Comprehensive Transcriptomic Analysis Reveals Cell-Type-Specific Roles of Human Odorant Receptors in Glioblastoma and the Tumor Microenvironment Authors: Hee Jin Cho, Dong Jun Yeo, HeeWoong Yang, JaeHyung Koo, Diego Guidolin, Maria A Mariggiò Published: 2024 ABSTRACT AbstractOdorant receptors (ORs), which constitute approximately 50% of all human G protein-coupled receptors, are increasingly recognized for their diverse roles beyond odor perception, including functions in various pathological conditions like brain diseases and cancers. However, the roles of ORs in glioblastoma (GBM), the most aggressive primary brain tumor with a median survival of only 15 months, remain largely unexplored. Here, we performed an integrated transcriptomic analysis combining The Cancer Genome Atlas RNA-seq and single-cell RNA sequencing data from GBM patients to uncover cell-type-specific roles of ORs within the tumor and its microenvironment. Our findings reveal that ORs display distinct expression patterns, with OR51E1 enriched in pericytes linked to vascular remodeling and angiogenesis, OR2B11 associated with tumor-associated macrophages supporting immunosuppressive phenotypes, and OR2L13 correlated with synaptic activity in recurrent tumors, potentially mediating treatment-induced neuronal adaptations. These results highlight ORs as potential therapeutic targets, offering new insights into their regulatory roles in GBM progression, immune modulation, and treatment resistance.Keywords: glioblastoma (GBM), odorant receptors (ORs), G protein-coupled receptors (GPCRs), tumor microenvironment (TME), single-cell RNA sequencing, vascular remodeling, angiogenesis, tumor-associated macrophages (TAMs), synaptic adaptation, therapeutic targets 1. Introduction Glioblastoma (GBM) remains one of the most lethal cancers, with a median survival of only 12–15 months and a recurrence rate of about 90% within two years [1,2]. The current standard treatment, which involves maximal surgical resection followed by concurrent chemoradiation with temozolomide, has shown limited effectiveness in improving patient outcomes. Despite extensive research efforts to enhance clinical outcomes, treatment resistance remains due to several key challenges. GBM exhibits remarkable inter- and intra-tumoral heterogeneity, leading to genomic and cellular evolution under therapeutic pressure [3,4,5,6]. Additionally, the plasticity of GBM’s malignant states enables tumor cells to evade treatment through state transitions [7]. The complex GBM microenvironment also plays crucial roles in disease progression and treatment resistance, characterized by hypoxia, high infiltration of tumor-associated microglia/macrophages (TAMs), and dynamic interactions with brain stromal cells [8,9,10,11]. Specifically, interactions between TAMs and glioma cells contribute to an immunosuppressive microenvironment that fosters tumor growth and facilitates resistance mechanisms [12]. These features collectively promote irregular vessel development, an immunosuppressive microenvironment with mesenchymal transition, and treatment-induced synaptic plasticity [4,8,9,13]. Furthermore, the blood–brain barrier (BBB) poses additional challenges by limiting drug delivery to tumor cells [14]. Therefore, understanding and modulating the tumor microenvironment (TME) has emerged as a key strategy to overcome current therapeutic limitations. Precision medicine has revolutionized cancer treatment by integrating advanced diagnostic and therapeutic strategies tailored to individual patients. Among these, theranostic nanosystems, which combine diagnostic and therapeutic capabilities, have emerged as a promising approach for treating aggressive cancers. These nanosystems are particularly advantageous in neuro-oncology as they can penetrate biological barriers like the BBB due to their optimized structure and solubility characteristics. Recent studies have demonstrated that theranostic approaches enhance diagnostic precision while reducing systemic side effects through targeted delivery mechanisms [15]. The integration of omics technologies has further expanded the potential of these systems, enabling the identification of novel theranostic targets and the development of personalized treatment strategies. In this context, odorant receptors (ORs), the largest family of G protein-coupled receptors (GPCRs), have emerged as potential theranostic targets. While best known for their role in olfactory signal transduction, emerging evidence reveals that they are widely expressed outside the olfactory system [16,17]. Their non-canonical functions have been highlighted in various diseases, including brain disorders and cancer [16,17,18,19,20,21,22,23,24]. Several studies have demonstrated their potential as therapeutic targets and prognostic biomarkers in cancer, with OR51E2, also known as the prostate-specific G-protein-coupled receptor, and OR2B6 showing overexpression in prostate and breast carcinomas, respectively [25,26]. OR51E2 overexpression