Ectopic expression of MYB46 identifies transcriptional regulatory genes involved in secondary wall biosynthesis in Arabidopsis

Authors

  • Jae-Heung Ko,

    1. Department of Forestry, Michigan State University, East Lansing, MI 48824-1222, USA
    2. Great Lakes Bioenergy Research Center, Michigan State University, East Lansing, MI 48824-1222, USA
    Search for more papers by this author
  • Won-Chan Kim,

    1. Department of Forestry, Michigan State University, East Lansing, MI 48824-1222, USA
    Search for more papers by this author
  • Kyung-Hwan Han

    Corresponding author
    1. Department of Forestry, Michigan State University, East Lansing, MI 48824-1222, USA
    2. Great Lakes Bioenergy Research Center, Michigan State University, East Lansing, MI 48824-1222, USA
    Search for more papers by this author

*For correspondence (fax +1 517 432 1143; e-mail hanky@msu.edu).

Summary

MYB46 functions as a transcriptional switch that turns on the genes necessary for secondary wall biosynthesis. Elucidating the transcriptional regulatory network immediately downstream of MYB46 is crucial to our understanding of the molecular and biochemical processes involved in the biosynthesis and deposition of secondary walls in plants. To gain insights into MYB46-mediated transcriptional regulation, we first established an inducible secondary wall thickening system in Arabidopsis by expressing MYB46 under the control of dexamethasone-inducible promoter. Then, we used an ATH1 GeneChip microarray and Illumina digital gene expression system to obtain a series of transcriptome profiles with regard to the induction of secondary wall development. These analyses allowed us to identify a group of transcription factors whose expression coincided with or preceded the induction of secondary wall biosynthetic genes. A transient transcriptional activation assay was used to confirm the hierarchical relationships among the transcription factors in the network. The in vivo assay showed that MYB46 transcriptionally activates downstream target transcription factors, three of which (AtC3H14, MYB52 and MYB63) were shown to be able to activate secondary wall biosynthesis genes. AtC3H14 activated the transcription of all of the secondary wall biosynthesis genes tested, suggesting that AtC3H14 may be another master regulator of secondary wall biosynthesis. The transcription factors identified here may include direct activators of secondary wall biosynthesis genes. The present study discovered novel hierarchical relationships among the transcription factors involved in the transcriptional regulation of secondary wall biosynthesis, and generated several testable hypotheses.

Introduction

Vascular plants have evolved to possess xylem fibers and vessels that provide mechanical support for their growing body and serve as a conduit for long-distance transport of water and solutes. A defining feature of these cells is the secondary cell wall, which allows them to resist the forces of gravity and/or tension associated with the transpirational pull on a column of water. Secondary wall is formed in a highly coordinated manner by successive encrustation and deposition of cellulose fibrils, hemicelluloses (i.e. xylan) and lignin as soon as cell growth has stopped (Lerouxel et al., 2006; Somerville, 2006; Zhong and Ye, 2007). Variability is found in the proportions of the three major components depending on the species, growing site, climate, age and part of the plant. However, little is known about the coordinated genetic regulation leading to secondary wall formation and the regulation of the variations in proportional composition of the major components.

Cellulose, the most abundant biopolymer on Earth, is ubiquitous among plants, and constitutes the major polysaccharide of cell walls (Tanaka et al., 2003; Gardiner et al., 2003; Taylor et al., 2003; Saxena and Brown, 2005; Bhandari et al., 2006; Suzuki et al., 2006). Hemicelluloses are complex polysaccharides that associate with cellulose microfibrils, providing a cross-linked matrix. Recent studies identified several genes involved in the glucomannan and xylan biosynthesis (Dhugga et al., 2004; Liepman et al., 2005; Zhong et al., 2005; Brown et al., 2009; Lee et al., 2009; Wu et al., 2009). Xylans are major components of the secondary wall in angiosperms (York and O’Neill, 2008). Lignin, a large cross-linked phenolic compound, fills the spaces between cellulose and hemicellulose in the secondary wall, and confers mechanical strength to the cell wall. Many of the genes involved in the biosynthetic pathway of lignin have been characterized (Boerjan et al., 2003; Vanholme et al., 2008).

Biosynthesis of the secondary wall requires coordinated transcriptional activation of genes responsible for secondary wall components, such as cellulose, xylan and lignin. However, only limited information is available about the molecular mechanisms of coordinated biosynthetic regulation of the secondary wall components. Understanding the transcriptional regulatory network controlling the biosynthetic pathways of individual secondary wall components will increase our knowledge of fundamental plant biology, as well as providing biotechnological means for feedstock improvement.

In Arabidopsis, a group of closely related NAC transcription factors, including ANAC043/NST1 (NAC SECONDARY WALL THICKENING PROMOTING FACTOR1), ANAC066/NST2, ANAC012/NST3/SND1 (SECONDARY WALL-ASSOCIATED NAC DOMAIN PROTEIN1), VASCULAR-RELATED NAC-DOMAIN6 (VND6) and VND7, have been suggested as key transcriptional regulators of secondary wall biosynthesis in various sclerenchyma cell types (Kubo et al., 2005; Mitsuda et al., 2005, 2007; Zhong et al., 2006; Ko et al., 2006, 2007; Yamaguchi et al., 2008). Over-expression of the NAC genes resulted in ectopic deposition of secondary walls in cells that are normally parenchymatous, while suppression of their function reduced the secondary wall thickness (Kubo et al., 2005; Mitsuda et al., 2005, 2007; Zhong et al., 2006; Yamaguchi et al., 2008). In addition, several MYB transcription factors have also been identified as important regulators of secondary wall biosynthesis in plants (Patzlaff et al., 2003a,b; Steiner-Lange et al., 2003; Goicoechea et al., 2005; Yang et al., 2007; Zhong et al., 2008). Among them, MYB46 (At5g12870) was reported to be a direct target of ANAC012/SND1/NST3, and was shown to function as a master switch that turns on the genes responsible for the biosynthesis and deposition of cellulose, xylan and lignin (Zhong et al., 2007).

In the present study, we confirmed that MYB46 does indeed act as a master switch for secondary wall formation in Arabidopsis. We then developed an inducible secondary wall thickening system in Arabidopsis thaliana by over-expressing MYB46 under the control of a dexamethasone-inducible promoter (Aoyama and Chua, 1997). Using this inducible system, we performed time-course whole-transcriptome profiling experiments using GeneChip and Illumina/Solexa digital gene expression analyses to obtain an insight into the transcriptional regulatory network for secondary wall formation. From the analyses, we derived a tentative hierarchical network of transcriptional regulators whose expression coincided with or preceded the induction of secondary wall formation. An in vivo transcriptional activation assay was used to confirm the relationship among the identified transcription factors in the model, and some of them were found to have the ability to activate secondary wall biosynthetic genes. The present study identified a novel C3H-type zinc finger protein AtC3H14 as a potential master regulator of secondary wall biosynthesis downstream of MYB46.

Results

MYB46 functions as a transcriptional switch for secondary wall formation

We generated transgenic Arabidopsis plants over-expressing full-length cDNA of MYB46 (At5g12870). Seventeen of the 52 T1 transgenic plants produced had stunted growth with small curled-up rosette leaves (Figure 1a). Consistent with previous report by Zhong et al. (2007), these plants showed ectopic secondary wall thickening in the parenchymatous cells of leaves, floral organs and inflorescence stems (Figure 1b–d). Even in the transgenic plants that did not show the stunted growth phenotype, ectopic secondary wall thickening was observed in epidermis, cortex and pith cells of stem cross-sections (Figure 1e–h). The severity of the ectopic secondary wall thickening phenotypes (e.g. dwarfing, leaf curling) was positively associated with the level of expression of the introduced MYB46 gene (data not shown), further confirming that MYB46 functions as a master switch for induction of secondary wall biosynthesis.

