Integrating mechanical signals into cellular identiy

Abstract

The large array of cell types in a multicellular organism are defined by their stereotypic size and/or morphology, and, for cells in vivo, their anatomic positions. Historically, this identity-structure-function correlation was conceptualized as arising from distinct gene expression programs that dictate how cells appear and behave. However, a growing number of studies suggest that a cell’s mechanical state is also an important determinant of its identity, both in lineage-committed cells and pluripotent stem cells. Defining the mechanism by which mechanical inputs influence complex cellular programs remains an area of ongoing investigation. Here, we discuss how the cytoskeleton actively participates in instructing the response of the nucleus and genome to integrate mechanical and biochemical inputs, with a primary focus on the role of the actomyosin-LINC (Linker of Nucleoskeleton and Cytoskeleton) complex axis.

Indirect and direct intracellular mechanotransduction mechanisms

For many years, cellular identity was thought to primarily arise as a result of biochemical activities that dictate a gene expression profile, which in turn defines each cell’s structure and function. In addition, each cell also resides in a unique, often dynamic microenvironment. For example, cells are subjected to contextual mechanical forces – including from interactions with the extracellular matrix (ECM), adjacent cells, and external forces exerted on the tissue as a whole (such as that imposed by blood flow). Characterization of mechanosensitive cellular structures, such as focal adhesions, has revealed that cells can sense and respond to changes in their mechanical environment [1]. This includes mechanical inputs into differentiation programs; for example early evidence that substrate mechanics impact the cell fate adopted by mesenchymal stem cells [2,3]. Building on this, numerous recent studies suggest that the mechanical environment of the cell has an active, rather than passive, role in regulating cellular activity and identity. Thus, mechanistically defining how mechanical inputs influence complex gene expression programs (and how they can be manipulated) is a growing, active area of investigation.

On first principles, the bulk mechanical properties of a cell could be considered as arising from a biological polymer gel, in which its gelling state (and therefore mechanics) is defined by the polymerization and crosslinking of cytoskeletal monomers. A key constituent molecule in this context is actin, since it is ubiquitous, abundant, undergoes dynamic conversion between monomers (G-actin) and polymers (F-actin), can be crosslinked into diverse polymeric networks, and can be acted on by myosin motors to become contractile [4]. Other cytoskeletal polymers such as microtubules [5] and intermediate filaments [6] are also integral components of the mechanotransduction system and frequently engage in crosstalk with the actomyosin cytoskeleton [7]. Moreover, all three cytoskeletal systems interact with the nucleus through various forms of the LINC complex, which spans the nuclear envelope by virtue of lumenal interactions between SUN proteins in the inner nuclear membrane and KASH or Nesprin proteins in the outer nuclear membrane [8]. The cell nucleus is a biochemically and mechanically distinct compartment, typically the stiffest organelle, which houses the biological polymer that underlies cell state: the chromatin. Here we discuss the communication of mechanical signals to the nucleus (referred to here as mechanotransduction) through the lens of two, parallel mechanisms: one indirect and the other direct.

One set of mechanotransduction mechanisms are indirect, in which the state of the actin network, specifically the ratio of G:F actin, regulates the nuclear import and activation of shuttling transcription factors. This type of system excels at sensing changes in the actin network and responding to maintain the G:F actin ratio at a homeostatic set point established through the negative feedback mechanism involving MKL1 (megakaryoblastic leukemia 1, MRTF-A) [9,10]. MKL1 is a ubiquitously expressed transcriptional co-activator that can bind to G-actin via its RPEL domains. G-actin-bound MKL1 localizes to the cytoplasm. Upon signals that induce actin polymerization, G-actin concentration drops thereby revealing the MKL1 nuclear localization signal to drive its translocation into the nucleus where it binds to the Serum Response Factor (SRF) to transactivate genes with CArG elements (CC(A/T)₆GG) in their promoters, including the gene encoding actin. When the G-actin level is restored, MKL1 returns to the cytoplasm. Its sensitivity toward G:F-actin ratio makes MKL1 a central regulator of mechanical signaling (Figure 1A) [10,11]. The other type of mechanotransduction mechanism is direct, in which mechanical signals are physically transmitted across an interconnected tripartite system: from the ECM to the cell interior (via adhesions); traversing the cytoplasm (via the cytoskeleton) and across the nuclear envelope (via LINC complexes) to the nuclear lamina and/or chromatin (Figure 1B) [12]. This type of system is ideal for communicating absolute and possibly more instantaneous mechanical information to the nucleus (e.g. substrate stiffness or critical cell stretching). Considering how signals arising from indirect, homeostatic circuit(s) are integrated with direct force transmission mechanisms to establish cell identity and state is the topic of this review.

