Production of Aromatic Compounds by Metabolically Engineered Escherichia coli with an Expanded Shikimate Pathway

Abstract

Escherichia coli was metabolically engineered by expanding the shikimate pathway to generate strains capable of producing six kinds of aromatic compounds, phenyllactic acid, 4-hydroxyphenyllactic acid, phenylacetic acid, 4-hydroxyphenylacetic acid, 2-phenylethanol, and 2-(4-hydroxyphenyl)ethanol, which are used in several fields of industries including pharmaceutical, agrochemical, antibiotic, flavor industries, etc. To generate strains that produce phenyllactic acid and 4-hydroxyphenyllactic acid, the lactate dehydrogenase gene (ldhA) from Cupriavidus necator was introduced into the chromosomes of phenylalanine and tyrosine overproducers, respectively. Both the phenylpyruvate decarboxylase gene (ipdC) from Azospirillum brasilense and the phenylacetaldehyde dehydrogenase gene (feaB) from E. coli were introduced into the chromosomes of phenylalanine and tyrosine overproducers to generate phenylacetic acid and 4-hydroxyphenylacetic acid producers, respectively, whereas ipdC and the alcohol dehydrogenase gene (adhC) from Lactobacillus brevis were introduced to generate 2-phenylethanol and 2-(4-hydroxyphenyl)ethanol producers, respectively. Expression of the respective introduced genes was controlled by the T7 promoter. While generating the 2-phenylethanol and 2-(4-hydroxyphenyl)ethanol producers, we found that produced phenylacetaldehyde and 4-hydroxyphenylacetaldehyde were automatically reduced to 2-phenylethanol and 2-(4-hydroxyphenyl)ethanol by endogenous aldehyde reductases in E. coli encoded by the yqhD, yjgB, and yahK genes. Cointroduction and cooverexpression of each gene with ipdC in the phenylalanine and tyrosine overproducers enhanced the production of 2-phenylethanol and 2-(4-hydroxyphenyl)ethanol from glucose. Introduction of the yahK gene yielded the most efficient production of both aromatic alcohols. During the production of 2-phenylethanol, 2-(4-hydroxyphenyl)ethanol, phenylacetic acid, and 4-hydroxyphenylacetic acid, accumulation of some by-products were observed. Deletion of feaB, pheA, and/or tyrA genes from the chromosomes of the constructed strains resulted in increased desired aromatic compounds with decreased by-products. Finally, each of the six constructed strains was able to successfully produce a different aromatic compound as a major product. We show here that six aromatic compounds are able to be produced from renewable resources without supplementing with expensive precursors.

INTRODUCTION

Aromatic compounds are an important class of chemicals that are used as organic solvents, dyes, and precursors in the processing of foods, pharmaceuticals, polymers, etc. However, these are currently manufactured from petroleum which is recognized as a limited resource and as a cause of global warming. Thus, renewable sources such as biomass feedstock are considered to be an alternative and sustainable source for manufacturing aromatic compounds.

Many bacteria are natural producers of aromatic compounds by virtue of a pathway that synthesizes aromatic amino acids known as the shikimate pathway (7, 13, 36). In this pathway, phosphoenolpyruvate (PEP) and erythrose 4-phosphate (E4P) converted from glucose through the central metabolic pathway are initially combined to form 3-deoxy-d-arabino-heptulosonate-7-phosphate (DAHP), which is then converted to chorismate. From chorismate, the pathway branches to form a variety of aromatic end products, including phenylalanine (Phe), tyrosine (Tyr), tryptophan (Trp), folic acid, etc. To date, the shikimate pathway has been exploited in the manufacture and investigation of several aromatic compounds, some of which are currently commercially available. Three aromatic amino acids, Phe, Tyr, and Trp, are produced by fermentation and used in a range of 150 to 12,000 tons/year (7). These amino acids are mainly used as food and feed additives, pharmaceutical intermediates, sweetener precursors, etc. Shikimate, an intermediate in the shikimate pathway, is a crucial starting material for the synthesis of neuramidase inhibitor GS4104 (Tamiflu), administered as a precaution against influenza infection (18).

To enhance the productivity of aromatic compounds, techniques for genetic modification of strains are frequently applied. For example, in Phe production, the most important steps are the first and last steps of the shikimate pathway. However, enzymes including these steps, DAHP synthase (encoded by aroG) and chorismate mutase/prephenate dehydratase (encoded by pheA), are strongly inhibited by Phe (20). Therefore, feedback-resistant mutants (fbr) have been studied and exploited for Phe production (16, 25). In addition, the levels of expression of both aroG and pheA are controlled by the transcriptional repressor TyrR (29) so that deletion of tyrR is also efficient for Phe production (5). Recent works address modifications of the central metabolic pathway to enhance the availability of the DAHP precursors PEP and E4P (7, 13, 36). Such modifications include overexpression of transketolase (tktA) and PEP synthase (pps) genes (28), deletion of PEP carboxylase gene (ppc) (23), deletion or overexpression of carbon storage regulator genes (csrA or csrB) (38, 44), and glucose transport system exchange from PEP-dependent sugar phosphotransferase system (PTS) to either galactose permease (GalP)-glucokinase (Glk) system (3, 45) or Zymomonas mobilis glucose facilitator (Glf)-Glk system (28). These modifications have been used in suitable combinations to enhance the production of aromatic compounds.

Expansion of the shikimate pathway by introducing homologous/heterologous genes into engineered Escherichia coli strains has been applied to a diversity of aromatic compounds. Production of p-hydroxybenzoic acid has been achieved by homologous overexpression of both ubiC-encoded chorismate lyase and aroF-encoded DAHP synthase in E. coli (4) and Pseudomonas putida (22, 40). In this strategy, PEP and E4P synthesized from glucose enter the shikimate pathway to form chorismic acid, which is then converted to p-hydroxybenzoic acid with the release of pyruvic acid. Phenol production from glucose has been studied using solvent-tolerant Pseudomonas putida S12 (43). In this study, a Tyr-overproducing strain of P. putida S12 was modified by introduction of Tyr phenol lyase gene (encoded by tpl from Pantoea agglomerans) to generate a phenol producer. Cinnamic acid (CA) (26) and coumaric acid (4HCA) (27) producers have also been constructed from aromatic amino acid-overproducing strains of P. putida S12 by introducing the pal gene encoding the bifunctional enzyme, Phe-ammonia lyase/Tyr-ammonia lyase from Rhodosporidium toruloides. The constructed strains were able to deaminate Phe or Tyr converted from glucose to produce copious amounts of CA or 4HCA. Introduction of pdc-encoding 4HCA decarboxylase from Lactobacillus plantarum into the 4HCA overproducer enabled p-hydroxystyrene (4HSTY) production from glucose (41). Production of 4HCA and 4HSTY from glucose also proved successful when a Tyr-overproducing strain of E. coli was used as the host strain (31, 33, 39). Recently, production of bio-based styrene (STY) was achieved by simultaneous overexpression of pal2 from Arabidopsis thaliana and fdc1 from Saccharomyces cerevisiae in a Phe-overproducing strain of E. coli (21).

We previously presented tyramine (TYM) and phenethylamine (PEA) production from glucose by expanding the shikimate pathway in E. coli (17). The TYM producer was constructed by overexpression of the tyramine decarboxylase gene from Lactobacillus brevis in a Tyr overproducer, whereas the PEA producer was constructed by overexpression of the aromatic amino acid decarboxylase gene from Pseudomonas putida in a Phe overproducer. In this report, we investigate the possibility of bioproduction of an additional six aromatic compounds, phenyllactic acid (PLA), 4-hydroxyphenyllactic acid (4HPLA), 2-phenylethanol (PE), 2-(4-hydroxyphenyl)ethanol (4HPE) (also known as tyrosol), phenylacetic acid (PAA), and 4-hydroxyphenylacetic acid (4HPAA) by metabolically engineered E. coli. The shikimate pathway was expanded by introducing T7 promoter-controlled homologous/heterologous genes into the chromosomes of Phe and Tyr overproducers to generate strains capable of producing the respective aromatic compounds from glucose. Target aromatic compounds have been widely used in some industries, and some of these compounds have the potential of being a biopolymer unit. PE is used for flavor and fragrance compound with a rose-like odor. PAA, 4HPAA, and 4HPE are used as building blocks for an antibiotic (penicillin G), pharmaceuticals, agrochemicals, etc. PLA is known as a broad-spectrum antimicrobial agent. In addition, PLA and 4HPLA have the possibility of being building blocks of aromatic polymers, e.g., blending with polylactate. To the best of our knowledge, this is the first assessment of the capability of metabolically modified E. coli to produce PLA, 4HPLA, PAA, and 4HPAA from glucose.

MATERIALS AND METHODS

Cloning of genes.

