Mechanosensitive Ion Channels, Axonal Growth, and Regeneration

Abstract

Cells sense and respond to mechanical stimuli by converting those stimuli into biological signals, a process known as mechanotransduction. Mechanotransduction is essential in diverse cellular functions, including tissue development, touch sensitivity, pain, and neuronal pathfinding. In the search for key players of mechanotransduction, several families of ion channels were identified as being mechanosensitive and were demonstrated to be activated directly by mechanical forces in both the membrane bilayer and the cytoskeleton. More recently, Piezo ion channels were discovered as a bona fide mechanosensitive ion channel, and its characterization led to a cascade of research that revealed the diverse functions of Piezo proteins and, in particular, their involvement in neuronal repair.

Introduction

Neurons use proteins that are capable of transducing force from physical stimuli such as touch and sound into biochemical signals, a process called mechanotransduction. Broadly speaking, there are two flavors of mechanotransduction: the sensing of mechanical stimuli at the organism level (e.g., touch sensation and hearing), which largely relies on mechanosensitive (MS) ion channels, and cellular mechanotransduction, through which cells grow, migrate, and interact, using mechanical molecular linkages and MS ion channels. This allows the nervous system, in particular, to adapt to its changing environment, for example, during neuronal growth (Gottlieb and others 2010) and pathfinding (Koser and others 2016) and process events including touch sensitivity (Kim and others 2012) and hearing (Friedman and others 2007). Some of the earliest work on mechanotransduction was performed in bull frogs, with in vitro microphonic recordings of bull frog vestibular hair cells showing rapid responses to mechanical stimuli. The response time between exposure to stimulus to electrical signal transduction was deemed too rapid for the involvement of intermediate chemical messengers (Corey and Hudspeth 1979). Therefore, the electric signals must have resulted directly from the activity of nearby transmembrane channels, later recognized as MS ion channels. MS ion channels are present in specialized cells of multiple organisms, and in mammals, the roles of MS proteins include touch response (Chalfie and Au 1989; Yan and others 2013), tendon-muscle stretch (Lansman and Franco-Obregon 2006; Woo and others 2015), sound detection (Brown and others 2008), and response to blood pressure changes (Chapleau and others 2007). Over the previous decades, research has revealed several classes of ion channels involved in mechanotransduction: DEG/ENaC, K_2P, TRP, TMEM16, and Piezo channels, and current work has focused on identifying additional protein candidates, determining their MS nature, characterizing their physiological roles, and discovering their associated molecular pathways. The topic on MS ion channels and sensation has been reviewed elsewhere. Recent work has also demonstrated that mechanosensation mediates the neural regenerative process, but the specific underlying mechanisms involving MS proteins remain less well understood. In this review, we highlight the latest progress in the field by first discussing how neurons sense mechanical stimuli through the adaptation of mechanically sensitive proteins with a focus on Piezo channels followed by how Piezo channels influence neural regeneration after injury.

How Neurons Sense Force

Cell-Extracellular Matrix Adhesion

The ability of a neuron to sense and respond to its environment is crucial for proper neural development, repair, and discovery of therapies. As motile structures at neurite tips, growth cones sense and respond to their environment to guide neurites to their proper targets. Growth cone behavior is regulated by cytoskeletal rearrangements, axon guidance cues, and interactions with the extracellular matrix (ECM) (Mai and others 2009; Wen and Zheng 2006). The ECM serves as an external scaffold comprising structural proteins, such as fibronectin and laminin, and directly couples to the intracellular cytoskeleton of the cell via integrin activation (Ingber 1997, 2006; Matthews and others 2007). The physical attachment of the cell to the ECM forms focal adhesion complexes that allow for the direct and indirect transmission of forces to locally placed membrane proteins (Cao and others 2013; Cheng and others 2007). The leading-edge protrusion at growth cones is driven by F-actin polymerizations that generate tensile forces and a rearward flow of F-actin (Carlier and Pantaloni 2007; Lin and others 1996; Mogilner and Oster 2003; Symons and Mitchison 1991). Nonmuscle myosin II binds to actin bundles to generate rearward contractile forces that, when coupled to the rearward flow of F-actin, contributes to what is known as “F-actin retrograde flow” (RF) (Brown and Bridgman 2003; Forscher and Smith 1988; Lin and Forscher 1995; Medeiros and others 2006; Shin and others 2014; Turney and Bridgman 2005). To oppose F-actin RF and achieve greater growth cone protrusions, “molecular clutches” at cell-ECM focal adhesion sites clutch and restrict myosin II–mediated contractile RF forces, thus increasing actin polymerization at the leading edges of growth cones and allowing the cell to propel forward. This is called the “molecular clutch” hypothesis and explains how local reductions in RF correlate with increased growth cone motility and growth cone forward propulsion (Bard and others 2008; Lin and Forscher 1995; Mitchison and Kirschner 1988; Myers and Gomez 2011; Santiago-Medina and others 2013; Shimada and others 2008; Suter and others 1998; Swaminathan and Waterman 2016; Toriyama and others 2013; Woo and Gomez 2006).

Mechanosensitive Cues

When growth cones adhere to their environment, adhesion complexes assemble and comprise several MS proteins to provide signaling and scaffolding functions at sites of cell-ECM adhesion. The ECM contains several ligands that influence growth cone motility, and each ligand activates a specific integrin receptor on growth cones. Upon integrin activation, scaffolding and signaling molecules, such as talin, vinculin, filamin, and focal adhesion kinase (FAK), are recruited to adhesion sites and reinforce cell-ECM attachment by bridging actin to the ECM (Cluzel and others 2005; Gomez and others 1996; Myers and Gomez 2011; Robles and Gomez 2006). These scaffold proteins act as mechanosensors by stretching to accommodate growth cone dynamics in different environments. FAK is an essential protein during point contact, and disrupting its activity diminishes cell adhesion. For example, disrupting FAK activity disrupts adhesion complexes and decreases traction force generation in growth cones induced by netrin-1 (Moore and others 2012). As growth cones migrate, contractile forces cause additional proteins such as talin to stretch and expose binding sites for vinculin and other scaffold proteins to attach and support integrin-actin linkages (del Rio and others 2009; Margadant and others 2011). Talin is another essential MS scaffold protein in point contact adhesion, and its disruption leads to decreased growth cone motility (Kerstein and others 2013; Sydor and others 1996). In another example, filamin is positioned between integrins and F-actin and unfolds under shear force to expose binding sites for signaling molecules such as RhoA, ROCK, and PKC (Furuike and others 2001; Razinia and others 2012).

Mechanosensitive Ion Channels

MS ion channels are abundantly found at cell-cell interfaces and integrin-ECM adhesions. In neurons, MS ion channels comprise a second group of MS cues and mediate axon growth. TRPC5 and Piezo ion channels, for example, negatively and positively regulate axon growth in neurons (Koser and others 2016; Oda and others 2020). Furthermore, mechanical stimuli caused by cortical folding during primate brain development differentially stretch neurons, depending on which layer the neurons reside, and therefore likely act on MS ion channels (Hilgetag and Barbas 2005; Martinac and Cox 2017).

