A High-Efficiency Artificial Synthetic Pathway for 5-Aminovalerate Production From Biobased L-Lysine in Escherichia coli

Bioproduction of 5-aminovalerate (5AVA) from renewable feedstock can support a sustainable biorefinery process to produce bioplastics, such as nylon 5 and nylon 56. In order to achieve the biobased production of 5AVA, a 2-keto-6-aminocaproate-mediated synthetic pathway was established. Combination of L-Lysine α-oxidase from Scomber japonicus, α-ketoacid decarboxylase from Lactococcus lactis and aldehyde dehydrogenase from Escherichia coli could achieve the biosynthesis of 5AVA from biobased L-Lysine in E. coli. The H2O2 produced by L-Lysine α-oxidase was decomposed by the expression of catalase KatE. Finally, 52.24 g/L of 5AVA were obtained through fed-batch biotransformation. Moreover, homology modeling, molecular docking and molecular dynamic simulation analyses were used to identify mutation sites and propose a possible trait-improvement strategy: the expanded catalytic channel of mutant and more hydrogen bonds formed might be beneficial for the substrates stretch. In summary, we have developed a promising artificial pathway for efficient 5AVA synthesis.


INTRODUCTION
Increasing concerns over global water pollution, climate change, public health, and petroleum shortages have attracted considerable attention to sustainable development as promising green alternatives to traditional petrochemical-derived chemicals renewable feedstock (Tsuge et al., 2016). Recently a variety of valuable chemicals such as 6-aminocaproate (Cheng et al., 2019), fructose (Yang et al., 2016), mandelic acid (Youn et al., 2020), vitamin B 12 (Fang et al., 2018), naringenin (Gao et al., 2020b), p-coumaric acid (Gao et al., 2020a), breviscapine (Liu et al., 2018), 4-hydroxybenzoic acid (Klenk et al., 2020), curcuminoids (Rodrigues et al., 2020) and hydroxytyrosol (Zeng et al., 2020) have been produced in microorganisms. As a kind of green alternative to petrochemical products, microbial bioplastics are composed of monomers containing appropriate functional groups, which have become the focus of metabolic engineering research. These compounds include amino acids such as methionine (Kromer et al., 2006) and leucine (Zhang et al., 2008), organic acids such as adipic acid (Zhao et al., 2018a) and glutarate (Zhao et al., 2018b), diamines such as 1,3-diaminopropane (Chae et al., 2015) and diaminopentane (Kind et al., 2010;Rui et al., 2020), as well as diols like 1,3propanediol (Nakamura and Whited, 2003) and 1,2-propanediol (Niimi et al., 2011). It is worth mentioning that two straight-chain amino acids-5-aminovalerate (5AVA) and 4-aminobutyrateare promising platform compounds for the synthesis of polyimides, serving as raw materials for disposable goods, clothes and automobile parts like nylon 5 (Adkins et al., 2013) and nylon 4 (Park et al., 2013a) because of its high temperature and organic solvent resistance.  Due to the high demand in the animal feed industry, the production of L-Lysine (L-lys) is saturated today and may even be in oversupply (Vassilev et al., 2018). As one of the most important bulk chemicals, 5AVA has become the precursor for the synthesis of δ-valerolactam (Zhang et al., 2017), glutarate (Rohles et al., 2016;Hong et al., 2018), nylon 5 (Adkins et al., 2013), 5hydroxyvalerate  and 1,5-pentanediol (Park et al., 2014). 5AVA is currently produced from petroleum feedstocks with aerobic oxidation of piperidine catalyzed by ceria-supported nanogold (Dairo et al., 2016). However, this chemical synthesis method not only requires higher temperature, but results in greater pollution (Dairo et al., 2016), so it is necessary to discover alternative approaches to produce 5AVA. Recently, with the rapid development of biotechnology, the synthesis of 5AVA by means of metabolic engineering and synthetic biology has attracted more and more attention (Hong et al., 2018).
In this study, 5AVA was synthesized using 2-keto-6aminocaproate as intermediate, which is related to the involvement of three key enzymes-RaiP, KivD, and aldehyde dehydrogenase (PadA)-as seen in Figure 1D. Compared with the wild type, the two mutants of KivD in residues F381 and M461 showed higher substrate recognition and catalytic efficiency. Moreover, the overexpression of KatE and LysP, contributes to the removal of H 2 O 2 and the transport of L-lys, thereby increasing the production of 5AVA, respectively. As can be expected, this artificial pathway has a potential prospect in industrial application, which enhances the value of L-lys and produces 5AVA efficiently in engineered E. coli.

