Enzyme-catalyzed reaction path optimization and biosynthetic design based on genetic algorithm

As an element with catalytic function, enzyme has great advantages compared with chemical catalysts, and it can form a composite catalytic system through rational design and multiple assembly to realize cascade catalytic function. Enzyme cascade catalysis can produce various chemical products with high conversion rates in a short time. Understanding how selected enzymes convert biological raw materials into biochemicals in multi-enzyme cascade catalytic reactions can serve as an important basis for biocatalysis, metabolic engineering and synthetic biology applications, thus realizing environmentally friendly green bioprocesses, while the utilization of enzyme catalysis is also very valuable for the conversion of non-natural compounds. The construction of multi-enzyme cascade reaction systems as well as in vitro organelles further enables the synthesis of complex structural compounds, thus solving a series of problems arising in biosynthetic applications. As more and more genome-scale metabolic networks are being reconstructed, the amount of data mined in biology is gradually increasing, so more theories, algorithms, and tools are needed [1].

However, in the process of exploring the synthetic pathways of multiple enzymes, exploring their optimal synthetic reaction pathways based on known products remains a challenge, and manual design based only on personal a priori knowledge suffers from limitations such as inexperience and incomplete data [2–3]. Based on this, a tool model based on artificial intelligence machine learning algorithm is proposed for inverse synthetic pathway analysis of target compounds. By defining the target molecule to be synthesized and then constructing it by using consistent biochemical reaction rules as templates to identify potential precursors and enzymes required for the reaction, the optimized reaction pathway has shorter steps and the reaction efficiency can be significantly improved [4–6].

Currently the most effective method for designing synthetic routes to known target compounds is to combine chemical and biological retrosynthetic analyses into a single tool to optimize synthetic pathways. Helmy, M. et al. elucidated that metabolic engineering is the process of maximizing the benefits of obtaining important materials by optimizing biogenetics, cellular processes, and growth conditions, and that with the explosion of biological data, the use of artificial intelligence techniques can optimize potential metabolic pathways and break through established metabolic engineering goals [7]. Siedentop, R. et al. discussed the optimization strategies of numerous in vitro enzyme cascade design systems, which have the advantages of synthesizing complex molecules without the need for intermediate separations, and which can be improved in terms of enzyme selection, reaction conditions, and process design by using design strategies including algorithmic optimization [8]. Liu, Q. et al. outlined the great promise of the application of computational enzyme design in combination with molecular evolution, showing that new methods based on precise and global structural dissection for generating highly efficient and stable enzymes open new frontiers in protein engineering, allowing for the rapid development of biocatalysis [9]. Mazurenko, S. et al. described the application of machine learning methods in enzyme engineering, where the use of intelligent algorithms to find patterns in the data to predict protein structures, improve enzyme reaction functions, and to understand substrate specificity is one of the important roles of machine learning methods, in addition to classical rational design and directed evolution [10]. Li, G. et al. analyzed the utility of machine learning technology for selective enzyme directed evolution assisted engineering, which is a suitable technology for protein engineering by learning the large amount of data generated by enzyme directed evolution to make decisions and predictions about reaction pathways as well as attributes of selective enzyme directed evolution [11]. Volk, M. J. et al. investigated the application of machine learning techniques to biological systems design, whose ability to identify patterns in complex biological data across multiple scales of analysis and its ability to predictively optimize the performance of new candidate objects can help enhance biological systems design applications at all scales [12]. Siedhoff, N. E. et al. described the advantages and challenges of machine learning methods in enzyme engineering, machine learning-assisted methods predict different enzyme properties by determining the relationship between protein structure and function, greatly accelerating the rate of directed enzyme evolution, but there are some limitations to this method [13]. BIAN, J. et al. showed that artificial intelligence models can learn the internal properties and relationships of protein structures from a given sequence function dataset to predict the serial properties of proteins and enzymes, and that this high-throughput screening approach plays an important role in the study of the mechanisms of protein folding and natural evolution of enzymes [14]. Saharan, V. et al. combined artificial neural networks with multi-objective genetic algorithms to statistically design Aspergillus niger VSRK09 to optimize its enzyme production capacity, and it was shown that the enzyme-catalyzed reaction products under the proposed model were highly active and had a good representation of performance [15]. Karim, A. S. et al. proposed an in vitro platform for prototyping and rapid optimization of biosynthetic enzymes (iPROBE), under the auspices of which cell lysates are enriched for biosynthetic enzymes through cell-free protein synthesis and metabolic pathways are assembled in a mix-and-match fashion, significantly enhancing the process of designing and optimizing biosynthetic pathways [16].

