Micrometer length scale

The fundamental structural unit providing physical resilience to cells and the extracellular environment is a fiber, which may range from tens of nanometers to multiple micrometers in diameter. While stiffness is the usual subject of investigation, other properties such as strain at failure and toughness have also been studied for biological fibers.^{Reference Guthold, Liu, Sparks, Jawerth, Peng, Falvo, Superfine, Hantgan and Lord19} The stress versus strain relationship is a relatively simple measurement at the macro-length scale, however, these measurements are tremendously challenging for fibers with a size of only a μm or less. Thus, many of the transformative studies in this field resulted from the development of a unique tool or approach to measure force and displacement at microscopic lengths. For example, fibrin fibers, which form the three-dimensional networks that form blood clots, were first characterized for their remarkable extensibility by depositing them on microtextured surfaces and using an atomic force microscope (AFM) combined with a fluorescence microscope to displace the fibers within the image plane.^{Reference Liu, Jawerth, Sparks, Falvo, Hantgan, Superfine, Lord and Guthold20} We elaborate on collagen here due to its prevalence in vivo and Fn due to its requirement for development^{Reference George, Baldwin and Hynes21,Reference George, Georges-Labouesse, Patel-King, Rayburn and Hynes22} and increased expression during the progression of some diseases such as atherosclerosis.^{Reference Astrof and Hynes23}

Collagens are well-studied materials due to their robust presence in numerous tissues. Some tissues, such as tendons and ligaments, have mechanical properties that are predominantly determined by their constituent collagen fibers and the cross-linked network they compose. There are many types of collagens, a fraction of which exist in a fibrillar state. Although most collagens form non-fibrillar meshes, we consider Type I collagen here since it is the most abundant protein in mammals. Type I collagen is the structural component of scars, tendons, and ligaments, and it is also abundant in corneas, blood vessels, and intervertebral discs. While this material is a major load-bearing structure in many tissues, it also plays an important function in defining the local rigidity sensed by cells. Collagens surround cells in many tissues and are contained within the thin sheets of connective tissue that underlie the epithelial cells that line the skin, blood vessels, and gut.

Collagen fibers have a complex hierarchical structure and are composed of fibrils that are themselves composed of microfibrils.^{Reference Banos, Thomas and Kuo24} Although mechanical measurements of collagen-rich tissues such as tendons could be used to infer collagen fibril properties, the microscopic properties of collagen need to be measured in order to link micro- and macroscale properties. The mechanical properties of single Type I collagen fibrils have been measured using a variety of techniques, generally utilizing a stretching technique with AFM, on single fibrils derived from a variety of sources.^{Reference Yang, Fitie, van der Werf, Bennink, Dijkstra and Feijen25–Reference Silver, Freeman and Seehra31} These studies are often complemented by simulations.^{Reference Buehler32,Reference Buehler33} In the hydrated state, these studies generally report elastic moduli between 0.1 GPa and 1.0 GPa and strains at failure less than 50%. These values are quite high when compared to the soft rigidities at which cells display dynamic rigidity sensing, discussed further below.

Fn fibers differ from collagen in that they are soft, but highly extensible. The first direct observations of the large extensibility of Fn fibers were made using cultured Chinese hamster ovary cells in culture that were transfected with green fluorescent protein-tagged Fn; thus, the Fn matrix formed by these cells was fluorescent and could be visualized using time-lapse microscopy.^{Reference Ohashi, Kiehart and Erickson34,Reference Ohashi, Kiehart and Erickson35} Time-lapse movies of Fn fibers in the ECM of cells expressing a fluorescent Fn matrix indicated that the Fn fibers are highly extensible and could be stretched to three to four times their stress-free length before mechanical failure. Such extreme extensibility supports the notion that Fn fiber stretch would alter Fn conformation in the fiber, thus altering its biochemical properties, and indeed this hypothesis is supported by a variety of recent findings.^{Reference Chabria, Hertig, Smith and Vogel36–Reference Cao, Zeller, Fiore, Strane, Bermudez and Barker39} Thus, the material properties of Fn are complicated by the strain dependence of its biological function.

