• The Arabidopsis thaliana mutants de-etiolated3 (det3), pom-pom1 (pom1) and ectopic lignification1 (eli1) all deposit lignins in cells where these polymers would not normally be found. Comparison of these mutants provides an opportunity to determine if the shared mutant phenotype arose by perturbing a common regulatory mechanism in each of the mutants.
• The mutants were compared using a combination of genetics, histochemistry, chemical profiling, transcript profiling using both Northern blots and microarrays, and bioinformatics.
• The subset of cells that ectopically lignified was shared between all three mutants, but clear differences in cell wall chemistry were evident between the mutants. Northern blot analysis of lignin biosynthetic genes over diurnal and circadian cycles revealed that transcript abundance of several key genes was clearly altered in all three mutants. Microarray analysis suggests that changes in the expression of specific members of the R2R3-MYB and Dof transcription factor families may contribute to the ectopic lignification phenotypes.
• This comparative analysis provides a suite of hypotheses that can be tested to examine the control of lignin biosynthesis.
Lignins are complex, heterogeneous, cell wall-bound polymers synthesized by land plants (Campbell & Sederoff, 1996). Lignins function as intra- and inter-molecular glues in specialized plant cells, where they reinforce the carbohydrate components of the cell wall (Boerjan et al., 2003). The biosynthesis of lignin is not only a key adaptation for plant support and water transport (Kubitzki, 1987), but also for plant defence (Hammond-Kosack & Jones, 1996). Given the roles played by lignins, the spatial and temporal control of lignin deposition is critical throughout plant growth and development (Boudet, 2000). Despite the importance of controlling lignin deposition, the mechanisms that underpin where and when lignins are made are not yet well understood (Rogers & Campbell, 2004).
Mutants with alterations in the quantity, quality or localization of lignins offer a unique opportunity to uncover the mechanisms that regulate lignin deposition. Arabidopsis thaliana mutants impaired in lignin deposition are readily identified in mutagenized populations by staining cross-sections of inflorescence stems with phloroglucinol-HCl or Maüle's reagent. The distribution of lignified cells found in A. thaliana is typical of a nonwoody dicotyledonous plant (Rogers & Campbell, 2004). Lignins are primarily deposited in the cell walls of xylem vessel elements; however, lignins are also found in the cell walls of sclerenchyma, phloem fibres and periderm. In mature stems, xylem bundles and interfasicular fibres form a ring of lignified tissue within the mature inflorescence stem. Arabidopsis thaliana can also form a vascular cambium under specific growth conditions, and will deposit extensive quantities of lignins in the secondary xylem derived from this meristem (Chaffey et al., 2002). Three main types of A. thaliana lignification mutants have been characterized: those that cause a modification of lignin composition, such as the ferulic acid hydroxylase (fah1) mutant (Chapple et al., 1992); those with a reduction in lignification, such as irregular xylem4 (irx4) (Jones et al., 2001); and those with an increase in lignified tissues such as ectopic lignification 1 (eli1) (Cano-Delgado et al., 2000) and ectopic lignification of pith1 (elp1) (Zhong et al., 2000). In the last class of mutants, lignin deposition is said to be ectopic in that lignins were deposited in cells other than xylem vessels and the sclerified parenchyma. Given that the localization and timing of lignification is perturbed in the ectopic lignification mutants, they provide an ideal opportunity to identify the components of the regulatory machinery that control the spatial and temporal deposition of lignins.
To date, the loci corresponding to three ectopic lignification mutants have been determined. The eli1 mutant was shown to be allelic to the mutant constitutive expression of VSP1 (cev1) (Cano-Delgado et al., 2003). These mutants arise owing to mutations in the CEV1 locus, which encodes the cellulose synthase subunit, CesA3 (Ellis et al., 2002; Cano-Delgado et al., 2003). Another ectopic lignification mutant, elp1, uncovered the ELP1 locus, which encodes a chitinase-like protein, AtCTL1 (Zhong et al., 2002). The elp1 mutant is allelic to the previously described mutant pom-pom1 (pom1) (Hauser et al., 1995; Mouille et al., 2003; Marie-Theres Hauser, pers. comm.). The de-etiolated3 (det3) was originally identified in a screen for dark photomorphogenic mutants (Cabrera y Poch et al., 1993), and was later shown also to lignify ectopically (Cano-Delgado et al., 2000; Newman et al., 2004). The DET3 locus encodes the C-subunit of the vacuolar-type ATPase (V-ATPase) (Schumacher et al., 1999). Despite the fact that mutations in CEV1, ELP1 and DET3 all give rise to broadly similar ectopic lignification phenotypes, the connection between lignin deposition and the three loci is not obvious. Hypothetically, mutations in these disparate loci may disrupt a common regulatory circuit that determines where and when lignins are deposited. Consequently, detailed comparison of the eli1, elp1 and det3 mutants may facilitate the discovery of central regulators involved in the spatial and temporal control of lignification.
In order to test the hypothesis that eli1, pom1 and det3 give rise to similar ectopic lignification phenotypes by disrupting a common regulatory circuit, the mutants were characterized in more detail using genetics, metabolite profiling and transcriptome analysis. Using mutants with new alleles of pom1 and eli1, in conjunction with the det3-1 mutant, genetic mechanisms that are likely to account for shared aspects of the ectopic lignification phenotypes were uncovered. Our data support the hypothesis that mutations at dissimilar loci misregulate lignin deposition at the transcriptional level, and that this occurs through both shared and distinct genetic mechanisms.
Materials and Methods
Plant growth conditions
Wild-type A. thaliana seeds (Col-0) were obtained from the Nottingham Arabidopsis Stock Centre (NASC), Nottingham, UK. The ethane methyl sulphonate (EMS)-mutagenized M2 generation seeds (Col-0 ecotype) were obtained from Lehle Seeds (Round Rock, TX, USA). Unless otherwise indicated, the pom1-26, eli1-3 and det3 mutants as well as pom1-26/eli1-3, pom1-26/det3 and eli1/det3 double mutants (all in a Col-0 ecotype background) were grown at 22°C in a mixture of Levington's Universal soil and Vermiperl vermiculite (3 : 1), under long-day conditions (16 h light : 8 h dark) at a light intensity of 130 µmol m−2 s−1. Inflorescence stem analysis was carried out after 5 wk of growth. The pom1-26, eli1-3 and det3 mutants seeds used for transcript abundance analyses were sterilized and grown in liquid Murashige and Skoog (MS) medium according to published methods (Newman et al., 2004).
For diurnal transcript abundance analyses, plants were grown on semisolid MS medium in Petri plates in a 16 h light : 8 h dark cycle for 14 d at a light intensity of 130 µmol m−2 s−1. At dawn on the day after the plants had reached growth stage 1.03 (Boyes et al., 2001), approx. 1 g of whole plant tissue was harvested from each genotype and snap frozen in liquid nitrogen. This collection was repeated every 4 h for 48 h, with collections during the dark taking place under a green safe light. Each collection contained > 20 seedlings.
For the circadian transcript abundance analyses, attempts were made to reproduce published conditions that had been used for the analysis of circadian changes in transcriptome activity (Harmer et al., 2000). Briefly, plants were grown on semisolid MS medium in Petri plates and entrained to a 12 h light: 12 h dark cycle at a light intensity of 130 µmol m−2 s−1. On the day that the plants had reached growth stage 1.03 (Boyes et al., 2001) during entrainment, the growth conditions were switched to continuous light to facilitate examination of the circadian rhythms of transcript abundance. At dawn on the day after the plants had reached growth stage 1.03, approx. 1 g of whole plant tissue was harvested from each genotype and snap frozen in liquid nitrogen. This collection was repeated every 4 h for 48 h. Each collection contained > 20 seedlings.
