The thanatos mutation in Arabidopsis thaliana cellulose synthase 3 (AtCesA3) has a dominant-negative effect on cellulose synthesis and plant growth


Author for correspondence:
Polydefkis Hatzopoulos
Tel:+30 210 529 4321/4329


  • • Genetic functional analyses of mutants in plant genes encoding cellulose synthases (CesAs) have suggested that cellulose deposition requires the activity of multiple CesA proteins.
  • • Here, a genetic screen has led to the identification of thanatos (than), a semi-dominant mutant of Arabidopsis thaliana with impaired growth of seedlings.
  • • Homozygous seedlings of than germinate and grow but do not survive. In contrast to other CesA mutants, heterozygous plants are dwarfed and display a radially swollen root phenotype. Cellulose content is reduced by approximately one-fifth in heterozygous and by two-fifths in homozygous plants, showing gene-dosage dependence. Map-based cloning revealed an amino acid substitution (P578S) in the catalytic domain of the AtCesA3 gene, indicating a critical role for this residue in the structure and function of the cellulose synthase complex. Ab initio analysis of the AtCesA3 subdomain flanking the conserved proline residue predicted that the amino acid substitution to serine alters protein secondary structure in the catalytic domain. Gene dosage-dependent expression of the AtCesA3 mutant gene in wild-type A. thaliana plants resulted in a than dominant-negative phenotype.
  • • We propose that the incorporation of a mis-folded CesA3 subunit into the cellulose synthase complex may stall or prevent the formation of functional rosette complexes.


Cellulose microfibrils, para-crystalline arrays of c. 36 chains of unbranched (1→4)-β-D-glucan chains, form the major scaffolding component in the cell wall (Carpita & McCann, 2000). Cellulose is synthesized by terminal complexes in the plasma membrane arranged in hexagonal rosette structures at the ends of microfibrils (Mueller & Brown, 1980; Herth, 1983). Each subunit of the rosette may contain six cellulose synthase (CesA) proteins that synthesize six (1→4)-β-D-glucan chains, which co-crystallize into the 36-glucan chain microfibril, with an estimated diameter of 8–10 nm (Doblin et al., 2002).

Arabidopsis thaliana has 10 CesA genes that encode proteins with 64% average sequence identity (Holland et al., 2000; Richmond, 2000). Maize (Zea mays) has at least 12 CesA genes (Appenzeller et al., 2004), while barley (Hordeum vulgare) has eight (Burton et al., 2004) and poplar (Populus trichocarpa) 18 (Djerbi et al., 2005). Functional genetic and co-expression analyses indicate that the A. thaliana CesA genes can be subdivided into two distinct groups. Group I contains genes required for synthesis of cellulose in primary (growing) cell walls: CesA1, CesA3 and CesA6, for which there are mutant alleles radial-swelling1 (rsw1; Arioli et al., 1998), isoxaben resistant1 (ixr1; Scheible et al., 2001), and procuste1 (prc1; Fagard et al., 2000), respectively. Three additional genes CesA2, CesA5 and CesA9 have sequence similarity to CesA6 and may have functions that show partial redundancy with those of CesA6 (Desprez et al., 2007; Persson et al., 2007). The expression patterns of the CesA2, CesA5, CesA6 and CesA9 genes are regulated according to developmental stage and tissue type. Group II consists of the genes CesA4, CesA7 and CesA8, for which there are mutant alleles irregular xylem5 (irx5; Taylor et al., 2003), irx3 (Taylor et al., 1999) and irx1 (Taylor et al., 2000), respectively. Group II genes are required for cellulose deposition in the thick secondary walls of vascular cells. Mutants are mainly characterized by the collapse of xylem vessels that are unable to withstand the negative pressure generated during water transport. The expression patterns of the secondary wall CesA genes show tight co-regulation, but this is less clear for primary wall CesA genes (Hamann et al., 2004; Brown et al., 2005; Persson et al., 2005). The similar effects on cellulose deposition of mutations in various single CesA genes have demonstrated that at least three different CesA gene products are required for a functional cellulose synthase complex in either primary (Desprez et al., 2007; Persson et al., 2007) or secondary walls (Taylor et al., 2000, 2003). Although numerous mutant alleles of A. thaliana CesA genes have been identified, a dominant growth phenotype is observed only upon overexpression of the fragile fiber5 (fra5) mutant allele of CesA7 (Zhong et al., 2003).

In this paper, we report the isolation of a novel missense mutation of AtCesA3 with a dominant-negative growth phenotype. The mutation occurs in the second cytoplasmic domain affecting the proline578 residue that is conserved among evolutionarily diverse, cellulose-biosynthetic organisms. In both homozygous and heterozygous seedlings, growth is impaired, dramatically so in the homozygous seedlings, consistent with an impact on primary-walled cells rather than on secondary-walled cells. Cellulose content is reduced in heterozygous and further reduced in homozygous plants, showing gene-dosage dependence. As the mutation is predicted to drastically change the secondary structure of the CesA3 subunit, we propose that the incorporation of a mis-folded CesA3 subunit into the cellulose synthase complex may stall or prevent the formation of heteromeric rosette complexes.

