Of the 20 standard L-amino acids, four are classified as aromatic amino acids (AAAs); L-phenylalanine (Phe), L-tyrosine (Tyr), L-tryptophan (Trp), and L-Histidine, which have an aromatic ring in their side chain. Three of them, Phe, Trp, and L-histidine, are essential amino acids for humans. Moreover, Phe, Trp, and Tyr are the starting compounds in the biosynthesis of various hormones and neurotransmitters.

Biosynthesis of the AAAs (hereafter, Phe, Trp, and Tyr) in microorganisms and plants has been well studied and is known to be regulated at the levels of transcription and enzyme activity. These AAAs have been used in animal feed and as precursors for the synthesis of industrial and pharmaceutical compounds. Based on knowledge of AAA biosynthetic pathways and their regulation, microbial cells have been engineered for fermentative AAA production. Recently, production of AAA derivatives using microbial cells has also been studied. In this review, we provide an overview of AAA biosynthesis and its regulation in Escherichia coli and Corynebacterium glutamicum, a coryneform bacterium used as a host for producing amino acids and other materials. Studies on production of AAAs and their derivatives using these bacteria are also introduced. It is known well that E. coli exhibits high growth rate and its genetic engineering tools, which can be used for breeding production host species, have highly been developed. As for C. glutamicum, various achievements in amino acid production have been conducted. Considering production of AAAs and their derivatives in particular, C. glutamicum exhibits higher tolerance to aromatic compounds, such as 4-hydroxybenzoate (Kitade et al. 2018) and p-aminobenzoate (Kubota et al. 2016), than other bacteria. Therefore, both E. coli and C. glutamicum are efficient host species for producing AAAs and their derivatives.

Biosynthesis of aromatic amino acids

In bacteria, yeasts, fungi, and plants, AAAs are biosynthesized via a common metabolic pathway, the shikimate pathway (Fig. 1). The enzymes, substrates, products, and genes involved in AAA biosynthesis in E. coli and C. glutamicum are listed in Supplementary Table S1.

Fig. 1

Metabolic pathways for chorismate biosynthesis (the shikimate pathway) in microorganisms

AAAs are biosynthesized from the common metabolite chorismate, which is produced via the shikimate pathway. In this pathway, phosphoenolpyruvate (PEP) and erythrose-4-phosphate (E4P), which are synthesized in the glycolysis and pentose phosphate pathway, respectively, are first condensed to yield 3-deoxy-D-arabinoheptulosonate-7-phosphate (DAHP) in a reaction catalyzed by DAHP synthase (DAHPS) (Reaction 1; the reaction numbers in this section are shown in Figs. 1 and 2 and Supplementary Table S1). Then, DAHP is converted to shikimate via three reactions (Reactions 2–4), and shikimate is converted to chorismate via three enzymatic reactions (Reactions 5–7). In the 5-enolpyruvylshikimate-3-phosphate synthase reaction (Reaction 6), PEP is condensed with shikimate-3-phosphate to yield 5-enolpyruvylshikimate-3-phosphate.

Fig. 2

Terminal metabolic pathways for biosynthesis of aromatic amino acids from chorismate in microorganisms

In the terminal pathways for Phe and Tyr biosynthesis (Fig. 2), chorismate is converted to prephenate via Claisen rearrangement with chorismate mutase (CM) (Reaction 8). In Phe biosynthesis, prephenate is metabolized to yield phenylpyruvate via decarboxylation by prephenate dehydratase (PDT) (Reaction 9), whereas in Tyr biosynthesis, chorismate is converted to 4-hydroxyphenylpyruvate via oxidative decarboxylation by prephenate dehydrogenase (PDH) (Reaction 10). Phenylpyruvate and 4-hydroxyphenylpyruvate are then converted to Phe and Tyr, respectively, via transamination by tyrosine aminotransferase (Reaction 11). Some bacteria including C. glutamicum and Pseudomonas aeruginosa have another Tyr biosynthesis pathway, in which Tyr is biosynthesized from arogenate (pretyrosine), as a conversion product of prephenate (Patel et al. 1977; Fazel and Jensen 1979).

