|Home | About | Journals | Submit | Contact Us | Français|
Tissue morphogenesis and homeostasis are dependent on a complex dialogue between multiple cell types and chemical and physical cues in the surrounding microenvironment. The emergence of engineered three-dimensional (3D) tissue constructs and the development of tractable methods to recapitulate the native tissue microenvironment ex vivo has led to a deeper understanding of tissue-specific behavior. However, much remains unclear about how the microenvironment and aberrations therein directly affect tissue morphogenesis and behavior. Elucidating the role of the microenvironment in directing tissue-specific behavior will aid in the development of surrogate tissues and tractable approaches to diagnose and treat chronic-debilitating diseases such as cancer and atherosclerosis. Toward this goal, 3D organotypic models have been developed to clarify the mechanisms of epithelial morphogenesis and the subsequent maintenance of tissue homeostasis. Here we describe the application of these 3D culture models to illustrate how the microenvironment plays a critical role in regulating mammary tissue function and signaling, and discuss the rationale for applying precisely defined organotypic culture assays to study epithelial cell behavior. Experimental methods are provided to generate and manipulate 3D organotypic cultures to study the effect of matrix stiffness and matrix dimensionality on epithelial tissue morphology and signaling. We end by discussing technical limitations of currently available systems and by presenting opportunities for improvement.
Tissue development depends on coordinated cycles of transcriptionally regulated cell growth, death, and migration that are controlled by exogenous soluble and physical stimuli and spatially dependent cell–matrix and cell–cell adhesion (Barros et al., 1995; Ingber, 2006; Jacobson et al., 1997; Locascio and Nieto, 2001). Regardless of length scale, understanding the molecular basis of tissue-specific differentiation and homeostasis requires an appreciation of adhesion-dependent cell behavior in the context of a three-dimensional (3D) extracellular microenvironment and a complex adhesion-dependent multicellular tissue. Through genetic and biochemical analysis, we have learned much about cell adhesion, including details of adhesion molecule structure and function, and about how various cell adhesion molecules likely mediate cell–extracellular matrix (ECM) interactions and facilitate cell–cell junctional complexes (Fuchs et al., 1997; Huttenlocher et al., 1998; Springer and Wang, 2004). We have also learned much about how exogenous growth, death, migration, and even mechanical cues activate signaling cascades to influence the fate of individual cells and undifferentiated 2D cell monolayers (Huttenlocher et al., 1995; McBeath et al., 2004; Stegemann et al., 2005; Thornberry and Lazebnik, 1998; Wang et al., 2001). Yet, all too often, experimental conclusions reached from observations of single cells and simplified 2D monolayer cell sheets do not accurately represent how cells behave within 3D tissues in vivo (Green et al., 1999; Sethi et al., 1999). Indeed, developmental models and transgenic animals consistently underscore the importance of studying cell behavior in the correct tissue context. However, live animal experimentation is so inherently complex that systematic assessment of the effect of individual variables, such as cell shape and matrix compliance, on cell behavior is extremely challenging and impractical (Sethi et al., 1999; Wang et al., 2005). At the interface between in vivo studies and 2D culture models are the organotypic culture systems that can faithfully recapitulate various aspects of tissue organization and function ex vivo. These organotypic 3D models have been employed with varying degrees of success to clarify some of the mechanisms, whereby biological processes such as adhesion-dependent survival (Weaver et al., 2002; Zahir and Weaver, 2004), polarity (O’Brien et al., 2002; Wang et al., 1998), proliferation (Zink et al., 2004), and even epigenetics (Bissell et al., 1999; Zink et al., 2004) regulate cell behavior as well as novel feedback/regulatory mechanisms (Bissell et al., 2002). Organotypic culture models have been effectively applied to study tissue-specific differentiation (Bissell et al., 2002), to understand factors controlling stem cell behavior (Hendrix et al., 2001), and even microenvironmental control of malignant transformation and tumor dormancy (Margulis et al., 2005; Weaver et al., 1997). Through the prudent use of organotypic 3D models, critical disparities between the molecular determinants of cell polarity (reviewed in O’Brien et al., 