Combined Single-Cell and Spatial Transcriptome Analysis Reveals Cellular Heterogeneity of the Hedgehog Pathway in Gastric Cancer

Academic Background

Gastric cancer (GC) is one of the most common and deadly malignancies worldwide, with high incidence and mortality rates. Despite advancements in chemotherapy, radiotherapy, and targeted therapy, the treatment of GC remains a significant challenge. The invasive and heterogeneous nature of GC makes it a difficult-to-treat cancer, particularly for patients with advanced stages, who have a very low survival rate. The invasion and metastasis of tumor cells are the main causes of recurrence and death, and existing treatments cannot fully address these issues. Immunotherapy, as a promising treatment modality, also faces challenges due to the complexity of the tumor microenvironment and immune system. Therefore, a comprehensive understanding of the pathogenesis of GC, especially at the genetic, molecular, and phenotypic levels, is crucial for better disease management and reducing the burden on patients.

The Hedgehog (Hh) pathway plays an important role in both embryonic development and adult tissues, participating in cell proliferation, differentiation, and tissue morphogenesis. In cancer, abnormal activation of the Hh pathway is closely associated with the development and progression of various tumors. Particularly in GC, abnormal activation of the Hh pathway is closely related to tumor proliferation, invasion, and metastasis. Additionally, the Hh pathway influences tumor treatment and prognosis by regulating the characteristics of cancer stem cells. Although therapeutic strategies targeting the Hh pathway have become a focus in cancer treatment, the Hh pathway as a therapeutic target for GC still requires further research and validation.

Single-cell sequencing technology has played a significant role in cancer research, revealing tumor diversity, evolutionary patterns, treatment resistance, and treatment responsiveness. However, single-cell sequencing cannot preserve spatial context information. Spatial transcriptomics (ST), as an emerging technology, addresses this limitation by revealing the spatial distribution and interactions of cells within tissues or tumors. By combining RNA sequencing data with spatial information, ST provides detailed insights into cell types, locations, and interactions within tissues. This study combines single-cell sequencing and spatial transcriptome data to deeply explore the potential molecular mechanisms of the Hh pathway in GC, aiming to provide a deeper understanding of tumor heterogeneity in GC and identify new therapeutic strategies.

Source of the Paper

This paper was co-authored by Guoliang Zhang, Guojun Xia, Chunxu Zhang, Shaodong Li, Huangen Wang, and Difeng Zheng, all from the Department of General Surgery, Central Hospital of Shaoxing, Zhejiang, China. The paper was published online on September 9, 2024, in the journal Genes & Immunity, with the DOI: https://doi.org/10.1038/s41435-024-00297-0.

Research Process and Results

Data Acquisition

This study utilized three types of GC data: bulk transcriptomic data, single-cell sequencing data, and spatial transcriptomic data. The bulk transcriptomic dataset and associated clinical information were obtained from The Cancer Genome Atlas (TCGA) database, covering data from 407 GC patients. The single-cell sequencing dataset was sourced from the Gene Expression Omnibus (GEO) database, specifically from entries GSE163558 and GSE184198, which included 3 GC samples and 2 paired normal samples. Spatial transcriptomic data were acquired from the CROST database (ID: visdp000078), which integrates 1033 spatial transcriptomic samples from 8 species using 5 technological platforms: 10x Visium, Slide-seqV2, MERFISH, 10x Xenium, and NanoString COO.

Single-Cell Sequencing Analysis

In the analysis of single-cell sequencing data, all samples were sequenced using the 10x single-cell sequencing method. Subsequently, the “Seurat” R package was used to generate Seurat objects, integrate all samples, and perform data filtering and correction. Quality control measures excluded cells with gene counts below 200 or above 6000, as well as cells with a mitochondrial gene percentage (pctMT) exceeding 15%. The SeuratNormalizeData function was used to normalize the expression profile of each cell, and the logNormalize method was applied for normalization. Principal component analysis (PCA) was then performed on the top 2000 genes, and the first 15 significant principal components were selected for subsequent analysis. Uniform Manifold Approximation and Projection (UMAP) was used to visualize cell distribution. Through manual annotation and reference to marker genes from previous studies, cells were classified into 10 populations, including T cells, B cells, monocytes, plasma cells, mast cells, epithelial cells, macrophages, endothelial cells, fibroblasts, and smooth muscle cells.

