Skip to main content
Download PDF

Combining biosensor and metabolic network optimization strategies for enhanced l-threonine production in Escherichia coli

Biotechnology for Biofuels and Bioproducts volume 18, Article number: 37 (2025) Cite this article

Abstract

l-threonine is an integral nutrient for mammals, often used in animal feeds to enhance growth and reduce breeding costs. Developing l-threonine engineered strains that meet industrial production specifications has significant economic value. Here, we developed a biosensor that monitors l-threonine concentration to assist in high-throughput screening to capture high-yielding l-threonine mutants. Among them, the PcysK promoter and CysB protein were used to construct a primary l-threonine biosensor, and then the CysBT102A mutant was obtained through directed evolution resulting in a 5.6-fold increase in the fluorescence responsiveness of biosensor over the 0–4 g/L l-threonine concentration range. In addition, the metabolic network of mutant was further optimized through multi-omics analysis and in silico simulation. Ultimately, the THRM13 strain produced 163.2 g/L l-threonine, with a yield of 0.603 g/g glucose in a 5 L bioreactor. The biosensor constructed here could be employed for iterative upgrading of subsequent strains, and these engineering strategies described provide guidance for other chemical overproducers.

Graphical Abstract

Background

As an essential nutrient, l-threonine is mainly used in additives, food products, and pharmaceuticals [1]. In recent years, the crucial role of l-threonine in animal feeds (such as piglet feeds) has become evident, making the development of l-threonine overproducing strains economically significant. Despite the titer of l-threonine exceeding 120 g/L in E. coli (Table S1) [2], the current production levels fall short when compared to l-lysine and l-glutamate [3]. With the gradual advancement in the understanding of metabolic networks, constructing microbial cell factories using systems metabolic engineering strategies to produce target chemicals has shown great promise [4, 5]. In particular, in the breeding of amino acid-producing industrial microorganisms, many reprogramming strategies have been employed to swiftly de novo develop amino acid overproducers. Although some strains have achieved significant success, their application to industrial production still faces many challenges. For example, products with high titers but low yields fail to meet the cost-effectiveness required for industrial production.

A constrained understanding of complex metabolic networks hinders the rapid identification of effective engineering targets for improving the yield of target products. Combining high-throughput screening techniques with non-rational engineering strategies has become an effective approach to achieving iterative strain evolution and efficient screening [6,7,8]. The utility of transcription factor biosensors, responsive to changes in specific intracellular metabolite concentrations, has become an effective practice in screening amino acid overproducers from mutant libraries [9,10,11]. Within the microorganism genome, the non-coding region sequence orchestrating intricate metabolic regulation serves as the natural library for these transcription factor biosensors [12, 13]. This innovative approach harnesses the intrinsic regulatory elements within the genome, offering a diverse range of potential biosensors. In addition, integrating transcriptomic analysis techniques into the screening of target biosensors, combined with rational design strategies, has paved the way for the development of more sensitive and effective identification systems [14, 15]. For example, Wu et al. pioneered the development of a biosensor device designed to respond to glucose-6-phosphate, efficiently modulating the synthesis of N-acetylglucosamine [16]. Liu introduced an in vivo mutagenesis system that disrupts DNA repair and replication, facilitating efficient and broad-spectrum iterative mutagenesis (MP6s) at the laboratory scale [17, 18]. These advancements in biosensor technology and screening methodologies enhance the precision and efficiency of strain construction.

In addition to the engineering strategies outlined earlier, an alternative approach to bolster strain productivity involves the exploration of effective genetic targets within the intricate global metabolic networks. For example, quantifying metabolic differences among different hosts through multi-omics analysis techniques to identify potential metabolic targets [19]. The optimal solution of metabolic fluxes is simulated in silico for maximizing the target product production. However, this approach requires optimizing the genome-scale metabolic network (GSMN) by removing specific catalytic reactions or incorporating new constraint equations to improve the accuracy of the simulation results [20].

In this study, we present engineering strategies to further increase the productivity of l-threonine producer. First, transcriptomic analysis was used to screen promoters that can sense exogenous l-threonine in E. coli. Next, a highly sensitive fluorescent reporter system responsive to changes in l-threonine concentration was constructed by combining CysB mutants with eGFP. Subsequently, the biosensor was used to assist the two-step high-throughput screening technique for strain iterative evolution to capture superior mutants. Next, beneficial targets are identified through multi-omics analysis, and intracellular carbon flux allocation is further optimized using GSMN to maximize l-threonine production. This study not only developed an l-threonine overproducing strain with potential industrial applications but also demonstrated a novel approach for constructing highly sensitive l-threonine biosensors, offering new avenues for the development of biosensors targeting other chemicals.

Materials and methods

Strain, plasmids, chemicals

Transcriptomic and genomic analysis, Oligonucleotides synthesis were implemented by Azenta (Suzhou, China). Multif seamless assembly mix were supplied by AB clonal (Wuhan, China). DNA polymerases (p525); gel DNA, plasmid, and genome extraction kits were produced by Vazyme (Nanjing, China). The strains and plasmids used in this study can be found in Supplementary Table S2.

Transcriptomics analysis of gene expression in response to l-threonine addition

An individual colony of wild-type MG1655 was inoculated in 10 mL high-salt Luria–Bertani (LB) medium (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl) for 12 h (220 rpm, 37 °C). 2 mL of culture was injected into 100 mL LB medium for 4 h and then all was transferred to 500 mL flasks containing 0, 30, and 60 g/L l-threonine, respectively. After continuing the incubation for 2 h, all cultures were harvested and centrifuged at 8000×g, 4 °C, for 10 min. After removing the supernatant, the tubes containing the cells were immediately placed in dry ice to ensure rapid freezing and prevent RNA degradation. Subsequently, these samples were shipped to Azenta (Suzhou, China) for transcriptomic analysis.

