Preparation method of high-purity extracellular vesicles

Abstract

The invention provides a preparation method of high-purity extracellular vesicles, and belongs to the technical field of extracellular vesicles, and the preparation method comprises the following steps: collecting cell culture supernatant generated by culturing cells, and setting a process parameter combination based on an orthogonal test principle; centrifuging the cell culture supernate for the first time; carrying out secondary centrifugation on the first centrifugal supernatant; filtering and purifying the first extracellular vesicles to obtain second extracellular vesicles; purifying the second extracellular vesicle to obtain a third extracellular vesicle; concentrating the third extracellular vesicles to obtain fourth extracellular vesicles; performing characteristic detection on the fourth extracellular vesicles, and recording detection data; establishing a preparation process physical and chemical model; performing parameter fitting on the preparation process physical and chemical model to obtain an optimal process parameter combination; carrying out an extracellular vesicle preparation experiment by adopting the optimal process parameter combination, and correcting a preparation process physical and chemical model to obtain a final process parameter combination; and preparing the extracellular vesicles.

Description

Technical Field

[0001] The present invention belongs to the technical field of extracellular vesicles, and more particularly, relates to a method for preparing high-purity extracellular vesicles. Background Art

[0002] Extracellular vesicles are a type of membrane vesicle particles secreted by cells, with diameters generally ranging from 30 to 500 nanometers. These vesicles carry bioactive molecules such as proteins, nucleic acids, and lipids from the parent cells and play important roles in physiological processes such as intercellular signal transduction, immune regulation, and tissue repair. In recent years, extracellular vesicles have shown broad application prospects in the fields of regenerative medicine, immunotherapy, targeted drug delivery, etc., and thus have attracted extensive attention.

[0003] To convert extracellular vesicles into clinical application products, it is necessary to achieve the preparation of vesicles with high purity and high concentration. Currently, common separation and purification methods include differential centrifugation, density gradient centrifugation, immunoaffinity chromatography, etc. Among them, differential centrifugation and density gradient centrifugation are the two most common methods. Differential centrifugation can perform a preliminary separation based on the differences in vesicle particle size and density, but it is difficult to remove small-scale impurity particles; density gradient centrifugation can further remove impurities, but it is prone to vesicle loss and damage. In addition, the above methods usually can only obtain product concentrations of millions to tens of millions of vesicles per milliliter, which is difficult to meet the requirements of clinical applications.

[0004] In view of the above problems, researchers have explored the application of membrane separation technology in the preparation of extracellular vesicles. Membrane filtration can separate vesicles in a size-exclusion manner and can adjust the separation efficiency through the membrane pore size; ultrafiltration concentration can further increase the concentration of vesicle products. However, the existing membrane separation methods still have the technical problem of being difficult to balance the requirements of high purity and high concentration. Summary of the Invention

[0005] In view of this, the present invention provides a method for preparing high-purity extracellular vesicles, which solves the technical problem that the traditional preparation methods cannot balance the requirements of high purity and high concentration. The present invention is implemented as follows:

[0006] The present invention provides a method for preparing high-purity extracellular vesicles, which includes the following steps:

[0007] Collect the cell culture supernatant produced by cultured cells, and set the process parameter combinations based on the principle of orthogonal experiment; perform the first centrifugation to remove cell debris and obtain the first centrifugation supernatant; perform the second centrifugation on the first centrifugation supernatant to collect extracellular vesicles and obtain the first extracellular vesicles; filter and purify the first extracellular vesicles to obtain the second extracellular vesicles; purify the second extracellular vesicles to obtain the third extracellular vesicles; concentrate the third extracellular vesicles to obtain the fourth extracellular vesicles; perform characteristic detection on the fourth extracellular vesicles, record the detection data and perform standardized processing to obtain a standardized data set, and establish a physicochemical model of the preparation process based on the standardized data set; perform parameter fitting on the physicochemical model of the preparation process to obtain the optimal process parameter combination; use the optimal process parameter combination to conduct an extracellular vesicle preparation experiment, obtain experimental verification data, correct the physicochemical model of the preparation process, and perform error correction to obtain the final process parameter combination for the preparation of extracellular vesicles.

[0008] Among them, the step of collecting the cell culture supernatant produced by cultured cells and setting the process parameter combinations based on the principle of orthogonal experiment specifically includes: filtering the cell culture supernatant through a 200-micron filter screen, a 100-micron filter screen, and a 50-micron filter screen in sequence to obtain a preliminary filtrate; centrifuging the preliminary filtrate at 300 revolutions per minute for 10 minutes and collecting the supernatant; centrifuging the supernatant at 3000 revolutions per minute for 20 minutes and collecting the supernatant; designing an orthogonal experiment plan according to 10 to 20 sample points, and recording the first centrifugation speed, the first centrifugation time, the second centrifugation speed, the second centrifugation time, the pore size of the filter membrane, the concentration of the density gradient medium, and the molecular weight cut-off of the ultrafiltration membrane; obtaining the initial particle size distribution data using a particle size analyzer.

[0009] Among them, the step of performing characteristic detection on the fourth extracellular vesicles and recording the detection data specifically includes: measuring the particle size distribution of extracellular vesicles using dynamic light scattering, controlling the laser power at 5 mW, the sampling time at 180 s, and the sampling interval at 1 s; measuring the concentration of extracellular vesicles using a nanoparticle tracking analyzer, setting the camera gain at 8, and the number of frames collected at 1000 frames; measuring the total protein content in extracellular vesicles using a protein quantification kit; observing the morphological characteristics of extracellular vesicles using a transmission electron microscope, setting the acceleration voltage at 80 kV; analyzing the expression of specific markers of extracellular vesicles using a flow cytometer.

[0010] Among them, the step of performing data cleaning and normalization processing on the detection data to obtain a standardized data set specifically includes: removing outliers in the detection data using the 3-fold standard deviation method; performing standardized processing on the detection data using the maximum-minimum normalization method; reducing the dimension of the detection data using the principal component analysis method; classifying the detection data using the hierarchical clustering method; analyzing the correlation between parameters of the detection data using the Pearson correlation coefficient.

[0011] Among them, the steps of establishing the physicochemical model of the preparation process specifically include: establishing the Stokes sedimentation equation to describe the movement of extracellular vesicles during centrifugation; establishing the Darcy penetration equation to describe the mass transfer of extracellular vesicles during membrane filtration; establishing the diffusion convection equation to describe the migration of extracellular vesicles in the density gradient; establishing the concentration polarization equation to describe the concentration of extracellular vesicles during ultrafiltration; and numerically solving the physicochemical equations using the finite element method.

[0012] Among them, the steps of parameter fitting for the physicochemical model of the preparation process specifically include: fitting the parameters in the physicochemical equations using the least squares method; evaluating the parameter fitting accuracy using the cross-validation method; optimizing the weight coefficients in the physicochemical equations using the gradient descent method; globally optimizing the physicochemical equations using the genetic algorithm; and evaluating the prediction error of the physicochemical equations using the Monte Carlo method.

[0013] Among them, the steps of conducting the extracellular vesicle preparation experiment using the optimal process parameter combination specifically include: inputting the optimal process parameter combination into the physicochemical model of the preparation process to obtain the theoretical prediction value; conducting the first centrifugation using the optimal process parameter combination, controlling the first centrifugation speed and the first centrifugation time, and collecting the first extracellular vesicles; conducting membrane filtration using the optimal process parameter combination, controlling the pore size of the filtration membrane, and collecting the second extracellular vesicles; conducting density gradient centrifugation using the optimal process parameter combination, controlling the concentration of the density gradient medium, and collecting the third extracellular vesicles; and conducting ultrafiltration concentration using the optimal process parameter combination, controlling the molecular weight cut-off of the ultrafiltration membrane, and collecting the fourth extracellular vesicles.

[0014] Among them, the steps of correcting the physicochemical model of the preparation process specifically include: calculating the root mean square error between the theoretical prediction value and the experimental verification data; correcting the parameters in the physicochemical model of the preparation process using the adaptive weight method; iteratively optimizing the physicochemical model of the preparation process using the Bayesian optimization algorithm; determining the key parameters in the physicochemical model of the preparation process using the sensitivity analysis method; and evaluating the robustness of the physicochemical model of the preparation process using the stability analysis method.

[0015] Among them, the steps of establishing the physicochemical model of the preparation process further include: establishing the basic Stokes sedimentation equation, introducing the Brownian motion correction term and the vesicle aggregation correction term; establishing the basic Darcy penetration equation, introducing the electrostatic interaction term, the surface adsorption term and the membrane fouling term; establishing the basic density gradient separation equation, introducing the thermal diffusion term and the convection term; establishing the basic ultrafiltration concentration equation, introducing the concentration polarization term and the membrane fouling kinetics term; and numerically solving the physicochemical model of the preparation process.

