Lactobacillus brevis carbon source screening method based on whole genome metabolism model

Abstract

The invention belongs to the crossing field of synthetic biology and fermentation engineering, and relates to a lactobacillus brevis carbon source screening method based on a whole genome metabolism model. The method comprises the following steps: firstly, carrying out whole genome sequencing on a target strain, and constructing an initial model by utilizing gapseq; a representative cheap carbon source is used for carrying out a'two-stage sequential correction 'experiment, sugar transport and energy metabolism are constrained and corrected by a specific growth rate, and then GAD branch flux is constrained and corrected cooperatively by a glutamic acid-GABA conversion rate and a substrate cost, so that the model precision is improved. And then simulating a plurality of carbon sources in batches in the correction model, screening an optimal combination with high biomass, high GABA and low cost, and performing closed-loop verification through high-throughput micro-feeding and blind test experiments. And finally, the yield of the GABA is increased through'starvation-pulse 'material supplementation and pH gradient optimization, and a rapid, low-cost and rational carbon source screening process from a genome to process amplification is realized.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of synthetic biology and fermentation engineering, and relates to a method for screening carbon sources in Lactobacillus brevis based on a whole-genome metabolic model. Background Technology

[0002] Lactobacillus brevis, due to its acid and osmotic resistance and high GABA production, has become a core strain in functional fermented foods, probiotic preparations, and green chemical synthesis routes. However, current processes generally use glucose as the sole or primary carbon source, resulting in raw material costs accounting for more than 50% of the total fermentation cost, severely weakening the market competitiveness of GABA products. More importantly, Lactobacillus brevis exhibits vastly different utilization efficiencies for various inexpensive by-product carbon sources (such as molasses, xylose mother liquor, glycerol, and whey permeate). Inappropriate carbon source selection can lead to low biomass and even carbon-nitrogen imbalance in the GABA synthesis pathway, resulting in a sharp drop in yield. Therefore, a low-cost carbon source alternative that can simultaneously enhance GABA synthesis has become a bottleneck for the industry.

[0003] Current mainstream methods rely on wet experimental validation (such as Biolog phenotypic microarrays and batch culture growth curve determination): they are time-consuming, requiring several days to weeks for a single screening; they have limited throughput, making it difficult to cover hundreds of potential carbon sources; they are not economically viable, and experiments with high-value / toxic carbon sources are risky; and they have blind spots, making it difficult to systematically explore unknown carbon sources.

[0004] Genome-wide metabolic models (GEMs), by integrating genome annotation and biochemical reaction networks, can simulate cellular metabolic flows (e.g., the FBA algorithm), providing a new pathway for phenotypic prediction. However, carbon source screening for *Lactobacillus brevis* still faces challenges: standardized model construction is lacking, high-quality *Lactobacillus brevis* GEMs are scarce in public databases, and the strain suitability of automated tools (such as CarveMe and ModelSEED) is not validated. Predictive accuracy is questionable; uncalibrated models easily overlook strain-specific metabolic constraints (such as cofactor requirements and transporter protein functions). The toolchain is fragmented; existing patents focus on general model algorithms, lacking a complete solution from strain-specific modeling to carbon source screening. Summary of the Invention

[0005] This invention addresses the shortcomings of existing carbon source screening methods, such as long screening cycles, low throughput, failure to balance the dual objectives of "high-density cultivation" and "high GABA production," and cost constraints. It proposes a highly efficient carbon source screening and process intensification method based on "small-scale calibration - high-precision prediction - process enhancement." After initial modeling using genome sequencing and GAPSE, this method innovatively introduces 15 representative inexpensive carbon sources and a two-stage sequential calibration pilot step: [The method is described in the original text, but the provided excerpt ends here.] -1The invention employs a unique constraint to correct sugar transport and energy regeneration reactions. Within the top 30% subset, a secondary correction of the GAD branch flux boundary is performed with a glutamate-to-GABA molar conversion rate ≥50% and a substrate cost ≤50% of the glucose price as a co-constraint, resulting in a high-precision GEM. Subsequently, a multi-objective flux balance analysis is conducted using COBRApy to virtually screen for optimized carbon sources that simultaneously meet the criteria of "growth rate not lower than glucose, GABA yield significantly higher than glucose, and cost much lower than glucose." Finally, a "starvation-pulse" feeding strategy (starvation for 2 h followed by a pulse of 10 g / L of the same carbon source, pH 8.5→5.0) and a 96-well plate pH gradient matrix are used to lock the optimal carbon source-pH window in one step, achieving a further increase in GABA yield of ≥55%. The screened carbon sources can be directly used as the core component of the industrial-scale GABA fermentation medium for *Lactobacillus brevis*, enabling low-cost, high-efficiency large-scale GABA production. This invention also provides a fermentation medium optimization method, applying a combination of screened and process-enhanced high-efficiency carbon sources to the industrial-scale culture of *Lactobacillus brevis*.

