Trans-α-xylosidase and trans-β-galactosidase activities, widespread in plants, modify and stabilize xyloglucan structures




Cell-wall components are hydrolysed by numerous plant glycosidase and glycanase activities. We investigated whether plant enzymes also modify xyloglucan structures by transglycosidase activities. Diverse angiosperm extracts exhibited transglycosidase activities that progressively transferred single sugar residues between xyloglucan heptasaccharide (XXXG or its reduced form, XXXGol) molecules, at 16 μm and above, creating octa- to decasaccharides plus smaller products. We measured remarkably high transglycosylation:hydrolysis ratios under optimized conditions. To identify the transferred monosaccharide(s), we devised a dual-labelling strategy in which a neutral radiolabelled oligosaccharide (donor substrate) reacted with an amino-labelled non-radioactive oligosaccharide (acceptor substrate), generating radioactive cationic products. For example, 37 μm [Xyl-3H]XXXG plus 1 mm XXLG-NH2 generated 3H-labelled cations, demonstrating xylosyl transfer, which exceeded xylosyl hydrolysis 1.6- to 7.3-fold, implying the presence of enzymes that favour transglycosylation. The transferred xylose residues remained α-linked but were relatively resistant to hydrolysis by plant enzymes. Driselase digestion of the products released a trisaccharide (α-[3H]xylosyl-isoprimeverose), indicating that a new xyloglucan repeat unit had been formed. In similar assays, [Gal-3H]XXLG and [Gal-3H]XLLG (but not [Fuc-3H]XXFG) yielded radioactive cations. Thus plants exhibit trans-α-xylosidase and trans-β-galactosidase (but not trans-α-fucosidase) activities that graft sugar residues from one xyloglucan oligosaccharide to another. Reconstructing xyloglucan oligosaccharides in this way may alter oligosaccharin activities or increase their longevity in vivo. Trans-α-xylosidase activity also transferred xylose residues from xyloglucan oligosaccharides to long-chain hemicelluloses (xyloglucan, water-soluble cellulose acetate, mixed-linkage β-glucan, glucomannan and arabinoxylan). With xyloglucan as acceptor substrate, such an activity potentially affects the polysaccharide’s suitability as a substrate for xyloglucan endotransglucosylase action and thereby modulates cell expansion. We conclude that certain proteins annotated as glycosidases can function as transglycosidases.


Xyloglucan is the major hemicellulose in primary cell walls of most land plants (Albersheim et al., 2010; Fry, 2011a). It hydrogen bonds to microfibril surfaces, coating them as a monolayer. An individual xyloglucan chain may adhere to multiple microfibrils through various segments along its length, tethering the microfibrils (Fry, 1989; McCann et al., 1990; Hayashi and Kaida, 2011), or locally penetrate them, firmly fastening the ends of the tethers (Pauly et al., 1999). Xyloglucan thereby contributes to wall architecture.

Xyloglucan may also generate oligosaccharins (biologically active oligosaccharides) (Darvill et al., 1992; Aldington and Fry, 1993). It undergoes partial hydrolysis in vivo, yielding apoplastic xyloglucan-derived oligosaccharides (XGOs) (McDougall and Fry, 1991), some of which, at least when supplied exogenously, modulate plant growth and morphogenesis (York et al., 1984; McDougall and Fry, 1989, 1990; Rakitin et al., 2001; Kaida et al., 2010).

Xyloglucan may thus play both structural and signalling roles, and enzymes that modify this polysaccharide or its oligosaccharides are therefore highly significant in plant biology.

Xyloglucan has a (1→4)-β-d-glucan backbone carrying α-d-xylose side chains. The major disaccharide unit [α-d-xylosyl-(1→6)-d-glucose; ‘isoprimeverose’] is abbreviated as X; glucose residues lacking side chains are G (Fry et al., 1993). Fifteen elaborations of X are known (Peña et al., 2008), including L [β-d-galactosyl-(1→2)-α-d-xylosyl-(1→6)-d-glucose] and F [α-l-fucosyl-(1→2)-β-d-galactosyl-(1→2)-α-d-xylosyl-(1→6)-d-glucose]. Common sequences include XXXG, XXFG and XLFG.

One reason why the structure of this polysaccharide is important concerns xyloglucan endotransglucosylase (XET) activity, which re-models xyloglucan chains in vitro (Fry et al., 1992; Nishitani and Tominaga, 1992; Rose et al., 2002; Eklöf and Brumer, 2010). XET action has also been demonstrated in vivo (Ito and Nishitani, 1999; Vissenberg et al., 2000; Mellerowicz et al., 2008), during both wall integration of new xyloglucan (Thompson et al., 1997) and re-structuring of existing material (Thompson and Fry, 2001). XET activity contributes to the control of wall extensibility and plant growth (Maris et al., 2009; Sasidharan et al., 2010; Harada et al., 2011; Miedes et al., 2011). The precise structure at the non-reducing terminus (usually X) of a potential acceptor substrate is crucial for XET activity (Lorences and Fry, 1993; Fanutti et al., 1996). Therefore, enzymes that modify this structure can alter the ability of xyloglucan to serve as an XET acceptor substrate.

Another reason why modifying xyloglucan structures is biologically interesting concerns oligosaccharins. For example, removal of the l-fucose residue inactivates XXFG as an oligosaccharin (York et al., 1984; McDougall and Fry, 1989). Two mechanisms of XXFG loss in vivo are enzymic hydrolysis (by α-fucosidase, α-xylosidase, β-glucosidase and β-galactosidase) (Augur et al., 1992, 1993; Iglesias et al., 2006) and sequestration by XET action (Baydoun and Fry, 1989). Enzymes that add, move or detach xyloglucan sugar residues can potentially create, modify or destroy biological signals.

We recently screened extracts from diverse plants, and reported numerous glycosidase and glycanase activities that potentially act on plant cell walls (see GHATAbase in Franková and Fry, 2011). Interestingly, five of the hydrolases screened showed accompanying transglycosylating activities that generated products with a higher molecular weight than the substrate. For example, sub-millimolar concentrations of (1→4)-β-d-xylohexaose underwent both endo- and exo-transglycosylation, demonstrating trans-β-xylanase and trans-β-xylosidase activities, respectively (Franková and Fry, 2011). Similar assays suggested the presence of trans-α-d-xylosidase and trans-α-l-arabinosidase activities acting on xyloglucan- and arabinan-derived oligosaccharides, respectively. Sampedro et al. (2010) also reported that an Arabidopsisα-d-xylosidase may catalyse transxylosylation in addition to its established hydrolytic role. Likewise, an XGO-acting trans-β-d-glucosidase has been described (Crombie et al., 1998).

This study explores the relocation of xyloglucan monosaccharide residues by transglycosidase activities – an interesting third mechanism, in addition to hydrolysis and ‘sequestering’, by which the biological activities of XGO oligosaccharins and the ability of long-chain xyloglucans to act as XET substrates in vivo may be modified.


A minority of plant glycosidases catalyse detectable transglycosylation on aryl glycosides

Of seven cauliflower glycosidase activities assayed on 1.4 mm nitrophenyl (NP) glycosides, only two (β-xylosidase and β-glucosidase) were accompanied by transglycosylation (Figure S1), transiently forming traces of NP disaccharides (e.g. Xyl-NP + Xyl-NP  →  Xyl-Xyl-NP + NP), in addition to the monosaccharide. The others (α-glucosidase, α- and β-galactosidase, and α- and β-mannosidase) produced free monosaccharides as the only detectable product.

