Medium and culture method for increasing the biomass and beta-1,3-glucan production of leptomitus lacteus

Abstract

This invention discloses a culture medium and cultivation method for increasing the biomass and β-1,3-glucan yield of Euglena filamentosa. The invention utilizes literature datasets for preliminary modeling and variable screening, then verifies and corrects the model recommendations using experimental datasets, forming a closed-loop path of model-driven experimental optimization. This approach fully leverages the broad coverage of literature data while mitigating the uncertainties caused by the heterogeneity of literature data through experimental data. Ultimately, it predicts the optimal culture medium and cultivation method that balances biomass and β-1,3-glucan yield. Scale-up fermentation experiments in a 200 L fermenter show that at 264 h of fermentation, the biomass reached 17.32 ± 1.35 g / L, and the β-1,3-glucan yield reached 12.82 ± 1.40 g / L, significantly higher than those obtained using other culture media and conditions.

Description

Technical Field

[0001] This invention relates to a method for culturing Euglena spp., and particularly to a culture medium and method for increasing the biomass and β-1,3-glucan yield of Euglena spp., belonging to the field of Euglena spp. cultivation. Background Technology

[0002] Euglena slenderis ( Euglena gracilis Euglena spp. is a single-celled protist exhibiting physiological characteristics of both plants and animals. Its unique metabolic mode allows it to synthesize substances using light energy under light conditions, grow using external organic carbon sources under dark conditions, and maintain growth even when both modes coexist. Due to its high metabolic plasticity and product diversity, Euglena spp. is considered a highly promising new type of feed additive. It can be used to directionally synthesize and accumulate various high-value-added products with industrialization prospects, including proteins rich in essential amino acids, various vitamins and their precursors, functional lipids, and the unique energy-storing polysaccharide β-1,3-glucan. This polysaccharide consists of linear, unbranched β-1,3- D - It is composed of dextran. It is stored in the cytoplasm in the form of insoluble particles and can be rapidly degraded when external carbon sources are insufficient, providing energy for the cell. (Previous research progress) Aemiro et al. [5] Studies have shown that in vivo experiments in sheep using Euglena spp. as a dietary supplement can increase nutrient intake and improve nitrogen digestibility and retention. (Levine et al.) [6] Studies have shown that adding fine Euglena gracilis powder to broiler diets can improve feed conversion ratio and reduce intestinal lesion scores, while enhancing intestinal immune responses. Kim et al. (KIM K, EHRLICH A, PERNG V, et al. Algae-derived β-glucan enhanced gut health and immune responses of weaned pigs experimentally infected with apathogenic E. coli[J]. Animal Feed Science and Technology, 2019, 248: 114-125.) showed that adding algal-derived β-glucan to feed can alleviate diarrhea symptoms in weaned piglets. In recent years, β-glucan... - 1,3-glucan has attracted much attention due to its wide range of biological activities. Studies have shown that it can not only significantly activate the immune system and exert antibacterial effects, but also has a variety of health-promoting functions such as lowering serum cholesterol and regulating glucose metabolism (KAWANO T, MIURA A, NAITO J, et al. Immune profiling of paramylon-rich Euglena gracilis inhumans: a randomized controlled trial[J]. Journal of FunctionalFoods, 2023,109:105804; RUSSO R, BARSANTI L, EVANGELISTA V, et al. Euglena gracilis Paramylon activates human lymphocytes[J]. Food Science & Nutrition, 2017, 5(2):205–214). However, the biomass and β-1,3-glucan content of Euglena slenderis are not determined by a single factor. The culture process is affected by multiple factors, including light intensity, temperature, carbon and nitrogen source supply, carbon-nitrogen ratio, trace element content, initial inoculation density, and culture mode (batch, fed, continuous, etc.). There are significant nonlinear relationships and interactions among these factors. Different culture modes (photoautotrophic, mixed nutrition, heterotrophic) and combinations of temperature and light conditions can significantly change cell growth kinetics and composition (RODRÍGUEZ-ZAVALA JS, ORTIZ-CRUZ MA, MENDOZA-HERNÁNDEZ G, et al. Synthesis of α-tocopherol, paramylon and tyrosineby). Euglena gracilis under high biomass conditions[J]. Journal of AppliedMicrobiology,2010, 109(6):2160–2172; ŠANTEK B, FELSKI M, FRIEHS K, et al.Production of paramylon by heterotrophic Euglena gracilisOn potato liquor[J].Engineering in Life Sciences, 2010, 10(2):165–170.). Nitrogen limitation is often used to induce the accumulation of reserve substances, but its coupled effects on growth rate and product synthesis are very complex, and there is often a trade-off between high biomass and high polysaccharide content. In addition, long-term adaptation may occur under open or semi-continuous culture conditions, making the growth and β-1,3-glucan accumulation patterns of Euglena slenderis different from those of short-term experiments (YAMADA K, SUZUKI H, TAKEUCHIT, et al. Efficient selective breeding of live oil-rich Euglena gracilis With fluorescence-activated cell sorting[J]. Scientific Reports, 2016, 6:26327.). Traditional culture optimization usually relies on single-factor experiments or empirical adjustments. However, in the case of multi-factor coupling, strong interactions, and multiple objectives to simultaneously increase biomass and β-1,3-glucan production, single-factor optimization is difficult to cover all combination parameter spaces and find the global optimal solution (UDAYAN A, ARUN A, ARULSELVI PI, et al. Optimization strategies for enhanced microalgalbioresource production[J]. Renewable and Sustainable Energy Reviews, 2022,153:111766.).

