A fermentation process for Pichia pastoris and its application

Abstract

A Pichia pastoris fermentation process and its application belong to the field of microbial fermentation technology. The technical problem to be solved is to provide an optimized Pichia pastoris fermentation process to address the shortcomings of existing technologies, such as complex purification of soluble MT, low yield, and high price. The key points of the technical solution are a Pichia pastoris fermentation process, including the following steps: (1) transferring Pichia pastoris carrying the MT gene to YPD plate medium for culture and activation to obtain the cultured strain; (2) inoculating the cultured strain obtained in step (1) into BMGY medium with an initial pH of 7-9 for shaking culture, and obtaining cell cells after centrifugation; (3) waiting for the OD of the cell cells obtained in step (2) 600 At 6-10, the medium was transferred to BMMY medium and induced metal was added to induce MT expression; (4) The process parameters were further optimized using a BP neural network model.

Description

Technical Field

[0001] This invention relates to the field of microbial fermentation technology, specifically to a Pichia pastoris fermentation process and its application. Background Technology

[0002] Metallothioneins (MT), chemically known as metalsulfhistidine trimethylammonium sulfate, are a class of cysteine-rich metal-binding proteins with unique biological properties, widely found in organisms. In 1957, Margoshes and Vallee first isolated a Cd(II)-binding protein from horse kidney while studying the biological functions of metals. Because this protein contains different metal ions and is rich in sulfur, Kagi et al. named it holoprotein metallothionein. Metallothioneins have multiple biological functions, such as heavy metal detoxification, free radical scavenging, anti-radiation, participation in trace element metabolism, anti-inflammatory, anti-cancer, skin care, and treatment of Alzheimer's and Parkinson's diseases.

[0003] With the rapid development of biotechnology and the pharmaceutical industry, the demand for metabolites (MT) is increasing year by year. As a highly inducible endogenous protein, MT can be induced to synthesize by various heavy metal ions, cytokines, hormones, organic chemical drugs, and stress stimuli. However, soluble MT produced using animals (rabbit liver, pig liver) and obtained through heavy metal induction has drawbacks such as specific raw material requirements, complex production processes, extremely low yields, and complicated purification steps. Therefore, there is currently no effective method for large-scale production of MT, resulting in its high price and limiting its application and research. The source of MT raw materials has become one of the bottlenecks restricting the development of this field.

[0004] Microbial gene recombination and fermentation technology is one of the important ways to obtain large quantities of high-purity marine-derived metallothionein (MT). In particular, probiotic expression systems, as the simplest and most reliable exogenous gene expression systems, have been favored by scholars and the market. Currently, there are a few literature reports on methods for producing MT using protein engineering technology and using probiotics as the preferred host cells for industrial protein production via recombinant technology. For example, Lü Pin et al. constructed a recombinant expression vector for metallothionein using green fluorescent protein (GFP) as a reporter gene, and obtained a Pichia pastoris strain expressing MT; Liao Anqi successfully constructed a Pichia pastoris surface-displaying strain by screening different fusion methods and metallothionein from different sources; Xu Liming reported the correct secretory expression of the human metallothionein gene in Pichia pastoris; Peng Delian et al. achieved the fusion expression of human metallothionein in lactic acid bacteria; in the early stages of this experiment, we also constructed recombinant MT Bacillus subtilis (a prokaryotic probiotic) and optimized the induced fermentation conditions to achieve extracellular secretion of MT, but it is far from meeting the requirements for large-scale industrial production.

[0005] Artificial Neural Networks (ANNs) are a class of artificial intelligence technologies that simulate the neural network structure and information processing mechanisms of the human brain, exhibiting unique advantages in the field of complex system modeling and prediction. Error Back Propagation Neural Networks (BPNNs), as one of the most widely used core models within ANNs, possess outstanding ability to fit complex nonlinear mapping relationships and have no strict limitations on the number of input and output endpoints. They have been widely applied in key aspects of fermentation industry processes, such as process optimization and performance prediction. In related research applications, Yang et al. combined artificial neural networks with genetic algorithms to construct a composite optimization model, which systematically regulated the composition of the culture medium for the two-step fermentation of vitamin C, effectively improving the fermentation yield of the mixed bacterial culture system; Ding et al. used this model to reliably predict the acid production of pear vinegar fermentation, verifying its effectiveness in predicting the content of fermentation products; Wei Quanzeng et al. used a BPNN model combined with a genetic algorithm to optimize the fermentation conditions for Streptomyces glucose isomerase production, with a model prediction error of only 1.76%, and the fermentation broth improved by 41.1% compared with the initial fermentation conditions after optimization; Ma Fengying et al. constructed a soft measurement model based on BPNN to achieve high-precision real-time monitoring of ethanol concentration in the fermentation system, solving the lag problem of traditional detection methods; Wang Qiang et al. used an optimization strategy of BPNN coupled with genetic algorithms to conduct targeted screening of the lycopene fermentation culture medium formula, significantly improving the fermentation yield of the target product; Tong Bing et al. optimized the component ratio of neomycin fermentation culture medium using a BPNN model, and under optimized conditions, the neomycin shake-flask fermentation potency was improved by 23% compared with that before optimization, providing a technical reference for industrial production.

