We have generated transcriptome expression profiles for 33 tumor tissues, excluding those lacking information on paracancerous tissues. By analyzing the data from both tumor and paracancerous tissues, we initially compared the alterations in CTHRC1 expression levels. Additionally, we extracted the expression patterns of the invasion gene set across various tumor tissues. Subsequently, we conducted a correlation analysis between CTHRC1 expression in different tumors and the invasion gene set. Finally, we assessed the association between CTHRC1 expression in diverse tumors and gene expression of immunosuppressant therapeutic targets.

The distinct characteristics of CTHRC1 in TNBC and non-TNBC were comprehensively elucidated. Breast cancer samples were classified into TNBC and non-TNBC subtypes, followed by an initial analysis of the association between CTHRC1 expression and tumor T-stage. The expression of CTHRC1 was stratified into high and low groups based on the median value. We examined the differential expression of the invasive gene set in conjunction with the trend in CTHRC1 expression. Furthermore, we corroborated this trend by analyzing expression data from multiple GEO databases for patients with TNBC.

Immunotherapy is a pivotal therapeutic modality for tumor patients, and thus we investigated the correlation between CTHRC1 expression and immune cell infiltration in TNBC patients (15). The abundance and activity of immune cell infiltration in triple-negative breasts were comprehensively analyzed using advanced algorithms including CIBERSORT, ssGSEA, and ESTIMATE. TIDE has emerged as a widely adopted tool for predicting immunotherapy response in tumor patients (16, 17). Moreover, we conducted predictive analyses to assess the potential efficacy of targeting CTHRC1 in patient immunotherapy.

The dataset GSE176078 consists of transcriptomic data obtained from single cells of nine patients diagnosed with TNBC. Initially, the data were processed individually to create Seurat objects. Rigorous quality control measures were implemented on individual samples to exclude those exhibiting (1) mitochondrial gene expression exceeding 10%, (2) bicellular contamination, and (3) erythrocyte gene expression surpassing 3%. Furthermore, cell cycle effects were accounted for and eliminated from the analysis. Subsequently, the harmony function was employed to integrate the datasets, followed by dimensionality reduction using Unified Mobility Approximation and Project (UMAP), enabling visualization of clustered cells in a two-dimensional map. Cell types were identified utilizing “SingleR” along with relevant high-quality literature references (18–21). Finally, specific cell types of interest underwent further dimensionality reduction clustering and comprehensive annotation at the subpopulation level.

GSVA enrichment analysis is widely used to assess the activity scores of gene sets in samples. We evaluated the correlation between 14 gene sets related to cell biology and CTHRC1. Firstly, we selected the expression matrix of TNBC and used the GSVA algorithm to calculate the scores of the 14 gene sets in the samples. Finally, we analyzed the correlation between CTHRC1 and their activity levels. In single-cell data, we annotated detailed cell types present in the tumor microenvironment of triple-negative breast cancer. Initially, we extracted “RNA” data from single-cell sequencing results and then assessed the activity scores of these gene sets in different cellular subsets using GSVA method.

The Monocle2 method utilizes unsupervised learning techniques on the transcriptome expression matrix of individual cells to categorize them into separate branches of the developmental path. Initially, we investigated the dynamics of CTHRC1 during cell differentiation, and subsequently employed Plot pseudotime heatmap to visualize gene expression patterns within the feature gene set across cellular evolution.

The communication between cells is achieved through the interaction of ligands and receptors. Within the tumor microenvironment, intercellular communication plays a pivotal role in facilitating immune evasion by tumor cells. CellChat utilizes gene expression levels to infer the pathways and magnitude of intercellular communication, while also characterizing the interactions between ligands and receptors. Overexpressed genes were identified, screened for ligand-receptor pairs, and mapped onto the protein-protein interaction network. To ensure high-quality data, a filter was applied excluding communication relationships involving low-quality cells with a minimum cell count of 3.

Based on bioinformatics discoveries, we corroborated our findings using multiple methodologies. Multiple immunofluorescence assays were utilized to authenticate the interactions between distinct molecules and cells.

The associations between variables were assessed using Pearson or Spearman coefficients. T-tests were utilized to compare continuous variables with a normal distribution, whereas the Mann-Whitney U test was employed for variables that did not follow a normal distribution. Categorical variables were compared using either the chi-square test or Fisher exact test. All data analysis was performed using R software (version 4.1.2, available at https://www.r-project.org/). Statistical significance was determined based on a two-sided P-value threshold of less than 0.05 (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

Significantly elevated expression levels of CTHRC1 were observed in breast, esophageal, and liver cancers compared to corresponding paraneoplastic tissues. Notably, breast cancer exhibited particularly high levels of CTHRC1 ( Figure 1A, B ). In non-TNBC tissues, the difference in CTHRC1 expression was not statistically significant. However, CTHRC1 expression was markedly elevated in TNBC tissues. Additionally, as the tumor T-stage and N-stage advanced, the expression of CTHRC1 exhibited an upward trend ( Figure 1C–G ). CTHRC1 was closely linked to the biology of TNBC.

