Arg-type dihydroflavonol 4-reductase genes from the fern Dryopteris erythrosora play important roles in the biosynthesis of anthocyanins

Dihydroflavonol 4-reductase (DFR), a key enzyme involved in the biosynthesis of anthocyanins, has been cloned from various species. However, little research has been conducted on this enzyme in ferns, which occupy a unique evolutionary position. In this study, we isolated two novel DFR genes from the fern Dryopteris erythrosora. In vitro enzymatic analysis revealed that DeDFR1 and DeDFR2 enzymes can catalyze dihydrokaempferol and dihydroquercetin but cannot catalyze dihydromyricetin. Amino acid sequence analysis showed that DeDFR1 and DeDFR2 have an arginine at the same substrate-specificity-determining site as that in the ferns Salvinia cucullata and Azolla filiculoides. Thus, we speculate that the Arg-type DFR is a new DFR functional type. To further verify the substrate preferences of the Arg-type DFR, an amino acid substitution assay was conducted. When N133 was mutated to R133, Arabidopsis DFR protein completely lost its catalytic activity for dihydromyricetin, as observed for DeDFR1 and DeDFR2. Additionally, heterologous expression of DeDFR2 in the Arabidopsis tt3-1 mutant resulted in increasing anthocyanin accumulation. In summary, DeDFR1 and DeDFR2 are considered to be a new type of DFR with unique structures and functions. The discovery of the Arg-type DFR provides new insights into the anthocyanin biosynthesis pathway in ferns.


Introduction
Anthocyanins represent an important subgroup of flavonoids, which are important secondary metabolites produced by plants to adapt to the terrestrial environment. Anthocyanins, serving as pigments, signaling molecules, photoprotectants, and protectants against biotic and abiotic stress, have a broad spectrum of functions in plant development [1][2][3]. Anthocyanins in food plants are potentially valuable components of the human diet. In addition, they have

Plant materials
Dryopteris erythrosora was grown in a field or in pots in a growth chamber at the College of Life Science, Shanghai Normal University, China. For gene expression analysis, we sampled rolled immature fronds, young fronds and mature fronds from plants grown in the field. All fronds were frozen in liquid nitrogen and stored at -80˚C. The Arabidopsis thaliana ecotype Columbia (Col) was a gift from Prof. Zhongnan Yang (Shanghai Normal University). The Arabidopsis tt3-1 mutant was purchased from SALK (SALK ID: CS2114).

Isolation and cloning of DeDFR1 and DeDFR2
Total RNA was extracted from 0.5 g of rolled immature fronds of D. erythrosora using the RNApure Plant Kit (Cwbiotech, China). First-strand cDNA was subsequently synthesized using the Superscript III First-Strand Synthesis Kit (Invitrogen, USA). All steps were performed by following the manufacturer's instructions supplied with the kits. The transcriptome was sequenced on the Illumina HiSeq2000 platform (Illumina, USA) at BGI (Shenzhen, China). Based on the assembled transcriptomic sequences of rolled immature fronds of D. erythrosora (TSA number: GHEW00000000), we designed pairs of specific primers (S1 Table) for DeDFR1 and DeDFR2. TaKaRa LA Taq (Takara, Japan) was used for PCR. The thermal cycling conditions were as follows: 94˚C for 1 min, followed by 30 cycles of 94˚C for 30 s, 55˚C for 30 s, and 72˚C for 60 s, and a final extension step at 72˚C for 5 min. The PCR products were ligated to the pEASY-T5 Zero vector (Transgen Biology, China), and single colonies were selected for sequencing.

Expression analysis of DeDFR1 and DeDFR2
The expression levels of DeDFR1 and DeDFR2 in D. erythrosora were analyzed using qRT-PCR. The cDNA used for detection of expression was synthesized by using PrimeScript™ RT Master Mix (Takara). The PCR primers were designed using Primer Premier 5.0 and are listed in S1 Table. All reactions were carried out with an ABI 7500 real-time PCR system (Applied Biosystems, USA) with a 20-μL reaction system consisting of 10 μL of SYBR Green™ Premix Ex Taq™ (Takara), 0.4 μL of ROX Reference Dye II, 0.2 μM each specific primer, and 10 ng of template cDNA. The thermal cycling conditions were as follows: an initial step of 95˚C for 30 s, followed by 40 cycles of 95˚C for 5 s and 60˚C for 34 s. A melting curve was generated for each sample at the end of each run to verify the purity of the amplified products, and at least three replicates of each sample were analyzed. The transcript levels were normalized using DeGAPDH (GenBank: MK920229) as the internal control and calculated using the 2 (−ΔΔCt) analysis method.