[0006] To further address the bottleneck issue of MT sourcing for industrial production, this experiment previously extracted the MT gene from marine shellfish and constructed a marine-derived MT-Pichia pastoris expression vector. This vector was successfully transformed into a food-grade Pichia pastoris strain, yielding two recombinant Pichia pastoris strains. Pichia pastoris, as one of the most important eukaryotic expression systems for recombinant proteins in modern molecular biology research, can highly express biologically active secretory functional exogenous target proteins. It is endotoxin-free, safe to use, and can achieve high cell density and protein yield during high-density fermentation. Furthermore, its fermentation process is mature and stable, meeting the requirements for large-scale industrial production. SMD1168 is a Pep4 gene mutation, resulting in the loss of proteolytic enzyme A activity. This characteristic protects the expressed exogenous protein from degradation, thereby increasing the expression level and stability of the target protein, which is beneficial for obtaining more active target proteins.

[0007] Relevant patent documents retrieved:

[0008] Publication number CN118684763A, publication date 2024.09.24, discloses a method for producing medical recombinant collagen via fermentation. The method comprises the following steps: S1, constructing a genetically engineered Pichia pastoris strain as a collagen-producing strain; S2, preparing a culture medium for cultivating the Pichia pastoris strain; S3, preparing a culture medium for amplifying the Pichia pastoris cells; S4, preparing a PBS buffer; S5, inducing the expression of recombinant collagen in the recombinant Pichia pastoris using a non-methanol inducer; S6, performing large-scale induction of recombinant collagen expression in the culture medium; S7, intermittent glycerol feeding during fermentation. This paper uses ammonium formate as a non-methanol inducer to induce the expression of exogenous proteins by the AOX1 promoter of Pichia pastoris, achieving efficient expression of exogenous proteins while avoiding the hazards of industrial use of methanol and the potential for methanol residues in recombinant collagen, which increases the methanol toxicity of downstream products.

[0009] Relevant non-patent literature retrieved: The journal *Chinese Journal of Biotechnology*, with the article titled "Recombinant Expression and Fermentation Optimization of Fibronectin FNIII8-10 in Pichia pastoris," published on December 23, 2025, discloses a strategy that significantly improves the secretory expression efficiency of fibronectin FNIII8-10 through a combination of rational signal peptide design and fermentation process regulation. Multiple signal peptides were screened, and the α-factor signal peptide was found to have the highest expression efficiency. Subsequently, site-directed mutagenesis was performed on its hydrophobic core region to obtain the highly efficient mutant α-V50A, increasing protein expression by 44.8%. Response surface methodology was used to optimize fermentation conditions, finding the optimal conditions to be 117 h of fermentation, pH 6.91, and methanol addition of 1.91%. Sorbitol was used as an auxiliary carbon source to alleviate methanol stress, resulting in a recombinant FNIII8-10 yield of 65.49 mg / L. Finally, high-density cell fermentation achieved a yield of 532.82 mg / L, a 7-fold increase compared to the highest yield achieved in shake flasks.

[0010] The prior art represented by the aforementioned documents has at least the following unresolved technical problems or defects: The fermentation time of Pichia pastoris is long. The relevant evidence is as follows: Publication No. CN118684763A, the technical solution disclosed in this literature is induced and maintained for about 78 hours; the literature title "Recombinant expression and fermentation optimization of fibronectin FNIII8-10 in Pichia pastoris", the fermentation time in this literature is 117 hours, the overall fermentation time is too long. Summary of the Invention

[0011] The purpose of this invention is to provide: A Pichia pastoris fermentation process and related technologies to solve technical problems such as increasing MT expression levels, or a combination thereof.

[0012] Terminology Explanation: Unless otherwise defined, all technical terms used herein have the same meanings as commonly understood by one of ordinary skill in the art to which this subject matter pertains. Unless otherwise stated, all patents, patent inventions, and disclosures cited throughout this document are incorporated herein by reference in their entirety. Where multiple definitions exist for terms herein, the definitions provided in this chapter shall prevail.

[0013] It should be understood that the above brief description and the following detailed description are exemplary and for illustrative purposes only, and do not limit the subject matter of the invention in any way. In this invention, the singular is used in conjunction with the plural unless otherwise specifically stated. It should also be noted that, unless otherwise stated, the use of “or” or “or” means “and / or”. Furthermore, the use of the term “comprising” and other forms such as “including,” “containing,” and “contains” are not limiting.

