Restructuring of wall-bound xyloglucan by transglycosylation in living plant cells


For correspondence (fax +44 131 650 5392; e-mail


Xyloglucan endotransglycosylases (XETs) cleave and then re-join xyloglucan chains and may thus contribute to both wall-assembly and wall-loosening. The present experiments demonstrate the simultaneous occurrence in vivo of two types of interpolymeric transglycosylation: ‘integrational’ (in which a newly secreted xyloglucan reacts with a previously wall-bound one) and ‘restructuring’ (in which one previously wall-bound xyloglucan reacts with another). Xyloglucans synthesised by cultured rose (Rosa sp.) cells in ‘heavy’ or ‘light’ media (with [13C,2H]glucose or [12C,1H]glucose, respectively) had buoyant densities of 1.643 and 1.585 g ml−1, respectively, estimated by isopycnic centrifugation in caesium trifluoroacetate. To detect transglycosylation, we shifted heavy rose cells into light medium, then supplied a 2-h pulse of l-[1−3H]arabinose. Light [3H]xyloglucans were thus secreted into heavy, non-radioactive walls and chased by light, non-radioactive xyloglucans. At 2 h after the start of radiolabelling, the (neutral) [3H]xyloglucans were on average 29% heavy, indicating molecular grafting during integrational transglycosylation. The [3H]xyloglucans then gradually increased in density until, by 11 h, they were 38% heavy. This density increase suggests that restructuring transglycosylation reactions occurred between the now wall-bound [3H]xyloglucan and other (mainly older, i.e. heavy) wall-bound non-radioactive xyloglucans. Brefeldin A (BFA), which blocked xyloglucan secretion, did not prevent the increase in density of wall-bound [3H]xyloglucan (2−11 h). This confirms that restructuring transglycosylation occurred between pairs of previously wall-bound xyloglucans. After 7 d in BFA, the 3H was in hybrid xyloglucans in which on average 55% of the molecule was heavy. Exogenous xyloglucan oligosaccharides (competing acceptor substrates for XETs) did not affect integrational transglycosylation whereas they inhibited restructuring transglycosylation. Possible reasons for this difference are discussed. This is the first experimental evidence for restructuring transglycosylation in vivo. We argue that both integrational and restructuring transglycosylation can contribute to both wall-assembly and -loosening.


The susceptibility of the primary cell wall to turgor-driven expansion is the principal factor controlling plant cell growth. The main load-bearing component of the primary wall in dicots is thought to be a network of cellulose microfibrils, tethered by chains of xyloglucan (a hemicellulose), which are attached to the microfibrils by hydrogen bonds (Fenwick et al., 1999; Fry, 1989a; Hayashi et al., 1987; McCann et al., 1990; Nishitani, 1998; Pauly et al., 1999; Whitney et al., 1999). Any agent that attacks this network is likely to promote the expansion of turgid plant cells, at least transiently. Such agents could include β-glucanases (Matsumoto et al., 1997; Nicol et al., 1998); xyloglucan endotransglycosylases (XETs) (Campbell and Braam, 1999; Fanutti et al., 1993; Fry et al., 1992; Nishitani and Tominaga, 1992; Purugganan et al., 1997; Smith and Fry, 1991; Sulováet al., 1998); expansins (Cosgrove, 1999; McQueen-Mason and Cosgrove, 1994); and hydroxyl radicals (Fry, 1998). The action of each of these agents is readily demonstrated on model substrates in vitro, but is difficult to demonstrate conclusively on wall-bound substrates in vivo. This is especially true of XET, which cleaves and then re-joins xyloglucan chains, causing no net chemical change.

The transglycosylation reaction catalysed by XET can be represented as

⊗OOOOOOOOOOO⊙+⊗○○○○○○⊙ →donor substrateacceptor(to be cleaved)


‘hybrid’ productleaving group

where each circle represents one of the Glc4-based repeat units of xyloglucan (Fry, 1989b; Hayashi, 1989); black and grey circles may be chemically identical; dotted circles are the original reducing termini; crossed circles are the original non-reducing termini. XETs appear to cleave the donor substrate at unsubstituted glucose residues chosen essentially randomly from along the length of the chain (Steele et al., 2001). Two main biological consequences of interpolymeric transglycosylation can be envisaged: these are referred to here as restructuring and integration.

Concerning restructuring transglycosylation, it has been proposed (Fry et al., 1990; Fry et al., 1992; Nishitani, 1998) that XETs can rearrange existing cell-wall material by catalysing transglycosylation between pairs of xyloglucan molecules, both partners having been previously hydrogen bonded to microfibrils in the cell wall. Transglycosylation of the restructuring type could potentially permit controlled cell expansion without irreversible weakening of the cell wall. For example, if the cleaved (donor) chain had been acting as a tether between two neighbouring microfibrils (Figure 1), its cleavage would locally disconnect these two microfibrils and permit incremental cell expansion. The subsequent re-formation of a glycosidic bond between the cleaved donor substrate and the non-reducing end of a different xyloglucan chain (the acceptor substrate), which had perhaps been brought within reach by the movement of molecules during incremental wall expansion, would restore much of the original strength of the cell wall (Figure 1). Reversible wall loosening by this type of transglycosylation, as opposed to a more permanent loosening caused by endo-β-glucanase action, would be comparable to the (penicillin-sensitive) transpeptidation reaction which is involved in the expansion of bacterial cell walls.

Figure 1.

Interpolymeric transglycosylation of the ‘restructuring’ type.

