Strawberry (Fragaria × ananassa) fruits contain high concentrations of flavonoids. In unripe strawberries, the flavonoids are mainly represented by proanthocyanidins (PAs), while in ripe fruits the red-coloured anthocyanins also accumulate. Most of the structural genes leading to PA biosynthesis in strawberry have been characterized, but no information is available on their transcriptional regulation. In Arabidopsis thaliana the expression of the PA biosynthetic genes is specifically induced by a ternary protein complex, composed of AtTT2 (AtMYB123), AtTT8 (AtbHLH042) and AtTTG1 (WD40-repeat protein).
A strategy combining yeast-two-hybrid screening and agglomerative hierarchical clustering of transcriptomic and metabolomic data was undertaken to identify strawberry PA regulators.
Among the candidate genes isolated, four were similar to AtTT2, AtTT8 and AtTTG1 (FaMYB9/FaMYB11, FabHLH3 and FaTTG1, respectively) and two encode putative negative regulators (FaMYB5 and FabHLH3∆). Interestingly, FaMYB9/FaMYB11, FabHLH3 and FaTTG1 were found to complement the tt2-1, tt8-3 and ttg1-1 transparent testa mutants, respectively. In addition, they interacted in yeast and activated the Arabidopsis BANYULS (anthocyanidin reductase) gene promoter when coexpressed in Physcomitrella patens protoplasts.
Taken together, these results demonstrated that FaMYB9/FaMYB11, FabHLH3 and FaTTG1 are the respective functional homologues of AtTT2, AtTT8 and AtTTG1, providing new tools for modifying PA content and strawberry fruit quality.
The popularity of strawberry as a fruit crop is mainly a result of its unique aroma, sweet taste, bright colour and nutritional value. These quality traits are largely determined by the metabolic composition of the fruit. For example, strawberry fruits are known as a valuable source of polyphenol compounds. The main polyphenol compounds found in strawberry fruits are flavonoids. About 70% of the total phenolics that accumulated in cv Queen Elisa ripe fruits correspond to proanthocyanidins (PAs), whereas anthocyanins, flavonols and the other phenolics (i.e. p-coumaric acid, ellagic acid) represent c. 20, 3 and 7%, respectively (Almeida et al., 2007; Carbone et al., 2009). Flavonoids have potential human health beneficial properties. They may act as antioxidants or as signalling molecules and were shown to be able to modulate some cellular functions (Nijveldt et al., 2001; Battino et al., 2009). Flavonoids such as PAs are also known to impact fruit flavour and astringency (Akagi et al., 2011).
The flavonoid biosynthetic pathway has been extensively studied in numerous plant species at the genetic, biochemical and molecular levels, and was recently described in strawberry (Lepiniec et al., 2006; Almeida et al., 2007; Carbone et al., 2009). This pathway (Fig. 1a) is initiated by the condensation of one molecule of 4-coumaroyl-CoA with three molecules of malonyl-CoA by the chalcone synthase (CHS) enzyme. The successive steps catalysed by the chalcone isomerase (CHI), flavonoid 3-hydroxylase (F3H/FHT), flavonoid 3′-hydroxylase (F3′H), dihydroflavonol-4-reductase (DFR), anthocyanidin synthase (ANS), and 3-glycosyltransferase (3-GT) enzymes lead to the production of the anthocyanin pigments. The F3H/FHT enzyme products dihydrokaempferol and dihydroquercetin are also the substrate of flavonol synthase (FLS) and lead, together with the involvement of several glycosyltransferases (GTs), to the production of flavonols. The products of the enzymatic reactions catalysed by DFR and ANS can also be converted into PA precursors (before polymerization/oxidation) by the leucoanthocyanidin reductase (LAR) and anthocyanidin reductase (ANR) enzymes, respectively. In strawberry, these two last steps produce both catechin- and epicatechin-based PAs that are found in different tissues (e.g. fruits, leaves or stems). Interestingly, Arabidopsis lacks a functional LAR gene and therefore produce exclusively epicatechin-based PA, which are limited to the seed coat (Abrahams et al., 2003; Xie et al., 2003; Routaboul et al., 2006).
In Arabidopsis, the transcriptional regulation of PA biosynthesis has been extensively studied. Using a genetic approach, based on the study of the transparent testa (tt) phenotype observed in mutant seeds, three regulatory proteins were identified as necessary for the production of PA in the seed coat, namely AtTT2/AtMYB123 (R2R3-MYB, Nesi et al., 2001; Dubos et al.,2010), AtTT8/AtbHLH042 (R/b-like bHLH; Nesi et al., 2000) and AtTTG1 (WD40-repeat protein; Walker et al., 1999). These three regulators were shown to act together in a ternary regulatory MYB-bHLH-WD40 (MBW) protein complex in order to regulate their target genes (e.g. BAN/ANR; Baudry et al., 2004). For fruit crops there is limited information available on transcriptional regulation of the PA biosynthesis pathway. In grape, three MYB TF genes have been identified which show homology to AtTT2 and promote PA biosynthesis early during grape berry development. The grape TFs VvMYB5b and VvMybPA2 are involved in the regulation of the PA pathway in skin tissue of young berries (Deluc et al., 2008; Terrier et al., 2009), whereas VvMybPA1 is mainly active in seeds (Bogs et al., 2007). For persimmon, a MYB TF, DkMyb4, was isolated which was shown to be involved in regulation of PA biosynthesis at an early stage of fruit development (Akagi et al., 2009). DkMyb4 was expressed in fruit flesh of astringent fruit types, but its expression was reduced in nonastringent persimmon fruits, resulting in down-regulation of PA biosynthesis and the resultant nonastringency trait.
