TcMYC2 regulates Pyrethrin biosynthesis in Tanacetum cinerariifolium

Abstract Pyrethrins constitute a class of terpene derivatives with high insecticidal activity and are mainly synthesized in the capitula of the horticulturally important plant, Tanacetum cinerariifolium. Treatment of T. cinerariifolium with methyl jasmonate (MeJA) in the field induces pyrethrin biosynthesis, but the mechanism linking MeJA with pyrethrin biosynthesis remains unclear. In this study, we explored the transcription factors involved in regulating MeJA-induced pyrethrin biosynthesis. A single spray application of MeJA to T. cinerariifolium leaves rapidly upregulated the expression of most known pyrethrin biosynthesis genes and subsequently increased the total pyrethrin content in the leaf. A continuous 2-week MeJA treatment resulted in enhanced pyrethrin content and increased trichome density. TcMYC2, a key gene in jasmonate signaling, was screened at the transcriptome after MeJA treatment. TcMYC2 positively regulated expression of the pyrethrin biosynthesis genes TcCHS, TcAOC, and TcGLIP by directly binding to E-box/G-box motifs in the promoters. The stable overexpression of TcMYC2 in T. cinerariifolium hairy roots significantly increased the expression of TcAOC and TcGLIP. Further transient overexpression and viral-induced gene-silencing experiments demonstrated that TcMYC2 positively promoted pyrethrin biosynthesis. Collectively, the results reveal a novel molecular mechanism for MeJA-induced pyrethrin biosynthesis in T. cinerariifolium involving TcMYC2.


Introduction
Tanacetum cinerariifolium is an economically important ornamental plant valued as a natural source of pyrethrins, a group of highly effective insecticidal secondary metabolites [1]. Natural pyrethrins have been widely used as an environmentally friendly insecticide, with low mammalian toxicity and no known insect immunity [2]. Consequently, T. cinerariifolium is cultivated in several countries, including Kenya, Tanzania, Uganda, Rwanda, China, and Australia [3], as a commercially important source of natural pesticides.
Natural pyrethrins comprise six types of monoterpene esters. The esterification involves an acid moiety (pyrethric acid and chrysanthemic acid) and an alcohol moiety (pyrethrolone, cinerolone, and jasomolone). The monoterpene acid moiety is derived from the methylerythritol-4-phosphate pathway and is exclusive to T. cinerariifolium and closely related species [4]. The first committed step is catalyzed by chrysanthemol synthase (CHS) [5,6], which catalyzes the condensation of two dimethylallyl diphosphate (DMAPP) molecules to produce transchrysanthemol. Subsequently, trans-chrysanthemol is oxidized by alcohol dehydrogenase (ADH) and aldehyde dehydrogenase (ALDH) to produce trans-chrysanthemic acid [7]. Simultaneously, trans-chrysanthemal or trans-chrysanthemic acid is converted production. Methyl jasmonate (MeJA) regulates the biosynthesis of certain secondary metabolites in plants via transcription factors (TFs), such as phenolic acid and tanshinone in Salvia miltiorrhiza [17], artemisinin in Artemisia annua [18,19], and flavor volatiles in Chrysanthemum indicum [20]. Typically, TFs positively or negatively regulate metabolite contents by directly binding to a downstream secondary metabolite biosynthesis target gene [21,22]. MYC family proteins are master regulators of the JA signaling pathway [23]. Many MYC TFs respond to the MeJA signal and play critical roles in regulating terpenoid biosynthesis in diverse plant species [24,25]. For instance, the JA-responsive TF AsMYC2 positively regulates agarwood sesquiterpene synthesis by binding to the sesquiterpene synthase 1 (ASS1) promoter [26]. AaMYC2 directly binds to the G-box motif in the promoters of CYP71AV1 and DBR2 to promote JA-mediated artemisinin biosynthesis [24]. In Taxus chinensis, TcMYC2a regulates taxol biosynthesis either directly or via ERF TFs depending on JA signaling [27].
In T. cinerariifolium, MeJA treatment and additional stimuli, such as mechanical wounding and volatile organic compounds, induce pyrethrin synthesis [16,28,29]. However, no TFs involved in the MeJA-mediated pyrethrin biosynthesis regulatory network have been reported in T. cinerariifolium. In this study, we found that short MeJA treatment applied to T. cinerariifolium leaves transiently induced the expression of pyrethrin biosynthetic genes and raised the pyrethrin content, and a continuous MeJA treatment even increased the trichome density, which further led to increasing pyrethrins content. Based on transcriptome analysis, we discovered a MeJA responsive TcMYC2 TF which could directly bind to the promoter regions of key pyrethrins biosynthetic genes TcCHS, TcAOC, and TcGLIP. Further transient overexpression and virus-induced gene silencing (VIGS) of TcMYC2 in T. cinerariifolium leaves confirmed that TcMYC2 was a positive factor involved in regulating pyrethrin biosynthesis.

