Cellulose microfibrils are insoluble cable-like structures that are typically composed of about 36 hydrogen-bonded glucan chains each of which contains between 500 and 14 000 [3-1,4- linked glucose molecules. Cellulose microfibrils comprise the core component of the cell walls that surround each cell. Studies from mutants deficient in secondary cell wall cellulose show very irregular deposition of non-cellulosic polysaccharides and lignin (4). Thus, it is apparent that cellulose is a central scaffold of cell walls.

The cellulose chains in microfibrils are parallel, and successive glucose residues are rotated 180°, forming a flat ribbon in which cellobiose is the repeating unit. The parallel chains are compatible with evidence that the chains in a microfibril are made simultaneously (3). The cellulose chains are held in a crystalline structure by hydrogen bonds and Van der Waals forces to form microfibrils. It is not yet known to what extent the “crystallization” of the nascent glucan chains into cellulose microfibrils is facilitated by proteins other than the catalytic enzyme. Jarvis (5) has shown that the two main forms of cellulose (i. e., cellulose Ia and Ip) can be interconverted by bending. He suggested that the sharp bend that is thought to take place when cellulose emerges from the rosette and becomes appressed to the overlying cell wall may be sufficient to induce the interconversion. Additional forms, which are primarily of interest in the context of industrial uses of cellulose, can be produced from natural cellulose by extractive treatments. For instance, in cellulose II, the chains are antiparallel — something that is unlikely to occur in native cellulose. Cellulose I is converted to cellulose II by extraction under strongly alkaline conditions.

The molecular weight of the individual glucan chains that comprise cellulose microfibrils has been difficult to determine because the extraction may lead to degradation. Analyses of secondary wall cellulose in cotton suggest a degree of polymerization (DP) of 14 000-15 000 (6). Primary wall cellulose appears to have lower molecular weight. Brown (7) reports a DP of 8000 for primary wall cellulose. However, Brett (6) reported a low molecular weight fraction of ~500 DP and a fraction with a DP of 2000-4000. Brett (6) suggested that the low molecular weight fraction may be chains at the surface of microfibrils whereas the high DP fraction maybe chains in the microfibril interior. Since a DP of 2000 corresponds to about 1 ^m of length, the implication is that the primary wall cellulose fibrils, which are frequently observed to be much longer than 1 ^m, must be composed of chains with breaks at various locations along the fibrils. As noted below, this is compatible with genetic evidence that a cellulase is required for cellulose synthesis in both plants and bacteria (8, 9). Whatever the exact length, it is apparent that in some cells the fibrils can be extremely long relative to other types of biological macromolecules.

Based on electron micrographs, the width of cellulose fibrils varies from about 25 to 30 nm in Valonia and other green algae, to about 5-10 nm in most plants (10). The variation in size may indicate that cellulose microfibrils from different sources contain different numbers of chains, and it may reflect variation in the kind or amount of hemicellulose coating on the fibrils. In a study of onion primary wall by solid state NMR (10), the spectral interpretation was consistent with the idea that the 8 nm wide microfibrils were composed of six 2-nm fibrils, each containing about 10 chains. Herth (11) estimated by electron microscopy that the microfibrils of Spirogyra contained 36 glucan chains. Thus, the measurements are generally consistent with the idea that each of the six globules in a rosette is composed of a number of subunits that synthesize 6-10 chains that hydrogen bond to form the 2 nm fibrils. Six of these 2 nm fibrils then bond to form the microfibrils.

The analyses of cellulose structure indicate that cellulose synthase is a highly processive enzyme, that it has many active sites that coordinately catalyze glucan polymerization, that alternating glucan units are flipped 180°, and that interspecies variation exists in the number of glucan chains per fibril, or possibly in the kind or amount of hemicellulose. What is not clear is whether the enzyme participates in facilitating the hydrogen bonding of the glucan chains or whether proximity of the glucan chains as they emerge from the enzyme is sufficient to cause formation of the highly ordered microfibrils.

Cellulose synthase can be visualized by freeze fracture of plasma membranes in vascular plants as symmetrical rosettes of six globular complexes approximately 25-30 nm in diame­ter. The rosettes have been shown to be cellulose synthase by immunological methods (12). The only known components of cellulose synthase in higher plants are the CESA proteins. The completion of the Arabidopsis genome sequence revealed that Arabidopsis has ten CESA genes that encode proteins with 64% average sequence identity (13, 14) and other species have been found to have similar numbers of CESA proteins (3). The proteins range from 985 to 1088 amino acids in length and have eight putative transmembrane (TM) domains. Two of the TM domains are near the amino terminus and the other six are clustered near the carboxyl terminus. The N-terminal region of each protein has a cysteine-rich domain with a motif that is a good fit to the consensus for a RING type zinc-finger. RING fingers have been implicated in mediating a wide variety of protein-protein interactions in complexes (15). Otherwise, the N-terminal domain is structurally heterogenous among the ten CESAs in Arabidopsis. The average overall sequence identity of the amino terminal domains is 40% compared with an average overall identity of 64%.

A large “central domain” of approximately 530 amino acids lies between the two regions of transmembrane domains and is thought to be cytoplasmic. Using this feature to anchor the topology of the protein indicates that the N-terminal domain is also cytoplasmic. The central domain is highly conserved among all the CESA proteins except for an approximately 64-91 residue region of unknown significance where there is weak sequence identity. The domain contains a motif (Q/RXXRW) that is associated with bacterial cellulose synthases and other processive glycosyltransferases (16), such as chitin and hyaluronan synthases, and glucosylceramide synthase (17). Additionally, a DXXD motif and two other aspartate residues have been associated with this class of enzymes and are referred to collectively as the D, D,D, Q/RXXRW motif. Site-directed mutagenesis experiments of the chitin synthase 2 of yeast showed that the conserved aspartic acid residues and the conserved residues in the QXXRW motif are required for chitin synthase activity (18). Similarly, Saxena etal. (19) replaced the aspartate residues in the A. xylinum cellulose synthase and found that they were required for catalytic activity.

Analysis of mutants with defects in secondary wall cellulose has revealed that three separate CESA proteins are required in the same cell at the same time (20) and that the proteins physically interact (21). Thus, it appears that within a cell type there is a single type of complex containing three types of CESA subunits. A detailed summary of the properties of mutations that alter cellulose accumulation has been published recently (3). In brief, null mutations in several of the primary wall CESA proteins are lethals, whereas others are not, presumably because of redundancy. Mutations that eliminate secondary cell wall cellulose are not lethal but impair the structural integrity of vascular cells. In addition to mutations in CESA genes, a number of other proteins have been implicated in the overall process but the role of these proteins is not understood.