New tools specifically dedicated to strawberry have recently been developed, including the recent availability of the whole genome sequence, expressed sequence tag (EST) collections, Agrobacterium-mediated transformation protocols, loss-of-function mutant collections, and protocols for large-scale metabolite analysis (Bombarely et al., 2010; Buendia et al., 2010; Oosumi et al., 2010; Shulaev et al., 2011; Aaby et al., 2012). Despite the fact that these tools may substantially help to improve our knowledge on strawberry fruit growth and development, little is still known about the regulation of flavonoid biosynthesis during the fruit-ripening process.
In order to get more insight into the regulation of PA biosynthesis during strawberry fruit development, a comparative genomics strategy was used, with the aim of identifying new PA-specific transcriptional regulators. We first found that AtTT2 expression in strawberry fruit was sufficient to redirect flavonoid biosynthetic pathway from anthocyanin to PAs. This result suggests that functional homologues of the AtTT8 and AtTTG1 (and also AtTT2) transcriptional regulators are present in strawberry. A green fruit cDNA library was thus screened using both yeast- and PCR-based approaches, in order to identify new PA-specific regulatory genes. Metabolic and expression data, recorded during strawberry fruit development in 14 cultivars, were analysed using the agglomerative hierarchical clustering method (Vilo et al., 2003). This approach revealed that nearly all the putative regulators identified were specifically expressed in the early phases of fruit development, when the accumulation of the PA biosynthetic gene transcripts and PA metabolites were the highest. Complementation experiments of Arabidopsis transparent testa null mutants (i.e. tt2-1, tt8-3 and ttg1-1), coupled with transient expression assay in Physcomitrella patens protoplasts, led to the characterization of four strawberry regulatory proteins which are the functional homologues of AtTT2, AtTT8 and AtTTG1, namely FaMYB9/FaMYB11, FabHLH3 and FaTTG1. The hypothetical function of two additional strawberry transcription factors (TFs; FaMYB5 and FabHLH3∆) that may act as negative regulators of PA biosynthesis is also discussed.
Materials and Methods
Plants from different strawberry (Fragaria × ananassa Duch.) cultivars were grown in the field at the experimental farm ‘the Santacker’, near Elst in the fruit-growing area in the Netherlands. Fruits at different developmental stages were harvested in June in three successive years and frozen in liquid nitrogen immediately upon harvesting.
Liquid chromatography quadrupole time-of-flight tandem mass spectrometry (LC-QTOF-MS) analysis of semipolar strawberry fruit compounds
The extraction and the LC-QTOF-MS analysis of semipolar compounds were performed according to the protocol for tomato fruit, as described in Moco et al. (2006). In short, 0.5 g frozen strawberry fruit powder (FW) of each of the 94 cultivars was extracted with 1.5 ml formic acid : methanol (1 : 1000, v/v) solution. The extracts were sonicated for 15 min and filtered through a 0.2 μm inorganic membrane filter (Anotop 10; Whatman, Maidstone, UK).
An LC-QTOF Ultima-MS platform was used for the profiling of the extracts. This platform consisted of a Waters Alliance 2795 HT HPLC system equipped with a Luna C18(2) pre-column (2.0 × 4 mm) and an analytical column (2.0 × 150 mm, 100 Å, particle size 3 μm; Phenomenex, Torrance, CA, USA) connected to an Ultima V4.00.00 QTOF mass spectrometer (Waters-Corporation, MS technologies, Manchester, UK). Degassed solutions of formic acid : ultra pure water (1 : 1000, v/v – eluent A) and formic acid : acetonitrile (1 : 1000, v/v – eluent B) were pumped into the HPLC system at 190 μl min−1 and the gradient linearly increased from 5 to 35% eluent B over a 45 min period, followed by 15 min of washing and equilibration of the column. The column, sample and room temperatures were kept at 40, 20 and 20°C, respectively. Ionization was performed using an electrospray ionization (ESI) source and masses were detected in positive mode. A collision energy of 5 eV was used for full-scan LC-MS in the range of m/z = 100–1500. Leucine encephalin was used for online mass calibration (lock mass).
PA extraction and determination using phloroglucinol
The procedure was performed according to Kennedy & Jones (2001) with slight modifications in the HPLC conditions, as previously published in Almeida et al. (2007). Phloroglucinol adducts were analysed by reverse-phase high-performance liquid chromatography (HPLC) with photodiode array detection (280 nm) and fluorescence detection (excitation at 275 nm, emission at 310 nm). (+)-Catechin and (−)-epicatechin, purchased from Sigma, were used as standards for quantification.
For overexpression of AtTT2, AtTT8 and AtTTG1, the corresponding open reading frames (ORFs) were introduced into the pCaMV35S expression cassette of the binary Gateway vector pGD625 (Chalfun et al., 2005) and transferred to the supervirulent Agrobacterium tumefaciens strain AGL0 (Lazo et al., 1991). Transgenic strawberry plants of the cv Calypso were produced according to Schaart et al. (2002). Initially, 50–100 kanamycin-resistant clones were obtained for each construct, from which 15 good growing lines were transferred to the glasshouse. Finally, for each construct three individual transgenic lines were selected for further experimentation. Calypso nontransgenic plants were used as control. For each line, two or three plants were maintained for harvesting fruits.