Exogenous MeJA upregulates pyrethrin biosynthesis gene expression
Pyrethrins are esters of an acid moiety and an alcohol moiety derived from two pathways: an irregular monoterpene biosynthesis pathway (specific to T. cinerariifolium) and the JA biosynthesis pathway (Fig. 1a). To investigate the mechanism of MeJAinduced pyrethrin biosynthesis, the transcriptome of T. cinerariifolium leaves was analysed at 0 (control), 2, 4, 6, 8, 10, 12, and 24 h after a single spray application of MeJA, and differentially expressed genes (DEGs) associated with pyrethrin synthesis were identified. Transcriptome assembly and analysis details are presented in Figs S1 and S2, see online supplementary material. Expression levels of TcLOX (KC441523.1), TcAOS (MH397471.1), and TcAOC (MH397472.1), which function upstream of JA, rapidly increased at 2 h and were maintained at a higher level compared with the control (Fig. 1b). The monoterpene pathway genes TcCHS (JX913536.1), TcADH (MF497444.1), TcALDH (MF497445.1), TcCHH (KC441525.1), and TcMT (MK139710.1) were strongly upregulated within 2 h but thereafter were dramatically downregulated to levels much lower than those of the control (Fig. 1b). We speculated that although the short MeJA treatment could transiently and sharply induce the pyrethrins biosynthesis genes, the expression of these genes, especially which involved the synthesis of the acid moiety unique to T. cinerariifolium, was finally downregulated. Moreover, the expression of genes involved in the final steps of pyrethrin biosynthesis, including TcJMH (MG189934.1), TcPYS (MG874680.1), and TcGLIP (JN418994.1), was decreased in response to MeJA treatment, and we speculate this may be due to feedback regulation to maintain cell homeostasis (Fig. 1b). The expression patterns of these genes was confirmed by real-time quantitative reverse-transcription PCR (qRT-PCR) (Fig. 1b).

Exogenous MeJA increases pyrethrin content in leaves
Methyl jasmonate stimulates overaccumulation of many secondary metabolites in plants [11]. Therefore, gas chromatography-mass spectrometry (GC-MS) analysis was conducted to detect pyrethrin accumulation in leaves of MeJA-treated plants (Fig. 1c). Contents of pyrethrin compounds, especially pyrethrin I and pyrethrin II, were significantly increased at 6 h after MeJA treatment, consistent with the timing of pyrethrin biosynthesis gene expression. At subsequent time points, the pyrethrin contents declined to levels lower than those at 0 h consistent with the gene expression profiles (Fig. 1d). Hence, we concluded that MeJA could promote pyrethrins production in response to a transient upregulation of pyrethrins biosynthetic genes; however, it did not maintain the higher pyrethrins content over time by a transient MeJA induction.

TcMYC2 is a potential regulator of MeJA-induced pyrethrin biosynthesis
To examine the role of JA pathway genes in pyrethrin biosynthesis, based on the FPKM expression matrix of 28 552 DEGs, a weighted gene coexpression network analysis (WGCNA) module was constructed. The 33 modules were obtained with a shear value of 0.2 (Fig. S3). The Cyan module (2320 genes) and Skyblue1 (211 genes) module were strongly correlated with pyrethrin synthesis. KEGG and GO enrichment analysis confirmed the genes were involved in secondary metabolite biosynthesis (Fig. S4). To explore the key regulatory genes, a candidate hub gene network was visualized with Cytoscape. Among the hub genes showing high connectivity, TcMYC2 was detected (Fig. S4).
To identify the TFs activated by MeJA that potentially regulated pyrethrin biosynthesis, the promoter sequences of TcCHS, TcALDH, TcAOC, and TcGLIP were cloned and putative TF-binding sites in the promoter sequences was analysed using PlantCARE and PlantPAN3.0 (Fig. 2a). MeJA-responsive elements were detected in the TcALDH, TcAOC, and TcGLIP promoters. In addition, E-box and G-box motifs, the binding sites of bHLH TFs, were identified in the promoter of each gene. Fifty-five bHLH TFs with FPKM >1 were identified with PlantTFDB (Fig. S3). To obtain candidate MeJA-regulated bHLHs regulating pyrethrin biosynthesis, a heatmap was constructed using the RNA-seq data for pyrethrin biosynthesis genes obtained from the MeJA treatment and different development stages (Fig. 2b). Based on Pearson distance measurement and Ward clustering algorithm, the candidate unigene66762 showed the most similar expression pattern to TcCHS or TcAOC. Phylogenetic analysis of bHLHs showed that unigene66762 belonged to the MYC2-like subclade (Figs S5 and S6, see online supplementary material) and contained a conserved MYC domain (Fig. 2c). Thus, we named unigene66762 as TcMYC2 and selected it for further analysis.

