E. Pichersky, Department of Molecular, Cellular and Developmental Biology, University of Michigan, 830 North University Street, Ann Arbor, MI 48109, USA. E-mail. email@example.com
The specific set of reactions that lead to the synthesis of benzoic acid in plants is still unclear, and even the subcellular compartment in which these reactions occur is unknown. Biosynthesis of both vegetative tissues and seeds of Arabidopsis thaliana contain a class of defense compounds termed glucosinolates, but only the seeds synthesize and store high levels of two glucosinolate compounds that contain a benzoic acid moiety. To identify genes involved in the synthesis of benzoic acid (directly or via benzaldehyde) in Arabidopsis, we analysed the levels of benzoylated glucosinolates in several lines that carry mutations in genes with homology to Pseudomonas fluorescens feruloyl-CoA hydratase, an enzyme that converts feruloyl-CoA to vanillin and acetyl-CoA, a reaction analogous to the conversion of cinnamoyl-CoA to benzaldehyde. We show here that mutations in the gene At5g65940, previously shown to encode a peroxisomal protein with β-hydroxyisobutyryl-CoA hydrolase activity and designated as Chy1, lead to a deficiency of benzoic acid-containing glucosinolates in the seeds. Furthermore, Chy1 exhibits cinnamoyl-CoA hydrolase activity with a Km of 2.9 μm. Our findings suggest that at least a part of benzoic acid biosynthesis occurs in the peroxisomes, although the specific pathway that leads to benzoic acid and the specific biochemical role of Chy1 remain unclear.
Benzoic acid (BA) and its derivatives serve as important structural elements in a large number of compounds in plants and have a range of functions. For example, both taxol and cocaine, produced by plants as defense compounds, contain a benzoyl group (Bjorklund & Leete 1992; Walker & Croteau 2000). Methylbenzoate and benzylbenzoate are common volatile esters found in the scent of many flowers, including snapdragon and petunia, and function to attract pollinators (Negre et al. 2003). Benzoyl moieties are not limited to specialized metabolites, as they are also present, for example, in some cytokinins, with meta-topolin being the most active compound among the aromatic compounds (Werbrouck et al. 1996).
Several hypothetical routes have been proposed for the biosynthesis of BA from a phenylpropanoid precursor in plants that, in principle, fall into two basic routes, β-oxidative and non-β-oxidative, and lead initially to either benzoyl-CoA or benzaldehyde, respectively (Fig. 1) (Beuerle & Pichersky 2002). It is possible that several of these pathways coexist in plants and might even be active simultaneously in the same tissue (Orlova et al. 2006). Experiments with stable isotope-labelled precursors in tobacco (Nicotiana tabacum) leaves (Ribnicky et al. 1998) suggest that benzoic acid is produced from phenylalanine-derived cinnamic acid via the β-oxidative pathway, mirroring the fatty acid β-oxidation pathway (Hertweck et al. 2001), and yielding benzoyl-CoA first, which is then hydrolysed by a thioesterase to free benzoic acid. In contrast, labelling experiments along with initial enzyme characterization in Hypericum androsaemum cell cultures (Abd El-Mawla & Beerhues 2002) support the existence of the non-oxidative conversion of cinnamic acid to benzaldehyde, with subsequent formation of benzoic acid, which can be further converted to benzoyl-CoA (Beuerle & Pichersky 2002). However, none of the genes or enzymes involved in any chain-shortening steps in BA biosynthesis has been identified or characterized in plants.
In the bacterium Pseudomonas fluorescens, a feruloyl-CoA hydratase/lyase was found that catalyses the metabolism of feruloyl-CoA to vanillin and acetyl-CoA via hydration and reverse aldol cleavage (Gasson et al. 1998; Mitra et al. 1999). This enzyme is active on a number of 4-hydroxycinnamoyl-CoA substrates, but not on cinnamoyl-CoA (Gasson et al. 1998; Mitra et al. 1999). This led to the postulation of a similar CoA-dependent non-oxidative pathway involved in BA biosynthesis in plants, and to a subsequent report of cinnamoyl-CoA hydratase/lyase activity in protein extracts of elicited H. androsaemum cells (Abd El-Mawla & Beerhues 2002). To date, this enzyme activity has not been further purified from H. androsaemum or any other plant species, so the identity of the protein possessing such activity is not known.
