Introduction
Glutaric acid, a C5 chain dicarboxylic acid, is used as the building block for synthesizing nylon and plasticizer (J. Adkins, J. Jordan, & D. R. Nielsen, 2013; Bermúdez, León, Alemán, & Muñoz-Guerra, 2000; W. Li et al., 2019; Zhang, Gao, Guo, Guo, Kang, Xiao, Yan, Tao, Zhang, & Dong, 2018). Currently, glutaric acid is prepared using nitric acid oxidation and separated from a mixture of dicarboxylic acids, but its yield is not high (Castellan, Bart, & Cavallaro, 1991; Nishikido, Tamura, & Fukuoka, 1979; Niu, Draths, & Frost, 2002; Polen, Spelberg, & Bott, 2013; Sato, Aoki, & Noyori, 1998). The most notable natural route for glutarate production was discovered inPseudomonas putida , which could utilize L-lysine degradation to produce glutarate by the 5-aminovalerate (AMV) pathway(JC & JR, 1977; Revelles, Espinosa-Urgel, Fuhrer, Sauer, & Ramos, 2005; Revelles, Wittich, & Ramos, 2007; Zhang, Gao, Guo, Guo, Kang, Xiao, Yan, Tao, Zhang, & Dong, 2018). In order to avoid the supply of precursor lysine and improve the glutarate production, the lysine overproducingCorynebacterium glutamicumoverexpressed the AMV pathway from P. putida to produce glutaric acid in high titers and yields (J. Adkins, J. Jordan, & D. R. Nielsen, 2013; Kim et al., 2019; Maria, Gideon, Michael, Christoph, & Judith, 2016; Rohles et al., 2018; Shin et al., 2016). At the same time, in order to better study the AMV pathway, those genes were also heterologous expressed in Escherichia coli(J. Adkins, J. Jordan, & D. R. Nielsen, 2013; Park et al., 2013). The maximum titer of 0.82 g/L for glutarate was reached after 48 hours of fermentation (J. Adkins, J. Jordan, & D. R. Nielsen, 2013). In order to obtain higher output, lysine and α-KG were supplemented to the cell catalytic system resulting in 1.7 g/L glutarate in E. coli WL3110 strain(Park et al., 2013). Recently, glutarate titer was also greatly improved through whole cell bio-catalyst or whole-cell immobilized (Hong et al., 2018; S.-Y. Yang et al., 2019). And Li et al used the native lysine catabolic pathway in E. coli to synthesize glutarate and finally achieved both high titer and high yield(W. Li et al., 2019).
The above all results are based on AMV pathway or lysine carbolic pathway. In addition to the AMV pathway, a variety of attempts for the glutarate production and exploration were made in E. coli . Recently, the reverse adipic acid degradation pathway (RADP) was utilized to balance the production of adipate and glutarate and finally achieved 4.8 g/L glutarate (Zhao, Huang, et al., 2018; Zhao, Li, & Deng, 2018). Then the glutarate in RADP was accumulated to 6.3 g/L by malonate absorption(Sui et al., 2020). The RADP seems to be a potential pathway to produce glutarate from glucose inE. coli . The RADP contained the following enzymes: Tfu_0875 (β-ketothiolase), Tfu_2399 (3-hydroxyacyl-CoA dehydrogenase), Tfu_0067 (3-hydroxyadipyl-CoA dehydrogenase), Tfu_1647 (5-carboxy-2-pentenoyl-CoA reductase) and Tfu_2576-7 (adipyl-CoA synthetase) (Sui et al., 2020; Zhao, Huang, et al., 2018; Zhao, Li, & Deng, 2018) (Fig. 1). The RADP could convert acetyl-CoA and malonyl-CoA to glutaric acid and malonyl-CoA was crucial for the glutarate production. Therefore, cerulenin was used for increasing the amount of malonyl-CoA available by inhibiting fatty acid synthesis pathway(Heath & Rock, 1995; Rogers & Church, 2016; Zhao, Li, & Deng, 2018). And matB(malonic acid synthetase) and matC (malonic acid carrier protein) were overexpressed, absorb the malonate to improve the intracellular malonyl-CoA(Sui et al., 2020; Wu, Du, Zhou, & Chen, 2013). However, only increasing the content of intracellular malonyl-CoA was not sufficient for glutarate production. Both precursors of glutaric acid should be considered to increase production.
The acetyl-CoA could be transformed to malonyl-CoA by catalyzing acetyl-CoA carboxylase (ACC) (S. J. Li & Cronan, 1992; Lussier, Colatriano, Wiltshire, Page, & Martin, 2012; Xu, Li, Zhang, Stephanopoulos, & Koffas, 2014; Zhu, Wu, Du, Zhou, & Chen, 2014). But acetyl-CoA mainly flew to the tricarboxylic acid (TCA) cycle to maintain basic metabolism and growth, whereas only a minor proportion was used by ACC to synthesize malonyl-CoA(Tokuyama et al., 2019), and a diminutive proportion was engaged in fatty acid synthesis(Zha, Rubin-Pitel, Shao, & Zhao, 2009) in E. coli. Therefore, it was important to balance the supply and demand of acetyl-CoA and malonyl-CoA to flow to the target product. Here, our research is based on the previous RADP pathway to produce glutaric acid (Fig. 1). Firstly, we constructed a transformation system of acetyl-CoA to malonyl-CoA for enhancing the available intracellular malonyl-CoA in E. coli . Furthermore, the acetic acid uptake pathway was introduced into the metabolic pathway and the degradation pathway was knocked out to increase and balance intracellular acetyl-CoA and malonyl-CoA. Meanwhile, the dissolved oxygen was effectively controlled so that more carbon source flew to CoA. Finally, the optimal strains, Bgl51464 was cultured in a 5‐L fermenter.
  1. Materials and methods
  2. Strains and plasmids
The strains, plasmids, and primers are shown in Supplementary Tables S1 and S2. In order to promote the transformation of acetyl-CoA to malonyl-CoA in vivo , theaccABCD (Xu et al., 2014) fromE. coli MG1655 was firstly linearized by PCR using accA-1 F and accA-1 R, accB-1 F and accB-1 R, accC-1 F and accC-1 R and accD-1 F and accD-1 R, respectively. Then the pACYCDuet-1 was digested usingBam H I and Avr II and all products were ligase by Gibson assembly (Gibson et al., 2009) and transformed into competent cells, forming strain pACC1. The accBCand dtsR1 were amplified from C. glutamicum (Zhu et al., 2014) using accBC-2 F and accBC-2 R and dtsR1-2 F and dtsR1-2 R primers, respectively. The pACYCDuet-1 was linearized using kpn I andAvr II. The all products were assembled like pACC1, resulting in strain pACC2. To increase the CoA in vivo , the acs fromE. coli MG1655 was amplified using acs-3 F and acs-3 R. Both the pACYCDuet-1 and the acs fragment were digested using Bam H I and Not I and ligated by T4 DNA ligase, naming pACC3. TheaccBC and dtsR1 were amplified from pACC2 using accBC-2 F and dtsR1-2 R. The pACC3 plasmid was digested by kpn I andAvr II and all products were assembled, forming strain pACC4. All above constructs were screened by colony PCR and Sanger sequencing using veri-pACYC F and veri-pACYC R. All subsequent plasmid constructions (packA, and ppoxB) also used the Gibson assembly method as described above. The sgRNA target gene and knock gene template are shown in Supplementary Table S3.