Taking Lactobacillus acidophilus mutant strain B-5 as an example, the parameters of the kinetic model of conjugated linoleic acid isomerase catalysis were solved by genetic algorithm, and the growth kinetic model, CLA synthesis kinetic model and substrate consumption kinetic model of the mutant strain were derived, describing the synthesis law of conjugated linoleic acid isomerase. The fit of the constructed kinetic model of enzyme synthesis was verified, and the CLA conversion effects of three linoleic acid isomerase-catalysed reaction systems, including phosphate buffer system, citrate buffer system and acetate reaction system, were compared. The effects of reaction temperature, substrate concentration and pH on the catalytic efficiency of CLA were analysed on the basis of the phosphate buffer system, respectively, and finally the pathway of linoleic acid enzyme-catalysed biosynthesis of CLA was optimally designed by orthogonal tests.

2

Conjugated linoleic acid biosynthesis

2.1

Conjugated linoleic acid

Conjugated linoleic acid (CLA) is an unsaturated fatty acid, a general term for all stereo and positional isomers of linoleic acid, with the molecular formula C17H31COOH.The double bonds of conjugated linoleic acid can be located in the 7 and 9, 8 and 10, 9 and 11, 10 and 12, 11 and 13, 12 and 14 positions, where each of the double bonds in turn has a cis (cis or c) and trans (trans or t) conformation. . To date, 28 conjugated linoleic acids are known, of which cis-9, trans-11 (c9t11CLA) and trans-10, cis-12 (t10c12CLA) are the most common among naturally occurring CLAs, respectively accounting for 85% and 10%. Naturally occurring CLA is mainly found in the rumen of ruminants and to a lesser extent in other animals and plants, thus dietary CLA is mainly derived from meat and dairy products of ruminants. In recent decades, CLA has been found to have positive effects in anti-cancer, anti-inflammatory, anti-obesity, anti-atherosclerosis, anti-diabetes, and osteogenesis, etc. Although the physiological mechanisms of these beneficial effects on human health are still not fully understood, it has been found in the course of the research that the CLA that plays a role in this process involves mainly the two isomers of c9t11 and t10c12. A lot of research has been carried out by scientists in order to develop techniques to produce high purity c9t11-CLA and t10c1CLA in large quantities [17].

2.2

Preparation of conjugated linoleic acid

2.2.1

Chemical synthesis

The most common method of chemical synthesis of CLA is to produce CLA by alkaline isomerisation of LA-rich oil crops such as safflower seed oil, sunflower seed oil, tobacco seed oil, chilli seed oil, etc. However, there are many problems with alkaline isomerisation: (1) large amounts of acid and alkali are required. (2) It is difficult to separate the physiologically active c9,t11-CLA and t10,c1CLA isomers from them after the reaction, resulting in low product purity. (3) The low heat transfer efficiency of the viscous mixture during the reaction limits the production scale. In addition to alkali isomerisation, metals such as Ru, Ni, Ru-MgAl, Ba, and Ru/MgZrO2 can be used as catalysts to efficiently catalyse CLA synthesis. However, such metal catalysts are expensive, difficult to recycle and the catalytic efficiency decreases with repeated use. In addition, iodine can also be used as a photosensitiser to generate CLA by photolysis of soybean oil using sunlight; this reaction has a low temperature and does not require expensive reactors and catalysts, but the efficiency of this reaction is low, and the content of c9,t11-CLA and t10,c1CLA is low. In addition, chemical synthesis of CLA may produce by-products that are harmful to humans.

