Functional properties of an alternative, tissue-specific promoter for rice NADPH-dependent dihydroflavonol reductase

NADPH-dependent dihydroflavonol reductase (DFR) plays an important role in both anthocyanin biosynthesis and proanthocyanidin synthesis in plants. A specific and quantitative RT-PCR assay for transcription from the DFR promoter detected high expression with limited variability in rice tissues. A 440 bp minimal promoter region was identified by transfection of β-glucuronidase (GUS) reporter constructs into Jeokjinju variety. Alignment of the region with orthologous promoters revealed three conserved segments containing both bHLH (-386 to -381) and Myb (-368 to -362) binding sites. Transfection of β-glucuronidase constructs with targeted point mutations in the minimal promoter defined two sites important for promoter function to the transcription factor binding consensus sequences. The expression study showed that the bHLH binding domain (-386 to -381) is essential for DFR expression, and that a Myb binding domain (-368 to -362) is also required for full expression of the DFR gene, while the two bHLH binding domains (-104 to -99 and -27 to -22) nearest to the transcriptional start site are not necessary for DFR expression.


Introduction
The accumulation of anthocyanins or proanthocyanidins in plant organs generally requires the presence of at least two families of regulatory proteins: Myc-type proteins containing a basic helix-loop-helix (bHLH) domain, and R2R3 Myb proteins. Genes belonging to these two families are present in numerous copies in plant genomes and encode similar proteins with varying patterns of expression that are associated with the regulation of diverse genes and pathways in different tissues and at different stages of growth [1,2]. Studies in maize report that a member of each of these families is necessary to activate anthocyanin biosynthesis in all tissues, and a direct physical interaction between Myc (bHLH) and Myb proteins is required for activation [3]. Subsequent research has revealed significant variations on this general theme in different plant species. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 In Arabidopsis and tomato, expression of the maize Lc (bHLH) gene induces anthocyanin production in both vegetative and reproductive organs, while in tobacco (Nicotiana tabacum), the same construct induces pigmentation only the floral organs [4,5]. Lc orthologs from Antirrhinum (Delila) and Perilla (Myc-RP) are unable to induce anthocyanin synthesis in Arabidopsis despite high sequence similarity between these proteins [6]. Similarly, the petunia bHLH homolog, Jaf13, was unable to activate anthocyanin synthesis when introduced into maize bHLH mutants [7]. These results suggest that interactions between bHLH proteins and their partner Myb proteins are very specific, and the expression of these genes is restricted to certain tissues and developmental stages, which may vary by species [8]. It also highlights the importance of the combinatorial interactions between these two families of transcription factors needed to impart transcriptional activation of the flavonoid pathway.
The combinatorial interactions of bHLH and Myb proteins for regulation of the flavonoid pathway may only be part of a larger picture. Work in maize, snapdragon, petunia, and Arabidopsis has revealed that bHLH and Myb proteins are only two of several factors potentially involved in the control of flavonoid biosynthetic genes. Additional molecular factors implicated in bHLH family-mediated flavonoid biosynthetic gene regulation include WD40 proteins, such as the TTG1 gene from Arabidopsis, the AN11 gene from petunia, and the PAC1 gene from maize [8][9][10][11]. While TTG1 has been shown to physically interact with a bHLH protein in Arabidopsis [10], AN11 from petunia was found to be localized in the cytosol [8]. In petunia and Arabidopsis, other regulatory factors, such as TT1, a zinc finger protein [12], and AN2, a homeodomain protein [13], have also been described where loss of function results in a complete lack of pigmentation. This further demonstrates that both a more general and a more specific understanding of bHLH-mediated transcriptional activation in diverse plant species is needed if we are to constructively harness the flavonoid and other metabolic pathways for applications in agriculture and medicine.
In this study, we focus on the regulation of proanthocyanidin accumulation in the pericarp of rice. Proanthocyanidins, or condensed tannins, give rice grains a characteristic red color, which is a feature of all wild Oryza species [14]. White grain color, on the other hand, is associated with domestication [15,16]. Proanthocyanidins are polymers of flavan-3,4-diols and flavan-3-ols, and are derived from the general flavonoid pathway [17]. These and related pigments are associated with a variety of biological functions in plants, including dormancy, pest and pathogen defense, and protection against UV radiation [18].
Pericarp color in rice is conditioned by genetic variation at the Rc and Rd loci [19,20]. The Rc gene encodes a Myc-type transcription factor containing a basic helix-loop-helix (bHLH) domain [15,21]. The Rd gene encodes dihydroflavonol 4-reductase (DFR) and is identical to the A locus in rice [15,22]. When the dominant, functional forms of Rc and Rd are present, the rice pericarp accumulates proanthocyanidins, giving the grains a red color [19]. Plants that have a nonfunctional rd, but a functional Rc have brown pericarp, while plants with non-functional alleles at Rc (rc or rc-s), regardless of the state of Rd, do not accumulate proanthocyanidins and consequently have white grains [15,21].
This suggests that Rd gene expression is dependent on the presence of a functional allele at Rc. However, Nakai et al. [22] showed constant expression of DFR transcripts in both Rc and rc varieties. Furukawa et al. [15] also reported no translational changes of the DFR protein in either red or white-pericarp varieties. These results suggested that the Rc gene may not be a direct, positive regulator of DFR transcription, but that the Rc-bHLH protein may regulate an unidentified gene involved in the biosynthetic pathway of proanthocyanidins, possibly at the step between leucoanthocyanidins and proanthocyanidins [15].
To investigate whether the Rc protein directly regulates DFR gene expression in rice, we constructed a set of deletion mutants in the DFR promoter region and used a promoter::GUS reporter system to monitor DFR expression. Using in silico analysis, we first identified eight cis-DNA binding elements in the DFR promoter corresponding to both bHLH and Myb protein binding domains. We then constructed five deletion mutants targeting these core cis-elements. Our results demonstrate that the expression of DFR in rice is regulated by intact binding domains in the DFR promoter for both bHLH and Myb transcription factors, and that Rc alone is not sufficient for DFR expression.