(a) Initially, two microfibrils (stippled) are tethered by a xyloglucan chain (solid line) which is hydrogen-bonded to them both; an additional xyloglucan chain (dotted line) is hydrogen-bonded to one of the microfibrils and has a non-reducing ‘loose end’. Both xyloglucan chains are shown with the non-reducing end to the left. XET (star) attacks the tethering chain, which is thereby cleaved, and a polysaccharide–enzyme covalent complex (star attached to line) is formed (Sulováet al., 1998).

(b) Turgor pressure causes the two microfibrils to move apart slightly so that the newly formed non-reducing end (of the leaving group) is no longer within reach of the polysaccharide–enzyme complex.

(c) Transglycosylation is later completed using the second xyloglucan chain as acceptor substrate: a new tether is thus produced in the form of a ‘hybrid’ molecule, and the XET is freed to act again. If the dotted xyloglucan chain (dotted line) was radioactive and the solid xyloglucan chain was heavy (or vice versa), the products would include a radioactive chain of intermediate density. The diagram shows only two of the many xyloglucan chains associated with these microfibrils. Adapted from Fry et al. (1990).

In integrational transglycosylation, a newly secreted xyloglucan chain is proposed to react with a xyloglucan chain that is already hydrogen bonded to one or more microfibrils in the cell wall. The incoming molecule (or part thereof) is thereby grafted to an existing wall-bound xyloglucan (or part thereof), and thus integrated into the wall architecture (Figure 2). We do not specify whether the incoming molecule is the acceptor substrate (and thus possibly integrated in its entirety; Figure 2b,c), or the donor substrate (such that part of it, the leaving group, fails to become integrated; Figure 2e). Newly secreted xyloglucan molecules can continue to become firmly integrated into the wall for up to a day after cellulose synthesis has been blocked by 2,6-dichlorobenzonitrile (Edelmann and Fry, 1992a); this result suggests that continued integration, at least over this time interval, is not dependent on direct hydrogen bonding of the incoming xyloglucan molecule to a recently produced length of microfibril. Integrational transglycosylation could account for these observations (Edelmann and Fry, 1992a). Over periods longer than about a day, continued integration of new xyloglucan fails (Shedletzky et al., 1990) as the existing cellulose–xyloglucan network becomes thinned by continuing growth.

Figure 2.

Five possible modes of integrational transglycosylation involving a newly secreted xyloglucan chain (dotted line) and a previously wall-bound chain (solid line).

In our experiments, the newly secreted xyloglucan chain (dotted line) was radioactive and the previously wall-bound chain (solid line) was heavy, so that the products would include radioactive chains of intermediate density. To simplify the diagram, we arbitrarily show the newly secreted chain as the shorter of the two. Both chains are shown with the non-reducing end to the left. In the initial condition (shown within the box), five distinct segments of the previously wall-bound chain are distinguished (Smith and Fry, 1991): (a) a non-reducing loose end; (b, b′) segments that are hydrogen-bonded to microfibrils (mf); (c) a tether between the two microfibrils; and (d) a reducing terminal loose end. Arrows a–d show the result expected when the previously wall-bound chain acts as the donor substrate and is cleaved by XET at sites a–d, respectively. Arrow e shows the result if the newly secreted chain acts as the donor substrate and is cleaved by XET at site e. Only in cases b, b′ and c is it possible for all the xyloglucan material under consideration to end up bonded to microfibrils. Note that any chains that are not hydrogen-bonded to any microfibrils will tend to be lost into the medium, and thus are not analysed in our experiments. The diagram shows only two of the many xyloglucan chains associated with these two microfibrils.

Polymer-to-polymer transglycosylation (whether of the integrational or the restructuring type) between endogenous xyloglucan molecules is difficult to demonstrate in vivo because the reaction products are chemically indistinguishable from the substrates. Dual labelling with the radioisotopes 3H and 14C (or with 3H and a fluorescent label) is not helpful because there is no convenient way of telling whether the two labels are present in a single xyloglucan molecule, or in two identical but separate molecules. Labelling experiments therefore critically depend on creating two physically separable populations of xyloglucan in vivo, and then monitoring the transfer of a label from one of these populations to the other.

We recently achieved this goal in cultured rose cells by dual labelling with a heavy, stable isotope (13C) and with a radioisotope (3H) (Thompson et al., 1997). When substrate concentrations of [13C]glucose were fed to the cells, they synthesized ‘heavy’ xyloglucan, which was partially resolvable from normal xyloglucan by isopycnic centrifugation in gradients of caesium trifluoroacetate (CsTFA). A trace of l-[3H]arabinose (a convenient precursor of the xylose residues of cell-wall polysaccharides) was then fed to the same cells to introduce a pulse of radioactive xyloglucan. By this approach we demonstrated the covalent grafting of portions of newly synthesized xyloglucans (recognized by their radioactivity) to portions of previously wall-bound xyloglucans (recognized by their density). This was attributed to integrational transglycosylation. The results suggested a role for XETs in the integration of new polysaccharide chains into the architecture of the primary cell wall (Thompson et al., 1997). In the dual labelling system described (Thompson et al., 1997), both the donor and the acceptor substrate are endogenous, natural polysaccharides; therefore the ‘grafting’ process detected can confidently be stated to occur in vivo.

However, there is currently no experimental evidence for restructuring transglycosylation of xyloglucan in vivo. The extensive integrational transglycosylation that occurred in the experiments reported (Thompson et al., 1997) would have masked any concurrent transglycosylation of the restructuring type. In the present work we used a novel dual-labelling protocol, with stable and radioactive isotopes, to provide the first evidence for restructuring transglycosylation in the walls of living plant cells.