Figure 1. Two mechanisms mechanotransduction: indirect and direct.

A) Mechanosensitive transcription factors like MKL1/SRF engage in indirect mechanotransduction. At homeostasis, MKL1 is sequestered in the cytosol by interacting with G-actin. A variety of inputs can lead to an increase in F-actin, and a concurrent decrease in G-actin. This frees MKL1 to translocate into the nucleus, where it forms a complex with SRF to activate target genes that contain CArG motif(s), including many genes that comprise and regulate the actin cytoskeleton. This surge in transcriptional activation leads to the production of more G-actin. The increase in G-actin inhibits MKL1 activity in the nucleus and again promotes its retention in the cytoplasm. As a result, the cell is now at a higher mechanical homeostatic point and poised to respond to another change. This indirect transduction mechanism allows cells to sense changes including in response to the mechanical environment. B) Cellular adhesions are physically coupled to LINC complexes in the nuclear envelope via the cytoskeleton, constituting a tripartite apparatus that can be relaxed or tensed. When cells are in a soft environment, their cell-matrix adhesions exert low levels of tension on LINC complexes. Accordingly, when cells are in a stiff environment, tension on LINC complexes is high. The nucleus can also undergo LINC complex-dependent shape changes in stiffer environments, become flattened and ovoid. This direct transduction mechanism allows cells to sense and transmit static or long-lasting mechanical propeties of the mechanical environment.

Indirect mechanotransduction circuits exert tight homeostatic control

The mechanical set point of the cell is maintained by a powerful feedback system that tightly regulates homeostasis. A major control circuitry involves actin dynamics and the transcription factor complex, MKL1/SRF, as described above (Figure 1A). Remarkably, genes encoding actin and many actin-binding or regulatory factors contain elements recognized by the MKL1/SRF complex in their promoters. Therefore, in response to the drop in G-actin concentration, more actin is produced after a surge in MKL1/SRF activity, leading to the nuclear export/inactivation of MKL1 and returning of the cells to a state of homeostatic F:G actin ratio. Through this mechanism, MKL1 effectively communicates a change, or deviation from the homeostatic mechanical state, and provides the means to adapt to this new mechanical set point via target gene transcription and new protein synthesis. Therefore, while this type of response could be slower in nature as compared to the direct mechanotransduction mechanism discussed below, it could also be longer lasting as part of a new cell identity, for example in the generation of myofibroblasts [13]. Besides directly transactivating expression of components of the actomyosin network, the MKL1/SRF pathway engages in crosstalk with other mechanosensitive transcriptional regulators, such as YAP/TAZ [14]. YAP/TAZ is known to transduce signals downstream of Hippo signaling and regulates organ size [15], but its activity is also sensitive to actin dynamics [16] and can regulate cell size directly [17]. Therefore, actin dynamics defines the cell’s mechanical state and integrates mechanical cues and biochemical singaling via mechanosensitive transcription factors to indirectly modulate gene expression and cellular activity.

Direct mechanotransduction – communicating static mechanical signals

Homeostatic circuits are well suited to respond to changes in the mechanical state of the cell, for example a stimulus that triggers actin polymerization. However, there is also abundant evidence that cells can sense constitutive mechanical signals, such as the stiffness and topology of the ECM in which they reside. For example, the epidermis is maintained by the continual differentiation of keratinocytes from a basal layer of epidermal progenitors that move apically as they differentiate [18]. These progenitors form β1 integrin-based adhesions and β4 integrin-based hemidesmosomes with the underlying ECM as well as cell-cell adhesions with one another [19]. As cells differentiate and begin to traverse apically, they lose integrin-based adhesions; accordingly, β1 integrins are genetically required to maintain the progenitor state and β1 integrin engagement inhibits differentiation [18,20–23]. Engaged focal adhesions elicit both mechanical and biochemical signals [8], both of which could contribute to the mechanisms by which β1 integrin engagement regulates epidermal differentiation. However, as described below, recent evidence supports an intuitive model in which a direct mechanical apparatus connecting the ECM to the genome regulates this genetic program in the context of a long-lived physical state of the microenvironment.

