The probiotic E. coli Nissle 1917 (EcN) is used as a host for the construction of engineered probiotics. Lkr and sdh genes were amplified from Saccharomyces cerevisiae BY4741 and inserted into the pTrc99a via XhoI and NheI. The recombinant plasmid was designated pTLS. The lys5 and lys2 genes were amplified from Saccharomyces cerevisiae BY4741 and inserted into pK18mobSacB via XbaI and HindIII. The recombinant plasmid was designated pK25. Both recombinant plasmids were verified by sequencing. All strains were cultured in LB medium at 37°C at 200 rpm. Ampicillin (100 μg/mL), spectinomycin (50 μg/mL), kanamycin (50 μg/mL), IPTG (1 mM) or X-gal (0.1 g/L), and L-arabinose (20 mM) were supplemented as appropriate. All primers used are listed in Supplementary Table S1 in the Supporting Information.

CRISPR-Cas9 technology was used to knock out the ldcC1 and ldcC2 genes, which are the first enzymes in the lysine degradation pathway in EcN. The pGRB plasmids were used as the backbone to express sgRNA by the PCR method. The constructed gRNA plasmids were named ldcC1-gRNA and ldcC2-gRNA, respectively. The donor DNA for knocking out the ldcC1/2 gene was obtained by three-step PCR. First, the primers ldcC1/2-up-R/ldcC1/2-up-F and ldcC1/2-down-R/ldcC1/2-down-F were used to amplify the upstream and downstream of the ldcC1/2 gene from the EcN genome. The ldcC1/2-up-R/ldcC1/2-down-R primers were then used for overlapping amplification of the two fragments to obtain ldcC1/2 donor DNA. Finally, the donor DNAs were ligated to the vectors by blunt-end ligation reactions and verified by sequencing, respectively. The ldcC1 and ldcC2 knockout genes were applied according to Li et al. (2015). The EcN(pREDCas9) competent cells were generated by electroporation. The donor DNA and gRNA plasmid were transformed into EcN(pREDCas9). Clones were identified by colony PCR and DNA sequencing. The correct clone was inoculated into LB with spectinomycin and 10% L-arabinose and cultured overnight to lose the gRNA plasmids. Finally, the EcN of the ldcC1 gene knockout was named EcN ∆ldcC1 (EcNO). The EcNO was used in the next round for ldcC2 gene editing to obtain EcN ∆ldcC1 ∆ldcC2 (EcNT). The recombinant plasmids were transformed into EcNT and the recombinant strains were named EcNT (pTLS) and EcNT (pK25). The primers used to amplify the pGRB plasmids or donor DNA are listed in Supplementary Table S2.

The metabolic characteristics of EcNT (pTLS) and EcNT (pK25) were investigated in shaken flask cultures. The engineered EcNs were cultured as seeds in LB medium overnight. The seeds were then cultured at 37°C in the fermentation medium (10 g/L glucose, 1 g/L yeast extract, 5 g/L KH₂PO₄, 5 g/L (NH₄)₂SO₄, 0.7 g/L KCl, 0.003 g/L FeSO₄ ·7H₂O, 0.003 g/L MnSO₄, 1 g/L MgSO₄). Cultures were grown at 37°C with an initial OD₆₀₀ of 0.1 and 0.1 mM isopropyl-B-D-thiogalactoside (IPTG). Samples were taken at regular intervals for measurement of cell growth (OD₆₀₀), sugar concentration, and metabolite analysis. All fermentations were performed with three samples in parallel. Glucose concentration was determined using a glucose assay kit with the glucose oxidase method. Lysine concentration was determined by high-performance liquid chromatography (HPLC LC-2030). Sample pretreatment was performed according to the Dikma protocol of using Diamonsil AAA 5 μm (4.6 mm × 250) column and ultraviolet detector at 360 nm. Mobile phase A was 0.02 mol/ L Na₂HPO₄ and 0.02 mol/ L NaH₂PO₄ aqueous solution, mobile phase B was methanol: acetonitrile =10:90 (V: V). The column temperature was maintained at 45°C and the flow rate was 1 mL/min. The analysis was performed using a binary gradient elution method according to the protocol of Dikma. Growth rates were calculated using the method (OD_6h-OD_0h)/OD_0h; rates of glucose as well as lysine metabolism were calculated using the method (C_6h-C_0h)/C_0h/OD.