Buoyant density of ‘heavy’ and ‘light’ xyloglucans

We first determined the buoyant density of fully heavy and fully light xyloglucans from rose cell cultures. Cultures G and H (Table 1), which had received only heavy substrate ([13C,2H]glucose) and only light substrate ([12C,1H]glucose), respectively, produced radioactive xyloglucans with mean buoyant densities of 1.6432 ± 0.0010 and 1.5846 ± 0.0006 g ml−1, respectively. In both cultures G and H, the buoyant density of the [3H]xyloglucans remained constant as the radiolabelled molecules ‘aged’ in the cell wall (0–7 days after radiolabelling).

Table 1.  Summary of treatments received by cell cultures
  • a Type of glucose fed: light circles = [12C, 1H]glucose (‘light’); dark circles = [13C, 2H]glucose (‘heavy’).

  • b

    Xyloglucan was extracted from each sample and analysed by isopycnic centrifugation in CsTFA.

−23 daysFresh medium to 50 mm glucoseaOOOOOOOO
−9 daysFresh medium to 25 mm glucoseaOOOOOOOO
−0.5 hFresh medium to 100 mm glucoseaOOOOOOOO
−0.1 hXGOs to 1 mm  +     
0 h[3H]Arabinose (to 1.9 µm)++++++++
+2 hSample 1.5 ml of cultureb++++++++
+2 hCold chase (to 5 mm arabinose)++++++ +
+2.5 hXGOs to 1 mm +  +   
+2.5 hBrefeldin to 48 µm   ++   
+2.5 hDMSO to 0.28%, v/v   +++  
+4 h to 7 daysSample 1.5 ml of cultureb++++++++

Evidence for integrational transglycosylation

Cultures A to F were pre-incubated in heavy medium for two passages, covering a total of 23 days (Table 1). To introduce a population of xyloglucan molecules with a different density so that we could monitor transglycosylation, we shifted cultures A to F from a medium containing heavy glucose as the sole carbon source to one containing light glucose as the sole carbon source. After a further 0.5 h, we supplied a pulse of [3H]arabinose so that a subpopulation of the new, light xyloglucan molecules would be radioactively labelled. The dose of [3H]arabinose supplied was sufficiently small (1.9 µm), and its specific radioactivity was sufficiently low (one 3H atom per approximately seven arabinose molecules), that it would not itself alter the buoyant density of the light xyloglucan being synthesized. The 0.5 h delay between the density shift and the [3H]arabinose feeding is known to be sufficient to replace essentially all the existing heavy intermediary metabolites (Glc-1-P; the Glc residue of UDP-Glc, etc.) with light ones (Thompson et al., 1997). Therefore the 3H-labelled xyloglucan chains would be fully light prior to their secretion from the endo-membrane system.

Cultures A, B, D, E and F received no additional treatments up to 2 h after the start of [3H]arabinose feeding (Table 1), by which time essentially all the [3H]arabinose had been consumed and the large majority of the [3H]xyloglucan would have been secreted into the apoplast (Edelmann and Fry, 1992b). At this time point the first samples were taken: the total neutral [3H]xyloglucan extractable from the cells at +2 h was found to have a mean buoyant density of 1.6016 ± 0.0004 g ml−1 (mean ± SE of cultures A, B, D, E and F; Figure 3). This value indicates that, within 2 h of synthesis, the initially light [3H]xyloglucans had become covalently grafted to older (heavy, non-radioactive) chains to form hybrid molecules in which, on average, 29% of the chain was heavy [calculated as 100 × (1.6016–1.5846) / (1.6432–1.5846)]. This grafting is attributed to integrational transglycosylation (Thompson et al., 1997), in which portions of newly synthesized, light [3H]xyloglucans (shortly after their secretion into the wall) become attached to portions of previously wall-bound, non-radioactive, heavy xyloglucans (Figure 2).

Figure 3.

Changes in buoyant density of pulse-labelled [3H]xyloglucan in cultures A–F (Table 1).

F or comparison, data for cultures A, G and H (dashed lines) are superimposed on the plots for each of the other cultures. The timing of the various treatments is summarized on each graph – a density shift (from heavy to light-medium) and a[3H]arabinose pulse (followed by a chase with non-radioactive arabinose, indicated by the second vertical line), preceded or followed by the addition of some or all of the following: XGOs (to 1 mm); BFA (to 48 µm); DMSO (to 0.28% v/v).

Preliminary evidence for restructuring transglycosylation

Cultures A and F received no further treatment (A) or only a small dose of DMSO (F) after the cold arabinose chase at +2 h (Table 1). In these two cultures, subsequent density changes gave evidence suggesting the occurrence of transglycosylation of the ‘restructuring’ type. The buoyant density of the recently deposited [3H]xyloglucan increased from 29% heavy at +2 h to a mean of 38% heavy at +11 h (Figure 3A,F). This increase is the expected result of restructuring transglycosylation involving reactions between the now wall-localized, intermediate-density [3H]xyloglucan and the previously wall-bound, heavy, non-radioactive xyloglucan (Figure 1).

The gradual increase in buoyant density of the radiolabelled xyloglucan was not expected to continue indefinitely. This is partly because, during the cells' continued growth on light substrate, the total population of wall-bound xyloglucans (from which a given wall-bound [3H]xyloglucan chain can potentially select a partner for restructuring transglycosylation) gradually becomes predominantly light. In addition, each new, light, non-radioactive chain (secreted after the pulse of 3H) has a chance of selecting one of the previously wall-bound 3H-labelled chains as its partner for integrational transglycosylation, and thus of decreasing the average density of the radioactive population. Both these factors would contribute to the steady decline in buoyant density of the [3H]xyloglucan seen between +11 h and +7 days.

