Regulation of flavonoid biosynthesis involves an unexpected complex transcriptional regulation of TT8 expression, in Arabidopsis

Authors

  • Wenjia Xu,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Damaris Grain,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
    • These authors contributed equally to this work.
  • José Le Gourrierec,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Erwana Harscoët,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Adeline Berger,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Vincent Jauvion,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Aurélie Scagnelli,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Nathalie Berger,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Przemyslaw Bidzinski,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Zsolt Kelemen,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Fabien Salsac,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Antoine Baudry,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Jean-Marc Routaboul,

    1. INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    2. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Loïc Lepiniec,

    Corresponding author
    1. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    • INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author
  • Christian Dubos

    Corresponding author
    1. AgroParisTech, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Current affiliation:
    1. Biochimie et Physiologie Moleculaire des Plantes, Montpellier Cedex 1, France
    • INRA, Institut Jean-Pierre Bourgin, Saclay Plant Sciences, Versailles, France
    Search for more papers by this author

Authors for correspondence:

Christian Dubos

Tel: +33 (0)4 9961 2613

Email: christian.dubos@versailles.inra.fr

Loïc Lepiniec

Tel: +33 (0)1 30 3686

Email: loic.lepiniec@versailles.inra.fr

Summary

  • TT8/bHLH042 is a key regulator of anthocyanins and proanthocyanidins (PAs) biosynthesis in Arabidopsis thaliana. TT8 transcriptional activity has been studied extensively, and relies on its ability to form, with several R2R3-MYB and TTG1 (WD-Repeat protein), different MYB-bHLH-WDR (MBW) protein complexes. By contrast, little is known on how TT8 expression is itself regulated.
  • Transcriptional regulation of TT8 expression was studied using molecular, genetic and biochemical approaches.
  • Functional dissection of the TT8 promoter revealed its modular structure. Two modules were found to specifically drive TT8 promoter activity in PA- and anthocyanin-accumulating cells, by differentially integrating the signals issued from different regulators, in a spatio-temporal manner. Interestingly, this regulation involves at least six different MBW complexes, and an unpredicted positive feedback regulatory loop between TT8 and TTG2. Moreover, the results suggest that some putative new regulators remain to be discovered. Finally, specific cis-regulatory elements through which TT8 expression is regulated were identified and characterized.
  • Together, these results provide a molecular model consistent with the specific and highly regulated expression of TT8. They shed new light into the transcriptional regulation of flavonoid biosynthesis and provide new clues and tools for further investigation in Arabidopsis and other plant species.

Introduction

Flavonoids are secondary metabolites that play important roles throughout the plant life cycle. For example, flavonoids were shown to be involved in the attraction of pollinators or seed dispersers, in defence against biotic or abiotic stresses and as in the regulation of plant growth and development (Lepiniec et al., 2006; Petroni & Tonelli, 2011; Grunewald et al., 2012). In Arabidopsis thaliana three main classes of flavonoid compounds are accumulated: anthocyanins in the vegetative tissues, proanthocyanidins (PAs) in seed integuments and flavonols in both, seeds and vegetative tissues.

The expression of the genes that encode enzymes involved in the specific steps that lead to the biosynthesis of anthocyanins and PAs (i.e. the late biosynthetic genes) are controlled by the combined action of specific R2R3-MYB (subgroup 5 and 6) and R/B-like bHLH (subgroup IIIf) transcription factors (TF), together with TTG1 (WD repeat protein), in a MYB-bHLH-WDR (MBW) ternary protein complex (Heim et al., 2003; Baudry et al., 2004; Lepiniec et al., 2006; Dubos et al., 2010; Thevenin et al., 2012). TT2/MYB123, TT8/bHLH042 and TTG1 have been demonstrated as being the three main regulators of PA biosynthesis (Baudry et al., 2004; Lepiniec et al., 2006). By contrast, the regulation of anthocyanin biosynthesis was shown to involve a larger set of TFs, which includes PAP1/MYB75, PAP2/MYB90, GL3/bHLH001, EGL3/bHLH002, TT8/bHLH042 and TTG1 (Zhang et al., 2003; Baudry et al., 2006; Feyissa et al., 2009; Appelhagen et al., 2011a). Similar MBW complexes have been also characterized in other plant species such as maize or petunia (Hernandez et al., 2004; Koes et al., 2005; Albert et al., 2011; Petroni & Tonelli, 2011). Transcriptional TTG1-dependent activities are not restricted to the regulation of flavonoid biosynthesis as, for example, additional MBW complexes involving R2R3-MYB belonging to subgroup 15 (in combination with GL3 and EGL3) were shown to regulate trichome formation and root hair patterning (Bernhardt et al., 2005; Zhao et al., 2008; Kang et al., 2009; Schiefelbein et al., 2009; Dubos et al., 2010; Song et al., 2011). The activity of MBW complexes is negatively regulated by a group of single MYB repeat proteins, the R3-MYBs. This group includes MYBL2 a negative regulator of anthocyanin biosynthesis in vegetative tissues that can repress PA biosynthesis when overexpressed in seeds (Dubos et al., 2008). Similarly to MYBL2, CPC (CAPRICE) has been proposed to regulate anthocyanin biosynthesis in response to nitrogen deprivation (Zhu et al., 2009). Moreover CPC, together with six additional R3-MYB, was shown to be involved in the regulation of trichomes and root hair patterning (Wang et al., 2008; Gan et al., 2011).