The EcNT strains were inoculated into LB medium, and cultured overnight at 37°C and 200 rpm for seeds. Seeds were inoculated into the fermentation medium as above with the addition of 10 g/L lysine and cultured at 37°C 200 rpm for 6 h. The activated bacterial solution was collected in a sterile 2 mL centrifuge tube, resuspended with deionized water to wash 3 times, and incubated at 10⁸ CFU/mL. 2% sodium alginate microbial colloid was dropped into a 4% CaCl₂ aqueous solution at a uniform rate and stirred on a magnetic stirrer. The microsphere was centrifuged for 10 min at 4°C at 5000 rpm, washed three times with sterile water to remove CaCl₂, and resuspended in sterile water. The microspheres are poured into the sterile medium and frozen in a lyophiliser for 12 h. After spraying with gold, the microspheres were photographed using an electron microscope. The microspheres were washed three times with sterilized water and cultured in 20 mL of fermentation medium at 37°C 150 rpm for 24 h for the stability test. Cell growth (OD600) and sugar concentration were tested as above. All samples were tested in triplicate.

To verify the cascade cocktail of engineered strains in mice, C57BL/6 mice (age, 6 weeks; weight, 18–20 g; half male and half female) were purchased from the Jinan Pengyue Experimental Animal Center (China) and fed under specific pathogen-free (SPF) conditions at the Experimental Animal Center of Binzhou Medical University. The experiment was conducted after 2 weeks of adaptive feeding. Male and female mice were divided into 4 cages with 5 mice in each cage. The mice were randomly assigned to control, EcN, EcNT, EcNT(pTLS), EcNT(pK25), and the probiotic cocktail of EcNT(pTLS) and EcNT(pK25) in a ratio of 1:2, with 10 mice in each group. The probiotic solution was mixed with the strains and inoculated into 50 mL of LB liquid medium at OD = 0.1 and cultured overnight at 37°C and 200 rpm. The next day, the cells were collected by centrifugation at 5000 rpm for 10 min, washed twice with sterilized ultrapure water, and resuspended in sterilized water to prepare a bacterial solution for probiotic administration. The probiotic groups were alternately fed with lysine solution for 1 day and the probiotic solutions for the next, while the control group was fed with lysine solution for 1 day and PBS solution for 1 day during the experiment. To maintain the activity of the probiotics, the fresh liquid of the probiotics was changed every feeding day. The body weight of the mice was recorded from the first day of administration and then every 5 days/time until the end of the experiment. Mice were sacrificed at 12 weeks of age after 6 weeks of drug-feeding. Three mice in the drug group and control groups were randomly selected and faecal; and plasma samples were collected. Fecal samples were collected with sterile forceps and 3–5 pieces of fresh mouse feces were placed in sterile cryopreservation tubes, immediately frozen with liquid nitrogen for 30 min, and then stored at −80°C. Blood was collected in heparin-sodium anticoagulant tubes at 5000 rpm for 5 min, and plasma was collected. The plasma sample was pretreated as follows: the sample was placed into a 1 mL centrifuge tube and the same volume of methanol was added for protein precipitation, centrifuged at 15,000 rpm after sonication for 2 min, the supernatant was taken and passed through a C₁₈-containing membrane and placed it in a sample bottle. The samples were analyzed by HPLC (SHIMADZU-20A) and Tandem mass spectrometry (MS/MS) (Applied Biosystems 5,500 Quadrupole Trap) using WAX column (5 cm × 4.0 mm) at 45°C and a flow rate of 0.5 mL/min. The mobile phase A was 5 mmol/L ammonium acetate 0.1% formic acid in water and B was methanol. The elution gradient was 0–2.0 min, B = 90%; 2.1–4.0 min, B = 70%; 4.1–6.0 min, B = 50%; 6.0–8.0 min, B = 30%; 8.1–10 min, B = 10%; 10.1–13.0 min, B = 90%. The mass spectrometry ion source was ESI. The condition was curtain gas (psi): 38; ionization voltage (V): positive ion 5,500; temperature (°C): 550; spray gas (psi): 55; auxiliary heating gas (psi): 55; scan mode: MRM.

