The Hippo pathway plays a vital role in controlling cell proliferation by phosphorylating the transcriptional co-activator YAP1/TAZ [
40], subject to regulation by multiple upstream factors [
41], although YAP and TAZ exhibit different translocation dynamics in some cases [
42]. Biochemical signals and mechanical forces both control the Hippo pathway [
43] (
Figure 2B). At low cell density, there is an increased production of amphiregulin, an EGF receptor ligand. This subsequently reduces Hippo pathway signaling, leading to the nuclear translocation of YAP1/TAZ. This process establishes a positive autocrine or paracrine feedback loop that encourages proliferation [
44]. On the contrary, at high cell density, confluent cells are expected to exhibit more paracrine and ECM signaling due to the increased secretion of paracrine factors and deposition of ECM. However, when low-density cells are cultured in tissue culture supernatant from high-density cell culture, they do not cease proliferation. Moreover, some cancer cells continue to grow even at high density, suggesting that the paracrine factors from high-density cells alone are not sufficient to regulate CIP. Instead, biochemical signaling for CIP is mediated by adhesion receptors such as E-cadherin. While E-cadherin regulates CIL through small GTPases and cytoskeletons, CIP is controlled through multiple pathways including growth factor receptors, merlin, the Hippo pathway, the Wnt signaling pathway, and small GTPases [
45,
46]. High cell density inhibits growth factor receptor signaling, such as EGFR and IGFR, in an E-cadherin-dependent manner. In other words, CIP reduces sensitivity to growth factors [
47]. Overexpression of E-cadherin or an increase in cell density leads to the suppression of cell growth by raising the threshold level of the growth factor required for proliferation. Conversely, in E-cadherin knockdown cells, CIP is eliminated, suggesting that growth factor-induced cell proliferation depends on E-cadherin rather than an increase in cell density [
45]. Interestingly, cleavage of the cytoplasmic domains of E-cadherin and N-cadherin by γ-secretase can abolish CIP and induce anchorage-independent growth [
48]. It is worth noting that while the cytoplasmic domains of cadherins induce YAP1/TAZ translocation to the nucleus, the expression of downstream genes like TAZ and connective tissue growth factor (CTGF) remains unaffected. This suggests that YAP1 nuclear localization alone is insufficient as an indicator of YAP1 activity, and additional factors are required to activate YAP1/TAZ. Supporting this observation, Yap1 S112A mice (corresponding to human YAP1 S127) are surprisingly normal despite the nuclear localization of the mutant YAP1, possibly due to homeostatic mechanisms that maintain physiological levels of YAP1 activity [
49]. In addition, resent research suggests that dynamic entry and exit of YAP1 defines its activity. More specifically, majority of Yorkie (the Drosophila ortholog of YAP1) dynamically fluctuates between the cytoplasm and nucleus and a cycle of fast exit of nuclear YAP1 to the cytoplasm followed by fast reentry to the nucleus (“localization-resets”) activates YAP1 target genes [
50]. Furthermore, the extracellular signal-regulated kinase 5 (ERK5)/mitogen-activated protein kinase 7 (MAPK7) is essential for YAP1/TEAD interaction and YAP1 recruitment on DNA in liver cells [
51]. Additionally, Necl-4 and 5 regulate CIP through ERK1/2 (MAPK3/MAPK1) [
52] (
Figure 2). Recently, Junctional adhesion molecule-A (JAM-A) has been found to interact with NF2 and LATS1, thereby activating the Hippo signaling pathway to suppress proliferation [
53].
