MYB103 is required for FERULATE-5-HYDROXYLASE expression and syringyl lignin biosynthesis in Arabidopsis stems


  • David Öhman,

    1. Department of Forest Genetics and Plant Physiology, Umeå Plant Science Centre, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author
    • These authors contributed equally to this paper.

  • Brecht Demedts,

    1. Department of Plant Systems Biology, Vlaams Instituut voor Biotechnologie, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Universiteit Gent, Gent, Belgium
    Search for more papers by this author
    • These authors contributed equally to this paper.

  • Manoj Kumar,

    1. Department of Forest Genetics and Plant Physiology, Umeå Plant Science Centre, Swedish University of Agricultural Sciences, Umeå, Sweden
    Current affiliation:
    1. Faculty of Life Sciences, University of Manchester, Manchester, UK
    Search for more papers by this author
  • Lorenz Gerber,

    1. Department of Forest Genetics and Plant Physiology, Umeå Plant Science Centre, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author
  • András Gorzsás,

    1. Department of Forest Genetics and Plant Physiology, Umeå Plant Science Centre, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author
  • Geert Goeminne,

    1. Department of Plant Systems Biology, Vlaams Instituut voor Biotechnologie, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Universiteit Gent, Gent, Belgium
    Search for more papers by this author
  • Mattias Hedenström,

    1. Department of Chemistry, Umeå University, Umeå, Sweden
    Search for more papers by this author
  • Brian Ellis,

    1. Michael Smith Laboratories, University of British Columbia, Vancouver, BC, Canada
    Search for more papers by this author
  • Wout Boerjan,

    1. Department of Plant Systems Biology, Vlaams Instituut voor Biotechnologie, Gent, Belgium
    2. Department of Plant Biotechnology and Bioinformatics, Universiteit Gent, Gent, Belgium
    Search for more papers by this author
  • Björn Sundberg

    Corresponding author
    • Department of Forest Genetics and Plant Physiology, Umeå Plant Science Centre, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author



The transcription factor MYB103 was previously identified as a member of the transcriptional network regulating secondary wall biosynthesis in xylem tissues of Arabidopsis, and was proposed to act on cellulose biosynthesis. It is a direct transcriptional target of the transcription factor SECONDARY WALL ASSOCIATED NAC DOMAIN PROTEIN 1 (SND1), and 35S-driven dominant repression or over-expression of MYB103 modifies secondary wall thickness. We identified two myb103 T-DNA insertion mutants and chemically characterized their lignocellulose by pyrolysis/GC/MS, 2D NMR, FT-IR microspectroscopy and wet chemistry. The mutants developed normally but exhibited a 70–75% decrease in syringyl (S) lignin. The level of guaiacyl (G) lignin was co-ordinately increased, so that total Klason lignin was not affected. The transcript abundance of FERULATE-5-HYDROXYLASE (F5H), the key gene in biosynthesis of S lignin, was strongly decreased in the myb103 mutants, and the metabolomes of the myb103 mutant and an F5H null mutant were very similar. Other than modification of the lignin S to G ratio, there were only very minor changes in the composition of secondary cell-wall polymers in the inflorescence stem. In conclusion, we demonstrate that F5H expression and hence biosynthesis of S lignin are dependent on MYB103.


Xylem cells typically deposit a secondary cell wall that makes up a considerable part of the biomass for industrial use provided by trees and bioenergy crops. The secondary wall is characterized by its ordered cellulose structure, organized in two or three layers with various microfibril angles, and by the presence of specific hemicelluloses, such as xylan, in dicot plants (Mellerowicz and Sundberg, 2008). Furthermore, secondary walled xylem cells are typically lignified throughout both the primary (compound middle lamella) and secondary wall layers. Formation of secondary walls utilizes a partly unique transcriptome for many cellular processes (Hertzberg et al., 2001; Geisler-Lee et al., 2006; Oakley et al., 2007; Zhong et al., 2010). However, there is a large plasticity in the ultra structure and chemical composition of secondary cell walls (in particular for fibres and parenchyma cells) (Mellerowicz and Sundberg, 2008; Gorzsás et al., 2011), which implies that subsets of the secondary wall transcriptome are likely to be under independent control.

Several regulators required for secondary wall formation in Arabidopsis have been identified. A null mutation of the tonoplast-located protein WALLS ARE THIN 1 (WAT1) caused a strong deficiency of secondary wall formation in fibres of both the inflorescence stem and the hypocotyl (Ranocha et al., 2010). SECONDARY WALL ASSOCIATED NAC DOMAIN PROTEIN 1 (SND1) and NAC SECONDARY WALL THICKENING PROMOTING FACTOR 1 (NST1) are NAC transcription factors (TFs) that act redundantly in the induction of secondary walls of fibres (Mitsuda et al., 2007; Zhong et al., 2007a). Fibres in both wat1 and the snd1 nst1 double mutant exhibit early phases of development, i.e. cell elongation and pointed tips, indicating that the corresponding proteins are specifically required for subsequent formation of the secondary wall. SND1 and NST1 are both down-regulated in wat1-1 plants, suggesting that WAT1 functions upstream of these NAC TFs (Ranocha et al., 2010). VASCULAR-RELATED NAC DOMAIN 6 (VND6) and VND7 are in the same sub-family of NAC TFs as SND1/NST1, and dominant repression of these genes specifically inhibits formation of vessel element secondary walls in metaxylem and protoxylem, respectively, in primary roots (Kubo et al., 2005). Together with other members of the VND family, they also appear to control vessel wall formation throughout the plant (Yamaguchi et al., 2008). Complementation studies have demonstrated that VND6/7 induce secondary cell walls in the snd1 nst1 double mutant background when driven by the SND1 promoter, suggesting that these TFs have similar function but operate in different cell types (Zhong et al., 2010; Yamaguchi et al., 2011). Based on these observations, SND1/NST1 and VND6/7 have been considered as ‘master switches’ for secondary wall formation in fibres and vessels, respectively.