Production of recombinant DFR proteins in Escherichia coli
DeDFR1 and DeDFR2 were cloned into the expression vector pET28a (+) (Novagen, USA) by using the In-Fusion PCR Cloning Kit (Takara). The PCR primers were designed using In-Fusion Cloning tools (https://www.takarabio.com). Bam HI and Sac I were used as the two restriction sites. The list of primers is provided in S1 Table. In-Fusion cloning was performed according to the manufacturer's instructions at 50˚C for 15 min with a 5-μL reaction system consisting of 1 μL of 5× In-Fusion HD Enzyme Premix, 50 ng of linearized vector and 25 ng of purified PCR fragment. Ligated products were introduced into competent Trans5α cells (Transgen Biology) for sequencing. After confirmation of the sequences, the subcloned vectors were used to transform E. coli strain BL21 (DE3) (Transgen Biology). The empty vector and pET28a-AtDFR were also introduced into E. coli as controls.
The expression of each recombinant DFR protein was induced by 0.2 mM isopropyl-thioβ-D-galactoside (IPTG) in LB culture for 20 h at 20˚C. The E. coli cells were harvested by centrifugation and resuspended in extraction buffer (25 mM Tris-HCl, pH 8.0). The cells were lysed by sonication, and the debris was removed by centrifugation at 8,000 rpm for 5 min at 4˚C. The protein concentration was estimated by using the Easy Protein Quantitative Kit (Transgen Biology), and the protein quality was examined using 10% SDS-PAGE.

Enzyme assay
Measurements of the activity of the recombinant DeDFR1 and DeDFR2 were performed according to the methods of Petit et al. and Katsu et al. with minor modifications [25,35]. A 500-μL reaction mixture consisted of 1.6 mM NADPH, 50 mM Tris-HCl (pH 7.0), 0.8 mM substrate, and 5 mg of total protein. Three dihydroflavonols were selected as experimental substrates: dihydrokaempferol (DHK), dihydroquercetin (DHQ) and dihydromyricetin (DHM) (Sigma-Aldrich). Each reaction was terminated after 30 min at 30˚C by extraction with ethyl acetate. Because the obtained leucoanthocyanidins were unstable in solution, the residues were dissolved after evaporation in 200 μL of n-BuOH-HCl (95:5) and incubated at 95˚C for 30 min to form anthocyanidins. The HPLC analysis of the DFR products was similar to the anthocyanidin detection described above. The process was repeated at least three times for each sample.

Amino acid point mutant construction
The 133rd amino acid residue of AtDFR protein (Asn) was modified to Asp or Arg by overlap PCR. The principle of overlap PCR is to design two pairs of primers to amplify sequences with overlapping modified sites and then extend the overlap to obtain the complete sequence. The list of primers is provided in S1 Table. The modified AtDFR N133D and AtDFR N133R genes were ligated into the vector pET28a (+), transformed into E. coli strain BL21 (DE3), and subjected to the above enzymatic assay.

Plant transformation
To further characterize the function of DeDFRs, we cloned DeDFR1 and DeDFR2 into pBI121 vector harboring the CaMV 35S constitutive promoter by using the In-Fusion PCR Cloning Kit. Xba I and Sma I were used as the two restriction sites. The list of primers is provided in S1 Table. The recombinant vector was transformed into Agrobacterium tumefaciens strain GV3101 using a freeze-thaw method. Seed germination and seeding growth followed the standard Arabidopsis growth protocol. Genetic transformation followed the floral dipping transformation protocol [36]. The growth of infected plants and seed harvesting followed the same protocol as previously reported [9,37]. T1 seeds were selected on PNS medium with 50 mg L −1 kanamycin to obtain T2 plants. Planting of T2 seeds provided more homozygous T3 plants by DNA identification and antibiotic screening. Wild type (Col) and tt3-1 mutant plants were grown as controls. The Arabidopsis tubulin gene was used as an internal control gene to detect the DeDFRs expression level.