2002), apoptosis resistance (Weaver et al., 2002), and growth factor responsiveness (Wang et al., 1998) in cells incorporated into a 3D tissue and those propagated as 2D monolayers have been revealed (reviewed in O’Brien et al., 2002). Yet, while 3D organotypic models can faithfully recapitulate some aspects of tissue behavior ex vivo, many of the systems routinely used to study tissue-like behaviors employ crudely defined natural ECM molecules that contribute to considerable experimental variance. In addition, many of the approaches used to assemble 3D tissue-like structures in culture operate by simultaneously modifying multiple variables, including restrictions on cell shape, matrix compliance, biochemical cues and metabolites, and even the spatial orientation of the ECM, thereby obscuring definitive experimental conclusions regarding individual experimental parameters (Kleinman et al., 1986; Paszek et al., 2005; Wozniak et al., 2003). Indeed, the engineering of surrogate tissues and the development of tractable approaches to diagnose and treat chronic-debilitating diseases such as cancer and atherosclerosis require both a comprehensive understanding of tissue-specific behavior at the molecular level and highly reproducible systems. Accordingly, considerable effort has been expended toward developing synthetic biomaterials in which individual material properties such as cell shape, matrix presentation (2D vs 3D), ligand density, and elastic modulus can be precisely modulated (Chen et al., 1998; Engler et al., 2004; Tan et al., 2003; Yamada et al., 2003). By applying one of the defined systems in which matrix compliance and ligand density could be rigorously controlled, the critical role of matrix stiffness and integrin adhesions as a key regulator of multicellular mammary epithelial cell (MEC) tissue morphogenesis and malignant transformation has been highlighted (Paszek and Weaver, 2004; Paszek et al., 2005). In this chapter, we discuss the rationale for applying well-defined 3D organotypic culture assays to study adhesion-dependent cell behavior. We describe the use of 3D MEC organotypic cultures to illustrate how matrix compliance plays a critical role in regulating mammary tissue function and signaling. Finally, we outline experimental methods to generate, manipulate, and study the effect of matrix stiffness and matrix dimensionality on epithelial tissue morphology and signaling, and discuss technical limitations of currently available systems and future opportunities for improvement.
During embryogenesis, the epithelium originates from the endoderm and ectoderm and develops into a specialized tissue whose primary functions in the organism are to protect and to control permeation or transport. Unlike skin and esophagus, which are stratified epithelia that provide a critical barrier, the secretory epithelium is composed of a simple layer of epithelial cells lining tubes and ducts, whose principal function is to facilitate secretion and transport of biological materials. In vivo, the secretory epithelium abuts on and is surrounded by a stroma, which consists of cellular and noncellular components, including ECM molecules, soluble factors, and various stromal cells such as fibroblasts, adipocytes, and endothelial cells. Directly interacting with the epithelium is the basement membrane (BM), which is a specialized, highly organized ECM composed primarily of laminins 1, 5, and 10, collagen IV, entactin, and heparin sulfate proteoglycans. The BM in turn intersects with the interstitial matrix, which consists of collagens I and III, fibronectin, tenascin, elastins, and various proteoglycans including lumican, biglycan, and decorin (Kleinman et al., 1986). Collectively, the various components of the ECM and stroma provide biochemical (composition) and biophysical (structural modification and organization) cues to the epithelium and operate in concert with soluble factors released from the resident stromal cells to maintain the epithelium’s organ-specific function. Perturbations in stromal–epithelial interactions and altered epithelial organization are hallmarks of cancer and many chronic degenerative diseases. Moreover, disrupting tissue organization or altering ECM integrity precipitates disease, and restoring tissue structure or proper ECM interactions normalizes tissue behavior (reviewed in Hagios et al., 1998; Jeffery, 2001). Accordingly, the goal of 3D organotypic culture models is to recreate tissue-specific interactions, organizations, functions, and behavior ex vivo through prudent control of the biochemical and biophysical properties of the ECM, in order to understand the role of stromal–epithelial interactions and tissue structures in tissue-specific functions.
