A novel R2R3 MYB transcription factor NtMYBJS1 is a methyl jasmonate-dependent regulator of phenylpropanoid-conjugate biosynthesis in tobacco


  • EST information first appearing in this paper can be found in the DNA Data Bank of Japan (DDBJ) (accession numbers BP525519–BP535535).

*(fax +81 45 503 9573; e-mail igalis@psc.riken.jp).


Target metabolic and large-scale transcriptomic analyses of tobacco (Nicotiana tabacum L.) Bright Yellow-2 (BY-2) cells were employed to identify novel gene(s) involved in methyl jasmonate (MJ)-dependent function in plants. At the metabolic level, we describe the specific accumulation of several phenylpropanoid–polyamine conjugates in MJ-treated BY-2 cells. Furthermore, global gene expression analysis of MJ-treated cells using a 16K cDNA microarray containing expressed sequence tags (ESTs) from BY-2 cells revealed 828 genes that were upregulated by MJ treatment within 48 h. Using time-course expression data we identified a novel MJ-inducible R2R3 MYB-type transcription factor (NtMYBJS1) that was co-expressed in a close temporal pattern with the core phenylpropanoid genes phenylalanine ammonia-lyase (PAL) and 4-coumarate:CoA ligase (4CL). Overexpression of NtMYBJS1 in tobacco BY-2 cells caused accumulation of specific phenylpropanoid conjugates in the cells. Subsequent microarray analysis of NtMYBJS1 transgenic lines revealed that a limited number of genes, including PAL and 4CL, were specifically induced in the presence of the NtMYBJS1 transgene. These results, together with results of both antisense expression analysis and of gel mobility shift assays, strongly indicate that the NtMYBJS1 protein functions in tobacco MJ signal transduction, inducing phenylpropanoid biosynthetic genes and the accumulation of phenylpropanoid–polyamine conjugates during stress.


Bright Yellow-2






4-coumarate:CoA ligase








HF medium containing 20 μm MJ


jasmonic acid


methyl jasmonate


ornithine decarboxylase


phenylalanine ammonia-lyase


Plants usually do not have efficient avoidance mechanisms to prevent injury caused by chewing insects, larger herbivores and pathogens. Thus, plants have evolved the capacity to make each cell competent for the activation of defense responses that largely depend on the transcriptional activation of specific genes. In many cases, the regulation of these genes is controlled by the complex cross-talk of stress-specific signal molecules, such as jasmonic acid, systemin, oligosaccharides, abscisic acid (ABA), ethylene, and salicylic acid. At the molecular level, the responses of plants to wound/pathogen signals involve the accumulation of proteins for: (i) repair of damaged tissues; (ii) production of substances that inhibit the growth of predator insects or pathogens; (iii) activation of wound/pathogen signaling pathways; or (iv) adjustment of plant metabolism to the imposed nutritional demands (Leon et al., 2001).

Jasmonic acid plays a central role in plant responses to wounding by directly activating diverse mechanisms involved in healing and further defense (Creelman and Mullet, 1997; Creelman et al., 1992; Leon et al., 2001). These plant mechanisms include repair and reinforcement of damaged tissues and the production of phytoalexins (natural bioproducts), such as phenylpropanoids and alkaloids (Croteau et al., 2000). Despite the enormous importance of natural bioproducts for plants and humans (as potent antioxidant compounds in human nutrition, anticancer drugs, etc.), the identification of genes corresponding to individual metabolic steps in their synthesis is far from complete. Additionally, important regulatory mechanisms controlling these enzymes remain largely unknown (Endt et al., 2002; Oksman-Caldentey and Inze, 2004).

Exogenous application of jasmonates can specifically elicit natural product biosynthesis in various plant species (Imanishi et al., 1998; La Camera et al., 2004; Lee-Parsons et al., 2004; Suzuki et al., 2005). Thus, jasmonates have become one of the most frequently used tools in the investigation and engineering of natural bioproduction in plants (Goossens et al., 2003; Oksman-Caldentey and Inze, 2004; Poulev et al., 2003). Previously, metabolic reprogramming of cells by jasmonates has been shown to involve de novo synthesis of specific transcription regulators (Goossens et al., 2003), suggesting that jasmonates may also be used for the identification of master regulatory genes that control the biosynthesis of these products in plants.

In tobacco (Nicotiana tabacum L.) Bright Yellow-2 (BY-2) cells, treatment with methyl jasmonate (MJ) induces the rapid accumulation of alkaloids, mainly represented by various nicotinic acid-derived compounds (Hakkinen et al., 2004). Using the cDNA-amplified fragment length polymorphism (AFLP) method, Oksman-Caldentey and colleagues reported transcriptional activation of the complete nicotine biosynthetic pathway by MJ (Goossens et al., 2003). Interestingly, other responses of this cell line to MJ, apart from the accumulation of alkaloids, include the accumulation of pigments (Goossens et al., 2003) and inhibition of the progression of the cell cycle (Swiatek et al., 2002, 2004).

In the current study, we employed an alternative comprehensive analytical method, i.e. cDNA microarray, to identify MJ-responsive genes and to isolate novel signal transduction components, connecting the MJ-hormone signal to phenylpropanoid-related gene expression. We have shown that the newly described tobacco MJ-inducible MYB-family transcription factor NtMYBJS1 positively regulates transcription of several early phenylpropanoid-related genes, causing accumulation of the MJ-specific natural bioproducts feruloylputrescine (FP) and caffeoylputrescine (CP) in cultured cells.


Induction of natural bioproduct metabolism in BY-2 cells treated with MJ

We characterized the metabolic profile of BY-2 cells in response to MJ. Methanolic extracts from 5-day-old cells cultivated with and without MJ were compared after separation by reversed-phase–high-performance liquid chromatography (RP-HPLC) and recording of ultraviolet (UV) absorbance at wavelengths suitable for the detection of phenylpropanoids and alkaloids in tobacco. MJ application induced the accumulation of several new compounds, which were tentatively named NA, NB, J, and S (Figure 1). The major components of the peaks NA and NB, determined after additional HPLC separation and purification (data not shown), displayed UV absorbance spectra with UVmax of 259 and 266 nm, respectively. The experimentally determined spectra of peak NA and NB compounds closely resembled the UV absorbance spectra of the major tobacco nicotine alkaloid anatabine (Hakkinen et al., 2004) and two other commercially available tobacco alkaloid standards, nicotine and anabasine (Figure S1). Because the accumulation of nicotine-like alkaloids in MJ-treated tobacco BY-2 cells has been described previously by other authors (Goossens et al., 2003; Hakkinen et al., 2004), we did not attempt further characterization of these compounds. Instead, we concentrated on the identification of compounds J and S, which showed properties characteristic of putative phenylpropanoid derivatives. Compound J appeared to be one of the dominant MJ-induced products in cells grown in standard medium supplemented with MJ (2,4-D/MJ medium). In contrast, cells cultured in hormone-free (HF) medium containing 20 μm MJ (HF/MJ medium) showed a smaller increase in compound J and low levels of compound S that were too low to be measured accurately.

Figure 1.

 Content of reversed-phase–high-performance liquid chromatography (RP-HPLC)-separated UV-absorbing methanol-extractable compounds in BY-2 cells. (a) Normal (2,4-D) medium; (b) 50 μm methyl jasmonate (MJ) in normal medium (2,4-D/MJ); (c) hormone-free (HF) medium; (d) 50 μm MJ in HF medium (HF/MJ). UVabs chromatograms detected at 254 nm (blue) and 280 nm (red) are shown. All cells were extracted in 80% methanol 5 days after subculture and addition of MJ. Representative chromatograms are shown. NA, NB pyridine derivatives; S, coumaroylputrescine; J, caffeoylputrescine (CP); L, feruloylputrescine (FP).

The RP-HPLC peaks for compounds S and J showed UVabs spectra similar to those of coumaric and caffeic acid standards, respectively (Figure S1). The fragmentation pattern obtained by tandem mass spectrometry (MS-MS) analysis (Figure S2) indicated that the major compounds in the S and J peaks were p-coumaric and caffeic acids conjugated with putrescine, i.e. p-coumaroylputrescine and CP, respectively. Another MJ-inducible derivative of the phenylpropanoid pathway that was also induced in cells cultured in HF medium was similarly identified as FP (peak L, Figure 1; Figures S1 and S2).

