Anthocyanins Profiling Analysis and RNA-Seq Revealed the Dominating Pigments and Coloring Mechanism in Cyclamen Flowers

Simple Summary Flower color is a complex plant trait that is mainly controlled by the accumulation of pigments, such as anthocyanins. However, the great diversity of flower color in plants cannot be attributed to a single pigment. Therefore, a more comprehensive approach is needed to fully understand the whole pigment spectrum. In this study, we used metabolomics to profile more than 100 plant pigment in red-flowered cyclamen. By comparing the anthocyanins metabolome in white-flowered cyclamen, we were able to identify key anthocyanins that are highly abundant in red flowers, but low in white flowers. Thus, we can pinpoint the pigments underlying specific flower color phenotype. To further reveal the gene expression network that ultimately controlling the production of pigments, we also performed RNA-Seq using the same plant materials. The transcriptomics revealed a significant altered gene expression profile between red flowers compared to white flowers. More importantly, differential expression analysis allowed us to identify key genes governing the pigment metabolome and flower color. Collectively, our study significantly advanced our understanding of the molecular basis of flower color in cyclamen. Abstract Pigments in cyclamen (Cyclamen purpurascens) endows flowers with great ornamental and medicinal values. However, little is known about the biosynthetic pathways of pigments, especially anthocyanins, in cyclamen flowers. Herein, anthocyanins profiling and RNA-Seq were used to decipher the molecular events using cyclamen genotypes of red (HXK) or white (BXK) flowers. We found that red cyclamen petals are rich in cyanidin-3-O-rutinoside, cyanidin-3-O-glucoside, delphinidin-3-O-glucoside, malvidin-3-O-glucoside, peonidin-3-O-rutinoside, quercetin-3-O-glucoside, and ruti. In addition, our transcriptomics data revealed 3589 up-regulated genes and 2788 down-regulated genes comparing the BXK to HXK. Our rich dataset also identified eight putative key genes for anthocyanin synthesis, including four chalcone synthase (CHS, g13809_i0, g12097_i0, g18851_i0, g36714_i0), one chalcone isomerase (CHI, g26337_i0), two flavonoid 3-hydroxylase (F3′H, g14710_i0 and g15005_i0), and one anthocyanidin synthase (ANS, g18981_i0). Importantly, we found a 2.5 order of magnitude higher expression of anthocyanin 3-O-glucosyltransferase (g8206_i0), which encodes a key gene in glycosylation of anthocyanins, in HXK compared to BXK. Taken together, our multiomics approach demonstrated massive changes in gene regulatory networks and anthocyanin metabolism in controlling cyclamen flower color.


Introduction
Cyclamen (Cyclamen persicum Mill.) is a pot crop extensively cultivated worldwide [1][2][3]. The flower color spans from white to purple, reflecting great genetic diversity and creating tion and stored at −80 • C for until use. Three biological repetitions were performed for all experiments.

RNA Sequencing
RNA sequencing was performed by Wuhan MetWare Biotechnology. Total RNA was extracted using RNAprep Pure Plant Plus (DP441, TIANGEN, Beijing, China) and then enriched for total mRNA using poly-T oligo-attached magnetic beads. For library construction, NEBNext1 Ultra™ RNA Library Prep Kit (NEB, Ipswich, MA, USA) was used according to the kit instructions. Briefly, mRNA was fragmented and first strand cDNA was synthesized using random hexamer primer and M-MuLV Reverse Transcriptase, followed by second strand cDNA synthesis. cDNA fragments were then methylated at the 3 ends and then ligated to adaptor for hybridization. Purificaiton was performed using AMPure XP beads (Beckman Coulter, Beverly, MA, USA). Then, PCR was performed after adding the Index (X) Primer, Universal PCR primers and High-Fidelity DNA polymerase. PCR products were purified and then assessed by Agilent Bioanalyzer 2100 system. Sequencing was performed on Illumina HiSeq2500™ (Illumina, San Diego, CA, USA). The entire experiment was repeated three times.

Splicing of Transcripts, Reads Mapping, Gene Expression Level Quantifying, and Differential Expression Analysis
Reads were filter by removing low quality ones (>10% unknown nucleotides or >50% low quality nucleotides). Then, transcripts were spliced using Trinity. The spliced transcripts were regarded as the reference sequences and the beads were mapped to the spliced transcripts using Hisat2. Relative gene expression was denoted with FPKM. Differential expression analysis was performed using DEGSeq. Two criteria were used for defining differentially expressed genes (DEGs): Fold Change ≥2 or ≤−2, and FDR < 0.01.

