Forces in stem cells and cancer stem cells

Abstract

Endogenous and exogenous forces are critical in physiology and pathology of the human body. Increasing evidence suggests that these forces, mechanics, and force-associated signaling are essential in regulating functions of living cells. Here we review advances in understanding the impact of forces and mechanics on functions and fate of embryonic stem cells, adult stem cells, and cancer stem cells and the pathways of mechanotransduction in cells. Stem-cells based models are useful in understanding how forces influence physiology, pathology, and embryonic development, which is incompletely understood, especially for mammals. We highlight increasing efforts and emerging favorable clinical outcomes in mechanomedicine, application of mechanobiology to medicine. Major progresses in mechanobiology, the pillar of mechanomedicine and mechanohealth (application of mechanobiology to health), are pivotal in understanding the life of force and making substantial advances in medicine and health.

Introduction

It has long been proposed that forces are critical in physiology and pathology of the human body [1, 2]. However, nowadays researchers are still trying to figure out how important forces are in shaping the human body at various stages of development and what are the underlying mechanisms of the force impact, i.e., how forces change the genes and direct patterns and morphogenesis. It is not surprising that the musculoskeletal system, the cardiovascular system, the pulmonary system, and the auditory system are all greatly influenced by force or pressure, but it is increasingly evident that other organs and tissues in the human body might also be substantially influenced by force and mechanical factors. Since tissues and organs are made of living cells, extracellular matrices, soluble molecules, and extracellular vesicles, it is reasonable to posit that if forces and mechanical signals are important in stem cells, cancer stem cells, and embryonic development, they should influence functions of all cell types in tissues and organs of living organisms. As such, in this review, we will highlight the advances in understanding the effects of forces in stem cell fate and cancer stem cells and discuss daunting challenges in quantifying forces and moduli in vivo. We will then summarize published favorable clinical outcomes of application of mechanobiology to medicine, i.e., mechanomedicine. Mechanomedicine and mechanohealth (application of mechanobiology to health) represent the ultimate goals of disseminating mechanobiology discoveries and technology for human medicine and health.

I. Force impact on embryonic development in vivo

An embryo starts from a fertilized egg that divides and grows into a multi-cellular organism. These early cells are connected via cell-cell adhesion protein E-cadherin. E-cadherin^−/− mouse embryos only form loose cell aggregates and fail to form blastocysts with a proper blastocoel [3]. Extracellular matrix protein fibronectin is expressed at the blastocyst stage [4]. Mouse embryos with fibronectin null mutations exhibit mesodermal defects and fail to develop notochord or somites [5]. Because cell-cell adhesion protein cadherins mediate force transmission between cells [6, 7] and cell-matrix proteins like integrins mediate force transmission between the cell and the matrix proteins [reviewed in ref. 8), it is believed that the forces generated inside the cells via actomyosin contractility play important roles in the development of the embryo. Ablation of nonmuscle myosin IIA or myosin IIB leads to mouse embryonic lethality [9–11]. A follow-up study on the isoforms of myosin IIs in mouse reveals the unique essential function for myosin IIA in placenta development and isoform-independent requirement for myosin II in visceral endoderm function [12]. A recent study reveals that maternal zygotic nonmuscle myosin heavy chain IIA mutation results in failed cytokinesis, increased duration of the cell cycle, weaker embryo compaction, and reduced differentiation in mouse embryos [13]. However, these findings only demonstrate the essential role of the presence of nonmuscle myosin II protein per se and not the force-generating capacity of myosin II driving these processes.

In fact, much of the understanding of force impact on early embryonic development comes from studies on lower animal models such as Drosophila, Xenopus, zebrafish, and Caenorhabditis elegans [14–22]. For example, during gastrulation, an early stage in development of the Xenopus embryo, cells undergo large deformation (possibly caused by patterned forces) resulting in morphogenetic movements of invagination, ingression, involution, epiboly, intercalation, and convergent extension [15, 23]. In the absence of active movements, formation of germ layers is incomplete and ensuing development is halted [15, 24]. When ventral ectodermal explants taken from early gastrula embryos of Xenopus laevis are stretched, the cells first move towards the stretching force and later reorient perpendicular to the stretching force, becoming intercalated between each other just like convergent extension in the embryo [25]. In Xenopus laevis the neural crest generates dynamic stiffness gradients [26]. Compressive forces in Drosophila are known to rescue the Twist protein expression and midgut formation in a mutant defective embryo in convergent-extension movement [27]. Mesoderm and endoderm invaginations are dependent on myosin II in Drosophila embryo gastrulation [28]. In the Drosophila embryo, while it is shown that planar cell intercalation and axis elongation depend on polarized localization of myosin [29], whether these processes depend on the force generated by myosin or the presence and accumulation of myosin protein remains unclear because the magnitude of the forces are not quantified and modulated. In one-cell stage Caenorhabditis elegans embryos, PAR proteins are shown to mediate asymmetric spindle positioning, possibly due to the imbalance of net pulling forces acting on the spindle poles [30]. However, the actual pulling force is not measured in this study. Other studies of Drosophila embryos suggest that forces play a role in gastrulation [31, 32] or planar cell polarity [33, 34] or growing wing discs [35], and oscillations of myosin II within cardioblasts are necessary to ensure appropriate cell-cell connection formation [36], but in all of these studies only myosin localization or accumulation, cell shape change, or cell movement speeds are quantified as a surrogate of force, not the force itself. The limitation is that alternative interpretations or mechanisms might be proposed to explain these findings and without quantifying the force or the stress itself it is difficult to know whether the force is driving these processes or the force is just playing some modulating or facilitating roles to accompany the dramatic changes of expressions and locations of certain genes and proteins and their interactions. Furthermore, it is conspicuous that actin network is critical in actomyosin contractility and it is found that Formin Frl is a key regulator of apical actin network and modulates epithelial cell deformability in Drosophila embryos [37]. A unique mechanotransduction pathway is discovered at tricellular junctions in the Drosophila embryo involving Abl tyrosine kinase and actin-binding Canoe/Afadin that stabilizes cell adhesion under tension [38]. A recent study reveals that self-organized Toll-8/GPCR (G protein coupled receptor) asymmetry is essential in forming polarized myosin contractile interface of Drosophila embryos [39]. It is reported that fluid pressure (~300 Pa) fracturing of cell-cell contacts followed by contractility-directed microlumen amalgamation appears to control the first axis of symmetry of the mouse embryo [40].