Transcriptomics analysis of THRM1 and THR36-L19 strains

Individual colonies of THRM1 and THR36-L19 were inoculated in 10 mL LB medium for 12 h (220 rpm, 37 °C), respectively. 2 mL cultures of THRM1 and THR36-L19 were injected into 100 mL LB medium for 10 h (220 rpm, 37 °C). Then, all cultures of THRM1 and THR36-L19 were transferred to a 5 L bioreactor containing 2 L fermentation medium (composition detailed in the “Fermentation conditions in shake flasks” section), respectively. The dissolved oxygen (DO) was maintained at 30%, and the pH was maintained at 7.0 by increasing the stirring speed and adding ammonia. After 14 h, 50 mL cultures of THRM1 and THR36-L19 were harvested, and each was centrifuged at 8000×g, 4 °C for 10 min. After removing the supernatant, the cultures of THRM1 and THR36-L19 were submitted to Azenta (Suzhou, China) for transcriptomics analysis (stored in dry ice).

De novo sequencing

An individual colony of the THRM1 strain was inoculated in 10 mL LB medium for 12 h (220 rpm, 37 °C). Then, 2 mL culture was injected into 100 mL LB medium for 12 h. All cultures were harvested and centrifuged at 8000×g, 4 °C, for 10 min. After removing the supernatant, the culture was submitted to Azenta (Suzhou, China) for genome sequencing (stored in dry ice). DNA sequence analysis was conducted using the genome of the model strain E. coli MG1655 as a reference.

Design and refinement of l-threonine biosensors

The complete non-coding regions of 21 genes were amplified by PCR using the E. coli genome as a template. Then, the non-coding region sequences, linearised pTrc99A and eGFP were ligated using the Seamless Cloning Kit and the reaction products were all transferred to E. coli DH5α and cultured on LB agar plates for 12 h at 37 °C to construct the initial fluorescent reporter system. To validate the system, E. coli DH5α cells were transformed with an eGFP plasmid driven by no promoter as a negative control to establish a fluorescence baseline. In addition, E. coli DH5α cells were transformed with an eGFP plasmid under the control of the strong constitutive promoter J23119 as a positive control to ensure the proper functioning of the fluorescence reporter system (Fig. S1). Positive transformants were selected from the plates and inoculated into 24-well plates containing LB medium with 0, 10, 20, and 30 g/L l-threonine. The cultures were incubated for 8 h (220 rpm, 37 °C). Then, the eGFP fluorescence of the transformants in different concentrations of l-threonine was measured, and plasmids with a linear positive response were screened. PcysK, PcysJ, PcysP, and PcysD truncation mutants were obtained by reverse amplification of the parental plasmid using primers deleting the CysB binding site. The plasmid linking the PcysK-egfp gene was linearized and then tandemly linked to the CysB protein to construct the pSensor biosensor. Subsequently, the pSensor plasmid was reverse amplified using primers containing the point mutation to get the pSensorThr mutant containing CysBT102A.

RBS library preparation and screening

The first 210 bp of the gdhA gene was selected to be ligated to the egfp gene using a flexible linker (GGGGS)3 and integrated onto the pTrc99A plasmid to construct a fluorescent reporter system. The parental plasmid was used as a template and amplified by primers containing 10 degenerate bases “N” (any one of A, T, G, C) for the construction of the RBS mutant library (Table S3). The PCR products were digested using Dpn I enzyme for transformation into E. coli DH5α and the cells were cultured on LB agar plates at 37 °C for 12 h. Single clones were selected from the plates and inoculated into 24-well plates, then incubated for 10 h (200 rpm, 37 °C). In a 24-well plate, the PgdhA promoter-driven gdhA (210 bp)-egfp plasmid was used as a negative control, while the expression of gdhA (210 bp)-egfp driven by the Ptrc promoter and the original RBS was used as a positive control. Strains with different fluorescence expression intensities were selected from 110 mutants for sequencing analysis. Following incubation, they were centrifuged at 8000×g for 10 min, and the pellets were suspended and washed twice with phosphate-buffered saline (PBS). Then, the mutants with higher fluorescence than the parental plasmid were screened from the library. The fluorescence output of cells was detected using a multifunctional enzyme marker Biotek (synergy H4).

Screening l-threonine overproducers from mutation libraries

First, the MP6ts plasmid was transformed into competent cells of the THR36-L19 strain to generate THR36-L19-MP6ts. A single clone of positive transformant was inoculated in LB medium and induced with 0.1 mM arabinose for 24 h (220 rpm, 30 °C) to generate mutant library. Removal of the MP6ts plasmid from the mutant was achieved by incubation at 42 °C. Transforming the pSensorThr plasmid into competent cells of mutant library was for screening l-threonine overproducing strains. Mutants containing the pSensorThr plasmid were inoculated into 50 mL fermentation medium for 12 h to accumulate l-threonine and induce eGFP expression. 2 mL culture was washed three times with PBS and resuspended to an OD600 of 0.4 for flow cytometry analysis. Based on forward scatter and side scatter signals, gating was performed to minimize interference from erroneous events. After gating, at least 20,000 events were collected for subsequent analysis, and the top 0.5% of cells with the highest fluorescence intensity were sorted for further sorting and fluorescence measurement.The enriched highly fluorescent population was coated on LB plates and incubated for 20 h at 37 °C. Then, 190 clones were inoculated into two 96-deep well plates with fermentation medium by Qpix 420 (Molecular Devices) for 12 h and then fluorescence was determined. From these, the top 10 mutants with the highest fluorescence values were selected and the above operation was repeated for iterative mutation of the strains. After five rounds of random mutagenesis, the 50 strains with the highest fluorescence values were selected for validation of fermentation performance in shake flasks. The production of l-threonine was determined by high-performance liquid chromatography.

Carbon flux balance analysis and target prediction for gene knockout

A model based on E. coli iML1515 (http://bigg.ucsd.edu/models/iML1515) was further modified for Flux Balance Analysis (FBA) and Optknock analysis experiments. The model incorporated the PYCRe-catalyzed oxaloacetate synthesis pathway, introducing the compound reaction equation: ATP + pyruvate + HCO3− + H+  → ADP + phosphate + oxaloacetate. In addition, the reaction fluxes of tdcC, tdh, sstT, ilvA, poxB, pflB, pykF, and ldhA genes were tuned to zero in the model. Further enhancement of l-threonine synthesis was achieved through the modification GSMN model with OptKnock, a computational framework for predicting knockout targets to achieve chemical overproduction [21]. The l-threonine synthesis reaction was designated as the target function. The in silico to simulation solution was implemented using COBRA Toolbox (v3.0) [22] and Gurobi Optimizer. The CRISPR/Cas9 gene editing system was used to knockout genes predicted by in silico analysis, following the experimental procedure outlined by Zhao et al. [23]. Shake flask fermentation was then employed to assess the impact of the gene knockouts on strain growth and l-threonine production.

