Introduction

Arabinose is quantitatively a major component of numerous plant polysaccharides including rhamnogalacturonan I (RG-I), arabinogalactan types I and II, and arabinoxylan (Ridley et al. 2001; O’Neill and York 2003). In these polysaccharides the arabinosyl residues exist predominantly in the α-L-arabinofuranosyl (α-L-Araf) form (Carpita and Gibeaut 1993). α-L-arabinopyranosyl (α-L-Arap) residues have been reported to be quantitatively minor components of arabinans isolated from several plants (Karácsonyi et al. 1975; Capek et al. 1983; Kiyohara et al. 1987; Swamy and Salimath 1991; Odonmažig et al. 1994). Soybean pectin contains galactan carrying an arabinopyranosyl residue at the nonreducing terminal residue (Huisman et al. 2001). A single α-L-Arap residue is also present in the aceric acid-containing side chain of rhamnogalacturonan II (O’Neill et al. 2004).

Immunolocalization studies using antibodies that recognize (1→4)-ß-D-galactan or (1→5)-α-L-arabinan have shown that the distribution of these side chains differs in the walls of different cell types and may be spatially and developmentally controlled (Knox 2002). Arabinan-deficient tobacco callus is brittle and exhibits weaker cell adhesion than wild-type callus (Iwai et al. 2002). Cell wall arabinan in the guard-cell would maintain the wall flexibility (Jones et al. 2003). Nevertheless, the function of the oligosaccharides with arabinosyl residues remains unclear.

Uridine-5′-diphospho-arabinopyranose (UDP-Arap) has been used as glycosyl donor for in vitro studies of arabinose-containing plant polysaccharides. Previous studies using this UDP-sugar and crude membrane preparation from mung bean (Vigna radiata) shoots (Odzurk and Kauss 1972), French bean (Phaseolus vulgaris) callus and hypocotyls (Bolwell and Northcote 1981; Rodgers and Bolwell 1992) provided evidence that arabinosyl residues from UDP-Ara incorporated into polysaccharides. However, neither the structures of the acceptor polysaccharides were fully characterized nor were the ring form, anomeric configuration, and glycosidic linkage of the incorporated arabinosyl residues determined. Nunan and Scheller (2003) have shown that [14C] Ara is transferred from UDP-ß-L-[14C] arabinopyranose onto exogeneous (1→5)-linked α-L-arabino-oligosaccharides by enzymes present in detergent-soluble extracts of mung bean hypocotyl microsomal membranes. These authors also provided evidence that the newly incorporated arabinosyl residue exists in the pyranose ring form. However, as far as we are aware, there are no reports describing plant glycosyltransferases that catalyze the formation of a glycosidic bond between a α-L-Arap and a β-D-Galp residue.

We, therefore, carried out this investigation to determine the activity of arabinopyranose transfer to β-D-Galp residue in plant extracts and to characterize the products formed. We describe that galacto-oligosaccharides containing a single α-L-Arap residue in the nonreducing end are generated by reacting UDP-ß-L-Arap with 2-aminobenzamide (2AB)-labeled (1→4)-linked ß-D-galacto-oligosaccharides in the presence of a detergent-solubilized microsomal membrane fractions of mung bean hypocotyls. We characterize the structure of the mono-arabinosylated oligosaccharides by NMR spectroscopy, MS, and sugar analysis and report some enzymic properties of the arbinopyranosyltransferase.

Materials and methods

Chemical materials and plant

Mung bean (Vigna radiata L. Wilezek) seeds were purchased from Tokita Seed and Plant (Saitama, Japan). 2AB-labeled galacto-oligosaccharides with a degree of polymerization (DP) between 2 and 7 were prepared as described (Ishii et al. 2004). UDP-ß-L-arabinopyranose (UDP-Arap) and UDP-β-D-galactopyranose (UDP-Gal) were obtained from Peptide Inst. Inc. (Osaka, Japan) and Kyowa Hakko Co. (Tokyo, Japan), respectively. Mes, Hepes, and Triton X-100 were obtained from Dojindo Laboratory (Kumamoto, Japan), and Sigma-Aldrich, respectively. α-L-arabinofuranosidase and endo-polygalacturonase (EPG; Megazyme, Ireland) were purchased from Biocon (Nagoya, Japan). All other chemicals and reagents used were purchased from Wako Pure Chemicals (Osaka, Japan).