Phloroglucinol staining of hand cross-section of inflorescence stems or dark grown hypocotyls was carried out as follows: samples were placed in a 1% phloroglucinol-HCl solution for 10 min, and mounted in 50% glycerol, 6 n HCl and observed immediately. Bright-field illuminated hand cross-sections were taken using a Leica DMRB microscope (Leica, Wetzlar, Germany). For Mäule staining, hand cross-sections of stems were treated for 10 min with 0.5% KMnO4 and then rinsed thoroughly with distilled H2O. The sections were then immersed 5 min in 10% HCl, rinsed and mounted in concentrated NH4OH for an immediate observation.
Mapping of mutations
Eight-thousand individuals of the M2 generation (representing approx. 1000 M1 parents) of an EMS-mutagenized population of Col-0 (parental group 51, obtained from Lehle seeds; Round Rock) were screened for plants with an abnormal lignification pattern in the inflorescence stem, which was visualized following staining with phloroglucinol-HCl. The two mutations identified in the mutant screen were mapped to a specific chromosome arm in the A. thaliana genome using simple sequence length polymorphisms (SSLPs) (Bell & Ecker, 1994). Mapping populations for the two mutants identified in the mutant screen were generated by crossing the mutants with the ecotype Landsberg erecta (Ler). The F1 plants from these crosses were allowed to self-pollinate and set seed. The F2 seed was then sown and screened for individuals that displayed an ectopic lignification phenotype. DNA was extracted from these individuals and used in PCR reactions with SSLP primer pairs. The SSLP primer pairs were based on known markers that allow even coverage of the 10 chromosome arms of the A. thaliana genome (http://signal.salk.edu/genome/SSLP_info/SSLPsordered.html).
Chemical analysis of lignins and carbohydrates
The chemical composition of stage 6.10 (Boyes et al., 2001) A. thaliana stems was determined using a modified Klason analysis. Greater than 20 stems were harvested for each sample, and three biological replicates were assayed. Briefly, freeze-dried A. thaliana stems were ground to pass a 40-mesh screen using a Wiley mill. The ground A. thaliana stems (1 g) were then Soxhlet-extracted with 100 ml acetone for 8 h to remove extractable components, and to minimize the formation of ‘pseudolignin’ during Klason analysis. The total weight of extractable components was determined gravimetrically by rotary-evaporation. The extracted lignocellulosic material was air-dried to remove solvent and then analysed in triplicate for sugar and lignin composition.
To assess lignin content, a 0.2 g sample of extracted A. thaliana stems was transferred to a 15 ml reaction vial in an ice bath. A 3 ml aliquot of 72% (w : w) H2SO4 was added to the sample and thoroughly mixed for 1 min. The test tube was immediately transferred to a water bath maintained at 20°C, and was subsequently mixed for 1 min every 10 min. After 2 h of hydrolysis, the contents of each test tube were transferred to a 125 ml serum bottle, using 112 ml nanopure H2O to rinse all residue and acid from the reaction vial. The serum bottles (containing 115 ml H2SO4 at 4% (w : w) plus A. thaliana stems) were sealed with septa and autoclaved at 121°C for 60 min. Samples were allowed to cool, and the hydrolysates were vacuum-filtered through preweighed medium coarseness sintered-glass crucibles, washed with 200 ml warm (c. 50°C) nanopure H2O to remove residual acid and sugars, and dried overnight at 105°C. The dry crucibles were weighed to determine Klason (acid-insoluble lignin) lignin gravimetrically. The filtrate was also analysed for acid-soluble lignin by absorbance at 205 nm according to TAPPI Useful Method UM250. Thioacidolysis of extracted stems was conducted according to published methods (Lapierre et al., 1999), with the volumes scaled to accommodate 50 mg of starting material.
The concentration of neutral sugars in the filtrate was determined using high-pressure liquid chromatography (HPLC) analysis. The HPLC system (Dionex DX-500; Dionex, Sunnyvale, CA, USA) was equipped with an ion-exchange PA1 (Dionex) column, a pulsed amperometric detector with a gold electrode, and a Spectra AS3500 autoinjector (Spectra-Physics, Mountain View, CA, USA). Before injection, samples were filtered through 0.45 µm HV filters (Millipore, Billerica, MA, USA) and a 20 µl volume of sample was loaded, containing fucose as an internal standard. The column was equilibrated with 250 mm NaOH and eluted with de-ionized water at a flow rate of 1.0 ml min−1. Detection of carbohydrates was facilitated with a post-column addition of 200 mm NaOH at 0.5 ml min−1. The concentration of acid sugars in the filtrate was determined with an identical HPLC system and column, with a different elution gradient: the sugars were eluted with a gradient of 0–400 mm sodium acetate in 100 mm NaOH over 50 min, followed by a 10 min 300 mm NaOH wash. The column was then equilibrated to 100 mm NaOH before the next injection. The sugars were detected by pulsed amperometry with the post-column hydroxide addition.
Northern blot analysis
Total RNA was extracted using published methods (Kirby, 1956; Goldsbrough & Cullis, 1981). For the Northern blot experiments, plants were grown to stage 1.03 on MS agar medium, the aerial tissues (comprising rosette leaves and hypocotyls) harvested, and frozen in liquid nitrogen. Each pool of RNA was derived from > 20 seedlings. Aliquots of 10 µg of total RNA were separated by electrophoresis in 1.2% agarose formaldehyde gel, and run in 1× 3-(N-morpholino) propanesulfonic acid (MOPS) buffer. RNA was then transferred onto a nylon membrane using standard capillary techniques and cross-linked under UV light (1200 µJ, Stratalinker; Stratagene, La Jolla, CA, USA). The amount of RNA that had been transferred to the blot in each lane was quantified using a Bio-Rad Fluor-S MultiImager and Bio-Rad multianalyst software (version 1.1) (Bio-Rad, Hemel Hempstead, UK). Radioactive probes were prepared using the DECAprime II DNA Labelling Kit (Ambion, Austin, TX, USA) and purified using the QIAquick Nucleotide Removal Kit (Qiagen, Crawley, UK). Hybridizations were carried out using membrane manufacturer's instruction. Membranes were exposed to a GS-525 Sample Exposure Platform (Bio-Rad,) for 2 d. The GS-525 Sample Exposure Platform was then scanned using a GS-525 Molecular Imager Laser Scanner, and the digitised image analysed using the Biorad multianalyst software (version 1.1). The RNA blots were used in Northern blot analysis and hybridized at high stringency to 32P-labelled gene specific fragments of the respective lignin biosynthetic genes. The resulting blots were exposed on the phosphoimager and the amount of hybridizing transcripts quantified using Bio-Rad multianalyst software. In each case, background values were also taken to account for any gradients across a blot. Relative transcript abundance was calculated by dividing the quantity of hybridizing transcript by the amount of RNA that had been transferred in a given lane.
