Biosynthesis of ethyl caffeate via caffeoyl-CoA acyltransferase expression in Escherichia coli

Hydroxycinnamic acids (HCs) are natural compounds that form conjugates with diverse compounds in nature. Ethyl caffeate (EC) is a conjugate of caffeic acid (an HC) and ethanol. It has been found in several plants, including Prunus yedoensis, Polygonum amplexicaule, and Ligularia fischeri. Although it exhibits anticancer, anti-inflammatory, and antifibrotic activities, its biosynthetic pathway in plants still remains unknown. This study aimed to design an EC synthesis pathway and clone genes relevant to the same. Genes involved in the caffeic acid synthesis pathway (tyrosine ammonia-lyase (TAL) and p-coumaric acid hydroxylase (HpaBC)) were introduced into Escherichia coli along with 4-coumaroyl CoA ligase (4CL) and acyltransferases (AtCAT) cloned from Arabidopsis thaliana. In presence of ethanol, E. coli harboring the above genes successfully synthesized EC. Providing more tyrosine through the overexpression of shikimate-pathway gene-module construct and using E. coli mutant enhanced EC yield; approximately 116.7 mg/L EC could be synthesized in the process. Synthesis of four more alkyl caffeates was confirmed in this study; these might potentially possess novel biological properties, which would require further investigation.


Introduction
Hydroxycinnamic acids (HCs) are abundant in nature. They are synthesized via phenylpropanoid pathway in plants and serve as the building blocks for other phenolic compounds, such as flavonoids, coumarin, lignin, proanthocyanidins, cutin, and suberin [1]. HCs form conjugates with other molecules as well. The conjugate of caffeic acid (an HC) and quinic acid is called chlorogenic acid [2]. Avenanthramides are amides formed from HCs and anthranilate derivatives [3]. Besides these, conjugates of HCs with other compounds (spermine, puterine, tyramine, dopamine, tryptamine, and glycine) have also been reported [4,5].
Ethyl caffeate (EC) is a conjugate of caffeic acid and ethanol, and is found in some plants, namely Prunus yedoensis [6], Polygonum amplexicaule [7], and Ligularia fischeri [8]. In particular, Ligularia fischeri is grown in eastern Asia and is used in herbal medicine. The biological effects of L. fischeri are derived from EC. EC is known to exhibit anticancer [8], anti-inflammatory [9], and antifibrotic activities [10]. In addition, it has the potential to regulate blood pressure by inhibiting aldosterone synthase [11].
The synthetic pathway for EC in plants has not been fully elucidated yet. Caffeic acid is synthesized from tyrosine by deamination and hydroxylation, and needs to be activated by the attachment of coenzyme A (CoA). Although high-molecular-weight alcohols, such as C18, C20, and C22 alkan-1-ols, have been reported to be synthesized from long-chain fatty acids in plants [12], how the ethyl group is provided during the synthesis of EC still remains a mystery. Ethanol is presumably synthesized in hypoxic condition [13], which is also suitable for the synthesis of EC; however, other ethyl group donors may Open Access *Correspondence: jhahn@konkuk.ac.kr Department of Integrative Bioscience and Biotechnology, Bio/Molecular Informatics Center, Konkuk University, Seoul 05029, Republic of Korea also be present. The mechanism underlying the conjugation of caffeoyl-CoA with ethyl group donor still remains unknown. Previous study on the biosynthesis of suberin might provide a clue to the process of generating an EC conjugate. Suberin is an ester of long-chain primary alcohols and HCs [14]. The conjugate reaction is mediated by aliphatic suberin feruloyl transferase in Arabidopsis thaliana [15]. Therefore, an enzyme of this family might catalyze the conjugation reaction to form EC, although the origin of ethyl group in plants still remains unknown.
Microbial systems (mainly Escherichia coli and Saccharomyces cerevisiae) have been used to synthesize phytochemicals [16]. Among the variety of phytochemicals, phenolic compounds are mostly synthesized in microbial systems, since the biological synthetic genes have been characterized in many plants and hosts are engineered to provide more substrates [17]. Introduction of caffeic acid synthesis pathway genes into E. coli and engineering of the latter to provide more tyrosine resulted in the successful synthesis of caffeic acid. Tyrosine-overexpressing E. coli strain, due to overexpression of tyrA fbr , ppsA, tktA, and aroG fbr and deletion of pheA, tyrA, and/or tyrR genes, was employed for the reaction, and TAL (Tyrosine ammonia-lyase) from either Rhodotorula glutinis or Saccharothrix espanaensis and p-coumaric acid hydroxylase genes (HpaBC from E. coli or Sam5 from S. espanaensis) were introduced into it. The final titers of caffeic acid, reported in various publications, are quite different; one report used tyrosine-overproducing E. coli by introducing tyrA fbr , ppsA, tktA, and aroG fbr while deleting pheLA and tyrA genes, and overexpressing TAL. Tyrosine from R. glutinis, together with HpaBC, enabled 766.7 mg/L titer of caffeic acid production [18]. Another report introduced TAL from S. espanaensis and Sam5 into tyrosineoverproducing E. coli strain, in which tyrR was disrupted and tyrA fbr and aroG fbr were overexpressed. The strain synthesized 150 mg/L caffeic acid [19]. In this article, we reported the successful synthesis of EC in E. coli. We selected a gene encoding a protein that would synthesize EC from caffeoyl-CoA and ethanol (Fig. 1). Subsequently, we manipulated the shikimate pathway in E. coli to increase the synthesis of caffeic acid. By introducing genes for the conjugation of caffeic acid and ethanol into an engineered E. coli strain, and providing ethanol, we could successfully synthesize EC.

