, 1981). However, because pectinase preparations are relatively undefined mixtures of enzymes, the exact type of polysaccharide into which the label was incorporated was not confirmed. Continuation of this work resulted in the partial purification of a protein with an approximate molecular mass of 70 kD (Rodgers and Bolwell, 1992), but no amino acid sequence of this protein has been reported. A monoclonal antibody that binds to a protein of this size was shown to inhibit the incorporation of Ara into a product (Bolwell and Northcote, 1984), lending greater support that the 70-kD protein is somehow involved in the transfer of Ara onto polysaccharides, but again, the nature of the polysaccharide acceptor(s) was never characterized.

Extensive studies on arabinan biosynthesis in mycobacteria have recently been carried out and although there are some apparent similarities between the bacterial and plant systems, there are also some obvious differences (Houseknecht and Lowary, 2001). In mycobacterial cell walls, the Ara residues are in the furanose ring confirmation as are the majority of Ara residues in plant cell walls. However, D-arabinans typify the cell walls of mycobacteria (McNeil and Brennan, 1991), whereas L-arabinans are present in plants (Mohnen, 1999). Furthermore, although UDP--L-arabinopyranose is the source of L-arabinofuranosyl residues in plants, UDP-arabinopyranose and UDP-arabinofuranose do not appear to be the donor of Ara in mycobacteria. Instead, it has been shown that the immediate donor of Ara in mycobacteria is -D-arabinofuranosyl-L-monophosphoryldecaprenol (Xin et al., 1997).

In this report, we follow the transfer of [¹⁴C]Ara from UDP--L-[¹⁴C]arabinopyranose onto endogenous products using microsomal membranes isolated from mung bean hypocotyls. Additionally, we describe for the first time to our knowledge a biochemical assay using detergent solubilized microsomal membranes and well-defined (15)-linked -L-arabino-oligosaccharides as acceptor molecules to study the activity of an arabinosyltransferase.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Arabinosyltransferase Activity Using Endogenous Acceptors

Microsomal membranes prepared from etiolated 2-d-old mung bean hypocotyls were incubated with UDP--L-[¹⁴C]Ara and the incorporation reaction was stopped by precipitation with methanol/chloroform. After centrifugation, the pelleted material was not washed to remove unincorporated substrate due to the difficulties associated with precipitating arabinans (Fry, 1988). The use of high concentrations of ethanol to ensure complete precipitation of arabinans resulted in incomplete removal of unincorporated substrate (results not shown). Pretreatment of the pellets with sodium hydroxide was necessary to ensure that maximum solubilization of high molecular mass polysaccharides was achieved. The solubilized material was filtered before separation by size exclusion chromatography (SEC). Most of the radiolabeled products eluted in the void volume, indicating a size greater than 25 kD according to the calibration of the column using Dextran standards. Fractions nine through to 19 (which represented the radiolabeled material) were collected, combined, dried, and subjected to enzymatic digestion before separation by SEC (Fig. 1A). Arabinofuranosidase (Megazyme, Bray, Ireland) was the only enzyme able to partly degrade the radiolabeled product (Fig. 1A). The enzyme released Ara corresponding to about 12% to 15% of the total radioactivity. This arabinofuranosidase can release (12)- and (13)-linked -L-arabinofuranosyl residues from branched arabinan and can also partly degrade linear (15)-linked -L-arabinofuranosyl chains (results not shown). Additionally, the enzyme also removes arabinosyl residues from arabinoxylan. Therefore, it could not be unambiguously determined from what sort of polysaccharide(s) the released Ara had originated. The endo-arabinanase, which is specific for the degradation of linear (15)-linked -L-arabinosyl chains, did not release any radiolabeled products, which suggested that the radiolabeled sugar was not attached to linear (15)-linked -L-arabinan. However, arabinans in mung bean cell walls could be highly branched and, therefore, the enzyme may be unable to gain access to the (15)-linked -L-arabinan backbone. Proteinase K also failed to release any radiolabeled components from the endogenous products, suggesting that no proteins were being labeled. However, when an incorporation reaction, which had only been precipitated with chloroform:methanol, was heated with an SDS/dithiothreitol (DTT) loading buffer and separated by SDS-PAGE, a radiolabeled protein band at approximately 42 kD was seen (results not shown). This is likely to be the same protein as the carbohydrate-binding protein reported by Bolwell (1986) and may also be related to a reversibly arabinosylated protein observed in wheat (Porchia et al., 2002). The 42-kD protein was obviously not seen on the SEC column due to its poor solubility.

View larger version (28K): [in this window] [in a new window]   Figure 1.   Analysis of radiolabeled endogenous products after incubation of microsomes with UDP-L-[14C]Ara. A, Digestion of high molecular mass products with no enzyme (), arabinanase (), arabinofuranosidase from Megazyme (), and proteinase K () followed by SEC. The number of counts in the collected fractions were determined by scintillation counting. Similar results were obtained in three separate experiments. The elution of Dextran standards and arabinobiose is indicated. B, Total hydrolysis and separation by TLC of fraction 10 (F10) and fraction 11 (F11) collected after SEC to determine sugar composition. The TLC plate was exposed to a phosphorimager screen to identify the radiolabeled sugars. The purity of the substrate used was confirmed by hydrolysis and TLC of the UDP-[14C]Ara substrate (Ara) and of UDP-[14C]Xyl (Xyl).

