Advertisement

Single-cell transcriptomic architecture and intercellular crosstalk of human intrahepatic cholangiocarcinoma

  • Author Footnotes
    † These authors contributed equally to this work.
    Min Zhang
    Footnotes
    † These authors contributed equally to this work.
    Affiliations
    College of Life Science and Bioengineering, Faculty of Environmental and Life Sciences, Beijing University of Technology, Beijing 100124, China

    Academy of Military Medical Sciences (AMMS), Academy of Military Sciences, Beijing 100071, China

    State Key Laboratory of Experimental Hematology, Fifth Medical Center of Chinese PLA General Hospital, Beijing 100071, China
    Search for articles by this author
  • Author Footnotes
    † These authors contributed equally to this work.
    Hui Yang
    Footnotes
    † These authors contributed equally to this work.
    Affiliations
    College of Life Science and Bioengineering, Faculty of Environmental and Life Sciences, Beijing University of Technology, Beijing 100124, China

    Stem Cell and Regenerative Medicine Lab, Institute of Health Service and Transfusion Medicine, AMMS, Beijing 100850, China

    South China Research Center for Stem Cell & Regenerative Medicine, SCIB, Guangzhou, Guangdong 510005, China
    Search for articles by this author
  • Author Footnotes
    † These authors contributed equally to this work.
    Lingfei Wan
    Footnotes
    † These authors contributed equally to this work.
    Affiliations
    College of Life Science and Bioengineering, Faculty of Environmental and Life Sciences, Beijing University of Technology, Beijing 100124, China

    Stem Cell and Regenerative Medicine Lab, Institute of Health Service and Transfusion Medicine, AMMS, Beijing 100850, China

    South China Research Center for Stem Cell & Regenerative Medicine, SCIB, Guangzhou, Guangdong 510005, China
    Search for articles by this author
  • Author Footnotes
    † These authors contributed equally to this work.
    Zhaohai Wang
    Footnotes
    † These authors contributed equally to this work.
    Affiliations
    Department of Hepatobiliary Surgery, Fifth Medical Center of Chinese PLA General Hospital, Beijing 100039, China
    Search for articles by this author
  • Author Footnotes
    † These authors contributed equally to this work.
    Haiyang Wang
    Footnotes
    † These authors contributed equally to this work.
    Affiliations
    Stem Cell and Regenerative Medicine Lab, Institute of Health Service and Transfusion Medicine, AMMS, Beijing 100850, China

    South China Research Center for Stem Cell & Regenerative Medicine, SCIB, Guangzhou, Guangdong 510005, China
    Search for articles by this author
  • Chen Ge
    Affiliations
    College of Life Science and Bioengineering, Faculty of Environmental and Life Sciences, Beijing University of Technology, Beijing 100124, China

    Stem Cell and Regenerative Medicine Lab, Institute of Health Service and Transfusion Medicine, AMMS, Beijing 100850, China

    South China Research Center for Stem Cell & Regenerative Medicine, SCIB, Guangzhou, Guangdong 510005, China
    Search for articles by this author
  • Yunhui Liu
    Affiliations
    College of Life Science and Bioengineering, Faculty of Environmental and Life Sciences, Beijing University of Technology, Beijing 100124, China

    Stem Cell and Regenerative Medicine Lab, Institute of Health Service and Transfusion Medicine, AMMS, Beijing 100850, China

    South China Research Center for Stem Cell & Regenerative Medicine, SCIB, Guangzhou, Guangdong 510005, China
    Search for articles by this author
  • Yajing Hao
    Affiliations
    Key Laboratory of RNA Biology, Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, China
    Search for articles by this author
  • Dongdong Zhang
    Affiliations
    Key Laboratory of RNA Biology, Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, China
    Search for articles by this author
  • Gaona Shi
    Affiliations
    State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China
    Search for articles by this author
  • Yandong Gong
    Affiliations
    Academy of Military Medical Sciences (AMMS), Academy of Military Sciences, Beijing 100071, China
    Search for articles by this author
  • Yanli Ni
    Affiliations
    State Key Laboratory of Experimental Hematology, Fifth Medical Center of Chinese PLA General Hospital, Beijing 100071, China
    Search for articles by this author
  • Chaojie Wang
    Affiliations
    Department of Pathology and Pathophysiology, Medical College, Jinan University, Guangzhou 510632, China
    Search for articles by this author
  • Yuan Zhang
    Affiliations
    Department of Integrated Traditional Chinese and Western Medicine, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022, China
    Search for articles by this author
  • Jiafei Xi
    Affiliations
    Academy of Military Medical Sciences (AMMS), Academy of Military Sciences, Beijing 100071, China

    Stem Cell and Regenerative Medicine Lab, Institute of Health Service and Transfusion Medicine, AMMS, Beijing 100850, China

    South China Research Center for Stem Cell & Regenerative Medicine, SCIB, Guangzhou, Guangdong 510005, China
    Search for articles by this author
  • Sen Wang
    Affiliations
    Department of Hepatobiliary Surgery, Fifth Medical Center of Chinese PLA General Hospital, Beijing 100039, China
    Search for articles by this author
  • Lei Shi
    Affiliations
    Department of Hepatobiliary Surgery, Fifth Medical Center of Chinese PLA General Hospital, Beijing 100039, China
    Search for articles by this author
  • Lina Zhang
    Affiliations
    College of Life Science and Bioengineering, Faculty of Environmental and Life Sciences, Beijing University of Technology, Beijing 100124, China
    Search for articles by this author
  • Wen Yue
    Correspondence
    Stem Cell and Regenerative Medicine Lab, Institute of Health Service and Transfusion Medicine, AMMS, Beijing 100850, China
    Affiliations
    Academy of Military Medical Sciences (AMMS), Academy of Military Sciences, Beijing 100071, China

    Stem Cell and Regenerative Medicine Lab, Institute of Health Service and Transfusion Medicine, AMMS, Beijing 100850, China

    South China Research Center for Stem Cell & Regenerative Medicine, SCIB, Guangzhou, Guangdong 510005, China
    Search for articles by this author
  • Xuetao Pei
    Correspondence
    Stem Cell and Regenerative Medicine Lab, Institute of Health Service and Transfusion Medicine, AMMS, Beijing 100850, China
    Affiliations
    Academy of Military Medical Sciences (AMMS), Academy of Military Sciences, Beijing 100071, China

    Stem Cell and Regenerative Medicine Lab, Institute of Health Service and Transfusion Medicine, AMMS, Beijing 100850, China

    South China Research Center for Stem Cell & Regenerative Medicine, SCIB, Guangzhou, Guangdong 510005, China
    Search for articles by this author
  • Bing Liu
    Correspondence
    State Key Laboratory of Experimental Hematology, Fifth Medical Center of Chinese PLA General Hospital, Beijing 100071, China
    Affiliations
    State Key Laboratory of Experimental Hematology, Fifth Medical Center of Chinese PLA General Hospital, Beijing 100071, China
    Search for articles by this author
  • Xinlong Yan
    Correspondence
    Corresponding authors. Addresses: College of Life Science and Bioengineering, Faculty of Environmental and Life Sciences, Beijing University of Technology, Beijing 100124, China
    Affiliations
    College of Life Science and Bioengineering, Faculty of Environmental and Life Sciences, Beijing University of Technology, Beijing 100124, China
    Search for articles by this author
  • Author Footnotes
    † These authors contributed equally to this work.
Open AccessPublished:June 04, 2020DOI:https://doi.org/10.1016/j.jhep.2020.05.039

      Highlights

      • Single cell transcriptomic datasets are a valuable resource to dissect cellular diversity and intercellular crosstalk of human ICCs.
      • Malignant cells displayed remarkable inter-tumor heterogeneity and Tregs revealed highly immunosuppressive characteristics.
      • Six distinct fibroblast subsets were defined in ICCs and adjacent tissues.
      • CD146+ vCAFs, comprising most of the fibroblasts, had tight interactions with malignant cells through IL-6/IL-6R axis.
      • Tumor exosomal miR-9-5p elicited IL-6 expression in vCAFs, contributing to ICC progression via upregulation of EZH2.

      Background & Aims

      Intrahepatic cholangiocarcinoma (ICC) is the second most common liver malignancy. ICC typically features remarkable cellular heterogeneity and a dense stromal reaction. Therefore, a comprehensive understanding of cellular diversity and the interplay between malignant cells and niche cells is essential to elucidate the mechanisms driving ICC progression and to develop therapeutic approaches.