Hybridization to the GeneChip Arabidopsis 8K Array (Affymetrix, Santa Clara, CA, USA), scanning of the hybridized array, and data generation were performed at the GARNET facility at NASC (http://nasc.nott.ac.uk/) according to standard Affymetrix protocols (http://affymetrix.com). Total RNA was extracted using published methods (Newman et al., 2004) from 6-d-old A. thaliana seedlings grown in the dark in liquid MS medium as described above. Each pool of RNA was derived from hundreds of seedlings. The RNA derived from three biological replicates of each experimental condition was submitted to the facility for microarray analysis. Raw data from the microarray experiments, as well as the description of the MIAME-compliant project can be found at http://affymetrix.arabidopsis.info/narrays/experimentpage.pl?experimented=14. Normalization of the data was carried out using the Affymetrix Microarray Suite (MAS) software.
Using data mining procedures in a Microsoft Excel spreadsheet containing normalized data exported from the Affymetrix MAS program, lists of genes of interest and present on the chip were generated. From this dataset, new expression matrices were generated containing the mean expression values from the three replicates for each gene of interest. Using the three normalized expression values for genes of interest, a mean expression value was calculated for each gene in each ‘condition’ (i.e. wild type grown on medium lacking sucrose, wild type grown on sucrose supplemented media, pom1 grown without exogenous sucrose, eli1 grown without exogenous sucrose, and det3 grown without exogenous sucrose). The mean expression value for each gene in wild-type plants grown without sucrose was used as the baseline for all subsequent comparisons. To generate relative data, each column in the expression matrix was divided by the expression value for wild-type plants grown without exogenous sucrose. Thus, the wild-type plants grown without exogenous sucrose became the experimental control against which the other chips were compared. The values were then standardized to have a mean of zero to give both positive and negative expression values relative to the control.
Following this, the data were log transformed to base 2. Given that only a limited number of genes identified to participate in lignin biosynthesis were present on the chip, Eisengrams were generated for all. In order to provide an indicator of data robustness, a one-way analysis of variance (anova) was performed for each gene set. The anova examined the transcript abundance data for each gene to determine if there was significant variation in transcript abundance between the five conditions examined. This produced a P-value representing the likelihood that the particular pattern of gene expression seen for each gene could occur by chance. The P-value for each gene set was added to the annotation file for gene function.
Log-transformed data were subjected to k-means clustering by the European Bioinformatics epclust program (http://ep.ebi.ac.uk/EP/). A k-value of 35 was used to generate 35 hypothetical profiles that traverse the range of possible transcript abundance profiles in the inputted data. Each gene was then allocated to the cluster whose centre matched its own expression profile most closely based on Euclidean distance (the square root of the sum of the squares of distances between each point on a profile and its equivalent the one whose distance is being measured) (Legendre & Legendre, 1998). The clustering process was iterative, such that when all the genes had been assigned to a cluster, new centres were calculated using the mean profile for the genes assigned to that cluster. These cluster centres were used for a new round of clustering, and this process continued until a stable set of clusters was generated.
The cluster data generated a set of genes that had a greater than threefold increase in transcript abundance in dark-grown wild-type plants grown in the presence of sucrose and the three mutants grown in the dark in the absence of sucrose, relative to dark-grown wild-type plants grown in the absence of sucrose. These genes were examined to determine if there were any over-represented motifs present in their 5′ noncoding sequences. To this end, 500 bp of upstream sequence for each gene in the cluster were obtained in a bulk download from The Arabidopsis Information Resource (TAIR) (http://www.arabidopsis.org/tools/bulk/sequences/index.html). These sequences were then subjected to analysis using the internet-based Motif Sampler software (http://www.esat.kuleuven.ac.be/~thijs/Work/MotifSampler.html). The following parameters were set on MotifSampler: motif length = 9; prior probability of finding 1 motif instance = 0.5; number of different motifs = 6, maximum number of motif instances per sequence = 0; maximum allowed overlap between different motifs = 2. Ten iterations of the algorithm were run, and those motifs found in the majority of iterations used for further analysis. The motifs identified from motif elucidation were used to query the PLACE (http://www.dna.affrc.go.jp/htdocs/PLACE/) and PlantCare (http://intra.psb.ugent.be:8080/PlantCARE) databases to search for previously characterized cis-acting regulatory elements.
Identification of ectopic lignification mutants
Previously, the det3 mutant was shown to ectopically lignify (Cano-Delgado et al., 2000; Newman et al., 2004). In order to identify similar mutants for comparative analysis, a mutant screen was initiated that involved histochemical staining of A. thaliana inflorescence stems. Phloroglucinol-HCl stains lignin magenta-pink when it reacts with coniferaldehyde constituents in the polymer. Mutants were identified on the basis of a characteristic ‘striped’ lignification pattern observed when whole inflorescence stems were stained with phloroglucinol-HCl. Two new mutants were identified in a screen of 8000 m2 generation, EMS-mutagenized plants.
The mutations corresponding to the two mutants were mapped to a specific chromosome arm in the A. thaliana genome using SSLPs (Bell & Ecker, 1994). A complete lack of recombination with the marker NGA63 placed the first of the two mutations at the top of chromosome I. AthS0392, another chromosome I marker, also cosegregated with this mutation. The map location of this first mutation coincided with that of pom1 and a complementation test confirmed that the first mutant was an allele of pom1. Consequently, the first mutant was named pom1-26, as it uncovered the 26th allele to be identified at this locus (M.-T. Hauser, pers. communication).
The second mutation mapped to the top of chromosome five, as revealed by linkage to marker NGA225, with 14.3% recombination occurring between the marker and the mutation, and NGA249 with 12.5% recombination. This interval coincided with another mutation that gives rise to ectopic lignification, eli1/cev1. Complementation testing revealed that the second mutant and eli1 were allelic, so the second mutant was named eli1-3.
Histochemical analysis of ectopic lignification mutants, and double mutants
Free-hand cross-sections were made of the primary inflorescence stems of all three mutants, at growth stage 6.10 (Boyes et al., 2001). Phloroglucinol staining consistently showed that det3 had the most intensely staining ectopic lignification phenotype, and pom1 the weakest. The sections revealed that the deposition of phloroglucinol-positive material in the pith tissue was amorphous in nature, with no identifiable or consistent pattern of deposition (Fig. 1). Consistent with previous findings for det3 (Newman et al., 2004), all three mutants also exhibited ectopic lignification in the leaves and hypocotyls at growth stage 1.03 (Boyes et al., 2001) (Fig. S1a, available online as supplementary material), and throughout the entire seedling in dark-grown seedlings 7 d after termination (Fig. S1b, available online as supplementary material).
On the basis of staining with Mäule's reagent, which provides a qualitative assessment of lignin subunit composition (syringyl vs guaiacyl), the ectopic lignin in the pith of the inflorescence stems of the eli1 mutant was different from that found in the other two mutants (Fig. 1). The pith tissue of det3 and pom1 stained red with Mäule's reagent, indicating the presence of an abundance of syringyl moieties in the lignin (Fig. 1). Syringyl-rich lignins are typically found in the interfasicular fibres of A. thaliana, in contrast to the lignin that is deposited in the cell walls of xylem vessels, which lacks syringyl subunits and is rich in guaiacyl units instead. Unlike det3 and pom1, the ectopic lignin deposited in the eli1 mutant stained a golden colour, indicative of a lower syringyl to guaiacyl (S : G) ratio of the lignins deposited in the walls of xylem cells (Fig. 1c). Furthermore, less intense golden staining was also observed in the cells of the cortex of the eli1 mutant, which may indicate that guaiacyl lignins were accumulating ectopically in these cells as well. However, as there was no phloroglucinol staining in these cells, it remains an open question as to whether the yellow staining observed with Mäule's reagent in the cortex of the eli1 mutant corresponds to lignins or not.