When two of the SEC fractions, fraction 10 (F10) and fraction 11 (F11), were hydrolyzed and separated by thin-layer chromatography (TLC), radiolabeled Ara and Xyl were detected (Fig. 1B), which indicated the presence of UDP-Xyl-4-epimerase activity. The presence of epimerase activity in microsomal preparations from mung beans as well as from other plant species has been reported previously (Odzuck and Kauss, 1972; Bolwell and Northcote, 1981; Porchia et al., 2002). A number of procedures were attempted to remove this epimerase activity, such as extensive washing of the microsomal preparation and fractionation of the preparation by Suc gradients, but enough of the epimerase activity remained so that radiolabeled Xyl was present in the endogenous product (results not shown). In addition, the UDP-Ara was quite rapidly broken down to Ara-1-phosphate and Ara (results not shown), indicating the presence of phosphodiesterase and phosphatase activities. Returning to the digestion of the radiolabeled product by arabinofuranosidase (Fig. 1A), it can now be concluded that because only one-half of the radiolabel is Ara, approximately 25% of the radiolabeled Ara was released by the enzyme.

Arabinosyltransferase Activity Using Exogenous Acceptors

To avoid the complication of analyzing products that contained [¹⁴C]Ara and [¹⁴C]Xyl, it was decided to add well-defined exogenous acceptors to the incorporation reaction with the hope that only radiolabeled Ara would be added. The microsomal preparation was first solubilized in detergent to allow access of the acceptors to the individual enzymes. A number of detergents was tested such as octyl glucoside, CHAPS, and reduced Triton X-100 at various concentrations. Additionally, a number of different potential acceptors were tested such as arabinan, debranched arabinan, galactan, RG I fragments prepared as previously described (Geshi et al., 2002), and short arabinofurano-oligosaccharides. The most successful combination was 1% (w/w) octyl glucoside and (15)-linked -L-arabino-octaose (Ara8; Fig. 2). The amount of radiolabel detected in the endogenous product peak was greatly reduced when detergent solubilized microsomal membranes were used (Fig. 2) in comparison with unsolubilized membranes (Fig. 1A). Due to the inability to precipitate the arabino-oligosaccharides, the fractions where the unincorporated substrate elutes (fractions 22-26) have been excluded from the SEC graphs. When the Ara8 acceptor was added to the reaction mix, there was a stimulation of incorporation at approximately the size of the added acceptor. With a concentration of UDP-Ara of 100 µM, approximately 19 pmol of Ara was incorporated into the 28 nmol of Ara8 acceptor present. It should be noted that the acceptor was in excess. After solubilization, activity was only detected in the supernatant fraction, and nothing in the pellet material indicating the arabinosyltransferase had successfully been released from the membranes. Complete acid hydrolysis of the product followed by TLC showed that Ara was the only radiolabeled sugar present (results not shown). Thus, even if UDP-Xyl-4-epimerase activity was present, the Ara8 oligosaccharide did not act as acceptor for Xyl. Other detergents were also successful with the Ara8 acceptor, but to a lesser degree (results not shown). Smaller oligomers of (15)-linked -L-arabinosyl residues were also tested (Fig. 3), but there was decreasing incorporation with decreasing size. The large polysaccharides (arabinan, debranched arabinan, linear arabinan, galactan, and RG fragments) were unsuccessful as acceptor molecules (results not shown). Obviously, the optimal acceptor size has not been found and therefore it would be very interesting to test arabino-oligomers larger than an octamer, but unfortunately, such oligomers are not currently available.

View larger version (14K): [in this window] [in a new window]   Figure 2.   Acceptor dependency of Ara incorporation. Detergent-solubilized microsomes were incubated with UDP-L-[14C]Ara in the presence () or absence () of Ara8 acceptor. The reaction mixture was then subjected to SEC and the collected fractions were counted by scintillation counting. The data represents the average of three separate experiments ± SE. The elution of Dextran standards and arabino-octaose is indicated.

View larger version (10K): [in this window] [in a new window]   Figure 3.   The effect of different (15)--L-arabino-oligosaccharides as acceptor molecules on levels of incorporation. Detergent-solubilized microsomes were incubated with UDP-L-[14C]Ara and various (15)--L-arabino-oligosaccharides. After separation by SEC, the acceptor-dependent incorporation was determined by scintillation counting. The data represents the average of three separate experiments ± SE. Ara5, Arabinopentaose; Ara6, arabinohexaose; Ara7, arabinoheptaose; Ara8, arabino-octaose.