However, in most of the above studies, the actual force is not measured since quantifying forces in vivo is extremely challenging, although some techniques have been developed to estimate the forces or stresses in situ, but most techniques are indirect and each method has its own limitations (reviewed in ref. 41). Only in a few studies the force itself is measured ex vivo or in vivo. For example, force is quantified using a gel-based force sensor in an isolated Xenopus laevis embryonic tissue and the mean compressive stress in convergent extension is found to be ~ 5 Pa [42]. Using an oil-droplet force sensor that can be microinjected into living tissues, anisotropic forces are quantified: in living dissected mouse mandibles at stage E11, the maximal anisotropic stress generated by mesenchymal cells is ~1.6 nN/μm² [43]. A different approach of using an elastic round microgel has been used to quantify forces in developing zebrafish embryos in vivo [44]. Substantial spatial and temporal traction variations are revealed inside the embryo from 3 to 10 hours post fertilization, suggesting that these tractions and their spatial variations might be important in the patterning and morphogenesis of the zebrafish embryos. Spatiotemporal viscoelastic properties of a zebrafish embryo have also been quantified by injecting oil droplets loaded with magnetic nanoparticles and the posterior elongating region appears to be less stiff and more fluid [45]. Supracellular stresses decrease from anterior to posterior regions and guide solid to fluid like transition in zebrafish embryos [46]. Furthermore, using atomic force microscopy, a stiffness gradient of ~1.0 Pa per μm, generated by the rising stiffness of tissue rostral to the optical tract as a result of cell proliferation, is found to be required for axons to change their turning direction caudally towards soft tissues in the developing Xenopus embryonic brain [47] (Fig. 1). The non-equilibrium phase transition from fluid phase to solid phase is proposed to interpret the changes in cell shape in first monolayer epithelial cells [48] and then in the Drosophila embryo [49]. Furthermore, the unjamming transition is distinct and different from the EMT (epithelial-to-mesenchymal) transition in primary epithelial cells [50]. All together these studies suggest that forces play some roles in embryonic development in vivo in various lower animals but there are some major differences in structures and functions among species [51], the details of how essential the forces are in driving these processes in embryonic development of mammals (especially humans) remain to be elucidated.

Figure 1. Force impact on embryogenesis and organogenesis.

Compelling evidence from published reports demonstrates that fate and differentiation of embryonic stem cells and adult stem cells depend on forces (shear and/or normal stress), substrate elasticity/viscoelasticity, and substrate topography. The observation from cell culture studies that mesenchymal stem cells undergo neurogenesis on soft substrates is consistent with the finding that a stiffness gradient is responsible for axons to change their turning direction caudally towards soft tissues in the developing Xenopus embryonic brain in vivo. Stem-cell based models are useful in understanding role of forces in the development of embryos, which are inaccessible in vivo for mammalian embryos. Blastoid formation via iPSCs is enhanced by 3D culture and substrate mechanics and may depend on endogenous forces (endo-force). Maturation of cardiomyocytes from iPSC differentiation is promoted by mechanical stretching. iPSC, induced pluripotent stem cells. Force is used here generically to represent any type of mechanical loading (force, torque, tensile or compressive stress, shear stress, torque per volume or specific torque) (exogenously or endogenously).

II. Effects of forces in embryonic stem cell fate

Since it is challenging to quantify forces in embryos in vivo, alternative approaches of studying these processes in culture have been developed. Cells are removed from an early-stage developing embryo to study how mechanical forces influence cell fate and differentiation. A fertilized oocyte is totipotent, capable of developing into a whole organism in vivo. Embryonic stem cells are cultured cell lines that are derived from inner cell mass cells (Fig. 1). The inner cell mass lies in the blastocyst, an early-stage (5–6 days post fertilization) pre-implantation embryo. Embryonic stem cells are pluripotent stem cells, being able to differentiate into all three germ layers (ecto, meso, and endo). Mouse embryonic stem cells are established in culture in early 1980s [52, 53] but stable human embryonic stem cells are cultured and established only in the late 90s [54]. Many early studies on the embryonic stem cells focus on the effects of soluble factors such as growth factors, cytokines, and transcription factors [55, 56] and the role of forces in embryonic stem cell fate is largely unknown in the early 2000s. When mouse embryonic stem cells are subjected to small exogenous forces, they are much more sensitive to force magnitudes than differentiated tissue cells [57, 58]. The reason is that the embryonic stem cells have low intrinsic cell stiffness (~0.5 kPa) and for a given magnitude of stress, the resultant strain (strain=stress divided by stiffness) is much greater. When mouse embryonic stem cells are cultured on soft substrates of a stiffness that mimics the stem cells’ intrinsic stiffness, mouse embryonic stem cells remained pluripotent without undergoing spontaneous differentiation, even in the absence of self-renewal promoting factor LIF (leukemia inhibitory factor) [59]. However, when an external force is applied to a single embryonic stem cell in culture, the cell starts to spread within a few minutes and elevate their endogenous forces to exert elevated tractions onto the underlying substrate and eventually differentiate [58] (Fig. 1). However, what causes this “external force” in the context of early embryo development is elusive, but there are several possibilities that are not mutually exclusive: elevated matrix stiffness as a result of fibronectin polymerization, shear stress generation during cell rearrangement, hydraulic pressure fracturing cell-cell contacts due to Ostwald ripening, or cell proliferation-induced tissue stiffening. One unique feature of the mouse embryonic stem cells is that their stiffness does not increase with substrate stiffness [60], different from human mesenchymal stem cells and other differentiated tissue cells whose stiffness increases with substrate elasticity [61], possibly because the embryonic stem cells have much fewer actin filaments than differentiated cells [58, 62]. In a recent report, using engineered patterned substrates to mimic early embryos, it is shown that cell-generated tension facilitates formation of gastrulation-like nodes and an exogenous stretching force promotes mesoderm specification in human embryonic stem cells [63]. Using a novel hydrogel that enables robust matrix tethering, it is shown that substrate mechanics impacts mechanosensitive signaling pathways and regulates self-renewal and differentiation of mouse embryonic stem cells and human pluripotent stem cells [64]. A recent report shows that a decrease in membrane-to-cortex attachment (cortex softening) is necessary but not sufficient for mouse embryonic stem cells to exit pluripotency [65]. All these findings suggest that forces (endogenous and exogenous) are critical in shaping embryonic stem cell fate and functions.