Mammary gland organotypic culture models have been used effectively to study the role of stromal–epithelial and ECM interactions in tissue-specific differentiation (Debnath et al., 2003; Petersen et al., 1992; Weaver et al., 1996; Wozniak et al., 2003). Unlike other tissues, the mammary gland undergoes unique developmental cycles in the adult organism and the gland can be readily accessed and manipulated in vivo and in culture. Additionally, reasonable quantities of breast tissue can be isolated and propagated ex vivo for culture experiments. As such, much of what we know regarding ECM-dependent epithelial differentiation has been derived from organotypic cultures of primary and immortalized MECs. Early studies demonstrated that MECs grown as 2D monolayers on rigid tissue culture substrates or within a physically constrained collagen I gel fail to assemble tissue-like structures (acini) and differentiate [no detectable expression of differentiated proteins such as whey acidic protein (WAP) or β-casein], despite the availability of appropriate growth factors and lactogenic hormones (reviewed in Roskelley et al., 1995). Yet, when the same MECs are grown within unconstrained collagen I gels and allowed to deposit and organize their own endogenous BM, or are embedded within a compliant reconstituted BM (rBM), they are able to assemble multicellular tissue-like structures (acini; reminiscent of terminal ductal lobular units in tissues in vivo) and differentiate in response to hormonal cues (expressed β-casein and WAP; reviewed in Roskelley et al., 1995; Fig. 1). Further studies using murine and human MECs have also consistently shown that the composition and spatial context of the ECM profoundly influence the responsiveness of an epithelium to exogenous growth, migration, and death stimuli (Wang et al., 1998, 2005; Weaver et al., 2002). For example, some human luminal epithelial breast tissues in vivo express the estrogen receptor (ER) and proliferate in response to hormonal fluctuations in estrogen. When these MECs are isolated and cultured on tissue culture plastic, they spread to form raised ER-negative, 2D cobblestone monolayer colonies that lack estrogenic responsiveness. However, if the isolated MECs are instead grown in the context of a compliant rBM, they retain their ER expression and maintain their estrogenic responsiveness (Novaro et al., 2003). Likewise, undifferentiated MECs grown on tissue culture plastic are highly sensitive to exogenous death cues, whereas their rBM-differentiated counterparts exhibit extremely high resistance to multiple apoptotic stimuli (Weaver et al., 2002). Analogous observations regarding the importance of biochemical and biophysical ECM cues for epithelial morphogenesis and tissue-specific differentiation have also been reported for thyroid, salivary gland, and kidney epithelia studies (Kadoya and Yamashina, 2005; O’Brien et al., 2001; Yap et al., 1995).
Many important discoveries have been made concerning the molecular mechanisms by which the ECM influences epithelial behavior, including the requirement of signaling through laminin-dependent ligation of α3β1 and α6β4 integrins. In addition, cooperative ERK-PI3 kinase and RacGTPase-NFκB signaling through epidermal and insulin growth factor receptors and prolactin-dependent activation of Stat3 have been identified as key biochemical events involved in directing MEC growth, survival, and differentiation (Akhtar and Streuli, 2006; Muschler et al., 1999; Paszek et al., 2005; Zahir et al., 2003). The ECM not only influences epithelial behavior through biochemical signaling but also through the mechanical properties of the microenivornment.
Early studies with constrained versus released collagen gels revealed the importance of ECM mechanics in directing the cell shape of MECs to promote differentiation (Emerman and Pitelka, 1977). MECs plated on constrained collagen gels or gluteraldehyde-crosslinked rBM fail to differentiate in response to lactogenic stimuli and instead spread to form a 2D cell monolayer despite appropriate integrin-ECM ligation and growth factor signaling (reviewed in Roskelley et al., 1995; Weaver and Bissell, 1999). Furthermore, laminin- and proteoglycan-mediated ligation of dystroglycan (DG) has been strongly implicated as the primary mediator of ECM-directed cell shape fate determination in MECs and as a critical component in establishing a continuous BM (Muschler et al., 1999). The hypothesized mechanism seems to depend only on DG’s extracellular domain and to involve DG binding to laminin, which then polymerizes on the cell surface and onto adjacent DG-expressing cells, ultimately establishing a continuous BM. This process of ligation-driven BM assembly is almost certainly in competition with integrin-based and other adhesive processes. The latter seem more mechanosensitive and might dominate on rigid substrates versus soft substrates.
