Advances in cell-derived extracellular matrix for 3D tumor modelling

Abstract

Cell-derived decellularized extracellular matrix (cdECM) is being increasingly explored as a supportive biomaterial for bioengineering biomimetic 3D tumor models. The potential to represent the biomolecular diversity of the extracellular matrix in a tumor-specific mode and its production using human cell cultures renders in vitro produced matrix a particularly attractive alternative to non-human counterparts. In this review, we describe well-established and emerging strategies to modulate human ECM production in vitro, as well as showcase recent advances in decellularization and processing methodologies that have been explored across different scales and diverse biological complexities. The incorporation of cdECM into the design stages of bioengineered tumor models and its biological relevance are also discussed considering currently available alternatives to model tumor ECM. Finally, we look at the current challenges in using cdECM-based biomaterials to develop preclinically relevant models and discuss the need for innovative production approaches and methodological developments in downstream processing to widen the use of cell-derived biomaterials for tumor modelling and other applications in the tissue engineering field.

Graphical abstract

1. Introduction

The extracellular matrix (ECM) is a key structural and regulatory network of the tumor microenvironment (TME), consisting of fibrillary proteins (e.g., collagens, fibronectin, laminin), proteoglycans and glycosaminoglycans (e.g., hyaluronic acid, heparan sulfate, chondroitin sulfate) [1]. During tumorigenesis, the ECM undergoes extensive structural and molecular remodeling driven by aberrant bidirectional interactions between cancer cells and stromal components of the TME. These alterations reprogram the biochemical and biophysical properties of the ECM, thereby promoting tumor growth, invasion, and metastasis [2,3]. This structural remodeling is driven by sustained synthesis and deposition of ECM components, together with continuous cycles of matrix degradation and de novo deposition. These processes are accompanied by enhanced ECM crosslinking, as well as increased expression and activity of matrix-remodeling enzymes, including lysyl oxidase (LOX) and matrix metalloproteinases (MMPs) [4]. Accompanying these alterations, cancer cells secrete a range of pro-fibrotic cytokines and growth factors, including TGF-β, PDGF, and EGF-family ligands which activate stromal cells, including cancer-associated fibroblasts (CAFs). Activated CAFs further amplify these signals and become major producers of ECM proteins such as fibronectin and collagen, leading to excessive matrix accumulation and increased tissue stiffness [5]. Accumulating evidence has demonstrated that increased ECM stiffness is highly correlated with increased tumor aggressiveness and a poorer prognosis in various types of cancer, including liver, pancreatic, and lung cancer [[6], [7], [8], [9], [10], [11]]. The resulting process of tumor-ECM dysregulation leads to the establishment of abnormal feedback loops, which ultimately results in further dysregulation of cancer cells and tumor-associated cells, supporting progression and resistance to therapeutic agents [12]. Owing to the key role that the cell-ECM interplay exerts in tumor progression, the inclusion of ECM-mimetic biomaterials that mimic the biomechanical and biomolecular complexity of the native tumor ECM is highly desirable to bioengineer three-dimensional (3D) models that aim to replicate human cancers in vitro.

Decellularized extracellular matrix (dECM)-based biomaterials have received increasing interest for generating 3D in vitro models owing to their ease of processing into fibrillar or hydrogel forms, as well as to their natural display of multiple cell bioinstructive cues when compared to those obtained with standalone biomaterials (i.e., collagen, fibronectin, etc.) [[13], [14], [15]]. More recently, in vitro-produced decellularized cell-derived matrices (cdECM) have emerged as a promising alternative to ECM derived from human patients or animals [16].

Despite being an attractive alternative, cdECM production still requires intensive optimization given its low yield when compared with tissue/organ decellularized counterparts [17]. Implementation of advanced culturing techniques, namely the addition of compounds that stimulate collagen biosynthesis (e.g., ascorbic acid) [18], the addition of macromolecular crowding agents (e.g., dextran sulfate, Ficoll, hyaluronic acid (HA) or carrageenan) [19], the introduction of mechanical stimuli, [20], the manipulation of culture media serum concentrations, or hypoxic conditioning [20], have been explored as strategies to improve the yield of cell-produced ECM. Yet significant advances in these strategies are still required. Careful considerations regarding cell culturing setups, two-dimensional (2D) or 3D/monoculture or multi-culture, as well as a rational selection of decellularization methods and downstream processing methodologies must be implemented for obtaining increasingly reproducible cdECM biomaterials [17]. In this context, herein we review the latest developments in the production, decellularization and processing of cdECM biomaterials and their use for engineering 3D tumor models. Particularly, we discuss the physiological significance of cdECM-based biomaterials and their integration into the design stages of bioengineered 3D models representing various tumor types. Finally, we examine the current bottlenecks hindering the widespread use of cdECM-based biomaterials in the context of bioengineering increasingly biomimetic human tumor models in vitro.

2. Cell-derived decellularized ECM as a promising biomaterial for developing 3D tumor models

To date, strategies for including ECM-mimetic biomaterials in the bioengineering stages of 3D in vitro models’ development have mostly relied on the use of unitary components such as collagen, gelatin, and hyaluronan, or on their combination. Despite leading to interesting outcomes these supporting matrices are not entirely representative of the biomolecular diversity of the tumor ECM found in human tumors [21]. Recent approaches aim at overcoming these limitations by exploring more complex and physiologically relevant matrices, such as dECM derived from healthy tissues, due to its heightened biomimetic and tissue-specific biomolecular content [1,22,23]. In this context, photocrosslinkable tissue-derived dECM sourced from healthy porcine pancreas was recently used to create compartmentalised Pancreatic Ductal Adenocarcinoma (PDAC) models that integrate both tumor and stromal cells. By combining tissue-specific dECM with hyaluronic acid-tyramine visible-light crosslinkable hydrogels, tumor and stromal regions were engineered within the same construct, enabling a more faithful recapitulation of pancreatic cancer desmoplasia and biological architecture [21]. This approach has enabled dual evaluation of tumor- and stroma-targeting therapies. Importantly, this model underscored that the biochemical specificity of pancreas-derived dECM contributes to altered cell behavior and therapeutic response. Tissue-specific dECM has also been explored as a bioink for bioprinting 3D tumor models. In a recent approach, decellularized breast tissue matrix was formulated into dECM microfibre bundles (dECM-μF) and mixed with photocrosslinkable hyaluronan to produce reproducible quasi-spherical tumor beads by extrusion printing into a superhydrophobic printing bed [15]. This methodology termed flow-on-repellent bioprinting (FLORE) emphasizes how tissue-specific dECM can be processed as a hybrid bioink to support rapid generation of tumor models and in a high-throughput mode. Notably, across these platforms, the presence of dECM, especially when accompanied by other stromal components such as CAFs, is frequently associated with the emergence of cellular phenotypes that promote increased drug resistance [15]. Such findings suggest that 3D tumor models incorporating dECM-biomaterials in preclinical screening stages may exhibit an increased predictive potential, however, a systematic benchmarking of such 3D models to data obtained from clinical outcomes remains a requirement to fully uncover the biomimicry potential of such strategies. In addition, off-the-shelf availability of dECM-containing 3D tumor models remains a key bottleneck to their wider use off-site, whenever needed. Strategies such as cryobioprinting approaches combining the principles of 3D bioprinting and cryopreservation are emerging as valuable strategies to mitigate these limitations [24,25]. In contrast to conventional bioprinting, which operates at room- or physiological temperatures, cryobioprinting employs a sub-zero cooled printing bed (typically operating between −10 °C and −40 °C). As the bioink is extruded and encounters this cooled printing bed, it undergoes a rapid, and controlled phase transition via freezing. This allows for the immediate preservation of printed constructs geometry without requiring immediate chemical/light-based crosslinking. In this approach, bioinks, such as dECM-based bioinks have been typically formulatedwith a cocktail of cryoprotective agents (CPAs) to support cell viability post-thawing [26]. Typically, this encompasses the use of intracellular permeating agents, most commonly 10% dimethyl sulfoxide (DMSO) in combination with non-permeating sugars like trehalose or glycerol to protect encapsulated cells. While conventional 3D bioprinted models are manufactured with a print-to-use rationale, cryobioprinted tumor models can be printed and generally stored at freezing temperatures such as at −80 °C for mid-term timeframes (spanning from 3 to 6 months) or in liquid nitrogen (−196 °C) for long-term storage (> 1 year) while preserving cell viability and matrix architecture post-thawing [25]. In this line, a recent report adapted porcine tissue-derived dECM bioinks for cryobioprinting shelf-stable human pancreatic tumor models. By integrating cryoprotective agents in the dECM-based constructs, the authors demonstrated the feasibility of fabricating tumor tissues that can be stored, transported, and revived without significant loss of viability or drug responsiveness [26]. This approach addresses the off-site usability of complex 3D models, positioning tissue-derived dECM as a cell supporting biomimetic material that is processable in cryobioprinting workflows. Despite these advances, the generation of storable 3D models employing human tumor tissue-specific ECM remains highly challenging to materialize. In this context, cell-derived ECM produced in vitro by using human cells as biofactories is a highly attractive alternative for recapitulating matrix cues found in human tumors [27]. The in vitro production of ECM can be designed as a highly modular and versatile process, with current approaches taking advantage of a wide selection and combination of ECM-producing cell types (e.g., malignant and stromal cells), their further modification (e.g., genetically), or exposure to certain stimuli (e.g., macromolecular crowding, cytokines/growth factors, etc.), all with the aim to improve ECM production [1,16]. However, given the relatively low yield of ECM when compared with tissue/organ decellularization-based approaches, producing cdECM encompasses a variety of challenges. Fine-tuning the production process will allow researchers to overcome the current limitations and harness the full potential of cdECM as a biomimetic biomaterial. Ultimately, the goal is to develop standardized protocols that can consistently produce cdECM-based biomaterials that mimic the biochemical and physical features of the native tumor ECM [17]. Holistically, the production of cdECM can be modulated by various factors, including cell source, cell culture conditions (e.g., viscosity, 2D/3D setups, static/dynamic fluid flow) and mechanical stimulation. Regarding the latter, rotating flasks/oscillating platforms have proven to be valuable systems for dynamic cell culture and dECM production. These setups allow for precise control of culture parameters such as oxygenation/hypoxia, shear stress and establishment of nutrient gradients, in turn directly modulating matrix deposition [28]. Other alternative production platforms such as microfluidic chips allow not only a continuous nutrient supply and waste removal (culture under flow), but also a precise control over shear stress/mechanical stimulation, as well as a dynamic interaction of different cell types under physiologically relevant flow conditions. Notably, the production of fibroblast-derived dECM sheets and on-chip-decellularization has been successfully shown [29], yet the use of such chips remains rather underexplored for fabricating tumor ECM. A careful selection of cell sources, culture methodologies and conditions is therefore essential for a robust production of cdECM biomaterials in vitro (Fig. 1). In the following sections, such concepts are elaborated and discussed considering their influence on cdECM production and biomolecular content.