Strains and Plasmids
The strains and plasmids involved in this work are listed in Table 2. The nucleotide sequences of genes raiP from  Lactococcus lactis was the α-ketoacid decarboxylase used in this study. All α-ketoacid decarboxylase genes were downloaded from NCBI by blastp against the nr database. The genes were from the species, L. lactis, Enterococcus crotali, Enterococcus termitis, Enterococcus Silesiacus, Enterococcus Caccae, Enterococcus rotai, Enterococcus haemoperoxidus, Streptococcus didelphis, Listeria thailandensis, Listeria kieliensis, and Brochothrix thermosphacta. The maximum-likelihood tree was constructed by MEGA (Li et al., 2019a,b). (B) An artificial pathway confirmed for the biosynthesis of 5AVA. 6.5 g/L of L-Lysine HCl was as substrate. All experiments were performed a minimum of three independent sets. All error bars represent standard deviations with n ≥ 3 independent reactions.
Frontiers in Bioengineering and Biotechnology | www.frontiersin.org S. japonicus, kivD from L. lactis and padA from E. coli are available in the GenBank database with the accession numbers of MG423617 (Cheng et al., 2018a), AIS03677.1 (McCulloch et al., 2014) and NP_415903.4 (Riley et al., 2006), respectively. In order to establish the synthetic pathway, the raiP, padA, and kivD genes were inserted into pET21a, and then the plasmid pET21a-raiP-padA-kivD was generated, which was also named as pETaRPK. Primers for saturation mutation of KivD are listed in Supplementary Table 1. kivD was replaced by kivD # (kivD with F381A/V461A mutations) to form the engineered pET21a-raiP-padA-kivD # , also named as pETaRPK # . The lysine permease gene lysP from E. coli (GenBank accession No. WP_000253273.1) was amplified from plasmid pLMAIP-04 (Cheng et al., 2018a), and the catalase gene katE (GenBank accession No. AAT48137.1) from E. coli MG1655. In order to remove H 2 O 2 , accelerate transportation of L-lys and reduce energy consumption, the katE, and lysP genes were firstly constructed in another single operon with the transcriptional order of katE-lysP, and then the engineered pZA22-katE-lysP was produced, also named as pZAKL. In addition, E. coli BL21 (DE3) with knocked out cadA was transformed with the plasmid pCJ01, pETaRPK, pETaRPK # , pETakatE, or pETaKL to obtain the strains CJ02, CJ06, CJ07, CJ08, or CJ09, respectively.

Cultivation Medium and Conditions
The E. coli strains harboring the corresponding plasmids were streaked onto Luria-Bertani (LB) agar plates with appropriate antibiotics at 37 • C for overnight. Engineering strains used for shake flask fermentation were cultured in the medium containing 5 g/L yeast extract, 10 g/L tryptone, 15 g/L glucose, 0.1 g/L FeCl 3 , 2.1 g/L citric acid·H 2 O, 2.5 g/L (NH 4 ) 2 SO 4 , 0.5 g/L K 2 PO 4 ·3H 2 O, 1.0 mM MgSO 4 , 3 g/L KH 2 PO 4 , and 0.5 mM thiamine diphosphate (ThDP) with appropriate antibiotics. After the OD 600 of the strains reached 0.5, 0.5 mM of isopropyl β-Dthiogalactoside (IPTG) and 6.5 g/L of L-lys HCl were added. Fed-batch biotransformation of engineering strains were conducted in a 5.0 L fermenter. The composition of the medium was described in our previous report as follows: glucose, 55 g/L; MgSO 4 ·7H 2 O, 1.6 g/L; FeSO 4 ·7H 2 O, 0.00756 g/L; (NH 4 ) 2 SO 4 , 1.6 g/L; citric acid, 2 g/L; K 2 HPO 4 ·3H 2 O, 7.5 g/L; Na 2 SO 4 , 0.02 g/L; ZnSO 4 , 0.0064 g/L; Cu 2 SO 4 ·5H 2 O, 0.0006 g/L; CoCl 2 ·6H 2 O, 0.004 g/L (Cheng et al., 2018a). The pH was controlled at 6.7-6.9 by the automatic addition of NH 3 ·H 2 O, and the temperature was set at 30 • C. Antifoam 289 was gradually added to prevent the formation of foam during biotransformation. The initial concentration of L-lys HCl was 40 g/L. The concentration of glucose and L-lys were maintained around 15 and 20 g/L during the whole fermentation process, respectively.