Fermentation conditions in shake flasks

For shake flask incubation, a single colony was selected from LB agar medium and inoculated in a 50 mL shake flask containing 10 mL LB medium at 220 rpm, 37 °C. After 12 h feeding, 2 mL culture was transferred to a 500 mL shake flask with 28 mL of seed medium. The seed medium composition included 15 g/L corn steep liquor, 3 g/L yeast extract, 2 g/L citrate, 0.5 g/L MgSO4·7H2O, 20 mg/L FeSO4·7H2O, 20 mg/L MnSO4·H2O, and 30 g/L glucose. Following an additional 10 h of feeding, 5 mL culture was shifted to a 500 mL shake flask with 25 mL fermentation medium. The fermentation medium contained 10.0 g/L corn steep liquor, 2.0 g/L yeast extract, 2.0 g/L citrate, 1.0 g/L KH2PO4, 1.0 g/L MgSO4·7H2O, 20.0 mg/L FeSO4·7H2O, 20.0 mg/L MnSO4·H2O, 3 g/L (NH4)2SO4, 0.2 mg/L biotin, 0.2 mg/L vitamin B1, 0.02 mg/L vitamin B12, and 15 g/L glucose. The seed medium is used during the initial cell expansion phase, while the fermentation medium is used after cell expansion for the fermentation phase to achieve efficient metabolic product synthesis.

Fed-batch fermentation in a 5 L bioreactor

A single colony of positive bacteria was harvested from LB agar medium and inoculated into 10 mL LB medium. After a 12-h incubation period, 2 mL culture was transferred to a 500 mL shake flask containing 100 mL LB medium and cultivated for 10 h. Then, 100 mL culture was transferred to a 5 L bioreactor with 1.9 L seed medium and cultivated for an additional 10 h at 37 °C. The cultivation continued until glucose was completely consumed. DO levels were controlled at 30.0% ± 2.0% by automatically adjusting the stirring speed, while the pH was maintained at pH 7.0 ± 0.02 through the automatic addition of ammonia. Following this, 400 mL the culture (20% inoculum) was transferred to a 5 L bioreactor filled with a 1.6 L fermentation medium. The parameters for DO and pH during fermentation were kept consistent with the previous stage.

Analytical methods

The biomass of the strain was detected by measuring the optical density at 600 nm using the UV-1200 spectrophotometer from AOE INSTRUMENTS, Shanghai, China. Glucose concentration in the fermentation broth was quantified by the Biosensor Analyzer S-10 from Sieman Technology, Shenzhen, China. Amino acids and organic acids in the fermentation broth were detected via high performance liquid chromatography (HPLC) utilizing pre-column derivatization with o-phthaldialdehyde and elution with 5 mM H2SO4, respectively [19]. l-threonine assay: Mobile phase A: 8 g sodium acetate dissolved in 1000 mL water, 225 μL triethylamine added, pH adjusted to 7.20 ± 0.05 with acetic acid, then 10 mL tetrahydrofuran added. Mobile phase B: 6 g sodium acetate dissolved in 200 mL water, pH adjusted to 7.20 ± 0.05 with acetic acid, then 400 mL methanol and 400 mL acetonitrile added. Column type: C18 column (250 × 4.6 mm); Column temperature: 40℃; Flow rate: 1 mL/min; UV detection wavelength: 338 nm; Gradient elution: 0–17.5 min: from 92% A, 8% B to 40% A, 60% B; 17.5–21.5 min: from 40% A, 60% B to 100% B; 21.5–24 min: 100% B; 24–30 min: from 100% B to 92% A, 8% B; Run time: 30 min. The FCS3 format data obtained were processed using FlowJo_V10 software. Experimental data were expressed as mean ± standard deviation on at least three parallel experiments.

Results and discussion

Screening for endogenous genetic elements responsive to changing l-threonine levels

With the continuous advancement of synthetic biology technology, the establishment of microbial cell factories for the production of target chemicals has become a popular direction [24, 25]. Alongside rational modification strategies, screening for phenotypes of interest from large mutation libraries is another viable approach for acquiring high-yielding strains [10, 26]. Multi-omics technology is an effective tool for rapidly mining relevant genetic factors from microorganisms to construct biosensors for target products [27, 28]. Jiang et al. successfully constructed an arginine overproducing strain using a biosensor-assisted high-throughput screening platform [29]. Liu et al. constructed a fusion promoter, PcysJH, for screening l-threonine mutants by proteomic analysis [30]. Inspired by this, we endeavored to develop a highly sensitive l-threonine biosensor for efficient screening.

To mine genetic coding elements in E. coli that could respond to changes in l-threonine concentration to construct a primary fluorescent reporter system, E. coli MG1655 was induced by exogenous addition of l-threonine at different concentrations (0, 30, and 60 g/L). Subsequently, the global gene expression changes were investigated by transcriptomics analysis, and 32 genes that were positively correlated with the l-threonine concentration were preliminarily screened based on their expression (Fig. 1A). The positional distribution of the 32 genes in the genome was investigated and it was found that some of the genes were in the same operon (such as cysDN, cysPUWA, cysJI, ygeWXY and hyuA, yeeED, xdhAB, proVWX), so the promoters of 21 of these genes were further screened and characterized. The non-coding regions of these 21 genes were ligated in tandem with the eGFP protein on the pTrc99A plasmid to create promoter library. Then, the responsiveness of these promoters at 0, 10, 20 and 30 g/L l-threonine levels was examined. Among them, PcysD, PcysJ, PcysP, and PcysK showed a linear positive correlation with l-threonine concentration and the PcysK promoter showed the best response performance (Fig. 1B), whereas the other promoters did not show a significant correlation (Fig. 1C). Although the ability of the ‘cys’ promoter to respond to l-threonine was determined by further characterization of the promoter library, the narrow range of fluorescence thresholds produced by this natural genetic element tends to lead to a high rate of false positives during the screening process.

Fig. 1
figure 1

Mining, design and characterization of l-threonine biosensor. A Cluster analysis of gene expression. Cluster analysis with linear positive correlation of gene expression levels in E. coli MG1655 after treatment with 0, 30, and 60 g/L l-threonine. B Fluorescent reporter system was constructed to characterize the changes in green fluorescence intensity of cells after induction of PcysK, PcysJ,PcysD,PcysP promoters at different concentrations of l-threonine (0, 10, 20 and 30 g/L). C To characterize the changes of green fluorescence intensity of cells after inducing the remaining 17 promoters with different concentrations of l-threonine (0, 10, 20 and 30 g/L)

Full size image

Engineering natural sensors to improve response to l-threonine

To address the issues of low sensitivity and response threshold in natural sensors, an artificial design was implemented to enhance screening performance. Initially, non-coding region sequences from cysD, cysJ, cysP, and cysK genes were compared (Fig. 2A), revealing the presence of CysB protein binding sites upstream of the -35 region [31, 32]. We speculated that the responsiveness of the “cys” promoters to l-threonine might be influenced by the CysB protein. Therefore, the CysB binding site in the PcysD, PcysJ, PcysP, and PcysK promoters were deleted to produce Ec-PcysD-1, Ec-PcysJ-1, Ec-PcysP-1, and Ec-PcysK-1, respectively. Testing revealed that these mutants lost their responsiveness to l-threonine and displayed only background expression (Fig. 2B). Subsequently, the biosensor “pSensor” was constructed by overexpressing CysB in the higher sensitive pTrc99A-PcysK-egfp plasmid (Fig. 2C). Compared to PcysK-egfp, pSensor exhibited a higher fluorescence threshold at l-threonine concentrations of 0–10 g/L but lacked correlation at 10–30 g/L (Fig. 2D). Further optimizing the induced concentration of l-threonine, the pSensor was determined to maintain a linear fluorescence response between 0 and 4 g/L l-threonine (Fig. 2E), and the fluorescence threshold was improved by about 2.1-fold compared with PcysK-egfp.

Fig. 2
figure 2

Development, test and optimization of the l-threonine biosensor. A Sequence analysis of promoters. Sequence analysis of PcysK, PcysJ, PcysD, and PcysP promoters, highlighting the binding site of CysB protein. B Fluorescence values of the sensor at 0, 10, 20, and 30 g/L l-threonine concentration after deletion of the binding site of CysB protein in promoters PcysD, PcysJ, PcysP, and PcysK. C Schematic diagram of the pSensor biosensor structure. D Characterization of the performance of biosensors pSensor and PcysK-egfp. The x-axis represents different concentrations of l-threonine, and the y-axis represents the change in intracellular green fluorescence intensity after induction. E Identification of the response range of the biosensor pSensor to l-threonine

Full size image

To further investigate the effect of CysB on the biosensor, the CysB protein crystals of Klebsiella aerogenes, which is highly similar to CysB of E. coli (https://doiorg.publicaciones.saludcastillayleon.es/10.2210/pdb1al3/pdb, 93.8% amino acid sequence similarity to CysBEc), were used for molecular docking with l-threonine. The binding of l-threonine to CysBKa was simulated using Autodock Vina, and the results were processed and analyzed by Pymol software. As a result, the amino acid residues T102, Q103, T149, E150, and T202 have hydrogen bonding force with l-threonine ligand (Fig. 3A). To further verify whether l- l-threonine affects PcysK expression by interacting with CysB protein, these amino acid residues were mutated to alanine. Among them, CysB T102A led to an approximately 2.4-fold increase in the response threshold of the biosensor compared to the pSensor (named the CysB T102A mutant as pSensorThr) (Fig. 3B), whereas Q103A, T149A, E150A did not show any significant changes, and T202A led to a loss of response of PcysK to concentration changes (Fig. 3C).

Fig. 3
figure 3

Development of pSensorThr biosensor. A Docking outcome between the CysBKa protein and l-threonine is illustrated. Hydrogen bonds are highlighted using yellow dotted lines. B Performance comparison of biosensors pSensor and pSensorThr. C Fluorescence values of the biosensor at concentrations of 0–4 g/L l-threonine after overexpression of the CysB mutant. D Analyzing the responsiveness of pSensorThr to endogenous l-threonine

Full size image

To validate the responsiveness of pSensorThr to endogenous l-threonine, we introduced the biosensor into two E. coli strains: one strain capable of producing a high yield of l-threonine (named THR36-L19, accumulating 2.8 g/L of l-threonine during fermentation in 24-well plates containing 10 g/L glucose), and the other being the non-producing E. coli THR01 strain [33]. The resulting strains were named THR36-L19-pSensorThr and THR01-pSensorThr. When incubated in 24-well plates containing fermentation medium, THR36-L19-pSensorThr exhibited a linear fluorescence response, whereas THR01-pSensorThr showed no significant change (Fig. 3D, Fig. S2). This suggested that the pSensorThr could effectively detect the endogenous l-threonine synthesis. In comparison with existing l-threonine biosensors, such as the one developed by Liu et al. [30], which fused the PcysJH promoter to drive the expression of the eGFP reporter gene, and the Thr l-Ptrc l-threonine biosensor constructed by Su et al. [34], the pSensorThr biosensor system exhibits a higher fluorescence response threshold. This broader fluorescence range helps reduce the risk of misjudgment caused by fluorescence errors during screening. However, the l-threonine concentration response threshold is lower than those of the biosensors developed by Liu (0–50 g/L) and Su (0–20 g/L). Therefore, further optimization of the response range of pSensorThr to l-threonine in future studies will enhance the applicability of this biosensor.

Assisting high-throughput screening to identify l-threonine overproducers

Mutation library construction with a broad mutational spectrum is essential for screening superior mutants. To construct a large mutation library of THR36-L19 strain, the temperature-sensitive MP6ts plasmid was transformed into THR36-L19 (Fig. 4A). The hosts carrying the MP6ts plasmid were induced and mutated at 30 °C, and the plasmid was eliminated by incubation at 42 °C [7]. Then, shake flask fermentation was conducted after incorporating the pSensorThr biosensor into mutant libraries. Highly fluorescent cell populations were then screened using fluorescence-activated cell sorting (FACS) (Fig. 4B). Then, Qpix screening system was used for further high-throughput sorting of single cells from highly fluorescent groups. The top 10 strains with the highest fluorescence values were selected for the next round of mutation and screening. After five rounds of screening, 50 mutant strains were obtained. Subsequently, the productivity of each strain was evaluated through shake flask culture. Most demonstrated higher l-threonine accumulation compared to THR36-L19. Ultimately, the mutant E. coli strain THRM1 was successfully identified, accumulating 38.97 g/L l-threonine in shake flask culture (Fig. 4C).

Fig. 4
figure 4

Application of pSensorThr biosensor. A Flowchart for the selection of l-threonine overproducers by random mutagenesis and two-step high-throughput screening. Transformation of MP6ts plasmid is introduced into competent cells of the THR36-L19 strain, creating a random mutation library. The pSensorThr plasmid is then introduced into the library to screen highly fluorescent cell populations using Fluorescence-Activated Cell Sorting (FACS). The Qpix system is employed to select single target clones from the cell population. Mutants with high fluorescence levels undergo multiple rounds of mutagenesis and are tested for fermentation performance in shake flasks and 5 L bioreactors. B FACS screening of mutant library. Fluorescence-activated cell sorting (FACS) was performed to select cells with higher fluorescence from the mutant library with pSensorThr installation. C l-threonine production in mutants. l-threonine production in 50 highly fluorescent mutants was obtained by multiple rounds of random mutagenesis and high-throughput screening. THR36-L19 served as a control. l-threonine production was determined by HPLC, and data represent means and standard deviations of independent triplicates

Full size image

Multi-omics-guided reverse metabolic engineering further enhances l-threonine production

To elucidate the reason for the higher l-threonine production in THRM1 strain, multi-omics analysis (genomics and transcriptomics) was applied to clarify the genetic mechanism of the mutant [19, 29]. Genomic analysis through de novo sequencing revealed four missense mutations (Tpx C61G, DeoB C350Y, TrmC A129P, SpoT R290H) and two nonsense mutations (cadAA400T, yeeIT188A) in the coding sequences. To verify the effect of these mutants on l-threonine synthesis, these mutations were individually integrated into the original locus of the parental strain THR36-L19. The recombinant THR36-L19 (Tpx C61G) showed a 3.12% increase in l-threonine production, while THR36-L19 (SpoT R290H) exhibited a 7.41% decrease. In addition, knocking out the tpx gene in THR36-L19 led to a 2.91% increase in l-threonine production, whereas knocking out the spoT gene severely inhibited both strain growth and l-threonine production (Fig. 5A). Subsequently, the Tpx G61C and SpoT H290R mutants were individually integrated into the native loci of the THRM1 strain, resulting in THRM2 and THRM3. The l-threonine yields of these strains decreased by 3.45% (37.63 g/L) and increased by 3.1% (40.18 g/L), respectively (Fig. 5B). The l-threonine titer of strains integrating DeoB C350Y, TrmC A129P, cadAA400T, and yeeIT188A mutants was similar to THR36-L19 (Fig. 5C). The effect of the combined mutants on l-threonine production could be further investigated in subsequent experiments.

Fig. 5
figure 5

Reverse metabolic engineering guided by multi-omics analysis to explain l-threonine overproduction. A, C Identification of beneficial targets. Integration of Tpx C61G, DeoB C350Y, TrmC A129P, SpoT R290H, cadAA400T, and yeeI.T188A mutants into THR36-L19 strain verify their effect on l-threonine production and cell growth. B Integration of Tpx G61C and SpoT H290R into THRM1 strain verify their effect on l-threonine production and cell growth. D Transcriptomic analysis. The log2 calculation of fragments per kilo bases per million reads for THRM1/THR36-L19 is shown. The red color indicates transcriptional up-regulation, while the green color indicates transcriptional down-regulation compared to THR36-L19

Full size image

Transcriptomic analysis of THRM1 and THR36-L19 strains revealed upregulation of genes involved in the EMP pathway and glyoxylate bypass were upregulated, while most genes in the TCA cycle were downregulated (Fig. 5D). This suggested faster synthesis of l-threonine from glucose in the THRM1 strain. Among these pathways, the expression level of the pps gene encoding phosphoenolpyruvate synthase (PPS) increased by 4.89-fold, facilitating PEP replenishment. Upregulation of pykF, poxB, and pflB genes were observed, potentially reducing carbon fluxes into the targeted metabolic pathways. By sequentially knocking out the pykF, poxB, and pflB genes in the THRM3 strain, the THRM4, THRM5, and THRM6 strains were generated, respectively. Among them, THRM6 achieved an l-threonine yield of 41.36 g/L (Fig. S2).

Genomics analysis and reverse metabolic engineering revealed that the Tpx C61G mutant and SpoT H290R revertant played pivotal roles in enhancing l-threonine production in THRM1. RNA-seq further unveiled significant metabolic differences between THRM1 strain and its parental strains. The up-regulation of the EMP pathway and the glyoxylate cycle, coupled with the down-regulation of the TCA cycle, indicated that THRM1 efficiently utilized glucose and produced l-threonine through a more optimal pathway. In addition, transcriptomic analysis revealed that up-regulation of the expression of pykF, poxB, and pflB genes resulted in leakage of some carbon sources. Although we identified new engineered targets to promote l-threonine synthesis, the regulatory mechanism leading to the increased yield was not resolved. An in-depth study of the genetic differences between strain THRM1 and THR36-L19 could be instrumental in further enhancing l-threonine production.