Microarray analysis of MJ-induced genes

Next, we used our 16K tobacco BY-2 cDNA microarray to identify genes that functioned in natural bioproduct formation as described in the previous section. To identify these genes, the following assumptions were made: (1) the genes should have higher expression levels in MJ-treated cells relative to their expression in HF (control) medium; (2) the genes should show elevated net expression compared with the washed (W; i.e. before the addition of MJ) cells that accumulate lower steady-state amounts of the phenylpropanoid–putrescine-type conjugates and nicotine-type alkaloids. In implementing these criteria, microarray data were processed in two steps as described below.

  • (i) A data set of 828 genes was produced (Table S2), the expression of which was MJ-induced to at least 2.5-fold above that of the HF control at any point in the analysis (3, 6, 24 or 48 h; see Experimental procedures for details on data processing). Similarly, the expression of 938 genes was downregulated by MJ (see Table S3).
  • (ii) From the upregulated genes, a subset of 439 MJ-induced genes with ≥2.5-fold expression relative to their expression in W cells was selected (Figure 2a). The MJ data for each time-point were processed individually (3 h/W, 6 h/W, 24 h/W and 48 h/W) and the genes that showed an increase in expression for at least one time-point were included in the analysis.
Figure 2.

 Classification of methyl jasmonate (MJ)-induced genes. (a) All genes upregulated by MJ. A group of 439 expressed sequence tags (ESTs) showed an increased expression of at least 2.5-fold relative to that in washed (W) cells. Time-points for cluster analysis are indicated in hours on the x-axes. Normalized relative expression levels are shown on the y-axes. Information about individual MJ-induced genes can be found in Table S2 (available as supplementary material online). (b) The group of 439 ESTs was classified by the self-organizing map clustering method into six subgroups, 1–6, on the basis of expression profiles during the time course of the experiment. (c) Representation of specific classes of regulatory proteins among 828 MJ-upregulated genes (E-value ≥10−4, Expression data in Table S4).

Subsequently, we categorized the selected genes based on their kinetic changes in expression level. All 439 genes were classified into six distinct groups using self-organizing map (SOM) clustering (Figure 2b). Groups 1, 2 and 3 contained genes with different amplitudes of constitutive induction over the cultivation period of the experiment (3–48 h), group 4 contained early but transiently induced genes (3–24 h), group 5 included genes with a late response to MJ (24–48 h), and most of the genes in group 6 had less regular patterns of weaker induction over the latter period of the experiment (24–48 h).

Phenylpropanoid pathway and putative regulators

A selection of MJ-upregulated genes that function in phenylpropanoid or polyamine metabolism, or possess potential regulatory functions in tobacco plants, are shown in Table 1.

Table 1.   Methyl jasmonate (MJ)-induced phenolic/polyamine-related genes and regulators in tobacco BY-2 cells
 e-valueAnnotationOrgExperiment AExperiment B
3 h6 h24 h48 h3 h6 h24 h48 h3 h6 h24 h48 h3 h6 h24 h48 h
  1. Regulators are shown in italics, and phenolic or polyamine-related genes in regular font. Organism abbreviations: At, Arabidopsis thaliana; Le, Lycopersicon esculentum; Nt, Nicotiana tabacum; Ph, Petunia × hybrida; Gm, Glycine max; Os, Oryza sativa; Ca, Capsicum annuum; Cr, Catharanthus roseus; Mx, Malus xiaojinensis; Md, Malus × domestica; Pc, Populus × canescens; Cc, Clarkia concinna; Rs, Rauvolfia serpentina; Ee, Euphorbia esula; Sb, Sorghum bicolor. DB cloned, database sequence used for cloning.

Group 1
 BP1298063.00E-26Leucine-rich repeat transmembrane protein kinase, putativeAt0.
 BP1307386.00E-15F-box protein familyAt0.
 BP1308024.00E-19F-box protein (SKP1 interacting partner 3-related)At0.
 BP1315018.00E-17Symbiotic ammonium transport SAT1& bHLH domain proteinGm0.
 BP1316783.00E-49bHLH transcriptional regulatorLe0.
 BP1373051.00E-64myb-related protein Ph2Ph0.
 BP5320672.00E-59Ethylene-responsive transcriptional coactivatorLe0.
 BP1336072.00E-40Phenylalanine ammonia-lyaseNt0.
 BP1356983.00E-22Probable amino acid acetyltransferaseAt0.
 BP5282371.00E-58Polyphenol oxidase E (Catechol oxidase)Le0.
 BP5290678.00E-71Hydroxycinnamoyl CoA quinate transferaseNt0.
 BP5300311.00E-66Phospho-2-dehydro-3-deoxyheptonate aldolase 1Nt0.
 BP5304053.00E-65Putative chorismate mutase/prephenate dehydrataseAt0.
 BP5317651.00E-58Hydroxycinnamoyl CoA quinate transferaseNt0.
 BP5320949.00E-64S-adenosylmethionine synthetase 3Nt0.
 D89984DB clonedOrnithine decarboxylaseNt0.
 D43773DB cloned4-coumarate:coenzyme A ligaseNt0.
 AB236952PCR/clonedCinnamate 4-hydroxylaseNt0.
Group 2
 BP5255637.00E-27Calmodulin-binding protein, pheromone receptor-like proteinAt0.
 BP1335243.00E-49Caffeoyl-CoA O-methyltransferase-like protein]At0.
Group 3
 BP5260412.00E-28Ornithine decarboxylaseGm0.
 BP532193 Putrescine methyl transferase (3′ UTR)Nt0.
Group 4
 BP1315168.00E-13Serine/threonine kinase-relatedAt0.
 BP5299855.00E-34Putative RING zinc finger proteinAt0.
 BP5328505.00E-20Sodium-inducible calcium-binding proteinAt0.
 BP1334005.00E-41Phospho-2-dehydro-3-deoxyheptonate aldolase 1Nt0.
 BP5258013.00E-40p-coumarate 3-hydroxylase cytochrome P450 98A3At0.
 BP5338427.00E-92Prephenate dehydratase familyAt0.
Group 5
 BP1287983.00E-25Jasmonic acid 1 homeobox associated leucine zipperLe0.
 BP1297131.00E-82Homeobox 2 proteinLe0.
 BP1308864.00E-23Histidine-containing phosphotransfer proteinCr0.
 BP1367445.00E-10Calcium-binding EF-hand family proteinAt1.
 BP5287597.00E-11bZIP proteinAt0.
 BP5303091.00E-11Transcription factor Myb1Mx1.
 BP5320224.00E-39Putative RING proteinPc0.
 BP5346898.00E-17Zinc finger (C3HC4-type RING finger) protein familyAt0.
 AF032465DB ClonedSerine threonine protein kinase WAPKNt0.
 BP1288652.00E-68Lignin forming anionic peroxidase precursorNt0.
 BP1302013.00E-09Hydroxycinnamoyl benzoyltransferase-relatedAt0.
 BP1330794.00E-38P450 geraniol 10-hydroxylaseCr0.
 BP5276111.00E-06Acetyl-CoA:benzylalcohol acetyltranferaseCc0.
 BP5285934.00E-35P450 geraniol 10-hydroxylaseCr0.
 BP5303762.00E-16Suberization-associated anionic peroxidase 2 precursorLe0.
 BP5312992.00E-72Hydroquinone glucosyltransferase (Arbutin synthase)Rs0.
 BP5319231.00E-24Caffeic acid 3-O-methyltransferase (COMT)Ca1.
 BP5322434.00E-14Caffeic acid 3-O-methyltransferase (COMT)Ca1.
 BP5344238.00E-18Suberization-associated anionic peroxidase 2 precursorLe0.
 X62343DB clonedCinnamyl-alcohol dehydrogenaseNt1.
Group 6
 BP1301717.00E-45Ethylene response element binding protein 1Nt0.
 BP1303043.00E-08OSJNBa0079M09.4 lectin protein kinaseOs0.
 BP1304959.00E-13F-box protein family, AtFBL11At1.
 BP1328875.00E-57Mitogen-activated protein kinase (MAPK), putative (MPK9)At1.
 BP1359432.00E-26Protein kinase familyAt0.
 BP1367382.00E-20Ankyrin repeat protein familyAt1.
 BP1369105.00E-52Putative myb-related proteinOs0.
 BP1370111.00E-15Similar to calmodulin-binding protein∼gene_id:K19B1.18At1.
 BP5262513.00E-19Probable serine/threonine kinase (EC 2.7.1.-) SNFL1Sb0.
 BP5291589.00E-15Armadillo repeat containing proteinAt1.
 BP5311002.00E-32Serine/threonine protein kinaseOs0.
 BP5319925.00E-59Putative flavonol synthase-like proteinEe1.