[0014] Definitions of standard chemical terms can be found in the references "Pichia Protocols (Methods in Molecular Biology)" and "Yeast Protocols Handbook (Invitrogen)".

[0015] Unless otherwise specified, conventional methods within the scope of the art, such as colony PCR identification and bacterial culture OD, shall be used. 600 Value determination, induced metal (Zn²) + / Cu² + / Cd² + Methods such as concentration determination.

[0016] Unless specifically defined herein, the use of all commercially available products herein employs standard techniques. For example, it may be carried out using the manufacturer's instructions for use with the kit, or in accordance with methods known in the art or the description of this invention. The techniques and methods described herein can generally be implemented according to conventional methods well known in the art, based on the descriptions in the various summary and more specific documents cited and discussed in this specification.

[0017] The term "Pichia pastoris" as used in this article refers to *Pichia pastoris*, a type of methanol-nutritional yeast capable of utilizing methanol as its sole carbon and energy source. During its asexual growth phase, it exists in haploid form, but can form diploid form under nutrient-limited conditions. Its methanol metabolism depends on alcohol oxidase (AOX) within peroxidase cells. The *Pichia pastoris* expression system uses the AOX1 gene promoter to support the stable integration of exogenous genes into Mut+ or Muts strains, achieving protein secretion expression through secretory signal sequences.

[0018] The term "OD" used in this article 600"OD" refers to the absorbance value of a sample at a wavelength of 600 nm, a commonly used indicator for rapid quantification of cell density in microbial culture. In Pichia pastoris fermentation, OD... 600 The value is positively correlated with the dry weight of the bacterial cells and can be used to determine the growth stage of the bacterial cells.

[0019] The term "induced metal" used in this article refers to metal ions that can promote increased MT expression, proper folding, and enhanced metal-binding activity in the Pichia pastoris fermentation system of recombinant marine-derived MT (metallothionein), with zinc ions (Zn²⁺) being a core component. + ), copper ions (Cu²⁺) + ), cadmium ions (Cd²) + It is not an inducer of the Pichia pastoris AOX1 promoter, but rather indirectly enhances the yield and activity of the target protein by activating the structural stability, anti-degradation and functional site formation of MT.

[0020] The term "BP neural network model" used in this article refers to the BackPropagation Neural Network, a multi-layer feedforward artificial neural network trained based on the backpropagation algorithm. It achieves feature mapping and output prediction of data through forward propagation, and propagates the error of the output layer backward along the network layers through backpropagation to correct the connection weights and biases of each layer, so as to minimize the error between the model's predicted value and the actual value, and complete the model training.

[0021] The term "YPD plate medium" as used in this article refers to a completely liquid / solid culture medium for yeast culture, which consists of yeast extract, peptone, and glucose as core nutrients. Agar can be added as needed to prepare solid plates, and antibiotics can be added when necessary for screening resistant strains.

[0022] The term "BMGY medium" used in this article refers to a liquid complete culture medium specifically designed for the proliferation stage of Pichia pastoris, used for high-density proliferation of Pichia pastoris cells, laying the cell mass foundation for subsequent methanol-induced expression.

[0023] The term "BMMY medium" as used in this article refers to a complete liquid culture medium specifically designed for the methanol-induced phase of Pichia pastoris. It is a buffered, complex medium containing methanol, used for methanol-induced secretion and expression of target proteins by recombinant Pichia pastoris strains, and is suitable for methanol-induced AOXI promoter-driven protein expression. BMMY medium has a controllable pH, which reduces protease activity. The yeast extract and peptone added to the medium improve the stability of secreted proteins, prevent or reduce their degradation, and simultaneously promote better yeast growth and the accumulation of large amounts of expressed products.

[0024] The term "MT" used in this article refers to metallothionein, a class of low molecular weight proteins rich in cysteine ​​with strong metal-binding capacity, possessing biological functions such as heavy metal detoxification, free radical scavenging, and anti-oxidative stress.

[0025] The term "YNB" used in this article refers to: Amino Acid-Free Yeast Nitrogen Source, a yeast-specific synthetic culture medium component that contains no amino acids, only basic nitrogen sources (ammonium salts), vitamins, trace elements, and minerals. It is primarily used for the precise culture of yeast strains such as Pichia pastoris, providing basic nutritional support for cell growth and exogenous protein expression.

[0026] In a first aspect, the present invention provides: a Pichia pastoris fermentation process, comprising the following steps: (1) Pichia pastoris loaded with the MT gene was cultured and activated to obtain the cultured strain; (2) The cultured strain obtained in step (1) is inoculated into BMGY medium with an initial pH of 7-9 and cultured to obtain bacterial cells; (3) OD of the bacterial cells obtained in step (2) 600 At 6-10 minutes, the cells were transferred to BMMY medium with a methanol content of 5-10 v / v% and induced metals were added to induce MT expression. (4) The process parameters were further optimized using a BP neural network model.

