Extracellular Matrix Sulfation in the Tumor Microenvironment Stimulates Cancer Stemness and Invasiveness

Abstract

Tumor extracellular matrices (ECM) exhibit aberrant changes in composition and mechanics compared to normal tissues. Proteoglycans (PG) are vital regulators of cellular signaling in the ECM with the ability to modulate receptor tyrosine kinase (RTK) activation via their sulfated glycosaminoglycan (sGAG) side chains. However, their role on tumor cell behavior is controversial. Here, it is demonstrated that PGs are heavily expressed in lung adenocarcinoma (LUAD) patients in correlation with invasive phenotype and poor prognosis. A bioengineered human lung tumor model that recapitulates the increase of sGAGs in tumors in an organotypic matrix with independent control of stiffness, viscoelasticity, ligand density, and porosity, is developed. This model reveals that increased sulfation stimulates extensive proliferation, epithelial-mesenchymal transition (EMT), and stemness in cancer cells. The focal adhesion kinase (FAK)-phosphatidylinositol 3-kinase (PI3K) signaling axis is identified as a mediator of sulfation-induced molecular changes in cells upon activation of a distinct set of RTKs within tumor-mimetic hydrogels. The study shows that the transcriptomic landscape of tumor cells in response to increased sulfation resembles native PG-rich patient tumors by employing integrative omics and network modeling approaches.

Keywords: cancer; extracellular matrix (ECM); hydrogels; tissue engineering; tumor microenvironment (TME); tumor models.

Figure 1

Elevated PG expression correlates with invasiveness and poor survival in LUAD patients. a) Hierarchical clustering heatmap of PG mRNA expression z‐scores in LUAD patient tumors relative to normal tissue samples where the rows represent PG genes and columns indicate patients. b) Percentages of PG+ and PG‐ tumor samples in TCGA LUAD patient cohort. A sample is considered PG+ if z‐score is greater than the cut‐off of three for at least three genes, while a sample is considered PG‐ if the z‐score is higher than the cut‐off, but the number of genes is equal to zero. c) Box plot comparison of PG scores of PG+ and PG‐ patient groups. Each box extends from the lower to the upper quartile where the horizontal line indicates the median value. Each dot represents the PG score of an individual patient. Statistical analysis was performed with Wilcoxon test, p = 1.5e⁻¹⁴. d) Survival plots of PG+ and PG‐ LUAD patients. Log‐rank test was used to assess significance, p = 0.025. e) Regression analysis on EMT scores and PG scores using the mRNA (p = 1.9e⁻¹⁰¹) and f) protein expressions (p = 5.22e⁻⁰⁷). Spearman rank correlation was used to calculate the correlation values (ρ). g) sGAG quantification in LUAD patient‐derived tumor and matched normal lung samples, n = 5 biological replicates. Data is represented as mean ± S.D and statistical significance was analyzed using a paired, two‐tailed student's t‐test, p < 0.01. h) Hematoxylin and eosin (H&E) and Alcian blue staining of normal lung parenchyma and tumor tissue (scale bar: 100 µm).

Figure 2

Mimicking the increased sulfation in the TME drives aberrant proliferation of lung tumor cells. a) Schematic illustration of the fabrication of AlgLung and S‐AlgLung hydrogels using decellularized lung ECM and alginate/alginate sulfate. b) Storage modulus and c) Loss tangent (G″/G′) of AlgLung and S‐AlgLung hydrogels, ns not significant. d) Molecular weight of alginate (Alg) and alginate sulfate (S‐Alg) characterized by SEC‐MALS, p < 0.001. e) Creep and Recovery test of AlgLung and S‐AlgLung hydrogels. f) Permanent strain of AlgLung and S‐AlgLung hydrogels obtained from creep and recovery tests, ns not significant. g) 10 kDa and h) 70 kDa dextran release from AlgLung and S‐AlgLung hydrogels over 72 h, ns not significant. i) Bright field images of A549 cells in AlgLung and S‐AlgLung hydrogels at day 1, 7, 14, and 28. j) Quantification of DNA content in AlgLung and S‐AlgLung hydrogels normalized to day 0, **p < 0.01, ***p < 0.001. k) Immunofluorescence image of A549 clusters stained for Phalloidin (magenta) and DAPI (blue) in AlgLung and S‐AlgLung hydrogels (scale bar: 100 µm). l) Quantification of cluster area (µm²) and m) invasiveness (%) of cells grown in AlgLung and S‐AlgLung hydrogels at day 28, *p < 0.05, ****p < 0.0001. n) Immunofluorescence staining of LUAD markers TTF‐1 and p63 (green) and DAPI (blue) in A549 cells grown in AlgLung and S‐AlgLung hydrogels (scale bar: 100 µm). Quantitative data is represented as mean ± S.D, and statistical significance was analyzed using an unpaired, two‐tailed student's t‐test.

