1. Construction of a cell factory for the production of 1, 5-pentanediol

There is no natural pathway for 1, 5-pentanediol (1,5-PDO) synthesis in wild-type Escherichia coli. Nowadays, the potential 1,5-PDO biosynthetic pathways have been proposed by other researchers. Here we designed a novel artificial 1,5-PDO biosynthetic pathway and assembled it in E. coli to obtain the cell factory for 1,5-PDO bioproduction. In the designed pathway, lysine is used as a director precursor, which is then converted into 5-hydroxyvalerate (5-HV) and 1,5-PDO via two synthetic modules (5-HV synthesis module and 1,5-PDO synthesis module). The 5-HV synthesis module converts lysine into 5-HV via the oxidase, hydrolase, transaminase, and aldehyde reductase, while 1,5-PDO synthesis module converts 5-HV into 1,5-PDO via carboxylate reductase and aldehyde reductase. As some of these enzymes are not present in E. coli, we added several heterologous enzymes into the metabolic network model of E. coli iML1515, and successfully established a 1,5-PDO metabolic network model (Figure 1). Based on a flux balance analysis (FBA), the maximum theoretical yield of 1,5-PDO from glucose based on our designed pathway was 0.847 mol/mol^Glucose, further identifying the theoretically feasibility of the artificial 1,5-PDO biosynthetic pathway.

Figure 1 The establishment of 1,5-PDO metabolic network model in E. coli

Next, we choose a lysine-producing strain of E. coli NT1003 as our chassis for the construction of 1,5-PDO pathway, which could avoid excessive modification of the front-end genes. In the designed pathway, a total of six enzymes are required for the bioconversion of lysine to the final product 1,5-PDO. The 5-hydroxyvalerate module of L-lysine monooxygenase (DavB) from Pseudomonas putida KT2440, 5-aminopentanamidase (DavA) from Pseudomonas putida KT2440 and 4-aminobutyrate aminotransferase (GabT) from E. coli were inserted into the vector pTrc99a (Figure 2), and the 1,5-PDO module of aldehyde reductase (YahK) from E. coli, phosphopantetheinyl transferase (sfp) from Bacillus subtilis, carboxylic acid reductase (MmCAR) from Mycobacterium marinum was inserted into the vector pACYCDuet-1 under the control of Ptrc (Figure 2). The two plasmids were co-transferred into E. coli NT1003 to assembly the 1,5-PDO biosynthetic pathways, and obtained the engineered strain of NT1003-P1, which was further validated by fermentation in shake flask.

Figure 2 The assembly of 1,5-PDO metabolic pathway in E. coli NT1003

We cultivated the engineered strain NT1003-P1 in a fermentation medium supplied with 20 g/L glucose. During the whole fermentation, the cell growth and 1,5-PDO production were measured. As shown in Figure 3, after fermentation of 80 h, cells could grow into an OD⁶⁰⁰ of 10, and 2.82 mM (0.293 g/L) of 1,5-PDO was produced from glucose. The results further indicated the feasibility of our designed pathway experimentally.

Figure 3 Production of 1, 5-pentanediol by fermentation with engineered strain NT1003-P1

2. The optimization of fermentation conditions

For the fermentation by the engineered strain NT1003-P1, we investigated the effect of IPTG addition time on the 1,5-PDO production. We divided the induction period into five periods according to the value of OD⁶⁰⁰, which are OD⁶⁰⁰ of 0-1,OD⁶⁰⁰ of 1-2, OD⁶⁰⁰ of 3-4, OD⁶⁰⁰ of 5-6, and OD⁶⁰⁰ of 7-8 respectively. When the fermentation was end, we analyzed the samples by HPLC. As the results shown in Figure 4, we determined that the production of 1,5-PDO exhibited the highest level when IPTG was added at the growth of OD⁶⁰⁰ 1~2.