Preparation of microsomal membranes and solubilization of microsomal membrane

Mung bean (Vigna radiata) seeds were soaked overnight in water and then placed on moist rock fiber and grown for 3 days in the dark at 25°C. The hypocotyls (1.5–2 cm in length from the cotyledons) were used for the preparation of the microsomal membranes as described (Nunan and Scheller 2003). The microsomal membranes (25 mg ml−1 protein) were treated with Triton X-100 (final concentration 0.5%, w/v) containing 150 mM KCl. The suspensions were vortexed and kept on ice for 30 min. The suspensions were then centrifuged for 20 min at 4°C and 100,000 g using a Beckman-Coulter MLA-130 ultracentrifuge rotor. The supernatant was collected and used immediately for subsequent reactions. Total protein in the microsomal membranes and in the solubilized microsomal fraction were determined with bovine serum albumin as a standard using a Protein assay reagent kit and a DC Protein assay reagent kit (Bio Rad, Hercules, CA, USA), respectively, according to the manufacturer’s instructions.

Assay procedure for α-(1→4)-arabinopyranosyltransferase

Arabinopyranosyltransferase activity was measured at 25°C for 20 min unless otherwise specified in a standard reaction mixture (total volume, 30 μl) containing solubilized microsomal fraction (80–100 μg protein), 10 μM 2AB-labeled galactoheptasaccharide (Gal7-2AB), 2 mM UDP-β-L-Arap, 40 mM Mes-KOH (pH 6.5), 25 mM MnCl2, 20 mM NaF, and 160 mM sucrose. The reaction was terminated by the addition of acetic acid (1.5 M, 30 μl) and then boiled for 1 min. The reaction mixture was centrifuged, and the supernatant was diluted with water to about 300 nM solution, and then analyzed by HPAEC with a linear gradient of sodium acetate (0 mM for 2 min, to 250 mM at 60 min in 100 mM NaOH), at a flow rate of 1.0 ml min−1. The products were detected by their fluorescence (λ ex=330 nm, λ em=420 nm). Enzyme activity is expressed as pmol Ara transferred min−1(mg protein)−1, based on the concentration curve of 2AB-Gal4 as the calibration standard. The apparent K m and V max values of AraT as the crude enzyme for the Gal7-2AB were determined at 25°C for 20 min using Gal7-2AB (1–100 μM) and the same concentrations of other components as used in the standard assay mixture. The apparent K max and V max values for UDP-Ara were determined at 25°C for 20 min with UDP-Ara (20–1,000 μM) and the same concentrations of other components as used in the standard assay mixture. The production of Ara1Gal7-2AB was quantified by HPAEC. The enzyme activity toward oligosaccharides with different DPs was determined at 25°C for 20 min by using 2-AB-labeled galacto-oligosaccharides with DP 3–7 (100 μM) and the same concentrations of other components as used in the standard assay mixture. The effect of various cations on the activity was examined by incubations in 25 mM final concentration and the same concentrations of other components as used in the standard assay mixture at 25°C for 20 min. Enzyme assays were performed, at least, in duplicate.