Different effects of exogenous oligosaccharides on integrational and restructuring transglycosylation

The application of exogenous, xyloglucan-derived oligosaccharides (XGOs) provided further evidence for the occurrence of two different types of transglycosylation. XGOs can act as acceptor substrates for all known XETs, typical Km values being 10–100 µm (Fry et al., 1992; Steele and Fry, 2000). We confirmed that culture filtrates normally contained only low concentrations of XGOs (undetectable by TLC under the conditions used here), and that when exogenous XGOs were added to cultures B, C and E to a final concentration of 1 mm, these oligosaccharides persisted in the medium for at least 7 days (Figure 4). There was a gradual conversion of the nonasaccharide (XLLG) to an octasaccharide (possibly XXLG), both of which, however, are known to be good acceptor substrates for XETs (Fry et al., 1992; Steele and Fry, 2000).

Figure 4.

Thin-layer chromatography of culture filtrates from cultures A–C and E (Table 1) sampled at 2 h (a); 11 h (b); 1 day (c); 3 days (d); 5 days (e); and 7 days (f) after the addition of [3H]arabinose. Marker mixtures of xyloglucan-derived hepta-, octa- and nonasaccharides (DP 7–9) were also chromatographed (m); DP7 had RF = 0.20. The solvent was butan-1-ol/acetic acid/water (2 : 1 : 1).

The addition of XGOs to a final concentration of 1 mm at +2.5 h (Figure 3b) prevented the gradual increase in buoyant density of [3H]xyloglucan which otherwise occurred between +2 and +11 h (Figure 3A,F). This indicates that the presence of competing acceptor substrates blocked restructuring transglycosylation reactions occurring between pairs of previously wall-bound xyloglucans. The [3H]xyloglucan in the presence of XGOs did still undergo a long-term decrease in buoyant density between +11 h and +7 days (Figure 3B). Thus, even in the presence of competing acceptor substrates, at least one type of interpolymeric transglycosylation still took place – presumably integrational transglycosylation involving a reaction between newly secreted, non-radioactive, light xyloglucans and the previously wall-bound [3H]xyloglucans of moderate density.

When XGOs were added to 1 mm just before the [3H]arabinose (Figure 3C), the new [3H]xyloglucan still quickly acquired an intermediate buoyant density, confirming that integrational transglycosylation was not prevented by the presence of competing acceptor substrates. At +2 h in culture C, the [3H]xyloglucan was found in hybrid molecules in which, on average, 22% of the chain was heavy (compared with about 29% in cultures A, B, D, E and F). Between +2 and +11 h, the [3H]xyloglucan of culture C increased in average buoyant density, eventually reaching 28% heavy (Figure 3C). As exogenous XGOs inhibit restructuring transglycosylation, we suggest that the gradual increase in buoyant density seen in culture C between +2 and +11 h may be due to somewhat delayed integrational transglycosylation. A decrease in buoyant density (between +11 h and +7 days), thought to be due to both types of transglycosylation operating concurrently (see above), still occurred (Figure 3C).

Additional evidence for restructuring transglycosylation

The effects of brefeldin A (BFA) provided the strongest evidence for restructuring transglycosylation. BFA inhibits the fusion of secretory vesicles with the plasma membrane (Driouich et al., 1993), and should therefore block integrational transglycosylation by preventing xyloglucan secretion. BFA had a negligible effect on cauliflower XET activity when assayed in vitro (data not shown); therefore any effects of BFA in vivo were not due to direct inhibition of XET action.

We confirmed that in rose cell-suspension cultures BFA inhibits the secretion of xyloglucan into the cell wall and culture medium for at least 8 h, and causes accumulation of newly synthesized [3H]xyloglucan within the protoplasts (Figure 5). BFA was added to the cultures as a solution in DMSO. We confirmed that DMSO alone had a negligible effect on xyloglucan metabolism (compare Figure 3A, without DMSO, and Figure 3F, with DMSO).

Figure 5.

Effect of 48 µm BFA (filled symbols) on the formation of [3H]xyloglucan in cultured rose cells and on its compartmentalization between protoplast contents (detergent-soluble), cell wall (alkali-soluble) and culture medium (filtrate). Controls (open symbols) received an equal dose of DMSO.

The addition of BFA at +2.5 h did not prevent the recently wall-bound [3H]xyloglucan increasing in buoyant density (between +2 and +11 h) at the same rate (Figure 3D) as in the control cultures (Figure 3A,F). This supports the idea that the increase in density was due to restructuring transglycosylation between pairs of xyloglucan molecules that were already located within the cell wall at the time the BFA was added, new secretion having been blocked by BFA.

If BFA had inhibited the increase in density in this experiment (Figure 3D), such an effect would have been difficult to interpret because the BFA would probably have blocked the secretion of new XET as well as of new xyloglucans. Fortunately, however, this ambiguity did not arise: BFA had no effect on the density changes of the [3H]xyloglucan. Evidently the XETs secreted before the application of BFA were sufficient for restructuring transglycosylation to continue.

The [3H]xyloglucan in the BFA-treated culture D continued to increase in buoyant density after +11 h, indicating the continued occurrence of restructuring transglycosylation between the wall-bound [3H]xylo glucan and the limited pool of previously deposited (heavy, non-radioactive) wall-bound xyloglucans (Figure 3D). By +7 days, the wall-bound [3H]xyloglucan segments had become grafted to segments of these older xyloglucan chains to form hybrid molecules in which, on average, 55% of the chain was heavy.

