The eli1 mutation reveals a link between cell expansion and secondary cell wall formation in Arabidopsis thaliana

The formation and differentiation of the cell wall plays a key role in plant morphogenesis. Two general types of plant cell walls can be distinguished: a thin primary wall which is synthesized during cell expansion and is capable of yielding to turgor pressure (Cosgrove, 1993), and a secondary, thicker wall deposited in the fully expanded cell between the primary wall and the plasma membrane. Secondary wall formation is restricted to specialized cells and provides mechanical strength and rigidity to support aerial structures and hydrophobicity for transport functions.

Several models of the organization of the plant cell wall have described the arrangement of the primary wall components and their structural modification during cell expansion (reviewed by Carpita and Gibeaut, 1993; McCann and Roberts, 1994; Cosgrove, 1999). The plant primary cell wall is mainly composed of cellulose microfibrils and a polysaccharide matrix of hemicellulose, pectins and proteins. The organization of cellulose microfibrils and their complex interaction with other cell wall components results in an extensive and dynamic network that can be modified by the action of several cell-wall enzymes, such as endoglucanases, xyloglucan-endotransglycoxylases and expansins (reviewed by Cosgrove, 1999). These proteins contribute to the continuous breaking and remaking of bonds necessary to maintain the integrity of the cell wall during cell expansion. The isolation and characterization of genes involved in the synthesis of major cell wall polysaccharides (Pear et al., 1996; Bonin et al., 1997; Turner and Somerville, 1997; Taylor et al., 1999; Arioli et al., 1998; Nicol et al., 1998) is starting to reveal the mechanisms of primary cell wall formation.

The formation of secondary cell walls has been studied most intensively during the development of xylem cells. This process is characterized by deposition of lignin into a matrix of polysaccharides (Roberts, 1989), and subsequent loss of cellular contents (Wardrop, 1981; O’Brien, 1981). The expression of specific genes associated with different stages of xylogenesis has been catalogued in differentiating Zinnia mesophyll cells. For example, genes encoding proteins involved in remodeling the endomembrane system in preparation for secondary thickening of the longitudinal walls, in the synthesis of lignin precursors and in lysis of cell contents have been identified (Fukuda et al., 1994).

The mechanisms governing the spatial and temporal patterns of secondary wall formation are beginning to be understood. Mutants defective in aspects of vascular development have been identified in Arabidopsis, for example the wooden leg (wol) mutant (Scheres et al., 1995), which fails to establish the appropriate numbers of vascular cells, and lion’s tail (lit),which is impaired in the radial expansion of vascular cells (Hauser et al., 1995). Mutants altered in different aspects of polysaccharide synthesis and secondary wall deposition have also been identified. An interfascicular fiberless mutant (ifl1) is disrupted in normal development of interfascicular fibres in Arabidopsis, leading to flaccid, mechanically weak stems (Zhong et al., 1997). The IFL1 gene encodes a homeodomain-leucine-zipper protein (Zhong and Ye, 1999). The irregular xylem (irx) mutants are characterized by a reduction in cellulose in stem tissue, primarily associated with xylem cells (Turner and Somerville, 1997). IRX3 encodes a catalytic subunit of a cellulose synthase complex specifically required for the synthesis of cellulose in the secondary cell wall (Taylor et al., 1999).

We have screened for mutants exhibiting abnormal patterns of lignification in the primary root of Arabidopsis. We report the isolation and characterization of one of these mutants, termed eli1 (ectopic lignification 1) and describe a role for the ELI1 gene in integrating cell expansion and secondary cell wall formation.

The Arabidopsis thaliana ecotype Columbia (Col) was used for mutant isolation. The lion’s tail (lit), cobra and pom-pom Col seeds (Hauser et al., 1995) were provided by Maria-Theres Hauser, wol Col seeds (Scheres et al., 1995) by Philip Benfey, kor1 WS seeds (Nicol et al., 1998) were provided by Herman Höfte, rsw1 Col seeds (Baskin et al., 1995) by Keith Roberts, and det3 Col seeds (Cabrera y Poch et al., 1993) by Joanne Chory.