      Methods

      Herein, we performed single-cell RNA sequencing (scRNA-seq) analysis on unselected viable cells from 8 human ICCs and adjacent samples to elucidate the comprehensive transcriptomic landscape and intercellular communication network. Additionally, we applied a negative selection strategy to enrich fibroblast populations in 2 other ICC samples to investigate fibroblast diversity. The results of the analyses were validated using multiplex immunofluorescence staining, bulk transcriptomic datasets, and functional in vitro and in vivo experiments.

      Results

      We sequenced a total of 56,871 single cells derived from human ICC and adjacent tissues and identified diverse tumor, immune, and stromal cells. Malignant cells displayed a high degree of inter-tumor heterogeneity. Moreover, tumor-infiltrating CD4 regulatory T cells exhibited highly immunosuppressive characteristics. We identified 6 distinct fibroblast subsets, of which the majority were CD146-positive vascular cancer-associated fibroblasts (vCAFs), with highly expressed microvasculature signatures and high levels of interleukin (IL)-6. Functional assays indicated that IL-6 secreted by vCAFs induced significant epigenetic alterations in ICC cells, particularly upregulating enhancer of zeste homolog 2 (EZH2) and thereby enhancing malignancy. Furthermore, ICC cell-derived exosomal miR-9-5p elicited high expression of IL-6 in vCAFs to promote tumor progression.

      Conclusions

      Our single-cell transcriptomic dataset delineates the inter-tumor heterogeneity of human ICCs, underlining the importance of intercellular crosstalk between ICC cells and vCAFs, and revealing potential therapeutic targets.

      Lay summary

      Intrahepatic cholangiocarcinoma is an aggressive and chemoresistant malignancy. Better understanding the complex transcriptional architecture and intercellular crosstalk of these tumors will help in the development of more effective therapies. Herein, we have identified important interactions between cancer cells and cancer-associated fibroblasts in the tumor stroma, which could have therapeutic implications.

      Graphical abstract

      Keywords

      Linked Article

      • Intrahepatic cholangiocarcinoma: A single-cell resolution unraveling the complexity of the tumor microenvironment
        Journal of HepatologyVol. 73Issue 5
        • Preview
          Cholangiocarcinoma (CCA) is one of the most aggressive epithelial cancer types with scarce therapeutic options, resulting in a dismal prognosis.1 Even if the understanding of the disease has improved over the last decade, researchers and clinicians are still faced with an enigmatic and worrisome cancer, with marked heterogeneity making therapeutic choices difficult. Besides the well-established anatomical criteria, by which CCA is categorized into 3 different tumor types (intrahepatic, perihilar and distal), the multifaceted tumor microenvironment (TME) is another factor that contributes to the complexity of CCA.
        • Full-Text
        • PDF
      See Editorial, pages 1007–1009

      Introduction

      Intrahepatic cholangiocarcinoma (ICC) is the second most common aggressive and chemotherapy-refractory liver malignancy. Over the past decade, it has become a significant global concern due to its increasing diagnostic incidence and accompanying mortality rates.
      • Peery A.F.
      • Crockett S.D.
      • Murphy C.C.
      • Lund J.L.
      • Dellon E.S.
      • Williams J.L.
      • et al.
      Burden and cost of gastrointestinal, liver, and pancreatic diseases in the United States: update 2018.
      The majority of patients have already reached advanced-stage disease at the time of diagnosis and only a few patients are suitable for curative hepatic resection. Moreover, the recurrence rate after surgical resection remains high, only 5% of patients survive beyond 5 years, and the prognosis is extremely poor for patients with unresectable ICC due to the lack of available treatment options.
      • Peery A.F.
      • Crockett S.D.
      • Murphy C.C.
      • Lund J.L.
      • Dellon E.S.
      • Williams J.L.
      • et al.
      Burden and cost of gastrointestinal, liver, and pancreatic diseases in the United States: update 2018.
      ,
      • Rizvi S.
      • Gores G.J.
      Emerging molecular therapeutic targets for cholangiocarcinoma.
      Large-scale genomic and transcriptomic studies have revealed that critical gene mutations and aberrant signaling pathways drive ICC pathogenesis,
      • Rizvi S.
      • Gores G.J.
      Emerging molecular therapeutic targets for cholangiocarcinoma.
      • Nakamura H.
      • Arai Y.
      • Totoki Y.
      • Shirota T.
      • Elzawahry A.
      • Kato M.
      • et al.
      Genomic spectra of biliary tract cancer.
      • Jusakul A.
      • Cutcutache I.
      • Yong C.H.
      • Lim J.Q.
      • Huang M.N.
      • Padmanabhan N.
      • et al.
      Whole-genome and epigenomic landscapes of etiologically distinct subtypes of cholangiocarcinoma.
      • Xue R.
      • Chen L.
      • Zhang C.
      • Fujita M.
      • Li R.
      • Yan S.M.
      • et al.
      Genomic and transcriptomic profiling of combined hepatocellular and intrahepatic cholangiocarcinoma reveals distinct molecular subtypes.
      • Farshidfar F.
      • Zheng S.
      • Gingras M.C.
      • Newton Y.
      • Shih J.
      • Robertson A.G.
      • et al.
      Integrative genomic analysis of cholangiocarcinoma identifies distinct IDH-mutant molecular profiles.
      including somatic mutations of TP53, KRAS, (ARID1A), IDH1/2 and FGFR gene fusions.
      • Rizvi S.
      • Gores G.J.
      Emerging molecular therapeutic targets for cholangiocarcinoma.
      ,
      • Farshidfar F.
      • Zheng S.
      • Gingras M.C.
      • Newton Y.
      • Shih J.
      • Robertson A.G.
      • et al.
      Integrative genomic analysis of cholangiocarcinoma identifies distinct IDH-mutant molecular profiles.
      ,
      • Dong L.Q.
      • Shi Y.
      • Ma L.J.
      • Yang L.X.
      • Wang X.Y.
      • Zhang S.
      • et al.
      Spatial and temporal clonal evolution of intrahepatic cholangiocarcinoma.
      However, these studies generally relied on data from bulk profiling, limiting their ability to accurately capture tumor heterogeneity. ICC tissues are typically characterized by a prominent desmoplastic stroma composed of various cell types such as CAFs, infiltrating immune cells, endothelial cells, and extracellular components that support tumor progression.
      • Hogdall D.
      • Lewinska M.
      • Andersen J.B.
      Desmoplastic tumor microenvironment and immunotherapy in cholangiocarcinoma.
      ,
      • Sirica A.E.
      • Gores G.J.
      Desmoplastic stroma and cholangiocarcinoma: clinical implications and therapeutic targeting.
      Targeting the rich tumor stroma of ICC in conjunction with targeted therapies for oncogene addiction or immunotherapies might be an effective strategy for ICC treatment.
      • Raggi C.
      • Invernizzi P.
      • Andersen J.B.
      Impact of microenvironment and stem-like plasticity in cholangiocarcinoma: molecular networks and biological concepts.
      Therefore, a deep understanding of cellular heterogeneity and the interplay between ICC cells and their microenvironment could allow the development of new therapeutic approaches for treating ICC.
      Single-cell RNA sequencing (scRNA-seq) analysis emerged as a powerful tool for revealing cellular diversity and intercellular communication at single-cell resolution.
      • Ren X.
      • Kang B.
      • Zhang Z.
      Understanding tumor ecosystems by single-cell sequencing: promises and limitations.
      • Baslan T.
      • Hicks J.
      Unravelling biology and shifting paradigms in cancer with single-cell sequencing.
      • Levitin H.M.
      • Yuan J.
      • Sims P.A.
      Single-cell transcriptomic analysis of tumor heterogeneity.
      Recently, scRNA-seq has been applied to dissect the complex tumor and immune landscapes of several cancers, including glioblastoma,
      • Yuan J.
      • Levitin H.M.
      • Frattini V.
      • Bush E.C.
      • Boyett D.M.
      • Samanamud J.
      • et al.
      Single-cell transcriptome analysis of lineage diversity in high-grade glioma.
      ,
      • Patel A.P.
      • Tirosh I.
      • Trombetta J.J.
      • Shalek A.K.
      • Gillespie S.M.
      • Wakimoto H.
      • et al.
      Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma.
      breast cancer,
      • Savas P.
      • Virassamy B.
      • Ye C.
      • Salim A.
      • Mintoff C.P.
      • Caramia F.
      • et al.
      Single-cell profiling of breast cancer T cells reveals a tissue-resident memory subset associated with improved prognosis.
      lung cancer,
      • Lambrechts D.
      • Wauters E.
      • Boeckx B.
      • Aibar S.
      • Nittner D.
      • Burton O.
      • et al.
      Phenotype molding of stromal cells in the lung tumor microenvironment.
      head and neck cancer,
      • Puram S.V.
      • Tirosh I.
      • Parikh A.S.
      • Patel A.P.
      • Yizhak K.
      • Gillespie S.
      • et al.
      Single-cell transcriptomic analysis of primary and metastatic tumor ecosystems in head and neck cancer.
      pancreatic ductal adenocarcinoma,
      • Peng J.
      • Sun B.F.
      • Chen C.Y.
      • Zhou J.Y.
      • Chen Y.S.
      • Chen H.
      • et al.
      Single-cell RNA-seq highlights intra-tumoral heterogeneity and malignant progression in pancreatic ductal adenocarcinoma.
      and liver cancer.
      • Zheng C.
      • Zheng L.
      • Yoo J.K.
      • Guo H.
      • Zhang Y.
      • Guo X.
      • et al.
      Landscape of infiltrating T cells in liver cancer revealed by single-cell sequencing.
      • Ma L.
      • Hernandez M.O.
      • Zhao Y.
      • Mehta M.
      • Tran B.
      • Kelly M.
      • et al.
      Tumor cell biodiversity drives microenvironmental reprogramming in liver cancer.
      • Zhang Q.
      • He Y.
      • Luo N.
      • Patel S.J.
      • Han Y.
      • Gao R.
      • et al.
      Landscape and dynamics of single immune cells in hepatocellular carcinoma.
      This technique has improved our understanding of cellular heterogeneity and facilitated the screening of promising molecular targets to guide anti-tumor therapies. However, tumor heterogeneity and the interplay between malignant cells and niche cells at single-cell resolution in human ICC remains poorly understood.
      In this study, we used a droplet-based scRNA-seq sequencing platform (10x Genomics) to profile single cells from human ICC and adjacent tissues. We identified high inter-tumor heterogeneity in ICC samples along with prominent immunosuppressive characteristics in CD4 regulatory T cells (Tregs). Furthermore, we defined 6 fibroblast subclusters in ICC and adjacent tissues, among which CD146+ vCAFs comprised the majority of CAFs in ICC tissues and could significantly induce enhancer of zeste homolog 2 (EZH2) upregulation and enhance ICC malignancy via the interleukin (IL)-6/IL-6 receptor (IL-6R) axis. Furthermore, exosomal miR-9-5p derived from ICC cells strongly induced vCAFs to promote ICC progression. Together, our results provide a comprehensive transcriptomic overview and dissect the intercellular crosstalk between ICC cells and vCAFs, suggesting potential targets for ICC therapy.

