CmHY5 functions in apigenin biosynthesis by regulating flavone synthase II expression in chrysanthemum flowers

The predominant flavones in the ray florets of chrysanthemum flowers are apigenin and its derivatives. CmHY5 participates in apigenin biosynthesis by directly regulating the expression of FNSII-1 in chrysanthemum. Chrysanthemum (Chrysanthemum morifolium) flowers have been used for centuries as functional food and in herbal tea and traditional medicine. The chrysanthemum flower contains significant amounts of the biologically active compound flavones, which has medicinal properties. However, the mechanism regulating flavones biosynthesis in chrysanthemum flowers organs is still unclear. Here, we compared the transcriptomes and metabolomes of different floral organs between two cultivars with contrasting flavone levels in their flowers. We identified 186 flavonoids by metabolome analysis. The predominant flavones in the ray florets of chrysanthemum flowers are apigenin and its derivatives, of which the contents are highly correlated with the expression of flavones synthase II gene CmFNSII-1. We also determined that CmHY5 is a direct upstream regulator of CmFNSII-1 transcription. We showed that CmHY5 RNAi interference lines in chrysanthemum have lower contents of apigenin compared to wild-type chrysanthemum. Our results demonstrated that CmHY5 participates in flavone biosynthesis by directly regulating the expression of FNSII-1 in chrysanthemum.


Introduction
Flavonoids are secondary metabolites that exist in various parts of plants, especially in flowers and leaves. Flavonoids can be divided into several classes according to their structures: flavones, flavonols, flavanones, flavanols, isoflavones, and anthocyanins (Andersen and Markham 2006). As one of the largest groups of flavonoids, flavones have diverse functions and play important roles in protection against ultraviolet (UV) irradiation (Schmitz-Hoerner and Weissenböck 2003), interaction with other organisms (Peters et al. 1986;Kong et al. 2007), auxin transport (Mathesius et al. 1998), copigmentation with anthocyanins (Yoshida et al. 2009), and lignification (Lan et al. 2015). In addition, flavones provide nutritional value and are beneficial to human health, since they have many antioxidant properties, as well as anticancer, antimicrobial, anti-inflammatory, and antiviral activities (Hostetler et al. 2017). Apigenin is one of the most widely distributed flavones and is abundant in a variety of vegetables and fruits, such as parsley (Petroselinum crispum), celery (Apium graveolens), chamomile (Matricaria chamomilla), oranges (Citrus × sinensis), thyme (Thymus vulgaris), and onions (Allium cepa), as well as in beverages derived from plants, such as tea, beer, and wine (Bak et al. 2016;Zhou et al. 2016). Apigenin in plants typically exists in glycosidic forms, which are catalysed by glycosyltransferases. The modification confers structural complexity, and is critical for the storage and transportation of apigenin in plants (Hostetler et al. 2017). Many studies conducted over the past several years have demonstrated that apigenin has potential antioxidant, antiinflammatory, and anticancer properties (Shukla and Gupta 2010).
Extensive studies have shown that the expression of structural genes in flavonoids biosynthesis is regulated in an independent or combinatorial manner by MYB and basic helix-loop-helix (bHLH) transcription factors, as well as WD40 proteins (Hichri et al. 2011). GtMYBP3 and GtMYBP4 in gentian flower, and MYB transcription factors P1 and P2 in maize positively regulate flavone biosynthesis (Zhang et al. 2003;Li et al. 2016). In chrysanthemum, CmMYB012 negatively regulates flavone biosynthesis in response to high temperature .
Chrysanthemum is an important ornamental crop worldwide whose flowers have been used in Chinese herbal medicine, for making tea, and as functional food for thousands of years (Lee et al. 2009;Xie et al. 2012;Li et al. 2019). Apigenin is the major active component in chrysanthemum flowers (Xie et al. 2012). Here, we investigated how apigenin biosynthesis is regulated in chrysanthemum flowers. Our results demonstrated that CmHY5 can directly regulate the expression of CmFNSII-1 to modulate flavone biosynthesis.

