Molecular domains of the cellulose/xyloglucan network in the cell walls of higher plants


  • Markus Pauly,

    Search for more papers by this author
    • Present address: Plant Biochemistry Laboratory, Department of Plant Biology, The Royal Veterinary and Agricultural University, 40 Thorvaldsensvej, DK-1871 Frederiksberg C, Denmark.

  • Peter Albersheim,

    1. Complex Carbohydrate Research Center and Department of Biochemistry and Molecular Biology, University of Georgia, 220 Riverbend Road, Athens, GA 30602-4712, USA
    Search for more papers by this author
  • Alan Darvill,

    1. Complex Carbohydrate Research Center and Department of Biochemistry and Molecular Biology, University of Georgia, 220 Riverbend Road, Athens, GA 30602-4712, USA
    Search for more papers by this author
  • William S. York

    Corresponding author
      *For correspondence (fax +1 706 542 4412; e-mail
    Search for more papers by this author

*For correspondence (fax +1 706 542 4412; e-mail


Cellulose and xyloglucan (XG) assemble to form the cellulose/XG network, which is considered to be the dominant load-bearing structure in the growing cell walls of non-graminaceous land plants. We have extended the most commonly accepted model for the macromolecular organization of XG in this network, based on the structural and quantitative analysis of three distinct XG fractions that can be differentially extracted from the cell walls isolated from etiolated pea stems. Approximately 8% of the dry weight of these cell walls consists of XG that can be solubilized by treatment of the walls with a XG-specific endoglucanase (XEG). This material corresponds to an enzyme-susceptible XG domain, proposed to form the cross-links between cellulose microfibrils. Another 10% of the cell wall consists of XG that can be solubilized by concentrated KOH after XEG treatment. This material constitutes another XG domain, proposed to be closely associated with the surface of the cellulose microfibrils. An additional 3% of the cell wall consists of XG that can be solubilized only when the XEG- and KOH-treated cell walls are treated with cellulase. This material constitutes a third XG domain, proposed to be entrapped within or between cellulose microfibrils. Analysis of the three fractions indicates that metabolism is essentially limited to the enzyme-susceptible domain. These results support the hypothesis that enzyme-catalyzed modification of XG cross-links in the cellulose/XG network is required for the growth and development of the primary plant cell wall, and demonstrate that the structural consequences of these metabolic events can be analyzed in detail.


All higher plant cells are encased in a cell wall, which defines the cell’s shape and thereby contributes to the structural integrity of the entire plant. Growing cells are surrounded by a metabolically active primary cell wall, which is capable of expanding ( Kerr & Bailey 1934). The primary cell wall of all higher plants consists of crystalline cellulose microfibrils, embedded in a hydrated (65% water; Fry 1988), amorphous matrix of hemicelluloses, pectins, and glycoproteins ( Darvill et al. 1980 ) that form a network around the cell. According to current cell-wall models, the cellulosic framework is interconnected by hemicellulosic polysaccharides, such as xyloglucan (XG) or arabinoxylan, forming a cellulose/hemicellulose network ( Carpita & Gibeaut 1993). Co-existing with the cellulose/hemicellulose network is another network that consists of the pectic polysaccharides homogalacturonan, rhamnogalacturonan I and rhamnogalacturonan II. In addition, primary cell walls often contain structural proteins such as the hydroxyproline-rich glycoprotein extensin.

The structural rigidity and strength of the wall is thought to depend on the integrity of the cellulose/hemicellulose network. Enzyme-catalyzed modification of the hemicellulosic component of this network is considered to be essential for wall expansion during cell growth ( Talbott & Ray 1992). A more complete knowledge of the structure, macromolecular organization, and metabolism of this network is necessary to understand the mechanisms leading to plant cell elongation and the biological regulation of this process.

The structural components of the cellulose/hemicellulose network are well known. Cellulose microfibrils consist of non-covalently associated, linear chains of β-1,4-linked d-glucopyranosyl residues. XG is the major hemicellulosic component in dicotyledonous and non-graminaceous monocotyledonous plants. Like cellulose, XG has a β-1,4 linked glucospyranosyl backbone, which is branched, with many of its β- d-glucosyl residues bearing an α- d-xylosyl residue at C6 ( Fig. 1). The xylosyl residues can be further substituted to form oligomeric side chains containing galactosyl, fucosyl, and/or arabinosyl residues. The structure and molecular distribution of these XG side chains varies in different plant tissues and species. The chemical structures of these side chains have been rigorously established by analysis of xyloglucan oligosaccharides (XGOs) that are generated upon endoglucanse digestion of the polymer. However, relatively little is known about the macromolecular organization of XG in the cell wall.

Figure 1.

Two xyloglucan-derived oligosaccharides (XGOs).

Endoglucanases such as cellulase or XEG hydrolyze the unbranched β- d-glucosyl residues of the XG backbone, to generate XGOs such as those shown here. XGO structures described in the text and in Table 4 are specified using a standard nomenclature ( Fry et al. 1993 ), in which an uppercase letter indicates the precise side-chain substitution pattern of each β- d-glucosyl residue in the oligosaccharide. The letter G represents an unbranched β- d-glucosyl residue, and the letter X represents a β- d-glucosyl residue with an α- d-xylosyl substituent at O6, etc. Thus, the sequences XXXG and XLFG unambiguously define the two structures shown.

