A Molecular Description of Cellulose Biosynthesis

Abstract

Cellulose is the most abundant biopolymer on Earth, and certain organisms from bacteria to plants and animals synthesize cellulose as an extracellular polymer for various biological functions. Humans have used cellulose for millennia as a material and an energy source, and the advent of a lignocellulosic fuel industry will elevate it to the primary carbon source for the burgeoning renewable energy sector. Despite the biological and societal importance of cellulose, the molecular mechanism by which it is synthesized is now only beginning to emerge. On the basis of recent advances in structural and molecular biology on bacterial cellulose synthases, we review emerging concepts of how the enzymes polymerize glucose molecules, how the nascent polymer is transported across the plasma membrane, and how bacterial cellulose biosynthesis is regulated during biofilm formation. Additionally, we review evolutionary commonalities and differences between cellulose synthases that modulate the nature of the cellulose product formed.

INTRODUCTION

In 1838, the French chemist Anselme Payen (1) described a fibrous component of plant tissues that resisted extraction with organic and aqueous solvents. Payen also determined the molecular formula of this material (C₆O₅H₁₀) and noted its similarity to that of starch. The material was first described as cellulose in 1839 in a report on Payen’s work (2). It was not until almost a century later that the polymeric structure of cellulose was determined, primarily through the research of Hermann Staudinger (3), who noticed that cellulose consists of covalently linked glucose units. Importantly, the bonds holding the cellulose polymer together are remarkably stable. Measurements of uncatalyzed hydrolysis reactions for methylglucopyranosides revealed a surprisingly low spontaneous hydrolysis rate at ambient temperatures with an estimated half-life of several million years—approximately two and four orders of magnitude longer than that of DNA and polypeptides, respectively (4).

We now know that cellulose is the major component of plant cell walls and is biosynthesized on a scale of several billion tons annually, thereby representing the most abundant biological polymer on Earth (5–7). Glucose, the monomeric constituent of cellulose, is a product of photosynthesis (8); thus, cellulose also accounts for a vast pool of chemically fixed carbon dioxide (9).

In the form of wood, cotton, and other plant fibers, cellulosic materials have been indispensable throughout human history as energy sources, building materials, and clothing (2). Modern applications of cellulosic materials include specialized filtration devices (10), additives in paints and coatings (2), wound treatments (11), food additives (12), and renewable energy (5).

Cellulose biosynthesis is by no means limited to higher plants. Indeed, it has been confirmed in almost all kingdoms of life, including bacteria (13), protists (14), algae (15), plants (16), and even animals (17). Having evolved almost 3.5 billion years ago, cyanobacteria might have been the first organisms to produce cellulose and the last common ancestor of cellulose biosynthesis genes (18).

As far as we know, cellulose is always secreted outside the cell to the extracellular matrix or cell wall. Plant cell walls contain cellulose chains organized into cable-like, paracrystalline structures (cellulose micro- and macrofibrils) that are embedded in a rich matrix of diverse polysaccharides and glycoproteins (reviewed in detail in References 5 and 19). Some bacteria, in particular Gluconacetobacter species, also produce ordered cellulose microfibrils and have been used as a valuable model system for several decades to study cellulose microfibril formation (summarized in Reference 20). However, most cellulose-producing bacteria likely produce amorphous aggregates of cellulose as an integral biofilm component (21).

Biofilms are sessile bacterial communities stabilized by an extracellular matrix of polysaccharides, proteins, and nucleic acids. Due to their ability to adhere to surfaces, as well as their reduced susceptibility to antimicrobial treatments, biofilm bacteria are responsible for numerous nosocomial infections (22). Thus, a detailed understanding of how biofilms form and how bacteria synthesize and secrete extracellular polysaccharides is urgently needed (23).

We review novel mechanistic insights into the evolutionarily conserved cellulose biosynthetic machinery, based on recent advances in structural and molecular biology of bacterial cellulose biosynthesis. We focus on how individual cellulose polymers are assembled from monomeric glucose units and secreted across the plasma membrane. We discuss regulatory mechanisms controlling bacterial cellulose biosynthesis and describe similarities and differences between pro- and eukaryotic cellulose synthases.

STRUCTURAL BASIS FOR CELLULOSE BIOSYNTHESIS

Structure of a Cellulose Polymer

Cellulose is a linear polymer of glucose molecules in which individual glucose units are connected via acetal linkages between the C1 and C4 carbons of the glucopyranose rings (Figure 1) (24). The anomeric C1 carbon is in the β-configuration, and every glucose unit is rotated by ~180° with respect to its neighbors, thereby forming a cellobiose disaccharide repeating unit. This internal symmetry, first described by Gardner & Blackwell (25) on the basis of fiber diffraction studies of algal cellulose, represents a twofold screw symmetry that is characteristic for β-1,4-glucans, in contrast to the water-soluble helical structures that α-1,4-glucans can adopt (26). Within the linear polymer, each glucose unit forms two hydrogen bonds with each of its neighbors. The C3 and C6 hydroxyls of each glucose unit donate a hydrogen bond to the ring oxygen and the C2 hydroxyl, respectively, of the glucose unit attached at its C4 carbon. Likewise, the ring oxygen and C2 hydroxyl of each glucose unit accept a hydrogen bond from the C3 and C6 hydroxyls of the glucose unit attached to its C1 carbon (Figure 1) (24). This interaction network stabilizes a coplanar orientation of the individual glucopyranose rings, resulting in a ribbon-shaped polymer with the equatorial hydroxyl groups forming its edges. One end of the cellulose polymer carries an unmodified C4-hydroxyl group and the opposite end a free C1-hydroxyl group; the latter (reducing) end is a convenient site for chemical modifications (27).

Figure 1.

Cellulose is a linear, ribbon-shaped polymer of glucose molecules. The individual glucose units are connected via glycosidic bonds between their C1 and C4 positions, and the anomeric C1 carbon adopts the β-configuration. Each glucose unit is rotated by ~180° relative to its neighbors and forms two hydrogen bonds with each adjacent unit (dashed black lines). The unmodified C1- and C4-hydroxyl groups of cellulose form the polymer’s reducing and nonreducing ends, respectively. Cellulose is elongated at its nonreducing end via a nucleophilic substitution reaction that transfers a glucose unit from UDP-glucose to the polymer’s C4-hydroxyl group. The reaction is facilitated by deprotonation of the acceptor hydroxyl (its hydrogen atom is labeled H) by a general base (labeled B). For clarity, gray spheres indicating hydrogen atoms are shown only in the right panel.

Cellulose polymers can reach an astonishing length; some studies (primarily on plant materials) report polymers of up to 15,000 glucose units (6). Because each glucose unit contains both equatorial hydroxyl groups that point radially away from the face of the pyranose ring and axial hydrogen atoms that are perpendicular to the face of the pyranose ring, cellulose has distinct hydrophilic and hydrophobic characters (Figure 1). The amphipathic macromolecule self-associates and becomes insoluble in water past a degree of polymerization of approximately six to eight (28). Aggregation is driven by van der Waals interactions between the glucopyranose rings (29), a property that has been exploited by many organisms to assemble cellulose polymers into supramolecular structures (30).

