The strawberry FaMYB1 transcription factor suppresses anthocyanin and flavonol accumulation in transgenic tobacco

Authors


For correspondence (fax +31 317 418094; e-mail a.aharoni@plant.wag-ur.nl).

Summary

Fruit ripening is characterized by dramatic changes in gene expression, enzymatic activities and metabolism. Although the process of ripening has been studied extensively, we still lack valuable information on how the numerous metabolic pathways are regulated and co-ordinated. In this paper we describe the characterization of FaMYB1, a ripening regulated strawberry gene member of the MYB family of transcription factors. Flowers of transgenic tobacco lines overexpressing FaMYB1 showed a severe reduction in pigmentation. A reduction in the level of cyanidin 3-rutinoside (an anthocyanin) and of quercetin-glycosides (flavonols) was observed. Expression of late flavonoid biosynthesis genes and their enzyme activities were aversely affected by FaMYB1 overexpression. Two-hybrid assays in yeast showed that FaMYB1 could interact with other known anthocyanin regulators, but it does not act as a transcriptional activator. Interestingly, the C-terminus of FaMYB1 contains the motif pdLNLD/ELxiG/S. This motif is contained in a region recently proposed to be involved in the repression of transcription by AtMYB4, an Arabidopsis MYB protein. Our results suggest that FaMYB1 may play a key role in regulating the biosynthesis of anthocyanins and flavonols in strawberry. It may act to repress transcription in order to balance the levels of anthocyanin pigments produced at the latter stages of strawberry fruit maturation, and/or to regulate metabolite levels in various branches of the flavonoid biosynthetic pathway.

Introduction

Several characteristics of fleshy fruit development and maturation might be attributed to changes in the levels of certain phenolic compounds, amongst them flavonoids (Seymour et al., 1993). In strawberry, for example, astringency associated with early fruit development is due to the presence of tannins, while anthocyanins are responsible for the characteristic red colour associated with late ripening (Cheng and Breen, 1991). In cultivated strawberry, the glucosylated anthocyanin pelargonidin (pelargonidin 3-glucoside) is the main anthocyanin present in ripe fruit (approximately 88%), along with other pelargonidin-glycosides and cyanidin 3-glucoside (Perkins-Veazie, 1995). During ripening, in addition to anthocyanins other flavonoid derivatives such as the flavonols quercetin and kaempferol are also produced (Hakkinen et al., 1999). The biological function of these compounds in fruit has been mainly credited to their role in attracting fruit-eating animals and protecting against harmful ultraviolet light and pathogens (Shirley, 1996).

The flavonoid biosynthesis pathway is a branch of the large phenylpropanoid pathway, which includes the biosynthesis routes to compounds such as lignins and phenolic acids (and their derivatives). The precursors for the synthesis of all flavonoids are malonyl-CoA and p-coumaroyl-CoA (Dixon and Steele, 1999). Virtually all genes encoding enzymes for the biosynthesis of flavonols and anthocyanins have been identified (Mol et al., 1998), including those modifying the basic skeleton (e.g. glycosylation, methylation, acylation and hydroxylation). In many plant species the structural genes of the flavonoid biosynthetic pathway appear to be largely regulated at the transcriptional level. Amongst the regulatory genes identified to date, members of the MYB family of transcription factors are chiefly represented (Jin and Martin, 1999). From the analysis of more than 100 R2R3-MYB protein sequences from different plant species (mainly Arabidopsis thaliana), a number of conserved motifs have been identified (Kranz et al., 1998; Riechmann and Ratcliffe, 2000). On the basis of these motifs, the MYB proteins have been classified into 22 subgroups. Some of these identified motifs may represent activation domains, while others may serve as repression domains or even domains for interaction with other transcription factors. The majority of MYB genes with assigned functions have been predicted to be transcriptional activators (Glover et al., 1998; Larkin et al., 1993; Martin and Paz-Ares, 1997; Urao et al., 1993).

Transcriptional regulation of gene expression is not mediated solely by activators, but also by the action of repressors; in some cases a transcription factor may perform both activities (Coffman et al., 1997). Temporal and spatial interactions between different combinations of activators and repressors give rise to a wide spectrum of expression patterns (‘combinatorial control’). Studying real transcriptional repression is complicated, due to effects caused by a dominant negative form of proteins (Schwechheimer and Bevan, 1998). Early studies on MYB transcription factors in maize reported on C1-I as a dominant inhibitor allele of the C1 gene, which is normally required for the synthesis of anthocyanin in the aleurone tissue (Paz-Ares et al., 1990). Transient expression assays showed that changes both in the C-terminus and in the DNA-binding domain were important for the C1-I inhibitory effect (Goff et al., 1991). Transient expression assays with another C1 homologue from maize, termed Zm38, showed that constructs retaining either the entire Zm38 cDNA, or a combination of the Zm38 3′ region and the 5′ region of the C1 cDNA, confer complete inhibition of the co-transformed wild-type C1 construct (Franken et al., 1994). The mode of action of these putative repressors was suggested to be either competition for binding sites in target promoters, or the formation of mixed dimers of C1-I and Zm38 with C1. Aligning the Zm38 C-terminus amino acid sequence reveals that it is structurally related to two other members of subgroup 4 of the MYB proteins (AmMYB308 and AtMYB4), recently reported to function as transcriptional repressors (Jin et al., 2000; Tamagnone et al., 1998). Overexpression of the Antirrhinum AmMYB308 gene in tobacco caused an inhibition of hydroxycinnamic acid and monolignol accumulation, by reducing expression of genes encoding enzymes that are part of their biosynthetic pathway. An Arabidopsis knockout mutant of AtMYB4, the proposed orthologous gene of AmMYB308, showed an increase in sinapate ester accumulation, which resulted in enhanced UV-B irradiation tolerance. Sinapate esters are utilized as sunscreens in Arabidopsis leaves. Under certain environmental conditions, when protection from irradiation is not essential (e.g. in the dark), AtMYB4 reduces the accumulation of sinapate esters by repressing the expression of the gene encoding the key enzyme in their biosynthesis (cinnamate 4-hydroxylase, C4H).