Perhaps the most important feature of XGs is their characteristic capacity to form strong, non-covalent associations with cellulose ( Valent & Albersheim 1974; Hayashi 1989; Hayashi et al. 1994 ). XGs bind to cellulose in vitro in a pH-dependent manner ( Hayashi et al. 1987 ), suggesting that the formation of hydrogen bonds is involved in the association of these polymers. Concentrated alkali (e.g. 24% KOH) is required to extract XG from the cellulose/XG complex ( Hayashi 1989; Hayashi & Maclachlan 1984). Antibody studies ( Vian et al. 1992 ) suggest that XGs are associated with the surface of cellulose microfibrils in muro. Cellulose microfibrils cross-linked by putative XG ‘tethers’ have been visualized by electron microscopy of rotary shadowed replicas of rapidly frozen, deep-etched cell-wall specimens ( McCann et al. 1990 ; McCann et al. 1992 ; Itoh & Ogawa 1997) and of artificially assembled bacterial cellulose/XG composites ( Whitney et al. 1995 ). XG cross-links were also suggested by studies in which cell-wall epitopes recognized by an XG-specific antibody were observed predominantly in the spaces between microfibrils ( Baba et al. 1994 ).

This paper describes a model for the cellulose/XG network based on the observation that differentially extractable xyloglucan fractions are structurally distinct. Three different macromolecular domains are proposed for the XG in this network. The data presented here suggest that the metabolism of XG associated with cell expansion occurs predominantly in the most accessible of the three domains.


Quantitative solubilization of xyloglucans from pea-stem cell walls by a sequential extraction procedure

The macromolecular organization of XG in the cell walls of etiolated pea (Pisum sativum L.) stems was investigated. This tissue was chosen because: (i) the structural features of XG in this tissue are well established ( Hayashi et al. 1984 ; Guillen et al. 1995 ); (ii) sufficient quantities of this tissue can be easily obtained; and (iii) metabolic turnover of XG in this tissue has been studied previously ( Hayashi & Maclachlan 1986; Talbott & Ray 1992). XG was solubilized from the pea-stem cell walls using a sequential extraction procedure (see Experimental procedures). Destarched cell walls were partially depectinated by treatment with a purified endopolygalacturonase (EPG) in combination with pectin methylesterase (PME). This procedure increases the amount of XG that can be solubilized by the subsequent steps (data not shown), confirming a previous observation ( Bauer et al. 1973 ) and suggesting that removal of pectin increases the enzyme accessibility of XG.

The depectinated cell walls were treated with an XG-specific endoglucanase (XEG), which selectively solubilizes XGOs ( Pauly et al. 1999 ). Under the conditions used, this treatment solubilized approximately 8.2% of the wall material ( Table 1). XEG was removed from the soluble extract by passing it through an anion-exchange resin and the salts were removed by size-exclusion chromatography (SEC) on Sephadex G-10. The sugar composition of the desalted fraction confirmed that only XGOs are present in this fraction ( Table 1). The use of additional XEG, extended digestion times, or multiple XEG treatments did not significantly increase the amount of wall material solubilized by this method (data not shown). Colorimetric analysis of the extract indicated that XEG solubilized 8.0% of the cell walls as XGOs, in good agreement with the gravimetrically determined decrease in the insoluble cell-wall material (8.2%) observed upon XEG treatment.

Table 1. . Yields and sugar compositions of XGO-enriched fractionsa, b obtained by sequential extraction of pea-stem CWM with XEG, KOH, and cellulase
  • a

    XGO-enriched fractions were prepared as described in Experimental procedures.

  • b

    Data are the average of two extraction experiments.

  • c

    Yields are wt% relative to the initial mass of the partially depectinated pea-stem cell wall.

  • d

    Determined by mass reduction of the residual CWM.

  • e

    Determined by the anthrone assay.

  • f

    Sugar compositions are normalized mol% of each XGO-enriched fraction.

  • g

    Not detected.

Yield (wt%) c
 Amount solubilized d8.233.827.0
 Recovered as XGOs e8.010.33.9
Composition (mol%) f
 Rhamnosend gndnd

Additional XG was solubilized when the XEG-treated walls were extracted with concentrated alkali (24% KOH). Approximately one-third of the initial mass of the partially depectinated wall material was solubilized by this treatment ( Table 1), but further base treatment did not release a significant amount of additional material. In addition to XG, the KOH extract contained acidic components of the cell wall, including pectic polymers and glucuronoarabinoxylan, which were removed by passing the extract through a column of Q-Sepharose. (Prior to Q-Sepharose chromatography, the desalted extract was treated with XEG to generate XGOs.) The monosaccharide composition of the neutral flowthrough (i.e. the XGO-enriched fraction, Table 1) was consistent with the presence of XG (glucose, xylose, galactose, and fucose), but arabinose was also abundant (42.4%), probably due to the presence of a neutral arabinoxylan or an arabinan. A small amount of mannose in the extract suggested that it also contained a mannan. Quantification of the XG present in this fraction by the anthrone method was not significantly compromised by the presence of the arabinose or xylose because the anthrone assay is insensitive to these pentoses. However, the anthrone assay is sensitive to mannose, and the sugar-composition data in Table 1 were used to estimate the colorimetric response of mannose in the KOH extract. After taking this ‘mannose factor’ into account, the anthrone assay indicated that XG in the KOH-solubilized material comprised 10.3% of the initial mass of the partially depectinated cell walls ( Table 1).

