Collagen remodeling-mediated signaling pathways and their impact on tumor therapy

Abstract

In addition to their traditional roles in maintaining tissue morphology and organ development, emerging evidence suggests that collagen (COL) remodeling—referring to dynamic changes in the quantity, stiffness, arrangements, cleavage states, and homo-/hetero-trimerization of COLs—serves as a key signaling mechanism that governs tumor growth and metastasis. COL receptors act as switches, linking various forms of COL remodeling to different cell types during cancer progression, including cancer cells, immune cells, and cancer-associated fibroblasts. In this review, we summarize recent findings on the signaling pathways mediated by COL arrangement, cleavage, and trimerization states (both homo- and hetero-), as well as the roles of the primary COL receptors—integrin, DDR1/2, LAIR-1/2, MRC2, and GPVI—in cancer progression. We also discuss the latest therapeutic strategies targeting COL fragments, cancer-associated fibroblasts, and COL receptors, including integrins, DDR1/2, and LAIR1/2. Understanding the pathways modulated by COL remodeling and COL receptors in various pathological contexts will pave the way for developing new precision therapies.

Collagen (COL) is the most abundant protein in mammals, accounting for 25%-35% of the protein content (1). It is an important component of the extracellular matrix (ECM), which maintains the morphology and function of tissues and organs such as skin, cartilage, tendons, ligaments, and internal organs (2). To date, 28 COL types have been identified and divided into several groups (3) (Fig. 1), encompassing fibrillar COLs, fibril-associated COLs with interrupted triple helices, short chain COLs, basement membrane COLs, multiple triple helix domains with interruptions (Multiplexin), membrane-associated COLs with interrupted triple helices, microfibril-forming COLs, anchoring fibrils, and other COL types XXVI and XXVIII (Fig. 1, Table 1) (4). Fibrillar COLs (types I, II, III, V, XI, XXIV, and XXVII) contain one major triple-helical domain and provide three-dimensional frameworks for tissues and organs. These networks confer mechanical strength as well as signaling and organizing functions through binding to cellular receptors and other components of the ECM (4, 5, 6, 7, 8, 9, 10). Fibril-associated COLs with interrupted triple helix (types IX, XII, XIV, XVI, XIX, XX, XXI, and XXII) is a subclass of COLs that contain two or more short triple-helical domains interrupted by noncollagenous domains with an N-terminal noncollagenous (NC) domain facing toward the interfibrillar space (5, 11, 12, 13, 14, 15, 16). These COLs are typically found alongside fibrillar COLs and contribute to the organization and stabilization of the ECM. Short chain COLs (types VIII and X) are a subgroup of nonfibrillar COLs, containing three distinct domains: a short NC amino (N)-terminal region, a collagenous domain, and a noncollagenous carboxyl (C)-terminal domain. Type VIII COL (COL-VIII) is highly expressed developmentally, then dramatically upregulated in aged and diseased vessels characterized by arterial stiffening (17, 18). COL-X is a homotrimer, deposited exclusively at sites of new bone formation (19, 20). Basement membranes are a specialized form of ECM that underlies epithelial, endothelial, muscle, fat, and nervous tissues. Basement membrane COLs (type IV) which contain a short N-terminal 7S domain representing critical nodes, a long central triple-helical domain, and a C-terminal NC1 domain are a key component required for basement membranes’ function, providing mechanical support for tissues, serving as a polyvalent ligand for cell adhesion receptors, and as a scaffold for other proteins (21, 22). Multiplexin COLs (types XV and XVIII) contain multiple triple-helix domains interrupted by NC sequences and have been shown to play an important structural role in maintaining the integrity of the ECM but also have significance in many physiological and pathological processes. For example, mutations in the COL18A1 gene are associated with Knobloch syndrome featured by retinal abnormalities (23, 24, 25). Membrane-associated COLs with interrupted triple helix (types XIII, XVII, XXIII, and XXV) are type II transmembrane proteins that consist of a short cytosolic domain, a transmembrane domain, and a large extracellular ectodomain that can be shed by furin convertases. They can contribute to the maturation of neuromuscular junctions in the case of COL-XIII, the development of the neuromuscular junction, and the survival of motor neurons in the case of COL-XXV. COL-XXIII is elevated in human prostate and head and neck cancer progression and COL-XXV may regulate the progression of Alzheimer’s disease (26, 27, 28, 29, 30, 31). Microfibril-forming COLs (type VI) are characterized by a short triple-helical domain and are predominantly composed of von Willebrand factor type A domains. These heteromeric, network-forming COLs are found in nearly all connective tissues, including cartilage, bone, tendons, muscles, and the cornea. Within these tissues, they contribute to the formation of abundant and structurally distinct microfibrils, which are organized into various suprastructural assemblies (32, 33). Anchoring fibrils (type VII) feature a central collagenous domain flanked by an N-terminal cysteine knot. The N- and C-terminal regions form two NC domains, NC1 and NC2. NC2 is cleaved off upon fibril formation. These fibrils function as specialized attachment complexes located at the epithelium–mesenchyme interface in various tissues, playing a crucial role in maintaining the structural integrity and stability of the skin (34). Although distinct types of COLs may consist of different α chains (Table 1), all COLs are composed of repeating peptide triplets of glycine-X-Y (X and Y are frequently proline and hydroxyproline, respectively) (8). COL biosynthesis is a complex process that takes place intracellularly and extracellularly. COL-I synthesis is the best studied. Three procollagen peptide chains can be coiled around each other to form a stable triple helix structure through interchain hydrogen bonds in the endoplasmic reticulum. Individual proα chains undergo numerous posttranslational modifications including hydroxylation of proline and lysine residues, glycosylation of hydroxylysine residues, and sulfation of tyrosine residues. These modifications are halted by the formation of the triple helix (35). After the three helical strands of COL are secreted into the extracellular space, the N-terminal and C-terminal peptide chains are hydrolyzed by specific procollagen proteinases, including bone morphogenetic protein 1, members of the short for a disintegrin and metalloproteinase with thrombospondin motifs protease family, and the more recently discovered meprins (36) (Fig. 2). This process results in the formation of mature COL trimers. These trimers can then polymerize into COL fibers under neutral pH conditions due to the mutual attraction between different charged regions of the molecules (Fig. 2).

Figure 1.

Classification of collagen types based on their structure and organization. Eight groups of COLs have been described: fibrillar COLs, fibril-associated COLs with interrupted triple helices (FACITs), short chain COLs, basement membrane COLs, multiple triple helix domains with interruptions (Multiplexin) COLs, membrane-associated COLs with interrupted triple helices (MACIT), microfibril-forming COLs, and anchoring fibrils.

Table 1.