This analysis highlighted two groups of genes that contained the majority of known phenylpropanoid-related genes: (i) constitutively induced transcripts at a moderate level in group 1, and (ii) transcripts with late induction in group 5. Notably, group 1 contained several early-induced, rate-limiting genes involved in phenylpropanoid metabolism, namely PAL, cinnamate-4-hydroxylase (C4H) and 4CL.

Classification of the genes into groups based on kinetic expression profiles enabled us to identify several putative regulators that showed similar expression profiles to target metabolic genes. For example, a group 1-classified BY-2 clone BP137305 (DNA Data Bank of Japan accession no. AB236951) was co-expressed in close temporal proximity to the PAL and 4CL genes, and encoded a novel full-length cDNA for the jasmonate-specific MYB transcription factor gene. Additionally, other possible regulators detected were an uncharacterized group 1-classified bHLH-like gene (BP131678) and a group 5 MYB-like gene BP530309. Previously, both types of genes, MYB and bHLH genes, have been shown to regulate natural bioproduct metabolism, especially phenylpropanoid and flavonoid biosynthesis. Apart from these two examples, dozens of other potential MJ-responsive regulators were identified (Figure 2c and Table S4).

Hydroxycinnamoyl–polyamine conjugate formation

The role of the rate-limiting enzyme ornithine decarboxylase (ODC) in polyamine biosynthesis is to provide the substrates required for the synthesis of polyamine–phenylpropanoid conjugates and nicotine-type alkaloids. The two ODC genes present on the microarray (D89984 and BP526041) were both strongly induced by MJ but classified differentially in group 1 (D89984) and group 3 (BP526041), respectively. We speculate that this differential expression could reflect the differential roles of these genes in phenylpropanoid and alkaloid metabolism (see Discussion).

Characterization of the novel MJ-induced NtMYBJS1 gene

We further characterized the candidate gene NtMYBJS1 (BP137305; DDBJ accession no. AB236951). The expression pattern of this gene was temporally nearly identical to those of the PAL and 4CL genes in MJ-treated BY-2 cells. Microarray analysis showed that transcription of the NtMYBJS1 gene was induced by MJ at an early time-point (3 h) and remained elevated for up to 48 h. Independent reverse transcriptase (RT)–PCR analysis confirmed the microarray data and showed that the gene was also expressed at a later time-point (72 h), and that its expression was not induced by other plant hormones in the HF medium, including salicylic and abscisic acids (Figure 3c).

Figure 3.

 Sequence comparison and expression analysis of expressed sequence tag (EST) BP137305 (DNA Data Bank of Japan accession number AB236951) encoding the MYB-like transcription factor NtMYBJS1. (a) Hierarchical clustering analysis of 125 Arabidopsis R2R3 MYB proteins (Stracke et al., 2001) supplemented with sequences of stress/pathogen/hormone-inducible MYBs (in red) or MYBs that function in phenylpropanoid biosynthesis from other species; AmMYBMIXTA, X79108; HvMYBGA, X87690; PhMYBAN2, AF146702; ZmMYBP1, AY702552; ZmMYBP, P27898; ZmMYBC1, X06333; ZmMYB-IF35, AF521880; AmMYB308, JQ0960; AmMYB305, JQ0958; AmMYB340, JQ0959; NbMYB1, AAS55703; NtMYBAS1, AAG28526; NtMYBAS2, AAG28525; NtMYBLBM4, BAA88224; NtMYBLBM2, BAA88222; NtMYBLBM1, BAA88221; NtMYBLBM3, BAA88223; PtMYB4, AAQ62540; Zm1, P20024; Zm38, S04899; ZmMYBPl, T03972; OsMYBC1, BAD04030; OsJAMYB, AAK08983; GmMYB29A1, BAA81730; and PsMYB26, CAA71992. (b) Alignment of NtMYBJS1-related proteins. The conserved repeated amino acid motif [I/V]D[E/D][S/N]FW–[M/Y]xFW in MYB genes is highlighted in yellow. (c) Expression of the NtMYBJS1 gene in hormone-treated BY-2 cells. Expression of NtMYBJS1 was examined by semiquantitative RT-PCR analysis of first-strand cDNAs prepared from stationary (S), washed (W), and hormone-treated cells at 3, 6, 24, 48 and 72 h after the addition of hormone. HF, hormone-free; SA, 20 μm salicylic acid in HF; MJ, 20 μm methyl jasmonate in HF; ABA, 20 μm abscisic acid in HF. α-tubulin (TUB) was used as the RT-PCR endogenous control.

The NtMYBJS1 gene encodes a protein with the typical R2R3 DNA-binding domain of the MYB transcription factor family. The clustalw (Chenna et al., 2003) comparison of the NtMYBJS1 protein with 125 Arabidopsis R2R3 MYB protein sequences (Stracke et al., 2001) and several other MYBs from different species is shown in Figure 3(a). The NtMYBJS1 protein was grouped with several wounding-, pathogen- and UV-B-inducible proteins (in red), suggesting that this group may be commonly involved in plant responses to environmental stress. Furthermore, we found a common repeated sequence motif [I/V]D[E/D][S/N]FW–[M/Y]xFW in NtMYBJS1 and in some of the NtMYBJS1 closely related genes (thicker-line-highlighted cluster of genes from Figure 3a is aligned in Figure 3b), further suggesting that these genes possibly have a conserved function in the plant stress response.

Overexpression of the NtMYBJS1 gene in tobacco BY-2 cells under the constitutive CaMV35S promoter resulted in cells with normal growth and morphology. However, in contrast to control cells transformed with vector alone, the cells overexpressing NtMYBJS1 accumulated a dark brown/gray substance(s) upon aging when grown on Gellan gum-supported plates (data not shown). Suspension cultures of the cells overexpressing NtMYBJS1 also appeared normal during a 1-week cultivation cycle in standard medium; however, when 7-day-old (stationary) cultures were left standing without shaking at room temperature for 1–2 days they turned brown/gray (Figure 4a). This suggested that the transgenic cells accumulated more dark-colored compounds, supposedly phenolics, when stressed.

Figure 4.

 Phenotypes and expression of phenylpropanoid genes in NtMYBJS1ox lines. (a) Phenotype of the NtMYBJS1ox line. Development of a dark-brown color in the representative stationary phase NtMYBJS1-overexpressing line T4 (right), after 3 days without shaking at room temperature, and the no-insert transformed control, C1 (left). (b) Expression of phenylpropanoid-related genes was examined by quantitative RT-PCR on first-strand cDNAs prepared from 3-day-old BY-2 cells. Five control (C1–C5; empty-vector-transformed) and nine NtMYBJS1 independently transformed cell lines (T1–T9) were examined. Each value represents the mean of three determinations ± standard error. PAL-A (B), phenylalanine ammonia-lyase A (B); C4H, cinnamate-4-hydroxylase; C3H, coumarate-3-hydroxylase; 4CL1(2), 4-coumarate:CoA ligase 1(2). Each data point is normalized to the α-tubulin endogenous control.