The amount of XG remaining in the cell walls after XEG and KOH treatment was determined by analysis of the monosaccharides released by Saeman hydrolysis of the residue. Saeman hydrolysis involves the use of concentrated H2SO4 (see Experimental procedures) to depolymerize all of the polysaccharide components of the wall, including microcrystalline cellulose ( Selvendran et al. 1979 ). The resulting monosaccharides are not significantly degraded ( Selvendran et al. 1979 ) and can be quantitated by GLC analysis of their alditol acetate derivatives ( Table 2). This analysis was also performed on cell walls that had not been treated with XEG and KOH. Under the conditions described above, 58% of the cell-wall mass remained after sequential extraction with XEG and KOH. Therefore, the mass percentages of the monosaccharides ( Table 2) present in the residual cell walls were normalized to give a total of 58%. Glucose is the most abundant sugar detected in both treated and untreated cell walls. Most of this glucose probably originated from the hydrolysis of cellulose, as cellulose accounts for up to 30% of the dry weight of the cell walls of higher plants ( Fry 1988). The monosaccharide-composition data shown in Table 2 indicate that sequential XEG and KOH treatment of the cell walls solubilized significant amounts of several cell-wall-matrix polysaccharides (XG, xylans, mannans, and pectic polysaccharides). Most of the cellulose remained in the insoluble residue (hereafter called the XEG/KOH residue). The XEG/KOH residue was treated with cellulase, which solubilized nearly all of the remaining matrix polysaccharides but less than half of the insoluble cellulose.

Table 2. . Sugar composition (mass %) of Saeman hydrolysates of cell-wall material
SugarsPartially depectinated
cell walls (untreated)
Cell walls treated
with XEG and KOH a
Cell walls treated
with XEG, KOH, and cellulase a
  • a

    Cell walls were sequentially extracted.

  • b The data are normalized by reference to the total mass of the untreated cell walls, which is set to 100%. As indicated in Tables 1, 58% of the cell wall remained insoluble after sequential treatment with XEG and KOH, and 31% remained insoluble after subsequent treatment with cellulase.

  • c

    Average of two extraction experiments.

  • d

    Not detected.

Total b1005831
Rhamnose0.1 cnd dnd

It is difficult to determine accurately the amount of XG present in the insoluble cell-wall fractions. The presence of cellulose in these fractions makes quantification of glucose a poor indicator of their XG content. However, quantitative analysis of xylose can be used to estimate the XG content of the XEG/KOH residue because xylose is present in all XGOs. The XEG/KOH residue contains xylose, which could be due to the presence of xylan and/or XG. However, xylans are more readily solubilized by KOH than are XGs ( York et al. 1985 ). Therefore, most of the xylose in the XEG/KOH residue is probably derived from XG that is not solubilized by this treatment. Analysis (described below) of the oligosaccharides released by cellulase treatment of the XEG/KOH residue is consistent with this conclusion.

Retreatment of the XEG/KOH residue with XEG was used in an attempt to solubilize more XGOs, as the KOH treatment might have exposed some of the XG that had previously been inaccessible to XEG. However, no additional wall material was released by the second XEG treatment (data not shown). The ability of a commercially available cellulase to solubilize additional XGOs was also tested. Cellulase treatment solubilized an additional 27% of the partially depectinated walls ( Table 1). Two carbohydrate-rich fractions were obtained when the cellulase-solubilized material was desalted by SEC (data not shown) ( Pauly et al. 1999 ). The sugar composition and elution volume of the very low-molecular-weight fraction indicated that it contained cellobiose and glucose, the products expected upon enzymatic degradation of cellulose. The sugar composition ( Table 1) and elution volume of the second fraction indicated that XGOs were its main constituents, and suggested that it may also have contained a small amount of larger cello-oligosaccharides that were co-eluted with the XGOs. Quantitative analysis of the second fraction by the anthrone method showed that XG released by this cellulase treatment consitituted an additional 3.9% of the cell wall ( Table 1). The insoluble residue obtained after sequential treatment with XEG, KOH, and cellulase comprised 31% of the mass of the depectinated cell wall. Therefore, the mass percentages of the monosaccharides ( Table 2) present in this α-cellulose fraction were normalized to give a total of 31%. As expected, glucose is the most abundant sugar present from this fraction. Xylose constituted only 0.2% of this fraction, indicating that xylose-containing wall polysaccharides, including XG, are almost quantitatively solubilized by the sequential extraction procedure.

These results show that the sequential extraction procedure, employing XEG, KOH (24%), and finally cellulase, solubilized three distinct domains of the XG present in the cell walls. Here, domain is not defined as a structural entity within a molecule, but rather as an ‘environmental’ domain based on its extractability. However, the results do not establish whether individual XG molecules occupy a single domain or span more than one domain.

Connectivities between the different solubilized XG domains

Additional insight into the physical relationships among the various XG domains was obtained by analysis of XG that was extracted by KOH with no prior XEG treatment. In this case, KOH-solubilized XG accounted for 18.4% of cell walls (data not shown). Thus, KOH alone solubilizes approximately the same amount of XG as sequential extraction with XEG and KOH (18.3%, Table 1), suggesting that XEG-extracTable XGs can also be solubilized with KOH. Furthermore, KOH treatment of the wall does not break glycosidic bonds in the XG backbone, releasing XG as a polysaccharide.