[0016] Among them, the step of establishing the physicochemical model of the preparation process further includes a parameter acquisition step: the particle size distribution of extracellular vesicles is measured by dynamic light scattering to obtain particle size parameters; the density distribution of extracellular vesicles is measured by linear sucrose density gradient centrifugation to obtain density parameters; the surface potential of extracellular vesicles is measured by electrophoretic light scattering to obtain surface potential parameters; the membrane permeability coefficient is measured by a standard protein penetration experiment to obtain permeability coefficient parameters; the diffusion coefficient is measured by fluorescence tracer method to obtain diffusion coefficient parameters.

[0017] Among them, the physicochemical model of the preparation process includes:

[0018] The differential centrifugation model is used to describe the separation behavior of extracellular vesicles during the first centrifugation and the second centrifugation. Based on Stokes sedimentation theory and centrifugal force field distribution, the input parameters of the differential centrifugation model include the first centrifugation speed, the first centrifugation time, the second centrifugation speed, the second centrifugation time, the particle size distribution of extracellular vesicles, and the density distribution of extracellular vesicles. The output parameters of the differential centrifugation model include the recovery rate and purity of the first extracellular vesicles;

[0019] The membrane filtration model is used to describe the separation behavior of the first extracellular vesicles during membrane filtration. Based on the sieving mechanism and surface adsorption mechanism, the input parameters of the membrane filtration model include the pore size of the filtration membrane, the transmembrane pressure difference, the membrane surface charge, and the extracellular vesicle surface charge. The output parameters of the membrane filtration model include the retention rate and flux of the second extracellular vesicles;

[0020] The density gradient separation model is used to describe the separation behavior of the second extracellular vesicles in the density gradient. Based on the buoyancy balance principle and diffusion mechanism, the input parameters of the density gradient separation model include the density gradient medium concentration, centrifugal force field distribution, temperature field distribution, and density distribution of extracellular vesicles. The output parameters of the density gradient separation model include the purity and separation efficiency of the third extracellular vesicles;

[0021] The ultrafiltration concentration model is used to describe the concentration behavior of the third extracellular vesicles during ultrafiltration. Based on the membrane mass transfer theory and concentration polarization theory, the input parameters of the ultrafiltration concentration model include the molecular weight cut-off of the ultrafiltration membrane, the transmembrane pressure difference, temperature, and membrane surface shear force. The output parameters of the ultrafiltration concentration model include the concentration and recovery rate of the fourth extracellular vesicles.

[0022] The differential centrifugation model takes into account centrifugal force, Brownian motion, vesicle aggregation, and hydrodynamic effects, and is specifically expressed as follows:

[0023]

[0024] In the formula, vs is the vesicle sedimentation velocity, with the unit of meters per second; d p is the vesicle particle size, with the unit of micrometers; ρ p is the vesicle density, with the unit of kilograms per cubic meter; ρ f is the medium density, with the unit of kilograms per cubic meter; ω is the angular velocity, with the unit of radians per second; r is the rotation radius, with the unit of meters; μ is the dynamic viscosity of the medium, with the unit of Pascal seconds; k B is the Boltzmann constant; T is the temperature, with the unit of Kelvin; k c is the aggregation coefficient; C p is the vesicle concentration; n is the aggregation index; η1 is the error correction term.

[0025] The first extracellular vesicle recovery rate model takes into account the particle size distribution and density distribution, and is specifically expressed as follows:

[0026]

[0027] In the formula, R1 is the recovery rate; d min , d max is the vesicle particle size range; ρ min , ρ max is the vesicle density range; f(d p , ρ p ) is the joint distribution function of vesicle particle size and density; h is the height of the centrifuge tube; t is the centrifugation time.

[0028] The membrane filtration model takes into account membrane pore sieving, electrostatic interaction and surface adsorption, and is specifically expressed as follows:

[0029]

[0030] In the formula, J is the permeation flux; P is the membrane porosity; Δp is the transmembrane pressure difference; R m is the intrinsic resistance of the membrane; R c is the resistance of the protein concentration polarization layer; R p is the resistance of the vesicle blocking layer; β is the blocking coefficient; α is the attenuation coefficient; is the surface potential of the vesicle; is the surface potential of the membrane; κ is the Boltzmann constant; K ad is the adsorption equilibrium constant; η2 is the error correction term.

[0031] The density gradient separation model takes into account buoyancy, diffusion and temperature effects, and is specifically expressed as follows:

[0032]

[0033] In the formula, C is the local concentration; t is the time; D is the diffusion coefficient; x is the position coordinate; v g is the gravitational sedimentation velocity; ST α is the thermal diffusivity; T is the temperature distribution; η3 is the error correction term.

[0034] The ultrafiltration concentration model takes into account membrane mass transfer, osmotic pressure, and concentration polarization, and is specifically expressed as follows:

[0035]

[0036] In the formula, J v is the volume flux; L p is the membrane permeability; σ is the reflection coefficient; Δπ is the osmotic pressure difference; δ is the boundary layer thickness; D is the diffusion coefficient; C w is the membrane wall concentration; C max is the maximum concentration; k f is the fouling coefficient; Q p is the cumulative permeation amount; η4 is the error correction term.

[0037] The derivation process and optimization ideas of each equation will be explained in detail below:

[0038] 1. Derivation process of the differential centrifugation model equation:

[0039] The first step: Establish the basic Stokes equation:

[0040] The second step: Considering the influence of Brownian motion, introduce the Einstein-Stokes diffusion term:

[0041] The third step: Analyze the vesicle aggregation effect and add the aggregation term:

[0042] The fourth step: Add the error correction term η1 to obtain the final equation:

[0043]

[0044] Innovation effect: This equation overcomes the defect of the traditional Stokes equation ignoring microscopic effects and improves the prediction accuracy of the sedimentation behavior of nanoscale vesicles.

[0045] 2. Derivation process of the recovery rate model:

[0046] The first step: Establish the expression of the recovery rate for a single particle size:

[0047] The second step: Considering the influence of particle size distribution, introduce integration:

[0048] The third step: Considering the influence of density distribution, expand it to a double integral:

[0049]

[0050] Innovative effect: Achieved accurate prediction of the separation effect for a non-uniform vesicle population.

[0051] 3. Derivation process of the membrane filtration model:

[0052] First step: Establish the basic equation of Darcy's law:

[0053] Second step: Consider the dynamic process of membrane fouling: (1 - βe -αt );

[0054] Third step: Introduce the electrostatic interaction term:

[0055] Fourth step: Add the surface adsorption term:

[0056] Fifth step: Improve the resistance expression to obtain the final equation:

[0057]

[0058] Innovative effect: Achieved a comprehensive description of multiple action mechanisms in the membrane separation process.

[0059] 4. Derivation process of the density gradient separation model:

[0060] First step: Establish the basic diffusion equation:

[0061] Second step: Introduce the gravitational sedimentation term:

[0062] Third step: Consider the temperature gradient effect:

[0063] Fourth step: Add error correction to obtain the final equation:

[0064]

[0065] Innovative effect: Achieved a coupled description of diffusion, sedimentation, and temperature effects in the density gradient separation process.

[0066] 5. Derivation process of the ultrafiltration concentration model:

[0067] First step: Establish the basic osmotic equation: J v = L p (Δp - σΔπ);

[0068] Second step: Introduce the concentration polarization term:

[0069] Third step: Consider the concentration limit:

[0070] Step 4: Add the pollution dynamic term to obtain the final equation:

[0071]

[0072] Innovation effect: It realizes a comprehensive description of the mass transfer resistance, concentration polarization, and dynamic changes of membrane pollution during the ultrafiltration process.

[0073] The innovation effects of the physical and chemical model of the preparation process include: establishing a complete mathematical description system covering the whole process of extracellular vesicle separation and purification; considering the coupling effect of multiple physical fields to improve the model prediction accuracy; introducing the description of the dynamic process to realize the real-time monitoring of the process; and it can be used for process parameter optimization and process scale-up.

[0074] Parameter acquisition method: The vesicle size distribution is measured by the dynamic light scattering method using a nanoparticle size analyzer; the density distribution is measured by the sucrose density gradient centrifugation method combined with tracer particles; the surface potential is measured by the electrophoretic light scattering method; the membrane parameters are calibrated through standard protein filtration experiments; the diffusion coefficient is measured by the fluorescence tracer method; the coefficients and error terms are obtained by least squares fitting of multiple groups of experimental data. Specifically:

[0075] The particle size parameter d p is obtained by measurement through the dynamic light scattering method using a nanoparticle size analyzer. First, a sample observation chamber is prepared using a plasma-cleaned silanized borosilicate glass slide. The extracellular vesicle sample is diluted with phosphate buffer to a concentration of 1×10 6 per milliliter, and the dynamic light scattering intensity is collected at 25 degrees Celsius for 30 minutes. The particle size distribution is analyzed by the cumulative method, and the average particle size value is obtained by weighted calculation of the signal intensity.

