Mechanosensitive membrane proteins: Usual and unusual suspects in mediating mechanotransduction

Some of the best and most important things in life have components that are mechanical—eating and excreting, moving and mating, touching and feeling, hearing and learning, developing and growing. These processes, as well as resistance to mechanical damage and the maintenance of turgor, tension, and other physical states, depend on membrane-based mechanosensors. Cells continuously face mechanical cues such as osmotic stress and stretch, and depend on the fast response (milliseconds) of mechanoelectrical transducers to control cellular cascades that occur on larger timescales (seconds to days). Mechanosensitive (MS) ion channels (the main focus of this Viewpoint) are the usual suspects for mediators of rapid responses to mechanical cues. G-protein coupled receptors (GPCRs) are examples of more unusual suspects, and evidence is emerging that they can mediate responses on slower time scales than MS ion channels.

Simple models for activation of MS ion channels and GPCRs posit that mechanical force catalyzes the transition between inactive and active conformations. Deciphering how this takes place is a vibrant area of research that is highlighted in this Special Issue. In the case of MS ion channels, the two most prominent mechanisms are the force-from-lipid (FFL; Kung, 2005; Teng et al., 2015) and the force-from-filament (FFF; Katta et al., 2015) principles of force delivery. In the FFL principle, active and inactive conformations differ in one or more characteristics, including cross-sectional area, thickness within the bilayer, and induced bilayer curvature. Switching between conformations depends on forces in the membrane bilayer, such as tension and lateral pressure, and its mechanical properties, such as stiffness. In the FFF principle, mechanical force is conveyed to the channel via displacement of one or more protein filaments that link the channel to extracellular and/or intracellular structures. Far from being mutually exclusive, the FFL and FFF principles may act in concert to elevate sensitivity to mechanical stress and enable each MS ion channel to operate consistently with its physiological role and cellular environment.

In this Viewpoint, we discuss features of established and suspected MS membrane proteins, their distribution across protein families and phyla, and current and future challenges in the study of these important proteins and their partners. As many of the studies in this virtual Special Issue pertain to the function of PIEZO1 and PIEZO2, we highlight these proteins in a separate section. We also note that rather than building a comprehensive catalog of MS membrane proteins, we draw examples from the literature in the hope of inspiring accelerated discovery of the biophysics and physiology of membrane mechanosensors. To date, there has been no single path leading to the discovery of MS membrane proteins. Some MS channels, like the MS channels of small and large conductances (MscS and MscL, respectively), have been uncovered through purification and biochemical reconstitution. Unbiased genetic studies identified MEC-4 and MEC-10, paralogous channels belonging to the DEG/ENaC/ASIC superfamily; NOMPC, a TRP channel that is absent from mammalian genomes; TMC1, a.k.a. transmembrane channel-like 1; and OSCA1, a hyperosmolarity-activated calcium channel. Candidate gene screens uncovered PIEZOs and additional OSCA-like channels, while explorations based upon homology yielded the MscS-like (MSL) channels. The idea that GPCRs are mechanosensitive is emerging from structure–function studies of adhesion GPCRs, and has been reinforced by genetic analyses linking their expression to mechanical sensing.

What evidence is needed to advance a given membrane protein from a suspected mechanosensor to a confirmed mechanosensor? At least two of the criteria previously established for mechanosensors involved in sensory mechanotransduction (Ernstrom and Chalfie, 2002; Arnadóttir and Chalfie, 2010; Katta et al., 2015) are broadly applicable to all mechanosensors: function and mimicry.

The protein must be required for responses to mechanical stimuli. These responses should occur on a timescale consistent with the type of putative mechanosensor: ion channels respond in milliseconds or less; other mechanosignaling pathways are slower. Genetic loss-of-function experiments could produce ambiguous results due to redundancy or indirect effects. Combining reverse genetics with the analysis of mutations that change the biophysical properties (e.g., ion selectivity, gating) of the response to mechanical stimulation can help to resolve the ambiguity. This approach was used successfully for channels involved in mechano-electrical transduction channels in Caenorhabditis elegans mechanoreceptor neurons (O’Hagan et al., 2005; Kang et al., 2010) and mouse auditory hair cells (Pan et al., 2018; Pan et al., 2013; Beurg et al., 2015a; Beurg et al., 2021).