Optimization of Ara Transfer onto Arabino-Oligosaccharides

Once it was established that detergent-solubilized microsomal membranes incorporated Ara onto arabino-octaose acceptors, optimization of incorporation was undertaken (Fig. 4). In all optimization experiments, 1% (v/v) octyl glucoside was used to solubilize the membranes. The age of the hypocotyl had a dramatic effect on incorporation, with the youngest tissue of 2 d after planting giving greatest incorporation of the label (Fig. 4A). The transferase activity was also pH dependent, with pH 6.5 as the optimal pH (Fig. 4B). The length of time that the incorporation was allowed to proceed also had an effect (Fig. 4C). There was linear incorporation of the label onto the acceptor for the first 20 min. Extended incubation time after 20 min had no improved effect on the amount incorporated. This was probably due to the substrate being broken down as well as the transferase being unstable at 30°C for an extended period of time. Another important requirement was Mn²⁺ ions. If MnCl₂ was not present in the reaction, then no incorporation was seen, but the inclusion of 1 to 15 mM MnCl₂ resulted in incorporation (results not shown). This requirement has previously been reported for arabinan biosynthesis in mung bean shoots (Odzuck and Kauss, 1972). Using optimal conditions, variable amounts of the microsomal preparation (equivalent to 0-175 µg of protein) were included in the reaction. The amount of incorporation was proportional to the amount of protein, with some leveling off at the higher protein amounts (Fig. 5A). To show substrate dependency, variable amounts of UDP-L-Ara were included in the incorporation reaction. At concentrations up to 200 µM UDP-L-Ara, the amount of incorporation of Ara was essentially proportional to the amount of UDP-L-Ara added to the reaction, but at 700 µM, the highest concentration used, the reaction was almost saturated (Fig. 5B). By fitting the data to the Michaelis-Menten equation, an apparent K_m of 0.33 ± 0.10 mM UDP-L-Ara can be estimated. However, it should be recalled that the breakdown and conversion of UDP-Ara is quite rapid and, therefore, until a more pure preparation of the arabinosyltransferase is obtained, the actual K_m cannot be accurately calculated. In a previous study, a partially purified arabinosyltransferase was reported as having an apparent K_m of 178 ± 45 µM (Bolwell and Northcote, 1981).

View larger version (15K): [in this window] [in a new window]   Figure 4.   Optimization of incorporation of [14C]Ara onto Ara8 acceptor. Only the counts that were acceptor dependent are shown. A, The effect of incorporation using mung bean hypocotyls of different ages based on the time from planting. The data represents the average of three separate experiments ± SE. B, pH optimum of incorporation of [14C]Ara onto Ara8 using 50 mM MES (), Tris/HCl (), and phosphate buffers (). C, Time dependence of incorporation of [14C]Ara onto Ara8. Percentage incorporation at each time point was calculated from the maximum dpm reached at 50 min (absolute values range between 78 and 168 dpm). The data represents the average of four separate experiments ± SE.

View larger version (15K): [in this window] [in a new window]   Figure 5.   Dependency of solubilized arabinosyltransferase activity on the amount of enzyme and substrate. Only the counts that were acceptor dependent are shown. A, Variable amounts of microsomal membrane preparation (0-175 µg of protein) were added to the incorporation reaction. B, Variable concentrations of UDP-Ara were added to the incorporation reaction as a mixture of UDP-[14C]Ara (0-10 µM) and unlabeled UDP-Ara (0-700 µM). The reactions contained 150 µg of protein and were incubated for 15 min in the presence of 0.56 mM Ara8 acceptor. The incorporation was calculated and the data fitted to the Michaelis-Menten equation by nonlinear regression.

The Arabinosyltransferase Activity Is Associated with Golgi Vesicles

Membranes isolated by separation on a discontinuous Suc gradient were probed with antibodies specific for Golgi and endoplasmic reticulum (ER) marker proteins (see "Materials and Methods"). The upper interface (0.25/1.1 M Suc) was enriched in Golgi, independent of whether the Suc solutions contained EDTA or not (Fig. 6A). When probed with the ER antibody, the two phases without EDTA appear to contain similar amounts of ER and when EDTA was included in the solution, there was a definite shift of ER into the upper phase (Fig. 6B). When arabinosyltransferase activity was studied in the two interface fractions, the majority of activity was found to be in the upper phase, which indicates that the transferase activity is associated with Golgi vesicles (Fig. 6C). It should be noted that, due to the high amount of ER present in the upper phase in the sample containing no EDTA, it cannot be totally excluded that some of the transferase activity is also associated with the smooth ER as also reported by Bolwell and Northcote (1983). Initiation of synthesis could possibly occur in the ER.

View larger version (35K): [in this window] [in a new window]   Figure 6.   Subcellular localization of the arabinan arabinosyltransferase activity. Microsomal membranes were fractionated on a Suc gradient in the presence and absence of 1 mM EDTA and the interfaces at 0.25/1.1 M and 1.1/1.3 M Suc were collected and subjected to SDS-PAGE before immunoblotting. A, Blot probed with an antibody against the reversibly glycosylated protein, which is representative of a Golgi protein. B, Blot probed with an antibody against calnexin/calreticulin, which are representative of ER proteins. C, Arabinosyltransferase activity in each membrane preparation. Detergent-solubilized membranes obtained at the interfaces of 0.25/1.1 M and 1.1/1.3 M Suc were incubated with UDP-L-[14C]Ara and Ara8. The reaction was subjected to SEC and the amount of acceptor-dependent incorporation was determined by scintillation counting. Lane 1, 0.25/1.1 M Suc interface, no EDTA included; Lane 2, 1.1/1.3 M Suc interface, no EDTA included; Lane 3, 0.25/1.1 M Suc interface, EDTA included; Lane 4, 1.1/1.3 M Suc interface, EDTA included.

Analysis of the [¹⁴C]Arabino-Oligosaccharide Products

When the radiolabeled product was subjected to SEC, it eluted only slightly earlier than the Ara8 acceptor (Fig. 2). Ara8 is quite well separated from Ara6, thus, it would be expected that a decamer would elute significantly earlier than Ara8. Therefore, we conclude that not more than one Ara unit was being added. Separation of the product by TLC was in agreement with this conclusion (results not shown). However, neither method gives any indication of how the single [¹⁴C]Ara residue was attached to the Ara8 acceptor molecule.