Regarding rigidity measurements, the first mechanical stretch experiment on Fn was motivated by the microtextured platforms used for collagen and fibrin,^{Reference Liu, Jawerth, Sparks, Falvo, Hantgan, Superfine, Lord and Guthold20,Reference Liu, Bonin and Guthold40,Reference Carlisle, Coulais and Guthold41} and it was confirmed that Fn is highly extensible with strain at failure greater than 600%.^{Reference Klotzsch, Smith, Kubow, Muntwyler, Little, Beyeler, Gourdon, Nelson and Vogel42} Furthermore, it was shown that Fn fibers strain harden from approximately 50 kPa at the onset of mechanical strain to more than 1 MPa at failure. However, detailed analysis of Fn fiber plasticity has yet to be performed. This behavior may have important implications in vivo since many cell types respond to gradients in stiffness.^{Reference Isenberg, Dimilla, Walker, Kim and Wong43,Reference Lo, Wang, Dembo and Wang44} In this context, cells can sense gradients in stiffness and respond by migrating up the stiffness gradient, a process termed durotaxis. Fn fiber mechanical properties have also been studied through computational approaches.^{Reference Bradshaw and Smith3,Reference Bradshaw, Cheung, Ehrlich and Smith45–Reference Peleg, Savin, Kolmakov, Salib, Balazs, Kroger and Vogel49} Other properties of Fn fibers, for example the density of binding sites for cell signaling molecules or cell adhesion receptors, may also be strain dependent. Using combined modeling and spectroscopic imaging, we recently estimated that the density of ligands (arginine-glycine-aspartate [RGD]^{Reference Klotzsch, Smith, Kubow, Muntwyler, Little, Beyeler, Gourdon, Nelson and Vogel50}) for some integrin cell adhesion molecules on the surface of Fn fibers is 1400 RGD/μm^{Reference Vogel and Sheetz2}, but this value decreases as the material is strained.^{Reference Bradshaw and Smith3,Reference Bradshaw, Cheung, Ehrlich and Smith45} Thus, tissues with gradients in Fn fiber tension may direct cell migratory behavior through both durotaxis and gradients in the density of cell adhesion ligands, which also can direct cell migration through a process termed haptotaxis.

The previous examples of collagen and Fn indicate a broad range of rigidities for individual fibers. However, these fibers are assembled into complicated networks that may provide a very different apparent rigidity to resident cells. In simple scenarios, such as a tendon or ligament, collagen fibers may be aligned such that the mechanical properties of the tissue are anisotropic, but the apparent rigidity of the collagen ECM in these tissues, when loaded along the fiber axis, should closely approximate the fiber rigidity. In more complicated network structures such as skin, fibers of multiple types are not uniformly aligned. Thus, resident cells sense an apparent rigidity that is the result of a composite matrix of fibers and that may be heavily dependent upon the strain or pre-stress of the tissue. Furthermore, the extent of cross-linking within the matrix can dramatically change the apparent rigidity of the matrix, both through alterations of the properties of single fibers^{Reference Kubow, Klotzsch, Smith, Gourdon, Little and Vogel51} and through alterations in the extent of interconnectedness between fibers within the network.^{Reference Marturano, Arena, Schiller, Georgakoudi and Kuo52} For example, lysyl oxidase, which is expressed in tissues that contain collagen, cross-links collagen matrix fibers by catalyzing the formation of aldehydes from lysine residues that spontaneously react with lysines on adjacent collagen fibers.^{Reference Marturano, Arena, Schiller, Georgakoudi and Kuo52–Reference Robins54} Similarly, stimuli such as mechanical injury and inflammation can lead to activation of transglutaminase 2, which catalyzes the formation of an isopeptide bond between glutamine and lysine residues on matrix components such as Fn and collagen.^{Reference Belkin55} It is not clear how different cross-linking mechanisms differ with regard to the resulting ECM stiffness, but it is clear that increased cross-linking leads to increased stiffness. These are important considerations for larger length scales as well.

Tissue length scale

Tissue rigidity is an important component of the regulatory system driving normal tissue homeostasis and dynamic processes such as tissue development and disease progression. For example, the rigidity of tumors and the layer of connective tissue that surrounds the borders of tumors have been shown to increase or decrease in different tumor systems.^{Reference Plodinec, Loparic, Monnier, Obermann, Zanetti-Dallenbach, Oertle, Hyotyla, Aebi, Bentires-Alj, Lim and Schoenenberger56} Other studies indicate cancerous cells may have a compromised ability to sense rigidity.^{Reference Liu, Tan, Zhang, Zhang, Xu, Chen, Poh, Tang, Wang and Huang9,Reference Macpherson and Montagnier57} The full impact of this loss of rigidity sensation is still being investigated, but this finding indicates that tissue rigidity is an important regulator of tissue function since the desensitization of cancer cell sensitivity may allow the disease to progress. We focus on the dynamic changes in rigidity resulting from dynamic matrix remodeling that occurs in developing tendon, as highlighted later.

