Volume 228, Issue 6 p. 1864-1879
Full paper
Free Access

The spatio-temporal biosynthesis of floral flavonols is controlled by differential phylogenetic MYB regulators in Freesia hybrida

Xiaotong Shan

Xiaotong Shan

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

These authors contributed equally to this work.

Search for more papers by this author
Yueqing Li

Yueqing Li

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

These authors contributed equally to this work.

Search for more papers by this author
Song Yang

Song Yang

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

Search for more papers by this author
Zhongzhou Yang

Zhongzhou Yang

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

Search for more papers by this author
Meng Qiu

Meng Qiu

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

Search for more papers by this author
Ruifang Gao

Ruifang Gao

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

Search for more papers by this author
Taotao Han

Taotao Han

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

Search for more papers by this author
Xiangyu Meng

Xiangyu Meng

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

Search for more papers by this author
Zhengyi Xu

Zhengyi Xu

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

Search for more papers by this author
Li Wang

Corresponding Author

Li Wang

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

Authors for correspondence:

Li Wang

Tel: +86 431 85099360

Email:[email protected]

Xiang Gao

Tel: +86 431 85099360

Email:[email protected]

Search for more papers by this author
Xiang Gao

Corresponding Author

Xiang Gao

Key Laboratory of Molecular Epigenetics of MOE and Institute of Genetics & Cytology, Northeast Normal University, Changchun, 130024 China

Authors for correspondence:

Li Wang

Tel: +86 431 85099360

Email:[email protected]

Xiang Gao

Tel: +86 431 85099360

Email:[email protected]

Search for more papers by this author
First published: 22 July 2020
Citations: 44

Summary

  • Floral flavonols play specific pivotal roles in pollinator attraction, pollen germination and fertility, in addition to other functions in vegetative organs. For many plants, the process of flavonol biosynthesis in late flower development stages and in mature flower tissues is poorly understood, in contrast to early flower development stages. It is thought that this process may be regulated independently of subgroup 7 R2R3 MYB (SG7 MYB) transcription factors.
  • In this study, two FLS genes were shown to be expressed synchronously with the flower development-specific and tissue-specific biosynthesis of flavonols in Freesia hybrida.
  • FhFLS1 contributed to flavonol biosynthesis in early flower buds, toruses and calyxes, and was regulated by four well-known SG7 MYB proteins, designated as FhMYBFs, with at least partial regulatory redundancy. FhFLS2 accounted for flavonols in late developed flowers and in the petals, stamens and pistils, and was targeted directly by non SG7 MYB protein FhMYB21L2. In parallel, AtMYB21 and AtMYB24 also activated AtFLS1, a gene highly expressed in Arabidopsis anthers and pollen, indicating the conserved regulatory roles of MYB21 against FLS genes in these two evolutionarily divergent angiosperm plants.
  • Our results reveal a novel regulatory and synthetic mechanism underlying flavonol biosynthesis in floral organs and tissues which may be exploited to investigate supplementary roles of flavonols in flowers.

Introduction

Flavonol, one of the most abundant and widespread major subgroups of flavonoids, is universally synthesized by different plant lineages traced back to liverworts and mosses (Bowman et al., 2017; Tsugawa et al., 2019). During the evolution and diversification of plant species, flavonols acquire a considerable diversity of functions, ranging from physiological to ecological roles in plant growth, development, and defense processes (Nakabayashi et al., 2014; Watkins et al., 2014; Ng et al., 2015; Silva-Navas et al., 2016; Wang et al., 2018; Tan et al., 2019). Besides roles in vegetative organs, flavonols have additional roles in floral organs and tissues. Firstly, flavonols are necessary for pollen viability, germination, and tube growth in some crop species, such as petunia, tobacco, tomato and maize (Mo et al., 1992; Pollak et al., 1993; Vogt et al., 1994; Ylstra et al., 1994; Muhlemann et al., 2018). Secondly, as UV-absorptive or reflective reagents which could be perceived by insects, flavonols contribute to attracting guilds of pollinators such as bees and moths (Sheehan et al., 2016; Yuan et al., 2016). Thirdly, flavonols could affect flower color hue as co-pigments of anthocyanins (Takahashi et al., 2007; Czemmel et al., 2009). Because of this functional versatility, flavonols are detected in almost all plant tissues and organs, especially in flowers, and exhibit complicated spatial and developmental accumulation profiles (Eldik et al., 1997; Owens et al., 2008b; Liu et al., 2018). Although competition between flavonol and anthocyanin biosynthesis has been extensively reported (Yuan et al., 2016), it is not a universal rule that flavonols are accumulated with biases in differently colored flowers. For instance, the total content of flavonols was not found to differ significantly between red and white flower phenotypes in some species (Luo et al., 2016). A significant negative correlation between anthocyanins and flavonols was also not found between white species or morphs and normally pigmented species in Iochrominae species (Larter et al., 2019), further demonstrating the complexity of flavonol biosynthesis in flowers. Consequently, the biosynthesis and regulation of flavonols in floral organs is of great importance and is much more complicated than was previously thought.

Generally, the biosynthesis of flavonols can be divided into three steps. The first step is the production of dihydroflavonols – catalyzed by cascade enzymes encoded by early flavonoid biosynthetic genes – which are then converted into flavonols through the action of flavonol synthase (FLS) in the second step. Subsequently, flavonols are further modified to increase their stability. Alternatively, dihydroflavonols can also be reduced to leucoanthocyanidins by dihydroflavonol 4-reductase (DFR), directing the metabolic flux into anthocyanin and proanthocyanidin branches (Luo et al., 2016; Yuan et al., 2016; Yonekura-Sakakibara et al., 2019). Therefore, FLS is one of the most important rate-limiting enzymes in terms of determining the components and quantities of flavonols in plants. Since the first FLS was verified in the protein extracts from irradiated parsley cells (Britsch et al., 1981), dozens of FLS enzymes have been characterized from various plant species. Generally, there is only one FLS gene characterized from various plants (Holton et al., 1993; Moriguchi et al., 2002; Nielsen et al., 2002; Toh et al., 2013; Kim et al., 2014; Vu et al., 2015; Liu et al., 2017; Park et al., 2017; Zhou et al., 2017; Liu et al., 2019; Sun et al., 2019). Gene expression analysis showed that the functional FLSs were mainly expressed in young floral buds, which did not correspond well with the disproportionately high total flavonol content in the flowers and tissues in later stages of development in some plant species (Davies et al., 1993; Holton et al., 1993; Nielsen et al., 2002; Liu et al., 2019). Comparatively, more FLS genes have been isolated in other plants (Downey et al., 2003; Owens et al., 2008a; Ferreyra et al., 2010; Ferreyra et al., 2012; Xu et al., 2012; Li et al., 2013; Akita et al., 2017; Zeng et al., 2017; Matsui et al., 2018; Sun et al., 2019). However, none of these FLS genes have been reported to be highly expressed in the flowers and tissues in later stages of development (e.g. anther and pollen) except FtFLS1 from Fagopyrum tataricum (Li et al., 2013). Considering the roles of flavonols in pollinator attraction and plant fertility, we cannot underestimate the importance of screening for new functional FLS genes specifically in fully opened flowers.