Evidence that the LINC complex is a conduit for direct mechanotransduction

In the example of epidermal progenitors, how is the presence of integrin-based adhesions sensed to promote the maintenance of their fate? LINC complexes are good candidates to act as conduits to transmit extracellular mechanical information directly to the nucleus[24,25]. For example, the nucleus deforms in response to mechanical challenges and this deformation is dependent on LINC complexes[26]. Recent work from our group has implicated the LINC complex in the maintenance of the the epidermal progenitor state, as genetic disruption of SUN proteins leads to precocious epidermal differentiation in the mouse epidermis [27]. In primary mouse keratinocytes we found that tension on LINC complexes is sensitive to substrate stiffness and requires β1 integrins using a fluorescence resonsence energy transfer (FRET)-based tension sensor module (Box 1). Briefly, this sensor takes advantage of a FRET donor connected to a FRET acceptor by an elastic linker, with FRET efficiency decreasing as the fluorophores are pulled apart when the sensor is under tension [28]. Using an in vitro model of keratinocyte differentiation in which cohesive colonies form, we found that tension on LINC complexes is high in focal adhesion-containing cells at the colony periphery, which maintain a progenitor fate. By contrast, keratinocytes at the interior of the colony lack integrin-based adhesions but maintain cell-cell adhesions; these cells display relaxed LINC complexes and are accordingly in a more differentiated state. From these observations we proposed a model in which tension from integrins is transmitted to LINC complexes, which in turn act to maintain a progenitor state by regulating chromatin accessibility of epidermal differentiation genes [27]. More broadly, these findings suggest that the LINC complex is a good candidate for directly transducing the mechanical state of the ECM to the nucleus. Indeed, it supports the existence of a mechanically tense system, connected from integrins to LINC complexes, required for maintaining the chromatin state of epidermal progenitors. These findings also align well with the prior observation that disruption of LINC complexes leads to changes in gene expression [29] and provide a mechanism by which other genes may be regulated by the LINC complex.

Box 1: Tools to modulate and monitor indirect and direct mechanotransduction.

From a technical perspective, it is easier to define downstream consequences of altering cellular mechanical state as opposed to directly and quantitatively measuring the mechanical force exerted on a cell, or protein within a cell. Below we summarize some of the technological advancements that have provided insight into the impact of mechanical forces on cell state (further reviewed in [40]). However, continued technical innovation, especially of methods that can directly measure forces within cells, is critical to elucidate the molecular mechanisms by which changes in mechanical forces (tension or compression) on the nucleus impacts gene expression and cell state / identity.

• Microrheology: Includes methods such as Atomic Force Microscopy (AFM) and micropipette aspiration to apply mechanical force to a sample of interest and measures the response of that sample quantitatively.

  • Pros: Direct, accurate, single cell

  • Cons: Requires specialized equipment; measures bulk properties; limited number of cells; limited time regime (mostly short term, minutes).

• FRET-based tension sensors: Constructed by inserting a tension sensor module (a FRET pair connected by an elastic linker) within a region of a protein of interest that is subject to mechanical force. Changes in FRET efficiency represent changes in tension across the module. These sensors have been used to measure tension on LINC complexes in varying mechanical contexts [27,30,31].

  • Pros: Measurements in single, live cells; can be applied to cells in complex culture conditions and in vivo

  • Cons: Moderate signal to noise; tension sensor insertion is a perturbation of protein architecture

• Laser ablation: Use of a laser to disrupt a cellular structure. This method has been used to disrupt cytoskeletal elements and measure recoil of a structure of interest.

  • Pros: Allows for a critical change to the system and an accurate measure of tension on a structure of interest

  • Cons: Lack of molecular specificity for the structure(s) being ablated.

• Cell/nuclear size/morphology coupled with gene expression and chromatin state: Includes methods such as quantification of nuclear circularity or wrinkling.