Figure 4 Comparison of 1,5-PDO fermentation yields at different induction time in strain NT1003-P1

3. Comparison of catalytic activities of carboxylate reductase, aldehyde reductase and transaminase

In the process of metabolic pathway construction, we also constructed expression plasmids containing carboxylic acid reductase (Figure 5a), aldehyde reductase (Figure 6a) and transaminase (Figure 7a) from different organism sources, aiming to compare their catalytic efficiency and obtain the desired enzyme with higher activity.
By using a whole-cell catalysis process, the enzyme activity of carboxylic acid reductase, and aldehyde reductase was comparatively analyzed by measuring the 5-HV consumption and 1,5-PDO production with HPLC. By comparing the 5-HV consumption, MmCAR and MpCAR displayed the higher substrate consumption activity. The further detection of 1,5-PDO production showed that MmCAR from Mycobacterium marinum resulted in the highest activity for 1,5-PDO production (11.5 mM) (Figure 5b). Therefore, MmCAR was selected as the carboxylate reductase in the following strain construction.

Figure 5 (a) The screening of carboxylate reductase for the establishment of functional modules for 1,5-PDO bioproduction. (b) Comparing the activity of carboxylate reductase from different organisms

After identifying the efficient carboxylate reductase for 1,5-PDO synthesis module, we performed the synthesis of 1,5-PDO by a whole-cell catalysis process to compare the activities of the four aldehyde reductases by detecting 1,5-PDO production with HPLC. The results confirmed that YahK from E. coli exhibited the highest activity for 1,5-PDO production (Figure 6b). Thus, Yahk was selected to assembly the 1,5-PDO synthesis module in the following strain construction.

Figure 6 (a) The screening of aldehyde reductase for the establishment of functional modules for 1,5-PDO bioproduction (b) Comparing the activity of aldehyde reductases from different organisms

For the 5-HV synthesis module, the high activity of DavB and DavA to convert lysine to 5-aminovalerate (5-AVA) has been demonstrated. The identification of the aminotransferase that could catalyze 5-aminovalerate efficiently was critical for this module. Here, we determined the activity of aminotransferases by detecting the 5-HV production from 5-AVA with HPLC. As the results shown in Figure 7b, all six aminotransferases exhibit ammonia-transamination activity towards 5-AVA. Among them, GabT from E. coli exhibited the highest ammonia-transamination activity and achieved the highest 5-HV production (Figure 7b). Thus, GabT was selected to assembly the 5-HV synthesis module in the following strain construction.

Figure 7 (a) The screening of transaminase for the establishment of functional modules for 1,5-PDO bioproduction (b) Comparing the activity of aminotransferases from different organisms

4. The predicted gene optimization targets by metabolic network model simulation

After obtaining the functional enzymes, we further assembly the pathway in E. coli NT1003, and obtained the engineering strain of NT1003-P1 that could produce 1,5-PDO shown in Figure 3. Based on a metabolic network model simulation by FSEOF algorithm, the overexpression of genes in 1,5-PDO synthetic pathway was predicted to be benefit for increasing 1,5-PDO production (Table 1). During the fermentation, we found that the intermediate of 5-HV was largely accumulated. In conjunction with the model simulation and experimental results, we supposed that the activity of 1,5-PDO synthesis module is not matched with the activity of 5-HV synthesis module. Increasing activity of 1,5-PDO synthesis module should be done. At the meantime, based on a metabolic network model simulation by OptKncok algorithm, several gene deletion targets were also proposed (Table 1).

Table 1 Predicted Gene optimization TARGETS based on the metabolic network simulation

Reaction ID	Reaction	Enzyme or gene	Optimization strategy
ARND2	5-Hydroxypentanal + Nicotinamide adenine dinucleotide phosphate --> 1,5-Pentanediol + H+ + Nicotinamide adenine dinucleotide phosphate - reduced	1.1.1.2	Overexpression
APENTAMAH2	5-Aminopentanamide + H2O + H+ --> 5-Aminopentanoate + Ammonium	3.5.1.30	Overexpression
CARs	5-hydroxypentanoate + ATP C10H12N5O13P3 + H+ + Nicotinamide adenine dinucleotide phosphate - reduced --> 5-Hydroxypentanal + AMP C10H12N5O7P + Nicotinamide adenine dinucleotide phosphate + Diphosphate	1.2.1.30	Overexpression
ASAD	L-Aspartate 4-semialdehyde + Nicotinamide adenine dinucleotide phosphate + Phosphate <=> 4-Phospho-L-aspartate + H+ + Nicotinamide adenine dinucleotide phosphate - reduced	1.2.1.11	Knockout
ACKr	Acetate +ATP <=>Acetyl phosphate+ ADP	2.7.2.15, 2.7.2.1	Knockout
PTAr	Acetyl-CoA + Phosphate <=> Acetyl phosphate + Coenzyme A	2.3.1.8	Knockout
PDH	Coenzyme A + Nicotinamide adenine dinucleotide + Pyruvate --> Acetyl-CoA + CO2 + Nicotinamide adenine dinucleotide - reduced	b0115 and b0116 and b0114	Knockout