In Trp biosynthesis (Fig. 2), chorismate is converted to anthranilate by anthranilate synthase (Reaction 12). Anthranilate is condensed with phosphoribosylpyrophosphate (PRPP) by anthranilate phosphoribosyltransferase to yield phosphoribosylanthranilate (Reaction 13). Phosphoribosylanthranilate is then metabolized by phosphoribosylanthranilate isomerase to produce carboxyphenylaminodeoxy-D-ribulose-5-phosphate (Reaction 14), which is further converted to indole-3-glycerol phosphate by indole-3-glycerol phosphate synthase (Reaction 15). Finally, indole-3-glycerol phosphate is converted to indole and 3-phosphoglycelaldehyde by tryptophan synthase α chain (TrpA), and then Trp is synthesized from indole and L-serine by tryptophan synthase β chain (TrpB) (Reaction 16).

In relation to the shikimate pathway, C. glutamicum possesses metabolic pathway for assimilating quinate and shikimate (Supplementary Fig. S1) (Teramoto et al. 2009; Kubota et al. 2014). It is thought that both quinate and shikimate are incorporated into the C. glutamicum cells through a protein encoded by pcaA, which belongs to the major facilitator superfamily. Quinate and shikimate are converted to 3-dehydroquinate and 3-dehydroshikimate, respectively, by shikimate 5-dehydrogenase encoded by qsuD. 3-Dehydroquinate is converted to 3-dehydroshikimate by 3-dehydroquinate dehydratase encoded by qsuC, which belongs to the shikimate pathway. 3-Dehydroshikimate is further metabolized to protocatechuate by 3-dehydroshikimate dehydratase encoded by qsuB. It is assumed that protocatechuate is finally metabolized to succinyl-CoA and acetyl-CoA, which are further metabolized in the TCA cycle. Additionally, qsuA, qsuB, qsuC, and qsuD genes constitute a single operon on the C. glutamicum genome and its expression is regulated by the chorismate-dependent transcriptional regulator encoded by qsuR which is located just upstream of the qsuABCD operon in the opposite direction (Kubota et al. 2014).

Regulation of aromatic amino acid biosynthesis in E. coli and C. glutamicum Regulation of aromatic amino acid biosynthesis by modulating enzyme activity

In the AAA biosynthesis pathway, carbon flow to the shikimate pathway is regulated through feedback inhibition, as the terminal AAAs control the enzymatic activity of DAHPS. E. coli has three isozymes of DAHPS, which are encoded by aroF, aroG, and aroH, and their activities are inhibited by Tyr, Phe, and Trp, respectively (Brown 1968). In contrast, C. glutamicum harbors two DAHPS isozymes, which are encoded by aroF and aroG. The activity of AroG is moderately inhibited by Trp, whereas that of AroF is inhibited by Tyr and Phe (Liu et al. 2008). AroG is the dominant enzyme in the shikimate pathway (Liu et al. 2008).

The biosynthesis of AAAs is also controlled by feedback inhibition of the enzymes catalyzing the terminal reactions, i.e. conversion of chorismate to the AAAs. During Phe biosynthesis in E. coli, the activity of the bifunctional enzyme CM-PDT, encoded by pheA, is inhibited by Phe (Dopheide et al. 1972). In Tyr biosynthesis in E. coli, Tyr inhibits activity of the bifunctional enzyme CM-PDH, which is encoded by tyrA (Hudson et al. 1983). In Trp biosynthesis in E. coli, Trp inhibits the activities of anthranilate synthase and anthranilate phosphoribosyltransferase (Ito and Crawford 1965).

