The monolignol pathway contributes to the biosynthesis of volatile phenylpropenes in flowers

Authors


Summary

  • Volatile phenylpropenes play important roles in the mediation of interactions between plants and their biotic environments. Their biosynthesis involves the elimination of the oxygen functionality at the side-chain of monolignols and competes with lignin formation for monolignol utilization. We hypothesized that biochemical steps before the monolignol branch point are shared between phenylpropene and lignin biosynthesis; however, genetic evidence for this shared pathway has been missing until now.
  • Our hypothesis was tested by RNAi suppression of the petunia (Petunia hybrida) cinnamoyl-CoA reductase 1 (PhCCR1), which catalyzes the first committed step in monolignol biosynthesis. Detailed metabolic profiling and isotopic labeling experiments were performed in petunia transgenic lines.
  • Downregulation of PhCCR1 resulted in reduced amounts of total lignin and decreased flux towards phenylpropenes, whereas internal and emitted pools of phenylpropenes remained unaffected. Surprisingly, PhCCR1 silencing increased fluxes through the general phenylpropanoid pathway by upregulating the expression of cinnamate-4-hydroxylase (C4H), which catalyzes the second reaction in the phenylpropanoid pathway.
  • In conclusion, our results show that PhCCR1 is involved in both the biosynthesis of phenylpropenes and lignin production. However, PhCCR1 does not perform a rate-limiting step in the biosynthesis of phenylpropenes, suggesting that scent biosynthesis is prioritized over lignin formation in petals.

Introduction

Flowers emit a prodigious amount of volatile organic compounds (VOCs), which notably serve as mediators in plant–pollinator interactions and pathogen/florivore defense (Raguso, 2008) and originate from a few biosynthetic pathways. After terpenoids, the second largest class of plant VOCs consists of phenylpropanoid volatiles, which are derived from the essential amino acid phenylalanine (Knudsen et al., 2006). This class of volatile compounds can be further divided into benzenoids (C6-C1 carbon skeleton), phenylpropanoid-related compounds (C6-C2) and phenylpropenes (C6-C3). Phenylpropenes, such as eugenol, isoeugenol, methyleugenol, isomethyleugenol, chavicol and methylchavicol, have roles beyond pollinator attraction, as they have been used in food flavoring and preservation, as well as in pharmacopoeias, since antiquity (Le Couteur & Burreson, 2004). The last biochemical steps in the biosynthesis of eugenol, isoeugenol and chavicol have only been elucidated recently in Clarkia breweri (Koeduka et al., 2008), Petunia hybrida (Koeduka et al., 2006, 2008, 2009; Dexter et al., 2007) and Ocimum basilicum (Koeduka et al., 2006; Vassao et al., 2006), which produce high levels of these phenylpropenes. It has been shown that chavicol and eugenol/isoeugenol are synthesized from the monolignols coumaroyl alcohol and coniferyl alcohol, respectively, whose side-chains undergo activation by acylation before NADPH-dependent reduction (Fig. 1). The genes responsible for these steps have been isolated and the corresponding enzymes have been characterized biochemically. They include the acetyl CoA:coniferyl alcohol acetyltransferase (PhCFAT) responsible for the conversion of coniferyl alcohol to coniferyl acetate in petunia flowers (Dexter et al., 2007), as well as eugenol synthase (EGS) and isoeugenol synthase (IGS), which reduce coniferyl acetate to the corresponding phenylpropenes in C. breweri, basil and petunia (Koeduka et al., 2006, 2008) (Fig. 1).

Figure 1.

Pathway leading to the formation of eugenol and isoeugenol and lignin monomers. The reactions catalyzed by cinnamoyl-CoA reductase are represented in a red box. Volatile compounds emitted by petunia flowers are evidenced by blue boxes and lignin monomers by red boxes. Stacked and dashed arrows indicate multiple enzymatic reactions. Abbreviations: 4CL, 4-coumarate:CoA ligase; C4H, cinnamate-4-hydroxylase; CAD, cinnamyl alcohol dehydrogenase; CCoAOMT, caffeoyl-CoA 3-O-methyltransferase; CCR, cinnamoyl-CoA reductase; CFAT, coniferyl alcohol acetyltransferase; EGS, eugenol synthase; IGS, isoeugenol synthase; PAL, phenylalanine ammonia lyase.

