Clade IVa Basic Helix–Loop–Helix Transcription Factors Form Part of a Conserved Jasmonate Signaling Circuit for the Regulation of Bioactive Plant Terpenoid Biosynthesis

Plants produce many bioactive, specialized metabolites to defend themselves when facing various stress situations. Their biosynthesis is directed by a tightly controlled regulatory circuit that is elicited by phytohormones such as jasmonate (JA). The basic helix–loop–helix (bHLH) transcription factors (TFs) bHLH iridoid synthesis 1 (BIS1) and Triterpene Saponin Activating Regulator (TSAR) 1 and 2, from Catharanthus roseus and Medicago truncatula, respectively, all belong to clade IVa of the bHLH protein family and activate distinct terpenoid pathways, thereby mediating monoterpenoid indole alkaloid (MIA) and triterpene saponin (TS) accumulation, respectively, in these two species. In this study, we report that promoters of the genes encoding the enzymes involved in the specific terpenoid pathway of one of these species can be transactivated by the orthologous bHLH factor from the other species through recognition of the same cis-regulatory elements. Accordingly, ectopic expression of CrBIS1 in M. truncatula hairy roots up-regulated the expression of all genes required for soyasaponin production, resulting in strongly increased levels of soyasaponins in the transformed roots. Likewise, transient expression of MtTSAR1 and MtTSAR2 in C. roseus petals led to up-regulation of the genes involved in the iridoid branch of the MIA pathway. Together, our data illustrate the functional similarity of these JA-inducible TFs and indicate that recruitment of defined cis-regulatory elements constitutes an important aspect of the evolution of conserved regulatory modules for the activation of species-specific terpenoid biosynthesis pathways by common signals such as the JA phytohormones.

Introduction

A plethora of bioactive, specialized metabolites are produced by plants as a protective measure against environmental adverse conditions such as herbivore and pathogen predation. One prominent class of such compounds, exhibiting broad structural diversity and bioactivities, are the terpenoids. Many plants produce a species-specific compendium of terpenoids, which are procured in specific biosynthesis pathways, the activation of which is tightly regulated by downstream regulators that act at the endpoint of a complex regulatory circuit (De Geyter et al. 2012, Zhou and Memelink 2016). These circuits are often conserved across the plant kingdom and are activated by phytohormones such as the jasmonates (JAs) that are perceived by a conserved co-receptor complex that transmits the signal to a conserved primary regulatory module (Goossens et al. 2016). Gradually, some of the downstream regulators have also been characterized. In this respect, one recurring family of transcription factors (TFs) are the basic helix–loop–helix (bHLH) TFs. Relevant examples are the recently identified homologous bHLH iridoid synthesis 1 (BIS1) and BIS2 in Catharanthus roseus (Madagascar periwinkle) and the homologous Triterpene Saponin Activating Regulator 1 (TSAR1) and TSAR2 in the model legume Medicago truncatula (barrel medic) (Van Moerkercke et al. 2015, Mertens et al. 2016, Van Moerkercke et al. 2016). These TFs constitute paramount activators of two plant- and species-specific terpenoid biosynthesis pathways: those for the monoterpenoid indole alkaloids (MIAs) and the triterpene saponins (TSs), respectively.

Terpenoids are assembled from individual 5-carbon isopentenyl pyrophosphate (IPP) units. In higher plants, IPP is delivered by two independent pathways: the cytosolic mevalonate (MVA) pathway and the plastidial 2-C-methyl-d-erythritol 4-phosphate (MEP) pathway (Vranová et al. 2013, Arendt et al. 2016). Terpenoids not only function as specialized metabolites in plant defense, but also play key roles in primary metabolism, because IPP is incorporated into several other relevant molecules such as phytohormones, membrane sterol lipids and photosynthetic pigments. Hence, a clear subcellular compartmentalization and a strict regulation of the IPP fluxes in the cell are pivotal (Vranová et al. 2013, Arendt et al. 2016). IPP used in monoterpenoid biosynthesis is generally supplied by the MEP pathway, whereas triterpenoid synthesis is usually fed with IPP originating from the MVA pathway (Moses et al. 2013).

The MIAs in C. roseus comprise the valuable cancer chemotherapeutics vinblastine and vincristine, for which C. roseus is the sole source (Almagro et al. 2015, Duge de Bernonville et al. 2015). MIAs are derived from strictosidine, which is the condensation product of the monoterpenoid compound secologanin, a seco-iridoid, and the indole compound tryptamine (Fig. 1). It was recently shown that CrBIS1 and CrBIS2 specifically and exclusively activate the same set of monoterpenoid biosynthesis genes up to the iridoid loganic acid in a JA-responsive manner (Van Moerkercke et al. 2015, Van Moerkercke et al. 2016). Other steps in the pathway, including the two-step conversion of loganic acid to secologanin, the one-step conversion of tryptophan to tryptamine and its subsequent condensation with secologanin to yield strictosidine, as well as several of the elucidated downstream steps are governed by other TFs, i.e. from the ethylene response factor family, which are also JA inducible like the CrBIS TFs (Zhou and Memelink 2016).

