1 Introduction

Gastrointestinal stromal tumors (GISTs) have complex epidemiology and varied treatment responses that require advanced therapeutic treatment strategies. GISTs are the most common mesenchymal tumors of the gastrointestinal tract worldwide. In the United States, approximately 5000 new cases of GISTs are detected each year, representing an estimated incidence of 10–15 cases per million people per year [1]. Surgical resection is the primary treatment for localized GIST. The use of tyrosine kinase inhibitors (TKIs) such as imatinib has revolutionized the treatment for GISTs, especially for patients with specific mutations in the KIT and PDGFRA genes. Imatinib is effective in both adjuvant and neoadjuvant therapies in the cases when complete surgical resection is not possible. In addition, sunitinib and regorafenib are approved for the treatment of GISTs with imatinib resistance [1,2,3]. At present, the treatment of gastrointestinal mesenchymal tumors still faces great difficulties, for example, the development of TKI resistance, and the heterogeneity of tumor mutations also complicates treatment choices. Understanding and reducing the side effects of these drugs may help improve patients' quality of life [4].

Recent studies have greatly deepened our understanding of GIST heterogeneity and stemness, highlighting their impact on disease progression and therapeutic challenges. Research has focused on genetic diversity such as mutations in MEN1 and MAX as well as the identification of stem cell properties through advanced approaches such as single-cell RNA sequencing analysis [5, 6]. Innovative models including 3D tumor models have been used to explore how tumor microenvironmental factors (stromal stiffness) affect cancer stem cell characteristics in GISTs. Studies showed that specific signaling pathways such as IL22RA1/STAT3 could promote tumorigenicity by enhancing stem cell-like features [7]. The integration of single-cell and bulk RNA sequencing has also been widely applied for the construction of gene signatures to elucidate the molecular basis of tumor heterogeneity [8, 9]. These findings all contributed to the development of more effective therapeutic strategies to target the complex biology of GISTs for improving the treatment outcomes [10].

This study was designed to investigate the heterogeneous and stemness characteristics of interstitial cells of Cajal (ICCs) and its role in the pathophysiology and therapeutic outcome of GISTs. Here, the cell types in GISTs were characterized using single-cell RNA sequencing (scRNA-seq) data. Subsequently, differential and enrichment analyses were performed on the identified ICC subpopulations to explore their stem cell properties. SCENIC analysis was used to explore the potential regulatory mechanisms of ICCs in GISTs. Finally, the interaction between ICCs and other cell types and its role in the tumor microenvironment (TME) were inferred applying intercellular communication analysis. Our study provided novel insights into the role of ICCs in GISTs, hoping to improve the targeted therapies for patients with GISTs.

2 Materials and methods

2.1 Dataset acquisition and preprocessing

The single-cell RNA sequencing dataset GSE162115, which contained four GIST samples of two tumor center samples (intra-tumor) and two tumor edge samples (peri-tumor), was downloaded from the Gene Expression Omnibus (GEO) database. Single-cell analysis and data processing were performed using the Seurat package [11, 12]. For quality control, genes expressed in fewer than 3 cells and cells expressing fewer than 200 genes were eliminated, retaining the cells expressing between 200 and 4000 genes with less than 15% mitochondrial genes. The proportion of mitochondria and rRNA was calculated using the PercentageFeatureSet function. The data from the 4 samples were normalized using the ScaleData function to identify highly variable genes, followed by downscaling with principle component analysis (PCA) to select 20 PCs as anchors (parameter dim = 20 was selected for the analysis). Batch effects between samples were eliminated using the Harmony package [13]. Dimensionality reduction was performed using the RunTSNE function. Clustering of cell subpopulations was performed applying the FindNeighbors and FindClusters functions at the resolution parameter of 0.1. Each cell type was annotated using marker genes provided in the CellMarker 2.0 database [14].

2.2 Differential expression analysis and functional enrichment analysis between cell subpopulations

Differential expression analysis of ICCs in intra-tumor and peri-tumor tissues was performed using the FindAllMarkers function in the Seurat package, and differentially expressed genes (DEGs) in ICCs of the two types of tissues were identified. In differential expression analysis, we set the analytical parameters to logfc.threshold = 0.25, min.pct = 0.25, only.pos = T. The clusterProfiler package was used for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional analysis [15, 16] to identify significant signaling pathways and biological processes (FDR < 0.05 and p < 0.05) in the ICCs in the two tissues. The gene set enrichment analysis Hallmarker Pathway was downloaded from the Molecular Signatures database (MSigDB, https://www.gsea-msigdb.org/gsea/msigdb/collections.jsp), and the AUCell package [17] was selected to calculate the activity scores of the Hallmarker Pathway in the ICCs in intra-tumor and peri-tumor tissues, respectively.