Analysis of reaction product

Transfer reactions were performed in 40 mM Mes-KOH, pH 6.5 (30 μl) containing UDP-Ara (2 mM), Gal6-2AB or Gal7-2AB (0.5 mM), detergent solubilized microsomal fraction (80–100 μg protein), and the same concentrations of the components as used in the standard assay mixture for 2 h at 25°C, and then kept for 16 h at 4°C. At this time the starting galacto-oligosaccharides converted into mono-arabinosyl galacto-oligosaccharides in almost 90% yield. The reaction mixture was then heated for 1 min in a boiling-water bath, centrifuged and the supernatant collected. The products from 50 separate reactions were combined and applied to a Bio-Gel P-2 (1.6 cm×90 cm and a Bio-Gel P-4 (1.6 cm×90 cm, Bio-Rad) column connected in series and eluted with water at 40°C at a flow rate of 0.4 ml min−1. The fluorescence-positive fractions were collected and freeze-dried. A solution of the residue in water (100 μl) was extracted with toluene (100 μl×10) to remove Triton X-100, and the aqueous fraction then freeze-dried. The residue was dissolved in ammonium acetate and separated by normal-phase liquid chromatography (LC) with UV detection at 254 nm. The UV-positive fractions (retention time 17 min and 19 min for Ara1Gal6-2AB and Ara1Gal7-2AB, respectively) were collected manually, concentrated by rotary evaporator, and freeze-dried. These procedures were repeated three times to obtain a sufficient amount of the products for NMR analysis (0.3–0.5 mg). The purified oligosaccharides were analyzed by electrospray-ionization mass spectrometry (ESI-MS), 1H and 13C NMR spectroscopy and by glycosyl residue composition and glycosyl linkage analyses. Normal-phase LC with UV (254 nm) and fluorescence (λ ex=330 nm, λ em=420 nm) detection was performed using a Phenomenex Luna NH2 column (150×4.6 mm Shimadzu GLC, Tokyo, Japan) eluted at 0.4 ml min−1, at 30°C. The column was eluted as follows: eluent A, 50 mM ammonium acetate (pH 4.5) and eluent B, aqueous 90% (v/v) acetonitrile, and a linear gradient of eluent B from 73%(v/v) to 55%(v/v) in 70 min. Normal phase LC-ESI-MS analysis was performed at 0.4 ml min−1 by connecting the outlet of the column to a LCQ classic mass spectrometer (Thermo electron, Waltham, MA, USA). Electrospray-ionization mass spectra were recorded in the positive-ion mode with a spray voltage of 3.5 KV, a capillary voltage of 5.0 V and a capillary temperature of 200°C. Spectra were obtained between m/z 260 and 2,000. For NMR spectroscopic analysis, the transfer product was dissolved in 99.96% isotopically enriched D2O and then freeze-dried. The residue was dissolved in 99.96% enriched D2O. One- and two-dimensional 1H and 13C spectra were recorded at 30°C and 800 MHz with a Brucker Avance 800 NMR spectrometer using the pulse sequences and software provided by the manufacturer (Ishii et al. 2005). The glycosyl residue and glycosyl linkage compositions of the oligosaccharide products were determined as described previously (York et al. 1986). Portions of the oligosaccharides were treated for 2 h at 40°C with α-L-arabinofuranosidase (0.1 unit) in 25 mM Na-acetate, pH 4.6. After heating the reaction mixture in boiling water the products were analyzed by HPAEC.

UDP-Ara and UDP-Gal coincubation with Gal6-2AB

For preparing the reaction product for structural analysis Gal6-2AB (0.1 mM) was coincubated with UDP-Arap (2 mM) and UDP-Gal (2 mM) and the same concentrations of other components as used in the standard assay mixture under the same conditions as those of the large scale preparation of transfer products. The products from ten separate reactions were combined and subjected to the same procedures as those of the large-scale preparation of the transfer products. The purified oligosaccharides were analyzed by normal phase LC-ESI-MS as described above.

Isolation of rhamnogalacturonan I

Alcohol-insoluble residue was prepared from the hypocotyls (1.5–2 cm in length from the cotyledon). The tissues were saponified with 0.1 N NaOH for 4 h at 4°C and then digested with EPG for 16 h at 35°C (Ishii et al. 2001).