These results suggested the following experiments to determine whether the XEG- and KOH-extractable domains are covalently connected. The molecular mass distributions of KOH-solubilized XG obtained with or without prior XEG treatment of the cell wall material (CWM) were determined by SEC on Superose 12 ( Fig. 2). Fractions were collected and assayed using the anthrone assay, which detects hexoses, and iodine staining, which selectively detects XGs with a molecular weight greater than approximately 20 kDa. The molecular weights of the KOH-solubilized XGs were estimated from their SEC elution volumes, using dextrans and tamarind XGOs as standards.

Figure 2.

The effects of prior XEG treatment on the Superose 12 profile of XGs solubilized by KOH treatment of partially depectinated pea-stem cell walls.

Fractions were collected and assayed for hexoses by the anthrone assay (●) and for XG by iodine staining (○). Both assays involved measuring the absorption at 620 nm. Dextrans (superscript a) and BEPS-XGOs (superscript b) of known molecular weight (expressed as kDa) were used to calibrate the column, as indicated by the arrows.

Two carbohydrate peaks dominate the chromatogram of the KOH extract prepared without prior treatment of the CWM with XEG ( Fig. 2, top panel). The iodine- and anthrone-positive peak indicated the presence of an XG with an average molecular weight of 100 kDa. The smaller anthrone-positive (iodine-negative) peak indicated the presence of a low-molecular-weight (∼2 kDa) glycan. The monosaccharide composition of the two peaks indicated that their main components were XG and arabinan/arabinogalactan, respectively (data not shown). Two peaks were also observed in the chromatogram obtained by SEC of the KOH extract prepared from XEG-treated cell walls ( Fig. 2, lower panel). A very small amount of material was eluted at the void volume of the column, indicating the presence of a large polymer with a molecular weight of more than 500 kDa. This polymer did not stain with iodine and consisted mainly of xylose (data not shown), indicating it was a xylan rather than an XG. The bulk of the material solubilized by KOH after XEG treatment was eluted at a volume corresponding to an average molecular weight of approximately 30 kDa. Iodine staining and monosaccharide composition of this material indicated that it consisted mainly of XG, but also had a high arabinose content (data not shown), indicating the presence of an arabinan. Thus, prior treatment with XEG decreased the molecular weight of the KOH-extracted XGs from approximately 100 to 30 kDa, suggesting that individual XG molecules span two domains, of which one can be solubilized by XEG while the other can be extracted only upon treatment with KOH. The results are also consistent with the hypothesis that the XEG-accessible XG domain is interdispersed within the KOH-extracTable XG domain. One intriguing possibility is that, on average, an approximately 28 kDa stretch at each end of a typical 100 kDa XG molecule is closely associated with cellulose (and thus resistant to XEG), and that the intervening approximately 44 kDa is susceptible to XEG attack. This hypothetical arrangement is consistent with both the SEC data ( Fig. 2) and the data in Table 1, as the two ‘cellulose-bound’ ends of the polysaccharides and the intervening ‘cross-link’ would represent 10.3 and 8.0%, respectively, of the dry weight of the cell wall.

In vitro experiments with an XG/Avicel composite

An artificial aggregate, composed of XG and microcrystalline cellulose (Avicel), was also analyzed by the sequential extraction method. This system is largely free of other cell-wall components that could complicate analysis of the composite. As the binding of XGs to cellulose in vitro does not involve enzymes, the molecular structure of the XG, including its chain length and molecular topology, is unlikely to be modified in the binding process. It is also unlikely that XG would become entrapped within a cellulose microfibril or irreversibly entangled with cellulose chains upon binding to Avicel. Thus, the molecular topology of this artificial system is likely to be much simpler than that of a plant cell wall synthesized in vivo. This relative simplicity makes it possible to directly examine the interaction of XG molecules with the surface of previously formed cellulose microfibrils. The results presented here demonstrate that this interaction, in itself, cannot lead to the complex topology of a real cell wall. Nevertheless, analysis of the XG/Avicel system indicated that association of XG with the cellulose surface gives rise to two distinct, differentially extracTable XG domains that are analogous to the XEG- and KOH-extracTable XG domains in pea stem cell walls.

XG obtained from the extracellular polysaccharides of a bean suspension culture (BEPS-XG) was bound to Avicel in vitro as described in Experimental procedures. The XG/Avicel composite contained approximately 1 mg BEPS-XG per 15 mg Avicel, as estimated by colorimetric analysis of XGs remaining in the supernatant after incubation with the cellulose (data not shown). The XG/Avicel composite and ‘naked’ Avicel were each treated with buffer, XEG, cellulase, or KOH under the same conditions used to extract the pea cell wall. Each treatment was considered ‘complete’, as repeated extraction with a given reagent did not solubilize additional carbohydrate (data not shown). The monosaccharide components of each of the solubilized fractions were quantified by GLC of their alditol acetate derivatives ( Table 3).

Table 3. . Glycosyl composition (μg) material extracted from Avicel (15 mg) or the XG/Avicel composite (1 mg BEPS-XG/15 mg Avicel)
  • a

    Not detected.

Rhamnosend andndndndndndnd

Buffer extraction of either naked Avicel or the XG/Avicel complex released a negligible amount of carbohydrate. XEG released a total of 156 μg of carbohydrate from the XG/Avicel complex. Thus, after accounting for the amount of carbohydrate (4.6 μg) released upon XEG treatment of naked Avicel, these results indicate that approximately 15% of the XG in the complex was released by XEG.