[0027] Preferably, the Pichia pastoris strain loaded with the MT gene is Pichia pastoris SMD1168-SUMO-MT; Preferably, the culturing and activation method in step (1) is: transferring the culture to YPD plate medium for culturing and activation; The YPD plate culture medium includes yeast extract, peptone, glucose, and agar powder; The preferred YPD plate culture medium comprises 10 g / L yeast extract, 20.0 g / L peptone, 20.0 g / L glucose, and 20.0 g / L agar powder. The activation temperature in step (1) is 25-30℃; preferably 30℃. The activation time in step (1) is 12-24 hours; preferably 24 hours. The BMGY culture medium mentioned in step (2) includes: YNB, biotin, potassium phosphate buffer, yeast extract and peptone; The BMGY culture medium in step (2) preferably comprises 13.4 g / L YNB, 0.0004 g / L biotin, potassium phosphate buffer (pH=6), 10 g / L yeast extract, 20 g / L peptone and 1 mg / L glycerol; The BMGY culture medium in step (2) also includes antibiotics; More preferably, the BMGY culture medium further includes Kana antibiotic, Amp antibiotic, and Zec antibiotic; More preferably, the BMGY culture medium further includes 100 μg / mL of Kana antibiotic, Amp antibiotic, and Zec antibiotic; Preferably, the initial pH value in step (2) is 7-8; More preferably, the initial pH value is 8.0; Preferably, the culture method in step (2) is oscillation culture; The temperature for the oscillation culture is 25-30℃; preferably 30℃. The duration of the oscillation culture is 12-24 hours; preferably 20 hours. The rotation speed of the oscillation culture is 180-240 r / min; The preferred rotation speed for the oscillation culture is 200-240 r / min; The rotation speed of the oscillation culture is further preferably 220 r / min; The inoculation amount in step (2) is selected from 1wt% to 8wt%; The preferred inoculation amount in step (2) is 4 wt%; The process after culturing in step (2) also includes a centrifugation step; Preferably, the methanol content in step (3) is 5v / v%-7.5v / v%. More preferably, the methanol content in step (3) is 7.7 v / v% The BMMY culture medium in step (3) contains YNB, biotin, potassium phosphate buffer, yeast extract and peptone; The preferred BMMY culture medium in step (3) is composed of 13.3 g / L YNB, 0.0004 g / L biotin, potassium phosphate buffer (pH=6), 10 g / L yeast extract and 20 g / L peptone; Preferably, the OD in step (3) 600 Selected from 8-10; More preferably, the OD in step (3) 600 It is 8.8; The induction conditions described in step (3) are: stirring speed 180-260 rpm, induction time 6-72 h; The preferred induction conditions in step (3) are: stirring speed 200-260 rpm and induction time 36-72 h; The induction conditions described in step (3) are further preferably: stirring speed 240 rpm, induction time 48 h; The inducing metal includes, but is not limited to, at least one of ZnSO4, ZnCl2, CuCl2, CdCl2, and CuSO4; The inducing metal is preferably at least one of ZnSO4, CuCl2 and CdCl2; The inducing metal is further preferably ZnSO4; The amount of the inducing metal added is selected from 50-700 μmol / L; The preferred amount of the inducing metal added is 50-200 μmol / L; The amount of the inducing metal added is further preferably 105 μmol / L; Preferably, the inducing metal is added in the form of a sterile solution; More preferably, the inducing metal needs to undergo sterilization treatment by filtration through a 0.22μm aqueous membrane; The BP neural network model described in step (4) uses five process factors as input variables: methanol volume fraction, inoculum amount, oscillation speed, induction time, and induction metal addition amount, and the corresponding measured final MT concentration as the output variable. Preferably, the construction process of the BP neural network model in step (4) is as follows: First, the data is randomly divided into a training set and a validation set at a ratio of 4:1 (train / test split=0.8 / 0.2), and the mean and standard deviation of the training set are used to standardize each input variable and the output variable. After obtaining a convergent neural network model with good generalization performance, the particle swarm optimization algorithm is used to perform global optimization on the five process parameters. Based on further solutions to the technical problems of the present invention, or simultaneous solutions to multiple technical problems, the preferred solution in the technical solution provided in the first aspect of the present invention includes: The first preferred solution is that the initial pH value in step (2) is 7-8; preferably 8.0. This technical solution solves the technical problem of "increasing MT expression level" and further solves the technical problem of "further increasing MT expression level".

[0028] The second preferred option: the OD described in step (3) 600 Selected from 8-10; preferably 8.8. This technical solution, based on solving the technical problem of "increasing MT expression level", further solves the technical problem of "further increasing MT expression level".