It has been proposed that monolignol biosynthesis in phenylpropene-emitting tissues is shared with the lignin biosynthetic pathway, despite the fact that these tissues often do not contain large amounts of lignin. However, genetic and biochemical evidence for the contribution of monolignol biosynthetic steps to the formation of phenylpropenes is still missing. The first committed step in the biosynthesis of lignin monomers involves the reduction of cinnamoyl-CoA thioesters to their respective cinnamaldehydes, which is catalyzed by cinnamoyl-CoA reductase (CCR, EC.1.2.1.44) (Fig. 1). CCR diverts carbon towards the synthesis of monolignols and therefore controls flux towards lignin (Piquemal et al., 1998; van der Rest et al., 2006; Leple et al., 2007; Ruel et al., 2009; Wagner et al., 2013). The plants examined thus far contain at least two CCR genes (Lauvergeat et al., 2001; Escamilla-Trevino et al., 2010; Zhou et al., 2010; Barakat et al., 2011), which are differentially expressed and possess distinct functions, namely lignin formation and pathogen defense (Lauvergeat et al., 2001; Escamilla-Trevino et al., 2010). Indeed, in Arabidopsis thaliana, AtCCR1 is expressed predominantly in lignifying stems, whereas AtCCR2 is poorly expressed during development but strongly induced under pathogen attack (Lauvergeat et al., 2001). The biochemical characterization of CCRs revealed that these enzymes have broad substrate specificity and are capable of converting a number of cinnamoyl-CoA thioesters (feruloyl-CoA, p-coumaroyl-CoA, sinapoyl-CoA, caffeoyl-CoA and 5-hydroxyferuloyl-CoA) to their respective aldehydes. Despite this broad substrate specificity, the highest catalytic efficiency was often found with feruloyl-CoA (Pichon et al., 1998; Baltas et al., 2005; Li et al., 2005; Ma, 2007; Escamilla-Trevino et al., 2010; Zhou et al., 2010).

To elucidate the potential involvement of CCR in volatile phenylpropene production, we used petunia (Petunia hybrida cv Mitchell) flowers, which emit high levels of phenylpropene compounds, including eugenol and isoeugenol, as a model system. We have isolated a petunia CCR (designated PhCCR1), which exhibits gene expression patterns typical for genes involved in scent production. Petal-specific silencing of PhCCR1 expression led to the accumulation of the CCR substrates p-coumaroyl-CoA and feruloyl-CoA, and a decrease in flux towards lignin and phenylpropene compounds, confirming that the biosyntheses of phenylpropenes and lignin share common enzymatic steps.

Materials and Methods

Growth conditions and generation of PhCCR1-RNAi transgenic lines

Petunia hybrida (Vilm.) cv Mitchell (Ball Seed Co., West Chicago, IL, USA) plants were grown under standard glasshouse conditions (Maeda et al., 2010). For the generation of the PhCCR1-RNAi construct, a 325-bp fragment was amplified by PCR using the forward and reverse primers 5′-CACCGGAGAAAGGTCATCTTCCAATTCC-3′ and 5′-GTAACATAAGGAAAAGAGCAATTCG-3′, respectively. The fragment was subcloned into the Gateway pENTR/D-TOPO entry vector (Invitrogen). After verification by sequencing, the 325-bp fragment was spliced into the two Gateway sites of the pLISG vector (in opposite orientation to each other) by LR clonase reaction (Maeda et al., 2010; Klempien et al., 2012). PhCCR1-RNAi transgenic plants were obtained via Agrobacterium tumefaciens (strain EHA105 carrying plasmid pLISG-PhCCR1) transformation using the standard leaf disk transformation method (Horsch et al., 1985). Plants were selected on shooting medium containing 200 μg ml−1 kanamycin.

RNA isolation and quantitative real-time polymerase chain reaction (qRT-PCR)

RNA was isolated using an RNeasy Plant Mini Kit (Qiagen) and treated with DNaseI using a TURBO DNA-free Kit (Ambion). One microgram of DNaseI-treated RNA was reverse transcribed using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Grand Island, NY, USA). Primer sequences were designed using PrimerExpress (Applied Biosystems) and the final concentrations of the primers are listed in Supporting Information Table S1. For expression analysis of Ph4CL1 and PhC4H1, primer sequences were as published (Colquhoun et al., 2011; Klempien et al., 2012). All primer concentrations were optimized to reach an amplification efficiency of 90–100%. For absolute quantification of PhCCR1 expression, the entire open reading frame of PhCCR1 was amplified using the forward and reverse primers 5′-CACCATGAGGTCAGTTTCCGGCCAAGTTG-3′ and 5′-AGGCTGAATTCGAATAATAGGCTC-3′, respectively. The amplicon was purified and quantified using a NanoDrop 1000 spectrophotometer (Thermo Scientific, Waltham, MA, USA). Standard curves for the absolute quantification of PhCCR1 were obtained by qRT-PCR using different concentrations of the PhCCR1 amplicon. For relative quantification, elongation factor 1-alpha (EF1a) was used as described previously (Mallona et al., 2010; Klempien et al., 2012). qRT-PCRs were performed in biological triplicates, each with three technical replicates. Reactions contained 5 μl of SYBR Green PCR Master Mix (Applied Biosystems), 2 μl of 50 times diluted cDNA and 1.5 μl of each of the primers. Amplification conditions were as described previously and qRT-PCRs were performed on a StepOne Real-Time PCR system (Applied Biosystems) (Maeda et al., 2010).