In C. glutamicum, AAA biosynthesis is also controlled by feedback inhibition of the enzymes that catalyze the terminal reactions. AroG from C. glutamicum forms a complex of its tetramer with a dimer of CM encoded by csm (Li et al. 2013; Burschowsky et al. 2018). Complex formation is not required for DAHPS activity but is essential for CM activity. Phe inhibits complex formation between AroG and Csm, resulting in inhibition of CM activity. Phe also inhibits PDT, which is encoded by pheA. Like in E. coli, in Trp biosynthesis pathway of C. glutamicum, Trp inhibits the activities of both anthranilate synthase and anthranilate phosphoribosyltransferase.

Transcriptional regulation of aromatic amino acid biosynthesis

The AAA biosynthesis are also transcriptionally regulated by AAA availability. In E. coli, production of DAHPS is controlled by transcriptional repression with the AAAs (i.e. feedback repression). Expression of aroH and aroF, which encode DAHPS isozymes, is negatively regulated by the transcriptional regulators TyrR and TrpR (Muday et al. 1991). These transcriptional regulators with AAAs bind to the regions upstream of target genes encoding DAHPS to repress their expression.

An important finding regarding the regulation of gene expression in the AAA biosynthesis is attenuation of the trpEDCBA operon in E. coli (Yanofsky 1981). Expression of this operon is regulated by the availability of Trp-charged tRNA, as trpL encodes a Trp-containing leader peptide upstream of the operon. Expression of the pheA gene encoding CM-PDT, which has the pheL leader peptide sequence upstream, is similarly regulated by the availability of Phe-charged tRNA (Zurawski et al. 1978; Gavini and Pulakat 1991). Expression of the pheST operon encoding the Phe-tRNA synthetase complex is also similarly regulated by attenuation (Springer et al. 1985).

Transcription attenuation of the trpEGDCFBA operon has also been reported in C. glutamicum (Neshat et al. 2014). The trpL encoding a leader peptide with Trp residues is located upstream of this operon, and transcription of this operon is enhanced by ribosome stalling at Trp codons in the trpL mRNA resulting from depletion of Trp-charged tRNA. The aroR gene also encodes a leader peptide upstream of the operon containing aroF, which encodes DAHPS (Neshat et al. 2014). The AroR leader peptide contains Phe-Tyr-Phe residues, and transcription of the operon containing aroF is enhanced by ribosome stalling at Phe and Tyr codons in aroR mRNA under Phe-limited conditions. However, Tyr availability does not affect transcription of this operon.

In addition to attenuation due to ribosome stalling, expression of the trpEGDCFBA operon is also regulated by the IclR-type transcriptional regulator LtbR (Brune et al. 2007). The ltbR gene is located upstream of the leuCD operon, which is related to L-leucine biosynthesis, and LtbR negatively regulates the expression of both the leuCD operon and the trpEGDCFBA operon. The LtbR consensus binding sequence in the –10 region of the promoter was also found in the promoter regions of the trpL and aroG genes, which encode the leader peptides for the trpEGDCFBA operon and DAHPS, respectively.

Fermentative production of aromatic amino acids by E. coli and C. glutamicum

In studies conducted in the 1970s, release from feedback inhibition of DAHPS and the enzymes in the terminal pathways for AAA biosynthesis enhanced the production of AAAs using microbial cells. In these studies, C. glutamicum mutant strains showing resistant to AAA analogs, which inhibit cell growth by affecting the related biosynthesis reactions, were isolated as AAA production hosts (Hagino and Nakayama 1973, 1974, 1975; Shiio et al. 1984). In the 1990s, AAA-producing strains were rationally created based on the concept of metabolic engineering (Bailey 1991; Stephanopoulos and Vallino 1991). In this section, we describe studies on fermentative production of AAAs based on metabolic engineering of E. coli and C. glutamicum reported after 2010 (Supplementary Table S2).

