Tumor extracellular matrix enhances invasive gene expression of breast cancer cells in 3D patient-derived scaffolds

Abstract

Extracellular matrix (ECM) remodeling in cancer provides an essential substructure for tumor progression. We utilized patient-derived scaffolds (PDS) to model tumor ECM changes and study their impact on cell behavior. PDS were obtained from breast tumor and normal healthy breast tissue by decellularizing surgically resected specimens. We used PDS to design a 3D culture of the breast cancer cell line MCF-7. We utilized bioinformatics pipeline to identify hub genes indicative of cell invasiveness, and assessed their expression using quantitative real-time PCR. Our decellularization protocol led to decellularization of tissues while preserving key ECM components. ECM components such as collagen, glycosaminoglycans, collagen IV, and vimentin were significantly overexpressed in tumor compared to normal PDS. In 3D cultures, cells cultured on normal PDS had significantly lower viability and proliferation. Moreover, cells cultured for 15 days on tumor PDS showed significant overexpression of hub genes, CAV1, CXCR4, CNN3, MYB, and TGFB1, and secreted higher levels of IL-6 (122.91 vs. 30.23 pg/10⁶ cells, P < 0.05), all markers of an aggressive breast cancer phenotype. Breast cancer cells fail to acquire aggressive features when cultured in an ECM lacking tumor-specific alterations. This underscores the potential of therapeutic approaches targeting the mechanobiological properties of the tumor ECM.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-12196-z.

Introduction

Breast cancer is the second leading cause of cancer mortality in women with an estimated number of 42,250 deaths and 310,720 new cases in 2024 in the United States¹. Although breakthroughs in cancer awareness, screening protocols, and novel therapeutic approaches have led to more than a 40% reduction in breast cancer mortality since its peak mortality rate in the year 1989¹, there are still so many unknown underlying mechanisms and undiscovered therapeutic targets remaining.

One of the emerging targets for anti-cancer therapies has been the cancer extra-cellular matrix (ECM)². Historically, cancer was viewed as stemming from tumorigenic mutations within the genome of cancer cells. However, during the last couple of decades, the significant role of other components in the tumor microenvironment (TME) such as the ECM has been revealed. ECM is a network of extra-cellular proteins and glycoproteins that form the structure and cohesivity of the tissue. It can also transfer mechanical and biochemical signals to the cells surrounded by it through a mechanism called mechanotransduction. During the process of tumorigenesis, the tumor ECM undergoes several changes. There is an interplay between the ECM and cancer cells; cancer cells and cancer-associated fibroblasts (CAFs) are responsible for ECM remodeling and dysregulation, and ECM can affect cancer cells by providing a suitable niche for their proliferation, progression, drug resistance, and even metastasis^3,4.

Utilizing scaffold-based 3D cancer cell cultures provides a platform to study the impact of mechanical and biochemical cues of ECM on cancer cells⁵. Earlier studies have used synthetic polymers or biostatic scaffolds to assess a limited number of selected components and features of ECM, such as stiffness or porosity, on cancer cells⁶. Whereas, using a scaffold that best mimics and encompasses all aspects of the tumor ECM in the patient’s body could provide a more realistic simulation of tumor ECM alterations and their interaction with cancer cells^7,8.

In this study, we employed an SDS-based decellularization protocol to decellularize the surgically resected breast tumor and normal breast tissue samples. We used these decellularized tissue samples or patient-derived scaffolds (PDS), first to characterize and compare the ECM components and structure in the tumor and normal ECM, and then to use them as cell culture platforms. Our study results demonstrated that MCF-7 breast cancer cells were unable to fully proliferate and failed to demonstrate an aggressive phenotype when cultured on normal PDS compared to when grown on tumor PDS. Notably, tumor PDS upregulated the expression of hub genes associated with cell invasiveness, migration, and metastasis, whereas, normal PDS hindered their expression in MCF-7 cells. These results underscored the role of tumor ECM alterations compared to the normal ECM in enhancing cancer progression and shed light on the promising future of a novel group of anti-cancer therapies targeting cancer ECM to reverse or eliminate its tumor-specific mechanobiological characteristics.

Results

The decellularization protocol led to complete decellularization of the tissue samples while preserving the key ECM contents

Figure 1 demonstrates a summary of the project design and conducted experiments (Fig. 1). The decellularization protocol resulted in complete cell removal of the samples compared to the native tissue. In gross evaluation of the samples, the tissue became opaque and less dense compared to the native samples before decellularization (Fig. 2a). H&E staining of the samples revealed the elimination of cell nuclei in both normal and tumor tissues while maintaining the ECM filaments after decellularization (Fig. 2b). The GAG content of both normal (1.62 vs. 1.90 µg/mg, 95% CI: −0.70-0.15, P-value: 0.12) and tumor (3.07 vs. 2.99 µg/mg, 95% CI: −0.34-0.49, P-value: 0.70) samples was preserved after decellularization and did not significantly differ between native and decellularized tissue (Fig. 2c). The decellularization protocol led to the maintenance of the collagen in the decellularized scaffold in both groups of the normal (186.94 vs. 226.71 µg/mg, 95% CI: −97.31-17.76, P-value: 0.10) and tumor (507.35 vs. 469.59 µg/mg, 95% CI: −159.10-234.62, P-value: 0.49) scaffold, as well (Fig. 2d).Scanning Electron Microscopy (SEM) analysis, in various magnifications, revealed the absence of cellular components, while also demonstrating increased porosity compared to the native samples. Additionally, the microstructural features were preserved, with an intact ECM structure and no signs of collagen degradation (Fig. 2e). DAPI staining verified optimal decellularization of the scaffolds, preserving the ECM while completely removing all cells (Fig. 2f). DNA quantification revealed a significant reduction in DNA content following decellularization of the scaffolds, from 527.1 ng/µL in native scaffolds to 7.9 ng/µL in decellularized scaffolds. The decellularized tissue samples are hereafter referred to as patient-derived scaffolds or PDS.

Fig. 1.