Phusion Hot Start DNA polymerase (Novagen) was used to amplify the relevant genes. The lactate dehydrogenase gene (ldhA) was amplified from genomic DNA from Cupriavidus necator JCM20644 (synonym Ralstonia eutropha) using the primer pair ReADH2-F-Nde (5′-CCAACCATATGCCTGCTCCCCAGATCCTCC-3′ [the NdeI site is underlined]) and ReADH2-R-Xho (5′-CACTCGAGTTACAGCACTGGCGTCAGCAC-3′ [the XhoI site is underlined]). The NdeI-XhoI-digested amplicon was introduced into the corresponding site of pET21a-FRT (17) to construct pARO133. The alcohol dehydrogenase gene (adhC) was amplified from genomic DNA from Lactobacillus brevis JCM1170 using the primer pair LbADH-F-Nde (5′-CCAACCATATGATGCAAATCAAAACAGCTTTTTC-3′) and LbADH-R-Xho (5′-CACTCGAGTTAGAATGTGATTACGGGC-3′). The NdeI-XhoI-digested amplicon was introduced into the corresponding site of pET21a-FRT to construct pARO131. The phenylacetaldehyde (PAAL) dehydrogenase gene (feaB) was amplified from genomic DNA from Escherichia coli BW25113. Two segments were initially amplified by PCR using primer pair EcPAALDH-Nde (5′-CCAACCATATGACAGAGCCGCATGTAGC-3′) and EcPAALDH-RM1-Nco (5′-CGGAAAGTTCCACGGCACAATTCCCGCC-3′) and primer pair EcPAALDH-FM1-Nco (5′-GGCGGGAATTGTGCCGTGGAACTTTCCG-3′) and EcPAALDH-Xho (5′-CACTCGAGTTAATACCGTACACACACCGAC-3′); the segments were then combined by overlap extension PCR (32) using the primer pair EcPAALDH-Nde and EcPAALDH-Xho. Aldehyde reductase genes (yqhD, yjgB, and yahK) were also amplified from genomic DNA from Escherichia coli BW25113. Primer pair yqhD-Nde (5′-CCAACCATATGAACAACTTTAATCTGCAC-3′) and yqhD-Xho (5′-CACTCGAGTTAGCGGGCGGCTTCG-3′), primer pair yjgB-Nde (5′-CCAACCATATGTCGATGATAAAAAGCTATG-3′) and yjgB-Xho (5′-CACTCGAGTCAAAAATCGGCTTTCAACACC-3′), and primer pair yahK-Nde (5′-CCAACCATATGAAGATCAAAGCTGTTGGTG-3′) and yahK-Xho (5′-CACTCGAGTCAGTCTGTTAGTGTGCG-3′) were used to amplify the respective genes. The NdeI-XhoI-digested amplicons of feaB, yqhD, yjgB, and yahK were introduced into the corresponding site of pET21a-FRT to generate pARO132, pARO167, pARO168, and pARO176, respectively. The phenylpyruvate (PP) decarboxylase gene (ipdC) was synthesized with optimizing codon usage for E. coli using OptimumGene algorithm (GenScript USA Inc.), and inserted into the NdeI-XhoI site of pET21a-FRT to construct pARO136.

Construction of strains.

Escherichia coli BW25113(DE3) was genetically modified by the previously developed method (17) to generate aromatic compound producers. The method consisted of Red-mediated recombination, FLP (flippase)/FRT (FLP recognition target) recombination, and P1 transduction. All of the first strains harboring a single desired gene were constructed by Red-mediated recombination. With adhC for example, the Km cassette flanking two FRT sites and adhC (FRT-Km-FRT-adhC) was amplified by PCR using pARO131 as the template DNA and then introduced into the acs locus on the chromosome to generate strain AR-G37 (acs::FRT-Km-FRT-T7p-adhC [T7p-adhC indicates that the adhC gene is controlled by the T7 promoter]). Similarly, strains AR-G38 (acs::FRT-Km-FRT-T7p-feaB [pARO132 used as the template]), AR-G39 (acs::FRT-Km-FRT-T7p-ldhA [pARO133 used as the template]), AR-G84 (mtlA::FRT-Km-FRT-T7p-ipdC [pARO136 used as the template]), AR-G85 (mtlA::FRT-Km-FRT-T7p-ldhA [pARO133 used as the template]), AR-G92 (acs::FRT-Km-T7p-FRT-yqhD [pARO167 used as the template]), AR-G93 (acs::FRT-Km-FRT-T7p-yjgB [pARO168 used as the template]), and AR-G94 (acs::FRT-Km-FRT-T7p-yahK [pARO176 used as the template]) were constructed. Next, genetic traits were assembled by P1 transduction, and a kanamycin resistance marker was excised by FLP/FRT recombination to generate aromatic compound producers. The genetic trait of strain AR-G39 (acs::FRT-Km-FRT-T7p-ldhA) was incorporated into strains AR-G91 (tyrR::T7p-aroF^fbr-pheA^fbr; Phe overproducer) and AR-G2 (tyrR::T7p-aroF^fbr-tyrA^fbr; Tyr overproducer) (17), which were derived from E. coli strain BW25113(DE3), to generate PLA producer PAR-57 and 4HPLA producer PAR-3, respectively. Strain PAR-57 was further modified by assembling the genetic trait of strain AR-G85 (mtlA::FRT-Km-FRT-T7p-ldhA) to generate PLA producer PAR-58 harboring two copies of the chromosomal T7p-ldhA gene. The genetic trait of strain AR-G84 (mtlA::FRT-Km-FRT-T7p-ipdC) was incorporated into strains AR-G91 and AR-G2 to generate strains PAR-60 and PAR-47, respectively. Both strains were further modified by assembling the genetic trait of AR-G38 (acs::FRT-Km-FRT-T7p-feaB) to generate PAA producer PAR-61 and 4HPAA producer PAR-51. Similarly, PE producer PAR-62 and 4HPE producer PAR-53 were constructed from strains PAR-60 and PAR-47, respectively, by assembling the genetic trait of AR-G37 (acs::FRT-Km-FRT-T7p-adhC). Instead of the genetic trait of strain AR-G37, the genetic traits of strains AR-G92 (acs::FRT-Km-T7p-yqhD), AR-G93 (acs::FRT-Km-FRT-T7p-yjgB), and AR-G94 (acs::FRT-Km-FRT-T7p-yahK) were incorporated into strains PAR-60 and PAR-47 to generate PE producer PAR-66, PAR-67, and PAR-68 and 4HPE producers PAR-63, PAR-64, and PAR-65. The strains constructed to produce aromatic compounds are summarized in Table 1.

Table 1.

E. coli strains used in this study

Straina	Relevant genotype	Relevant characteristic or phenotype
BW25113(DE3)	lacI rrnBT14 ΔlacZWJ16 hsdR514 ΔaraBADAH33 ΔrhaBADLD78 dcm(DE3)	Host strain in this study (9, 17)
AR-G2	tyrR::T7p-aroFfbr-tyrAfbr	Tyr overproducer (17)
AR-G37	acs::FRT-Km-FRT-T7p-adhC	Donor of T7p-adhC
AR-G38	acs::FRT-Km-FRT-T7p-feaB	Donor of T7p-feaB
AR-G39	acs::FRT-Km-FRT-T7p-ldhA	Donor of T7p-ldhA
AR-G49	feaB::FRT-Km-FRT	feaB deficient
AR-G51	paaK::FRT-Km-FRT	paaK deficient
AR-G78	tyrA::FRT-Km-FRT	tyrA deficient
AR-G79	pheA::FRT-Km-FRT	pheA deficient
AR-G84	mtlA::FRT-Km-FRT-T7p-ipdC	Donor of T7p-ipdC
AR-G85	mtlA::FRT-Km-FRT-T7p-ldhA	Donor of T7p-ldhA
AR-G91	tyrR::T7p-aroFfbr-pheAfbr	Phe overproducer (17)
AR-G92	acs::FRT-Km-FRT-T7p-yqhD	Donor of T7p-yqhD
AR-G93	acs::FRT-Km-FRT-T7p-yjgB	Donor of T7p-yjgB
AR-G94	acs::FRT-Km-FRT-T7p-yahK	Donor of T7p-yahK
AR-G98	yahK::FRT-Km-FRT	yahK deficient
JW2978	yqhD::FRT-Km-FRT	yqhD deficient (2)
JW5761	yjgB::FRT-Km-FRT	yjgB deficient (2)
PAR-3	tyrR::T7p-aroFfbr-tyrAfbr acs::T7p-ldhA	4HPLA producer
PAR-47	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC	4HPAAL synthesis
PAR-51	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC acs::T7p-feaB	4HPAA producer
PAR-53	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC acs::T7p-adhC	4HPE producer
PAR-57	tyrR::T7p-aroFfbr-pheAfbr acs::T7p-ldhA	PLA producer
PAR-58	tyrR::T7p-aroFfbr-pheAfbr acs::T7p-ldhA mtlA::T7p-ldhA	PLA producer
PAR-60	tyrR::T7p-aroFfbr-pheAfbr mtlA::T7p-ipdC	PAAL synthesis
PAR-61	tyrR::T7p-aroFfbr-pheAfbr mtlA::T7p-ipdC acs::T7p-feaB	PAA producer
PAR-62	tyrR::T7p-aroFfbr-pheAfbr mtlA::T7p-ipdC acs::T7p-adhC	PE producer
PAR-63	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC acs::T7p-yqhD	4HPE producer
PAR-64	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC acs::T7p-yjgB	4HPE producer
PAR-65	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC acs::T7p-yahK	4HPE producer
PAR-66	tyrR::T7p-aroFfbr-pheAfbr mtlA::T7p-ipdC acs::T7p-yqhD	PE producer
PAR-67	tyrR::T7p-aroFfbr-pheAfbr mtlA::T7p-ipdC acs::T7p-yjgB	PE producer
PAR-68	tyrR::T7p-aroFfbr-pheAfbr mtlA::T7p-ipdC acs::T7p-yahK	PE producer
PAR-75	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC yqhD	4HPAAL synthesis
PAR-77	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC yjgB	4HPAAL synthesis
PAR-79	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC yahK	4HPAAL synthesis
PAR-82	tyrR::T7p-aroFfbr-pheAfbr mtlA::T7p-ipdC acs::T7p-feaB paaK	PAA producer
PAR-83	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC acs::T7p-yahK feaB	4HPE producer
PAR-84	tyrR::T7p-aroFfbr-pheAfbr mtlA::T7p-ipdC acs::T7p-yahK feaB	PE producer
PAR-89	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC yqhD yjgB	4HPAAL synthesis
PAR-90	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC yqhD yahK	4HPAAL synthesis
PAR-91	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC yjgB yahK	4HPAAL synthesis
PAR-92	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC yqhD yjgB yahK	4HPAAL synthesis
PAR-97	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC acs::T7p-feaB yahK	4HPAA producer
PAR-100	tyrR::T7p-aroFfbr-pheAfbr mtlA::T7p-ipdC acs::T7p-feaB tyrA	PAA producer
PAR-102	tyrR::T7p-aroFfbr-tyrAfb mtlA::T7p-ipdC acs::T7p-feaB yahK pheA	4HPAA producer
PAR-104	tyrR::T7p-aroFfbr-tyrAfbr mtlA::T7p-ipdC acs::T7p-yahK feaB pheA	4HPE producer
PAR-105	tyrR::T7p-aroFfbr-pheAfbr mtlA::T7p-ipdC acs::T7p-yahK feaB tyrA	PE producer