Phenylalanine

Modification of the shikimate and terminal biosynthesis pathways is one engineering strategy for producing AAAs in E. coli and C. glutamicum. Another strategy is enhancement of the supply of substrates for DAHPS, PEP and E4P. Liu et al. (2013) demonstrated that overexpression of a truncated pheA that has the coding region for the catalytic domain of CM-PDT and a mutant aroG encoding feedback-resistant DAHPS enhanced Phe production by E. coli wild-type and elevating expression of the ydiB and aroK genes encoding shikimate dehydrogenase and shikimate kinase, respectively, further improved the productivity. Ding et al. (2016) found that increase in the amount of shikimate kinase and 5-enolpyruvylshikimate-3-phosphate synthase encoded by aroL and aroA, respectively, enhanced Phe production in E. coli based on absolute quantification of the enzymes related to shikimate synthesis and an in vitro system using purified enzymes. Overexpression of the aroA gene successfully increased Phe production by the recombinant E. coli strain to about 62 g L–1 in fed-batch culture. Wu et al. (2019) applied a dynamic regulation strategy to generate Phe-producing strains of E. coli. In this study, modified promoters for tyrP, whose transcription is upregulated by the transcriptional regulator TyrR in the presence of Phe, were screened and used for dynamic control of the expression of aroK, which encodes shikimate kinase, in the previously constructed Phe-producing E. coli mutant strain.

Recently, Wang et al. (2024) reported breeding a Phe-producing strain by expressing endogenous and exogenous genes related to the shikimate pathway and Phe biosynthesis from various promoters in a shikimate-producing E. coli strain. In addition, they used adaptive laboratory evolution to isolate E. coli cells with tolerance to high Phe concentration and found that marA, which encodes a transcriptional regulator, was responsible for tolerance to Phe. Integration of enhanced flux for the shikimate and terminal Phe biosynthesis pathway with high Phe tolerance by overexpression of marA yielded about 80 g L–1 Phe in fed-batch cultivation.

As with Phe production using engineered E. coli strains, increased flux of the shikimate and terminal Phe biosynthesis pathways also enhanced Phe production in C. glutamicum (Zhang et al. 2013, 2014). Zhang et al. (2015) investigated the effect of Phe biosynthesis gene overexpression on Phe production to identify the key enzymes involved in Phe production. Subsequently, they introduced various expression modules for identified genes encoding key enzymes in the wild-type strain and evaluated their effects on Phe productivity. Phe production was further improved by modifying the phosphotransferase system (PTS), which is responsible for sugar uptake coupled with conversion of PEP to pyruvate, to supply PEP and blocking the production of lactate and acetate.

Recently, Kataoka et al. (2023) conducted stepwise metabolic engineering of C. glutamicum for Phe production. They achieved about 8 g L–1 Phe production by overexpressing wild-type aroH and mutant pheA genes from E. coli cloned on a plasmid and the aroE gene on the genome combined with disruption of hdpA, qsuB, qsuD, tyrA, and ppc to avoid utilizing intermediate metabolites in the shikimate pathway for other metabolic pathways, reduce competing Tyr production, and enhance the PEP supply to the DAHPS reaction.

Tachikawa et al. (2024) metabolically engineered C. glutamicum for Phe production using adaptive laboratory evolution based on long-term repetitive passage cultures to isolate mutants showing resistance to a Phe analog. They found that analog-resistant mutants had the potential to produce both Phe and Tyr. Since the mutants carried mutations in the aroG and pheA genes, they analyzed AAA production by the wild-type C. glutamicum strain overexpressing both mutant aroG and pheA, which produced about 3 g L–1 Phe. Then, Phe production was further improved up to 6 g L–1 by disrupting the aroP gene, which encodes AAA permease.