CCR enzyme assay

The activity of CCR was measured in protein extracts from wild-type and PhCCR1 transgenic corollas harvested at 20:00 h. Corolla tissue was ground in liquid nitrogen and protein was extracted in 100 mM Tris-HCl, pH 7.5, 2% (w/v) PEG6000, 5 mM dithiothreitol (DTT) and 2% (w/v) polyvinylpyrrolidone (PVP). The crude extracts were clarified by centrifugation, desalted on a Sephadex G50 (GE Healthcare Life Sciences, Pittsburgh, PA, USA) column and stored in 100 mM Tris-HCl, pH 7.5, 5 mM DTT. The total protein concentration was determined by the Bradford method. CCR enzyme assays were performed with 9.67 μg of desalted crude extract as described by Pan et al. (2014).

Subcellular localization

The coding region of PhCCR1 was cloned into the pDH51-GW-YFP plasmid (Nottingham Arabidopsis Stock Centre) (Zhong et al., 2008), resulting in a C-terminal yellow fluorescent protein (YFP) fusion under the control of the cauliflower mosaic virus 35S promoter. Transient transformation of Arabidopsis protoplasts was performed using the polyethylene glycol (PEG) method as described by Yoo et al. (2007). Fluorescence in transformed protoplasts was monitored by confocal laser scanning microscopy (Zeiss LSM710), as described previously (Klempien et al., 2012).

Targeted metabolite profiling

Volatiles emitted from wild-type and PhCCR1-RNAi flowers were collected between 18:00 and 21:00 h on day 2 postanthesis by the closed-loop stripping method, and analyzed as described previously (Orlova et al., 2006). Corolla tissue harvested at 21:00 h on day 2 postanthesis was used for the quantification of internal pools of VOCs, organic acids and amino acids, which were analyzed as described by Klempien et al. (2012), with minor modifications. Amino acids were derivatized with 100 μl methoxyamine in pyridine (20 mg ml−1) for 90 min at 30°C and 100 μl N-methyl-N-tert-butyldimethylsilyltrifluoroacetamide (MTBSTFA) (Campbell Science, Rockford, IL, USA) for 60 min at 70°C. Extraction and quantification of CoA thioesters were performed as described by Qualley et al. (2012a), with minor modifications. The electrospray ionization (ESI) source was operated in positive mode and multiple reaction monitoring (MRM) was performed using the following ions (corresponding to [M + H] of our analytes of interest): cinnamoyl-CoA, 898; [2H5]benzoyl-CoA, 877; benzoyl-CoA, 872; p-coumaroyl-CoA, 914; caffeoyl-CoA, 930; feruloyl-CoA, 944; 4-chlorobenzoyl-CoA, 907. The internal standards 4-chlorobenzoyl-CoA and [2H5]benzoyl-CoA were synthesized via an N-hydroxysuccinimide ester of 4-chlorobenzoic acid and [2H5]benzoic acid, as described by Stöckigt & Zenk (1975). 4-chlorobenzoyl-CoA was used as an internal standard for quantification of CoA thioesters from 2H5-phenylalanine fed petal tissue, while [2H5]benzoyl-CoA was used as an internal standard with unfed tissue.

Quantification of total lignin

Total lignin was quantified by the thioglycolic acid method as described by Campbell & Ellis (1992).

Labeling experiments with 2H5-phenylalanine (2H5-Phe)

Feeding experiments with 150 mM 2H5-Phe (Cambridge Isotopes, Tewksbury, MA, USA) were performed from 18:00 to 22:00 h with day 2 postanthesis corollas from wild-type and PhCCR1-RNAi lines, as described previously (Boatright et al., 2004).