Cellulose Is Synthesized by Cellulose Synthase Complexes

Cellulose is produced by membrane-integrated cellulose synthase (CeS) complexes. Although the composition of these assemblies varies considerably by kingdom, all manifestations share a conserved catalytic subunit, termed BcsA (bacterial cellulose synthase subunit A) in prokaryotes and CesA in eukaryotes (hereafter, CeS refers to both pro- and eukaryotic enzymes).

CeS belongs to family 2 of glycosyltransferases (GTs) (31) and processively synthesizes a linear β-1,4-glucan from cytosolic UDP-activated glucose (UDP-Glc, the donor sugar) (32, 33). A processive polymerization mechanism has been inferred because the enzyme remains tightly bound to its polymer product (34). This binding enabled the purification of CeSs and related chitin synthases by the product entrapment method (35), in which detergent-solubilized membrane extracts are incubated with substrate, leading to the formation and precipitation of water-insoluble products together with the biosynthetic machinery. In addition to polymer synthesis, the enzyme translocates the polysaccharide across the cell membrane through a channel formed by its membrane-embedded domain (36). CeSs were first visualized in freeze-fracture images of algal membranes, in which regular arrays of protein “terminal complexes” (TCs) were observed at the membrane-anchored end of growing cellulose fibrils (37). Later, cellulose synthesis was achieved in vitro with cell-free extracts from the bacterium Acetobacter xylinum (recently reclassified as Gluconacetobacter xylinus) (38, 39), and the catalytic subunit was purified and sequenced (40). Homologous plant synthases were later identified in cotton fiber extracts by sequence similarity (16). To date, ceS genes have been discovered in almost all kingdoms of life (13–15, 17, 19, 41).

Although sequence-based predictions suggested that CeSs from disparate phylogenies may vary in topology and domain organization, the best-characterized enzymes from bacteria, algae, and plants likely share similar folds consisting of a core of at least six transmembrane (TM) helices. Two of these helices are N-terminal and four are C-terminal to the intervening cytosolic GT domain (Figure 2) (recently reviewed in Reference 42). The GT domain includes a conserved motif consisting of three variably spaced aspartates (D,D,D, designating DDG, DxD, and TED motifs) followed by a QxxRW pentapeptide (43, 44), which is conserved among all putatively processive GTs, including hyaluronan, chitin, and alginate synthases (44–47).

Figure 2.

Cellulose synthases (CeSs) from different kingdoms share similar features for cellulose synthesis and translocation. Membrane topologies were predicted for the indicated enzymes by the TOPCONS program suite (143), and the consensus predictions are shown as gray cartoons. Pc stands for Phytophthora capsici CesA3 (UniProt entry H6U2P7); Md, Micrasterias denticulata (gi|293413208|); At, Arabidopsis thaliana CesA3 (UniProt entry Q941L0); Cs, Ciona savignyi (UniProt entry Q6RCS2); Gh, Gossypium hirsutum CesA1 (UniProt entry I1T421); and Rs, Rhodobacter sphaeroides (UniProt entry Q3J125). The putative gating loop of each sequence is indicated by a wavy line and is labeled for Rs. The transmembrane (TM) helices contacting the cellulose polymer in Rs BcsA (bacterial cellulose synthase subunit A) are colored brown. In all enzymes, the TM helices frame an intracellular glycosyltransferase (GT) domain, whose architecture is represented by the Rs BcsA structure (Protein Data Bank entry 4P00) (center). Cellulose (cyan) and a UDP molecule (violet) are shown as sticks. Mg²⁺ is shown as a yellow sphere. Conserved motifs are labeled with Roman numerals and shown as consensus sequences obtained for the listed enzymes. For each motif, the Cαatom for the underlined residue is shown as a blue sphere. The insertions of the plant-conserved region (P-CR) and class-specific region (CSR), primarily found in plants, are represented as orange triangles.

Mechanism of Glycosyl Transfer and the Direction of Chain Elongation

CeSs are known as inverting GTs because they invert the configuration of the donor sugar’s anomeric C1 carbon from α when it is attached to UDP to β when it is attached to the cellulose polymer. The catalytic mechanism of inverting GTs has been studied in great detail for non-processive enzymes—structurally through a wealth of crystallographic snapshots, functionally by kinetic and mutagenesis studies, and in silico by simulation analyses (48, 49). Overall, the catalytic mechanism has been described as a classical S_N2-like substitution reaction, in which a nucleophile, namely the polymer’s nonreducing C4-hydroxyl group (acceptor), attacks the anomeric C1 carbon of the donor glucose (Figure 1). This reaction is facilitated by simultaneous deprotonation of the nucleophile and stabilization of the substrate’s pyrophosphate group via a divalent metal cation, usually Mg²⁺ or Mn²⁺. As mentioned above, this mechanism inverts the configuration of the donor sugar’s anomeric carbon from α to β and generates UDP as a leaving group (32, 33).

The cellulose polymer could conceivably be extended at the C1 or the C4 position at its reducing or nonreducing end (Figure 1), respectively. To determine the direction of chain elongation, Koyama and colleagues (50) combined electron microscopy analyses with silver-enhanced staining methods on cellulose fibrils emerging from intact Acetobacter aceti cells to visualize the polymer’s reducing end. These studies revealed that the polymer’s reducing end points away from the bacterium, suggesting that its nonreducing end is attached to the biosynthetic machinery for elongation. This model was further corroborated by Lai-Kee-Him and colleagues (51), who used CesA complexes (CSCs) extracted from blackberry membranes and visualized the reducing ends by gold labeling. Recent crystallographic analyses of a bacterial CeS containing a nascent cellulose polymer (36, 52) provided additional support for a nonreducing-end elongation mechanism (see the section titled Cellulose Translocation and the Membrane Channel, below).

Architecture of Cellulose Synthase and the Active Site

The recently determined crystal structures of CeS from Rhodobacter sphaeroides, containing the catalytic BcsA and a periplasmic BcsB subunit (referred to as the BcsA–B complex), provided the first insights into CeS’s architecture at intermediate states during cellulose synthesis and translocation (36, 52). BcsA forms eight TM helices flanked by three amphipathic helices that run parallel to the membrane at the cytosolic water–lipid interface (Figure 3). The TM helices, in particular TM3–8, form a cellulose-conducting channel across the membrane that originates right above BcsA’s cytosolic active site and ends at the periplasmic BcsA–B interface (Figures 2 and 3) (36). The catalytic GT domain, inserted between BcsA’s TM4 and TM5, forms a classical GT-A domain that is shared by many other GTs, including several nonprocessive enzymes (Figure 2) (48).

