Extracellular Matrix Remodeling in Tumors: Forces Beyond Fibroblasts

Abstract

Aberrant extracellular matrix (ECM) remodeling in tumors is characterized by altered deposition, enzymatic cross-linking, degradation, and organization of matrix molecules that in turn can contribute directly to tumor invasiveness, immune evasion, and resistance to therapy. Traditionally, cancer-associated fibroblasts (CAFs) have been considered the principal architects of ECM remodeling in tumors. However, the role of non-fibroblast cell populations in modulating ECM structure and function and how these processes intersect with immune regulation, metabolism, and metastasis is also crucial but currently underappreciated. Challenging the prevailing fibroblast-centric paradigm, this review provides an overview of how non-fibroblast cell types engage in dynamic crosstalk and collectively shape the biochemical and biomechanical landscape of tumors. Therapeutic strategies targeting these alternative stromal players could disrupt pro-tumor ECM dynamics, enhance anti-tumor immunity, and restore tissue homeostasis. Thus, a comprehensive understanding of the intricate network of non-fibroblast stromal cells within the tumor microenvironment (TME) can help elucidate opportunities for precision medicine and personalized cancer care, leading to the development of innovative therapeutic interventions targeting the TME.

1. Introduction

The TME is a complex milieu comprising cancer cells, stromal cells, immune cells, and the extracellular matrix (ECM). The ECM is a dynamic network of fibrous proteins, adhesive glycoproteins and hydrophilic macromolecules such as proteoglycans and glycosaminoglycans including collagens, elastin, fibronectin (FN), tenascin, laminin, hyaluronic acid (HA), and leucine-rich proteoglycans that sustains the architecture and function of the tumor [1]. In the context of cancer, the TME undergoes extensive ECM remodeling characterized by altered deposition, cross-linking, degradation, architectural reorganization, cellular sensing through matrix receptors, and the generation of bioactive ECM fragments (matrikines) which together coordinate tumor–stromal interactions [2–7]. This structural network of ECM molecules is now recognized as an active regulator of tumor progression and immune modulation, with its biochemical composition, physical properties, and spatial organization actively evolving in concert with malignant transformation [4]. Table 1 provides an integrated overview of the key ECM components and highlights their role in normal and tumor-associated ECM

Table 1.

Components of the ECM^*

Category	Component	Key Features	Physiological Role(s)	Pathological Role(s) in Cancer	Cellular expression
Structural ECM	Collagen	Triple-helical fibrillar proteins; highly cross-linked; major ECM scaffold	Provides tensile strength and structural integrity; supports tissue architecture	Increased deposition and PTMs leads to increased stiffness and fiber thickness regulate invasion of tumor and immune cells; promote proliferation, migration, and chemoresistance [11, 16, 17]	CAFs (S), Activated fibroblasts (S), Tumor cells (S/R)
	Fibrin(ogen)	Plasma glycoprotein cleaved into fibrin clot	Hemostasis; provisional matrix for wound healing and angiogenesis	Tumor-associated hypercoagulability; fibrin scaffolds enhance tumor cell adhesion, extravasation, and metastasis [177]	Hepatocytes (S), Fibroblasts (S), endothelial cells (S/R), keratinocytes (S)
	Fibronectin	Multidomain glycoprotein; soluble and fibrillar forms	ECM organization; regulates adhesion, migration, differentiation; growth factor reservoir	Promotes migration, tumor cell clustering, and metastasis; contributes to pre-metastatic niche formation[39, 177]	CAFs (S), Tumor cells (S/R), Endothelial cells (S/R), Macrophages (S/R)
	Hyaluronan	Large, hydrophilic glycosaminoglycan forming viscoelastic matrix	Tissue hydration, mechanical buffering, inflammation regulation	Accumulates in TME; promotes proliferation, migration, inflammation [130]	CAFs (S), tumor cells (S)
	Mucins	Highly glycosylated secreted/transmembrane proteins	Barrier protection; epithelial integrity; anti-inflammatory roles	Promote survival signaling, immune evasion, chemoresistance, metastasis [178]	Goblet cells (specialized epithelial cells) (S)
	Elastin	Cross-linked elastic polymer	Provides elasticity and recoil to tissues	Degradation fragments promote migration, angiogenesis, and stromal recruitment [177]	Fibroblasts (S)
	Laminin	Basement membrane heterotrimers	Establish epithelial polarity; regulate adhesion and signaling	Isoform-specific effects: LM-332 promotes migration; loss of LM-111 anchoring enhances tumor progression [177]	CAFs (S), endothelial cells (S)
	Osteopontin	Secreted phosphoprotein; integrin/CD44 ligand	Bone remodeling; immune modulation	Promotes tumor proliferation, invasion, metastasis; circulating biomarker [179]	Tumor cells (S), Macrophages (S), dendritic cells (S), NK cells (R), CAFs (S), endothelial cells (S)
	Tenascins	Modular ECM glycoproteins	Tissue repair and development	Support migration, proliferation, and metastatic niche formation [177]	CAFs (S), tumor cells (S)
	Periostin	Matricellular protein regulating collagen organization	Tissue remodeling and repair	Promotes metastasis, cancer stem cell niches, and therapy resistance [13]	CAFs (S)
ECM Receptors	Integrins (e.g., αβ heterodimers)	Transmembrane adhesion receptors linking ECM to cytoskeleton	Mediate adhesion, mechanotransduction, migration, and survival signaling	Dysregulated signaling enhances invasion, survival, EMT, and therapy resistance [101]	Tumor cells (S/R), CAFs (S/R), endothelial cells (S/R), immune cells including macrophages (S/R), and T cells (S/R),
	DDR1/DDR2	Collagen-activated receptor tyrosine kinases	Regulate cell adhesion, differentiation, and matrix sensing	Promote proliferation, migration, and chemoresistance upon collagen engagement [14, 165]	DDR1: tumor cells (S/R) DDR2: fibroblasts (S/R/M), CAFs (S/R/M)
	CD44	HA and osteopontin receptor	Cell adhesion, migration, immune regulation	Drives tumor cell motility, stemness, and interaction with HA-rich TME [70]	Tumor cells (S/R), CAFs (S/R), endothelial cells (S/R), T cells (S/R), macrophages (S/R), B cells (S/R)
	Endo180 (uPARAP)	Collagen-binding endocytic receptor	Mediates collagen internalization and lysosomal degradation	Supports tumor-associated collagen turnover and invasion; coordinates with MMP-dependent cleavage [60]	CAFs (S/M)
	LAIR1	Collagen-binding inhibitory immune receptor (ITIM-containing)	Maintains immune homeostasis by dampening activation upon collagen binding	Suppresses anti-tumor immune responses in collagen-rich TME; contributes to immune evasion [15]	T cells (S), NK cells (S), macrophages (S)
ECM remodeling enzymes	Lysyl hydroxylase (PLOD1–3)	Enzymes catalyzing hydroxylation of lysine residues in collagen	Enable collagen cross-linking and fibril stability	Enhance collagen cross-linking, matrix stiffness, and metastasis [107]	CAFs (S/M), tumor cells (S/M)
	Lysyl oxidases (LOX, LOXL1–4)	Copper-dependent enzymes catalyzing covalent collagen/elastin cross-links	ECM maturation and mechanical strength	Increase ECM stiffness; promote invasion, metastasis, and pre-metastatic niche formation [46]	CAFs (S/M), tumor cells (S/M), TAM (S/M)
	Prolyl hydroxylase (P4HA1–3)	Hydroxylate proline residues in collagen triple helix	Stabilize collagen structure and enable proper folding	Promote collagen deposition and ECM accumulation; associated with hypoxia-driven tumor progression [117]	CAFs (S/M), tumor cells (S/M), activated stellate cells (S/M)
ECM Degraders	MMPs (e.g., MMP1, 2, 9, 13, 14)	Zinc-dependent ECM-degrading proteases	Controlled ECM turnover during development and repair	Promote invasion, metastasis, and growth factor release; remodel TME architecture [42, 102]	CAFs (S/M), tumor cells (S/M), TAMs (S/M), TANs (S/M), TIMs (S/M), T-cells (S/M), Adipocytes (S/M), endothelial cells (S/M)
	ADAMs / ADAMTSs	Metalloproteinases with shedding and ECM remodeling roles	Regulate cell signaling and development	Dysregulated; promote invasion, EMT, and tumor progression [180]	CAFs (S/M), tumor cells (S/M)
	Cathepsins (e.g., CTSK)	Lysosomal proteases degrading ECM proteins	Tissue homeostasis; bone remodeling	Enhance ECM degradation, invasion, immune modulation, and therapy resistance [58]	TAMs (S/M), CAFs (S/M), tumor cells (S), TAN (R)
Matrikines	Fibrin(ogen) degradation products	Plasmin-generated fragments (e.g., D-dimer, fragment E)	Maintain fibrinolytic balance; support vascular remodeling	Promote angiogenesis; correlate with metastasis and poor prognosis; circulating biomarkers [181]	Produced within tumors following fibrinolysis and coagulation
	Endotrophin	the cleaved COL6α3 fragment	Tissue remodeling and adipogenesis	promotes fibrosis, angiogenesis, and inflammation [139]	CAFs (S/M)
	Fibronectin fragments (e.g., anastellin)	Proteolytic FN fragments with signaling activity	Support tissue repair and survival	Can inhibit angiogenesis and tumor growth or modulate tumor progression [182]	Produced from CAF-derived fibronectin via cleavage by proteases released by CAFs, tumor cells and immune cells

