Mesenchymal stromal cell quality evaluation system

Abstract

The present invention provides a stem cell quality evaluation system, which includes a quality scoring module for calculating a stem cell quality score according to the expression level of characteristic genes related to stem cell quality and the weight coefficient of the characteristic genes, a quality evaluation module for evaluating stem cell quality according to the quality score of the stem cell, and a result output module for outputting a stem cell quality result report, and further includes a subpopulation clustering module for obtaining single cell gene expression data and specific quality attributes of stem cells, and a subpopulation identification module for determining a stem cell quality prediction model, characteristic genes related to stem cell quality and the weight coefficient of the characteristic genes according to the single cell gene expression data and specific quality attributes of stem cells. The present invention can achieve the effect of accurately and quantitatively evaluating stem cell quality, and the system can be used to screen high quality stem cells.

Description

[Technical Field]

[0001] [CROSS-REFERENCE TO RELATED APPLICATIONS] This application claims priority from Chinese Patent Application No. 202210043060.6, filed on January 14, 2022, the entire contents of which are incorporated herein by reference.

[0002] [Technical field] The present invention relates to the technical field of mesenchymal stromal cells, and to a system for evaluating the quality of mesenchymal stromal cells. [Background technology]

[0003] Mesenchymal stromal cell therapy is expected to fundamentally change the clinical dilemmas of intractable diseases faced by conventional medicine by restoring tissue function and treating the root causes of degenerative diseases, and is a future direction for medical development.

[0004] Obtaining sufficient mesenchymal stromal cells through appropriate expansion culture is a necessary prerequisite for supporting mesenchymal stromal cell research and applications. However, during the expansion process, different microenvironments can lead to changes in gene expression and heterogeneity among mesenchymal stromal cells of the same origin. This heterogeneity significantly hinders scientific research on mesenchymal stromal cells and poses a critical risk to their clinical application. Therefore, assessing heterogeneity during the expansion process of mesenchymal stromal cells is an important prerequisite for the clinical development of mesenchymal stromal cell therapy.

[0005] Single-cell RNA sequencing (scRNA-seq) offers the possibility of investigating cell-to-cell heterogeneity and can preliminarily analyze the heterogeneity of cell subpopulations based on gene expression profiles. However, it cannot clarify the relationship between cell heterogeneity and cell quality, nor can it quantitatively determine mesenchymal stromal cell quality.

[0006] CN113061638A provides a mesenchymal stromal cell evaluation system that performs sterility tests, safety tests, cell activity tests, and cell morphology tests on mesenchymal stromal cells.

[0007] However, existing technologies lack sound and unified standards for mesenchymal stromal cell quality assessment, failing to accurately demonstrate the effects of the microenvironment on mesenchymal stromal cells and failing to resolve safety issues in mesenchymal stromal cell therapy. The mesenchymal stromal cell industry still faces major challenges, including incomplete quality control systems, incomplete mechanism research, and non-standard clinical applications. Summary of the Invention

[0008] In response to the shortcomings of the prior art and practical needs, the present invention provides a mesenchymal stromal cell quality assessment system that establishes important classification standards for mesenchymal stromal cell quality at the single-cell level. Single-cell gene expression datasets of mesenchymal stromal cells labeled with quality attributes were obtained based on single-cell transcriptome analysis and functional clustering of cell subpopulations. A mesenchymal stromal cell quality prediction model was constructed using supervised machine learning methods, and feature genes and feature gene weighting coefficients associated with mesenchymal stromal cell quality were determined. The quality risk due to heterogeneity of mesenchymal stromal cells was quantitatively assessed.

[0009] First, the present invention provides a mesenchymal stromal cell quality evaluation system including an evaluation subsystem, wherein the evaluation subsystem: a quality scoring module for calculating a quality score of the mesenchymal stromal cells based on the expression levels of the characteristic genes associated with the mesenchymal stromal cell quality and the weight coefficients of the characteristic genes; and a quality assessment module for assessing the quality of the mesenchymal stromal cells based on the quality score of the mesenchymal stromal cells.

[0010] Although clinical mesenchymal stromal cells share the same cell biological properties, they are subject to heterogeneity due to the influence of the microenvironment. To demonstrate cell heterogeneity, predict cell state / fate, and further evaluate the quality of mesenchymal stromal cells, bioinformatics is used to determine the characteristic genes and characteristic gene weighting factors associated with mesenchymal stromal cell quality, and the quality of mesenchymal stromal cells is further evaluated based on the expression levels of the characteristic genes and the characteristic gene weighting factors.

[0011] In the present invention, a supervised machine learning model is used to train a single-cell gene expression dataset labeled with quality attributes, and feature genes and weighting factors for the feature genes that can accurately characterize the differences between different mesenchymal stromal cells and are related to mesenchymal stromal cell quality are determined. Based on the expression levels of the feature genes and the weighting factors for the feature genes of the mesenchymal stromal cell samples being tested, a quality score for the mesenchymal stromal cells is calculated to quantitatively evaluate the quality of the mesenchymal stromal cells.

[0012] Preferably, the quality scoring module includes: (1) a feature gene expression level acquisition unit for obtaining expression levels of feature genes related to mesenchymal stromal cell quality; and (2) a calculation unit for calculating a quality score of mesenchymal stromal cells based on the expression levels of the feature genes and the weighting coefficients of the feature genes, and the calculation function of the quality score of the mesenchymal stromal cells is as follows:

[0013]

number

[0014] Preferably, the method for obtaining the expression level of the characteristic gene includes common gene quantification methods in the art, such as single-cell sequencing, high-throughput sequencing, microarray chip, qPCR, etc. Preferably, the expression level of the characteristic gene is obtained by single-cell sequencing.

[0015] Preferably, the quality assessment module includes: (1) a mesenchymal stromal cell quality risk threshold determination unit for analyzing the quality score of mesenchymal stromal cells of a dataset using a receiver operating characteristic curve and an area under the curve, wherein the numerical value of the highest point of the receiver operating characteristic curve is a mesenchymal stromal cell quality risk threshold, and the dataset includes single-cell gene expression data of mesenchymal stromal cells having known specific quality attributes; and (2) a comparison and judgment unit for comparing the quality score of the mesenchymal stromal cells with the mesenchymal stromal cell quality risk threshold, wherein if the quality score of the mesenchymal stromal cells is greater than or equal to the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are quality risk mesenchymal stromal cells, and if the quality score of the mesenchymal stromal cells is less than the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are non-quality risk mesenchymal stromal cells.

[0016] Preferably, the evaluation subsystem further includes a result output module for outputting a result report from the mesenchymal stromal cell quality evaluation module.

[0017] Secondly, in the present invention, a subpopulation clustering module for obtaining single-cell gene expression data and specific quality attributes of mesenchymal stromal cells; and a subpopulation identification module for determining and / or optimizing a mesenchymal stromal cell quality prediction model, feature genes associated with mesenchymal stromal cell quality, and weighting coefficients of the feature genes based on the single-cell gene expression data and specific quality attributes of the mesenchymal stromal cells.

[0018] Preferably, the subpopulation clustering module includes: (1) a single-cell gene expression data acquisition unit for preprocessing single-cell RNA sequencing data to obtain single-cell gene expression data of mesenchymal stromal cells; (2) a pathway score matrix acquisition unit for performing pathway enrichment analysis on the single-cell gene expression data of mesenchymal stromal cells, calculating pathway scores in each mesenchymal stromal cell for each gene-enriched pathway, and obtaining a score matrix of single-cell pathway enrichment of mesenchymal stromal cells; and (3) a specific quality attribute determination unit for performing standardization, dimensionality reduction, clustering, and visualization processes on the pathway score matrix of single-cell pathway enrichment of mesenchymal stromal cells, and obtaining clustering results of single-cell subpopulations of mesenchymal stromal cells as specific quality attributes of mesenchymal stromal cells.

[0019] In the present invention, a pathway score matrix is ​​constructed by combining traditional cell subpopulation clustering, differential gene analysis, and pathway enrichment analysis. In this pathway score matrix, each row represents the expression status of a pathway in different mesenchymal stromal cells, each column represents a cell index, and each grid data represents the expression status of a specific pathway in a specific mesenchymal stromal cell. This pathway-based functional clustering method has the effect of quickly identifying functional differences in mesenchymal stromal cells.

[0020] Preferably, the subpopulation identification module includes: (1) a dataset creation unit for forming single-cell gene expression data of mesenchymal stromal cells labeled with specific quality attributes into a dataset; (2) a dataset division unit for dividing the dataset into a training set and a test set; (3) a model training unit for training a supervised machine learning model using the training set, adjusting parameters of the supervised machine learning model using cross-validation and the test set, and confirming and / or optimizing a mesenchymal stromal cell quality prediction model; and (4) a feature gene output unit for outputting feature genes and weighting coefficients related to mesenchymal stromal cell quality according to the mesenchymal stromal cell quality prediction model.

[0021] In the present invention, the subpopulation identification module of the mesenchymal stromal cell quality evaluation system functions to use single-cell gene expression data of mesenchymal stromal cells labeled with known quality attributes as a dataset, which is randomly divided into a training set and a test set at a certain ratio, and train a supervised machine learning model (the feature quantities of the supervised machine learning model are determined using the training set, and the parameters are adjusted using the test set to optimize the supervised machine learning model), thereby obtaining a mesenchymal stromal cell quality prediction model with good accuracy, precision, recall, and F1 score.

[0022] Preferably, the supervised machine learning model includes any one of a perceptron model, a K-nearest neighbor model, a naive Bayes model, a decision tree model, a logistic regression, a support vector machine, a random forest, a boosting model, an EM algorithm, or a conditional random field.

[0023] Preferably, the mesenchymal stromal cells include any one or a combination of at least two of adult mesenchymal stromal cells, embryonic mesenchymal stromal cells, induced pluripotent mesenchymal stromal cells, or mesenchymal stromal cells transformed from mature somatic cells, and their derivative cells.

[0024] Preferably, the mesenchymal stromal cells include any one or a combination of at least two of mesenchymal stromal cells, mesenchymal stromal cells, multipotent stromal cells, multipotent mesenchymal stromal cells, and medicinal signaling cells.

[0025] Preferably, the mesenchymal stromal cells include any one or a combination of at least two of adipose-derived mesenchymal stromal cells, umbilical cord mesenchymal stromal cells, placenta-derived mesenchymal stromal cells, bone marrow mesenchymal stromal cells, dental marrow mesenchymal stromal cells, menstrual blood-derived mesenchymal stromal cells, amniotic epithelial mesenchymal stromal cells, and bronchial basal cells.