has been linked to cancer cell proliferation and migration, while OR2B6 overexpression in breast carcinoma contributes to invasive behavior and metastatic potential [26,27]. In immune regulation, Olfr78, the mouse ortholog of OR51E2, partners with a GPCR to influence macrophage polarization toward a protumoral M2 phenotype [27]. Beyond cancer and immune regulation, OR dysregulation in brain stromal cells, including microglia, astrocytes, and neurons—central players in neuroinflammatory and degenerative processes—links OR function to mechanisms underlying brain disorders, highlighting their potential as therapeutic targets [21,23,28,29]. While our previous systematic analysis of adult gliomas using The Cancer Genome Atlas (TCGA) bulk RNA-seq data suggested potential therapeutic roles for ORs, the specific functions of ORs within distinct cell types of the TME remained unclear due to the limitations of bulk sequencing approaches that average signals across heterogeneous cell populations [18,19]. Given the complex cellular composition of GBM and the emerging roles of ORs in both tumor cells and the TME, a comprehensive analysis focusing on cell-type specific OR functions and their interactions with the TME is crucial. Here, we performed an integrated transcriptomic analysis by combining TCGA RNA-seq data with expanded single-cell RNA sequencing of GBM patients to identify the cell-type specific roles of ORs and their biological implications. This approach reveals previously undetected interactions between ORs and the TME that were not evident in bulk RNA-seq analyses. It also offers valuable insights into potential therapeutic strategies targeting GBM’s cellular heterogeneity and complex tumor–TME interactions. 2. Results 2.1. Comprehensive Analysis Reveals Distinct OR Expression Patterns in GBM 2.1.1. Distinct OR Expression Patterns in GBM Tissues Principal component analysis (PCA) of OR gene expression in TCGA GBM (IDH wildtype), TCGA lower grade glioma (LGG, IDH mutant), and Genotype-Tissue Expression (GTEx) cerebral cortex samples revealed distinct clustering patterns, demonstrating unique OR expression patterns in GBM compared to LGG and normal brain tissues (Figure 1A). To better understand the relationship between PCA clusters and OR expression patterns, we performed differential expression gene (DEG) analysis. The analysis revealed that the GBM-specific cluster was characterized by 46 up-regulated and 92 down-regulated ORs compared to normal brain tissue (Table S1). When compared to LGG samples, this cluster showed elevated expression of 10 ORs and decreased expression of 12 ORs. Notably, among these dysregulated ORs, eight up-regulated and seven down-regulated ORs maintained their expression patterns in both comparisons (versus normal brain and LGG), suggesting their potential involvement in GBM-specific processes (Figure 1B,C, Table S1). Visualization of OR expression levels on the PCA plot confirmed that these top up-regulated ORs showed enriched expression specifically in the GBM cluster (Figure S1). Interestingly, several of these up-regulated ORs have been previously reported in other cancers. For instance, OR2B6 overexpression has been reported in breast cancer, and OR51E1 has been found to be elevated in advanced prostate cancer and other solid tumors, consistent with our findings [18,26,30,31,32]. The persistent dysregulation of certain ORs, such as OR2B6 and OR51E1, in comparisons with both normal and LGG tissues suggests that these ORs may be involved in critical GBM-specific processes, potentially contributing to malignancy or treatment resistance. Analysis of OR family distribution revealed that OR family 7 was the most enriched (21%) in up-regulated ORs, while OR family 2 comprised the largest portion (28%) of down-regulated ORs (Figure 1D). Interestingly, a significant proportion of dysregulated ORs were pseudogenes, with 32% of up-regulated and 52% of down-regulated ORs in this category, indicating that both functional OR genes and pseudogenes may contribute to GBM pathogenesis and its microenvironment (Table S1). Distinct OR expression patterns in GBM compared to normal brain and LGG. (A) PCA of OR gene expression using TCGA and GTEx RNA-seq data. PCA was performed on the expression profiles of 847 OR genes. PCA plot including GBM, LGG (TCGA), and normal cortex (GTEx) samples (left panel). PCA plot showing the distribution of TCGA samples only (right panel). Each dot represents an individual sample, and samples from TCGA-GBM, TCGA-LGG, and GTEx normal brain are indicated in red, blue, and green, respectively. (B) DEG analysis of OR in GBM. Volcano plots showing differentially expressed OR genes filtered from genome-wide differential expression analysis. GBM versus normal cortex samples (left panel). GBM versus LGG samples (right panel). For clarity, only non-overlapping OR gene labels are displayed. (C) Venn diagrams illustrating the overlap of differentially expressed OR genes between GBM versus normal cortex and GBM versus LGG comparisons. The diagram shows OR genes that are up-regulated (left panel) and down-regulated (right panel) in GBM. (D) Bar plot showing the percentage of dysregulated OR genes within each OR family (calculated as the number of dysregulated OR genes divided by the total number of OR genes per family). Comparison of up- (pink) and down-regulated (green) OR percentages across 14 detected OR families. 