Plant materials and growth conditions
Chrysanthemum cultivars and wild chrysanthemum (Chrysanthemum indicum) were obtained from Chrysanthemum Germplasm Resource Preservation Center of Beijing Academy of Agriculture and Forestry (Beijing, China). The chrysanthemum plants were propagated by tissue culture. Chrysanthemum young branch cuttings were grown on 1/2 Murashige and Skoog (MS) medium for 45 days. The plantlets of wild chrysanthemum were planted into a mixture of 1:1 (v/v) peat and vermiculite and growth in a growth chamber (22 °C ± 2 °C, 20-45% relative humidity, and an 8/16 h light/dark photoperiod).
For transcriptome and metabolomic analyses, two cultivars were cultivated in a greenhouse at day/night temperatures of 24 °C/18 °C, 40% relative humidity, and 8/16 h light/dark photoperiod. Floral bud and organs (ray florets and disk florets) at different developmental stages (S1, S2, and S3) were harvested as three biological replicates. The samples were divided into two parts: one for transcriptome analysis and the other for metabolomic analysis.

Metabolite profiling
The sample preparation, extraction, metabolites' identification, and quantification were carried out according to the standard procedures described by Peng et al. (2017) and performed at MetWare Biotechnology Co., Ltd. (Wuhan, China). Metabolites were qualified following the method described by Chen et al. (2013). Flavonoids were quantified according to standards of apigenin.

RNA-seq analysis
Total RNA was extracted from the flowers of the cultivars '14_C_1' and 'Yutai' at three developmental stages after flowering (three independent plants for each stage and replicate) using the Plant RNA Extraction Kit (TaKaRa, Beijing, China). Thirty libraries were prepared and sequenced on an Illumina HiSeq 4000 instrument by Gene Denovo Biotechnology Co. (Guangzhou, China) (Zhong et al. 2011). The workflow of data processing, assembling, and annotation was described by Yang et al. (2014).

Genes cloning and phylogenetic analysis
Total RNA was extracted from chrysanthemum flowers at different developmental stages using the Plant RNA Extraction kit (TaKaRa). cDNAs for cloning were synthesised using PrimeScript™ RT Master Mix (TaKaRa). Amino acid sequence alignment and phylogenetic analysis were performed in MEGA 6.0 software with neighbour-joining method running 1000 bootstrap replications (Tamura et al. 2013).

Reverse transcription quantitative PCR (RT-qPCR) analysis
Total RNA was extracted from chrysanthemum flowers using the Plant RNA Extraction kit (TaKaRa). 1 μg of total RNA was used to synthesise first-strand cDNAs following the manufacturer's instructions of the PrimeScript™ RT Master Mix (TaKaRa). RT-qPCR reactions (using 2 μL cDNA as the template) were performed using the StepOne Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) with the TB Green Fast qPCR Mix (TaKaRa). The chrysanthemum Ubiquitin gene was used as an internal control (Wei et al. 2017). Yeast one-hybrid (Y1H) assay Y1H assays were performed according to the Yeast Protocols Handbook (TaKaRa). CmFNSII-1 promoter fragments were inserted into the pLacZi vector, and the coding sequences of CmMYB32, CmMYB330, CmbHLH79, CmHY5, CmP, and CmBIM1 were individually cloned into the pJG4-5 vector. The plasmids were then transformed into yeast strain EGY48. Individual colonies were selected on synthetic defined (SD) medium lacking uracil and tryptophan and containing X-gal for blue/white screening.

Dual-luciferase assay
According to a previous protocol (Yin et al. 2010), the fulllength coding sequences of CmHY5 or CmbHLH79 were cloned individually into the pGreen II 62-SK vector. The promoter fragment of CmFNSII-1 was inserted into upstream of the firefly luciferase (LUC) reporter gene in the pGreen II 0800-LUC vector. The internal control in vector pGreen II 0800-LUC is Renilla luciferase (REN) gene promoted by the cauliflower mosaic virus (CaMV) 35S promoter. The above plasmids were transiently infiltrated in pairs in Nicotiana benthamiana leaves via Agrobacterium (Agrobacterium tumefaciens)-mediated infiltration (strain GV3101). The activation of transcription was determined 3 days after infiltration and expressed as the ratio between LUC and REN activity, as measured on a Modulus Luminometer (Promega, Beijing, China). For each reporter/effector pair, each Agrobacterium culture was resuspended in infiltration buffer (10 mM 2-(N-morpholino)-ethanesulfonic acid, 10 mM MgCl 2 , 0.2 mM acetosyringone, pH 5.7) to a final OD 600 of 1.0, before mixing 10 mL of the effector cell suspension with 1 mL of reporter cell suspension. As a negative control and for calibration, the empty vector SK was co-infiltrated with the reporter plasmids, and the resulting LUC/REN ratio was set to 1. After 48 h infiltration, the leaves were treated in the dark for 5 min before measuring LUC-derived luminescence by a CDD camera (CHEMIPROHT 1300B/LND, 16 bits; Roper Scientific, Sarasota, FL, USA).