[0029] Secondly, the present invention provides the application of the Pichia pastoris fermentation process in MT production.

[0030] This includes technical features: applications in MT production.

[0031] Preferably, the application is as follows: by using the protease-deficient Pichia pastoris SMD1168 as the host strain, the SUMO-MT fusion gene is integrated into the host genome, and the efficient production of MT is achieved through a three-stage fermentation process of "seed culture-glycerol fed-through high-density proliferation-methanol-induced expression"; by leveraging the strong inducibility of the Pichia pastoris AOX1 promoter, the anti-degradation effect of the SUMO tag, and the low protease background of the SMD1168 strain, the expression level and activity of MT are significantly improved, making it suitable for large-scale production of MT.

[0032] In this invention, embodiments 1-5 at least support the protection scope of "induced metal" and "BP neural network model".

[0033] The term "inducible metal" is derived from the foregoing explanation and / or the corresponding technical features in Examples 1-5, such as "the induced metal includes, but is not limited to, at least one of ZnSO4, ZnCl2, CuCl2, CdCl2, and CuSO4" and "the amount of the induced metal added is selected from 50-700 μmol / L," and is summarized by the common feature "metal ions that can promote the increase of MT expression, correct folding, and enhanced metal-binding activity in the Pichia pastoris fermentation system." Therefore, those skilled in the art can reasonably infer that "inducible metal," its subordinate concepts, its essentially equivalent technical means, and technical means that can replace it within the scope of conventional technical means and common knowledge based on the existing technical level should all fall within the protection scope of "inducible metal." For example, replacing "inducible metal" with "zinc ion," "copper ion," etc., while keeping other technical features unchanged, still falls within the protection scope of this invention.

[0034] The “BP neural network model” is summarized from the aforementioned explanation and / or the corresponding technical features in Examples 1-5, such as “the BP neural network model uses five process factors as input variables: methanol volume fraction, inoculum amount, oscillation speed, induction time, and induction metal addition amount, and the corresponding measured final MT concentration as the output variable” and “the construction process of the BP neural network model in step (4) is: firstly, the data is randomly divided into training set and validation set at a ratio of 4:1”, and the common feature “a multi-layer feedforward artificial neural network trained based on the error backpropagation algorithm, which realizes the feature mapping and output prediction of the data through forward propagation, and transmits the error of the output layer in reverse along the network hierarchy through backpropagation and corrects the connection weights and biases of each layer to minimize the error between the model prediction value and the actual value, and completes the model training”. Therefore, those skilled in the art can reasonably presume that the "BP neural network model", its subordinate concepts, its basically equivalent technical means, and technical means that can replace it within the scope of conventional technical means and common knowledge based on the existing technical level should all fall within the protection scope of the "BP neural network model". For example, if the "BP neural network model" is replaced with "error backpropagation neural network" or "error backpropagation type artificial neural network" while other technical features remain unchanged, it still falls within the protection scope of this invention.

[0035] Examples 1-5 of this invention at least support the scope of protection for "applications in MT production".

[0036] The term "application in MT production" is summarized from the foregoing explanation and / or the corresponding technical features in Examples 1-5: "By using the protease-deficient Pichia pastoris SMD1168 as the host strain, the SUMO-MT fusion gene is integrated into the host genome, and a three-stage fermentation process of 'seed culture-glycerol fed-through high-density proliferation-methanol-induced expression' is achieved to achieve efficient MT production; leveraging the strong inducibility of the Pichia pastoris AOX1 promoter, the anti-degradation effect of the SUMO tag, and the low protease background of the SMD1168 strain, the expression level and activity of MT are significantly improved, making it suitable for large-scale MT production." Therefore, those skilled in the art can reasonably presume that "application in MT production," its subordinate concepts, its essentially equivalent technical means, and technical means that can replace it within the scope of conventional and common knowledge based on the existing level of technology should all fall within the protection scope of this invention.

[0037] The beneficial effects of this invention are as follows: The present invention has at least the following beneficial effects: 1. Compared with the prior art, the present invention has better technical effects in terms of MT expression level, etc.

[0038] According to experimental tests, this invention increases the expression level of MT to over 0.2141 mg / mL.

[0039] 2. Compared with existing technologies, this invention provides a technical solution with a different technical concept, and its technical effect is equivalent to or slightly improved compared with existing technologies. The difference between the technical concept of this invention and existing technologies includes, but is not limited to, "in the Pichia pastoris fermentation process, by further defining and preferably specifying a particular pH range, methanol content, and OD..." 600 "Values ​​and other conditions are considered, and a BP neural network model is used for optimization."