Arabidopsis and related species synthesize a class of compounds known as glucosinolates that serve in defense throughout the plant. These compounds all have the same basic β-thioglucose structure, but differ in the structure of the aliphatic side chain. Developing Arabidopsis seeds, but not other parts of the plant, synthesize large amounts of two types of glucosinolate whose side chain contains a benzoyl moiety (Graser et al. 2001). Because of the high amounts of benzoylated glucosinolates in Arabidopsis seeds, these seeds constitute an excellent system to examine the biosynthesis of benzoic acid. In this report, we demonstrate that Arabidopsis plants homozygous for null alleles in the gene At5g65940 (Chy1), a distant homolog of P. fluorescens feruloyl-CoA hydratase/lyase that was previously shown to encode a peroxisomal protein that affects β-oxidization (Zolman et al. 2001), are deficient in benzoylated glucosinolates in the seeds.
Material and Methods
Arabidopsis thaliana (ecotype Columbia) plants were grown on soil at 23 °C under 16 h light/8 h dark for 4–6 weeks. SALK mutant lines were obtained from the ABRC. In addition, plants homozygous for two mutant Chy1 alleles, chy1-1 and chy1-3, were obtained from Dr Bonnie Bartel (Rice University, Houston, USA).
Extraction and analysis of benzoylglucosinolates
Extracts were made using methods adapted from previously published protocols (Reichelt et al. 2002). Seeds (20 mg) were extracted in a 1-ml solution of 80% MeOH to which 30 μl of 10 mm sinigrin (Sigma-Aldricht, Allentown, PA, USA) were added as an internal standard (final concentration 300 μm). The extracts were centrifuged in a microcentrifuge at 2000 g for 10 min and the supernatant loaded onto a small (50 mg) column of DEAE Sephadex A25 (Sigma-Aldricht). Water was added to the DEAE Sephadex A25 to induce swelling. The column was rinsed successively with 1 ml 80% (aqueous) MeOH, 1 ml deionized water, and 1 ml 20 mm NaOAc, pH 5.5. After washing, 25 μl sulfatase (Sigma-Aldricht) solution, prepared according to Graser et al. (2001), were loaded on the column and the column was left standing overnight. The resulting desulfoglucosinolates were eluted from the column with 0.5 ml 20% (aqueous) MeOH.
Extraction and HPLC analysis of benzoic acid
The protocol for extraction of BA was modified from Chong et al. (2001). Plant material (10–20 mg seeds) was extracted with 2 ml 90% MeOH containing 10 μl 10 mm cinnamic acid as an internal standard. After centrifugation in a microcentrifuge for 5 min at 14,000 rpm, the residue was extracted again with 2 ml 100% MeOH. The combined extracts were reduced to dryness under nitrogen. Samples were either analysed directly by HPLC for BA after resuspension in methanol, or dried extracts were resuspended in 1 ml water at 50 °C, saponified with 100 μl 1 m NaOH for 30 min at room temperature, neutralized with 30 μl 1 n HCL, centrifuged in a microcentrifuge for 5 min at 14,000 rpm, and then injected into the HPLC.
Cloning and expression of a cDNA of the Arabidopsis At5g65940 (Chy1) gene
RNA from the siliques of A. thaliana Columbia ecotype Col-0 was isolated using methods adapted from previously published protocols (Maes & Messens 1992). Total RNA (5 μg) was reversed-transcribed using an oligo dT primer and MLV RT (Stratagene, La Jolla, CA, USA). The open-reading frame of the Chy1 gene (At5g65940) was amplified by PCR using the primers 5′-ATGGCAGTCGAGATGGCCTCTCA-3′ and 5′-CAGCTTTGCGATCCCTAAAGCAGGCAAGTTATT C-3′, which eliminated the termination codon. The high-purity KOD-DNA polymerase (Novagen, Madison, WI, USA) was used and constructs were verified by DNA sequencing. The cDNA was ligated into the vector pEX5-CT/TOPO TA (Invitrogen, Carlsbad, CA, USA) to create a fusion of the open-reading frame with a His tag-coding extension at the C-terminus. The plasmid was transferred into Escherichia coli BL21 (DE3) cells. Protein expression was induced with 1 mm isopropyl-1-thio-β-d-galactopyranoside at 18 °C for 14 h. Bacteria were lysed and soluble protein purified by Nickel chromatography and assayed for purity by SDS–PAGE as previously described (Koeduka et al. 2008).