Figure 1.

 Over-expression of MYB46 results in ectopic deposition of secondary walls in Arabidopsis plants.
A full-length MYB46 cDNA driven by CaMV 35S promoter was expressed in wild-type Arabidopsis plants.
(a) Three-week-old T1 plants showing dwarf stature with leaf curling (red arrow) and a wild-type phenotype (blue arrow).
(b–d) Lignified secondary wall-thickened cells from the leaf surface (b, c) and pedicel (d), stained with phloroglucinol.
(e–h) Ectopically secondary wall-thickened cells in a stem cross-section of wild-type-like transgenic plants. Arrows indicate secondary wall-thickened epidermal cells (e), cortex cells (f) and pith cells (g, h), stained with phloroglucinol (e–g) or toluidine blue (h).
Scale bars = 100 μm.

Establishment of an inducible secondary wall development system in Arabidopsis

In order to study the mechanism by which MYB46 triggers secondary wall biosynthesis, we established an inducible ectopic secondary wall biosynthesis system in transgenic Arabidopsis plants by over-expressing the MYB46 gene under the control of a dexamethasone-inducible promoter (Figure 2a) (Aoyama and Chua, 1997). The induction of ectopic secondary wall thickening was tested in the rosette leaves of 2-week-old transgenic (T3 homozygous) seedlings, at which growth stage no apparent secondary wall thickening is observed in wild-type plants (Figure 2). Within 24 h of dexamethasone (DEX) treatment, we observed upward curling of the leaf blades (Figure 2b), consistent with the phenotypic consequences of constitutive over-expression of MYB46 (Figure 1).

Figure 2.

 MYB46--nducible secondary wall thickening system.
(a) Diagram of the vector used for inducible expression of MYB46 in pTA7002.
(b) Leaf-curling phenotypes were observed after 24 h of DEX (dexamethasone) treatment (arrows).
(c) Dramatic induction of MYB46 and several secondary wall biosynthesis genes in a time-dependent manner after DEX treatment. Semi-quantitative RT-PCR was performed using the ACTIN8 gene as an internal control. Expression of selected genes involved in cellulose, xylan and lignin biosynthesis is shown.

We then performed a time-course semi-quantitative RT-PCR analysis to examine the changes in expression of genes that are known to be involved in secondary wall biosynthesis. Expression of MYB46 was induced as early as 30 min after the DEX treatment, but no expression of MYB46 was detected in the leaves of control plants (Figure 2c). As expected, the genes involved in the biosynthesis of major secondary wall components such as cellulose, xylan and lignin were dramatically up-regulated following MYB46 induction (Figure 2c). In particular, transcripts of CesA4, FRA8/IRX7, IRX8 and IRX9 started to accumulate after 3 h of MYB46 induction.

ATH1 GeneChip analysis identifies genes that are up-regulated during the early stage of secondary wall biosynthesis

In an effort to identify genes that are differentially expressed during the induction stage of secondary wall thickening, we performed whole-transcriptome ATH1 GeneChip analysis. The ATH1 GeneChip has a total of 22 745 probe sets that can interrogate 22 591 gene models of Arabidopsis. Based on the preliminary RT-PCR analysis (Figure 2c), we narrowed down the time course experiment to 0, 1, 3 and 6 h after induction. Only young rosette leaves (5–7th leaves) without petiole were used for RNA extraction to avoid any compounding effect of an endogenous developmental program of secondary wall thickening.

Within 6 h of the induction treatment, a total of 282 genes were threefold or higher up-regulated (Table S1). Their expression patterns were similar to the RT-PCR data in Figure 2(c). Many of the differentially expressed genes were involved in the biosynthesis of cellulose, xylan and lignin (Figure 3 and Table S2).

Figure 3.

 ATH1 GeneChip expression profiles of the secondary wall biosynthesis genes up-regulated by MYB46 induction.
The genes that were previously known to be involved in the biosynthesis of secondary wall components (e.g. cellulose, xylan and lignin) and in ‘other cell wall biosynthesis’ were used for hierarchical cluster analysis with Euclidean similarity and centroid linkage using GeneSpring software (Agilent). Genes classified as ‘other cell wall biosynthesis’ were adopted from the Cell Wall Navigator database (http://bioweb.ucr.edu/Cellwall/). Several genes co-expressed with MYB46 (ATTED-II; http://atted.jp/) but of ‘unknown function’ are also included. The color range (log2 scale) indicates the expression level of each gene. Locus identifier and gene names are shown on the right.

Illumina digital gene expression analysis confirms the ATH1 GeneChip data

According to a recent estimate (TAIR8, released on 28 April 2008), the Arabidopsis genome has approximately 33 282 genes, including non-coding RNAs, and thus the ATH1 GeneChip does not cover a third of the estimated genes in the genome. In order to cover the missing genes in our transcriptome analysis and to validate the ATH1 GeneChip data, we carried out digital gene expression (DGE) analysis using the Illumina ultrahigh-throughput deep sequencing system (http://www.illumina.com). In addition to providing a means for rare transcript discovery and quantification of gene expression, this digital gene expression system offers a cost-effective and robust method for validating microarray hybridization data (Barakat et al., 2007; ‘t Hoen et al., 2008; Hafner et al., 2008; Hanriot et al., 2008; Marioni et al., 2008; Wang et al., 2008b).

All of the seven samples, which are biological replicates of the ATH1 GeneChip analysis, generated more than four million filter-passed reads (Table S3). After alignment to the ‘tag-ome’ of Arabidopsis (see Experimental procedures), we obtained expression reads, expressed as transcripts per million (TPM), from a total of 19 389 loci. The ATH1 GeneChip carries 16 418 of the 19 389 loci, allowing direct comparison of the expression profiles between the two transcriptome analysis methods. A statistic analysis (F test, two-sample for variances) showed that the results are not significantly different from each other (α = 0.05) (Figure 4a). Table 1 shows the expression profiles of MYB46-induced secondary wall biosynthesis genes obtained by Illumina DGE analyses, which have very similar expression patterns to the ATH1 GeneChip data (Figure 3 and Table S2). The genes identified as ‘up-regulated’ in the ATH1 GeneChip analysis are also invariably marked as up-regulated in the Illumina DGE, cross-validating the two methods (Table 1 and Figure 4b,c).

Figure 4.

 Reproducibility of the two gene expression analysis methods used in the study: GeneChip (ATH1) data and Illumina DGE.
(a) Similarity test between the ATH1 and Illumina data sets. The observed data set was selected based on their ‘presence’ expression calling at the indicated time points of ATH1 and their matching Illumina data. The similarity of ATH1 and Illumina data sets was tested using the ‘F test, two-sample for variances’. The F ratios are smaller than the F critical values in all data set comparisons, indicating that the similarities between the data sets are statistically significant.
(b, c) Expression patterns of selected genes from ATH1 and Illumina DGE data.

Table 1.   Illumina DGE profiles of the secondary wall biosynthesis genes up-regulated by MYB46 induction
Affy IDaAGIbIllumina DGE (TPM)cDGE (FC)eATH1 (FC)fGene description
−DEXd+DEX(+DEX/−DEX)(+DEX/−DEX)
0 h1 h3 h6 h1 h3 h6 h1 h3 h6 h1 h3 h6 h
  1. aAffymetrix probe identification number.