Tyrosine

As with Phe production, microorganisms were metabolically engineered for Tyr production by modifying the shikimate and terminal AAA biosynthesis pathways. Juminaga et al. (2012) performed proteomic and metabolomic analyses to identify bottlenecks in Tyr production, which revealed that the activity level of the shikimate dehydrogenase YdiB and low expression of the dehydroquinate synthase AroB are bottlenecks in shikimate production. Based on their bottleneck analysis, they employed expression modules of the genes related to the shikimate pathway in E. coli and examined Tyr production, which showed that expressing shikimate pathway-related genes as operons on medium copy number plasmids resulted in more than 2 g L–1 Tyr production, which is 80% of the theoretical yield. Moreover, modification of Tyr transport system and the acetic acid biosynthesis pathway, expression of phosphoketolase (fpk) gene from Bifidobacterium adolescens with endogenous phosphotransacetylase (pta) gene and engineering of cofactor balance together with adaptive evolution to confer acid resistance in Tyr-producing strain of E. coli, in which the shikimate pathway and AAA biosynthesis pathway were modulated, enhanced Tyr production and Tyr production by the engineered strain reached 92.8 g L–1 in fed-batch cultivation using a jar fermenter (Ping et al. 2023).

In C. glutamicum, Kurpejović et al. (2023) engineered a Tyr-producing strain by modifying the shikimate pathway. In the modified strain, mutant aroG was expressed and the initiation codons for pheA and trpE were replaced with a minor initiation codon (TTG) to decrease their translation efficiency; the resulting strain produced 3.1 g L–1 Tyr. However, unlike Phe production, modification of the PTS to enhance the supply of PEP did not improve Tyr production.

The shikimate pathway is absent in humans and Phe and Trp are essential amino acids. Instead, Tyr is biosynthesized by the tetrahydrobiopterin (BH4)-dependent phenylalanine hydroxylase PheH, which catalyzes the formation of Tyr from Phe, molecular oxygen, and BH4 (Fitzpatrick 2023) (Fig. 3). Although some bacteria have homologs that use tetrahydromonapterin (MH4) as a cofactor (Pribat et al. 2010) (Fig. 3), these enzymes have never been used for fermentative production. The main challenge is the supply of tetrahydropterin, which is stoichiometrically consumed during the reaction in host cells. Satoh et al. (2012b) showed that this issue could be overcome by using the human BH4 regeneration system, which consists of pterin-4α-carbinolamine dehydratase (PCD) and dihydropteridine reductase (DHPR) (Fig. 3), when producing 3,4-dihydroxyphenylalanine (DOPA) with mouse tyrosine hydroxylase (TyrH), a homolog of PheH. Indeed, an engineered E. coli strain expressing these regeneration system-related genes and PheH from Gulbenkiania sp. SG4523, which was screened for high activity among eight enzymes from rat and bacteria, produced 4.63 g L–1 Tyr from 5 g L–1 of Phe in test tubes (Satoh et al. 2023). After further optimization and chromosomal engineering of E. coli, a strain was obtained that produced 5.19 g L–1 Tyr in test tube cultivation (Shen et al. 2024). Another group also reported production of Tyr (0.401 g L–1) from glucose by E. coli harboring PheH from Xanthomonas campestris and the MH4 regeneration system, including the PCD homolog PhhB from Pseudomonas aeruginosa and the dihydromonapterin reductase FolM from E. coli, in shake flasks (Huang et al. 2015), suggesting that this route is also available for Tyr production.

Fig. 3

Tyrosine formation via hydroxylation of phenylalanine

Tryptophan

Similar to Phe and Tyr production, Trp production in E. coli was engineered by improving metabolic flux of the shikimate pathway and Trp biosynthesis. Zhao et al. (2011) generated a Trp-producing strain of E. coli by enhancing flux of the shikimate and Trp biosynthesis pathways, avoiding Trp degradation, and blocking the competing Phe and Tyr biosynthesis pathways; the resulting strain produced about 13 g L–1 Trp in fed-batch cultivation. Gu et al. (2012) generated a Trp-producing strain of E. coli by expressing tktA encoding transketolase to improve the supply of E4P for enhancing shikimate pathway flux, and preventing Trp degradation; the resulting strain produced 1.3 g L–1 Trp in batch cultivation. Trp production in the strain was further enhanced by replacing the leader sequence and trpEDCBA operon promoter with a stronger promoter to about 1.7 and 10 g L–1 Trp in batch and fed-batch cultivation, respectively. Interestingly, Trp production by this strain was further improved by expressing polyhydroxybutyrate (PHB) biosynthesis genes from Cupriavidus necator (Gu et al. 2013). In the Trp production strain expressing heterologous PHB biosynthesis genes, expression of the trpDCBA genes were upregulated compared with that in the parental strain and this phenomenon may result in improved Trp production. However, the mechanism of upregulation of trpDCBA expression is not understood.