      Materials and methods

      ICC samples

      We obtained human ICC samples from the Fifth Medical Center of Chinese People's Liberation Army General Hospital (Beijing, China), with Institutional Review Board approval (2017041D).

      Library preparation and sequencing

      Single-cell transcriptomic amplification and library preparation were performed by Capitalbio Technology Corporation and Berry Genomics Corporation using single-cell 3′ v2 or v3 (10x Genomics) according to manufacturer's instructions.
      For further details regarding the materials and methods, please refer to the supplementary information and the CTAT table.

      Results

      Single-cell transcriptomic analysis revealed the complexity of human ICCs

      To investigate cellular diversity and molecular signatures in ICC tissues, we generated scRNA-seq profiles from 4 treatment-naïve ICC samples, 1 recurrent ICC sample, and 3 adjacent tissues using 10x Genomics sequencing (Fig. 1A; Dataset 1, 43,721 cells, GSE138709). The clinical characteristics and H&E staining of these patients are presented in Table S1 and Fig. S1. After stringent filtering, 31,302 cells from Dataset 1 were retained for further analysis. Following gene expression normalization, we conducted dimensionality reduction and clustering using principal component analysis and t-distributed stochastic neighbor embedding (tSNE), respectively. Thereafter, copy number variation (CNV) analysis was employed to distinguish malignant and non-malignant cells (Fig. S2A–C). These cells could be assigned to 10 distinct cell types (Fig. 1B) using known marker genes: malignant cells (11,601 cells, 37.1%, marked with EPCAM, keratin 19 [KRT19], and KRT7); cholangiocytes (546 cells, 1.7%, marked with FYXD2, TM4SF4, and ANXA4); hepatocytes (328 cells, 1.0%, marked with APOC3, FABP1, and APOA1); B cells (827 cells, 2.6%, marked with MS4A1 and CD79A); T cells (10,883 cells, 34.8%, marked with CD2, CD3D, and CD3E); natural killer (NK) cells (1,597 cells, 5.1%, marked with CD7, FGFBP2, and KLRF1); macrophages (3,407 cells, 10.9%, marked with CD14); dendritic cells (765 cells, 2.4%, marked with CLEC9A and CD1C); fibroblasts (498 cells, 1.59%, marked with ACTA2 and COL1A2); and endothelial cells (823 cells, 2.6%, marked with ENG and vWF; Fig. 1C, Table S2). Remarkably, the malignant subclusters were highly patient-specific, suggesting prominent molecular inter-tumor heterogeneity in the ICC samples (Fig. 1D, and Fig. S2C), while the proportion of each cell type varied greatly by sample (Fig. 1E). The differentially expressed genes (DEGs) and marker genes, as shown in the tSNE plots, confirmed the accuracy of cell identity (Fig. 1F, and Fig. S2D).
      Figure thumbnail gr1
      Fig. 1Comprehensive cellular overview of human ICC.
      (A) Schematic diagram of scRNA-seq analysis workflow. ICC and adjacent tissues were dissociated into single cells, sorted by FACS, and sequenced using 10x Genomics platform. (B) tSNE plots for the cell type identification of 31,302 high-quality single cells. (C) Violin plots showing marker genes for 10 distinct cell types. (D) Box plots showing the CNV signals for each cell type. (E) Bar plots showing the proportion of cell types in each sample. (F) Heatmap showing the top DEGs (Wilcoxon test) in each cell type. CNV, copy number variation; DEGs, differentially expressed genes; ICC, intrahepatic cholangiocarcinoma; scRNA-seq, single-cell RNA-sequencing; tSNE, t-distributed stochastic neighbor embedding.

      Transcriptomic inter-tumor heterogeneity of malignant cells in human ICCs

      In this study, we identified 6 main subclusters following the reclustering of all the malignant and normal epithelial cells (Fig. 2A). The malignant cells (subclusters 0–3) displayed a high degree of inter-tumor heterogeneity, consistent with a previous study which reported that somatic mutations are patient-specific.
      • Tirosh I.
      • Izar B.
      • Prakadan S.M.
      • Wadsworth 2nd, M.H.
      • Treacy D.
      • Trombetta J.J.
      • et al.
      Dissecting the multicellular ecosystem of metastatic melanoma by single-cell RNA-seq.
      Subclusters 4 (cholangiocytes, top DEGs: FXYD2 and RHOB) and 5 (hepatocytes, top DEGs: ASGR1 and APOE) possessed a lower CNV signal and were localized in adjacent tissues (Fig. 2A). Interestingly, subcluster 0 malignant cells were characterized by high expression levels of mesenchymal markers such as COL1A1, fibronectin, and IGFBP7, indicating epithelial-mesenchymal transition (EMT) characteristics. Subcluster 1 displayed high expression levels of the malignancy-promoting factors S100P and FABP5; subcluster 2 exhibited high expression levels of the immune-associated genes CD74 and HLA-DRA; and subcluster 3 malignant cells, from the recurrent patient, displayed high expression levels of SPINK1 (Fig. 2B). Gene set variation analysis (GSVA) indicated that these subclusters shared common activated signatures, such as IL6/STAT3, WNT, TGF, and TNF signals, alongside the following unique molecular signatures: EMT-dominant signature (subcluster 0), cell-cycle and hypoxia-dominant signature (subcluster 1), and interferon response-dominant signature (subcluster 2; Fig. 2C). In addition, we employed the single-cell regulatory network inference and clustering (SCENIC) method to identify SNAI2, MYC, and STAT1 as the underlying transcription factors in the different signatures (Fig. 2D, and Fig. S3).
      Figure thumbnail gr2
      Fig. 2Transcriptomic heterogeneity of malignant cells in ICC tissues.
      (A) tSNE plots and CNV box plots for 6 distinct epithelial cell subclusters. (B) Heatmap showing the top 10 DEGs (Wilcoxon test) in 6 epithelial cell subclusters. (C) Differences in pathway activity (scored per cell by GSVA) in 4 malignant cell subclusters. (D) Heatmap of the t-value for the area under the curve score of expression regulation by transcription factors, as estimated using SCENIC. (E) Violin plots showing the expression of marker genes in distinct malignant subclusters. (F) IHC staining showing the expression of CK19, TM4SF4, PROM1, CDH6, and SPINK1 in recurrent sample (ICC-20T) and adjacent liver tissue. (G) Kaplan–Meier survival curve of TM4SF4, SPINK1, and KRT19 expression using the optimal group cut-off point in patients with ICC (from TCGA). p values were obtained from 2-sided log-rank tests. CNV, copy number variation; DEGs, differentially expressed genes; GSVA, gene set variation analysis; ICC, intrahepatic cholangiocarcinoma; IHC, immunohistochemistry; tSNE, t-distributed stochastic neighbor embedding.
      Moreover, our scRNA-seq data revealed that malignant subclusters exhibited diverse expression levels of the mature cholangiocyte marker KRT19 and progenitor markers CDH6, ANXA4, TM4SF4, and PROM1 (Fig. 2E). Immunohistochemistry (IHC) staining verified the scRNA-seq analysis (Fig. 2F). Furthermore, Kaplan-Meier survival analysis indicated that higher expression of SPINK1 in patients with ICC was associated with worse survival (Fig. 2G). quantitative reverse transcription PCR and IHC staining further showed that SPINK1 was significantly upregulated in tumor spheroids (Fig. S4), while functional assays indicated that SPINK1 was required for tumor sphere propagation, colony formation, invasion, and drug resistance (Fig. S5, S6). Taken together, these results reveal a high degree of inter-tumor heterogeneity in patients with ICC and show that SPINK1 is closely associated with poor prognosis.

