The transcription factor EMISSION OF BENZENOIDS II activates the MYB ODORANT1 promoter at a MYB binding site specific for fragrant petunias


(fax +31 20 525 7934; e-mail


Fragrance production in petunia flowers is highly regulated. Two transcription factors, ODORANT1 (ODO1) and EMISSION OF BENZENOIDS II (EOBII) have recently been identified as regulators of the volatile benzenoid/phenylpropanoid pathway in petals. Unlike the non-fragrant Petunia hybrida cultivar R27, the fragrant cultivar Mitchell highly expresses ODO1. Using stable reporter lines, we identified the 1.2-kbp ODO1 promoter from Mitchell that is sufficient for tissue-specific, developmental and rhythmic expression. This promoter fragment can be activated in non-fragrant R27 petals, indicating that the set of trans-acting factors driving ODO1 expression is conserved in these two petunias. Conversely, the 1.2-kbp ODO1 promoter of R27 is much less active in Mitchell petals. Transient transformation of 5′ deletion and chimeric Mitchell and R27 ODO1 promoter reporter constructs in petunia petals identified an enhancer region, which is specific for the fragrant Mitchell cultivar and contains a putative MYB binding site (MBS). Mutations in the MBS of the Mitchell promoter decreased overall promoter activity by 50%, highlighting the importance of the enhancer region. We show that EOBII binds and activates the ODO1 promoter via this MBS, establishing a molecular link between these two regulators of floral fragrance biosynthesis in petunia.


Tight regulation of metabolic pathways is necessary to manage the large number of compounds present in the cell, and to ensure that substrates, which are often used by competing pathways, are used at the appropriate time. To a large extent this occurs through the coordinate expression of the genes in the pathway, which involves the action of transcription factors binding to cis-elements in the promoter of these genes, thereby activating or inhibiting their expression, often in a combinatorial manner (Singh, 1998; Hartmann et al., 2005).

For many pathways, such as the flavonoid and lignin pathways (Winkel-Shirley, 2001; Boerjan et al., 2003), several transcription factors that regulate biosynthetic genes or genes encoding other transcription factors have been identified (Winkel-Shirley, 2001; Boerjan et al., 2003; Hartmann et al., 2005; Koes et al., 2005; Zhong and Ye, 2009). Regulation of the volatile benzenoid/phenylpropanoid pathway in plants, however, is less well understood (Schuurink et al., 2006), despite the fact that this pathway produces important compounds in plant secondary metabolism (Wildermuth, 2006; Van Moerkercke et al., 2009). In petunia flowers, both the flavonoid (Koes et al., 2005) and volatile benzenoid/phenylpropanoid pathways are active, and competition for the common precursor Phe is avoided because both pathways are active at different developmental stages of the flower (Verdonk et al., 2005). Transcription of genes in the volatile benzenoid/phenylpropanoid pathway starts when the flower opens and peaks after anthesis. Additionally, the expression of volatile benzenoid/phenylpropanoid genes is regulated rhythmically (Colquhoun et al., 2010b). Together with the tissue-specific expression pattern, this means that the volatile benzenoid/phenylpropanoid pathway is highly regulated, both temporally and spatially.

Few transcription factors (TFs) regulating the volatile benzenoid/phenylpropanoid pathway have been identified, and little of their target genes are known. The petunia R2R3-MYB TF ODORANT 1 (ODO1) was shown to activate the promoter of 5-enol-pyruvylshikimate-3-phosphate synthase (EPSPS), which regulates precursor availability via the shikimate pathway (Verdonk et al., 2005). Silencing of ODO1 (ir-ODO1) resulted in a severe decrease of volatile production in petunia flowers, but did not affect the production of Phe-derived flavonols and anthocyanins (Verdonk et al., 2005), which are produced earlier in flower development. Two genes, benzoic acid/salicylic acid methyl transferase (BSMT) and benzoyl-CoA:benzylalcohol/phenylethylalcohol benzoyltransferase (BPBT), producing the volatile compounds methylbenzoate/methylsalicylate and benzylbenzoate/phenylethylbenzoate, respectively, were upregulated in ir-ODO1 petunia flowers, whereas several genes including phenylalanine ammonia lyase 1 and 2 (PAL1/2) and chorismate mutase (CM) were downregulated. However, direct activation of their promoters by ODO1 was not investigated (Verdonk et al., 2005). Recently, a second petunia R2R3-MYB TF, EMISSION OF BENZENOIDS II (EOBII), was shown to regulate the volatile benzenoid/phenylpropanoid pathway in petunia. This homolog of Antirrhinum majus MYB305 influences transcript abundance of several biosynthetic genes in the pathway (Spitzer-Rimon et al., 2010), whereas direct activation of the petunia isoeugenol synthase (IGS) and Nicotiana tabaccum (tobacco) phenylalanine ammonia lyase B (PALB) promoters was shown in Arabidopsis leaf protoplasts (Spitzer-Rimon et al., 2010). Interestingly, ODO1 transcript levels were lower in petals in which EOBII transcript levels were reduced via virus-induced gene silencing (VIGS), but overexpression of EOBII did not result in higher ODO1 mRNA levels. This led to the conclusion that other factors must be needed for ODO1 activation in planta (Spitzer-Rimon et al., 2010). Finally, the petunia MYB4, a homolog of the Arabidopsis MYB4, was identified as a repressor of cinnamate-4-hydroxylase (C4H), thereby fine-tuning the production of p-coumaric acid-derived volatile compounds, such as isoeugenol and eugenol (Colquhoun et al., 2010a).

Although many target genes of R2R3-MYB TFs involved in developmental processes, (a)biotic stress responses, and primary and secondary metabolism are known in plants, few MYB binding sites have been identified (Dubos et al., 2010). Binding studies using several promoters of phenylpropanoid genes resulted in the identification of an MYB consensus binding motif NMACCTWAAM for A. majus MYB305 (Sablowski et al., 1994). Recently, another homolog of AmMYB305, the Nicotiana langsdorffii × Nicotiana sanderae MYB305 was shown to bind an MYB binding site (TCACCTAAT) in the promoter of the nectarin1 (nec1) gene (Liu et al., 2009). Regulation of lignin genes by an R2R3 MYB TF was shown in Zea mays (Fornaléet al., 2006), and the binding site for ZmMYB31, ACCT/AACC, was inferred using gel-shift and SELEX experiments (Fornaléet al., 2010). However, only few TFs that regulate the expression of plant R2R3-MYB TFs have been identified (Suzuki et al., 1997; Koshino-Kimura et al., 2005; Shangguan et al., 2008; Espley et al., 2009; Kang et al., 2009; Zhang et al., 2009).