In culture E, BFA was added at +2.5 h (to block integrational transglycosylation) as well as XGOs (to block restructuring transglycosylation). The [3H]xyloglucan of culture E was thus expected to undergo little change in buoyant density after +2.5 h. The data (Figure 3E) generally agree with this prediction.


Integrational transglycosylation

In this paper we provide new evidence for the occurrence in vivo of integrational transglycosylation, that is, that involving a reaction between a xyloglucan molecule that is already bound within the wall architecture and a newly secreted (‘incoming’) xyloglucan molecule.

The data show that a newly secreted xyloglucan chain (or part thereof), at about the time of its integration into the wall matrix, becomes attached to another xyloglucan molecule (or part thereof) which was already wall-bound (Figure 2). The product of this grafting process is detectable because it is radioactively labelled and has an intermediate buoyant density, indicating that it is a heavy–light hybrid molecule. A priori, several hypotheses could be proposed as to the nature of the attachment between the two formerly separate xyloglucan molecules. Most of these hypotheses are ruled out by the available evidence (Table 2). Although we are interested in the formation of xyloglucan–pectin conjugates (Thompson and Fry, 2000), we deliberately removed such conjugates in the present experiments by ion-exchange chromatography. The only known, naturally occurring bond that could account for the observed properties of the neutral, heavy–light hybrid xyloglucan molecules (Table 2) is a glycosidic bond between one xyloglucan chain and another:

Table 2.  Possible bonds that could theoretically lead to the formation of heavy–light ‘hybrid’ xyloglucan chains in the plant cell wall
Possible linkageNature of bond(s)Evidence against
  • a

    In principle such a linker could be any of a very wide range of non-anionic compounds with at least two hydroxy groups, e.g. glycerol or inositol.

  • b

    In principle such a linker could be any compound with two carboxy groups, e.g. oxalate, succinate or diferulate.

xyloglucan–xyloglucanHydrogen bonds and/or
van der Waals
Would be unstable under conditions used, as shown by
the fact that 3H- and 14C-labelled xyloglucans, run as a mixture,
can form discrete peaks in CsTFA (Thompson et al., 1997)
xyloglucan–[linker]–xyloglucanIonic bondsNo charged groups known in xyloglucan; bonds would be reversibly
cleaved during extraction in 6 M NaOH
xyloglucan→pectin←xyloglucanTwo glycosidic bondsWould be removed at the anion-exchange chromatography step
(Thompson and Fry, 2000)
xyloglucan→cellulose←xyloglucanTwo glycosidic bondsWould be insoluble in 6 M NaOH
xyloglucan→protein←xyloglucanTwo glycosidic bondsWould be zwitterionic and thus removed at the anion-exchange
chromatography step
xyloglucan→[linkera]←xyloglucanTwo glycosidic bondsNo such linker known, but no definitive counter-evidence
xyloglucan–[linkerb]–xyloglucanTwo ester bondsNo such linker known; would be permanently cleaved during extraction in
6 M NaOH
xyloglucan–phenolic–xyloglucanTwo phenolic ether
No such bonds known in xyloglucan; the rose cell walls used have no
measurable ferulate, coumarate or lignin
xyloglucan→xyloglucanGlycosidic bondNone [favoured hypothesis]

inline image

(where the arrow represents a glycosidic bond). Once such a bond had been made, there need be no chemical distinction between it and any of the numerous other Glc–β–(1→4)–Glc bonds in the backbone of the two formerly separate polysaccharide chains. Such a bond could be formed by the action of an XET, and this is our favoured hypothesis. The alternative hypothesis, a glycosidically linked complex of the type

inline image

or a larger complex, for example

inline image

would require us to propose an unknown linker group (Table 2) as well as an unknown enzyme activity; this hypothesis is therefore less plausible.

Glycosidic bonds are formed by transglycosylation, and the only known enzymes that could catalyse the formation of a xyloglucan→xyloglucan hybrid molecule within the apoplast are XETs. We conclude that the integration of xyloglucan into the cell wall involves an interpolymeric transglycosylation reaction, catalysed by an apoplastic enzyme which is probably an XET.

Integrational transglycosylation was not blocked by high concentrations of exogenous XGOs, which would potentially act as competing acceptor substrates for XETs. However, this observation does not provide evidence against the proposed role of XETs in integrational transglycosylation. Possible reasons why XET-catalysed integrational transglycosylation was not blocked by exogenous XGOs include the following.

(i) Integrational transglycosylation may occur at a local site where the endogenous xyloglucan concentration is very high, and therefore exogenous XGOs at 1 mm compete only weakly with the polysaccharides as acceptor substrates for XETs. This site of active integration would be on the inner face of the cell wall at a position where a Golgi-derived vesicle had recently discharged its contents by exocytosis. The contents of the vesicle are an aqueous solution of matrix polysaccharides and other solutes. We would expect most of the water released from such a vesicle to be rapidly squeezed through the fabric of the existing cell wall by the pressure of the turgid protoplast against the wall. On the other hand, most of the polysaccharides released from the vesicle would be caught on the inner face of the wall owing to its molecular sieving properties (Baron-Epel et al., 1988; Tepfer and Taylor, 1981). At this local site of high xyloglucan concentration, exogenous XGOs would have little chance to compete with the intended (polymer-to-polymer) transglycosylation reaction.

(ii) Integrational transglycosylation may be catalysed by XET isoenzymes having a much lower affinity (Purugganan et al., 1997) or catalytic efficiency (Steele and Fry, 2000) for XGOs than for xyloglucans as the acceptor substrate.

