Bulk anda single-cell transcriptome profiling reveals the molecular characteristics of T cell-mediated tumor killing in pancreatic cancer

Abstract

Background: Despite the potential of immune checkpoint blockade (ICB) as a promising treatment for Pancreatic adenocarcinoma (PAAD), there is still a need to identify specific subgroups of PAAD patients who may benefit more from ICB. T cell-mediated tumor killing (TTK) is the primary concept behind ICB. We explored subtypes according to genes correlated with the sensitivity to TKK and unraveled their underlying associations for PAAD immunotherapies.

Methods: Genes that control the responsiveness of T cell-induced tumor destruction (GSTTK) were examined in PAAD, focusing on their varying expression levels and association with survival results. Moreover, samples with PAAD were separated into two subsets using unsupervised clustering based on GSTTK. Variability was evident in the tumor immune microenvironment, genetic mutation, and response to immunotherapy among different groups. In the end, we developed TRGscore, an innovative scoring system, and investigated its clinical and predictive significance in determining sensitivity to immunotherapy.

Results: Patients with PAAD were categorized into 2 clusters based on the expression of 52 GSTTKs, which showed varying levels and prognostic relevance, revealing unique TTK patterns. Survival outcome, immune cell infiltration, immunotherapy responses, and functional enrichment are also distinguished among the two clusters. Moreover, we found the CATSPER1 gene promotes the progression of PAAD through experiments. In addition, the TRGscore effectively predicted the responses to chemotherapeutics or immunotherapy in patients with PAAD and overall survival.

Conclusions: TTK exerted a vital influence on the tumor immune environment in PAAD. A greater understanding of TIME characteristics was gained through the evaluation of the variations in TTK modes across different tumor types. It highlights variations in the performance of T cells in PAAD and provides direction for improved treatment approaches.

Keywords: Immunotherapy; Pancreatic adenocarcinoma; Single-cell sequencing; T cell-mediated tumor killing; Tumor immune microenvironment.

Fig. 1

Overview of the process flow. GSTTK: genes regulating the sensitivity of T cell-mediated tumor killing, TTK: T cell-mediated tumor killing.

Fig. 2

Unsupervised consensus clustering method (K-means) was used to generate consensus matrices for patients in the TCGA(A), ICGC(B), and meta-GEO(C) cohorts. Survival analysis of the different clusters in the TCGA (D) and ICGC (E) cohorts and meta-GEO (F). Assessment of the TME in the two TTK associated designs(G).

Fig. 3

Variability in levels of immune cell infiltration was observed in two clusters analyzed in TCGA(A) and ICGC(B) cohorts using MCPcounter, and in the meta-GEO(C) cohort using CIBERSORT. Visualization of immune reactions within the TTK-associated clusters using ssGSEA(D) heatmap.

Fig. 4

Both patterns exhibit expression of chemokines, receptors, MHC molecules(A), and immune checkpoint-related genes(B). In the TCGA-PAAD cohort, boxplots showed distinct TTK-related patterns based on TME-related signatures (C, D).The score comparison of 10 pivotal cancerogenic signaling pathways amongs two TTK related patterns were illustrated(E). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 5

Biological pathways activation status was compared between two patterns using ssGSEA enrichment analysis (Kruskal-Wallis test, P < 0.05), where red indicates activation and blue indicates inhibition(A). Correlation between the TMB and the two TTK related patterns(B). Significantly mutated genes in the patients with PAAD in the different patterns. There were significant mutations in the genes of the PAAD patients according to the different patterns. Genes that have been mutated (rows, top 20) are ranked by mutation rate. The right panel displays the percentage of mutations, while the top panel shows the total number of mutations. Mutation types are indicated by color coding (C, D).

Fig. 6

Verification of hub gene.Venn diagram showing 4 Intersection genes as the candidate (A). CATSPER1's median expression level served as the threshold for overall survival in both ICGC (B) and TCGA cohort (C). The difference of level of CD8⁺ T cells between CATSPER1 -high and -low groups (D). CATSPER1 was immunohistochemically stained in PAAD and normal tissues using samples from the First Affiliated Hospital of Wenzhou University cohort. Bar = 50 μm (E). PCR analysis revealed the associated CATSPER1 expression in Panc-1 and PATU-8988 cell lines when compared to normal cell lines (F). PCR testing revealed the impact of reducing CATSPER1 in Panc-1 and PATU-8988 cell lines(G). Cell proliferation was assessed using the CCK-8 assay following the knockdown of CATSPER1 in Panc-1 and PATU-8988 cell lines (H). Inhibition of CATSPER1 using the Transwell invasion assay resulted in decreased cellular invasion of Panc-1 and PATU-8988. Visual depiction showing the quantity of invasive Panc-1 and PATU-8988 cells in each microscopic area. Bar = 100 μm (I). Contrast in the relative levels of CATSPER1 expression in PAAD and healthy tissues. *p < 0.05,**p < 0.01,***p < 0.001.

Fig. 7

Quantification of TTK related signatures based on TRGscore. Survival analysis was conducted for the overall survival (OS) in the TCGA-PAAD (A) and ICGC (B) cohorts, respectively. Correlations between the two TTK-relevant patterns and TRGscore in TCGA-PAAD (C) and ICGC(D) cohort, respectively. The TCGA-PAAD cohort(E) showed similar levels of immune checkpoint-related gene expression in both groups. Relationships between TRGscore and specific pathway patterns in the TCGA-PAAD cohort(F). Scatter plots show a strong positive correlation between TRGscore and TMB in the TCGA-PAAD cohort(G). MCPcounter(H) analyzed the variation in levels of immune cell infiltration between two groups in TCGA(H). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 8

TRGscore predicts the response of PAAD to Immunotherapy. Forecasting the response to immunotherapy (anti-PD-1 and anti-CTLA4) in the TRGscore high and low subgroups (A). TIS scores (B). TIDE score (C). The Kaplan–Meier curve based on high- and low- TRGscore groups in the IMvigor210 cohort (D). The Wilcoxon test was conducted to analyze the variation in TRGscore related to anti-PD-L1 responsiveness in the IMvigor210 cohort (E). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 9

ScRNA-seq analysis. Low TRGscore suggests an immune-active TME in the single cell cohort. t-SNE plot (left) and UMAP (right) plot showing the composition of 7 main subgroups derived from pancreatic cancer samples (A). The dynamics of TRGscore in 7 main cell types are showed in the t-SNE plot and UMAP plot (B). Distribution in the TRGscore of five pancreatic cancer samples (separated into 2 patterns) (C). The distribution (D) and the proportion (E) of 7 main subtypes in five pancreatic cancer samples. Differences in cytotoxic score (F) and exhaustion (G) of T cells between two TRGscore related groups. Circos plots illustrating the CCL signaling pathways between high-TRGScore (left) and low-TRGScore (right) group (H). Circos plots illustrating the SEMA3 signaling pathways between high-TRGScore (left) and low-TRGScore (right) group (I).

figs1

figs2

figs3

figs4

figs5

figs6

figs7

figs8