Isolation of DeDFRs and phylogenetic analysis
Based on the assembled transcriptomic sequences of D. erythrosora, we screened and obtained two DeDFR genes (CL10493.Contig3 and CL16087.Contig1) and designated them DeDFR1 (GenBank: MK920230) and DeDFR2 (GenBank: MK920231). The open reading frame (ORF) of each DeDFR was analyzed, and a pair of specific primers that contained the start and stop codons was designed. The full-length cDNA of each gene was obtained by PCR. The ORF of DeDFR1 was a 987-bp segment encoding 328 amino acids (Fig 1), and the theoretical molecular weight and isoelectric point were 36.3 kD and 5.41, respectively (calculated using the ExPASy website https://web.expasy.org/compute_pi/, S2 Table). DeDFR2 was a 1026-bp segment encoding 341 amino acids, and the theoretical molecular weight and isoelectric point were 37.8 kD and 5.54. The identity of the deduced amino acid sequences between two DeDFRs and GbDFR were 45.94% and 47.20%, respectively. These percentages were higher than the identity between the two DeDFRs and AtDFR of 43.48% and 46.34%. In addition, the alignment showed that DeDFR1 and DeDFR2 belonged to the SDR superfamily, including a conserved NAD(P)-binding domain and a substrate-binding domain.
Phylogenetic analysis of the deduced amino acid sequences of DeDFR1 and DeDFR2 was performed, and the sequences were compared to those of homologous DFRs from different species (Fig 2). To further investigate the characteristics of DFR proteins in ferns, we found five Salvinia cucullata DFRs (ScDFR1-5) and six Azolla filiculoides DFRs (AfDFR1-6) based on genomic data. The DFR sequences of the lycophyte S. moellendorffii were used as the root of the tree. This phylogenetic tree indicated that DFRs of ferns and seed plants were clearly classified into different branches. Spermatophyte DFRs further diverged into different groups of gymnosperms, monocots, and eudicots. The DFR sequences of ferns were located at the base of the phylogenetic tree. The figure shows that AfDFR1-3, ScDFR1-2 and DeDFR1-2 proteins clustered within a branch, which is more similar to seed plants. It is worth noting that all the DFRs we found from fern S. cucullata, A. filiculoides and D. erythrosora belonged to neither Asn-nor Asp-type DFR. The results indicated that the sequences of DFRs of ferns and seed plants were different; thus, their protein functions may also be different.

Expression of DeDFRs and anthocyanidin accumulation during frond development
In D. erythrosora, the fronds showed a significant color change during the growth process ( Fig  3A). The anthocyanidin component in different stages of frond development was detect by HPLC. At 520 nm, the chromatograms showed that the standard peaks of cyanidin, delphinidin and pelargonidin appeared at approximately 11.09 min, 9.16 min, and 13.22 min,  respectively (Fig 3B). The detection results showed that cyanidin and delphinidin, but almost no pelargonidin, were present in these fern fronds (Fig 3C). The contents of both cyanidin and delphinidin were enriched in young fronds (Fig 3D). Cyanidin could be detected in all stages, and delphinidin could hardly be detected in rolled immature fronds. Transcriptomic analysis revealed that DeDFR2 was present at high levels in raw fragments of the rolled immature fronds, indicating that DeDFR2 might be closely associated with frond development (Table 1). qRT-PCR was used to examine the expression patterns of DeDFR1 and DeDFR2 (Fig 3D). The transcripts of DeDFR1 and DeDFR2 were detected in all fronds tested. The expression of DeDFR1 and DeDFR2 showed an increasing trend at first, followed by a decreasing trend, with frond development. The highest expression was observed in young fronds, followed by rolled immature fronds, and relatively low expression was observed in mature fronds, which is consistent with the change trend observed for cyanidin and delphinidin. Based on the accumulation of anthocyanins and the expression of DeDFRs, we speculated that DeDFR1 and DeDFR2 might be involved in anthocyanin synthesis.