Collagen family and functions

Groups	Type	α chain	Tissue distribution	Functions
Fibrillar	I	[α1 (I)]2α2 (I)	Skin, bones, tendons, and ligaments	Providing strength and structural support to tissues (4)
	II	[α1 (II)]3	Cartilage	Providing elasticity and resistance to pressure (7)
	III	[α1 (III)]3	Skin, blood vessels, and internal organs	Offering flexibility and structural support (8)
	V	[α1 (V)]2 α2 (V), [α1 (V)α2 (V)α2 (V)], [α1 (V)]3	Skin, bones, and placenta	Participating in fiber formation and structural stability of tissues (9)
	XI	[α1 (XI)α2 (XI)α2 (XI)]	Cartilage and vitreous humor	Formation and maintenance of cartilage (5)
	XXIV	[α1 (XXIV)]3	Cornea and bones	Formation and maintenance of the skeletal system (10)
	XXVII	[α1 (XXVII)]3	Cartilage	Calcification of cartilage (5)
FACIT	IX	[α1 (IX)α2 (IX)α3 (IX)]	Cartilage, vitreous humor, and cornea	Formation and maintenance of cartilage (5)
	XII	[α1 (XII)]3	Bones, ligaments, and tendon	Protecting bone and muscle integrity (13)
	XIV	[α1 (XIV)]3	Skin, tendon, and cartilage	Involving in the regulation of collagen fibrillogenesis (14)
	XVI	[α1 (XVI)]3	Skin, cartilage, and intestine	Modulating integrity and stability of the extracellular matrix (15)
	XIX	α1 (XIX)	Blood vessels, central neurons, and internal organs	Formation of hippocampal synapses, modifying ECM superstructure (25)
		[α1 (XIX)]2
		[α1 (XIX)]3
		[α1 (XIX)]4
		[α1 (XIX)]5
		[α1 (XIX)]6
	XX	α1 (XX)	Skin, tendons, and cartilage	Formation and maintenance of skeletal and cartilage (5)
	XXI	[α1 (XXI)]3	Blood vessel	Maintaining tissue junction and maintaining vascular stability (8)
	XXII	α1 (XXII)	Tissue junction, blood vessel, and cartilage	Maintaining tissue junction, maintaining vascular stability (11)
Short-chain	VIII	[α1 (VIII)]2α2 (VIII)	Blood vessels and the cornea	Formation of the Descemet’s membrane (17)
	X	[α1 (X)]3	Cartilage	Participating in endochondral ossification and establishment of a hematopoietic niche (19)
Basement membrane	IV	[α1 (IV)]2α2 (IV)	Basement membranes	Structural support, barrier function, regulation of cell behavior, filtration, angiogenesis (21)
		[α1 (IV)]2α3 (IV)
		[α1 (IV)]2α4 (IV)
		[α1 (IV)]2α5 (IV)
		[α1 (IV)]2α6 (IV)
Microfibril-forming	VI	[α1 (VI)α2 (VI)α2 (VI)], [α1 (VI)α2 (VI)α3 (VI)]	Uterus, skin, cartilage, placenta, lung, and blood vessel	organizing the three-dimensional tissue architecture of skeletal muscles, tendons, bone and cartilage, interacting with ECM and other collagen (32)
Multiplexin	XV	α1 (XV)	Heart, placenta, kidney, and ovary	Modeling of the ECM, antiangiogenic (23, 24)
	XVIII	α1 (XVIII)	Kidney, liver, and lungs
MACIT	XIII	α1 (XIII)	Lung, skin, bones, hair follicles, neurons, endomysium, intestine, chondrocytes, lungs, and liver	Modeling of the ECM, formation, and maintenance of the neuromuscular junction (30, 31)
	XVII	[α1 (XVII)]3	Skin	Mediating the interactions of stem cells with surrounding cells and the matrix, and regulating skin homeostasis, aging, and wound repair (28)
	XXIII	α1 (XXIII)	Heart, retina, epidermis, tongue, gut, brain, kidney, and cornea	Modeling of the ECM and associated with tumor progression (28, 31)
	XXV	α1 (XXV)	Muscle, brain, heart, testis, and eyes	Modeling of the ECM, intramuscular innervation (28, 29)
Anchoring fibrils	VII	[α1 (VII)]3	Skin	Connecting the basement membrane to the dermis (34)

Figure 2.

The processing of COLs. All COLs are composed of repeating peptide triplets of glycine-X-Y (X and Y are frequently proline and hydroxyproline, respectively) and each COL molecule consists of three α chains. After procollagens are secreted extracellularly, the N- and C-terminal nonhelical procollagen segments are removed to form mature COL trimers which can be then polymerized into COL fibers. COLs can undergo extensive remodeling during tumor progression, including dynamic changes in amounts, arrangement (linear/wavy COL), cleavage states (cleaved COL/intact COL), and homogenous/heterotrimers.

COLs are known to play a vital physiological role in providing structural support for tissue mechanical properties, organization, and shape. Mutations in COL genes can lead to a wide range of connective tissue disorders, such as osteogenesis imperfecta and alport syndrome, and abnormal deposition of the COL matrix contributes to tumor formation. Irregular COL expression is associated with the prognosis and progression of various tumors, including gastric, liver, bladder, ovarian, melanoma, colon, lung, and pancreatic cancers, which have been well described and discussed in many excellent reviews (37). Recent studies indicate that, in addition to serving as structural components, COLs play crucial roles as metabolic regulators and immunoregulatory signals. This review will focus on the signaling pathways mediated by COL cleavage states, arrangement, and trimer heterogeneity, particularly concerning tumor bioenergetics, growth, progression, and immunity. We will also discuss the roles of key COL receptors in these pathways and explore therapeutic strategies targeting these receptors. Understanding how variations in COL networks affect cancer growth and spread could potentially result in more personalized treatment approaches targeting each patient's distinct needs while minimizing side effects associated with traditional therapeutic methods.

COL remodeling-mediated signaling pathways and cancer progression

During tumor progression, COLs may undergo extensive remodeling including dynamic changes in amounts, arrangements, cleavage states, and homo-/hetero-trimers (Fig. 2). COLs, primarily secreted by cancer-associated fibroblasts (CAFs) in response to TGF-β, are upregulated in multiple tumors (38). Increased COL deposition leads to elevated ECM stiffness, inducing morphological changes and metabolic reprogramming in cancer cells. Additionally, it creates a physical barrier within the tumor microenvironment (TME), hindering immune cell infiltration and further promoting CAF activation. This positive feedback loop results in further stiffening of the COLs (39, 40, 41, 42, 43). Several reviews have discussed the relevance of ECM stiffness in cancer progression and therapy (44, 45, 46). However, deletion of COL-I, the most abundant protein in both the ECM and the human body, in α-smooth muscle actin-positive myofibroblasts has been shown to accelerate the development of pancreatic ductal adenocarcinoma (PDAC). Additionally, denatured COL induced by thermal ablation has been found to inhibit the migration and growth of neuroblastoma tumor spheres. These findings suggest that the dynamics of COL architecture play crucial roles in tumor growth, beyond mere changes in its quantity (47). Recent findings on COL arrangements, cleavage states, and homo-/hetero-trimers provide new insights into the sophisticated roles of COLs (Fig. 3).

Figure 3.

Schematic diagram of different types of COL remodeling-mediated signaling pathways.A, (i) a high degree of wavy COL-III fibers maintains tumor dormancy via DDR1-mediated STAT1 signaling, while linear COL-III plays a reversed effect (50, 51). (ii) the extracellular domain (ECD) of DDR1 which can be cleaved off by sheddase promotes COL fiber alignment to instigate immune exclusion possibly via LAIR1-induced CD8⁺ T cell exhaustion (52, 53, 54, 55). B, (i) the cleaved COL-I promotes mitochondria biogenesis and macropinocytosis via the DDR1-NF-κB-p62-NRF2 signaling pathway, leading to a poor patient outcome. In contrast, the intact COL-I has the opposite effect on mitochondria biogenesis, macropinocytosis, and patient outcome via inhibiting DDR1 and its downstream effectors (58). (ii) cleaved COL-I may induce CXCL5 production to promote tumor microenvironment (TME) inflammation via DDR1-PKCθ-SYK-NF-κB signaling (60). C, COL-I homotrimers, resulting from epigenetic suppression of the Col1a2 gene, significantly promote pancreatic ductal adenocarcinoma (PDAC) tumor growth (63). These homotrimers bind to α3β1 integrin on cancer cells, activating oncogenic signaling pathways, including FAK-PI3K-AKT and FAK-MAPK-ERK. COL-I homotrimers are associated with a microbiota enriched in anaerobic Bacteroidales in immunosuppressed tumors.

