According to the ninth edition of the Diabetes Atlas published by the International Diabetes Federation in 2019, more than 463 million adults are suffered from diabetes worldwide, and this number will reach 578.4 million by 2030 and up to 700.2 million by 2045 (Saeedi et al. 2019). Among these populations, the populations with type 2 diabetes mellitus (T2DM) accounts for approximately 90%. These patients are characterized by various degrees of insulin deficiency and extensive insulin resistance. Encouragingly, (2S, 3R, 4S)-4-hydroxyisoleucine (4-HIL), which was found in fenugreek (Trigonella foenum-graecum) (Neelakantan et al. 2014), exhibits the particular glucose-dependent insulin-stimulating and insulin-sensitizing activities. Thus, 4-HIL becomes a promising drug for treating the T2DM (Zafar and Gao 2016). Traditionally, 4-HIL is extracted from fenugreek seeds, but its yield is only 150 mg/kg (Jetté et al. 2009). Fortunately, the α-ketoglutarate (α-KG)-dependent L-isoleucine dioxygenase (IDO) derived from Bacillus thuringiensis was found to specifically convert L-isoleucine (Ile) to 4-HIL (Ogawa et al. 2011). 4-HIL was then synthesized by overexpressing the IDO gene (ido) in Escherichia coli, however, the addition of Ile makes such process costly and inefficient (Smirnov et al. 2010).
Corynebacterium glutamicum is globally recognized as safe and is widely used in the production of various amino acids (Becker et al. 2018; Tsuge and Matsuzawa 2021). In our previous study, through overexpressing ido gene derived from B. thuringiensis YBT-1520 in an Ile-producing C. glutamicum ssp. lactofermentum strain SN01, 65.44 ± 2.27 mM 4-HIL was synthesized from the self-produced Ile (Shi et al. 2015). In order to improve 4-HIL production, metabolic engineering was performed, such as synergistically enhancing the substrates supply and IDO activity by particular gene overexpression and deletion (Shi et al. 2016; Shi et al. 2018; Shi et al. 2019). The engineered strain SZ06 could produce 111.99 ± 2.15 mM 4-HIL (Shi et al. 2019). Although the 4-HIL production of SZ06 increased significantly, a by-product, L-lysine (Lys) was also accumulated, because the first several enzymes of the Ile biosynthetic pathway are involved in Lys biosynthesis simultaneously. Furthermore, diaminopimelate, an intermediate metabolite of the Lys biosynthetic pathway, is required for cell growth, making it difficult to weaken the Lys biosynthesis. The accumulation of Lys limits the further increase of 4-HIL production.
Adaptive laboratory evolution (ALE) has become a powerful strategy in metabolic engineering. Compared with traditional metabolic design, ALE generates non-intuitive beneficial mutations that can occur in many different gene regions simultaneously under specific selection pressures (Portnoy et al. 2011). This natural selection provides an effective basis for obtaining new characteristics, phenotypes, and beneficial mutations in microorganisms, which can be fixed after multiple rounds of continuous evolution (Desai and Fisher 2007). It is precisely because of the rapid adaptation of microorganisms to different environments, the application of ALE for phenotypic optimization has been extended to various aspects, such as isolating the best strains, activating latent pathways, and improving environmental tolerance of production strains, etc (Portnoy et al. 2011; Sandberg et al. 2019). This phenotypic optimization not only improves overall cell function but also improves the physiological adaptation of microbial strain. Because normally, the adaptive rate under specific selection pressures is directly coupled to the growth rate, this coupled metabolic design will result in significantly higher production and lower by-products (Fong et al. 2005). For example, ALE was successfully applied to improve cellular tolerance of a C. glutamicum strain to high concentrations of methanol, thereby enhancing methanol biotransformation, meanwhile, the cell growth was also improved (Tuyishime et al. 2018; Wang et al. 2020). ALE was used to develop potential evolutionary strains of Zymomonas mobilis that can co-utilize glucose and xylose (Millán et al. 2020). Besides lowering effects by toxic chemicals, ALE has also been successfully used to improve production with C. glutamicum. For example, ALE combined with rare codon-rich markers was applied to select a mutant strain with 3.7-fold increased production of L-arginine (Arg) (Zheng et al. 2018). ALE combined with metabolic engineering was successfully used to improve the production of putrescine in C. glutamicum (Li et al. 2018). Through ALE driven by an L-valine (Val) biosensor Lrp-PbrnFE in combination with atmospheric and room temperature plasma mutagenesis, an mutant strain HL2-7 was successfully obtained and its Val production increased by 21.47% (Han et al. 2020).