A set of TFs from the MYB, NAC and KNAT families with vascular expression patterns were found to be down-regulated in SND1-RNAi plants (Zhong et al., 2008). Although down-regulation of these TFs by RNAi or T-DNA mutations did not induce any visible phenotype, in many cases their over-expression and dominant repression affected the secondary wall thickness of inter-fascicular fibres and xylem cells, suggesting a role as downstream targets of SND1/NST1 (Zhong et al., 2008). Among these TF genes, the MYB46 and MYB83 promoters are both direct targets of SND1/NST1 and VND6/7, and the myb46-1 myb83-1 double mutant is strongly dwarfed (Zhong et al., 2007b; McCarthy et al., 2009). As dominant repression of each gene reduced the wall thickness of both vessels and fibres, it was suggested that MYB46/83 act redundantly as regulators of secondary wall biosynthesis in both fibres and vessels. Downstream of MYB46/83, other TFs identified from SND1-RNAi plants have been proposed to regulate the biosynthesis of specific cell wall polymers, in particular MYB58, MYB63 and MYB85, which have been suggested to induce lignification (Zhong et al., 2008; Zhou et al., 2009). More recent studies explored direct target genes for SND1 and VND7 by using a global approach (Zhong et al., 2010; Yamaguchi et al., 2011). Induction of their expression from 35S-driven constructs in protoplasts, as well as in 10-day-old seedlings treated with cycloheximide, activated more than 100 genes (with large overlap between SND1 and VND7), including many TFs, secondary cell-wall biosynthesis genes and cell death-related genes.

MYB103 (At1g63910) was initially identified as one of the TFs that were down-regulated in SND1-RNAi plants (Zhong et al., 2008). It was shown to be a direct transcriptional target of SND1, NST1/2 and VND6/7, and promoter–GUS analysis demonstrated that it is expressed primarily in inter-fascicular fibres and xylem tissues. RNAi-based down-regulation of MYB103 did not result in any visible phenotype, but its over-expression resulted in thicker secondary cell walls, while expression of a dominant repression construct resulted in thinner walls. MYB103 was also shown to be a target of MYB46/83 in an in planta transactivation system (Nakano et al., 2010; Yamaguchi et al., 2010, 2011). As MYB103 activated the CESA8 promoter in a protoplast transactivation system, it was proposed that it may specifically act on cellulose biosynthesis (Zhong et al., 2008). Here, we have used a combination of pyrolysis/GC/MS (Py/GC/MS), 2D NMR, FT-IR microspectroscopy and wet chemistry to characterize the cell-wall chemotype of the inflorescence stem of myb103-1 and myb103-2 T-DNA mutants. We show that loss of functional MYB103 results in a 70–75% decrease in syringyl (S) lignin content accompanied by an equivalent increase in guaiacyl (G) lignin, whereas other wall polymers show only minor changes. The expression of FERULATE-5-HYDROXYLASE (F5H), which encodes the cytochrome P450 isoform that directs lignin biosynthesis towards formation of S lignin, was strongly decreased in both myb103 mutants, whereas expression of genes encoding other major lignin biosynthesis enzymes was not affected. Our results therefore demonstrate that MYB103 is required for F5H expression and biosynthesis of S lignin in the Arabidopsis inflorescence stem.


Identification of two mutant myb103 T-DNA lines

Two allelic myb103 (At1g63910) mutant lines were identified in the Columbia (Col-0) background. myb103-1 harbours a T-DNA insertion in the second exon, and myb103-2 harbours a T-DNA insertion located 163 bp upstream of the translational start codon (Figure 1a). The locations of both inserts were verified by genomic sequencing, and no additional disruption of nearby genomic DNA was detected in the vicinity of the inserts.

Figure 1.

Description of myb103 T-DNA insertion mutants. (a) Position of the T-DNA insertion sites in myb103-1 and myb103-2. Grey boxes, 5′ and 3′ untranslated (UTR) regions; black boxes, exons; lines, introns. Arrows indicate the position of left and right quantitative PCR primers. (b) Quantitative PCR analysis of MYB103 in basal stems of WT and mutants. Values are means ± SD for = 3 biological replicates, each consisting of six pooled stem segments. *< 0.05, ***< 0.001 (Student's t test) for comparison with WT.

Quantitative PCR analysis using primers targeting the third exon detected essentially no MYB103 transcripts in basal stem tissues of myb103-1, but no such reduction of MYB103 transcripts was detected in myb103-2 (Figure 1b). However, both mutants showed similar modifications in their stem tissue chemotype and their transcriptomes (described below). Sequencing of amplified transcript from myb103-2 showed that this also contained part of the T-DNA insert. The T-DNA contained a start codon located 201 bp upstream of the native start codon, as well as several stop codons downstream and in frame with the novel start codon (Figure S1a–d). This shows that, despite the apparent wild-type (WT) abundance of MYB103-derived transcripts in myb103-2, this transcript is unlikely to produce a functional MYB103. Allelic complementation analysis (described below) confirmed that this was the case. Both mutants showed normal development and stem anatomy (Figure S2a–c).

myb103 mutant lines are deficient in syringyl lignin units and show decreased expression of the FERULATE-5-HYDROXYLASE gene

Py/GC/MS analysis of the lower part of the inflorescence stem was used to screen for cell-wall modifications of myb103 mutants. This analysis provides a chemical fingerprint of the cell walls, and is particularly suited for rapid determination of lignin chemistry (Meier et al., 2005; Gerber et al., 2012). The Py/GC/MS chromatograms were evaluated by multivariate OPLS-DA (orthogonal projections to latent structures discriminant analysis) (Bylesjö et al., 2006), which yielded good models for separation of the two mutants from WT (Figure S3). Integration of peaks originating from lignin showed a 70–75% reduction in S lignin, with a corresponding increase in G lignin for both mutant alleles (Table 1). To confirm these alterations in S and G lignin, and to detect other possible cell-wall modifications, the dissolved stem tissue was analysed by 2D-NMR combined with OPLS-DA, as described by Hedenström et al. (2009). The scores plot showed clear separation between the two mutants and WT (Figure 2a). The loadings revealed that the peaks from S and G lignin in the aromatic region of the spectra, together with the peak from O-methoxy groups, were the most important variables responsible for the separation between mutants and WT observed in the model, and that S lignin was decreased in the mutants (Figure 2b). When a lower cut-off filter was applied to reveal minor contributors to the observed separation between mutants and WT, peaks corresponding to glucomannan and cellulose (both decreased in the mutant) and xylan (increased in the mutant) were observed for both mutants (Figure S4).

