The R2R3-MYB transcription factor HAG1/MYB28 is a regulator of methionine-derived glucosinolate biosynthesis in Arabidopsis thaliana


(fax + 49 221 4705039; e-mail


Methionine-derived glucosinolates belong to a class of plant secondary metabolites that serve as chemoprotective compounds in plant biotic defense reactions and also exhibit strong anticancerogenic properties beneficial to human health. In a screen for the trans-activation potential of various transcription factors toward glucosinolate biosynthetic genes, we could identify the HAG1 (HIGH ALIPHATIC GLUCOSINOLATE 1, also referred to as MYB28) gene as a positive regulator of aliphatic methionine-derived glucosinolates. The content of aliphatic glucosinolates as well as transcript levels of aliphatic glucosinolate biosynthetic genes were elevated in gain-of-function mutants and decreased in HAG1 RNAi knock-down mutants. ProHAG1:GUS expression analysis revealed strong HAG1 promoter activity in generative organs and mature leaves of A. thaliana plants, the main sites of accumulation of aliphatic glucosinolates. Mechanical stimuli such as touch or wounding transiently induced HAG1/MYB28 expression in inflorescences of flowering plants, and HAG1/MYB28 over-expression reduced insect performance as revealed by weight gain assays with the generalist lepidopteran herbivore Spodoptera exigua. Expression of HAG1/MYB28 was significantly induced by glucose, indicating a novel transcriptional regulatory mechanism for the integration of carbohydrate availability upon biotic challenge. We hypothesize that HAG1/MYB28 is a novel regulator of aliphatic glucosinolate biosynthesis that controls the response to biotic challenges.


Arabidopsis thaliana plants, along with other Brassicaceae species, produce an enormous diversity of aliphatic glucosinolates (Kliebenstein et al., 2001; Mithen, 2001; Rask et al., 2000), which play an important role in the defense against micro-organisms (Brader et al., 2006; Chung et al., 2005; Kliebenstein et al., 2005a; Manici et al., 2000; Mari et al., 1996) and herbivores (Giamoustaris and Mithen, 1995; Levy et al., 2005; Mewis et al., 2005; Reymond et al., 2004). The chemoprotective and anti-carcinogenic properties of glucosinolates, on the other hand, have produced an increasing interest in the biosynthesis and regulation of methionine-derived glucosinolates and their breakdown products (Gross et al., 2000; Jeffery, 2005; Mithen et al., 2003; Nho and Jeffery, 2004;Shapiro et al., 2006). Furthermore, Brassicaceae cultivars are increasingly used as fumigants to suppress soil-borne pests and diseases in agricultural and horticultural crops (Vaughn et al., 2005; Zasada and Ferris, 2004). In plants, glucosinolates form a defense system known as ‘mustard oil bomb’: upon tissue damage, non-toxic glucosinolates are hydrolysed by myrosinases into biologically active isothiocyanates, which have been shown to be highly toxic for both generalist and specialist herbivores (Kliebenstein et al., 2005b; Rask et al., 2000; Wittstock and Gershenzon, 2002).

The most prominent glucosinolate biosynthetic precursors are the amino acids methionine, tryptophan and phenylalanine, leading to aliphatic, indolic and aromatic glucosinolates, respectively. The biosynthesis of glucosinolates proceeds in several steps (Figure 1) (Grubb and Abel, 2006; Halkier and Gershenzon, 2006). First, methionine or other amino acids are transaminated by a cytoplasmic branched-chain aminotransferase (BCAT4) leading to corresponding α-keto-acids (Schuster et al., 2006). Subsequent side-chain elongation of the amino acid involves the action of chloroplastic methylthioalkylmalate synthases (MAM1–3), followed by formation of the glucosinolate core structure mediated by cytochrome P450 mono-oxygenases, which catalyse the formation of aldoximes and the subsequent oxidation and conjugation of aldoximes with a sulfur donor to form S-alkyl thiohydroximates. In the case of methionine-derived glucosinolates, the cytochrome P450 mono-oxygenases CYP79F1, CYP79F2 and CYP83A1 are involved in these reactions (Chen et al., 2003; Hemm et al., 2003; Naur et al., 2003; Reintanz et al., 2001; Tantikanjana et al., 2004). The resulting compounds are further metabolized to glucosinolates by consecutive actions of C–S lyases, S-glucosyltransferases and sulfotransferases (Grubb et al., 2004; Klein et al., 2006; Mikkelsen et al., 2004; Piotrowski et al., 2004). Secondary modifications then give rise to the considerable diversity of glucosinolates (Fahey et al., 2001).

Figure 1.

 Proposed pathways for the biosynthesis of methionine-derived (left) and tryptophan-derived glucosinolates (right).

The glucosinolate pattern varies between tissues and organs during ontogenesis (Brown et al., 2003; Petersen et al., 2002), in response to environmental stimuli (Bednarek et al., 2005; Brader et al., 2001; Kliebenstein et al., 2002; Mewis et al., 2005; Mikkelsen et al., 2003) or in different A. thaliana ecotypes (Kliebenstein et al., 2001; Reichelt et al., 2002). The highest glucosinolate concentrations have been found in reproductive organs, including inflorescences, flowers and siliques, followed by leaves and the root system. Indolic glucosinolates are mainly found in mature rosette leaves and roots, whereas aliphatic glucosinolates accumulate ubiquitously in plants, being predominantly present in generative organs.

The control of glucosinolate biosynthesis is poorly understood, and only recently have the first regulatory components of the glucosinolate biosynthetic pathways in A. thaliana been described. A novel calmodulin-binding nuclear protein, IQD1, was shown to be a positive regulator of aliphatic and indolic glucosinolate formation (Levy et al., 2005). In addition, the transcription factor AtDof1.1 was identified as part of a regulatory network controlling CYP83B1 activity and glucosinolate accumulation (Skirycz et al., 2006). The R2R3-MYB transcription factor HIG1/MYB51 has been recently shown to regulate specifically indolic glucosinolate biosynthesis (Gigolashvili et al., in press). Another MYB factor, ATR1/MYB34, has previously been reported to regulate auxin and indolic glucosinolate homeostasis (Celenza et al., 2005); however, ATR1/MYB34 expression appears to be almost absent in leaves, the main site of indolic glucosinolate accumulation, and, moreover, ATR1/MYB34 over-expression led to a high-auxin growth phenotype (Gigolashvili et al., in press).