Mass Spectrometry
Mass spectrometry was performed on 6500 QTRAP equipped with an ESI Turbo Ion-Spray interface as described. Positive ion mode was used. Analyst 1.6.3 software (AB Sciex, Framingham, MA, USA) was used for setting up instrument method and data acquisition [23].

Metabolite Identification and Quantitative Analysis
Metabolites identification and quantification were performed using MultiaQuant software 3.0.3 with the MWDB database (Metware Biotechnology, Woburn, MA, USA). The spectra of each metabolite were calibrated based on peak pattern and retention time. The reproducibility among different samples was determined by overlapping display the total ion chromatography (TIC) [28][29][30][31].

LC-MS/MS Analyzes the Anthocyanin Metabolic Profiling in Two Cyclamen Varieties with Different Colors
In this study, metabolic analysis of anthocyanin was carried out using an LC-MS-based metabolomics approach.

RNA Sequencing Analysis of Flowers of Two Cyclamen Varieties with Different Color
Six cDNA libraries derived from flowers of HXK and BXK (namely HXK_1, HXK_2, HXK_3, BXK_1, BXK_2, and BXK_3, respectively) were constructed and sequenced. After quality control, 14,8922 transcripts, 11,0942 unigenes, and 21,622 conden genes were acquired ( Table 2). The detailed sequence of spliced transcripts, unigenes, and conden genes are presented in Dataset S2, S3, and S4. Detailed information of beads and probability of mapping of the six samples are presented in Table 3.  DEGs, including 3589 up-regulated genes and 2788 down-regulated genes, were detected comparing the two varieties ( Figure 2A, Dataset S5). To verify the RNA-Seq data, the expression level of nine randomly chosen DEGs was validated. As shown in Figure 2C, qRT-PCR analysis showed that the relative expression patterns of the genes were consistent with RNA-Seq data.

KEGG Analysis of DEGs
Functional subcategories of DEGs were shown in Figure 2B. The most enriched 24 KO terms were beta-alanine metabolism, stilbenoid, diarylheptanoid and gingerol biosynthesis, starch and sucrose metabolism, sesquiterpenoid and triterpenoid biosynthesis, ribosome, photosynthesis-antenna proteins, phenylpropanoid biosynthesis, phenylalanine metabolism, pentose and glucuronate interconversions, PPAR signaling pathway, NODlike receptor signaling pathway, metabolism of xenobiotics by cytochrome P450, linoleic acid metabolism, gap junction, flavonoid biosynthesis, estrogen signaling pathway, DNA replication, cysteine and methionine metabolism, cutin, suberine and wax biosynthesis, bile secretion, antigen processing and presentation, anthocyanin biosynthesis, and ABC transporters. Notably, a large number of genes related to flavonoid biosynthesis, anthocyanin biosynthesis, and phenylpropanoid biosynthesis were significantly differentially expressed in BXK compared to HXK ( Figure 2B). Thus, we next analyzed the expression patterns of the DEGs in these pathways.

KEGG Analysis of DEGs
Functional subcategories of DEGs were shown in Figure 2B. The most enriched 24 KO terms were beta-alanine metabolism, stilbenoid, diarylheptanoid and gingerol biosynthesis, starch and sucrose metabolism, sesquiterpenoid and triterpenoid biosynthesis, ribosome, photosynthesis-antenna proteins, phenylpropanoid biosynthesis, phenylalanine metabolism, pentose and glucuronate interconversions, PPAR signaling pathway, NODlike receptor signaling pathway, metabolism of xenobiotics by cytochrome P450, linoleic acid metabolism, gap junction, flavonoid biosynthesis, estrogen signaling pathway, DNA replication, cysteine and methionine metabolism, cutin, suberine and wax biosynthesis, bile secretion, antigen processing and presentation, anthocyanin biosynthesis, and ABC transporters. Notably, a large number of genes related to flavonoid biosynthesis, anthocyanin biosynthesis, and phenylpropanoid biosynthesis were significantly differentially expressed in BXK compared to HXK ( Figure 2B). Thus, we next analyzed the expression patterns of the DEGs in these pathways.