2.3 The tumor heterogeneity and stem cell characterization of ICCs

Identified ICCs were further clustered at the resolution of 0.1 to determine the cell subpopulations. Subsequently, the cancer stem cells (CSCs) characteristics of the ICCs were assessed by the CytoTRACE method, a biological tool used to quantify the state of cellular differentiation and to represent a continuous spectrum of undifferentiated cells to highly differentiated cells. Based on single-cell transcriptomic data, the CytoTRACE predicts a cell's differentiation potential by analyzing the complexity and diversity of gene expression. A high CytoTRACE score indicates that the cell expresses a low number of genes and a low complexity of gene expression, which means that the cells are in an undifferentiated state with more stem cell characteristics [18].

2.4 Single-cell regulatory network inference and clustering analysis

Single-cell regulatory network inference and clustering (SCENIC) analysis was performed using the SCENIC package [17, 19]. The GENIE3 method in the SCENIC package was used to calculate the potential target genes of each transcription factor (TFs), and top5perTarget was used to construct the TF regulatory network. Finally, highly reliable TF-target gene pairs were identified, and the AUCell function was used to calculate the activity of regulons in each cell. The activity of each regulon in individual cells was scored by the SCENIC package based on the expression value of the gene, with higher scores representing higher levels of activation of the gene set.

2.5 Analysis of cellular communication

To analyze the intercellular communication and interactions between ICCs and other cell types (T cells, B cells, macrophages), cell communication analysis was performed using the CellChat package [20], which is a specifically designed tool for analyzing and visualizing the intercellular communication networks based on single-cell RNA sequencing data.

2.6 Statistical analysis

Statistical analysis of the data was performed using R software (3.6.0), and comparisons between groups were performed using a two-sided independent samples t-test, with p < 0.05 indicating a statistically significant difference.

3 Results

3.1 Single-cell mapping of intra-tumor and peri-tumor tissues in GISTs

A total of 34,861 eligible cells were identified from GSE162115 for subsequent analyses, and using these cells we classified 8 cell clusters (Fig. 1A). According to the marker genes in the CellMarker2.0 database, the 8 clusters were annotated to B cells, endothelial cells, interstitial cells of Cajal (ICCs), macrophages 1, macrophages 2, MKI67+ progenitor cells, NK cells, T cells (Fig. 1B). It was found that KIT and PDGFRA were high-expressed in ICC (Fig. 1C), and that the proportion of ICC in intra-tumor tissues was lower than that in peri-tumor tissues (Fig. 1D).

Fig. 1
figure 1

Single-cell mapping in GIST. A After dimensionality reduction and clustering analysis, 8 cell clusters were identified in GIST. B After annotation of marker genes, 8 cell types were identified in GIST. C Expression of marker genes in the 8 cell subtypes. D Proportion of the 8 cell subtypes in intra-tumor and peri-tumor tissues

Full size image

3.2 Biological differences of ICCs in intra-tumor and peri-tumor tissues

DEGs with altered expression in intra-tumor and peri-tumor tissues were identified and we performed GO and KEGG analysis. Specifically, the ICCs in intra-tumor tissues were significantly associated with cadherin binding, transcription regulator inhibitor activity, disordered domain specific binding, DNA -binding transcription activator activity, RNA polymerase II-specific, DNA-binding transcription activator activity. The ICCs in peri-tumor tissues was significantly associated with MHC class II protein complex binding, MHC protein complex binding, peptide antigen binding, MHC class II receptor activity, antigen binding. All these mentioned biological processes were closely associated with immune response (Fig. 2A). KEGG analysis revealed that the ICCs in intra-tumor tissues were involved in NGF-stimulated transcription, nuclear events (kinase and transcription factor activation) in the organism, attenuation phase, signaling by NTRK1 (TRKA), signaling by NTRKs pathway in the organism. The ICCs in peri-tumor tissues were involved in interferon signaling, interferon gamma signaling, interferon alpha/beta signaling, translocation of ZAP-70 to immunological synapse, and phosphorylation of CD3 and TCR zeta chains pathway (Fig. 2B). Subsequently, the pathways differentially activated between intra-tumor and peri-tumor tissues were compared by GSEA, and we found that biological processes such as cell cycle, response to hypoxia, and glycolytic process were activated in the ICCs in intra-tumor, while related biological processes such as immune response and antigen presentation were inhibited (Fig. 2C). This was also consistent with the previous results from GO and KEGG analyses that the ICCs in peri-tumor were actively involved in immune processes in the host body. Finally, HALLMARK pathway activities in the ICCs from different tissues were calculated by AUCell scores. The results showed that pathways related to cell cycle, reproduction, metabolism, cell signaling, stress response, growth regulation, and disease progression were more active in intra-tumor tissues, while those related to immune regulation, inflammatory response, cell proliferation, survival, and differentiation were more active in peri-tumor tissues (Fig. 2D).