We then utilized the tobacco microarray to determine possible transcriptional targets of the NtMYBJS1 gene product. Two control (C1 and C2) and two transgenic (T3 and T4) cell lines grown in standard medium were analyzed at day 3 (log phase) and day 5 (early stationary phase). Data analysis with the genespring software (Agilent Technologies, Palo Alto, CA, USA; Find Similar function; minimum correlation coefficient of at least 0.90) identified the 4CL2 spot (U50846) as the gene most significantly co-expressed with the NtMYBJS1 gene (approximately 8-fold induction at both time-points; Table S5). Furthermore, the transcription of another phenylpropanoid gene, C4H (AB236952), was coordinately induced approximately 3-fold in transgenic cell lines. Additionally, expression of the PAL-B gene (BP133607) and the coumarate-3-hydroxylase gene (BP525801) was also induced, but at lower levels (1.5–2-fold). Manual processing of the data, using an induction threshold at least 2.5-fold higher than that of the control, resulted in a similar list, with the gene encoding 4CL2 at the top (Table S6). Several other genes were also significantly induced at a single time-point (day 3 or 5), suggesting that NtMYBJS1 may interact with these gene targets in a growth-dependent fashion or, alternatively, that their induction is indirectly related to changes in the metabolic composition of the cells induced by overexpression of the NtMYBJS1 protein. The latter may be particularly relevant for the polyamine biosynthesis gene ODC (D89984) and the S-adenosylmethionine decarboxylase gene (BP530380) which were induced mainly at the later time-point in the culture (day 5).

We then compared the expression levels of PAL, C3H, C4H, and 4CL genes in 3-day-old cells in five control (no insert) and nine cell lines overexpressing the NtMYBJS1 transgene by quantitative RT–PCR. All genes tested, except the coumarate-3-hydroxylase gene, showed significantly increased expression in the transgenic cell lines (Figure 4b). Of these, the PAL-A and 4CL2 genes showed patterns of expression closest to that of the NtMYBJS1 gene. Basal expression of PAL-B, C4H, and 4CL1 genes was also detected in the control cell lines, suggesting that these genes are expressed constitutively in BY-2 cells, and are further induced by MJ through the action of the NtMYBJS1 gene. Consistent with our data, the transcription of PAL-A, but not PAL-B, has been shown to be substantially induced by MJ and elicitor in tobacco cells (Taguchi et al., 1998).

Analysis of the metabolites in 7-day-old NtmybJSox lines showed that the transgenic cells contained increased amounts of FP (peak L) and CP (peak J; Figure 5a–c) relative to the control transformants and untransformed cells. The accumulation of CP (and FP) (Figure 5c) was consistent with the observed transgene/target gene expression data (Figure 4b). Interestingly, and in contrast to MJ-treated cells, cells overexpressing NtMYBJS1 did not contain any nicotinic-acid-derived compounds (peaks NA and NB in Figure 1b,d). In fact, NtMYBJS1 overexpression did not substantially induce any of the nicotine synthesis-related genes on the array.

Figure 5.

 UV-absorbance reversed-phase–high-performance liquid chromatography (RP-HPLC) profiles for wild type (WT) and NtMYBJS1ox BY-2 cells cultivated in standard medium. (a) Methanolic extract from control cell line C1 (no-insert-transformed) separated by RP-HPLC. (b) Extract from a NtMYBJS1-overexpressing representative line, T3. Extracts were prepared from 7-day-old (stationary) cells. Representative chromatograms are shown. UVabs spectra detected at 254 nm (blue) and 280 nm (red) are shown. J, caffeoylputrescine (CP); L, feruloylputrescine (FP); X, unknown glucoside. (c) Accumulation of CP and FP in the stationary stage of NtMYBJS1 transgenic (T1–T9) and control (no-insert-transformed, C1–C5) cell lines. WT, wild-type cells; WT/MJ, wild-type cells treated with 50 μm methyl jasmonate (MJ). Each value represents the mean of four replicates of the experiment ± standard error.

Binding of NtMYBJS1 to the promoter region of PAL genes in vitro

Direct interaction of the NtMYBJS1 protein with target promoters was examined using the electromobility shift assay (EMSA). Initially, we attempted to express the full-length recombinant NtMYBJS1 protein tagged with a six-histidine tag in Escherichia coli. However, we could not obtain the full-length protein from E. coli because its expression appeared to be highly toxic to the host cells. We thus tested the expression of a carboxy-terminal deletion derivative of the protein, where part of the protein encoding the potential activator domain was removed (amino acids 127–275). A similar truncation strategy has been used previously (Ramsay et al., 1992; Romero et al., 1998; Solano et al., 1997; Yang et al., 2001) to assess the interaction of the MYB DNA-binding domain and to probe DNA by EMSA. The truncated protein, NtMYBJS1ΔC1, containing an intact conserved MYB DNA-binding domain (amino acids 1–126) and a six-histidine tag, showed very low toxicity to the host cells and the protein was efficiently expressed at high levels.

Two PAL promoters belonging to the PAL-A and PAL-B genes were used for analysis. The expression of PAL-B gene (BP133607) was upregulated by MJ in microarray experiments and the expression of both genes was elevated in NtMYBJS1 gene overexpressing lines. Each promoter was divided into four subfragments and the capacities of the recombinant NtMYBJS1ΔC1 to bind to these DNA fragments were examined (Figure 6a).

Figure 6.

 Electromobility shift assay (EMSA) analysis of NtMYBJS1ΔC1–DNA binding affinity. (a) PAL-A and PAL-B promoter DNA fragments used for binding assays. (b) Binding of the NtMYBJS1ΔC1 protein to isolated promoter fragments A1–B4 at an equimolar DNA/protein ratio. (c) Binding of the NtMYBJS1ΔC1 protein to A1 and A3 fragments shows the presence of multiple binding sites in the A3 DNA fragment. (d) Binding of the NtMYBJS1ΔC1 protein to consensus and mutant synthetic oligonucleotide fragments; O1, consensus 1; O2, consensus 2; M1, mutant 1; M2, mutant 2. (e) Control reaction with NtMYBJS1ΔC1 protein and negative protein control extract from Escherichia coli expressing empty pET23a vector using either promoter fragment A3 or O1 synthetic oligonucleotide.

At lower protein concentrations, the NtMYBJS1ΔC1 protein bound mainly to the A3 and A4 fragments derived from the PAL-A promoter, and to B3 and B4 fragments of the PAL-B promoter (Figure 6b) in a single shifted band. When the molar ratio of the NtMYBJS1ΔC1 protein to probe DNA was increased, multiple shifted bands were observed (Figure 6c). This multiple-band pattern has previously been observed in EMSA with DNA probes containing more than one binding site (Fournier et al., 2000). Close analysis of the PAL promoter fragments revealed that the number of shifts (under maximum binding conditions) roughly correlated with the frequency of the proposed AACC MYB-binding core motif in the PAL promoter fragments (Table 2).

Table 2.   Potential NtMYBJS1ΔC1 protein binding sites in PAL-A and PAL-B promoter fragments containing the ‘aacc’ conserved motif
  1. Sequences in italics are located on the (−) DNA strand; location of the fragments relative to the PAL-A and PAL-B promoters is described in Figure 6. Underlined sequences are identical to oligo-DNA O1, and the boxed sequence corresponds to the O2 motif tested in an electromobility shift assay (EMSA) using the NtMYBJS1ΔC1 protein.

Sequencetccaaccaatcaaaaccaaaaacaacccctaataaccagg tacaaccgagcttaacccctctcaaccgtt
aacaacctaaaaaaaccgacaccaacccccaataacccaa  aacaacccaaaacaaccacc
cctaaccaattcaaaccaacaacaacccaaacaaaccaac  tctaaccttctacaacccct
  ccaaacctttaccaacccct  accaaccaaccacaaccttt
  accaaccccc   accaacccaa 
  ctcaaccgct   ataaacccat 
Max binding1–21–24–53–40–1123

Based on promoter sequence data and the number of shifted bands observed, two oligo-DNAs containing the putative NtMYBJS1ΔC1-binding sites ACCAACCCC (O1) and AACAACCAC (O2) were designed and tested in EMSA with NtMYBJS1ΔC1 (Figure 6d). Both binding sites were embedded into the neutral part of the B1 fragment (Ng) that showed no binding activity with the NtMYBJS1ΔC1 protein. As expected, the protein showed selective binding to oligo-DNAs that contained putative consensus-binding sites (Figure 6d). The binding capacity of oligonucleotide O2 was lower than that of oligonucleotide O1 and was further diminished by the introduction of mutations into the core binding site. A single A to G transition in the AACC core motif in the M1 oligonucleotide dramatically reduced protein binding, and two mutant bases in the M2-oligo almost completely abolished binding capacity. In summary, the EMSA results showed that the MYB DNA-binding domain of the NtMYBJS1 protein can bind specifically to PAL-A and PAL-B promoters.

