CXCL8/SDC1 axis mediates tumor stem cell interactions to drive remote transfer in thyroid cancer

Abstract

This study explores the molecular mechanisms behind the remote transfer of thyroid cancer (THCA) by investigating the interaction network of C-X-C motif chemokine ligand 8⁺ (CXCL8⁺ monocytes and syndecan-1⁺ (SDC1⁺) tumor stem cells using single-cell and spatial transcriptome sequencing. Tumor samples from THCA patients were analyzed using single-cell RNA sequencing (scRNA-seq), spatial transcriptome sequencing, and tumor tissue transcriptome analysis. Data were processed with Seurat and CellChat R packages, integrated via the SPOTlight package, and correlated with clinical data from the UCSC Xena database. Functional pathway enrichment analyses were performed using Gene Set Enrichment Analysis (GSEA), Gene Ontology (GO), and Kyoto Encyclopedia of Genes and Genome (KEGG). In vitro, a co-culture system of monocytes and THCA stem cells was developed, and protein levels were measured via enzyme-linked immunosorbent assay (ELISA) and Western blotting. The self-renewal and migration of follicular thyroid carcinoma (FTC) 238-S cells were assessed through sphere formation, colony formation, Cell Counting Kit-8 (CCK-8), and Transwell assays. In vivo, a subcutaneous tumor xenograft model and a lung metastasis model were established in nude mice. Transcriptomic analyses identified the CXCL8/SDC1 axis as a key mediator of Janus kinase-signal transducer and activator of transcription (JAK-STAT) signaling activation, promoting THCA stem cell self-renewal, invasion, and metastasis. CXCL8/SDC1 expression was significantly higher in the high-risk C1 subtype of THCA patients and correlated with a worse prognosis. In vitro and animal studies confirmed that the CXCL8/SDC1 axis drives tumor progression and metastasis. The interaction between CXCL8⁺ monocytes and SDC1⁺ tumor stem cells activates the JAK-STAT pathway, facilitating the remote transfer of THCA. Targeting the CXCL8/SDC1 axis may provide novel therapeutic strategies for improving THCA patient outcomes.

Graphical abstract

Highlights

• •In THCA, CXCL8⁺ monocytes may interact with SDC1⁺ tumor stem cells.
• •CXCL8 and SDC1 are highly expressed in high-risk C1 subtype tumor tissues of THCA patients and are significantly associated with malignant tumor classification.
• •CXCL8 may activate the JAK-STAT signaling pathway by interacting with SDC1 to promote remote transfer of THCA tumors.
• •The CXCL8/SDC1 axis can promote in vivo tumor formation and metastasis of THCA stem cells.
• •This study provides a new theoretical basis and molecular targets for treating THCA.

1. Introduction

Thyroid cancer (THCA) is one of the most common malignant tumors of the endocrine system, with an increasing incidence in recent years [1]. Although the prognosis for most THCA patients is relatively good, those with remote transfer do not respond well to treatment and have a poor prognosis [2,3]. Therefore, exploring the molecular mechanisms of its metastasis is important for developing new treatment strategies and improving patient outcomes.

The tumor microenvironment plays a crucial role in cancer development and metastasis, involving various cell types such as tumor cells, immune cells, and fibroblasts [4,5]. Recent studies increasingly indicate the significant role of monocytes in regulating the tumor microenvironment and associated signaling pathways. In the immune microenvironment of THCA, monocytes interact with other immune cells, such as macrophages and lymphocytes or secrete cytokines and chemokines, exerting a notable influence on tumor progression and immune surveillance mechanisms [6]. Studies suggest that monocytes play an inhibitory role in the anti-tumor immune microenvironment, potentially promoting cancer development by affecting the differentiation status of immune cells and their accumulation in the tumor microenvironment [7].

C-X-C motif chemokine ligand 8 (CXCL8) is an inflammatory cytokine widely secreted by monocytes, known for its role in promoting tumor proliferation, angiogenesis, and metastasis in various cancers [8,9]. In THCA, THCACXCL8 binds to its receptors C-X-C motif chemokine receptor 1 (CXCR1) and CXCR2, activating the phosphoinositide 3-kinase/protein kinase B (PI3K/AKT) and mitogen-activated protein kinase (MAPK) signaling pathways in tumor cells, thereby promoting cancer cell growth and survival [10]. Studies indicate that high expression of CXCL8 is significantly associated with poor prognosis in THCA patients [11], and CXCL8 can enhance tumor cell resistance to chemotherapy drugs [12].

Tumor stem cells possess high self-renewal ability and multilineage differentiation potential, crucial for tumor metastasis [13,14]. In THCA, tumor stem cells are vital in maintaining tumor heterogeneity and shaping the tumor microenvironment, which is crucial for disease progression [14]. Syndecan-1 (SDC1), CD138, is a cell surface protein widely expressed in tumors, particularly associated with tumor stem cells in various cancers [15]. Studies show that SDC1 in the lung microenvironment can induce breast cancer metastasis [16]. Although SDC1 is closely linked to the activity of tumors and stem cells in various cancers, research on how SDC1 influences the self-renewal and differentiation of tumor stem cells in THCA remains scarce. Furthermore, there is a lack of detailed characterization of spatial heterogeneity in the tumor microenvironment of THCA, limiting our understanding of the changes in tumor stem cell behavior under different environmental conditions. By employing high-resolution single-cell and spatial transcriptome analysis techniques, exploring the role of SDC1 in tumor stem cells of THCA and its specific function in the tumor microenvironment can provide a more profound molecular basis for precise treatment of THCA.

The Janus kinase-signal transducer and activator of transcription (JAK-STAT) pathway is a crucial regulator of various cellular processes, including proliferation, differentiation, and migration [17]. In the context of tumor progression, the JAK-STAT pathway drives tumor cell migration and invasion by promoting epithelial-mesenchymal transition (EMT), cytoskeletal remodeling, and tumor microenvironment remodeling [18]. Additionally, activation of this pathway in tumor stem cells enhances their self-renewal and migration capabilities, further exacerbating metastatic spread. Given its critical role in tumor progression, dysregulation of the JAK-STAT pathway is closely associated with poor prognosis in various cancers, including THCA [19,20].

This study utilizes single-cell transcriptome and space transcriptome sequencing technologies to provide a detailed depiction of cellular heterogeneity and interactions in the tumor microenvironment of THCA. Single-cell transcriptome sequencing technology can identify and analyze the gene expression characteristics of monocytes and tumor stem cells in tumors, while Space transcriptome sequencing technology offers spatial distribution information of cell expression data, crucial for understanding how cells interact in the physical space of the tumor microenvironment. This study can comprehensively analyze the dynamic interaction between CXCL8⁺ monocytes and SDC1⁺ tumor stem cells in the microenvironment by combining these two technologies. Additionally, applying this approach offers a new perspective to better understand the migration mechanisms of THCA and guide the development of potential therapeutic targets in the future.

2. Materials and methods

2.1. Single cell transcriptome sequencing and data analysis

Tumor tissue samples from THCA patients treated at our hospital between December 2021 and December 2022 were collected, including 2 samples from thyroid tumors within the gland (Situ_thyroid tumor: FTC1/2) and 2 samples from distant metastatic tumors (Metastatic_thyroid tumor: FTC3/4). This study was approved by Huashan Hospital, Fudan University’s clinical ethics committee, and informed consent was obtained from the patients, strictly adhering to the Helsinki Declaration (Approval No.: 2020-100). The experimental procedures and animal usage protocols were approved by the Animal Ethics Committee of Huashan Hospital, Fudan University (Approval No.: 2020-JS-344).

The collected tumor tissues were washed with cold phosphate-buffered saline (PBS) and digested with 1 mg/mL of collagenase (C2674, Sigma-Aldrich, St. Louis, MO, USA) at 37 °C for 10 min. Subsequently, trypsin/ethylenediamine tetraacetic acid (EDTA) (25200072, Gibco, Grand Island, NY, USA) was used at 37 °C for 5 min to prepare a single-cell suspension. The C1 Single-Cell Automated Preparation System (Fluidigm, South San Francisco, CA, USA) was used to capture individual cells, lysed within the chip, releasing messenger RNA (mRNA) that underwent reverse transcription to generate complementary DNA (cDNA). The lysed and reverse-transcribed cDNA underwent pre-amplification in a microfluidic chip for subsequent sequencing. The amplified cDNA was used for library construction and subjected to single-cell sequencing on the HiSeq 4000 Illumina platform (Illumina Inc., San Diego, CA, USA) with paired-end reads, with read length set at 2 × 75 base pair (bp), and approximately 20,000 reads per cell.