Remarkably, a specific regulator of aliphatic glucosinolate biosynthesis has not yet been described. Here, we demonstrate that the R2R3-MYB transcription factor HAG1/MYB28 represents a key component in the regulation of aliphatic methionine-derived glucosinolate biosynthesis in A. thaliana.


HAG1/MYB28 activates genes of the aliphatic glucosinolate biosynthetic pathway

A co-transformation assay (Berger et al., in press) was used as a tool to identify new regulatory genes of glucosinolate biosynthetic pathways. Reporter constructs, consisting of promoters of the aliphatic glucosinolate biosynthesis genes MAM3, CYP79F1 and CYP83A1, and the indolic glucosinolate biosynthesis genes TSB1, CYP79B2 and CYP79B3, fused to the uidA (GUS) reporter gene were generated. Likewise, several effector constructs, i.e. putative transcription factors or nuclear-localized proteins (MYB factors, Dof1.1 and IQD1), putatively involved in the control of glucosinolate biosynthetic genes, were cloned into a binary vector under the control of the CaMV 35S promoter. Both reporter and effector constructs were simultaneously expressed in cultured A. thaliana cells using the supervirulent Agrobacterium strain LBA4404.pBBR1MCS.virGN54D (Koroleva et al., 2005), and the trans-activation potential of the effector was subsequently assayed as GUS activity (Berger et al., in press). Alternatively, a simple histochemical staining procedure using X-Gluc was applied.

As shown in Figure 2, cultured A. thaliana cells transiently expressing only the GUS reporter construct fused to promoters of putative target genes without an effector showed only weak GUS activity. However, three transcription factors (effectors), namely MYB28, MYB76 and MYB29, were shown to specifically trans-activate genes of the aliphatic glucosinolate biosynthetic pathway, i.e. MAM3, CYP79F1, and CYP83A1, but not those of the indolic glucosinolate biosynthetic pathway (TSB1, CYP79B2, CYP79B3). These three genes are therefore referred to as high aliphatic glucosinolate (HAG) genes: HAG1/MYB28, HAG2/MYB76 and HAG3/MYB29.

Figure 2.

 Identification of new regulators of methionine-derived glucosinolates in co-transformation assays using cultured A. thaliana cells.
Co-transformation assays for the determination of target gene specificity of various transcription factors (effectors) towards target promoters of aliphatic and indolic biosynthetic pathway genes are shown. The promoters of MAM3, CYP79F1, CYP83A1, TSB1, CYP79B2 and CYP79B3 genes were fused to the uidA (GUS) reporter gene (target promoter:GUS vectors). Cultured A. thaliana Col-0 cells were inoculated with the supervirulent Agrobacterium strain LBA4404.pBBR1MCS.virGN54D containing either only the reporter construct (target promoter:GUS:pGWB3i), or the reporter construct and, in addition, various effector constructs (Pro35S:effector:pGWB2). GUS staining indicates trans-activation of a promoter by an effector. TF, transcription factor.

On the other hand, ATR1/MYB34 and HIG1/MYB51, which have been previously identified as regulators of the indolic glucosinolate biosynthetic pathway (Celenza et al., 2005; Gigolashvili et al., in press), exclusively trans-activated only promoters of genes involved in the indolic but not the aliphatic glucosinolate biosynthetic pathway (Figure 2). The co-transformation assay was validated by demonstrating that MYB12, involved in the control of the phenylpropanoid biosynthetic pathway, could not interact with any promoters of glucosinolate biosynthetic genes but strongly activated promoters of CHS (chalcone synthase) and FLS (flavone synthase) as previously demonstrated (Figure 2) (Berger et al., in press; Mehrtens et al., 2005;). Interestingly, the recently described regulators of glucosinolate biosynthesis, Dof1.1 and IQD1 (Levy et al., 2005; Skirycz et al., 2006), did not activate any tested promoter of glucosinolate biosynthetic genes in trans, and did not change the observed gene activation by HIG1/MYB51 and HAG1/MYB28 in triple assays (not shown).

Properties of the HAG1/MYB28 protein

HAG1/MYB28 (At5g61420) is a member of the large family of R2R3-MYB transcription factors in A. thaliana (Stracke et al., 2001). HAG1/MYB28 clusters into subgroup 12 of the A. thaliana MYB factors together with five other MYB factors (HIG1/MYB51, HIG2/MYB122, ATR1/MYB34, HAG2/MYB76 and HAG3/MYB29), all of which have been identified as regulators of either aliphatic or indolic glucosinolate biosynthetic genes (Celenza et al., 2005; Gigolashvili et al., in press; this work). The HAG1/MYB28 gene structure of three exons and two introns was confirmed by isolation and sequencing of the full-length cDNA. The predicted protein is 336 amino acid residues in length, with a molecular mass of about 41.1 kDa and a calculated isoelectric point of 5.71. HAG1/MYB28 does not contain a typical nuclear localization signal (NLS) as revealed by PredictNLS (; Nair and Rost, 2005). However, an amino acid residue sequence LKKRL was detected (amino acid residues 112–116) that might act as a SV40-type NLS motif. Similar or identical motifs are also present in sequences of other members of the HAG1/MYB28 subfamily (LKKRL and LKKLL in HAG2/MYB76 and HAG3/MYB29, respectively), suggesting that these motifs do actually serve as nuclear localization signals.

HAG1/MYB28 is directed to the nucleus of A. thaliana cells

To determine the subcellular localization of HAG1/MYB28 in plant cells, a co-translational Pro35S:HAG1:GFP fusion plasmid was generated and transiently transferred into a homologous system, i.e. cultured A. thaliana cells.

As shown in Figure 3 (a,b), the nuclei showed a strong green fluorescence signal resulting from expression of the HAG1:GFP fusion protein, which co-localized with DNA-specific DAPI staining (Figure 3c), thus demonstrating the nuclear localization of the HAG1:GFP fusion protein. By contrast, the GFP signal was distributed uniformly in the cytosol when the cytosolic BCAT4 protein (Schuster et al., 2006) was used as a control (Figure 3d).

Figure 3.

