Adaptive laboratory evolution initiator strain design method with the aim of relying solely on a single carbon source

Abstract

This invention proposes a method for designing starting microbial strains in adaptive laboratory evolution. The method employs a rational reaction knockout strategy to construct adaptive starting microbial strains for laboratory evolution, enabling them to gradually adapt and eventually grow independently on a previously unavailable carbon source. This method utilizes genome-scale metabolic models and a two-layer mixed integer programming algorithm to precisely design starting strains that enhance adaptability to the target carbon source, providing clear guidance and reference for strain optimization in experiments. Creating strains capable of growing on a specified carbon source is of great significance for expanding the application of industrial microorganisms in the utilization of renewable resources; therefore, this invention has significant practical value.

Description

Technical fields:

[0001] This invention belongs to the fields of bioengineering and metabolic engineering, and relates to a method for designing starting strains in adaptive laboratory evolution (ALE) processes, particularly for designing starting strains that gradually adapt to ALE processes and eventually grow solely on a previously unavailable carbon source. Background technology:

[0002] With the development of industrial biotechnology, industrial microorganisms face increasing challenges in production. Traditionally used carbon sources such as glucose are no longer sufficient to meet the needs of certain industrial applications. Therefore, utilizing previously unavailable carbon sources, especially renewable resources, has become a research hotspot. However, many industrial microorganisms lack the natural ability to utilize these carbon sources, which greatly limits their application in industrial production.

[0003] Adaptive laboratory evolution (ALE) is an effective method that forces microorganisms to adapt and eventually utilize new carbon sources by gradually adjusting environmental conditions. However, existing ALE methods typically require co-substrate assistance and struggle to achieve complete dependence of strains on a single target carbon source. To address this issue, this invention proposes the AdaptUC-B method, which, through a rationally designed reaction knockout strategy, enables strains to gradually adapt during the ALE process and ultimately grow solely dependent on a previously unavailable carbon source. Summary of the Invention:

[0004] This invention proposes a method for designing starting strains during ALE (Alternating Lesion of Organisms), called AdaptUC-B. This method uses a genome-scale metabolic model and a mixed-integer programming algorithm to predict and design a response knockout strategy (e.g., through gene knockout) that allows the strain to gradually adapt and eventually rely solely on a previously unavailable carbon source for growth. This method provides clear guidance for strain design and optimization in experiments, enabling the designed strains to successfully adapt during ALE and ultimately independently utilize the target carbon source.

[0005] This invention provides a method for designing starting strains in adaptive laboratory evolution processes, characterized by comprising the following steps:

[0006] S1 provides the following data, including bacterial species, metabolic network model, substrate and strain culture conditions for maximum uptake rate, target carbon source, and co-substrate carbon source;

[0007] S2 uses a two-layer mixed integer programming algorithm based on a genome-scale metabolic model to determine the response knockout strategy, enabling microorganisms to gradually adapt during evolution and eventually rely solely on a carbon source that was previously unavailable for growth.

[0008] S3 applies the above-mentioned reaction knockout target to microorganisms as the evolutionary initiator strain for the adaptive laboratory evolution ALE process, so as to gradually enhance its ability to utilize the carbon source through the adaptive laboratory evolution ALE process, ensuring that the microorganism can eventually grow independently of the carbon source in the absence of other carbon sources.

[0009] Furthermore, the aforementioned two-level mixed integer programming algorithm further includes the following steps:

[0010] a) In the inner layer optimization problem, simulate the growth performance of the strain after reaction knockout under the conditions of target carbon source and other co-substrate;

[0011] b) In the outer layer optimization problem, maximize the utilization of the target carbon source and ensure its necessity for strain growth.

[0012] Specifically, the reaction knockout strategy mentioned above includes:

[0013] a) Select reaction knockout combinations that can couple the target carbon source with growth;

[0014] b) Ensure that the knockout strain can grow in a medium without other carbon sources.