As ODO1 is highly transcribed in fragrant petunia cultivars, but barely in a non-fragrant cultivar, and because suppression of ODO1 expression dramatically reduces volatile production (Verdonk et al., 2005), it appears there is a correlation between volatile production and high ODO1 expression. This would implicate that transcriptional regulation of ODO1 is an essential factor determining volatile emission. Low ODO1 transcript abundance can result from: (i) the absence of positive regulatory promoter elements; (ii) the presence of negative regulatory elements in the promoter region or elsewhere in the gene; (iii) the lack of an activating TF(s); (iv) the presence of a repressor(s); or (v) any combination of (i)–(iv), in addition to chromatin structure and post-transcriptional regulation (Moore, 2005). To investigate the role of the ODO1 promoter in the regulation of its transcript abundance, we performed a detailed analysis of the petunia ODO1 promoter from fragrant and non-fragrant petunias to identify cis-elements. In addition we identified EOBII as one of the trans-factors, responsible for ODO1 expression.


Regulation of ODO1 expression in petunia flowers occurs to a great extent at the promoter level

ODO1 transcripts accumulate predominantly in the petals of fully developed flowers (Verdonk et al., 2005). To investigate to what extent the ODO1 promoter contributes to this tissue specificity, we made stable reporter lines with the Petunia hybrida cv. Mitchell ODO1 promoter. A 1880-bp promoter fragment was isolated and a transcriptional fusion was made with the β-glucuronidase (uidA) coding sequence (CDS) (M19:GUS; Figure 1). Following stable introduction in Mitchell plants, seven independent primary transformants were obtained, six of which showed detectable GUS levels in their petals (Figure S1). Histochemical analysis of GUS activity in these lines showed petal-specific activation of the ODO1 promoter (Figure S2). No activity was seen in sepals, flower petioles (Figure S2), filament stigma or stem (data not shown). Staining was observed in anthers, but to a similar extent as in wild-type plants (data not shown). Three independent lines were chosen for detailed quantitative analyses of GUS activity. In these lines, leaves had on average only 0.5% activity of that of petal limbs. The activity in petal tubes was on average 17% of that in petal limbs (Figure 2a). Depending on the line, activity in petals of a 3-cm-long bud was 0.6–8%, and in a 4-cm-long bud was 3–24%, of that of an open petal (Figure 2b). We also generated stable lines with a 1207-bp promoter fragment fused to uidA (M12:GUS; Figure 1). Three independent primary transformants were obtained, and although the absolute GUS activity in petal limbs differed between these three lines (Figure S1), all showed a similar developmental and tissue-specific pattern as the M19:GUS lines (Figure 2a,b).

Figure 1.

 Reporter constructs used in this study.
The ODO1 promoters of three Petunia hybrida cultivars were used to create reporter constructs: M, P. hybrida cv. Mitchell; R, P. hybrida cv. R27; and V, P. hybrida cv. V26. Mutations in the M12 promoter are denoted by m1 and m2. The bar on top indicates the length of the promoter fragment relative to the ATG codon. The location of the primers used to isolate each fragment are shown on top of each fragment. F, forward primer; R, reverse primer.

Figure 2.

 The 1.2-kbp ODO1 promoter fragment directs tissue-specific, developmental and rhythmic expression.
(a, b) Normalized GUS activities in leaves, petals and petal tubes (a), and in petals at different developmental stages (b), of M19:GUS and M12:GUS T0 lines. Values are represented relative to those of a stable CaMV 35S:GUS line (n = 3, average ± SE).
(c) RT-PCR showing GUS transcript levels in M12:GUS plants during a day/night cycle. The bar indicates the artificial day/night period. Three independent T0 lines showed a similar pattern and a representative gel is shown. FBP1 (Floral Binding Protein 1) was used as a reference gene. The numbers on the left indicate the number of PCR cycles.
(d) Quantitative RT-PCR showing rhythmic expression of GUS driven by the M12 promoter and endogenous ODO1 transcript levels in two T0 independent M12:GUS lines. The bar indicates the artificial day/night period. ACTIN was used as a reference gene.

ODO1 transcripts accumulate rhythmically, peaking 2–3 h before the onset of the dark period (Verdonk et al., 2005). We therefore assessed our M12:GUS reporter lines for rhythmic expression from the 1207-bp ODO1 promoter. As the GUS protein is relatively stable and therefore not suitable to investigate transient activity of a promoter, we assessed GUS mRNA accumulation at different time points during the day/night cycle. Semiquantitative RT-PCR on petal RNA isolated every 4 h during a 24-h period showed that GUS transcripts driven by the ODO1 promoter accumulate before the onset of the dark period, and decline during the course of the dark period (Figure 2c), thus showing that the 1207-bp promoter is capable of driving rhythmic expression. We additionally used quantitative RT-PCR on cDNA from M12:GUS T0 lines to confirm the RT-PCR data, and to compare GUS expression from the ODO1 promoter with the endogenous ODO1 transcript levels. Figure 2(d) shows a pattern of gene expression of GUS that is similar to that of endogenous ODO1. With regard to tissue specificity, developmental regulation and rhythmic expression of ODO1, we conclude that the necessary cis-elements must be contained within this 1207-bp promoter of cv. Mitchell.

Whereas the fragrant cv. Mitchell highly expresses ODO1, flowers of the non-fragrant cv. R27 do not (Figure 3a). To examine whether cis-elements or trans-factors cause this difference, we analysed the activity of the 1.2-kbp ODO1 promoters of both cultivars reciprocally in their petals. To this end, we additionally made a transcriptional fusion of the 1.2-kbp promoter of R27 with the uidA reporter (R12:GUS; Figure 1), and assessed GUS activity driven by both promoters in petals after agroinfiltration. Figure 3(b) shows that in the fragrant Mitchell petals the R27 promoter has only 20% activity of that of the Mitchell promoter. To determine whether the ODO1 promoters are active in R27, we assessed both reporter constructs in petals of this cultivar. GUS activity was detected using the Mitchell promoter, which showed fourfold higher activity compared with the R27 promoter in R27 petals (Figure 3c). In conclusion, these experiments suggest that the difference in ODO1 expression levels between the Mitchell and R27 petals is in part caused by differences in the promoter, rather than the lack of upstream activators.