As long as a suspected mechanosensor protein functions autonomously, it should retain its MS activity when expressed in another cell or when reconstituted in a lipid bilayer. However, this criterion may not be met if the suspected mechanosensor requires protein or lipid partners that are absent from the heterologous system, if it fails to traffic properly in transfected cells, or if tools available for mechanical stimulation are insufficient to activate it in a heterologous system. Although the mimicry concept is straightforward, experiments testing for it have been problematic. For instance, endogenous MS channels in heterologous cells can function as false mimics (Gottlieb et al., 2008). Complications from endogenous MS channels can be reduced by CRISPR/Cas9 gene editing techniques (Cahalan et al., 2015; Dubin et al., 2017; Moroni et al., 2018). Furthermore, purification and reconstitution of a putative mechanosensor could uncover responses to mechanical stimuli that are suppressed in native tissues, leading to a false positive. These conceptual complications imply that missteps in the field are likely. For instance, several TRP channels initially suspected to be mechanosensitive did not meet one or both of these criteria (Geffeney et al., 2011; Nikolaev et al., 2019; Corey et al., 2004). In summary, gathering and solidifying the multiple lines of evidence needed to establish that a suspected target protein is a true biological mechanosensor is a significant undertaking and is rarely, if ever, accomplished in a single paper.

Force-gated or MS channels are found in protein superfamilies that vary in their phylogenetic distribution (Fig. 1). This is not the only axis of variation, however. Several MS channel monomers have only two transmembrane helices, while others are predicted to contain many more than four. For instance, the bacterial MscK ion channel has 11 transmembrane helices (Mount et al., 2022) and animal PIEZO channel monomers have 38 transmembrane helices (Yang et al., 2022; Wang et al., 2019a). Some MS channels assemble as dimers or trimers, others as tetramers, and still others as pentamers or heptamers. Some, like DEG/ENaC/ASICs and TMCs, operate together with many protein partners. Others in bacteria (MscS, MscL), plants (MSLs, PIEZOs, OSCAs), and animals (PIEZOs, two-pore domain K⁺ channels) can operate autonomously.

Curiously, there is no known amino acid sequence or structural motif that distinguishes MS channels from their cousins who do not appear to be affected by mechanical force. At present, it is not known if ancestral isoforms in these multifunctional superfamilies were all mechanosensitive, and this feature was lost in some proteins, or if mechanosensitivity is a derived innovation. Phylogenetic studies of some MS channels favor the latter viewpoint (Pivetti et al., 2003; Nishii et al., 2021).

Several MS channels help cells sense changes in osmotic pressure and initiate the appropriate physiological responses. For instance, MscS and MscL operate as part of a concerted response to protect bacteria from osmotic challenges of different magnitudes (Levina et al., 1999; Martinac et al., 1987; Sukharev, 2002; Sukharev et al., 1994). They were discovered and cloned via painstaking purification and reconstitution from bacteria and allele replacement, respectively (Sukharev et al., 1994; Levina et al., 1999). The eukaryotic MSL channels have diverse functions in plant and fungal cells (Basu and Haswell, 2017). For instance, the Arabidopsis MSL8 protein plays pivotal roles in pollen rehydration and germination (Hamilton et al., 2015), and MSL10 is key for cellular responses to swelling (Basu and Haswell, 2020). MSL10 and its homologs are implicated in other specialized functions, including long-distance damage signaling (Moe-Lange et al., 2021) and prey detection by carnivorous plants (Procko et al., 2021).

The awesome power of plant genetics led to the discovery of another major family of eukaryotic MS ion channels, the OSCA family of cation channels. OSCA1 (reduced hyperosmolality induced [Ca²⁺]_i increase 1) was discovered in a screen for Arabidopsis thaliana mutants with altered calcium signaling in response to hyperosmotic shock and is required for guard cell closing during drought stress (Yuan et al., 2014; Hou et al., 2014). Multiple cryo-EM structures of plant OSCA channels in the closed state reveal a dimeric channel with two pores and lipid-filled crevices (Zhang et al., 2018; Maity et al., 2019; Liu et al., 2018; Jojoa-Cruz et al., 2022), but the biophysical mechanism of gating remains poorly understood. Other key questions for the future include how OSCA signaling is linked to guard cell function and whether OSCA homologs involved in immune signaling (Thor et al., 2020) and implicated in Venus flytrap closure (Procko et al., 2021; Iosip et al., 2020; Scherzer et al., 2022) are also mechanosensitive.