Embryonic tendons begin developing as highly cellular structures that increase in ECM protein content, collagen cross-linking, and matrix organization during development.^{Reference Marturano, Arena, Schiller, Georgakoudi and Kuo52,Reference Marturano, Xylas, Sridharan, Georgakoudi and Kuo53,Reference Kuo, Petersen and Tuan59,Reference Brown, Lind, Burzesi and Kuo60} In the contexts of embryonic tissue development and engineered tissue formation, mechanical property increases are typically attributed to increases in collagen and other ECM protein content, such as glycosaminoglycans (GAG). We recently used force volume (FV)-AFM to characterize the topography and elastic modulus of developing embryonic tendon at the nano- and microscales, demonstrating discrete and significant increases in tissue modulus and matrix organization as a function of developmental stage^{Reference Marturano, Arena, Schiller, Georgakoudi and Kuo52} (Figure 2). Increases in collagen and GAG content, cross-link density, and matrix density and alignment were all confirmed via biochemical assays, histological staining, second harmonic generation imaging, and mass spectrometry analyses.^{Reference Marturano, Arena, Schiller, Georgakoudi and Kuo52,Reference Marturano, Xylas, Sridharan, Georgakoudi and Kuo53} Additionally, immunohistochemical characterization showed that the spatial distribution and density of ECM proteins as well as cell receptors for biochemical factors change during development.^{Reference Marturano, Arena, Schiller, Georgakoudi and Kuo52,Reference Kuo, Petersen and Tuan59,Reference Brown, Lind, Burzesi and Kuo60} These studies demonstrate that embryonic tendon tissue is dynamic in its biochemical composition (ECM molecules, adhesive ligands, and soluble factors) and presentation (spacing and density), which is consistent with the general understanding that cell differentiation during development in vivo is regulated by various biochemical factors. However, discrete increases in nano- and microscale tissue rigidity and architecture as a function of embryonic developmental stage^{Reference Isenberg, Dimilla, Walker, Kim and Wong43} suggest that cells are also responding to mechanical cues during development. The elucidation of mechanical cues on progenitor cell behavior is not easily studied in vivo due to various confounding chemical and physical influences. Thus, researchers have turned to engineered 2D and 3D in vitro systems to investigate the contributing influences of matrix rigidity on cell behavior.

Figure 2.

(a) Force volume atomic force microscopy (FV-AFM) topography (top row; scale 2 µm), nanoscale elastic modulus maps (second row; scale 2 µm), second harmonic imaging (SHG; third row; scale 2 µm), and trichrome staining (bottom row; scale 20 µm) of chick embryonic tendon at the Hamburger Hamilton stages of development, HH 28–43. FV-AFM images showed increasing heterogeneity in structure and elastic modulus during development. SHG imaging showed increases in collagen density and organization during tendon development. Trichrome staining showed a decrease in cellularity (red) and increase in collagen content (blue) with development. (b) Tendon elastic modulus measured via FV-AFM as a function of developmental stage, calculated with agarose gel standards. Nano- and microscale moduli increased non-linearly during development, with the greatest increases between HH 28–30 and after HH 38 (N = 5 different animals). Figure adapted with permission from Reference 52.

2D platforms

Though they are over-simplified representations of the matrix environment, 2D systems have advanced our understanding of mechanical influences on cell behavior. Tunable properties at the micron and nanometer length scales enable regulation of cell behavior at the cellular and subcellular levels, respectively, via biochemistry, topography, and rigidity. Studies on these customized 2D substrates have significantly advanced our understanding of proteins involved in cell-cell and cell-matrix adhesion,^{Reference Chen, Tan and Tien1,Reference Leckband, le Duc, Wang and de Rooij65,Reference Burute and Thery66} the role of cytoskeletal tension,^{Reference Schwarz and Gardel67} and mechano-regulated signaling pathways.^{Reference Ingber68}

Numerous studies have demonstrated that the mechanical properties of a 2D substrate can guide differentiation of stem cells toward various tissue lineages, including bone, muscle, and nerve,^{Reference Engler, Sen, Sweeney and Discher69–Reference Young, Choi, Engler and Christman73} which has important implications for tissue engineering and regenerative medicine. However, while there is still much to learn about the specific effects of rigidity on signaling pathways and subsequent cell responses, in vivo cell fate is likely influenced by a combination of matrix mechanics, adhesion ligand presentation, and tissue microstructure. Toward understanding cell responses in vivo, researchers have demonstrated that ligand composition and presentation on a substrate can couple with rigidity to influence stem cell differentiation.^{Reference Trappmann, Gautrot, Connelly, Strange, Li, Oyen, Cohen Stuart, Boehm, Li, Vogel, Spatz, Watt and Huck74–Reference Klim, Li, Wrighton, Piekarczyk and Kiessling78} For instance, the presentation of collagen and Fn coupled with different substrate rigidities has been shown to differentially influence stem cell differentiation toward tendon and bone lineages.^{Reference Sharma and Snedeker75}