Yeast-two-hybrid screening and interaction study
For the yeast-two-hybrid studies, the ORFs of the different MYB, bHLH and TTG1 TF genes described were recombined into the pBDGAL4 bait vector (pDEST™32; Invitrogen) and pADGAL4 prey vector (pDEST™22; Invitrogen) by a Gateway™ LR reaction. For the yeast-two-hybrid screening, an expressed green strawberry fruit cDNA library was recombined into the pADGAL4 prey vector. The resulting bait and prey vectors were transformed into yeast strains PJ69-4a (MATa) and PJ69-4α (MATα) (James et al., 1996), respectively, and used in yeast-two-hybrid experiments as described by de Folter et al. (2005). GenBank accession number for the genes identified are FaMYB5, JQ989280; FaMYB9, JQ989281; FaMYB11, JQ989282; FaMYC1, JQ989283; FabHLH3, JQ989284; FabHLH3∆, JQ989285; FabHLH33, JQ989286; and FaTTG1, JQ989287.
RNA extraction and quantitative real-time reverse transcription PCR (RT-PCR)
Total RNA isolation, cDNA synthesis and quantitative RT-PCR was performed as described previously (Schaart et al., 2002) SYBR Green was used for detection of PCR products on a MyiQ Single-Color Real-Time Detection System (Bio-Rad). All the primers used in this study are described in Supporting Information, Table S1.
Arabidopsis thaliana complementation assays
FaMYB9, FaMYB11, FabHLH3 and FaTTG1 cDNAs were introduced into the binary vector pMDC32 (Curtis & Grossniklaus, 2003) by a Gateway LR recombination. Agrobacterium-mediated transformation of the Arabidopsis tt2-1, tt8-3 and ttg1-1 null mutants were carried out as described in Baudry et al. (2006). Wildtype Landsberg erecta (Ler) and Wassilewskija (WS) were used as controls where appropriate. Fourteen independent transgenic lines per construct were isolated, genotyped and checked for seed colour, from which three were chosen to measure PA accumulation in triplicate (in accordance with Routaboul et al., 2006; Dubos et al., 2008).
Physcomitrella patens transient expression study
FaMYB9, FabHLH3 and FaTTG1 cDNAs were introduced into the expression vector pBS TPp-A by a LR recombination (Thevenin et al., 2012). The reporter proBAN : GFP vector used (containing a 75-bp-long promoter) as well as all the steps from transfection to quantitative analysis, are described in detail elsewhere (Thevenin et al., 2012).
Large variation in flavan-3-ols within strawberry varieties
To investigate the genetic variation in soluble flavan-3-ol monomers in cultivated strawberry, aqueous-methanol extracts from red fruit of 57 different strawberry genotypes were analysed using accurate mass LC-MS. This showed that in all genotypes the monomeric flavan-3-ols catechin and epicatechin were present (Table 1, Fig. S1). In addition to these monomeric forms, two major PA dimers (B1 and B3) were detected in all genotypes (Figs S1, S2). The concentrations of the monomeric flavan-3-ol catechin and of their dimeric PA B1 and B3 derivatives in ripe fruits seemed to be strongly correlated (R2 = 0.84 and 0.91; Fig. S2), whereas epicatechin concentrations did not correlate with either catechin or PA dimer concentrations (not shown). Relative flavan-3-ol monomer content varied up to 8.5 times between the most extreme genotypes (Fig. S1).
Table 1. Proanthocyanidin (PA) composition obtained by phloroglucinolysis of selected strawberry (Fragaria × ananassa) cultivars
Total PA, total soluble PA; mDP, mean degree of polymerization; nd, not detected.
Since it is difficult to extract and quantify all individual oligo- and polymeric PAs, the total PA content in red fruits of a selection of 14 contrasting strawberry cultivars (grown in three consecutive years) were determined using an HPLC photodiode array detector after hydrolysis with phloroglucinol (Table 1). Total PA concentrations varied between 0.54 and 1.32 g kg−1 FW among genotypes. On average, the amount of total PA is 15 times higher than the amount of free PA monomers. Catechin and epicatechin accounted, on average, for 44% (± 5%) and 56% (± 5%) of total PA units, respectively. Catechin was mainly present as the end unit, whereas both catechin and epicatechin were present as extension units, the latter being prominent. The mean degree of polymerization of these soluble PAs varied between 4.8 (± 0.76) and 7.03 (± 0.82).
We examined PA accumulation during fruit development by staining cross-sections of Elsanta fruits at different developmental stages with 4-dimethylaminocinnamaldehyde (DMACA), which specifically stains PA in blue (Fig. 1b,c). Fig. 1(c) shows that PAs were localized throughout the fruit tissue and seemed to be more concentrated in the green fruits compared with later ripening stages. Then we also measured soluble PA content at different fruit developmental stages with four different strawberry genotypes (i.e. Elsanta, Sonata, Holiday× Mirande, and Senga sengana) using HPLC. For all genotypes tested, a similar gradual decrease was observed during fruit ripening to a value of c. 25% of that measured in small green fruits (Fig. 1d).