Figure 3

Increased sulfation in the ECM modulates EMT and cancer stemness. a) Immunofluorescence staining for E‐cadherin, N‐cadherin, vimentin and fibronectin (green, top to bottom) and DAPI (blue) in A549 cells grown AlgLung and S‐AlgLung hydrogels (scale bar = 100 µm). b) mRNA expression of EMT regulators and markers in A549 cells grown in AlgLung and S‐AlgLung hydrogels, ns not significant, *p < 0.05, *p < 0.01. c) Schematic illustration showing the inverse correlation between epithelial phenotype and ECM sulfation. d) Immunofluorescence staining of MUC5B (green) and DAPI (blue) in A549 cells grown in AlgLung and S‐AlgLung hydrogels (scale bar: 100 µm). e) mRNA expression of MUC5AC and MUC5B in A549 cells grown in AlgLung and S‐AlgLung hydrogels, *p < 0.05. f) Immunofluorescence staining of SOX2 (top) and beta‐catenin (bottom) (green) and DAPI (blue) in A549 cells grown in AlgLung and S‐AlgLung hydrogels (scale bar: 100 µm). g) mRNA expression of stemness markers in A549 cells grown in AlgLung and S‐AlgLung hydrogels, *p < 0.05, ***p < 0.001. h) Immunofluorescence image of A549 spheroids stained for Phalloidin (magenta) and DAPI (blue) in AlgLung and S‐AlgLung hydrogels (scale bar: 100 µm). i) Quantification of cluster invasiveness (%) of spheroids grown in AlgLung and S‐AlgLung hydrogels at day 21, *p < 0.0001. j) Regression analysis of cancer stem cell (CSC) genes and PGs using the mRNA expression scores in TCGA LUAD patient cohort. Spearman rank correlation was used to calculate the correlation values (ρ) (p = 2.52e⁻²⁶). b,e,g) Relative quantification (RQ) was used with normalization to AlgLung samples. Data is represented as mean ± S.D and statistical significance was analyzed using an unpaired, two‐tailed student's t‐test.

Figure 4

RTK signaling mediates sulfation‐induced tumorigenic phenotype. a) Immunofluorescence staining of EGFR (top) and integrin β1 (bottom) (green) and DAPI (blue) in A549 cells grown in AlgLung and S‐AlgLung hydrogels (scale bar: 100 µm). b) Representative images of phalloidin (magenta) and DAPI (blue) stained A549 cells in AlgLung and S‐AlgLung hydrogels treated with EGFR and integrin β1 inhibitors (scale bar: 100 µm). c) Quantification of cluster area (µm²) of A549 cells in S‐AlgLung hydrogels treated with EGFR and integrin β1 inhibitors, ns not significant. d) Representative brightfield images of A549 cells grown in S‐AlgLung and treated with RTK inhibitors at day 14, (scale bar: 100 µm). e) Metabolic activity analysis of A549 cells in S‐AlgLung hydrogels and treated with RTK inhibitors. Values were normalized to control group, ns not significant, *p < 0.05, **p < 0.01, ***p < 0.001. f) Human Phospho‐RTK Array performed on A549 cells grown in AlgLung and S‐AlgLung hydrogels for 21 days. RTKs were framed up and footnoted. g) Representative images of phalloidin (magenta) and DAPI (blue) stained A549 cells in S‐AlgLung hydrogels treated with vofatamab (scale bar: 100 µm). h) Metabolic activity analysis of A549 cells in S‐AlgLung and vofatamab treated hydrogels using CellTiter‐Glo 3D assay, *p < 0.05. i) Representative images of phalloidin (magenta) and DAPI (blue) stained A549 cells S‐AlgLung hydrogels treated with indicated inhibitors (scale bar: 100 µm). j) Quantification of cluster area (µm²) of A549 cells in S‐AlgLung hydrogels treated with indicated inhibitors, ns not significant, **p < 0.01. k) Quantification of DNA content of A549 cells grown in S‐AlgLung hydrogels and treated with indicated inhibitors. Values were normalized to control group, ns not significant, *p < 0.05, **p < 0.01. Quantitative data is represented as mean ± S.D and statistical significance was analyzed using ordinary one‐way Anova analysis.