^*CAFs represent the major cell type responsible for producing these ECM components

^*S: Sender (produces/secretes the ECM component)

^*R: Receiver (senses/binds/signals through the ECM component)

^*M: Modifier (enzymatically modifies/cross-links/degrades the ECM component)

The composition and mechanical properties of ECM have emerged as important prognostic indicators across multiple cancer types. Specific ECM proteins have been associated with clinical outcomes; for instance, high expression of COL5A1 correlates with poor prognosis in cervical cancer [8], type VI collagen, especially an overmodified form, is associated with lung and other types of cancer [9–11], and type XVII collagen has been implicated in promoting invasion and metastasis in colorectal cancer (CRC) [12]. Similarly, overexpression of periostin (POSTN) is associated with unfavorable outcomes in non-small cell lung cancer (NSCLC) [13]. Beyond composition, ECM provides biomechanical cues that directly influence tumor cell states. In breast cancer models, moderate matrix stiffness was shown to activate integrin β1/β3–cytoskeletal– Autoimmune Regulator (AIRE) signaling, promoting stemness and tumorigenic potential, whereas higher stiffness triggered Discoidin domain-containing receptor 2 (DDR2)/ Signal transducer and activator of transcription 1 (STAT1)/P27 signaling, leading to cell cycle arrest and quiescence [14]. In addition, emerging evidence reveals that ECM stiffness primarily driven by the accumulation of stable type I collagen fibrils [15, 16] critically regulates immune cell functions such as activation, polarization, migration, and cytotoxicity, thereby affecting the efficacy of immunotherapies [17]. ECM deposition and stiffness can serve as a physical barrier of cancer cells in the TME and hinder immune infiltration and accessibility to tumor cells, contributing to immune exclusion. The type VI collagen-rich TME observed in prostate cancer has been reported to impair CD4⁺ T cell motility, resulting in their accumulation in the stromal compartment and potentially limiting effective anti-tumor immune responses [17].

Beyond affecting anti-tumor immune responses, fibroblast-mediated ECM remodeling has profound implications for clinical outcomes. Cancer-associated fibroblasts (CAFs) actively remodel the ECM composition and organization, thereby creating a microenvironment that promotes tumor progression [18]. In HER2-positive breast cancer, trastuzumab resistance has been associated with a population of Transforming growth factor-beta (TGF-β)-activated CAFs characterized by reduced interleukin 2 (IL2) activity. Restoration of IL2 signaling in resistant tumors enhanced the antitumor efficacy of trastuzumab, highlighting the contribution of CAFs to therapeutic response through not only ECM remodeling but also modulation of patient response to different therapies [19]. Consistent with these findings, analysis of pretreatment biopsies from ovarian, breast, and colorectal cancer patients demonstrated elevated expression of ECM-related genes in chemoresistant tumors. Moreover, fibroblast-associated lysyl oxidase (LOX)-mediated collagen covalent cross-linking promotes matrix stiffening, whereas pharmacological inhibition of LOX restores chemosensitivity in preclinical models, linking fibroblast-driven ECM remodeling to poor therapeutic response and adverse clinical outcomes [20].

However, not all ECM remodeling promotes tumor progression, and its functional consequences are highly context-dependent. Certain ECM components can act as physical barriers that restrict tumor spread or serve as reservoirs for anti-tumorigenic signals. Notably in pancreatic ductal adenocarcinoma (PDAC), experimental depletion of type I collagen or αSMA⁺ fibroblasts accelerated tumor progression, reduced survival, increased infiltration of myeloid derived suppressor cells (MDSCs) and suppressed CD8+ T cell activity [21]. Similarly, in liver metastases of colorectal and pancreatic cancers, CAFs derived type I collagen, primarily from hepatic stellate cells, acts as a mechanical constraint on tumor expansion, and its depletion promoted metastatic growth, outweighing the pro-tumorigenic effects of CAF-derived factors such as HA and hepatocyte growth factor [22]. Additional complementary evidence for HA-mediated tumor suppression is provided by studies in the naked mole rat, where extremely high molecular weight HA accumulates due to both decreased degradation by hyaluronidases and a unique sequence of hyaluronan synthase 2 (HAS2). Enhanced HA–CD44 signaling promotes ‘early contact inhibition’ in naked mole rat fibroblasts, thereby conferring resistance to malignant transformation. Disruption of high–molecular weight HA, either through HAS2 knockdown or increased HA degradation, renders naked mole-rat cells susceptible to malignant transformation, highlighting a direct functional link between ECM composition and tumor suppression [23]. It is important to note, however, that the deposition and overall quantity of type I collagen are not the only factors influencing tumor cell behavior as post-translational modifications (PTMs) have been shown to modify type I collagen quality and promote metastasis [24]. Brisson et al. demonstrated that stromal type III collagen restrains tumor growth and spread in a murine breast cancer model by regulating stromal organization and limiting myofibroblast activation [25, 26]. Type III collagen-deficient mice injected with 4T1 cells developed larger primary tumors and more lung metastases having highly aligned, myofibroblast-rich stroma compared to wild-type mice. Type III collagen modulates the lateral growth of type I collagen fibrils and limits the formation of densely aligned collagen fibers that facilitate directional tumor cell migration and invasion [24, 25]. Consistent with this, type III collagen-rich matrices inhibited cancer cell adhesion, proliferation, migration, and invasion, while also inducing apoptosis, highlighting the role of type III collagen in opposing metastasis. [22]. This tumor restraining property of type III collagen is therefore closely linked to its ability to regulate stromal architecture and attenuate pro-invasive mechanical and structural signaling cues that promote tumor dissemination. These observations highlight the multifaceted role of the ECM in cancer biology, emphasizing that indiscriminate ECM targeting may be detrimental. Instead, precise, modulatory strategies are needed to reshape the TME to benefit therapeutic interventions.

Traditionally, CAFs have been regarded as the primary drivers of ECM remodeling within the TME, given their well-characterized roles in paracrine signaling and in collagen expression, PTMs, and deposition leading to matrix stiffening [27]. While CAFs are indeed central, a broad body of work has also established the contributions of non-fibroblast populations including immune cells, endothelial cells, pericyte cells, and even tumor cells in shaping the TME by actively participating in ECM remodeling either directly or indirectly through the regulation of CAFs and other stromal cell activity via secreted cytokines, growth factors, and proteases [28]. Nonetheless, much of the current literature remains inconclusive, often focusing on isolated pathways or individual cell types rather than capturing the spatial and temporal complexities of ECM regulation [29–32]. A more integrated perspective is essential to identify effective therapeutic windows and optimize stromal-targeted interventions.

This review aims to challenge the prevailing fibroblast-centric paradigm by comprehensively examining current insights into the diverse cellular and molecular mechanisms governing ECM remodeling across cancer types, and beyond fibroblasts. We delineate how these non-fibroblast cell types engage in dynamic crosstalk and collectively shape the biochemical and biomechanical landscape of tumors. Furthermore, we discuss therapeutic strategies targeting these alternative stromal players to disrupt pro-tumor ECM dynamics, enhance anti-tumor immunity, and restore tissue homeostasis. By highlighting the multifaceted, non-fibroblast contributions to ECM regulation, this review advocates for a broader and more granular understanding of tumor–stroma interactions that can provide insight into the development of next-generation anti-cancer therapies.

2. Involvement of Immune Cells in ECM Remodeling

Besides being the defense machinery against the tumor, infiltrating immune cells within the TME have demonstrated to regulate the ECM. Various immune populations actively participate in ECM remodeling through the secretion of proteolytic enzymes, cytokines, and growth factors that influence matrix deposition, degradation, and organization (Figure 1). These immune-mediated alterations in the ECM can modulate tissue stiffness, regulate tumor cell invasion, and shape immune cell infiltration and function. This section discusses the role of these immune cell populations in shaping the tumor ECM and their implications for tumor progression and immune regulation.

Figure1. Contribution of immune cells in ECM remodeling.