Heterologous expression and catalytic activity of DeDFRs
To identify the catalytic function of DeDFR enzymes on dihydroflavonols, prokaryotic expression was implemented. Using the In-Fusion PCR cloning method, the ORFs of DeDFR1 and DeDFR2 were inserted into the expression vector pET28a (+) to obtain recombinant proteins. The SDS-PAGE results showed that these recombinant proteins were similar in total protein content (S1 Fig). The cumbersome purification process tended to inactivate the recombinant protein, and the impure protein obtained after purification affected the accuracy of the protein content determination. Therefore, our activity analysis was based directly on the total protein content. Crude extracts of Escherichia coli expressing DFRs were subjected to enzyme assays using DHK, DHQ, and DHM as substrates in the presence of NADPH. The reaction product with an empty vector was colorless, and the HPLC chromatogram showed no significant product peak at 520 nm. Additionally, the results for the positive control AtDFR, an Asn-type DFR, showed that all three dihydroflavonols could be converted to pink anthocyanins, which could be detected by HPLC at 520 nm (S1 Fig). DeDFR1 and DeDFR2 had the ability to catalyze the DHK and DHQ reaction to produce the desired products. However, they could not catalyze DHM to produce the products of AtDFR (S2 Fig). The product peak area in HPLC is shown in Fig 4. The results showed that the catalytic products of DeDFR1 and DeDFR2 were lower than those of AtDFR regardless of the use of DHK or DHQ as a substrate. The amount of product of DeDFR1 and DeDFR2 was different. When using DHK as a substrate, the target product of  DeDFR1 was comparable to that of DeDFR2. However, when using DHQ as a substrate, the target product of DeDFR2 was lower than DeDFR1 by approximately 24%. The above data provide evidence that the DeDFR1 and DeDFR2 in D. erythrosora had DFR activity. In addition, the substrate specificity of these enzymes differed from that of AtDFR, which could catalyze DHM.

Site-directed mutagenesis of the substrate-specificity-determining site
We found that both DeDFR1 and DeDFR2 had an Arg at the substrate-specificity-determining site, and they did not have the same ability to catalyze DHM like the Asn-type DFR. To verify the substrate preferences of the Arg-type DFR, the overlap PCR method was used to modify AtDFR to AtDFR N133D and AtDFR N133R (S3 Fig). Because modification of the 133rd residue hardly affected the solubility of AtDFR, we analyzed the activity of the modified recombinant proteins by quantifying the total protein content. Crude extracts of E. coli expressing modified AtDFRs were subjected to enzyme assays using DHK, DHQ, and DHM as substrates in the presence of NADPH (S2 Fig). The results showed that the enzyme activity was changed after the site was modified (Fig 5). The catalytic capabilities of AtDFR, AtDFR N133D and AtDFR N133R were quite different. When DHK was used as a substrate, the AtDFR N133D activities decreased significantly. However, when DHQ was used as a substrate, the target product of AtDFR N133R yielded approximately half of that obtained with AtDFR. Notably, when DHM was used as a substrate, products were not detected with AtDFR N133R , as observed for DeDFR1 and DeDFR2. The site-directed mutagenesis results showed that the 133rd amino acid residue was indeed an important site affecting substrate specificity. They also showed that Arg-type DFRs could hardly catalyze DHM.

Overexpression of DeDFRs in Arabidopsis tt3-1 Mutant
The Arabidopsis tt3-1 mutant is characterized by the knockout of DFR, resulting in a lack of anthocyanins, so the seed coat is yellow due to absence of pigment. To investigate the function of DeDFRs in plants, they were ectopically expressed in Arabidopsis tt3-1 mutant. Using the flower dip transformation method, transgenic DeDFR1 and DeDFR2 plants were obtained ( Fig  6A). The results showed that the DeDFR1 gene could not be expressed in the transgenic plants, and the seeds of the transgenic plants were still yellow. By analyzing the codon preference, we found that unlike AtDFR or DeDFR2, multiple arginines were encoded by CGC in DeDFR1, which might have led to the lack of DeDFR1 expression in Arabidopsis (S4 Fig). Phenotypic observation showed that transgenic plants expressing DeDFR2 exhibited a restored brown color of the seed coat similar to wild type seeds. The anthocyanidin content (mainly cyanidin) in seeds of transgenic DeDFR2 Arabidopsis was higher than in the mutant and wild type ( Fig  6B). The presence and expression of the foreign gene in the transgenic lines were further confirmed by RT-PCR. The amplification of DeDFR2 was observed in transgenic lines, not in wild type or mutants plants (Fig 6C). These results indicated that the overexpression of DeDFR2 in the tt3-1 mutant reconstructed the mutated anthocyanin biosynthesis.