Summary of the project design and experiments. (a) acquisition of normal and tumor patient-derived scaffolds (PDS) by decellularization of the surgically-resected breast tumor and normal breast samples from patients with breast cancer and patients undergoing breast reduction mammoplasty, respectively. (b) characterization of the differences in normal and tumor ECM by histological and immunohistochemical evaluations, scanning electron microscopy (SEM), and biochemical analyses. (c) utilization of tumor and normal PDS as a platform for scaffold-based 3D culture of MCF-7 breast cancer cells. Compared to the tumor PDS, normal PDS led to significantly lower viability and IL-6 secretion in breast cancer cells. The error bars represent the standard deviation (SD). Designed using images from bioicons.com.

Fig. 2.

Investigation into the structure and components of the extracellular matrix (ECM) in the native tissue and decellularized scaffold. (a) Gross appearance of the tissue samples before and after decellularization, demonstrating a change in color, while maintaining a similar form as its native tissue. (b) Hematoxylin and eosin (H&E) staining images of the normal and tumor tissue before and after decellularization. (c) The concentration of glycosaminoglycans (GAGs) in the normal and tumor tissue before and after decellularization (µg in mg of wet tissue). (d) The concentration of collagen in the normal and tumor tissue before and after decellularization (µg in mg of wet tissue). (e) Scanning electron microscopy (SEM) images of both native and decellularized samples showing the absence of cellular components in the latter, accompanied by increased porosity and cross-linked components in the decellularized tumor compared to the decellularized normal samples. (f) DAPI images of the tissue before and after decellularization, showing complete removal of cells after decellularization.

ECM key proteins are more abundant in the tumor compared to normal PDS

Histological assessments based on trichrome, PAS, Sirius red, and alcian blue staining confirmed the preservation of ECM structural and biochemical components in the PDS. Moreover, they revealed a higher density, abundance, and cross-linking among collagen and other ECM key proteins and glycoproteins in the tumor PDS (Fig. 3a-f). This was consistent with the results of the tensile test that revealed the tumor PDS had a significantly higher stiffness, measured by the Young’s modulus, compared to the normal PDS (Fig. 3g-i). Evaluation of IHC images revealed that collagen IV, and vimentin were significantly overexpressed in the tumor compared to the normal PDS (Supplementary Table 3).

Fig. 3.

Histological, Immunohistochemical, and mechanical investigations on normal and tumor patient-derived scaffolds (PDS). Masson’s trichrome (a), periodic acid-Schiff (b), Sirius red (c), and alcian blue (d) staining in normal and tumor PDS revealing higher accumulation and denser organization of collagen and glycoproteins in the tumor compared to the normal extracellular matrix (ECM). Immunohistochemical evaluations demonstrating higher expression of collagen IV (e) in tumor PDS compared to the normal one. The absence of cytokeratin 7 (CK7) (f), a protein expressed in breast epithelial cells, confirms the complete removal of cells in both the tumor and normal PDS. The results of the tensile test (g), (h) show that the tumor PDS had a significantly higher Young’s modulus, which is a measure of tissue stiffness, than normal PDS (102 MPa vs. 0.19 MPa) (i). The scale bar is 100 μm.

Tumor PDS facilitates tumor cell growth

To compare the contribution of normal and tumor PDS to the viability and proliferation of cancer cells, an MTT cell viability assay was performed on day 7 and day 15 of cell culture on the PDS. Tumor PDS resulted in higher optical density (OD) values compared to the control group, indicating increased cell metabolic activity (Fig. 4a). Moreover, the analysis of DAPI images on day 15, exhibited a mean of 59 DAPI-stained nuclei in each slide obtained from cell-cultured tumor PDS, whereas this number was only 6 for the normal PDS. These results indicated that the tumor PDS group exhibited significantly higher DAPI-stained nuclei compared to the normal PDS group (P-value: 0.016) (Fig. 4b-d).

Fig. 4.

Characterization of scaffold-based 3D cell culture of MCF-7 breast cancer cells on normal compared to tumor patient-derived scaffolds (PDS). a) the result of MTT cell viability assay demonstrating higher optical density (OD) in 3D cell cultures on tumor PDS compared to normal PDS, 7 days and 15 days after cell culture (N = 3 in each group). DAPI images of tumor (b) and normal (c) PDS after PDS-based 3D cell culture with MCF-7 cells for 15 days reveal a higher number of DAPI-stained nuclei on tumor PDS compared to the normal PDS, as quantified by automated count of DAPI-stained cell nuclei (d) in the obtained images (N = 8 in each group). e) concentration of IL-6 in the cultured media of tumor and normal PDS-based 3D cell culture of MCF-7 cells after 15 days (N = 3 in each group). *P-value < 0.05. The error bars represent the standard deviation (SD). The scale bar is 50 μm.

Tumor PDS induces IL-6 secretion by tumor cells

In our study, we also aimed to investigate the impact of scaffolds on the secretome and cytokine profile of MCF-7 breast cancer cells. Therefore, we assessed the concentration of IL-6, as a representative of cytokines that are closely correlated with tumor progression and metastasis, in the cultured media of both normal and tumor PDS-based 3D cell cultures. In 15 days after PDS-based cell culture, MCF-7 cells cultured on tumor PDS secreted 122.91 pg IL-6/10^6 cells, whereas cells on normal PDS released 30.23 pg IL-6/10^6 cells (P-value: 0.048) (Fig. 4e).

Identifying gene markers of invasiveness in MCF-7 breast cancer cell lines through bioinformatic analyses

To identify differentially-expressed genes (DEGs) associated with invasiveness and migration in breast cancer cell lines, the microarray gene expression profiles of MCF-7, as a non-metastatic, non-invasive, and non-migratory breast cancer cell line, and MDA-MB-231, HCC1937, BT549, and Hs578t cell lines as its metastatic, more invasive and migratory counterparts were compared in two independent GSE datasets (GSE111653 and GSE48213). To identify DEGs we utilized a minimum fold change in expression of more than 2 and an FDR-corrected P-value cutoff of 0.01 and found 701 DEGs in GSE111653 (105 up-regulated and 596 down-regulated DEGs) and 597 DEGs in the GSE48213 dataset (126 up-regulated and 471 down-regulated DEGs), as shown by volcano plots in Fig. 5 (Fig. 5a, b). The expression patterns of the DEGs across different samples in the datasets are also visualized in the heatmaps (Fig. 5c, d). The samples were clustered similarly using DEGs from the two different datasets.