In an attempt to study the genetic regulation of strawberry ripening, we have cloned and characterized the FaMYB1 gene. FaMYB1 encodes an R2R3 MYB protein homologue which is primarily expressed in the red ripe strawberry fruit. Heterologous expression in tobacco primarily affects the metabolism of anthocyanins and flavonols, and suggests a role for FaMYB1 as a transcriptional regulator of late flavonoid biosynthesis genes.

Results

Strawberry FaMYB1 is a short R2R3 MYB protein

Using a random expressed sequence tag (EST) sequencing approach, in which 1100 cDNAs derived from a strawberry red fruit cDNA library were partially sequenced (Aharoni et al., 2000), we identified and isolated two strawberry transcripts putatively encoding R2R3 MYB proteins (FaMYB1 and FaMYB2). FaMYB1 appeared to be a full-length cDNA (1135 bp) encoding a protein of 187 amino acids. FaMYB2 was a truncated cDNA (732 bp) that, when translated, was nearly identical to the putative translation product of FaMYB1 except for two base-pair substitutions, which either did not alter the amino acid composition or produced a conserved amino acid substitution. Both transcripts showed large differences in their 3′ untranslated regions (UTR), primarily due to a deletion of 143 bp in FaMYB2. We assumed that both clones were derived from different alleles within the genome of the octaploid variety used for their isolation (cv. Elsanta).

FaMYB1 encodes a relatively short MYB protein (187 amino acids compared to an average of 230–300 amino acids for most other plant MYB proteins). However, it has a relatively long 3′ UTR (496 bp) compared to its predicted ORF (564 bp; Figure 1a). We confirmed that the protein encoded by FaMYB1 was not a truncated R2R3 MYB protein but a functional entity by itself, by performing reverse transcription PCR (RT–PCR) and 3′ RACE experiments (data not shown).

Figure 1.

FaMYB1 cDNA structure and homology to plant R2-R3 MYB proteins.

(a) Schematic drawing of FaMYB1 cDNA showing the coding region (rectangle) and the 3′ and 5′ untranslated regions (lines). R2 and R3 are the two repeats of the MYB DNA-binding domain. C1 and C2 are two conserved motifs located outside the MYB domain and identified in other members of subgroup 4 MYB proteins (Kranz et al., 1998). Numbers in the upper and lower parts correspond to amino acid and nucleic acid, respectively.

(b) Amino acid sequence alignment of FaMYB1 and other R2-R3 MYB protein homologues in the MYB domain. The line above indicates the R2 and R3 MYB repeats and the regularly spaced tryptophan (W) residues (or phenylalanine, F) are shaded in green. Black background indicates 100% conservation; grey or light grey represent 80 and 60% conservation, respectively. The four amino acid residues required for the interaction with the maize bHLH protein R (Grotewold et al., 2000) present in the MYB proteins C1, AN2 and GL1, but not in the P MYB protein, are shown (box) and marked in all proteins (shaded red). Two additional amino acids important for interaction in plant cells are marked above with asterisks (residues GR), both are present in C1, AN2 and GL1. Abbreviations: At, Arabidopsis thaliana; Fa, Fragaria ananasa; Ph, Petunia hybrida; TM, tomato; Am, Antirrhinum majus; Zm, Zea mays; Hv, Hordeum vulgare; Gh, Gossypium hirsutum; Os, Oryza sativa. GenBank accession numbers or reference for the sequences: FaMYB1, AF401220, PhMYB27, Mur, 1995; TMH27, X95296; AmMYB308, JQ0960; ZmMYB38, X78846; HvMYB1, X78845; AtMYB4, AF062860; HvMYB5, X70880; AmMYB330, JQ0957; GhMYBA, L04497; AtMYB, genomic clone T22J18.19; GhMYB (GhMYB6), AF034134; OsMYB1, D88617; AtMYB7, U26937; AtMYB6, U26936.

(c) Amino acid sequence alignment of FaMYB1 and other R2-R3 MYB protein homologues outside the MYB domain. The arrow indicates the end of the R3 repeat in the MYB domain. Black background indicates 100% conservation; grey or light grey represent 80 and 60% conservation, respectively. The motifs C1, C2 and a zinc finger are indicated above their location in the protein sequences, as described by Kranz et al. (1998) for members of MYB subgroup 4. Upper or lower case letters indicate amino acids found in all or more than 50% of the subgroup genes, respectively. If two amino acids are found at the same position, both are given and they are separated by a slash.

The FaMYB1 protein sequence showed greatest homology to GhMYB6, a cotton ovule protein (GenBank accession no. AF034134), with 60.1% overall identity and 84.5% identity in the R2R3 DNA-binding domain (Figure 1b,c). As for other MYB proteins, FaMYB1 shared very little homology to other MYB proteins in the C-terminus region. However, it contained two conserved motifs in either side of its short C-terminus (C1 and C2, Figure 1a,c), which relates it to several other MYB proteins previously clustered as members of subgroup 4 (Kranz et al., 1998). Unlike a number of its homologues, FaMYB1 does not contain a zinc-finger motif in the C-terminus (Figure 1c). Some FaMYB1 homologues were previously suggested to act as transcriptional repressors or weak activators (AmMYB308 and AmMYB330, Tamagnone et al., 1998; Zm38, Franken et al., 1994; C1-I, Paz-Ares et al., 1990; PhMYB27, Mur, 1995; AtMYB4, Jin et al., 2000). Additional evidence was recently provided by Jin et al. (2000) that the motif pdLNLD/ELxiG/S (C2 motif in FaMYB1, Figure 1a,c) forms part of the region involved in the repression of transcription.