COL arrangement-mediated signaling

COL arrangement refers to the specific organization and spatial configuration of COL fibers within ECMs or tissues. COL fibers can be organized in different orientations—parallel, woven, or randomly arranged—depending on the mechanical demands of the tissue. For instance, tendons have a parallel arrangement for tensile strength, while skin has a more woven structure for flexibility. Recent studies indicate that the orientation and alignment of COLs play a significant role in tumor progression. Wavy COLs are curved COLs with a low degree of orientation and linear organization, found to correlate with benign tumors (48) and inhibit the growth of tumor cells in head and neck squamous cell carcinoma (HNSCC) (49). Dormant HNSCC cells produce an ECM enriched in COL-III, characterized by a significantly lower degree of linear organization than COL-I fibers. These wavy COL-III fibers activate the COL receptor discoidin domain receptor 1 (DDR1) on the surface of tumor cells, enhancing the transcriptional activity of signal transducer and activator of transcription 3 (STAT3) through an unknown mechanism, which in turn promotes further production of COL-III. This positive feedback loop helps maintain the dormancy of tumor cells (Fig. 3A). However, the reasons why dormant tumor cells initially preferentially produce wavy COL-III, as well as how these wavy COLs sustain tumor dormancy through metabolic and immune adaptations require further investigation. In contrast, linear COL fibers exert the opposite effects (50, 51). Oriented or linear fibers significantly inhibit antitumor immune infiltration and facilitate intravasation during the metastasis of triple-negative breast cancer (TNBC), whereas the wavy COLs play a reversed effect. In response to COL treatment, the extracellular domain (ECD) of DDR1 can be cleaved by sheddase (52). It has been shown that DDR1 ECD in the TME binds to COL, reinforcing aligned COL fibers and obstructing immune infiltration (Fig. 3A). However, the mechanisms by which DDR1 exerts opposing effects on tumor growth in response to wavy versus linear COLs remain unclear. It is crucial to investigate whether wavy and linear COLs have distinct effects on DDR1 shedding.

The leukocyte-associated immunoglobulin-like receptor 1 (LAIR-1), which is highly expressed by intratumoral myeloid cells, inhibits the activation, proliferation, and cytokine production of T cells in response to increased COL deposition (53). Blockade of LAIR-1, in combination with programmed death-ligand 1 (PD-L1) inhibition, significantly reduces tumor size in several cancers (53, 54, 55). Based on these findings, it is worth investigating whether aligned COLs promote immune exclusion by activating LAIR-1 in myeloid cells. Taken together, these findings suggest that linear COLs support tumor growth, while wavy COLs oppose it. More details on the COL receptors, DDR1 and LAIR-1 will be discussed in Section 3 COL Receptors.

COL cleavage states-modulated signaling

Retrospective clinical studies have revealed that patients with PDAC whose tumors exhibit a fibrogenic but inert stroma—characterized by extensive ECM deposition, low expression of the myofibroblast marker α-SMA, and low matrix metalloproteinase (MMP) activity—experience improved progression-free survival compared to those with tumors characterized by a fibrolytic stroma. The latter stroma is marked by a low content of COL fibers, high expression of α-SMA, and elevated MMP activity (56, 57). The mechanism by which stromal state influences clinical outcomes was previously unknown until a recent study shed light on this issue. Extracellular COLs can be cleaved by MMPs, generating a variety of COL fragments. Intact COL-I (iCOL-I) and MMP-cleaved COL-I (cCOL-I) have been shown to exert opposing effects on PDAC tumor growth and metastasis (58). Specifically, cCOL-I engages DDR1 to activate an NF-κB-p62-NRF2 signaling cascade, stimulating macropinocytosis (a form of fluid-phase endocytosis in which cells use plasma membrane ruffles to internalize soluble material into primary endocytic vesicles), mitochondrial biogenesis, bioenergetics, tumor growth, and metastasis. In contrast, iCOL-I triggers DDR1 downregulation, leading to opposite effects on these processes (Fig. 3B). Patients with PDAC whose tumors are enriched with iCOL-I—and thus exhibit low DDR1 and NRF2 expression—show significantly improved median survival compared to those with tumors enriched in cCOL-I and high MMP expression, where malignant cells express elevated levels of DDR1 and NRF2. These findings support the development of therapeutics targeting the DDR1-NF-κB-NRF2 metabolism-stimulating pathway. Similar observations have been made in hepatocellular carcinoma (HCC). ECM containing degradable COL-I, but not ECM with iCOL-I, activates DDR1 and downstream tumor-promoting pathways such as AKT in HCC tumor cells; these effects can be blocked by the DDR1 inhibitor 7rh (59). Unlike in TNBC, no change in COL fiber alignment is observed between tumors from COL-I-cleavable and noncleavable pancreata. However, patients with high levels of cCOL-I in their tumors correlate with a more immune-suppressive TME, characterized by high inflammatory markers. cCOL-I-DDR1 signaling may enhance TME inflammation by inducing cancer cell expression of CXCL5 via a PKCθ-SYK-NF-κB cascade, leading to the recruitment of tumor-associated neutrophils and the formation of neutrophil extracellular traps (60) (Fig. 3B). Further investigation into the role of COL cleavage states in immune cell composition could pave the way for novel immunotherapies against PDAC. Understanding how DDR1 interacts with different remodeling types of COLs in various cancers requires the cooperation of biologists with protein structuralists and will facilitate precision medicine.

Homo-/hetero-COL trimers-mediated signaling

COL trimers are the fundamental building blocks of COL fibers, consisting of three individual COL polypeptide chains. Depending on the COL subtype and the type of COL-producing cells, these chains can be identical (homotrimers) or different (heterotrimers). For example, COL-II consists of homotrimers of the α1 chain, while COL-VI is mainly composed of one α1 chain and two α2 chains (Table 1). COL-I typically occurs as a heterotrimer with two α1(I) chains and one α2(I) chain (Table 1), although it can form homotrimers in adult skin (61), embryonic tissues (62), and pancreatic cancer cells (63), where it consists of three α1(I) chains. These homotrimers are highly resistant to MMP cleavage, which prevents the activation of DDR1 signaling. In contrast, COL-I α2 homotrimers have never been observed in natural settings, and the α2 chain is more unfolded than the α1 chain near the MMP cleavage site (64, 65). Recent studies have shown that COL-I homotrimers, resulting from epigenetic suppression of the Col1a2 gene, significantly promote PDAC tumor growth. These homotrimers bind to α3β1 integrin on cancer cells, activating oncogenic signaling pathways, including FAK-PI3K-AKT and FAK-MAPK-ERK (Fig. 3C). This effect is associated with a microbiota enriched in anaerobic Bacteroidales in hypoxic and immunosuppressed tumors. Deletion of COL-I homotrimers has been shown to increase overall survival in PDAC-bearing mice, correlating with a tumor microbiota enriched for microaerophilic Campylobacterales. Furthermore, deletion of COL-I homotrimers enhances CD8⁺ T cell infiltration and improves responsiveness to anti-PD-1 immunotherapy. However, COL-I homotrimer deletion does not affect the total COL-I content in the TME, raising the question of how even a small amount of cancer cell–derived COL-I homotrimers can influence the tumor microbiome and the immunobiology of tumor islets surrounded by large amounts of TME-derived COL-I. An earlier study demonstrated that COL-I homotrimers, derived from lathyritic chick embryo tendons and calvaria, exhibit greater solubility than COL-I heterotrimers (66). Therefore, cancer cell–derived COL-I homotrimers may have a similar tumor-promoting effect as soluble cCOL-I derived from the TME, influencing both the tumor microbiome and immune responses. These emerging studies expand our understanding of the diverse roles of COLs, from their structural functions in organ morphology and development to their roles as signaling molecules that regulate tumor growth, metabolism, immunity, and the intestinal microbiota, through various COL receptors and their downstream effectors (49, 67, 68, 69).