Expression profiles and subcellular localization of TcMYC2
To validate MeJA-mediated induction of TcMYC2 expression, qRT-PCR was used to analyse TcMYC2 expression. The TcMYC2 transcript abundance strongly increased at 2 h, and was maintained at a high level after prolonged MeJA exposure (4, 6, and 8 h), until returning to the original level at 24 h (Fig. 3a). These . Green bars are the qRT-PCR results normalized with the 2 − Ct method using GADPH as an internal reference. The experiment was performed three times with three biological replicates. Error bars represent mean ± SD. c GC-MS chromatograms of leaves treated for 0, 6, 12, and 24 h. The relative pyrethrin content was determined using tetradecane as an internal standard, with three biological replicates. d Relative contents of total pyrethrins. The y-axis represents peak area intensity. Error bars represent the SD of the mean of three replicates. Different lower-case letters indicate a significant difference (P < 0.05) from one-way ANOVA followed by a post-hoc Tukey test. results were consistent with the expression pattern of TcAOC in MeJA-treated leaves (Fig. 1b). To investigate potential similarities in expression patterns between TcMYC2 and other pyrethrin biosynthesis genes, the relative expression levels of TcMYC2 were analysed in different tissues and flowers at different developmental stages. The TcMYC2 expression level was lowest in leaves, low in the receptacle and pedicel, and highest in ray florets and disc florets (Fig. 3b). The expression profiles of TcMYC2 at different stages of flowering ( Fig. 3c) were similar to those of other pyrethrin biosynthesis genes, including TcCHS, TcALDH, and TcGLIP, which are highly expressed at the first two stages of flowering. These results suggest that TcMYC2 is involved in the regulation of pyrethrins biosynthesis. Thus, we speculated that TcMYC2 co-regulates pyrethrin production by interacting with pyrethrins biosynthetic gene promoters.
To determine the subcellular localization of TcMYC2, the coding sequence of TcMYC2 was fused in-frame to the green fluorescent protein (GFP) under the control of the Cauliflower mosaic virus (CaMV) 35S promoter. When TcMYC2-GFP was transiently expressed in Nicotiana benthamiana leaves, strong fluorescence was localized to the nucleus. The GFP signal co-localized with the RFP nuclear localization signal (NLS; Fig. 3d). These results indicate that TcMYC2 is a nuclear-localized protein according to its potential role as a TF.

TcMYC2 activates and directly binds to TcCHS, TcAOC, and TcGLIP promoters
Given that binding sites of bHLH TFs were predicted in the TcCHS, TcALDH, TcAOC, and TcGLIP promoter sequences, and the TcMYC2 expression pattern was similar to that of four pyrethrin biosyn- MeJA-treated 1-month-old tissue-cultured seedlings with three biological replicates. The fold change in relative expression level was normalized to the 0 h nontreated samples. Error bars represent mean ± SD with three biological replicates and two technical replicates. Different lower-case letters indicate a significant difference (P < 0.05) from one-way ANOVA followed by a post-hoc Tukey test. b qRT-PCR analysis of TcMYC2 transcript abundance in leaf and flower tissues at different developmental stages, normalized against the pd transcript level. S2 flower stages (pd: pedicel of immature bud; rp: receptacle; bc: involucral bract; rf: ray floret; df: disc floret); S4 flower stages (os4: fully open ray floret, cs4: unopened ray floret); L1: immature developing leaf; L2: mature leaf. Different letters indicate a significant difference (P < 0.05) from one-way ANOVA followed by a post-hoc Tukey test. c qRT-PCR analysis of TcMYC2 transcript abundance from S1 to S6 stages, normalized against the S1 transcript level. Different letters indicate a significant difference (P < 0.05) from one-way ANOVA followed by a post-hoc Tukey test. thesis genes, yeast one-hybrid (Y1H) assays were conducted to test the hypothesis that TcMYC2 regulates the expression of these genes by directly binding to the corresponding promoter. The promoter of each gene was cloned and inserted into the pHis2.1 vector to generate reporter constructs. TcMYC2 was fused to the GAL4 activation domain to generate the effector construct pGADT7-TcMYC2 (Fig. 4a). The reporter and effector constructs were cotransformed into yeast cells. Yeast grew on SD/−Leu/−Trp/−His medium only when pGADT7-TcMYC2 was co-transformed with pHis2.1-TcCHS, pHis2.1-TcAOC, or pHis2.1-TcGLIP but not with pHis2.1, suggesting that TcMYC2 can directly bind to the TcCHS, TcAOC, and TcGLIP promoters (Fig. 4b). The direct binding motif in the promoters was verified by electrophoretic mobility shift assays (EMSAs). The shifted band was separately detected in the presence of both TcMYC2 and a labeled probe containing the Gbox in the TcAOC promoter or E-box in the TcCHS and TcGLIP promoters (Fig. 4c). The intensity of the shifted band decreased with addition of 50-fold concentrations of cold competitor (unlabeled probe) and recovered with addition of mutant competitor probe.