MS ion channels help control growth cone activity by regulating the influx of ions as growth cones navigate through the ECM. The stiffness of the ECM changes with changes in substrate composition (such as laminin, fibronectin, and collagen), and neurons often adapt to such changes by using Ca²⁺ signaling (Franze and others 2009; Jacques-Fricke and others 2006; Kerstein and others 2013; Lee and others 1999; Wei and others 2009). Since Ca²⁺ signals localize to adhesion sites and areas of higher traction forces, cell-substratum interactions can activate MS channels in motile cells, with traction force increasing with substrate stiffness (Doyle and others 2004; Franze and others 2009; Gomez and others 2001; Janmey and others 2020). In addition, stiffer substrata promote Ca²⁺ transients in cells, thus providing evidence that substratum stiffness (the mechanical stimulus) modulates Ca²⁺ influx and restricts axon extension (Kerstein and others 2013; Kim and others 2009). Intriguingly, the interaction between cells and their environment is inherently bidirectional, as is the concept of mechano-reciprocity (Paszek and Weaver 2004). It is defined as an iterative process, through which cells modify the organization and elastic response of the environment (e.g., substrate stiffness) and reciprocally adjust their behavior (Paszek and Weaver 2004; van Helvert and others 2018). It is thus an adaptive process and may affect the generation, detection, processing, and response to mechanical forces.

While several MS ion channels have been identified, their underlying mechanisms have not been fully defined. Pharmacologically, MS ion channels can be inhibited by GsMTx4, Gd³⁺, and gentamicin, leading to reduced secondary messenger-mediated signaling. However, the transduction process is thought to involve interactions with the cytoskeleton, lipids, and second messengers, including Ca²⁺, that activate biochemical pathways, such as the CaMKII signaling, to regulate axon growth (Song and others 2019). Generally speaking, the cell membrane can experience endogenous forces, which are exerted largely by the cytoskeleton and the acto-myosin machinery, and external forces that can generate contractile tension, compression, shear stress, membrane swelling, and membrane curvature (Chen 2008; Ingber 1997, 2005). MS channels can be gated by one or more types of forces.

Several strategies have been used to identify MS ion channels and their key structures responsible for mechanotransduction. Some of the earliest known MS channels were discovered in bacteria, where it was observed that membrane stretch of spheroplasts opened a channel that resulted in a large conductance (Martinac and others 1987; Sukharev and others 1993, 1994). These channels were subsequently named MscL and MscS (mechanosensitive channel of large and small conductance, respectively), which have been recently reviewed in detail (Cox and others 2018), and here we mainly focus on metazoan MS channels. Genetic screens were used to first identify MS ion channels in mutants with touch-insensitivity in Caenorhabditis elegans (DEG/ENaC), bristle mechanosensation defects in Drosophila (TRP channels), and hearing loss in mice (TRPA1) (Brown and others 2008; Chalfie and Au 1989; Chalfie and Sulston 1981; Corey and others 2004; Friedman and others 2007; Kernan 2007; Sulston and others 1975). From then on, most MS channels were discovered through candidate-based and genetic screens such as with the discovery of PIEZO channels using microarray analysis (Coste and others 2010). The identification of PIEZO proteins allowed subsequent mechanistic studies revealing direct linkages of PIEZO1 and PIEZO2 mutations in diseases such as dehydrated hereditary stomatocytosis (PIEZO1) (Albuisson and others 2013; Andolfo and others 2013; Bae and others 2013; Fotiou and others 2019; Lukacs and others 2015; Masingue and others 2019; Yamaguchi and others 2019). Studying the role of Piezo channels in disease has helped map the location of Piezo channels in different tissue types in both mammalian and nonmammalian subjects: red blood cells, somatic stem cells, neurons, primary articular chondrocytes, and cardiomyocytes, to name a few (He and others 2018; Hennes and others 2019; Jiang and others 2021a; Lee and others 2014).

DEG/ENaC Channels

Selective for Na⁺ ions, the DEG/ENaC superfamily of ion channels comprises degenerin (DEG), mammalian epithelial Na⁺ channel (ENaC), the acid sensing ion channel (ASIC) proteins, Drosophila pickpocket (PPK) and ripped pocket (RPK), and Hydra sodium channels (HyNaCs), and certain members have been shown to be mechanosensitive (Adams and others 1998; Assmann and others 2014; Kellenberger and Schild 2002; O’Hagan and others 2005; Stockand 2015; Zhong and others 2010). DEG/ENaC channels share a common trimeric structure, and genes for DEG/ENaC family were first identified in C. elegans during a screen of mutants defective in light touch sensitivity (Chalfie and Au 1989; Chalfie and Sulston 1981). In Drosophila, PPK encodes for a subunit of DEG/ENaC channels and is required for mechanical nociception in larvae but not for thermo nociception (Zhong and others 2010). There is little evidence of HyNaCs and ASICs conferring mechanosensitivity, but their structures still offer insights into mechanosensation of other MS channels.

Degenerin proteins include UNC, MEC, DEG, and DEL, but not all of these proteins represent ion channel subunits (Martinac and Cox 2017). The core structure of the ion transduction apparatus appears to be formed by MEC-4 homotrimers and MEC-4/MEC-10 heterotrimers that interact with intra- and extracellular auxiliary MEC/DEG counterparts to form a macromolecular touch-sensitive complex (Chen and others 2015; Driscoll and Chalfie 1991; Jasti and others 2007; O’Hagan and others 2005). Mutations of the C. elegans mechanosensory abnormality mec-4 and mec-10 alter channel selectivity and kinetic properties based on in vivo electrophysiological recordings of touch neurons and result in touch-insensitive animals (Arnadottir and others 2011; Chalfie and Sulston 1981; O’Hagan and others 2005). Other degenerin proteins exhibit varying degrees of mechanosensitivity (Martinac and Cox 2017), and some have weaker evidence of mechanosensitivity while others (UNC-7) demonstrate more direct evidence of mechanosensation in which their removal results in touch insensitivity (Walker and Schafer 2020).

The mammalian structural homolog of DEG is the epithelial sodium channel (ENaC), the second component of the DEG/ENaC superfamily. There are conflicting results on the true MS nature of ENaCs, as some studies demonstrate that ENaCs activate via laminar shear stress while lipid bilayer reconstitution assays fail to show mechanically activated (MA) currents (Awayda and others 1995; Awayda and Subramanyam 1998; Carattino and others 2004, 2005; Ji and others 1998; Rossier 1998).