Plant extracts catalyse exo-transglycosylation of XXXG

The model substrate NP-α-xyloside was not hydrolysed by enzymes from cauliflower (Figure S1) or soya bean (Glycine max) (Koyama et al., 1983). However, incubating diverse plant extracts with 1.4 mm XXXG (heptasaccharide) yielded larger products with lower mobility on thin-layer chromatography (TLC), indicating transglycosylation, in addition to the smaller products expected as a result of hydrolase action (Figure 1). Products larger than XXXG were regularly spaced on the TLC plate, suggesting degrees of polymerization (DP) of 8, 9 and 10, corresponding to XXXG with one, two or three additional monosaccharide residues, respectively (Figure 1). On the basis of staining intensity, the yields were DP8 > DP9 > DP10, indicating sequential addition of monosaccharide residues, e.g. DP7 + DP7 ↔ DP8 + DP6, DP7 + DP8 ↔ DP9 + DP6 and DP7 + DP9 ↔ DP10 + DP6. This observation indicates transglycosidase rather than transglycanase action (which would have added oligosaccharide segments, as with trans-β-xylanase; Franková and Fry, 2011). Most plant extracts tested catalysed these transglycosidase reactions, including dicots and poalean and non-poalean monocots. The reaction was least prominent in three non-angiosperms: Equisetum, Selaginella and Marchantia (Figure 1).

Figure 1.

 Products formed by the action of plant extracts on XXXG.
NaCl-extracted enzymes were dialysed, and incubated with 1.4 mm XXXG at 22°C for 24 h. Products (3 μg) were analysed by TLC. Blue text indicates hydrolysis products or small products left after loss of a sugar residue during transglycosylation; red text indicates transglycosylation products formed by adding sugar residue(s) to XXXG or one of its major smaller products; et, etiolated; hypo, hypocotyl; sdlg, seedling; spr, sprout.

The major products smaller than XXXG co-migrated with XXG, XG, glucose and xylose – a pattern expected for slow α-xylosidase action coupled with rapid β-glucosidase action (Koyama et al., 1983; Iglesias et al., 2006; Franková and Fry, 2011). The short-lived intermediates GXXG, GXG and cellobiose are faintly visible. Both XXG and XG were accompanied by slower-migrating compound(s), presumably transglycosylation products based on XXG or XG. Each major oligosaccharide (XXXG, XXG and XG) appeared to be capable of accepting one additional sugar residue per existing xylose residue (Figure 1).

To resolve some of the activities, we (NH4)2SO4-fractionated cauliflower proteins (Figure 2). Some fractions (50–70% saturated (NH4)2SO4 cuts) catalysed extensive conversion of XXXG to larger oligosaccharides, whose concentrations eventually exceeded those of unreacted XXXG. Release of free glucose by these cuts greatly exceeded that of free xylose, indicating that most xylose cleavage (a pre-requisite for subsequent β-glucosidase action) is by transxylosylation rather than hydrolysis. The cuts showing maximal XET activity were 20 and 40% saturated (NH4)2SO4 (data not shown), confirming that XXXG-acting transglycosidase activity is not an XET side reaction.

Figure 2.

 Products formed by the action of cauliflower leaf (NH4)2SO4 cuts on XXXG.
NaCl-extracted enzymes were dialysed (unless marked ‘crude’), and sequentially (NH4)2SO4-precipitated (unless marked ‘no AS-pptn’). ‘% AS’, percentage saturation of (NH4)2SO4. Each fraction was incubated with XXXG and analysed as in Figure 1. Each lane contained products generated by 0.7-9.4 μg protein from 1.5-84 mg fresh weight of cauliflower tissue, as indicated. Green text indicates endogenous plant sugars.

A time-course analysis (Figure S2a) showed that DP8–10 transglycosylation products formed gradually and concurrently with hydrolysis. Another time-course analysis, in which a more active enzyme preparation was used (Figure S2b), showed that the transglycosylation products were longer-lived than XXXG or its hydrolysis products (XXG and XG). For example, XXXG was 90% consumed within approximately 1 h, whereas the DP8 and DP9 products required approximately 18 h for 90% disappearance.

To study the effect of substrate concentration on transglycosylation versus hydrolysis, we used [1-3H]XXXGol of various specific radioactivities. The initial radioactivity in the reaction mixture was always 1.0 kBq μl−1, but the total concentration was varied between 8 and 4096 μm (Figure 3a–d). Selected bands, or pools of related bands, were quantified (Figure 3e–i). Using 1.4 mm XXXGol instead of 1.4 mm XXXG made little difference to the TLC pattern (Figures 2 and 3). β-Glucosidase action gradually shortened the oligosaccharide backbone, most rapidly (when expressed as a percentage of starting material) at low initial substrate concentrations: the Glc4-based pool (XXXGol plus DP8 and DP9 transglycosylation products) gradually decreased between 0 and 12 h (Figure 3e), the Glc3-based pool was formed and lost with kinetics typical of metabolic intermediates (Figure 3f), and the Glc1–2-based pool appeared after a lag period and with kinetics typical of end products (Figure 3g). Transglycosylation was detectable concurrent with this β-glucosidase action, even at substrate concentrations as low as 16 μm, as revealed by rapid DP8 production (Figure 3h). Thus the transglycosidase(s) catalysed detectable transglycosylation in the presence of a 3 × 106-fold molar excess of H2O, the ‘acceptor substrate’ for hydrolysis. The percentage yield of DP8 increased at higher concentrations (optimally ≥ 256 μm), as the oligosaccharide acceptor substrate out-competed H2O.

Figure 3.

 Action of cauliflower lamina enzymes on [3H]XXXGol at various concentrations.
The 50-60% saturated (NH4)2SO4 cut (Figure 2) was incubated with [1-3H]XXXGol (1.0 kBq μl−1) at various specific radioactivities, producing 8–4096 μm final concentrations. After 1 h (a), 3 h (b), 6 h (c) and 12 h (d), a 5-μl aliquot was analysed by TLC; fluorograms are shown. Lane 8(C) contains [1-3H]XXXGol incubated without enzyme. The origin, which showed negligible radioactivity, has been omitted, except in (d).
All bands were quantified for 3H (as illustrated for example in Figure S3), and regions of interest are shown in (e)–(i) as percentage of the total radioactivity in the lane: (e) Glc4-based oligosaccharides (XXXGol, DP8 and DP9); (f) Glc3-based oligosaccharides (XXGol and DP6); (g) Glc1–2-based compounds (XGol, isoprimeveritol and glucitol); (h) DP8 alone; (i) DP9 alone. The grey dashed line indicates radioactivity recorded for four runs of unreacted [1-3H]XXXGol plus its trace impurities (the SE is arbitrarily plotted at 5 μm).

As transglycosylation is reversible whereas hydrolysis is irreversible, the latter will ultimately predominate. Therefore, at later time points, the concentration of DP8 decreased. However, DP8 loss was much slower than XXXGol loss: with 256 μm initial substrate, the concentration of DP8 exceeded that of remaining XXXGol 1.37-fold at 6 h. Thus, the newly introduced bond rendered DP8 more stable to cauliflower enzymes. DP9 (Figure 3i) and DP10 (Figure 3c,d) were also produced, but somewhat later and more slowly than DP8.