Protein Expression and Purification
The media for protein expression was supplemented by 0.5 mM ThDP in LB at 37 • C. At an OD 600 of 0.5, 0.5 mM of IPTG was added and then cultured at 20 • C for 16 h, cells were washed with potassium phosphate buffer (KPB, 50 mM, pH 8.0) and disrupted by sonication in an ice bath of 50 mM KPB. The enzymes were purified with AKTA Purifier 10 using a Ni-NTA column (Cheng et al., 2019). The concentration of protein was measured by SpectraMax M2 e at 280 nm. The detections of 5AVA and L-lys were reported in our previous work (Cheng et al., 2018b).

Enzyme Assay
The oxidation activity of RaiP was measured according to the concentration of hydrogen peroxide (Cheng et al., 2018b). The decarboxylation activity of KivD and KivD mutations (KivD * ) were determined at 30 • C, using a coupled enzymatic assay . The reaction mixture contained 1.0 mM NAD + , 1.1 µM PadA, 1.1 µM RaiP, 0.85 µM KivD, or KivD * and different concentrations of L-lys in assay buffer (50 mM KPB, pH 8.0, 1 mM MgSO 4 , 1.0 mM TCEP, 0.5 mM ThDP). The reactions began with the addition of the substrate L-lys, and the formation of NADH was monitored at 340 nm with the extinction coefficient of 6.22 mM −1 cm −1 .

Homology Modeling, Substrate Docking, and Molecular Dynamic Simulation
The theoretical structure of native KivD and mutant KivD # (KivD with F381A/V461A mutations) (PDB: 2VBF), both were generated by SWISS-MODEL online Server 1 . The 3D structural comparison between KivD and KivD # was revealed using PyMOL 2.2. The ligand 2K6AC was docked into the pocket of KivD or KivD # using AutoDock 4.2.6 package, where the lowest energy conformation in the largest cluster was considered to be the approximately natural complex model (Xie et al., 2019;Tahara et al., 2020). Molecular dynamic (MD) simulation was used to simulate the relationship between structure and function of biomacromolecules in solution in this study (Wu et al., 2020). Two comparative MD simulations at 300 K were executed for KivD and KivD-2K6AC systems with AMBER 18 package (Zuo et al., 2017;Wu et al., 2020).

Construction of an Artificial Synthetic
Route for the Biosynthesis of 5AVA in E. coli Figure 1D showed a heterogeneous artificial route for the bioconversion of L-lys to 5AVA. The designed artificial biosynthetic pathway of 5AVA consists of three steps: (1) deamination of L-lys to form intermediate 2K6AC via RaiP; (2) decarboxylation of 2K6AC to produce 5-aminopentanal via KivD; (3) oxidation of 5-aminopentanal to 5AVA via PadA. Firstly, a plasmid pETaRPK was constructed and introduced into E. coli ML03 to obtain the strain CJ05, with the co-expression of RaiP, KivD, and PadA under a T7 promoter. To reduce the degradation of L-lys to cadaverine, the lysine decarboxylase gene cadA was knocked out to obtain the strain CJ06. The maximumlikelihood tree was displayed in Figure 2A. Notably, 5AVA could be produced in strains CJ01, CJ02, CJ05, and CJ06. As shown in Figure 2B, the control strain CJ00 only produced 0.06 g/L 5AVA from 6.5 g/L L-lys HCl with the consumption of 0.01 g/g L-lys. For engineered strain CJ01, a titer of 0.23 g/L 5AVA was acquired. Moreover, the strain CJ05 produced 1.66 g/L of 5AVA by this artificial pathway (see Figure 1D), with a yield increase of 774% compared to the single gene pathway (see Figure 1C). These results demonstrate the feasibility of this proposed artificial 5AVA pathway.

