A method for analyzing molecular differences in glioblastoma and its application

Abstract

The present invention provides a method for analyzing molecular differences in glioblastoma and its application, belonging to the field of biotechnology. Using differential gene expression analysis, spatial transcriptomics analysis, deconvolution analysis, single-cell RNA sequencing data, and immunofluorescence, this application identified specific molecular signatures associated with pseudopalisading necrosis (PAN) and microvascular proliferation (MVP), revealing differences between these tissue structures. The results showed that PAN and MVP shared spatial expression of specific marker genes and were enriched for endothelial cells of distinct phenotypes. Single-cell trajectories inferred pseudotemporal trajectories of glioblastoma stem cells (GSCs) to EC differentiation. This study reveals distinct roles for endothelial cells in the PAN and MVP regions of GBM, highlighting the complexity and diversity of the tumor microenvironment and providing new insights into endothelial cell subpopulations and GSC differentiation into endothelial cells in GBM.

Description

Technical Field

[0001] The present invention relates to the field of biotechnology, and in particular to a method for analyzing molecular differences in glioblastoma and an application thereof. Background Art

[0002] Glioblastoma (GBM) is a highly lethal malignant intracranial tumor characterized by histopathological features including pseudopalisading cells surrounding necrosis (PAN) and microvascular proliferation (MVP). These features distinguish GBM from low-grade gliomas and have been recognized as markers of poor prognosis. PAN refers to a high-density cellular area surrounding necrotic areas in GBM, while MVP is a form of angiogenesis characterized by the proliferation of endothelial cells in new blood vessels. PAN and MVP are considered indicators of the high malignancy and invasiveness of GBM.

[0003] A widely accepted theory holds that the overgrowth of GBM cells leads to intratumoral vascular damage, which in turn triggers endothelial damage and exacerbates tissue hypoxia. Malignant cells migrate out of the hypoxic environment, forming a wave-like structure that moves toward the periphery, appearing as pseudoapoptotic cells under a microscope. In addition, tumor cells in hypoxic areas secrete vascular endothelial growth factor (VEGF) and IL-8, which promote microvascular proliferation and accelerate the outward expansion of tumor cells accompanied by the formation of new blood vessels. Although considerable progress has been made in the molecular mechanisms of GBM tumorigenesis over the past few decades, the molecular and cellular heterogeneity of these landmark areas and their spatial characteristics remain poorly understood.

[0004] Angiogenesis is crucial for tumor growth and progression. Glioma, including GBM, is a highly vascularized tumor. Endothelial cells (ECs), which constitute the blood vessels, are the most critical, forming capillaries that supply blood to glioma cells. Studies have shown that ECs are heterogeneous in GBM patients and models of different tumor sites. GBM contains a large number of non-tumor components, such as stromal cells, blood vessels, cytokines, immune cells, and fibroblasts, which constitute a complex tumor microenvironment (TME) with many anatomically distinct regions. Necrotic cores surrounded by hypoxic and perivascular areas within the GBM TME are considered to be standard morphological indicators of GBM invasiveness. However, the pathogenesis of these areas in malignant GBM is not yet fully understood. Therefore, elucidating the molecular characteristics of specific regions and targeting therapies based on these characteristics may become a promising therapeutic strategy.

[0005] The Ivy Glioblastoma Atlas Project (Ivy GAP) collects genomic and transcriptomic data from anatomical regions within tumors, including PAN, MVP, and cellular tumors (CT). Therefore, this database can help understand the molecular diversity associated with histological heterogeneity. However, few studies have revealed the spatial heterogeneity of molecular features across different cellular states within PAN and MVP. The development of spatial transcriptomics (SRT) technology has enabled us to detail the correlations between characteristic gene expression profiles while preserving information about the spatial context and organization of the tumor tissue. Here, we integrated RNA sequencing, SRT, and single-cell RNA sequencing (scRNA-seq) data to characterize differences between GBM structures, revealing cellular heterogeneity and molecular stratification within GBM tissue architecture and identifying prognostic and predictive markers corresponding to GBM localization. However, the molecular and cellular heterogeneity of these pathological features and their spatial nature have not been fully elucidated. Comprehensively characterizing histological features in a spatial context will help improve the diagnosis and treatment of GBM. Single-cell transcriptomics technologies have revealed valuable insights into heterogeneity and potential therapeutic molecular targets. SRT can classify adjacent points into histological regions, thus enabling the identification of spatial heterogeneity between patients.