[0003] To overcome the aforementioned bottlenecks, machine learning (ML) algorithms, especially decision tree-based ensemble models (such as Random Forest and XGBoost), have provided a new paradigm for microalgae cultivation optimization. These models can efficiently capture the complex nonlinear relationships between environmental parameters (such as light intensity, temperature, pH, and nutrient concentration) and algal cell growth and metabolism, accurately predicting key indicators such as biomass and β-1,3-glucan yield, and identifying the optimal parameter combinations for achieving optimal output. Compared to deep neural networks, decision tree-based ensemble models offer higher interpretability, can quantify key influencing factors, and are well-suited to small sample data. However, deep neural networks may have an advantage in prediction accuracy on very large datasets. This framework has successfully achieved significant increases in biomass and target product yield in other microalgae systems such as *Chlorella vulgaris*, demonstrating its feasibility as a precise optimization tool. However, there is a lack of ML optimization research specifically for *Euglena spp.*, and currently, there is a lack of effective culture media and cultivation methods to improve the biomass and β-1,3-glucan yield of *Euglena spp.* Summary of the Invention

[0004] The main objective of this invention is to construct a predictive model based on machine learning (ML) algorithms to predict the optimal culture medium and culture method for significantly improving key indicators such as biomass and β-1,3-glucan production of Euglena filamentosa, and to establish a fermenter-scale fermentation culture method that significantly improves the biomass and β-1,3-glucan production of Euglena filamentosa using the aforementioned optimal culture medium and culture method.

[0005] The above-mentioned objectives of the present invention are mainly achieved through the following technical solutions: Based on literature data, this invention explored the effects of three machine learning models—DT, RF, and XGBoost—on predicting the biomass and β-1,3-glucan yield of *Euglena spp.* The results showed that the XGBoost model performed best. Based on the model's output and the importance features obtained through SHAP analysis, and after laboratory validation, the optimal culture medium and method for significantly improving key indicators such as *Euglena spp.* biomass and β-1,3-glucan yield were finally obtained, thus completing this invention.

[0006] One aspect of the present invention is to provide a culture medium and a cultivation method for increasing the biomass and β-1,3-glucan yield of Euglena filamentosa, wherein the culture medium comprises: glucose 12.0 g / L, sodium acetate trihydrate 2.5 g / L, sodium nitrate 1.5 g / L, ammonium chloride 1.0 g / L, yeast extract 1.2 g / L, tryptone 1.3 g / L, beef extract 0.5 g / L, vitamin B1 2.0 mg / L, vitamin B6 3.0 mg / L, and vitamin B2. 12 4.0 μg / L.

[0007] In a preferred embodiment of the present invention, the culture conditions are: a culture temperature of 25-29°C, preferably 27.5°C.

[0008] In a preferred embodiment of the present invention, the cultivation conditions are as follows: the light intensity is 4000-5000 lux; preferably, the light intensity is 4500 lux.

[0009] In a preferred embodiment of the present invention, the culture conditions are: light time 12-16h / dark time 8-12h; preferably, the culture conditions are: light time 14h / dark time 10h.

[0010] The Euglena filamentosa mentioned in this invention can be any commercially available Euglena filamentosa or Euglena filamentosa preserved in a strain preservation center; for example, the Euglena filamentosa can be Euglena filamentosa with the microbial preservation number CGMCC NO.41188.

[0011] Another aspect of the present invention provides a fermentation method for increasing the biomass and β-1,3-glucan yield of Euglena filamentosa in a fermenter, wherein the culture medium comprises: glucose 12.0 g / L, sodium acetate trihydrate 2.5 g / L, sodium nitrate 1.5 g / L, ammonium chloride 1.0 g / L, yeast extract 1.2 g / L, tryptone 1.3 g / L, beef extract 0.5 g / L, vitamin B1 2.0 mg / L, vitamin B6 3.0 mg / L, and vitamin B2. 12 4.0 μg / L; culture temperature 27.5℃, light intensity 4500 lux, light duration 14 h / dark duration 10 h; In a preferred embodiment of the present invention, during fermentation culture, the initial inoculum volume of the seed culture is 15%, the working volume of the fermenter is 120L, and after three days of light culture to accumulate biomass, polysaccharide enrichment is carried out; the aeration rate is controlled at 3m³ / h, and DO is automatically controlled by adjusting the stirring and aeration in conjunction. In a preferred embodiment of the present invention, glucose is fed once on the 3rd, 5th and 7th days of fermentation culture. The feed solution is a 500 g / L glucose solution, and the volume of each addition is 3.6 L.

[0012] In a preferred embodiment of the present invention, the stirring speed is dynamically adjusted within the range of 80–200 r / min to maintain DO at no less than 20% and avoid growth restriction or insufficient mass transfer caused by hypoxia.

[0013] The results of the scale-up fermentation experiment using the fermenter described above in this invention show that, at 264 hours of fermentation, the biomass reached 17.32 ± 1.35 g / L and the β-1,3-glucan yield reached 12.82 ± 1.40 g / L, which are significantly higher than the biomass and β-1,3-glucan content of Euglena filamentosa cultured using other culture media and conditions.

[0014] This invention systematically compared various machine learning models based on a literature dataset and ultimately selected XGBoost as the primary prediction model for subsequent key factor analysis and culture condition parameter recommendation. Gradient boosting trees achieve function approximation by progressively fitting residuals, effectively characterizing nonlinear response relationships and potential threshold effects under multi-factor coupling conditions. This characteristic makes it more advantageous than linear models or single decision trees when dealing with the combined effects of culture medium components, environmental parameters, and their effects. In the development of microalgae cultivation and fermentation processes, culture medium components and environmental factors often change simultaneously, and the range of variable values ​​varies significantly between different studies, leading to certain correlations and interactions between variables. Simply relying on empirical judgment or single-factor experiments is insufficient to fully reflect the comprehensive impact of multi-variable synergistic effects on biomass and metabolite accumulation. The model construction results of this invention show that XGBoost's fitting accuracy on the test set for both biomass and β-1,3-glucan yield prediction targets is superior to decision trees and random forests, indicating that this model can better adapt to the complex and heterogeneous parameter distribution characteristics in the literature dataset, providing a reliable modeling foundation for subsequent variable selection and parameter optimization. At the same time, the application of machine learning models in the optimization of fermentation processes not only depends on prediction accuracy, but also requires a certain degree of interpretability to support experimental design and parameter decision-making.