Fig. 5.

Identification of differentially-expressed genes (DEGs) in MCF-7 cell line compared to a group of more invasive breast cancer cell lines including MDA-MB-231, HCC1937, BT549, and Hs578t in two GEO datasets. Volcano plot of DEGs with a log2 fold change > 1 and P-value < 0.01 in (a) GSE111653, and (b) GSE48213. Gene expression patterns of DEGs in cell lines clustered with the Pearson method in (c) GSE111653, and (d) GSE48213. .

We identified pairs of strongly co-regulated genes in each dataset, i.e. pairs with a correlation coefficient of more than 0.9 and an adjusted P-value of less than 0.01, to perform gene co-expression network analysis. Then based on the gene co-expression network analysis (Fig. 6a, b), we selected genes with the highest ranks in the co-expression clusters of the GSE111653 and GSE48213. The GO enrichment analysis demonstrated the biological processes overrepresented in the largest clusters of each dataset (Fig. 6c, d), with most co-expressed DEGs involved in cell migration, cell motility, and cell adhesion.

Fig. 6.

Gene co-expression network and gene ontology (GO) analyses using genes with a strongly correlated expression. Gene co-expression network derived from (a) GSE111653, and (b) GSE48213. Larger nodes represent genes with a higher rank. GO analysis results showcasing the most overrepresented biological processes in the biggest network clusters from (c) GSE111653, and (d) GSE48213. These results show the enrichment of genes involved in cell migration, motility, and adhesion.

Subsequently, seven genes with the highest ranks in the gene co-expression networks, including CAV1, CAV2, CNN3, CXCR4, MYB, TGFB1, and ZNF518B, were selected as significant markers of cell motility and migration in breast cancer cell lines. We subsequently sought to assess the expression of these gene markers in MCF-7 cells cultured on tumor or normal PDS utilizing qRT-PCR.

Tumor PDS upregulates the expression of genes characteristic of cancer cell invasiveness

We assessed the expression of selected genes, as hub genes indicative of cell invasiveness and migration, in MCF-7 cells cultured on tumor or normal PDS for 15 days. The results demonstrated that 3D culture on tumor PDS led to the significant upregulation of CAV1 and CAV2, multifaceted genes with reported roles in tumor’s aggression and resistance to therapies, CNN3, CXCR4, a part of CXCL12 axis as a driver of breast cancer invasiveness and metastasis, MYB, again as a promoter of invasiveness and metastasis in breast cancer, and TGFB1, compared to the MCF-7 cells culture on normal PDS (Fig. 7). This highlights the role of the absence of cancer-specific ECM alterations and remodeling in the normal ECM to downregulate the invasive and migratory gene expression profile in the MCF-7 cells cultured on normal PDS, compared to the cells cultured on tumor PDS.

Fig. 7.

Relative expression of genes in MCF-7 cells cultured on tumor compared to normal patient-derived scaffolds (PDS). The bar plot depicts the relative expression (log₂ scale) of genes CAV1, CAV2, CNN3, CXCR4, MYB, TGFB1, and ZNF518B in MCF-7 breast cancer cells cultured on tumor PDS compared to those cultured on normal PDS. Gene expression levels were determined using qRT-PCR, normalized to a reference gene, and analyzed using the ΔΔCt method. The p-values were calculated using an independent sample t-test or Mann-Whitney U test based on the normality test results. *P-value < 0.05, **P-value < 0.01. The error bars represent the standard error (SE).

Tumor PDS promotes cell infiltration and migration

To functionally confirm the impact of tumor PDS on invasiveness and migratory behavior of cancer cells, we assessed the infiltration of the cells within the scaffold across different time points (Fig. 8a). On day 2, cells on normal PDS were predominantly localized to the top regions, with minimal infiltration into the central or bottom regions, while in tumor PDS infiltration to the center of the PDS could already be observed. By day 7, cells on the tumor PDS demonstrated markedly deeper migration, with nuclei frequently observed not only at the top, but also in the central and bottom regions of the scaffold. In contrast, cells on normal PDS remained mostly restricted to the upper layers, with limited penetration into the bottom region of the scaffold (Fig. 8b). These observations suggest that cancerous scaffolds not only influence gene expression profiles but also promote greater migratory capacity and scaffold infiltration by MCF-7 cells over time.

Fig. 8.

Migration patterns of the MCF-7 cells cultured on normal and tumor PDS across the scaffolds. Fluorescent microscopy on DAPI-stained samples derived from the top (t), where the cells were initially cultured, center (c), and bottom (b) of the PDS in day 2 and 7 after cell culture (a) is demonstrated. Quantitative analysis of the DAPI-stained nuclei (b) in samples (N = 3 in each group and time point) revealed a higher number of cells in the center and bottom regions of the tumor PDS compared to normal PDS in both day 2 and day 7 suggesting not only a higher proliferation, but also higher migration rates of MCF-7 cells cultured on tumor PDS. The error bars represent the standard deviation (SD). The scale bar is 400 μm.

Discussion

During the process of tumorigenesis, the tumor ECM obtains various alterations in biochemical and structural features compared to healthy tissue. At the same time, this altered ECM can contribute to the behavior of cancer cells and promote their proliferation and aggressiveness. In the present study, we investigated the changes in the ECM of breast tumors compared to normal breast tissue by decellularization of samples from both tissues and obtaining a decellularized bioscaffold, i.e. PDS, for cell culture studies. To summarize, tumor PDS contained a significantly higher amount of total collagen, collagen IV, GAGs, and vimentin. The analysis of SEM images revealed higher porosity, density, and cross-links in the tumor compared to normal PDS. We used PDS as a 3D-culture platform to model the role of tumor ECM in breast cancer and compare the impact of tumor and normal PDS on the behavior of MCF-7 breast cancer cells. Our results demonstrated higher viability and proliferation of MCF-7 cells, as inferred by MTT assay and supported by DAPI staining quantification, and also higher secretion of IL-6 while cultured on tumor PDS. Compared to normal PDS, tumor PDS induced the expression of hub genes involved in breast cancer invasiveness, migration, and metastasis. The hindered proliferation of tumor cells on a normal PDS, which lacks the features of tumor ECM, sheds light on the importance of a new era of anti-cancer therapies that target reversing the changes in the ECM of tumors.