Figure 3.

The Rhodobacter sphaeroides bacterial cellulose synthase subunits A and B (BcsA and BcsB) form an inner membrane protein complex. The intracellular domain of the BcsA–B complex consists of a glycosyltransferase (GT) domain ( green) inserted between BcsA’s transmembrane (TM) helices 4 and 5, as well as a regulatory C-terminal PilZ domain (red ) that binds the allosteric activator cyclic di-GMP (c-di-GMP). The TM region is formed from BcsA’s eight TM helices (brown and yellow) and BcsB’s membrane anchor ( purple cylinder). TM3–8 of BcsA (brown) primarily contribute to forming the cellulose-conducting channel. BcsB’s periplasmic domain contains two copies of a repeating unit containing a carbohydrate-binding domain (CBD) (blue) linked to a flavodoxin-like domain (FD) ( gray). The translocating glucan is shown as cyan and red spheres, and UDP at the active site and c-di-GMP at the PilZ domain are shown in spheres colored according to elements. The putative membrane area is shaded gray. Coordinates were obtained from Protein Data Bank entry 4P00.

Earlier hydrophobic cluster analyses and sequence alignments contrasting processive and non-processive GTs predicted a two-domain organization of the active site of processive GTs. Domain A, containing the first two aspartic acid residues of the D,D,D motif, was thought to be involved in substrate binding, whereas domain B, harboring the final aspartic acid residue and the QxxRW sequence, was predicted to be required for processivity and binding of the polymer’s acceptor terminus (43, 53). The BcsA structure revealed that domains A and B are combined into a single GT-A fold, where they indeed mediate substrate and acceptor binding, respectively.

BcsA’s GT-A fold contains an N-terminal, four-stranded, parallel β-sheet that is surrounded by four α-helices and forms the binding site for UDP-Glc (Figure 2). The β-sheet is extended by another three β-strands to one side that, together with an additional three α-helices, mediate acceptor binding. Thus, BcsA’s GT domain is best described as a seven-stranded, mixed β-sheet surrounded by seven α-helices. The organization of the GT domain was recently reproduced independently of the BcsA crystal structure in a theoretical model of the cotton CesA1 GT domain, utilizing a large number of nonprocessive GTs as templates (54).

Sequence Motifs Implicated in Catalysis

The donor- and acceptor-binding sites in CeS are formed by characteristic sequence motifs. The uracil moiety of the substrate fits into a pocket formed by the DDG, HxKAG, and FxVTxK motifs (Figure 2). An adjacent DxD motif coordinates a Mg²⁺ or Mn²⁺ ion required for catalysis; the ion also interacts with the β-phosphate of UDP and likely additional water molecules unresolved in the crystal structures due to limited resolution. The pyrophosphate is further coordinated by the glutamine and arginine residues of the QxxRW motif as well as the threonine and lysine residues from the FxVTxK motif. These interactions probably stabilize the leaving group during glycosyl transfer (Figure 4) (32, 52).

Figure 4.

The cellulose synthase (CeS) donor- and acceptor-binding sites are formed from evolutionarily conserved sequences. Residues of Rhodobacter sphaeroides (Rs) bacterial cellulose synthase subunit A (BcsA) contacting either the UDP molecule or the acceptor glucose (sticks and spheres) are labeled with Roman numerals corresponding to the consensus sequences in Figure 2. Mg²⁺ coordinated by the DxD motif is shown as a yellow sphere. Asp343 of the TED motif hydrogen-bonds with the acceptor’s C4-hydroxyl group and likely functions as a general base during catalysis. The likely donor glucose–binding pocket is indicated by a dashed ellipsoid. The coordinates were derived from Protein Data Bank entry 4P00.

Although a UDP-Glc-bound structure of BcsA has not yet been obtained, the position of the donor glucose can be inferred from other donor sugar–bound GTs, such as a catalytically inactive galactosyltransferase bound to UDP-galactose (55, 56). These comparisons position the donor glucose in a conserved pocket formed by the HxKAG and TED signature motifs (Figure 4). The TED motif, invariant among CeSs but sometimes replaced with a DD motif in other processive family 2 GTs, forms hydrogen bonds with the acceptor glucose through the threonine and aspartic acid residues (52) and may also contact the donor glucose through the glutamate residue (55, 56). Due to its close proximity to the acceptor’s C4-hydroxyl group, the TED motif ’s aspartic acid residue likely functions as the general base during catalysis (Figure 1) (36).

Structures of BcsA–B both with and without a bound UDP nucleotide revealed yet another mechanistic feature that is generally not well appreciated (52). The FxVTxK motif implicated in substrate nucleotide binding (Figure 4) is part of a flexible loop, termed a gating loop, that runs across the cytosolic opening of the GT domain. In a nucleotide-dependent manner, this loop inserts into and retracts from the active site, thereby either mediating substrate coordination at or allowing product diffusion from the active site (52). Interestingly, a threonine-to-isoleucine mutation within the FxVTxK motif in Arabidopsis thaliana CesA3 (AtCesA3) confers resistance to the herbicide isoxaben (57). Although the exact inhibitory mechanism of isoxaben in plants is unknown, the increased resistance toward the herbicide in this mutant underlines the functional importance of this motif.

Similarly, Ramakrishnan and colleagues (55, 58) have identified a mechanism involving loop insertion and retraction from the active site through several crystal structures of donor- and/or acceptor-bound β-1,4-galactosyltransferases. In this case, the inserting loop provides crucial interactions with the donor as well as the acceptor sugars.

Cellulose Translocation and the Membrane Channel

A processive polymer extension requires CeS to move the elongated cellulose polymer by one glucose unit after each round of catalysis, such that the newly added glucose unit is positioned to become the acceptor in a subsequent reaction. The mechanism by which cellulose is translocated is currently unknown. Importantly, in vitro synthesis and translocation studies with purified cellulose and hyaluronan synthases, either reconstituted into lipid bilayer discs, termed nanodiscs (59), or in detergent-solubilized states, demonstrate that the enzymes efficiently couple polymer translocation with synthesis, independently of additional contributions from TM gradients or other subunits (32, 39, 51, 60, 61). Therefore, translocation may be coupled to conformational changes at the active site, relaxation of the newly added sugar molecule into the plane of the cellulose polymer, or binding of another substrate molecule (36, 52).

A common mode of carbohydrate–protein interaction occurs through van der Waals contacts between the faces of the sugar rings and aromatic amino acids (Figure 5) (62). These CH–π stacking interactions (63) likely recognize the secondary structure of polysaccharides, whereas hydrophilic interactions with the sugar’s hydroxyls confer specificity (64–67).

Figure 5.