However, when the selection pressure coupled to growth is not available, the desired mutant strains are difficult to be screened out by ALE. In addition, the mutation rate is too low when ALE is performed under natural evolution. Considering these two problems, ALE was ameliorated by using feed-back regulated evolution of phenotype (FREP) to increase the mutation rate and the population diversity (Binder et al. 2013). The FREP is actually a programming adaptive control system, in which the mutation rate is improved under the initial condition and decreased under the final condition when the concentration of target metabolites increased, thus resulting in higher positive screening rate (Chou and Keasling 2013). FREP is implemented in two parts: a biosensor for sensing the concentration of the target metabolite to generate the transcriptional signal and an actuator for changing the mutation rate when receives transcriptional signal. In this case, the sensor consists of two components: a transcription factor (TF) that binds to the target metabolite and a promoter that is regulated by the TF. The actuator is usually a mutagenesis gene, which can shorten the evolution time. The effect of FREP was successfully verified by evolving E. coli strains to increase the production of tyrosine and isoprenoid, meanwhile, a fluorescent protein, another actuator, was expressed in order to detect the visible changes in the evolved strains (Chou and Keasling 2013). Delightfully, more and more researches have been conducted in the field of TF-based biosensors. However, sensors that can be used for metabolic engineering of C. glutamicum is limited, for example, the branched-chain amino acids (BACCs) and L-methionine biosensor Lrp-PbrnFE (Mustafi et al. 2012; Tan et al. 2020), Lys, L-histidine (His), and Arg biosensor LysG-PlysE (Schendzielorz et al. 2014; Kortmann et al. 2019), and shikimic acid biosensor ShiR-PshiA (Liu et al. 2017).
lysE encodes the Lys, His, and Arg export protein, while LysG, an LysR-Type TF, is a positive regulator of lysE transcription. In the presence of intracellular Lys, Lys binds to LysG, leading LysG to bind in the promoter region of lysE, thereby promoting the transcription of lysE (Bellmann et al. 2001). Recently, an optical Lys sensor pSenLys was constructed based on LysG-PlysE, and several new Lys producer were generated by fluorescence-activated cell sorting (FACS) of mutant strains harboring this sensor (Binder et al. 2013). Subsequently, semi-conserved LysG was designed to construct the biosensor pSenHis with focused His and Arg specificity and the His-producing mutant strains of C. glutamicum was obtained through this pSenHis-based FACS (Della Corte et al. 2020).
In this study, in order to improve the 4-HIL production and reduce the by-product Lys production, a programming ALE driven by the Lys-sensing evolution system was employed. Firstly, a Lys-sensing evolution system pMK was constructed. It containes two parts: the Lys biosensor LysG-PlysE and the actuator composed of a mutagenesis gene cdd and a fluorescent reporter gene egfp. cdd gene encodes the cytosine deaminase. It has been used in genome editing to cause specific mutations at target sites and generate gene silencing (Banno et al. 2018; Wang et al. 2019). cdd was used here to accelerate mutation and evolution. Thereby, the mutation rate and eGFP fluorescence intensity of cells may be positively regulated in this system according to the intracellular Lys concentration. Secondly, the evolutionary strain SZ06/pMK was constructed and programming ALE was conducted to generate the evolved positive mutants. Finally, the whole genome of these mutant strains was sequenced and the mutations in these evolved strains were analyzed.