^aStrains BW25113(DE3), JW2978, and JW5761 were derived from E. coli BW25113, whereas the other strains were derived from E. coli BW25113(DE3).

Gene deletion.

The feaB, paaK, yahK, tyrA, and pheA genes were individually deleted from the chromosome of E. coli BW25113(DE3) as previously reported (2, 9), yielding strains AR-G49 (feaB::FRT-Km-FRT), AR-G51 (paaK::FRT-Km-FRT), AR-G98 (yahK::FRT-Km-FRT), AR-G78 (tyrA::FRT-Km-FRT), and AR-G79 (pheA::FRT-Km-FRT). Two derivatives of strain BW25113, strains JW2978 (yqhD::FRT-Km-FRT) and JW5761 (ΔyjgB::FRT-Km-FRT), which are members of a single-gene knockout mutant library known as the Keio collection (2), were obtained from the National BioResource Project (NIG, Japan) on E. coli strains. The traits of gene-deficient strains were transferred to aromatic compound producers by P1 transduction, and then the kanamycin resistance marker was excised by FLP/FRT recombination (17) to generate strains PAR-75, PAR-77, PAR-79, PAR-82, PAR-83, PAR-84, PAR-89, PAR-90, PAR-91, PAR-92, PAR-97, PAR-100, PAR-102, PAR-104, and PAR-105, which are listed in Table 1.

Production of aromatic compounds.

Each strain was precultivated overnight in 5 ml LB-G medium (10 g polypeptone, 5 g dried yeast extract D-3, 10 g NaCl per liter [pH 7.0]) at 37°C. Fifty microliters of the preculture was inoculated into 5 ml M9M medium (10 g glucose, 6 g Na₂HPO₄, 3 g KH₂PO₄, 0.5 g NaCl, 2 g NH₄Cl, 1 mM MgSO₄, 0.1 mM CaCl₂, 50 μM FeSO₄, 1 μM ZnSO₄, 1 μM CoCl₂, and 0.001% thiamine per liter), M9M medium plus 1 mM Phe, or M9M medium plus 1.5 mM Tyr. The timing to induce gene expression by the addition of isopropyl-β-d-thiogalactopyranoside (IPTG) was varied depending on the strains because the growth of some constructed strains was depressed by IPTG addition and glucose consumption became poor. As a standard, the strains were cultivated at 37°C with shaking at 250 rpm/min (BR-23FH·MR; Taitec Co. Ltd.) in M9M medium until an optical density at 660 nm (OD₆₆₀) of 0.3 was attained (17). Strains PAR-47, PAR-51, PAR-53, PAR-63, PAR-64, PAR-65, PAR-75, PAR-77, PAR-79, PAR-83, PAR-89, PAR-90, PAR-91, PAR-92, and PAR-97, which were AR-G2 derivatives harboring T7p-ipdC at the mtlA locus of the chromosome, were cultivated in M9M medium at 37°C (250 rpm/min) until an OD₆₆₀ of 4.0 was attained. Phe auxotroph strains PAR-102 and PAR-104 and Tyr auxotroph strains PAR-100 and PAR-105 were cultivated at 37°C (250 rpm/min) in M9M medium plus 1 mM Phe and M9M medium plus 1.5 mM Tyr, respectively, until an OD₆₆₀ of 2.0 was attained. IPTG was added to the culture at a final concentration of 1 mM to induce gene expression, and the strains were continuously cultivated at 27°C with shaking at 250 rpm/min (TC-300; Takasaki Scientific Instruments Corp.). The total cultivation time at 37°C and 27°C was 48 h. If necessary, cultivation time was prolonged for up to 72 h. In order to determine volumetric and specific productivity parameters, an aliquot of the culture (0.2 ml) was periodically sampled.

Following cultivation, 1/20 volume of 2 M HCl and 1/5 volume of methanol were mixed with 1/10 diluted culture. The mixture was centrifuged at 10,000 × g for 5 min, and the obtained supernatant was filtered by Cosmonice filter W (0.45 μl) (Millipore). The filtrate was preserved for a high-performance liquid chromatography (HPLC) assay.

Enzyme assay.

E. coli BW25113(DE3) harboring pARO131 (adhC), pARO133 (ldhA), pARO136 (ipdC), pARO167 (yqhD), pARO168 (yjgB), or pARO176 (yahK) was cultivated in LB-G medium containing ampicillin until an OD₆₆₀ of 0.5 was attained. IPTG was added to the culture at a final concentration of 1 mM to induce gene expression, and the strain was continuously cultivated at 27°C for 16 h. After the strain was cultivated at 27°C for 16 h, cells were harvested from 1 ml of culture by centrifugation at 10,000 × g for 5 min, washed once with 1 ml of 100 mM phosphate buffer (PB) (pH 7.2), and then dissolved with 100 μl PB. The cells were disrupted by sonication in ice-chilled water and centrifuged at 10,000 × g for 15 min. The supernatant was used as cell extract (CFE [named CFE for cell-free extract]) in the enzyme assay. The protein concentration of the CFE was measured using Coomassie protein assay reagent (Pierce) with bovine serum albumin as a standard. For assay of PLA or 4HPLA dehydrogenase activity, the reaction mixture contained 2 mM PP or 4-hydroxyphenylpyruvate (4HPP), 2 mM NADH, and each CFE in PB. For assay of PAAL reductase activity, the reaction mixture contained 2 mM PAAL, 2 mM NADH, 0.2 mM ZnSO₄, and each CFE in PB. The reactions were started by the addition of 10 μl of each CFE. For assay of 4-hydroxyphenylacetaldehyde (4HPAAL) reductase activity, the reaction mixture contained 2 mM 4HPP, 2 mM NADH, 2 mM thiamine pyrophosphate (TPP), 0.2 mM ZnSO₄, 0.2 mM MgSO₄, 100 μl of CFE (4 mg/ml) from E. coli BW25113(DE3)/pARO136 (IpdC solution), and each CFE. The reaction was started by the addition of 100 μl of IpdC solution and 10 μl of each CFE. The reactions were performed at 30°C for 10 min and then stopped by adding 50 μl of 2 M HCl. After 0.2 ml acetonitrile was mixed into the reaction mixture, the amount of each product was determined by HPLC.

Analytical method.

An HPLC system fitted with a photodiode array (SPD-M10AVP; Shimadzu Corp.) and an octadecyl silica column (Cosmosil 5C₁₈-MS-II column [3.0 by 150 mm] from Nacalai Tesque Inc.) was used to measure the concentrations of aromatic compounds. The compounds were eluted at a flow rate of 0.4 ml/min of mobile phase containing methanol and 0.1% trifluoroacetic acid. The methanol concentration was increased from 20% to 80% for 5 min, and this concentration was then maintained for 10 min. Under such conditions, Tyr (3.8 min), 4HPLA (5.1 min), Phe (5.2 min), 4HPE (5.4 min), 4HPAA (5.7 min), PLA (6.8 min), PAA (7.2 min), and PE (7.3 min) were eluted with acceptable separation. The wavelength used to detect Phe, PLA, PAA, and PE was 206 nm, whereas the wavelength used to detect Tyr, 4HPLA, 4HPE, and 4HPAA was 222 nm. The compounds were quantified using standard curves of the respective commercial chemicals (except for PAA, which is difficult to purchase). For PAA quantification, PE was used as the standard. To determine enzyme activity (see “Enzyme assay” above), acetonitrile was used for the mobile phase instead of methanol. Under such conditions, PLA, 4HPLA, PE, and 4HPE were eluted at 5.2, 3.2, 5.9, and 3.9 min, respectively.

To analyze the chirality of produced PLA and 4HPLA, an HPLC system fitted with a Sumichiral OA-5000 column (4.6 by 150 mm; Sumika Chemical Analysis Service Ltd.) was used. The compounds were eluted at a flow rate of 1.0 ml/min from the mobile phase containing 15% isopropanol and 2 mM CuSO₄. Under such conditions, the standard chemical compounds l-4HPLA, d-4HPLA, l-PLA, and d-PLA were eluted at 19, 25, 72, and 94 min, respectively, and detected at 254 nm.