Cellulose synthase (CeS) forms a cellulose-conducting transmembrane channel. The membrane channel of bacterial cellulose synthase subunit A (BcsA) is lined with aromatic residues (brown sticks) that form CH–π stacking interactions with alternating faces of the cellulose polymer. Additionally, throughout the pore, the polymer’s equatorial hydroxyl groups form hydrogen bonds with hydrophilic BcsA residues ( gray sticks or yellow surfaces). One of two possible registers for the cellulose polymer is shown, as represented in Protein Data Bank entry 4P00.

Quite often, protein–cellulose interactions must be dynamic, allowing either the protein to slide along rigid cellulose fibrils (as observed for a large family of cellulases) or isolated glucan chains to snake through pores or narrow clefts (as observed in, for example, CeS and some cellobiohydrolases) (36, 67). Meyer & Schulz (68) performed a detailed study of the contributions of hydrophobic and hydrophilic interactions to maltodextrin translocation through an outer membrane maltoporin. The authors separated the hydrophilic and hydrophobic energy contributions to translocation in a statistical analysis. These studies revealed that the hydrogen-bonding profile is offset relative to the CH–π stacking profile; thus, both contributions combine into a rather smooth net energy landscape (68). This finding suggests that polar interactions can compensate for the energetic costs of moving polysaccharides between aromatic residues, and vice versa. Other researchers have drawn similar conclusions based on crystallographic analyses of Cel6A cellobiohydrolase in complex with nonhydrolyzable oligosaccharides (67), as well as NMR studies of carbohydrate-binding modules (66, 69–72).

The glucan channel formed by BcsA likely also minimizes the energetic costs of cellulose translocation. The channel is continuous with the active site, and its dimensions resemble a thick ribbon. Ten glucose units span the entire pore and are coordinated by CH–π stacking interactions with aromatic residues (36), some of which are highly conserved from bacteria to vascular plants (Figure 5) (reviewed in Reference 42). While the CH–π interactions alternate between the polymer’s faces, its hydroxyl groups are coordinated through a continuous pattern of hydrogen bonds throughout the channel (36).

The BcsA structure suggests a plausible mechanism of how CeSs and other processive GTs, such as hyaluronan, chitin, alginate, and poly N-acetylglucosamine (PNAG) synthases, couple polymer extension with translocation (43). The tryptophan residue of the characteristic QxxRW motif is located right above the active site and forms part of the entrance to BcsA’s TM channel. In this position, it coordinates the terminal glucose unit of the nascent polymer, such that the acceptor’s C4-hydroxyl group is within hydrogen-bond distance of the general base, likely provided by the TED motif (52). Acceptor dissociation after catalysis is prevented because the growing polymer is coordinated inside the TM channel, thereby enabling its stepwise elongation.

Processive GTs are unlikely to “bind” the growing polysaccharide tightly. Instead, the enzymes must prevent backsliding of the translocated polymer into the cytosol as well as its premature release to the extracellular space. In BcsA, the acceptor-binding site at the entrance to the TM channel may serve this purpose. It represents the only section of the TM channel that tightly interacts with the glucan (i.e., its terminal glucose molecule), in that highly conserved residues hydrogen-bond with every hydroxyl group of the acceptor (52).

Cellulose polymers can be thousands of glucose units long (73) and may span a distance of several micrometers in an extended conformation. The mechanism controlling cellulose length is currently unknown. It is conceivable that high–molecular weight glucans disengage from the biosynthetic machinery due to shearing forces or that the enzymes stall due to increasing polymer interactions in the extracellular space.

LONG-STANDING QUESTIONS CONCERNING THE MECHANISM OF CELLULOSE BIOSYNTHESIS

Primer Hypotheses

Biochemical analyses of a diverse set of GTs suggest that many enzymes initiate synthesis by elongating oligosaccharide primers (74–78). Thus, it is possible that CeS also requires a primer to form the initial acceptor for cellulose biosynthesis; however, neither the strict dependence on nor the identity of a primer has been demonstrated in a purified system to date. In 1995, Matthysse et al. (79) performed cellulose biosynthesis studies with Agrobacterium tumefaciens, which described mutations in genes of the bcs operon that led to the incorporation of ¹⁴C-labeled glucose into a membrane-associated compound. These observations inspired a model by which CeS would polymerize preformed lipid-linked oligosaccharides to a high–molecular weight polymer. Subsequently, in 2002 the Delmer laboratory (80) reported the use of sitosterol-β-glucoside as a primer for plant cellulose biosynthesis. In these experiments, yeast membranes containing recombinantly expressed cotton CesA1 extended synthetic, radioactively labeled sitosterol-β-glucoside in vitro in the presence of UDP-Glc. Furthermore, these CesA1-expressing yeast membranes were able to polymerize cellulose from radioactively labeled UDP-Glc only when cotton fiber membrane extracts were added to the reaction, suggesting dependence on a plant-specific membrane component (80).

Structural analyses of bacterial BcsA–B suggest yet another possibility. The coordination of the acceptor glucose unit by the QxxRW motif and surrounding polar residues strongly resembles the coordination of a single galactose molecule in the sodium-dependent sugar transporter SGLT (52, 81). Thus, BcsA might be able to bind a monomeric glucose molecule at the acceptor site, thereby providing the initial acceptor for glycosyl transfer. Indeed, cellulose biosynthetic activity of detergent-solubilized plant and bacterial enzymes in the absence of supplemented lipids further argues against a direct involvement of lipid-linked reaction intermediates (32, 39, 51, 61, 82).

One-Site Versus Two-Site Model

The twofold screw symmetry of cellulose (Figure 1) inspired several different models of its biosynthesis mechanism. Some models explained the internal symmetry with a twofold symmetric substrate-binding site, formed by either one enzyme or a CeS dimer. To this end, CeS would bind two UDP-Glc molecules with opposite orientations, as well as an acceptor glucan, such that glycosyl transfer would connect the incoming two glucose units with the acceptor glucan, thereby elongating the polymer by a cellobiosyl unit (83). In contrast, Delmer (7) proposed a mechanism by which the polymer is extended by a single glucosyl unit, assuming that each newly added glucose unit freely rotates around the acetal linkage to adopt the characteristic twofold screw symmetry of cellulose. The BcsA–B structure tipped this discussion toward the latter model. All of the conserved CeS signature motifs form a single, deeply buried binding site for UDP-Glc (Figure 4), suggesting a stepwise polymer extension that occurs one glucosyl unit at a time. Thus, the donor glucose is always bound in the same orientation relative to the acceptor glucan, requiring it to rotate around the newly formed linkage during or after catalysis to align with the polymer prior to translocation (36). Indeed, the first crystal structure of BcsA–B, which represents a cellulose biosynthesis state after glycosyl transfer but before translocation, showed a weakly coordinated terminal glucose unit extending from the acceptor-binding site into the active-site pocket, where it is free to rotate around the glycosidic linkage (36).