Double mutants were constructed to examine the extent to which ectopic lignification was due to distinct vs overlapping effects of the mutant loci. All of the double mutants had additive phenotypes with respect to plant growth and development. In particular, det3/pom1 double mutants were severely compromised. Many of the det3/pom1 double mutants died before bolting. Similarly, the pom1/eli1 double mutant was compromised in growth and the rate of development was slow relative to the single mutants. The pom1/eli1 double mutants had very small rosettes, and their leaves accumulated anthocyanins.
By contrast to the overall morphology, the double mutants had clear overlap in the manifestation of the ectopic lignification phenotype (Fig. 1). No significant alteration in the cellular distribution of the ectopic lignification pattern was observed in the double mutants. The double mutants tended to exhibit the ectopic lignification phenotype of the more prominent parental phenotype. For example, on the basis of Mäule staining, ectopic lignin in the det3/eli1 and det3/pom1 double mutants was most similar to the det3 mutant, while ectopic lignification in the pom1/eli1 double mutant was similar to the more severe eli1 phenotype.
Metabolite profiling of the cell walls ectopic lignification mutants
While pom1/elp1, eli1/cev1, and det3 are referred to as ectopic lignification mutants, the characterization of the lignins in these mutants has generally been limited to histochemical staining, such as that described above. To better characterize the ectopic lignification phenotypes, inflorescence stems, including cauline leaves, from the mutants at growth stage 6.10 (Boyes et al., 2001) were subjected to metabolite profiling and compared with wild-type plants. Freeze-dried stems were extracted with acetone, and then subjected to extensive chemical analyses.
While the stems of all three mutants exhibited a decrease in total extractives, and a concomitant increase in total lignins relative to wild-type plants, the mutants were clearly different with respect to the chemical composition of the insoluble residue (Table 1). For example, while the stems of all three mutants had increased quantities of acid-insoluble lignins relative to wild-type plants, this increase was particularly striking in the det3 mutant. The mutants also varied in the quantity of acid-soluble lignins relative to wild-type plants, with no difference observed for eli1, and a decrease observed for pom1 and det3. Acid-soluble lignins are liberated during the 4% acid treatment, under high temperature, when the carbohydrate component of the cell wall is dissolved, and are thought to represent either low molecular weight lignins, or low molecular weight lignins complexed with dissolved carbohydrates.
Table 1. Chemical analysis of inflorescence stems from Arabidopsis plants at growth stage 6.10 (Boyes et al., 2001) from wild-type (WT) plants and three ectopic lignification mutants
More than 20 stems were harvested for each sample, and three biological replicates were assayed. Values that are significantly different from wild type are in bold font (Student's t-test, P < 0.05).
H, hydroxyphenyl; G, guaiacyl; S syringyl.
The total yield is the quantity of monolignols recovered by thioacidolysis based on the starting lignin concentration.
5.95 (± 0.46)
11.39 (± 0.15)
8.34 (± 0.4)
21.9 (± 28.2)
3.99 (± 0.5)
17.17 (± 0.26)
6.57 (± 0.14)
58.6 (± 16.2)
2.49 (± 0.25)
12.76 (± 0.19)
8.62 (± 0.34)
128.4 (± 13.8)
3.79 (± 0.37)
12.75 (± 0.18)
6.86 (± 0.18)
116.1 (± 76.5)
There was also significant variation in the monomer composition of the lignins (Table 1). For example, the ratio of para-hydroxyphenyl (H) to guaiacyl (G) to syringyl (S) units was roughly the same when comparing wild type with det3. By contrast, there were more H units relative to wild-type lignins in the pom1 and eli1 mutants, and less S units in the eli1 mutant. These chemical analyses are consistent with the results obtained with histochemical staining, which indicated that the eli1 mutant had a much lower proportion of S units than that observed in wild-type lignins. Monolignol release by thioacidolysis (represented by ‘total yield’ in Table 1) is contingent on the relative proportion of S and G units, with an increased S : G ratio generally resulting in increased release of monolignols. Consistent with the higher S : G ratio in the pom1 mutant, the proportion of monolignols released by thioacidolysis was greater in this mutant relative to wild-type plants. Similarly, consistent with the decreased S : G ratio in the eli1 mutant relative to wild-type plants, the thioacidolysis monolignol recovery was also lower. The eli1 mutant also exhibited a greater accumulation of cell wall-esterified ferulate relative to wild-type plants.
The carbohydrate composition of the cell walls also varied significantly among the mutants (Table 2). Notably, while fucose was more abundant in all three mutants relative to wild type, arabinose was only more abundant in the insoluble residue from pom1 and eli1 stems. There was a clear decrease in the abundance of rhamnose in pom1 stems, whereas there was a obvious increase in rhamnose in det3 stems. There was also a notable decrease in the abundance of glucose in eli1 stems, which might be expected in this mutant with a defective cellulose synthase subunit, and, presumably, a lower capacity to incorporate glucose into the cell wall. Unlike pom1 and eli1 plants, the stems of det3 plants were enriched in rhamnose and xylose residues relative to wild-type plants. While the mutants are superficially described as ‘ectopic lignification’ mutants, they clearly differ in other aspects of cell wall composition.
Table 2. Carbohydrate composition of insoluble extracts from inflorescence stems from Arabidopsis plants at growth stage 6.10 (Boyes et al., 2001) for wild-type (WT) plants and three ectopic lignification mutants
More than 20 stems were harvested for each sample, and three biological replicates were assayed. Quantities are expressed as nmol mg−1 tissue. Values that are significantly different from wild type are in bold type (Student's t-test, P < 0.05).
120.97 (± 6.1)
192.4 (± 3.98)
194.32 (± 4.6)
125.6 (± 3.26)
11.16 (± 0.4)
18.2 (± 0.4)
14.17 (± 0.3)
18.24 (± 0.3)
161.51 (± 2.1)
178.43 (± 1.38)
179.37 (± 0.67)
164.7 (± 2.6)
171.34 (± 9.9)
208.21 (± 15.92)
175.7 (± 13.26)
170.97 (± 5.6)
915.7 (± 77.47)
921.83 (± 54.4)
602.75 (± 54.97)
959.48 (± 55.94)
70.32 (± 4.07)
80.08 (± 0.92)
68.76 (± 2.19)
81.69 (± 3.11)
172.26 (± 15.4)
74.87 (± 6.9)
162 (± 16.6)
257.99 (± 11.8)
211.96 (± 30.96)
291.4 (± 27.3)
160.34 (± 23.21)
331.83 (± 20.62)
Diurnal and circadian transcriptional regulation of genes encoding lignin biosynthetic enzymes
To test the hypothesis that the ectopic deposition of lignins in the mutants arises from changes in the transcriptional regulation of the genes encoding lignin biosynthetic enzymes, Northern blot analysis was used to assess transcript abundance over a period of 48 h in the aerial tissues, comprising the rosette leaves and hypocotyls, collected from > 20 plants grown to stage 1.03 (Boyes et al., 2001) (Fig. 2). The rosette leaves and hypocotyls of all three mutants exhibit ectopic lignification at this stage (Fig. S1a). The temporal expression profiles of 11 genes encoding lignin biosynthetic enzymes were examined every 4 h over a 48-h diurnal cycle (16 h light : 8 h dark). The 11 genes included three that encode the entry-point enzyme into phenylpropanoid metabolism, phenylalanine ammonia-lyase (PAL1, At2g37040; PAL2, At3g53260; PAL3, At5g04230). These genes also included two others that encode enzymes related to general phenylpropanoid metabolism, cinnamate 4-hydroxylase (C4H, At2g30490) and hydroxycinnamate:CoA ligase (4CL1, At1g51680). The other genes that were examined encode enzymes that are hypothesized to be more central to the biosynthesis of the monomeric precursors of lignins, the monolignols, including: coumarate 3-hydroxylase (C3H1, At2g40890), caffeoyl-CoA O-methyltransferase (CCoAOMT1, At4g34050), hydroxycinnamoyl-CoA reductase (CCR1, At1g15950), ferulate/coniferaldehyde 5-hydroxylase (F5H1, At4g36220), caffeic acid/5-hydroxyconiferaldehyde O-methyltransferase (COMT, At5g54160), hydroxycinnamyl alcohol dehydrogenase (CAD6, At4g34230). The probes that were used for lignin biosynthetic genes had previously been shown to be part of the lignin ‘toolbox’ in A. thaliana (Raes et al., 2003). Transcript abundance for the gene encoding the entry-point enzyme for flavonoid biosynthesis, chalcone synthase (CHS, At5g13930), was also examined. All of the probes that were used were gene specific. The mean per cent of wild-type transcript abundance was also calculated for each gene (Fig. 3). This value also provides a relative indicator of the cumulative transcript abundance in the mutants relative to wild type.