2.2.2

Biosynthesis

In contrast, microbial synthesis of CLA has many advantages. Microbial culture is flexible and convenient, and many microorganisms are able to use their linoleic acid isomerase enzyme to exclusively convert to physiologically active CLA, which have been found to be Ruminalia, Lactobacillus, Propionibacterium, and so on. Currently, Lactobacillus is the more desirable strain, which contains linoleic acid isomerase that can convert linoleic acid into biologically active CLA isomers. At present, screening and improvement of highly active linoleic acid isomerase-producing microbial strains as well as optimisation of fermentation culture conditions are the important contents and research hotspots of the current bioisomerisation method for CLA synthesis. And the use of genetic engineering technology to modify microbial strains, improve linoleic acid isomerase activity or cell wall spreading to express linoleic acid isomerase to simplify the CLA extraction process will further increase the prospect of industrial application of CLA synthesis by bioisomerisation method [18].

3

Catalytic kinetic modelling of isomerases based on genetic algorithms

3.1

Genetic Algorithm Solution Steps

Genetic algorithm is a random iteration and evolution based on natural selection and population genetics mechanism, with a wide applicability of the search method, with a strong global optimisation search capability. The basic problem-solving idea of genetic algorithm is to start from a population that represents a possible potential solution to the problem, a population is composed of a certain number of individuals after coding each individual is actually a chromosome with the characteristics of the entity chromosome as the main carrier of the genetic material that is, a collection of multiple genes, the internal performance is a combination of certain genes determines the shape of the individual's external performance, so the beginning of the need to achieve from the The mapping from expression to genotype, i.e. coding, is done after the initial population is created, and the principle of survival of the fittest and survival of the fittest is followed to produce better and better approximations of the solution generation by generation In each generation, individuals are selected according to the fitness of the individuals in the problem domain, and with the help of natural genetics genetic algorithms, combinatorial crossover and mutation are carried out to produce populations that represent the new set of solutions The process will result in the subsequent generations of populations as in the case of natural evolution The population is more adapted to the environment than its predecessors and finally the optimal individuals in the population are decoded to be the optimal solution to the problem [19].

A variety of different genetic algorithms are formed by different encoding methods and different genetic operators. However, these genetic algorithms have a common feature, which is to complete the adaptive search process for the optimal solution of the problem by mimicking the mechanism of selection, crossover and mutation in the process of biological inheritance and evolution. Based on this common feature, Goldberg summarised and proposed a unified most basic genetic algorithm - the basic genetic algorithm (SGA).

Components of the basic genetic algorithm:

1)

Chromosome coding methods. The basic genetic algorithm uses fixed-length strings of binary symbols to represent individuals in a population whose alleles are made up of binary-valued symbol sets {0,1}.

2)

Individual fitness evaluation. The basic genetic algorithm decides how much chance each individual in the current population has to be inherited into the population of the next generation by a probability proportional to the individual fitness. To calculate this probability correctly, it is required here that the fitness of all individuals must be positive or zero. In order to satisfy the requirement that the fitnesses take non-negative values, the basic genetic algorithm generally transforms the objective function value f (X) into the fitnesses of the individuals F(X) using one of the following two methods.

Method 1: For the optimisation problem of finding the maximum value of the objective function, the transformation method is: 1 $F (X) = {\begin{matrix} f (X) + C_{\min}, i f f (X) + C_{\min} > 0 \\ 0, i f f (X) + C_{\min} \leq 0 \end{matrix}$

Eq:

C_min - a suitably relatively small number, which can be chosen in several ways: a pre-specified smaller number, the minimum objective function value up to the current generation of evolution, the minimum objective function value in the current generation or in the population of recent generations.

Method 2: For the optimisation problem of finding the minimum value of the objective function, the transformation method is: 2 $F (X) = {\begin{matrix} C_{\max} - f (X), i f f (X) < C_{\max} \\ 0, i f f (X) \geq C_{\max} \end{matrix}$

Eq:

C_max - an appropriately relatively small number, which can be chosen in several ways: a prespecified larger number, the maximum objective function value up to the current generation of evolution, the maximum objective function value in the current generation or in the population of recent generations.

3)

Genetic operators. The basic genetic algorithm uses the following three genetic operators:

(1)

The selection operation uses the proportional selection operator. The proportional selection operator means that the probability of an individual being selected and passed on to the next generation of the population is proportional to the fitness of that individual. The specific implementation process of the proportional selection operator is: firstly, the sum of the fitness of all individuals in the population is calculated, secondly, the size of the relative fitness of each individual is calculated, that is, the probability of each individual to be inherited into the next generation of the population, and finally, the simulated roulette board operation (i.e., a random number between 0 and 1) is used to determine the number of times each individual is selected.

