Biomechanical regulation of mesenchymal cell function

Myofibroblast differentiation and function

Fibroblasts help to establish tissue architecture and determine tissue mechanical properties through regulated deposition and maintenance of extracellular matrix. In addition, these cells also deform the matrix through cell-generated forces, and respond to mechanical cues, including exogenously applied stretch, and matrix tension and rigidity sensed through internally generated forces. Although multiple markers of diverse fibroblast phenotypes have been described, the best-studied phenotypic conversion is between fibroblast and myofibroblast states, typically associated with acquisition of α-smooth muscle actin (ASMA) expression and increased contractile function. Recent work by several groups has shown that fibroblast function and myofibroblast differentiation are tightly linked to the deformability (rigidity or stiffness) of the extracellular matrix [1,2,3▪,4▪;,5▪▪,6,7]. Intriguingly, several of these studies [1,2,5▪▪,6] indicate that the effects of matrix stiffness may be independent of profibrotic factors, such as transforming growth factor-beta (TGF-β), or may selectively promote the effects of these factors. The effect of matrix stiffness on contractile function of fibroblasts, which has been inferred from substrate wrinkling and three-dimensional gel compaction, has recently been quantified in a rigorous fashion [8▪]. These measurements revealed that cell-generated forces vary dramatically across matrix stiffness conditions, and confirmed that matrix stiffness alters forces over a range associated with contractile integrin-mediated liberation of active TGF-β (see section below on TGF-β). The discovery that matrix stiffening by itself promotes enhanced matrix deposition [2] has potentially profound implications for understanding progressive fibrosis. This discovery suggests that fibroblasts by themselves are not mechanically homeostatic, but rather that in the absence of external controls and in the presence of a permissive environment, these cells are intrinsically programmed for matrix production, contraction, and stiffening. Interestingly, other cell types also appear to be capable of expressing mesenchymal markers through an epithelial mesenchymal transition that is also dependent on matrix rigidity [9▪,10] and stretch [11], furthering amplifying the potential implications of matrix stiffening and mechanical forces for fibrotic pathologies.

Externally applied forces also provoke diverse fibroblast signaling responses, including activation of mitogen-activated protein (MAP) kinases [12], Akt [13], and focal adhesion kinase (FAK). Interestingly, stretch has been shown to augment TGF-β release and signaling [14▪,15], and promote myofibroblast and profibrotic phenotypes [16], but has also been shown to suppress myofibroblastic differentiation in lung fibroblasts [17▪]. These differences may reflect differences in loading specifics (cyclic vs. static), cell of origin (lung vs. dermal), or both. Most of the work on matrix rigidity and stretch responses originated in two-dimensional planar systems, and while early results suggest reasonable predictive power when translated into three-dimensional systems, and ultimately back into physiological contexts, more work is needed. In addition, it has only recently become possible to separately tune both matrix rigidity and stretch within the same system [18▪▪], and the possibilities for investigating these mechanical cues in tandem to enhance physiological relevance is appealing.

Endothelial barrier function and angiogenesis

Endothelial cells have long been recognized to sense and integrate biomechanical cues encoded in vascular flow and shear stress patterns. More recent work has focused on matrix rigidity as a key regulator of integrated endothelial functions. For example, Mammoto et al. [19] described a unique transcriptional program that regulates capillary network formation through the Rho inhibitor p190RhoGAP and the angiogenic growth factor receptor VEGFR2. The activity of this transcriptional program was shown to be sensitive to extracellular matrix elasticity, resulting in optimal angiogenesis and capillary network formation in a relatively narrow range of matrix stiffness. Subsequent work by Rao et al. [20] combined endothelial cells with mesenchymal stem cells to study vascular network formation in three-dimensional matrices. These experiments identified an inverse relationship between matrix stiffness and network formation. Further investigation using a matrix cross-linker to vary stiffness independently of matrix density confirmed the inhibitory effect of increasing matrix stiffness on vascular network formation. Together, these findings suggest that tissue stiffness must be within a permissible range to promote vascularization, which may have interesting implications for angiogenesis in wound healing and fibrosis.

Work from three groups [21▪▪,22,23] has shown that matrix rigidity also importantly regulates endothelial barrier function, and influences the transmigration of leukocytes across endothelial monolayers. Based on alterations in vessel wall stiffness with aging, such functional effects on endothelial barrier function could have important implications for the development of atherosclerosis. Given the importance of restoring endothelial barrier function during repair and resolution of injury, such biomechanical influences could importantly impact on the divergence between successful resolution of injury and persistent dysfunction leading to aberrant wound healing or fibrosis.