[0015] 4. The method according to any one of claims 1 to 3, characterized in that the method for implementing the two-level mixed integer programming algorithm is as follows: minimizing ∑y j

[0016] Obey:

[0017] Maximize v1 bio =cv1

[0018] Obey

[0019]

[0020] Maximize v2 bio =cv2

[0021] Obey

[0022]

[0023] Maximize v3 bio =cv3

[0024] Obey

[0025]

[0026] minimize

[0027] Obey

[0028]

[0029] v1 bio ≥0.5v0 bio_UC for AdaptUC B

[0030] v1 bio ≤0.1v0 bio_UC for AdaptUC A

[0031] v2 bio ≤0.1v0 bio_cosub

[0032] v3 bio ≥v0 bio_cosub

[0033] ∑y j <K

[0034] v1.v2, v2, v4∈R, y∈[0, 1]

[0035] in,

[0036] S ij The matrix representing the stoichiometric coefficients of metabolite i in reaction j, v1 j It is the flux variable of reaction j, LB1 j and UB1 j They are v1 j The lower and upper bounds of the binary variable y j Indicator j: Is response j deactivated? j =1 or not deactivated y j =0.

[0037] This invention provides a design system for performing the method, comprising the following modules:

[0038] a. Data entry module, used to enter relevant information, including bacterial species, metabolic network model, substrate and strain culture conditions with maximum uptake rate, target carbon source, and co-substrate carbon source;

[0039] b. Data processing module, used for calculation and processing;

[0040] c. Results output module.

[0041] The present invention further provides a bioinformatics tool based on the system, comprising:

[0042] a. User interface that allows users to input relevant information;

[0043] b. The system described;

[0044] d. Results display module, used to show the prediction results to users;

[0045] Optionally, it also includes: e. a report generator for creating results.

[0046] The main features and advantages of this invention include:

[0047] Mixed Integer Programming Algorithm: This algorithm combines genome-scale metabolic models and optimization techniques to effectively handle complex metabolic networks and predict the optimal response knockout combinations.

[0048] Prediction of Target Knockout: This invention predicts and designs a response knockout strategy that promotes microbial dependence on the target carbon source during ALE by simulating the growth of knockout strains under target carbon source and other co-substrate conditions. The principle is that, using a genome-wide metabolic network model as the main input, which contains thousands of major intracellular small molecule metabolic pathways of the target bacteria, and employing a flux balance analysis algorithm, given an objective function and culture medium composition, the intracellular metabolic flux distribution can be predicted. Each subproblem of the AdaptUC-B method is a flux balance analysis problem. All subproblems constitute the inner-layer problem, and the inner-layer problem and the outer-layer problem together form AdaptUC-B.

[0049] As a case study, we present metabolic engineering strategies for ALE-initiating strains, including strategies that co-utilize methanol with other carbon sources and strategies that use methanol as the sole carbon source. Our case study considers important industrial species such as *Escherichia coli* and *Corynebacterium glutamicum*. We evaluate these strategies based on the ease with which the initial strains grow on unadapted carbon sources and the evolutionary drive to assimilate more unadapted carbon sources. To validate our approach, we compare our predictions with experimentally effective strategies. In-depth metabolic analysis reveals new strategies with even greater potential. We anticipate that our work will accelerate the transformation of industrial biology, enabling more industrial microorganisms to utilize previously unadapted but promising renewable substrates. Attached image description:

[0050] Figure 1 AdaptUC-B Concept Diagram Detailed implementation method:

[0051] Example 1: Establishment of AdaptUC-B

[0052] This invention proposes a computational framework called AdaptUC-B, designed to design reactive knockout (K0) strategies for ALE process initiating strains. AdaptUC-B completely couples strain growth to the maladapted substrate, ensuring growth only in its presence. AdaptUC-B retains the potential to utilize the maladapted substrate as the sole carbon source. AdaptUC-B comprises three internal problems (design requirements), each with an objective that must satisfy external constraints with minimal modifications; the global optimization objective is to achieve the minimum number of knocked-out genes. Figure 1 Each internal problem and its corresponding outer constraint represent a design requirement for the ALE starting strain, defining the expectations for the strain at the end of the ALE process.