Constructs
Caffeoyl-CoA transferases from A. thaliana (AtCAT1 [AT5G63560.1] and AtCAT2 [At5g41040]) were cloned using reverse transcription-polymerase chain reaction (RT-PCR). The primers were as follows: aaacatATG GCC GAC TCA TTC G and aaggtaccTCA TAT ATC CAT AAT CCC TTGGA for AtCAT1, and aaacatATG GTT GCT GAG AAC AATAA and aaggtaccTTA TAT CTG TAA AAA CTG TTC TTG A for AtCAT2. The restriction enzyme recognition sites (NdeI and KpnI) are underlined in the primer sequences. The PCR product was sequenced to verify the nucleotide sequence. The previously cloned Os4CL [20] was subcloned into EcoRI and HindIII sites of pCDF-duet1 (Novagen) and the resulting construct was named pC-OS4CL. AtCAT1 and AtCAT2 were subcloned into NdeI/KpnI site of pC-OS4CL, respectively, and became pC-OS4CL-AtCAT1 and pC-OS4CL-AtCAT2, accordingly. The previously cloned SeTAL and HpaBC [21,22] were subcloned into EcoRI and Hin-dIII sites and NdeI and XhoI sites of pET-duet1 (Novagen), respectively, resulting in the construct named pE-SeTAL-HpaBC.

Synthesis of EC
Escherichia coli BL21 (DE3) transformant, containing pC-OS4CL-AtCAT1 (B-EC1 in Table 1), was grown in Luria-Bertani (LB) broth with 50 μg/mL spectinomycin overnight at 37 °C. The culture was inoculated into fresh LB medium and grown at 37 °C until the OD 600 reached 1.0. Thereafter, isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to the medium to a final concentration of 1 mM and the cells were incubated at 18 °C for 16 h. The amount of cells, corresponding to OD 600 of 3 in 1 mL, was harvested and resuspended in 1 mL of M9 medium containing 1% yeast extract, 2% glucose, 50 μg/ mL spectinomycin, and 1 mM IPTG. Caffeic acid and ethanol were also added to the medium resulting in a final concentration of 100µM and 1%, respectively. The culture was incubated at 30

Analysis of EC
The reaction product was purified using thin layer chromatography (TLC; TLC silica gel 60 F254; Millipore, Burlington, MA, USA). The mobile phase was a mixture of dichloromethane, ethyl acetate, and formic acid (9:1:0.25). The structure was determined using NMR [23]; HPLC analysis of the product was performed as earlier [24].

Screening of genes for EC synthesis
EC is an ester of caffeic acid and ethanol. Plants contain genes that encode an enzyme capable of forming alkyl hydroxycinnamate ester [25,26]. These enzymes have been involved in suberin and cutin biosynthesis, and are known to use long-chain alcohol (i.e. dodecan-1-ol) as an alkyl group donor. We assumed these enzymes to possibly use low-molecular-weight alcohol. Two genes (AtCAT1 and AtCAT2), which encoded fatty alcohol pA-aroG-tyrA pACYCDuet carrying aroG and tyrA from E. coli [21] pA-aroG f -tyrA f pACYCDuet carrying aroG f and tyrA f from E. coli [21] pA-aroG f -ppsA-tktA-tyrA f pACYCDuet carrying aroG f , ppsA, tktA, and tyrA f from E. coli [21] pA-aroL-aroG f -ppsA-tktA-tyrA f pACYCDuet carrying aroL, aroG f , ppsA, tktA, and tyrA f from E. coli [30] pA-aroL-aroE-aroD-aroB-aroG f -ppsA-tktA-tyrA f pACYCDuet carrying aroL, aroE, aroD, aroB, aroG f , ppsA, tktA, and tyrA f from E. coli [28] pA-aroC-aroA-aroL-aroE-aroD-aroB-aroG f -ppsA-  (Fig. 2). We further tested other low-molecular-weight alcohols as substrates. E. coli strain B-EC1 was fed caffeic acid and various low-molecular-weight alcohols, including methanol, ethanol, propanol, butanol, and pentanol. Analysis of the supernatant showed that all the alcohols tested were substrates for the synthesis of corresponding alkyl caffeates (data not shown). Propanol turned out to be the best substrate among all alcohols tested, based on the unreacted caffeic acid remaining. The different permeability of each alcohol into the E. coli, and other factors, such as volatility, were found to have an influence on the titer of each alkyl caffeate. Together, the results indicated that different alkyl caffeates could be synthesized by providing different alcohols. We did not explore the alcohols with more than six carbon atoms, since the solubility of such alcohols would be significantly low (in case of hexanol, the water solubility was 5.9 g/L).
We investigated the maximum amount of EC synthesized when E. coli was fed 1% ethanol. E. coli harboring pC-OS4CL-AtCAT1 (B-EC1) was fed different concentrations of caffeic acid (0.5, 0.7, 1.0, 1.2, 1.5, and 1.8 mM) and 1% ethanol. Caffeic acid was completely converted into EC up to the concentration of 1.2 mM. However, at 1.5 mM caffeic acid, approximately 0.3 mM was not converted into EC, which suggested that AtCAT1 could convert more than 1.2 mM (216.2 mg/L) caffeic acid and 1% ethanol (about 0.36 M) was not the limiting factor.