5. The optimization of 1,5-PDO cell factory by multiple engineering strategies

Based on experimental result analysis in conjunction with the model simulation guidance as descried above, we employed several engineering strategies to optimize strain performance, including optimizing gene expression by the high copy number plasmid, the engineering limiting enzyme as well as the assembly of limiting enzyme. What was more, the deletion of several genes in the branched pathway was also performed.

Due to the low copy number of the pACYCDuet plasmid, we supposed that the expression of MmCAR was insufficient, and thus affected the synthesis of 1,5-PDO. We replaced pACYCDuet in engineered bacteria with a high copy number plasmid of pRSFDuet to express genes in 1,5-PDO synthesis module, and obtained engineered strain NT1003-P2. The yield of 1,5-PDO produced by the engineered strain NT1003-P2 was determined by the HPLC. After a total fermentation period of 84 h, 8.09 mM (0.843 g/L)1,5-PDO was produced by the engineered strain E. coli NT1003-P2, which is 3.5-fold higher than in E. coli NT1003-P1 (Figure 8), indicating the optimization of gene expression by changing the high copy number plasmid was benefit for 1,5-PDO production in our work. This result also confirmed that the optimization of strain in conjunction with the metabolic network model simulation is an efficient approach.

Figure 8 Comparison of fermentation yields between NT1003-P1 and NT1003-P2

By using the OptKncok algorithm, multiple bypass genes that might affect the synthesis of 1,5-PDO were predicted (Table 1). In combination with the knockout results without affecting the normal growth of E. coli and the occurrence of the exchange reaction, we decided to knock out the two genes of ackA and pta, which encode the enzymes (acetate kinase and phosphate acetyltransferase) that convert pyruvate to acetate, the main by-product for E. coli fermentation. In addition, gene involving in 1,5-PDO degradation was also evaluated.
The results in Figure 9 exhibited the effect of deleting genes of ackA and pta on the final 1,5-PDO production. In comparison to engineered strain NT1003-P2, the knockout of ackA-pta increased the 1,5-PDO titer by 16%, and the titer in strain NT1003-P2-Δacka-pta reached 9.3 mM (0.97 g/L) (Figure 9), indicating the positive effect on 1,5-PDO production by deleting branched pathway of acetic acid synthesis. This result is also another success of strain optimization guided by the metabolic network model simulation.

Figure 9 Comparison of production between NT1003-P2 and NT1003-P2-ΔackAa-pta.

During the fermentation, we also evaluated the 1,5-PDO degradation ability of E. coli. Within 24 h, E. coli NT1003-P1 could consume (100.77 mM) 10.5 g/L 1,5-PDO (Figure 10a). From the previous study in the lab, the gene of YcjQ was identified to be responsible for 1,5-PDO degradation. The deletion of YcjQ significantly decreased the 1,5-PDO degradation (Figure 10a). Subsequently, we evaluated the 1,5-PDO production in YcjQ-deleting strain. We found that the deletion of YcjQ gene markedly improved the final 1,5-PDO titer, 1.75-fold higher than the strain NT1003-P2 (Figure 10-b). These results demonstrates that NT1003-P2-ΔYcjQ performed better, and we decided to adopt this strain in the following experiments.