[0053] Subproblem 1 has an inner key constraint of only taking in C1, and an outer constraint of being able to grow. Subproblem 2 has an inner key constraint of only taking in conventional substrates, and an outer constraint of reducing the growth rate by more than 90%. Subproblem 3 has an inner key constraint of only taking in C1, and an outer constraint of a growth rate no lower than that achieved by taking in conventional substrates alone. The operand is reaction knockout, and the outer constraint is the total number of reaction knockouts.

[0054] Figure 1 This diagram illustrates the concept of AdaptUC-B proposed in this invention. The first internal problem states the necessity for the post-ALE strain to grow well on a medium with an adapted substrate as the sole carbon source (AdaptUC-B). The second requirement specifies that the post-ALE strain should not grow well in the absence of an adapted carbon source, even in the presence of a co-substrate. Combining these two requirements effectively couples the assimilation of the adapted carbon source with growth, creating a strong driving force for cellular uptake of the adapted substrate. In the final internal problem, when the adapted substrate can be taken up (as opposed to the second internal problem), the goal is to ensure that the optimal growth of the post-ALE strain is not inferior to that of the reference strain on the co-substrate. This design requirement ensures that the addition of a conventional substrate to a medium containing an adapted carbon source can aid the survival and growth of the initiating strain during the ALE process. We assume that the gene deletions in the post-ALE strain and the ALE initiating strain are similar, or that their differences have little impact on growth. AdaptUC is specifically designed for constructing ALE initiating strains.

[0055] Design of AdaptUC-B

[0056] AdaptUC-B comprises three design requirements, all of which must be met simultaneously. The first design requirement is that the genetically modified strain has the potential to evolve to grow on unused carbon sources. Growth rate was simulated using flux balance analysis. v1, v2, and v3 represent the intracellular metabolic flux distribution (in millimoles / g dry weight cells / hour) under the three design requirements, LB1, LB2, and LB3 represent the lower limit of reaction flux under the three design requirements, and UB1, UB2, and UB3 represent the upper limit of reaction flux under the three design requirements.

[0057] Maximize v1 bio

[0058] Obey

[0059]

[0060] Among them, S ij Let v1j be the stoichiometric matrix of metabolite i in reaction j, v1j be the flux variable of reaction j, and LB1 be the stoichiometric matrix. j and UB1 j They are v1 j The lower and upper limits of these boundaries can be used to thermodynamically define the reversibility of a reaction. Binary variable y j Indicator j: Is response j disabled (y)? j =1) or not deactivated (y j =0). The outer layer constraint of the first internal problem is that, for AdaptUC-B, it must still be able to grow on unadapted substrates after deletion, and the growth reduction caused by gene deletion must not exceed 50%.

[0061] v1 bio ≥0.5v0 bio_UC (6)

[0062] Where v0 bio_UC The growth rate of the reference strain on an unadapted carbon source can be calculated by solving the following FBA problem:

[0063] Maximize v0 bio_UC =cv

[0064] Obey

[0065]

[0066] Where v j It is the flux variable of reaction j.

[0067] The second design requirement is that, after gene deletion, the strain does not grow or its growth is severely restricted on conventional co-substrate, forcing the strain to utilize the C1 substrate in its presence. Considering that a 90% reduction in growth satisfies this design requirement, it is achieved through the following constraints:

[0068] v1 bio ≤0.1v0 bio_UC (10)

[0069] The calculation of the growth rate of the reference strain on the co-substrate and v0 bio_UC Similarly, the difference lies in setting the boundary for the co-substrate exchange reaction. The growth rate v2 is calculated through the following internal problem. bio :

[0070] Maximize v2 bio =cv2

[0071] Obey

[0072]

[0073] The third design requirement is that adding an unadapted substrate to a culture medium containing only the co-substrate should restore the growth rate before gene deletion. This design requirement is achieved through the following outer layer constraints:

[0074] v3 bio ≥v0 bio_cosub (16)

[0075] Under these conditions, the growth rate is v3. bio Calculated through the third internal problem:

[0076] Maximize v3 bio =cv3

[0077] Obey

[0078]

[0079] Minimum uptake rate v of unadapted carbon source UC It can be obtained through the following internal question.