3D platforms

While some cells in the body exist on matrices that are somewhat 2D in nature (e.g., endothelial and many epithelial cell types), many cells in vivo exist within a 3D environment, surrounded by matrix or cell-cell contacts on all sides. A 3D system enables a spatial arrangement that more closely mimics the 3D environment in vivo in that adhesion occurs on all sides of the cell membrane, which markedly affects cell morphology, adhesion, and numerous related cell functions. Both 2D and 3D systems can elicit cell responses, including cell proliferation, motility, differentiation, tumorigenicity, and response to ECM molecules and soluble factors.^{Reference Li, Livi, Gourd, Deweerd and Hoffman-Kim79–Reference Nii, Lai, Keeney, Han, Behn, Imanbayev and Yang87} Although most studies have been performed with 2D systems, a rapidly growing class of hydrogel systems is advantageous in that hydrogels can encapsulate cells in a 3D environment (Figure 3e), and their mechanical properties can be tuned by changing cross-link density, polymer concentration, and molecular weight.^{Reference Lutolf and Hubbell88,Reference DeVolder and Kong89}

Commonly utilized 3D systems that have been used to explore matrix stiffness-guided cell response are based on natural protein materials such as Matrigel, collagen, and fibrin. Cells bind to peptide sequences of the hydrogel’s matrix proteins and physically contract the gel. Cell traction has been shown to be integral to varying cell responses and can be modulated as a function of stiffness.^{Reference McBeath, Pirone, Nelson, Bhadriraju and Chen16,Reference Engler, Sen, Sweeney and Discher69} However, these natural protein hydrogel systems do not allow for de-coupling of matrix biochemistry and stiffness to explore alternative combinations. Therefore, the specific contributions of biological and physical factors are difficult to elucidate in these gels.

Synthetic (e.g., polyesters) and natural (e.g., alginate) polymers that do not contain cell-recognition sequences allow for tuning of ligand presentation via chemical functionalization of the polymer and mechanical properties via polymer concentration and cross-link density.^{Reference Kuo and Ma90} However, even in these more highly controlled 3D systems, cell-matrix interactions within the 3D environment are complex, affecting delivery of mechanical and biochemical cues as well as interpretation of cell responses.

For instance, increasing polymer concentration or cross-link density to increase matrix stiffness can also influence porosity and ligand density. In turn, while cells may sense a change in matrix stiffness, other parameters that may change include nutrient diffusion, cell morphology, and access to ligand binding sites. Another major challenge is to determine the mechanical properties actually experienced by cells. Cell traction is itself modulated by matrix compliance. In a more compliant gel that may yield when pulled upon, but simultaneously strain harden, cell traction can alter the matrix density surrounding a cell, consequently resulting in a higher local modulus than that at the original bulk scale. In addition to rigidity, local matrix compaction may also increase cell-ligand interactions, possibly affecting adhesion, morphology, migration, and biological signaling. Importantly, inconsistency in the amount of cell traction between gels of varying rigidities further compromises the ability to compare the effects of 3D matrix rigidity on cell behavior. Therefore, it is unclear what exactly is affecting the cells in a 3D system to elicit a response when mechanical properties are altered.

In the near term, further study is needed to more closely link the properties of natural ECMs with synthetic platforms. These studies should focus on elaboration of the complexity of natural ECMs. For example, single fiber studies have significantly advanced our understanding of the material properties of the building blocks of the ECM, but the ECM is a complex network of multiple types of these fibers. Thus, network effects of fibrillar mats as well as the impact of multi-component matrices are needed. Furthermore, the impact of the multiple cross-linking and degradation mechanisms should be studied in more detail to determine their effects on material properties of the ECM, thus complementing our current understanding of their biological effects. Finally, the interactions of the ECM with signaling molecules will likely be elaborated in great detail in the near future. Clearly the ECM controls cell functions directly, but the ability of these structures to bind and present cell signaling molecules such as growth factors may be a function with importance second only to the mechanical stabilizing properties of the ECM.