Our evidence for interpolymeric transglycosylation in vivo is especially pertinent because both donor and acceptor substrates are natural, endogenous polysaccharide chains, distinguished from each other only by isotopic labelling.

Is the incoming xyloglucan the donor or the acceptor substrate?

It is not possible to deduce from the present experiments whether the incoming (= radioactive) partner was usually the donor or usually the acceptor substrate in the integrational transglycosylation reaction. This is a significant question for future research, because the donor is cleaved in the process and a portion of it may therefore be lost from the cell wall (Figure 2a,d,e)– presumably an undesirable side effect in a process designed for efficient wall assembly.

As we envisage it, there are five possible outcomes of transglycosylation between a newly secreted chain and a previously wall-bound chain (Figure 2). Integration without the undesirable side effect of leaving any xyloglucan unattached to microfibrils could, in principle, be achieved if XET selected as the donor substrate a xyloglucan chain that was already wall-bound, and cleaved it within a part of its length that was either directly hydrogen-bonded to cellulose (Figure 2b,b′) or acting as a tether between two microfibrils (Figure 2c). An XET could in principle select tethers (in the wall of a turgid cell) by recognizing their tautness (Fry, 1989a); it could in principle select segments that are directly cellulose-bonded by recognizing their characteristic conformation (Levy et al., 1997). It might be suggested that directly cellulose-bonded segments would be a poor substrate because of steric hindrance. However, future work is required to determine whether any isoenzymes of XET do possess this kind of donor substrate specificity.

Cleavage may thus occur predominantly in hydrogen-bonded and tethering zones of xyloglucan; however, cleavage is unlikely to be absolutely restricted to these zones. This is shown by the observation that, in vivo, some xyloglucan is sloughed from the cell wall into the culture medium. Although some of these sloughed molecules are newly synthesized (Becker et al., 1964; Edelmann and Fry, 1992b) and may represent incoming molecules that escaped the integration mechanism, others are older molecules (or parts thereof), released from the cell wall slowly over a period of days (Thompson and Fry, 1997). This latter process, described as ‘trimming’, may result from XET- or glucanase-catalysed cleavage of wall-bound xyloglucan chains in zones a or d (or twice in succession in zone c) (Figure 2). Unfortunately, the sloughed xyloglucan is of low Mr and so it fails to form a tight band in CsTFA gradients; we have therefore not analysed its density.

As an alternative experimental approach for distinguishing donor and acceptor substrates, it is possible to monitor the covalent attachment of exogenous, fluorescently labelled XGOs to wall-bound xyloglucans in vivo (Ito and Nishitani, 1999). This approach has provided evidence for the ability of one specific partner (an existing wall-bound xyloglucan) to act as the donor substrate (Vissenberg et al., 2000). Unfortunately, this experimental approach does not provide complementary data on the ability of incoming molecules to act as the donor substrate. This is because the observable incoming molecule is an XGO, which is too small to act as a donor for XET. In addition, the incoming XGO is small enough freely to permeate the wall, whereas in natural circumstances the incoming molecule would be a polysaccharide secreted by the protoplast and trapped by a sieving effect at the wall's inner face (see above). There is, therefore, no guarantee that the majority of the fluorescent XGO molecules react at the same site or in the same way as do newly secreted polysaccharide chains. Thus, while the observed attachment of wall-bound xyloglucans to exogenous, fluorescent XGOs (acceptor substrates) (Ito and Nishitani, 1999; Vissenberg et al., 2000) indicates that existing wall-bound xyloglucan chains are indeed cleaved by endogenous XETs in vivo, it does not indicate whether the natural acceptor substrate, in the absence of the fluorescent XGOs, would have been any of the following possibilities: (i) a newly secreted xyloglucan chain, leading to its integration; (ii) another existing wall-bound xyloglucan chain, leading to restructuring; (iii) the same xyloglucan as had just been cleaved (an ‘idling’ reaction); (iv) an endogenous XGO; or (v) H2O (with the XET slowly acting as a hydrolase in the absence of accessible carbohydrate acceptor substrates).

Restructuring transglycosylation

In this paper we provide the first in vivo evidence for interpolymeric transglycosylation of the restructuring type (a reaction between a pair of xyloglucan molecules, both of which had previously been integrated into the cell wall; Figure 1). Again, the advantage of our approach is that both donor and acceptor substrates were natural, endogenous polysaccharide chains, distinguished from each other only by isotopic labelling. The results support the speculative models (Fry et al., 1990; Fry et al., 1992; Nishitani, 1998) in which XET was proposed to catalyse the cleavage and re-formation of xyloglucan tethers within the growing cell wall.

The increase in buoyant density of [3H]xyloglucan, occurring between 2 and 11 h after [3H]arabinose feeding, provided initial evidence for restructuring transglycosylation in vivo. It is interesting to note that in the 2 to 11 h interval, a subpopulation of [3H]xyloglucans is known to be converted from weakly to firmly wall-bound, as judged by the concentration of NaOH required for solubilization (Edelmann and Fry, 1992b). This subpopulation was described as being in a ‘slow lane’ towards firm wall-binding, unlike the major population of xyloglucans in the same cell culture, which were in the ‘fast lane’ as they very quickly became firmly wall-bound. It is possible that the xyloglucan molecules in the ‘slow lane’ initially (within the first 2 h) bind to the wall relatively weakly by hydrogen bonding to the surfaces of microfibrils, without undergoing integrational transglycosylation, and then gradually (between 2 and 11 h) become more firmly wall-bound by restructuring transglycosylation. The xyloglucan molecules in the fast lane, in contrast, are proposed to be those that undergo integrational transglycosylation immediately after secretion across the plasma membrane. Further density-labelling experiments, comparing the densities of weakly and firmly wall-bound xyloglucan chains, would permit this hypothesis to be tested.