Overexpression of the Arabidopsis AtTT2, AtTT8 and AtTTG1 PA biosynthesis-related transcriptional regulators in strawberry
Expression of the Arabidopsis PA biosynthesis regulators AtTT2/AtMYB123, AtTT8/AtbHLH042 and AtTTG1 in strawberry was carried out first to assess the potential effect of these proteins on strawberry PA biosynthesis (Walker et al., 1999; Nesi et al., 2000, 2001). Transgenic ripe fruits showed a loss of red coloration in the flesh (Fig. 2a) associated with an increased PA content (Fig. 2b) and a decreased anthocyanin accumulation (Fig. 2c). This increased PA content was also observed in leaves and stems of plants overexpressing simultaneously AtTT2, AtTT8 and AtTTG1 (Fig. S3). The observed anthocyanin loss was mainly the result of a strong decrease in pelargonidin accumulation (Fig. 2d), whereas cyanidins were slightly, but not significantly, increased (Fig. 2e). In order to elucidate the molecular mechanism behind this effect of Arabidopsis MBW complex on PA and anthocyanin biosynthesis in strawberry fruit, a transcript accumulation study of key biosynthetic genes (namely: FaF3′H, flavonoid 3′-hydroxylase FaDFR, dihydroflavonol-4-reductase; FaANS, anthocyanidin synthase; FaANR, anthocyanidin reductase; and FaLAR, leucoanthocyanidin reductase; Fig. 1a) was carried out using quantitative RT-PCR (qRT-PCR). This study revealed that FaF3′H transcript accumulation was slightly increased, FaANS, FaANR and FaLAR transcript accumulation were strongly increased, and that FaDFR transcript accumulation remained unchanged by the Arabidopsis MBW complex (Fig. 2f). The effects on the transcript abundance were consistent with those observed at the metabolic level (Figs 1a, 2a–e). Interestingly, the plants overexpressing AtTT2 alone (but not AtTT8 or AtTTG1 alone) showed the same changes at both metabolite and transcript levels. In ripe fruits of AtTT2-overexpressors, a significant increase in the amount of total PAs was also found which coincided with a decrease in pelargonidin accumulation. Moreover, these fruits also showed elevated gene expression levels of FaF3′H, FaANS, FaANR and FaLAR. This indicates that, in strawberry fruits, AtTT2 alone is sufficient to redirect the flavonoid pathway towards the biosynthesis of PA at the expense of the pelargonidin-type of anthocyanins (Figs 2, S3).
Identification of putative strawberry homologues of AtTT2, AtTT8 and AtTTG1
Two consecutive yeast-two-hybrid screens were carried out in order to identify putative homologues of AtTT2, AtTT8 and AtTTG1 (Figs 3a, S4). The first yeast-two-hybrid screen was performed using AtTT2, AtTT8 and AtTTG1 as bait and an expressed green strawberry fruit cDNA library as prey. Because AtTT2 showed self-activation, a series of C-terminal deletions was made, and one deleted AtTT2 form (of 218 aa) that lost its self-activation property was selected. However, using this partial TF as bait, no interactions were identified. Using AtTT8 as bait, three different strawberry TF cDNAs called FaMYB5, FaMYB11 and FaTTG1 were identified. When AtTTG1 was used as bait, two proteins, FaMYC1 and FabHLH33, showed interaction. In the second yeast-two-hybrid screen, two of the identified strawberry TF genes, FaMYC1 and FaTTG1, were used as bait, but this did not result in new interacting partners. Screening with FaMYC1 identified FaTTG1 as interacting partner, while using FaTTG1 as bait confirmed the putative interactions with FaMYC1 and FabHLH33.
Next to the yeast-two-hybrid screening approach, putative candidate cDNA sequences were amplified from a mixed green and red strawberry fruit cDNA sample with degenerate primers that were designed based on the cDNA sequences of homologues of both AtTT2 and AtTT8. This resulted in three additional candidate cDNAs, FaMYB9, FabHLH3 and FabHLH3Δ.
Interaction of the putative flavonoid regulators in yeast
The yeast-two-hybrid strategy was also used to study the ability of the entire set of identified strawberry transcriptional regulators, including FaMYB1, a repressor previously described by Aharoni et al. (2001), to interact with each other and with AtTT2, AtTT8 and AtTTG1 (Fig. 3a). Before starting this interaction study, the self-activation property of these proteins was evaluated. We found that FaMYB5, FaMYB9 and FaMYB11 displayed a strong self-activation when fused to the GAL4 DNA-binding domain (BD); therefore, their interaction with other TFs was only assessed when fused to the GAL4 DNA-activating domain (AD). From this preliminary analysis, a total of 108 combinations were tested (12 × 9 matrix; Fig. S4), from which 45 interactions were found. Interestingly, most of the observed interactions in yeast were independent of the fused domain (i.e. BD or AD for the bait or prey protein, respectively), as only about one-third (37%) of the interactions were specific to the fusion to the BD domain.
When AtTT8, FabHLH3 and FabHLH3Δ were used as bait, they were found to be able to interact with all the tested MYB TFs, whereas only AtTT8 and FabHLH3 (not FabHLH3Δ) displayed an interaction with the Arabidopsis and strawberry TTG1 proteins. When FaMYC1 or FabHLH33 were used as bait, they were also found to interact with both TTG1 proteins, but no interaction was identified with the MYB TF used in this study. FaMYC1 and FabHLH33 were able to form homo- and heterodimers, a characteristic of some basic helix–loop–helix (bHLH) proteins (Feller et al., 2006). When both TTG1 proteins were used as prey, interactions were found with FaMYB9, FaMYB11, AtTT8, FaMYC1, FabHLH3 and FabHLH33. Finally, the BD fusion to AtTT2 gives AtTT8, FabHLH3 and FabHLH3Δ as well as TTG1 proteins as interacting partners, and when the BD was fused to FaMYB1, it only interacted with FabHLH3Δ.
Phylogenetic analysis of the identified strawberry transcriptional regulators
Phylogenetic analysis of the deduced amino acid sequences using the neighbour-joining method was employed to determine the relationships that may exist between the identified strawberry MYB, bHLH and WD40-repeat regulatory proteins and the known TF genes from Arabidopsis, petunia (Petunia × hybrida), maize (Zea mays), apple (Malus × domestica) and grape (Vitis vinifera). The predicted proteins were similar in length to their homologues, except for FaMYB5 and FabHLH3Δ, for which the predicted proteins were much shorter (Fig. S5).