DFR expression in various tissues
DFR expression at 20 days after flowering (DAF) in Gopum (white pericarp) and Jeokjinju (red pericarp) varieties was observed in leaf, leaf blade, leaf sheath, stem and roots, and also in grains at 10, 15, and 20 DAF as shown in Fig 1. The DFR expression in leaf blade, leaf sheath, and stem was very low in both varieties, but the expression in leaf was 3 times higher in Jeokjinju than that in Gopum. The expression level in roots was approximately 10-fold higher than in leaves. In rice grains, DFR expression was highest at 20 DAF (ripening stage), with expression levels in Jeokjinju being about 8-fold higher than in Gopum (Fig 1).  (Table 1).

In planta expression assay of DFR::GUS
To identify which of the core cis-elements controlled the expression of the DFR gene, the 1.2 kb fragment upstream of the start codon was isolated and subcloned into the GUS expression vector pCAMBIA1301 ( Fig 3A). The DEL1 construct (1,203 bp) contained all six bHLH and two Myb binding motifs. DEL2 contained all DNA binding motifs except the first (most distal) bHLH (-736 to -731) motif. DEL3 had deleted the three distal bHLHs (-736 to -731, -637 to -632 and -552 to -547) and the Myb (-540 to -535), but retained the three bHLH (-386 to -381, -104 to -99 and -27 to -22) and the single MYB (-368 to -362) that were proximal to the start site of the DFR gene. DEL4 contained only the most proximal two bHLHs (-104 to -99 and -27 to -22) and DEL5 contained only bHLH (-27 to -22). These constructs were transformed into A. tumefaciens LBA4404 and then introduced into the red pericarp variety Jeokjinju (RcRd genotype) using Agrobacterium-mediated transformation. To determine whether the expression of DFR is dependent on the presence of a functional Rc allele, we transformed the DEL1 construct, which had a putative full-length promoter region containing all eight cis-DNA binding motifs, into a red-pericarp variety, Jeokjinju (Rc), and into a white-pericarp variety, Gopum (rc). Gus staining results showed similar expression patterns of the GUS gene in both rice varieties, regardless of which Rc allele was present (Fig 3B and 3C). To determine which of the motifs were essential for DFR expression, we introduced each individual construct, DEL1 to DEL5, into the Jeokjinju variety. T 1 seed staining results showed that the DFR promoter drove strong expression of GUS in DEL1, DEL2 and DEL3, but expression was very weak in DEL4 and DEL5 (Fig 3B and 3C).
We also carried out GUS staining of developing grains for each of the deletion lines, from the flowering stage to the mature seed stage. GUS activity was highest at 20 DAF. DEL1, DEL2, and DEL3 transgenic lines showed constant, high levels of GUS expression in pericarp tissue from early embryogenesis to mature seed, whereas, lines carrying DEL4 and DEL5 showed very weak expression of the GUS gene (Fig 4). These results suggest that DEL3, which contains three bHLH and a single MYB binding motif located over 300 bp away from the start site of the DFR gene has the minimal configuration of cis-elements necessary for DFR specific expression.