2.1.2. Distinct OR Expression Patterns in Neoplastic and Non-Neoplastic Cells of GBM To delineate OR expression patterns at the single-cell level, we analyzed combined GBM single-cell RNA sequencing (scRNA-seq) datasets [33], comprising 1.1 million cells from 240 patients across 26 independent cohorts. The cells were hierarchically classified across three levels: first by neoplastic (tumor-derived) versus non-neoplastic (microenvironment-derived) populations (level 1), followed by major cell types (level 2), and finally by specific cell subtypes (level 3) (Figure 2A). Consistent with our TCGA GBM analysis, OR family 7 and 2 showed the highest enrichment in neoplastic and non-neoplastic cells, respectively, despite limited OR expression across cell populations (Figure 1D and Figure 2B). DEG analysis revealed that OR4N2 was up-regulated in neoplastic cells as well as GBM tissues, suggesting its potential tumor-intrinsic function (Figure 2C and Table S1). Conversely, OR2B11, OR52K1, and OR3A2 were up-regulated in GBM tissues and non-neoplastic cells, implying their roles in modulating the TME to promote tumorigenesis (Figure 2C and Table S1). Furthermore, cell-type-specific analysis (annotated according to the reference methodology [33]) indicated that OR4N2 was expressed in neoplastic cells, while other genes, such as OR2B11, were specifically expressed in TME cells such as TAMs (Figure 2D and Table S2). The specific expression of OR4N2 in neoplastic cells suggests a possible role in tumor cell survival or proliferation, whereas the up-regulation of OR2B11 in TAMs points to a role in shaping the TME to support tumor growth. The cell-type specificity of OR expression was further validated through correlation analysis between OR expression levels and cell-type-specific gene set scores in TCGA GBM tissues (Figure S2). Cell-type-specific OR expression patterns in GBM at single-cell resolution. (A) Uniform manifold approximation and projection (UMAP) visualization of scRNA-seq data showing the distribution of cell populations in GBM. Each cell (dot) is colored by cell identity at three annotation levels (left: level 1, middle: level 2, right: level 3), following the reference publication annotation scheme [33]. (B) OR family distribution analysis showing the percentage of OR-positive cells across cell populations, demonstrating predominant enrichment of OR family 7 in neoplastic cells and OR family 2 in non-neoplastic cells. Cells with positive expression of any OR in each OR family were designated as OR-positive. (C) Differential expression analysis of OR genes between neoplastic and non-neoplastic cells. Colored dots represent differently expressed ORs (|log2 fold change| > 1, adjusted p-value < 0.05). (D) Cell-type-specific differentially expressed OR genes based on annotation level 3 (adjusted p-value < 0.05). Statistical significance was assessed by Wilcoxon rank-sum testing with Benjamini–Hochberg correction. 2.2. Heterogeneous OR Expression Across Different GBM Malignant States To examine the heterogeneity of OR expression within tumor cells, we analyzed OR expression patterns across four distinct GBM neoplastic cell states. Among the neoplastic-up-regulated OR genes, OR11A1 showed uniformly weak expression, while OR7D2 exhibited significant expression across all malignant states (Figure 3A,B, Table S2). In contrast, OR4N2 and OR7E14P demonstrated state-specific patterns, with predominantly enriched in neural precursor cell (NPC)/oligodendrocyte precursor cell (OPC)-like and mesenchymal (MES)-like states, respectively (Figure 3C,D). These findings highlighted the distinct regulation of ORs in different GBM cellular states. Despite minor discrepancies due to cellular plasticity and intra-tumoral heterogeneity, where multiple malignant states coexist within bulk tissue samples [7], validation using TCGA GBM bulk tissue data corroborated these patterns. OR4N2 exhibited significantly higher expression in tumors with NPC-like features, while OR7E14P showed elevated expression in tumors with MES-like characteristics compared to other tumors (Figure 3E,F). Neoplastic cell-expressing OR genes in GBM. (A–D) UMAP visualization of scRNA-seq data showing OR expression distribution across neoplastic and non-neoplastic cells. The analysis demonstrates expression patterns of OR11A1 (A), OR7D2 (B), OR4N2 (C), and OR7E14P (D). (E) Heatmap depicting state-specific OR gene expression in TCGA-GBM data. DEG analysis identified up-regulated OR genes in each GBM state-enriched group (OPC-, NPC-, MES-, astrocyte (AC)-like) compared to other