In plants, R2R3 MYB regulators can be categorized into at least 25 subgroups (Stracke et al., 2001). Among them, members of the subgroup 7 (SG7) family have been extensively characterized and found to be responsible for FLS expression in crops, vegetables and medical plants (Czemmel et al., 2009; Ferreyra et al., 2010; Liu et al., 2015; Fernandez-Moreno et al., 2016; Huang et al., 2016; Liu et al., 2016; Wang et al., 2017; Allan & Espley, 2018; Kocabek et al., 2018; Matsui et al., 2018; Cao et al., 2019; Zhai et al., 2019). However, very few MYB regulators have been shown to control FLS genes expressed in floral organs and tissues, and the identified flower-specific MYB regulators were either highly expressed in young petals (Nakatsuka et al., 2012) or were found to prepattern the corolla pigmentation (Yuan et al., 2016); whether they could control the FLS genes dominant in the fully opened flowers or tissues is largely unknown. For instance, studies in Arabidopsis revealed that AtFLS1 was expressed in anther or pollen independent of AtMYB11/12/111 (Stracke et al., 2010), suggesting that a combination of regulators might participate in flavonol biosynthesis in these tissues. In fact, AtMYB99 was found to be involved in the pollen flavonol biosynthesis in a regulatory triad with AtMYB21 and AtMYB24; however, whether the triad could directly target the AtFLS1 gene was not reported (Battat et al., 2019). Hence, the transcription regulation of the FLS genes in mature flowers might be controlled by diverse regulators – this requires further investigation.

To date, numerous flavonoid biosynthetic genes and their regulators have been functionally identified in Freesia flowers (Sun et al., 2016; Li et al., 2019; X. Meng et al., 2019; Shan et al., 2019a). In this study, the biosynthesis and transcriptional regulation of flavonol accumulation during floral development processes and in different tissues of fully opened flowers were further investigated. First, two FLS genes, which might be responsible for the spatio-temporal biosynthesis of floral flavonols were isolated and functionally characterized. Second, it was found that FhFLS1, which mainly contributes to the flavonol synthesis in early flower buds, is regulated by the well-known SG7 MYB proteins. Third, the expression of FhFLS2, a gene responsible for flavonols found in the late developed flowers and mature flower tissues such as petal, stamen and pistil, was targeted directly by non SG7 MYB proteins. In addition, it was also confirmed that AtMYB21 and AtMYB24, the orthologs of Freesia non SG7 MYB proteins in Arabidopsis, activate the expression of AtFLS1 in anther and pollen of mature flowers. Our study provides new insights into the flavonol biosynthesis and regulation in floral organs and tissues, which will pave the way for further clarification of the roles of flower accumulated flavonols in pollinator attraction, pollen germination and fertility.

Materials and Methods

Plant materials and growth conditions

Red River®, a representative cultivar of Freesia hybrida, was cultivated under conditions described elsewhere, and the samples used for metabolite and gene expression analysis were prepared as previously reported (Gao et al., 2018).

The surface-sterilized young inflorescences of Red River® were cut into segments for callus induction following a previously established system (Uwagaki et al., 2015). Subsequently, 3-wk old sub-cultured calli were used in Freesia protoplast isolation, as described elsewhere (Shan et al., 2019b).

Nicotiana tabacum cv K326 and Arabidopsis ecotype Columbia (Col-0) were grown in a greenhouse at 22°C under a 16 h : 8 h, light : dark photoperiod. Healthy tobacco plants aged c. 4 wk were used for transfection with Agrobacterium (Sparkes et al., 2006). Fully bloomed tobacco flowers were collected to evaluate gene expression and flavonoid accumulation profiles. Four-week-old rosette leaves from Arabidopsis were chosen for protoplast isolation.

Measurement of flavonol and anthocyanin contents

The preparation of flavonol and anthocyanin samples was carried out according to procedures detailed in previous studies (Pandey et al., 2015; Shan et al., 2019a). Flavonol derivatives were separated, identified and quantified according to the method described in our own previous study, with minor modifications (Sun et al., 2016). The gradient elution method was changed as follows: 0 min, 14% solvent B; 10 min, 18% solvent B; 35 min, 37% solvent B; 60 min, 46% solvent B; 67 min, 14% solvent B; 70 min, 14% solvent B. Quantitative analysis of flavonols was performed using the external standard curve calibration of quercetin 3-O-glucoside.

For the analysis of the anthocyanin content in tobacco flowers, samples were analysed using ultraviolet spectroscopy and the concentration of anthocyanins was determined as follows: QAnthocyanins = (A530−0.25 × A657) g−1 (fresh weight).

Gene and promoter sequence screening and isolation

AtFLS1 (GenBank no. NP_196481.1) and AtMYB12 (GenBank no. CAB09172) were used as probe baits to screen the orthologs in the well-established Freesia transcriptomic database using the tBlastn algorithm (Li et al., 2019, see Supporting Information Dataset S1). The sequences obtained were subjected to a manual Blastx search of the National Center for Biotechnology Information (NCBI) database, and the best hits were defined as candidate FhFLS and FhMYBF transcripts. Moreover, the established Freesia regulator database was employed to screen for FhFLS2 candidate factors using Pearson’s correlation of reads per kilobase per million mapped reads (RPKM) values at the 1% significance level, as described in our recent study (Yang et al., 2020, Dataset S2). For promoter isolation, the 5′ flanking sequences of Freesia flavonol biosynthetic genes were cloned using a Genome Walking Kit (Takara, Dalian, China) according to the manufacturer’s instructions. The −1000 to −1500 bp upstream of the initiation codon ‘ATG’ were tentatively employed as promoter sequences. For promoter isolation of AtFLS1, specific primers (Table S1) were designed to clone the −1551 bp upstream of the ATG according to the Arabidopsis genomic sequences from Phytozome v.12 (https://phytozome.jgi.doe.gov/pz/portal.html#). All the candidate sequences were cloned into the pGEM-Teasy vector (Promega, Madison, WI, USA) for sequencing confirmation.