Figure 2.

2D NMR analysis of dissolved cell-wall material from basal stems of WT, myb103-1 and myb103-2 plants. (a) OPLS-DA scores plot showing the separation between WT and mutants. Each symbol represents a biological replicate consisting of 50 pooled basal stem segments. Model details: one predictive + two orthogonal components, total variance explained, R2Y(cum) = 1, predictability, Q2(cum) = 0.88. (b) Loadings plot for the predictive component from the OPLS-DA model with WT and myb103-1. Black peaks have higher relative intensity in WT, and green peaks have higher relative intensity in myb103-1. A cut-off filter of 0.95 was applied to only show the variables with the highest weight in the model. The loadings plot for myb103-2 showed similar peaks (Figure S4). G, guaiacyl lignin; Met, O-methyl (mainly from syringyl lignin); S, syringyl lignin.

Table 1. Lignin and cellulose content of basal stem cell wall material (CWM) from WT, myb103-1 and myb103-2 plants
GenotypeSyringyl (%)Guaiacyl (%)Klason lignin (mg g−1 CWM)Updegraff cellulose (mg g−1 CWM)
  1. Syringyl and guaiacyl lignin units were estimated using Py/GC/MS and are expressed as percentages of total ion counts. Values are means ± SD for = 6 biological replicates, each consisting of an individual plant for Py/GC/MS, and of 25 pooled plants for Klason lignin and Updegraff cellulose determinations. ***< 0.001 (Student's t test) for comparison with WT.

WT4.1 ± 0.314.8 ± 0.4125 ± 13267 ± 25
myb103-1 1.1 ± 0.1***20.0 ± 0.3***129 ± 35248 ± 45
myb103-2 1.2 ± 0.1***19.5 ± 0.3***119 ± 13244 ± 14

The same plant material was further analysed for Klason lignin, monosaccharides released after treatment with 2 m trifluoroacetic acid and Updegraff cellulose (Tables 1 and 2). The Klason lignin content did not differ between mutants and WT, demonstrating that the decrease in S lignin was essentially offset by an increase in G lignin. Similarly, no significant difference was observed for Updegraff cellulose, although there was a trend towards less cellulose in the mutants (Table 1). The monosaccharide analysis showed a trend for increased xylose and decreased mannose (not significant) in both mutants, but increases were observed in fucose, rhamnose, arabinose, galactose and galacturonic acid in myb103-2 only (Table 2). Finally, FT-IR microspectroscopy using a focal plane array detector, combined with OPLS-DA, was used to study chemical modifications specifically in vessel elements, xylem fibres and inter-fascicular fibres of the inflorescence stem, as described by Gorzsás et al. (2011). In WT plants, the chemotypes of all three cell types showed a clear separation (Figure 3a,b). Pairwise OPLS-DA between cell types provided good models for each comparison (Figure S5), and a loadings plot showed a higher proportion of S lignin in inter-fascicular fibres and a higher proportion of G lignin in xylem fibres and vessel elements, as deduced from the relative intensities of the aromatic –C=C– bands at 1510 and 1595 cm−1 (Figure 3c–e). The results also show higher relative lignin amounts in vessel elements compared to xylem fibres, as the positive correlation of the 1510 cm−1 band is not countered by a negative correlation of the 1595 cm−1 band in Figure 3(c). The sugar composition/structure also differed between cell types, as concluded from the carbohydrate-related bands (950–1070 cm−1 region, marked in grey) and the –C=O band at 1740 cm−1 (Figure 3c–e). Although the various carbohydrate-related bands (950–1070 cm−1) in the FT-IR spectra are less conclusive with regard to their origin in cellulose or hemicellulose, the loading corresponding to the –C=O band (which is non-existent in cellulose) at 1740 cm−1 suggests that xylem fibres are rich in hemicelluloses (Figure 3c,d). When myb103-1 was compared with WT in an OPLS-DA, it was found to separate well for each cell type (Figure 4a–c). A similar pattern of bands contributed to the separation in all cell types (Figure 4d–f), demonstrating similar cell-wall modification in them. This is consistent with the observation that MYB103 is expressed in inter-fascicular fibres, xylem fibres and vessel elements (Zhong et al., 2008). A set of bands that were more intense in myb103-1 (1140, 1250 and 1510 cm−1) are all related to G lignin (Faix, 1991a), and reflect a major shift in lignin type in the mutant.

Figure 3.

FT-IR microspectroscopic analysis of various xylem cell types in WT plants. (a) Transverse sections showing the positions of sampled cell types. IF, inter-fascicular fibres (triangles); XF, xylem fibres (squares); VE, vessel elements (circles). (b) OPLS-DA scores plot showing the separation between IF, XF and VE. Each symbol represents one cell. Spectra were collected from three plants with ten spectra/plant. (c–e) Loadings plots for the predictive component, showing (c) factors separating VE from XF: bands with positive loadings are more intense in VE, whereas bands with negative loadings are more intense in XF; (d) factors separating IF from XF: bands with positive loadings are more intense in IF, whereas bands with negative loadings are more intense in XF; (e) factors separating VE from IF: bands with positive loadings are more intense in VE, whereas bands with negative loadings are more intense in IF. 950–1070 cm−1, unspecific carbohydrate vibrations (McCann et al., 1993); 1140 cm−1, asymmetric –COC– stretch (McCann et al., 1993) with contribution from aromatic –CH in-plane deformation of G-type lignin (Faix, 1991a); 1510 and 1595 cm−1, aromatic –C=C– vibrations with higher 1510/1595 cm−1 band ratios, indicating more G-type lignin (Faix, 1991a); 1740 cm−1, –C=O vibration, absent in cellulose.

Figure 4.