[0026] In the present invention, The subpopulation clustering module of the optimization subsystem provides a method for optimizing the mesenchymal stromal cell quality evaluation system, including preprocessing newly obtained single-cell RNA sequencing data to obtain single-cell gene expression data, performing pathway enrichment analysis on the single-cell gene expression data to calculate pathway scores for each mesenchymal stromal cell to obtain a pathway score matrix, and performing standardization, dimensionality reduction, clustering, and visualization processes on the pathway score matrix to obtain clustering results as specific quality attributes of the mesenchymal stromal cells.

[0027] The subpopulation identification module of the optimization subsystem incorporates single-cell gene expression data of mesenchymal stromal cells labeled with specific quality attributes into the original dataset to create a new dataset, which is divided into a training set and a test set. A supervised machine learning model is trained using the training set, and parameters of the supervised machine learning model are adjusted through cross-validation and testing using the test set, thereby optimizing the mesenchymal stromal cell quality prediction model and outputting feature genes and weighting coefficients related to mesenchymal stromal cell quality.

[0028] The quality scoring module of the evaluation subsystem obtains expression levels of feature genes related to mesenchymal stromal cell quality, and calculates a quality score of the mesenchymal stromal cell based on the expression levels of the feature genes and the weighting coefficients of the feature genes.

[0029] The calculation function for the quality score of the mesenchymal stromal cells is as follows:

number

[0030] The quality assessment module of the assessment subsystem compares the quality score of the mesenchymal stromal cells with a quality risk threshold, and the dataset includes single-cell gene expression data of the mesenchymal stromal cells labeled with known specific quality attributes.

[0031] If the quality score of the mesenchymal stromal cells is greater than or equal to the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are quality risk mesenchymal stromal cells.

[0032] If the quality score of the mesenchymal stromal cells is less than the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are non-quality risk mesenchymal stromal cells.

[0033] Thirdly, in the present invention, obtaining single-cell gene expression data and specific quality attributes of mesenchymal stromal cells to form a dataset divided into a training set and a test set; and A method for constructing a mesenchymal stromal cell quality prediction model is provided, which includes training a supervised machine learning model using a training set, adjusting parameters of the supervised machine learning model through cross-validation and testing using a test set, and determining the mesenchymal stromal cell quality prediction model.

[0034] Preferably, the method for obtaining single cell gene expression data and specific quality attributes of mesenchymal stromal cells comprises: Obtain single-cell gene expression data of mesenchymal stromal cells by performing single-cell RNA sequencing on mesenchymal stromal cells. Performing pathway enrichment analysis on single-cell gene expression data of mesenchymal stromal cells and calculating the pathway enrichment score for each mesenchymal stromal cell to obtain a pathway score matrix; and This involves performing bioinformatics analysis on the pathway score matrix to obtain clustering results of mesenchymal stromal cells as specific quality attributes of mesenchymal stromal cells.

[0035] Preferably, the bioinformatics analysis of the pathway score matrix of the mesenchymal stromal cells includes: This involves performing dimension reduction and clustering processes on the pathway score matrix.

[0036] Preferably, the construction method further includes determining characteristic genes and characteristic gene weighting coefficients associated with mesenchymal stromal cell quality using a mesenchymal stromal cell quality prediction model.

[0037] Compared with the prior art, the present invention has the following beneficial effects: (1) The mesenchymal stromal cell quality evaluation system of the present invention accurately and quantitatively evaluates the quality of mesenchymal stromal cells using a quality scoring module and a quality evaluation module based on the determined mesenchymal stromal cell quality-related feature genes and the weight coefficients of the feature genes. The system outputs the mesenchymal stromal cell quality based on the weighted sum of the mesenchymal stromal cell quality-related feature genes and the quality risk threshold, and is a regular, complete, and unified mesenchymal stromal cell quality evaluation system. (2) The mesenchymal stromal cell quality assessment system of the present invention continuously updates a mesenchymal stromal cell quality standard map at the single-cell level using a subpopulation clustering module and a subpopulation identification module based on accumulated single-cell RNA sequencing data of mesenchymal stromal cells. Using this mesenchymal stromal cell quality standard map as a dataset, a mesenchymal stromal cell quality prediction model is optimized based on a supervised machine learning model, thereby improving the accuracy of mesenchymal stromal cell quality assessment results. (3) The mesenchymal stromal cell quality evaluation system of the present invention achieved the effect of accurately and quantitatively indicating the cellular heterogeneity of mesenchymal stromal cells due to the influence of the microenvironment. (4) The mesenchymal stromal cell quality evaluation system of the present invention can be used to screen for high-quality mesenchymal stromal cells. [Brief explanation of the drawings]

[0038] [Figure 1A] FIG. 1A is a growth curve of D1M1-P5. [Figure 1B] FIG. 1B is a growth curve of D1M2-P5. [Figure 1C] FIG. 1C is a cell cycle analysis of D1M1-P5. [Figure 1D] FIG. 1D is a cell cycle analysis of D1M2-P5. [Figure 1E] FIG. 1E shows the apoptotic group of D1M1-P5. [Figure 1F] FIG. 1F shows the apoptotic group of D1M2-P5. [Figure 1G] FIG. 1G shows adipogenic, osteogenic, and chondrogenic differentiation of D1M1-P5. [Figure 1H] FIG. 1H shows adipogenic, osteogenic, and chondrogenic differentiation of D1M2-P5. [Figure 2A] FIG. 2A shows lung tissue and HE staining results when D1M1-P5, D1M2-P5, or saline was injected into mice (black arrows indicate venous thrombus formation). [Figure 2B]FIG. 2B shows the density of emboli expressed in each 10× field (*p<0.05). [Figure 2C] FIG. 2C shows the results of lung fluorescence in mice injected with D1M1-P5, D1M2-P5, or saline. [Figure 2D] FIG. 2D shows the number of PKH26+ expressing cells in each 10× field. [Figure 3A] FIG. 3A shows the clustering results of D1M1-P5 and D1M2-P5 cell subpopulations (0, 1, 2, 3, 4, and 5 represent different mesenchymal stromal cell subpopulations, respectively). [Figure 3B] Figure 3B shows the expression of different mesenchymal stromal cell subpopulations for risk genes based on the GO-BP database (C0, C1, C2, C3, C4, and C5 (corresponding to mesenchymal stromal cell subpopulations 0, 1, 2, 3, 4, and 5 in Figure 3A) represent different mesenchymal stromal cell subpopulations, respectively). [Figure 3C] Figure 3C shows the expression of risk genes in different mesenchymal stromal cell subpopulations based on the KEGG database (C0, C1, C2, C3, C4, and C5 (corresponding to mesenchymal stromal cell subpopulations 0, 1, 2, 3, 4, and 5 in Figure 3A) represent different mesenchymal stromal cell subpopulations, respectively). [Figure 4] FIG. 1 is a schematic diagram of the mechanism of the functional clustering method. [Figure 5A] Figure 5A shows the functional clustering results of cell subpopulations obtained by the ssGSEA scoring function (A2105C2P5 (i.e., D1M1-P5) were quality risk mesenchymal stromal cells, and A2105C3P5 (i.e., D1M2-P5) were both non-quality risk mesenchymal stromal cells). [Figure 5B] Figure 5B shows the functional clustering results of cell subpopulations obtained using the AUCell scoring function (A2105C2P5 (i.e., D1M1-P5) were all quality risk mesenchymal stromal cells, and A2105C3P5 (i.e., D1M2-P5) were all non-quality risk mesenchymal stromal cells). [Figure 5C]Figure 5C shows the functional clustering results of cell subpopulations obtained using the Seurat scoring function (A2105C2P5 (i.e., D1M1-P5) were all quality risk mesenchymal stromal cells, and A2105C3P5 (i.e., D1M2-P5) were all non-quality risk mesenchymal stromal cells). [Figure 6] FIG. 1 is a schematic diagram of cross-culture of mesenchymal stromal cells. [Figure 7A] Figure 7A shows the functional clustering results of cell subpopulations obtained by the ssGSEA scoring function (D1M1-P3, D1M1-P5, and D1M2 / M1-P5 were all quality-risk mesenchymal stromal cells, while D1M2-P3, D1M2-P5, and D1M1 / M2-P5 were all non-quality-risk mesenchymal stromal cells). [Figure 7B] Figure 7B shows the functional clustering results of cell subpopulations obtained using the AUCell scoring function (D1M1-P3, D1M1-P5, and D1M2 / M1-P5 were all quality risk mesenchymal stromal cells, and D1M2-P3, D1M2-P5, and D1M1 / M2-P5 were all non-quality risk mesenchymal stromal cells). [Figure 7C] Figure 7C shows the functional clustering results of cell subpopulations obtained using the Seurat scoring function (D1M1-P3, D1M1-P5, and D1M2 / M1-P5 were all quality-risk mesenchymal stromal cells, while D1M2-P3, D1M2-P5, and D1M1 / M2-P5 were all non-quality-risk mesenchymal stromal cells). [Figure 8A] Figure 8A shows lung tissue and HE staining results when mice were injected with D1M1-P3, D1M2-P3, D1M2 / M1-P5, D1M2 / M1-P5, or saline (black arrows indicate venous thrombus formation). [Figure 8B] FIG. 8B shows the density of emboli expressed in each 10× field (*p<0.05). [Figure 9] FIG. 1 is a schematic diagram of the construction mechanism of the mesenchymal stromal cell quality prediction model. [Figure 10]This figure shows the change curves of cross-validation accuracy with increasing gene quantity in the recursive feature reduction (RFE) process, and enlarged inflection points of eight different RFE change curves (M1, M2, M3, M4, M5, M6, M7, and M8). [Figure 11] This is the quality score threshold with the highest sensitivity and specificity for classifying mesenchymal stromal cells into quality risk or non-quality risk mesenchymal stromal cells in the four measurement datasets. [Figure 12A] FIG. 12A is a density distribution of mesenchymal stromal cell quality scores for test set 1. [Figure 12B] FIG. 12B is a density distribution of mesenchymal stromal cell quality scores for test set 2. [Figure 12C] FIG. 12C is a density distribution of mesenchymal stromal cell quality scores for test set 3. [Figure 12D] FIG. 12D is a density distribution of mesenchymal stromal cell quality scores for test set 4. [Figure 13A] FIG. 13A is a schematic diagram of the evaluation subsystem 10 of the mesenchymal stromal cell quality evaluation system (110: quality scoring module, 120: quality evaluation module). [Figure 13B] FIG. 13B is a schematic diagram of the quality scoring module 110 (1110: feature gene expression level acquisition unit, 1120: calculation unit). [Figure 13C] FIG. 13C is a schematic diagram of the quality assessment module 120 (1210: quality risk threshold determination unit, 1220: comparison and judgment unit). [Figure 13D] FIG. 13D is a schematic diagram of the optimization subsystem 20 of the mesenchymal stromal cell quality evaluation system (210: subpopulation clustering module, 220: subpopulation identification module). [Figure 13E] FIG. 13E is a schematic diagram of the subpopulation clustering module 210 (2110: single-cell gene expression data acquisition unit, 2120: pathway score matrix acquisition unit, 2130: specific quality attribute determination unit). [Figure 13F]FIG. 13F is a schematic diagram of the subpopulation identification module 220 (2210: dataset creation unit, 2220: dataset division unit, 2230: model training unit, 2240: feature gene output unit). [Figure 13G] FIG. 13G is a schematic diagram of the mesenchymal stromal cell quality evaluation system (10: evaluation subsystem, 20: optimization subsystem). [Figure 14] This is a prediction of the quality of D1M1 / M2-P5 and D1M2 / M1-P5 based on the characteristic genes and characteristic gene weight coefficients determined by the mesenchymal stromal cell quality prediction model. DETAILED DESCRIPTION OF THE INVENTION