  2. bArabidopsis Gene Index number.

  3. cTranscripts per million.

  4. dWith (+) or without (−) dexamethasone treatment.

  5. eFold change based on Illumina DGE data.

  6. fFold change based on ATH1 GeneChip data (Table S2).

  7. gGenes involved in the cell wall metabolism from the Cell Wall Navigator database (http://bioweb.ucr.edu/Cellwall/).

  8. hGenes co-expressed with MYB46 (ATTED-II; http://atted.jp/) but of unknown function. ‘UN’ indicates that the value is undefined because of the denominator is zero. DUF, domain of unknown function. Threefold or up-regulated expression is indicated by shading.

Cellulose biosynthesis
249070_atAt5g4403011.611.810.413.111.2179.4219.50.917.316.81.17.09.2AtCesA04, IRX5
246425_atAt5g174203.75.55.48.52.08.669.10.41.68.10.91.06.8AtCesA07, IRX3
254618_atAt4g187800.50.00.10.00.80.30.5UN3.050.01.03.312.0AtCesA08, IRX1
Xylan biosynthesis
248121_atAt5g546900.20.30.20.40.23.420.00.717.050.00.63.622.8IRX8, glycosyltransferase (GT) family 8
265463_atAt2g370900.90.60.70.20.813.719.81.319.699.00.814.317.0IRX9, GT43
266156_atAt2g281101.94.24.44.02.0106.183.80.524.121.01.317.88.8FRA8, GT47
264493_atAt1g274400.90.52.01.81.03.78.22.01.94.60.82.92.4GT47, AtGUT2, IRX10
247496_atAt5g6184086.1127.7105.4117.1126.41098.2382.61.010.43.31.19.55.5GT47, AtGUT1, IRX10-L
Lignin biosynthesis
259149_atAt3g103401.30.50.52.20.728.322.61.456.610.31.312.315.0PAL4
267470_atAt2g30490110.394.978.0121.776.0408.2130.80.85.21.10.85.31.2C4H
258047_atAt3g212403.61.16.27.02.224.017.42.03.92.51.14.71.74CL2
258037_atAt3g212300.70.83.02.22.07.25.02.52.42.31.22.71.34CL4
245101_atAt2g4089013.110.811.514.510.534.033.01.03.02.30.83.71.5C3H1
259878_atAt1g7679010.714.934.318.445.8125.442.93.13.72.33.23.42.2COMT-like8
253985_atAt4g2622076.73.77.47.213.56.61.13.60.92.1158.6230.7CCoAOMT
258023_atAt3g1945075.730.229.973.928.8428.4518.41.014.37.01.26.64.0CAD2
267094_atAt2g380803.21.92.01.12.211.145.31.25.641.21.45.115.6IRX12, LACCASE 4
251131_atAt5g011901.10.20.20.70.6111.783.93.0558.5119.90.5114.2237.2LACCASE 10
250958_atAt5g0326024.728.115.333.730.372.8118.61.14.83.50.97.44.4LACCASE 11
Other cell wall biosynthesisg
256038_atAt1g191701.43.22.81.56.245.728.11.916.318.71.211.24.9Glycoside hydrolase (GH) family 28
261266_atAt1g267708.97.711.725.25.2151.7404.50.713.016.11.48.83.6At-EXP10, expansin family protein
260941_atAt1g4497000.00.00.00.00.03.0UNUNUN0.62.35.4Peroxidase
262444_atAt1g474801.60.54.02.35.511.319.911.02.88.72.25.77.7Similar to CXE carboxylesterase
264433_atAt1g6181000.00.40.00.05.819.0UN14.5UN5.541.754.2GH family 1
259801_atAt1g722300.90.91.01.70.211.746.00.211.727.10.719.7159.5Plastocyanin-like protein
262351_atAt1g7299021.334.128.126.433.4824.4396.61.029.315.01.025.314.2GH family 35/BGAL17
263628_atAt2g0478027.447.437.335.633.8289.5363.50.77.810.20.75.46.5AtFLA7
266086_atAt2g3806000.00.00.00.08.44.1UNUNUN1.011.97.1Carbohydrate transmembrane transporter
259174_atAt3g036900.90.40.70.30.04.53.30.06.411.00.94.73.9GT family 14
256252_atAt3g1134000.414.411.30.037.155.70.02.64.93.63.29.2UDP-glucosyl transferase family
257896_atAt3g169203.41.12.43.03.312.7150.13.05.350.00.42.839.1GH family 19
257876_atAt3g171300.50.40.30.30.026.930.70.089.7102.31.039.711.7Pectin methylesterase inhibitor family
258143_atAt3g181705.112.97.08.113.132.660.71.04.77.51.17.596.2Similar to GT
252781_atAt3g4295019.528.820.318.728.0263.8101.31.013.05.40.911.45.4GH family 28
252563_atAt3g459708.332.628.08.6109.3190.870.83.46.88.22.05.48.5At-EXPL1, expansin family protein
252490_atAt3g4672000.00.00.00.01.15.7UNUNUN0.83.64.2GT family 1
253998_atAt4g2601000.00.30.20.00.00.0UN0.00.00.8480.746.2Peroxidase
253380_atAt4g333306.79.27.78.810.914.610.71.21.91.25.36.519.8GT family 8
253379_atAt4g333306.79.27.78.810.914.610.71.21.91.20.27.621.3GT family 8
250499_atAt5g097305.25.14.02.84.519.732.00.94.911.42.05.96.5GH family 3, ATBXL3
250437_atAt5g1043002.70.40.50.80.110.20.30.320.40.51.26.3AtAGP4
250157_atAt5g151805.41.91.52.31.21.58.60.61.03.71.11.09.4Prx10, peroxidase
246555_atAt5g154707.512.99.34.610.836.933.40.84.07.31.13.83.5GT14/GAUT14
255860_atAt5g349402.12.93.61.00.926.827.00.37.427.01.017.76.7GH family 79
Unknown functionh
260886_atAt1g292000.20.00.00.20.20.37.3UNUN36.51.00.630.3DUF246
261999_atAt1g338001.61.92.91.51.364.656.00.722.337.30.929.414.0DUF579
266244_atAt2g277400.20.00.60.20.33.739.9UN6.2199.51.77.845.4DUF662
246344_atAt3g567301.72.30.42.02.14.824.00.912.012.01.81.58.3DUF537
255005_atAt4g099902.82.62.12.01.121.935.60.410.417.80.36.817.6DUF579
247590_atAt5g607200.41.40.71.01.210.127.00.914.427.01.53.47.0DUF547
247030_atAt5g672100.90.40.20.81.429.933.83.5149.542.31.317.312.9DUF579

Several unknown genes in the list are of particular research interest in understanding the biological function of MYB46 as they are not only highly up-regulated by MYB46 but also co-expressed with MYB46 (ATTED-II co-expression analysis; Obayashi et al., 2009) (Figure 3 and Table 1).

Identification of transcription factors involved in secondary wall thickening

The whole-transcriptome analysis performed here identified a total of 42 transcription factors that are up-regulated within 6 h of the induction treatment (Table 2), 32 of which were up-regulated as early as 3 h after MYB46 induction. All of them were significantly up-regulated by DEX treatment, and were expressed at the basal level without DEX treatment during the entire 6 h time span after the induction treatment (Figure S1).