K_2p Channels

Within the potassium channel family, several members of the two-pore domain (K_2p) subfamily exhibit MS properties. Present in both excitable and nonexcitable cells, K_2p channels leak potassium conductance to maintain negative resting membrane potential, cell excitability, and the transport of solutes (Hughes and others 2017; O’Connell and others 2002; Schewe and others 2016). The 15 types of K_2p channels are organized into the following subgroups: TWIK, THIK, TASIK, TALK, TREK, and TRESK and are distributed in multiple cell types (central nervous system, peripheral nervous system, myocytes, endothelial cells, lungs, kidneys, liver) (Bang and others 2000; Bockenhauer and others 2001; Fink and others 1996, 1998; Girard and others 2001; Lesage and others 1996, 2000; Rajan and others 2001; Salinas and others 1999). K_2p channels can also be activated by physical, chemical, and biochemical stimuli (Honore and others 2002, 2006; Maingret and others 1999b, 2000a, 2000b). To date, TREK-1, TREK-2, and TRAAK (all from the TREK family) are known to be mechano-gated in mammals. Studies of TREK-1, TREK-2, and TRAAK activation dynamics in reconstituted liposomes suggested that their channel activities were directly dependent on inherent changes in the lipid membrane in the absence of additional cellular components. Aryal and others (2017) showed that TREK-2 switches from a “down” to “up” conformation in the presence of negative pressure that causes membrane stretch. Berrier and others (2013) showed that TREK-1 is reversibly closed in the presence of positive (but not negative) pressure. Meanwhile, Brohawn and others (2014a, 2014b) demonstrated that TREK-1 and TRAAK channels were activated to both positive and negative pressures. These studies concluded that TREK-1, TREK-2, and TRAAK2 activate through membrane tension alone, much like bacteria MS channels, thus establishing the MS function of these K_2p channels. Interestingly, K_2p channels can be activated not only by mechanical stimulation but also by changes in cytosolic alkalinity, acidosis, and temperature. In addition, some evidence suggests that K_2p channel activation can be regulated by amphipathic molecules, depending on how these molecules induce local membrane curvature (Bavi and others 2016; Maingret and others 1999a; Patel and Honore 2001; Patel and others 1998). Thus, K_2p channels (such as TREK and TRAAK channels) are both thermosensitive and mechanically sensitive, and our current understanding of their structure could provide valuable insights into how other ion channels sense mechanical stimuli. Furthermore, TREK-1 channels can be modulated pharmacologically, offering K_2p channels to be potential therapeutic targets (Dong and others 2015; Heurteaux and others 2004; Kennard and others 2005; Patel and Honore 2001; Patel and others 1999).

TRP Channels

Initially discovered in the fly trp mutant, transient receptor potential (TRP) channels respond to light and play important roles in calcium signaling and sensory physiology (Cosens and Manning 1969; Minke 1977; Ramsey and others 2006). TRP channels are nonselective toward cations and are activated by ligands, membrane potential, osmolarity, and both physical and chemical stimuli. The 33 genes of the mammalian TRP superfamily are categorized into seven subcategories (TRPC, TRPV, TRPA, TRPM, TRPP, TRPML, and TRPN) (Martinac and Cox 2017; Montell 2001) and confer MS properties (Julius 2013; Ramsey and others 2006; Zheng 2013) by activating upon changes in cell membrane. Of the TRP channels, TRPN fits the criteria of a true MS channel (Zhang and others 2015) while the MS nature of other TRP channels is still debated such as with TRPC1 and TRPC6, in which MA currents were detected in some studies but not in others (Gottlieb and others 2008; Kerstein and others 2013; Maroto and others 2005; Spassova and others 2006).

NompC (no mechanoreceptor potential C) of the TRPN family is the most well-characterized TRP MS channel and is activated through a “dual-tether” gating mechanism via its ankyrin repeats (see “Dual-Tethered Model“ below) (Wang and others 2021; Zhang and others 2015). Like many other ion channels, NompC was discovered through a genetic screen and is essential in Drosophila mechanosensory responses (Walker and others 2000). NompC homologs were found in other organisms such as C. elegans (TRP4) and in zebrafish (TRPN1), but no homologs have yet been identified in mammals (Kang and others 2010; Shin and others 2005). Despite our understanding of certain TRP channels’ role in mechanotransduction such as with NompC, studies on the true MS nature of certain TRP channels, such as with TRPC channels, have yielded more correlative results than definitive links to mechanosensation (Suchyna and others 2004b). For example, TRPC6 has been implicated in several mechanosensory processes such as myogenic tone and the vasoconstriction. When the expression of TRPC6 was reduced while in the presence of pressure, there was a decrease in arterial smooth muscle depolarization and blood vessel constriction in mice and rat, but a causative link between TRPC6 and actual pressure sensing has yet to established (Dietrich and others 2005; Welsh and others 2002).

TMEM16 Superfamily

The transmembrane member 16 (TMEM16) superfamily, also referred to as the anoctamin (ANO) family, is made up of a diverse set of transmembrane proteins (TMEM16A–TMEM16K) that are widely expressed across eukaryotes (Duran and others 2012; Whitlock and Hartzell 2016). There are also several distinct families of proteins that are considered evolutionarily related to TMEM16, including the transmembrane channel-like (TMC) and hyperosmolarity-gated calcium-permeable OSCA/TMEM63 families, which have MS properties (Hou and others 2014; Liu and others 2018; Medrano-Soto and others 2018; Whitlock and Hartzell 2017).

TMC Channels

TMC channels are composed of three subgroups based on amino acid sequence: TMC1–3, TMC5/6, and TMC4/7/8 (Kurima and others 2003). TMC1 and TMC2 are possibly pore-forming subunits of MS channels concentrated at the tip of hair cell stereocilia, the site of cochlear mechanotransduction, and serve a crucial function in hearing (Beurg and others 2009; Corey and Holt 2016; Kawashima and others 2011; Kurima and others 2015; Qiu and Muller 2018). However, there has been some debate as to whether TMC1 and TMC2 are intrinsically mechanosensitive or constituents of a larger transduction channel, given their functional reliance on auxiliary proteins (Corey and others 2019; Gleason and others 2009; Maeda and others 2014; Tang and others 2020; Xiao and Xu 2010; Zhao and others 2014). To address this question, Jia and others (2020) reconstituted green sea turtle TMC1 (CmTMC1) and budgerigar TMC2 (MuTMC2) into liposomes and measured channel current in response to mechanical pressure. These results, taken with the observation that CmTMC1 mutants had diminished ion channel activity, provide physiological evidence that CmTMC1 and MuTMC2 are pore-forming MS channels (Jia and others 2020). Given the broad conservation of homologous TMC genes across species, it is possible that mouse and human TMC channels are mechanosensitive, as well. Structurally, TMC channels resemble TMEM16 anion channels and lipid scramblases, assembled as dimers with discrete ions conduction pathways in each monomer (Brunner and others 2014; Corey and others 2019; Middleton and others 1996; Whitlock and Hartzell 2016). This is in contrast with other MS channels that form as tetramers, pentamers, or heptamers and contain a single ion pore (Corey and others 2019; Haswell and others 2011). Although topological studies have not been definitive, several groups have proposed that TMC channels are composed of 10 transmembrane domains, with ionselective S4 to S7 forming the pore (Ballesteros and others 2018; Corey and others 2019; Pan and others 2018). It remains to be determined whether the pore is entirely bound by these helices or stabilized by auxiliary factors (Corey and others 2019).