III. Mechanoregulation in organoids and embryo models

Mammalian embryos are inaccessible to in vivo visualization and mechanical intervention and thus it is difficult to determine if forces and mechanics play any significant roles in embryonic development of mammals. When a strategy of culturing a single mouse embryonic stem cell in a 3D soft matrix and then transferring the cell colony to a 2D matrix that mimics the stiffness of early embryos’ microenvironment is employed, the single cell is able to develop into a highly ordered and proper 3-germ layer arrangement of ecto-, meso-, and endoderm [66], suggesting that forces and mechanics are important in directing germ layer organization and order. A different method of using a micropatterned coverslip for geometric confinement is shown to be sufficient to trigger self-organized patterning in human embryonic stem cells [67]; the patterns appear to be under the control of edge sensing of the colony, suggesting mechanical factors are at play but the exact mechanism remains unclear. Using human pluripotent stem cells in mechanically-designed microfluidics cell culture environments, it is found that substrate stiffness and 3D matrix play an important role in embryonic patterning that mimic early development in humans [68, 69]. Recent studies using a blastoid model to model human embryos by growing human embryonic stem cells or reprogrammed human fibroblasts in a 3D culture system demonstrate that the blastoid can model the overall architecture of the human blastocyst [70, 71] (Fig. 1), but how important cellular endogenous forces are in driving the pattern formation is elusive nor it is clear what is the underlying mechanism. Moreover, while it might be possible to study the effects of forces in human stem cell-based embryo models in vitro, substantial technical and ethical challenges remain in studying human embryos or embryo models (e.g., blastoid or gastoid), even with the updated 14-day rule [72]. A recent study using artificial intelligence designed shape of a stem cell colony of Xenopus laevis reveals that the cell assembly can replicate kinematically by moving and compressing dissociated cells in their environment into functional self-copies [73], but the exact role of cellular forces and/or matrix proteins in the replication process remains unclear. The organoids and stem cell-based embryo models of mammals represent useful models to study how force and mechanics regulate the embryonic development, but it is difficult to know how truthfully these models can mimic embryonic development in vivo. In addition, because of the incomplete understanding of the physiological microenvironment around the mammalian embryos in vivo (especially human embryos) it would be rather challenging to use stem cell-based embryo models to completely recapitulate rodent and human embryos and study the interactions among the cell force and mechanics, the genes and proteins, and the microenvironment of the embryos. These fundamental issues remain to be resolved in the future.

IV. Effects of Forces in adult stem cell fate

All adult humans have various types of multipotent and unipotent adult stem cells in their bodies. Human naïve mesenchymal stem cells, a type of adult stem cells, are sensitive to the elastic stiffness of the substrate that they are attached to and differentiate into neural cell-like cells when the substrate stiffness is ~ 1 kPa, skeletal muscle cell-like cells when the substrate is ~10 kPa, and osteogenic lineage cells when the substrate is ~30 kPa [61]. The substrate elasticity directed stem cell lineage fate and specification is blocked when nonmuscle myosin II is inhibited [61], suggesting that endogenous forces are required for the cell response to substrate matrix elasticity (Fig. 1). However, a number of studies show that leptin-receptor-expressing mesenchymal stem cells contribute to fat, cartilage and bone lineages in vivo [74, 75] but not necessarily neurogenesis and myogenesis. The endogenous contractile forces must be balanced inside the cell all the time and hence they are called cytoskeletal tension or cytoskeletal pre-existing tensile stress or prestress [8]. Other early studies shown that mechanical signals regulate adult stem cell fate [76–79]. Exogenous mechanical strains regulate differentiation and proliferation of human mesenchymal stem cells and the orientation of cells with respect to the strain axis determines what genes are upregulated and downregulated [80] and equiaxial and uniaxial strains exert differential effects on gene expressions in these stem cells [81]. Muscle stem cells show sensitivity to underlying substrate stiffness [82]. In regular plastic dishes muscle stem cells have been shown to lose stemness. But when these muscle stem cells are cultured on 10–12 kPa substrates, resembling in vivo muscle stiffness, cells retain their self-renewal and rejuvenation capacity [82, 83]. Mesenchymal stem cell fate and activity are also regulated by viscoelasticity of the substrate and enhancement of mesenchymal stem cell differentiation by fast matrix stress relaxation appears to be mediated through local clustering of RGD ligands and actomyosin contractility, possibly because forces in the viscoelastic matrix (but not in the pure elastic matrix) can be relaxed over time due to mechanical yielding and remodeling of the matrix [84]. Adult neural stem cells have been reported to undergo neurogenesis depending via regulation of substrate stiffness mediated by YAP-beta catenin interactions [85]. In contrast to mechanical tension that often leads to stem cell differentiation, mechanical compression is shown to enhance intestinal stem cell self-renewal in organoids [86]. These findings demonstrate that functions and fate of adult stem cells are greatly influenced by force and mechanics (Fig. 1) such as moduli (storage modulus and loss modulus) and rheological properties of the cells [87] and viscoelastic properties of the extracellular matrix.

