Single-cell sequencing reveals the heterogeneity of pancreatic neuroendocrine tumors under genomic instability and histological grading

Abstract

Histological grading is the key factors affecting the prognosis and instructive in guiding treatment and assessing recurrence in non-functional pancreatic neuroendocrine tumor (NF-Pan-NET). Approximately one-third of patients without copy number variation (CNV) alteration and the prognosis of these patients are better than that of patients with CNV alteration. However, the difference between CNV and histological grading is unclear. Here, we analyzed the heterogeneity of tumor cells according to two classification criteria, genomic instability (including CNV alteration and tumor mutation burden) and histological grading. We revealed that the activated core pathways of tumor cells were significantly different under different histological grading's and genomic instability patterns. We also found that tip cells, lymphatic endothelial cells, macrophages, CD1A + dendritic cell, Treg, MAIT, ILC, and CAFs might participate in the process of hepatic metastases, which will facilitate the understanding of the patterns to decode the malignant potential and of NF-Pan-NET.

Keywords: Cancer; Microenvironment; Omics; Transcriptomics.

Graphical abstract

Figure 1

Single-cell profiling of the tumor ecosystem in NF-Pan-NET (A) The demographic and pathological data of the 17 randomly selected NF-Pan-NET cases in our center, single-cell RNA-sequence, and multiple-omics (WES, RNA-sequencing and proteomics) of bulk tissues were conducted in some of cases. (B) Schematic diagram showed the design and purpose of this study. (C) The cells were visualized by UAMP. (D) Marker genes of each cluster were shown in the heatmap. (E) The proportion of selected cell types in bulk RNA-seq was calculated by CIBERSORTx (left) and the proportion of selected cell types in our enrolled scRNA-seq data. The data of scRNA-seq are represented as mean ± SEM. The statistical analysis in the left showed the comparation of cell proportion between bulk RNA-seq and scRNA-seq data. The statistical analysis in the right showed the comparation of cell proportion in scRNA-seq data between group NT and PT. (F) Proportion of cell types in different groups. (G) Heatmap showed the proportion of cell types in each specimen.

Figure 2

Tumor cells show marked heterogeneity in copy number variation (A) CNV score of all endocrine cells in each sample. The ordinate represents the frequency. (B) CNV score of selected CNV-high endocrine cells in each sample. The ordinate represents the frequency. Percentage represents the proportion of inferCNV-high neuroendocrine cells to total neuroendocrine cells. (C) Copy number variations in the WES data. (D) TMB were analyzed in the WES data. (E) Signal pathway of endocrine cells in different specimens.

Figure 3

Revealing the heterogeneity of tumor cells under the pattern of genomic instability and histological grading (A) The subcluster of endocrine cells was presented on the tSNE map. (B–D) The source of endocrine cells was presented on the tSNE map. PT-G1, endocrine cells from the patients with histological grading G1. PT-G2, endocrine cells from patients with histological grading G2. HM, hepatic metastases. NT-a, endocrine cells of normal pancreas from patients in our center. NT-b, endocrine cells of normal pancreas from Peng’s study. PT non-metastatic, endocrine cells from patients without hepatic metastases. PT metastatic, endocrine cells of primary lesion from patients with hepatic metastases. (E) The proportion of CNV^high tumor cells, CNV^low tumor cells, and normal endocrine cells in each cluster. (F) The proportion of tumor cells from patients with G1 grade, G2 grade, hepatic metastases, and normal adjacent tissues in each cluster. (G) The proportion of tumor cells from group GIL, GIH, PHM, and normal adjacent tissues in each cluster. (H) Differentiated degree of endocrine cells in each cluster was analyzed by CytoTRACE. (I) Gene-module analysis of endocrine cells in each cluster. (J and K) The activation of KEGG pathway in endocrine cells of different group was analyzed by QuSAGE.

Figure 4

Identifying subtype of fibroblasts associated with malignant progression (A) tSNE map showed the sub-cluster of fibroblasts. (B) The proportion of sub-cluster in different groups. (C) Selected marker genes of fibroblasts in cluster 2 and 6. (D) Differentiated degree of fibroblasts in each cluster was analyzed by CytoTRACE. (E) Pathway analysis of fibroblasts in cluster 2 and 6. (F) Gene-module analysis of fibroblasts in each cluster. (G) The activation of KEGG pathway in fibroblasts in each cluster was analyzed by QuSAGE. (H) Part interleukin family gene expression of fibroblasts in each cluster.

Figure 5

Analysis of endothelial cell heterogeneity based on genome instability and histological grading (A) tSNE map showed the subpopulations of endothelial cells. (B) Selected marker genes of each endothelial cell subpopulation. (C) The proportion of each endothelial cell subpopulation in different groups. (D) Gene-module analysis of endothelial cells in each cluster. (E) The expression of chemokines in each endothelial cell subpopulation. (F) Immunohistochemistry assay showed the typical staining of PDPN and CCL21 in a sample of PHM (primary lesion). (G) The proportion of PDPN⁺/CCL21⁺ endothelial cells in different groups. (H) The Chi-square test compared the composition ratio of CCL21⁺ to PDPN⁺ endothelial cells. (I) Kaplan-Meier depicted the effect of PDPN⁺/CCL21⁺ endothelial cells on PFS probability.

Figure 6

The heterogeneity of monocytic cells based on genome instability and histological grading (A) Selected marker genes of monocytic cells. (B) tSNE map showed the subpopulations of monocytic cells. (C) The proportion of subpopulations of monocytic cells in different groups. (D) The proportion of subpopulations of macrophages in different groups. (E) Gene-module analysis of macrophages in each cluster. (F) The expression of IL-1b in each cluster of macrophages.