Peripheral mechanisms of tactile sensation in fish

Abstract

Touch is a fundamental sensory modality across vertebrates, yet its neuronal basis has been primarily investigated in amniotes. How tactile information is detected, transduced, and encoded in fishes, a diverse group of aquatic vertebrates, remains poorly understood. Here, we focus on the peripheral mechanisms of touch detection in the piscine somatosensory system, integrating anatomical, physiological, and molecular lines of evidence. Existing work reveals broadly distributed slowly and rapidly adapting tactile afferents across fish body regions and species. The terminal end-organs of these afferents have simpler anatomical structures than in amniotes. However, these tactile afferents detect a similarly broad range of mechanical cues, including sustained pressure, transient touch, velocity, and vibrations over a broad frequency range, and use similar encoding and mechanotransduction strategies. This perspective offers insights into general principles of vertebrate touch across aquatic and terrestrial environments and a comparative framework for the evolution of sensory specialization among tactile receptors.

Introduction

Fishes constitute the largest group of vertebrates, containing more than 35,000 extant species. Over more than 500 million years of evolution, fishes have not only diversified across temperate freshwater and marine habitats but also developed adaptations to colonize extreme environments such as polar waters, geothermal springs, lightless caves, and acidic lakes [1]. As part of these adaptive trajectories, fishes have evolved striking morphological and behavioral diversity and a wide range of senses to navigate their surroundings, locate prey, and avoid predators.

Many of these senses, including broadly conserved chemosensation and vision, as well as systems prominent in fish, such as the lateral line and electrosensory organs, have been extensively investigated. In contrast, the sense of touch through the skin via the somatosensory system, particularly its cellular and molecular mechanisms, has received comparatively little attention, as noted in the major review on fish tactile sensation published in 2011 [2]. Even a comprehensive fish physiology encyclopedia released in 2024 still lacks a dedicated chapter on this topic [3].

An argument can be made that, in aquatic habitats, the lateral line — a vestibulocochlear system that detects hydrodynamic cues through hair cells inside neuromasts — is particularly well suited for encoding water flow disturbances, compared with the somatosensory system, which detects touch through direct physical contact using cutaneous afferents from the trigeminal and dorsal root ganglia. However, fish in general, and benthic species in particular, regularly interact with their surrounding surfaces, conspecifics, and other animals, indicating extensive exposure to direct tactile cues. Despite this, the sensory capacity of the fish tactile system has received considerably less attention than the lateral line.

As in other vertebrates, tactile information in fish is conveyed from the skin or equivalent cutaneous tissues to the central nervous system via primary afferents of mechanically sensitive somatosensory neurons known as mechanoreceptors (Box 1). A subset of mechanoreceptors that detect non-damaging forces is known as low-threshold mechanoreceptors (LTMRs). However, the tactile sensitivity, functional specialization, and encoding strategies of fish LTMRs remain poorly understood.

Box 1. Common terminology.

Mechanoreceptor: a primary somatosensory neuron (e.g., a trigeminal or dorsal root ganglion neuron) specialized to detect mechanical stimuli. In the cutaneous system, mechanoreceptors are subdivided based on their sensitivity into low-threshold mechanoreceptors (LTMRs, also called tactile receptors), tuned to innocuous mechanical stimuli, and high-threshold mechanoreceptors (HTMRs), tuned to noxious mechanical stimuli.

Mechanosensory end-organ: a specialized terminal structure of a mechanoreceptor, located in the skin. Some end-organs have a multicellular organization, in which the afferent terminals are surrounded by specialized non-neuronal accessory cells (Box 2).

Mechanotransduction: the cellular process that converts a mechanical stimulus into a depolarizing ionic current, producing a receptor (generator) potential in the afferent terminal, and sometimes in the non-neuronal components, of a mechanosensory end-organ. Mechanotransduction is a prerequisite for action potential firing in a mechanoreceptor.

Mechanotransducer: the molecular machinery that performs mechanotransduction. In cutaneous mechanosensation, this is typically a mechanically gated ion channel that opens in response to mechanical force (e.g., Piezo2).

Recent studies combining behavioral, anatomical, physiological, and genetic approaches have begun to illuminate this understudied sensory modality in fish. In this review, we integrate recent discoveries within the piscine somatosensory system, and organize them into four themes spanning tactile organs, afferent terminal morphology, functional properties, and mechanotransduction mechanisms. By comparing tactile receptors in fish with their much better characterized counterparts in amniotes, we seek to deepen our understanding of general principles and evolutionary adaptations of tactile sensation across vertebrates.

Tactile organs

A major question, deceptively simple but foundational for framing this field, is: what anatomical regions of a fish body detect touch? As in amniotes, fish LTMRs are broadly distributed across the body surface, providing the anatomical substrate for tactile sensing in multiple behavioral contexts (Figure 1a–b).

Figure 1. Comparative anatomy and function of peripheral tactile mechanisms in amniotes and fish.

