The strains and plasmids used in this study are listed in Supplementary Tables S1, S2. E. coli DH5α was used for molecular cloning and manipulation of plasmids. All primers used are listed in Supplementary Table S3. All plasmids were constructed using the Gibson assembly method (Gibson et al., 2009). E. coli K-12 MG1655 was used for RBS strength characterization. RBS (B0029, B0030, B0031, B0032, B0033, B0034, B0035, B0064 and J61101) of different strengths were analyzed by the ratio of RFP to OD₆₀₀. Nine plasmids were assembled. The expression of thrAB and thrC was regulated by using the primers listed in Supplementary Table S3, which replaced various RBS to construct L-threonine production plasmids (PA-29thrAB-29thrC, PA-29thrAB-01thrC, PA-29thrAB-35thrC, PA-01thrAB-29thrC, PA-01thrAB-01thrC, PA-01thrAB-35thrC, PA-35thrAB-29thrC, PA-35thrAB-01thrC and PA-35thrAB-35thrC). The TH strain producing L-threonine was provided by the Fufeng Group (Qingdao, China) and derived from E. coli K-12 MG1655. Three genes thrA, thrB and thrC increase from one copy number to four copies in the genome. Comparison of partial genes of strains TH and MG1655 are listed in Supplementary Table S4. The expression level of thrAB and thrC in L-threonine production was balanced by transforming the above plasmids into strain TH to generate strains 2929, 2935, 2901, 0129, 0101, 0135, 3529, 3535 and 3501. P _flic , P _1.1 and P _2.1 promoter sequences were derived from reports (Jaishankar and Srivastava, 2020; Zhang et al., 2021). The ptsG gene deletion experiment was carried out by homologous recombination (Datsenko and Wanner, 2000). All fragments were amplified with Phanta HS ultra-fidelity DNA polymerase (Vazyme Biotech, Nanjing, China).

RBS characterization was achieved by monitoring the RFP fluorescence and cell density in real-time using a multiple detection microplate analyzer (SynergyHT, BioTek, Winooski, VT, United States). The details are as follows. The seeds of RBS characterization strains were prepared by transferring a single colony to a 12-well microassay plate containing 2 mL LB medium with 34 μg/mL chloramphenicol. The cells were grown at 37°C for 12 h. Next, 2% (v/v) of the seeds were inoculated into 0.2 mL LB medium with 34 μg/mL chloramphenicol in a 96-well microassay plate to detect red fluorescence. The 96-well plate was incubated at 37°C with oscillation. During RBS characterization, strain growth was measured at 600 nm. The red fluorescence was detected by excitation at 590 nm and emission at 645 nm.

LB medium was used for plasmid construction and RBS strength characterization. L-threonine fermentation medium consists of 15 g/L (NH₄)₂SO₄, 2 g/L KH₂PO₄, 1 g/L MgSO₄·7H₂O, 2 g/L yeast extract and 0.02 g/L FeSO4 (Kruse et al., 2002). Glucose (40 g/L) was added as the initial carbon source and 20 g/L CaCO₃ was used to adjust the pH during fermentation. For shake flask fermentation, a single colony was incubated in fresh LB medium at 37°C for 12 h. The precultured seeds were then transferred with 1% (v/v) inoculation to a 300 mL shake flask that contained 20 mL fermentation medium. Fermentation was carried out at 220 rpm and 37°C. Cultures were supplemented with 40 g/L glucose when the glucose level was lower than 15 g/L.

The fed-batch culture was carried out in a 5-L bioreactor containing 4L medium (20 g/L (NH₄)₂SO₄, 3 g/L yeast extract, 2 g/L KH₂PO₄, 2 g/L MgSO₄·7H₂O, 5 mg/L FeSO₄·7H₂O, 5 mg/L MnSO₄·4H₂O, 0.5 g/L betaine). Temperature was maintained at 37°C, the aeration rate at 1.5 vvm, pH was maintained automatically at 7.0 with NH₄OH, and the dissolved oxygen value was maintained below 30%. 20 g/L initial amount of sterilized glucose was added in the working culture, and glucose concentration was controlled by continuous feeding.