Fig. 1

Schematic overview of the MVA/triterpene saponin and the MEP/monoterpenoid indole alkaloid pathways in Medicago truncatula and Catharanthus roseus, respectively. Genes in turquoise are up-regulated by TSAR1 and TSAR2. Genes marked in blue and green are specifically activated by TSAR1 and TSAR2, respectively. Genes in orange and lilac boxes are regulated by BIS1 and ORCA3, respectively. Dashed arrows represent multiple enzymatic steps. AACT, acetyl-CoA acetyltransferase; BAS, β-amyrin synthase; CMK, CM kinase; CMS, 4-(cytidine 5’-diphospho)-2C-methyl-d-erythritol (CM) synthase; CYP, Cyt P450-dependent monooxygenase; 7DLGT, 7-deoxyloganetic acid glucosyl transferase; 7DLH, 7-deoxyloganic acid hydroxylase; DXR, 1-deoxy-5-xylulose-5-phosphate reductase; DXS, 1-deoxyxylulose 5-phosphate synthase; FPPS, farnesyl pyrophosphate synthase; GES, geraniol synthase; G8O, geraniol-8-oxidase; G3P, glyceraldehyde 3-phosphate; GPPS, geranyl pyrophosphate synthase; HDR, HD reductase; HDS, hydroxymethylbutenyl 4-diphosphate (HD) synthase; 8HGO, 8-hydroxygeraniol oxidoreductase; HMGS, hydroxymethylglutaryl-CoA synthase; HMGR, 3-hydroxy-3-methylglutaryl-CoA reductase; IDI, isopentenyl diphosphate Δ-isomerase; IO, iridoid oxidase; IPP, isopentenyl pyrophosphate; IS, iridoid synthase; LAMT, loganic acid O-methyltransferase; MECS, 2C-methyl-d-erythritol synthase; MEP, 2-C-methyl-d-erythritol 4-phosphate; MVA, mevalonate; MVK, MVA kinase; MVD, MVA diphosphate decarboxylase; PMK, phosphomevalonate kinase; SGD, strictosidine-β-glucosidase; SLS, secologanin synthase; STR, strictosidine synthase; SQE, squalene epoxidase; SQS, squalene synthase; TDC, tryptophan decarboxylase; UGT, UDP-dependent glycosyltransferase.

Medicago truncatula in turn provides a myriad of oleanane-type TSs that display an array of biological activities including anticarcinogenic, cholesterol-lowering, insecticidal and antimicrobial effects (Pollier and Goossens 2012, Gholami et al. 2014, Moses et al. 2014). These TSs can be subdivided into two classes: the hemolytic saponins in which C-28 is fully oxidized to a carboxyl group, and the non-hemolytic soyasaponins that carry a hydroxyl substituent at C-24 (Gholami et al. 2014). The TFs MtTSAR1 and MtTSAR2 specifically boost non-hemolytic and hemolytic TS levels, respectively, when overexpressed in M. truncatula hairy roots. Both MtTSARs induce the MVA pathway genes and all consecutive genes involved in the generation of the β-amyrin backbone, after which they activate only the respective branch-specific Cyt P450 oxidase and UDP-dependent glycosyltransferase (UGT) enzymes leading to non-hemolytic and hemolytic TSs (Fig. 1).

The CrBIS and MtTSAR TFs belong to clade IVa of the bHLH superfamily (Van Moerkercke et al. 2015, Mertens et al. 2016, Van Moerkercke et al. 2016). As bHLH TF sequences display a broad sequence diversity, they are categorized based upon the sequence homology between their signature bHLH motifs that comprise about 60 amino acids (Heim et al. 2003, Toledo-Ortiz et al. 2003, Carretero-Paulet et al. 2010, Pires and Dolan 2010). The N-terminal region of this domain consists of approximately 15 mainly basic amino acids, and is responsible for the specific recognition of promoter cis-regulatory elements (CREs). Approximately 45 amino acids located at the C-terminus form two amphipathic α-helices that are connected by a variable loop region and engage in the formation of homo- or heterodimeric protein complexes. MtTSAR1 and MtTSAR2 exert their activity through binding of a hexanucleotide N-box [CACG(A/C)G] in the promoter region of their target genes. It is generally known that many bHLH TFs recognize E-box sequences (CANNTG) or variants of it. In planta, recognition of N-boxes has only been shown for the TSARs of M. truncatula, and two bHLH TFs from the monocot Oryza sativa (rice), i.e. diterpenoid phytoalexin factor (DPF) and phytochrome-interacting factor protein PIF14 (Yamamura et al. 201, Cordeiro et al. 2016, Mertens et al. 2016). To date, the MtTSAR and CrBIS TFs are the only identified specific regulators of specialized terpenoid biosynthesis that reside together in one subclade. In this study, we show that these homologous TFs recognize the same CREs in the respective species-specific gene promoters and that they are functionally equivalent when ectopically expressed in the respective host plants.