[0039] To further explain the technical means and effects of the present invention, the present invention will be further described below in conjunction with examples and drawings. It should be understood that the specific embodiments described herein do not limit the present invention, but merely serve to interpret the present invention. Various modifications and variations of the method and system of the present invention will be obvious to those skilled in the art and do not depart from the scope and substance of the present invention. While the present invention has been described in conjunction with specific preferred embodiments, it should be understood that the present invention should not be unduly limited to these specific embodiments, as defined by the claims, and that various modifications and additions can be made to the embodiments within the scope of the present invention. Of course, various modifications to the embodiments made by those skilled in molecular biology and related fields for carrying out the present invention are all within the scope of the claims.

[0040] Examples that do not specify specific techniques or conditions are performed in accordance with the techniques and conditions described in the literature or product specifications. Reagents or equipment for which the manufacturer is not specified are all genuine products that can be purchased through official channels.

[0041] definition As used above and below, "mesenchymal stromal cells" refer to a type of cell that is relatively undifferentiated and capable of differentiation, is capable of actively dividing and cycling, and provides appropriate stimuli for mature, differentiated, and functional cell lines. The defining properties of mesenchymal stromal cells include: (a) the mesenchymal stromal cells themselves are not terminally differentiated; (b) they can divide indefinitely throughout the life of an animal; (c) they are characterized by consistent cell markers and are not a mixture of multiple types of mesenchymal stromal cells and / or somatic cells, but are a single type of mesenchymal stromal cell; and (d) when mesenchymal stromal cells divide, each daughter cell can either remain as a mesenchymal stromal cell or undergo a process that irreversibly induces terminal differentiation.

[0042] As described above and below, "multipotent mesenchymal stromal cells" or "mesenchymal stromal cells" are multipotent mesenchymal stromal cells that can differentiate into multiple cell types. Multipotent mesenchymal stromal cells have been shown to differentiate in vitro or in vivo into cell types including osteoblasts, chondrocytes, myocytes, and adipocytes. Mesenchyme is an embryonic connective tissue derived from the mesoderm that differentiates into hematopoietic and connective tissues, although multipotent mesenchymal stromal cells do not differentiate into hematopoietic cells.

[0043] As mentioned above and below, "mesenchymal stromal cell quality" refers to any one of the above-mentioned factors related to the safety of mesenchymal stromal cells. Mesenchymal stromal cells can be heterogeneous due to the influence of the microenvironment, and such heterogeneity can pose a quality risk. Clinical-grade mesenchymal stromal cells contain a single type of mesenchymal stromal cell, rather than a mixture of multiple types of mesenchymal stromal cells and / or mesenchymal stromal cells and somatic cells. They require rigorous third-party testing and laboratory testing closely related to the safety, efficacy, and consistency of mesenchymal stromal cells, including cell viability, biological function, tumorigenicity, thrombogenicity, immunogenicity, microbial, mycoplasma, and endotoxin testing. To avoid acute or subacute serious adverse reactions, such as fever, allergy, and bacteremia, during or after transplantation, qualified mesenchymal stromal cells must undergo release testing and compatibility testing against microorganisms, mycoplasma, and endotoxins before transplantation.

[0044] As described above and below, "mesenchymal stromal cell quality-related signature genes" refer to genes that determine mesenchymal stromal cell quality categories. Increased expression of these genes increases or decreases the quality risk of mesenchymal stromal cells.

[0045] As stated above and below, "expression level" refers to the expression level of a gene.

[0046] As described above or below, the "mesenchymal stromal cell quality score" refers to a score calculated using the following calculation function based on the expression levels of characteristic genes and the weighting coefficients of each characteristic gene determined by the mesenchymal stromal cell quality prediction model.

[0047]

number

[0048] As described above and below, the term "gene expression level" refers to the expression level of a specific gene in a cell measured by a common method in the field of molecular biology, including, for example, the hybridization level (measurement data) measured as the fluorescence intensity between probe nucleic acids immobilized on the surface of a DNA chip plate, and an estimated value of the gene expression level obtained based on that value.

[0049] As described above or below, a "specific quality attribute" refers to the clustering result of a single subpopulation of mesenchymal stromal cells determined by the subpopulation clustering method, i.e., "quality risk mesenchymal stromal cells" or "non-quality risk mesenchymal stromal cells."

[0050] As described above or below, a "pathway score matrix" has pathway identities as columns and cell indices as rows, and data in each grid represents the expression status of a particular pathway in a particular mesenchymal stromal cell. Data analysis methods (including supervised and unsupervised data analysis, as well as bioinformatics methods) are presented in Brazma and ViIo J, 2000, FEBS Lett 480(1):17-24.

[0051] Although the following examples describe the risk of mesenchymal stromal cell-induced embolism closely related to mesenchymal stromal cell quality, those skilled in the art will understand that the tumorigenicity and immunogenicity of mesenchymal stromal cells can also be identified using substantially the same methods and means according to the present invention. For example, the tumorigenicity-related characteristic gene may be c-myc, and the immunogenicity-related characteristic gene may be dnam-1 or mcp-1.

[0052] The risk of mesenchymal stromal cell-induced embolism is the most typical risk associated with mesenchymal stromal cell application and one of the most important factors affecting mesenchymal stromal cell quality. Over the past 20 years, numerous clinical cases have been reported in which embolic complications occurred after mesenchymal stromal cell therapy (Woodard, J.P. et al., Pulmonary cytolytic thrombosis: a newly recognized complication of stem cell transplantation. Bone Marrow Transpl 25, 293-300 (2000). Tatsumi, K. et al., Tissue factor triggers procoagulation in transplanted multipotent mesenchymal stromal cells leading to thromboembolism. Biochem Biophys Res Commun 431, 203-209 (2013).). Therefore, those skilled in the art will understand that assessment of this risk can be used to evaluate the quality of mesenchymal stromal cells. [Example]

[0053] Example 1: Obtaining and culturing multipotent mesenchymal stromal cells 1. Obtaining Multipotent Mesenchymal Stromal Cells (1) Collection of adipose tissue Adipose tissue was collected from donors (negative for AIDS virus, hepatitis B virus, hepatitis C virus, human T-cell virus, EB virus, cytomegalovirus, and Treponema pallidum) in a sterile environment. 50-150 mL of adipose tissue was placed in a sealed container to which 100 mL of tissue preservation solution (purchased from TIAN JIN HAO YANG BIOLOGICAL MANUFACTURE Co., Ltd.) had been added in advance, and the tissue was stored at 2-8°C. 30 mL of tissue preservation solution was pipetted and tested for bacterial, endotoxin, and mycoplasma contamination, and the tissue was then used for isolation of pluripotent mesenchymal stromal cells.

[0054] (2) Isolation of multipotent mesenchymal stromal cells An equal volume of Dulbecco's phosphate-buffered saline (dPBS) was added to the adipose tissue, the container containing the tissue was sealed, vigorously shaken for 20 seconds, and left to stand for 5 minutes. After the adipose tissue and dPBS were completely separated, the underlying liquid was discarded and the adipose tissue was repeatedly washed with dPBS until the underlying liquid was no longer red. A 20 mL aliquot of washed adipose tissue was placed in a 50 mL centrifuge tube, and an equal volume of dPBS was added. The mixture was centrifuged at 400 g for 5 minutes. The solution was separated into an upper oily layer, a middle adipose tissue layer, and a lower layer of dPBS and blood cell sediment. The upper oily layer, lower dPBS and blood cell sediment were removed. Two volumes of 1 mg / mL type I collagenase (purchased from Gibco, catalog number: 17100-017) were added to the adipose tissue. The container containing the tissue was sealed and transferred to a thermostatic air shaker preheated to 37°C. The tissue was digested with collagenase at 120 rpm / min for 1 hour.

[0055] (3) Collection of multipotent mesenchymal stromal cells The digested tissue was centrifuged at 500 g for 8 minutes at room temperature. After centrifugation, the tissue was separated into an upper oily layer, a middle adipose tissue layer, a lower digestion solution, and a bottom cell pellet. The upper oily layer, middle adipose tissue layer, and lower digestion solution were discarded. The bottom cell pellet was resuspended in dPBS, filtered through a 100 μm filter, and centrifuged at 500 g for 5 minutes in a 50 mL centrifuge tube. The supernatant was removed, yielding a cell pellet containing primary human adipose-derived stromal cells (hADSCs). The complete medium was added to the centrifuge tube in the same volume as the adipose tissue, and mixed evenly to thoroughly release the digested cells, yielding a cell suspension containing primary human adipose-derived stromal cells.

[0056] 2. Culturing Multipotent Mesenchymal Stromal Cells Primary human adipose-derived stromal cells were cultured in different media M1 (αMEM + 10% FBS, αMEM purchased from Thermo Fisher, FBS purchased from ExCell Bio) or M2 (DMEM / F-12 + 5% Helios UltraGRO-Advanced, DMEM / F-12 purchased from Thermo Fisher, Helios UltraGRO-Advanced purchased from Helios BioScience), and the specific steps were as follows:

[0057] (1) Primary culture The cell pellet was resuspended in M1 medium / M2 medium, and 1.5 mL of the cell suspension was inoculated into a T75 cell culture flask to which 8.5 mL of M1 medium / M2 medium had been added beforehand. After labeling the T75 cell culture flask, the flask was transferred to a cell culture incubator and cultured at 37°C with 5% CO2. After 24 hours, most of the primary human adipose-derived stromal cells had adhered to the cell wall. The supernatant was removed, and 10 mL of M1 medium / M2 medium was added. The medium was then changed every 3 days. Microscopic observation revealed that in addition to primary human adipose-derived stromal cells, the resulting primary cells contained many heterocytic cells and matrix components, and had the typical long spindle shape of human adipose-derived stromal cells.