  • Pros: Can measure cells regardless of sample heterogeneity; cells are measured in their endogenous state

  • Cons: Indirect (e.g. cannot distinguish if a change in shape is due to more force on the nucleus or the nucleus becoming softer)

In line with our observations in keratinocytes, initial work by Arsenovic et al. indicated that integrin engagement on fibronectin-coated substrates is sufficient to increase tension on LINC complexes [30]. Moreover, Déjardin et al. [31] showed that LINC complexes in cells at the wound edge of a cell monolayer are under higher tension than those at the interior of the monolayer, which have primarily cell-cell adhesions. The cells at the wound edge in this model have acquired more mesenchymal-like characteristics, evocative of a process often referred to as epithelial-to-mesenchymal transition (EMT) in which cells lose markers of epithelia such as expression of E-cadherin, become more contractile and engage focal adhesions. The authors focus their interpretation on the relationship between the extent of LINC complex tension in cells at different positions in the monolayer relative to the wound edge (extent of EMT) and Wnt/β-catenin signaling [31]. However, their findings also reinforce the connection between high tension on LINC complexes and integrin engagement at the wound edge while cells at the interior of the monolayer have primarily cell-cell adhesions and relatively relaxed LINC complexes. In addition, Déjardin et al., demonstrated that as cells traverse through a constriction, LINC complexes located in the constricted region of the nucleus are under less tension, suggesting that LINC complexes may also be sensitive to compressive forces. Last, cells plated at a higher packing density show lower LINC complex tension than cells in a lower packing density[31]. This may provide an additional mechanism by which LINC complexes convey information about the physical environment within a tissue that reflects a relatively long-lived mechanical state.

The relationship between LINC complexes and the actomyosin network

Our work and that of others suggest that cellular engagement with the extracellular environment regulates tension on LINC complexes [27,30,31]. What is the mechanism by which this might occur? One intuitive hypothesis is that tension on LINC complexes is responding to the increase in actomyosin contractility that occurs when cells interact strongly with the ECM. Indeed, it has long been known that tension can be transmitted directly from integrin-based adhesions to the nucleus even before the LINC complex was discovered [32]. After the initial identification of nuclear envelope bridges composed of SUN and KASH proteins, an intuitive model has arisen in which mechanical force can be transmitted from the ECM through cellular adhesions, to the cytoskeleton, and ultimately across the LINC complex to the nucleus, where this mechanical signal can impact activities such as gene expression[24,32]. In line with this model, the first description of a FRET-based Nesprin tension sensor suggested that disrupting myosin II activity leads to a decrease in LINC complex tension [30]. There is also abundant evidence that LINC complexes lie upstream of cytoskeleton organization. In cells lacking SUN proteins or expressing dominant-negative constructs that disrupt LINC complexes, there are changes in the organization of both the actin and microtubule networks [33], consistent with the ability of LINC complexes to work with both actin and microtubules to regulate nuclear positioning in cells [33,34]. Additionally, disruption of LINC complexes leads to an increase in actomyosin contractility (as measured by phosphorylated myosin light chain) in a three-dimensional context (but intriguingly not a two-dimensional context) in MCF-10A and MDCK cell models [35]; we also observed changes in the organization of actin stress fibers in cultured mouse keratinocytes [33]. A similar gain in global contractility observed in breast cancer cells lacking LINC complexes interestingly requires MKL1 [36], again hinting at integration of these mechanoresponsive pathways. Taken together, these findings suggest that LINC complexes are able to regulate the actomyosin cytoskeleton.

Surprisingly, however, the extent of actomyosin contractility may not likewise determine tension on LINC complexes. Indeed, we have made several observations that challenge the simplistic view that tension on LINC complexes necessarily reflects the global extent of actomyosin network tension. For example, our work shows that there is a loss of tension on LINC complexes in keratinocytes lacking β1 integrins [27] despite the fact that these cells exhibit a hypercontractile actomyosin network [37,38]. Additionally, in the context of both keratinocyte colonies and an epithelial monolayer wound edge, engagement of cell-cell adhesions, which drive extensive contractile actin networks [33,39], appears insufficient to induce high levels of LINC complex tension [27,31]. In fact, tension on LINC complexes scales inversely with the presence of E-cadherin in the epithelial monolayer wounding model [31]. The actin network no doubt plays a role in building tension on LINC complexes, and indeed treatment with the actin depolymerizing compound latrunculin releases tension on LINC complexes in keratinocytes [27]. However, taken together, these observations argue that rather than sensing the bulk F:G actin ratio like mechanosensitive transcription factors or global non-muscle myosin activity, LINC complexes instead transmit forces from specific actin-based networks or structures within the cell, such as focal adhesions. Thus, these intriguing observations suggest that, while LINC complexes can modulate the activity of the actomyosin network, the LINC complex may be unresponsive to global actomyosin contractility, at least in some contexts (Figure 2). This nuanced model of mechanotransduction would allow cells to detect the absolute stiffness of their extracellular environment (i.e. do they reside on a soft or stiff ECM) through direct mechanotransduction to LINC complexes independent of global actomyosin contractility. However, since integrin engagement also influences actomyosin contractility, it is likely that the direct and indirect mechanosensing mechanisms are ultimately integrated (Figure 2). Defining the mechanisms by which specific aspects of the actin network engage LINC complexes and how the direct and indirect mechanotransduction pathways are decoded individually remain key challenges for the future.