Recently, Grotewold et al. (2000) reported on the identification of amino acid residues in the maize transcriptional activator C1, which specify the interaction between the MYB domain of C1 and the N-terminal of the basic helix–loop–helix (bHLH) protein R. C1 depends on the interaction with R for its regulatory function in anthocyanin biosynthesis. With yeast two-hybrid experiments, the authors identified four residues in the first helix (all four residues must be present) and two additional residues in the second helix of the R3 repeat of C1, which are necessary for the interaction with R in plant cells (Figure 1b). The maize MYB protein P, closely related to C1, is not dependent on R for its activity and does not have the necessary interacting residues. Apart from C1, two additional MYB transcription factors, GL1 and AN2, previously shown to interact in yeast two-hybrid experiments with R, contain the required residues (Grotewold et al., 2000). FaMYB1 was the only protein amongst its closest homologues, which contained the six required amino acid residues (Figure 1b).

FaMYB1 is highly expressed in the red ripe strawberry fruit

FaMYB1 expression in both vegetative and reproductive strawberry plant tissues was analysed by RNA gel blot (Figure 2). Hybridization was performed using a probe derived from the 3′ region of the FaMYB1 cDNA (nucleotides 618–1005), in order to avoid cross-hybridization with other members of the MYB gene family in strawberry. FaMYB1 expression commenced at the turning stage, and reached the highest levels in the red ripe fruit stage. Low expression levels were detected in the flower and green fruit tissues. We also checked the expression of FaMYB1 in the achenes (seeds) and in the dark red over-ripe stage of fruit development (data not shown). No expression could be detected in achenes. However, a strong expression similar to that shown for red ripe fruit tissue was observed in over-ripe fruit tissue.

Figure 2.

FaMYB1 gene expression in strawberry.

RNA gel-blot analysis of the expression of FaMYB1 in different tissues of strawberry. The blot was hybridized with a fragment containing the 3′ region outside the conserved MYB domain of FaMYB1 and then re-hybridized for loading control with a strawberry cDNA probe showing homology to a gene encoding a ribosomal protein (RIBO). Quantification of the results was performed by measuring the OD per mm2 (using the tina software; Raytest, Straubenhardt, Germany) of the bands and calculating the ratios between the band in the gene specific gel blot and its corresponding band in the loading control blot (transcript ratio).

Tobacco plants overexpressing FaMYB1 and their progeny are affected in flower pigmentation

In order to ascertain a putative function for the strawberry FaMYB1, we used tobacco as a model system for heterologous expression experiments. FaMYB1 was expressed under the control of an enhanced 35S-CaMV promoter. Twenty-five primary transformant lines and three control wild-type lines were grown to maturity. One-third of the flowering transformed lines showed clear phenotypic changes in petal pigmentation, already visible at the early stages of flower development (prior to anthesis; Figure 3a). In several lines, alterations in petal colour resulted in a complete loss of pigmentation (severe phenotype). In other cases a distinct colour patterning was observed in which the main veins of the flower often remained red and a few regions were pigmented with pale red colour, while the remainder of the flower appeared white (medium phenotype). Apart from the reduction in petal pigmentation, the weak red colouring normally encountered in the distal part of the mature stamen filaments of wild-type tobacco was no longer visible in several transformed lines. Additionally, these lines showed premature browning of the mature flowers.

Figure 3.

Changes in tobacco flowers induced by expression of strawberry FaMYB1.

(a) Non-transformed (Nt), three FaMYB1 expressing lines exhibiting different phenotypic characteristics (strong, Tr-s; medium, Tr-m; weak, Tr-w), and a red (Tr-bcred) and a white (Tr-bcwhite) back-cross line [Tr-s line was back-crossed with a non-transformed (Nt) wild-type line] at four different flowering stages (1–4). Flowering plants showed a clear phenotypic change in petal pigmentation, already visible from early stages of flower development (prior to anthesis). In several plants, alterations in petal colour resulted in a complete loss of pigmentation. In other cases, colour patterning was observed in which the main veins of the flower became red while the rest of the flower appeared white.

(b) HPLC analysis of anthocyanin levels in FaMYB1 and non-transformed lines. Non-hydrolysed extracts of petals from stage 4 tobacco flowers were used. Tr-w-1, 2, 5, 6, 8, 9 lines are the progeny of the weak phenotype line (Tr-w) selected visually for analysis according to their flower pigmentation.

(c) FaMYB1 affects total quercetin levels in primary transformants and progeny. HPLC results of hydrolysed extracts from petals (stage 4 tobacco flowers) of primary transformants (Tr-s, Tr-m, Tr-w); a non-transformed line (Nt); two back-cross lines (Tr-bcred and Tr-bcwhite); and six lines from the progeny of the selfed weak phenotype primary transformant (Tr-w-1, 2, 5, 6, 8, 9). Bars indicate the levels (means and SD, n = 3 plants) of the flavonols quercetin and kaempferol relative (in percentage) to the mean levels in the non-transformed line (Nt).

To examine inheritance of the phenotype, two lines showing different phenotypic phenomena (strong, Tr-s; weak, Tr-w) were selected. The Tr-s line was back-crossed with a non-transformed wild-type line (Nt), and the Tr-w line and one Nt line (as a control) were self-fertilized. A similar phenotypic effect on flower colour was also detected in the progeny (Figure 3a). We analysed the changes in anthocyanin levels and their correlation with the severity of the FaMYB1 phenotype using HPLC. Flowers of control tobacco lines contain the anthocyanin cyanidin-3-rutinoside as their main pigment component (>90% of all anthocyanins), formed through the flavonoid biosynthetic pathway (Figure 7). We observed clear reductions in the levels of cyanidin-3-rutinoside in: (i) flowers of primary transformed lines (strong, Tr-s; medium, Tr-m; weak, Tr-w); (ii) flowers of Tr-s back-crossed progeny (white petal line, Tr-bcwhite); and (iii) flowers of Tr-w selfed progeny (Tr-w 1,2,5,6) (Figure 3b). The levels of reduction correlated very well with the phenotypic response observed by eye (Figure 4c).

Figure 7.

Effect of FaMYB1 overexpression on metabolic flux in flavonoid metabolism.