(A) Tumor-infiltrating monocytes (TIMs) are recruited to the TME by tumor-derived chemokines such as CCL2. SPON1⁺ TIMs directly signal to tumor cells via the LRP8–TGF-β1 axis, promoting fibrillar collagen deposition. (B) Tumor-associated macrophages (TAMs) regulate the ECM both directly and indirectly by activating fibroblasts, enhancing collagen remodeling, cross-linking, fibrosis, and tissue stiffening. (C) T cells interact with the ECM via receptors such as CD44 and integrins, contributing to ECM remodeling through both direct and indirect mechanisms. (D) Tumor-associated neutrophils (TANs) remodel the ECM through secretion of neutrophil elastase (NE), matrix metalloproteinases (MMPs), and neutrophil extracellular traps (NETs). PMN-MDSCs, recruited and activated by tumor-derived galectin-1 via the STING–NF-κB–CXCL2 signaling axis, further remodel the ECM within premetastatic niches. Created in BioRender. Dawoud, A. (2026) https://BioRender.com/rc05wju.

i. Monocytes

Infiltrating tumor inflammatory monocytes (TIMs) were once considered merely as transitional immature monocytes that mature into tumor-associated macrophages (TAMs) – and only the latter of which are responsible for immunosuppressive and pro-tumorigenic functions in the TME (Figure 1A) [33]. Nevertheless, recent evidence demonstrated their independent activity in tumor progression, metastasis, and immune modulation [34]. TIMs lie under the umbrella of MDSCs which are cells physiologically reside in the bone marrow; however, tumor-secreted signals facilitate their recruitment to target organs aiding oncogenesis [35]. Indeed, bioinformatic analysis of Lung Squamous Cell Carcinoma (LUSC) reported that tumor overexpression of the chemokine CCL2 chemokine mediates TIM recruitment to the TME, a process shown to be both necessary and sufficient to promote LUSC metastasis [36]. Mechanistically, recruited TIMs were shown to express and release high levels of coagulation factor XIII-A (FXIII-A), promoting fibrin cross-linking within the TME, thereby creating a stabilized, pro-migratory fibrin scaffold that facilitates tumor cell invasion and metastatic dissemination [36]. Supporting this notion, FXIII-A is not the only coagulation-related mediator produced by monocytes that contributes to ECM remodeling. A study by Poitevin et al. reported that type I collagen–activated monocytes secrete tissue factor, the main initiator of the coagulation cascade, along with MMP9, through a ROS-dependent pathway involving NADPH oxidase and mediated by NF-κB signaling [37].

A study using a PDAC murine model reported that the anti-fibrotic activity observed with CD40 activation is linked to systemic Interferon-gamma (IFNγ) and CCL2 released in response to CD40 agonist which promotes Ly6C⁺CCR2⁺ TIM infiltration to the TME. This was followed by Matrix metalloproteinases (MMP)-dependent ECM remodeling and degradation, marked by depletion of type I collagen and FN [38]. Surprisingly, such TIM-derived antifibrotic activity highlighted a TME-specific vulnerability that can be harnessed for therapeutic interventions. Upon in vitro and in vivo analysis, myeloid cells overexpressed MMP13 and its activator MMP14 in response to IFNγ, demonstrating the key mediators of this antifibrotic function [38]. Consistent with this paradigm, Marom and colleagues demonstrated that the pro-inflammatory cytokine Tumor Necrosis Factor-alpha (TNFα) induces monocytes to secrete MMP9, which in turn promotes FN fragmentation [39]. This process is governed by a positive feedback–like mechanism whereby intact FN suppresses TNFα-induced pro-MMP secretion at the translational, but not transcriptional, level. However, local accumulation of FN fragments alleviates this inhibition, further enhancing MMP9 secretion and facilitating monocyte migration. Direct cell-to-cell contact further amplify protease release, as contact between monocytes and vascular endothelial cells induces MMP1 secretion by both cell types [40]. Co-culture of monocytes with human umbilical vein endothelial cells (HUVECs) was shown to increase MMP1 levels in the culture medium by approximately fivefold, highlighting the role of cell–cell communication in regulating ECM-degrading protease activity [40].

Tumor cells can also interact with TIMs to create in their pro-metastatic ECM remodeling independent of CAFs. In a recent study by Whately et al., F-Spondin (SPON1)–expressing TIMs were found to bind to low-density lipoprotein receptor-related protein 8 (LRP8) on NSCLC cells, leading to TGF-β1 activation. This SPON1-LRP8 mediated activation of TGF-β1 in cancer cells resulted in pronounced production of fibrillar collagens by cancer cells independent of CAFs, resulting in promotion of metastases. Furthermore, using multiplex immunohistochemical staining of a large tumor microarray of clinically annotated NSCLC tumors, the SPON1-LRP8-TGF-β1 axis was associated with worsened overall survival [41]. Another study on Her2⁺ breast cancer model has shown that overexpression of MMP11 in macrophages, rather than cancer cells, promotes monocyte recruitment and Her-2+ breast cancer cells migration [42]. MMP11-overexpressing macrophages also secreted the CCL2 chemokine, resulting in activation of CCR2 receptors on breast cancer cells, which in turn activated MAPK signaling and MMP9 expression in breast cancer cells [42].

ii. Tumor-associated macrophages

Within tumors, TAMs often acquire an M2-like, pro-fibrotic phenotype that promotes ECM stiffening, desmoplasia, and immune exclusion [43]. One of the earliest demonstrations of macrophage involvement in ECM remodeling showed that macrophages co-localize with type I collagen in fibrous capsules surrounding dormant tumors, suggesting their capacity to directly produce collagen, a process further enhanced by T lymphocytes [44]. Since then, a growing body of evidence has expanded our understanding of TAMs as both direct remodelers and indirect modulators of the ECM (Figure 1B).

A particularly compelling demonstration comes from breast cancer, where TAMs have been shown to drive stromal cell-dependent collagen cross-linking and stiffening as well as respond to the stiffened fibrotic TME by initiating collagen biosynthesis directed by TGF-β that alters metabolic reprograming thereby compromising CD8+ T cell function in breast cancer [45, 46]. Depletion of TAMs reduced stromal expression of LOX that initiates collagen cross-linking and lysyl hydroxylase 2 (LH2/PLOD2) that is critical for the formation of stable cross-links. This leads to decreased total hydroxylysine aldehyde–derived collagen cross-links (HLCCs), including dehydro-dihydroxylysinonorleucine (DHLNL) -maybe it is better to remove this since this can be formed from Lys-aldehyde, pyridinoline (Pyr), and deoxypyridinoline (d-Pyr)-cross-links, and significantly softened the stroma. These changes were accompanied by reduced metastatic burden and impaired mechanotransduction signaling (pFAK) in epithelial cells. Mechanistically, TAM-derived TGF-β activates CAFs, creating a feedback loop that enhances matrix remodeling, with CD163⁺/RELMα⁺ TAMs strongly associated with tumor aggressiveness [45].

In PDAC, TAMs secrete oncostatin M (OSM), which upregulates Lysyl Oxidase-Like 2 (LOXL2) in tumor cells that initiates collagen cross-linking and drives fiber alignment and stiffening. Consequently, targeting TAMs in orthotopic mouse models reduced Osm and Loxl2 expression, collagen alignment, suppressed metastasis and improved survival, establishing a mechanistic link between macrophage signaling, ECM remodeling, and cancer progression [47]. This study found support in a recently published single cell RNA-seq analysis that showed β-catenin signaling to be specifically activated in TAMs, and its inhibition altered TAM polarization and reduced TAM-derived OSM expression in PDAC cells. Moreover, inhibition of STAT3 reduced OSM-induced LOXL2 expression and subsequent epithelial-to-mesenchymal (EMT) programming [48]. Notably, Afik et al demonstrated that depletion of TAMs, in an orthotopic mouse model of CRC led to reduced expression of types I and XIV collagens in CAFs [49]. Similarly, the scavenger receptor Stabilin-1 expressed on TAMs was shown to mediate the endocytosis of secreted protein acidic and rich in cysteine (SPARC or osteonectin), a collagen-binding glycoprotein [50, 51]. In glioblastoma (GBM), deletion of Stabilin-1 disrupted collagen clearance and impaired tumor growth via ECM accumulation [52]. Consistent with this, Atkins et al used an engineered 3D HA-based hydrogel, recapitulating the mechanical features of GBM TME, to demonstrate that M2-polarized macrophages induce transcriptional and phenotypic reprogramming in GBM stem cells closely associated with the highly invasive GBM mesenchymal subtype, in part through secretion of pro-invasive ECM-associated factors such as TGF-β–induced protein (TGFBI, also referred as BIGH3) [53]. Mechanistic and transcriptomic analysis in a murine orthotopic PDAC model demonstrated that CXCR4+ macrophages secreting SPARC through CXCR4/PI3K/Akt pathway are involved in ECM remodeling, tumor proliferation and migration [54]. In High-grade serous carcinoma (HGSOC), TAMs in malignant ascites secrete pro-migratory ECM proteins including TGF-β, tenascin-C, and FN1, which are linked to poor progression-free survival and metastasis, as confirmed by neutralization and knockdown studies [55]. Another study reported the role of M2-like macrophages in actively contributing to ECM remodeling by epigenetically upregulating LOX via H3K27 demethylation by Jmjd3 demethylase, thereby elevating collagen levels to facilitate cross-linking and enhanced tumor cell migration; an effect reversible by H3K27 demethylase inhibitors [56]. Moreover, in lung fibrosis, macrophages promote expression of type VI collagen (ColVI) through the AP-1 transcription factor Fra-2, activating myofibroblasts. Disruption of the Fra-2/ColVI axis in macrophages significantly reduces fibrogenesis, highlighting a paracrine pro-fibrotic role of macrophages [57]. Macrophages serve as the primary source of cathepsin S (CTSS) in fibrotic liver tissue, driving ECM remodeling by cleaving type XVIII collagen to release endostatin and promoting hepatic stellate cell activation via integrin α5β1 signaling, thereby contributing to liver fibrogenesis [58].