Figure 2. Model of integrating direct mechanotransduction through LINC complexes and indirect mechanotransduction through mechanosensitive transcription factors to influence cell state.

Model of mechanotransduction in which LINC complexes participate in detecting the stiffness of the mechanical environment (i.e. soft or stiff substrate – top versus bottom). In response to mechanical signals and other signaling inputs that drive increased actin polymerization, mechanosensitive transcription factors translocate into the nucleus (from left to middle panels) where their activity leads to the production of more cytoskeletal monomers and a return to a homeostatic but higher tension state (right panels). This general scheme can repeat itself as tension increases to produce a cell with highly polymerized cytoskeleton.

Both indirect and direct mechanotransduction mechanisms are attenuated in pluripotent stem cells

Despite evidence that the mechanical environment impacts cell fate decisions[2,3], to date the bulk of studies on mechanotransduction have focused on somatic and lineage-committed cells, including those discussed above. Even in cases where stem cells were examined, their fate potential is developmentally limited, such as mesenchymal stem cells or epithelial stem cells. In part, the lack of knowledge into the mechanical differences between lineage-restricted stem cells and their more differentiated progeny reflects that physical properties are more challenging to measure than changes in biochemical characteristics such as marker gene expression, although an expanding toolkit will hopefully make filling this gap increasingly feasible [40] (Box 1).

One exception is the clear evidence for a change in cellular mechanical state during the conversion of somatic cells into pluripotent stem cells by the Yamanaka factors. Pluripotent stem cells, whether embryonic stem cells or induced pluripotent stem cells, are mechanically unique, being overall softer than differentiated cells [41–43]. Genes encoding many components of the mechanotransduction system including transcription factors, channels and numerous cytoskeletal proteins are expressed at much lower levels in pluripotent stem cells (see gene cluster VI in Polo et al. for example [44]). In the context of somatic cell reprogramming into pluripotency, this dramatic and concerted change in the expression of these mechanotransduction genes is widely referred to as the down-regulation of the mesenchymal program [45,46], and mesenchymal-to-epithelial transition (MET) is considered a prerequisite for pluripotency to emerge. Consistent with a mechanically attenuated state, loss of function for most, if not all, genes implicated in both indirect and direct mechanotransduction have minimal deleterious effects on pluripotent cells. For example, although mouse embryos null for SRF [47], MKL1/2 [48,49], lamins [50–52], emerin [53], Nesprins [54,55], and the broadly expressed SUN proteins, Sun1 and Sun2 [56,57], or YAP/TAZ [58,59] have various defects, the pluripotent stem cells themselves invariably appear normal.

We note that cortical actin tension has recently been reported to be required for pluripotency maintainence [60,61]. Indeed, we have also observed that almost all the phalloidin-positive actin network appears to be associated with the cell cortex in pluripotent stem cells [42]. Whether and how cortical actins modulate intracellular mechanotransduction, especially in relation to the nucleus and genome, is unclear. However, the existance of cortical actin tension in pluripotent stem cells lends further support to the concept that bulk properties of the actomyosin network might not always be a faithful indicator of more specific, subcompartmentalized cellular mechanics.

Resetting the mechanotransduction system in reprogramming cell fate

Given the vastly different mechanical properties of most differentiated cells compared to pluripotent cells, it is interesting to consider the role(s) that mechanotransduction systems play in the process of cell fate reprogramming. During transcription factor induced reprogramming, somatic cells, such as fibroblasts, which possess abundant actomyosin contractility, must “reset” their mechanotransduction apparatus (e.g. by downregulating the expression and/or activity of actomyosin components); if they cannot undergo this transition, they fail to enter pluripotency [42,62].

