Mechanical systems biology of C. elegans touch sensation

Abstract

The sense of touch informs us of the physical properties of our surroundings and is a critical aspect of communication. Before touches are perceived, mechanical signals are transmitted quickly and reliably from the skin’s surface to mechano-electrical transduction channels embedded within specialized sensory neurons. We are just beginning to understand how soft tissues participate in force transmission and how they are deformed. Here, we review empirical and theoretical studies of single molecules and molecular ensembles thought to be involved in mechanotransmission and apply the concepts emerging from this work to the sense of touch. We focus on the nematode Caenorhabditis elegans as a well-studied model for touch sensation in which mechanics can be studied on the molecular, cellular, and systems level. Finally, we conclude that force transmission is an emergent property of macromolecular cellular structures that mutually stabilize one another.

Introduction

The way we experience our environment and each other is deeply influenced by the body’s ability to detect and respond to mechanical stimuli. The sense of touch provides immediate and intuitive access to physical properties of objects and our bodies such as density, texture, and shape [1, 2]. The perception of sound underlies language and allows us to enjoy many facets of music, from the deep grooves of house music to the high pitches of a soprano’s aria. Less obviously, we also rely on mechanical cues for every beat of our heart [3, 4].

This flow of mechanical signals is influenced by the physical properties of the intervening materials. Just as the seismic waves of an earthquake travel faster on land than in water, mechanical energy travels faster through bone than through interstitial fluid. The process that links touching to feeling is no different: forces applied to the skin surface are transmitted through millimeters of tissue before reaching mechano-electrical transduction (MeT) channels that convert mechanical signals into electrical ones. Molecules [5], cells [6], tissues [7], and whole animals [8] all deform in response to externally applied forces. Importantly, any living structure that deforms under force could, in principle, be mechanosensitive (MS) [9]. The extent and dynamics of the deformation depends on constitutive material properties, such as elasticity and isometric tension. In general, stiff structures deform less than soft ones subjected to the same force, and tense structures propagate mechanical stimuli further than relaxed ones.

Here, we review concepts of force propagation along cytoskeletal filaments and suggest a framework for understanding how mechanical loads applied to the skin might be transferred to MeT channels that decorate mechanoreceptor neurons. The approach we propose combines our understanding of the biophysical mechanisms of force transmission within and between living cells and our knowledge of the physics and physiology of touch sensation in the nematode C. elegans. Both arenas have been covered separately in several excellent reviews [9–13]. Here, we bring them together to develop an understanding of how the mechanical loads delivered in a touch result in neural responses.

Cytoskeleton mechanics affect mechanical signal transmission

The actin cytoskeleton experiences mechanical tension [14] generated by myosin contraction [6, 15], which is counterbalanced by structures that include microtubules (MTs) [16, 17], anchoring to the extracellular matrix (ECM) [11] or the osmotic pressure of the cytoplasm [18]. This preexisting mechanical tension has been proposed to help convey mechanical signals over long distances [13]. Examples include force transfer from the membrane to the nucleus [19], which elicits changes in gene expression and/or nucleolar organization [20, 21], and src kinase activation at cellular sites distant from the location of the applied force [22, 23], which initiates phosphorylation of kinase targets.

Mechanical pre-stress in the actin cytoskeleton plays a central role in transmitting force between physically distant parts of the cell [21, 22, 24–26]. Similar to the string in a tin can telephone, a cytoskeletal element under tension transmits mechanical deformation faster and further than a relaxed one [27]. Put differently, if the string is completely slack, then no mechanical energy can be transported along its length. In support of this idea, experimental manipulations that decrease actin tension or destroy actin stress fibers impair force propagation in cells [22, 25].

