Cell–3D Matrix Interactions: Recent Advances and Opportunities

Abstract

Tissues consist of cells and their surrounding extracellular matrix (ECM). Cell-ECM interactions play crucial roles in embryonic development, differentiation, tissue remodeling, and diseases including fibrosis and cancer. Recent research advances in characterizing cell-matrix interactions include detailed descriptions of hundreds of ECM and associated molecules, their complex intermolecular interactions in development and disease, identification of distinctive modes of cell migration in different 3D ECMs, and new insights into mechanisms of organ formation. Exploring the roles of the physical features of different ECM microenvironments and the bidirectional regulation of cell signaling and matrix organization emphasize the dynamic nature of these interactions, which can include feedback loops that exacerbate disease. Understanding mechanisms of cell-matrix interactions can potentially lead to targeted therapeutic interventions.

New advances in research on cell-matrix interactions

The interactions of cells with the 3D extracellular matrix (ECM) (see Glossary) play crucial roles during cell migration and organ formation in embryonic development, differentiation, adult tissue homeostasis, and pathogenesis of diseases such as fibrosis and cancer. This research field has expanded explosively after the discovery of many new ECM molecules and their cell surface receptors. Our goal in this brief review is to highlight recent conceptual and experimental advances that should provide exciting opportunities for future research into cell-3D matrix interactions.

Diversity of cell-3D extracellular matrix interactions

A recent development in the ECM field starting in ~2019 has been widespread adoption of the term “matrisome” – i.e., the molecules comprising the extracellular matrix as it changes during embryonic development, organ differentiation, and disease pathogenesis. This holistic concept of matrisomes moves beyond classical studies focusing on a single protein or protein family to include not only classical ECM structural proteins, such as the collagens, elastin, proteoglycans, and fibronectin, but also matrix-associated enzymes and their inhibitors, matrix-bound growth factors, and in some cases the receptors for ECM molecules [1,2]. Among many examples, recent studies have used matrisome analyses to characterize different basement membranes (https://bmbase.manchester.ac.uk/), to discover that ECM-associated genes have more genomic alterations and mutations than other genes [2], to identify thrombospondin and tenascin links to collagen alignment in breast cancer [3], and to characterize new bioengineered models of human pancreatic cancer [4].

Many publications on the matrisome are still at a descriptive level. There are also considerable overlaps between matrisome components and the “adhesome,” which comprises molecules associated with cell-matrix adhesions, especially focal adhesions (e.g., see www.adhesome.org). Both of these “omics” approaches provide major opportunities for applying increasingly sophisticated methods to understand embryonic development and diseases involving changes in networks of ECM molecules rather than alterations in just a few selected proteins as in the past. Unexpected findings may arise in terms of new functions for groups of proteins regulating other matrisome components. Exemplifying such crosstalk, the planar basement membrane and two of its biochemically unrelated constituents, laminin and collagen IV, can strongly regulate the assembly of a major fibrillar ECM component, fibronectin, in a variety of cell types [5]

We now know that modes of cell migration in 3D microenvironments can vary widely depending not only on the cell type and biochemical composition of the ECM, but also on the physical and mechanical properties of the ECM [Figure 1]. Multiple mechanisms of 3D cell migration have been characterized recently that range from classical lamellipodial migration characteristic of mesenchymal fibroblast-like cells to rounded amoeboid cell migration of immune and certain cancer cells [6,7], or to lobopodial and multiple other modes of migration in cross-linked linearly elastic or spatially confined microenvironments involving intracellular pressure, cortical actin flow, ion fluxes, and other mechanisms [8–12]

Figure 1.

Biochemical or mechanical properties of the extracellular matrix governing cell-ECM interactions. (A) The specific molecules comprising the ECM can modulate cell behavior, e.g., via biochemical signals from collagens, elastin, fibronectin, proteoglycans, etc. (B) The concentration of each ECM component can alter cell interactions, e.g., different densities of collagen or proteoglycans. (C) Topology of the ECM can guide different responses, e.g., when organized as 1D fibrils, 2D basement membranes, or 3D fibrillar matrix; in addition, the architectural organization of ECM can guide cell interactions, e.g., by the orientation of collagen fibrils. (D) ECM can be crosslinked, which can strongly alter cell interactions by altering matrix stiffness, viscoelasticity, and pore size. (E) The pore sizes between ECM molecules can affect cell migration and molecular diffusion. (F) ECM stiffness or compliance can have major effects on cell adhesion formation and stem cell differentiation. (G) Gradients of ECM components can alter cell migration and differentiation, whether involving structural molecules or embedded molecules such as growth factors. (H) Matrix can have distinctive elasticity – linear elastic in analogy to springs or rubber bands, non-linear elastic in which applying different levels of stress can result in non-linear changes in ECM elasticity, or plasticity and viscoelastic behavior. (I) ECM can often be remodeled by proteases or post-translational modifications, resulting in altered topology, pore size, or elasticity. (J) The ECM can also provide confinement and squeezing of cells, e.g., cells that migrate into dense, aligned collagen or cells becoming encased by increasingly dense or fibrotic ECM.

The microarchitecture of the matrix can affect not only cell adhesions and migration, but also differentiation [13,14]. For example, the microarchitecture of fibrillar collagen networks, i.e., fiber thickness and pore size, can regulate the differentiation of adipose stromal cells toward myofibroblast differentiation in a process independent of the overall stiffness of matrix previously known to regulate stem cell differentiation [14,15].

Matrix physical properties

The local physical and biochemical features of the surrounding microenvironment of cells affect their migratory speed and directionality (Figure 1). For example, cells show differences in migration depending not only on matrix molecular composition, but also on its elasticity and viscoelasticity [15]. Examining whether an extracellular microenvironment is soft or stiff as in numerous previous studies [16,17] should now ideally be complemented by analyses of cell behavior in environments of differing viscoelasticity. The reason is that biological matrices and proteins are often viscoelastic, i.e., they display a mechanical combination of the viscosity of a thick fluid and elasticity that attempts to return the material to its original form or organization after a deforming force is released. This combination of properties results in plastic deformation with slipping, creep and “stress relaxation” of the ECM in response to being deformed, often without returning to its original form. Viscoelasticity of the ECM can be modulated by the extent of inter-molecular crosslinking [15] The complex mechanical property of viscoelasticity can have major effects on cell behavior, although the underlying molecular mechanisms are not yet clear. For example, altered viscoelasticity and tissue rheology can regulate filopodial versus lamellipodial protrusions from the leading edge of a cell, rates of cell spreading on a surface and overall speed of migration, and more complex processes including stem cell morphogenesis, epithelial-mesenchymal transition, cancer cell invasion, and fibrosis [18–24].

The fact that the mechanical and chemical nature of the surrounding matrix can substantially alter cell migration mode, speed, and directionality means that one cannot simply “work in 3D” due to these many matrix factors (Figure 1). An ongoing challenge will be to generate 3D matrix models that more accurately mimic specific in vivo microenvironments or to use ex vivo tissues. Development of such more-physiological microenvironments than simple collagen and laminin gels or current synthetic hydrogels will be valuable to provide more accurate insights into the mechanisms of cell migration and tissue remodeling, as well as platforms for testing translational approaches.

Mechanotransduction

Different physical properties of the ECM elicit distinct mechanical responses from cells. 2D studies revealed that cells can test the properties of the underlying matrix by repetitively probing with their focal adhesions [25], analogous to hikers testing their footing when crossing unstable terrain. Cells can sense and respond to ECM stiffness and ECM-transmitted forces such as tension or stress [15]. Cell mechanosensing involves integrin-based focal adhesions in the process of mechanotransduction, where actomyosin-mediated contractile forces are transmitted to and from ECM substrata [26,27]. In 3D environments, mechanotransduction at the focal adhesion level in response to matrix stiffness resembles cellular responses under 2D conditions –stiffer substrates stabilize adhesions while adhesions on softer, flexible substrates have shorter lifetimes and faster dynamics [13].

Mechanotransduction becomes more complex for cellular forces associated with cell migration, particularly concerning the 3D migration cycle. Recent evidence from fibroblasts migrating in 3D collagen reveals that prior to translocation, cells initially deform (pre-strain) collagen fibrils, increasing self-generated tension by contracting and transmitting it to the ECM – essentially first “pulling up the slack in the rope” (i.e., collagen fibrils [28]). Interestingly, although epithelial cancer cells often display relatively equal-and-opposite strain transmission in anterior and posterior directions during 3D migration (Figure 2A) [29], fibroblasts and mesenchymal (non-epithelial) cancers can display constant anterior pre-strains two-fold greater at the front than at the rear of the cell, suggesting a disconnect between strain propagation at the front and rear of the cell (Figure 2A) [28]. This anterior pre-strain by fibroblasts is likely genetically primed for a contractility-centric mode of migration, with higher expression of contractile molecules (e.g., myosin II) and enhanced integrin-based adhesion to the microenvironment. The sequence of events during 3D migration is also distinctive, with anterior actomyosin contractions preceding leading edge protrusive activity, suggesting that mechanotransduction in 3D helps to establish a unique mesenchymal 3D cell migration cycle distinct from the classical 2D cell migration cycle (Figure 2B–C) [28]. Discrepancies between 2D and 3D migration cycles in different cells highlight the importance of testing numerous cell types and matrix conditions.

Figure 2.

Mechanotransduction of contractile forces in 3D microenvironments during cell migration. (A) Schematic showing the differences in ECM surrounding single cells migrating directionally in 3D collagen gels. Mesenchymal fibroblasts (left) that have high myosin II expression and extensive integrin ligation during migration generate large anterior ECM strains 2-fold larger than at the cell rear. Similar attributes are seen in fibrosarcoma cancer cells, while the majority of epithelial cancer cells (right) generate smaller, transient ECM strains at the front and rear, and they have lower myosin II expression and integrin ligation (see graph summarizing the latter concept, which still requires further testing). (B and C) This schematic shows a fibroblast migrating directionally through a 3D collagen matrix. Yellow arrows depict the local directional contractile forces applied to the matrix, and magenta arrows indicate the relative summed forces in a given region. During a 3D migration cycle (C) a retrograde pull stabilizes adhesions (gray ovals) at the leading edge. A contralateral anterograde pull (increased force in the direction of migration) leads to a pinching of the matrix, which is followed by an increase in leading edge protrusion (dashed white line).

In addition, the type of ECM (elastic versus nonlinear elastic versus viscoelastic, etc.) alters mechanotransduction. Recent studies suggest that viscoelastic ECMs that mimic in vivo 3D matrix properties elicit distinctive adhesion formation on 2D and within 3D matrices [15]. Viscoelastic ECMs may be more effective at eliciting a mechanical cellular response compared to elastic matrices [15]. Future research should evaluate whether protein dynamics are altered within focal adhesions interacting with ECMs with different properties, as well as whether viscoelasticity changes the nature of cell adhesions and cell migration modes in in vitro and in vivo 3D microenvironments.

The nucleus in 3D cell migration

Some cells can use a “nuclear piston” mode of 3D cell migration in which the nucleus is pulled anteriorly by actomyosin contractility to pressurize the anterior of a migrating cell to drive a “lobopodial” cell protrusion forward [30]. In confining hydrogels that are viscoelastic (plastic), mechanosensitive ion channels in the anterior protrusion generated by the nuclear piston respond to the elevated hydrostatic pressure by triggering ion channel activation; the resulting influx of sodium and calcium ions enhances intracellular osmotic pressure and provides additional forces for extending the protrusion and promoting efficient cell migration [31].

Another intriguing recent finding is that cells migrating in 3D collagen environments can use their stiff, bulky nucleus as a ruler to help guide their choice of a direction for migration towards a wider, more readily traversed passageway [32,33]. Besides serving as a ruler, the nucleus in conjunction with its associated cytoskeleton can also play roles in mechanosensing – in this function, the nucleus functions as an elastic deformation gauge to activate signaling and epigenetic pathways (Figure 3) [8,34–36].

Figure 3.

Multiple functions of the cell nucleus during cell migration, mechanosensing, and mechanotransduction. The nucleus, as the largest cell organelle, can function as a ruler to test entry of migrating cells into narrow spaces or channels. A nucleus can also be pulled anteriorly by actomyosin contraction, thereby increasing hydraulic pressure at the front of a cell and promoting cell protrusions termed lobopodia. The nucleus can also serve as a stress sensor by responding to local confinement. Finally, a nucleus can also function as a signal transducer by initiating signaling or altered gene expression.

Cancer cell – ECM dynamics

Cancer cell invasion continues to be another very active field of investigation using various 3D in vitro and in vivo models of ECM. Before malignant cells can invade the interstitial matrix and tissues, they must usually breach the basement membrane barrier that surrounds epithelial tissues (Figure 4). Although cancer cells use proteases to degrade basement membranes during invasion, protease-independent breaching can occur. Physical protrusion of a cellular extension can penetrate and expand holes in basement membranes without proteases in a process using local ATP production in C. elegans [37]. Human cells can also use non-proteolytic physical protrusive forces to penetrate the basement membrane using repetitive probing by microspikes that widen into filopodia for enlarging basement membrane perforations; these protruding filopodia subsequently probe fibrillar collagen ECM [38–40]. Cellular metabolic activation is also important for successful invasion, associated with bidirectional crosstalk between the ECM and cellular metabolic pathways [37,41–43].

Figure 4.

Different mechanisms for breaching the basement membrane. Cancer cells or transiently invasive cell during embryonic development can breach basement membrane barriers using differing mechanisms. Cells can use proteases, such as matrix metalloproteinases (MMPs) to degrade the basement membrane locally, e.g., using MMPs on the tips of invasive processes termed invadopodia. Even if proteases are inhibited, some cells can expand tiny holes or perforations in the basement membrane for breaching and invasion in a process requiring energy production (orange shading) and force (yellow arrows) to push the basement membrane laterally to form a ridge around an expanding hole (red mounds). Cells can also use both processes, i.e., both proteolysis and force to expand basement membrane perforations.

Although generation of force to breach basement membranes is thought to involve actomyosin contractility, the relationship between levels of cellular non-muscle myosin and cancer invasiveness is complex. Some cancer cells have elevated myosin II isoforms or actomyosin-associated proteins as predicted to facilitate breaching the basement membrane or migratory invasion, yet others have decreased myosin II consistent with function as a tumor suppressor [28,44,45]. Similar bi-functional complexity is seen for myosin-X in cell-ECM interactions, both enhancing invasive cell migration yet suppressing breast cancer cell invasion by modulating basement membrane assembly [46].

Although myosin II contractility is an obvious source of forces for disrupting basement membrane structure to promote breaching, another contributor may be the Arp2/3 complex generation of lateral actin branches to increase local actin density and provide extension forces for a cellular protrusion to breach the basement membrane [47]. Crosstalk between these two major cytoskeletal systems (myosin II and Arp2/3) can govern the mode of cell migration [9]. Multiple opportunities in this research area include more precisely evaluating the mechanisms used by different tumor cell types for initial membrane breaching and then for subsequent squeezing the large, stiff cell nucleus through a basement membrane perforation, and then using collective- versus single-cell migration for invasion in vivo [48].

A pattern of aligned ECM collagen fibrils outward from human tumors correlates with poor patient survival from breast cancer [49], yet the source and mechanisms of such collagen alignment and its role in facilitating local cancer migration remain to be further clarified. One possibility is that in analogy to non-cancerous fibrosis, fibroblasts (especially cancer-associated fibroblasts and potentially myofibroblasts) can generate contractile forces that can affect cell migration/invasion either directly by aligning collagen fibers directed outward from tumors to guide collective or single-cell migration [50,51], or by altering matrix composition [3]. Moreover, this alignment of collagen by cancer cells can promote the outward diffusion of exosomes that can stimulate the generation of cancer-associated fibroblasts [52], which could further potentiate this 3D collagen alignment process. Because different types of cancer differ widely in their modes of metastatic dissemination, mechanisms of cell-ECM remodeling likely also differ.

Another feature of the ECM to which both normal and cancer cells respond is termed durotaxis [53], in which cells migrate toward stiffer ECM regions [53,54]. The efficiency of durotaxis depends on the initial local ECM stiffness: cells in a very soft environment respond most effectively to stiffness gradients [54,55]. This responsiveness to local mechanical stiffness of tissues may provide clues in the future not only for how malignant cells invade or disseminate, but also for how to attain appropriate tissue positioning of therapeutic stem or progenitor cells for cell-based regenerative medicine. While most durotaxis studies have focused on single cells on 2D substrates, evidence from 3D models establish that long-distance tension sensing is likely involved in cell migration of cancer cells [56]. Future studies should focus on how cell-ECM and cell-cell tension within collectively invading cells are involved in metastatic progression.

Cell – matrix feedback and reciprocal feedback loops

A well-accepted current concept is that of bi-directional interactions between cells and their surrounding extracellular matrix (e.g., [57]). Cells both directly modify the ECM and respond to ECM stiffness, forces, and other physical inputs. For example, the actomyosin cytoskeleton of cells stiffens in response to a locally stiff external microenvironment [58], yet it is also cells that generate the ECM microenvironment. When cells form focal adhesions to a matrix substrate, the normal rearward flow of F-actin (termed actin retrograde flow) slows as intermolecular complexes assemble, resulting in a combination of rearward forces on the focal adhesions and the attached ECM, as well as a stimulation of forward cell protrusion at the leading edge, both of which promote cell migration (Figure 5). A stiffer substrate can promote both this formation of a clutch-like molecular complex and contractility of myosin II [59].

Figure 5.

Schematic representation of dynamic 3D ECM-cell-actin cytoskeletal interactions. (A) Actin polymerization at the leading edge (righthand side) of a cell produces “pushing” forces that protrude the plasma membrane (light blue arrow) along with the rearward translocation of F-actin filaments in the process of actin retrograde flow (light purple arrows). F-actin side branches are generated by activity of the Arp2/3 complex. (B) Formation of focal adhesions (FA, red boxes) to the matrix substrate slows actin retrograde flow locally in a clutch-like manner, resulting in enhanced anterior protrusion of the plasma membrane (blue arrow) and rearward forces on the local matrix. These effects increase with increasing stiffness of the substrate. (C) ECM barriers to cell protrusion and migration by increased 3D ECM density and stiffness are predicted to enhance activity of the Arp2/3 complex, resulting in increased numbers of actin side branches, higher overall F-actin density, and increased protrusive force at the leading edge of the cell to push back against the obstruction. Box: In general, increased stiffness of the ECM results in increased stiffness of cells and their cytoskeleton, but increased cellular actomyosin cytoskeletal activity can produce increased tension and stiffness of the 3D ECM.

If cells encounter resistance to migration due to local dense, stiff ECM that produces compressive forces at the leading edge of cells, recent in vitro studies predict resultant changes in actin polymerization. Increased membrane tension and mechanical load at the leading edge are known to stimulate the formation of lateral F-actin branches due to Arp2/3 activity, resulting in higher overall actin density that permits cells to push back against impediments [47,60].

In human diseases, deleterious self-enhancing “feed-forward” mechanistic loops appear to contribute to worsening of disease processes (Figure 6). For example, fibrosis involves increased cellular production of extracellular matrix that produces a stiffer local microenvironment, but that stiffness can itself induce cellular responses that further worsen this condition. These effects can be seen in fibrotic conditions [61–63] where a progressive feed-forward loop has been implicated in idiopathic fibrosis of the human lung. A similar process may occur in cancer, which induces stiff desmoplasia and where ECM stiffness can promote tumor progression, e.g., by selective transcription factor induction and microRNA suppression [16,64] and alignment of collagen fibers. In addition, however, invading cancer cells can also degrade and remodel the ECM – inhibiting proteases can restore a more normal mode of nuclear piston migration to certain malignant cells [30]

Figure 6.

Positive cell-matrix feedback during cancer metastasis, skin scarring, and epithelial elongation. Cancer cells and cancer-associated fibroblasts can generate contractile forces to increasingly align collagen fibers which can direct collective or single-cell migration and promote cancer metastasis. During skin wound healing, increasing tension on the wound can promote the conversion of Engrailed-1-negative fibroblasts (blue cells) to Engrailed-1-positive fibroblasts (red cells) with enhanced nuclear YAP expression. This induces a fibrotic response, thereby generating more traction and depositing more collagen, leading to skin scarring. During epithelial remodeling embryonic morphogenesis, epithelial cell aggregates can break symmetry and exert local strain on collagen fibrils. The polarized collagen can feed back to the cells to promote local cell proliferation and direct elongation of epithelial aggregates for tissue remodeling.

YAP (Yes-associated protein) and TAZ (transcriptional coactivator with PDZ-binding motif) are transcriptional regulators that act as intracellular mechanosensors responding to the mechanical signals that cells receive from the surrounding ECM. Compared to normal tissue, fibrotic tissue has more prominent nuclear expression of YAP/TAZ, which seems to occur early after organ injury and is sustained during fibrogenesis. Inhibition or knockout of YAP/TAZ can attenuate fibrosis in organs such as lung, liver, and kidney [65]. Mechanistically, YAP/TAZ activation can induce a pro-fibrotic response in fibroblasts, thereby generating more traction, depositing more collagen, and eventually promoting tissue fibrosis (Figure 5). Interestingly, a recent study found that inhibition of the YAP pathway can prevent this tension-associated fibroblast transition, reducing skin fibrosis and permitting effective skin regeneration [66]. These findings suggest that tissue regeneration without scarring may be possible by preventing this mechanically driven fibrotic response.

Paradoxically, however, the YAP/TAZ pathway is also known to induce tissue regeneration in multiple organs (e.g., [67,68]), avoiding elevated mechanical tension that can eventually promote fibrosis [69]. Consequently, the bidirectional functions of YAP/TAZ may require precise tuning. Inhibition of YAP/TAZ function in mesenchymal cells can reverse in vitro extracellular matrix stiffening and in vivo tissue fibrosis in lung and liver in mouse models [70]. To summarize, an appropriate balance appears to be critically important in the reciprocal regulatory loops involving 3D matrix stiffness/tension on cells linked to the positive and negative roles of the intracellular YAP/TAZ pathway in tissue regeneration and fibrosis.

A normal positive-feedback process can drive collagen alignment and directional epithelial elongation in a model of morphogenesis in vitro (Figure 5). Cells proliferating isotropically in non-polarized collagen can break symmetry of cell aggregates, exerting local strain on collagen fibrils to produce low-level polarization. This initial collagen polarization can feed back to the cells to promote local cell proliferation and elongation of the epithelial aggregate along the oriented collagen. This process can continue in a feed-forward fashion so that collagen, cell proliferation, and elongation of the epithelial cell aggregate can increasingly extend and further bundle and polarize the collagen, which in turn promotes more epithelial extension [71]

Interestingly, negative feedback can also modulate matrix organization. An in vitro model of wound closure revealed that the degree of ECM alignment can negatively regulate matrix assembly [72]. Specifically, greater ECM alignment suppresses the fibronectin matrix assembly characteristic of the early provisional matrix during wound repair, at least in vitro [72]. Additional research to determine the mechanisms of these feed-forward and other mechanistic regulatory loops, and especially exploring how to interrupt them by targeted therapy, will provide intriguing future challenges.

ECM–cell biophysics and modeling

Biophysics, physics, and computational modeling are becoming major tools for understanding the mechanisms of cell-ECM interactions and especially pathophysiology. Selected examples include fibroblast signaling in fibrosis, fibroblast regulation of macrophage migration, breast cancer invasion, and budding and branching morphogenesis in embryonic development [50,68,73–75]. In all these biological phenomena, cell interactions with an adjacent ECM induces signaling responses. Associated changes in cell-cell adhesion can lead to cancer cell invasion, formation of buds, or epithelial branching. These processes have been modeled by applying physical principles and widely different types of computational modeling.

In mechanically driven fibrosis, myofibroblasts can generate contractile force that is transmitted to other fibroblasts through collagen fibers, resulting in tensile force-induced fibroblast activation. This process can involve Piezo1, integrin, and calcium signaling that can be quite rapid and long-range (less than 1 second and at a distance of 70 μm) [68]. In vitro and in silico experiments combined with mathematical modeling demonstrate that this 3D matrix mechanical signaling by myofibroblasts can promote expansion of fibrotic tissue into neighboring normal tissue. Model studies predict that blocking this myofibroblast-fibroblast crosstalk can slow the expansion of fibrosis [68].

Interestingly, myofibroblasts can also transmit long-range 3D mechanical signaling to macrophages located hundreds of microns away to regulate their migration. In this process involving α2β1 integrins and stretch-activated ion channels, modeling reveals that it is the rate of strain/displacement of collagen rather than altered stiffness/durotaxis that mediates this response [50].

In studies of mammary tumor cells, the extent of ECM confinement of the cells and their strength of cell-cell adhesion function cooperatively to govern whether cells in a tumor are immobilized in a physical state termed “jammed” or can transition to move more fluidly or even disperse into single-cell invasion. These different states can be predicted using an in silico lattice-gas cellular automaton model incorporating the concept of jamming (immobile) or unjammed migration to model both collective and individual cell migration [73].

During organ formation involving budding morphogenesis, adhesion to the surrounding basement membrane combined with a programmed loss of cadherin-based cell-cell adhesion in the outer layer of cells plus cell proliferation can generate numerous buds. This process can be analyzed by applying concepts of free energy differences underlying the sorting out of cells with lower and higher levels of cell-cell adhesion and during the outward buckling of a stratified epithelial sheet and basement membrane to form buds. This process can be engineered, even using cells that do not normally bud [74].

In mouse mammary epithelial branching during puberty, a focal increase in ECM accumulation and stiffness constrains the angle of bifurcation of terminal end buds by influencing the dynamics of motile epithelial cells. According to modeling and observation, these effects do not require instruction from pre-aligned collagen fibers as previously hypothesized, but instead local stiffness. Modeling by a finite element computational analysis and in vitro cultures showed that the key effect of enhanced local ECM stiffness is to constrain the angle of bifurcation to establish the global orientation of the remodeling epithelium [75].

The value of these types of physical and computational models is to generate testable hypotheses that challenge the generality and effectiveness of a model. We can safely predict that computational and physical modeling will become increasingly valuable, not only for understanding developmental and tissue dynamics, but also for promoting rational tissue engineering and regenerative medicine. For example, the nature of the local microenvironment for stem cell regenerative medicine will be critical, since disease often results in local fibrosis with altered stiffness and viscoelasticity, which are likely to be deleterious for cell-based tissue engineering. A combination of direct physical characterization and modeling should facilitate the process of developing novel regenerative medicine approaches.

Concluding remarks

Recent research on the mechanisms and key biological roles of bidirectional cell interactions with the ECM has expanded explosively, generating many exciting opportunities and questions for the future (see Outstanding questions). Recent reviews of this field can provide more details (e.g., [1,15,27,42,62,63,76–81]), though they may become outdated. As this field matures further, we suggest a caution concerning generalizations and reproducibility. As noted above, modes of cell migration can clearly vary widely depending on the type of cell and its surrounding ECM, making it dangerous to generalize about universal principles of 3D cell migration, invasion, and tissue remodeling. In fact, current concerns in biomedical science concerning reproducibility of findings from different laboratories [82] may often be attributable to over-generalizing from research using only a single cell type under laboratory-specific artificial experimental conditions. It remains important for us to continue to apply the classical approach of asking whether phenomena can be reproduced in cells of different tissue origins, and whether one can apply orthogonal experimental methods to reach the same conclusions.

Outstanding Questions Box.

What are the intermolecular binding and functional network interactions of the many components of the ECM matrisome in each tissue, and how do they change in development and diseases?

How do cells assemble and remodel complex in vivo 3D ECMs in different tissues?

How do the physical attributes and microarchitecture of different 3D ECM microenvironments affect cell interactions, adhesions, and mechanotransduction?

By which mechanisms can ECM viscoelasticity regulate so many cellular responses?

How is actin polymerization at the leading edge of cells regulated by 3D ECM stiffness and compression?

What is the super-resolution organization and ultrastructure of 3D cell-matrix adhesions, and what are the dynamics of their matrisome constituents?

Can we generate more physiologically relevant 3D in vitro ECM model systems that mimic tissue- and disease-specific in vivo ECM conditions?

What are the different modes of in vivo cell migration in different ECM microenvironments, and do they use novel biophysical mechanisms?

How does altering cell-ECM and cell-cell tension and dynamics – both short-range contacts and long-range ECM strains – within embryonic and cancerous tissues help to govern developmental and cancer progression?

Are there undiscovered feedback loops regulating cell-ECM interactions, and can regulatory loops be targeted therapeutically?

How is collagen reorganized, aligned, and oriented in cancers, and how important is this process for pathogenesis and potential therapy?

Can scarring and fibrosis be blocked by manipulating YAP and other mechanotransduction signaling pathways?

Will increasingly sophisticated computational models not only provide new mechanistic concepts, but also identify key regulatory points to permit therapy of diseases and disorders?

What principles of cell-ECM interactions can facilitate personalized regenerative medicine, including establishing optimized stem cell niches and generating replacement organs?

The ongoing discovery of new underlying concepts will continue to expand our understanding of the complex bidirectional reciprocity of signaling between cells and their ECM microenvironment, the vast numbers of molecules implicated in matrisome-cell interactions, the crucial contributions of feedback loops between physical/mechanical and intracellular signal transduction, and their many roles in development, tissue function, and pathophysiology. Future research will likely play increasingly important roles in regenerative medicine as translational research with novel approaches to high-throughput drug screening [83] target the interactions of stem cells, differentiating cells, or pathological tissues with their local ECM microenvironments. Understanding the complex networks of matrisome components and their feedback loops linked to cellular signaling will provide many possibilities for sophisticated, targeted biomedical interventions.

Highlights.

• The diversity of hundreds of extracellular matrix (ECM) molecules in different tissues and their interactions are now being documented in “matrisome” databases.

• Physical properties of the 3D ECM, including viscoelasticity and microarchitecture, can govern cell adhesion, mechanotransduction, and multiple modes of cell migration.

• New advances in ECM biology are identifying mechanisms of cancer progression and fibrosis, as well as potential therapeutic targets.

• Characterizations of cell-ECM feedback loops and computational modeling are providing new insights and potential opportunities for intervention in diseases and disorders.

Glossary

Adhesome

the molecular components of cell-to-matrix adhesions, particularly of focal adhesions

Arp2/3 complex

a complex of 7 proteins that initiates a lateral actin filament at a 70° angle from a pre-existing F-actin filament; the new actin filament points toward the plasma membrane

Basement membrane

the flat, thin but dense, sheet-like form of ECM located under epithelia separating them from connective tissue (stroma), as well as under certain other cells, such as muscle and fat cells

Budding and branching morphogenesis

a process during embryonic development in which epithelia form buds and undergo branching to form a tree-like architecture that maximizes epithelial surface area in compact organs

Desmoplasia

the formation and progression of fibrous tissue, e.g., in association with cancer

Durotaxis

migration of cells along an ECM stiffness gradient toward regions of increased stiffness or rigidity

Elasticity

a physical property of ECM in which deformation by force is followed by springing back after removing the force

Exosomes

vesicles released by cells containing cytoplasmic materials surrounded by a plasma membrane

Extracellular matrix (ECM)

the non-cellular molecules of all tissues and organs, which provide tissue structural support and induce both biochemical and mechanical signaling

Feed-forward loop

a regulatory mechanism in which a stimulus is transmitted forward and then enhanced by feedback from the target in a loop that increases its strength

Fibrosis

a pathological form of wound repair in which connective tissue rich in collagen accumulates excessively and replaces normal tissue

Filopodia

slender cellular protrusions extending from the leading edge of cells that contain actin

Focal adhesion

a dynamic, micron-sized subcellular structure composed of hundreds of structural and signaling proteins that act as a mechanical link between transmembrane integrins with their extracellular matrix ligand and the actomyosin cytoskeleton

Finite element computational analysis

a method for solving differential equations by numerical analysis to model dynamic processes such as fluid flow and mass movements of particles and cells

Lattice-gas cellular automaton model

a computational model derived from fluid dynamics in which cells are considered as particles interacting with other particles under various physical states

Matrisome

the molecular components of the ECM at a particular time in embryonic development, in an adult tissue, or in a disease or disorder.

Mechanosensing

the sensing by cells of mechanical information or stimuli

Mechanotransduction

the conversion of mechanical information by cells into signaling or a subsequent contractile response

Microspikes

tiny spike-like protrusions from a cell

Rheology

the study of the flow or deformation of materials including viscoelastic solids and fluids

Strain

deformation of a material due to applied stress

Stress

force applied to a cell or ECM

Stress relaxation

a physical property of ECM in which deformation by a stress results in reduced resistance of the material to the stress over time due to irreversible, plastic deformation

Viscoelasticity

a physical property of ECM in which its deformation by a force results in plasticity changes in shape that are irreversible, i.e., the ECM does not return to the original state or form

YAP/TAZ

key transcriptional regulators with full names “Yes-associated protein 1” and “transcriptional coactivator with PDZ-binding motif,” respectively, that can regulate genes involved in cell proliferation, suppression of cell death, and ECM accumulation. YAP/TAZ can respond to ECM stiffness during mechanotransduction by translocating into the nucleus to enhance specific gene expression