 HAG1/MYB28 is localized to the cell nucleus.
Subcellular localization of a Pro35S:HAG1:GFP translational fusion in cultured A. thaliana cells.
(a) The Pro35S:HAG1:GFP construct is localized to the nucleus (bright-field with GFP).
(b) GFP filter only.
(c) DAPI staining of the nucleus.
(d) Cytosolic GFP distribution of the Pro35S:BCAT4:GFP construct (Schuster et al., 2006).

Generation of Pro35S:HAG1-over-expressing and HAG1 RNAi plants

To demonstrate that the accumulation of aliphatic glucosinolates can be triggered by HAG1/MYB28, several transgenic CaMV 35S over-expression and RNAi plants were created using Gateway-compatible vectors and A. tumefaciens-mediated transformation. About 60 primary transformants were analysed for both HAG1/MYB28 over-expression and silencing constructs, and three representative over-expressing and RNAi lines with different steady-state HAG1/MYB28 mRNA levels were analysed in more detail.

Constitutive over-expression of HAG1/MYB28 led either to an almost unaltered growth phenotype (Pro35S:HAG1-12), or plant growth was moderately (Pro35S:HAG1-15) or strongly affected (Pro35S:HAG1-11) (Figure 4). The observed growth phenotypes nicely correlated with the expression levels of HAG1/MYB28 and decreased contents of both I3M (Figure 5d) and auxin (data not shown). HAG1 RNAi plants did not show visible effects on plant morphology (Figure 4).

Figure 4.

 Growth phenotypes of Pro35S:HAG1 and HAG1 RNAi mutants.
Various independent transgenic lines are shown. Upper row, from left to right: Pro35S:HAG1-11, Pro35S:HAG1-15, Pro35S:HAG1-12. Lower row, from left to right: HAG1-RNAi-10, HAG1-RNAi-14 and HAG1-RNAi-12. All transgenic lines are in the Col-0 wild-type background (furthest left).

Figure 5.

 Glucosinolate contents in rosette leaves of 5-week-old HAG1/MYB28 over-expression and RNAi plants.
(a–d) Glucosinolate contents (GS) in the HAG1/MYB28-over-expressing lines Pro35S:HAG1-11, Pro35S:HAG1-15 and Pro35S:HAG1-12. Means and SD, n = 3.
(e–h) Glucosinolate contents (GS) in the HAG1/MYB28 RNAi lines HAG1-RNA1-10, -14 and -12. Means and SD, n = 3.
4MSOB, 4-methylsulfinylbutyl-GS; 3 MSOP, 3-methylsulfinylpropyl-GS; 5MSOP, 5-methylsulfinylpentyl-GS; 4MTB, 4-methylthiobutyl-GS; 8MSOO, 8-methylsulfinyloctyl-GS; I3M, indol-3-ylmethyl-GS; 4MOI3M, 4-methoxyindol-3-ylmethyl-GS.

Metabolite profiling of HAG1/MYB28 over-expressors and RNAi plants

To analyse the levels of aliphatic glucosinolates in transformants with an altered expression of HAG1/MYB28, the glucosinolate pattern of HAG1/MYB28 over-expressors and RNAi plants was determined. Glucosinolates (GS) were extracted from freeze-dried rosette leaves of 5-week-old plants, and the levels of 3-methylsulfinylpropyl-GS (3 MSOP), 4-methylsulfinylbutyl-GS (4MSOB), 5-methylsulfinylpentyl-GS (5MSOP), 4-methylthiobutyl-GS (4MTB), 8-methylsulfinyloctyl-GS (8MSOO), and of the indolic glucosinolates indol-3-ylmethyl-GS (I3M) and 4-methoxyindol-3-ylmethyl-GS (4MOI3M) were determined.

As shown in Figure 5, the HAG1/MYB28 over-expression lines Pro35S:HAG1-11, -15 and -12 contained approximately two- to sevenfold higher levels of the main aliphatic glucosinolate 4MSOB as compared to the wild-type. Also, a two- to three-fold increase in the accumulation of other (minor) short-chain aliphatic glucosinolates such as 3MSOP and 5MSOP was observed, whereas the content of long-chain aliphatic glucosinolates such as 8MSOO remained almost unchanged. Also, the content of 4MTB did not significantly increase or decrease in loss-of-function or gain-of-function mutants, indicating that the levels of methylthioglucosinolates are presumably not regulated by HAG1/MYB28 (Figure 5b,f). The graduation in glucosinolate levels reflects well the corresponding HAG1/MYB28 transcript levels, with Pro35S:HAG1-11 representing the strongest over-expressor line (Figure 5a). Remarkably, the increased accumulation of glucosinolates upon HAG1/MYB28 over-expression was rather specific for methionine-derived glucosinolates. Synthesis of indolic glucosinolates was not influenced in the lines with a moderate or slight increase in the steady-state level of HAG1 (lines Pro35S:HAG1-15 and -12). However, there was a decrease in the content of indolic glucosinolates in line Pro35S:HAG1-11 (Figure 5d) in which the methionine-derived glucosinolate biosynthetic pathway is highly activated.

Analysis of HAG1 RNAi plants showed that these lines contained lesser amounts of both long- and short-chain aliphatic glucosinolates, namely 4MSOB, and also 3MSOP, 5MSOP and 8MSOO (Figure 5e–g). The levels of the indolic glucosinolates I3M and 4MOI3M were notably not affected in the RNAi lines (Figure 5h).

Transcription profiling of HAG1/MYB28 over-expressors and HAG1 RNAi plants

Real-time RT-PCR assays were used to assess whether the steady-state mRNA levels of genes encoding enzymes involved in aliphatic glucosinolate metabolism are affected in HAG1/MYB28 over-expressors and HAG1 RNAi knock-down plants. For this, samples from rosette leaves of wild-type plants, Pro35S:HAG1 plants (Pro35S:HAG1-11, -15 and -12) and RNAi plants (HAG1-RNAi-10, -14 and -12) were analysed for relative mRNA levels. As shown in Figure 6, starting from the chloroplastidic MAM1 and MAM3 (MAML), enzymes involved in the side-chain elongation of methionine, to the last enzymes in the biosynthesis of the aliphatic glucosinolate core structure, the desulfoglucosinolate sulfotransferases AtST5b and AtST5c (Figure 1), HAG1/MYB28-over-expressing lines had consistently higher steady-state mRNA levels of aliphatic glucosinolate biosynthetic genes than the wild-type (Figure 6a).