Cyanidin and pelargonidin are the main anthocyanin groups formed in tobacco flower and the cultivated strawberry fruit, respectively. Both are modified and accumulate in the vacuole. The enzymes catalysing the reactions are: PAL, phenylalanine ammonia-lyase; C4H, cinnamate 4-hydroxylase; 4CL, 4-coumaroyl-CoA ligase; CAD, cinnamyl alcohol dehydrogenase; CHS, chalcone synthase; CHI, chalcone flavanone isomerase; F3H, flavanone 3-hydroxylase; DFR, dihydroflavonol 4-reductase; ANS, anthocyanidin synthase; GT, glucosyl transferase; RT, rhamnosyl transferase; GST (an9), glutathione S-transferase; FLS, flavonol synthase; F3′H, flavonoid 3′-hydroxylase. Genes and enzymes showing reduced expression or activity in the FaMYB1-overexpressing lines are marked with (–); those that did not show any change in expression with (=). Metabolites with reduced levels in the FaMYB1-overexpressing lines are depicted in bold and underlined.

Figure 4.

Correlation between T-DNA insertions and the FaMYB1 phenotype.

(a) DNA gel-blot of primary transformants and a non-transformed line of genomic DNA digested with Xbal, which cuts in the T-DNA on one side of the FaMYB1 fragment. The FaMYB1 coding region fragment was used as a probe. Different amounts of T-DNA copies were detected in the FaMYB1 lines, while no T-DNA was present in the non-transformed line. Numbers at the left indicate the lengths of marker DNA fragments in kilobases.

(b) Gel blot of genomic DNA digested with Xbal from two back-cross lines (Tr-s crossed with non-transformed) showing the wild-type phenotype (Tr-bcred) and the FaMYB1 phenotype (Tr-bcwhite) and a line from a progeny of a selfed, non-transformed line (Nt). The FaMYB1 coding-region fragment was used as a probe. Numbers at the left indicate the lengths of marker DNA fragments in kilobases.

(c) Gel blot of genomic DNA digested with Xbal from nine lines (Tr-w, 1–9) out of a selfed progeny of the primary transformant showing the weak phenotype (Tr-w) and two lines from a selfed progeny of a non-transformed line (Nt1 and 2). Tr-w contains a single T-DNA insertion as seen in (a). The FaMYB1 coding region fragment was used as a probe. Tr-w, 1–9 lines were selected visually for analysis according to their flower pigmentation phenotype.

HPLC analyses reveal specific reduction in the flavonol quercetin in the petals of FaMYB1-expressing lines

In addition to anthocyanins, tobacco flowers accumulate the flavonols quercetin and kaempferol, in the form of various glycosides. These flavonols may function as co-pigments for the anthocyanins, and also as sunscreens. Together with the anthocyanins, they share the compound dihydrokaempferol as a common precursor (Figure 7).

HPLC analyses of flavonol-glycosides revealed a clear reduction in the levels of quercetin-glycosides in the primary transformant lines and in the back-crossed and selfed progeny displaying the FaMYB1 phenotype, while levels of kaempferol-glycosides were not altered (data not shown). We did not detect any new flavonol glycoside in the lines showing the FaMYB1 phenotype. Flowers of control and transgenic lines and their progeny were also compared for total kaempferol and quercetin levels, after acid hydrolysis of extracts (Figure 3c). A marked reduction in quercetin levels was observed in the primary transformant lines (Tr-s, Tr-m, Tr-w), and in the back-crossed progeny of Tr-s (Tr-bcwhite) and selfed progeny of Tr-w (Tr-w 1,2,5,6), while kaempferol levels remained unchanged. Reduction in quercetin levels correlated well with the reduction in anthocyanin levels in these lines (Figure 3b).

The phenotype correlates with FaMYB1 T-DNA insertions

DNA gel-blot analysis was performed to confirm the transgenicity (to determine the number of T-DNA inserts) of the primary transformant lines and the back-crossed and selfed progeny of the Tr-s and Tr-w lines, respectively. The Tr-w line showing the weak phenotype had a single insertion, while the lines showing a medium (Tr-m) phenotype and a strong (Tr-s) phenotype contained at least two (possibly three in the case of Tr-s) FaMYB1 T-DNA insertions. Weak bands were observed in the non-transformed line (Nt), but the same bands appeared in all samples (e.g. the 6.5 kb band, Figure 4a) and were probably due to cross-hybridization with endogenous tobacco MYB genes.

The Tr-s back-crossed progeny line (Tr-bcred), showing full red coloured petals, and the Tr-bcwhite line, showing completely white petals, were subjected to DNA gel-blot analysis (Figure 4b). A clear association between the presence of the two bands corresponding to the two FaMYB1 insertions in the Tr-s parent (12.0 kb and 4.8 kb) and petal colour is shown. The segregation of the phenotype was also investigated in the Tr-w selfed progeny (Tr-w 1–9), each containing a single FaMYB1 insertion. As was the case for the back-crossed progeny, the DNA gel-blot results of the Tr-w progeny correlated well with the phenotype and demonstrated the dominant effect of FaMYB1 (Figure 4c).

FaMYB1 expression levels in petals and leaves of primary transformed lines were investigated by RNA gel-blot analysis (Figure 5). Relatively high FaMYB1 expression levels were found in lines with the strong phenotype, and relatively low expression levels were observed in lines with the weak phenotype. We could therefore conclude that the strength of the FaMYB1 phenotype in the transgenic lines was dependent on the relative expression levels of the introduced FaMYB1 gene.

Figure 5.

Correlation between FaMYB1 expression in transformants and severity of the phenotype.

Total RNA from leaves and petals (stage 4 tobacco flowers) of primary transformants (Tr-s, Tr-m, Tr-w), a non-transformed line (Nt), and ripe fruit stage strawberry tissue (St) were used for RNA gel-blot analysis. The blot was hybridized with a FaMYB1 probe, then rehybridized for loading control with a petunia cDNA probe showing homology to a gene encoding ubiquitin (UBI). Quantification of the results was performed by measuring the OD per mm2 (using the tina software; Raytest, Straubenhardt, Germany) of the bands and calculating the ratios between the band in the gene-specific gel blot and its corresponding band in the loading control blot (transcript ratio).