Figure 3.

ODO1 expression and promoter activity in fragrant and non-fragrant petunia cultivars.
(a) RT-PCR was performed using ODO1-specific and FBP1-specific primers. FBP1 (Floral Binding Protein 1) was used as a reference gene. RNA was isolated from petals of Petunia hybrida cv. Mitchell and cv. R27, 2 h before the onset of the dark period. The numbers to the left indicate the number of PCR cycles.
(b, c) Analysis of the 1.2-kbp ODO1-promoter activity using transcriptional uidA (GUS) fusions in petals of Mitchell (b) and R27 (c) by means of Agrobacterium-mediated transient transformation. Flowers were co-infiltrated with Agrobacterium tumefaciens harbouring a CaMV 35S-driven luciferase (35S:LUC) construct to enable normalisation (average ± SE; n = 5). The experiment was repeated twice and the results of a representative experiment are shown. The normalized values of the Mitchell promoter were set to 100.

The ODO1 promoter is polymorphic between different P. hybrida cultivars

Because differences in ODO1 transcript levels between cultivars depend on the promoter strength, we isolated the ‘full-length’ODO1 promoter fragments of the fragrant cv. V26 and the non-fragrant cv. R27 using Mitchell primers, for sequence comparison and to identify putative regulatory motifs. The Mitchell promoter contains a 252-nt fragment, located between 1814 and 1561 nt upstream of the ATG, which is absent in the R27 and V26 promoters. This 252-nt fragment is identical to part of the 11th intron of the petunia PSK6 (Tichtinsky et al., 1998). If this fragment is not considered, the promoters share 87.6% (V26) and 87.9% (R27) identity with the Mitchell promoter. A selection of cis-elements contained in the Mitchell ODO1 promoter is depicted in Figure S3. The Mitchell promoter contains two evening elements (EE; AAAATATCT) (Harmer et al., 2000), which are located at 1116–1108 and 1003–995 nt upstream of the ATG (Figure 4), whereas in the R27 and V26 promoters, the upstream EE has one nucleotide difference (AAcATATCT; Figure 4). In the Mitchell and V26 promoters, there are two MYB consensus sequences (AAACCTAAT) (Sablowski et al., 1994; Liu et al., 2009), which are mutated in the R27 promoter. These two MYB consensus sequences are located directly upstream of each EE in the Mitchell promoter: 1127–1119 nt (MYB binding site, MBS1) and 1014–1006 nt (MBS2). The EE and MBS are part of a 26-bp perfect repeat, which are separated by 87 bp (Figure 4). Additionally, the Mitchell promoter contains two motifs over-represented in light-induced promoters (SORLIP1 and SORLIP2) (Hudson and Quail, 2003), six I BOX cores (Giuliano et al., 1988), three CIACADIANLELHC elements (Piechulla et al., 1998) and one CCA1ATLHB1 element (Wang et al., 1997). The sequence also includes one ethylene responsive element (ERELEE4) (Montgomery et al., 1993) and three MYBCORE elements (Solano et al., 1995) (Figure S3).

Figure 4.

 Sequence alignment of the enhancer of the ODO1 promoter from Mitchell with R27 and V26.
The alignment was generated using jalview ( The sequences upstream of the ATG between positions 1204 and 964 for Mitchell, 1189 and 972 for V26, and 1170 and 960 for R27 are shown. Arrowheads indicate the polymorphisms in the MYB binding sites (MBSs). The dashed lines represent the perfect repeat in the Mitchell promoter. Circadian boxes, light-related boxes and MYB boxes are indicated. EE, evening element; SORLIP, sequence over-represented in light-induced promoters.

Identification of an enhancer region in the ODO1 promoter

We generated a series of 5′ deletion reporter constructs of the 1880-bp Mitchell promoter (Figure 1): (i) to investigate if there were additional regulatory elements upstream of the 1207-bp promoter that determine the level of ODO1 transcription in petals; and (ii) to identify regulatory regions within the 1.2-kbp ODO1 promoter (Figure 5a). Petals of petunia Mitchell plants were infiltrated with Agrobacterium tumefaciens strains that harboured these reporter constructs. The highest expression was seen using M19:GUS and M12:GUS, which showed approximately fourfold higher GUS activity than M10:GUS, and approximately twofold higher activity than M3:GUS (Figure 5a). Thus, cis-elements for high promoter activity seem to be located in the 240-bp region between −1207 and −967 nt in the Mitchell promoter and, with respect to promoter strength, the 1207-bp promoter is comparable with the full-length promoter. To verify if this putative enhancer region was specific for the Mitchell promoter, we compared the activity of R10:GUS and R12:GUS (comparable R27 promoter deletions; Figure 1) in Mitchell flowers. The activity of these constructs was similarly as low as that of the M10:GUS construct (Figure 5b), indicating that the R27 promoter lacks this enhancer. To further investigate the contribution of the enhancer, we generated chimeric reporter constructs, in which this 240-bp enhancer fragment was exchanged between the promoters of Mitchell and R27. Swapping this fragment in the Mitchell promoter with the R27 fragment (R2M10; Figure 1) resulted in decreased GUS activity, comparable with that of M10:GUS and R12:GUS (Figure 5b). Conversely, introducing the Mitchell enhancer to the 1-kbp R27 promoter (M2R10; Figure 1) enhanced GUS activity by 2.5-fold, on average, which is up to 70% of the activity observed in M12:GUS (Figure 5b). We also made stable transformants with the chimeric Mitchell and R27 reporter constructs. Seven out of seven transformants (100%) showed GUS activity for M2R10:GUS, whereas in only two out of five (40%) R2M10:GUS lines GUS activity could be detected in petals (Figure S1).

Figure 5.