The experimental data were processed and analyzed using Prism 9. The lysine metabolic efficiency was calculated following lysine metabolic efficiency = (1- Conc. lysine in treated group/ Conc. lysine in EcNT) *100%. Relative saccharopine concentrations were calculated following Relative Conc. saccharopine = Conc. saccharopine in treated group/ Conc. saccharopine in EcNT *100%. Error bars represent the coefficient of variation. All data analyses were considered statistically significant at p < 0.05. Gut microbial diversity was tested at Shanghai Qrigingene Biotechnology Co., Ltd. for 16 s RNA sequencing analysis. The Illumina PE250 high-throughput sequencing platform was used for sequencing. Animal protocols were approved by the Ethics committee of the Binzhou Medical University No. 2021049 and all experiments were conducted following the National Guidelines for the Ethical Review of Laboratory Animal Welfare of the People’s Republic of China (GB/T 35892–2018).

Hyperlysinemia, which involves high levels of amino acid lysine in the blood, is caused by the absence of the AASS enzyme. Our proposal involves introducing gut bacteria similar in activity to the AASS enzyme to improve the breakdown of lysine in the blood and urine. To investigate this hypothesis, we formulated a probiotic mixture that would metabolize lysine and its downstream metabolites sequentially (Figure 1). Since the inherent lysine metabolism in probiotic E. coli Nissle 1917 (EcN) is mediated by the cadaverine pathway, whose metabolites are highly toxic to the host cells, we initially modified the host EcN strain by using CRISPR-Cas9 technology to eliminate its LdcC1 and LdcC2 genes. Subsequently, we developed two probiotic strains that can carry out lysine metabolism with the saccharopine pathway. Given that the expression of the human-derived AASS protein in E. coli is unstable, whereas yeast has the same lysine degradation pathway as humans with better solubility and stability in bacteria, two enzymes involved in yeast lysine metabolism, lysine-ketoglutarate reductase (EC1.5.1.7, abbreviated as LKR, encoded by the lys1 gene) and saccharopine dehydrogenase (EC1.5.1.10, SDH, encoded by the lys9 gene) were selected for the production of the engineered probiotic bacteria. One EcN strain has been engineered to convent lysine to α-aminoadipic acid semialdehyde by heterologous overexpression of LKR and SDH from Saccharomyces cerevisiae. The alternative EcN strain has been genetically modified to convent α-aminoadipic acid semialdehyde to α-aminoadipic acid by overexpression of Saccharomyces cerevisiae’s aminoadipate semialdehyde dehydrogenase (EC1.2.1.3, ALDH). Finally, an engineered probiotic cocktail consisting of two strains was used to achieve balanced lysine metabolism.

We chose the probiotic EcN as the host to create an engineered probiotic cocktail utilizing two cascade-metabolic E. coli. Initially, two recombinant plasmids were developed to overexpress the AASS-like enzymes in the saccharopine pathway. A plasmid called pTLS was constructed by utilizing pTrc99a and expressing ketoglutarate reductase (LKR) and saccharopine dehydrogenase (SDH) from Saccharomyces cerevisiae BY4741 (Figures 2A,B). This plasmid was able to convert lysine into saccharopine through to saccharopine semialdehyde. The additional plasmid, pK25, was produced from pK18mob featuring kanamycin resistance and a perpetually expressed aminoadipate semialdehyde dehydrogenase (ALDH, encoded by lys2 and lys5). This facilitated the conversion of saccharopine semialdehyde to α-aminoadipic acid. Both plasmids metabolize excess lysine to α-aminoadipic acid, which is degraded via the tricarboxylic acid (TCA) cycle. Unlike mammalian cells, the lysine catabolism pathway in E. coli cells produces harmful cadaverine. To address this, we used CRISPR-Cas9 technology to remove the ldcC genes responsible for breaking down lysine in EcN. The engineered EcN strain was then designated as EcNT. Subsequently, we introduced the pTLS and pK25 plasmids into EcNT, producing two recombinant strains named EcNT (pTLS) and EcNT (pK25), respectively. These two engineered bacteria collaborate to form a cascade. EcNT (pTLS) converts lysine to α-aminoadipate semialdehyde, followed by the conversion of α-aminoadipate semialdehyde to non-toxic alpha-aminoadipic acid by EcNT (pK25). This combination of strains achieves the dual purposes of providing supplements for pathogenic AASS and balancing the dynamic cascade metabolites from lysine simultaneously.