[0040] Furthermore, based on the present invention: 1. Based on the comparison of Examples 1-5 and Comparative Example 3, the present invention employs a combination of technical means such as "optimized Pichia pastoris fermentation process" and "BP neural network model," achieving a new technical effect of "further improving the expression level of MT." The combined technical effect is superior to the sum of the effects of each individual technical means.

[0041] 2. Based on the comparison of Examples 1-5 and Comparative Examples 1-2, the present invention addresses the wide range of pH ranges and OD values ​​disclosed in the prior art. 600 In the "values", a narrow range not mentioned in existing technologies was intentionally selected: "pH range of 7-9" and "OD value". 600 The value was 6-10, which achieved unexpected technical results such as "further improving the expression level of MT". Attached Figure Description

[0042] Figure 1 To investigate the effects of initial pH, oscillation speed, and inoculum size on the cultivation of Pichia pastoris strain in step (2) of Example 2; wherein, Figure (a) shows the effects of initial pH, oscillation speed, and inoculum size on the cultivation of Pichia pastoris strain; Figure (b) shows the effects of oscillation speed on the cultivation of Pichia pastoris strain; and Figure (c) shows the effects of inoculum size on the cultivation of Pichia pastoris strain.

[0043] Figure 2 To detect the methanol content and OD in step (3) of Example 2. 600 The effects of oscillation speed and induction time on MT expression levels are shown in Figure (a), where methanol content affects MT expression levels; and OD200 is shown in Figure (b). 600 Effects on MT expression levels; Figure (c) shows the effect of oscillation speed on MT expression levels; Figure (d) shows the effect of oscillation speed on MT expression levels.

[0044] Figure 3 To investigate the effect of the type and amount of induced metal in step (3) of Example 2 on the expression level of MT.

[0045] Figure 4This section analyzes the training and prediction results of the BP neural network model in Example 1. In Figure (a), the training and validation losses are shown; in Figure (b), the learning rate scheduling is shown; and in Figure (c), the training R... 2 (0.8430) and verification R 2 (0.8337), (d) in the figure represents the relative importance of process parameters, and (e) in the figure represents the PSO optimization curve. Detailed Implementation

[0046] The following non-limiting embodiments are intended to enable those skilled in the art to gain a more comprehensive understanding of the present invention, but do not limit the invention in any way. The following content is merely an exemplary description of the scope of protection claimed by the present invention, and those skilled in the art can make various changes and modifications to the present invention based on the disclosed content, and such changes should also fall within the scope of protection claimed by the present invention.

[0047] The present invention will be further described below by way of specific embodiments. Unless otherwise specified, all instruments, devices, equipment, reagents, products, etc., used in the embodiments of the present invention are obtained through conventional commercial means.

[0048] Specific raw material information is shown in Table 1: Table 1. Raw Material Information

[0049] Construction process of Pichia pastoris SMD1168-SUMO-MT in basic embodiment For the specific construction process of Pichia pastoris SMD1168-SUMO-MT, please refer to the literature "Construction and expression activity study of recombinant strain of marine metallothionein Pichia pastoris" (Yu Lei, Yan Guangyu, Sun Jipeng, et al. Construction and expression activity study of recombinant strain of marine metallothionein Pichia pastoris [J]. Journal of Jilin Agricultural Science and Technology College, 2018, 27(3):10-14,30.).

[0050] Example 1: A Pichia pastoris fermentation process, comprising the following steps: (1) Pichia pastoris SMD1168-SUMO-MT stored at -80℃ was transferred to YPD plate medium and activated at 30℃ for 24 h to obtain the cultured strain; (2) The cultured strain obtained in step (1) was inoculated into BMGY medium with an initial pH of 8 at a 4% inoculation rate (containing 100 μg / mL of Kana antibiotic, Amp antibiotic and Zec antibiotic) and cultured with shaking. The culture was carried out at 30℃ and 220 r / min for 20 h. After centrifugation, the bacterial cells were obtained. (3) OD of the bacterial cells obtained in step (2) 600At 8.8, when the bacteria reached the stable growth phase and had almost completely consumed the glycerol, they were transferred to BMMY medium with a methanol content of 7.5 v / v%, and then inducing metal ZnSO4 (sterilized by 0.22 μm aqueous membrane filtration) was added, wherein Zn²⁺ + A concentration of 105 μmol / L was used to induce Zn-MT expression. The induction temperature was 30℃, the stirring speed was 240 rpm, and the induction time was 48 h. (4) Further optimization of process parameters using a BP neural network model: Five process factors—methanol content, inoculum size, oscillation speed, induction time, and induction metal addition—were used as input variables, with the corresponding measured final MT concentration as the output variable. First, the data was randomly divided into a training set and a validation set at a ratio of 4:1 (train / test split = 0.8 / 0.2). The mean and standard deviation of the training set were used to standardize each input and output variable to reduce the impact of dimensional differences on model training. After obtaining a convergent neural network model with good generalization performance, a particle swarm optimization algorithm was used to globally optimize the five process parameters.