(2)

The crossover operation uses the one-line crossover operator. The specific implementation process of the single crossover operator is: first of all, two individuals in the population of two randomly paired, followed by each pair of individuals paired with each other randomly set a certain position after the genome as the crossover point, and finally each pair of individuals paired with each other in accordance with the set crossover probability of p_c at the crossover point of the exchange of part of the chromosomes of the two individuals, resulting in the creation of two new individuals.

(3)

The mutation operation uses the basic positional mutation operator. The specific implementation process of the basic position mutation operator is: firstly, each locus of an individual is designated as a mutation point according to the mutation probability p_m, and secondly, each designated mutation point is inverted or replaced by other allele values, and finally, a new individual is generated.

4)

Operating parameters of the basic genetic algorithm.

(1)

M : Population size, i.e. the number of individuals contained in the population, generally taken as 20~100.

(2)

T : The number of evolutionary generations of termination of the genetic operation, generally taken as 100~500.

(3)

p_c : Crossover probability, generally taken as 0.4~0.99.

(4)

p_m : mutation probability, generally taken as 0.0001~0.1.

Formal definition of the basic genetic algorithm: the basic genetic algorithm can be defined as an 8-tuple: 3 $S G A = (C, E, P_{0}, M, Φ, Γ, Ψ, T)$

Eq: C -

the coding method of the individual, the

E -

individual fitness evaluation function, the

P₀ -

initial population.

M -

population size.

Φ -

selection operator, the

Γ -

crossover operator.

Ψ -

variation operator.

Τ -

Genetic operator termination condition.

For a real-world application problem that requires optimisation calculations, a genetic algorithm for solving the problem can generally be constructed by following the steps described below.

Step 1:

Determine the decision variables and their various constraints, i.e., determine the individual expression type × and the solution space of the problem.

Step 2:

Establish the optimisation model, i.e. determine the type of objective function (whether it is the maximum or the minimum of the objective function) and the form of mathematical description or quantification.

Step 3:

Determine the chromosome coding method that represents the feasible solution, and identify the individual genotype × and the search space of the genetic algorithm.

Step 4:

Determine the decoding method, and determine the correspondence or conversion relationship from individual genotype X to individual phenotype X.

Step 5:

Determine the quantitative evaluation method of individual fitness, and determine the conversion rules from the objective function value f(X) to the individual fitness F(X).

Step 6:

Design genetic operators, and determine the specific operation methods of genetic operators such as selection operation, crossover operation and mutation operation.

Step 7:

Determine the relevant operating parameters of the genetic algorithm, and determine the parameters M, T, p_c and p_m of the genetic algorithm.

In the process of constructing the basic genetic algorithm, it can be seen that the encoding method of feasible solutions and the design of genetic operators are the two main issues to be considered when constructing the genetic algorithm, and they are also the two key steps in designing the genetic algorithm. The basic genetic algorithm is run according to the following steps after it is constructed successfully.

Step 1:

Initialisation. Set the evolutionary generation counter t ← 0, set the maximum evolutionary generation T, and randomly generate M individuals as the initial population P(0).

Step 2:

Individual evaluation. Calculate the fitness of each individual in population P(t).

Step 3:

Selection operation. Based on the fitness of each individual obtained in the previous step and according to some rules, select the individuals into the next generation.

Step 4:

Crossover operation. Perform crossover operations on the individuals in the population according to the designed crossover operator to generate new individuals.

Step 5:

Variation operator. The designed mutation operator is applied to the population. Population P(t) undergoes selection, crossover and mutation operations to get the next generation population P(t + 1).

Step 6:

Termination condition judgement. If t ≤ T, then: t ← t + 1, go to step 2, if t > T, then take the individual with the maximum fitness obtained in the evolution process as the optimal solution output, terminate the calculation.