 Promoter activity of selected reporter constructs in Mitchell petals using agro-infiltration.
(a, b) Petals of Mitchell plants were transiently transformed using Agrobacterium tumefaciens harbouring a series of 5′-deletion reporter constructs of the Mitchell promoter, the R27 promoter, or chimeric Mitchell and R27 promoters, depicted at the left. The bar on top shows the length of the promoter fragments: M, Mitchell; R, R27. To enable normalization, petals were co-infiltrated with A. tumefaciens harbouring a CaMV 35S:LUC construct (average ± SE, n = 5). Independent experiments with different constructs (n ≥ 3) gave similar results. Letters indicate significant differences (anova, P < 0.05 according to least significant difference post hoc analysis).

Two MYB binding sites in the ODO1 promoter contribute to the strength of the promoter

Sequence comparison with the Mitchell enhancer region (Figure 4) identified 17 polymorphisms in the R27 promoter in addition to 22 nucleotides at the 3′ end, which are absent in the Mitchell promoter. The V26 promoter lacks 31 nucleotides at the 3′ end of the enhancer, and differs from the R27 promoter by only eight nucleotides (Figure 4). Two of these eight nucleotides are identical in the Mitchell and V26 promoters. Like Mitchell, V26 highly expresses ODO1 (Verdonk et al., 2005), and V12:GUS (V26 promoter) is more active than R12:GUS when transiently expressed in Mitchell petals (Figure S4). Interestingly, these two nucleotides are located within the two MBSs. To investigate the possible contribution of the nucleotide polymorphism in MBS1 and MBS2 to promoter strength, we mutagenized both of them in the 1207-bp promoter of Mitchell by changing the cytosine at position 1124 to thymine in MBS1 and the cytosine at position 1012 to adenine in MBS2 (m2M12:GUS; Figure 6), which are the corresponding nucleotides in the ‘MBSs’ of the R27 promoter. This resulted in as much as a 50% reduction of the promoter activity in Mitchell petals (Figure 6). When only MBS1 was mutated (m1M12:GUS), a similar result was obtained (Figure 6), suggesting both elements are necessary for 100% activity. These results suggest that an MYB-type transcription factor can activate the ODO1 promoter.

Figure 6.

 Mutation of the MYB binding sites alters ODO1 promoter activity.
Agroinfiltration of wild-type and mutated Mitchell ODO1 promoter reporter constructs in petals of Mitchell. Black boxes in the promoter represent wild-type MYB binding sites (MBSs), as found in the Mitchell promoter (M12:GUS). Hatched boxes represent MBSs, as found in the R27 promoter, which have been mutated in the Mitchell promoter (m1M12:GUS and m2M12:GUS). In MBS1 and MBS2, a cytosine residue was changed to adenosine and thymidine, respectively (inset). Petal corollas were co-infiltrated with CaMV 35S:LUC for normalization (average ± SE, n = 5). The experiment was repeated twice, and the results of a representative experiment are shown.

EOBII can trans-activate the ODO1 promoter in N. benthamiana leaves, and binds to a MYB binding domain in vitro

Recently, the petunia R2R3-MYB EOBII was identified as a regulator of fragrance biosynthesis in petunia flowers, and silencing of EOBII resulted in the downregulation of ODO1, among other genes such as PAL2 and IGS (Spitzer-Rimon et al., 2010). Unlike ODO1 (Figure 3a), EOBII is expressed in the flowers of the non-fragrant R27 cultivar (Figure 7a). As the Mitchell promoter is active in the R27 background (Figure 3b), and is potentially activated by a MYB (Figure 6), we set out to investigate whether EOBII can activate the ODO1 promoter. Because EOBII is not expressed in N. benthamiana leaves (data not shown), we chose this system for promoter trans-activation assays. Agroinfiltration of N. benthamiana leaves has previously been used to analyse plant promoters (Yang et al., 2000). To investigate EOBII activation of the ODO1 promoter, we first transiently co-expressed CaMV35-driven EOBII (35S:EOBII) with the M19:GUS reporter construct in the leaves of N. benthamiana (Figure 7b). As a control, the reporter construct was separately co-infiltrated with CaMV 35S-driven ODO1 (35S:ODO1) and CaMV 35S-driven RED FLUORESCENT PROTEIN (35S:RFP). EOBII enhanced reporter activity up to sevenfold compared with the RFP-infiltration and ODO1-infiltration, indicating that EOBII can activate the ODO1 promoter independently from other petunia factors (Figure 7c).

Figure 7.

Trans-activation of the ODO1 promoter of cv. Mitchell by EOBII in Nicotiana benthamiana leaves.
(a) EOBII is expressed in the corollas of both Mitchell and R27. RT-PCR using RNA isolated from petal corollas at the peak of EOBII expression. FBP1 (Floral Binding Protein 1) was used as a reference gene. The numbers on the left indicate the number of PCR cycles.
(b) Constructs used for the trans-activation assay.
(c) Normalized GUS activity after co-infiltration with Agrobacterium harbouring any of the effector constructs with the M19:GUS reporter construct. Co-infiltration with the CaMV 35S:ODO1 or CaMV 35S:RFP effectors were used to show the specificity of EOBII. Co-infiltration with 35S:LUC enabled normalization (average ± SE, n = 6). RFP: red fluorescent protein.

In order to detect which part of the ODO1 promoter is necessary for activation by EOBII, we performed trans-activation assays using the 5′ deletion, as well as the chimeric and the mutagenised promoter reporter constructs, in N. benthamiana. Because the M19:GUS and M12:GUS reporters were equally strongly activated by EOBII (Figure S5), we argued that the binding site for EOBII was contained within the 1207-bp promoter. The highest activation by EOBII co-expression was seen for M12:GUS, resulting in 2.5–3.5-fold higher activity compared with M3:GUS and M10:GUS (Figure 8). The latter result suggests the EOBII binding site is located within the previously identified enhancer region.

Figure 8.