Results

Isolation and gene expression of petunia CCRs

A search of a petunia petal-specific expressed sequence tag (EST) database (Boatright et al., 2004), the Sol genomics network (http://solgenomics.net) and the Gene Indices was performed for sequences with homology to LeCCR1 and LeCCR2 from tomato. Tblastx queries resulted in three ESTs representing PhCCR1 (Pan et al., 2014) and one EST (FN003816) with 78% nucleotide identity to PhCCR1. To identify the CCR involved in phenylpropene formation, the relative expression of both candidates was analyzed in corollas, as the genes involved in the biosynthesis of scent compounds exhibit the highest expression in this tissue (Klempien et al., 2012; Qualley et al., 2012b). qRT-PCR with gene-specific primers revealed that, in contrast with PhCCR1, which is highly expressed in corollas of 2-d-old petunia flowers, FN003816 expression was undetectable (Fig. 2a). Further analysis of tissue-specific PhCCR1 gene expression showed that PhCCR1 is expressed in all floral tissues, as well as in leaves and stems (Fig. 2b, Supporting Information Fig. S1). Within the flower, the highest expression was found in the scent-producing corollas and tubes. PhCCR1 expression in other floral tissues, as well as in stems, leaves and sepals, was at least 40% lower than that in corollas. In corollas, PhCCR1 gene expression changed rhythmically over the day : night cycle with a maximum at c. 15:00 h (Fig. 2b), whereas this rhythmicity was not observed in old or young stem tissues (Fig. S1). Lastly, PhCCR1 expression in corollas was differentially regulated over flower development, reaching a maximum at day 3 postanthesis (Fig. 2b), thus exhibiting a positive correlation with phenylpropanoid emission in petunia flowers (Dexter et al., 2007).

Figure 2.

Expression profiles of petunia cinnamoyl-CoA reductases (CCRs). (a) Expression of PhCCR1 and FN003816 in corollas of flowers collected on day 2 postanthesis at 15:00 h. nd, not detected. (b) Tissue-specific expression is shown relative to the expression in corollas. Rhythmic expression of PhCCR1 in corollas harvested between the day of anthesis (day 1) at 15:00 h and day 3 at 03:00 h. Gray bars in the rhythmicity panel indicate dark hours. Levels are shown relative to the expression on day 2 at 15:00 h. Expression levels of PhCCR1 in corollas harvested over flower development, from mature buds to day 7. Levels are shown relative to the expression of PhCCR1 in corollas on day 2. Data are means ± SE (= 3 biological replicates).

Subcellular localization of PhCCR1

Based on in silico localization analysis using the PSORT program (http://www.psort.org/), PhCCR1 was predicted to be localized in the cytosol. Furthermore, the rice CCR OsCCR1 (Kawasaki et al., 2006), as well as several other enzymes involved in phenylpropanoid metabolism, have been shown to be localized in the cytosol. To experimentally determine the subcellular localization of PhCCR1, the complete coding region of the gene was fused to the N-terminus of YFP. The construct was transiently expressed in Arabidopsis protoplasts under the control of the cauliflower mosaic virus 35S promoter, and confocal laser scanning microscopy was used to assess the localization of the corresponding fluorescent fusion protein. The subcellular distribution of the fluorescence signals of the PhCCR1-YFP was found to be identical to the distribution of the green fluorescent protein (GFP), which served as a cytosolic marker (Fig. S2). Moreover, PhCCR1-YFP signals did not overlap with chlorophyll autofluorescence, confirming the in silico prediction for PhCCR1 cytosolic localization (Fig. S2).

Effects of PhCCR1 silencing on the phenylpropanoid network

To determine the contribution of PhCCR1 to phenylpropene formation in planta, transgenic lines were generated in which PhCCR1 transcript levels were decreased in petunia petals using an RNAi approach under the control of a petal-specific LIS promoter (Cseke et al., 1998). Initial screening of transgenic plants was performed based on CCR activity in petal tissue. Three independently transformed lines (H, Q and V) with the highest reduction in CCR activity in petal crude extracts (> 80% reduction relative to untransformed controls) (Fig. 3b) were chosen for further analysis and metabolic profiling. Consistent with the reduction in CCR activity, PhCCR1 expression in these lines was decreased by 90% when compared with wild-type flowers (Fig. 3a). Metabolic profiling of transgenic and control flowers revealed that this 90% decrease in PhCCR1 transcript levels did not affect the emission and internal pools of phenylpropene and benzenoid compounds, whereas it increased the emission rate and internal pools of vanillin by c. 1.6- and c. 2.8-fold, respectively (Figs 3c, S3). Moreover, the quantification of hydroxycinnamoyl-CoA thioesters uncovered an increase in feruloyl-CoA and p-coumaroyl-CoA, the PhCCR1 substrates (Pan et al., 2014), on average by c. 420% and 95%, respectively, in transgenic lines relative to controls (Fig. 4a). No statistically significant changes were observed in transgenic lines for the pool sizes of the other detected CoA esters, benzoyl-CoA and cinnamoyl-CoA, which are involved in the benzenoid network in petunia flowers (Fig. 4a). Similarly, levels of free organic acids, with the exception of benzoic acid, and the phenylalanine pool size were not affected in transgenic lines (Fig. 4b,c). Total lignin content, however, was reduced by 30% relative to controls (Fig. 4d).