Theoretical modeling of cellular force propagation along cytoskeletal filaments has suggested that the bending rigidity, visco-elasticity, and pre-stress of the fiber as well as cytosolic viscous damping influence force transmission [12–14, 24–28]. For instance, when force is applied transversely (perpendicular to the direction of the fiber) [25, 27, 28], the fiber bends and is slightly stretched (Fig. 1A). The bending mode and resulting deformation depends on how the fiber ends are coupled to the boundaries (Fig. 1B). Due to the low flexural rigidity of both actin and MTs, bending has a minor contribution to the restoring force after deformation. As a result, the extent and rate of the propagation of this perturbation increases dramatically with pre-stress [27, 28]. In contrast, when a stimulus is applied longitudinally (along the fiber), the propagation speed is independent of the tension, but depends instead on the filament’s elastic modulus [27, 28]. The distinction between these different modes of force application may have important biological consequences. For example, during touch sensation in C. elegans, forces are applied transverse to the long (anterior-posterior) axis of sensory neurites and such stimuli are known to activate MeT channels in vivo [29]. Consequently, pre-stress in the neurite might be necessary to facility fast long-range forces transmission to induce a behavioral response. It is conceivable, however, that biological materials do not experience purely transverse or axial deformations during mechanical stimulations, suggesting that both elasticity and tension can greatly enhance the spread of a mechanical signal through cells or tissues.

Figure 1.

Force transmission along cellular filaments. A: A slender rod, e.g., a hypothetical cytoskeletal fiber, with fixed (upper schematic) and hinged ends that are free to pivot (lower schematic) is subjected to a transverse point load (green arrow). Red arrows indicate the restoring force due to pre-stress acting on the deformed fiber (direction of signal propagation). Black arrows indicate the restoring force due to bending elasticity. The two different boundary conditions determine the relative contributions of pre-stress and bending modulus to the force transmission properties of the fiber. B: Cytoskeletal filaments are subjected to a transverse deformation in cultured cardiac myocytes. A thin glass pipette in the middle of the cell is moved up and down while imaging cytoskeleton-associated mitochondria. Reproduced from Dyachenko et al. [118]. C: Classical Euler buckling of an axially compressed rod. The ends are fixed in the upper image and free to pivot below. The energy of deformation is dominated by the bending rigidity of the rod. D: A single microtubule in a dumbbell configuration with hinged ends of an optical trap assay subjected to axial compression. The initially straight filament buckles into a single arc. Reproduced from Kurachi, et al. [119]. Scale bar: 10μm. E: Attenuated short wavelength buckling of an axially compressed rod embedded in an elastic material. The deformation of the rod now depends on the mechanical interaction with its surrounding. An energetic competition between matrix deformation and bending of the rod drives local buckling behavior. Importantly, due to the low flexural rigidity of the rod, oscillatory buckling minimizes the energetic cost of matrix deformation. The extent of buckling, however, not only depends on the presence of the matrix but also how the rod adheres to it. Blue: undeformed matrix; red: extension; green: compression deformation. F: A single microtubule within the cytoplasm of a Cos7 cell is subjected to axial forces by a micropipette. Local, attenuated short wavelength buckling is visible. Reproduced from Brangwynne et al. [16]. Scale bar: 10μm.

Whereas actin fibers and networks are the predominant cytoskeletal structures that provide shape and rigidity, other components, notably MTs, endow specific cell types with more specialized mechanical properties. MTs resist axial loads and experience pure compression up to a critical limit [16, 30]. This limit depends on MT length, L, and its flexural rigidity (or bending modulus), k_MT. Beyond this limit, the MT will buckle into an arc with a characteristic wavelength λ=2L, because the total deformation energy is smaller if the rod is bent. This behavior arises as a consequence of the high-compression and low-bending modulus of the MT, which makes it more favorable to reduce end-to-end distance by buckling than by pure axial compression (Fig. 1C,D). For reasonable values of κ and L [31, 32], the critical force threshold is about 1 pN – less than the 4–6 pN force exerted by a single kinesin motor protein [33–35]. This begs the question—how are MTs protected from buckling in cells? They are embedded in a viscoelastic cytoplasm that includes high-molecular weight actin and intermediate filament (IF) cytoskeletons and constrains lateral movement. This is analogous to a composite material, like rebar embedded in concrete. Under these circumstances, MTs can withstand axial forces up to 100 pN before collapsing [16]. Moreover, the deformation mode changes and the buckling wavelength is reduced to $\lambda =2\pi {\left(\frac{\kappa }{G}\right)}^{\frac{1}{4}}$ ([16, 30] and Fig. 1E,F), where κ is the MT bending modulus and G is the elasticity of the surrounding matrix, because the deformation of the matrix is energetically less favored than the bending of the MT. Under these conditions, the wavelength is independent of MT length and inversely proportional to G. As a result, measurements of buckling wavelength can be used to derive G (if κ is known or can be estimated) [16,36–38]. In addition to the elasticity of the surrounding matrix, friction between the MT and the matrix also affects the buckling mode. Specifically, such frictional, viscous forces cause the buckling amplitude to decrease exponentially with distance from a site of force application (Fig. 1F) [16, 39, 40].