Beyond matrix deposition, TAMs also facilitate ECM degradation through mannose receptor (MRC1)-mediated endocytosis of collagen, followed by lysosomal degradation [59]. In the context of dermal tissue remodeling, pathways such as CCL2 signaling, GM-CSF, IL-13, IL4-IL4 receptor a (IL4Ra) signaling instruct phenotypically diverse macrophage subsets to mediate collagen and fibrin endocytosis utilizing MRC1 and urokinase-type plasminogen activator receptor-associated protein (uPARAP) [60]. In desmoplastic tumors such as PDAC, TAMs undergo metabolic reprogramming that supports a profibrotic phenotype. The lysosomal degradation of collagen results in the intracellular accumulation of collagen-derived amino acids, notably elevating arginine biosynthesis, inducing elevated nitric oxide synthase (iNOS) levels, and reactive nitrogen species production thereby establishing a feed-forward loop that reinforces stromal remodeling and tumor progression [61]. Madsen et al. also reported that collagen-degrading TAMs, derived from CCR2⁺ monocytes, are present across tumors of epithelial, mesenchymal, and neural crest origin [62]. In CRC, M2-like macrophages promote ECM remodeling and tumor invasion by inducing MMP9 expression and EMT in cancer cells via TNFα and Leukotriene D4 (LTD4) signaling. Consequently, inhibition of these pathways reduces MMP9 activity and invasiveness [63].

Collectively, these findings establish TAMs as not merely immunomodulatory entities of the stromal compartment but key regulators of ECM dynamics in the TME.

iii. T cells

Beyond their classical immune-regulatory and cytotoxic roles in the TME, T lymphocytes have been recognized as active participants in ECM remodeling (Figure 1C). Recent evidence indicates that activated CD8⁺ T cells can produce the collagen remodeling enzyme, LOX [64]. Haj-Shomaly et al. (2022) demonstrated that paclitaxel (PTX) chemotherapy induced host-mediated ECM remodeling of the pulmonary metastatic niche with an increase in LOX expression. Moreover, adoptive transfer of CD8⁺ T cells from PTX-treated mice to naïve immunodeficient mice induced ECM remodeling in the lungs, confirming the source of LOX as coming from the CD8⁺ T cells. Pharmacological inhibition of LOX mitigated PTX-induced ECM remodeling and subsequent metastasis [64].

In endometrial cancer, infiltrating CD3⁺ T lymphocytes correlated spatially with areas of high MMP9 protein expression, however, no corresponding increase in MMP9 or MMP2 mRNA was detected in the lymphocytes, suggesting that while T cells may not directly produce ECM-degrading enzymes, their presence could modulate local protease activity and facilitate ECM breakdown, aiding tumor invasion [65]. However, earlier work demonstrated the intrinsic capacity of normal human T cells to produce gelatinases A (MMP2) and B (MMP9), with MMP9 being expressed constitutively, and both enzymes induced upon activation, enabling them to degrade key basal lamina components such as type IV and V collagens, which could be specifically blocked by the metalloproteinase inhibitor GM6001 [66]. Complementary findings in T-cell lymphomas further showed that T-cells overexpress cell surface molecule emmprin (also known as CD147 or basigin) that stimulates adjacent fibroblasts to produce pro-MMP2 and active MMP2, with emmprin-positive lymphoma cells inducing fibroblast-derived MMP2 both in vitro and in vivo [67].

Similarly, CD4⁺ T cell-derived signals can indirectly drive ECM remodeling by activating resident stromal cells that reshape the tissue microenvironment in both cancer and chronic inflammation. In an inflammatory model of tissue remodeling, activated CD4⁺ T cells promoted fibroblast-mediated collagen contraction and degradation through the induction of MMP9 activity and secretion of pro-inflammatory cytokines such as TNFα and IL-6 [68]. Pro-tumoral Regulatory T (T regs) cell ablation in the MMTV-PyMT murine breast cancer model led to marked alterations in the tumor ECM, including reduced collagen content and fiber organization, along with increased fibronectin and laminin levels impairing migratory ability of tumor cells resulting in reduced lung metastases (bioRxiv 2025.10.25.684515). Tregs were demonstrated to orchestrate a TAM-dependent ECM program mediated by IFN-γ signaling that facilitates invasion and metastasis (bioRxiv 2025.10.25.684515).

T cells also possess diverse ECM-interacting receptors which are dynamically regulated by inflammatory cues. A previous study identified α4β1 integrin as a distinct fibronectin receptor on T lymphocytes, functioning alongside α5β1. While α5β1 binds the canonical RGD (Arginine–Glycine–Aspartic acid)-containing domain of fibronectin, α4β1 specifically recognizes the alternatively spliced CS-1 region within the IIICS domain, mediating adhesion at focal contacts [69]. Pro-inflammatory cytokines and chemokines rapidly enhanced CD44 receptor activation on T-lymphocytes and adhesion to its ligand, HA [70] thereby promoting CD44-dependent adhesion, spreading, and polarization of T cells under both static and shear flow conditions. This process involves cytoskeletal reorganization and PTM of CD44, enabling firm T cell–matrix interactions that facilitate migration into inflamed tissues [71].

iv. Neutrophils

Neutrophils, traditionally considered the first-line of defense in innate immunity, have recently gained recognition as active modulators of ECM architecture within the TME (Figure 1D) [72]. Tumor-associated neutrophils (TANs) infiltrate solid tumors in response to chemokines and release a repertoire of proteolytic enzymes, including neutrophil elastase (NE), cathepsin G, and MMP9, which are capable of degrading structural ECM components, thereby loosening matrix barriers and facilitating tumor cell invasion and angiogenesis [72]. Neutrophils also remodel ECM through the formation of neutrophil extracellular traps (NETs) [73]. NETs are composed of decondensed neutrophil-derived nuclear chromatin bound with serine proteases and MMPs [74]. During the progression of Hepatocellular carcinoma (HCC), particularly those with portal vein tumor thrombosis, the HCC-derived cytokine IL-8 was demonstrated to elevate NET formation, correlating with poor prognosis in this cancer type. NET-associated cathepsin G was shown to enhanced cancer cell invasion in vitro and in vivo [75]. In breast cancer, neutrophils were shown to induce EMT in tumor cells via tissue inhibitor of matrix metalloprotease (TIMP-1). Reciprocally, breast cancer cells undergoing EMT enhanced neutrophil TIMP-1 secretion by CD90 in a cell-contact manner [76]. In prostate cancer xenografts, tumor-infiltrating polymorphonuclear MDSCs (PMN-MDSCs) were demonstrated to exert protumorigenic actions partially through neutrophil elastase (ELANE), released upon granulocyte activation, stimulating proliferation, migration, and invasion of prostate cancer cells [77]. Besides, in Head and neck squamous cell carcinoma (HNSCC) models, PMN-MDSCs contribute to ECM remodeling within the premetastatic niche, where they are recruited and activated by tumor-derived Galectin-1 via STING–NF-κB–CXCL2 signaling, thereby facilitating metastasis [78].

3. Vascular Cells in ECM Dynamics

Endothelial cells (ECs) in developing blood vessels are stabilized by pericyte recruitment, a process that occurs in both normal tissues and in pathological angiogenesis, including tumors [79, 80]. ECs and pericytes are central regulators of ECM remodeling within tumors, shaping the vascular niche and influencing tumor progression, invasion, and metastasis [81, 82]. Tumor endothelial cells (TECs) arise from multiple origins and exhibit profound phenotypic diversity and functional differences compared to normal ECs, including structural disorganization and a heterogeneous vasculature with functional abnormalities in permeability, proliferation, mechanosensing, and inflammatory signaling [83]. Early studies demonstrated that ECs synthesize a basement-membrane matrix rich in laminin, collagen and other ECM proteins, providing structural stability and organizing cues for new vessel formation [84, 85].