Mechanical forces can also influence the translocation of YAP1 in a Hippo-independent manner [
54,
55]. Even when the phosphorylation of YAP1 is inhibited by deleting LATS1/2, YAP1 remains in the cytosol of cells on a soft substrate [
54]. This suggests that dephosphorylation of YAP1 alone is not sufficient for its nuclear entry, and a mechanical cue is necessary. The force-induced stretching of the nuclear pore complex (NPC), mediated by the LINC complex and actin cytoskeleton, is responsible for promoting the nuclear entry of YAP1[
56]. On soft substrates, alterations in substrate stiffness specifically impact the nuclear entry of YAP1, while leaving its export unaffected. Consequently, the nuclear import of YAP1 is lower than its export on soft substrates. Conversely, on stiff substrates, the actin cytoskeleton stretches and curves the NPC, exposing its cytoplasmic side, which enhances YAP1 import. The interaction of YAP1 with importin 7 is crucial for the nuclear entry of YAP1, but this interaction is abolished by the phosphorylation of YAP1 on serine 127 (or 112 in mice) by LATS1/2, induced by inhibition of myosin and high cell density [
57]. These findings suggest that the mechanical cue remains crucial even for the Hippo-dependent nuclear entry of YAP1. Even in the presence of the S127A mutant YAP1, which is not affected by LATS-dependent phosphorylation, the nuclear entry of YAP1 is still sensitive to substrate rigidity and mechanical signals [
56].
Recent research has revealed a more intricate mechanism governing the nuclear entry of YAP1. YAP1 interacts with Enigma and Enigma-like proteins (PDLIM7 and PDLIM5, respectively) through their C-terminal PDZ-binding motif, which is vital for YAP1's full nuclear localization and activity. Silencing the expression of PDLIM5/7 leads to a reduction in YAP1's nuclear entry and transcriptional activity, indicating that the opening of the NPC alone is not enough for YAP1's nuclear entry [
58]. Enigma PDLIM7 predominantly localizes to the cytoplasm in densely populated cells. However, in low-density cells, it translocates to F-actin stress fibers, focal adhesions, and F-actin fibers at adherens junctions through alpha-actinin. This localization promotes the tyrosine phosphorylation of YAP1 by Src family kinases, including Yes, in the integrin-Src signaling complex, leading to the activation of YAP1. Phosphorylation on tyrosine 357 sets YAP1 apart from Hippo-dependent phosphorylation on serine 127, as it triggers YAP1 activation and translocation to the nucleus, where it forms a transcriptionally active complex [
59]. Within the integrin-Src mediated mechanotransduction, ANKHD1 and ANKRD17, members of the mask family proteins, play a vital role in facilitating YAP1's nuclear entry [
60]. These proteins possess two ankyrin repeat domains that bind to YAP1 and also contain a conserved nuclear localization sequence and nuclear export sequence, which together regulate YAP1's nuclear import and stability. In Drosophila epithelia, physiological mechanical strain at the apical membrane has been shown to decrease Hippo kinase dimerization, leading to downregulation of Hippo signaling and activation of Yki (YAP1) [
61]. However, it remains to be investigated whether this mechanism can be applied in animal cells.
3.1. Common signaling pathways of CIP and mechanotransduction for the regulation of gene expression
Both CIP and mechanotransduction share common signaling pathways that regulate gene expression through the generation of external or internal forces (
Figure 3). External forces can be sensed by mechanosensitive channels like Piezo and TRP families [
66] and other transmembrane proteins, including integrins, glycocalyx, and anthrax toxin protein receptor 1 [
67,
68]. These forces induce conformational changes in the channels, opening them to allow influx of ions that stimulate downstream signaling events [
69]. For instance, the activation of Piezo1 induces an influx of calcium ions, subsequently activating calcium-dependent myosin II [
70]. The mechanical force that triggers Piezo1 activation also originates from activated myosin II [
71], indicating the presence of a feedback loop that sustains Piezo1 activation. Interestingly, not only activation but also expression of Piezo1 is mechanosensitive, while expression of Piezo2 appears to be less mechanosensitive [
72] and is distinct from Piezo1 for some functions [
73]. Additionally, Piezo1 activates YAP1, and its knockout leads to increased phosphorylated YAP1 [
73].