Restructuring transglycosylation, unlike integrational transglycosylation, was blocked by the addition of high concentrations of XGOs (Figure 3B), which are competing acceptor substrates for XETs. The inhibition by XGOs of restructuring transglycosylation supports the conclusion that it was XET-catalysed.

The cell-suspension culture used in the present work routinely undergoes considerable cell expansion (the cells are highly vacuolated) as well as rapid cell division. However, unlike some excised organs (e.g. segments cut from coleoptiles, hypocotyls, etc.), the cultured rose cells do not exhibit any dramatic promotion of cell expansion in response to low pH or to elevated auxin concentrations. It was therefore not feasible to test whether the rate of ‘restructuring’ transglycosylation increases during an acid- or hormone-induced step up in the rate of cell expansion. Such an increase might be predicted because XET activity (and XET gene expression) is often positively correlated with the rate of cell expansion (Campbell and Braam, 1999; Fry et al., 1992; Vissenberg et al., 2000). Future work involving the application of our methods to excised organs will be necessary to explore this aspect.

Biological roles of integrational and restructuring transglycosylation

It is interesting to compare the possible biological roles of integrational and restructuring transglycosylation. We suggest that both types of transglycosylation may be involved in both cell expansion and wall assembly. For example, cleavage of a tethering xyloglucan chain by XET, regardless of whether this cleavage will culminate in restructuring transglycosylation (Figure 1) or integrational transglycosylation (Figure 2c), would locally disconnect a pair of neighbouring microfibrils and could thus transiently loosen the wall, permitting incremental cell expansion. Integrational transglycosylation could clearly also play a role in wall assembly, by enabling newly secreted segments of xyloglucan to link to the existing xyloglucan–cellulose network; but so too could restructuring transglycosylation if it resulted in the conversion of weakly wall-bound to firmly wall-bound xyloglucan, as suggested above to account for the results of Edelmann and Fry (1992b).

Integrational and restructuring transglycosylation may be in competition with each other. Cleavage of a xyloglucan chain by XET (whether acting in integrational or restructuring mode) initially results in the formation of a polysaccharide–enzyme covalent complex (Figure 1b), which can persist for minutes or even hours in the absence of a suitable acceptor substrate (Steele and Fry, 1999; Sulováet al., 1998). In the cell wall, acceptor substrates (the non-reducing ends of xyloglucan chains) may sometimes be relatively rare or inaccessible to the enzyme. Under these circumstances, a high percentage of the wall's XET molecules may be tied up at any given time in polysaccharide–enzyme complexes, unsuccessfully seeking acceptor substrates. The release of these enzyme molecules, required for the cleavage of additional tethers, and thus for continued wall expansion, could be limited by the availability of acceptor substrates. In this situation, the secretion of new xyloglucan molecules (enabling the XET to proceed with integrational transglycosylation and thus to be freed to act again) could be a growth-controlling factor. Thus wall loosening could be promoted not only by XET secretion, but also by secretion of new wall material (xyloglucans capable of acting as acceptor substrates), even though the key requirement for cell expansion is the loosening of existing wall material.


The main finding of the present paper is that both integrational and restructuring transglycosylation occur concurrently, and probably in competition with each other, during the normal growth of cultured cells. Integrational transglycosylation occurs very soon after xyloglucan secretion; restructuring transglycosylation occurs more gradually over a period of hours or days. We suggest that, by catalysing both types of transglycosylation, XETs serve important roles in both assembly and loosening of the cell wall, together enabling long-term plant cell expansion with minimal loss of wall strength.

Experimental procedures


Cell-suspension cultures of rose (Rosa sp. Paul's Scarlet) were subcultured fortnightly by 10-fold dilution into medium (Fry and Street, 1980) containing 2% d-glucose as sole carbon source. ‘Heavy’ glucose was d-[U-13C; 1,2,3,4,5,6,6-2H]glucose (in which ≈99% of the C atoms were 13C and ≈50% of the non-exchangeable H atoms were 2H) from Aldrich, (Gillingham, Dorset, UK). l-[1-3H]Arabinose (148 TBq mol−1) was from Amersham International, (Little Chalfont, Bucks, UK). A mixture of XGOs containing the three major Glc4-based XGOs, in order of abundance XLLG > XXLG > XXXG (for nomenclature see Fry et al., 1993), was kindly donated by Mr K. Yamatoya (Dainippon Pharmaceutical Co., Osaka, Japan), and the hepta- to nonasaccharide fraction was further purified by gel-permeation chromatography on Bio-Gel P-2. CsTFA was from Pharmacia Biotech, (Uppsala, Sweden), and BFA was from Sigma, (Poole, Dorset, UK).