Results

CrBIS1 and MtTSAR1 recognize the same CREs in the promoters of M. truncatula TS biosynthesis genes

We expanded our recently published phylogenetic tree of plant bHLH TFs (Mertens et al. 2016) with the recently identified C. roseus BIS and the rice DPF TFs (Supplementary Fig. S1) to illustrate the phylogenetic relationship between all known bHLH-type TFs that are involved in terpenoid biosynthesis and their relative position among the closest Arabidopsis thaliana clades. This analysis showed that OsDPF does not belong to clade IVa with the CrBIS and MtTSAR TFs, but to the distinct clade IVd instead. Considering that the MtTSAR and CrBIS genes are JA inducible, we were wondering how conserved their role and mode of action in the regulation of JA-responsive terpenoid pathways in the plant kingdom is. The clade IVa bHLH TFs MtTSAR1 and MtTSAR2 have been shown to transactivate TS gene promoters by binding to an N-box (Mertens et al. 2016), whereas CRE specificity for the CrBIS TFs has not been established yet. Hence, if the homologous CrBIS and MtTSAR TFs would act through a conserved mechanistic mode of action, we reasoned that CrBIS1 would also transactivate the promoter of the M. truncatula gene 3-hydroxy-3-methylglutaryl-CoA reductase 1 (MtHMGR1), encoding the enzyme catalyzing the rate-limiting step in MVA synthesis. To assess this, we assessed the transactivation of a reporter construct consisting of the fusion between a minimal MtHMGR1 promoter, proMtHMGR1[−101, −281], that includes an N-box, and the firefly LUCIFERASE (fLUC) gene in tobacco protoplasts (Supplementary Fig. S2). Co-transfection of protoplasts with this reporter construct and a Cauliflower mosaic virus 35S promoter (proCaMV35S)-driven CrBIS1 effector plasmid demonstrated a 17-fold increase of luciferase activity when compared with a control transfection (Fig. 2). Notably, when the N-box of proMtHMGR1[−101, −281] was substituted with TGAATT, the transactivation by CrBIS1 was completely abolished (Fig. 2; Supplementary Fig. S2), indicating that, like the MtTSARs, CrBIS1 can mediate transactivation through an N-box. Also the more recently discovered CrBIS2 strongly transactivated proMtHMGR1 by 25-fold (Supplementary Fig. S3). Such transactivation activity was not observed for bHLH TFs that are co-regulated with the CrBIS and MtTSAR TFs and belong to bHLH clades other than IVa, such as MtbHLH35 and CrbHLH13 (Supplementary Fig. S3), thus supporting the specific activity of the CrBIS and MtTSAR TFs.

Fig. 2

CrBIS1 transactivates the M. truncatula proHMGR1 through the N-box in transfected N. tabacum protoplasts. proMtHMGR1[−101, −281] encompasses the nucleotide stretch starting at base pair 101 and ending at base pair 281 upstream of the translational start site of MtHMGR1 (Mertens et al. 2016). proMtHMGR1[−101, −281]mut represents the same fragment but with a replacement of the N-box 5’-CACGAG-3’ with 5’-TGAATT-3’ (Mertens et al. 2016). Values on the y-axis are normalized fold changes relative to protoplasts co-transfected with the reporter constructs and the proCaMV35S:GUS control plasmid (black). The error bars indicate the SEM (n = 8). The statistical significance was calculated by a Student’s t-test to compare with the control (black asterisk) and to compare proMtHMGR1[−101, −281] with proMtHMGR1[−101, −281]mut (orange asterisk) (***P < 0.001).

Next, we assessed whether CrBIS1 could transactivate the same range of TS gene promoters as the MtTSAR TFs and whether it would display specificity for one of the TS pathway branches, namely the non-hemolytic or the hemolytic TSs. β-Amyrin is the common precursor for both branches (Fig. 1). The first committed step in the non-hemolytic branch, i.e. the oxidation of C-24 on β-amyrin, is performed by MtCYP93E2 (Fukushima et al. 2013). MtCYP72A67, which acts in the non-hemolytic branch, catalyzes the hydroxylation of C-2 (Biazzi et al. 2015). In contrast to the CrBIS proteins that activate the same set of genes (Van Moerkercke et al. 2016), MtTSAR1 and MtTSAR2 display overlapping but different functionalities (Mertens et al. 2016). Indeed, whereas MtTSAR1 mainly directs the conversion of β-amyrin into the non-hemolytic branch, MtTSAR2 exclusively induces the production of hemolytic TSs. Utilizing the gene promoters of the two aforementioned Cyt P450 enzymes as a diagnostic, we observed strong transactivation by CrBIS1 of proMtCYP93E2 by 60-fold, which is comparable with the transactivation effect caused by MtTSAR1 (Fig. 3; Supplementary Fig. S3). The proMtCYP93E2 harbors two N-boxes and, when both N-boxes were mutated in the minimal promoter proMtCYP93E2[−160, −300], transactivation by CrBIS1 was abolished (Supplementary Fig. S4). Transactivation by CrBIS1 of proMtCYP72A67 was far less pronounced, which is again comparable with the activity of MtTSAR1 (Fig. 3; Supplementary Fig. S3). Conversely, and corroborating previous results (Mertens et al. 2016), MtTSAR2 exhibits the exact opposite pattern (Fig. 3).

Fig. 3

CrBIS1 mimics MtTSAR1 behavior in promoter transactivation assays. proMtCYP93E2 and proMtCYP72A67 comprise 1,000 and 1,500 bp upstream of the start codon of MtCYP93E2 and MtCYP72A67, respectively (Mertens et al. 2016). Values on the y-axis are normalized fold changes relative to protoplasts co-transfected with the reporter constructs and the control plasmid (black). The error bars indicate the SEM (n = 4 for proMtCYP93E2; n = 8 for proMtCYP72A67). The statistical significance was calculated with a Student’s t-test (*P < 0.05, **P < 0.01, ***P < 0.001).