Stable transformation of TcMYC2 in hairy roots
To verify the function of stably overexpressed TcMYC2, TcMYC2 was inserted in the pBI121 vector, and the recombinant vector was transferred to T. cinerariifolium via Agrobacterium rhizogenes to obtain stable transgenic hairy roots. After 30 d (Fig. 5a), positive transgenic hairy roots were selected via PCR amplification of rolB and 35 s-TcMYC2 (Fig. 5b). We first checked the gene expression of TcMYC2 and its target genes TcCHS, TcAOC, and TcGLIP in transgenic hairy root. As expected, TcAOC and TcGLIP expression was significantly upregulated, suggesting that TcMYC2 may positively regulate its target genes (Fig. 5c). However, the C t value of TcCHS was close to 40 and was barely expressed in hairy roots, owing to the lack of trichome tissue in which TcCHS is specifically expressed [30]. The GC-MS analysis revealed that no pyrethrins were produced in hairy roots except for (E)-β-farnesene (EβF), which shares the substrate DMAPP with pyrethrins. Thus, TcCHS expression in the trichome was indispensable for de novo pyrethrin biosynthesis even though the substrate was supplied. No pyrethrins were detected in stable transformed green callus, which further supported this hypothesis (Fig. 5d). Thus, based on stable transformation in hairy roots, TcMYC2 positively upregulated its target genes TcAOC and TcGLIP in T. cinerariifolium.

Transient overexpression of TcMYC2 upregulates pyrethrin biosynthesis genes and pyrethrin accumulation
The role of TcMYC2 in regulating pyrethrin biosynthesis in T. cinerariifolium was investigated. Given that stable transformation of T. cinerariifolium is not available and to mimic short-term MeJAinduced upregulation of pyrethrin biosynthesis, we transiently overexpressed TcMYC2 in T. cinerariifolium leaves, and determined the change in expression of pyrethrin biosynthesis genes by qRT-PCR at 2, 4, and 6 d post transformation. Compared with mock plants transformed with blank pGreenII62-SK, TcMYC2 transcript abundance was significantly increased in transformed plants, peaking at 4 d, and thereafter declined; however, at 6 d post transformation, TcMYC2 transcript abundance in transformed plants remained higher than that in the mock plants (Fig. 6a). Expression of TcCHS and TcAOC was markedly elevated in TcMYC2overexpressing plants, peaking at 4 d, which was 11.7-to 55.7fold higher than that in the mock plants (Fig. 6a). Expression of TcGLIP increased by 2.2-fold at 2 d compared with the control, and thereafter decreased to levels lower than the control (Fig. 6a). Significant differences were detected in the total pyrethrin contents between TcMYC2-overexpressing and mock leaves (Fig. 6bc). These

Transient silencing of TcMYC2 decreases pyrethrin biosynthesis gene expression and pyrethrin accumulation
A viral-induced gene silencing (VIGS) assay was conducted to confirm the indispensable role of TcMYC2 in regulating pyrethrin biosynthesis. Compared with the mock (empty pTRV2 vector), TcMYC2 expression in T. cinerariifolium leaves was significantly reduced by ∼93% in silenced plants (Fig. 6d). Downregulation of TcMYC2 expression resulted in a significant decrease (∼1.2to 2.5-fold) in the expression level of TcCHS, TcAOC, and TcGLIP (Fig. 6d). The total pyrethrin content was also reduced in TcMYC2silenced T. cinerariifolium leaves (Fig. 6). These results suggest that TcMYC2 is an important positive regulator of pyrethrins biosynthesis, which acts by directly activating the pyrethrins biosynthetic genes.