Discussion
The Arabidopsis tt3-1 mutant is characterized by knockout of DFR, resulting in a lack of anthocyanins. Seed plant studies have shown that red/purple leaves or flowers are enriched in anthocyanins [38][39][40][41]. In Malus crabapple, the abundance of anthocyanins in red-leaf cultivars decreased with leaf maturation, but almost no anthocyanins were detected in the evergreen cultivar [42]. Similar to the above result, we found that the red young fronds had the highest anthocyanin content and that the greenish mature fronds had relatively low anthocyanin levels in D. erythrosora. Furthermore, researchers found that anthocyanin concentrations were positively correlated with DFR expression in a variety of plants, such as B. oleracea, G. biloba and Pyrus communis [16,43,44]. By analyzing the transcripts, we found that the anthocyanin content was highly consistent with the trend of changes in DeDFR1 and DeDFR2 expression. In D. erythrosora, DeDFR1 and DeDFR2 were initially expressed in rolled immature fronds to produce anthocyanins. The expression levels of DeDFR1 and DeDFR2 and total anthocyanins peaked in young fronds. After the fronds matured, the expression of DeDFR1 and DeDFR2 decreased, the anthocyanin content was diluted, and the color of the mature fronds gradually deceased in intensity. Similar phenomena also occur in M. crabapple; McDFR1 is mainly expressed in the early stage of red leaf growth [42]. Thus, DeDFR1 and DeDFR2 may be associated with anthocyanin coloration in fronds of D. erythrosora.
Multiple DFR genes are present in plants. At least five different DFRs exist in L. japonicus [13]. The P. hybrida genome contains three DFR genes located on chromosomes 2, 4 and 6 [45]. Both Medicago truncatula and Fagopyrum esculentum contain two different DFRs [12,35]. The Arabidopsis genome contains only one DFR gene [19]. However, there have been few reports of DFRs in ferns, mainly due to the substantial differences between the genomes of ferns and seed plants, making it almost impossible to obtain functional genes by homologous cloning. Conversely, there is a lack of genomic data for ferns-only the genomes of A. filiculoides and S. cucullata have been reported [46]-so there are few molecular studies on ferns. Transcriptome sequencing technology provides a wealth of data for functional genetic studies of species without genomic data. In the present study, we found two DFRs from the transcriptomic data of D. erythrosora, five DFRs from the genomic data of S. cucullata and six DFRs from the genomic data of A. filiculoides. All of these genes belong to the SDR superfamily, with a conserved NAD(P)-binding domain and a substrate-binding domain. DFR evolution can be clearly understood by generating a phylogenic tree with a broad range of plant species. As shown in Fig 2, DFRs of ferns and seed plants were clearly classified into different branches. DeDFR1, AfDFR1 and ScDFR1 clustered into one branch, while DeDFR2, AfDFR2, AfDFR3, and ScDFR2 clustered into another branch. Thus, it can be deduced that DeDFR1 and DeDFR2 are independently evolved and may have different functions in D. erythrosora.
Previous studies have indicated that amino acid residue 133 plays an important role in substrate specificity and classified DFRs into different types based on this site, i.e., Asn-type DFRs, Asp-type DFRs, and non-Asn/Asp-type DFRs, with the amino acid residue at the 133 position being asparagine (Asn), aspartic acid (Asp) and neither Asn nor Asp, respectively [37,47]. Most previous reports have suggested that Asn-type DFRs and Asp-type DFRs have the function of catalyzing dihydroflavonol [13,47,48]. However, we did not find an Asn-type or Asptype DFR in D. erythrosora, S. cucullata and A. filiculoides. Unlike seed plants, most of these DFR proteins in the three ferns have an arginine at the substrate-specificity-determining site. On the other hand, unlike Asn-type or Asp-type DFRs, DeDFR1 and DeDFR2 proteins cannot catalyze DHM. Thus, we speculated that there is a new DFR functional type in ferns, and we named it Arg-type DFR. The Arg-type DFR was not reported in the current research on seed plants, and it might be unique to ferns.