Seeds were surface sterilized in 5% sodium hypochlorite, rinsed with sterile water and plated in 1 MS salts (Murashige and Skoog,) containing 30 mM or 100 mM sucrose, 0.3% phytagel (pH 5.8), and imbibed at 4°C for 48 hours in darkness. Seedlings were grown in a vertical position in continuous light (80 μeinsteins/m²/second) at 21°C. After 7 days plants were transferred to soil and grown in a greenhouse. rsw1 seeds were grown at 21°C for 5 days and transferred to 31°C for a further 2 days to express the conditional phenotype.

Seedlings were grown as described above on media supplemented with different concentrations of indole-3-acetic acid (IAA) and kinetin (0.5 nM, 50 nM, 0.5 μM and 5 μM), epi-brassinosolide (1 μM) and gibberellin GA4 (10⁻⁴ M and 10⁻⁵ M). For ethylene response assays seedlings were germinated on either 10⁻⁴ M ACC or 10⁻⁴ M silver nitrate. Root phenotypes were analysed 7 days after germination.

The Landsberg erecta (Ler) ecotype was used for mapping.

Col seeds were treated with 0.5% ethyl methane sulphonate (EMS) in water for 8 hours at 22°C, and sown on soil for growth in the greenhouse. Single siliques representing a total of 4,000 M₂ families were harvested. An additional 1800 families containing multiple copies of an autonomous En transposon (Wisman et al., 1998) were used in the same screen. A total of 87,000 M₂ seeds were plated in 4,000 individual families and grown on vertical plates for 7 days before screening. The screen for mutants with altered lignin distribution was carried out in seedlings in chloral hydrate:glycerol:distilled water (8:1:2) (Berleth and Jürgens, 1993). Cleared seedlings were stained and mounted with a saturated solution of phloroglucinol:HCl and viewed under dark-field illumination. Siblings from segregating populations were kept and their progeny re-tested in the M₃ generation.

The eli1 mutant was back-crossed twice to wild-type Columbia, and progeny of the final back-cross were used for subsequent experiments. The F₁ progeny of crosses between all three alleles of eli1 and Col wild type were phenotypically wild type, while the F₂ progeny from self-pollinated F₁ showed a segregation ratio of 3:1 for wild-type:mutant phenotype for the three isolated alleles of eli1 (251 ELI1-1/+, 79 eli1-1/eli1-1, χ²=2.4, P>0.05; 301 ELI1-2/+, 87 eli1-2/eli1-2; χ²=1.37, P>0.05; and 287 ELI1-3/+, 92 eli1-3/eli1-3; χ²=0.09, P>0.05), indicating that eli1 is a single locus, nuclear recessive mutation. Double mutants between eli1 and lit, and eli1 and wol mutants were obtained by crossing plants homozygous for the mutations and screening the F₂ progeny.

The eli1 mutant was crossed to Landsberg erecta (Ler) and segregating F₂ populations were prepared for mapping with simple sequence length polymorphism (SSLP) markers as described by Bell and Ecker (1994). Each marker was tested on the DNA of 90 single eli1 seedlings from the segregating F₂ of these crosses.

Transverse sections (2 μm) of the primary root were made in Historesin-embedded material as described by Schneider et al. (1997) and stained with 0.05% Toluidine Blue O (Sigma) for 30 seconds. Inflorescence stem sections were cut by hand and stained with 0.05% Toluidine Blue O for 1 minute, rinsed and mounted in water. Cell wall material was stained with 0.001% Calcofluor in semi-thin sections.

Roots of 7-day-old seedlings were cleared with methanol for 1 hour at room temperature, and then with 10% NaOH at 60°C for 12 hours, and stained with 0.01% basic Fuchsin to visualize lignin (Dharmawardhana et al., 1992). Optical sections in the longitudinal plane were made using a Bio-Rad MRC-1024 Confocal Scanning Laser Microscope (CSLM). To image all xylem vessels, Z-series were captured and stacked using the NIH-image programme.