Gene expression in a NtMYBJS1 antisense-suppressed cell line

We then used an antisense RNA strategy to suppress the expression of the NtMYBJS1 gene in the presence of MJ to assess the effects of the gene on its putative target genes (PAL-A, PAL-B, C4H, 4CL1 and 4CL2). The full-length NtMYBJS1 gene was expressed in an antisense orientation under a constitutive CaMV 35S promoter. The A4 cell line, which showed consistently lower levels of induction of NtMYBJS1 by MJ compared with the five control (vector-only-transformed) and wild-type cells, was chosen from the eight examined putative antisense lines and used in further investigations. The two cell lines (representative control C14 and antisense A4) were used to compare the expression levels of the PAL-A, PAL-B, C4H, 4CL1 and 4CL2 genes upon induction with 50 μm MJ in standard medium.

All these potential NtMYBJS1 target genes showed consistently lower levels of expression in the antisense A4 cell line compared with the C14 control (Figure 7). This observation further supports the notion that a causative relationship exists between the expression of the NtMYBJS1 gene and that of these target genes. Despite the partial reduction in expression of the early phenylpropanoid pathway genes, there was no consistent reduction in accumulation of CP or FP in the A4 cell line compared with the C14 cultures when incubated for 3 days in the presence of MJ (data not shown). We assume that the partial suppression of NtMYBJS1 gene expression by the antisense approach was not sufficient to exert an effect on the total cumulative amount of CP and FP over the 3-day cultivation period.

Figure 7.

 Expression of NtMYBJS1 putative target genes in the empty-vector-transformed control cell line C14 and NtMYBJS1 antisense cell line A4. Cells were cultivated in the presence of 50 μm methyl jasmonate (MJ) for 3 days and cDNA was prepared from total RNA extracts. Transcript levels of the NtMYBJS1, PAL-A, PAL-B, C4H, 4CL1 and 4CL2 genes were determined by quantitative RT-PCR. Tubulin-normalized relative expression levels are shown on the y-axes. Each value represents the mean of three replicates of the experiment ± standard error. The significance of decreases in expression measured in three independent replicates of the experiment was determined using Student's t-test for unpaired sample means; P ≤ 0.01 (**); P ≤ 0.10 (*).


Methyl jasmonate activates the biosynthesis of many natural compounds, previously know as secondary metabolites, in plants. In this study, we used tobacco BY-2 cells and large-scale gene expression analysis to further investigate MJ-dependent metabolic changes in this model system. One of the main aims of our work was the characterization of novel MJ signal transduction components. Time-course expression data for various MJ-responsive putative regulators were aligned with the expression data for known metabolic genes, resulting in the identification of a novel MJ-regulated MYB gene in tobacco. This transcriptionally controlled NtMYBJS1 gene was shown to interact with several phenylpropanoid biosynthesis genes in an MJ-dependent fashion, leading to the enhanced accumulation of hydroxycinnamoyl–polyamine conjugates in the cells.

NtMYBJS1 is an MJ-dependent regulator of phenylpropanoid biosynthesis

A number of phenylpropanoid-related MJ-inducible genes were identified using current tobacco microarray (see Matsuoka and Galis, 2006, for further details). In addition, we found good agreement between metabolic data and classification of several of these genes in the constitutively MJ-induced group 1. Because the MJ responsiveness of the basic phenylpropanoid biosynthesis genes has already been documented (4CL: Lee and Douglas, 1996; Suzuki et al., 2005, and this report; PAL: Ellard-Ivey and Douglas, 1996; Sharan et al., 1998; Suzuki et al., 2005; Taguchi et al., 1998; Yang et al., 2001, and this report; CHS: Richard et al., 2000), we concentrated on the identification of potential MJ-signal transduction elements in this study.

Among the possible candidates, MYB proteins are known regulators of diverse metabolic and developmental processes in plants. For example, MYB proteins are involved in control of trichome differentiation (AtMYB0/GLABROUS 1; Oppenheimer et al., 1991), root hair development (AtMYB66/WEREWOLF; Lee and Schiefelbein, 1999), leaf patterning (AtMYB91/AS; Byrne et al., 2000) and regulation of cell cycle by MYB3R factors (Ito et al., 2001). However, the best established role of MYB proteins, and especially of their largest family, the R2R3-MYB proteins, is in the regulation of metabolic pathways in plants.

For example, the AtMYB34/ATR1 gene has been shown to regulate tryptophan biosynthesis (Bender and Fink, 1998) and AtMYB2 induced alcohol dehydrogenase gene expression during the response to low oxygen (Hoeren et al., 1998). In addition, a significant number of MYB transcription factors, including that encoded by the tobacco NtMYBJS1 gene described in this paper, have been implicated in the control of phenolic biosynthesis in plants (for reviews, see e.g. Endt et al., 2002; Jin and Martin, 1999; Weisshaar and Jenkins, 1998).

Among the best characterized examples of MYB-phenylpropanoid pathway regulators are the maize R- or B-cofactor-dependent regulators of anthocyanin biosynthesis C1 (Pazares et al., 1986, 1987) and Pl (Cone et al., 1993), and the P gene, which acts alone in activating gene transcription (Grotewold et al., 1991, 1994). The overexpression of another maize (Zea mays) MYB gene, ZmMYB-IF35, in a way similar to that of NtMYBJS1 in tobacco, induced ectopic accumulation of ferulic and chlorogenic acids and other related compounds in the cultured cells (Dias and Grotewold, 2003). The results presented in this paper suggest that the NtMYBJS1 protein also does not require a cofactor for its activity, as the overexpression of the MYB protein alone induced phenylpropanoid genes and accumulation of phenylpropanoid conjugates in tobacco cells. Other well-characterized flower-specific MYB regulators, AmMYB305, AmMYB340, AmMYB308 and AmMYB330 (Moyano et al., 1996; Sablowski et al., 1994; Tamagnone et al., 1998), were proposed to regulate accumulation of flavonoids in the flowers of Antirrhinummajus L. (snapdragon) plants. The overexpression of AmMYB308 (AmMYB330) in heterologous tobacco host plants resulted in suppressed expression of the C4H, 4CL and CAD genes (Tamagnone et al., 1998), suggesting a strong conservation of MYB DNA-binding motifs among plant species. In tobacco, an anther-specific NtmybAS1 gene has been shown to regulate two different PAL gene promoters when expressed in tobacco leaf protoplasts (Yang et al., 2001) and a lignin-related MYB protein, NtLim, which is mainly expressed in tobacco stem tissues, can control the expression of the PAL, 4CL, and CAD genes (Kawaoka et al., 2000). Another good example of MYB regulators may include the Arabidopsis AtMYB75/PAP1 gene which controls transcription levels of the PAL, chalcone synthase (CHS) and dihydroflavonol reductase (DFR) genes in anthocyanin biosynthesis (Borevitz et al., 2000; Tohge et al., 2005). Finally, the Arabidopsis AtMYB12 protein, with a high degree of functional and structural similarity to the P gene from maize, was recently identified as a flavonol-specific regulator of phenylpropanoid biosynthesis (Mehrtens et al., 2005). In the context of the many examples described in this paragraph, the NtMYBJS1 gene represents a novel member of the large family of MYB genes which specifically regulate phenylpropanoid biosynthesis in plants. Moreover, we have shown that the expression of this novel gene is controlled by MJ in a hormone-dependent manner.