Optical density for cell growth was measured at 660 nm (UV-160A; Shimadzu Corp). Cell weight (dry weight) (CDW) values were calculated from OD₆₆₀ values using the formula CDW (g liter⁻¹) = OD₆₆₀ × 0.5. The glucose concentration was colorimetrically determined using Glucose CII Test Wako (Wako Pure Chemical Industries, Ltd.).

RESULTS

Overview of the production of aromatic compounds by expanding the shikimate pathway.

E. coli synthesizes Tyr and Phe from glucose through the central metabolic pathway and the shikimate pathway (Fig. 1). Glucose is converted to PEP and E4P in the central metabolic pathway, and then DAHP was synthesized by DAHP synthases encoded by aroF, aroG, and/or aroH. Seven steps into the shikimate pathway, prephenic acid, precursor to both PP and 4HPP, is synthesized. Here, pheA- or tyrA-encoded bifunctional enzyme acts on prephenic acid to form PP or 4HPP, respectively, which is converted to Phe or Tyr, respectively, by tyrB-encoded aminotransferase.

Fig 1.

Overview of production of aromatic compounds by expanding the shikimate pathway. Solid lines indicate inherent pathways in E. coli, whereas broken lines indicate pathways that may be constructed by introducing heterologous genes. The genes in brackets are the introduced and overexpressed genes to produce the respective aromatic compounds. The shaded (gray shading) compounds are the target aromatic compounds in the present study.

Expansion of the shikimate pathway by introduction and overexpression of homologous/heterologous genes may allow tailoring the production of aromatic compounds from glucose (Fig. 1). For instance, PLA can be produced from PP by introducing a dehydrogenase gene that facilitates reduction of the PP carbonyl group. PP can also be converted to PAAL by introducing a decarboxylase gene. PAAL is further converted to PE or PAA by introduction and overexpression of an aldehyde reductase gene or an aldehyde dehydrogenase gene, respectively. Phe synthesized from PP by transamination can be converted to PEA or CA by introduction of a decarboxylase gene or an ammonia lyase gene. CA can be further converted to STY by introducing a decarboxylase gene. These compounds are derived by expansion of the Phe synthetic route; a similar pathway may be constructed by expanding the Tyr synthetic route. In such cases, compounds possessing hydroxyl groups on their aromatic ring, namely, 4HPLA, 4HPE, 4HPAA, TYM, 4HCA, and 4HSTY, are obtainable from glucose. A PAA dehydrogenase gene (feaB) is known to reside in the genome of E. coli, enabling the potential conversion of PAAL and 4HPAAL to PAA and 4HPAA in this organism. However, E. coli lacks a decarboxylase gene for conversion of PP or 4HPP to PAAL or 4HPAAL. So far, production of CA, 4HCA, STY, and 4HSTY from glucose has been well studied by using metabolically engineered E. coli and/or Pseudomonas strains (21, 26, 27, 31, 33, 41). Production of PEA and TYM from glucose by metabolically engineered E. coli has been demonstrated in our previous study (17). In the present study, we address the issue of whether PLA, 4HPLA, PE, 4HPE, PAA, and 4HPAA can be produced via a similar mechanism (shaded products in Fig. 1).

Genes for expansion of the shikimate pathway.

To achieve the production of six aromatic compounds, PLA, 4HPLA, PE, 4HPE, PAA, and 4HPAA, the desired genes were surveyed using the BRENDA database (http://www.brenda-enzymes.org/) and some biological tools (e.g., BLAST search) and information from the National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/). Because E. coli is used as a host strain in the present study, bacteria (rather than fungi, yeasts, or archaea) were targeted as natural sources of the genes. Finally, selected genes were cloned from three members of the phylum Proteobacteria (including E. coli) and one member of the phylum Firmicutes (Lactobacillus brevis). The cloned genes and the organisms from which they were derived are listed in Table 2.

Table 2.

Genes for production of aromatic compounds

Gene	GenBank protein ID	Enzyme	Species of origin
ldhA	CAJ91827	Lactate dehydrogenase or related dehydrogenase	Cupriavidus necator JCM20644
ipdC	CAA67899	Indol-3-pyruvate/phenylpyruvate decarboxylase	Azospirillum brasilense NBRC102289
feaB	AAC74467	Phenylacetaldehyde dehydrogenase	Escherichia coli BW25113
adhC	ABJ64046	Zn-dependent alcohol dehydrogenase	Lactobacillus brevis JCM1170
yqhD	AAC76047.1	Aldehyde reductase, NADPH dependent	Escherichia coli BW25113
yjgB	AAC77226.2	Predicted alcohol dehydrogenase, Zn dependent and NAD(P) binding	Escherichia coli BW25113
yahK	AAC73428.1	Predicted oxidoreductase, Zn dependent and NAD(P) binding	Escherichia coli BW25113

A reductase capable of acting on PP and 4HPP is required for PLA and 4HPLA production. Suitable known reductases are (R)-4HPLA dehydrogenase (EC 1.1.1.222) and 4HPP reductase (EC 1.1.1.237). There are some eukaryotic enzymes, such as 4HPP reductase from Solenostemon scutellarioides (GenBank protein identification [ID] CAD47810.2), known to belong to these categories, so the corresponding genes were surveyed in bacteria by BLAST search using the amino acid sequence of S. scutellarioides 4HPP reductase as a query sequence. The amino acid sequence of ldhA-encoded dehydrogenase from C. necator showed 45% identity, and its gene was cloned for our purpose. Although E. coli harbors ldhA gene encoding lactate dehydrogenase in the genome, the similarity of the amino acid sequence against 4HPP reductase from S. scutellarioides and LdhA from C. necator was low (<24%).

PP decarboxylase (EC 4.1.1.43) is widely distributed in bacteria. The enzyme encoded by the ipdC gene derived from Azospirillum brasilense has been extensively investigated, and its three-dimensional structure has been published (35, 42). The enzyme acts not only on PP but also on 4HPP to produce the corresponding aldehydes. Although the native ipdC gene is expressed in active form in E. coli, codon modification (without changing the amino acid) was undertaken to reduce the GC content, rendering it suitable for amplification by PCR.

PAAL dehydrogenase, which is encoded by feaB and involved in PEA and TYM degradation pathways in E. coli, has been well-studied. The feaB gene was cloned for conversion of PAAL and 4HPAAL to PAA and 4HPAA, respectively.

Reduction of PAAL and 4HPAAL yields PE and 4HPE, but no PAAL or 4HPAAL reductase gene has yet been found in E. coli. Instead, the heterologous adhC gene, which encodes Zn-dependent alcohol dehydrogenase, was cloned from L. brevis. The adhC gene exhibited similarity (>50%) with aryl-alcohol dehydrogenase (EC 1.1.1.90) derived from a variety of bacteria. However, our data (described below) suggest that reductase genes capable of acting on PAAL and 4HPAAL do exist in E. coli. Therefore, we surveyed such genes in E. coli by BLAST search using query amino acid sequences of some heterologous aromatic reductases, phenylacetaldehyde reductase of Rhodococcus sp. strain ST-10 (GenBank protein ID BAD51480), cinnamyl alcohol dehydrogenase of Helicobacter pylori (AAD08150.1), and benzyl alcohol dehydrogenases of Acinetobacter calcoaceticus (AAC32671.1) and of Pseudomonas putida (BAJ06499 and BAJ06503). Twelve genes, adhE, adhP, eutG, ydjJ, ydjL, frmA, gatD, yiaY, yjgB, yjiN, yphC, and yqhD, exhibited similarities to the query sequences sufficiently and were introduced into the multicloning site (MCS) of pET21a-FRT or pET21d-FRT as 4HPAAL reductase candidates. 4HPAAL-producing strain PAR-47 (Table 1) was transformed with the respective recombinant plasmids, and 4HPE and 4HPAA production in M9M medium by the transformants was preliminarily examined (data not shown). In conclusion, three aldehyde reductase gene candidates seemed to increase 4HPE production, so these genes were used to generate 4HPE-producing strains.

Enzyme assays.

Since two exogenous genes, ldhA and adhC, and two endogenous genes, yjgB and yahK, encode the enzymes without experimental data on the function, the genes were expressed in E. coli, and their enzymatic activities were evaluated (Table 3). Although two kinds of enzymatic reactions, PP to PLA and 4HPP to 4HPLA, were carried out by using CFE prepared from the E. coli BW25113(DE3) host strain as a control, such reductase activities were not detected. In contrast, the reductase activities of PP to 4HPLA and 4HPP to 4HPLA were distinctly detected in CFE prepared from an ldhA-expressing strain. On the other hand, the reductase activity of PAAL to PE was detected in strains overexpressing adhC, yjgB, or yahK as well as in a strain with the yqhD gene, which is known to encode a broad-substrate alcohol dehydrogenase, and their reductase activities were considerably higher than those of host strain BW25113(DE3).

Table 3.

