Analytical Biochemistry 302, 305–312 (2002) doi:10.1006/abio.2001.5574, available online at http://www.idealibrary.com on

Enzymatic Synthesis and Purification of Aromatic Esters1

Till Beuerle and Eran Pichersky2 Department of Molecular, Cellular and Developmental Biology, University of Michigan, 830 North University Street, Ann Arbor, Michigan 48109-1048

Received November 14, 2001; published online February 13, 2002

ipate in reductive reactions catalyzed by oxidoreducta- Two recombinant His-tagged proteins, a plant 4-cou- ses (3) and in aldol-type reactions catalyzed by Claisen marate:coenzyme A ligase (EC 6.2.1.12) and a bacterial (1). benzoate:coenzyme A ligase (EC 6.2.1.25), were ex- The 4CL catalyzes the formation of CoA pressed in Escherichia coli and purified in a single thioesters of hydroxy-cinnamic acids in a process uti- step using Ni-chelating chromatography. Purified en- lizing ATP (4). These activated hydroxy-cinnamic acids zymes were used to synthesize cinnamoyl-coenzyme A serve as building blocks for numerous secondary com- (CoA), p-coumaroyl-CoA, feruloyl-CoA, caffeoyl-CoA, and benzoyl-CoA. Conversions up to 95% were pounds including flavonoids or anthocyanins (5, 6), achieved. Using a rapid solid-phase extraction proce- (7), and other phenolic products (8–10) which dure, the target CoA esters were isolated with yields of fulfill diverse functions as phytoalexins, cell wall com- up to 80%. Structures were confirmed by analytical ponents, UV protectants, flavor and defense com- comparison with chemically synthesized reference pounds, or pigments. Since 4CL is involved in lignifi- compounds and electrospray ionization–mass spec- cation, a pathway found in practically all plants, this trometry. The recombinant enzymes were stable for enzyme has been extensively characterized from nu- several months at ؊80°C, thus providing a reliable and merous plant species, and many variants of this gene facile method to produce these delicate biological in- are available (11–13). termediates. © 2002 Elsevier Science (USA) BZL catalyzes the analog reaction of Key Words: coenzyme A; electrospray ionization– with CoA. So far this activity has been best described mass spectrometry; affinity chromatography; solid- in the benzoate degradation pathway in microorgan- phase extraction; 4-coumarate:CoA ligase; benzoate: isms (14, 15). In plants, benzoyl-CoA is reported as a CoA ligase; recombinant proteins. substrate in various enzymatic benzoylations in the biosynthesis of natural compounds such as cocaine (16), Taxol (17), dianthramide B (18), benzoylated glu- CoA3 thioesters represent an important class of ac- cosinolate esters in Arabidopsis thaliana (19), or ben- tivated intermediates in various biological pathways. zylbenzoate in Clarkia breweri (20). Although no BZL This type of activation can facilitate the transfer of the has yet been characterized from plants, an enzyme acylated moiety by enzymes known as acyltransferases that catalyzes the formation of 3-hydroxybenzoyl-CoA, (1, 2). Additionally, acylated intermediates can partic- an intermediate in the biosynthesis of xanthone, in cell cultures of Centaurium erythraea has been reported, but the gene encoding this enzyme was not isolated 1 This work was supported by a National Science Foundation (21). Grant MCB-9974463 to E.P. and by a DAAD fellowship (Gemein- sames Hochschulprogramm III von Bund und La¨ndern) to T.B. During our investigations of coumaric acid and ben- 2 To whom correspondence and reprint requests should be ad- zoic acid metabolism in plants, we encountered the dressed. Fax: 1-734-647-0884. E-mail: [email protected]. need to generate CoA esters of the above mentioned 3 Abbreviations used: CoA, coenzyme A; BZL, benzoate:CoA ligase; aromatic acids. We initially followed protocols for 4CL, 4-hydroxyxinnamate:CoA ligase; DTE, dithioerythritol; LB me- chemical synthesis of these compounds, but the estab- dium, Luria–Burrous medium; IPTG, isopropyl ␤-D-thiogalactopyr- anoside; SPE, solid-phase extraction; PAL, ammonia- lished protocols have many drawbacks: they involve lyase. multiple steps, they result in low yields with more side