Gene expression analysis

SYBR Green (Osaka, Japan) based quantitative reverse transcription polymerase chain reaction (qRT-PCR) assays were carried out using a StepOnePlus Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) in a total reaction volume of 10 µl containing 5 µl 2 × Master Mix (Toyobo, Osaka, Japan), 1 µl cDNA and 0.5 µM of each primer set. The reaction was performed in triplicate under the following conditions: 95°C for 60 s, then 40 cycles at 95°C for 5 s and 60°C for 60 s. Freesia 18S rRNA was used to normalize the Ct values, and relative quantities were calculated using the 2−ΔΔCт formula (Livak & Schmittgen, 2001). All primer sequences are listed in Table S1.

Multiple alignment and phylogenetic analysis

Amino acid sequences were submitted to Clustal Omega (http://www.ebi.ac.uk/Tools/msa/clustalo/) with the default parameters to perform multiple sequence alignment. For phylogenetic analysis, the alignments were further processed by Mega v.6 (Tamura et al., 2013) to generate a neighbor-joining tree with bootstrap analysis (1000 replicates), and the gaps were removed using pairwise deletion.

Construction of plasmids

Vectors were constructed using the Minerva Super Fusion Cloning Kit (US Everbright R Inc., Suzhou, China) following the provided instructions. For vectors used in purification of FhFLS proteins and plant transformation, the open reading frames (ORFs) of FhFLS were seamlessly cloned into the BamHI and SacI digested pET-32a vector and pBI121 binary vector respectively. To obtain the purified proteins used in the electrophoretic mobility shift assay (EMSA), the ORFs of FhMYB21L2, AtMYB21 and AtMYB24 were assembled into PGEX-4T-2 digested with BamHI and XhoI. To construct the plasmids used in the transient protoplast transfection assays, the modified pUC19 vectors containing CaMV (cauliflower mosaic virus) 35S promoted HA (human influenza hemagglutinin), GD (GAL4 DNA binding domain), GFPN (N terminal 174 residues of green fluorescent protein) or GFPC (C terminal 66 residues of green fluorescent protein) tags were linearized by NdeI and AflII (Wang et al., 2015), and the linearized pUC19 was then used to generate pUC19-GFP(C)-FhFhMYBFs, pUC19-GD-FhMYBFs and pUC19-HA-FhMYBFs vectors. In addition, the termination codons of HA-tagged FhMYBFs were substituted with GFP to generate vectors for subcellular localization. To construct the β-glucuronidase (GUS) reporter plasmids, the potential promoters were seamlessly cloned into the PstI and SacI digested DFR3-Fpro: GUS vector, as used in previous studies (Li et al., 2019; Shan et al., 2019a). All other constructs used in the present study have been described previously (Li et al., 2019; Shan et al., 2019a,b).

Purification of FhFLS proteins in Escherichia coli and in vitro enzymatic assays

The prokaryotic proteins were made following earlier studies (X. Meng et al., 2019). In brief, the pET-32a vectors containing FhFLS1 and FhFLS2 were transformed into E. coli BL21 (DE3), and recombinant proteins were induced using 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG) at 16°C for 18 h. After induction, the cells were harvested by centrifugation and resuspended in phosphate-buffered saline (PBS, pH 7.4) following cell lysis by sonication. Before enzymatic activity assays, the recombinant FhFLS proteins were purified using Ni Sepharose columns (GE Healthcare, Buckinghamshire, UK), and then confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis.

The enzymatic assays were performed according to the method described by Xu et al., (2012), with some modifications. Briefly, 50 μg of recombinant protein, 2 µl of 100 mM DTT, 2 µl of 100 mM ferrous sulfate containing 10 mM ascorbic acid, 2 µl of 100 mM 2-oxoglutaric acid, 2 µl of 10 mg ml−1 BSA and 5 µl of 1 mg ml−1 substrate (dihydrokaempferol (DHK), dihydroquercetin (DHQ) and dihydromyricetin (DHM) from Sigma) were mixed in 100 mM Tricine (Sangon Biotech Co. Ltd, Shanghai, China) buffer (pH 7.5) to a final volume of 200 µl. The mixtures were incubated at 30°C for 15 min, and an equal volume of ethyl acetate was then added to the reactions with vigorous mixing and brief centrifugation. The supernatant was transferred into new tubes and evaporated in a SpeedVac (Eyela, Tokyo, Japan). 30 µl of 80% methanol was used to resolve the residues, which were analyzed by high performance liquid chromatography (HPLC) with the aforementioned C18 column. The column was eluted with an isocratic mobile phase consisting of 4% formic acid/100% methanol (40/60, v/v) at 1 ml min−1. Specific wavelengths were monitored simultaneously at 288 nm for dihydroflavonols and 360 nm for flavonols.

Plant transformation

Agrobacterium mediated tobacco transformation was conducted according to the procedures described in Sparkes et al. (2006). In brief, the Agrobacterium (strain GV3101) transformant was inoculated in 5 ml lysogeny broth (LB) media and grown overnight at 30°C, shaking at 170 rpm. Then, the cells from c. 1.5 ml of the culture were harvested by centrifugation at 1000 g for 10 min at room temperature, then resuspended in infiltration media to a final OD600 of 0.8. After 4 d, infiltrated leaves were sterilized and cut into pieces before being placed onto Petri plates of shooting medium with 50 µg ml−1 kanamycin for shooting induction. The shoots were removed from the tissue and transferred to plates of rooting medium. The rooted plantlets were transferred to soil for further analysis.

Transient protoplast expression assay

The transient protoplast expression assays were performed as described earlier (Zhou et al., 2014; Shan et al., 2019b). All the plasmids were extracted and purified by GoldHi EndoFree Plasmid Maxi Kit (CWBIO, Beijing, China). Arabidopsis leaves or Freesia calli were harvested for lysing with Cellulase R10 and Macerozyme R10 (both from Yakult Pharmaceutical Ind. Co. Ltd, Tokyo, Japan). Plasmids (10 µg for each plasmid) were introduced into protoplasts by PEG3350 and incubated for 20–22 h at 22°C in the dark. Afterwards, the protoplasts were collected by centrifugation at 120 g for 3 min for further fluorescence observation, GUS activity detection or transcript analysis. All transfection assays were performed in triplicate, and each experiment was carried out with at least three biological replicates.

Chromatin immunoprecipitation ChIP assay

The ChIP assay was performed using the EpiQuik™ Plant ChIP kit (Epigentek, Brooklyn, NY, USA) according to the manufacturer’s instructions. In brief, the transfected protoplasts were linked using 1% formaldehyde and then sheared by sonication to reduce the average DNA fragment size to around 500 bp. The sonicated chromatin complex was further trapped by anti-HA antibodies (Cell Signaling Technology, Boston, MA, USA) in the microwell. A normal mouse IgG provided in the kit was included as negative control. After reverse cross-linking, the DNA samples were analyzed by qRT-PCR with specific primers (Table S1) targeting different cis-elements.

