Single-cell and spatial architecture of primary liver cancer

Abstract

Primary liver cancer (PLC) poses a leading threat to human health, and its treatment options are limited. Meanwhile, the investigation of homogeneity and heterogeneity among PLCs remains challenging. Here, using single-cell RNA sequencing, spatial transcriptomic and bulk multi-omics, we elaborated a molecular architecture of 3 PLC types, namely hepatocellular carcinoma (HCC), intrahepatic cholangiocarcinoma (ICC) and combined hepatocellular-cholangiocarcinoma (CHC). Taking a high-resolution perspective, our observations revealed that CHC cells exhibit internally discordant phenotypes, whereas ICC and HCC exhibit distinct tumor-specific features. Specifically, ICC was found to be the primary source of cancer-associated fibroblasts, while HCC exhibited disrupted metabolism and greater individual heterogeneity of T cells. We further revealed a diversity of intermediate-state cells residing in the tumor-peritumor junctional zone, including a congregation of CPE⁺ intermediate-state endothelial cells (ECs), which harbored the molecular characteristics of tumor-associated ECs and normal ECs. This architecture offers insights into molecular characteristics of PLC microenvironment, and hints that the tumor-peritumor junctional zone could serve as a targeted region for precise therapeutical strategies.

Fig. 1. Molecular architecture of PLC.

a Experimental workflow using scRNA-seq, multi-omics and ST, validated by RNAscope ISH and mIHC. b Demographic and clinical characteristics of 9 donors and related sample collection of tumor, peritumor, lymph node and peripheral blood (P121-P127 for scRNA-seq and multi-omics, and P128-P129 for ST). c UMAP distribution of single cells from 7 donors’ perioperative samples (P121-P127). d UMAP distribution of ST spots from 2 donors (P128-P129). e UMAP distribution of cell populations (247,515 cells from 22 peri-operative samples). T, T cells; NK, natural killer cells; B, B cells; myeloid, myeloid cells; endothelial, endothelial cells; fibroblast, fibroblast cells or hepatic stellate cells; M1, M2, M3, supra-clusters related to malignant cells. f UMAP distribution based on tumor type, cell cycle phase, tissue type. It shows an equilibrium distribution that was not affected by these factors. g UMAP distribution based on canonical marker gene of supra-clusters. h Bar plots of supra-clusters based on tissue type, patient, tumor type, cell cycle phase and cell number. ST, spatial transcriptomics; UMAP, uniform manifold approximation and projection; G1, gap 1 Phase; S, synthesis phase; G2M, gap 2 and mitotic phase. Mix, mixed and unassigned cells or clusters.

Fig. 2. Spatial patterns of tumor, peritumor and tumor-peritumor junctional zones.

a Annotated solid tissue cryosection on ST slices. T, tumor zone; P, peritumor zone; J, tumor-peritumor junctional zone (zone between blue and red dotted lines); S, stroma zone (black or white dotted lines); n-F, no-fatty infiltration region; F, fatty infiltration region; P128T, tumor tissue slice from P128; P128P1, peritumor tissue slice 1 from P128; P128P2, peritumor tissue slice 2 from P128; P129TP1, tumor-peritumor junctional zone tissue slice from P129 main lesion; P129TP2, tumor-peritumor junctional zone tissue slice from P129 sub-lesion. Scale bars, 1 mm. b, c Clustering (b) and UMAP distribution (c) of ST spots. Colors represent different clusters. d GSVA of P128 (upper left) and P129 (upper right). SCENIC analysis of P128 (lower left) and P129 (lower right). Light gray indicates lower enrichment and red indicates higher enrichment. e Expression of key genes related to cholesterol homeostasis in HCC ST slices. f Spatial expression pattern of marker genes related to fibroblast activation in P129TP2 ST slice (upper) and corresponding consecutive cryosection slice (lower). ACTA2, smooth muscle aortic alpha-actin; FN1, fibronectin-1; VIM, vimentin; Scale bars, 1000 µm.