 Fine-tuning of the EOBII binding region in the ODO1 promoter.
Leaves of Nicotiana benthamiana were co-infiltrated with Agrobacterium harbouring selected reporter constructs, indicated left, and the 35S:EOBII effector construct: M, Mitchell; R, R27. A control co-infiltration of each reporter construct with CaMV 35S:RFP was used to set the GUS activity to 1 (indicated by the vertical line). Activation by EOBII for each reporter construct thus indicates fold induction relative to CaMV 35S:RFP. Co-infiltration with 35S:LUC enabled normalization (average ± SE, n = 6). The experiment was repeated once and the results of a representative experiment are shown. Letters indicate significant differences (anova, P < 0.05 according to least significant difference post hoc analysis).

Because R27 expresses EOBII, but not ODO1, we tested if EOBII could activate the R27 promoter, and, as expected, the R12:GUS reporter construct could not be activated by EOBII (Figure 8). Interestingly, the 1-kbp ODO1 promoter of R27 with the 240-bp enhancer region added (M2R10:GUS) was only weakly activated by EOBII in N. benthamiana leaves (Figure 8).

Mutation of the MBSs in the enhancer region of Mitchell affected promoter activity in petals (Figure 6). Therefore, we wanted to know if it would affect activation by EOBII as well. The activity of m2M12:GUS dropped to approximately 30% compared with the activity of M12:GUS (Figure 9), indicating that, to a large extent, EOBII activation depends on these two nucleotides in the MBSs.

Figure 9.

 Mutations in the MYB binding site disrupt activation by EOBII.
Leaves of Nicotiana benthamiana were co-infiltrated with Agrobacterium harbouring a CaMV 35S:EOBII activator and 1.2-kbp Mitchell reporter constructs with the wild-type (M12:GUS) and mutated (m1M12:GUS) MYB binding sites (MBSs) 1 and 2. Co-infiltration with 35S:LUC enabled normalization (average ± SE, n = 6). The experiment was repeated twice and the results of a representative experiment are shown.

Finally, to verify EOBII binding at the MBS, we performed electrophoretic mobility shift assays (EMSAs) using recombinant GST-tagged EOBII. Figure S6 shows that EOBII specifically binds to the MBS in the ODO1 promoter. When the MBS is replaced by an oligoA stretch, or when the cytosine at position 1124 is changed to thymine within the MBS, the binding of EOBII to the promoter is strongly reduced. Thus, binding of EOBII to the MBS in vitro (Figure S6) is in agreement with the capacity to activate the promoter at the MBS in vivo (Figure 9).


Transcriptional regulation is an important mechanism to control metabolic pathways. These pathways are generally influenced by environmental stimuli and the plant’s developmental program. The regulatory network of plant volatile benzenoid/phenylpropanoid biosynthesis is largely unknown, although evidence indicates that the pathway is strongly influenced by the circadian clock (Kolosova et al., 2001; Colquhoun et al., 2010b) and flower development (Dudareva et al., 2000; Verdonk et al., 2003).

By using stable reporter lines and a transient transformation assay in petals (Shang et al., 2007) with a series of reporter constructs, we identified a crucial enhancer region in the promoter of ODO1, a regulator of volatile benzenoid/phenylpropanoid biosynthesis. In addition, we were able to identify EOBII, another regulator of the pathway, as a direct activator of ODO1 transcription. Interestingly, two polymorphisms in the ODO1 enhancer, distinguishing a fragrant from a non-fragrant petunia, affected EOBII binding and trans-activation, and contributed to overall ODO1 transcript abundance in fragrant petunias.

Significance of the ODO1 promoter in fragrant and non-fragrant P. hybrida cultivars

In petunia, the ability to produce scent appears to be correlated with ODO1 expression. When ODO1 transcript accumulation is prevented by RNAi, precursors cannot be synthesized, and as a consequence low quantities of volatiles are produced (Verdonk et al., 2005). In addition, in the fragrant P. hybrida cultivars Mitchell and V26, ODO1 is highly expressed, whereas in the non-fragrant cultivar R27 ODO1 transcripts are not detectable (Figure 3a; Verdonk et al., 2005). To investigate the contribution of cis-elements to its expression level, the ODO1 promoters from Mitchell, V26, and R27 were isolated and compared. The promoters are polymorphic, but overall nucleotide identity is high. In Mitchell petals, the promoter of Mitchell is four times more active than the R27 promoter (Figure 3b). A similar result was obtained in R27 petals (Figure 3c), confirming the presence of cis-regulatory sequences in the Mitchell promoter, which are absent in the R27 promoter and are necessary for high activity. This only partially explains the difference in ODO1 transcript abundance between both petals, as ODO1 transcripts are not detectable in R27 using RT-PCR (Figure 3a). One explanation for this discrepancy could be that cis-binding sites for repressors are present in the R27 promoter, located more upstream in the promoter. Alternatively, other types of regulation than one would predict from the strength of the promoter alone can cause reduced transcript abundance, like chromosome context, regulatory sequences elsewhere in the gene and transcript stability. Finally, it should be noted that the transient nature of our assay can result in high numbers of transgenes in the cell, potentially overestimating the strength of a weak promoter. In agreement with this, only approximately 25–40% of the stable reporter lines that lacked the enhancer region showed detectable levels of GUS activity in their petals, whereas this rose to 85–100% for reporter lines that included the enhancer (Figure S1). These numbers suggest a significantly higher promoter strength in lines that contain the enhancer:GUS construct.

Surprisingly, normalized activity of both promoters in transiently transformed petals is higher in R27 than in Mitchell (data not shown), which could reflect the absence of repressors in R27. Future work is needed, not only to identify activators, but also repressors of ODO1 transcription, to unravel the complex regulatory mechanisms that determine overall ODO1 transcript abundance. In conclusion, our experiments show that trans-acting factors in the non-fragrant R27 flowers function to activate the fragrant Mitchell promoter, and that to a large extent Mitchell-specific cis-regulatory elements in the promoter are responsible for high ODO1 expression.