Electrophoretic mobility shift assay (EMSA)

GST tagged FhMYB21L2, AtMYB21 and AtMYB24 and the peptide GST were induced in E. coli BL21 (DE3) following the method outlined in the section ‘Purification of FhFLS proteins in ‘Escherichia coli and in vitro enzymatic assays’, above. The recombinant proteins were purified using immobilized glutathione beads (Sangon Biotech Co. Ltd, Shanghai, China) and incubated with biotin labeled probes in the binding buffer supplied in the EMSA kit (Beyotime, Shanghai, China). The mixtures were further processed using 6.6% nondenatured polyacrylamide gel and analyzed on a positively charged nylon membrane. Details regarding the procedures and detection process can be found in the manual provided by the manufacturer (Beyotime).

Data availability statement

The sequence data of genes in this article have been deposited into Genbank under the following nos.: FhMYB21L1 (MT741683), FhMYB21L2 (MT741684), FhFLS1 (MT741686), FhFLS2 (MT741687), FhMYBF1 (MT741688), FhMYBF2 (MT741689), FhMYBF3 (MT741690), FhMYBF4 (MT741691).

Results

Expression profiles of two FhFLS genes correlate with the flavonol spatio-temporal accumulations in Freesia flowers

To determine the flavonol spatio-temporal accumulation patterns in Freesia hybrida, represented in this study by Red River®, the flower development process was divided into five developmental stages (Fig. 1a), and flowers at stage 5 were separated into floral tissues (Fig. 1b). Two groups of flavonol derivatives (i.e. kaempferol derivatives and quercetin derivatives) were detected in the samples (Fig. S1). The concentrations of flavonol derivatives plunged, whereas the total flavonols, especially kaempferol derivatives, had an overall upward trend as flower development progressed (Fig. 1c). In the fully opened flowers, kaempferol derivatives were the predominant flavonol depositions in petal, stamen, and pistil, whereas quercetin derivatives mainly accumulated in petal and torus (Fig. 1d). In comparison with other floral tissues, total flavonols were most highly accumulated in petals.

Details are in the caption following the image
Flavonol accumulation and its biosynthetic gene expression in flowers of Freesia hybrida cv Red River®. (a, b) Freesia flowers at five developmental stages, and different tissues stripped from fully blooming flowers. S1 (stage 1), <10 mm long with unpigmented buds; Stage 2, 10–20 mm long with slightly pigmented buds. Stage 3, 20–30 mm long with pigmented buds. Stage 4, fully pigmented flowers before complete opening. Stage 5, fully opened flowers. Ca, calyx; Pe, petal; Pi, pistil. St, stamen; To, torus. (The torus is enclosed by the calyx in Freesia flowers, and the petals, pistil and stamens are attached to the torus.) Bars, 2 cm. (c, d) Flavonol contents at different flower development stages and in different flower tissues. The pillars and broken lines indicate the total mass and concentrations in fresh materials, respectively. FW, fresh weight. (e, f) Expression patterns of flavonol biosynthetic genes at different flower development stages and in different flower tissues. The data represent the mean ± SD of three biological replicates. The gene expression data are calculated as log2 and are hierarchically clustered based on average Pearson’s distance metric. Red and green boxes indicate high and low expression levels, respectively.

To unravel the molecular basis of flavonol spatio-temporal accumulation patterns, flavonol biosynthetic genes were screened. In total, 10 candidate genes were isolated (Figs S2, S3) and their expression profiles were examined (Fig. 1e, f). FhCHS1, FhCHI2 and Fh3GT1 expression levels were higher in almost all flower developmental stages and floral tissues compared to their paralogs, indicating that they might play more important roles in floral flavonol or anthocyanin biosynthesis. By contrast, FhCHS6 was more highly expressed than FhCHS1 in stage 1, probably indicating that it has specific roles in flavonol biosynthesis during the early developmental stages, when flowers are unpigmented. In general, Fh3GT2 might also play crucial roles in the early developmental stage as it exhibited high catalytic efficiency towards kaempferol aglycone, compared with Fh3GT1 (X. Meng et al., 2019). FhCHI2 and F3H, however, were consistently expressed in both early and late developmental stages and flower tissues. Interestingly, two FhFLS genes showed divergent expression profiles: FhFLS1 had an extremely high expression level compared with FhFLS2 in stage 1 and then decreased, whereas FhFLS2 expression levels increased as flower development progressed and were higher than FhFLS1 from stage 2 to stage 5. In addition, the divergent expression profiles of FhFLS were also observed in flower tissues. FhFLS1 appeared to be highly expressed in toruses, while FhFLS2 was predominantly expressed in petals, pistils and stamens. In conclusion, the biosynthesis of flavonol in developing Freesia flowers and tissues is strictly regulated by the structural genes, especially FhFLS genes.

FhFLS1 and FhFLS2 catalyze the formation of flavonols in vitro and in planta

The full-length sequences of the two FhFLS genes were amplified from flowers of Red River® using specific primers (Tables S1, S2). Sequence alignment analysis (Fig. S3A) revealed conserved HxDxnH motifs (His217, 213, Asp219, 215, and His273, 269) for binding ferrous iron, RxS motifs (Arg283, 279 and Ser285, 281) participating in the 2-oxoglutarate binding, active sites for binding ascorbate or substrate, and other residues probably conferring flexibility for proper folding in plant 2-Oxoglutarate-Dependent Dioxygenases (2-ODD proteins) (Wellmann et al., 2002; Xu et al., 2012; Park et al., 2017). Moreover, both FhFLS proteins possessed the conserved ‘PxxxIRxxxEQP’ and ‘SxxTxLVP’ motifs (Fig. S3A), which were reported to be FLS-specific characteristics distinct from other plant 2-ODDs, such as ANS, F3H or FNS (Stracke et al., 2009; Park et al., 2017). Further phylogenetic analysis re-confirmed that FhFLS1 and FhFLS2 might function as candidate Freesia flavonol synthases (Fig. S3B).