The EPG-soluble fraction was digested with α-amylase and dialyzed with molecular cut off 1,000 dialysis tube and then freeze-dried. The EPG-soluble, α-amylase treated material was separated with a Superdex 75 column (1.6 cm×40 cm) eluted with 50 mM HCOONH4 (pH 5.0). The void volumn fraction was manually collected and dialyzed, and then freeze-dried.

Results

Mung bean microsomes contain enzymes that transfer arabinose from UDP-β-L-Arap onto galacto-oligosaccharides

Reacting 2AB-labeled galacto-heptasaccharide (Gal7-2AB) and UDP-β-L-Arap with a detergent-solubilized microsomal fraction from mung bean hypocotyls generated a product whose elution time differed from that of standard Gal7-2AB (35.1 min) and Gal8-2AB (44.0 min; Fig. 1a). The LC-ESI-MS spectrum of the newly formed product contained an ion at m/z 1,427 that corresponds to the [M+Na]+ ion of a compound that is composed of one pentose, six hexoses, a hexitol, and 2AB (Fig. 2a). The product ion spectrum of the sodium adduct ion at m/z 1,427 contained ions at m/z 1,295, 1,133, 971, 809, and 647. These series of ions are the Y n ions (Costello and Vath 1993), and correspond to the sequential loss of a pentose, and a series of hexoses (Fig. 2b). The product ion mass spectrum also contained ions at m/z 1,127, 965, 803, and 641 that correspond to the Z n ions that result from the loss of 2AB-Gal, 2AB-Gal2, 2AB-Gal3 and 2AB-Gal4 (Fig. 2b). These results are consistent with the existence of the oligosaccharide Ara1Gal7-2AB. The results of ESI-MS-MS analysis of the product formed by reacting Gal6-2AB with UDP-β-L-Arap in the presence of the detergent-solubilized microsomal fraction were consistent with the formation of Ara1Gal6-2AB. No peak corresponding to Ara1Gal7-2AB was detected when Gal7-2AB and UDP-β-L-Arap were reacted together in the absence of the solubilized microsomal fraction or in the presence of the heat-denatured microsomal extract (Fig. 1b), or when UDP-β-L-Arap was omitted from the reaction mixture. Taken together these results provide strong evidence that an enzyme that transfers a single arabinose to the nonreducing terminal galactosyl residue of 1,4-linked β-D-galacto-oligosaccharides is present in mung bean hypocotyl microsomes.

Fig. 1
figure 1

HPAEC profiles of the products formed when Gal7-2AB and UDP-Ara are reacted together at 25°C for 20 min with the Triton X-100 solubilized extract of the microsomal fractions from mung bean (Vigna radiata) under the standard conditions. a The reaction mixture contained Gal7-2AB (100 μM), UDP-Ara (2 mM) and the solubilized microsomal fraction (100 μg protein). b Denatured one. The elution position of Gal8-2AB is indicated by an arrow

Full size image
Fig. 2
figure 2

Electrospray-ionization mass spectra of the enzymatically formed Ara1Gal7-2AB. a Positive-ion mode ESI mass spectrum of Ara1Gal7-2AB. b Positive-ion mode MS/MS spectrum of the product ion ([Ara1Gal7-2AB + Na]+, m/z 1427) of Ara1Gal7-2AB

Full size image

The primary structures of enzymically formed Ara1Gal6-2AB and Ara1Gal7-2AB

The primary structures of the enzymically formed Ara1Gal6-2AB and Ara1Gal7-2AB (Fig. 3) were determined by 1H and 13C NMR spectroscopy and by glycosyl residue and glycosyl linkage composition analyses. Ara1Gal6-2AB and Ara1Gal7-2AB in amounts sufficient for these analyses were generated by reacting Gal6-2AB and Gal7-2AB (0.5 mM) with UDP-β-L-Arap (2 mM). The products from fifty separate reactions were combined and purified by size-exclusion and normal-phase liquid chromatography and this procedure was repeated three times to obtain a sufficient amount of the products for NMR analysis (see “Materials and methods”).