Previously, several MYB genes, in a similar way to the NtMYBJS1 gene, were tentatively shown to be involved in abiotic stress and pathogen response. These include, for example, the UV-B-inducible AtMYB4 (Jin et al., 2000) and soybean (Glycine max L.) GmMYB29A1 (Shimizu et al., 2000) genes, the MJ-inducible rice (Oryza sativa L.) OsJAMYBgene (Lee et al., 2001), the ABA/dehydration-responsive AtMYB2 gene (Urao et al., 1993) and the wounding/elicitor-inducible NtMYBLBM1 gene (NtMYB2; Sugimoto et al., 2000; Takeda et al., 1999). It is noteworthy that the NtMYBLBM1, GmMYB29A1, NtMYBJS1 and Arabidopsis AtMYB13-15 proteins share certain common structural features (Figure 3b), suggesting the functional conservation of these genes in the stress/pathogen response. In support of this proposal, we found that the NtMYBJS1 gene was induced in tobacco leaves in response to wounding or direct exogenous application of MJ (IG and KM, unpublished results).

Binding of the MYB DNA-binding domain of NtMYBJS1 protein to PAL promoters (Figure 6) further strengthens the previous conclusion that NtMYBJS1 is an intrinsic transcriptional regulator of phenylpropanoid genes in tobacco. This protein preferably recognized the ACCAACCCC DNA motif in the PAL promoter sequence, which is consistent with the previously reported maize P-gene core binding sequence CC(T/A)ACC (Grotewold et al., 1994). Recently, a binding capacity of the elicitor and UV-B-inducible carrot (Daucus carota L.) DcMYB1 recombinant protein to the same core sequence tctcACCAACCCttg (box-L5) was reported (Maeda et al., 2005). In Arabidopsis, three corresponding MYB genes (AtMYB13, AtMYB14, and AtMYB15) encode proteins that have conserved motifs in common with NtMYBJS1 (Figure 3). Based on a clustalw-derived phylogenetic tree, these genes also have the most similar DNA-binding domain to the NtMYBJS1 gene (data not shown), suggesting that they could be involved in the control of similar genes in Arabidopsis. To find potential targets of these MYB genes in Arabidopsis, we carried out database searches for genes with the ACCAACCCC sequence in the promoter region, using the pattern-matching search site in The Arabidopsis Information Resource (http://www.arabidopsis.org/cgi-bin/patmatch/nph-patmatch.pl). Consequently, we found a possible linkage of this motif and phenylpropanoid genes. Among 27 184 genes, 65 contained the ACCAACCCC sequence motif within the 500-bp upstream sequence. Among these 65 genes, 12 (about 18%) encoded proteins for shikimate/phenylpropanoid or S-adenosylmethionine (SAM)/polyamine metabolism; these included two 4CL genes (At1g51680 and At3g21240), a caffeoyl-CoA 3-O-methyltransferase gene (At4g34050), a cinnamoyl-CoA reductase gene (At1g15950), two cinnamoyl-alcohol dehydrogenase genes (At4g34230 and At1g09510), two shikimate kinase genes (At4g39540 and At2g21940) and an S-adenosylmethionine synthetase gene (At2g36880). Furthermore, the wound-responsive Arabidopsis PAL1 (At2g37040) and PAL2 (At3g53260) genes contain a very similar motif (ACCAACCGC) in their promoter regions.

Taken together, these observations indicate that the NtMYBJS1 gene and its homologs are direct signaling components between MJ (hormone signal) and the accumulation of phenylpropanoid compounds in plants during stress.

The role of hydroxycinnamoyl–polyamine conjugates in plant defense

Our analysis of metabolites in MJ-treated BY-2 cells indicates that two major classes of compounds, nicotine alkaloids (Goossens et al., 2003; Hakkinen et al., 2004) and hydroxycinnamoyl–polyamine conjugates, accumulate in the cells. The latter were further identified as putrescine conjugates of caffeic acid and p-coumaric acid. One of these compounds, CP, is also the main differentially accumulated metabolite in MJ-treated Nicotiana attenuata leaves (Keinanen et al., 2001). The importance of polyamines and their conjugates in plant disease was recently reviewed by Walters (2003). In Nicotiana tabacum, hydroxycinnamoyl–polyamine conjugates accumulate in the reproductive organs, and the formation of putrescine conjugates is stimulated by viral or fungal infections (Wink, 1997). For example, hydroxycinnamic acid amides accumulate around or in necrotic lesions during the hypersensitive response to tobacco mosaic virus (TMV) infection (Martintanguy et al., 1976; Rabiti et al., 1998; Torrigiani et al., 1997). Torrigiani et al. (1997) suggested that high levels of polyamine conjugates may be required to limit virus movement and prevent systemic infection. Additionally, Martintanguy et al. (1976) showed that tobacco leaf discs treated with coumaroylputrescine and CP undergo a 90% reduction in local lesion formation following TMV inoculation.

Further to these observations, TMV infection causes a transient production of jasmonic acid (Seo et al., 2001) and induces patatin-like lipase (Dhondt et al., 2000), which is likely to be the enzyme involved in jasmonic acid biosynthesis. Thus, our result that MJ induces the production of hydroxycinnamoyl–polyamine conjugates through the action of NtMYBJS1 could explain the link between TMV infection and the accumulation of these polyamine conjugates. Future analysis of the effect of NtMYBJS1 on pathogen infection will clarify this possible linkage.

MJ-induced genes for hydroxycinnamoyl–polyamine conjugate formation

The accumulation of putrescine conjugates in BY-2 cells (Figure 1) includes two major biosynthetic pathways in tobacco: the PAL-driven phenylpropanoid pathway and the agrinine decarboxylase/ODC-controlled biosynthesis of polyamines. Interestingly, the two rate-limiting ODC genes on the BY-2 microarray showed differential induction during MJ treatments. One ODC gene, D89984, which is identical to the NtODC-2 gene (Xu et al., 2004), was classified in group 1 together with several key phenylpropanoid genes. This observation suggests that the NtODC-2 gene might be one of the main sources of putrescine for the synthesis of phenylpropanoid conjugates. The expression pattern of the NtODC-2 gene in NtMYBJS1ox cell lines lends further support to this hypothesis. This gene was induced in a 5-day culture of cell lines overexpressing NtMYBJS1 (Table S6), which overproduce phenylpropanoid substrates in the cells. In contrast, the alternative ODC gene on the array (BP526041) encodes a novel tobacco ODC protein, dissimilar to the proteins summarized by Xu et al. (2004), yet similar to the putative ODC from Solanum demissum. This ODC gene was classified into group 3 of the MJ-induced genes which typically show very low basal levels of gene expression in untreated cells and a strong increase in amplitude after MJ application. Significantly, the nicotine biosynthesis-controlling gene putrescine N-methyltransferase (PMT; BP532193; Xu and Timko, 2004) was co-classified with the BP526041 gene. Based on this co-expression data, we propose that the main function of this ODC enzyme may be the synthesis of putrescine required for alkaloid biosynthesis (Shoji et al., 2000). Additional experiments showing the proposed metabolic compartmentalization of ODC enzymes in plant cells are required to investigate this hypothesis.

In the previous paragraph, the role of two ODC genes in polyamine and alkaloid biosynthesis was discussed. However, the contribution of arginine decarboxylase (ADC), together with agmatine iminohydroalse and N-carbamoylputrescine amidohydrolase, in the formation of the putrescine pool in plants must also be considered. While ADC was not among the genes investigated on the current tobacco microarray, previous data suggest that this gene is also induced by MJ in BY-2 cells (Goossens et al., 2003). In addition, the EST BP528797, which is a close structural homolog of the Arabidopsis NLP1 gene with N-carbamoylputrescine amidohydrolase activity (Piotrowski et al., 2003), was upregulated in group 6 on the current microarray.

The enzymatic activity of putrescine hydroxycinnamoyl-CoA transferase required for CP and FP formation in tobacco has been previously characterized (Meurer-Grimes et al., 1989; Negrel et al., 1989, 1991, 1992). Additionally, the activity of this enzyme significantly increases in barley (Hordeum vulgare L.) during the hypersensitive reaction to powdery mildew fungus (Blumeria graminis f. sp hordei) (Cowley and Walters, 2002). Despite well-characterized enzymatic activity, the corresponding N-acetyltransferase gene, which encodes plant putrescine hydroxycinnamoyl-CoA transferase, has not been identified. Interestingly, reasonably good candidates for this gene could be found among our MJ-upregulated transcripts. For example, the BP135698 transcript, encoding a putative amino acid acetyltransferase gene, was co-regulated with the PAL, 4CL, and ODC (D89984) genes in group 1 during MJ elicitation. Additionally, the GNAT-like acetyltransferase genes, encoded by transcripts BP129058 and BP533687, not only showed induction in the presence of MJ but were also approximately twofold upregulated in cell lines overexpressing NtMYBJS1 (data not shown) with enhanced production of CP and FP (Figure 5c). Further characterization of these gene products will be required to identify the N-acetyltransferase gene.