FT-IR microspectroscopic analysis of xylem cell types in WT and myb103-1. (a–c) OPLS-DA scores plots showing the separation between WT and myb103-1 in vessel elements (a), inter-fascicular fibres (b) and xylem fibres (c). Each symbol represents one cell. Spectra were collected from three plants per genotype, with ten spectra/plant. Model details: vessel elements (a): one predictive + seven orthogonal components, R2Y(cum) = 0.77, Q2(cum) = 0.52; inter-fascicular fibres (b): one predictive + two orthogonal components, R2Y(cum) = 0.74, Q2(cum) = 0.67; xylem fibres (c): one predictive + six orthogonal components, R2Y(cum) = 0.78, Q2(cum) = 0.63. (d–f) Loadings plot for the predictive component, showing factors separating WT from myb103-1 in vessel elements (d), inter-fascicular fibres (e) and xylem fibres (f). Positive bands are more intense in myb103-1, whereas negative bands are more intense in WT. 1250 cm−1, centre of a composite band, including –CC–, –CO and –C=O vibrations, involving the G-type lignin ring (Faix, 1991a). For explanation of other marked bands, see Figure 3.

Table 2. Monosaccharide content of basal stem cell-wall material (CWM) from WT, myb103-1 and myb103-2 plants
GenotypeMonosaccharide (mg g−1 CWM)
  1. Fuc, fucose; Rha, rhamnose; Ara, arabinose; Gal, galactose; Glc, glucose; Xyl, xylose; Man, mannose; GalAc, galacturonic acid; GlcAc, glucuronic acid.

  2. Values are means ± SD for = 6 biological replicates, each consisting of 25 pooled plants. *< 0.05, **< 0.01, ***< 0.001 (Student's t test) for comparison with WT.

WT1.7 ± 0.25.5 ± 0.37.6 ± 0.413.3 ± 0.410.9 ± 0.787.5 ± 0.77.3 ± 0.815.5 ± 0.41.3 ± 0.1
myb103-1 1.9 ± 0.56.3 ± 0.68.6 ± 0.414.0 ± 0.810.9 ± 0.792.5 ± 1.45.8 ± 0.717.1 ± 0.81.0 ± 0.2*
myb103-2 2.7 ± 0.3**6.8 ± 0.2**13.7 ± 0.5***16.9 ± 0.5***12.8 ± 0.690.2 ± 0.76.0 ± 0.520.8 ± 0.8**1.3 ± 0.2

Taken together, it may be concluded that the mutant alleles of MYB103 cause a strong decrease (70–75%) in S lignin, with a corresponding increase in G lignin such that the proportion of total lignin was not altered. The NMR data also show a decrease in the proportion of cellulose and glucomannan, and an increase in the proportion of xylan in both mutant alleles. However, these differences were small and not statistically different in the less sensitive wet chemistry analysis. The increase in fucose, rhamnose, arabinose, galactose and galacturonic acid observed in myb103-2 hydrolysates was not seen in myb103-1, and we therefore do not directly associate these changes with loss of MYB103 function.

F5H is the enzyme responsible for directing the products of the phenylpropanoid pathway toward S lignin biosynthesis, and null mutants in F5H fail to accumulate S lignin (Chapple et al., 1992; Meyer et al., 1996, 1998). When we performed quantitative PCR analysis of F5H expression in myb103-1 and myb103-2 plants, we found that F5H transcript abundance was strongly reduced in both mutants (Figure 5). Allelic complementation of the two mutants by reciprocal crossing generated F1 heterozygous offspring that exhibited the same reduced level of F5H expression and S lignin content as the homozygous parental lines (Figure S1g,h). This strongly suggests that F5H expression is dependent on MYB103, and is consistent with the decrease in S lignin observed in the myb103 mutants. However, the residual F5H expression (and S lignin content) in myb103 mutants shows that other factors are also involved in its transcriptional regulation, and that its dependence on MYB103 may be indirect.

Figure 5.

Quantitative PCR analysis of F5H, CESA4, CESA7 and CESA8 in basal stem segments of WT, myb103-1 and myb103-2. Values are the means ± SE for = 3 biological replicates, each consisting of six pooled basal stem segments. *< 0.05, **< 0.01 (Student's t test) for comparison with WT.

Metabolite analysis of myb103 shows specific effects on lignin biosynthesis

To further investigate the specificity of the effect of the myb103 mutation on S lignin biosynthesis, we compared the metabolome of myb103-1 with that of f5h1-4, a null mutant of F5H. A principal component analysis scores plot shows that, although the metabolomes of both myb103-1 and f5h1-4 differ from WT, they also differ from each other (Figure 6a). The major loading separating f5h1-4 from WT is the absence of sinapoyl malate in f5h1-4 plants (Figure S6), which is in line with the role of F5H in the biosynthesis of sinapic acid esters (Ruegger et al., 1999). Sinapoyl malate accumulation is not affected in the myb103-1 mutant (Figure S6). However, when only oligolignols originating from oxidative coupling of coniferyl and sinapyl alcohol are analysed, the profiles of myb103-1 and f5h1-4 are similar (Figure 6b,c). In both mutants, increased levels of all detected oligolignols containing only G units were observed, while the levels of oligolignols containing S units are lower. Because biosynthesis of S lignin is not completely blocked in the myb103 mutant and the level of G units is increased, the proportion of S and G lignin in the oligolignols determines the extent of the decrease in these mixed oligolignols. In this respect, it is worth noting that trilignols with two S units and one G unit are significantly decreased in the myb103 mutant, whereas, for oligolignols with relatively more G units, only a trend to lower levels is seen in most cases. Taken together, the metabolite profiling data show that the metabolites most affected in myb103 plants are related to biosynthesis of S and G monolignols, consistent with the cell-wall chemistry data and with a model in which MYB103 is involved in the regulation of F5H expression. As sinapoyl malate is produced predominantly in epidermal tissue (Li et al., 2010) and lignin is produced predominantly in xylem tissues, we believe that the specific effect of MYB103 on lignin-related monolignols reflects its xylem-specific expression pattern, demonstrated by Zhong et al. (2008). This may be predicted to lead to tissue-specific down-regulation of F5H activity, although expression analysis in isolated tissues is required to support this.

Figure 6.