      Immunosuppressive tumor-infiltrating Tregs were enriched in human ICC tumors

      Tumor-infiltrating immune cells are highly heterogeneous and have been shown to play important roles in immune evasion and response to immunotherapy.
      • Bindea G.
      • Mlecnik B.
      • Tosolini M.
      • Kirilovsky A.
      • Waldner M.
      • Obenauf A.C.
      • et al.
      Spatiotemporal dynamics of intratumoral immune cells reveal the immune landscape in human cancer.
      In this study, the T and NK cells exhibited 8 distinct subclusters (Fig. 3A) and the tSNE plot revealed the subcluster distribution among sample pathology (Fig. 3B). According to their top 30 DEGs, T cell and NK cell subclusters were designated as C0–CD8T–GZMK (CD8A+GZMK+, subcluster 0), C1–CD8T–GZMB (CD8A+GZMB+, subcluster 1), C2–CD4T–naïve (CD4+IL7R+, subcluster 2), C3–CD8T–naïve (CD8A+IL7R+, subcluster 3), C4–NK–GZMH and C5–NK–GZMK (KLRF1+, subcluster 4, 5), C6–CD8T–proliferation (CD8A+TOP2A+MKI67+, subcluster 6), and C7–Tregs–FOXP3 (CD4+FOXP3+, subcluster 7; Fig. 3C, D). In addition, we found that CD8 T cell subpopulations (subclusters 0, 1, 3, and 6) expressed different levels of cytotoxic markers such as GZMK, perforin (PRF1), and interferon γ (IFNγ) (Fig. 3E and Fig. S7A), and the proliferating CD8 T cells (subcluster 6) expressed a certain number of exhaustion markers, such as lymphocyte-activation gene 3 protein (LAG3), T cell immunoreceptor with Ig and ITIM domains (TIGIT), and T cell immunoglobulin mucin receptor 3 (TIM3, or HAVCR2), suggesting that proliferating CD8 T cells (subcluster 6) became exhausted. Furthermore, CD4 Tregs (C7–Tregs–FOXP3) were characterized by the prominent expression levels of immunosuppression markers such as TIGIT, cytotoxic T lymphocyte antigen 4 (CTLA4), and TNFR-related protein (GITR, or TNFRSF18). Two NK clusters (C4–NK–GZMH and C5–NK–GZMK) mainly derived from adjacent tissues were characterized by high GZMB, GZMK, PRF1, and KLRF1 expression, indicating that these cells remained cytotoxic or activated (Fig. 3D, E). Pervasive changes in cancer-associated T and NK clusters were revealed by enriched signaling pathways, including increased hypoxia, apoptosis, and IFN response, alongside decreased oxidative phosphorylation (Fig. S7B). Intercellular communication inferred by ligand-receptor analysis indicated that the TIGIT-PVR pair was enriched between Tregs and malignant cells (Fig. 3F), suggesting that blocking the TIGIT-PVR axis may affect the interaction of Tregs with malignant cells and could be an effective therapeutic strategy for ICCs. Taken together, our single-cell analyses reveal that Tregs display highly immunosuppressive characteristics and suggest that manipulation of Tregs could present a novel therapeutic strategy for ICCs.
      Figure thumbnail gr3
      Fig. 3Distinct subtypes of infiltrating T cells and NK cells in the ICC ecosystem.
      (A,B) tSNE plots for T cells and NK cells, color-coded for 8 subclusters (A) and sample pathology (B). (C) tSNE plots, color-coded for the expression (gray to red) of marker genes for each cell type, as indicated. (D) Heatmap expression showing the top 30 DEGs (Wilcoxon test) in each cluster. (E) Violin plots of selected cytotoxicity, proliferation, and suppressive genes in distinct T cell, regulatory T cell, and NK cell subclusters. (F) Interaction analysis showing enriched receptor-ligand pairs in T cell or NK cell subclusters and malignant cells. DEGs, differentially expressed genes; ICC, intrahepatic cholangiocarcinoma; NK, natural killer; tSNE, t-distributed stochastic neighbor embedding.