Materials and Methods

Plant material and growth conditions

The thanatos (than) mutant allele was isolated from a genetic screen of an M2 ethylmethanesulfonate-mutagenized Arabidopsis thaliana Columbia-0 (Col-0) population. The than mutants were backcrossed four times into the Col-0 background. Seeds were surface-sterilized and sown on 0.5× Murashige & Skoog (MS) medium (Duchefa, Haarlem, the Netherlands), pH 5.7, supplemented with 1% sucrose and solidified with 0.35% phytagel (Sigma, Gillingham, UK). After 48 h of stratification at 4°C, plates of seedlings were placed vertically for 7 d at 22°C in a growth chamber with a 16 : 8 h day:night cycle. Isoxaben pestanal (Sigma-Aldrich, Seelze, Germany) was applied at 100 nm to solidified media, allowing the seeds to grow up to the fifth day after germination to permit measurement of than/+ primary root length. Transgenic T1 and T2 plants were selected on 30 mg l−1 hygromycin.

Morphometric analysis

Measurements of primary root and hypocotyl elongation were taken from digital scans of 7-d-old seedlings grown on vertical plates. Morphometric analyses of mature plant and organ sizes were derived from digital photographs. Images were analyzed using the imagej software package (

Map-based cloning

The than mutant was crossed to the A. thaliana polymorphic ecotype Landsberg erecta. Genomic DNA was isolated from F2 individuals having a wild-type phenotype. Positional cloning was performed using the combinations of insertion/deletion (InDel) and single nucleotide polymorphism (SNP) markers provided in Supporting Information Table S1. The markers were designed based on DNA polymorphisms made available by Monsanto Company (St. Louis, MO, USA) and The Arabidopsis Information Resource (TAIR) (

PCR amplification of the than mutant locus

The At5g05170 locus was amplified from genomic DNA, isolated from than homozygous seedlings, using a set of primers to generate overlapping PCR products of 600 bp average size, which were sequenced to identify the mutation. For dominant-negative complementation, DNA from than homozygous seedlings was used as a template to amplify the AtCesA3 (P578S) gene using Phusion High-Fidelity DNA Polymerase (Finnzymes Oy, Espoo, Finland) with forward 5′-TCCCGGGATGTCTCTTGATCTCATCTCTGC-3′ and reverse 5′-TCCCGGGAAGACACCTTTAACTCAGCCT-3′ primers carrying a SmaI site (underlined sequence). The PCR product was ∼8150 bp and contained the entire AtCesA3 coding region, 2800 bp of upstream regulatory sequence and 705 bp of downstream sequence. The AtCesA3 mutant allele was cloned into the SmaI site of the pGPTV-HPT binary vector. The Agrobacterium tumefaciens strain C58C1 RifR containing the pGV3101 Ti plasmid was transformed with the pGPTV-HPT construct by electroporation (Gene Pulser II; Bio-Rad, Hercules, CA, USA). The construct was introduced into Col-0 plants by vacuum infiltration.

Quantitative real-time PCR assays

The concentrations of the PCR template nucleic acids (RNA and DNA) were determined spectrophotometrically and verified by ethidium bromide staining on agarose gels. PCR primers listed in Table S2 were manually designed for each gene according to the parameters provided by the Methods and Applications Guide for Quantitative PCR (IN 70200 B; to assure maximal efficiency and sensitivity. The A. thaliana CesA3 gene was amplified by primers giving a specific product of 228 and 135 bp for the genomic DNA and RT-PCR assays, respectively, as a result of the presence of a spanning intron. A specific pair of primers was designed to amplify CER2B, a single molecular locus at the upper arm of A. thaliana chromosome II, which permits normalization of the quantitative data for gene copy analysis. The house-keeping glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene was amplified to normalize the quantitative data for expression analysis.

Quantitative PCR reactions were performed in the Mx3005P Stratagene QPCR System using SYBR Green I as the DNA-binding dye provided in Brilliant SYBR Green QPCR Master Mix (Cat. No. 600548; Stratagene, La Jolla, CA, USA) applying the following cycler conditions: 10 min of pre-incubation at 95°C, followed by 40 cycles of 30 s at 95°C, 1 min at 55°C and 1 min at 72°C. The reactions were carried out in a volume of 25 µl in optical PCR tubes. All quantitative PCR reactions were performed as triplicates with two biological repeats. At the end of each reaction, the cycle threshold (Ct) was automatically set up at the level that reflected the best kinetic PCR parameters by the Stratagene MxPro QPCR software package and melting curve analysis was performed to monitor primer specificity. The ΔΔCt method of relative quantification (described in the Methods and Applications Guide) was adapted to estimate the AtCesA3 (P578S) transgene copy number and expression level in than phenotypic variants. Genomic DNA of Col-0 seedlings was used for calibration of the test sample harboring a single copy of the CesA3 gene (n-fold = 1). Similarly, the expression level of CesA3 in the wild-type background was set up as a reference unit for relative expression analysis. The average and standard deviation of the Ct values were calculated for each gene and sample. Results for each sample are expressed as n-fold relative to the Col-0 test sample used for calibration. For data analysis using the ΔΔCt method the amplification efficiencies of the target and reference genes were determined by constructing standard curves for each gene.