Considerably more carbohydrate was released by cellulase treatment of the XG/Avicel complex than by XEG treatment. Most of the extracted material was glucose generated by enzymatic depolymerization of the Avicel itself. As expected, cellulase treatment of naked Avicel also solubilized a considerable amount of glucose-containing polysaccharides. Cellulase solubilized only 2.5 μg of xylose from naked Avicel but 113.2 μg of xylose from the Avicel/XG complex. Considering that xylose represents approximately 30% of the mass of BEPS-XG, and that 1 mg of BEPS-XG is bound in the XG/cellulose complex, XG contributes approximately 300 μg of xylose to the composite. Hence, cellulase treatment of the complex solubilized approximately 37% of the bound XG.

KOH solubilized a large amount of glucose from naked Avicel but only minute quantities of other sugars (e.g. mannose, xylose, and arabinose). However, KOH treatment of the XG/Avicel complex released 301.4 μg of xylose, indicating that essentially all of the XG in the complex was solubilized by this treatment.

XGO composition of the XG-enriched fractions obtained from pea stems

The XGO compositions of the sequentially solubilized XEG, KOH, and cellulase fractions were determined in order to establish whether any structural feature of the XG is correlated with its presence in a particular domain. Solubilized XGOs were reductively aminated with p-nitrobenzylhydroxy-amine (PNB, Pauly et al. 1996 ). The resulting PNB-XGOs were separated by reversed-phase chromatography and quantified by integrating their UV absorbance ( Pauly 1999) ( Table 4). O-Acetylated XGOs present in the XEG-solubilized fraction were separated and quantified by this method. However, base treatment of the walls hydrolyzes any O-acetyl substituents that may be present, and O-acetylated glycans were not present in the fractions solubilized by KOH and subsequent cellulase treatment. The overall compositions of the three fractions are very similar, with XXXG and XXFG being the main oligosaccharide subunits in all XG domains – see Fig. 1 and Fry et al. (1993) for a description of the XGO nomenclature. However, only the XEG-accessible domain contains the ‘xylose-deficient’ oligosaccharides GXXG and GXFG. Furthermore, the amount of XXG, an oligosaccharide that may be formed by enzymatic trimming of the non-reducing end of the XG, is correlated to the accessibility of the domain, and is almost undetectable in the cellulase-extracted domain.

Table 4. . XGO-compositiona, b of fractions solubilized by sequential XEG, KOH, and cellulase treatment
Oligosaccharide cNot acetylatedMono-acetylatedDi-acetylatedKOH
Not acetylated d
Not acetylated d
  • a See Fry et al. (1993) and Fig. 1 for a description of the XGO nomenclature used here.

  • b

    Normalized mol%, average of two extraction experiments.

  • c See Fig. 1 for nomenclature.

  • d

    All acetyl substituents are hydrolyzed by KOH treatment.

  • e

    Not identified: no standards are available for these particular O-acetylated forms, as we have never observed them as constituents of cell wall XGs.

  • f

    Not detected.

XXG12.7ni eni5.00.3
GXXG0.9ninind fnd


An extended model for the XG/cellulose network in pea stems

Most current models of the primary cell wall include a cellulose/XG network in which the XG cross-links the cellulose microfibrils ( Hayashi & Maclachlan 1984). These models predict that at least two XG domains exist in the cell wall, with one domain bound directly to cellulose microfibrils and a second domain forming the cross-links. We extend this model for the macromolecular organization of the cellulose/XG complex based on the analysis of pea-stem cell walls ( Fig. 3). The pea epicotyl tissues used in this study are heterogeneous mixtures of cell types. Thus, this analysis cannot fully describe changes that occur in the XG/cellulose network of any specific cell. However, the overall trends that were observed reveal the presence of three XG domains that are likely to occur in the cell walls of most non-solanaceous, dicotyledonous plants. In our extended model, XG that is extractable by XEG corresponds to a molecular domain that constitutes the XG cross-links between cellulose microfibrils and any exposed tails or loops of XG that extend away from the microfibril surface. The XEG-extractable domain is covalently attached to the KOH-extracTable XG domain, which is non-covalently bound to the surface of cellulose microfibrils. Close association of the KOH-extractable domain with the microfibril surface makes it inaccessible to the XEG. KOH treatment of cell walls that have not been pretreated with XEG releases both the XEG- and KOH-extractable domains. A third XG domain is entrapped within or between cellulose microfibrils, where neither XEG or KOH can gain access. The entrapped XG may reside in relatively non-crystalline (amorphous) regions of the microfibril. This entrapped XG can be released by cellulase treatment, which depolymerizes amorphous regions of the cellulose and hydrolyzes the unbranched glucosyl residues in the XG backbone. The entrapped XG domain consists of molecules that are rarely, if ever, contiguous with the XEG- and KOH-extracTable XG domains, as the XEG and KOH domains can be solubilized in their entirety (by KOH) without breaking covalent bonds (i.e. without XEG pretreatment) and without releasing any of the entrapped XG. These data do not rule out the possibility that some of the entrapped XG molecules may extend onto the surface of cellulose microfibrils. Such molecular extensions may not be accessible to XEG, due to their close association with the microfibril surface, and would not be solubilized by KOH, due to their covalent attachment to the entrapped XG domain.

Figure 3.

Model accounting for differential extractability of XG domains in the XG/cellulose network.