Fig. 3
figure 3

Structure of Ara1Gal6-2AB and Ara1Gal7-2AB

Full size image

Ara1Gal6-2AB and Ara1Gal7-2AB were shown, by glycosyl residue composition analysis, to contain Ara:Gal in the molar ratio of 1:3.5 and 1:4.5, respectively (Table 1). Glycosyl linkage composition analysis gave 1,4,5-tri-O-acetyl-2,3,6-tri-O-methylgalactitol (derived from 4-linked galactopyranosyl residue) and 1, 5-di-O-acetyl-2, 3, 4-tri-O-methylarabinitol in the molar ratio of 5.2:1 and 6.5:1 for Ara1Gal6-2AB and Ara1Gal7-2AB, respectively (Table 1). The arabinitol derivative must be formed from a nonreducing terminal arabinopyranosyl residue because it contains O-methyl groups at O-2, O-3, and O-4. No arabinose was released when Ara1Gal6-2AB was treated with a commercial arabinofuranosidase that is known to hydrolyze terminal α-L-Araf residues (data not shown). These results strongly suggest that an arabinopyranosyl rather than an arabinofuranosyl residue is linked to the galacto-oligosaccharides.

Table 1 Characterization of arabinosyltransferase transferred products
Full size table

Additional evidence for the ring form and anomeric configuration of the arabinosyl residue linked to the galacto-oligosaccharides was obtained by 1H and 13C NMR spectroscopic analyses of Ara1Gal6-2AB. The signals in the 1H NMR spectrum of this product (Fig. 4) were assigned (see Table 2) by 2D-double quantum filtered correlation spectroscopy (DQFCOSY), and 2D-total correlation spectroscopy (TOCSY). The doublet at δ4.555 (J1,2 7.6 Hz) is assigned to the H-1 resonance of the nonreducing terminal arabinopyranosyl residue (Arap). This chemical shift value and the magnitude of the J1,2coupling constant are consistent with a α-linkage (Glushka et al. 2003). The chemical shift values of the H-1 of the former reducing Gal residue (residue A) and the H-1s of the internal galactosyl residues (B, I, E, and T) and the magnitude of their J1,2 coupling constants (7.9 Hz) are consistent with a ß-linkage (Agrawal 1992). The signals corresponding to H-2, H-3, and H-4 of the Ara and Gal residues were assigned using DQFCOSY and TOCSY. The two H-5s of the Ara residue and H-5 and two H-6s of the Gal residue were assigned by HSQC and HMBC analyses. The anomeric configuration and ring conformation of an arabinopyranosyl residue is readily determined by measuring its 3JH,H scalar coupling constant (Glushka et al. 2003). The values of 3JH1,H2 and 3JH2,H3 are >7 Hz, which is characteristic of a trans-diaxial arrangement of H-1, H-2 and H-3 in a pyranose ring with a 4C1 chair conformation, and thus confirms the α-Ara p configuration. The 13C NMR spectra of the 2AB-labeled Ara1Gal6 and Ara1Gal7 were assigned using HSQC and HMBC spectroscopy (see Table 3). The HMBC spectrum gave a cross peak between H-1 of Arap and C-4 of Gal residue T, thereby, confirming that the terminal arabinose is linked to O-4 of this Gal residue (Fig. 5). Taken together, these results show that mung bean hypocotyls contain α-(1→4)-arabinopyranosyltransferase (AraT) activity and that this enzyme transfers Arap from UDP-β-L-Ara to O-4 of the terminal Gal residue of exogenous 2AB-labeled galacto-oligosaccharides.

