Single-cell transcriptomics reveals stage- and side-specificity of gene modules in colorectal cancer

Abstract

Background: An understanding of mechanisms underlying colorectal cancer (CRC) development and progression is yet to be fully elucidated. This study aims to employ network theoretic approaches to analyse single cell transcriptomic data from CRC to better characterize its progression and sided-ness.

Methods: We utilized a recently published single-cell RNA sequencing data (GEO-GSE178341) and parsed the cell X gene data by stage and side (right and left colon). Using Weighted Gene Co-expression Network Analysis (WGCNA), we identified gene modules with varying preservation levels (weak or strong) of network topology between early (pT1) and late stages (pT234), and between right and left colons. Spearman's rank correlation (ρ) was used to assess the similarity or dissimilarity in gene connectivity.

Results: Equalizing cell counts across different stages, we detected 13 modules for the early stage, two of which were non-preserved in late stages. Both non-preserved modules displayed distinct gene connectivity patterns between the early and late stages, characterized by low ρ values. One module predominately dealt with myeloid cells, with genes mostly enriched for cytokine-cytokine receptor interaction potentiallystimulating myeloid cells to participate in angiogenesis. The second module, representing a subset of epithelial cells, was mainly enriched for carbohydrate digestion and absorption, influencing the gut microenvironment through the breakdown of carbohydrates. In the comparison of left vs. right colons, two of 12 modules identified in the right colon were non-preserved in the left colon. One captured a small fraction of epithelial cells and was enriched for transcriptional misregulation in cancer, potentially impacting communication between epithelial cells and the tumor microenvironment. The other predominantly contained B cells with a crucial role in maintaining human gastrointestinal health and was enriched for B-cell receptor signalling pathway.

Conclusions: We identified modules with topological and functional differences specific to cell types between the early and late stages, and between the right and left colons. This study enhances the understanding of roles played by different cell types at different stages and sides, providing valuable insights for future studies focused on the diagnosis and treatment of CRC.

Keywords: WGCNA; early tumor stage; functional enrichment; late tumor stage; left colon; modularity; right colon; scRNA-seq.

Figure 1

Workflow for the approach used in our analysis

Figure 2

UMAP plot for all cells. A Before down-sampling. B After down-sampling late stage (pT234) cells. C After down-sampling right colon cells

Figure 3

Topological and functional analysis of the modules detected for the early stage (pT1). A Median rank of module preservation between the early (reference) and late stages (test). B Cellcoverage of various modules detected for the early stage. The modules are arranged in ascending order based on their sizes, from the smallest to the largest. The number after each module’s name represents the number of genes in that module. Non-preserved (or preserved) modules are outlined by a dashed (or solid) black rectangle. C-D Circle plot for top 10% unique edges of modules E1 and E2, respectively. Each yellow (or brown) edge represents the co-expression value between the two corresponding genes in the early stage (or late stages). The white (colored) circle near each gene represents the sum of weights of edges connected to that gene in the early (or late) stages. Circles are colored based on the z-score between late stages vs. the early stage, ranging from blue for z-score ≤ −2, white for z-score = 0, to red for z-score ≥ 2. E-F Top 10 enriched KEGG pathways (based on p-values) for modules E1 and E2, respectively.

Figure 4

Topological and functional analysis of the modules detected for the right colon. A Median rank of module preservation between the right (reference) and left colon (test). B Cell coverage of different modules detected for the right colon. The modules are arranged in ascending order based on their sizes, from the smallest to the largest. The number after each module’s name represents the number of genes in that module. Non-preserved (or preserved) modules are highlighted by a dashed (or solid) blue rectangle around them. C-D Circle plot for top 10% unique edges of modules R1 and R3, respectively. Each green (or crimson red) edge represents the co-expression value between the two corresponding genes in the right colon (or left colon). The left (or right) circle near each gene represents the sum of weights of edges connected to that gene in the right (or left) colon. Circles are colored based on the z-score between the left and right colon, ranging from blue for z-score ≤ −2, white for z-score = 0, to red for z-score ≥ 2. E-F Top 10 enriched KEGG pathways (based on p-values) for modules R1 and R3, respectively.