Enhancing fibroblast-epithelial cell communications: Serpine2 as a key molecule in Fusobacterium nucleatum-promoted colon cancer

Abstract

Background: Fusobacterium nucleatum (Fn) has been identified as a causative factor in the progression of colon cancer. This study aims to integrate bulk RNA-seq with single-cell RNA-seq (scRNA-seq) to elucidate the molecular mechanisms by which Fn facilitates colon cancer progression.

Methods: The scRNA-seq data from tumor tissues of Fn intervention were analyzed to screen cells with significant proportion changes. Differentially expressed genes of cells with different proportions were extracted and intersected with those identified in the bulk RNA-seq analysis. Three machine learning algorithms were employed to identify characteristic genes. Clinical tissue samples and external datasets, along with in vitro co-culture experiments, were utilized to validate these findings.

Results: Following Fn intervention, there was an observed increase in the fibroblast iso-cellular ratio and interaction levels. Utilizing machine learning algorithms, we identified five key genes. The differential expression of Serpine2 was validated using clinical samples and external datasets. Furthermore, patients with metastatic colon cancer exhibited significantly higher Serpine2 expression compared to those without metastasis. Fn was found to significantly enhance the expression of Serpine2 in fibroblasts and to promote the proliferation and migration capabilities of tumor cells.

Conclusion: This study elucidates the role of Fn in promoting colon cancer progression through the enhancement of fibro-macrophage-epithelial cell interactions. Furthermore, Serpine2 has been identified as a potential molecular marker associated with Fn-mediated colon cancer progression and metastasis. These findings contribute novel insights that may inform the development of therapeutic strategies for colon cancer.

Keywords: Fusobacterium nucleatum; SerpinE2; colon cancer; fibroblast; single cell RNA-seq.

Figure 1

The work roadmap.

Figure 2

Fn promotes colon cancer progression/metastasis, which is closely related to metal ion signaling (GSE122183). (A) Difference analysis volcano plot. (B) Clustering heat map. (C) GO enrichment analysis histogram. (D) KEGG enrichment analysis bubble plot.

Figure 3

Single-cell transcriptome reveals a significant increase in the proportion of fibroblasts after Fn intervention (GSE172334). (A, B) Umap plots of cell clustering. (C) Maker genes of different cell clusters. (D, E) Comparison of cell proportions in different samples and between groups.

Figure 4

Identified differential genes between fibroblasts and dendritic cells (GSE172334). (A) The number of fibroblast–epithelial cell communication was significantly higher after Fn intervention. (B) The number and intensity of cell communication. (C) Volcano plots of fibroblast, macrophage, and dendritic cell differential analysis. (D, E) Extracted upregulated and downregulated differential genes to take the intersection, respectively.

Figure 5

Machine learning identifies five feature genes (GSE122183): JUNB, SERPINE2, NR2F2, LTBP4, and MMP7. (A) Lasso regression. The panel shows a plot of the mean squared error in Lasso regression as a function of Log(λ), thus helping us to select the best model. The upper horizontal coordinate, showing the number of variables required for the corresponding model, decreases from left to right; the lower horizontal coordinate, the logarithm of the penalty coefficient λ; and the vertical coordinate indicates that the mean squared error (MSE) refers to the degree to which the predicted value of the computational model differs from the true value. (B) The model error rate remains stable after the random forest decision tree reaches 200. (C) Random forest screening of importance variables. (D) The blue box-and-line plot on the left of Boruta algorithm corresponds to the minimum, average, and maximum Z scores for a shadowed feature. (E) The green line corresponds to a confirmed feature, the red color indicates a rejected feature, the yellow color indicates a feature to be determined, and the blue color indicates the importance of a minimal, average, and maximal shadowed feature, respectively. (F) The three algorithms take the intersection to obtain the five key feature genes.

Figure 6

Fn co-culture significantly increased Serpine2 gene expression in different tumor cells. (A, B) Sequentially, the expression box plots of 5 featured genes were plotted using two external datasets (A, GSE173549, LoVo cells; B, GSE191257, HT29 cells). (C) Expression histograms of five featured genes were plotted using the training dataset. (D) The expression level of JUNB was detected in 60 clinical samples by using q-PCR, Serpine2, ltbp4, and mmp7E mRNA levels. (E) Western blotting was used to assess the expression levels of Serpine2 protein in 10 clinical samples. "*" means p < 0.05, "**" means p < 0.01, "***" means p < 0.001, and "****" means p < 0.0001.

Figure 7

Serpine2 was positively correlated with pathological stage of colon cancer. (A, B) In the TCGA dataset, Serpine2 was not observed to be correlated with OS and DFS outcomes of patients. (C) The TCGA dataset showed a positive correlation between Serpine2 mRNA expression and pathological stage. (D) In our clinical samples, correlation was observed between Serpine2 mRNA expression and pathological stage (p = 0.092). (E) In Fn+ colon cancer tumor tissues, Serpine2 mRNA expression was higher in metastatic samples (M) than in non-metastatic samples (NM). (F) In Fn+ metastatic colon cancer patients, Serpine2 mRNA expression was higher in tumor samples than in paired normal tissue (paired t-test). "*" means p < 0.05 and "**" means p < 0.01.

Figure 8

Analysis of Serpine2 localization and regulatory network. (A–C) Single-cell analysis suggested that Serpine2 was mainly localized on fibroblasts and the Serpine2 mRNA expression level was higher in the Fn group than that in the PBS group (GSE172334). (D) Constructed transcription factor–miRNA–gene regulatory network.

Figure 9

The supernatant derived from the co-culture of Fn and CCD-18Co cells significantly promotes the growth and migratory capabilities of tumor cells. (A) The schematic representation of the co-culture setup involving Fn and CCD-18Co cells. (B) The CCK-8 assay to evaluate the impact of the Fn-fibroblast co-culture supernatant on the proliferation of HCT116 cells, where MOI of 0 indicates the use of supernatant from fibroblasts cultured with uninfected Fn, whereas MOIs of 100 and 200 correspond to fibroblasts cultured with varying levels of Fn infection. (C) The scratch assay to assess the influence of the Fn-fibroblast co-culture supernatant on the migratory capacity of HCT116 cells. (D) The Western blot analysis, indicating a significant increase in the protein levels of Serpine2 in fibroblasts following Fn infection. Data are expressed as the mean ± standard deviation (SD), and statistical significance was evaluated using one-way ANOVA, with results denoted as *p < 0.05, **p < 0.01 and ****p < 0.0001. ns, Not Significant.