Abstract
Biosynthesis of medicinally valuable terpenoid indole alkaloids (TIAs) in Catharanthus roseus is regulated by transcriptional activators such as the basic helix–loop–helix factor CrMYC2. However, the transactivation effects are often buffered by repressors, such as the bZIP factors CrGBF1 and CrGBF2, possibly to fine-tune the accumulation of cytotoxic TIAs. Questions remain as to whether and how these factors interact to modulate TIA production. We demonstrated that overexpression of CrMYC2 induces CrGBF expression and results in reduced alkaloid accumulation in C. roseus hairy roots. We found that CrGBF1 and CrGBF2 form homo- and heterodimers to repress the transcriptional activities of key TIA pathway gene promoters. We showed that CrGBFs dimerize with CrMYC2, and CrGBF1 binds to the same cis-elements (T/G-box) as CrMYC2 in the target gene promoters. Our findings suggest that CrGBFs antagonize CrMYC2 transactivation possibly by competitive binding to the T/G-box in the target promoters and/or protein–protein interaction that forms a non-DNA binding complex that prevents CrMYC2 from binding to its target promoters. Homo- and heterodimer formation allows fine-tuning of the amplitude of TIA gene expression. Our findings reveal a previously undescribed regulatory mechanism that governs the TIA pathway genes to balance metabolic flux for TIA production in C. roseus.
Introduction
Terpenoid indole alkaloids (TIAs; vinca alkaloids) are produced in a limited number of plant species in the Apocynaceae, Loganiaceae, and Rubiaceae families (Memelink and Gantet, 2007; De Luca et al., 2014). The medicinal plant Catharanthus roseus (Madagascar periwinkle) from the Apocynaceae family produces over 130 TIAs (Thamm et al., 2016). Catharanthus roseus is the exclusive natural source of two anticancer drugs, vinblastine and vincristine. TIA biosynthesis is a highly coordinated biological process, which involves multiple enzymatic steps and transcriptional regulators (Courdavault et al., 2014; Stavrinides et al., 2015, 2016; Tatsis et al., 2017; Carqueijeiro et al., 2018; Qu et al., 2018; Caputi et al., 2018). Moreover, TIA biosynthesis requires at least four cellular compartments and complex intra- and intercellular translocations of the intermediates (Patra et al., 2013b; Courdavault et al., 2014; Thamm et al., 2016). In TIA biosynthesis, the shikimate pathway provides the indole moiety, tryptamine, whereas the methyl erythritol phosphate (MEP) and iridoid pathways contribute the monoterpenoid moiety, secologanin (see Supplementary Fig. S1 at JXB online). Condensation of tryptamine and secologanin by strictosidine synthase (STR) results in the production of the first TIA, strictosidine, which is deglucosylated by strictosidine-β-D-glycosidase to form strictosidine aglycon. Subsequently, the TIA biosynthetic pathway branches to synthesize various other TIAs and precursors of bisindole alkaloids (Supplementary Fig. S1; Stavrinides et al., 2015, 2016; Tatsis et al., 2017; Carqueijeiro et al., 2018; Qu et al., 2018; Caputi et al., 2018). The bisindole alkaloids vinblastine and vincristine are synthesized by coupling catharanthine and vindoline, a reaction catalysed by peroxidase 1 (Sottomayor et al., 2004; Costa et al., 2008).
A number of transcriptional activators and repressors have been isolated and characterized for their roles in regulating TIA biosynthesis. The octadecanoid-derivative responsive Catharanthus AP2-domain (ORCA) transcription factors (TFs) ORCA2 and ORCA3 are known to activate key steps in the TIA pathway (Menke et al., 1999; van der Fits and Memelink, 2000; Pan et al., 2012). Recently, two other AP2/ERF TFs, ORCA4 and ORCA5, which form a physical cluster with ORCA3 (Kellner et al., 2015), have been shown to regulate the TIA pathway through overlapping yet distinct mechanisms (Paul et al., 2017). The basic helix–loop–helix (bHLH) TFs CrMYC2, bHLH IRIDOID SYNTHESIS 1 (BIS1), and BIS2 are also involved in the regulation of the TIA pathway (Zhang et al., 2011; Van Moerkercke et al., 2015, 2016). CrMYC2 acts upstream of ORCA3 and co-regulates TIA pathway genes, such as tryptophan decarboxylase (TDC), concomitantly with ORCA3 (Zhang et al., 2011; Paul et al., 2017). BIS1 and BIS2 are major activators of the iridoid branch of the TIA pathway, capable of activating all seven genes encoding enzymes that catalyse the sequential conversion of geranyl diphosphate to loganic acid (Van Moerkercke et al., 2015, 2016). We recently show that ORCA4 also activates several iridoid branch genes, in addition to co-regulating other TIA pathway genes with ORCA3. In addition, CrMYC2 and the ORCAs act downstream of a mitogen-activated protein kinase cascade to modulate TIA pathway gene expression (Paul et al. 2017). The WRKY TF CrWRKY1 has also been shown to influence TIA biosynthesis in C. roseus (Suttipanta et al., 2011). Together with the transcriptional activators, a number of transcriptional repressors function in the TIA pathway. The zinc finger TFs ZCT1, ZCT2, and ZCT3 bind to the TDC and STR promoters to repress their activities (Pauw et al., 2004). Other TIA pathway repressors include the bHLH factor repressor of MYC2 targets 1 (RMT1) and the jasmonate ZIM-domain (JAZ) TFs (Patra et al., 2018), as well as the basic leucine zipper (bZIP) TFs G-box binding factor 1 (GBF1) and GBF2 (Sibéril et al., 2001a).
In plants, bZIPs comprise a large TF family involved in diverse biological processes, including light and stress signaling, seed maturation, pathogen defense, flower development, and secondary metabolite biosynthesis (Jakoby et al., 2002). The 75 bZIP TFs in Arabidopsis are divided into 10 groups based on a conserved basic domain and additional sequence motifs (Jakoby et al., 2002). The Arabidopsis group G bZIP factors GBF1 (bZIP41), GBF2 (bZIP54), GBF3 (bZIP55), and bZIP16 have been implicated in light signaling (Schindler et al., 1992; Mallappa et al., 2008; Hsieh et al., 2012). In addition to the bZIP domain, GBFs contain three conserved proline-rich motifs (PRMs) at the N-termini (Jakoby et al., 2002). The PRM in GBFs can be an activation domain, as shown in Arabidopsis, or a repression domain, as in soybean and Catharanthus (Sibéril et al., 2001b; Jakoby et al., 2002). CrGBF1 and CrGBF2 bind to the T/G-box elements in the STR and TDC promoters in vitro and repress STR activity in C. roseus cells (Pasquali et al., 1999; Sibéril et al., 2001a). However, the influence of CrGBFs on TIA biosynthesis and the molecular mechanism for regulation of TIA accumulation have not been thoroughly investigated.