Table 2.   Transcription factors up-regulated within 6 h of MYB46 induction identified by whole-transcriptome analyses
Affy IDaAGIbATH1Illumina DGETissuedMR to MYB46eSecondary wall-relatedfTestedgGene description
(+DEX/−DEX)c(+DEX/−DEX)
1 h3 h6 h1 h3 h6 h
  1. N/A, data not available.

  2. aAffymetrix probe identification number.

  3. bArabidopsis Gene Index number.

  4. cWith (+) or without (−) dexamethasone treatment.

  5. dtissue specific expression obtained from theArabidopsis eFP browser (http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi).

  6. eMR, mutual rank of genes co-expressed with MYB46 (ATTED-II; http://atted.jp/).

  7. fgenes reported to be involved in secondary wall biosynthesis (Zhong et al., 2008; Zhou et al., 2009).

  8. gGenes tested in this study. ‘UN’ indicates that the value is undefined because of the denominator is zero.

261564_atAt1g017208.44.74.16.96.42.9Seed   ANAC002/ATAF1
264148_atAt1g022201.06.91.73.719.01.5Petal   ANAC003
260776_atAt1g145801.27.25.32.17.48.3Seed   Zinc finger (C2H2 type) protein
262715_atAt1g164901.71.64.40.85.91.4Stem246.2Yes AtMYB58
255903_atAt1g179500.81.57.6UN8.0UNStem149YesYesAtMYB52
261717_atAt1g184000.62.23.61.30.93.4Leaf   AtbHLH044/BEE1
261139_atAt1g197000.94.12.00.75.82.2Senescent leaf   BEL1-LIKE HOMEODOMAIN 10
262096_atAt1g560102.56.01.41.17.47.7Senescent leaf   NAC1
245628_atAt1g566500.92.84.51.54.37.2Senescent leaf   MYB75/PAP1
261106_atAt1g629900.88.316.81.035.4115.3Stem312.9YesYesKNAT7 (KNOTTED-LIKE HOMEOBOX 7)
256367_atAt1g668100.74.82.60.017.0UNStem8.3 YesAtC3H14, C3H-type zinc finger protein
259854_atAt1g722001.313.620.9UN6.193.3Seed/stem   RING-H2 zinc finger protein
259828_atAt1g722200.814.145.3UN43.0UNSeed/stem155.6  RING-H2 zinc finger protein
245726_atAt1g733601.74.61.40.7UNUNShoot apex   HOMEODOMAIN GLABROUS11 (HDG11)
245735_atAt1g7341010.412.821.0UNUNUNStem/seed YesYesAtMYB54
264119_atAt1g791800.521.365.60.0127.0UNStem147.1YesYesAtMYB63
265359_atAt2g167206.027.810.829.029.736.8Stem/seed  YesAtMYB7
265621_atAt2g273007.24.54.06.037.0UNImbibed seed   ANAC040
267141_atAt2g380901.34.87.40.35.25.6Stem  YesMYB-R, MYB-related transcription factor
263380_atAt2g402001.12.06.7N/AN/AN/APedicel129.8  bHLH family protein
267077_atAt2g409703.47.14.23.410.215.4Root   GARP-G2-like family
263775_atAt2g464107.04.61.2N/AN/AN/AShoot apex   CAPRICE (CPC), MYB transcription factor
258758_atAt3g108101.13.92.0UN28.04.2Seed   RING zinc finger protein
258366_atAt3g142301.43.81.41.74.42.9Imbibed seed   ERF/AP2 transcription factor
257267_atAt3g150301.23.52.01.14.72.6Petal/leaf   TCP transcription factor
251763_atAt3g557300.94.72.30.36.53.4Stamen   AtMYB109
255125_atAt4g082501.02.14.53.70.74.0Stamen   Scarecrow transcription factor family protein
253710_atAt4g292300.61.932.00.70.20.8Stem/seed  YesNAC/ANAC075
253245_atAt4g345900.73.22.2N/AN/AN/APetal   G-BOX BINDING FACTOR 6 GBF6/bZIP11
253247_atAt4g346100.513.83.03.80.30.5Stem   BELL1-LIKE HOMEODOMAIN 6
253219_atAt4g349901.022.24.6UN0.8UNStem/stamen  YesAtMYB32
252958_atAt4g386201.511.03.5N/AN/AN/AStem/petal  YesAtMYB4
250322_atAt5g1287077.5217.472.8UN866.3582.5Stem YesYesAtMYB46
246520_atAt5g157900.95.91.81.07.22.0Shoot apex/stem   RING zinc finger protein
246913_atAt5g258301.03.913.0UN1.84.0Stem40  Zinc finger protein (GATA13)
249189_atAt5g427800.75.52.3UN28.58.3Seed   ATHB27 (HOMEOBOX PROTEIN 27)
249115_atAt5g438101.54.42.51.74.92.2Shoot apex   AGO10 PNH/ZLL
249087_atAt5g442101.925.417.14.028.018.0Seed   ATERF9
248460_atAt5g509151.93.74.13.02.812.2Seed   bHLH family protein
247767_atAt5g588900.91.53.5N/AN/AN/ASeed   AGL82
N/AAt3g09230N/AN/AN/A1.77.31.6N/A   AtMYB1
N/AAt3g51080N/AN/AN/A0.45.02.7N/A   Zinc finger protein (GATA9)

Based on this observation, we hypothesize that the transcriptional factors showing an early response to DEX treatment may act as transcriptional regulators of downstream secondary wall biosynthesis genes. Previous studies (Zhong et al., 2008; Zhou et al., 2009) also found that five of the transcription factors we identified in this study (KNAT7, MYB52, MYB54, MYB58 and MYB63) were required for secondary wall biosynthesis in Arabidopsis (Table 2). In order to test our hypothesis by using in vivo transcriptional activation analysis (TAA; Yoo et al., 2007), we selected 10 transcription factor genes (MYB4, MYB7, MYB32, MYB52, MYB54, MYB63, MYB-R, KNAT7, ANAC075 and AtC3H14) from the list based on their abundant expression secondary wall-forming tissues and co-expression patterns with MYB46 [Arabidopsis eFP browser (Winter et al., 2007) and ATTED-II (Obayashi et al., 2009)].

MYB46 activates the selected transcription factors in vivo

We used transient transcriptional activation analysis (TAA) to investigate the MYB46-triggerred regulatory network for secondary wall biosynthesis. In the protoplast-based transient gene expression system, the transcription and translation machineries are transiently manipulated by delivering macromolecules, such as DNA, to investigate cell-autonomous regulation and responses (Walker et al., 1987; Skriver et al., 1991; Ballas et al., 1993; Sheen, 2001; Yoo et al., 2007). Arabidopsis mesophyll protoplasts have been successfully used in transient analysis of promoter activity in vivo (Abel and Theologis, 1994; Sheen, 2001; Tiwari et al., 2006; Yoo et al., 2007; Zhong et al., 2007, 2008; Zhou et al., 2009).