Experimental procedures

Cell cultivation and hormone treatments

Tobacco BY-2 cells were cultivated as described previously (Matsuoka and Nakamura, 1991). For hormone treatments, 7-day-old stationary BY-2 cells were washed with hormone (auxin)-free (HF) medium as follows: 35 ml of culture in a 50-ml conical tube was centrifuged at 130–190 g in a swing rotor (TS-7, Tomy LC 100 centrifuge; TOMY TECH USA, Fremont, CA, USA) for 10 min. The supernatant was decanted and the cells washed with 20 ml of fresh HF medium. After four washes, the cells were resuspended in a 1:1 ratio with HF medium. A 5-ml aliquot of washed cells was added to 90 ml of HF medium in 300-ml Erlenmeyer flasks, supplemented with hormones dissolved in methanol (0.1% final concentration in the medium). The hormone treatments used for cDNA library construction were: HF = 0.1% methanol, 50 μm MJ, 10 μm ABA, 20 μm benzyladenine, 40 μm salicylic acid, 5 μm brassinolide, and 10 μm gibberellic acid (GA3). Sugar-starved cells were cultured in standard medium without sucrose. Cells were sampled at 2 and 20 h for RNA isolation and cDNA preparation. For microarray experiments, 20 μm MJ was added to HF medium to produce HF/MJ medium. The samples were collected by filtration and were snap-frozen in liquid nitrogen. The time-points selected for microarray analysis were: −1 h (7-day-old stationary cells), 0 h (washed cells) and cells at 3, 6, 24, and 48 h after the addition of MJ to washed cells.

RNA isolation

In all experiments, the total RNA from BY-2 cells was prepared using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) as described in the manufacturer's protocols. The poly(A)-rich RNA fraction was isolated from total RNA using the GenElute mRNA Miniprep Kit (SIGMA-ALDRICH Co., St. Louis, MO, USA) in two subsequent runs of purification following the Miniprep Kit protocol. Under these conditions the level of contaminating rRNA was usually below 5% as determined by 2100 BioAnalyzer (Agilent Technologies, Palo Alto, CA, USA).

Source libraries for EST sequencing and microarray preparation

Construction of the first cDNA library in the pGEM-Teasy vector (Promega, Madison, WI, USA) from lag, log, and stationary stage tobacco BY-2 cells was performed as described previously (Demura et al., 2002; Matsuoka et al., 2004). This library was extensively normalized by three rounds of normalization (Kohchi et al., 1995), which resulted in an enriched proportion of cDNAs with low transcript levels. Conversely, abundantly expressed genes, including transcripts encoding ribosomal proteins, were sometimes almost completely eliminated from the library. The second library was designed as a hormone-induced transcript-enriched library, using hormone-treated BY-2 cells as an initial source of RNA. This library was prepared as described for the first library with the following improvements: (i) a single round of normalization was employed to avoid complete elimination of abundant transcripts, and (ii) the pBII KS+ cloning vector (Stratagene, La Jolla, CA, USA) was used for directional cloning of the cDNAs. This strategy resulted in a better representation of abundantly expressed genes and the presence of a number of hormone-related clones in the library (see Results).

Amplification of DNA and printing of microarrays

DNA for microarrays was prepared by amplification of cDNA cloned inserts using conventional PCR and the universal M13 forward (GTAAAACGACGGCCAGT) and reverse (CAGGAAACAGCTATGAC) vector primers (pGEM-Teasy), or M13 forward (GTAAAACGACGGCCAGT) and KS reverse (TCGAGGTCGACGGTATC) vector primers (pBII KS+), with cDNA-containing plasmids as PCR templates. The microarray slides contained cDNA/EST clones described previously (Matsuoka et al., 2004), new clones BY21001–BY35384 (accession numbers BP530024–BP535535), selected clones from BY10201–BY14996 (accession numbers BP525519–BP530023), and some tobacco genes of interest that were not among the ESTs but could be uploaded from public databases (BY39001 and higher; databank accession numbers for these clones are supplied in the text and in tables; primers used to amplify the corresponding gene fragments are described in Table S1a). For these gene sequences, the DNA was PCR-amplified from tobacco BY-2 cDNA that was used as a PCR template. Corresponding PCR products were purified, cloned into the pBluescript vector (pBII KS+; Stratagene) and verified by DNA sequencing. The plasmid clones were then processed together with the cDNA/ESTs during microarray preparation.

A complete set of 16 224 amplified cDNA inserts was used to print a custom 16K BY-2 microarray. The complete microarray consisted of four microarray Bar Coded Slides, Type 7 star* (Amersham Biosciences UK Limited, Buckinghamshire, UK), with each PCR-amplified cDNA sample spotted in duplicate on the slides. The Array Spotter Generation III (Molecular Dynamics, Sunnyvale, CA, USA) was used for printing of slides.

Probe labeling and microarray hybridization

We used 1 μg of labeled mRNA for single-slide hybridizations. Labeling was performed using SuperScript II reverse transcriptase (Invitrogen), random 9-mer primers, poly(A)-specific poly (dT) primers, and Cy-3/Cy-5-labeled dCTP nucleotides (Amersham Biosciences) for first-strand cDNA synthesis. Before use, the DNA on the spotted microarray slides was immobilized by cross-linking the slides under UV light (energy 50 mJ cm−2) and slides were prehybridized for 1 h at 55°C in prehybridization buffer [1% bovine serum albumin (BSA), 5× saline sodium citrate (SSC) and 0.1% sodium dodecyl sulfate (SDS)]. Hybridization of microarrays with Cy-5-labeled sample cDNA and Cy-3-labeled vector sequence as control was performed in ExpressHyb hybridization solution (Clontech, Palo Alto, CA, USA) at 60°C for 4 h as described previously (Demura et al., 2002; Matsuoka et al., 2004). The protocol was modified with the use of a Cy-3-labeled polylinker DNA fragment located between the M13 forward/reverse or M13 forward/KS reverse primer sites of the respective vectors (pGEM-Teasy and pBII KS+), instead of the labeled oligo-DNA that was used previously (Matsuoka et al., 2004).

The Cy-5 (sample signal) and Cy-3 (control signal) were obtained by scanning the hybridized slides after washing in SSC/SDS buffer solutions at 55°C (final wash in 0.1× SSC and 0.1% SDS) using a GenePix 4000A microarray scanner (Molecular Dynamics). The scans were subsequently analyzed by the MicroArray Suite iplab program (Scananalytics, BD Biosciences, Rockville, MD, USA).

Normalization and processing of microarray data

Duplicate spots for each cDNA (EST) on the microarray slides from two independent biological replicates of experiments A and B provided four data points that were processed using genespring software (Agilent Technologies, Palo Alto, CA, USA) as follows: (i) each average sample value of the duplicated spots (Cy-5-labeled cDNA) was normalized relative to the average Cy-3-labeled vector control signal (per gene normalization), and (ii) median polishing (per gene and per chip) normalization was determined by the implemented default method of the genespring software program (normalized data for all data points in both experiments A and B are shown in Table S7). The HF/MJ data spots, with an average normalized signal value at least 2.5-fold higher (or lower for suppressed clones) than the corresponding spot in the HF medium, were selected independently for experiments A and B. Only signals that simultaneously increased or decreased in both biological replicates of the experiments were considered as significant and were selected for this study.

A similar procedure was used to identify the potential targets of the MYB transcription factor NtMYBJS1. Transcription profiling data from two cell lines overexpressing NtMYBJS1, T3 and T4, were compared with two independent control samples, C1 and C2. Two time-points, day 3 and day 5 after subculture, were chosen to examine the possible influence of the growth stage of BY-2 cells on transgene interaction with BY-2 cell gene expression (normalized data for all data points are shown in Table S8).