Evaluation of enzyme activities for putative lactate dehydrogenase and aldehyde reductases

Reactiona	Supplemented cofactor(s) and enzymeb	Strain	Overexpressed gene	Sp act (U/mg)c
PP → PLA	NADH	BW25113(DE3) (control)		ND
		BW25113(DE3)/pARO133	ldhA	1.60 ± 0.01
4HPP → 4HPLA	NADH	BW25113(DE3) (control)		ND
		BW25113(DE3)/pARO133	ldhA	0.58 ± 0.02
PAAL → PE	NADH + ZnSO4	BW25113(DE3) (control)		0.02 ± 0.00
		BW25113(DE3)/pARO131	adhC	0.06 ± 0.01
		BW25113(DE3)/pARO167	yqhD	0.40 ± 0.00
		BW25113(DE3)/pARO168	yjgB	2.80 ± 0.04
		BW25113(DE3)/pARO176	yahK	2.24 ± 0.09
4HPP → 4HPAAL → 4HPE	NADH + ZnSO4 + TPP + MgCl2 + IpdC	BW25113(DE3) (control)		ND
		BW25113(DE3)/pARO131	adhC	ND
		BW25113(DE3)/pARO167	yqhD	0.08 ± 0.00
		BW25113(DE3)/pARO168	yjgB	0.44 ± 0.01
		BW25113(DE3)/pARO176	yahK	0.84 ± 0.02

^aThe assay for conversion of 4HPP to 4HPE was performed by a coupling reaction of IpdC [CFE from BW25113(DE3)/pARO136] and each overexpressed reductase.

^bThiamine pyrophosphate (TPP) and MgCl₂ were cofactors for IpdC.

^cThe values are means ± standard deviations of 3 independent enzyme assays. One unit of activity was defined as the amount of enzyme that produced 1 μmol of product per minute at 30°C. ND, not detected.

An enzymatic assay for reduction of 4HPAAL to 4HPE was not performed because sufficient amounts of 4HPAAL could not be obtained. Instead, an assay for conversion of 4HPP to 4HPE was performed by a coupling reaction of phenylpyruvate decarboxylase (IpdC) and each putative reductase. Although 4HPP (2 mM) was almost completely converted to 4HPAAL (1.9 mM) in the control reaction [using CFE from E. coli BW25113(DE3)], 4HPE was not detected. In contrast, the conversion of 4HPP to 4HPE was advanced by using CFEs prepared from yqhD-, yjgB-, and yahK-overexpressing strains, indicating that yqhD-, yjgB-, and yahK-encoded reductases have 4HPAAL reductase activity. These results suggest that the cloned putative genes are candidates for construction of PLA, 4HPLA, PE, and 4HPE producers.

Construction and evaluation of PLA and 4HPLA producers.

To generate PLA- and 4HPLA-producing strains, T7 promoter-controlled ldhA (T7p-ldhA) was introduced into the chromosomes of E. coli strains AR-G91 (Phe overproducer) and AR-G2 (Tyr overproducer). The resulting strains, PAR-57 and PAR-3, produced PLA and 4HPLA, respectively (Table 4). During PLA production, accumulation of a small amount of Phe was observed. The accumulated Phe was reduced by introducing additional T7p-ldhA into the PAR-57 chromosome, yielding strain PAR-58. Under the same cultivation conditions, the level of accumulated PLA in strain PAR-58 (6.0 ± 0.3 mM) was identical to the level of accumulated Phe in strain AR-G91 (7.6 ± 0.3 mM). Surprisingly, although strain AR-G2 produced only 4.7 ± 0.5 mM Tyr, strain PAR-3 produced a considerably higher concentration of 4HPLA (8.1 ± 0.8 mM). By HPLC chirality analysis, the PLA and 4HPLA produced by strains PAR-58 and PAR-3, respectively, were identified with standard chemicals of d-forms, and the optical purities were above 99%ee (enantiomeric excess). These results indicate that we had successfully constructed strains that produced d-PLA and d-4HPLA from glucose.

Table 4.

Production of PLA and 4HPLA by the genetically modified strains^a

Strainb	Strain description and relevant inserted genec	Amt (mM) of glucose consumedd,e	OD660d,f		Amt (mM) of major product producedg				Yieldh (%, mol/mol)
			Ind.	Fin.	PLA	4HPLA	Phe	Tyr	Yp/s	Ybp/s
Strains constructed to produce PLA
AR-G91	Phe overproducer	54.1	0.3	7.2	ND	ND	7.6 ± 0.5	ND	NC	13.7
PAR-57	Phe overproducer + ldhA	53.3	0.3	6.4	5.5 ± 0.2	ND	2.4 ± 0.2	ND	9.9	4.3
PAR-58	Phe overproducer + ldhA + ldhA	54.0	0.3	6.4	6.0 ± 0.3	ND	1.3 ± 0.1	ND	10.8	2.3
Strains constructed to produce 4HPLA
AR-G2	Tyr overproducer	55.5	0.3	7.9	ND	ND	ND	4.7 ± 0.2	NC	8.5
PAR-3	Tyr overproducer + ldhA	54.0	0.3	8.6	ND	8.1 ± 0.8	ND	ND	14.6	NC

^aThe strains were cultivated in M9M medium at 37°C (before induction of gene expression) and 27°C (after induction of gene expression) for 48 h.

^bStrains AR-G91 and AR-G2 were phenylalanine and tyrosine overproducers, respectively, and used as control strains.

^c+ before a gene indicates that the gene was inserted. Expression of the inserted genes was controlled by the T7 promoter.

^dValues were obtained from three independent cultures, and the maximum coefficient of variation was below 10%.

^eThe initial glucose concentration was 55.6 mM (10 g/liter).

^fInd., OD₆₆₀ when the induction of gene overexpression was started; Fin., final OD₆₆₀.

^gThe values are means ± standard deviations obtained from 3 independent cultures. ND, not detected.

^hY_p/s, yield in accumulated target product (mM) per initial glucose (55.6 mM) × 100%; Y_bp/s, yield in accumulated by-product (mM) per initial glucose (55.6 mM) × 100%. NC, not calculable.

Construction and evaluation of PE and 4HPE producers.

To generate strains PAR-60 and PAR-47, strains capable of synthesizing PAAL and 4HPAAL, respectively, T7 promoter-controlled ipdC (T7p-ipdC) was incorporated into the chromosomes of the respective strains, strains AR-G91 (Phe overproducer) and AR-G2 (Tyr overproducer). T7 promoter-controlled adhC (T7p-adhC) was then incorporated into the chromosomes of strains PAR-60 and PAR-47 to generate PAR-62 (PE producer) and PAR-53 (4HPE producer), respectively (Table 1).

The PAR-60 and PAR-47 strains and their derivatives, which harbored T7p-ipdC, did not accumulate Phe and Tyr when cultured. In addition, accumulation of aldehydes (PAAL and 4HPAAL) was not observed in cultures of strains PAR-60 and PAR-47. Instead, PE, 4HPE, PAA, and 4HPAA accumulated, suggesting that PAAL and 4HPAAL produced from glucose were readily reduced and oxidized to their corresponding alcohols and acids. The titers and yields of PE and 4HPE by the constructed strains are summarized in Table 5. Overexpression of both T7p-ipdC and T7p-adhC genes in strain PAR-62 increased the PE production of that strain relative to that of PAR-60. However, a similar effect was lacking in strain PAR-53, which was constructed as a 4HPE producer. The reductase encoded by adhC appeared to act on PAAL in vivo but not on 4HPAAL. During the production of PE by strain PAR-62, a significant amount of 4HPE by-product, which was synthesized from the Tyr synthetic route (described below), accumulated in the culture.

Table 5.

Production of PE and 4HPE by the genetically modified strains^a

Strainb	Strain description and relevant inserted and deleted gene(s)c	Mediumd	Amt (mM) of glucose consumede,f	OD660e,g		Amt (mM) of major product producedh (mM)				Yieldi (%, mol/mol)
				Ind.	Fin.	PE	4HPE	PAA	4HPAA	Yp/s	Ybp/s
Strains constructed to produce PE
AR-G91	Phe overproducer	M9M	54.1	0.3	6.5	ND	ND	ND	ND	NC	NC
PAR-60	Phe overproducer + ipdC	M9M	48.4	0.3	4.5	4.6 ± 0.5	2.4 ± 0.1	1.7 ± 0.2	0.9 ± 0.1	8.3	9.0
PAR-62	Phe overproducer + ipdC + adhC	M9M	53.4	0.3	4.2	5.7 ± 0.4	2.0 ± 0.3	0.2 ± 0.0	0.2 ± 0.0	10.3	4.3
PAR-66	Phe overproducer + ipdC + yqhD	M9M	54.0	0.3	5.5	5.1 ± 0.4	2.0 ± 0.2	<0.2	<0.2	9.2	<4.3
PAR-67	Phe overproducer + ipdC + yjgB	M9M	54.1	0.3	5.4	5.8 ± 0.4	2.3 ± 0.2	<0.2	<0.2	10.4	<4.9
PAR-68	Phe overproducer + ipdC + yahK	M9M	53.9	0.3	5.3	6.5 ± 0.1	2.7 ± 0.2	<0.2	<0.2	11.7	<5.6
PAR-84	Phe overproducer + ipdC + yahK − feaB	M9M	51.7	0.3	4.9	7.7 ± 0.2	3.4 ± 0.1	ND	ND	13.8	6.1
PAR-105	Phe overproducer + ipdC + yahK − feaB − tyrA	M9M+ Tyr	54.1	2.0	3.2	6.9 ± 0.2	1.3 ± 0.1	ND	ND	12.4	2.3
Strains constructed to produce 4HPE
AR-G2	Tyr overproducer	M9M	55.5	0.3	7.9	ND	ND	ND	ND	NC	NC
PAR-47	Tyr overproducer + ipdC	M9M	50.2	4.0	5.1	0.3 ± 0.0	2.6 ± 0.1	<0.2	1.7 ± 0.0	4.7	<4.0
PAR-53	Tyr overproducer + ipdC + adhC	M9M	52.4	4.0	5.2	0.3 ± 0.0	2.4 ± 0.6	<0.2	1.6 ± 0.1	4.3	<3.8
PAR-63	Tyr overproducer + ipdC + yqhD	M9M	55.4	4.0	5.0	0.3 ± 0.0	3.0 ± 0.1	<0.2	1.6 ± 0.0	5.4	<3.8
PAR-64	Tyr overproducer + ipdC + yjgB	M9M	54.3	4.0	5.1	0.4 ± 0.0	3.9 ± 0.7	<0.2	1.5 ± 0.5	7.0	<3.8
PAR-65	Tyr overproducer + ipdC + yahK	M9M	54.1	4.0	4.8	0.5 ± 0.0	4.1 ± 0.2	<0.2	0.5 ± 0.1	7.4	<2.2
PAR-83	Tyr overproducer + ipdC + yahK − feaB	M9M	53.5	4.0	5.4	2.4 ± 0.7	3.8 ± 0.3	ND	ND	6.8	4.3
PAR-104	Tyr overproducer + ipdC + yahK − feaB − pheA	M9M+ Phe	51.7	2.0	3.2	1.0 ± 0.1	8.3 ± 0.2	ND	ND	14.9	1.8

^aThe strains were cultivated at 37°C (before induction of gene expression) and 27°C (after induction of gene expression) for 48 h.