Although the detailed molecular mechanisms of the mechanosensitivity of MEC differentiation remain to be delineated, recent studies using both nontransformed and transformed human MECs suggest that Rho GTPase-dependent cell contractility regulates adhesion-directed, cell shape-dependent, epithelial tissue-specific functions (Paszek et al., 2005). Transformed human mammary epithelial tumor cells propagated on top of constrained collagen I gels assembled aberrant invasive structures with high Rho and ROCK activity, whereas they could form cell aggregates reminiscent of nontransformed tubules when grown in unconstrained collagen I gels (Keely et al., 1995). In concert with these in vitro observations, transformed mammary tumors were recently shown to exhibit enhanced Rho GTPase activity and exert elevated myosin-dependent cell contractility and aberrant integrin adhesions when compared to nontransformed MECs. Normalizing tumor cell contractility through application of pharmacological inhibitors of Rho, ERK signaling, or myosin could phenotypically revert the malignant phenotype (Paszek et al., 2005). Consistent with a critical role for matrix compliance in epithelial behavior, nontransformed MECs grown within highly compliant rBM gels or nonconstrained and compliant collagen I/rBM gels competently assemble polarized, growth-arrested acini-like structures. However, when grown within constrained collagen I/rBM gels or collagen I/rBM gels of higher concentration and stiffness, they form progressively disrupted, disorganized, and continuously proliferating colonies (Paszek et al., 2005; see also Sections III.A.1 and III.A.2).
Through the application of defined, synthetic, rBM-crosslinked polyacrylamide gels, it was concluded that matrix stiffness and not matrix density or physical presentation constitutes a critical regulator of multicellular epithelial morphogenesis (Paszek et al., 2005; Fig. 2). These studies clearly emphasize the importance of myosin contractility and integrin adhesion maturation as matrix-regulated cell shape and force regulators. They have also identified altered ERK-dependent cell growth and survival, destabilization of cell–cell adhesions, and perturbed matrix assembly, as central mechanisms for further study. Indeed, the proper assembly of an endogenous cell-derived matrix plays a key role in epithelial differentiation, as has been illustrated by the necessity of proper laminin–nidogen interactions for mammary tissue differentiation and gene expression in culture (Pujuguet et al., 2000) and for multiple epithelial tissues including the kidney in vivo (Willem et al., 2002). Indeed, in lung development, increasing the compliance of the chest wall or decreasing the skeletal muscle fibers that aid in breathing modifies the biophysical properties of the tissue microenvironment by decreasing the applied force to the developing lung, leading to a decrease in lung growth, which further perturbs the tissue ECM and compromises tissue function (reviewed in Liu et al., 1999).
Key to engineering tissue-specific function is the application of an appropriate ECM in which the biochemical, biophysical, and spatial cues can be defined and controlled. An array of natural ECMs and a growing list of synthetic biomaterials, each with advantages and disadvantages, are available to the experimentalist. Ideally, a comprehensive assessment of what constitutes normal ECM composition, mechanical properties, and organization should be taken into consideration. Unfortunately, our comprehension of these variables has lagged behind, due to the complexity, lack of homogeneity, and anisotropy of biological materials.
rBMs isolated from Engelbreth–Holm–Swarm (EHS) mouse sarcomas have been routinely used to assemble tissue-like structures in culture and have been successfully applied to study mammary, thryoid, salivary gland, lung, and kidney epithelial cell morphogenesis and differentiation, and to distinguish between normal and transformed epithelial cells (Azuma and Sato, 1994; Debnath et al., 2003; Nogawa and Ito, 1995; O’Brien et al., 2001; Petersen et al., 1992; Yap et al., 1995). Similarly, fibrin gels have also been successfully used to assemble 3D normal and transformed tissue-like structures in culture (Alford et al., 1998). However, given that rBM is directly isolated from tissues, the matrix is inherently complex, poorly defined, and subject to complications with lot to lot variability and limitations due to the specific nature of the biochemical and biophysical environment associated with sarcomas. Fibrin gels, while attractive, also suffer from preparation variance. Additionally, fibrin gels are easily proteolyzed by cell-derived MMPs and consequently are not viable for long-term culture experiments. While alternative fibronectin sources that are less proteolytically sensitive have proven useful, these matrices have yet to be routinely applied to epithelial organ culture models.
