Efficient cell factories for the production of N-methylated amino acids and for methanol-based amino acid production.

The growing world needs commodity amino acids such as L-glutamate and L-lysine for use as food and feed, and specialty amino acids for dedicated applications. To meet the supply a paradigm shift regarding their production is required. On the one hand, the use of sustainable and cheap raw materials is necessary to sustain low production cost and decrease detrimental effects of sugar-based feedstock on soil health and food security caused by competing uses of crops in the feed and food industries. On the other hand, the biotechnological methods to produce functionalized amino acids need to be developed further, and titres enhanced to become competitive with chemical synthesis methods. In the current review, we present successful strain mutagenesis and rational metabolic engineering examples leading to the construction of recombinant bacterial strains for the production of amino acids such as L-glutamate, L-lysine, L-threonine and their derivatives from methanol as sole carbon source. In addition, the fermentative routes for bioproduction of N-methylated amino acids are highlighted, with focus on three strategies: partial transfer of methylamine catabolism, S-adenosyl-L-methionine dependent alkylation and reductive methylamination of 2-oxoacids.

Introduction

Amino acid production by fermentation is a success story that started more than six decades ago (Lee and Wendisch, 2017). The market demand is steadily rising, even though African Swine Fever and the COVID‐19 pandemic slowed the growth. The very efficient L‐glutamate and L‐lysine production processes that are operated at a huge scale (million tons per year) benefit from the so‐called “economy of scale” (Wendisch, 2020). However, since the margins are very low, two trends have emerged: a shift from commodities towards specialty amino acids (Ajinomoto, 2020) and a shift from traditional substrates towards alternatives carbon sources (Wendisch et al., 2016).

Traditional amino acid fermentation is based on sugars and molasses and costs for these feedstocks contribute notably to the operational expenditures. Considering substrate availability, costs and competing uses in the food and feed industries, a flexible feedstock concept was realized for amino acid producer strains enabling access to sustainable alternatives, for example, lignocellulosic, aqua‐ and agricultural sidestreams (Wendisch et al., 2022).

Specialty amino acids find applications in the pharmaceutical industry (e.g. infusions, injections, intermediates in active substance syntheses or as active pharmaceutical ingredients). Among others, Escherichia coli and Corynebacterium glutamicum strains have been engineered to produce the blood pressure‐lowering L‐arginine (Park et al., 2014), the insulinotropic (2S, 3R, 4S)‐4‐hydroxyisoleucine (Smirnov et al., 2010; Zhang et al., 2018), 5‐hydroxy‐L‐tryptophan that can be used against depression and obesity (Mora‐Villalobos and Zeng, 2018), and the cyclic amino acid L‐pipecolic acid used as cell protectant and precursor of, for example, the immunosuppressant rapamycin and the antitumor agent swainsonine (Pérez‐García et al., 2016; 2017; 2019).

In this review, we address these trends by focusing on how methanol, a feedstock without competing food and feed uses, can be harnessed for production of L‐glutamate, L‐lysine, L‐threonine and their derivatives by bacteria. In recent years, there has been substantial progress in the development of methods for methanol synthesis particularly through not only CO2 hydrogenation but also isothermal methane conversion into methanol catalysed by copper‐containing zeolites or production of methanol from crude glycerol (Haider et al., 2015; Tomkins et al., 2016; Mbatha et al., 2021). In this review methanol‐based production of L‐serine, an intermediate of serine cycle for formaldehyde assimilation, will not be presented as it has been thoroughly summarized elsewhere (Eggeling, 2007). Moreover, we cover how access to N‐methylated amino acids, a particular class of specialty amino acids, has been gained by metabolic engineering.

Engineering cell factories for methanol‐based amino acid production

Production of L‐glutamate and its derivatives from methanol

Bacillus methanolicus MGA3 is a methanol‐utilizing bacterium known for its capacity to overproduce L‐glutamate up to 60 g l−1 in methanol‐controlled fed‐batch fermentations (Table 1) (Schendel et al., 2000; Heggeset et al., 2012), and in flasks under magnesium or methanol limitation (Schendel et al., 2000; Brautaset et al., 2003). There are several factors that may contribute to L‐glutamate accumulation in B. methanolicus: (1) overflow metabolism due to inactive tricarboxylic acid (TCA) cycle during methylotrophic growth, (2) production of L‐glutamate as compatible solute in response to osmotic stress.

Table 1

Production of amino acids from methanol by fermentation. Characteristic of production strains and titres of illustrative processes are listed. For abbreviations see either text or below.