Seven-day-old seedlings were mounted on the surface of an aluminium stub with OCT compound, immersed into liquid nitrogen slush, and put onto the cold stage of a CT1500HF cryotransfer system attached to a Phillips XL30 FEG scanning electron microscope. Sublimation was performed at −95°C for 3 minutes before sputter-coating the sample with platinum for approximately 2 minutes at 10 mA.

For transmission electron microscopy (TEM), 7-day-old root tissues were fixed in 4.5% formaldehyde, dehydrated in an ethanol series, and infiltrated in LR White resin as described by Wells (1985). Rabbit polyclonal antibodies against lignin dehydrogenation polymer (GSzt L4, mixed anti-guaiacyl-syringyl lignin) were used (Joseleau and Ruel, 1997). Ultrathin tissue sections were blocked with 1% BSA and Tween 20, and treated with a 1/30 dilution of the primary antibody in 0.1% BSA and incubated for 3 hours. Sections were washed three times with PBS pH 7.4 for 10 minutes before incubation for an hour with the secondary antibody (10 nm gold-goat anti-rabbit, BioCell) diluted in 0.1% BSA. Sections were then post-fixed in 1% glutaraldehyde, stained with 2% uranyl acetate and alkaline lead citrate, and observed on a JEOL 1200 electron microscope. Semi-thin sections were used for immunofluorescence detection of lignin with a secondary cy3-labelled goat anti-rabbit antibody diluted 1/100 in 0.1% BSA. Sections were incubated for 3 hours. Pre-immune antisera and the omission of the primary antibody were used as immunochemical controls.

Images were composed using Adobe® Photoshop 5.0.

A total of 4,000 individual M₂ EMS families and 1800 En transposon insertion lines were screened for altered patterns of lignification. In wild-type tissues, secondary thickened xylem cells are lignified, therefore the pattern of phloroglucinol staining can reveal alterations in secondary wall deposition and alterations in xylem development.

Three groups of mutants were identified on the basis of patterns of phloroglucinol staining in the primary root. mux (multiple xylem) mutants exhibited an increased number of xylem strands in the primary root, tpx mutants displayed an altered timing of protoxylem development in the primary root, and eli (ectopic lignification) mutants exhibited an altered pattern of lignification. The mux and tpx mutants isolated were the result of monogenic recessive mutations. All the mutants with altered xylem development resulted in a shortened root system.

Three ectopic lignification mutant lines, two from the EMS mutagenized population and one from the En transposon lines, were identified. They exhibited essentially the same phenotype, with a characteristic lignification pattern in the primary root. The three mutants belong to the same complementation group, therefore they were designated eli1-1 and eli1-2 (the EMS mutagenized lines alleles) and eli1-3 (the En transposon insertion line allele). The first line identified, eli1-1, was used for all the subsequent studies. Using a population of 90 F₂eli1-1 seedlings in crosses with wild-type Ler, the eli1 locus was found to map to the short arm of chromosome 5, between CTR1 and the SSLP marker nga249 (Bell and Ecker, 1994).

Light-grown 7-day-old eli1 seedlings exhibited a primary root approximately 3 times shorter and thicker than the wild type (Table 1; Fig. 1A). Scanning electron microscopy (SEM) of the primary root revealed that epidermal cells of eli1 (Fig. 1E) were shorter than wild-type cells (Fig. 1D), and they exhibited irregular, often bulbous protrusions. The eli1 primary root had a reduced elongation zone with shorter cells, evident by the early appearance of root hairs in the differentiation zone (Fig. 1E). eli1 lateral roots showed a similar shortened phenotype to the primary root, and fewer lateral roots were formed compared to wild type (data not shown). eli1 seedlings stained with phloroglucinol exhibited an abnormal lignification pattern, with a higher than normal staining of lignin in the xylem strands, and a patchy mosaic of intense lignin staining in non-vascular cells (Fig. 1B). This staining pattern was observed in other tissues, but was most intense in the root.