As an alternative, collagen I gels have been extensively used as a 3D tissue matrix. The application of defined collagen gels to replace the more complex and biologically accurate rBM and fibrin gels has several advantages, including the fact that collagen I is a more biologically defined substrate, is relatively inexpensive to prepare or purchase, and is much more readily available. Because collagen I is the most common protein found in vertebrate animals and is structurally highly conserved, it is generally well tolerated for in vivo studies, and multiple cell types readily adhere to this substrate. In addition, the elastic moduli of a collagen I gel can be readily manipulated by varying collagen orientation, fibril crosslinking, concentration, or even biochemical modification or mutation (Christner et al., 2006; Girton et al., 1999; Martin et al., 1996; Roeder et al., 2002), thereby increasing its biological versatility (Elbjeirami et al., 2003; Grinnell, 2003). The magnitude and directional orientation of externally imposed tension can also be easily manipulated with collagen preparations. For example, through the release of collagen gels from the culture vessel, the isometric tension within the gel can be dramatically reduced (Rosenfeldt and Grinnell, 2000). Collagen gels can also be biochemically modified to facilitate epithelial functionality, as for example through the addition of either rBM, purified laminin, or derivatized peptides (Gudjonsson et al., 2002).
Purified, biologically derived materials, such as rBM and collagen I, have an intrinsic amount of biochemical and biophysical variability due to the inherent variability between animals and preparations. This variability leads to inconsistencies between experiments, as well as a high degree of heterogeneity within single gels. Additionally, the dynamic range of elastic moduli that can be reasonably achieved with these systems is limited by biochemical and biophysical constraints of these unique macromolecules. Therefore, although these materials have proven to be useful for clarifying the general influence of matrix on cell and tissue phenotypes, they are not as tractable for defining precise molecular mechanisms mediating mechanotransduction.
To address the issues listed above, especially control over matrix compliance, we and many others use a system first developed by Pelham and Wang (1997) that involves functionalizing synthetic polyacrylamide gels for cell culture (by cross-linking them with precise concentrations of ECM ligands) as 2D model systems for cell spreading, adhesion, and migration. Polyacrylamide gels represent tractable materials to allow studies of molecular pathways and signaling events of cells grown in various mechanical environments. The mechanical properties of these gels, which have been defined using rheology and atomic force microscopy (Engler et al., 2004; Guo et al., 2006; Yeung et al., 2005; Chapter 22 by Engler et al., this volume), can be manipulated by changing the relative concentration of acrylamide and the crosslinker, bis-acrylamide, yielding a system with precisely controlled biochemical and biophysical properties. Polyacrylamide is an exemplary material for studying cell behavior, as it is nonreactive, resistant to nonspecific binding and protein adhesion, and optically clear. The most significant downside to the polyacrylamide gel system is that acrylamide is cytotoxic in its monomeric form, which precludes the extension of its use to 3D cultures in which cells are embedded before polymerization. To overcome this limitation, we have used these polyacrylamide gels to reconstitute 3D conditions by overlaying MECs plated on top of rBM-crosslinked polyacrylamide gels with a blanket layer of rBM. Although the cells undergo normal morphogenesis under these conditions, there are some limitations inherent in this unique technique. Namely, and most importantly, this is neither a true 2D nor a complete 3D system, and the cells behave differently than they do in full 3D cultures (Leight et al., unpublished observations). Although this drawback leads to difficulty in interpretation and definition of these experiments, this system is suitable for approximating the physiological mechanical conditions under which epithelial cells grow and thrive. Alternatively, polyethylene glycol gels combined with bioactive peptides, such as fibronection- and laminin-binding sites, are also attractive biomaterials. However, their 3D organization is significantly different from that found in naturally occurring matrices and in vivo in that they typically have a greater matrix density and altered spatial orientation (reviewed in Zhang, 2004). Furthermore, because matrix remodeling is a critical aspect of epithelial morphogenesis, expensive bioactive peptides that can be proteolytically remodeled need to be incorporated into these synthetic biomaterials to permit proper tissue morphogenesis, migration, and to support long-term cell and tissue viability (reviewed in Lutolf and Hubbell, 2005). As an attractive new strategy in the arsenal of synthetic materials, novel matrices that incorporate recombinant natural and synthetic proteins and biopeptides are currently being developed and offer new hope for future applications.