One way to determine how this [¹⁴C]Ara residue was linked to the acceptor was to digest the radiolabeled product with a number of different enzymes. When the Ara8 acceptor was digested with endo-arabinanase [specific for (15)-linked -L-arabinofuranosyl residues], the product was mostly a dimer with small amounts of monomer. When the radiolabeled product was digested under the same conditions, small radioactive products were released, which eluted as a trimer or larger (Fig. 7A). It has been reported that this arabinanase digests linear arabinans down to a trimer (Skjøt et al., 2001), but under the conditions used here, the Ara8 acceptor was digested to mainly a dimer, presumably due to the different conditions for digestion compared with those used by Skjøt and coworkers or because the substrate used by those authors contained branch points. Therefore, the digestion pattern of the radiolabeled product suggests that the [¹⁴C]Ara was not a (15)-linked -L-arabinofuranosyl residue. When digested with arabinofuranosidase C [specific for (12)- and (13)-linked -L-arabinofuranosyl residues; Scott et al., 1997] and therefore unable to digest the Ara8 acceptor), no change in the size of the radiolabeled product was seen (Fig. 7B), which suggests that the [¹⁴C]Ara was not a (12)- or (13)-linked arabinofuranosyl residue. Finally, digestion with arabinofuranosidase from Megazyme, which acts on (12)-, (13)-, and (15)-linked -L-arabinofuranosyl residues, also produced a trimer and tetramer (Fig. 7C). Under the same conditions, this enzyme digests the Ara8 acceptor completely to Ara. Overall, the digestions indicate that no radiolabeled (12)-, (13)-, or (15)-linked -L-arabinofuranosyl residues were being attached to the acceptor.

View larger version (19K): [in this window] [in a new window]   Figure 7.   Enzymatic characterization of the radiolabeled product obtained with Ara8 as acceptor. After incubation of UDP-[14C]Ara with detergent-solubilized microsomes, the reaction was separated by SEC and fractions 17, 18, and 19 (i.e. the Ara8 acceptor-dependent peak) were collected and combined. This sample was then divided into four aliquots and digested with no enzyme (), endo-arabinanase (; A), arabinofuranosidase C from Danisco (; B), or arabinofuranosidase from Megazyme (; C) before separation by SEC. Radioactivity in the fractions was determined by scintillation counting. Similar results were obtained in two separate experiments. The elution of Ara (Ara1), arabinotriose (Ara3), and arabino-octaose (Ara8) is indicated.

One possibility was that an arabinopyranosyl residue was being attached instead of an arabinofuranosyl residue, which would explain why the enzymes could not release the [¹⁴C]Ara. This was easily tested by treating the radiolabeled product with mild acid, which will hydrolyze the glycoside bond between two furanose residues but will not hydrolyze the bond of a pyranose residue attached to a furanose residue (Fry, 1988). The mild acid treatment efficiently hydrolyzed the acceptor completely to Ara. In contrast, mild acid treatment of the radiolabeled product produced a dimer, whereas treatment with strong acid was required for complete hydrolysis (Fig. 8), indicating that the [¹⁴C]Ara is an arabinopyranosyl residue.

View larger version (18K): [in this window] [in a new window]   Figure 8.   Determination of the ring conformation of the radiolabeled Ara residue by acid hydrolysis. After incubation of UDP-[14C]Ara with detergent-solubilized microsomes, the reaction was separated by SEC and fractions 17, 18, and 19 (i.e. the Ara8 acceptor-dependent peak) were collected and combined. This sample was then divided into three aliquots and treated with mild acid (0.1 M trifluoroacetic acid [TFA], 100°C; ), strong acid (2 M TFA, 100°C; ), or no acid () followed by SEC. Similar experiments were obtained in two separate experiments. The elution of Ara (Ara1), arabinobiose (Ara2), and arabino-octaose (Ara8) is indicated.

To further confirm that the radiolabeled Ara was in the pyranose conformation, a procedure for linkage analysis was used that separates partially methylated arabinosyl residues by TLC (Hayashi and Maclachlan, 1984). Methylation of free hydroxyl groups and the subsequent hydrolysis of the oligosaccharide allow the identification of hydroxyl groups that were involved in a linkage. Combined with the reduction of the sugars to their corresponding alditols, these procedures enable the separation and identification of differently linked sugars. When the radiolabeled product was methylated and hydrolyzed, the labeled compound had the same R_F value as a terminal Ara in the pyranose conformation that had been methylated (Fig. 9). When the product was methylated and reduced, the radiolabeled compound migrated the same distance as the methylated and reduced terminal arabinopyranosyl residue, confirming the results found with the mild acid hydrolysis. The linkage analysis clearly excludes the possibility that an arabinofuranosyl residue carries the radiolabel. Notably, the methylated arabinofuranosylwhether terminal or substitutedmigrates more slowly upon reduction, whereas methylated arabinopyranosyl migrates faster upon reduction (Fig. 9). Unfortunately, it was not possible to determine how the arabinopyranosyl residue was attached to the arabino-oligosaccharide due to the limited amount of material available. It is known that the label is not attached to the reducing end of the acceptor because when the acceptor was reduced with sodium borohydride before being added to the reaction, no effect on incorporation was detected (results not shown).