[0006] Intraoperative necrosis, PET (positron emission tomography) tracer flumisonidazole (FMISO) imaging, and histological vascular pathology have highlighted hypoxia as a unique pathological feature of GBM, promoting its invasiveness. The hypoxia-regulated gene N-myc downstream regulatory gene-1 (NDRG1) in GBM plays an important role in regulating migration, angiogenesis, and drug resistance in GBM. Studies have shown that overexpression of NDRG1 can reduce angiogenesis and tumor growth. The study by Yanming Ren et al. also highlighted the enrichment of CH3CL1 in hypoxic and invasive glioma environments. However, the biological significance of these differentially expressed genes still requires further functional verification, and this cautious attitude is crucial for scientific research, especially when exploring potential therapeutic targets. Summary of the Invention

[0007] The purpose of the present invention is to provide a method for analyzing molecular differences in glioblastoma and its application, which provides new insights into the subpopulations of endothelial cells in GBM and the differentiation of GSCs into endothelial cells, and clarifies the molecular characteristics of specific regions, which can be used to guide the targeted treatment of glioblastoma.

[0008] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0009] The present invention provides a method for analyzing molecular differences in glioblastoma, comprising the following steps:

[0010] (1) Based on the Ivy-GAP dataset, we extracted gene expression profiles of glioblastoma tumor cells (CT), pseudopalisading cells (PAN) around necrosis, and microvascular proliferation (MVP) areas, and performed differential gene expression analysis to identify PAN and MVP marker genes.

[0011] (2) Based on the spatial transcriptomic dataset of GBM samples without IDH mutations, spatial transcriptomic analysis was performed to identify spatially variable genes in the pseudopalisading cells (PAN) and microvascular proliferation (MVP) areas surrounding necrosis, and to achieve spatial localization of PAN and MVP marker genes;

[0012] (3) using a spatial transcriptomic dataset of GBM samples to identify different cell types through deconvolution analysis;

[0013] (4) Single-cell RNA sequencing data analysis was performed based on the scRNA-seq dataset to obtain the gene expression of PAN and MVP marker genes in various cell types, and pseudo-time analysis was performed on all cell types;

[0014] (5) Based on the heterogeneity of endothelial cells, unsupervised cluster analysis was performed to determine the different roles played by endothelial cells in the PAN and MVP areas of GBM.

[0015] Preferably, the Ivy-GAP dataset in step (1) is an RNA-seq dataset of different anatomical regions of primary human glioblastomas derived from the Ivy Glioblastoma Atlas Project;

[0016] The steps of the differential gene expression analysis are as follows: converting RNA-seq datasets of different anatomical regions of primary human glioblastoma into log2(n+1) format, and then performing quality control and differential expression analysis on paired data of different anatomical regions using the limma package; performing intersection query using the UPsetR package, and converting overlapping gene IDs from the SYMBOL format to the ENTREZID format using the clusterProfiler package, and performing gene ontology functional enrichment analysis and KEGG enrichment analysis of biological processes, molecular functions and cellular components to obtain differentially expressed gene data of different anatomical regions of primary human glioblastoma, and determine PAN and MVP marker genes.

[0017] Preferably, the steps of spatial transcriptomics analysis in step (2) are:

[0018] The spatial transcriptomics dataset of GBM samples without IDH mutations was filtered, and gene expression at each location was normalized using the SCTransform function. Principal components were then extracted from highly variable genes, and resolution parameters were set for clustering. Seurat and the SpaGCN toolkit were used to identify robust spatially variable genes in the 10X Visium dataset. Seurat was used to integrate points within the same cluster within each sample into pseudo-blocks to achieve spatial localization of PAN and MVP marker genes.

[0019] The filtering conditions are: the number of genes is between 500 and 8000, the number of transcripts is less than 30000, and the proportion of mitochondrial genes is less than 30%; the number of principal components extracted is 30, and the resolution parameter is set to 0.5.

[0020] Preferably, in step (3), the deconvolution analysis step includes:

[0021] Using GBM spatial transcriptome data as a reference dataset, the "Harmony" method was used to integrate the scRNA-seq dataset from GSE131928 to obtain a merged spatial transcriptome dataset. The integrated data were visualized by UMAP dimensionality reduction and unsupervised clustering, and annotated and manually proofread by SingleR. Using RCTD, the merged spatial transcriptome dataset was used for deconvolution analysis to obtain possible constituent cell types, and the deconvolution ratios of each cell type were visualized using the Stdeconvolve and ggplot2 packages.

[0022] Preferably, in step (4), the single-cell RNA sequencing data analysis step includes: analyzing the count table of the scRNA-seq dataset using the Seurat4.0 package, following the standard workflow and default settings; inferring copy number variation CNV in the scRNA-seq data using the InferCNV package, with cutoff = 1; then combining CNV classification with UMAP- and gene set-based classification to define GSCs and GBM-associated macrophages / microglia, inferring intercellular communication in the GSC environment using the CellPhoneDB-R 2.1.7 algorithm, and performing pseudo-time analysis on all cell types.