TMEM63 proteins belong to the same superfamily as OSCA and are also proposed to mediate response to hyperosmotic shock (Zhao et al., 2016). When expressed in mammalian cells, plant and animal OSCA/TMEM63 channels activate in response to both hyper- and hypo-osmotic stimulation (Du et al., 2020). Exactly how these channels are activated within their native cellular context(s) remains to be resolved, however. For instance, TMEM63 may function in insect hygrosensation because it is activated in antennal sensory neurons that bend in response to changes in humidity (Li et al., 2022). In mammals, the TMEM63B protein is expressed in outer hair cells in the inner ear of mammals and is needed for hearing (Du et al., 2020). The deafness phenotype results from degeneration of the outer hair cells, however, leaving open the question of how TMEM63B is needed for outer cell survival. Disparate findings regarding how OSCA and TMEM63 proteins respond to osmotic stimulation may in part reflect the specialized ways in which plant and animal cells respond to osmotic challenges.

Among other functions, animal PIEZO1 channels are important regulators of red blood cell volume (Ma et al., 2018; Zarychanski et al., 2012; Glogowska et al., 2017), wound healing (Holt et al., 2021b), and neuromuscular function (Millet et al., 2022; Bai et al., 2020). The paralogous PIEZO2 channels underpin many aspects of somatosensation and proprioception in mammals (Chesler et al., 2016; Ranade et al., 2014; Woo et al., 2015; Ikeda et al., 2014), fish (Faucherre et al., 2013), and birds (Schneider et al., 2017). In nematodes and insects, PIEZO proteins are vital for feeding (Min et al., 2021; Millet et al., 2022; Wang et al., 2020; Hughes et al., 2022) and mechanical nociception (Kim et al., 2012). PIEZO channels are also widely distributed throughout the plant lineage where they localize to the membrane of the vacuole (Radin et al., 2021) rather than to the plasma membrane. Plant PIEZOs are required for normal calcium transients and vacuole morphology in tip-growing cells in moss protonemata and Arabidopsis pollen tubes (Radin et al., 2021); for normal root growth on hard media, calcium transients in response to touch (Radin et al., 2021; Mousavi et al., 2021); and for defense against systemic viral infection (Zhang et al., 2019).

Among the first ion channels shown to function as mechano-electrical transduction channels in sensory neurons were the C. elegans members of the DEG/ENaC/ASIC superfamily, MEC-4 and MEC-10 (O’Hagan et al., 2005), and the Drosophila TRP superfamily protein NOMPC (Walker et al., 2000; Yan et al., 2013) and its C. elegans ortholog TRP-4 (Li et al., 2006; Kang et al., 2010). MEC-4, MEC-10, and NOMPC were identified through unbiased genetic screens for touch-defective mutant animals (Driscoll and Chalfie, 1991; Huang and Chalfie, 1994; Chalfie and Au, 1989; Kernan et al., 1994). TRP channels are absent from land plants and bacteria, but present in the genomes of most, if not all, animals and single-celled eukaryotes including green algae, paramecia, and some yeast species (Himmel and Cox, 2020; Liebeskind et al., 2015). The DEG/ENaC/ASICs are more exclusive: they are present only in metazoan animals (Liebeskind et al., 2015).

The C. elegans MEC-4 and MEC-10 proteins are co-expressed in touch receptor neurons and contribute to low-threshold touch sensitivity (a.k.a. gentle touch; Driscoll and Chalfie, 1991; Huang and Chalfie, 1994; O’Hagan et al., 2005; Arnadóttir et al., 2011; Chatzigeorgiou et al., 2010a; Suzuki et al., 2003) and substrate vibration (Kubanek et al., 2018; Zhou et al., 2022). Other DEG/ENaC/ASIC proteins contribute to nociception in C. elegans and Drosophila larvae (Geffeney et al., 2011; Chatzigeorgiou et al., 2010b; Zhong et al., 2010; Mauthner et al., 2014) and to proprioception (Tao et al., 2019; Jang et al., 2019; Adams et al., 1998).

The NOMPC channel plays a central role in insect hearing (Kamikouchi et al., 2009; Effertz et al., 2011; Lehnert et al., 2013; Walker et al., 2000) and proprioception in insects and nematodes (Das et al., 2021; Li et al., 2006;Wang et al., 2019b; Cheng et al., 2010). This channel also functions in the lateral line hair cells of zebrafish (Sidi et al., 2003), indicating that NOMPC’s contribution as a mechanosensor is not limited to invertebrates. Mammals lack a NOMPC or TRPN channel ortholog (Peng et al., 2015; Goodman and Schwarz, 2003), suggesting that mechanosensitivity in these channels was either lost in vertebrate lineages or selectively acquired in invertebrates. Mechanical gating of the NOMPC channel depends on links to the microtubule cytoskeleton (Zhang et al., 2015; Wang et al., 2021; Liang et al., 2013), and structural studies reveal a spring-like structure composed of 29 ankyrin repeats in its N-terminal domain (Jin et al., 2017). Molecular dynamics simulations support a model in which channels activate in response to compression of the ankyrin repeat domain (Wang et al., 2021), a process that seems likely to be mediated by NOMPC-microtubule linkages in native cells.