As described above, the indirect mechanotransduction homeostasis circuitry involving actin and MKL1 resists fluctuation and perturbation, perhaps presenting a key barrier to the ability to execute the transition required for reprogramming. Indeed, we found that expressing a constitutively active MKL1 mutant that is defective in G-actin binding (caMKL1) completely blocks the induction of pluripotency [10]. Leveraging this condition, we next attempted to break this circuit. Besides directly knocking down MKL1 itself, we uncovered multiple ways to rescue the blocked cells to effectively release them into pluripotency. For example, caMKL1 blocked cells can be released into pluripotency by inhibiting Arp2/3, a regulator of polymeric branched actin networks; this suggests that the state of the actin network plays a role in regulating reprogramming beyond the activation of MKL1. Interestingly, genetic ablation of LINC complex components also rescues the induction of pluripotency despite caMKL1 expression, suggesting a possible role for direct mechanotransduction. Furthermore, the blocked cells can be released into pluripotency by inhibiting heterochromatin formation using an inhibitor of H3K9me2/3 transferase, suggesting a connection between the mechanical and chromatin states, a notion suggested by recent studies of cell stretching [63]. While it remains possible that sustained MKL1 activity alters a function not related to the mechanical state, the observations that the overwhelming transcriptional changes caused by expressing caMKL1 are related to the actomyosin system and that disrupting the actomyosin network or contractility releases the cells into pluripotency lend strong support to the hypothesis that resetting the cellular mechanical state is required for the emergence of a new cell identity (Figure 3). Given this, manipulating the actin-MKL1 circuitry could provide effective means for cell fate engineering: indeed, multiple small molecule inhibitors in related pathways, such as Rho kinase inhibitors and TGFβ inhibitors, are already staples of pluripotent stem cell derivation or cultivation protocols [45,64–66].

Figure 3. Model of how LINC complex engagement and mechanosensitive transcription factors integrate to influence cell identity.

During differentiation (top left to top right), MKL/SRF are activated and LINC complex engagement increases (the nucleus is under more tension). Reprogramming by the Yamanaka factors from a differentiated to a pluripotent state is blocked by the expression of caMKL1 (bottom right), which is constitutvely nuclear. However, ablation of LINC complexes (or disrupting actin nucleation and polymerization) relieves tension on the nucleus, allowing cells to return to a pluripotent state despite nuclear caMKL1 (bottom left), suggesting that direct mechanotransduction through LINC complexes contributes to maintaining cell fate decisions.

Although the process of resetting the cellular mechanical state may be exaggerated during pluripotency induction from somatic cells, it also likely occurs in physiological contexts, including along the trajectory of developmental cell fate changes and during both tissue homeostasis and pathogenesis [67,68]. Because many of the genes constituting the mesenchymal program encode elements of the cytoskeletal system, we propose that the conversion of cells between set points along the mesenchymal to epithelial spectrum (MET and EMT) reflects the resetting of the cellular mechanotransduction system. Seen in this light, the prevalence of transitting through the EMT/MET landscape co-occurring with many cell fate changes, including many pathological ones, could be understood as the resetting of the cellular mechanical state. Thus, similar to signatures of gene expression and chromatin landscapes, we propose that each cell state is also determined and reflected by a specific mechanical state. In this context, cell identity is established and stabilized by the concerted action of lineage specific transcription factors, epigenetic remodelers and mechanical state modulators, the latter of which could be resistant to change due to its tight homeostatic control.

Concluding Remarks:

As we have articulated above, how mechanical inputs impinge on cell fate depends on their nature (indirect and/or direct), the regime (response to a change in actomyosin contractility or a static/lasting mechanical signal, such as that provided by engaging the ECM), and the context – e.g. lineage commitment point, cues from the tissue microenvironment on cell fate, etc. It is intuitive to expect that the direct and indirect mechanisms are coupled. The indirect, homeostatic circuitry provides the cytoskeletal components that make up the physical apparatus on which forces are directly sensed, transmitted and balanced. For simplicity of discussion, we highlight the particular importance of the LINC complex given its juxtaposition with the genome, and our recent findings demonstrating its role in instructing chromatin organization in pluripotent stem cells and epithelial progenitors [27,42]. In the future, it will be important to define the relevant mechanisms at each junction of the physical apparatus, as well as crosstalk with other components of the mechanotransduction system not covered here (e.g. mechanosensitive channels). Additionally, while the molecular mechanism by which indirect mechanotransduction pathways impact gene expression have been well-characterized (i.e. how mechanical cues impact the nuclear translocation of transcription factors to directly regulate gene expression), analogous molecular details for the direct mechanotransduction via LINC complexes remain unclear and is an important area of ongoing research. Furthermore, even though the mechanosensitive transcription factors do indeed regulate gene expression, their targets rarely dictate cell identity – leaving open the question of how mechanical cues influence cell identity programs at a mechanistic level.

