The irregular xylem 2 mutant is an allele of korrigan that affects the secondary cell wall of Arabidopsis thaliana

Authors


For correspondence (fax +44 161 275 3938; e-mail simon.turner@man.ac.uk).

Current address: CNAP, Department of Biology, University of York, Heslington, York YO10 5DD, UK.

Current address: Environmental Sciences, Inveresk Research International, Tranent EH33 2NE, Scotland, UK.

Summary

The irregular xylem 2 (irx2) mutant of Arabidopsis thaliana exhibits a cellulose deficiency in the secondary cell wall, which is brought about by a point mutation in the KORRIGAN (KOR) β,1-4 endoglucanase (β,1-4 EGase) gene. Measurement of the total crystalline cellulose in the inflorescence stem indicates that the irx2 mutant contains approximately 30% of the level present in the wild type (WT). Fourier–Transform Infra Red (FTIR) analysis, however, indicates that there is no decrease in cellulose in primary cell walls of the cortical and epidermal cells of the stem. KOR expression is correlated with cellulose synthesis and is highly expressed in cells synthesising a secondary cell wall. Co-precipitation experiments, using either an epitope-tagged form of KOR or IRX3 (AtCesA7), suggest that KOR is not an integral part of the cellulose synthase complex. These data are supported by immunolocalisation of KOR that suggests that KOR does not localise to sites of secondary cell wall deposition in the developing xylem. The defect in irx2 plant is consistent with a role for KOR in the later stages of secondary cell wall formation, suggesting a role in processing of the growing microfibrils or release of the cellulose synthase complex.

Introduction

Higher plant cell walls are dynamic, complex networks of polysaccharides, proteins and phenylpropanoid polymers that form a gel support matrix surrounding the plasma membrane (Carpita, 1998). They can be classified into primary and secondary types, each varying in composition. Primary cell walls consist of two phases: the cellulose/hemicellulose phase and the pectin phase (McCann and Roberts, 1991), while secondary cell walls often contain an additional polyphenolic lignin component incorporated into the cellulose/hemicellulose network (Campbell and Sederoff, 1996).

Cellulose is the world's most abundant biopolymer, essential for the proper structure and function of the cell walls of higher plants (Arioli et al., 1998; Delmer, 1999). The chemical structure of cellulose comprises β,1-4-linked glucose residues, each orientated 180° to its neighbour, such that the repeating unit is cellobiose. The configuration of glucose residues dictates that the cellulose polymer adopts a strong ribbon-like structure resistant to extension. These polymer chains associate to form crystalline microfibrils, which are arranged in parallel within the cell wall. The physical properties of individual cellulose chains, along with their organised arrangement, enable cellulose to function as the major load-bearing element of the cell wall composite network, governing the processes of cell division, elongation and differentiation (Carpita, 1998; Kerstens et al., 2001).

While other polysaccharides that form the cell wall matrix are synthesised and assembled in the Golgi apparatus (Driouich et al., 1993), the site of cellulose biosynthesis is at the plasma membrane. Large multimeric rosette complexes are embedded in the plasma membrane, and they extrude cellulose into the cell wall (reviewed by Delmer and Amor, 1995). Little is known concerning the identity of the components of this complex, which appears to comprise of six lobes. Genes that were presumed to encode the catalytic subunit of the cellulose synthase complex were identified through analysis of expressed sequence tags that showed homology to bacterial cellulose synthases. These genes are now referred to as the CesA gene family (Delmer, 1999; Pear et al., 1996). The subsequent analysis of Arabidopsis thaliana mutants, which exhibited a cellulose-deficient phenotype, indicated that multiple CesA proteins are required for cellulose synthesis in both the primary and the secondary cell walls (Arioli et al., 1998; Fagard et al., 2000; Scheible et al., 2001; Taylor et al., 1999, 2000, 2003). The presence of CesA proteins in the plasma membrane rosette complexes has been shown by labelling studies using antibodies recognising CesA proteins (Kimura et al., 1999). Other components of this complex, however, have yet to be identified.

One of the problems in taking a genetic approach to studying cellulose synthesis is highlighted by work on cyt1, knf and rsw3 (Burn et al., 2002; Gillmor et al., 2002; Lukowitz et al., 2001). These mutants exhibit a severe phenotype, consistent with a dramatic reduction in cellulose content. The defects are caused by mutations in mannose-1-phosphate guanylyltransferase, glycosidase I and glycosidase II, respectively. These enzymes are part of an essential pathway that processes carbohydrates during the assembly and folding of membrane proteins in the endoplasmic reticulum. So, while a genetic approach has been successful in identifying candidate genes that may be involved in cellulose synthesis, it has become increasingly hard to distinguish those core components that are absolutely required for cellulose synthesis from other components that are required for more general housekeeping functions within the cell.