Chrysanthemum transformation
To construct the RNA interference (RNAi) plasmid, two 288-bp CmHY5 fragments containing AscI/SwaI and BamHI/PacI sites were prepared, respectively. Each fragment was cut with the corresponding restriction enzymes and cloned into the pFGC1008 vector to form an intron-containing "hairpin" RNA (iphRNA) construct. The resulting plasmid was transformed into chrysanthemum according to the method described by Hong et al. (2006). Hygromycinresistant primary transformants were screened by PCR to confirm the presence of the transgene. Fig. 2 Levels of the major flavones in different floral organs of the two chrysanthemum cultivars 14_C_1 and Yutai. S1B, bud at opening stage 1; S2D, disk floret at opening stage 2; S2R, ray floret at opening stage 2; S3D, disk floret at opening stage 3; S3R, ray floret at opening stage 3. Bars represent means ± SD (n = 3). Asterisks indicate statistically significant differences between the two cultivars, as determined by two-tailed Student's t test (*P < 0.05, **P < 0.01)

Integrative analysis of the transcriptome and metabolome revealed the correlation of structural genes' expression with the major flavonoid contents
Based on the KEGG pathway annotation of unigenes and metabolites from transcriptome and metabolome datasets, we identified eight genes encoding enzymes associated with the flavonoid biosynthetic pathway (Fig. 3a). Among the eight putative structural genes, CHS and FNSII (Uni-gene0057769) were more highly expressed in ray florets than in disk florets, with higher expression levels in stage 3 ray florets of the cultivar 14_C_1 relative to Yutai. Moreover, the expression of FLS and F3H was higher in disk florets than in ray florets (Fig. 3b). Gene-to-metabolite correlation indicated that contents of apigenin and its derivative were tightly correlated with FNSII expression levels (Fig. 3c).
The phylogenetic analysis of the predicted protein encoded by Unigene0057769 with other FNSII proteins determined that Unigene0057769 is most closely related to CmFNSII and CiFNSII, two known FNSII enzymes from C. morifolium and C. indicum (Jiang et al. 2019;Wang et al. 2021) (Fig. 4a). Accordingly, we renamed Unigene0057769 as CmFNSII-1. We also validated the RNA-seq results by RT-qPCR (Fig. 4b). Moreover, RT-qPCR results showed that CmFNSII-1 reaches its highest expression levels in ray florets compared to other organs, such as roots, stems, leaves, and floral organs (Fig. 4c).

CmHY5 is a direct upstream regulator of CmFNSII-1
To investigate the regulatory mechanism of CmFNSII-1, we performed a co-expression analysis between genes encoding transcription factors and CmFNSII-1. Genes from the transcription factor gene families MYB, bZIP, basic helix-loop-helix (bHLH), WRKY, and MADS showed relatively high co-expression (Pearson's correlation coefficient Relative CmFNSII-1 expression levels in different floral organs of the two cultivars 14_C_1 and Yutai, as determined by RT-qPCR. c Relative CmFNSII-1 expression levels in different floral organs of cultivar 14_C_1, determined by RT-qPCR analysis. S1B, bud at opening stage 1; S2D, disk floret at opening stage 2; S2R, ray floret at opening stage 2; S3D, disk floret at opening stage 3; S3R, ray floret at opening stage 3. Chrysanthemum Ubiquitin was used as the internal control. Bars represent means ± SD (n = 3). Asterisks indicate statistically significant differences between the two cultivars, as determined by two-tailed Student's t test (**P < 0.01). Letters indicate significant differences according to Tukey-Kramer test (P < 0.05) r > 0.6) with CmFNSII-1 (Fig. 5a, Dataset S3). To narrow down which transcription factor(s) can directly regulate the expression of CmFNSII-1, we cloned a 2-kb CmFNSII-1 promoter fragment and looked for putative cis-acting regulatory elements in the PlantCARE database (Fig. S2) (Lescot et al. 2002). The promoter of CmFNSII-1 contained MYBbinding sites, MYC-binding sites, and cis-acting regulatory elements involved in light responsiveness, such as G-box and GATA-motif (Fig. S2). Based on co-expression analysis and putative cis-elements along the CmFNSII-1 promoter, we selected six candidate transcription factors whose gene expression patterns positively correlated with apigenin contents (Fig. 5b). We tested their binding to the CmFNSII-1 promoter by Y1H assay. We established that CmbHLH79 and CmHY5 can bind to the proximal 1-kb fragment (P1) of the CmFNSII-1 promoter (Fig. 5c). We also conducted a dual-LUC reporter assay in N. benthamiana leaves to assess the regulation of the CmFNSII-1 promoter by CmbHLH79 and CmHY5 in vivo. We observed that leaf cells expressing CmHY5 drastically activate the transcription of the CmFNSII-1pro:LUC reporter (Fig. 5d). Indeed, expressing the CmHY5 effector construct resulted in a 2.2-fold increase in LUC activity from the CmFNSII-1pro:LUC reporter construct, whereas the CmbHLH79 effector construct did not increase LUC activity (Fig. 5d).

