The Balance of Expression of Dihydroflavonol 4-reductase and Flavonol Synthase Regulates Flavonoid Biosynthesis and Red Foliage Coloration in Crabapples

Red leaf color is an attractive trait of Malus families, including crabapple (Malus spp.); however, little is known about the molecular mechanisms that regulate the coloration. Dihydroflavonols are intermediates in the production of both colored anthocyanins and colorless flavonols, and this current study focused on the gene expression balance involved in the relative accumulation of these compounds in crabapple leaves. Levels of anthocyanins and the transcript abundances of the anthocyanin biosynthetic gene, dihydroflavonol 4-reductase (McDFR) and the flavonol biosynthetic gene, flavonol synthase (McFLS), were assessed during the leaf development in two crabapple cultivars, ‘Royalty’ and ‘Flame’. The concentrations of anthocyanins and flavonols correlated with leaf color and we propose that the expression of McDFR and McFLS influences their accumulation. Further studies showed that overexpression of McDFR, or silencing of McFLS, increased anthocyanin production, resulting in red-leaf and red fruit peel phenotypes. Conversely, elevated flavonol production and green phenotypes in crabapple leaves and apple peel were observed when McFLS was overexpressed or McDFR was silenced. These results suggest that the relative activities of McDFR and McFLS are important determinants of the red color of crabapple leaves, via the regulation of the metabolic fate of substrates that these enzymes have in common.

Scientific RepoRts | 5:12228 | DOi: 10.1038/srep12228 The flavonoid biosynthetic pathway lies downstream of the phenylpropanoid pathway and leads to the formation of anthocyanins and flavonols (Fig. 1). Chalcone synthase (CHS; EC 2.3.1.74) uses beta-coumaroyl-CoA and 3 Malonyl-CoA as substrates to form naringenin chalcone 15 . This condensation reaction is a key step in the pathway leading to the formation of flavonoids 16 . Next, the F3H converts naringenin to dihydrokaempferol. Then dihydrokaempferol and dihydroquercetin are converted to kaempferol and quercetin by FLS, respectively. DFR and FLS also catalyze competing reactions to generate products leading to the spectrum of downstream anthocyanins and flavonols 17,18 .
MYB family transcriptional factors are known to be involved in regulating the expression expression of flavonoid biosynthesis genes 19 . For example, in Arabidopsis thaliana, AtMYBL2 interacts with TT8 (TRANSPARENT TESTA 8) to reduce anthocyanin biosynthesis 20 . In the context of flavonoid biosynthesis, AtMYB12 is thought to effect flavonol production 19,21 by regulation CHS, CHI, F3H and FLS1 gene expression levels 22 . In addition, the expression of three allelic apple (Malus × domestic) genes (MYB10, MYB1 and MYBA) was shown to strongly correlate with the accumulation of anthocyanins in fruit [23][24][25] .
The pigmentation of many plant organs results from the presence of some of these flavonoid classes and, in many instances, a primary determinant of color is the accumulation of anthocyanins. The synthesis of anthocyanins occurs via a key branch in the flavonoid biosynthesis pathway, involving the action of the enzyme DFR on the dihydroflavonols dihydroquercetin, dihydrokaempferol or dihydromyricetin. Indeed, DFR genes have been shown to play an important role in determining the total anthocyanin content in A. thaliana 26,27 . Additionally, the expression level of the DFR gene TfDFR1 has been shown to be positively correlated with red pigment accumulation in the petals of tulip (Tulipa fosteriana) 28 . It has also been reported that the expression level of DFR positively correlates with the abundance of anthocyanins in peanut (Arachis hypogaea) 29 .
Flavonols provide another important co-pigment in the colorful organs of terrestrial plants, such as the yellow petals of Lathyrus chrysanthus 30 , and they also influence pollen tube growth 31 . Flavonols are derived from 2, 3-dihydroquercetin and their formation is catalyzed by FLS, which belongs to the 2-oxoglutarate dependent dioxygenase family 4,32 . Following the identification of an FLS gene from Petunia hybrida 33 , homologous FLS genes have been identified from A. thaliana, Solanum tuberosum, Matthiola incana, Malus domestica and Eustoma russellianum. In addition, it has been reported that increases in transcript levels of an FLS gene from satsuma mandarin (Citrus unshiu Marc.), CitFLS, in the fruit peel correlate with flavonol accumulation during fruit development 34 . Recently, it has been found that major floral color changes are a consequence of FLS expression in petunia (Petunia hybrida Vilm.), Lisianthus (Eustoma grandiflorum) and camellia (C. nitidissima) 33,[35][36][37] . In contrast, an indirect effect of a camellia FLS gene (CnFLS1) on anthocyanin accumulation during floral coloration was suggested following an experiment where its overexpression in transgenic tobacco (N. benthamiana) plants resulted in an increase in flavonol content, but a reduction in anthocyanin levels in petals 37 .
Leaf color is a key determinant of the commercial value of many ornamental plant species; however, much remains to be learnt about the mechanisms of color formation in leaves at the molecular level. The study of pigmentation mechanisms in leaves is therefore significant for both breeding and genetic engineering of ornamental plants. An example of an important ornamental woody plant is Malus crabapple, which belongs to the Rosaceae, Malus Mill family. The numerous plant landscape species in this family provide an excellent source of research material for studying the mechanism of color formation and accumulation, due to their colorful leaves, flowers and fruits 38 . To date, little is known about the mechanism of anthocyanin and flavonol biosynthesis in ornamental crabapples.
In this current study, we investigated the function of the crabapple DFR (McDFR) and FLS (McFLS) genes in regulating leaf color in different cultivars. We overexpressed and silenced each gene to determine their interaction in controlling flavonol and anthocyanin biosynthesis, and evaluated the gene expression ratio of McDFR and McFLS that is required for leaf color production. We also discuss the metabolic flux between McDFR and McFLS during flavonoid biosynthesis in leaves and fruit. We propose that the finding from this study will assist future attempts to enhance anthocyanin or flavonol accumulation in the leaves of target ornamental species by altering the balance between the McDFR and McFLS enzyme activities.