CrBIS1 overexpression increases the accumulation of soyasaponins in M. truncatula hairy roots

To establish CrBIS1 further as a functional ortholog of MtTSAR1, we ectopically expressed CrBIS1 in stably transformed M. truncatula hairy roots. Transformation of the generated lines was confirmed by PCR on genomic DNA amplifying rolA and the transgene of interest (GUS or CrBIS1; Supplementary Fig. S5).Three independent CrBIS1 expression (CrBIS1^E) lines and three independent control (CTR) lines expressing β-glucuronidase (GUS) were generated and compared by quantitative PCR (qPCR) analysis (Fig. 4). A slight, but significant, increase in MtHMGR1 transcript accumulation levels could be observed. As for the protoplast transactivation assays, we used MtCYP93E2 and MtCYP72A67 as representatives for the non-hemolytic and the hemolytic Cyt P450s, respectively. A prominent effect, i.e. an average 11-fold increase, on MtCYP93E2 transcript levels but no significant effect on the MtCYP72A67 transcripts was observed (Fig. 4). To corroborate further the lack of effect of CrBIS1 on haemolytic TS synthesis genes, we measured the expression of an additional hemolytic Cyt P450, MtCYP716A12, and for this gene the transcript levels also remained unaltered in two of the three CrBIS1^E lines. These data are in accordance with previous data of MtTSAR1-overexpressing hairy roots where the most prominent effect at the transcript level was the elevated level of MtCYP93E2 (Mertens et al. 2016). However, in contrast to MtTSAR1 (Mertens et al. 2016), CrBIS1 overexpression did not affect transcript levels of the β-amyrin synthase (MtBAS) gene (Fig. 4).

Fig. 4

qPCR analysis of the effect of CrBIS1 expression in M. truncatula hairy roots on the transcript levels of TS biosynthesis genes. Three independent control lines (CTR), which harbor a proCaM35S:GUS construct, and three independent CrBIS1^E lines were used. Expression ratios were plotted relative to the normalized CTR-1. The error bars designate the SEM (n = 3). The statistical significance was calculated with a Student’s t-test (*P < 0.05, **P < 0.01, ***P < 0.001).

To assess the effects of CrBIS1 expression in M. truncatula hairy roots on the metabolite content, untargeted metabolite profiling of hairy root extracts by liquid chromatography–mass spectrometry (LC-MS) was performed. A principal component analysis (PCA) of the LC-MS data clearly showed a separation of the CTR and CrBIS1^E hairy root lines (Fig. 5B). Differential peaks between the CTR and CrBIS1^E lines were identified using a partial least squares discriminant analysis (PLS-DA) model that was used to make an S-plot that visualizes the correlation and covariance of all present m/z peaks (Fig. 5C, D). Peaks were considered significant when the absolute covariance was >0.03 and the absolute correlation value >0.8. Only peaks higher in the CrBIS1^E hairy roots added to the observed differences. The metabolites corresponding to the latter peaks were identified based on their MSn spectra and were shown to be non-hemolytic soyasaponins (Fig. 5A; Supplementary Table S1). For instance, soyasaponin I amounts were increased by 4-fold, Rha-Gal-GlcA-Soyasapogenol E levels by 15-fold and dHex-Hex-HexA-dHex-soyasapogenol B by 9.5-fold. The levels of the hemolytic TS did not increase in the CrBIS1^E hairy roots as compared with CTR roots (Fig. 5A), further confirming the specificity of CrBIS1 for the non-hemolytic TS pathway. Therefore, we can conclude that CrBIS1 mimics MtTSAR1 behavior when ectopically expressed in M. truncatula.

Fig. 5

Ectopic expression of CrBIS1 in M. truncatula hairy roots leads to increased accumulation of non-hemolytic saponins. (A) Average total ion current (TIC) of peaks corresponding to TS. Three independent control (CTR) and CrBIS1^E lines were incorporated with five or four technical repeats. The error bars designate the SEM (with variable n, depending on whether peaks could be measured in the different samples). The statistical significance was calculated with a Student’s t-test (*P < 0.05, **P < 0.01, ***P < 0.001). (B) PCA of CTR and CrBIS1^E roots. (C) PLS-DA of CTR and CrBIS1^E roots. (D) S-plot for correlation (p(corr)[1]) and covariance (w*c[1]) derived from PLS-DA. Metabolites in the bottom left and top right quadrants (marked by dotted red lines) are significantly higher and lower, respectively, in abundance in the CrBIS1^E roots.

MtTSAR1 and MtTSAR2 transactivate C. roseus monoterpenoid gene promoters

Having established orthologous activity of CrBIS1 in M. truncatula, we subsequently assessed the inverse situation. Therefore, we first examined whether the MtTSAR proteins can transactivate C. roseus MIA gene promoters that are transactivated by CrBIS1. Indeed, in tobacco protoplasts, both MtTSAR1 and MtTSAR2 could transactivate the promoters of the geraniol-8-oxidase (proCrG8O) and iridoid synthase (proCrIS) genes (Fig. 6; Supplementary Fig. S6), suggesting that the MtTSARs can be functional equivalents of CrBIS1. In contrast to the situation in M. truncatula, where CrBIS1 mimics MtTSAR1 more, for the C. roseus gene promoters, MtTSAR2 appeared to be more equivalent to CrBIS1, at least in the transient expression assay in tobacco protoplasts.