View larger version (10K): [in this window] [in a new window]   Figure 9.   Determination of the ring conformation of the radiolabeled Ara by linkage analysis. After incubation of UDP-[14C]Ara with detergent-solubilized microsomes, the reaction was separated by SEC and fractions 17, 18, and 19 (i.e. the Ara8 acceptor-dependent peak) were collected and combined. This sample was divided into two aliquots and methylated or methylated and reduced (see "Materials and Methods") before separation by TLC. The radiolabeled sugars were detected by cutting the lanes on the TLC plate into 2.5-mm strips and counting the strips by scintillation counting. The elution profiles for the methylated only () and the reduced methylated () products are shown. Position of standards: 1 = 2,3,4-tri-O-methyl-Ara ( methylated terminal arabinopyranose) and 2,3-di-O-methylarabinitol ( reduced methylated C5-substituted arabinofuranose); 2 = 2,3-di-O-methyl-Ara ( methylated C5-substituted arabinofuranose), 2,3,4-tri-O-methylarabinitol ( reduced methylated terminal arabinopyranose) and 2,3,5-tri-O-methylarabinitol ( reduced methylated terminal arabinofuranose); 3 = 2,3,5-tri-O-methyl-Ara ( methylated terminal arabinofuranose).

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Previous to this report, products being synthesized in microsomal preparations using UDP-Ara as a substrate have not been well characterized. Odzuck and Kauss (1972) reported arabinosyltransferase activity in mung bean microsomal preparations, but did not characterize the type of polymer onto which the labeled Ara was being added. Extensive work on an arabinosyltransferase has been carried out using microsomal membranes from French bean, but the product was only characterized using a heterogeneous pectinase preparation and, therefore, the authors were never able to fully characterize the acceptor molecule nor was the conformation of the Ara residue determined (Bolwell and Northcote, 1981, 1983, 1984). Difficulties in characterizing the synthesized product were also encountered here when using intact mung bean microsomal preparations and relying on endogenous acceptors, and even the use of specific enzymes that are available today could not significantly improve on this characterization. The Megazyme arabinofuranosidase, which can act on (12)-, (13)-, and (15)-linked -L-arabinofuranosyl residues, released approximately 25% of the radiolabeled Ara. Treatment of high M_r polysaccharides with a single hydrolytic enzyme is never completely efficient, so the indication is that a significant proportion of the labeled sugars in the synthesized endogenous product were (12)-, (13)-, and/or (15)-linked -L-arabinofuranosyl residues. The arabinanase, which is specific for (15)-linked -L-arabinofuranosyl residues, released no radiolabeled products. These digestion patterns suggest that the product(s) being synthesized are highly branched arabinans or other polysaccharides not associated with pectin. The breakthrough came when solubilized membranes were used and well-defined arabino-oligosaccharides were included as exogenous acceptor molecules. Because (15)-linked -L-arabinans are generally considered to be a component of the side chains of RG I, it is likely that this solubilized arabinosyltransferase activity is involved with pectin biosynthesis.

Upon solubilization, the ability of the mung bean membrane preparation to transfer Ara onto an endogenous acceptor was greatly reduced. The addition to the incorporation reaction of (15)-linked -L-arabino-oligosaccharides as acceptors stimulated the transfer of radiolabeled Ara onto these acceptor molecules, suggesting that it was the loss of endogenous acceptors that resulted in reduced incorporation. The success of using oligosaccharides as acceptor molecules in solubilized microsomal systems has also been reported for other transferases, including homogalacturonan galacturonosyltransferase (Doong and Mohnen, 1998), galactomannan galactosyltransferase (Edwards et al., 1999), xylan xylosyltransferase (Kuroyama and Tsumuraya, 2001), and xyloglucan xylosyltransferase (Faik et al., 2002). The addition of large polysaccharides such as debranched arabinan and galactan did not show a stimulation of incorporation using membranes prepared from mung bean hypocotyls. Interestingly, in a report on the solubilization and partial purification of an arabinosyltransferase from French bean suspension-cultured cells, incorporation of labeled Ara using solubilized membranes into an endogenous product was achieved and no acceptor molecule was required (Rodgers and Bolwell, 1992).

One question raised by the results presented in this paper is whether arabinopyranose attached to (15)-linked arabinan is a naturally occurring component of mung bean cell walls. A comprehensive study of cell walls isolated from mung bean hypocotyls has not been made. Instead, there have been studies focusing on readily soluble pectic polysaccharides from mung bean hypocotyls, which only reported the presence of (15)-linked -L-arabinofuranosyl residues, but no arabinopyranosyl residues and no (12)- or (13)-linked -L-arabinofuranosyl residues, which are characteristic of branched arabinans (du Penhoat et al., 1987; Goldberg et al., 1989). However, a number of other plant species, including some closely related to mung bean, have arabinopyranosyl residues present in the wall. An arabinan isolated from pigeon pea (Cajanus cajan) cotyledons was found to be highly branched, having a (15)-linked -L-arabinofuranosyl backbone with O-2- and O-3-linked Ara residues attached, and, notably, some of the terminal Ara residues were in the pyranose conformation (Swamy and Salimath, 1991). In spinach (Spinacia oleracea), feruloylated pectins have been shown to contain domains with the structure feruloyl-arabinopyranose-(arabinofuranose)_n-Ara where n = 0 to 7 (Fry, 1983). Of most interest is that this oligosaccharide was unbranched and therefore is very similar to that being synthesized in the mung bean preparation reported here. However, a different study found that the Ara in feroylated oligosaccharides from sugar beet (Beta vulgaris) were all arabinofuranosyl residues as determined by NMR spectroscopy and that arabinopyranose residues were only observed as artifacts caused by recyclization of the arabinofuranose under acidic conditions (Colquhoun et al., 1994). Arabinopyranosyl residues also exist in acidic arabinogalactan from Angelica acutiloba in which highly branched Ara chains possess some four-linked arabinopyranosyl residues (Kiyohara et al., 1987). Pectic arabinogalactan from soybean (Glycine max) is composed of (14)-linked Gal residues bearing an arabinopyranosyl residue as the nonreducing terminal residue (Huisman et al., 2001), and in arabinogalactan type II from Japanese larch (Larix leptolepis), there exists the structure 3-O--L-arabinopyranosyl-L-Ara (Aspinall et al., 1968). Taken together, these studies suggest that arabinopyranose is a normal component of arabinans and galactans but has often been overlooked due to the much higher occurrence of arabinofuranose.

