In Arabidopsis thaliana, proanthocyanidins (PAs) accumulate in the innermost cell layer of the seed coat (i.e. endothelium, chalaza and micropyle). The expression of the biosynthetic genes involved relies on the transcriptional activity of R2R3-MYB and basic helix-loop-helix (bHLH) proteins which form ternary complexes (‘MBW’) with TRANSPARENT TESTA GLABRA1 (TTG1) (WD repeat protein). The identification of the direct targets and the determination of the nature and spatio-temporal activity of these MBW complexes are essential steps towards a comprehensive understanding of the transcriptional mechanisms that control flavonoid biosynthesis.
In this study, various molecular, genetic and biochemical approaches were used.
Here, we have demonstrated that, of the 12 studied genes of the pathway, only dihydroflavonol-4-reductase (DFR), leucoanthocyanidin dioxygenase (LDOX), BANYULS (BAN), TRANSPARENT TESTA 19 (TT19), TT12 and H+-ATPase isoform 10 (AHA10) are direct targets of the MBW complexes. Interestingly, although the TT2–TT8–TTG1 complex plays the major role in developing seeds, three additional MBW complexes (i.e. MYB5–TT8–TTG1, TT2–EGL3–TTG1 and TT2–GL3–TTG1) were also shown to be involved, in a tissue-specific manner. Finally, a minimal promoter was identified for each of the target genes of the MBW complexes.
Altogether, by answering fundamental questions and by demonstrating or invalidating previously made hypotheses, this study provides a new and comprehensive view of the transcriptional regulatory mechanisms controlling PA and anthocyanin biosynthesis in Arabidopsis.
Flavonoids are secondary metabolites that fulfil a multitude of functions during plant growth and development (Winkel-Shirley, 2001; Lepiniec et al., 2006). Three main classes of flavonoids are found in Arabidopsis thaliana, namely flavonols, anthocyanins and proanthocyanidins (PAs). PAs, also called condensed tannins, are flavonoid polymers resulting from the condensation of flavan-3-ol units that specifically accumulate in the innermost cell layer of the testa (chalaza, micropyle and endothelium), conferring to the mature seed its characteristic brown colour (Debeaujon et al., 2003; Pourcel et al., 2005; Lepiniec et al., 2006). In Arabidopsis, most of the mutants impaired in flavonoid accumulation have been isolated through screening for altered seed pigmentation which results in the transparent testa (tt) phenotype (Koornneef, 1990). The molecular characterization of the tt loci has allowed the flavonoid biosynthetic pathway to be deciphered.
The structural genes leading to PA biosynthesis in Arabidopsis seeds are usually divided into two groups, the so-called early (EBGs) and late (LBGs) biosynthetic genes (Pelletier et al., 1999; Lepiniec et al., 2006). The EBGs include chalcone synthase (CHS), chalcone isomerase (CHI), flavonol 3-hydroxylase (F3H) and flavonol 3′-hydroxylase (F3′H), which are involved in precursor biosynthesis for the three classes of Arabidopsis flavonoids. The LBGs comprise dihydroflavonol-4-reductase (DFR), leucoanthocyanidin dioxygenase (LDOX) and BANYULS/anthocyanidin reductase (BAN/ANR). A third group of structural genes, which includes TT10 (laccase 15), TT12 (MATE transporter), TT19 (glutathione-S-transferase) and AHA10 (H+-ATPase), has been shown to be involved in PA modification, transport and oxidation (Kitamura et al., 2004; Baxter et al., 2005; Pourcel et al., 2005; Marinova et al., 2007; Routaboul et al., 2012). Finally, a gene whose function has not yet been clearly elucidated, namely TT15/UGT80B1 (UDP-glucose:sterol-glucosyltransferase), has been shown to affect the PA accumulation process in seeds. TT15 catalyses the synthesis of steryl glycosides, which, in turn, affect the trafficking of lipid polyester precursors and, most probably indirectly, PA accumulation (DeBolt et al., 2009).