Silencing CmHY5 reduces apigenin contents in floral organs
To investigate whether CmHY5 influences the production of flavones in chrysanthemum, we silenced CmHY5 in wild chrysanthemum (C. indicum), leading to the isolation of 10 CmHY5-RNAi lines. We confirmed lower CmHY5 transcript levels in the RNAi lines by RT-qPCR (Fig. 6a). We noticed that the RNAi lines were taller than wild-type (WT) non-transgenic plants (Fig. 6b, c), whereas no significant difference were observed in flower diameter between WT and RNAi lines (Fig. S3). In addition, RT-qPCR results indicated that except of FLS, the expression of CmFNSII-1 and other structural genes was significantly reduced in CmHY5-RNAi plants compared to WT plants (Fig. 6d). We also measured the contents of apigenin and apigenin-7-O-glucoside by LC-MS, which revealed that RNAi lines accumulated less apigenin and apigenin-7-Oglucoside in their leaves and flowers than the WT (Fig. 7). Taken together, our results demonstrated that CmHY5 is Fig. 5 CmHY5 binds to the CmFNSII-1 promoter. a Transcriptional regulatory network of CmFNSII-1. Different transcription factor families are shown in different colours. The grey lines indicate high correlation between the expression of CmFNSII-1 and transcription factor genes (Pearson's correlation coefficient r > 0.6). Nodes with red circles were identified as candidate CmFNSII-1 regulators. b Candidate CmFNSII-1 regulators identified from co-expression analysis (left) and expression of the candidate regulator genes in different floral organs in the two cultivars, shown as a heatmap. S1B, bud at opening stage 1; S2D, disk floret at opening stage 2; S2R, ray floret at opening stage 2; S3D, disk floret at opening stage 3; S3R, ray floret at open-ing stage 3. c CmHY5 binds to the CmFNSII-1 promoter, as shown by yeast one-hybrid assay. The CmFNSII-1 promoter was divided into two fragments (P1, P2). Interactions were determined by yeast cell growth and confirmed by colour indication on SD-Trp-Ura mediumcontaining X-gal. d Dual-luciferase assays using different combinations of CmHY5 and bHLH79 proteins as effectors and a CmFN-SII-1 promoter construct in N. benthamiana leaves. LUC activity was tested 3 d after infiltration. Bars represent means ± SD (n = 4-5). Asterisks indicate statistically significant differences, Student's t test (**P < 0.01) 7 Page 8 of 11 involved in flavone biosynthesis by regulating the expression of CmFNSII-1 in chrysanthemum flowers.