Fig. 1.

Schematics of the key factors to consider during in vitro production of cell-derived extracellular matrix (cdECM). Factors include careful selection of the cell source, cell culture conditions, the composition of the culture medium, and cell culture conditions, among others, as these have an impact on both productivity and biomolecular content. The image was created using PowerPoint, with icons from BioRender. Mano, J. (2026) https://BioRender.com/ifpvrh7.

2.1. Design blueprints for cell-derived extracellular matrix production

2.1.1. Cell source and culture strategies

The cell types selected for in vitro production of ECM are key as they directly impact the biomolecular composition of the resulting cdECM. In the context of using cells as ECM biofactories, researchers have explored the production of cdECM using both primary cells and cell lines [30]. Primary cells are the closest to their original in vivo phenotype and may be better able to generate a matrix that mimics the native microenvironment. Therefore, they are theoretically considered the best cell source for cdECM generation [17]. Yet, primary cells are cultured in vitro for limited passages, and it is challenging to explore them as biofactories for producing cdECM indefinitely. As an alternative, immortalized cell lines can be employed to produce cdECM due to their ease of culture and capacity to produce large cell batches.

On the other hand, immortalized cell lines can differ substantially in phenotype, gene expression, and secretome, when compared with primary cells [31]. Either way, ensuring reproducibility across cell passages and donors remains a challenge that must be fully addressed. Following cell selection, another key design consideration is cell culturing conditions to optimize their use as ECM biofactories. Specifically for tumor modeling applications, the inclusion of CAFs for producing ECM in vitro is particularly valuable, as they are the main ECM-producing cells in the native TME [32]. This has remained rather underexplored to date and advances in this direction are envisaged in the near future.

Among the cell culture strategies used to produce cdECM, the most common approaches include culturing cells as adherent monolayers in a 2D setting, either in a scaffold-free or scaffold-based mode (e.g., coated surfaces) [33], as multicellular aggregates in 3D, or through encapsulation in a suitable carrier (e.g., bulk or porous microparticles) [34] until ECM is deposited [35].

2.1.2. Culture media composition

The selection of growth factors and supplements to be added to the culture medium during cell culture for ECM production can greatly impact cell phenotype and bioactivity, as well as ECM deposition and its final biomolecular diversity. To date, several classes of small molecules have been included in standard culture media to modulate ECM production in vitro. Particularly, ascorbic acid has been one of the most used as a culture medium additive to promote cell-derived ECM production in vitro [36]. This water-soluble vitamin with antioxidant properties plays a key role in collagen biosynthesis and is a cofactor for enzymes such as lysyl and prolyl hydroxylases, being also essential for collagen fibrillogenesis [[37], [38], [39]]. Depending on its concentration, ascorbic acid has been shown to act as an antioxidant at lower physiological ranges (10 to 100 μM) [40] or as pro-oxidant at higher experimental concentrations (>100 μM to the low mM range) [41], which consequently activates transcription factors and increases the expression of TGF-β1, thereby regulating collagen gene expression [39,42]. However, as this small molecule is susceptible to rapid degradation, it is important to perform frequent removal/addition cycles to the culture medium, a key limitation if one envisions high-scale production of dECM in an in vitro setting. To overcome this limitation, recent approaches have focused on 2-phospho-L-ascorbic acid, a highly stable alternative [18,37,43,44]. In addition, it has been demonstrated that the direct addition of growth factors to the medium, such as TGF-β1 (e.g., at a concentration of 5 ng/mL), increases the production of glycosaminoglycans (GAGs) and collagen [45]. It is important to emphasize that collagen deposition in vitro is substantially slower than in vivo, as the more dilute medium makes it more difficult for the specific C and N-proteinases to cleave the pro-peptides, thereby resulting in a lower conversion of pro-collagen into collagen I [46]. Therefore, to favor chemical reactions, macromolecular crowding (MMC) agents have been increasingly used to improve ECM production in vitro [47]. Macromolecular crowding (MMC) is based on the principle of excluded volume, according to which the simultaneous presence of numerous macromolecules reduces the free space available in the medium [48,49]. As a result, diffusion and proximity between molecules are restricted, significantly altering the dynamics of biochemical reactions. This strategy has attracted particular interest due to its ability to increase collagen deposition. In diluted cell cultures devoid of MMC agents, water-soluble pro-collagen tends to dissolve, and N and C proteinases tend to be deactivated before they meet, resulting in reduced collagen deposition. In contrast, in cultures containing MMC agents, diffusion is restricted, increasing the likelihood of interaction between procollagen and proteinases, which accelerates and intensifies collagen deposition. Thus, MMC acts as an essential physicochemical modulator promoting increased ECM deposition [49,50]. MMC has been explored mainly by using dextran sulfate, Ficoll and/or carrageenan [37]. The use of these MMC agents has shown to trigger a faster conversion of pro-collagen into collagen in vitro [46]. Specifically, the use of Ficoll in the cell culture medium seems to contribute to increased collagen I in human corneal fibroblasts cultivated in 2D [19]. Also, in 3D, it was shown that the presence of Ficoll 70 + Ficoll 400 in human mesenchymal cell culture provided a significant increase in gene expression levels of key extracellular matrix-associated genes, namely COL1A1, COL3A1, and FN1. On the other hand, carrageenan has been suggested as the most suitable macromolecule for promoting deposition of ECM components, due to its greater polydispersity compared to other molecules such as Ficoll 70, Ficoll 400, and DxS [46]. In fact, the use of carrageenan as a MMC agent has shown to increase the deposition of collagen I, III, IV, V, and VI, proteins such as laminin and fibronectin, as well as proteoglycans, and enzymes important in collagen and elastin crosslinking, such as Lysyl oxidase (LOX) [46]. Interestingly, it has been shown that the incorporation of λ-carrageenan into the culture medium of human bone marrow-derived mesenchymal stromal cells grown in 2D results in a higher GAG content, which allows for greater retention of growth factors capable of binding to heparin, for example. [51]. The joint addition of carrageenan and growth factors, such as TGF-β3, to the culture medium of human tenocytes grown in 2D was also explored, resulting in increased collagen deposition [52]. Alternatively, the addition of HA to the culture medium was also evaluated in human dermal fibroblasts also cultured in 2D, and it was found that concentrations of 0.05% of HA enhance collagen deposition and an increase in matrix rigidity when compared to Ficoll 70/400 combinations, suggesting that in the latter case the cells were in a proliferative rather than secretory state [53]. Interestingly, higher concentrations of HA, such as 0.5%, may result in greater viscosity and consequently less collagen deposition [53]. For DxS, its addition to the culture medium of bone marrow-derived mesenchymal stromal cells cultured in 2D has been shown to accelerate the removal of C and N peptides from pro-collagen and thus also seems to positively impact collagen deposition [54]. The addition of MMC agents thus seems to favor collagen deposition, without impairing the physiological function of the cells, but it is necessary to assess which agent and its concentration is the most appropriate to use [46]. Although MMC appears to be a valuable methodology to promote ECM in vitro production, it is important to discuss that more studies should be carried out to better elucidate the impact that this technique can have on gene expression, cell metabolism, and regulation of cytoskeleton dynamics [47].