      Distinct fibroblast subpopulations in human ICC ecosystem

      An intense desmoplastic reaction is a hallmark of ICC, as demonstrated by picrosirius red and α-SMA IHC staining (Fig. 4A). The 498 fibroblasts in our unselected viable strategy of scRNA-seq analyses (Dataset 1) accounted for only 1.59% of all viable cells, which reflects the challenge of capturing these populations (Fig. 4B). Therefore, we used a negative selection strategy (EpCAMCD45CD31 cells) to enrich the fibroblast population in 2 ICC samples (Fig. 4B; Dataset 2, 13,150 cells). After stringent filtering, we obtained 2,941 high-quality fibroblasts that were clustered into 6 subpopulations, of which 5 fibroblast clusters (subcluster 0, 1, 2, 3, 4) were mainly enriched in ICC tissues, whereas subcluster 5 was mainly present in adjacent tissues (Fig. 4C). All 6 subclusters expressed high levels of canonical fibroblast markers such as ACTA2 (α-SMA), COL1a2, and PDGFRβ, confirming their identity as fibroblasts; however, each subcluster displayed distinct transcriptomic signatures (Fig. 4D).
      Figure thumbnail gr4
      Fig. 4Distinct fibroblast subpopulations detected in human ICC.
      (A) H&E, picrosirius red, CK19, and α-SMA staining in ICC and adjacent tissues. (B) Experimental scRNA-seq strategy for fibroblasts in viable (7-AAD-negative) cells (n = 498) or (7-AAD/CD45/EpCAM/CD31-negative) FACS-enriched fibroblasts (n = 2,443) in ICC samples and adjacent tissues. (C) tSNE plots of total fibroblasts color-coded for 6 clusters, sorting strategy and sample pathology, respectively. (D) Heatmap showing the top 5 DEGs (Wilcoxon test) for each cluster. (E) tSNE plots color-coded for the expression (gray to red) of marker genes for the distinct cell subclusters. (F) GO analysis of DEGs in distinct fibroblast subclusters. (G) Multiplex immunofluorescence staining for panCK, α-SMA, PDGFRβ, MCAM, POSTN, and DAPI in ICC. DEGs, differentially expressed genes; GO, gene ontology; ICC, intrahepatic cholangiocarcinoma; scRNA-seq, single-cell RNA sequencing; tSNE, t-distributed stochastic neighbor embedding.
      Subcluster 0 fibroblasts accounted for the majority of the fibroblast populations (57.6%) and were characterized by microvasculature signature genes such as CD146 (MCAM), MYH11, GJA4, and RGS5, as well as inflammatory chemokines such as IL-6 and CCL8 (Fig. 4E). Thus, we designated them as vascular CAFs (vCAFs, vCAFs-c0-MCAM). Gene ontology (GO) analysis of vCAFs indicated significant enrichment for muscle contraction, response to hypoxia, and mesenchymal cell proliferation, consistent with their microvascular signatures (Fig. 4F). Subcluster 1 fibroblasts expressed low levels of α-SMA but high levels of extracellular matrix (ECM) signatures, including collagen molecules (COL5A1, COL5A2, and COL6A3), periostin (POSTN), FN1, LUM, DCN, and VCAN. Interestingly, the GO terms enriched for this subtype were associated with ECM and collagen fibril organization, so we accordingly designated them as matrix CAFs (mCAFs, mCAFs–c1–POSTN, Fig. 4E, F). Like mCAFs–c1–POSTN, subcluster 2 fibroblasts expressed low levels of α-SMA but high levels of FBLN1, IGFI, CXCL1, IGFBP6, SLPI, SAA1, and complement genes (C3 and C7). In addition, the GO terms enriched for this subcluster were related to ECM, inflammatory response regulation, and complement activation, indicating that this subcluster may engage in immune modulation. Accordingly, fibroblasts in this subcluster were named inflammatory CAFs (iCAFs, iCAFs–c2–FBLN1; Fig. 4E, F). Consistent with a previous report of mouse KPC tumors (Kras+/LSL-G12D; Trp53+/LSL-R172H; Pdx1-Cre) and human pancreatic ductal adenocarcinoma (PDAC),
      • Elyada E.
      • Bolisetty M.
      • Laise P.
      • Flynn W.F.
      • Courtois E.T.
      • Burkhart R.A.
      • et al.
      Cross-species single-cell analysis of pancreatic ductal adenocarcinoma reveals antigen-presenting cancer-associated fibroblasts.
      we found that subcluster 3 fibroblasts expressed major histocompatibility complex II (MHC-II) genes such as CD74, HLA-DRA, and HLA-DRB1. Moreover, the GO terms enriched in this subcluster were related to leukocyte cell-cell adhesion, response to IFN-γ, antigen processing, and antigen presentation via MHC-II; we therefore termed them antigen-presenting CAFs (apCAFs, apCAFs–c3–CD74; Fig. 4E, F). Subcluster 4 fibroblasts mainly expressed epithelium-specific marker genes such as KRT19, KRT8, and SAA1, which we designated as EMT-like CAFs (eCAFs, eCAFs–c4–KRT19; Fig. 4D). Finally, subcluster 5 fibroblasts were mainly derived from adjacent tissues and expressed high levels of lipid metabolism and processing related genes, including APOA2, FABP1, FABP4, and FRZB, therefore, we named them lipofibroblast–c5–FABP1 (Fig. 4D).
      Next, we investigated the correlation between vCAFs and mCAFs using scRNA-seq data and human ICC bulk RNA-seq data, finding that the vCAF signature in ICC was highly correlated with proliferation and microvasculature metagenes, whereas the mCAF signature was closely associated with ECM and stroma metagenes (Fig. S8).
      • Bartoschek M.
      • Oskolkov N.
      • Bocci M.
      • Lovrot J.
      • Larsson C.
      • Sommarin M.
      • et al.
      Spatially and functionally distinct subclasses of breast cancer-associated fibroblasts revealed by single cell RNA sequencing.
      Notably, we re-analyzed the scRNA-seq data from patients treated with immune checkpoint inhibitor therapy, and our results were clearly verified in Ma et al.'s ICC scRNA-seq data (Fig. S9A–C).
      • Ma L.
      • Hernandez M.O.
      • Zhao Y.
      • Mehta M.
      • Tran B.
      • Kelly M.
      • et al.
      Tumor cell biodiversity drives microenvironmental reprogramming in liver cancer.
      We further confirmed the presence of the main fibroblast subsets in ICC samples using IHC staining (Fig. S9D) and multiplex immunofluorescence staining (Fig. 4G, Fig. S10). Interestingly, CD146+ vCAFs were mainly localized in the tumor core and microvascular region, suggesting that they undergo intense interactions with ICC cells. Conversely, POSTN+ mCAFs were localized in the invasive front of tumor nests, predominantly within collagen-rich stromal streaks, suggesting that mCAFs are closely associated with ICC invasion. Taken together, our findings reveal different fibroblast subpopulations and indicate a highly heterogeneous stromal microenvironment that improves our understanding of ICC pathogenesis.

      The IL-6/IL-6R axis was enriched in the interplay between vCAFs and ICC cells

      To explore the interactions between ICC cells and niche cells, we conducted intercellular interaction analyses based on ligand-receptor pairs. Interestingly, fibroblasts expressed more ligands corresponding to the receptors expressed by malignant cells (Fig. 5A). In addition, the IL-6/IL-6R pair was enriched in the interactions between vCAFs and malignant cells (Fig. 5B, Fig. S11A), consistent with the finding that malignant subclusters display common activation signatures such as IL-6-STAT3 signaling upregulation (Fig. 2C). Moreover, IHC staining verified that IL-6 expression was mainly localized in the CD146+ microvascular region (Fig. 5C), in accordance with our scRNA-seq findings that IL-6 was highly expressed in the vCAFs (Fig. 4D,E). We also isolated and expanded CD146+ vCAFs from ICC tissues. Flow cytometry analysis and immunofluorescence staining demonstrated that the vCAFs were positive for canonical mesenchymal markers and IL-6 expression (Fig. 5D and Fig. S11B). Furthermore, we assessed the cytokines secreted by vCAFs using antibody microarray profiling,
      • Chen W.J.
      • Ho C.C.
      • Chang Y.L.
      • Chen H.Y.
      • Lin C.A.
      • Ling T.Y.
      • et al.
      Cancer-associated fibroblasts regulate the plasticity of lung cancer stemness via paracrine signalling.
      finding that IL-6 was highly secreted by vCAFs. Importantly, even higher IL-6 secretion levels were detected in the vCAFs when they interacted directly with ICC cells (Fig. 5E). Additionally, IL-6 expression in the vCAFs was significantly upregulated when vCAFs were treated with conditioned medium from ICC cells (Fig. 5F, and Fig. S11C) or transwell co-cultured with ICC cells (Fig. 5G, and Fig. S11D). Quantitative PCR (qPCR), western blotting, and ELISA further verified that IL-6 was highly expressed in the vCAFs (Fig. S11E). Collectively, these results indicate that the IL-6/IL-6R axis is enriched in the interplay between vCAFs and ICC cells.
      Figure thumbnail gr5
      Fig. 5vCAFs interact with malignant cells via the IL-6/IL-6R axis.
      (A) Bar plot presenting the numbers of putative ligand-receptor pairs between malignant cells and indicated cell types using scRNA-seq data. (B) Dot plot showing receptor-ligand pair analysis of the interactions between malignant cells and distinct cell types. (C) IHC staining of CD146 and IL-6 in ICC tissues and adjacent samples. (D) Flow cytometry analysis showing that vCAFs isolated from ICC tissues were positive for IL-6 and mesenchymal markers. (E) Representative cytokine array image for cytokines secreted from vCAFs, ICC cells, or vCAFs co-cultured with ICC cells. (F, G) Quantitative PCR analysis of IL-6 expression in vCAFs when treated with conditioned medium of ICC cells or transwell co-cultured with ICC cells. Data was analyzed by unpaired t test. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001. ICC, intrahepatic cholangiocarcinoma; IHC, immunohistochemistry; scRNA-seq, single-cell RNA sequencing; vCAFs, vascular cancer-associated fibroblasts.