The regulation of PA biosynthesis involves some additional transcriptional regulators that belong to other families of TF, namely TT1/WIP1 (Zn finger), TT16/ABS/AGL32 (MADS box) and TTG2/DSL1/WRKY44, for which the precise mode of action remain to be elucidated (Johnson et al., 2002; Nesi et al., 2002; Sagasser et al., 2002; Lepiniec et al., 2006). It is hypothesized that TT1 and TT16 regulate the competency and the differentiation, respectively, of cells that are destined to accumulate Pas, whereas TTG2 is proposed to control the overall seed coat development, including the integument cell elongation process (Johnson et al., 2002; Garcia et al., 2005).

If the regulation of PA and anthocyanin biosynthesis by the MBW complexes is currently relatively well described and understood, the means by which the expression of the transcriptional regulators is regulated have been less investigated. It has been demonstrated that some environmental cues such as light intensity or temperature influence anthocyanin accumulation in the vegetative tissues by modulating the expression of the R2R3-MYB (e.g. PAP1 and PAP2) and R/B-like bHLH (e.g. TT8, GL3 and EGL3) TF (Dubos et al., 2008; Olsen et al., 2009; Rowan et al., 2009). Similarly, the carbon to nitrogen balance as well as the concentration of certain phytohormones, such as jasmonic acid, gibberelins or cytokinins, are also important factors that have an impact on the expression of these TF (Lea et al., 2007; Maes et al., 2008; Feyissa et al., 2009). By contrast, the regulation of TT2 and TT8 expression in seeds appeared to be developmentally regulated (Lepiniec et al., 2006). However, the precise nature of the molecular mechanisms that govern the expression of these anthocyanin and PA biosynthesis regulators is still elusive.

The regulation of TT8 expression was chosen as a first step in the elucidation of this fundamental question because of the key role of TT8 in the regulation of both anthocyanin and PA biosynthesis in different tissues. In a previous study, it was found that TT8 was regulating its own expression through a positive feedback regulatory loop, which implied a TTG1-dependent regulatory mechanism (Nesi et al., 2001; Baudry et al., 2006). It was also proposed that different MBW complexes involving closely related bHLH (i.e. EGL3 and GL3) could regulate TT8 expression in the mucilage cell layer of the seed coat and in the margins of seedling cotyledons (Baudry et al., 2006; Gonzalez et al., 2008). Additional studies revealed that TT8 mRNA accumulation in developing siliques was controlled by TT2 and MYB5 (Gonzalez et al., 2008). Together, these data indicate that more than one MBW complex is involved in the regulation of TT8 expression in both seeds and vegetative tissues.

With the aim of providing a comprehensive model for the complex pattern of TT8 expression, we first carried out a functional dissection of the TT8 promoter, revealing its modular structure. Two modules were sufficient to drive TT8 expression in PA- and anthocyanin-accumulating cells, whereas the third one was regulating the strength of the promoter activity. The activity of these two-first modules was assayed in different regulatory mutants affected in the formation of the MBW complexes, as well as in tt1, tt16, or ttg2. Each of the regulatory modules integrated the regulatory signals from these different regulators in a tissue-specific manner. We also found that TT1, TT16 and TTG2 were involved in the regulation of TT8 expression, and that some new regulatory factors are most probably yet to be discovered. Furthermore, this approach revealed that six different MBW complexes are involved in the regulation of TT8 promoter activity, with three being specifically dedicated to the PA-accumulating cells in seeds and the other three to the cotyledon margins where anthocyanins accumulate. Finally, specific cis-regulatory elements through which TT8 expression is regulated have been identified and characterized.

Materials and Methods

Histochemical detection of β-glucuronidase (GUS) activity, fluorimetric (quantitative) GUS analyses, microscopy, Physcomitrella patens transfection assays and yeast one-hybrid experiments were carried out as described elsewhere (Jefferson et al., 1987; Baud et al., 2003; Debeaujon et al., 2003; Baudry et al., 2006; Berger et al., 2011; Thevenin et al., 2012). All methods and conditions used for plant growth, plant transformation and selection for transgenic lines were as previously reported (Nesi et al., 2000).

Plant material

The following Arabidopsis thaliana (L.) Heynh. mutants were used in this study: tt2-1 (CS83), myb5-1 (salk_030942), tt2 myb5-1 (salk_005260, salk_030942), pap1 (PST_16228), pap2 (salk_093731), tt8-3 (DEB122), egl3-1 tt8-1, egl3-1 gl3-1 tt8-1, ttg1-1 (CS89), tt1-1 (CS82), tt16-1 (DXT32), ttg2-4 (CTA18).