XGOs carrying galactose residues were also substrates for transglycosidases from cauliflower and numerous other species (Figure S4, and data not shown). Free galactose was released by β-galactosidase, mainly through conversion of XLLG to XXLG, which accumulated (Franková and Fry, 2011). Each major substrate (XLLG, XXLG and XXXG) was accompanied by its own series of slightly slower-migrating transglycosylation products: XXLG produced prominent DP9 and DP10 products (Figure S4).

In conclusion, plants possess transglycosidase activities that transfer monosaccharide residues between XGOs, and the presence of galactose side chains does not interfere with this process.

Xylose, not glucose, residues are transferred by the transglycosidase

To test whether the transferred monosaccharide is xylose, we devised a dual-labelling strategy using a neutral donor substrate possessing radiolabelled xylose residues and a non-radioactive acceptor substrate, XGO-NH2 (amino-labelled XGO). Transxylosylation products would be both cationic and radioactive (Figure 4a), as revealed by high-voltage paper electrophoresis (HVPE) (Figure S5a) and scintillation counting. This strategy, optimized for exo-transglycosylases, is related to the dot-blot approach used for endo-transglycosylases (Fry, 1997; Kosík et al., 2010).

Figure 4.

 Dual-labelling strategy detecting transglycosidase activities.
The principle is illustrated for trans-α-xylosidase and trans-β-galactosidase. A radiolabelled neutral substrate, either [Xyl-3H]XXXG (a) or [Gal-3H]XXLG (b), was mixed with an amino-labelled (cationic) non-radioactive substrate (e.g. XXLG-NH2). Transfer of a radiolabelled residue to the latter generates a ‘hybrid’ product that is both cationic and radioactive. Radioactive sugars: black stars, [3H]xylose; black circles, [3H]galactose. Non-radioactive sugars: white stars, xylose; white squares, glucose; white circles, galactose. Grey square with plus symbol, non-radioactive 1-amino-1-deoxyglucitol.

Extracts of five species were assayed with 37 μm [Xyl-3H]XXXG as donor and 0, 1 or 10 mm XGO-NH2 as acceptor. Radioactive cationic products were formed by all extracts only when XGO-NH2 was present. Asparagus products are shown (Figure 5a–c); the other species gave similar results (data not shown).

Figure 5.

 Products formed by action of asparagus enzymes on [Xyl-3H]XXXG plus XGO-NH2.
(a–c) NaCl-extracted asparagus enzymes were incubated with 37 μm [Xyl-3H]XXXG plus 0, 1 or 10 mm XGO-NH2 for 20 h; products were electrophoresed. 2,4-Dinitrophenyl-lysine (DNP-Lys) was added as an internal marker.
(d–f) Neutral products (−1 to +6 cm in a–c) were eluted and paper-chromatographed in butan-1-ol/pyridine/water (4:3:4 by volume).
Short horizontal lines indicate external markers.

Cationic products (Figure 5b) were eluted from the electrophoretogram, then acid-hydrolysed. Paper chromatography revealed that [3H]xylose was the sole radioactive product. The data thus provide unambiguous evidence for a trans-α-xylosidase (TαX) reaction (Figure 4a).

The major radioactive cations were all at least as large as XXXG-NH2; smaller products (e.g. XXG-NH2) would have migrated faster than 2,4-dinitrophenyl-lysine owing to their higher charge : mass ratio (Miller et al., 2007; Fry, 2011b). In contrast, [Xyl-3H]XXXG itself was partially degraded in the absence of XGO-NH2 to [3H]XXG, [3H]XG and free [3H]xylose (Figure 5d), the latter indicating α-xylosidase activity.

Formation of cationic radioactive products was time-dependent, although the rate decreased during prolonged incubations (Figure 6). The detected transglycosylation rate was generally higher with 1 mm XGO-NH2 as acceptor than with 10 mm XGO-NH2.

Figure 6.

 Kinetics of the trans-α-xylosidase reaction.
Extracts of cauliflower, parsley, asparagus, chicory and snowdrop were incubated with 37 μm [Xyl-3H]XXXG plus 0, 1 or 10 mm XGO-NH2. After electrophoresis, 3H-labelled cations were assayed for radioactivity (±counting error).

Lack of interference from XET activity

All extracts contained XET activity (Table S1). If they also contained co-extracted xyloglucan, some of the [3H]XXXG in the assays would have undergone the XET reaction, generating [3H]xyloglucan, perhaps limiting the [3H]XGO-NH2 yield. Traces of 3H-polysaccharide were indeed formed during TαX assays (Table S1), but the yields were only 0.04–0.23% of the [3H]XXXG supplied; therefore, XET activity did not limit TαX reactions.

Remarkable stability of trans-α-xylosidase products

The DP8 and DP9 products generated from XXXG by TαX remained intact in the presence of crude plant extracts even after most XXXG had disappeared [e.g. cress (Figure 1) and the 50–60% saturated (NH4)2SO4 cut (Figure 2)]. Thus the transferred xylose residue(s) are remarkably resistant to hydrolysis by plant xylosidases.

Plant extracts contain trans-α-xylosidase and trans-β-galactosidase, but not trans-α-fucosidase activity

The dual-labelling method was extended to five 3H-oligosaccharides as potential donors, with or without 1 mm XGO-NH2 as acceptor (Figure 4 and Table 1). Negligible radioactivity (<0.1%) was incorporated into cationic products when XGO-NH2 was omitted (Table 1). After correction for these low control values, incorporation demonstrated TαX and trans-β-galactosidase (TβGa) activities but not trans-α-fucosidase activity (Table 1).

Table 1.   Transglycosidase activities with diverse donor substrates
Potential donor substrateIncubation time (h)Yield of transglycosylation products (% of supplied 3H) generated by extracts from
  1. *Extract was diluted compared with that used in Figure 6.

  2. Plant extracts were incubated with tritium-labelled potential donor substrates ± 1 mm XGO-NH2. Cationic transglycosylation products (excised from the zone shown in Figure S5a) were assayed for radioactivity. Any 3H incorporation in the absence of XGO-NH2 was subtracted from these data (which are presented ± the counting errors). Boiled-extract controls were completely inactive.

[Xyl-3H]XXXG (14 μm)165.52 ± 0.379.57 ± 0.499.17 ± 0.485.26 ± 0.366.73 ± 0.41
646.72 ± 0.3313.74 ± 0.4812.8 ± 0.4610.96 ± 0.4319.94 ± 0.57
[Xyl-3H]XXFG (58 μm)162.62 ± 0.134.14 ± 0.164.72 ± 0.171.42 ± 0.092.84 ± 0.13
643.42 ± 0.128.09 ± 0.188.54 ± 0.193.96 ± 0.135.80 ± 0.15
[Gal-3H]XXLG (155 μm)160.64 ± 0.060.87 ± 0.070.10 ± 0.021.09 ± 0.080.03 ± 0.01
640.41 ± 0.040.74 ± 0.060.19 ± 0.030.87 ± 0.060.25 ± 0.03
[Gal-3H]XLLG (77 μm)160.92 ± 0.081.49 ± 0.100.18 ± 0.032.97 ± 0.140.03 ± 0.01
640.45 ± 0.040.87 ± 0.060.14 ± 0.021.87 ± 0.090.01 ± 0.01
[Fuc-3H]XXFG (16 μm)160.03 ± 0.010.02 ± 0.010.09 ± 0.020.05 ± 0.010.02 ± 0.01
640.01 ± 0.010.04 ± 0.010.07 ± 0.010.00 ± 0.010.01 ± 0.01