Numerous studies have indicated that the expression of the genes involved in the flavonoid biosynthetic pathway is controlled by distinct mechanisms in a tissue- or species-specific manner (Winkel-Shirley, 2001; Koes et al., 2005; Lepiniec et al., 2006; Petroni & Tonelli, 2011; Schaart et al., 2013). This regulation involves different sets of transcriptional regulators that are specific to each class of flavonoids. In Arabidopsis, EBGs and FLAVONOL SYNTHASE 1 (FLS1, which leads to the accumulation of flavonols) expression is controlled by at least three R2R3-MYBs, namely PFG1/MYB12, PFG2/MYB11 and PFG3/MYB111 (Stracke et al., 2007). The regulation of PA and anthocyanin biosynthesis involves the combined action of specific R2R3-MYB (subgroup 5 and 6) and R/B-like basic helix-loop-helix (bHLH) (subgroup IIIf) transcription factors (TFs), together with TRANSPARENT TESTA GLABRA1 (TTG1) (WD repeat protein), in a MYB–bHLH–WDR (MBW) ternary protein complex (Heim et al., 2003; Baudry et al., 2004; Zimmermann et al., 2004; Lepiniec et al., 2006; Dubos et al., 2010; Thévenin et al., 2012). In vegetative tissues, anthocyanin biosynthesis is regulated by different MBW complexes involving PRODUCTION OF ANTHOCYANIN PIGMENT 1 (PAP1)/MYB75 and PAP2/MYB90, in combination with GLABRA3 (GL3)/bHLH00, ENHANCER OF GLABRA3 (EGL3)/bHLH002, TT8/bHLH042 and TTG1 (Zhang et al., 2003; Baudry et al., 2006; Feyissa et al., 2009; Appelhagen et al., 2011a). In seeds, TT2/MYB123, TT8 and TTG1 are the main PA biosynthesis regulators (Lepiniec et al., 2006). The TT2–TT8–TTG1 protein complex directly regulates the expression of BAN, which encodes the first enzyme committed to the PA biosynthetic pathway (Baudry et al., 2004). In addition, DFR and LDOX are direct targets of the TTG1:GR chimeric protein (a translational fusion between TTG1 and the glucocorticoid receptor; Gonzalez et al., 2008). Moreover, two recent studies on TT1/WIP1 (WIP zinc finger), a TF that has been hypothesized to regulate the competency of endothelium cells to synthesize and accumulate PAs, have suggested that the transcriptional regulatory network that controls PA biosynthesis and accumulation may be more complicated than previously thought (Appelhagen et al., 2010, 2011b). TT1 may act synergistically with the TT2–TT8–TTG1 complex to regulate the expression of the EBGs, such as CHS, as already suggested for other MBW complexes during vegetative development (Borevitz et al., 2000; Tohge et al., 2005; Appelhagen et al., 2011a,b). In addition, the overexpression of MYBL2 (R3-MYB), a negative regulator of MBW complex transcriptional activity in PA-accumulating cells, leads to a lower accumulation of PAs and LBGs, CHS, F3H and F3′H mRNA levels (Dubos et al., 2008). This latter finding may be the result of either a negative metabolic feedback or a negative regulation of a transcriptional activator of EBGs. In this respect, it has also been shown that TT8 promoter activity is itself partially regulated by TT1 (Xu et al., 2013). From these studies, it is tempting to speculate that MYBL2 may also inhibit the activity of both the TT2–TT8–TTG1 complex and the TT1 TF (Zimmermann et al., 2004; Dubos et al., 2008; Appelhagen et al., 2011a). Finally, functional redundancy involving MYB5 and either EGL3 or GL3 could be at play in this regulatory process, but the set of structural genes that may be directly regulated by these additional TFs remains to be identified (Baudry et al., 2004; Lepiniec et al., 2006; Gonzalez et al., 2009).
TFs regulate the expression of their target genes through the interaction with specific DNA cis-regulatory elements usually localized upstream of the transcribed region (within the promoters). Therefore, the identification of these cis-elements is an essential step towards a comprehensive understanding of the transcriptional regulation. The means by which the TT2–TT8–TTG1 complex regulates its target genes has been explored previously through the study of the BAN promoter activity, allowing the identification of a 236-bp-long minimal promoter upstream from the translational start site (Debeaujon et al., 2003; Baudry et al., 2004; Thévenin et al., 2012). In these studies, specific R2R3-MYB and bHLH binding sites through which the TT2–TT8–TTG1 complex regulates BAN expression were identified and characterized. Similar cis-regulatory motifs were also identified in the promoter of key Arabidopsis genes involved in trichome and root hair differentiation (i.e. GL2, CAPRICE (CPC), TTG2 and MYB23), from which additional MBW complexes (involving GL1/MYB0, WEREWOLF (WER)/MYB66 or MYB23 together with GL3, EGL3 and TTG1) regulate their expression (Koshino-Kimura et al., 2005; Ryu et al., 2005; Ishida et al., 2007; Kang et al., 2009; Song et al., 2011). Altogether, these studies suggest that the MBW transcriptional regulatory complexes may regulate the expression of the entire set of their target genes through specific interaction with MYB and bHLH cis-regulatory motifs, although this remains to be proven.
In this study, a combination of genetic, biochemical and molecular methods was undertaken in order to dissect the regulation of PA biosynthetic genes orchestrated by the MBW complexes. The regulation of gene expression was investigated through the characterization of the mRNA accumulation level (using quantitative reverse transcription-polymerase chain reaction, qRT-PCR) and promoter activity (using the β-glucuronidase (GUS) reporter gene) for the so-called EBGs and LBGs, as well as for TT10, TT12, TT15, TT19 and AHA10, in single and multiple loss-of-function mutant backgrounds. This strategy revealed that only DFR, LDOX, BAN, TT12, TT19 and AHA10 are regulated by the TT2–TT8–TTG1 complex. The use of transgenic plants that express the inducible TTG1:GR chimeric protein showed that these genes are direct targets of the MBW complex. Genetic analyses demonstrated that partially overlapping and redundant transcriptional regulatory activities were at play for all of the direct target genes, except TT19, in both seed and seedlings. We found that MYB5 acts redundantly with TT2 only in the endothelium, whereas TT8, EGL3 and GL3 act together in the chalaza area. The means by which these MBW complexes interact with the promoters of their target genes were investigated through a promoter deletion strategy using both transient and stable transformation assays in Physcomitrella patens protoplasts and Arabidopsis seeds, respectively. Taken together, these results provide a comprehensive view of the transcriptional regulation of the flavonoid biosynthetic pathway by the MBW regulatory complexes in Arabidopsis seed.
Materials and Methods
Unless otherwise stated below, all the primers used in this study are described in Supporting Information Table S1.
Arabidopsis thaliana (L.) Heynh accessions Wassilewskija (WS) and Landsberg erecta (Ler) were used as wild-type (WT) controls where appropriate. The Arabidopsis thaliana tt2-1, tt8-3, ttg1-1, tt8 egl3, tt8 gl3 egl3 and myb5-1 mutant lines, as well as the lines overexpressing TTG1 fused to the GR domain, have been described elsewhere (Debeaujon et al., 2003; Zhang et al., 2003; Baudry et al., 2004; Gonzalez et al., 2009). The tt2 myb5 double mutant line used in this study was obtained by crossing tt2-1 with myb5-1. Plant growth, transformation and selection for transgenic lines were performed as reported previously (Nesi et al., 2000).
The proOsActine:TT2, proOsActine:MYB5, proOsActine:TT8, proOsActine:GL3, proOsActine:EGL3 and proOsActine:TTG1 vectors used to express the corresponding genes in protoplasts of the moss P. patens have been described elsewhere (Thévenin et al., 2012; Xu et al., 2013).
All promoters (ranging from 0.5 to 2 kb, which correspond to the whole intergenic regions) in this study were amplified by PCR using high-fidelity Phusion DNA polymerase (Thermo Scientific–Finnzymes, Waltham, MA, USA) from WS genomic DNA, and subsequently recombined into pDONR207 vector (BP Gateway® reaction) according to the manufacturer's instructions, and sequenced to ensure their integrity. Promoters were then inserted into the destination vector by LR reactions (Gateway®). The destination vectors used for the promoters were pBS TPp-B (green fluorescent protein (GFP) reporter gene; Thévenin et al., 2012) and pGWB3 (GUS reporter gene; Nakagawa et al., 2007) for P. patens protoplast transfection assays and Arabidopsis stable transformation, respectively.
The 5′ deletion fragments were amplified from the promoters inserted in the pGWB3 vector as templates and cloned as described above. Primers used for amplification were modified by the addition of the 35S cauliflower mosaic virus (CaMV) minimal promoter sequence before the 5′-end of the reverse primers.
The proTT8:MYB5 vector was generated by an LR recombination as described above from the following two vectors: proTT8:GTW (TT8 promoter) and pDONR207-MYB5 (MYB5 coding sequence). Both vectors have been described elsewhere (Dubos et al., 2008; Xu et al., 2013).
Transient expression assays in P. patens protoplasts
Moss culture, protoplast preparation, protoplast transformation and flow cytometry measurement analysis were carried out as described previously (Thévenin et al., 2012). Each experiment was repeated at least three times, with three technical repetitions per experiment.
Total RNA was extracted from 4-d-old siliques, treated with DNase, reverse transcribed into cDNA and assayed using qRT-PCR as described previously (Baudry et al., 2004, 2006). The primer pairs used in this study for CHS, CHI, F3H, F3′H, DFR, LDOX, BAN/ANR and TT10 have been described elsewhere (Baudry et al., 2004, 2006; Dubos et al., 2008). The primers used for TT12 (At_TT12_1_SG QuantiTect primer assay) and TT15 (At_AT1G43620_1_SG QuantiTect primer) were purchased from Qiagen. For each experiment (i.e. mutant analysis or dexamethasone (DEX) induction), cDNA samples were generated from three different plants per genotype, each being assayed three times by qRT-PCR (technical repetitions). All the experiments were repeated at least three times.
DEX induction experiments and RNA analysis
For each treatment, four siliques (4 d after pollination) were taken from three independent plants for each line, opened and incubated in 24-well plates in the presence of Mock buffer alone or with 100 μM cycloheximide (CHX; Sigma-Aldrich) before a 30-min vacuum treatment to ensure effective infiltration of CHX. DEX was then added to a final concentration of 10 μM, and another 30-min vacuum treatment was applied. After 3 h, the reaction media were renewed, and incubation was continued for 3 h. After 6 h, samples were collected in 1.5-ml Eppendorf Safe-Lock Tubes™ (Hamburg, Germany) and immediately frozen in liquid nitrogen (Baudry et al., 2004).
Histochemical detection of GUS activity
GUS staining for seeds expressing promoter:uidA gene fusion (and deleted versions) constructs was performed in the presence of 2 mM potassium ferri-/ferrocyanide, when necessary, as described elsewhere (Debeaujon et al., 2003; Berger et al., 2011). For each construct, 8–24 independent transgenic plants were analysed, and representative observations are presented.
Complementation of the transparent testa phenotype of tt2
The tt2-1 mutant plants were stably transformed with the proTT8:MYB5 vector, from which 12 independent transgenic lines displayed a WT seed colour (brown).
Each experiment (i.e. qRT-PCR analysis and transient assays in P. patens protoplasts) was repeated at least three times (biological repetitions). For each sample, three measures were carried out (technical repetitions). Student's t-tests were used to compare the samples against their corresponding controls. The null hypothesis was rejected when the P value was below 5% (P <0.05).
Contribution of TT2, TT8 and TTG1 to the expression of PA biosynthetic genes
The involvement of TT2, TT8 and TTG1 in the expression of 12 characterized flavonoid biosynthetic genes leading to PA accumulation was investigated by comparing mRNA accumulation levels between WT and the tt2, tt8 and ttg1 mutants (Fig. 1). The mRNA steady-state levels were quantified in 4-d-old siliques using qRT-PCR as, at this stage of silique development (globular stage), most of the mRNAs accumulate to their maximum (Nesi et al., 2000, 2001; Kleindt et al., 2010). A significant decrease in F3H, DFR, LDOX, BAN, TT12, TT19 and AHA10 mRNA accumulation was observed in all three mutants when compared with the WT (Fig. 1). F3′H also displayed a lower transcript accumulation in the tt2 and ttg1 mutants when compared with the WT, whereas no variation was observed in the tt8 mutant. A weak decrease in TT10 mRNA accumulation was also observed in the tt8 and ttg1 mutants when compared with the WT, but not in the tt2 mutant. A slight but insignificant increase in CHS mRNA accumulation was observed in the tt2 mutant and for CHI in the tt2 and ttg1 mutants. Finally, a significant increase in CHS mRNA accumulation was observed in tt8 and ttg1.
Activity of 12 PA pathway promoter sequences in Arabidopsis
The promoters (i.e. intergenic region of c. 0.5–2 kb upstream from the translation initiation start (TIS) site) of the 12 studied structural genes were fused to the uidA (GUS) reporter gene and stably introduced into the WT and various regulatory mutant backgrounds.
As shown previously for the F3H, BAN, TT10, TT12, TT15 and AHA10 promoters, all the promoters analysed in this study, except for F3′H, were active in PA-accumulating cells (i.e. chalaza, micropyle and endothelium) of WT seeds (Figs 2, S1; Harper et al., 1994; Debeaujon et al., 2003; Pourcel et al., 2005; Marinova et al., 2007; DeBolt et al., 2009; Berger et al., 2011). Although c. 24 transgenic plants were analysed, no GUS activity was found for the F3′H promoter, preventing further analyses.
The activity remained unchanged for CHS, CHI, F3H, TT10 and TT15 promoters in the seed coat of tt2, tt8 and ttg1 mutants. This result is fully consistent with the mRNA accumulations found for CHS, CHI and TT15. It also suggests that the lower mRNA accumulation observed for F3H and TT10 does not result from a direct transcriptional regulation by the TT2–TT8–TTG1 complex in the seed coat.
By contrast, DFR, LDOX, BAN, TT12, TT19 and AHA10 promoter activities were strongly altered in the three studied mutant backgrounds (Fig. 2). No activity could be observed for the six genes in the ttg1 mutant, even though a faint and diffuse GUS activity was observed in a few lines for the DFR promoter. Interestingly, DFR, LDOX and TT12 showed similar GUS patterns in the seed coat of tt8 (in chalaza) and tt2 (in endothelium), whereas no activity was found in tt2 and tt8 for both TT19 and BAN promoters. It is noteworthy that, in the present study, the previously characterized BAN minimal promoter was used (236-bp fragment upstream from the TIS site; Debeaujon et al., 2003). Finally, when AHA10 promoter activity was tested in these three mutants, blue staining was only found in the chalaza of tt8 (Fig. 2).
Because anthocyanins accumulate in vegetative tissues, the activity of the different promoters was assayed in WT and mutant 10-d-old seedlings. No activity was found for the promoters of BAN, TT10, TT12 and AHA10 (which code for proteins specifically involved in PA metabolism). By contrast, DFR, LDOX and TT19 promoter activity was detected in the cotyledon margin, independent of TT2 and TT8. Nevertheless, these activities were totally lost in ttg1 (Fig. S2a). Finally, the activities of the CHS, CHI, F3H and TT15 promoters were found throughout the whole seedling, with the exception of the roots for the promoter of TT15, and were unaffected in the three mutant backgrounds tested (Fig. S2b).
Finally, promoter activities were assayed in different organs of 5-wk-old plants (Fig. S3). BAN, TT10 and TT12 promoters were found to specifically drive GUS activity in seeds, and not in any other of the organs tested. In rosette leaves, depending on the assayed promoter, blue staining was found in the whole organ (CHS and TT15), in the leaf margin (CHI and F3H) or mainly in the mid-vein and lamina (DFR, LDOX, TT19 and AHA10). In cauline leaves, GUS activity was detected in the whole tissues or in the margin, for CHS and TT15, and CHI, F3H, DFR and LDOX, promoters, respectively. The CHS, CHI, F3H and TT15 promoters also displayed some GUS activity in petals, whereas the TT10 promoter was active in anthers and stigmas. In the apical part of the stems, blue staining was found for the CHS, F3H and TT15 promoters. In the lower part of the stem, GUS activity was found after cutting induction in the cortex and the cambium for the CHS, CHI, DFR, LDOX and TT19 promoters. By contrast, TT15 promoter activity was found throughout the stem (except in the xylem).
Identification of the primary targets of TTG1-dependent regulatory complexes
The characterization of promoter activities and mRNA accumulations suggested that DFR, LDOX, BAN, TT12, TT19 and AHA10 are primary targets of TT2, TT8 and TTG1. This is consistent for BAN, which has been shown previously to be directly regulated by TT2, TT8 and TTG1 (Baudry et al., 2004, 2006).
This latter finding was based on the analysis of plants that constitutively overexpressed the chimeric TTG1:GR protein, which is unable to translocate into the nucleus without the addition of DEX. Primary target genes were identified as being induced at the transcriptional level after the simultaneous treatment of DEX and CHX, an inhibitor of protein translation. Thus, the same transgenic lines were used to determine whether a TTG1-dependent regulatory complex directly controls the expression of DFR, LDOX, TT12, TT19 and AHA10. The corresponding mRNA steady-state levels were determined by qRT-PCR on 4-d-old siliques after 6-h treatments (Fig. 3). Significant and reproducible inductions of BAN mRNA accumulation were obtained after DEX, and DEX plus CHX, treatments, compared with the Mock and CHX controls. By contrast, no effect on TT10 mRNA accumulation was observed with any of these treatments. These positive (BAN) and negative (TT10) controls demonstrated that the experimental conditions were effective. Interestingly, DEX treatment resulted in a significant induction of DFR, LDOX, TT12, TT19 and AHA10 mRNA levels, which was also observed when DEX was used in combination with CHX. Altogether, these data demonstrate that DFR, LDOX, BAN, TT12, TT19 and AHA10 are primary targets of TTG1-dependent regulatory complexes.
Characterization of functionally redundant MYB–bHLH–TTG1 complexes in seeds
MYB5 and the two bHLH GL3 and EGL3 are known to be involved in TTG1-dependent regulatory complexes (Heim et al., 2003; Zhang et al., 2003; Gonzalez et al., 2009; Li et al., 2009). In order to test whether these proteins could also regulate the activity of the LBG promoters in seeds, two complementary strategies were used.
First, the different MBW combinations were assayed against the DFR, LDOX, TT12 and AHA10 promoters using the P. patens protoplast transient expression system (Thévenin et al., 2012). These four promoters have been chosen because they are direct targets of TTG1-dependent complexes that still display some activity in the tt8 and/or tt2 mutant seeds (Fig. 2). This approach showed that EGL3 could functionally replace TT8 in combination with TT2 and TTG1 to induce the activity of the four promoters, and that MYB5 was also able to functionally replace TT2 with either TT8 or EGL3, but to a lesser extent (Fig. 4). Weak but significant activations of the DFR and AHA10 promoters were also observed when GL3 was assayed in combination with MYB5.
Second, genetic validation was carried out in planta by analysing mRNA accumulation or promoter activity in double or triple loss-of-function mutants (i.e. tt2 myb5, tt8 egl3 or tt8 egl3 gl3). The accumulation of DFR, LDOX and TT12 mRNAs in the tt2 myb5 double mutant revealed that the expression of these three genes is regulated in the endothelium by both TT2 and MYB5 (Fig. 5a). AHA10 was not investigated in the tt2 myb5 double mutant because no promoter activity was detected in tt2 (Fig. 2). In this respect, no GUS activity was detected in the seeds of the tt2 myb5 double mutant when the promoters of DFR, LDOX and TT12 were analysed (Fig. 5b). These findings are supported by the ability of MYB5 to partially complement the transparent testa phenotype of tt2 when overexpressed in seeds (i.e. light brown colour of the complemented seeds) under the control of the TT8 promoter (Fig. 5c; Baudry et al., 2006; Dubos et al., 2008). In addition, GUS analysis showed that, in the chalaza, TT8 and EGL3, but not GL3, regulate the expression of LDOX, TT12 and AHA10, whereas the three bHLHs are involved in the regulation of DFR expression (Fig. 5d,e).
Functional dissection of the MBW target gene promoters
Transient expression in P. patens protoplasts was used to characterize, within the DFR, LDOX, TT12, TT19 and AHA10 promoters, the DNA regions (modules) involved in the regulation of their activity by the MBW complexes. The TT2–TT8–TTG1 complex was chosen as a model. First, a 5′-end deletion series was generated and assayed using GFP as reporter gene (Fig. 6). The different promoter fragments used in this experiment were selected in accordance with the position of the putative R2R3-MYB and bHLH binding sites that were present on the studied promoters, with the aim to remove one putative target of the MBW complex in each of the generated deletions. This search was carried out using the PLACE web tool (http://www.dna.affrc.go.jp/PLACE/; Higo et al., 1999). For this search, only the MYB core (CNGTTA/G) and AC-rich DNA motifs were considered as putative TT2 (or MYB5) binding sites, as similar cis-elements have been identified in the promoter of several genes regulated by specific R2R3-MYB involved in secondary metabolism, or by the TT2–TT8–TTG1 complex (Grotewold et al., 1994; Patzlaff et al., 2003; Hartmann et al., 2005; Prouse & Campbell, 2012; Thévenin et al., 2012; Xu et al., 2013). By contrast, all the consensus bHLH binding sequences (CANNTG) were searched. The effect of the MBW complex on each deletion was quantified relative to the activity of the longest promoters. From this approach, key deletion fragments (corresponding to drastic changes in GFP intensity) were then selected and assayed in planta using the GUS reporter gene. This strategy was chosen as it has been used successfully for the study of the BAN and TT8 promoters, for which cis-target motifs have been identified and characterized (Debeaujon et al., 2003; Thévenin et al., 2012; Xu et al., 2013).
Deletions of the DFR promoter fragment did not have a significant impact on its activity until −302 bp (proDFR-3) before the TIS site. Removal of an additional fragment of 84 bp (proDFR-4) led to a strong decrease (≈45%) in the activity (Fig. 6a). Consistent with this result, when assayed in planta, proDFR-4 did not display any GUS activity. By contrast, blue staining was still found in PA-accumulating cells when proDFR-3 was tested, indicating that this fragment contains the minimal information necessary to specifically drive transcriptional activity in this tissue. Similarly, LDOX promoter deletion analysis led to the identification of a 336-bp promoter fragment upstream of the ATG, namely proLDOX-3 (Fig. 6b). Intriguingly, proTT12-3 (−464 bp) was identified as containing the minimal nucleotide sequence that specifically drives TT12 expression in PA-accumulating cells, in contrast with the observations in transient expression assays (Fig. 6c). Such results suggest that some domains upstream of proTT12-5 have a negative impact on TT12 promoter activity in P. patens protoplasts (i.e. the occurrence of some specific repressors able to recognize proTT12-6 in P. patens). A similar hypothesis can be made concerning the proTT19-2 (259 bp before TIS site) fragment, which confers seed expression and displays a higher activity than proTT19-1 in transient expression assays (Fig. 6d). Finally, this strategy has allowed the identification of a 328-bp promoter fragment (proAHA10-2) that specifically drives AHA10 expression in PA-accumulating cells (Fig. 6e).
As the longest DFR, LDOX and TT19 promoters were also active in WT seedlings, proDFR-3, proLDOX-3 and proTT19-2 were assayed in this tissue. As expected, these three promoter fragments were sufficient to drive GUS activity in the cotyledon margins and the upper part of the hypocotyls in cells in which anthocyanins accumulate, whereas shorter promoter versions (i.e. proDFR-4, proLDOX4 and proTT19-3) were not (Fig. S4).
Identification of putative cis-regulatory elements
The characterization of short promoter domains able to confer MBW regulation in seeds and seedlings revealed the presence of various R2R3-MYB and bHLH binding sites (Fig. S5). When the sequences of the different classes of cis-regulatory motif were compared with each other (Fig. S5), two subgroups per class were identified as follows: MYB core, CC/TGTTA and CA/CGTTG; AC-rich element, A/CCCAACC/G and ACCTAA/C (ACI or ACIII); bHLH binding site, CANNTG (E-box) and CACGTG (G-box). At least one conserved DNA motif belonging to the three classes (i.e. MYB core, AC-rich element and bHLH binding site) of cis-regulatory elements was found for all the short functional promoters, with the exception of proLDOX-3, for which no bHLH binding site was found (Table S2).
Our main objective was to carry out a comprehensive functional analysis of the MBW complexes involved in PA biosynthesis and their target gene promoters in Arabidopsis thaliana seeds. Because most of the studied structural genes are also involved in the anthocyanin pathway, we looked by extension for similar regulatory mechanisms in seedlings (as there is no anthocyanin accumulation in seeds). For this purpose, genetic, molecular and biochemical approaches were used. In this study, we have identified the direct targets of the complexes, functionally characterized their promoters and determined the minimal promoters involved in the regulation by MBW complexes. The results presented here demonstrate that the LBGs, as well as TT12, TT19 and AHA10, are direct targets of the TT2–TT8–TTG1 complex and other TTG1-dependent transcriptional regulatory complexes involving MYB5, GL3 and/or EGL3. These results are discussed below.