[0015] The predictive model constructed in this invention demonstrates that process factors such as carbon source level, culture time, temperature, and light environment have significant impacts on Euglena gracilis biomass and β-1,3-glucan production. These factors exhibit high stability in both literature and laboratory datasets, suggesting they may have common regulatory roles in different culture systems. Glucose and related carbon source variables consistently rank among the top contributors to biomass and polysaccharide predictions. From a model perspective, the high contribution of carbon source variables not only reflects their direct role as energy and carbon skeleton sources but also indicates that, in the literature datasets, differences in culture media among different studies largely stem from variations in carbon source supply levels. Therefore, the model tends to identify carbon source as one of the key factors explaining yield differences.

[0016] The importance of culture time, temperature, and light environment in the model during fermentation suggests that the fermentation process has distinct stage characteristics. The model interpretation results indicate that the contribution of culture time does not simply reflect "longer culture, higher yield," but more likely represents the differential effects of different culture stages on biomass growth and energy storage polysaccharide (β-1,3-glucan) deposition. This provides a basis for setting fixed photoperiods and culture durations in subsequent laboratory datasets.

[0017] Besides conventional culture medium optimization, adjustments to substrate composition in mixed nutrient or complex culture systems can simultaneously affect cell growth and metabolite formation. This invention incorporates inorganic, organic, and compound nitrogen sources in a comprehensive formulation to balance biomass and polysaccharide yield. Furthermore, while vitamin factors did not consistently rank highest in contribution in the model interpretation, they showed a stable contribution in laboratory datasets. From the perspective of model and experimental operability, including vitamin factors as core control variables helps improve the stability and reproducibility of the system in subsequent culture condition optimization.

[0018] Based on the overall prediction results, carbon source supply, process conditions, and nutrients such as nitrogen source and vitamins showed good consistency across different datasets and model analyses, laying the foundation for further variable interaction analysis and integrated culture medium design.

[0019] In multi-factor fermentation systems, the responses of biomass growth and β-1,3-glucan accumulation to culture conditions are not determined independently by a single factor, but are influenced by the synergistic effects of multiple nutritional and environmental factors. Based on single-factor contribution analysis, this experiment further utilized the SHAP method to evaluate the interaction effects between variables, in order to identify factor combinations that are significant for yield prediction. The interaction importance results showed that in both the biomass and β-1,3-glucan prediction models, "glucose × sodium nitrate" was the interaction term with the highest contribution, significantly higher than other variable combinations. This result indicates that the ratio of carbon source to inorganic nitrogen source plays a central role in the model, and its synergistic changes have a significant impact on both biomass formation and polysaccharide accumulation. Compared to simply increasing the level of carbon or nitrogen source, this interaction characteristic better reflects the comprehensive effect of the overall composition of the culture medium on cell metabolic state. In addition to the carbon-nitrogen interaction, the interaction between nitrogen source and environmental conditions also showed a certain contribution. For example, the "sodium nitrate × culture temperature" factor ranked highly in both the biomass model and the β-1,3-glucan model, suggesting that the nitrogen supply effect may adjust with changes in culture temperature, thereby affecting cell growth efficiency and metabolite accumulation levels. This phenomenon emphasizes that during the optimization of culture conditions, the nitrogen source concentration needs to be matched with environmental parameters, rather than being adjusted as an isolated variable. Among the light-related variables, the combined interaction of light intensity and light duration contributed to both target models, reflecting the synergistic regulatory effect of light conditions as an environmental input on the culture system. Compared to culture time itself, the interaction between light parameters is more directly operable and more conducive to fine-tuning in experimental design. Although culture time and multiple variables showed some contribution in the interaction analysis, culture time mainly reflects the stage-based cumulative effect of the fermentation process, and its interaction characteristics are difficult to use as an independent control method in actual operation. Therefore, in the discussion of interaction effects and comprehensive culture medium design, this invention focuses more on the interaction relationships between directly controllable variables such as carbon source, nitrogen source, and environmental conditions.

[0020] In summary, the interactive SHAP analysis results indicate that the synergistic configuration of carbon and nitrogen sources in the culture medium composition, as well as their matching with key environmental factors, is a crucial foundation for achieving the dual objectives of synergistically increasing biomass and β-1,3-glucan yield. Based on this, further optimization of the combined culture medium and conditions will help improve the stability and feasibility of the model-recommended scheme in experimental validation. Based on the screening results of both objectives, the optimal culture medium formulation that balances biomass and β-1,3-glucan yield was ultimately predicted and determined to be: glucose 12.0 g / L, sodium acetate trihydrate 2.5 g / L, sodium nitrate 1.5 g / L, ammonium chloride 1.0 g / L, yeast extract 1.2 g / L, tryptone 1.3 g / L, beef extract 0.5 g / L, vitamin B1 2.0 mg / L, vitamin B6 3.0 mg / L, and vitamin B2... 12 4.0 μg / L; the optimal culture conditions were: culture temperature 27.5℃, 4500 lux, light time 14 h, dark time 10 h.

[0021] This invention uses literature datasets for preliminary modeling and variable screening, and then uses experimental datasets to verify and correct the model recommendations, forming a closed-loop path of model-driven experimental optimization. This fully utilizes the advantage of the wide coverage of literature data, while using experimental data to compensate for the uncertainty caused by the heterogeneity of literature data. Finally, it predicts and determines the optimal culture medium and culture method that balances biomass and β-1,3-glucan yield. The results of scale-up fermentation of the predicted optimal culture medium and culture method in a 200 L fermenter show that at 264 h of fermentation, the biomass reached 17.32 ± 1.35 g / L, and the β-1,3-glucan yield reached 12.82 ± 1.40 g / L, which are significantly higher than the biomass and β-1,3-glucan content of Euglena filamentosa cultured using other culture media and culture conditions. Attached Figure Description

[0022] Figure 1 shows the results of Pearson correlation analysis of continuous variables in the literature dataset.