Analysis of DEGs Related to Anthocyanin Biosynthesis
As shown in Figure 3, the transcript levels of most genes involved in phenylpropanoid biosynthesis and flavonoid biosynthesis were significantly up-regulated in BXK-VS-HXK comparison. Interestingly, the expression levels of many potentially key genes for anthocyanin synthesis, including four potential chalcone synthase genes (CHS, g13809_i0, g12097_i0, g18851_i0, g36714_i0), one potential chalcone isomerase gene (CHI, g26337_i0), two potential flavonoid 3-hydroxylase gene (F3 H, g14710_i0 and g15005_i0), and one potential anthocyanidin synthase gene (ANS, g18981_i0), in flower of HXK were higher than that of BXK ( Figure 3). These genes may be responsible for the synthesis of anthocyanidin, which colored the flower of HXK with red.
VS-HXK comparison. Interestingly, the expression levels of many potentially key genes for anthocyanin synthesis, including four potential chalcone synthase genes (CHS, g13809_i0, g12097_i0, g18851_i0, g36714_i0), one potential chalcone isomerase gene (CHI, g26337_i0), two potential flavonoid 3-hydroxylase gene (F3′H, g14710_i0 and g15005_i0), and one potential anthocyanidin synthase gene (ANS, g18981_i0), in flower of HXK were higher than that of BXK (Figure 3). These genes may be responsible for the synthesis of anthocyanidin, which colored the flower of HXK with red.

g8206_i0, a Potential UFGT Responsible for Glycosylating Anthocyanins
Anthocyanins can be further glycosylated by UFGT (uridine diphosphate-glucose: flavonoid 3-O-glucosyltransferase), leading to the formation of glycosylated anthocyanins. Significantly, glycosylated anthocyanins are the major anthocyanins that contribute to the color variation among different varieties. We found the abundance of 3-O-glucoside and 3-O-rutinoside glycosylated cyanidin, including delphinidin, malvidin, pelargonidin, and peonidin, are higher than other derivatives in HXK flowers (Table 1). Therefore, we next focused on glucosyltransferase, the key enzyme for the glycosylation of cyanidin, delphinidin, malvidin, pelargonidin, and peonidin. KEGG analysis of DEGs between in BXK and HXK showed that two UFGTs were up-regulated (Figure 3). The expression of g8206_i0 in HXK, encoding a potential UFGT responsible for glycosylating cyanidin, delphinidin, and pelargonidin and producing cyanidin-3-O-glucoside, delphinidin-3-O-glucoside, and pelargonidin-3-O-glucoside, is more than 2.5 orders of magnitude more than that in BXK (Figure 4). Similarly, the expression of g5626_i0 in the flower of HXK is more than 4 orders of magnitude more than that in BXK (Figure 4). Moreover, the expression of g8206_i0 is prominent higher than g5626_i0 in flower of HXK. Thus, we reasoned that g8206_i0 is a key UFGT responsible for glycosylating cyanidin, delphinidin, and pelargonidin to produce cyanidin-3-O-glucoside, delphinidin-3-O-glucoside, and pelargonidin-3-O-glucoside, which play an important role in accumulating glycosylated anthocyanins in flowers of HXK.

Discussion
Flavonoids are further divided into anthocyanins, flavonoids, flavonols, and flavanones. Anthocyanins are anthocyanins that control the pink, red, blue, purple, and redpurple colors of flowers. Anthocyanins can be divided into three types: pelargonin, delphinidin, and cyanidin, which usually exist alone in different flowers. There are also three methoxyl substitution products of anthocyanins in plants, namely paeoniflorin, morning glory pigment, and mallow pigment. Geranium pigments are brick red, corn-flower pigments and peony pigments are purple-red, and delphinium pigments, morning glory pigments, and mallow pigments are blue-violet [32]. Flavonoids and flavonols are yellow or colorless; orangeone and chalcone often appear in the same petal together and appear dark yellow [33].
Cyclamen flowers' color diversity are largely due to the differential accumulation of two categories of metabolites-carotenoids (yellow or deep orange pigments) and flavonoids (red, blue, or purple pigments). In general, the flower color can be affected by several factors. The first is pH as anthocyanins show different colors at different pH [34]. Second, anthocyanins can form metalloanthocyanin complex with Al 3+ and Mg 2+ , which can give distinct flower colors [35]. Third, anthocyanins can also interact with other colorless molecules, such as copigments [35] or other anthocyanins [36], to enhance the color. Such a theory is also supported by the later intermolecular stacking theory [37]. For instance, polyacylated anthocyanins can form stable hydrophobic interactions that give blue color even without copigment or metal ions in a wide pH range. While these theories can explain the flower color to different extent, the molecular identifies underlying the color variation remain elusive. In this study, metabolic profiling of anthocyanin was performed to explore the dominating pigments endowing the cyclamen petal with red using white flowers as a control. In addition, we performed comparative transcriptome analysis to delineate key genes controlling the anthocyanin biosynthetic pathways.