OSCA/TMEM63 Superfamily

Of the known MS ion channels, the OSCA/TMEM63 superfamily may be the largest with 15 distinct members (Hou and others 2014; Murthy and others 2018; Yuan and others 2014). It has been suggested that changes in osmotic stress alter membrane tension through cell shrinking and swelling, and hyperosmolarity activates OSCA proteins and the initiation of the hyperosmotic stress response (Jojoa-Cruz and others 2018; Liu and others 2018; Sachs 2010). Furthermore, recordings of MA whole-cell currents demonstrated that the Arabidopsis thaliana OSCAs, AtOSCA1.1 and AtOSCA1.2, are activated by cell membrane stretch caused by indentation (Murthy and others 2018) but maintain a higher threshold than mouse PIEZO1 (Dubin and others 2017). Additional evidence of MS properties of OSCA1.2 was shown in reconstitution assays in which stretch-activated currents were recorded in excised proteoliposome patches (Murthy and others 2018). Other OSCA proteins produce stretch-activated currents, including OSCA1.8, OSCA2.3, and OSCA3.1, although whether these qualify as bona-fide MA ion channels remains debated (Murthy and others 2018; Zhang and others 2018). The mechanotransduction machinery of OSCA family proteins remains largely unknown, but recent structural studies in AtOSCA have highlighted the anchor region that connects TM0 and TM1 and the cytosolic linker that connects TM1 and TM2 as potentially involved in osmosensing (Jojoa-Cruz and others 2018; Liu and others 2018). Further evidence of the MS characteristics of OSCA proteins was shown when TMEM63, the mammalian homolog of OSCA, was heterologously expressed in Drosophila, mouse, and humans and negative pressure was applied to an excision patch to activate the channels (Murthy and others 2018). These results indicate that the MS function of OSCA proteins is also evolutionarily conserved across species despite an 80% difference in their primary amino acid sequences (Jin and others 2020).

Piezo Channels

More recently discovered, PIEZO channels (of the Fam38a/b family) control cell migration in nonneuronal cell types and were shown to inhibit axon regeneration of rat hippocampal neurons and the outgrowth of mouse peripheral neurons (Li and others 2021; Song and others 2019). A more detailed description of Piezo channels will be provided in later sections of this review, but briefly, Piezo proteins can be activated by tension from the plasma membrane, the tensile force generated by the cytoskeleton (Coste and others 2010; Cox and others 2016; Lewis and Grandl 2015), shear stress (Hyman and others 2017; Li and others 2014; Ranade and others 2014), and compressive forces (Lee and others 2014; Luo and others 2022). Cells that lack Fam38a activity revert to an amoeboid-like migration independent of integrin activation, require less anchorage, and require extensive cytoskeletal rearrangement to push through the ECM. In neurons, axons extend farther in their environment than they would if Fam38 is present (McHugh and others 2012). Additional details will be discussed below under “Piezo.”

Gating Models of MS Ion Channels

Force-from-Lipids Model

In the force-from-lipids (FFL) model, MS ion channels are gated only by forces generated by the lipid membrane, making it the simplest method of activation for MS ion channels, because the FFL model does not require additional external adaptor protein structures (Teng and others 2015). When the membrane tension changes, the lipid bilayer exerts forces on the protein’s structural elements and induces conformational changes that allow the channel to find a new state of equilibrium. Thus, when membrane tension crosses a certain threshold, lipid molecules pull the MS channel into its open state (Kocer 2015; Zhang and others 2016). Studies on the FFL model began in bacterial channels, MscL and MscS, and were found to be gated by changes in lateral membrane tension more so than by membrane curvature or indentation (Moe and Blount 2005; Teng and others 2015). Later studies identified additional channels that are gated by forces from lipids: TREK/TRAAK and PIEZO1 channels (Brohawn and others 2012; Coste and others 2010; Lolicato and others 2014; Ranade and others 2015; Saotome and others 2018; Zhao and others 2018).

Single-Tether Model

While the force-from-lipid model suggests that changes in lateral membrane tension are sufficient to gate certain MS channels, the single-tether model proposes that the repositioning of channels exposes the channels to different forces in the membrane. Channels could either be moved into or out of the plane of the membrane and can therefore be exposed to and associate with different ECM components that then act as a single tether. In doing so, dual directional signaling can occur, but unlike the dual-tether model (see below), showing concrete examples of tethering through assays such as reconstitution or expression in heterologous systems would be difficult as both components, including the ECM itself, would need to be replicated (Bounoutas and Chalfie 2007; Kung 2005).

Dual-Tether Model

In this model, the MS channel is tethered intra- and extracellularly and activates/inactivates by undergoing conformational changes due to its linkages to the tethers (Gillespie and Walker 2001; Martinac and Cox 2017; Sotomayor and others 2005). One of the earliest studies on the dual-tether model was performed on the MS receptors of the auditory and vestibular systems of vertebrate inner ear (Howard and Hudspeth 1987, 1988). It was found that soundwaves (the physical stimulus) deflect hair cell bundles, and within hair cell bundles, specialized microvilli (stereocilia) are arranged in rows of three and interconnected by tip links at the top. MS ion channels located at the top of the shorter stereocilia are gated by mechanical forces caused by the deflection of the hair cells. The gating spring is thought to be either the tip link or the ankyrin repeats of TRPA1 and TRPN1 (Sotomayor and others 2005). In another example, NompC ion channels are tethered to the cytoskeleton via a filament of 29 ankryin repeats, and both structures are required for NompC activation in Drosophila touch sensation (Zhang and others 2015). Interestingly, when the ankyrin repeats were transferred to voltage-gated potassium channels, mechanosensation properties were conferred on these previously non-MS channels (Zhang and others 2015). Recently, mouse PIEZO1 was shown to be the first mammalian MS channel to use a tether model for mechanogating by tethering to the actin cytoskeleton through the mechanotransduction complex (made of E-cadherin, β-catenin, and vinculin) (Wang and others 2022).

Piezo Proteins

Discovery

Piezo was first identified and characterized in 2010 in the Patapoutian lab through the efforts of Coste and others (2010). Through a screen that used whole-cell patch clamp recordings, it was determined that mouse neuro2A (N2A) cells responded to different mechanical stimuli while producing mechanically activated (MA) currents most similar to that of primary cells. Therefore, mouse N2A cells were used in subsequent screens for cation channels and proteins that were predicted to span the membrane at least twice. Over 70 different genes that were enriched in N2A cells based on Affymetrix microarrays were targeted with multiple small interfering RNAs (siRNAs) before the knockdown of the then-known Fam38a, now called “Piezo1” (after the Greek word for “pressure”), led to significant decreases in MA currents (Coste and others 2010; Satoh and others 2006). Piezo2 (Fam38b) was identified as Piezo1’s vertebrate homolog. Functional analyses, including siRNA knockdown in dorsal root ganglion (DRG) neurons, showed that its absence led to a depletion of fast inactivating currents (Coste and others 2010). Cloning of mouse genes revealed approximately 2500 and 2800 amino acids (for PIEZO1 and PIEZO2, respectively) with 26 to 38 resolved transmembrane domains per protomer in PIEZO1 and 38 transmembrane domains in PIEZO2 (Coste and others 2010; Saotome and others 2018; Wang and others 2019; Zhao and others 2018).