[0023] The present invention also provides an application of a method for analyzing molecular differences in glioblastoma in guiding targeted therapy of glioblastoma.

[0024] The beneficial effects of the present invention compared with the prior art are:

[0025] (1) This application uses differential gene expression analysis, spatial transcriptomics analysis, deconvolution analysis, single-cell RNA sequencing data and immunofluorescence to identify specific molecular features associated with PAN and MVP, and reveal the differences between these tissue structures, providing new insights into the subpopulations of endothelial cells in GBM and the differentiation of GSCs into endothelial cells, and elucidating the molecular characteristics of specific regions that can be used to guide targeted therapy for glioblastoma. The results showed that PAN and PNZ showed similar expression due to histological similarities, while HBV and MVP highly expressed genes that promote angiogenesis, such as IL-8, ADM, VEGFA and NDRG1. Compared with other regions, HBV and MVP showed high expression of endothelial cell and stemness-related molecules, such as ESM1 and CRIP1. In addition, MVP also showed high expression of tumorigenesis and migration-related molecules, such as HIGD1B and MMP9.

[0026] (2) Differential expression analysis showed that PAN-associated endothelial cells highly expressed CHI3L1, SNRPD3, CLU, LAMTOR4, and AQP1, while MVP-associated endothelial cells expressed MGP, FABP5, collagen, and modifying enzymes such as COL1A2, COL3A1, COL4A1, and COL4A2. Functionally, MVP endothelial cells were particularly active in cell adhesion, angiogenesis, and cytoplasmic translation pathways. These insights revealed the different roles played by endothelial cells in the PAN and MVP regions of GBM, highlighting the complexity and diversity of the tumor microenvironment, and providing new insights into the subpopulations of endothelial cells in GBM and the differentiation of GSCs into endothelial cells. Moreover, NDRG1 and EPAS1 were specifically expressed in the PAN and MVP regions, revealing that two different phenotypes of endothelial cells were enriched in the MVP and PAN regions, respectively. Single-cell trajectory inference showed the differentiation trajectory of glioblastoma stem cells (GSCs) into endothelial cells, providing a new perspective for further understanding the pathogenesis of GBM.

[0027] (3) Studies have shown that GSCs can differentiate into pericyte-like cells that support blood vessel and tumor growth. GSCs are located in the perivascular environment close to endothelial cells. Here, GSCs secrete pro-angiogenic VEGF, which drives the proliferation, survival, migration and permeability of ECs. The receptor PTPRZ1 is preferentially expressed in GSCs, and inhibition of PTPRZ1 can inhibit GBM growth and prolong survival. PTN secreted by TAMs has been shown to stimulate PTPRZ1 on GSCs and promote tumor malignant development. The results of the present invention indicate that PTN / PTPRZ1 and MDK / PTPRZ1 may also play a role in the communication between endothelial cells and GSCs. This suggests that PTN and MDK not only promote malignant transformation by stimulating PTPRZ1 on GSCs, but may also play a key role in the interaction between endothelial cells and GSCs, thereby further promoting tumor progression. Furthermore, live cell imaging showed that GSCs can also transdifferentiate into endothelial progenitor cells and endothelial cells, which may explain the poor effect of anti-angiogenic glioma treatment.

[0028] (4) The present invention found that NDRG1 is highly expressed in the hypoxic PAN region but not in the MVP. MVP endothelial cells exhibit angiogenesis and high collagen expression, while PAN endothelial cells show activation of the hypoxic pathway. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1Figure 1 is an analysis of different molecular features of expression heterogeneity within glioblastoma tumors in Example 1 of the present invention, wherein A is a schematic diagram of multi-level RNA sequencing data processing and integration, RNA-seq dataset (n=41), spatial transcriptomics dataset (SRT, n=28) and single-cell RNA-seq dataset (n=28, GSE131928); B is a heat map of differentially expressed genes (rows) in different anatomical regions (columns), comparing the expression differences (>2-fold change, FDR<0.05) between the perivascular region (including hyperplastic blood vessels [HBV] and microvascular proliferation [MVP]) and cellular tumor (CT) and the peri-necrotic region (perinecrotic zone [PNZ], pseudopalisading cells around necrosis [PAN]) and CT, IT, invasive tumor; LE, leading edge zone; identification of PAN and MVP-related marker genes, wherein C is a Venn diagram representing the significantly upregulated genes in the Ivy GAP data and the Seurat and SpaGCN spatial variant genes (SVGs) in PAN, and D is the Ivy Overlap of significantly upregulated genes in GAP data with those in Seurat and SpaGCN spatially variant genes (SVGs) in MVP; E is the expression of PAN-related marker genes in GBM patients, and F is the expression of MVP-related marker genes in GBM patients. The bar on the right indicates the number of spatially variant expressions of each gene in the patient; color indicates expression level;