V. Force and mechanics regulate iPSC-derived cells

Because of the ethical issues associated with the use of human embryonic stem cells and the need to generate patient-specific autologous cells and tissues, induced pluripotent stem cells derived from differentiated somatic cells are highly desirable. Reprogramming somatic cells to pluripotent stem cells is achieved from mouse fibroblasts [88] and then from human fibroblasts [89, 90] (Fig. 1), paving the way towards personalized tissue engineering and regenerative medicine. Various mechanical approaches such as microgrooved surfaces have been employed to promote mesenchymal-epithelial transition (MET) in adult fibroblasts and thus reprogramming efficiency [91] and a microfluidics device induces somatic cells into human iPSCs [92]. Since engineering heart tissues is an important step toward repairing or replacing damaged hearts, protocols have been established on differentiation of cardiomyocytes from human iPSCs (induced pluripotent stem cells) and 3D assembly of the cardiomyocytes into engineered heart tissues [93] (Fig. 1). It is reported that mechanical stimulation promotes maturation and force generation of human-iPSC-derived cardiac tissues [94]. In addition, using human embryonic stem cells or human iPSCs-derived cardiomyocytes to assemble engineered heart muscle, it is shown that mechanical stretch induces structural and functional maturation of the heart tissue [95]. A similar finding shows that mechanical stretch enhances growth and maturation of 3D human-iPSC-derived myocardium [96] (Fig. 1). Human iPSC-derived cardiac cell sheet-tissues can generate a mean force per area of 3.3 mN/mm² [97], suggesting that these tissues are functional. Electrical stimulation to early-stage iPSC-derived cardiomyocytes promotes human cardiac tissue maturation [98]. A report finds that substrate stiffness modulates differentiation of iPSC-derived human neural crest stem cells in vivo and stiff substrates induce smooth muscle cell genes whereas soft substrates induce glial genes [99]. Substrate stiffness has also been shown to contribute to maturation of human-iPSC-derived cardiomyocytes [100]. Interestingly, mechanical nonuniformity of hybrid matrices composed of fibers with different diameters induces maladaptive hyper-contractile phenotypes of cardiac microtissues, assembled from human-iPSC-derived cadiomyocytes [101], suggesting that the hybrid matrices-model may be used to create pathological cardiac tissues. However, acute perturbation to actomyosin stress or matrix stiffness causes rapid changes in Lamin A and DNA damage of human-iPSC-derived cadiomyocytes and Lamin A is stress-stabilized to protect the DNA [102]. Together these studies demonstrate that force and substrate stiffness have profound impacts on human- iPSC-derived cells and tissues. However, since iPSCs are pluripotent, just like embryonic stem cells, they express key pluripotency markers such as Oct4, Nanog and self-renewal marker Sox2, etc. Even after iPSCs are differentiated into cadiomyocytes or other cell types, there is still a risk that iPSC-derived cells could form tumors in vivo. For example, a recent paper shows that iPSCs dedifferentiated from adult cardiomyocytes can drive heart regeneration by stimulating exiting cardiomyocytes and a strategy of transient expression of pluripotency markers to minimize the potential of tumor formation is used [103]. However, one still cannot rule out the possibility of tumorigenesis from these grafted cells that might contain a tiny number of undifferentiated iPSCs. Recently a genome-editing approach is reported that can deplete undifferentiated human iPSCs and kill all iPSCs-derived cell-types if necessary to mitigate the risks [104], but the question of the safety issue of iPSCs in any autologous tissue-engineered tissues persist since a few undifferentiated iPSCs might still survive the treatment and killing all iPSCs-derived cell-types would defeat the purpose of using iPSCs in the first place.

VI. Mechanoregulation of cancer stem cells

Cancer stem cells are a small subpopulation of blood cancer cells or tumor cells within tumors that are stem-like cells and exhibit enhanced cancer-initiating or tumor-initiating capability. Since the first isolation of leukemia cancer stem cells [105, 106], cancer stem cells have been found in many types of solid tumors: breast [107], brain [108], skin [109], prostate [110], ovary [111], and lung [112]. However, the exact nature of cancer stem cells is still not well understood. While there is evidence that activation of EMT (epithelial-mesenchymal transition) gives rise to cancer stem cells [113], non-cancer stem cells can de-differentiate into cancer stem cells [114], suggesting interconversion and plasticity in these cells. Alternatively, some undifferentiated tumor stem cells or partially differentiated cells may exist in the general population of differentiated tumor cells. For example, recent findings suggest that there exists a partial EMT program [115, 116] or MET program [117] that may give rise to stem-like cancer cells that have enhanced self-renewing, invasive and metastatic capability.