Metabolite analysis in basal stem segments of WT, myb103-1 and f5h1-4 plants. (a) Principal component analysis scores plot showing separation between WT, myb103-1 and f5h1-4. Each symbol represents a biological replicate. (b) G lignin-containing oligolignols accumulate in the mutants. Values are means ± SE for = 8 biological replicates. **< 0.01, ***< 0.001 (Student's t test) for comparison with WT. Nomenclature of oligolignols is according to Morreel et al. (2004). Identical labelling reflects different stereoisomers. (c) S lignin-containing oligolignols are reduced in the mutants. Values are means ± SE for = 8 biological replicates. **< 0.01, ***< 0.001 (Student's t test) for comparison with WT. Identical labelling reflects different stereoisomers.

The transcriptome of myb103 mutant lines reveals mis-regulation of several secondary wall-related TFs and biosynthesis genes

A microarray experiment was performed using both myb103 mutants to provide an overview of the effect of the mutation on their transcriptomes. Genes whose expression was more than twofold different between WT and either of the two myb103 mutants (of which the majority were down-regulated) and that were reasonably highly expressed (signal strength >50) produced a list of 40 genes. All of these transcripts were found to respond in a similar fashion in both mutants, supporting a model in which their mis-regulation was caused by mutation of MYB103 (Table 3). The marked down-regulation of F5H was confirmed, and it was one of the most down-regulated genes. In addition, a number of TFs and cell wall-related genes were down-regulated in the myb103 mutants. Of these, MYB20, MYB63, MYB69, SND2 and SND3 are also hypothesized to be part of the SND1/NST1 transcriptional network regulating secondary wall biosynthesis (McCarthy et al., 2009). Among the down-regulated cell wall-related genes were MAP65-8 and MAP70-5, which encode two microtubule-associated proteins that are thought to be important in secondary wall patterning (Pesquet et al., 2010; Oda and Fukuda, 2012), LACCASE17, which has been shown to be important for lignification (Berthet et al., 2011), and CELLULOSE SYNTHASE LIKE 9, which is thought to be involved in mannan biosynthesis (Goubet et al., 2009; Davis et al., 2010). Other than F5H, the microarray analysis did not show any other genes encoding enzymes involved in lignin biosynthesis that were significantly mis-regulated, and nor were any of the secondary wall ‘master switch’ genes. It is noteworthy that, although MYB103 was previously shown to induce expression of CESA8 in a promoter transactivation assay (Zhong et al., 2008), none of the secondary wall CESA genes were found to be significantly mis-regulated in the microarray analysis of the myb103 mutants. However, when the expression level of all secondary wall-associated CESA genes was quantified by quantitative PCR analysis, we detected a small down-regulation of CESA4 in myb103-2 (Figure 5). When this is considered together with the small decrease in cellulose in myb103 plants indicated in the NMR analysis, a role for MYB103 in cellulose biosynthesis cannot be excluded.

Table 3. Microarray data showing the most mis-regulated genes for myb103-1 and myb103-2 mutants in comparison with WT
IDTAIR9 annotationmyb103-1 (log2 value)myb103-2 (log2 value)
  1. The list shows genes that are mis-regulated more than twofold in either of the two mutants, with signal intensity values >50. Values are the mean of three biological replicates, each consisting of six pooled stem sections. The expression values for all genes listed were significantly different from WT (Student's t test, < 0.05).

AT3G62730Unknown protein−3.7−3.3
AT2G16980Tetracycline transporter−1.6−0.9
AT1G56720Protein kinase family protein−1.6−1.8
AT3G50220Unknown protein−1.5−1.4
AT1G28470SND3 (ANAC010)−1.4−1.0
AT2G20680Glycosyl hydrolase family 5 protein/cellulase family protein−1.4−1.3
AT4G28500SND2 (ANAC073)−1.4−1.0
AT1G08340Rac GTPase-activating protein, putative−1.3−1.2
AT1G05340Unknown protein−1.2−1.7
AT5G23750Remorin family protein−1.2−0.9
AT3G14280Unknown protein−1.1−1.6
AT4G08160Glycosyl hydrolase family 10 protein/carbohydrate-binding domain-containing protein−1.1−0.9
AT1G79420Unknown protein−1.1−1.1
AT3G23090Unknown protein−1.0−0.8
AT1G05170Galactosyltransferase family protein−0.9−1.1
AT4G14760Unknown protein−0.5−1.3
AT4G31140Glycosyl hydrolase family 17 protein0.61.0
AT1G31720Unknown protein0.71.0
AT4G26220CCoAOMT, putative0.81.1
AT2G19970Pathogenesis-related protein, putative1.00.8
AT1G22890Unknown protein1.11.1
AT1G18980Germin-like protein, putative1.21.1
AT5G25840Unknown protein1.51.7

MYB103 does not transactivate the F5H promoter

A promoter transactivation assay was used to investigate the potential interaction between MYB103, or the MYB and NAC TFs mis-regulated in myb103 plants, and promoters representing the most important genes in the early phenylpropanoid and monolignol biosynthesis pathways (Table 4). MYB103 did not show strong transactivation of any of the promoters tested. Among the other TFs examined, only SND2 had a small positive effect on the F5H promoter, while MYB69 had a similar small negative effect. The most striking interactions were those mediated by MYB20 and MYB63, which both showed strong positive regulation of several of the lignin biosynthesis promoters. Thus, the transactivation assay does not provide any evidence in support of a simple mechanism for specific down-regulation of F5H in the myb103 mutants. It should also be noted that we did not find any interaction between SND1 and the Arabidopsis F5H promoter, whereas SND1 did transactivate the Medicago F5H promoter, as previously reported by Zhao et al. (2010b).

Table 4. Protoplast transactivation analysis
  1. Values are log2 values for the fold up-regulation (activation) or down-regulation (repression) of the firefly luciferase reporter by the 35S-driven effector compared to control reporter activity without the effector. Reporter activities were normalized to the activities from a 35S-driven Renilla luciferase construct. Values are the means of four independent experiments. Numbers indicate differences between the effector and the control, that were significantly different (< 0.05, Student's t test). Analyses where no significant difference was found are indicated with a dash.



Lignin in xylem cell walls of angiosperms is to a major extent formed by polymerization of coniferyl and sinapyl alcohol, which differ with regard to the presence of a second O methyl group at the 5-position of the ring in the S monomer (Bonawitz and Chapple, 2010; Vanholme et al., 2010). The proportion of S and G units in the lignin polymer is important for its chemical properties, and affects both the processing properties of lignocellulose for industrial use and its forage digestibility. The molecular mechanisms that control the lignin S/G ratio have therefore attracted considerable interest (Baucher et al., 2003). There is a large variation in the S/G ratio of lignin between taxonomic plant groups, but also between individual plants of a species and even between different cell types in the xylem, as a combined result of genetic and environmental regulation. In the inflorescence stem of Arabidopsis, for example, the cell walls of inter-fascicular fibres are rich in S lignin relative to the vessel elements and xylem fibres in the vascular bundle (Figure 3) (Meyer et al., 1998; Patten et al., 2010). S-type lignin monomers are formed by a branch in the monolignol pathway, whereby the C5 position of the aromatic ring is hydroxylated by the activity of F5H (Meyer et al., 1998). Null F5H mutants have almost no S lignin, whereas over-expression of F5H in Arabidopsis and poplar results in xylem cells with highly enriched S lignin, demonstrating that expression of F5H is a main determinant of the lignin S/G ratio (Marita et al., 1999; Huntley et al., 2003). However, aside from the genetic dependence on F5H, little is known about the molecular mechanisms that regulate the S/G ratio or F5H expression in developing secondary xylem walls. We show here that a null mutation in the MYB103 transcription factor gene results in a large reduction of S lignin in the Arabidopsis stem and a strong decrease in F5H expression. We also show that the metabolic profile of oligolignols in myb103 plants mimics that of an F5H null mutant, but with weaker effects, in line with the incomplete down-regulation of F5H. Overall, this indicates a specific effect of MYB103 on F5H expression rather than on general lignin biosynthesis, as supported by our microarray experiments, and demonstrates that MYB103 is important in regulating lignin composition in Arabidopsis.

In Medicago truncatula, it was recently shown that loss of function of MtNST1 (a Medicago homologue of SND1/NST1) results in a phenotype similar to the Arabidopsis snd1 nst1 double mutant, including a lack of secondary cell-wall formation in the inter-fascicular fibres and xylem fibres (Zhao et al., 2010a). It was further demonstrated that the stem of the Mtnst1 mutant plants contained much lower levels of S lignin, and that AtSND1 bound the MtF5H promoter (Zhao et al., 2010b). Based on these observations, it was concluded that SND1 directly regulates F5H expression and is required for biosynthesis of S lignin in angiosperms. However, the decrease in the proportion of S lignin in the Mtnst1 mutant may also be explained by its lack of secondary-walled inter-fascicular fibres, which, at least in Arabidopsis, are the main source of S lignin in stem tissues (Meyer et al., 1998; Patten et al., 2010), rather than specific regulation of biosynthesis of S lignin by MtNST1. Moreover, the promoter binding and transactivation assays in the study by Zhao et al. (2010b) were performed using the Arabidopsis SND1 gene as an effector and the Medicago F5H promoter in the reporter construct. This is a potential problem as there is considerable sequence difference between the Arabidopsis and Medicago F5H promoters, which is reflected in the different number of TF binding motifs when comparing the two promoters (Table S1). For example, the Arabidopsis F5H promoter lacks the MYB1AT promoter-binding element (a type of MYB-binding AC element), whereas these motifs are abundant in the Medicago promoter (Figure S7a). Therefore, the observed binding of AtSND1 to the Medicago F5H promoter cannot be assumed to predict a similar affinity for the Arabidopsis F5H promoter. Indeed, AtSND1 activated the Medicago F5H promoter but not the Arabidopsis F5H promoter when assayed side by side in our transactivation assay (Figure S7b). It is also noteworthy that Zhong et al. (2010) did not report any activation of F5H in their global analysis of direct targets of AtSND1. Thus, the conclusion that SND1 acts as a direct activator of F5H and biosynthesis of S lignin appears not to be a general one.

MYB103 has also been suggested to act specifically on cellulose biosynthesis, because it was observed to activate the CESA8 promoter in a protoplast transactivation assay (Zhong et al., 2008). Our 2D NMR data from myb103 plants analysed by OPLS-DA did show a decrease in peaks corresponding to cellulose. However, these peaks contributed less to the OPLS-DA model that separates the myb103 mutants from the WT than the peaks originating from S and G lignin did (Figure S4). The fact that the decrease in cellulose in myb103 plants revealed by 2D NMR is minor is also supported by the wet chemistry analysis, which, at the number of replicates used, only showed a statistically insignificant trend towards a decreased proportion of cellulose (Table 1). Quantitative PCR analysis of all the secondary wall CESA genes similarly showed a small decrease in CESA4 expression only (Figure 5). We cannot eliminate the possibility that MYB103 does activate CESA8 expression, but this activation may be redundant with other activators and thus the final expression level is not affected, or it may be too small to be readily confirmed by quantitative PCR. On the other hand, the possibility cannot be excluded that the reported activation of the CESA8 promoter by MYB103 reflects non-specific binding due to high protein abundance resulting from 35S-driven over-expression, rather than its true function in planta. Taken together, our data suggest that, in addition to the major effect on the lignin S/G ratio, myb103 mutants only display minor effects on biosynthesis of the carbohydrate matrix.

Dominant repression of MYB103 results in thinner cell walls, and its over-expression leads to thicker walls (Zhong et al., 2008). This may reflect additional roles of MYB103 in cell-wall biosynthesis that were not observed in myb103 mutants because of functional redundancy of MYB103 with other protein(s). Indeed, the transcriptome analysis of the myb103 mutants does suggest a more complex role for MYB103. Among the more abundant transcripts that were mis-regulated were a number of TF and cell wall-related genes, as well as genes with unknown function (Table 3). Some of the mis-regulated genes may be more indirectly related to biosynthesis of cell-wall components, but it is also possible that MYB103 has other cellular functions that were not detected by the analyses used in this study.

We did not find any interaction between MYB103 and the Arabidopsis F5H promoter in our protoplast transactivation system (Table 4). This indicates that the down-regulation of F5H expression in myb103 mutants is indirect, possibly mediated through some of the mis-regulated TFs in the myb103 mutant. However, none of the independently tested TFs had significant effects on the F5H promoter in the transactivation assay, and it therefore appears unlikely that these TFs mediate the MYB103 effect on F5H expression. Interestingly, MYB20 and MYB63 were very efficient activators of promoters for most genes in the monolignol biosynthesis pathway other than F5H. Although MYB63 has been suggested to be specifically involved in lignin biosynthesis (Zhou et al., 2009), the function of MYB20 has not yet been described.

It has been suggested that the cascade of TFs that are induced, directly or indirectly, by the ‘master switches’ have different functions in regulating the biosynthesis of specific cell-wall biopolymers (Zhong et al., 2008). MYB58, MYB63 and MYB85 have all been suggested to be specific activators for lignin biosynthesis. This proposal is based on the observation that their over-expression induces ectopic lignification without biosynthesis of other cell-wall polymers in parenchyma cells, and (for MYB58 and MYB63) that dominant repression and RNAi down-regulation of both genes result in thinner cell walls and a 40% decrease in total lignin (Zhong et al., 2008; Zhou et al., 2009). It has also been shown that MYB58 directly targets several lignin biosynthesis genes. It is not known whether MYB58 also affects other cell-wall biosynthesis genes and biosynthesis of other cell-wall polymers. Our data provide evidence that a null allele of MYB103 results in a strong reduction in the proportion of S lignin in secondary cell walls, as a consequence of decreased F5H expression. However, 2D NMR analysis of dissolved cell walls shows additional minor modifications of the cell-wall polymers, and transcriptome analysis shows alterations in the expression of several TFs that are important for secondary wall biosynthesis, as well as of cell wall biosynthesis-related genes other than F5H. These observations suggest, in agreement with recently published data (Zhao and Dixon, 2011), that transcriptional regulation of secondary cell-wall synthesis may be more complex than previously thought.

Experimental procedures

Plant material and growth conditions

myb103-1 (GABI-361D06) and myb103-2 (SAIL_337_C03) in the Col-0 background were obtained from the Nottingham Arabidopsis Stock Centre ( Homozygous lines were identified using gene-specific primers (Table S2). To localize the insert, the flanking DNA was sequenced by Eurofins MWG Operon ( The f5h1-4 T-DNA mutant was obtained as a homozygote from the GABI-KAT collection at the Nottingham Arabidopsis Stock Centre. Homozygous seeds were obtained and plants were grown in soil (16 h light/8 h dark, 24°C/19°C; 150 μE m−2 sec−1; 70% humidity). For the metabolite experiment, plants were grown under 8 h light/16 h dark for 6 weeks, followed by 16 h light/8 h dark for 5 weeks. In all experiments, the position of the various genotypes was carefully randomized during growth.

RNA isolation and transcript analysis by quantitative PCR

The basal 1–9 cm of 25 cm high stems was frozen and homogenized. Total RNA was isolated and DNase-treated using an Aurum total RNA mini kit (Bio-Rad,, and used to generate cDNA using an iScript cDNA synthesis kit (Bio-Rad). The reaction mixture for quantitative PCR contained 10 μl 2× iQ SYBR Green supermix (Bio-Rad), 8 μl nuclease-free H2O, 1 μl of a mixture of 10 μm forward/reverse primers, and 1 μl cDNA template (diluted 1:10). Analysis was performed using a Bio-Rad CFX96 real-time PCR detection system; 40 quantitative PCR cycles were run under the following conditions: denaturation step, 95°C for 10 sec; annealing step, 55°C for 10 sec; elongation step, 72°C for 30 sec. ELONGATION FACTOR 1-α (EF1-α) (At5g60390) and POLYUBIQUITIN 10 (UBQ10) (At4g05320) were used as reference genes. Primers are listed in Table S3.

Microarray analysis

Total RNA (prepared as described above) was purified and concentrated using an RNeasy MinElute clean-up kit (Qiagen, RNA quality and quantity were checked using an Agilent 2100 bioanalyzer ( and a Thermo Scientific NanoDrop 2000. (, respectively, and 4 μg of each sample was submitted to ATLAS Biolabs ( for whole-transcriptome analysis using GeneChip Arabidopsis ATH1 genome arrays (Affymetrix, http://www. RNA for myb103-1 and myb103-2 was hybridized against WT with three biological replicates per genotype, each sample consisting of tissue from six pooled stems. The microarray data were used here to identify the most mis-regulated genes with signal strength >50, and have been deposited at the EMBL/EBI database ArrayExpress Archive ( under accession number E-MEXP-3680.

Chemical analysis

Basal 1–4 cm stem segments were harvested from 35 cm high plants for Py/GC/MS and FT-IR, or senesced plants, approximately 43 cm in height, for 2D NMR and wet chemistry, freeze-dried and ball-milled using an MM400 mixer mil (Retsch,; 2 min, 30 Hz, 1 ml stainless steel jars, ball diameter 7 mm). Cell-wall material was obtained by extracting twice with a solution of 80% ethanol and twice with a solution of 50% ethanol, each containing 4 mm HEPES buffer, pH 7.5. Samples were incubated at 80°C for 30 min between each sequential extraction step. Pelleted cell-wall material was washed in acetone and dried completely in a centrifugal evaporator.

Py/GC/MS analysis was performed as described by Gerber et al. (2012). Briefly, samples were pyrolysed at 450°C using an on-line pyrolizer (PY-2020iD and AS1020E, FrontierLab ) mounted on an 7890A/5975C GC/MS (Agilent Technologies) equipped with a J&W DB-5 capillary column, 30 m length, diameter 250 μm, 25 μm film thickness (Agilent Technologies). The temperature program was 40°C initially, 32°C min−1 to 100°C, 6°C min−1 to 118.75°C, 15°C min−1 to 250°C, and 32°C min−1 to 320°C. The total run time was 19 min, and full-scan spectra were recorded in the range 35–250 m/z. Data processing, including peak detection, integration, normalization and identification, was performed as described by Gerber et al. (2012). Chemical fingerprints, consisting of the integrated peaks, were then subjected to multivariate data analysis by OPLS-DA. Quantitative estimates of S- and G-type lignin were calculated based on integrated peak areas from selected m/z channels, as described by Faix et al. (1991b).

2D NMR analysis of acetylated cell-wall material was performed as described by Hedenström et al. (2009) with a few modifications. Briefly, 2D 13C–1H HMQC spectra (pulse program hmqcetgp) were acquired at 25°C on a Bruker DRX 600 MHz instrument (Bruker Biospin GmbH, equipped with a triple-resonance cryo-probe with z-gradients. Forty-eight transients, 320 t1 increments and an inter-scan delay of 1 sec resulted in an experiment duration of approximately 5 h for each spectrum. Spectra were manually phase- and baseline-corrected using Topspin 2.0 (Bruker Biospin GmbH). The residual CHCl3 peak was used as an internal reference (δC 77.0 ppm, δH 7.27 ppm). The spectra were prepared for multivariate analysis using an in-house MATLAB script (Hedenström et al., 2009). Loading vectors from the OPLS-DA models calculated from variables scaled to unit variance were imported to MATLAB, and variables (individual data points in the spectra) with a correlation lower than a manually defined threshold were set to zero. This is the cut-off value referred to in Figures 2b and S4. The remaining variables were multiplied by their standard deviation and the resulting vector was transformed into a 2D loading plot with the same dimensions as the original spectra. This back-scaling approach has previously been used for 1H NMR spectra (Cloarec et al., 2005), and results in loading plots that are more readily interpretable compared to loading plots constructed directly from the correlations themselves. MATLAB scripts are available from the authors upon request.

FT-IR microspectroscopy was performed as described by Gorzsás et al. (2011). Briefly, 20 μm thick cross-sections were cryo-sectioned from the stem sampled at 1 cm above the soil. Spectra were recorded at a spectral resolution of 4 cm−1 using a Bruker Tensor 27 spectrometer equipped with a Hyperion 3000 microscopy accessory, including a 64 × 64 focal plane array detector (Bruker Optik GmbH, Cell-specific spectra were extracted, baseline-corrected and area-normalized in the region 900–1850 cm−1, using custom-built MATLAB scripts (Stenlund et al., 2008), before multivariate analysis. Spectra were converted to data point tables using opus version 5.0.53 (Bruker Optik GmbH). Data processing was performed using custom scripts programmed within matlab software version 7.0 (Mathworks,

Chemical fingerprints yielded by FT-IR microspectroscopy, 2D NMR and Py/GC/MS were evaluated by principal component analysis and OPLS-DA (Bylesjö et al., 2006). Principal component analysis and OPLS-DA models were created using simca-p+ version 12.0 (Umetrics,

Wet chemical analysis was performed on cell-wall material. Klason lignin was measured as described by Ona et al. (1995), and crystalline cellulose was measured as described by Updegraff (1969) and Scott and Melvin (1953). For quantification of the cell-wall monosaccharides, 10 mg cell-wall material was hydrolysed in 2 m trifluoroacetic acid for 1 h at 120°C. Samples were dried completely in a centrifugal evaporator overnight to evaporate the trifluoroacetic acid. Cell-wall material not hydrolysed by trifluoroacetic acid was resuspended in Milli-Q water (Millipore ) and pelleted by centrifugation (20800 RCF, room temperature, 10 min). The upper layer of the supernatant was carefully collected and transferred to LC vials. The cleaned-up monosaccharides were separated by HPAEC on a CarboPac PA20 (Dio nex, in two separate runs, as described by Currie and Perry (2006). Quantification by pulsed amperometric detection using a MetrOhm 850 professional ion-chromatograph ( was performed using external calibration standards for all detected sugars.

Metabolome analysis

Basal 1–11 cm segments of 22–28 cm high stems were analysed by LC-MS, as described in Data S1. The instrument used was a Waters Acquity UPLC system (, equipped with an Acquity BEH C18 column inner diameter, 2.1 mm; length, 150 mm; particle size, 1.7 μm (Waters Corp) connected to a Synapt HDMS Q-TOF mass spectrometer equipped with an electrospray ionization source and lockspray interface (Waters).

Protoplast transactivation assay

The protoplast transactivation assay was performed as previously described (De Sutter et al., 2005), using 500 protoplasts per μl. Briefly, tobacco BY-2 protoplasts were transformed in a Ca2+/poly(ethylene glycol)-dependent manner using 2 μg of each effector, reporter and normalization vector. Protoplasts were incubated for 24 h before lysis and read-out of the luciferase activities, using a dual luciferase kit (Promega, As a positive control, MYB63 was used and showed equally high activation of lignin promoters as described previously (Zhou et al., 2009). Reporter activities were normalized to the activities from a 35S-driven Renilla luciferase construct. As an internal control, the reporter and normalization vector were used in the absence of an effector transcription factor.

The promoter of F5H from Medicago truncatula was cloned from genomic DNA using the primers described in Zhao et al. (2010b) but with adaptor sequences added to the original primers (Data S1), into pDONR P4P1R (Invitrogen,, and sequenced. Arabidopsis thaliana promoter amplicons were obtained from the Benhamed et al. (2008), cloned into pDONR P4P1R (Invitrogen) and sequenced. Promoter entry clones were sub-cloned into destination vector pm42GW7 (Karimi et al., 2002), driving expression of the firefly luciferase gene. ORF clones were obtained from the Arabidopsis Biological Resource Center (Yamada et al., 2003; accession numbers G85112, GC103511 and DQ056658 for MYB20, MYB69 and SND2, respectively) or cloned from cDNA (MYB103, NST1, SND1 and SND3), cloned into pDONR207 (Invitrogen) and sequenced, before cloning into destination vector p2GW7 (Karimi et al., 2002). Primers for cloning ORFs are listed in Table S3. Accession numbers for phenylpropanoid biosynthesis genes and transcription factors are given in Data S1.


We thank Kjell Olofsson for technical assistance. This work was supported by grants from FORMAS (FuncFiber/BioImprove), the Swedish Research Council, VINNOVA and Bio4Energy (the Swedish programme for renewable energy) (B.S.). We also thank the Universiteit Gent Multidisciplinary Research Partnership ‘Biotechnology for a Sustainable Economy’ and the Hercules program of Ghent University for the Synapt Q-TOF (grant number AUGE/014). B.D. thanks the Agency for Science and Technology for a pre-doctoral fellowship.