Lignin biosynthetic genes fell into four different groups with regard to the levels of expression in the four different genotypes. The first group comprised PAL3 and C3H1 (Figs 2 and 3). These genes exhibited very little difference in transcript accumulation profiles between all four genotypes. The second group contained PAL2, 4CL1 and CAD6 (Figs 2 and 3). The level of PAL2, 4CL1 or CAD6 transcript accumulation in the det3 mutant exceeded that seen in any of the other genotypes. The other mutants, pom1 and eli1, had wild-type levels of PAL2, 4CL1 or CAD6 transcript accumulation. CCR1 and COMT comprised the third group of genes (Figs 2 and 3). These genes were grouped together because eli1 and det3 had significantly higher total transcript accumulation than either the wild type or pom1. The last group was the largest and included PAL1, C4H1, CCoAOMT1 and F5H1 (Figs 2 and 3). For this group of genes, all three ectopic lignification mutants exhibited a higher level of transcript accumulation than the wild type. Of all the genes examined, those contained in this group had the greatest overall increase in transcript abundance in all genotypes (Fig. 3). Unlike the genes encoding lignin biosynthetic enzymes, there was no obvious or consistent difference in the transcript accumulation for CHS between wild-type plants and any of the mutants (Figs 2 and 3).
The increase in transcript abundance that was observed for lignin biosynthetic genes in the ectopic lignification mutants may be attributable to an alteration in the normal circadian regulation of these genes. To test this hypothesis, a second set of temporal Northern blot analyses were undertaken, but with the conditions altered so that they precisely replicated those that allow the examination of circadian effects (Harmer et al., 2000). To this end, plants were entrained to a 12 h light : 12 h: dark cycle. When the plants had reached growth stage 1.03 (Boyes et al., 2001) with entrainment, the growth conditions were switched that morning to continuous light to facilitate examination of the circadian rhythms of transcript abundance in the three mutants compared with wild type. Comparable to the trends observed in the diurnal temporal northern analysis, the ectopic lignification mutants exhibited a pattern of transcript abundance over the 44 h that was similar to that exhibited by the wild-type plants (Figs S2 and S3, available online as supplementary material). Of the eight genes examined, only C3H1 and CAD6 exhibited significant increases in transcript abundance in all three mutants, whereas CCoAOMT1, CCR1, and COMT had significant increases in transcript abundance in only eli1 and det3.
Conditional nature of lignin deposition
To determine the relationship between the ectopic lignification phenotype and the provision of sugar, wild-type and mutant seeds were germinated and grown in the dark for 14 d. Seedlings were grown in liquid medium in the absence of an exogenously supplied carbon source, or in medium supplemented with 30 mm sucrose. As the seedlings were germinated and grown in the dark, they were photosynthetically inactive, so any sugars available to the developing seedling must be derived from seed reserves or provided exogenously. No phloroglucinol staining was observed in wild-type plants grown in the absence of sucrose, whereas the xylem vessels were clearly phloroglucinol positive when the seedlings were grown in the dark in the presence of 30 mm sucrose (Fig. 4a). By contrast, all the ectopic lignification mutants had phloroglucinol-positive material in the absence of sucrose (Fig. 4a).
Microarray analysis of the lignin biosynthetic pathway in the ectopic lignification mutants
The Affymetrix 8K GeneChip was used to simultaneously examine the transcript abundance of genes encoding additional steps in the lignin biosynthetic pathway. A list of genes putatively involved in lignification was compiled and compared against the genes present on the 8K array. Nine of the 14 genes that comprise the A. thaliana‘lignin toolbox’ (Raes et al., 2003) are found on the 8K GeneChip. An additional 12 genes on the 8K GeneChip encode gene products that have functions that are annotated as being ‘like’ those of monolignol biosynthetic enzymes. All 21 genes were included in the analyses.
The experiment examined the A. thaliana transcriptome under four ‘conditions’ where the plants differed in the extent of lignification. Wild-type A. thaliana seedlings were grown in the dark in the absence of an exogenous sugar supply to provide a ‘nonlignified’ control, against which other lignification conditions could be examined. The first lignin accumulation ‘condition’ corresponded to dark-grown wild-type plants grown in medium supplemented with sucrose. As the three ectopic lignification mutants produce lignin throughout the length of the seedling when grown in the dark without a sucrose supplement (Fig. S1b), transcript abundance data was also obtained for these three ‘conditions’. Three biological replicates were obtained for each condition, so the dataset comprises 15 individual GeneChips. The details of the conditions, as well as all data, are available online at NASCArrays (Craigon et al., 2004), as Hybridization Set: Campbell AGA Arrays (http://affymetrix.arabidopsis.info/narrays/experimentpage.pl?experimented=14). In general, increases in transcript abundance, accordant with conditions that resulted in lignin accumulation, could be documented for almost every gene class that was interrogated (Fig. 4b). Furthermore, in every instance but one (CCR), these changes were significant with P < 0.05. In almost every case where there were two or more gene family members encoding a given step in the lignin biosynthetic pathway, one gene family member consistently had increased transcript abundance in all three mutants, in addition to wild-type plants grown in the presence of sucrose.
Cluster analysis of microarray data for the ectopic lignification mutants
A bioinformatics approach uncovered regulatory factors that may account for shared aspects of the phenotypes observed in dark-grown, sucrose-supplemented, wild-type plants, and in the ectopic lignification mutants. Hierarchical cluster analysis of all of the genes on the 8K GeneChip identified a well-supported cluster of 31 genes that had a transcript abundance pattern that was shared between dark-grown, sucrose-supplemented, wild-type seedlings and dark-grown ectopic lignification mutants, relative to wild-type seedlings grown in the dark in the absence of sucrose (Fig. 5a). This well-supported cluster contained genes that showed a large increase in transcript abundance under conditions where lignins accumulated. In addition to containing two genes in the lignin biosynthetic toolbox (PAL1 and 4CL1) (Raes et al., 2003), this cluster also contained genes encoding proteins related to photosynthetic carbon fixation (10/31), photorespiration (2/31), lipid metabolism (6/31), auxin biosynthesis (1/31) and one-carbon metabolism (1/31).