[0076] The density parameter ρ p is obtained by measurement using the linear sucrose density gradient centrifugation method. A sucrose density gradient solution of 0 to 60% is prepared with ultrapure water and added to the centrifuge tube in order of decreasing concentration using a density gradient preparation instrument, controlling the thickness of each layer to be 800 microliters. At the same time, standard silica tracer particles with densities of 1.05, 1.10, 1.15, and 1.20 grams per cubic centimeter are added. The extracellular vesicle sample is slowly added to the top layer, and after centrifugation at 100,000 revolutions per minute for 120 minutes using an ultracentrifuge, it is collected layer by layer at 500-microliter intervals from top to bottom, and the density of each layer is measured by the refractive index method.

[0077] The surface potential parameter is obtained by measurement through the electrophoretic light scattering method. The ionic strength of the solution is adjusted to 0.01 mol / L using phosphate buffer, and the extracellular vesicle sample is diluted to a concentration of 1×10 6Per milliliter, it is filled into a surface-treated U-shaped electrophoresis cell, and the surface potential is measured using a laser Doppler electrophoresis instrument at 25 °C. The electric field strength is set to 15 volts per centimeter, and each sample is measured 10 times and the average value is taken.

[0078] The permeability coefficient parameter L p It is obtained through a standard protein permeability experiment. Bovine serum albumin with a molecular weight of 66,000 daltons is selected as the standard protein, and a protein solution with a concentration of 1 g per liter is prepared. At 25 °C, an ultrafiltration device is used to record the permeation flux within a pressure difference range of 0.5 to 3.0 bar, and the permeability coefficient value is obtained by linear fitting through Darcy's law.

[0079] The diffusion coefficient parameter D is obtained by fluorescence tracer method. Carboxyfluorescein is used to label extracellular vesicles, and fluorescently labeled vesicles with a concentration of 1×10 7 per milliliter are prepared. A confocal microscope is used to construct a diffusion observation system, and fluorescence intensity distribution images are continuously collected for 120 minutes at 25 °C, and the diffusion coefficient is calculated through Fick's diffusion law.

[0080] The aggregation coefficient k c and the aggregation index n are obtained by dynamic light scattering method. Vesicle samples with a concentration of 1×105 to 1×108 per milliliter are prepared, and a nanoparticle size analyzer is used to continuously monitor the particle size change for 120 minutes at 25 °C, and the parameter values are obtained by fitting the vesicle aggregation kinetic equation through the least squares method.

[0081] The reflection coefficient σ is obtained by osmotic pressure method. An osmometer is used to measure the osmotic pressure of vesicle solutions with different concentrations, and it is compared with the theoretical osmotic pressure calculated by the van't Hoff equation, and the reflection coefficient is calculated through the Starling equation.

[0082] The boundary layer thickness δ is obtained by laser scattering imaging technology. Nanofluorescent particles are used as tracers, and during the ultrafiltration process, confocal microscopy is used to collect the concentration distribution images on the membrane surface, and the concentration gradient curve is obtained through image analysis to determine the boundary layer thickness.

[0083] The error correction terms η1, η2, η3, η4 are obtained by fitting a large amount of experimental data. More than 50 groups of experimental data are collected under different process conditions, the deviation between the predicted value of the theoretical model and the experimental value is calculated, and the least squares method is used to optimize and determine the values of the error correction terms, and cross-validation is used to ensure the reliability of the correction terms. Additionally: The error correction terms can also directly use empirical values, and the default empirical values are 0.12 to 0.19.

[0084] In view of the problems existing in the existing extracellular vesicle separation and purification technologies, the present invention proposes a method for preparing high-purity extracellular vesicles. This method makes full use of various separation technologies such as differential centrifugation, membrane filtration, density gradient centrifugation, and ultrafiltration concentration, and establishes a corresponding physical and chemical model to achieve precise control of the entire process of separation and purification. Compared with the existing technologies, the method of the present invention has the following innovative effects:

[0085] 1. A complete physical and chemical model of the extracellular vesicle separation and purification process is established. This model covers the entire process of differential centrifugation, membrane filtration, density gradient centrifugation, and ultrafiltration concentration, realizes the description of the coupling effect of multiple physical fields, and greatly improves the prediction accuracy of the model. At the same time, this model also introduces the description of the dynamic process, can monitor the process in real time, and provides a basis for determining the optimal process parameters;

[0086] 2. The orthogonal experimental design is used to optimize the process parameters, and parameter fitting and verification are carried out in combination with the physical and chemical model. This strategy can not only quickly determine the optimal combination of process parameters, but also ensure a high degree of coincidence between the model prediction results and the experimental data, ensuring the reliability and repeatability of the method;

[0087] 3. The preparation of high-purity and high-concentration extracellular vesicles is realized. After optimizing multiple processes such as differential centrifugation, membrane filtration, density gradient centrifugation, and ultrafiltration concentration, the final extracellular vesicle sample is obtained.

[0088] In summary, the method for preparing extracellular vesicles proposed by the present invention fully combines the advantages of the existing separation and purification technologies, and innovatively establishes a physical and chemical model for the whole process, realizes the precise regulation of process parameters, and solves the technical problem that the traditional preparation method cannot take into account the requirements of high purity and high concentration. Brief Description of the Drawings

[0089] Figure 1 is the flow chart of the present invention;

[0090] Figure 2 is the particle size distribution change diagram of Example 2;

[0091] Figure 3 is the principal component analysis result diagram of Example 2;

[0092] Figure 4 is the optimal process parameter response surface diagram of Example 2;

[0093] Figure 5 is the recovery rate-purity relationship diagram of Example 2. Detailed Description of the Invention

[0094] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0095] As Figure 1 shown, it is a flow chart of the present invention, including the following steps:

[0096] S01. Collect the cell culture supernatant produced by cultured cells, and set a combination of process parameters based on the principle of orthogonal experiment. The combination of process parameters includes the first centrifugation speed, the first centrifugation time, the second centrifugation speed, the second centrifugation time, the pore size of the filter membrane, the density gradient medium concentration, and the molecular weight cut-off of the ultrafiltration membrane;

[0097] S02. Perform the first centrifugation on the cell culture supernatant by differential centrifugation. The first centrifugation uses the first centrifugation speed and the first centrifugation time to remove cell debris, and obtain the first centrifugation supernatant;

[0098] S03. Perform the second centrifugation on the first centrifugation supernatant. The second centrifugation uses the second centrifugation speed and the second centrifugation time to collect extracellular vesicles, and obtain the first extracellular vesicles;

[0099] S04. Filter and purify the first extracellular vesicles by membrane filtration. The membrane filtration method uses the pore size of the filter membrane for filtration and classification, and obtain the second extracellular vesicles;

[0100] S05. Purify the second extracellular vesicles by density gradient centrifugation. The density gradient centrifugation method uses the density gradient medium concentration for separation, and obtain the third extracellular vesicles;

[0101] S06. Concentrate the third extracellular vesicles by ultrafiltration concentration. The ultrafiltration concentration method uses the molecular weight cut-off of the ultrafiltration membrane for concentration, and obtain the fourth extracellular vesicles;

[0102] S07. Perform characteristic detection on the fourth extracellular vesicles, and record the detection data. The detection data includes vesicle size distribution, vesicle concentration, vesicle purity, vesicle protein content, and vesicle morphological characteristics;

[0103] S08. Perform data cleaning and normalization processing on the detection data to obtain a standardized data set, and establish a physicochemical model of the preparation process based on the standardized data set;

[0104] S09. Perform parameter fitting on the physicochemical model of the preparation process by the least squares method to obtain an optimal combination of process parameters;

[0105] S10. Use the optimal process parameter combination to conduct experiments on extracellular vesicle preparation, obtain experimental verification data, correct the physicochemical model of the preparation process, and obtain the final process parameter combination;

[0106] S11. Prepare extracellular vesicles according to the final process parameter combination.

[0107] The following describes the specific implementation manners of the above steps in detail:

[0108] The specific implementation manner of step S01 is to pretreat the cell culture supernatant by differential centrifugation. First, filter the cell culture supernatant through 200-micron, 100-micron, and 50-micron filters in sequence to obtain a preliminary filtrate. The purpose of this step is to remove cell debris and large-sized impurity particles, creating good conditions for the subsequent separation and purification of extracellular vesicles. Next, centrifuge the preliminary filtrate at a speed of 300 revolutions per minute for 10 minutes, and collect the supernatant. The purpose of this step is to remove larger cell organelles and cell debris. Then, centrifuge the supernatant at a speed of 3000 revolutions per minute for 20 minutes, and collect the supernatant. The purpose of this step is to preliminarily enrich extracellular vesicles. After that, design an orthogonal experiment plan according to 10 to 20 sample points, and record process parameters such as the first centrifugation speed, the first centrifugation time, the second centrifugation speed, the second centrifugation time, the pore size of the filter membrane, the concentration of the density gradient medium, and the molecular weight cut-off of the ultrafiltration membrane. The purpose of this step is to determine the process parameter combination according to the orthogonal experiment principle, laying a foundation for subsequent process optimization. Finally, use a particle size analyzer to obtain the initial particle size distribution data. The purpose of this step is to provide necessary input parameters for establishing the physicochemical model of the preparation process.