Shaken flask experiments were conducted to explore the metabolic traits of the engineered bacteria in vitro. Initially, these bacteria must endure the anaerobic gut environment to attain the curative objective. Although the wild-type E. coli is a facultative anaerobium capable of survival in anaerobic intestinal conditions, the competence of the engineered derivative in performing within the gut environment necessitates authentication. The engineered strains’ metabolic capacity and growth were examined with and without oxygen, in which shaken flask experiments illustrated that EcN (pTrc99a) and EcN (pTLS) had comparable growth rates under both aerobic and anaerobic conditions (Figure 2C). The engineered modifications did not impede EcN’s oxygen tolerance, i.e., the engineered EcNs can thrive and metabolize in the presence or absence of oxygen. Furthermore, the ldcC1 and ldcC2 genes within EcN’s genetic makeup are responsible for the catabolism of lysine into cadaverine, which is both highly toxic and irritating. Consequently, we utilized CRISPR-Cas9 technology to remove ldcC1 and ldcC2 genes from the wild-type strain of EcN, resulting in an engineered strain named EcNT. To assess the differences in growth and metabolism between the wild-type EcN strain and the engineered EcNT strain with plasmid pTLS, we carried out a shake-flask experiment. The findings revealed that both the modified strain and wild-type strain exhibited similar rates of growth and sugar consumption. However, the EcNT(pTLS) strain depleted lysine earlier than the EcN(pTLS) strain (Figure 2D). Hence, the genetic modifications made have had no significant impact on the EcN strain’s growth and metabolism. Additionally, the recombinant plasmid exhibits consistent expression.

The engineered bacteria must cooperate in the complex gut environment to achieve therapeutic objectives. However, various bacteria colonize different locations within the gut, where spatial partitions can interrupt material exchange between the two bacteria. Consequently, we investigated an artificial microsphere system to encourage the metabolic cascade reactions between different strains that have spatial colocalizations. We refined the metabolic properties of the cascade cocktail and established an artificial microsphere system comprising EcNT (pTLS), herein referred to as pTLS, and EcNT (pK25), also known as pK25 (Figure 3A). Bacteria cultured for 6 h at a 1:2 ratio were coated with sodium alginate to produce microspheres (Figure 3B). Microsphere morphology was examined using a scanning electron microscope. The microspheres are approximately 1 mm in diameter and have a rough surface. Microsphere dissolution and metabolism were evaluated via fermentation (Figure 3C). The study demonstrated that the microbial microspheres remained intact and there was minimal bacterial liquid dissolution in the medium within 24 h, thus indicating the stability and effectiveness of the artificial ecosystem microspheres.

To investigate the growth and metabolic capacity of the engineered strains, we initially analyzed the ratio of bacteria and the concentrations of medium components in vitro. We investigated the prevalent strains of two different bacteria in a cascade cocktail and analyzed their impact on growth, sugar consumption, and lysine metabolism. To identify the dominant strain, we established the ratio of EcNT (pTLS) and EcNT (pK25) was set as 1:1, 1:3, and 3:1 while monitoring the growth rate of the engineered strains under varying ratios of glucose, and lysine. We measured the metabolic rates at 6, 12, and 24 h to determine their effects. Figure 3D illustrates that engineered strains with differing dominant strains exhibited comparable growth rates over the same period. However, the metabolic rate at 6 h was significantly higher than that of the other two periods. Notably, the engineered microbiome displayed a higher level of lysine metabolism compared to the other two groups when EcNT (pK25) was the dominant bacterium. Therefore, the engineered strains with EcNT (pK25) as the dominant bacterium demonstrated faster lysine metabolism. We conducted a comparison of the ratios of EcNT (pTLS) and EcNT (pK25) at 1:1, 1:2, 1:3, and 1:4 to examine the effect of EcNT (pK25) on lysine metabolism in the cocktail. Figure 3E displays that the growth and glucose metabolic rates of each system were comparable, while the lysine metabolic rate at a 1:2 ratio of pTLS:pK25 was greater than the others. Moreover, we compared the tolerance of the engineered strains to glucose and lysine in the medium at a 1:1 ratio. Glucose concentrations of 10 g/L, 20 g/L, and 40 g/L were compared in shake flasks. The results are demonstrated in Figure 3F, indicating that high initial glucose concentration impeded both glucose and lysine metabolism. Furthermore, the outcome of lysine concentrations of 10 g/L, 20 g/L, and 40 g/L as the initial medium components (Figure 3G) reveals that the high lysine concentration decreased the speed of lysine metabolism.