Fig. 4
figure 4

1H NMR spectrum of the enzymatically formed Ara1Gal6-2AB 2AB-labeled Ara1Gal6 is composed of galactitol (R), five internal galactosyl residues (A, B, I, E, and T), and a nonreducing terminal arabinopyranosyl residue (Arap). Signals of 2AB are not shown. (see Fig. 3)

Full size image
Fig. 5
figure 5

Contour plot of a portion of the HMBC spectrum of Ara1Gal6-2AB. Ara1T4 is the cross-peak observed between H-1 of terminal Ara residue (Ara) and C-4 of former terminal Gal residue (T)

Full size image
Table 2 1H Chemical shifts (δ, ppm) and first-order coupling constants (Hz) for compounds Ara1Gal6-2AB and Ara1Gal7-2AB
Full size table
Table 3 13C chemical shifts (δ, ppm) for compounds Ara1Gal6-2AB and Ara1Gal7-2AB
Full size table

Characterization of α-(1→4)-arabinopyranosyltransferase

The transfer of Ara from UDP-β-L-Ara to Gal7-2AB increased with time and was linear over 20 min (Fig. 6a), was dependent on the protein concentration (Fig. 6b) and was maximum at pH 6.5 (Fig. 6c). Maximal enzyme activity was obtained at 20°C (Fig. 6d). The activity increased in the presence of Mn2+ (5.3), but not in the presence of Mg2+ (0.9), Ca2+ (0.6), Co2+ (1.2), Cu2+ (1.3), and Zn2+ (1.1; numbers in parenthesis indicate the relative activity to none as 1.0). The apparent K m values for the Gal7-2AB and for UDP-β-L-Ara were 45 μM and 330 μM, respectively. The apparent V max values for Gal7-2AB and for UDP-β-L-Ara were 200 pmol min−1(mg protein)−1 and 250 pmol min−1(mg protein)−1, respectively. Gal6- and Gal7-2AB are more effective acceptors than Gal4- and Gal5-2AB (Fig. 7). Virtually no arabinose was transferred to Gal3- and Gal2-2AB (Fig. 7). The oligosaccharides with DP more than 8 were not determined because it was not possible to separate completely oligosaccharides with DP 8, 9 and 10 from a mixture by size-exclusion chromatography with BioGelP-2 and P-4 columns.

Fig. 6
figure 6

Activity of AraT in the Triton X-100 soluble fraction from mung bean hypocotyl microsomes. a Time course of the formation of Ara1Gal7-2AB from Gal7-2AB. The membrane extract (100 μg protein) was incubated with 100 μM Gal7-2AB and 2 mM UDP-Ara at 25°C in the presence of 25 mM MnCl2 under the standard conditions. b Effect of membrane-protein concentration on the formation of Ara1Gal7-2AB from Gal7-2AB. The reaction was allowed to proceed at 25°C for 20 min using 10 μM Gal7-2AB and 2 mM UDP-Ara under standard conditions. c Effects of pH on the formation of Ara1Gal7-2AB from Gal7-2AB. Mung bean membrane extract (100 μg protein) was incubated with 10 μM Gal7-2AB and 2 mM UDP-Ara at 25°C for 20 min under the standard conditions. The buffers used to control the reaction pH were 40 mM Mes-KOH (filled circle), Hepes-KOH (open square) and Tris-HCl (filled tirangle). d Effects of temperature on the formation of Ara1Gal7-2AB from Gal7-2AB. Mung bean membrane extract (100 μg protein) was incubated with 10 μM Gal7-2AB and 2 mM UDP-Ara at 25°C for 20 min under the standard conditions

Full size image
Fig. 7
figure 7

Effects of the DP of the galacto-oligosaccharide acceptors on solubilized microsomal AraT activity. The reaction was conducted with solubilized microsome (100 μg protein), 2 mM UDP-Ara and 2AB-labeled galacto-oligosaccharides (100 μM) with DPs between 3 and 7 at 25°C for 20 min. The transferase activities are shown relative to that of the Gal7-2AB. Data are the average of duplicates

Full size image

Galactosyltransferase activity to the galacto-oligosaccharides containing terminal arabinopyranosyl residue