Biosynthesis of specialized metabolites in plant cells, often in response to various abiotic or biotic stresses, is an energy-intensive process. Moreover, many metabolites, such as TIAs, are toxic to plant cells (Payne et al., 2017). Plants have thus evolved unique mechanisms to tightly regulate the synthesis of these specialized metabolites. The phytohormone jasmonic acid (JA) is a major elicitor of the TIA pathway. Most TIA pathway activators and repressors are coordinately induced in response to JA treatments (Memelink and Gantet, 2007; Thamm et al., 2016), suggesting that TIA biosynthesis is under stringent regulatory control by the concerted actions of activators and repressors that fine-tune TIA accumulation in C. roseus. This notion is further strengthened by a number of studies in which overexpression of an activator or suppression of a repressor has limited or no effects on TIA accumulation in C. roseus hairy roots (Peebles et al., 2009; Van Moerkercke et al., 2015; Li et al., 2015; Rizvi et al., 2016). However, key questions remain as to whether and how these regulators interact to modulate TIA pathway gene expression and alkaloid accumulation.
In this work, we demonstrated that overexpression of CrMYC2 or RNAi-mediated suppression of CrGBF1 perturbs gene expression in both the indole and iridoid branches of the TIA pathway and reduces alkaloid accumulation in transgenic C. roseus hairy roots. Notably, the expression of key activators, such as ORCA3, BIS1, and BIS2, and the repressors CrGBFs and ZCTs was significantly altered. We investigated whether these unexplained transcriptional and metabolic consequences are the results of a counter-effect from one or more of these repressors to prevent over-accumulation of TIAs. We demonstrated that CrGBFs form homo- and heterodimers that repress the transcriptional activity of key gene promoters in both the indole and iridoid branches of the TIA pathway. Furthermore, CrGBFs form heterodimers with CrMYC2 and antagonize the transactivation activity of CrMYC2 on target promoters. The homo- and heterodimers possibly allow further fine-tuning of the amplitude of gene repression. Collectively, our findings highlight the biological significance of cross-family TF interactions and reveal a novel regulatory mechanism governing the expression of TIA pathway genes, likely to balance the metabolic flux for TIA accumulation in C. roseus.
Materials and methods
Plant materials, RNA isolation, cDNA synthesis and cloning
Catharanthus roseus var. ‘Little Bright Eyes’ (NE Seed Commercial and Garden, USA) seeds were surface sterilized and germinated on half-strength Murashige and Skoog (MS) basal medium (Caisson Labs, USA). Two-week-old seedlings were treated with 100 μM methyl jasmonate (MeJA) for 6, 12, and 24 h. Total RNA isolated from MeJA-treated seedlings using the RNeasy Plant Mini Kit (Qiagen, USA) was used for cDNA synthesis. The full-length cDNAs of CrMYC2 (GenBank accession no. AF283507), CrGBF1 (GenBank accession no. AF084971), and CrGBF2 (GenBank accession no. AF084972) were amplified from first-strand cDNA using gene-specific primers with suitable restriction enzyme sites (see Supplementary Table S1). PCR products were cloned into the pGEM-T Easy vector (Promega, USA) and confirmed by sequencing.
Plasmid construction, plant transformation and generation of hairy roots
For overexpression, the full-length cDNA of CrMYC2 was cloned into the plant transformation vector pKYLX71 under the control of the CaMV 35S promoter and the rbcS terminator (Schardl et al., 1987). For RNAi-mediated suppression, a 216 bp fragment was amplified from CrGBF1 and cloned in both sense and antisense orientation in pKYLX80 containing the soybean omega-3 fatty acid desaturase (FAD3; accession no. DQ672337) intron (Schardl et al., 1987). The sense and antisense fragments with the FAD3 intron were excised from pKYLX80 and ligated into modified binary vector pCAMBIA2300 containing the CaMV 35S promoter and rbcS terminator. The plasmids were mobilized into Agrobacterium rhizogenes R1000 using the freeze–thaw method (Weigel and Glazebrook, 2006). Transformation of C. roseus seedlings with A. rhizogenes and generation of hairy roots were conducted as previously described (Choi et al., 2004; Suttipanta et al., 2011). Briefly, roots of 2-week-old C. roseus seedlings were removed by scalpel and the seedlings were immersed in A. rhizogenes suspension for 45 min. The seedlings were then blotted on sterile filter paper to remove excess Agrobacterium, and transferred to the half-strength MS medium. After 2 d of co-culture at 28 °C in the dark, the explants were rinsed with half-strength MS basal solution containing 800 mg l−1 cefotaxime and then transferred to half-strength MS basal medium containing 400 mg l−1 cefotaxime. New roots usually protrude from cutting areas of the explants within 3 weeks. The resulting roots were excised and selected on one-third-strength Schenk and Hilderbrandt (SH) basal medium (Caisson Labs, USA) supplemented with sucrose (30 g l−1), agar (10 g l−1), kanamycin (100 mg l−1), and cefotaxime (400 mg l−1). The actively growing hairy roots were selected and cultured in one-third-strength SH liquid medium on a shaker at 100 rpm in the dark at room temperature. After 2 weeks, the newly generated hairy roots were collected, frozen immediately in liquid nitrogen, and stored at −80 °C for subsequent gene expression analysis and metabolite measurement.
cDNA synthesis and quantitative real time PCR
Total RNA was isolated from CrMYC2 overexpression (CrMYC2-OX) and empty vector (EV) control hairy roots using the RNeasy Plant Mini Kit following the manufacturer’s instructions (Qiagen, USA). About 2 μg of total RNA was used for DNase I digestion. Synthesis of first strand cDNA was performed using Superscript III reverse transcriptase (Invitrogen) in a total volume of 20 μl (Suttipanta et al., 2007). Quantitative real-time PCR (qRT-PCR) was used to measure transcripts levels of TIA pathway genes in hairy root lines. All primers used for qRT-PCR are listed in Supplementary Table S1. The C. roseus Ribosomal Protein Subunit 9 (RPS9) gene was used as an internal control. All qRT-PCRs were performed in triplicate and repeated twice.
Analysis of transgenic status of hairy roots
To verify the transgenic status of CrMYC2-OX, CrGBF1-RNAi, and EV control hairy root lines, gene-specific primers (see Supplementary Table S2) were used to PCR-amplify the rol B, rol C, vir C, and kanamycin resistant (nptII) genes. PCR products were analysed on a 1% ethidium bromide-stained agarose gel.
RNA sequencing library preparation and Illumina sequencing
RNA sequencing libraries of CrMYC2-OX and EV control hairy root lines were made using 2 µg of total RNA following a previously reported protocol (Hunt, 2015). Equal-amount libraries were pooled and sequenced using the Illumina Hiseq 2500 platform. Deep sequencing was performed in triplicate for each line with a 50 cycle single end run. The data quality was checked at the Sequencing and Genotyping Center, Delaware Biotechnology Institute at the University of Delaware, USA, and sequencing reads were provided in FASTq format. Demultiplexing, and adapter and barcode sequence trimming were performed with CLC Genomics Workbench version 8.0 (CLC bio, 2015).