[0109] The specific implementation manner of step S02 is to conduct a detailed characterization analysis of the separated extracellular vesicles using various characterization means. First, use dynamic light scattering to measure the particle size distribution of extracellular vesicles. Specifically, control the laser power at 5 milliwatts, the sampling time at 180 seconds, and the sampling interval at 1 second. The purpose of this step is to obtain the particle size parameters of extracellular vesicles. Secondly, use a nanoparticle tracking analyzer to measure the concentration of extracellular vesicles. Specifically, set the camera gain at 8 and the number of frames collected at 1000 frames. The purpose of this step is to obtain the concentration parameters of extracellular vesicles. Subsequently, use a protein quantification kit to measure the total protein content in extracellular vesicles. The purpose of this step is to obtain the protein content parameters of extracellular vesicles. Next, use a transmission electron microscope to observe the morphological characteristics of extracellular vesicles. Specifically, set the acceleration voltage at 80 kilovolts. The purpose of this step is to obtain the morphological parameters of extracellular vesicles. Finally, use a flow cytometer to analyze the expression of specific markers of extracellular vesicles. The purpose of this step is to obtain the purity parameters of extracellular vesicles.

[0110] The specific implementation of step S03 is to perform data preprocessing on the collected detection data. First, the 3-sigma method is used to remove outliers from the detection data. The purpose of this step is to clean the original data and improve data quality. Next, the maximum-minimum normalization method is used to standardize the detection data. The purpose of this step is to eliminate the influence of dimension on data analysis and ensure the comparability of each parameter. Then, the principal component analysis method is used to reduce the dimension of the detection data. The purpose of this step is to extract key features, reduce data redundancy, and prepare for subsequent model establishment. After that, the hierarchical clustering method is used to classify the detection data. The purpose of this step is to identify the potential structure of the data and provide a basis for process parameter optimization. Finally, the Pearson correlation coefficient is used to analyze the correlation between the parameters of the detection data. The purpose of this step is to reveal the internal relationship between variables and provide a reference for the establishment of the physical and chemical model of the preparation process.

[0111] The specific implementation of step S04 is to establish a physical and chemical model for the whole process of extracellular vesicle isolation and purification. First, the Stokes sedimentation equation is established to describe the movement of extracellular vesicles during centrifugation. The purpose of this step is to predict the separation behavior of extracellular vesicles during differential centrifugation. Next, the Darcy permeation equation is established to describe the mass transfer of extracellular vesicles during membrane filtration. The purpose of this step is to predict the separation behavior of extracellular vesicles during membrane filtration. Then, the diffusion-convection equation is established to describe the migration of extracellular vesicles in the density gradient. The purpose of this step is to predict the separation behavior of extracellular vesicles during density gradient centrifugation. After that, the concentration polarization equation is established to describe the concentration of extracellular vesicles during ultrafiltration. The purpose of this step is to predict the concentration behavior of extracellular vesicles during ultrafiltration. Finally, the finite element method is used to numerically solve the above physical and chemical equations. The purpose of this step is to achieve the coupled description and simulation of the entire separation and purification process.

[0112] The specific implementation of step S05 is to optimize the parameters of the established physical and chemical model for the preparation process. First, the least squares method is used to fit the parameters in the physical and chemical equations. The purpose of this step is to determine the optimal values of the model parameters. Next, the cross-validation method is used to evaluate the accuracy of the parameter fitting. The purpose of this step is to test the reliability of the model parameters. Then, the gradient descent method is used to optimize the weight coefficients in the physical and chemical equations. The purpose of this step is to improve the prediction accuracy of the model. After that, the genetic algorithm is used to globally optimize the physical and chemical equations. The purpose of this step is to find the optimal process parameters in a wider parameter space. Finally, the Monte Carlo method is used to evaluate the prediction error of the physical and chemical equations. The purpose of this step is to quantify the uncertainty of the model prediction results.

[0113] The specific implementation of step S06 is to conduct an experiment on the preparation of extracellular vesicles using the optimized process parameters. First, the optimal process parameter combination is input into the physical and chemical model of the preparation process to obtain the theoretical prediction value. The purpose of this step is to provide theoretical guidance for the experimental design. Next, the first centrifugation is carried out using the optimal process parameter combination, controlling the first centrifugation speed and the first centrifugation time, and the first extracellular vesicles are collected. The purpose of this step is to initially separate the extracellular vesicles according to the conditions predicted by the theoretical model. Then, membrane filtration is carried out using the optimal process parameter combination, controlling the pore size of the filtration membrane, and the second extracellular vesicles are collected. The purpose of this step is to further purify the extracellular vesicles according to the conditions predicted by the theoretical model. After that, density gradient centrifugation is carried out using the optimal process parameter combination, controlling the concentration of the density gradient medium, and the third extracellular vesicles are collected. The purpose of this step is to remove impurities according to the conditions predicted by the theoretical model and further improve the purity of the extracellular vesicles. Finally, ultrafiltration concentration is carried out using the optimal process parameter combination, controlling the molecular weight cut-off of the ultrafiltration membrane, and the fourth extracellular vesicles are collected. The purpose of this step is to increase the concentration of the extracellular vesicles according to the conditions predicted by the theoretical model to obtain the final product.

[0114] The specific implementation of step S07 is to correct and optimize the physical and chemical model of the preparation process. First, calculate the root mean square error between the theoretical prediction value and the experimental verification data. The purpose of this step is to quantify the difference between the model and the experimental results. Next, the parameters in the physical and chemical model of the preparation process are corrected using the adaptive weight method. The purpose of this step is to reduce the deviation between the model and the experimental results. Then, the Bayesian optimization algorithm is used to iteratively optimize the physical and chemical model of the preparation process. The purpose of this step is to further improve the prediction accuracy of the model while ensuring the reliability of the model. After that, the sensitivity analysis method is used to determine the key parameters in the physical and chemical model of the preparation process. The purpose of this step is to identify the key factors that have the greatest impact on the separation and purification process. Finally, the stability analysis method is used to evaluate the robustness of the physical and chemical model of the preparation process. The purpose of this step is to ensure that the model can still maintain good prediction performance under uncertain conditions.

[0115] In the process of establishing the physicochemical model of the preparation process, first, the Stokes sedimentation basic equation is established to describe the motion law of extracellular vesicles during centrifugation. On this basis, the Brownian motion correction term and the vesicle aggregation correction term are introduced to improve the prediction accuracy of the model for the sedimentation behavior of extracellular vesicles at the nanoscale. Secondly, the Darcy penetration basic equation is established to describe the mass transfer process of extracellular vesicles during membrane filtration. On this basis, the electrostatic interaction term, the surface adsorption term, and the membrane fouling term are introduced to achieve a comprehensive description of the multiple action mechanisms during the membrane separation process. Then, the density gradient separation basic equation is established to describe the migration behavior of extracellular vesicles in the density gradient. On this basis, the thermal diffusion term and the convection term are introduced to achieve a coupled description of diffusion, sedimentation, and temperature effects during the density gradient separation process. Finally, the ultrafiltration concentration basic equation is established to describe the concentration behavior of extracellular vesicles during ultrafiltration. On this basis, the concentration polarization term and the membrane fouling kinetics term are introduced to achieve a comprehensive description of the mass transfer resistance, concentration polarization, and dynamic changes in membrane fouling during the ultrafiltration process. The above process of establishing the physicochemical model of the preparation process fully considers the multiple physical mechanisms in the whole process of extracellular vesicle separation and purification, improving the prediction accuracy and applicability of the model.

[0116] In terms of parameter acquisition, first, the dynamic light scattering method is used to measure the particle size distribution of extracellular vesicles to obtain the particle size parameter. This method obtains the average particle size value by collecting the dynamic light scattering intensity of the vesicle sample under laser irradiation and using signal intensity weighted analysis. Secondly, the linear sucrose density gradient centrifugation method is used to measure the density distribution of extracellular vesicles to obtain the density parameter. This method determines the density value of each layer of vesicles by establishing a linear sucrose density gradient in the centrifuge tube and adding silica gel tracer particles with known density, and using the density calculation formula. Then, the electrophoretic light scattering method is used to measure the surface potential of extracellular vesicles to obtain the surface potential parameter. This method measures the migration speed of vesicles in the liquid under the action of an electric field and calculates the surface potential by combining with the Teeter formula. Next, the standard protein penetration experiment is used to measure the membrane permeability coefficient to obtain the permeability coefficient parameter. This method measures the permeation flux of the standard protein solution under a certain pressure difference condition and calculates the permeability coefficient by combining with Darcy's law. Finally, the fluorescence tracer method is used to measure the diffusion coefficient to obtain the diffusion coefficient parameter. This method calculates the diffusion coefficient by monitoring the diffusion process of the fluorescence tracer in the vesicles in the liquid and according to Fick's diffusion law. The above parameter acquisition methods fully consider the physicochemical properties of extracellular vesicles, providing reliable parameter inputs for the establishment of the physicochemical model of the preparation process.