Cytoskeletal fibers rarely work alone: in specialized sensory and nerve cells, MTs are frequently assembled into cross-linked bundles. Crosslinking individual MTs into a coherent bundle increases the resistance to bending deformation and persistence length of the bundle as compared to an isolated MT [41]. Thus, MT bundles might be an important aspect of mechanical signal transmission. In the context of an MT bundle, it is important to consider the material properties of individual MTs as well as those of the bundled super-structure. The flexural rigidity of the bundle (k_B.) increases with filament number N according to k_B.= N · k_MT. for a loosely crosslinked and k_B.= N² · k_MT for a completely crosslinked bundle [42]. More complex models that incorporate crosslinker dynamics and compliance have been reported [43]. As with single MTs, if an MT bundle is embedded in an elastic matrix, the bundle can withstand higher forces before undergoing buckling instabilities. The shape and spatial dynamics of the buckling MT bundle is intimately coupled to the mechanical properties of the surrounding medium and the bundle itself. In principle, it should be possible to use MT shape as a local probe to read-out the forces acting in cells with high spatio-temporal resolution [36], similar to what has been achieved with carbon nanotubes on elastomeric substrates [38]. Taken together, MTs can in principle transmit compressive forces in living cells, and, due to their well-characterized mechanical properties, can be used as experimental probes to monitor cell mechanics and force propagation [16, 22, 39, 44].

IFs also contribute to mechanics, tissue integrity, and mechanical signal propagation [45–48]. IFs show high tensile strength and protect skin against mechanical damage [49]. However, compared to actin and MTs, IFs have received less attention regarding their function in cell mechanics. Stable assembly of neurofilaments (NFs) is critical to maintain neuron shape, and defects in NFs lead to morphological changes reminiscent of neurodegeneration [50]. Interestingly, all NFs have a domain that inhibits MT polymerization and helps to determine the number of MTs within an axon [51]. In theoretical work, it has been proposed that IFs stabilize MTs and protect them from undergoing buckling instability under axial compression [30]. A direct role of keratin IFs in mechanotransduction was revealed in a study in Xenopus; local tugging forces applied to isolated embryonic cells cause reorganization and reinforcement of the IF network towards the site of force application, and lead to a change in their migratory behavior [52]. Just recently, nuclear lamins have been identified as crucial determinants of cell mechanics and differentiation [48]. Finally, the IF protein IFB-1A both colocalizes with mechanosensory neurons in C. elegans and is required for mechanical integrity of the epidermis during muscle contraction [53], suggesting that it plays a critical role in bidirectional mechanical signal transmission between the cuticle and underlying cells.

Membrane mechanics and force transmission to ion channels

The mechanical properties of lipid bilayers are thought to be crucial for the activation of MS ion channels, including MeT channels that give rise to the senses of touch, proprioception, and hearing. The principles linking the physical properties of membranes to ion channel gating have been discussed in several excellent reviews [9, 54–58]. Some of these physical properties enable membranes to store mechanical energy: resistance to changing the angle between two lipid molecules (bending rigidity); preferred lateral spacing of lipid molecules (in-plane tension); and an adjustable, but well-defined thickness of the bilayer due to the length of the fatty acids chains [59]. When proteins such as ion channels are embedded in membranes, the lipid bilayer is deformed, creating a tension along this interface with a force proportional to the length of the interface between the protein and the bilayer. In principle, such a force could contribute to gating [60].