PIEZO1 and PIEZO2

PIEZO1 and PIEZO2 are both nonselective cation MS channels, are expressed abundantly in diverse tissue types, and share some functional redundancy. PIEZO1 is expressed highly in the lung, bladder, and skin and less so in the stomach, kidney, and colon. Meanwhile, high levels of PIEZO2 have been found in DRGs and sensory neurons, among others. PIEZO2 was found to be necessary for rapidly adapting MA currents in DRG neurons, and overexpression of both PIEZO1 and PIEZO2 led to a 17- to 300-fold increase in MA currents. Despite being variously expressed in different cell types, PIEZO1 and PIEZO2 functions are not mutually exclusive, and both can work synergistically as observed in cartilage cells in diarthrodial joints where there is high-strain mechanical stress. Directed expression of PIEZO1 and PIEZO2 in articular chondrocytes sustained potentiated mechanically induced Ca²⁺ signals and electrical currents compared to single expression of each PIEZO protein (Lee and others 2014).

Inhibitors and Agonists

Mechanically activated currents generated by Piezo channels can be blocked by both chemical (ruthenium red and Gd³⁺) and a peptide inhibitor (GsMTx). Ruthenium red (RR) blocks inward but not outward currents by entering the channel through the extracellular side, while all enantiomeric forms of GsMTx block stretch channels by positioning itself at the protein-lipid interface to block channel activation (Bae and others 2011; Gnanasambandam and others 2017a; Suchyna and others 2000, 2004a). Proteins can also diminish PIEZO1-mediated MA currents as seen with the expression of Polycystine-2, which leads to reduced currents in renal proximal convoluted tubules (Peyronnet and others 2013). A high-throughput screen of over a million small chemical compounds was carried out to identify potential agonists for PIEZO channels and yielded one candidate, Yoda1, now known as a specific PIEZO1 agonist with no effect on PIEZO2 activation (Syeda and others 2015). A separate small-molecule screen yielded two additional PIEZO1 specific agonists, Jedi1 and Jedi2, that triggered PIEZO1-mediated currents. Jedi1/2 are effective agonists for both mouse and human PIEZO1 (Wang and others 2018).

Cellular Localization

PIEZO proteins are primarily membrane bound, and their localization was initially studied with PIEZO antibodies. Recent work within red blood cells (RBCs) shows that PIEZO1 can exist in submicrometric clusters under native conditions and during PIEZO1 activation with Yoda1, where PIEZO1 clusters increase in areas of higher membrane tension and lower membrane curvature. Dumitru and others (2021) also found evidence that these PIEZO1 clusters might be held in place by anchoring to the cytoskeleton network, thus providing more evidence of the interplay between the force-from-lipids and the dual-tether model. Furthermore, in gliomas, PIEZO1 localizes to focal adhesion sites to activate FAK-integrin signaling, maintain tissue stiffness, and promote tumor cell proliferation (Chen and others 2018).

Structure

To date, the sequences of PIEZO1 and PIEZO2 do not resemble that of other ion channels and share approximately 42% sequence homology. The mechanosensitivity of PIEZO1 is better understood than PIEZO2 (Coste and others 2010). PIEZO cation channels are homotrimeric, and pore permeation in both PIEZO1 and PIEZO2 is possibly regulated by an intrinsically disordered region located adjacent to the beam and functions as a plug (Taberner and others 2019). Work since its initial characterization indicates that the pore function of mouse PIEZO1 (mPIEZO1) is evolutionarily conserved and directly gated by membrane stretch. However, it is now accepted that charged residues positioned between the beam and the intracellular C-terminal domain (CTD) are required for normal mechanosensitivity in PIEZO2, and a transmembrane gate may be controlled by the cap domain (Wang and others 2019).

Recent high-resolution structures of mouse PIEZO1 indicate a unique set of 38 transmembrane (TM) domains per subunit, with domains 37 and 38 forming the inner helix (IH) and outer helix (OH) that enclose the pore region of the channel, while the remaining 36 TM domains form blade-like structures (Fig. 1A,C) (Zhao and others 2018). A topological prediction model by Kamajaya and colleagues found that the central cap consists of the C-terminal extracellular domain (CED) at residues 2210 to 2457, and deletion within this region (Δ2219–2453) resulted in the absence of the central cap (Ge and others 2015; Kamajaya and others 2014). Furthermore, the CED exists in a trimeric complex that contains the extracellular vestibule (EV) with openings (Ge and others 2015; Zhao and others 2018). Together, the IH-CED-OH of one subunit is turned clockwise into the neighboring unit by an anchor region (Fig. 1B). This anchor region is a hairpin structure that maintains the integrity of the ion-conducting pore, and mutations in the anchor region have been implicated in disease (Andolfo and others 2013; Zhao and others 2018). The importance of the anchor region is further demonstrated by the observation that binding of the anchor-OH region by SERCA (now known as a PIEZO1 binding protein) leads to decreased PIEZO1 activity (Zhang and others 2017). Meanwhile, the ion-conducting pore contains the last two transmembrane domains and consists of three parts: the EV within the cap, a transmembrane vestibule (MV) within the membrane, and an intracellular vestibule (IV) underneath the membrane (Fig. 1C) (Jiang and others 2021b). The efficiency and the selectivity for cations of the ion-conducting pore region is determined by a stretch of negatively charged residues at the opening of the central cap. Furthermore, PIEZO1’s selectivity for calcium ions, pore blockages, and unitary conductance might also be mediated by two important acidic residues, E2495 and E2496 (Zhao and others 2018).

Figure 1.

PIEZO1 structure and opening. (A–C) PIEZO1 is composed of three repeating units (demarcated in red, blue, green), which make up the extracellular, transmembrane, and intracellular vestibules (Jiang and others 2021b). (A) PIEZO1’s cap remains closed until the blades respond to outside stimuli (Chesler and Szczot 2018), causing (B) the twisting and (C) subsequent opening of the pore, allowing for the passage of cations, including calcium, through the cell membrane. CED = C-terminal extracellular domain; EV = extracellular vestibule; IH = inner helix; IV = intracellular vestibule; MV = transmembrane vestibule; OH = outer helix; PH = peripheral helices. The illustration was generated using Biorender.com.

While the structure of the pore region is crucial in regulating ion conductance and selectivity, the homotrimer propeller structure (Fig. 1A), perhaps one of the more recognized PIEZO-distinct architectures, confers mechanotransduction into the pore region and induces membrane curvature (Guo and MacKinnon 2017; Lin and others 2019; Saotome and others 2018; Zhao and others 2018). Highly conserved across species, each blade of the propeller structure contains nine transmembrane helical units. When viewed from the top, each “blade” twists clockwise and connects to the pore via an intracellular beam that behaves like a lever. The beam couples the distal blade to the central hydrophobic pore and amplifies the force input from the external blades (Fig. 1A–C). The mechanical force detected by the blade region is passed through the lever to the ion-conducting pore, but conformational changes at the distal blades lead to only openings small enough to allow for cation permeation (Zhao and others 2018). Interestingly, both the beam and blade structure were shown to also curve the membrane plane in which the PIEZO1 ion channels reside, thus aligning with previous work that demonstrates PIEZO1 channels to be gated by lateral membrane tension and changes in membrane curvature (Cox and others 2016; Lewis and Grandl 2015; Syeda and others 2016; Zhao and others 2018).