Glutamate production
Host strain	Phenotype/ Genotype or Relevant enzymatic characteristic	Overexpressed gene(s)	Titre [g l−1]	Fermentation mode	References for established processes
B. methanolicus MGA3	Wild type	‐	0.8/ 59	Shake flask/ Fed‐batch	Heggeset et al. (2012); Krog et al. (2013)
B. methanolicus MGA3	Wild type	gltAB	1.1	Shake flask	Krog et al. (2013)
B. methanolicus MGA3	Wild type	gltA2	1.1	Shake flask	Krog et al. (2013)
M. glycogenes RV3	Phe+ (auxotrophy revertant)	‐	9.3/ 38.8	Test tubes/ Jar fermentor	Motoyama et al. (1993a)

AECR, S‐(2‐aminoethyl)‐L‐cysteine resistance; ThrR, L‐threonine resistance; Phe−, phenylalanine auxotrophy; Phe+, phenylalanine prototrophy; Ile−, isoleucine auxotrophy; fbr, feedback inhibition resistance.

GABA production

Lysine production

Cadaverine production

5AVA production

Threonine production

It is a widespread property of methylotrophs that they do not need a complete TCA to fulfil their energy requirements (Chistoserdova et al., 2009). While B. methanolicus is equipped with a full gene set for a functional TCA cycle and a functional glyoxylate shunt (Heggeset et al., 2012; Muller et al., 2015; Drejer et al., 2020), during growth on methanol the levels of some TCA enzymes were decreased and the carbon flux through the TCA cycle stopped after isocitrate with only small remaining flux needed to support the synthesis of biomass precursors (Müller et al., 2014; Delépine et al., 2020). For example, the activity of 2‐oxoglutarate dehydrogenase (encoded by odhAB) in crude extract of B. methanolicus was lower than in other Bacillus species that do not overproduce L‐glutamate, and its restoration through plasmid‐based overexpression of odhAB decreased L‐glutamate accumulation confirming importance of low carbon flux through TCA cycle for L‐glutamate synthesis (Carlsson and Hederstedt, 1986; Brautaset et al., 2003; Krog et al., 2013).

While being able to grow in seawater‐based media (Komives et al., 2005), B. methanolicus possesses a restricted ability to cope with sustained osmotic stress through synthesis of the moderately effective compatible solute L‐glutamate (Frank et al., 2021). The cellular L‐glutamate pool increased concomitantly with increasing external osmolarity, and a large portion of the newly synthesized L‐glutamate was excreted (Frank et al., 2021). The expression gltAB and gltA2 encoding two glutamate synthases was upregulated in response to high salinity along with that of gltC, which encodes a transcriptional activator of the glutamate synthase operon (Frank et al., 2021). Plasmid‐based overexpression gltAB and gltA2 boosted secretion of L‐glutamate by B. methanolicus, but not that of gltA2, yweB and glnA encoding glutamate synthase, glutamate dehydrogenase and glutamine synthetase (Table 1), respectively, indicating the major role of GltAB and GltA2 in L‐glutamate biosynthesis (Krog et al., 2013).

The mechanism of L‐glutamate secretion in B. methanolicus is still not elucidated. In C. glutamicum, a known microbial L‐glutamate producer, MscCG, a MscS‐like channel, is the major L‐glutamate export system (Nakamura et al., 2007; Nakayama et al., 2016; 2018; Wang et al., 2018). No homologues of an MscCG channel are encoded in the genome of B. methanolicus and the question if the MscS‐type channel‐encoding gene (locus BMMGA3_16700) present in the genome is involved in L‐glutamate secretion remains to be solved (Heggeset et al., 2012; Frank et al., 2021).

Bacillus methanolicus is not the only methylotrophic candidate to become platform strain for methanol‐based L‐glutamate production, the classical mutant of Methylobacillus glycogenes, a Gram‐negative obligate methylotroph, secreted 38.8 g l−1 of L‐glutamate in an 84‐h 5‐l methanol‐based fermentation supplemented with 10 g l−1 yeast extract (Table 1) (Libudzisz et al., 1983; Urakami and Komagata, 1986; Motoyama et al., 1993a).

Production of an L‐glutamate‐derivative, γ‐aminobutyric acid (GABA), which is a precursor of a 2‐pyrrolidone building block of biodegradable polyamine, nylon 4, was established in B. methanolicus through heterologous overexpression of glutamate decarboxylase gene (gad) derived from Sulfobacillus thermosulfidooxidans (Table 1) (Irla et al., 2017; Fukuda and Sasanuma, 2018). While the choice of this thermophilic donor circumvented the issue of thermolability of E. coli‐derived Gad, it did not alleviate the problem of its low activity at neutral pH (Irla et al., 2017; Fan et al., 2018). Bacterial Gads participate in acid stress response and are only active at low pH (Capitani et al., 2003). In two‐phase, methanol‐controlled fed batch fermentation 9 g l−1 of GABA were produced by engineered B. methanolicus, however, this approach did not support full L‐glutamate conversion, with almost 13 g l−1 of L‐glutamate left in fermentation broth (Irla et al., 2017). The purification of GABA from fermentation broth was achieved to 99.1% purity in a multistep process composed among others of flocculation, filtration, ultrafiltration, decolouration, ion exchange chromatography and lastly crystallization (Gao et al., 2013).