The ECM is the interconnected and porous non-cellular component of the tissue, which provides a 3D habitat for cell growth and localization. Beyond its role as a physical scaffold for cells, the ECM plays a vital role by providing essential chemical, physical, and mechanical cues necessary for tissue morphogenesis, differentiation, and homeostasis^9,10. These signals are conveyed to the cells through a process called mechanotransduction. Integrins, acting as cell surface receptors, bridge the gap between the ECM and the actin cytoskeleton, enabling the transduction of mechanical forces into intracellular signals that regulate vital cell functions mentioned above^11–13. Vimentin is another protein found in ECM, engendering mechanotransduction and mechanoprotection, controlling collagen deposition in the ECM, and more importantly cell migration in cancer cells^14–16. Our study results showed a higher abundance of vimentin in tumor PDS, compared to the normal PDS, which could have contributed to higher colocalization and proliferation of cancer cells on tumor PDS.

The fundamentals of tumorigenesis have been long revealed to be genetic alterations disrupting the cell growth and division cycle. However, there has been rising evidence about the key role of other “partakers” in the TME, including ECM, during this process. In the last couple of decades, studies sought to investigate the impact of different mechanical properties of ECM on cancer cell behavior, and vice versa. One of the first observations was that tumor stroma stiffness can be partly responsible for the disruption in adherens junctions, focal adhesions, tissue polarity perturbations, and lack of lumen formation in tumor tissue, leading to metastasis through various pathways of mechanotransduction^17–20. This was consistent with the results of the mechanical testing in our study, where the tumor PDS had a substantially higher tissue stiffness compared to the normal PDS. The role of tumor tissue mechanical matrix in metastasis has been an area of great interest. For instance, metastatic cancer cells have been shown to have more traction stress forces. Collagen density and stiffness of the cell culture matrix can contribute to this traction force^21,22. While research on ECM in cancer has a history spanning several decades, most studies have primarily relied on 2D cell cultures or 3D cultures using biostatic scaffolds, such as collagen scaffolds. These approaches do not fully capture the intricate features and dynamic changes present in the cancer ECM. However, our study leverages biomimetic scaffolds, which closely mimic the tumor ECM environment, allowing a better understanding of its impact on cultured cells. Previous studies have utilized 3D co-cultures of cancer cells and other cellular components of the TME, such as dendritic cells²³, or peripheral blood immune cells^24–26. These co-cultures allow for a more comprehensive take on the interactions of the ECM and the cellular components within the TME, and provide a platform to study and predict the tumor response to a variety of anti-tumor therapies, particularly immunotherapies^27,28. Further studies could incorporate PDS into such coculture systems to better capture the native TME.

Some former studies have been able to shed light on the tumor ECM characteristics by bioengineering the actual tumor samples, as well. Lv et al. used decellularized tumor xenografts, created from MDA-MB-231 cell line injection to mouse models, to obtain decellularized tumor scaffolds with different stiffness and study the impact of the scaffold stiffness on cancer cell behavior²⁹. Another study using decellularized tumor xenograft samples demonstrated the role of collagen VI in the tumor ECM as a promoter of cell invasion in breast cancer³⁰. Compared to our study methodology, i.e. using patient-derived tissue samples, utilizing xenografts can omit the patient variability. However, the tumor xenograft ECM might differ from the actual tumor ECM in the human body. Using samples from patients can also provide the opportunity to link the ECM characteristics to the clinical outcomes and use these PDS-based 3D cultures as a personalized platform for tumor characterization and drug screening studies³¹. Moreover, cell culture and animal conditions might affect the ECM characteristics. Nevertheless, the latter could be an advantage in studies that opt to investigate the impact of a variety of desirable features in tumor-derived bioscaffolds^29,30.

Besides the higher proliferation of MCF-7 cells on tumor PDS, normal PDS led to the downregulation of genes involved in epithelial-to-mesenchymal transition (EMT), invasiveness, cell migration, and metastasis. We utilized a meticulous bioinformatics pipeline to identify gene markers that can best reflect invasiveness in MCF-7 cells by finding the differentially-expressed hub genes in these cells compared to their metastatic, more invasive counterparts, i.e., MDA-MB-231, HCC1937, BT549, and Hs578t³². Based on the bioinformatic analyses, we chose seven hub genes to evaluate their expression in cells cultured on tumor versus normal PDS using qRT-PCR. These gene markers were CAV1 and CAV2, CNN3, CXCR4, a part of the CXCL12 axis as a driver of breast cancer invasiveness and metastasis³³, MYB, again as a promoter of invasiveness and metastasis in breast cancer³⁴, and TGFB1. Moreover, the expression of CAV1, which can induce the expression of EMT-involved genes and alter the endocytic trafficking of integrins to promote cell migration³⁵ and lead to chemotherapy resistance by diminishing oxidative stress³⁶, was also downregulated in cells cultured on normal PDS. As mentioned before, in our study, the tumor PDS had a substantially higher stiffness than normal PDS. Aligned with previous studies, the mechanical features of the tumor ECM, particularly tissue stiffness, can be a key contributing factor to the observed differences in gene expression³⁷. For instance, the expression and activity of TGF-1 have been shown to be significantly regulated by tissue stiffness in hepatocellular carcinoma (HCC)³⁸. Interestingly, CXCR4 was introduced as a key mediator in mechanotransduction in HCC, promoting cancer cell proliferation as the matrix stiffness increases³⁹. Matrix stiffness can promote EMT through c-MYB and its downstream protein, discoidin domain receptor 2 (DDR2), in lung cancer cells, as well⁴⁰.