[0058] (2) Subculture When primary human adipose-derived stromal cells reached 50%–70% confluence, the medium was removed and the cells were washed once with 10 mL of dPBS. TM 1.5 mL of 1x-Express (1x) (purchased from Gibco, catalog number: 12604-021) was added and digestion was carried out for 1-2 minutes. After some cells had rounded up and fallen off, the culture flask was gently tapped and 4.5 mL of dPBS was added to terminate the digestion. The liquid was collected in a 50 mL centrifuge tube. After washing once with 10 mL of dPBS, the tubes were centrifuged at 400 g for 5 minutes. The upper layer was a mixture of the digestion solution and dPBS, and the lower layer was a white precipitate, i.e., a precipitate containing primary human adipose-derived stromal cells. The supernatant was removed, and the white precipitates from the multiple centrifuge tubes were collected in one centrifuge tube and resuspended in M1 medium / M2 medium until the total volume reached 30 mL. The cell suspension was then mixed uniformly and used for cell counting. The counted cells were resuspended in M1 medium / M2 medium to obtain a cell density of 5,000–6,000 cells / cm. 2 The cells were passaged at a density of 1000 x g. The cell culture flask was labeled with information such as the batch of cells, passage number, and culture time, and placed in a cell culture incubator. When the cells reached approximately 90% confluence, they were passaged again. The pluripotent mesenchymal stromal cells at the third and fifth passages were collected and stored under refrigeration. They were designated D1M1-P3 (representing pluripotent mesenchymal stromal cells obtained by subculturing primary pluripotent mesenchymal stromal cells from donor 1 in M1 medium up to passage 3), D1M2-P3 (representing pluripotent mesenchymal stromal cells obtained by subculturing primary pluripotent mesenchymal stromal cells from donor 1 in M2 medium up to passage 3), D1M1-P5 (representing pluripotent mesenchymal stromal cells obtained by subculturing primary pluripotent mesenchymal stromal cells from donor 1 in M1 medium up to passage 5), and D1M2-P5 (representing pluripotent mesenchymal stromal cells obtained by subculturing primary pluripotent mesenchymal stromal cells from donor 1 in M2 medium up to passage 5), respectively.

[0059] Example 2 Quality control of pluripotent mesenchymal stromal cells In this example, refrigerated samples were rapidly thawed in a 37°C water bath with continuous stirring, and the multipotent mesenchymal stromal cells were resuspended in pre-warmed dPBS, centrifuged at 400g for 5 minutes, and washed twice with dPBS. Finally, the multipotent mesenchymal stromal cells (D1M1-P3, D1M2-P3, D1M1-P5, D1M2-P5) were counted and used for quality control. 1. Microbiological safety testing (1) Sterility test According to the Chinese Pharmacopoeia 2020 Edition (Part 4) General Provision 1101 Sterility Test Method, the steps are briefly described as follows: After wetting the filter membrane of a disposable triple bacterial collector (purchased from Zhejiang Tailin Bioengineering Co., Ltd.) with 100 mL of 0.9% saline (purchased from SJZ No. 4 Pharmaceutical), each hADSC sample was introduced into the bacterial collector and filtered. After filtration, the filter membrane was washed twice with 300 mL of 0.9% saline. Two different media were used: liquid thioglycollate medium for anaerobic and aerobic bacterial testing, and trypticase soy broth (TSB), a soybean-casein-digest medium, for anaerobic and aerobic bacterial testing. Two incubators of the triple bacterial collector containing the sample were filled with 100 mL of liquid thioglycollate medium, and the other incubator was filled with 100 mL of TSB. 1 mL of 0.9% saline in place of the sample served as a negative control, and Staphylococcus aureus (bacterial loading less than 100 CFU) served as a positive control. After inoculation, these media were gently shaken and incubated under the conditions recommended by the sterility test: the liquid thioglycollate medium was kept at 30-35°C, and the remaining medium was kept at 20-25°C for 14 days. After a total of 14 days of incubation, bacterial growth was observed and recorded every working day during the incubation period.

[0060] (2) Mycoplasma test According to the Chinese Pharmacopoeia 2020 Edition (Part 4) General Provision 3301 Mycoplasma Testing Method, the steps are briefly described as follows: Mycoplasma broth medium, arginine-containing mycoplasma broth medium, mycoplasma semi-solid medium, and arginine-containing mycoplasma semi-solid medium were prepared according to the conventional recipe and sterilized. Next, 800,000 units of penicillin sodium for injection (purchased from Jiangxi Dongfeng Pharmaceutical Co., Ltd.) were reconstituted with 1 mL of 0.9% saline to prepare the medium. For every 800 mL of sterilized medium, 200 mL of fetal bovine serum and 800,000 units of penicillin sodium for injection were added, mixed uniformly, and stored at 2-8°C. Four mycoplasma broth media (10 mL / bottle), four arginine-containing mycoplasma broth media (10 mL / bottle), two mycoplasma semi-solid media (10 mL / bottle), and two arginine-containing mycoplasma semi-solid media (10 mL / bottle) were inoculated with 1.0 mL of cell sample, cultured at 36 ± 1°C for 21 days, and observed every 3 days. On the seventh day after inoculation, two tubes of mycoplasma broth medium inoculated with the cell sample and two tubes of arginine-containing mycoplasma broth medium inoculated with the cell sample were subcultured. Each tube of mycoplasma broth medium was subcultured into two arginine-containing semi-solid mycoplasma broth medium and two arginine-containing mycoplasma broth medium, respectively, with an inoculum volume of 1 mL. Each tube of arginine-containing mycoplasma broth medium was subcultured into two arginine-containing semi-solid mycoplasma broth medium and two arginine-containing mycoplasma broth medium, respectively. The tubes were cultured at 36°C ± 1°C for 21 days and observed every 3 to 5 days.

[0061] (3) Endotoxin test According to the Chinese Pharmacopoeia 2020 Edition (Part 4) General Provision 1143 Bacterial Endotoxin Testing Method, the steps are briefly described as follows: The secondary standard (purchased from Zhanjiang A&C Biological Ltd.) was reconstituted in 1 mL of endotoxin test water (purchased from Zhanjiang A&C Biological Ltd.) and mixed uniformly on a vortex shaker for 15 minutes, followed by gradient dilutions. For each dilution, the solution was mixed on a vortex shaker for 30 seconds, and finally diluted to 4λ and 2λ endotoxin standard solutions. The cell suspension was diluted with endotoxin test water and mixed for 30 seconds on a vortex shaker at each dilution. The diluted solution was used as the test sample. The dilution rate was equal to or less than the maximum valid dilution factor (MVD), which was calculated using the formula MVD = C × L / λ, where L is the endotoxin limit of the sample, C is the concentration of the test sample, and λ is the sensitivity of the Limulus reagent. One test sample was added to the 4λ endotoxin standard solution at a 1:1 volume ratio to serve as an endotoxin positive control. Eight Limulus reagents (purchased from Zhanjiang A&C Biological Ltd.) were reconstituted with 0.1 mL of endotoxin test water. Two Limulus reagents were mixed with 0.1 mL of endotoxin positive control solution to serve as parallel cell positive controls (PPC). Two Limulus reagents were mixed with 0.1 mL of 2λ endotoxin standard solution to serve as parallel positive controls (PC). Two Limulus reagents were mixed with 0.1 mL of endotoxin test water to serve as parallel negative controls (NC). Two Limulus reagents were mixed with 0.1 mL of cell solution at a dilution rate below the MVD to serve as parallel cell tests. The reaction tube was placed in a preheated endotoxin gel tester and a 60-minute countdown was started. One minute before the end of the 60-minute period, the reaction tube was removed and the results were observed and recorded. As shown in Table 1, after 14 days, there was no bacterial growth in the collectors containing the mesenchymal stromal cell samples, and the sterility test results for D1M1-P3, D1M2-P3, D1M1-P5, and D1M2-P5 were passed. The mycoplasma test results for D1M1-P3, D1M2-P3, D1M1-P5, and D1M2-P5 were negative, and the test results were passed. The endotoxin test results for D1M1-P3, D1M2-P3, D1M1-P5, and D1M2-P5 were negative, and the test results were passed.

[0062] [Table 1]

[0063] 2. Phenotype analysis The expression of surface markers specific to multipotent mesenchymal stromal cells (CD73, CD90, CD105, CD11b, CD19, CD34, CD45, and HLA-DR) in hADSC samples was analyzed by flow cytometry (reference: M. Dominici et al., Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy (2006) Vol. 8, No. 4, 315-317). The steps were as follows:

[0064] hADSC samples at passage 3 or 5 were TM Digest with -Express (1x) for 2-3 minutes at 37°C. When the cells become round and fall off, TM The digestion was terminated by adding PBS (1x) in a volume three times greater than the volume of 1x PBS. The cell suspension was then pipetted into a 50 mL centrifuge tube and centrifuged at 300 g for 5 minutes. The cells were washed twice with PBS (1x) until the viable cell density reached (0.5–1) x 10. 7 The cells were resuspended to a concentration of 100 cells / mL.

[0065] 100 μL of the cell suspension was pipetted into a flow tube and incubated with 5 μL of pre-labeled antibodies (FITC-labeled anti-human CD34, FITC-labeled anti-human CD45, FITC-labeled anti-human CD11b, FITC-labeled anti-human HLA-DR, FITC-labeled anti-human CD73, FITC-labeled anti-human CD90, APC-labeled anti-human CD19, and PE-labeled anti-CD105) for 15 minutes at room temperature in the dark. As controls, FITC-labeled mouse IgG1, APC-labeled mouse IgG1, and PE-labeled mouse IgG1 were added. The antibodies used were purchased from Biolegend.

[0066] Two milliliters of sheath fluid was added to each flow tube, vortexed, and centrifuged at 300 g for 5 minutes. The supernatant was removed. Cells were resuspended in 300 μL of 1% paraformaldehyde in PBS (1×) and analyzed by flow cytometry.