Mesenchymal stem cell fate decisions

Several groups have recently followed up on the seminal findings of Engler et al. [24] that indicated that extracellular matrix rigidity influences mesenchymal stem cell (MSC) fate decisions. Dupont et al. [25▪▪] showed that MSC differentiation along adipogenic vs. osteogenic lineages as determined by matrix rigidity was under the control of the transcriptional co-activators YAP and TAZ, and that exogenous control of these factors could override matrix stiffness cues. Park et al. [26] provided more context for this issue by showing that MSCs on stiff substrates express smooth muscle cell markers ASMA and calponin-1, whereas MSCs on soft substrates express both chondrogenic and adipogenic markers. They further examined the influence of TGF-β, which increased smooth muscle cell marker expression on stiff substrates, while selectively increasing chondrogenic marker expression and suppressing adipogenic marker expression on soft substrates. These data suggest that in addition to direct effects of matrix rigidity on cell fate, mechanical cues also strongly influence cellular responses to biochemical stimuli. Recently, Guvendiren and Burdick [27▪] developed a novel dynamic matrix system, which allows stiffness to be changed while cells are resident. They provided evidence for relatively early cell commitment to fate decisions under the influence of mechanical cues, consistent with other recent findings [28].

Transforming growth factor-β

TGF-β is synthesized and secreted as part of a latent complex that can be anchored to the extracellular matrix [29]. The latent complex includes a peptide sequence recognized by integrins, and a pivotal study Munger et al. [30] demonstrated that integrin-mediated activation of TGF-β is critical to pulmonary fibrosis. Recent work has reinforced the concept that physical forces dramatically influence TGF-β bioavailability and signaling [14▪], and have shown that cell-generated mechanical forces play a critical role in triggering bioavailability of active TGF-β1 from its extracellular matrix-bound latent complex [31,32▪▪,33▪▪]. The recent solution of the latent TGF-β crystal structure identified the preferred orientation of force that is needed to unfasten the ‘straitjacket’ that encircles each TGF-β monomer [33▪▪], and Buscemi et al. [32▪▪] recently estimated that ≈40pN force is needed to release TGF-β from the latency-associated peptide. Forces in this range are consistent with the biophysical limits of individual integrin-matrix linkages [34]. Fibroblasts cultured across a range of matrix stiffness conditions generate tractions that increase dramatically with matrix stiffness, reaching local values as high as 6000 Pa (or 6000 pN/µm²) in isolated areas [8▪]. Based on a recent estimate that the number of cell-substrate interacting structures is ≈170 per µm² within cell matrix adhesions [35], and assuming equivalent force distribution across 170 linkages, local tractions of 6000 Pa translates to 36pN per molecule, similar to the estimate of force needed for TGF-β activation (40 pN). Consistently with these estimates, Wipff et al. [31] observed a modest matrix stiffness-dependent increase in fibroblast activation of TGF-β1 of ≈1.3-fold between matrices of 5 and 20 kPa stiffness. They also used thrombin stimulation to induce fibroblast contraction, and observed increased TGF-β1 activation on 20 kPa matrices by ≈1.75-fold, but found no effect on 6 kPa matrices [31]. Such findings are consistent with recent observations that exogenous stimuli selectively augment cell tractions on stiff, but not soft matrices [8▪]. Taken together with the broad functional effects of TGF-β on matrix remodeling, these findings suggest the potential for an adverse feedback cycle of increased fibroblast contraction on stiff matrices, promoting TGF-β activation and further enhancing matrix deposition, with potentially important implications for fibrotic pathologies.

Wnt and β-catenin

The Wnt/β-catenin pathway is an evolutionarily conserved pathway critical for development and increasingly recognized to play an important role in fibrosis [45,46▪]. Recent evidence indicates that both exogenous stretch [11] and internal actomyosin-generated forces [47▪▪] can lead to signaling through the Wnt/β-catenin pathway. Though the mechanisms by which mechanical cues are transduced into activation of this signaling pathway remain to be delineated, the implications are potentially profound given the developmental and disease roles for Wnt signaling and the broad interconnectedness of Wnt with other important signaling pathways. For instance, β-catenin appears to interface with additional mechanically activated pathways, including a critical role in TGF-β-induced myofibroblast differentiation of aortic valve interstitial cells [6], and a collaborative role with MRTF in promoting TGF-β-dependent epithelial-mesenchymal transition [48].