To verify whether FhFLS1 and FhFLS2 harbored FLS activities, purified recombinant proteins were prepared (Fig. S4) for in vitro biochemical assays. As shown in Fig. 2(a–c), both recombinant FLS proteins were functional in converting DHK to kaempferol or DHQ to quercetin. Moreover, it seemed that the two FhFLSs catalyzed DHK more efficiently than DHQ, and FhFLS2 had comparatively higher catalytic efficiency under the same conditions. Interestingly, FhFLS2 appeared to produce myricetin using DHM as substrate, albeit no myricetin was detected in Freesia flowers. To further investigate the potential roles of FhFLS isozymes in planta, both FhFLS genes were overexpressed in tobacco plants. As shown in Fig. 2(d) and (e), transgenic tobacco flowers displayed pure white or pale pink color alterations. Consistent with the phenotypic changes, significantly reduced anthocyanin concentrations and increased flavonol concentrations were detected in transgenic tobaccos (Fig. 2f). In parallel with the biochemical analyses, we also found that FhFLS preferred DHK as substrate, as there seemed to be a more pronounced increase in kaempferol than quercetin in transgenic tobaccos (Fig. 2f). Collectively, these results indicate that two FhFLS genes encode bona fide FLS enzymes responsible for flavonol biosynthesis in Freesia flowers.

Details are in the caption following the image
Freesia FhFLS1 and FhFLS2 are functional in flavonol biosynthesis. (a–c) The enzyme activities of FhFLS1 and FhFLS2 on DHK, DHQ and DHM, respectively. DHK, dihydrokaempferol. K, kaempferol. DHQ, dihydroquercetin. Q, quercetin. DHM, dihydromyricetin. M, myricetin. (d) Phenotypic changes in transgenic tobacco flowers overexpressing FhFLS1 or FhFLS2. (e) Expression analysis of FhFLS1 or FhFLS2 by reverse transcription polymerase chain reaction (RT-PCR) in the wild-type and transgenic tobaccos. (f) Flavonol derivatives and total anthocyanins in different tobacco flowers. EV, plant transformed with empty vector; FW, fresh weight; OX, plant overexpressing FhFLSs; WT, wild-type plant. Several flowers were randomly sampled at the same time from the same line to form one replicate. Data represent the mean ± SD of three biological replicates. Student's t-test was used to analyze the level of significance (*, P < 0.05; **, P < 0.01).

SG7 group FhMYBFs positively influence flavonol biosynthesis in Freesia flowers

As previously reported, the most extensively characterized positive regulators involved in plant flavonol biosynthesis are members of SG7 MYB families. In F. hybrida, four putative SG7 MYB regulatory genes named FhMYBF1, FhMYBF2, FhMYBF3 and FhMYBF4 were obtained from Red River® (Table S2). Sequence alignment analyses revealed that the four SG7 group FhMYBFs were defined by an N-terminal R2R3 MYB domain and two signature motifs, SG7 and SG7–2 (Fig. 3a). Additionally, another signature of SG7 MYB protein – the absence of a [D/E]Lx2[R/K]x3Lx6Lx3R motif in the R3 domain, which is required to interact with bHLH proteins – was also absent in the FhMYBFs (Fig. S5). Phylogenetic analysis revealed that all FhMYBF proteins were categorized into the cluster of flavonol regulators (Fig. S6A), and they were found to localize to the nucleus (Fig. S6B). Furthermore, Gal4-based transient Arabidopsis protoplast assays were conducted to determine the trans-regulation properties of FhMYBFs, and the four FhMYBFs were determined to be R2R3 MYB transcription activators (Fig. 3b).

Details are in the caption following the image
FhMYBF1, FhMYBF2, FhMYBF3 and FhMYBF4 are R2R3 MYB transactivators and function in Freesia flavonol biosynthesis. (a) Protein alignments between FhMYBFs and Arabidopsis flavonol related MYB factors. The conserved R2 and R3 repeat domains are indicated by yellow and green lines, respectively. Subgroup 7 and SG7-2 motifs are outlined by red boxes. Numbers indicate the position of the last amino acid in each line of the proteins. *, identical amino acids, or similar amino acids. (b) Transactivation capacities of Freesia FhMYBFs. Effectors and reporters for transient assays are shown by schematic diagrams. (c) Relative expression levels of flavonol biosynthetic genes promoted by FhMYBFs in protoplasts isolated from Red River®. Data represent the mean ± SD of three biological replicates. Student’s t-test was used to analyze the level of significance (*, P < 0.05; **, P < 0.01).

To confirm which flavonol biosynthetic genes could be upregulated by these four FhMYBF activators, a fast and reliable Freesia protoplast isolation and transient expression system was set up to partially complement the shortcomings of the long life cycle of Freesia and recalcitrant character of monocots against Agrobacterium (Shan et al., 2019b). As shown in Fig. 3(c), almost all the flavonol biosynthetic genes were strongly or slightly activated when FhMYBF genes were overexpressed in Freesia protoplasts, except FhCHI1 and FhFLS2. Specifically, FhCHS1 could be upregulated by FhMYBF2 or FhMYBF1, while FhCHS6, FhCHI2, Fh3GT2 and FhFLS1 could be induced by all four FhMYBFs with different folds. By contrast, none of the FhMYBF proteins could activate anthocyanin- or proanthocyanidin-related genes, such as FhDFR and FhLDOX (Fig. S7). In conclusion, these results strongly suggest that SG7 group FhMYBFs could positively regulate flavonol biosynthesis in Freesia flowers.

SG7 group FhMYBFs activate FhFLS1 and early flavonoid biosynthetic genes by directly binding to promoter regions

Protoplast transient expression assays were carried out to further confirm the interaction between FhMYBFs and the main flavonol biosynthetic gene promoters. Consistent with the results shown in Fig. 3(c), FhMYBFs could activate the promoters of the tested flavonol biosynthetic genes, including FhCHS1, FhCHI2, FhF3H, Fh3GT2 and FhFLS1, whereas no changes were found in FhFLS2 (Fig. S8). This result indicates that the two FhFLS genes might be controlled by different regulators. To further prove that FhFLS1 might be the target of FhMYBFs, the expression correlation between them was calculated. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis showed that the transcripts of FhMYBFs were most abundant at early stages of flower development and decreased throughout the flower development process, consistent with FhFLS1 (Fig. S9a). In addition, FhMYBFs were predominantly expressed in torus followed by calyx, in contrast to petal, pistil and stamen, and a synchronous relationship between FhMYBFs and FhFLS1 was observed in torus (Fig. S9b).