Liu et al. (2012) investigated the effect of deletion of aroP gene, which encodes AAA permease, and expression of yddG gene, which encodes an aromatic amino acid exporter, on Trp production in a Trp-producing E. coli strain. Wang et al. (2013) reported that deletion of pta and mtr encoding phosphotransacetylase and a high-affinity Trp transporter, respectively, combined with overexpression of yddG, reduced acetate production as a by-product and increased Trp production.

Li et al. (2020) reported the effects of optimizing the supply of precursor and cofactor on Trp production in E. coli. Trp biosynthesis requires L-glutamine, L-serine, and PRPP (Fig. 2). In this study, heterologous gene encoding glutamine synthetase and the endogenous icd and gdhA genes encoding isocitrate dehydrogenase and glutamate dehydrogenase, respectively, were expressed in an engineered Trp-producing strain to enhance the L-glutamine supply. Introduction of additional copies of the prs gene encoding phosphoribosylpyrophosphate synthase into the genome for improving PRPP supply and expression of mutant serA and thrA genes encoding 3‐phosphoglycerate dehydrogenase and bifunctional aspartokinase/homoserine dehydrogenase, respectively, with feedback resistance to L-serine for improving L-serine supply were additionally conducted; L-serine inhibits the activity of aspartokinase/homoserine dehydrogenase encoded by thrA and expression of mutant thrA gene is expected to maintain L-threonine biosynthesis even if L-serine supply is enhanced. To maintain redox balance, genes encoding transhydrogenases, which catalyze the interconversion of NADPH + NAD+ and NADP+  + NADH, were overexpressed in the engineered strain. Trp production in the final engineered strain reached 1.7 g L–1 in batch cultivation.

Guo et al. (2022b) metabolically engineered E. coli to improve and optimize the supply of precursors and modify the membrane transporters for Trp production. Trp biosynthesis was improved by removing the negative transcription factor TrpR, preventing the formation of competing by-products (Phe and Tyr), and overexpressing the trpEDCBA operon, in which trpE was replaced with a mutant encoding a feedback-resistant anthranilate synthase. To enhance the supply of PEP, the pathways for production of acetate, formate, lactate, and ethanol were disrupted. Moreover, to optimize the supply of substrates for DAHPS (i.e. PEP and E4P), recombinant strains of E. coli expressing ppsA, tktA, and mutant aroG under promoters with different strengths were constructed, and Trp production was evaluated. In addition, the expression of serA, serB, and serC was optimized by combinatorial screening of promoters to improve the L-serine supply, and the yggG gene, encoding the Trp exporter, was overexpressed. The resulting engineered E. coli cells produced 52.1 g L–1 Trp.

Fermentative production of various aromatic amino acid derivatives

AAAs are important starting compounds for the synthesis of various aromatic derivatives that are widely used in chemicals, food, polymers, and pharmaceuticals. Here, we briefly summarize fermentative production of these derivatives using E. coli and C. glutamicum (Fig. 4, Supplementary Figs. S2, S3, S4 and S5 and Supplementary Table S3).

Fig. 4

Synthetic pathways for various aromatic compounds from aromatic amino acids

Phenylalanine derivatives

In E. coli, phenylethylamine was produced by decarboxylation of Phe using aromatic amino acid decarboxylase from Pseudomonas putida (Koma et al.