Figure 6.

 Transcript levels of glucosinolate pathway genes in rosette leaves of 5-week-old HAG1/MYB28 over-expression and RNAi plants.
(a) Real-time RT-PCR analysis of aliphatic glucosinolate pathway genes measured in over-expression lines Pro35S:HAG1-11, Pro35S:HAG1-15 and Pro35S:HAG1-12.
(b) Real-time RT-PCR analysis of aliphatic glucosinolate pathway genes measured in RNAi plants HAG1-RNAi-10, -14 and -12.
Relative gene expression values are given compared to the wild-type (= 1). Values are means and SD (n = 3).

The increase in mRNA transcript levels of putative target genes corresponded nicely with the increased levels of HAG1/MYB28 transcript in the particular over-expression lines, i.e. the strongest over-expression line showing the strongest growth phenotype (Pro35S:HAG1-11) also contained the highest transcript levels of biosynthetic genes. Remarkably, the transcript level of CYP79B2 (and other genes of the indolic glucosinolate biosynthetic pathway including AtST5a, data not shown) remained unchanged in all Pro35S:HAG1 over-expression lines compared to the wild-type (Figure 6a).

In contrast, HAG1/MYB28 RNAi knock-down plants showed a significant decrease in the transcript levels of aliphatic glucosinolate biosynthetic genes in leaves (Figure 6b). The same pattern was also observed in inflorescences (not shown). Again, the reduction in transcript levels corresponded well to the remaining levels of HAG1/MYB28 transcript in the respective RNAi lines, with line HAG1-RNAi-10 being most strongly affected and line HAG1-RNAi-12 showing only a partial decrease in the HAG1/MYB28 transcript level. Notably, the transcript level of CYP79B2, a representative indolic glucosinolate biosynthetic gene, was slightly increased in all RNAi lines.

HAG1/MYB28: tissue-specific expression and induction by mechanical stimuli

To study the tissue-specific expression of HAG1/MYB28, the promoter region (−1995 to +157 bp) was used to generate a translational fusion construct with the uidA (GUS) reporter gene that was stably transferred to A. thaliana plants. About 25 transgenic ProHAG1:uidA lines were analysed in detail, most of them showing a similar tissue-specific pattern of HAG1/MYB28 expression.

The analysis of GUS activity revealed the absence of reporter gene expression in cotyledons, hypocotyls and roots of 3–7-day-old seedlings (Figure 7). HAG1/MYB28 promoter activity first appeared in stems, petioles and the main veins of true leaves in 10- to 14-day-old seedlings. Three-week-old seedlings showed increasing GUS expression, which was also detectable in lateral roots. Reporter gene activity gradually increased in expanding leaves, reaching a maximum in fully expanded leaves of non-flowering plants. However, during transition from the vegetative to the generative stage, GUS expression gradually decreased in leaves and appeared in inflorescences, being maintained during the whole generative stage. During senescence, GUS expression disappeared almost completely in leaves, but was still detectable in inflorescences.

Figure 7.

 Histochemical GUS staining in tissues of ProHAG1:GUS plants.
(a) 14-day-old seedling. (b) Three-week-old plant. (c) Roots of three-to-four-week old plants. (d) Five-week-old plant. (e) Adult leaves. (f) Inflorescences of flowering plant. (g) Roots of adult plants. (h) Leaf of adult plant after transition from vegetative to generative stage. (i) Silique. (j) Leaf of senescent plants. (k) GUS induction at cut sites of inflorescences. (l) GUS induction at touch sites of inflorescences.

Together, the highest HAG1/MYB28 promoter activity could be detected in inflorescences of flowering plants and in adult leaves, the sites of accumulation of aliphatic glucosinolates. These observations are consistent with AtGenExpress data from the Genevestigator microarray database (Zimmermann et al., 2004;

Interestingly, HAG1/MYB28 expression was also induced by touch and wounding (Figure 7k,l) as has been previously reported for IQD1 (Levy et al., 2005). To obtain more information about the time-dependent induction of HAG1/MYB28 by mechanical stimuli, leaves and inflorescences were harvested 1, 5, 15, 60, 90 and 120 min after wounding, and used for real-time PCR analysis. As shown in Figure 8, HAG1/MYB28 expression rapidly increased about threefold after wounding (by 1 min), and decreased within 15 min to below the level of unwounded plants. Similarly, an increased steady-state transcript level of CYP79F1 was observed after approximately 5 min of induction by wounding, returning to its original level after 15 min.

Figure 8.

 Transient expression of HAG1/MYB28 (a) and CYP79F1 (b) in leaves and inflorescences induced by wounding.
The plant material was punctured, and, after 1, 5, 15, 60, 90 and 120 min, used for real-time RT-PCR experiments (means and SD, n = 3). Relative expression values are given compared to non-wounded plants (= 1).

HAG1/MYB28 hinders performance of the generalist Spodoptera exigua

Mechanical stimuli- and herbivory-induced tissue damage linked to myrosinase activity causes glucosinolate degradation and the release of bioactive molecules, which are known to act as repellents against generalist insects (Giamoustaris and Mithen, 1995; Kliebenstein et al., 2002; Lambrix et al., 2001; Levy et al., 2005).

To test whether increased HAG1/MYB28 expression correlates with enhanced resistance to a generalist lepidopteran herbivore, the performance of Spodoptera exigua larvae (Lepidoptera: Noctuidae), which are known to feed on cruciferous and other plant species, was studied. The weight gain of 5-day-old larvae, first kept on an artificial diet, was examined in feeding experiments on both wild-type and HAG1/MYB28-over-expressing plants over 5 days. As shown in Figure 9, larvae developing on HAG1/MYB28 over-expressor plants were significantly smaller and gained about 70% less fresh weight than larvae on wild-type plants within 5 days, indicating that over-expression of HAG1/MYB28 significantly reduces herbivore performance.

Figure 9.

HAG1/MYB28 over-expression reduces insect performance.
Weight gain assays of Spodoptera exigua larvae on 5-week-old A. thaliana plants. Larvae were kept on an artificial diet for 5 days and second-instar larvae were transferred to plants. Values are the mean fresh weight (and SE) of larvae feeding for 5 days on A. thaliana wild-type (Col-0) and Pro35S:HAG1-15 lines. Day 1 (Col-0 and Pro35S:HAG1-15: n = 15 per line; t test, = 0.45); day 3 (Col-0 and Pro35S:HAG1-15: n = 12; test, = 0.042); day 5 (Col-0 and Pro35S:HAG1-15: n = 9; t test, = 0.0016).