The transmembrane channel-like (TMC) proteins came to light through genetic studies of deafness in humans (Kurima et al., 2002) and mice (Vreugde et al., 2002). Based on sequence similarity and a cryo-EM structure (Jeong et al., 2022), TMC proteins assemble as dimers and resemble the TMEM16 family of lipid scramblases (Ballesteros et al., 2018; Pan et al., 2018). There is ample evidence supporting the idea that TMC1 is a pore-forming subunit of the MS channel responsible for mechanotransduction during hearing (Pan et al., 2018; Beurg et al., 2015a; Beurg et al., 2021). Consistent with a conserved role in animal sensory mechanotransduction, fruit flies rely on TMC to discern food texture (Zhang et al., 2016). Although TMC isoforms do seem to traffic to the plasma membrane in heterologous cells (Kawashima et al., 2011; Labay et al., 2010; Beurg et al., 2015b; Wang et al., 2016; Zhao et al., 2014), two vertebrate isoforms form functional and MS channels when purified and reconstituted in liposomes (Jia et al., 2020). Additional studies are needed to decipher what regulates the assembly of endogenous TMC-containing complexes and what factors are needed to reconstruct these complexes in heterologous cells.

Purified and reconstituted MscS, MscL, MSL1, MSL8, MSL10, TRAAK, TREK-1, TREK-2, and PIEZO1 channels are activated by increases in membrane tension and, therefore, are thought to gate using an FFL mechanism (Lee et al., 2016; Hamilton et al., 2015; Maksaev and Haswell, 2012; Sukharev, 2002; Sukharev et al., 1994; Brohawn et al., 2014; Aryal et al., 2017). Consistent with the idea that the FFL principle also holds in cells, PIEZO1 channels are mobile and do not colocalize with the actin cytoskeleton in mammalian red blood cells (Vaisey et al., 2022). Consistent with theoretical predictions of the energetics of force-gating (Sukharev and Corey, 2004), angstrom-scale structures of MS channels suggest that forces from lipids increase the cross-sectional area that a channel occupies in the plasma membrane (Fig. 2).

Closed and open state cryo-EM structures of AtMSL1, a MscS homolog that localizes to plant mitochondria, suggest an analogous gating mechanism with a literal wrinkle. The transmembrane domains of MSL1 are cup-shaped in the closed state but linear in the open state structure, suggesting that the closed-state channel deforms the membrane into tiny dimples, which are then flattened out once membrane tension is added, driving the transition to the open-state channel (Deng et al., 2020). A cryo-EM structure of the Venus flytrap homolog of MSL10 has a similar membrane-bending transmembrane domain (Procko et al., 2021; Jojoa-Cruz et al., 2022). A related gating mechanism is proposed for PIEZO channels, which are structurally and evolutionarily unrelated to MscS and MSL channels. Specifically, giant PIEZO monomers assemble into trimers that appear to bend the membrane bilayer (Guo and MacKinnon, 2017; Wang et al., 2019a; Yang et al., 2022), an effect proposed to prime the channel for activation in response to membrane stretch (Lin et al., 2019). Collectively, these findings suggest a conserved interplay of channel architecture, channel–membrane interactions, and mechanosensitivity among MS channels operating in the FFL gating mode.

The FFF principle is thought to govern activation of the NOMPC and TMC1/2 channels. NOMPC is connected to the intracellular microtubule cytoskeleton (Liang et al., 2013; Zhang et al., 2015). By contrast, TMC1/2 channels are thought to be connected to tip links, specialized extracellular filaments that attach stereocilia to one another in vertebrate hair cells (reviewed in Zheng and Holt, 2021; Holt et al., 2021a). Although the filament linking NOMPC to microtubules is integral to the channel protein itself (Zhang et al., 2015; Jin et al., 2017; Liang et al., 2011), the connection between TMC1/2 channels and tip links is proposed to be mediated by other proteins such as LHFPL5 and TMIE (Zheng and Holt, 2021). The proteins comprising this mechanotransduction apparatus have been uncovered through genetics in humans, mice, and fish (Holt et al., 2021a). Although nematodes do not have ears, they do harbor a TMC complex that includes orthologs of the proteins found in vertebrates (Jeong et al., 2022; Tang et al., 2020).