[0023] Figure 2 To assess the predictive performance of different models on Euglena slenderis biomass and polysaccharide production.

[0024] Figure 3 The results of SHAP analysis show the importance of characteristics during the culture of Euglena filamentosa.

[0025] Figure 4 The predictive performance of Euglena filamentosa on biomass and polysaccharide production in the experimental dataset was evaluated.

[0026] Figure 5The SHAP analysis results contributed to the single-factor analysis of the Euglena filamentosa experimental data.

[0027] Figure 6 The results of SHAP analysis were used to analyze the experimental data of Euglena spp. Detailed Implementation

[0028] The present invention will be further described below with reference to specific experimental examples, and the advantages and features of the present invention will become clearer with the description. However, these experimental examples are merely exemplary and do not constitute any limitation on the scope of the present invention. Those skilled in the art should understand that modifications or substitutions to the details and form of the present invention can be made without departing from the spirit and scope of the present invention, but all such modifications and substitutions fall within the protection scope of the present invention.

[0029] Example 1: Construction of a predictive model for Euglena biomass and β-1,3-glucan production, and prediction of optimal culture medium and conditions. 1. Experimental Methods 1.1 Data Sources and Dataset Construction 1.1.1 Literature Dataset (Dataset 1) This study investigated the biomass and β-1,3-glucan (parastarch / polysaccharide) during the cultivation of Euglena filamentosa. Literature searches were conducted in databases such as Web of Science, Google Scholar, and Scopus, with keywords including " Euglena gracilis The search scope included literature on the following terms and combinations: *Euglena slenderis*, *biomass*, *paramylon*, and β-1,3-glucan. The search period was from 1984 to 2022. Inclusion criteria were: literature providing clearly defined culture conditions (environmental parameters, culture medium components, culture system, etc.) and reporting at least one response indicator (biomass and β-1,3-glucan). Exclusion criteria were: missing key culture conditions or target indicators, incompatibility in unit conversion, only qualitative descriptions, or data that could not be verified.

[0030] Data was extracted from literature tables and text. When data was presented only graphically, PlotDigitizer 2.6.9 was used for digitization, and units and numerical ranges were verified. Initially, 97 records were obtained, and after cleaning, 93 were retained for modeling. Input variables included environmental and process parameters (temperature, culture time, initial pH, light intensity, light duration, initial inoculum size, culture device type, etc.), carbon / organic substrate, nitrogen / phosphorus sources and compound nutrient sources, trace elements / chelating agents, and vitamins. Output variables were biomass (g / L) and β-1,3-glucan index (g / L).

[0031] To characterize the overall culture conditions, derived features were constructed based on the original variables: light dose (lux·h), total carbon source (g / L), total nitrogen source (g / L), and C / N ratio (total carbon source / total nitrogen source). Categorical variables (such as culture device type) were numerically coded, while "whether added" variables such as trace elements / chelating agents were coded using binary codes (0 / 1). Variable names, units, and ranges are shown in Table 1.

[0032] Table 1 Input variables, types, and amounts of the literature dataset

[0033] 1.1.2 Laboratory Dataset (Dataset 2) To conduct controlled validation and establish a data foundation for model validation and parameter recommendation, a laboratory dataset (778 records, 17 variables) was constructed. The experimental design fixed the culture apparatus (specifications / types) and trace element levels. Initial inoculum size, light intensity, light duration, carbon and nitrogen source formulations (types and concentrations), and vitamins B1, B6, and B6 were also included. 12 The amount added was used as a controllable variable combination. The output index was the correlation between biomass and β-1,3-glucan production. The determination method is described in 1.2.

[0034] 1.2 Indicator Measurement Methods 1.2.1 Biomass determination After the culture is completed, take 5 mL of algal solution and add it to a pre-weighed centrifuge tube (empty tube mass). m １ The algal cell pellet was collected by centrifugation at 8,000 × g for 10 min. The pellet was washed with double-distilled water and centrifuged twice more. After freeze-drying the pellet to constant weight, the total mass of the centrifuge tube and the algal cells was weighed. m 2).

[0035] DW (g / L)=( m 2 -m 1 )×200 / V (5 mL).

[0036] in: DW Biomass (g / L); m 1 represents the mass of the centrifuge tube (g); m 2 represents the total dry weight (g) of the centrifuge tube and algal cells; V The sampling volume is 5 mL. 1.2.2 Determination of β-1,3-glucan Collect algal cells by centrifugation with 5 mL of algal solution, wash twice with deionized water, freeze-dry, and weigh (empty tube mass). m1. Quality of "tube + algal cells" after freeze-drying m 2). The lyophilized samples were processed sequentially as follows: washed once with deionized water. Washed twice with 1 mL of 99.9% anhydrous ethanol to remove lipids; all centrifugation steps were performed at 7500×g for 10 min. To remove insoluble proteins, the cell pellet was added to 1 mL of 100 mM sodium phosphate buffer (containing 1 g / L trypsin) and incubated overnight at 37 °C. After centrifugation and discarding the supernatant, the sample was washed twice with 1 mL of 500 g / L urea solution to remove protein hydrolysis products. Finally, the sample was washed twice with deionized water, pre-frozen, lyophilized, and weighed to obtain the total mass of "tube + β-1,3-glucan". m 3 (GRIMM P, RISSE JM, CHOLEWA D, et al. Applicability of Euglena gracilis for biorefineries demonstrated by the production of α-tocopherol and paramylon followed by anaerobic digestion[J]. Journal of Biotechnology, 2015, 215:72–79.).

[0037] β-1,3-glucan (g / L)=( m 3- m 1)×200 / V (5 mL).