Discussion
Flavonoids are further divided into anthocyanins, flavonoids, flavonols, and flavanones. Anthocyanins are anthocyanins that control the pink, red, blue, purple, and red-purple colors of flowers. Anthocyanins can be divided into three types: pelargonin, delphinidin, and cyanidin, which usually exist alone in different flowers. There are also three methoxyl substitution products of anthocyanins in plants, namely paeoniflorin, morning glory pigment, and mallow pigment. Geranium pigments are brick red, cornflower pigments and peony pigments are purple-red, and delphinium pigments, morning glory pigments, and mallow pigments are blue-violet [32]. Flavonoids and flavonols are yellow or colorless; orangeone and chalcone often appear in the same petal together and appear dark yellow [33].
Cyclamen flowers' color diversity are largely due to the differential accumulation of two categories of metabolites-carotenoids (yellow or deep orange pigments) and flavonoids (red, blue, or purple pigments). In general, the flower color can be affected by several factors. The first is pH as anthocyanins show different colors at different pH [34]. Second, anthocyanins can form metalloanthocyanin complex with Al 3+ and Mg 2+ , which can give distinct flower colors [35]. Third, anthocyanins can also interact with other colorless molecules, such as copigments [35] or other anthocyanins [36], to enhance the color. Such a theory is also supported by the later intermolecular stacking theory [37]. For instance, polyacylated anthocyanins can form stable hydrophobic interactions that give blue color even without copigment or metal ions in a wide pH range. While these theories can explain the flower color to different extent, the molecular identifies underlying the color variation remain elusive. In this study, metabolic profiling of anthocyanin was performed to explore the dominating pigments endowing the cyclamen petal with red using white flowers as a control. In addition, we performed comparative transcriptome analysis to delineate key genes controlling the anthocyanin biosynthetic pathways.
The biosynthesis pathway of anthocyanins has been divided into two stages: anthocyanidin formation and anthocyanin modification such as glycosylation, methylation, and acylation [39]. Key enzymes for anthocyanidin biosynthesis include ANS, CHI, CHS, DFR, F3H, F3 H, and F3 5 H [6,8,9]. In our data, we found eight potentially key genes for anthocyanin synthesis, including four CHSs (g13809_i0, g12097_i0, g18851_i0, g36714_i0), one CHI (g26337_i0), two F3 Hs (g14710_i0 and g15005_i0), and one ANS (g18981_i0). Not surprisingly, our data showed that the expression of these eight genes are higher in the flower of HXK than that of BXK.
Anthocyanin modifications facilitate the formation of more anthocyanin subtypes. Glycosylation promotes the solubility, chemical stability, storage, and transportation of anthocyanins [40]. 3GT/UFGT (flavonoid 3-O-glucosyltransferase) is commonly considered as the first enzyme that catalyzes the glycosylation with several genes characterized in diverse plant species such as Vitis vinifera, Petunia hybrida, and Freesia hybrida [41,42]. Recently, an anthocyanin 5-O-glucosyltransferase (Cpur5GT) catalyzing the glycosylation of 3-glucoside-type anthocyanidins at the 5-O-position was identified from cyclamen [27]. Here, our transcriptomics data identified a potential UFGT (g8206_i0) that may execute the glycosylation of cyanidin (to give cyanidin-3-O-glucoside), delphinidin (to give delphinidin-3-O-glucoside), and pelargonidin (to give pelargonidin-3-O-glucoside). The mRNA of g8206_i0 in HXK flower is more than 2.5 orders more than that in BXK. The exact role of this gene needs further validation.