3.2

Isomerase catalytic kinetic model solving

3.2.1

Growth kinetics modelling of Lactobacillus acidophilus mutants

In this paper, three mathematical equations were chosen to model the enzyme production kinetics of Lactobacillus acidophilus mutant strain B-5. The most commonly used model for describing bacterial growth is the Monod equation, but it is not appropriate to use the Monod equation to describe the batch fermentation bacterial model because the limiting effects of substrate and bacterial concentration on growth during the batch fermentation process cannot be ignored [20].The logistic equation can well reflect the general law of bacterial growth during the batch fermentation process, and it is one of the equations that are more commonly used, commonly used to description of cell growth kinetics [21]. At the same time, since the cell growth curve is nearly S-shaped, its application to the growth of mutant strain B-5 leads to the Logistic integrable to algebraic equation for the growth of mutant strain B-5: 4 $\frac{d X}{d t} = μ_{m} X (1 - \frac{X}{X_{m}})$

Logistic equations can be integrated into algebraic equations: 5 $X = \frac{X_{0} X_{m} e^{μ_{m} t}}{X_{m} - X_{0} + X_{0} e^{μ_{m} t}}$

Genetic algorithms were applied to fit the algebraic equations Eq. (5) and the experimental values of bacterial growth non-linearly under the Matlab software platform. Parameter estimation of the kinetic model of bacterial growth using genetic algorithm resulted in: X₀ = 0.05325g/L, X_m = 7.17381, μ_m = 0.38152. Substituting the obtained parameter values into equation (5) there is a change in dry weight of the bacterium (X) as a function of fermentation time as: 6 $X = \frac{0.20418 e^{0.42582 t}}{7.14854 + 0.02848 e^{0.42582 t}}$

The correlation coefficient of this model was R² = 0.9971, indicating that 99.71% of the experimental data coincided with the model, so the equation has high rationality and credibility, and can better reflect the growth pattern of mutant strain B-5.

3.2.2

Kinetic modelling of CLA synthesis

In the process of microbial batch fermentation, the intracellular biosynthesis pathway is very complex, Gaden according to the relationship between the rate of product generation and the rate of cell growth, the kinetic model of the relationship between the formation of products and microbial cell growth is divided into three categories: I product formation and cell growth coupled type, II product formation and cell growth partially coupled type, product formation and cell growth non-coupled type. On this basis, Luedeking and Piret proposed the famous Luedeking-Piret equation to describe the relationship between product formation and cell growth.

Luedeking-Piret equation: 7 $\frac{d P}{d t} = α \frac{d X}{d t} + β X$

Where: α ≠ 0, β = 0 denotes class I fermentation, α ≠ 0, β ≠ 0 denotes class II fermentation, α = 0, β ≠ 0 denotes class III fermentation. From the results of the batch fermentation test of the mutant strain B-5, it can be judged that the fermentation process of CLA production belongs to class II fermentation, and the Luedeking-Piret equation can be integrated into an algebraic equation by combining with equation (7): 8 $P = p_{0} - α X + α \frac{X_{0} X_{m} e^{μ_{m} t}}{X_{m} - X_{0} + X_{0} e^{μ_{m!}}} + β \frac{X_{m}}{μ_{m}} \ln (\frac{X_{m} - X_{0} + X_{0} e^{μ_{m} t}}{X_{m}})$

P: CLA enzyme activity, U/L, p₀ : initial CLA enzyme activity, and

α: growth coupling coefficient.

β : non-growth coupling coefficient.

Since the effect of P₀ = 0, X₀ on the product at 0h is negligible, equation (8) can be simplified as: 9 $P = α \frac{X_{0} X_{m} e^{μ_{*}}}{X_{m} - X_{0} + X_{0} e^{μ_{m}}} + β \frac{X_{m}}{μ_{m}} \ln (\frac{X_{m} - X_{0} + X_{0} e^{μ_{m - 1}}}{X_{m}})$

Genetic algorithms were applied to nonlinearly fit the algebraic equations Eq. (8) and the experimental values of bacterial growth under the Matlab software platform, and the distribution of the optimal fitness for each generation and the average individual spacing for each generation converged after up to 100 generations. Parameter estimation of the kinetic model of bacterial growth using the genetic algorithm resulted in the following: α = 2.28982, β = 0.08928. Substitution of the obtained parameter values into Eq. (9) yielded the amount of CLA synthesis (P) as a function of fermentation time: 10 $P = \frac{0.49735 e^{0.42582 t}}{7.14817 + 0.02882 e^{0.42582 t}} + 1.4568 \times \ln \frac{7.14882 + 0.02833 e^{0.42582 t}}{7.1781}$

Eq. (10) can fit the obtained experimental number well enough, R² = 0.9942, which better depicts the synthesis of CLA during the fermentation process, CLA in the late stage of synthesis, the depletion of nutrients in the fermentation broth as well as the accumulation of other limiting factors, CLA viability has declined, so that the experimental value in the late stage of fermentation and the fitted value will be deviated from the fitted value.