Normal PDS altered the secreted cytokine profile of the cells by hindering the release of IL-6 in the cell culture media. IL-6 plays a pivotal role in cancer progression and aggression, exhibiting pro-tumorigenic characteristics^41,42. It has even been revealed to be correlated with providing an immunosuppressive and pre-metastatic niche for cancer cells⁴³. This implies that tumor PDS was able to induce a more aggressive phenotype in MCF-7 cancer cells. Whereas the normal PDS, which lacks the specific features of tumor ECM, failed to engender such a response in cancer cells. This highlights the importance of utilizing a 3D scaffold encapturing features of tumor ECM in cell culture cancer studies to perfectly simulate the impact of TME on cell phenotype. Future studies incorporating a broader panel of cytokines would provide a more comprehensive understanding of how the tumor ECM modulates cancer cell secretory behavior, which was out of the scope of the current study.

Our study demonstrated that MCF-7 cancer cells were unable to proliferate on a bioscaffold obtained from normal breast tissue. Hence, therapeutic approaches opting for reversing the changes in the tumor ECM and providing an ECM similar to normal tissue could restrain the proliferation and activity of tumor cells. Our improved understanding of ECM changes in cancer has led to targeting mechanobiological characteristics of tumors. These approaches, known as mechanobiological-based therapies target ECM by hindering the production of key ECM proteins, i.e. collagen, disrupting the ECM cross-linking, or cell-matrix mechanotransducers, i.e. integrin or vimentin⁴⁴. As mentioned earlier, vimentin is another key transducer in the ECM, which was found to be higher in the tumor than in normal PDS. Preliminary studies on breast and rectal cancer cell lines suggested the inhibition of vimentin expression through microRNAs could result in lower cell migration and metastasis^45,46.

Due to the significantly higher abundance of collagen in tumor ECM compared to the normal tissue, some other studies are focused on producing “collagen-binding drugs” by fusion of a collagen-binding domain or protein to active anti-cancer agents, specifically immunotherapy drugs, as means of targeted therapies⁴⁷. Although most of these collagen-binding drugs have been assessed in a preclinical stage, results from their studies suggest higher accumulation and retention of the agents in the tumor site along with decreased systemic toxicity. For example, Ishihara et al. used the collagen-binding von Willebrand factor A3 domain to develop collagen-binding immune checkpoint inhibitors and IL-2 in murine cancer models⁴⁸. Their results revealed higher anti-cancer efficiency and safety when administrating collagen-binding drugs compared to their unmodified counterparts.

In this study, we specifically obtained tissue from patients who were eligible for surgical resection and had no metastatic disease. Additionally, the patients had not undergone neoadjuvant chemotherapy or radiotherapy to preserve the integrity of the tumor ECM and avoid treatment-related alterations that could confound the interpretation of the results. Therefore, our patient cohort included patients with early-stage disease (stage I or II) and different molecular subtypes. While we recognize that ECM properties may vary by molecular subtype and stage, our sample size was limited and did not allow for full stratification in this regard. Future studies with larger, clinically homogeneous cohorts will be valuable to investigate the subtype-specific or stage-specific contributions of ECM to tumor behavior. Moreover, comparing PDS obtained from patients with different cancer stages and different survival or clinical outcomes could be highly insightful into the ECM modifications through the process of tumor progression. Additionally, we only studied the impact of decellularized ECM on cancer cell behavior, whereas incorporating other components of the TME, such as immune cells and stromal components, into the 3D culture model could better characterize their role in shaping the cancer secretome, including other cytokines than IL-6, and promoting tumor progression. 3D cultures of primary tumor cells can also better represent the in vivo interactions of these cells with their TME. The other limitation is that we only utilized transcriptomic analyses to identify our hub genes regulating cell invasion. Integrating multiomics analyses, including proteomics, could better and more robustly shed light on the key regulators of cell invasion by incorporating post-translational modifications and protein-level data.

Conclusions

Our study characterized the changes in tumor ECM compared to normal ECM in breast cancer. It also demonstrated that cancer cells are unable to fully proliferate and obtain an aggressive gene expression profile and phenotype while cultured on a normal PDS. This underscores the importance of emerging anti-cancer therapies targeting tumor ECM components, such as collagen and vimentin, as cells are more likely to lose their pro-tumorigenic characteristics when confined in an ECM lacking cancer-specific changes. Although we could underscore the ECM-induced changes in the proliferation and phenotype of cancer cells at a deep level of gene expression, our study had limitations that could be addressed in future studies. Studies on samples from larger, clinically-stratified cohorts with a broader range of tumor stages and subtypes could better elucidate how ECM characteristics vary with disease progression and clinical outcomes. In further studies, the investigation of meticulous mechanisms by which these mechanical cues are transferred to cancer cells could reveal numerous novel targets to hinder and reverse the tumor ECM impact on the behavior of cancer cells.

Methods

Patient samples

We obtained breast tumor samples from 8 patients with pathologically-confirmed invasive ductal carcinoma who were candidates for surgery, didn’t have metastatic disease, and did not receive any chemotherapy, radiotherapy, or any other type of neoadjuvant therapy before surgery. Patients had a median age of 47 years (range: 40–64). The tumors represented diverse molecular subtypes, including ER+/PR+ (n = 4), ER+/HER2+ (n = 2), and triple-negative (n = 1), with 5 out of 8 samples being Stage II and 3 being Stage I. Lymphovascular invasion was present in 5 cases. (Supplementary Table 1) Patients underwent surgical resection of the breast tumor from August to December 2023. We also collected normal breast tissue samples from 5 patients undergoing the Reduction Mammoplasty procedure. All methods were carried out in accordance with relevant guidelines and regulations, and in compliance with the Declaration of Helsinki. Written informed consent was obtained from all patients and the study design was approved by the ethical review board of Tehran University of Medical Sciences (IR.TUMS.VCR.REC.1396.3529). Both tumor and normal samples were first thoroughly rinsed in a phosphate-buffered saline (PBS) buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4; pH 7.4; Sigma Aldrich, St. Louis, MI, USA) for 18 h to completely remove blood. Each tissue sample was dissected into pieces with an approximate size of 5 mm x 10 mm x 20 mm. The dissected tissue samples were then stored at a temperature of – 70 °C until use.