TαX operated with either [Xyl-3H]XXXG or [Xyl-3H]XXFG as donor substrate: XXXG was two- to threefold preferred over XXFG (Table 1). The highest yield of radioactive cations was 20% of the supplied [3H]XXXG, catalysed by snowdrop extracts. With all species, TαX products increased between 16 and 64 h (Table 1), demonstrating a long-lived enzyme and relatively stable products, despite the activity, in most extracts, of α-xylosidase which released [3H]xylose from [3H]XXXG and [3H]XXFG (Figure 7a,b) and degraded [1-3H]XXXGol (Figure 7f). As expected, the presence of competing XGO-NH2 strongly decreased [3H]xylose release and [1-3H]XXXGol degradation (Figure 7). A slow-migrating radioactive spot (presumably [3H]Xyl·XGO-NH2) is visible (Figure 7a,b); however, neutral transglycosylation products are faint (e.g. at the position marked ‘Xyl·XXFG’), suggesting that XGO-NH2 is a less effective as a xylosyl donor than as an acceptor. This is helpful in our assays, minimizing non-radioactive xylosyl transfer. Products such as DP8–10 (Figures 1–3, and Figures S2 and S4) were undetectable in the ‘–’ lanes (Figure 7a) because the XXXG concentration in the latter was 100-fold lower (0.014 versus 1.4 mm).

Figure 7.

 Products formed by action of plant enzymes on 3H-oligosaccharides ± XGO-NH2.
Extracts were incubated for 64 h with (a) 14 μm [Xyl-3H]XXXG, (b) 58 μm [Xyl-3H]XXFG, (c) 155 μm [Gal-3H]XXLG, (d) 77 μm [Gal-3H]XLLG, (e) 16 μm [Fuc-3H]XXFG or (f) 49 μm [1-3H]XXXGol; each with (+) or without (−) 1 mm XGO-NH2. Enzymes were from (1) cauliflower, (2) parsley, (3) asparagus, (4) chicory and (5) snowdrop. Products (neutral plus cationic) were subjected to TLC, then fluorographed. Loadings were 100 Bq (a), 410 Bq (b–d), 350 Bq (e) or 180 Bq (f). Each ‘+’ lane contained products from 5 nmol XGO-NH2. After fluorography, plates were thymol-stained (pink), as shown for (a) and (b); all others appeared similar.

TβGa activity was detected with [Gal-3H]XXLG or [Gal-3H]XLLG (Figure 4b) as the donor. Rates were lower than for TαX, but the reliability of the data is strengthened by the close correlation between yields obtained with XXLG or XLLG as donor (Table 1). XLLG was roughly twice as effective as XXLG. Chicory had the highest TβGa activity, monocots had the lowest. Cationic trans-β-galactosylation products were unstable, with the yield usually decreasing between 16 and 64 h. One degradation product was [3H]galactose, indicating hydrolysis (Figure 7c,d). However, there were also unidentified products, including some that migrated faster than galactose. In some incubations, especially with asparagus and chicory, 3H recovery was incomplete (lanes c3–, d3– and d4–; Figure 7), suggesting [3H]water formation. This 3H loss was decreased by the presence of competing XGO-NH2. Release of [3H]water may be due to oxidation of [6-3H]galactose residues to galacturonic acid or arabinopyranose (found in xyloglucan building blocks M, N, P and Q; Peña et al., 2008).

Although XXFG was a TαX donor substrate, the same oligosaccharide failed to transfer its fucose residue to XGO-NH2 (Table 1). Absence of transfucosylation was not due solely to hydrolysis of XXFG by α-fucosidase; some [Fuc-3H]XXFG remained after 64 h incubation, especially when XGO-NH2 was present (Figure 7e). Therefore, under conditions that successfully revealed TαX and TβGa activities, no trans-α-fucosidase activity was observed.

The transferred xylose residue forms a Driselase-resistant trisaccharide and retains α-anomerism

Treatment of XGO-NH2 with Driselase (which lacks α-xylosidase activity; Fry, 2000) cleaved the galactosyl and most glucosyl linkages, yielding isoprimeverose, galactose and a cationic trimer (XG-NH2): XXLG-NH2 + 3H2O → galactose + 2 isoprimeverose + XG-NH2. In contrast, Driselase digestion of the radioactive TαX products (Table 1) eluted from electrophoretograms produced three neutral radioactive products: isoprimeverose, a trisaccharide and a tetrasaccharide, irrespective of whether the xylosyl donor was [Xyl-3H]XXXG (Figure 8a) or [Xyl-3H]XXFG (Figure 8b). Driselase digestion of [Xyl-3H]XXXG (Figure 8c) and [Xyl-3H]XXFG themselves produced [3H]isoprimeverose as the sole radioactive product, as expected. Thus, production of the 3H-trimer and 3H-tetramer (Figure 8a,b) indicated that asparagus transxylosidase generates new, Driselase-resistant structures that are not present in the substrates.

Figure 8.

 Analysis of cationic transglycosylation products.
Asparagus extract was incubated with [Xyl-3H]XXXG (a), [Xyl-3H]XXFG (b) or [Gal-3H]XLLG (d), plus 1 mm XGO-NH2, as in Figure 5(b). Cationic transglycosylation products (CTPs) were isolated by electrophoresis, then exhaustively digested with Driselase; products were separated by paper chromatography. Driselase products for the pure donor substrates [Xyl-3H]XXXG (c) and [Gal-3H]XLLG (e) are also shown. The material labelled ’[3H]DP3‘ in (a) and (b) was then exhaustively digested with α-xylosidase (g,j) or β-xylosidase (h,k) or left undigested (f,i), then re-analysed by paper chromatography. Chromatography solvents: (a–c) butan-1-ol/acetic acid/water (12:3:5 by volume) for 16 h followed by ethyl acetate/pyridine/water (10:4:3 by volume) for 40 h; (d–k) butan-1-ol/acetic acid/water (12:3:5 by volume) for 54 h alone. Short horizontal lines and dashed vertical arrows indicate approximate positions of external markers and OGi (internal marker Orange G). Solid arrows indicate the proposed identity of radioactive peaks. Cell-2, cellobiose; Gal, galactose; IP, isoprimeverose; Xyl2–4, β-(1→4)-xylobiose, xylotriose and xylotetraose. Data are presented ± the counting error.

The presence of [3H]isoprimeverose in the cationic products (Figure 8a,b) indicates that the asparagus enzyme(s) catalysed ‘futile’ xylosyl exchange, releasing a non-radioactive xylose from XGO-NH2 and replacing it with [3H]xylose at the same position.

To determine the anomerism of the newly incorporated [3H]xylose, we treated the Driselase-resistant trimer (DP3; Figure 8a,b) with commercial xylosidases. α-Xylosidase released all radioactivity from the trisaccharide as free [3H]xylose, whereas β-xylosidase released none (Figure 8f–k). Thus, the new xylosyl linkage created by plant extracts showed α-anomerism, and the enzyme is a retaining TαX. The 3H-trimer has been characterized as α-d-[3H]xylosyl-(1→4)-α-d-xylosyl-(1→6)-d-glucose, a new xyloglucan repeat unit abbreviated V (Franková and Fry, 2012), which is formed by [3H]xylosylation of a non-radioactive isoprimeverose unit.