Ion channels required for hearing are not the only membrane mechanosensors likely to depend on protein filaments for their in situ activation. Other candidates that may depend on extracellular filaments include the MEC-4–dependent channels responsible for touch in C. elegans (Das et al., 2022,Preprint; Emtage et al., 2004; Sanzeni et al., 2019; Katta et al., 2019) and adhesion GPCRs proposed to activate in response to mechanical forces applied to their extracellular domains (Lin et al, 2022a; Wilde et al, 2022).

It has also become apparent that channels activated by FFL mechanisms could be modulated by FFF and vice versa. For instance, PIEZO2 requires cytoskeletal elements such as filamentous actin and tubulin for normal function (Eijkelkamp et al., 2013; Romero et al., 2020; Verkest et al., 2022). Furthermore, both PIEZO1 and PIEZO2 are reported to be biochemically and functionally tethered to the actin cytoskeleton via the cadherin-β-catenin mechanotransduction complex in MDCK epithelial cells (Wang et al., 2022). Together, these data imply that complex interactions between the membrane and filaments work in concert to tune PIEZO channel function and identify these proteins as channels sensitive to forces delivered by both lipids (FFL) and filaments (FFF).

Several lines of evidence indicate that the composition of the membrane bilayer helps to fine-tune the sensitivity of MS channels. For instance, genetic perturbation of lipid biosynthesis in C. elegans implicates lipids containing the polyunsaturated fatty acid arachidonic acid as regulators of neuronal membrane stiffness and MEC-4–dependent touch sensation (Vásquez et al., 2014). More direct evidence for the role of lipid composition in mechanosensitivity comes from reconstitution of MscL and MscS channels in bilayers (Ridone et al., 2018). When MscL is reconstituted in bilayers composed of lipids with short-chain fatty acids, it is more sensitive to membrane tension than when reconstituted with longer-chain fatty acids (Perozo et al., 2002). Structurally, MscS channels reconstituted in lipid nanodiscs appear to have lipids tightly bound to the channel (Zhang et al., 2021; Rasmussen et al., 2019; Reddy et al., 2019). A non-conducting state is favored in thicker bilayers, and thinning the bilayer by removing lipids from the nanodisc destabilizes this conformation (Zhang et al., 2021).

The two-pore domain K⁺ channels, TRAAK and TREK-1, are activated by perfusion of free fatty acids (Fink et al., 1998; Maingret et al., 2000; Kim et al., 2001) and seem to bind lipids (Schrecke et al., 2021; Cabanos et al., 2017), suggesting that their mechanosensitivity also depends on protein–lipid interactions. Since PIEZO1 channels induce membrane bending (Guo and MacKinnon, 2017; Yang et al., 2022; Lin et al., 2019), they are sensitive to the mechanical and biochemical properties of membrane phospholipids and the presence of cholesterol (Ridone et al., 2020; Shi et al., 2020; Romero et al., 2019). This feature is shared by PIEZO2, as found in coarse-grained molecular dynamics simulation (Lin et al, 2022b; this Special Issue). Indeed, enriching cell membranes in margaric (heptadecanoic) acid, an odd-chain saturated fatty acid, increases bending stiffness and inhibits PIEZO1 and PIEZO2 channel activity (Romero et al., 2019; Romero et al., 2020). These examples illustrate that mechanosensitivity depends on the intimate interplay of lipid bilayers and MS channels. New experimental tools for determining the biochemical composition and mechanical properties of native lipid bilayers and for perturbing these factors would accelerate efforts to fully understand MS channel function in context.

GPCRs play a conserved role in light and chemical sensing, signaling via trimeric G proteins and/or β-arrestin to modulate ion channels or soluble second messengers like Ca²⁺ ions, cAMP, or cGMP. However, evidence is emerging that some GPCRs may respond to mechanical forces in addition to light or chemical ligands. For instance, genetic dissection implicates light-activated opsins (Rh5, Rh6, and NINAE) and an adhesion GPCR latrophilin (dCIRL) in hearing and proprioception in adult and larval fruitflies, respectively (Senthilan et al., 2012; Zanini et al., 2018). Deficits seen in Rh6, NINAE, and dCIRL mutant larvae include disrupted locomotion and reduced mechanically activated neural responses in proprioceptors (Zanini et al., 2018; Scholz et al., 2015, 2017). Although all of these GPCRs satisfy the first rule of evidence for functioning as a membrane mechanosensor, additional studies are needed to directly investigate their contribution to cellular and behavioral responses.

