Xyloglucans exist as cell wall components in most species and as storage polymers in seeds of some species (94). Xyloglucan (XG) comprises 20-25% of primary walls of dicots but graminaceous monocots typically contain much less. XG is defined by a (3-1,4-glucan back­bone in which many glucosyl residues contain a-1,6-linked xylose branches. Xyloglucan from pea had an average molecular mass of 330 kDa representing a backbone of about 1100 glucose residues of about 500 nm in length (95). In many species the xylose residues are further substituted with (31,2-linked galactose which may in turn be linked at the 2-position

Table 5.1 Common elements of single letter code for xyloglucan structure

Code Structure represented

G p—d-GIc p*-a

X a—D-Xylp-(1^6)-p-D-Glcp*-

L p—D-Galp-(1^2)-a-D-Xylp-(1^6)-p-D-Glcp*-

F a—L-Fucp-(1^2)-p-D-Galp-(1 ^2)-a-D-Xylp-(1^6)-p-D-Glcp*- a*- D-Glucose in chain or at reducing terminus. See Fry and coworkers (97) for details.

to a-L-fucose or arabinose (94). X-ray fiber diffraction studies of tamarind XG indicated a twofold helix similar to cellulose (96). A single letter code has been developed to describe the structure of xyloglucans (Table 5.1) (97).

Most species have an XXXG type of XG. However, members of Poacea and Solanaceae have an XXGG type in which a pair of arabinose residues replace fucose (98, 99). In most monocots XG contains less xylose and galactose and does not contain terminal fucose. The structure and molecular distribution of the side chains varies in different plant tissues and species (100-102).

XG maybe extensively acetylated (103). In Sycamore cells, the O-2-linked-p-D-galactosyl residue of the nonasaccharide was found to be the dominant site of O-acetyl substitution in XG. Both mono-O-acetylated and di — O-acetylated p-D-galactosyl residues were detected. Thedegreeof O-acetylation ofthe p-D-galactosyl residue was estimated to be 55-60% at O-6, 15-20% at O-4, and 20-25% at O-3. Approximately, 50% ofthe p-D-galactosyl residues were mono — O-acetylated, 25-30% were di — O-acetylated, and 20% were not acetylated. In tomato (Lycopersicon esculentum), O-acetyl substituents were located at O-6 of the unbranched backbone p-D-Glcp residues, O-6 of the terminal p-D-Galp residue, and/or at O-5 of the terminal a-L-Arap residues (104). Acetylation of XG does not affect the degree to which XG hydrogen bonds to cellulose in vitro (100) and the role of acetylation is unknown. Similarly, the enzymes that acetylate XG are unknown. O-acetylation of galactose residues was considerably reduced in Fuc-deficient mutants (atfutl, murl, and mur2) that synthesize XG containing little or no Fuc (105). These results suggest that fucosylated XG is a suitable substrate for at least one O-acetyltransferase in Arabidopsis.

Immunoelectron microscopy using antibodies against XG indicates that XG is localized to the cellulose-containing region of the cell wall (106). Hayashi (94) proposed that XG does not have covalent cross-links to other components or if there are links they must be alkali-labile linkages such as O-esters. However, Thompson and Fry (107) have observed cross-links between XG and pectins in Rose cells. Brett and coworkers (108) have also observed such cross-links and found that they form in the Golgi. Feruloyl esters of XG have also been observed in maize cell cultures (109).

Pure XG binds to cellulose in vitro inapH-dependent manner (110). Levy etal. (111) have presented evidence that the structure of the XG side branches may facilitate the binding of XG to cellulose. Native XG-cellulose complexes contain higher ratios than can be obtained in vitro, suggesting that XG maybe intercalated into the cellulose microfibrils (110). Also, mild alkali does not completely dissociate the complex and concentrated alkali (e. g., 4M KOH) is required to completely extract XG. The proposed function of XG binding to cellulose is to prevent aggregation of cellulose fibrils (110) but because single strands of XG maybe hydrogen bonded to different cellulose microfibrils (112-114); it may also provide some degree of crosslinking. Thus, XG hydrolysis maybe required for growth. Based on the com­bined chemical and cytological evidence, Pauly and coworkers (100) have developed a model for the cellulose/XG network that posits that XG can have three configurations; hydrogen bonded to the surface of cellulose, cross-linked, and embedded within the microfibril. They propose that the cross-links are the domain that is accessible to enzymes such as xyloglucan endotransglycosylases that are thought to play a role in cell wall expansion. This is supported by observations of XET-mediated incorporation of fluorescent XG fragments into XG in expanding cell walls (115). Pauly and coworkers (100) also note that it is not clear to what extent the various XG structures participate in determining the nature of the XG-cellulose association.