The single-cell RNA sequencing (scRNA-seq) data was analyzed using the Seurat software package. Firstly, quality control measures were applied with standards: nFeature_RNA > 500, nCount_RNA > 1000, nCount_RNA < 20,000, and percent.mt < 5. The data was normalized using the LogNormalize function, followed by principal component analysis (PCA) on the top 2000 highly variable genes using the RunPCA function. Important principal components (PCs) were selected for uniform manifold approximation and projection (UMAP) clustering analysis with JackStrawPlot and ElbowPlot functions. The FindAllMarkers function was used to identify marker genes for each cell cluster, and cell annotation was performed using the CellMarker database. Expression of marker genes in different cell clusters was visualized using FeaturePlot and VlnPlot functions. The “CellChat” R package was utilized for pathway activity analysis between different cells [21].

2.2. Space transcriptome sequencing and data analysis of THCA patient tumor tissue

The 10 × Genomics Visium platform was employed for spatial transcriptomic analysis. Fresh frozen human thyroid tissue sections (10 μm) were extracted from octamer-binding transcription factor 4 (OCT4)-embedded material and placed onto Visium spatial slides, followed by a 30-min permeabilization step to release mRNA. The mRNA was bound to spatial barcode oligonucleotides on the slide bottom and underwent reverse transcription according to the manufacturer’s protocol. Libraries prepared from these cDNA samples were sequenced on the Illumina NextSeq 2000 platform (San Diego, CA, USA), generating over 400 million reads per slice, with over 50,000 reads per position. Spaceranger software (version 3.1.0, 10 × Genomics, Pleasanton, CA, USA) was used to align each position on the Visium spatial transcriptomic slide to the GRCh38 human genome reference data to obtain raw counts.

The 10x Visium spatial transcriptomic data were analyzed using Seurat’s Load10X_spatial function to integrate the spatial transcriptomic raw gene expression matrix, location information, and tissue hematoxylin and eosin (H&E) images, creating a Seurat object. After normalization, PCA was conducted to reduce dimensions to the top 20 PCs. The FindAllMarkers function within Seurat was used to detect marker genes and analyze differential gene expression. The FindSpatially Variable Features function, with default settings, was used to identify genes with spatially variable expression.

To spatially locate cells in cancer tissues from THCA patients with different origins, an anchor-based integration pipeline within Seurat was employed to integrate the combined scRNA-seq dataset with the 10x Visium spatial transcriptomic data. This facilitated the transfer of cell type annotations from scRNA-seq to spatial transcriptomics. Cell type predictions were loaded into Seurat, and the SPOTlight R package was used for annotating and visualizing cell types at each spatial location [22].

2.3. THCA tissue transcriptome sequencing

Tumor tissue samples from 7 THCA patients who underwent surgery at our hospital were collected. Among them, 3 cases had no metastasis, while 4 had distant metastases. RNA high-throughput next-generation sequencing was conducted for the study. Total RNA was extracted using Trizol reagent (15596026, Invitrogen, Carlsbad, CA, USA). The concentration and purity of RNA samples were determined using the Nanodrop 2000 spectrophotometer (1011U, ThermoFisher, Waltham, MA, USA). RNA samples meeting specific criteria (RNA integrity number (RIN) ≥ 7.0 and 28S:18S ratio ≥1.5) were used for downstream experiments.

The sequencing libraries were generated and sequenced by CapitalBio Technology (Beijing, China). Each sample used 5 μg of RNA. The Ribo-Zero™ Magnetic Kit (MRZE706, Epicentre Technologies, Madison, WI, USA) was used to remove ribosomal RNA (rRNA) from total RNA. The NEB Next Ultra RNA Library Prep Kit (#E7775, NEB, Ipswich, MA, USA) was used to construct libraries for sequencing. RNA was fragmented in NEB Next First Strand Synthesis Reaction Buffer (5 × ) to approximately 300 bp fragments. First-strand cDNA was synthesized using reverse transcriptase primers and random primers, followed by second-strand cDNA synthesis in the second-strand synthesis reaction buffer containing deoxyuridine triphosphate (dUTP) Mix (10 × ). The cDNA fragments were end-repaired, and polyA tails and sequencing adapters were added. The USER Enzyme (#M5508, NEB, Ipswich, MA, USA) was used to digest the second strand of cDNA to build strand-specific libraries. Library DNA was amplified, purified, and subjected to polymerase chain reaction (PCR) enrichment. Library identification was performed using Agilent 2100 (Agilent, Santa Clara, CA, USA), and quantification was conducted with the KAPA Library Quantification Kit (KK4844, KAPA Biosystems, Wilmington, MA, USA). Finally, paired-end sequencing was carried out on the NextSeqCN500 platform (Illumina, San Diego, CA, USA).

Quality control of the raw sequencing data was performed using FastQC software v0.11.8. Preprocessing of the raw data was done with Cutadapt software 1.18, which removed Illumina sequencing adapters and poly(A) tail sequences. Reads with N content exceeding 5% were discarded using a Perl script. The FASTX Toolkit software 0.0.13 was used to extract reads with a base quality of over 20, covering 70% of the bases. Paired sequences were fixed using the BBMap software. Finally, the filtered high-quality read fragments were aligned to the reference genome using Hisat2 software (version 0.7.12) [23].

2.4. Protein-protein interaction (PPI) analysis

Differential gene expression data were imported into the STRING database platform (https://string-db.org/), with Homo sapiens selected as the research species for PPI analysis. The interaction confidence score was set at medium confidence (0.700), and hidden unconnected nodes and other parameters were left at default settings [24].

2.5. The Cancer Genome Atlas (TCGA) data acquisition

RNA-Seq data for THCA were downloaded from the UCSC Xena database (Santa Cruz, CA, USA; https://xena.ucsc.edu/) along with phenotype and clinical prognosis data for THCA cancer tissue samples. The RNA-Seq dataset included 510 cancer tissue samples and 58 adjacent normal tissue samples. Samples were grouped using Perl scripts, and the ensemble IDs of the samples were converted using the GENCODE Gene Set-09.2019 version annotation file. Ensemble IDs for long noncoding RNAs (lncRNAs) and mRNAs not included in the GENCODE database were excluded. As the data from the UCSC Xena database are publicly available, no ethical approval or informed consent was required for this study.

Differential expression analysis was conducted using the “limma” package in R software, with differential P-values corrected using the false discovery rate (FDR) method. A filtering threshold of FDR < 0.05 was applied to identify significantly differentially expressed proteins. Visualizations, including heatmaps and volcano plots, were generated using the “pheatmap” and “ggplot2” packages in R. Kinase-substrate enrichment analysis was performed with the “KSEAapp” package, and the kinase-substrate network was visualized and mapped using Cytoscape v3.6.0 software (San Diego, CA, USA). All analyses were conducted using R version 4.2.1 (Vienna, Austria; R Foundation for Statistical Computing) [25].

2.6. Differential gene selection

Gene differential expression analysis was performed on the sequencing and TCGA datasets using the “limma” package in R software. The FDR method was applied to correct differential P-values, with a threshold of FDR < 0.05 for selecting significant differentially expressed genes. Heatmaps and volcano plots were generated using the “pheatmap” and “ggplot2” packages. Data comparisons between groups were performed using the Wilcox test. All analyses were conducted with R version 4.2.1 (Vienna, Austria; R Foundation for Statistical Computing) [25].

2.7. Weighted gene co-expression network analysis (WGCNA)

WGCNA was performed using the “WGCNA” package in R software. Initially, hierarchical clustering analysis was carried out using the Hclust function. A suitable soft threshold β was selected using the “pickSoftThreshold” function, and the adjacency matrix was transformed. The Topological Overlap Matrix (TOM) was computed, and a hierarchical clustering dendrogram was constructed to divide gene expression into modules, with 50 set as the minimum gene count per module. Modules with potentially similar traits were merged by defining 0.25 as the cut height. Module eigengene (ME) were summarized, and their correlation with traits was calculated [26].

2.8. Survival analysis

Survival analysis was performed using TCGA data containing survival time, status, and gene expression information for THCA patients. The “survival” package in R software was used, with a significance threshold of P < 0.05 considered meaningful [27].

2.9. THCA patient tumor classification

Subtype analysis of THCA patients was conducted based on prognostic gene expression using the “ConsensusClusterPlus” package in R. Differential protein analysis was performed within high-risk subtypes [28].

2.10. Clinical relevance analysis

Clinical relevance analysis was carried out based on THCA patient tumor classification and clinical characteristics. The analysis was visualized using the “ComplexHeatmap” package in R software (Vienna, Austria; R Foundation for Statistical Computing) [26].