Arabidopsis gene IDs studied

MYB5, At3g13540; PAP1/MYB75, At1g56650; PAP2/MYB90, At1g66390; TT2/MYB123, At5g35550; GL3/bHLH001, At5g41315; EGL3/bHLH002, At1g63650; TT8/bHLH042, At4g09820; TTG1, At5g24520; TTG2/WRKY44/DSL1, At2g37260; TT1/WIP1, At1g34790; TT16/AGL32/ABS, At5g23260.

Constructs

All the PCR products were obtained using high-fidelity Phusion DNA polymerase with HF buffer (Thermo Scientific–Finnzymes, Waltham, MA, USA). The primers (from Sigma-Aldrich) used in this study are described in the Supporting Information, Table S1. Mutagenesis of proTT8-1 and proTT8-6 were carried out using the QuikChange Site-Directed Mutagenesis Kit according to manufacturer instructions (Stratagene-Agilent, Massy, France). The integrity of each construct was ensured by DNA sequencing. Following Arabidopsis transformation, between 12 and 36 independent lines were assayed for GUS activity, for each construct.

proTT8:GUS fusions

The TT8 promoter fragments used for the deletion studies were amplified from the Wassilewskija Arabidopsis accession. The position of each fragment refers to the 3′-end of the previously described ‘full length’ TT8 promoter (Baudry et al., 2006). These DNA fragments were then introduced into the pDONR207 vector by a BP recombination (Invitrogen), and finally transferred into the binary vector pBI-101GUS containing the Gateway cassette (Figs 1, 5) or in pGWB3 (Figs 2b, 3b,c), by a LR recombination (Baudry et al., 2006; Nakagawa et al., 2007). The ‘full length’ TT8 promoter fused to GUS was introduced into the different mutants background by crossing (Figs 2a, 3a). The 3′-end deletion fragments were fused by PCR (Table S1) to a modified minimal 35S promoter from cauliflower mosaic virus (from the pRTL2 vector) carrying a transcription start site, prior to the BP recombination.

Figure 1.

Molecular and functional dissection of the TT8 promoter. (a) Scheme of 5′-end deletion constructs. (b) Quantitative β-glucuronidase (GUS) analysis of the 5′-end deletion constructs. t-test significance at P < 0.01 when compared to (a) proTT8-1, (b) proTT8-2 and (c) proTT8-6. Error bars + SE. (c) The GUS activity driven by the 5′-end deletion constructs in Arabidopsis thaliana seeds (globular stage) and seedlings. m, micropyle; c, chalaza; endo, endothelium; emb, embryo. (d) Scheme of 3′-end deletion constructs. (e) The GUS activity driven by the 3′-end deletion constructs in seeds (globular stage) and seedlings. Bars: 0.1 mm (seeds), 2 mm (seedlings). The ‘full length’ TT8 promoter was used as reference for numbering. + and − refer to the visual observation of the GUS strength.

Figure 2.

TT8 promoter activity in tt1, tt16 and ttg2 Arabidopsis thaliana mutant seeds. Analysis of the β-glucuronidase (GUS) activity driven by the (a) ‘full length’ promoter fragment in seeds or (b) proTT8-1 in comparison with, proTT8-6 and proTT8-17 (globular stage), in the tt1, tt16 and ttg2 backgrounds. Bars: 0.1 mm. m, micropyle; c, chalaza; endo, endothelium. (c) Schematic regulation of TT8 promoter activity by TT1, TT16 and TTG2. Solid arrows, link previously demonstrated; dashed arrows, regulations identified in this study; dotted arrows, putative regulations. The ‘full length’ TT8 promoter was used as reference for numbering.

Figure 3.

TT8 promoter activity in R2R3-MYB (tt2, myb5, pap1 and pap2), R/B-like bHLH (tt8, gl3 and egl3) and ttg1 Arabidopsis thaliana mutants. The activity of the (a) ‘full length’ promoter fragment was studied in seeds (globular stage), whereas proTT8-1, proTT8-6 and proTT8-17 activities were studied in (b) seeds (globular stage) and (c) seedlings. m, micropyle; c, chalaza; endo, endothelium; hyd, hydathodes. Bars: 0.1 mm (seeds), 2 mm (seedlings).

proTT8:green fluorescent protein (GFP) fusions

The two TT8 promoter fragments were used in P. patens transient expression assays were recombined into the pBS TPp-B vector (Thevenin et al., 2012). In order to be able to compare the results obtained for both promoters, the proTT8-6 TATA box was removed and replaced by the 35S minimal promoter, as in proTT8-17 (Fig. 1d).

Coding sequence (cds) cloning and expression

MYB5, PAP1, GL3 and EGL3 cds were amplified from cDNA that were reverse transcribed from mRNA extracted from 4-d-old seedlings grown in vitro on Murashige and Skoog (MS) media supplemented with 3% sucrose. The fragments obtained were then recombined first into the pDONR207 vector and subsequently into pBS TPp-A expression vector (Thevenin et al., 2012). Expression of TT2, TT8 and TTG1 in P. patens protoplast and yeast was carried out as described elsewhere (Baudry et al., 2006; Thevenin et al., 2012).