[0006] Technical solution: The first aspect of this invention provides a method for screening carbon sources in Lactobacillus brevis based on a whole-genome metabolic model, the method comprising the following steps: 1) The FASTA file obtained from the whole genome sequencing of Lactobacillus brevis was input into gapeq for automated metabolic network construction. This process automatically completes metabolic pathway prediction, transport system identification, draft model generation and gap filling optimization based on bioinformatics principles. 2) Manually adjust the carbon source components in the input file according to the CDM medium formula, delete glucose and retain 10-20 representative carbon sources with "monovalent value ≤ 50% glucose and metabolic coverage ≥ 85%", and set their concentration to 5-15 mmol gDW. -1 h -1 We fixed 18 amino acid nitrogen sources and standardized trace elements to ensure consistent testing conditions for different carbon sources. 3) Conduct a two-stage sequential correction of "growth-product decoupling": with a specific growth rate μ ≥ 0.06 h -1"With the sole constraint, the true μ was determined using 48-well plate microculture, and the sugar PTS and glycolysis kcat / Km were reverse-corrected to make the μ prediction error <8%. Within the first 30% subset of μ, the GABA yield was determined through microfeeding inflection point experiments. The GAD branch flux boundary was corrected twice with the glutamic acid → GABA molar conversion rate ≥50% and substrate cost ≤50% of glucose price as a synergistic constraint to make the GABA prediction error <5%; 4) The corrected GEM was used to simulate ≥15 kinds of inexpensive carbon sources in batches, and COBRApy was called to perform multi-objective flux balance analysis to screen carbon sources in the synergistic region that simultaneously meet the requirements of "μ not lower than glucose, unit substrate cost GABA yield > glucose, and carbon source unit price ≤50% of glucose" as industrially recommended usable carbon sources; 5) Establish a "blind test" closed-loop verification. Randomly select two carbon sources that are not involved in the modeling. The GEM outputs OD600, GABA, and cost ranking in advance. After single-well verification, the deviation between the measured value and the predicted value is <15%, ensuring the reliability of the model extrapolation. 6) Implement a "starvation-pulse" feeding strategy for the recommended carbon source: first, incubate with the carbon source for 6 h to mid-logarithmic growth, then starve for 2 h in a high pH 8.5 buffer without carbon source, followed by a one-time pulse feeding of 10 g / L of the same carbon source + 0.05 mM PLP and adjusting the pH back to 5.0, continue for 6 h to further increase GABA yield by ≥55%. At the same time, use a 96-well plate pH gradient matrix (each row with the same carbon source, each column with pH 4.0, 4.5, 5.0, 5.5, 6.0) to lock the peak pH at once and complete the scale-up verification.

[0007] Preferably, the Lactobacillus brevis strain is AR455. Preferably, the carbon source screening is a carbon source suitable for GABA production; Preferably, the gapeseq version is 1.40, and its automated modeling process performs the following standardized steps: identifying metabolic pathways in the genome by comparing protein sequence evidence from the three major databases MetaCyc, KEGG, and ModelSEED (the existence of a pathway is determined if more than 85% of the reactions are sequence-supported); matching carbon source transport genes based on the transporter classification database TCDB (if a homolog of xylose transporter TC#3.A.1.1.2 is detected, the corresponding reaction is added); and employing a gap-filling algorithm to ensure biomass synthesis capacity. The key improvement is the use of CDM medium instead of standard M9 medium for missing reaction completion, making the model more closely resemble the real nutritional environment of Lactobacillus brevis.

[0008] Preferably, in step 2), 15 representative carbon sources are retained, with a concentration set at 10 mmol gDW. - 1 h -1 ; Preferably, in step 2), the manual adjustment of the culture medium formulation must strictly follow these steps: delete reference carbon sources such as glucose from the original file and replace them with ≤3 representative target carbon sources; fix the nitrogen source components to 18 amino acids; standardize the concentration of trace elements to ensure consistency of test conditions for different carbon sources.