Fig. 6

MtTSAR1 and MtTSAR2 transactivate the C. roseus G8O promoter in transient expression assays in N. tabacum protoplasts. Values on the y-axis are normalized fold changes relative to protoplasts co-transfected with the reporter constructs and the proCaMV35S:GUS control plasmid (black). proCrG8O comprises 533 bp upstream of the translational start of CrG8O (Van Moerkercke et al. 2015). The error bars designate the SEM (n = 4). Statistical significance was calculated by a Student’s t-test (**P < 0.01, ***P < 0.001).

Like proMtHMGR1, proCrIS contains an N-box (Van Moerkercke et al. 2016). Hence, we tested whether this box would be required to mediate the transactivation by the MtTSAR and CrBIS TFs. To this end, we created a truncated version of proCrIS that lacks the N-box (proCrISdel) and a mutated version in which the N-box is substituted for CCATGG (proCrISmut; Supplementary Fig. S2). Transactivation of proCrISdel by either CrBIS1, MtTSAR1 or MtTSAR2 was markedly reduced compared with that of proCrIS (Fig. 7). When only the N-box was mutated, significant reductions in transactivation were observed for MtTSAR1 and MtTSAR2, but not for CrBIS1 (Fig. 7). Together, this indicates that as for the TS genes in M. truncatula, the transactivation of iridoid genes in C. roseus by clade IVa bHLH TFs involves recognition of the N-box element. Notably, however, transactivation of proCrIS by CrBIS1 and the MtTSARs was not completely abolished by deletion of the N-box, in contrast to proMtHMGR1 (Fig. 2; Mertens et al. 2016), which may suggest that additional CREs recognized by the CrBIS and MtTSAR TFs may still be present in proCrISdel and proCrISmut. However, we could not readily detect any apparent additional N-box-like sequence in the proCrIS sequence; hence, this was not further investigated.

Fig. 7

MtTSAR1 and MtTSAR2 transactivate the C. roseus IS promoter in transient expression assays in N. tabacum protoplasts. Values on the y-axis are normalized fold changes relative to protoplasts co-transfected with the reporter constructs and the GUS control plasmid (black). proCrIS comprises 1,479 bp upstream of the translational start codon of CrIS (Van Moerkercke et al. 2015). proCrISdel corresponds to a truncated version of proCrIS that lacks the N-box, whereas in proCrISmut the N-box is replaced by 5’-CCATGG-3’ (Supplementary Fig. S2). The error bars designate the SEM (n = 4). Statistical significance was calculated by a Student’s t-test to compare with the control (black asterisks) and to compare proCrIS with proCrISdel and proCrISmut (colored asterisks) (*P < 0.05, **P < 0.01, ***P < 0.001).

Expression of MtTSAR1 and MtTSAR2 in C. roseus petal limbs activates monoterpenoid genes

To assess whether the MtTSARs act similarly to CrBIS1 in positively regulating iridoid biosynthesis genes in C. roseus in planta, we transiently expressed MtTSAR1 and MtTSAR2 in C. roseus petal limbs via agroinfiltration. Previously, this had already been successfully carried out with CrBIS2, showing that CrBIS2 overexpression resulted in increased transcript accumulation of MEP and iridoid pathway genes, but not of other MIA genes or triterpenoid genes (Van Moerkercke et al. 2016). CrBIS1 expression in C. roseus petals essentially mimicked the effect of CrBIS2 observed previously (Van Moerkercke et al. 2016), as illustrated by the strongly increased levels of CrG8O, iridoid oxidase (CrIO) and CrIS transcripts, and the lack of an effect on loganic acid O-methyltransferase (CrLAMT), strictosidine-β-glucosidase (CrSGD) and tryptophan decarboxylase (CrTDC) transcripts (Fig. 8). As in the tobacco protoplast assays, the MtTSARs also positively affected iridoid gene expression. CrG8O transcript levels increased on average 25- and 404-fold by ectopic MtTSAR1 and MtTSAR2 expression, respectively (Fig. 8). Likewise, CrIO transcript levels increased 10- and 79-fold by the expression of MtTSAR1 and MtTSAR2, respectively. CrIS transcript levels increased tremendously, by 980-fold and 11,297-fold, when MtTSAR1 and MtTSAR2 were expressed, respectively. Conversely, no such pronounced changes in the CrLAMT, CrSGD and CrTDC transcript levels could be observed by ectopic expression of MtTSAR1 and MtTSAR2 in flower petals (Fig. 8). Together, these data indicate that both MtTSARs can activate the same set of genes as the CrBIS TFs in C. roseus, although MtTSAR2 was clearly more active than MtTSAR1. In addition, we investigated whether the MtTSARs could affect expression of the MEP pathway as was described for CrBIS1 (Van Moerkercke et al. 2015). Transcript levels of several but not all tested MEP pathway genes were indeed significantly increased by MtTSAR expression (Supplementary Fig. S7). Again, MtTSAR2 caused a markedly stronger effect, suggesting that it displays more functional similarity to CrBIS1 than MtTSAR1.

Fig. 8

qPCR analysis showing the effects of ectopic BIS1, TSAR1 and TSAR2 expression on the transcript levels of MIA biosynthesis genes in agroinfiltrated C. roseus petals. (A) qPCR analysis of the ectopically expressed TFs. (B) qPCR analysis of MIA genes. Three independent flower infiltrations were carried out with each effector construct as well as a control (CTR) proCaM35S:GUS construct. Expression ratios were plotted relative to the normalized CTR-1. The error bars designate the SEM (n = 3). The statistical significance was calculated with a Student’s t-test (*P < 0.05, **P < 0.01, ***P < 0.001).