Like other latrophilins, dCIRL has a very large extracellular region that includes an autoproteolytic GAIN domain. Increasing the size of the large extracellular domain of dCIRL disrupts mechanically evoked neural responses, but disabling autoproteolysis by the GAIN domain leaves these responses intact (Scholz et al., 2017). The C. elegans latrophilin ortholog, LAT-1, is also expressed in mechanoreceptor neurons and seems to be required for the function of the male-specific mechanosensory organs essential for mating (Matúš et al., 2022). Transgenic expression of the extracellular N-terminal fragment of LAT-1 rescues these phenotypes in lat-1 mutants, again implicating the extracellular regions of latrophilin in sensory mechanotransduction.

Several other GPCRs are implicated in mechanosensing in the vascular, immune, and nervous systems (see Dunn et al., 2019, and Table 1). Much of the evidence that GPCRs are MS consists of findings that cytoplasmic signals induced by shear stress, vibration, or centrifugation depend on GPCR receptor expression. Intramolecular FRET has also been applied to monitor conformational changes in GPCRs evoked by osmotic stimulation and shear stress (Erdogmus et al., 2019). This approach showed that the histamine receptor HR1 responds to hypo-osmotic saline, independent of agonist binding, and these responses depend on the C-terminal helix 8 (Erdogmus et al., 2019). GPR126 is activated by its binding partners in the extracellular matrix (ECM), collagen IV, and laminin211 (Petersen et al., 2015; Paavola et al., 2014), an effect that is potentiated by mechanical force applied using an atomic force microscope (AFM; Mitgau et al., 2022).

GPR68 (a.k.a. OGR1) is emerging as a candidate membrane mechanosensor based on a cell-based screen (Xu et al., 2018) and for its responses to cell stretching (Wei et al., 2018). GPR68 is conserved and proposed to play an important role in the vascular system based on its expression in endothelial cells and its role in shear-stress sensing in cell lines and primary endothelial cells (Xu et al., 2018). Building on these findings and engineering of other GPCRs, Ozkan et al. (2021) transformed GPR68 into a fluorescent reporter of shear stress by inserting a circularly permuted GFP into an intracellular loop in the protein. The resulting probe, iGlow, is sensitive both to chemical ligands and shear stress (Ozkan et al., 2021). Although they have attracted less attention than channels, the investigation of MS GPCRs could provide new insight into the role of mechanics in cell–cell and cell–matrix interactions and the molecular basis of shear stress sensing. Future studies may uncover both novel biology and biophysics of membrane mechanosensors.

The features that enable mechanosensitivity in cells that differ vastly in size, turgor, and other mechanical aspects remain mysterious and will continue to draw attention from researchers seeking to decipher the interplay of mechanics and biological function. For known and emerging MS membrane proteins, however, some conclusions can be drawn. First, these proteins exist in at least two conformations. Second, the application of mechanical force favors the activated conformation that enables transmembrane ion flux (ion channels) or induces intracellular biochemical signaling pathways (GPCRs). For proteins operating in the FFL mode, the active conformation typically occupies a larger area in the lipid bilayer than the closed conformation. The active conformation may also involve reduced curvature or thinning within the plane of the membrane (Fig. 2). How the active conformation of proteins operating under a FFF principle differs from the closed and/or inactive state is murkier, and clarification will almost certainly depend on future structural studies. Third, no MS protein functions in isolation; all depend on the physicochemical properties of the membrane in which it is embedded, and some also depend on filaments linking the membrane to the ECM, cell wall, or cytoskeleton. In summary, force sensing in biology underpins many fundamental and evolutionarily adaptive functions, including growth and homeostasis, external and internal sensation, moving, eating, and mating. The catalog of MS transmembrane proteins enabling these functions is diverse and continues to expand. Biophysical studies of this MS protein catalog have uncovered some unifying themes for activation, and future studies will help to clarify these mechanisms and reveal new ones.

Jeanne M. Nerbonne served as editor.

We thank the members of our laboratories, A. Ricci, and J. Art, for their feedback on early drafts of the manuscript.

The authors are supported by grants from the National Institutes of Health (R35NS105092 to M.B. Goodman; R01GM133845 to V. Vásquez), the National Science Foundation (MCB 1929355), and by a Howard Hughes Medical Institute-Simons Faculty Scholar Grant (55108530 to E.S. Haswell).

The authors declare no competing financial interests.