Gene expression and enzymatic activities of late flavonoid biosynthetic pathway are affected in the FaMYB1 transgenic tobacco lines

The effect of FaMYB1 expression on various genes and enzymes of flavonoid, general phenylpropanoid and lignin metabolism was examined in flowers by RNA gel blots and enzymatic activity assays. Expression of chalcone synthase (CHS) and flavanone 3-hydroxylase (F3H) genes, encoding enzymes active in the upper part of the flavonoid biosynthetic pathway (up to the formation of dihydrokaempferol), were unaffected by FaMYB1 overexpression (Figures 6a and 7). In addition, no alteration of the expression of genes encoding enzymes related to general phenylpropanoid metabolism (phenylalanine ammonia-lyase, PAL; cinnamate 4-hydroxylase, C4H; 4-coumaroyl-CoA ligase, 4CL) and lignin metabolism (cinnamyl alcohol dehydrogenase, CAD) was observed.

Figure 6.

Effect of FaMYB1 overexpression on gene expression and enzymatic activity in different branches of phenylpropanoid metabolism.

(a) Total RNA from petals (stage 4 tobacco flowers) of primary transformants (Tr-s, Tr-m, Tr-w) and a non-transformed line (Nt) was used for RNA gel blot. The blots were hybridized with different probes corresponding to genes encoding the enzymes of flavonoid metabolism [CHS, F3H, DFR, ANS, GST (AN9), RT], general phenylpropanoid metabolism (PAL, C4H, 4CL) and lignin metabolism (CAD). Each of the blots was checked for RNA amounts by rehybridizing with either a petunia cDNA probe showing homology to a gene encoding ubiquitin (UBI) or a potato ribosomal RNA probe (RIBO). Quantification of the results was performed by measuring the OD per mm2 (using the tina software, Raytest, Straubenhardt, Germany) of the bands and calculating the transcript ratio between the band in the gene specific gel blot and its corresponding band in the loading control blot (transcript ratio).

(b) Flavonoid-UDP-glucose transferase enzyme activity is decreased in flowers of FaMYB1 expressing tobacco lines. On each experimental day enzyme extracts were simultaneously prepared from flowers of control lines and FaMYB1-transformed lines (six flowers per extract, two replicates) and tested for specific UDP-glucose transferase activity with the flavonoid substrates indicated. For each flavonoid tested, bars indicate glucosyl transferase activities in transformed lines relative to the activities in control lines (Nt = 100%) and represent means ± SD of different independent experiments (quercetin, n = 3; kaempferol, n = 2; cyanidin, n = 1).

In contrast, the gene encoding anthocyanidin synthase (ANS), an enzyme from the lower end of the flavonoid pathway, was significantly affected in its expression when compared with the non-transformed controls (Figure 6a). The effect on another gene from the lower part of the pathway, encoding dihydroflavonol 4-reductase (DFR), is uncertain as a modest reduction in DFR expression was observed only in the Tr-s line.

Enzyme assays conducted on protein extracts from flowers of the Tr-s and wild-type tobacco (Nt) lines were performed in order to examine the effect on flavonoid-UDP-glucosyl transferase (GT) enzyme activity (Figure 6b). GT is also active at the lower end of the flavonoid pathway, and showed a significant and reproducible reduction in specific activity (3–4-fold) in the Tr-s line compared to the NT line. This reduction in activity was observed when quercetin, kaempferol or cyanidin were used as substrates. The result implied a possible repression of GT gene expression. Gene expression of UDP-rhamnosyl transferase (RT), encoding another sugar-transfer enzyme, was not altered in the FaMYB1-expressing lines. The RT enzyme adds a rhamnose group to the anthocyanidin glucoside molecule produced by GT, to generate anthocyanidin rutinosides.

Another crucial step in anthocyanin metabolism that might be affected by FaMYB1 is the export of anthocyanins from their site of synthesis in the cytoplasm to their site of permanent storage in the vacuole. Glutathione S-transferases (GSTs) are assumed to be involved in this activity, either by conjugating anthocyanin to the tripeptide GSH (–Glu–Cys–Gly) or by binding the anthocyanins and ‘escorting’ them for sequestration without conjugate formation (Mueller et al., 2000). As anthocyanin-GSH conjugates were never demonstrated (both in vitro and in vivo), we used the common substrate for most GSTs, 1-chloro-2,4-dinitrobenzene (CDNB), for analysing GST activity. The results did not show any difference in CDNB glutathionation between FaMYB1-overexpressing lines and the controls. RNA gel-blot analysis using the petunia flavonoid glutathione S-transferase homologue (An9) as a probe showed no significant changes in GST expression in the FaMYB1 transgenic lines.

Our results suggest that the main effect of FaMYB1 overexpression is the repression of the expression of genes at the lower end of the flavonoid biosynthetic pathway, more directly related to the biosynthesis of anthocyanins and the flavonol quercetin.

FaMYB1 interacts with known regulators of anthocyanin biosynthesis and does not contain a functional activation domain in yeast two-hybrid system

In order to examine whether FaMYB1 can interact with known regulators of anthocyanin metabolism and/or act as a transcriptional activator, we performed two-hybrid assays in yeast. The yeast strain PJ69-4A was transformed with the FaMYB1 gene-coding region fused to either the Gal4 activation domain (Gal4AD) or the Gal4 DNA-binding domain (Gal4BD). The coding regions of four Petunia genes encoding known regulators of anthocyanin, namely JAF13[Quattrocchio et al., 1998, basic helix–loop–helix (bHLH) protein]; AN1 (Spelt et al., 2000, bHLH protein); AN2 (Quattrocchio et al., 1998, MYB domain protein); and AN11 (De Vetten et al., 1997, WD40 repeat protein), were fused to the Gal4AD and used for co-transformation with the Gal4BD–FaMYB1 construct. Co-transformation with the Gal4BD–FaMYB1 construct and the Gal4AD was performed in order to test whether Gal4BD–FaMYB1 contains an activation domain. The Gal4AD–FaMYB1 and Gal4BD were co-transformed as a control, to check whether FaMYB1 can bind to the GAL4 promoter. The results indicated that in yeast, FaMYB1 does not contain a functional activation domain (Figure 8). However, FaMYB1 did interact with the two bHLH proteins (JAF13 and AN1), while no binding was observed between FaMYB1 and the MYB protein AN2 and the WD40 protein AN11 (Figure 8). The AN1 and JAF13 proteins were previously demonstrated to be factors necessary for anthocyanin synthesis in petunia tissues. This interaction validates the structural integrity of the FaMYB1 fusion and thus confirms that FaMYB1 was properly expressed. The Gal4AD–FaMYB1 experiment showed that FaMYB1 itself did not activate the Gal4 promoter.