What are the similarities and fundamental differences between normal stem cells and cancer stem cells? It is known that cancer stem cells from solid tumors are capable of self-renewing, just like normal embryonic stem cells and iPSCs [118], but cancer stem cells of solid tumors are often unipotent and not pluripotent. In sharp contrast, cancer stem cells exhibit a unique capability of invasion and metastasis whereas normal stem cells do not. The exact mechanism of what contributes to this difference is poorly understood. Cancer stem cells are traditionally selected based on the expression of stem cell surface markers on their plasma membrane, but the surface-marker selection approach has been rather controversial since some tumor-initiating cells that do not express typical stem cell surface markers can still efficiently generate tumors [119–121]. In addition, human colon cancer harbors a tiny tumorigenic subpopulation that is uncorrelated with stem cell markers [122]. As such, there is a need to develop alternative approaches to select cancer stem cells. A mechanical approach of seeding the cancer cells inside a 3D soft (~100-Pa) fibrin (but not collagen-1) gels has been reported [123], independent of stem cell surface markers. Melanoma cells selected by this approach are highly tumorigenic and metastatic: only a few cells are needed to metastasize to the lung [123] or generate local subcutaneous tumors [124]. Because these tumorigenic cells do not depend on surface stem cell markers for selection and isolation, they are called tumor-repopulating cells [123,125]. Stem-cell-like tumor-repopulating cells, like normal embryonic stem cells, express high levels of Sox2 and are self-renewing [123, 125]. The soft-fibrin selection approach has been successfully used to select and culture tumor-repopulating cells from a general population of tumor cells from skin, breast, lung, liver, cervical, and ovarian cancer of murine or human cancer cell lines [123, 125–127]. The tumor-repopulating cells are very soft (~100 Pa, quantified using magnetic twisting cytometry (MTC)), a few folds softer than their differentiated counterpart tumor cells [125]. Other types of highly metastatic cancer cells measured using other probes (e.g., optical stretcher or atomic force microscopy (AFM)) are also softer than those non-metastatic cancer cells [128–130]. Is the softness of tumor cells just an insignificant byproduct of other changes in the tumor cells or a potentially important physical trait? Using a microfluidic-based method, a softness feature has been reported [124], which separates soft tumor cells (<300 Pa, quantified using AFM, 2–3-fold higher than the values measured by MTC, as demonstrated from a comparison between AFM with MTC [131]) from stiff tumor cells (>800 Pa) in the suspended tumor cell population. Since this method can predict the metastatic potential of the soft tumor cells, it suggests that the cell softness is an important physical marker for malignant solid tumors, possibly adding an additional physical trait to the known four physical traits (tumor microarchitecture, tumor tissue stiffness, tumor tissue solid stress, and interstitial fluid pressure) of solid tumors that hinder successful treatment of malignant tumors [132]. Interestingly, these stem-like tumor-repopulating cells do not stiffen with increasing substrate elasticity [125], similar to what is observed in embryonic stem cells [60]. These soft tumor-repopulating cells, as a result of high Sox2 that leads to low Cdc42 and low F-actin, are more efficient than the stiff differentiated counterpart tumor cells in extravasation in the in vivo zebrafish model [133]. Depleting Sox2 or Cdc42 or elevating F-actin in these tumor-repopulating cells substantially reduces the extravasation efficiency of these cells [133]. A mechanics-based model to explain the tumor cell heterogeneity in cancer invasion has been proposed can be proposed: The matrix stiffening is reported to be a result of excessive collagen-1 [134], which might be a protective response trying to impede the tumor growth. As a result of the differential or heterogeneous tumor tissue stiffening, some tumor stem cells are stiffened and differentiated (the EMT process). Some undifferentiated tumor cells or partially differentiated tumor cells, such as the soft tumor-repopulating cells, follow those differentiated tumor cells to invade and to intravasate. The soft undifferentiated or partially differentiated tumor-repopulating cells are likely the real culprit in establishing metastatic colonization [125]. Consistent with this model is the report that demonstrates that differential tissue stiffness is essential in initiating skin tumor stem cell invasion in a developing embryo [135]. In fact, uniform stiff tumor matrix inhibits tumor cell growth and promotes dormancy of stem-cell-like soft tumor repopulating cells in mouse models for both murine and primary human melanoma [136]. Stiff fibronectin matrices surrounding the tumor cells promote breast cancer dormancy and exits from dormancy require MMP-2-mediated fibronectin degradation [137]. In addition, a recent study reports that tumor-derived type III collagen is required to sustain tumor dormancy and increased in tumors from patients with lymph node-negative head and neck squamous cell carcinoma [138]. The exact roles of excessive fibrin, fibronectin, and/or type III collagen in promoting dormancy of various types of tumors in patients remain to be elucidated in the future. Nevertheless, these findings support the stiffness matching model [139] that intrinsically soft stem-cell-like tumorigenic cells that have low endogenous forces thrive in a soft 3D matrix environment whereas stiff differentiated tumor cells survive in a stiff 3D matrix of the primary tumor site, supporting the proposition that the ECM is a (physical) barrier to restrain tumor progression [140]. A recent study reveals that long liver cancer patient survival duration is positively correlated with increasing type-1 collagen content [141]; in addition, type-1 collagen only scales with fibrotic genes in many types of cancers [141], consistent with the notion that extracellular matrix proteins such as type-1 collagen are protective barriers and not facilitators to solid tumor progression. From these published reports, we propose a mechanobiology model of cancer progression that highlights the importance of softness-based mechanoregulation in cancer stem cell growth, dormancy, and metastasis (Fig. 2) and illustrates a role of inflammation in cancer and an element of the Rudolf Virchow’s postulate that a tumor is a wound that does not heal [142]. A recent report shows that mechanical compression can induce a homogeneous population of non-small-cell lung carcinoma into heterogeneous subpopulations [143] but which subpopulation is more malignant remains to be seen. In vivo studies are needed in the future to determine what subtypes of tumor cells are the primary culprit of tumor metastasis and metastatic colonization.

Figure 2. A mechanobiology model of tumor cell self-renewal and metastasis.