[0009] Preferably, the determination of carbon source availability and GABA yield in steps 3) and 4) employs a dual-threshold verification mechanism: during the computational simulation phase, the COBRApy basic function is called to perform flux balance analysis, predicting a specific growth rate μ_max ≥ 0.06 h. -1 Furthermore, the strain was deemed usable when its GABA production rate was significantly higher than that of glucose; during the experimental verification phase, the strain was cultured anaerobically at 37 °C for 48 h, and the OD was recorded using a microplate reader. 600 Dynamic data, and with endpoint OD 600 A positive standard is a dual standard that uses a GABA yield (Berthelot colorimetric method) significantly higher than that of glucose, with a result greater than 0.2. This standard can effectively avoid false positives.

[0010] Preferably, the method further includes establishing a manual verification and correction mechanism: when the predicted result of a specific carbon source (such as L-arabinose) does not match the experiment, the gap filling record is retrieved from the gapeseq running log. If a key reaction is found to be missing, the reaction is manually added to the model and recalculated until the error converges to a reasonable range.

[0011] A second aspect of the present invention provides a fermentation medium for the industrial production of GABA from Lactobacillus brevis, characterized in that the core carbon source component of the medium is composed of an industrially recommended carbon source screened and process-enhanced by the method described in the first aspect.

[0012] The beneficial effects of this invention include: 1) Significantly shortens the carbon source-product co-screening cycle and reduces industrial costs: Compared to traditional phenotypic chips (PM) or batch shake-flask experiments, which take 2–4 weeks and can only test dozens of carbon sources at a time, this invention uses a strategy of "≤3 representative carbon sources with dual-stage sequential correction + high-precision GEM" to compress the experimental cycle to 7 days. The growth of 96-well plates and the verification of GABA only take 48 hours. It can virtually screen ≥15 carbon sources at a time, increasing the screening throughput by more than 10 times and reducing labor and reagent costs by 80%. At the same time, the unit price of the selected carbon source is only 1 / 8 to 1 / 2 of that of glucose, and the unit GABA raw material cost can be reduced by 50%–90%, significantly enhancing the industrial competitiveness of Lactobacillus brevis in functional fermented foods, probiotic preparations and green biomanufacturing.

[0013] 2) The model prediction accuracy has been improved to the industrial scale-up level: CDM medium is used for interstitial filling to overcome the defect of poor compatibility of standard M9 with lactic acid bacteria; after "μ first and then GABA" dual threshold correction, the model comparison growth rate prediction error is <8%, the GABA yield prediction error is <5%, the extrapolation blind test deviation is <15%, and the comprehensive accuracy of the "growth + GABA" dual indicators is 92% verified by 15 types of carbon sources, which can directly output a scale-up low-cost high-yield formula.

[0014] 3) One-step process enhancement: Based on virtual screening, a "starvation-pulse" feeding strategy is coupled (starvation for 2 h + pulse 10 g / L of the same carbon source + PLP 0.05 mM, pH 8.5→5.0), which increases GABA yield by ≥55% compared to constant culture; at the same time, a 96-well plate pH gradient matrix is ​​used to lock the peak pH in one go, realizing a one-time closure from screening to process window optimization. Attached Figure Description

[0015] Figure 1 This is a flowchart of carbon source screening based on the Lactobacillus brevis whole-genome metabolic model. Figure 2 Experimental validation and qualitative prediction of growth phenotypes with different carbon sources Figure 3 Growth curves for different carbon sources Figure 4 GABA content in CDM medium with different carbon sources Detailed Implementation

[0016] The following detailed embodiments further illustrate the concept and technical effects of the present invention to fully understand its purpose, features, and effects. Unless otherwise specified, all methods described are conventional methods. Unless otherwise specified, all materials are available from publicly available commercial sources. The illustrative embodiments and descriptions of the present invention are used to explain the invention and do not constitute an undue limitation thereof. It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0017] Example 1: Construction of a genome metabolic model of Lactobacillus brevis AR455 according to Figure 1 The flowchart in this embodiment first constructs a metabolic model based on the Lactobacillus brevis AR455 genome: The laboratory-preserved *Lactobacillus brevis* strain AR455 was removed from a -80℃ glycerol tube and anaerobically activated on MRS solid medium at 37℃ for 48 hours. A single colony was picked and inoculated into 5 mL of MRS solid medium and cultured at 37℃ for 16 hours. Then, a 2% inoculum was added to 50 mL of MRS medium and incubated statically for 16 hours until mid-log phase (OD50). 600=1.0±0.3). The bacterial culture was centrifuged at 10,000 rpm for 10 minutes to collect the bacterial cells. The cells were washed three times with pre-cooled phosphate-buffered saline (PBS, pH 6.5) to thoroughly remove any culture medium residue, leaving only bacterial sludge. Whole-genome sequencing was outsourced to a commercial organization using Illumina + PacBio sequencing. The high-quality genome FASTA file obtained was directly uploaded to gapseq software for automated model construction.