[0031] Figure 2 Figure 1 is a Venn diagram analysis of down-regulated genes in the MVP region and down-regulated genes in the PAN region and up-regulated genes in the MVP region and up-regulated genes in the PAN region in Example 1 of the present invention; Figure 2 is a Gene Ontology (GO) biological process network diagram (left) of the MVP region (down-regulated) and a bubble diagram of the top 5 pathways enriched in the Kyoto Encyclopedia of Genes and Genomes (KEGG) (right), where the X-axis is the enrichment score, the bubble size indicates the number of genes included in the project, and the bubble color from purple to red indicates the P value; Figure 3 is a GO and KEGG enrichment analysis of the PAN region (down-regulated); Figure 4 is a GO and KEGG enrichment analysis of the MVP region (up-regulated) and the PAN region (up-regulated); Figure 5 is a GO and KEGG enrichment analysis of the MVP region (up-regulated) and the PAN region (up-regulated);

[0032] Figure 3 This is the spatial cluster identification of GBM patient tissue in Example 1 of the present invention;

[0033] Figure 4 Figure 1 shows the spatial localization of PAN and MVP marker genes in GBM in Example 1 of the present invention, wherein A is a representative section of different regions based on H&E staining and spatially variable genes (SVGs); B is the characteristically expressed genes in each region; C is the expression of 8 genes in the detailed spatial location set of patient UKF423;

[0034] Figure 5This is a comparison of the expression levels of PAN and MVP marker genes in LGG and GBM patients in the TCGA database in Example 1 of the present invention, where A is a marker gene for the PAN region; B is a marker gene for the MVP region;

[0035] Figure 6 This is the Kaplan-Meier survival analysis of low-expression and high-expression genes in glioma patients in Example 1 of the present invention. The blue line represents low-expression patients, and the yellow line represents high-expression patients.

[0036] Figure 7 The expression of NDRG1 and SPAS1 in GBM in Example 1 of the present invention, wherein A is an in situ hybridization (ISH) image from the Ivy GAP database showing the expression pattern of NDRG1 in PAN (left) and the expression pattern of EPAS1 in MVP (right) in GBM tissue; B is a representative multiple immunofluorescence image of NDRG1 (green), HIF2A (green), CD31 (pink), and CD44 (red) in GBM sections;

[0037] Figure 8 Figure 1 is a UMAP visualization of human GBM SRT cluster and cell type annotations in Example 1 of the present invention, where A is a UMAP plot of GBM spatial transcriptome data from 16 patients, with cells color-coded according to defined clusters; B is a UMAP plot of 16 patients, with cells color-coded according to patient origin; C is the proportion of each cell subset in the 16 patients; D is a heat map showing the expression of PAN (top) and MVP (bottom) regional characteristic genes in different cell types;

[0038] Figure 9 The different states of endothelial cells in the PAN and MVP regions in Example 1 of the present invention; wherein, A is the composition of different types of cells in the PAN region; B is the composition of different types of cells in the MVP region; C is the UMAP map of PAN glioma endothelial cells; D is the UMAP map of MVP glioma endothelial cells; E is the heat map of the top 10 differentially expressed genes in PAN and MVP related endothelial cells; F is the heat map of the top 10 enriched GO terms in different endothelial cells based on the differentially expressed genes between PAN (748) and MVP (746);

[0039] Figure 10Figure 1 is the differentiation of GSCs into endothelial cells in Example 1 of the present invention; wherein, A and B are violin plots showing the normalized expression levels of marker genes in different cell types; C is a dot plot showing intercellular ligand and receptor pairs between endothelial cells and other cell types; D is a spatial heat map of PTN and PTPRZ1 in patient UKF423; E is a t-SNE plot of 9 cell types identified in GBM; F is a pseudo-time analysis of single-cell RNA-seq data; G is a correlation heat map showing the relationship between pairs of infiltrating cell types in GBM;

[0040] Figure 11 This is the cell type identification of single-cell sequencing data in the GBM microenvironment in Example 1 of the present invention, where A is the UMAP cluster analysis of 9 cell types in GBM; B is the proportion of different cell types in patients. DETAILED DESCRIPTION

[0041] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as limiting the present invention, but rather as a more detailed description of certain aspects, features, and embodiments of the present invention.