[0117] In terms of error correction, first, collect experimental data under more than 50 different combinations of process parameters, including indicators such as separation effect. Then, calculate the deviation between the predicted values of the physical and chemical model of the preparation process and the experimental data. Next, use the least squares method to determine the numerical values of the error correction terms in each physical and chemical equation. The purpose of this step is to reduce the difference between the model prediction results and the experimental results. After that, use the cross-validation method to evaluate the reliability of the error correction terms. The purpose of this step is to ensure that the error correction terms have good applicability and generalization ability. Finally, use the Bootstrap method to estimate the confidence interval of the error correction terms. The purpose of this step is to quantify the uncertainty of the model prediction results and provide a reference basis for process optimization. Through the above error correction steps, the accuracy and robustness of the physical and chemical model of the preparation process can be further improved.

[0118] The following provides a specific Embodiment 1 of the present invention. The specific implementation methods of each step in this Embodiment 1 are described in detail as follows:

[0119] The specific implementation method of step S01 is as follows:

[0120] First, pass the cell culture supernatant through a 200-micron filter, a 100-micron filter, and a 50-micron filter in sequence to obtain a preliminary filtrate. The purpose of this step is to remove cell debris and large-sized impurity particles and create good conditions for the subsequent separation and purification of extracellular vesicles.

[0121] Next, centrifuge the preliminary filtrate at a speed of 300 revolutions per minute for 10 minutes and collect the supernatant. The purpose of this step is to remove larger cell organelles and cell debris.

[0122] Then, centrifuge the supernatant at a speed of 3000 revolutions per minute for 20 minutes and collect the supernatant. The purpose of this step is to preliminarily enrich extracellular vesicles.

[0123] After that, design an orthogonal experiment plan according to 10 to 20 sample points, and record the first centrifugation speed ω1, the first centrifugation time t1, the second centrifugation speed ω2, the second centrifugation time t2, the pore size d of the filter membrane pore , the concentration ρ of the density gradient medium gra , the molecular weight cut-off M of the ultrafiltration membrane cut and other process parameters. The purpose of this step is to determine the combination of process parameters according to the orthogonal experiment principle and lay a foundation for subsequent process optimization.

[0124] Finally, use a particle size analyzer to obtain the initial particle size distribution data. The purpose of this step is to provide the necessary input parameters for the establishment of the physical and chemical model of the preparation process.

[0125] The specific implementation method of step S02 is as follows:

[0126] First, the dynamic light scattering method was used to determine the particle size distribution of extracellular vesicles. Specifically, the laser power was controlled to P las =5 mW, sampling time is t smp = 180 seconds, and the sampling interval is Δt = 1 second. The purpose of this step is to obtain the particle size parameter d of the extracellular vesicles p .

[0127] Secondly, the concentration of extracellular vesicles was measured using a nanoparticle tracking analyzer. The camera gain was set to G cam =8, the number of acquisition frames is N frm = 1000 frames. The purpose of this step is to obtain the concentration parameter C of extracellular vesicles p .

[0128] Subsequently, a protein quantification kit was used to determine the total protein content in the extracellular vesicles. The purpose of this step is to obtain the protein content parameter C of the extracellular vesicles. pro .

[0129] Next, the morphological characteristics of extracellular vesicles were observed using transmission electron microscopy. The specific acceleration voltage was set to V acc = 80 kV. The purpose of this step is to obtain the morphological parameters S of extracellular vesicles mor .

[0130] Finally, flow cytometry was used to analyze the expression of specific extracellular vesicle markers. The purpose of this step is to obtain the purity parameter P of extracellular vesicles. pur .

[0131] The specific implementation of step S03 is as follows:

[0132] First, we use the 3 times standard deviation method Eliminate outliers in the test data. The purpose of this step is to clean the original data and improve the data quality.

[0133] Next, the maximum and minimum normalization method is used The test data is standardized. The purpose of this step is to eliminate the impact of dimension on data analysis and ensure that each parameter is comparable.

[0134] Then, using principal component analysis, X = TP T +E reduces the dimension of the detection data. The purpose of this step is to extract key features, reduce data redundancy, and prepare for subsequent model building.

[0135] Then, hierarchical clustering method was used Classify the test data. The purpose of this step is to identify the potential structure of the data and provide a basis for process parameter optimization.

[0136] Finally, the Pearson correlation coefficient is adopted to analyze the correlation between parameters of the detection data. The purpose of this step is to reveal the internal relationship between variables and provide a reference for the establishment of the physical and chemical model of the preparation process.

[0137] The specific implementation of step S04 is as follows:

[0138] First, establish the Stokes sedimentation equation to describe the motion law of extracellular vesicles during centrifugation. The purpose of this step is to predict the separation behavior of extracellular vesicles during differential centrifugation.

[0139] where v s is the sedimentation velocity of the vesicle, d p is the vesicle diameter, ρ p is the vesicle density, ρ f is the medium density, ω is the angular velocity, r is the rotation radius, μ is the dynamic viscosity of the medium, k B is the Boltzmann constant, T is the temperature, k c is the aggregation coefficient, C p is the vesicle concentration, n is the aggregation index, and η1 is the error correction term.

[0140] Next, establish the Darcy penetration equation to describe the mass transfer of extracellular vesicles during membrane filtration.

[0141] where J is the permeation flux, P is the membrane porosity, Δp is the transmembrane pressure difference, R m is the intrinsic resistance of the membrane, R c is the resistance of the protein concentration polarization layer, R p is the resistance of the vesicle plugging layer, β is the blocking coefficient, α is the attenuation coefficient, is the surface potential of the vesicle, is the membrane surface potential, κ is the Boltzmann constant, K ad is the adsorption equilibrium constant, and η2 is the error correction term.

[0142] Then, establish the diffusion convection equation to describe the migration behavior of extracellular vesicles in the density gradient.

[0143] where C is the local concentration, t is the time, D is the diffusion coefficient, x is the position coordinate, v g is the gravitational sedimentation velocity, S T is the thermal diffusion coefficient, T is the temperature distribution, and η3 is the error correction term.

[0144] After that, establish the concentration polarization equation to describe the concentration behavior of extracellular vesicles during ultrafiltration.

[0145] where Jv is the volume flux, L p is the membrane permeability coefficient, σ is the reflection coefficient, Δπ is the osmotic pressure difference, δ is the boundary layer thickness, D is the diffusion coefficient, C w is the membrane wall concentration, C max is the maximum concentration, k f is the fouling coefficient, Q p is the cumulative permeation amount, and η4 is the error correction term.

[0146] Finally, the finite element method is used to numerically solve the above physical and chemical equations. The purpose of this step is to achieve the coupled description and simulation of the entire separation and purification process.

[0147] The specific implementation of step S05 is as follows:

[0148] First, the least squares method min||y - Xβ|| 2 is used to fit the parameters in the physical and chemical equations. The purpose of this step is to determine the optimal values of the model parameters.

[0149] where y is the experimental data vector, X is the independent variable matrix, and β is the parameter vector to be estimated.

[0150] Next, the cross-validation method is used to evaluate the accuracy of the parameter fitting. The purpose of this step is to test the reliability of the model parameters.

[0151] Then, the gradient descent method is used to optimize the weight coefficients in the physical and chemical equations. The purpose of this step is to improve the prediction accuracy of the model.

[0152] where θ is the weight coefficient vector, α is the learning rate, and J is the loss function.

[0153] After that, the genetic algorithm is used to globally optimize the physical and chemical equations. The purpose of this step is to find the optimal process parameters in a wider parameter space.

[0154] where x i+1 is the individual in the next generation, x i,j is the individual in the current generation, and w j is the crossover weight.

[0155] Finally, the Monte Carlo method is used to evaluate the prediction error of the physical and chemical equations. The purpose of this step is to quantify the uncertainty of the model prediction results.

[0156] The specific implementation of step S06 is as follows:

[0157] First, the optimal process parameter combination Input the physical and chemical model of the preparation process to obtain the theoretical prediction value. The purpose of this step is to provide theoretical guidance for experimental design.

[0158] Next, perform the first centrifugation with the optimal process parameter combination, controlling the first centrifugation speed and the first centrifugation time Collect the first extracellular vesicles. The purpose of this step is to preliminarily isolate extracellular vesicles according to the conditions predicted by the theoretical model.

[0159] Then, perform membrane filtration with the optimal process parameter combination, controlling the pore size of the filter membrane Collect the second extracellular vesicles. The purpose of this step is to further purify extracellular vesicles according to the conditions predicted by the theoretical model.