[0018] The software first calls the MetaCyc, KEGG, and ModelSEED databases for pathway prediction: it then compares the results with DIAMOND (e-value < 1e...). -10 The study identifies enzyme-encoding genes associated with metabolic responses, determining the presence of a pathway when ≥70% of responses in that pathway show sequence evidence. Simultaneously, it identifies carbon source transport systems based on the Transporter Protein Classification Database (TCDB). These results are then integrated to generate an initial draft model.

[0019] Key model adaptation procedures: Replace the software's default culture medium with the experimental CDM formulation, delete the original glucose exchange reaction, and add exchange reactions for candidate carbon sources (including xylose, inulin, mannose, etc.) (flux range set to -10 to 0 mmol / g DCW / h). Perform flux balance analysis (FBA) using COBRApy, with the constraint of an anaerobic environment (upper limit of oxygen uptake flux = 0 mmol / g DCW / h), using maximizing biomass response BIOMASS as the objective function. The criteria for determining the availability of a carbon source are: carbon source uptake flux < -1.0 mmol / g DCW / h and biomass flux > 0.06 h. -1 Then, under the constraint of μ≥0.9×μ_Glc, the synthesis rate of GABA ratio is maximized to provide a quantitative basis for subsequent industrial screening.

[0020] Example 2: Model Optimization and Experimental Verification To calibrate the accuracy of the model predictions, a 96-well microplate carbon source growth experiment was conducted: Microplates containing 180 μL of basic CDM (carbon source-free) were prepared in a clean bench. 20 μL of filtered and sterilized carbon source stock solution (final concentration 20 g / L) was added to each well, covering 20 candidate carbon sources. 5 μL of bacterial suspension was inoculated, and carbon source-free negative control, glucose positive control, and sterile blank control were set up (three replicates each). The plates were anaerobic at 37 °C for 48 h, and OD values ​​were read every 0.5 h. 600 The supernatant from the 48-hour fermentation broth was then analyzed using the Berthelot colorimetric method to determine GABA content. The endpoint OD was used. 600 A result greater than 0.2 and with GABA production exceeding glucose levels is considered positive. The experimental results are directly compared with the model predictions. Figure 2 The model falsely reported four types of unusable carbon sources (such as the measured OD of sucrose). 600=0.14). Using gapeq for targeted gap filling: removing erroneous transport reactions related to sucrose, improving the model's prediction accuracy to 90%. Ultimately, this allows for the rapid identification of low-cost, high-efficiency industrial-grade carbon sources. Figure 3 ).

[0021] Example 3: Prediction and Optimization of Carbon Source Performance In a comparative experiment involving 100 candidate carbon sources, the measured GABA production of D-mannitol, D-fructose, and D-galactose was significantly higher than that of glucose, while the model predictions were close to those of glucose. Maltose showed zero GABA production in the model, but 0.1 mg / ml of GABA was detected experimentally. D-ribose exhibited a phenomenon of "model overestimation and measured underestimation." These discrepancies mainly stem from insufficient annotation of inducible glutamate decarboxylase expression, cofactor supply, and the activity of certain transport proteins in existing genomic metabolic models. Therefore, this embodiment uses experimental results as the standard, employing the model only as a preliminary screening tool. The aforementioned biases do not affect the final selection of carbon sources. The model simulates the metabolic phenotype of AR455 under 100 carbon sources: with glucose as the baseline (μ=0.472h). -1 ), maltose (μ = 0.549 h) -1 One carbon source was listed as a high-efficiency carbon source (μ≥0.495 h⁻¹). -1 ); Meanwhile, using the 48-hour fermentation broth GABA yield as an indicator, the yield using D-mannitol, D-fructose, and D-galactose as carbon sources was significantly higher than that using glucose as a carbon source. Figure 4 It can serve as a preferred carbon source for the industrial production of GABA from Lactobacillus brevis.

[0022] The model also features exchange reactions of common organic acids such as acetic acid, lactic acid, and succinic acid. Utilizing the high-precision GEM after dual-stage calibration of this invention, simply replacing the objective function from "maximizing GABA" to "maximizing acetic acid / lactic acid / succinic acid" allows for rapid identification of high-yield, low-cost carbon source combinations for the corresponding acids within the same "small-scale calibration-virtual screening-process intensification" process. This enables the direct expansion of *Lactobacillus brevis* from a single GABA plant to a multi-acid co-production platform, providing a low-cost raw material solution for applications such as food preservatives, bio-based material precursors, and feed acidifiers.