Co-expression analysis
For co-expression analysis, CrMYC2-OX and EV control transcriptome, and transcriptome data from five different tissues (flower, mature leaf, immature leaf, stem, and root) obtained from the Sequence Read Archive (SRA) database at the National Center for Biotechnology Information (accession number SRA030483) were used. Raw reads were processed and reads per kilobase of transcript per million mapped reads (RPKM) values were calculated as previously described (Singh et al., 2015). Pair-wise Pearson correlation coefficients for each transcript were calculated using the RPKM. Matrix distances for the expression heat-map were computed with Pearson correlations of gene expression values (RPKM) by the heatmap.2 function of gplots (version 3.0.1) Bioconductor package in R (version 3.2.2) (Wickham, 2009).
Amino acid sequence alignment and phylogenetic analysis
Deduced amino acid sequences of CrGBF1, CrGBF2, and CrMYC2 were aligned with their corresponding Arabidopsis orthologues using the ClustalW program with the default parameters. A phylogenetic tree was constructed using the neighbor-joining (NJ) method through MEGA v.6 software. The statistical reliability for nodes in the phylogenetic tree was assessed by bootstrap analyses with 1000 replications.
Yeast two-hybrid assay
The full-length cDNAs of CrMYC2, CrGBF1, and CrGBF2 were cloned into the two yeast expression plasmids, pAD-GAL4-2.1 and pBD-GAL4 Cam (Stratagene, USA), respectively. Different combinations of bait and prey plasmids were transformed into yeast strain AH109 using the polyethylene glycol/LiCl method (Clontech, USA), and the transformants were selected on synthetic dropout (SD) medium lacking leucine and tryptophan (−Leu−Trp). Transformed colonies were then streaked on SD medium lacking histidine, leucine, and tryptophan (−His−Leu−Trp) to check protein–protein interaction.
Bimolecular fluorescence complementation assay
The full-length cDNAs of CrMYC2, CrGBF1, and CrGBF2 without their stop codons were individually cloned into pSAT6-nEYFP-N1 and pSAT6-cEYFP-N1 vectors (Citovsky et al., 2006). pSAT6-nEYFP-N1 and pSAT6-cEYFP-N1 vectors contain the N (nYFP) or C (cYFP) terminal fragment of yellow fluorescent protein (YFP), respectively. The resulting constructs were then co-electroporated into tobacco protoplasts in different combinations to check homo-/heterodimerization (Pattanaik et al., 2010b). Protoplasts were stained with 4′,6-diamidino-2-phenylindole (DAPI) as previously described (Patra et al., 2013a) and visualized using fluorescence microscopy (Nikon, Japan). Primers used for generating the bimolecular fluorescence complementation (BiFC) constructs are listed in Supplementary Table S2.
Plasmid construction and protoplast transient assay
The reporter plasmids used in the protoplast assay contained the firefly luciferase coding sequence driven by TDC, STR, geraniol 10-hydroxylase (G10H), ORCA3, and BIS1 promoters and rbcS terminator (Suttipanta et al., 2007; Paul et al., 2017; Patra et al., 2018). The T/G-box motif (AACGTG) in ORCA3/TDC promoters was changed to AAAATG to generate mORCA3/mTDC by site-directed mutagenesis (Zheng et al., 2004). The effector plasmids were generated by cloning the full-length coding sequences of CrMYC2, CrGBF1, and CrGBF2 into a pBlueScript vector containing the CaMV 35S promoter and rbcS terminator. A plasmid containing the β-glucuronidase (GUS) gene driven by the CaMV 35S promoter and rbcS terminator was used as an internal control. Protoplast isolation, electroporation, and luciferase assay were performed as previously described (Suttipanta et al., 2007; Pattanaik et al., 2010a,b).
Alkaloid extraction and quantification
For alkaloid extraction, 150 mg fresh tissue collected from actively growing hairy roots was ground in liquid nitrogen and extracted with 2 ml methanol. The extracts were concentrated to 250 µl with a rotary evaporator under vacuum. For preliminary analyses, 15 µl volumes of concentrated extracts of hairy root and commercial alkaloid standard mixture (catharanthine, tabersonine, and ajmalicine, Sigma-Aldrich) were applied onto thin-layer chromatography (TLC) plates (silica G60; Merck). The TLC was run on the solvent ethanol:chloroform (1:9, v/v) mixture and alkaloids were verified at 254 and 366 nm UV wavelength, respectively, and by color reaction with ceric (IV) ammonium sulfate stain. TIAs in each hairy root sample were quantified using HPLC followed by electrospray ionization–tandem mass spectrometry (LC-ESI-MS/MS) as previously described (Suttipanta et al., 2011).
Results
Ectopic expression of CrMYC2 perturbs TIA pathway gene expression and represses alkaloid accumulation in C. roseus hairy roots
MYC2 acts as a regulatory hub involved in diverse developmental and metabolic processes including JA signaling and specialized metabolite biosynthesis in plants (Dombrecht et al., 2007; Kazan and Manners, 2013). A previous study showed that CrMYC2 overexpression in a C. roseus cell line moderately induced ORCA3 expression; however, expression of downstream targets, such as STR, remained unchanged. Moreover, the effect of CrMYC2 overexpression on metabolite accumulation was not investigated (Zhang et al., 2011). To determine the effects of CrMYC2 on TIA pathway gene expression and metabolic outcomes, we generated transgenic C. roseus hairy root lines overexpressing CrMYC2 (CrMYC2-OX) (see Supplementary Fig. S2A). Empty vector (EV) hairy root lines served as control. The transgenic status of independent hairy root lines was verified by PCR (Supplementary Fig. S2B) and two independent lines (CrMYC2-OX-A and CrMYC2-OX-B) were chosen for further analysis. Compared with the EV control, relative expression of CrMYC2 was significantly higher in CrMYC2-OX lines (Supplementary Fig. S2C). The hairy root line (OX-B), showing consistently higher expression of CrMYC2 compared with the EV control, was chosen for RNA sequencing. We performed a hierarchical clustering analysis to provide an overview of the expression profiles of structural and regulatory genes in the TIA pathway in EV control and CrMYC2-OX lines (Supplementary Fig. S3). The analysis showed that CrMYC2 co-expressed with ORCA1, T3O, CrGBF1, ZCTs, and the box P-binding factor-1 (BPF1), whereas ORCA3 and CrGBF2 clustered with other indole and iridoid biosynthetic pathway genes. Expression of positive regulators, such as ORCA3, BIS1, and BIS2, was repressed in the CrMYC2-OX line compared with the EV control. Interestingly, the expression of the negative regulators, such as CrGBF1 and CrGBF2, was significantly up-regulated, whereas ZCTs were moderately induced, in CrMYC2-OX hairy roots. Notably, the expression of genes encoding key enzymes in the MEP, indole, and iridoid branches was repressed in CrMYC2-OX hairy roots.