Figure 3.

Cinnamoyl-CoA reductase (CCR) activity, gene expression, volatile emission and internal pools of volatiles in PhCCR1 transgenic lines. (a) PhCCR1 gene expression levels and (b) CCR activity were determined in corollas of petunia wild-type (WT; open bars) and three independent transgenic lines (H, Q, V; closed bars) harvested on day 2. Data are means ± SE (= 3 biological replicates). (c) Emission rates of volatile compounds are shown in the graphs with a white background, and levels of the corresponding internal pools are shown in the graphs with a gray background. Emitted volatiles were collected from detached flowers on day 2 from 18:00 to 2100 h by a closed-loop stripping method. Corolla tissue for internal pools was harvested on day 2 at 21:00 h. Data are means ± SE (n ≥ 9 biological replicates for emission and = 5 biological replicates for internal pools). Significant difference between wild-type and the corresponding PhCCR1 transgenic line by a paired t-test: *, < 0.05.

Figure 4.

Quantification of phenylalanine-derived metabolites in petunia wild-type (WT) and cinnamoyl-CoA reductase 1 (PhCCR1) transgenic lines (H, Q, V). (a) Hydroxycinnamoyl-CoA esters (pCACoA, p-coumaroyl-CoA; FACoA, feruloyl-CoA; CACoA, cinnamoyl-CoA; BACoA, benzoyl-CoA), (b) free organic acids, (c) phenylalanine and (d) total lignin were extracted from corolla tissue harvested on day 2 at 21:00 h of wild-type (open bars) and PhCCR1 transgenic lines (closed bars). Data are means ± SE ( 3 biological replicates). Significant difference between wild-type and the corresponding PhCCR1 transgenic line by a paired t-test: *, < 0.05.

Fluxes through the phenylpropanoid network on PhCCR1 silencing

To investigate the effects of PhCCR1 silencing on fluxes within the phenylpropanoid network, wild-type and PhCCR1 transgenic corollas were fed with deuterium ring-labeled Phe (2H5-Phe), and pool sizes and isotopic abundances of phenylpropanoid metabolites were monitored over 4 h. The labeling kinetics of the emitted phenylpropene isoeugenol (C6-C3) were significantly different between wild-type and PhCCR1 transgenic corollas (< 0.05, t = 2.90) (Fig. 5a). Indeed, an 18% decrease in labeling was observed in PhCCR1 transgenic lines compared with wild-type corollas after 4 h of 2H5-Phe feeding. Over the same time frame, no differences in isotopic labeling were observed for the phenylpropanoid-related volatile phenylacetaldehyde (C6-C2) or the benzenoid volatile benzaldehyde (C6-C1) between wild-type and PhCCR1 transgenic corollas (> 0.05, t = 1.66 and > 0.05, t = 2.23, respectively) (Fig. 5a). Pool sizes for all of these emitted volatiles were not significantly altered between wild-type and PhCCR1 transgenic corollas (isoeugenol: > 0.05, t = 2.43; phenylacetaldehyde: > 0.05, t = 1.41; benzaldehyde: > 0.05, t = 0.48) (Fig. 5b).

Figure 5.

Isotopic abundances and pool sizes of emitted volatiles in petunia wild-type (WT) and a cinnamoyl-CoA reductase 1 (PhCCR1)-RNAi line on 2H5-phenylalanine (2H5-Phe) feeding. (a) Isotopic abundances and (b) pool sizes of emitted volatiles were analyzed during a 4-h time course of 2H5-Phe feeding to wild-type (open squares) and PhCCR1-RNAi line Q (closed triangles) corollas. Data are means ± SE (= 4 biological replicates).