3.2.3

Kinetic modelling of substrate consumption

During fermentation of the mutant strain B-5CLA, a carbon source is used as the limiting substrate, and its consumption is mainly used for growth of the bacterium as well as for the maintenance of the bacterial cells, and the substrate consumption can be expressed by an equation similar to that of Luedeking-Piret: 11 $- \frac{d S}{d t} = \frac{d X}{Y_{X / S} d t} + γ X$

Combining equations (7) and (8), the integral of equation (11) is obtained: 12 $S = S_{0} + X_{0} λ - λ \frac{X_{0} X_{m} e^{μ_{m} 1}}{X_{m} - X_{0} + X_{0} e^{μ_{κ} 1}} + γ \frac{X_{m}}{μ_{m}} \ln (\frac{X_{m} - X_{0} + X_{0} e^{μ_{m}} t}{X_{m}})$

Where Y_X/S : the rate constant of carbon source used for bacterial growth, λ = 1/Y_X/S.

γ: yield constant of carbon source used for product accumulation, h⁻¹.

Genetic algorithms were applied to nonlinearly fit the algebraic equations Eq. (12) and the experimental values of bacterial growth under the Matlab software platform. Parameter estimation of the kinetic model of bacterial growth using the genetic algorithm resulted in S₀ = 33.32727g/L, λ = 3.01727, γ = –0.03381. Substituting the obtained parameter values into Eq. (12) yielded the equation of substrate consumption variation with fermentation time as: 13 $S = 33.47828 - \frac{0.60135 e^{0.42582 t}}{7.14838 + 0.02899 e^{0.42582 t}} - 0.68458 \times \ln \frac{7.14884 + 0.02837 e^{0.42582 t}}{7.1838}$

Equation (13) provides a good depiction of substrate consumption during the synthesis of conjugated linoleic acid isomerase R² = 0.9979, with some deviation of the experimental values from the fitted values at the later stages of fermentation due to autolysis of a small amount of the bacterium at the later stages of fermentation.

4

Model validation and case analysis

4.1

Validation of the kinetic model of isomerase catalysis

According to the mathematical expression of the established kinetic model of enzyme catalytic reaction and the values of the model parameters, the fitted values of the changes of bacterial volume, residual sugar volume and CLA production with fermentation time during batch fermentation were obtained by applying origin software, and the results are shown in Table 1.

Table 1.

The comparison of computed and experimental value of dynamic model

Culture time(h)	Fungi dry weight(g/L)		CLA production(g/L)		Residual sugar(g/L)
Culture time(h)	Experimental value	Computed value	Experimental value	Computed value	Experimental value	Computed value
0	0.0067	0.0153	0.0000	0.0000	30.5400	30.5522
3	0.0712	0.0877	0.0000	0.0000	27.4436	27.4523
6	0.3390	0.3246	0.0000	-0.0107	23.5625	23.5694
9	0.7547	0.7545	0.0019	0.0039	20.7066	20.7165
12	1.1476	1.1485	0.0647	0.0651	17.5478	17.5601
15	1.2526	1.2370	0.0609	0.0666	16.1969	16.2030
18	1.2840	1.2849	0.1548	0.1601	14.5413	14.5275
21	1.2854	1.2748	0.2907	0.3125	13.5045	13.5213
24	1.3314	1.3409	0.3496	0.3370	13.0553	13.0245

As can be seen from Table 1, most of the fitted values were closer to the experimental values, and most of the points had smaller errors except for a few points with larger errors, which was a better fitting situation, indicating that the kinetic model constructed could better reflect the CLA batch fermentation process.

4.2

Optimisation of the pathway for the bio-enzyme catalysed synthesis of CLA

4.2.1

Selection of reaction system catalysed by linoleic acid isomerase

The CLA conversion systems were set as phosphate buffer system, citrate buffer system and acetate reaction system, the holding time was l h, the substrate concentration was l5 mg/mL, the reaction temperature was 36°, and the pH of the reaction system was 6.0. Three validation experiments were carried out with the predicted conversion rate of the mutant strain as a control, and the conversion rates of CLA were determined in different systems to get the conversion rate of CLA as shown in Figure 1. 1 shows.