Results
The anthocyanin and flavonol content of the leaves of two crabapple cultivars. Two extreme leaf color cultivars, 'Royalty' and 'Flame' , have ever-red and ever-green leaves, respectively. We evaluated the abundance of anthocyanins and flavonols in the leaves of these cultivars at 5 development stages of the crabapple leaf growing season by high-performance liquid chromatography (HPLC) (Fig. 2). The chromatography results showed that cyanidin 3-O-glucoside was the predominant anthocyanin, and we found that that the major flavonols were quercetin derived compounds, such as quercetin 3-O-diglucoside, quercetin 3-O-glucoside and quercetin 3-O-glycosidase isomer (Fig. 2B). As shown in Fig. 2C, anthocyanin levels in the ever-red leaves of 'Royalty' were significantly higher than those in the ever-green leaves of 'Flame' . A gradual decrease in anthocyanin content was observed in 'Royalty' leaves during their development, while anthocyanins were almost undetectable in 'Flame' leaves. In contrast, the abundance of flavonols increased during the development of 'Flame' leaves, except at stage 5. We also transiently over-expressed the McDFR gene in the stem tips of the ever-green cultivar 'Strawberry Jelly' , which promoted anthocyanin accumulation at 20 dpi, and a deep green coloration in most of the new buds (Fig. 4A). The anthocyanin content showed a slight increase in the McDFR-overexpressing plants (Fig. 4B) and we confirmed that these plants indeed had higher McDFR transcript levels in the new buds (Fig. 4C). We also detected an increase in the transcript levels of anthocyanin biosynthetic genes and a decrease in McFLS expression, compared with the non-transformed plants (Fig. 4C).
Collectively, these results indicated that McDFR expression is associated with red color formation in crabapple leaves, and that changes in McDFR expression can affect the expression of downstream genes (e.g. ANS and UFGT) involved in the anthocyanin biosynthetic pathway. McFLS is involved in flavonol biosynthesis in crabapple leaves. To confirm the prediction, based on sequence homology, that McFLS is a key flavonol biosynthetic gene, we suppressed its expression in the leaves of 'Strawberry Jelly' using the VIGS system and the TRV vector. Approximately 14 days after Agrobacterium infiltration, red coloration was seen in the margin and other areas of the infected leaves (Fig. 5A). HPLC analysis confirmed that the levels of anthocyanins were significantly higher in the silenced leaves than in control leaves infiltrated with TRV alone (Fig. 5B). Finally, as seen in