HAG1/MYB28 expression is induced by glucose but not by plant hormones

Because glucosinolates are involved in various biotic defense responses, their regulation by multiple signal transduction pathways such as methyl jasmonate (MeJa), salicylic acid (SA) or the ethylene precursor aminocyclopropane carboxylate (ACC) has been suggested (Brader et al., 2001; Cipollini et al., 2004; Devoto and Turner, 2005; Kliebenstein et al., 2002; Mewis et al., 2005; Mikkelsen et al., 2003; Sasaki-Sekimoto et al., 2005). Glucose is another important signaling molecule that may induce transcriptional regulatory mechanisms integrating carbohydrate availability and hormone action. We used ProHAG1:uidA plants to study whether or not HAG1/MYB28 expression responds to these signaling molecules.

Two-week-old seedlings were transferred for 24 h to a sucrose-free medium for adaptation to autotrophic conditions. Subsequently, MeJa, SA, ACC or glucose were added, and plants were analysed for HAG1/MYB28 expression after 4, 12 and 24 h of exposure. As shown in Figure 10, MeJa, SA and ACC did not exert any positive effect on HAG1/MYB28 expression. Rather, exposure to SA even suppressed HAG1/MYB28 expression after 4 h of treatment. On the other hand, HAG1/MYB28 expression was significantly induced by glucose. Thus, glucose seems to be an important signalling molecule inducing HAG1/MYB28 expression, indicating a special mechanism of transcriptional regulation of HAG1/MYB28 in response to carbohydrate availability.

Figure 10.

HAG1/MYB28 expression is induced by glucose.
Elicitor-treated ProHAG1:uidA plants are shown. C, water control; MeJa, 0.45 mM methyl jasmonate; ACC, 10 μM 1-aminocyclopropane carboxylate; SA, 0.2 mM salicylic acid; Glu, 1% glucose. Mannitol (Man, 1%) was used as an osmotic control.


HAG1/MYB28 functions as a regulator of aliphatic glucosinolate biosynthesis and plant defense against herbivores

The biosynthetic pathways leading to the great diversity of glucosinolates have been elucidated in the past years, including identification of almost the whole set of corresponding biosynthetic genes (Grubb and Abel, 2006; Halkier and Gershenzon, 2006). However, the first regulatory factors directly or indirectly activating glucosinolate pathway genes have only recently been described, e.g. IQD1, ATR1/MYB34, AtDof1.1 and HIG1/MYB51 (Levy et al., 2005; Celenza et al., 2005; Skirycz et al., 2006; Gigolashvili et al., in press). HIG1/MYB51 was shown to specifically regulate indolic glucosinolate biosynthesis (Gigolashvili et al., in press), whereas the other genes are either involved in the control of both aliphatic and indolic glucosinolates or activate genes of the indolic glucosinolate pathway along with the biosynthesis of auxin, mainly because the indolic glucosinolate and indole-3-acetic acid (IAA, auxin) biosynthetic pathways share indole-3-acetaldoxime as a common intermediate. Nothing was known about the regulation of the aliphatic methionine-derived glucosinolates that play important roles in the plant defense response against herbivores and pathogens.

Here, we show that HAG1/MYB28 is involved in the specific regulation of aliphatic glucosinolates. Several lines of evidence indicate that HAG1/MYB28 functions as a key regulator of aliphatic glucosinolate biosynthesis and plant defense against herbivores.

First, an increase and/or decrease in glucosinolate accumulation nicely corresponded to the level of HAG1/MYB28 transcripts in both the various over-expression lines as well as the RNAi lines (Figures 5 and 6). HAG1/MYB28-over-expressing lines contained up to sevenfold higher levels of short-chain aliphatic glucosinolates, especially of the major aliphatic glucosinolate, 4MSOB. These effects were restricted to aliphatic methylsulfinylglucosinolates; the levels of methylthioglucosinolates and indolic glucosinolates remained almost unaffected in plants with moderate expression of HAG1/MYB28 (Figure 5). However, lines strongly over-expressing HAG1/MYB28 and containing high levels of methionine-derived glucosinolates (Pro35S:HAG1-11, Figure 5d) showed a decrease in the content of the main indolic glucosinolate I3M. This observation may result from metabolic cross-talk between the different branches of glucosinolate biosynthesis. Alternatively, and as proposed by Grubb and Abel (2006), the competition of cytochrome P450 mono-oxygenases involved in either the aliphatic or indolic glucosinolate biosynthetic pathways for electrons could be the reason for reciprocal negative feedback regulation between both branches of glucosinolate biosynthesis. Interestingly, the CYP79F1 null mutant (bus1-1f) that completely lacks short-chain aliphatic glucosinolates accumulated increased levels of indolic glucosinolates (Chen et al., 2003; Reintanz et al., 2001). These observations provide additional evidence for an interplay between both biosynthetic pathways.

The levels of indolic glucosinolates in HAG1 RNAi plants were also unchanged, but those of aliphatic glucosinolates were significantly reduced. In contrast to mutants detective in CYP79F1 or CYP79F2 functions (Tantikanjana et al., 2004), HAG1 RNAi plants did not completely lack aliphatic glucosinolates. This may be due to either the only partial knock-down of HAG1/MYB28 in HAG1 RNAi plants and/or complementation of HAG1/MYB28 function by other transcription factors with similar roles, e.g. HAG2/MYB76 or HAG3/MYB29. However, both HAG2/MYB76 and HAG3/MYB29 were apparently unable to fully compensate for the reduced HAG1/MYB28 activity in HAG1 RNAi lines. A number of scenarios may be considered to explain this observation. The three transcription factors may have distinct sites of expression or could functionally act only in concert with other (transcription) factors and may thus differ in their trans-activation potential toward glucosinolate biosynthetic genes. As revealed by the Genevestigator database (Zimmermann et al., 2004), HAG2/MYB76 shows a similar expression pattern as HAG1/MYB28, and should, in the case of functional redundancy, be able to complement the deficiency in HAG1/MYB28 activity. However, and in contrast to HAG1/MYB28, HAG2/MYB76 does not appear to be co-regulated with glucosinolate biosynthetic genes such as BCAT4, MAM1, CYP79F1 and CYP83A1 (Toufighi et al., 2005, It may be functional as an enhancer of the expression of aliphatic glucosinolate biosynthetic genes, but only in concert with other factors, e.g. HAG1/MYB28. Future work will focus on understanding of the regulatory mechanism underlying the spatial and temporal trans-activation potential of HAG2/MYB76 and HAG3/MYB29 in comparison with that of HAG1/MYB28.