^bStrains AR-G91 and AR-G2 were phenylalanine and tyrosine overproducers, respectively, and used as control strains.

^c+ before a gene indicates that the gene was inserted, whereas − before a gene indicates that the gene was deleted. Expression of the inserted genes was controlled by the T7 promoter.

^dThe concentrations of supplemented Phe and Tyr were 1.0 and 1.5 mM, respectively.

^eValues were obtained from 3 independent cultures, and the maximum coefficient of variation was below 10%.

^fThe initial glucose concentration was 55.6 mM (10 g/liter).

^gInd., OD₆₆₀ when the induction of gene overexpression was started; Fin., final OD₆₆₀.

^hThe values are means ± standard deviations obtained from 3 independent cultures. ND, not detected.

ⁱY_p/s, yield in accumulated target product (mM) per initial glucose (55.6 mM) × 100%; Y_bp/s, yield in accumulated by-product (mM) per initial glucose (55.6 mM) × 100%. NC, not calculable.

As strains PAR-60 and PAR-47 accumulated PE and 4HPE, it is thought that endogenous aldehyde reductases are able to target PAAL and 4HPAAL in E. coli. Simultaneous overexpression of the T7p-ipdC gene and T7 promoter-controlled candidate genes (T7p-yqhD, T7p-yjgB, or T7p-yahK) increased PE production by strains PAR-66, PAR-67, and PAR-68 and 4HPE production by strains PAR-63, PAR-64, and PAR-65 (Table 5). This was especially true in strain PAR-65, in which the yahK gene was overexpressed and the accumulation of 4HPAA by-product was considerably decreased.

To identify which reductases contribute to 4HPE production, three reductases encoded by yqhD, yjgB, and yahK were individually or simultaneously deleted from the chromosome of strain PAR-47. Since 4HPE and 4HPAA were produced from 4HPAAL by endogenous aldehyde reductase and feaB-encoded aldehyde dehydrogenase, respectively, in strain PAR-47, deletion of the relevant aldehyde reductase would cause a decrease in 4HPE production, coinciding with an increase in 4HPAA production. The ratio of 4HPE/4HPAA accumulation by strain PAR-47 was 1.5, whereas the ratios by strain PAR-75 (yqhD deficient), PAR-77(yjgB deficient), and PAR-79 (yahK deficient) were 0.7, 1.0, and 1.0, respectively (Table 6). Furthermore, the ratio was reduced to 0.5 by strain PAR-90 lacking both yqhD and yahK. However, 4HPE titers were not significantly reduced in all deletion variants, suggesting that another endogenous reductase acts on the production of 4HPE.

Table 6.

Effect of gene deletion on production of 4HPE and 4HPAA^a

Strain	Deleted gene(s)	Amt (mM) of product formedb		Ratio of 4HPE/4HPAA formed
		4HPE	4HPAA
PAR-47		2.6 ± 0.1	1.7 ± 0.0	1.5
PAR-75	yqhD	3.4 ± 0.1	4.8 ± 0.3	0.7
PAR-77	yjgB	4.3 ± 0.3	4.5 ± 0.4	1.0
PAR-79	yahK	2.7 ± 0.3	2.9 ± 0.4	0.9
PAR-89	yqhD yjgB	2.0 ± 0.7	2.7 ± 0.8	0.7
PAR-90	yqhD yahK	2.2 ± 0.4	4.3 ± 0.2	0.5
PAR-91	yjgB yahK	2.5 ± 0.4	2.4 ± 0.3	1.0
PAR-92	yqhD yjgB yahK	3.0 ± 0.1	2.2 ± 0.1	1.4

^aThe strains were cultivated at 37°C (before induction of gene expression) and 27°C (after induction of gene expression) for 48 h.

^bThe values are means ± standard deviations obtained from 3 independent cultures.

The pathways for PE and 4HPE syntheses include some branching points. The first branching point is the locus of prephenic acid, the second branching point is the locus of PP or 4HPP, and the third one is the locus of PAAL or 4HPAAL (Fig. 2). Because undesirable metabolic flow would reduce the yield of each target compound and increase the amount of by-products, undesirable pathways should be blocked. Indeed, throughout PE and 4HPE syntheses, undesired PAA and 4HPAA syntheses occurred, due to the presence of endogenous feaB-encoded PAAL dehydrogenase (Table 5). Endogenous feaB was therefore deleted from PAR-68 (PE producer) and PAR-65 (4HPE producer) to block the PAA and 4HPAA synthetic routes, generating strains PAR-84 and PAR-83, respectively (modification of the third branching point in Fig. 2). The amounts of PE and 4HPE produced by strains PAR-84 and PAR-83, respectively, were nearly equal to those by strains PAR-68 and PAR-65 (Table 5). PAA and 4HPAA by-products were not detected after deletion of endogenous feaB, as shown by strains PAR-84 and PAR-83.

Fig 2.

PE, 4HPE, PAA, and 4HPAA synthesis and degradation pathways in E. coli BW25113(DE3). Three metabolic branching points are surrounded by broken lines, whereas degradation pathways are emphasized by gray shading. Host strain BW25113(DE3), which is a derivative of E. coli strain K-12, lacks the hpa operon which encodes enzymes involved in 4HPAA degradation. TCA, tricarboxylic acid.

4HPE by-product was accumulated in PE production by strain PAR-84, whereas the accumulation of PE by-product was increased in 4HPE production by strain PAR-83 (Table 5). Because the host strain BW25113 accumulated PE and PAA but not 4HPE and 4HPAA when cultivated in M9M medium supplemented with 1 mM PAAL, we inferred that 4HPE by-product had accumulated through leakage into the Tyr synthetic route in strain PAR-84 (the first branching point in Fig. 2). Similarly, it was thought that accumulation of PE by-product by strain PAR-83 was due to leakage into the Phe synthetic route. To avoid such leakage, the tyrA gene of PE producer PAR-84 and the pheA gene of 4HPE producer PAR-83 were deleted to generate strains PAR-105 and PAR-104, respectively (modification of the first branching point in Fig. 2). Since these strains became Tyr or Phe auxotrophic, each aromatic amino acid was added to the medium: the addition of aromatic amino acids contributes to cell growth rather than production of target product. As a result, accumulation of 4HPE by-product during PE production by strain PAR-105 was appreciably decreased (Table 5). Similarly, PE by-product during 4HPE production by strain PAR-104 was also decreased. In addition, the amount of 4HPE produced by strain PAR-104 considerably increased more than that by strain PAR-83.

Blockage of degradation pathways of the target compounds produced is one of the considerable issues that must be resolved (Fig. 2). As the host strain BW25113(DE3) is a K-12 derivative, it can degrade PAA through the PAA degradation pathway, components of which are encoded by the paa operon, but it cannot degrade 4HPAA because it lacks the hpa operon (10). On the other hand, degradation pathways of PE and 4HPE are unclear. The strain could not degrade 4HPE in M9M medium supplemented with 1 mM 4HPE, nor could 4HPE be used as the sole carbon and energy source in glucose-free M9M medium, suggesting that 4HPE is the final product. In contrast, PE was slightly degraded by the strain (less than 10%) when the strain was cultivated at 27°C for 48 h in M9M medium supplemented with 1 mM PE. Although the degradation pathway of PE is unclear, it is thought that PE was oxidized to PAA through PAAL formation and then degraded via the PAA degradation pathway. In fact, PE was not degraded by a feaB-deficient strain (AR-G47) when the strain was cultivated at 27°C for 48 h in M9M medium supplemented with 1 mM PE. Therefore, we think that deletion of feaB, shown as strains PAR-84 and PAR-105, might contribute to prevent PE degradation.

Construction and evaluation of PAA and 4HPAA producers.

The PAR-60 and PAR-47 strains were further modified by incorporation of the T7 promoter-controlled feaB (T7p-feaB) gene to generate PAR-61 (PAA producer) and PAR-51 (4HPAA producer), respectively (Table 1). Overexpression of both T7p-ipdC and T7p-feaB in strains PAR-61 and PAR-51 resulted in accumulation of PAA and 4HPAA (Table 7). However, a small amount of accumulated PAA by-product was detected in 4HPAA production by strain PAR-51. Similarly, PAR-61 acquired small quantities of 4HPAA by-product during PAA production.

Table 7.