Results

Functional dissection of TT8 promoter

In order to get better insight into how TT8 expression is regulated, a 5′-end promoter deletion series, every c. 100 bp on a 1005 bp DNA fragment, was carried out (Fig. 1a–c). These different TT8 promoter fragments were fused to the uidA (GUS, β-glucuronidase) reporter gene, and transferred into wild-type (WT) plants. The first six distal promoter constructs (until proTT8-6, −511 bp) displayed the same GUS pattern in seeds and seedlings as the previously studied, and so-called ‘full length’, promoter fragment (1518 bp, Fig. 1a,c; Baudry et al., 2006). By contrast, no GUS activity was detected when proTT8-7 (−356 bp) fragment was tested (Fig. 1a–c). This experiment demonstrated that proTT8-6 (−511 bp) was sufficient to specifically drive GUS activity in PA- and anthocyanin-accumulating cells in seeds and seedlings, respectively. In order to obtain further insights into the observed decrease in GUS staining (Fig. 1a), the activity of TT8 promoter constructs was quantified in 4-d-old siliques at the time when the BAN gene (a direct target of TT8) reaches its peak of expression (Debeaujon et al., 2003; Baudry et al., 2004). The activity was clearly affected in a promoter length-dependent manner (Fig. 1b). This experiment revealed that 50% of proTT8-1 (−1005 bp) activity was lost with proTT8-2 (−905 bp) and that an additional 40% of proTT8-1 activity was lost with proTT8-3 (−821 bp). Finally, this experiment also revealed that only 10% of proTT8-1 (−1005 bp) activity was driven by proTT8-6 (−511 bp). These results showed that some sequences located between −1005 and −821 bp were involved in the strength of TT8 promoter. As a second step, a 3′-end deletion series, between −1005 and −511 bp, was carried out. This approach led to the analysis, in WT plants, of 11 additional TT8 promoter fragments (proTT8-8 to proTT8-18) placed upstream of a minimal 35S promoter (Fig. 1d,e). The results demonstrated that a second promoter fragment located between −821 and −614 bp (corresponding to proTT8-17) was sufficient to reproduce the activity of the previously studied ‘full length’ TT8 promoter fragment (Baudry et al., 2006). These analyses showed that a first module (located between −1005 and −821 bp) is responsible for the strength of the TT8 promoter, whereas two others modules (i.e. proTT8-6 and proTT8-17) control its tissue specificity (Fig. 1).

Determination of the genetic relationships existing between TT8 promoter activity and the TT1/WIP1, TT16/ABS/AGL32 and TTG2/WRKY44 transcription factors (TFs)

When the previously characterized ‘full length’ TT8 promoter fragment fused to GUS was introduced into tt1, no difference with the WT GUS pattern was observed in micropyle, chalaza or endothelium, suggesting that TT1 was not playing a major role in the control of TT8 promoter activity (Fig. 2a). Nevertheless, in a tt1 mutant background, the activities of both proTT8-6 (−511 bp) and proTT8-17 (−821 to −614 bp) were restricted to the chalazal area and the distal part of the endothelium, respectively, whereas proTT8-1 (−1005 bp) was found to be active in both tissues, but not in micropyle (Fig. 2b). This later result revealed TT1-independent activity in the micropyle of the region located upstream the −1005 bp DNA fragment. Thus, although TT1 is not necessary for basal activity of the TT8 promoter in PA-accumulating cells (Fig. 2c), it modulates this activity by acting on the different modules in a tissue-specific manner.

In a tt16 mutant background, the activity of the ‘full length’ promoter was essentially restricted to the micropyle and the chalazal areas, although some weak activity was also observed in the endothelium in later developmental stages (from heart stage onward, Fig. 2a). This GUS pattern perfectly matches the pattern of PA accumulation observed in this mutant (Nesi et al., 2002). Similarly, proTT8-1 (−1005 bp) activity was detected in the same two tissues, whereas the activity of proTT8-6 (−511 bp) and proTT8-17 (−821 to −614 bp) were restricted to the chalaza and the micropyle, respectively (Fig. 2b). These data indicated that TT16 was necessary for the activity of TT8 promoter in the endothelium, and that TT16 signalling was integrated through the proTT8-6 and proTT8-17 modules (Fig. 2c).

In the ttg2 mutant the activity of the ‘full length’ TT8 promoter was restricted to the chalaza area at early stage of seed development (globular) and was detected in all the PA-accumulating cells in the later stages (from heart stage onward; Fig. 2a). Consistent with this result, proTT8-1 (−1005 bp) activity is observed in the chalaza strand (Fig. 2b). When proTT8-6 (−511 bp) and proTT8-17 (−821 to −614 bp) were assayed in ttg2, no blue staining was detected (Fig. 2b). Together, these data showed that the activity of the TT8 promoter in the endothelium was strictly dependent on TTG2 early during seed development, but this was not the case later during seed maturation or in the chalaza. Moreover, the detection of GUS activity in chalaza with proTT8-1 (−1005 bp) suggested that either TTG2 regulation in this tissue was independent of proTT8-6 and proTT8-17 or that a threshold effect was at play (Fig. 2c).