Matrix metalloproteinases (MMPs) from the primary tumor site soften the extracellular matrix (ECM) of the tumor microenvironment and break tumor cell dormancy, leading to tumor cell invasion. Stiff (>800 Pa) differentiated tumor cells and soft (<300 Pa) undifferentiated tumor stem cells such as tumor-repopulating cells [124] enter blood vessels (intravasation), arrest at narrow vessels, and exit blood vessels (extravasation) to metastasize to distance sites and form micrometastasis. In some cases, soft tumor stem cells proliferate and self-renew within a soft matrix (e.g., bone marrow, brain, lung, and liver) to establish metastatic colonization and grow into macroscopic metastases, or survive and enter dormancy within the stiff matrix, whereas stiff differentiated tumor cells die in the soft or stiff matrix of a different tissue (denoted by an X). In some other cases, when the matrix of tumor microenvironment of metastatic sites becomes inflamed and then softened, soft dormant tumor stem cells will exit dormancy, self-renew, and grow into clinically-detectable macroscopic metastases. Note that this simple model illustrates an element of the Virchow’s postulate and highlights softness-based mechanoregulation of cancer progression, which is also regulated by other physical and soluble factors and cells such as tumor-associated fibroblasts and immune cells.

VII. Mechanical interactions between cancer cells and immune cells

The immune system is the key host defense system against infection from viruses, bacteria, fungi, parasites, etc. There are two subtypes of the immune system: innate and adaptive systems. The rapid nonspecific innate system is made of innate immune cells (granulocytes (neutrophils, eosinophils and basophils), macrophages, dendritic cells, NK (natural killer) cells, and mast cells) and complement components. The slow adaptive immune system is specific and is made of T cells and B cells. Neutrophils that migrate on soft substrates generate tractions [144] and exert tractions on vascular endothelial cells during the process of diapedesis to open a junctional gap and push themselves across the gap [145]. Neutrophils also generate tension on VE-cadherin during trans-endothelial migration [146]. Monocytes are able to cross the endothelium to enter the interstitial space where they differentiate into macrophages. Macrophages phagocytize varieties of alien molecules and apoptotic or senescent cells to maintain the human body’s homeostasis. The phagocytosis process is tightly regulated by the target cell’s modulus and the contractile forces [147–149]. There is evidence that monocytes and monocyte-derived macrophages enhance their immune functions in response to force and cyclic pressure [150]. Since phagocytosis requires substantial membrane distortion and deformation of the macrophages, these cells are relatively soft. Activated macrophages can be differentiated toward M1 or M2 phenotype and a published report shows that elongation of macrophages, without exogenous cytokines, leads to the expression of M2 phenotype markers and reduces secretion of inflammatory molecules [151]. In addition, soft hydrogels reduces macrophage inflammatory response and YAP is a key molecule for controlling inflammation and substrate stiffness sensing by macrophages [152]. Directed cell migration and activation of dendritic cells derived from bone marrows can be controlled by fluid shear stress in a microfluidic channel model [153]. It appears that adaptive immune cells’ priming and activation are also dependent on endogenous forces. Antibody production is critically dependent on B cell internalization of antigens and its presentation of antigens to helper T cells. But antigen-presenting cells (e.g., macrophages) must first form immune synapses with B cells and present antigens to B cells. B cells, in turn, extract antigens using actomyosin contractile forces (~10-pN per antigen molecule) that regulate affinity discrimination [154] and this process is mediated by Arp2/3-branched-actin foci and formin activity [155]. As expected, stiffness of the substrate that presents antigens to a B cell regulates B cell activation [156].

It is known that contractile forces regulate T cell activation [157–159]. Applied forces on T cell receptor and CD8 trigger calcium entry and stiffened antigen presenting cells enhance the calcium response of the T cells [160]. Cytotoxic T cells exert their cytoskeletal contractile forces onto tumor cells to kill tumor cells [158]. It is reported that within the immune synapse cytotoxic T cells release perforin at the base of actin protrusions that are required for synaptic force exertion and efficient killing of target cells [161]. However, different types of tumor cells respond differently to T cell killing. For example, cytotoxic T cells do not kill soft undifferentiated tumor-repopulating cells as effectively as they kill differentiated tumor cells. The underlying mechanism is that tumor-repopulating cells are very soft and thus they impair perforin pore formation on their plasma membrane (Fig. 3). This result also explains why anti-PD-1 antibody is not effective in killing some malignant tumor cells. Elevating the cell stiffness of tumor-repopulating cells restores T cell-mediated cytolysis of tumor-repopulating cells and together with treatment with anti-PD-1 antibody the cytotoxic T cells kill tumor-repopulating cells much more effectively in mice [162]. This finding is supported by an independent study that cholesterol depletion mediated cancer cell stiffening enhances cytotoxicity by T-cells as their forces arise at the synapse [163]. The impact of cell softness in preventing T cell killing also applies to macrophages as it has been reported that macrophages engulf stiff target cells more avidly than soft target cells [164]. It is also known that tumors tend to skew to the M2 phenotype of macrophages than the M1 phenotype [165]. In addition to the immune cells at the tumor microenvironment, cancer-associated fibroblasts also regulate tumor cell activities. Cancer-associated fibroblasts promote tumor invasion by exerting forces on cancer cells via N-cadherin at the CAF membrane and E-cadherin at the cancer cell membrane [166] and four subpopulations of cancer-associated fibroblast are identified in metastatic lymph nodes [167]. CD8⁺ cytotoxic T cells interact with tumor cells at the immune synapse site and release perforin and granzymes to mediate the killing. The space within the immune synapse is very small and relatively sealed off and so the released perforin should also attack T cells but how T cells evade this “self” killing is not understood, possibly because they are also very soft (Fig. 3). Taken together these findings suggest that forces are critical in mediating interactions between immune cells and tumorigenic cells to elicit effective immune responses against tumor growth.

Figure 3. Cell softness regulates cytotoxic T cell killing of tumor cells.