In addition, with a single substrate-binding site, the alternating orientation of glucose units in cellulose requires that the polymer’s terminal glucose unit function as the acceptor in two distinct orientations. Indeed, the coordination of the terminal glucose by the QxxRW motif and surrounding residues likely positions its C4-hydroxyl group in close proximity to the donor glucose and the catalytic base in either orientation. Further detailed experimental and theoretical studies are required to corroborate this model and to elucidate how relaxation of the newly added glucose unit affects polymer translocation.

FACTORS REQUIRED FOR BACTERIAL CELLULOSE SYNTHASE ACTIVITY

BcsB, a Periplasmic Subunit

Genetic studies in Gluconacetobacter xylinus and Agrobacterium tumefaciens, as well as biochemical analyses with purified components from Rhodobacter sphaeroides and Escherichia coli, demonstrated that bacterial cellulose biosynthesis depends on the periplasmic noncatalytic BcsB subunit (32, 79, 84, 85). The bcsB gene is found in all bcs operons identified to date (13). BcsB is anchored to the inner membrane via a C-terminal TM helix that, together with a preceding periplasmic helix, tightly interacts with the catalytic BcsA subunit (Figure 3) (36). The dome-shaped protein extends ~60 Å into the periplasm and forms two copies of a repeating unit of a carbohydrate-binding domain (CBD) that is C-terminally fused to an α/β domain resembling a flavodoxin-like domain (FD) (36). The CBD is structurally related to carbohydrate-binding modules (CBMs) (86, 87), which form classical β-strand-rich jelly-roll motifs and interact with carbohydrates either via the β-sheet surface or with the edges of the jelly roll (31, 88). So far, no direct interaction between BcsB’s CBDs and cellulose has been demonstrated experimentally, yet the potential implication of such an interaction in guiding the polymer across the periplasm is intriguing.

In vitro cellulose biosynthesis studies with N-terminally truncated BcsB constructs revealed that only its C-terminal TM anchor, together with the preceding amphipathic helix, is required for BcsA activity (32). This functional dependence of BcsA on BcsB is further reflected in some Gluconacetobacter strains containing apparently fused bcsA and bcsB genes (20, 85, 89). Interestingly, purification of functional CeSs from these strains in different laboratories did not yield a polypeptide corresponding to a fused bcsA–bcsB gene product, but instead two peptides of ~83 and 93 kDa were isolated, suggesting posttranslational processing into the bona fide BcsA and BcsB proteins (85, 89).

The functional dependence on noncatalytic, membrane-anchored periplasmic subunits is a common theme among bacterial exopolysaccharide synthases, including alginate and PNAG synthases (90). For example, Alg8 and Alg44 are the inner membrane components of the bacterial alginate synthase; Alg8 is the catalytically active subunit similar to BcsA. Alg44, just like BcsB, contains a single TM region followed by an extended, although somewhat shorter, periplasmic domain (47). In vivo studies in Pseudomonas aeruginosa, which produces alginate as a biofilm component, revealed the strict dependence of Alg8’s catalytic activity on Alg44 (91). Similarly, PNAG biosynthesis in Escherichia coli requires the membrane-embedded GT PgaC, together with an associated PgaD subunit that spans the membrane twice (92). In contrast to BcsB, however, the noncatalytic Alg44 and PgaD proteins are also important for the allosteric activation of the synthases by cyclic di-GMP (c-di-GMP; discussed in the following section), which complicates the interpretation of deletion mutants. The commonalities and differences between bacterial exopolysaccharide synthases were recently reviewed in detail by Whitney & Howell (90).

Cyclic di-GMP, a Bacterial Signaling Molecule

It has long been appreciated that bacterial cellulose biosynthesis requires a small-molecule activator. This activator was initially thought to be GTP (39, 82); however, in a later seminal study, Ross et al. (93) identified c-di-GMP, which is enzymatically formed from two GTP molecules, as a potent activator of cellulose biosynthesis. The discovery and characterization of c-di-GMP opened the door to the first purification of BcsA and the identification of the UDP-Glc-binding subunit by product entrapment, in which in vitro–synthesized, water-insoluble cellulose is sedimented along with its biosynthetic machinery (34, 94).

Since its discovery as the BcsA activator, c-di-GMP has been shown to be involved in numerous bacterial processes, including cell motility, biofilm formation, and cell division (see Reference 95 for an up-to-date review). However, the mechanism by which c-di-GMP activates BcsA had remained a mystery until recently. Initially, BcsB was thought to be the c-di-GMP-binding subunit (96), but subsequent bioinformatics and functional analyses demonstrated that the catalytic BcsA subunit contains a C-terminal cytosolic domain, termed PilZ, that is responsible for c-di-GMP binding (97, 98).

PilZ domains, originally identified as a bacterial pilus component (97), bind c-di-GMP via RxxxR and DxSxxG motifs (99); the first motif is located within a flexible loop and the second on the surface of a β-sheet or β-barrel (99, 100). Fujiwara et al. (101) determined the crystal structure of a lone BcsA PilZ domain in the absence of c-di-GMP and performed mutagenesis experiments to determine how PilZ’s signature motifs affect the affinity and stoichiometry with which it binds c-di-GMP. Subsequently, Omadjela et al. (32) reconstituted the purified BcsA–B complex into nanodiscs, determined the enzyme’s affinity for c-di-GMP at ~2 μM, and found that increasing c-di-GMP concentrations do not affect BcsA’s Michaelis constant (K_m) for UDP-Glc. Instead, titration of c-di-GMP seemed to increase the fraction of catalytically active enzymes. These results led to a mechanistic model in which BcsA’s active site is blocked and inaccessible to UDP-Glc in the absence of c-di-GMP. Allosteric activation through c-di-GMP binding was thus proposed to remove this block and expose the active site, thereby allowing BcsA to function at its maximum catalytic rate.

This model was recently confirmed by the crystal structure of c-di-GMP-activated BcsA–B (Figure 3) (52). In the absence of c-di-GMP, BcsA’s active site is blocked by the so-called gating loop, which contains the conserved FxVTxK motif discussed above. The gating loop is tethered in this active site–occluding position by an interaction between the first arginine residue of the PilZ domain’s RxxxR motif and the backbone of the gating loop. In the presence of c-di-GMP, however, this arginine residue rotates away from the active site to coordinate c-di-GMP, thereby releasing the gating loop to move to a position clear of the active site, allowing UDP-Glc binding. Mutagenesis studies replacing the first arginine residue of the PilZ motif with an alanine residue generated a constitutively active enzyme, thereby confirming that BcsA is indeed autoinhibited by this interaction in the absence of c-di-GMP (52). Interestingly, Matthysse et al. (79) observed in 1995 that the CeS from the C58 strain of Agrobacterium tumefaciens showed no increase in catalytic activity in the presence of c-di-GMP. The same group also published the sequence of this BcsA protein, which showed that the first arginine residue of the RxxxR motif was replaced with serine (84), further validating the mutagenesis studies reported for Rhodobacter sphaeroides BcsA (52).

