Molecular Analysis of Cellulose Biosynthesis in Arabidopsis

See allHide authors and affiliations

Science  30 Jan 1998:
Vol. 279, Issue 5351, pp. 717-720
DOI: 10.1126/science.279.5351.717


Cellulose, an abundant, crystalline polysaccharide, is central to plant morphogenesis and to many industries. Chemical and ultrastructural analyses together with map-based cloning indicate that the RSW1 locus of Arabidopsis encodes the catalytic subunit of cellulose synthase. The cloned gene complements the rsw1 mutant whose temperature-sensitive allele is changed in one amino acid. The mutant allele causes a specific reduction in cellulose synthesis, accumulation of noncrystalline β-1,4-glucan, disassembly of cellulose synthase, and widespread morphological abnormalities. Microfibril crystallization may require proper assembly of the RSW1 gene product into synthase complexes whereas glucan biosynthesis per se does not.

Cellulose, a crystalline β-1,4-glucan, is the world's most abundant biopolymer. Its biomass makes it a global carbon sink and renewable energy source, and its crystallinity provides mechanical properties central to plant morphogenesis and the fiber industries. The mechanisms that plants use in synthesis have not yielded to biochemistry or cloning by hybridization to genes encoding prokaryotic cellulose synthases (1). By combining chemical and ultrastructural analyses with map-based cloning, we show that the Arabidopsis RSW1 locus encodes a glycosyl transferase that complements the rsw1mutant (2). The temperature-sensitive rsw1 allele disassembles cellulose synthase complexes in the plasma membrane (“rosettes”), alters cellulose crystallinity, and disrupts morphogenesis. The gene product, which is closely related to the putative cellulose synthase catalytic subunit from cotton fibers (3), can therefore be used to manipulate the production and physical properties of cellulose, while the mutant links plant morphogenesis and cellulose production.

Mutants impaired in cellulose production were selected with the use of a radial swelling phenotype (rsw), which mimics responses of wild-type roots to cellulose synthesis inhibitors such as dichlorobenzonitrile. Shoots of rsw1 seedlings grown at the restrictive temperature (31°C) have less cellulose than wild-type seedlings (159 ± 19 versus 363 ± 28 nmol of glucose per milligram of plant dry weight) but more of an ammonium oxalate–extracted glucan (195 versus 58 nmol mg 1), which methylation analysis (Fig.1A) and enzyme digestion show is β-1,4–linked (4). Facilitated extraction and digestion by enzymes and trifluoroacetic acid indicate low crystallinity, the property that makes cellulose resistant to extraction and digestion. Smaller changes in Golgi-synthesized polysaccharides show thatRSW1 is specifically involved in cellulose biosynthesis.

Figure 1

(A) Gas chromatography of alditol acetates of methylated sugars from laminarin (top) and cellulose (middle) standards and from the glucan purified from the ammonium oxalate fraction from shoots of rsw1 grown at 31°C (bottom). Coincident peaks show that the rsw1 glucan is 1,4-linked. (B through E) Roots frozen in nitrogen slush without cryoprotection were freeze-fractured with the use of double replica holders in a Balzers BAF 400T (18). (B) Normal microfibrils in rsw1 (18°C). (C) Rosettes (P face) of the mutant are indistinguishable from the wild type at 18°C, but (D) are rare and sometimes irregular in rsw1 seedlings within 3 hours of transfer to 31°C. (E) Particles in rows (arrows) can curve or cluster (box) under longer (18-hour) exposures. Scale bars, 50 nm.

Rosettes (terminal complexes) are the putative hexameric cellulose synthase complexes of higher plant plasma membranes (5). Freeze-fractured root cells of wild type and mutant grown at 18°C show cellulose microfibrils (Fig. 1B). Rosettes on the P face of the mutant plasma membrane at 18°C (Fig. 1C) resemble those of the wild type, but transferring the mutant to 31°C reduces rosette numbers within 30 min, with extensive loss after 3 hours (Fig. 1D) and a loss of definition to the terminal globules on the E face. Plasma membrane particles tend to align in the mutant after prolonged exposure to the restrictive temperature (Fig. 1E). Cortical microtubules that align cellulose microfibrils and Golgi bodies that synthesize other wall polysaccharides appeared unchanged.

The rsw1 mutation therefore disassembles cellulose synthase complexes, reduces cellulose accumulation, and causes β-1,4-glucan to accumulate in a noncrystalline form. It maps (6) to a region of chromosome 4 (Fig.2A) to which a mapping program had assigned an expressed sequence tag (EST) that, it was deduced, might show weak similarities to a bacterial cellulose synthase (7). Full sequence of the EST partial cDNA indeed showed all except the first D of a D,D,D,QXXRW signature (8) characterizing a heterogeneous group of processive β-glycosyl transferases and more extended but still weak similarities to a subset (9). Correcting radial swelling by transformingrsw1 (Fig. 2C) with full-length genomic clones (Fig. 2B) identical to sequences found on a yeast artificial chromosome (YAC) covering the mapped site proves that the gene is RSW1. The 3.8-kb RSW1 transcript is widespread, as are misshapen cells in mutant plants grown at 31°C (Fig. 2D). A similarly sized transcript in the mutant is consistent with the mutant allele substituting Val for Ala549 after a C to T nucleotide change (7).