[0038] in, m 3 represents the total mass (g) of the centrifuge tube and the starch byproduct; m 1 represents the mass of the centrifuge tube (g); V The sampling volume is 5 mL. 1.3 Machine Learning Model Construction and Evaluation 1.3.1 Prediction Target and Data Division Independent models were constructed using biomass (biomass_gL) and β-1,3-glucan yield (β-1,3-glucan_gL) as regression prediction targets, respectively. The dataset was randomly divided into training and test sets in an 8:2 ratio, with the test set used only for final performance evaluation.

[0039] 1.3.2 Model Establishment and Hyperparameter Optimization This study compares three models: Decision Tree (DT), Random Forest (RF), and Extreme Gradient Boosting (XGBoost). A pipeline workflow including data preprocessing and model building is employed. Five-fold cross-validation is performed on the training set, and hyperparameter optimization is performed using Randomized SearchCV with a fixed random seed to ensure repeatability. After obtaining the optimal parameter combination, the model is retrained on the training set data and evaluated on the test set.

[0040] 1.3.3 Evaluation Indicators Model performance is primarily evaluated using the coefficient of determination (R²) and root mean square error (RMSE). The calculation formulas are as follows:

[0041] in, y i Let i be the measured value of the i-th sample. ŷ i Let be the predicted value for the i-th sample. m This represents the number of samples.

[0042] The coefficient of determination (R²) ranges from (-∞, 1]. R A higher 2 indicates a better model fit, while a lower root mean square error (RMSE) indicates better model prediction performance.

[0043] 1.4 Model Interpretation and Statistical Processing 1.4.1 SHAP Interpretability Analysis SHAP was used to perform interpretability analysis on the tree model (Lundberg SM, Lee S I. A unified approach to interpreting model predictions[C]. Advances in Neural Information Processing Systems 30. Long Beach: Curran Associates, 2017:4765–4774.). The contribution value (SHAP value) of each feature to the prediction of a single sample was calculated using TreeExplainer from the SHAP library. The mean absolute SHAP value (mean|SHAP|) was used to characterize global importance and was used for feature ranking. The effects of key features were further analyzed using dependency graphs and other methods.

[0044] 1.4.2 Statistics and Software Correlation analysis of input variables used the Pearson correlation coefficient, with a significance level of 100%. P<0.05. Data preparation, modeling, and interpretability analysis were completed in a Python 3.13.5 environment, mainly using libraries such as pandas, numpy, scikit-learn, xgboost, and shap.

[0045] 2. Experimental Results 2.1 Correlation Analysis of Variables in the Literature Dataset (Pearson) Before model building, Pearson correlation analysis was performed on continuous variables in the literature dataset to assess the linear correlation between variables. Figure 1 Overall, the correlations among most variables were weak, indicating that components such as carbon sources changed largely independently in the experiment. However, a clear "modular" correlation structure was observed: organic nitrogen sources showed strong positive correlations, while inorganic nitrogen sources showed high correlations and were generally strongly negatively correlated with organic nitrogen sources, suggesting that organic and inorganic nitrogen sources may have been designed as mutually exclusive or linked gradients in the experimental design. In addition, vitamins B1, B6, and B... 12 The variables also exhibit highly synchronized changes, suggesting their inclusion as a linkage factor in the formulation. This correlation structure indicates significant collinearity and coupling changes in the input features, providing a basis for subsequent model construction and feature processing (such as avoiding unstable coefficients in linear models, using tree models, or merging redundant variables).

[0046] 2.2 Comparison of Models and Determination of Master Model in Literature Dataset (Dataset 1) Based on a literature dataset (93 cleaned valid samples), decision tree (DT), random forest (RF), and XGBoost regression models were constructed with biomass and β-1,3-glucan yield (g / L) as prediction targets, respectively, and their predictive performance was compared on the test set. Figure 2 Coefficient of determination on the comprehensive test set () R ²) In terms of both root mean square error (RMSE) and overall performance, XGBoost outperforms DT and RF, and was therefore selected as the master model for subsequent analysis and parameter recommendation.

[0047] XGBoost test set for biomass prediction R The value was 0.939, and the RMSE was 1.243 g / L; in the test set for β-1,3-glucan prediction. R The mean square value was 0.914 and the RMSE was 1.002 g / L, indicating good fitting ability and generalization performance.

[0048] 2.3 Identification of Key Influence Factors Based on XGBoost (Dataset 1) To clarify the differences in the contributions of cultivation conditions to the two response variables in the literature data, the SHAP method was used to perform interpretability analysis on the XGBoost model trained on dataset 1, and the global contribution of the features was represented by the mean absolute SHAP value (mean|SHAP|). Figure 3 ).

[0049] The prediction results showed that in the biomass model, glucose concentration, culture time, and culture temperature were the three factors with the highest contribution, followed by total nitrogen source level, light dose, and pH. In the β-1,3-glucan model, glucose concentration and culture time were also among the top contributors, while culture temperature, C / N ratio, and total carbon source level also showed some influence. It is worth noting that the initial inoculum size was related to vitamins (B1, B6, B1, B2). 12 Although not ranked highest in contribution, as an "initial state variable" and "metabolic cofactor supply variable" in the culture system, it may play a crucial fundamental regulatory role in final biomass and polysaccharide accumulation by influencing early growth kinetics, substrate consumption rhythm, and carbon and nitrogen metabolic flux allocation, and interact with dominant factors such as carbon source, nitrogen source, and light. Therefore, considering both contribution ranking and operability, the initial inoculum size, photoperiod, light intensity, carbon source, nitrogen source, and vitamins B1, B6, and B1 were ultimately selected. 12 As a core control variable, targeted laboratory culture experiments were conducted.