[0023] The embodiments described above are only some embodiments of the present invention, not all embodiments. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are all within the scope of protection of the present invention.

Claims

1. A carbon source screening method based on a whole-genome metabolic model of Lactobacillus brevis, characterized in that... It includes the following steps: A. Perform whole-genome sequencing on the target Lactobacillus brevis to obtain its whole-genome data; B. Using the gapeseq software, an initial genome-scale metabolic model (GEM) was constructed based on the whole genome data obtained in step A, and the potential available carbon sources for this strain were predicted; C. Small-scale two-stage sequential experiments were conducted using 10-15 representative inexpensive carbon sources to determine the growth-product decoupling process: first, a specific growth rate μ ≥ 0.3 h was established. -1 To constrain the correction of sugar transport and energy regeneration reactions, a secondary correction of the GAD branch flux boundary was performed using "glutamic acid → GABA molar conversion rate ≥ 50% and substrate cost ≤ 50% glucose value" as a co-constraint, thus obtaining a high-precision GEM. D. Based on the high-precision GEM obtained in step C, flux balance analysis was used to simultaneously simulate the strain's specific growth rate and GABA specific synthesis rate; E. Based on the simulation results, the carbon source with the highest GABA yield per unit substrate cost and biomass accumulation not lower than that of the glucose control was selected as the recommended carbon source for industrial production. F. Implement a "starvation-pulse" feeding strategy for the carbon source determined in step E: first, incubate with the carbon source for 6 h to mid-logarithmic growth, then starve for 2 h in a high pH 8.5 buffer solution without carbon source, followed by a one-time pulse addition of 10 g / L of the same carbon source and adjustment of pH back to 5.0, and continue incubation for 6 h to further increase GABA yield by ≥55%; G. A 96-well plate with a "dual pH gradient" matrix was used, with each row containing the same carbon source and each column containing pH values ​​of 4.0–6.0, to complete the determination of the optimal carbon source-pH window in one go and lock in the acidity conditions corresponding to the peak GABA.

2. The method according to claim 1, characterized in that, In step B, the Cobrapy package in Python is used to perform flux balance analysis (FBA). By solving the objective function of maximizing biomass, the carbon source uptake flux is calculated. If the carbon source uptake flux is less than 0 and the biomass flux is greater than the threshold, the carbon source is determined to be a potential available carbon source.

3. The method according to claim 1 or 2, characterized in that, The experimental measurements in step C include the determination of the growth curve of the micro-filled 48-well plate and the quantification of GABA content by Berthelot colorimetric method. The experimental data are used to fill gaps in the model and correct the reaction boundary constraints.

4. The method according to any one of claims 1 to 3, characterized in that, Step D uses the Python COBRApy package to perform multi-objective flux balance analysis, with the objective function being to maximize biomass flux while simultaneously analyzing GABA synthesis flux.

5. The method according to any one of claims 1 to 4, characterized in that, The screening rule for step E is as follows: In the simulation results, only carbon sources with a specific growth rate not lower than that of the glucose control and a GABA yield per unit substrate cost higher than that of glucose are retained as industrially recommended carbon sources.

6. The method according to any one of claims 1 to 5, characterized in that, The "starvation-pulse" feeding strategy described in step F has a starvation time of 2 h, a pulse carbon source concentration of 10 g / L, and a post-pulse pH of 5.

0.

7. The method according to any one of claims 1 to 6, characterized in that, In step G, the pH range of the dual pH gradient matrix is ​​4.0, 4.5, 5.0, 5.5, and 6.

0. The buffer solution and carbon source are added simultaneously for each range, and GABA is measured and the peak value is taken after 24 hours.

8. The method according to any one of claims 1 to 7, characterized in that, The number of carbon sources in step C is 15.

9. The method according to any one of claims 1 to 8, characterized in that, The method also includes establishing a manual verification and correction step: when the predicted results of a specific carbon source do not match the experiment, the gap filling records are retrieved from the gapseq running log. If a key reaction is found to be missing, the reaction is manually added to the model and recalculated until the error converges to a reasonable range.

10. A fermentation medium for *Lactobacillus brevis* for industrial production of GABA, characterized in that, The core carbon source component of the culture medium consists of an industrially recommended carbon source that has been screened and process-enhanced by the method described in any one of claims 1 to 9.

Images