Methanol‐based production of L‐lysine, its derivatives, and L‐threonine

B. methanolicus wild type produced up to 0.4 g l−1 of L‐lysine in high cell density fed‐batch fermentations, and its mutant strain NOA2#13A52‐8A66 up to 65 g l−1 under the same conditions which is caused by several mutation in its genome (Hanson et al., 1996; Brautaset et al., 2010). The amino acid exchange in one of its three aspartokinases (AKs), catalysing the phosphorylation of L‐aspartate to L‐aspartate‐4‐phosphate, encoded by dapG abolished feedback inhibition by meso‐diaminopimelic acid (DAP) (Naerdal et al., 2011; 2017). Due to mutation in homoserine dehydrogenase (Hom) gene Hom activity decreased and metabolic flux was redirected from reduction of aspartate 4‐semialdehyde to homoserine catalysed by Hom towards synthesis of 4‐hydroxy‐tetrahydrodipicolinate by its synthase (DapA) in L‐lysine biosynthetic pathway (Naerdal et al., 2011; 2017). Finally, the mutation of the region upstream of lysA (Table 1) increased its expression in comparison to the wild‐type strain, presumably enhancing the decarboxylation of meso‐DAP to L‐lysine by LysA (Naerdal et al., 2011; 2017). Apart from mutations in L‐lysine biosynthesis pathway, NOA2#13A52‐8A66 strain exhibits changes in enzyme activities in central carbon metabolism. Decreased pyruvate dehydrogenase activity in comparison to the wild type caused by point mutation in pdhD gene can potentially decrease carbon flux towards TCA cycle and direct it towards oxaloacetate through activity of pyruvate decarboxylase instead (Brautaset et al., 2003; Naerdal et al., 2017).

Methylophilus methylotrophus AS1 is an obligate methylotroph with a ribulose monophosphate (RuMP) pathway for formaldehyde assimilation (Jenkins et al., 1987; Gunji et al., 2004). M. methylotrophus wild type naturally produces less than to 0.01 g l−1 of L‐lysine in test tube cultivations (Jenkins et al., 1987; Gunji et al., 2004), however, L‐lysine titre was increased to 0.08 g l−1 in strain G49 due to mutations in asd and dapA genes (Table 1) that caused partial resistance to feedback inhibition of aspartate semialdehyde dehydrogenase (Asd) catalysing formation of L‐aspartate‐semialdehyde in the reductive dephosphorylation of L‐aspartate‐4‐phosphate, and DapA further converting L‐aspartate‐semialdehyde to 4‐hydroxy‐tetrahydrodipicolinate (Gunji et al., 2004; Tsujimoto et al., 2006).

The mutant DHL122 derived from M. glycogenes ATCC 21276 produced 5.6 g l−1 of L‐threonine and 3.1 g l−1 of L‐Iysine in 72‐h, 5‐liter jar fermentation (Table 1) (Motoyama et al., 1993a). The AKDHL122 was completely insensitive to inhibition by L‐lysine in contrast to that of parental strain ATCC 21276, and it was activated with increasing concentrations of L‐threonine (Motoyama et al., 1993b, 2001). Moreover, the L‐lysine feedback inhibition of DapADHL122 was partially alleviated compared to wild type, possibly due to amino acid exchanges located in the region relevant for interaction with the allosteric effector, L‐lysine (Motoyama et al., 1993b, 2001).

One strategy to increase the L‐lysine titre in B. methanolicus is plasmid‐based overexpression of genes belonging to L‐lysine biosynthesis pathway (Table 1). Upon overexpression of AK‐encoding genes dapG, lysC and yclM in B. methanolicus L‐lysine titres either did not increase or increased 8‐ and 20‐fold in flask cultivation, and 2‐, 10‐ and 60‐fold in high cell density methanol fed‐batch fermentations, respectively, with a final titre in the fed‐batch fermentation for a yclM‐expressing strain of 11 g l−1 (Jakobsen et al., 2009; Naerdal et al., 2011). Interestingly, overexpression of NOA2#13A52‐8A66‐derived mutated dapG coding for a previously mentioned AK desensitized to feedback inhibition led to 17‐fold increase in L‐lysine titre compared to the control strain in flask cultivation (Naerdal et al., 2011).