Continuous MeJA application increases trichome density and pyrethrin content
Phytohormones promote the formation of leaf trichomes in plants [31,32]. In T. cinerariifolium, the ovary epidermis has a high density of glandular trichomes in which pyrethrin precursors are produced [33]. However, the trichome number on leaves is limited. To examine the effect of MeJA treatment on trichome development in T. cinerariifolium, seedlings at the four-leaf stage were subjected to continuous MeJA treatment for 2 weeks. A significant (1.6-fold) increase in trichome density was observed on MeJA-treated leaves compared with mock leaves (Fig. 7a and b), suggesting that prolonged application of MeJA promotes the initiation of glandular trichomes. Given that the short MeJA treatment only transiently enhanced the pyrethrin content in the first few hours, we performed high-performance liquid chromatography (HPLC) analysis to detect pyrethrins in samples treated with MeJA for 2 weeks. Compared with mock plants, the pyrethrin I and pyrethrin II contents were significantly increased in MeJA-treated plants by vector; pTRV2-TcMYC2: TcMYC2 silenced leaves. The fold change in relative expression level was normalized to the mock expression. e Representative HPLC chromatogram of total pyrethrins extracted from TRV-VIGS treatment leaves. CK: VIGS empty pTRV2 vector; pTRV2-TcMYC2: TcMYC2 silenced leaves. f Pyrethrin content in T. cinerariifolium leaves after TRV-VIGS treatment of TcMYC2. The ordinate represents the intensity peak area detected by HPLC; the abscissa represents the six major pyrethrin compounds. Values are the mean ± SD of three biological replicates. * P < 0.05, * * P < 0.01 (two-tailed Student's t-test).
14% at 2 d post treatment, in contrast to the rapid decline in pyrethrin content at 8 h after short-term MeJA treatment (Fig. 7c). Thus, prolonged application of MeJA positively regulated glandular trichome development and pyrethrin accumulation in leaves.