The endoglucanase KORRIGAN (KOR) appears to be essential for cellulose synthesis. Originally isolated from a mutant exhibiting a lack of elongation in the absence of light, it was initially thought to be defective in some aspect of cell wall loosening (Nicol et al., 1998). Additional alleles of kor were subsequently isolated that exhibit defects in cell plate formation (Zuo et al., 2000) and a specific reduction in cellulose content of the primary cell wall (Lane et al., 2001; Sato et al., 2001). A Brassica napus orthologue of the KOR endoglucanase protein (Cel16) has been demonstrated to have activity against low-substituted carboxymethylcellulose and amorphous cellulose (Molhoj et al., 2001). Circumstantial evidence that KOR may be directly involved in plant cellulose biosynthesis is gained from the fact that a β,1-4 endoglucanase (β,1-4 EGase) is essential for normal cellulose synthesis in the bacteria Agrobacterium tumefaciens and Acetobacter xylinum (Matthysse et al., 1995; Standal et al., 1994).

Sitosterol cellodextrins have been identified as primers for cellulose biosynthesis (Peng et al., 2002). It has been suggested that KOR may act to remove glucan units from these primers, which are subsequently organised into the growing cellulose microfibrils (Peng et al., 2002); however, despite the considerable information available on KOR, its function in cellulose synthesis is unclear. In this paper, we report that both alleles of the irregular xylem 2 (irx2) complementation groups are caused by mutations in kor. Further analysis of irx2-1 shows that while it has a severe defect in cellulose synthesis in the secondary cell walls, it shows no detectable decrease in cellulose content in the primary cell wall. While expression and localisation supports a role for KOR in cellulose biosynthesis, it does not appear to be an integral part of the cellulose synthase complex.

Results

Map position of irx2 locus

Two alleles of the irx2 complementation group were isolated from different M2 populations of ethylmethylsulphonate-mutagenised Arabidopsis plants. Mutants were isolated on the basis of their collapsed xylem phenotype (Figure 1a,b; Turner and Somerville, 1997). irx2-1 was isolated from a Landsberg erecta (Ler) background, while irx2-2 was derived from ecotype Columbia.

Figure 1.

Phenotype of WT and irx2 plants.

(a) Wild-type Ler.

(b) irx2-1.

(c) irx2-1 transformed with pCKS5.

(d) irx2-1 transformed with pCNHisKOR.

(e, f) kor-1.

Stem sections stained with toluidine blue showing the tracheary element of mature xylem. Magnification 400×. xe, xylem elements.

Preliminary linkage analysis, involving 142 test-cross plants, suggested the irx2 locus mapped to the bottom arm of chromosome V between the markers LFY (17 recombinants) and DFR (10 recombinants). In order to identify the gene affected by the irx2 mutation, a total population of 948 test-cross individuals was used to fine-map irx2 relative to microsatellite markers. Analysis of the recombination data suggested that the irx2 locus was located in a region spanned by two closely linked markers located on bacterial artificial chromosome (BAC) K9P8 (three recombinants) and transformation-competent bacterial artificial chromosome (TAC) K2I5 (one recombinant). The microsatellite marker K2I5s is 11.8 kb from the KOR gene on TAC K2I5. Although irx2 plants contained none of the seedling defects exhibited by kor mutants, as both mutants affect cellulose synthesis, it was considered a good candidate for the irx2 locus.

Isolation of mutant alleles

The nature of the lesions in irx2-1 and irx2-2 were determined by sequencing multiple clones isolated by RT-PCR. These sequences were compared to the wild-type (WT) KOR sequence from ecotype Columbia (GenBank Accession number AF073875). For both alleles, single point mutations were identified that lead to the substitution of a single amino acid. To ensure that these were not polymorphisms between ecotypes, the Ler sequence in these regions was also determined by sequencing RT-PCR products and was found to be the same as the Columbia sequence. Both mutations resulted in the alteration of a Pro residue to Leu at 250Pro in irx2-1 and 553Pro in irx2-2. Both these mutations occur in highly conserved amino acids. From DNA sequence analysis, Arabidopsis contains 25 endoglucanases that are all members of family 9 (http://afmb.cnrs-mrs.fr/~cazy/CAZY/index.html). The prolines mutated in irx2-1 and irx2-2 are both conserved in all 25 family members, as well as a diverse range of fungi and bacteria that includes Cellulomonas fimi, Clostridium stercorarium and Thermobifida fusca.

Complementation of irx2 with the WT KOR endoglucanase gene

Measurements of crystalline cellulose, i.e. that which is resistant to acid hydrolysis, revealed that the irx2 mutant was deficient in cellulose, the base of mature stems containing approximately 30% of the level present in WT (Figure 2). Furthermore, there was no increase in the glucose in the non-cellulose fraction, suggesting that irx2 causes a specific decrease in cellulose synthesis in the secondary cell wall (Turner and Somerville, 1997). In order to verify that the irx2-1 mutation was within the KOR gene, a 5-kb StuI restriction fragment, containing the Ler WT gene and 2252 bp of promoter and no other complete open-reading frames, was cloned from TAC K2I5 into the binary transformation vector pC2300. The resulting construct (pCKS5) was transformed into Agrobacterium and was used to transform irx2-1 plants. Transformants were selected and grown for 6 weeks alongside Ler WT and untransformed irx2-1. None of the 15 transformants examined exhibited the irx phenotype and were indistinguishable from the WT (Figure 1c). Similarly, analysis of the cellulose content from the base of mature stems of these plants show that cellulose levels in irx2-1 transformed with pCKS5 were returned to WT levels (Figure 2).

Figure 2.

Cellulose measurements showing complementation of irx2-1.

Error bars represent SE. Values are the mean of three replicate measurements from five independent transformants.

Further characterisation of the irx2 phenotype

Mutant plants were slightly smaller than the WT plants, but both the altered morphology and the degree of xylem collapse were less severe than for other irx mutants, such as irx3-1 (Turner and Somerville, 1997).

A number of plants with decreased cellulose content exhibit altered seedling growth patterns. When grown in the dark on high concentrations of sucrose, procuste (prc) seedlings exhibit reduced elongation (Fagard et al., 2000), as does the kor-1 mutant (Nicol et al., 1998). In contrast, irx2 mutants elongate normally under these conditions (Table 1) and do not exhibit any significant elongation defect. Furthermore, neither alleles of irx2 exhibit any radial swelling, abnormal cell shape or incomplete cell plate phenotypes associated with mutants with decreased cellulose in the primary cell wall (data not shown).

Table 1.  Hypocotyl lengths of irx2-1 plants
LineHypocotyl length (mm)
  1. Plants were grown vertically on plates for 9 days in the dark. SEs are shown (n ≥ 15). The cellulose-deficient mutant prc is shown for reference.

Columbia10.4 ± 0.63
irx2-29.6 ± 0.55
prc5 ± 0.43

To compare the irx2 phenotype with that of kor-1, the xylem from the inflorescence stems of kor-1 plants was examined. The xylem from the inflorescence stems from kor-1 exhibits a severe collapsed xylem phenotype (Figure 1e,f), demonstrating that kor-1 also affects secondary cell walls.

FTIR analysis of cell walls

FTIR microscopy was used to determine both the amount and the structure of cellulose in cell walls from irx2 plants. Briefly, the hydroxyl groups of cell wall carbohydrates were deuterated by exchange with D2O vapour. The internal chains of cellulose microfibrils are resistant to deuterium exchange and the complex of stretching bands from their –OH groups remains detectable at around 3350 cm−1 in the FTIR spectrum (Figure 3). Its total intensity is a measure of the content of crystalline cellulose, and its shape would be altered by any major change in crystal structure, for example, conversion to cellulose II would be immediately obvious. Crystal-surface chains are not detected, as these are deuterated.

Figure 3.

FTIR spectra of cell walls from cortical parenchyma of mature irx2 and WT stems.

Spectra were recorded following vapour-phase deuteration of hydroxyl groups on all but the crystal-interior chains of cellulose.

Tissue sections, whole flattened hypocotyls or isolated cell walls can be used for FTIR microscopy. Isolated cell walls are preferable, provided that the isolation procedure yields cell wall fragments from specific, identifiable cell types and that their thickness is suitable. This is normally the case for primary-walled tissues, where the use of isolated cell walls avoids interference from crystalline chain segments in granular starch and from internal segments of proteins, containing deuteration-resistant –OH and –NH groups, respectively. Isolated primary cell walls were used here for FTIR. However, it was not possible to use isolated xylem and interfascicular cell walls. Procedures that gave clean preparations of vascular cell walls from WT and irx3 stems (Ha et al., 2002) and other plant materials did not do so with irx2, apparently because of a change in fracture properties. Instead, the xylem and interfascicular tissues were hand-sectioned for FTIR microscopy.

The xylem and interfascicular cell walls from irx2-1 contained less of the crystalline cellulose fraction, measured by deuteration-FTIR, than the cells walls of the WT (Table 2). The difference from the WT was less than that observed by chemical measurement, probably because the FTIR spectra were obtained from close to the mid-point of the flowering stem, where the phenotype is more weakly expressed, than in the mature basal region. The corresponding cell walls in the basal portion of the stem were too dense for quantification of cellulose by FTIR.

Table 2.  Relative content of crystalline cellulose in xylem/interfascicular and cortical cell walls of mature irx2-1 and WT stems
 Relative content of crystalline cellulose
MeanSEMP
  1. Cortical cell walls of kor-1 and its WT (Ler) are also included for comparison. Measurements are based on the residual FTIR O–H stretching band area after deuteration of hydroxyl groups on other polysaccharides. The area of the O–H stretching band is normalised relative to the area of the C–H stretching band in each spectrum. The cellulose content is then relative to unit mass of cell wall carbohydrate (see Experimental procedures), although comparisons between different cell types are not valid. Data presented as mean and SEM for n = 7–25. Comparisons are based on one-way anova. ns, not significant.

Stem vascular tissue
 irx21.440.060.01
 WT1.710.08 
Cortical parenchyma cell walls
 irx22.290.08ns
 WT2.270.09 
 kor-11.600.170.05
 WT2.200.18 

Isolated primary cell walls from the cortical parenchyma of the flowering stem showed no significant difference in crystalline cellulose content between irx2-1 and the WT (Table 2), whereas a significant reduction in crystalline cellulose content was observed in the corresponding primary cell walls from mature kor-1 plants. In case this effect was specific to the flowering stems, epidermal cell walls from flowering stems and rosette leaves were also examined. Again, there was no significant difference in crystalline cellulose content between irx2-1 and the WT (Table 3).

Table 3.  Relative content of crystalline cellulose in epidermal cell walls of mature irx2-1 and WT plants
 Content of crystalline cellulose
MeanSEMP
  1. Measurements are based on the residual FTIR O–H stretching band area after deuteration of hydroxyl groups on other polysaccharides. Cellulose content is relative to unit area of epidermal cell wall. Data presented as mean ± SEM for n = 10–14. Comparisons are based on one-way anova. ns, not significant.

Stem epidermis cell walls
 irx20.0430.001ns
 WT0.0450.002 
Rosette-leaf epidermis cell walls
 irx20.0190.003ns
 WT0.0140.002 

The FTIR spectra of deuterated primary cell walls from irx2-1 stems showed no evidence of any change in the structure of the crystalline cellulose or in the non-cellulosic polysaccharide composition in comparison to the WT (Figure 3).

Expression and localisation of the KOR gene

RNA gel blot analysis was employed to investigate the level of expression of the KOR gene in WT and irx2-1. A 925-bp probe was generated from the 5′ region of the KOR gene. Probing RNA isolated from different tissues demonstrated high levels of expression in stem tissue and comparatively low levels of expression in leaves (Figure 4). The pattern of expression is similar to that obtained using a probe for caffeic O-methyl transferase (COMT), a gene successfully employed as a marker of secondary cell-wall-specific transcription (Jones et al., 2001), indicating that in WT, the KOR gene is significantly upregulated during secondary cell wall synthesis (Figure 4). Expression of KOR is also detectable in the stem tips, as well as at a very low level in leaves, whereas COMT is undetectable in these tissues. Small differences occur in the expression of KOR between WT and irx2-1. These differences are probably a result of small difference in the development of the mutant compared to the WT.

Figure 4.

RNA gel blots showing expression of the irx2 gene.

Blots containing RNA from developing stem segments and leaves from WT and irx2-1 plants were probed with KOR and COMT. Ethidium bromide-stained rRNA is shown as a loading control.

Tissue prints of stems probed with this antibody demonstrated that KOR was expressed in the same cells as IRX3, a cellulose synthase catalytic subunit specific for secondary cell wall cellulose synthesis (Figure 5a).

Figure 5.

Localisation of KOR during secondary cell wall formation.

(a) Tissue prints from developing stems of WT Ler plants showing the distribution of KOR and IRX3 in the xylem and interfascicular region undergoing secondary cell wall formation. Black (xylem) or white (interfascicular region) arrows indicate the same points in serial tissue prints from the same stem section.

(b) Immunolocalisation of KOR and α-tubulin in whole mounts of developing Arabidopsis roots. The panel is designed to show the variation in KOR distribution in developing xylem vessels. The top panel shows some evidence of banding similar to the MTs, while the bottom row shows no apparent correlation between KOR and the MTs. Magnification 400×.

Subcellular localisation of the KOR protein

Immunofluorescence, using a specific antibody raised against the KOR protein demonstrated on whole mounts of Arabidopsis roots, demonstrated that KOR was expressed in developing xylem, as well as in other parts of the root (Figure 5b). Developing xylem vessels are characterised by a banded pattern of tubulin labelling, indicating cortical microtubules (MTs), that mark the sites of secondary cell wall deposition and facilitate their identification (Figure 5b). In general, KOR appears to be localised throughout most of the cell. In some instance, there appears to be some banding of KOR in xylem vessels; however, only a minor proportion of the KOR protein co-localises with the cortical MTs that mark the sites of secondary cell wall deposition (Figure 5b).

Association of KOR with other genes involved in cellulose synthesis

Previous investigations have demonstrated the interaction of the detergent-solubilised proteins CesA4, CesA7 and CesA8, such that they can be co-purified using His-affinity resin (Taylor et al., 2000, 2003). KOR may be solubilised from membranes using the detergents Nonidet P-40 (NP-40) and Triton X-100 (data not shown). Using an RGSHis-tagged version of IRX3 under conditions in which IRX1 co-purifies, there is no evidence that KOR also co-purifies (Figure 6a). The absence of any KOR in these co-purifications suggested the possibility that KOR is not a part of a complex that contained the secondary cell wall-specific CesA proteins CesA4, CesA7 and CesA8. The low signal obtained using the anti-KOR antibody, however, could have prevented the detection of small amounts of KOR that co-precipitated with IRX3.