0003-2697/02 $35.00 305 © 2002 Elsevier Science (USA) All rights reserved. 306 BEUERLE AND PICHERSKY products, and they require sophisticated lab equip- structs were transformed into BL21(DE3)pLysS cells ac- ment. An enzymatic approach allows a single-step re- cording to the manufacturer’s instructions. A similar ex- action under mild conditions in an aqueous solvent pression and purification protocol for both proteins was system. established. A published procedure to synthesize some of these A single isolated bacterial colony from freshly chemicals by enzymatic means took advantage of the streaked plates (grown on LB agar medium containing high levels of 4CL in wheat seedlings, but because the 50 ␮gmlϪ1 ampicillin and 34 ␮gmlϪ1 chloramphenicol) investigators used crude seedling extract, the product was used to inoculate 10-ml liquid cultures in LB me- yield was relatively low, due mostly to thioesterase dium containing 50 ␮gmlϪ1 ampicillin and 34 ␮gmlϪ1 activity (22). Here we present a facile and reliable chloramphenicol and grown overnight at 37°C. One method to enzymatically synthesize (hydroxy)cinna- aliquot of 1 ml of each culture was used to inoculate moyl-CoA and benzoyl-CoA esters that results in high 50-ml liquid cultures containing 50 ␮gmlϪ1 ampicillin. yields (up to 80%) and high specific activities, taking Once the cultures reached a cell density of 0.4–0.5 advantage of cloned 4CL and BZL and an Escherichia OD600, recombinant protein expression was induced by coli expression system. the addition of 0.8 mM isopropyl ␤-D-thiogalactopyr- anoside (IPTG), and the culture was grown for 20–24 h MATERIALS AND METHODS at room temperature. Cells were harvested by centrif- ugation at 3000g for 10 min at 4°C. Pellets were resus- Chemicals, Solvents, and Reagents pended in 10 times the volume with buffer of 50 mM Chemicals, solvents, and reagents were purchased Bis–Tris, pH 7.0, containing 10% glycerol, 2 mM DTE, from Sigma, Fluka, and Aldrich (St. Louis, MO) unless 1 mM EDTA, 10 mM NaCl. After three cycles of freeze/ otherwise stated. Benzoyl-CoA, coenzyme A thaw at Ϫ80°C/37°C, cells were disrupted by three 20-s 14 salt, and L-[U- C]phenylalanine with a specific activity intervals of sonication on ice. The resulting homoge- of 460 mCi/mmol and [7-14C]benzoic acid with a specific nate was centrifuged at 20,000g for 10 min to pellet the activity of 16.6 mCi/mmol were purchased from Sigma. debris. The supernatant was assayed for activity and Acetonitrile UV-grade and liquid scintillation cock- stored at Ϫ80°C prior to protein purification. tail Econo-Safe were purchased from Burdick & Jack- Since Ni2ϩ chelating chromatography is incompatible son (Muskegon, MI) and Research Products Interna- with sulfur-containing reducing agents and EDTA, the tional (Mount Prospect, Il), respectively. buffer was exchanged to binding buffer conditions (5 mM imidazole, 500 mM NaCl, 20 mM Tris–HCl, pH Source of 4CL and BZL Genes 7.9) using PD-10 columns (Amersham Pharmacia, Pis- cataway, NJ). Plasmids pQE-19(11) and pPE204(14), containing Hi-Trap chelating columns of 1 ml bed vol (Amer- the coding region of tobacco 4CL and BZL, were the sham Pharmacia) were conditioned with 10 ml of wa- kind gifts of Drs. C. J. Douglas and C. S. Harwood, ter, 5 ml of charging buffer (50 mM NiSO ),and5mlof respectively. 4 binding buffer (5 mM imidazole, 500 mM NaCl, 20 mM Tris–HCl, pH 7.9). After loading the protein solution Cloning, Expression, and Purification of 4CL and BZL (3.5 ml in binding buffer) the column was rinsed with Primers were designed to remove the native stop 10 ml of binding buffer and 8 ml of washing buffer (80 codon and place the gene of interest in frame with the mM imidazole, 500 mM NaCl, 20 mM Tris–HCl, pH DNA encoding a C-terminal peptide containing a poly- 7.9). His-tagged 4CL was eluted with elution buffer histidine region. The gene for 4CL was amplified from (400 mM imidazole, 500 mM NaCl, 20 mM Tris–HCl pQE-19 using the primer pair 4CL-CT-His 5Ј-ATG- pH 7.9). His-tagged BZL was eluted with elution buffer GAGAAAGATACAAAACAGG and 4CL-CT-His 3Ј-AT- containing1Mimidazole. Fractions of 2.5 ml were TTGGAAGCCCAGCAGCC and the following PCR pro- collected throughout the procedure, but only the first gram: 94°C 1 min, 30 cycles 94°C 0.5 min, 54°C 0.5 400mMand1Mimidazole fractions contained 4CL min, 72°C 2 min, and then 72°C for 7 min. The gene for and BZL, respectively. For long-time storage, the BZL was amplified from pPE204 in a reaction mix buffer was changed to 50 mM Bis–Tris pH 7.0, 10% containing 10% dimethyl sulfoxide using the primer glycerol, 2 mM DTE using PD-10 columns, and the pair BZL-CT-His 5Ј-ATGAATGCAGCCGCGGTCAC and sample was stored at Ϫ80°C. BZL-CT-His 3Ј-GCCCAACACACCCTCGCG and follow- ing the PCR program: 96°C 1 min, 30 cycles 96°C 0.5 min, Enzyme Activity Assays 54°C 0.5 min, 72°C 2 min, and then 72°C for 7 min. The amplified genes were transferred into the pCRT7/ Cinnamic and (hydroxy)cinnamic acid:CoA ligase ac- CT-TOPO expression vector (Invitrogen Inc., Carlsbad, tivity assay. Enzyme activity was measured spectro- CA) according to the manufacturer’s instructions. Con- photometrically at room temperature with a 1-ml mix- SYNTHESIS OF COENZYME A ESTERS 307 ture containing 100 mM Tris–HCl, pH 7.5, 2.5 mM terminated by extraction (3ϫ) with ethylacetate (phase

MgCl2, 2.5 mM ATP, 0.2 mM (hydroxy)cinnamic acids separation achieved by centrifugation). CoA esters and 0.2 mM CoA. The assay was started by the addi- were purified using solid-phase extraction cartridges tion of CoA. The change in absorbance of the reaction (1000 mg Chromabond C 18 ec, Macherey-Nagel) pre- mixture was monitored at the wavelengths of 311, 333, conditioned with consecutive washes of methanol,