Discussion
Due to the central role that the DFR enzyme plays in the anthocyanin biosynthetic pathway, DFR genes have been studied in several monocotyledonous and dicotyledonous species, such as Forsythia intermedia 40 , Torenia fournieri 31 , Triticum aestivum 41 , Vitis vinifera 42 and Ascocenda spp. 43 , and in some cases at the transcriptional level 44 . The key structural gene, FLS, completes the last step of flavonol synthesis. To our knowledge, although FLS and DFR have been identified and characterized in many land plant species, these genes have yet to be studied in ornamental crabapples. There is growing evidence that anthocyanins and flavonols to contribute to the ornamental and economic value of crabapples, which have highly colorful leaves, fruits and flowers 38 (Table 1). This result is congruent with a recent study showing a negative correlation of CnFLS1 expression and anthocyanin synthesis during floral coloration in the petals of transgenic tobacco expressing this gene 37 . We infer from these observations that there is a competitive relationship between McDFR and McFLS in flavonoid biosynthesis.

The function of McDFR and McFLS in flavonoid biosynthesis. Genetic transformation has been
used to test the functions of several genes in the flavonoid biosynthetic pathway in model experimental plants, such as tobacco and A. thaliana, as well as in some crop species. For example, overexpression of petunia CHI in tomato fruit was reported to lead to an ~65% increase in flavonol levels 45 , while silencing of a CHS gene in apple fruit resulted in changes in growth and developmental phenotypes 46 . In Gentiana triflora and apples (Malus spp.), the silencing of an ANS gene caused a reduction in anthocyanin content and, consequently, a much weaker color 11 . However, few studies to date have focused on genetic transformation using DFR or FLS genes and none has targeted woody ornamental species, such as crabapple.
Virus-induced gene silencing (VIGS) is a technology that allows the analysis of genes function by suppressing the expression of target genes. In a previous study, we assessed the effect of gene silencing in several plant species, including Nicotiana benthamiana, rose (Rose hybrida) and strawberry (Fragaria × ananassa) [47][48][49] (Table 1). We propose that there is a competitive relationship between the expression McDFR and McFLS that results in the production of different classes of flavonoid compounds (i.e. anthocyanins or flavonols) (Figs 4-7). Since dihydroflavonols are substrates for DFR and FLS, they lie at an important branch point in flavonoid biosynthesis, where precursor substrates are channeled toward either anthocyanin or flavonol production. In this regard, the regulation of DFR expression and the competing DFR and FLS activities may be particularly important. We conclude that the expression of McDFR and McFLS may represent a key mechanism for regulating color in crabapple leaves. Moreover, when AtMYB12 was over-expressed in a tissue-specific manner in tomato, the flavonol biosynthesis pathway was activated 51 . However, expression of the DFR gene was not induced in AtMYB12 and AtMYB111 over-expressing transgenic plants, as well as in AtMYB111 transgenic lines 1,44,51 . In apple and crabapple, MYB10, which regulates anthocyanin accumulation and coloration of various organ (e.g. fruit, petals and leaves), can activate the expression of DFR and bind to several the promoters of several anthocyanin biosynthetic genes 52,53 . In grape berries, expression of the regulatory gene VvMYBF1 is light inducible, and is involved in the control of VvFLS1 transcription and flavonol synthesis in fruit 54 . Thus, we speculated that MYB TFs may similarly regulate flavonoid biosynthesis in crabapple, and that their expression levels may vary at different development stages or in response to different environment conditions. Moreover, changes in the transcript levels of MYB TFs control the biosynthesis of flavonoids, by regulating the expression of various members of the flavonoid biosynthetic pathway. We propose that MYB TFs promote anthocyanin biosynthesis by increasing the transcript levels of DFR during fruit or leaf development, leading to red coloration. However, in response to environmental stresses, MYB TFs activate the transcription of FLS resulting in increased production of flavonols. Leaves of 'Royalty' and 'Flame' were collected at five different developmental stages (Fig. 1A,B) for gene expression analyses and anthocyanin and flavonol quantification. Wild-type 'Royalty' and 'Strawberry' seedlings were grown in a greenhouse, as above. Apple (Malus × domestica 'Fuji') fruits were used for analysis of McDFR and McFLS expression. All samples were frozen in liquid nitrogen upon collection, and stored at − 80 °C until further use.  Table S1. Virus-induced gene silencing vectors carrying the target gene fragments, as well as pTRV1 and pTRV2 [47][48][49] , were transformed into Agrobacterium tumefaciens strain GV3101 competent cells using a freeze-thaw method 55 and selected on kanamycin-rifampicin-containing (50 mg/L) LB (Luria Bertani media) plates. Positive clones were verified by restriction enzyme digestion and by sequencing the vector-insert junctions. The harvested bacterial cells were then resuspended to an OD 600 of 0.5 in infiltration medium (10 mM 2-morpholinoethanesulfonic acid [MES], 200 mM acetosyringone, and 10 mM MgCl 2 ) and incubated at room temperature for 3 h. Before infiltration, bacteria carrying pTRV1 and pTRV2 were mixed in a 1: 1 volume ratio.