In a previous study we showed that mung bean microsomes contain an enzyme that catalyzes the transfer of Gal from UDP-β-D-Galp to Gal7-2AB to from Gal8-2AB through Gal13-2AB (Ishii et al. 2004). We now provide evidence that Ara1Gal6-2AB, Ara1Gal7-2AB, Ara1Gal8-2AB, and Ara1Gal9-2AB are formed (see Fig. 8) when Gal6-2AB reacts with UDP-β-D-Galp together with UDP-β-L-Arap and the solubilized microsomal fraction. ESI-MS-MS analysis showed that the pentosyl residue is attached to the nonreducing terminal Gal residue of the galacto-oligasaccharides. No peaks were detected that correspond to 2AB-labeled galacto-oligosaccharides that contain more than one pentosyl residue. Moreover, no galactosyl residues were added to Ara1Gal7-2AB when this derivative was reacted with UDP-Gal and the solubilized microsomal fraction (data not shown). Such results show that Ara1Gal7-2AB is not an acceptor substrate for galactosyltransferase (GalT) and suggests that the presence of Arap prevents the elongation of galactan chains.

Fig. 8
figure 8

HPEAC profile and ESl-MS spectra of the products formed when Gal6-2AB (100 μM), UDP-Ara (2 mM) and UDP-Gal (2 mM) are reacted with solubilized microsomal fractions (100 μg protein). a Fluorescence detection of the enzymatically formed 2AB-labeled mono-arabinosylated galacto-oligosaccharides. b Positive-ion mode ESl mass spectra of 2AB-labeled mono-arabinosylated galacto-oligosaccharides with DP 6–10. c Positive-ion mode MS/MS spectra of the sodium adduct ion [M + Na]+ of 2AB-labeled mono-arabinosylated galacto-oligosaccharides with DP 6–10

Full size image

Mung bean hypocotyls contain nonreducing terminal Arap Residue

Rhamnogalacturonan I was isolated from EPG solubilized fraction of mung bean hypocotyls. Glycosyl linkage analysis gave 2-linked and 4-linked Rha residues (Table 4), showing that the polysaccharide was RG-I and that the side chains are attached to the backbone at O-4 of the Rha residues (Ridley et al. 2001). The presence of 4-linked Gal and 5-linked Araf in molar ratio of 1:12, indicated that the side chain contained a short linear (1→4)-linked galactan and a long linear (1→5)-linked arabinan. Terminal Arap and terminal Ara f were detected in molar ratio of 2:7, indicating that terminal arabinopyranosyl residues are present in the side chain of RG-I.

Table 4 Glycosyl-linkage analysis of RG-I from 3-day-old mung bean
Full size table

Discussion

We have shown that mung bean microsomes contain an enzyme that transfers the arabinopyranosyl residue from UDP-ß-L-arabinopyranose onto 1,4-linked β-D-galacto-oligosaccharides. Our chemical and spectroscopic data show unequivocally that a single α-L-arabinopyranosyl residue is linked to O-4 of the terminal Gal residue of the galacto-oligosaccharide. Such results are consistent with a previous report showing that mung bean hypocotyls contain arabinopyranosyltransferases (Nunan and Scheller 2003).

Studies of the biosynthesis of arabinose-containing plant polysaccharides are complicated by the fact that most of the arabinose residues exist in the α-L-Araf form (Carpital and Gibeaut 1993) yet the only readily available glycose donor for in vitro studies is UDP-β-L-Arap. Early studies using this UDP sugar and crude membrane preparations provided evidence for the incorporation of arabinosyl residues into polysaccharides (Odzurk and Kauss 1972; Bolwell and Northcote 1981). However, neither the structures of the acceptor polysaccharides were fully characterized nor were the ring form, anomeric configuration, and glycosidic linkage of the incorporated arabinosyl residues determined. Our data together with the results of Nunan and Scheller (2003) show that mung bean microsomes contain enzymes that add a single α-L-Arap residue to exogenous α-L-(1→5) linked arabino-oligosaccharide and β-D-(1→4) linked galacto-oligosaccharides. RG-I isolated from 3-day-old mung bean hypocotyls contained nonreducing terminal Arap residues. Some terminal Arap residues detected here might be derived from the arabino-oligosaccharides. Some plant species have been reported to synthesize polysaccharides that do contain quantitatively small amounts of this pyranosyl residue. For example, terminal Arap residues have been detected in arabinans from white willow (Karácsonyi et al. 1975), marsh mallow (Capek et al. 1983), and Mongolian larchwood (Odonmažig et al. 1994). Arap residues are believed to terminate some of the galactose-containing side chains of soybean RG-I (Huisman et al. 2001). Thus, we believe our data is consistent with the notion that Arap-containing oligosaccharides are formed naturally in plants (Nunan and Scheller 2003) and that this process is likely to account for some of Arap that are present in RG-I of mung bean hypocotyls.