To validate the transcriptome data, we first measured the transcript levels of key regulatory genes of the TIA pathway in the EV control and CrMYC2-OX-B lines by qRT-PCR (Fig. 1A). Expression of BIS1 and BIS2, major activators of the iridoid branch, was repressed by 50–63% in CrMYC2-OX hairy roots. ORCA3 is a positive regulator of a number of TIA pathway genes including TDC and STR and known to be activated by CrMYC2 (van der Fits and Memelink, 2000; Zhang et al., 2011). However, inconsistent with such previous reports, ORCA3 expression was deceased by 51% in CrMYC2-OX hairy roots compared with the EV control. This result points to the possibility of up-regulation of repressors in CrMYC2-OX. Consistent with this hypothesis, the expression of CrGBF1 and CrGBF2 was significantly higher in CrMYC2-OX hairy roots; notably, CrGBF1 transcript was 4.5-fold higher in CrMYC2-OX compared with the EV control (Fig. 1A). CrGBF1 and CrGBF2 are known to repress the expression of a key TIA pathway gene, STR, by binding to its promoter (Sibéril et al., 2001a). Moreover, expression of three zinc finger repressors, ZCT1, ZCT2, and ZCT3, was significantly increased in CrMYC2-OX hairy roots compared with the EV control (Fig. 1A). ZCTs are known to bind the TDC and STR promoters in C. roseus cell lines and repress their activity (Pauw et al., 2004).
qRT-PCR validation of TIA pathway gene expression, and alkaloid accumulation in empty vector (EV) control and CrMYC2 overexpression (CrMYC2-OX) hairy root lines. (A) The relative expression levels of TIA pathway genes encoding enzymes and transcriptional regulators in EV control and CrMYC2-OX hairy root lines. Ribosomal protein subunit 9 (RPS9) was used as an internal control. For abbreviations of pathway genes, see Supplementary Fig. S1. (B) Accumulation of catharanthine, tabersonine, and ajmalicine in EV control and two independent CrMYC2-OX lines. An equal amount of hairy root was used for each line for total alkaloid extraction. Alkaloid extracts from EV control and CrMYC2-OX lines were analysed by LC-MS/MS, and the levels of catharanthine, ajmalicine, and tabersonine were estimated based on peak areas. Each sample was tested in triplicate. Data represent means ±standard deviation (SD) of three replicates. Significant differences among different lines were determined at P<0.05 by Student’s t-test, and are indicated by an asterisk.
Next, we measured the expression of structural genes in the TIA pathway using qRT-PCR. Expression of two indole branch genes, anthranilate synthase α (ASα) and TDC, was decreased by 34% and 31%, respectively, in CrMYC2-OX hairy roots. Expression of the iridoid branch genes, including G10H, iridoid synthase (IRS), iridoid oxidase (IO), and 7-deoxyloganic acid hydroxylase (7-DLH), was decreased by 32%, 25%, 26%, and 22%, respectively, in CrMYC2-OX hairy roots compared with the EV control (Fig. 1A). In addition, CrMYC2 overexpression significantly repressed the expression of 7-deoxyloganetic acid glucosyl transferase (7-DLGT) and cytochrome P450 reductase (CPR) by 40% and 25%, respectively, although the expression of geraniol synthase (GES) and 10-hydroxygeraniol oxidoreductase (10HGO) was not significantly altered (Fig. 1A). Expression of STR was decreased by 36% in CrMYC2-OX hairy roots compared with the EV control. Collectively, these finding suggests that ectopic expression of CrMYC2 negatively affects indole and iridoid pathway gene expression.
Expression of JA biosynthetic genes is regulated by MYC2 in other plant species, including Arabidopsis (Dombrecht et al., 2007) and tomato (Du et al., 2017). To determine whether CrMYC2 overexpression affects JA biosynthetic gene expression, we analysed the transcriptomes of CrMYC2 overexpression and control hairy root lines. Expression of two key genes, allene oxidase synthase (AOS) and allene oxidase cyclase (AOC), was induced by 2.3- and 1.5-fold, respectively, whereas oxophytodienoate-reductase 3 (OPR3) expression did not change significantly in CrMYC2 overexpression lines compared with the EV control.
To determine whether repression of key regulatory and structural genes in CrMYC2-OX hairy roots has any effect on the metabolic outcome, we measured the alkaloid contents in the EV control and CrMYC2-OX hairy roots. Before quantifying the amounts of different alkaloids, we performed TLC. TLC plates were initially visualized under long and short wave UV light, and compared with catharanthine, tabersonine, and ajmalicine standards. Authentication of metabolites was conducted using the stain ceric (IV) ammonium sulfate, which produces different colorimetric reactions with various vinca alkaloids. The TLC results suggested a decreased accumulation of TIAs in CrMYC2-OX hairy roots. For further validation, the samples were then analysed by LC-ESI-MS/MS (Suttipanta et al., 2011). The EV control hairy root lines produced 1027.0, 735.2, and 924.0 (ng mg−1) catharanthine, tabersonine, and ajmalicine, respectively (Fig. 1B). In contrast, the two CrMYC2-OX lines showed significant decrease in the accumulation of catharanthine (391.5 ng mg−1), tabersonine (121.0 ng mg−1), and ajmalicine (155.0 ng mg−1) (Fig. 1B). The transcriptional and metabolic consequences of CrMYC2 overexpression suggest the counter-effects of repressors, such as CrGBFs and ZCTs, in the TIA pathway.
Expression of CrMYC2 and CrGBFs is highly correlated and JA-responsive
CrMYC2 is an activator whereas CrGBF1 and CrGBF2 are repressors of the TIA pathway. We hypothesized that coordinated expression and interactions between activators and repressors control the dynamics of TIA pathway gene expression and alkaloid accumulation in C. roseus. To test our hypothesis, we first analysed the expression profile of CrMYC2 and CrGBFs in different tissues and in response to JA treatment. We performed a co-expression analysis of TIA pathway genes in different tissues including young leaves, mature leaves, roots, and flowers using the C. roseus transcriptomes available through the SRA database (accession number SRA030483). Figure 2A shows two major clusters: the first includes CrMYC2, CrGBFs, ORCAs, and ZCTs; and the other includes BIS1, BIS2, G10H, and genes related to vindoline biosynthesis. We also verified the expression patterns of CrMYC2 and CrGBFs in young and mature leaves, as well as roots, using qRT-PCR (Fig. 2B). Transcript levels of CrMYC2 and CrGBFs were significantly higher in roots compared with leaves (Fig. 2B). JA-responsive expression of structural and regulatory genes is a hallmark of the TIA pathway. We speculated that the co-expressed CrMYC2 and CrGBFs have similar induction patterns in response to JA treatment. We thus examined the expression of CrMYC2 and CrGBFs in C. roseus roots, where they are highly expressed, in a time course of JA treatment using qRT-PCR. The transcript levels of CrMYC2 and CrGBFs were reduced 6 and 12 h following JA treatment, but increased by 5- to 16-fold after 24 h (Fig. 2C). Such biphasic kinetics of expression has also been observed in CrMYC2-regulated genes, such as ORCA2 (Menke et al., 1999). Taken together, our results indicate that the spatial and temporal expression profiles of CrMYC2 and CrGBFs are highly correlated, which is likely to be relevant in modulating expression of TIA regulatory and biosynthetic genes.