While overexpression of asd and dapA had no positive effect on L‐lysine production in B. methanolicus wild type, the L‐lysine titre increased almost two‐ and fivefold in L‐lysine producing mutant B. methanolicus strain M168‐20 overexpressing asd and dapA in comparison to empty vector control (Naerdal et al., 2011). Similarly, the overexpression of the gene encoding feedback inhibition resistant DapADHL122 in DHL122 and its parent strain AL119 elevated the specific activity of DapA 20‐fold in both strains and L‐lysine production two‐ and threefold, respectively, with concomitant reduction of L‐threonine accumulation in test tube cultures. AL119 overexpressing dapA DHL122 produced 8 g l−1 of L‐lysine in a 5‐liter jar fermentor from methanol as a substrate (Motoyama et al., 2001).

Finally, through the overexpression of the gene encoding diaminopimelate decarboxylase (LysA), the last enzyme of the L‐lysine biosynthesis pathway, 20‐fold increase in L‐lysine accumulation was achieved in B. methanolicus wild type in comparison to empty vector control (Naerdal et al., 2011). In this respect, it has to be noted that expression of lysA was increased in NOA2#13A52‐8A66 due to previously mentioned point mutation in promoter region (Naerdal et al., 2011).

Co‐expression of several genes of L‐lysine biosynthesis had a cumulative effect on L‐lysine production in B. methanolicus (Table 1), when dapA was overexpressed together with yclM the L‐lysine titre increased 30‐fold, and addition of lysA to this pair resulted in an 83‐fold rise in comparison to the wild‐type strain (Naerdal et al., 2011). Heterologous expression of mutated versions of E. coli‐derived dapA and lysC encoding enzymes with reduced sensitivity to feedback inhibition and wild‐type version of dihydrodipicolinate reductase gene (dapB) in M. methylotrophus G49 improved L‐lysine titre to 0.4 g l−1 in test tube cultivation compared to 0.08 g l−1 for empty vector strain, with final titre of 1 g l−1 in jar fermentor (Tsujimoto et al., 2006).

Another strategy to increase L‐lysine titres is the overexpression of exporter encoding gene (lysE) (Table 1). Heterologous expression of mutated lysE gene derived from C. glutamicum in M. methylotrophus AS1 increased L‐lysine titre eightfold in the test tube in comparison to empty vector control strain (Gunji and Yasueda, 2006). The strain AS1 overexpressing lysE with dapA produced 1 g l−1 L‐lysine in shake flask cultivation and 11.3 g l−1 in 72 h jar fermentation (Gunji and Yasueda, 2006). A methionine auxotrophic M. methylotrophus mutant with deletion of 10‐methylenetetrahydrofolate reductase gene (metF) overexpressing lysE and dapA produced 1.2 g l−1 L‐lysine in shake flasks and more than 9.0 g l−1 in 1‐liter jar fermentors (Ishikawa et al., 2008a, 2008b). The metF deletion presumably positively affected L‐lysine biosynthesis due to homocysteine accumulation that inhibited activity of homoserine kinase (HK) encoded by thrB (Ishikawa et al., 2008a). Inhibition of HK activity decreased accumulation of intracellular L‐threonine, an AK inhibitor, subsequently averting feedback inhibition of AK by L‐threonine and increasing L‐lysine production (Gunji et al., 2004; Ishikawa et al., 2008a). B. methanolicus strain co‐expressing lysC with lysE produced almost sevenfold more L‐lysine in flask cultivation in comparison to strain expressing only lysC, while expression on native lysE had no effect on L‐lysine titres, leading to question whether the latter protein serves as L‐lysine exporter in B. methanolicus MGA3 (Naerdal et al., 2017).

Based on the presented results for three different methylotrophic bacterial species, several approaches seem to be particularly successful in strain engineering for L‐lysine production: (i) expression of the genes encoding for the enzymes relieved from feedback inhibition or introduction of genomic modifications to alleviate the feedback inhibition, (ii) overexpression of genes coding for L‐lysine export systems and (iii) deactivation of competing pathways. Furthermore, it seems that overexpression of genes of enzymes of the pathways that are not feedback regulated brings the least positive effects,