To determine the initial appearance of ectopic lignification, different developmental stages of eli1 and wild-type embryos were stained with phloroglucinol and observed by light microscopy. No lignin staining was detected in embryos of wild-type and eli1 plants (data not shown), indicating that the defect in eli1 occurs in postembryonic tissues.

Since ectopic lignification was associated with reduced cell elongation, the eli1 mutant was subjected to conditions that increased cell elongation in wild-type seedlings. Seven-day-old dark-grown eli1 seedlings displayed a shorter hypocotyl and roots with shorter cells than wild type grown under the same conditions (Table 1; Fig. 1C).

eli1 seedlings developed into mature plants which had a stunted aerial phenotype with reduced apical dominance, small leaves, siliques and flowers (Fig. 2A,B), suggesting that the cell expansion defect in eli1 occurs in several tissues and organs. The reduced stamen elongation observed may account for the decreased fertility of eli1 plants.

Transverse sections of the primary root of Col and eli1 seedlings revealed an abnormal cellular organization in eli1 mutant plants (Fig. 3). All the cell layers of the eli1 primary root were present, but radial cell expansion was increased (Fig. 3B) compared to the wild type (Fig. 3A). In the stele the number of cells differentiating into tracheids was higher in eli1 (Fig. 3G,H) than in wild type (Fig. 3A,B). The number and position of the additional xylem elements in eli1 changed from one section to another (as shown in Fig. 3E and F compared with 3G and H) suggesting that they were isolated tracheids that failed to form files of continuous cells.

Lignification in the primary root of eli1 is not restricted to the xylem cells, as in wild-type plants, but also occurred in other vascular, pericycle, cortical or endodermal cells, as indicated by arrowheads in Figs 3F and 3H. This ectopic lignification appeared as amorphous patches of differing intensities of lignin staining, in contrast to the structured pattern of lignification observed in both wild-type and eli1 tracheids. Ectopically lignified cells also stained more intensely with Toluidine Blue (Fig. 3E,G) as did the secondary wall thickenings of tracheids of eli1 and wild-type plants (Fig. 3C).

To investigate the eli1 phenotype in other plant organs, sections from the basal part of 6-week-old inflorescence stems were observed. The diameter of the eli1 inflorescence stem sections (Fig. 2G) was reduced to about two-thirds of the wild-type stem diameter (Fig. 2D). Toluidine blue staining of inflorescence stem sections showed an increased number of tracheids and disruption of xylem strands in eli1 (compare Fig. 2E, wild type, with 2H, eli1). In addition, the pith parenchyma cells and some cortex cells contained unusual cell structures, revealed by Toluidine Blue staining, that also stained for lignin, that were not found in wild type (Fig. 2G and 2H, eli1). Longitudinal sections of the eli1 inflorescence stem revealed that lignin occurred in a mosaic surrounding small groups of cortical and pith cells (compare Fig. 2C, wild type, with Fig. 2F, eli1). This was also observed in the primary roots of eli1 seedlings (Fig. 1B) and other tissues of eli1 plants (data not shown).

Calcofluor staining of eli1 revealed an altered distribution of cellulosic material in the cell walls of the mutant compared with wild type. Cell apices exhibited a higher intensity of fluorescence (indicated by arrowheads in Fig. 8F) compared to the uniform staining observed in the wild type (Fig. 8C).