Co-expression of CrMYC2 and CrGBFs with TIA pathway genes in different tissues of C. roseus. (A) Hierarchical clustering and heatmap showing that CrGBF1 was clustered with ASα and ZCTs, while CrGBF2 was co-expressed with CrMYC2 and ORCA3, and their targets, STR, TDC, LAMT, and SLS. (B) Relative expression levels of CrMYC2, CrGBF1, and CrGBF2 in young and mature leaves and roots. (C) Relative expression levels of CrMYC2, CrGBF1, and CrGBF2 in roots treated with MeJA for different times. Relative expression levels of target genes were determined by qRT-PCR. RPS9 was used as an internal control. Data represent means ±SD of three replicates. Statistical significance was determined at P<0.05 by Student’s t-test, and is indicated by an asterisk.
RNAi-mediated suppression of CrGBF1 affects TIA pathway gene expression and alkaloid accumulation in C. roseus hairy roots
To determine the role of CrGBF1 in TIA pathway gene expression and metabolite accumulation, we generated transgenic hairy root lines in which CrGBF1 expression was suppressed using RNAi. The transgenic status of hairy root lines was confirmed by PCR (see Supplementary Fig. S2B) and two independent RNAi lines showing 58–72% suppression of CrGBF1 transcripts were chosen for further analysis (Supplementary Fig. S2C). We analysed the expression of TIA pathway genes that were changed significantly in the CrMYC2-OX hairy root lines. Expression of CrMYC2 was decreased by 35–49% in both CrGBF1-RNAi hairy root lines compared with the control (Fig. 3A). Expression of ORCA3 was decreased by 31–57%, whereas BIS1/2 expression was decreased by 50% in CrGBF1-RNAi hairy roots (Fig. 3A). Expression of CrGBF2, ZCT1, ZCT2, and ZCT3, which were up-regulated in CrMYC2-OX hairy roots, was decreased by 42–62% compared with control (Fig. 3A). Expression of iridoid pathway genes such as GES, G10H, CPR, IRS, IO, 7-DLGT, and 7-DLH, which were repressed in the CrMYC2-OX line, was not significantly altered in CrGBF1-RNAi roots (Fig. 3B). Moreover, the expression of TDC was increased by 2.0- to 2.5-fold and STR by 1.2-fold compared with the control (Fig. 3C).
qRT-PCR analysis of TIA pathway gene expression and alkaloid accumulation in empty vector (EV) control and CrGBF1-RNAi hairy root lines. (A) The relative expression levels of TIA pathway genes encoding enzymes and transcriptional regulators in EV control and CrGBF1-RNAi hairy root lines. Ribosomal protein subunit 9 (RPS9) was used as an internal control. For abbreviations of pathway genes, see Supplementary Fig. S1. (B) Accumulation of catharanthine, tabersonine and ajmalicine in two EV control and two independent CrGBF1-RNAi lines. An equal amount of hairy root was used for each line for total alkaloid extraction. Alkaloid extracts from EV control and CrGBF1-RNAi lines were analysed by LC-MS/MS, and the levels of catharanthine, ajmalicine, and tabersonine were estimated based on peak areas. Each sample was tested in triplicate. Data represent means ±standard deviation (SD) of three replicates. Significant differences among different lines were determined at P<0.05 by Student’s t-test, and are indicated by an asterisk.
We also measured the alkaloid contents in the EV control and CrGBF1-RNAi hairy roots using LC-ESI-MS/MS (Suttipanta et al., 2011). The EV control hairy root lines produced 825.5, 606.2 and 797.5 ng mg−1 catharanthine, tabersonine, and ajmalicine, respectively (Fig. 3D), whereas the two CrGBF1-RNAi lines showed significant decrease in accumulation of catharanthine (259.0–448.1 ng mg−1), tabersonine (279.0–432.2 ng mg−1), and ajmalicine (310.2–636.0 ng mg−1) (Fig. 3D).
CrGBF1 and CrGBF2 interact with CrMYC2 in yeast and plant cells
Interfamily TF interactions have been shown to regulate developmental as well as metabolic processes in plants (Patra et al., 2013b; Pattanaik et al., 2014; Bemer et al., 2017). In Arabidopsis, MYC2 and GBF1 (AtbZIP41) interaction regulates hypocotyl elongation (Maurya et al., 2015). CrMYC2 shares high sequence similarity with AtMYC2 in Arabidopsis. CrGBF1 and CrGBF2 share high sequence similarity with Arabidopsis GBF3 (AtbZIP55) and AtbZIP16, respectively (see Supplementary Fig. S4). As CrMYC2 and CrGBF expression was correlated, we decided to test if there was a protein–protein interaction between these two factors.
In a yeast two-hybrid assay, using CrGBF1 or -2 as bait and CrMYC2 as prey, we observed the interaction of CrMYC2 with CrGBF2, but not CrGBF1 (Fig. 4A). We then used a protoplast-based two-hybrid assay (Mitsuda and Ohme-Takagi, 2009; Patra et al., 2013a; Paul et al., 2017) to test the interaction between CrMYC2 and CrGBF1/2. CrMYC2 was fused to the GAL4 DNA binding domain to generate BD-CrMYC2, which significantly activated the expression of a luciferase reporter driven by minimal CaMV 35S promoter with five tandem repeats of GAL4 RESPONSE ELEMENTS (5XGAL4RE-35S) in plant cells, demonstrating CrMYC2 as a strong activator. However, the reporter activity was significantly decreased when BD-CrMYC2 was co-expressed with CrGBF1 or CrGBF2 (Fig. 4B). Because the co-expressed CrGBF1 and -2 lack the GAL4 DNA binding domain, the repression of CrMYC2 activity was likely due to the interaction of CrGBF1 or -2 with CrMYC2. To further confirm the interaction between CrMYC2 and CrGBFs, we also performed a BiFC assay in tobacco protoplasts (Pattanaik et al., 2010a). The EYFP signal was detected in the nucleus when CrMYC2-nEYFP and CrGBF1-cEYFP or CrMYC2-nEYFP and CrGBF2-cEYFP were co-electroporated into tobacco protoplasts. The EYFP signal was not detected when only one of the fusion proteins, CrMYC2-nEYFP or CrGBF1-cEYFP, was co-expressed with the cEYFP or nEYFP EVs (Fig. 4C). Taken together, our results suggest that CrMYC2 directly interacts with CrGBF1 and CrGBF2 in plant cells.