Second, using co-transformation assays, it could be demonstrated that HAG1/MYB28 specifically activates genes of the aliphatic glucosinolate biosynthetic pathway, i.e. MAM1, MAM3, CYP79F1, CYP79F2, CYP83A1, AtST5b and AtST5c, without affecting genes involved in the indolic glucosinolate pathway, e.g. TSB1, CYP79B2, CYP79B3 (Figures 2 and 6). Thus, this list of target genes encompasses genes involved in the first reactions of aliphatic glucosinolate biosynthesis up to the very last steps (Figure 1). It may be noted that the expression patterns of these genes overlap with those of HAG1/MYB28 (; Toufighi et al., 2005). HAG1 RNAi plants, on the other hand, contained lower transcript levels of aliphatic glucosinolate pathway genes. Remarkably, AtDof1.1 and IQD1 did not possess any trans-activation potential towards promoters of aliphatic and indolic glucosinolate biosynthetic genes, indicating that these proteins are not able to directly activate glucosinolate biosynthetic genes, i.e. that additional factors are required to exert activation of glucosinolate biosynthesis. It may be noted that the recently discovered regulators of the indolic glucosinolate pathway, ATR1/MYB34 and HIG1/MYB51 (Celenza et al., 2005; Gigolashvili et al., in press), specifically activate indolic but not aliphatic glucosinolate biosynthetic genes (Figure 2).

Third, HAG1/MYB28 expression overlaps well with the expression of aliphatic glucosinolate biosynthetic genes (Figure 7). For example, BCAT4 and CYP79F1 show a strong expression in the vasculature of young and mature rosette leaves and in stems (Chen et al., 2003; Reintanz et al., 2001; Schuster et al., 2006). This observation is consistent with the role of HAG1/MYB28 as a positive regulator of this pathway. However, expression of glucosinolate biosynthesis genes often appeared to be present in non-photosynthetic organs such as flowers and seeds, in which HAG1/MYB28 is not expressed. To provide reproductive organs with glucosinolates, long-distance transport of glucosinolates from photosynthetic tissues to siliques is envisaged (Brudenell et al., 1999; Chen and Andreasson, 2001). Remarkably, the expression of HAG1/MYB28 overlaps not only with the expression of glucosinolate biosynthetic genes, but also covers the main sites of methionine-derived glucosinolate accumulation. Young rosette leaves exhibit a high concentration of aliphatic glucosinolates, which declines during maturation and flowering, resulting in a predominance of indolic glucosinolates in rosette leaves (Brown et al., 2003). This distribution correlates well with the switch in HAG1/MYB28 expression during transition from the vegetative to the generative phase of development. Starting with flowering, aliphatic glucosinolates are found mainly in generative organs, which are also the main sites of HAG1/MYB28 expression.

Finally, HAG1/MYB28 expression is very rapidly induced by mechanical stimuli, resulting in increased levels of aliphatic glucosinolates (Figure 8). Moreover, weight gain assays using larvae of the generalist herbivore Spodoptera exigua showed that larvae fed with HAG1/MYB28-over-expressing plants gained markedly less weight as compared to larvae fed with wild-type plants (Figure 9). Thus, over-expression of HAG1/MY28 reduces herbivore performance.

Pro 35S :HAG1 plants exhibited, dependent on the strength of over-expression, retardations in growth, whereas HAG1 RNAi plants with lowered levels of aliphatic glucosinolates have a wild-type appearance (Figure 4). We occasionally observed that some HAG1/MYB28 over-expression lines with extremely increased levels of aliphatic glucosinolates and decreased levels of I3M and auxin exhibited a drastic retardation in growth. These plants were unable to produce seeds and could not be further analysed.

HAG1/MYB28 expression is induced by glucose but not by MeJa, SA and ACC

Environmental stimuli such as herbivore attack, wounding or hormone treatment (e.g. MeJa, SA, ACC) are known to have an impact on glucosinolate regulation, and several glucosinolate biosynthesis pathway genes have been shown to be induced by these treatments (Brader et al., 2001; Kliebenstein et al., 2002; Mewis et al., 2005; Mikkelsen et al., 2003). Although HAG1/MYB28 expression clearly responds to mechanical stimuli, as has also been shown for IQD1, BCAT4 and MAM1 (Levy et al., 2005; Schuster et al., 2006), none of the tested plant hormones were able to induce HAG1/MYB28 expression (Figure 10), indicating that HAG1/MYB28 may function independently of these hormones. However, HAG1/MYB28 expression clearly responded to sugars, e.g. glucose, suggesting that HAG1/MYB28 is a component of regulatory mechanisms integrating signals for carbohydrate availability (Li et al., 2006;Zimmermann et al., 2004). Glucose is released during myrosinase action on glucosinolates, thus possibly generating a signal for HAG1/MYB28 induction. It has been generally shown that biotic stress, wounding or infection by viruses, bacteria and fungi can modulate the carbohydrate status of plant tissues. An extracellular invertase appears to play a prominent role in carbon partitioning and to integrate sugar, stress and hormone signals (Roitsch, 1999; Rolland et al., 2002; Sonnewald et al., 1991; Sturm, 1999). As shown previously, sugars can regulate expression of the wound-inducible proteinase inhibitor II, lipoxygenase genes (Johnson and Ryan, 1990;Sadka et al., 1994) and pathogenesis-related genes (Herbers et al., 1996; Xiao et al., 2000). Remarkably, analysis of microarray data for A. thaliana seedlings led to the identification of HAG1/MYB28 as one of the very rapidly up-regulated genes in response to glucose treatment (Li et al., 2006). In accordance with this observation, the HAG1/MYB28 promoter contains several glucose regulated motifs (e.g. AAACCCTAA, GTTAGGTT, RCCGAC), as listed by Li et al. (2006).