Calcium ions can also activate other signaling pathways to regulate gene expression [
74]. For example, binding of calcium ions to DREAM disrupts its tetramer formation, leading to the derepression of gene expression [
75]. Furthermore, calcium ions might influence epigenetic modifications like DNA methylation and alternative splicing patterns, thereby modulating gene expression [
74]. These findings indicate that cell-cell contact-induced relaxation of the plasma membrane could inactivate mechanosensitive channels, halting growth by altering gene expression. Glycocalyx and integrins also participate in mechanotransduction to regulate gene expression. For instance, fluid shear stress induces remodeling of the endothelial glycocalyx, activating gene expression [
76]. Similarly, tensile force applied to integrins at the focal adhesion site promotes the activation of focal adhesion kinase and Src kinase, connecting to the LINC complex through the cytoskeleton [
77]. This connection stretches the nuclear pore complex, facilitating nuclear translocation of YAP1 to induce gene expression [
56,
78].
Mechanosensing can occur at the plasma membrane but also at the cellular organelles. For example, mechanical properties of the nucleus and mitochondria can be altered by internal or external forces [
79]. The internal mechanical forces can be generated by myosin contraction and polymerization of the cytoskeleton. Inhibition of myosin II by a specific inhibitor, blebbistatin, or by inhibition of upstream Rho kinase by Y27632 prevents nuclear localization of YAP1 and UBE2A/B [
54,
80,
81]. Treatment that disrupts actin polymerization also opposes nuclear localization of YAP1 and UBE2A/B, presumably because myosin cannot generate force without filamentous actin (F-actin) and/or pushing force generated by polymerization is reduced. Surprisingly, the depletion of F-actin-capping/severing proteins such as cofilin, capZ, or gelsolin is individually sufficient to rescue YAP1 nuclear localization and YAP1-dependent gene expression in high-density cells and cells on a soft substrate [
55]. However, it is worth noting that transgenic gelsolin-null mice appear to have normal embryonic development and longevity [
82].
The typical pushing forces generated by a growing individual F-actin and microtubule are in the range of 0.76 ± 0.22 pN and 3–4 pN, respectively [
83,
84,
85]. In comparison, a single myosin, kinesin, and dynein motors can generate forces of approximately 3-4 pN, 5-6 pN, and ~7pN forces, respectively, though these values may vary depending on the methods used [
86,
87,
88]. This range of forces appears to be sufficient to induce conformational changes in mechanosensitive protein molecules (
Figure 3) [
89].
Unfolding-folding – Force-induced conformational changes expose cryptic binding site or dissociate a binding protein (talin[
90,
91,
92], filamin[
93,
94], vinculin[
95], alpha-catenin[
96,
97], alpha-actinin 1 [
98], lamin A/C[
99], von Willebrand factor [
100,
101], fibronectin[
102,
103,
104]) or substrate site (phosphorylation of p130Cas[
105,
106], FAK[
107], PDZ proteins[
108]).
Catch-slip bond –Catch bonds resist dissociation when subjected to a mechanical force, leading to longer lifetimes with increasing force. On the other hand, slip bonds weaken or have shorter lifetimes when exposed to mechanical force (examples include integrin[
109,
110], L-selectin[
111], P-selectin glycoprotein ligand 1 (PSGL-1)/P-selectin[
112], TCR/pMHC [
113,
114,
115], LFA-1/ICAM-1[
116], cadherin-catenin complex/F-actin[
117], DNAM-1/ CD155[
118]). Catch bond engineering can be employed to tune high-sensitivity TCRs for T cell therapy [
119].
Enzyme activity – Mechanical force exposes the catalytic site of titin kinase by rearranging the autoinhibitory tail [
120]. Forces generated via the actin cytoskeleton in focal adhesions release kinase domain of focal adhesion kinase (FAK) from the FERM domain [
107].
Channels – *Different mechanosensitive channels use different mechanisms to open channels [
66,
121,
122]. Piezo1/2[
123], transient receptor potential channels (TRPs: TRPM, TRPV, TRPC) [
124,
125,
126], TRAAK[
127], TREK [
128,
129], BK channels[
130], OSCA/TMEM63[
131], DEG/ENaC [
132].
Release –Integrin αVβ6 binds pro-TGF-β1. A pulling force transmitted by the β6-subunit through F-actin releases matured TGF-β1[
133,
134].