[0160] After that, perform density gradient centrifugation with the optimal process parameter combination, controlling the concentration of the density gradient medium Collect the third extracellular vesicles. The purpose of this step is to remove impurities and further improve the purity of extracellular vesicles according to the conditions predicted by the theoretical model.

[0161] Finally, perform ultrafiltration concentration with the optimal process parameter combination, controlling the molecular weight cut-off of the ultrafiltration membrane Collect the fourth extracellular vesicles. The purpose of this step is to increase the concentration of extracellular vesicles and obtain the final product according to the conditions predicted by the theoretical model.

[0162] The specific implementation of step S07 is as follows:

[0163] First, calculate the root mean square error between the theoretical prediction value and the experimental verification data y. The purpose of this step is to quantify the difference between the model and the experimental results.

[0164] Next, use the adaptive weight method to correct the parameters in the physical and chemical model of the preparation process. The purpose of this step is to reduce the deviation between the model and the experimental results.

[0165] where α (j) = α (0) ·γ j is the adaptive learning rate and γ is the decay coefficient.

[0166] Then, use the Bayesian optimization algorithm to iteratively optimize the physical and chemical model of the preparation process. The purpose of this step is to further improve the prediction accuracy of the model while ensuring the reliability of the model.

[0167] Among them, p(D|θ) is the likelihood function, p(θ) is the prior distribution, and p(D) is the evidence.

[0168] After that, the sensitivity analysis method is used to determine the key parameters in the physicochemical model of the preparation process. The purpose of this step is to identify the key factors that have the greatest impact on the separation and purification process.

[0169] Finally, the stability analysis method is used to evaluate the robustness of the physicochemical model of the preparation process. The purpose of this step is to ensure that the model can still maintain good prediction performance under uncertain conditions.

[0170] In the process of establishing the physicochemical model of the preparation process, first, the Stokes sedimentation equation is established to describe the movement law of extracellular vesicles during centrifugation. This equation takes into account the centrifugal force, Brownian motion, vesicle aggregation, and hydrodynamic effects, improving the prediction accuracy of the sedimentation behavior of vesicles at the nanoscale.

[0171] Next, the Darcy penetration equation is established to describe the mass transfer process of extracellular vesicles during membrane filtration. This equation takes into account multiple action mechanisms such as membrane pore sieving, electrostatic interaction, and surface adsorption, realizing a comprehensive description of the membrane separation process.

[0172] Then, the diffusion-convection equation is established to describe the migration behavior of extracellular vesicles in the density gradient. This equation takes into account the coupling of diffusion, sedimentation, and temperature gradient effects, realizing a comprehensive description of the density gradient separation process.

[0173] After that, the concentration polarization equation is established to describe the concentration behavior of extracellular vesicles during ultrafiltration. This equation takes into account factors such as membrane mass transfer, osmotic pressure, and concentration polarization, realizing a comprehensive description of the ultrafiltration process.

[0174] Finally, the finite element method is used to numerically solve the above physicochemical equations, realizing the coupled description and simulation of the entire separation and purification process. This modeling method fully considers the multiple physical mechanisms in the whole process of extracellular vesicle separation and purification, improving the prediction accuracy and applicability of the model.

[0175] In terms of parameter acquisition, first, the dynamic light scattering method is used to measure the particle size distribution of extracellular vesicles to obtain the particle size parameter d p . This method obtains the average particle size value by collecting the dynamic light scattering intensity of the vesicle sample under laser irradiation and using signal intensity weighted analysis. Secondly, the linear sucrose density gradient centrifugation method is used to measure the density distribution of extracellular vesicles to obtain the density parameter ρ pThis method determines the density values of vesicles in each layer by establishing a linear sucrose density gradient in a centrifuge tube and adding silica tracer particles with known density, and using the density calculation formula.

[0176] Then, the electrophoretic light scattering method is used to measure the surface potential of extracellular vesicles and obtain surface potential parameters. This method measures the migration speed of vesicles in a liquid under the action of an electric field and calculates the surface potential by combining with the Tate formula. Next, the standard protein penetration experiment is used to measure the membrane permeability coefficient and obtain the permeability coefficient parameter L. p This method measures the osmotic flux of a standard protein solution under a certain pressure difference condition and calculates the permeability coefficient by combining with Darcy's law.

[0177] Finally, the fluorescence tracer method is used to measure the diffusion coefficient and obtain the diffusion coefficient parameter D. This method monitors the diffusion process of the fluorescent tracer in the liquid in the vesicles and calculates the diffusion coefficient according to Fick's diffusion law. The above parameter acquisition methods fully consider the physicochemical properties of extracellular vesicles and provide reliable parameter inputs for the establishment of the physicochemical model of the preparation process.

[0178] In terms of error correction, first, more than 50 groups of experimental data under different combinations of process parameters, including indicators such as separation effect, are collected. Then, the predicted values of the physicochemical model of the preparation process are calculated. The deviation from the experimental data y is calculated. Next, the least squares method is used to determine the error correction terms η1, η2, η3, η4 in each physicochemical equation. The purpose of this step is to reduce the difference between the model prediction results and the experimental results. After that, the cross-validation method is used to evaluate the reliability of the error correction terms. The purpose of this step is to ensure that the error correction terms have good applicability and generalization ability. Finally, the Bootstrap method is used to estimate the confidence interval of the error correction terms. The purpose of this step is to quantify the uncertainty of the model prediction results and provide a reference basis for process optimization. Through the above error correction steps, the accuracy and robustness of the physicochemical model of the preparation process can be further improved.

[0179] The extracellular vesicle preparation method of the present invention focuses on establishing a complete physicochemical model to guide the optimization and control of process parameters. The model mainly includes four parts: differential centrifugation model, membrane filtration model, density gradient separation model, and ultrafiltration concentration model.

[0180] First, the differential centrifugation model is used to describe the separation behavior of extracellular vesicles during centrifugation. Based on Stokes sedimentation theory, this model comprehensively considers centrifugal force, Brownian motion, vesicle aggregation, and hydrodynamic effects, and can accurately predict the movement trajectories of vesicles with different particle sizes and densities in the centrifugal force field. Through this model, the optimal rotational speed and time parameters for the first and second centrifugations can be determined to achieve the purpose of preliminary separation and enrichment.

[0181] Secondly, the membrane filtration model is used to describe the mass transfer behavior of extracellular vesicles during membrane separation. This model takes into account multiple mechanisms such as membrane pore sieving, electrostatic interactions, and surface adsorption, and can accurately predict the separation effect under different membrane pore sizes and transmembrane pressure differences. Through this model, the optimal membrane pore size parameter can be determined to further purify extracellular vesicles.

[0182] Then, the density gradient separation model is used to describe the migration behavior of extracellular vesicles during density gradient centrifugation. This model comprehensively considers buoyancy equilibrium, diffusion mechanism, and temperature gradient effect, and can accurately predict the separation process of vesicles in the centrifugal force field under different density gradient conditions. Through this model, the optimal density gradient medium concentration parameter can be determined to further improve the purity of extracellular vesicles.

[0183] Finally, the ultrafiltration concentration model is used to describe the concentration behavior of extracellular vesicles during ultrafiltration. This model takes into account multiple factors such as membrane mass transfer, osmotic pressure, and concentration polarization, and can comprehensively predict the concentration effect under different membrane cut-off molecular weight conditions. Through this model, the optimal ultrafiltration membrane cut-off molecular weight parameter can be determined to increase the concentration of extracellular vesicles.

[0184] By coupling and optimizing the above four physicochemical models, the method of the present invention can precisely control the entire process of extracellular vesicle separation and purification, thereby obtaining products with high purity and high concentration. Specifically, the differential centrifugation model and the membrane filtration model are used for preliminary separation and purification, the density gradient separation model is used to further improve the purity, and the ultrafiltration concentration model is used to concentrate the product. This model-driven optimization strategy based on physicochemical mechanisms can not only quickly determine the optimal process parameters but also ensure the repeatability and reliability of the process results.

[0185] In addition, the method of the present invention also makes innovations in parameter acquisition and error correction. On the one hand, a series of advanced characterization methods such as dynamic light scattering, electrophoretic light scattering, and fluorescence tracing are used to accurately measure the required key parameters. On the other hand, through the fitting of a large amount of experimental data, the error terms in the physicochemical models are corrected, greatly improving the prediction accuracy of the models. These strategies ultimately enable the method of the present invention to reliably prepare high-quality extracellular vesicle products.

[0186] To better understand and implement the present invention, Example 2 of a specific application scenario of the present invention is provided below:

[0187] A certain biomedical company intends to develop a regenerative medicine product based on extracellular vesicles. To this end, the company commissions a laboratory to conduct research on the preparation of high-purity and high-concentration extracellular vesicles.

[0188] The experimental personnel first collected the culture supernatant of human umbilical cord mesenchymal stem cells as the raw material source of extracellular vesicles. Referring to the process flow of the present invention, a specific implementation plan for the isolation and purification of extracellular vesicles was designed.