The protein–protein interaction between CrMYC2 and CrGBFs in yeast and plant cells. (A) Yeast two-hybrid assay demonstrates the interaction between CrMYC2 and CrGBF2 proteins. CrMYC2 fused to the GAL4 activation domain (pAD-CrMYC2) was co-transformed with CrGBF1 or CrGBF2 fused to GAL4 DNA binding domain (pBD-CrGBF1/CrGBF2). The transformants were grown on either double (SD−Leu−Trp) selection medium or quadruple (SD−Ade−His−Leu−Trp) selection medium. (B) A protoplast-based two-hybrid assay demonstrates the interactions of CrMYC2 with CrGBF1 and CrGBF2. The reporter plasmid contains the firefly luciferase gene under the control of GAL4 response elements fused to minimal CaMV 35S promoter and rbcS terminator. CrMYC2-BD: CrMYC2 fused to GAL BD. The reporter plasmid was either transformed alone or co-transformed with different effectors into tobacco protoplasts. A plasmid containing the β-glucuronidase (GUS) reporter driven by the CaMV 35S promoter and rbcS terminator was used as a control for normalization. Luciferase activity was normalized against GUS activity. Data represent mean ±SD of three replicates. Significant differences among different combinations were determined at P<0.05 by Student’s t-test, and are indicated by an asterisk. (C) BiFC assay in tobacco protoplasts showing the protein–protein interaction between CrMYC2 and CrGBFs. CrGBF1 and CrGBF2, and CrMYC2 were fused to cEYFP and nEYFP, respectively, and were co-transformed into tobacco protoplasts and visualized using a fluorescence microscope. Nuclei are indicated by DAPI staining. As a negative control, the CrMYC2–nEYFP and an empty cEYFP plasmid were also co-transformed into tobacco protoplasts.
CrGBF1 and CrGBF2 form homo- and heterodimers
The bZIP TFs contain a basic domain, which is involved in DNA binding, and a leucine-zipper domain that mediates homo- and heterodimer formation (Sibéril et al., 2001b). To determine whether the CrGBF proteins form homo- and heterodimers, we performed a yeast two-hybrid assay that used CrGBF1 and CrGBF2 as bait and/or prey. Our results showed that CrGBF1 and CrGBF2 formed homodimers and heterodimers in yeast cells (Fig. 5A). To further validate the interaction observed in yeast cells, we performed a BiFC assay in tobacco protoplasts. A strong EYFP fluorescence signal was only detected in the nucleus when co-expressing CrGBF1-nEYFP and CrGBF1-cEYFP, CrGBF2-nEYFP and CrGBF2-cEYFP, CrGBF1-nEYFP and CrGBF2-cEYFP, or CrGBF2-nEYFP and CrGBF1-cEYFP. In contrast, the EYFP fluorescence signal was not observed when only one of the fusion proteins was co-expressed with the complementary portion of EYFP (Fig. 5B). Our findings suggest that CrGBFs form homo- and heterodimers in plant cells.
The homo-/heterodimerization between CrGBF1 and CrGBF2 in yeast and plant cells. (A) Yeast two-hybrid assay demonstrates homo-/heterodimerization of CrGBF1 and CrGBF2. The plasmid harboring a GAL4 AD/CrGBF1 or CrGBF2 fusion (pAD-CrGBF1/CrGBF2) was co-transformed with the pBD-CrGBF1 or pBD-CrGBF2 plasmids. The transformants were grown in either double (SD−Leu−Trp) or quadruple (SD−Ade−His−Leu−Trp) selection medium. (B) BiFC assay in tobacco protoplasts showing the homo- and heterodimerization of CrGBF1 and CrGBF2. CrGBF1 and CrGBF2 fused with the cEYFP and nEYFP, respectively, were co-transformed into tobacco protoplasts and visualized using a fluorescence microscope. Nuclei are indicated by DAPI staining. As a negative control, the CrGBF1–cEYFP, CrGBF2–cEYFP and an empty nEYFP plasmid were also co-transformed into tobacco protoplasts.
CrGBFs act as repressors of TIA pathway
Significant up-regulation of CrGBF1/2 and down-regulation of several TIA pathway genes in CrMYC2-OX hairy roots led us to hypothesize that CrGBFs regulate expression of a number of regulatory and structural genes in the TIA pathway. A previous study showed that CrGBFs bind to the T/G-box element in STR and TDC promoters in vitro and repress transcriptional activity of STR promoter in plant cells (Sibéril et al., 2001a). To validate the previous observation and to discover additional targets, we cloned the promoter sequences of selected indole and iridoid pathway genes, ORCA3, BIS1, STR, TDC, and G10H, that contain the putative CrGBF binding sites (T/G-box elements) and used a protoplast-based assay to determine the ability of CrGBFs to repress their activities. The ORCA3, BIS1, STR, TDC, or G10H promoters, fused to the firefly luciferase (LUC) reporter gene, were electroporated into tobacco protoplasts alone or with plasmids expressing CrGBF1 and/or CrGBF2. Individually, CrGBF1 or CrGBF2 significantly repressed the transcriptional activities of all tested promoters by 28–85% compared with the control (Fig. 6A). To show CrGBF1 binds to the T/G-box (AACGTG) element, we mutated the AACGTG sequence to AAAATG in ORCA3 and TDC promoters. Transcriptional activities of the mutant promoters of ORCA3 (mORCA3) or TDC (mTDC) were not repressed when co-electroporated with CrGBF1 into tobacco protoplasts (Fig. 6B), suggesting that CrGBF binds the T/G-box element in TIA pathway promoters to repress their expression.
Transcriptional activities of the promoters of key structural and regulatory genes in TIA pathway were repressed by CrGBFs. (A) Protoplast transient assays demonstrate that CrGBF1 and CrGBF2 either individually or in combination repressed transcriptional activities of TDC, STR, G10H, ORCA3, and BIS1 promoters. Schematic diagrams show the T/G-box elements in the promoters of TDC, STR, G10H, ORCA3, and BIS1. The positions of the T/G-box are relative to translation start site of corresponding promoters. TDC, STR, G10H, ORCA3, and BIS1 promoters fused to luciferase reporter were electroporated into tobacco protoplasts either alone or co-transformed with CrGBF1 and/or CrGBF2. A plasmid containing the GUS reporter, controlled by the CaMV 35S promoter and rbcS terminator, was used as a control for normalization. Luciferase activity was normalized against GUS activity. The control represents the reporter alone without effectors. The values represent means ±SD of three biological replicates. Statistical significance was determined at P<0.05 by Student’s t-test and is indicated by an asterisk. (B) Transactivation of the mutated TDC (mTDC) and ORCA3 (mORCA3) promoters by CrGBF1. The mTDC or mORCA3 promoter fused to the luciferase reporter was electroporated alone or with CrGBF1 into tobacco protoplasts. A plasmid containing GUS reporter, controlled by the CaMV 35S promoter and rbcS terminator, was used as a control for normalization.
Compared with CrGBF2, CrGBF1 exhibited stronger repression on all tested promoters, except BIS1. In addition, compared with individual expression of each factor, co-expression of CrGBF1 and CrGBF2 showed various degrees of repression. For example, CrGBF1/2 co-expression resulted in additive repression of STR and BIS1 promoters, whereas additive repression was not apparent on other tested promoters. These results suggest that CrGBF homodimers and heterodimers mediate a wide magnitude of repression on indole and iridoid pathway genes.