To test whether the decreased labeling in isoeugenol was caused by a decreased flux through the PhCCR1-catalyzed step or resulted from a decreased labeling in the PhCCR1 substrate feruloyl-CoA, labeling kinetics and pool sizes of hydroxycinnamoyl-CoA thioesters were analyzed. This analysis revealed that isotopic abundances of hydroxycinnamoyl-CoA thioesters, with the exception of caffeoyl-CoA, were unaltered in PhCCR1 transgenic relative to wild-type corollas (p-coumaroyl-CoA: > 0.05, t = 2.46; caffeoyl-CoA: < 0.05, t = 3.16; feruloyl-CoA: > 0.05, t = 2.66) (Fig. 6a), suggesting that the observed decrease in isoeugenol labeling is caused by a reduced flux through PhCCR1. Interestingly, although the isotopic abundances of feruloyl-CoA were nearly identical in transgenic and wild-type petals, the feruloyl-CoA pool was significantly increased in PhCCR1 transgenics relative to the wild-type (< 0.05, t = 5.00) (Fig. 6b), indicating higher flux towards this hydroxycinnamoyl-CoA thioester in transgenic lines.

Figure 6.

Isotopic abundances and pool sizes of hydroxycinnamoyl-CoA esters in petunia wild-type (WT) and a cinnamoyl-CoA reductase 1 (PhCCR1)-RNAi line on 2H5-phenylalanine (2H5-Phe) feeding. (a) Isotopic abundances and (b) pool sizes of hydroxycinnamoyl-CoA esters were analyzed during a 4-h time course of 2H5-Phe feeding to wild-type (open squares) and PhCCR1-RNAi line Q (closed triangles) corollas. Data points for each biological replicate are shown. Lines represent means of the biological replicates (= 1–2 biological replicates).

Gene expression in the phenylpropanoid network in PhCCR1-RNAi flowers

To gain a better understanding of the regulatory mechanisms leading to the observed phenotype of PhCCR1 transgenic flowers, we quantified the expression of multiple genes involved in the general phenylpropanoid pathway. Since we observed an increase in carbon fluxes through the general phenylpropanoid pathway in PhCCR1 transgenic flowers (Fig. 6), we analyzed the expression of cinnamate-4-hydroxylase 1 (PhC4H1) and 4-coumarate:CoA ligase 1 (Ph4CL1), the genes encoding the enzymes catalyzing the second and third reactions of this pathway. PhC4H1 expression exhibited an approximately two-fold increase in PhCCR1 transgenic flowers relative to the control, whereas Ph4CL1 expression remained unaltered by PhCCR1 silencing (Fig. 7a). These results suggest that increased fluxes through the general phenylpropanoid pathway are probably a result of transcriptional upregulation of PhC4H1. In petunia flowers, PhC4H1 expression is controlled by the transcriptional repressor PhMYB4 (Colquhoun et al., 2011). However, expression of this transcriptional repressor was not altered by PhCCR1 silencing (Fig. 7b), indicating that the upregulation of PhC4H1 occurs in a PhMYB4-independent manner. Alternative mechanisms for PhC4H1 upregulation could include an as yet unidentified transcription factor.

Figure 7.

Expression of genes involved in the phenylpropanoid pathway in petunia wild-type (WT) and cinnamoyl-CoA reductase 1 (PhCCR1) transgenic lines (H, Q, V). Expression of the (a) cinnamate-4-hydroxylase (PhC4H1), 4-coumarate:CoA ligase (Ph4CL1) and (b) the transcription factor PhMYB4 was measured in wild-type (open bars) and PhCCR1 transgenic flowers (closed bars). Data are means ± SE (= 3 biological replicates). Significant difference between wild-type and the corresponding PhCCR1 transgenic line by a paired t-test: *, < 0.05.

As an 80% decrease in CCR activity in PhCCR1 transgenic flowers did not affect the internal and emitted pools of phenylpropenes, we investigated whether this observation could be explained by the upregulation of genes involved in the final steps of phenylpropene biosynthesis (i.e. PhCFAT, PhIGS and PhEGS). qRT-PCR with gene-specific primers revealed that the expression of these genes was not influenced by PhCCR1 silencing (Fig. S4).