During tumorigenesis, TECs exhibit altered ECM remodeling capacities where endothelial secretion of MMP2 can facilitate breast cancer cell transmigration across the endothelial–basement membrane barrier, highlighting EC-derived proteolysis as a direct contributor to tumor invasion and vascular remodeling [86]. In a breast cancer model, a subpopulation of TECs was demonstrated to undergo a spectrum of TGFβ-driven endothelial-to-mesenchymal transitions (EndMT), showing strong upregulation of α-SMA and reduced motility, contributing to plasticity and dysfunction of the tumor vasculature [87]. In addition, TEC-derived extracellular vesicles (EVs) further modulate the ECM and promote tumor progression by carrying proteins, such as Intercellular adhesion molecules, vascular endothelial-cadherin, E-selectin, platelet EC adhesion molecule-1, endoglin, and endothelial nitric oxide synthase, fostering a tumor-permissive niche [88]. Moreover LOXL2 and type I collagen were reported to be components of non-neoplastic EVs secreted by ECs subjected to hypoxic stress conditions [89].

Recruitment of pericytes to ECs is driven by the Platelet Derived Growth Factor Subunit B (PDGFB)/ Platelet-Derived Growth Factor Receptor (PDGFR) axis, which also promotes PDGFR-dependent vitronectin deposition in the vascular basement membrane. Vitronectin then engages integrin αvβ3 on ECs leading to NFκB activation-dependent expression of VEGFA [90]. In contrast, autocrine VEGFA transcription in TECs leads to upregulation of the antiapoptotic protein Bcl-w which reinfores pro-survival signaling within the angiogenic tumor vasculature [90, 91].

In pilocytic astrocytomas and GBM, but not in lower-grade gliomas, PDGFRβ⁺ pericytes were established as the source of POSTN, where it is essential for vascular growth, branching, and endothelial junction integrity [92]. Single-cell RNA sequencing of tumor pericytes (TPCs) from CRC patients, with and without liver metastases, identified a TCF21^high subpopulation, termed “matrix-pericytes,” that correlated strongly with metastatic potential and poor patient prognosis. Conditional knockout of Tcf21 in pericytes suppressed ECM remodeling and liver metastasis in mice, while co-injection of TCF21^high pericytes with CRC cells enhanced metastatic colonization. Loss of integrin α5 attenuated DNA hypermethylation through downregulation of a DNMT1/FAK/PI3K/AKT axis, thereby maintaining TCF21 expression in pericytes [93, 94]. Pericytes undergo tumor-induced phenotypic switching that enhances their fibronectin-rich ECM production [95], and studies using pericyte-specific deletion of the transcription factor Klf4 demonstrate that this reprogramming is essential for pericyte expansion and lung pre-metastatic niche formation, thereby promoting metastatic seeding [96]. Besides the direct role of pericytes in ECM remodeling, these cells also undergo pericyte–fibroblast transition (PFT) via PDGF-BB-PDGFRβ signaling, activated pericytes then acquire mesenchymal progenitor features [97], an observation also shown to be enhanced by exposure to radiation [98]. Overall, ECs and pericytes act as both architects and modulators of the tumor ECM.

4. ECM remodeling by tumor cells

Tumor cells remodel ECM through both direct and indirect mechanisms to create a microenvironment that supports malignancy. They modify the ECM by secreting a range of enzymes that modify the biochemical and mechanical characteristics of collagen. This includes PTMs of procollagen, such as hydroxylation of lysine residues by the 2-oxoglutarate 5-dioxygenases (PLODs), which encode the family of lysyl hydroxylases (LH1–3), and hydroxylation of proline residues by the prolyl 4-hydroxylases (P4Hs) [99]. Moreover, covalent inter-molecular cross-linking of collagen and elastin by tumor-derived LOX and its isoforms (LOXL1–LOXL4) promotes ECM stiffening. Tumor cells can also directly interact and respond to matrix components that act as ligands to receptors such as integrins and DDRs expressed on tumor cells [100, 101]. Additionally, malignant cells can also influence ECM remodeling by recruiting stromal cells and releasing EVs, which propagate pro-remodeling signals both locally and systemically [102]. These mechanisms are discussed in detail in this section and illustrated in Figure 2.

Figure 2. Tumor-driven ECM remodeling.

(A) Secretion of ECM-modifying enzymes: Hypoxia-induced NF-κB signaling elevates LH1 which ameliorate tumor progression, while LH2 promotes stable HLCCs and stroma stiffening. LH3 enhances collagen IV deposition and inhibits FOXO3, activating Survivin and Twist-1/β-catenin/AKT pathways. Collagen could bind integrin β1 on endothelial cells leading to the disruption of their barrier integrity. Tumor cells overexpress METTL3 stabilizes P3H4 mRNA and its transcription factor ETV4, while the invasive front secretes MMP-2/14 along with TIMP-2, driving localized ECM remodeling. (B) Tumor cell–matrix interactions and mechanical signaling: Tumor cells mechanically interact with the ECM via receptors such as DDR1 and DDR2, which regulate collagen remodeling and fiber alignment and limits CD8⁺ T cell infiltration to TME. Metastatic cells deposit collagen III whose presence initiates disseminating cells dormancy; however, dormant cells can be reactivated via DDR1–STAT3 signaling. Tumor cells also migrate along fibronectin (FN) gradients through α5β1 integrin–MENA interactions, sensing haptotactic cues to direct invasion. (C) Indirect ECM remodeling via stromal cell recruitment: Tumor cells secrete IL-1α, activating stellate cells to overexpress MMP-1/3 while downregulating TIMP-2/3. At the invasive front, cancer cells activate adipocytes to secrete MMP-11. IL-6 secretion by adipocytes stimulate JAK/STAT3 and PI3K/AKT pathways in tumor cells, upregulating their PLOD2/LH2 expression and enhancing their metastatic potential. Tumor-derived osteopontin (OPN) binds CD44 and α5β3 integrins on fibroblasts, activating pro-inflammatory gene programs and increasing fibroblast motility. (D) Tumor cell derived-Extracellular vesicle mediated ECM remodeling: Under hypoxic conditions, tumor cells release extracellular vesicles carrying MMP-2 and MMP-9, promoting fibronectin and collagen deposition at distant organs and facilitating CD11b⁺ myeloid cells recruitment to the tumor microenvironment, establishing pro-metastatic niches. Created in BioRender. Dawoud, A. (2026) https://BioRender.com/fjvvpzr.

i. Secretion of ECM modifying enzymes

Within this context, the PLOD/LH family members are increasingly recognized as key drivers of tumor aggressiveness through ECM remodeling and modulation of oncogenic signaling (Figure 2A). Among these, PLOD1/LH1 hydroxylates lysine residues in the helical domain of fibrillar collagens and is highly expressed in aggressive tumors such as GBM, particularly the mesenchymal (MES) subtype, where its elevated levels correlate with poor patient survival. Functionally, PLOD1/LH1 was shown to promote cell proliferation, migration, MES transition, and resistance to apoptosis in patient-derived glioma stem cells, while its overexpression accelerated tumor growth in vivo. These effects were linked to activation of the NF-κB pathway, which were amplified under hypoxic conditions [103].

Extending this paradigm, gain- and loss-of-function studies revealed that LH2 activity, the enzyme encoded by the gene PLOD2, revealed an important role in controlling the ‘quality’ rather than the ‘quantity’ of cross-linked type I collagen [24, 104]. Lysine hydroxylation in collagen telopeptides, catalyzed by LH2, is a prerequisite for the formation of stable cross-links. Whereas unmodified type I collagen that has been cross-linked by LOXs remains proteolytically labile, LH2-mediated lysine hydroxylation in the telopeptides of fibrillar collagens promotes the formation of stable HLCCs over lysine aldehyde–derived cross-links (LCCs), which shifts the collagen matrix toward a biochemically stiffer and stable protease-resistant stroma that facilitates a more invasive tumor phenotype [104, 105]. Notably, enzymatically active LH2 was demonstrated to be secreted by lung cancer cells into the extracellular space, where it modifies collagen by promoting the formation of HLCCs [106]. This extracellular activity provided further evidence that tumor cell-derived LH2 drives matrix remodeling to support the metastatic potential of lung cancer [104]. Consistent with this, LH2 was also shown to be significantly upregulated in late-stage Oral squamous cell carcinoma and in cases with regional lymph node metastasis, and its expression was associated with poor survival outcomes [107]. Supporting this observation, a mechanistic study using HNSCC cells and an orthotopic mouse model demonstrated that LH2 specifically increased HLCCs without altering the quantity of collagen cross-links. This qualitative shift in the biophysical properties of collagen cross-links led to a mechanically stiffer and stable ECM, which in turn promoted tumor cell migration, invasion, and metastasis [24]. Further expanding its role across tumor types, in sarcoma, hypoxia-induced PLOD2/LH2 was shown to modify and promote secretion of type VI collagen (COLVI) by tumor cells, which binds to integrin β1 on lung endothelial cells, disrupting barrier integrity via F-actin polymerization, promoting trans-endothelial migration and metastasis [108]. Together, these findings indicate LH2 as a tumor cell–intrinsic regulator of collagen architecture and mechanical properties within the tumor microenvironment that favors tumor aggression.