Site-directed mutagenesis of DEL3 construct
To test more specifically the effect of the bHLH and Myb binding site mutations on DFR::GUS expression, we used site-directed mutagenesis to introduce single substitutions into three bHLH and one Myb binding site in the DEL3 promoter ( Fig 5A) and then used the constructs for transient expression using the calli of the red pericarp variety, Jeokjinju (Fig 5B). DEL3-1 contains a single mutation, CAGGTG to CCGGTG, at the first bHLH (-386 to -381) binding site. DEL3-2 includes the mutation, ACCTACC to ATCTACC, at the Myb (-368 to -362) binding site, and DEL3-3 carries the mutation, CAAGTG to CCAGTG, in the last two bHLHs (-104 to -99 and -27 to -22). DEL3-4 has a single functional Myb (-368 to -362) and DEL3-5 contains a single functional bHLH (-386 to -381) binding motif (Fig 5A). Each construct containing a mutated site was transformed into A. tumefaciens LBA4404 and subsequently used to infect Jeokjinju calli to test for transient expression of each construct. In DEL3-1, infected calli expressed faintly or did not express the GUS gene at all (Fig 5B). DEL3-2 showed GUS expression, but not as strong as DEL3-3, which contained mutations in the last two bHLHs. These results demonstrated that the presence of both the bHLH (-386 to -381) and the Myb (-368 to -362) binding sites are necessary for complete expression of the DFR gene (Fig 5B). DEL3-3 also confirmed that the two proximal bHLH motifs (-104 to -99 and -27 to -22) are not necessary for DFR expression, while mutations in either of the distal bHLH motifs (-386 to -381) or (-368 to -362) completely prevent DFR expression. The faint GUS expression observed in DEL3-2 suggests that a single bHLH binding site is sufficient for low levels of DFR expression, but the higher level of GUS expression in DEL3-3 demonstrates that the combination of closely linked bHLH and Myb binding sites in the DFR promoter drives higher levels of gene expression (Fig 5B).