[0189] In the first step, the cell culture supernatant was pretreated. The supernatant was filtered successively through 200-micron, 100-micron, and 50-micron filters to remove large particulate impurities, obtaining a preliminary filtrate. Then, the preliminary filtrate was centrifuged at a speed of 300 revolutions per minute for 10 minutes, and the supernatant was collected. Subsequently, the supernatant was centrifuged at a speed of 3000 revolutions per minute for 20 minutes, and the supernatant was collected. The main purpose of this centrifugation was to further remove larger cell organelles and cell debris.

[0190] In the second step, an orthogonal test plan was designed to optimize the process parameters. According to the present invention, the first centrifugation speed ω1, the first centrifugation time t1, the second centrifugation speed ω2, the second centrifugation time t2, the pore size d of the filter membrane pore , the density gradient medium concentration ρ gra , the molecular weight cut-off M of the ultrafiltration membrane cut were selected as the optimization targets. Referring to the literature, an orthogonal test plan with 10 sample points was designed, as shown in Table 1 specifically.

[0191] Table 1. Orthogonal test plan for the extracellular vesicle preparation process

[0192]

[0193] In the third step, the characteristics of the samples were detected. First, the dynamic light scattering method was used to measure the particle size distribution of extracellular vesicles. During the experiment, the laser power was controlled at 5 milliwatts, the sampling time was 180 seconds, and the sampling interval was 1 second. The light scattering signals were analyzed by the cumulative method to obtain the average particle size and particle size distribution data.

[0194] Secondly, a nanoparticle tracking analyzer was used to measure the concentration of extracellular vesicles. During the experiment, the camera gain was set at 8, and the number of frames collected was 1000 frames. By analyzing the movement trajectories of the particles in the sample, quantitative data on the vesicle concentration were obtained.

[0195] Then, a protein quantification kit was used to determine the total protein content in the extracellular vesicles. The experimental results showed that the protein content of samples under different process conditions varied greatly, ranging from 0.2 to 1.5 mg / ml.

[0196] Next, the morphological characteristics of extracellular vesicles were observed using transmission electron microscopy. In the experiment, the accelerating voltage was set at 80 kilovolts. Electron microscopic images showed that the vesicles had a relatively regular spherical morphology, and the diameters were mostly concentrated in the range of 50 to 150 nanometers.

[0197] Finally, flow cytometry was used to analyze the expression of specific markers of extracellular vesicles. The results showed that the purity of samples under different process conditions was significantly different, and the proportion of CD9 and CD81 positive cells was between 60% and 90%.

[0198] Figure 2 The changes in the particle size distribution of extracellular vesicles at different separation stages are shown. The horizontal axis is the particle size (with a subscript dp), and the vertical axis is the relative number distribution percentage. The three curves represent the particle size distribution after pretreatment, after the first centrifugation, and after the second centrifugation.

[0199] The fourth step is to preprocess the test data. First, the 3-times standard deviation method is used to eliminate outliers. Statistical analysis shows that about 5% of the data points are identified as outliers.

[0200] Next, the maximum and minimum normalization method was used to normalize the data. Taking the vesicle particle size as an example, the original data ranged from 50 to 150 nanometers, which was converted into a dimensionless index between 0 and 1 after normalization.

[0201] Then, principal component analysis was used to reduce the data dimension. The analysis results showed that the first three principal components could explain about 90% of the total variance, so these three principal components were selected as input variables for subsequent modeling.

[0202] Afterwards, the hierarchical clustering method was used to classify the samples. Cluster analysis showed that the 10 samples could be classified into 4 different categories, corresponding to different separation and purification effects.

[0203] Finally, the Pearson correlation coefficient was used to analyze the correlation between the parameters. The results showed that the vesicle size was negatively correlated with the concentration (r = -0.72), while the vesicle concentration was positively correlated with the protein content (r = 0.84). This provides a reference for the subsequent establishment of a physical and chemical model.

[0204] Figure 3 The three-dimensional scatter plot of the principal component analysis results shows the distribution of the 10 samples in the first three principal component spaces. Different colors in the figure represent different clustering groups, and the coordinate axis shows the percentage of variance explained by each principal component.

[0205] Step 5: Establish a physicochemical model for the isolation and purification of extracellular vesicles.

[0206] First, establish a differential centrifugation model. This model can be expressed as:

[0207]

[0208] where v s is the sedimentation velocity of the vesicles, d p is the particle size of the vesicles, ρ p is the density of the vesicles, ρ f is the density of the medium, ω is the angular velocity of the centrifuge, r is the radius of rotation, μ is the dynamic viscosity of the medium, k B is the Boltzmann constant, T is the temperature, k c is the aggregation coefficient, C p is the vesicle concentration, n is the aggregation index, and η1 is the error correction term.

[0209] This model takes into account centrifugal force, Brownian motion, vesicle aggregation, and hydrodynamic effects, and can accurately predict the separation of vesicles with different particle sizes and densities in the centrifugal force field.

[0210] Based on the particle size and density data obtained in Step 3 and combined with the aforementioned centrifugation process parameters, the vesicle recovery models for the first and second centrifugations can be calculated as:

[0211]

[0212] where R1 is the recovery rate, d min , d max is the vesicle particle size range, ρ min , ρ max is the vesicle density range, f(d p , ρ p ) is the combined particle size and density distribution, h is the height of the centrifuge tube, and t is the centrifugation time.

[0213] Secondly, establish a membrane filtration model, which can be expressed as:

[0214]

[0215] where J is the permeation flux, P is the membrane porosity, Δp is the transmembrane pressure difference, R m is the intrinsic resistance of the membrane, R c is the resistance of the protein concentration polarization layer, R p is the resistance of the vesicle blocking layer, β is the blocking coefficient, α is the attenuation coefficient, is the surface potential of the vesicles, is the surface potential of the membrane, κ is the Boltzmann constant, Kad is the adsorption equilibrium constant, and η2 is the error correction term.

[0216] This model takes into account factors such as membrane pore sieving, electrostatic interaction, and surface adsorption, and can accurately predict the separation effect under different membrane pore sizes and pressure differences.

[0217] Then, a density gradient separation model is established, which can be expressed as:

[0218]

[0219] where C is the local concentration, t is the time, D is the diffusion coefficient, x is the position coordinate, v g is the gravitational sedimentation velocity, S T is the thermal diffusion coefficient, T is the temperature distribution, and η3 is the error correction term.

[0220] This model comprehensively considers buoyancy balance, diffusion mechanism, and temperature gradient effect, and can accurately predict the separation process of vesicles in the centrifugal force field under different density gradients.

[0221] Finally, an ultrafiltration concentration model is established, which can be expressed as:

[0222]

[0223] where J v is the volume flux, L p is the membrane permeability coefficient, σ is the reflection coefficient, Δπ is the osmotic pressure difference, δ is the boundary layer thickness, D is the diffusion coefficient, C w is the membrane wall concentration, C max is the maximum concentration, k f is the fouling coefficient, Q p is the cumulative permeation amount, and η4 is the error correction term.

[0224] This model takes into account factors such as membrane mass transfer, osmotic pressure, and concentration polarization, and can comprehensively predict the concentration effect under different membrane retention molecular weight conditions.

[0225] Step 6: Use an optimization algorithm to determine the optimal process parameters.

[0226] First, use the least squares method to fit the parameters in the above physicochemical model. Statistical analysis shows that the goodness of fit R between the model and the experimental data 2 is all above 0.85, indicating that the model prediction results are in good agreement with the actual situation.

[0227] Next, use the cross-validation method to evaluate the reliability of the model parameters. The results show that 8 out of 10 sample points can be successfully predicted, verifying the generalization ability of the model.

[0228] Then, the gradient descent method is used to optimize the weight coefficients in the physicochemical model. After 20 iterations, the prediction error RMSE of the model decreases from the initial 0.16 to 0.08, and the prediction accuracy is significantly improved.

[0229] After that, the genetic algorithm is used to globally optimize the process parameters. After 50 generations of evolution, the finally determined optimal parameter combination is: revolutions per minute, minutes, revolutions per minute, minutes, micrometers, grams per milliliter, kilodaltons.

[0230] Finally, the Monte Carlo method is used to evaluate the prediction uncertainty of the entire preparation process. Statistical analysis shows that at the 95% confidence level, the predicted vesicle purity by the model is between 92% and 98%, and the concentration is between 8×10^8 and 1.2×10^9 per milliliter. This provides an important reference for subsequent process scale-up.

[0231] Figure 4 is a three-dimensional response surface plot showing the effects of the first centrifugation speed (ω1) and the first centrifugation time (t1) on the vesicle recovery rate (R1). The color change in the figure represents the magnitude of the recovery rate, and the optimal parameter combination can be intuitively seen.

[0232] Step 7, verify the actual effect of the optimal process parameters.