In addition to the actomyosin system focused on here, the integration of other mechanosensitive cellular pathways and structures represent critical areas of study. First, in addition to the actomosin cytoskeleton, the role of microtubules and intermediate filaments in LINC-dependent mechanotransduction, and how they integrate with the integrin-actomyosin-LINC axis, remains to be fully defined. In particular, lamins, type V intermediate filaments present at the inner nuclear membrane, are themselves mechanosensitive. As LINC complexes interact with lamins [8], forces transmitted from LINC complexes to lamins, and the subsequent cellular responses, could represent another component of the direct mechanotransduction pathway described here. Interestingly, lamins have also been shown to influence actin polymerization and could therefore regulate the F:G actin ratio [69], suggesting that lamins could serve as an integration point for direct and indirect mechanical pathways. The nucleus as an organelle has also been suggested to respond to compressive forces to influence actyomyosin contractility [70,71]; further interrogation of the molecular mechanisms underlying this activity await further study. Second, in addition to its interaction with MKL1/SRF signaling, as discussed here, LINC complex-dependent tension has been suggested to influence the translocation of the mechanosensitive transcription factor YAP [72]; furthering our knowledge of the details of this relationship is an important future direction. Finally, recent work has pointed to the nuclear pore complex, the conduit for nuclear-cytoplasmic transport, being responsive to nuclear envelope tension, and may represent another mechanism by which direct mechanotransduction may operate, including to regulate YAP [72–74].

Last, critical consideration of the experimental contexts in which observations are made and reported will be important to arriving at a full and unified understanding of how mechanotransduction contributes to cell fate decisions. Several powerful tools to investigate the impact of mechanical force on cells rely on critical changes to the mechanical environment, which is likely to influence both indirect and direct mechanotransduction systems. Such experimental systems include stretching cell monolayers, pulling on regions of cells by micropipettes or applying forces by cantilevers, all of which induce changes in both the cytoskeleton and forces exerted on adhesions [40]. Insights from such perturbations include the observations that subjecting epithelial monolayers to cyclic biaxial strain leads to an increase in actomyosin contractility, an increase in the repressive H3K27me3 chromatin mark and a global decrease in gene expression [75]. Similarly, subjecting monolayers to uniaxial stretch leads to cell realignment perpendicular to the force, and caused a calcium-dependent transient loss of the repressive H3K9me2,3 mark [63]. Additionally, subjecting human mesenchymal stem cells to cyclic tensile strain leads to changes in post translational modification of the LINC complex component SUN2 and its loss from nuclear envelope; loss of SUN2 alters gene expression in this system, including chromatin-modifying proteins [76]. Relating such in vitro systems, and particularly the strength, frequency, duration and directionality of the mechanical perturbations, to the in vivo context(s) that they recapitulate represents an important future direction. In addition, tools to monitor and manipulate forces in vivo will prove essential (Box 1); molecular tension sensors are already being adapted to meet this need, for example in the study of embryogenesis [77]. Finally, how to manipulate and deliver proper mechanical cues to facilitate cell fate changes to heal wounds, regenerate tissues or to intervene in pathological processes represent the ultimate goal of understanding how mechanotransduction determines cell identity.

Outstanding Questions.

How is the specificity of mechanical cues from the ECM transmitted and decoded?

How do LINC complexes impact the homeostatic setpoint of cell contractility?

What are the nucleus-intrinsic mechanisms that translate mechanical signals to drive changes in gene expression and cell identity?

How are biochemical and mechanical cues integrated during development and tissue homeostasis?

Highlights.

Cells can sense intrinsic and extrinsic mechanical inputs through indirect and direct pathways.

Linker of Nucleoskeleton and Cytoskeleton (LINC) complexes are a conduit for direct mechanotransduction.

Mechanical state impacts cell identity and fate decisions both in lineage-committed progenitors and pluripotent stem cells.

Resetting the cellular mechanical state is a key step of induced pluripotent stem cell reprogramming.

Further developments in tools to robustly measure intracellular mechanical forces will be critical to advance the understanding of the molecular mechanisms of mechanotransduction.