Progress has been made in recapitulating tissue-specific morphology ex vivo either for tissue transplantation or for the study of tissue-specific function, but the application of these organotypic model systems to dissect the molecular basis of tissue homeostasis and disease has lagged behind significantly. The failure to exploit current 3D model systems for the study of cell behavior and signaling in the context of a tissue-like microenvironment and structure resides primarily in the lack of appropriate, cost-effective, easy, and reproducible strategies to manipulate, analyze, and assess cell function, signaling, and gene expression in these model systems. We have been studying the effect of cell shape, matrix compliance, adhesion, and dimensionality on cell behavior at the molecular levels, and here we provide a detailed description of the methods we have successfully used to do so.
Anticipated results: After 14 days in culture, MECs plated on top of functionalized polyacrylamide gels with a rBM blanket layer, similar to cells overlaid on rBM gels (Debnath et al., 2003), form larger acini than their counterparts embedded within natural matrices. Analogous to cells grown within a natural matrix or grown on top of rBM with an overlay of rBM, by day 4, cells grown on the rBM PA gels should have acquired detectable cell–cell E-cadherin/β-catenin junctions as well as basal and apical polarity. Thus, by day 4, the cells should be highly proliferative but have acquired basal polarity, detectable by basally localized β4 integrin and deposition of a laminin-5- and collagen IV-rich endogneous BM as well as apically localized cortical actin. Studies have revealed that while the growth rate of MCF10As plated on soft (E = 140 Pa) and stiff (E = 5000 Pa) rBM-functionalized polyacrylamide gels are similar, cells interacting with a matrix stiffer than 1000 Pa fail to fully growth arrest. We have previously shown that MECs plated with a 3D rBM blanket layer on soft rBM-functionalized polyacrylamide gels undergo normal morphogenesis, while morphogenesis is perturbed in those plated on a stiff gel under the same conditions (Paszek et al., 2005; Fig. 2). A protocol outlining basic immunofluorescence techniques for each cell culture method is described in the Section III.E, for the visualization of characteristics indicative of morphogenesis. A comparison detailing morphogenetic characteristics of all of the described methods is currently in progress (Leight et al., unpublished observations).
Epithelial tissues are highly complex, organized 3D structures that evolve incrementally during development to generate these specialized functional tissues through spatially and temporally controlled stromal–epithelial interactions. The tissue microenvironment of the epithelium is composed of multiple stromal cell types, and these cellular components, together with the epithelium, are embedded within a proteinaceous ECM. It is the combination of cellular and ECM interactions, operating through controlled biochemical and physical cues, that ultimately regulates epithelial cell fate and function. The goal of an epithelial experimentalist is to recreate at least some of the intricate relationships that exist between the various cell types and the ECM in vivo, but in a simple format in culture, such that the recreated system is more amenable to molecular studies without severely compromising the epithelial cell’s normal tissue behavior. The idea is that experimental observations made using such contrived but simplified systems will ultimately be distilled into the critical information that is necessary to systematically engineer surrogate tissues for replacement therapy or to develop tractable treatments to prevent and cure various diseases. Toward this lofty goal, considerable research has been successfully directed at determining how each individual cell variable and microenvironmental component influences epithelial cell behavior (Mostov et al., 2005; Paszek et al., 2005; Petersen et al., 1992).