Fig. 2
figure 2

Biological differences of ICC cells in intra-tumor and peri-tumor. A Biological processes involved in ICC cells in intra-tumor and peri-tumor. B Biological pathways involved in ICC cells in intra-tumor and peri-tumor. C Biological pathways involved in the differential activity of ICC cells in intra-tumor and peri-tumor. D AUCell score heatmap of pathway activity in ICC cells in intra-tumor and peri-tumor

Full size image

3.3 Heterogeneity of the ICCs in tumor tissues

To reveal the heterogeneity of ICCs, clustering analysis of the cells in GISTs was performed and we obtained 4 cell subpopulations, namely, subcluster 1, subcluster 2, subcluster 3, and subcluster 4 (Fig. 3A). Specifically, subcluster 1, subcluster 3, and subcluster 4 were slightly more in intra-tumor samples than peri-tumor, while subcluster 2 was more in peri-tumor samples (Fig. 3B). We found that the AUCell scores of the HALLMARK pathway were all higher in subcluster 1 but lower in subcluster 3. Compared to subcluster 1, OXIDATIVE_PHOSPHORYLATION, PROTEIN_SECRETION, ADIPOGENESIS, FATTY_ACID_METABOLISM, NOTCH_SIGNALING, INTERFERON_ALPHA_RESPONSE, APICAL_JUNCTION, and GLYCOLYSIS pathways were more active in subcluster 2 (Fig. 3C). These pathways were associated with cancer cell energy acquisition, immune escape, metastasis, and invasion.

Fig. 3
figure 3

Heterogeneity of ICC cells in tumor tissues. A Four cell subtypes of ICC cells exist in GIST, subcluster 1, subcluster 2, subcluster 3, and subcluster 4. B Proportion of the four cell subtypes in intra-tumor and peri-tumor samples. C Heatmap of AUCell score of HALLMARK pathway in the four cell subtypes

Full size image

3.4 Stem cell characterization score and potential regulatory mechanisms of the ICCs

The ICC were extracted for CytoTRACE analysis to further analyze the stem cell characteristics. According to their differentiation status, the cells were labeled with different colors from red (least differentiated) to blue (most differentiated) to present the heterogeneity of cells within the tumor. Specifically, subcluster 3 had the highest CytoTRACE score, indicating the lowest differentiation potential and the strongest stem cell characteristics. Subcluster 2 had the lowest CytoTRACE score, indicating the highest differentiation potential and least stemness characteristics (Fig. 4A, B). After defining high CSCs and low CSCs groups based on CytoTRACE score, we found that the cells in subcluster 3 had the lowest proportion of low CSCs (Fig. 4C). Cells characterized by high CSCs properties had the lowest percentage of antigen processing and presentation, antigen processing and presentation of peptide antigen, production of molecular mediator of immune response, antigen processing and presentation of exogenous antigen, activated antigen processing and presentation of exogenous peptide antigen pathways, and inhibited amide biosynthetic process, ribosome biogenesis, peptide biosynthetic process, translation, and cytoplasmic translation (Fig. 4D). SCENIC analysis was performed to further explore the TFs that play key regulatory roles between high and low CSCs groups. Nine important TFs were identified, including FOSB, ATF3, XBP1, CEBPD, REL, IRF1, STAT1, NCOR1, and PITX1. The AUC scores of XBP1 and CEBPD in the low CSCs group were higher than in the high CSCs group, indicating a higher transcriptional regulatory activity of XBP1 and CEBPD in the low CSCs group, but the transcriptional regulatory activity of STAT1 was higher in the high CSCs group than in the low CSCs group (Fig. 4E). The set of XBP1-regulated target genes actively participated in hepaticobiliary system development, cellular response to topologically incorrect protein, Golgi vesicle budding, vesicle budding from membrane, and face development (Fig. 4F). The set of CEBPD-regulated target genes were closely involved in rhythmic process, cellular response to steroid hormone stimulus, circadian regulation of gene expression, negative regulation of stem cell differentiation (Fig. 4G). The STAT1-regulated target gene set was actively involved in response to interferon-gamma, antigen processing and presentation of peptide antigen, type I interferon signaling pathway, cellular response to type I interferon, and response to type I interferon (Fig. 4H).