Other relevant parameters that must be addressed during cell-derived ECM production are serum concentration and culture time. Recent evidence points to the fact that an increase in serum concentration (e.g., fetal bovine serum (FBS)), and culture time, contributed to the reduction of collagen deposition, possibly due to the presence of a higher content of MMPs that prompt matrix degradation [19,46,55]. Furthermore, the use of human serum has been suggested as an alternative to FBS, to reduce animal sourcing for cells culture, and also owing to its relatively lower amount of MMPs [56]. The latter appears to favor greater collagen deposition in vitro [46]. Thus, cell culture with low serum concentrations may be advantageous, as it reduces the likelihood of ECM degradation, as well as an imbalance in its natural remodeling rate. It is important to emphasize that most of the existing culture media modulation approaches focus on collagen as the readout. However, the ECM biomolecular landscape is considerably more complex than its major structural constituents.

This complexity is encompassed by the concept of the matrisome, defined as the complete set of ECM proteins and ECM-associated molecules [57]. The matrisome comprises core matrisome, including structural scaffolding proteins such as collagens, proteoglycans, as well as matrisome-associated proteins, such as ECM-affiliated proteins including glycoproteins, ECM remodeling enzymes (e.g., matrix metalloproteinases), and secreted signalling factors (e.g., TGF-β) [58]. Despite increasing efforts to engineer and characterize in vitro-produced cdECMs, the role of many matrisome components in determining ECM bioactivity remains largely unexplored. Therefore, comprehensive studies addressing the composition and dynamics of the in vitro matrisome are needed to identify key molecular similarities that recapitulate tumor tissue dECMs. Such insights will provide a foundation for the rational optimization of cdECM production processes, allowing culture conditions, cell sources, and bioprocessing parameters to be tailored toward the generation of tumor-specific biomaterials.

2.1.3. Cell culture conditions

Cell culture conditions must be strictly controlled as these have a direct effect on cell proliferation and ECM deposition. This includes key factors such as oxygen levels and culture environment dynamicity, as well as regular changes in the culture medium to provide cells with the necessary nutrients and growth factors.

The influence of oxygen tension on ECM deposition has been evaluated, showing that hypoxic conditions may be favorable for increasing the deposition of fibronectin and collagen I, V, and VI, especially when these are coupled with the use of MMC agents [55,59]. In this context, it was recently demonstrated that the use of carrageenan with low serum concentrations and hypoxic conditions substantially increased ECM deposition in 2D culture of human corneal fibroblasts [55]. The use of Ficoll 400 associated with these two conditions has also been shown to favor the deposition of fibronectin and collagen [60]. However, one must consider the susceptibility of different cells to these conditions, as this may vary depending on the type of cell in culture [55].

Furthermore, considering that in vivo cells are subjected to different traction and compression forces, it has been suggested that mechanical stimuli can also modulate ECM deposition. Recently, 2D co-cultures of bone mesenchymal stem cells and chondrocytes showed greater deposition of collagen II and GAGs, but also of regulatory factors, such as TGF-β, when subjected to a mechanical stimulus, particularly a cyclic sinusoidal dynamic traction [61]. Also, in a study involving dermal fibroblasts obtained from adult donors, it was found that when subjected to a mechanical stimulus, in this case, mechanical stretching, greater proliferation and deposition of collagen occurred. Furthermore, the effect of stirring on cdECM production was evaluated, suggesting that its presence contributed to a greater expression of collagen I and fibronectin and consequently to an increase in the stiffness of the ECM [16]. Optimizing dynamic stimuli for cdECM deposition is therefore highly valuable [62,63]. In tensile stretching systems, this may involve calibrating both strain amplitude and cyclic frequency to reproduce tissue-relevant mechanical loading conditions. Depending on the tissue and experimental model, cyclic strain amplitudes commonly range from 5% to 20%, while loading frequencies are typically selected to mimic the physiological dynamics of the target tissue, often within the sub-hertz to hertz range [64]. Conversely, in dynamic fluid systems, the focus shifts to stirring velocities and perfusion conditions capable of generating increasingly biologically relevant shear stress regimes. Replicating these parameters is vital because matrix biophysics adapt dramatically during tumor progression. While many healthy soft tissues exhibit elastic moduli in the range of approximately 0.1–5 kPa [65], tumor-associated fibrosis can increase matrix stiffness by one to two orders of magnitude, often reaching tens of kilopascals depending on tumor type (for example in breast tissue, normal tissue:∼1.13 kPa; tumor tissue: ∼55 kPa [66]). Importantly, different measurement techniques can yield distinctive tissue stifness values and targeting definitive values for human tumors should be carefully addressed [67]. Elevated interstitial fluid flow is also another hallmark of solid tumors, generating low-magnitude shear stresses (typically 0.001–0.1 dyn/cm²), this in turn influences cellular mechanotransduction, migration, and ECM remodeling [[68], [69], [70]]. Therefore, incorporating tumor-relevant interstitial flow conditions during cdECM production in vitro may provide valuable means to steer matrix deposition toward a more physiologically representative tumor matrisome and biomechanical profile.

In conclusion, several factors can be adjusted during in vitro fabrication of cdECM biomaterials, and their selection should be ideally based on the specific requirements of the 3D tumor model to be developed. In this context, studies focusing on the exploitation and optimization of these variables have increased substantially (Table 1).

Table 1.

Summary of biomolecular and microenvironmental modulation approaches for tailored in vitro cdECM production.

	Parameters		Conditions	Cell line	Outcome	Ref.
Biomolecular Modulation	Ascorbic acid		Increasing dose (10, 100, 300 and 600 μM)	Human dermal fibroblast	Increased collagen deposition	[18]
	AA2P		Increasing dose (10, 100, 300 and 600 μM)	Human dermal fibroblasts	Effect analogous ot hat of ascorbic acid, but less susceptible to degradation	[18]
	TGF-β1		5 ng/mL	Human neonatal dermal fibroblasts	Increased the deposition of GAGs and collagen, resulting in a denser matrix	[45]
	MMC agents	Ficoll	37.5 mg/mL Ficoll 70 + 25 mg/mL, Ficoll 400	Human corneal fibroblast	Increased collagen I deposition	[19]
		Carrageenan	75 μg/mL	Human osteoblasts, tenocytes and lung fibroblasts	Increased the deposition of fibronectin, laminin and collagen I, III, IV, V and VI	[46]
		HA	0.05% w/v	Human neonatal fibroblast	Increased collagen deposition, however, 0.5% w/v may result in higher viscosity and lower collagen deposition	[53]
		DxS	100 μg/mL	Bone marrow mesenchymal stromal cells	Accelerated the removal of C and N peptides from pro-collagen and the consequent deposition of collagen	[54]
	Serum concentration		0.5% v/v	Human skin and lung fibroblasts	Seems to contribute to a lower presence of MMPs and consequently to a greater deposition of collagen	[46]
Environmental Modulation	Oxygen tension		2%	Human corneal fibroblast	Increased the deposition of fibronectin and collagen I, V, VI	[55]
	Mechanical stimulus	Dynamic sinusoidal cyclic traction	Strain amplitude from 5 to 10% at 0.5 Hz	Co-culture of chondrocytes and rabbit bone mesenchymal stem cells	Increased the deposition of collagen II, GAGs and TGF-β1	[61]
		Mechanical stretching	-	Adult human dermal fibroblasts	It increased proliferation and collagen deposition	[62]
		Stirring	35 rpm	Human adipose mesenchymal stem cells	Increased the expression of fibronectin and collagen I, as well as the stiffness of the matrix	[20]

2.2. Methods for decellularizing cdECM

The decellularization process is a highly relevant aspect of the downstream processing of ECM biomaterials. Decellularization must be optimized according to the target tissues, to strike a balance between retaining the bioactive components of the ECM and removing the cellular and nuclear components [71]. Decellularization methods can be classified as physical, chemical, or enzymatic [22,72,73]. Physical methods can involve pressure gradients, freezing and thawing cycles, mechanical stirring, supercritical CO₂, or osmotic pressure. These methods are often used as a first stage of decellularization and can be used in combination with the use of chemical or enzymatic methods to encourage the entry of compounds into the tissues and the consequent removal of cellular components. These methods focus on the destruction of the cell membrane and are defined as fast and easy-to-operate processes. However, it is important to note that they can result in incomplete decellularization of the matrix or damage to its structure. Chemical methods use different compounds, including alkaline solutions, acids, detergents, alcohols, and chelating agents [38]. The use of detergents such as Triton X-100, sodium deoxycholate, and sodium dodecyl sulfate (SDS) is quite common and is used to promote the solubilization of lipid membranes and proteins, as well as to solubilize cellular components [22,38]. In addition to solubilizing cytoplasmic components, acid solutions are also important for destabilizing nucleic acids [22,38]. Alkaline solutions, which promote DNA denaturation, and chelating agents, which sequester divalent cations, interfering with integrin binding and consequent cell adhesion, are also options used in decellularization approaches that leverage chemical methods [38]. However, it is important to note that the use of certain reagents, such as acids and bases, can result in a loss of ECM microstructure and key components, such as growth factors or glycosaminoglycans (GAGs) [22,73]. Finally, enzymatic methods can also be used, using proteases such as trypsin to remove proteins, nucleases to remove nucleic acids, and lipases to remove lipids [22,38,73]. Guanidine can also be used to remove adipocytes and ethylenediamine tetraacetic acid (EDTA) to break intercellular bonds [73]. The decision of decellularization methodology must also consider the final application since a host inflammatory response can be triggered, due to the presence of possible traces of reagents used in the decellularization process [73]. The use of combinatorial decellularization approaches, including the use of EDTA and another enzymatic method have been shown to reduce the risk of an inflammatory response [73]. Enzymatic decellularization with deoxyribonuclease has been shown to reduce the loss of sulfated GAGs compared to methods involving detergents, while also allowing endogenous growth factors to be preserved [51]. The selection of decellularization methodologies is therefore key to attaining a compromise between effective decellularization and maintenance of key ECM components [22]. Decellularization of cdECMs tends to be gentler than applied to complex tissues, such as those of animal or human origin. In tissues, both the strategy used, and the duration of the process vary depending on material density, lipid content, and degree of vascularization, often requiring the combination of several methods and extended decellularization periods. Although simpler, the protocols applied to cdECMs also require extensive optimization and adaptation to the specific type of cdECM being studied, in order to ensure efficient removal of cell content without compromising matrix integrity [1,73]. In this sense, the protocols used for cdECMs are often based on the use of chemical agents such as Triton X-100, ammonia, EDTA, or SDS, and sometimes enzymes such as DNase, RNase, or collagenase [1,[74], [75], [76], [77]]. In addition, for cdECMs decellularization can be achieved relatively quickly, occurring from a few minutes to about 24 h, depending on the thickness of the cell layer and the applied methodology. Still, this process requires exhaustive optimization for each specific type of cdECM. After cellular removal, the dECM, whether from tissue or cell cultures, is usually sterilized before being integrated into tumor models. This process can involve different agents, such as ethanol, peracetic acid, antibiotics, and/or UV or gamma radiation. Ensuring the elimination of potential contaminants without compromising the structural and biochemical integrity of the matrix is key in this process [1]. Efficient decellularization of the biomaterial requires carefully defined protocols and careful selection of decellularizing agents and decellularization times to ensure complete removal of cellular or genomic material, while simultaneously ensuring the preservation of the biochemical composition and architecture of the native ECM. After this process is completed, the matrix must be subjected to rigorous characterization to determine the existence of specific biomolecular contents including growth factors, as well as to assess potential structural modifications to its components and that may influence its performance as a supporting and bioinstructive matrix [1,73].

2.3. Characterization of cell-derived extracellular matrix

Characterizing in vitro-produced dECM is crucial to access the effectiveness of decellularization methodologies, its biomolecular composition, and downstream processability [73]. Characterizations can be carried out by using different techniques, including scanning electron microscopy, atomic force microscopy, antibody assays such as western blotting and immunocytochemistry, mass spectrometry, Raman microscopy, and others [37]. One key characterization is that pertaining to the removal of nuclear components, which is generally assessed by staining the cell nuclei with haematoxylin or by quantifying the DNA present. In order to reduce the risk of unwanted immune responses and assure that the biomolecular cues provided to cells are mainly attributed to ECM components in the formulations, DNA content should be less than 200 base pairs in length and less than 50 ng in weight compared to the weight of the ECM non-decellularized in milligrams [22,73]. In addition, fluorescent labelling of cell components is very useful for assessing the removal of residual cellular material following decellularization. For example, staining for nuclear DNA using dyes such as the widely used 4′,6-diamidino-2-phenylindole (DAPI) and comparing the results with those obtained from the corresponding non-decellularized cell-derived matrix can reveal the extent of nuclear material removal [22,78]. Detection of collagen, GAGs, elastin, fibronectin, laminin, and other non-nucleic acid components in order to assess the biomolecular composition of the ECM is often qualitatively performed through histological analysis or quantitatively through commercially available absorbance-based methods [22,73]. In addition, to obtain a more detailed assessment of the composition of the ECM, matrisome analysis can also be carried out by using advanced mass spectrometry methodologies [79]. Nevertheless, the biomolecular complexity of the ECM can make its characterization challenging with this technique in the absence of optimized protocols [37]. Scanning electron microscopy and transmission electron microscopy also play a crucial role in assessing the macro- and microstructure, allowing researchers to compare the processed dECM to the original tissue matrix [22]. Finally, to fully characterize the resultant matrix, it is important to assess the mechanical properties of the dECM, since decellularization generally affects the native structure of the ECM [73].

Analyzing the cellular and biomolecular contents is a critical step in producing cdECM-based biomaterials. Progress in analytical technologies, particularly omics-based approaches, can enable high-throughput functional assessments of the cell's genome, epigenome, transcriptome, proteome, and metabolome. Such methodologies provide a more thorough understanding of the molecular mechanisms underlying ECM production in vitro, thereby opening the door to developing tissue-specific ECM biomaterials that, for example, support 3D tumor models bioengineering [80].

2.4. Cell-derived ECM biomaterials

After decellularization and characterization, the downstream processing of cdECM can be carried out to generate biomimetic hydrogels with specific biomolecular signatures of each tissue (i.e., in a tissue-specific mode). To do this, cdECM is conventionally lyophilized, fragmented to form a powder-like material, and subsequently solubilized [21,81]. This solubilization is usually conducted by exogenous enzymes, such as pepsin or papain [21,81,82]. Proteases are added in defined proportions and under optimal pH and temperature conditions to ensure that their proteolytic activity is maintained. After digestion, gelation may occur in a thermo-dependent manner due to the collagen content present in dECM formulations, typically at 25-37 °C [21,81,82]. However, this process is cumbersome, requires extensive optimization, is dependent on enzymes activity and leads to the presence of unnatural protein residues in final dECM formulations [83]. To overcome these limitations, alternative methodologies based on the application of ultrasonic waves in liquid medium have been explored [81,84]. Interestingly, the matrix processed using this methodology exhibits thermo-dependent crosslinking, but at low temperatures, in contrast to conventionally digested matrices by enzymes [81]. Although ECM composition is better recapitulated in dECM hydrogels when compared to the use of standalone biomaterials (i.e., collagen-only hydrogels, etc.), these platforms are frequently soft and have poorly controlled mechanical properties [85]. Thus, to improve control of the mechanical characteristics of dECM hydrogels, several crosslinking strategies have been explored. These approaches resort to chemical crosslinking of reactive groups widely present in the macromolecular components of ECM, such as naturally occurring tyrosine residues [21,26]. The use of a photoinitiator, such as ruthenium (Ru) and its combination with sodium persulfate (SPS), under visible light (400–450 nm) allows for the formation of covalent bonds between tyrosine residues, resulting in more robust hydrogels with easily modulable mechanical properties [81]. Another strategy already well described is the chemical modification of dECM through methacrylation, subsequently allowing its crosslinking under visible light (e.g., 400 - 450 nm) using LAP (lithium phenyl-2,4,6-trimethylbenzoylphosphinate) as a photoinitiator [86,87]. However, methacrylation implies chemical modifications that are not present in vivo and increases processing time. Light-based dECM crosslinking with the inclusion of various amounts of methacrylated hyaluronic acid has also been explored to increase the existence of this component in hydrogels and better recapitulate its diversity in the TME of different human tumors [15]. An even more advantageous option is the reported use of hyaluronic acid modified with tyrosine residues, which enables the formation of covalent bonds with tyramine residues naturally present in dECM formulations, thus constituting a more straightforward biomimetic solution [21]. After obtaining hydrogel, it is essential to proceed with its characterization, particularly at the level of mechanical properties. For this, various tests are used, including tensile testing, compression, rheology, dynamic mechanical analysis, and atomic force microscopy, allowing to evaluate the mechanical behavior at nano, micro, and macro scales and consequently, determine the viscoelastic properties of dECM hydrogels [16]. Importantly, a significant gap remains with regard to the optimization of cdECM processing, which, in most cases, continues to be digested by exogenous enzymes of animal origin that ultimately remain incorporated into matrices derived from human cell cultures [[88], [89], [90]]. We envision advances in the future where cdECM processing will be devoid of peptidases, indeed, we and others have demonstrated that the use of ultrasound processing can contribute to phasing out the use of such exogenous proteins [91,92]. This will greatly contribute to increasing the purity of such formulations. Regarding downstream processing of cdECMs these biomaterials have for example, been combined with polysaccharides like alginate to allow hydrogel formation [93,94]. Despite being effective, this cdECM/alginate combination introduces materials that are not natively present in the TME and thus should be used carefully when attempting to emulate human tumors in vitro. The modification of cdECMs to react with activated alkynes has also been recently reported [95]. In this approach, the matrix was produced from human fibroblasts cultured in vitro, in 2D, using metabolic engineering to introduce azide groups into cell surface glycans. A specific culture medium containing a monosaccharide analog Ac4GalNAz (1,3,4,6-tetra-O-acetyl-N-azidoacetylgalactosamine) and sodium L-ascorbate was used. This sugar is absorbed by cells and incorporated into sialic acid residues, remaining in the matrix post-decellularization. The crosslinking process occurred through the presence of azide groups incorporated in the produced ECM and strained alkyne-containing moieties (dibenzocyclooctyne (DIBO)) immobilized in silicon wafers, without the need for using toxic catalysts (i.e., copper). The resulting biomaterials demonstrated to be cytocompatible, and suitable for promoting adequate keratinocytes adhesion on these surfaces. Importantly, it was demonstrated that azides incorporated in cdECM remained accessible even after decellularization and gentle homogenization (acetic acid 0.1% followed by addition of HEPES buffer (2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid, in PBS to adjust pH = 7.0). This enabled the formation of covalent bonds with activated alkynes on substrates functionalized with DIBO, resulting in a stable ECM coating without compromising keratinocyte cell adhesion. It remains however unclear whether the introduced modification interferes with biomolecular signaling features of the produced matrix [95]. Also, within the framework of cdECMs crosslinking strategies, an alternative approach based on the incorporation of a natural compound, Genipin, was also evaluated [96]. In this strategy, cdECM obtained from rabbit adipose stem cells (ADSCs), cultured in 2D, was decellularized, lyophilized, solubilized in an acid medium, digested with pepsin and subsequently crosslinked with Genipin. The physicochemical properties of the resulting hydrogel, as well as its cytocompatibility with L929 fibroblasts, primary anterior cruciate ligament fibroblasts, and rabbit-derived ADSCs, were analyzed. It was observed that the Genipin crosslinking significantly improved the mechanical properties of the hydrogel, raising its complex viscosity and reducing its degradation rate. In addition, the use of Genipin did not result in any type of cytotoxicity, maintaining cell viability levels comparable to other hydrogels. Thus, it was demonstrated that Genipin can represent a suitable natural crosslinking agent for cdECM hydrogels, providing a high crosslinking efficiency [96]. Nevertheless, further investigating different processing and crosslinking strategies for cdECM is still in high demand to allow for the fabrication of hydrogel matrices that more faithfully mimic the structural and biofunctional complexity of tissues.

3. Cell-derived ECM biomaterials in tumor modelling

Conventional tumor models remain insufficient to capture the full complexity of the TME, in large part because they fail to reproduce the biochemical composition, structural organization, and dynamic remodeling of the ECM in vivo. This limitation reduces the ability of these models to accurately predict tumor progression and therapeutic responses. Emerging evidence positions cdECMs as a promising strategy to bridge this gap, offering a biologically relevant alternative that more closely reflects tumor-specific matrix cues. Despite this promise, the majority of studies remain confined to 2D systems. In such approaches, cdECMs are typically generated under 2D culture conditions and most often derived from monocultures of fibroblasts or stem cells. Consequently, the resulting matrices lack the cellular heterogeneity and structural complexity characteristic of native tumors. Moreover, these matrices are frequently not tumor-derived, do not fully preserve tissue-specific biochemical and biophysical properties, and are rarely produced within a 3D architectural context. In addition, selecting an optimal culture medium remains particularly challenging. Different cell types require specific formulations, and this challenge becomes even more critical in coculture, where a medium suitable for one cell line may compromise the performance of another [97]. Thus, when the goal is to compare the ECMs produced by different cell types, it is equally essential to ensure that any observed differences do not result from variations in the composition of the culture medium, including FBS concentration or growth factors, which can influence ECM production [46,98]. Using a common medium can ensure experimental uniformity, but it does not guarantee that all cells are equally well adapted. In contrast, using lineage-specific media ensures optimal growth conditions, although it introduces the risk that media components may influence the composition of the deposited matrix. Thus, identifying a culture medium that adequately supports all cells and, simultaneously, does not interfere with cdECM production constitutes a significant challenge in the field. Nevertheless, even within these constraints, cdECMs have revealed potential to instruct cancer cells behavior. In fact, the biomolecular composition of the matrix has been shown to regulate invasive phenotypes, support metastatic progression, and promote therapeutic resistance.

3.1. Role in the modulation of cell behavior and tumor dissemination

The cdECM indeed plays a decisive role in regulating cell behavior, influencing processes such as adhesion, proliferation, migration, and metastasis. In this regard, several studies have explored the incorporation of cdECMs into in vitro models to gain a deeper understanding of key interactions happening in the TME. These studies have investigated not only the integration of cdECM into models, but also how different matrix compositions can influence cell behavior and tumor progression.

To evaluate the effect of different types of ECM on cell behavior, cdECMs were produced from 2D cultures of mesenchymal stem cells (MSCs), fibroblasts, osteoblasts, and MG-63 cells, a cell line isolated from the bone of a patient with osteosarcoma (Fig. 2Ai), using culture medium specific to each cell type. The cdECMs were produced on tissue culture polystyrene plates. Subsequently, the effect of different cdECMs on the adhesion and proliferation of MSCs and MG-63 was evaluated (Fig. 2A ii). A distinct effect of cdECMs was observed depending on the cell type. MSC proliferation increased on surfaces coated with cdECM when compared to TCPs. In contrast, MG-63 cell proliferation was inhibited in the presence of cdECM compared to TCPs. The effect on adhesion was similar in both cell types, being lower than that observed in cdECMs derived from normal cells, but higher than that observed in the presence of tumor cdECM. Regarding the effect of different cdECMs, there was an improvement in the adhesion and proliferation of MSCs and MG-63 cells when cultured on matrices derived from normal cells (MSCs, fibroblasts, and osteoblasts). This effect may be associated with the high fibronectin content present in these cdECMs, which is an essential component for promoting cell adhesion. In contrast, cdECMs derived from MG-63 cells had a high versican content, giving them anti-adhesive properties. In addition to the effects on cell adhesion and proliferation, phenotypic differences were also observed when MSCs and MG-63 cells were cultured in normal, tumor cdECMs, or TCPs alone (Fig. 2A iii) [74]. These results reinforce the importance of selecting a matrix that is specific to tumor tissue and tumor stage to more accurately replicate the in vivo scenario. Evidence suggests that dECMs derived from normal cells and those derived from cancer cells may play complementary roles in modeling distinct stages of tumor progression. Matrices derived from normal cells, generally rich in adhesion-promoting components such as fibronectin, may more faithfully reproduce the microenvironment characteristic of early-stage tumors, in which the stromal architecture remains relatively preserved and continues to support the initial expansion of malignant cells. Tumor derived dECMs, characterized by a profoundly remodeled composition [74,99,100] and high levels of anti-adhesive molecules such as versican, more closely resemble the dysregulated ECM typical of advanced-stage tumors, where progressive structural and biochemical disorganization contributes to more aggressive and resistant phenotypes. Thus, the strategic selection between matrices derived from normal or tumor cells may enable the bioengineering of stage-specific models, elucidating how the changes in ECM with disease progression can influence cellular behavior and, consequently, response to therapeutics [74].

Fig. 2.

A) Effect of different cdECMs, prepared by culture of MSCs cells, fibroblasts (fbs), osteoblasts (Obs), and MG63 in culture plates of treated polystyrene tissue (TCPS) (i), in the adhesion and proliferation (ii), and morphology (iii), of osteosarcoma (MG-63) and mesenchymal stem cells (MSCs). Adapted from Ref. [74]. B) Influence of the composition of decellularized extracellular matrix (dECM), obtained from normal fibroblasts and TGF-β1-activated fibroblasts, combined with adjustable stiffness, on the induction of epithelial-mesenchymal transition (EMT) in MCF10A mammary epithelial cells, adapted from Ref. [93]. C) Effect of different cdECMs, derived from fibroblasts and osteoblasts, on the role of a collagen receptor, Endo180, and the LOX enzyme to evaluate prostate cancer metastasis and consequent identification of possible therapeutic targets. BAPN: Inhibitor of LOX. Adapted from Ref. [75]. The image was created using PowerPoint, with icons from BioRender. Mano, J. (2026) https://BioRender.com/8v4wg7.

Still within the scope of studying the influence of the matrix on cellular behavior, another report produced cdECM in 2D culture and subsequently evaluated its integration into 3D tumor models. To this end, mammary fibroblasts were used to generate a cdECM representative of physiological tissue, and TGF-β1-activated mammary fibroblasts were used to recapitulate the composition of the breast TME. This study evaluated the impact of these matrices on cell behavior, particularly in the epithelial-mesenchymal transition, a central process in tumor progression. After production and decellularization, cdECM was conjugated with alginate to generate cdECM-alginate hybrid hydrogels [93]. The ability of alginate hydrogels to be mechanically tunable, combined with different cdECM compositions (tumor and non-tumor), allowed to evaluate different degrees of epithelial-mesenchymal transition in MCF-10A cells. It was observed that the presence of a cdECM with an aberrant composition, representative of a tumor context, associated with high hydrogel stiffness, promoted an epithelial-to-mesenchymal transition in MCF-10A cells, reproducing what occurs in breast cancer in vivo (Fig. 2B) [93]. These findings thus highlight the importance of both the composition and mechanical properties of the matrix in modulating cell behavior.

With the aim of clarifying the possible influence of the matrix on cell behavior modulation and metastasis, cdECMs derived from human fibroblasts and osteoblasts were produced in 3D cultures under controlled conditions (DMEM + 10% v/v FBS, 1 mM penicillin/streptomycin, and 2 mM L-glutamine for both cell types) in order to mimic visceral and bone metastasis of prostate cancer, respectively (Fig. 2C). Metastatic prostate cancer cells were then cultured on these matrices. The study evaluated the role of the Endo180 receptor and LOX enzyme activity in cell migration, being observed that Endo180 expression promoted the rounded phenotype and migration, while its inhibition eliminated this phenotype and abolished migration in both matrices. On the other hand, LOX inhibition revealed distinct behaviors, with migration maintained in osteoblast-derived ECM, while no cell migration was observed in fibroblast-derived ECM. It was suggested that this discrepancy may be related to the fact that ECM associated with osteoblasts, being mineralized, has sufficient stiffness to allow Endo180-dependent mechanotransduction and migration. In fibroblast-derived ECM, however, the reduction in stiffness resulting from LOX inhibition may not be sufficient to support these processes. The evidence indicates that matrix properties, namely its stiffness and composition, play a decisive role in modulating cell behavior and, consequently, in tumor progression [75]. However, once again, this study remains limited by the use of 2D models, which may condition the responses observed and differ significantly from what would be obtained in a 3D context.

Considering the mentioned reports, it is clear that the inclusion of the matrix in tumor models cannot be viewed solely as inserting a structural support. On the contrary, dECM acts as an active regulator and determinant of tumor phenotypes, and, although further studies are still needed in the field, there is already ample evidence that its presence and biological features significantly increase the ability of these models to better mimic the TME present in vivo.

3.2. Role in resistance to therapeutics

Beyond the matrix's impact on cell behavior and metastasis, further aspects warrant consideration, particularly regarding cells response to therapeutics. In fact, both the intrinsic properties of the matrix and its interactions with cellular components can favor the acquisition of more resistant phenotypes, reinforcing the importance of considering these factors in therapeutic screening trials.

The use of a tissue-specific tumor matrix is essential for a more accurate recapitulation of the in vivo TME. However, it is equally important to consider different stages within the same disease, as these can also influence tumor severity and therapeutic resistance. To elucidate the differences, present in cdECMs associated with different tumor stages, cdECMs were produced using three breast cell lines: a cancer cell line (MCF-7), a non-malignant cell line (MCF-10A), and a metastatic line (MDA-MB-231). These were co-cultured in 2D with normal human dermal fibroblasts (Nuff) to enhance ECM deposition. The cell lines were cultured in their specific medium and, in coculture, were seeded with NuFF cells in a 1:1 ratio, using a mixture prepared with equal volumes of NuFF medium and the specific medium for each cell line. In this context, marked differences in ECM deposition were observed, especially between co-cultures and isolated fibroblast cultures, highlighting the potential of co-culture to increase matrix production. Regarding the comparison between the different cdECMs, variations in the expression of essential ECM components were observed, namely the absence of collagen I in the ECM derived from MDA-MB-231-Nuff co-cultures and insignificant levels of laminin in the ECM derived from Nuff. In addition, differences were observed in the fibrils deposited in each ECM, particularly in their organization, and it is known that changes in matrix orientation and density may be associated with different levels of tumor severity. These changes reinforce the importance of including tissue- and tumor stage-specific cdECMs that mimic the desired conditions [76]. It should be noted that, in this study, the specific medium for each cell line was used, and in coculture conditions, a mixture of media proportional to the ratio of the cells. However, there is still no consensus in the literature regarding the most appropriate medium for cdECM-based models, particularly in cocultures, where differences in proliferation rates between cell lines may render the simple use of a proportional mixture inadequate. Furthermore, it is not guaranteed that such formulations do not interfere with matrix deposition, which reinforces this issue as a significant methodological limitation in the field. Based on the evident differences in the composition of cdECMs associated with different stages of breast cancer, a study was developed to understand how these variations can influence cell behavior. MCF-10A (benign), MCF-7 (non-invasive tumor), and MDA-MB-231 (invasive tumor) cells were cultured on TCPs using the same culture medium, DMEM/F-12 containing 10% fetal bovine serum, to produce different cdECMs. The performance of these matrices was compared with TCPs coated with fibronectin, TCPs coated with bovine serum albumin (negative control), and uncoated TCPs, when grown with MCF-10A, MCF-7, and MDA-MB-231 (Fig. 3A). All cell lines showed low adhesion to the BSA-coated substrate but adhered well to fibronectin and uncoated TCP. The proliferation of all cell lines was also higher on the fibronectin-coated and uncoated substrates. Matrices mimicking different stages of breast tumorigenesis showed greater adhesive activity than the negative control, although adhesion was lower in tumor matrices derived from MDA-MB-231 and MCF-7 compared to those derived from MCF-10A. MDA-MB-231 and MCF-7 cells exhibited increased proliferation in ECMs derived from MDA-MB-231 [77]. This finding appears to be associated with the high levels of LAMA3 and LAMA5 expressed in MDA-MB-231 cells, which encode subunits of laminins 332 and 511/521, respectively. These laminins preferentially bind to α3β1 and α6β4 integrins, triggering activation of the PI3K/Akt pathway, resulting in cell survival, cell cycle progression, and resistance to apoptosis [[102], [103], [104]]. This mechanism suggests that the laminin-rich matrix composition provides key biomolecular cues that maintain tumor growth and invasion, in contrast to the normal ECM, which tends to restrict proliferation to preserve tissue homeostasis. In turn, MCF-10A cell proliferation was stimulated only by matrices derived from MCF-10A cells themselves, possibly due to the expression of tenascin-C, an ECM glycoprotein that regulates tissue homeostasis [77,105]. As for cytotoxicity, this was evaluated by culturing MDA-MB-231 in different matrices, revealing greater chemoresistance to 5-fluorouracil (5-FU) in matrices derived from MDA-MB-231, while the response to doxorubicin did not vary between the different types of ECM. Together, this data indicates that the specific ECM of different stages, normal, tumor, and invasive tumor, can influence cell proliferation and adhesion, in addition to modulating the therapeutic response in a manner dependent on both the type of matrix and the drug tested [77]. It should be noted that, both in this study, in which a common medium was used [77], and in the previous study [76], which employed specific media, differences in ECM were observed between different tumor stages. Using the same method for producing cdECMs in TCPs, researchers engineered a colorectal cancer incorporating matrix derived from different colorectal cancer cell lines, HT-29 (high malignancy), SW480 (low malignancy), and CCD-841-CoN (benign), using the same culture medium for all cell lines, DMEM/F-12 containing 10% fetal bovine serum. These models were produced to evaluate the resistance of HT-29 and SW480 cancer cells to 5-FU. In this context, it was found that HT-29-derived ECM confers increased drug resistance, a phenomenon that appears to result from the upregulation of the ABCB1 and ABCC1 efflux transporters, as well as the activation of the Akt pathway. This interpretation is supported by the increased phosphorylation of Akt observed when HT-29 cells were cultured in cell-derived matrices, while ERK levels remained unchanged across the different stages of tumor-mimicking conditions. In contrast, SW480 cells did not exhibit significant changes in Akt or ERK phosphorylation across the multiple stages of tumor modeling. Additionally, the expression levels of ABCB1 and ABCC1 were approximately twice as high in HT-29 and SW480 cells when cultured in matrices derived from HT-29, compared to matrices derived from SW480 or CCD-841-CoN. This increase appears to be associated with the activation of c-Jun and the AP-1 complex, a transcriptional regulator known to promote ABCB1 expression, suggesting that HT-29-derived ECM favors drug efflux. Thes findings suggest that the Akt-mediated pathway, in conjunction with the regulation of the ABCB1 and ABCC1 efflux transporters, may represent a set of promising therapeutic targets for overcoming chemoresistance in highly malignant tumors. A more pronounced chemoresistance was observed in the presence of ECMs derived from more aggressive cancer cells, reinforcing the importance of screening candidate therapeutics in in vitro models cultured in matrices that are tumor-stage specific [106]. It should be noted that, although relevant, these results are based on a 2D-produced cdECM applied to equally generated 2D models, which limits their ability to accurately reproduce the structural and functional complexity of the TME in vivo.

Fig. 3.

A) Effect of different cdECMs derived from cells corresponding to different tumor stages of the breast tissue, particularly of cells MCF-10A (normal), MCF-7 (non-invasive), and MDA-MB-231, on cell adhesiveness and proliferation, as well as therapeutic resistance. Adapted from Ref. [77]. B) Optimization (i) and 3D production of cdECM from MSCs for integration into 3D models comprising colorectal cancer cells (ii) and comparison of their resistance to doxorubicin with collagen hydrogels and 2D cultures (iii). Adapted from Ref. [20]. C) Evaluation of the impact of tissue origin and degree of malignancy of cdECM biomaterials (i), as well as of the initial substrate where the matrix was deposited, and the resulting 5-fluorouracil (5-FU) chemoresistance (ii). Adapted from Ref. [101]. The image was created using PowerPoint, with icons from BioRender. Mano, J. (2026) https://BioRender.com/aq4xh1l.

Approaches for generating matrix in a more 3D context are highly desirable. Recently, a 3D matrix produced from human mesenchymal stem cells (hAMSCs) and subsequently incorporated into 3D models engineered with colorectal cancer cells. The matrix was generated using PLA microcarriers coated with fibronectin, to favor cell adhesion, and MMC agents were used to enhance matrix deposition (Fig. 3B). During the process, different amounts of microcarriers and MMCs were screened, and static cultures were compared with dynamic cultures, where shaking proved to be more advantageous because it promoted better nutrient diffusion and greater protein deposition, resulting in increased matrix deposition, particularly of collagen (Fig. 3Bi). After decellularization, cdECM structure remained intact, with well-defined pores corresponding to the space previously occupied by the microcarriers. The matrix showed predominance of type I collagen, type III collagen, and fibronectin, proteins normally characteristic of tumor ECM. This cdECM was subsequently recolonized with colorectal cancer cells (HT29 and SW480), allowing the creation of a living 3D tumor model (Fig. 3B ii). The models containing cdECM were then exposed to doxorubicin and compared with 2D cultures and collagen hydrogels. Greater resistance to the drug was observed in the presence of cdECM compared with the 2D models or collagen hydrogels, likely due to the physical barrier provided by the ECM. This denser and more complex 3D organization may limit drug diffusion and penetration, reducing its effective availability to the cells and thus contributing to the observed increase in chemoresistance [20,99] (Fig. 3B iii). This effect is associated with increased stiffness resulting from the deposition of ECM, which can activate various mechanotransduction pathways capable of converting physical stimuli into biochemical responses, thereby significantly modulating tumor behavior. Inactivation of the Hippo pathway in rigid matrices has been shown to promote dephosphorylation and nuclear translocation of YAP/TAZ, which, in complex with TEAD, induces programs associated with EMT, proliferation, and invasion. In parallel, integrin-dependent pathways, such as FAK/Src and Rho-ROCK, reinforce cytoskeletal organization, increase contractility, and may even suppress tumor immunogenicity through cGAS degradation. Matrix stiffness may also activate the EPHA2–LYN–ERK/RSK1 cascade, stabilizing TWIST1 in the nucleus and promoting EMT and metastasis [107,108]. However, it is important to note that these mechanisms are presented as potential explanatory pathways supported by the literature and have not been directly investigated or confirmed in the present study [20]. Thus, although the cdECM used was not yet tissue-specific, this study clearly demonstrated the feasibility of producing human ECM in a 3D context and of functionally integrating it into tumor models. Overall, the results showed that the presence of cdECM allowed for a more accurate recapitulation of the complexity of the TME, bringing these in vitro models closer to the conditions observed in vivo and reinforcing their potential as drug screening platforms [20].

In another study, a 3D tumor model aimed at reproducing the microenvironment of head and neck squamous cell carcinoma (HNSCC) more accurately was developed and explored to evaluate the antiproliferative effects of melatonin. To this purpose, cdECM was produced in vitro from the human fibroblast monolayer cultures, stimulated with ascorbic acid and proline to increase matrix deposition. After decellularization, cdECM preserved collagen, fibronectin, laminin, and several growth factors relevant to tumor progression. The cdECM was later combined with alginate and gelatin, resulting in a mechanically robust 3D hydrogel. In this hydrogel, the cellular component comprised tumor spheres of cancer stem cells (CSCs) derived from the cell line Cal-27, primary fibroblasts, and mesenchymal stem cells (MSCs). The simultaneous integration of matrix, CSCs, and stromal cells (fibroblasts and MSCs) enabled the creation of a complex 3D model capable of reproducing the essential components of the TME. In addition, the matrix supported cell adhesion, viability, and proliferation, also favoring the acquisition of a characteristic spherical phenotype. The effect of melatonin was evaluated in tumor spheres of isolated CSCs, in matrix-embedded CSCs spheres, and in the complete tumor-stroma model. Thus, it was found that melatonin exerted an antiproliferative effect both on isolated CSCs tumor spheroids and when incorporated into the matrix. However, in the presence of cdECM, higher doses of melatonin were required, demonstrating the protective barrier provided by the matrix, whose dense and organized structure limits the diffusion of the compound and reduces its availability in the TME. Thus, a smaller amount of melatonin effectively reaches the cells, diminishing the antiproliferative impact of the treatment. In addition to this physical effect, the dECM retains growth factors and cytokines, such as VEGF, PDGF, EGF, FGF, IL-4, IL-6, and HGF, which modulate the behavior of cancer cells and favor the maintenance of more resistant phenotypes. These bioactive components reinforce survival pathways and cellular plasticity, contributing to an environment that supports the persistence of subpopulations more tolerant to treatment. The addition of melatonin to the complete tumor-stroma model did not exert cytotoxicity on healthy stromal cells, reinforcing its potential as a safe therapeutic agent in more complex biomimetic contexts. However, although the oncostatic effects of melatonin in head and neck squamous cell carcinoma have been reported, the molecular mechanisms responsible for these effects remain poorly understood, underscoring the need for further research. In summary, this study demonstrated the ability to generate a complex and highly mimetic 3D model of the in vivo TME simultaneously integrating cancer cells, stromal cells, and a human matrix derived from primary fibroblasts cultured in 2D [94]. Despite being a notable advance, further developments regarding the production of cdECM in a tumor tissue-specific mode is highly desirable to better emulate the biochemical complexity of the matrix present in vivo.

Recently this was addressed in a report that evaluated the use of tissue-specific matrices, adjusted to the tumor stage. In this approach, different cdECMs were developed to evaluate how three variables influenced chemoresistance to 5-fluorouracil (5-FU), particularly the tissue origin of the cells that produced the matrix, the degree of malignancy of these cells, and the type of initial substrate where the matrix was deposited (Fig. 3C). For this, six cell lines were used, three representing breast tissue (MDA-MB-231, invasive tumor; MCF-7, non-invasive tumor; MCF-10A, benign) and three representing the colon (HT-29, invasive tumor; SW480, non-invasive tumor; CCD-841-CoN, benign), using the same culture medium for all cell lines, DMEM/F-12 containing 10% fetal bovine serum. In the first phase, these cells were cultivated in 2D on polystyrene substrates (Tissue culture treated polystyrene - TCPS), allowing the production of six distinct cdECMs. Subsequently, the invasive cancer cells MDA-MB-231 and HT-29 were then re-cultured in 2D, on each of the six matrices produced. The results showed that chemoresistance increased significantly when cells were grown on matrices derived from the same source tissue, showing a clear tissue-specific effect. Furthermore, within each tissue, the matrices produced by more malignant cells were those where re-cultured cells exhibited higher resistance to 5-FU. This finding demonstrates that both the tissue origin and the tumor stage of ECM significantly influence the therapeutic response (Fig. 3C i). It has already been shown that chemoresistance increases with the progression of tumor malignancy, in part due to the activation of pathways associated with the epithelial-mesenchymal transition (EMT) and the expression of transporters. This study adds that ECM remodeling constitutes another mechanism associated with this progression. However, the details of how matrix remodeling contributes to the acquisition of chemoresistance were not explored in the study in question. In the second phase, matrices derived from the MDA-MB-231 and HT-29 cell lines were also produced on three different types of substrates: uncoated TCPs, PMEA-coated TCPs, and PTHFA-coated TCPs. These materials were selected due to their different protein adsorption capacities, which influence how cells adhere and deposit the ECM. Upon reculturing the invasive MDA-MB-231 and HT-29 cell lines in 2D, the cdECMs produced on PTHFA were found to confer greater chemoresistance in both cell types. This effect is likely due to two mechanisms. First, it may be associated with increased ECM deposition on this substrate and greater availability of fibronectin binding sites, which enhances the activation of integrin-dependent signaling pathways and, consequently, promotes increased cell survival and proliferation (Fig. 3C ii). Second, the activation of integrin signaling, particularly focal adhesion kinase (FAK), may trigger the expression of genes encoding fibrotic ECM proteins, whose accumulation reinforces the intracellular signaling associated with resistance (Fig. 3C ii). It should be noted that, despite our understanding of these potential mechanisms, the specific processes responsible for the increased chemoresistance in PTHFA remain to be elucidated. In summary, current evidence indicates that not only cell origin and malignancy, but also the initial substrate, play a determining role in the functionality of cdECMs and tumor models' response to chemotherapeutics [101]. In addition to ensuring a tissue-specific ECM adjusted to the tumor stage, the selection of in vitro matrix production methodologies, namely the use of 2D or 3D in vitro fabrication approaches is key as it can substantially alter cdECM components and consequently influence bioengineered 3D models response to therapeutics.

4. Outlook and future perspectives

Reconstructing the multitude of cellular and non-cellular components of the native TME in vitro remains a central challenge in cancer modelling. In vitro-produced human cdECM is emerging as an increasingly viable ECM-mimetic biomaterial for bioengineering tumor-relevant 3D models. The in vitro production of tumor cdECM is highly modular and intrinsically dependent on cell source selection, biochemical stimulation, and mechanical conditioning. This dynamic process shifts ECM production from a descriptive extraction process to a highly adjustable production approach that can ultimately be tailored to better mimic the matrix present in the in vivo TME, ideally in a tumor-specific manner.

This conceptual transition is particularly relevant given the critical limitations associated with widely used animal-derived matrices, such as basement membrane extracts (BMEs (e.g., Matrigel). Although these extracts have been instrumental in establishing early 3D tumor models, their animal origin and poorly tunable mechanical properties, raise significant ethical and scientific concerns. Moreover, these conventional ECM-mimetic biomaterials largely reflect the biomolecular content of their tumor tissue of origin rather than a tumor-specific biomolecular profile, acting as a generic “one-fits-to-all“ matrix. In this context, cdECM may provide a more human-relevant alternative that drastically reduces reliance on animal sourcing and offers highly tunable biochemical and biophysical properties dictated by selected cells and tuned culture conditions [109]. Batch-to-batch variability in BMEs is a prevalent aspect, thus it will be increasingly relevant to also address potential batch-to-batch variability aspects in cdECM biomaterials. Importantly, batch variability has been mostly characterized with regard to protein contents (e.g., using proteomics-based analysis), additional elements (i.e., glycosaminoglycans, lipid or extracellular vesicles), and cdECM strutral properties are highly desired to be additionally characterized, so that batch variations could be further framed as resulting from a multi-parametric characterization workflow. Efforts to characterize potential batch-to-batch variations in terms of cdECM preparations obtained from different batches of the same donor, and batches of different donors are already being reported [110]. In the foreseeable future, as the technologies for cdECM production and analysis evolve, comprehensive chracterization approaches will become increasingly valuable for identifying and understanding sources of production variability.

To unlock the full potential of cell-derived matrices, efforts are also increasingly shifting from producing these in conventional 2D cultures toward opting for 3D cell culture systems. Within these 3D environments, cells exhibit more physiologically relevant phenotype, establish suitable cell-cell contacts, as well as present distinctive gene expression profiles, supporting an enhanced ECM deposition, when compared with traditional 2D cultures. Yet, producing cdECM at this dimension introduces additional challenges. While in vitro 3D matrix production strategies (i.e., microcarrier-based, etc.), better recapitulate spatial tumor geometry, the establishment of oxygen and nutrien-diffusion gradients and necrotic regions formation are to be considered. Moreover, downstream processing poses a major bottleneck. Achieving thorough decellularization within dense 3D tissues may require specific chemical or physical approaches which potentially impact the structural integrity and the biomolecular content/activity of cdECM biomaterials. Scaling up these 3D culture systems while maintaining high batch-to-batch reproducibility also remains technically demanding [16,17]. Tumor ECM is not simply a product of one cell type, but the outcome of dynamic crosstalk among cancer cells, fibroblasts, immune cells, endothelial cells, and systemic factors [111]. While stromal and/or cancer cells cultured in vitro can deposit tumor-relevant components, the resulting matrix may reflect a simplified and relative reductionist snapshot of a far more complex microenvironment. Expanding these systems to include immune cells, although not primary producers of dECM, could be key for recapitulating immune–matrix–tumor interactions that are recognized to influence the TME [112]. Bioengineering multicellular 3D tumor models for cdECM production also introduces an additional challenge associated with the formulation of culture media capable of simultaneously supporting the distinct biological requirements of cancer, stromal, and immune cell populations while minimizing unintended effects on ECM deposition in vitro. Achieving this balance remains an important production bottleneck, irrespective of whether the models are established from commercially available immortalized cell lines or patient-derived cells. The use of patient-specific cells could potentiate the bioengineering of tumor models that recapitulate the disease stage and physiology in a more personalized mode. In this context, achieving ‘tumor-stage-specific’ cdECM formulations is envisioned as a possible milestone, specifically if one envisions a design strategy where primary cells isolated directly from patient biopsies at precise clinical stages (such as early-stage localized lesions versus advanced metastatic niches) are used to fabricate patient-specific tumor ECM in vitro. This could potentially allow cdECMs to shift to a clinically correlatable biomaterial. Importantly, maximizing the translational utility of these patient-specific matrices requires not only optimization of their production, but also innovation in how they are physically processed and utilized in vitro. Moving beyond intact 3D scaffolds, the formulation of cdECM into alternative physical states unlocks entirely new methodological avenues. Fully capturing the dynamics of disease progression introduces additional structural and microenvironmental demands, particularly the ability to reproduce spatial organization, biochemical gradients, and dynamic flow conditions. In this context, microfluidic organ-on-chip technologies further expand the available toolbox to generate or process dECM biomaterials [113]. Integrating patient-specific cdECM with matched tumor, stromal and immune cells within these platforms can bring about significant advances, especially considering that cdECM production and use on-chip has already been reported [29]. It is envisaged that coupling this in-chip production with cells re-seeding and drug screening in dynamic flow conditions will be increasingly prevalent in future developments in the field.

Overall, the potential value of cdECM may lie not in achieving perfect mimicry but in enabling controlled and programmable reconstruction of tumor-specific microenvironmental features when different components are present. Cell-derived ECM has not yet reached its full biological potential, but if the field can address current limitations in standardization, scalability, and rigorous characterization, these biomaterials may become a foundational tool for next-generation 3D tumor modeling, drug screening, and fundamental oncology research.

CRediT authorship contribution statement

Patrícia S. Rodrigues: Data curation, Formal analysis, Methodology, Writing – original draft. Margarida Henriques-Pereira: Data curation, Formal analysis, Writing – original draft. João F. Mano: Conceptualization, Funding acquisition, Supervision, Writing – review & editing. Vítor M. Gaspar: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Gaspar VM reports financial support was provided by Foundation for Science and Technology. Gaspar VM reports financial support was provided by Horizon Europe. Gaspar VM reports a relationship with Cellularis Biomodels that includes: board membership. Mano JF reports a relationship with Cellularis Biomodels that includes: board membership. The corresponding authors, Gaspar VM and Mano JF, are Editors of Materials Today Bio Journal. Given their role as editorial board members, they had no involvement in the peer review of this article and had no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to another journal editor. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

No data was used for the research described in the article.