CrGBFs antagonize transactivation of CrMYC2 on key TIA pathway gene promoters
To test whether and how the CrMYC2–CrGBF complexes regulate TIA pathway gene expression, we analysed their effects on the promoters of ORCA3 and TDC. We and other groups have shown that CrMYC2 binds the T/G-box element in ORCA3 and TDC promoters to regulate their expression (Zhang et al., 2011; Patra et al., 2018). Recently, we have also demonstrated that CrMYC2 co-regulates TDC with ORCA3 by binding to the T/G box element (Paul et al., 2017). The TDC-LUC and ORCA3-LUC reporter plasmids were electroporated into tobacco protoplasts alone or with plasmids expressing CrMYC2, CrGBF1, or CrGBF2. As expected, CrMYC2 significantly activated the ORCA3 and TDC promoters (Fig. 7A, B). However, the activations were significantly reduced when CrGBF1 and/or CrGBF2 was co-electroporated with CrMYC2 (Fig. 7A, B). The repressive effect was more evident on TDC promoter when both CrGBF1 and CrGBF2 were co-electroporated with CrMYC2 (Fig. 7B). CrMYC2 and CrGBF are shown to bind the T/G-box on TDC and ORCA3 promoter (Fig. 6B) (Sibéril et al., 2001a; Zhang et al., 2011; Paul et al., 2017). Next, we performed a competition assay by increasing the ratio of plasmid expressing CrMYC2 to that of CrGBF1 in the protoplast assay. The CrGBF1-mediated repression of luciferase activities of ORCA3 promoter gradually decreased with increasing CrMYC2 concentration (0-, 2-, and 4-fold that of CrGBF1; Fig. 7A). Similarly, CrMYC2-induced activation of ORCA3 promoter gradually declined with increasing concentrations of CrGBF1 (0-, 2-, and 4-fold that of CrMYC2; Fig. 7A). These findings suggest that CrMYC2 and CrGBF antagonize each other’s activities in a dose-dependent manner by binding to the same cis-elements on a target promoter.
CrGBFs repress the transactivation of CrMYC2 and compete with CrMYC2 for binding to TIA pathway gene promoters in tobacco protoplasts. (A) Protoplast transient assays showed that transcriptional activity of ORCA3 promoter was regulated by CrMYC2 and CrGBF1 in a dose-dependent manner. Plasmid DNA of CrMYC2 (0, 2, 4 µg) was used for co-transformation with CrGBF1. Similarly, 0, 2, 4 μg of CrGBF1 plasmid DNA was co-transformed with CrMYC2. (B) TDC promoter fused to firefly luciferase reporter was electroporated into tobacco protoplasts either alone or co-transformed with different combinations of CrMYC2, CrGBF1, and CrGBF2. (C) STR promoter fused to firefly luciferase reporter was electroporated into tobacco protoplasts either alone or co-transformed with different combinations of CrMYC2, CrGBF1, and CrGBF2. Data represent means ±SD of three biological replicates. Statistical significance was determined at P<0.05 and P<0.01, by Student’s t-test, indicated by * and **, respectively.
To test the possibility of CrGBFs sequestering CrMYC2 by forming a non-DNA binding complex, we tested the STR promoter, which is repressed by CrGBFs (Fig. 6A) but not activated by CrMYC2 (Zhang et al., 2011). We found that the transcriptional activity of STR was significantly repressed individually or in combination by CrGBF1 and CrGBF2 (Fig. 6A); however, the STR-driven luciferase activity was not significantly affected when CrGBF1 or CrGBF2 was co-electroporated with CrMYC2 into tobacco protoplasts (Fig. 7C), suggesting that CrMYC2 interacts with CrGBF1 or CrGBF2 to form a non-DNA binding complex that can attenuate the repression caused by the individual CrGBF. However, transcriptional activity of STR promoter is significantly repressed by the co-electroporation of CrGBF1 and CrGBF2 with CrMYC2 (Fig. 7C), suggesting that the CrGBF heterodimer possesses higher affinity to the STR promoter than the CrGBF homodimers or CrGBF–MYC2 heterodimers.
Discussion
MYC2 regulates biosynthesis of a number of specialized metabolites, including nicotine, flavonoids, glucosinolates, taxol, and steroidal glycoalkaloids, in many plant species (Dombrecht et al., 2007; Shoji and Hashimoto, 2011; Zhang et al., 2012; Schweizer et al., 2013; Lenka et al. 2015; Shen et al., 2016; An et al., 2016; Cárdenas et al., 2016). MYC2 overexpression has contrasting effects in different plant species. MYC2 overexpression leads to increased anthocyanin accumulation in apple calli (An et al., 2016) and artemisinin production in Artemisia annua (Shen et al., 2016). However, MYC2 overexpression significantly represses the expression of paclitaxel pathway genes in Taxus (Lenka et al., 2015) and nicotine pathway genes in tobacco (Wang et al., 2015). The mechanism by which MYC2 restricts biosynthesis of specialized metabolites is insufficiently understood. CrMYC2 is known to activate key TIA pathway genes in C. roseus cell lines. Initially it was perplexing as to why ectopic expression of CrMYC2 in C. roseus hairy roots down-regulated a large number of TIA genes and reduced alkaloid accumulation. Our analysis of CrMYC2 overexpression in C. roseus hairy roots revealed the up-regulation of several repressor genes, significantly among which were CrGBF1 and CrGBF2. In addition, TIA accumulation was reduced in CrMYC2-OX hairy roots compared with the control (Fig. 1B). Our findings are in close agreement with several previous studies that showed that overexpression of an activator, such as ORCA3 (Peebles et al., 2009) or BIS1 (Van Moerkercke et al., 2015), or silencing of a repressor, ZCT1 (Rizvi et al., 2016), is not sufficient to boost TIA production in C. roseus hairy roots. A reasonable interpretation is that a network of repressors (e.g. CrGBFs) acts in concert with the activator (CrMYC2) to prevent over-accumulation of TIAs. We found that the spatial and temporal expression of CrMYC2 and CrGBFs are tightly correlated in different tissues and in response to JA treatment (Fig. 2). Interestingly, we found that repression of CrGBF1 in C. roseus hairy roots altered TIA pathway gene expression and reduced TIA accumulation (Fig. 3). This is likely the result of down-regulation of key transcriptional regulators (Fig. 3A). The exact mechanism by which down-regulation of CrGBF1 leads to repression of several co-expressed regulators remains unclear. In the CrGBF1-RNAi lines, expression of the iridoid branch genes largely remained unchanged (Fig. 3B), and the expression of indole branch gene TDC was up-regulated by approximately 2.5-fold (Fig. 3C). These results suggest that other regulators in the complex TIA regulatory network also respond to the down-regulation of CrGBF1. Nevertheless, our findings suggest that CrMYC2 and CrGBF1/2 jointly modulate TIA pathway gene expression and metabolite accumulation in C. roseus.
Cross-family TF interactions occur frequently in plants and are an important feature of gene regulatory networks (Bemer et al., 2017). In Arabidopsis, of the 2331 TF–TF interactions, 1124 involve cross-family TF interactions (Bemer et al., 2017). The bHLH and bZIP TFs are known to form homo- and heterodimers within the same family; however, the interconnections between these two families are not well characterized (Maurya et al., 2015). Our results showed that CrGBFs interact with CrMYC2 in yeast and plant cells (Fig. 4). Previously, CrGBF1 and CrGBF2 have been shown to bind preferentially to the T/G-box motifs in the STR and TDC promoters (Sibéril et al., 2001a). However, their influence on other TIA pathway genes and the regulatory mechanism were not known. We showed that CrGBF homo- and heterodimers (Fig. 5) repress the transcriptional activities of a wide range of regulatory and structural gene promoters in both the indole and iridoid branches, including those of ORCA3, BIS1, TDC, G10H, and STR. We also showed that mutations in the T/G box of the TDC and ORCA3 promoters affected the repression by CrGBF1 (Fig. 6). CrGBF1 and CrGBF2 repressed the transcriptional activities of the promoters with different intensities, suggesting that the binding of the GBF homo- or heterodimers to the corresponding promoter is influenced by the T/G-box motif and the surrounding DNA sequences (Fig. 6).