Construction of VIGS vectors and
For vacuum infiltration, whole plants were submerged in the Agrobacterium suspension and subjected to a vacuum (− 25 kPa). When the rate of air bubbles being released from the plants started to decrease, the vacuum was released quickly to allow bacteria to enter the plant tissues. The vacuum treatment time varied from 30 s up to 3 min, depending on the vacuum source used. After vacuum infiltration, plants were rinsed with sterile water, and cultured on Murashige and Skoog medium. Fifteen plants from each cultivar were treated, and 'Royalty' was used for silencing of McDFR expression, while 'Strawberry Jelly' was used for silencing of McFLS expression. All experiments were repeated three times.

Overexpression of McDFR and McFLS in crabapple leaves and fruits. The full length McDFR
and McFLS open reading frames (ORFs) were cloned from the cDNA library described above and inserted into the pBI121 vector 53 using the XbaI and SacI sites. Primers used for these constructs are shown in Table S1. Transient expression in Malus crabapple leaves was performed using the 'Strawberry' Jelly cultivar and Agrobacterium-mediated transformation, as described above. Agrobacterium cells containing the different constructs were harvested and resuspended in infiltration buffer (10 mM MES, 0.2 mM acetosyringone, and 10 mM MgCl 2 ) to a final concentration of OD 600 = 0.5. Vacuum infiltration was performed as described above, and infiltration with an empty vector was used as a negative control. Seven days after infiltration, the infected leaves and fruit were collected to observe phenotypic features and to evaluate differences in expression.

RNA extraction.
To analyze the effects of VIGS and overexpression on target genes expression, tissue samples from areas showing the silencing and enhancing phenotypes were collected. For controls, corresponding samples were collected from tissues infected by Agrobacterium carrying vectors with no host gene fragment insert, or from non-infected plants. Samples from three independent biological replicates were analyzed. Total RNA was extracted from crabapple leaves using the RNA plant plus Reagent (TIANGEN BIOTECH) according to the manufacturer's instructions. DNase (TIANGEN BIOTECH) treatment was performed to remove any genomic DNA according to the manufacturer's instructions. First-strand cDNA was synthesized from total RNA using the Reverse Transcriptase M-MLV (RNase H − ) kit (TaKaRa).
Quantitative RT-PCR analysis. qRT-PCR was performed using the SYBR ® Premix Ex TaqTM II (Perfect Real Time) (TaKaRa, Ohtsu, Japan) and the CFX96TM Real Time System (Bio-Rad, USA). The PCR amplification conditions were as previously described 56 , and transcript levels were determined by relative quantification using the Malus 18S ribosomal RNA gene (DQ341382) as the internal control and the 2^ (−∆∆CT) analysis method was applied. Specific primers (Table S1) for semi-quantitative RT-PCR and qRT-PCR analysis were designed using the primer 5 software 57 . HPLC analysis. Crabapple leaf samples (approximately 0.8-1.0 g fresh weight) were subjected to extraction with 10 mL extraction solution (methanol: water: formic acid: trifluoroacetic acid= 70: 27: 2: 1) 58 at 4 °C in the dark for 72 h, shaking every 6 h. The supernatant was isolated by filtration through filter paper and a further filtration through a 0.22 μ m Millipore TM filter (Billerica, MA, USA). For the HPLC analysis, trifluoroacetic acid: formic acid: water (0.1: 2: 97.9) was used as mobile phase A and trifluoroacetic acid: formic acid: acetonitrile: water (0.1: 2: 48: 49.9) was used as mobile phase B. The gradients used were as follows: 0 min, 30% B; 10 min, 40% B; 50 min, 55% B; 70 min, 60% B; 30 min, 80% B. Detection was performed at 520 nm for anthocyanin and 350 nm for flavonol 58 , respectively. All samples were analyzed in three biological triplicates (extracted from three different batches of leaves).