Functional Diversification of the Dihydroflavonol 4-Reductase from Camellia nitidissima Chi. in the Control of Polyphenol Biosynthesis

Plant secondary metabolism is complex in its diverse chemical composition and dynamic regulation of biosynthesis. How the functional diversification of enzymes contributes to the diversity is largely unknown. In the flavonoids pathway, dihydroflavonol 4-reductase (DFR) is a key enzyme mediating dihydroflavanol into anthocyanins biosynthesis. Here, the DFR homolog was identified from Camellia nitidissima Chi. (CnDFR) which is a unique species of the genus Camellia with golden yellow petals. Sequence analysis showed that CnDFR possessed not only conserved catalytic domains, but also some amino acids peculiar to Camellia species. Gene expression analysis revealed that CnDFR was expressed in all tissues and the expression of CnDFR was positively correlated with polyphenols but negatively with yellow coloration. The subcellular localization of CnDFR by the tobacco infiltration assay showed a likely dual localization in the nucleus and cell membrane. Furthermore, overexpression transgenic lines were generated in tobacco to understand the molecular function of CnDFR. The analyses of metabolites suggested that ectopic expression of CnDFR enhanced the biosynthesis of polyphenols, while no accumulation of anthocyanins was detected. These results indicate a functional diversification of the reductase activities in Camellia plants and provide molecular insights into the regulation of floral color.


Introduction
Plants produce multifarious secondary metabolites that play pivotal roles in various aspects related to development, growth and survival. Flavonoids (including polyphenolic compounds) are a major class of secondary metabolites that are produced in diverse plant lineages; they play an important role in the coloration and resistance of plants [1][2][3]. Many specific chemicals of flavonoids from plants (e.g., tea, grape and cacao) are found to be beneficial to human health [4][5][6]. Currently, the biosynthesis pathway of flavonoids is extensively characterized; at the initial step, phenylalanine is catalyzed to 4-coumaryol-CoA to generate chalcones which are backbones of flavonoids [7,8]. After multiple steps of enzymatic reaction, flavones, flavonols, polyphenols and anthocyanins are finally synthesized [9][10][11].

Cloning of CnDFR
Total RNAs were isolated using RNAprep Pure Extraction Kit (DP441, TIANGEN biochemical Technology, Beijing, China) and we determined the integrity of RNA based on the results of the 1.5% agarose gel electrophoresis analysis. According to the instructions of PrimeScript II 1st Strand cDNA Synthesis Kit (6210, TaKaRa, Tokyo, Japan), we synthesized the cDNA for gene cloning experiments. We designed a pair of specific primers (Table S1. 1) by Primer 3 (http://www.primer3plus.com/cgi-bin/ dev/primer3plus.cgi) according to the transcriptome data. PCR products were obtained, and we cloned them into a T-vector (CT501, TransGen Biotech Co., Ltd., Beijing, China) for sequencing. The full length of DFR was assembled and verified based on sequence analysis.

Sequence Alignment and Phylogenetic Analysis
The BioEdit software and NCBI Blast online (https://blast.ncbi.nlm.nih.gov/Blast.cgi) were used to align the sequences [45]. We used NCBI ORFfinder (http://www.ncbi.nlm.nih.gov/projects/gorf/) to find reading frames [46]. Protparam online (https://web.expasy.org/protparam/) was used to analyze protein molecular weight and isoelectric point and so on [47]. Meanwhile, amino acid sequence alignment was performed by DNAMAN and a phylogenetic tree was constructed with MEGA 5.0 software, using the neighbor-joining (NJ) method and 1000 bootstrap replicates [48].

Quantitative PCR Analysis of CnDFR
GAPDH as the reference gene (Table S1. 2) was used for quantitative PCR analysis (Table S1. 3). Using PrimeScript RT reagent Kit with gDNA Eraser (RR047, TaKaRa, Japan), the cDNA first strand was synthesized. Using SYBR Prime Ex Tap II (Tli RNaseH Plus) (RR420, TaKaRa, Japan), the quantitative PCR analysis was performed according to the user's manual. The reaction was performed on QuantStudio ® 7 Flex (Applied Biosystem, Foster City, CA, USA) and the reaction procedure was as follows: pre-denaturation at 95 • C for 30 s; 98 • C 5 s, 60 • C 34 s, 40 cycles; 95 • C 15 s, 60 • C 1 min, 95 • C 15 s. The relative expression level of CnDFR in different organs and different development periods was measured with GAPDH as the housekeeping gene by the 2 (−∆∆CT) method [49].