      Contribution of vCAFs to ICC tumorigenesis and cancer stemness via the IL-6/IL-6R axis

      To further explore whether vCAFs could contribute to ICC progression, we subcutaneously injected luciferase-labeled ICC cells into nude mice alone or with vCAFs. We found that vCAFs significantly accelerated tumor growth, as demonstrated by in vivo luciferase imaging, tumor volume, and tumor weight analysis (Fig. 6A). Moreover, cancer stem cell markers such as CD13, CD90, and EGFR were markedly increased when vCAFs were directly co-cultured with GFP-positive ICC cells (Fig. 6B). We also established a vCAF/sphere transwell co-culture system and found that vCAFs significantly enhanced tumor sphere formation (Fig. 6C); however, this effect was significantly abrogated when treated with IL-6R neutralizing antibody (tocilizumab) or small molecule inhibitors of IL-6/IL-6R signaling SC144 (gp130 inhibitor) and INCB018424 (JAK1/2 inhibitor). LY364947, a TGF-β1 receptor inhibitor, was used as a control (Fig. 6D). Consistent with these results, tumor sphere formation was significantly enhanced by the exogenous addition of IL-6, but not TGF-β1 (Fig. 6E), while silencing IL-6Ra in ICC cells significantly repressed the sphere-promoting effects of vCAFs (Fig. 6F). Interestingly, CD146-depleted (shCD146) and IL-6-depleted (shIL-6) vCAFs displayed a significant reduction of IL-6 expression (Fig. 6G) and their sphere-promoting effect of vCAFs on tumor spheres was potently abrogated (Fig. 6H). Collectively, these data suggest that IL-6 secreted by vCAFs could significantly enhance the malignancy of ICC cells.
      Figure thumbnail gr6
      Fig. 6vCAFs significantly enhance tumorigenesis and cancer stemness.
      (A) Tumor luciferase activity, volumes, and weights were examined to analyze the effect of vCAFs on tumor propagation. (B) Flow cytometry analysis of cancer stem cell markers (CD13, CD90, and EGFR) in GFP-positive ICC cells alone and directly co-cultured with vCAFs. (C) Tumor sphere formation was analyzed in ICC cells – alone or transwell co-cultured with vCAFs. (D) Tumor sphere formation in ICC cells transwell co-cultured with vCAFs and treated with Tocilizumab (IL-6 receptor neutralizing antibody, 50 μg/ml), LY364947 (TGF-β1 receptor inhibitor, 5 μM), SC144 (gp130 inhibitor, 5 μM) or INCB018424 (JAK1/2 inhibitor, 15 μM), respectively. (E) Tumor sphere formation in ICC cells was analyzed when cells were treated with IL-6 (10 ng/ml) or TGF-β1 (10 ng/ml). (F) Diminished tumor sphere formation in ICC cells with IL-6 Ra depletion alone or co-cultured with vCAFs. (G) Quantitative PCR analysis of IL-6 expression in vCAFs with depleted CD146 (shCD146) or IL-6 (shIL-6). (H) Tumor sphere formation was analyzed using the transwell co-culture system in vCAFs depleted of CD146 (shCD146) or IL-6 (shIL-6). Data was analyzed by unpaired t test. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001. ICC, intrahepatic cholangiocarcinoma; vCAFs, vascular cancer-associated fibroblasts.

      Tumor exosomal miR-9-5p elicited IL-6 expression in vCAFs, thereby leading to epigenetic alterations in ICC

      We used cDNA microarray analysis to explore the gene expression in GFP-positive ICC cells after direct co-culture with vCAFs (Fig. 7A). GO analysis revealed that the DEGs were mainly enriched for epigenetic alterations including chromatin modification and methyltransferase activity (Fig. 7B, Fig. S12A). We further verified these differentially expressed epigenetic modification factors in ICC samples compared to adjacent tissues using our bulk cDNA microarrays (ICC and adjacent, n = 3) and human ICC TCGA bulk sequencing data (ICC, n = 36; adjacent, n = 9). We obtained consistent results showing that the epigenetic modification factor EZH2 was significantly upregulated in ICC samples (Fig. S12B). Depletion of EZH2 in ICC cells attenuated the proliferative capacity of tumors (Fig. S12C). Moreover, tumor sphere formation was remarkably repressed by the addition of small molecule EZH2 inhibitors (GSK126 and GSK343) to ICC cells in transwell co-culture with vCAFs (Fig. S12D). Together, these results suggest that vCAFs contribute to epigenetic alterations, such as EZH2 upregulation, fostering ICC malignancy.
      Figure thumbnail gr7
      Fig. 7ICC-derived exosomal miR-9-5p elicits IL-6 expression in vCAFs, leading to ICC epigenetic alterations.
      (A) Microarray analyses examining altered gene expression in GFP-positive ICC cells co-cultured with or without vCAFs and sorted by flow cytometry. (B) GO terms indicating that epigenetic modification-related genes were enriched in ICC cells co-cultured with vCAFs. (C) Immunohistochemical staining of EZH2 and IL-6Ra in ICC tissues and adjacent samples. (D) Western blotting assays detecting EZH2 expression in ICC cells overexpressing IL-6 (oeIL-6) or with IL-6Ra depletion (shIL-6 Ra). (E) Western blotting assays of EZH2 and p-STAT3 expression in ICC cells treated with IL-6 (10 ng/ml) or IL-6 signaling inhibitors: SC144 (gp130 inhibitor, 7 μM) and INCB018424 (JAK1/2 inhibitor, 20 μM). (F) IL-6 expression in vCAFs treated with ICC cell-derived exosomes detected by quantitative PCR analysis. (G) quantitative PCR analysis of mRNAs enriched in ICC cell-derived exosomes. (H) ELISA and western blot analysis of IL-6 expression in vCAFs transfected with the indicated miRNAs. (I) Tumor sphere formation in ICC cells transwell co-cultured with vCAFs and transfected with miR-9-5p or a control vector. (J) Schematic illustration of intercellular crosstalk between ICC cells and vCAFs. Data was analyzed by unpaired t test. ∗p <0.05, ∗∗p <0.01, ∗∗∗p <0.001. GO, gene ontology; ICC, intrahepatic cholangiocarcinoma; vCAFs, vascular cancer-associated fibroblasts.
      Next, we explored whether IL-6 secreted by vCAFs could induce EZH2 upregulation in ICC cells. Our scRNA-seq data revealed that EZH2 and IL-6 receptors such as IL-6ST (gp130), IL-6 Ra, and F3 were highly expressed in ICC cells (Fig. S13), while IHC staining verified that EZH2 and IL-6R were highly expressed in tumor nests (Fig. 7C). Exogenous IL-6 treatment significantly upregulated EZH2 and p-STAT3 expression. Moreover, EZH2 upregulation was confirmed in ICC cells overexpressing IL-6, while EZH2 downregulation was observed following IL-6Ra knockdown in ICC cells (shIL-6Ra; Fig. 7D, E). Consistently, EZH2 expression was markedly downregulated in ICC cells when treated with the IL-6 signaling inhibitors SC144 (gp130 inhibitor) and INCB018424 (JAK1/2 inhibitor; Fig. 7E). Since exosomes secreted by tumor cells can instruct the stroma to foster tumor progression, we purified exosomes from conditioned medium of ICC cells according to a standard protocol, as verified by electron microscopy and western blotting (Fig. S14A). Intriguingly, IL-6 expression was remarkably upregulated in the vCAFs treated with exosomes from ICC cells (Fig. 7F). Exosome-encapsulated miRNAs are abundant and have recently been implicated in the crosstalk between tumor and stromal cells
      • Garcia-Silva S.
      • Peinado H.
      Melanosomes foster a tumour niche by activating CAFs.
      • Becker A.
      • Thakur B.K.
      • Weiss J.M.
      • Kim H.S.
      • Peinado H.
      • Lyden D.
      Extracellular vesicles in cancer: cell-to-cell mediators of metastasis.
      • Costa-Silva B.
      • Aiello N.M.
      • Ocean A.J.
      • Singh S.
      • Zhang H.
      • Thakur B.K.
      • et al.
      Pancreatic cancer exosomes initiate pre-metastatic niche formation in the liver.
      ; therefore, we hypothesized that exosomes derived from ICC cells might encapsulate miRNAs to mediate the upregulation of IL-6 expression in vCAFs. To validate this hypothesis, we used qPCR assays to detect the exosomal miRNA profiles of ICC cell-conditioned medium. Considering the close association between IL-6 and inflammation, ICC cells stimulated with lipopolysaccharide (LPS) were used as the positive control. A panel of miRNAs, including miR-9-5p, miR-1246, miR-1247-3p, and miR-16-5p, were enriched in ICC-derived exosomes and upregulated by LPS stimulation (Fig. 7G, Fig. S14B). Thereafter, ELISA, western blotting, and qPCR assays indicated that miR-9-5p markedly promoted IL-6 expression in the vCAFs (Fig. 7H, and Fig. S14C). When vCAFs were stably transfected with miR-9-5p, tumor sphere formation in the transwell co-culture system was significantly elevated compared to the control vector (Fig. 7I). Collectively, these findings suggest that ICC cell-derived exosomal miR-9-5p significantly induces IL-6 expression in vCAFs, in turn leading to EZH2 upregulation and enhancing malignancy in ICC cells (Fig. 7J).