FaMYB5 was most similar to VvMYBCs1 and VvMYB5b from grape, AtMYB5 from Arabidopsis (52% identity) and PhPH4 from petunia (Fig. 3b). All these MYBs form a subgroup that was previously defined when PhMYB4 was first identified. This MYB subgroup contains two conserved C-terminal motifs named ‘G20’ (Quattrocchio et al., 2006). These G20 motifs are also present in FaMYB5, but not in AtTT2 (Fig. S5a). Unlike the other members of the G20 subgroup, the FaMYB5 deduced amino acid sequence missed the initial start codon, leading to a 70-amino-acid deletion at the N-terminus. The 70-amino-acid deletion results in a loss of the R2 domain that is present in all members of the R2R3-MYB family (Fig. S5a). The R2R3-MYB FaMYB9 and FaMYB11 are homologues of the apple genes MdMYB9 and MdMYB11 (67 and 68% identity, respectively), and both share some similarity with AtTT2 (46 and 45% identity, respectively; Fig. 3a). All three identified strawberry MYB TF genes map to different locations on the genome of woodland strawberry (Fragaria vesca), which was published recently (Shulaev et al., 2011). Although FaMYB9 and FaMYB11 both map to LG6, their mapping position is located 13.3 Mb apart. FaMYB9 and FaMYB11 are the respective homologues of the F. vesca TFs FvMYB58 and FvMYB26, which are both part of a clade of sequences with homology to AtTT2 of a F. vesca phylogenetic tree of R2R3-MYB-predicted proteins (see the Supporting Information of Shulaev et al., 2011). Next to FvMYB58 and FvMYB26, this clade represents four other R2R3-MYB-predicted proteins. FvMYB33, which is also present in this clade, was reported to be highly expressed in F. vesca fruits and was therefore suggested to have a key function in PA synthesis (Shulaev et al., 2011). Phylogenetic analysis shows that FvMYB33 is less related to AtTT2 than to FaMYB9 and FaMYB11 (Fig. 3b).
Two different forms of the bHLH TF gene FabHLH3 were isolated, which differed by several single nucleotide polymorphisms (SNPs) and a 152 bp insert. This insert causes a frame shift mutation, putatively leading to a short and truncated FabHLH3 protein named FabHLH3Δ, which is missing the typical bHLH domain and the ACT-like dimerization domain (Fig. S5b; also see Feller et al., 2006). When blasting both FabHLH3 sequences against the F. vesca genomic sequence, only a single homologous sequence with high similarity was found, suggesting that the two FabHLH3 sequences are not paralogous. The FabHLH3 amino acid sequence is very similar to MdbHLH3 from apple (80% identity) and related to AtTT8 (57% identity) and to other AtTT8-like TFs such as the petunia PhAN1 or the grape VvMYC1 (Fig. 3c). FabHLH33 and FaMYC1 share strong similarity at the predicted protein level, and are more closely related to AtEGL3/AtbHLH002 (43 and 45% identity, respectively) than to AtTT8 (Fig. 3c). Moreover, FabHLH33 appears to be homologous to MdbHLH33 from apple (75% identity). FabHLH33 and FaMYC1 map on unique positions at the F. vesca linkage groups LG7 and LG5, respectively.
Finally, FaTTG1 displayed a very high similarity to its homologue from Arabidopsis, AtTTG1 (80% identity protein level); however, the highest similarity was found with the apple homologue MdTTG1 protein (91% identity; Fig. 3d). Also FaTTG1 was mapped on a single location at the F. vesca genome.
Metabolomic and transcriptomic data clustering analysis during fruit development
Metabolomic (HPLC) and transcriptomic (qPCR) data from 14 genotypes at the mature stage (R), and on the four studied genotypes (Elsanta, Sonata, Holiday× Miranda, and Senga sengana) during fruit development (G1, small green; G2, green; W, white; T, turning; and R, red) were collected and organized by the agglomerative hierarchical clustering (AHC) method, using the EPCLUST tool (Fig. 4; Vilo et al., 2003). The metabolites measured mainly belonged to the PA biosynthetic pathway (i.e. free or polymerized catechin and epicatechin subunits, total flavan-3-ols, and polymerization degree), the anthocyanin pathway (i.e. pelargonidins or cyanidin conjugates and derivatives, total anthocyanins) and the flavonol pathway (i.e. kaempferol or quercetin conjugates). Some other phenolic compounds were also annotated, such as the flavonoid precursor p-coumaric acid (and derivative) or the antioxidant ellagic acid (as conjugate and derivative). Transcript accumulation was measured for key structural genes involved in both PA and anthocyanin biosynthesis (FaF3′H, FaDFR, FaANS, Fa3-GT, FaANR and FaLAR), and for the identified transcriptional regulators (FaMYB1, FaMYB5, FaMYB9, FaMYB11, FaMYC1, FabHLH3, FabHLH33, and FaTTG1).
This global analysis confirms that PA-related metabolites were mainly accumulating early during fruit development, and then decreased during ripening. This analysis also confirmed that the accumulation of the transcripts corresponding to the structural genes specifically involved in PA biosynthesis (i.e. FaANR and FaLAR) as well as FaF3′H, and all the identified regulators, except FaMYB1, were following the same trend.
By contrast, anthocyanin- and flavonol-related metabolites were largely accumulated in ripe fruits (R stage), concomitant with the increase in the degree of PA polymerization and the accumulation of transcripts corresponding to known anthocyanin biosynthetic genes, such as FaDFR, FaANS and Fa3-GT. p-Coumaric and ellagic acid-related compounds were also mostly accumulated in this later stage.