Site-directed mutagenesis is a good way to verify the contribution of a specialized amino acid residue to the activity and substrate preference of DFRs. Based on the experiments, we found that modification of the 133rd amino acid residue of AtDFR could change the enzymatic activity efficiency. The present investigation showed that the modified AtDFR N133D exhibited a reduced ability to catalyze the conversion of DHK to leucopelargonidins, which is consistent with studies of M. truncatula and L. japonicus [12,13]. Although Asp-type DFRs exhibit a certain catalytic ability toward DHK in vitro, pelargonidin-based anthocyanins are not detected in plants with Asp-type DFRs [26,47]. We speculate that in plants with Asp-type DFRs, leucopelargonidins are converted to the highly stable cyanidin/delphinidin-based anthocyanins by F3'H or F3'5'H, making it difficult for the plants to produce pelargonidinbased anthocyanins. Conversely, we found that the modified AtDFR N133R exhibited a reduced ability to catalyze DHQ, and we did not detect the product of DHM, which is similar to the catalytic effects observed for DeDFR1 and DeDFR2. This finding demonstrates that Arg-type DFRs are significantly different from the Asp and Asn types in terms of function. Notably, the product yields of DeDFR1 and DeDFR2 were significantly lower than AtDFR N133R . This finding indicates that in Arabidopsis, the structure of DFR is favorable for substrate binding, so the modified AtDFR N133R is highly efficient. It is possible that other sites can affect binding efficiency via the spatial configuration.
In the present study, the 133rd amino acids of the functional DFRs were Arg, Asn and Asp, all of which are polar amino acids [35]. Studies on the crystal structure of DFRs from V. vinifera have shown that the side chains of N133 interact with both hydroxyl groups of ring B of DHQ. Moreover, the carboxyl and amino groups of N133 can make contacts with the ligand at C3' and C4' of DHQ, respectively [25]. Based on the docking model [35], we analyzed the differences in binding among proteins with different 133rd amino acids and for different ligands. The Asp-type DFR is able to catalyze DHQ via the carboxyl group on the side chain of D133; however, it is difficult to catalyze DHK due to the lack of an amino group in contact with the hydroxyl group on C4'. The Arg-type DFR catalyzes DHQ inefficiently, mainly due to the lack of a carboxyl group on the side chain of R133. The Arg-type DFR can catalyze DHK as efficiently as the Asn-type DFR, mainly via the interaction of the amino group with the hydroxyl group on C4'. These findings explain why different types of DFRs have different catalytic capabilities for dihydroflavonols. In summary, the 133rd amino acid residue of DFRs is an important substrate recognition site, and different amino acid residues at this site directly affect the substrate binding ability.
Catalytic activity analysis showed that the function of DeDFR1 and DeDFR2 is mainly to catalyze DHK and DHQ. Therefore, the expected products are pelargonidin-based and cyanidin-based anthocyanins, but we only detected cyanidin-based and delphinidin-based anthocyanins in D. erythrosora. The lack of pelargonidin-based anthocyanins might be due to the influence of downstream genes, such as F3'H and F3'5'H, which convert the product of DeDFR1 and DeDFR2 to a more stable structure with two or three hydroxyl groups on ring B (Fig 7). In previous studies, a few glycosylation compounds of quercetin, kaempferol, and myricetin were detected in D. erythrosora [29,30], indicating that ferns have enzymes similar to FLS, which can compete with DFRs for substrates. Interestingly, Zhang et al. detected DHM in D. erythrosora [49], indicating that the functional enzymes acting on DHM are not sufficient for complete digestion of this substrate. This finding is consistent with the results showing that DeDFR cannot catalyze DHM. Whether the synthetic process of cyanidin-based and delphinidin-based anthocyanins is consistent with our hypothesis is worth exploring in the future. The development of plants is a synergistic process, and the synthesis of anthocyanins is not controlled by a single gene. The mechanism by which DeDFR and F3'H/F3'5'H synergistically control the anthocyanin type, the competition mechanism between DeDFR and FLS, and whether the process of downstream anthocyanin synthesis in ferns is the same as in seed plants remain to be further studied. The discovery of the new functional Arg-type DFR also provides a new direction for flower cultivation.

Conclusions
The newly identified DeDFR1 and DeDFR2 genes were associated with the accumulation of anthocyanins in D. erythrosora. Moreover, both DeDFR1 and DeDFR2 enzyme proteins can catalyze DHK and DHQ but do not catalyze DHM. DeDFR1 and DeDFR2 are Arg-type DFRs, a previously unreported DFR type. This finding might provide new insights into the anthocyanin biosynthesis pathway in ferns.