      Discussion

      In this study, we employed scRNA-seq to comprehensively delineate the transcriptomic landscape of human ICCs and revealed novel cellular interactions between ICC cells and vCAFs at single-cell resolution. ICC tissues are characterized by an intense desmoplastic reaction during which activated CAFs surrounding ICC cells are believed to play a pivotal role in ICC progression. However, cellular diversity of CAFs and how the CAF subsets interact with ICC cells at single-cell resolution have not been well defined. ScRNA-seq analysis has been used to elucidate constituent cell types including CAFs in human melanoma, lung cancer,
      • Lambrechts D.
      • Wauters E.
      • Boeckx B.
      • Aibar S.
      • Nittner D.
      • Burton O.
      • et al.
      Phenotype molding of stromal cells in the lung tumor microenvironment.
      and head and neck cancer.
      • Puram S.V.
      • Tirosh I.
      • Parikh A.S.
      • Patel A.P.
      • Yizhak K.
      • Gillespie S.
      • et al.
      Single-cell transcriptomic analysis of primary and metastatic tumor ecosystems in head and neck cancer.
      Due to the restrictions in the number of fibroblasts analyzed and the limitations of scRNA-seq methodology, clearly discriminating CAF subsets at single-cell resolution remains a challenge. Several pioneer groups have used a negative selection strategy to enrich fibroblasts by depleting EpCAM+CD45+CD31+ cells in scRNA-seq analysis. Bartoschek et al. have defined 3 distinct CAF subsets in genetically engineered mouse models of breast cancer, revealing the spatial separation of CAF subpopulations of different origins.
      • Bartoschek M.
      • Oskolkov N.
      • Bocci M.
      • Lovrot J.
      • Larsson C.
      • Sommarin M.
      • et al.
      Spatially and functionally distinct subclasses of breast cancer-associated fibroblasts revealed by single cell RNA sequencing.
      Elyada et al. applied this strategy to enrich fibroblasts from mouse and human PDAC samples and demonstrated the presence of 3 CAF subtypes: myofibroblastic CAFs, iCAFs and apCAFs, indicating that apCAFs had the ability to present antigens to CD4 T cells and potentially modulate the immune response.
      • Elyada E.
      • Bolisetty M.
      • Laise P.
      • Flynn W.F.
      • Courtois E.T.
      • Burkhart R.A.
      • et al.
      Cross-species single-cell analysis of pancreatic ductal adenocarcinoma reveals antigen-presenting cancer-associated fibroblasts.
      We defined 6 distinct fibroblast subtypes in human ICC and adjacent tissue, of which vCAFs, mCAFs, iCAFs, apCAFs, and eCAFs were found in ICC tissues and lipofibroblasts were mainly found in adjacent tissues, revealing that the desmoplastic microenvironment of ICC is highly heterogeneous.
      In this study, vCAFs were the most prevalent fibroblast subpopulation and expressed high levels of microvasculature signature genes and IL-6. Multiplex immunofluorescence and IHC staining verified that CD146+ vCAFs expressed high IL-6 levels and were specifically distributed in the tumor core and microvascular region, suggesting their close interactions with ICC cells. Furthermore, we found that the IL-6/IL-6R pair was significantly enriched in the interactions between vCAFs and ICC cells, consistent with our GSVA assays which indicated that ICC subclusters shared common activated signatures such as IL6-STAT3 signaling. In addition, CD146 depletion in vCAFs resulted in a remarkable decrease of IL-6 expression. Functional assays indicated that IL-6 secreted by vCAFs could upregulate EZH2 in ICC cells, thereby enhancing their malignancy. Interestingly, a recent study revealed that CD10+GPR77+CAFs derived from patients with chemoresistant breast cancer can maintain cancer stemness and induce chemoresistance by abundantly expressing IL-6 and providing a survival niche for breast cancer stem cells.
      • Su S.
      • Chen J.
      • Yao H.
      • Liu J.
      • Yu S.
      • Lao L.
      • et al.
      CD10(+)GPR77(+) cancer-associated fibroblasts promote cancer formation and chemoresistance by sustaining cancer stemness.
      Additionally, Fang et al. found that exosomal miR-1247-3p secreted by metastatic hepatocellular carcinoma could induce high IL-6 and IL-8 expression in CAFs, resulting in the metastasis of liver cancer to the lung.
      • Fang T.
      • Lv H.
      • Lv G.
      • Li T.
      • Wang C.
      • Han Q.
      • et al.
      Tumor-derived exosomal miR-1247-3p induces cancer-associated fibroblast activation to foster lung metastasis of liver cancer.
      In this study, miR-9-5p was enriched in the exosomes of ICC cells and significantly induced IL-6 expression in vCAFs, upregulating EZH2 in ICC cells and enhancing malignancy. Collectively, our scRNA-seq data and functional analyses indicated that vCAFs, which account for the majority of fibroblasts in ICCs, could secrete high levels of IL-6 to modulate tumor epigenetic alterations and promote ICC progression.
      Taken together, our findings provide a comprehensive transcriptomic landscape of human ICC at single-cell resolution and present a well-established resource for elucidating ICC diversity.

      Abbreviations

      apCAFs, antigen-presenting CAFs; CAFs, cancer-associated fibroblasts; CTLA4, cytotoxic T lymphocyte 4; CNV, copy number variation; DEGs, differentially expressed genes; eCAFs, EMT-like CAFs; ECM, extracellular matrix; EMT, epithelial-mesenchymal transition; EZH2, enhancer of zeste homolog 2; GO, gene ontology; GSVA, gene set variation analysis; HCC, hepatocellular carcinoma; iCAFs, inflammatory CAFs; ICC, intrahepatic cholangiocarcinoma; IFNγ, interferon-γ; IL-6, interleukin-6; IL-6R, IL-6 receptor; LAG3, lymphocyte-activation gene 3 protein; LPS, lipopolysaccharide; mCAFs, matrix CAFs; MHC-II, major histocompatibility complex-II; NK, natural killer; PDAC, pancreatic ductal adenocarcinoma; POSTN, periostin; PRF1, perforin; scRNA-seq, single-cell RNA-sequencing; TIGIT, T cell immunoreceptor with Ig and ITIM domains; TIM3, T cell immunoglobulin mucin receptor 3; Tregs, regulatory T cells; tSNE, t-distributed stochastic neighbor embedding; vCAFs, vascular CAFs.

      Financial support

      The research was supported by grants from National Natural Science Foundation of China ( 81772617 , 81472341 ), Great Wall Scholar Project ( CIT&TCD20190311 ), Beijing Municipal Education Commission ( KM201710005031 ), and Guangzhou Health Care and Cooperative Innovation Major Project ( 201803040005 ). National Key Research and Development Program of China ( 2017YFA0103100 , 2017YFA0103103 , 2017YFA0103104 ). We thank Mr Zeng Fan for the assistance in FACS experiments. We thank Editage (www.editage.cn) and LetPub (www.letpub.com) for language editing.

      Authors' contributions

      X.L.Y., B.L., W.Y., and X.T.P. designed the study. M.Z. performed experiments and analyzed the single-cell RNA sequencing data. H.Y., L.F.W., H.Y.W., C.G., Y.J.H., D.D.Z., Y.H.L., N.G.S, Y.D.G., Y.L.N., C.J.W., Y.Z., J.F.X., and L.N.Z. performed experiments and analyzed data, Z.H.W., S.W. and L.S. contributed biopsy samples and pathology analysis. M.Z., X.L.Y., and B.L. wrote the manuscript. All authors read and approved the manuscript.

      Data availability

      The scRNA-seq data has been deposited in Gene Expression Omnibus (GEO) with accession number: GSE138709, GSE142784. The accession number of microarray data of vCAF modulating ICC cells: GSE148773.

      Conflict of interest

      The authors declare no conflicts of interest that pertain to this work.
      Please refer to the accompanying ICMJE disclosure forms for further details.