Decellularization and patient-derived scaffold preparation

To decellularize the tumor and normal breast tissue samples, they were individually transferred to a falcon and incubated at room temperature on a shaking set at 400 rpm using the following protocol: every day, the samples were first washed for two periods, each lasting for 3 h in a decellularization buffer, consisting of deionized water (DI H₂O) containing 1% sodium dodecyl sulfate (SDS) (Sigma-Aldrich), followed by another two 3-hour periods of wash in another decellularization buffer of 0.5% SDS. The samples were rinsed with PBS buffer for 30 min in between each 3-hour decellularization period to remove any remaining cell debris. The solutions were refreshed after the completion of each period. At the end of these four periods, the samples were then rinsed in a 0.1% solution of SDS overnight for the next 10 h. This protocol was repeated for 5 days to ensure complete cell removal.

At the end of the 5th day of the protocol, the samples were immersed in the PBS buffer for another 6 h and rinsed with deionized water every 30 min for any residual material to be removed. The complete decellularization of the samples was determined using histological evaluations, 4′, 6-diamidino-2- phenylindole (DAPI) staining, and DNA quantification.

Histology and immunohistochemistry

For histological evaluations, a section from each decellularized tumor and normal PDS and their native tissue samples were fixed in neutral buffered formalin 10% solution for 24 h. After rinsing with deionized water, samples were dehydrated using graded alcohol and were then embedded in paraffin. Obtained paraffin blocks from each tissue sample were sliced into sections with a thickness of 5 μm before staining. Specimens were first stained with hematoxylin and eosin (H&E) to evaluate the complete decellularization and the preservation of the scaffolds’ integrity in the PDS compared to the native tissues.

To compare the structure and the expression of key biochemical components in the ECM of the tumor versus normal breast tissue, the specimens obtained from the tumor and normal PDS underwent further staining. Masson’s trichrome and Sirius Red staining were used to evaluate the ECM architectural integrity and collagen content. Periodic acid-Schiff (PAS) and the Alcian blue staining to detect ECM polysaccharides were also performed.

To detect the expression level of specific ECM or intracellular proteins, immunohistochemical staining for vimentin, type IV collagen, and cytokeratin 7 (CK7) was performed on sections from tumor and normal PDS. In summary, after embedding the samples in paraffin, we obtained thin sections of the tissue with a thickness of 5 μm using a microtome. To unmask antigens, we employed heat-induced epitope retrieval (HIER). To minimize the occurrence of nonspecific antibody binding, we treated sections with a protein-based blocking solution comprising bovine serum albumin (BSA). Next, we incubated the tissue with a primary antibody specific to the target antigen. For visualization, we used chromogenic detection with DAB. Hematoxylin served as a counterstain to highlight cell nuclei. We included positive and negative controls to validate staining specificity, as well. The expression of IHC target proteins was quantified using ImageJ software (NIHImage, USA) as a percentage of the stained area on five random images taken from each slide.

After the recellularization of the PDS as explained in the following subsection, these specimens were then placed on specially designed slides for fixed tissue. The next step involved embedding the tissue in an optimal cutting temperature (OCT) compound. The OCT-embedded tissue was rapidly frozen to a temperature of approximately − 20 °C. Once frozen, the specimens underwent cryosectioning and were stained with H&E.

Biochemical analyses

We used biochemical analyses to measure the collagen and glycosaminoglycans (GAG) content of the native and decellularized scaffolds to ensure that our decellularization protocol preserved the key ECM contents. We also opted to compare the collagen and GAG content in the two groups of tumor and normal PDS.

To quantify collagen content in our samples, we employed the hydroxyproline assay. Hydroxyproline, an amino acid found predominantly in collagen, serves as a reliable marker for collagen quantification. The assay is based on the principle that hydroxyproline residues are released upon acid hydrolysis of collagen. Briefly, our samples were homogenized, hydrolyzed in 6 M HCl, and then subjected to colorimetric determination using chloramine-T and p-dimethylaminobenzaldehyde (Ehrlich’s reagent). Samples with a determined weight of 30–40 mg were homogenized in 100 µL deionized water and hydrolyzed in 6 N HCl at 120 °C for 4 h. Subsequently, the samples were incubated at 90 °C until they became completely dry and all the solute evaporated. The released hydroxyproline was oxidized by chloramine-T (1.4% w/v in acetate-citrate buffer, pH 6.0) for 15 min at room temperature and then incubated in Ehrlich’s reagent at 60 °C for 45 min. The color density was quantified at 550 nm using a spectrophotometer. A standard curve generated from known hydroxyproline concentrations was utilized to calculate the collagen content in each sample as µg collagen in mg of wet tissue (µg/mg).

To quantify GAG content in our samples, we employed a GAG assay kit based on the dimethyl methylene blue (DMMB) method. The DMMB dye specifically binds to GAGs, resulting in a color change that can be measured. 20 mg of each sample was homogenized and treated with papain enzyme overnight at 65 °C to remove proteins. GAGs from the samples were extracted by protein precipitation using a chaotropic agent (guanidine hydrochloride). The remaining solution after protein precipitation was also used for DNA quantification as described by the manufacturer’s manual. The extracted GAG solution was then mixed with the DMMB dye. The absorbance of the GAG-DMMB complex was measured at 546 nm wavelength. Using a standard curve from known GAG concentrations, we calculated the GAG content in samples as µg GAG in mg of wet tissue (µg/mg).

DNA quantification

As described above, a part of the samples incubated in papain enzyme was used for DNA quantification after undergoing protein precipitation. The extracted DNA solutions from the tumor and normal native tissue and their corresponding PDS were then assessed for DNA content using a Nanodrop spectrophotometer at a wavelength of 260 nm. A near-zero measurement of DNA concentration in extracts from PDS ensured complete decellularization of the scaffolds.