Figure 8.

FaMYB1 interacts with known anthocyanin regulators in yeast two-hybrid assays.

Results of the yeast two-hybrid assays on a selection plate lacking histidine with double-transformed yeast strain PJ96-4A. Two independent transformants were used for each of the six construct combinations (below). Construct combinations 1 and 2 were performed in order to check whether FaMYB1 can bind to the GAL4 promoter and if FaMYB1 contains a functional activation domain in yeast, respectively.

Discussion

In this paper we have shown that constitutive overexpression in tobacco of FaMYB1, a MYB transcription factor isolated from red ripe strawberry fruit, resulted in severe phenotypic changes in the transformed lines. HPLC analysis revealed a dramatic decrease in the levels of anthocyanins and of the flavonol quercetin in the flowers. The phenotype was inherited as a dominant trait to the progeny, and its severity correlated well with FaMYB1 transgene expression levels. This suggested that the phenotype was in direct correlation with the cellular concentration of FaMYB1.

The expression of genes encoding enzymes catalysing reactions in the upper part of the flavonoid pathway, general phenylpropanoid and lignin metabolism were unaffected by FaMYB1 overexpression. However, expression of the ANS gene and activity of flavonoid-UDP-glucose transferase enzyme, which suggests a possible reduction in expression of the GT gene, both active in the lower end of the flavonoid pathway, were significantly reduced in all transgenics showing the phenotype.

The specific reduction in quercetin compared to kaempferol in the FaMYB1-overexpressing tobacco lines suggests a reduction in the activity of the enzyme flavonoid 3′-hydroxylase (F3′H). F3′H catalyses the reaction forming dihydroquercetin, which is the direct precursor of both quercetin and cyanidin (Figure 7). Our attempts to analyse F3′H gene expression and enzyme activity proved unsuccessful, and as a result we could not verify whether F3′H is affected in the FaMYB1-expressing plants. These results suggest that the strawberry transcription factor FaMYB1 affects several metabolic steps at the lower end of the flavonoid biosynthetic pathway, by reducing mRNA levels of genes encoding enzymes involved in anthocyanin and flavonol biosynthesis.

In addition to the phenotypic effect in the FaMYB1 overexpressing plants, several other lines of reasoning lead us to believe that FaMYB1 may function in strawberry as a transcriptional regulator of genes related to flavonoid metabolism. Many MYB proteins from different plant species have previously been shown to take part in the regulation of the phenylpropanoid metabolism and branch pathways such as the flavonoid biosynthetic pathway (Weisshaar and Jenkins, 1998). In the flavonoid pathway itself, MYB proteins have been identified as key regulators of the metabolism of different groups of compounds such as phenolic acids, flavonols, and the pigments anthocyanin and phlobaphenes (Martin and Paz-Ares, 1997). FaMYB1 showed the highest expression levels during the ripening stage of strawberry fruit, when anthocyanin levels reach their maximum. The importance of anthocyanin biosynthesis at this stage of development was further supported by the fact that out of 1100 ESTs from a red ripe strawberry fruit library, we identified six putative genes encoding enzymes involved in the metabolism of anthocyanins (Aharoni et al., 2000). Identifying a putative regulator for the flavonoid pathway (represented twice in our EST collection) was therefore not unexpected.

Additional evidence for the association of FaMYB1 with anthocyanin metabolism arose from two-hybrid assays in yeast, which indicated that FaMYB1 could interact with known bHLH anthocyanin regulators previously identified in Petunia, AN1 and JAF13. Both AN1 and JAF13 are involved in the regulation of the lower part of the flavonoid pathway, although they are functionally and evolutionary distinct (different affinities for partner proteins and/or target DNA sequences) (Spelt et al., 2000). The presence of all six amino acid residues required for the interaction with the known bHLH anthocyanin regulator R in the R3 MYB repeat of FaMYB1, in particular the glycine–arginine pair (GR), strengthens the possibility that the interaction observed in the yeast two-hybrid system in this study may occur in planta. However, the interaction with the R regulator might not always associate a protein to a role in flavonoid metabolism, as in the case of the Arabidopsis GL1 protein that interacts with R, but does not play a role in flavonoid biosynthesis (Payne et al., 2000).

DNA-binding transcriptional repressors act by a variety of mechanisms, including repression mediated by passive steric hindrance mechanisms or active repression (Hanna-Rose and Hansen, 1996). Active repression is thought to involve inhibitory protein–protein interactions with components of the basal transcriptional machinery (e.g. direct repression model) or positive transcriptional regulators (e.g. quenching model), or possibly induction of inactive chromatin structure at the regulated promoter. It might also involve recruitment of additional proteins as co-activators and co-repressors. In contrast to activators, each repressor generally contains a single small repression domain. Each type of repression motif might represent a unique interaction surface for contacting a particular target within the basal or regulatory transcriptional machinery, thus resulting in repression.

Information on repression mechanisms in plants is limited (Schwechheimer and Bevan, 1998), and the only R2R3 MYB protein identified as a repressor, containing a putative repressor domain, was reported recently (AtMYB4; Jin et al., 2000). AtMYB4 was suggested to act both by direct repression and by competition with activators on binding motifs located on the promoters of target genes. Both the C-terminus of the protein and a region in it containing the peptide sequence NLELRISLPDDV were shown to be required for the repression activity. The same motif (pdLNLD/ELxiG/S) is also present in the C-terminus of FaMYB1 and other members of the MYB protein subgroup 4 (Kranz et al., 1998).