Driselase digestion of cationic transgalactosylation products formed from [Gal-3H]XLLG (Figure 8d) and [Gal-3H]XXLG produced free [3H]galactose as the sole radioactive product, as did digestion of the donor substrates [Gal-3H]XLLG (Figure 8e) and [Gal-3H]XXLG. Thus, the [3H]galactose residue transferred by asparagus TβGa does not acquire unusual Driselase resistance.

Polysaccharides also act as acceptor substrates for trans-α-xylosidase

Enzyme from cauliflower leaves catalysed not only oligosaccharide-to-oligosaccharide xylosyl transfer (Figure 2) but also oligosaccharide-to-polysaccharide transfer (Table S2). Polysaccharides acting as xylosyl acceptor substrates (with [Xyl-3H]XXFG as donor) included xyloglucan, water-soluble cellulose acetate (WSCA), mixed-linkage (1→3),(1→4)-β-d-glucan (MLG), glucomannan and arabinoxylan. With xyloglucan as the acceptor, TαX assays are complicated by the XET activity also present. However, the XET and TαX reactions are distinguishable because XET generates 3H-polymers from either [Xyl-3H]XXFG or [1-3H]XXFGol, whereas TαX does so only from [Xyl-3H]XXFG. Moreover, TαX is much more strongly affected by polysaccharide concentration than is XET in the range studied: increasing xyloglucan from 0.1 to 0.5% (w/v) caused only a 1.28-fold increase in XET activity (16-h [1-3H]XXFGol data; Table S2), whereas with [Xyl-3H]XXFG (indicating TαX plus XET activity), increasing xyloglucan from 0.1 to 0.5% resulted in a 1.93-fold increase in 3H-polymeric products (16-h [Xyl-3H]XXFG data; Table S2). If the 854 cpm of 3H-product formed over 16 h with 0.1% xyloglucan + [Xyl-3H]XXFG were all due to XET, we would expect only 1.28 × 854 = 1093 cpm of product with 0.5% xyloglucan; the observed yield of 1650 cpm implies at least 1650−1093 = 557 cpm of TαX products (i.e., not only XET products). The 40-h data lead to similar conclusions. Such calculations are unnecessary for the other polysaccharides tested, because they are poor XET donor substrates. In conclusion, the polymeric acceptor substrates for TαX were, in order of effectiveness, xyloglucan ≈ WSCA > MLG > glucomannan ≈ arabinoxylan (Table S2).

To explore the attachment site of the transferred [3H]xylose on polysaccharide acceptor substrates, we digested the ethanol-insoluble TαX products using Driselase. The control in which pure [Xyl-3H]XXFG was used yielded [3H]isoprimeverose as the sole radioactive product (Figure 9a,b). Another control, in which the ‘acceptor’ xyloglucan was added after the enzyme had been denatured, produced no ethanol-insoluble 3H-polymers (Figure 9c). 3H incorporated into xyloglucan by non-denatured TαX was released (Figure 9d) as [3H]isoprimeverose plus [3H]V (DP3; compare Figure 8f,i). [3H]Isoprimeverose may arise from either XET- or TαX-generated polymers, but V is attributed solely to TαX. [Note: each XET reaction recruits three [3H]xylose residues (intact [Xyl-3H]XXFG) into the polymeric form, whereas the TαX reaction introduces only one; therefore the DP3 peak in Figure 9(d) can legitimately be tripled for the purposes of TαX/XET comparison.] Digesting the same [3H]xyloglucan with xyloglucan endoglucanase (XEG, which cleaves at unsubstituted G units) produced only oligosaccharides migrating with or faster than XXFG, well resolved from Glc8-based XGOs (Figure S6); thus the [3H]isoprimeverose residues (Figure 9d) did not originate from XEG-resistant structures such as XXXXXXXG (where the bold X is radioactive), which would have indicated α-xylosyl transfer onto a mid-chain G unit of xyloglucan.

Figure 9.

 Polysaccharides as trans-α-xylosidase acceptor substrates.
Reaction mixtures contained [Xyl-3H]XXFG (donor), the named polysaccharide (0.5% w/v) (acceptor) and cauliflower leaf protein. 3H-Polymers present after 16 h incubation were Driselase-digested, and their fragments resolved by paper chromatography in butan-1-ol/acetic acid/water (12:3:5 by volume). (a) Undigested [Xyl-3H]XXFG; (b) Driselase-digested [Xyl-3H]XXFG; (c) control with xyloglucan added to denatured cauliflower enzyme; (d–h) Driselase digestion of high-molecular-weight products formed by TαX from the named polysaccharide. In (i), DP3 from (h) was eluted, mild acid-hydrolysed, and re-chromatographed. Horizontal grey lines, non-radioactive markers; double grey lines, replicate markers. Abbreviations: Cell2, cellobiose; IP, isoprimeverose; MLG, mixed-linkage β-glucan; int OG, internal marker Orange G; TFA, trifluoroacetic acid; WSCA, water-soluble cellulose acetate.

3H-Polymers formed from WSCA, MLG and glucomannan also produced [3H]isoprimeverose on Driselase digestion; the WSCA also produced a trace of the 3H-trisaccharide (Figure 9e–g). As these polysaccharides are relatively poor substrates for XET (see [1-3H]XXFGol data; Table S2), the results in Figure 9 imply α-xylosyl transfer from XXFG to glucose residues of these polysaccharides. The glucomannan produced in addition a series of low-mobility radioactive peaks, possibly due to xylosylation of mannose residues to form structures such as α-Xyl-(1→6)-Man, which may not be releaseable as a disaccharide by Driselase.

Arabinoxylan was also a TαX acceptor substrate (Figure 9h). The major Driselase product in this case migrated near ‘DP3’ in Figure 9(d), suggesting a structure such as α-[3H]Xyl-(1→?)-β-Xyl-(1→4)-Xyl, which would be Driselase-resistant, like 5-O-feruloyl-α-Ara-(1→3)-β-Xyl-(1→4)-Xyl (Wende and Fry, 1997). This ‘DP3’ product (Figure 9h) was eluted and found to be resistant to mild acid under conditions optimized (Kerr and Fry, 2003) for selective hydrolysis of arabinofuranosyl linkages (Figure 9i). Thus TαX α-xylosylates the β-d-xylose rather than the α-l-arabinose residues of arabinoxylan.


The results of this study show that plants possess transglycosidase activities, transferring single α-d-xylose and β-d-galactose (but not α-l-fucose) residues between XGO molecules and from oligosaccharides to polysaccharides.

Transglycosidases are interesting because wall polysaccharide structures are determined not only by Golgi-localized nucleoside-diphosphate--sugar-dependent synthases, but also by reactions occurring in the wall after secretion. Such apoplastic reactions include those involving hydroxyl radicals (Müller et al., 2009) and those catalysed by oxidoreductases (Burr and Fry, 2009), hydrolases (de Alcântara et al., 1999; Sampedro et al., 2010; Günl and Pauly, 2011) and transglycosylases (Thompson and Fry, 2001).