Figure 2

(A) Part of contig IV fromArabidopsis chromosome 4 (19) refined with the use of primers based on partially sequenced left (L) and right (R) ends to establish YAC overlap by PCR (vertical lines), by converting g8300 to a cleaved amplified polymorphic sequence (CAPS) marker and by detecting new Co/Ler polymorphisms in three YAC ends (CAPS marker for yUP5C8RE; RFLP markers for EG6C4LE and yUP17GLE; white boxes on the genetic map). (B) Sequences hybridizing to EST T20782 lie centrally in cosmid 23H12 and within pRSW1 [cloned into pBIN19 (20)]. The 14-exon RSW1 gene produces a predicted protein product including transmembrane helices (hatched) and the D,D,D,QVLRW (8) signature. Conserved regions (black), variable regions (lighter), and residue numbers are based on the supplementary material (17). (C) Complementation of rsw1. T1 seeds of rsw1plants transformed (21) with cosmid 23H12 were selected for kanamycin resistance (21°C over 10 to 12 days). Two days at 31°C after 5 days at 21°C caused swelling of rsw1 (left) but not of T2 (center) or wild-type (right) seedlings. (D) Wild-type (inset) and rsw1 seedlings grown for 10 days at 31°C and viewed by cryoscanning electron microscopy. Epidermal cells in all organs of rsw1 plants are misshapen.

Four pieces of evidence make a compelling case that the RSW1gene product encodes the catalytic subunit of cellulose synthase: (i) The rsw1 mutation selectively inhibits cellulose synthesis and promotes accumulation of a noncrystalline β-1,4-glucan; (ii)rsw1 disassembles plasma membrane rosettes, a plausible mechanism for reducing cellulose and placing the RSW1product in the rosettes or interacting with them; (iii) the D,D,D,QXXRW signature identifies the RSW1 gene product as a processive glycosyl transferase (9) in family 2 of inverting nucleotide-diphospho-sugar glycosyltransferases (10) and with demonstrated uridine 5'-diphosphate–glucose binding ability in the highly similar cotton celA1 gene (3); and (iv) the wild-type allele corrects the mutant's radial swelling that results from reduced cellulose synthesis.

The deduced 122-kD RSW1 product (Fig.3) closely resembles the products of Ath-A and Ath-B [two full-length Arabidopsis cDNAs (11)], of the cotton celA genes proposed as cellulose synthase catalytic subunits (3), and of rice ESTs D48636(3) and D39394 (11). Architecture is conserved (Fig. 2B): Six predicted membrane-spanning regions lie close to the COOH terminus, and two others separate an extended NH2-terminal region from a central, probably cytoplasmic domain weakly similar to prokaryotic glycosyl transferases (3, 7,9, 12). Strikingly variable regions interrupt extended, highly conserved regions, which are particularly prominent in the central domain (11). The NH2-terminal regions are heterogeneous except for a cysteine-rich domain that may cause protein-protein binding (13). The predicted products of fiveArabidopsis genomic sequences (14) diverge further from RSW1, Ath-A, and Ath-B: they are smaller (710 to 828 amino acids versus 1081 in RSW1), lack an extended NH2 terminus, vary in the number and position of predicted transmembrane helices, retain extensive sequence similarities in the central domain but have major insertions and deletions, and differ in their D,D,D spacings and QXXRW motifs. Arabidopsis EST fragments recently proposed as cellulose synthases (15) show little sequence similarity to RSW1. All belong to the large class of Arabidopsis Csl genes (cellulose-synthase–like) (16), but weak similarities do not prove a function in cellulose synthesis given the widely different polymers produced by enzymes sharing weakly related sequences (10). In our view, only Ath-A and Ath-B of the full-length Arabidopsis genes sufficiently resemble the functionally characterized RSW1 to be prime candidates for additional cellulose synthases.

Figure 3

Sequence of the predicted RSW1gene product. The D,D,D,QXXRW signature is bold, conserved Cys residues are underlined, and Ala549 (substituted with Val inrsw1) is bold and underlined.

In conclusion, chemical and ultrastructural changes in the cellulose-deficient mutant combine with gene cloning, complementation of the mutant, and sequence analyses to show that the RSW1locus encodes the catalytic subunit of cellulose synthase. The noncrystalline β-1,4-glucan in the shoot of the rsw1mutant suggests that the mutant allele interrupts assembly of glucan chains into microfibrils. We hypothesize that at the restrictive temperature, mutant synthase complexes disassemble to monomers (or smaller oligomers) undetectable by freeze etching. The monomers continue producing β-1,4-glucan, but the dispersed chains fail to crystallize in an acid-resistant form. Crystallization—with consequences for wall mechanics that are central to morphogenesis and industrial fiber usage—therefore requires assembled rosettes.

  • * To whom correspondence should be addressed at Plant Cell Biology Group, Research School of Biological Sciences, Australian National University, Post Office Box 475, Canberra, ACT 2601, Australia. E-mail: richard{at}


Stay Connected to Science

Navigate This Article