L‐Lysine can be converted to cadaverine, a monomer for bio‐polymer synthesis, in one reaction catalysed by lysine decarboxylase encoded by cadA (Table 1). Cadaverine, called also 1,5‐diaminopentane, has a plethora of applications in agriculture, medicine and industry (Wendisch et al., 2018b). It can be purified from fermentation broth by solvent extraction followed by a subsequent two‐step distillation process (Kind et al., 2014). Polymerization of bio‐based cadaverine with appropriate bio‐blocks, such as succinic acid or sebacic acid yields completely bio‐based polyamides PA‐5,4 and PA‐5,10 respectively (Kind et al., 2014; Yang et al., 2019). Overexpression of E. coli‐derived cadA in B. methanolicus wild type resulted in full conversion of L‐lysine to cadaverine and accumulation of the latter to a final titre of 6.5 g l−1 in a high cell density methanol‐controlled fed‐batch fermentation (Table 1), later improved to 10.2 g l−1 through use of stable Θ‐replication vector for expression of cadA (Naerdal et al., 2015; Irla et al., 2016).

5‐Aminovalerate (5AVA) is one of the intermediates of different L‐lysine degradation pathways. It is a precursor of valerolactam which can be used for the development of novel polyamides (PAs), and can be separated from its precursor, L‐lysine, through chromatography (Kim et al., 2020). Evaluation of five pathways for 5AVA biosynthesis in B. methanolicus resulted in the establishment of its production either using activity of lysine α‐oxidase (RaiP) (Table 1) or via a pathway with cadaverine as intermediate composed of CadA, putrescine transaminase (PatA), and 5‐aminopentanal dehydrogenase (PatD) (Table 1) (Brito et al., 2021). Initial titre of 0.02 g l−1 for the latter pathway in flask cultivation was increased fourfold through external supplementation with cadaverine (Brito et al., 2021). While B. methanolicus wild type exhibited low tolerance to 5AVA, mutant stains with increased 5AVA tolerance were selected by adaptive laboratory evolution (ALE) (Haupka et al., 2021).

Strains AL119 (derived from M. glycogenes ATCC 21276) and ATR80 (derived from ATCC 21371) produced 11.0 g l−1 and 8.5 g l−1 of L‐threonine (Table 1), respectively, in 5‐liter jar fementors at 72 h (Motoyama et al., 1993a). While the AKs of ATCC 21276 and ATCC 21371 were sensitive to L‐threonine and partially to L‐lysine, AKAL119 was completely insensitive to inhibition by L‐lysine and its activity was gradually enhanced with increasing concentrations of L‐threonine similarly to AKDHL122, whereas AKATR80 was completely insensitive to inhibition by L‐lysine, and partially inhibited by L‐threonine (Motoyama et al., 1993b).

The inhibition of the HKATR80 activity by L‐threonine was slightly reduced compared with that of parental wild‐type strain, and the DapA of both AL119 and ATR80 were somewhat desensitized to L‐lysine inhibition in comparison to parental strains (Motoyama et al., 1993b). The expression of the hom‐thrC genes, encoding homoserine dehydrogenase and threonine synthase (Table 1), respectively, in ATR80 and its L‐isoleucine auxotroph, A513, led to up to 12‐fold elevated activities of respective enzymes (Motoyama et al., 1994). The hom‐thrC expressing A513 strain produced about 40% more L‐threonine in test tube cultivation in comparison to empty vector control with final titre of 16.3 g l−1 after 72 h in 5‐liter jar fermentors (Motoyama et al., 1994).

Engineering cell factories for production of N‐methylated amino acids

Amino acids are functionalized, for example, by phosphorylation, acetylation, hydroxylation or halogenation. These modifications may affect either the free amino acid or an amino acid residue in a protein, and they typically alter bioactivity. For example, phosphorylation of free aspartic acid yields aspartyl‐phosphate, an activated intermediate of the lysine, methionine and threonine biosynthesis pathways (Wittmann and Becker, 2007). However, specific aspartyl residues in regulatory proteins are phosphorylated to control their activities, for example, the response regulator PhoR of C. glutamicum is activated upon phosphorylation of aspartic acid residue 59 (Kocan et al., 2006).

Alkylation and in particular methylation of the amino group of free or protein‐bound amino acids is abundant in nature. For example, N‐methylated amino acids are components of secondary metabolites such as the anti‐cancer compound actinomycin D (Mindt et al., 2020), or they have bioactivity themselves, such as the flavour compound of green tea, L‐theanine (Benninghaus et al., 2021). In peptide‐based drugs, alkylated amino acids provide stabilization against proteolytic attack and they increase lipophilicity for better membrane permeability and pharmacokinetics (Di Gioia et al., 2016), as shown, for example, for the anti‐prostate and anti‐breast cancer drug leuprolide (Haviv et al., 2002).