Regulation of ODO1 transcription

In order to understand the transcriptional network underlying secondary metabolism in plants, the identification of trans-activating factors is essential. Of particular interest are TFs that regulate other TFs, because they can be seen as master regulators of a pathway. Because ODO1 is a key regulator for the precursor availability of the volatile benzenoid/phenylpropanoid pathway, controlling regulation of ODO1 would implicate control over the pathway. Unlike for other metabolic pathways in plants, regulation of the volatile benzenoid/phenylpropanoid pathway is far from understood. Deletion analysis with reporter constructs revealed the importance of a 240-bp fragment in the ODO1 promoter of the fragrant cultivar Mitchell (Figure 4), which turned out to be necessary for high promoter activity in petals (Figure 5). Replacing this fragment with the corresponding fragment of R27 significantly decreased the activity of the promoter. Conversely, swapping the R27 fragment with the Mitchell counterpart in the ‘minimal’ R27 promoter could restore the ODO1 promoter strength to as much as 70% (Figure 5b). This means that the 240-bp fragment acts as an enhancer, which is specific for a fragrant, and thus ODO1-expressing, cultivar. It also means the enhancer is necessary and to a large extent sufficient for full promoter strength in petunia petals.

Although ODO1 transcript abundance was reduced in petals suppressed in EOBII transcription using VIGS, transient overexpression of EOBII at the peak of ODO1 expression did not result in higher ODO1 transcript levels (Spitzer-Rimon et al., 2010). This leads to the hypothesis that additional factors are involved for ODO1 expression (Spitzer-Rimon et al., 2010). We show that EOBII is able to activate the ODO1 promoter in N. benthamiana leaves, without the involvement of additional petunia factors (Figure 7). This does not, however, exclude additional interactions with the ODO1 promoter in petunia petals. The fact that the transient overexpression of EOBII did not result in increased ODO1 mRNA abundance in petals (Spitzer-Rimon et al., 2010) could be because the measurements were performed at the peak of ODO1 expression, which would make it difficult to detect an increase. We further show that EOBII activates the Mitchell promoter more strongly than the R27 promoter (Figure 8). This is in agreement with the activity of both promoters in petunia petals (Figure 5b). Mutational analysis within the enhancer region identified two nucleotides specific for fragrant petunias (Figure 4) that are necessary for high promoter activity in petals (Figure 6). These polymorphisms occurred in two MYB consensus sequences (Sablowski et al., 1994) within the enhancer region (Figure 4), which correspond to a core sequence recently identified in the promoter of nec1, and which is a target for MYB305 from ornamental tobacco (Liu et al., 2009). The same core sequence (MBS) is present once in the PALB promoter of tobacco, which has been shown to be activated by EOBII as well (Spitzer-Rimon et al., 2010). By changing nucleotide C1124 to T in MBS1, and C1012 to A in MBS2, we show that activation of the ODO1 promoter by EOBII in tobacco leaves can be reduced significantly (Figure 9). When analyzing the M2R10:GUS construct, which is essentially the ‘minimal’ 1187-bp R27 promoter in which the two MBSs are restored within the context of the enhancer region, this chimeric promoter cannot be strongly activated by EOBII in N. benthamiana (Figure 8). This means that the MBSs in the enhancer region are necessary, but not sufficient, for activation by EOBII. This result is different from what is seen in petunia petals (Figure 5b), and suggests that an additional binding site for EOBII is present in the Mitchell promoter. It also suggests that an additional petunia factor causes high activity of M2R10:GUS in petals, as compared with trans-activation in N. benthamiana (Figure 8). Generally, R2R3-MYB TFs bind a single consensus sequence in a promoter (Dubos et al., 2010). However, MYB305 of ornamental tobacco was shown to bind two related core sequences, separated by 813 bp, in the nec1 promoter in vitro. Furthermore, both core sequences were necessary for high GUS activity in planta, although weak activation on a single core sequence was seen (Liu et al., 2009). Our results suggest a similar mechanism for ODO1 activation, although we could not find an additional MBS in the ODO1 promoter sequence. In agreement with this, the IGS promoter lacks the MYB consensus sequence we describe here, despite its activation by EOBII (Spitzer-Rimon et al., 2010). We showed binding of EOBII to the MBS (Figure S6), of which two copies exist approximately 1.2 kb upstream of the ATG (Figure 4). The identification of additional binding sites within the promoter needs further investigation. EOBII is expressed before anthesis when ODO1 expression is low (Verdonk et al., 2005; Spitzer-Rimon et al., 2010), indicating that ODO1 expression is repressed before anthesis or that an additional factor is needed for high expression after anthesis. Although our results show that EOBII can activate ODO1 without additional petunia factors, they also suggest combinatorial regulation at the ODO1 promoter that might be developmentally controlled. Future work is needed to identify putative EOBII-interacting factors at the ODO1 promoter, as well as to assess the importance of the two EEs in the Mitchell promoter for rhythmic expression.

Experimental Procedures

Plant material

Petunia hybrida cultivars Mitchell (also referred to as W115), R27 and V26, and N. benthamiana plants, were grown in standard glasshouse conditions (16-h photoperiod, 300–500 μmol m−2 sec−1 light intensity, 60–65% humidity and day/night temperatures of 22/17°C). Petunia plants were moved to a controlled growth chamber (16-h photoperiod, 250–350 μmol m−2 sec−1 light intensity, 70% relative humidity and constant temperature of 21°C) at least 3 days prior to the experiments, unless stated otherwise.

Promoter isolations

Genomic DNA was isolated from leaves with extraction buffer (7 m urea, 0.3 m NaCl, 50 mm Tris–HCl, 20 mm EDTA, 1% lauroylsarcosyl, pH 8.0), followed by phenol/chloroform purification and isopropanol precipitation. First, a 2133-bp fragment of the Mitchell promoter, including the 5′ untranslated region (5′UTR), was obtained using a genome walking protocol (Siebert et al., 1995). For this, genomic DNA was digested with EcoRV, HinCII, PvuII, ScaI, SmaI or SspI. Next, an adapter was created by annealing a short (5′-P-acctgcccgggc-N-3′) oligonucleotide and a long (5′-ctaatacgactcactatagggctggagcggccgcccgggcaggt-3′) oligonucleotide, which was then ligated to the digested gDNA. Upstream promoter fragments were PCR amplified using an adapter-specific forward (5′-ctaatacgactcactatagggc-3′) and an ODO1-specific reverse primer. A nested PCR was performed on the first PCR product with a second adapter-specific forward (5′-tcgagcggccgcccgggcaggt-3′) and a second ODO1-specific reverse primer. Finally, the PCR products were cloned in pCR2.1 (Invitrogen, and sequenced. The promoters of the other cultivars were PCR amplified using Mitchell-specific primers, cloned and subsequently sequenced.