345, 346, and 352 nm according to the reported absorp- dH2O, and 4% ammonium acetate solution (5 column tion maxima for cinnamoyl-CoA, p-coumaroyl-CoA, vol each). After evaporation of the organic solvent, feruloyl-CoA, caffeoyl-CoA, and sinapoyl-CoA, respec- ammonium acetate was added to the water phase to a tively (23, 24). final concentration of 4%, and the mixture was loaded Benzoyl:CoA ligase activity assay. Radioisotopic as- onto the SPE cartridge. The column was rinsed with says were performed with 100 ␮l buffer containing 100 4% ammonium acetate solution until the flowthrough mM Tris–HCl, pH 7.5, 2.5 mM MgCl2, 2.5 mM ATP, 0.2 showed the absence of free CoA (determined by spec- mM CoA, and 10,000 to 50,000 dpm [7-14C]benzoic acid. trophotometry). The CoA esters were recovered by elu- The assay was started by the addition of CoA and kept tion with distilled water. Fractions containing the CoA at room temperature for 15 to 120 min. The reaction esters were identified by their UV spectrum and lyoph- was stopped by the addition of 5 ␮l 50% trichloracetic ilized overnight. Yields varied between 15 and 45%. acid and 180 ␮l , vortexed, and phase The purity was checked by TLC and reverse-phase separated by a 1-min centrifugation at 14,000g. The HPLC analysis with UV detection at 216 nm on a upper organic phase was removed and the ethyl ace- Nova-Pak column (Nova-Pak C 18 60A 4 ␮m, 3.9 ϫ 300 tate extraction was repeated. The remaining aqueous mm, Waters, Milford, MA) and was higher than 90%. phase was counted in a liquid scintillation counter. The TLC was performed on Silicagel plates using the sol- amount of radioactivity in the aqueous phase indicated vent system 1-butanol:water:acetic acid (60:35:25), R f’s the amount of synthesized benzoyl-CoA. of 0.37, 0.4, 0.45, 0.54, and 0.56 were observed for Thioesterase activity assay. Reactions (total sample caffeoyl-CoA, p-coumaroyl-CoA, feruloyl-CoA, cinna- volume 1 ml) containing 100 mM Tris–HCl, 2.5 mM moyl-CoA, and benzoyl-CoA, respectively. For HPLC, a flow of 1 ml/min and a gradient of solvent A (acetoni- MgCl2, 0.05 mM cinnamoyl-CoA, and either 50 ␮g crude supernatant of BL21(DE)LysS cells containing trile) and solvent B (20 mM KH2PO4, pH 2.9) were an empty vector or 50 ␮g purified 4CL were incubated applied: 0–5 min, 5% A; 5–32 min, 5–38% A linear; at room temperature. The reaction was started by ad- 32–35 min, 38–75% A linear; 35–40 min, 75–5% A dition of cinnamoyl-CoA. The decrease in absorbance linear; 40–45 min 5% A. R t’s of 16.1, 18.4, 19.2, 19.8, at 311 nm over time was recorded. Thioesterase activ- and 21.9 min were recorded for caffeoyl-CoA, benzoyl- ity for benzoyl-CoA was measured in a similar assay to CoA, p-coumaroyl-CoA, feruloyl-CoA, and cinnamoyl- that described above. Released radiolabeled benzoic CoA, respectively. CoA esters were stored in solution at acid was extracted with ethyl acetate and counted in a pH 6 or lyophilized and stored over Drierite for several liquid scintillation counter. months at 80°C without noticeable degradation. Chemical synthesis and purification of cinnamoyl- Electrospray ionization–mass spectrometry (ESI-MS) CoA, p-coumaroyl-CoA, caffeoyl-CoA, feruloyl-CoA, and analysis of hydroxy-cinnamoyl and benzoyl-CoA esters. benzoyl-CoA. Coenzyme A esters of cinnamic and sev- For ESI-MS analysis, purified CoA esters were diluted eral (hydroxy)cinnamic acids were synthesized by a in distilled water at a concentration of 0.1 mM. Sam- modified imidazolide method. Imidazolides of the cor- ples were analyzed on a Micromass Quattro LC (Mi- responding acids were prepared as previously de- cromass, Beverly, MA) by direct infusion with a flow ␮ Ϫ1 scribed (25). The reaction was monitored for purity and rate of 8 lmin . Samples were ionized in positive and completion by TLC (Polygram Sil G/UV, Macherey- negative mode with a capillary voltage of 2.9 and 3.5 V, Nagel, Easton, PA), developed in diethylether acidified respectively. Nitrogen was used for nebulization (65 Ϫ1 Ϫ1 with 1% acetic acid, and visualized under UV light. liters h ) and desolvation gas (420 liters h , 250°C). Reactions were allowed to proceed until the disappear- The source temperature was set at 120°C and cone ance of the acid on the TLC. The acid imidazolides were voltages between 20 to 30 V were applied. Standard used without further purification for the next reaction mass spectra were recorded for a total of 30 s with a step. Corresponding CoA esters were obtained by a total scan time of 2 s over the mass range from 200 to modification of the method of Pabsch et al. (25). Acid 1000 m/z. Data acquisition and evaluation were con- imidazolides were used in twofold excess compared to ducted with MassLynx version 3.5 software (Micromass, the CoA-sodium salt. The reaction was monitored with Manchester, UK) on a personal computer. TLC (Silicagel) and a solvent system of n-butanol:ace- Preparative enzymatic synthesis of p-coumaroyl-CoA. tic acid:water 63:10:27. CoA esters were identified with To synthesize coumaroyl-CoA, 3.3 mg coumaric acid, 2 delayed nitroprusside reaction (26). The reaction was mg CoA, and 6.9 mg ATP were dissolved in a total 308 BEUERLE AND PICHERSKY

volume of 10 ml of 50 mM Tris–HCl buffer containing three times with 250 ␮l diethylether to remove resid-