Kinetics: Activation

PIEZO1 channels open directly by lipid bilayer membrane tension and tension caused by tethered linkages to the extracellular matrix or the cytoskeleton (Coste and others 2010; Cox and others 2016; Lewis and Grandl 2015; Syeda and others 2016). Recent studies indicate that PIEZO1 channels can locally deform the lipid membrane into dome-like shapes, and based on calculated free energy, the projected area of PIEZO1 in closed to open states is also important for the mechanosensitivity of PIEZO1 (Guo and MacKinnon 2017; Lin and others 2019; Wang and others 2019). The transmembrane helices of PIEZO1 do not lie within the membrane plane, but instead, the closed conformation likely curves the lipid membrane into a spherical dome that projects into the cell. This dome serves as a reservoir of potential energy for MS gating, and as PIEZO1 flattens along with a flattening membrane, the “dome” proportionally flattens in parallel with PIEZO1. The curved membrane leaves a “membrane footprint” of deformed lipid molecules outside the perimeter of the PIEZO1 dome and contributes to the increased sensitivity of PIEZO1 to changes in membrane tension. PIEZO1’s membrane foot printing makes PIEZO1 exceptionally sensitive in low membrane tension areas, rendering PIEZO1 ready to respond to even the smallest changes in membrane tension (Haselwandter and MacKinnon 2018). Furthermore, high-speed atomic force microscopy (HS-AFM) was used to indent and apply force to the membrane that resulted in channel activation. This implies that if tethers can act on the membrane and PIEZO1 channels in similar manners, then PIEZO1 can be gated by similarly directed forces (Guo and MacKinnon 2017; Lin and others 2019).

Kinetics: Inactivation

PIEZO channel kinetics are regulated by temperature, alternative splicing, osmotic swelling, membrane lipid composition, pH, coexpression of other membrane proteins, and G-protein coupled pathways. PIEZO1 can rapidly lose current with time, and its activation is followed by a rapid inactivation despite prolonged exposure to mechanical stimuli (Gnanasambandam and others 2017b; Honore and others 2006). Meanwhile, PIEZO2 showed faster inactivation kinetics than PIEZO1. In addition, PIEZO1 has faster inactivation at negative membrane potentials and slower inactivation at more positive membrane potentials (Coste and others 2010). Current understanding of regulators and protein structures essential to the inactivation process grows through both structural studies and studies of diseases involving mutations of PIEZO channels (Albuisson and others 2013; Murthy and others 2017; Zarychanski and others 2012). For example, in dehydrated hereditary xerocytosis (DHS), a cluster of mutations around the central pore of PIEZO channels slows PIEZO inactivation, implying that the IH, OH, external cap, and CTD domains regulate inactivation (Andolfo and others 2013). Furthermore, three small subdomains in the extracellular cap and potentially two inactivation gates in the IH and CTD region appear to be essential in regulating inactivation, thus providing additional evidence for the importance of the central pore region in determining PIEZO channel inactivation kinetics (Lewis and Grandl 2020). Del Marmol and others (2018) found that PIEZO1 inactivation kinetics is slow in mouse embryonic stem (mES) cells but becomes smaller and faster inactivating during differentiation. Heterologous expression of mPiezo1 cDNA from mES cells resulted in faster inactivating MA currents, whereas endogenous expression of PIEZO1 showed an otherwise slower inactivation, implying the involvement of additional regulatory mechanisms other than amino acid sequence (Del Marmol and others 2018).

Piezo Functions

Both Piezo1’s and Piezo2’s expression has been demonstrated in nearly every eukaryotic model organism, except in a select few such as yeasts, and nearly every organ in said organisms (Gillespie and Walker 2001; Xiao and Xu 2010). Here, we will focus on and discuss several functions of PIEZO1 (Fig. 2). PIEZO1’s structure lends itself extremely well to the external activation of different systems through Ca²⁺ (and other cations) influx initiated by mechanical force. Well studied in mouse and human models, PIEZO1 is essential in multiple aspects of the cardiovascular system: through chemical and mechanical activation, PIEZO1 promotes angiogenesis in vitro and in vivo (Kang and others 2019), determines developmental and adult vascular structure in vivo (Li and others 2014), and regulates RBC volume in response to hydration in vitro and in vivo (Cahalan and others 2015). In the pancreas, PIEZO1 responds to mechanical stimuli to induce pancreatitis in vivo (Romac and others 2018), while in the musculoskeletal system, PIEZO1 promotes myotubule formation in vitro and in vivo (Ducret and others 2010; Tsuchiya and others 2018) and osteoblast differentiation in vitro and in vivo (Kao and others 2019; Sun and others 2019), exemplifying diverse functions necessary for either disease or survival. In the excretory system, PIEZO1 holds critical roles in maintaining urine osmolarity in vitro and in vivo (Martins and others 2016), mediating urination in general in vitro (Miyamoto and others 2014; Wei and others 2019), and is necessary for peristalsis in the gut in vitro and in vivo (Sugisawa and others 2020). Further highlighted later in this review, PIEZO1’s involvement in the nervous system is wide-ranged and includes neuroregeneration (Fig. 5 and Fig. 6) in vitro and in vivo (Li and others 2021; Song and others 2019), glioma aggression through a positive feedback mechanism in vitro and in vivo (Chen and others 2018), and neural stem cell differentiation through myosin II mechanical activation in vitro (Pathak and others 2014). As demonstrated, PIEZO1’s ability to be activated by a span of physical forces idealizes it for the indiscriminate meditation of a large domain of systems.

Figure 2.

Functions of PIEZO1 in different cells and tissues. PIEZO1 has been well implicated in many systems, such as the cardiovascular system, including vascular development in mouse (Kang and others 2019; Li and others 2014) and red blood cell volume regulation in mouse (Cahalan and others 2015); the pancreas in mouse (Romac and others 2018); musculoskeletal system, including myotubule in rats and in vitro (Ducret and others 2010; Tsuchiya and others 2018) and osteogenesis in mouse (Kao and others 2019; Sun and others 2019); renal system in mouse and in vitro (Martins and others 2016; Miyamoto and others 2014; Wei and others 2019); gut in mouse (Sugisawa and others 2020); and nervous system, including neuroregeneration in fly and mouse (Li and others 2021; Song and others 2019), glioma aggression in mouse (Chen and others 2018), and neuroblast differentiation in vitro (Pathak and others 2014). Expanded upon in the bottom leftmost and rightmost panels are two of PIEZO1’s distinct roles in the nervous system: strengthening gliomas through positive feedback tissue stiffening (left) and myosin II’s activation of PIEZO1, resulting in neural stem cell differentiation through Yap/Taz signaling (right). In addition, the organism in which the study was performed is displayed in gray alongside the tissue in which the study focused on. The illustration was generated using Biorender.com.