For scanning electron microscopy, etiolated hypocotyls were fixed in 3% glutaraldehyde, washed in buffer, dehydrated in a series of acetone, dried in a Polaron E3000 (POLARON, Watford, UK) critical-point drier and sputter-coated with gold. Specimens were imaged using a JEOL 6360 (JEOL Ltd., Tokyo, Japan) scanning electron microscope. Fresh toluidine blue-O (0.05%) was applied to 3–5-µm-thick sections for 1 min followed by a quick wash in water. Vascular bundles were examined with an Olympus BX-50 light microscope equipped with a DP71 camera, using Cell^A (Olympus Soft Imaging Solutions). Phloroglucinol was used to stain lignin as described by Cano-Delgado et al. (2003).

Bioinformatics analysis  Multiple alignments were performed with ClustalX 1.83 ( using default algorithms and settings. The results were exported in GCG/MSF format and analyzed using the GeneDoc MFC application, version ( The neighbor-joining phylogenetic tree was inferred from the multiple sequence alignment with 1000 bootstraps to obtain support values for each internal branch and exported to phylip (Phylogeny Inference Package) 3.65. Image representation of the calculated tree as Tagged Image File Format (TIFF) was constructed using TreeView 1.6.5 ( Protein structures were ab initio predicted using the HMMSTR/Rosetta Server ( and the resulting pdb files analyzed using pymol (

Fourier transform infrared microspectroscopy (FTIR)  For microspectroscopy, cell walls were mounted in the wells of IR-reflective, gold-plated microscope slides (Thermo-Electron, Madison, WI, USA). The slides were supported on the stage of a Nicolet Continuum (Thermo Fisher Scientific Inc., Waltham, MA, USA) series microscope accessory coupled to a 670 IR spectrophotometer with a liquid nitrogen-cooled mercury-cadmium telluride detector (Thermo-Electron, Madison, WI, USA). An area of sample (up to 125 × 125 µm) was selected for spectral collection in ‘transflectance’ mode. In transflectance, the beam is transmitted through the wall sample, reflected off the gold-plated slide, and then transmitted through the sample a second time. One hundred and twenty-eight interferograms were collected with 8 cm−1 resolution and co-added to improve the signal-to-noise ratio for each sample.

Cell wall materials were prepared from A. thaliana seedlings grown in long-day conditions for 2 wk axenically on a defined agar medium, with the last 2 d of incubation in darkness to lower starch content, a contaminating feature of infrared spectra. The entire shoots of 2-wk-old A. thaliana seedlings were harvested into liquid nitrogen, crushed to a powder in 2-ml Eppendorf centrifuge tubes, and suspended in 50 mm Tris[HCl], pH 7.2, containing 1% sodium dodecyl sulfate (SDS); they were then heated to 70°C to extract protein and other nonwall components, which were collected on a 47-µm square nylon mesh filter, and washed with water. Three 5-mm stainless-steel balls were added to the tubes and the samples were homogenized at 1200 cycles min−1 in a reciprocating shaker (Geno/Grinder; SPEX CertiPrep, Metuchen, NJ, USA) for 10 min. The cell walls from the homogenate were collected on nylon mesh filters, and washed with 50% hot ethanol and then water at 70°C. The cell wall material in water was homogenized a second time at 1200 cycles min−1 in the reciprocating shaker for 10 min. The cell walls were then collected on the nylon mesh and washed sequentially with water (70°C), 50% (v/v) ethanol (70°C), and water at ambient temperature. The walls were suspended in deionized water and allowed to settle. A twofold mechanical homogenization of the cell wall materials and collection of the walls on nylon filters were essential to yield isolated wall material free of starch and most cytoplasmic protein. At least 35 spectra, from populations of 40 or more seedlings, were obtained for each subpopulation of seedlings. For each sample, spectra were area-averaged and baseline-corrected before generating average spectra and digital subtraction spectra. Baseline-corrected and area-normalized data sets of spectra were then used in the chemometric analyses. Exploratory principal component analysis (PCA) was carried out using win-das software (Kemsley, 1998).

Cellulose determination

On the last day of the light period, 10-d-old seedlings grown vertically were transferred to clear Plexiglas fumigation chambers, and 10 µl of water containing 0.67 µCi of D-[14C]-glucose (290 mCi mmol−1; ICN, Irvine, CA, USA) was added to each seedling along the hypocotyls and roots, and incubation was continued in the light at 23°C for 7 h. Residual 14CO2 was trapped in alkali by in-line air-flow, and the fumigation chambers were vented.