Production of PAA and 4HPAA by the genetically modified strains^a

Strainb	Strain description and relevant inserted and deleted gene(s)c	Mediumd	Amt (mM) of glucose consumede,f	OD660e,g		Amt (mM) of major product producedh				Yieldi (%, mol/mol)
				Ind.	Fin.	PE	4HPE	PAA	4HPAA	Yp/s	Ybp/s
Strains constructed for PAA production
AR-G91	Phe overproducer	M9M	54.1	0.3	6.5	ND	ND	ND	ND	NC	NC
PAR-60	Phe overproducer + ipdC	M9M	48.4	0.3	4.5	4.6 ± 0.5	2.4 ± 0.1	1.7 ± 0.2	0.9 ± 0.1	3.1	12.9
PAR-61	Phe overproducer + ipdC + feaB	M9M	48.4	0.3	4.5	ND	ND	5.2 ± 0.0	1.8 ± 0.0	9.4	3.2
PAR-82	Phe overproducer + ipdC + feaB − paaK	M9M	46.2	0.3	5.1	ND	ND	2.9 ± 0.0	2.3 ± 0.0	5.2	4.1
PAR-100	Phe overproducer + ipdC + feaB − tyrA	M9M + Tyr	47.0	2.0	3.1	ND	ND	8.8 ± 0.1	1.3 ± 0.0	15.8	2.3
Strains constructed for 4HPAA production
AR-G2	Tyr overproducer	M9M	55.5	0.3	7.9	ND	ND	ND	ND	NC	NC
PAR-47	Tyr overproducer + ipdC	M9M	50.2	4.0	5.1	0.3 ± 0.0	2.6 ± 0.1	<0.2	1.7 ± 0.0	3.1	<5.6
PAR-51	Tyr overproducer + ipdC + feaB	M9M	50.2	4.0	5.1	ND	ND	0.5 ± 0.0	5.3 ± 1.0	9.5	0.9
PAR-97	Tyr overproducer + ipdC + feaB − yahK	M9M	48.4	4.0	4.6	ND	ND	0.6 ± 0.0	4.9 ± 0.5	8.8	1.1
PAR-102	Tyr overproducer + ipdC + feaB − yahK − pheA	M9M + Phe	44.5	2.0	2.4	ND	ND	0.8 ± 0.0	6.1 ± 0.4	13.2	1.4

^aThe strains were cultivated at 37°C (before induction of gene expression) and 27°C (after induction of gene expression) for 48 h.

^bStrains AR-G91 and AR-G2 were phenylalanine and tyrosine overproducers, respectively, and used as control strains.

^c+ before a gene indicates that the gene was inserted, whereas − before a gene indicates that the gene was deleted. Expression of the inserted genes was controlled by the T7 promoter.

^dThe concentrations of supplemented Phe and Tyr were 1.0 and 1.5 mM, respectively.

^eValues were obtained from three independent cultures, and the maximum coefficient of variation was below 10%.

^fThe initial glucose concentration was 55.6 mM (10 g/liter).

^gInd., OD₆₆₀ when the induction of gene overexpression was started; Fin., final OD₆₆₀.

^hThe values are means ± standard deviations obtained from 3 independent cultures. ND, not detected.

ⁱY_p/s, yield in accumulated target product (mM) per initial glucose (55.6 mM) × 100%; Y_bp/s, yield in accumulated by-product (mM) per initial glucose (55.6 mM) × 100%. NC, not calculable.

As strain PAR-61 can operate the PAA degradation pathway (10), we modified the strain by deleting paaK, which encodes phenylacetyl-coenzyme A (CoA) ligase (catalyzing the first step of the PAA degradation pathway), to prevent PAA degradation. However, the accumulated PAA by the resulting strain, PAR-82, was considerably decreased (Table 7); the reason is unclear. On the other hand, endogenous yahK of strain PAR-51 was deleted to block the 4HPE synthetic route (the third branching point in Fig. 2), but it was not meaningful with respect to 4HPAA production, as shown by strain PAR-97. To prevent the production of 4HPAA by-product via leakage into the Tyr synthetic route (the first branching point in Fig. 2) in strain PAR-61, the tyrA gene was deleted to generate strain PAR-100. Similarly, the pheA gene of 4HPAA producer PAR-97 was deleted to generate strain PAR-102. The deletion of tyrA or pheA allowed strain PAR-100 or PAR-102 to produce a larger amount of PAA or 4HPAA and a smaller amount of by-product.

Productivities of aromatic compounds by the constructed strains.

Production of aromatic compounds by the respective constructed strains, which contained PLA by strain PAR-58, 4HPLA by strain PAR-3, PE by strain PAR-105, 4HPE by strain PAR-104, PAA by strain PAR-100, and 4HPAA by strain PAR-102, was periodically analyzed (Fig. 3). For comparison, time course data of the production of Phe (by strain AR-G91) and Tyr (by strain AR-G2) were also shown in Fig. 3. Gene overexpression was induced by the addition of IPTG at an OD₆₆₀ of 0.3 when strains AR-G91, AR-G2, PAR-3 and PAR-58 were cultivated. On the other hand, since the cell growth of strains PAR-105, PAR-104, PAR-100, and PAR-102 were depressed by IPTG addition, gene overexpression was induced at an OD₆₆₀ of 2.0. Under such conditions, the final OD₆₆₀ of the former strains ranged from 5.8 to 8.8, whereas that of the latter strains ranged from 2.4 to 3.4.

Fig 3.

Time course analyses of the production of the respective aromatic compounds by the constructed strains on a shaking test tube platform. E. coli strains AR-G91 (Phe overproducer), AR-G2 (Tyr overproducer), PAR-58 (PLA producer), and PAR-3 (4HPLA producer) were cultivated in M9M medium. Strains PAR-105 (PE producer) and PAR-100 (PAA producer) were cultivated in M9M medium plus 1.5 mM Tyr, whereas strains PAR-104 (4HPE producer) and PAR-102 were cultivated in M9M medium plus 1 mM Phe. The black arrows indicate the time of the addition of IPTG for gene overexpression. The strains were cultivated at 37°C (before IPTG addition) and 27°C (after IPTG addition). The data are averages of triplicate cultures, and the maximum variation was less than 10%. The respective symbols represent as follows: +, glucose concentration; ×, OD₆₆₀; open circle, Phe concentration; open triangle, Tyr concentration; open square, PLA concentration; open diamond, 4HPLA concentration; closed circle, PE concentration; closed triangle, 4HPE concentration; closed square, PAA concentration; closed diamond, 4HPAA concentration. The broken lines represent by-products. Abbreviations are summarized in Materials and Methods.

Time course experiments revealed the occurrence of by-products during the production of the respective aromatic compounds. In 4HPLA production by strain PAR-3, Tyr by-product was detected in the early stage (within 24 h) of cultivation (Fig. 3). However, the by-product disappeared later (after 24 h). In contrast, Phe by-product was gradually increased during PLA production by strain PAR-58. On the other hand, by-products of PE, 4HPE, PAA, and 4HPAA production arose in the early stage (within 10 h) and their accumulated levels were maintained until the end of cultivation, suggesting that these by-products were derived from supplemented Phe or Tyr but not from glucose.

Maximum volumetric and specific productivities of PLA, 4HPLA, PE, 4HPE, PAA, and 4HPAA by the respective constructed strains were determined from the time course data and summarized in Table 8. Due to the by-product produced, the maximum specific productivity (q_p, _max) of PLA by strain AR-G58 was slightly lower (2.6 μmol g⁻¹ CDW min⁻¹) than that of Phe by strain AR-G91 (3.3 μmol g⁻¹ CDW min⁻¹). In contrast, the maximum specific productivity of 4HLA by strain PAR-3 (2.5 μmol g⁻¹ CDW min⁻¹) was similar to that of Tyr by strain AR-G2 (2.2 μmol g⁻¹ CDW min⁻¹). On the other hand, the maximum specific productivities of PE by strain PAR-105 (4.2 μmol g⁻¹ CDW min⁻¹), 4HPE by strain PAR-104 (4.5 μmol g⁻¹ CDW min⁻¹), PAA by strain PAR-100 (5.8 μmol g⁻¹ CDW min⁻¹), and 4HPAA by strain PAR-102 (4.6 μmol g⁻¹ CDW min⁻¹) were obviously higher than those of Phe by strain AR-G91 and Tyr by strain AR-G2.

Table 8.

Volumetric and specific productivities^a

Strain	Target product	Mediumb	rp, max (μmol l−1 min−1)	qp, maxc (μmol g−1 CDW min−1)
AR-G91	Phe	M9M	10.0 ± 0.6	3.3 ± 0.1
AR-G2	Tyr	M9M	4.8 ± 0.6	2.2 ± 0.1
PAR-58	PLA	M9M	6.4 ± 0.2	2.6 ± 0.1
PAR-3	4HPLA	M9M	7.6 ± 0.3	2.5 ± 0.2
PAR-105	PE	M9M + Tyr	6.2 ± 0.1	4.2 ± 0.2
PAR-104	4HPE	M9M + Phe	7.9 ± 0.3	4.5 ± 0.5
PAR-100	PAA	M9M + Tyr	8.2 ± 0.2	5.8 ± 0.1
PAR-102	4HPAA	M9M + Phe	5.8 ± 0.2	4.6 ± 0.6

^aThe strains were cultivated at 37°C (before induction of gene expression) and 27°C (after induction of gene expression) for 72 h. An aliquot of the culture was periodically sampled to determine volumetric and specific productivity parameters.