In the process of conversion of LA to CLA by linoleic acid isomerase, the conversion of CLA was affected by the reaction system, and the binding efficiency of linoleic acid isomerase and substrate varied with different reaction systems. From Fig. 1, it can be seen that under the phosphate buffer system, linoleic acid isomerase converted the substrate to generate CLA with the best efficiency and the highest catalytic performance, which may be mainly due to the stronger dispersion of the substrate linoleic acid under the action of phosphate in the phosphate buffer system and thus better affinity with the linoleic acid isomerase, which in turn improves the conversion efficiency of CLA.

4.2.2

Effect of reaction temperature on the catalytic efficiency of CLA

The CLA conversion system was set as a phosphate buffer system, the holding time was 1 h, the substrate concentration was 15 mg/mL, and the pH of the reaction system was 6.0. The conversion rate of CLA at different temperatures was determined, and the effect of reaction temperature on the conversion rate of CLA is shown in Fig. 2, which demonstrates the violin plots plotted based on the three experiments, and the mean values of each group of data were calculated and connected with a straight line.

As can be seen in Fig. 2, from 26-36 °C, the catalytic efficiency of linoleic acid isomerase increased with the increase of reaction temperature, reaching a maximum value of 26.44% at 36 °C. With the further increase of temperature, the CLA conversion rate decreased sharply, which was due to the fact that enzyme-catalysed substrates have their optimal temperatures, at the optimal temperature, the reaction rate of linoleic acid isomerase combined with the substrate was faster and more efficient, and the active molecules inside the linoleic acid isomerase continued to increase as it carried out the enzyme reaction in the process of temperature increase. When the temperature rose to 36°C, the internal structure of the substrate affecting enzyme catalysis was changed, making the enzyme unstable in both geometric configuration and spatial structure, which in turn led to a decrease in its catalytic performance. Therefore, the optimal reaction temperature of linoleic acid isomerase generated by this mutant strain was 36°C.

4.2.3

Effect of substrate concentration on the catalytic efficiency of CLA

The CLA conversion system was set as a phosphate buffer system with a holding time of 1 h, a reaction temperature of 36 °C, and a reaction system pH of 6.0. The conversion rates of CLA at different substrate concentrations were determined, and the results of all the conversions at various concentrations are shown in Fig. 3, in which the solid line indicates the average conversion rate, and the dashed line is the connecting line for the specific experimental results. By adding different concentrations of LA to the reaction system, the highest conversion rate of CLA, 27.51%, was achieved when the LA concentration was 20 mg/mL. During the increase of LA concentration from 0 to 20 mg/mL, the conversion rate of CLA was increasing because the linoleic acid isomerase was sufficient at this time and the substrate LA concentration was not high. Linoleic acid isomerase is an inducible enzyme that will be continuously produced by enzyme-producing strains in the medium containing LA. Meanwhile, the concentration of LA has a great influence on the catalytic efficiency of linoleic acid isomerase when it reacts with the substrate LA. When LA accumulates and the concentration is too high, linoleic acid isomerase does not react sufficiently with the substrate LA, and its enzymatic reaction is inhibited, which leads to a decrease in the conversion rate of CLA.

4.2.4

Effect of pH on the catalytic efficiency of CLA

The reaction system was set to O.1 M different buffers: sodium citrate buffer solution (pH 3.0, 3.5, 4.0, 4.5). Sodium phosphate buffer solution (pH 5.0, 5.5, 6.0, 6.5, 7.0, 7.5). Sodium carbonate buffer solution (pH 8.0, 8.5, 9.0). The concentration of substrate LA was 0.6% (w/v), and the air in the test tube was driven off with high-purity nitrogen before the reaction. The experimental results are shown in Fig. 4. As seen in Fig. 4, the highest amount of CLA was generated at pH 6.0. Therefore, the optimum pH for the catalytic generation of CLA by linoleic acid isomerase is 6.0.