Discussion
Jasmonates play an important role in controlling secondary metabolite synthesis through a network of TFs that regulate metabolite biosynthesis genes [11]. In this study, transient shortterm MeJA treatment of T. cinerariifolium leaves strongly induced expression of upstream genes in the pyrethrin biosynthesis pathway, although the expression levels quickly declined. A transient increase in pyrethrin accumulation resulted, which lasted for only several hours and thereafter declined. The pyrethrin biosynthetic pathway is long and complex. Most pyrethrin biosynthesis genes and precursors are expressed in and exported from trichomes to synthesize pyrethrins, which mainly accumulate in the secretory cavity of the ovary wall to protect the seed [33]. Mature leaves contain no corresponding storage structures and the sudden excessive accumulation of pyrethrins might disturb the cellular homeostasis. This was shown previously in T. cinerariifolium with overexpression of TcCHS driven by the Chrysanthemum RUBISCO promoter. Overexpression of TcCHS increased the pyrethrin content but led to a significant decrease in chlorophyll content [34]. A pyrethrin content of 0.05% or 0.001% still exerts good control of pyrethrin-sensitive herbivorous insects [35,36]. The average pyrethrin content in leaves is ∼0.1% (dry weight), which is still toxic to pyrethrin-sensitive insects in leaves without external stimuli. We presume that it is not necessarily energetically costly to respond to JAs to increase the pyrethrin content.
Continuous MeJA treatment of seedlings at the four-leaf stage for 2 weeks increased the glandular trichome density and pyrethrin accumulation in the leaves. We speculate that under long-term MeJA exposure, seedlings gradually adapt to exogenous hormone signals, which results in morphological changes that enable increased pyrethrin production. These changes have been reported in other Asteraceae species, such as Asteraceae annua, which showed upregulation of the artemisinin content in leaves together with trichome development on the leaf surface [37]. Similarly, a significant correlation between the monoterpene and sesquiterpene content and gland density on leaves in chrysanthemum has been reported [38].
MYC, a member of the bHLH superfamily that harbors the conserved HLH domain, can bind to the G-box or E-box motifs in the promoters of target genes to regulate their involvement in secondary metabolite biosynthesis [39]. As expected, TcMYC2 could bind to the TcCHS, TcAOC, and TcGLIP promoters and positively regulate expression of downstream genes in vitro. Plants transiently silenced for TcMYC2 expression showed significant downregulation of pyrethrin biosynthesis genes and dramatically reduced pyrethrin contents (Fig. 6d and e). Thus, TcMYC2 positively regulated pyrethrin biosynthesis in T. cinerariifolium leaves. In JA-treated Arabidopsis plants, MYC2 positively regulates the expression of flavonoid biosynthesis genes [40]. MYC2 orthologs of tobacco regulate nicotine biosynthesis [41]. These MYC2 proteins show the strongest binding activity to the G-box followed by the 5 -CACGTG-3 sequence. Thus, similar mechanisms operate in the regulation of the biosynthesis of various secondary metabolites, indicating that MYC2-like genes have conserved functions in these species [42]. MYCs generally exhibit functional redundancy and weakening of the function of one MYC can be compensated by other MYCs. In Arabidopsis, MYC5 has functions redundant with other MYCs, which participate in root growth inhibition, leaf senescence, and plant defense against insect attack and pathogen infection regulated by JA [43]. In Nicotiana tabacum, NtMYC2a coordinates with NtMYC2b to positively regulate nicotine biosynthesis [44]. In the current study, the molecular phenotype of TcMYC2silenced T. cinerariifolium plants was not rescued by other potential MYCs, suggesting that TcMYC2 might be an indispensable, nonredundant positive regulator of pyrethrin biosynthesis in T. cinerariifolium.
Although TcMYC2 showed strong activity on pyrethrin biosynthesis genes in vitro and a strong decrease in pyrethrin content in VIGS plants, gene expression and pyrethrin content in leaves with transient overexpression of TcMYC2 showed only a slight and noncontinuous increase. In other studies, MYC-mediated continuous activation of JA responses inhibit growth, and are potentially harmful or fatal [45]. For example, overexpression of TwMYC2a and TwMYC2b results in downregulation of triptolide biosynthesis in hairy roots of Tripterygium wilfordii [46]. A possible reason is that TcMYC2 has multiple functions and may activate other negative transcription factors simultaneously. For instance, MYC2 directly binds to the promoter of JAZ repressors, which restrain JA response [47,48]. Plants overexpressing MYC2 constitutively resynthesize JAZ repressors in response to JA stimulation to close activation of the pathway in a controlled manner [48].
To further explore the function of TcMYC2, stable transformation of hairy roots with TcMYC2 was conducted. The target genes TcAOC and TcGLIP were significantly upregulated, whereas TcCHS transcripts were not detected, and no pyrethrins were produced either with incubation with substrates. TcCHS is almost exclusively expressed in glands, but no pyrethrins and glands are produced by hairy roots of T. cinerariifolium [49]. Furthermore, tissue differentiation is a prerequisite for production of certain secondary metabolites [50]. Typical examples include some Hypericum species that yield hypericin, but their hairy roots are devoid of hypericin [51]. To clarify whether glands are essential for pyrethrin synthesis, the function of TcMYC2 requires further study in other tissues and experimental systems.
In summary, this study demonstrates that MeJA induces TcMYC2 expression in T. cinerariifolium and subsequently activates the expression of pyrethrin biosynthesis genes by binding to their promoters. Prolonged exposure of T. cinerariifolium seedlings to MeJA slightly enhances production of glandular trichomes, and consequently pyrethrin accumulation, but only in immature leaves (Fig. 8). This study reports the first of likely several TFs that are indispensable to promoting the expression of pyrethrin biosynthesis genes and pyrethrin accumulation. Elucidation of the full spectrum of relevant TFs may provide a novel means of enhancing pyrethrin yield through metabolic engineering of T. cinerariifolium.

Plant materials
T. cinerariifolium 'W99' was used in this study because it is easy rooting, grows rapidly, and has high pyrethrin content. T. cinerariifolium is a self-incompatible, highly heterogenous species [52], and individual genotypes have a unique pyrethrin content. Onemonth-old uniformly sized 'W99' seedlings grown in vitro on halfstrength Murashige and Skoog medium in large airtight containers of identical size were used for MeJA treatment. The seedlings were grown at 25 • C under a 16 h/8 h (light/dark) photoperiod.
Pyrethrum flower heads and leaves at different developmental stages (S1-S6 [3] and L1-L2 [15], respectively,) as well as flower tissues, were collected in triplicate, immediately frozen in liquid nitrogen, and stored at −80 • C until use.