2.5 mM MgCl2. The reaction was started by the addi- ual cinnamic acid. After evaporating the remaining tion of 0.25 mg purified 4CL. After5hatroom tem- organic solvent, ammonium acetate was added to a perature, 6.9 mg ATP, 2 mg CoA, and 0.25 mg enzyme final concentration of 4% (w/v). The aqueous solution were added and the reaction continued. After an addi- was loaded on a preconditioned 100-mg SPE cartridge tional 12 h, 0.4 g of ammonium acetate was added and (Chromabond C 18 ec, Macherey-Nagel) as described the crude enzyme reaction was loaded on precondi- above. The column was successively rinsed with 5 ml of

tioned 1000 mg C18 SPE cartridge and purified as 4% ammonium acetate solution to remove unreacted described above. Four 5-ml fractions containing p-cou- CoA and protein and with 5 ml of distilled water to maroyl-CoA were lyophilized and 3.4 mg p-coumaroyl- elute the CoA ester. Fractions of 1 ml were collected. CoA were obtained, which represented 80% yield based Most of the radioactivity was recovered in one fraction. on CoA used. p-Coumaroyl-CoA was characterized and The overall yield in this procedure was Ͼ80% based on stored as described above. radioactivity. Chemical and radiochemical purity of 14 more 95% was verified with HPLC and TLC (Silicagel, Enzymatic synthesis of [U- C]cinnamic acid. L-[U- 14C]Phenylalanine was dissolved in up to 600 ␮lof30 R f 0.4) and authentic nonlabeled standards, as de- mM Tris–HCl buffer, pH 8.5, containing 0.05 Units scribed above. In addition, purity was also checked by of phenylalanine ammonia-lyase (PAL, EC 4.3.1.5, hydrolysis of the reaction product at pH 12 (NaOH) for Sigma) at 30°C. Small aliquots of the reaction mixture 30minat60°C and detection of liberated cinnamic acid were acidified and extracted with ethyl acetate to check with HPLC and TLC as descried above. 14 for reaction completion. The ratio of counts in the or- Enzymatic synthesis of [7- C]benzoyl-CoA. A total ganic phase (cinnamic acid) vs the aqueous phase (phe- volume of 500 ␮l of 100 mM Tris–HCl, pH 8.5, contain- ␮ 14 nylalanine) indicated the conversion of L-phenylala- ing 2.5 mg CoA, 1.7 mg ATP, 25 Ci [7- C]benzoic acid ␮ nine to cinnamic acid. Reaction times varied from 18 to (16.6 mCi/mmol, Sigma), and 50 g purified BZL was 72 h. The reaction was performed on different scales of incubated at room temperature. The conversion was 10, 25, and 250 ␮Ci, with yields between 85 and 93%. checked every hour by extracting a small portion as The mixture was acidified with 20 ␮l of 6 N HCl and described above. After 4 h, 95% was converted to ben- extracted (3ϫ) with 300 ␮l each to isolate [U-14C]cin- zoyl-CoA and the reaction was stopped by the addition ␮ ϫ ␮ namic acid. The organic phases were combined and of 75 l6NHCl,followed by 3 extraction with 350- l concentrated to dryness and then redissolved in a portions of diethylether. Benzoyl-CoA was purified from the aqueous phase with SPE as described for small volume of 25% ethanol and stored at 4°C. Chem- 14 ical and radiochemical purity were checked with TLC [U- C]cinnamoyl-CoA. Chemical and radiochemical purity of more 95% was verified with HPLC and TLC (Silicagel, R f 0.73) using the solvent system pentane: 14 diethylether:acetic acid (100:100:1) and by reverse- (Silicagel, R f 0.35) as described for [U- C]cinnamoyl- phase HPLC analysis with UV detection at 216 and CoA. The overall yield was 75%, based on recovered 270 nm. The HPLC was operated at a flow rate of 1 radioactivity. ml/min and a gradient of solvent A (acetonitrile) and RESULTS solvent B (water, pH 2.9, adjusted with H2SO4) was applied: 0–5 min, 5% A; 5–45 min, 5–75% A linear; Thioesterase Activity in Crude Extracts of E. coli 45–50 min, 75–100% A linear; 50–53 min, 100% A The soluble protein fraction of cell-free extracts of E. linear; 53–55 min, 100–5% A linear; 55–60 min, 5% A. coli BL-21 cells expressing 4CL or BZL showed high Fractions were collected every minute and counted 14 activity of the corresponding CoA ligase activity. How- in a liquid scintillation counter. [U- C]cinnamic acid ever, attempts to use the crude extracts for the synthe- coeluted with an authentic reference compound at R t sis of CoA esters resulted in unsatisfactory yields of 25.7 min. Ͻ40%, and the yield could not be increased by varying Enzymatic synthesis of [U-14C]cinnamoyl-CoA. An the reaction parameters. Incubation of cinnamoyl-CoA almost complete enzymatic conversion of cinnamic acid with crude extracts of BL-21 cells containing an empty to cinnamoyl-CoA was achieved after 90 min at room expression vector indicated that soluble thioesterase temperature. The reaction mixture contained 50 mM activity (0.26 nkatal/mg crude protein extract) was Tris–HCl, pH 7.5, 2.5 mM MgCl2, 2.5 mM ATP, 0.2 mM present in these cells. CoA, 25 ␮Ci [14C-U]cinnamic acid, and 50 ␮g of purified It appeared that poor yields using crude enzyme 4CL in a total volume of 500 ␮l. The conversion rate preparations might therefore reflect a balance between was checked as described above. The ratio of counts in CoA ligase and thioesterase activity. Crude enzyme the aqueous phase vs the organic phase was calculated extracts might also contain other ATP-utilizing en- as conversion into cinnamoyl-CoA. The reaction mix- zymes which compete for the limited amount of ATP. ture was acidified with 50 ␮l 6 N HCl and extracted Once the supply of ATP is used up, no new ligation SYNTHESIS OF COENZYME A ESTERS 309