Transglycosidase assays on neutral XGOs

When XXXG (or [1-3H]XXXGol) was used as both donor and acceptor substrate (Figures 1–3), either TαX or trans-β-glucosidase could have generated the patterns observed. For example, an octasaccharide could be formed from XXXG either in one step by TαX


or in two steps by α-xylosidase


then trans-β-glucosidase (Crombie et al., 1998)


However, the octasaccharide Xyl3·Glc5 (e.g. GXXXG) formed in the latter case would be short-lived in the presence of β-glucosidase-rich plant extracts.

Dual-labelling strategy for transglycosidase assays

To obtain proof of intermolecular xylosyl transfer, we devised a ‘dual-labelling’ strategy (Figure 4a). The production of radioactive cationic products confirms intermolecular transxylosylation. [Xyl-3H]XXXG served as the xylosyl donor substrate and XGO-NH2 as the acceptor. Inevitably, reactions involving the three other donor:acceptor combinations (3H:3H, cation:cation and cation:3H) occurred simultaneously, but produced only neutral or non-radioactive products. By disregarding such products, we were able to demonstrate the occurrence of ‘3H:cation’ reactions.

This dual-labelling strategy has wide applicability in transglycosidase studies. Alternatives to HVPE could be used, including cation-exchange methods. The assays reported here were for homo-transglycosidase reactions, as both donor and acceptor were XGOs. Nevertheless, the ability to selectively detect only chosen donor:acceptor combinations will be useful in future assays for hypothetical hetero-transglycosidases.

Xyloglucan structures formed by trans-α-xylosidase

The TαX activity reported here created a novel xyloglucan building block (the Driselase-resistant trisaccharide α-d-xylosyl-(1→4)-α-d-xylosyl-(1→6)-d-glucose), recently given the code letter V (Franková and Fry, 2012).

TαX activity may be due to an α-xylosidase similar to that described by Sampedro et al. (2010), who presented evidence compatible with transglycosylation. This is a family GH31 enzyme (, which, as a retaining hydrolase, catalyses some transglycosylation in addition to hydrolysis (Kang et al., 2008). As a hydrolase, it attacks the non-reducing ends of XGOs and possibly also high-molecular-weight xyloglucans (Koyama et al., 1983; Augur et al., 1993; Crombie et al., 2002). It probably selects the same end residue for its donor substrate during transxylosylation. It remains unknown whether the acceptor substrate site is similarly confined to the non-reducing extremity or whether TαX can graft α-xylose residues onto alternative sites, e.g. the second xylose residue of XXXG (generating XVXG).

The acceptor substrate may also be a β-(1→4)-linked hemicellulose, e.g. β-glucans, β-mannan or β-xylan (Figure 9). With glucans and mannans, it is possible that TαX usually selects only the –CH2OH group (6-position) at the non-reducing extremity of the backbone as its acceptor site, producing [3H]isoprimeverose or α-[3H]xylosyl-(1→6)-mannose (albeit, in the case of xyloglucan, only after removal of the existing non-reducing end xylose residue). In agreement with this, we showed that the –CH2OH group of a mid-chain G unit in xyloglucan cannot serve as acceptor (Figure S6). However, in the case of β-xylan, TαX transferred α-[3H]xylose onto a β-xylopyranose residue, which has no –CH2OH group, showing that TαX can create products that are quite dissimilar from isoprimeverose. It is therefore also possible that, with high-molecular-weight xyloglucan as acceptor, TαX can transfer new xylose residues onto isoprimeverose units to form V (α-Xyl-α-Xyl-β-Glc) at multiple positions along the polysaccharide’s backbone.

Xyloglucan structures formed by trans-β-galactosidase activity

Dual-labelling also demonstrated TβGa activity (Figure 4b). All known plant β-galactosidases, e.g. the 17 such enzymes in Arabidopsis, belong to CAZy families GH35 (Gantulga et al., 2009) or GH2, and are thus predicted to be ‘retaining’ enzymes.

XXLG and XLLG were both effective donor substrates for T{β}Ga. This contrasts with the major plant β-galactosidases, which hydrolyse XLLG (and XLXG) but not XXLG (de Alcântara et al., 1999; Franková and Fry, 2011). Our present findings thus indicate the existence of a specialized TβGa that differs from the major plant β-galactosidases in substrate preference.

It will be interesting to determine whether the transferred galactose residue is grafted at a ‘conventional’ position (e.g. converting XXLG-NH2 to [3H]XLLG-NH2) or elsewhere (producing novel structures).

Transglycosylation/hydrolysis competition

Retaining glycosidases operate by double displacement, involving transient formation of a substrate–enzyme covalent complex (Sinnott, 1990; Davies and Henrissat 1995Hrmová and Fincher, 2007), which then transfers the glycosyl group onto an acceptor substrate: either H2O (hydrolysis) or an alcohol (transglycosylation) (Figure 10). The transglycosylation:hydrolysis ratio is determined by the alcohol concentration and the enzyme’s properties. ‘Mechanistic’ transglycosylation, which is observed only at high alcohol concentrations (e.g. >100 mm sugar), is probably of little biological significance. However, if the enzyme shows appreciable transglycosylation even at millimolar concentrations of alcohol, this may be relevant in vivo.

Figure 10.

 Hydrolysis/transglycosylation competition during trans-α-xylosidase action.
Enzyme (E) severs [3H]xylose from the donor substrate (top), forming a [3H]xylosyl–enzyme bond. This then reacts with H2O (hydrolysis, dashed arrows) or XGO-NH2 (transglycosylation, solid arrows). Symbols are defined in Figure 4. Measured hydrolysis:transglycosylation ratios are indicated for four enzyme sources supplied with 1 mm (left) or 10 mm (right) XGO-NH2.

Variation in the transglycosylation:hydrolysis ratio between cauliflower glycosidases was illustrated for seven activities (α- and β-glucosidase, α- and β-galactosidase, α- and β-mannosidase and β-xylosidase) using NP glycosides (all probable retaining enzymes; (Figure S1). Only two exhibited detectable transglycosylation. Thus detectable transglycosylation is not an inevitable side reaction of enzymic hydrolysis, but is a feature of certain glycosidases.

In dual-labelling TαX and TβGa assays, transglycosylation occurred with 1 mm acceptor substrate, and may thus be physiologically significant. TαX-catalysed transglycosylation was even detectable with 16 μm XXXGol (Figure 3). Hydrolysis occurred simultaneously, but transglycosylation often predominated (Figures 1, 2, 5 and 10). Under standardized conditions, the transxylosylation:hydrolysis ratio varied between species (Figure 10), indicating that TαX activity is not simply a predictable side reaction of α-xylosidase; rather, certain isoenzymes have evolved to favour transglycosylation.

During TαX assays, both [Xyl-3H]XXXG and XGO-NH2 presumably underwent both transglycosylation and hydrolysis, leading to a range of reactions. As radioactive cation production (indicating [Xyl-3H]XXXG:XGO-NH2 transglycosylation) was higher with 1 mm XGO-NH2 than with 10 mm XGO-NH2 (Figure 6), the latter apparently competed with [Xyl-3H]XXXG as the xylosyl donor substrate, leading to ’XGO-NH2:XGO-NH2’ transglycosylation, which is undetectable because the products are non-radioactive. However, with all extracts except snowdrop, this competition between radioactive and non-radioactive xylosyl donors was slight. For example, in the 20-h asparagus sample, 10 mm XGO-NH2 inhibited XXXG-to-cation [3H]xylosyl transfer by only 8% compared with 1 mm XGO-NH2 (Figure 6c).