The cytosolic concentration of c-di-GMP is controlled by the finely tuned activities of a large family of diguanylate cyclases and phosphodiesterases (95). Detailed studies by the Römling and Hengge laboratories delineated a signaling cascade that leads to elevated c-di-GMP levels and activation of cellulose biosynthesis in Enterobacteriaceae (102–104). Entering the stationary phase, the general stress-response master regulator σ^s is upregulated, which activates expression of the MerR-like (105) transcriptional activator MlrA. MlrA, in turn, is a positive transcriptional regulator of CsgD (curli subunit gene D, homologous to AgfD in Salmonella typhimurium), a transcriptional regulator controlling cellulose biosynthesis as well as the assembly of amyloid-like proteinaceous fibers, termed curli (106, 102). Increasing CsgD activity leads to the expression of YaiC (known as AdrA in S. typhimurium) (107, 108), which is a membrane-integrated diguanylate cyclase containing an N-terminal membrane-associated sensor (MASE) and a C-terminal cytosolic diguanylate cyclase domain (Figure 6) (109). Expression of YaiC/AdrA correlates with strong activation of cellulose biosynthesis, perhaps due to the localized production of c-di-GMP (107). Recent studies by Hufnagel et al. (110) identified a CsgD-independent cellulose biosynthesis pathway in Escherichia coli. This pathway appears to be coupled to the disulfide-bonding (DSB) machinery and leads to the expression of YfiN, another membrane-anchored diguanylate cyclase.

Figure 6.

The bacterial cellulose synthase (Bcs) contains multiple subunits at the cell envelope. The inner membrane (IM)-integrated BcsA–B complex may interact with the periplasmic domain of BcsC, thereby forming a transenvelope conduit for the nascent cellulose polymer (orange circles). The periplasmic cellulase BcsZ may cleave the translocating glucan stochastically or degrade mislocalized cellulose in the periplasm. C-di-GMP binds to the C-terminal PilZ domain of BcsA and is formed by the membrane-integrated diguanylate cyclase AdrA. The BcsD subunit is found primarily in cellulose microfibril–forming bacteria, and its biological function, as well as its cellular localization, is unclear. Abbreviations: c-di-GMP, cyclic-di-GMP; GT, glycosyltransferase; OM, outer membrane.

Regulation by c-di-GMP is a common feature of bacterial extracellular polysaccharide synthases involved in biofilm formation (90). However, the mechanism by which c-di-GMP activates BcsA contrasts sharply with that of PNAG synthase. In the latter case, c-di-GMP stabilizes a complex between the catalytic PgaC and the regulatory PgaD subunits, neither of which contains a PilZ domain (92). Further research is required to characterize the mechanisms by which c-di-GMP activates these promising antimicrobial targets.

SUPRAMOLECULAR ASSEMBLY OF BACTERIAL CELLULOSE SYNTHASE

BcsC, a Putative Outer Membrane Pore

The genes necessary for efficient cellulose synthesis in gram-negative bacteria are encoded by the bcsABZC operon, of which BcsA and BcsB form the catalytic core at the inner membrane (Figure 6) (13, 32, 36). Although only BcsA, BcsB, and c-di-GMP are necessary for robust cellulose synthesis in vitro (32), genetic analyses in Gluconacetobacter xylinus and Agrobacterium tumefaciens demonstrated that the BcsZ and BcsC subunits, as well as several genes for c-di-GMP metabolism, are necessary for maximal cellulose synthesis and secretion in vivo (79, 84, 85, 107, 111, 112).

BcsC is a large protein, consisting of ~1,100 residues, that carries an N-terminal cleavable signal sequence for translocation to the periplasm. Its C-terminal 300 residues are predicted to form an 18-stranded β-barrel in the outer membrane, which is preceded by a large N-terminal domain. This periplasmic region includes tetra-tricopeptide (TPR)-like repeats, consisting of 34-residue tandem repeats that adapt helix-turn-helix tertiary structures and are frequently involved in mediating protein–protein interactions (113–115).

BcsC is similar to components of the bacterial alginate synthase; however, in the alginate synthase complex, the corresponding periplasmic N-terminal and outer membrane–integrated C-terminal domains represent individual subunits, AlgK and AlgE, respectively. AlgK forms a right-handed α-helical superhelix containing 10 TPR-like repeats, onto which several putative sites for protein–protein interactions have been mapped (113). Alginate, unlike cellulose, is covalently modified during transit across the periplasm (47, 90, 116); thus, the AlgK scaffold may provide a docking platform for these enzymes.

The structure of the outer membrane component, AlgE, was determined both in a detergent-solubilized state and by the lipidic cubic-phase crystallization method (117, 118). As predicted, the protein forms an 18-stranded β-barrel with a funnel-shaped pore that is sufficient to allow passage of a linear polysaccharide, and halide efflux assays demonstrated that the pore forms an anion-conducting channel, consistent with the polyanionic nature of alginate. The anion conductivity was further enhanced after deleting periplasmic and extracellular loops that fold back onto the pore in the crystal structure (117), inviting speculations that the interaction between AlgK and AlgE opens the outer membrane pore for alginate translocation (118).

We do not know whether the inner membrane CesS components (BcsA and BcsB) directly interact with subunits at the outer membrane (BcsC). BcsB extends ~60 Å into the periplasm, which, combined with an AlgK-like extended structure of BcsC’s N terminus, may suffice to establish a direct interaction. It is also unclear whether the periplasmic domain of BcsC interacts with the translocating polysaccharide and how the polymer enters the outer membrane channel.

BcsZ, a Periplasmic Cellulase

All cellulose-producing bacteria identified to date also contain a gene coding for a β-1,4-glucanase (termed BcsZ or CMCax) that is either part of or closely associated with the bcsA-containing operon (13). BcsZ is a single-domain protein of ~370 residues that belongs to family 8 of the glycosyl hydrolases (31) and is synthesized with an N-terminal cleavable signal sequence for secretion into the periplasm. The gene was first identified in mutant Gluconacetobacter xylinus strains that showed a significantly reduced cellulose biosynthesis level combined with reduced crystallinity of the cellulose microfibrils (111, 112).