Among the best-studied examples is the mechanosensitive channel of small conductance (MscS) present in bacteria and plants [56], whose open conformation is stabilized by stresses in the lipid bilayer by a mechanism of hydrophobic mismatch [61, 62]. In this model, increased tension in the cell membrane leads to thinning of the bilayer and hence a mismatch between the hydrophobic core of the fatty acids and the hydrophobic region of the transmembrane helices of the channel. Because this is energetically unfavorable, the helices tilt with respect to the bilayer plane, thinning the channel into an open conformation that allows ions to pass through the channel. However, in an alternative model, tension applied locally to the plasma membrane leads to phase separation of different lipid moieties in the bilayer, which then stabilizes different conformations of an integral membrane protein [54].

Very little is known about how gating happens in eukaryotic MS channels, notably those that mediate the senses of touch and hearing. We note that tension-dependent interactions between MS channels and the surrounding lipid membrane could operate in parallel with additional factors, like cytoskeletal tension [63, 64]. Moreover, the sufficiency of membrane tension does not exclude a role for protein tethers linking such channels to extracellular or intracellular structures. Indeed, it is possible that channels such as TREK and TRAAK, which are known to be activated by increased membrane tension in reduced and reconstituted systems [65], may also depend on protein tethers for mechanical signal transmission in vivo [66].

Mechanical systems govern force transmission during touch sensation in the worm

The process of touch sensation begins with mechanical signal transmission from the skin to MeT channels expressed in touch receptor neurons (TRNs), and culminates in MeT when external mechanical loads activate the MeT channel. To illustrate what is already understood as well as the key open questions, we focus on the roundworm C. elegans and its six TRNs as a model for mechanosensory signaling. Figure 2 shows the general anatomy of the TRNs and their position relative to other tissues. The epidermis is a single epithelial cell layer that secretes the collagenous cuticle, a protective, acellular layer important for the animal’s shape, development, and physiology [67, 68], from its apical surface, and is bounded by a basement membrane along its basal surface. Body wall muscles assemble into quadrants attached to the dorsal and ventral aspects of the body. Also shown schematically are the MeT channels that localize to puncta with an average inter-punctum spacing of about 2 μm [69, 70].

Figure 2.

Body and tissue environment of C. elegans TRNs. Schematic of C. elegans body morphology and anatomy, highlighting mechanical elements thought to dominate body mechanics (see text). Four of the six touch receptors are shown in the schematic drawing of a worm: AVM, ALM, PLM, and PVM. PLM and ALM exists as pairs on the left and right side of the body, of which only the left neuron is shown. A slab through the worm shows the cuticle ridges as circumferential rings and the alae, longitudinal projections on the ventral and dorsal sides. The cuticle is secreted by the epidermis below (pink), which attaches to the muscles (white) and the touch receptor neurons (purple). The hemidesomosomes, which attach the muscles and the TRNs to the hypodermis, are shown as white dots.

C. elegans offers many experimental advantages for studies in basic biology, biophysics, and neurobiology, including a short generation time (2–3 days), a fully sequenced genome, and a fully mapped nervous system consisting of only 302 neurons. Aspects particularly important for our discussion include a transparent body that enables direct visualization of neuron shape and function in living animals, and the ability to deliver precise mechanical stimuli (force, indentation) to both immobilized and freely behaving animals [8, 71]. Because the neurons responsible for converting touch into simple behaviors, and the protein partners forming the MeT channel complex, are known, C. elegans enables us to build a unique perspective on mechanosensation that is integrated from molecules to behavior.