Due to the incomplete stereoselectivity, use of genotoxic alkylating agents and low yields of chemical synthesis of N‐methylated amino acids, enzymatic (Hyslop et al., 2019; Yao et al., 2021) and fermentative (Mindt et al., 2020) routes for their bioproduction have been developed. Three strategies for the fermentative production of N‐methylated amino acids (Table 2) will be discussed.

Table 2

Production of N‐methylated amino acids by fermentation. Key enzymatic reactions of reductive methylamination, SAM‐dependent methylation and partial transfer of methylamine catabolism are depicted. Titres, yields and productivities of illustrative processes are listed. For abbreviations see text.

Partial transfer of methylamine catabolism
Product	Titre [g l−1]	Yield [g g−1]	Productivity [g l−1 h−1]	Fermentation mode	References for established processes
N‐methyl‐l‐glutamate	17.9	0.11	0.13	Fed‐batch	Mindt et al. (2018b)
l‐Theanine
by E. coli	70.6	0.42	2.72	Fed‐batch	Fan et al. (2020)
by C. glutamicum	42.0	0.20	0.88	Fed‐batch	Ma et al. (2020)
by P. putida	21.0	0.03	0.38	Fed‐batch	Benninghaus et al. (2021)

SAM‐dependent methylation

Reductive methylamination

Fermentative production by partial transfer of methylamine catabolism

N‐Methylglutamate is an intermediate in monomethylamine catabolism of some methylotrophs such as Methylobacterium extorquens. Assimilation of a C1 compound, methylamine as the sole carbon and nitrogen source by M. extorquens involves three specific enzymes (Ochsner et al., 2015). γ‐Glutamylmethylamide synthetase (GMAS) first methylamidates glutamate at its C5 position before N‐methylglutamate synthase (NMGS) transfers the N‐methyl group of γ‐glutamylmethylamide to 2‐oxoglutarate yielding glutamate and N‐methylglutamate (Table 2). Next, N‐methylglutamate dehydrogenase (NMGDH) catalysed oxidative demethylation of N‐methylglutamate to glutamate and formaldehyde, the latter being fixed in the serine cycle. Upon expression of the M. extorquens‐derived GMAS and NMGS genes in the non‐methylotrophic P. putida, N‐methylglutamate was produced to about 18 g l−1 in 2‐liter bioreactor fed‐batch cultivation with a yield of 0.11 g g−1 glycerol and a volumetric productivity of about 0.13 g l−1 h−1, if methylamine was added to the growth medium (Mindt et al., 2018b).

When the C2 compound monoethylamine was provided to one of the non‐methylotrophic hosts E. coli, C. glutamicum and P. putida that expressed a GMAS‐encoding gene, efficient production of L‐theanine resulted with differences in final titres of L‐theanine resulting from the supply of precursor, L‐glutamate, with extensive genetic work performed for E. coli, and only limited changes, deletion of L‐glutamate exported or overexpression of gdh, for C. glutamicum and P. putida, respectively (Fan et al., 2020; Ma et al., 2020; Benninghaus et al., 2021). The ethylamide L‐theanine is the major free amino acid and bioactive component of green tea and it is known for its favourable physiological and pharmacological effects (Vuong et al., 2011). The L‐theanine‐producing E. coli strain, for example, expressed the GMAS‐encoding gene from Paracoccus aminovorans. It had improved glutamate availability due to overexpression of the endogenous citrate synthase gene, the glutamate dehydrogenase and pyruvate carboxylase genes from C. glutamicum, the phosphoenolpyruvate carboxykinase gene from Mannheimia succiniciproducens, and a deletion of the succinyl‐CoA synthetase genes (Fan et al., 2020). The resulting E. coli strain produced about 71 g l−1 L‐theanine in a 5‐liter bioreactor fed‐batch cultivation with a yield of 0.42 g g−1 glucose and a volumetric productivity of about 2.7 g l−1 h−1(Fan et al., 2020).

Notably, the addition of the C2 compound ethylamine could be circumvented by metabolic engineering of its biosynthesis in E. coli (Hagihara et al., 2021). To this end, acetyl‐CoA, a central carbon metabolite, was reduced to acetaldehyde by endogenous acetaldehyde dehydrogenase EutE and the ω‐transaminase SpuC‐II from P. putida transferred the amino group of the co‐substrate L‐alanine to acetaldehyde yielding pyruvate and ethylamine. The resulting E. coli strain produced about 16 g l−1 L‐theanine without the requirement to add ethylamine to the growth medium (Hagihara et al., 2021).