the preferred substrate of 4CL has been reported to be , followed by p-coumaric acid, cinnamic acid, and (11), the His-tagged protein had a slightly different specificity, being more active with ferulic acid than with p-coumaric acid (Table 1). This observed difference in relative activity might be due of the introduction of the C-terminal His-tag, or the orig- inal report may be in error as a result of the presence of bacterial thioesterase activity in crude extracts that may have degraded the hydroxy-cinnamoyl-CoA esters FIG. 1. SDS–PAGE analysis of purified recombinant BZL and 4CL. (Lane A) Molecular mass markers; (lane B) soluble extract from at different rates. At any rate, the difference in activity uninduced BL21(DE3)pLysS cells harboring the His-tagged BZL of 4CL for the substrates can be neglected in practice, gene construct (10 ␮g protein); (lane C) soluble extract from because it can be easily overcome by adding more en- BL21(DE3)pLysS cells harboring the His-tagged BZL gene construct zyme to the reaction. 20 h after induction with IPTG (10 ␮g protein); (lane D) recombinant BZL after Ni2ϩ-chelating chromatography (2 ␮g protein); (lane E) soluble extract from uninduced BL21(DE3)pLysS cells harboring the Preparative Enzymatic Synthesis of p-Coumaroyl-CoA His-tagged 4CL gene construct (10 ␮g protein); (lane F) soluble extract from BL21(DE3)pLysS cells harboring the His-tagged 4CL We chose the synthesis of p-coumaroyl-CoA to dem- gene construct 20 h after induction with IPTG (20 ␮g protein); onstrate the use of an enzymatic approach to generate (lane G) recombinant 4CL after Ni2ϩ-chelating chromatography (2 ␮g CoA esters. Our procedure, detailed under Materials protein). and Methods, gave a 80% yield after purification, based on the amount CoA utilized (Table 2A). reaction occurs, while the already formed CoA esters continue to be cleaved by the thioesterase. Since 4CL Enzymatic Synthesis of Labeled Cinnamoyl-CoA is inhibited at ATP concentrations higher than 5 mM, this problem cannot be simply solved by adding large Since radiolabeled cinnamic acid was not commer- amounts of ATP. cially available, we combined two enzymatic steps to synthesize U-14C-labeled cinnamoyl-CoA. Starting 14 14 Purification of 4CL and BZL by Affinity from L-[U- C]phenylalanine, [U- C]cinnamic acid was Chromatography obtained utilizing the enzyme PAL (EC 4.3.1.5) from Rhodotorula glutinis. Subsequent conversion to cin- To circumvent these difficulties, we used His-tagging namoyl-CoA was achieved using purified recombinant and affinity chromatography to purify the expressed 4CL. The product obtained showed high chemical and heterologous proteins. Most importantly, no thioester- radiochemical purity (Table 2A). ase activity was detectable after the affinity chroma- ␮ tography step. When 2 g of the purified protein frac- 14 tions was analyzed by SDS–PAGE and Coomassie blue Enzymatic Synthesis of [7- C]Benzoyl-CoA staining of the gel, the purified BZL and 4CL samples The radiolabeled benzoic acid that we used had a both were shown to contain only one major band of the relative low specific activity. Since we limited ourselves expected size for these ligases (Fig. 1). The molecular for practical reasons to a relatively small, easy to han- weight of the purified His-tagged BZL protein was estimated from the SDS–PAGE analysis to be 61.5 kDa, and that of the His-tagged 4CL to be 63 kDa (Fig. TABLE 1 1), corresponding to the calculated molecular weights Substrate Specificity of Recombinant Purified His-Tagged of 60,115 and 62,900 Da, respectively. In both samples, Tobacco 4CL Compared to Crude Extracts of E. coli Express- one additional, unidentified protein band of approxi- ing Recombinant Tobacco 4CL (11) mately 75 kDa was barely detected (in this system, the minimum detection level is 50 ng protein per band). Relative activity (%) of Relative activity (%) of In our purification procedure, 50-ml cultures typi- recombinant His-tagged recombinant tobacco 4CL cally yielded 3.4 mg of recombinant 4CL and 3.5 mg Substrate tobacco 4CL as reported in (11) recombinant BZL. Expression in E. coli and enzyme p-Coumaric acid 100 100 purification were repeated several times with similar Ferulic acid 109 60 results. The enzymes were stored at Ϫ80°C in a buffer Cinnamic acid 74 25 containing 10% glycerol and were stable for several Caffeic acid 64 20 month and multiple freeze/thaw cycles. Sinapic acid 0 0 Recombinant 4CL was also tested with cinnamic, Note. The specific enzyme activity of purified 4CL with coumaric ferulic, caffeic, and sinapic acid as substrates. While acid as substrate was 31.3 nkatal/mg protein. 310 BEUERLE AND PICHERSKY