The third PLOD family member (PLOD3/LH3), possesses galactosyl hydroxylysine-glucosyltransferase activity for collagen [109, 110]. In gastric cancer, elevated PLOD3 levels have been linked to increased cell proliferation, tumor growth, and resistance to Trastuzumab therapy, mediated through repression of the tumor suppressor FoxO3 and upregulation of survivin [111]. PLOD3 overexpression in renal cell carcinoma was shown to activate TWIST1, which in turn stimulates β-catenin and AKT signaling pathways, leading to enhanced tumor cell proliferation, migration, invasion, EMT, and tumor growth [112]. Further, PLOD3 knockdown inhibited these malignant features and disrupted ECM remodeling, underscoring its central role in nurturing a pro-tumorigenic microenvironment [112].

Tumor-initiating cells (TICs) in hepatocellular carcinoma contribute to a stiff ECM architecture by secreting the LOX and LOXL families of enzymes. In turn, TICs sense and respond to this increased stiffness by upregulating integrin α7 (ITGA7)-mediated signaling pathways, activating FAK and ERK1/2 phosphorylation, thereby fostering a supportive niche for self-renewal and tumorigenicity [113]. A similar mechanism was observed in invasive ductal carcinoma, where the inflammatory cytokine OSM induced the expression of LOXL2 in tumor cells. The secretion of enzymatically active LOXL2 induced extensive type I collagen cross-linking, leading to fiber alignment and stromal stiffening, thereby enhancing tumor cell invasion [114]. A recent study demonstrated LOXL2-mediated collagen remodeling and immune exclusion in CRC patients and in vivo models, the expression of LOXL2 being regulated by a G-protein coupled receptor GPR4 [115].

Complementing the role of cross-link associated enzymes, prolyl 3 hydroxylase family member 4 (P3H4) plays a crucial role in ECM remodeling driven by tumor cells across multiple cancers. P3H4 was demonstrated to be significantly overexpressed in bladder cancer where it contributed to tumor proliferation, migration, invasion, and EMT. Its expression was regulated both by the m⁶A methyltransferase METTL3, which stabilizes P3H4 mRNA, and the transcription factor ETV4, which directly activated P3H4 transcription [116]. Functional studies confirmed that reducing P3H4 levels impaired bladder cancer growth and metastatic traits in vitro and in vivo [117]. Similarly, in LUAD, elevated expression of P3H4 correlated with advanced tumor stage, poor patient survival, and promoted metastasis and proliferation by interacting with EGFR signaling to influence metabolic processes that support tumor progression [118].

Proteolytic remodeling of the ECM is also shaped by tumor-secreted protease inhibitors such as SerpinE2 (protease nexin-1), elevated level of which was demonstrated to enhance metastatic potential of breast cancer by promoting ECM remodeling through regulation of collagen breakdown and matrix-degrading enzymes. Its suppression led to dense type I collagen accumulation, increased TIMP-1 (a natural MMP inhibitor), and reduced infiltration of pro-tumorigenic, M2-polarized macrophages[119]. Similarly, in melanoma, co-expression of MMP2, MT1-MMP (MMP14), and TIMP-2 at invasive fronts highlighted the spatial regulation of proteolysis in facilitating tumor invasion. Invasive potential was closely linked to tumor cell–intrinsic upregulation and activation of MMP2, coordinated by membrane-bound protease complex MT1-MMP and its cofactor TIMP-2 [120].

ii. Tumor Cell-Matrix Interactions and Mechanical Signaling

Tumor cells not only secrete ECM-modifying enzymes but also engage with the ECM through transmembrane receptors that transmit mechanical cues into intracellular signaling events (Figure 2B). Mechanical features of the remodeled ECM including increased matrix stiffness due to excessive deposition, altered fiber organization, and tensile forces arising from collagen crosslinking and alignment, are sensed primarily through integrins and other adhesion receptors, leading to focal adhesion assembly and activation of downstream signaling pathways such as FAK/Src, Rho/ROCK, and downstream ERK/MAPK and YAP/TAZ signaling [121–123]. These signaling cascades regulate cytoskeletal tension, adhesion dynamics, gene expression, cellular proliferation, survival, and migration [124]. Thus, mechanical cues are not merely a consequence of remodeling but play an active role in driving tumor progression.

In a triple-negative breast cancer (TNBC) model, genetic deletion of DDR1, a collagen-binding receptor on tumor cells, significantly enhanced the penetration of CD8⁺ T cells and suppressed tumor growth in immunocompetent, but not immunodeficient, mice. Importantly, the extracellular domain (ECD) of DDR1, independent of its kinase activity, was shown to bind collagen and orchestrate the linear alignment of fibers that creates an ECM-based barrier against immune infiltration [100]. A study by Albrengues et al. revealed that disseminated tumor cells secrete and organize type III collagen to construct an autocrine ECM niche that can maintain cellular dormancy at secondary sites. Loss of tumor-derived type III collagen disrupted this niche and led to tumor cell reactivation through DDR1-STAT1 signaling, indicating a critical role for tumor-intrinsic ECM production in dormancy regulation. Clinically, elevated type III collagen levels in lymph node-negative HNSCC specimens further support its association with reduced metastatic activity [125]. Recent work in GBM demonstrated that tumor cell–derived type VI collagen, a microfibril-forming collagen, can regulate the mechanical properties of the microenvironment by functioning as a matrix crosslinker that increases ECM stiffness, activating integrin β1 and integrin αvβ3-dependent mechanotransduction pathways. This promotes β-catenin signaling, ZEB1 expression, and mesenchymal reprogramming, enhancing invasive behavior, particularly under hypoxic and bevacizumab-resistant conditions [126].

Apart from collagen, tumor cells also exploit gradients of other ECM proteins such as FN to guide migration, a phenomenon known as haptotaxis. In breast cancer, this directed movement was mediated by the synergistic interaction between α5β1 integrin and MENA^INV, a pro-metastatic isoform of the actin-regulator MENA. The interaction with FN enhanced focal adhesion signaling, and tumor cells concurrently restructured the surrounding ECM, reinforcing their migratory trajectory. Elevated MENA^INV expression has been associated with increased FN deposition, higher recurrence rates, and worse patient survival outcomes [127].

iii. Indirect ECM Remodeling via Stromal Cell Recruitment

Cancer cells also alter the ECM indirectly by recruiting and reprogramming stromal cells such as fibroblasts, stellate cells, and immune cells (Figure 2C). In pancreatic tumors, cancer cell-derived IL-1α was demonstrated to promote expression of MMP1 and MMP3 in pancreatic stellate cells while suppressing TIMP-2 and TIMP-3, shifting the balance toward ECM degradation and enhancing cancer cell invasion [128]. In a parallel study, breast cancer cells at the invasive front induced adjacent adipocytes to express MMP11 (Stromelysin-3, ST3), a MMP linked to poor prognosis. This tumor-induced ST3 expression remodeled the adipocyte membrane, facilitating cancer cell infiltration and survival in connective tissue, while also suppressing adipocyte differentiation and fat homeostasis, suggesting a dual role in both ECM remodeling and adipocyte reprogramming [129].

Beyond protease-mediated remodeling, tumor cells can also reprogram stromal populations to actively produce ECM components that support invasion. In GBM, tumor-associated mesenchymal stem-like cells enhance ECM remodeling by increasing HA production through HA synthase (HAS2) induction, driven by autocrine C5a–ERK MAPK signaling. The accumulated HA was shown to activate the Receptor for Hyaluronan-Mediated Motility (RHAMM) on GBM cells, triggering intracellular pathways that support tumor invasion [130]. Another study revealed that adipocyte-secreted IL-6 and leptin drive breast cancer metastasis by activating JAK/STAT3 and PI3K/AKT signaling to induce PLOD2 expression in tumor cells [131]. Co-culture of breast cancer cells with adipocytes increased PLOD2 expression, and enhanced cell migration and invasion whereas, knockdown of PLOD2 reversed these effects. Clinically, high PLOD2 levels correlated with poor prognosis, highlighting its role in adipocyte-mediated tumor progression [131].

In breast cancer, tumor-derived Osteopontin (OPN) was shown to reprogram normal mammary fibroblasts into CAF-like cells mediated through its interaction with key receptors, CD44 and αvβ3 integrin, thereby inducing a proinflammatory gene signature and enhancing their motility [132]. Notably, OPN alone was sufficient to drive fibroblast reprogramming, while neutralization of OPN blocked the tumor-induced CAF phenotype both in vitro and in vivo [132]. This shows an indirect mechanism of ECM remodeling, whereby tumor cells reprogram fibroblast behavior rather than directly altering the matrix. In GBM, OPN-mediated activation of CD44 within the perivascular niche has been shown to promote cancer stem cell phenotypes and radiation resistance through γ-secretase–dependent release of the CD44 intracellular domain, which engages transcriptional co-activators CBP/p300 to potentiate HIF-2α driven transcriptional programs associated with hypoxia, stemness, and aggressive tumor growth, thereby reinforcing niche-driven invasion [133].