groups. The analysis reveals OR genes were specifically up-regulated in MES- and NPC-like groups, while no significant OR genes were detected in OPC- and AC-like groups. (F) Box plots showing differential expression of OR4N2 (left panel) and OR7E14P (right panel) across GBM state-enriched groups, with statistical significance determined by Student’s t-test. Given emerging evidence suggesting that pseudogenes can function as regulatory elements in cancer, we performed weighted correlation network analysis (WGCNA) using TCGA-GBM data (Figure S3). This analysis identified 27 distinct modules, with OR7E14P showing significant correlation with the pink module (gene significance = 0.61, p = 1 × 1017 (Figure S3C). Within this module, we identified 16 genes that were highly connected to OR7E14P. Gene ontology analysis of these co-expressed genes revealed significant enrichment in cytoskeleton-related pathways, particularly in cilium and microtubule bundle formation (Figure S3D,E). These findings are particularly relevant as cytoskeletal reorganization is a key feature of proneural to mesenchymal transition in GBM [34], consistent with our cell type-specific analysis showing OR7E14P enrichment in the MES-like state. Although these state-associated OR genes showed no significant association with clinical outcomes (Figure S4), these distinct expression patterns suggest that ORs may play specific roles in the unique biology of each malignant state, potentially influencing pathways linked to tumor plasticity and adaptive behaviors. 2.3. Pericyte-Enriched OR51E1 Promotes Vascular Remodeling and Angiogenesis in GBM Cell-type-specific OR expression analysis identified vascular-associated OR genes, including OR51E1, in endothelial cells and pericytes (mural cells) (Figure 2D and Figure 4A). Sub-clustering analysis of vascular cells using scRNA-seq data revealed 10 distinct clusters (Figure 4B), with clear separation between mural and endothelial cell populations (Figure 4C). Clusters 0, 5, and 6 were enriched for endothelial cells, while clusters 1, 3, 4, 7, 8, and 9 were dominated by mural cells, with cluster 2 showing mixed populations. Among vascular-associated ORs, OR51E1, OR51E2, OR3A2, and OR7E38P demonstrated cluster-specific expression patterns (Figure 4D,E, Table S3). Notably, OR51E1 exhibited prominent expression in clusters 1, 4, and 9, which were characterized by elevated levels of pericyte markers RGS5 and PDGFRB (Figure 4F). VEGFA, a key regulator of vascular remodeling and angiogenesis [35], was predominantly expressed in GBM neoplastic cells rather than vascular cells (Figure S5), suggesting potential paracrine signaling between VEGFA-secreting tumor cells and OR51E1-positive pericytes. Analysis of the TCGA GBM dataset revealed OR51E1’s strong correlations with vascular cell populations as well; correlation analysis with cell type-specific gene sets demonstrated significant positive associations with both mural (r = 0.63) and endothelial cells (r = 0.4), with mural cells showing stronger correlation (Figure S6). Consistently, OR51E1 expression showed significant positive correlations with gene sets of blood vessel remodeling and angiogenesis (Figure 4G). Furthermore, OR51E1-high GBM tissues showed significantly elevated expression of RGS5, PDGFRB, and VEGFA compared to OR51E1-low samples (Figure 4H). The anatomical analysis using the Ivy Glioblastoma Atlas dataset further substantiated these findings, demonstrating significant OR51E1 up-regulation in regions of microvascular proliferation and hyperplastic blood vessels (Figure 4I). Taken together, these results suggest that OR51E1 plays a crucial role in GBM vasculature, particularly in pericyte-mediated vascular remodeling and tumor angiogenesis, potentially influencing tumor progression and therapeutic resistance. OR51E1 contributes to vascular remodeling and angiogenesis in GBM. (A) UMAP visualization of scRNA-seq data showing the OR51E1 expression distribution across neoplastic and non-neoplastic cells, with notable expression in vascular regions. (B) Sub-clustering analysis of 11,974 vascular cells. Each cell (dot) is colored by cell type (left panel) and assigned Leiden cluster (right panel). (C) Bar plot showing the proportion of each cell type within individual Leiden clusters. (D) DEG analysis of OR genes across Leiden clusters (adjusted p-value < 0.05). (E) UMAP visualization demonstrating OR51E1 expression distribution across vascular cells, with prominent expression in clusters 1, 4, and 9. (F) Violin plots showing expression levels of pericyte markers and angiogenic factors: RGS5 (left), PDGFRB (middle), and VEGFA (right panel). (G) Dot plot showing the results of correlation analysis between OR51E1 expression and angiogenesis-associated gene sets from the Molecular Signatures Database (MSigDB) in TCGA-GBM. **** p < 0.0001; *** p < 0.001; ** p < 0.01; * p < 0.05. (H) Box plots showing differential expression of PDGFRB, RGS5, and VEGFA between OR51E1-high (>median) and -low (