[0051] The specific performance parameters of the BP artificial neural network model in Example 1 are shown in the table below. The analysis of the BP neural network model training and prediction results is as follows: Figure 4 : Table 2. Performance parameters of the BP artificial neural network model

[0052] The results show that the established BP neural network model has small error and excellent fitting and prediction accuracy, and global optimization of fermentation parameters can be achieved based on this model. The R² values ​​for the training and validation sets reached 0.8430 and 0.8337, respectively, indicating that the model has good generalization ability and no obvious overfitting phenomenon. The optimal fermentation conditions are: methanol addition 7.7%, OD600 value 8.2, stirring speed 234 rpm, induction time 50 h, Zn² + The concentration was 125 μmol / L. Under these conditions, the MT expression level reached 0.2141 mg / mL, which was 2.93 times higher than the initial fermentation yield (the initial MT expression level was 0.073 mg / mL), further greatly optimizing the process and increasing the MT yield.

[0053] Example 2 The difference from Example 1 is that the initial pH value in step (2) is changed to "the initial pH value is 7", and the rest are the same.

[0054] Example 3 The difference from Example 1 is that the initial pH value in step (2) is changed to "the initial pH value is 9", and the rest are the same.

[0055] Example 4 Unlike Example 1, the OD of the bacterial cells obtained in step (2) was used... 600 "to 8.8" should be changed to "OD of bacterial cells obtained in step (2)". 600 "Up to 6", the rest are the same.

[0056] Example 5 Unlike Example 1, the OD of the bacterial cells obtained in step (2) was used... 600 "to 8.8" should be changed to "OD of bacterial cells obtained in step (2)". 600 "Up to 10", the rest are the same.

[0057] Comparative Example 1 The difference from Example 1 is that the initial pH value in step (2) is changed to "the initial pH value is 5.5", and the rest are the same.

[0058] Comparative Example 2 Unlike Example 1, the OD of the bacterial cells obtained in step (2) was used... 600 "to 6-10" should be changed to "OD of bacterial cells obtained in step (2)". 600 "Up to 4", the rest are the same.

[0059] Comparative Example 3 Unlike Example 1, the "BP neural network model" in step (4) is replaced with the "response surface methodology". The specific steps of the response surface methodology are as follows: Based on the single-factor experiments, response surface experimental factors and levels were set according to Table 3. A mathematical model (a second-order response surface design without center point replication) was established using Box-Behnken Design, with methanol addition (A) and OD levels as the main parameters. 600 (B) Stirring speed (C) Fermentation time (D) and Zn² + The concentration (E) was used as the independent variable, with +1, 0, and -1 representing high, medium, and low levels, respectively. The MT concentration was used as the response value, and the experiment was repeated 5 times. The optimal fermentation conditions for MT expression by the recombinant Pichia pastoris strain were determined. The experimental results are shown in Table 4.

[0060] Regression analysis was performed on the experimental results using Design-Expert 11, yielding the regression equation: R1 = 0.1972 + 0.0005A + 0.0095B + 0.0024C + 0.0007D + 0.0178E + 0.0032AB + 0.0031AC + 0.0102AD + 0.0001AE + 0.0081BC + 0.0016BD - 0.0028BE - 0.0013CD + 0.0092CE - 0.0009DE - 0.04171A 2 -0.0106B2 -0.0140C 2 -0.0129D 2 -0.0243E 2 .

[0061] Table 3. Response surface methodology factors and levels for MT expression in recombinant Pichia pastoris strains

[0062] Table 4. Response surface methodology analysis of fermentation conditions and results of recombinant Pichia pastoris MT.

[0063] Analysis of variance and significance analysis showed that the model P < 0.05, indicating that the regression equation model is significant; the lack-of-fit term P = 0.0559 > 0.05, indicating no significant difference, suggesting that the model fits the experiment well and can predict the experimental results; the correlation coefficient R... 2 =0.8375, indicating that variables less than 2% cannot be explained by the model, and the regression equation can explain the true relationship between each factor and the response value well. The signal-to-noise ratio (SNR) of 8.51 > 4 indicates good model fit and reliability, and this model can be used to predict and optimize experimental results. Combining response surface analysis and optimization, the optimal values ​​were determined to be 7.5% methanol addition and OD... 600 =8. Stirring speed 240 rpm / min, fermentation time 50 hours and Zn² + The optimal fermentation conditions for this Pichia pastoris were 100 μmol / L.

[0064] Everything else is the same as in Example 1.