Tumor cell-derived factors drive ECM remodeling by modulating macrophage behavior within the TME. In CRC, cancer cells secrete collagen triple helix repeat containing 1 (CTHRC1), that not only directly remodels the ECM but also indirectly promotes fibrosis by enhancing M2-like polarization and recruitment of TAMs by upregulating CCL15 via the TGFβ/Smad pathway, facilitating macrophage chemotaxis and sustaining a pro-tumorigenic, immunosuppressive microenvironment [134]. Similarly, CRC cells were also shown to secrete Spondin-2 (SPON2), an ECM glycoprotein that facilitates cytoskeletal remodeling and trans-endothelial migration of monocytes via the integrin β1/PYK2 axis. SPON2 not only promoted tumor cell migration but also enhanced recruitment and M2-polarization of TAMs by upregulating immunosuppressive cytokines (IL-10, CCL2, CSF1). accelerating tumor growth and invasion [135].

Collectively, these studies highlight that tumor cells reshape the ECM indirectly by co-opting diverse stromal populations through distinct but convergent signaling pathways, ultimately coordinating matrix remodeling, stromal reprogramming, and invasive behavior.

iv. Extracellular Vesicle–Mediated ECM Remodeling

Tumor-derived EVs, especially exosomes, particularly under hypoxic conditions, play a key role in conditioning distant microenvironments to support metastatic colonization (Figure 2D). Prostate cancer cells release exosomes enriched with proteases such as MMP2 and MMP9, which promote fibronectin and collagen deposition while enhancing immune cell infiltration at distant, pre-metastasis niches. In murine models, systemic delivery of these hypoxia-derived exosomes increased collagen and fibronectin deposition along with CD11b⁺ myeloid cell recruitment in peripheral metastatic sites. Proteomic analyses revealed distinct ECM-remodeling cargo in these vesicles, suggesting a specialized role in priming the pre-metastatic niche [102].

5. ECM remodeling by other, non-tumor cell types

Natarajan et al. demonstrated that under hypoxic conditions, mesothelial cells promote ECM remodeling in HGSOC by enhancing type I collagen production LOX mediated cross-linking, thereby promoting tumor invasion during peritoneal dissemination. Pharmacologic inhibition of LOX significantly reduced ECM remodeling and tumor burden in omental metastases [136].

Early work demonstrated that breast cancer cells convert neighboring adipocytes into cancer-associated adipocytes (CAAs), characterized by de-lipidation and elevated IL-6, IL-1β, and MMP11, changes that heighten tumor cell invasiveness and are detectable in human tumors [137]. Beyond this CAA phenotype, adipocytes contribute to breast cancer progression by releasing type VI collagen, which engages the NG2/chondroitin sulfate proteoglycan receptor on malignant ductal epithelial cells and triggers sequential activation of Akt and β-catenin signaling pathways [138]. Moreover, the cleaved COL6α3 fragment, endotrophin (ETP), acts in concert with elevated TGF-β activity to further drive fibrosis, angiogenesis, inflammation, EMT, and metastatic dissemination [139]. Since adipocytes are a crucial part of breast cancer microenvironment, they can potentially influence tumor-intrinsic programs of ECM remodeling. Wu et al. reported that exposure of TNBC cells exposed to adipogenic conditioned medium and palmitic acid resulted in upregulation of POSTN, MMPs, and stemness-related molecules through MAPK/ERK signaling, thereby enhancing ECM degradation, invasiveness, and chemoresistance [140].

A recent study by Dessauge and colleagues showed that lymphoma B cells engage in bidirectional crosstalk with murine leptin receptor-positive (LepR⁺) bone marrow stromal cells, inducing their reprogramming toward a CAF-like phenotype [141]. This interaction drives ECM and TGFβ dysregulation within the bone marrow niche, contributing to the establishment of a supportive stromal microenvironment that may facilitate lymphoma persistence.

Epithelial cells can actively contribute to ECM remodeling and tumor progression by shaping the physical and biochemical properties of the surrounding microenvironment. A study by Lee et al. has demonstrated that mammary epithelial cells (MECs) promote early breast cancer cell dissemination by secreting soluble laminin, which induces basement membrane remodeling and facilitates carcinoma cell invasion [142]. MEC-derived laminin activates integrin β1 signaling in neighboring breast cancer cells, driving the formation of microtubule-dependent membrane protrusions that enhance cellular contractility and matrix invasion. These findings reveal that non-malignant epithelial cells are not merely passive components of the tumor microenvironment but can actively regulate ECM dynamics and create permissive niches that facilitate tumor cell dissemination. Accumulating evidence suggests that lymphatic endothelial cells (LECs) engage in dynamic bidirectional interactions with ECM remodeling [143]. Detry et al. demonstrated that LEC-derived MMP2 promotes collagen fiber remodeling and facilitates a mesenchymal-like migratory phenotype, thereby enhancing lymphangiogenesis [144]. These observations raise the possibility that elevated MMP2 expression in LECs may similarly promote tumor lymphatic invasion by remodeling the ECM and facilitating lymphatic vessel infiltration into the tumor microenvironment. Reciprocally, ECM stiffening functions as a biomechanical regulator of LEC behavior by promoting proliferation and migration through the mechanotransducer FAT1, which facilitates β-catenin nuclear translocation and regulates cytoskeletal remodeling and focal adhesion dynamics [145].

Mast cell infiltration has also been implicated in TME remodeling through the secretion of ECM-modifying enzymes, including MMP9, tryptase, and chymase [146]. Recently, mast cell-derived heparanase was shown to promote breast cancer stemness by cleaving ECM-associated heparan sulfate proteoglycans, generating bioactive heparan sulfate fragments that stimulate the release of soluble factors and activate the mucin 1 (MUC1)/estrogen receptor signaling axis in HER2-negative tumor cells [147]. This pathway enhances mammosphere formation, induces stemness-associated gene expression, and promotes tamoxifen resistance, highlighting ECM remodeling as a mechanism linking mast cells to subtype-specific breast cancer progression.

6. Integrated Cellular Crosstalk in ECM Remodeling

The cell type-specific mechanisms described above do not operate independently but instead constitute an interconnected ecosystem in which all cells cooperatively drive ECM remodeling. Tumor cells and CAFs secrete cytokines, chemokines, growth factors, and extracellular vesicles that recruit and activate immune and vascular cells, which then directly secrete ECM components or enzymes that modify the existing ECM. This intricate functional interplay was recently demonstrated in PDAC, where triggering receptors expressed on myeloid cells 2 (TREM2⁺) macrophages drove vascular remodeling while MMP12⁺ macrophages promoted CAF-mediated ECM reorganization at the invasive front [148]. Superimposed on these multicellular interactions, TGF-β and hypoxia act as central regulators that amplify this circuit, coordinately driving fibroblast activation, matrix deposition, ECM crosslinking, angiogenesis, and immunosuppression. The resulting ECM accumulation and reorganization increases tissue stiffness and generates biomechanical stress, including solid stress, elevated interstitial fluid pressure, and altered shear forces, which further promotes tumor progression and sustains stromal activation through feed-forward mechano-transduction [149]. Consistent with this concept, solid stress and interstitial fluid pressure co-evolve during tumor progression, reaching levels sufficient to compress intratumoral blood vessels while deforming peripheral vasculature [150]. Moreover, genetic ablation of the collagen-binding integrin α11β1 in stromal fibroblasts reduces interstitial fluid pressure and disrupts collagen organization, directly linking fibroblast mechano-sensing to tumor biomechanics [151]. Thus, ECM remodeling is best understood as an integrated ecosystem-level process orchestrated by continuous communication between malignant and stromal cell populations.

7. Therapeutic Implications

Growing recognition of diverse contributors to ECM remodeling, beyond the fibroblast-centric paradigm, highlights multiple therapeutic targets across the TME. Upregulated LOX/LOXL enzymes and LH2 generate abundant, stabilized covalent collagen cross-links, driving pathological stiffening. LOX blockade with β-aminopropionitrile (BAPN) disrupts collagen stiffening and reduces endothelial cell migration and angiogenesis [152], and alleviate EMT in paraquat-induced pulmonary fibrosis [153]. While Minoxidil reduced activity of LH1–3 [154], several high-throughput drug screens and in silico approaches identified small molecule inhibitors specific to LH2 including KD-1 (4,4,4-Trifluoro-1-(pyridine-3-yl)-1,3-butanedione) and KS122–0485428 [155–158]. Simtuzumab, a monoclonal antibody against LOXL2, was investigated in patients with Idiopathic pulmonary fibrosis (IPF) in a phase II trial but it did not improve progression-free survival [159]. These enzymes are induced by key upstream regulators such as TGF-β and HIF-1α, placing ECM mechanics directly under the influence of oncogenic and microenvironmental signals. A LOX function‐blocking antibody in combination with gemcitabine was shown to alter stroma and increase tumor vascularization, immune cell infiltration and increase survival by manipulating a ‘LOX/hypoxia’ signature in a PDAC model [160]. Bintrafusp Alfa, a bifunctional fusion protein targeting TGF-β receptor II (a TGF-β “trap”) and programmed cell death ligand 1 (PD-L1), was used in second-line treatment of patients with NSCLC in a Phase I trial [161]. Galunisertib, a small molecule inhibitor of TGFβ receptor combined with gemcitabine resulted in improved survival of patients with unresectable pancreatic cancer in a phase I/II trial [162]. Vactosertib, a TGFβ inhibitor, and imatinib combination has demonstrated to have promising clinical activity in patients with progressive, locally advanced desmoid tumors in phase Ib/II study [163].