Hedgehog Pathway-Related Gene Set Scoring

To evaluate the expression of Hh pathway-related genes in the GC tumor microenvironment, this study employed five scoring methods: AUCell, UCells, singscore, GSVA, and AddModuleScore. By integrating the results from these five methods, normalization and standardization were performed to generate a comprehensive score for subsequent analysis. The results showed that Hh pathway-related genes were highly expressed in epithelial cells, fibroblasts, smooth muscle cells, and macrophages, with significantly higher expression in tumor-associated fibroblasts compared to normal samples.

Fibroblast Developmental Trajectory Analysis

Based on the Hh score, this study selected the fibroblast population for further analysis. Using the Monocle software package, the developmental trajectory of fibroblasts was constructed. It was found that Cluster 24 fibroblasts were in the early and middle stages of development, while Cluster 21 fibroblasts were in the late stage. Since Cluster 21 fibroblasts were derived from tumor samples, this finding suggests that normal fibroblasts may transform into tumor-associated fibroblasts. Further analysis revealed that the expression of Wnt2, Gli3, CCND1, and Hip1 genes gradually increased during fibroblast development, indicating that these genes may play an important role in the transformation of normal fibroblasts into tumor-associated fibroblasts.

Cell Interaction Analysis

To further explore the role of Hh in GC fibroblasts, this study divided fibroblasts into Hh high-expressing fibroblasts (Hh_high_fib) and Hh low-expressing fibroblasts (Hh_low_fib) and performed cell-cell communication analysis using the CellChat software package. The results showed that Hh_high_fib directly interacted with other cell populations, particularly as signal senders to epithelial cells and endothelial cells. These findings suggest that Hh-associated epithelial cells play a critical role in GC tumors.

Spatial Transcriptome Analysis

In the analysis of spatial transcriptome data, this study used the 10x spatial transcriptome method to process the data and performed normalization and batch effect removal using the Seurat R package. Through dimensionality reduction clustering and deconvolution analysis, nine distinct cell types were identified, including endothelial cells, epithelial cells, Hh_high_fib, Hh_low_fib, macrophages, mast cells, plasma cells, smooth muscle cells, and T cells. Correlation analysis revealed that Hh high-expressing fibroblasts were significantly positively correlated with smooth muscle cells and negatively correlated with Hh low-expressing fibroblasts, indicating that Hh expression plays an important role in the development and differentiation of fibroblasts.

Role of CCND1 Fibroblasts in GC

Based on the important role of CCND1 in fibroblast development, this study divided fibroblasts into CCND1-positive fibroblasts (CCND1+ fib) and CCND1-negative fibroblasts (CCND1- fib). Through spatial transcriptome data analysis, it was found that CCND1-positive fibroblasts were highly infiltrated in Clusters 1, 5, and 6. Further analysis revealed that Cluster 1 cells were in the early stage of development, while Clusters 5 and 6 cells were in the late stage, and CCND1 expression gradually increased during the middle and late stages of development, suggesting that it may play an important role in the middle and late stages of tumor development.

Bulk Transcriptome Survival Analysis

By integrating bulk transcriptome data, this study evaluated the impact of CCND1+ fibroblast infiltration on the prognosis of GC patients. The results showed that CCND1 levels were significantly higher in GC tumor tissues compared to normal tissues, and patients with high CCND1+ fibroblast infiltration had a poorer prognosis. Additionally, CCND1+ fibroblasts were positively correlated with tumor immune escape capability, suggesting that they may influence the efficacy of immunotherapy.

Conclusion

This study combined single-cell sequencing and spatial transcriptome data to deeply explore the role of the Hh pathway in GC. Based on bulk transcriptome data, it was verified that high infiltration of CCND1+ fibroblasts is a risk factor for GC patients and may affect the outcomes of immunotherapy and chemotherapy. The study provides unique insights into GC and the Hh pathway and offers new directions for cancer treatment strategies.

Research Highlights

1. Key Findings: The Hh pathway is highly expressed in GC fibroblasts, and high infiltration of CCND1+ fibroblasts is a risk factor for GC patients.
2. Methodological Innovation: The combination of single-cell sequencing and spatial transcriptome data comprehensively reveals the cellular heterogeneity and interactions within the GC tumor microenvironment.
3. Application Value: The study provides new directions for GC treatment strategies, particularly targeting the Hh pathway and CCND1.

Future Research Directions

Future research could expand to multi-center data and diverse patient populations to enhance the generalizability and robustness of the results. Additionally, integrating other omics technologies to further explore the molecular mechanisms of GC progression and treatment resistance will help develop more effective personalized treatment strategies.