[0233] First, substitute the above optimal process parameters into the physicochemical model to obtain the theoretical prediction values. The results show that the predicted vesicle recovery rate for the first centrifugation is 85%, and the purity is 82%; the vesicle retention rate for membrane filtration is 92%, and the flux is 380 liters per square meter per hour; the vesicle purity by density gradient centrifugation can reach 95%, and the separation efficiency is 88%; the vesicle recovery rate for ultrafiltration concentration is 90%, and the final concentration can reach 1.1×10 9 per milliliter.

[0234] Next, prepare extracellular vesicles according to the optimal parameter combination. The specific operations are as follows:

[0235] (1) Perform the first centrifugation under the conditions of 1200 revolutions per minute and 18 minutes, and collect the supernatant.

[0236] (2) Filter using a membrane with a pore size of 0.08 micrometers, and control the transmembrane pressure difference at 0.2 megapascals.

[0237] (3) Centrifuge under the conditions of a sucrose density gradient of 1.12 grams per milliliter, 100000 revolutions per minute, and 120 minutes.

[0238] (4) Ultrafiltration membrane with a molecular weight cut-off of 300 kDa was used, and the transmembrane pressure difference was 0.5 MPa for concentration.

[0239] The experimental results showed that for the finally obtained extracellular vesicle sample, the particle size distribution was concentrated between 80 and 150 nm, with an average particle size of 105 nm; the concentration reached 1.08×10 9 per milliliter; the protein content was 1.2 mg per milliliter; the proportion of CD9- and CD81-positive cells reached 96%. These indicators were highly consistent with the model prediction results, verifying the reliability of the method of the present invention.

[0240] Figure 5 It is a double Y-axis chart showing the changing trends of the recovery rate and purity during the entire separation and purification process. The blue curve (left axis) represents the recovery rate, the red curve (right axis) represents the purity, and the horizontal axis represents the process steps.

[0241] The variables of the present invention and their explanations are shown in Table 2:

[0242] Table 2, Variable Explanation Table

[0243]

[0244]

[0245] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention.

Claims

1. A method for preparing high-purity extracellular vesicles, characterized in that, Including: Collecting the cell culture supernatant produced by cultured cells, and setting process parameter combinations based on the principle of orthogonal experiment; Performing the first centrifugation to remove cell debris, and obtaining the first centrifugation supernatant; Performing the second centrifugation on the first centrifugation supernatant to collect extracellular vesicles, and obtaining the first extracellular vesicles; Filtering and purifying the first extracellular vesicles to obtain the second extracellular vesicles; purifying the second extracellular vesicles to obtain the third extracellular vesicles; concentrating the third extracellular vesicles to obtain the fourth extracellular vesicles; Performing characteristic detection on the fourth extracellular vesicles, recording the detection data and performing standardization processing to obtain a standardized data set, establishing a physicochemical model of the preparation process based on the standardized data set; performing parameter fitting on the physicochemical model of the preparation process to obtain the optimal process parameter combination; using the optimal process parameter combination to perform an extracellular vesicle preparation experiment, obtaining experimental verification data, correcting the physicochemical model of the preparation process, and performing error correction to obtain the final process parameter combination for preparing extracellular vesicles.

2. The method for preparing high-purity extracellular vesicles according to claim 1, wherein The step of collecting the cell culture supernatant produced by cultured cells and setting process parameter combinations based on the principle of orthogonal experiment specifically includes: filtering the cell culture supernatant through a 200-micron filter screen, a 100-micron filter screen, and a 50-micron filter screen in sequence to obtain a preliminary filtrate; centrifuging the preliminary filtrate at 300 revolutions per minute for 10 minutes, and collecting the supernatant; centrifuging the supernatant at 3000 revolutions per minute for 20 minutes, and collecting the supernatant; designing an orthogonal experiment scheme according to 10 to 20 sample points, and recording the first centrifugation speed, the first centrifugation time, the second centrifugation speed, the second centrifugation time, the pore size of the filter membrane, the concentration of the density gradient medium, and the molecular weight cut-off of the ultrafiltration membrane; obtaining the initial particle size distribution data by using a particle size analyzer.

3. The method for preparing high-purity extracellular vesicles according to claim 2, wherein The step of performing characteristic detection on the fourth extracellular vesicles and recording the detection data specifically includes: measuring the particle size distribution of extracellular vesicles by dynamic light scattering method, controlling the laser power to be 5 mW, the sampling time to be 180 s, and the sampling interval to be 1 s; measuring the concentration of extracellular vesicles by a nanoparticle tracking analyzer, setting the camera gain to be 8, and the number of collected frames to be 1000 frames; measuring the total protein content in extracellular vesicles by using a protein quantification kit; observing the morphological characteristics of extracellular vesicles by transmission electron microscopy, setting the acceleration voltage to be 80 kV; analyzing the expression of specific markers of extracellular vesicles by flow cytometry.

4. The method for preparing high-purity extracellular vesicles according to claim 3, characterized in that, The step of performing data cleaning and normalization processing on the detection data to obtain a standardized data set specifically includes: removing outliers in the detection data by the 3-fold standard deviation method; performing standardization processing on the detection data by the maximum-minimum normalization method; reducing the dimension of the detection data by principal component analysis; classifying the detection data by hierarchical clustering; analyzing the correlation between parameters of the detection data by Pearson correlation coefficient.

5. The method for preparing high-purity extracellular vesicles according to claim 4, wherein, The steps of establishing the physicochemical model of the preparation process specifically include: establishing the Stokes sedimentation equation to describe the movement of extracellular vesicles during centrifugation; establishing the Darcy permeability equation to describe the mass transfer of extracellular vesicles during membrane filtration; establishing the diffusion-convection equation to describe the migration of extracellular vesicles in the density gradient; establishing the concentration polarization equation to describe the concentration of extracellular vesicles during ultrafiltration; and numerically solving the physicochemical equations using the finite element method.

6. The method for preparing high-purity extracellular vesicles according to claim 5, wherein The steps of parameter fitting for the physicochemical model of the preparation process specifically include: fitting the parameters in the physicochemical equations using the least squares method; evaluating the parameter fitting accuracy using the cross-validation method; optimizing the weight coefficients in the physicochemical equations using the gradient descent method; globally optimizing the physicochemical equations using the genetic algorithm; and evaluating the prediction error of the physicochemical equations using the Monte Carlo method.

7. The method for preparing high-purity extracellular vesicles according to claim 6, characterized in that, The steps of conducting the extracellular vesicle preparation experiment using the optimal process parameter combination specifically include: inputting the optimal process parameter combination into the physicochemical model of the preparation process to obtain the theoretical prediction value; conducting the first centrifugation using the optimal process parameter combination, controlling the first centrifugation speed and the first centrifugation time, and collecting the first extracellular vesicles; conducting membrane filtration using the optimal process parameter combination, controlling the pore size of the filtration membrane, and collecting the second extracellular vesicles; conducting density gradient centrifugation using the optimal process parameter combination, controlling the concentration of the density gradient medium, and collecting the third extracellular vesicles; and conducting ultrafiltration concentration using the optimal process parameter combination, controlling the molecular weight cut-off of the ultrafiltration membrane, and collecting the fourth extracellular vesicles.

8. The method for preparing high-purity extracellular vesicles according to claim 7, wherein, The steps of modifying the physicochemical model of the preparation process specifically include: calculating the root mean square error between the theoretical prediction value and the experimental verification data; modifying the parameters in the physicochemical model of the preparation process using the adaptive weight method; iteratively optimizing the physicochemical model of the preparation process using the Bayesian optimization algorithm; determining the key parameters in the physicochemical model of the preparation process using the sensitivity analysis method; and evaluating the robustness of the physicochemical model of the preparation process using the stability analysis method.

9. The method for preparing high-purity extracellular vesicles according to claim 8, wherein, The steps of establishing the physicochemical model of the preparation process further include: establishing the basic Stokes sedimentation equation, introducing the Brownian motion correction term and the vesicle aggregation correction term; establishing the basic Darcy permeability equation, introducing the electrostatic interaction term, the surface adsorption term, and the membrane fouling term; establishing the basic density gradient separation equation, introducing the thermal diffusion term and the convection term; establishing the basic ultrafiltration concentration equation, introducing the concentration polarization term and the membrane fouling kinetics term; and numerically solving the physicochemical model of the preparation process.

10. The method for preparing high-purity extracellular vesicles according to claim 9, wherein The steps of establishing the physicochemical model of the preparation process further include the parameter acquisition steps: measuring the particle size distribution of extracellular vesicles using dynamic light scattering to obtain the particle size parameters; measuring the density distribution of extracellular vesicles using linear sucrose density gradient centrifugation to obtain the density parameters; measuring the surface potential of extracellular vesicles using electrophoretic light scattering to obtain the surface potential parameters; measuring the membrane permeability coefficient using the standard protein permeation experiment to obtain the permeability coefficient parameters; measuring the diffusion coefficient using fluorescence tracer method to obtain the diffusion coefficient parameters.

Images