The 5′ noncoding sequences of the genes identified by cluster analysis were interrogated to identify conserved motifs. In this analysis, 500 bp of 5′ noncoding sequence was obtained from the TAIR database for each gene of interest. The 5′ sequences for the set of coregulated genes were then subjected to Gibbs sampling to identify DNA motifs that were over-represented within these putative regulatory regions. All motifs were compared against the PLACE database to determine if these motifs had a previously characterized role in transcriptional regulation. Of those motifs that have been previously characterized, one, MACWCCTYCK, bears similarity to known binding sites for R2R3-MYB transcription factors, which lack G residues and are frequently referred to as AC elements (reviewed in Rogers & Campbell, 2004); whereas another, TwACTTTTT, is similar to a target for Dof transcription factors (Mena et al., 2002; Isabel-LaMoneda et al., 2003). Of the 31 genes in the cluster, six (At1g20340, At1g51680, At2g36880, At2g37040, At4g13310 and At4g39950) had the 9 bp AC element-like sequence in their 5′ noncoding sequence. As there are 304 of these 9 bp sites in the 5′ noncoding region of genes in the A. thaliana genome, this frequency of occurrence in the list of 31 genes is approx. 15 times greater than would be expected by chance alone. In keeping with this finding, 22 of the 31 genes contained at least one canonical AC-element (ACCWAHM) – elements that are over-represented in the regulatory regions of genes encoding lignin biosynthetic enzymes (Raes et al., 2003). Only four (At1g20340, At3g16140, At4g12880, At5g35630) of the 31 genes contained the putative Dof-binding site, which is only 1.5 times greater than expected by chance alone, based on the frequency of this element in the 5′ noncoding region of genes in the A. thaliana genome.
The microarray data were queried to determine if there were any R2R3-MYB or Dof family members with patterns of transcript abundance that were consistent with playing a role in lignin accumulation. Similarly, two other transcription factor families with previously reported roles in lignin deposition were examined: KNOX family members related to BREVIPEDICELLUS/KNAT1 (Mele et al., 2003), and transcription factors related to Nicotiana tabaccum LIM1 (NtLIM1) (Kawaoka et al., 2000). Cluster analysis was conducted using these genes, together with those corresponding to the lignin biosynthetic toolbox (Raes et al., 2003). Of the members of these families found on the 8K GeneChip, only 2 out of 33 had transcript abundance patterns that clustered together with the lignin biosynthetic genes that define the lignin toolbox, all of which had increased transcript abundance under conditions where lignin accumulation occurred (Fig. 5b). Another subset had completely opposite transcript abundance patterns, that is, the abundance of their transcripts decreased under conditions where lignin accumulation occurred.
Mutants impaired in normal lignin deposition provide a powerful tool to unravel the complexities underpinning the spatial and temporal control of lignin biosynthesis. The mutants pom1/elp1, eli1/cev1 and det3 all accumulate lignins in cells where they are not normally found. The three mutants arise because of mutations in three very different loci, with no obvious link between the functions of the three gene products. That all three mutations give rise to a shared phenotype is intriguing, and it may be that this occurs because the defects arising from the mutations impinge on a common regulatory pathway that governs the timing and localization of lignin deposition. Alternatively, it is possible that ectopic lignification arises in the three mutants as a result of completely nonoverlapping mechanisms. The results presented herein suggest that at least part of the ectopic lignification phenotype may be attributable to misregulation of lignin biosynthetic genes, and that this misregulation may involve both common and distinct regulators in the three different mutants.
The ectopic lignification phenotype varies between mutants
There are obvious differences in the ectopic lignification phenotypes observed in the det3 mutant relative to the pom1 and eli1 mutants that were isolated in this study. The det3 phenotype is the most striking of the three in that it stains with the greatest intensity and, on the basis of chemical analysis, has the greatest abundance of extractable lignins. The pom1 allele isolated in this study had the weakest lignification phenotype, both in terms of histochemical staining and chemical analysis. The pom1 mutant did not have significantly greater concentrations of total (acid insoluble + acid soluble) lignins compared with wild-type plants, which was a surprising finding. A decrease in the relative abundance of lignified vascular cells relative to ectopically lignifying cells could account for the apparent lack of increase in total lignins in the pom1 mutant, but there was no evidence to support this hypothesis. Alternatively, it is possible that the presence of ectopically lignifying cells without a concomitant increase in total lignin accumulation in the stem was accommodated by decreasing the amount of lignin deposition that occurred in the cell walls of the cells that are normally lignified. This hypothesis is difficult to test, but could be the subject of future investigations. Regardless of the mechanism, the pattern of lignification, and the composition of the lignins were clearly different in the pom1 mutant. The eli1 mutant was different again, with a lower S : G ratio, and an increase in extractable lignins superimposed upon the ectopic lignification phenotype
The variation in the quantity and quality of lignins was surpassed by the variation in the quality and quantity of other cell wall components. There was notable variation in cell wall carbohydrates, suggesting that there is no obvious relationship between cell wall carbohydrate composition and the manifestation of an ectopic lignification phenotype per se. Conversely, the specific changes in cell wall carbohydrates may be related to the specific changes in lignin deposition found in each mutant. Chemical analysis of cellulose synthase mutants related to eli1, such as rsw1, may determine whether there are consistent changes in cell wall carbohydrate composition that can be correlated with specific changes in lignin chemistry.
Despite the variation between the mutants in terms of lignin composition and the quantity of extractable lignins, there was one striking similarity between the mutants – the location of the ectopic lignins. While the double mutants suggested that the intensity of lignification, and the nature of the subunits in the mutants could be additive, the localization of ectopic lignin deposition, as determined by phloroglucinol staining, remained the same. Ectopic lignification, as defined by the localization of phloroglucinol staining, was only observed in the pith and occasionally the phloem fibres. Cortical cell types also stained positively with Mäule's reagent in the eli1 mutant, but this staining did not coincide with phloroglucinol staining, suggesting that whatever the Mäule's reagent was reactive with, it was not lignins. Therefore, while it is possible to deposit different kinds of lignins, it appeared that only certain cell types acquire the competency to lignify in the mutants. This suggests that the three mutants may impinge upon a common regulatory pathway to confer competency to these cells alone.
Ectopic lignification mutants provide further evidence for the central role of transcriptional regulation in the control of lignin deposition
There is a large body of evidence that suggests that lignin biosynthesis is regulated, at least in part, at the transcriptional level (Hertzberg et al., 2001; Anterola et al., 2002; Demura et al., 2002; Israelsson et al., 2003; Patzlaff et al., 2003; Raes et al., 2003). Consistent with this evidence, a high degree of accordance was found between transcript abundance and lignin accumulation in the three ectopic lignification mutants examined in this study. So strong was the correlation that it is tempting to speculate that the differences in lignin accumulation that were observed between the three mutants may be solely attributable to differences in transcriptional regulation of genes encoding lignin biosynthetic enzymes.
In examining the transcriptional regulation of lignin biosynthetic genes, it is important to examine the diurnal changes in transcript abundance. Some previous studies have hinted at the diurnal regulation of lignin biosynthetic genes (Logemann et al., 1995; Mizutani et al., 1997), but a systematic analysis of the diurnal changes in transcript abundance of lignin biosynthetic genes has not yet been reported. The data provided herein reveal that it is important to investigate the diurnal component of the regulation of lignin biosynthetic genes, otherwise important differences in transcript abundance might be overlooked. Microarray analysis of lignin biosynthetic genes revealed that they are also likely to be under circadian regulation (Harmer et al., 2000). Consequently, in dissecting the basis for differences in lignin deposition, this aspect of regulation should also be addressed.