Discussion

PhCCR1 silencing uncovers a contribution of the monolignol pathway to phenylpropene formation

Volatile phenylpropenes are widely distributed in the plant kingdom and play important roles in pollinator attraction, plant defense and human nutrition (Koeduka et al., 2006). They constitute the second largest subset (after benzenoids) of Phe-derived VOCs and originate from the general phenylpropanoid pathway, which begins with the deamination of Phe to trans-cinnamic acid by Phe ammonia lyase (Bonawitz & Chapple, 2010). The last two steps in the biosynthesis of phenylpropenes have been elucidated recently in petunia flowers, which emit isoeugenol and eugenol (Koeduka et al., 2006, 2008; Dexter et al., 2007). It has been shown that PhIGS (Koeduka et al., 2006) and PhEGS (Koeduka et al., 2008) catalyze the conversion of coniferyl acetate to isoeugenol and eugenol, respectively. Coniferyl acetate is formed from the monolignol precursor coniferyl alcohol in a reaction catalyzed by PhCFAT (Dexter et al., 2007) (Fig. 1). Monolignols are also the principal lignin polymer constituents, and thus it has been proposed that a common set of enzymes contribute to both lignin monomers and phenylpropene synthesis (Koeduka et al., 2006). The first committed step towards monolignols and monolignol-derived compounds is catalyzed by CCR, which is responsible for the NADPH-dependent reduction of cinnamoyl-CoA thioesters to the corresponding cinnamaldehydes. To assess the contribution of CCR(s) to the biosynthesis of volatile phenylpropenes, we identified two P. hybrida CCRs, one of which, PhCCR1, is highly expressed in the scent-emitting flower tissues and exhibits rhythmic and developmental expression profiles typical of genes involved in scent production (Fig. 2) (Klempien et al., 2012; Qualley et al., 2012b). Although PhCCR1 expression is highest in corollas, the presence of expression, albeit at lower levels, in stems, leaves and sepals suggest its potential contribution to the lignification of these tissues (Figs 2, S1). Our recent biochemical characterization of PhCCR1 revealed that it is a bona fide CCR and, like other characterized CCRs, prefers feruloyl-CoA as a substrate (Pan et al., 2014). Similar to previously characterized CCRs, PhCCR1 can also utilize sinapoyl-CoA and p-coumaroyl-CoA, whilst having a very low activity with caffeoyl-CoA and benzoyl-CoA (Pan et al., 2014). Indeed, the accumulation of the hydroxycinnamoyl-CoA thioesters feruloyl-CoA and p-coumaroyl-CoA, which serve as monolignol precursors, was observed in PhCCR1 transgenic flowers, confirming the involvement of PhCCR1 in monolignol biosynthesis in planta. The pool sizes of benzoyl-CoA and cinnamoyl-CoA, however, remained unaltered in transgenic flowers relative to controls. In petunia petals, benzoyl-CoA is an essential intermediate in benzenoid biosynthesis (Orlova et al., 2006; Van Moerkercke et al., 2009). This CoA thioester is not part of the lignin biosynthetic pathway, but could serve as a potential substrate for PhCCR1 based on in vitro enzyme assays (Pan et al., 2014), thus linking the β-oxidative and non-β-oxidative benzenoid pathways by converting benzoyl-CoA to benzaldehyde. Although benzoyl-CoA is synthesized in the peroxisome (Van Moerkercke et al., 2009; Klempien et al., 2012; Qualley et al., 2012b), a pool of benzoyl-CoA is also present in the cytosol and serves as the substrate for the cytosolic benzoyl CoA:benzylalcohol/phenylethanol benzoyltransferase (Orlova et al., 2006). Benzoyl-CoA is therefore potentially available to the cytosolically localized PhCCR1 (Fig. S2). However, the lack of changes in the benzoyl-CoA pool size and benzenoid emission between wild-type plants and PhCCR1 transgenic lines suggests that PhCCR1 is not involved in the benzenoid network in planta.

The total lignin content in PhCCR1 transgenic flowers was reduced by 30% (Fig. 4d), consistent with the role of CCR in monolignol biosynthesis (Piquemal et al., 1998; van der Rest et al., 2006; Leple et al., 2007; Ruel et al., 2009; Wagner et al., 2013). In contrast with lignin levels, emitted and internal pool sizes of benzenoid/phenylpropanoid volatiles remained unchanged in PhCCR1 transgenic flowers (Figs 3c, S3). Despite the lack of differences in phenylpropene levels between PhCCR1 transgenic and wild-type flowers, the flux towards isoeugenol was decreased, as evidenced by the reduced incorporation of [2H5-Phe] into isoeugenol in PhCCR1 transgenic lines (Fig. 5a). Taken together, these results provide clear evidence that the monolignol pathway contributes to the biosynthesis of phenylpropene volatiles and is shared with lignin biosynthesis.