Overall, the two-strain cascade cocktail system was affirmed through in vitro testing, with the highest metabolic capacity found in EcNT (pTLS): EcNT (pK25) at a 1:2 ratio. Additionally, an artificial microsphere system was created to address the issue of spatial colocalizations.

Next, we examined the effect of the probiotics on a hyperlysinemia-like mice model. C57BL/6 mice were raised on a lysine diet of 40 g/L to obtain high blood lysine concentrations mimicking the hyperlysinemia condition (Figure 4A). To evaluate the mice’s blood lysine sensitivity, they were split into three groups and fed with water, EcN, or EcNT for 6 weeks, respectively. During the 40-day treatment period, there were no noteworthy differences observed among the various groups (Figure 4B). Following the treatment, the plasma of the mice was examined to evaluate the concentrations of lysine and saccharopine via LC–MS/MS. As depicted in Figure 4C, in comparison with the control group, the EcN group was able to decrease lysine with a significant accumulation of saccharopine. It is plausible that the decreased lysine was metabolized using the cadaveramines pathway from E. coli. This can be demonstrated by the elevated blood concentration of lysine level in the EcNT group with the deleted ldcC1 and ldcC2 genes that regulate EcNT’s cadaverine pathway. Furthermore, the findings indicate that the hyperlysinemia-like mice model is responsive to the assessment of blood lysine levels.

We further conducted research on probiotics in mice with hyperlysinemia-like symptoms to determine their impact on hyperlysinemia. The hyperlysinemia-like mice were administered with EcNT(pTLS), EcNT(pK25), and a mix of probiotics, EcNT(pTLS) and EcNT(pK25) at a ratio of 1:2 for 6 weeks. During the 40-day treatment, we observed that the weight changes in the treatment group were not significantly different (Figure 4B) compared to the control group. The mouse plasma was analyzed using LC–MS/MS after treatment. The EcNT(pTLS) treatment efficiently lowered the levels of plasma lysine and saccharopine, as anticipated. The metabolic rates of lysine and saccharopine were higher in the EcNT(pTLS) group by roughly 89 and 95%, respectively when compared to the EcNT group (Figures 4D,E). The rapid rates at which lysine is metabolized by EcNT(pTLS) suggest that the synthetic bacteria’s lysine degradation pathway can effectively process free lysine in vivo. Furthermore, EcNT(pTLS) could lower the quantity of lysine and saccharopine in plasma immediately after protein digestion.

A suitable concentration of lysine is essential for protein synthesis, and it is necessary to maintain a balanced metabolism. Excessive consumption of lysine and saccharopine can hinder metabolism maintenance. The cocktail probiotics of EcNT(pTLS) and EcNT(pK25) reduce the lysine concentration by 45% and balance saccharopine concentration, which is more appropriate for balancing the lysine concentration required for protein synthesis and the downstream metabolites compared to EcNT(pTLS). Furthermore, the cocktail group could balance the downstream metabolites through large fluxes of lysine metabolism. To sum up, the cascade cocktail of engineered probiotics could effectively decrease lysine content, maintain saccharopine concentration equilibrium in plasma, and potentially mitigate the side effects of lysine’s incomplete metabolism in hyperlysinemia-like mice.

To compare the gut microbiota of mice with and without the different strains, we conducted 16S rRNA sequencing to detect changes in the gut microbiome. The results of the Shannon analysis showed that mice administered with EcN exhibited a significant increase in microbiome diversity. Additionally, mice administered with the engineered strains, namely EcNT, EcNT(pTLS), EcNT(pK25), and the cocktail, also showed increased microbiome diversity compared to the control group (Figure 4G). Diversity analysis indicated that the cocktail treatment rendered the composition of the microbiome similar to that of the EcN group, suggesting a positive effect on the gut’s healthy microbiome (Figure 4F). Various taxa of commensal bacteria were found to increase in number in the cocktail group as observed in the taxonomic composition of microorganisms, including bacteria belonging to the genus E. coli (Figure 4H). We also found that Bifidobacterium, Escherichia, Cetobacterium, and Faecalibaculum were significantly increased in the cocktail group (Figure 4I). In addition, Streptococcus was significantly reduced in the control group and Christensenllaceae was significantly reduced in the single-engineered strain group. (Figure 4I). These results demonstrate that the engineered strains can reach the intestine of mice and that the engineered cocktail slightly affects the composition of the gut microbiota without inducing microbiome dysregulation.