Figure 10 (a) Comparison of residual 1,5-PDO catalyzed by NT1003-P1 and NT1003-P1-∆YcjQ. (b) Comparison of fermentation yields between NT1003-P2 and NT1003-P2-∆YcjQ

In the 1,5-PDO module, MmCAR was a critical limited enzyme. It has been reported that MmCAR has low catalytic efficiency for medium and long chain carboxylic acids. To improve the activity of MmCAR, we employed a rational engineering strategy to design the mutant MmCAR with higher activity. In combination with protein structure simulation, and molecular autodock, we focused on the amino site of Q302, and mutant it to E302 to change the polarity of the residues. After obtaining the mutated MmCARQ302E, we replaced the original MmCAR in the engineering bacteria, obtained NT1003-P3-∆YcjQ. As the results shown in Figure 11, the mutation of MmCARQ302E could moderately increase the final 1,5-PDO production, and 14 mM (1.45g/L) of 1,5-PDO in the engineered E. coli NT1003-P3-ΔYcjQ was finally obtained from glucose (Figure 11).

Figure 11 Comparison of fermentation yields between NT1003-P2-∆YcjQ and NT1003-P3-∆YcjQ

In the reported study, enzyme assembly based on protein scaffolds has been widely used to enhance the catalytic efficiency of enzymes. After reviewing the literature and discussing within the group, we focused on the EutM from Salmonella enterica. EutM is known to self-assemble as filament-like structure in vivo, and has been developed as a powerful protein scaffolding system to rapidly co-immobilize enzymes for improving the biocatalysis efficiency. We added a peptide of SpyTag on the N-terminus of MmCARQ302E, and also ligated a gene encoding EutM-SpyCatcher in the plasmid for enzyme assembly in vivo . The engineered strain containing the assembled enzymes was named as NT1003-P4-ΔYcjQ. As shown in Figure 12, the 1,5-PDO production of the engineered strain NT1003-P4-ΔYcjQ was determined by the HPLC analysis. After fermentation of 96 h, the assembly of MmCAR could increase 1,5-PDO production by 60%, and 23 mM (2.4 g/L) of 1,5-PDO was finally obtained in the engineered E. coli NT1003-P4-ΔYcjQ .

Figure 12 Comparison of fermentation yields between NT1003-P3-∆YcjQ and NT1003-P4-∆YcjQ

[1].Xuecong Cen. et al. Metabolic Engineering of Escherichia coli for De Novo Production of 1,5-Pentanediol from Glucose. ACS Synthetic Biology 2021 10 (1), 192-203

[2].Cheng, J. et al. Efficient whole-cell catalysis for 5-aminovalerate production from L-lysine by using engineered Escherichia coli with ethanol pretreatment. Sci Rep 10, 990 (2020)

[3].Chen Chen. et al. Metabolic engineering of Escherichia coli for the efficient production of 5-hydroxyvaleric acid, Process Biochemistry, Volume 130, Pages 625-633 (2023).

[4].Xuecong Cen. et al. Metabolic engineering of Escherichia coli for high production of 1,5-pentanediol via a cadaverine-derived pathway. Metabolic Engineering, Volume 74, Pages 168-177 (2022).

[5].Xin Wang. Et al. Efficient production of 5-aminovalerate from l-lysine by engineered Escherichia coli whole-cell biocatalysts. Journal of Molecular Catalysis B: Enzymatic, Volume 134, Part A, Pages 115-121 (2016)

[6].Tayyab Islam. et al. Metabolic engineering of Escherichia coli for biological production of 1, 3-Butanediol, Bioresource Technology, Volume 376, 128911 (2023)

[7].Zhang, G. et al. Self-Assembling Protein Scaffold System for Easy in Vitro Coimmobilization of Biocatalytic Cascade Enzymes. ACS Catalysis, 8(6), 5611-5620.

[8].Guoqiang Zhang. et al. Chapter Thirteen-Protein-based scaffolds for enzyme immobilization. Claudia Schmidt-Dannert, Maureen B. Quin, Methods in Enzymology, Academic Press, Volume 617, Pages 323-362 (2019)

[9].Xiao-Bao Sun. et al. SpyTag/SpyCatcher molecular cyclization confers protein stability and resilience to aggregation, New Biotechnology, Volume 49, Pages 28-36 (2019)