Enzyme activity measurements
Enzyme assays were carried out in a buffer containing 50 mm Tris/HCl pH 7.0, with 20 μm substrate and 1–500 ng of total protein in a total volume of 100 μl. The assays were incubated at 30 °C for 30 min, after which reactions were terminated by the addition of 20 μl 3.5% TCA in 50% acetonitrile. The reaction products were analysed by reverse phase liquid chromatography on a Inertsil ODS-2 5 μm C18 column (150 × 4.6 mm i.d., 5 μm particle size) operated at 0.8 ml·min−1 and 30 °C, and eluted with a gradient (solvent A: 1.5% phosphoric acid in H2O, solvent B: acetonitrile) of 0–10% B (2 min), 10–45% B (12 min), 45–50% B (16 min) 50–95% (17 min), 95% (1 min), 95–10% (25 min). Compound elution was monitored at 200–400 nm with a Waters 996 UV/visible photodiode array detector (Waters, Milford, MA, USA).
Characterization of Chy1 kinetic parameters
In all enzyme activity measurements, appropriate enzyme concentrations and incubation times were chosen so that the reaction velocity was linear during the incubation time, and at least three replicates were performed. pH optima were measured in 50 mm Tris-buffer ranging from pH 5.5 to 9.0.
Enzymatic synthesis and purification of aromatic coenzyme A esters
Cinnamoyl-CoA, p-coumaroyl-CoA, and 3-hydroxy-3-phenoylpropionoyl-CoA were prepared using tobacco 4Cl as described by Beuerle & Pichersky (2002).
A bioinformatic search for an Arabidopsis gene homologous to Pseudomonas fluorescens feruloyl-CoA hydratase/lyase
A bioinfomatic search for plant proteins with significant sequence similarity to Pseudomonas fluorescens feruloyl-CoA hydratase/lyase did not yield any protein with sequence identity >30%. However, in the Arabidopsis genome there are at least 13 genes encoding proteins with low (<30%) but statistically significant sequence similarity to P. fluorescens feruloyl-CoA hydratase/lyase (Fig. 2). These include the two peroxisomal multifunctional proteins encoded by genes At4g29010 and At3g06860, each containing a hydratase domain known to be involved in the β-oxidation degradative pathway of fatty acids (Richmond & Bleecker 1999; Rylott et al. 2006), as well as the protein encoded by At5g65940, designated Chy1 and also shown to be localized to peroxisomes (Zolman et al. 2001). chy1 mutants have been shown to be defective in the β-oxidization of fatty acids (Zolman et al. 2001; Lange et al. 2004). They are also defective in the conversion of indole-3-butyric acid (IBA) to indole-3-acetic acid (IAA), a reaction that requires the removal of two carbons from the butyric side chain and that has been postulated to occur through the β-oxidation pathway (Zolman et al. 2001). The only enzymatic activity so far demonstrated for Chy1 is hydrolysis of β-hydroxyisobutyryl-CoA to β-hydroxyisobutyrate and CoA, a reaction that is part of the valine degradation pathway in animals (Shimomura et al. 1994; Zolman et al. 2001). The inhibitory affect of the chy1 mutants on β-oxidation was hypothesized to occur via the accumulation of a toxic intermediate in the valine degradation pathway (Zolman et al. 2001). However, whether valine degradation in plants occurs in peroxisomes is still unknown (Lange et al. 2004; Lucas et al. 2007).
Seeds of mutants in the Chy1 gene are deficient in benzoylated glucosinolates
We obtained T-DNA knockout lines (in Arabidopsis Col-0 background) of many of the 13 Arabidopsis homologs of P. fluorescens feruloyl-CoA hydratase/lyase and tested their seeds for the presence of benzoylated glucosinolates. Null mutations in the two multifunctional proteins of fatty acid β-oxidation are either lethal or sterile (Richmond & Bleecker 1999; Rylott et al. 2006) so it was not possible to test seeds from such mutants. Homozygous lines for a T-DNA insertion in genes At4g16800, At4g16210, and At1g60550 also could not be obtained. Seeds of homozygous lines with T-DNA insertions in genes At5g43280, At2G30650, At3g06860, At2g30660, and At4g16210 had normal levels of the two benzoylated glucosinolates found in Arabidopsis Col-0 wild-type seeds, 3-benzoyloxypropylglucosinolate (3BZO) and 4-benzoyloxybutylglucosinolate (4BZO) (see for example T-DNA insertion mutant of At2g30650, Fig. 3A and B). In contrast, seeds of plants from a homozygous line with a T-DNA insertion (SALK_025417) in gene At5g65940 (Chy1) had no detectable levels of these two benzoylated glucosinolates. Since several EMS-induced point mutations in Chy1 in the Arabidopsis Col-0 background have been reported (Zolman et al. 2001), we obtained two of those, chy1-1 and chy1-3, and tested their seeds for the presence of 3BZO and 4BZO. These lines also had no detectable levels of either compound (see Fig. 3 for chy1-1).