[0067] As a result, as shown in Table 2, the expression rates of the positive markers CD73, CD90, and CD105 on the surface of D1M1-P3, D1M2-P3, D1M1-P5, and D1M2-P5 were over 95%, and the expression rates of the negative markers CD11b, CD19, CD34, CD45, and HLA-DR on the surface of D1M1-P3, D1M2-P3, D1M1-P5, and D1M2-P5 were less than 2%.

[0068] [Table 2]

[0069] 3. Cellular activity test (1) Cell viability analysis The cell suspension was diluted with 0.9% saline and mixed with 0.4% trypan blue staining solution at a volume ratio of 9:1 until homogeneous. 10 μL of the mixture was pipetted into the counting chamber of the counting plate and observed under a 10x objective lens. The total number of live cells and the total number of dead cells in each of the four grids were recorded, and the cell viability was calculated using the following formula: Cell viability (%) = total number of live cells / (total number of live cells + total number of dead cells) x 100% The cell viabilities of D1M1-P3, D1M2-P3, D1M1-P5, and D1M2-P5 were all over 80%.

[0070] (2) Cell proliferation kinetics measurement When the hADSC samples at passage 5 reached 80%–90% confluence, TM The cells were digested with 1x PBS-Express (1x). The cell density was adjusted to 3.2 x 10 5 pieces / mL, 1.6×10 5 pieces / mL, 0.8×10 5 pieces / mL, 0.4×10 5 pieces / mL, 0.2×10 5 pieces / mL, 0.1×10 5 The cells were adjusted to cells / mL and inoculated into a 96-well microplate at 100 μL / well. 100 μL of complete medium was added to the control wells, and six duplicate wells were prepared for each group. After culturing for 4 hours at 37°C and 5% CO2, the medium in each well was discarded, and 110 μL / well of CCK8 solution (DMEM / F12 medium (phenol red-free):CCK8 (v / v) = 100:10) was added to each well, followed by incubation for 2 hours at 37°C and 5% CO2. The optical density (OD450) at a wavelength of 450 nm was measured using a multi-function microplate reader. The average OD450 of the control wells was normalized to obtain the delta OD450 values ​​of wells with different cell densities. Linear regression curve fitting was performed using delta OD450 as the horizontal axis and cell number as the vertical axis. In a parallel experiment, a passage 5 hADSC sample was cultured at 1 × 10 4The cells were plated at a density of 100 cells / well into a 96-well microplate. 100 μL of complete medium was added to the control wells. Eight plates were prepared, with six replicate wells for each group. The cells were counted every day until day 8. The medium in each well was discarded, and 110 μL of CCK8 solution (DMEM / F12 medium (phenol red-free):CCK8 (v / v) = 100:10) was added to each well, followed by incubation at 37°C and 5% CO for 2 hours. OD450 was measured using a multi-function microplate reader. The average OD450 of the control wells was normalized to obtain the delta OD450 value of the well. The cell number in each well was calculated based on the curve by linear regression. A growth curve was created using the average values, and the population doubling time was calculated from the growth curve. Figures 1A and 1B show the growth curves of D1M1-P5 and D1M2-P5, respectively. hADSCs entered the logarithmic growth phase after 3 days of culture, the quiescent phase after 6 days, and the proliferation ability of the cells began to decline after 7 days. The population doubling time of D1M1-P5 was 37.5 hours, and that of D1M2-P5 was 21.9 hours.

[0071] (3) Cell cycle analysis hADSC samples at passage 3 or 5 were TM The cell pellets were digested with PBS (1x) at 37°C for 2-3 minutes, centrifuged at 1000 rpm for 3-5 minutes, and the supernatant was carefully discarded. The cell pellets were washed twice with 1 mL of pre-chilled PBS (1x) to obtain a cell pellet with a density of 1x10. 6 The cells were resuspended to a density of 100 cells / mL. 1 mL of the cell suspension was added dropwise to 4 mL of pre-chilled 95% ethanol solution while vortexing at low speed (operated on ice). After uniformly mixing the cells, they were fixed at 4°C for at least 2 hours. Next, the cells were centrifuged at 1000 rpm for 3-5 minutes to pellet them. The cells were washed twice with 5 mL of pre-chilled PBS (1x), and the bottom of the centrifuge tube was gently tapped to disperse the cells and avoid clumping. Using Table 3, a cell cycle and apoptosis detection kit (purchased from Beijing 4A Biotech Co., Ltd.) was used to prepare propidium iodide solution according to the number of samples to be tested. Next, 0.4 mL of propidium iodide solution was added to the cell sample, and the cell pellet was gently resuspended. The cells were then incubated in the dark at 37°C for 30 minutes. After washing twice with PBS (1x), the cells were resuspended in PBS (1x). Cell cycle analysis using flow cytometry was completed within 24 hours.

[0072] [Table 3]

[0073] As shown in Figure 1C and Figure 1D, the percentages of D1M1-P5 in the G1, S, and G2 phases were 85.69%, 12.56%, and 1.75%, respectively, and the percentages of D1M2-P5 in the G1, S, and G2 phases were 89.07%, 6.42%, and 4.51%, respectively.

[0074] (4) Apoptosis test Tryptophan 5 hADSC samples TM The cells were digested with 1x Binding Buffer (purchased from Beijing 4A Biotech Co., Ltd.) for 2-3 minutes at 37°C and centrifuged at 1000 rpm for 3-5 minutes. The supernatant was carefully discarded. The cell pellets were resuspended in 0.8 mL of 1x Binding Buffer (purchased from Beijing 4A Biotech Co., Ltd.). Density is (2~5)×10 5 200 μL of hADSC sample (100 μg / mL) was added to each flow tube and incubated with 5 μL of Annexin-V-FITC for 10 min in the dark. After centrifugation, cells were resuspended in 200 μL of binding buffer and then incubated with 5 μL of propidium iodide before flow cytometry analysis. As a result, as shown in Figures 1E and 1F, D1M1-P5 had a cell viability of 92.0% and an apoptosis rate of 5.75%, while D1M2-P5 had a cell viability of 91.4% and an apoptosis rate of 0.52%.

[0075] 4. Biological activity analysis (1) Adipogenic differentiation Before the experiment, solutions A and B were prepared according to the instructions for the OriCell kit (purchased from Cyagen Biosciences Inc., catalog number: HUXMD-90031) used for adipogenic differentiation of human adipose-derived mesenchymal stromal cells, and then the following steps were performed. 2 x 10 cells 4 cell / cm 2 The cells were seeded into a 6-well plate at a density of 100%, 2 mL of complete medium was added to each well, and the cells were cultured at 37°C and 5% CO2 until they reached 100% confluence. The supernatant was discarded. The cells were incubated in 2 mL of solution A for 3 days, then replaced with 2 mL of solution B and incubated for 24 hours. This was repeated three times, and the cells were subsequently cultured in solution B for 4 to 7 days until the lipid droplets became sufficiently large and round. The cells were washed, fixed with 4% paraformaldehyde solution, and stained with 0.5% Oil Red O for 20 min at room temperature. After washing three times with PBS (1×), images were taken with an inverted phase-contrast microscope.

[0076] (2) Osteogenic differentiation Before the experiment, osteogenic medium was prepared according to the instructions for the OriCell kit (purchased from Cyagen Biosciences Inc., catalog number: HUXMD-90021) used for osteogenic differentiation of human adipose-derived mesenchymal stromal cells, and then the following steps were performed. 2 x 10 cells 4 cell / cm 2 The cells were seeded into a 6-well plate at a density of 1000 × 1000, 2 mL of complete medium was added to each well, and the cells were cultured at 37 °C and 5% CO until the cells reached 80–90% confluence. The culture supernatant was discarded, and the cells were incubated in 2 mL of osteogenic medium, which was replaced once every 3 days, for 2 to 4 weeks, at which time significant calcium deposition was observed under an inverted microscope. The cells were washed, fixed with 4% paraformaldehyde solution, and stained with Alizarin Red S for 5 min at room temperature. After washing three times with PBS (1×), images were taken with an inverted phase-contrast microscope.

[0077] (3) Chondrogenic differentiation Before the experiment, chondrogenic medium was prepared according to the instructions for the OriCell kit (purchased from Cyagen Biosciences Inc., catalog number: HUXMD-90041) used for chondrogenic differentiation of human adipose-derived mesenchymal stromal cells, and then the following steps were performed. 0.1% gelatin was added to a 6-well plate, gently shaken to cover the bottom of the wells, allowed to stand for 30 minutes, the gelatin was discarded, and the plate was allowed to dry. 1 x 10 cells of passage 5 4 cell / cm 2 The cells were seeded at a density of 1000 μg / well onto a 6-well plate coated with 0.1% gelatin, 2 mL of complete medium was added to each well, and the cells were cultured at 37°C and 5% CO2 until the cells reached 80-90% confluence. The supernatant was discarded. Cells were induced in 2 mL of fresh chondrogenic medium (containing 20 μL of TGF-β3) and the chondrogenic medium was replaced every 2–3 days for 2 weeks. Control wells were continuously cultured in complete medium. The cells were washed, fixed with 4% paraformaldehyde solution, and stained with Alcian blue for 30 minutes at room temperature. After washing three times with PBS (1x), images were taken using an inverted phase-contrast microscope. As shown in Figure 1G and Figure 1H, the differentiation status of multipotent mesenchymal stromal cells in vitro was demonstrated. The multipotent mesenchymal stromal cells were induced to undergo adipogenic, osteogenic, and chondrogenic differentiation. As a result, D1M1-P5 and D1M2-P5 were successfully induced into adipogenic cells, osteoblasts, and chondroblasts, respectively. These results demonstrate that both passage 3 and passage 5 mesenchymal stromal cells cultured in different media are multipotent mesenchymal stromal cells and meet the recognized quality control standards for mesenchymal stromal cells.

[0078] Example 3 Animals treated with pluripotent mesenchymal stromal cells 1. Intravital Injection of Multipotent Mesenchymal Stromal Cells Six to eight-week-old male NCG mice (purchased from Gempharmatech Co., Ltd.) were randomly assigned to each group. After the mice were immobilized, the injection site was disinfected. The fifth-passage hADSC samples (D1M1-P5 or D1M2-P5) that had passed quality control were resuspended in 0.9% saline and injected at a concentration of 1 × 10 6 The cells were slowly injected into each mouse via the tail vein at a dose of 100 cells / mouse. The control group was injected with 0.9% saline. The survival rate of mice was recorded within 3 minutes after injection. All 6 mice injected with D1M1-P5 died within 3 minutes, while all 6 mice injected with D1M2-P5 and 6 mice injected with 0.9% saline remained alive within 3 minutes. After completion of the observation period, the mice were anesthetized with Avertin and euthanized by cutting the abdominal aorta.