4.2.5

Optimisation of the pathway for the catalytic generation of CLA by linoleuclease in orthogonal tests

The results of the one-factor test showed that the main factors affecting the yield of CLA were 3 factors: linoleic acid concentration, pH and reaction temperature. Three levels were taken for each factor. The preparation of CLA orthogonal test protocol and results are shown in Table 2.

Table 2.

Orthogonal test of parameter optimization on producing CLA

Factor	A	B	C	CLA production (g/L)
Number	LA concentration (%)	Ph value	Temperature	CLA production (g/L)
1	1.5(1)	5(1)	30(1)	4.282±0.238
2	1.5(1)	6(2)	36(2)	6.564±0.342
3	1.5(1)	7(3)	40(3)	3.813±0.238
4	2(2)	5(1)	30(1)	6.727±0.381
5	2(2)	6(2)	36(2)	7.681±0.443
6	2(2)	7(3)	40(3)	5.317±0.452
7	2.5(3)	5(1)	30(1)	5.057±0.281
8	2.5(3)	6(2)	36(2)	5.813±0.387
9	2.5(3)	7(3)	40(3)	5.782±0.138
R	1.538	1.552	1.547
Superior level	A₂	B₂	C₂
Factor sort	C>A>B
Best combination	A₂ B₂ C₂

The table of orthogonal test results shows that the factors affecting the synthesis of CLA by mutant strain B-5 are in the following order of priority: C>A>B, i.e.: reaction temperature>substrate concentration>pH, with reaction temperature having the greatest effect on CLA yield.

Table 3 shows the analysis of variance (ANOVA) of the results of the orthogonal test. The ANOVA results showed that the effects of each factor reached significant differences (P<0.05). Combining the changes in the levels of the factors, the optimal combination of conditions for the synthesis of CLA by the mutant strain B-5 was obtained as A₂ B₂ C₂, i.e., substrate concentration of 2%, pH value of 6, and reaction temperature of 36°C.

Table 3.

Analysis of variance

Factor	Deviation squares	df	Mean square sum	F	Sig.	Best combination
A	3.64	2	1.82	72.8	<0.05	A₂
B	3.46	2	1.73	70.2	<0.05	B₂
C	4.76	2	2.38	87.9	<0.05	C₂
Test error	0.05	2	0.025
Sum.	11.91	8

5

Conclusion

The fitting test revealed that the established kinetic model of enzyme-catalysed reaction had a small fitting error and a good fitting effect, which was a guide for CLA production increase. Experiments showed that the use of phosphate buffer system in the conversion of LA to CLA by linoleic acid isomerase had the best effect on the improvement of CLA conversion efficiency. In addition, the maximum CLA catalytic efficiency was achieved at a reaction temperature of 36°C (26.44%), and the highest conversion of CLA (27.51%) was achieved at a substrate concentration of 20 mg/mL, and it was also found that the optimum pH for the catalytic generation of CLA by linoleic acid isomerase in mixed buffer was 6.0. The results of orthogonal experiments showed that the reaction temperature had the greatest effect on the yield of CLA, followed by the substrate concentration and pH. Analysis of variance (ANOVA) tested the optimal combination of conditions for CLA synthesis by mutant strain B-5, i.e., substrate concentration of 2%, pH of 6, and reaction temperature of 36°C. The results indicated that the conjugated substituents were produced under these conditions. It indicated that the theoretical synthesis yield of conjugated linoleic acid was maximum under this condition.

Language:: English

Publication timeframe:: 1 times per year
Journal Subjects:: Life Sciences, Life Sciences, other, Mathematics, Applied Mathematics, General Mathematics, Physics, Physics, other

Journal RSS Feed

Enzyme-catalyzed reaction path optimization and biosynthetic design based on genetic algorithm

Xiaoqian Zhang

Tong Sun

Zhen Xu

Haiying Sun

Xinyao Su

Published Online: Sep 23, 2025

Received: Jan 03, 2025

Accepted: Apr 30, 2025

DOI: https://doi.org/10.2478/amns-2025-1110

KeywordsGenetic algorithm, Conjugated linoleic acid isomerase catalysis, Pathway optimisation, Kinetic modelling

© 2025 Xiaoqian Zhang et al., published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Keywords
Genetic algorithm, Conjugated linoleic acid isomerase catalysis, Pathway optimisation, Kinetic modelling