Whole body mechanics influence touch sensation and behavior

One of the first detailed experimental characterizations of C. elegans body mechanics was carried out using self-sensing, piezo-resistive cantilevers [72] to measure the deformation of the whole animal under a given pre-set force [73]. The resulting linear force-deformation relationship was modeled as a cortical shell, consisting of cuticle, muscles, and epidermal tissue under hydrostatic pressure. In these experiments, cuticle puncture and adaptation to hyperosmotic environments had only modest effects on body stiffness, especially when compared to the impact of mutations that altered the composition or collagen crosslinking of the cuticle [73]. These and other measurements support the idea that the cortical shell is a major determinant of C. elegans body stiffness [74, 75].

Body wall muscle tone also modulates body mechanics and touch sensitivity [8, 71]. Hyper-contraction of body wall muscles leads to stiffer worms and a reduced sensitivity of the worm to external forces. This observation is consistent with the idea that C. elegans body mechanics modulates the efficiency of force transfer to sensory neurons and therefore is critical to set optimal touch sensitivity [8, 71]. Taken together, these data strongly suggest that body mechanics influence touch sensitivity by modulating mechanical signal transmission in living animals.

Subsequent work linked changes in body stiffness to changes in touch sensitivity, showing that both softer and stiffer animals respond less efficiently than wild-type animals when exposed to the same force [71]. This result implies that worms are sensitive to indentation rather than force per se, and suggests that indentation-induced changes in neuronal strain might account for activation of TRNs. However, it is not yet known whether sensitivity to indentation is a conserved feature of mechanoreceptor neurons across species and contexts. Indeed, a recently published mechanical model of touch in mammals suggests that variations in skin thickness across individuals is more likely to favor behavioral control by force rather than indentation [76]. Additional experiments are needed to address whether worms sense stress or strain and whether or not this represents a conserved feature in metazoan animals.

How do mechanical signals propagate during touch?

In order for a touch to the cuticle to be sensed by the TRNs, the mechanical deformation has to be transmitted through successive tissue layers (Fig. 3, 4A). Though little is currently known about the molecules that comprise this mechanical signal transmission pathway, we can garner insight from investigations of the structures that attach muscles to the cuticle [77], which contain many of the same molecular components (Fig. 4). Among these proteins are LET-805 myotactin [78] (Fig. 4A), the IFB-1A IF [53] (Fig. 4B), and MUA-3 [79] (Fig. 4C), a protein that links the IF cytoskeleton to sites assumed to assist in the transfer of mechanical stress. Myotactin is a large transmembrane protein in the epidermis with 32 type-3 fibronectin repeats and at least four integrin-binding RGD/RLD motifs, that localizes to both muscles and TRNs ([78] and Fig. 4B). It has been suggested that myotactin binds to integrins in muscle cells and ensures their firm connection to the epidermis [78, 80]. C. elegans has two α-integrins encoded by the ina-1 and pat-2 genes, and one β-integrin encoded by pat-3. Intriguingly, reduced expression of the RGD-binding α-integrin PAT-2, but not the laminin-binding α-integrin INA-1, impairs touch sensitivity [81]. A myotactin-integrin interaction may therefore be a crucial part of the force transmission pathway in the TRNs as it is in muscle. Together, these observations suggest that hemidesmosomes transmit force from muscles to the cuticle [77], and, conversely, from the animal’s exterior to TRNs (Fig. 4A). The exact pathway of force transmission to the MeT channel and how it is activated during touch is unknown in C. elegans or in other metazoans. Below, we consider structures that could support mechanical signal transmission in C. elegans TRNs and perhaps other mechanoreceptor neurons.

Figure 3.

The molecular ‘touch’ transmission machine. A: Electron micrograph and schematics of a TRN, embedded into the epidermis, underneath the cuticle. TRNs are densely filled with cross-linked MTs that are connected to the electron-dense cortical cytoskeleton by numerous tethers. The TRN and the epidermis are separated by a specialized extracellular matrix called the mantle, which is necessary for MeT channel trafficking and mechanosensation. Scale bar: 100nm. B: Hypothetical assembly of mechanical components separating the stimulus from the touch response. Touch deforms the cuticle collagens and with it the hypodermal cells. Structural proteins in the epidermal cell are needed both to position the TRNs correctly and to provide mechanical coupling to the cuticle. A specialized ECM surrounding the neuron is required either for mechanical signal propagation and/or channel trafficking. Tension and bending rigidity of the membrane might influence ion channel gating and open probability. MTs and the spectrin cytoskeleton inside TRNs are critical for function on multiple levels. Both cytoskeletal elements likely contribute to TRN mechanical integrity (see text).