2.11. Gene Set Enrichment Analysis (GSEA) analysis

Samples were grouped based on THCA patient tumor classification, and GSEA was performed to reveal pathway differences enriched between the two groups. Gene sets go.cc_analysis.Gsea.1689305913241, go.mf_analysis.Gsea.1689304082899, and kegg_analysis.Gsea.1689303926237 from MSigDB was used as a reference [29].

2.12. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional enrichment analysis

The “ClusterProfiler” package in R was used to conduct GO and KEGG enrichment analysis on differential genes, and the results were visualized using the “ggplot2” package. Differential gene analysis focused on cellular functions and signaling pathways primarily enriched with a significance level of P < 0.05 [30].

2.13. Construction and grouping of lentiviruses

To acquire The lentivirus interference vector, pSIH1-H1-copGFP(sh-), an interference vector, SI501A-1 from System Biosciences in Palo Alto, CA, USA was purchased. Additionally, the lentivirus overexpression vector, pCDH-CMV-MCS-EF1α-copGFP(oe-), another product from System Biosciences in the USA, was obtained to construct lentivirus-based vectors for SDC1 gene interference and CXCL8 gene overexpression. The lentivirus packaging kit (A35684CN, Invitrogen, Carlsbad, CA, USA) was used to package the lentivirus particles in HEK-293T cells (iCell-h237, Cyagen Biosciences, Shanghai, China). The cell culture supernatant was collected 48 h later as lentivirus, with a titer of 1 × 10⁸ TU/mL. Specifically, the sequences used were as follows: short hairpin RNA-negative control (sh-NC) sequence: AGGCTACAATGATCAGACTAAT; short hairpin RNA-SDC1 (sh-SDC1) sequence: CCGCAAATTGTGGCTACTAAT; sh-SDC1-2 sequence: GAGCAGGACTTCACCTTTGAA.

2.14. Cell culturing and grouping

THCA cells FTC238 (MZ-8205, Ningbo Mingzhou Biotechnology Co., Ltd., Ningbo, China) were cultivated using 10% Dulbecco's Modified Eagle's Medium (DMEM) basic fetal bovine serum (FBS, S9020, Beijing Solaibao Technology Co., Ltd., Beijing, China). Tohoku Hospital Pediatrics-1 (THP-1) cells (CL-0233, Procell, Wuhan, China) were cultured in RPMI1640 medium containing 10% FBS and 1% penicillin/streptomycin.

Once the FTC238 cells reached 90% confluency, CD133+cells were isolated using immunomagnetic beads (130-097-049, Miltenyi Biotec, Bergisch Gladbach, Germany), obtaining THCA stem cells (FTC238-S). The CD133+ cells were cultured in serum-free DMEM/F12 medium, and the stem cell markers CD133 and Nestin were identified using immunofluorescence techniques.

According to the research requirements, FTC238-S was divided into the following 5 groups: (1) sh-NC (FTC238-S cells transfected with sh-NC), (2) sh-SDC1 (FTC238-S cells transfected with sh-SDC1), (3) PBS + sh-NC (FTC238-S cells transfected with sh-NC and supplemented with an equal amount of PBS), (4) CXCL8 + sh-NC (FTC238-S cells transfected with sh-NC and treated with CXCL8), and (5) CXCL8 + sh-SDC1 (FTC238-S cells transfected with sh-SDC1 and treated with CXCL8). Specifically, 1 mL of the corresponding lentivirus was added to the FTC238-S cells, followed by CXCL8 treatment at a concentration of 100 ng/mL for 24 h (200-08M-5UG, ThermoFisher, Waltham, MA, USA) [31]. Monocytes were divided into two groups: oe-NC (monocytes transfected with oe-NC) and oe-CXCL8 (monocytes transfected with oe-CXCL8). The operation was to add 1 mL of the corresponding lentivirus to the monocytes and then check the lentivirus infection effect after 48 h [32].

2.15. Construction and grouping of co-culture systems

Monocytes from each group were incubated in RPMI 1640 medium (21875091, ThermoFisher, Waltham, MA, USA) for 24 h. The conditioned medium was collected and used to co-culture FTC238-S cells. To block or neutralize CXCL8, 10 μg/mL of anti-CXCL8 (ab289967, Abcam, Cambridge, UK) was added to the conditioned medium, with anti-immunoglobulin G (anti-IgG) (ab172730, Abcam, Cambridge, UK) as a control [33]. Additionally, the JAK-STAT pathway inhibitor SD-1008 (HY-107595, MCE, Shanghai, China) was used to divide cells into dimethyl sulfoxide (DMSO) and SD-1008 groups. SD-1008 (10 μM) was applied to cells for 30 min treatment, while DMSO was used as a control.

The grouping of the co-culture system was as follows: (1) M_oe-NC + S_sh-NC (FTC238-S cells transfected with sh-NC and treated with conditioned medium from monocytes transfected with oe-NC), (2) M_oe-CXCL8 + S_sh-NC (FTC238-S cells transfected with sh-NC and treated with conditioned medium from monocytes transfected with oe-CXCL8), (3) M_oe-CXCL8 + S_sh-SDC1 (FTC238-S cells transfected with sh-SDC1 and treated with conditioned medium from monocytes transfected with oe-CXCL8), (4) M_oe-CXCL8 + S_DMSO (FTC238-S cells treated with conditioned medium from monocytes transfected with oe-CXCL8 and supplemented with an equal amount of DMSO), and (5) M_oe-CXCL8 + S_SD_1008 (FTC238-S cells treated with conditioned medium from monocytes transfected with oe-CXCL8 and supplemented with SD-1008).

2.16. Immunofluorescence staining

CD133+ FTC238 cells were seeded in 12-well cell culture plates. After attachment, the medium was aspirated, and the cells were washed with PBS three times for 5 min each. Cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and blocked with 10% donkey serum. Skin tissue sections were permeabilized with 0.5% Triton X-100, blocked with 5% bovine serum albumin (BSA), and then incubated overnight with the primary antibody CD133 (ab19898, Abcam, Cambridge, UK) at 4 °C. After washing with PBS, the sections were incubated with the secondary antibody goat anti-rabbit IgG H&L (ab150079, Abcam, 1:500, Cambridge, MA, USA) at room temperature for 1 h. 4',6-diamidino-2-phenylindole (DAPI) staining was performed for 5 min before slides were mounted with anti-fade mounting medium and observed under a fluorescence microscope (FV-1000/ES, Olympus, Tokyo, Japan) [34].

2.17. Enzyme-linked immunosorbent assay (ELISA) experiment

ELISA was conducted following the kit instructions (ab174446, Abcam, Cambridge, UK). Standards were prepared, and 100 μL of samples were added to wells and incubated at 37 °C for 90 min. After sequential incubation with biotinylated antibody and enzyme conjugate reagent, the substrate solution was added and incubated for 15 min at 37 °C. The reaction was terminated with a stop solution, and optical density (OD) values were measured using a microplate reader (BioTek Synergy 2, Winooski, VT, USA). A standard curve was plotted to determine CXCL8 levels in the samples [35].

2.18. Western blot

Total protein from the cells was extracted using RIPA lysis buffer containing PMSF (P0013C, Biyuntian, Shanghai, China). The lysates were incubated on ice at 4 °C for 30 min, centrifuged at 8000 g for 10 min, and the supernatant was collected. Protein concentration was measured using a BCA assay kit (23227, ThermoFisher, Waltham, MA, USA). Fifty micrograms of protein were dissolved in 2 × sodium dodecyl sulfate (SDS) loading buffer, boiled for 5 min, and separated by SDS-polyacrylamide gel electrophoresis (PAGE). Proteins were transferred to a polyvinylidene difluoride (PVDF) membrane and blocked with 5% skim milk at room temperature for 1 h. The PVDF membrane was incubated overnight at 4 °C with primary antibodies against CXCL8 (1:1000, ab235584, Abcam, Cambridge, UK), SDC1 (1:2000, ab128936, Abcam, Cambridge, UK), NESTIN (1:100, ab105389, Abcam, Cambridge, UK), OCT4 (1:10000, ab200834, Abcam, Cambridge, UK), SRY-Box transcription factor 2 (SOX2, 1:1500, ab92494, Abcam, Cambridge, UK), phosphorylated STAT3 (p-STAT3) (1:1000, NB100-82213, Novus Biologicals, Centennial, CO, USA), STAT3 (1:1000, MAB1799, Novus Biologicals, Centennial, CO, USA), phosphorylated-JAK2 (p-JAK2) (1:1500, ab32101, Abcam, Cambridge, UK), JAK2 (1:1000, ab108596, Abcam, Cambridge, UK), nuclear factor kappa B (NF-κB, 1:1000, 8242T, CST, Boston, MA, USA), phosphorylated NF-κB (p-NF-κB) (1:1000, 3033T, CST, Boston, MA, USA), glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (1:2500, ab9485, Abcam, Cambridge, UK), and Tubulin (1:5000, ab7291, Abcam, Cambridge, UK). After washing with Tris-buffered saline with Tween-20 (TBST) three times for 10 min each, the membrane was incubated with an horseradish peroxidase (HRP)-conjugated goat anti-rabbit immunoglobulin G (IgG) heavy and light chains (H&L) secondary antibody (1:2000, ab97051, Abcam, Cambridge, UK) for 1 h. The membrane was treated with a mixed solution from the enhanced chemiluminescence (ECL) Fluorescence Detection Kit (abs920, Abcam, Cambridge, UK) in a darkroom. Imaging and analysis were performed using Quantity One v4.6.2 software (Bio-Rad, Hercules, CA, USA). The relative protein content was normalized to GAPDH as an internal control. Each experiment was repeated three times, and the average was calculated.