When ectopically expressed, CrBIS1 and the MtTSARs regulate distinct IPP-providing pathways as they do in their ‘own’ species

Catharanthus roseus produces oleanane- and ursane-type triterpenoids in the cuticular wax layer of the leaves (Murata et al. 2008, Huang et al. 2012). Unlike the situation in M. truncatula, however, triterpenoid production in C. roseus is not subjected to JA regulation (Van Moerkercke et al. 2013). Accordingly, it has been shown that the CrBIS TFs did not affect the expression of MVA or triterpenoid pathway genes, such as CrHMGR (Van Moerkercke et al. 2015, Van Moerkercke et al. 2016). Considering that the MtTSARs strongly regulate MtHMGR expression in M. truncatula (Mertens et al. 2016), we investigated whether they would, like CrBIS1, also ‘lose’ their capacity to activate CrHMGR expression in C. roseus. Indeed, the transcript levels corresponding to the CrHMGR gene in C. roseus remained unaltered when either MtTSAR1 or MtTSAR2 was ectopically expressed in C. roseus flower petals (Supplementary Fig. S8A). Vice versa, ectopic CrBIS1 expression in stably transformed M. truncatula hairy roots did not affect the MEP pathway genes encoding 1-deoxy-5-xylulose-5-phosphate reductase (MtDXR) and (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate synthase (MtHDS) (Supplementary Fig. S8B). Together, these data further support the orthologous behavior of the CrBIS and MtTSAR TFs when expressed in their heterologous plant hosts.

Discussion

CrBIS and MtTSAR TFs can be considered as orthologs despite their distinct targets in their species of origin

With 162 identified members in Arabidopsis, the bHLH TFs comprise the second largest TF superfamily in plants (Bailey et al. 2003). Subclade IVa harbors the JA-inducible M. truncatula TSAR and C. roseus BIS TFs, which both positively govern their respective JA-inducible species-specific terpenoid biosynthesis pathways, i.e. of TSs and iridoids, respectively. In this study, we established that these homologous TFs recognize the same CREs and exert the same function as their counterparts, when ectopically expressed in the other species. MtTSAR1 and particularly MtTSAR2 largely phenocopied CrBIS1 when expressed in C. roseus flowers, whereas CrBIS1 behaved like a functional copy of MtTSAR1 and boosted soyasaponin levels when expressed in M. truncatula hairy roots. Hence, CrBIS and MtTSAR TFs can be considered as orthologs in the JA signaling cascade despite the fact that they activate distinct terpenoid pathways in the species of their origin. This is a remarkable finding considering that C. roseus and M. truncatula are quite distantly related, with M. truncatula belonging to the Fabaceae family from the Rosid clade and C. roseus to the Apocynaceae family from the Asterid clade.

Clade IVa bHLH TFs are part of a conserved regulatory module for JA-inducible species-specific terpenoid biosynthesis

From our findings, we can infer that both C. roseus and M. truncatula have implemented an analogous, probably evolutionarily conserved, JA signaling circuit to control their respective defensive terpenoid production. Our findings suggest that clade IVa bHLH TFs are an integral part of such a conserved circuit, as well as clade IIIe bHLH TFs such as MYC2 and its homologs that are well-established members of the primary JA signaling module that is conserved across the plant kingdom, including the gymnosperms (Kazan and Manners 2013, Goossens et al. 2016). Given the fact that plant specialized metabolism is orchestrated mainly at the transcriptional level by environmental cues that involve JA signaling, the involvement of such conserved TF modules is plausible (De Geyter et al. 2012).

Both clade IVa and IIIe bHLH TFs recognize E-boxes, the latter with high affinity for G-boxes (CACGTG) (Fernández-Calvo et al. 2011, Mertens et al. 2016). Protein binding arrays have shown that MtTSARs bind in vitro to CACGHG in which H can be C, A or T, i.e. thus a G-box or an N-box (Mertens et al. 2016). Accordingly, all the promoters of the C. roseus iridoid synthesis genes contain either a G- or an N-box (Van Moerkercke et al. 2016), whereas all the promoters of the known M. truncatula TS synthesis genes, except for MtCYP716A12, harbor at least one N-box (Mertens et al. 2016). However, E-boxes and their variants can be found in many gene promoters that are not necessarily regulated by JAs, and therefore different layers of specificity must be imposed on TF–CRE interaction. This is well illustrated by the fact that the strictosidine synthase gene promoter, which contains a G-box (Pasquali et al. 1999), cannot be transactivated by the CrBIS TFs (Van Moerkercke et al. 2015, Van Moerkercke et al. 2016).

Our transient expression assays in tobacco protoplasts illustrate clearly that the same CREs are recognized by the MtTSARs and CrBIS1 in planta. In addition to the conserved bHLH domains, no additional conserved domains or primary sequence similarities could be detected in the clade IVa TFs, in contrast to many of the other clades of the bHLH protein family, including the clade IIIe MYC TFs (Heim et al. 2003, Pires and Dolan 2010). Hence, it is plausible that the target specificity of clade IVa proteins can be attributed to their bHLH domain. However, additional elements are needed to explain, for instance, the specialization of MtTSAR1/CrBIS1 and MtTSAR2 for the non-hemolytic and hemolytic TSs, respectively. Such elements remain elusive to date, but may involve specific post-translational modifications of the TFs, integration of the TFs in distinct protein complexes with distinct CRE specificity, or alternatively, epigenetic effects that alter the nucleotide landscape surrounding the CRE. Such additional regulatory levels have been reported to affect the activity of bHLH TFs (Pires and Dolan 2010, Feller et al. 2011, Pauwels and Goossens 2011, Figueroa and Browse 2012, Goossens et al. 2016).