Synthesis of EC without feeding caffeic acid
We attempted to synthesize EC without feeding caffeic acid to E. coli. Caffeic acid was synthesized from tyrosine using SeTAL and HpaBC [27]. SeTAL converts tyrosine into p-coumaric acid, which is converted to caffeic acid by HpaBC. The two genes were introduced into E. coli harboring pC-OS4CL-AtCAT1. The resulting E. coli strain B-EC3 was fed ethanol thereafter. The culture filtrate from the strain B-EC3 showed a peak that had the same retention time as EC. Furthermore, no detectable caffeic acid was observed, indicating that all the synthesized caffeic acid had been converted to EC, and synthesis of more caffeic acid could increase the final titer of EC in E. coli. Therefore, we engineered E. coli to synthesize more tyrosine, which is a precursor of caffeic acid, by introducing genes, which are known to be involved in the synthesis of tyrosine in E. coli. We tested six constructs (pA-aroG-tyrA, pA-aroG f -tyrA f , pA-aroG f -ppsA-tktA-tyrA f , pA-aroL-aroG f -ppsA-tktA-tyrA f , pA-aroL-aroE-aroD-aroB-aroG f -ppsA-tktA-tyrA f , and pA-aroC-aroA-aroL-aroE-aroD-aroB-aroG f -ppsA-tktA-tyrA f ) with various combination of genes relevant to the shikimate pathway of E. coli, including tyrA, aroG, tyrA f , aroG f , aroC, aroA, aroL, aroE, aroD, aroB, ppsA, and tktA [24,28]. Each construct was transformed into E. coli strain B-EC3 and the resulting strains (B-EC4 -B-EC9) were tested for the synthesis of EC. The E. coli strains overexpressing any of these constructs showed better titer than B-EC3, which did not overexpress the shikimate pathway gene-module construct. The E. coli strain B-EC7, containing five genes, namely aroL, aroG fbr , ppsA, tktA, and tyrA fbr , produced the highest titer of EC (78.8 mg/L), followed by B-EC8 (68.6 mg/L), B-EC6 (68.5 mg/L), B-EC5 (65.2 mg/L), B-EC9 (59.1 mg/L), B-EC4 (44.8 mg/L), and B-EC3 (38.6 mg/L) (Fig. 3). This result agreed with the previous studies in which the overexpression of shikimate gene module increased the final titer of the synthesized compound; however, the best gene-module construct was different depending on the compound synthesized [24,28,29].
Our feeding study had shown that approximately 216.2 mg/L (1.2 mL) caffeic acid was converted to EC. The current EC titer was approximately 78.8 mg/L (0.38 mM), which indicated that the final titer could be increased further. In order to increase the final titer of EC, we used E. coli mutants with deletion of genes from the shikimate pathway, for further production of tyrosine. The BT strain, in which tyrR was deleted, and the BTP strain, in which tyrR and tyrA were deleted, were used to increase the production of tyrosine [21]. We introduced pC-OS4CL-AtCAT1 and pA-aroL-aroG fbr -ppsA-tktA-tyrA fbr into BT and BTP (BT-EC7, BTP-EC7) to examine the titer of EC. Both the mutant strains showed better titer than the wild type. The E. coli strain BT-EC7 synthesized approximately 116.7 mg/L EC, which was 1.48-fold higher than in the wild-type strain (Fig. 4).
EC is a natural compound found in some plants, and its biosynthetic pathway is not yet known. Here, we designed the biosynthetic pathway, assembled genes for the synthesis of EC, and successfully synthesized the compound. Rational design of the biosynthetic pathway and selection of genes from diverse sources can make the synthesis of diverse compounds feasible. Furthermore, this could lead to the synthesis of unnatural compounds, derived from natural compounds. For example, we could synthesize at least four other alkyl caffeates, depending on the alcohol, which have not been found in nature yet, and the compounds could potentially have different biological activities.