2.19. Suspension culture experiment of stem cells

FTC238-S cells from each group were seeded at a density of 2 × 10⁷ cells/mL into low-adhesion 96-well plates. Cells were cultured in serum-free DMEM/F12 medium (11320033, ThermoFisher, Waltham, MA, USA) supplemented with 20 ng/mL EGF (PHG0311, ThermoFisher, Waltham, MA, USA) and 20 ng/mL FGF-β (100-18B-1 MG, ThermoFisher, Waltham, MA, USA). A partial medium change was performed every two days. After 10 days of continuous culture, cells were observed, photographed, and counted under an inverted light microscope CKX41 (Olympus, Tokyo, Japan).

2.20. Clonogenic assay

FTC238-S cells were seeded in 6 cm culture dishes at 2000 cells per dish. After 14 days of culture in fresh medium, colonies were stained with 0.5% (wt/vol) crystal violet (C8470, Beijing Solarbio Science & Technology Co., Ltd., Beijing, China) for quantitative analysis. Colonies with more than 50 cells per cluster were counted as clones. The experiment was repeated three times [36].

2.21. Transwell assay

For the invasion experiment, Matrigel gel (356234, Shanghai Haoyang Biotechnology Co., Ltd., Shanghai, China) was thawed overnight at 4 °C. Under 4 °C conditions, 200 μL of Matrigel gel was diluted with 200 μL of serum-free culture medium, and 50 μL was added to each Transwell chamber. Plates were incubated for 2–3 h at 37 °C until the gel solidified. Cells were digested, counted, and suspended in a serum-free culture medium. A 200 μL cell suspension was added to the upper chamber, while 800 μL of culture medium containing 20% FBS was added to the lower chamber. After incubation at 37 °C for 24 h, non-invading cells were removed from the upper surface with a cotton swab. The remaining cells were fixed with formaldehyde for 10 min, washed, and stained with 0.1% crystal violet for 30 min. Cells were observed and counted under an inverted microscope (Nikon TE2000, Tokyo, Japan).

Migration experiments were performed similarly, but without Matrigel coating, and incubation lasted 24 h. Cell counting was done in four randomly selected microscopic fields. Each experiment was repeated three times [37].

2.22. Mouse tumor model

Fifty-four 6-week-old BALB/c nude mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd (401, Beijing, China) and individually housed in specific pathogen-free (SPF) grade animal facilities with a humidity level of 60%–65% and a temperature of 22–25 °C. After an adaptation period of one week, the mice’s health status was observed before the experiments commenced. The experimental procedures and animal usage protocols were approved by the Animal Ethics Committee of the Department of Thyroidology, Yanjing Medical College, Capital Medical University (Approval No.: 202011013). The study involved the construction of subcutaneous transplantation tumors, primary tumors, and lung metastasis models in nude mice. The mice were divided into four groups: (1) anti-IgG group (tail vein injection of 100 μL 50 μg IgG twice a week after tumor model construction), (2) anti-CXCL8 group (tail vein injection of 100 μL 50 μg anti-CXCL8 twice a week after tumor model construction), (3) S_sh-SDC1 group (FTC238-S cells transfected with sh-NC for tumor model construction), and (4) S_sh-SDC1 group (FTC238-S cells transfected with sh-SDC1 for tumor model construction).

Subcutaneous Xenograft Tumor Model in Nude Mice: FTC238-S cells from each group were prepared into a cell suspension with a concentration of 2 × 10⁷ cells/mL. Using a 1 mL syringe, 0.2 mL of the cell suspension was injected subcutaneously into the left axilla of nude mice. After injection, all nude mice were housed in a specific pathogen-free (SPF) animal facility. On the 28th day post-inoculation, the nude mice from each group were euthanized at the cervical dislocation site, and the tumor tissues were removed and weighed. Simultaneously, the occurrence of lymph node metastasis in the axillary region of the nude mice from each group was recorded.

Primary Tumor Model: FTC238-S-lucifer cells (2 × 10⁷) were injected into the thyroids of the mice after 4 weeks for bioluminescence imaging. Bone metastasis was observed using H&E staining upon euthanasia.

Lung Metastasis Model: FTC238-S-lucifer cells (2 × 10⁷) were intravenously injected into the mice for bioluminescence imaging of lung metastasis. Lung tissues were collected on day 28 post-injection and analyzed using H&E staining to quantify liver metastasis [38,39].

2.23. H&E staining

Tissue sections of lung and bone from the different mouse models were processed for H&E staining following the manufacturer’s protocol (PT001, Shanghai Bogoo Biotechnology Co., Ltd., Shanghai, China). The staining steps were conducted as follows: Hematoxylin staining was performed at room temperature for 10 min, followed by rinsing with running water for 30–60 s. Sections were differentiated in 1% hydrochloric acid alcohol for 30 s, rinsed under running water, and soaked for 5 min. Eosin staining was performed at room temperature for 1 min. Sections were then dehydrated in an alcohol gradient (70%, 80%, 90%, 95%, and 100%), with each step lasting 1 min, and cleared in xylene for 1 min. Two rounds of transparency were conducted in xylene I and II for 1 min each. Finally, the sections were mounted with neutral resin in a fume hood. Morphological changes in the lung tissues of each group were observed under an optical microscope (Olympus BX50, Tokyo, Japan) [40].

2.24. Statistical analysis

Data analysis was performed using GraphPad Prism 9 (GraphPad Software, Inc., San Diego, CA, USA) and R language (Vienna, Austria; R Foundation for Statistical Computing). Quantitative data were represented as mean ± SD. Statistical significance was determined using unpaired Student’s t-test or the Wilcox test for non-normally distributed data. One-way analysis of variance (ANOVA) followed by Tukey’s post hoc test was used for multiple group comparisons, with significance set at P < 0.05.

3. Results

3.1. Transcriptomic sequencing reveals potential interaction between CXCL8 and SDC1 activating JAK-STAT signaling pathway to promote remote transfer of THCA tumors

THCA is a common malignant tumor, and metastasis is a critical factor affecting its prognosis. Therefore, we conducted comprehensive transcriptomic sequencing, single-cell sequencing, and Space transcriptome sequencing to unravel the mechanisms of THCA metastasis and identify therapeutic targets. The bioinformatics flowchart is depicted in Fig. 1.

Fig. 1.

Flowchart for analyzing the letter portion of this study. WGCNA: weighted gene co-expression network analysis; CXCL8: C-X-C motif chemokine ligand 8; SDC1: syndecan-1.

Initially, transcriptomic sequencing analysis was performed on three non-metastatic THCA cases and four metastatic THCA cases. Differential analysis revealed (Figs. 2A and B) 459 significantly downregulated genes and 559 significantly upregulated genes in the metastatic group compared to the non-metastatic group, with the heatmap displaying the top 50 differentially expressed genes. Subsequently, the differential genes were input into the STRING database to construct a PPI network. Only two protein pairs, mitochondrial assembly of ribosomal large subunit 1-mitochondrial ribosomal protein L34 (MALSU1-MRPL34) and CXCL8-SDC1, demonstrated physical interaction (Fig. 2C). The chemokine CXCL8, a member of the chemotactic factor family, has been reported to play a significant role in the progression and development of THCA [41]; however, there is scarce documentation regarding the involvement of SDC1.

Fig. 2.