Despite the efforts, our understanding of what controls the epithelial cells’ behavior within the complex 3D tissue-like structure and how combinations of microenvironmental cues might cooperate to influence epithelial function remains rudimentary at best. Moreover, although we and others have been successful in generating functional data using these “crude” systems, it remains difficult to isolate specific responses to allow the identification of the precise molecular mechanisms linked to the generation of a given tissue phenotype. For example, MECs grown within a 3D rBM simultaneously and acutely change their shape, matrix adhesion, cell contractility, and signaling, as well as growth factor and apoptosis responsiveness, as compared to MECs interacting with a 2D rigid substrate (Debnath et al., 2003; Wang et al., 1998; Weaver et al., 2002). To address this difficulty, tractable culture models that reproducibly reconstruct individual aspects of tissue organization and function and that encompass controllable homotypic and heterotypic cell–cell interactions and ECM cues are needed. Preferably, these newly engineered culture systems will be amenable to precise biochemical and physical manipulation and will be sufficiently robust and versatile for routine experimentation. Additionally, they should be inexpensive and lend themselves to easy and reproducible manipulation.
Conventional organotypic systems are often expensive, labor intensive to generate, and suffer from experimental inconsistencies. It is now feasible to synthesize biocompatible matrices to study the effect of individual parameters, such as matrix binding and ECM orientation, receptor expression and activity, cell shape, matrix compliance, and even ECM dimensionality through a combination of nonreactive hydrogels with cell-adhesive sites. Recombinant synthetic proteins have also been used to promote specific adhesion and to foster cell-specific degradation and remodeling of the matrix by incorporating proteolytically degradable peptide sequences. We and others have applied similar strategies to successfully study the phenotypic behavior of individual cells in response to various physical, architectural, and biochemical cues including issues pertaining to the regulation of cell survival (Buckley et al., 1999; Capello et al., 2006; Chen et al., 1997; Friedland et al., unpublished observations), migration (Gobin and West, 2002; Wong et al., 2003), stem cell fate (McBeath et al., 2004), differentiation (Bokhari et al., 2005; Mauck et al., 2006), and growth regulation (Bokhari et al., 2005; Georges et al., 2006; Paszek et al., 2005). However, such specialized systems have limited application for studying cell behavior in multicellular structures and 3D tissues and have only sparingly been applied to the study of heterotypic cell–cell interactions (Georges et al., 2006). Moreover, many of the currently available synthetic biomaterials exhibit incompatible material properties such as high stiffness, elevated matrix density, and random matrix presentation that render them less than suitable for the study of epithelial tissue morphogenesis (reviewed in Zhang, 2004). To address these concerns, newer generations of biomaterials are currently being developed, including highly compliant synthetic matrices generated using combinations of polyethylene glycol and methylcellulose conjugated with various bioactive peptides and MMP-cleavable proteins (Leach JB, personal communication), polyethylene glycol gels with functionalized recombinant proteins (Rizzi and Hubbell, 2005), electrically spun collagen gels with precisely controlled orientations (Matthews et al., 2002), and synthetic gels with gradients of ECM compliance that recreate durotactic-directed cell migration during development, wound closure, and tumor metastasis (Lo et al., 2000; Wong et al., 2003; Zaari et al., 2004). The application of these novel materials together with the availability of pluripotent and tissue-specific stem cells provide encouragement that we are at least moving closer to our idealized model systems, to begin to elucidate the mechanisms regulating multicellular epithelial tissue-specific structure and function.
In addition to these important considerations, it is recognized that tissues develop progressively and evolve through reciprocal and dynamic dialogues between the cellular and stromal components and tissue milieu, and this temporal relationship must also provide the mature tissue with physiological advantages that need to be identified and assessed. For example, although bioengineers have been able to successfully reconstruct blood vessels that are phenotypically and functionally identical to differentiated arteries in vivo, the engineered vessels rapidly fatigue when transplanted into a host in vivo. One must also consider that our ultimate goal should be the engineering of complex 3D microenvironments that are amenable to dynamic physical and biochemical modification. When seeded with pluripotent and tissue-specific stem cells, they should allow systematic development ex vivo of viable, live tissues to be used for routine and faithful experimentation and for various clinical applications. Clearly, we have our work cut out for us.
We thank J. C. Friedland and J. N. Lakins for their contributions. This work was supported by NIH grants CA078731 and BRP HL6438801A1 (to V.M.W.) and T32HL00795404 (to K.R.J.), DOD grants W81XWH-05-1-330 and DAMD17-01-1-0367 (to V.M.W.), and a NSF graduate fellowship (to J.L.L.).