Figure 5

PI3K is a key regulator of sulfation‐induced tumorigenic phenotype. a) Representative brightfield and confocal microscopy images of A549 cells expressing shPIK3CA or control vectors in S‐AlgLung hydrogels. Cells were stained with phalloidin (magenta) and DAPI (blue), (scale bar: 100 µm). b) Quantification of cluster area (µm²) of shPIK3CA‐expressing A549 cells in S‐AlgLung hydrogels compared to control, ***p < 0.001. c) mRNA expression of EMT regulators in shPIK3CA‐expressing A549 cells grown in S‐AlgLung hydrogels, ns not significant, *p < 0.05, **p < 0.01, ***p < 0.001. d) mRNA expression of stemness markers in shPIK3CA‐expressing A549 cells grown in S‐AlgLung hydrogels, ns not significant, **p < 0.01, ***p < 0.001. e) Representative brightfield and confocal microscopy images of A549 cells overexpressing PIK3CA or control vectors in AlgLung hydrogels. Cells were stained with phalloidin (magenta) and DAPI (blue), (scale bar: 100 µm). f) Metabolic activity analysis of PIK3CA‐overexpressing A549 cells in AlgLung hydrogels using CellTiter‐Glo 3D assay, ****p < 0.0001. g) mRNA expression of EMT regulators in PIK3CA‐overexpressing A549 cells grown in AlgLung hydrogels, ns not significant, *p < 0.05. h) mRNA expression of stemness markers in PIK3CA overexpressing A549 cells grown in AlgLung hydrogels, *p < 0.05, **p < 0.01. i) Representative brightfield images of PIK3CA‐overexpressing A549 cells grown in S‐AlgLung hydrogels and treated with FAK inhibitor (scale bar: 100 µm). j) mRNA expression of EMT regulators in PIK3CA‐overexpressing A549 cells grown in S‐AlgLung hydrogels and treated with FAK inhibitor, ns not significant, *p < 0.05, **p < 0.01. k) Schematic illustration of the sulfated ECM‐induced signaling cascade in lung tumor cells grown in S‐AlgLung hydrogels. Sulfated ECM exerts affinity to bioactive ligands that leads to the activation of FGFR3 and RYK receptors and their downstream signaling. PI3K acts as a hub in sulfation‐induced proliferation, EMT activation and stemness phenotype in A549 cells. All quantitative data is represented as mean ± S.D and statistical significance was analyzed using an unpaired, two‐tailed student's t‐test. In qRT‐PCR data, relative quantification (RQ) was used with normalization to control group.

Figure 6

Sulfated ECM significantly alters the transcriptional program of cancer cells. a) Heatmap shows hierarchically clustered normalized expression values of differentially expressed genes across samples. Negative z‐scores are in blue color‐scale, and positive z‐scores are in purple color‐scale. b) Volcano plot shows DEGs in purple (up‐regulated genes), blue (down‐regulated genes), and gray (other genes). Thresholds to find DEGs (adj‐pvalue<0.01 and abs(log2(FC))>1) are shown as black dashed horizontal and vertical lines. c) Reconstructed signaling network orients from receptors to significant transcription factors that regulate the differentially expressed genes. In this network, RYK, FGFR3, PTK2, and PIK3CA are source nodes and significant transcription factors are target nodes. Pathlinker is used for network reconstruction. d) Bar plot shows the functionally enriched KEGG pathways of intermediate nodes. P‐values were determined using a hypergeometric test. e) Heatmap shows TPM‐normalized values of 483 DEGs, comparing TCGA/PG+ samples with S‐AlgLung and AlgLung samples. Hierarchical clustering reveals that patient samples are more closely related to S‐AlgLung samples. f) Average z‐scores of mRNA expression data of intermediate nodes in TCGA/PG+ samples. Negative z‐scores are in blue color‐scale, and positive z‐scores are in purple color‐scale. g) Schematic summary of the results from our three‐step computational pipeline (RNA‐seq data analysis, identifying significant transcription factors and reconstructing signaling network) used for integrative network modeling (label 1). ECM‐cell interactions, influenced by sulfation, initiate a cascade of signaling events. The reconstructed signaling network promotes 34 significantly active transcription factors, including MYC, leading to widespread changes in the transcriptional program of the cells. These changes result in the differential expression of 483 genes (DEGs), with 277 genes being upregulated and 206 downregulated. Importantly, the DEGs are more highly correlated with patient tumor data in sulfated ECM compared to non‐sulfated ECM (label 2). Pathway enrichment analysis on sulfated ECM‐induced network revealed pathways regulating cell cycle, glycan biosynthesis, ECM‐receptor interaction, and cytoskeleton organization (label 3).