[0079] 2. Immunohistochemistry (1) Gathering materials The mouse's skin and muscles were incised, the thoracic cavity was exposed, and the right ventricle was punctured with a syringe. 5 mL of 0.9% saline was slowly perfused throughout the body until the outflowing fluid was relatively clear and not obviously bloody. The mouse's lungs were then promptly harvested. Visual pathological observation was performed, and the tissues were fixed in 10% formalin solution for 2 days.

[0080] (2) Hematoxylin-eosin staining According to standard laboratory procedures, the lung tissues were dehydrated in gradient ethanol, embedded in paraffin, sectioned, stained with hematoxylin and eosin (HE), and finally observed under a light microscope. Figure 2A shows the pathological results of lung tissue after injection of D1M1-P5, D1M2-P5, or saline into mice. Compared to the control group, mice injected with D1M1-P5 exhibited pulmonary congestion and severe pulmonary embolism, whereas mice injected with D1M2-P5 did not exhibit any of these adverse reactions. Figure 2B also shows that mice injected with D1M1-P5 had a large number of venous thrombi in the lungs, with the density of embolism significantly higher than those in the D1M2-P5 and control groups. Furthermore, fluorescent PKH26-labeled D1M1-P5 or D1M2-P5 was injected into a mouse model, and the number of PKH26-positive cells in the lungs was counted. As a result, as shown in Figures 2C and 2D, consistent with the immunohistochemical results, a large amount of PKH26 was detected in the lungs of mice with thrombus formation. + D1M1-P5 were observed, indicating that mesenchymal stromal cells grown under different culture conditions underwent different biological processes or formed different lineages.

[0081] Example 4 Single-cell RNA sequencing To identify the heterogeneity of mesenchymal stromal cells in different culture media, we performed single-cell RNA sequencing to examine the gene expression profiles of mesenchymal stromal cells at the single-cell level. The steps were as follows: 1. Preparation of Single-cell Suspension The fifth-passage hADSC samples (D1M1-P5 and D1M2-P5) were diluted with sample buffer to a cell suspension with a concentration of <1000 cells / μL. 1 μL of Calcein AM dye and 1 μL of Draq7 dye were added to 200 μL of the cell suspension for cell staining. Filter the stained cell suspension through a 40 µm filter and transfer to BD Rhapsody TMThe cells were placed in a scanner to measure cell density and cell viability. Cell suspensions were prepared by dilution based on the stock cell and buffer volumes obtained from the scanner's sample calculator function.

[0082] 2. Single Cell Sorting The diluted cell suspension was loaded onto a Cartridge workflow equipped with 200,000 microwells (Cartridge Kit, purchased from BD Biosciences, catalog number: 633733), and the cell loading and doublet rate were analyzed to evaluate the effectiveness of single cell isolation. After washing to remove unloaded cells, BD Rhapsody beads were loaded into the Cartridge workflow, and bead and cell loading and doublet rates were analyzed to assess the number of beads bound within single-cell wells. After washing away excess beads, the cell lysis solution was added to the Cartridge workflow to lyse the cells. The mRNA contained in each cell was captured by the probe via the polyA / polyT on the surface of the BD Rhapsody beads, which had the same cell label (CL) and various molecular barcodes (UMI). The BD Rhapsody beads were then collected from the Cartridge workflow into a centrifuge tube.

[0083] 3. Single-cell cDNA Synthesis and Library Construction First-strand cDNA from single cells was reverse synthesized, and libraries were constructed using a Cartridge kit (purchased from BD Bioscience, catalog number: 633731) and a Whole Transcriptome Analysis (WTA) Amplification Kit (purchased from BD Bioscience, catalog number: 633801). The following procedures were performed according to the kit's instructions, which are briefly described below. The collected beads were washed, and the reverse transcription reagent (Table 4) was added and mixed with the beads, followed by incubation at 37°C for 45 minutes.

[0084] [Table 4]

[0085] Exonuclease was added, and the mixture was incubated at 37°C for 30 minutes and then at 80°C for 20 minutes to remove probes that were not connected to mRNA on the bead surface. Random primer mix (Table 5) was added and incubated at 95°C for 5 minutes, 37°C at 1200 rpm for 5 minutes, and 25°C at 1200 rpm for 15 minutes; primer extension mix (Table 6) was added and incubated at 25°C at 1200 rpm for 10 minutes, 37°C at 1200 rpm for 15 minutes, 45°C at 1200 rpm for 10 minutes, and 55°C at 1200 rpm for 10 minutes, and the extended first-strand cDNA was eluted with bead-free eluent.

[0086] [Table 5] [Table 6]

[0087] The random primer extension (RPE) products were added to a PCR amplification mixture (Table 7) containing universal and specific primers, and amplification was carried out using the process in Table 8, followed by enrichment and purification of the amplified products.

[0088] [Table 7] [Table 8]

[0089] The amplified product was used as a template for PCR using the Whole Transcriptome Index PCR amplification mixture (Table 9), and amplification was performed according to the process in Table 10 (if the molar concentration of the amplified product is 1-2 nM, amplification was performed for 9 cycles; if the molar concentration of the amplified product is >2 nM, amplification was performed for 8 cycles). The new amplified product was then enriched and purified to obtain a single-cell sequencing library.

[0090] [Table 9] [Table 10]

[0091] 4. Single-cell Sequencing Library Quality The concentration of the single-cell sequencing library was measured using a Qubit instrument, and the fragment length of the single-cell sequencing library was measured using an Agilent 2100 bioanalyzer. The library concentration was 0.1–100 ng / μL, and the fragment length of the library was 460–550 bp.

[0092] 5. Single-cell Sequencing The molar concentration of the single-cell sequencing library, calculated based on the library concentration and fragment length, was 1 to 100 nM. After diluting it to a standard molar concentration of 0.2 to 2 nM, the sequencing control library PhiX, which has the same molar concentration as the single-cell sequencing library, was mixed in a ratio of single-cell sequencing library:sequencing control library = 1:(0.05 to 0.5) and sequencing was performed.

[0093] 6. Sequencing Data Quality Sequencing data were analyzed by BD cwl-runner 3.1 to assess the quality of the raw sequencing data. 5095 D1M1-P5 and 3249 D1M2-P5 were sequenced, with an average sequencing depth of 50K / cell.

[0094] Example 5 Clustering of Subpopulations The raw sequencing data was converted to FASTQ format and the quality of the sequencing data was analyzed using the BD Rhapsody analysis pipeline v1.9.1 (BD Biosciences) for cell barcode identification, read alignment, and UMI quantification using default parameters. 1. Data Preprocessing Quality control: Sequences with read1 length < 60, read2 length < 42, read1 and read2 base quality < 20, and read1 single nucleotide frequency (SNF) ≥ 0.55 or read2 SNF ≥ 0.80 were filtered out. Alignment and annotation: Quality-controlled valid reads were aligned to the human reference genome GRCh38, and the alignment results were annotated. Gene expression matrix: The expression read counts for each gene across all samples were merged (collapsed) and adjusted to molecular barcode (UMI) counts using recursive permutation error correction (RSEC). Second-derivative analysis of all RSEC-adjusted UMI counts distinguished putative cells from background noise. The resulting output was a gene expression matrix with gene identities as columns and cell indices as rows.

[0095] 2. Cell Filtering The RSEC-adjusted UMI count matrix was imported into R 4.1.0, and gene expression data were analyzed using Seurat software 4.0.3. After singlets were identified, abnormal cells were excluded from downstream analysis using the median absolute deviation (MAD) method. Cells with a median mitochondrial read percentage greater than 3 MAD, a median number of expressed genes less than 3 MAD, or a median UMI count less than 3 MAD were considered abnormal. To avoid the influence of confounding factors such as cell cycle stage, sequencing depth, and percent mitochondrial content, these confounding factors were removed from the analysis by Seurat.

[0096] 3. Dimensionality Reduction To obtain a two-dimensional projection of population dynamics, we processed the top 2000 highly variable genes in the normalized gene barcode matrix using Seurat's principal component analysis (PCA), and then performed dimensionality reduction on the matrix to obtain low-dimensional spatial information. We then performed uniform manifold approximation and projection (UMAP) to process the top 30 principal components (PCs), achieving cell visualization in two-dimensional space. The steps included the following: Normalize the data using the NormalizeData function (normalization.method = "LogNormalize") Select the top 2000 genes, ranked based on variance, as highly variable genes (HVG) using the FindVariableFeature function (selection.method="vst", nfeatures=2000); The ScaleData function was used to normalize 2000 highly variable genes and remove noise due to factors such as the cell cycle. Reduce the dimension of the data using the RunPCA function (features=VariableFeatures(object=adsc)). Constructing a neighborhood graph using the shared nearest neighbor similarity algorithm (SNN) of the FindNeighbors function; Use the FindClusters function to adjust the parameters of the SNN model results (resolution = 0.1 ~ 1) and determine the number of cell subpopulations. Visualization and dimensionality reduction analysis can be performed using the RunUMAP function. Figure 3A shows the clustering results for D1M1-P5 and D1M2-P5 cell subpopulations, which included six distinct subpopulations (0–5). There were clear differences in the occupancy of each subpopulation between D1M1-P5 and D1M2-P5. Risk genes for mesenchymal stromal cells were identified using GO and KEGG. As shown in Figures 3B and 3C, there were also differences in the expression of risk genes among each subpopulation. This indicates that mesenchymal stromal cells exhibit heterogeneity in different culture media, resulting in distinct gene expression profiles for D1M1-P5 and D1M2-P5.