Exploration of possible molecular mechanisms of malignant progression of thyroid cancer (THCA) tumors through transcriptome sequencing. (A) Volcano plots of differentially expressed genes in the non-metastatic group (3 cases) and metastatic group (4 cases); red dots represent upregulated genes, blue dots represent downregulated genes. (B) Heatmap of gene expression levels in the non-metastatic group (3 cases) and metastatic group (4 cases), with a hierarchical clustering dendrogram on the left based on gene expression levels, color scale on the right, where red indicates upregulated genes, blue indicates downregulated genes, and the upper histogram shows blue for the non-metastatic group and red for the metastatic group. (C) Protein-protein interaction (PPI) network constructed from differentially expressed genes. (D) Box plots showing the differential expression of C-X-C motif chemokine ligand 8 (CXCL8) in the non-metastatic group (3 cases) and metastatic group (4 cases). (E) Box plots displaying the differential expression of syndecan-1 (SDC1) in the non-metastatic group (3 cases) and metastatic group (4 cases). (F) Volcano plots of differentially expressed genes in the low-SDC1 group (2 cases) and high-SDC1 group (2 cases), with red dots representing upregulated genes and blue dots representing downregulated genes. (G) Heatmap of gene expression levels in the low-SDC1 group (2 cases) and high-SDC1 group (2 cases), with a hierarchical clustering dendrogram on the left based on gene expression levels, color scale on the right indicating red for upregulated genes, blue for downregulated genes, and the upper histogram showing blue for the low-SDC1 group and red for the high-SDC1 group. (H) Pathways enriched by Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of differentially expressed genes, with the color scale on the right and circle size representing the number of enriched genes. MALSU1: mitochondrial assembly of ribosomal large subunit 1; MRPL34: mitochondrial ribosomal protein L34; JAK-STAT: Janus kinase-signal transducer and activator of transcription.

In the sequencing results, both CXCL8 and SDC1 showed significantly elevated expressions in the metastatic group (Figs. 2D and E). Furthermore, we conducted differential analysis on the metastatic group samples based on the high and low expressions of SDC1 (Figs. 2F and G). KEGG analysis results revealed (Fig. 2H) that these differentially expressed genes were significantly enriched in the JAK-STAT signaling pathway, which is significantly associated with cancer pathways.

In conclusion, it is speculated that CXCL8 may activate the JAK-STAT signaling pathway by interacting with SDC1 to promote THCA tumor remote transfer.

3.2. High expression of CXCL8 and SDC1 in high-risk subtype tumor tissues of THCA patients, significantly associated with tumor migration and invasion

To further investigate the molecular mechanisms regulating THCA remote transfer, we downloaded a THCA-related transcriptome dataset from the UCSC Xena database. Differential analysis revealed (Fig. 3A) 385 significantly downregulated genes and 495 significantly upregulated genes in THCA tissues compared to normal tissues. WGCNA analysis identified the turquoise module genes significantly positively correlated with tumor tissues (Fig. 3B). Subsequently, prognostic-related genes were selected from the turquoise module, resulting in 20 genes significantly related to prognosis (Table S1), dividing THCA patients into two subtypes, C1 and C2 (Fig. 3C).

Fig. 3.

Differential expression of C-X-C motif chemokine ligand 8 (CXCL8) and syndecan-1 (SDC1) in the C1/C2 subtypes of thyroid cancer (THCA) patients. (A) Volcano plot of differentially expressed genes in the THCA transcriptome dataset, with red dots indicating upregulated genes and blue dots indicating downregulated genes. The THCA tissue samples (510 cases) are compared to normal control tissue samples (58 cases). (B) Heatmap showing the correlation between modules and traits, with each cell containing the corresponding correlation and P-value. The color scale on the right indicates red for positive correlation and blue for negative correlation between modules and traits. (C) Hierarchical clustering diagram of THCA subtypes: C1 subtype includes 369 cases, C2 subtype includes 132 cases, with 9 cases removed due to missing survival information out of the initial 510 THCA patients for survival analysis. (D) Heatmap depicting the clinical relevance of THCA C1/C2 subtypes, where clinical traits are divided by the corresponding THCA C1/C2 subtype classification indicated below. Clinical traits annotated with ∗P < 0.05, ^∗∗P < 0.01, ^∗∗∗P < 0.001. T: Tumor; N: Node; M: Matastasis. (E) Distribution of clinical traits in C1/C2 subtypes. (F) Pathways enriched by Gene Set Enrichment Analysis (GSEA) in the C1/C2 subtypes. C1 subtype consists of 369 cases, and C2 subtype consists of 132 cases. (G) Volcano plot of differentially expressed genes in the THCA patients' C1/C2 subtypes, with red dots representing upregulated genes and blue dots representing downregulated genes. (H) Heatmap showing the differential gene expression levels in THCA patients' C1/C2 subtypes, with a hierarchical clustering dendrogram on the left based on gene expression levels, color scale on the right indicating red for upregulated genes, blue for downregulated genes, and the upper histogram showing blue for C2 subtype cancer tissue samples (132 cases) and red for C1 subtype cancer tissue samples (369 cases). (I) Violin plot illustrating the differential expression of CXCL8 and SDC1 in the C1/C2 subtypes. (J) Violin plot showing the differential expression of CXCL8 and SDC1 in different tumor stages.

To investigate the relationship between C1/C2 subtypes and the degree of malignancy in patients with THCA, we conducted an in-depth exploration. Initially, through a clinical correlation analysis of the C1/C2 subtypes, we found (Figs. 3D and E) that THCA patients with the C1 subtype exhibited a higher proportion of advanced tumor node metastasis staging (TNM) grading and staging compared to those with the C2 subtype. This suggests a potential association between the C1 subtype and more aggressive tumor characteristics, leading us to consider the C1 subtype as a high-risk subtype for THCA patients. Additionally, utilizing GSEA to investigate the enriched pathways in the C1/C2 subtypes, the results revealed (Fig. 3F) that pathways related to tumor migration and invasion (CELL_ADHESION_MEDIATOR_ACTIVITY, CYTOKINE_CYTOKINE_RECEPTOR_INTERACTION, JAK-STAM_SIGNALING_PATHWAY) were aberrantly activated in the C1 subtype. Overall, this indicates a significant enrichment of pathways related to tumor migration and invasion in the high-risk C1 subtype of THCA patients.

To further validate the roles of CXCL8 and SDC1 in the malignant progression of THCA tumors, we examined the expression of CXCL8 and SDC1 in the C1/C2 subtypes. Differential analysis was performed to identify the differentially expressed genes in tumor tissues of the C1/C2 subtypes (Figs. 3G and H), revealing (Fig. 3I) that CXCL8 and SDC1 were highly expressed in tumor tissues of the high-risk C1 subtype. Furthermore, CXCL8 and SDC1 showed significantly elevated expression in tumors with higher malignant grades (Fig. 3J). In conclusion, CXCL8 and SDC1 are highly expressed in tumor tissues of the high-risk C1 subtype of THCA patients and are significantly associated with tumor migration and invasion.

3.3. Significant heterogeneity of cells in THCA tissue and communication between CXCL8⁺ monocytes and SDC1⁺ tumor stem cells

In recent years, single-cell sequencing technology has provided a powerful tool for the in-depth exploration of tumor heterogeneity and metastatic mechanisms [42]. To further investigate the cellular-level mechanisms of the CXCL8/SDC1 axis, we conducted scRNA-seq analysis on THCA. The sequenced samples included 2 thyroid tumor samples (Situated thyroid tumor: FTC1/2) and 2 tumor samples with distant metastases (Metastatic thyroid tumor: FTC3/4). Utilizing the “Seurat” package to analyze scRNA-seq data, low-quality cells were removed after quality control and normalization (Fig. S1A). The correlation coefficients of nCount with percent. mt and nCount with nFeature were −0.07 and 0.94, respectively (Fig. S1B), indicating good cell quality post-filtering. Screening for the top 2000 highly variable genes in THCA tissues revealed gene enrichment with high variability (Fig. S1C). Dimensionality reduction analysis did not show batch effects (Fig. S1D). Using the JackStraw and JackStrawPlot functions, evaluation was conducted on the first 40 PCs, with a selection of the top 6 PCs for UMAP analysis to demonstrate the significant differences by showcasing the key gene composition (Figs. S1E and F) and a heatmap (Fig. S1G). The analysis through UMAP further examined the cellular heterogeneity in THCA tissue, revealing (Fig. S2A) a clustering of all cells into 17 distinct cell clusters. Grouping of cell clusters based on the origin of THCA samples showed, in the results (Figs. S2B and C), no significant differences between cell clusters from Situ_thyroid tumor and Metastatic_thyroid tumor groups. Additionally, the expression profiles of the top 4 specific marker genes for each cell cluster in THCA tumor tissue were depicted (Fig. S2D).