COL receptors

According to their structural features, binding preference, and operational mechanisms, COL receptors fall into two primary groups–integrin COL receptors and nonintegrin COL receptors. The human genome encodes 18 distinct α subunit genes and eight unique β subunit genes, leading to 24 potential α-β pairings observed at the protein level (70). Here we will discuss the functions of COL receptors in cancer development associated with four integrin COL receptors—α1β1, α2β1, α10β1, and α11β1, along with six nonintegrin COL receptors—DDR1, DDR2, MRC2, LAIR-1, and GPVI according to the latest progress (Table 2). However, it is still unclear whether all the functions are related to COL remodeling-mediated signaling pathways.

Table 2.

COL receptor’s distribution and function

Receptor	Cell types	Functions
Integrin α1β1	Endothelial cells, smooth muscle cells, myofibroblasts, monocytes, and T cells (77)	Immunosuppressive response via COL-VI-α1β1 interaction (40)
		Chemoresistance via COL-XI-α1β1-Src-PI3K-Akt-NF-κB-IAP axis (80)
		Invasion via COL-V-α1β1-MAPK signaling (81)
		Angiogenesis via COL-I/IV-α1β1 axis (82)
Integrin α2β1	Epithelial cells, fibroblasts, T cells, myeloid cells, megakaryocytes, platelets, keratinocytes, and endothelial cells (77)	Proliferation via COL-IV-α2β1-FAK-Src-PI3K-Akt axis (97).
		Migration via COL-I-α2β1-FAK-p38-Akt axis (98)
Integrin α10β1	Mesenchymal stem cells, osteocytes, fibroblasts, chondrogenic cells lining the endosteum and periosteum (77)	Proliferation via COL-I/II-α10β1-ERK signaling (111); or COL-II-α10β1-Rac-PAK axis (114); or COL-II-α10β1-Akt-mTOR axis (114)
Integrin α11β1	Mesenchymal stem cells, osteocytes, and fibroblasts (77)	Proliferation via COL-I-α11β1 interaction (122)
DDR1	Epithelial cells (127, 128)	Metabolism, tumor growth, and metastasis via cCOL-I-DDR1-NF-κB-p62-NRF2 axis (59, 143)
		Sustain stemness via COL-I-DDR1-YAP axis (134); or COL-I-DDR1-14-3-3-Beclin1-Akt-mTOR axis (137, 138)
		Immunosuppressive response via cCOL-I-DDR1 interaction (141, 142)
DDR2	Fibroblasts, smooth muscle cells, chondrocytes (127, 128), and tumor cells (146)	Glycolysis via fibrillar COL-DDR2-SNAI1 axis (143)
		Cell quiescent via COL-I-DDR2-STAT-p27 axis (146)
MRC2	Macrophages and fibroblasts (149)	Immunosuppressive response and maintaining myofibroblast (151)
LAIR-1/LAIR-2	Immune cells and tumor cells (157, 158)	Induction of CD8+ T cell exhaustion possibly via cCOL-LAIR-1-SHP-1/SHP-2 axis (159, 161, 162)
		Sustaining stem cell via COL-LAIR-1-SHP1-CAMK1-CREB axis (159, 160)
GPVI	Megakaryocytes and platelets (167)	Metastasis via galectin-3-GPVI interaction (171, 172)
		Platelet activation (174)

Integrin

The integrin family is predominantly expressed in vertebrates but exhibits lower expression levels in other animal species, mirroring the adaptive demands of living systems regarding cell–matrix interactions. Subfamily members that primarily interact with COLs share common β1 subunits (34). The β2 integrin is a leukocyte-specific integrin found exclusively on the surfaces of leukocytes, including neutrophils, lymphocytes, and monocytes. Its key ligands include complement components (iC3b) and intercellular adhesion molecules (71, 72), though it has limited involvement as a COL receptor. Matricryptins, products of enzymatic cleavages from COLs yielding bioactive peptides, utilize diverse integrin subfamilies such as β3, β5, etc., as their respective receptors (73). Integrin β4 is found to predominantly bind to laminin (74). Although immunoprecipitation-based studies suggest potential interactions between integrin β4 and COL-I, contributing to gastric cancer progression and altered glucose homeostasis through activation of the FAK/SOX2/HIF-1α cascade (75), further research is needed to fully understand the molecular mechanisms and confirm its role as a COL receptor. Four key members of the β1 integrin subfamily—integrin α1β1, α2β1, α10β1, and α11β1—are primary COL receptors, distinguished by large αI ECDs that specifically recognize the GFOGER motif in COLs (76).

Integrin α1β1

Integrin α1β1 (also known as very late antigen-1) is commonly expressed in mesenchymal cells like endothelium, smooth muscle cells, myofibroblasts, and monocytes (77, 78). Its major ligands are fibrous-type COLs (i.e., COL-I and III). Integrin α1β1 is found to either promote or inhibit cancer development through various pathways after binding with different types of COLs (79). In prostate cancer, loss of integrin α1β1, which binds to COL-VI but not COL-I, impairs traction between T cells and matrix, limiting T cell movement. This results in the accumulation of CD4⁺ T cells at the tumor site, triggering an immunosuppressive response (40). In ovarian cancer, integrin α1β1 combined with DDR2 promotes the expression of apoptosis inhibitory proteins through Src-PI3K-Akt-NF-κB-IAP cascade in response to COL-XI, leading to resistance to cisplatin, a chemotherapy drug used to treat cancer (80) (Table 2). Additionally, COL-V increases breast cancer cell invasion via α1β1 integrins and MAPK signaling, while also increasing the alignment of COL-I, which has been associated with increased invasion (81) (Table 2). Moreover, in endothelial cells, integrin α1β1 is involved in physiological and pathological angiogenesis (82). For example, disturbing integrin α1β1 and COL-I/COL-IV interaction using lebetin 2—a 4 KD peptide derived from macrovipera lebetina transmediterranea venom and previously displayed a potent anti-platelet activity—exhibits a potent anti-angiogenic effect and prevents the adhesion and migration of pheochromocytoma cells and CHO-α1 cells, indicating lebetin 2 as a potential therapeutic agent against cancer by targeting the integrin receptor function (83).

Tumors are often associated with an increase in stiffness, as with breast and pancreatic cancer (∼4 kPa), raising the possibility that chronic disruption to tensional homeostasis may act as a precursor to malignant transformation. The stiff collagenous stroma has been found to elevate mammary epithelial stem-progenitor cell frequency and promote tumor initiation through the tension-dependent enhancement of β1-integrin mechanosignaling that drives EGFR-dependent ERK activity to potentiate progesterone receptor–induced receptor activator of NF-κB ligand (RANKL)-RANK activity which is known to control the expansion of the stem cell (84, 85, 86). These findings suggest targeting either EGFR, ERK, or RANK signaling might have therapeutic implications for patients with highly fibrotic breast cancers.

Integrin α2β1

Different from integrin α1β1, integrin α2β1 is widely expressed in epithelial cells (77, 87, 88, 89) (Table 2). Integrin α2β1 exhibits a high affinity for COL-I (90, 91). As the mechanical properties and distribution patterns of COL-I have a substantial impact on tumorigenesis, integrin α2β1 has received considerable interest in tumor research. It is thought to directly or indirectly modulate key signal transduction pathways within malignant cells, promoting their primary settlement and secondary dissemination (3, 92, 93, 94, 95). At different stages of cancer development, COL-IV is found to interact with distinct integrins to modulate tumor growth. Bioinformatics analysis has identified an interactive regulatory network in which COL4A1/2 directly binds to integrin α2β1, initiating a complex signaling cascade that accelerates the cell cycle and promotes HCC tumorigenesis (96). Among the pathways involved, the PI3K–Akt pathway is notably enriched in cooperative mutations and correlation analyses. Additionally, Wang et al. have demonstrated that COL4A1 promotes HCC cell growth and metastasis by activating the FAK-Src-PI3K-Akt pathway (97) (Table 2). These findings imply that COL-IV, a marker of liver fibrosis, contributes to HCC progression through the PI3K–Akt pathway, mediated by its binding to integrin α2β1. In nonsmall cell lung cancer, CAFs exhibit preferential binding to COL-XI through integrin α2β1, resulting in the inhibition of CAF-dependent COL-I matrix remodeling and cellular migration. This competition between COL-XI and COL-I for integrin α2β1 binding downregulates downstream FAK/p38/Akt signaling, ultimately suppressing CAF-mediated lung tumor cell invasion (98) (Table 2). However, COL-XI also exerts oncogenic functions when bound to DDR2 and integrin α1β1, triggering Src/Akt signaling and increasing HSP27 levels, conferring chemoresistance in ovarian cancer cells (99) (Table 2). These findings suggest that understanding how distinct COL–receptor interactions contribute to specific stages of malignancy and individual forms of cancer will facilitate the design of targeted therapies.