Figure 5.

Summary of regeneration phenotypes in the Piezo-Atr pathway. (A) Brief summary of the pathway from injury to Piezo activation to downstream regeneration regulation. (B) Drosophila class III dendritic arborization (da) neuron regeneration phenotypes are illustrated with varying genotypes in the Piezo-Atr pathway. The upper and lower boxes show the neurons at 24 and 72 hours after injury via two-photon axotomy. Boxes highlighted in green show regeneration, while those in red fail to regenerate. The site of injury is surrounded by a dotted red circle. Regenerating axons are shown in green and marked by arrowheads, while stagnant axons are identified by an arrow.

Figure 6.

Piezo1’s role in inhibiting axon regeneration through the Atr-Chek1 pathway. (A) Uninjured neurons display baseline Piezo1 expression (Song and others 2019). (B) Upon axonal ablation, Piezo1 channels accumulate around the growth cone and open, allowing calcium ion influx (right). Calcium ions bind to calmodulin-dependent protein kinase II (CaMKII) and subsequently activate nitric oxide synthase (NOS) (Mittal and Jadhav 1994; Weaver and others 2002), which causes the nuclear activation of Atr (left). The Atr protein complex is assembled/activated, leading to Chek1 phosphorylation and activation and subsequent axon regeneration inhibition. A glia-neuron interaction that alters the extracellular matrix is hypothesized (Li and others 2021). The illustration was generated using Biorender.com.

Piezo in Neurodevelopment and Regeneration

Mechanical Force in Axon Growth

During development, animals can undergo enormous growth, suggesting that the nervous system must also expand to accommodate the growing animal. Integrated neurons no longer have pronounced growth cones and must undergo rapid and sustained growth, likely mediated by mechanotransduction, specifically mechanical tension caused by axon stretching (Pfister and others 2004; Smith 2009). The relationship between tension and neurite outgrowth was explored by attaching needles to the growth cone of single axons and towing the axons to study the capacity of neurons to stretch and grow without breaking (Fig. 3A) (Bray 1984; Chada and others 1997; Dennerll and others 1989; Heidemann and others 1995; Zheng and others 1991). Axons have two distinct steps in growth: axons reach their targets by being pulled slowly by growth cones, followed by the stretching of integrated axons over an extended period. Tension created by neurite pulling was thus found to be a key regulator for successful axon elongation (Bray 1979, 1984; Heidemann and others 1990; Lamoureux and others 1989; Weiss 1941). “Strain” and “acclimation” define the boundaries of the long-term stretching period. At the start of elongation, strain is applied to short integrated neurons that are vulnerable to breakage. To prevent axons from breaking, strain is applied at displacement steps to provide neurons the necessary time to acclimate and undergo cellular conditioning needed to accommodate axon growth without sacrificing axon volume (Pfister and others 2004). Pfister and others (2006a) grew DRG neurons in an axon stretching system where two populations of integrated neurons were pulled apart to elongate the connected axons (Fig. 3B). It was observed that neurons eventually acclimated to stretching as cellular conditioning processes adapted to accommodate axon elongation. The elongated axons maintained their thickness and propagated action potentials, suggesting that stretch activated protein synthesis and transport pathways (Pfister and others 2006a). However, if the stretch rate exceeded the cell’s ability to compensate, tension accumulated and resulted in axon breakage (Pfister and others 2004).

Figure 3.

Axons can be mechanically stretched. (A) Individual growth cones of axons are towed by glass microelectrodes. The microelectrode is moving from left to right at 100 μm/hour or less and is towing the axon to at least 100 μm in length. Towing at these rates results in elongated neurites with a normal or slightly increased diameter (Bray 1984). (B) Dorsal root ganglion (DRG) neurons are cultured in an axon stretch-growth system with an axon stretching frame. Two populations of neurons are attached to two adjoining substrates and allowed time to connect with neurons on the opposite side. The stretching frame moves one population of cells away from the other and the axons undergo a period of stretch before acclimating into the elongation process during which axons elongate while maintaining their normal diameters without any thinning or breakage. The stretching frame moves at a rate of 10 cm/day and stretches the integrated axons up to a total of 10 cm (Pfister and others 2006b). The illustration was generated using Biorender.com.

Therapeutically, understanding the mechanisms that underlie the extreme stretching of neurons in culture can be used to repair severe nerve damage, and long nerve tracts might be used as transplant material to treat damage in the nervous systems. Our knowledge of the molecular and cellular machinery underlying mechanosensation during neuronal outgrowth remains limited. In vitro studies have suggested an inhibitory role of MS ion channels in neuronal outgrowth because inhibiting MS channels pharmacologically with GsMTx4 promoted neurite outgrowth in PC12 cells in the presence of NGF (Gottlieb and others 2010; Jacques-Fricke and others 2006). In addition, MS channels mediate transient calcium influx, in part through TRPC1 in Xenopus growth cones to slow neurite extension and regulate growth cone turning on rigid surfaces (Kerstein and others 2013).

Piezo Regulates Axon Growth and Regrowth

Through single-molecule Forster resonance energy transfer (smFRET) and total internal reflection fluorescent microscopy (TIRFM), Piezo channels were seen widely distributed and mobile throughout the plasma membrane of human neural stem/progenitor cells (hNSPCs) with concentrated Ca²⁺ “flickers” in areas of higher force-producing adhesions and traction forces. These cellular traction forces trigger Piezo activation, resulting in downstream biochemical signals that are crucial in regulating development and neural injury (Ellefsen and others 2019).