Regulation of TT8 promoter activity by the MYB-bHLH-WDR (MBW) protein complexes

It has been previously shown that TT2 controls the specific expression of TT8 in PA-accumulating cells through its interaction with TTG1 and TT8 itself (Baudry et al., 2006; Lepiniec et al., 2006). Therefore, the activity of the TT8 promoter was assayed in tt2 (as well as in tt8 and ttg1 as controls). This experiment confirmed that the activity of the TT8 promoter was restricted to the chalaza in both tt8 and ttg1 mutants (Fig. 3a). In tt2 seeds, although no activity was detected in micropyle and chalazal areas, it was apparently unaffected in the endothelium. These results suggest that putative redundant MBW complexes regulate TT8 promoter in a partially overlapping manner in seeds (Fig. 3a). In order to determine which MBW complexes could be involved, the promoter activity of proTT8-1 (−1005 bp), proTT8-6 (−511 bp) and proTT8-17 (−821 to −614 bp) were assayed in different mutant backgrounds (Fig. 3b). proTT8-1 (−1005 bp) and proTT8-17 (−821 to −614 bp) conferred GUS activity in the endothelium when expressed in tt2, whereas no more activity was detected in the myb5 tt2 double mutant. The mutation of tt2 was sufficient to block the activity of proTT8-6 (−511 bp). In tt8, some blue staining was observed in the chalaza when proTT8-1 (−1005 bp) and proTT8-6 (−511 bp) were assayed, which disappeared in the tt8 egl3 double mutant. Conversely, proTT8-17 (−821 to −614 bp) was inactive in tt8. Finally, none of the three promoters was active in ttg1.

In 4-d-old seedlings the activities of these promoters were found to be dependent on PAP1 (and independent of PAP2; Fig. 3c). In addition, proTT8-6 (−511 bp) and proTT8-17 (−821 to −614 bp) were regulated by TT8, EGL3, and TTG1. Interestingly, proTT8-1 (−1005 bp) was still active in the hydathodes of the tt8 egl3 double mutant, but not in the tt8 egl3 gl3 triple mutant, suggesting that a threshold effect involving GL3 was at play in this tissue (Figs 3c, S1). In ttg1, proTT8-1 (−1005 bp) was also active in the hydathodes, suggesting that either some TTG1-independent regulators may be involved in the regulation of this promoter in these specific cells, or that some activities could be driven by the R2R3-MYB and R/B-like bHLH factors independently of TTG1. This second hypothesis implies a threshold effect as no GUS activity was detected in ttg1 seedlings carrying either the proTT8-6 (−511 bp) or the proTT8-17 (−821 to −614 bp) construct (Fig. 3c). Together, these data exemplified how different MBW complexes regulate TT8 activity in both, seeds and cotyledons.

In order to provide molecular support to the genetic analyses the different putative MBW complexes were assayed in transient activation assays in P. patens protoplasts using proTT8-6 (−511 bp) and proTT8-17 (−821 to −614 bp) fused to the GFP reporter gene (Fig. 4a). Physcomitrella patens protoplasts were used as this system was previously described as a fast, sensitive and robust method to carry out transient expression assays (Thevenin et al., 2012). This approach allows quantitative analysis of gene expression using up to four different regulatory proteins in one experiment. In our assays, three R2R3-MYBs (i.e. TT2, MYB5 and PAP1) were tested in combination with three R/B-like bHLHs (i.e. TT8, EGL3 and GL3) and TTG1. For both modules the strongest activation was obtained with TT2–TT8–TTG1 and TT2–EGL3–TTG1 complexes. The other seven possible complexes were able to significantly activate both promoters, but to a much lower extent. Interestingly, R2R3-MYBs together with the R/B-like bHLH TFs were able to slightly activate both modules, independently of TTG1 (Fig. S2). This observation is consistent with previous results obtained using a BAN promoter in a yeast one-hybrid experiment and in transient activation assays in P. patens protoplasts (Baudry et al., 2004; Thevenin et al., 2012).

Figure 4.

Regulation of TT8 expression by the MBW complexes. (a) Green fluorescent protein (GFP) intensity (arbitrary units) measured in Physcomitrella patens protoplasts co-transfected with R2R3-MYB (TT2, MYB5 or PAP1), R/B-like bHLH (TT8, GL3 and EGL3), TTG1 and either proTT8-6 or proTT8-17. Error bars ± SE. t-test significance: *, P < 0.01. (b,c) Schematic representation of TT8 promoter regulation by the MBW complexes in (b) seeds and (c) cotyledons. Solid arrows, interaction demonstrated in this study; dashed arrows, putative regulatory mechanism. The ‘full length’ TT8 promoter was used as reference for numbering.