Present results show that mung bean hypocotyls contain an enzyme that catalyzes the formation of a glycosidic bond between α-L-Arap and O-4 of the nonreducing terminal Gal of 1,4-linked β-D-galacto-oligosaccharides. Such a reaction is likely to be catalyzed by an α-(1→4)-arabinopyranosyltransferase. However, there will be some possibilities that Arap-containing oligosaccharides form. One possibility is that plants contain other enzymes that convert Arap to Araf, in addition to transferase. Thus, the enzymes that convert UDP-β-L-Arap to UDP-β-L-Araf, if they exist, may be labile and loose activity when they are solubilized with detergent. In this study only Triton X-100 was used for solubilization of the membranes. However, when a range of different detergents were used to study transfer of arabinose to arabino-oligosaccharides only Arap was detected in the product (Nunan and Scheller 2003). Indeed, reacting crude microsomal membranes of wheat (Triticum aesstivum) with UDP-β-L-[14C] Arap results in the generation of endogenous polysaccharides that contain [14C] Araf residues. However, only Arap residues are transferred to exogenous acceptors using detergent-soluble extracts of these membranes (Porchia et al. 2002). Studies with bacteria have shown that UDP-L-Araf and UDP-L-Arap are interconverted by a UDP-galactopyranose mutase (Houseknecht and Lowary 2001). However, the reaction favors the formation of the thermodynamically more stable pyranose because only quantitatively small amounts of UDP-Araf are formed when the mutase is reacted with UDP-Arap (Zhang and Liu 2001). Moroever, there is no evidence showing that such mutases exist in plants. As neither UDP-L-arabinofuranose nor mutase were available, it was not possible to determine the activity using UDP-L-arabinofuranose as a donor. Another possibility is the involvement of galactopyranosyltransferase. UDP-β-L-Arap and UDP-β-D-Galp differ only in the nature of the substituents at C-5 of the glycose (Fig. 9). L-Arap has two protons whereas D-Galp has a proton and a hydroxymethyl group. Thus, a galactosyltransferase may be capable of forming both a β-D-Galp and a α-L-Arap glycosidic linkage because it does not readily distinguish between UDP-β-D-Galp and UDP-β-L-Arap. However, it would be not possible to distinguish arabinopyranosyltransferase activity from the apparent activity due to that galactopyranosyltransferase may be able to transfer Arap from UDP-Arap when the activity is determined by using detergent-solubilized protein fractions.

Fig. 9
figure 9

Reaction scheme for formation of galacto-oligosaccharide and mono-arabinosylated galacto-oligosaccharide

Full size image

In conclusion, an α-(1→4)-arabinopyranosyltransferase activity has been solubilized from microsomal membranes prepared from mung bean hypocotyls. It transfers a single arabinopyranosyl residue from UDP-ß-L-arabinopyranose onto 2AB-labeled galacto-oligosaccharides. The transferred arabinose is α-L-arabinopyranosyl form linked to O-4 of the terminal galactopyranosyl residue of 2AB-galacto-oligosaccharides. The nonreducing terminal α-L-arabinopyranosyl residue of galacto-oligosaccharides terminates galactan chain elongation. Using biochemical and bioinformatic strategies, identification and purification of α-(1→4)-arabinopyranosyltransferases will be helpful for understanding the function of arabinopyranosyl residue in galactan.