TABLE 2 obtained in aqueous solution ready to use in enzymatic assays or in in vivo labeling experiments. The products Acid moiety Yield (%) Amount (mg) Purity (%) of the large-scale syntheses of unlabeled CoA esters A. Yields, amount, and purity of enzymatically synthesized were lyophilized overnight and the compounds were CoA esters obtained as fluffy powders. Aqueous solutions and

14 a b dry powders of CoA esters were stored at Ϫ80°C over [U- C]Cinnamic acid 80 0.032 Ͼ95 p-Coumaric acid 80c 3.5 Ͼ95 Drierite without noticeable degradation for several [7-14C]Benzoic acid 75a 0.275b Ͼ95 months.

B. Yields, amount, and purity of CoA esters chemically synthesized via imidazolide activation ESI-MS of Hydroxy-cinnamoyl-CoA’s and Benzoyl-CoA Cinnamic acid 45 16 Ͼ95 p-Coumaric acid 15 5 Ͼ90 At cone voltages between 20 and 30 V in both ion- Benzoic acid 43 15 Ͼ95 ization modes (ESI positive and ESI negative), mass Caffeic acid 43 16 90 Ferulic acid 42 16 Ͼ90 spectra of hydroxy-cinnamoyl-CoA and benzoyl-CoA esters were dominated by ions that stand in good a Yield based radioactivity. agreement with calculated singly and doubly charged b Calculated, based on radioactivity. pseudo-molecular ions of the corresponding CoA ester c Yield based on CoA. (Fig. 2 and Table 3). At ESI negative mode (cone 30 V), in-source fragmentation occurred and an ion with a mass of 80 m/z lower than [M Ϫ H]Ϫ was observed (Fig. dle volume, we had to use higher concentrations of 2B, ion m/z 790.2), most likely representing loss of benzoic acid in our reaction (3 mM). Since BZL shows HPO3 from the phosphoadenosine moiety of the CoA product inhibition (80% at 0.1 mM benzoyl-CoA) (14), ester. The results of ESI-MS analysis of all hydroxy- we tested several conditions to optimize the conver- cinnamoyl and benzoyl-CoA esters described here are sion. Even though CoA esters are more likely to be summarized in Table 3. ESI-MS spectra of enzymati- cleaved in basic solutions, it proved to be necessary to cally synthesized CoA esters were identical to those raise the pH to 8.5 to increase the conversion rate. CoA esters obtained via the chemical method. Despite product inhibition, using excess of protein forced the reaction toward high yields (Table 2A). Thus, using larger volumes should provide a facile DISCUSSION route for a preparative synthesis of benzoyl-CoA. Despite some reports over failed synthesis of ␣,␤- unsaturated CoA esters using 1-acyl imidazolide acti- Chemical Synthesis of CoA Esters vation (29), the chemical method used to synthesize To verify the authenticity of products from the enzy- caffeoyl-CoA via the imidazolide (25) can be applied matic reactions, reference compounds were synthe- generally to synthesize other hydroxy-cinnamoyl-CoA sized using a method of imidazolide activation for the and benzoyl-CoA esters. Nevertheless, multiple steps acids, recently published for caffeoyl-CoA (25). Yields are necessary and our yields did not exceed 45%. and purity of all chemically synthesized CoA esters are Several reports have recently appeared describing summarized in Table 2B. the use of crude enzyme extracts of various source to generate hydroxy-cinnamoyl-CoA esters (22, 27). As previously noted, and as we learned from our own Solid-Phase Extraction Procedure experiments (unpublished data), crude plant extracts