To verify whether FhMYBFs could directly target the promoters of flavonol biosynthetic genes, FhCHI2 and FhFLS1 were selected to represent general flavonoid and flavonol specific pathway genes, respectively. Their promoters were analyzed for the presence of MYB recognition sequence elements (MREs) using the New PLACE database (https://www.dna.affrc.go.jp/PLACE/?action=newplace; Higo et al., 1999) (Fig. 4a). The key MREs in recruiting FhMYBFs were confirmed by differently truncated promoters in transient protoplast transfection assays. As shown in the results depicted in Fig. 4(b) and (c), decreased GUS activity was observed following successive promoter truncations. The sequence between −1240 and −830 bp of the FhFLS1 promoter and that between −247 and −154 bp of FhCHI2 promoter were randomly chosen for further binding analysis of FhMYBFs. Chromatin immunoprecipitation assays revealed that the selected sequences containing MYBCORE and AC-rich elements were significantly enriched (Fig. 4d). Altogether, FhMYBF1, FhMYBF2, FhMYBF3 and FhMYBF4 could regulate Freesia flavonol biosynthesis by direct binding to the promoters of FhFLS1 and other flavonol biosynthetic genes. However, none of the FhMYBFs seemed to regulate FhFLS2.

Details are in the caption following the image
FhMYBF1, FhMYBF2, FhMYBF3 and FhMYBF4 can directly bind to FhFLS1 and FhCHI2 promoters. (a) Potential MYB binding sites in the promoters of FhFLS1 and FhCHI2. The predicted sites are indicated by colored boxes. (b, c) Activation effects of FhMYBFs on differently truncated FhFLS1 and FhCHI2 promoters detected by transient protoplast assays. Data represent the mean ± SD of three biological replicates. Student's t-test was used to analyze the level of significance (*, P < 0.05; **, P < 0.01). (d) The direct binding of FhMYBFs to FhFLS1 and FhCHI2 promoters, assessed by chromatin immunoprecipitation–quantitative polymerase chain reaction (ChIP-qPCR). The potential binding sites are indicated in the left diagram. Freesia linalool related FhTPS1 was used as a negative control. The data represent the mean ± SD of three biological replicates.

FhMYB21L2, a non SG7 group MYB regulator, activates FhFLS2 directly

In order to screen the regulators responsible for the expression of FhFLS2 that could not be activated by the well-known SG7 MYB proteins, 12 flavonoid biosynthesis related MYB regulators expressed in Freesia flowers were tested for their activation effects on FhFLS2. However, none of them could activate FhFLS2 (Fig. S10). Based on the fact that regulatory genes and their targets are usually expressed synchronously with each other (Yang et al., 2020), Pearson's correlation of the fragments per kilobase million (FPKM) values was calculated for FhFLS2 and the potential regulators; 36 candidates were obtained (Fig. S11), which were further co-transfected with FLS2-Fpro: GUS into protoplasts. Three proteins encoded by unigene-49278, unigene-36442 and unigene-82591 were found to be able to significantly activate the FhFLS2 promoter (Fig. S12). Phylogenetic analysis together with the Arabidopsis MYB regulators revealed that unigene-36442 and unigene-49278 fell into the subclade containing AtMYB21 and AtMYB24, and were thus tentatively designated as FhMYB21L1 and FhMYB21L2, respectively (Fig. S13). As unigene-49278 showed the most dramatic activation of FhFLS2, it was subjected to further analysis. As shown in Fig. S14, FhMYB21L2 is a nuclear localized protein and displayed transcriptional activation activity; furthermore, it was confirmed to activate the expression of FhFLS2 rather than FhFLS1. In conclusion, it is reasonable to conclude that FhMYB21L2 might be a specific regulator targeting FhFLS2. A positive correlation was consistently found in the expression patterns of FhMYB21L2 and FhFLS2 in the late flower development stages and in different tissues of mature flowers, such as petal, stamen and pistil (Fig. S15).

Sequence analysis revealed 13 potential MREs in the promoter of FhFLS2 (Fig. 5a). A set of vectors containing FhMYB21L2 and GUS reporter constructs driven by differently truncated promoters of FhFLS2 were co-transfected into Arabidopsis protoplasts. As GUS activities plunged when the promoter was truncated to −154 bp upstream of ‘ATG’ (Fig. 5b), the sequence located between −254 bp and −164 bp upstream of the initiation codon seemed to play critical roles in FhMYB21L2 binding to the FhFLS2 promoter, which was enriched nearly eightfold in the subsequent chromatin immunoprecipitation–quantitative polymerase chain reaction (ChIP-qPCR) analysis (Fig. 5c). In addition, biotin labelled probes were synthesized and GST tagged FhMYB21L2 was purified (Fig. 5d), and they were subsequently subjected to EMSA to validate the binding of FhMYB21L2 to the FhFLS2 promoter. As shown in Fig. 5(e), FhMYB21L2 could directly bind to either the MYBPLANT or MYBCORE element, respectively. Taken together, these results indicate that FhMYB21L2 participates in Freesia flavonol biosynthesis by directly regulating the expression of FhFLS2.

Details are in the caption following the image
FhMYB21L2 could directly bind to the FhFLS2 promoter. (a) Potential MYB binding sites in the FhFLS2 promoter. The predicted sites are indicated with colored boxes. (b) Activation effects of FhMYB21L2 on differently truncated FhFLS2 promoters, detected by transient protoplast assays. Data represent the mean ± SD of three biological replicates. Student's t-test was used to analyze the level of significance (**, P < 0.01). (c) The direct binding of FhMYB21L2 to the FhFLS2 promoter, assessed by chromatin immunoprecipitation–quantitative polymerase chain reaction (ChIP-qPCR). Arabidopsis anthocyanin related AtDFR was used as a negative control. The data represent the mean ± SD of three biological replicates. (d) FhMYB21L2 protein purified from prokaryotic cells, detected by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). (e) The direct binding of FhMYB21L2 to MYBPLANT and MYBCORE elements of the FhFLS2 promoter, assessed by electrophoretic mobility shift assay (EMSA).

AtMYB21 and AtMYB24 directly bind to the promoter of AtFLS and activate its transcription

Previous studies have indicated that the transcription factors responsible for AtFLS1 expression in Arabidopsis anthers have not yet been revealed (Stracke et al., 2010; Battat et al., 2019), and the potential regulatory roles of AtMYB21 and AtMYB24, the FhMYB21L2 homologs, in Arabidopsis anther-specific flavonol biosynthesis require further clarification. The results will help to verify whether the transcriptional regulation of FLS genes by MYB21 in monocot plants is also functional in eudicot plants.