Alternatively, arabinopyranosyl residues may not be an intrinsic part of the cell wall. For example, the arabinopyranose residue may only be present in arabino-oligosaccharides while being synthesized and transported from the site of synthesis, but then the arabinopyranosyl residue is removed or modified before insertion of the oligosaccharide into the cell wall. Therefore, detection of arabinopyranose in this system would be almost impossible. Arabinopyranosyl groups are very uncommon in small glycosides, but the possibility that the arabinosyltransferase has a noncarbohydrate substrate in vivo may be considered. We find it unlikely because the arabinosyltransferase clearly prefers the larger oligomeric substrates.

Considering that arabinofuranosyl residues are the predominant form of Ara found in plant cell wall polysaccharides, it is surprising that there was no incorporation of arabinofuranosyl residues onto the arabino-oligosaccharide acceptor molecules. This may indicate that the transferase does not perform ring conversion and sugar transfer, as suggested earlier (Fry and Northcote, 1983). Potentially, this theory could be tested by the addition of UDP-L-arabinofuranose directly to the reaction mix or by the addition of a UDP-galactopyranose mutase that has the ability to convert UDP-L-arabinopyranose to UDP-L-arabinofuranose (Zhang and Liu, 2001). However, no homolog of UDP-galactopyranose mutase from Escherichia coli appears to be encoded by the genome of Arabidopsis or any other plant. Alternatively, an intermediate step may bind the UDP-arabinopyranose and perform the ring conversion. In the intact mung bean membrane system, it appears that at least 25% of the incorporated Ara exists in the furanose conformation. Additionally, using UDP-L-[¹⁴C]arabinopyranose and intact wheat (Triticum aestivum) microsomal or Golgi membranes, [¹⁴C]arabinofuranose was incorporated into several products (Porchia et al., 2002). Together, these results suggest that upon solubilization of the microsomal membranes, an important component is lost that is essential for the conversion of arabinopyranose to arabinofuranose. Possibly, the transferase is present in a complex with another enzyme, which performs the ring conversion, but upon solubilization, this complex is disrupted.

How the arabinopyranose is attached to the arabino-oligosaccharide still needs to be ascertained. This task is very difficult due to the small quantities of radiolabeled oligosaccharide produced, and, therefore, more traditional techniques for determining polysaccharide linkages such as gas chromatography-mass spectrometry of methylated polysaccharides are not possible. Degradation using specific enzymes is a sensitive technique used to determine linkage type within a polysaccharide. Unfortunately, the currently available enzymes, arabinofuranosidases and arabinanases, were not suitable in this case and arabinopyranosidases have not yet been purified and characterized. Other techniques such as matrix-assisted laser desorption ionization/time of flight may offer an alternative and more sensitive approach in the attempt to determine the type of linkage.

Many organisms have now had the complete sequence of their genomes determined, and the number of nucleotide and amino acid sequences for putative glycosyltransferases involved in polysaccharide biosynthesis has been increasing rapidly. Many strategies to identify these transferases can be applied (Perrin et al., 2001), and it could be expected that results will come quickly. However, because no L-arabinosyltransferases have been identified in any organism, the bioinformatics approach will be difficult. Assumptions need to be made on deciding which candidate genes could be arabinosyltransferases, such as whether it will be a retaining versus an inverting enzyme, whether it will have a nucleotide binding sequence motif, and whether it will have homology to similar plant transferases such as D-galactosyltransferases.

In conclusion, an arabinosyltransferase activity has been solubilized from microsomal membranes prepared from mung bean hypocotyls. It transfers an arabinopyranosyl residue from UDP-Ara onto arabino-oligosaccharides. Future experiments will focus on determining the nature of the glycosidic linkage between the pyranose residue and the acceptor molecule. Additionally, using biochemical and bioinformatic strategies, identification and purification of arabinosyltransferases will be attempted.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant Material

Mung bean (Vigna radiata) seeds were soaked overnight in water and spread onto moist vermiculite before being grown in darkness at 24°C for 2 to 6 d. The root and cotyledons were removed using a scalpel, and the hypocotyl tissue was used for the preparation of microsomal membranes.

Preparation of Microsomal Membranes

All procedures were carried out at 4°C unless stated otherwise. Hypocotyls (3-5 g) from 2- to 6-d-old mung bean seedlings were gently ground using a mortar and pestle in 20 mL of homogenization buffer consisting of 5 mM phosphate buffer, pH 7.5, 0.4 M Suc, and 1 mM DTT. The suspension was filtered through nylon mesh (30 µm) to remove cell debris and was centrifuged at 3,000g for 10 to 15 min. The supernatant was transferred to a new tube and centrifuged at 50,000g for 1 h. The microsomal pellet was resuspended in homogenization buffer (normally 500 µL). The protein concentration of the microsomal preparation was determined using Bradford Coomassie Blue reagent (Bradford, 1976) and using bovine serum albumin as the protein standard.