Molecular Docking and MD Simulation of KivD and KivD #
In order to explore the mechanism of the 5AVA increase in mutants, molecular docking and MD simulation were discussed (Xiang et al., 2019). The structures of KivD and KivD # both are mainly composed of 23 α-helices and 17 β-strands, containing a large activity pocket. Compared with that of KivD, the structure of KivD # remains almost unchanged. Nevertheless based on homology modeling analysis, the catalytic channel of mutant KivD # was enlarged. According to bioinformatics and crystal structure information (PDB: 2VBF) (Berthold et al., 2007), residues F381 and V461 are the two key residues for KivD catalysis (see Figure 3). Modeling and molecular docking of KivD with ligand 2K6AC further highlight the residues involved in substrate recognition. As shown in Figure 3, the substrate docking results indicated that the distances of ligand 2K6AC with F381A, V461A active sites both became farther. The docking results of KivD and 2K6AC showed that 2K6AC formed eight hydrogen bonds with the side chain Q377, T379, N456, T460, and V461. 2K6AC formed nine hydrogen bonds with the side chain D429, N456, G458, T460, A461, and E462 of KivD # (Figure 4). At the same time, the surface hydrophobicity of the catalytic pocket in mutated protein KivD # has also changed (Figure 4). We speculated that the increase in catalytic activity of KivD # may be due to the expansion of catalytic channel and the formation of more hydrogen bonds, the expansion that is likely to result in a change in the conformation of the small molecule 2K6AC which was beneficial to stretch. Through the MD simulations, the results of the root mean squared deviation (RMSD) showed that the RMSD of the KivD system and the complex system KivD-2K6AC were basically maintained at 1.639 ± 0.240 Å and 1.738 ± 0.152 Å (see Figure 5B), which indicated that the MD simulation process was reliable (Zuo et al., 2017). As seen in Figure 5A, there are four fragments of the KivD with lower root mean squared fluctuation (RMSF) values, that is G58-L69, T212-N223, T379-F388, and TABLE 3 | Kinetic parameters of α-ketoacid decarboxylase KivD mutants (KivD*) on 2-keto-6-aminocaproate (2K6AC). D457-H466. These four fragments are located near ThDP, which may be related to the activity of KivD (Zuo et al., 2017;Liu et al., 2019). KivD mutations (F381A/V461, F381L/V461, F381/V461A, F381/V461L, and F381A/V461A) displayed enhanced activities in Table 3. The KivD F381A/V461A (KivD # ) showed the greatest activity shown in Table 3. KivD # displays a K m value of 2.52 mM, a K cat value of 562.16 s −1 and a K cat /K m value of 223.08 mM −1 s −1 with 2K6AC used as the substrate shown in Table 3.