In contrast, 10 mm XGO-NH2 caused 86% inhibition of [3H]XXXG hydrolysis (free [3H]xylose production) compared with 1 mm XGO-NH2 (Figure 5e,f). Thus, 10 mm XGO-NH2 had little competitive effect on [3H]xylose severance from [Xyl-3H]XXXG, but increased the transglycosylation:hydrolysis ratio by providing a higher concentration of acceptor substrate.

Transxylosylation/hydrolysis competition can be quantified by the relative yield of 3H-labelled cations (transglycosylation) versus free [3H]xylose (hydrolysis). On this basis, transglycosylation accounted for 61–95% of total xylosyl severance (Figure 10). These values under-estimate transglycosylation because (i) they do not include reactions in which a [3H]xylose residue is transferred from one neutral oligosaccharide to another (Figure 1), and (ii) some initially formed 3H-labelled cations are lost by subsequent hydrolysis during lengthy incubations.

Absence of trans-α-fucosidase activity

In similar experiments, we found no XXFG-acting trans-α-fucosidase, although α-fucosidase activity is widespread (Baydoun and Fry, 1989; Augur et al., 1993; Franková and Fry, 2011) (Figure 7f). Plant α-1,2-l-fucosidases belong to an inverting family, GH95 (, and are therefore unlikely to catalyse transglycosylation (Koshland, 1953; Scigelova et al., 1999; Tramice et al., 2007; Rudsander et al., 2008).

Biological significance of transglycosidase action on oligosaccharins

Enzymes acting on XGOs are interesting because of the reported signalling (oligosaccharin) roles of these apoplastic solutes. Apoplastic transglycosidases may generate novel oligosaccharins. The presence of additional α-xylose residues would also enhance the longevity of oligosaccharins. Conversely, transglycosidases could inactivate oligosaccharins by adding new residues as ‘blocking groups’.

Transglycosidase action on polysaccharides

Our dual-labelling TαX and TβGa assays used only oligosaccharides as substrates. Nevertheless, certain polysaccharides also serve as TαX acceptor substrates (Figure 9 and Table S2). In this capacity, it is possible that TαX might add new xylose residues only at the non-reducing extremity, at which α-xylosidase probably acts on xyloglucan (O’Neill et al., 1989; Crombie et al., 2002). Alternatively, it may add xylose residues in mid-chain.

Transglycosidase action confined to a polysaccharide’s non-reducing extremity may initially seem rather insignificant. However, this specific molecular site has paramount importance as the only site in the whole xyloglucan chain known to serve as an acceptor substrate for XET. This role is important for wall assembly (‘integrational transglycosylation’) and wall modification (‘re-structuring transglycosylation’), both of which occur in vivo (Thompson and Fry, 2001) and are implicated in cell expansion.

TαX action removes an α-xylose residue and re-attaches it elsewhere. Removing the non-reducing end α-xylose residue, whether by TαX or α-xylosidase action, may temporarily incapacitate a xyloglucan chain as an XET acceptor substrate (Franková and Fry, 2011), as suggested by the observation that GGGG and related xylose-deficient oligosaccharides are poor XET acceptor substrates (Nishitani and Tominaga, 1992; Fry et al., 1992; Lorences and Fry, 1993; Fanutti et al., 1996; Saura-Valls et al., 2008). Conversely, adding a new α-xylose at the polysaccharide’s non-reducing end (converting, say, XXXG… to VXXG…) may more permanently modify susceptibility to XET. It will be interesting to discover whether V is formed preferentially at the non-reducing end, and whether such a terminus serves as a favoured, or disfavoured, XET acceptor substrate. As TαX-generated oligosaccharides (e.g. DP8; Figures 1, 3 and S2) are remarkably resistant to hydrolysis by crude plant extracts (Figures 1 and 2), we suggest that a V unit in a long xyloglucan chain is a durable structural feature.

Concluding remarks

The existence and catalytic properties of the reported transglycosidase activities would not readily have been predicted from gene sequence data, but were revealed by enzymic assays. Definitive evidence for XGO→XGO intermolecular grafting of α-xylose and β-galactose residues was provided by dual-labelling experiments. One of the labels (3H) was radioactive; the other (–NH3+) provided a reliable means to physically separate molecules possessing it from all others. This strategy has general applicability for detecting both homo- and hetero-transglycosidase activities. The biological roles of the new xyloglucan domains created by TαX and TβGa now invite detailed exploration.

Experimental procedures

Polysaccharides and oligosaccharides

XXXG, XXXGol, konjac (Amorphophallus konjac) glucomannan and wheat (Triticum aestivum) arabinoxylan were from Megazyme ( A mixture of non-fucosylated XGOs was kindly provided by Dr K. Yamatoya (Dainippon Pharmaceutical Co., Osaka, Japan). Barley (Hordeum vulgare) MLG was from Sigma ( WSCA was prepared as described by Fry et al. (2008).

Analytical hydrolysis

Oligosaccharides and polysaccharides were enzymically hydrolysed in 1% Driselase (Sigma Chemical Co.; partially purified as described by Fry, 2000) in pyridine/acetic acid/water (1:1:98) containing 0.5% chlorobutanol at 37°C for 3 days; reactions were stopped with formic acid (0.2 volumes). Acid hydrolysis was performed for 1 h in 2 m trifluoroacetic acid at 120°C (complete hydrolysis) or 0.1 m trifluoroacetic acid at 85°C (partial hydrolysis).

Thermostable α-xylosidase was obtained from CPC Biotech ( The oligosaccharide sample (50 μl) was incubated with 100 μl 1%α-xylosidase in pyridine/acetic acid/water (11:3:2000 by volume, pH 5.5, containing 0.5% chlorobutanol) at 65°C for 24 h.

Bacillus pumilusβ-xylosidase (catalogue reference E-BXSEBP) was obtained from Megazyme. The oligosaccharide sample (50 μl) was mixed with 100 μl of enzyme solution (75 U/ml in 1% 2,4,6-collidine/0.24% acetic acid, pH 7.4, containing 1 mg ml−1 bovine serum albumin and 0.5% chlorobutanol), and incubated at 20°C for 24 h.

Chromatography and electrophoresis

Reaction products were analysed on Merck ( silica-gel ‘60’ TLC plates in butan-1-ol/acetic acid/water (2:1:1 by volume). Non-radioactive sugars were stained with thymol/H2SO4 (Franková and Fry, 2011).

Paper chromatography was usually performed on Whatman ( 3CHR paper in butan-1-ol/acetic acid/water (12:3:5 by volume), ethyl acetate/pyridine/water (10:4:3 by volume) or butan-1-ol/pyridine/water (4:3:4 by volume). Marker sugars were stained with AgNO3 or aniline hydrogen phthalate (Fry, 2000).

Cationic products were resolved by HVPE at pH 2.0 on Whatman 3CHR (Fry, 2011b); 4.5 kV was applied for 30–45 min. The dried paper was stained with ninhydrin, aniline hydrogen phthalate or AgNO3 (Fry, 2000), or cut into strips for 3H assay.