[0097] 4. Functional Clustering Analysis of Cell Subpopulations To convert the sparse gene expression matrix into a pathway score matrix, all genes in the gene expression matrix were scored based on classical pathways. Then, the pathway score matrix was subjected to dimensionality reduction and visualization to obtain functional clustering results of cell subpopulations. A schematic diagram of the mechanism is shown in Figure 4. The steps are briefly described as follows: Pathway enrichment analysis was performed on single-cell gene expression data, and the canonical pathway scores for each mesenchymal stromal cell were calculated using the ssGSEA, AUCell, and Seurat scoring functions to obtain a pathway score matrix. Dimensionality reduction and visualization were then performed based on the pathway score matrix as described above. As shown in Figures 5A, 5B, and 5C, the clustering results using the three different scoring functions were consistent. The D1M1-P5 (i.e., A2105C2P5) and D1M2-P5 (i.e., A2105C3P5) cell subpopulations were distinct and separated from each other. Both D1M1-P5 were quality-risk mesenchymal stromal cells, while both D1M2-P5 were non-quality-risk mesenchymal stromal cells. This indicates that mesenchymal stromal cells underwent significant functional changes after culturing in different media, consistent with the results of the animal experiments in Example 2. The functional clustering process developed here can provide a valuable tool for identifying specific functional subpopulations based on their transcriptome profiles. Using the cell culture method shown in Figure 6, multipotent mesenchymal stromal cells derived from donor 1 were subcultured from the first to the third passage using M1 or M2 medium, and then the medium was replaced and the cells were subcultured up to the fifth passage. The quality of the mesenchymal stromal cells at the third and fifth passages was assessed using functional clustering analysis. As shown in Figures 7A, 7B, and 7C, the mesenchymal stromal cells cultured in M1 medium and those cultured in M2 medium contained different cell subpopulations. All mesenchymal stromal cells cultured in M1 medium were quality-risk mesenchymal stromal cells, while all mesenchymal stromal cells cultured in M2 medium were non-quality-risk mesenchymal stromal cells. Heterogeneity occurs when mesenchymal stromal cells are grown under different culture conditions. According to the animal experiment results shown in Figures 8A and 8B, D1M1-P3 and D1M2 / M1-P5 induced pulmonary embolism in mice, but D1M2-P3 and D1M1 / M2-P5 did not, demonstrating the accuracy of the functional clustering results.

[0098] Example 6: Identification of subpopulations In this example, a mesenchymal stromal cell quality prediction model was constructed based on decision tree, random forest, or support vector machine (SVM). The datasets are listed in Table 11. A schematic diagram of the mechanism is shown in Figure 10. D1M1-P5 and D1M2-P5 represent pluripotent mesenchymal stromal cells obtained by subculturing primary pluripotent mesenchymal stromal cells derived from donor 1 and cultured in M1 or M2 medium up to passage 5, respectively. D1M1-P3 and D1M2-P3 represent pluripotent mesenchymal stromal cells obtained by subculturing primary pluripotent mesenchymal stromal cells derived from donor 1 and cultured in M1 or M2 medium up to passage 3, respectively. D2M1-P5 and D2M2-P5 represent pluripotent mesenchymal stromal cells obtained by subculturing primary pluripotent mesenchymal stromal cells derived from donor 2 and cultured in M1 or M2 medium up to passage 5, respectively. D2M3 / M2-P5 represent pluripotent mesenchymal stromal cells obtained by subculturing primary pluripotent mesenchymal stromal cells derived from donor 2 and cultured in M3 medium (αMEM + 5% Helios UltraGRO-Advanced) up to passage 3 and then in M2 medium up to passage 5.

[0099] [Table 11]

[0100] The steps were as follows: 1. Initial hyperparameters In the random forest model, the estimator (n_estimator) was 100, 200, 300, 400, 500, 600, 700, 800, 900, or 1000, and the maximum tree depth (max_depth) was 3, 5, or 7. In the support vector machine model, the regularization parameter C was 0.2, 0.6, 0.8, 1.0, 1.2, 1.6, 2.0, 2.2, 2.6, or 3.0, and the kernel parameter (kemel) was linear, "poly", "rbf", or "sigmoid".

[0101] 2. Feature Selection A machine learning method combining recursive feature reduction and cross-validation was used to rank each gene in the training set for its importance in distinguishing between quality risk and non-quality risk mesenchymal stromal cells. Starting from the most significant gene, one gene was added in order, and the accuracy rate of 10-fold cross-validation was calculated to determine the matching feature genes and the number of feature genes. As shown in Figure 11, when the most important gene was selected as a feature gene and the models with different normalization parameters C had 10-fold cross-validation accuracy rates of over 94% in the training set, the most important 13 genes (TAGLN, EFEMP1, TPM1, CLU, PTX3, IER3, IGFBP7, MFAP5, IL6, LUM, SERPINE2, CRIM1, and RHOB) were selected as feature genes, and when the models with different normalization parameters C had 10-fold cross-validation accuracy rates of 100% in the training set, the most important genes (TAGLN, EFEMP1, TPM1, CLU, PTX3, IER3, IGFBP7, MFAP5, IL6, LUM, SERPINE2, CRIM1, and RHOB) were selected as feature genes.

[0102] 3. Model derivation A linear SVM model was developed using 13 feature genes, and the model coefficient matrix (model weight matrix) was optimized by cross-validation to indicate the importance scores of the feature genes. The models were trained using test set 1, test set 2, test set 3, and test set 4, respectively, and the regularization parameter C was adjusted according to the prediction accuracy. Based on the test results shown in Table 12, the linear SVM model with the best representation in all test sets (C=0.0005) was selected as the mesenchymal stromal cell quality prediction model.

[0103] [Table 12]

[0104] As shown in Table 13, the established mesenchymal stromal cell quality prediction model (support vector machine model, model complexity parameter C = 0.0005) predicted different types of mesenchymal stromal cells with good measurement accuracy, precision, recall, and F1 score in the four test sets. The 13 established feature genes and their corresponding weighting coefficients are shown in Table 14.

[0105] [Table 13] [Table 14]

[0106] 4. Single-cell Level Mesenchymal Stromal Cell Quality Score Based on the expression levels of the identified 13 characteristic genes and the weight coefficients of each characteristic gene determined by the mesenchymal stromal cell quality prediction model, the quality score of mesenchymal stromal cells at the single-cell level was calculated using the following function to quantify the quality risk of a single mesenchymal stromal cell.

[0107]

number

[0108] where Gi is the expression level of the i-th feature gene in a single mesenchymal stromal cell, Wi is the weight coefficient of the i-th feature gene, and n is the quantity of feature genes. A positive value of Wi indicates that increased expression of the feature gene promotes the quality risk of mesenchymal stromal cells, and a negative value of Wi indicates that increased expression of the feature gene suppresses the quality risk of mesenchymal stromal cells.

[0109] The receiver operating characteristic (ROC) curve and area under the curve (AUC) were used to evaluate the quality scores and risk score thresholds of mesenchymal stromal cells in the four test sets.

[0110] Figure 12 shows the ROC curves and corresponding AUCs for Test Set 1, Test Set 2, Test Set 3, and Test Set 4. The numerical value at the highest point of the ROC curve (the value with the highest sensitivity and specificity) was used as the threshold for determining whether the mesenchymal stromal cells in that test set were quality risk mesenchymal stromal cells or non-quality risk mesenchymal stromal cells. Based on the ROC curves, the quality score thresholds for the four test sets were defined as 3.961 (AUC = 1), 3.961 (AUC = 1), 5.312 (AUC = 0.986), and 6.680 (AUC = 0.993), respectively. Specific results are shown in Figures 12A, 12B, 12C, and 12D.

[0111] Example 7 Mesenchymal stromal cell quality evaluation system In this embodiment, the mesenchymal stromal cell quality evaluation system includes an evaluation subsystem 10 . As shown in FIG. 13A, the evaluation subsystem 10 a quality scoring module 110 for calculating a quality score of the mesenchymal stromal cells based on the expression levels of the feature genes associated with the quality of the mesenchymal stromal cells and the weight coefficients of the feature genes; and a quality assessment module 120 for assessing the quality of the mesenchymal stromal cells based on the quality score of the mesenchymal stromal cells. As shown in FIG. 13B, the quality scoring module 110: a characteristic gene expression level obtaining unit 1110 for obtaining the expression level of a characteristic gene related to the quality of mesenchymal stromal cells; and a calculation unit 1120 for calculating a quality score of the mesenchymal stromal cells based on the expression levels of the feature genes and the weight coefficients of the feature genes. The calculation function for the quality score of the mesenchymal stromal cells is as follows:

[0112]

number

[0113] In the formula, Gi is the expression level of the i-th feature gene, Wi is the weight coefficient of the i-th feature gene, and n is the quantity of the feature gene. As shown in FIG. 13C, the quality assessment module 120 a mesenchymal stromal cell quality risk threshold determination unit 1210 for analyzing the quality score of mesenchymal stromal cells in a dataset using a receiver operating characteristic curve and an area under the curve, wherein the numerical value of the highest point of the receiver operating characteristic curve is the threshold of the mesenchymal stromal cell quality risk, and the dataset includes single-cell gene expression data of mesenchymal stromal cells having known specific quality attributes; a comparison and judgment unit 1220 for comparing the quality score of the mesenchymal stromal cells with a quality risk threshold; If the quality score of the mesenchymal stromal cells is greater than or equal to the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are quality risk mesenchymal stromal cells, and if the quality score of the mesenchymal stromal cells is less than the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are non-quality risk mesenchymal stromal cells. The evaluation subsystem 10 includes a result output module for outputting a report of the evaluation results of the mesenchymal stromal cell quality. The mesenchymal stromal cell quality evaluation system further includes an optimization subsystem 20 .

[0114] As shown in FIG. 13D, the optimization subsystem 20: a subpopulation clustering module 210 for obtaining single cell gene expression data and specific quality attributes of mesenchymal stromal cells; and a subpopulation identification module 220 for determining and / or optimizing mesenchymal stromal cell quality prediction models, feature genes associated with mesenchymal stromal cell quality, and feature gene weighting coefficients based on single-cell gene expression data and specific quality attributes of the mesenchymal stromal cells.

[0115] As shown in FIG. 13E, the subpopulation clustering module 210: a single-cell gene expression data acquisition unit 2110 for preprocessing the single-cell RNA sequencing data to obtain single-cell gene expression data of mesenchymal stromal cells; a pathway score matrix acquisition unit 2120 for performing pathway enrichment analysis on the single-cell gene expression data of the mesenchymal stromal cells and calculating the pathway score of each mesenchymal stromal cell; and a specific quality attribute determination unit 2130 for performing dimension reduction and clustering on the pathway score matrix to obtain clustering results as specific quality attributes of mesenchymal stromal cells.

[0116] As shown in FIG. 13F, the subpopulation identification module 220 a dataset creation unit 2210 for forming single-cell gene expression data of mesenchymal stromal cells with specific quality attribute labels into a dataset; a dataset division unit 2220 for dividing the dataset into a training set and a test set; a model training unit 2230 for training a supervised machine learning model using the training set, adjusting parameters of the supervised machine learning model using cross-validation and a test set, and determining and / or optimizing the mesenchymal stromal cell quality prediction model; and a feature gene output unit 2240 for outputting feature genes and weighting coefficients related to the mesenchymal stromal cell quality according to the mesenchymal stromal cell quality prediction model.