Dual-labelling protocol

The protocol is summarized as a timetable of treatments for each of eight cultures, A–H, in Table 1. Time-point 0 in the timetable is the moment of addition of [3H]arabinose. Aseptic conditions were maintained throughout the incubation. At time-point −23 days, for cultures A–G, cells from a 14-day-old rose culture were washed three times in sugar-free medium, then inoculated into medium containing 50 mm heavy glucose as the sole carbon source, and allowed to grow for 2 weeks. At −9 days the culture was diluted 10-fold into fresh medium containing 25 mm heavy glucose. After 9 days in this medium, the cells had consumed 90% of the glucose (assayed in the medium) and the suspension culture was dispensed into seven Petri dishes (A–G) at 18 ml per dish. A control culture, which had been treated as above but in media containing only ‘light’ glucose throughout, was dispensed at 18 ml into Petri dish H. At time-point −0.5 h, a 20% (w/v) solution of light glucose was added to seven of the dishes (A–F, H) to a final concentration of 100 mm; as a control, dish G simultaneously received heavy glucose. After a further 0.4 h incubation period, which enabled the cells' pools of all major polysaccharide precursors (hexose phosphates, the glucose residue of UDP-glucose, etc.) to become fully saturated with the new carbon source (Thompson et al., 1997), XGOs were added to culture C to a final concentration of 1 mm. At time zero, 5 MBq of [3H]arabinose was added to each of the eight cultures. At time-point +2 h (by which time ≈90% of the 3H had been taken up), a chase of non-radioactive arabinose (5 mm) was applied to each culture so that no further appreciable synthesis of [3H]xyloglucan could occur (except to culture G, in which it was important to exclude all ‘light’ carbohydrate). Also at time-point +2 h, a 1.5 ml portion of each culture was sampled. At time-point +2.5 h, XGOs were added (to a final concentration of 1 mm) to cultures B and E; BFA (to 48 µm, added in DMSO) was added to cultures D and E; and an equivalent dose of DMSO (final concentration 0.28%, v/v) was added to culture F. Further 1.5 ml aliquots of each culture were harvested at intervals up to time-point +7 days.

Paper chromatography of culture filtrates showed that glucose and non-radioactive arabinose were available to the cells up to at least +7 days (data not shown).

Analysis of [3H]xyloglucan

The cells in each 1.5 ml sample were quickly washed through a Bio-Rad, (Bio-Rad Laboratories Ltd., Hemel Hempstead, Herts. UK) ‘PolyPrep’ tube fitted with a polypropylene filter, rinsed with sugar-free medium and stored at −80°C. The culture filtrates plus washings were analysed chromatographically, as described below. The frozen cells were thawed into 5 ml of 6 m NaOH containing 1% NaBH4 and shaken at 37°C for 24 h, a treatment optimized for the solubilization of hemicelluloses from the rose cultures (Edelmann and Fry, 1992c). The extract was neutralized, dialysed against water and freeze-dried. The hemicellulose obtained was redissolved in 0.5 ml 8 mm pyridinium acetate (PyOAc, pH 4.7); xyloglucan was precipitated from this solution by addition of 50% ethanol followed by incubation at 25°C for 18 h, and the resultant pellet was again dried, redissolved in 1 ml 8 mm PyOAc containing 8 m urea (pH 5.3) and applied to a 2 ml bed-volume column of Q-Sepharose FastFlow (pre-equilibrated with 8 m urea in 8 mm PyOAc). Neutral xyloglucan was eluted from the column with a further 2 ml of 8 m urea in 8 mm PyOAc. This eluate contained 70% of the total cellular xyloglucan, at a radiochemical purity of 94.6 ± 2.28% (mean ± SE; n = 34) (Thompson and Fry, 2000). Analysis of the sugar residues by non-radiochemical methods revealed a purity of 96.7% xyloglucan (Thompson and Fry, 2000).

A 2.5 ml portion of this purified solution of neutral xyloglucan was mixed with 9 ml CsTFA solution (density 2.0 g ml−1) plus 2.75 ml 500 mm PyOAc containing 1.7 kBq [14C]acetylated tamarind xyloglucan (Thompson et al., 1997), added as an internal marker. The resulting solution, which had a density of 1.6 g ml−1, was centrifuged in a vertical rotor [Sorvall StepSaver 65V13 (Stevenage, Herts. UK) at 255 000 g for 40 h at 18°C, with slow deceleration] to produce a linear density gradient of ≈1.45–1.75 g ml−1. Each such gradient was fractionated into 250 µl fractions by downward displacement with white mineral oil, pumped at 0.5 ml min−1. The density of each fraction was estimated from the weight (±0.1 mg) of a 200 µl portion, which was then assayed for 3H and 14C by scintillation counting. Curves were fitted to the radioactivity profiles, and mean buoyant densities of the [3H]xyloglucan samples were estimated, as described earlier (Thompson et al., 1997).

Analysis of mono- and oligosaccharides in the culture medium

Aliquots of each culture filtrate (plus washings), equivalent to 5 µl of neat filtrate, were assayed for free glucose and arabinose by paper chromatography (Fry, 2000; Thompson and Fry, 1997). Additional aliquots (equivalent to 13 µl of neat filtrate) of selected culture filtrates were analysed for XGOs by TLC on silica gel in butan-1-ol/acetic acid/water (2 : 1 : 1 by vol) and stained by spraying with 0.5% (w/v) thymol in EtOH/H2SO4 (19 : 1, v/v) followed by heating at 105°C for 20 min

Possible side effects of brefeldin A

The effect of BFA (0.1–100 µm) on cauliflower XET activity was tested in vitro, using tamarind xyloglucan and [3H]XLLGol as donor and acceptor substrates, respectively, in the assay described earlier (Fry et al., 1992). The effect of BFA on integration of xyloglucan into the cell wall was tested in vivo: 15 min after the addition of BFA (to a final concentration of 48 µm) to a 50 ml, 7-day-old rose cell culture, 4 MBq l-[3H]arabinose was added. Samples (5 ml) of the culture, taken 0.25–8 h later, were fractionated into compartments (protoplasts, cell walls and medium), and each compartment was assayed for [3H]xyloglucan as described before (Thompson et al., 1997).


We thank the BBSRC for a grant in support of this work.