Tensile test

To assess the mechanical properties of the decellularized PDS, tensile tests were performed on both tumor-derived and normal PDS. Rectangular scaffold specimens in the approximate dimensions of 10 mm by 5 mm by 1 mm were prepared following decellularization. Mechanical testing was conducted using a universal testing machine (STM-400, SANTAM). Samples were clamped with minimal preload and stretched at a constant extension rate of 5 mm/min until break. Force and extension data were recorded continuously throughout the test. Stress was calculated by normalizing the measured force to the original cross-sectional area of the scaffold. Strain was computed as the extension divided by the original gauge length. Young’s modulus was determined as the slope of the linear region of the stress–strain curve using linear regression.

Cell culture

Breast cancer cell line MCF-7 (ATCC) was cultured in Dulbecco’s modified Eagle’s medium (DMEM) complete media (Gibco), supplemented with 10% fetal bovine serum (FBS), 1% penicillin/streptomycin (all ThermoFisher Scientific). Cells were grown in an incubator with a humidified atmosphere with 5% CO₂ and 20% O₂ at 37 °C. The medium was changed every 1–2 days, and the cells were passaged when they reached about 80–90% confluency. To passage cells, the culture medium was aspirated, and the cell monolayer was rinsed with PBS. Cells were detached using 0.25% trypsin-EDTA solution. The cell suspension was centrifuged at 200 × g for 5 min, and the resulting pellet was resuspended in fresh medium before reseeding in new flasks. All experiments of cell cultures and recellularization were repeated three times (experimental replicates).

Patient-derived scaffolds recellularization

Both normal and tumor PDS slices, with a maximum thickness of 5 mm, were placed in 96-well plates. The slices were then coated with 150 µL DMEM with 10% FBS and put in the incubator for 20 h. Afterward, 5 × 10^3 MCF-7 cells were added dropwise on the surface of each PDS in 150 µL of the cell line growth medium mentioned above. The PDS were transferred to a new well 48 h after recellularization. The wells were visually checked for cell growth every other day and the PDS were again transferred to a new well if the cells were seen on the plastic surface of the well. The medium in the wells was changed every 5–6 days.

Cell viability assay

To investigate and compare cell viability and proliferation on the tumor and normal PDS, we utilized the MTT assay on days 7 and 15 after recellularization. The same recellularization protocol was employed as described above. A number of 5 × 103 MCF-7 cells cultured on tumor and normal PDS were uniformly incubated in 96-well plates. Both tumor and normal PDS groups consisted of 10 replicate wells. MTT was added to 5 replicates of each group on day 7, and the rest of the replicates on day 15, at a final concentration of 5 mg/ml. The cells were subsequently incubated at 37 °C for 4 h. Afterward, 100 µl of Dimethyl sulfoxide (DMSO) was added to each well following removing the medium. After shaking the plates for 30 min at 400 rpm, the absorbance of each well was measured at 570 nm with a spectrophotometer.

IL-6 secretion assessment

After 15 days of PDS-based 3D cell culture of the MCF-7 cells, the culture media were collected and underwent two rounds of centrifuge to remove cellular debris. The resulting supernatant was used for IL-6 assessment. We utilized the ELISA kit for the measurement of human IL-6 (CN: KPG-HIL6-48, Karmania Pars Gene). Briefly, we added 50 µL of either sample supernatants or standards in duplicate to a 96-well antibody-coated plate. The plate was incubated and washed. Afterwards, 50 µL of conjugated detection antibody was added to each well, and then treated with Streptavidin-HRP. After adding the substrate, the reaction was stopped, and absorbance was measured at the 450 nm wavelength. The concentration of IL-6 in the MCF-7 culture media was determined based on the standard curve generated using human IL-6 standard solutions in picograms per milliliter (pg/mL) and then converted to pg per 10^6 cells based on the estimated number of cells.

DAPI staining

We evaluated the PDS after the completion of the decellularization process to ensure complete elimination of the DNA material. We also performed DAPI staining on PDS after recellularization to visualize the seeding of cells on the scaffold. For DAPI staining of the recellularized PDS, we used the slides obtained from the cryosectioning of these samples which were stored at – 70 °C. For visualization of dsDNA in both decellularized and recellularized scaffolds, we used 0.5 mg/mL blue-fluorescent DAPI (Sigma, St Louis, MO, USA) diluted with PBS. 300 µL of the final solution was placed directly on each slide and incubated in a dark room for 30 min. Afterward, the slides were rinsed in dH2O. We used the Leica TCS SPE confocal microscope (Leica Microsystems, Germany) to obtain fluorescent images.

Cell migration assessment within scaffolds

To evaluate the migration behavior of MCF-7 cells within the PDS, cells were cultured on top of decellularized scaffolds derived from cancerous and normal breast tissues for 2 and 7 days, using the recellularization protocol described above. The PDS were prepared with approximate dimensions of 6 mm by 6 mm by 1 mm. At each time point, scaffolds were cryosectioned along the vertical axis into top, central, and bottom regions within approximately 300–400 μm intervals, and stained with DAPI to visualize cell nuclei. Images were acquired using fluorescence microscopy with an objective lens of 10x. Quantitative analysis using ImageJ software (NIHImage, USA) was performed to count the number of DAPI-stained nuclei in each region and assess the depth and extent of cell migration across scaffold regions.

Scanning electron microscopy analysis

SEM images were taken from both native and decellularized PDS to evaluate the efficiency of the decellularization process to preserve the ECM integrity while vanishing the cellular components completely. To prepare the specimens for SEM imaging, they were incubated in a 4% glutaraldehyde solution for 1 h and then were dehydrated in graded ethanol. The samples were eventually coated with a 3.5-nm-thick chromium layer using a Gatan ion beam coater for electrical conductivity. Field emission SEM (FE-SEM; JSM-6340 F, JEOL, Tokyo, Japan) was administered to obtain images at a working distance of 8 mm and an acceleration voltage of 10 kV. The images were evaluated by two independent expert pathologists to provide interpretations.