The first progress in defining the genes involved in XG synthesis was the identification of the fucosyltransferase that adds the terminal fucose to XG side chains. A 60-kDa fucosyltrans — ferase (FTase) that adds this residue was purified from pea epicotyls (116). Peptide sequence information from the pea FTase allowed the cloning of a homologous gene, AtFUT1, from Arabidopsis. AtFUT1 expressed in mammalian COS cells resulted in the presence of XG FTase activity in these cells. AtFUT1 shows very little identity with FucTs from other organ­isms. AtFUT1 andPsFUT1 (the peaXyG FucT homologue) are 62.3% identical (117). Both enzymes contain motifs that had been identified in other FucTs but combine these motifs in a unique manner (117). Three motifs had been identified in (1^2)a — and (1^6)a-FucTs. Motifs I and II had been present in both (1^2)a — and (1^6)a-enzymes, but a particular version of motif III had appeared to be characteristic of each group. AtFUT1 and PsFT1, however, contain a hybrid motif III that has features of both the (1^2)a — and (1^6)a — versions. There are ten genes in Arabidopsis with identity of encoded amino acid sequences to AtFUT1 ranging from 35 to 73.8% (118). The AtFUT1 gene was found to correspond to the fucose-deficient mur2 mutant of Arabidopsis (119).

The galactosyltransferase that contributes to the synthesis of XG side chains was identi­fied by map-based cloning of the MUR3 gene of Arabidopsis, which had previously been identified based on a screen for variation in cell wall polysaccharide composition (120). MUR3 belongs to a large family of Type II membrane proteins that is evolutionarily con­served among higher plants. The enzyme shows sequence similarities to the glucuronosyl transferase domain of exostosins, a class of animal glycosyltransferases that catalyze the syn­thesis of heparan sulfate, a glycosaminoglycan with numerous roles in cell differentiation and development. Arabidopsis has ten genes encoding proteins with significant sequence similarity to the MUR3 xyloglucan GalT (121).

One of the XG xylosyltransferases (XT1) was identified from Arabidopsis based on se­quence similarity to the fenugreek mannan a-1,6-galactosyltransferase (122). Expression of the gene in Pichia pastoris resulted in a protein with cello-oligosaccharide-dependent xylosyltransferase activity. Characterization of the products obtained with cellopentaose as acceptor indicated that the pea and the Arabidopsis enzymes transfer xylose mainly to the second glucose residue from the non-reducing end in an a(1,6)-linkage to the glucan chain. Arabidopsis has seven related genes, some of which may catalyze addition of xylose to other positions in the repeating unit of XG.

In vitro assays of the glucan synthase involved in synthesis of the XG backbone exhibit maximal activity only if both UDP-glucose and UDP-xylose are present, suggesting that the glucan synthase acts in concert with a xylosyltransferase that adds side chains (94, 123).

The glucan synthase extends existing XG by addition to the non-reducing end but cannot be primed with exogenous primers (94). Also the enzyme does not add xylose to preformed glucans.

Recently, Cocuron and coworkers (124) have obtained evidence that proteins of the cellulose synthase-like 4 (CSLC4) family catalyze synthesis of the XG backbone (99). They expressed CSLC4 genes from Arabidopsis and tamarind along with the Arabidopsis XT 1 gene in Pichia pastoris and observed the formation of p-1,4-linked glucan. However, they were unable to detect XG synthase activity in extracts from the cells. Mutations in the Arabidopsis CSLC4 gene are deficient inxyloglucan, supporting the proposed role inxyloglucan synthesis (Milne and Somerville, unpublished). If substantiated by further work, the work of Cocuron and coworkers (124) appears to represent a long-awaited breakthrough in understanding the synthesis of XG (99). The identification of the genes involved in XG synthesis should pave the way for an analysis of how the amount of XG is regulated and what the consequences are to plant growth and development and the properties of cell walls of genetic variation in the amount of XG.

Some information about genetic variation in XG is available from analysis of Arabidopsis mutants that were recovered by screening for alterations in total cell wall sugar compo­sition (125, 126). Comparison of the mechanical responses of mur2 (AtFUT1) and mur3 (XG galactosyltransferase), indicated that galactose-containing side chains of xyloglucan make a major contribution to overall wall strength, whereas xyloglucan fucosylation plays a comparatively minor role (127). Thus, it seems unlikely that it will be possible to develop biomass feedstock crops with significant alterations in the structure of XG without also making compensating changes in another cell wall component. Because Arabidopsis has a number of CSLC genes, it has not yet been possible to develop mutant plants with major reductions in the amount of XG to assess the phenotypic consequences of such alterations.