Discussion
Flavonoids are particularly interesting features of secondary metabolism in plants due to their roles in the production of pigments. Since plant organ pigmentation is an easily observable phenotype, the flavonoid pathway has been an ideal system for the study of metabolic pathways and gene regulation. Extensive genetic and molecular characterization of the biosynthetic and regulatory genes involved in the flavonoid pathway has provided some of the most detailed knowledge of gene interactions in plants [23].
The ability to regulate gene expression often depends on combinatorial interactions between transcription factors. In maize, a direct physical interaction between bHLH and Myb proteins is required for activation of anthocyanin biosynthetic genes [3]. Studies in mammals have shown that bHLH proteins are capable of binding to DNA cis-elements called "E-boxes" (CANNTG) in the promoters of their target genes [24][25][26], and these binding sites are conserved for plant bHLH proteins (sometimes referred to as "G-boxes") [27,28]. Variability in the capacity of bHLH proteins to bind DNA directly may be a function of their interactions with other molecular factors, including Myb proteins. Myb proteins contain three α-helices,   forming a helix-turn-helix structure, with the third helix making sequence-specific contacts with DNA [29,30]. The P and C1 genes from maize encode Myb proteins that bind directly to CC(T/A)ACC sites in the A1 promoter for activation of the Al biosynthetic gene (which encodes the maize DFR) [31]. The C1 Myb protein can also bind to a variety of sequences that resemble a consensus site A(C/A)C(T/A)A(C/A)C [30]. Other consensus sites for Myb protein binding have been experimentally described in the promoters of multiple flavonoid biosynthetic genes [23,32]. It is therefore likely that DNA binding by the bHLH, Myb, or both transcription factors together may dictate the tissue-specific activation of their target biosynthetic genes.
In this study, we dissected the role of cis-elements in the DFR promoter using a promoter:: GUS fusion system to understand which elements were critical to the regulation of DFR expression. The analysis assumes that the functionally important region of the target DFR promoter is located within a 2 kb region upstream of the transcriptional start site of the gene. While it is possible that additional regulatory elements may lie further upstream, work on flavonoid pathway regulation in maize and Arabidopsis has revealed that a small region of the promoter, near the transcription start site, contains Myb and bHLH binding sites and that this region is sufficient for tissue-specific activation of several flavonoid biosynthetic genes [23,[32][33][34][35].
Using constructs containing serial deletions within the rice DFR promoter fused to the coding sequence of a GUS reporter gene, our transformation experiments showed that a promoter containing three bHLH and a single MYB binding motif located >300 bp from the transcriptional start site was capable of driving DFR expression in intact seeds, pericarp tissue and embryonic callus of the red pericarp rice variety, Jeokjinju (Rc).
Using site-directed mutagenesis, we further showed that the bHLH (-386 to -381) binding domain is essential for DFR expression, and that the Myb (-368 to -362) binding domain is also required for full expression of the DFR gene), while the two bHLHs (-104 to -99 and -27 to -22) nearest to the transcriptional start site of the DFR gene are not necessary for GUS expression in embryonic callus of the red pericarp rice variety. These findings are consistent with a previous study by Furukawa et al. [15] who suggested that Rc alone did not regulate the function of the DFR gene. They hypothesized that the Rc-bHLH gene interacts with an unidentified gene involved in the biosynthetic pathway of proanthocyanidins [15]. In this study, we identified eight cis-DNA binding elements and showed that the expression of DFR in intact seeds, pericarp tissue and embryonic callus in rice, monitored by a promoter::GUS reporter system, required the presence of both bHLH and Myb binding domains, with binding sites located at -381 bp and -362 bp, respectively, upstream of the transcriptional start site of the DFR gene.

Plant materials and growth condition
The rice variety Jeokjinju (Rc, red pericarp) was used as the wild type for the generation of DFR::GUS lines. Regenerated plants were transplanted into soil medium (50% compost and 50% earth soil) in a greenhouse and acclimatized for 2 weeks. Young leaf samples were collected for genomic DNA analysis using hpt-specific primers. Plants confirmed as having the gene insert were harvested, and the T 1 seeds were used in the subsequent planting cycle in the field. Transgenic individuals, together with the wild type plants, were sown and grown up to the T 2 generation.

Analysis of cis-acting regulatory DNA element of DFR promoter
Cis-acting regulatory DNA elements in the DFR promoter (1.2 kb upstream of the transcriptional initiation site) were analyzed using the online software NEW PLACE (http://www.dna. affrc.go.jp/PLACE/) [36].
DFR promoter cloning and vector construction A 1.2 kb region of the DFR promoter was amplified by PCR using DFR deletion primers (DEL1 to DEL6) and cloned into the pGEMT-easy cloning vector (Promega USA) ( Table 1). All serial deletions were constructed using the pCAMBIA1301 (Cambia, Australia) binary vector, which contains a promoterless GUS gene and a CaMV 35S promoter driving the hpt gene encoding hygromycin B phosphotransferase. The 5 serial deletions were amplified by PCR using genomic DNA from the red pericarp variety Jeokjinju ( Table 1). The 5 sets of forward primers (DFR DEL1, DFR DEL2, DFR DEL3, DFR DEL4, and DFR DEL5) used for PCR had 5' overhang HindIII restriction enzyme sites and were respectively specific to 1238, 837, 581, 457, and 186 bp positions upstream of the start codon of DFR. The reverse primer, DFR DEL6 carrying the NcoI restriction enzyme site was designed from the 5' UTR region of the DFR gene. Each PCR product was digested with HindIII and NcoI restriction enzymes (New England Biolabs, UK) and digested products were ligated into the HindIII and NcoI predigested pCAMBIA1301 vector to create promoter::GUS constructs. The recombinant clones were confirmed by PCR, restriction digestion analysis, and sequencing.