This model emphasizes the likely arrangement of the various XG domains in different micro-environments, but is not meant to quantitatively represent the mass ratio of these domains. Cellulose microfibrils are indicated by the grey bars. The XG domain that can be solubilized with XEG (blue bars) is proposed to be accessible to enzyme-catalyzed modification in vivo. This domain may form loops, dead ends, or cross-links between cellulose microfibrils. The XG domain that can be solubilized by KOH (red bars), with or without prior XEG treatment, is bound to the surface of the cellulose microfibrils. The XG domain that remains insoluble after XEG/KOH extraction of the cell wall (light green bars and dots in the cross-sections of some of the cellulose microfibrils) is trapped within or between cellulose microfibrils. Some of this XG domain may also reside on the surface of cellulose microfibrils, remaining resistant to XEG due to its close association with the microfibril, and resistant to KOH extraction due to its covalent attachment to entrapped XG. Little or none of the XG that can be solubilized by XEG appears to be covalently attached to entrapped XG.

Further support for this model comes from analysis of an XG/cellulose composite produced by mixing the two components in vitro. This system is simpler than the XG/cellulose network in the cell wall because XG is bound exclusively to the surface of the microfibrils, and none is internalized (trapped) as may occur in vivo when cellulose is synthesized in the presence of XG. XG that binds to the surface of cellulose in vitro may form loops or cross-links between microfibrils. XEG treatment of this artificial system solubilized approximately 15% of the bound XG; cellulase treatment solubilized approximately 34% of the bound XG; and KOH treatment solubilized virtually 100% of the bound XG ( Table 3). Even multiple XEG treatments solubilize only a small portion of the XG, consistent with the hypothesis that XEG can attack only the XG domain that constitutes XG cross-links and/or loops. XG was more efficiently solubilized by treating the XG/Avicel system with cellulase than by treating it with XEG, consistent with the hypothesis that partial depolymerization of cellulose exposes additional XG, making it accessible to enzyme. Virtually all of the XG in the artificial XG/cellulose system was solubilized by KOH treatment, supporting the hypothesis that the KOH-extractable domain is closely associated with the surface of the cellulose microfibril.

It is clear from the results presented here that a significant portion of the XG molecules in the cell wall span two molecular domains that are selectively solubilized by sequential XEG and KOH treatment. These two domains almost certainly correspond to different micro-environments within the cell wall. The ‘enzyme-accessible domain’, operationally defined by its extractability by XEG, represents the putative XG cross-link. The ‘surface-bound domain’ can be solubilized by KOH but not by XEG ( Fig. 3). This model predicts that the enzyme-accessible domain can be modified in vivo by plant cell-wall enzymes, including endoglucanases (EGs), xyloglucan endotransglycosylases (XETs), and exoglycosidases such as α-fucosidases ( Augur et al. 1992 ) or β-galactosidases ( Ross et al. 1993 ). The model also predicts that surface-bound domain cannot be modified by enzymes such as EGs and XETs that attack the glucosyl backbone of the XGs, but might be modified by exoglycosidases that attack XG side chains extending away from the microfibril. A third ‘trapped domain’, corresponding to XG that is not solubilized by XEG and KOH, is likely to be completely inaccessible to cell-wall enzymes in vivo. Although the enzyme-accessible XG domain is the most likely to be modified during the assembly and expansion of the cell wall, it is important to note that our results do not establish whether a given XG molecule can evolve from one domain to another during plant cell development.

The physical basis for segregating XG into separate domains in the primary cell wall is poorly understood. This process is likely to be affected by the primary structure of the XG and any enzyme-catalyzed modifications that affect its primary structure or topology. For example, it is possible that the presence of O-acetyl substituents could affect the rate and extent to which an XG molecule binds to cellulose. However, the degree of O-acetylation in the surface-bound domain cannot be readily determined because solubilization of this domain requires strong alkali. Nevertheless, cellulose-binding experiments using native (O-acetylated) and de-O-acetylated BEPS-XG indicated that the presence of O-acetyl groups does not affect the amount of XG that binds to cellulose in vitro ( Pauly 1999). Another possibility is that binding to cellulose may be modulated by the local sequence of XGO subunits in the polysaccharide. Our data do not allow us to evaluate this hypothesis because, although the XGO subunit composition was determined for each of the three XG domains, the order of the various XGOs in the original polysaccharide was not.

Most XG subunits have an even number of glucosyl residues (e.g. XXXG, XXFG), leading to a regular topology that facilitates binding to cellulose ( Levy et al. 1991 ). It has been proposed that the binding of XG to cellulose can be disrupted by a ‘topological reversal’ that occurs at sites consisting of XG subunits (e.g. XXG) that have an odd number of glucosyl residues. This proposal is consistent with the observation that the occurrence of XG subunits (such as XXG) having an odd number of residues decreases as the XG domain is more tightly associated with the cellulose microfibril ( Table 4). However, it is also possible that this trend is a result of intimate contact with cellulose, rather than its cause. That is, XG subunits such as XXG may arise by sequential trimming of exposed non-reducing ends of the polysaccharide, converting, for example, XXXG to GXXG and then to XXG. Such enzyme-mediated transformations are most likely to occur in the enzyme-accessible domain, consistent with the data presented in Table 4.

The model presented here does not represent a radical departure from previously published hypotheses concerning the macromolecular structure of XG in the primary cell wall. However, the results demonstrate that there are three distinct XG domains in the primary walls of pea stem cells, and that it is possible to study the individual characteristics of each of these domains. Defining the genesis and metabolism of these domains and analyzing their conformational, topological, and dynamic properties are extremely challenging problems, the solutions to which will require the application of a broad range of genetic, biochemical, and physical techniques.