One milliliter of the culture was mixed vigorously, and 0.1 mL was transferred to a 1.5 mL centrifuge tube. One millimolar HCl (0.9 mL) was added to this culture, and the sample was mixed to remove residual CaCO₃. Subsequently, the OD₆₀₀ was measured using a spectrophotometer (Shimadzu, Kyoto, Japan). For glucose and acetate assays, the cultures were centrifuged at 12,000 rpm for 2 min to collect the supernatant. The collected supernatant was filtered through a 0.22 μm water membrane for analysis. A refractive index detector (RID-10A; Shimadzu, Kyoto, Japan) and an AminexHPX-87H ion exclusion column (Bio-Rad Laboratories, Hercules, CA, United States) were used with 5 mM H₂SO₄ as the mobile phase and a flow rate of 0.6 mL/min.

For the detection of L-threonine, the collected supernatant was deproteinized with 5% trichloroacetic acid. Subsequently, the pretreated supernatant was derivatized with triethylamine and phenyl isothiocyanate, followed by extraction with n-hexane (Wang et al., 2022a). Briefly, 0.2 mL of sample and standard L-threonine were pretreated with a mixture of triethylamine-acetonitrile (1.4 mL of triethylamine mixed with 8.6 mL of acetonitrile). Next, we added phenylisothiocyanate-acetonitrile (25 μL of phenylisothiocyanate mixed with 2 mL of acetonitrile) to pretreat samples and the L-threonine standard for 1 h at room temperature. n-Hexane (0.4 mL) was added, and the sample was shaken vigorously. The lower layer (0.2 mL) was collected and diluted with 0.8 mL deionized water. The solution was filtered with a 0.22 μm organic membrane, and samples were detected using an HPLC equipped with a diode array detector (SPD-M20A; Shimadzu, Kyoto, Japan) and a VenusilAA (4.6 × 250 mm, 5 μm, AgelaTechanology) column at 40°C. The mobile phase consisted of (A) 15.2 g sodium acetate dissolved in 1850 mL of ultrapure water and mixed with 140 mL of acetonitrile and (B) 80% (v/v) acetonitrile and 20% (v/v) ultrapure water. The flow rate was 1 mL/min. The L-threonine concentration was quantified using the corresponding standard curve and peak area.

In the biosynthesis of L-threonine in E. coli, genes thrA, thrB and thrC encoding the last three key enzymes of this biosynthesis are located on the thrABC operon. ThrC catalyzes the final step of L-threonine synthesis, and this reaction is slow (Lee et al., 2013). As the rate-limiting step of L-threonine production, it is necessary to enhanced the expression of thrC. We used the L-threonine-producing strain TH available from FuFeng Group as the initial strain, which was derived from MG1655 (Supplementary Table S2). Initially, we enhanced the expression of thrC on the low copy PCL-1920 plasmid and transformed the plasmid into the initial strain TH to obtain the pcl-thrC strain. However, the production of L-threonine and cell growth of the strain was not significantly improved compared with the original strain TH (Supplementary Figure S1A,B). We analyzed that the reason is the weak degree of enhanced expression. Next, we enhanced the expression of thrC on the medium copy PACYC-Duet plasmid and transformed the plasmid into the initial strain TH to obtain the PA-thrC strain. Although the production of L-threonine increased, cell growth and the glucose consumption rate were seriously impeded during early cultivation (Figures 2A–C). In addition, thrA, thrB, thrC are located on the thrABC operon. So the overexpressed thrC might disrupt the balance between thrAB and thrC. Therefore, we analyzed the influence of various ratios of thrAB and thrC on L-threonine synthesis and cell growth. Ribosome binding sites (RBS) define the translation efficiency of genes and are typically used in gene regulation studies (Hur et al., 2020; Wang et al., 2022b). We selected three RBS with different intensities by fluorescence characterization, RBS0035 (2615), J61101 (958) and RBS0029 (585), to regulate thrAB and thrC expression levels, with an intensity ratio between them of 13:5:3 (Figure 2D). We constructed nine strains containing various combinations and levels of thrAB and thrC expression.

The results showed that different combinations of thrAB and thrC expression levels had different effects on cell growth. High expression of thrC promoted growth and L-threonine accumulation simultaneously; however, there were also ratios between thrAB and thrC when the level of L-threonine decreased because the proportion of thrC was too high. Among the nine strains constructed, the 2901 strain (thrAB:thrC = 3:5) yielded the highest L-threonine titer. The 2935 strain (3:13) grew slowly because of the large difference in the expression levels of thrAB and thrC. Optimal growth was observed for the 0135 strain (5:13) (Figure 3A; Figure 4). Accumulation of L-threonine was poor when thrC was expressed weakly, and different ratios had minimal effect on L-threonine production (Figure 3B, Figure 4). Equal expression levels of thrAB and thrC in strain 3535 yielded a relatively good L-threonine titer, but production was lower than that of strain 2901 (Figure 3C, Figure 4). The results indicated that high expression of thrC in an appropriate thrAB:thrC ratio yielded optimal cell growth and L-threonine production.

In summary, the results showed that when the expression of thrC was higher than thrAB, it was beneficial to the growth and production of L-threonine, and the level of L-threonine decreased when the expression of thrC was much higher than thrAB in particular strains. Using a thrAB:thrC expression ratio of 3:5 yielded the highest production of L-threonine. Production of L-threonine by proportional expression of thrAB and thrC was found to be a more effective approach to produce L-threonine than direct overexpression of thrABC (Zhao et al., 2018).

As described above, L-threonine production was promoted by carefully regulating the expression levels of thrAB and thrC and optimizing the thrAB:thrC ratio; however, constitutive overexpression of the thrABC operon inhibited cell growth (Figures 3A–C). We adopted a dynamic regulation strategy to ensure that the metabolic burden caused by the premature introduction of the thrABC operon was alleviated. Instead of using IPTG and other chemical inducers, a growth-related promoter was used to control thrABC expression after obtaining a high cell density. Several stationary phase promoters with different strengths have been obtained previously by random mutagenesis of a wild-type stationary phase promoter (Shimada et al., 2004). The strength of the P _1.1 and P _2.1 promoters obtained was equivalent to E. coli promoters (Jaishankar and Srivastava, 2020). The promoters of fliA, fliC and flgC involved in flagella construction were also identified as stationary phase promoters. The transcriptional intensity of these promoters is active during the lag and early exponential phases but inhibited throughout the late exponential and stationary phases (Zhang et al., 2019).

The effect of dynamic regulation of thrABC on strain growth and production was verified by using three stationary phase promoters, P _fliC , P _1.1 and P _2.1 , for thrABC regulation in the PA-29thrAB-01thrC plasmid (Figure 5A). The results showed that the strains controlled by the three stationary phase promoters displayed improved growth in the early stage, and L-threonine production increased significantly compared with the control group. Among these strains, growth was strongest for strain P _2.1 -2901, which was controlled by P _2.1 , and L-threonine production was similar to that of the strain controlled by the tac promoter (Figures 5B,C). This observation is because the P _2.1 promoter is the strongest among the three tested (Jaishankar and Srivastava, 2020). We measured L-threonine production in each period and found that L-threonine accumulation in the P _2.1 -2901 strain was the fastest, with the L-threonine titer and productivity reaching 26.02 g/L and 1.08 g/L/h at 24 h, respectively. L-threonine production by strain P _2.1 -2901 was 44.56% and 65.00% higher than that of the control strain TH and strain 2901 controlled by the tac promoter, respectively (Figure 5C). This result arises because the replaced stationary phase promoter is more active during the early exponential phase, which promotes the overexpression of the thrABC operon to accelerate the accumulation of L-threonine. This phenomenon suggests that this strategy can enhance product titer and shorten the fermentation period. The overexpression of the thrABC gene controlled by the stationary phase promoter can improve L-threonine yield (Figure 5D).