MeJA treatment
For short-term MeJA treatment, 1-month-old 'W99' seedlings grown in vitro were sprayed with 5 mL of 2 mM MeJA dissolved in 0.8% ethanol. Leaves were collected in triplicate after 0 (control), 2, 4, 6, 8, 12, and 24 h. The samples were immediately frozen in liquid nitrogen and stored at −80 • C until use.
For long-term MeJA treatment, 'W99' seedlings at the four-leaf stage were sprayed with 5 mL of 300 μM MeJA dissolved in 0.8% ethanol, and the roots were irrigated with 5 mL every 2 d for a total of 2 weeks. Plants sprayed with 0.8% ethanol served as a mock control. Subsequently, leaves were collected and oven-dried at 50 • C for 48 h to constant dry weight.
The qRT-PCR assay was performed on the Roche LightCycler ® 96 Real-time PCR System (Roche, Basel, Switzerland), in accordance with the manufacturer's instructions, using SYBR Premix Ex Taq II (Takara, Kusatsu, Japan) and sequence-specific primers described previously [12]. Other primers were based on the transcriptome and checked using TBtools [64]. Primers used for qRT-PCR are listed in Table S1 (see online supplementary material). Three biological replicates were performed for each gene, with two technical repeats per biological replicate. GADPH was used as the internal reference gene. Relative gene expression was calculated using the 2 − Ct method [65]. Student's two-tailed t-test was used to determine statistical significance at the 5% and 1% significance levels.

Transcription factor and TcMYC2 identification
PlantTFDB 5.0 was used to predict all TFs [66]. Protein domains were identified using the SMART online tool [67]. After filtering out weakly expressed genes and proteins lacking conserved domains, all bHLH TFs were visualized. The FPKM values of genes from the MeJA treatment and flower stage transcriptome were used to perform a cluster analysis. Heatmaps were generated using TBtools [64].
The open reading frame of TcMYC2 was cloned from T. cinerariifolium cDNA using sequence-specific primers (Table S1, see online supplementary material). Sequences of T. cinerariifolium genes homologous to Arabidopsis genes were identified by a BLAST search against The Arabidopsis Information Resource database, and to genes of other plant species by a BLAST search against the National Center for Biotechnology Information database. The AtMYC proteins were scanned using HmmerSearch v3.0 software with the PfamA domain database [68]. TcMYC2 and AtMYC family coding sequences were aligned using mafft v7.158 software [69]. A maximum-likelihood phylogenetic tree was constructed using IQtree2 software [70], using the best-fit model (JTT + F + I + G4) selected by ModelFinder [71] and with 1000 bootstrap replications.

Cloning and bioinformatic analysis of gene promoters
The promoter sequences of TcCHS, TcALDH, TcAOS, and TcGLIP were mined from the draft genome sequence of T. cinerariifolium [72]. The associated cis-elements and their biological functions were determined using PlantCARE [73], and the cis-regulatory elements (E-box, G-box, and G-box-like) were identified using PlantPAN3.0 [74].

Dual-LUC transient expression assay
To generate reporter constructs, promoter fragments of downstream genes (TcCHS, TcGLIP, TcALDH, and TcAOC) were cloned into the HindIII-linearized pGreenll0800-LUC vector. The TcMYC2 coding sequence was cloned into the HindIII-linearized pGreenII62-SK vector under the control of the CaMV 35S promoter. The resultant vectors were transiently coexpressed in N. benthamiana leaves. Luminescence was detected using the LB 985 Nightshade system (Berthold, Bad Wildbad, Germany). Introduction of the pGreen-62-SK empty vector and of pGreen-62-SK/pGreenII-0800-LUC:proTcCHS/TcGLIP/TcAOC constructs served as a negative control.

EMSA
The open reading frame without the terminator codon of TcMYC2 was inserted in the pET6HN-C vector protein, fused in the Nterminal frame of 6 × His. The vector pET6HN-C-TcMYC2-6 × His was transformed into Escherichia coli strain Rosetta (DE3). For induction of the recombinant protein, 0.5 mM IPTG was used, and the cultures were incubated at 18 • C for 16 h. Then, Ni-NTA was used to purify the recombinant proteins.
For EMSA, the promoter fragments of TcCHS, TcAOC, and TcGLIP contained E-box or G-box cis-elements. FAM-labeled probes and the same unlabeled DNA fragments, cis-element mutant DNA fragments were used as competitors in the assay. The EMSA assays were performed as described previously [75]. FAM-labeled DNA was detected by the chemiluminescence method.

Subcellular localization of TcMYC2
The coding sequence of TcMYC2 was amplified by PCR using sequence-specific primers containing the SwaI and KpnI restriction sites at their ends (Table S1, see online supplementary material). The PCR product was cloned into pSuper1300-GFP, derived from pCambia1300, using the ClonExpress ® II One Step Cloning Kit (Vazyme Biotech, Nanjing, China). The resultant plasmid was introduced into Agrobacterium tumefaciens strain GV3101, and then co-infiltrated into N. benthamiana leaves together with the RFP-NLS plasmid (nuclear marker) following methods described previously [76]. The empty vector was used as a control. After 72 h of weak light exposure, fluorescence signals were observed with a confocal laser scanning microscope (Leica TCS-SP8; Leica, Wetzlar, Germany).