At the functional level, the clade IIIe MYC TFs exhibit pleiotropic effects on different types of specialized metabolism in different species across the plant kingdom, such as terpenoids and phenolics, but also affect primary metabolism and numerous growth and developmental processes (De Geyter et al. 2012, Kazan and Manners 2013, Goossens et al. 2016, Zhou and Memelink 2016). In addition to being central players in the primary JA signaling cascade, evidence exists that the MYC TFs can directly modulate terpenoid biosynthesis by direct regulation of terpenoid synthesis genes. For instance, in Arabidopsis, MYC2 recognized one out of several E-boxes present in the promoters of sesquiterpenoid synthase genes (Hong et al. 2012). Similar results have been obtained in the medicinal plant Artemisia annua, in which a MYC2 homolog was reported to regulate biosynthesis of the antimalarial sesquiterpenoid artemisinin (Shen et al. 2016). Interestingly, C. roseus MYC2 can activate ORCA3 gene expression, and thereby indirectly induce the up-regulation of the last two steps of the iridoid branch and the indole branch of MIA synthesis, as well as further downstream MIA synthesis (Zhang et al. 2011). In addition to the MtTSAR and CrBIS TFs, the only other bHLHs from clade IVa that have been characterized to some extent are bHLH20/NAI1 and bHLH25 from Arabidopsis, which are involved in the formation of endoplasmic reticulum bodies, a process which is also modulated by JAs (Matsushima et al. 2004), and susceptibility to cyst nematodes (Jin et al. 2011), respectively. Expression of bHLH20/NAI1 and bHLH25 is also JA inducible. However, whether these and/or the two other Arabidopsis clade IVa bHLH TFs (Supplementary Fig. S1) play a role in the regulation of specialized (terpenoid) metabolism has not been investigated yet. Further in-depth functional characterization of the clade IVa TFs from Arabidopsis and/or other plant species will be required to assess to what extent the role of these TFs in JA-responsive terpenoid production is conserved in the plant kingdom.

Materials and Methods

DNA constructs

All DNA constructs were previously generated (Van Moerkercke et al. 2015, Mertens et al. 2016, Van Moerkercke et al. 2016), except for those described here. The proCrISdel reporter construct, which comprises 792 bp upstream of the translational start site, was obtained by PCR using the primer pair listed in Supplementary Table S2. The proCrISdel PCR amplicon was recombined into pDONR207, sequence-verified and subsequently recombined into the destination vector pGWL7. proCrISmut was obtained by introducing a mutation in proCrIS on the pGWL7 plasmid. The mutation was generated using the primer pair described in Supplementary Table S2 and the gene tailor site-directed mutagenesis system (Invitrogen). CrbHLH13 and MtbHLH35 open reading frames were PCR amplified and recombined in the donor plasmid pDONR207 (Supplementary Table S2). The entry clones were sequenced and recombined in the destination plasmid p2GW7 required for the transient expression assays in tobacco protoplasts.

Transient expression assays in N. tabacum protoplasts

Nicotiana tabacum ‘Bright Yellow-2’ protoplasts were generated and co-transfected with three types of plasmids as described (De Sutter et al. 2005, Vanden Bossche et al. 2013). The normalizer plasmid carries Renilla LUCIFERASE (rLUC) driven by proCaMV35S. The effector plasmid harbors proCaMV35S stitched to a TF gene or GUS in the case of the control. The reporter plasmid harbors the specific promoter fragment fused to fLUC. Normalization is carried out by division of the fLUC values by the rLUC values, after which the normalized fLUC values are averaged and plotted relative to the control values. In all experiments we used four or eight biological repeats.

Creation of transgenic M. truncatula hairy roots

Generation of M. truncatula (ecotype Jemalong J5) hairy roots via Agrobacterium rhizogenes-mediated transformation and maintenance of the hairy roots was carried out as described (Pollier et al. 2011). In brief, M. truncatula seedlings were infected with A. rhizogenes (LBA9402/12) carrying the destination vector pK7WG2D, which has GUS or CrBIS1 under the control of proCaMV35S. Seedlings were grown on Murashige and Skoog (MS) medium supplemented with vitamins (MS + VIT) for 1 week, followed by a 1-week growing period on the same medium supplemented with cefotaxime. Roots were cut off from the green parts and grown in a dark room for 10 d in 2 ml of MS + VIT supplemented with 1% sucrose and cefotaxime. Green fluorescent protein (GFP)-positive hairy roots were grown on solid medium for 2.5 weeks, followed by 2.5 weeks on solid medium without cefotaxime. Integration of T-DNA was confirmed by PCR with primers for the rolA and CaMV35S promoter and terminator sequences (Supplementary Table S2). Genomic DNA for PCR was extracted using Edwards’ buffer. To generate the material for metabolomics, hairy roots were incubated in liquid medium (MS + VIT supplemented with 1% sucrose) for 3 weeks in a dark room.