C18 SPE using cartridges was found to be most effi- contain thioesterase activities that result in the unde- cient and convenient method to purify the products. sired cleavage of hydroxy-cinnamoyl and benzoyl-CoA Aromatic CoA esters bind to the column in the pres- esters. This makes it difficult to predict reaction con- ence of ammonium acetate, and rinsing the columns ditions or maximize conversion rates, and therefore with 4% ammonium acetate solutions washes off the yields are low and do not exceed 40% (22, 27). The use free and oxidized CoA and proteins. The CoA esters of purified enzymes expressed in E. coli is a way to were recovered by rinsing the column with distilled circumvent this problem. water. The progress of each rinsing step can be easily The tobacco 4CL protein we employed in our synthe- followed using a spectrophotometer. Under the above ses allowed us to generate preparative amounts of ci- described conditions, the excess hydroxy-cinnamic ac- nammoyl-CoA and coumaroyl-CoA and, based on our ids and benzoic acid were retained by the C18 material, preliminary experiments (Table 1), should also be use- thus separate from the target compounds. An addi- ful in the generation of preparative amounts of the tional advantage of this method is that CoA esters were caffeic and ferulic acid CoA esters. The Rhodopseudo- SYNTHESIS OF COENZYME A ESTERS 311 monas palustris BZL gene was reported to use benzoic TABLE 3 acid, halogenated benzoic acid derivatives, cyclohexen- Dominant Ions in ESI Mass Spectra of Hydroxy-cinnamoyl carboxylic acids, and pyridine carboxylic acids as sub- and Benzoyl-CoA Ester in Positive and Negative Mode strates (14), and therefore it is likely that our method ESI negativea should be capable of generating CoA esters of these Ϫ m/z of [M Ϫ H] / ESI positiveb acids as well. [M Ϫ 2H]2Ϫ/ m/z of [M ϩ H]ϩ/ Ϫ 2ϩ Together, the one-step enzymatic synthesis and the Compound M r [M Ϫ H Ϫ HPO3] [M ϩ 2H] one-step SPE purification described here are simple and efficient, providing an excellent method especially Benzoyl-CoA 871.6 870.2/434.7/790.2 872.2/436.8 Cinnamoyl-CoA 897.7 896.2/447.7/816.2 898.2/449.7 p-Coumaroyl-CoA 913.7 912.3/455.8/832.3 914.2/457.8 Caffeoyl-CoA 929.7 928.2/463.7/848.2 930.3/465.8 Feruloyl-CoA 943.7 942.2/470.7/862.2 944.3/472.8

a Cone voltage 30 V. b Cone voltage 20 V.

for the generation of radiolabeled CoA esters, where a minimum of handling is desirable. Furthermore, we were able to isolate these CoA esters using SPE to high purity in aqueous solvents, making them immediately usable in enzyme assays and in in vitro feeding ex- periments. We also used a two-step enzymatic method, starting 14 with L-[U- C]phenylalanine, to make radiolabeled cin- namoyl-CoA. Since PAL is also active with tyrosine (28), an analogous approach starting with radiolabeled L-tyrosine would lead to p-coumaroyl-CoA. The enzymatic syntheses as well as the isolation procedure can be easily scaled up to generate milli- gram amounts of CoA esters. Considering the fact that CoA is the most costly component and that the enzy- matic method described here uses much less CoA per product formed, this method is superior to the chemical method.

ACKNOWLEDGMENTS

We thank Drs. C. J. Douglas and C. S. Harwood for their kind gifts of 4CL and BZL, respectively.

REFERENCES 1. Mishra, P. K., and Drueckhammer, D. G. (2000) Coenzyme A analogues and derivatives: Synthesis and applications as mech- anistic probes of coenzyme A ester-utilizing enzymes. Chem. Rev. 100, 3284–3309. 2. St-Pierre, B., and De Luca, V. (2000) in Recent Advances in Phytochemistry Evolution of Metabolic Pathways (Romeo, J. T., Ibrahim, R., Varin, L., and De Luca, V., Eds.), Vol. 34, pp. 285–317, Elsevier, Oxford. FIG. 2. Formular and ESI-MS spectra of benzoyl CoA. (A) Formula 3. Lacombe, E., Hawkins, S., VanDoorsselaere, J., Piquemal, J., and calculated molecular weight of benzoyl-CoA; (B) ESI-MS of ben- Goffner, D., Poeydomenge, O., Boudet, A. M., and GrimaPette- zoyl-CoA in negative mode (cone 30 V) with singly and doubly nati, J. (1997) Cinnamoyl CoA reductase, the first committed charged pseudo molecular ion [M Ϫ H]Ϫ at m/z 870.2 and [M Ϫ 2H]2Ϫ enzyme of the lignin branch biosynthetic pathway: Cloning, ex- Ϫ pression and phylogenetic relationships. Plant J. 11, 429–441. 434.7, respectively, as well as the fragment ion [M Ϫ H Ϫ HPO3] ; (C) ESI-MS of benzoyl-CoA in positive mode (cone 20 V) with singly 4. Gross, G. G., and Zenk M. H. (1974) Isolation and properties of and doubly charged pseudo molecular ion [M ϩ H]ϩ at m/z 872.2 and hydroxycinnamate: CoA ligase from lignifying tissue of For- [M ϩ 2H]2ϩ at m/z 436.8. sythia. Eur. J. Biochem. 42, 453–459. 312 BEUERLE AND PICHERSKY