      References

        • Peery A.F.
        • Crockett S.D.
        • Murphy C.C.
        • Lund J.L.
        • Dellon E.S.
        • Williams J.L.
        • et al.
        Burden and cost of gastrointestinal, liver, and pancreatic diseases in the United States: update 2018.
        Gastroenterology. 2019; 156: 254-272.e11
        • Rizvi S.
        • Gores G.J.
        Emerging molecular therapeutic targets for cholangiocarcinoma.
        J Hepatol. 2017; 67: 632-644
        • Nakamura H.
        • Arai Y.
        • Totoki Y.
        • Shirota T.
        • Elzawahry A.
        • Kato M.
        • et al.
        Genomic spectra of biliary tract cancer.
        Nat Genet. 2015; 47: 1003-1010
        • Jusakul A.
        • Cutcutache I.
        • Yong C.H.
        • Lim J.Q.
        • Huang M.N.
        • Padmanabhan N.
        • et al.
        Whole-genome and epigenomic landscapes of etiologically distinct subtypes of cholangiocarcinoma.
        Cancer Discov. 2017; 7: 1116-1135
        • Xue R.
        • Chen L.
        • Zhang C.
        • Fujita M.
        • Li R.
        • Yan S.M.
        • et al.
        Genomic and transcriptomic profiling of combined hepatocellular and intrahepatic cholangiocarcinoma reveals distinct molecular subtypes.
        Cancer Cell. 2019; 35: 932-947.e8
        • Farshidfar F.
        • Zheng S.
        • Gingras M.C.
        • Newton Y.
        • Shih J.
        • Robertson A.G.
        • et al.
        Integrative genomic analysis of cholangiocarcinoma identifies distinct IDH-mutant molecular profiles.
        Cell Rep. 2017; 19: 2878-2880
        • Dong L.Q.
        • Shi Y.
        • Ma L.J.
        • Yang L.X.
        • Wang X.Y.
        • Zhang S.
        • et al.
        Spatial and temporal clonal evolution of intrahepatic cholangiocarcinoma.
        J Hepatol. 2018; 69: 89-98
        • Hogdall D.
        • Lewinska M.
        • Andersen J.B.
        Desmoplastic tumor microenvironment and immunotherapy in cholangiocarcinoma.
        Trends Cancer. 2018; 4: 239-255
        • Sirica A.E.
        • Gores G.J.
        Desmoplastic stroma and cholangiocarcinoma: clinical implications and therapeutic targeting.
        Hepatology. 2014; 59: 2397-2402
        • Raggi C.
        • Invernizzi P.
        • Andersen J.B.
        Impact of microenvironment and stem-like plasticity in cholangiocarcinoma: molecular networks and biological concepts.
        J Hepatol. 2015; 62: 198-207
        • Ren X.
        • Kang B.
        • Zhang Z.
        Understanding tumor ecosystems by single-cell sequencing: promises and limitations.
        Genome Biol. 2018; 19: 211
        • Baslan T.
        • Hicks J.
        Unravelling biology and shifting paradigms in cancer with single-cell sequencing.
        Nat Rev Cancer. 2017; 17: 557-569
        • Levitin H.M.
        • Yuan J.
        • Sims P.A.
        Single-cell transcriptomic analysis of tumor heterogeneity.
        Trends Cancer. 2018; 4: 264-268
        • Yuan J.
        • Levitin H.M.
        • Frattini V.
        • Bush E.C.
        • Boyett D.M.
        • Samanamud J.
        • et al.
        Single-cell transcriptome analysis of lineage diversity in high-grade glioma.
        Genome Med. 2018; 10: 57
        • Patel A.P.
        • Tirosh I.
        • Trombetta J.J.
        • Shalek A.K.
        • Gillespie S.M.
        • Wakimoto H.
        • et al.
        Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma.
        Science. 2014; 344: 1396-1401
        • Savas P.
        • Virassamy B.
        • Ye C.
        • Salim A.
        • Mintoff C.P.
        • Caramia F.
        • et al.
        Single-cell profiling of breast cancer T cells reveals a tissue-resident memory subset associated with improved prognosis.
        Nat Med. 2018; 24: 986-993
        • Lambrechts D.
        • Wauters E.
        • Boeckx B.
        • Aibar S.
        • Nittner D.
        • Burton O.
        • et al.
        Phenotype molding of stromal cells in the lung tumor microenvironment.
        Nat Med. 2018; 24: 1277-1289
        • Puram S.V.
        • Tirosh I.
        • Parikh A.S.
        • Patel A.P.
        • Yizhak K.
        • Gillespie S.
        • et al.
        Single-cell transcriptomic analysis of primary and metastatic tumor ecosystems in head and neck cancer.
        Cell. 2017; 171: 1611-1624.e24
        • Peng J.
        • Sun B.F.
        • Chen C.Y.
        • Zhou J.Y.
        • Chen Y.S.
        • Chen H.
        • et al.
        Single-cell RNA-seq highlights intra-tumoral heterogeneity and malignant progression in pancreatic ductal adenocarcinoma.
        Cell Res. 2019; 29: 725-738
        • Zheng C.
        • Zheng L.
        • Yoo J.K.
        • Guo H.
        • Zhang Y.
        • Guo X.
        • et al.
        Landscape of infiltrating T cells in liver cancer revealed by single-cell sequencing.
        Cell. 2017; 169: 1342-1356.e16
        • Ma L.
        • Hernandez M.O.
        • Zhao Y.
        • Mehta M.
        • Tran B.
        • Kelly M.
        • et al.
        Tumor cell biodiversity drives microenvironmental reprogramming in liver cancer.
        Cancer Cell. 2019; 36: 418-430.e6
        • Zhang Q.
        • He Y.
        • Luo N.
        • Patel S.J.
        • Han Y.
        • Gao R.
        • et al.
        Landscape and dynamics of single immune cells in hepatocellular carcinoma.
        Cell. 2019; 179: 829-845.e20
        • Tirosh I.
        • Izar B.
        • Prakadan S.M.
        • Wadsworth 2nd, M.H.
        • Treacy D.
        • Trombetta J.J.
        • et al.
        Dissecting the multicellular ecosystem of metastatic melanoma by single-cell RNA-seq.
        Science. 2016; 352: 189-196
        • Bindea G.
        • Mlecnik B.
        • Tosolini M.
        • Kirilovsky A.
        • Waldner M.
        • Obenauf A.C.
        • et al.
        Spatiotemporal dynamics of intratumoral immune cells reveal the immune landscape in human cancer.
        Immunity. 2013; 39: 782-795
        • Elyada E.
        • Bolisetty M.
        • Laise P.
        • Flynn W.F.
        • Courtois E.T.
        • Burkhart R.A.
        • et al.
        Cross-species single-cell analysis of pancreatic ductal adenocarcinoma reveals antigen-presenting cancer-associated fibroblasts.
        Cancer Discov. 2019; 9: 1102-1123
        • Bartoschek M.
        • Oskolkov N.
        • Bocci M.
        • Lovrot J.
        • Larsson C.
        • Sommarin M.
        • et al.
        Spatially and functionally distinct subclasses of breast cancer-associated fibroblasts revealed by single cell RNA sequencing.
        Nat Commun. 2018; 9: 5150
        • Chen W.J.
        • Ho C.C.
        • Chang Y.L.
        • Chen H.Y.
        • Lin C.A.
        • Ling T.Y.
        • et al.
        Cancer-associated fibroblasts regulate the plasticity of lung cancer stemness via paracrine signalling.
        Nat Commun. 2014; 5: 3472
        • Garcia-Silva S.
        • Peinado H.
        Melanosomes foster a tumour niche by activating CAFs.
        Nat Cell Biol. 2016; 18: 911-913
        • Becker A.
        • Thakur B.K.
        • Weiss J.M.
        • Kim H.S.
        • Peinado H.
        • Lyden D.
        Extracellular vesicles in cancer: cell-to-cell mediators of metastasis.
        Cancer Cell. 2016; 30: 836-848
        • Costa-Silva B.
        • Aiello N.M.
        • Ocean A.J.
        • Singh S.
        • Zhang H.
        • Thakur B.K.
        • et al.
        Pancreatic cancer exosomes initiate pre-metastatic niche formation in the liver.
        Nat Cell Biol. 2015; 17: 816-826
        • Su S.
        • Chen J.
        • Yao H.
        • Liu J.
        • Yu S.
        • Lao L.
        • et al.
        CD10(+)GPR77(+) cancer-associated fibroblasts promote cancer formation and chemoresistance by sustaining cancer stemness.
        Cell. 2018; 172: 841-856.e16
        • Fang T.
        • Lv H.
        • Lv G.
        • Li T.
        • Wang C.
        • Han Q.
        • et al.
        Tumor-derived exosomal miR-1247-3p induces cancer-associated fibroblast activation to foster lung metastasis of liver cancer.
        Nat Commun. 2018; 9: 191