Bioinformatics analyses

To identify genes related to migration, metastatic potential, and invasion of our breast cancer cell line, we searched the Gene Expression Omnibus (GEO) for mRNA expression datasets of different breast cancer cell lines with different behavioral features, including MCF-7 as a non-metastatic, non-invasive cell line, and MDA-MB-231, HCC1937, BT549, and Hs578t as a group of metastatic breast cancer cell lines with more migratory and invasive behavior. Based on the homogeneity and consistency of the results, two datasets (GSE111653, GSE48213) with expression profiles of these cell lines assessed by high throughput sequencing were selected. The online GEO2R tool was used to identify DEGS between two cell line groups, with a minimum |log 2-fold change (FC)|>1 and Benjamini & Hochberg (False discovery rate (FDR)) corrected P-value < 0.01. We used the R “psych” package to calculate the Pearson correlations between every pair of DEGs and obtain a correlation matrix⁴⁹. Pearson correlation was utilized to visualize the expression patterns of the DEGs in samples as heatmaps. We filtered for DEGs with |r| > 0.9 and adjusted P-values < 0.01 to be used for building co-expression networks. We then calculated network statistics, i.e. betweenness and degree, utilizing the R package “igraph”⁵⁰, and visualized the co-expression network and its corresponding statistics using Cytoscape⁵¹. “community clustering” (GLay) with default options in clusterMaker2 plugin was used to perform cluster analysis⁵². Genes with the highest rank were selected for further experiments as markers of invasiveness.

We also performed a gene ontology (GO) enrichment analysis to identify and visualize the biological function of the DEGs using the ClueGO application⁵³. ClueGO facilitates the identification of GO categories that are statistically overrepresented within the specific gene set in the network. By analyzing these categories, we could gain insights into the biological processes and functions that are most prevalent in the data, and enhance the understanding of the underlying molecular mechanisms.

RNA extraction

Total RNA was extracted from MCF-7 cells cultured on tumor and normal PDS using the TRIzol™ reagent (Thermo Fisher Scientific) following the manufacturer’s protocol by lysing the cells directly on the scaffolds. RNA was isolated through a phase separation with chloroform, followed by isopropanol precipitation. The resulting RNA pellets were washed with 75% ethanol, air-dried, and dissolved in RNase-free water. RNA concentration and purity were measured utilizing a Nanodrop spectrophotometer by measuring the absorbance ratios at 260/230 nm and 260/280 nm.

cDNA synthesis and qRT-PCR

Reverse transcription of 1 µg of total RNA was performed using the AddScript cDNA Synthesis Kit (ADDBIO) with oligo(dT) and random hexamer primers. Quantitative real-time PCR (qRT-PCR) was carried out on a LightCycler^® 96 System (Roche) using Add SYBR Master (ADDBIO). Gene-specific primers were designed to target the following genes: CAV1, CAV2, CNN3, CXCR4, MYB, TGFB1, and ZNF518B (Supplementary Table 2). We used GAPDH as the housekeeping gene for normalization. The amplification process included an initial step at 95 °C for 900 s for denaturation, followed by 40 cycles of denaturation at 95 °C for 15 s, annealing at 60 °C for 40 s, and 10 s at 72 °C for extension phase. The relative expression levels of target genes were determined using the ΔΔCt method, comparing cells cultured on tumor versus normal PDS. Fold changes were expressed as log₂-transformed values. The standard error (SE) of the relative gene expressions between tumor and normal group was also calculated using the following formula: SE =.

Statistical analysis

For Statistical analyses, we used R version 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria)⁵⁴. An independent-sample t-test or Mann-Whitney U test was used to compare the results of IHC staining, biochemical analyses, and qRT-PCR-derived gene expression depending on the results of the Shapiro-Wilk test for normality. p values lower than 0.05 were considered significant. Some of the plots were generated using GraphPad Prism (version 10.3.1; GraphPad Software, San Diego, CA). All figures were created using Inkscape version 1.3.2 (Free Software Foundation, Inc. Boston, MA, USA).

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

3D

Three-Dimensional

CAF

Cancer-Associated Fibroblast

Ct

Cycle Threshold

DAPI − 4',6

Diamidino-2-Phenylindole

DMMB

Dimethyl Methylene Blue

DMSO

Dimethyl Sulfoxide

DMEM

Dulbecco’s Modified Eagle Medium

ECM

Extracellular Matrix

EDTA

Ethylenediaminetetraacetic acid

ELISA

Enzyme-Linked Immunosorbent Assay

EMT

Epithelial-to-Mesenchymal Transition

FBS

Fetal Bovine Serum

FC

Fold Change

FDR

False Discovery Rate

GAG

Glycosaminoglycan

GO

Gene Ontology

H&E

Hematoxylin and Eosin

HIER

Heat-Induced Epitope Retrieval

IHC

Immunohistochemistry

IL

6-Interleukin 6

MCF7

Michigan Cancer Foundation-7 (Breast Cancer Cell Line)

MTT

Methylthiazolyldiphenyl-Tetrazolium Bromide

OCT

Optimal Cutting Temperature

PBS

Phosphate-Buffered Saline

PDS

Patient-Derived Scaffold

qRT

PCR-Quantitative Real-Time Polymerase Chain Reaction

SDS

Sodium Dodecyl Sulfate

SEM

Scanning Electron Microscopy

TGF

β-Transforming Growth Factor Beta

TME

Tumor Microenvironment

Author contributions

PS.P. : Conceptualization, Writing - original draft, Methodology, Visualization, Formal analysis; N.MG.: Conceptualization, Methodology, Validation, Formal analysis; A.A.: Methodology, Data curation; M.MZ.: Validation, Conceptualization, Project administration, Supervision; H.M.: Methodology, Resources; SR.M.: Resources; H.M.: Resources; A.K.: Project administration, Investigation, Funding acquisition, Writing - review & editing, Validation, Supervision.

Funding

This project was funded by the Tehran University of Medical Sciences (Grant Number: IR.TUMS.VCR.REC.1396.3529).

Data availability

The data that support the findings of this study are available on request from the corresponding author, A.K.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

All patients provided written informed consent prior to sample acquisition. The study design received approval from the ethical review board at Tehran University of Medical Sciences (IR.TUMS.VCR.REC.1396.3529).