The data demonstrate that diverse MBW complexes have the ability to regulate TT8 promoter activity: three MBW complexes are specific to PA-accumulating cells in seeds, namely TT2-TT8-TTG1, TT2-EGL3-TTG1 and MYB5-TT8-TTG1, whereas PAP1-TT8-TTG1, PAP1-EGL3-TTG1 and PAP1-GL3-TTG1 are specific to the cotyledon margins (Fig. 4b,c). Interestingly, these data also indicated that TT8 expression in the chalaza and cotyledon hydathodes might involve some TTG1-independent regulatory mechanisms (Fig. 4b,c).

Identification of the cis-regulatory targets of the MBW complexes

We selected proTT8-6 (−511 bp) as a model to be studied in depth in order to determine through which cis-regulatory elements the MBW complexes could regulate TT8 expression. As a starting point, proTT8-6 sequences were amplified in five additional accessions (Bay-0, Shahdara, Ler, Cvi-0 and Col-0), aligned and analysed using the PLACE web tool (Fig. S3; http://www.dna.affrc.go.jp/PLACE/, Higo et al., 1999) with the aim of identifying conserved putative MYB and bHLH binding sites. This analysis led to the identification of two putative binding sites (MYB-core binding site,CAGTTA, and G-box like CACGTC) that display strong similarities with the cis-regulatory elements of the BAN promoter shown to be the target of the TT2–TT8–TTG1 complex (Thevenin et al., 2012).

The MYB-core and G-box sites were mutagenized and replaced by the CATACA (proTT8-6-m1) and CACGCC (proTT8-6-m2) sequences, respectively. The mutated promoters were assayed in yeast one-hybrid experiments, as described in (Baudry et al., 2006) and in transient expression assays in P. patens protoplasts (Fig. 5a,b). These experiments confirmed that TT2 and TT8 regulate proTT8-6 activity through these two putative cis-regulatory elements. The mutated promoters were subsequently tested in planta (Fig. 5c). Both promoters lack activity in seed endothelium and cotyledon margins, their activity being restricted to chalaza and cotyledon hydathodes. These data demonstrated that these two cis-regulatory elements played an important role in the regulation of TT8 promoter activity, at least in the endothelium and cotyledon margins. However, no similar binding sites were identified in proTT8-17, suggesting that MBW complexes may also recognize different types of regulatory sequences (Fig. S4).

Figure 5.

Identification and characterization of cis-regulatory elements involved in the regulation of TT8 promoter activity. (a) Yeast one-hybrid experiment. Yeast were stably transformed with wild-type (WT) or mutated proTT8-6:HIS3 constructs. These different yeast strains were then co-transfected with either the empty vectors (AD + BD) or with the same vectors containing TT2 and TT8 (TT2 + TT8). Upper panel, growth on control media deprived of W and L amino acids. Lower panel, growth on selective media deprived of W, L and H amino acids. AD, GAL4 activation domain; BD, GAL4 binding domain. (b) Green fluorescent protein (GFP) intensity measured in Physcomitrella patens protoplasts co-transfected with TT2, TT8, TTG1 and the WT or the mutated versions of proTT8-6:GFP. Error bars + SE. t-test significance: *, P < 0.01. (c) WT or mutated proTT8-6:GUS activity in WT seeds (globular stage) and seedlings. (d) Schematic representation of the position of the AC-element analysed on the WT and the mutated proTT8-1. The ‘full length’ TT8 promoter was used as reference for numbering. + and − refer to the visual observation of the β-glucuronidase (GUS) strength. (e) WT or mutated proTT8-1:GUS activity in WT seeds (globular stage) and seedlings. (f) Impact of the AC-element mutations on proTT8-6:GUS and proTT8-17:GUS activities in WT seeds and seedlings. c: chalaza, hyd: hydathode. Bars: 0.1 mm (seeds), 2 mm (seedlings).

Identification of AC-rich regulatory sequences involved in TT8 promoter activity

The implication of AC-rich sequences (AC-element, AC) in the regulation of structural genes involved in secondary metabolism has been extensively reported (Patzlaff et al., 2003; Prouse & Campbell, 2012). Two palindromic motifs, which differed by only one nucleotide and are similar to the well-described ACII MYB-binding site (ACCAACC; Patzlaff et al., 2003), were identified in proTT8-1 (−1005 bp) – the first motif in proTT8-6 (ACCAACCA; −511 bp, Fig. S3) and the second motif in proTT8-17 (ACCAAACCA; −821 to −614 bp, Fig. S4).

In order to determine if these two putative cis-regulatory sequences are involved in the regulation of TT8 promoter activity, they were mutated alone or in combination in the proTT8-1 context (Fig. 5d). The impact of these mutations was then assessed through the analysis of GUS activity in seeds and cotyledons of WT plants (Fig. 5e). The mutation of the proTT8-6 (−511 bp) element into ATGGATCA (proTT8-1-m3) did not affect activity in seeds, whereas in cotyledons the blue staining was only observed in the hydathodes. Conversely, mutation of the proTT8-17 (−821 to −614 bp) element into AATGGATCA (proTT8-1-m4) did not modify GUS activity. However, when both mutations were combined (proTT8-1-m3m4), no more blue staining was detected in either seeds or cotyledons (Fig. 5d,e). This later result suggests that these two AC-rich sequences have partly overlapping function in the regulation of proTT8-1 activity. To test this hypothesis, the m3 and m4 mutations were introduced into proTT8-6 and proTT8-17, respectively, and assayed in planta. This experiment showed that GUS activity was totally lost in seeds and seedlings of WT plants expressing these constructs, supporting the important role played by these AC-elements in the regulation of TT8 promoter activity (Fig. 5f).

Discussion

Modular structure of the TT8 promoter

Previous studies aimed at understanding the role and the mode of action of TT8 in PA accumulation during Arabidopsis seed development revealed that TT8 regulates PA biosynthesis through its involvement in a MBW ternary protein complex. In this MBW complex TT8 was shown to control the expression of both the so-called late biosynthetic genes (which encode the enzyme specifically committed to the biosynthesis of PAs and anthocyanins) and its own expression (Nesi et al., 2000; Debeaujon et al., 2003; Baudry et al., 2004, 2006). A ‘full length’ promoter sufficient to drive reporter gene expression in the PA-accumulating cells (chalaza, micropyle and endothelium) in seeds, and in the cotyledon margins (including the hydathodes) where anthocyanins accumulate in young seedlings, has been isolated (Baudry et al., 2006). Analysis of the TT8 promoter revealed that a 1 kb fragment (proTT8-1) was sufficient to drive gene expression in a similar way to the previously isolated ‘full length’ promoter. proTT8-1 is composed of at least three regulatory modules: two specifically driving gene expression in PA- and anthocyanin-accumulating cells (proTT8-6, −511 bp and proTT8-17, between −821 and −614 bp) and one influencing the overall promoter strength (Fig. 1).

The activity of the TT8 promoter is strongly connected to the seed coat differentiation

In order to explore how TT8 promoter activity is influenced by the developmental stage of the Arabidopsis seed coat, the role of three key genes, namely TT1 (Zn finger), TT16 (MADS box) and TTG2 (WRKY), involved in this process was studied (Nesi et al., 2002; Sagasser et al., 2002; Garcia et al., 2005; Appelhagen et al., 2010, 2011b). The activity of the TT8 promoter was found to be regulated by these three TFs, and the specific pattern driven by proTT8-1 (−1005 bp) and the ‘full length’ promoter, corresponded essentially to the sum of the activities driven by the proTT8-6 (−511 bp) and proTT8-17 (−821 to −614 bp) regulatory modules (Fig. 2). The fact that proTT8-1 was active in the chalaza of all three mutants, indicated that another regulation was at play in this tissue. It could be hypothesized that any of these three TF is sufficient to activate proTT8-1 in this tissue, or the involvement of another regulatory mechanism. Interestingly, this work also revealed an unpredicted positive feedback regulatory loop between TT8 and TTG2, and an unexpected regulatory link between TT1 and TT8 expression (Johnson et al., 2002; Ishida et al., 2007; Gonzalez et al., 2009; Albert et al., 2011; Appelhagen et al., 2011b). Together, these data indicate that proTT8-6 and proTT8-17 differentially integrate the developmental signals from TT1, TT16 and TTG2, and that perhaps some unknown regulators are likely to be discovered.

Several MBW protein complexes control the spatio-temporal expression of TT8

Various studies have indicated that different sets of MBW complexes regulate the biosynthesis of PAs and anthocyanins, by modulating the expression of the structural genes specifically dedicated to these biosynthetic processes. From these studies, redundant TTG1-dependent activities between proteins belonging to the same TF subgroups were identified. For example, the Arabidopsis R2R3-MYB PAP1 and PAP2 were shown to regulate anthocyanin biosynthesis in the vegetative tissues, whereas TT2 and MYB5 were proposed to regulate PA biosynthesis in seeds (Baudry et al., 2006; Gonzalez et al., 2008, 2009; Appelhagen et al., 2011a). Similarly, two additional Arabidopsis R/B-like bHLH factors, namely EGL3 and GL3, were also shown to regulate these two branches of the flavonoid pathway (Zhang et al., 2003; Baudry et al., 2006; Feyissa et al., 2009). Interestingly, TT8 expression was found to be directly regulated by TT8 itself through a positive feedback regulatory loop involving redundant MBW complexes (Nesi et al., 2001; Baudry et al., 2006; Gonzalez et al., 2008, 2009). We took advantage of the characterization of the different modules involved in the regulation of TT8 promoter activity to further explore these mechanisms. This approach revealed that the control of TT8 expression much more complex than previously hypothesized, as it involves the combined action of at least six functionally redundant MBW complexes interacting with different modules in a tissue-dependent manner (Fig. 3). This level of complexity is particularly well exemplified for the PA-accumulating cells where the sole TT2-TT8-TTG1 complex was thought to regulate TT8 expression, and for which two additional complexes have been identified (Baudry et al., 2006). It would be interesting to investigate if this level of complexity also occurs in other aspects of plant growth and development that are regulated by MBW complexes.

TT8 promoter activity is orchestrated by a highly diverse set of cis-regulatory DNA sequences

In-depth analysis of proTT8-6 (−511 bp) resulted in the identification of a functional MYB binding site similar to the one characterized in BAN (which encodes the first enzyme of the flavonoid pathway specifically dedicated to PA biosynthesis) promoter, and closely related to the canonical MYB binding site (CNGTT[G/A]; Weston, 1992; Thevenin et al., 2012). Equivalent MYB binding sites were identified on the promoter of some MBW complex target genes in maize or Arabidopsis (Roth et al., 1991; Koshino-Kimura et al., 2005; Ryu et al., 2005; Song et al., 2011). By contrast, proTT8-17 did not display any canonical MYB binding site, suggesting that divergent MYB target DNA sequences recognized by the MBW complexes remain to be identified, as has been described for other promoters (Figs S4, S5; Koshino-Kimura et al., 2005; Ryu et al., 2005; Ishida et al., 2007; Kang et al., 2009; Song et al., 2011). In addition, two AC-rich MYB binding sites (also called AC-elements) were also found to be functionally redundant and necessary for the tissue-specific activity of the TT8 promoter. Similar DNA motifs were shown in various plant species to play a central role in the regulation of the promoter activity of some structural genes involved in secondary metabolism (Patzlaff et al., 2003; Prouse & Campbell, 2012). These results suggest that another R2R3-MYB protein could participate in the regulation of TT8 promoter activity independent of any R/B-like bHLH TF, as has been shown for the maize P protein (Grotewold et al., 1994). However, it cannot be excluded that the MBW complexes, or another type of TF, may regulate TT8 promoter activity through these AC-rich sequences. On the other hand, TT8, GL3, and EGL3 were predicted, like other plant bHLH proteins to bind to the palindromic G-box DNA sequences (Heim et al., 2003; Toledo-Ortiz et al., 2003; Baudry et al., 2004; Thevenin et al., 2012). Interestingly, a G-box like sequence (CACGTC) was found to be necessary for the regulation of proTT8-6 (−511 bp) activity, suggesting that the consensus sequence to which TT8, EGL3 and GL3 bind could be CACGT[G/C], in which the ACGT core sequence is conserved (Figs 5, S3). This observation is consistent with some results obtained with Caenorhabditis elegans, suggesting that bHLH proteins may recognize divergent DNA motifs, depending on their interacting protein partners (Grove et al., 2009; De Masi et al., 2011; Feller et al., 2011). This level of complexity in the bHLH binding capacity may explain why no canonical bHLH binding site (i.e. E-box or ACGT core sequence) was found in proTT8-17 (−821 to −614 bp). These results point out that some important efforts remain to be done in order to determine the entire diversity of cis-regulatory sequences recognized by the bHLH proteins in plants, which would be of great interest in order to better understand how MBW complexes regulate their target genes. High throughput genomic approaches (e.g. ChIP-chip or ChIP-seq) would be helpful to build a comprehensive view of this regulation, although it seems that the relevance of each interaction found has to be confirmed in vivo (Morohashi & Grotewold, 2009).

Modularity and complexity of spatio-temporal regulation of promoter activity

In this study we found that the regulation of TT8 promoter activity, and consequently TT8 expression, requires different regulatory modules in the promoter, several partially redundant proteins complexes and various cis-regulatory DNA sequences to which TF bind. The partially redundant DNA regulatory modules and proteins complexes identified, probably reflect the underlying evolutionary processes. It would be interesting to investigate similar regulatory networks in distantly related plant species in order to understand the structure of the ancestral network and its evolution. Overall, the surprising complexity of this regulation indicates that the study of genes expression through the regulation of their promoter activity cannot be reduced to the search for, and the analysis of either a minimal or a full-length promoter if the aim is to fully understand the intricate relations that exist, at the spatio-temporal levels, between the various TF involved and their target DNA sequences.

Acknowledgements

We thank Dr Bertrand Dubreucq for useful discussions, and Johanne Thévenin, Louise Vaisman, Cecile Brousse and Sophie Bobet for their excellent technical support, as well as the ‘Plateforme de cytologie et imagerie végétale (PCIV)'. The egl3 tt8, egl3 gl3 tt8 and tt2 myb5 mutant seeds were kindly provided by Dr Alan Lloyd, and the pap1 and pap2 seeds by were provided Prof Bernd Weisshaar and Dr Ralph Stracke. This research was carried out as part of the EU-funded project FLAVO (Food CT-2004-513960). W.X.'s work was generously supported by the China Scholarship Council (CSC).

Ancillary