PA biosynthesis is regulated by more than one MBW complex in Arabidopsis
GUS staining analysis gained from the study of the DFR, LDOX, BAN, TT12, TT19 and AHA10 promoters in immature seeds showed that the promoter of these six genes is active in PA-accumulating cells (i.e. endothelium, micropyle and chalaza) in a TTG1-dependent manner (Fig. 2). This result confirms that the tissue-specific expression of these six genes is largely controlled at the transcriptional level. Interestingly, the weak promoter activities remaining in the tt2 and/or tt8 mutants, but not in ttg1, suggest that (with the exception of the TT19 promoter) functionally redundant R2R3-MYB and R/B-like bHLH proteins are involved in the transcriptional regulation of this specific set of promoters. Such TTG1-dependent redundant transcriptional activities have been reported previously in the regulation of trichome and root hair patterning, anthocyanin and PA biosynthesis, and mucilage production in seed (Heim et al., 2003; Zhang et al., 2003; Lepiniec et al., 2006; Gonzalez et al., 2008, 2009; Ishida et al., 2008; Pesch & Hulskamp, 2009; Schiefelbein et al., 2009; Dubos et al., 2010; Feller et al., 2011; Xu et al., 2013).
Taking into account all of these results, we made the assumption that the functional homologues of TT2 and TT8 in the regulation of the DFR, LDOX, BAN, TT12, TT19 and AHA10 promoter activities could be MYB5, and EGL3/bHLH002 and/or GL3/bHLH001, respectively (Figs 4, 5). By combining transient expression assays, quantification of mRNA accumulation and promoter analysis in multiple loss-of-function mutants, we demonstrated that EGL3 was able to fully replace TT8 in the MBW complex to activate the DFR, LDOX, BAN, TT12, TT19 and AHA10 promoters, whereas GL3 was only able to slightly activate the DFR promoter. Interestingly, EGL3 has also been shown to play a predominant role in the regulation of anthocyanin biosynthesis when plants are grown in non-stressful conditions, with TT8 and GL3 being involved to a lesser extent (Zhang et al., 2003; Cominelli et al., 2008; Feyissa et al., 2009). MYB5 was also able to induce the DFR, LDOX, BAN, TT12, TT19 and AHA10 promoter activities, but to a lesser extent than TT2 (Fig. 4). Nevertheless, this result is consistent with genetic analyses showing that TT2 is necessary for PA biosynthesis in seeds and that, by contrast, MYB5 plays a secondary role (i.e. no PAs are synthesized in tt2, whereas decreased PA accumulation is observed in myb5; Nesi et al., 2001; Gonzalez et al., 2009). This observation is strengthened by the lack of GUS activity driven by the promoters DFR, LDOX and TT12 in the seeds of the tt2 myb5 double mutant, and by the ability of MYB5 to partially complement the transparent testa phenotype of tt2 when its expression is increased in cells that specifically accumulate PAs (Fig. 5).
Altogether, these data show that the TT2–TT8–TTG1 complex plays the main role in the regulation of DFR, LDOX, BAN, TT12, TT19 and AHA10 expression in developing seeds, although three additional MBW complexes have partially overlapping transcriptional regulatory function. In the endothelium, the MYB5–TT8–TTG1 complex is involved in the regulation of DFR, LDOX, TT12 and AHA10 expression. In the chalaza, two complexes are at play, namely the TT2–EGL3–TTG1 complex, which induces DFR, LDOX, BAN, TT12 and AHA10, and the TT2–GL3–TTG1 complex, which only controls the expression of DFR (Fig. 7). Interestingly, the above results are fully consistent with a recent tissue-specific transcriptomic analysis carried out on developing Arabidopsis seeds using laser capture microdissection coupled with GeneChip analysis (ATH1 Affymetrix; Le et al., 2010). EGL3 and MYB5 mRNA preferentially accumulate from pre-globular to heart stage of embryo development (i.e. when PA biosynthesis is occurring) in the chalaza and the endothelium, respectively (Figs S6, S7). Moreover, the results presented in this study are congruent with previously published data in which the promoter activity of the MBW complex members, as well as the mRNA or protein accumulation of their direct target (as revealed by in situ hybridization or GFP fusion), was found to be specific to the PA-accumulating cells in Arabidopsis developing seeds (Devic et al., 1999; Debeaujon et al., 2001, 2003; Baudry et al., 2006; Gonzalez et al., 2009; Li et al., 2009; Kitamura et al., 2010; Appelhagen et al., 2011b; Xu et al., 2013).
Interestingly, the analyses carried out in seedlings suggest that similar redundancies occur in vegetative tissues (Fig. S2). This latter observation is fully consistent with previous reports showing that mutations of the three bHLHs (namely TT8, EGL3 and GL3) or two R2R3-MYBs (PAP1 and PAP2) are necessary for the complete prevention of anthocyanin accumulation in seedlings (Zhang et al., 2003; Appelhagen et al., 2011a).