Fermentative production via S‐adenosyl‐L‐methionine‐dependent alkylation

S‐Adenosyl‐L‐methionine (SAM) is a universal cofactor of cellular metabolism. SAM‐dependent methyltransferases that catalyse regioselective methylation reactions and show a defined substrate spectrum have found wide applications in enzyme catalysis (Struck et al., 2012; Zhang and Zheng, 2015). For amino acids, SAM‐dependent methylation of C‐, N‐ and O‐atoms has been described, for example, in the synthesis of N‐methylarginine, an inhibitor of nitric oxide synthase (Stefanovic‐Racic et al., 1994), 3‐methyl‐arginine, a suppressor of bacterial blight of soybean (Braun et al., 2008), and the grape flavours O‐methylanthranilate and N‐methyl‐O‐methylanthranilate (Lee et al., 2019). Anthranilate can be methylated to N‐methyl‐O‐methylanthranilate by sequential reactions of an N‐methyltransferase and an O‐methyltransferase (Table 2). The intermediate may be either N‐methylanthranilate or O‐methylanthranilate (Table 2). While only little N‐methyl‐O‐methylanthranilate was produced by a recombinant E. coli strain (Lee et al., 2019), E. coli and C. glutamicum have recently been engineered for efficient production of O‐methylanthranilate (Luo et al., 2019). About 5.7 g l−1 O‐methylanthranilate was produced by a recombinant C. glutamicum strain with a yield of 0.02 g g−1 glucose and a volumetric productivity of 0.052 g l−1 h−1, and about 4.5 g L‐1 by an E. coli strain with a yield and productivity of 0.02 g g−1 glucose and of 0.052 g L−1 h−1 respectively (Luo et al., 2019). To this end, the gene coding for anthranilic acid methyltransferase 1 (AAMT1) from the plant Zea mays was expressed in strains that were engineered for overproduction of the immediate precursor anthranilate, an intermediate of L‐tryptophan biosynthesis, and for improved regeneration of SAM. Product toxicity was avoided by using a tributyrin overlay as second AAMT1 phase that captured the product O‐methylanthranilate leading to its in situ extraction which can facilitate downstream purification (Luo et al., 2019). In both strains, accumulation of the precursor, anthranilate, was observed suggesting that the methylation reaction was limiting for formation of O‐methylanthranilate, either due to arability of co‐substrate SAM or activity of AAMT1. The other monomethylated anthranilate, N‐methylanthranilate, is a precursor in plant secondary metabolism leading to acridone alkaloids and avenacin, which have anticancer, cytotoxic and antimicrobial properties relevant for pharmaceutical and therapeutic applications purposes (Rohde et al., 2007). Expression of the gene for N‐methytransferase (ANMT) from the plant herb‐of‐grace Ruta graveolens enabled a genome‐reduced C. glutamicum chassis strain engineered for overproduction of anthranilate as precursor and for improved regeneration of SAM to produce 0.5 g l−1 of N‐methylanthranilate with a yield of about 0.005 g g−1 glucose and a volumetric productivity of 0.01 g l−1 h−1 (Walter et al., 2020). Similarly, in this process an excess of precursor, anthranilate, accumulated suggesting that methylation was a limiting step of the N‐methylanthranilate production.

Fermentative production via reductive methylamination of 2‐oxoacids

2‐Oxo acids are converted to the respective amino acids by transamination or reductive amination using ammonium as substrate. The enzyme DpkA from P. putida has been described to catalyse reductive alkylamination of 2‐oxo acids with methylamine or ethylamine instead of ammonium as substrate. In nature, DpkA reduces the imine bond of piperideine‐2‐carboxylate to yield L‐pipecolic acid in D‐lysine catabolism (Muramatsu et al., 2005a; 2005b). 2‐Oxo acids and methylamine spontaneously form imines that are reduced by DpkA to yield the respective N‐methylated amino acids, for example, N‐methyl‐L‐alanine from pyruvate or N‐methyl‐L‐leucine from 2‐oxoisocaproate (Mihara et al., 2005).

Expression of dpkA in C. glutamicum strains engineered to overproduce glyoxylate, pyruvate or phenylpyruvate as 2‐oxoacid precursor (Wieschalka et al., 2012; 2013; Zahoor et al., 2014) enabled fermentative production of about 37 g l−1  N‐methyl‐L‐alanine (Mindt et al., 2018a), about 8.7 g l−1 sarcosine (Mindt et al., 2019b), about 1.6 g l−1  N‐ethylglycine (Mindt et al., 2019a) and about 0.7 g l−1  N‐methyl‐L‐phenylalanine (Kerbs et al., 2021) upon addition of (m)ethylamine to the growth medium (Table 2). Production of N‐methyl‐L‐phenylalanine did not only require systems metabolic engineering for provision of phenylpyruvate as substrate but also engineering of the enzyme DpkA. Native DpkA from P. putida prefers pyruvate over phenylpyruvate, however, upon introduction of the amino acid exchanges P262A and M141L in the substrate binding pocket of DpkA comparable catalytic efficiencies with phenylpyruvate and pyruvate resulted (Kerbs et al., 2021). When the xylose isomerase gene xylA from Xanthomonas campestris and the endogenous xylulokinase gene xylB were expressed, sustainable production of N‐methyl‐L‐phenylalanine from the lignocellulosic pentose sugar xylose to a titre of 0.6 g l−1 with a yield of 0.05 g g−1 xylose was achieved (Kerbs et al., 2021). Further extension of this concept is possible, but has not been realized experimentally.