[0065] Example 1: MT production The MT production results of Examples 1-5 and Comparative Examples 1-3 are shown in the table below: Table 5. MT Production Results

[0066] The data above shows that Comparative Example 3 has a higher MT yield than Examples 2-5. However, the response surface methodology used in Comparative Example 3 has significant shortcomings compared to the BP neural network model used in the embodiments of this application. Specifically, while optimizing fermentation conditions through response surface design can select a better parameter combination within a certain range, it still has two limitations: First, it requires pre-assuming the form of the mathematical model, making it difficult to fully characterize the strong nonlinear relationships in the actual fermentation process; second, when multiple factors act simultaneously and there are interactive effects, the model dimension and experimental quantity increase rapidly, limiting the optimization efficiency.

[0067] Based on this, to further overcome these limitations, we constructed a fully connected BPNN neural network regression model using a large amount of experimental data from previous studies. This model is used to simultaneously analyze the combined effects of multiple variables on fermentation. Compared with traditional response surface methodology, BPNN has three advantages: ① stronger nonlinear fitting ability, automatically capturing complex effects such as saturation, inhibition, and thresholding; ② the ability to handle more factors and their interactions simultaneously, without being limited by the form and order of the equations; ③ higher prediction accuracy and generalization ability, maintaining lower errors on independent validation data.

[0068] Example 2: Single-factor test 1. Effects of initial pH, shaking speed and inoculum size in step (2) on the cultivation of Pichia pastoris strain For detailed test results, please see [link / details]. Figure 1 .

[0069] 2. Methanol content and OD in step (3) 600 Effects of oscillation speed and induction time on MT expression levels For detailed test results, please see [link / details]. Figure 2 .

[0070] 3. Effects of the type and amount of induced metal in step (3) on MT expression levels For detailed test results, please see [link / details]. Figure 3 .

[0071] Verification of technical effectiveness and / or analysis of technical problem solving Verification has shown that the BP neural network model is a method that can be used to optimize fermentation conditions and improve the yield of marine-derived MT fermentation by recombinant Pichia pastoris. This study provides theoretical support and technical reference for the precise scale-up and industrial application of recombinant MT fermentation technology by constructing a technical system of "process optimization-model prediction-global optimization".

[0072] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, and is not intended to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions made by those skilled in the art to the technical solution of the present invention do not depart from the essence and scope of the technical solution of the present invention.

Claims

1. A Pichia pastoris fermentation process, characterized by, The method comprises the steps of: (1) culturing and activating the Pichia pastoris carrying the MT gene to obtain the cultivated strain; (2) inoculating the cultivated strain obtained in step (1) into a BMGY medium with an initial pH value of 7-9 to obtain the bacterial cells; (3) The OD of the bacterial cell obtained in step (2) is measured. 600 When the OD reaches 6-10, the BMMY medium with a methanol content of 5v / v%-10v / v% is used, and the expression of MT is induced by adding an inducing metal. (4) further optimizing the process parameters by using a BP neural network model.

2. The Pichia fermentation process according to claim 1, characterized in that, In step (1), the temperature for activation is 25-30°C, and the time length is 12-24 h.

3. The Pichia fermentation process according to claim 1, characterized in that, In step (2), the BMGY medium further comprises antibiotics.

4. The Pichia fermentation process according to claim 3, characterized in that, The BMGY medium further comprises Kana antibiotics, Amp antibiotics and Zec antibiotics.

5. The Pichia fermentation process according to claim 1, characterized in that, In step (2), the initial pH value is 7-8.

6. The Pichia fermentation process of claim 1, wherein, In step (2), the culture is performed by oscillation, the temperature is 25-30°C, the time length is 12-24 h, and the rotation speed is 180-240 r / min.

7. The Pichia fermentation process of claim 1, wherein, In step (2), the inoculation amount is selected from 1wt%-8wt%.

8. The Pichia fermentation process of claim 1, wherein, In step (3), the methanol content is 5v / v%-7.5v / v%. The OD of step (3) 600 selected from 8-10; In step (3), the induction conditions are as follows: the stirring rotation speed is 180-260 rpm, and the induction time is 6-72 h. In step (3), the induction metal is selected from at least one of ZnSO4, ZnCl2, CuCl2, CdCl2 and CuSO4. In step (3), the addition amount of the induction metal is selected from 50-700 μmol / L.

9. The Pichia fermentation process of claim 1, wherein, In step (4), the BP neural network model takes the methanol volume fraction, the inoculation amount, the oscillation rotation speed, the induction time and the addition amount of the induction metal as the input variables, and takes the measured MT final concentration as the output variable. In step (4), the construction process of the BP neural network model is as follows: first, the data is randomly divided into a training set and a validation set according to a ratio of 4:1, and the mean value and the standard deviation of the training set are used to standardize each input variable and the output variable; after obtaining the neural network model, the particle swarm optimization algorithm is used to perform global optimization on the five process parameters.

10. The application of the Pichia pastoris fermentation process according to any one of claims 1-9 in MT production.

Images