TcMYC2 stable transformation of hairy roots
The full-length TcMYC2 coding sequence was inserted into pBI121 vector digested with BamHI and SacI using homologous recombination, and transferred into A. rhizogenes strain MSU440. The infection process followed that of previous studies [49].
Genomic DNA was extracted from transgenic hairy roots after 30 d using the CTAB method. Successful transfer of the transgenepositive cells was confirmed by PCR. The primers for rolB of the A. rhizogenes Ri plasmid were designed in accordance with a previous study [77]. Transgenic hairy roots were determined by qRT-PCR and GC-MS analysis.

Transient overexpression of TcMYC2 in T. cinerariifolium leaves
The full-length TcMYC2 coding sequence was cloned into the HindIII-linearized pGreenII62-SK vector downstream of the CaMV 35S promoter using gene-specific primers (Table S1, see online supplementary material). The pGreenII62-SK: TcMYC2 vector was co-transformed together with the helper plasmid pSoup19 into A. tumefaciens strain GV3101. Leaves of T. cinerariifolium were placed in a 500 mL beaker containing A. tumefaciens suspension. The leaves were infiltrated under vacuum of 0.23 atm for 5 min. The agroinfiltrated leaves were dried with filter paper and stored in a Petri dish lined with a moisture-retaining filter paper at the bottom, as described previously [78]. After 2, 4, and 6 d of incubation, the agroinfiltrated leaf samples were used for qRT-PCR, and the 4 d samples were used for HPLC analysis.

VIGS assay
Tobacco rattle virus (TRV)-based vectors, pTRV1 and pTRV2, were used for the VIGS assay. A 256 bp fragment complementary to the TcMYC2 coding sequence was designed, searched in the local transcriptome library using siRNA Finder [79], and cloned into the BamHI-linearized pTRV2 vector by homologous recombination using gene-specific primers (Table S1, see online supplementary material). The pTRV2:TcMYC2 and pTRV1 plasmids were transformed into chemically active GV3101 cells using the liquid nitrogen freeze-thaw method. The VIGS assay was conducted as described previously [80] with three biological replicates. Empty vectors were used as controls. After 14 d, the samples were subjected to qRT-PCR and HPLC analyses.

Measurement of pyrethrin content by GC-MS
Leaf samples exposed to short-term MeJA treatment were ground into a fine powder in liquid nitrogen. Powdered tissue (0.5 g) was transferred to a tube containing 2 mL methyl tert-butyl ether (MTBE), with 0.01 ng/mL tetradecane as an internal standard. The tube was vortexed for 3 min at maximum speed and incubated at 24 • C with rotation of 50 rpm for 2 h. The sample was dried using Na 2 SO 4 and filtered through a 0.22-μm mesh filter. The pyrethrin content was measured using a GC/MS-QP2010Ultra apparatus (Shimadzu Corporation, Kyoto, Japan) equipped with a HP-5 MS column. Three biological replicates were performed for each sample. The GC process was performed as described previously [81]. Commercial pyrethrum extract (Sigma-Aldrich, St Louis, MI, USA) was used as the standard.

Pyrethrin content measurement by HPLC
The transiently transformed samples and samples exposed to long-term MeJA treatment were oven-dried at 50 • C for 48 h to constant dry weight. The dried leaves were milled into a fine powder and dissolved in n-hexane at a concentration of 1 g per 6 mL in glass tubes with screwcaps. After vortexing for 30 s, the samples were extracted in an ultrasound water bath for 10 min, then vortexed again for 30 s, as described previously [74]. The samples were filtered through a 0.22-μm mesh filter. The pyrethrin content was measured by HPLC, with three biological replicates, as described previously [6]. Commercial pyrethrum extract (Sigma-Aldrich, St Louis, MI, USA) was used as the standard.

Analysis of the effect of MeJA on trichome density
Seedlings at the four-leaf stage were sprayed with 300 μM MeJA every 2 d for 2 weeks. The mock control was treated with 0.8% ethanol. The sixth leaf, which emerged after MeJA treatment, was excised and prepared for scanning electron microscopy as described previously [82]. The trichome number was determined from the micrographs of six biological replicates. Statistical significance was determined by Student's t-test at P < 0.05.