5. Cooman, L. D., Everaert, E. S. W., Fache, P., Casteele, K. V., and 16. Bjorklund, J. A., and Leete, E. (1992) Biosynthesis of the benzoyl Van Sumere, C. F. (1993) biosynthesis in petals of moiety of cocaine from cinnamic acid via (R)-(ϩ)-3-hydroxy-3- Rhododendron simsii. Phytochemistry 33, 1419–1426. . Phytochemistry 31, 3883–3887. 6. Yabuya, T., Yamaguchi, M., Fukui, Y., Katoh, K., Imayama, T., 17. Walker, K., and Croteau, R. (2000) Taxol biosynthesis: Molecular and Ino, I. (2001) Characterization of anthocyanin p-coumaroyl- cloning of a benzoyl-CoA:taxane 2 alpha-O-benzoyltransferase transferase in flowers of Iris ensata. Plant Sci. 160, 499–503. cDNA from Taxus and functional expression in Escherichia coli. Proc. Natl. Acad. Sci. USA 97, 13591–13596. 7. Bernards, M. A., Susag, L. M., Bedgar, D. L., Anterola, A. M., and Lewis, N. G. (2000) Induced metabolism 18. Yang, Q., Reinhard, K., Schiltz, E., and Matern, U. (1997) Char- during suberization and lignification: A comparative analysis. acterization and heterologous expression of hydroxycinnamoyl/ Plant Physiol. 157, 601–607. benzoyl-CoA: anthranilate N-hydroxycinnamoyl/benzoyltrans- ferase from elicited cell cultures of carnation, Dianthus 8. Petersen, M., and Alfermann, A. W. (1988) Two new enzymes of caryophyllus L. Plant Mol. Biol. 35, 777–789. biosynthesis from cell-cultures of Coleus 19. Graser, G., Oldham, N. J., Brown, P. D., Temp, U., and Gersh- blumei—Hydroxyphenylpyruvate reductase and rosmarinic acid enzon, J. (2001) The biosynthesis of benzoic acid glucosinolate synthase. Z. Naturforsch. 43c, 501–504. esters in Arabidopsis thaliana. Phytochem. 57, 23–32. 9. Back, K., Jang, S. M., Lee, B. C., Schmidt, A., Strack, D., and 20. Dudareva, N., Raguso, R. A., Wang, J. H., Ross, E. J., and Kim, K. M. (2001) Cloning and characterization of a hydroxycin- Pichersky, E. (1998) Floral scent production in Clarkia breweri. namoyl-CoA:tyramine N-(hydroxycinnamoyl) transferase in- III. Enzymatic synthesis and emission of benzenoid esters. Plant duced in response to UV-C and wounding from Capsicum an- Physiol. 116, 599–604. nuum. Plant Cell Physiol. 42, 475–481. 21. Barillas, W., and Beerhues, L. (2000) 3-Hydroxybenzoate:coen- 10. Abe, I., Takahashi, Y., Morita, H., and Noguchi, H. (2001) Ben- zyme A ligase from cell cultures of Centaurium erythraea: Isola- zalacetone synthase—A novel polyketide synthase that plays a tion and characterization. Biol. Chem. 381, 155–160. crucial role in the biosynthesis of phenylbutanones in Rheum 22. Obel, N., and Scheller, H. V. (2000) Enzymatic synthesis and palmatum. Eur. J. Biochem. 268, 3354–3359. purification of caffeoyl-CoA, p-coumaroyl-CoA, and feruloyl-CoA. 11. Lee, D., and Douglas C. J. (1996) Two divergent members of a Anal. Biochem. 286, 38–44. tobacco 4-coumarate:coenzyme A ligase (4CL) gene family— 23. Gross, G. G., and Zenk, M. H. (1966) Darstellung und Eigen- cDNA structure, gene inheritance and expression, and proper- schaften von Coenzyme A-Thioestern substituierter Zimtsa¨uren. ties of recombinant proteins. Plant Physiol. 112, 193–205. Z. Naturforsch 21b, 688–690. 12. Ehlting, J., Shin, J. J. K., and Douglas, C. J. (2001) Identification 24. Sto¨ckigt, J., and Zenk, M. H. (1975) Chemical syntheses and of 4-coumarate:coenzyme A ligase (4CL) substrate recognition properties of hydroxycinnamoyl-coenzyme A derivatives. Z. domains. Plant J. 27, 455–465. Naturforsch 30c, 352–358. 13. Ehlting, J., Bu¨ ttner, D., Wang, Q., Douglas, C. J., Somssich, I. E., 25. Pabsch, K., Petersen, M., Rao, N. N., Alfermann, A. W., and and Kombrink, E. (1999) Three 4-coumarate:coenzyme A ligases Wandrey, C. (1991) Chemoenzymatic synthesis of rosmarinic in Arabidopsis thaliana represent two evolutionarily divergent acid. Rec. des Trav. Chim. Pays-Bas 110, 199–205. classes in angiosperms. Plant J. 19, 9–20. 26. Stadtman, E. R. (1957) in Methods in Enzymology, Vol. III, pp. 14. Egland, P. G., Gibson, J., and Harwood, C. S. (1995) Benzoate 931–941, Academic Press, New York. coenzyme A ligase, encoded by BADA, is one of 3 ligases able to 27. Huabin, M., and Campbell, W. H. (1997) Facile enzymic synthe- catalyze benzoyl-coenzyme A formation during anaerobic growth sis of caffeoyl CoA. Phytochemistry 44, 605–608. of Rhodopseudomonas palustris on benzoate. J. Bacteriol. 177, 28. Friederich, S., Maier, U., Deus-Neumann, B., Asakawa, Y., and 6545–6551. Zenk, M. H. (1999) Biosynthesis of cyclic bis(bibenzyls) in 15. Altenschmidt, U., Oswald, B., and Fuchs, G. (1991) Purification Marchantia polymorpha. Phytochemistry 50, 589–598. and characterization of benzoate-coenzyme-A ligase and 2-ami- 29. Semler, U., Schmidtberg, G., and Gross, G. G. (1987) Synthesis nobenzoate-coenzyme-A ligases from a denitrifying Pseudomo- of piperoyl coenzyme-A thioester. Z. Naturforsch. 42c, 1071– nas sp. J. Bacteriol. 173, 5494–5501. 1074.