Structural homology between MYB proteins from different plant species might point to general similarity in the pathways regulated by them, and in the type of regulation (activation or repression). This was shown recently for the Arabidopsis AtMYB75 (termed PAP1; Borevitz et al., 2000) that was demonstrated to act as an activator of anthocyanin biosynthesis genes, as predicted earlier from its sequence similarity to known anthocyanin biosynthesis regulators from other species (both at the MYB domain and the C-termini) (Jin and Martin, 1999; Quattrocchio et al., 1999). The overall structural homology between FaMYB1 and other MYB proteins from different plant species containing the pdLNLD/ELxiG/S motif, and the fact that several of them were proposed to act as transcriptional repressors, suggests that FaMYB1 might function as a true repressor in strawberry.

Several mechanisms could be proposed to explain how FaMYB1 acts to reduce transcription of late flavonoid biosynthesis genes in tobacco. One possibility is that FaMYB1 is a repressor, acting similarly to AtMYB4 and recognizing its normal target genes, which are different from those recognized by AtMYB4 (different target site selectivity). The comparison between the DNA-binding domains of FaMYB1 and AtMYB4 (Figure 1b) shows that 90 residues out of 104 are identical or contain conserved amino acid substitutions, and 14 have non-conserved substitutions, of which two are amongst the four amino acids identified as necessary for protein interaction. The remaining 12 residues might be important for the difference in target site selectivity between the FaMYB1 and AtMYB4 proteins. In this case ANS, GT and possibly DFR might be the primary target sites for FaMYB1. Another possibility could be that FaMYB1, in high concentrations, might bind non-specifically to target sites without the capability to activate gene transcription and therefore may act by hindering the function of a tobacco regulator. Repression might also be an indirect effect in which FaMYB1 binds non-specifically to another tobacco transcription factor that would be inhibited in its function (Tamagnone et al., 1998). Another indirect effect could be the titration of an endogenous bHLH protein which, similarly to the R protein from maize, normally interacts with an endogenous tobacco MYB counterpart and controls expression of anthocyanin biosynthesis genes (Glover et al., 1998; Payne et al., 1999).

Micro-array gene-expression data showed that transcripts of ANS and GT accumulate co-ordinately during strawberry ripening, and attain their maximum at the turning stage of development (Aharoni et al., 2000). No significant difference in expression of these genes between the turning and red stages of development could be detected. Repression of anthocyanin and flavonol-related genes would be an efficient way to control metabolic flux throughout the phenylpropanoid pathway during strawberry fruit development. Reducing the levels of anthocyanins in late strawberry fruit ripening (when fruit are entirely red) would be beneficial in negating their potential toxic effects on the plant cell (Mueller et al., 2000), and would substantially reduce the carbon and energy consumption needed for their biosynthesis and transport into the vacuole.

Co-ordinated reduction in expression of anthocyanin pathway genes on late development was demonstrated in flowers of Petunia (Mur, 1995). In corollas of red Petunia flowers, flavonoid gene expression starts at a very early stage and declines around anthesis. Interestingly, in the same tissue the gene PhMYB27, encoding a MYB protein homologue, shows high levels of expression when flavonoid gene expression ceases. Apart from being a relatively short protein (184 amino acids) with no obvious acidic activation domain, PhMYB27 shows homology to other members of the MYB subgroup 4 proteins (Mur, 1995). It also contains a complete LlsrGIDPxT/SHRxI/l motif and a partial pdLNLD/ELxiG/S motif outside the MYB domain. PhMYB27 was shown to be expressed in cells that accumulate flavonols or anthocyanins, including leaves, seed coat, seedlings and different floral tissues. Further evidence that PhMYB27 is involved in the regulation (possibly in repression) of flavonoid biosynthesis was supported by the fact that it is controlled by the anthocyanin regulatory loci An1, An2, An4 and An11. AN1 (shown in this study to interact in yeast with FaMYB1) directly activates the expression of the dfrA gene and of PhMYB27 (Spelt et al., 2000).

It is possible that FaMYB1 and PhMYB27 may have a similar function: that is, they may act as transcriptional repressors of late flavonoid biosynthesis genes in strawberry fruit and Petunia flowers, respectively. One way to verify this would be to overexpress or downregulate FaMYB1 in strawberry and PhMYB27 in a red petunia variety. It might be also of interest to analyse the putative repression motif (present in MYB subgroup 4, including in FaMYB1 and PhMYB27; pdLNLD/ELxiG/S) by introducing it to the C-terminus of a known anthocyanin transcriptional activator, and analysing the overexpression effect in transgenic plants.

Experimental procedures

We used the strawberry (Fragaria × ananasa) cultivar Elsanta and tobacco (Nicotiana tabacum) cultivar SR1. Transformation of tobacco was performed as described by Horsch et al. (1985) using the Agrobacterium tumefaciens strain AGL0 (Lazo et al., 1991). Plants were kept in a containment greenhouse with a 16 h photoperiod and a 21°C/17°C day/night temperature.

FaMYB1 overexpression construct

A 736 bp fragment including the FaMYB1 ORF was inserted in a sense orientation to the pFLAP10 subcloning vector using Xbal and Ncol restriction sites introduced to the fragment by PCR at the 5′ and 3′ ends, respectively. The fragment was inserted between a double 35S-CaMV promoter and a nopaline synthase terminator. From the pFLAP10 vector the fragment was excised with Pacl and Ascl restriction digestions and introduced to the pBinPlus binary vector containing a kanamycin resistance gene inside the T-DNA for selection of transformants (Engelen et al., 1995).