Cell walls were prepared by homogenization by glass-glass grinding in 3 ml of 1% SDS in 50 mm Tris[HCl], pH 7.0 (grinding buffer), heated to 70°C to extract protein, and washed sequentially with additional grinding buffer, and water. The cell walls were homogenized a second time in water, and washed sequentially with water, 50% (v/v) ethanol, and water. Cell walls were collected by centrifugation after each washing step. Samples of the cell wall preparation were hydrolyzed in 0.8 ml of acetic-nitric reagent (acetic acid:nitric acid:water; 8 : 1 : 2 (v/v/v); Updegraff, 1969) at 100°C for 90 min in 1-ml glass conical Reacti-vials (Pierce, Rockford, IL, USA). The insoluble residue (cellulose) was pelleted. The cellulose pellet was washed several times with water, 50 µl of 12 m H2SO4 was added to hydrolyze the cellulose, and 300 µl of 2 N NaOH and 200 µl of 1 m Tris[HCl], pH 7, were added to neutralize for counting. Radioactivity in each sample was determined by liquid scintillation spectroscopy. The liquid supernatant from the acetic-nitric digestion was also neutralized with the NaOH-Tris reagent before counting by liquid scintillation spectroscopy. Relative cellulose synthesis was determined as the proportion of the total radioactivity incorporated.


The than mutation impairs seedling morphology and growth

In a screen to identify mutants with impaired root growth, we isolated a mutant with primary root length reduced by 40% compared with wild type (Fig. 1). The primary root also had a radially swollen phenotype, with a significant increase in diameter. The normal pattern of cell types in epidermal files was disrupted by ectopic cell files (Fig. 1e,f). Etiolated hypocotyls were less than half the length of wild-type hypocotyls, and siliques, floral organs and stems were also reduced in length (Fig. 1b–d).

Figure 1.

Developmental phenotypes of thanatos (than) plants. Seven-day-old than/+ (middle) and than/than (right) seedlings grown vertically on plates in the light (a) or dark (b) have reduced root and hypocotyl lengths compared with wild type (left). (c) Six-week-old than/+ (right) plants are dwarfed compared with wild type (left). (Inset) Silique and floral phenotype of than/+ (right) and wild-type (left) plants. (d) Morphometric analysis of than/+ (gray bars) plants showing the difference in size of several organs expressed as relative percentages of wild type (Columbia-0 (Col-0); black bars); n = 100. The root diameter in than/+ is increased compared with wild type. Error bars represent standard deviations of the means. (e) In the Arabidopsis thaliana primary root, cell files are organized in layers of lateral root-cap (lrc), epidermis (ep), cortex (co), endodermis (en) and stele (st). (f) In than/+ the positions of root cell files are variable, causing a radially swollen root phenotype. Bars: (a, b) 1 mm; (c) 1 cm; (e, f) 10 µm.

The mutant was allowed to self-pollinate and the phenotypic segregation ratio of progeny was examined. Analysis of 780 F1 progeny showed a 1 : 2 : 1 segregation ratio corresponding to 219 wild type : 373 with reduced growth phenotype : 188 seedlings that died after germination (χ2 = 3.86, P > 0.05). The ratio is consistent with a dominant-negative effect of the mutation on seedling morphology and growth. Because of the lethal segregating phenotype, the mutation was named thanatos (than), after the name of the ancient Greek entity for death. Thus, than represents a single Mendelian semi-dominant mutant allele segregating 1 wild-type : 2 than/+ heterozygous plants : 1 than/than homozygous lethal. The heterozygous plants have an intermediate phenotype between the wild-type and homozygous seedlings before lethality (Fig. 1a,b).

The than locus maps to A. thaliana CesA3

The locus of the than mutation was mapped by positional cloning. Coarse mapping placed the mutation on the upper arm of chromosome V (Fig. 2a). Fine-interval mapping positioned the than locus in a zero-recombination interval of ∼40 kb, flanked by proximal markers nga225 and k2a11. This region consists of 10 candidate genes with AGI annotation numbers At5g05110–At5g05200. The ciw18 marker, localized close to At5g05170, identified no recombinants among 897 F2 individuals. Sequencing of than homozygous genomic DNA revealed a nucleotide transition from G to A at the 10th exon of At5g05170 coding for cellulose synthase catalytic subunit3 (CesA3). The mutation results in amino acid substitution of proline578 to serine (Fig. 2b).

Figure 2.

Identification of the thanatos (than) mutation. (a) Recombination mapping of the than mutation within Bacterial Artificial Chromosome (BAC) K2A11 clone on the upper arm of chromosome V. The Arabidopsis Genome Initiative (AGI) annotation numbers for the BAC clones (horizontal lines), the relative positions of the molecular markers (vertical lines), and the number of recombinants (in parentheses) per 1794 chromosomes used in the positional cloning are indicated. The than mutation is linked to At5g05170 (Arabidopsis thaliana cellulose synthase 3 (AtCesA3)) which contains 14 exons (solid boxes) separated by 13 introns (lines) and causes a G-to-A mutation within the 10th exon resulting in amino acid substitution of proline578 to serine. (b) Sequence chromatograms showing the G-to-A base transition identified in than/+ (rectangle; middle) and than/than (right) seedlings compared with the wild type (left). Arrowheads indicate direction of translation. (c) Schematic representation of the domain structure of AtCesA polypeptides. Bars indicate the sites of amino acid substitutions in previously characterized mutations of AtCesA genes (Supporting Information Table S3). Gray ovals indicate AtCesA3 mutant alleles. The sites of mutations are indicated relative to domains rather than by precise residue number. CSR, class-specific region; HVR, hypervariable region; ZFD, zinc-finger domain; D, aspartic acid; Q, glutamine; R, arginine; W, tryptophan; X, any amino acid; cev, constitutive expression of the vegetative storage protein; eli, ectopic lignification; fra, fragile fiber; irx, irregular xylem; ixr, isoxaben resistant; lew, leaf wilting; mur, murus; prc, procuste; rsw, radial swelling.

Many A. thaliana CesA mutant alleles have been characterized previously (Fig. 2c, Table S3). As two CesA3 mutant alleles confer semi-dominant resistance to isoxaben, a cellulose synthesis inhibitor (Scheible et al., 2001), the effect of isoxaben on than/+ seedlings was examined. The than heterozygotes were more sensitive to the application of isoxaben compared with ixr1-1, showing a wild-type phenotypic response (Fig. S1).

Wild-type plants expressing the than allele of AtCesA3 display dominant-negative phenotypes

In order to verify that the AtCesA3 (P578S) mutant allele is responsible for the than mutant phenotype, we transformed wild-type A. thaliana plants with the AtCesA3 (P578S) transgene isolated from than homozygous mutant plants. The transgenic lines at the seedling stage demonstrated a drastic reduction in primary root lengths, and a short stem phenotype on reaching maturity (Fig. 3a,b). As wild-type plants acquired than-like developmental phenotypes when they harbored the mutated allele, we conclude that expression of AtCesA3 (P578S) results in a dominant-negative complementation.

Figure 3.

The dominant-negative effect of the thanatos (than) allele expressed in wild-type plants. (a) Seven-day-old Columbia-0 (Col-0) seedlings expressing the Arabidopsis thaliana cellulose synthase 3 (AtCesA3; P578S) transgene have reduced root lengths (mild). Soon after germination on selective media, the phenotype of several hygromycin-resistant lines was severe (strong). Bars, 1 mm. (b) The T2 generation of 6-wk-old Col-0 plants transformed with the AtCesA3 (P578S) transgene showed a range of mild to strong phenotypes. Bars, 1 cm. (c) Plants with strong than-like phenotypes have higher copy numbers of the AtCesA3 (P578S) transgene than phenotypically mild variants or the wild-type plants, which harbor a single copy of the CesA3 gene. (d) Plants with strong than-like phenotypes have increased transcript levels of the transgene compared with phenotypically mild variants or the wild-type plants, used as reference for relative expression.

To address whether the range of phenotypic variation in the transgenic lines is related to gene dosage effects of the inserted mutant allele, the relative AtCesA3 gene copy number of the phenotypic variants was assessed. Individual transgenic plant lines were categorized as strong or mild than phenotypic variants (Fig. 3a,b). Quantitative PCR analysis indicated that, as the number of copies of the inserted mutant allele increased, the phenotype of the transgenic A. thaliana plants became more severe (Fig. 3c). Α correlation between the relative number of integrated mutant allele copies and the level of AtCesA3 expression was also observed in the transgenic lines (Fig. 3c,d). These results confirmed the dominant-negative effect of the than mutation.

Cellular abnormalities in than genetic segregants

The than mutation severely disturbed the pattern of the cell files and the diameter of the primary root. To determine whether there were any cellular abnormalities, we observed embryos at the bent cotyledon stage in than genetic segregants. Mutant embryos carrying the rsw1-2 and prc1-1 alleles of CesA1 (Gillmor et al., 2002) and CesA6 (Fagard et al., 2000), respectively, were chosen because they are also expected to compromise subunits of the primary cell wall complex (Fig. 4a–e). The rsw1-2 mutant allele has a lethal phenotype, while the growth pattern of prc1-1 seedlings is similar to that of than heterozygotes. The morphology at the bent cotyledon stage of than homozygous embryos resembled the extremely swollen embryo phenotype produced by the rsw1-2 allele, with changes in cell shape and the direction of cell elongation. In contrast to than homozygous embryos, the heterozygotes had recognizable shoot and root meristems and were similar to the wild-type and prc1-1 embryos, with a slightly swollen embryo phenotype. Scanning electron micrographs of etiolated than heterozygous hypocotyls showed that the than mutation resulted in a partially hookless phenotype, with a more severe phenotype for than homozygotes (Fig. 4f).

Figure 4.

The thanatos (than) mutants have impaired embryonic development, cell expansion and elongation. Both than/than (c) and radial-swelling1-2 (rsw1-2)/rsw1-2 (e) homozygous lethal embryos (upper panel; bars, 100 µm) have swollen phenotypes. Epidermal cell morphology of than/+ (b) embryos at the bent cotyledon stage is also altered compared with procuste1-1 (prc1-1)/prc1-1 (d) and wild type (a) (lower panel; bars, 1 µm). (f) Scanning electron micrographs of the shoot apical hook of 7-d-old etiolated wild-type (left), than/+ (middle) and than/than (right) seedlings. Bars, 200 µm. (g) Magnification of the hypocotyl region. Bars, 100 µm.