Firstly, AtMYB21 and AtMYB24 together with GUS reporter constructs driven by the AtFLS1 promoter were co-transfected into Arabidopsis protoplasts. The results showed that both AtMYB21 and AtMYB24 exhibited significant activation effects on the AtFLS1 promoter (Fig. 6a). Fourteen MREs, including one MYBCORE, two MYBPZMs, two MYBCOREATCYCB1s, four MYB1ATs, two MYBST1s and three MYBGAHVs, were identified in the AtFLS1 promoter (Fig. 6b). Subsequently, differently truncated AtFLS1 promoters were used primarily to screen for the key MREs. Consequently, the MYB1AT and MYBPZM cis-elements located between −561 bp and −324 bp of the AtFLS1 promoter were supposed to be of significance for the binding of AtMYB21 to proAtFLS1, whereas the MYBCORE cis-element located between −755 bp and −561 bp of the AtFLS1 promoter was assumed to be important for AtMYB24 (Fig. S16). To further confirm this assumption, EMSA was performed using AtMYB24 and AtMYB21 recombinant proteins (Fig. S17). As shown in Fig. 6(c) and (d), the MYBCORE (TAACTG) and MYBPZM (CCAACC) were experimentally confirmed to play pivotal roles in the binding of AtMYB24 and AtMYB21 to the AtFLS1 promoter, respectively. This evidence leads us to conclude that AtMYB21 and AtMYB24 are candidate regulators in Arabidopsis anther and pollen flavonol biosynthesis.

Details are in the caption following the image
AtMYB21 and AtMYB24 could directly bind to the AtFLS1 promoter. (a) Transactivation activities of AtMYB21 and AtMYB24 on the AtFLS1 promoter, detected by transient Arabidopsis protoplast assay. Data represent the mean ± SD of three biological replicates. Student's t-test was used to analyze the level of significance (**, P < 0.01). (b) Potential MYB binding sites in the AtFLS1 promoter. Data represent the mean ± SD of three biological replicates. The predicted sites are indicated by colored boxes. (c, d) The direct binding of AtMYB24 and AtMYB21 to MYBCORE, MYB1AT and MYBPZM elements of the AtFLS1 promoter, assessed by electrophoretic mobility shift assay (EMSA).

Discussion

The spatio-temporal biosynthesis of flower accumulated flavonols is controlled by single or multiple FLS genes

FLS genes have been widely isolated and functionally characterized as indispensable in plant flavonol biosynthesis. However, the discrepancies between FLS expression and flavonol biosynthesis described in the Introduction section suggest that the function of FLS in flavonol biosynthesis might be more complicated than was previously thought. Firstly, flavonol synthase encoded by a single FLS gene might be able to sustainably synthesize the flavonols in the late developmental stages when the FLS gene is not highly expressed. This assumption awaits further confirmation through detailed and robust evidence, however. Secondly, the spatio-temporal biosynthesis of floral flavonols are controlled by a single functional FLS protein in some plants, whereas the expression of the single FLS gene might be controlled by diverse regulators in different floral tissues (Stracke et al., 2010). Thirdly, there are probably additional new copies of FLS genes that are functional in the late stages of development or in certain tissues in some plants. As demonstrated in the present study, two FLS genes, FhFLS1 and FhFLS2, with different expression patterns were isolated and verified to function independently in Freesia flavonol biosynthesis. Therefore, it is reasonable to conclude that the transcriptional regulations of FhFLS1 and FhFLS2 differentiated after gene duplication, though both of them are functionally the same (flavonol synthase). However, whether the spatio-temporal biosynthesis of floral flavonols can be explained by multiple FLS genes in other plants needs to be further clarified. Moreover, members of the 2-ODD protein family such as FLS, LDOX/ANS and F3H might be interchangeable or complementary to some extent. For instance, AtLDOX/ANS has also been reported to participate in flavonol production in planta (Preuss et al., 2009), while AtFLS and AtLDOX/ANS could partially complement AtF3H function in Arabidopsis (Owens et al., 2008a). In conclusion, flavonols accumulated in the early and late flower development stages might be differently controlled through regulation of the expression of single or multiple FLS genes.

The spatio-temporal expression of FLS genes is controlled by different clades of R2R3 MYB regulators in flowers

As has been widely investigated, the SG7 group of MYB regulators have been shown to specifically control FLS expression in numerous plant species (Liu et al., 2015, 2016; Allan & Espley, 2018). In this study, four SG7 R2R3 MYB regulators, named FhMYBF1, FhMYBF2, FhMYBF3 and FhMYBF4, were isolated and confirmed to activate the expression of FhFLS1 (Figs 3, 4). The four FhMYBFs are likely to be separate genes rather than alleles, as the highest identity of the translated amino acid sequences is no more than 60%, which is rather low for alleles considering the domestication of modern F. hybrida in less than 200 yr (Manning and Goldblatt, 2010). Nevertheless, their differential expression patterns imply that the four FhMYBFs might be functionally divergent in the biosynthesis of Freesia floral flavonols (Fig. S9), among which FhMYBF1 might act as the predominant SG7 MYB regulator, considering its abundant expression levels. Functionally divergent SG7 MYB regulators have also been found in other plants. In developing Arabidopsis seedlings, MYB12 mainly controls flavonol biosynthesis in the root, while MYB111 is primarily functional in cotyledons (Stracke et al., 2007). The divergence between duplicated MYB regulators seems to be common in plants, as it was also found in regulators responsible for anthocyanin biosynthesis (Albert et al., 2014). In addition to the expression patterns, the functional divergence of the duplicated MYB regulators might also be reflected by their transactivation capacities towards the flavonol biosynthetic genes. In gentian flowers, GtMYBP3 can significantly activate the expression of CHS and F3′H, which was not targeted, or was weakly targeted, by GtMYBP4 (Nakatsuka et al., 2012). Likewise, FhMYBF1 and FhMYBF2 could activate the expression of FhCHS1, whereas FhMYBF3 and FhMYBF4 were incapable. All four FhMYBFs have the ability to up-regulate the expression of FhCHI2 and FhFLS1, whereas the transactivation capacities of FhMYBF1 and FhMYBF3 are higher than the other two FhMYBFs on FhFLS1 (Figs 3, 4). The transactivation discrepancy on target genes has also been characterized among MYB orthologs responsible for anthocyanin biosynthesis (Li et al., 2020).