Functional analysis and regulation of the MBW target gene promoters
We have shown that DFR, LDOX, BAN, TT12, TT19 and AHA10 are the direct targets of different MBW complexes. Nevertheless, how the complexes control the transcription of this defined set of target genes remains to be addressed. Functional analyses of the promoters were undertaken using both transient and stable expression assays in moss (P. patens) protoplasts and transgenic plants (Arabidopsis), respectively (Fig. 6, Table S2). These approaches allowed the identification of a short promoter fragment (regulatory module) conferring an MBW-dependent expression in seed for each of the six genes (Fig. 6, Table S2).
Sequence analyses revealed that all the regulatory modules contain at least three types of conserved binding site, a MYB core (CNGTTR), an AC-rich element ([A/C]CC[A/T]A[A/C]) and a bHLH binding sequence (G-box [C/G]ACGT[A/G] and/or E-box CANNTG), with the exception of proLDOX-3, for which no obvious bHLH binding site was found (Devic et al., 1999; Debeaujon et al., 2003; Patzlaff et al., 2003; Hartmann et al., 2005; Dare et al., 2008; Prouse & Campbell, 2012; Thévenin et al., 2012; Table S2). Interestingly, a similar finding has been reported recently for the promoter of TT8, another target of the MBW complexes (Xu et al., 2013). Briefly, the promoter of TT8 contains two domains through which the MBW complexes regulate its expression in seeds, with one of them not displaying any already described MYB and bHLH binding sites. Similarly, this study demonstrates that the BAN promoter also contains at least two MBW-regulated domains, with partially redundant functions. Indeed, although the minimal BAN promoter is totally inactive in tt2, tt8 or ttg1 (Fig. 2), a clear GUS activity was detected in the chalaza area of the tt8 mutant when a 2-kb promoter fragment was used (Debeaujon et al., 2003).
Extending the LBG group to TT12, TT19 and AHA10 in Arabidopsis thaliana seed
The results presented here show that TT12, TT19 and AHA10 display the same patterns of mRNA accumulation, promoter activity and regulation by the TT2–TT8–TTG1 complex as the LBGs (Figs 1, 2, S7). Moreover, similar to the LBGs, TT12, TT19 and AHA10 have been shown to be direct targets of TTG1-dependent transcriptional regulatory complexes (Fig. 3). The EBGs and LBGs were initially categorized accordingly to their coordinate expression in response to environmental cues, such as light, at distinct developmental stages, in a species-dependent manner (Pelletier et al., 1999; Lepiniec et al., 2006). As such, the results presented above strongly support the idea that TT12, TT19 and AHA10 should be included in the LBG group, at least when referring to Arabidopsis seed development. This assumption is reinforced by the analysis of the GUS pattern driven by the DFR, LDOX, TT19 and AHA10 promoters in seedlings, which is specific to the cotyledon margins, and strictly dependent on TTG1 activity (Fig. S2).
This study also allowed us to refine our understanding of the transcriptional role played by the MBW complexes in the regulation of the flavonoid biosynthetic pathway in Arabidopsis seeds. Indeed, we found that the MBW complexes control specifically the transcription of the extended LBG group, but, more interestingly, we also found that this regulation occurs in a gene- and tissue-specific manner (Fig. 7). The TT2–TT8–TTG1 complex plays the major role in the development of seeds by regulating DFR, LDOX, BAN, TT19, TT12 and AHA10 in all the PA-accumulating cells (i.e. micropyle, chalaza and endothelium), whereas the MYB5–TT8–TTG1 complex is only active in the endothelium where it regulates DFR, LDOX and TT12 expression. Finally, the TT2–EGL3–TTG1 and TT2–GL3–TTG1 complexes regulate the expression of DFR, LDOX, BAN and AHA10, and DFR, respectively, in the chalaza.
Taken together, the results presented here highlight the complexity and robustness of the mechanisms involved. Several partially overlapping transcriptional regulations have been characterized involving functionally redundant DNA cis-elements and/or trans-acting factors. Further analyses of the structure–function relationships existing between the DFR, LDOX, BAN, TT12, TT19 and AHA10 promoters and the MBW complexes will be necessary to understand how they interact at the molecular level (e.g. 3′-end deletions and point mutation analyses of the promoters, chromatin immunoprecipitation experiments or electrophoresis mobility shift assay). This knowledge should help to identify new direct target genes involved in the flavonoid pathway that cannot be easily or directly detected by classical genetic approaches.
We thank Dr N. Terrier, Dr K. Hematy and Dr J. M. Routaboul for their useful comments. We thank the ‘Plateforme de cytologie et imagerie végétale (PCIV)’ from the Institut Jean-Pierre Bourgin (IJPB) for excellent technical support. The egl3 tt8 and egl3 gl3 tt8 mutant seeds were kindly provided by Dr Alan Lloyd. W.X.'s work was generously supported by the China Scholarship Council (CSC) of China.