[0117] Example 8 Quality assessment of mesenchymal stromal cell samples using the system Based on the 13 identified feature genes and weighting coefficients (Table 14) determined by the mesenchymal stromal cell quality prediction model, the expression levels of the feature genes at the single-cell level for D1M1 / M2-P5 and D1M2 / M1-P5 were examined using subsystem 10 of the mesenchymal stromal cell quality evaluation system. To evaluate the quality of mesenchymal stromal cells, the quality scores of D1M1 / M2-P5 and D1M2 / M1-P5 were calculated using a formula. The quality score threshold was 3.961.

[0118] As a result, as shown in Figure 14, in D1M2 / M1-P5, 99.90% of mesenchymal stromal cells were quality risk mesenchymal stromal cells and 0.10% were non-quality risk mesenchymal stromal cells, whereas in D1M1 / M2-P5, 0.24% of mesenchymal stromal cells were quality risk mesenchymal stromal cells and 99.76% were non-quality risk mesenchymal stromal cells. This prediction result is consistent with the functional clustering results of cell subpopulations shown in Figures 7A, 7B, and 7C and the animal experiment results shown in Figures 8A and 8B. This indicates that the quality prediction model can accurately predict the quality risk of mesenchymal stromal cells.

[0119] Example 9 Analysis service using the system When a user uploads single-cell RNA sequencing data of mesenchymal stromal cells to the analysis website, the quality scoring module 110 of the evaluation subsystem 10 in the mesenchymal stromal cell quality evaluation system obtains the expression levels of feature genes related to the mesenchymal stromal cell quality of the mesenchymal stromal cells, and calculates the quality score of the mesenchymal stromal cells based on the expression levels of the feature genes and the weight coefficients of the feature genes.

[0120] The calculation function for the quality score of the mesenchymal stromal cells is as follows:

[0121]

number

[0122] The quality assessment module 120 of the assessment subsystem 10 compares the quality score of the mesenchymal stromal cells with a mesenchymal stromal cell quality risk threshold. If the quality score of the mesenchymal stromal cells is equal to or greater than the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are quality risk mesenchymal stromal cells; If the quality score of the mesenchymal stromal cells is less than the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are non-quality risk mesenchymal stromal cells. The result output module of the evaluation subsystem 10 outputs an evaluation report of the mesenchymal stromal cell quality.

[0123] When the administrator upgrades the analysis website, the subpopulation clustering module 210 of the optimization subsystem 20 in the mesenchymal stromal cell quality assessment system preprocesses the single-cell RNA sequencing data to obtain single-cell gene expression data, performs pathway enrichment analysis on the single-cell gene expression data to calculate pathway scores for each mesenchymal stromal cell, and obtains a pathway score matrix. Dimensionality reduction and clustering are performed on the pathway score matrix, and the clustering results are obtained as specific quality attributes of mesenchymal stromal cells.

[0124] The subpopulation identification module 220 integrates single-cell gene expression data of mesenchymal stromal cells with specific quality attributes into the original dataset to form a new dataset divided into a training set and a test set. The training set and the test set are used to optimize the mesenchymal stromal cell quality prediction model. The module outputs feature genes and weighting coefficients related to mesenchymal stromal cell quality. The quality scoring module 110 in the evaluation subsystem 10 obtains the expression levels of the feature genes associated with the quality of the mesenchymal stromal cells, and calculates the quality score of the mesenchymal stromal cells based on the expression levels of the feature genes and the weighting coefficients of the feature genes. The calculation function for the quality score of the mesenchymal stromal cells is as follows:

[0125]

number

[0126] The quality assessment module 120 of the assessment subsystem 10 compares the quality score of the mesenchymal stromal cells with a quality risk threshold, and the dataset includes single-cell gene expression data of mesenchymal stromal cells with known specific quality attributes. If the quality score of the mesenchymal stromal cells is equal to or greater than the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are quality risk mesenchymal stromal cells; Mesenchymal stromal cell quality score <間葉系間質細胞品質リスクの閾値であると、間葉系間質細胞が非品質リスク間葉系間質細胞である。The result output module of the evaluation subsystem 10 outputs an evaluation report of the mesenchymal stromal cell quality. Although the present invention has been described in the above examples as specific methods of the present invention, the present invention is not limited to these specific methods, i.e., it does not mean that the present invention must be carried out by relying on the above specific methods. Any improvements to the present invention, equivalent replacement of each raw material of the product of the present invention, addition of auxiliary ingredients, selection of specific methods, etc. are all within the protection scope and disclosure scope of the present invention.

Claims

1. a quality scoring module for calculating a quality score of the mesenchymal stromal cells based on the expression levels of the characteristic genes associated with the mesenchymal stromal cell quality and the weight coefficients of the characteristic genes; a quality evaluation module for evaluating the quality of mesenchymal stromal cells based on the quality score of the mesenchymal stromal cells, the quality scoring module: (1) a characteristic gene expression level acquisition unit for obtaining the expression level of a characteristic gene related to the quality of a mesenchymal stromal cell; (2) a calculation unit for calculating a quality score of the mesenchymal stromal cells according to the expression levels of the feature genes and the weight coefficients of the feature genes; The calculation function for the quality score of the mesenchymal stromal cells is as follows: [Equation 1] (where Gi is the expression level of the i-th feature gene, Wi is the weighting coefficient of the i-th feature gene, and n is the number of feature genes.) The quality assessment module: (1) a mesenchymal stromal cell quality risk threshold determination unit for analyzing the quality score of mesenchymal stromal cells in a dataset using a receiver operating characteristic curve and an area under the curve, wherein the numerical value at the highest point of the receiver operating characteristic curve is the mesenchymal stromal cell quality risk threshold, and the dataset includes single-cell gene expression data of mesenchymal stromal cells having known specific quality attributes; (2) a comparison and judgment unit for comparing the quality score of the mesenchymal stromal cells and the threshold value of the mesenchymal stromal cell quality risk; A mesenchymal stromal cell quality evaluation system, wherein if the quality score of the mesenchymal stromal cells is greater than or equal to the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are quality risk mesenchymal stromal cells, and if the quality score of the mesenchymal stromal cells is less than the mesenchymal stromal cell quality risk threshold, the mesenchymal stromal cells are non-quality risk mesenchymal stromal cells.

2. The mesenchymal stromal cell quality evaluation system according to claim 1 , further comprising a result output module for outputting a result report of mesenchymal stromal cell quality.

3. The mesenchymal stromal cell quality evaluation system according to claim 1, wherein the mesenchymal stromal cells comprise any one or a combination of at least two of adult mesenchymal stromal cells, embryonic mesenchymal stromal cells, induced pluripotent mesenchymal stromal cells, mesenchymal stromal cells transformed from mature somatic cells, and their derivative cells.

4. The mesenchymal stromal cell quality evaluation system according to claim 1 or 2, wherein the mesenchymal stromal cells comprise any one or a combination of at least two of mesenchymal stromal cells, mesenchymal stromal cells, multipotent stromal cells, multipotent mesenchymal stromal cells, and medicinal signaling cells.

5. 3. The mesenchymal stromal cell quality evaluation system according to claim 1 or 2, wherein the mesenchymal stromal cells comprise any one or a combination of at least two of adipose-derived mesenchymal stromal cells, umbilical cord mesenchymal stromal cells, placenta-derived mesenchymal stromal cells, bone marrow mesenchymal stromal cells, dental marrow mesenchymal stromal cells, menstrual blood-derived mesenchymal stromal cells, amniotic epithelial mesenchymal stromal cells, and bronchial basal cells.

6. a subpopulation clustering module for obtaining single-cell gene expression data and specific quality attributes of mesenchymal stromal cells; 3. The mesenchymal stromal cell quality evaluation system according to claim 1 or 2, further comprising an optimization subsystem including: a subpopulation identification module for determining and / or optimizing a mesenchymal stromal cell quality prediction model, feature genes associated with mesenchymal stromal cell quality, and weighting coefficients of the feature genes based on single-cell gene expression data and specific quality attributes of the mesenchymal stromal cells.

7. the subpopulation clustering module: (1) a single-cell gene expression data acquisition unit for preprocessing single-cell RNA sequencing data to obtain single-cell gene expression data of mesenchymal stromal cells; (2) a pathway score matrix acquisition unit for performing pathway enrichment analysis on single-cell gene expression data of mesenchymal stromal cells and calculating the pathway score of each mesenchymal stromal cell; (3) a specific quality attribute determination unit for performing dimensionality reduction and clustering on the pathway score matrix to obtain clustering results as specific quality attributes of mesenchymal stromal cells.

8. the subpopulation identification module: (1) a dataset creation unit for forming single-cell gene expression data of mesenchymal stromal cells having specific quality attributes into a dataset; (2) a dataset splitting unit for splitting the dataset into a training set and a test set; (3) a model training unit for training a supervised machine learning model using the training set, adjusting parameters of the supervised machine learning model using cross-validation and a test set, and determining and / or optimizing a mesenchymal stromal cell quality prediction model; (4) a feature gene output unit for outputting feature genes and weighting coefficients related to mesenchymal stromal cell quality according to the mesenchymal stromal cell quality prediction model.

9. the subpopulation identification module: (1) a dataset creation unit for forming single-cell gene expression data of mesenchymal stromal cells having specific quality attributes into a dataset; (2) a dataset splitting unit for splitting the dataset into a training set and a test set; (3) a model training unit for training a supervised machine learning model using the training set, adjusting parameters of the supervised machine learning model using cross-validation and a test set, and determining and / or optimizing a mesenchymal stromal cell quality prediction model; (4) a feature gene output unit for outputting feature genes and weighting coefficients related to mesenchymal stromal cell quality according to the mesenchymal stromal cell quality prediction model.

10. The mesenchymal stromal cell quality evaluation system according to claim 6, wherein the supervised machine learning model comprises any one of a perceptron model, a K-nearest neighbor model, a naive Bayes model, a decision tree model, a logistic regression, a support vector machine, a random forest, a boosting model, an EM algorithm, or a conditional random field.

11. The mesenchymal stromal cell quality evaluation system according to claim 7, wherein the supervised machine learning model comprises any one of a perceptron model, a K-nearest neighbor model, a naive Bayes model, a decision tree model, a logistic regression, a support vector machine, a random forest, a boosting model, an EM algorithm, or a conditional random field.

12. The mesenchymal stromal cell quality evaluation system according to claim 8, wherein the supervised machine learning model comprises any one of a perceptron model, a K-nearest neighbor model, a naive Bayes model, a decision tree model, a logistic regression, a support vector machine, a random forest, a boosting model, an EM algorithm, or a conditional random field.

Images