For example, the cryptic vinculin-binding site of talin can be exposed by 2-12 pN of stretching force [
90] and the filamin mechanosensing domain can be exposed with 2-5 pN force [
135]. Unfolding of α-catenin can occur at 5-15 pN force to promote vinculin binding [
97]. Membrane-embedded Piezo1 channel can open in the range of 6-50 pN force applied to the supporting membrane [
123]. In addition, it was observed that the average force exerted by a single vinculin molecule in stationary focal adhesion is approximately ~2.5 pN [
95]. While a single motor protein may not generate sufficient force, the presence of multiple motors could result in much stronger forces. Since actin–actin bond breaking force and the minimum force required to rupture a microtubule are estimated to be around ~600 pN and ~500 pN, respectively [
136,
137], even with the collective force from multiple motor proteins, these filaments do not break. These findings strongly suggest that mechanotransduction can indeed be regulated by physiological forces within cells [
138].
When cells exert pressure on each other, stress stiffening and stress softening of actin filaments in the leading edge may arise due to the resistance of actin filaments and their buckling against compression [
139]. Although the potential impact of these changes on mechanotransduction has been hypothesized, it has not been demonstrated yet.
It seems that the mechanical stretching of NPC is crucial, at least for the nuclear entry of YAP1 [
56]. Nevertheless, it is possible that signaling molecules downstream of mechanotransduction or cell-cell contact could facilitate the nuclear entry of other TAFs without altering NPC (
Figure 4). For instance, small molecules with a molecular weight of less than 20 kDa might be less sensitive to force-induced NPC opening and could freely enter the nucleus unless import and export are actively regulated [
56]. Additionally, other regulatory mechanisms such as post-translational modifications or ion binding might also control nuclear entry. Furthermore, it is feasible that signals downstream of mechanotransduction or CIP directly stimulate TAFs within the nucleus (
Figure 4). For example, calcium ions might directly or indirectly activate nuclear TAFs [
74]. However, as of now, such mechanisms have not been demonstrated.
Hypothesis: How mechanical forces are converted to biological signals. External or internal mechanical forces are sensed and transmitted to a converter, effector, or regulator (eg. NPC and channel) that controls the localization of TAFs such as YAP1/TAZ and ion influx. TAFs include not only activator and repressor transcription factor proteins but also long noncoding RNAs (lncRNAs) such as Neat1 [
140]. FLNA acts as a sensor, transmitter, and converter [
93]. These converted signals eventually regulate cell growth, differentiation, cell adhesion, shape change, migration, metabolism, etc.
The stretching of NPC facilitates the passive transport of large molecules, but this effect diminishes as the size of the molecule increases, which may account for the differential extent of nuclear entry for UBE2A/B (152aa) and YAP1 (504aa) [
56,
81]. The NPC stretching model proposes that not only YAP1 but also other proteins could potentially move between the nucleus and cytoplasm upon mechanical stimulation. Up to now, a total of 41 nucleocytoplasmic shuttling molecules sensitive to mechanical forces and/or cell density have been identified (
Table S1) [
141]. Surprisingly, even relatively large molecules (eg. SREBP1: 1147aa) can shuttle, indicating that active transport assists in the translocation of large molecules. According from this model, even more nucleocytoplasmic shuttling molecules could be involved in CI and mechanotransduction because there are many small proteins involved in gene regulation. However, the identification of such molecules faces significant challenges primarily due to the lack of appropriate methodologies. Generally, TAFs such as transcription factors (TFs) are expressed at low levels compared to other cellular proteins. As a result, mass spectrometry-based proteomics may detect numerous irrelevant proteins, making it difficult to identify mechano- and CIP-sensitive TAFs.