ECM-directed approaches are also emerging in immunotherapy. Along similar lines, dual inhibition of HASs and hyaluronidases (HYAL) in a patient derived xenograft model of GBM altered mechanical and metabolic properties of the microenvironment by targeting tumor-derived hyaluronan (bioRxiv 2024.01.05.574065). Moreover, suppressing hyaluronan synthesis with 4-methylumbelliferone increased γδ T-cell infiltration in PDAC mice xenograft model [164].

Additional lines of investigation point toward therapeutic targeting ECM–cell interactions, specifically matrix-associated receptors. Ablation of Ddr1 in tumors was shown to reverse immune exclusion caused by stromal trapping in TNBC mouse model [165]. DDR1 ECD-neutralizing antibodies were able to disrupt collagen fiber alignment and enhance intratumoral penetration of immune cells [165]. A novel DDR1 inhibitor, KI-301690, was shown to enhance anti-cancer efficacy of gemcitabine in pancreatic cancer xenograft model [166]. Inhibition of leukocyte-associated immunoglobulin-like receptor-1 (LAIR-1) signaling using NC410, a fusion protein consisting of 2 LAIR-2 molecules on an IgG1 backbone, in combination with bintrafusp alfa remodeled the tumor collagenous matrix, enhanced tumor infiltration and activation of CD8+ T cells, and repolarized suppressive macrophage populations, resulting in high cure rates and long-term tumor-specific protection across murine models of colon and mammary carcinoma [167].

While future therapeutics will likely combine these approaches to normalize tissue mechanics, dismantle stromal barriers, and enhance drug and immune cell penetration, realizing this potential, however, requires confronting the challenges that have limited earlier ECM-targeted approaches and pursuing the strategies now emerging to address them.

i. Challenges in ECM-Targeted Therapy

Despite the growing repertoire of ECM-targeted therapeutics, most first-generation ECM-directed agents have failed to translate into clinical benefit, and understanding why is as important as cataloguing new targets. Three major factors have emerged as recurrent barriers to successful ECM-targeted therapy. The first is the biological redundancy of ECM remodeling pathways. Broad inhibition of ECM-modifying enzymes is often insufficient, because multiple enzymatically redundant or compensatory pathways can maintain fibrosis and tissue stiffness even when a single node is blocked. This limitation is exemplified by simtuzumab’s failure to improve progression-free or overall survival in a randomized phase II trial on metastatic PDAC patients despite compelling preclinical activity (NCT01472198) [168], mirroring its earlier failure in IPF discussed above. Similarly, marimastat and related broad-spectrum MMP inhibitors, despite compelling preclinical anti-invasive activity, failed to demonstrate consistent survival benefit across phase III trials in pancreatic, gastric, and small-cell lung cancer, and in some cohorts were associated with dose-limiting musculoskeletal toxicity from off-target inhibition of physiological MMP activity (NCT00003010, NCT00003011, NCT00002911) [169]. Together, these trials illustrate that ECM-remodeling enzymes perform indispensable physiological functions in tissue homeostasis and wound repair, limiting the therapeutic window of broad-spectrum inhibition.

A second challenge is the functional heterogeneity of the tumor stroma. Some ECM-producing stromal components actively restrain tumor growth, and their depletion can accelerate disease. In PDAC, two landmark studies demonstrated this directly: genetic depletion of αSMA⁺ myofibroblasts led to more invasive, undifferentiated tumors with increased hypoxia, EMT, and cancer stem cell features, increased CD4⁺Foxp3⁺ Tregs leading to suppressed immune surveillance, and correlated with reduced survival in both mouse models and PDAC patients [170]; and genetic or pharmacological ablation of Sonic Hedgehog-driven stroma similarly produced more aggressive, poorly differentiated tumors with increased vascularity [171]. These findings reframed the field’s understanding of desmoplasia, rather than a uniformly protective barrier for the tumor, at least part of the fibrotic stroma constrains tumor progression, meaning indiscriminate stromal depletion strategies can backfire.

A third challenge is disease stage as advanced fibrotic ECM may become therapeutically irreversible. Once collagen has been extensively cross-linked by LOX/LOXL enzymes and further stabilized by LH2 -mediated hydroxylysine cross-links, resolving this architecture pharmacologically becomes substantially harder than preventing its formation [172]. This therapeutic irreversibility may partly explain why several agents effective in earlier or pre-clinical disease stages have underperformed in patients with established, advanced fibrotic tumors.

ii. Future Directions

These lessons argue for strategies that move beyond simple ECM elimination toward more precise, context-aware interventions. Rather than attempting to dismantle an already-established fibrotic matrix, one emerging alternative is to enhance the capacity of immune and effector cells to function effectively within mechanically restrictive environments. For instance, engineered lymphocytes capable of actively degrading or traversing dense stromal barriers may improve antitumor immunity without requiring systemic matrix depletion [173]. Menon et al. demonstrated this concept by reprogramming T and NK cells to express ECM-degrading enzyme MMP8, enabling these lymphocytes to locally remodel dense stromal barriers and improve intratumoral infiltration in NSG mice engrafted with SKOV3 tumors [173].

Artificial intelligence (AI)-enabled drug discovery is emerging as a complementary strategy to identify selective ECM regulators that would have likely been overlooked by conventional target-based screening. This advanced strategy is exemplified by the Traf2- and Nck-interacting kinase (TNIK) inhibitor, rentosertib (formerly INS018_055/ISM001–055), developed by Ren and colleagues, in which generative AI was used throughout the discovery pipeline to identify the therapeutic target and design the small-molecule inhibitor. The resulting compound, rentosertib, attenuated fibrosis across preclinical and clinical models like IPF and renal fibrosis by suppressing profibrotic transcriptional programs [174]. Consistent with these findings, rentosertib has been further validated in two independent phase I clinical trials, both demonstrating favorable safety, tolerability and pharmacokinetic profiles [174], and antifibrotic biomarker activity validated in a recent randomized, placebo-controlled phase IIa trial in IPF patients (NCT05938920) [175]. This end-to-end AI-driven pipeline, from target identification through molecule design to clinical validation, may offer a faster, more capital-efficient route to identifying tractable, selective ECM-pathway targets than conventional high-throughput screening. Multi-omics profiling combined with AI-based patient stratification may help resolve the heterogeneity that has undermined prior trials [148]. Rather than treating all desmoplastic tumors as a single therapeutic category, integrating transcriptomic, proteomic, and spatial data could define context- and cancer-type-specific ECM signatures, distinguishing tumors in which the matrix is predominantly tumor-restraining, as in some PDAC contexts, from those in which it is predominantly tumor-promoting, and identifying which patients are most likely to benefit from a given matrix-targeted agent before treatment begins [148]. Chakravarthy et al. analyzed transcriptomic data across thousands of patient samples spanning multiple cancer types and identified a specific TGF-β-associated ECM gene signature, driven largely by CAFs, that predicts immune cell exclusion and failure of PD-1 checkpoint blockade independent of previously-proposed biomarkers [176].

Taken together, future ECM-directed therapeutics will likely need to combine selective, context-informed target engagement, AI-accelerated discovery, and molecular patient stratification including defining context-specific ECM signatures, identifying patient subsets most likely to benefit, and integrating matrix-modulating agents with standard-of-care therapies, to normalize rather than indiscriminately eliminate tumor mechanics and stromal architecture.

8. Conclusion

ECM remodeling is a central orchestrator of tumor biology—shaping cancer cell behavior, stromal activation, immune infiltration, and therapeutic response. Rather than being fibroblast-driven alone, ECM dynamics emerge from coordinated interactions among multiple stromal, immune, and tumor cell populations. Emerging studies demonstrate that targeting ECM composition, crosslinking, and proteolysis can significantly alter tumor progression, reduce metastatic niche formation, and enhance therapy penetration as well as response. Effective ECM-directed therapies must integrate mechanical, biochemical, and immunological remodeling rather than relying solely on fibroblast-directed strategies. Understanding the complexities of ECM regulation will enable development of therapies that inhibit tumor growth by fundamentally reshaping the microenvironment to support durable clinical responses.