Confocal Scanning Laser Microscopy (CSLM) on the primary root of 7-day-old seedlings stained with basic fuchsin was carried out in order to determine the longitudinal organization of primary xylem in the eli1 mutant (Fig. 4). In wild-type plants, the transition from early stages of differentiation (Fig. 4C) to later stages of differentiation (Fig. 4B) is defined by the appearance of metaxylem, shown in Fig. 4A. Projections of the CSLM optical sections of the xylem cell files revealed the alignment of xylem cells in eli1 was severely altered (Fig. 4D,E), in that a number of individually lignified cells formed out of the xylem axis (Fig. 4E). While exhibiting the normal thickening of Arabidopsis metaxylem, these ectopic tracheids in eli1 formed neither files nor perforation plates. In addition, the protoxylem cell files had irregular complex thickenings (Fig. 4E). The misshapen xylem cells in eli1 were more isometric than their wild-type counterparts, with increased diameters and decreased length compared to the wild type (Fig. 4B,C). The Casparian strip seen in the wild type (Fig. 4B,C) was replaced in eli1 by more strongly fluorescing amorphous deposits of lignin (Fig. 4D,E, arrowed). UV autofluorescence detection of lignin confirmed the lignification patterns observed in thin sections stained with Toluidine Blue, and in whole mounts stained with phloroglucinol.

The ultrastructure of eli1 and wild-type cell walls in transverse sections of roots of 7-day-old seedlings was investigated. Transmission electron microscopy (TEM) revealed substantial differences in the cell walls of different cell types in the eli1 mutant (Fig. 5). In wild type, the secondary cell walls of the differentiated xylem cells were characterised by dense uranyl acetate staining (Fig. 5A). Sections of eli1 primary roots revealed the presence of similar densely staining structures inside some non-vascular cells (Fig. 5B), but these lacked the regular striated appearance of normal secondary thickenings. To define the subcellular localisation of lignin, immunolocalization with an anti-lignin antibody, GSzt L4, which binds specifically to guaicyl and syringyl residues in maize lignin (Joseleau and Ruel, 1997), was carried out. Discrete localization of the anti-lignin antibodies to the xylem axis in wild-type roots was observed (Fig. 6B) indicating that the antibody was suitable for specifically localizing lignin in Arabidopsis. Preimmune sera did not show specific labelling (Fig. 6F). The pattern of antibody labelling in sections of eli1 mutant roots revealed several non-xylem cells and other structures that contained large amounts of lignin (Fig. 6D). TEM and immunogold labelling revealed lignin was specifically localized to the secondary wall thickenings of proto-and metaxylem cells (electron-dense particles in Fig. 5C). In eli1 roots the lignin antibody was localized to the ectopic secondary thickenings found in endodermal cells for example (Fig. 5F). In contrast, the anti-lignin antibody did not react with cell walls of an endodermal cell from wild-type roots (Fig. 5E).

A number of cell expansion mutants have been described (Baskin et al., 1995; Hauser et al., 1995) which have root phenotypes related to eli1. Phloroglucinol staining of lit roots revealed an abnormal pattern of lignification similar to that seen in the primary root of eli1, but localised mainly to the root tip region (Fig. 8H). Genetic interactions between eli1 and lit were studied to reveal relationships between these two loci. Crosses between eli1 and lit mutants yielded 100% wild-type F₁ progeny and self-pollination of this F₁ progeny gave an F₂ progeny segregating in a 9:3:3:1 ratio (233 wild type: 92 lit:85 eli1: 27eli1lit; χ² =2.52; P>0.05) for wt: eli1: lit: eli1 lit phenotypes. The F₂ segregating families of crosses between eli1 and lit were grown for 7 days in media containing 30 mM sucrose, to distinguish the lit and eli1 root phenotypes (Fig. 7A). The double eli1 lit mutant was characterized by a more extreme whole plant phenotype, with a dwarfed seedling that ultimately died at 3 weeks, and shorter fatter roots than the parental mutants. A highly compressed cell morphology was also observed in roots (Fig. 7A). Phloroglucinol staining of the eli1 lit double mutant (Fig. 7C) showed more extensive ectopic lignification and disruption of the vascular system than was observed in either single mutant. This was confirmed by CLSM of the primary root, which revealed severely disrupted xylem formation, with many additional single tracheids and disrupted tracheid cell files of nearly isometric cells (Fig. 7G). SEM analysis of hypocotyl epidermal cells also showed nearly isometric cells often with bulbous protrusions (arrow in Fig. 7B). These results indicate a possible additive effect of the eli1 and lit mutations on cell expansion, xylem organization and ectopic lignification, and the non-viability of the double mutant revealed the importance of both genes for plant development.

The wol mutant is characterised by a narrower vascular cylinder composed mainly of protoxylem cell files, with no apparent metaxylem or phloem (Scheres et al., 1995). The primary defect in wol is thought to be the absence of a specific cell division in the stele during embryogenesis, leading to fewer vascular initials. Potential genetic interactions between wol and eli1 were studied in double mutants (Fig. 7D). Analysis of roots from the F₂ segregation of the crosses showed 310 wt: 116 wol: 119 eli1: 22 eli1wol χ² =4; P>0.05) corresponding to a 9:3:3:1 ratio.

Seven-day-old eli1wol double mutants revealed a seedling phenotype similar to the wol single mutant (Fig. 7E). CSLM analysis of the eli1 wol double mutant in 7-day-old seedlings (Fig. 7I) revealed only a few protoxylem strands in the narrow vascular cylinder, characteristic of wol (Fig. 7H). eli1 wol seedlings had reduced ectopic lignification compared to eli1 (Fig. 7I, compare to Fig. 4D), and nearly normal cell expansion compared to the eli1 mutant.

The lit phenotype is conditional upon sucrose levels (Hauser et al., 1995), therefore the phenotype of eli1 seedlings grown in media supplemented with 30 mM or 100 mM sucrose was assessed (Table 1). Growth on different sucrose concentrations had no discernable effects on either cell elongation or lignification in eli1 seedlings. In contrast, growth of lit seedlings on higher sucrose levels caused reduced cell expansion and increased ectopic lignification. In addition, dark grown lit hypocotyls were longer than dark grown eli1 hypocotyls and did not exhibit ectopic lignification (Table 1). This suggested a possible dependence of ectopic lignification upon cell expansion. To test this, several mutants with reduced cell expansion phenotypes were assessed for lignin content and altered cell wall composition. The rsw1 mutant (Baskin et al., 1995; Arioli et al., 1998) is a temperature sensitive mutant of a catalytic subunit of cellulose synthase. At a restrictive temperature of 31°C rsw1 exhibits radial swelling of the root caused by misshapen and often smaller cells, as shown in Fig. 8J,K L. Fig. 8K shows phloroglucinol-stained roots of rsw1, and several ectopically lignified cells are visible. The extent of ectopic lignification is not as high as that seen in eli1 (Fig. 8E) or in lit (Fig. 8H). In addition to phloroglucinol staining, thin sections of roots of selected cell expansion mutants were stained with Calcofluor to reveal possible differences in cell wall composition. As shown previously in Figs 6C and 8F, eli1 exhibited patches stained brightly by Calcofluor, and these appear to are associated with secondary-thickened walls as shown in sections of wild-type roots (Figs 6A, 8C). Fig. 8L shows a Calcofluor-stained section of an rsw1 root. Bright fluorescence is seen in the apices of junctions between cell walls, a pattern not seen in wild type (Fig. 8C). The korrigan1 (kor1) mutation alters cell expansion in roots and hypocotyls, and is defective in an endo-1,4-β-D-glucanase (EGase) (Nicol et al., 1998). Calcofluor staining of thin sections of kor1 roots revealed a patchy pattern similar to that observed previously (Fig. 8O; Nicol et al., 1998), and similar to that observed in eli1 and rsw1. Calcofluor staining of lit revealed less cellulosic material in the cell walls (Fig. 8I). kor1 also exhibited phloroglucinol staining of non-vascular cells (Fig. 8N, compared to wild type in 8B). Finally, light-grown det3 plants exhibit reduced cell elongation and altered cell morphology due to reduced levels of a vacuolar ATPase subunit (Schumacher et al., 1999). Phloroglucinol staining of det3 roots revealed extensive ecotopic lignification (compare Fig. 8S with wild-type roots in 8B) and Toluidine Blue-stained sections of det3 inflorescence stems revealed extensive lignification of cortical and pith cells (dark blue staining in Fig. 8R). These experiments demonstrated an association of ectopic lignification in several cell expansion mutants which are deficient in both structural components of the cell wall (cellulose synthase and an endoglucanase) and in a V-ATPase that is predicted to promote solute uptake into the vacuole and thus drive cell expansion. This suggested a linkage between cell expansion and the initiation of secondary cell wall formation and subsequent lignification.

The control of cell wall formation in higher plants plays a central role in the processes of cellular differentiation and morphogenesis. Cell walls are dynamic structures capable of expansion during cell growth, and become fixed into a final shape upon the cessation of growth (reviewed by Carpita and Gibeaut, 1993). Often the primary cell wall is further elaborated by the deposition of a secondary wall to provide additional mechanical strength and water resistance. The processes involved in determining the final shape of plant cells and subsequent secondary wall formation must be subject to multiple levels of regulation, because the processes that modify cell walls after cessation of growth are essentially irreversible. Little is known about how cell expansion is integrated with secondary wall formation.

We have identified a genetic locus, ELI1, that when mutated results in the production of large amounts of lignin in cells that do not normally lignify. We discuss the phenotypes associated with three independent alleles at the eli1 locus, reveal that several other cell expansion mutants also exhibit aberrant secondary wall formation, interpret the phenotypes resulting from genetic interaction studies, and describe a possible role for ELI1 in controlling secondary wall formation in Arabidopsis.

Phloroglucinol staining revealed extensive lignin deposition in many different cell types in the root, hypocotyl and inflorescence stem of eli1 mutants. Unlike the sculpted patterns of lignin deposition in secondary thickenings of tracheids and fibres, the ectopic lignin characteristic of eli1 plants occurred in amorphous patches that were not associated with specific groups of cells. Immuno-gold electron microscopy using a lignin-specific antibody identified lignin in large electron-dense bodies, typical of normal secondary thickening, in non-vascular cells of eli1, and in the secondary thickenings of xylem cells in both eli1 and wild-type roots. Calcofluor staining of the cell walls of eli1 seedlings revealed additional cellulosic deposits in non-vascular cells as well as in secondary-thickened xylem cells (compare Figs 6A with 6C, and 8C with 8F), indicating that not only additional lignin but other cell wall components are deposited aberrantly in eli1 seedlings. Electron microscopy revealed that these deposits are remarkably similar in structure (although distributed differently in cells as described above) to normal secondary thickening in xylem cells, supporting the interpretation that secondary wall formation, and not specifically lignin synthesis and deposition, is de-regulated in eli1 plants.

In cells destined to form xylem, the same cell expansion defect is manifest in eli1 and lit plants, and in addition tracheids are more heavily lignified, cell files are disrupted and there are more tracheid cells. These effects appear additive in the eli1 lit double mutant, with severely disrupted xylem and many near-isodiametric tracheids, leading to death several weeks after germination. These results suggested that the ELI1 gene product may also be involved in specifying xylem vessel development.

To test this hypothesis, interactions between eli1 and wol, a mutation that alters xylem specification, were studied. The primary defect in wol plants is thought to be the absence of specific cell divisions during axis formation (Scheres et al., 1995) resulting in the loss of metaxylem and phloem cells in the primary root. The eli1 wol double mutant phenotype is of near-normal cell expansion, continuous xylem files, absence of metaxylem and reduced ectopic lignification. This suggests that ELI1 and WOL gene products may interact additively with respect to cell elongation and ectopic lignification, because of near-normal cell elongation and reduced ectopic lignification. This supports the interpretation derived from eli1 lit double mutants, that a primary defect in eli1 is in cell expansion, as the near-wild-type cell expansion seen in the eli1 wol double mutant leads to reduced ectopic lignification of non-xylem cells. The continuous xylem strands typical of wol in the double mutant indicates that wol is epistatic to eli1 with respect to this phenotype. This may indicate a separate role for ELI1 in promoting formation of continuous xylem strands during tracheid formation. An alternative explanation of this epistasis is that the defect in wol plants, the absence of progenitor metaxylem cells, occurs earlier in development than the defect in eli1 plants.

Another cell expansion mutant, lit, also displays ectopic lignification. This phenotype is associated with the extent of cell expansion in lit, shown by the dependent of cell expansion and lignification on external sucrose levels. Analysis of double eli1 lit mutants revealed an additive phenotype of significantly shorter, almost isodiametric cells with increased ectopic lignification, indicating that ELI1 and LIT may act in parallel in at least two independent mechanisms governing normal cell expansion. These observations strongly suggest that reduced cell expansion in eli1 leads to ectopic secondary thickening and lignification. In this model the ELI1 gene product acts to promote cell size increase and inhibit secondary thickening and subsequent lignification in non-xylem cells. The observation that ELI1 acts during later stages of development, such as in the elongation zone of the primary root, is consistent with this proposed role of the ELI1 gene product in normal cell expansion.

Ectopic lignification was also observed in rsw1, defective in a catalytic subunit of cellulase synthase (Arioli et al, 1998), in korrigan, defective in a plasma membrane associated β-1,4 endoglucanase (Nicol et al., 1998), and in det3 (Schumacher et al., 1999), with reduced vacuolar ATPase levels. The varying degree of ectopic lignification observed in these mutants was correlated with the degree of altered cell expansion, suggesting a meaningful link between these processes. This linkage was clearly demonstrated by the temperature dependence of both cell expansion and ectopic lignification in the rsw1 allele.

However, ectopic lignification is not an inevitable consequence of reduced cell expansion, because several cell expansion mutants such as pom-pom and cobra (Hauser et al., 1995) do not exhibit ectopic lignification (unpublished data). Therefore specific features of mutants such as eli1, lit, rsw1 and korrigan lead to the induction of inappropriate secondary wall formation and lignification.

Mutations in the cellulose synthase gene RSW1 (Arioli et al., 1998) cause severe alterations in cell shape due to disassembly of cellulose synthase complexes and consequent accumulation of non-crystalline cellulose in the primary cell wall. The β-1,4 endoglucanase encoded by korrigan is required for the correct assembly of the cellulose-hemicellulose network in expanding primary walls. The V-ATPase encoded by DET3 may regulate cell expansion by controlling solute uptake into the vacuole that drives osmosis, the motive force for cell expansion. The additive effect of lit and eli1 indicate there are at least two independent mechanisms modulating the attainment of normal cell volume, consistent with the need for multiple components for normal cell expansion described above. These postulated regulatory mechanisms require the components and forces driving expansion of the cell wall to be monitored and integrated to maintain the correct geometry of the primary wall. This study demonstrates that this postulated mechanism(s) may also signal differentiation pathways such as lignification, which is normally only initiated in specific cells that have reached a defined shape and size, such as tracheids. Brassinosteroids regulate korrigan transcription (Nicol et al., 1998) and have been postulated to affect V-ATPase function (Schumacher et al., 1999), and also promote secondary wall formation in differentiating tracheids (Yamamoto et al., 1997). They are therefore candidates for integrating cell size and differentiation. However, initial experiments have not yet demonstrated a dependence of cell size and ectopic lignin formation on exogenous brassinosteroids or other growth regulators such as auxin, gibberellins, ethylene or cytokinins in eli1 and other expansion mutants (data not shown). Consequently, the identification of the ELI1 gene and determination of its function will provide important new insights into the mechanisms integrating morphogenesis and differentiation.

We are grateful to Marie-Theres Hauser, Philip Benfey, Herman Höfte, and Joanne Chory for providing mutant lines, Katia Ruel and Jean-Paul Joseleau for providing anti-lignin antibodies, and Brian Wells, Pablo González-Melendi and Kim Findlay for their valuable help with microscopy. We thank Liam Dolan for comments on the manuscript, advice and encouragement. A. I. C.-D. was supported by a PhD Studentship from the John Innes Foundation.