Although FhMYBF1 was expressed in mature flowers and tissues, its activation effects on FhFLS2 were not detected. The independent transcriptional regulation of FLS genes in mature flowers was also found in Arabidopsis and maize (Neuffer et al., 1997; Stracke et al., 2010). Gene expression correlation analysis is effective in the screening of regulatory genes responsible for certain target genes (Zhao et al., 2014; Zheng et al., 2015; Yang et al., 2020). Based on this method, an R2R3 MYB activator, FhMYB21L2, was isolated and confirmed to act as a potential regulator targeting for FhFLS2, which could directly bind to the MYBCORE and MYBPLANT (AC-rich) cis-elements in the promoter of FhFLS2 and significantly activate its expression (Fig. 5). The MYBCORE element (C/T)NGTT(G/A) and AC-rich element ACC(A/T)(A/C/T)(A/C/T) have previously been shown to be recognized by other plant R2R3 MYB-family proteins (Gomez-Maldonado et al., 2004; Xu et al., 2015; Gao et al., 2016; Zhu et al., 2017; Y. Meng et al., 2019). However, how the hundreds of MYB regulators define their specificities in specific development events or metabolite biosynthesis processes remains poorly understood. Recently, AtMYB21 and AtMYB24, the homologs of FhMYB21L2 in Arabidopsis, were associated with exclusive production of flavonols and HCAAs for coating pollen, together with AtMYB99 as an MYB triad (Battat et al., 2019). However, whether AtMYB21 and AtMYB24 were able to target AtFLS1 has not been verified. In the present study, both AtMYB21 and AtMYB24 were found to directly bind to MYBCORE or AC-rich elements in the AtFLS1 promoter, respectively, and then activate its expression. These results indicate that the regulation of MYB21 on FLS genes expressed in late flower development stages and certain floral tissues in mature flowers might be conserved in these two evolutionarily divergent angiosperm plants. Interestingly, Arabidopsis AtFLS1 could be regulated by both SG7 member AtMYB11/12/11 and non-SG7-member AtMYB21, whereas the corresponding MYB orthologs FhMYBFs and FhMYB21 tend to specifically regulate FhFLS1 and FhFLS2, respectively. More experimental evidence, especially in other representative plants, is therefore required to reveal the conservation or diversity of flavonol biosynthesis regulation during evolution.

MYB21 homologs act as pleiotropic regulators in flower development and specialized metabolite biosynthesis

As pleiotropic regulators, MYB21 homologous genes have been widely investigated in multiple plant species. In Arabidopsis, AtMYB21 and AtMYB24 were reported to regulate gibberellin and jasmonate-mediated stamen development, and myb21myb24 double mutants exhibited defects specifically in pollen maturation, anther dehiscence, filament elongation, gynoecium and petal development, which could lead to male sterility (Mandaokar et al., 2006; Cheng et al., 2009; Song et al., 2011; Reeves et al., 2012). Afterwards, the two Arabidopsis MYBs were found to cooperatively regulate seed production together with IIIe-bHLH proteins (Qi et al., 2015). Moreover, in the myb21myb24 double mutant flowers, the release of volatile sequiterpenes was decreased, and this was accompanied by the downregulation of terpene synthase genes AtTPS11 and AtTPS21, indicating the involvement of these MYB regulators in the biosynthesis of volatile compounds. Recently, we verified that the MYB-bHLH complex consisted of MYB21 and IIIe-bHLH proteins, and that this complex could also cooperatively regulate the expression of TPS genes (Yang et al., 2020). In accordance with the MYB triad in Arabidopsis (Battat et al., 2019), a similar MYB triad containing PhEOBI, PhEOBII and PhODO1 was also shown to coregulate genes responsible for the flower volatile compounds in Petunia hybrida (Van Moerkercke et al., 2011; Spitzer-Rimon et al., 2012). In the current study, we have broadened the roles of MYB21 homologs in flavonoid biosynthesis, which has long been considered to be one of two apparently well diversified pathways, the other being terpenoid biosynthesis. The flavonoid and terpenoid pathways are synthesized from carbon skeletons generated in primary metabolism and are thought to operate independently. However, several studies have begun to uncover metabolic and regulatory connections between these two major specialized metabolites. For example, MYB factors have been shown to function in coordinating metabolic activity between the flavonoid and terpenoid biosynthetic pathways (Bendon et al., 2010; Ben-Zvi et al., 2012; Kang et al., 2014). SlCHI1 is not only required for flavonoid synthesis but also plays a role in terpenoid accumulation in glandular trichomes (Kang et al., 2014). However, the precise nature of a direct bridge between the flavonoid and terpenoid pathways is still unknown. In the present study, MYB21 was found to regulate flavonol biosynthesis through targeting the FLS gene. It might therefore be reasonable to conclude that there exists a regulatory network mediated by MYB21 to regulate flower development as well as specialized metabolite biosynthesis; flavonol and volatile components accumulated or released from matured flowers might attract insect pollinators and/or repel pathogens and predators in order to benefit the fertility of the plants (Lucas-Barbosa et al., 2016). The findings from this study provide new insights into flavonol and perhaps even specialized metabolite biosynthesis and regulation in flower organs and tissues in general, paving the way for further investigation into how flower development and metabolite biosynthesis are coordinated in flowering plants.

Conclusions

A hypothetical model of the regulation of flavonol biosynthesis in Freesia flowers has been proposed here (Fig. 7). In the early developmental stages, FhMYBFs are highly expressed and account for the expression of FhFLS1 as well as other flavonol biosynthetic genes, giving rise to abundant flavonol accumulation in these stages. As flower development progresses, the transcripts of FhMYBFs and FhFLS1 are decreased, and FhFLS2 is activated by FhMYB21 to maintain flavonol biosynthesis in mature flowers. When flowers are fully opened, FhMYBFs are mainly in charge of flavonol biosynthesis in torus and calyx, while FhMYB21 plays a predominant role in flavonol biosynthesis in petal, stamen and pistil. In addition, we also found that the regulatory roles of MYB21 proteins might be conserved at least in F. hybrida and Arabidopsis, two evolutionarily divergent plant species.

Details are in the caption following the image
The hypothetical regulation model of flavonol biosynthesis during Freesia flower development. Stages 1 to 5 (S1–S5) indicate the five flower development stages. Ca, calyx; Pe, petal; Pi, pistil; St, stamen; To, torus. In early flower buds, the contribution of FhFLS1 to flavonol synthesis is regulated by FhMYBFs. As flower development progresses, the transcripts of FhFLS1 are decreased, while the expression of FhFLS2 is activated by FhMYB21 to maintain the accumulation of flavonols in mature flowers. At S5, FhMYBFs are mainly involved in flavonol biosynthesis in the calyx and torus, while FhMYB21 plays pivotal roles in petal, stamen and pistil flavonol biosynthesis.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (31972445, 31900252), the China Postdoctoral Science Foundation funded project (2018M641761), and the Department of Science and Technology of Jilin Province (20190201299JC, 20190303095SF). In addition, we would like to thank Dr Shadrack Kimani for revising this manuscript. We declare that this research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

    Author contributions

    XS, YL, SY, ZY, MQ, RG, TH and XM performed the experiments and ZX helped analyze data. XG and XS wrote and revised this manuscript, XG designed the experiments and discussed them with LW. All authors participated in this research and approved the final manuscript. XS and YL contributed equally to this work.