[0042] It should be understood that the terms described herein are intended only to describe particular embodiments and are not intended to limit the present invention. In addition, for numerical ranges herein, it should be understood that each intermediate value between the upper and lower limits of the range is also specifically disclosed. Each smaller range between any intermediate value within a stated value or stated range and any other stated value or intermediate value within the stated range is also encompassed by the present invention. The upper and lower limits of these smaller ranges may be independently included or excluded within the scope.

[0043] Unless otherwise indicated, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art. Although only preferred methods and materials are described herein, any methods and materials similar or equivalent to those described herein may also be used in the practice or testing of the present invention. All documents mentioned in this specification are incorporated by reference to disclose and describe the methods and / or materials associated with the documents. In the event of any conflict with any incorporated document, the contents of this specification shall prevail.

[0044] It will be apparent to those skilled in the art that various modifications and variations may be made to the specific embodiments described herein without departing from the scope or spirit of the invention. Other embodiments will be apparent to those skilled in the art from the description of the invention. The description and examples are intended to be exemplary only.

[0045] The words “include,” “including,” “have,” “contain,” etc. used in this document are open-ended terms, meaning including but not limited to.

[0046] Example 1

[0047] To resolve the heterogeneity of MVP and PAN anatomy hidden in large-scale analyses and to explore unique molecular signatures associated with GBM malignancy, we conducted a comprehensive bioinformatics analysis combining tissue-specific transcriptional profiles, spatial transcriptome datasets, and single-cell transcriptome datasets. The specific methods are as follows:

[0048] (1) Data collection

[0049] We collected RNA-seq datasets from 41 primary human glioblastoma (GBM) samples from different anatomical regions from the Ivy Glioblastoma Atlas Project (Ivy-GAP). In addition, we obtained spatial transcriptomic (SRT) datasets from 16 GBM samples without IDH mutations from data provided by Ravi et al. and analyzed these data using Visium technology (10XGenomics) (https: / / doi.org / 10.5061 / dryad.h70rxwdmj).

[0050] The Smart-seq2 glioblastoma single-cell RNA sequencing (scRNA-seq) dataset (n = 28, GSE131928) was downloaded from the GEO database. The Cancer Genome Atlas (TCGA) GBM dataset (n = 155), including RNA sequencing and clinical data, was obtained from the UCSC Xena browser (https: / / xenabrowser.net / ). RNA sequencing data and clinical data were used for validation analysis of the PAN and MVP signature gene sets and survival.

[0051] (2) Differential gene expression analysis

[0052] Gene expression profiles of cell tumor (CT), pseudopalisading cells around necrosis (PAN), and microvascular proliferation (MVP) areas were extracted from the Ivy-GAP dataset. To normalize gene expression values, the counts in the Fragments PerKilobase Million (FPKM) format were log2(n+1) transformed using the dplyr package (R 1.0.10). Quality control and differential expression analysis (logFoldChange cutoff=1) of paired data between the three groups were performed using the limma package (R 3.54.0). Intersection queries were performed using the UPsetR package (R 1.4.0), and overlapping gene IDs were converted from SYMBOL format to ENTREZID format using the clusterProfiler package (R4.6.0). Gene ontology functional enrichment analysis was performed in biological processes (BP), molecular functions (MF), and cellular components (CC). KEGG enrichment analysis was also performed, and the results are shown in Figure 3. Figure 1 、 2 shown.

[0053] Figure 1 、 2 The results show that in the Ivy-GAP dataset, the present invention compares the characteristics of 270 laser microdissected GBM tissues, including 111 CT (cellular tumor), 22 HBV (hypervascular), 40 PAN, 28 MVP, 26 PNZ (perinecrotic zone), 24 IT (invasive tumor) and 19 LE (leading edge zone) ( Figure 1 (B) Pairwise comparisons of expression differences between perivascular regions (HBV and MVP) and CT, as well as between perinecrotic regions (PNZ and PAN) and CT were performed (>2-fold change, FDR < 0.05). As expected, PAN and PNZ showed similar expression due to their histological similarities, while HBV and MVP highly expressed genes that promote angiogenesis, such as IL-8, ADM, VEGFA, and NDRG1. Compared to other regions, HBV and MVP showed elevated expression of endothelial cell and stemness-related molecules, such as ESM1 and CRIP1. Furthermore, MVP also showed elevated expression of molecules involved in tumorigenesis and migration, such as HIGD1B and MMP9. Differentially expressed genes were more prominent in PAN and MVP than in PNZ and HBV.

