Method for inducing differentiation of human pluripotent stem cells into pancreatic progenitor cells

Abstract

Provided is a method for inducing differentiation of human pluripotent stem cells into pancreatic progenitor cells. Compared with conventional induced differentiation processes, the present method not only shortens a process that takes more than 30 days and requires 6-7 stages to only 19 days and 5 stages, but also significantly increases the proportion of β-like cells to 60%–70%, effectively improving hyperglycemic symptoms in diabetic mouse models. Furthermore, single-cell transcriptome detection of the grafts revealed that 60%-70% β cells could be detected, which were more mature than the cell state before transplantation. A method for assessing cell quality is also provided, which is used to evaluate the quality of in vitro induced cells. The method no longer relies on conventional detection based on the expression of a few specific genes and proteins, but uses single-cell omics technology to qualitatively and quantitatively evaluate the characteristics and differentiation efficiency of the induced cells based on a "three-module gene co-expression network," providing a new evaluation standard for the field.

Description

A method of inducing differentiation of human pluripotent stem cells into pancreatic precursor cells

[0001] Cross-reference to Related Applications

[0002] This application claims the benefit of Chinese Application No. 202410431275.4, filed April 10, 2024. The application number 202410431275.4 is hereby incorporated by reference in its entirety. TECHNICAL FIELD

[0003] The present application relates to the field of biotechnology, in particular, the present application relates to a method of inducing differentiation of human pluripotent stem cells into pancreatic precursor cells. BACKGROUND

[0004] Diabetes, a global chronic disease marked by persistent high blood glucose, is mainly caused by insufficient insulin secretion or resistance to insulin. A long-term high blood glucose environment can lead to damage and dysfunction of multiple tissues, including the heart, blood vessels, kidneys, eyes, and nerves. Current treatment strategies, such as exogenous insulin injection and islet cell transplantation, can alleviate the symptoms of diabetes, but have problems such as insufficient accuracy, risk of hypoglycemia, and shortage of donor cells. With the development of stem cell technology, especially the strategy of obtaining β-like cells and islet-like structures with insulin secretion ability from human pluripotent stem cells (such as human embryonic stem cells and human induced pluripotent stem cells) through in vitro directed differentiation technology, a new treatment possibility for diabetes is provided.

[0005] The process of mimicking in vivo pancreatic lineage development by adding small molecules and recombinant proteins to start from human pluripotent stem cells (including human embryonic stem cells and human induced pluripotent stem cells) is generally believed to be mainly through stepwise induction into definitive endoderm (DE), primitive gut tube (PGT), posterior foregut (PF), pancreatic endoderm (PE) or pancreatic progenitor (PP), and endocrine progenitor (EP), and finally obtain islet-like tissue containing β-like cells. Among them, the induction of high-quality pancreatic progenitor cells is crucial for the differentiation efficiency and function of subsequent endocrine cells. Due to the limited understanding of human pancreatic development, the clues for directed differentiation mainly come from animal models, especially mice. Although there are corresponding specific genes or proteins for the above stages to judge the differentiation efficiency, the differentiation path and cell state in the whole process lack comprehensive evaluation methods, and the similarities and differences with the real development process in vivo are still unknown. Therefore, it is extremely important to guide the differentiation of human pluripotent stem cells into high-quality pancreatic progenitor cells according to the differentiation path and regulatory network of human pancreatic lineage, as well as the conservation and specificity of pancreatic lineage development regulation between species.

[0006] The current existing scheme obtains induced cells expressing PP cell characteristic genes PDX1 and NKX6-1 to a certain extent by gradually using different small molecule compound combinations and processing sequences, and different culture times, and on this basis, a certain proportion of β-like cells with certain functions are obtained, but at the same time, a high proportion of multi-hormone cells (co-expressing insulin and glucagon) or α-like cells (secreting glucagon) are also produced. The main reason for the limited differentiation efficiency and purity of the current induction scheme is that the understanding of human pancreatic lineage differentiation process is not complete, and there is a lack of system evaluation of in vitro differentiation path and induced cell state based on in vivo differentiation process at single cell level, and the suboptimal early differentiation process and PP cell state may be one of the reasons for the subsequent differentiation problem. SUMMARY

[0007] In view of the problems in the prior art, the present application deeply analyzes the development process of mouse and human pancreas, and finds that the development path is conserved between species but has certain specificity in the regulation network. From the development path, the pancreatic precursor cell PP has two origins of the ventral and dorsal pancreatic endoderm (VPE and DPE), which respectively come from the DE cell through the AL cell (AL-P) of the pancreatic differentiation fate and the MG cell (MG-P) of the pancreatic differentiation fate. By comparing the PP cell development paths of humans and mice, it is found that in humans, the similarity of the ventral path (AL-P and VPE) to the PP cell is higher than that of the dorsal path (MG-P and DPE), which indicates a higher differentiation efficiency, while this phenomenon is not obvious in mice. Through gene co-expression network analysis, it is further confirmed that the fine differentiation process from DE to endocrine lineage includes three key gene modules, and the PP cell is at a key position connecting the first two modules.

[0008] The inventors then globally analyze the existing induction scheme process at the single cell level, identify the cell properties, differentiation path and regulation network of the existing scheme. By comparing with the in vivo cell differentiation process, the inventors find that there are mainly two points different from in vivo: one is that a stable gene co-expression network similar to in vivo development is not established. The differentiation process from DE to endocrine cell in vitro is divided into four gene modules, not three modules in vivo, and the early gene module is not stably established. The second is that the formation of PP cell in the existing scheme fails to simulate the efficient process of the ventral differentiation path in human body, resulting in insufficient similarity between the transition state of PP cell and the AL-P cell in human body.

[0009] Therefore, the present application provides an improved method: by prolonging the induction time of S2 stage, promoting the formation of early gene network, and imitating the natural conversion process of the ventral pancreatic endoderm to PP cell in human embryo development (DE-AL-P-VPE-PP), the generation of AL-P state cells is promoted by adding specific signal substances to activate the TGF-beta Nodal pathway (adding Activin A) and the WNT pathway (adding CHIR-99021). On this basis, the S3 and S4 induction periods of the control method can be degenerated, and the concentration of FGF10 can be reduced to improve the quality of PP cells. Without changing the S5 and S6 induction scheme of the control method, the proportion of beta-like cells can be significantly increased, and the proportion of multi-hormone or alpha-like cells can be reduced. Analysis of the islet-like cells transplanted into animals at different induction stages shows that the last induction stage (S7) of the control method can be omitted in the present process, so that the whole induction process is shortened to 19 days of 5 stages, and the islet-like cells after transplantation can effectively maintain blood glucose stability.

[0010] Specifically, the present application provides the following technical solutions:

[0011] In a first aspect, the present application provides a method for inducing differentiation of human pluripotent stem cells into pancreatic precursor cells, said method comprising the following steps:

[0012] S1: simultaneously activating TGF-beta Nodal pathway and WNT pathway;

[0013] S2: simultaneously activating FGF pathway, TGF-beta Nodal pathway, WNT pathway and adding final concentration of 0.25 mM vitamin C;

[0014] S3: simultaneously activating FGF pathway, inhibiting SHH pathway, activating Retinoic acid (RA) pathway, inhibiting BMP pathway, activating PKC and adding final concentration of 0.25 mM vitamin C;

[0015] S4: simultaneously inhibiting SHH pathway, activating Retinoic acid (RA) pathway, inhibiting BMP pathway, inhibiting TGF-beta RI kinase and adding final concentration of 1 µM thyroid hormone T3;

[0016] S5: simultaneously inhibiting BMP pathway, activating Notch pathway, inhibiting TGF-beta RI kinase and adding final concentration of 1 µM thyroid hormone T3.

[0017] In an embodiment, the steps of the method are as follows:

[0018] The human pluripotent stem cells are cultured in mTeSR1 pluripotent stem cell medium with final concentration of 10 µM Y-27632, and the induction of differentiation is initiated after 24 hours:

[0019] S1 for 2 days: final concentration of 100 ng / mL Activin A and CHIR-99021 is added in the culture medium, final concentration of 3 µM is used on the first day and final concentration of 0.3 µM is used on the second day;

[0020] S2 for 4 days: final concentration of 50 ng / mL FGF10, 10 ng / mL Activin A, 0.3 µM CHIR-99021 and 0.25 mM vitamin C is added in the culture medium;

[0021] S3 for 3 days: final concentration of 2 ng / mL FGF10, 0.25 µM SANT-1, 1 µM RA, 100 nM LDN-193189, 200 nM TPPB and 0.25 mM vitamin C is added in the culture medium;

[0022] S4 for 3 days: final concentration of 0.25 µM SANT-1, 0.05 µM RA, 100 nM LDN-193189, 10 µM ALK5i II, 1 µM thyroid hormone T3 is added in the culture medium;

[0023] S5 for 7 days: final concentration of 100 nM LDN-193189, 100 nM GSIXX, 10 mM ALK5i II, 1 mM thyroxine T3 was added to the medium;

[0024] wherein the basal medium used in S1 and S2 is MCDB 131 medium, with additional final concentration of 1.5 g / L NaHCO3, 1x GlutaMAX, 10 mM D-glucose, 0.5% BSA, 100 U / mL penicillin-streptomycin;

[0025] The basal medium used in S3 is MCDB 131 medium, with additional final concentration of 2.5 g / L NaHCO3, 1x GlutaMAX, 10 mM D-glucose, 2% BSA, 0.5x ITS-X, 100 U / mL penicillin-streptomycin;

[0026] The basal medium used in S4-S5 is MCDB 131 medium, with additional final concentration of 1.5 g / L NaHCO3, 1x GlutaMAX, 20 mM D-glucose, 2% BSA, 0.5x ITS-X, 10 mM ZnSO4, 10 pg / mL heparin, 100 U / mL penicillin-streptomycin.

[0027] In a second aspect, the present application provides the use of the above method in inducing pancreatic islet-like tissue.

[0028] In one embodiment, the use comprises the following steps:

[0029] S6 for 3-14 days: final concentration of 10 mM ALK5i II, 2 mM R428, 10 mM vitamin E analogue Trolox, 1 mM N-acetyl cysteine, 1 mM thyroxine T3 was added to the medium; The basal medium used in S6 is MCDB 131 medium, with additional final concentration of 1.5 g / L NaHCO3, 1x GlutaMAX, 20 mM D-glucose, 2% BSA, 0.5x ITS-X, 10 mM ZnSO4, 10 pg / mL heparin, 100 U / mL penicillin-streptomycin.

[0030] In a third aspect, the present application provides a method for evaluating in vitro cell differentiation process, comprising the following steps:

[0031] 1) First, in order to balance the number of cells in different cell populations on the path, we down-sample each population of cells, in principle to make the number of cells close and the network stable for subsequent analysis;

[0032] 2) For the down-sampled dataset, calculate the correlation coefficient between all transcription factor genes, taking transcription factor genes as nodes, and the correlation coefficient as the weight of the edge, to build a fully connected network;

[0033] 3) Cut off 90-95% of the low-weight edges, delete nodes with too low (<5-10%) or too high (>80-95%) expression ratio in the down-sampled dataset of cells, and perform network sparsification to retain the largest subnetwork after sparsification;

[0034] 4) Cluster the remaining network using algorithms such as Louvain algorithm, remove communities (Community) related to cell cycle and batch effect, and merge the remaining communities according to similarity to obtain multiple modules of the network;

[0035] 5) By analyzing whether the three-module co-expression network structure of in vitro differentiation process establishes the in vivo differentiation process, the differentiation efficiency and cell characteristics of in vitro differentiation process are evaluated.

[0036] Compared with the prior art, the present application has the following significant technical progress and beneficial effects:

[0037] 1) By extending the second stage (S2) induction time to 4-6 days, the method effectively promotes S2 cells to reach a state similar to AL-P cells in the human body and establishes a stable gene network. Tests have found that an induction time of 4 days is optimal for H1 embryonic stem cell lines (H1 ESC).

[0038] 2) Based on the extension of S2 induction time, the present application activates the TGF-beta Nodal pathway and WNT pathway by adding Activin A and CHIR-99021, respectively, enhancing the pancreatic development potential of S2 induced cells and improving their similarity to AL-P cells in the human body. For H1 ESC, the recommended final concentrations of Activin A and CHIR-99021 are 10 ng / mL and 0.3 μM, respectively.

[0039] 3) The present application replaces the 5-day induction period of S3 and S4 in the control method with an adjusted S3 induction scheme, and reduces the FGF10 concentration from 50 ng / mL in the control method to 2 ng / mL. This improvement not only shortens the induction time and reduces costs, but also promotes the production of PP cells that are more similar to in vivo, establishes a gene regulation network closer to in vivo PP cells and eliminates their potential to differentiate into other endodermal cell lineages.

[0040] 4) The induction process from S1 to S3 not only optimizes the generation of pancreatic precursor cells (PP), but also directly affects the differentiation path and efficiency of subsequent EP cells, resulting in a higher proportion of beta-like cells and their precursors and a significant reduction in the production of multi-hormone or alpha-like cells compared to the control method.

[0041] 5) The islet-like tissue formed by the PP cells induced by the present application shows excellent therapeutic effect in a diabetic mouse model, significantly improving the symptoms of hyperglycemia.

[0042] 6) Through post-transplantation detection of islet-like cells transplanted at different induction stages, the present application can reduce the entire induction process to 5 stages in 19 days, and the islet-like cells can still effectively regulate blood glucose levels after transplantation, indicating that the last stage of the control induction method can be omitted in the present process.

[0043] 7) Through single-cell transcriptome detection of the graft, it is found that the present application can increase the proportion of beta cells in the graft to 60%-70%, which is much higher than the proportion of beta cells in the graft of the control method, and the cell function is further matured after transplantation.

[0044] 8) The present application deeply analyzes the path and regulation mechanism of human pancreatic differentiation, especially by establishing a stable three-module gene co-expression network, providing a new standard for evaluating the effect of in vitro induction, thereby improving the accuracy and reliability of the method. BRIEF DESCRIPTION OF DRAWINGS

[0045] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, together with the embodiments of the present application, to explain the present application, and do not constitute a limitation on the present application. In the drawings:

[0046] Figure 1 is a schematic diagram of the control method and the process of the present application.

[0047] Figure 2 Single-cell transcriptome analysis of human embryonic pancreas lineage development: (a) Single-cell differentiation paths from DE cells to each lineage of pancreas in human embryonic period. DE, definitive endoderm; MG-P, midgut cell of pancreatic differentiation fate; AL-P, anterior lip cell of pancreatic differentiation fate; DPE, dorsal pancreatic endoderm cell; VPE, ventral pancreatic endoderm cell; PP, pancreatic precursor cell; Tip, tip cell; Trunk, trunk cell; Acinar, acinar cell; Duct, duct cell; EP, endocrine precursor cell. SS, somite stage; CS, Carnegie stage; W, weeks post-conception. (b) The differentiation model of pancreas lineages corresponding to (a). (c, d) The co-expression network (c) and expression heatmap (d) of transcription factors from DE cells to pancreatic endocrine cells in human embryonic period. I-III represent 3 gene modules. (e) The relationship tree between MG-P, AL-P, DPE, VPE and PP-early cells in human and mouse.

[0048] Figure 3 Single-cell transcriptome analysis of control method: (a) Flow cytometry analysis statistics of S1D2 cells. CXCR4, CD117, FOXA2 and SOX17 are the characteristic genes of DE. (b) Cell types (left) and induction periods (right) displayed in control method. GI, gastrointestinal cell; EC, enterochromaffin cell. (c) The proportion of each cell type at each induction period in control method. The color is consistent with the cell types in (b). (d, e) The co-expression network (d) and expression heatmap (e) of transcription factors from DE cells to pancreatic endocrine cells in control method. I-IV represent 4 gene modules. (f) Similarity analysis of each cell type in control method and the cell types related in vivo. Stomach, stomach cell; Duodenum, duodenum cell; SI, small intestine cell; Liver, liver cell; GI, gastrointestinal cell; EC, enterochromaffin cell.

[0049] Figure 4. Single-cell transcriptome analysis of the method of the present application: (a) Cell types (left) and induction stages (right) in the method of the present application. (b) Proportion of each cell type at each induction stage in the method of the present application. The color is consistent with the cell types in Figure 4a. (c, d) Co-expression network (c) and expression heatmap (d) of transcription factors from DE cells to pancreatic endocrine cells in the method of the present application. I-III represent three gene modules. (e) Similarity analysis of each cell type in the method of the present application with the relevant cell types in vivo. Stomach, stomach cells; Duodenum, duodenum cells; SI, small intestine cells; Liver, liver cells; GI, gastrointestinal cells; EC, enterochromaffin cells. (f) Comparison of gene network establishment of PP cells in vivo during human embryonic period, in the control method and in the method of the present application. (g) Expression of transcription factors in DE to PP cells in the control method and in the method of the present application. Blue words represent the control method, and red words represent the method of the present application. (h) Comparison of single-cell paths from DE to PP cells in the control method and in the method of the present application. (i) Comparison of single-cell paths from EP to endocrine cells in the control method and in the method of the present application. (j) Expression of transcription factors in EP1-4 cells in the control method and in the method of the present application. Blue words represent the control method, and red words represent the method of the present application.

[0050] Figure 5. Comparison of effects of different induction durations of S2 in the present application: (a) Single-cell transcriptome analysis of S2 with different induction durations in the present application and the control method. (b) Similarity analysis of S2 cells after induction with different S2 durations with the relevant cell types in vivo. (c) Expression levels of genes in S2 cells after induction with different S2 durations. (d) Expression levels of genes in S4D3 cells of the control method and S3D3 cells induced with different S2 durations in the present application.

[0051] Figure 6. Comparison of effects of different factors added to S2 in the present application: (a) Single-cell transcriptome analysis of S2D4 cells under different factor treatment conditions. AA represents addition of Activin A, CH represents addition of CHIR-99021, and Ctrl represents the control group. (b) Similarity analysis of S2D4 cells (DE4) under different factor treatment conditions with the relevant cell types in vivo. (c) Projection display of DE, AL-L, AL-P and VPE cells in vivo and S2D4 cells under different factor treatment conditions. (d) Proportion of cells under different factor treatment conditions of S2. (e) Expression levels of genes in S2D4 cells (DE4) under different factor treatment conditions of S2.

[0052] Figure 7. Functional analysis of islet-like tissue in control and the present method: (a, b) Immunofluorescence staining showing the expression of characteristic proteins in islet-like tissue in control and the present method. (c) Glucose tolerance test analysis of control S7D14 islet-like tissue and the present method S6D14 islet-like tissue transplanted into NOD-SCID mice, a diabetic model, for one month. n represents the number of mice. (d) Changes in fasting blood glucose of control S7D14 islet-like tissue, the present method S6D14, S6D3 and S5D7 islet-like tissue transplanted into NOD-SCID mice, a diabetic model. Fasting blood glucose is the blood glucose level after 8 hours of fasting. (e) Immunofluorescence staining showing C-peptide and GCG expression in the graft. (f) Smart-seq3 single-cell transcriptome analysis of the graft one month after islet-like tissue transplantation. UI represents cell types with unclear identity. (g) Characteristic gene expression of each cell type in Figure 7f. (h) Cell proportion composition of the graft one month after islet-like tissue transplantation. The color is consistent with the cell type in Figure 7f. (i) 10xGenomics single-cell transcriptome analysis of the present method S5D7 islet-like tissue showing cell types (left) and cell proportions (right). (j) 10xGenomics single-cell transcriptome analysis of the present method S5D7 islet-like tissue transplanted for one month showing cell types (left) and cell proportions (right). DETAILED DESCRIPTION

[0053] The preferred embodiments of the present application will be described herein below with reference to the accompanying drawings, in which the preferred embodiments of the present application are illustrated. It should be understood that the preferred embodiments described herein are merely for the purpose of illustration and explanation, and are not intended to limit the present application in any manner.

[0054] The English name abbreviations used in the embodiments and their Chinese counterparts are listed as follows:

[0055] Example 1 (Comparative Example) Existing method for inducing human pluripotent stem cells to differentiate into pancreatic precursor cells

[0056] The existing method for inducing human pluripotent stem cells to differentiate into pancreatic precursor cells is a method developed by Timothy Kieffer's laboratory, and many optimization schemes of research groups are related to this method. These schemes use various combinations of recombinant proteins or small molecules to regulate signaling pathways, and are treated for different lengths of time. Generally, PE / PP-like cells are formed through four induction stages, and islet-like tissue or beta-like cells are formed through two or three subsequent induction stages. In this process, the characteristics of each differentiation stage in vitro are defined by using limited cell type-related characteristic genes.

[0057] The following will be described in the present application, the control method of 7 stages (Stage, hereinafter referred to as S), see Figure 1a, the whole process is 37℃, 5% carbon dioxide culture conditions.

[0058] The human embryonic stem cell line H1 is used as the starting point, which is maintained in the mTeSR1 pluripotent stem cell culture medium. TrypLE TM Express enzyme is used to disperse H1 embryonic stem cells into single cells, which are plated on a pre-prepared 1:30 diluted Matrigel incubated culture plate, with a cell concentration of 0.53 x 10 6 cells / cm 2 , and cultured in mTeSR1 pluripotent stem cell culture medium with a final concentration of 10 μM Y-27632, and the induction differentiation stage is started after 24 hours.

[0059] The basic medium for stages 1 and 2 (S1 and S2) is MCDB 131 medium with a final concentration of 1.5 g / L NaHCO3, 1x GlutaMAX, 10 mM D-glucose, 0.5% BSA, 100 U / mL penicillin-streptomycin.

[0060] The basic medium for stages 3 and 4 (S3 and S4) is MCDB 131 medium with a final concentration of 2.5 g / L NaHCO3, 1x GlutaMAX, 10 mM D-glucose, 2% BSA, 0.5x ITS-X, 100 U / mL penicillin-streptomycin.

[0061] The basic medium for stages 5-7 (S5-S7) is MCDB 131 medium with a final concentration of 1.5 g / L NaHCO3, 1x GlutaMAX, 20 mM D-glucose, 2% BSA, 0.5x ITS-X, 10 μM ZnSO4, 10 μg / mL heparin, 100 U / mL penicillin-streptomycin.

[0062] S1 lasts for 2 days (Day, hereinafter referred to as D), by simultaneously activating the TGF-β Nodal pathway (adding a final concentration of 100 ng / mL Activin A) and activating the WNT pathway (adding CHIR-99021, a final concentration of 3 μM is used on the first day (D1), and a final concentration of 0.3 μM is used on D2), to reach the DE stage of CXCR4+ / FOXA2+ / SOX17+.

[0063] S2 lasts for 2 days, by simultaneously activating the FGF pathway (adding a final concentration of 50 ng / mL FGF10) and adding a final concentration of 0.25 mM vitamin C, to reach the PGT stage, which currently has no better indicator genes.

[0064] S3 for 2 days, reaching the PF stage of PDX1+ by simultaneously activating the FGF pathway (addition of FGF10 at a final concentration of 50 ng / mL), inhibiting the SHH pathway (addition of SANT-1 at a final concentration of 0.25 mM), activating the Retinoic acid (RA) pathway (addition of RA at a final concentration of 1 mM), inhibiting the BMP pathway (addition of LDN-193189 at a final concentration of 100 nM), activating PKC (addition of TPPB at a final concentration of 200 nM) and addition of Vitamin C at a final concentration of 0.25 mM.

[0065] S4 for 3 days, reaching the PE / PP stage of PDX1+ / NKX6-1+ by simultaneously activating the FGF pathway (addition of FGF10 at a final concentration of 2 ng / mL), inhibiting the SHH pathway (addition of SANT-1 at a final concentration of 0.25 mM), activating the Retinoic acid (RA) pathway (addition of RA at a final concentration of 0.1 mM), inhibiting the BMP pathway (addition of LDN-193189 at a final concentration of 200 nM), activating PKC (addition of TPPB at a final concentration of 100 nM) and addition of Vitamin C at a final concentration of 0.25 mM.

[0066] At the end of S4, cells were dispersed into single cells using TrypLE TM Express enzyme and aggregated in 5 mL of S5 medium at a concentration of 5 x 10 6 cells / well in a low-attachment 6-well plate, the first day of S5 added Y-27632 at a final concentration of 10 mM to maintain single-cell activity and incubated at 100 rpm on an orbital shaker. Further differentiation into EP cells and islet-like tissue containing beta-like cells was achieved by 3 further stages.

[0067] S5 for 3 days, reaching the EP stage by simultaneously inhibiting the SHH pathway (addition of SANT-1 at a final concentration of 0.25 mM), activating the Retinoic acid (RA) pathway (addition of RA at a final concentration of 0.05 mM), inhibiting the BMP pathway (addition of LDN-193189 at a final concentration of 100 nM), inhibiting TGF-beta RI kinases (addition of ALK5i II at a final concentration of 10 mM) and addition of Thyroxine T3 at a final concentration of 1 mM.

[0068] S6 for 7 days, obtaining immature islet-like tissue by simultaneously inhibiting the BMP pathway (addition of LDN-193189 at a final concentration of 100 nM), activating the Notch pathway (addition of GSIXX at a final concentration of 100 nM), inhibiting TGF-beta RI kinases (addition of ALK5i II at a final concentration of 10 mM) and addition of Thyroxine T3 at a final concentration of 1 mM.

[0069] S7 for 14 days by simultaneous inhibition of TGF-beta RI kinase (addition of final concentration 10 μΜ ALK5i II), inhibition of tyrosine protein kinase receptor (addition of final concentration 2 μΜ R428), and addition of final concentration 10 μΜ vitamin E analogue Trolox, 1 mM N-acetyl cysteine, 1 μΜ thyroid hormone T3, to obtain further matured islet-like tissue.

[0070] Example 2 Improved method of inducing differentiation of human pluripotent stem cells into pancreatic precursor cells

[0071] The present application also starts with human embryonic stem cell line H1, maintained in culture in mTeSR1 pluripotent stem cell medium, dispersed into single cells using TrypLE Express enzyme, plated onto pre-prepared 1:30 dilution Matrigel incubated culture plates, with a plating cell concentration of 0.53 x 10 TM Express enzyme into single cells, plated onto pre-prepared 1:30 dilution Matrigel incubated culture plates, with a plating cell concentration of 0.53 x 10 6 cells / cm 2 in mTeSR1 pluripotent stem cell medium with final concentration 10 μΜ Y-27632, and initiation of differentiation induction after 24 hours. The present application induces human pluripotent stem cells into pancreatic precursor cells PP for 3 stages, on this basis to continue 3 induction stages to obtain islet-like tissue, the whole process is 37°C, 5% carbon dioxide culture conditions, see Figure 1b.

[0072] The base medium for S1 and S2 is the same as S1 and S2 of the control method, which is MCDB 131 medium, with final concentration 1.5 g / L NaHCO3, 1 x GlutaMAX, 10 mM D-glucose, 0.5% BSA, 100 U / mL penicillin-streptomycin.

[0073] The base medium for S3 is the same as S3 and S4 of the control method, which is MCDB 131 medium, with final concentration 2.5 g / L NaHCO3, 1 x GlutaMAX, 10 mM D-glucose, 2% BSA, 0.5 x ITS-X, 100 U / mL penicillin-streptomycin.

[0074] The base medium for S4-S6 is the same as S5-S7 of the control method, which is MCDB 131 medium, with final concentration 1.5 g / L NaHCO3, 1 x GlutaMAX, 20 mM D-glucose, 2% BSA, 0.5 x ITS-X, 10 μΜ ZnSO4, 10 μg / mL heparin, 100 U / mL penicillin-streptomycin.

[0075] S1 for 2 days by simultaneous activation of the TGF-beta Nodal pathway (addition of Activin A at a final concentration of 100 ng / mL) and activation of the WNT pathway (addition of CHIR-99021 at a final concentration of 3 mM on D1 and 0.3 mM on D2).

[0076] S2 for 4 days by simultaneous activation of the FGF pathway (addition of FGF10 at a final concentration of 50 ng / mL), activation of the TGF-beta Nodal pathway (addition of Activin A at a final concentration of 10 ng / mL), activation of the WNT pathway (addition of CHIR-99021 at a final concentration of 0.3 mM) and addition of Vitamin C at a final concentration of 0.25 mM.

[0077] S3 for 3 days by simultaneous activation of the FGF pathway (addition of FGF10 at a final concentration of 2 ng / mL), inhibition of the SHH pathway (addition of SANT-1 at a final concentration of 0.25 mM), activation of the Retinoic acid (RA) pathway (addition of RA at a final concentration of 1 mM), inhibition of the BMP pathway (addition of LDN-193189 at a final concentration of 100 nM), activation of PKC (addition of TPPB at a final concentration of 200 nM) and addition of Vitamin C at a final concentration of 0.25 mM.

[0078] At the end of S3, cells were dispersed into single cells using TrypLE TM Express enzyme and aggregated in 5 mL of S4 medium at a concentration of 5 x 10 6 cells / well in a low-attachment 6-well plate, with the addition of Y-27632 at a final concentration of 10 mM on the first day of S4 to maintain single-cell activity, and incubation on an orbital shaker at 100 rpm.

[0079] The induction protocol of S4-S6 is identical to S5-S7 of the control method, but the duration of S6 of the present application can be shortened or omitted.

[0080] S4 for 3 days by simultaneous inhibition of the SHH pathway (addition of SANT-1 at a final concentration of 0.25 mM), activation of the Retinoic acid (RA) pathway (addition of RA at a final concentration of 0.05 mM), inhibition of the BMP pathway (addition of LDN-193189 at a final concentration of 100 nM), inhibition of the TGF-beta RI kinase (addition of ALK5i II at a final concentration of 10 mM) and addition of Thyroxine T3 at a final concentration of 1 mM.

[0081] S5 for 7 days by simultaneous inhibition of the BMP pathway (addition of LDN-193189 at a final concentration of 100 nM), activation of the Notch pathway (addition of GSIXX at a final concentration of 100 nM), inhibition of the TGF-βRI kinase (addition of ALK5i II at a final concentration of 10 μΜ) and addition of thyroid hormone T3 at a final concentration of 1 μΜ.

[0082] S6 for 3-14 days by simultaneous inhibition of the TGF-βRI kinase (addition of ALK5i II at a final concentration of 10 μΜ), inhibition of the tyrosine-protein kinase receptor (addition of R428 at a final concentration of 2 μΜ) and addition of the vitamin E analogue Trolox at a final concentration of 10 μΜ, 1 mM N-acetylcysteine, 1 μΜ thyroid hormone T3.

[0083] Example 3 result analysis

[0084] 1. Assess the induction effect of both methods using the human in vivo pancreatic differentiation pathway and gene network, respectively:

[0085] By analyzing the single-cell transcriptome of human in vivo pancreas during embryonic stages (from 2 somites to 19 weeks of gestation), the cell types, differentiation pathway and gene network of the pancreatic lineage were resolved (Fig. 2a-d), which were used as the standard to assess the in vitro induction effect. The analysis found that the gene network during the development of human in vivo from DE cells to endocrine cells can be divided into 3 modules, and the PP cells are at the key position between the first and second module transitions (Fig. 2c, d). And in vivo PP cells are developed from the ventral (DE-AL-P-VPE-PP) and dorsal (DE-MG-P-DPE-PP) endoderm pathways, respectively, to converge (Fig. 2a, b), where the cells on the ventral pathway (AL-P and VPE) are more similar to PP cells than the cells on the dorsal pathway (MG-P and DPE), which is not the case in mice (Fig. 2e). This suggests that mimicking the ventral development pathway in humans is a more efficient way to form PP cells.

[0086] (1) Using the human embryonic stem cell line H1 as a starting point, the cells were subjected to directed differentiation using the control method in Example 1. Flow cytometry analysis showed that the proportion of CXCR4+CD117+ cells and FOXA2+SOX17+ cells at the end of S1D2 could reach more than 97% (Figure 3a). At the same time, Smart-seq3 single-cell transcriptome analysis was performed on the differentiated cells, and the analysis found that the control method could be divided into 20 cell types during the entire differentiation process (Figure 3b). Among them, the SC-β identified in the last stage S7 accounted for 18%-29%, the SC-α accounted for 37%-64%, the SC-δ accounted for 2%-5%, the EC1 accounted for 5%-15%, the EC2 accounted for 1%-5%, the EC3 (β precursor cell) accounted for 0.8%-2% (Figure 3c), and the SC-β and its precursor cells accounted for 19%-32%. From the gene co-expression network, the differentiation process from DE cells to endocrine cells in vitro can be divided into four modules (Figure 3d, e), which indicates the difference in gene network construction between the control scheme and in vivo development. From the similarity of in vivo and in vitro cells, the PE and PE / PP cells obtained by the control method have similar properties to both DPE and VPE in vivo, and the PP cells obtained have certain similarity with the PP cells in vivo, and also have strong similarity with the gastrointestinal cells (Figure 3f), which indicates that the cell properties obtained by the control scheme are mixed, and the unstable gene network construction leads to unstable cell state. In addition, EC3 cells have similarity with β cells in vivo, combined with characteristic gene expression and path characteristics, it is considered to be β precursor cells, while EC1 / 2 has similarity with EC cells in vivo.

[0087] (2) By analyzing Example 1, it was found that there were problems in the construction of the gene network in the current induction process. Similarly, using the method of the present application (Example 2), the human embryonic stem cell line H1 was differentiated directionally, and the Smart-seq3 single-cell transcriptome analysis was performed on the differentiated cells. Analysis found that the method of the present application could be divided into 20 cell types during the entire differentiation process (Figure 4a). Among them, the last stage S6 identified β-like cells (SC-β) accounted for 55%-68%, α-like cells (SC-α) accounted for 11%-21%, δ-like cells (SC-δ) accounted for 0.8%-2%, EC1 accounted for 11%-17%, EC2 accounted for 0%-3%, EC3 (β precursor cells) accounted for 2%-5% (Figure 4b), and SC-β and its precursor cells accounted for 60%-70%. Compared with the control method, the cell proportion of SC-β and EC3 (β precursor cells) was significantly increased, the cell proportion of SC-α was significantly decreased, and the cell proportion of SC-δ was slightly decreased. From the gene co-expression network, the gene network in the differentiation process from DE cells to endocrine cells in vitro of the present application could be divided into three modules (Figures 4c, d), which was similar to the transition in the in vivo development process. From the similarity of in vivo and in vitro cells, the DE4 cells obtained by the method of the present application at S2D4 were similar to AL-P in vivo, and the PP cells obtained were similar to PP cells in vivo, and did not have similarity with gastrointestinal cells in vivo (Figure 4e), which indicated that the gene network construction of the present application was stable, and the cell properties were clear and stable. In addition, by comparing the gene network formed with the PP cells in vivo, the method of the present application could better construct the gene network connection of the PP cells in vivo than the control method, and express higher characteristic genes of PP cells, such as SOX9, PDX1, NKX6-1 and PTF1A, etc. (Figures 4f, g). By comparing the differentiation paths of the control method and the method of the present application, it was found that from the optimization treatment of S2, the two methods showed different differentiation paths, including the path of PP generation and the path of EP differentiation (Figures 4h, i). In addition, the EP cells generated in the method of the present application highly expressed the regulatory transcription factors related to β cells, such as PAX4, PDX1, NKX6-1 and MAFB, while the EP cells in the control method highly expressed the regulatory transcription factors related to α cells, such as ARX and ETV1, which indicated that the EP cells of the two methods had different endocrine lineage differentiation tendencies (Figure 4j).

[0088] 2. Extending the induction time of S2 can enhance the similarity of induced DE cells to AL-P cells in vivo

[0089] Using human embryonic stem cell line H1 as starting material, S2 condition is: MCDB 131 medium, adding 1.5 g / L NaHCO3, lx GlutaMAX, final concentration 10 mM D-glucose, 0.5% BSA, 100 U / mL penicillin-streptomycin, 50 ng / mL FGF10, 10 ng / mL Activin A, 0.3 mM CHIR-99021 and 0.25 mM Vitamin C. S2 lasts for 3, 4, 5, 6 days respectively (denoted as groups A-D), and continues to be induced to S3 according to the induction method of S3 in Example 2, and the cells at the end of S2 and S3 are subjected to Smart-seq3 single cell transcriptome analysis (Figure 5a). At the same time, the control method of Example 1 is used for directional differentiation to S4, and the cells at the end of each stage of S2-S4 are subjected to Smart-seq3 single cell transcriptome analysis (Figure 5a).

[0090] Analysis found that cells treated with S2 for 4-6 days have stronger similarity to AL-P cells in vivo compared to cells treated with S2 for 3 days (Figure 5b). Among them, S2 induction for 4 days is the best duration, S2D4 up-regulates the expression of transcription factors SOX4, HHEX, PROX1 and ONECUT1 related to AL-P, and down-regulates the expression of transcription factors SOX17, GATA4 and PITX2 related to DE and MG-P cells (Figure 5c). Continuing to extend S2 to 5 days and 6 days cannot further enhance the properties of AL-P, and will cause down-regulation of PP cell related transcription factor PDX1 and up-regulation of non-PP cell related gene AFP at S3D3 (Figure 5d).

[0091] 3. S2 activates TGF-β Nodal pathway and activates WNT pathway to enhance the similarity of induced DE cells to AL-P cells in vivo

[0092] Using human embryonic stem cell line H1 as starting material, S2 lasts for 4 days, and the condition is: MCDB 131 medium, adding 1.5 g / L NaHCO3, lx GlutaMAX, final concentration 10 mM D-glucose, 0.5% BSA, 100 U / mL penicillin-streptomycin, 50 ng / mL FGF10 and 0.25 mM Vitamin C (as Ctrl group), or adding different concentrations of TGF-β Nodal pathway activators (adding final concentration 10 ng / mL, 25 ng / mL, 50 ng / mL Activin A, denoted as AA1-AA3 groups) or WNT pathway activators (adding final concentration 0.3 mM, 1 mM, 2 mM CHIR-99021, denoted as CH1-CH3 groups) at the same time, and the cells at S2D4 are subjected to Smart-seq3 single cell transcriptome analysis (Figure 6a).

[0093] It was found that the addition of CHIR-99021 could promote the differentiation of DE4 cells of S2D4 to AL-P and AL-L cells in vivo, while the addition of Activin A could effectively inhibit the differentiation to AL-L cells (Fig. 6b, c). With the increase of the concentration of CHIR-99021, the proportion of off-target cells increased, and the addition of Activin A also delayed the off of the expression of related genes of DE cells, delaying the differentiation process (Fig. 6b-d). Therefore, the combination of the two factors with the optimal concentration of 10 ng / mL Activin A and 0.3 μM CHIR-99021 and the functional hedge allowed the early cells to inhibit the differentiation to liver cells and promote the differentiation to AL-P cells, and inhibited the expression of liver differentiation-related genes such as AFP, FGB and AMBP, and DE-related genes PITX2 and GATA4 (Fig. 6e).

[0094] 4. Function detection of the islet-like tissue induced by the present application (Example 2) and the control method (Example 1)

[0095] Using the human embryonic stem cell line H1 as the starting material, the islet-like tissue induced at the last stage of 14 days was detected by immunofluorescence staining, and it was found that the islet-like tissue of the present application had a high proportion of PDX1+NKX6-1+C-peptide+cells and a reduced proportion of GCG+cells (Fig. 7a, b), which was consistent with the analysis results of single-cell transcriptome data.

[0096] In addition, the islet-like tissue induced in the last stage (S6D14 of the present application, S7D14 of the control method) was transplanted into immunodeficient mice NOD-SCID, a model of diabetes, with about 2.5 million cells per mouse. NOD-SCID mice were purchased from Vantianlihua, and a model of diabetic mice was constructed by injecting a low dose (35 mg / kg body weight) of streptozotocin (STZ) for 5 consecutive days. It was found that the islet-like tissue obtained by the method of the present application had better blood glucose-lowering effect in the glucose tolerance test one month after transplantation, and faster blood glucose-lowering function (Fig. 7c, d). Immunofluorescence staining was performed on the recovered grafts one month after transplantation, and it was found that the grafts of the present application had a higher proportion of C-peptide+ cells, and a high proportion of such cells was stably observed in the grafts six months after transplantation (Fig. 7e). Smart-seq3 single-cell transcriptome analysis was performed on the grafts one month after transplantation, and it was found that there were various cell types in the grafts, including SC-β, SC-α, SC-δ and EC cells, as well as a group of cells with unclear cell properties (UI, unidentified), which mixedly expressed PDX1 and exocrine characteristic genes such as CPA2 and SOX9 (Fig. 7f, g). Among them, the SC-β, SC-α, SC-δ and EC in the S6D14 grafts of the present application accounted for 48.8%, 29.0%, 8.8% and 12.5%, respectively, and less than 1% of UI cells; while the SC-β, SC-α, SC-δ and EC in the S7D14 grafts of the control method accounted for 8.9%, 70.0%, 1.8% and 3.4%, respectively, and as high as 15.9% of UI cells (Fig. 7h). The grafts of the present application had a higher proportion of C-peptide+ β-like cells and a lower proportion of GCG+ α-like cells, but the cell properties did not differ significantly from the control method in the transcriptome.

[0097] In view of the fact that the proportion of SC-β cells in the method of the present application at S5D7 is already high, further transplantation of islet-like tissue of the method of the present application at an early stage (S5D7 and S6D3) to the immunodeficient mouse NOD-SCID of the diabetes model (same as the method described above) was carried out. Functional detection found that mice transplanted with S5D7 and S6D3 islet-like tissue had better blood glucose regulation effect than mice transplanted with S6D14 islet-like tissue (Fig. 7c, d), and Smart-seq3 single-cell transcriptome analysis also showed that the proportion of SC-β cells in the grafts was higher (67%) (Fig. 7h). In addition, analysis found that although there were some differences in the transcriptome between S5D7 cells and S6 cells before transplantation in the method of the present application, there were no obvious differences after transplantation (Fig. 3a, 7f), which indicated that transplantation into the body can significantly promote the maturation of islet-like tissue and eliminate the relative immaturity difference during in vitro induction. Through 10xGenomics single-cell transcriptome analysis of S5D7 islet-like tissue before transplantation in the method of the present application and its grafts one month after transplantation, it was found that the cell types and proportions were consistent with the Smart-seq3 single-cell transcriptome analysis (Fig. 7i, j). Through Smart-seq3 single-cell transcriptome analysis of the grafts after transplantation of the early S6D7 islet-like tissue of the control method, it was found that the UI cells accounted for as high as 46% (Fig. 7h), which indicated that the control method shortened by one period would result in incomplete differentiation of endocrine lineage, and could not achieve the differentiation effect after shortening by one period in the present application. In summary, the method of the present application can significantly shorten the induction time to as short as 5 induction stages for a total of 19 days, and obtain islet-like tissue with better function after transplantation.

[0098] Example 4 Analysis of gene co-expression network

[0099] During development, the speed of cell fate transition is not constant, but there are alternations of smooth transition and leap. The transition of gene modules in the transcription factor co-expression network represents the leap of cell fate, and the number of modules reflects the number of stages of smooth transition of cell state during development. The present application deeply analyzes the dynamic changes of transcription factor co-expression network during the development of DE to endocrine lineage in embryonic period of human body by analyzing single-cell transcriptome, and reveals its three-module structure (Fig. 2c, d). The first module covers the differentiation of DE to PP-early cells, and the cells gradually close the expression of genes related to endoderm, and open the expression of genes related to pancreatic differentiation. The leap between the first and second modules indicates that the transcription factors related to pancreatic differentiation lose the co-expression characteristics of transcription factors related to the identity of endoderm cells in the first module, establish the co-expression characteristics of transcription factors related to the identity of endocrine precursor cells, and smoothly transition to endocrine precursor cells. The transition between the second and third modules represents the regulatory transition of endocrine lineage fate.

[0100] The analysis of the transcription factor co-expression network in the in vivo process in the present application is completed by the following bioinformatics analysis method:

[0101] 1) First, in order to balance the number of cells in different cell populations in the pathway, we randomly sample each population of the 21 cell populations identified in Figure 2a to achieve down-sampling. If the number of cells in a cell population is greater than 80, 80 cells are randomly sampled therefrom, and if it is less than or equal to 80, all the cells are retained.

[0102] 2) For the down-sampled data set, the Pearson correlation coefficient between all transcription factor genes is calculated by R software, with the transcription factor genes as nodes and the correlation coefficient values as the weight of the edges to establish a fully connected network.

[0103] 3) The edges with a weight less than 0.25 are cut off, the nodes with an expression ratio less than 10% or greater than 80% in the down-sampled cell set are deleted, and then the nodes without connection and the sub-networks with less than 10 genes are removed.

[0104] 4) The remaining network is clustered using the Louvain algorithm in the igraph software, the communities (Community) related to the cell cycle and batch effect are removed, and the remaining communities are merged according to the similarity to obtain multiple modules of the network. Considering the inherent differences in gene expression between in vivo development system and in vitro directed induction system, the inventors believe that the module architecture of the in vivo gene co-expression network is more reliable and more accurate as a standard for comparison with in vitro, and can evaluate the in vitro differentiation process from the global perspective of cell fate regulation change.

[0105] By performing the above similar analysis on the in vitro differentiation process from embryonic stem cells to endocrine cells, the inventors found that the control method presents a four-module gene co-expression network feature in this process (Figures 3d, e), which is inconsistent with the in vivo characteristics, and its unstable early gene network is the core of the present application. Through analysis, it is found that the method of the present application presents a three-module gene co-expression network consistent with in vivo (Figures 4c, d).

[0106] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A method for inducing the differentiation of human pluripotent stem cells into pancreatic progenitor cells, characterized in that: The method comprises the following steps (S): S1: Simultaneous activation of the TGF-βNodal pathway and the WNT pathway; S2: Simultaneous activation of the FGF pathway, TGF-β Nodal pathway, and WNT pathway and addition of vitamin C to a final concentration of 0.25 mM; S3: Simultaneously activate the FGF pathway, inhibit the SHH pathway, activate the Retinoic acid (RA) pathway, inhibit the BMP pathway, activate PKC, and add a final concentration of 0.25 mM vitamin C; S4: Simultaneously inhibit the SHH pathway, activate the Retinoic acid (RA) pathway, inhibit the BMP pathway, inhibit TGF-βRI kinase, and add thyroxine T3 at a final concentration of 1 μM; S5: Simultaneously inhibit the BMP pathway, activate the Notch pathway, inhibit TGF-βRI kinase, and add thyroxine T3 at a final concentration of 1 μM.

2. The method according to claim 1, wherein The specific operations of each step are: Human pluripotent stem cells were cultured in mTeSR1 pluripotent stem cell medium supplemented with a final concentration of 10 μM Y-27632, and differentiation was initiated after 24 hours: S1 lasts for 2 days: Activin A and CHIR-99021 were added to the culture medium at a final concentration of 100 ng / mL, with a final concentration of 3 μM on the first day and 0.3 μM on the second day; S2 lasted for 4 days: the culture medium was supplemented with a final concentration of 50 ng / mL FGF10, 10 ng / mL Activin A, 0.3 μM CHIR-99021, and 0.25 mM vitamin C; S3 lasts for 3 days: the culture medium is supplemented with a final concentration of 2 ng / mL FGF10, 0.25 μM SANT-1, 1 μM RA, 100 nM LDN-193189, 200 nM TPPB, and 0.25 mM vitamin C; S4 lasted for 3 days: the culture medium was supplemented with a final concentration of 0.25 μM SANT-1, 0.05 μM RA, 100 nM LDN-193189, 10 μM ALK5i II, and 1 μM thyroxine T3; S5 lasted for 7 days: the culture medium was supplemented with a final concentration of 100 nM LDN-193189, 100 nM GSIXX, 10 μM ALK5i II, and 1 μM thyroxine T3; The basal medium used in S1 and S2 was MCDB 131 medium, supplemented with a final concentration of 1.5 g / L NaHCO 3 , 1× GlutaMAX, 10 mM D-glucose, 0.5% BSA, and 100 U / mL penicillin-streptomycin. The basal medium used in S3 was: MCDB 131 medium supplemented with a final concentration of 2.5 g / L NaHCO , 1× GlutaMAX, 10 mM D-glucose, 2% BSA, 0.5× ITS-X, and 100 U / mL penicillin-streptomycin; The basal medium used in S4-S5 was: MCDB 131 medium, supplemented with a final concentration of 1.5 g / L NaHCO3, 1×GlutaMAX, 20 mM D-glucose, 2% BSA, 0.5×ITS-X, 10 μM ZnSO4, 10 μg / mL heparin, and 100 U / mL penicillin-streptomycin.

3. Use of the method according to claim 2 in inducing pancreatic islet-like tissue.

4. The use according to claim 3, characterized in that The application comprises the following steps: S6 lasts for 3-14 days: the culture medium is added with a final concentration of 10μM ALK5i II, 2μM R428, 10μM vitamin E analog Trolox, 1mM N-acetylcysteine, and 1μM thyroxine T3; the basal culture medium used in S6 is: MCDB 131 culture medium, additionally supplemented with a final concentration of 1.5g / L NaHCO3, 1×GlutaMAX, 20mM D-glucose, 2% BSA, 0.5×ITS-X, 10μM ZnSO4, 10μg / mL heparin, and 100U / mL penicillin-streptomycin.

5. A method for evaluating a cell differentiation process in vitro, characterized in that The method comprises the following steps: 1) First, in order to balance the number of cells in different cell populations on the path, we downsampled each cell population, in principle to make the cell number close and the network stable for subsequent analysis; 2) For the downsampled dataset, the correlation coefficients between all transcription factor genes were calculated, and a fully connected network was established with the transcription factor genes as nodes and the correlation coefficient values ​​as the weights of the edges; 3) 90-95% of the low-weight edges were cut off, and nodes with too low (<5-10%) or too high (>80-95%) expression ratios in the cells of the downsampled dataset were deleted to thin out the network, retaining the largest sub-network after thinning; 4) The remaining network was clustered using an algorithm, such as the Louvain algorithm, to remove communities related to the cell cycle and batch effect, and the remaining communities were merged based on similarity to obtain multiple modules of the network; 5) By analyzing and determining whether the in vitro differentiation process establishes the three-module co-expression network structure of the in vivo differentiation process, the differentiation efficiency and cell characteristics of the in vitro differentiation process were evaluated.

Images