Subsequently, annotation of cell clusters based on known cell marker genes classified the cellular components in the tumor microenvironment into 7 cell types (Figs. 4A and S2E): epithelial cells (KRT7+), T cells (CCL5+), monocyte (C1QC+), B cells (MS4A1+), smooth muscle cells (RGS5+), endothelial cells (FLT1+), and cancer stem cells (SOX9+). Further analysis of the proportion of each cell type in THCA sample origins, as shown in Figs. 4B and C, revealed no significant differences between the cell types in the Situated thyroid tumor and Metastatic thyroid tumor groups. However, individual cell type proportions varied greatly among different THCA samples, such as epithelial cells, T cells, monocytes, and B cells. Subsequent analysis, illustrated in Fig. 4D, showed significantly high expression of CXCL8 in Monocytes, while SDC1 exhibited significantly high expression in cancer stem cells and epithelial cells. The connection between SDC1 in tumor stem cells and CXCL8 in monocytes was particularly interesting. To investigate the potential communication between monocytes and cancer stem cells, a comparison was made between the Situated thyroid tumor and Metastatic thyroid tumor groups regarding cellular communication, revealing that compared to the Situated thyroid tumor group, the number and strength of interactions between cells were higher in the Metastatic thyroid tumor group (Figs. 4E and F). Notably, the interaction intensity between Monocytes and cancer stem cells in the Metastatic thyroid tumor group was significantly enhanced (Fig. 4G).

Fig. 4.

Cell annotation and communication analysis of thyroid cancer (THCA) tissue samples' single-cell RNA sequencing (scRNA-seq) data. (A) 17 cell clusters annotated as 7 different cell types. (B) Display of 7 cell types categorized by Situ_thyroid tumor group and Metastatic_thyroid tumor group. (C) Proportions of different cell types in Situ_thyroid tumor group and Metastatic_thyroid tumor group of THCA tissue. (D) Expression levels of C-X-C motif chemokine ligand 8 (CXCL8) and syndecan-1 (SDC1) in different cell types, with darker shades of blue indicating higher gene expression levels. (E) Cell communication network graphs in Situ_thyroid tumor and Metastatic_thyroid tumor tissue samples. (F) Statistical comparison of the differences in the number and strength of cell interactions between Situ_thyroid tumor group and Metastatic_thyroid tumor group. (G) Circular graphs depicting the differences in the number and strength of cell interactions between Situ_thyroid tumor group and Metastatic_thyroid tumor group, with blue lines indicating significant downregulation in the number and strength of interactions in Metastatic_thyroid tumor group compared to Situ_thyroid tumor group, and red lines indicating significant upregulation.

In conclusion, the results indicate significant cellular heterogeneity in THCA tissues, with CXCL8 primarily expressed in Monocytes and SDC1 predominantly expressed in tumor stem cells, suggesting communication and interaction.

3.4. CXCL8⁺ monocytes may regulate THCA remote transfer by interacting with SDC1⁺ tumor stem cells

Various cellular and non-cellular components in the tumor microenvironment collectively drive tumor growth, invasion, metastasis, and response to treatment [43]. The latest spatial profiling technology, spatial transcriptome sequencing (ST-seq), provides a systematic view and elucidates the physics of the tumor microenvironment components.

To confirm the communication between monocytes and tumor stem cells in scRNA-seq, ST-seq was performed on frozen sections of Situ_thyroid tumor and Metastatic_thyroid tumor tissues. After integrating the ST-seq data using the Seurat package, low-quality cells were removed (Figs. S3A and B). CellCycleScoring was used to calculate the cell cycle (Fig. S3C) and then standardized (Figs. S3C and D). The top 3000 high variable genes were selected based on variance (Fig. S3E), and dimensionality reduction was carried out using PCA (Fig. S3F), followed by evaluation of PC standard deviations using ElbowPlot (Fig. S3G). The top 6 PC-related gene expression heat maps are shown in Fig. S3H.

Furthermore, we utilized the t-distributed stochastic neighbor embedding (t-SNE) algorithm to reduce nonlinear dimensionality on the top 30 PCs, followed by cluster analysis with resolution of 0.4. By assessing the overlap between genes specifically mapped to certain regions and genes of specific cell types identified via scRNA-seq data, we inferred the enrichment of specific cell types in given tissue regions. Annotations were performed on cells in the ST-seq data, highlighting the distribution of monocytes and cancer stem cells. Comparing the distribution of monocytes and cancer stem cells between Situ_thyroid tumor and Metastatic_thyroid tumor revealed a higher coincidence in Metastatic_thyroid tumor (Figs. 5A and B). Additionally, using the R language spatial positioning of cells by transcriptomics (“SPOTlight”) package, information on cell interactions in space was obtained, and a circle graph depicting the strength of cell-cell interactions revealed that cancer stem cells interact with monocytes (Fig. 5C). Further analysis of the colocalization of CXCL8 and SDC1 in Situ_thyroid tumor and Metastatic_thyroid tumor demonstrated partial overlap of CXCL8 and SDC1 expression sites in Situ_thyroid tumor, whereas nearly complete overlap was observed in Metastatic_thyroid tumor (Figs. 5D and E). Combined with the previous findings on the distribution of monocytes and cancer stem cells, the enhanced interaction between CXCL8⁺ monocytes and SDC1⁺ cancer stem cells in the remote transfer tissue was highlighted.

Fig. 5.

Analysis results of integrated single-cell RNA sequencing (scRNA-seq) and spatial transcriptome sequencing (ST-seq). (A) Distribution of different cell types on tissue slices of Situ_thyroid tumor group and separate distributions of Monocyte and syndecan-1 (SDC1)⁺ cancer stem cells in ST-seq data, with the distribution proportions of different cells in each spot represented by pie charts. (B) Distribution of different cell types on tissue slices of Metastatic_thyroid tumor group and separate distributions of Monocyte and SDC1⁺ cancer stem cells in ST-seq data, with the distribution proportions of different cells in each spot represented by pie charts. (C) Circle graph showing the strength of interactions between different cells in ST-seq data, where thicker and darker lines indicate stronger interaction strengths. (D) Expression distribution of C-X-C motif chemokine ligand 8 (CXCL8) and SDC1 on tissue slices of Situ_thyroid tumor group in ST-seq data. (E) Expression distribution of CXCL8 and SDC1 on tissue slices of Metastatic_thyroid tumor group in ST-seq data.

Based on the scRNA-seq results, we hypothesize that CXCL8⁺ monocytes may regulate THCA remote transfer by interacting with SDC1⁺ tumor stem cells.

3.5. CXCL8⁺ monocytes may facilitate self-renewal and migration invasion of tumor stem cells by interacting with SDC1⁺ tumor stem cells

The conclusion drawn from the bioinformatic analyses suggests a crucial regulatory role of the potential interaction between CXCL8⁺ monocytes and SDC1⁺ tumor stem cells in THCA remote transfer. To further validate this conclusion and ensure its biological significance, additional experiments are conducted in vitro.

It is known that tumor stem cells can maintain tumor growth and promote the spread of primary tumors to distant tissues, leading to tumor recurrence and metastasis [44]. Initially, we isolated CD133+ THCA stem cells (FTC238-S) from THCA cells FTC238 through immunomagnetic bead sorting. Immunofluorescence analysis (Fig. S4A) revealed positive expression of stem cell markers CD133 and Nestin in the sorted cells, indicating successful isolation of FTC238-S. Subsequently, we subjected FTC238-S to grouped treatment with lentiviruses carrying CXCL8 and SDC1. Western blot analysis (Figs. S4B and C) demonstrated that overexpression of CXCL8 significantly increased its expression in monocytes, while silencing of SDC1 led to a significant decrease in its expression in FTC238-S cells, with sh-SDC1-1 showing the most efficient silencing effect for subsequent experiments.

Next, FTC238-S cells were treated with CXCL8 (Fig. S4D), and their self-renewal ability was assessed using Western blot, sphere formation assays, and colony formation assays. The results (Figs. S4E–G) indicated that following CXCL8 treatment, the stemness markers NESTIN, OCT4, and SOX2 expression levels in FTC238-S significantly increased. Additionally, the number of spheres and clones formed by FTC238-S significantly increased, suggesting a notable enhancement in the self-renewal capacity of FTC238-S. Silencing SDC1 reversed the impact of CXCL8 treatment on the self-renewal ability of FTC238-S. Evaluation of cell proliferation and migration invasion capabilities of FTC238-S using Cell Counting Kit-8 (CCK-8) and Transwell assays showed that following CXCL8 treatment, the proliferation and migration invasion abilities of FTC238-S significantly increased (Figs. S4H and I). Silencing SDC1 reversed the effects of CXCL8 treatment on the proliferation and migration invasion capabilities of FTC238-S. These findings indicate that the CXCL8/SDC1 axis promotes the self-renewal and migration invasion of tumor stem cells.