As shown in Figure 5(a), MYB46 was co-expressed in Arabidopsis leaf protoplasts with a GUS reporter gene driven by the promoter of a selected transcription factor to assess the effect of MYB46 on the promoter after 16 h of incubation. First, the transfection and expression of MYB46 as an effector in this TAA system was confirmed by semi-quantitative RT-PCR analysis (Figure 5b). Expression of MYB46 and ANAC012 was undetectable in the protoplasts transfected with reporter construct only (control), but a high level of MYB46 and ANAC012 expression was observed in the protoplasts transfected with the effector constructs 35S::MYB46 or 35S::ANAC012. Moreover, we detected the induction of genes located downstream of ANAC012 and MYB46 (Figure 5b). For example, MYB46, AtC3H14 and KNAT7 genes were up-regulated in the protoplast co-transfected with ANAC012 as effector. In contrast, MYB7 was up-regulated only with MYB46 as effector. No expression of ANAC012 was detected in the protoplasts transfected with MYB46 as effector, confirming that the ANAC012/SND1/NST3 is located upstream of MYB46 (Zhong et al., 2007).

Figure 5.

 MYB46 activates the expression of selected transcription factors and genes involved in the biosynthesis of secondary wall components.
MYB46 was co-expressed in Arabidopsis leaf protoplasts with the GUS reporter gene driven by the promoters of selected transcription factors and secondary wall biosynthetic genes. Activation of the promoter by MYB46 was measured by assaying the GUS activity after 16 h of incubation.
(a) Diagram of the effector and reporter constructs used in this transcriptional activation assay. The effector construct contains the MYB46 gene driven by the CaMV 35S promoter. The reporter constructs consist of the GUS reporter gene driven by the promoters of selected transcription factors and secondary wall biosynthetic genes.
(b) Confirmation of the protoplast transfection and over-expression of ANAC012 and MYB46 as effector constructs. Control indicates transfection with the reporter construct; 35S::ANAC012 and 35S::MYB46 indicate the effector constructs. Semi-quantitative RT-PCR analysis shows that the effector genes were highly expressed in protoplasts transfected with the effector constructs. ACTIN8 was used as a loading control.
(c) Transcriptional activation analysis showing the effects of 35S::MYB46 on activation of the promoters of transcription factor genes.
(d) Transcriptional activation analysis showing the effects of MYB46 on induction of the promoters of secondary wall biosynthetic genes. The expression level of the GUS reporter gene in the protoplasts transfected with no effector was used as a control and was set to 1. Error bars indicate SEs of three biological replicates.

MYB46 strongly activated expression of all of the transcription factors tested, with exception of ANAC075 (At4g29230) and MYB46 itself (Figure 5c). It seems that MYB46 does not show any negative auto-regulation and ANAC075 is probably not directly regulated by MYB46 (Figure 5c). This analysis successfully confirmed that the transcription factors identified by the transcriptome analyses are indeed direct or indirect downstream targets of MYB46.

We then performed additional TAA experiments using the selected transcription factors as effectors and their promoters as reporters to further investigate the relationships among the transcription regulators (Figure S2a). The results revealed a complex network of cross-talk among the transcription factors. For example, MYB63 positively regulates the transcription of MYB7, MYB4, MYB54 and AtC3H14, while MYB7, MYB32 and MYB4 negatively regulate their own transcription and that of MYB52 (Figure S2b). Previous reports (Jin et al., 2000; Zhao et al., 2007) demonstrated that MYB4 binds to its own promoter and represses its own transcription as well as that of cinnamate-4-hydroxylase. MYB32, which is a close homolog of MYB4, also negatively affected the expression of genes involved in the phenylpropanoid pathway (Preston et al., 2004). Interestingly, AtC3H14 strongly activates the transcription of MYB63 and MYB52.

Transcriptional activation of secondary wall biosynthesis genes in vivo

Whole-transcriptome analyses indicated that MYB46 induced a battery of genes involved in secondary wall biosynthesis within 6 h of induction treatment (Figure 3 and Table 1). To confirm this result, the promoter regions of eight secondary wall biosynthesis genes (CesA4, CesA7 and CesA8 for cellulose, IRX8 and IRX9 for xylan, and PAL4, CCoAOMT and LAC10 for lignin) were used to drive GUS reporter gene expression in TAA experiments using MYB46 as an effector (Figure 5a). MYB46 activated all of the secondary wall biosynthetic genes tested (Figure 5d), further confirming the validity of our experimental system and approach.

To test our hypothesis that the selected transcription factors are responsible for activation of individual downstream genes involved in the biosynthesis of secondary walls, we studied the transcriptional regulation of secondary wall biosynthetic genes using the selected transcription factors as effectors (Figure 6). Both MYB63 and MYB52 strongly activated expression of PAL4 gene, which is involved in the first step of the monolignol biosynthetic pathway (Figure 6b). The results also revealed that a C3H-type zinc finger protein, AtC3H14, activated all of the secondary wall biosynthetic genes tested (Figure 6b), suggesting that AtC3H14 may function as a master regulator of secondary wall biosynthesis, located downstream of MYB46.

Figure 6.

 Transcriptional activation of genes involved in the biosynthesis of secondary wall components by selected transcription factors.
Transcription factors, including MYB46, were co-expressed in Arabidopsis leaf protoplasts with the GUS reporter gene driven by the promoters of secondary wall biosynthetic genes. Activation of the promoters by MYB46 was measured by assaying GUS activity after 16 h incubation.
(a) Diagram of the effector and reporter constructs used in the transcriptional activation assay. The effector constructs contains the transcription factors identified in this study, including MYB46, driven by the CaMV 35S promoter. The reporter constructs consist of the GUS reporter gene driven by the promoters of the secondary wall biosynthetic genes CesA4, CesA7, CesA8, IRX8, IRX9, PAL4, CCoAOMT and LAC10.
(b) Transcriptional activation analysis showing the effects of the transcription factor on induction of the promoters of secondary wall biosynthetic genes. The result for 35S::MYB46 (Figure 7b) is included as a positive control. The expression level of the GUS reporter gene in the protoplasts transfected with no effector was used as a control and was set to 1. Error bars indicate SEs of three biological replicates.

Coordinated regulation of downstream target genes by ANAC012/SND1/NST3, MYB46 and AtC3H14

ANAC012 is known to be an upstream direct regulator of MYB46. Both ANAC012 and MYB46 function as master regulators of the entire secondary wall biosynthesis pathway (Zhong et al., 2007, 2008). Both of them regulated common downstream transcription factors and activated all of the cell wall biosynthetic genes tested (Figure S3), but there were some unexpected differences in their roles as a master switch for secondary wall biosynthesis. MYB46 strongly activated the transcription of MYB7, MYB32 and MYB-R, while ANAC012 did not (Figure S3a). Furthermore, the activation levels of individual secondary wall biosynthetic genes were different between ANAC012 and MYB46. For example, ANAC012 activated the expression of IRX8 and IRX9 much more strongly than MYB46 did. The opposite was true in the case of PAL4 (Figure S3b). These results suggest that MYB46 may not be the only downstream target for ANAC012-mediated regulation of secondary wall biosynthesis.

To further investigate the functional relationship among ANAC012, MYB46 and AtC3H14, we performed TAA experiments to study the temporal regulation of downstream target genes (Figure 7). GUS activities were measured at 1, 3, 6, 12 and 18 h after transfection of the vector constructs into protoplasts. As expected, strong activation of MYB46 by ANAC012 was observed as early as 6 h after transfection (Figure 7a). AtC3H14 was activated by both ANAC012 and MYB46 (Figure 7a,b), which confirms that AtC3H14 is located downstream of these two master regulators. Within 6 h of transfection, transcription of MYB52, MYB54, MYB63 and KNAT7 was induced by MYB46, ANAC012 and AtC3H14 (Figure 7). This result suggests that MYB52, MYB54, MYB63 and KNAT7 may be common downstream targets of ANAC012, MYB46 and AtC3H14 in the ANAC012-directed transcriptional network. Interestingly, MYB7, MYB4 and MYB32 were activated only by MYB46 during the entire experiment, consistent with the previous results (Figure S3a). In particular, MYB7 was activated early and very strongly by MYB46.