Cytotoxic T cells enter tumor parenchyma, where the T cells use T cell receptor (TCR) to recognize MHC (major histocompatibility complex)-tumor antigenic peptide complex and form the synapse. The activated T cells then release perforin and granzymes to the synapse space where perforin forms pores on the plasma membrane of target tumor cells and allows the entry of granzymes into the cytoplasm, activating caspases 3 and 7 and leading to tumor cell apoptosis. However, drilling pores by perforin is not only a chemical but also a mechanical process. Cell stiffness (>600 Pa) is required for the pore formation by perforin and cell softness impairs perforin pore formation. Thus, soft (<300 Pa) tumor stem cells such as tumor-repopulating cells use their softness to evade T cell cytolysis by impeding perforin pore formation. On the other hand, activated T cells might be very soft, avoiding autolysis. In addition, it is possible that TCR-MHC binding may be equal in the molecular number but the interacting force may be weaker between the soft tumor cell and the immune cell than between the stiff tumor cell and the immune cell; as a result, the released granzyme and perforin from the immune cell may be less, contributing to less killing of the soft tumor cell. Stiff target cells are engulfed avidly by macrophages, suggesting that this model may be applied to other immune cells.

VIII. Mechanisms of Nuclear Mechanotransduction

It is increasingly evident that not only cell-matrix molecules integrins and focal adhesions [168], cell-cell adhesion molecules cadherins [6], and PECAM-1 (platelet endothelial cell adhesion molecule-1) [169] are mechanosensors that respond to forces and substrate mechanics but also is the nucleus [170]. Mechanosensing is critical in embryonic development, adult physiology, and diseases such as cancer, fibrosis, and cardiovascular diseases. At the cell surface where the cell experiences external forces, force-induced protein (such as integrin, talin, and vinculin) stretching and unfolding have been reported at the focal adhesions [171–174]. Recent work shows that restoration of substrate rigidity sensing to cancer cells by restoring tropomyosin (which associates with actin and stress fibers) levels inhibits tumor formation [175]. On the other hand, inside the nucleus, chromatin is a self-organized, complex architecture of DNA, histone proteins, and other molecules in the form of fibers, loops, domains, and compartments [176] in the cell nucleus, which in general is the largest organelle inside a eukaryotic cell. A primary function of the chromatin is to package long molecules of DNA into compact and dense structures for dynamic regulation of gene transcription, DNA replication, and DNA repair. The LINC (linker of nucleoskeleton and cytoskeleton) complex consists of KASH-domain proteins across the outer nuclear membrane that link the cytoskeleton in the cytoplasm with SUN proteins across the inner nuclear membrane that connect to the nuclear lamina (Lamin A/C and Lamin B) [170], providing a physical pathway for force transmission from the cell surface to the chromatin. However, we know little about how force regulates gene expression in the chromatin.

Experimental evidence shows that there exist multiple pathways of mechanical regulation of transcription [177]. Mechanically-activated cation channels Piezo1/2 at the plasma membrane [178] and mechanosensitive calcium channels play important roles in mechanical-chemical signaling in the cytoplasm. However, for Piezo1/2, applying stresses via a patch-clamping electrode or a glass pipette is generally the approach to deform the cell membrane. Hence the magnitude of the applied stress is not controlled precisely or measured accurately for Piezo1/2 activation, but it appears to involve high stresses and large plasma membrane deformation. For calcium channels, the stress has to be >100 Pa in order to open the calcium channels in endothelial cells [179] and ultra-rapid (<4 ms) activation of calcium influx through transient receptor potential vanilloid 4 (TRPV4) ion channels has been reported [180]. Stretching cells triggers Piezo1-mediated calcium release from the endoplasmic reticulum 30 sec post stretching and the elevated cytoplasmic calcium is responsible for the decrease in H3K9me3 levels in the nucleus and the increase in perinuclear F-actin rings [181]. Other molecules that transduce cell surface deformation into nuclear activities are the extracellular matrix (ECM) rigidity responsive elements YAP (Yes-associated protein) and TAZ (transcriptional coactivator with PDZ-binding motif) that translocate to the nucleus to relay the signals from the cytoplasm to regulate cellular differentiation and proliferation [182]. However, it generally takes many minutes to hours for YAP to translocate into the nucleus in response to mechanical forces. For example, YAP translocation from the cytoplasm into the nucleus occurs only after four-hour cyclic stretching of the flexible substrate [183], although shorter duration of YAP translocation after loading has been observed in response to compression forces [184], possibly because of the large magnitude of the compressive stress. Other cytoplasmic molecules such as Twist1 are found to translocate into the nucleus independent of YAP/TAZ [185]. Published work has shown that Lamin A/C is a mechanosensor that responds to tissue stiffness, and regulates differentiation [186]. Lamin A/C also regulates translocation and signaling of a mechanosensitive transcription factor MKL1 (megakaryoblastic leukaemia 1) [187]. Furthermore, decondensation of chromatin depends on cell shape, cell spreading, and cytoskeletal contractile forces [188]. Activation of mechanosensitive genes is impaired when nuclear envelope proteins nesprin 1 and nesprin 2 are depleted [189]. Lamin A/C depletion causes defective nuclear mechanics and mechanotransduction [190] and Lamin mutation leads to nuclear envelope rupture and DNA damage [191]. Actomyosin tension exerted on the nucleus through nesprin-1 is found to be essential in cyclic-strain induced endothelial reorientation [192]. Even isolated nuclei can respond to mechanical forces with nuclear stiffening and protein phosphorylation [193]. Together these findings demonstrate that those proteins (e.g., nesprins) that link the cytoskeleton with the nuclear envelope and those proteins such as lamins that link SUN1/2 at the nuclear membrane with the chromatin play critical roles in regulating nuclear mechanics and mechanotransduction.