Incorporation into Endogenous Product

UDP-L-[¹⁴C]arabinopyranose with specific activity of 9.9 GBq mmol¹ was prepared from UDP-D-[¹⁴C]xylopyranose (NEN Life Science Products, Boston) as described previously (Pauly et al., 2000). Reaction mixtures contained freshly prepared microsomal membranes (approximately 175 µg of protein), 1 µM UDP-L-[¹⁴C]Ara (approximately 25,000 dpm), 50 mM potassium phosphate buffer, pH 6.5, and 3 mM MnCl₂ in a final volume of 50 µL. Reactions were performed at 30°C for 30 min. The reactions were stopped by the addition of 1 mL of chloroform:methanol (3:2, v/v) and then vortexed. After centrifugation at 15,000g for 5 min at 4°C, 200 µL of 1 N NaOH was added to the pellets, and the samples were vortexed and placed at 4°C for 16 h. The samples were centrifuged and the supernatants were transferred to new tubes before being neutralized with glacial acetic acid. After centrifugation at 20,000g for 5 min, the collected supernatants were spin-filtered using 0.2-µm nylon filters (Lida Manufacturing [Nalg Nunc], Rochester, NY). SEC was performed using a Superdex Peptide gel filtration column (30 cm long, 1 cm i.d.; Amersham Pharmacia Biotech, Uppsala). The filtered samples were applied to the column and eluted with 50 mM ammonium formate, pH 6.5, at a flow rate of 0.4 mL min¹. Fractions were collected every 2 min and 3 mL of scintillation fluid (Ecoscint; National Diagnostics, Manville, NJ) was added to each vial before quantification by scintillation counting (Wallac 1414 Liquid Scintillation Counter; EG&G, Turku, Finland). The molecular masses of eluted polysaccharides were estimated by comparing retention times with Dextran molecular mass standards (Fluka, Buchs, Switzerland).

Solubilization of Microsomal Membranes

Microsomal membranes were prepared as described above. The suspension of microsomal membranes was solubilized at a protein concentration of 5 mg mL¹ with a final detergent concentration of 1% (w/v) octyl -D-glucopyranoside (Sigma Chemical Company, St. Louis), and a final KCl concentration of 150 mM. The samples were vortexed lightly and incubated on ice for 10 min. The solubilized membranes were centrifuged at 100,000g for 30 min and the supernatant was collected and used immediately for subsequent incorporation reactions.

To optimize incorporation, the solubilized microsomal preparation was added to reaction mixtures containing buffers of various pH values (5.0-9.0), various MnCl₂ concentrations (0-15 mM), and various concentrations of nonradioactive UDP-Ara (0-700 µM). A number of different acceptor molecules was tested, including arabinan, debranched arabinan, and linear arabinan from sugar beet (Beta vulgaris), galactan from lupin, and (15)--L-arabino-oligosaccharides (all from Megazyme International), as well as RG acceptors (RG-B), which were a kind gift from Dr. Naomi Geshi and were prepared as described previously (Geshi et al., 2002). The arabinan was approximately 18 kD and consisted mainly of a (15)-linked -L-arabinofuranosyl backbone, and 60% of the Ara residues were substituted with single (13)- and possibly (12)-linked arabinofuranosyl residues. Approximately 10% (w/v) of the arabinan preparation consisted of Gal, rhamnose, and GalUA residues. The debranched arabinan was approximately 18 kD and had been obtained by treating the above arabinan with arabinofuranosidase, which removed all of the (13)- and (12)-linked arabinosyl residues. The linear arabinan was approximately 18 kD and had been obtained by treating the debranched arabinan with chloroacetic acid resulting in a polysaccharide consisting mainly of linear chains of (15)-linked -L-arabinofuranosyl residues of approximately 18 kD from which most of the charged pectic fraction had been removed. The galactan consisted mainly of (14)-linked -D-galactosyl residues with some rhamnose, Ara, and Xyl residues. The (15)--L-arabino-oligosaccharides were linear chains of arabinofuranosyl residues. The RG acceptors were approximately 21 kD and contained mainly uronic acid, rhamnose, and Gal residues (Geshi et al., 2002).

Using optimized conditions, solubilized microsomal membranes (30-35 µL, equivalent to 150-175 µg of protein) prepared from 2-d-old mung bean hypocotyls were incubated for 30 min at 30°C with 1 µM UDP-L-[¹⁴C]Ara, 50 mM potassium phosphate buffer, pH 6.5, 3 mM MnCl₂, and 30 µg of (15)--L-arabino-octaose (equivalent to 0.56 mM) in a final reaction volume of 50 µL. Reactions were terminated by the addition of chloroform:methanol and were further treated as described for the endogenous products. To show dependency of the arabinosyltransferase activity on the amount of enzyme supplied, variable amounts of the solubilized microsomal preparation (0-175 µg of protein) were included in the reaction mixture, whereas other components and conditions remained the same as described above. To show substrate dependency, variable amounts of UDP-L-[¹⁴C]Ara (0-10 µM) and variable amounts of cold UDP-Ara (100, 200, and 700 µM) were included in the reaction mixture, which contained the same components as described above with the exception that only 150 µg of protein was included. The reaction was stopped after 15 min by the addition of chloroform:methanol and was treated as described above.