Figure 6.

Absence of association between KOR and IRX3.

Protein gel blots showing purification of epitope-tagged proteins.

(a) Extracts from NHisIRX3 plants probed with anti-IRX3 (top), anti-IRX1 (middle) and anti-KOR (bottom) antibodies.

(b) Extracts from NHisKOR plants probed with anti-RGSH4 and anti-IRX3 antibodies.

Molecular mass markers are shown at right in kilodaltons.

In order to confirm whether the KOR protein does associate with detergent-solubilised secondary cell wall CesA proteins, the converse purification experiment was performed. An epitope-tagged version of KOR was constructed with an RGS(H)6 epitope 31 amino acids from the N-terminus. The tag did not interfere with the function of the protein as this construct (NHisKOR) was still able to complement the mutant phenotype when transformed into irx2-1 plants (Figure 1d). Subsequent purification of the detergent-solubilised epitope-tagged polypeptide was carried out using Cobalt ion resin, which binds contiguous His amino acid residues. The epitope tag enabled the detection of the tagged protein using a monoclonal antibody, anti-RGS(H)4. Significant amounts of native NHisKOR were purified using the His-affinity resin (Figure 6b). The diffuse signal at approximately 78 kDa suggests that the purified protein is highly glycosylated (Figure 6b) as previously suggested (Molhoj et al., 2001). The same extracts were probed with an antibody specific to IRX3 (AtCesA7). No detectable IRX3 co-purified with the tagged KOR protein. No signal was detectable resulting from any background non-specific binding to the column of the untagged WT KOR protein (data not shown). These results suggest that under the conditions used, KOR does not associate with the secondary cell wall CesA protein complex.

Discussion

Three lines of evidence demonstrate that irx2 is an allele of KOR. irx2 maps very closely (1 recombinant from 948 chromosomes) to a microsatellite marker on TAC K2I5 located on the bottom arm of chromosome V. This marker is located only 11 kb from the KOR gene. Both irx2 alleles contain a single nucleotide mutation in the KOR gene compared to the WT sequence. Finally, transformation of irx2 plants with a WT copy of the KOR gene fully restores the WT levels of cellulose and normal xylem morphology.

The irx2 mutation specifically affects the deposition of cellulose, with little effect on other cell wall polymers (Turner and Somerville, 1997). The fact that both alleles of irx2 are caused by mutations in the KOR gene strongly supports a role for KOR in cellulose deposition in the secondary cell wall. While previous studies have demonstrated that mutations in kor affect cellulose deposition in the primary cell wall (His et al., 2001; Lane et al., 2001; Sato et al., 2001), it is now clear that the KOR gene is required for cellulose deposition in both primary and secondary cell walls (Turner and Somerville, 1997). This is supported by studies on the kor-1 allele that exhibits a collapsed xylem phenotype consistent with a secondary cell wall defect (Figure 1e,f). Furthermore, expression of the KOR transcript (Figure 4) and high expression of KOR in cells also expressing the secondary cell-wall-specific IRX3 protein (Figure 5) are consistent with a role for KOR in secondary cell wall synthesis, and suggest that levels of KOR are strongly correlated with the rate of cellulose synthesis. This is consistent with the observation that the expression of the cotton orthologue of KOR (Cel3) is high in cotton fibres that are undergoing secondary cell wall cellulose synthesis (Peng et al., 2001).

KOR is predicted to be an β,1-4 EGase, and a B. napus orthologue has been demonstrated to catalyse the hydrolysis of amorphous cellulose in vitro (Molhoj et al., 2001). The lack of a recognisable cellulose binding domain, however, has lead to the suggestion that substrate might be the hemicellulose xyloglucan (Zuo et al., 2000). The hemicellulose in secondary cell wall of Arabidopsis is almost exclusively xylan, and this study supports the idea that in vivo, the substrate for KOR is cellulose rather than any other β,1-4 glucan, such as xyloglucan (Turner and Somerville, 1997).

All previously identified alleles of kor exhibit a primary cell wall defect that is consistent with a decrease in cellulose deposition (His et al., 2001; Lane et al., 2001; Sato et al., 2001; Zuo et al., 2000). Several lines of evidence suggest that neither irx2-1 nor irx2-2 exhibits a primary cell wall defect. irx2 seedlings do not exhibit any cell elongation, cell shape or cell plate formation phenotypes (Table 1, data not shown), which are phenotypes of kor that may be attributed to a primary cell wall defect. The initial measurements of cellulose content of irx2 showed that only tissues with high secondary cell wall content exhibited a large difference (Turner and Somerville, 1997). This result has been confirmed using FTIR analysis, which provides an alternative more sensitive method for measuring cellulose in the cell wall. It is clear, however, that there is no reduction in cellulose content in the primary cell wall (Table 2).