When MS ion channels were blocked by GsMTx4 in Xenopus laevis retinal ganglion cells (RGCs) grown on stiff surfaces, RGC axons were significantly shorter than the controlled group containing fully functioning Piezo1 channels (Koser and others 2016). Inhibiting Piezo1 function with GsMTx4 or downregulating Piezo1 expression via a morpholino knockdown in vivo resulted in RGC axons dispersing widely from their normal pathfinding trajectory with reduced directionality and fasciculation (Koser and others 2016). According to Koser and colleagues, this resembles the behavior of RGC axons grown in “soft” environments, because removing the channel’s activity impairs the cell’s ability to detect “stiff” and therefore results in a “soft”-response phenotype. Here, Koser and colleagues defined “soft” and “stiff” in their system by synthesizing polyacrylamide (PAA) hydrogels of differing stiffnesses of shear moduli, 1.0 kPa (“stiff”) and 0.1 kPa (“soft”), to represent the upper and lower bounds of brain tissue stiffnesses. Furthermore, RGC axon bundles display correct pathfinding by turning toward softer surfaces, but this directionality and neurite growth are diminished during treatments with GsMTx4 and Piezo1 morpholinos, suggesting that Piezo1 channels are essential in promoting axon growth and normal axon pathfinding in vivo and in vitro (Fig. 4A) (Koser and others 2016). In a separate study by Li and colleagues, PIEZO1 was found to inhibit axon growth in adult mouse DRG neurons. Piezo1 was conditionally knocked out (cKO) in adult mouse DRG neurons. Mutant and control mouse DRG neurons were cultured on PAA gels of varying stiffnesses (shear moduli: 0.1, 0.3, 1.0, 5.0, and 30.0 kPa), and the effect of PIEZO1 on total neurite outgrowth was assessed (Li and others 2021). Interestingly, in the absence of PIEZO1, DRG neurons had greater total neurite lengths when grown on “intermediate” and “soft” stiffnesses (defined by Li and colleagues as 1.0 and 0.3 kPa, respectively) (Fig. 4B). Meanwhile, there was no difference in total neurite lengths in DRG neurons grown on higher stiffnesses, suggesting that a certain range of stiffnesses provides the adequate mechanical stimulus needed for PIEZO1 activation. Thus, it is likely that within a specific range of environmental stiffnesses, PIEZO1 activates and plays a role in limiting neurite outgrowth. However, at stiffnesses outside this range (such as at 30 kPa), PIEZO1 may not be activated, or additional factors might become dominant and negate PIEZO1’s inhibitory role. Although studies by the two groups show seemingly opposite effects of Piezo on neurite outgrowth, these results highlight the complexities of axon growth and pathfinding in different organisms and tissue types as well as the importance of Piezo1 and mechanotransduction in neurite behavior.

Figure 4.

Mechanosensation mediates axon pathfinding and outgrowth. (A) Schematic of axon bundles navigating through a substrate with a gradient of stiffnesses. Axons grown on stiffer surfaces travel faster and straighter than on softer substrates. Upon encountering softer surfaces, the axons fasciculate and are pulled toward the softer substrate (Koser and others 2016). (B) Control and Piezo1 conditionally knocked out (cKO) dorsal root ganglion (DRG) neurons are plated on polyacrylamide (PAA) gels of “soft,” “intermediately stiff,” and “extremely stiff” surfaces. Piezo1 cKO neurons grown on “soft” (not shown) and “intermediately stiff” surfaces have greater total neurite lengths than control neurons, while there is no difference at “extremely stiff” surfaces, suggesting that PIEZO1 mediates neurite outgrowth within a range of stiffnesses (Li and others 2021). The illustration was generated using Biorender.com.

As mentioned earlier, the stiffness of the substrate is largely determined by the ECM characteristics, which are heavily influenced by the composition and densities of the structural ECM molecules, such as laminins, collagens, and chondroitin sulfate proteoglycans (CSPGs) in the nervous system in mammals and flies. The tissue environment can also affect stiffness, and cellular constituents and density may lead to a specific stiffness within a certain area. Piezo1 is a mechanosensor that detects and responds to the environmental stiffness (Koser and others 2016). On the other hand, mechanical stimulation of Piezo1 also depends on the ECM proteins (Gaub and Muller 2017), adding another potential layer of its regulation during neuroregeneration.

When studying in vivo axon regeneration in Drosophila melanogaster and in rat hippocampal neurons (in culture), Piezo takes on an inhibitory role (Fig. 5) (Li and others 2021; Song and others 2019). Using a two-photon injury system, Drosophila class III dendritic arborization (da) neurons were axotomized and allowed to regenerate over 72 hours. In fly larvae lacking Piezo, axons regenerated significantly more than neurons of their wild-type counterparts. In addition, GCaMP6f imaging showed increased local Ca²⁺ influxes at the axon tip in wild-type compared to Piezo-null flies during the axon regeneration period. This influx precedes obvious formations of filopodia in growth cones, implying that Piezo functions locally to diminish regeneration in Drosophila. Rat hippocampal neurons were next grown in microfluidic chambers that spatially separated the axon and cell body compartments, allowing for the isolated study of biological events at either just the soma or the axon tips. In the microfluidics devices, only the axons of rat hippocampal neurons were injured via shear stress, and the injured axon “stumps” were locally treated with Yoda1, the PIEZO1-speicific agonist. Chambers that were treated with Yoda1 regenerated less than neurons grown in chambers treated with only DMSO. Song and others’ (2019) work in Drosophila and rat hippocampal neurons indicate that Piezo plays an inhibitory role in axon regrowth following injury.

Pathway Downstream of Piezo

It was demonstrated that upon Piezo activation, calcium/calmodulin-dependent protein kinase II (CaMKII) and nitric oxide synthase (NOS) are activated (Fig. 6) (Song and others 2019). Further downstream, ataxia telangiectasia and Rad3-related (Atr) and checkpoint kinase-1 (Chek1) pathway (traditionally known to mediate the single-strand DNA damage response) noncanonically inhibit axon regrowth (Li and others 2021). It is hypothesized that that during axon regrowth, the growth cone dynamically probes the environment, for example, by interacting with glial processes, leading to Piezo activation at the growth cone tip. Piezo channel opening leads to local calcium influx and the activation of nitric oxide synthase (Nos), the producer of NO. NO acts as a second messenger that propagates to the nucleus, where it activates Atr and the associated complex, including Atrip (Atr interacting protein), the 9-1-1 complex (Rad9-Hus1-Rad1), Rad17, and Topbp1 (DNA topoisomerase II binding protein 1). Atr phosphorylates and activates Chek1, which phosphorylates and inactivates Cdc25 (cell division cycle 25), inhibiting its ability to dephosphorylate and activate Cdk1 (cyclin-dependent kinase 1). The phosphorylated and inactive Cdk1 impinges on downstream effectors, causing regeneration arrest (Fig. 6) (Li and others 2021). Therefore, the Piezo-Nos-Atr-Chek1 axis enables regulation of axon regeneration through mechanosensation. To summarize, the role of Piezo channels in axon growth and regrowth warrants further investigation, and it is likely that Piezo channels have distinct roles depending on the type of neuron and organism in which Piezo is localized, as well as the organism’s differing development stages.

Conclusion

Mechanosensation is pivotal in our daily lives and mediates how we sense and interact with the world. At the cellular level, mechanotransduction allows the cell to react to its physical environment by translating physical cues into biological signals, leading to diverse functions. Research over the past several decades has unearthed numerous MS ion channels shown to be gated directly by mechanical tension within the membrane itself. Recent work has characterized the structure and function Piezo channels, now known to be important in pain, touch, and, excitingly, neuroregeneration. Despite our remarkable progress in understanding mechanotransduction and neuroregeneration, many outstanding questions remain. Importantly, we still do not fully understand the pathophysiological processes involving mechanotransduction. Research that maps out mechanical forces and determines how Piezo channels are gated during axon regeneration is much warranted. Investigating how MS ion channels regulate mechanotransduction in different cell types will also provide crucial insights into how mechanical forces orchestrate cellular architecture and signaling within specific types of cells. Last, it is essential to determine how we can leverage our current understanding of the structure and function of ion channels to help develop novel therapeutics for diseases by targeting MS ion channels such as Piezo.