Fig. 4
figure 4

Stem cell characterization scores and potential regulatory mechanisms of ICC cells. A Cytotrace analysis identified high and low tumor stem cell populations in ICC cells, colored by predicted differentiation grade, from undifferentiated (red) to highly differentiated (blue). B Cytotrace scores in four ICC subpopulations. C Proportion of cells characterized by high CSCs and cells characterized by low CSCs in the four ICC subpopulations. D Biological pathways activated and inhibited in high CSCs. E Key transcription factor activities in the high and low CSCs characterized cell groups. F Biological processes involved in the XBP1-regulated target gene set. G Biological processes involved in the CEBPD-regulated target gene set. H Biological processes involved in the set of target genes regulated by STAT1

Full size image

3.5 Cell communication analysis

Finally, we performed cell communication analysis on the ICCs characterized by high and low CSCs with other 7 cell types in the GIST tissues to explore the potential interactions between the ICCs and with other cells. The results demonstrated that the ICCs characterized by high CSCs had a higher proportion and intensity of interactions with other cells (Fig. 5A). The ICCs characterized by high CSCs mainly had the cell-to-cell interactions with endothelial cells, macrophages 1, macrophages 2, MKI67+ progenitor cells, NK cells, and T cells via MIF-CD74 (CD44) (Fig. 5B). The ICCs characterized by low CSCs mainly interacted with MKI67+ progenitor cells, NK cells, and T cells via GZMA-F2RL2 (Fig. 5C). Finally, the interaction strength between the ICCs characterized by high and low CSCs and the 7 cell types were visualized into heatmaps (Fig. 5D, E). The results also confirmed that the ICCs characterized by high CSCs mainly interacted with other cells via MIF-CD74 (CD44), while the ICCs characterized by low CSCs mainly interacted with other cells via GZMA-F2RL2.

Fig. 5
figure 5

Cell communication analysis. A Bar graph showing the number of gene interactions and average interaction strength in the high CSCs and low CSCs subgroups. B Bubble graph demonstrating the strength of interactions between ICC cells characterized by high CSCs and the seven cell types. C Bubble plots demonstrating the interaction strength of ICC characterized by low CSCs with 7 types of cells. D Cell communication relationship heatmap demonstrating the interaction strength of ICC characterized by high and low CSCs with 7 kinds of cells via MIF-CD74 (CD44) receptor/ligand. E Cell communication relationship heatmap demonstrating the interaction strength of high and low characterized ICC with 7 kinds of cells via GZMA-F2RL2 receptor/ligand

Full size image

4 Discussion

In this study, the intra-tumor and peri-tumor tissues of GISTs were compared applying scRNA-seq study to analyze the cellular heterogeneity and TME. Eight cell clusters including B cells, endothelial cells, ICCs and macrophages were identified from the GIST samples. The proportion of ICCs in the tumor core was lower than in the marginal tumor tissues and the cells were involved in different biological processes. Previous study used the GSE162115 dataset from the GEO database to classify major cell types including T cells, macrophages, tumor cells, and NK cells for GISTs applying scRNA-seq technology. They found that the tumor cells could be divided into two groups, one was highly proliferative and associated with a high risk of, while the other was “dormant” and associated with a low metastasis risk [21]. Consistently, the present study observed a significant difference in the proportion and characteristics of the ICCs in the tumor core and marginal tissues. ICC are considered to be the cells of origin of GISTs and exhibit different behaviors depending on their location within the tumor, especially between the intra-tumor and peri-tumor regions. Compared to the peri-tumor region, intratumor ICCs are usually exposed to a more aggressive tumor environment characterized by higher rates of cell proliferation and hypoxia that can induce the activation of specific pathways in the ICCs [22]. For example, the ICCs in intra-tumor region are more related to pathways involved in cell cycle control and metabolic stress response, which are critical for cell survival and proliferation under stressful conditions. However, peri-tumor ICCs tend to interact more with immune cells and participate in immune-related pathways. This suggested that ICCs play a crucial role in regulating the immune environment around tumors and influences immune surveillance and tumor-host interactions. Thus, the locus-dependent behaviors of ICCs in GISTs may have a significant impact on the progression of the tumors and their responses to therapy, indicating a potential of targeting specific microenvironmental factors for the treatment of GISTs [23]. Furthermore, understanding the molecular mechanisms (particularly the KIT signaling pathway) through which ICCs adapt to these different conditions could also provide potential therapeutic targets. Study indicated that modulation of these pathways according to the microenvironmental context may be a more effective targeted therapy, especially when therapeutic responses vary with the type of KIT mutation and intra-tumor localization [24, 25]. These results contributed to the current understanding of GIST biology and highlighted the complexity of targeting the heterogeneity of GISTs and the need for personalized treatment.