Additionally, COL-integrin signaling is involved in fibrosis, a common complication of inflammation. Recent evidence suggests that reducing the communication between COL–integrin interactions may help prevent fibrotic diseases (100, 101). Studies have shown that activation of the integrin α2β1-FAK-PI3K-AKT axis promotes increased COL-I expression during liver fibrogenesis (102). In contrast, the integrin α2β1-COL-I interaction has dual effects in pulmonary fibrosis (103). These results highlight the essential role of integrin α2β1 in mediating COL deposition during fibrotic conditions. Furthermore, integrin α2β1's involvement in fibroproliferative diseases associated with precarcinogenesis calls for further investigation into the underlying mechanisms driving tumorigenesis triggered by fibrosis (96, 100, 104, 105).

Integrin α10β1 and integrin α11β1

While integrins α10β1 and α11β1 are both found on mesenchymal stem cells, osteocytes, and fibroblasts, they appear to perform separate functions (77, 106). Considering that integrin α10β1 has a unique charge-dependent constraint mechanism, it may serve as an essential COL receptor (107). Given the prominence of COLs in bones (108), previous investigations primarily concentrated on integrin α10β1's functions in the skeletal system (109, 110). However, recent efforts have sought to uncover its potential roles in cancer biology (Table 2).

The interaction of a cryptic regulatory region of COL termed the HU177 epitope, with integrin α10β1 expressed on mesenchymal cells, enhances the stromal cell–mediated release of protumoral cytokines, such as interleukin-6. This interaction leads to upregulated ERK signaling and promotes ovarian carcinoma progression (111). Integrin α10β1 is also able to interact with COL-I and COL-II to promote the adhesion of glioblastoma cancer cells, possibly contributing to elevating their proliferative capacity (112). The development of antibodies targeting integrin α10β1 based on these interactions holds promise for therapeutic applications (113). In highly aggressive myxofibrosarcomas, COL-II elicits Rac/PAK and Akt/mTOR pathways via engagement with integrin α10β1 overexpressed on tumor cells, which appears independent of cell adhesion modulation of COL-II since unimpeded matrix attachment can be achieved by interaction of COL-II with other integrin partners (α1β1 or α2β1) even loss of α10β1 (114). These findings suggest that integrin α10β1 acts as a signaling molecule rather than a traditional adhesion scaffold, highlighting its pivotal role in cancer progression.

Integrin α11β1 plays an important role in the transformation of fibroblasts into myofibroblasts (115). It is highly expressed in myofibroblast within lung adenocarcinomas, PDAC, and breast cancer (116). Given the potential oncogenic nature of myofibroblasts (117, 118), integrin α11β1 likely exerts an important role in tumor progression. Actually, in breast and bladder cancer, a specific subpopulation of CAFs, identified based on the high levels of integrin α11β1 expression, has been linked to increased proliferation (119), metastatic dissemination (120), and lymphovascular invasion (121). Although the specific type of COL that serves as a ligand for integrin is not well-established in most studies, COL-I is believed to facilitate the migration of CAFs toward areas that promote tumor spread by binding to integrin α11β1 (122). While integrin α11β1 has primarily served as a diagnostic tool for discerning CAF diversity (116, 123), further investigations into its intricate molecular networks of COLs may aid our comprehension of neoplastic processes.

DDR1 & DDR2

The DDR family consists of two members—DDR1 and DDR2—both possessing single transmembrane-spanning structures and intracellular tyrosine kinase activities. Amongst DDR1 variants (i.e., DDR1a, DDR1b, DDR1c, DDR1d, and DDR1e), only DDR1a-c possess catalytically active tyrosine kinase domains. DDR2 exhibits just one isoform. The overall structural organization encompasses four core elements: extracellular regions comprising discoidin homology and discoidin homology–like modules, juxtamembrane segments featuring external and internal components, and a C-terminal segment harboring a tyrosine kinase domain (124). While DDR family members play critical roles across multiple signal transduction cascades governing diverse malignancies such as those affecting the lungs, breasts, pancreas, liver, or colon, their expression patterns differ significantly between tissues and disease states (125, 126). Specifically, DDR1 predominantly appears in epithelia where upregulation associates with advanced stages of pathogenesis in PDAC, HCC, colorectal cancer, and prostate cancer, whereas DDR2 expression predominates among mesenchymal cell lineages including connective tissue, muscles, and bones (58, 127, 128, 129, 130, 131).

In PDAC, DDR1 has been found to play an important role in collagenolysis-mediated tumor metabolism. cCOL-I activates DDR1-NF-κB-p62-NRF2 cascade promoting PDAC tumor growth and metastasis to the liver, as well as a poor outcome of patients characterized by enhanced mitochondrial activity and increased macropinocytic uptake capacity. In contrast, iCOL-I exerts a reversed effect on them (58). However, how DDR1 responds to distinct cleavage states of COL-I to perform opposite roles in cancer development is still unknown. DDR1 is also known to affect cancer stem cells. COL-I–activated DDR1 cooperatively interacts with other receptors such as CD44 to preserve the stemness characteristic of HCC tumor cells by inhibiting Hippo signaling, which leads to exaggerated YAP activation (130) (Table 2). The YAP signaling pathway has been shown to promote dormancy in human colon cancer stem cells through the interaction of COL17A1 with an unidentified COL receptor (132). Activated DDR1 also sustains the stemness of glioblastoma by forming a 14-3-3-Beclin-1-AKT1 protein complex, which leads to increased mTOR activity and decreased autophagy, a process also referred to as a self-eating pathway, another lysosome-dependent nutrient-scavenging mechanism (133, 134) (Table 2). Inhibition of autophagy has been shown to upregulate macropinocytosis in PDAC by accumulating the autophagy chaperone SQSTM1/p62, which binds to KEAP1 and induces the activation of the transcription factor NRF2 (135, 136). This activation has recently been linked to the cCOL-I–DDR1 axis (58). These findings position DDR1 as a crucial nutrient sensor, regulating the interplay between intracellular energy states and the external microenvironment and the crosstalk between two key nutrient-scavenging pathways: autophagy and macropinocytosis. Investigating whether cCOL-I–activated DDR1 maintains stemness in PDAC stem cells by upregulating mitochondrial biogenesis and macropinocytosis presents a valuable direction for future research. In colon cancer, DDR1 has also been identified as a key component of the immune exclusion signature within the TME (137), and it plays a role in immune cell infiltration (138) and tumor cell metastasis (139). Recent studies have also underscored the significant role of DDR1-mediated COL barriers in conferring drug resistance to colon cancer (140). Given the multifaceted contributions of YAP signaling in intestinal research, further exploration of the potential interactions between COLs-DDR1 and the YAP signaling pathway, as well as their associated mechanisms in tumor immunity, is warranted. It is also important to understand how DDR1, a rather weak RTK, exerts such profound effects on PDAC metabolism, although the answer may be somewhat trivial and related to the very high concentration of cCOL I in the TME, far surpassing the concentration of any growth factor by many orders of magnitude.