Why the irx2 mutations only appear to reduce the cellulose content in the secondary cell wall remains an intriguing question. One explanation is that irx2 alleles represent relatively weak mutations of KOR and consequently only when cells have very high rates of cellulose synthesis, such as during secondary cell wall formation, does the defect become obvious. If this is the case, then plants must be very sensitive to KOR activity as alleles such as kor-1 and rsw2 have clear primary cell wall defect, but are relatively weak mutations compared to mutations, such as kor-2, that abolish KOR expression (Lane et al., 2001; Sato et al., 2001; Zuo et al., 2000). Both alleles of irx2 result in the amino acid change of proline to leucine. Both of these prolines are highly conserved in all the 25 putative endoglucanase genes found in the Arabidopsis genome (http://afmb.cnrs-mrs.fr/~cazy/CAZY/). 553Pro is found in a particularly well-conserved run of seven amino acids, six of which are found in an analogous position in endoglucanases from a wide range of organisms, including C. fimi, C. stercorarium and T. fusca, in addition to all 25 Arabidopsis genes. The program prosite (http://www.ebi.ac.uk/interpro/) identifies a conserved amino acid motif in all family 9 endoglucanases that contain amino acids known to be important in catalysis. 553Pro lies only five amino acids from this consensus, suggesting that it resides in or is close to the catalytic site. The extensive conservation of these amino acids suggests that their alteration in the irx2 mutants may affect protein activity. It is apparent, however, that if this is the case, there is sufficient residual activity to allow the cells to synthesise cellulose at the rate required by primary cell walls.

An alternate explanation for why irx2-1 and irx2-2 only affect the secondary cell wall could be that KOR is part of a large protein complex. Other single amino changes found in kor alleles, such as rsw2-1, rsw2-4 and rsw2-2/acw1, are predicted to occur on the surface of the protein (Molhoj et al., 2002) and may well affect the ability of KOR to form a complex. It is possible that plants may be more sensitive to mutations that affect the assembly of a protein complex than they are to mutations that reduce the catalytic activity of KOR. Whether this hypothesis is correct should be resolved once the protein complex containing KOR has been purified and characterised.

Despite the increasingly large amount of information that is available on KOR, its function during cellulose synthesis remains to be established. This is further complicated by the lack of definitive evidence proving the subcellular localisation of KOR. KOR is consistently found localised in at least two membrane fractions, one of these being the plasma membrane (Brummell et al., 1997; Nicol et al., 1998). One possibility is that KOR is not targeted to the plasma membrane, but is instead retained in another membrane compartment. In this scenario, the hypothetical topology of the protein would place the predicted active site of the protein within a membrane compartment, allowing it to supply intermediate ‘priming’ glycan molecules to the CesA protein for chain elongation. A recent model has suggested a role for KOR at the plasma membrane, where it might cleave glucan residues from a sitosterol-linked primer that is subsequently incorporated into a growing cellulose microfibril by CesA proteins (Peng et al., 2002). A variety of data have shown that cellulose microfibrils originate from a large plasma membrane-bound complex that is visualised as a rosette, and that CesA proteins are part of this complex and are essential for its proper organisation (Arioli et al., 1998; Kimura et al., 1999). Any protein that functions to provide a substrate for cellulose microfibrils or that is involved in the formation of some kind of intermediate may be expected to associate with the cellulose synthase complex. Using either tagged KOR or IRX3, we are not able to detect any interaction between KOR and the CesA proteins, suggesting that if such an interaction exists, it is either weak or transient. Furthermore, the localisation of KOR is not consistent with a direct role for KOR in the synthesis of cellulose microfibrils. There is only relatively poor localisation of KOR with cortical MTs that mark the site of secondary cell wall deposition (Figure 5). This is in contrast to the localisation found with IRX3, which is believed to be the catalytic subunit of the cellulose synthase complex and which exhibits very good co-localisation with cortical MTs during secondary cell wall formation (Gardiner et al., 2003). In irx2 mutants, secondary cell wall deposition only occurs at the corners of the cells (Turner and Somerville, 1997), which are the earliest sites of secondary cell wall deposition (Altamura et al., 2001). These observations are consistent with KOR being required in the later stages of cellulose deposition in the secondary cell wall. These data are also consistent with the predicted topology of the protein, which, at the plasma membrane, would place the active site on the cell wall side of the plasma membrane. This would allow KOR to have at least two possible roles. It may function in the ‘editing’ of growing microfibrils to maintain proper packing of individual chains. In this case, KOR would remove parts of individual chains that are incorrectly positioned in the growing microfibril. An alternative role for KOR is in the severing of microfibrils such that synthesis of the cellulose chains is terminated.

The data presented here provide evidence to support the activity of KOR in cellulose synthesis in both the primary and the secondary cell walls. Elucidating the role of the KOR in the cellulose biosynthetic mechanism is likely to require the definitive subcellular localisation of the KOR protein, along with the purification of the KOR-containing complex and identification of any interacting proteins.

Experimental procedures

Light microscopy

Stems were sectioned with a razor blade to an approximate thickness of 200 µm and stained in 0.05% toluidine blue for 1 min, rinsed in distilled water and mounted in water.

Etiolated hypocotyl screen

Sterilised seeds were plated onto Murashige and Skoog 1% agar plates containing 4.5 or 1% sucrose, and after 48 h at 4°C, they were transferred to 22°C for 5 days, as described by Nicol et al. (1998).

FTIR analysis of cell walls

Primary cell walls were prepared from whole flowering stems by the homogenisation and wet sieving procedure described by Ha et al. (2002), except that the particle size range of 53–125 µm was collected after wet sieving. This particle size range contained cell wall fragments from cortical parenchyma and epidermal tissues, which were readily distinguished from one another under the FTIR microscope, and from which spectra were separately recorded. The same isolation procedure was used to prepare cell walls from excised rosette leaves. In this case, only the cell wall fragments from epidermal cells were readily identifiable in the 53–125-µm fraction, and FTIR spectra were collected from these.

This cell wall isolation protocol normally also yields a clean vascular (xylem plus interfascicular) cell wall preparation of larger particle size. However, when it was applied to irx2, the >125-µm cell wall fragments remained several cells thick, giving absorbances too high for collection of undistorted spectra, and cortical and epidermal cell walls remained adherent to them. Instead, near-longitudinal hand-sections were prepared from near the mid-point of frozen flowering stems after stripping the epidermis and cortex. The thinnest parts of these sections, containing the tangential walls of xylem and interfascicular tissues, were used for FTIR microscopy after washing with water and air-drying.

FTIR spectra were obtained on a Nicolet Nexus FTIR spectrometer, with Continuum microscope attachment and purpose-built continuous-flow vapour deuteration cell, as described by Ha et al. (2002). Spectral resolution was normally 4 cm−1 with 64–256 scans per spectrum. Initial spectrum handling was in nicolet omnic software. Linear baselines between the extremities of the C–H and O–H stretching bands were subtracted in Excel prior to the calculation of the band areas. Normally, the most appropriate way to quantify the intensity of the hydroxyl stretching band, used to quantify crystalline cellulose, is to ratio its area against the total area of the C–H stretching bands. This procedure corrects for variability in the thickness of the cell wall fragments and estimates cellulose as a proportion of the total carbohydrate present, although the major differences in non-cellulosic polysaccharide composition make it inadvisable to compare primary with secondary cell walls. For the vascular tissue and cortical parenchyma cell walls, 7–25 spectra were normalised on the area of the C–H stretching band (2800–3000 cm−1) before averaging and before statistical analysis (anova) of the area of the O–H stretching band (3150–3600 cm−1).

A different procedure was necessary for epidermal cell walls because cutin hydrocarbons contribute to the total area of the C–H stretching bands, while the fragments were consistently only one cell thick. For the epidermal cell walls, we recorded the absolute hydroxyl band intensity without normalisation. In these cases, the data correspond to mass of cellulose per unit area of epidermal cell walls.

Mapping of irx2

A test-cross population was generated by crossing the mutant line, SRT228-4, isolated in a Ler background, with Columbia gl1, as previously described by Turner and Somerville (1997).

Test-cross plants, aged approximately 5 weeks, were scored phenotypically for the irx2 mutation by visual assessment, using light microscopy, and for recombination events with microsatellite markers, using isolated DNA, as described previously by Turner and Somerville (1997). Successful microsatellite polymorphisms were identified through a screen of BAC clones spanning the region between genetic markers LFY3 and DFR on chromosome V (A. thaliana genomic sequence, National Centre for Biotechnology Information (NCBI); http://www.ncbi.org). Polymorphic dinucleotide repeats >18 bp were amplified as part of a 80–250-bp PCR product, which were screened on an agarose electrophoresis gel.

RT-PCR of KOR cDNA

Stem RNA was isolated from 6-week-old plants using an Rneasy Plant Mini Kit (Qiagen, Crawley, West Sussex, UK). First-strand cDNA was synthesised from 20 ng stem RNA using a gene-specific reverse primer and 200 U reverse transcriptase (Promega, Southampton, UK), according to the manufacturer's instructions, in a total volume of 25 µl at 42°C for 60 min. This first-strand product was then used in a PCR and cloned into pGEM-T Easy (Promega) for sequencing. Two independent clones were isolated and sequenced to negate the effect of mis-incorporation of nucleotides by Taq DNA polymerase.

DNA sequencing

Plasmid templates purified using QIAprep spin miniprep kits (Qiagen) were primed with gene-specific primers of high-purity salt-free grade (MWG Biotech UK; Milton Keynes, UK) and sequenced automatically using ABI PRISM Big Dye Terminators (Applied Biosystems Inc., Foster City, CA, USA). DNA sequences were analysed using vector nti software (Informax, Oxford, UK).

RNA gel blot analysis

Total RNA was isolated from 6-week-old plants using an RNeasy Plant Mini Kit (Qiagen). Five micrograms of electrophoresed RNA was transferred onto Hybond N+ membrane (Amersham Pharmacia Biotech UK Ltd., Chaffont, UK) and probed for signals from COMT (Arabidopsis Biological Resource Centre, Columbus, OH, USA; stock centre clone 115 N5, EcoRI–HindIII 1.5-kb fragment) and KOR (Arabidopsis Biological Resource Centre, clone TAC K2I5, PCR-product-amplified using the following primers: 5′-GTACGGAAGAGATCCATGGGG-3′ and 5′-CACGGGCCTTTTATAGTCCAT-3′). After 30 cycles of 96°C for 30 sec, 59°C for 17 sec and 72°C for 40 sec, the 925-bp product was purified using High Pure PCR Product Purification column (Roche Diagnostics, Lewes, UK).

Complementation of irx2

A 5-kb StuI fragment from TAC K2I5 carrying the WT KOR gene was cloned into pC2300. An epitope-tagged KOR (NHisKOR) was constructed by annealing two complementary primers (5′-GAGGGGATCCCATCACCATCACCATCACCC-3′ and 5′-GGGTGATGGTGATGGTGATGGGATCCCCTC-3′) and ligating the annealed product into a unique MamI restriction site. irx2 mutant plants were transformed using A. tumefaciens (GV3101) carrying these constructs (Bent and Clough, 1998). Primary transformants (T1) were selected by plating sterilised T1 seeds on Murashige and Skoog 1% agar plates containing 50 µg ml−1 kanamycin sulphate. After 3 weeks, the kanamycin-resistant plants were transplanted into pots containing commercial soil/peat/perlite mixture. Stems from mature T1 plants, together with equivalent specimens of Ler WT and irx2, were sectioned and stained with toluidine blue.

Protein purification

One gram of stem from homozygous T2 NHisKOR plants was ground well in lysis buffer (50 mm KH2PO4 (pH 7.0) and 300 mm NaCl) containing 10 mm imidazole. After clarification by centrifugation, NP-40 was added to a final concentration of 1%. Next, 50 µl of TALON resin (Clontech Laboratories Inc., San Diego, USA) was added to these solubilised extracts, which were mixed end-over-end for 60 min. After centrifugation, the resin was washed three times with 250 µl of lysis buffer containing 20 mm imidazole. Proteins were eluted from the resin three times with 30 µl lysis buffer containing 250 mm imidazole. The entire purification procedure was performed at 4°C in the presence of protease inhibitors (protease inhibitor cocktail for mammalian cell extracts; Sigma, Poole, Dorset, UK). Fifteen-microlitre aliquots were denatured in loading buffer for 60 min at 37°C before electrophoresis through 10% SDS–polyacrylamide gels.

Epitope-tagged IRX3 was purified as described by Taylor et al. (2000) except that the lysis buffer consisted of 50 mm NaCl and 50 mm sodium phosphate buffer (pH 6.5).

Immunological techniques

After transfer to immuno-blot polyvinylidene difluoride membrane (Bio-Rad, Hercules, CA, USA), protein gel blots were performed according to standard protocols (Harlow and Lane, 1998). A rabbit polyclonal antibody was raised to and affinity-purified against the peptide MYGRDPWGGPLEINT by Eurogentec (Seraing, Belgium). This antibody was used for detecting KOR in Western blots (Figure 6b). Figures 5 and 6(a) used a previously described antibody to KOR (Nicol et al., 1998). Epitope-tagged KOR was detected by using an anti-RGSHis monoclonal antibody (Qiagen); IRX3 was detected using an anti-IRX3 antibody (Taylor et al., 2000).

Tissue prints were carried out as described previously by Taylor et al. (2000, 2003). Immunofluorescent staining of Arabidopsis roots was carried out essentially as described by Harper et al. (1996), except for the digestion step, where we used 0.5% w/v pectolyase and 0.5% w/v cellulase for 25 min. Prepared seedlings on slides were incubated with 1/2000 monoclonal antibody B512 against α-tubulin or anti-KOR, secondary antibodies were sheep–antimouse–FITC (Fluorescein isothiocyanate isomer 1), and 1/80 goat–antirabbit–FITC. Where more than one secondary antibody was required, they were added sequentially so as to avoid cross-reaction. Samples were viewed using a Nikon Eclipse TE300 fluorescence microscope equipped with a 40× Nikon Plan Fluor oil-immersion lens and a Bio-Rad MRC1024 MP confocal scanning head with krypton/argon laser. Excitation lines of 488 and 568 nm were used to view FITC and Alexa Fluor 568, respectively.

Acknowledgements

We thank Herman Hofte for the gift of the anti-KOR antibody, NASC for supplying prc and kor seeds and the BBSRC for funding.

Ancillary