We observed a good correlation (R2 > 0.95) between expression of FaMYB9 and FaMYB11, their putative target genes FaF3′H, FaANS, FaANR and FaLAR, and total PA concentrations at the small green fruit stage. This indicates that the expression of these genes is important in determining genotypic differences in total PA content at this fruit developmental stage. This correlation was absent for FaTTG1 or any of the FabHLH/MYC genes. We did not observe correlation of gene expression at the green or red stage and total PA concentrations at the red ripe fruit stage.
Complementation of Arabidopsis transparent testa mutants with strawberry regulators
To characterize the function of the isolated strawberry transcriptional regulator genes, we performed complementation studies by expressing strawberry R2R3-MYB, bHLH and WDR regulatory genes into Arabidopsis lines mutated for AtTT2 (tt2-1), AtTT8 (tt8-3) and AtTTG1 (ttg1-1), respectively (Fig. 5a–d). The strawberry MYB and bHLH genes were expressed under the control of the double CaMV35S promoter, which provides a constitutive expression pattern.
Based on the results accumulated in the frame of this work, and on available data in the literature, we only selected a subset of genes susceptible to form an active MBW complex in Arabidopsis to be tested. FaMYB5 and FabHLH3Δ were not included as they encoded truncated proteins missing structurally important domains, the R2 repeat and the typical bHLH domain, respectively (Fig. S5). As FaMYC1 and FabHLH33 were not able to interact with any of the studied MYB TFs (Fig. 3a), these two bHLH proteins were also not studied further.
The overexpression of FaMYB9 and FaMYB11 in tt2-1, FabHLH3 in tt8-3 and FaTTG1 in ttg1-1 mutants was able to complement the associated transparent testa phenotype (i.e. bright yellow seed colour), giving the mature seeds their typical brown colour (Fig. 5a–c). These observations were further confirmed by measuring PAs accumulated in the corresponding transgenic seeds (Fig. 5d).
These results showed that FaMYB9 and FaMYB11, FabHLH3 and FaTTG1 are functional homologues of AtTT2, AtTT8 and AtTTG1, respectively, and therefore suggest that they are probably acting together through the formation of an MBW complex.
Assay for functional interactions in P. patens protoplasts
As a final attempt to determine if the regulatory gene products from FaMYB9, FabHLH3 and FaTTG1 can form a MBW complex as suggested by the yeast-two-hybrid experiments, a transient expression assay in the P. patens protoplast (Thevenin et al., 2012) was carried out. In this assay, the concerted action of the selected transcriptional regulators on the activity of the BAN/ANR promoter (proBAN) fused to green fluorescent protein (GFP) was tested.
When these genes were transfected alone or two by two, no clear induction of proBAN activity (i.e. increased GFP intensity) was observed (Fig. 5e). However, a strong activation of proBAN activity was observed when protoplasts were cotransformed simultaneously with FaMYB9, FabHLH3 and FaTTG1 (Fig. 5e), emphasizing the fact that these three proteins act together to activate their target promoters.
PA biosynthesis decreases during strawberry fruit development
It is generally known that ripe strawberry fruits contain high concentrations of the red coloured anthocyanins, but next to this high concentrations of PAs are found in strawberry fruits (Almeida et al., 2007; Buendia et al., 2010). In unripe green fruits, in particular, we also found a relatively high concentration of PAs; however, this decreased considerably during fruit ripening (Fig. 1b–d), which is in agreement with previously results (Carbone et al., 2009). In accordance with Buendia et al. (2010) we found that the concentration of PA in strawberry fruits is genotype-dependent, showing the important role played by the genetic background (Fig. S1).
Expression of AtTT2 in strawberry fruits is sufficient to modify PA biosynthesis
To determine if the PA regulatory network identified in Arabidopsis was functionally conserved in strawberry, we studied the effect of expression in strawberry plants of the well-described Arabidopsis transcriptional regulators AtTT2 (R2R3-MYB), AtTT8 (bHLH) and AtTTG1 (WD40 repeat protein), which specifically modulate the expression of genes involved in PA biosynthesis in seeds (Baudry et al., 2004; Lepiniec et al., 2006). We found that the combined overexpression of AtTT2, AtTT8 and AtTTG1 resulted in a rechannelling in the flavonoid pathway in ripe fruits, leading to increased concentrations of PAs and reduced concentrations of anthocyanins, mainly through a partial loss of pelargonidin content (Fig. 2a–e). These metabolic modifications were accompanied by an increase in transcript accumulation for some key structural genes (Figs 1a, 2f), namely FaANS (anthocyanidin synthase), FaANR (anthocyanidin reductase) and FaLAR (leucoanthocyanidin reductase), and, to a lesser extent, FaF3′H (flavonoid 3′-hydroxylase). These transcriptional modifications were in agreement with what was observed in Arabidopsis where the AtTT2-AtTT8-AtTTG1 MBW complex was shown to regulate the later steps of the PA biosynthetic pathway (Baudry et al., 2004; Lepiniec et al., 2006).
Interestingly, we found that these metabolic and expressional changes were also observed when AtTT2 was expressed alone (and not when either AtTT8 or AtTTG1 was expressed). The simplest hypothesis for this is that, in strawberry, functional AtTT8-like and AtTTG1-like transcriptional regulators are present which are able to act as AtTT2 interacting partners and form a functional MBW regulatory complex. As such R2R3-MYBs are considered to be highly pathway-specific (Lepiniec et al., 2006), this implies that in strawberry TT2-like R2R3-MYB TF may have a key function in regulating PA biosynthesis. These results prompted us to isolate the strawberry homologues of AtTT2, AtTT8 and AtTTG1.