Interactions between TFs in response to various biotic and abiotic factors dictate the metabolic outcomes in plants. In Arabidopsis, regulation of flavonoid biosynthesis and trichrome development by a group of activators and repressors is a hallmark of combinatorial gene regulation. The MBW complex, consisting of R2R3MYB (PAP1/PAP2/GL1), bHLH (GL3/EGL3/TT8), and WD40 (TTG1) TFs, positively regulates flavonoid biosynthesis and trichrome development. On the other hand, the R3-MYBs (MYBL2/CPC/TRY) act as negative regulators by forming a repressor complex through competition with the R2R3 MYBs in the MBW complex (Patra et al., 2013b; Pattanaik et al., 2014). The sub-group IIId bHLH factors (bHLH3/13/14/17) also act as negative regulators of anthocyanin biosynthesis and the JA response in Arabidopsis (Song et al., 2013). Unlike R3-MYBs, these bHLH factors do not interact with components of the MBW complex; rather, they antagonize the activity of the MBW complex by binding to the same cis-elements in target promoters. AtMYC2 is a positive regulator of the Arabidopsis wounding- and JA-responsive gene TAT1. The sub-group IIId bHLH factors antagonize the activity of AtMYC2 on the TAT1 promoter by binding to the same cis-elements (Song et al., 2013). Recently, we identified a C. roseus sub-group IIId bHLH factor, RMT1, to be a negative regulator of TIA biosynthesis. RMT1 antagonizes CrMYC2 activity by binding to the same cis-elements on the target TIA gene promoters (Patra et al., 2018). In Arabidopsis, the interaction between AtMYC2 and AtGBF1 forms a non-DNA binding complex that is unable to repress the transcriptional activity of the promoter of HYH, a gene involved in photomorphogenic growth of seedlings (Maurya et al., 2015). Here, we showed that CrGBF1 antagonizes the transactivation of the ORCA3 promoter by CrMYC2 in a dose-dependent manner. Similarly, CrMYC2 was able to overcome the CrGBF1-mediated repression of the ORCA3 promoter in a dose-dependent manner. Furthermore, addition of CrMYC2 overcame the repression of the STR promoter mediated by homodimers of either CrGBF1 or -2, but not that mediated by the CrGBF1/2 heterodimer (Fig. 7C).
Based on our findings, we propose a further refined model that illustrates a balanced regulatory network controlling TIA biosynthesis in C. roseus. CrMYC2 activates ORCA3 (and other ORCA-cluster members) which, in turn, activates the expression of downstream genes, such as TDC and STR. CrMYC2 and ORCA3 also co-regulate TDC. BIS1 and -2 are key activators of the iridoid pathway genes, such as G10H. CrMYC2 and BIS1 activate CrGBFs, which, in turn, repress the activators (e.g. ORCA3 and BIS1), as well as their target genes in both the indole and iridoid branches of the TIA pathway (Fig. 8A). CrGBF1 antagonizes CrMYC2 activity by two possible mechanisms: (i) binding to the same cis-elements (T/G box) in the target promoters as a homo- or heterodimer, and/or (ii) forming a non-DNA-binding CrGBF–MYC2 complex that is likely to be transcriptionally inactive (Fig. 8B).
A model depicting CrMYC2–CrGBFs-mediated regulation of TIA biosynthetic pathway in C. roseus. (A) TIA biosynthesis is coordinately regulated by various transcriptional activators and repressors in C. roseus. CrMYC2 and CrGBFs act as a regulatory module that not only controls activators (ORCA3), but also modulates the expression of other key pathway genes and regulators of TIA biosynthesis, such as TDC (indole pathway), BIS1, and G10H (iridoid pathway), and STR (initial TIA biosynthesis pathway). (B) CrGBFs interact to form a homo- or heterodimer and bind to the T/G-box in promoters of TIA pathway. Physical interaction between CrMYC2 and CrGBF inhibits their activity on target promoter possibly due to formation of a non-DNA binding complex. Dashed lines represent regulation that may be direct or indirect. Solid arrows represent transactivation, and T-bars represent repression.
Several hypotheses can be put forth to explain the abundance of repressors that are involved in the fine-tuning of the TIA regulatory circuit. One hypothesis is to buffer cytotoxicity. A recent study has shown that virus-induced gene silencing of a tonoplast transporter of strictosidine causes cell death due to increased accumulation of strictosidine (Payne et al., 2017). Another hypothesis involves the trade-off between growth and defense, which is crucial for growth and survival of plants (Yang et al., 2012; Fan et al., 2014). Production of specialized metabolites, including TIAs, which are known defense compounds, is an energy-intensive process. TIA production, in response to biotic or abiotic signals, likely comes with a trade-off for C. roseus growth and development. Transcriptional repressors likely attenuate TIA pathway hyper-activation and enable regulatory outputs favoring plant growth and development. Our work illustrates that CrBGF1 and CrGBF2 are among the repressors responding to CrMYC2 activation, and provides insights into the underlying mechanisms by which the CrGBFs antagonize CrMYC2 function. The interaction between the activators and repressors in the TIA pathway reported here exemplifies yet another layer of control that modulates gene expression and fine-tunes metabolite accumulation in C. roseus.
Supplementary data
Supplementary data are available at JXB online.
Fig. S1. Terpenoid indole alkaloid (TIA) biosynthetic pathway in Catharanthus roseus.
Fig. S2. Schematic representation of expression vectors used for generating transgenic hairy roots and the verification of their transgenic status.
Fig. S3. Co-expression analysis of TIA pathway genes and regulators in empty vector (EV) control and CrMYC2 overexpression (CrMYC2-OX) hairy root lines.
Fig. S4. Sequence alignment of MYC2 and GBF proteins from Catharanthus and Arabidopsis.
Table S1. Primer sequences used for qRT-PCR in this study.
Table S2. Primer sequences used for gene cloning and checking transgenic status.
Acknowledgements
We express our gratitude to Drs A. Hunt, G. Hao, and M. Chakrabarti (Department of Plant and Soil Sciences, University of Kentucky) for assistance in RNA-seq library preparation. We thank Mr J. May (Department of Civil Engineering and Environmental Research Training Laboratories, University of Kentucky) for assistance in LC-MS/MS. This research was supported by a graduate scholarship to XS from the Department of Plant and Soil Sciences, University of Kentucky. This work is supported partially by the Harold R. Burton Endowed Professorship to LY and by the National Science Foundation under Cooperative Agreement no. 1355438 to LY.
Comments