Experimental procedures

Chemicals, reagents, substrates and enzymes

Buffer salts, acids, bases, and organic solvents were obtained from J.T. Baker (Philipsburg, NJ, USA); other reagents were purchased from Sigma Chemical Co. (St Louis, MO, USA).

XGO fragments of the soluble XG secreted by suspension-cultured bean cells (BEPS-XGOs) were prepared as described ( Wilder & Albersheim 1973).

Endopolygalacturonase (EPG) from Aspergillus niger was obtained from Dr Carl Bergmann, CCRC. Pectin methylesterase (PME) and xyloglucan-specific endoglucanase (XEG) ( Pauly et al. 1999 ) from Aspergillus aculeatus was obtained from Novo-Nordisk (Copenhagen, Denmark). XEG was purified as described ( Pauly et al. 1999 ).

Colorimetric assays

The total carbohydrate content of samples was quantified using the anthrone assay for hexoses as described ( Dische 1962). A known amount of BEPS-XGOs was used as a standard when determining the XG or XGO content of a sample. The XG content of samples was estimated by an iodine-staining assay ( Kooiman 1960). Samples were dissolved in 100 μl water, Gram-stain (75 μl aqueous KI, 6.6 g l−1, and I2, 3.8 g l−1) and sodium sulfate (500 μl aqueous Na2SO4, 0.2 g ml−1) were added, and the solution was incubated for 1 h. A 200 μl aliquot of the solution was transferred to a microtiter plate and the absorbance at 620 nm was measured. The uronic acid content (e.g. galacturonic acid from pectic polysaccharides) of samples was quantitated using the m-hydroxybiphenyl assay ( Blumenkrantz & Asboe-Hansen 1973).

Isolation of cell walls

Cell walls were prepared from whole stems of 9-day-old etiolated pea plants ( Guillen et al. 1995 ). The entire isolation procedure was carried out at 4°C. Harvested tissue was suspended (1 g FW ml−1) in potassium phosphate buffer (100 m m, pH 7.0) containing 5 m m Na2S2O5 as an antioxidant, and homogenized in a Polytron (Brinkman, Westburg, NY, USA) at maximum speed for 5 min. The homogenized tissue was filtered through a triple layer of nylon mesh (1 mm pore size) and washed three more times with the same buffer. The solid residue was resuspended in the same volume of 500 m m phosphate buffer (pH 7.0) containing 5 m m Na2S2O5, filtered through the nylon mesh and washed three more times with the same buffer. The residue was then resuspended in the same volume of 0.5% aqueous SDS containing 3 m m Na2S2O5, and stirred for 20 h. The suspension was filtered and the solid residue was washed with water and then resuspended in a 1 : 1 mixture of chloroform and methanol, homogenized for 3 min with the Polytron, and filtered. The insoluble residue (CWM) was then washed with acetone and dried under vacuum at 30°C.

To remove starch, the CWM was suspended (0.5 g DW ml−1) in 100 m m potassium phosphate, pH 7.0, containing antibiotic (0.01% thimerosal) and α-amylase (EC, Type IIA from Sigma, 2.5 mg g−1 FW). The suspension was stirred at room temperature for 48 h, then filtered through four layers of nylon mesh, washed three times with deionized water, washed with acetone, and dried.

The CWM was partially depectinated using a combination of EPG and pectin methylesterase PME. Dried CWM was suspended (10 mg DW ml−1) in 100 m m sodium-acetate buffer (pH 5.2) containing 0.01% thimerosal. EPG (five units) and PME (five units) were added and the suspension was incubated for 24 h at 37°C. Solubilized material was then removed by filtration through a nylon membrane (10 μm pore size; Micron Separations Inc., Westboro, MA, USA) in a polysulfone filter funnel (50 ml capacity; Gelman Sciences, Ann Arbor, MI, USA). The partially depectinated CWM (retained on the filter) was then dried.

Solubilization of XGOs using XEG

XG was solubilized from cell walls using XEG ( Pauly et al. 1999 ) that had been purified to remove glycosidases that might otherwise degrade XGOs. CWM was suspended (100 mg DW) in 10 m m sodium acetate (pH 4.5, 10 ml) containing 0.02% thimerosal. XEG (10 units) was added and the suspension was incubated for 24 h at 37°C. After the XEG digestion, the suspension was filtered through a nylon membrane (10 μm pore size) in a polysulfone funnel to separate ‘XEG-XGOs’ (filtrate) from the XEG-treated CWM.

Solubilization of XGs using 24% KOH

XEG-treated CWM was suspended (100 mg DW) in 24% (w/v) aqueous KOH (10 ml) containing 0.1% NaBH4. The suspension was stirred at room temperature for 24 h and then filtered through a nylon membrane (10 μm pore size) in a polysulfone funnel. The filtrate was neutralized with glacial acetic acid and salts were removed by dialysis [3500 MWCO tubing (Spectrum, Houston, TX, USA)] versus deionized water (4°C). The retentate, containing the KOH-solubilized XG, was buffered by adding 1 m sodium acetate, pH 4.5, to a final concentration of 10 m m. XGOs were generated by adding purified XEG (five units) and incubating the solution for 24 h at 37°C.

Solubilization of XGs using cellulase

A commercial cellulase (Megazyme, Bray, Ireland) was used to solubilize XG from the KOH-extracted CWM. The insoluble residue recovered after KOH-treatment was suspended (100 mg DW) in 10 m m sodium acetate, pH 4.5 (10 ml), containing 0.02% thimerosal. Cellulase (10 units) was added and the suspension was incubated for 48 h at 37°C. After the cellulase digestion the suspension was filtered through a nylon membrane (10 μm pore size) in a polysulfone funnel.