In the ectopically lignifying aerial tissues of all three mutants grown to growth stage 1.03, there did not appear to be a significant disruption in diurnal or circadian trends in transcript abundance; however, all three mutants exhibited marked increases in transcript abundance for many of the key genes encoding enzymes in the monolignol biosynthetic pathway, the genes that define the ‘lignin toolbox’ (Raes et al., 2003). The greatest changes in transcript abundance were observed for det3, which also accumulated the greatest amounts of lignins. Similarly, pom1, which had the lowest amount of lignins of all three mutants, also had the lowest transcript abundance. Under diurnal conditions, the genes encoding PAL1, C4H1, CCoAOMT1 and F5H1 had significant increases in transcript abundance in all three mutants relative to wild-type plants. In keeping with the relationship between lignin content and transcript abundance, in the eli1 and det3 mutants, but not pom1, CCR1 and COMT also exhibited significant increases in transcript abundance relative to wild type, and in the det3 mutant, but not the other two mutants, PAL2, 4CL1 and CAD6 had increased transcript abundance relative to wild type. These findings suggest that PAL1, C4H1, CCoAOMT and F5H1 may be under a common regulatory control in all three mutants; whereas, CCR1 and COMT are only affected in the mutants with the greatest accumulation of lignins. This could reflect either the expression of different regulators in the different mutants, or a threshold mechanism that results in the expression of different suites of genes depending on the strength of some common stimulus.
It is important to note that the differences in transcript abundance did not completely explain differences in lignin subunit composition in the three mutants. There are several reasons why this may be the case. In the first instance, while the Northern blot analysis was relatively comprehensive in capturing multiple time points in the expression of genes encoding lignin biosynthetic enzymes, it still only represented 2 d during the lifetime of the plants. It may be that the genes modifying lignin subunit composition are expressed at later points in development that were missed by the analyses described herein. In addition, the differences in transcript abundance documented in the aerial tissues of growth stage 1.03 plants undoubtedly do not reflect those that take place in the inflorescence stem for which the lignin composition was determined. Finally, transcript abundance only represents a portion of the picture in terms of how a biosynthetic pathway may be regulated, and it may be that translational or post-translational mechanisms operate to modify lignin subunit composition. Future studies should aim to investigate these possibilities.
Under circadian conditions, significant increases in relative transcript abundance were documented for only C3H1 and CAD6 in all three mutants relative to wild-type plants, while CCoAOMT, CCR1 and COMT had elevated transcript abundance in only eli1 and det3. These findings emphasize the correlation between the extent of transcript accumulation across the pathway and lignin accumulation. Moreover, these results suggest that diurnal and circadian stimuli have effects on overall transcript abundance that vary between the different members of the lignin biosynthetic toolbox, and that this variance is genotype dependent. Nevertheless, while there was an increase in overall transcript abundance in the mutants, the pattern of diurnal and circadian changes was not appreciably different from wild-type plants. This suggests that the mutants do not accumulate lignins by altering the regulatory mechanisms that control diurnal or circadian fluctuations in lignin biosynthesis, but rather simply alter the overall accumulation of transcripts.
A role for sugar in lignin deposition
As the patterns of diurnal and circadian changes in transcript abundance were not significantly altered in the mutants relative to wild-type plants, an alternative hypothesis was considered that may account for the ectopic lignin production in the mutants. As importance of sugar signalling in the deposition of lignins in wild-type A. thaliana plants has been shown (Rogers et al., 2005), the possibility that this may be altered in the mutants was examined. As has previously been observed (Rogers et al., 2005), on the basis of phloroglucinol staining, the ability to detect lignins in dark-grown, wild-type xylem vessels was contingent on the presence of exogenous sucrose. That is, exogenous sugar was necessary for visible phloroglucinol staining in dark-grown, wild-type seedlings. This was not the case for the ectopic lignification mutants, which accumulated phloroglucinol-positive material in the dark even in the absence of sucrose. Consistent with this finding, many of the lignin biosynthetic genes that had increased transcript abundance in wild-type plants grown in the dark in the presence of sucrose (Rogers et al., 2005) also had increased transcript abundance in the mutants. This suggests that the regulatory factors that are deployed to give rise to sugar-responsive lignification in dark-grown, wild-type plants may also be deployed in the mutants even when sugar is not present.
Some of the genes show very consistent trends in transcript accumulation between the plants grown in the light to growth stage 1.03 for Northern blot analysis and those grown in the dark for microarray analysis, including PAL1 and F5H. By contrast, other genes, such as PAL2 and 4CL1, had increased transcript abundance in only a limited subset of mutants under one of the conditions and had increased transcript abundance in all three mutants under the other conditions. This observation underlines the fact that different conditions have varying effects on overall transcript abundance of different members of the lignin biosynthetic toolbox, and that this variance in overall abundance is genotype dependent. It suggests that different genes, including some of those genes that are ‘like’ those in the lignin biosynthetic toolbox, respond to the mutations in a manner that depends on growth stage or the environmental conditions. Importantly, it suggests that different suites of genes, including those not normally thought to be in the lignin biosynthetic toolbox, may be involved in the deposition of lignins, depending on developmental stage or growth conditions, and suggests that multiple stimuli must feed into the regulation of lignin biosynthesis. It is striking that these stimuli all impinge upon transcript abundance of genes involved in lignin deposition and suggests that transcription factors play a key role in regulating the process.
Relationship between transcriptome activity in dark-grown wild-type plants grown in the presence of sucrose vs dark-grown ectopic lignification mutants grown without sucrose
Comparative transcriptome analysis of 14-d-old seedlings that had been germinated and grown in the dark was undertaken, as these growth conditions accentuated difference in the ectopic lignification phenotype of the mutants relative to wild-type plants. Moreover, analysis of plants grown under these conditions simplified the dissection of regulatory mechanisms that may be involved in giving rise to lignin accumulation by eliminating diurnal and circadian inputs. Importantly, these analyses revealed regulatory factors that may be involved in the control of lignin deposition in wild-type plants in response to sucrose, which are constitutively active in the ectopic lignification mutants.
Cluster analysis of microarray data identified a set of 31 genes with pronounced increases in transcript abundance in dark-grown, wild-type plants grown in the presence of sucrose, and in the three ectopic lignification mutants. Two genes of the lignin biosynthetic toolbox are included in this group of genes, PAL1 and 4CL1 (Raes et al., 2003), in keeping with the lignin accumulation that occurs in wild-type plants grown in the presence of sugar, and in the three mutants grown without sugar. Under these simple growth conditions, it suggests that the mechanism that controls lignin accumulation in wild-type plants in response to sugar, does so by altering the expression of these two genes, and that this mechanism is constitutively active in the three mutants. Of the genes that show this pattern of regulation, 12 are involved in chloroplast function, which is curious, given that the plants were germinated and grown in the dark. Exogenous sucrose can function as a proxy for light signalling in dark-grown plant tissues (Cheng et al., 1992), and, in these experiments, it may be that the exogenously supplied sucrose ‘misinforms’ wild-type plants, indicating that they are undergoing photosynthesis, and activating gene expression related to chloroplast function. That these same genes are also activated in the three ectopic lignification mutants, in the absence of sucrose, suggests that this sugar-signalling pathway is constitutively active in the mutants. These findings are in keeping with the fact that all three mutants are also dark photomorphogenic (Cano-Delgado et al., 2000, 2003; Zhong et al., 2002; Mouille et al., 2003).