[0050] 2.4 Experimental Dataset (Dataset 2) XGBoost Model Prediction Performance On the experimental dataset (dataset 2,778 samples), an XGBoost regression model was built using the same modeling workflow as the literature data, and its predictive performance was evaluated on the independent test set. Results showed that the biomass model on the test set… R ² was 0.817, RMSE was 0.886 g / L; β-1,3-glucan model test set R The mean square value was 0.750, and the RMSE was 0.632 g / L. Compared to the literature dataset, the experimental data conditions were more controllable, and the model still maintained good predictive ability on this dataset, indicating that the XGBoost model has a certain degree of robustness and can be used for subsequent parameter selection and combination optimization.

[0051] 2.5 Single-factor contribution and interaction analysis (SHAP) of dataset 2 Based on the XGBoost model in dataset 2, the SHAP method was further used to analyze the univariate contribution. Figure 5 ) and interactions between variables ( Figure 6The univariate SHAP results showed that in the biomass model, variables such as sodium nitrate, vitamin B1, glucose, light duration, and initial inoculum size had high contributions; in the β-1,3-glucan model, glucose, light duration, sodium nitrate, and initial inoculum size were the main contributing factors, while sodium acetate and vitamin B1 were the least contributing factors. 12 Secondly, the contributions of the other variables were relatively low.

[0052] Interaction analysis results showed a high degree of consistency in the ranking of interaction contributions between the two target models. The interaction with the highest contribution was "glucose × sodium nitrate," followed by "sodium nitrate × culture time" and "glucose × culture time." Furthermore, interactions related to process parameters were also prominent, such as "light intensity × culture time," "sodium nitrate × culture temperature," and "light duration × initial inoculum size," all ranking highly. These results suggest that, under laboratory conditions, there is a significant coupling effect between substrate supply and culture process factors, representing an important model characteristic influencing biomass and β-1,3-glucan yield.

[0053] 2.6 Parameter recommendation, comprehensive culture medium determination and experimental validation based on XGBoost (Dataset 2) Based on the XGBoost model in Dataset 2, a sampling grid search was performed on 15 controllable input variables to screen for potential high-yield cultivation combinations. Considering the theoretical number of combinations is as high as 10... 15 This experiment employed a sampling evaluation strategy, predicting yields from 30,000 candidate combinations and identifying the top 10% of predicted combinations as high-yield candidate regions. Based on the screening results considering both objectives, the optimal culture medium formulation, balancing biomass and β-1,3-glucan yield, was ultimately predicted to be: glucose 12.0 g / L, sodium acetate trihydrate 2.5 g / L, sodium nitrate 1.5 g / L, ammonium chloride 1.0 g / L, yeast extract 1.2 g / L, tryptone 1.3 g / L, beef extract 0.5 g / L, vitamin B1 2.0 mg / L, vitamin B6 3.0 mg / L, and vitamin B2... 12 4.0 μg / L; the optimal culture conditions were: culture temperature 27.5℃, 4500 lux, light time 14 h, dark time 10 h.

[0054] Example 2: Using the optimal culture medium and conditions predicted in Example 1, a 250 mL shake flask fermentation of Euglena spp. was conducted, and the biomass and β-1,3-glucan yield were measured. 1. Test reagents and biological materials The test reagents are shown in Table 2.

[0055] Euglena slendera HP6 (CGMCC NO.41188, CN 118240664 A) was a mutagenized algal strain obtained in our laboratory. Sterile single-clone algal strains were obtained by dilution plate method with 1.5% agar added to solid plates and then transferred to liquid medium. The strains were then cultured in a light incubator (1000 lux) for 12 h / 12 ​​h (light / dark) at 27.5℃.

[0056] 2. Test Methods 2.1 Sterilization of culture medium Based on their chemical properties and thermal stability, the culture medium components were divided into three categories for treatment: Basic nutrient solution: sodium acetate trihydrate 2.5 g / L, sodium nitrate 1.5 g / L, ammonium chloride 1.0 g / L, yeast extract 1.2 g / L, tryptone 1.3 g / L, beef extract 0.5 g / L; Carbon source concentrate: Glucose 12.0 g / L; Heat-sensitive vitamin components: Vitamin B1 2.0 mg / L, Vitamin B6 3.0 mg / L, Vitamin B... 12 4.0 μg / L; Sterilization process flow: Basic nutrient solution for autoclaving: Dissolve the basic nutrient solution in 90% volume deionized water and autoclave at 121 ℃ for 16 min using an autoclave.

[0057] Carbon source concentrate (glucose solution): To avoid significant browning reaction in complex systems, glucose was prepared as a 10-fold concentrate and sterilized separately at 115°C for 30 minutes.

[0058] Physical filtration to sterilize heat-sensitive vitamin components: Prepare a concentrated stock solution of heat-sensitive vitamin components from three vitamins. Filter the solution through an organic microporous membrane with a pore size of 0.22µm in a clean bench. The filtered sterile solution should be stored in a sterile brown test tube away from light to prevent light-induced decomposition.

[0059] 2.2 Seed liquid inoculation The seed culture of the Euglena slenderis mutant strain (HP6) (CGMCC No. 4 1188) was inoculated at a rate of 15% into a 250 mL shake flask containing 150 mL of culture medium and cultured twice daily, once in the morning and once in the afternoon. The culture conditions were: 27.5℃, 4500 lux, 14 h light / 10 h dark.

[0060] 3. Experimental Results Table 3. Changes in HP6 biomass and β-1,3-glucan yield at different fermentation times.

[0061] As shown in Table 3, when the Euglena filamentosa mutant strain (HP6) was cultured in a 250 mL shake flask using the optimal culture medium and culture conditions predicted in Experiment Example 1, the biomass of Euglena filamentosa reached 9.98 g / L and the β-1,3-glucan reached 6.02 g / L after 216 hours of culture.

[0062] Experiment 3: A 200 L fermenter-scale culture experiment of Euglena filamentosa was conducted using the optimal culture medium and culture conditions predicted in Experiment 1. 1. Experimental Methods 1.1 Test reagents and instruments The microbial preservation number of Euglena filamentosa used in this experiment is CGMCC No. 41188.

[0063] The main reagents used in this experiment are the same as those in Table 1 of Experiment Example 1, and the main instruments used are shown in Table 4.