Figure 7.

 Temporal regulation of selected transcription factors and secondary wall biosynthesis genes by ANAC012, MYB46 and AtC3H14.
ANAC012, MYB46 and AtC3H14 were co-expressed in Arabidopsis leaf protoplasts with the GUS reporter gene driven by the promoters of transcription factors (a–c) or secondary wall biosynthetic genes (e, f). Induction of GUS gene expression by ANAC012 (a, d), MYB46 (b, e) and AtC3H14 (c, f) was measured by assaying the GUS activity at the indicated time. The expression level of the GUS reporter gene in the protoplasts transfected with no effector was used as a control and was set to 1. Error bars indicate SEs of three biological replicates.

Although the activation levels were different, ANAC012, MYB46 and AtC3H14 activated transcription of all the cell wall biosynthetic genes tested as early as 6 h after transfection (Figure 7d–f). Interestingly, MYB46 started to activate the transcription of the lignin biosynthesis genes PAL4, CCoAOMT and LAC10 at 6 h, while ANAC012/SND1 activated the genes at 12 h after transfection. This difference implies that regulation of these lignin biosynthetic genes by ANAC012 might be accomplished through MYB46.

ANAC012 and MYB46 bind to the promoter of AtC3H14

To test whether ANAC012 and MYB46 bind to the AtC3H14 promoter, we used an electrophoretic mobility shift assay (EMSA) using recombinant glutathione S-transferase (GST) fusion proteins and AtC3H14 promoter fragments. As a positive control, we first confirmed the previously reported binding of recombinant ANAC012 protein to the MYB46 promoter (Figure 8a) (Zhong et al., 2007). Figure 8 shows that both ANAC012 and MYB46 proteins bind to the AtC3H14 promoter fragment. The specificity of the binding between the recombinant transcription factor proteins and the AtC3H14 promoter was confirmed by the observations that (i) addition of cold (unlabeled) AtC3H14 promoter fragments competed with the specific binding in a dose-dependent manner, and (ii) the mobility shift was not observed when the AtC3H14 promoter fragment was incubated with GST alone (Figure 8b,c). We used two fragments of the AtC3H14 promoter in the experiments, from −1 to −250 and −250 to −500 bp from the start codon. ANAC012 binds to both of the fragments (Figure 8b), while MYB46 only binds to the −250 to −500 bp fragment (Figure 8c). These results clearly indicate that both ANAC012 and MYB46 directly interact with the AtC3H14 promoter. However, it appears that binding of MYB46 to the AtC3H14 promoter is stronger than that of ANAC012, because the MYB46 binding to the labeled probes continues until addition of a 100-fold concentration of competitor, whereas ANAC012 binding was eliminated by a 30-fold concentration (data not shown).

Figure 8.

 ANAC012 and MYB46 bind to the promoter of AtC3H14.
The ANAC012–GST and MYB46–GST recombinant proteins were incubated with 32P-labeled DNA probes (fragments of the AtC3H14 promoter) and were subjected to EMSA by PAGE. GST was used as a control protein in the EMSA experiments.
(a) ANAC012 binds to the 260 bp MYB46 promoter fragment (−455 to −714 bp), resulting in retardation of its mobility. This experiment serves as a positive control.
(b) ANAC012 binds to the two 250 bp promoter fragments of AtC3H14 (−1 to −250 and −250 to −500bp).
(c) MYB46 binds to the 250 bp AtC3H14 promoter fragment −250 to −500 bp.
The promoter regions used for the DNA probes in each experiment are indicated below the gel images. Competition for the protein–DNA binding was performed using 30× cold (unlabeled) probes. The free unbound DNA probes are indicated by arrows.

Discussion

For a better understanding of the molecular mechanisms underlying secondary wall biosynthesis, it is critical to identify the transcriptional regulatory network controlling the biosynthetic process and characterize the functional roles of the transcription factors in the network. Recently, a MYB domain transcription factor, MYB46, was identified as a key switch activating the developmental program leading to biosynthesis of the secondary wall (Zhong et al., 2007, 2008). However, only limited information is available regarding the MYB46-mediated regulation of secondary wall biosynthesis. Mellerowicz et al. (2001) suggested that a few hundred genes are probably involved in the biosynthesis of secondary walls during wood formation. Therefore, it is plausible that a set of common as well as pathway-specific transcription factors regulates the biosynthetic pathways for the individual components of secondary wall.

Whole-transcriptome analyses to identify the transcriptional network controlling the secondary wall biosynthesis

Using the inducible ectopic secondary wall biosynthesis system that we developed, we attempted to identify the transcription factors whose expression coincided with or preceded the induction of secondary wall biosynthesis. Both GeneChip (ATH1, Affymetrix) and Illumina DGE analyses were used to obtain a series of transcriptome profiles during the one of induction of secondary wall thickening.

Unlike traditional EST analysis, the Illumina ultra-high-throughput sequencing system allowed much deeper coverage of the mRNA population and identification of additional regulators of secondary wall biosynthesis. The transcriptome profiles obtained by Illumina DGE were confirmed by ATH1 GeneChip and vice versa (Figure 4 and Table 1). The Illumina DGE produced expression profile data matching a total of 59 466 inter-genic loci. We found that at least 100 inter-genic loci were differentially expressed during MYB46-induced secondary wall thickening (Table S5). Furthermore, the Illumina DGE yielded expression profiles of an additional 2902 genes that are not available on the ATH1 GeneChip (Table S4). Among them, 126 genes are predicted to have transcription factor activity. The information obtained from Illumina DGE could be utilized to identify additional regulators such as non-annotated novel genes or regulatory small RNAs that otherwise cannot be identified by ATH1 GeneChip analysis.

The utility of our approach for identification of transcriptional regulators of secondary wall biosynthesis was successfully demonstrated by the fact that the genes significantly up-regulated during the induction of secondary wall thickening (Table S1) include most of the genes previously known to be involved in secondary wall biosynthesis (Figure 3 and Table 1). Combining the Illumina DGE and ATH1 GeneChip data, we identified a total of 42 transcription factors as potential regulators of secondary wall biosynthesis. The list includes several transcription factors previously known to be regulators of secondary wall biosynthesis (e.g., KNAT7, MYB52, MYB54, MYB58 and MYB63) as well as many unidentified transcription factors.

MYB46 transcriptionally activates the downstream transcription factors

We have demonstrated that all of the selected transcription factors identified in the transcriptome analyses, with the exception of ANAC075, are transcriptionally activated by MYB46 in the TAA using Arabidopsis mesophyll protoplasts (Figures 5, 7b and S3a). Zhong et al. (2008) suggested that MYB85 (At4g22680) is a direct target of MYB46. However, our transcriptome analyses showed no changes in MYB85 expression during MYB46 induction, suggesting that MYB85 may not be a part of the MYB46-directed secondary wall formation pathway.

Previous studies (Zhong et al., 2007, 2008) reported that both KNAT7 and MYB46 are direct targets of ANAC012, so it is interesting to note that MYB46 could transcriptionally activate KNAT7 despite the fact that the protoplasts have no detectable ANAC012 expression (Figures 5, 7b and S3a). Therefore, our data suggest that, in the hierarchical transcriptional regulatory network of secondary wall biosynthesis, KNAT7 may be positioned downstream of MYB46 and under the control of both ANAC012 and MYB46. Our electrophoresis gel mobility shift assay (EMSA) appears to corroborate the possibility that KNAT7 is directly regulated by MYB46 (W.-C.K., unpublished data).