[0080] minimize

[0081] Obey

[0082]

[0083] cv4≥v0 bio_cosub (27)

[0084] Where: v4 is the reaction flux distribution under the minimum uptake condition for an unadapted carbon source (unit: millimoles / g dry weight cells / hour), and LB4 and UB4 are the lower and upper limits of the reaction flux under the minimum uptake condition for an unadapted carbon source, respectively.

[0085] The total number of reactions to be deleted, K, must be limited; the default value for K is 5, to speed up the solution process.

[0086] ∑y j <K (29)

[0087] In summary, the formula for the AdaptUC algorithm is as follows. This optimization algorithm is a whole, an indivisible series of steps executed simultaneously. The preceding description explains the various parts of this indivisible series:

[0088] Minimize ∑y j

[0089] Obey:

[0090] Maximize v1 bio =cv1

[0091] Obey

[0092]

[0093] Maximize v2 bio =cv2

[0094] Obey

[0095]

[0096] Maximize v3 bio =cv3

[0097] Obey

[0098]

[0099] minimize

[0100] Obey

[0101]

[0102] v1 bi o≥0.5v0 bio_UC for AdaptUC B

[0103] v1 bio ≤0.1v0 bio_UC for AdaptUC A

[0104] v2bio ≤0.1v0 bio_cosub

[0105] v3 bio ≥v0 bio_cosub

[0106] ∑y j <K

[0107] v1.v2, v2, v4∈R, y∈[0, 1]

[0108] Example 2: Validation of AdaptUC

[0109] Example 1: Design of E. coli starting strains with methanol as the target carbon source

[0110] Model building:

[0111] First, the methanol assimilation pathway (including methanol dehydrogenase, 3-hexose-6-phosphate synthase, and 6-phosphate hexose isomerase) was integrated into the genome-scale metabolic model iML1515 of *E. coli* (Monk, JM, Lloyd, CJ, Brunk, E., Mih, N., Sastry, A., King, Z., Takeuchi, R., Nomura, W., Zhang, Z., Mori, H., Feist, AM, Palsson, BO, 2017. iML1515, a knowledgebase that computes *Escherichia colitraits*. Nat Biotechnol 35.904-908. https: / / doi.org / 10.1038 / nbt.3956), and some unnecessary metabolic pathways in the model were removed.

[0112] Escherichia coli: Three heterogeneous reactions encoded by the RuMP methanol assimilation pathway (medh: methanol dehydrogenase, hps: 3-hexanonealdehyde-6-phosphate synthase, and phi: 3-hexanonealdehyde-6-phosphate isomerase) were added to the E. coli model iML1515. The reactions 'PFL' (catalyzed by pyruvate-formate lyase) and 'OBTFL' (catalyzed by 2-oxobutyrate-formate lyase) were deactivated in the model because they are active only under anaerobic conditions. Furthermore, the 'DRPA' and 'PAI2T' reactions were also removed because they were reported as impractical in the prior art.

[0113] Determining the reaction knockout strategy:

[0114] Using the AdaptUC-B algorithm, several reactive knockout strategies were simulated and predicted to enhance the growth ability of *E. coli* under methanol conditions. Details are shown in the table below:

[0115]

[0116]

[0117]

[0118]

[0119] The reaction knockout combination ΔtpiΔrpeΔf6pa was ultimately determined as the optimal strategy. This combination can gradually enhance the utilization efficiency of methanol by E. coli during ALE and reduce the dependence on other co-substrate.

[0120] Implementation of the ALE process:

[0121] First, the modified *E. coli* strain was cultured in a medium containing methanol, and the supply of other co-substrates in the medium was gradually reduced, forcing the strain to gradually increase its dependence on methanol. After multiple generations, an evolved strain capable of growing in a medium containing only methanol was obtained. The AdaptUC-B algorithm of this invention is used to determine the modification strategy of the starting strain in adaptive evolutionary subculturing experiments. Therefore, the effectiveness of the modification strategy predicted by AdaptUC-B is demonstrated by comparing successful cases in the literature. For example, the AdaptUC-B strategy using pyruvate as a co-substrate and deleting tpi has also been verified in the literature; furthermore, the AdaptUC-B strategy using pyruvate as a co-substrate and deleting tpi has also been verified in the literature (Keller, P., Noor, E., Meyer, F., Reiter, MA, Anastassov, S., Kiefer, P., Vorholt, JA, 2020. Methanol-dependent *Escherichia coli* strains with a complete ribulosemonophosphate cycle. Nat Commun 11, 1-10). https: / / doi.org / 10.1038 / s41467-020- 19235-5; Keller, P., Reiter, MA, Kiefer, P., Gassler, T., Hemmerle, L., Christen, P., Noor, E., Vorholt, JA, 2022. Generation of an Escherichia coli strain growing onmethanol via the ribulose monophosphate cycle. Nat Commun 13.1-13. https: / / doi.org / 10.1038 / s41467-022-).

[0122] The results showed that the growth efficiency of the modified Escherichia coli in methanol medium was significantly improved by using the reaction knockout strategy predicted by AdaptUC-B, which verified the effectiveness and applicability of the algorithm.

[0123] This invention provides an effective method for designing starting strains for adaptive laboratory evolution processes, called AdaptUC-B. This method, through predictive reaction knockout strategies, enables microorganisms to gradually adapt and eventually rely solely on a previously unavailable carbon source for growth. This method can significantly improve the efficiency of industrial microorganisms in utilizing previously unused carbon sources, contributing to the innovation and development of biosynthesis.

Claims

1. A method for designing starting strains in adaptive laboratory evolution processes, characterized in that, Includes the following steps: S1 provides the following data, including bacterial species, metabolic network model, substrate and strain culture conditions for maximum uptake rate, target carbon source, and co-substrate carbon source; S2 uses a two-layer mixed integer programming algorithm based on a genome-scale metabolic model to determine the response knockout strategy, enabling microorganisms to gradually adapt during evolution and eventually rely solely on a carbon source that was previously unavailable for growth. S3 applies the target after the above reaction knockout to microorganisms as the evolutionary initiator strain for the adaptive laboratory evolution ALE process, so as to gradually enhance its ability to utilize the carbon source through the adaptive laboratory evolution ALE process, ensuring that the microorganism can eventually grow independently of the carbon source in the absence of other carbon sources. The aforementioned two-level mixed integer programming algorithm further includes the following steps: a) In the inner layer optimization problem, simulate the growth performance of the strain after reaction knockout under the conditions of target carbon source and other co-substrate; b) In the outer layer optimization problem, maximize the utilization of the target carbon source and ensure its necessity for strain growth; The reaction knockout strategy mentioned above includes: a) Select reaction knockout combinations that can couple the target carbon source with growth; b) Ensure that the knockout strain can grow in a medium without other carbon sources; The method for implementing the aforementioned two-level mixed integer programming algorithm is as follows: in, Minimize ∑y j Obey: Maximize v1 bio =cv1 Obey Maximize v2 bio =cv2 Obey Maximize v3 bio =cv3 Obey Minimize v c1 =-v4j |j∈J UC Obey v1 bio ≥0.5v0 bio_UC for AdaptUC B v2 bio ≤0.1v0 bio_cosub v3 bio ≥v0 bio_cosub ∑y j <5 v1, v2, v3, v4 ∈ R S ij V1 represents the stoichiometric coefficient of metabolite i in reaction j. j It is the flux variable of reaction j, LB1 j and UB1 j They are v1 j The lower and upper bounds of the binary variable y j Indicator j: Is response j deactivated? j =1 or not deactivated y j =0.

2. A design system for performing the method as described in claim 1, characterized in that, Includes the following modules: a. Data entry module, used to enter relevant information, including bacterial species, metabolic network model, substrate and strain culture conditions with maximum uptake rate, target carbon source, and co-substrate carbon source; b. Data processing module, used for calculation and processing; c. Results output module.

3. A bioinformatics tool system, comprising: a. User interface that allows users to input relevant information; b. Design the starting strain using the design system described in claim 2; d. Results display module, used to display the design results to users.

4. The bioinformatics tool system as described in claim 3, characterized in that, Also includes: e. Report generator, used to create results.

Images