Despite the numerous observations, it is evident that cell-cell contact alone does not instantaneously induce CIP. For instance, confluent cells exhibit only about a 30% reduction in proliferation and only partial cytoplasmic localization of YAP1, whereas high-density cells fully stall proliferation [
55]. Therefore, it appears to be true that “So far, there is no clear mechanistic connection between CIL and CIP, and they should not be thought of as interrelated processes as some people have suggested” [
6]. However, recent evidence has revealed overlapping signaling pathways between CIL and CIP, as described earlier. To reconcile this apparent discrepancy, several possibilities can be considered. One possibility is that there might be a time lag between cell-cell contact and the full induction of CIP and/or the mechanical changes associated with high cell density. For instance, it has been observed that the size of the nucleus depends on the density of cells, and the size remains relatively constant even when cells come into contact with each other, indicating that NPC is still being stretched at confluency [
55,
81]. Another explanation could be that the duration of signaling defines the cell's fate. It is plausible that prolonged signaling is necessary to fully switch from CIL to CIP. Perhaps a more extended exposure to specific signaling cues is required to trigger the complete transition from cell migration to growth inhibition. Consistent with this concept, prolonged mechanical muscle loading results in an increase in the expression of mechanosensor genes and a switch in muscle fiber type [
142].
In future research, it is essential to investigate the potential common and distinctive signaling pathways for CIP and mechanotransduction as depicted in
Figure 4. However, before exploring into these pathways, it is necessary to identify all the key players, including sensors, transmitters, converters, and TAFs.
3.2. Nuclear lamina in CIP and mechanotransduction
The nuclear lamina, depicted in
Figure 5A, is composed of lamins and nuclear lamin-associated membrane proteins, forming a meshwork lining the inner surface of the nuclear envelope. Lamins are intermediate filaments that can be classified into A-type (lamin-A/C) or B-type (lamin-B1 and B2). The expression of lamin-A/C, but not lamin-B, correlates with tissue stiffness and is upregulated when cells are seeded on stiff substrates or subjected to stretching [
72,
143,
144]. Conversely, culturing cells on a soft substrate or inhibiting myosin II reduces lamin-A/C levels, presumably by increasing their phosphorylation and solubilization [
143]. Adhesion cells express higher levels of lamin-A/C compared to non-adherent cells, underscoring the importance of lamin-A/C in mechanotransduction. Mutations in the LMNA gene lead to defective mechanotransduction and human diseases known as laminopathies. For instance, LMNA mutations causing muscular dystrophies increase YAP1 nuclear entry in muscle stem cells, even at high cell density [
145]. Reduced expression of lamin A/C is also frequently associated with cancer phenotypes [
146]. These mutations and decreased lamin A/C expression can induce genomic instability [
147], potentially by disrupting the normal association of the genome with the nuclear lamina. Dysfunction of lamin A/C would likely impact the stability of the genome, as lamina-associated domains (LADs) occupy about 30–40% of the total genome, and genes localized within LADs are typically transcriptionally silenced [
148].
The LINC complexes play a crucial role in connecting chromatin to the cytoskeletons (
Figure 5A) [
149]. These LINC complexes consist of nesprin proteins located on the outer nuclear membrane (ONM) that bind to cytoskeletons. Nesprins, in turn, interact with SUN proteins, which are linked to inner nuclear membrane (INM) proteins such as lamins, chromatin, and other nuclear envelope proteins. As a result, both external and internal forces are transmitted through the cytoskeleton to the nucleus, leading to changes in nuclear morphology and gene expression [
150]. This force transmission to chromatin may influence DNA mechanics and structure, thereby regulating early transcription initiation events [
151].
The nuclear lamin-associated membrane proteins are found embedded in or associated with the INM or lamins. Among these proteins, lamina-associated polypeptides 1 and 2 (LAP1, LAP2) and lamin B receptor (LBR) have been identified as INM transmembrane proteins that bind heterochromatin through the adapter protein [
152]. Another INM transmembrane protein is emerin, which interacts with other INM proteins, including lamins, NPC proteins, and chromatin [
153]. Moreover, emerin binds to several transcription regulators, such as ß-catenin and BAF. Notably, the loss of emerin has been shown to reduce mechanosensitive gene expression induced by strain stimuli [
154], suggesting that emerin plays a role in mechanotransduction. In addition, non-transmembrane protein PRR14 also connects the nuclear lamina to H3K9me3-modified heterochromatin [
155,
156]. These findings suggest that lamin-associated proteins may play a role in mechanotransduction through cytoskeletons subjected to mechanical force, , although a comprehensive characterization is still pending. Additionally, both lamin A/C and B have the ability to directly bind core histones through a specific sequence element in their tail domain [
157].