Detection of radioactivity

Strips of chromatography paper were assayed for 3H by scintillation counting in OptiScint HiSafe (PerkinElmer; (efficiency for mono- and oligosaccharides approximately 7%, efficiency for polysaccharides approximately 30%). 3H-Labelled samples were located on TLC plates by fluorography (Fry, 2000), and quantitatively profiled on a LabLogic ( AR2000 radioisotope imaging scanner with LabLogic’s Laura data analysis software.

Preparation of radiolabelled oligosaccharides

[Xyl-3H]XXXG, [Xyl-3H]XXFG and [Fuc-3H]XXFG were isolated chromatographically from walls of cultured spinach (Spinacia oleracea) cells that had been fed l-[1-3H]arabinose or l-[1-3H]fucose (Fry, 2000), then digested with XEG. [Gal-3H]XXLG and [Gal-3H]XLLG were prepared by XEG digestion of tamarind (Tamarindus indica) xyloglucan that had been treated with galactose oxidase (Parikka et al., 2010) then NaB3H4-reduced. [1-3H]XXXGol and [1-3H]XXFGol were prepared by NaB3H4 reduction of XXXG and XXFG. The specific radioactivities, estimated after TLC, were: [Xyl-3H]XXXG and [Xyl-3H]XXFG, 1.6 MBq μmol−1; [Fuc-3H]XXFG, 6.7 MBq μmol−1; [Gal-3H]XXLG, 0.6 MBq μmol−1; [Gal-3H]XLLG, 1.2 MBq μmol−1; [1-3H]XXXGol, 1.3 MBq μmol−1; [1-3H]XXFGol, 34 MBq μmol−1.

On acid hydrolysis, each preparation gave mainly the expected radioactive product as revealed by paper chromatography: [Xyl-3H]XXXG, 94.9% xylose; [Xyl-3H]XXFG, 97.7% xylose; [Fuc-3H]XXFG, 97.8% fucose; [Gal-3H]XXLG, 92.6% galactose; [Gal-3H]XLLG, 97.3% galactose; [1-3H]XXXGol, 71.5% glucitol plus 14.2% mannitol. This small proportion of [1-3H]XXX-mannitol is not expected to influence the results.

Preparation of XGO-NH2

Tamarind XGOs were reductively aminated and freed of reagents (Miller et al., 2007). The major products are oligosaccharidyl-1-amino-1-deoxyalditols (XLLG-NH2, XXLG-NH2 and XXXG-NH2, or in general XGO-NH2, where G-NH2 is 1-amino-1-deoxyglucitol). Four independent XGO-NH2 preparations were analysed by HVPE. Staining for amino groups, reducing sugars and total sugars (Figure S5b–d, respectively) demonstrated the presence of non-reducing cationic oligosaccharides and the absence of neutral and reducing sugars. Preparation 3 (Figure S5) was used as acceptor substrate in transglycosylation assays.

Extraction and fractionation of enzymes

Seedlings of onion (Allium cepa), cress (Lepidium sativum), lettuce (Lactuca sativa), lucerne (Medicago sativa), maize (Zea mays) and broad bean (Vicia faba) were grown in the dark. Asparagus officinalis, parsley (Petroselinum crispum), chicory (Cichorium endivia), cauliflower (Brassica oleracea var. botrytis) and bean sprouts (Vigna radiata) were purchased from a supermarket. Snowdrop (Galanthus nivalis), horsetail (Equisetum arvense), liverwort (Marchantia polymorpha) and Selaginella apoda were grown at the King’s Buildings (Edinburgh, UK).

Enzymes were extracted as described previously (Franková and Fry, 2011). Plant material was homogenized at 4°C in 0.2 m succinate (Na+), pH 5.5, 10 mm CaCl2, containing 2% polyvinylpolypyrrolidone (= extractant A), or 1 m NaCl in extractant A (= extractant B). The homogenate was stirred for 3 h at 4°C, filtered and centrifuged at 11500 g and 4°C for 40 min. In some experiments, the supernatant was dialysed against water, then against 0.04 m succinate (Na+) buffer. Extracts were stored at −80°C.

For (NH4)2SO4 fractionation, proteins were sequentially precipitated at between 20 and 70% saturation, centrifuged at 10 500 g for 40 min, and then redissolved in buffer B. After dialysis (as above), enzyme solutions were centrifuged at 2000 g for 1 min, and the supernatants were used in assays.

TLC-based assay of glycosidases and transglycosidases acting on XGOs

Routinely, assay mixtures contained 30 μg XXXG or XGO, added in 5 μl water to 15 μl of buffered enzyme extract, giving approximately 1.4 mm substrate and 70–950 μg ml−1 protein. After incubation at 22°C for 24 h, 2–10-μl aliquots were analysed by TLC. In a study of the effect of XXXGol concentration, 20 kBq [1-3H]XXXGol plus a variable mass of non-radioactive XXXGol were dried into a vial under vacuum, and then 20 μl of enzyme preparation was added; 5 μl aliquots were removed for TLC at intervals (1–12 h).

Transglycosidase assays by the dual-labelling method

Reaction mixtures contained (final concentrations): 14 or 37 μm [Xyl-3H]XXXG, 58 μm [Xyl-3H]XXFG, 155 μm [Gal-3H]XXLG, 77 μm [Gal-3H]XLLG, 16 μm [Fuc-3H]XXFG or 49 μm [1-3H]XXXGol (control) as the putative glycosyl donor, 1 or 10 mm XGO-NH2 (omitted in controls) as the acceptor, 0.5% chlorobutanol and 40% v/v enzyme extract. The enzymes tested were prepared from cauliflower leaf mid-veins (40–50% saturated (NH4)2SO4 pellet), parsley shoots, asparagus sprouts, chicory leaves and snowdrop leaves in extractant B (see ‘Extraction and fractionation of enzymes’) and dialysed. After 1.5, 5 and 20 h, 12.5-μl aliquots were analysed by HVPE. Routinely, the XGO-NH2 zone was assayed for 3H; in some experiments, additional zones were assayed too.

Trans-α-xylosidase assays with polysaccharides as acceptor substrate

Reaction mixtures contained (final concentrations): 75 Bq μl−1 (47 μm) [Xyl-3H]XXFG as donor substrate, or 75 Bq μl−1 [1-3H]XXFGol as a ‘mock donor’, 0.1 or 0.5% polysaccharide (tamarind xyloglucan, barley MLG, WSCA, glucomannan or arabinoxylan) as acceptor substrate, 0.1% chlorobutanol and 0.62 mg/ml protein from cauliflower leaf lamina (extracted with extractant B, precipitated in the 50–60% saturated (NH4)2SO4 cut, and dialysed). After 16 and 40 h, 20 μl of the reaction mixture was added to 20 μl 20 m ammonia and incubated for 16 h (stopping transglycosylation and de-acetylating WSCA), then 50 μl acetic acid was added followed by 14 ml 75% ethanol. Precipitated polysaccharide was washed in 80% ethanol until the supernatants were non-radioactive, dried, and digested with 40 μl Driselase. A 10-μl aliquot of the digest was assayed for total radioactivity, and 30 μl was analysed by paper chromatography in butan-1-ol/acetic acid/water (12:3:5 by volume) for 64 h. In the case of the arabinoxylan products, the major 3H-oligosaccharide was then eluted from the paper with water, dried, subjected to partial acid hydrolysis, and re-chromatographed.


We thank the UK Biotechnology and Biological Sciences Research Council for financially supporting this work.