Construct design

Constructs were made using standard molecular techniques. To create the 5′-deletion GUS-reporter constructs, promoter fragments were PCR-amplified, introducing a HindIII site upstream of the promoter and a PscI site at the ATG (Table S1), and cloned HindIII/NcoI in a shuttle vector containing the uidA CDS with intron followed by the Nos terminator (tNos) between an NcoI site at the ATG and an EheI site downstream of the tNos. The pODO1:GUS(I)tNos cassettes were HindIII/EheI ligated in the MCS of the binary vector pBINPLUS (Vanengelen et al., 1995) between HindIII and SmaI. To create the chimeric promoter GUS reporter constructs (R2M10:GUS and M2R10:GUS), the respective promoter fragments were fused using an overlapping PCR strategy and subsequently cloned as described above (Figure 1; Table S1). All constructs were sequenced. For the construction of the single mutated promoter reporter construct (m1M12:GUS), the Mitchell promoter was PCR amplified using a forward primer that contained the mutation (primer F5; Table S1) and a reverse primer. A second PCR on this product was performed (primers F3 and R1; Table S1), introducing a HindIII and PscI site, and cloned as described above. For the double mutated promoter reporter construct (m2M12:GUS), the promoter region containing the two mutations was first PCR amplified using forward and reverse primers, which each contained the mutation (primers F5 and R3; Table S1) and an overlapping PCR, and subsequent cloning was performed as above using primers F3 and R1 (Table S1). For the EOBII overexpression construct, the EOBII CDS was PCR amplified, introducing an NcoI site at the ATG and a SacI site downstream of the stop codon (forward primer, 5′-catgccATGgataaaaaaccatgcaac-3′; and reverse primer, 5′-cgagctcTTAatcaccattaagcaattg-3′; restriction sites are in bold, and start and stop codons are in upper case). The PCR product was ligated NcoI/SacI between a CaMV 35S promoter and the Nos terminator of a shuttle vector, and the expression cassette was cloned AscI/SfoI in the binary vector pBINPLUS (Vanengelen et al., 1995). The 35S:ODO1 construct is described by Verdonk (2006). To create a glutathione-S-transferase (GST)-tagged fusion protein construct, the EOBII CDS was NcoI/SacI digested from the 35S:EOBII overexpression construct, and cloned NcoI/SacI into pGEX-KG.

Stable transformations and reporter assays in petunia flowers

Stable P. hybrida cv. Mitchell transformants were made using a standard leaf disc method (Horsch et al., 1985) with minor modifications, as described by Van Moerkercke et al. (2009). Agrobacterium tumefaciens GV3101 (pMP90) harbouring different reporter constructs was used (Figure 1). Rooted plants were transferred to soil and screened for transgene integration by PCR using construct-specific primers. For each T0 transgenic line, GUS activity in petals was determined. For M19:GUS and M12:GUS, three independent T0 lines were chosen for the detailed quantitative assessment of GUS activity in leaves, petal tubes and petals of 1-cm-long buds, 3-cm-long buds and 1-day-old open flowers, collected from glasshouse-grown plants at 18.00–20.00 h in winter, 2 h before artificial illumination ceased. For each sample, at least three flowers/leaves were pooled. Tissues were homogenized in liquid N2 and proteins were extracted from 50 mg of tissue in 150 μl of cold luciferase cell culture lysis reagent [CCLR; 25 mm Tris-phosphate, pH 7.8; 2 mm DTT, 2 m 1,2-diaminocyclohexane-N,N,N′,N′-tetraacetic acid (CDTA), 1% Triton®-X-100, 10% glycerol; Promega,, supplemented with complete protease inhibitor (Roche, Samples were vortexed for 30 sec and centrifuged at 4°C for 30 min at 13 000 g. GUS activity of the crude extract was determined spectrophotometrically using 4-methylumbelliferyl-d-glucuronide (MUG) as a substrate (1 mm MUG in CCLR supplemented with 10 mmβ-mercaptoethanol) after a 30-min incubation period at 37°C. Measurements were performed in a FluoroCount Microplate Fluorometer (Packard BioScience Company, using a 360-nm excitation filter and a 460-nm emission filter. Activities were corrected for protein content, measured using Bradford reagent. For each tissue and developmental stage values of a 35S:GUS line were set to 100%. In planta GUS staining was performed as described by Jefferson et al. (1987).

Transient transformation and reporter assays in petunia flowers

Agrobacterium tumefaciens-mediated transient transformations of petal limbs were performed as described by Verweij et al. (2008). A. tumefaciens GV3101 (pMP90) harbouring different reporter constructs was used (Figure 1). In order to normalize for transformation efficiency and protein extraction efficiency, we co-infiltrated with a construct containing firefly luciferase (LUC) driven by the CaMV 35S promoter. Because the LUC used in this study does not contain an intron, we first determined potential LUC expression by A. tumefaciens after infiltration and incubation for 48 h. For this we used the A. tumefaciens strain GV3101 cured of plasmid pMP90 (Holsters et al., 1980), which confers virulence. Essentially no LUC activity was measured using this strain compared with the virulent, pMP90-containing strain in N. benthamiana leaf extracts (Figure S7), indicating that LUC is expressed by the plant and not by A. tumefaciens under these conditions.

The Agrobacteria containing the reporter and CaMV 35S:LUC constructs were mixed 5:1, respectively, prior to infiltration. One-day-old Mitchell petal limbs were infiltrated and incubated for 36 h in the growth chamber prior to analyses. After the incubation period, five petal discs of infiltrated limbs per flower were pooled and frozen in liquid nitrogen. At least five petals were used per construct (n = 5), and each experiment was independently repeated at least once (n ≥ 2). In vitro GUS activity was determined as described above. For LUC measurements, 20 μl of extract was analysed in 80 μl of luciferase assay buffer (20 mm Tricine, 2.67 mm MgSO4, 0.1 mm EDTA, 33 mm DTT, 270 μm co-enzyme A, 530 μm ATP, 470 μm d-Luciferin, pH 7.8; van Leeuwen et al., 2000), and measured in a FluoroCount Microplate Fluorometer (Packard BioScience Company) using a 560-nm emission filter.