HAG1/MYB28 is obviously a key player in a signaling pathway connecting external signals such as mechanical stimuli to glucose availability and further to the activation of transcriptional machinery for the production of aliphatic glucosinolates, promoting sustained plant resistance against herbivores.

Experimental procedures

Plant materials and growth conditions

Plants (Arabidopsis thaliana ecotype Columbia) and corresponding gain-of-function and loss-of-function mutants were grown in a temperature-controlled greenhouse under a 16 h light/8 h dark regime or in a growth chamber under a 8 h light/16 h dark regime at day/night temperatures of 21°C/18°C and at 40% humidity. For selection of kanamycin-resistant plants, seeds were surface-sterilized and plated on agar-solidified half-strength Murashige and Skoog (MS) medium containing 2.15 g/l MS salts, 0.5% sucrose, pH 5.8. Plates were grown in a culture chamber (16 h light periods, 75% humidity, 21°C).

Arabidopsis thaliana cell culture

An A. thaliana Col-0 suspension-cultured cell line generated from roots was maintained in 50 ml of A. thaliana (AT) medium (Mathur et al., 1998). The AT medium contained 4.3 g/l MS basal salt medium (Duchefa;, 1 mg/l 2,4-dichlorophenoxyacetic acid (2,4-D), 4 ml vitamin B5 mixture (Sigma; and 30 g/l sucrose (pH 5.8). Cells were gently agitated at 160 rpm in the dark at 22°C.

Co-transformation assays

To generate reporter constructs, the promoter regions of the aliphatic glucosinolate biosynthetic genes MAM1 (−1739 to +270 bp), CYP79F1 (−1388 to +45 bp) and CYP83A1 (−919 to +63 bp), and of indolic glucosinolate biosynthesis genes TSB1 (−2320 to +132), CYP79B2 (−1383 to +81 bp) and CYP79B3 (−1353 to +84 bp) were amplified from genomic DNA of A. thaliana plants and cloned into the pDONOR-207 vector (Invitrogen Life Technologies; To drive Agrobacterium-mediated expression of the uidA (GUS) gene under the control of these promoters, the binary plant transformation vector pGWB3i containing an intron within the GUS gene was generated. A 189 bp intron fragment was amplified from pPCV 6NFHyg GUS Int vector (Vancanneyt et al., 1990) using a proof-reading polymerase. The blunt-ended PCR product was ligated to the pGWB3 vector using the SnaBI restriction site. The pGWB3i vector was recombined with the pDONOR-207 vector containing corresponding promoter sequences using LR reactions (Invitrogen Life Technologies). Gateway entry clones for creating reporter constructs for CHS and FLS and for the generation of an effector construct for MYB12 were kindly provided by R. Stracke (University of Bielefeld, Germany).

To generate effector constructs, the coding sequences of HAG1/MYB28, HAG2/MYB76, HAG3/MYB76, HIG1/MYB51 and ATR1/MYB34 were amplified by RT-PCR and cloned into the pDONOR207 vector. To recombine the insert from the entry clone into the destination vector, LR reactions betweenpDONOR207 and pGWB2 clones were performed. The reporter and effector constructs were used to transform the supervirulent Agrobacterium strain LBA4404.pBBR1MCS.virGN54D.

For transient expression assays, Agrobacteria containing effector constructs, the anti-silencing 19 K protein (Voinnet et al., 1999) and each of the reporter constructs were taken from fresh YEB plates, grown overnight, resuspended in 1 ml of A. thaliana (AT) medium and used for the transformation of dark-grown cultured A. thaliana Col-0 cells generated from roots (Mathur et al., 1998). To estimate the trans-activation potential, three clones of Agrobacterium, containing effector, reporter and anti-silencing constructs (19 K) were mixed in a 1:1:1 ratio, and 75 μl of this suspension was added to 3 ml of cultured A. thaliana cells, grown for 3–5 days in the dark and stained afterwards with X-Gluc. GUS activity could be analysed earlier, but reliable results were obtained at later time points.

Generation of transgenic HAG1/MYB28 over-expression and RNAi plants

To generate HAG1/MYB28 over-expression plants, the Pro35S:HAG1 construct, also used in trans-activation assays, was transformed into A. tumefaciens (strain GV3101) by electroporation and further into A. thaliana plants by vacuum infiltration.

Constructs for silencing of the HAG1/MYB28 gene were generated using a Gateway-based pJawohl17 vector kindly provided by I. Somssich (Max Planck Institute for Plant Breeding Research, Cologne, Germany). First, a suitable target region for silencing of the HAG1/MYB28 gene was selected. Two complementary 5′→3′ and 3′→5′HAG1/MYB28 fragments were amplified from a wild-type cDNA using Gateway attB1- and attB2-extended primers (attB1-TTAATGGCTTCACTGAGCAGATTC; attB2-TGATGAGACTTCTTGGGAAACATC) and cloned into the pDONOR-207 vector. To recombine inserts from the entry clone into the pJawohl17 destination vector supplied with two recombination cassettes, LR reactions between two pDONR-207 clones and the pJawohl17 destination clones were performed. The entire HAG1-pJawohl17 silencing construct was transformed into the Gateway-compatible A. tumefaciens strain GV3101 by electroporation and into A. thaliana plants by vacuum infiltration. All transformants were selected with kanamycin, and the HAG1/MYB28 transcript level was monitored by quantitative real-time PCR analysis.

Construction of the Pro35S:HAG1:GFP fusion plasmid and transformation of cultured A. thaliana cells

To generate an expression clone with the Pro35S:HAG1:GFP fusion, the pDONOR-207 vector containing the HAG1/MYB28 coding sequence was recombined with the pGWB5 destination vector using an LR reaction. Transformation of dark-grown cultured A. thaliana cells was performed using the supervirulent Agrobacterium strain LBA4404.pBBR1MCS.virGN54D (kindly provided by Dr Memelink, University of Leiden, The Netherlands) as described by Koroleva et al. (2005). GFP expression patterns and DAPI (4,-6-diamidino-2-phenylindole) staining of the nucleus were recorded using a fluorescence microscope (Eclipse E800, Nikon;

Histochemical analysis of transgenic plants expressing the ProHAG1:uidA fusion construct

The HAG1/MYB28 promoter region (−1995 to +157 bp) was amplified from genomic DNA of A. thaliana plants and cloned into the pDONOR-207 vector. To drive expression of the uidA gene under the control of the HAG1/MYB28 promoter, the binary Gateway-compatible plant transformation vector pGWB3 was recombined with the HAG1/MYB28 entry clone using an LR reaction. The ProHAG1:uidA clone in pGWB3 was transformed into the GV3101 Agrobacterium strain and A. thaliana wild-type plants.