Identification of putative AtTT2, AtTT8 and AtTTG1 homologues in strawberry
To identify the putative homologues of the Arabidopsis AtTT2, AtTT8 and AtTTG1, two strategies were used. First, the Arabidopsis genes were used as bait in a yeast-two-hybrid screen of an expressed strawberry fruit cDNA library, which led to the identification of several transcriptional regulators from three different protein families, the MYB (FaMYB5 and FaMYB11), the bHLH (FaMYC1 and FabHLH33) and the WD40-repeat proteins (FaTTG1; Figs 3a, S4). As TF interactions are known to be functionally conserved (Immink & Angenent, 2002), the identified strawberry proteins may be functional counter partners in the formation of the MBW complex regulating the expression of genes involved in PA biosynthesis. Secondly, using a PCR-based strategy, additional MYB (FaMYB9) and bHLH (FabHLH3 and FabHLH3∆) genes were isolated from strawberry fruit. Comparison of deduced protein sequences of the identified strawberry TF genes with characterized TFs from other plant sources show that all shared homology with TFs involved in the regulation of flavonoid biosynthetic genes, and in particular with genes involved in PA biosynthesis (Fig. 3b–d).
When all the identified regulators were assayed against each other in yeast (12 × 9 combination matrix), some interactions were found to be specific to the domain (i.e. GAL4 DNA-binding, BD, or activation, AD, domains) to which the proteins were fused. This usual result could be explained by conformational changes of the proteins fused to AD or DB domains (Burbulis & Winkel-Shirley, 1999) and has previously been described for MYB, bHLH and WDR proteins (e.g. by Baudry et al., 2004). From this study we found that AtTT8, FabHLH3 and FabHLH3∆ were able to interact with all the studied MYBs, whereas FaMYC1 and FabHLH33 were not. FaMYC1 and FabHLH33 could form homo- and heterodimers, which was previously reported for other bHLH proteins (Feller et al., 2006). Both AtTTG1 and FaTTG1 were found to interact with nearly all the tested MYB and bHLH TFs, with the exception of FaMYB1, FaMYB5 and FabHLH3∆.
The results obtained through this yeast-two-hybrid interaction matrix suggested that FaMYB9 and FaMYB11 could be the AtTT2 homologues in strawberry, as they were the only R2R3-MYB found to interact with FaTTG1. Interestingly, as FabHLH3 was found to interact with FaTTG1, as well as with FaMYB9 and FaMYB11, FabHLH3 and FaTTG1 could likely be the strawberry orthologues of AtTT8 and AtTTG1, respectively.
Coregulation study confirm the putative involvement of most of the studied transcriptional regulator in PA biosynthesis
The expression pattern of the genes encoding the enzymatic steps of the PA biosynthetic pathway have been previously studied during strawberry development; however, there was hardly any information available on their regulation (Almeida et al., 2007; Carbone et al., 2009). One way to identify putative regulators that are specific to a particular biosynthetic pathway is to search for coregulation. Such coregulation can be found between TF and putative targets like structural genes, but can also include any compound related to the studied pathway. We took the advantage of the availability of a large set of strawberry genotypes (14 different) to measure in ripe fruit (all genotypes), and during fruit development (four genotypes, Fig. 1d), the accumulation of diverse flavonoid compounds (including flavonols, anthocyanins, PAs, and the p-coumaric acid precursor) as well as the antioxidant ellagic acid. In parallel, transcript accumulation was also measured for six key structural genes (i.e. FaF3′H, FaDFR, FaANS, Fa3-GT, FaANR, and FaLAR) and eight of the studied transcriptional regulators (i.e. FaMYB1, FaMYB5, FaMYB9, FaMYB11, FaMYC1, FabHLH3, FabHLH33, and FaTTG1). These data were clustered with the aim of identifying MYB and bHLH TFs that were specific to PA biosynthesis in strawberry fruits (Vilo et al., 2003).
The results obtained from this global correlation analysis support the hypothesis that all identified transcriptional regulators are probably involved in PA biosynthesis, and were not involved either in anthocyanin or flavonol biosynthesis, or in p-coumaric and ellagic acid accumulation (Figs 1b–d, 4). The involvement of FaMYB1 in the regulation of the anthocyanin biosynthetic pathway (Aharoni et al., 2001) was confirmed.
FaMYB9/FaMYB11, FabHLH3 and FaTTG1 are the strawberry functional homologues of AtTT2, AtTT8 and AtTTG1, respectively
During the course of this study, we found that FaMYB9 and FaMYB11 were able to interact with AtTT8 and AtTTG1, and that FaMYB9 and FaMYB11 were encoding two R2R3-MYB TFs highly similar at the amino acid level to AtTT2. When expressed in the Arabidopsis yellow seeded tt2-1 mutant, these two strawberry R2R3-MYB TFs were able to restore the wildtype brown seed colour phenotype. Because FaMYB9 and FaMYB11 map to different positions on the F. vesca genome, it is most likely that FaMYB9 and FaMYB11 are paralogous genes (i.e. generated by a gene duplication). However, despite the paralogous nature of FaMYB9 and FaMYB11, we were not able to demonstrate a clear divergence in gene expression. Both show a similar expression pattern during fruit development (decreasing from the G1 to the R stage), independent of the studied cultivars (Fig. 4). Altogether, these data suggested that FaMYB9 and FaMYB11 are two potential functional homologues of AtTT2 (Figs 3–5, S4).
Similar analysis leads us to postulate that the strawberry bHLH gene FabHLH3 could be functionally orthologous to AtTT8. FabHLH3 encodes an R/B-like (class IIIf) bHLH TF highly similar to AtTT8, which was found to interact in yeast with FaMYB9, FaMYB11 and FaTTG1 (Heim et al., 2003; Zimmermann et al., 2004). Expression of FabHLH3 in the Arabidopsis tt8-3 null mutant resulted in complementation of the transparent testa phenotype, through the accumulation of PAs in the seed coat, confirming that FabHLH3 is a functional homologue of AtTT8 (Figs 3, 5, S4).