Cellulose-degradation assays using acid-swollen and carboxymethyl cellulose revealed a very low hydrolytic activity of BcsZ toward these substrates (119), prompting initial discussions that BcsZ might instead function as a transglycosylase connecting oligosaccharides as they emerge from the BcsA–B complex. However, recent in vitro cellulose biosynthesis assays performed in the presence of BcsZ demonstrated that the enzyme efficiently degrades individual glucan chains (32), yet it fails to degrade crystalline polymers represented by most commercial celluloses. Several crystal structures of free and product-bound BcsZ further established its nature as an inverting glycosyl hydrolase-8 cellulase (119, 120). Despite increasing biochemical data on BcsZ, very little is known about its biological function. We do not know whether BcsZ interacts with other components, such as BcsB or BcsC, or whether it freely diffuses in the periplasm to degrade mislocalized glucans. Some reports on Gluconacetobacter CMCax even suggest that the protein may be secreted to perform its function on the cell surface (112, 120). Interestingly, plant CeSs also associate with hydrolytic enzymes during cellulose biosynthesis (121, 122), perhaps to release fiber torque or cleave misaligned strands that would otherwise impede synthesis. Accordingly, one study recently suggested that BcsZ may reduce the twisting of cellulose microfibrils formed by Gluconacetobacter xylinus (123), yet the observation that BcsZ is part of bcs operons, irrespective of whether the final product is amorphous or fibrillar cellulose, argues for additional, as-yet-unidentified biological functions of this enzyme (13).

Additional Factors Associated with Bacterial Cellulose Biosynthesis

Several other genes that are peripheral to the bcsABZC operon have been implicated in cellulose biosynthesis and biofilm formation in enteric bacteria. Genes of the bcsEFG operon (formerly yhjSTU) are all necessary for cellulose synthesis in S. typhimurium and S. enteritis, and insertional mutations produce the same nonbiofilm phenotype observed upon disruption of the bcsABZC genes. In these mutants, rigid pellicle formation and colony fluorescence on calcofluor plates (a common assay to detect cellulose biosynthesis) were not observed, but could be rescued by complementation in trans (107, 124). Recently, BcsE was further characterized in a proteomic screen for c-di-GMP-binding factors. Subsequent studies in BcsE-disrupted strains revealed that cellulose production in S. typhimurium could be rescued only by BcsE mutants capable of binding c-di-GMP, demonstrating that BcsA’s PilZ domain is not the only sensor through which this biofilm inducer regulates cellulose biosynthesis (125). However, the exact signaling pathway has yet to be determined.

The bcsQ gene (formerly yhjQ) is part of a two-member operon situated between the bcsABZC and bcsEFG operons in several species of Enterobacteriaceae, and its nonpolar deletion in E. coli results in a loss of colony fluorescence on calcofluor plates (126). Interestingly, a recent report by Serra et al. (21) suggested that the “domesticated” E. coli K12 strain does not produce cellulose due to a putative premature stop codon in the bcsQ gene. Removing the stop codon restored a colony morphology indicative of cellulose secretion. BcsQ colocalizes with secreted cellulose polymers at the cell poles of clustered bacteria, a finding that is consistent with the observation that E. coli biofilms form chainlike pole-to-pole aggregates (127). Thus, BcsQ may be involved in localizing the cellulose biosynthetic machinery to the cell poles (126).

FEATURES RELATED TO CELLULOSE MICROFIBRIL–FORMING BACTERIA

Linear Terminal Complexes in Gluconacetobacter

Because of their ability to produce cellulose microfibrils that resemble cellulosic materials isolated from plants, Gluconecatobacter species have long been a model system for cellulose biosynthesis (20). Freeze-fracture electron micrographs of cellulose-producing Gluconacetobacter xylinus visualized these microfibrils emerging from so-called linear TCs in the cell membrane (128, 129). Subsequent advances of this technique in sample preparation have allowed immunogold labeling and localization of the CeS complexes within the membrane-embedded TCs. The CeSs were linearly arranged along the cell axis, suggesting that cellulose microfibril formation is a consequence of CeS supramolecular organization (128). However, this interpretation requires further validation with other cellulose microfibril–forming strains because linearly arranged CeSs have so far been visualized only in Gluconacetobacter.

BcsD, an Accessory Cellulose Synthase Subunit

Bacteria that produce cellulose microfibrils usually contain an additional CeS subunit, known as BcsD or CesD (13, 20), which is a small protein of ~160 residues. The exact biological function of BcsD and its cellular localization are unknown. Fractionation of Gluconacetobacter hansenii combined with immunostaining suggested that BcsD is a periplasmic protein, but the protein does not contain an identifiable signal sequence and appears not to be posttranslationally processed, as would be expected for a translocated protein (20, 130). Although BcsD is not an essential subunit for cellulose biosynthesis, some results indicate significantly reduced in vivo cellulose production and crystallinity in its absence (20, 85).

The crystal structure of BcsD from Gluconacetobacter xylinus revealed its organization as a barrel-shaped tetramer of dimers (131). Interestingly, four cellopentaose molecules bind inside the barrel when the crystals are soaked with the oligosaccharide. These observations gave rise to a yet-to-be-tested model by which BcsD localizes between BcsA–B and BcsC subunits in the periplasm (Figure 6). In this model, the barrel forms four passageways for emerging glucans, thereby guiding them across the periplasm and/or facilitating their alignment into cellulose microfibrils (131).

A direct interaction between BcsD and the membrane-associated BcsA and BcsB subunits has not yet been demonstrated. Sunagawa et al. (133), however, colocalized a so-called cellulose-complementing protein (Ccp) with the Gluconacetobacter BcsD subunit and demonstrated their interaction through pull-down experiments and isothermal titration calorimetry. Tn5 transposon mutagenesis of Gluconacetobacter hansenii by the Kao laboratory (132) later identified a non-cellulose-producing Ccp mutant that could be rescued by genetic complementation to near wild-type levels. In the Ccp mutant, Western blot analyses showed that expression of BcsA was not altered, yet expression of BcsB and BcsC was reduced. Thus, although the exact biological functions of Ccp and BcsD are unclear, the proteins may be required to stabilize or organize BcsA–B complexes in the plasma membrane (133).

A COMPARISON BETWEEN BACTERIAL AND PLANT CELLULOSE SYNTHASES

Catalytic Mechanism

The biochemistry of plant CesAs and their assembly into macromolecular complexes have recently been reviewed (134). Despite the enzymological similarities between cellulose-producing enzymes from different kingdoms of life, significant differences exist with respect to the products formed. Cellulose produced by most eukaryotes represents a paracrystalline assembly of individual glucan chains, namely cellulose micro- and macrofibrils (6). Intriguingly, plants organize CesAs into CSCs (135), which appear in freeze-fracture images as sixfold symmetric rosettes at the ends of growing cellulose microfibrils (136). Thus, an attractive model is that CSCs position CesAs such that the emerging glucan chains spontaneously align into microfibrils, driven primarily by van der Waals interactions (30).