Preparation of Golgi Vesicles

Membrane preparations enriched in Golgi vesicles were prepared essentially as described previously (Muñoz et al., 1996) with minor modifications. The Suc solutions used for the gradients consisted of 50 mM Tris-HCl buffer, pH 7.5, 1 mM DTT, ±1 mM EDTA, and various concentrations of Suc. All procedures were carried out at 4°C unless otherwise stated. Mung bean hypocotyls (4-5 g) were gently pressed using a mortar and pestle to minimize breakage of Golgi vesicles in the presence of 10 mL of buffer containing 50 mM Tris/HCl buffer, pH 7.5, 0.5 M Suc, 1 mM DTT, and ±1 mM EDTA. The suspension was filtered through nylon mesh (30 µm) and centrifuged at 3,000g for 10 min. The supernatant (approximately 10 mL) was loaded onto a 4-mL 1.3 M Suc solution and was centrifuged at 100,000g for 1 h and 30 min. The upper phase was carefully removed so as not to disturb the material at the interface. A discontinuous gradient was formed by adding 5 mL of 1.1 M and 4 mL of 0.25 M Suc solutions to the surface of the bottom phase. The gradient was centrifuged at 100,000g for 1 h and 40 min and the interfaces at 0.25/1.1 M and 1.1/1.3 M Suc were collected. The vesicles were solubilized immediately using a final concentration of 1% (w/v) octyl glucoside and were subsequently used in incorporation reactions as described above for exogenous acceptors. The identity and purity of the two fractions were determined by immunoblotting using polyclonal antibodies raised against reversibly glycosylated protein from pea (Pisum sativum; Dhugga et al., 1997), which is representative of a Golgi protein, and calnexin/calreticulin from barley (Hordeum vulgare; Møgelsvang and Simpson, 1998), which is representative of ER proteins. The immunoblots were visualized using secondary antibodies conjugated with horseradish peroxidase (Dako, Glostrup, Denmark) and a chemiluminescence detection kit (Amersham Pharmacia Biotech).

Analysis of Radiolabeled Products

To determine which type of sugars were radiolabeled, the high molecular mass material collected after SEC representing incorporation into endogenous products (eluted in fractions 10 and 11) or the combined lower molecular mass material incorporated into Ara8 (eluted in fractions 17, 18, and 19) were freeze dried and then hydrolyzed in 2 M TFA for 1 h at 120°C. The TFA was evaporated and the samples were resuspended in water before separation by TLC (Silica Gel 60 F₂₅₄ plates; Merck, Darmstadt, Germany) with ethyl acetate:methanol:acetic acid:water, 12:3:3:2 (v/v). The radiolabeled sugars were determined by exposing the TLC plate to a phosphorimager screen and were detected using a Storm PhosphorImager (Molecular Dynamics, Sunnyvale, CA). The position of the standard sugars, Ara and Xyl, was determined by dipping the plate in 10% (v/v) sulfuric acid in ethanol and charring the plate.

Additionally, enzymatic digestion of the ¹⁴C-labeled products was performed. The collected peaks after SEC were digested with a number of different enzymes. These included -L-arabinofuranosidase from Aspergillus niger (0.025U, 1U releases 1 µM Ara from sugar beet arabinan min¹; Megazyme International), endo-arabinanase from Aspergillus aculeatus (0.05 U, 1U releases 1 µM of Ara from sugar beet arabinan min¹; Novozymes, Copenhagen), -L-arabinofuranosidase C from A. niger (0.05U, 1U releases 1 µM of Ara from sugar beet arabinan min¹; Danisco Biotechnology, Copenhagen), and Proteinase K (20 µg per reaction; Boehringer Mannheim, Mannheim, Germany). The final reaction volumes were 120 µL and the digestions were performed overnight in 50 mM ammonium formate containing 0.02% (w/v) sodium azide, pH 4.5 for the -L-arabinofuranosidases (40°C), and pH 5.5 for endo-arabinanase (30°C) and in phosphate buffer, pH 6.5, for proteinase K (40°C). Nonenzymatic treatments of samples were in 50 mM ammonium formate containing 0.02% (w/v) sodium azide, pH 4.5 (40°C). The digests were then separated by SEC and the collected fractions were counted by scintillation counting.

The radiolabeled oligosaccharides were treated with various concentrations of acid to determine sugar ring conformation. The collected peak from SEC for the exogenous acceptor was treated with water (for 1 h at 100°C), mild acid (0.1 M TFA for 1 h at 100°C), or strong acid (2 M TFA for 1 h at 120°C). The samples were then separated again by SEC and the collected fractions were counted by scintillation counting.

Methylation and reduction of the radiolabeled oligosaccharides were performed as described previously (Nunan et al., 1997). Briefly, collected radiolabeled peaks from four reactions were methylated using NaOH/CH₃I and were hydrolyzed using 2.5 M TFA for 4 h at 100°C. Some of the hydrolyzed samples were then additionally reduced using 1 M NaBH₄ in 1 M NH₄OH. The dried samples (methylated only or methylated and reduced) were resuspended in water and separated by TLC as described by Hayashi and Maclachlan (1984). Standards were prepared from sugar beet arabinan and arabinotriose (Ara3; Megazyme) and arabinopyranosyl myo-inositol (Sigma).