Concluding remarks

In this review, we have presented how the C1 metabolism can be harnessed for the production of amino acids or their methylated derivatives, either by use of methylotrophic cell factories or activity of specific enzymes involved in methylotrophy. Regarding biosynthesis of methylated amino acids, we focused on three strategies relying on the activity of different enzymes or enzymatic cascades (i) GMAS and NMGS derived from methylotrophic M. extorquens where they function as part of methylamine assimilation pathway, (ii) ANMT and AAMT derived from plants or (iii) DpkA derived from P. putida where it functions in D‐lysine degradation. Here, supply of precursors and co‐factors, as well as the activity of the biosynthetic enzymes seem to play major roles in the process efficiency, becoming major strain engineering targets. As an outlook, we foresee that the development of methylated amino acids may respond to market needs to a certain extent. N‐Methylated amino acids do not only play a role as free bioactives or in peptide drugs, but they may also be co‐translationally incorporated into proteins at specific locations by codon engineering (Hoesl and Budisa, 2012). For example, translational amber stop codons have been re‐coded using an evolved pyrrolysyl‐tRNA synthetase‐pylT pair (Blight et al., 2004) to incorporate meta‐nitrophenylacetate‐photocaged Nε‐L‐lysine residues. Upon photolysis in vivo, the labelled proteins were converted to proteins with monomethylated lysine residues (Wang et al., 2010).

The strategies used for methanol‐based production of amino acids by natural methylotrophs generally include use of classical mutagenesis and selection of best‐performing strains, or expression of genes encoding feedback inhibition alleviated enzymes or amino acid exporters. In case of non‐natural products, such as the diamine cadaverine, or the non‐proteinogenic amino acids 5AVA and GABA, expression of heterologous pathways was necessary. Considering that all these compounds are bulk chemicals, with L‐glutamate and L‐lysine serving as food and feed additives, and cadaverine, 5AVA and GABA as building blocks of polyamines of platform chemicals, it is worthwhile to investigate their methanol‐based productions. Methanol is considered a promising raw material for bioprocesses due to its stable prices, easiness of transport and storage and the fact that it can be produced sustainably from non‐food sources.

We foresee that the development of new and more efficient processes for production of amino acids from methanol will be driven by a technology push. Specifically, we anticipate that the use of various CRISPR technologies will revolutionize producer strain development (Schultenkamper et al., 2019; 2020). Adaptive laboratory evolution (Hu et al., 2016; Sandberg et al., 2019; Hennig et al., 2020; Wang et al., 2020) and enforcement of production by coupling it to growth (Haupka et al., 2020) will allow for efficient selection procedures of superior strains (Prell et al., 2021). Moreover, development of novel genetic tools will facilitate strain engineering of methylotrophic production hosts (Irla et al., 2016; Irla et al., 2021). In addition, synthetic consortia of different microorganisms may be developed to divide labour, for example, between conversion of a substrate such as methanol to an intermediate by one microorganism and product formation from the intermediate by another (Sgobba and Wendisch, 2020). In this respect it has to be noted that methanol initially is oxidized to formaldehyde and there are other sources of formaldehyde that may be used as substrates for fermentation. However, formaldehyde has to be liberated from these, for example, by degradation of formaldehyde oligomers such as trioxymethylene and hexamethylenetetramine (Kaszycki and Koloczek, 2002) or by demethylation of vanillin and other methylated aromatic compounds that are present in lignin (Wendisch et al., 2018a; Costa et al., 2021). Albeit attractive, this is clearly unchartered terrain and it is questionable whether these compounds will be available at reasonable cost and quantities.

Taken together, production of amino acids from methanol and production of N‐methylated amino acids has seen substantial success. It is anticipated that future developments driven by technology push and/or market demand will shape this exiting field of microbial biotechnology.

Funding information

VFW is a self‐funded partner in the ERA CoBioTech project MCM4SB (327216).

Conflict of interest

The authors declare no competing financial interest.