Histochemical localization of GUS in several independent transgenic lines harboring the ProHAG1:uidA construct was performed as described by Jefferson et al. (1987) with some modifications. Sample tissues were infiltrated with the reaction buffer [50 mm Na2HPO4-NaH2PO4, pH 7.0, 0.5 mm K3Fe(CN)6, 0.5 mm K4Fe(CN)6, containing 2 mm 5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid (X-Gluc) as substrate] under vacuum, and incubated at 37°C overnight. Plant pigments were destained with 80% ethanol, and the GUS staining patterns were recorded under a binocular microscope (SMZ-U, Nikon).

Preparation of methanolic extracts and HPLC analysis of glucosinolates

Leaves (50–100 mg) were placed into a 2 ml reaction tube and frozen in liquid nitrogen. Frozen leaf samples were lyophilized and homogenized in an MM301 mill (Retsch; Glucosinolates were extracted in 80% methanol after addition of 20 μl of 5 mm benzyl glucosinolate as an internal standard ( Extracts were applied onto a DEAE Sephadex A25-column (0.1 g powder equilibrated in 0.5 m acetic acid/NaOH, pH 5). Glucosinolates were converted to desulfoglucosinolates by overnight incubation with a purified sulfatase (E.C. designated ‘type H-1, from Helix pomatia, 16 400 units/g solid’ (Sigma). For analysis of desulfoglucosinolates, samples were subjected to HPLC analysis on a Hewlett-Packard 1100 series chromatograph with a quaternary pump and a 1040M diode array detector and using a Supelco C-18 column (Supelcosil LC-18, 250 × 4.6 mm, 5 μm internal diameter; Elution was accomplished with a gradient of 0–5% solvent B (10 min), 5–38% solvent B (24 min), followed by a cleaning cycle (38–100% solvent B in 4 min, 6 min hold, 100 to 0% solvent B in 5 min, 7 min hold) (solvent A, water; solvent B, methanol). Peaks were quantified by the peak area at 229 nm (bandwidth 4 nm) relative to the area of the internal standard peak, applying the response factors as used by Brown et al. (2003). The identities of the desulfo derivatives of glucosinolates were confirmed by LC-ESI-MS as described previously (van Dam et al., 2003; Müller and Martens, 2005). Three different technical replicates and two independent sets of plants were used for the analyses.

Reverse transcriptase-mediated first-strand synthesis and real-time RT-PCR analysis

Total RNA was extracted from rosette leaves of adult plants from different mutant lines using TRIsure buffer (Biolone; followed by treatment with RNase-free DNase (Roth; to remove genomic DNA contaminations. Total RNA (7–10 μg) was reverse-transcribed using the First-Strand cDNA Synthesis SSII Kit (Invitrogen Life Technologies) according to the manufacturer’s instructions.

The expression of glucosinolate biosynthesis genes was analysed by real-time quantitative RT-PCR analysis using the fluorescent intercalating dye SYBR Green in a GeneAmp® 5700 sequence detection system (Applied Biosystems; The Arabidopsis ACTIN2 gene was used as a standard. A two-step RT-PCR analysis was performed. First, total RNAs (10 μg per reaction) were reverse-transcribed into cDNAs, using the First-Strand cDNA Synthesis SSII Kit according to the manufacturer’s instructions. Subsequently, the cDNAs were used as a template in real-time PCR experiments with gene-specific primers (for primer sequences, see Table 1). Real-time PCR was performed using the SYBR Green master kit system (Applied Biosystems) according to the manufacturer’s instructions. The Ct, defined as the PCR cycle at which a statistically significant increase of reporter fluorescence is detected, is used as a measure of the starting copy number of the target gene. Relative quantification of expression levels was performed using the comparative Ct method (manufacturer’s instructions, bulletin 2, Applied Biosystems). The relative value for the expression level of each gene was calculated by the equation Y = 2-ΔΔCt, where ΔCt is the difference between control and target products (ΔCt = Ct(GENE) − Ct(ACT)), and ΔΔCt = ΔCt(mutant) − ΔCt(wt). Thus, the calculated relative expression values are normalized to the wild-type expression level (wild-type = 1). The efficiency of each primer pair was tested using wild-type (Col-0) cDNA as a standard template, and the RT-PCR data were normalized according to the relative efficiency of each primer pair. Three different technical replicates and two independent sets of plants were used for the analyses.

Table 1.   Primers used in real-time RT-PCR experiments
Primer namePrimer sequence (5→3)

Elicitor treatments

Several independent transgenic plants harboring the ProHAG1:GUS construct were grown for 7 days on a half-strength Murashige and Skoog agar plates containing 0.5% of sucrose in a 22–24°C growth chamber with a 16 h light/8 h dark cycle. Before elicitor induction, plants were adapted to autotrophic growth conditions in sucrose-free liquid MS culture. For induction, methyl jasmonate (MeJA, 0.45 mm), 1-aminocyclopropane carboxylate (ACC, 10 μm), salicylic acid (SA, 0.2 mm), glucose (1% w/v) or mannitol (1% w/v) were added. After 4, 12 and 24 h, the plants were histochemically analysed for GUS activity.

Weight gain assays with the generalist herbivore Spodoptera exigua

Eggs of the lepidopteran herbivore, Spodoptera exigua (Lepidoptera: Noctuidae), were obtained from Bayer CropScience (Monheim, Germany), and larvae were kept on an artificial diet for 5 days. Second-instar larvae (15 per line) were taken and transferred to 5-week-old plants of either wild-type or HAG1/MYB28-over-expressing plants that had been grown in soil under short-day conditions (8 h light/16 h dark). Larvae were kept on plants at 27°C and a 12 h light/12 h dark cycle. After 1, 3 and 5 days of feeding, the fresh weights of larvae were individually determined. Student’s t-tests were performed to compare larval weights on both plant lines.


This work was supported by the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie. We thank Kerstin Kunze for excellent technical assistance.