Transient ectopic expression assays in C. roseus flower petals

Fully developed C. roseus cv. Sunstorm plants were used and grown in a greenhouse (21°C) with additional artificial illumination to create a 16 h photoperiod. Flowers were infiltrated as described (Van Moerkercke et al. 2016). In brief, in attached flowers with undehisced anthers, an incision was made perpendicular to the main vein on the adaxial side of each petal limb, and buffered Agrobacterium solution was infiltrated using a 1-ml needleless syringe. Incisions were made so that only the upper layers of the petal limb were damaged, enabling easy entry of the solution into the petal and complete infiltration of the petal limb. Flowers were co-incubated for 3 d prior to sampling. Agrobacterium tumefaciens were grown overnight in liquid selective medium, washed once and resuspended in infiltration buffer (50 mM MES, 2 mM Na₂HPO₄, 1% glucose, 0.1 mM acetosyringone) to an OD₆₀₀ of 0.3.

Transcript profiling

Total RNA was extracted from ground roots and cDNA was generated with the RNeasy Mini Kit (Qiagen) and the iSCRIPT cDNA Synthesis Kit (Bio-Rad), respectively, following the manufacturer’s guidelines. qPCR primers were developed with the Beacon Designer 4 software (Premier Biosoft International) and are listed in Supplementary Table S2. The reaction was conducted with a LightCycler 480 (Roche) and the LightCycler 480 SYBR Green I Master Kit (Roche) following the manufacturer’s instructions. Each reaction was performed in triplicate for M. truncatula and in duplicate for C. roseus. Multiple reference genes were used to calculate relative expression levels with qBase (Hellemans et al. 2007). Normalization of gene expression was conducted using experimentally validated reference genes. For M. truncatula, these are the 40S ribosomal protein S8 (Mt40S) and translation elongation factor 1α (MtELF1α) and for C. roseus the N2227-like family protein (CrN2227) and the SAND family protein (CrSAND) (Pollier et al. 2013, Pollier et al. 2014).

Metabolite profiling

Medicago truncatula hairy roots were harvested and extracted as reported (Pollier et al. 2011), and LC-MS analysis was carried out as described (Mertens et al. 2016). Chromatograms were integrated and aligned with the XCMS package (Smith et al. 2006) in R version 2.6.1. using the following parameter values: xcmsSet (fwhm = 10, max = 300, snthresh = 2, mzdiff = 0.5), group (bw = 8, max = 300), rector (method = loess, family = symmetric). A second grouping was carried out with the same parameter values. The PCA and PLS-DA were performed with the SIMCA-P 11 software package (Umetrics) with Pareto-scaled mass spectrometry data. Peaks with an absolute covariance value >0.03 and an absolute correlation value >0.8 were considered significantly different.

Phylogenetic analysis

The bHLH domains were defined and the group nomenclature was adopted as reported (Heim et al. 2003, Mertens et al. 2016). Alignment of the sequences was done in BioEdit7. Gaps were removed and a Neighbor–Joining tree was created with the MEGA5 software (Tamura et al. 2011). As an amino acid substitution model, the Jones, Taylor and Thornton model was chosen (Jones et al. 1992). The bootstrap analysis was performed with 10,000 replicates.

Supplementary data

Supplementary data are available at PCP online.

Funding

This work was supported by the European Union Seventh Framework Program [FP7/2007–2013 under grant agreement number 613692–TRIFORC]; the Research Foundation Flanders [through pre- and post-doctoral fellowships to J.M. and J.P., respectively]; and the European Molecular Biology Organization [EMBOCOFUND2010] and the European Commission support from Marie Curie Actions [GA-2010-267154] [fellowships to A.V.M.].

Acknowledgments

We thank Annick Bleys for help in preparing the manuscript.

Disclosures

The authors have no conflicts of interest to declare.

Abbreviations

Abbreviations
 
• BAS

  β-amyrin synthase

 
• bHLH

  basic helix–loop–helix

 
• BIS

  bHLH iridoid synthesis

 
• BIS1^E

  BIS1 expression

 
• CaMV 35S

  Cauliflower mosaic virus 35S promoter

 
• CRE

  cis-regulatory element

 
• CTR

  control

 
• DPF

  diterpenoid phytoalexin factor

 
• DXR

  1-deoxy-5-xylulose-5-phosphate reductase

 
• fLUC

  firefly LUCIFERASE

 
• G8O

  geraniol-8-oxidase

 
• GUS

  β-glucuronidase

 
• HDS

  (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate synthase

 
• HMGR1

  3-hydroxy-3-methylglutaryl-CoA reductase 1

 
• IO

  iridoid oxidase

 
• IPP

  isopentenyl pyrophosphate

 
• IS

  iridoid synthase

 
• JA

  jasmonate

 
• MEP

  2-C-methyl-d-erythritol 4-phosphate

 
• LAMT

  loganic acid O-methyltransferase

 
• LC-MS

  liquid chromatography–mass spectrometry

 
• MIA

  monoterpenoid indole alkaloid

 
• MVA

  mevalonate

 
• PCA

  principal component analysis

 
• PLS-DA

  partial least squares discriminant analysis

 
• qPCR

  quantitative PCR

 
• SGD

  strictosidine-β-glucosidase

 
• TDC

  tryptophan decarboxylase

 
• TF

  transcription factor

 
• TS

  triterpene saponin

 
• TSAR

  Triterpene Saponin Activating Regulator

 
• UGT

  UDP-dependent glycosyltransferase