High-Performance Liquid Chromatography Analysis
The NF555 colorimeter (Nippon Denshoku Industries Co., Ltd., Tokyo, Japan) was used to detect the color indicator of petals. HPLC analysis was performed to measure the flavonoids, polyphenols and anthocyanins constituents. We grinded a fresh sample of 0.6 g weight in liquid nitrogen, supplied with 5 mL extraction solution (methanol/water/formic acid/trifluoroacetic acid = 70:27:2:1). Then, we extracted in the dark for 24 h, shaking in the middle a few times. After extraction, we filtered the sample with absorbent cotton to remove residue and pass it through the organic microporous filter membrane (0.22 cm) (ANPEL Laboratory Technologies (Shanghai) Inc., Shanghai, China). The filtrate was used in the machine analysis.

Agroinfiltration-Based Transient CnDFR-EGFP Gene Expression in Nicotiana Benthamiana Domin for Subcellular Localization of CnDFR
We designed a pair of primers (Table S1. 4) according to the EXclone Kit instructions (exv09, Hangzhou Biogle Co. Ltd., Hangzhou, China) for the vector construction. The overexpression vector was transformed to the Agrobacterium GV3101 strain by the thermal shock method [50]. To perform tobacco infiltration analysis, the transformed agrobacterium was suspended using induction medium (10 mM/L MES + 10 mM/L MgCl 2 + 100 uM/L AS) and injected into the Nicotiana benthamiana Domin leaf [51]. Then, the GFP signals were detected 2~5 days after injection by a LSM510 Meta device (Zeiss, Jena, Germany) [52].

Tobacco Transformation Analysis of CnDFR
To verify the functionality of CnDFR, we performed transformation of Nicotiana benthamiana Domin using the leaf plate method [53]. We used T5 Direct PCR Kit (Plant) (TSE011, TSING KE Biological Technology, Tianjin, China) for positive identification of rooting plants with PCR primers (Table S1. 5). PCR procedure was as follows: pre-denaturation at 98 • C for 3 min, denaturation at 98 • C for 10 s, annealing at 65 • C for 10 s, extension at 72 • C for 1 min and 30 s, 30 cycles, extended at 72 • C for 5 min and detected by 1% agarose gel electrophoresis. After the positive plants flowering, we collected the flowers, froze them with liquid nitrogen and stored them at −80 • C. The quantitative PCR was performed to measure the relative expression of CnDFR and determine total flavonoids, total polyphenols and total anthocyanins in flowers by a spectrophotometer [39]. Constituents of flavonoids, anthocyanin and polyphenols in flowers were also determined by HPLC and compared with the control group.

Molecular Characterization of CnDFR Reveals Lineage-Specific Amino Acid Sites
Based on transcriptome sequences of C. nitidissima Chi. [41], the full-length CDSsequence of CnDFR was obtained through gene-specific amplification (GenBank accession number MN276188). The CnDFR transcript encoded a protein of 342 amino acids, with a NADPH binding site of 21 amino acids (VTGAAGFIGSWLVMRLLERGY; Figure 1A). We performed sequencing alignment analysis of CnDFR with other homologs from various plant species. Then, it was found that all sequences included the conserved NADPH binding motif at the N terminal, and the substrate specificity-determining amino acid was identical in all other species but different from c hybrida L. ( Figure 1A). In the substrate specificity-determining region, there was a phenylalanine (F) in all DFRs of Camellia species, different from other plants ( Figure 1A); there were an additional four amino acids that were unique to Camellia species ( Figure 1A), which may result in different functions of Camellia DFRs.
To evaluate the phylogenetic relationships of DFRs, we aligned 14 DFR sequences from various plant species to construct a phylogenetic tree ( Figure 1B). The results showed that the three Camellia DFRs formed a clade together that was close to Actinidia chinensis ( Figure 1B). These results suggested that CnDFR was potentially functionally conserved in the secondary metabolism pathway, and DFRs from Camellia species might have unique functions compared to other plants.
unique to Camellia species ( Figure 1A), which may result in different functions of Camellia DFRs.
To evaluate the phylogenetic relationships of DFRs, we aligned 14 DFR sequences from various plant species to construct a phylogenetic tree ( Figure 1B). The results showed that the three Camellia DFRs formed a clade together that was close to Actinidia chinensis ( Figure 1B). These results suggested that CnDFR was potentially functionally conserved in the secondary metabolism pathway, and DFRs from Camellia species might have unique functions compared to other plants.   T  T  T  T  T  T  T   I  I  I  I  I  I I  I  I  I  I  I  I  i   I  I  I  I  I  I  I I  I  I  I  I  I  I  s   S  S  S  S  S  S  S  i   I  I  I  I  I  I  I  i   I  I  I  I  I  I  I I  I  I  I  I  I  I  i   I  I  I  I  I  I  I T  T  T  T  T  T  I  i   I  I  I  I  I  I  I    CnDFR sequence was found to be 75% to 99% similar to homological DFR genes.