Discussion
The expression of genes participating in the same biosynthesis pathway is often coordinately regulated, which can be assessed by generating co-expression modules. If the compounds of interest of these gene products exhibit spatial, temporal, and/or condition-specific expression, the underlying co-expression patterns would become apparent after quantification of expression, as done here by RNA-seq. Therefore, the integration of transcriptomic and metabolomic datasets is a powerful tool to identify candidate genes responsible for the accumulation of specific metabolites. In this study, we implemented this approach to identify candidate regulatory genes involved in flavone biosynthesis in chrysanthemum flowers. We showed that flavonoids accumulated distinctively in disk florets and ray florets. In addition, apigenin and its derivatives accumulated abundantly in ray florets (Fig. 1b). Organ-specific accumulation of metabolites is of special importance for the survival and adaptation of plant species.
Mounting evidence has illustrated the crosstalk between flavonoids and various phytohormones in various developmental processes, including seed size and fertility (Buer 2013;Doughty et al. 2014;Maloney et al. 2014). The expression of FNSII and FLS in chrysanthemum was higher in flower than other tissues (Jiang et al. 2019;Wang et al. 2021). However, the details of their expression patterns in flower organs have not been studied. In this work, we noted RNAi−4,−5, and −7 are three independent CmHY5-RNAi lines. Bars represent means ± SD (n = 3). Asterisks indicate statistically significant differences between the WT and each RNAi line, Student's t test (**P < 0.01) that CHS and FNSII transcripts reach high accumulation levels in ray florets, while those of F3H and FLS accumulated in disk florets (Fig. 3b), suggesting that flavonoids may play different roles in the development of disk and ray florets. Since chrysanthemum flower is composed of multiple individual ray and disk florets, and the details of the expression patterns of structural genes in florets will be beneficial for breeding of high flavonoids' chrysanthemum cultivars.
Among the regulators of flavonoid biosynthesis, MYB transcription factors have been comprehensively studied (Liu et al. 2015). Recently, in chrysanthemum, a novel transcription factor CmMYB012 has been shown to inhibit flavone and anthocyanin biosynthesis by directly regulating CmFNS and other structural genes in response to high temperature ). In our co-expression analysis, we found that the expression of MYB gene families was highly correlated with the expression of CmFNSII-1 (Fig. 5a). However, none of the MYB transcription factors that we tested, directly bind to the promoter of CmFNSII-1 (Fig. 5c).
In addition, other potential regulators, which have been reported to be involved in the flavonoid pathway, such as bHLH, WRKY, MADS, and bZIP transcription factors (Ishida et al. 2007;Jaakola et al. 2010;Wang et al. 2016;Shi et al. 2018), are also found in the co-expression network (Fig. 5a). Yeast one-hybrid experiments showed that CmbHLH79 and CmHY5 can directly bind to the promoter P1 of CmFNSII-1. However, only CmHY5 can activate the transcription of CmFNSII-1 (Figs. 5c, d), suggesting a role of CmHY5 in flavonoid biosynthesis.
To date, most of the genes and transcription factor genes involved in anthocyanin biosynthesis appear to be controlled by HY5 (Gangappa and Botto 2016). By contrast, the functions and regulatory mechanisms of HY5 in flavone biosynthesis remain uncharacterised, perhaps because Arabidopsis does not harbour FNS genes in its genome (Martens and Mithöfer 2005). In maize, two FNS enzymes participating in the biosynthesis of apigenin are regulated by UV-B, suggesting that FNS-catalysed apigenin accumulation might be dependent on the HY5 signalling pathway (Righini et al. 2019). In this study, yeast one-hybrid and dual-luciferase assays revealed that CmHY5 can activate the transcription of CmFNSII-1 via directly bind to its promoter, indicating that HY5 positively regulates flavone biosynthesis. The expression of CHS, CHI, F3H, and F3'H was downregulated in CmHY5-RNAi lines (Fig. 6d), which are in agreement with previous studies (Oyama et al. 1997;Holm et al. 2002;Shin et al. 2007;Song et al. 2008;Stracke et al. 2010). In addition, CmFNSII-1 was also downregulated in CmHY5-RNAi lines, which was consistent with the reduction of apigenin accumulation in CmHY5-RNAi lines (Figs. 6d and 7). RNAi−4,−5, and −7 are three independent CmHY5-RNAi lines. Bars represent means ± SD (n = 3). Asterisks indicate statistically significant differences between the WT and each RNAi line, Student's t test (**P < 0.01) 7 Page 10 of 11

Conclusion
In this study, we compared the transcriptomes and metabolomes of different flora organs between two cultivars with contrasting flavone levels in their flowers. The analysis of the metabolome revealed that the major flavonoids in the ray florets were apigenin and its derivatives, which were accumulated higher in 14_C_1 than Yutai. We further established that CmHY5 participates in flavone biosynthesis by directly regulating the expression of CmFNSII-1 in chrysanthemum. Silencing of CmHY5 reduced apigenin contents in floral organs and leaves. Our results provide a strategy of molecular breeding for improving the quality of medicinal chrysanthemum.
Author contribution statement CM, CL, and CH conceived and designed the experiments. CL, LL, JZ, and YX performed experiments. HL, DC, XC, and JG provided technical support and conceptual advice. CM, CL, BH, and CH analysed the data and wrote the manuscript.