Maximizing l-threonine production by optimizing metabolic flux through in silico simulation

The complexity of metabolic target screening presents challenges in metabolic engineering [35]. In silico simulation of intracellular metabolic flux allocation and optimization using GSMN is an emerging approach in metabolic engineering [36]. Among these simulation methods, optKnock is a computational technique designed to predict knockout genes that balance cell growth and maximize product production [21].

To identify new metabolic targets for enhancing l-threonine production, the GSMN model iML1515 of E. coli guided the metabolic engineering of the THRM6 strain [37]. Initially, the reaction equations of PYCRe enzyme (ATP + pyruvate + HCO3− + H+  → ADP + phosphate + oxaloacetate) was introduced into the iML1515 model, and the metabolic response flux boundaries for the tdcC, tdh, sstT, ilvA, poxB, pflB, pykF, and ldhA genes were set to 0 for simulating gene knockout. FBA was used to simulate the allocation of intracellular carbon fluxes under conditions that maximized l-threonine biosynthesis (Fig. 6A) [36]. In silico simulation indicated an upregulation of glutamate dehydrogenase (encoded by the gdhA gene) when maximizing l-threonine production, suggesting the amino donor provided by l-glutamate was a crucial limiting factor for l-threonine synthesis. To ensure rational expression of GdhA, a reporter system for indirect quantification of gdhA gene expression level was developed by ligating the first 210 bp of the gdhA gene to eGFP (Fig. 6B). Subsequently, a ribosome-binding site (RBS) library of gdhA gene was generated, resulting in RBS mutations that increased reporter gene expression by 1.2–3.5 folds compared to the original Ptrc-RBS combination (Fig. 6C). Substituting the native promoter of the gdhA gene with different Ptrc-RBS combinations, resulting an increase and then a decrease in l-threonine production with the GdhA expression level. In addition, it was observed that 1.2 g/L and 2.3 g/L of l-glutamate were accumulated in the RBS4 and RBS5 mutants, respectively, with no significant change in biomass (Fig. S4). THRM7 strain, overexpressing gdhA with the promoter Ptrc-RBS3, resulted in 44.12 g/L l-threonine. This indicated that it was essential to carefully regulate GdhA expression to avoid limiting l-threonine synthesis due to insufficient l-glutamate supply, while also preventing a decrease in l-threonine production due to excessive l-glutamate accumulation.

Fig. 6
figure 6

Predicting new metabolic engineering targets in silico. A Flux balance analysis and prediction of knockout targets. Flux balance analysis and prediction of new knockout targets using an expanded GSMN. B RBS library construction and screening. The first 210 bp of the target gene (gdhA) is fused with the egfp gene using a flexible linker (GGGGS)3 and inserted behind the Ptrc-RBS promoter. RBS random mutation libraries are constructed using primers containing degenerate bases. Transformants are cultured in 96-well plates for further screening based on green fluorescence. C Fluorescence analysis of Ptrc-RBS mutants. Analysis of fluorescence in Ptrc-RBS mutants using the gdhA-egfp fusion gene as a reporter gene. The effect of replacing the native promoter of gdhA gene with Ptrc-RBS mutants on l-threonine production is assessed. D Effect of optKnock-predicted target gene knockout. The effect of optKnock-predicted target gene knockout on l-threonine synthesis and cell growth is evaluated. E Fermentation parameters in a 5 L bioreactor. Fermentation parameters of the THRM13 strain in a 5 L bioreactor are presented for further analysis

Full size image

Subsequently, in silico five engineered targets were predicted that needed to be knocked out under conditions of maximized l-threonine production by combining the extended GSMN with the OptKnock software. These five genes (ydbk, ompN, aroL, phoA, puuE) were individually knockout in THRM7 strain to produce THRM8 (Δydbk), THRM9 (ΔompN), THRM10 (ΔaroL), THRM11 (ΔpuuE), THRM12 (ΔphoA). As shown in l-threonine production increased by 1.63% and 2.17% when the ompN and phoA genes were knockout, respectively. By further deleting the ompN gene in the THRM12 strain, the THRM13 strain was generated, achieving an l-threonine yield of 46.02 g/L. Finally, the fermentation performance of the THRM13 strain was validated through fed-batch fermentation in a 5 L bioreactor, achieving an accumulation of 163.2 g/L, 0.603 g/g glucose (Fig. 6E). This is the highest level of l-threonine production reported so far in E. coli. These findings advance our understanding and offer novel strategies for enhancing aspartic acid family derivative production.

Conclusions

In this study, a highly sensitive and high fluorescence threshold biosensor was developed to assist the high-throughput platform to screen for superior mutants. Subsequently, cellular metabolic flow distribution was optimized and l-threonine production was further enhanced by multi-omics analysis guided reverse metabolic engineering and in silico simulations. Ultimately, engineered strain, THRM13, demonstrated performance by producing 163.2 g/L of l-threonine in a 5 L bioreactor, with a yield of 60.3%. This represents the highest reported l-threonine production without the use of exogenous inducers and antibiotics. In addition, this high-throughput screening strategy could still be used for subsequent iterative evolution of strains.

Availability of data and materials

Data is provided within the manuscript or supplementary information files.

References

  1. Lee KH, Park JH, Kim TY, Kim HU, Lee SY. Systems metabolic engineering of Escherichia coli for L-threonine production. Mol Syst Biol. 2007;3:149.

    CAS  PubMed  PubMed Central  Google Scholar 

  2. Liu J, Li H, Xiong H, Xie X, Chen N, Zhao G, Caiyin Q, Zhu H, Qiao J. Two-stage carbon distribution and cofactor generation for improving L-threonine production of Escherichia coli. Biotechnol Bioeng. 2019;116(1):110–20.

    CAS  PubMed  Google Scholar 

  3. Xu JZ, Ruan HZ, Yu HB, Liu LM, Zhang W. Metabolic engineering of carbohydrate metabolism systems in Corynebacterium glutamicum for improving the efficiency of L-lysine production from mixed sugar. Microb Cell Fact. 2020;19(1):39.

    CAS  PubMed  PubMed Central  Google Scholar 

  4. Choi KR, Jang WD, Yang D, Cho JS, Park D, Lee SY. Systems metabolic engineering strategies: Integrating systems and synthetic biology with metabolic engineering. Trends Biotechnol. 2019;37(8):817–37.