Overexpression of Catalase KcatE and Lysine Permease LysP Favoring the Increase of 5AVA Production
There are four strategies used in this study to increase the production of 5AVA. Firstly, lysine decarboxylase gene cadA was knocked out and L-lys HCl was selected as the industrial substrate for enhancing the utilization of L-lys (Cheng et al., 2018a(Cheng et al., ,b, 2020. Thirdly, H 2 O 2 could inhibit cell growth, thus affecting the production of goal production (Niu et al., 2014). In Liu's experiments, through the expression of catalase, the content of H 2 O 2 was significantly reduced, and the output of α-ketoglutarate was greatly increased . In this study, the co-expression of katE, raiP, kivD # , and padA in strain CJ08 yielded 1.88 g/L of 5AVA without addition of catalase, there was no significant difference compared to strain CJ07 (Table 4). In fact, the H 2 O 2 generated by RaiP in this work was instantly eliminated by KatE. The data in rows 5 and 7 of Table 4 showed that the overexpression of katE did not significantly increase the OD 600 and the production of 5AVA during shake flask fermentation. On the contrary, it decreased the OD 600 , possibly because the increase in gene expression caused an increase in cell burden (Camara et al., 2017). However, in the fermentation tank, H 2 O 2 could significantly inhibit cell growth, resulting in limited production of 5AVA (Cheng et al., 2018b. In addition, a lysine transporter gene lysP was overexpressed and inserted into the plasmid pZAkatE to form a new plasmid pZAKL. As shown in Table 4, strain CJ09 produced 1.93 g/L of 5AVA. Figure 6 showed the results of the fed-batch biotransformation in E. coli strain CJ09. Recombinant E. coli strain CJ09 grew quickly throughout the biotransformation, reaching the highest cell concentration of an OD 600 of 142 in 18 h. After the addition of L-lys HCl, 5AVA was accumulated to 48.3 g/L between 18 and 36 h. With the fermentation time increasing to 48 h, 52.24 g/L of 5AVA was successfully acquired. The productivity and yield of 5AVA were 1.09 g/L/h and 0.65 g/g L-lys, respectively. The control strain CJ02 just produced 9.16 g/L 5AVA with a yield of 0.11 g/g L-lys. Interestingly, the expression of KatE in strain CJ08 had no effect on the production of 5AVA in shake flask ( Table 4), but it could significantly improve the production of 5AVA to 45.92 g/L in fermentation tank compared to strain CJ07 with a titer of 16.48 g/L. This is because H 2 O 2 can significantly inhibit the growth of strain CJ07, resulting in OD 600 of only 40. Data are presented as means ± STDV calculated from three replicate biotransformation experiments. Statistics were performed by the two-tailed student t-test. * P < 0.05; ns, not significant. a Statistic analysis of the 5AVA production were performed with every two separated lines. b The yield of 5AVA was calculated based on L-lys consumption. 6.5 g/L L-lys HCl, 15 g/L glucose, 0.5 mM IPTG, 1.0 mM MgSO 4 and 0.5 mM ThDP were added. The above results advocated that the synthetic route developed in this work can effectively produce 5AVA. Compared with a previous whole cell transformation, the titer of 5AVA based on this synthesis pathway increased by about 79.4% from 29.12 to 52.24 g/L as seen in Table 1; the inhibition of cell growth and enzyme activity by H 2 O 2 both resulted in the lower yields of 5AVA (Cheng et al., 2018b). Compared with another new synthesis pathway for the fermentative production of 5AVA, in which the titer was only 5.1 g/L (seen in Table 1; Jorge et al., 2017), and the titer was greatly increased in this study. Compared with another whole-cell catalysis work, this synthetic pathway increased the titer of 5AVA by about 3.20% from 50.62 to 52.24 g/L . Importantly, the industrial production of 5AVA without the addition of ethanol and H 2 O 2 was more safe and economical in this study. In terms of reaction mechanism, the new 5AVA synthesis strategy proposed in this work mainly includes three steps: (1) the accumulation of intermediate 6A2KCA by RaiP; (2) the decarboxylation of 6A2KCA to 5-aminopentanal by KivD; (3) the oxidization of 5-aminopentanal to 5AVA by PadA.

CONCLUSION
From renewable feedstocks, an artificial pathway in E. coli was proposed and optimized to produce 5AVA in this study. Since the inhibition of enzyme activity and cell growth by H 2 O 2 is the main limiting factor in the production of 5AVA, catalase KatE was overexpressed to decompose H 2 O 2 to achieve high yield of 5AVA. Finally, an engineered strain CJ09 with RaiP, KivD, PadA, KatE, and LysP overexpression successfully produced 5AVA from biobased L-lys HCl at a final titer of 52.24 g/L. The renewable substrate and simple culture conditions were adopted in this work, while possessing higher yield and less environmental pollution. The improvement of substrate utilization and H 2 O 2 decomposition efficiency contributes to the increase in the yield of 5AVA, which has the potential to become a common strategy for the sustainable production of other chemicals.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
JC and WT performed the experiments, analyzed the data, and drafted the manuscript. ZL, QL, and XG analyzed the data. JC and DW conceived and coordinated the study. JC, DW, and JZ finalized the manuscript. All authors contributed to the article and approved the submitted version.