In addition to functioning as an essential component in multiple oncogenic signaling networks triggered by engagement with distinct COLs, DDR1 could also mediate inflammation—cancer transformation governing lineage commitment decisions taken by neoplastic progenitors. Myofibroblastic hepatic stellate cells, a subset of liver-resident fibrogenic cells, augment tumoral rigidity by depositing COL-I. This facilitates the transdifferentiation of premalignant hepatocytes toward HCC tumor cells through elevated DDR1 and TAZ expression levels within resulting neoplasms (59). In vascular smooth muscle cells, COL-stimulated translocation of YAP/TAZ towards the nucleus drives increased DDR1 gene transcription, establishing a positive-feedback loop whereby DDR1 further modulates nucleocytoplasmic shuttling dynamics of these critical mechanotransduction effectors (141).

DDR2, primarily expressed on stromal cells within cancerous tissues, has been associated with increased metastatic potential and invasive behavior in malignant tumors. In ovarian carcinoma, for example, DDR2 responds to fibrillary COL signals by modulating metabolic processes that drive tumor proliferation and dissemination. Specifically, mesenchymal cells within the tumor stroma stimulate COL synthesis through the induction of SNAI1 gene expression, which leads to elevated arginase-1 levels (142) or enhanced glycolytic activity (143). Additionally, SNAI1, the key regulator of epithelial-mesenchymal transition, promotes excessive deposition of ECM proteins, contributing to more aggressive tumor phenotypes and a higher likelihood of metastasis (144, 145). In breast cancer, DDR2 can also be expressed on the surface of tumor cells. Upon binding to COL-I, DDR2 activates STAT1/p27 signaling, helping to maintain the quiescent state of breast cancer cells (146) (Table 2). Cancer cells often evade chemotherapy and immune detection by entering a quiescent state, which contributes to distant metastasis and tumor recurrence (49, 132, 147, 148). Furthermore, it is not only individual ECM components but the entire ECM network that plays a critical role in promoting metastatic behavior (146). Therefore, further investigation into how DDR2-regulated ECM networks influence cancer progression is essential.

Mannose receptor type C2

Mannose receptor type C2 (MRC2), also known as uPARAP/Endo180, is a COL-binding C-type lectin and a constitutively recycling endocytic receptor characterized by high structural conservation (149). The protein consists of an amino-terminal signal peptide, a cysteine-rich domain, a fibronectin type II domain, eight to ten C-type lectin-like domains, a transmembrane segment, and a short cytosolic tail (149). MRC2 is abundantly expressed on the surfaces of macrophages and fibroblasts, where it plays a key role in the recognition and uptake of COL types I, IV, and V (150) (Table 2). Using mouse models and single-cell RNA sequencing, it has been shown that MRC2 is primarily expressed in myofibroblastic cancer-associated fibroblasts in breast cancer. The absence of MRC2, along with reduced levels of fibrillar COLs, results in the depletion of myofibroblastic cancer-associated fibroblasts marked by α-smooth muscle actin expression, providing evidence that MRC2 may function as a tumor promoter (151). MRC2 KO mice exhibit increased CD8⁺ T-cell infiltration and heightened sensitivity to immune checkpoint blockade therapy compared to WT controls (151). Moreover, MRC2 expression has been associated with poor clinical outcomes in HCC, PDAC, and HNSCC (152, 153, 154). As such, MRC2 could serve as a prognostic marker or a potential therapeutic target, though the specific role of individual COL subtypes in this association remains unclear (155, 156).

LAIR-1 and LAIR-2

LAIR-1 is a COL-binding immunosuppressive receptor belonging to the immunoglobulin family, primarily expressed on the surface of immune cells (157, 158). It features an extracellular immunoglobulin-like domain, a single-pass transmembrane segment, and an intracellular immunoreceptor tyrosine-based inhibition motif (157). Several studies have demonstrated that LAIR-1 can suppress immune responses through interactions between COL and immune cells within the TME (55, 159, 160). Specifically, transmembrane COL or COL fragments generated by MMP1 or MMP9—rather than laminin-rich ECM—have been shown to induce CD8⁺ T cell exhaustion and reduce IFN-γ production via interaction with LAIR-1 (55, 159, 160, 161, 162). This interaction triggers tyrosine phosphorylation events, leading to the recruitment of Src homology 2-containing tyrosine phosphatases 1 (SHP-1) and 2 (SHP-2), which suppress immune cell activation and function, thereby promoting lung cancer progression (55, 160, 161) (Table 2). The LAIR-1–SHP1 axis has also been shown to recruit calcium/calmodulin-dependent protein kinase type 1, leading to the activation of downstream cyclic AMP response element-binding protein (CREB), which sustains the survival and self-renewal of acute myeloid leukemia stem cells (159). Enhanced expression of LAIR-1, along with myeloid markers of the M2 phenotype, has been observed in areas rich in COL-VI deposited by CAFs (163). These regions often overlap with tumor and perivascular areas in high-grade gliomas. This suggests a mechanism wherein immune cells bind to COL-VI produced by CAFs near blood vessels or specific tumor locations, engaging in immunosuppressive processes that promote T cell exhaustion through the LAIR-1 receptor. Furthermore, elevated LAIR-1 mRNA levels in various malignancies are associated with shorter overall survival times than lower expression levels (162). These findings highlight LAIR-1 as a promising target for enhancing immunotherapy. This hypothesis is supported by studies on LAIR-2, a soluble homolog of LAIR-1 that competes with COL binding. Due to its stronger binding affinity for COL, the fusion protein antibody NC410 (a dimeric LAIR-2 with a functional IgG1 Fc tail) can effectively counteract the protumor effects of LAIR-1 (164). While NC410 alone is ineffective in immune-competent mouse models, it enhances the efficacy of anti-PD-1/PD-L1 or anti-TGF-β therapies (54, 165, 166), further supporting its therapeutic potential.

GPVI

Platelet glycoprotein VI (GPVI) is a transmembrane glycoprotein receptor primarily expressed on the surface of platelets and megakaryocytes (Table 2). GPVI is activated by ligands such as COLs in the subendothelial matrix and fibrinogen in thrombi (167). The structure of GPVI consists of three main regions: an extracellular immunoglobulin-like domain, a membrane-spanning segment, and a short cytoplasmic tail. Concerning molecular interactions, COLs bind to the D1 domain of GPVI through specialized sequences known as the glycine-proline-hydroxyproline (GPO motif). The helical peptide containing this motif is referred to as a COL-related peptide (167). Recent studies have identified key residues, including Trp76, Arg38, and Glu40, as critical for the binding of GPVI to fibrillar COLs and COL-related peptide (168). It is further proposed that the binding site for COL interaction with GPVI is structured around two distinct COL chains, each featuring overlapping GPO-containing segments. These segments are arranged around a central "core" element with a recurring pattern of OGPOGP, which is essential for efficient binding (168).

Platelets play a crucial role in tumor metastasis through various regulatory mechanisms (169, 170), with their receptors—particularly GPVI—being closely linked to platelet-mediated tumor metastasis. Previous studies have shown that GPVI enhances the metastatic potential of lung cancer and melanoma cell lines (171). More recent research has further indicated that GPVI contributes to tumor growth in established metastatic tumors (172). This highlights the potential for developing antibodies targeting GPVI, as GPVI deficiency leads to intratumoral hemorrhage without causing systemic bleeding and may also enhance the efficacy of chemotherapeutic agents. While the interactions between GPVI and COL are often studied in the context of platelet adhesion to the vascular endothelium, their role in promoting tumor growth or metastasis requires further exploration. One recent study offers a plausible explanation: galectin-3, by mimicking the high-affinity interaction between GPVI and COL-I or COL-III, promotes cancer cell survival and significantly increases lung metastasis in mouse models of colorectal cancer and breast cancer. In this context, GPVI facilitates platelet adhesion to tumor cells through interactions with galectin-3, which is produced by cancer cells (173). This suggests that cancer cells and other components of the TME may regulate tumor progression by expressing COL-like domains, enabling communication with platelets. Analysis of data from the PanCanAtlas dataset, which includes patients receiving immune checkpoint blockade therapy across 19 cancer types, including metastatic melanoma and renal cell carcinoma, has revealed associations between GPVI-mediated platelet activation signaling pathways and tumor immunity (174). These findings suggest that GPVI could serve as a promising target for future cancer therapies, particularly in combination with anti-PD-1 and other immune checkpoint blockade treatments.