Seeds of mutants in the Chy1 gene contain no benzoic acid
To determine if mutants in the Chy1 gene do not accumulate 3BZO and 4BZO because of a defect in their synthesis rather than in the process of linking them to the glucosinolate moiety, we measured the total levels of BA in these seeds. Methanolic extraction of seed material detected no free BA in any seed line, including the wild type (data not shown). When methanolic extracts were subjected to base hydrolysis, BA was detected in extracts from wild-type plants and from the line homozygous for a T-DNA insertion in gene At2g30650 but not from seeds of the chy1 mutants SALK_02541 and chy1-1 (Fig. 4A and B).
Enzymatic activity of the Chy1 protein
The three amino acids at the C-terminus of the 378-residue Chy1 protein are AKL, a known peroxisomal targeting signal (Nakai & Horton 1999; Emanuelsson et al. 2000; Reumann 2004), and previous reports have demonstrated that Chy1 is present in peroxisomes (Zolman et al. 2001). Since Chy1 mutants are defective in BA biosynthesis as well as in peroxisomal β-oxidation (Zolman et al. 2001), we tested Chy1 for activity with the substrate cinnamoyl-CoA, which may be an intermediate in the biosynthesis of BA through β-oxidation (Fig. 1). Because Chy1 is homologous to the P. fluorescens feruloyl-CoA hydratase/lyase, we also checked to see if Chy1 is capable of producing benzaldehyde directly from cinnamoyl-CoA or 3-hydroxy-3-phenoylpropionoyl-CoA in a reaction analogous to that catalysed by P. fluorescens feruloyl-CoA hydratase/lyase (Fig. 1).
We used His-tagging and affinity chromatography to purify the E. coli-expressed Arabidopsis Chy1 protein (Fig. 5). Incubation of cinnamoyl-CoA with purified Chy1 led to the production of free cinnamic acid, and incubation of 3-hydroxy-3-phenoylpropionoyl-CoA with Chy1 led to the production of free 3-hydroxy-3-phenoylpropionate. In addition, incubation of p-coumaroyl-CoA with Chy1 led to the production of free p-coumaric acid. In contrast, Chy1 had no hydrolase activity with benzoyl-CoA, consistent with the previously published report (Zolman et al. 2001).
The Km of Chy1 with cinnamoyl-CoA was 2.9 ± 1.0 μm, with a Vmax of 12.6 ± 1.9 μm·s−1·mg−1 (Fig. 6). The enzyme had a broad pH optimum of 7.0 (Fig. 7), and the cations Mg2+, Fe2+, Cu2+, Zn2+, and Mn2+ at concentrations 1 mm had no effect on the hydrolase activity of Chy1, while Cu2+ at this concentration caused a >90% inhibition of Chy1 enzyme activity.
The biosynthesis of BA in plants has remained unclear despite repeated attempts at elucidation (Wildermuth 2006). While labeling experiments have given conflicting results (summarized in Wildermuth 2006), no clear enzymatic reactions responsible for the shortening of the propyl side chain have been conclusively demonstrated. In addition to the lack of biochemical data, mutants directly defective in the synthesis of BA have not yet been reported, suggesting that perhaps the BA biosynthetic pathway (or pathways) is either essential to plant viability or that it contains redundancies.