To further confirm that monocytes can regulate the characteristics of tumor stem cells by secreting the chemokine CXCL8, monocytes overexpressing CXCL8 were cultured, and the cell-conditioned medium was collected. ELISA results revealed (Fig. 6A) a significant increase in CXCL8 levels in the cell-conditioned medium after CXCL8 overexpression, indicating a substantial elevation in CXCL8 secretion by monocytes. Subsequently, these monocytes overexpressing CXCL8 were co-cultured with SDC1-silenced FTC238-S cells (Fig. 6B). Functional assays demonstrated that following the overexpression of CXCL8 in monocytes, the stemness (Fig. 6C), self-renewal (Figs. 6D and E), proliferation (Fig. 6F), and migration invasion capabilities (Fig. 6G) of FTC238-S significantly increased. Conversely, silencing SDC1 in FTC238-S reversed the self-renewal, proliferation, and migration invasion capabilities of FTC238-S, indicating that monocytes can promote the self-renewal and migration invasion of tumor stem cells through the CXCL8/SDC1 axis.

Fig. 6.

Impact of monocytes on tumor stem cell self-renewal and migration invasion through the C-X-C motif chemokine ligand 8/syndecan-1 (CXCL8/SDC1) axis. (A) Enzyme-linked immunosorbent assay (ELISA) detection of CXCL8 content in cell-conditioned media. (B) Schematic illustration of co-culturing monocytes with FTC‑238 Human Follicular Thyroid Carcinoma Cells, Lung Metastasis (FTC238-S) cells. (C) Protein expression levels of stemness markers neuroepithelial stem cell protein (NESTIN), octamer-binding transcription factor 4 (OCT4), and SRY-box transcription factor 2 (SOX2) in FTC238-S cells from each group examined by Western blot. (D) Cell sphere formation assay assessing the sphere formation capability of FTC238-S cells in co-culture systems of each group. (E) Clonogenic assay evaluating the clonogenic capacity of FTC238-S cells in co-culture systems of each group. (F) Cell Counting Kit-8 (CCK-8) assay measuring the proliferation ability of FTC238-S cells in co-culture systems of each group. (G) Transwell assay determining the migration and invasion capability of FTC238-S cells in co-culture systems of each group, ∗P < 0.05 compared between the two groups, all cell experiments were repeated three times. M_oe-NC + S_sh-NC: FTC238-S cells transfected with short hairpin RNA-negative control (sh-NC) and treated with conditioned medium from monocytes transfected with oe-NC; M_oe-CXCL8 + S_sh-NC: FTC238-S cells transfected with sh-NC and treated with conditioned medium from monocytes transfected with oe-CXCL8; M_oe-CXCL8 + S_sh-SDC1: FTC238-S cells transfected with sh-SDC1 and treated with conditioned medium from monocytes transfected with oe-CXCL8; OD: optical density.

Additionally, the addition of anti-CXCL8 to the medium of monocytes to block or neutralize CXCL8 effects, followed by co-culturing with FTC238-S (Fig. S5A), resulted in significantly reduced self-renewal (Figs. S5B and C), proliferation (Fig. S5D), and migration invasion capabilities (Fig. S5E) of FTC238-S. This implies that inhibiting CXCL8 secreted by monocytes can suppress the self-renewal and migration invasion capabilities of tumor stem cells, consistent with the co-culture results and further confirming that monocytes can affect the functionality of tumor stem cells through the CXCL8 secretion pathway.

3.6. CXCL8⁺ monocytes may activate the JAK-STAT signaling pathway by interacting with SDC1⁺ tumor stem cells

To investigate whether the interaction between CXCL8⁺ monocytes and SDC1⁺ tumor stem cells can activate the JAK-STAT signaling pathway in tumor stem cells, we conducted Western blot experiments to assess the expression of JAK-STAT signaling pathway-related proteins in co-culture systems. The results revealed (Fig. 7A) that overexpression of CXCL8 in monocytes led to a significant increase in the phosphorylation levels of JAK2 and STAT3 in FTC238-S. Furthermore, silencing SDC1 in FTC238-S resulted in a substantial decrease in the phosphorylation levels of JAK2 and STAT3. Additionally, the addition of anti-CXCL8 in the conditioned medium resulted in a notable reduction in the phosphorylation levels of JAK2 and STAT3 in FTC238-S cells. There were no changes in the expression of NF-κB pathway proteins, indicating that the interaction between CXCL8⁺ monocytes and SDC1⁺ tumor stem cells can activate the JAK-STAT signaling pathway in tumor stem cells (Fig. 7B). Moreover, treatment of FTC238-S cells with the JAK-STAT signaling pathway inhibitor SD-1008 (Fig. 7C) demonstrated (Fig. 7D) a significant decrease in the phosphorylation levels of JAK2 and STAT3 post-treatment, indicating the inhibitory effect of SD-1008 on the JAK-STAT signaling pathway. Functional experiments (Figs. 7E–I) revealed that SD-1008 could reverse the effects of CXCL8 overexpression on the stemness, self-renewal, proliferation, migration and invasion capabilities of FTC238-S cells.

Fig. 7.

Impact of the Janus kinase-signal transducer and activator of transcription (JAK-STAT) signaling pathway on tumor stem cell self-renewal and migration invasion. (A, B) Western blot analysis of JAK-STAT and nuclear factor kappa B (NF-κB) signaling pathway-related proteins: protein expression levels of phosphorylated-Janus kinase 2 (p-JAK2), JAK2, phosphorylated signal transducer and activator of transcription (p-STAT3), STAT3, NF-κB, and phosphorylated NF-κB (p-NF-κB) in FTC238-S cells co-cultured with different modified myeloid‑derived suppressor cells (MDSCs) (A), and protein levels of p-JAK2, JAK2, p-STAT3, and STAT3 in FTC238-S cells treated with anti-immunoglobulin G (anti-IgG) or anti-C-X-C motif chemokine ligand 8 (anti-CXCL8) antibodies (B). (C) Schematic diagram of monocyte treatment and co-culturing with FTC238-S cells. (D) Western blot analysis of JAK-STAT signaling pathway-related proteins in FTC238-S cells from each group. (E) Protein expression levels of stemness markers neuroepithelial stem cell protein (NESTIN), octamer-binding transcription factor 4 (OCT4), and SRY-box transcription factor 2 (SOX2) in FTC238-S cells from each group examined by Western blot. (F) Cell sphere formation assay assessing the sphere formation capability of FTC238-S cells in co-culture systems of each group. (G) Clonogenic assay evaluating the clonogenic capacity of FTC238-S cells in co-culture systems of each group. (H) Cell Counting Kit-8 (CCK-8) assay measuring the proliferation ability of FTC238-S cells in co-culture systems of each group. (I) Transwell assay determining the migration and invasion capability of FTC238-S cells in co-culture systems of each group, ∗P < 0.05 compared between the two groups, and all cell experiments were repeated three times. M_oe-NC + S_sh-NC: FTC238-S cells transfected with short hairpin RNA-negative control (sh-NC) and treated with conditioned medium from monocytes transfected with oe-NC; M_oe-CXCL8 + S_sh-NC: FTC238-S cells transfected with sh-NC and treated with conditioned medium from monocytes transfected with oe-CXCL8; M_oe-CXCL8 + S_sh-SDC1: FTC238-S cells transfected with short hairpin RNA-SDC1 (sh-SDC1) and treated with conditioned medium from monocytes transfected with oe-CXCL8; M_oe-CXCL8 + S_DMSO: FTC238-S cells treated with conditioned medium from monocytes transfected with oe-CXCL8 and supplemented with an equal amount of dimethyl sulfoxide (DMSO); M_oe-CXCL8 + S_SD_1008: FTC238-S cells treated with conditioned medium from monocytes transfected with oe-CXCL8 and supplemented with SD-1008; THP-1 cells: Tohoku Hospital Pediatrics-1 cells; OD: optical density.

In conclusion, the interaction between CXCL8⁺ monocytes and SDC1⁺ tumor stem cells may activate the JAK-STAT signaling pathway, promoting tumor stem cells' self-renewal, migration and invasion abilities.

3.7. The CXCL8/SDC1 axis may promote THCA stem cell tumorigenesis and tumor growth and metastasis by activating the JAK-STAT signaling pathway

Finally, we investigated the effects of the CXCL8/SDC1 axis on the tumorigenesis, tumor growth, and metastasis of THCA stem cells in vivo through the establishment of subcutaneous xenograft tumors, primary tumor models, and lung metastasis models (Fig. 8A). We observed that blocking CXCL8 or silencing SDC1 significantly inhibited the JAK-STAT signaling pathway (Fig. 8B). Additionally, blocking CXCL8 or silencing SDC1 markedly suppressed the growth of subcutaneous and primary tumors in nude mice (Figs. 8C and D). Further assessment of subcutaneous xenograft tumor axillary lymph node metastasis and lung metastasis via tail vein injection revealed that blocking CXCL8 or silencing SDC1 led to a significant reduction in axillary lymph node metastasis of subcutaneous xenograft tumors (Fig. 8E), primary tumor bone metastasis (Fig. 8F), and lung metastasis (Figs. 8G and H). These results indicate that the CXCL8/SDC1 axis may mediate the activation of the JAK-STAT signaling pathway to promote THCA stem cell tumorigenesis, tumor growth, and metastasis in vivo.