Matrix stiffness as a therapeutic target for fibrotic diseases

Through the mechanoresponsive pathways of cellular activation described above, increased matrix stiffness promotes fibroblast ASMA expression and differentiation into myofibroblasts [1,2,3▪,4▪,5▪▪,6,7]. Pathologically increased matrix stiffness is a hallmark of essentially all fibrotic diseases, and myofibroblast differentiation and activation are central to their pathogenesis [59]. Therapeutic interventions designed to reduce matrix stiffness could therefore potentially benefit the entire class of fibrotic diseases, which when taken together account for enormous burdens of morbidity and mortality [60]. Tissue stiffness in fibrotic diseases is regulated by the cross-linking of matrix proteins, in addition to the amounts of these proteins that are deposited. In health, this protein cross-linking stabilizes and orients extracellular matrix assembly for correct function, and is facilitated by a series of enzymes including transglutaminases, lysyl oxidases, and prolyl hydroxylases [61]. In fibrotic disease, these enzymes represent potential therapeutic targets for reducing matrix stiffness. Lysyl oxidase–like-2 (LOXL2) is a cross-linking enzyme that demonstrates increased expression in human fibrotic lung and liver tissues compared with its limited expression in healthy tissues, and that has recently been targeted in animal fibrosis models. In commonly used mouse models of liver and lung fibrosis, an inhibitory monoclonal antibody to LOXL2 decreased the accumulation of tissue collagen and of ASMA-positive myofibroblasts, and decreased TGF-β pathway signaling [62]. Transglutaminase 2 (TG2), the most widely expressed member of the transglutaminase family of cross-linking enzymes, has also been shown to contribute to fibrosis in animal models. TG2-deficient mice are protected in commonly used mouse models of kidney [63] and lung fibrosis [64▪]. In the kidney model, both myofibroblast accumulation and TGF-β activation were reduced in the absence of TG2 expression [63]. In humans, TG2 expression and activity are increased in lung biopsy sections from pulmonary fibrosis patients compared with normal controls [64▪]. There now is substantial interest in translating the promising preclinical results of targeting matrix cross-linking enzymes in animal models to human fibrotic diseases. A phase I study of an inhibitory monoclonal antibody to LOXL2 in patients with pulmonary fibrosis is currently recruiting patients (ClinicalTrials.gov identifier: NCT01362231).

Tissue tension as a therapeutic target for hypertrophic scarring

Skin exhibits intrinsic tension, and mechanotransduction of this tension is central to normal healing of skin wounds [65]. Much like in fibrotic diseases, aberrant wound healing can result in hypertrophic scar formation that is associated with contractile fibroblasts [66] and mechanical signaling through FAK [39▪▪,67]. Recently developed experimental models have been used to show that exogenous application of mechanical tension to healing wounds is sufficient to produce hypertrophic scars in mice [40,68▪]. Based on this view of mechanical tension as central to exuberant scarring, stress-shielding silicone polymer sheets have been developed that can be prestrained and applied across cutaneous wounds [69▪▪]. The stress shielding devices effectively carry tensile loads normally borne by the skin, thereby modifying the mechanical environment by reducing tension across the wound bed. Preclinical studies in a swine model of cutaneous wound healing demonstrated that the stress-shielding approach resulted in reduced expression of profibrotic markers and reduced scarring [69▪▪]. In a phase I clinical trial (ClinicalTrials.gov identifier: NCT00766727), stress shielding after abdominal incisions demonstrated improved scar appearance relative to within patient controls.

Cell-matrix interactions as targets for discovery and screening

In addition to directly modifying biomechanical cues through devices or targeted alterations of matrix cross-linking, it is becoming possible to study cellular responses to defined biomechanical environments in new ways, which should provide both new insights and new targets for intervention. For instance, recent proteomics-based efforts to analyze the composition of focal adhesions have revealed myriad changes in their composition that depend on myosin contractility, indicating the profound effect mechanical loads have on signaling networks emanating from integrin based cell-matrix adhesions [70▪▪]. RNA interference-based screening has been used to elucidate gene targets that control matrix stiffness-dependent fibroblast polarization and traction force generation, important functions relevant to directed cell motility and tissue remodeling. Finally, the ability to control matrix mechanical properties in high-throughput platforms is leading to new insights into stem cell function [71], and opening the possibility for molecular screening within defined mechanical environments that mimic normal and pathologic tissue stiffness [72]. Such approaches will ultimately inform our understanding of how cells process cues from their mechanical environment, and allow us to test interventions aimed at altering this reciprocal interaction within relevant mechanical environments, with the ultimate goal of targeting the mechanobiological interface to improve clinical outcomes.