    CAS  PubMed  Google Scholar 

  5. Lee SY, Kim HU, Chae TU, Cho JS, Kim JW, Shin JH, Kim DI, Ko Y-S, Jang WD, Jang Y-S. A comprehensive metabolic map for production of bio-based chemicals. Nat Catal. 2019;2(1):18–33.

    CAS  Google Scholar 

  6. Park SH, Kim HU, Kim TY, Park JS, Kim SS, Lee SY. Metabolic engineering of Corynebacterium glutamicum for L-arginine production. Nat Commun. 2014;5:4618.

    CAS  PubMed  Google Scholar 

  7. Long M, Xu M, Ma Z, Pan X, Rao Z. Significantly enhancing production of trans-4-hydroxy-L-proline by integrated system engineering in Escherichia coli. Sci Adv. 2020;6(21):eaba2383.

    CAS  PubMed  PubMed Central  Google Scholar 

  8. Lin JL, Wagner JM, Alper HS. Enabling tools for high-throughput detection of metabolites: metabolic engineering and directed evolution applications. Biotechnol Adv. 2017;35(8):950–70.

    CAS  PubMed  Google Scholar 

  9. Zhou LB, Zeng AP. Exploring lysine riboswitch for metabolic flux control and improvement of L-lysine synthesis in Corynebacterium glutamicum. ACS Synth Biol. 2015;4(6):729–34.

    CAS  PubMed  Google Scholar 

  10. Liu J, Xu JZ, Rao ZM, Zhang WG. An enzymatic colorimetric whole-cell biosensor for high-throughput identification of lysine overproducers. Biosens Bioelectron. 2022;216: 114681.

    CAS  PubMed  Google Scholar 

  11. Zhang Y, Cortez JD, Hammer SK, Carrasco-Lopez C, Garcia Echauri SA, Wiggins JB, Wang W, Avalos JL. Biosensor for branched-chain amino acid metabolism in yeast and applications in isobutanol and isopentanol production. Nat Commun. 2022;13(1):270.

    CAS  PubMed  PubMed Central  Google Scholar 

  12. Zhang F, Carothers JM, Keasling JD. Design of a dynamic sensor-regulator system for production of chemicals and fuels derived from fatty acids. Nat Biotechnol. 2012;30(4):354–9.

    CAS  PubMed  Google Scholar 

  13. Wu Y, Chen T, Liu Y, Tian R, Lv X, Li J, Du G, Chen J, Ledesma-Amaro R, Liu L. Design of a programmable biosensor-CRISPRi genetic circuits for dynamic and autonomous dual-control of metabolic flux in Bacillus subtilis. Nucleic Acids Res. 2020;48(2):996–1009.

    CAS  PubMed  Google Scholar 

  14. Schallmey M, Frunzke J, Eggeling L, Marienhagen J. Looking for the pick of the bunch: high-throughput screening of producing microorganisms with biosensors. Curr Opin Biotechnol. 2014;26:148–54.

    CAS  PubMed  Google Scholar 

  15. Mahr R, Frunzke J. Transcription factor-based biosensors in biotechnology: current state and future prospects. Appl Microbiol Biotechnol. 2016;100(1):79–90.

    CAS  PubMed  Google Scholar 

  16. Wu Y, Li Y, Jin K, Zhang L, Li J, Liu Y, Du G, Lv X, Chen J, Ledesma-Amaro R, et al. CRISPR-dCas12a-mediated genetic circuit cascades for multiplexed pathway optimization. Nat Chem Biol. 2023;19(3):367–77.

    CAS  PubMed  Google Scholar 

  17. Badran AH, Liu DR. Development of potent in vivo mutagenesis plasmids with broad mutational spectra. Nat Commun. 2015;6:8425.

    CAS  PubMed  Google Scholar 

  18. Zheng B, Ma X, Wang N, Ding T, Guo L, Zhang X, Yang Y, Li C, Huo YX. Utilization of rare codon-rich markers for screening amino acid overproducers. Nat Commun. 2018;9(1):3616.

    PubMed  PubMed Central  Google Scholar 

  19. Zhao Z, Cai M, Liu Y, Hu M, Yang F, Zhu R, Xu M, Rao Z. Genomics and transcriptomics-guided metabolic engineering Corynebacterium glutamicum for L-arginine production. Bioresour Technol. 2022;364: 128054.

    CAS  PubMed  Google Scholar 

  20. Xu Y, Wang X, Zhang C, Zhou X, Xu X, Han L, Lv X, Liu Y, Liu S, Li J, et al. De novo biosynthesis of rubusoside and rebaudiosides in engineered yeasts. Nat Commun. 2022;13(1):3040.

    CAS  PubMed  PubMed Central  Google Scholar 

  21. Burgard AP, Pharkya P, Maranas CD. Optknock: a bilevel programming framework for identifying gene knockout strategies for microbial strain optimization. Biotechnol Bioeng. 2003;84(6):647–57.

    CAS  PubMed  Google Scholar 

  22. Heirendt L, Arreckx S, Pfau T, Mendoza SN, Richelle A, Heinken A, Haraldsdóttir HS, Wachowiak J, Keating SM, Vlasov V, et al. Creation and analysis of biochemical constraint-based models using the COBRA Toolbox v.3.0. Nat Protoc. 2019;14(3):639–702.

    CAS  PubMed  PubMed Central  Google Scholar 

  23. Zhao Z, You J, Shi X, Zhu R, Yang F, Xu M, Shao M, Zhang R, Zhao Y, Rao Z. Engineering Escherichia coli for L-threonine hyperproduction based on multidimensional optimization strategies. J Agric Food Chem. 2024;72(41):22682–91.

    CAS  Google Scholar 

  24. Lee SY, Kim HU. Systems strategies for developing industrial microbial strains. Nat Biotechnol. 2015;33(10):1061–72.

    CAS  PubMed  Google Scholar 

  25. Li Y, Wei H, Wang T, Xu Q, Xie X. Current status on metabolic engineering for the production of L -aspartate family amino acids and derivatives. Bioresour Technol. 2017;245:1588–602.