To validate the microarray data by an independent method, we selected 14 genes and amplified the target sequences by PCR using specific primers and template cDNA identical to that used in the microarray experiments. In all cases, the microarray signal and the intensity of the PCR-amplified bands corresponded well, confirming that our microarray system reflects the actual gene expression in the cells (Figure S3).

BY-2 cell transformation, and sense and antisense NtMYBJS1 gene constructs

The full-length DNA sequence of the NtMYBJS1 gene (bp 3–947*, DDBJ entry AB236951*) was introduced into BglII-NotI sites of the pMAT137 binary vector (Yuasa et al., 2005) in both sense and antisense orientations under the control of an enhancer-duplicated CaMV35S promoter. BY-2 cells were transformed with the binary vector pMAT137-NtMYBJS1sense and -NtMYBJS1antisense constructs as described previously (Matsuoka and Nakamura, 1991).

Reverse transcriptase–polymerase chain reaction (RT-PCR)

For RT-PCR, the first-strand cDNA was prepared with M-MLV reverse transcriptase (Promega) using 1 μg of DNase-treated total RNA (1 U of RQ1 RNase-free DNase/μg total RNA) as a template in a 25-μl reaction volume. Quantitative RT-PCR was performed in a Light Cycler (Roche Applied Science, Indianapolis, IN, USA) with the QuantiTect SYBR Green PCR Kit (Qiagen, Valencia, CA, USA) as described in the manufacturer's instructions, using a specific set of primers for each amplified gene (see Table S1b), 1 μl of a 5-fold dilution of first-strand cDNA, and a 56°C annealing temperature. Conventional PCR for microarray validation purposes was performed with the same cDNA templates and GeneAmp PCR system 9700 (Applied Biosystems, Foster City, CA, USA) using 28 cycles of amplification and an annealing temperature of 56°C. The specificity of the primers, in case of close gene family members, was verified by the restriction digestion pattern of PCR-amplified products (data not shown).

Phenolic analysis and mass spectrometry

Cells for metabolite analysis were collected on filter paper and freeze-dried before homogenization and extraction in 80% methanol (0.1 ml per mg dry weight). HPLC analysis of the phenolic compounds shown in Figure 1 and 5 (chromatograms) was performed with the ÄKTA Explorer RP-HPLC system (Amersham Pharmacia Biotech AB, Uppsala, Sweden) with a 5 μm RP8 Symmetry Shield column, 250 × 4.6 mm inner diameter (i.d.) (Waters, Milford, MA, USA); solvent A: 0.5 mm tetrabutylammonium hydroxide and 0.1% acetic acid; solvent B: methanol; gradient conditions 0–10 min isocratic 100% A, 10–20 min linear gradient to 50% B, and 20–25 min linear gradient to 100% B. The amounts of CP and FP in Figure 5 (graphs) were estimated after resolution of methanolic extracts using RP-HPLC and the gradient of solvents A: 20 mm sodium acetate, pH 5; solvent B: methanol; gradient conditions 0–4.5 min isocratic 25% B, 4.5–15 min linear gradient to 70% B, and 25–30 min linear gradient to 100% B. The compounds were quantified on the basis of caffeic and ferulic acid standards as described previously (Galis et al., 2004).

For the identification of unknown compounds, the mass spectrometer (MS-MS) was set as described previously (Galis et al., 2004), except for liquid chromatographic separation, where a Synergi Polar-RP 150 × 2 mm i.d. column (Phenomenex, Torrance, CA, USA) was used for sample separation before infusion into the electrospray ion source.

UVabs spectra (220–500 nm) of RP-HPLC purified peaks were recorded using a UV spectrophotometer (Beckman Coulter DU640, Beckman Coulter, Inc., Fullerton, CA, USA).

Recombinant protein preparation and purification

Purified histidine-tagged NtMYBJS1 protein extracts were prepared by cloning of the full-length (amino acids 1–275) or truncated version of the protein (DNA-binding MYB domain amino acids 1–126; NtMYBJS1ΔC1) into a pET23a expression vector (Novagen, Madison, WI, USA) using the NdeI/XhoI restriction sites of the vector. NdeI and XhoI restriction sites were introduced into the primers used for PCR amplification of the NtMYBJS1 DNA with PfuUltra High-Fidelity DNA polymerase (Stratagene). The forward primer used in both constructs was 5′-atatatcatATGGGCAGAGCTCCTTGCTGTG-3′, and reverse primers for the full-length NtMYBJS1 protein and truncated NtMYBJS1ΔC1 protein were 5′-acacacctcgagTAATTCAGGCAATTCTAACATCAACTC-3′ and 5′-acacacctcgagCTGAGTTTCTTGAGGTTG-3′, respectively.

The truncated protein was directly purified from E. coli BL21(DE3) cells (Invitrogen), which produced a sufficient amount of the protein even without isoprophyl B-D-7-thiogalactophyranoside (IPTG) induction. IPTG induction led to complete lethality of full-length protein expression. The NtMYBJS1ΔC1 protein was purified over a Ni+ agarose column (Qiagen) under native conditions according to the manufacturer's instructions. Proteins were eluted from the column using a gradient of increasing imidazole concentration in the wash buffer (50 mm NaH2PO4, 300 mm NaCl and 130–250 mm imidazole). Subsequent dialysis was performed in buffer containing 10% glycerol and 1× Tris-buffered saline at 4°C overnight. As a negative control, protein extract from the pET23a empty vector expressing BL21(DE3) cells was processed in the same way as the expressed proteins. Aliquots of the same volume corresponding to MYB protein fractions were used as nonbinding negative control protein extracts in gel shift (EMSA) experiments (Figure 6e).

Electromobility gel shift assays

PAL-A (AB008199; bases 79–1108) and PAL-B (AB008200; bases 1–1033) promoters were first subcloned into the pENTR/D-TOPO cloning vector (Invitrogen) using PfuUltra High-Fidelity DNA polymerase (Stratagene) and tobacco BY-2 genomic DNA as a template. After sequence verification of the clones, eight fragments with an average size of approximately 250 bp (A1–A4, B1–B4; Figure 6a) were amplified by PCR and terminally labeled with Cy3-dCTP (Amersham Biosciences, Uppsala, Sweden) using terminal deoxynucleotidyl transferase (TdT; Promega) and short 3-min reactions at 37°C. Labeled probes were separated from unincorporated Cy3-dCTP using Micro Bio-Spin6 chromatography columns (Bio-Rad, Hercules, CA, USA). Oligonucleotide probes (Figure 6d; O1, O2, M1 and M2) were custom synthesized as single-stranded oligo-DNAs, annealed in equimolar concentrations, and labeled identically to the promoter fragments.

Approximately 0.1 pmol of Cy-3-labeled promoter fragments and 0.05 pmol of labeled oligonucleotide dsDNA were incubated with 0.15–3 pmol of purified recombinant NtMYBJS1ΔC1 protein in 1× binding buffer (10 mm Tris-HCl, pH 7.5, 50 mm NaCl, 0.5 mm DTT, 0.5 mm EDTA, 1 mm MgCl2 and 4% glycerol). To prevent nonspecific binding, 0.5 μg of poly(dI-dC) was included in the reaction buffer. Reactions were set up on ice and incubated for 30 min at room temperature after the addition of Cy3-labeled probes. Reaction mixtures were resolved on 3% agarose gels in 0.5× Tris-borate-EDTA buffer while placed on ice to prevent overheating of running buffer during electrophoresis at constant voltage (100 V). Gels were directly scanned using Typhoon 8600 Variable Mode Imager (Molecular Dynamics) using an emission/excitation 555 BP20/532-nm filter set and PMT voltage of 700 V at high sensitivity. The focal depth of the laser beam was set to 3 mm above the glass surface. At least three independent repeats for each experiment with PAL promoters and synthetic oligonucleotide probes were performed with qualitatively similar results. A representative example of each experiment is shown in Figure 6(b–e).


We thank Professor Hiroetsu Wabiko (Akita Prefectural University, Akita, Japan) for useful comments on the manuscript and Ms Yumiko Suzuki for patient help with processing and input of the microarray data P.S. was supported by the grant AVOZ50070508.