Benzoic acid biosynthesis in developing Arabidopsis seeds for the production of benzoylated glucosinolates appears to constitute a system that is amenable to genetic analysis. A mutant that lacks seed benzoylated glucosinolates has recently been found in Arabidopsis, and further analysis identified the mutated gene as BZL1 (At1g65880) (Kliebenstein et al. 2007). Biochemical analysis suggested that BZL1 is capable of catalysing the formation of BA-CoA using BA, CoA, and ATP (Kliebenstein et al. 2007). BA-CoA is believed to be the substrate of a benzoyltranferase that links the benzoyl moiety to the glucosinolate moiety (Graser et al. 2001; Kliebenstein et al. 2007). Interestingly, BZL1 has a peroxisomal targeting sequence at its C-terminus (Kliebenstein et al. 2007), although the synthesis of benzoylated glucosinolates is believed to be in the cytosol, not the peroxisomes (Chen & Andreasson 2001).
Here, we show that mutations that inactivate the Chy1 protein, previously demonstrated to be in peroxisomes, also lead to a deficiency in benzoylated glucosinolates in developing Arabidopsis seed. Extracts of seeds of chy1 mutants also contain no BA after hydrolysis, suggesting that the BA detected in similarly treated extracts of wild-type seeds is derived from hydrolysis of benzoylated glucosinolates. The absence of BA in chy1 mutant seeds suggests that its biosynthesis occurs, at least partly, in the peroxisomes, although some reactions in this pathway may occur elsewhere.
However, this observation by itself does not distinguish between the possibilities that BA is synthesized viaβ-oxidation or retro-aldol cleavage. Since Chy1 is homologous both to the hydrates domain of the multi-functional protein (MFP) of β-oxidation, as well as to P. fluorescens feruloyl-CoA hydratase/lyase, we considered two competing hypotheses: that Chy1 binds cinnamoyl-CoA and hydrates it to 3-hydroxy-3-phenoylpropionoyl-CoA, which is then further metabolized through the β-oxidation pathway, or that it uses cinnamoyl-CoA to produce benzaldehyde, perhaps through a 3-hydroxy-3-phenoylpropionoyl-CoA intermediate (Fig. 1). Since both hypotheses postulate that cinnamoyl-CoA is the substrate of Chy1, we incubated Chy1 with cinnamoyl-CoA, but the only product of this reaction observed was free cinnamic acid. Incubation of Chy1 with 3-hydroxy-3-phenoylpropionoyl-CoA or with p-coumaroyl-CoA also gave the respective free acids.
These results suggest that Chy1 acts in vitro primarily as a hydrolase. Zolman et al. (2001) previously reported that Chy1 could specifically hydrolyse β-hydroxyisobutyryl-CoA but no other acyl-CoAs tested, including BA-CoA and several other straight-chain and branched-chained acyl-CoAs. Cinnamoyl-CoA, however, was not tested in that study. The Km for β-hydroxyisobutyryl-CoA was 3.7 μm, similar to the Km we measured for Chy1 with cinnamoyl-CoA (2.9 μm). Since Chy1 mutants are defective in β-oxidation of fatty acids as well as the conversion of the IAA precursor IBA to IAA, Zolman et al. (2001) hypothesized that the mutation in Chy1 leads to accumulation in the peroxisomes of methylacrylyl-CoA, a precursor of β-hydroxyisobutyryl-CoA in the valine degradation pathway, to toxic levels and that the β-oxidation pathway is consequently inhibited. Lange et al. (2004) presented labeling data that suggest that valine degradation indeed occurs in the peroxisomes, but recent results presented by Lucas et al. (2007) concluded that valine degradation in plants is more likely to occur in the mitochondria and not in the peroxisomes. While our results do not resolve this question, we show here that cinnamoyl-CoA is as good a substrate of Chy1 as β-hydroxyisobutyryl-CoA. We note, however, that hydrolysis of cinnamoyl-CoA is not part of either of the two alternative pathways to BA depicted in Fig. 1. One possibility is that the activity of the Chy1 protein on cinnamoyl-CoA changes when associated with other proteins. Alternatively, Chy1 exerts its influence on BA synthesis indirectly. Thus, while our data strongly suggest that BA is synthesized at least partly in the peroxisomes, and further indicate that Chy1 has a strong affinity for cinnamoyl-CoA, the exact role that Chy1 plays in BA biosynthesis remains to be determined.
The authors thank Dr Bonnie Bartel for the gift of chy1-1 and chy1-3 mutants and for valuable discussion. We also thank Drs. Jonathan Gershenzon and Michael Reichelt for providing glucosinolate standards as well as advice on the isolation and analysis of glucosinolates.