The LINC-anchored actin cap connects the extracellular matrix to the nucleus for the possibility of ultrafast mechanotransduction [194]. Increasing experimental evidence demonstrates that a local stress via integrins propagates along tensed actin bundles to the LINC complex to stretch chromatin to upregulate genes in living cells. The chromatin is directly stretched milliseconds after application of a local cell surface stress (via integrins) of physiologic magnitudes and frequencies (so that there is no chromatin tearing or breaking) and the extent of transgene DHFR (dihydrofolate reductase) upregulation depends on the degree of chromatin stretching, which is dependent on the angle of the applied stress (the magnetic bead rolling direction) relative to the long axis of the cell (i.e., alignment of the majority of the stress fibers) [195]. The dependence of direct chromatin stretching and rapid gene upregulation on the stress angle for a given stress magnitude (i.e., for a fixed mechanical energy input) is significant because it suggests that the diffusion-based cytoplasmic biochemical processes may not be necessary for stress-induced rapid gene activation. Chromatin stretching is a form of ‘deformation’, which is a result of tensile stress-induced positive changes in displacements of chromatin. Some researchers call compacted chromatin as ‘condensed’ that can be ‘folded’, which could decondense or unfold via various factors including soluble factors. Here the term “stretching” by external forces means to locally “decondense” or “unfold” the chromatin domain. It is known that when the stress or the deformation is very large, it can cause nuclear envelope rupture and DNA damage, such as during cell migration through highly constricted spaces [196, 197]. In response to the magnetic bead stress of physiologic magnitudes, nuclear lamin breaking or chromatin tearing has not been observed in various types of cells [195, 198]. Disrupting F-actin (filamentous actin) or inhibiting actomyosin-mediated contractile forces abrogates or attenuates force-induced DHFR transcription, whereas elevating endogenous contractile stresses upregulates force-induced DHFR transcription [195]. These findings suggest that myosin II dependent contractile forces regulate external stress induced transcription and that local stresses applied to integrins propagate from the tensed actin cytoskeleton to the LINC complex and then through lamina-chromatin interactions to directly stretch chromatin and upregulate transcription. Additional findings show that force-induced rapid upregulation of the transgene DHFR applies to simultaneous upregulation of endogenous genes [199] egr-1 (early growth response-1) and Cav1 (caveolin-1), suggesting that forces may act as a “supertranscription factor” on the chromatin [200]. A recent study reveals that force transmission from nuclear lamina to chromatin via nuclear protein LAP2β is critical to chromatin domain stretching (but not compression) dependent rapid transcription upregulation [201], in addition to the force sensor proteins BAF (barrier-to-autointegration factor) and HP1 (heterochromatin protein 1) [195]. The requirement of the intact F-actin-nesprin physical pathway has also been shown for force-induced differentiation of mouse embryonic stem cells [202]. In addition, fibroblasts that are persistently activated in fibrotic tissue exhibit more condensed chromatin than transiently activated fibroblasts and the persistent activation depends on direct force transmission from the matrix to the nucleus via F-actin [203]. It is likely that the rapid direct force transmission pathway from the cell surface into the nucleus and the slow indirect force transduction pathways (plasma membrane ion channels of Piezo1/2 and others, cytoplasmic YAP/TAZ translocation, etc.), together with soluble factors such as growth factors, cytokines, and extracellular vesicles (exosomes and microparticles or microvesicles), synergistically activate numerous genes and assemble abundant proteins at various times to impact functions and fate of living cells including normal stem cells and cancer stem cells. Nevertheless, how forces interact with other factors and the genes to regulate gene transcription and cell fate remains to be elucidated in normal stem cells and cancer stem cells in the future.

IX. Outlooks

It is increasingly evident that forces play essential roles in regulating functions and fate of normal stem cells and cancer stem cells. Substantial insights have been gained on mechanical regulation of embryonic development and cancer progression by seeding these cells in 2D culture and/or in 3D organoids. However, because of technical and ethical issues and incomplete knowledge of the in vivo physiological microenvironment of the embryos and tissues, major challenges remain in understanding of embryonic development of mammals, especially humans, and designing and carrying out intervention strategies in pathological conditions regarding these higher organisms. Various mechanobiology-based strategies have been developed recently toward applications in mechanomedicine. For example, a human airway-on-a-chip, a microfluidic device that contains two parallel microchannels separated by an extracellular matrix-coated porous membrane to mimic human airways in vivo, can rapidly identify candidate antiviral therapeutics and prophylactics [204]; a mesenchymal stromal cell inside a soft gel is used to inhibit fibrotic lung injury in mouse models [205]; a soft implant surface and inhibition of TGF-β1 activation are able to reduce the fibrotic encapsulation of silicone implants in mice [206]; in contrast, a tough hydrogel with an adhesive side and high drug-loading capacity is developed to boost healing in a rat model of Achilles-tendon rupture [207]. During the last decade, there are a few successful examples of using mechanobiology-based approaches for clinical application to treat various human diseases with improved clinical outcomes: a tumor microenvironment normalization strategy to improve immunotherapy [208], fluid shear stress-activated technology for selectively treating local thrombosis [209], a biomaterial-based cancer vaccine technology [210], and a tumor-cell-derived microparticle technology that reverses drug resistance in stage-IV cancer patients [211]. The next few decades should be high time for scientists working in the field of mechanobiology, together with researchers in other fields, to think creatively to develop novel strategies to advance mechanomedicine and mechanohealth (application of mechanobiology to health) to make substantial improvements in medicine and health.

Highlights.

Here we review advances in understanding the impact of forces and mechanics on functions and fate of embryonic stem cells, adult stem cells, and cancer stem cells and the pathways of mechanotransduction in cells. We highlight increasing efforts and emerging favorable clinical outcomes in mechanomedicine, dissemination of mechanobiology discovery and technology to medicine. Major progresses in mechanobiology are pivotal in understanding the life of force and making substantial advances in medicine and health.

Data Availability

All data in this article are from the cited original articles and no original data are generated in this review.