Expression of CnDFR in C. nitidissima Chi. Is Positively Correlated with Polyphenols Contents
To study the expression profiles, the quantitative PCR (qPCR) analysis of CnDFR in different tissues of C. nitidissima Chi. was performed with GAPDH as the housekeeping gene (Figure 2A,B). It is found that CnDFR expressed in all tissues including the root, leaf, fruit, flower, sepal, petal, stamen and pistil ( Figure 2C). During flowering, the expression of CnDFR is maintained at a high level in floral bud differentiation stages ( Figure 2D) and reaches the highest expression when the flowers are half open, and then decreases rapidly after the blooming stage ( Figure 2D). DFR genes play a key role in the biosynthesis of plant secondary metabolites. In order to study the relationship between CnDFR and flower color in C. nitidissima Chi., the yellow color index b * and content of flavonoids and polyphenols were determined in the petals in five stages of C. nitidissima Chi., and we analyzed their relationships to the expression of CnDFR (Figure 3). By correlation analysis, it is found that the relative expression of CnDFR (DFR-RQ) is negatively correlated with TF (the content of total flavonoids), Qu7 G (quercetin-7-O-β-D-glucopyranoside) and b* (yellow color index of petals). However, it is positively correlated with the contents of total polyphenols and some components of polyphenols ( Figure 3B). The results indicate that a high level of CnDFR expression is correlated with enhanced polyphenols biosynthesis and potentially causes the yellow color to become lighter in C. nitidissima Chi. (Figure 3C).

Subcellular Localization of CnDFR Was in the Nucleus and Cell Membrane
To investigate the subcellular localization of CnDFR, we performed transient expression analysis using tobacco infiltration (Figure 4). A constitute expression construct harboring the fusion protein of CnDFR with green fluorescent protein (EGFP) was introduced into tobacco leaf. We found that free EGFP signals appeared in the nucleus, cell membrane and cytoplasm, and the signals were scattered throughout the whole cell ( Figure 4A). Meanwhile, the signals of CnDFR-EGFP were found to be localized in the cell membrane as well as the nucleus (Figure 4B), suggesting an extremely likely dual subcellular localization of CnDFR.

Overexpression of CnDFR Enhanced the Biosynthesis of Polyphenols But Not Anthocyanins
To study the biochemical functions of CnDFR, we generated overexpression lines using transgenic tobacco. Six transgenic lines with positive resistance were validated using construct-specific primers (Table S1, Figure S1). Further, we measured the relative expression levels of CnDFR in transgenic tobacco lines; the results showed that all tested lines displayed ectopic expression of CnDFR compared to the wild type plant ( Figure 5A), while no visible phenotypes were observed in the transgenic lines ( Figure 5B). Meanwhile, the contents of the secondary metabolites in flowers, including flavonoids, polyphenols and anthocyanins, were determined. It is found that there is no total anthocyanins content detected and the total polyphenols are significantly higher in transgenic lines than those in the wild type ( Figure 5C). Besides, the contents of flavonoids are increased slightly and the content of total flavonoids is much lower than total polyphenols ( Figure 5C).
In order to understand the compositional changes of secondary metabolites, the contents of different chemicals were measured by HPLC. Six flavonoid and six polyphenol standards were used to reveal the changes in contents in transgenic lines. We detected that the EGC and GC were significantly higher than the wild type in all six transgenic lines ( Figure 6A), while GCG, EGCG, CG and ECG were significantly increased in five transgenic lines ( Figure 6B). In most of the transgenic plants, the flavonoids contents (including Qu7 G, Qu3 G, Ka3 G, DHQ and Qu) were not significantly different from the wild type ( Figure 6C,D), and only the contents of Ka were significantly higher in all transgenic lines ( Figure 6C). This indicates that enhanced expression of CnDFR promotes the biosynthesis of polyphenols extraordinarily.