    CAS  PubMed  Google Scholar 

  26. Lv Y, Qian S, Du G, Chen J, Zhou J, Xu P. Coupling feedback genetic circuits with growth phenotype for dynamic population control and intelligent bioproduction. Metab Eng. 2019;54:109–16.

    CAS  PubMed  Google Scholar 

  27. Uchiyama T, Miyazaki K. Product-induced gene expression, a product-responsive reporter assay used to screen metagenomic libraries for enzyme-encoding genes. Appl Environ Microbiol. 2010;76(21):7029–35.

    CAS  PubMed  PubMed Central  Google Scholar 

  28. Zhang J, Jensen MK, Keasling JD. Development of biosensors and their application in metabolic engineering. Curr Opin Chem Biol. 2015;28:1–8.

    PubMed  Google Scholar 

  29. Jiang S, Wang R, Wang D, Zhao C, Ma Q, Wu H, Xie X. Metabolic reprogramming and biosensor-assisted mutagenesis screening for high-level production of L-arginine in Escherichia coli. Metab Eng. 2023;76:146–57.

    CAS  PubMed  Google Scholar 

  30. Liu Y, Li Q, Ping Z, Zhang Z, Liu Y, Sun C, Cao G, Zhou W, Wang X, Zhang D. Developing a high-throughput screening method for threonine overproduction based on an artificial promoter. Microb Cell Fact. 2015;14:121.

    PubMed  PubMed Central  Google Scholar 

  31. Kredich NM. The molecular basis for positive regulation of cys promoters in Salmonella typhimurium and Escherichia coli. Mol Microbiol. 2010;6(19):2747–53.

    Google Scholar 

  32. Malo MS, Loughlin RE. Promoter elements and regulation of expression of the cysD gene of Escherichia coli K-12. Gene. 1990;87(1):127–31.

    CAS  PubMed  Google Scholar 

  33. Zhao Z, You J, Shi X, Cai M, Zhu R, Yang F, Xu M, Shao M, Zhang R, Zhao Y, et al. Multi-module engineering to guide the development of an efficient L-threonine-producing cell factory. Bioresour Technol. 2025;416: 131802.

    CAS  PubMed  Google Scholar 

  34. Su B, Lai P, Deng M-R, Zhu H. Design of a dual-responding genetic circuit for high-throughput identification of L-threonine-overproducing Escherichia coli. Bioresour Technol. 2024;395: 130407.

    CAS  PubMed  Google Scholar 

  35. Somerville V, Grigaitis P, Battjes J, Moro F, Teusink B. Use and limitations of genome-scale metabolic models in food microbiology. Curr Opin Food Sci. 2022;43:225–31.

    CAS  Google Scholar 

  36. Orth JD, Thiele I, Palsson BØ. What is flux balance analysis? Nat Biotechnol. 2010;28(3):245–8.

    CAS  PubMed  PubMed Central  Google Scholar 

  37. Monk JM, Lloyd CJ, Brunk E, Mih N, Sastry A, King Z, Takeuchi R, Nomura W, Zhang Z, Mori H, et al. iML1515, a knowledgebase that computes Escherichia coli traits. Nat Biotechnol. 2017;35(10):904–8.

    CAS  PubMed  PubMed Central  Google Scholar 

Download references

Funding

The National Key Research and Development Program of China (2021YFC2100900). The National Natural Science Foundation of China (32071470). The National First-class Discipline Program of Light Industry Technology and Engineering (QGJC20230201). The Major Scientific and Technological Project for “unveiling and commanding” of Hohhot (2023-unveiling and commanding-He-1). The Open Funding Project of Key Laboratory of Industrial Biotechnology, Ministry of Education, Jiangnan University (KLIB-KF202102). The Program of Introducing Talents of Discipline to Universities (111-2-06) and Postgraduate Research & Practice Innovation Program of Jiangsu Provence (KYCX22-2369).

Author information

Author notes

  1. Zhenqiang Zhao and Rongshuai Zhu contributed equally to this work.

Authors and Affiliations

  1. Key Laboratory of Industrial Biotechnology, Ministry of Education, School of Biotechnology, Jiangnan University, Wuxi, 214122, Jiangsu, China

    Zhenqiang Zhao, Rongshuai Zhu, Xuanping Shi, Fengyu Yang, Meijuan Xu, Minglong Shao, Rongzhen Zhang, Jiajia You & Zhiming Rao

  2. Institute of Future Food Technology, JITRI, Yixing, 214200, China

    Zhenqiang Zhao, Rongshuai Zhu, Xuanping Shi, Fengyu Yang, Meijuan Xu, Minglong Shao, Jiajia You & Zhiming Rao

  3. College of Biochemical Engineering, Beijing Union University, Beijing, 100023, China

    Youxi Zhao

Authors
  1. Zhenqiang Zhao
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Rongshuai Zhu
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Xuanping Shi
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Fengyu Yang
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Meijuan Xu
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Minglong Shao
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Rongzhen Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. Youxi Zhao
    View author publications

    You can also search for this author inPubMed Google Scholar

  9. Jiajia You
    View author publications

    You can also search for this author inPubMed Google Scholar

  10. Zhiming Rao
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

Z.Q.Z.: investigation, visualization, writing—original draft, funding acquisition. R.S.Z.: investigation, visualization. X.P.S.: investigation. F.Y.Y.: visualization. M.J.X.: conceptualization, visualization. M.L.S.: visualization. R.Z.Z.: investigation, visualization. Y.X.Z.: visualization. J.J.Y.: investigation, visualization. Z.M.R.: conceptualization, funding acquisition.

Corresponding authors

Correspondence to Jiajia You or Zhiming Rao.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

All authors approved the final manuscript and the submission to the journal.

Competing interests

The authors declare no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhao, Z., Zhu, R., Shi, X. et al. Combining biosensor and metabolic network optimization strategies for enhanced l-threonine production in Escherichia coli. Biotechnol. Biofuels Bioprod. 18, 37 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13068-025-02640-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13068-025-02640-7

Keywords

Biotechnology for Biofuels and Bioproducts

ISSN: 2731-3654

Contact us