Figure 4.

Bi-directional force signaling of fibrous organelles. A: Many molecular components, such as myotactin, intermediate filaments and integrins, are shared in structures that attach muscles to the basal lamina and TRNs to the mantle and epidermis, and ultimately to the cuticle. Muscle contraction produces a force that is transmitted outwards to the cuticle to drive body bending and translocation. Conversely, touch produces a stress in the cuticle, which can be transmitted along the fibrous organelles inwards to the TRNs. In the case of a touch to the anterior, a reversal response is elicited. B: Antibody stain (MH46) of myotactin, an epidermal cell adhesion molecule. C: Antibody stain (MH4) of intermediate filaments. Note that tracks are visible engulfing the TRNs. D: Antibody stain of MUA-3, a hypodermal protein necessary for hypodermis-cuticle attachment. Adapted from Bercher et al. (2001) [79]. Scale bars: 10 μm.

Is force transmitted through the membrane?

Whereas it is well established that forces due to osmotic swelling are transmitted through the membrane to MscS [56], the gating mechanism of eukaryotic MS channel during touch is less clear. Several reviews have discussed the potential effects of membrane mechanical properties on the gating mechanism of eukaryotic ion channels [9, 54, 57, 58, 82].

Piezo1 and Piezo2 are two members of a family of evolutionary conserved ion channels involved in blood flow [4] and the sensation of touch (reviewed in [83]) in mammals, respectively. In addition, the sole Piezo ortholog in Drosophila has been shown to sense noxious stimuli and be activated by mechanical force [84]. C. elegans also has one Piezo ortholog, which has yet to be characterized [85]. Piezo ion channels have been proposed to rely on membrane tension for activation [55]. While Piezo1 is sufficient to induce MS curents in heterologous cells, mechanosensitivity has yet to be detected in Piezo channels reconstituted in pure lipid bilayers [86]. Thus, it remains uncertain whether or not the Piezo1 protein is sufficient to confer sensitivity to bilayer tension [82].

In C. elegans, mutations that prevent the synthesis of poly-unsaturated phospholipids increase the resistance of TRN membranes to deformation and also impair touch sensitivity [87]. These observations provide indirect evidence of a role for membrane mechanics in MeT channels activation. Additional evidence comes from work showing that the stomatin homolog, MEC-2 is a palmitoylated, cholesterol-binding protein and that mutations that reduce or eliminate lipid binding result in decreased touch sensation in vivo. Another stomatin homolog, UNC-24, co-localizes with MEC-2 and has a putative lipid-transfer domain. Thus, it has been proposed that these stomatin-related proteins regulate MeT channel function by modifying the lipid environment in the TRNs (reviewed in [10]). Whether or not the MeT channels in TRNs are directly activated by membrane stretch remains to be determined.

Is force transmitted through the extracellular matrix and basement membrane?

TRNs in C. elegans are surrounded by a specialized basement membrane, called the mantle. Several extracellular proteins have been identified [70, 88, 89] that might link the C. elegans MeT channel to its specialized ECM (reviewed in [10]). At least three proteins thought to localize to this ECM are crucial for touch sensation [70, 89]: MEC-1 and MEC-9 are EGF/Kunitz domain-containing proteins, and MEC-5 is a type IV collagen[70]. It is plausible that this matrix plays an important role in force transmission to the MeT channels, especially given that mutations in the MEC-1, MEC-5, and MEC-9 proteins all result in defects in touch sensitivity. However, all three proteins are also required for proper localization of the MeT channel in the neurite [70], complicating interpretation. Another collagen IV, LET-2, was proposed to be important for mechanical activation of a presumptive stretch-activated ion channel of the DEG/ENaC family expressed in muscles, UNC-105 [90]. However, uncertainty about this model remains [91].