^bPhe and Tyr were supplied at 1.0 and 1.5 mM, respectively.

^cMaximum specific productivity was calculated by the formula q_p = r_p/Mx (27), where r_p is the volumetric productivity (μmol liter⁻¹ min⁻¹) and Mx is the cell dry weight (CDW) (g l⁻¹).

DISCUSSION

E. coli has become a promising host organism for the microbial production of a variety of valuable chemicals, including biofuel (8), lactic acid (19), succinic acid (6), etc., from renewable resources. The aim of this study was to generate E. coli strains able to produce a range of aromatic compounds from glucose. Our strategy—expanding the shikimate pathway—is exceedingly simple. Homologous/heterologous genes were introduced into the chromosomal DNA of aromatic amino acid overproducers. To accomplish this, the heterologous genes were surveyed from bacteria to be expressed in soluble and active form in E. coli. Conclusively, we successfully constructed metabolically engineered E. coli strains capable of producing six kinds of aromatic compounds, namely, PLA, 4HPLA, PE, 4HPE, PAA, and 4HPAA. To the best of our knowledge, this is the first report of the bioproduction of PLA, 4HPLA, PAA, and 4HPAA from glucose by metabolically engineered E. coli.

PLA is known as a broad-spectrum antimicrobial agent which is naturally produced by some classes of microorganisms, especially lactic acid bacteria. An enzyme responsible for converting PP to PLA is required in PLA production. Batch and fed-batch fermentation by Lactobacillus, optimized by PP feeding and pH control, has been proposed as one mechanism of PLA production (24). Whole cells of Bacillus coagulans SDM have also been shown to effectively convert PP to PLA (46). In both cases, PP was used as a substrate for PLA production; however, PP is too unstable and expensive for industrial and/or commercial use. In our study, glucose could be used as a substrate, because the dehydrogenase gene (ldhA), responsible for PP reduction activity, was introduced into the chromosomal DNA of Phe-overproducing E. coli. Applying this strategy to the Tyr overproducer, we found that 4HPLA could be similarly produced from a glucose substrate.

PE is an important flavor and fragrance compound with a rose-like odor. Little is currently known about whether bacteria can synthesize PE, but Microbacterium sp. and Brevibacterium linens are known to do so (15). By comparison, many fungal and yeast species are PE producers. Indeed, yeasts are the most prominent natural PE-synthesizing microorganisms, and PE production by yeast has been extensively studied (11). Yeasts produce PE through the Ehrlich pathway which involves transamination to the corresponding alpha-keto acids, followed by decarboxylation to an aldehyde, and final reduction to an alcohol. Current yeast production processes require Phe as a substrate. On the other hand, we successfully produced PE by metabolically engineering E. coli, using glucose as a substrate instead of Phe. In our study, the shikimate pathway of a Phe overproducer was expanded by introducing two bacterial heterologous genes (T7p-ipdC and T7p-adhC) whose effects mimic the Ehrlich pathway. Our attempt was successful, but an unexpected phenomenon arose. Introducing the PP decarboxylase gene (T7p-ipdC) induced automatic reduction of PAAL in vivo, supposedly by endogenous aldehyde reductases present in E. coli, leading to PE accumulation. By surveying such reductase genes, three candidate genes, yqhD, yjgB, and yahK, were identified. Of these, yqhD-encoded reductase alone has been experimentally investigated. This reductase is a broad-substrate alcohol dehydrogenase (14), and as such, has been used to generate strains capable of producing 1,3-propandiol (37) and isobutanol (1). We propose here that other genes, yjgB and yahK, could be used for PE production from glucose. The Ehrlich pathway was also simulated in a Tyr overproducer to generate a 4HPE-producing strain. In this case, the T7p-adhC gene product did not sufficiently function as a 4HPAAL reductase (Table 5). However, three of the genes encoding aldehyde reductases (yqhD, yjgB, and yahK) were implicated in successful 4HPE production. Introducing T7p-yahK resulted in lower accumulated levels of 4HPAA by-product. In recent work, it was shown that an E. coli strain overexpressing both the tyramine oxidase gene from Micrococcus luteus and the Tyr decarboxylase gene from Papaver somniferum, without any exogenous 4HPAAL reductases, produced 4HPE from glucose, suggesting the presence of a reductase capable of acting on 4HPAAL (34). The enzymatic data in the present study indicated that yqhD-, yjgB-, and yahK-encoded reductases can act as 4HPAAL reductases (Table 3). However, the results of the gene deletion experiment suggest that other reductases can also promote 4HPE production (Table 6).

Oxidation of PAAL to PAA and 4HPAAL to 4HPAA are well-known processes. In our study, oxidation was carried out using feaB-encoded PAAL dehydrogenase, which is involved in E. coli PEA and TYM degradation pathways (10, 30). Introduction of T7p-ipdC alone thus resulted in PAA and/or 4HPAA production via decarboxylation of Phe and Tyr by ipdC-encoded decarboxylase, followed by oxidation by the endogenous feaB-encoded dehydrogenase in strains PAR-60 and PAR47 (Table 7). PAA and 4HPAA production was enhanced when feaB was overexpressed together with ipdC in strains PAR-61 and PAR-51.

We encountered several obstacles in the production of the respective aromatic compounds, which we attempted to resolve. During 4HPE production by strain PAR-65, small amounts of PAA, 4HPAA, and PE by-products were accumulated (Table 5). In this case, we thought that deletion of the endogenous feaB gene encoding aldehyde dehydrogenase might prevent PAA and 4HPAA production (10, 12). In fact, the constructed strain PAR-83, a PAR-65 derivative lacking feaB, did not produce such by-products. The same effect was observed in PE producer PAR-84 lacking the feaB gene (Table 5).

The most serious problem was that hydroxyphenyl compounds, 4HPE and 4HPAA, accumulated as by-products in the production of PE and PAA. That is, a certain amount of 4HPE was accumulated during PE production by strain PAR-68 (Table 5), while the PAA produced by strain PAR-61 was contaminated with 4HPAA (Table 7). For reasons that are not entirely clear, 4HPE and 4HPAA leakages may have occurred through Tyr synthesis. In such cases, we thought that blocking the Tyr synthetic route might reduce accumulation of unwanted by-products. In addition, blockage of the Tyr synthetic route might contribute to the enhanced yield of target product by depressing carbon flux into the cell. Indeed, in a previous study of Phe production, deletion of the tyrA gene in E. coli strains overexpressing both aroF and pheA was shown to markedly enhance Phe production (36). Blockage of the Tyr synthetic route by deletion of tyrA in PE- and PAA-producing strains did indeed repress accumulation of 4HPE (strain PAR-105 in Table 5) and 4HPAA (strain PAR-100 in Table 7). Furthermore, the titers and yields of 4HPE and 4HPAA were increased. However, deletion of the tyrA gene rendered strains PAR-105 and PAR-100 Tyr auxotrophic, so they required supplemental Tyr in the M9M medium for growth. The observed lower levels of 4HPE (PAR-105) and 4HPAA (PAR-100) by-products accumulated were thought to mean that supplemental Tyr for cell growth was converted to 4HPE and 4HPAA through the Ehrlich pathway. Although PP decarboxylase encoded by the ipdC gene from A. brasilense, which acts on both PP and 4HPP, was used in the present study, use of a PP-specific enzyme might encourage PE and PAA production without amassing 4HPE and 4HPAA as unwanted by-products.

Maximum theoretical yield coefficients of produced Phe from glucose have been calculated from the known stoichiometry of Phe biosynthesis from glucose in different metabolic scenarios (28). Although PEP is converted to pyruvate during glucose transport by PTS without recycling of pyruvate to PEP in wild-type E. coli, the calculated maximum theoretical yield is 0.275 g/g, which corresponds to 30% (mol/mol). On the other hand, by PTS inactivation or pyruvate recycling in an engineered strain, the maximum theoretical yield increased 2-fold (0.55 g/g), which corresponds to 60% (mol/mol); these modifications have already been achieved in Phe biosynthesis (7, 13, 36). These scenarios are also applicable to the yields of Tyr and our target aromatic compounds. Since Phe overproducer AR-G91 and Tyr overproducer AR-G2, which are used as host strains for aromatic compound producers, take up glucose through PTS and do not recycle pyruvate, the maximum theoretical yields of Phe and Tyr by these strains are 30% (mol/mol). Similarly, maximum theoretical yields of PLA, 4HPLA, PE, 4HPE, PAA, and 4HPAA by the constructed strains derived from strains AR-G91 and AR-G2 are also 30% (mol/mol). Therefore, the constructed strains in this study should be further genetically modified to increase the maximum theoretical yield up to 60% (mol/mol) as a future issue. Such modifications would also enhance the volumetric and specific productivities of the respective aromatic compounds.

The replacement of petroleum with renewable sources in industrial chemical processing is of urgent concern, and here we address this rapidly expanding field of research. As an initial step toward our goal, we verified that six kinds of aromatic compounds can be produced from glucose, a primary component of renewable resources such as starch and cellulose. The productivities of six aromatic compounds by the constructed strains were the same or higher in comparison to those of Phe (by strain AR-G91) and Tyr (by strain AR-2) (Table 8), so that the possibility of the production of aromatic compounds from renewable resources was shown here. We further found that yjgB- and yahK-encoded enzymes probably act as aldehyde reductases on PAAL and 4HPAAL. Although these reductases require enzymological data, novel functions of some putative genes could be revealed through metabolic engineering studies such as that presented here.