To investigate whether the dwarf etiolated hypocotyl phenotype of than is caused by defects in cell expansion or division, the length and number of epidermal hypocotyl cells were assessed. The length of than homozygous hypocotyl epidermal cells (0.032 ± 0.007 mm; mean ± SE) was drastically reduced to one-tenth of that of the segregating wild type (0.273 ± 0.063 mm). The length of heterozygous cells (0.113 ± 0.025 mm) was less than half that of the wild type. These data represent mean ± SE for measurements of ∼60 cells from four etiolated hypocotyls (Fig. 4g). However, the ratio between the hypocotyl length and the average length of its epidermal cell was the same for the than genetic segregants and the wild-type plants, indicating no difference in the number of cells. The short hypocotyl phenotype caused by alteration in cell size suggests a reduction of anisotropic cell expansion.

We further examined the cellular morphology of the vascular bundles in inflorescence stems (Fig. 5). The wild-type vascular bundles contained phloem, procambium and xylem vessel elements with a relatively round shape. The shape of vascular bundles from than heterozygous and wild-type plants expressing the AtCesA3 (P578S) transgene was distorted and the cell walls were abnormal. These cross-sections showed intense toluidine blue staining potentially indicating lignin deposition. Lignin hyperaccumulation in than genetic segregants was confirmed with phloroglucinol cytochemical staining (Fig. S2).

Figure 5.

Vascular morphologies of thanatos (than)/+ mutant and transgenic plants. Cross-sections of inflorescence stems stained with toluidine blue from wild-type (left), than/+ (middle) and Columbia-0 (Col-0) plants expressing the Arabidopsis thaliana cellulose synthase 3 (AtCesA3; P578S) transgene (right) showing abnormal cell walls. ph, phloem; pc, procambium; xe, xylem elements. Bar, 50 µm.

Substitution of conserved proline578 is predicted to alter the structure of the CESA3 catalytic domain

Multiple sequence alignment of the protein region flanking proline578 revealed a consensus sequence between plant and nonplant CesA genes and members of the A. thaliana Cellulose synthase-like (Csl) gene family (Fig. 6a, Table S4). There are three clusters of CesA3 orthologs (Fig. 6b). Clade I exclusively contains the nonplant Eucaryota, whereas Clade II comprises bacterial orthologs. Clade III is subdivided into two subclusters. Clade IIIa predominantly includes the Csl orthologs, while Clade IIIb includes the majority of plant CesA3 genes. Members of AtCslD gene subfamily group together with CesA, in contrast to other AtCsl genes, as previously described by Lerouxel et al. (2006). Intriguingly, proline578 was conserved among all the diverse protein accessions.

Figure 6.

Bioinformatic analysis of cellulose synthase 3 (CesA3) proteins. (a) Multiple sequence alignment of the catalytic and substrate binding domain surrounding proline578 from plant and nonplant CesA3 homologs, and selected members of the Arabidopsis thaliana Cellulose synthase-like (Csl) family. The cross marks the conserved proline residue. The rectangle in green indicates the DXD motif of processive β-glycosyltransferases. H4 and S5, S6 represent the predicted secondary structures of helices and β-strands, respectively, from Acetobacter xylinum. (b) Unrooted neighbor-joining tree computed from the multiple sequence alignment. The group of proteins within each eclipse represents evolutionarily different clades identified in the analysis. Bootstrap values between 50 and 80% are depicted with black spots, whereas the remaining bootstrap values are > 80%. (c) Ab initio prediction of the catalytic domain by the HMMSTR/Rosetta Server showed that the amino acid substitution of proline578 to serine alters the secondary structure.

In the absence of a crystal structure for any CesA protein, we used the HMMSTR/Rosetta Server (Bystroff & Shao, 2002) to model whether Ser substitution of the conserved Pro might alter the three-dimensional structure of CesA3. We compared the secondary structure of the A. thaliana CesA3 globular region with that from Acetobacter xylinum cellulose synthase as previously identified by Saxena et al. (2001). The predicted α-helices and β-strands of A. thaliana CesA3 were perfectly aligned with the H4, S5 and S6 of A. xylinum (Fig. 6a,c). The model also predicted that Pro interrupts the secondary structure of the catalytic domain of CesA3, leading to the formation of two successive antiparallel β-strands. Substitution to Ser changed this structure to a long single β-strand (Fig. 6c). These predictions suggest that Pro578 is a critical residue for maintenance of the three-dimensional architecture of the catalytic domain.

CesA3 (P578S) gene dosage effect on cellulose synthesis

Cell walls were isolated from 8-d-old plate-grown seedlings of than genetic segregants, and 40 infrared spectra obtained from wild-type, than homozygous and heterozygous seedlings. The spectra were averaged for each of the three populations and compared by digital subtraction. An example of the analysis is shown in Fig. 7 comparing than/than and +/+ genetic segregants. The average spectra are baseline-corrected and area-normalized to each other (Fig. 7a) before digital subtraction and therefore the difference spectrum represents proportional differences (Fig. 7b). Positive peaks characteristic of cellulose (1165, 1107, 1057 and 1030 cm−1) and the ester peak at 1743 cm−1 are relatively enriched in +/+ segregants, while negative peaks characteristic of proteins (1647 and 1539 cm−1) and phenolics (1520 cm−1) are relatively enriched in than/than segregants. A PCA indicates that 95% of the total variance between the two populations is accounted for by the first PC – the loading for PC1 has the same features as those of the digital subtraction, indicating that features characteristic of cellulose (1160, 1103, 1057 and 1026 cm−1) and the ester group (1743 cm−1) can be used to discriminate between the two spectral populations (Fig. 7c,d). The same features are revealed when we compare wild-type with heterozygous segregants (data not shown). Cell walls from Col-0 and wild-type segregants cannot be resolved from each other (data not shown).