According to our transcriptome analysis, MYB7 was rapidly up-regulated (within 1 h of MYB46 induction) (Table 2), and this early up-regulation was further confirmed in our TAA experiment (Figure 7b), implying that MYB7 may be a novel direct target of MYB46. However, the functional significance of MYB7 in secondary wall formation or promoter binding of MYB46 has yet to be established.

Identification of transcriptional activators of the secondary wall biosynthetic genes

We tested our hypothesis that the selected transcription factors are responsible for activation of the individual genes involved in biosynthesis of the secondary wall (Figure 6). Using MYB63 as the effector in the TAA analysis, we successfully demonstrated that MYB63 is an upstream regulator of PAL4. Very recently, Zhou et al. (2009) reported that MYB63 is a direct transcriptional activator of the lignin biosynthetic pathway during secondary wall formation in Arabidopsis. MYB52 also strongly activated expression of PAL4 (Figure 6).

In addition, we found that a novel C3H-type zinc finger protein, AtC3H14, could activate all of the secondary wall biosynthetic genes tested (Figure 6). This result suggests that AtC3H14 may act as another master regulator of secondary wall biosynthesis, under the direct control of ANAC012 and/or MYB46. However, the functional significance in planta and the detailed molecular function of AtC3H14 remain to be elucidated. Several studies have shown that C3H-type motif-containing zinc finger proteins have an RNA-binding function in RNA processing (Berg and Shi, 1996; Li et al., 2001; Delaney et al., 2006; Wang et al., 2008a,b).

Arabidopsis appears to have at least three layers of transcriptional control systems for secondary wall formation. Each layer of the control system uses a different transcription factor family, such as NAC, MYB or C3H. Given the fundamental importance of secondary walls in land plants, it is not surprising to find such multiple strategies for secondary wall biosynthesis.

Interestingly, seven of the ten transcription factors used as the effector in the TAA experiment described in Figure 6, including KNAT7, MYB4, MYB32 and MYB54, did not have any significant impact on expression of the secondary wall biosynthetic genes tested. Several possibilities can explain this finding: (i) they are not involved in the transcriptional control of these particular genes, (ii) they require participation of other transcription factor(s) to function, or (iii) they are not involved in secondary wall formation. As KNAT7 and MYB54 are known to be required for secondary wall biosynthesis (Zhong et al., 2008), the results suggest that KNAT7 and MYB54 may not be direct regulators of the particular secondary wall biosynthesis genes tested here. This might also be true in the case of MYB4 and MYB32, which are known to be regulatory genes involved in the phenypropanoid pathway (Jin et al., 2000; Preston et al., 2004). It is a characteristic of eukaryotic transcriptional regulation that a combination of multiple transcription factors binds to the promoters of target genes to activate or repress their expression (Ogata et al., 2003). Therefore, we cannot exclude the possibility that these transcription factors function in a protein complex with other partners. Further combinatorial activation analysis using these transcription factors is required to support this hypothesis.

Further analysis of additional transcriptional effectors identified in this study and extending the list of target reporter genes of secondary wall biosynthesis are likely to identify other novel transcriptional regulators of secondary wall biosynthesis.

Complex transcriptional regulatory network leading to secondary wall biosynthesis

The present study on the regulatory relationship of the selected transcription factors uncovered a tentative transcriptional regulatory network with complex cross-talk (Figures 5, 7 and S3a), providing an unprecedented opportunity to dissect the downstream regulation path of ANAC012 and MYB46. For example, MYB52, MYB54, MYB63 and KNAT7 are probably common downstream targets of ANAC012, MYB46 and AtC3H14 (Figure 7a–c). MYB46 activates the transcription of PAL4, CCoAOMT and LAC10 much earlier and more strongly than ANAC012 does (Figure 7d,e). These results suggest that ANAC012-mediated transcriptional regulation leading to secondary wall biosynthesis might involve multi-level regulation, including MYB46 and possibly AtC3H14. AtC3H14 may be one of the direct upstream regulators of MYB63 because transcription of MYB63 was activated more strongly and earlier by AtC3H14 than by MYB46 (Figures 7 and S3). As MYB63 is involved in the lignin biosynthesis pathway (Zhou et al., 2009), it is possible that MYB46-directed lignin biosynthesis might be achieved through AtC3H14 and MYB63. It is interesting to note that MYB7, MYB4 and MYB32 were activated only by MYB46, and might be negatively regulated by ANAC012 because induction of ANAC012 strongly up-regulates MYB46 expression but does not affect the expression of these genes (Figure 7). In addition, MYB46 strongly activated MYB52 expression (Figures 5 and S3a), but MYB7, MYB4 and MYB32 negatively regulate the transcription of MYB52 (Figure S2b). These results suggest that the ANAC012-mediated transcriptional regulation might involve some form of fine-tuning mechanism, providing extra flexibility of transcriptional control by activators and repressors on common targets (Jin et al., 2000).

In summary, we have described our strategy to identify transcription factors regulating secondary wall biosynthesis by using an inducible secondary wall thickening system and whole-genome transcriptome profiling. Our results show that the transcription factors identified here most likely include direct activators of many genes involved in the biosynthesis of secondary wall components in Arabidopsis. While additional functional characterization of these genes is necessary to confirm the tentative transcriptional network, the current study discovered novel relationships between transcription factors, and generated several hypotheses that may be tested in order to fill the gap in our knowledge regarding transcriptional regulation of secondary wall biosynthesis.

Experimental procedures

Plant materials and growth conditions

Arabidopsis thaliana, ecotype Columbia (Col-0), was used in both the wild-type and transgenic experiments. Plants were grown on soil in a growth chamber (16 h light/8 h dark) at 23°C. All experiments were performed in triplicate and repeated at least three times.

Histological analysis

To visualize secondary xylem, rosette leaves (after removing chlorophyll by 70% ethanol extraction) and hand-cut stem cross-sections were stained with 2% phloroglucinol/HCl or 0.05% toluidine blue O for 1 min.

Plasmid construction and plant transformation

All of the constructs used in this study were verified by DNA sequencing. The full-length cDNA of MYB46 (At5g12870) was amplified from Arabidopsis stem cDNA by PCR using primers with attB1 and attB2 tails (shown in lower-case letters): 5′-aaaaagcaggctATGAGGAAGCCAGAGGTAGCCAT-3′ (forward) and 5′-agaaagctgggtTCATATGCTTTGTTTGAAGTTGA-3′ (reverse). The PCR product was cloned into pDONR201 (Invitrogen, http://www.invitrogen.com/) according to the manufacturer’s instructions. Subsequently, the MYB46 gene was recombined into destination vector pB2GW7, in which the CaMV 35S promoter drives the gene (Karimi et al., 2002), using the Gateway LR reaction (Invitrogen).

The construct for inducible expression of MYB46 (pTA7002-MYB46) was created by inserting the MYB46 gene behind the upstream activation sequence of the dexamethasone (DEX)-inducible system (Aoyama and Chua, 1997). A full-length cDNA of MYB46 was PCR-amplified using primers 5′- CCCCTCGAGATGAGGAAGCCAGAGGTAGCCAT-3′ (forward) and 5′- GGGACTAGTTCATATGCTTTGTTTGAAGTTGA-3′ (reverse), designed to contain XhoI (forward) and SpeI (reverse) sites (underlined). The resulting product was inserted in the XhoI and SpeI restriction sites of the pTA7002 binary vector. The vector constructs were introduced into Agrobacterium tumefaciens strain C58, which was used to transform Arabidopsis thaliana (Col-0) by the floral-dip method as described by Clough and Bent (1998).