FaTTG1 is the only one gene encoding a WD40 repeat protein highly similar to AtTTG1 that was identified. TTG1-like proteins were described as being able to interact with several proteins simultaneously, and seem to have a docking function, stabilizing the ternary TF complex, rather than a direct regulatory function (Baudry et al., 2004; Lepiniec et al., 2006; Hichri et al., 2011). This latter specificity was in agreement with the interaction studies where FaTTG1 was found to interact with FaMYB9, FaMYB11 and FabHLH3. Complementation experiments carried out in Arabidopsis confirmed that FaTTG1 was the functional homologue of AtTTG1.
In addition, the transactivation assays carried out in P. patens protoplasts also supported the idea that FaMYB9, FabHLH3 and FaTTG1 functionally interact to regulate PA synthesis during strawberry fruit development (Fig. 5f).
FaMYB5 and FabHLH3∆: two new negative regulators of flavonoid biosynthesis?
FaMYB5, which was identified as an AtTT8-interacting partner, is more similar to AtMYB5 than to AtTT2. In Arabidopsis, AtMYB5 in combination with AtTT8 and AtTTG1 form a ternary complex that was shown to regulate outer seed coat development and trichome morphogenesis, and was proposed to play a role in regulating PA biosynthesis in seeds (Gonzalez et al., 2009; Li et al., 2009). In petunia, the AtMYB5 homologue PhPH4 is involved in vacuolar acidification in petal cells (Quattrocchio et al., 2006), whereas the grapes homologues VvMYB5a and VvMYB5b play a role in the regulation of anthocyanin and PA biosynthesis in berries (Deluc et al., 2006, 2008). Altogether, these data suggest that the R2R3-MYB that are characterized by the presence of two ‘G20’ motifs (Quattrocchio et al., 2006) can have divergent functions (i.e. development vs secondary metabolism), even if their main role in planta seems to be the regulation of flavonoid biosynthesis. Beside this general sequence feature, FaMYB5 differs from the other G20-subgroup members by the lack of the R2-repeat in its DNA binding domain (Fig. S5). Recently, similar single-repeat R3-MYB TFs were identified in Arabidopsis as negative regulators of flavonoid biosynthesis (Dubos et al., 2008; Matsui et al., 2008; Zhu et al., 2009). For example, ectopic expression of AtMYBL2 in seeds inhibited both the expression of genes involved in PA biosynthesis and PA accumulation, whereas atmybl2 loss-of-function mutants presented an enhanced anthocyanin accumulation, and transcript accumulation of the corresponding genes in vegetative tissues. Moreover, AtMYBL2 was found to interact in yeast with AtTT8, and to be able to inhibit the AtTT2–AtTT8–AtTTG1 complex activity in both Arabidopsis and the P. patens protoplast, demonstrating the repression role of AtMYBL2 through the direct interference of the MBW complex activity (Dubos et al., 2010; Thevenin et al., 2012). The fact that FaMYB5 interacted with AtTT8 and its strawberry homologue FabHLH3 in yeast suggests that this TF may have a similar inhibitory role in strawberry flavonoid biosynthesis (Fig. 5f). Because FaMYB5 displayed a steady amount of transcript accumulation during strawberry fruit development, it can be hypothesized that FaMYB5 may have a role in fine-tuning both PA biosynthesis during early fruit development and anthocyanin biosynthesis during fruit ripening. Interestingly, it was recently found in petunia that the R3-MYB, PhMYBx, acts in concert with an R2R3-MYB repressor, PhMYB27, to negatively regulate the PhPHZ–PhAN1–PhAN11 MBW complex activity in order to modulate anthocyanin biosynthesis in vegetative tissues (Albert et al., 2011). Based on this work, it is tempting to speculate that such a mechanism may exist in strawberry fruits, where FaMYB5 could act in combination with the negative regulator FaMYB1 (Aharoni et al., 2001) to regulate the activity of the MBW complexes involved in anthocyanin and/or PA biosynthesis (Fig. 5f).
Along the same lines, FabHLH3∆ encodes a truncated TF that misses the bHLH and ACT-like dimerization domains (Fig. S5a). This small bHLH may act as dominant repressor because the remaining MYB-interacting domain is still present and able to bind to R2R3-MYB proteins (such as FaMYB9 or FaMYB11), thereby blocking the binding sites for functional bHLH TFs. A similar regulatory mechanism was proposed for the maize ZmIN1 bHLH factor. The mis-splicing of ZmIN1 results in premature termination of translation, producing a protein lacking the bHLH domain, which was proposed to be a putative anthocyanin biosynthesis negative regulator (Burr et al., 1996). Because FabHLH3∆ was also found to interact with FaMYB5, and because both were expressed early during fruit development, it is possible that FabHLH3∆ may acts as an inhibitor of the MYB negative regulator (or vice-versa), in a negative feedback regulatory loop (Figs 4, 5f).
The possibility that FaMYB5 and FabHLH3∆ may repress flavonoid (i.e. PAs and/or anthocyanins) biosynthesis in strawberry fruits will be further investigated using both genetic and molecular approaches (e.g. loss-of-function in strawberry, heterologous mutant complementation, protein activity characterization, etc.).
We thank Dr Richard Immink and Marco Busscher for their excellent help in setting up yeast-two-hybrid screening. This research was carried out as part of the EU-funded project FLAVO (Food CT-2004-513960). The work of W.X. was generously supported by the China Scholarship Council (CSC).