Figure 7.

Fourier transform infrared (FTIR) spectroscopic analysis of thanatos (than)/than and +/+ genetic segregants. (a) Average spectra after baseline correction and area normalization of populations of 40 wild-type segregants (black line) and 40 than/than segregants (red line). (b) Digital subtraction spectrum of the spectral differences between +/+ and than/than average spectra. Positive peaks characteristic of cellulose (1165, 1107, 1057 and 1030 cm−1) and the ester peak at 1743 cm−1 are relatively enriched in +/+ segregants, while negative peaks characteristic of proteins (1647 and 1539 cm−1) and phenolics (1520 cm−1) are relatively enriched in than/than segregants. (c) Principal component analysis (PCA) of FTIR spectra from +/+ (closed circles) and than/than (open circles) genetic segregants. PC1 accounted for 95% and PC2 accounted for 2% of total variance. (d) Loading plot of PC1 (black line) and PC2 (blue line). The PC1 loading is very similar to the digital subtraction spectrum shown in (b).

To assay cellulose content, Col-0 and the segregating population for than were grown axenically for 10 d on agar plates before 10-µl droplets of D-[U-14C]-Glc dissolved in water were applied to each seedling along the hypocotyl and roots. Seedlings were incubated in the light for 7 h. The seedlings were phenotyped into +/+, than/+, and than/than populations and frozen in liquid nitrogen, and cell walls were isolated. The percentage of radioactivity in cellulose was determined by the fraction resistant to acetic-nitric digestion. Homozygous wild-type segregants had 31.7 ± 1.9% cellulose, compared with 25.7 ± 1.4% for the than/+ heterozygous seedlings, and 18.4 ± 2.1% for homozygous than mutants.


thanatos is a nonconditional semi-dominant mutant allele of AtCesA3

Cellulose is synthesized at the plasma membrane by hexameric rosette terminal complexes to form crystalline microfibrils that act as a framework for the deposition of the noncellulosic wall polymers. Genetic mutant analyses and physical interaction experiments show that a functional complex requires at least three different CesA isoforms (Taylor et al., 2000, 2003; Desprez et al., 2007; Persson et al., 2007). All CesA proteins share a domain structure of two transmembrane domains (TMDs) close to the amino-terminus, six TMDs towards the carboxy-terminus and a large hydrophilic domain that faces the cytosol (Fig. 2c). This domain contains the conserved DXD motif, two other single aspartate residues and the Q/RXXRW consensus that are collectively referred to as the D,D,D,Q/RXXRW motif (Pear et al., 1996; Saxena & Brown, 1997). This motif is characteristic of the class of processive β-glycosyltransferases (GTs) that belong to the GT2 family ( The cytoplasmic catalytic domain bridging TMD2 and TMD3 contains a class-specific region (CSR; Vergara & Carpita, 2001) that may contribute to subfunctionalization within the gene family. The relatively short hydrophilic N-terminus contains the first hypervariable region (HVR1) and the cysteine-rich zinc-finger domains (ZFD) that are common among CesA proteins. The ZFDs act as redox-regulated multimerization domains and may be involved in the formation of homo- or heterodimers of CesA monomers for assembly of the rosette complex (Kurek et al., 2002).

thanatos is a novel mutant allele of CesA3 located downstream of the conserved DXD motif in the globular region that modulates substrate catalysis and binding of processive glycosyltransferases (Saxena et al., 2001). Map-based cloning and sequencing of the than locus revealed a single-nucleotide transition from G to A causing the amino acid substitution of proline578 to serine. Ab initio analysis predicted a radical change in the three-dimensional configuration of the mutated AtCesA3 domain flanking the proline residue. Curiously, the same proline has been mutated in the fra5 allele of AtCesA7, a gene involved in secondary cell wall formation (Zhong et al., 2003). Noticeable growth phenotypes are only observed when the fra5 mutant cDNA is overexpressed in wild-type plants. Nevertheless, the fra5 mutation causes a dominant negative effect in terms of fiber cell wall thickness and cellulose content. By contrast, than has a dramatic nonconditional growth phenotype even in heterozygous plants, probably because primary walled rather than secondary walled cells are affected. Both than and fra5 mutants reveal the critical role of the proline residue in the structure and function of cellulose synthase complexes. The incorporation of a than defective CesA3 subunit into the rosette complex could impair the production of normal (1→4)-β-D-glucan chains. Our results indicate that the predicted drastic conformational change caused by proline to serine substitution even in one of the three subunits may impair the construction of an active rosette or prevent its assembly. Nevertheless, further experimental evidence is required to show whether the defective than polypeptide associates into the complex and how it affects the activity of the whole rosette complex.