Discussion
The DFR enzymes are key players in the formation of plant pigments and antioxidative flavonoids [23]. The functions of DFRs in different plant lineages have been found to be conserved, which catalyze the reduction step of dihydroflavonols (including DHK, DKQ and DHM) [14,16]. In plants with red anthocyanins, the DFR gene is recognized as the first enzyme committed to anthocyanin biosynthesis, after the common phenylpropanoid pathway [28,54]. However, the diversity of the substrate specificity of DFRs is also revealed in some plant lineages [15,27,28]. In this study, we showed that CnDFR possessed conserved catalytic domains and some amino acids specific within Camellia species. The sequences of DFR from the genus Camellia were highly similar; the phylogenetic analysis also demonstrated that CnDFR was closely related to Camellia (Figure 1). Meanwhile, we have discovered that there was a conserved phenylalanine (F) of Camellia DFRs that differed from other plants in the substrate specificity-determining region (Figure 1). This indicates that Camellia DFRs might have different catalytic functions.
Many studies have shown that the expression patterns of DFRs have certain tissue specificity and are related to their functions. For instance, Nakatsuka et al. [26] found the DFR gene of the Asiatic hybrid was largely expressed in the colored tepals, anthers, filaments, pistils and red scales, while the expression was not detected in the uncolored tissues of the yellow variety. It is shown that, in C. nitidissima Chi., CnDFR is broadly expressed in various tissues including roots, leaves, fruits and flowers ( Figure 2). Detailed analysis during floral development has revealed that the expression pattern of CnDFR is positively correlated with polyphenol accumulation (Figure 3). This result also suggests that CnDFR is not a determinant directing the anthocyanin biosynthesis in C. nitidissima Chi. The yellow pigments in C. nitidissima Chi. have been identified majorly as quercetin derivatives [40] belonging to flavonoids. The expression profile of CnDFR during floral petal development was positively correlated with polyphenols but negatively with yellow coloration. Therefore, the roles of CnDFR in the regulation of the floral color of C. nitidissima Chi. need to be further characterized. Through the subcellular localization analysis, we have shown that CnDFR is likely to localize in the nucleus and cell membrane ( Figure 4). This result is different from the analysis of Vitisbellula, which found VbDFR was mainly located in the cytosol of onion epidermal cells [33].
In the overexpression analysis, it is found that there are no anthocyanins in the flowers of transgenic tobacco lines ( Figure 5). These results are not consistent with analyses from other plants. For example, overexpression of DFR of Agapanthus praecox into Petunia hybrida "W85" resulted in a change in floral color from white to fuchsia [32]. Further, down-regulation of DFRs from Nicotiana tabacum and Petunia hybrida reduced the anthocyanin contents and changed the floral color from pink to light pink and white [25,55]. All these studies indicate that DFR plays a key role in promoting the formation of anthocyanin and changing floral color, which is different from CnDFR. In overexpression lines of CnDFR, no anthocyanins were detected, which was also probably related to the transformation plants (Nicotiana benthamiana) with white flowers. The flavonoid pathway in N. benthamiana may be interrupted and does not have a complete synthesis pathway, leading to less synthesis of the final colored products and no color rendering, which need to be further researched. However, we found that a lot of polyphenols were accumulated in transgenic positive tobacco lines ( Figure 6). This indicates a functional diversification of the molecular function of CnDFR.
Studies in C. sinensis have shown that overexpression CsDFR enhanced the biosynthesis of polyphenols and stress resistance of plants [29,35], which supported a functional conservation in Camellia species. However, C. nitidissima Chi. is unique in its yellow floral color, and the petals of C. nitidissim Chi. accumulated a high level of flavonoids. Since most C. sinensis species bear white petals, it is not known if CnDFR has specified functions of flavonoid biosynthesis that is related to the yellow pigments. Future studies comparing different DFRs from several Camellia species might be required to investigate the molecular functions of DFRs.

Conclusions
We identified the DFR homolog from C. nitidissima Chi. (CnDFR), and its sequence analysis showed that CnDFR possessed some amino acids peculiar to Camellia species, among which a specific phenylalanine (F) was observed in Camellia DFRs in the substrate specificity-determining region. Phylogenetic analysis showed that DFRs of Camellia species formed a clade that was close to Actinidia chinensis. Gene expression analysis revealed that the expression of CnDFR was positively correlated with polyphenols but negatively with yellow coloration. Subcellular localization of CnDFR showed a likely dual localization in the nucleus and cell membrane. Furthermore, in the transgenic tobaccos, it was found that ectopic expression of CnDFR enhanced the biosynthesis of polyphenols, while no accumulation of anthocyanins was detected. These results suggest a functional diversification of DFR activities in Camellia plants and provide molecular insights into the regulation of floral color.

Conflicts of Interest:
The authors declare that they have no conflict of interest.