Collagen IV is not the only component of the ECM in C. elegans. Amongst other members of the laminin superfamily, α-laminin EPI-1 is also needed for axon outgrowth and recruitment of fibulin-1C FBN-1 during the assembly of hemidesmosomes [92]. Although FBN-1 is dispensable for the sense of touch in C. elegans [93], it is worth mentioning in light of recent data showing that a tether of yet unknown identity connects murine dorsal root ganglion neurons to a laminin-111 matrix and is important for mechanosensation [94, 95]. Importantly, removal of these tethers by protease treatment abolishes the response to mechanical stimuli [94], suggesting that anchoring of the neurons to the surrounding ECM is required for touch sensitivity [96]. These tantalizing hints suggest that mechanical connections between stretch-sensitive channels and the ECM are likely important for physiological function, at least in some cases. How this connection is established at the molecular level, to our knowledge, remains unknown.

The MT cytoskeleton has a significant role during C. elegans touch

The C. elegans TRN neurite contains a cross-linked bundle of as many as 50 MTs [69, 97, 98] (Fig. 3A). These distinctive MTs, which contain 15 protofilaments, depend on expression of MEC-12 α-tubulin, MEC-7 β-tubulin, and MEC-17 α-tubulin acetyltransferase [98,99]. Mutations that result in a loss of these MTs cause impaired touch sensation and altered sensitivity of MeT channels to external forces [29, 100]. The integrity and stability of the cross-linked bundle depends on the C. elegans doublecortin protein ZYG-8 [101] and the MT minus-end binding protein PTRN-1 [102]. As discussed above, doublecortin and other MT-associated proteins can increase the bending rigidity of the MTs [32], suggesting that the partial defects in touch sensitivity seen in zyg-8 and ptrn-1 mutants could result from a change in the mechanics of the MT bundle and the TRN. Consistent with this idea, MT-associated proteins PTL-1 and ELP-1, which are known to increase MT bending rigidity in vitro [32], are also necessary for full touch sensitivity [103,104]. Experiments have yet to reveal whether or not MTs participate directly in relaying mechanical signals along TRN neurites.

The presence of a specialized MT cytoskeleton in TRNs suggests that the MT bundle directly or indirectly involved in mechanosensation, e.g. provides a rigid mechanical support. Many filaments connect the MTs to the cortex of the TRNs [69, 97]. It is not known whether these tethers connect to the membrane directly or the membrane-bound actin-spectrin network. As suggested by Bounoutas et al. [100], unbinding of these tethers could release tension and help regulate MeT channel closure following activation. Such a mechanism is analogous to the tension release model [105] of hair cell adaptation. It will be interesting to discover whether such tethers contribute to membrane tension, MeT channel closure, or both.

Does force propagate along the actin-spectrin cytoskeleton?

A membrane subjacent actin-spectrin network characterizes many cell types, including neurons [106]. Spectrin assembles into tetramers containing two molecules each of α- and β-spectrin [107]. Such tetramers form through interactions between the C-terminus of β-spectrin and the N-terminus of α-spectrin; mutations in this tetramerization domain interfere with network formation and elasticity [108–111]. Whereas actin has been implicated in mechanical activation of MeT channels by direct force transfer [112], the role played by spectrin in mechanotransduction is less well studied.

Leveraging mutations in UNC-70 β-spectrin, we recently showed that TRNs are held under spectrin-dependent mechanical pre-stress, and interact mechanically with the engulfing epidermis [111]. In parallel, we also showed that defects in UNC-70 β-spectrin alter touch sensitivity. Both the severity of mechanical defects (buckling under compressive stress) and touch defects were correlated with the severity of the molecular defect in UNC-70. Taken together, these findings imply that the mechanical properties of TRNs are tuned to provide optimal sensitivity to external forces, similar to the overall body mechanics.