Isolation of nucleic acids from strawberry and tobacco and gel-blot analyses

Total RNA was isolated from different strawberry tissues as described by Schultz et al. (1994), and from tobacco leaf and petal material as described by Verwoerd et al. (1989). For the isolation of genomic DNA from tobacco, we used the protocol described by Doyle and Doyle (1990). For DNA gel-blot analyses, aliquots of 5 µg tobacco genomic DNA were digested with Xbal (cuts once outside the FaMYB1 fragment in the T-DNA) and separated on a 0.7% TAE (0.04 m Tris–acetate pH 8.0, 1 mm EDTA pH 8.0) agarose gel. The DNA was then depurinated in 0.4 m NaOH for 30 min and transferred to a Hybond N+ membrane (Amersham, Buckinghamshire, UK) in 0.4 m NaOH. After fixation (2 h at 80°C) blots were hybridized as described by Angenent et al. (1992). The hybridization probes were made by random labelling oligonucleotide priming (Feinberg and Vogelstein, 1984) of the entire FaMYB1 cDNA. Washing was performed under low stringency conditions [two times for half an hour each in 2 × SSC at 52°C (1 × SSC = 0.15 m NaCl, 0.015 m sodium citrate) and 0.1% SDS] and further under stringent conditions (65°C, 0.1 × SSC and 0.1% SDS twice every half hour). RNA gel-blot analyses (strawberry and tobacco) were performed as described previously (Aharoni et al., 2000), except that the region outside the FaMYB1 MYB domain was used as a probe in order to avoid cross-hybridization. For analysing expression of different flavonoid genes, we used cDNAs of tobacco (C4H and 4CL), petunia (PAL, An9, RT, F3′H, ANS, DFR), tomato (F3H) and strawberry (CHS, CAD) as probes.

Enzymatic activity assays

For each experiment, enzyme extracts were prepared by isolating and pooling petal limbs from six flowers [developmental stages 2, 3 and 4 (Figure 3a); two flowers per stage per plant, in triplicate]. For flavonoid-UDP-glucosyl transferase activity, extracts were prepared by grinding flower tissue in ice-cold 0.2 m potassium phosphate buffer pH 7.5 containing 10 mm DTT, 0.5 g DOWEX 1 × 2–100 anion exchanger and quartz sand. After centrifugation (10 min, 14 000 g), enzyme reactions were started by mixing up to 150 µl supernatant with 25 µl 4 mm flavonoid substrate (dissolved in ethanol) and 25 µl of 12 mm UDPG. Reactions were incubated for 30 min at 30°C and stopped by adding 800 µl of a chloroform/methanol (2 : 1) mixture. Flavonoid glycosides in the aqueous layer were detected by HPLC. For glutathione-S-transferase activity, petal tissues were ground with quartz sand in 5 ml 0.1 m sodium phosphate pH 7.4 containing 1 mm EDTA, and 0.25 g polyvinylpyrrolidone. After centrifugation (10 min, 17 500 g) extracts were desalted using Sephadex G-25. Enzyme activity was recorded at 30°C, in a 1 ml cuvette with 1 mm GSH and 1 mm chloro-2,4-dinitrobenzene at pH 6.5.

HPLC analyses of flavonoids

Petals of mature flowers (at least three per line) were used. Flavonoids were determined as aglycons or their glycosides by preparing hydrolysed and non-hydrolysed extracts, respectively. Hydrolysed extracts were prepared by heating 0.15 g of petal tissue in 2 ml of an acid aqueous methanol (MeOH, HPLC quality) solution, consisting of 50% MeOH, 0.16% ascorbic acid, 0.16% t-butylhydroquinon (TBHQ) and 1.2 m HCl, at 90°C for 1 h. After hydrolysis, the extracts were diluted with 2 ml 100% MeOH and sonicated for 5 min. Non-hydrolysed extracts were prepared by sonicating 0.15 g petal tissue in 1.5 ml 75% MeOH, 0.1% trifluoroacetic acid (TFA) and 0.1% TBHQ for 15 min. All extracts were filtered over a 0.2 µm Teflon filter before analysis by HPLC. Flavonoids were separated on a 150 × 3.9 mm NovaPak C18 column (Waters Chomatography, Etten-Leur, The Netherlands), using isocratic conditions (25% acetonitril in 0.1% TFA; flow rate 0.9 ml min−1) for flavonol aglycons, and a gradient of 5–50% acetonitril in 0.1% TFA for anthocyanins and flavonol glycosides, and detected with a photodiode array detector (Waters 996).

Yeast two-hybrid screen

The PJ69-4a yeast strain was used, which possesses the genotype MATa trp-901 leu2-3, 112 ura3-52 his3-200 gal4D gal80D GAL-ADE2 LYS::GAL1-HIS3 met::GAL7-lac Z (James et al., 1996). All yeast transformations were performed according to the lithium acetate method (Gietz and Schiestl, 1995). The ORFs of AN1, AN2, AN11 and JAF13 genes (containing the EcoRl and Xhol restriction sites generated by PCR) were introduced in-frame with the GAL4 activation or binding domains in the pAD and pBD vectors, respectively (Stratagene, La Jolla, CA, USA). The FaMYB1 ORF was introduced to the same vectors as described above using Mfel and Xhol restriction sites introduced by PCR at the 5′ and 3′ ends, respectively. The yeast transformants were tested for interaction/activation on media without histidine or histidine and adenine.

Sequence analyses

Sequencing was performed using the Applied Biosystems (Foster City, CA, USA) dye-terminator cycle-sequencing ready reaction kit and the Applied Biosystems 310 DNA sequencer. Comparison and analyses of the sequences were conducted with the advanced basic local alignment search tool, blast (Altschul et al., 1990) at the National Center for Biotechnological Information (http://www.ncbi.nlm.nih.gov) Nonredundant Protein Database. Software used for DNA and protein analysis was the dnastar program (DNAstar Inc., Madison, WI, USA).

Acknowledgements

We are grateful to David Weiss, Cathie Martin and Francesca Quattrocchio for providing the cDNA clones; Arnaud Bovy for providing the pFLAP10 vector; P. James for the PJ69-4A yeast strain; Dirk Bosch for comments and advice on the manuscript; and Geert Scholten for his care of the plants.

Genbank accession number Af401220.

Ancillary