Fig. 3. Distinct phenotypes of PLC’s malignant cells at a single-cell level.

a UMAP distribution of M1, M2 and M3. Clustered cells present a distinctive patient-specific way. b Bar plot of cell cycle phase in terms of M1, M2 and M3. c Identification of somatic chromosomal CNV based on scRNA-seq data of patient-specific malignant sub-clusters from M1. d Metascape pathway enrichment based on malignant sub-clusters from M1. black frame, relatively enriched pathways in ICC sub-clusters; white frame, relatively enriched pathways in HCC sub-clusters. e SCENIC analysis of malignant sub-clusters from M1. f Pseudotime trajectory of malignant sub-clusters from M1.

Fig. 4. Identification of cancer-associated fibroblasts from PLCs.

a UMAP distribution of re-clustered fibroblasts. ICC-derived, intrahepatic cholangiocarcinoma derived fibroblasts; HCC-derived, hepatocellular carcinoma derived fibroblasts; CHC-derived, combined hepatocellular-cholangiocarcinoma derived fibroblasts. b Stacked violin plot showing expression of canonical marker genes and differential expression of genes of sub-clusters. c Comparison of somatic chromosomal CNV between fibroblasts and tumor cells. d Bar plots of fibroblast sub-clusters based on tissue type, patient, tumor type. e Metascape pathway enrichment based on differentially expressed genes of each sub-cluster. f Differences of metascape pathway enrichment between HCC-CAFs and ICC-CAFs. CAFs, cancer-associated fibroblasts. g Differences of metascape pathway enrichment between CHC-CAFs and ICC-CAFs. h Differences of metascape pathway enrichment between CHC-CAFs and HCC-CAFs. i Pseudotime trajectory of CAFs derived from 3 PLC types. HCC-T, cells from HCC-derived tumor tissue; ICC-T, cells from ICC-derived tumor tissue; CHC-T, cells from CHC-derived tumor tissue. j RNAscope ISH stained with marker genes of subcluster C1-CAFs and C2-CAFs. Staining shows expression of C1 marker gene MYH11, C2 marker gene CLO10A1, and a common co-localization marker gene TAGLN. All nuclei were DAPI-stained. Scale bars, 200μm (upper), 20μm (lower). k Box plots presenting differential expression of MYH11 + C1 CAFs (left) and COL10A1⁺C2 CAFs (right) between HCC-T (n = 89 biologically independent samples, 2 replicates) and ICC-T (n = 48 biologically independent samples) external cohort. Two-sided t-test: ns, no significant; ****, p < 0.0001. Lower, inside and upper horizontal line of the box plot indicate first quartile, median, and third quartile, separately.

Fig. 5. Identification of intermediate-state endothelial cells.

a UMAP distribution of re-clustered endothelial cells. LSECs, liver sinusoidal endothelial cells; TAECs, tumor-associated endothelial cells; intermediate-state ECs, intermediate-state endothelial cells; inflammatory ECs, inflammatory endothelial cells; CAFs-like ECs, cancer-associated fibroblasts-like endothelial cells. b Box plots presenting the proportion of endothelial cells in terms of tissues (upper) and tumor types (lower). Two-sided t-test: ns, no significant; *, p < 0.05. c Bar plots of endothelial cell sub-clusters based on tissue, patient and tumor type. d GSVA of endothelial cell sub-clusters (upper) and SCENIC analysis of endothelial cell sub-clusters (lower). Light gray indicates lower enrichment and red indicates higher enrichment. e Expression heatmap of selected genes in endothelial cell sub-clusters. Blue indicates lower expression and yellow indicates higher expression. f Pseudotime trajectory of LSECs, TAECs and intermediate-state ECs. g RNA velocity showing dynamic flows among endothelial cells. Arrows show the directions. h Spatial distribution prediction of intermediate-state sub-clsuters C5 and C8 in P129 ST slices using MIA and MNN algorithm. Red color shows higher possibility of cells located in clusters of ST spots. i RNAscope ISH stained with marker genes of intermediate-state sub-cluster C5. Staining shows expression of C5 marker gene CPE, 2 common colocalization marker genes VWF and STAB2. All nuclei were DAPI-stained. T, tumor zone; P, peritumor zone; J, tumor-peritumor junctional zone. Scale bars, 2000μm (left), 50μm (right three). j Box plots presenting expression of CPE⁺C5 intermediate-state ECs based on different zones of slices over CHC (upper), HCC (middle) and ICC (lower). Two-sided t-test: ns, no significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001. Lower, inside and upper horizontal line of the box plot indicate first quartile, median, and third quartile, separately.