Several point mutations of plant CesA GT domains have been characterized that give rise to profound in vivo phenotypes, mostly due to a reduction in cellulose content and/or crystallinity in the cell wall. Mapping these mutations onto the BcsA structure further emphasizes the similarity between prokaryotic and eukaryotic enzymes. For example, an alanine-to-valine mutation of the HxKAG motif in AtCesA3 reduces cellulose biosynthesis and increases the amount of noncrystalline glucans in the cell wall, leading to widespread morphological abnormalities (137–139). The HxKAG motif in BcsA localizes to one side of the putative donor glucose–binding site (Figure 4), and modifications thereof could affect the affinity and specificity for the substrate. Another CesA point mutation replaces a conserved proline residue with serine or threonine and has a dominant-negative phenotype in Arabidopsis (referred to as fra5 in AtCesA7 and thanatos in AtCesA3) (140, 141). The corresponding proline residue in BcsA positions a following histidine residue such that it hydrogen-bonds with and stabilizes the acceptor glucose for glycosyl transfer. This histidine corresponds to a glutamine in plants, which could perform a similar function. An up-to-date list of known CesA mutations localizing to the GT domain is discussed in detail in References 42 and 54.

Plant-Specific Insertions into the Glycosyltransferase Domain

Plant genomes contain multiple CesA isoforms that are differentially expressed during primary and secondary cell wall formation (134); additionally, the enzymes contain two insertions within the evolutionarily conserved GT domain. One is moderately conserved, termed the plant-conserved region (P-CR), and one is specific for cesA gene subclasses, termed the class-specific region (CSR) (142). Both domains seem necessary for proper in vivo function, yet their precise roles during cellulose biosynthesis are unclear (54, 142). The domains are ~150 residues long and, on the basis of their sequence, currently do not have any homologs of known structure. They are inserted into the GT-A fold between highly conserved β-strands and are predicted to form separate domains that protrude into the cytosol (Figure 2). The Yingling laboratory (54) undertook a de novo structure prediction approach to model both domains as part of a theoretical model of the GT domain from cotton CesA1. Accordingly, both regions form individual α-helical protrusions from the GT domain, where they may interact with CesA-interacting components. Additionally, some models assume that P-CR and CSR mediate CesA oligomerization and thereby affect cellulose microfibril formation (54). However, in the absence of experimental data with, for example, domain-swapped CesA enzymes, these models must await further validation.

Topology

An evolutionarily conserved biosynthesis mechanism suggests that plant and bacterial CeSs have a similar TM architecture. However, inconsistent topology predictions, combined with the lack of experimental data verifying the predictions, have caused significant confusion with regard to the architecture of CesAs (42). Since the first amino acid sequences became available, the enzymes have been predicted to contain eight TM helices, two N-terminal and six C-terminal of the GT domain (16). However, this organization localizes the gating loop, which contains the FxVTxK motif involved in substrate coordination in BcsA (Figure 2), to the extracellular side of the plasma membrane, raising the possibility that the predicted CesA topology is incorrect (42). Improved topology prediction algorithms, such as those implemented in the TOPCONS program suite (143), appear to be superior in discriminating proper TM helices from peripheral amphipathic helices, both of which are common features of BcsA’s architecture (36). Notably, the gating loop in BcsA follows an amphipathic interface helix that could be misinterpreted as a proper TM helix, thereby inverting the downstream topology. The TOPCONS-predicted topologies for Arabidopsis and cotton CesAs are consistent with the BcsA fold (Figure 2). Despite these topological uncertainties, it is widely accepted that all CeSs couple cellulose synthesis with translocation through a pore formed by their membrane-embedded domains (6, 73, 83, 144).

CesA-Associated Factors

Similar to the importance of BcsB for bacterial cellulose biosynthesis, plant CesAs may also require additional components for proper function. This hypothesis is underscored by the notorious difficulties in isolating catalytically active CesAs from plant sources, which may be due to the loss of essential components during solubilization and purification. Several candidates, some of which have unknown biological functions, have been suggested to associate with plant CesAs. Examples include the putatively glycophosphatidylinositol (GPI)-linked protein COBRA (145), the membrane-anchored KOBITO (146) and endo-β-1,4-glucanase KORRIGAN proteins (147), and sucrose synthase (148). Recently, live-cell imaging and split ubiquitin and fluorescence complementation assays suggested that KORRIGAN is indeed an integral component of CSCs in Arabidopsis membranes; however, transient interactions with other components cannot be ruled out at this point (121). Thus, identifying CesA-interacting proteins and analyzing their effects on cellulose biosynthesis will remain important goals toward unraveling the mechanism of cellulose microfibril formation in plants.

CONCLUDING REMARK

Cellulose is found in essentially all kingdoms of life and thus affects humans on multiple levels. Recent advances in cellulose research aid the prevention of biofilm formation, the design of superior materials for biomedical and biotechnological applications, and the use of a vast natural energy resource.

SUMMARY POINTS.

1. Cellulose is an integral component of plant cell walls and many bacterial biofilms.

2. The CeS GT domain adopts a classical GT-A fold, and all characteristic sequence motifs form a single binding site for UDP-Glc.

3. The GT domain tightly interacts with a membrane-embedded region that forms a cellulose-conducting channel.

4. Processive polymer elongation is achieved through the coupling of glycosyl transfer and translocation of the polysaccharide into the TM pore.

5. BcsA requires BcsB to form a catalytically active synthase.

6. c-di-GMP allosterically activates BcsA by releasing an autoinhibited state in which BcsA’s active site is occluded.

7. The functional core of CeS is most likely evolutionarily conserved, containing at least six TM helices, a GT domain, and a flexible gating loop.

FUTURE ISSUES.

1. What is the mechanism of cellulose membrane translocation? How is mechanical force generated and exerted to drive polymer translocation?

2. How are glucan chains guided through the periplasm and across the outer membrane? How do other subunits interact with BcsA–B to facilitate secretion and to generate amorphous or fibrillar cellulose?

3. Does c-di-GMP activate other exopolysaccharide synthases in a manner similar to BcsA? Can mechanistic insights be applied to develop novel antimicrobial drugs?

4. What is the supramolecular organization of CeS in the membrane, how is it achieved, and how does it affect cellulose microfibril formation?

5. How can we reconstitute plant cellulose biosynthesis for an in-depth understanding of cellulose microfibril formation and cell wall synthesis?

Glossary

Cellulose microfibril

a cable-like, paracrystalline structure composed of multiple parallel β-1,4-glucan chains

Glucan

polymer of glucose molecules

Donor

compound that provides the sugar unit for glycosyl transfer

Acceptor

compound that receives the sugar unit during glycosyl transfer

GT-A fold

a glycosyltransferase (GT) domain containing a central β-sheet surrounded by α-helices implicated in donor and acceptor binding

Jelly roll

a protein fold in which an antiparallel β-sheet wraps around to form a barrel-shaped structure

Cyclic di-GMP (c-di-GMP)

a bacterial signaling molecule that is a potent activator of biofilm formation and exopolysaccharide biosynthesis

PilZ

a regulatory protein domain that binds c-di-GMP and affects multiple cellular processes in bacteria

Plant-conserved region (P-CR) and class-specific region (CSR)

regions representing GT domain insertions first identified in plant cellulose synthases