COL remodeling-related tumor therapy

Numerous inhibitors and therapeutic strategies targeting COL receptors are currently under development, undergoing clinical investigation, or already available. Given the significant progress made in this field, we will summarize these inhibitors and therapies. While individual MMP plays crucial roles in physiological and pathological collagenolysis, this review will not delve into the biological characteristics of MMPs as several excellent reviews have already thoroughly addressed their properties (175, 176, 177).

COL fragments serve as biomarkers for cancer diagnosis

Elevated levels of cCOL have been associated with increased tumor aggressiveness, metastasis, and poor prognosis in various cancers. Monitoring cCOL levels, which can be detected in the circulation, offers the potential for predicting patient prognosis and tumor progression in clinical settings (Table 3). For instance, ELISA analysis of pretreatment serum from a randomized phase III clinical trial involving patients with stage III/IV PDAC revealed that higher levels of COL fragments were significantly correlated with shorter overall survival (178). In addition, serum markers such as MMP-degraded COL-I (C1M) and COL-III (C3M) have been shown to reflect tissue inflammation in patients with osteoarthritis (179). Similarly, the detection of serum COL-IV has been utilized for the clinical diagnosis of liver fibrosis and cirrhosis (180, 181). Circulating serum COL (C1M, C3M, COL-XVI, and others) levels have been positively correlated with colorectal cancer and ulcerative colitis (182, 183). These findings suggest that serum cCOL levels have significant potential as diagnostic and prognostic biomarkers for cancer progression. However, there is a critical need to standardize assays for measuring COL fragments in clinical settings. Current challenges include variability in assay sensitivity and differences in fragment profiles across cancer types, which must be addressed to enhance the clinical utility of these biomarkers. Furthermore, the ECM within tumors is a complex and heterogeneous environment. Different COL types (e.g., types I, II, III, IV) exhibit distinct degradation patterns in response to various cancers and treatments. It is therefore essential to identify specific COL fragments that are most reliable for monitoring each cancer type.

Table 3.

COL remodeling-related tumor therapies

Cancer diagnosis	Function	Clinical trial/Application
Serum COL fragments detection	Higher levels of serum COL fragments were significantly correlated with shorter overall survival of patients with PDAC (178)	Phase III
Serum MMP-degraded COL-I (C1M) and COL-III (C3M)	Serum C1M and C3M were related to tissue inflammation in patients with osteoarthritis (179), colorectal cancer, and ulcerative colitis (182, 183)	Preclinical
Serum COL-IV	Serum COL-IV was utilized for the diagnosis of liver fibrosis and cirrhosis (180, 181)	Clinical application

CAFs-targeted therapy	Function	Clinical trial/Application
FAPI-PET imaging	Using the high expression of FAP in various cancers to increase the sensitivity and specificity of PET technology (190)	As an imaging tracing aid in clinical practice
CAR-TEAM therapy	Binding to antigens CD3 and FAP and lead to T-cell activation and cytotoxicity of the target cell in PDAC (191)	Preclinical
NIR-PIT	Eliminating chemotherapy resistance caused by CAFs by targeting FAP (192, 193)	Preclinical
Dual-targeting CAR-T therapy	Targeting both cancer cells and CAFs to overcome the immunosuppressive TME (195)	Preclinical
Ocoxin	Combined with the standard-of-care inhibitor Vemurafenib enhances apoptotic effects against melanoma cells (196)	Approved (nutritional supplement); Phase II (chemotherapy adjuvant drugs)
TGF-β inhibitor (LY2109761) and silencing miR-423-5p	Inhibition of TGF-β and silencing miR-423-5p enhanced the in vivo sensitivity of PDAC cells to chemotherapy (197)	Preclinical
IGF2 neutralizing antibody	Combined with the autophagy inhibitor 3-MA was found to reduce tumor recurrence (203)	Preclinical

Inhibitor/antibody of DDR1 and DDR2	Function	Clinical trial/Application
7rh	Reduced melanoma liver metastasis by inhibiting DDR1 to prevent activation of SOX2/STAT3 (204) and had antiproliferative properties against liposarcomas, PDAC, and HCC (130, 205, 206)	Phase I
Nilotinib	For leukemia treatment by targeting DDR1 and shows limited potential for broader oncological applications (207)	Approved for Philadelphia chromosome-positive leukemia
PRTH-101	Modulating ECM remodeling, preventing reduced immune infiltration in TNBC via inhibiting DDR1 (208)	Phase I
Regorafenib	Promoting immune infiltration in colorectal cancer and HCC by targeting DDR2 (209, 210)	Approved

Inhibitor/antibody of LAIR-1 and LAIR-2	Function	Clinical trial/Application
NC410	A dimeric LAIR-2, promoting T cell infiltration (162, 164)	Phase I/Phase II
NC525	Promise for managing gliomas that contain abundant COL fibers via targeting LAIR-1 (163)	Phase I

Inhibitor/antibody of GPVI and MRC2	Function	Clinical Trial/Application
JAQ1 F(ab’)2	Enhancing intratumoral bleeding and inducing cancer cell apoptosis (212, 213)	Preclinical
A5/158	Causing cell death specifically in MRC2-expressing sarcoma cell lines (155)	Preclinical

CAF-targeted therapy

Fibroblast activation protein (FAP) is highly expressed in multiple subpopulations of CAFs and exerts a synergistic effect with CAFs during tumorigenesis. Its pro-oncogenic effects have been observed in various malignancies (184, 185, 186, 187). Research suggests that FAP mediates the transformation of tumor-promoting inflammatory CAFs by activating the STAT3–CCL2 pathway through a uPAR-dependent FAK-Src-JAK2 signaling cascade, which in turn recruits myeloid-derived suppressor cells (188). Targeting approaches, such as vaccines based on FAP (189), fibroblast activation protein inhibitors-positron emission tomography imaging (190), and CAR-TEAM (T cells with an antimesothelin chimeric antigen receptor (CAR) and a secreted T-cell-engaging molecule) (191), highlight the importance of FAP-CAFs–based targeting strategies in clinical applications (Table 3). FAP-targeted near-infrared photoimmunotherapy has been shown to effectively overcome esophageal cancer resistance to conventional treatments (192, 193) and demonstrates promising activity in preclinical trials involving colon cancer cell lines (Table 3) (194).

CAFs contribute to the formation of a physical barrier by remodeling the ECM within the TME, hindering drug delivery, and influencing immune cell responsiveness—key factors that affect the success of immunotherapies. To address these challenges, researchers have developed innovative approaches. For example, dual-targeted CAR-T cell technology, designed to target both cancer cells and CAFs, has shown improved outcomes in patients with relapsed/refractory multiple myeloma by overcoming the immunosuppressive TME (195). Combination therapies provide another strategy for addressing the limitations of CAF-targeted treatments. For instance, combining the CAF-targeting agent ocoxin with the standard-of-care inhibitor vemurafenib enhances apoptotic effects against melanoma cells, reduces lung metastasis, and helps overcome drug resistance (Table 3) (196). CAFs also play a critical role in conferring chemoresistance to cancer cells. In studies investigating exosomes released from CAFs in pancreatic cancer, researchers identified elevated levels of microRNA-423-5p, which directly targets GREM2 and induces chemotherapeutic drug resistance via activation of the TGF-β signaling pathway (197). Treatment with the TGF-β–specific inhibitor LY2109761 increased apoptosis in pancreatic cancer cells while silencing microRNA-423-5p improved the in vivo sensitivity to chemotherapy (Table 3) (197).