A fundamental aspect of the functions of ICCs in GIST involves the KIT signaling pathway, which plays an important role in the survival and proliferation of ICCs. Mutations in the KIT gene are common in GISTs and will lead to sustained activation of downstream signaling pathways (Ras-Erk) to promote tumor growth and progression. Targeting these pathways, particularly through tyrosine kinase inhibitors (imatinib), has been the key to the treatment of GISTs, even through resistance often arises due to secondary mutations [24, 26]. Resistance to therapy (especially imatinib) is usually associated with reactivation of the MAPK signaling pathway that weakens the therapeutic effect, suggesting the need to counteract these adaptive responses in tumor cells. The protein LIX1 regulates the early adaptive response of GIST cells to targeted therapy by affecting MAPK signaling reactivation after treatment, suggesting the potential of using LIX1 protein as a therapeutic target to enhance the treatment efficacy of GISTs [27]. ICCs in GISTs are also involved in immune-associated pathways that regulate the immune environment around tumors. Understanding these interactions can help better design immunotherapeutic strategies [28]. These insights highlighted the complexity of ICC function in GISTs and emphasized the importance of targeting specific molecular pathways to improve therapeutic efficacy and overcome drug resistance. Further exploration and understanding of these pathways are also essential for the development of more effective GIST treatments.

Studies of ICC stemness features in GISTs have revealed their significant impact on tumor behaviors and treatment outcomes. Stem features in cancer cells including ICCs are associated with enhanced tumorigenicity (proliferation, metastasis, and resistance to conventional therapies). These features are regulated by complex signaling pathways, including key regulatory genes and the TME, which can alter gene expression and cellular behavior [29]. Stemness in ICCs is particularly affected by the KIT signaling pathway, a key component in the pathogenesis of GISTs. Mutations in the KIT gene lead to sustained activation of the signaling pathway to promote cancer cell survival and proliferation, which is the hallmark features of stem cell-like cancer cells. Studies suggested that overcoming drug resistance may require targeting other pathways that interact with or compensate for KIT signaling, for example, the Ras-Erk and PI3K-AKT pathways, so as to inhibit stem cell-driven tumor growth and improve therapeutic efficacy [24]. In addition, the TME plays a critical role in maintaining cancer stem cells by secreting various cytokines and growth factors to enhance stem cell self-renewal and survival. Targeting these microenvironmental interactions (possibly through immunomodulatory therapies) can suppress the stem cell properties of tumors and enhance patients’ responsiveness to treatment [30,31,32]. Finally, understanding and quantifying stem cells based on molecular markers and gene expression profiling could provide prognostic value and help design more effective therapies against tumors with high stem cell indices and overcome problems such as drug resistance and recurrence at the same time [33]. In conclusion, exploring the stemness features of ICCs in GISTs could enhance our understanding of tumor biology and facilitate the development of targeted therapies to reduce the stemness and overcome drug resistance, thereby improving the prognosis of patients with GISTs.

5 Limitations

Limitations in this study should be equally noted. Although the heterogeneity of ICCs in GIST tissues at different sites was analyzed in this study, the biological pathways and molecular mechanisms involved in different sites were still unclear, and the biological functions of ICC in intra-tumor and peri-tumor tissues should be verified using animal models and cellular assays, as the importance of method validation in the analysis of biomarker [34]. In addition, this study was conducted based on only a single public dataset, therefore future studies are encouraged to integrate multiple independent datasets from different patient populations to validate the current findings.