[0064] Table 4. Main Instruments for 200 L Pilot-Scale Scale-Up Experiment

[0065] 2.2 Scale-up Test Operation and Process Control Strategy After cleaning, the 200 L fermenter scale-up test was performed online. The fermenter and its auxiliary piping system were treated with steam-in-place (SIP) sterilization. After sterilization, sterile air was introduced to cool the fermenter. Inoculation was carried out after the internal temperature of the fermenter stabilized to the set culture temperature.

[0066] Scale-up culture employed a mixed nutrient culture mode combining light and dark, with a working volume of 120 L and an inoculum volume fraction of 15% (v / v). The culture medium composition adopted the optimal formulation determined in Experiment 1, namely: glucose 12.0 g / L, sodium acetate trihydrate 2.5 g / L, sodium nitrate 1.5 g / L, ammonium chloride 1.0 g / L, yeast extract 1.2 g / L, tryptone 1.3 g / L, beef extract 0.5 g / L, vitamin B1 2.0 mg / L, vitamin B6 3.0 mg / L, and vitamin B2... 12 4.0 μg / L. The culture temperature was controlled at 27.5 ℃, and constant temperature control was achieved through a cold water circulation system in conjunction with the tank jacket. The initial pH was not adjusted, but its change trend was monitored and recorded online.

[0067] During fermentation, a combination of mechanical agitation and air aeration is used to maintain system mixing and gas mass transfer. Air is supplied by an air compressor, and after oil removal and sterile filtration, it is introduced into the bottom of the fermenter at a controlled aeration rate of 3 m³ / h. DO is controlled automatically through a linkage between agitation and aeration: the agitation speed is dynamically adjusted within the range of 80–200 r / min to maintain DO at no less than 20%, thus preventing growth restriction or insufficient mass transfer due to anoxic conditions.

[0068] To alleviate the potential carbon source limitation during the later stages of culture, glucose was added on days 3, 5, and 7 of culture. The feed solution was 500 g / L glucose solution, and the volume added each time was 3.6 L.

[0069] Under combined light and dark culture conditions, the light source system operated according to a preset light-dark cycle, with a light intensity set at 4500 lux, a light duration of 14 h, and a dark duration of 10 h. During fermentation, parameters such as temperature, dissolved oxygen (DO), pH, and operating status were recorded regularly and archived simultaneously with the sampling and testing data to ensure that the entire scale-up experiment was conducted under stable and controllable conditions, thereby improving the reliability and reproducibility of the data.

[0070] 2.3 Sampling Plan and Data Acquisition Specifications During the scale-up test in the 200 L fermenter, a timed sampling strategy was adopted to dynamically monitor the cultivation process. Sampling was conducted every 24 hours (at fixed times) to track the growth status of *Euglena filamentosa* and changes in β-1,3-glucan accumulation. For cultivation days with scheduled feeding (days 3, 5, and 7), to avoid the impact of instantaneous dilution caused by feeding on the comparability of results, sampling was uniformly arranged before feeding, and the feeding time and volume were marked in the records.

[0071] Sampling was conducted through the aseptic sampling port of the fermenter. The sampling procedure was as follows: Before sampling, the outer surface and interface of the sampling valve were disinfected with 75% ethanol. Then, the steam sterilization branch of the sampling port was opened, and the sampling valve, sampling channel, and outlet were sterilized online using high-temperature steam provided by the steam generator. After opening the sampling valve, the initial sample liquid was discarded (used to flush the sampling channel and reduce the influence of local residual liquid and disinfection condensate on representativeness). Then, the sample was collected under aseptic conditions. Three parallel samples were collected at each sampling time point, 50 mL each, for a total of 150 mL, for subsequent testing, and the average value was taken as the measurement result for that time point. The parallel sampling setting was used to reduce random errors in the sampling and testing process and improve the repeatability and statistical reliability of the scaled-up experimental data. After sampling, the sampling valve was immediately closed, and high-temperature steam was introduced again to flush the sampling port and sampling channel to remove residual sample liquid and maintain the aseptic state of the sampling port, reducing the risk of contamination in subsequent sampling processes.

[0072] Samples are immediately processed after sampling: samples for biomass determination are centrifuged and washed within a short time. Samples for β-1,3-glucan determination are processed according to established pretreatment steps. A unified data recording table is established throughout the scale-up experiment, including sampling time points, raw online parameters, offline detection results, and calculated data. All data are summarized and organized according to specifications for subsequent statistical analysis and comparison with model prediction results.

[0073] 2.4 Determination of scale-up test indicators and evaluation of results To verify the stability and reproducibility of the optimal process, three batches of repeated fermentation were conducted under the same experimental conditions. The methods for determining the biomass and β-1,3-glucan content of the scaled-up test samples were consistent with the laboratory methods described in Example 1 to ensure the comparability of data from different scales of experiments.

[0074] All data processing and statistical analysis were completed in Python 3.13.5, primarily using libraries such as pandas, numpy, scikit-learn, xgboost, and shahap. This ensured consistency and traceability in the calculation process. Through this standardized data collection and analysis workflow, effective integration of small-scale and scale-up experimental results was achieved.

[0075] 2. Experimental Results The optimal culture formula and key environmental parameters obtained from the laboratory comprehensive optimization in Experiment 1 were applied to a 200 L fermenter for scale-up verification. During the scale-up culture, the biomass of Euglena filamentosa and the yield of β-1,3-glucan both showed an overall trend of first increasing and then decreasing, as shown in Table 5.

[0076] Table 5. Dynamic changes in biomass and β-1,3-glucan yield during the 200L fermentation scale-up experiment.

[0077] Note: β-1,3-glucan yield g / L = biomass × polysaccharide content.