Isolation of XGOs

Solubilized XGOs were subjected to anion-exchange chromatography on a Q-sepharose column (Sigma) to remove the enzymes (cellulase or XEG) or pectic polysaccharides (cosolubilized by 24% KOH). The column (5 ml in a 1.5 × 15 cm plastic tube) was equilibrated with 10 m m imidazole/HCl pH 7.0, and the solubilized XGOs were then applied in the same buffer. The column was washed with two column volumes of the same buffer to elute the neutral components (XGOs), which were desalted by SEC on a Sephadex G-10 column (1.5 × 80 cm, Sigma). The G-10 column was eluted with deionized water, and collected fractions were assayed for salt content by conductivity and for carbohydrate content using the anthrone assay. The resulting ‘XGO-enriched fractions’ were characterized as described below.


The XGO composition of each XGO-enriched fraction was determined by HPLC of PNB-XGOs prepared by reductive amination with p-nitrobenzylhydroxylamine hydrochloride, as described ( Pauly et al. 1996 ). PNB-XGOs were separated by HPLC on a Vydac 238TP54 reversed-phase C-18 column (Vydac, Hesperia, CA,USA). The sample was eluted from the column with a linear gradient from 7 to 12% aqueous acetonitrile (1 ml min−1) over 40 min, followed by a linear gradient from 12 to 23% aqueous acetonitrile over 20 min. PNB-XGOs were detected by monitoring the UV absorbance (275 nm) of the eluant with a Beckman UV 163 variable wavelength detector.

Saeman hydrolysis

Saeman hydrolysis ( Selvendran et al. 1979 ) was used to hydrolyze plant cell walls for monosaccharide composition analysis. Plant cell walls (2–4 mg) were wet with 100 μl aqueous myo-inositol (1 mg ml−1, internal standard) in a screw-capped borosilicate test tube. Concentrated H2SO4 (300 μl) was added, the test tube was capped, and the suspension was left for 3 h at RT with occasional vortexing. The suspension was then diluted with 6.6 ml of distilled water and heated for 2 h at 100°C. After cooling, the solution was transferred to a 50 ml Falcon tube and titrated with saturated barium hydroxide [∼35–40 ml] to pH 6–7 to remove sulfate ions, which precipitate as Ba(SO4). The hydrolysate was centrifuged for 5 min at 2000 g and the supernatant filtered through filter paper (Whatman, Fairfield, NJ, USA) into a round-bottomed flask. The pellet was resuspended in water (40 ml) by vortexing and centrifuged again. The resulting supernatant was filtered as described above and combined with the first supernatant. The pooled solution was concentrated by rotary evaporation to approximately 5–6 ml. The monosaccharide composition of the Saeman hydrolysate was determined by analyzing (see below) the alditol acetate derivatives prepared from 500 μl of the solution.

Monosacharide composition analysis

The monosaccharide compositions of the various CWM fractions were determined by gas–liquid chromatography (GLC) of alditol acetate derivatives ( York et al. 1985 ). Monosaccharides were prepared by Saeman hydrolysis ( Selvendran et al. 1979 ) of the insoluble residues recovered after various extractions, or by hydrolysis of water-soluble oligo- and polysaccharides in aqueous trifluoroacetic acid. The internal standard (myo-inositol) was added to the samples before hydrolysis. The resulting monosaccharides were reduced with sodium borohydride and O-acetylated. A standard sample containing a mixture of rhamnose, fucose, arabinose, xylose, mannose, galactose, and glucose was prepared and derivatized in parallel with the experimental samples.

XG binding to cellulose

Cellulose (Avicel PH101, Fluka, Milwaukee, WI, USA) was prewashed at least three times by suspending it (15 mg) in 15 ml incubation buffer (50 m m sodium acetate, pH 4.5, room temperature), centrifuging at 2000 g for 5 min in a Marathon 6K table-top centrifuge (VWR, McGaw Park, IL, USA), and decanting the supernatant. The washed Avicel was resuspended in 1 ml of the incubation buffer and BEPS-XG (1.5 mg) was added. The suspension was vigorously shaken in an incubator (180 r.p.m.) for 12 h at room temperature. After this binding period, the bound and unbound XG was separated by centrifuging the suspension (2000 g for 5 min) and decanting the supernatant. The insoluble Avicel–XG complex was washed twice with 500 μl incubation buffer. The Avicel–XG complex and a control consisting of Avicel without added XG were then treated with XEG, cellulase, or KOH, under the conditions (described above) used to extract XG from CWM. After each treatment the insoluble residue (XG/cellulose or cellulose) was removed by centrifugation and 200 μl of the supernatant was transferred to a fresh test tube. The KOH supernatant (200 μl) was also neutralized and dialyzed to remove the salts. All samples were dried, 25 μg myo-inositol was added as an internal standard, and their glycosyl compositions were determined by GLC of the alditol acetate derivatives of the monosaccharides released by acid hydrolysis (see above).


This research was supported in part by funds from the US Department of Energy grant DE-FG05-93ER20115 and DE-FG09-93ER20097. We would like to thank Carl Bergmann for the pure EPG, and Lene Anderson and Hans Peter Heldt Hansen from Novo Nordisk, Denmark for the XEG-preparation and the PME.