To date, six mutant alleles of CesA3 have been characterized. The constitutive expression of the vegetative storage protein1(cev1; Ellis & Turner, 2001; Ellis et al., 2002), ectopic lignification1-1 (eli1-1) and eli1-2 (Cano-Delgado et al., 2003) mutations occur in the central catalytic domain. These mutants show constitutive expression of defense-related genes or ectopic lignin deposition. The rsw5 mutant is a temperature-sensitive allele of CesA3 in the carboxy-terminal domain following TMD8 (Wang et al., 2006). Whereas these CesA3 missense mutations are recessive, the conditional semi-dominant alleles ixr1-1 and ixr1-2 that lie between TMDs 7 and 8 confer resistance to the potent cellulose synthesis inhibitor isoxaben (Scheible et al., 2001). However, than is sensitive to isoxaben application. It appears that isoxaben resistance is coupled with missense mutations located in close proximity to the carboxy-terminus of cellulose synthases which is quite distant from the catalytic and substrate binding domain (Fig. 2c). Consistent with this observation, ixr2-1, the isoxaben-resistant allele of CesA6, also carries a missense mutation close to the carboxy-terminus (Desprez et al., 2002). The ixr1-1, ixr1-2 and ixr2-1 semi-dominant isoxaben-resistant alleles suggest that isoxaben affects the sensitivity of the CesA3- and CesA6-containing complex (Desprez et al., 2002, 2007).

Measurements of cellulose content in segregating populations showed a gene dose-dependent effect of the than mutation on cellulose synthesis, with cellulose content reduced by 19% in heterozygotes and by 42% in homozygotes. This gene dosage effect was also apparent when genomic DNA containing the promoter and coding sequence of the AtCesA3 (P578S) mutant gene was introduced into the wild-type A. thaliana background. The than-like phenotypic variation in transgenic lines appears to be correlated with the copy-number-dependent expression of the integrated transgene.

FTIR spectra showed changes in ester content, which appear to be characteristic of cellulose-deficient mutants (Fagard et al., 2000; Schindelman et al., 2001). As in the eli1-1 and eli1-2 mutant alleles of CesA3, the reduction of cellulose synthesis in the than mutant induces ectopic lignification. A regulatory mechanism may monitor cell wall integrity by inducing lignification as a mechanical barrier when the tensile strength of the wall is compromised (Cano-Delgado et al., 2003). In contrast to all AtCesA mutant alleles isolated to date, only than has a nonconditional dominant-negative effect on growth under normal conditions as a result of defective cellulose deposition during primary cell wall formation, while fra5 exhibits a dominant growth phenotype only upon overexpression of the mutant allele (Zhong et al., 2003).

thanatos perturbs cell morphology and plant growth

Microscopic analysis of than embryos revealed that the mutation caused defects in cell expansion and elongation. The embryos had a swollen phenotype and impaired development, and the formation of the shoot apical hook in than etiolated seedlings was dramatically affected. The swollen phenotype persisted during the elongation of the primary root and resulted in abnormal organization of the root cell files. The cellular abnormalities in than genetic segregants probably originate from the reduction of anisotropic cell expansion (Wasteneys, 2004).

Transverse sections through the vascular bundles of inflorescent stems showed abnormal cell shapes and organization both in than heterozygous plants and in wild-type plants expressing the AtCesA3 (P578S) transgene. Similar phenotypes have been previously described for either the hypocotyls of wild-type dark-grown seedlings treated with herbicides that inhibit cellulose biosynthesis (Desprez et al., 2002) or mutant alleles of CesA proteins involved in primary wall cellulose synthesis, including rsw1-1/cesa1 (Arioli et al., 1998) and prc1/cesa6 (Fagard et al., 2000). The phenotype of than heterozygous seedlings is similar to that produced by the prc1-1/cesa6 mutant allele, whereas the than homozygous lethal phenotype resembles that of rsw1-2, the strong nonconditional mutant allele of CesA1. These data are consistent with observations that both CesA3 and CesA1 are unique components of the primary wall cellulose synthase complex, while CesA6 may be functionally redundant with other CesA6-related CesAs (Persson et al., 2007).

Cellulose deficiency may also induce feedback mechanisms affecting the synthesis of other wall components. Recent fluorescent live-cell imaging of CesA6 (Paredez et al., 2006) and CesA3 (Desprez et al., 2007) identified significant intracellular reservoirs of CesA proteins in the Golgi that do not exclusively coincide with cellulose synthase complex assembly. The intracellular trafficking of CesAs could play an important role in the developmental and environmental regulation of cell wall composition. The semi-dominant thanatos mutation provides a further tool for use in dissecting the interactions of cellulose synthase subunits in assembly of a functional rosette and contributes to better understanding of plant growth mechanisms.


The authors thank NASC, UK for kindly providing plant materials. This work was supported by Pythagoras I and ΙI-Ministry of Education and PENED-01 GSRT to PH. GD is indebted for funding to the Greek State Scholarships Foundation. MCM and NCC gratefully acknowledge support from the NSF Plant Genome Research Program DBI-0217552.