[0078] As can be seen from the experimental results in Table 5, the biomass and β-1,3-glucan yield of the scaled-up fermentation in the 200 L fermenter showed an overall upward trend during the 24–264 h period. After 264 h, both indicators showed a decline to varying degrees. Under the current operating conditions, it is not conducive to maintaining the target product by continuing to extend the cultivation time. Therefore, around 264 h can be considered a relatively optimal harvest window under pilot-scale conditions. According to the experimental results in Table 5, at 264 h of fermentation, the biomass reached 17.32 ± 1.35 g / L and the β-1,3-glucan yield reached 12.82 ± 1.40 g / L. The biomass and β-1,3-glucan yield at 264 h and the time around (240 h and 288 h) were significantly better than those of the inventor's patent (invention title: High Polysaccharide Euglena filamentosa strain and its breeding method and application, CN 118240664 A) at 264 h and the time around (240 h and 288 h).

[0079] Overall, the optimal culture medium and process parameters obtained in the laboratory stage of Experiment 1 showed good scale-up adaptability in the 200L fermenter and formed a clear high-yield time range, providing a basis for subsequent process effect evaluation.

[0080] Experiment 4: Cultivation of Euglena scabra under different culture media and conditions, and detection of biomass and β-1,3-glucan production. 1. Experimental Methods Thirty-nine different culture media were prepared according to the components and dosages described in Table 4; the preparation method of the culture media was the same as in Experiment 2; the seed liquid of Euglena scabra mutant strain (HP6) (CGMCC No. 4 1188) was inoculated into 250 mL shake flasks containing 150 mL of culture medium according to the inoculation amount described in Table 4, and cultured by shaking the flasks once in the morning and once in the afternoon. The culture conditions (including culture temperature, light intensity and light duration) of each experimental group are shown in Table 4.

[0081] The biomass and β-1,3-glucan content in the fermentation product were measured every 24 hours, using the same method as in Experiment 1.

[0082] 2. Experimental Results The fermentation peak was reached at 216 hours. Table 6 shows the results of the determination of biomass and β-1,3-glucan content in the fermentation products at 216 hours of fermentation.

[0083] Table 6. Results of determination of biomass and β-1,3-glucan content in fermentation products under different culture media and conditions.

[0084] Comparative Test Example 1 1. Experimental Methods Fermentation strain: Euglena slenderis mutant (HP6) (CGMCC No. 4 1188); The fermentation medium in a 250mL Erlenmeyer flask consists of: glucose 15 g / L, sodium acetate trihydrate 1 g / L, beef extract 1 g / L, yeast extract 2 g / L, tryptone 2 g / L, sodium nitrate 1.5 g / L, ammonium chloride 1.0 g / L, vitamin B1 2.0 mg / L, vitamin B6 3.0 mg / L, and vitamin B2. 12 4.0 μg / L; Fermentation culture conditions: culture temperature was 27.5℃, light intensity was 1000 lux, light exposure was 12 h / darkness was 12 h.

[0085] The biomass and β-1,3-glucan content of the fermentation products were measured at 24, 192, 216, and 240 hours of fermentation, respectively, using the same methods as in Example 1. 2. Experimental Results

[0086] Table 7. Results of biomass and β-1,3-glucan content determination at different fermentation times. The measurement results are shown in Table 7.

Claims

1. A culture medium and cultivation method for increasing the biomass and β-1,3-glucan yield of Euglena filamentosa, characterized in that, The culture medium consists of: glucose 12.0 g / L, sodium acetate trihydrate 2.5 g / L, sodium nitrate 1.5 g / L, ammonium chloride 1.0 g / L, yeast extract 1.2 g / L, tryptone 1.3 g / L, beef extract 0.5 g / L, vitamin B1 2.0 mg / L, vitamin B6 3.0 mg / L, and vitamin B2. 12 4.0 μg / L.

2. The culture medium and culture method according to claim 1, characterized in that, The cultivation conditions are as follows: the cultivation temperature is 25-29℃; preferably, the cultivation temperature is 27.5℃.

3. The culture medium and culture method according to claim 1, characterized in that, The cultivation conditions are as follows: light intensity of 4000-5000 lux.

4. The culture medium and culture method according to claim 3, characterized in that, The cultivation conditions were as follows: light intensity of 4500 lux.

5. The culture medium and culture method according to claim 1, characterized in that, The culture conditions are: 12-16 hours of light exposure / 8-12 hours of darkness.

6. The culture medium and culture method according to claim 5, characterized in that, The culture conditions are: 14 h light exposure / 10 h dark exposure.

7. The culture medium and culture method according to claim 1, characterized in that, The microbial preservation number of the aforementioned Euglena filamentosa is: CGMCC NO.41188.

8. A fermentation method for scaled-up fermentation in a fermenter to increase the biomass and β-1,3-glucan yield of Euglena filamentosa, characterized in that, The culture medium consists of: glucose 12.0 g / L, sodium acetate trihydrate 2.5 g / L, sodium nitrate 1.5 g / L, ammonium chloride 1.0 g / L, yeast extract 1.2 g / L, tryptone 1.3 g / L, beef extract 0.5 g / L, vitamin B1 2.0 mg / L, vitamin B6 3.0 mg / L, and vitamin B2. 12 4.0 μg / L; culture temperature 27.5℃, light intensity 4500 lux, light duration 14 h / dark duration 10 h; the microbial preservation number of the described Euglena slenderis is: CGMCC NO.41188.

9. The cultivation method according to claim 8, characterized in that, When scaling up fermentation culture in a fermenter, the initial inoculum of seed liquid is 15%, the working volume of the fermenter is 120L, and biomass is accumulated by light culture for the first three days before polysaccharide enrichment is carried out. The aeration rate is controlled at 3 m³ / h, and the stirring speed is dynamically adjusted within the range of 80–200 r / min to maintain DO not less than 20%.

10. The cultivation method according to claim 8, characterized in that, Glucose was added on days 3, 5, and 7 of fermentation. The feed solution was 500 g / L glucose solution, and the volume added each time was 3.6 L.

Images