Interestingly, a drastic increase in the endogenous pool of vanillin and a slight elevation in its emission were observed in PhCCR1 transgenic flowers (Fig. 3c). Vanillin is the most commonly used flavoring compound, but the biosynthesis of this aldehyde has not been fully elucidated (Dixon, 2011). It has been proposed that vanillin biosynthesis proceeds via feruloyl-CoA as an intermediate (Zenk, 1965). Therefore, the observed increase in vanillin production in the PhCCR1 transgenic lines (Fig. 3c) could be the result of a redirection of accumulated feruloyl-CoA into vanillin.

Reduced flux through PhCCR1 leads to higher perturbations in total lignin production than in phenylpropene emission

Silencing of PhCCR1 in petunia flowers had a larger impact on the total amounts of lignin than on the emitted and internal pools of phenylpropene volatiles. A possible explanation for this observation is that enzymes competing for the monolignol substrate differ in their relative affinities. At the coniferyl alcohol branch point, coniferyl alcohol is subject to two major fates. It can be utilized by ABC transporters which export monolignols to the cell wall (Miao & Liu, 2010; Alejandro et al., 2012), and by PhCFAT which directs carbon flux towards phenylpropene formation (Dexter et al., 2007). In principle, the branch with the highest apparent Km for coniferyl alcohol will probably be the most sensitive to reduction in its concentration. It has previously been shown that the apparent Km value of PhCFAT for coniferyl alcohol is 27.5 μM (Dexter et al., 2007), whereas the apparent Km value of the ABC transporter moving coniferyl alcohol across Arabidopsis plasma membranes is 71.4 μM (Miao & Liu, 2010). These values provide a mechanistic explanation for the higher perturbation in total lignin production than in phenylpropene emission by PhCCR1 silencing.

Another explanation for the larger impact of PhCCR1 silencing on total lignin relative to phenylpropene production is that reduced flux through PhCCR1 could be partially compensated by an increase in PhCFAT, PhEGS and PhIGS expression. These biosynthetic genes are ultimately responsible for the conversion of coniferyl alcohol to eugenol and isoeugenol. However, the expression of these biosynthetic genes was unaltered in PhCCR1 transgenic flowers (Fig. S4).

PhCCR1 silencing upregulates carbon fluxes upstream of PhCCR1 in a PhMYB4-independent manner

The observation that PhCCR1 silencing leads to increased fluxes through hydroxycinnamoyl-CoA thioesters in the general phenylpropanoid pathway (Fig. 6) implies the activation of an early step within this pathway. The first gene, cinnamate-4-hydroxylase (PhC4H1), which diverts carbon from cinnamate towards the general phenylpropanoid pathway, exhibited an approximately two-fold increase in expression in PhCCR1 transgenic flowers relative to controls. Interestingly, a similar increase in C4H expression was observed in Arabidopsis ccr1 mutant plants (Vanholme et al., 2012).

In the PhCCR1 transgenic lines, the increase in feruloyl-CoA (c. 420%) was higher than for p-coumaroyl-CoA (c. 95%). This higher accumulation of feruloyl-CoA can be explained by different scenarios which are not mutually exclusive. First, the accumulation of p-coumaroyl-CoA as a result of PhCCR1 blockage could increase the flux through the route leading from p-coumaroyl-CoA to feruloyl-CoA via a caffeoyl-CoA intermediate. Second, this higher accumulation in feruloyl-CoA could reflect the kinetic properties of the petunia CCR1, which has a higher catalytic efficiency with feruloyl-CoA than with p-coumaroyl-CoA. Therefore, blockage of PhCCR1 has stronger effects on the feruloyl-CoA pool size than on the p-coumaroyl-CoA pool.

In summary, we have provided genetic evidence for the involvement of the monolignol-specific pathway in both lignin formation and volatile phenylpropene biosynthesis. Indeed, silencing of a CCR, which catalyzes the first committed step in monolignol biosynthesis, resulted in decreased total lignin and flux towards the volatile phenylpropene isoeugenol in petunia flowers. This work has advanced our understanding of how perturbations in the monolignol biosynthetic pathway affect carbon flux towards both lignin and phenylpropene volatiles.

Acknowledgements

We thank Christine Whittinghill and Funmilayo Adebesin for technical assistance with gene isolation and qRT-PCR, and Dr Pierre-Alexandre Vidi for help with confocal microscopy. This work was supported by a grant from the National Science Foundation MCB-0919987 to N.D.

Ancillary