Although we increased L-threonine productivity by using a stationary phase promoter, the strains with better growth in the early stage also produced a large amount of acetate (Figure 5E). Studies have suggested that the rapid use of glucose during the early stage of cell growth will lead to an acetate overflow (Eiteman and Altman, 2006). Thus, we next tried to reduce the acetate overflow.

The phosphotransferase system (PTS) is the major glucose transport system in E. coli. Deletion of PTS has been reported to reduce acetate overflow and increase the production of recombinant proteins and biochemicals (Wang et al., 2006; Kang et al., 2009). We deleted the ptsG gene in three threonine-producing strains, TH, 2901 and P _2.1 -2901, to examine the effect of PTS activity on L-threonine production.

The growth of the three strains was retarded after deleting ptsG, which was caused by the decrease in the glucose uptake rate by the cells. The growth of the 2901ΔptsG strain was inhibited significantly because of the combined effects of thrABC overexpression and ptsG deletion. Optimal growth was observed for strain P _2.1 -2901ΔptsG, reaching a stable growth state after 36 h, whereas the 2901ΔptsG strain showed slower growth up to 60 h (Figures 6A,B). This observation also reflects the advantages of dynamic regulation. Using the P _2.1 -2901ΔptsG strain can shorten the fermentation period and increase L-threonine production. L-threonine production by the three strains was improved when compared with the corresponding strains without the ptsG deletion, with an increase in L-threonine yield (Figure 6C). At 60 h, L-threonine production by strain P _2.1 -2901ΔptsG reached 40.06 g/L, which was 36.30% and 7.95% higher than that of THΔptsG and 2901ΔptsG strains, respectively (Figure 6D). As expected, acetate was not generated during fermentation.

To further verify the properties of P _2.1 -2901ΔptsG for L-threonine production, fed-batch fermentation was performed in 5-L bioreactor, using TH as a control. The cell density, glucose consumption levels (Figure 7A) and L-threonine titer (Figure 7B) are depicted. Following 48 h of cultivation, P _2.1 -2901ΔptsG could produce 121.05 g/L L-threonine, leading to a yield of 0.60 g/g glucose and productivity of 2.52 g/L/h. TH could produce 98.88 g/L L-threonine in 48 h, with a yield and productivity of 0.49 g/g glucose and 2.06 g/L/h, respectively. After 48 h of incubation, unfavorable growth conditions and irreparable cellular damage lead to a death phase and long-term stationary phase (Finkel, 2006), resulting in the lose viability of L-threonine production and glucose consumption. In contrast to the results of other reports, P _2.1 -2901ΔptsG achieved the highest productivity and yield. Therefore, with respect to the three major parameters in industrial production, P _2.1 -2901ΔptsG possesses substantial L-threonine production ability.

The reaction catalyzed by ThrC is the rate-limiting step of L-threonine production and overexpression of thrC inhibits cell growth. Here, we showed that balanced regulation of thrAB and thrC expression levels promotes E. coli growth and L-threonine production. L-threonine production was further improved by dynamically regulating the overexpression of thrABC. Finally, the deletion of ptsG reduced acetate production and further increased the titer and yield of L-threonine. With the above approaches, we obtained strain P _2.1 -2901ΔptsG, with the titer of L-threonine by fermentation in a shake flask for 60 h reaching 40.06 g/L, which is 96.85% higher than that of the control strain TH (20.35 g/L) and the highest titer reported in a shake flask (Table 1). The yield reached 0.84 g/g, which was 44.82% higher than the control strain TH (0.58 g/g). P _2.1 -2901ΔptsG could produce 121.05 g/L L-threonine in 48 h, with a yield and productivity of 0.60 g/g glucose and 2.52 g/L/h by fed-batch fermentation, respectively. The maximum L-threonine yield and productivity was obtained in reported fed-batch fermentation, and L-threonine titer is close to the maximum (127.30 g/L) (Table 1). This study provided a new concept for improving L-threonine production and downstream products.