Transient trans-activation assay in N. benthamiana leaves

Agrobacterium tumefaciens GV3101 (pMP90) cultures harbouring reporter or effector constructs were grown overnight and diluted in infiltration buffer [50 mm 2-(N-morpholino)ethanesulfonic acid (MES), pH 5.8, 0.5% glucose, 2 mm Na3PO4, 100 μm acetosyringone] to an OD600 of 0.3 prior to infiltration in N. benthamiana leaves. Combinations of reporter and effector constructs were co-infiltrated in leaves of glasshouse-grown N. benthamiana plants and incubated for 48 h. Regions of infiltration were marked, and five leaf discs were pooled per sample. Six leaves on two plants were infiltrated for each effector/reporter combination (n = 6), and the experiment was repeated twice. To enable normalization, leaves were co-infiltrated with a CaMV 35S:LUC-harbouring A. tumefaciens. GUS and LUC activity were determined as described above.

RNA-isolation, cDNA synthesis, RT-PCR and quantitative RT-PCR

RNA was isolated from petals harvested 2 h before the onset of the dark period using TriZol reagent (Invitrogen, A 1-μg portion was used for first-strand cDNA synthesis using an oligo-dT(18) primer and M-MuLV Reverse Transcriptase (Fermentas, after DNase I (Ambion Turbo DNA-free; Applied Biosystems, treatment to remove contaminating gDNA. Semi-quantitative RT-PCR was performed on diluted cDNA samples using a standard PCR program and DreamTaq polymerase (Fermentas). We used primers 5′-gtgttctttgtgatgctcgtg-3′ and 5′—caacctctcctgcaaatttgg-3′ for the amplification of FBP1, primers 5′-catgccatggataaaaaaccatgcaac-3′ and 5′-cgagctcttaatcaccattaagcaattg-3′ for EOBII, primers 5′-gttggtggtagctgagagtcag-3′ and 5′-gactctaagcaaatctaacttcc-3′ for ODO1, and primers 5′-cagactgaatgcccacaggccgtcgag-3′ and 5′-ctgaatgcccacaggccgtcgag-3′ for uidA. For quantitative RT-PCR (qRT-PCR), we used forward primer 5′-gtgccttagcttgcttctttagagg-3′ and reverse primer 5′-ccttttctttgtggacctttttgg-3′ to amplify ODO1, and forward primer 5′-ttcgttggcaatactccacatc-3′ and reverse primer 5′-ttacaggcgattaaagagctgatag-3′ for uidA. We amplified ACTIN (forward primer 5′-tgctgatcgtatgagcaaggaa-3′ and reverse primer 5′-ggtggagcaacaaccttaatcttc-3′; Spitzer-Rimon et al., 2010) to enable normalization. For analyses, the cDNA equivalent of 10 ng of total RNA was used per reaction, along with SYBR Green qPCR SuperMix UDG (Invitrogen), ROX reference dye and 300 nm of each primer. Primer pair efficiencies were calculated by analysis of amplification curves of a standard cDNA dilution range. Two technical and two biological replicas were used.

Recombinant protein production and EMSA

The GST-EOBII fusion protein construct was transformed into C41 (DE3) cells (Dumon-Seignovert et al., 2004). Cells were grown until they reached OD600 of 0.5, induced with 0.5 mm isopropyl-β-d-thio-galactoside (IPTG) and grown for an additional 4 h at 20°C. Proteins were extracted in 20 mm Tris–HCl, pH 7.5, 100 mm NaCl, 0.2% Triton X-100, 0.05% Tween 20, 5 mm EDTA, 1 mm EGTA and 5 mm DTT, supplemented with Roche complete protease inhibitor (Roche). The recombinant protein was purified using glutathione sepharose 4B beads (GE Healthcare,, and the size of the fusion protein was confirmed using SDS-PAGE and western blot analysis. For electrophoretic mobility shift assays (EMSAs), we created three double stranded probes. For this, three oligo pairs (probe M, 5′-aaaataggacataAACCTAATaaaaaatatcttg-3′ and 5′-caagatattttttATTAGGTTtatgtcctatttt-3′; probe m1, 5′-aaaataggacataAAAAAAAAaaaaaatatcttg-3′ and 5′-caagatattttttTTTTTTTTtatgtcctatttt-3′; probe m2, 5′-aaaataggacataAACTTAATaaaaaatatcttg-3′ and 5′-caagatattttttATTAAGTTtatgtcctatttt-3′; the MBSs are indicated in upper case, and the mutations in the MBSs in probes m1 and m2 are indicated in bold) were separately phosphorylated using T4 polynucleotide kinase (Fermentas) with [γ-32P]ATP and purified using mini Quick Spin Oligo Columns (Roche). The labelled oligos were then boiled for 2 min and slowly cooled to room temperature (20°C) in a glass beaker with water. A 25-fmol portion of labelled DNA (20.000 cpm) was incubated with 5 pmol of recombinant protein in 10 mm Tris–HCl, pH 7.8, 150 mm NaCl, 5% glycerol, 0.5 μg of poly(dI-dC), 0.1 mm DTT for 30 min at 20°C. Mixtures were separated on a 6% polyacrylamide gel and visualized on photographic film. Specificity was shown by adding cold competitor in a 50-fold excess. We used a GST-tagged control protein to show the binding specificity of EOBII.


The authors would like to thank Miriam Kroon for help with promoter isolations, Carlos Galván-Ampudia for help with the recombinant GST-EOBII protein production and purification, Magdalena Julkowska and Fionn McLoughlin for the GST-SnRK2.4 control protein, Marcella Holsters (Ghent University, VIB, Belgium) for the cured Agrobacterium tumefaciens GV3101, Joop Vermeer for the 35S:RFP binary vector, Julian Verdonk for the 35S:ODO1 binary vector and Francesca Quattrocchio (Free University, the Netherlands) for the 35S:LUC construct. We thank Petra Bleeker for critically reading the article. This research was supported by the University of Amsterdam.

Accession numbers: Sequences corresponding to the ODO1 promoters of Petunia hybrida cv. Mitchell (HQ901078), cv. R27 (HQ901079) and cv. V26 (HQ901080) have been deposited in GenBank.