Fig. 8.

Impact of the C-X-C motif chemokine ligand 8/syndecan-1 (CXCL8/SDC1) axis on tumor initiation, growth, and metastasis of thyroid cancer (THCA) stem cells in Vivo. (A) Diagram of the in vivo animal experiment protocol, with the green syringe representing anti-immunoglobulin G (anti-IgG) and anti-CXCL8 treatments. (B) Western blot analysis of Janus kinase-signal transducer and activator of transcription (JAK-STAT) signaling pathway-related proteins in subcutaneous tumor tissues of nude mice from each group. (C) Gross anatomy of subcutaneous transplant tumors in nude mice (left) and corresponding weight statistics (right). (D) Images of primary tumors in nude mice from each group (left) and diameter analysis (right). (E) Statistical analysis of axillary lymph node (LN) metastasis in subcutaneous transplant tumors of nude mice from each group. (F) Hematoxylin and eosin (H&E) staining to assess bone metastasis of primary tumors in nude mice from each group. (G) Images of lung metastasis in nude mice from each group (left) and diameter analysis (right). (H) H&E staining to evaluate lung metastasis in nude mice from each group. ∗P < 0.05 compared between two groups, ^∗∗P < 0.01 with 6 nude mice in each group. p-JAK2: phosphorylated-JAK2; p-STAT3: phosphorylated STAT3; GAPDH: glyceraldehyde 3-phosphate dehydrogenase; S_sh-NC: FTC238-S cells transfected with short hairpin RNA-negative control (sh-NC); S_sh-SDC1: FTC238-S cells transfected with short hairpin RNA-SDC1 (sh-SDC1).

4. Discussion

In the history of THCA research, a continuous focus has been on how tumor cells influence the ability to transfer remotely. In recent years, more studies have begun to investigate the tumor microenvironment and its interactions with tumor stem cells [45,46]. This study delves into an in-depth exploration of the interaction between CXCL8⁺ monocytes and SDC1⁺ tumor stem cells in THCA, providing new insights into remote transfer.

The study found that CXCL8⁺ monocytes may activate the JAK-STAT signaling pathway by interacting with SDC1⁺ tumor stem cells, promoting tumor remote transfer. This result aligns with previous research on other cancer types, indicating the vital role of CXCL8 and its related pathways in cancer progression and metastasis [10,47]. However, applying this mechanism specifically to THCA brings important new understanding to the field. By further validating the significance of the CXCL8/SDC1 axis in THCA using data from the TCGA database, it was found that migration and invasion-related pathways were significantly enriched in high-risk C1 subtype patients. Similar to findings in other cancers in the TCGA database, this signaling pathway may be a common mechanism promoting cancer [48].

Furthermore, the study identified the interactive relationship of the CXCL8/SDC1 axis in the tumor microenvironment through the combination of single-cell and Space transcriptome sequencing technologies. Subsequent in vitro cell experiments confirmed that the interaction between CXCL8⁺ monocytes and SDC1⁺ tumor stem cells could activate the JAK-STAT signaling pathway, further enhancing tumor stem cells' self-renewal and migration and invasion capabilities. This experimental evidence supported the role of this axis in THCA remote transfer. Animal results demonstrated that this axis promoted tumorigenesis, growth, and metastasis of THCA stem cells in vivo, enhancing our understanding of the functioning of this signaling pathway in a biological context. While the involvement of SDC1 and its related pathways in the progression and metastasis of other cancers has been established, research specific to THCA remains relatively scarce. This study not only fills this gap but also corroborates the key role of the CXCL8/SDC1 axis in THCA through various methods.

We recognize that the close relationship between monocytes and macrophages in the tumor microenvironment might lead to confusion. Through single-cell and spatial transcriptomic analyses, our study confirmed that CXCL8 is predominantly expressed by C1QC⁺ monocytes and not by CD68⁺ or CD163⁺ macrophages. This finding highlights CXCL8⁺ monocytes as a distinct immune cell population that directly interacts with SDC1⁺ tumor stem cells, promoting their migration and invasion via JAK-STAT pathway activation. While our study focuses on the role of CXCL8⁺ monocytes, the potential involvement of macrophages in the CXCL8/SDC1 axis was not explored in detail. Monocytes recruited to the tumor microenvironment can differentiate into macrophages, which play critical roles in immunosuppression, matrix remodeling, and tumor migration. Future studies could investigate whether CXCL8 indirectly influences the tumor microenvironment by modulating macrophages, thereby enhancing the understanding of the CXCL8/SDC1 axis.

This study, through single-cell and spatial transcriptomic data, confirmed that CXCL8 is primarily secreted by C1QC⁺ monocytes, with relatively low expression in macrophages, neutrophils, and other immune cells. Functional experiments further validated that CXCL8⁺ monocytes directly interact with SDC1⁺ tumor stem cells, significantly enhancing the self-renewal, migration, and invasion capabilities of tumor stem cells. Blocking CXCL8 or silencing SDC1 markedly suppressed these effects, indicating that the CXCL8-SDC1 axis is a critical mechanism regulating tumor stem cell functions.Nonetheless, CXCL8 secreted by other immune cells, such as neutrophils or macrophages, might exert synergistic effects on the tumor microenvironment, particularly by indirectly modulating tumor stem cell functions through interactions with other immune or stromal cells. This possibility was not thoroughly explored in this study. Future research could utilize multi-cell co-culture systems and single-cell isolation experiments to further validate the roles of other CXCL8 sources. Such studies will contribute to a more comprehensive understanding of the multi-layered roles of CXCL8 in the tumor microenvironment. This content has been incorporated into the revised manuscript to enrich the discussion of the study’s findings. Thank you for your valuable input.

In addition, the PPI network analysis identified another physically interacting protein pair, MALSU1 and MRPL34. MALSU1 is a mitochondrial ribosome assembly factor that plays a key role in mitochondrial protein synthesis and metabolic homeostasis, while MRPL34 is a component of the mitochondrial large ribosomal subunit. Previous studies suggest that these proteins may function cooperatively in regulating mitochondrial dysfunction and metabolic reprogramming in cancer cells. The identification of this pair hints at a possible role of mitochondrial translation and energy metabolism in the distant metastasis of THCA. However, due to limited existing research on these genes in thyroid cancer, their function was not further explored in this study. Future investigations are warranted to elucidate their potential involvement in THCA metastasis.

Considering the importance of the CXCL8/SDC1 axis in THCA, future research may delve into how to design drugs and optimize treatment strategies targeting this pathway. Additionally, further studies could explore the role of this axis in other cancers or diseases to provide more information for clinical applications. Based on the collective research findings, we can preliminarily conclude (Fig. 9) that CXCL8⁺ monocytes in THCA may activate the JAK-STAT signaling pathway by interacting with SDC1⁺ tumor stem cells to promote tumor remote transfer. The integrated analysis of single-cell and Space transcriptome sequencing and in vitro cell and in vivo animal experiments provide new theoretical foundations and molecular targets for treating THCA, revealing that targeting the CXCL8/SDC1 axis may be an effective therapeutic approach for THCA.

Fig. 9.

Schematic diagram of the molecular mechanism by which monocytes promote self-renewal, migration, and invasion of thyroid cancer (THCA) stem cells through the C-X-C motif chemokine ligand 8/syndecan-1 (CXCL8/SDC1) axis. JAK2: Janus kinase 2; STAT3: signal transducer and activator of transcription 3.

This study has deeply elucidated the interaction mechanism between CXCL8⁺ monocytes and SDC1⁺ tumor stem cells in THCA by employing a combination of single-cell and Space transcriptome sequencing technologies. This in-depth molecular mechanistic research aids the scientific community in better understanding the occurrence, development, and metastasis processes of THCA. The study confirms that the CXCL8/SDC1 axis and the JAK-STAT signaling pathway play critical roles in the remote transfer of THCA, opening up possibilities for developing novel therapeutic strategies targeting these molecules or signaling pathways.

CRediT authorship contribution statement

Wenjuan Wang: Writing – original draft, Formal analysis, Data curation, Conceptualization. Jian Zhou: Writing – original draft, Project administration, Methodology, Investigation. Baorui Tao: Writing – review & editing, Supervision, Software, Resources. Ning Kong: Writing – review & editing, Visualization, Validation, Supervision. Jie Shao: Writing – review & editing, Visualization, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: