SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Matrix Dimensionality and Cell Behavior
  4. The Influence of Cell–Matrix Interaction on Extracellular and Intracellular Compartments
  5. Potential Future Therapeutic Applications
  6. Conclusions
  7. Acknowledgements
  8. Literature Cited

The complex interactions of cells with extracellular matrix (ECM) play crucial roles in mediating and regulating many processes, including cell adhesion, migration, and signaling during morphogenesis, tissue homeostasis, wound healing, and tumorigenesis. Many of these interactions involve transmembrane integrin receptors. Integrins cluster in specific cell–matrix adhesions to provide dynamic links between extracellular and intracellular environments by bi-directional signaling and by organizing the ECM and intracellular cytoskeletal and signaling molecules. This mini review discusses these interconnections, including the roles of matrix properties such as composition, three-dimensionality, and porosity, the bi-directional functions of cellular contractility and matrix rigidity, and cell signaling. The review concludes by speculating on the application of this knowledge of cell–matrix interactions in the formation of cell adhesions, assembly of matrix, migration, and tumorigenesis to potential future therapeutic approaches. J. Cell. Physiol. 213:565–573. © 2007 Wiley-Liss, Inc.

The extracellular matrix (ECM) provides scaffolds for cellular support that are present in all tissues and organs. The ECM is a complex mixture of matrix molecules, including the glycoproteins fibronectin, collagens, laminins, proteoglycans, and non-matrix proteins including growth factors. Cell adhesion to the ECM induces discrete cell surface structures tightly associated with the matrix termed cell–matrix adhesions, which mediate direct interactions of the cell with its extracellular environment. Cell–matrix adhesions are essential for cell migration, tissue organization, and differentiation, and as a result they play central roles in embryonic development, remodeling, and homeostasis of tissue and organ systems. Matrix adhesion signals cooperate with other pathways to regulate biological processes such as cell survival, cell proliferation, wound healing, and tumorigenesis. Thus, elucidating the structure and function of cell–matrix adhesions provides a critical vantage point for understanding the regulation of eukaryotic cellular phenotypes in vivo. For recent reviews see references Miranti and Brugge (2002); Danen and Sonnenberg (2003); Guo and Giancotti (2004); Wozniak et al. (2004); Ginsberg et al. (2005); Li et al. (2005); Mitra et al. (2005); Vicente-Manzanares et al. (2005); Janes and Watt (2006); Luo et al. (2007). Due to constraints on the length of the article, we apologize for not being able to cite all relevant references.

Integrins are the principle cell surface adhesion receptors mediating cell–matrix adhesion. Integrins are heterodimeric receptors generated by selective pairing between 18 α and 8 β subunits; there are 24 distinct integrin receptors that bind various ECM ligands with different affinities (Luo et al., 2007). Some integrin subunits are ubiquitously expressed, while other subunits are expressed in a tissue- or stage-restricted manner (Humphries et al., 2006). For instance, integrin β1 is ubiquitously expressed, whereas the β6 subunit is only expressed in the adult during wound healing.

The extracellular domains of integrin receptors bind ECM ligands and divalent cations, but they can also associate laterally with other proteins at the cell surface, such as tetraspanins, growth factor receptors, matricellular proteins, and matrix proteases or their receptors (Miranti and Brugge, 2002). Integrins influence cell behavior not only by providing a docking site for the ECM at the cell surface, but also by actions of the integrin intracellular domains. Integrin intracellular domains are short regions of roughly 50 amino acids in length, except for integrin β4 (∼1,000 amino acids). Integrin cytoplasmic domains form multi-molecular complexes with proteins involved in cell signaling and with adaptor proteins that provide a connection to the cytoskeleton (Hynes, 2002).

Integrins provide a bi-directional conduit for mechanochemical information across the cell membrane, providing a major mechanism for connecting the intracellular and extracellular compartments. Cell adhesion to the ECM transmits information via integrin receptors that regulates intracellular signaling via outside-in signaling, which is important, for example, in cell spreading and cell migration. Conversely, intracellular signals can induce changes in integrin conformation and activation that alter its ligand-binding activity in a process termed inside-out signaling. Integrin engagement with matrix can also affect integrin activation, providing bi-directional crosstalk between inside-out and outside-in signaling (Ginsberg et al., 2005; Luo et al., 2007).

Integrin clustering follows the engagement of integrins with the naturally multivalent ECM, and it promotes the localized intracellular concentration of signaling molecules. Clustering of integrin receptors, or particularly integrin β cytoplasmic domains, activates non-receptor tyrosine kinases such as focal adhesion kinase (FAK) leading to localized increases in the levels of tyrosine-phosphorylated proteins. Serine/threonine kinases including members of the protein kinase C family, lipid kinases such as PI 3-kinase, and phosphatases such as RPTPα are also regulated by integrin engagement and clustering. These kinase and phosphatase signaling pathways induce post-translational modifications regulating the protein interactions and/or enzymatic activity of the substrates (Li et al., 2005). They specify protein recruitment to adhesion complexes and thereby selectively link matrix adhesions to various downstream signaling cascades that control cytoskeletal organization, gene regulation, and diverse cellular processes and functions (Fig. 1) (Hervy et al., 2006).

thumbnail image

Figure 1. General model of cell–matrix adhesions and their downstream regulation. Cell-extracellular matrix adhesions containing clusters of integrins recruit cytoplasmic proteins, which in cooperation with other cell surface receptors control diverse cellular processes, functions, and phenotypes.

Download figure to PowerPoint

More than 50 cytoplasmic proteins are present in cell–matrix adhesion structures (Lo, 2006). Because integrin receptors lack intrinsic enzymatic activity, they must recruit signaling proteins to control adhesion-dependent processes (Liu et al., 2000; Mitra et al., 2005). Three basic categories of proteins are recruited to cell–matrix adhesions: (1) integrin-binding proteins, (2) adaptors and/or scaffolding proteins that lack intrinsic enzymatic activity, and (3) enzymes. Talin is an example of a protein that directly binds to integrin cytoplasmic domains and is important for regulating integrin activation and signaling (Calderwood, 2004). Adaptors and/or scaffolding factors link integrin-associated proteins with actin or other proteins, and examples include vinculin, paxillin, and α-actinin. Enzymes that modify integrin downstream effectors include the non-receptor tyrosine kinases FAK and Src. The profiles of proteins recruited to matrix adhesions specify the biochemical signals and biophysical properties of matrix adhesions (Li et al., 2005).

The cytoskeleton contains three general classes of filamentous structure, F-actin, intermediate filaments, and microtubules. It has become clear that cell migration and tissue remodeling require coordinated crosstalk between the actin, intermediate filament, and microtubule cytoskeletal networks (e.g., see reference Even-Ram et al. (2007)). Actin polymerization and proteins important for regulating actin organization are essential regulators of membrane protrusion and cell migration. Rearrangements of the actin cytoskeleton are mediated by complex molecular pathways that promote actin polymerization, actin depolymerization to renew the intracellular pool of monomeric actin, and modifications of actin-crosslinking proteins (Vicente-Manzanares et al., 2005; Pollard, 2007). Matrix adhesions associate with bundles of actin filaments, and bi-directional interactions mediated by the actin cytoskeleton involving actomyosin contraction and clustering of integrins bound to matrix combine to increase cell contractility, also referred to as endogenous tension. The dynamic assembly and disassembly of adhesion structures applies different levels of force to the matrix that, in turn, regulates endogenous tension (Wozniak et al., 2004). Conversely, endogenous tension is transmitted through integrins to the ECM and can increase matrix rigidity, referred to as exogenous tension. Adhesion structures recruit cytoplasmic proteins that induce downstream effectors involved in regulating matrix deposition or remodeling. Thus, integrin engagement with the ECM generates bi-directional signals that can alter endogenous tension, exogenous tension, and matrix composition (Fig. 2) (Katsumi et al., 2004; Peterson et al., 2004; Wang, 2007).

thumbnail image

Figure 2. Coordinate regulation of exogenous tension (matrix rigidity) and endogenous tension (contractility). A: Cells and matrix mutually interact to regulate tension. B: Tension in the cell microenvironment is thought to be distributed by integrin receptors that signal bi-directionally between extracellular and intracellular compartments. Tension levels may alter outside-in and inside-out integrin signaling. Integrin engagement with extracellular matrix (ECM) regulates endogenous cellular tension by triggering actin cytoskeletal organization and actomyosin contractility. Endogenous tension levels can indirectly or directly control exogenous tension (matrix rigidity) as indicated in the figure.

Download figure to PowerPoint

Directional cell migration requires the establishment of cell polarity to create a leading edge and a trailing edge (Moissoglu and Schwartz, 2006). The leading edge undergoes membrane protrusive activities driven by actin polymerization that establish new matrix contacts, whereas at the trailing edge cell adhesions are disassembled to promote retraction of the cell rear and forward cell movement (Vicente-Manzanares et al., 2005). Local actin polymerization can induce membrane protrusions and favor formation of matrix contact by localizing integrin receptors in an active conformation at cell protrusions (Galbraith et al., 2007). The rate of cell migration can be limited by the rate of rear retraction, and thus the dynamic formation and disassembly of cell–matrix adhesions are critical to cell migration (Ridley et al., 2003).

The Rho GTPase family of GTP-binding proteins including RhoA, Rac1, and Cdc42 are critical regulators of cell contractility, lamellipodial and filopodial formation, and cellular polarity. RhoA GTPase downstream signals such as activation of RhoA-kinase (ROCK) and inhibition of myosin phosphatase increase myosin light chain phosphorylation, leading to clustering of actin stress fibers to regulate actomyosin contractility and endogenous tension. Rho downstream signals regulate membrane retraction, thereby significantly contributing to leading and trailing edge cell polarity in cell migration. The balance of Rac and Rho activation coordinates membrane protrusion, retraction, and numbers of protrusions during cell migration. Sites of high levels of active Rac1 will suppress RhoA and induce lamellipodia; in contrast, regions containing concentrated active RhoA will have low Rac1 activity and membrane retraction (Clark et al., 1998; Nimnual et al., 2003; Burridge and Wennerberg, 2004). Furthermore, integrin engagement with matrix regulates the activity of Rho GTPase family members and localization with downstream effectors. Thus, matrix adhesions establish feedback loops that control membrane protrusion and retraction during cell migration by coordinating and integrating the activities of individual Rho GTPase family members in the leading and trailing edges of the cell (Ridley et al., 2003).

The remainder of our review will focus on the instructive role of cell–matrix adhesions on both cell phenotype and extracellular environment. First, we compare how cell morphology, cell migration, and cell–matrix adhesion structures respond to two-dimensional (2D) compared to three-dimensional (3D) matrices. Changes in cell–matrix adhesions associated with cancer are described. We then examine how alterations of the extracellular environment influence the intracellular environment and vice versa. Finally, we speculate about ways in which our rapidly expanding knowledge about cell–matrix adhesion might be exploited for therapeutic purposes.

Matrix Dimensionality and Cell Behavior

  1. Top of page
  2. Abstract
  3. Matrix Dimensionality and Cell Behavior
  4. The Influence of Cell–Matrix Interaction on Extracellular and Intracellular Compartments
  5. Potential Future Therapeutic Applications
  6. Conclusions
  7. Acknowledgements
  8. Literature Cited

Eukaryotic cells adapted to grow in vitro are routinely cultured on a 2D substratum. Many studies have characterized cellular responses to growth on a 2D ECM-coated substratum. They have identified complex molecular and biochemical pathways activated or modified by integrin-mediated adhesion and have provided insights into mechanisms that regulate adhesion-dependent cellular processes such as cell spreading, cell proliferation, cell differentiation, and cell survival. Fibroblasts cultured in 2D matrices interact with a rigid substratum at the ventral surface of the cell. The binding of integrin ligands and the differences in exogenous tension (matrix rigidity) between the dorsal and ventral cell surfaces can selectively trigger focal adhesion formation at the ventral surface, thereby inducing cell polarity (Giannone and Sheetz, 2006). Fibroblasts in 3D matrices are typically not exposed to these large differences in exogenous tension, and hence lack dorsal–ventral polarity. However, observations in 2D cultures suggest that regional variations in 3D-matrix exogenous tension (rigidity) may influence the distribution of cell–matrix adhesions and cell behavior in vivo (Ingber, 2006). Intriguing new insights into the effects of 3D matrix on cell behavior elucidate the synergistic relationship of cell and ECM in vivo and the dynamic function of cell–matrix adhesions in a 3D environment.

Cell morphology

Three-dimensional ECMs can have striking effects on cell morphology, which differ depending on whether the cells are epithelial cells or fibroblasts. Mammary epithelial cells grown in 2D versus 3D matrices have dramatic differences in organization. In 3D, the mammary epithelial cells aggregate, form cell–cell contacts, polarize, and establish spherical acini (Debnath and Brugge, 2005). In contrast, when cultured on a 2D ECM, these cells grow as a simple monolayer (Nelson and Bissell, 2006). Fibroblasts in vivo are normally embedded within a collagen-rich ECM. Fibroblasts adherent to a 2D matrix attach, spread out, and flatten with large prominent lamellipodia. When placed back into a 3D matrix, the fibroblasts re-acquire an elongated spindle-shaped phenotype devoid of large, flat lamellipodia. The differences between fibroblast and epithelial cell morphologies in 2D and 3D suggest that the physical configuration of the matrix itself provides spatial signals that control cell morphology (Larsen et al., 2006). In fact, mechanically flattening a 3D matrix to a relatively 2D surface returns morphology to a 2D phenotype even though the same molecules and growth factors are present (Cukierman et al., 2001) yet sandwiching cells between two 2D surfaces can mimic a 3D environment (Beningo et al., 2004).

Cell migration

Migration of cells adherent to 2D matrices is based on cycles of lamellipodial extension, attachment, cell body translocation, and retraction of the cell rear. There are at least two major modes of cell migration through 3D matrices. In the mesenchymal mode of cell migration, just as on 2D surfaces, there are cycles of membrane protrusion at the leading edge, formation of matrix adhesions, dynamic cellular contractility, and retraction of the rear (Ridley et al., 2003; Li et al., 2005). For both 2D and 3D mesenchymal migration, cell traction forces are established by matrix-bound integrin receptors and transmitted by the actin cytoskeleton, and subsequent actomyosin contractility induces centripetal movements of actin and adhesion structures from the front and rear of cells. A second mode of 3D cell migration is termed amoeboid, similar to the process used by amoebae and leukocytes; this migration mode depends on non-adhesive conformational adaptation of cell shape to the local surrounding matrix. Cells adapt their shape to match the path of least resistance within the matrix, and migration is achieved by propulsive squeezing forward through gaps in the matrix (Friedl and Wolf, 2003). Fibroblasts in a 3D cell-derived matrix appear to migrate in a mesenchymal mode. Thus, the regulation of membrane protrusion, retraction, and endogenous tension by cell–matrix adhesions are likely to be important factors in fibroblast migration in 3D matrices. Levels of total Rac activity are lower in fibroblasts in a 3D matrix, resulting in fewer lamellae and more directional migration (Pankov et al., 2005).

An additional mode of cell–matrix interaction has been defined for cells on a 2D substratum interacting with a collagen fiber. In this mode, the cell translocates the collagen fiber toward the cell body using cycles of membrane protrusion to establish fiber contact and membrane retraction with the fiber attached. The repetitive protrusion and retraction cycles elicit a “hand-over-hand” membrane dynamic associated with movement of the collagen fiber (Meshel et al., 2005). It will be interesting to determine the role of hand-over-hand membrane dynamics in 3D matrices.

Molecular composition of cell–matrix adhesions

At least four different types of adhesion structures have been defined in fibroblasts, termed focal complexes, focal adhesions, fibrillar adhesions, and 3D-matrix adhesions. We will focus on matrix–integrin–actin adhesion structures that are important contributors to regulating endogenous and exogenous tension (Katsumi et al., 2004). Focal complexes are small, transient matrix contact structures that provide early cell attachment at the leading edge. If stabilized, they will subsequently form focal adhesions, which can in turn transition to fibrillar adhesions. For cells on 2D matrices, the assembly and disassembly of matrix adhesions are regulated in a dynamic fashion in response to cell signals and the exogenous tension (matrix rigidity). The biological relevance of focal adhesions was initially questioned, since equivalent structures to these prominent 2D adhesion structures were not observed in most tissues. However, focal adhesions have been found at points of high fluid shear stress in blood vessels (Romer et al., 2006). The fourth adhesion structure was identified in fibroblasts cultured within a cell-derived 3D matrix and was also detected in tissues, indicating that adhesion structures containing integrins and actin do indeed form in 3D and in vivo, albeit with differing morphology and composition (Cukierman et al., 2001).

The ability of integrins to localize to adhesion structures is not restricted to a particular integrin receptor. However, certain integrin receptors are preferentially concentrated at different cell–matrix adhesion structures. For example, fibroblasts adherent to a 2D fibronectin matrix will form focal complexes and focal adhesions that are rich in αvβ3. While α5β1 is often excluded from the focal adhesion core, fibrillar adhesions contain α5β1. Three-dimensional-matrix adhesions contain primarily α5β1, but αvβ3 can be observed at the adhesion periphery. It seems likely that different integrin receptors will recruit different cytoplasmic factors and differentially control cell signaling and cellular tension (Cukierman et al., 2001).

Subsets of proteins are recruited to different adhesion structures suggesting that adhesions may have signaling specificity. For instance, focal adhesions contain vinculin and numerous tyrosine-phosphorylated proteins including FAK and paxillin. Fibrillar adhesions contain abundant tensin, low levels of protein tyrosine phosphorylation, and α5β1 instead of αvβ3 integrins. Three-dimensional-matrix adhesions are similar to fibrillar adhesions regarding α5β1 localization; however, 3D-matrix adhesions contain high levels of vinculin, α-actinin, and phosphorylated paxillin. Proteins that are tyrosine phosphorylated localize to both 3D-matrix adhesions and focal adhesions, but levels of FAK Y397 phosphorylation are low in 3D-matrix adhesions, indicating that integrin signaling can differ substantially in 3D compared to 2D environments (Fig. 3) (Zamir and Geiger, 2001; Yamada et al., 2003).

thumbnail image

Figure 3. Comparison of focal adhesions, fibrillar adhesions, and 3D-matrix adhesions. These matrix adhesion structures recruit distinct cytoplasmic proteins and differ in signaling, for example, in the levels of protein tyrosine phosphorylation (pY) of adaptor and signaling proteins. For example, FAK pY 397 (phosphorylation at tyrosine 397) levels are high in focal adhesions but substantially lower in fibrillar and 3D-matrix adhesions. Paxillin pY 31 and FAK pY 861 levels are high in both focal adhesions and 3D-matrix adhesions but are lower in fibrillar adhesions. Based on these differences, distinct protein complexes are likely to form at each type of matrix adhesion to trigger specific signaling pathways.

Download figure to PowerPoint

Levels of particular integrin receptors can be altered in various diseases, including cancer progression. Normally, integrin β6 is restricted in expression to epithelial cells during embryonic development and is not typically expressed in adult tissues. Integrin β6 is expressed at the cell surface in association with αv as a receptor for fibronectin, tenascin, and vitronectin. Expression of αvβ6 is induced on epithelial cells during wound healing and in carcinomas of the colon, lung, oral cavity, breast, and cervix (Janes and Watt, 2006). Increased expression of the integrin β1 subunit correlates with decreased breast cancer survival. Other integrin receptors such as αvβ3 and α6β4 are induced in highly metastatic melanoma cells and pancreatic adenocarcinoma progression, respectively. The shifting profile of integrin receptor expression may influence endogenous (contractility) or exogenous (matrix rigidity) tension or facilitate tumor cell survival and migration in multiple tissues with different matrix compositions (Guo and Giancotti, 2004; Danen, 2005; Wilhelmsen et al., 2006).

Changes in the expression of the integrin-associated proteins FAK, paxillin, α-actinin, and vinculin are also observed during tumor progression. Tumor cells appear to take advantage of the ability of FAK to regulate pathways important for cell proliferation, cell migration, gene expression, and survival (Mitra et al., 2005; Slack-Davis et al., 2007). For instance, FAK expression and activity are enhanced in metastatic tumors of the oral cavity, colon, rectum, thyroid, prostate, and cervix. In ovarian cancer, increased FAK expression correlates with decreased patient survival. In breast carcinoma, FAK activity is important for VEGF expression and tumor angiogenesis (McLean et al., 2005). Similar to FAK, increased expression of paxillin is observed in breast carcinoma, and α-actinin expression levels are increased in melanomas and in tumor cell lines with faster migration rates (Vadlamudi et al., 1999). In contrast, vinculin expression appears to have an inverse relationship with tumor metastasis. Vinculin expression is elevated in weakly metastatic melanoma cells. In highly metastatic melanoma cell lines, vinculin expression is reduced, and vinculin localizes to small punctate adhesion structures (Lifschitz-Mercer et al., 1997). It will be important to determine whether these changes in expression of integrin-associated proteins contribute directly to tumor progression or are secondary responses to altered tumor or tumor stromal microenvironments.

Cellular interactions with the matrix are coordinated with local actin polymerization and F-actin organization. Fibroblasts adherent to a 2D matrix contain prominent bundles of actin filaments or stress fibers that insert into focal adhesions. The Arp2/3 complex that nucleates actin polymerization is localized in a broad band at the leading edge within prominent lamellipodia (Vicente-Manzanares et al., 2005). In a 3D matrix, thin stress fibers are located at the periphery of membrane extensions. Focal adhesions are rare and appear as dot-like structures near the protrusive edge. The Arp2/3 complex is in foci at the tips of extensions and lamellipodia are smaller and narrow (Beningo et al., 2004). Actin and nucleators of actin polymerization such as the Arp 2/3 complex assume different spatial configurations in cells engaged with 2D and 3D matrices. Thus, dynamics in exogenous tension or rigidity within 3D matrices may influence the cellular distribution of actin and the Arp2/3 complex.

The Influence of Cell–Matrix Interaction on Extracellular and Intracellular Compartments

  1. Top of page
  2. Abstract
  3. Matrix Dimensionality and Cell Behavior
  4. The Influence of Cell–Matrix Interaction on Extracellular and Intracellular Compartments
  5. Potential Future Therapeutic Applications
  6. Conclusions
  7. Acknowledgements
  8. Literature Cited

Matrix control of cell phenotype

Matrix ligand density and exogenous tension levels

Conversion between motile and stationary phenotypes is critical for controlling developmental processes and tissue remodeling, for example, the recruitment and organization of epithelial cells and fibroblasts during wound healing. The density of matrix ligands regulates cell migration. There is a biphasic response of cell motility to increasing the matrix ligand density in both 2D and 3D matrices. Cells migrate poorly on substrates of relatively low or high matrix ligand density, and they preferentially migrate at an intermediate density (Li et al., 2005). The matrix rigidity or exogenous tension of collagen gels regulates fibroblast cell migration and cell signaling (Rhee et al., 2007). Fibroblasts migrate efficiently on less rigid matrices with lower exogenous tension. The relative migration rate is impeded on rigid matrices with high exogenous tension. The sensitivity of cell migration to the matrix ligand density and matrix rigidity indicates that cellular mechanisms link exogenous tension (matrix rigidity) and ligand density to cell migration. Interestingly, if the migratory path of a cell includes differentials in either matrix ligand density or matrix rigidity, the cells will move toward the ligand-dense or more rigid regions and away from the low ligand density or less rigid matrix in phenomena termed haptotaxis and durotaxis, respectively (Lo et al., 2000).

Mammary tumors and the adjacent tumor stroma have high exogenous tension compared to normal mammary gland. Metastatic breast tumors, unlike non-metastatic tumors, frequently express high levels of lysyl oxidase, a collagen-crosslinking enzyme, and form rigid stroma in vivo. Overexpression of an isoform of lysyl oxidase in a non-metastatic breast tumor cell line is sufficient to induce both tumor fibrosis and tumor invasion (Kirschmann et al., 2002). In breast tumor biopsies, invasive potential correlates with the presence of fibrotic foci associated with the tumor stroma (Hasebe et al., 2002).

The influence of tumor stroma tension on cell phenotype was recapitulated in 3D cultures using the basement membrane extract Matrigel crosslinked to polyacrylamide gels of variable tension. Exogenous tension (rigidity) was adjusted to the levels observed in normal and tumor-containing tissue by changing the polyacrylamide gel composition. Elevation of exogenous tension enhances Rho activity, induces cytoskeletal tension, increases focal adhesions, decreases cell–cell contact, perturbs tissue polarity, and increases growth. Exogenous tension (rigidity) at levels comparable to the tumor stroma induces integrin clustering and activation of intracellular signaling, such as ERK phosphorylation and ROCK-dependent contractility. Inhibiting Rho and ERK signaling was sufficient to decrease endogenous tension in cells adhering to a rigid matrix, and this resulted in decreased formation of focal adhesions and increased cell–cell contacts (Fig. 4) (Paszek et al., 2005). Thus, the high exogenous tension of the tumor stroma can influence the endogenous tension and cell–matrix adhesions that form in tumor cells (Bershadsky et al., 2006). The levels of exogenous tension or rigidity also regulate preosteoblast proliferation, differentiation, and focal adhesion dynamics (Kong et al., 2005).

thumbnail image

Figure 4. Increasing the exogenous tension (matrix rigidity) has dramatic effects on intracellular signaling, matrix adhesions, and endogenous tension (contractility). An increase in exogenous tension alters RhoA activity, the actin cytoskeleton, focal adhesions, cell–cell contacts, tissue polarity, and importantly the growth rate. The figure shows a comparison of cellular phenotypes on matrices of different rigidity ([UPWARDS ARROW] indicates relative increase and [DOWNWARDS ARROW] relative decrease). Thus, dense “desmoplastic” tumor stroma will influence the endogenous tension, matrix adhesion structures, and cell signaling pathways. If some mechanism could be developed to reduce both the exogenous tension (rigidity) and endogenous tension (contractility) of rigid tumors, it might be possible to inhibit tumor growth or progression. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

It is interesting to speculate that due to durotaxis, the gradient in stroma exogenous tension at the tumor boundary—high at the tumor and low in surrounding tissue—may induce a migratory response of cells in the surrounding tissue, leading to preferential recruitment of cells to the tumor and its stroma to continually increase tumor size. The idea of cells responding to the presence of a rigid body in a matrix is not novel. Chicken heart fibroblasts are known to align and induce membrane protrusions toward an embedded rigid body from adjacent compliant areas (Boocock, 1989).

Composition of the ECM

Rarely will cells encounter only a single isolated ECM protein in vivo. The basement membrane and connective tissue are specialized matrices containing networks of multiple proteins including several collagen isoforms, laminin isoforms, and fibronectin. Migration from a collagen plug and collagen gel contractility by fibroblasts adherent to collagen or mixtures of collagen and fibronectin indicate that the matrix mixture elicits a cell phenotype that is distinct from cells adherent to only collagen (Greiling and Clark, 1997; Liu et al., 2006). Consequently, this and many other studies indicate that the composition of the ECM, rather than simply the presence of an extracellular scaffold, is critical for regulating cell phenotype.

Cell control of the matrix

Matrix porosity

The matrix is a meshwork of fibrillar and non-fibrillar proteins containing pores. At a constant matrix concentration, crosslinking of the matrix will reduce pore size and enhance the barrier function of the matrix (McKegney et al., 2001). Lysyl oxidase is a cell-derived collagen-crosslinking enzyme that associates with fibronectin to form a complex that retains crosslinking activity (Fogelgren et al., 2005). Thus, cell-derived lysyl oxidase may utilize fibronectin as a scaffold to locally increase collagen crosslinking and decrease matrix porosity.

Matrix cleavage

Cells use matrix proteases to remodel the matrix. ECM remodeling is required for normal physiological processes such as embryonic development, morphogenesis, and wound repair. Tissues normally harbor low levels of matrix protease activity that are controlled by inhibitors such as tissue inhibitors of metalloproteinases (TIMPs) (recently reviewed in references Nagase et al. (2006); Page-McCaw et al. (2007)). In many disease states, or following tissue damage, there is a shift favoring matrix protease activity, and matrix remodeling ensues. The triple-helical structure of fibrillar collagen confers resistance to many proteases except collagenase. Collagenase promotes uncoiling of the triple helix and exposes additional sites that become susceptible to proteolysis. Thus, the activity of collagenases can disrupt the fibrous meshwork of the ECM and increase matrix porosity.

Endothelial cells provide an excellent example of inherent differences between 2D and 3D matrix adhesions and associated downstream pathways that regulate matrix protease activity and cell migration. Endothelial cell migration occurs independent of matrix protease activity in a 2D matrix. In contrast, in a 3D matrix environment, endothelial cell migration is dependent on protease activity. Interestingly, the activation of matrix metalloproteinase-2 also occurs selectively for endothelial cells cultured in a 3D matrix. Thus, 2D and 3D matrix adhesions induce different cell signaling pathways that control matrix protease activity and the mode of cell migration (Koike et al., 2002; Fisher et al., 2006).

Invasive tumor cells shift the proteolytic balance and display enhanced matrix protease activities. Tumor-associated matrix proteases can promote cell invasion, in part, by increasing matrix porosity or generating pro-migratory matrix peptides (Giannelli et al., 1997; Hotary et al., 2006). However, the requirement for matrix remodeling during tumor cell migration in vivo is controversial, since cells can reportedly shift between protease-dependent and independent migration modes (Friedl and Wolf, 2003; Wolf et al., 2003). Nonetheless, carcinoma cells acquire the ability to invade through the basement membrane and connective tissues, potentially utilizing both matrix protease-dependent and independent modes of cell migration.

Invadopodia are membrane protrusions on the cell surface of tumor cells that mediate matrix cleavage. Many of the cytoplasmic proteins that are recruited to matrix adhesions in primary cells are also recruited to invadopodia in tumor cells. Invasive cancer cells can extend multiple invadopodia that induce matrix cleavage, leading to increases in matrix porosity and liberation of pro-migratory matrix peptides. Because of their matrix remodeling properties, invadopodia are suggested to promote tumor cell invasion (Artym et al., 2006; Weaver, 2006; Linder, 2007).

Cellular contractility or endogenous tension

Cell–matrix interactions appear to utilize a molecular clutch mechanism that provides a bi-directional conduit for controlling mechanical tension across the cell membrane (Evans and Calderwood, 2007). Endogenous intracellular tension fluctuates in response to disruption or stabilization of the linkages between matrix-engaged integrins, the F-actin cytoskeleton, and myosin. Actomyosin contractility increases endogenous cell tension through sliding of actin filaments and subsequent force applied to matrix adhesions. Increasing the cytoplasmic tension of matrix-engaged integrins can transfer tension to the tethered matrix and increase exogenous tension (matrix rigidity) (Smilenov et al., 1999; Brown et al., 2006; Hu et al., 2007). As an example, during fibroblast migration, cellular force applied through matrix adhesions in protrusions at the leading edge will displace the matrix centripetally, and this local matrix stretch increases the exogenous tension.

Fibronectin is comprised of a series of modular domains that fold into tertiary structures that can undergo conformational rearrangements in response to tension. Elevating actomyosin contractility or endogenous tension induces cell-associated fibronectin to undergo a conformational change that reveals previously concealed fibronectin-binding sites and triggers fibronectin matrix assembly (Pankov and Yamada, 2002; Mao and Schwarzbauer, 2005). Adjusting either the endogenous (contractility) or exogenous tension (matrix rigidity) can potentially regulate fibronectin matrix assembly and matrix structure.

Potential Future Therapeutic Applications

  1. Top of page
  2. Abstract
  3. Matrix Dimensionality and Cell Behavior
  4. The Influence of Cell–Matrix Interaction on Extracellular and Intracellular Compartments
  5. Potential Future Therapeutic Applications
  6. Conclusions
  7. Acknowledgements
  8. Literature Cited

The major advances in knowledge of the mechanisms and sequelae of cell–matrix interactions that we have summarized in this mini review should lead to new therapeutic approaches in the future. Principles such as the roles of matrix composition, three-dimensionality, and rigidity, as well as the existence of distinct types of cell–matrix adhesions and bi-directional signaling responses provide a rational foundation for the development of novel approaches to tissue repair and intervention in disease processes. For example, there are already many applications of matrix molecules and synthetic biomaterials to tissue engineering that speed wound repair and potentially replace failed organs (Lutolf and Hubbell, 2005; Maskarinec and Tirrell, 2005; Clark et al., 2007; Kong and Mooney, 2007; Metcalfe and Ferguson, 2007). Applying new knowledge of the principles of the 3D organization and rigidity of the microenvironment to controlling the cell signaling response should accelerate research in engineering tissues. Similarly, increasing the understanding of the roles of local tension and feedback mechanisms may lead to the development of approaches to facilitate wound repair and prevent scarring and fibrosis (Xiao et al., 2004; Metcalfe and Ferguson, 2007).

Since integrin-mediated adhesion and signaling are crucial for thrombosis and inflammation, there are major efforts already underway to develop novel therapies by manipulating integrin activation and functions (Rose et al., 2000; Coller, 2001; Meadows and Bhatt, 2007). Complex diseases with an inflammatory and mechanical component, such as atherosclerosis, may eventually include management of cell–matrix interactions, since they play crucial roles in cell recruitment, adhesion, and tissue remodeling (Hamm, 2003). A less-explored area of future application of cell–matrix biology involves cancer, and we propose below a few possible areas of future research. Although highly speculative, they provide examples of novel possible approaches based on new findings in matrix biology.

Potential therapeutic approaches based on altering cell–matrix adhesion in cancer

Selective expression of integrin receptors on tumor cells presents an opportunity for targeting drugs, peptides, or radioisotopes using anti-integrin antibody conjugates. In particular, integrin β6 expression is induced on several carcinomas, whereas β6 subunit expression is normally only detectable in the adult during wound healing (Jones et al., 1997). Tumor cells retain expression of the β6 subunit at metastatic sites, such as in lymph nodes (Hazelbag et al., 2007). Thus, anti-αvβ6 antibody conjugates may provide a tool for delivering cytotoxic agents to the primary tumor and metastatic sites for potential therapy at both early and late tumor stages (Weinreb et al., 2004; Sheppard, 2005; Hehlgans et al., 2007). Clearly, this approach is limited to tumors that express the integrin β6 subunit.

The localized expression of lysyl oxidase in metastatic tumor cells provides a unique opportunity to target proteins to invasive tumors or the associated stroma (Kirschmann et al., 2002; Erler and Giaccia, 2006), but the challenge will be how to obtain drug targeting. Invasive tumor cells expressing lysyl oxidase might be targeted with a fusion protein containing as a fusion partner the collagen substrate domain for lysyl oxidase. The substrate domain could crosslink the fusion protein to the tumor matrix (Lucero and Kagan, 2006).

Once a fusion protein or antibody is targeted to the tumor by taking advantage of selective tumor cell expression of an integrin or lysyl oxidase, a number of therapeutic approaches are possible (Fig. 5); some can take further advantage of knowledge of matrix biology. The fusion protein could contain a toxin that is specifically released in active form by engineering sites specific for matrix metalloprotease(s) known to be expressed by the particular tumor or its stroma. Alternatively, the fusion protein could carry a selective matrix protease inhibitor or non-cleavable substrate to inhibit local matrix remodeling and invasion. The strategy in this case would be to restrain a potentially metastatic tumor within a molecular cage designed to prevent cell invasion and tumor cell matrix remodeling. Engineered fusion proteins could be injected into sites of tumor excision after surgery to target any remaining tumor cells.

thumbnail image

Figure 5. Speculative applications of matrix biology: lysyl oxidase (LOX)-targeted inhibitors and a molecular cage for tumors. Certain metastatic tumors frequently express elevated lysyl oxidase, which leads to a rigid stroma with high exogenous tension. This selective expression of the collagen-crosslinking enzyme lysyl oxidase in metastatic tumors could potentially provide a unique opportunity to target and crosslink proteins to the tumor stroma. For example, modular proteins for targeting could contain the lysyl oxidase collagen substrate domain (LOX substrate domain) fused to effector domains such as inhibitors of matrix remodeling or toxins to be crosslinked to tumor stroma. The modular protein could be engineered to contain a protease-resistant matrix protein to be crosslinked to the tumor stroma, which together with targeted matrix protease inhibitors could trap tumor cells within a molecular cage designed to resist invasion and matrix remodeling. In addition, a targeted toxin could also be crosslinked to tumor stroma that is tailored to the matrix protease activity signature of the tumor for local activation and release. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Breast and other tumors can associate with dense, collagen-rich desmoplastic stroma. The high exogenous tension or rigidity associated with breast tumor stroma is known to induce signaling and cytoskeletal changes, disrupt tissue polarity, and stimulate cell growth (Paszek et al., 2005). Therefore, strategies to reduce the exogenous and endogenous tension of breast and other tumors characterized by high exogenous tension may provide a novel therapeutic opportunity to control tumor progression or recurrence (Fig. 4). Developing approaches to decrease exogenous and/or endogenous tension of tumors in vivo will require creative new technologies.

Conclusions

  1. Top of page
  2. Abstract
  3. Matrix Dimensionality and Cell Behavior
  4. The Influence of Cell–Matrix Interaction on Extracellular and Intracellular Compartments
  5. Potential Future Therapeutic Applications
  6. Conclusions
  7. Acknowledgements
  8. Literature Cited

Over the past few decades, there has been exciting progress in understanding the molecular mechanisms that regulate the formation and function of cell–matrix adhesions. Many cellular processes are now known to be regulated by signals from cell–matrix adhesion structures that are transmitted bi-directionally across the cell membrane and dynamically link the intracellular and extracellular microenvironments. Future studies will clarify further the roles of cell–matrix interactions in determining cellular fate in vivo. This knowledge should provide a foundation for developing molecular tools to specifically modify the cell–matrix interface in order to control particular cellular functions such as gene expression, cell migration, and differentiation. The ability to control cellular phenotype in vivo could be translated to therapeutic technologies to prevent progression of cancer and other diseases.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Matrix Dimensionality and Cell Behavior
  4. The Influence of Cell–Matrix Interaction on Extracellular and Intracellular Compartments
  5. Potential Future Therapeutic Applications
  6. Conclusions
  7. Acknowledgements
  8. Literature Cited

We appreciate the insightful comments of Marinilce dos Santos and Andrew Doyle in the preparation of this article. This research was supported by the Intramural Research Program of the NIH, the National Institute of Dental and Craniofacial Research, and the National Center on Minority Health and Health Disparities. We would like to dedicate this review to the memory of Dr. Suzanne Bernier.

Literature Cited

  1. Top of page
  2. Abstract
  3. Matrix Dimensionality and Cell Behavior
  4. The Influence of Cell–Matrix Interaction on Extracellular and Intracellular Compartments
  5. Potential Future Therapeutic Applications
  6. Conclusions
  7. Acknowledgements
  8. Literature Cited