One of the genes contained within the 31 gene cluster, which has increased transcript abundance under conditions that result in lignin accumulation, encodes a cytochrome P450 that is involved in auxin biosynthesis, CYP79B2 (Zhao et al., 2002). CYP79B2 is a critical component of the auxin biosynthetic machinery in A. thaliana (Zhao et al., 2002). Auxin is well-known to be important in the differentiation of lignifying cells, such as xylem, and auxin can induce lignification (Demaggio, 1972; Sugiyama & Komamine, 1990; Sieburth, 1999; Sachs, 2000; Mellerowicz et al., 2001). It may be that the elevated level of CYP79B2 gene expression results in increased auxin biosynthesis in dark-grown wild-type plants growing in the presence of sucrose, and in the three ectopic lignification mutants, even in the absence of sucrose. Increased auxin levels, in turn, may induce the lignin biosynthesis observed in these plants.
Two other genes in the cluster of 31 genes may be linked to the ectopic lignification phenotype. The first of these genes encodes glutamine synthetase (GS). Like glycine decarboxylase, which is also encoded by a gene in the cluster of 31 genes, GS is implicated in photorespiration (Lam et al., 1996). It also plays a central role in nitrogen recycling in plant tissues (Lam et al., 1996). The requirement for nitrogen recycling is particularly high in lignifying cells, as the first enzyme in the lignin toolbox, PAL, catalyses the deamination of phenylalanine and thereby liberates large quantities of nitrogen. This nitrogen would be lost to the plant if there was not an efficient recycling mechanism. Glutamine synthetase is hypothesized to play a role in recovering the nitrogen liberated during lignification (Razal et al., 1996). Recent evidence suggests that the regulation of the transcription of genes encoding lignin biosynthetic enzymes and GS may be linked in some plants (Gomez-Maldonado et al., 2004), consistent with what has been observed here. The second of the two genes encodes S-adenosylmethionine synthetase (SAM synthetase). This is involved in one-carbon metabolism, and plays a central role in generating the SAM that is used as a methyl donor in methylation of the 3- and 5-hydroxyl groups of the phenyl ring during the elaboration of monolignols (Whetten & Sederoff, 1995). The level of SAM synthetase is important in the maintenance of lignin biogenesis (Shen et al., 2002), and therefore the elevated transcript abundance observed for the gene encoding this protein in the ectopic lignification mutants, and in wild-type plants grown in the presence of sucrose, may reflect an increased requirement for this activity for lignin biosynthesis.
Candidate regulators of lignin deposition
When the 5′ noncoding regions of the 31 genes in the cluster described above were subjected to a Gibbs sampling-based analysis to identify statistically over-represented sequence motifs (Thijs et al., 2002), two motifs emerged that had previously known regulatory function. One of these motifs was similar to known R2R3-MYB binding sites, and the other is a known binding site for members of the Dof family of transcription factors. R2R3-MYB binding sites, also known as AC elements, have previously been shown to be present in the putative regulatory regions of all but one of the genes that define the lignin toolbox in A. thaliana (Raes et al., 2003), and have been shown to be important in the regulation of lignin biosynthetic gene expression in a number of different plant species (Hauffe et al., 1993; Hatton et al., 1995; Lacombe et al., 2000). Consistent with this, R2R3-MYB family members that can bind to these elements have been shown to have the capacity to modulate lignin deposition (Tamagnone et al., 1998; Patzlaff et al., 2003; Feng et al., 2004). To date, Dof binding sites have not been implicated in the regulation of lignin biosynthetic genes.
Cluster analysis was used to identify specific R2R3-MYB or Dof family members that might be involved in the regulation of the genes in the lignin toolbox under conditions that increase lignin deposition. Genes encoding two other classes of transcription factors postulated to be involved in lignin deposition, KNOX family members related to KNAT1/BREVIPEDICELLUS (Mele et al., 2003) and A. thaliana homologues of Nicotiana tobaccum LIM1 (Kawaoka et al., 2000), were also included in this analysis. As might have been predicted, all of the genes in the lignin toolbox had increased transcript abundance relative to the control in all three ectopic lignification mutants and in dark-grown wild-type plants in the presence of sugar. The genes in the lignin toolbox formed a cluster at one end of the Eisengram. Embedded in this cluster were genes encoding several R2R3-MYB family members (AtMYB84, At3g49690; AtMYB61, At1g09540; AtMYB92, At5g10280; AtMYB42, At4g12350) and one Dof family member (AtDof2.1, At2g34140). It might be predicted that the gene products encoded by these genes play a role in increasing the transcript abundance of the genes of the lignin toolbox. Consistent with this hypothesis, AtMYB61 has been shown to contribute to the ectopic lignin deposition observed in the det3 mutant (Newman et al., 2004). Clearly, AtMYB61 is not involved in the manifestation of the eli1 phenotype, as AtMYB61 expression is lower in this mutant relative to the other lignin accumulation conditions, and therefore, other regulators must be involved in giving rise to the eli1 phenotype.
While it is possible that lignin deposition is controlled solely by positive regulators of gene expression, it may also be that the downregulation of transcriptional repressors of lignin biosynthetic genes contributes to increased lignin deposition. Several genes, encoding either R2R3-MYB, Dof, or NtLIM-like transcription factors, all had decreased transcript abundance relative to the control in all three ectopic lignification mutants and in dark-grown wild-type plants in the presence of sugar. These genes may encode repressors of lignin deposition that, when they themselves are repressed, allow the expression of lignin biosynthetic genes to proceed apace. Consistent with this hypothesis, AtMYB4 (At1g22640), a known repressor of some of the genes involved in lignin biosynthesis (Jin et al., 2000), has decreased transcript abundance under those conditions that result in increased lignin deposition. It may be that the other transcription factors in this category function in the same manner.
The approach described herein has identified several putative regulators of the lignin biosynthetic pathway. Thorough gain-of-function/loss-of-function analyses has already shown that one of these regulators, AtMYB61, is both necessary and sufficient to explain the ectopic lignin deposition phenotype in the det3 mutant (Newman et al., 2004). Future studies will clarify the roles of the other candidate regulators in the control of lignin deposition.
We are very grateful to members of the Campbell laboratory for their kind assistance in various aspects of this work, particularly two undergraduate project students, Joanne Chick and Mark Walker for conducting the original mutant screen. We also extend our deepest gratitude to Drs Fiona Brew and Stephen Weber-Hall of Affymetrix UK Ltd, Dr Sean May and John Okyere at the Nottingham Arabidopsis Stock Centre, and the Genomic Arabidopsis Resource Network (GARNet) for outstanding assistance with the microarray experiments. John Baker provided excellent assistance with photography. We thank Drs Marie-Theres Hauser and Mike Bevan for providing mutant seeds for complementation analyses, and Marie-Theres Hauser for sharing data before publication. We are grateful to three anonymous reviewers for useful comments on the text. S.D.M. is a Canada Research Chair in Wood and Fibre Quality. This work was generously supported by funding from the Natural Science and Engineering Research Council of Canada (NSERC), funding by the Canadian Fund for Innovation (CFI) to S.D.M., by a studentship from the UK Biotechnology and Biological Sciences Research Council (BBSRC) to L.A.R. and by competitive grant funding from the BBSRC to M.M.C.