For multiple samples, we utilized the GSVA enrichment score to evaluate the invasive gene set of tumors. The results demonstrated a strong positive correlation between CTHRC1 expression and tumor invasion scores specifically in TNBC. Conversely, no significant correlation was observed between CTHRC1 expression and tumor invasion scores in non-TNBC tissues. Conversely, there was no similar pattern found in non-TNBC. After conducting an in-depth analysis, we proceeded to investigate whether the invasive gene set displayed similar characteristics in both TNBC and non-TNBC. Intriguingly, the expression of a majority of genes within the invasive gene set exhibited an upward trend with elevated CTHRC1 gene expression specifically in the TNBC cohort ( Figure 2 ). CTHRC1 was served as a uniquely distinctive biological molecule in TNBC relative to non-TNBC.

In TNBC, CTHRC1 exhibited limited association with most immune cells. However, there was a significant difference in the abundance of macrophage infiltration between the CTHRC1 high-expression and low-expression groups ( Figure 3A ). The infiltration abundance of M2-type macrophages significantly increased with higher levels of CTHRC1 expression ( Figure 3C ). The ESTIMATE algorithm indicated no disparities in immune scores but revealed discrepancies in mesenchymal scores between the high and low expressing groups of CTHRC1 ( Figure 3B, D ). TIDE predicted that patients in the high-expression group of CTHRC1 would derive less benefit from immunotherapy compared to those with lower expression levels. Similarly, while there were no variations observed in T-cell escape between the high and low expressing groups of CTHRC1, a difference was noted in M2-type macrophage scores ( Figure 3E ). CTHRC1 played a crucial role in the distinct tumour microenvironment of TNBC.

The GSE176078 dataset comprises single-cell sequencing results from 9 patients diagnosed with TNBC. Each sample underwent rigorous quality control, and the data were integrated using the “Harmony” package in R programming language. Our study encompassed a total of 24272 cells and investigated expression patterns across 29626 genes. Cell type classification was performed based on established literature in the field of high-quality single-cell sequencing, revealing intriguingly predominant expression of both CTHRC1 and invasion-related genes in CAFs. Notably, no detectable expression was observed in immune cells, tumor-associated macrophages, or malignant cells. CTHRC1 exhibits a distinctive expression pattern in CAFs.

Based on these findings, we isolated tumor-associated fibroblasts and macrophages from the single-cell data for recolonization analysis ( Figure 4A-E ). We identified three subtypes of macrophages, M0, M1 and M2. Furthermore, based on the expression levels of CTHRC1 in CAFs, they were categorized into high-expression and low-expression groups ( Figure 5A, B ). GSVA enrichment analysis of the single-cell data revealed that only cells in the high-expression group exhibited the highest activity of the invasive gene set ( Figure 5C, D ). To validate this observation, transcriptomic data was utilized to confirm a strong correlation between CTHRC1 expression and activity of genes associated with tumor invasion. Finally, AUCell was employed to score cellular invasion gene set activity, demonstrating a significantly higher score for the high expression group compared to the low expression group ( Figure 5E-G ). Single-cell analysis revealed a significant positive correlation between CTHRC1 expression and tumour invasion score.

The tumor microenvironment undergoes dynamic changes. We utilized the “Monocle” package to elucidate the dynamic interplay between CTHRC1 and the invasive gene set ( Figure 6A-C ). CAFs were classified into five distinct differentiation trajectories, wherein the expression of CTHRC1 exhibited a significant upregulation concomitant with CAFs differentiation ( Figure 6D, E ). This augmented expression of CTHRC1 was accompanied by a notable increase in the expression levels of most members within the invasive gene sets ( Figure 6F ). Alterations in CTHRC1 expression were associated with modifications in the invasive activity of TNBC.

Receptor-ligand interactions serve as the foundation of intercellular communication. Utilizing the ‘Cellchat’ function for single-cell data analysis, potential signaling between cells was speculated upon ( Figures 7A, B ). CTHRC1 modulates communication between tumor-associated fibroblasts and macrophages. The group with high levels of CTHRC1 demonstrated increased functionality in the complement signaling pathway and annexin signaling pathway when compared to the group with low levels of CTHRC1. Conversely, a contrasting phenomenon was observed in the PTN signaling pathway, PROS signaling pathway, ANGPTL signaling pathway, and CHEMERIN signaling pathway ( Figure 7C-H ). CTHRC1-positive CFAs and macrophages exhibited close communication.

The expression levels of CTHRC1 and invasive genes were increased in TNBC patients ( Figure 8B ). Immunofluorescence analysis of tumour and adjacent non-tumour tissues revealed a significant increase in M2 macrophage infiltration and elevated CTHRC1 expression levels ( Figure 8A ). These results validate our intriguing findings based on bioinformatics ( Figure 8C ). CTHRC1-positive CAFs facilitated the infiltration of M2-polarized macrophages.