[0054] The Venn diagram of differentially expressed genes showed that 312 genes were specifically upregulated in PAN and 1048 genes were upregulated in MVP ( Figure 2 ). Gene ontology and KEGG analysis of MVP differentially expressed genes showed that they were associated with amoebae migration, PI3K-Akt, and focal adhesion pathways ( Figure 1Middle C). PAN differentially expressed genes are enriched in hypoxia response and HIF-1 signaling ( Figure 1 Among them, 151 down-regulated genes overlapped in PAN and MVP, accounting for about one-third of the PAN down-regulated genes. These genes are involved in nervous system development and cell adhesion ( Figure 2 ), suggesting that cells in PAN and MVP are highly malignant and that reduced intercellular adhesion promotes permeability and invasion.

[0055] (3) Spatial transcriptomic analysis

[0056] The present invention follows the standard procedures recommended by Seurat (R 4.0) for spatial transcriptomics analysis. Based on the spatial transcriptomics dataset of GBM samples without IDH mutations, the transcriptome spatial data of 16 GBM patients were merged, and the filtering parameters included: the number of genes was between 500 and 8000, the number of transcripts was less than 30,000, and the proportion of mitochondrial genes was less than 30%. The gene expression at each position was normalized using the SCTransform function. Then, the present invention extracted the top 30 principal components from the highly variable genes and set the resolution parameter to 0.5 for clustering. Seurat and SpaGCN toolkits were also used to identify robust spatially variable genes (SVGs) in the 10X Visium dataset. The present invention integrated the points of the same cluster in each sample into pseudo blocks ( Figure 3 ), the results are shown in Figures 3 to 7 .

[0057] 1000 and 524 SVGs were identified by Seurat and SpaGCN, respectively. The overlap with PAN and MVP upregulated genes generated 63 PAN and 53 MVP candidate marker genes ( Figure 4 These genes were ranked by frequency in the patient's SVGs. The top genes for PAN included IGFBP5, MTIX, CHI3L1, NDRG1, and VEGF. The top genes for MVP included IGFBP, SPARC, IFITM3, IGFBP3, and A2M ( Figure 4 Most genes related to migration, invasion, stemness, and angiogenesis were consistent with the histological characteristics and functions of PAN and MVP. TCGA RNA sequencing data showed that most marker genes were more highly expressed in GBM than in low-grade glioma (LGG) ( Figure 5 In addition, high expression is associated with poor prognosis ( Figure 6 ), including IGFBP5, IGFBP3, IGFBP7, and CHI3L1. This suggests that spatial upregulation of these genes in PAN and MVP may promote the malignancy of GBM.

[0058] In order to clarify whether PAN and MVP marker genes have specific spatial expression patterns, the present invention further compared their spatial expression in GBM patients. The present invention selected UFK243 as a typical sample, which has obvious necrosis and vascular proliferation. The points of UFK243 were clustered into 6 regions by Seurat and SpaGCN. The present invention observed that APOC1 and NDRG1 expression increased in region 0, EPAS1 and VEGFA expression increased in region 1, and IGFBP7, CHI3L1 and TAGLN expression increased in region 5. PAN genes such as IGFBP5, MT1X, CHI3L1 and NDRG1 have similar spatial patterns, mainly concentrated in regions 0 and 5. At the same time, IGFBP7, IFITM3, A2M and EPAS1 are scattered in regions 1, 3 and 5. PAN and MVP genes also showed different spatial patterns in UFK243. Ivy GAP in situ hybridization further verified the increase in NDRG1 in PAN and the increase in EPAS1 in MVP ( Figure 7 Immunofluorescence also confirmed this difference, with high expression of NDRG1 in PAN and high expression of HIF2A in vascular-rich areas ( Figure 7 These results confirmed the accuracy of the obtained region-specific marker genes.

[0059] (4) Deconvolution analysis

[0060] The complex glioma microenvironment includes non-tumor cell types such as endothelial cells, pericytes, microglia, macrophages and fibroblasts. In order to fully characterize these cell types, the present invention uses the GBM scRNA-seq data set (spatial transcriptomics data set of GBM samples) of Bhaduri (Bhaduri A, DiLullo E, Jung D, et al. Outer Radial Glia-like Cancer Stem Cells Contribute to Heterogeneity of Glioblastoma. Cell Stem Cell.) people and Neftel (Neftel C, Laffy J, FilbinMG, et al. An Integrative Model of Cellular States, Plasticity, and Genetics for Glioblastoma. Cell) et al. as a reference data set. The scRNA-seq data set from GSE131928 was integrated using the “Harmony” method to obtain a merged spatial transcriptome data set. The integrated reference data set contains 40,791 cells, 22,346 genes and 20 cell types. Cell types with less than 1000 cells were filtered out. The integrated data were visualized by UMAP dimensionality reduction and unsupervised clustering ( Figure 8 Through SingleR annotation and manual proofreading, the present invention identified 9 major cell types: astrocytes, macrophages / microglia, neurons, oligodendrocytes, OPCs, fibroblasts, endothelial cells, T cells and adipocytes ( Figure 8 Middle C).

[0061] The merged spatial transcriptome dataset was used as input for subsequent deconvolution analysis. The transcriptome of each point was deconvolved using RCTD (v1.1.0, https: / / github.com / dmcable / spacexr) to identify the possible constituent cell types. The deconvolved proportions of individual cell types were visualized using the stdeconvolve and ggplot2 packages.

[0062] The module scores of PAN and MVP signature features also showed relatively independent distributions ( Figure 9 In A and B), it is shown that the endothelial cells in these regions have different expression patterns and functional states. In order to understand the heterogeneity of endothelial cells, the present invention conducted an unsupervised cluster analysis ( Figure 9Differential expression analysis showed that PAN-associated endothelial cells highly expressed CHI3L1, SNRPD3, CLU, LAMTOR4, and AQP1, whereas MVP-associated endothelial cells expressed MGP, FABP5, collagen, and modifying enzymes such as COL1A2, COL3A1, COL4A1, and COL4A2 ( Figure 9 Middle E). Functionally, MVP endothelial cells are particularly active in cell adhesion, angiogenesis, and cytoplasmic translation pathways. These insights reveal distinct roles for endothelial cells in the PAN and MVP regions of GBM, highlighting the complexity and diversity of the tumor microenvironment.

[0063] (5) Single-cell RNA sequencing data analysis

[0064] The scRNA-seq dataset (GSE131928) contains 6863 GBM cells and 1067 normal cells from 28 patients. The count table was analyzed using the Seurat 4.0 package, following the standard workflow and default settings. Subsequently, the copy number variation (CNV) in the scRNA-seq data was inferred using the InferCNV package (R 1.3.3), with cutoff = 1. The present invention then combines CNV classification with UMAP- and gene set-based classification to define GSCs and GBM-associated macrophages / microglia. Marker genes for macrophages, T cells, and oligodendrocytes are from Neftel et al. (GSE131928). In order to study the molecular interaction network between each cell subtype, the present invention uses the CellPhoneDB (R2.1.7) algorithm to infer intercellular communication in the GSC environment.

[0065] The above results establish the connection between PAN / MVP-related characteristic genes and different functional states of endothelial cells in GBM. Due to the lack of single-cell resolution of spatial transcriptomics (SRT), the present invention analyzed 5 GBM single-cell RNA sequencing (scRNA-seq) datasets, including 2 paired SRT datasets. Different cell types were identified by Seurat: T cells, microglia / macrophages, oligodendrocytes, astrocytes, monocytes, endothelial cells and B cells ( Figure 11 ). It is noteworthy that many PAN / MVP signature genes are highly or specifically expressed in endothelial cells, such as AKAP12. In addition, some genes such as MT1X, MT2A, SPARC, EPAS1, and THY1 are highly expressed in both GSCs and endothelial cells ( Figure 10 (A, B).

[0066] To explore intercellular communication, the present invention visualized the ligand-receptor pairs between each cell type. Multiple pairs involved endothelial cells, among which the PTN / PTPRZ1 ligand-receptor pair is a key regulator between GSCs and endothelial cells ( Figure 10 Middle C). Spatial heatmap shows that PTN and PTPRZ1 have similar localization ( Figure 10 Middle D). Pseudo-temporal analysis reveals the trajectory of transition from GSCs to endothelial cells ( Figure 10 Correlation analysis confirmed the strong correlation between GSCs and endothelial cells ( Figure 10 Middle G).

[0067] (6) Immunofluorescence

[0068] The present invention obtained clinical glioma paraffin sections from Wuhan Union Hospital. All samples were collected after signing the informed consent form in accordance with the requirements of the internal review and ethics committee of Wuhan Union Hospital. The experimental procedures were approved by the Ethics Committee of Wuhan Union Hospital. After dewaxing and rehydration, the sections were boiled in Tris-EDTA antigen retrieval buffer (pH = 8.0) for 10 minutes. Then, the sections were treated with 3% hydrogen peroxide for 10 minutes to bleach endogenous peroxidase. After blocking with donkey serum in PBS for 30 minutes, the sections were incubated with primary antibodies at 4°C overnight. Glioma sections were probed with primary antibodies to the following proteins: CD44 (eBioscience, San Diego, USA), HIF2A (Signalway Antibody, Greenbelt, USA), NDRG1 and CD31 (Cell Signaling Technology, MA, USA). The sections were then washed with PBS and incubated with DyLight 488 / Fluor™ Plus 647-conjugated donkey anti-rabbit, mouse, or rat secondary antibodies (Invitrogen) at room temperature in the dark for 2 hours, and then stained with DAPI for 10 minutes. After washing, the sections were mounted with anti-fluorescence quenching sealant and fluorescence was observed using an Olympus VS200 slide scanner.

[0069] This was confirmed by double immunofluorescence, which showed colocalization of the GSC marker CD44 with the endothelial cell marker CD31 ( Figure 7 B) This result is consistent with the evidence that GSCs differentiate into tumor vascular endothelium.

[0070] Ethical approval and informed consent: All glioma samples were collected after signed informed consent according to the requirements of the Internal Review and Ethics Committee of Wuhan Union Hospital. The experimental procedures were approved by the Ethics Committee of Wuhan Union Hospital.

[0071] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for analyzing molecular differences in endothelial cells in glioblastoma, characterized in that: The steps include: (1) Based on the Ivy-GAP dataset, we extracted gene expression profiles of glioblastoma tumor cells (CT), pseudo-palisading cells (PAN) around necrosis, and microvascular proliferation (MVP) areas, and performed differential gene expression analysis to identify PAN and MVP marker genes. (2) Based on the spatial transcriptomic dataset of GBM samples without IDH mutation, spatial transcriptomic analysis was performed to identify spatially variable genes in the pseudopalisading cells (PAN) and microvascular proliferation (MVP) areas surrounding necrosis, and to achieve spatial localization of PAN and MVP marker genes; (3) Using a spatial transcriptomic dataset of GBM samples, different cell types were identified through deconvolution analysis; (4) Single-cell RNA sequencing data analysis was performed based on the scRNA-seq dataset to obtain the gene expression of PAN and MVP marker genes in various cell types, and pseudo-time analysis was performed on all cell types; (5) Based on the heterogeneity of endothelial cells, unsupervised cluster analysis was performed to determine the different roles played by endothelial cells in the PAN and MVP areas of GBM; The Ivy-GAP dataset in step (1) is an RNA-seq dataset of different anatomical regions of primary human glioblastomas derived from the Ivy Glioblastoma Atlas Project; The steps of the differential gene expression analysis are as follows: converting RNA-seq datasets of different anatomical regions of primary human glioblastoma into log2(n+1) format, and then performing quality control and differential expression analysis on paired data of different anatomical regions using the limma package; performing intersection query using the UPsetR package, and converting overlapping gene IDs from SYMBOL format to ENTREZID format using the clusterProfiler package, and performing gene ontology functional enrichment analysis and KEGG enrichment analysis of biological processes, molecular functions and cellular components to obtain differentially expressed gene data of different anatomical regions of primary human glioblastoma, and determine PAN and MVP marker genes; The steps of spatial transcriptomics analysis in step (2) are as follows: The spatial transcriptomics dataset of GBM samples without IDH mutations was filtered, and gene expression at each location was normalized using the SCTransform function. Principal components were then extracted from highly variable genes, and resolution parameters were set for clustering. Seurat and the SpaGCN toolkit were used to identify robust spatially variable genes in the 10X Visium dataset. Seurat was used to integrate points within the same cluster within each sample into pseudo-blocks to achieve spatial localization of PAN and MVP marker genes. The filtering conditions are: the number of genes is between 500 and 8000, the number of transcripts is less than 30000, and the proportion of mitochondrial genes is less than 30%; the number of principal components extracted is 30, and the resolution parameter is set to 0.

5.

2. The method for analyzing molecular differences in glioblastoma according to claim 1, characterized in that: In step (3), the deconvolution analysis step includes: Using GBM spatial transcriptome data as a reference dataset, the "Harmony" method was used to integrate the scRNA-seq dataset from GSE131928 to obtain a merged spatial transcriptome dataset. The integrated data were visualized by UMAP dimensionality reduction and unsupervised clustering, and annotated and manually proofread by SingleR. Using RCTD, the merged spatial transcriptome dataset was used for deconvolution analysis to obtain the constituent cell types. The deconvolution ratios of each cell type were visualized using the Stdeconvolve and ggplot2 packages.

3. The method for analyzing molecular differences in glioblastoma according to claim 1, characterized in that: In step (4), the single-cell RNA sequencing data analysis step includes: The scRNA-seq datasets were analyzed for count tables using the Seurat 4.0 package, following the standard workflow and default settings. Copy number variation (CNV) in the scRNA-seq data was inferred using the InferCNV package with a cutoff of 1. CNV classification was then combined with UMAP- and gene set-based classification to define GSCs and GBM-associated macrophages / microglia. Intercellular communication in the GSC environment was inferred using the CellPhoneDB-R 2.1.7 algorithm, and pseudotime analysis was performed for all cell types.

Images