Site-directed mutagenesis of the DFR promoter
For site-directed mutagenesis, primers used to mutagenize the regulatory elements were designed using the web program (http://labtools.stratagene.com/QC) ( Table 1) and purified by polyacrylamide gel electrophoresis. The mutant strand synthesis reaction mixture (50uL) contained 1X PfuUltra™ buffer, 200 uM each dNTPs, about 10 ng of template plasmid, 2.5 unit of PfuUltra™ (Stratagene, USA), and 125 ng of each primers. The PCR runs were 20 cycles with each cycle at 95ycle at 95sec, 55˚C for 30 sec, and 68˚C for 20 min to allow for the completion of the polymerizations. After mutant strand synthesis, PCR products were digested with 10 units of DpnI restriction enzyme at 377the completion of the polymerizations. After muL of DpnI digested DNA was transformed into DH5α, and five independent clones were sequenced.

Generation of transgenic plants and histochemical staining of GUS expression
The five promoter GUS constructs were electroporated into A. tumefaciens LBA4404 and introduced into pre-soaked rice seed using the method of Lee et al. [37] with minor modifications. Intact seeds, excised young pericarp tissue, and embryogenic calli originating from the red pericarp variety Jeokjinju and transgenic plants containing each deletion construct were used to estimate GUS expression. Histochemical GUS assays were performed as described by Jefferson et al. [38], and relative GUS expression was estimated using a fluorescence spectrophotometer (Bio-Rad, USA) with the excitation and emission wavelengths set at 360 and 465 nm, respectively. The GUS reaction solution consisted of sodium phosphate (Na 4 PO 4 , pH7.0), Triton-X-100, potassium ferricyanide (K 3 FeCN 6 ), potassium ferrocyanide (K 4 FeCN 6 ), and 5-bromo-4-chloro-3-indolyl-ferricyanide (Ksphagluc). The samples were vacuum-infiltrated using a GUS reaction solution and incubated at 37˚C for 24 hr. The reaction was stopped by adding 70% (v/v) ethanol, and the pigments and chlorophylls were removed by repeated ethanol treatment.

Analysis of GUS gene expression using RT-PCR
Total RNA was extracted from leaves collected from each transgenic rice line and the control Jeokjinju variety using the RNeasy Plant Mini Kit (QIAGEN, Maryland, USA). The relative purity and concentration of extracted RNA was estimated using a NanoDrop-1000 spectrophotometer (NanoDrop Technologies, Inc. Wilmington, USA), and RNA samples were then stored in a freezer at -80˚C. Total RNAs were purified using DNase I, and first-strand cDNA synthesis was performed by reverse transcription of mRNA using Oligo (dT) 20 primer and SuperScript™ III Reverse Transcriptase (Invitrogen, USA). The specific sequences of the primer pairs used in a semi-quantitative reverse transcription-PCR (RT-PCR) are GUS-F (5' -AGT GTGGGTCAATAATCAGGAAGT-3') and GUS-R (5AATCATGGAAGTAAGACTGCTTTT TCT-3'). For normalization of the quantitative RT-PCR reaction, the actin gene was used as an internal control and amplified using the primers ACT-F (5'-TGTATGCCAGTGGTCGTACC-3') and ACT-R (5'-CCAGCAAGG TCGAGACGAA-3') [39].

Expression of DFR gene in different tissues
Rice seeds of Gopum (white rice) and Jeokjinju (red rice) were dehulled and surface sterilized before plating in half strength MS media. After ten days of growth, leaf, stem and root tissues were harvested for RNA extraction. TRIzol 1 Reagent was used to isolate high quality total RNA. First-strand cDNA was synthesized with an oligo(dT) primer. The primers used to detect DFR expression by a semi-quantitative reverse transcription-PCR (RT-PCR) were 5'-ACTTCTCGTCGTGGAAGCTC-3' and 5'-GAAGCCCCTTCTCCCTGC-3'.