Recent studies have revealed that CAFs regulate tumor metabolism through the activation of autophagy, suggesting the potential benefit of combining autophagy inhibitors with other CAF-targeted therapies, such as chemotherapy. In a glutamine-deprived PDAC mouse model, CAFs secrete nucleosides via the NUFIP1-dependent autophagy pathway to promote glucose uptake, compensating for glutamine deficiency and ultimately supporting tumor growth (198). Autophagy is known to help dormant tumor cells resist apoptosis and protect cancer cells during chemotherapy (199, 200, 201, 202). In a subcutaneous tumor-forming mouse model of malignant melanoma, the combination of the autophagy inhibitor 3-MA with an insulin-like growth factor 2–neutralizing antibody effectively reduced tumor recurrence after radiation therapy (Table 3). Insulin-like growth factor 2, secreted by CAFs, supports the completion of autophagy in cancer cells following radiation by increasing reactive oxygen species levels and suppressing mTOR signaling (203). These findings highlight the potential of combinatorial approaches targeting both CAFs and autophagy, alongside conventional chemoradiotherapies, to enhance antitumor efficacy.

Small molecule inhibitors targeting DDR1 and DDR2

Small molecule inhibitors targeting COL-related pathways primarily exert antitumorigenic effects by disrupting multiple signaling pathways associated with COLs. Downstream signaling components of COLs, such as small molecules targeting their receptors, are known to influence COL remodeling activities, affecting cell migration and chemotaxis, and subsequently regulating oncogenesis. These agents have garnered significant attention in recent years for their potential in cancer therapeutics. In particular, DDR1 and DDR2 have been identified as key mediators of COL-associated inflammation and carcinogenesis, prompting extensive investigations into therapeutic strategies targeting these receptors.

One DDR1 inhibitor, 7rh (DDR1-IN-2), has been extensively investigated in mouse cancer models (Table 3). It suppresses activation of the SOX2-dependent STAT3 pathway by inhibiting DDR1, which is triggered by activated HSCs, leading to successful treatment of melanoma liver metastases (204). Additionally, 7rh demonstrated potent antiproliferative effects against liposarcomas and PDAC expressing high levels of DDR1 (205, 206). When combined with the YAP/TAZ inhibitor verteporfin, it exhibited synergistic inhibition of tumor proliferation in HCC (130). Despite promising preclinical results, the development of small molecule drugs targeting DDR1 in the clinic has been slow. To date, only nilotinib, an approved drug targeting DDR1, has secured marketing authorization in Switzerland, the USA, Japan, and China. However, it is primarily used for leukemia and shows limited potential for broader oncological applications (207). Other small molecule inhibitors, such as ANG-3070 and AVI-4015, are currently in phase II clinical trials for treating kidney disease and dry eye syndrome, respectively. A monoclonal antibody, PRTH-101, targeting DDR1 in TNBC, recently began phase I clinical trials. This antibody modulates ECM remodeling without altering COL abundance, effectively preventing reduced immune infiltration caused by ECM rearrangement by binding to DDR1 and inhibiting its phosphorylation (208). Regorafenib, a small molecule inhibitor targeting DDR2, completed preclinical testing and has been approved for use in the USA, China, and other regions for various indications, including colorectal cancer and HCC. Its efficacy in treating multiple malignancies has been well-established (209, 210).

Small molecule inhibitors/antibodies targeting LAIR-1 and LAIR-2

LAIR-2, a soluble homolog of LAIR-1, shares 84% sequence similarity with LAIR-1 (161). Its ability to compete with LAIR-1 for COL binding has attracted attention as a potential target for antibody development. The fusion protein antibody NC410, a dimeric LAIR-2 with a functional IgG1 Fc tail, has been shown to reverse the inhibitory effects of MMP9-generated COL-I fragments on T-cell infiltration, suggesting significant clinical potential (162, 164). In addition to NC410, another antibody targeting LAIR-1, NC525 (NextCure, NCT05787496), is currently undergoing Phase I clinical trials. NC525 functions as both an antagonist and agonist of LAIR-1 (Table 3) and shows promise for managing gliomas that contain abundant COL fibers (163). Preclinical studies have demonstrated its ability to induce apoptosis in leukemic stem cells, potentially offering synergistic benefits when combined with conventional chemotherapy agents such as azacitidine and venetoclax (211).

One of the main strategies for inhibiting GPVI or MRC2 is the development of monoclonal antibodies that specifically block its interaction with COL or other molecules involved in tumor progression. These antibodies could either directly inhibit GPVI or MRC2's function or serve as vehicles for drug or toxin delivery to the TME. For example, GPVI antibody JAQ-1 is found to specifically enhance intratumoral bleeding and induce cancer cell apoptosis in both prostate and breast cancer mouse models (Table 3) (212, 213). An anti-Endo180 monoclonal antibody (A5/158) conjugated to the antimitotic agent, monomethyl auristatin E via a cleavable linker, is rapidly internalized into target cells and trafficked to the lysosome for degradation, causing cell death specifically in MRC2-expressing sarcoma cell lines, reducing off-target effects on healthy tissues (155) (Table 3). As of now, clinical trials directly investigating GPVI or MRC2 inhibition for cancer treatment are still limited. However, the potential for such inhibitors to target platelet–tumor cell interactions, disrupt the TME, reduce fibrosis, and modulate immune responses is gaining traction. Some studies have explored the potential of antiplatelet therapies (such as aspirin) in cancer, but these drugs are not specifically targeting GPVI and do not have the precision or selectivity needed for a GPVI-specific therapeutic effect (214). The overview provided regarding integrins and their corresponding pharmacological interventions is extensive and well-established (70, 215, 216), thus precluding repetition in this context.

Concluding remarks and future perspectives

Different from the traditional function of COLs—“physical scaffold”—numerous studies revealed that COL remodeling plays important roles in shaping tumor occurrence, growth, and metastasis through COL receptors and their downstream signaling pathways (3, 37, 217). Notably, different types of COL remodeling, or the same COL remodeling—receptor axis in distinct cancers may elicit various outcomes (49, 58, 60, 218, 219). This needs further investigations on how COL receptors recognize distinct COL states—for instance, how DDR1 distinguishes between iCOL-I and cCOL-I—and the exact downstream effectors of COL remodeling in each cancer. Such insights will help refine existing therapies based on manipulating COL remodeling pathways. Despite demonstrating efficacy in addressing therapy resistance in prior studies (192), FAP-targeted near-infrared photoimmunotherapy faces obstacles when dealing with diverse CAF populations. Single-cell genomic profiling performed on breast cancer uncovered a subset of cells concurrently displaying FAP and CAF-S1 expression patterns that exhibit primary resistance to near-infrared photoimmunotherapy (220), underscoring the complexity associated with targeting heterogeneous CAF populations. Therefore, in-depth and accurate research on the role of heterogeneity of CAFs is necessary. The involvement of COL cleavage state or COL arrangement in shaping malignancy through DDR1 highlights its importance as a potential anticancer target (58). Nevertheless, existing DDR1 antagonists often present unwanted properties such as multitarget engagement or off-target interactions resulting in undesirable side effects like hypertension and proteinuria (221, 222). Advancements in bioinformatic tools combined with artificial intelligence–driven methodologies show promise for identifying novel inhibitors tailored exclusively towards DDR1 (223). Despite offering substantial benefits, therapeutic agents targeting COL remodeling and their downstream effectors face complex hurdles in safe and efficient deployment.

Conflicts of interests

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Robert Haltiwanger