Chemical treatment of Arabidopsis plants

The pTA7002-MYB46 transgenic plants were grown on soil for 14 days and treated with DEX (dexamethasone, Sigma, http://www.sigmaaldrich.com/). DEX was applied by spraying at 10 μm with 0.02% silwet surfactant (Lehle Seeds, http://www.arabidopsis.com/). Control plants were treated with water containing the same concentration of surfactant. After treatment, young rosette leaves (5–7th leaves) without petioles were harvested at the indicated times, frozen immediately with liquid N2, then stored until used.

RNA extraction and RT-PCR

Total RNAs were extracted using Trizol reagent (Invitrogen, http://www.invitrogen.com/) according to the manufacturer’s instructions. Five micrograms of total RNA were reverse-transcribed using Superscript II reverse transcriptase (Invitrogen) in 20 μl reactions. RT-PCR was performed using 1 μl of the reaction products as a template. Amplified DNA fragments were separated on 1% agarose gels and stained with ethidium bromide. The primers used for RT-PCR are shown in Table S6.

Transcriptome analysis I: Affymetrix GeneChip

Total RNAs were isolated from young rosette leaves (5–7th leaves) of pTA7002-MYB46 plants that were treated with or without DEX as described above. A total of seven samples were prepared: 0 h control, and 1, 3 or 6 h with or without DEX treatment. The methods for preparation of cRNA from mRNA, as well as subsequent steps for hybridization and scanning of the ATH1 GeneChip arrays, were performed as described previously (Ko et al., 2004; Ko and Han, 2004; Ko et al., 2007). The signal intensities and ‘presence/absence’ expression calls were determined using MAS 5.0 (Affymetrix, http://www.affymetrix.com) with default parameters. Further data normalization was performed using GeneSpring software (Agilent, http://www.agilent.com). The resulting signal intensity values, a reflection of the abundance of a given mRNA species relative to the total mRNA population, were used to calculate the fold change of expression of individual genes between the samples. Differentially regulated genes were identified using a threefold threshold in their signal intensities as statistically significant (‘change’ calling by MAS 5.0). The microarray data may be accessed from the GEO database under accession number GSE16143.

Transcriptome analysis II: Illumina DGE

Total RNAs were prepared as described above. Libraries were prepared from total RNA using the NlaIII DGE kit (Illumina Inc., http://www.illumina.com). Briefly, mRNA was isolated by hybridization to oligo(dT)-coated magnetic capture beads. Reverse transcription to form double-stranded DNA was performed on the bead. This DNA was digested using the NlaIII restriction enzyme, leaving a short 3′ portion of the gene attached to the bead. An adapter that contains an internal MmeI type II restriction site was then ligated to the NlaIII site. Digestion with MmeI resulted in release of the construct which contains a short 17 base tag associated with the corresponding gene. Sequencing of these populations of gene tags was performed according to the Illumina protocol (see step-by-step protocol described by Hafner et al., 2008) at the Research Technology Support Facility of Michigan State University using an Illumina GA2 sequencer.

Each filter-passed read was aligned to the predicted ‘tag-ome’ of Arabidopsis thaliana (based on the TAIR version 7 release). The tag-ome consists of 17 bp sequences downstream of an NlaIII restriction site. For alignment, up to two mismatches were allowed, and tags were classified as ‘genic’ or ‘inter-genic’. Genic tags lie within the known gene model, while inter-genic tags lie completely outside any gene model. Using genic tags, expression profiles from 19 389 loci were obtained that had at least two reads in at least one of the samples. The counts were normalized to the number of filter-passed reads obtained for each sample, and are expressed as transcripts per million (TPM). In addition, a total of 59 466 tags from inter-genic regions, which lie outside predicted genes, were observed. The expression profiles of each tag and the chromosomal location/position are shown in Table S5, with a hyperlink to the TAIR GBrowser of each tag.

Transcriptional activation analysis

Preparation of Arabidopsis leaf protoplasts and transient transfection of reporter and effector constructs were performed as described by Yoo et al. (2007). Briefly, transfected protoplasts were lysed, and the soluble extracts were used for analysis of GUS after 16 h of incubation. For the time-course experiment, transfected protoplasts were harvested at the indicated time for GUS activity measurement. In each experiment, the expression level of the GUS reporter gene in the protoplasts transfected with the reporter construct alone was used as the control. The data are the means of three biological replications ± SE.

We used the pTrGUS vector to produce both effector and reporter constructs. pTrGUS is homemade vector designed to reduce the vector size to approximately 4 kb to improve transfection efficiency, and was provided by Dr Soo-Un Kim (Department of Agricultural Biotechnology, Seoul National University, Korea). The pSMGFP vector backbone (GenBank accession number U70495) was used to replace the GFP region with a 35S promoter–GUS–NOS (nopaline synthase) terminator region from pBI121 (Clontech, http://www.clontech.com/). For effector constructs, full-length cDNAs of transcription factors were ligated between the CaMV 35S promoter and the NOS terminator after removing GUS from the pTrGUS vector. The reporter constructs were created by placing 1 kb promoter fragments (located between −1 to −1000 bp relative to the start codon) of target genes in front of the GUS reporter gene after removing the 35S promoter region of pTrGUS. Transfection and transient expression using this vector system were successful (Figure 5b). Primers used for PCR amplification of full-length genes and promoters are summarized in Table S6.

Electrophoretic mobility shift assay (EMSA)

ANAC012 and MYB46 were fused in-frame with GST and expressed in Escherichia coli strain Rosetta gami (Novagen, http://www.emdbiosciences.com/novagen/). The recombinant ANAC012–GST and MYB46–GST proteins were purified using glutathione-immobilized particles (MagneGST; Promega, http://www.promega.com/) and used for EMSA with the AtC3H14 promoter fragments. Two AtC3H14 promoter fragments between −1 and −250 and −250 and −500 bp from the start codon of AtC3H14 were PCR-amplified and labeled with 32P-γ-ATP using T4 polynucleotide kinase (NEB, http://www.neb.com/). The labeled DNA fragments were incubated for 25 min with 30 ng of ANAC012–GST or MYB46–GST in binding buffer [10 mm Tris pH 7.5, 50 mm KCl, 1 mm DTT, 2.5% glycerol, 5 mm MgCl2, 100 μg ml−1 BSA and 50 ng μl−1 poly(dI-dC)]. Five per cent polyacrylamide gel electrophoresis (PAGE) was used to separate the recombinant protein-bound DNA fragments from the unbound ones. The gel was dried, and radioactive fragments were visualized by autoradiography.

Acknowledgements

This work was supported by the Department of Energy/Great Lakes Bioenergy Research Center and the Michigan State University Office for Biobased Technology. We thank Dr Sangmin Kim for technical assistance, Dr Soo-Un Kim (Department of Agricultural Biotechnology, Seoul National University, Korea) for the pTrGUS vector, Dr Jeff Landgraff (RTSF, MSU) and Mr Kevin Carr (RTSF, MSU) for help with the Illumina analysis, and Dr Annette Thelen (RTSF, MSU) for the ATHI GeneChip analysis.

Ancillary