A mechanical systems perspective

A central question remains: How do these components interact to enable mechanical signal transmission and touch sensation? In C. elegans, external mechanical loads activate 20–30 MeT channels in one millisecond or less [29]. As the channels localize to puncta separated by 2–4 μm [29, 69], these experimental findings imply that mechanical signals travel over distances of 60–100 μm in, at most, one millisecond. As discussed above, theoretical considerations indicate that when mechanical deformation occurs orthogonal to the axis of a fiber, it propagates much farther along a pre-stressed fiber than along a slack one [27, 28, 113]. As touches delivered to the C. elegans cuticle and transverse to the TRNs are sufficient to elicit robust behavioral responses, we speculate that pre-stress may facilitate force transmission by increasing the distance of mechanical signal propagation along TRN neurites.

The contribution of a pre-stressed actin-spectrin cytoskeleton to mechanical signal transmission could exist on other levels. For instance, a pre-stressed actin-spectrin network could have a role in stabilizing MTs inside TRNs. As suggested from theoretical studies, embedding MTs in an active matrix such as a pre-stressed actin-spectrin network would increase their effective stiffness [114]. Thus, mechanical interactions between sub-structures within the TRN cytoskeleton could result in a mutual stabilization. But, the converse could also be true. Tension in the spectrin network could depend on the state of MTs inside axons [115]. Future experiments directed towards deciphering whether or not such a tight mechanical interaction between MTs and the neuronal spectrin network exists will be illuminating.

Pre-stressed actin-spectrin networks would also help keep the plasma membrane under tension and could, therefore, contribute to MeT gating according to the force-from-lipid model [58]. Alternatively, the actin-spectrin network might bind directly to the MeT channel, and thus transmit tension directly to the channel itself. In either case, tension in the network could shift the MeT channel to a point of maximal responsiveness, analogous to the gating spring in hair cells [116]. Tension release within the actin-spectrin network or between the membrane and its associated proteins could account for closure of MeT channels following their activation [29, 53, 100, 117]. However, whether mechanical pre-stress in the spectrin-actin network in fact tunes MeT activity remains to be established directly.

Conclusions and outlook

Here, we combined basic biophysical considerations about mechanical signal transmission within living cells with knowledge about the structures and proteins needed for touch sensation in C. elegans. The picture emerging from this synthesis is that a mechanical signal transmission pathway links the physical stimulus of touch to neuron activation, and hence to perception and behavior (Fig. 5). In the case of C. elegans, we envision a transverse pathway that transmits cuticle indentation through the epidermis and extracellular matrix, culminating in a local deformation of the TRN itself. It will be exciting to uncover the influence of the mechanics of fibrous organelles on touch sensation of freely moving animals, and whether or not they directly participate in mechanical signal transmission. Further, we speculate that spectrin-dependent tension within the TRNs contributes to lateral transmission and activation of the MeT channels, which in turn convert the mechanical signal into an electrical one. Thus, touch sensation depends on multiple cellular machines that act in concert to direct the mechanical signal from its source on the surface of the skin to MeT channels within the sensory neuron. It follows from this scenario that any genetic defect that alters the mechanical properties of the intervening links has the potential to impair touch sensation. Within this context, challenges for the future are to develop a holistic understanding of how proteins interact genetically, biochemically, and mechanically, and to take physical principles into account when developing mechanisms and models of sensory mechanotransduction. Therefore, quantitative and predictive models that link molecular properties to physiological function will be indispensable to understand how our body communicates with the physical world using the sense of touch.

Figure 5.

Hypothetical force transmission pathway during touch. A mechanical stimulation at the cuticle has to be transferred to the MeT channel through stratified tissues such as the cuticle, epidermis, and ECM surrounding the TRNs. Lipid bilayer mechanics and the cytoskeleton likely influence MeT function. Alternatively, a direct force transmission pathway via fibrous organelle (red dotted arrow) might be a fast track to MeT channel activation and behavior.

Abbreviations

ECM

extracellular matrix

MeT

mechano-electrical transduction

MS

mechanosensitive

MT

microtubule

TRN

touch receptor neuron