(a-b) In both amniotes and fish, LTMRs are distributed across the body surface, enabling touch in diverse behavioral contexts. (a) In amniotes, LTMRs innervate both hairy/feathered and glabrous (hairless/featherless) skin, exemplified by the primate fingertip and the duck bill. (b) In fish, LTMR innervation spans both scaled skin, where the epidermis typically overlies the scales, and glabrous (scaleless) skin. Glabrous skin covers the mouth, head, and fins (not shown), as well as specialized tactile appendages such as the “walking legs” of sea robins (Prionotus carolinus) and the Schnauzenorgan of elephantnose fish (Gnathonemus petersii). (c) Within the glabrous skin of amniotes, at least four LTMR subtypes can be distinguished by their characteristic terminal morphologies and functional specializations. Aβ SAI-LTMRs and Merkel cells form Merkel cell–neurite complexes via chemical synapses, together driving the sustained, irregular firing characteristic of SAI afferents in response to sustained pressure. In contrast, Aβ SAII-LTMRs, which may innervate Ruffini corpuscles, fire with regular inter-spike intervals and have been reported as sensitive to skin stretch in some species (uncertainties indicated by question marks). Aβ RAI-LTMRs innervate Meissner corpuscles at the apex of dermal papillae, adapt rapidly, and are most sensitive to vibrations around 40–60 Hz. Meissner corpuscles may also be innervated by a second, molecularly distinct afferent that adapts more slowly but remains vibration sensitive (not shown). Within Meissner corpuscles, LTMR endings are interdigitated with non-myelinating terminal Schwann cells, termed lamellar cells. Aβ RAII-LTMRs innervate Pacinian corpuscles and are tuned to high-frequency vibrations in the 100–400 Hz range. Each Pacinian corpuscle comprises a single RAII afferent surrounded by concentric terminal Schwann cell-derived lamellar cells and a multilayered outer core. (d) Within fish glabrous (scaleless) skin, the predominant tactile receptors are myelinated afferents with free nerve endings. To date, there is no convincing evidence that fish skin contains canonical Meissner or Pacinian corpuscles. In the Schnauzenorgan of elephantnose fish, at least three LTMR subtypes exhibit distinct functional profiles: Aβ SA-LTMRs fire throughout sustained pressure with regular inter-spike intervals; Aδ RA-LTMRs and Aβ RA-LTMRs both adapt rapidly to static indentation and are tuned to low-frequency vibrations around 30 Hz and high-frequency vibrations around 80 Hz, respectively. The mechanisms that generate subtype-specific specialization, and the extent to which these response properties generalize across fish species, remain to be determined. Merkel-like cells have also been reported in multiple fish species, but the functional properties of these cells and their associated LTMRs remain to be established (indicated by a question mark).

For example, the skin around the mouth and head is often scaleless, richly innervated by LTMRs, and actively engaged during foraging and exploration [4–6]. Fins are likewise exposed to frequent environmental contact and densely innervated [7,8]. In ray-finned fishes, a typical fin comprises bony rays interconnected by a membrane, both of which are mechanosensitive and innervated by proprioceptors and LTMRs [7–13]. In addition, scales in most fishes are covered by epidermis that is densely innervated by LTMRs, indicating substantial tactile capability in addition to their protective role [14–16].

In some lineages, tactile sensing is further enhanced through the evolution of specialized appendages, such as barbels, modified fins, and rostra, accompanied by corresponding neuronal substrates [2]. These structures parallel the roles played by fingers in primates or bills in tactile-specialist birds.

In sea robins, the membrane connecting the first three pectoral fin rays regresses during development, producing three free fin rays that function as “walking legs” [17]. In these digging fishes, such as Prionotus carolinus, the free rays function as chemotactile organs aiding in excavating and probing the substrate to locate buried food [18–20].

Adhesive discs in remoras (Echeneidae), or sharksuckers, represent another example of fins modified for tactile function [21]. These discs, thought to have evolved from dorsal fins, enable remoras to securely attach to hosts such as sharks, thereby improving locomotor and foraging efficiency. To establish and maintain attachment to large and fast-moving hosts, the discs must provide reliable mechanical feedback [21].

Another example is the Schnauzenorgan of the elephantnose fish (Gnathonemus petersii), an elongated chin appendage that functions as a tactile rostrum. Long regarded primarily as an electroreception fovea enriched with electroreceptors, the Schnauzenorgan is now recognized as a sophisticated tactile appendage that mediates a wide range of active tactile behaviors, including substrate probing and object displacement during benthal foraging [22,23].

Architecture of mechanosensory end-organs

Our current understanding of the sense of touch in vertebrates is shaped primarily by work in mammals, and, more recently, birds, whose capacity to discern the nuanced complexity of tactile information relies on a diverse array of thickly myelinated fast-conducting (Aβ-type) LTMRs (Figure 1c) [24–26]. Based on electrophysiological recordings from primary afferents, Aβ-LTMRs are classified according to the pattern of action potential (AP) firing in response to dynamic and static mechanical stimulation. Slowly adapting type I and II (SAI and SAII) LTMRs fire APs throughout the duration of a sustained stimulus, whereas SAI-LTMRs produce irregularly spaced spikes, and SAII-LTMRs display highly regular discharge patterns. In contrast, rapidly adapting type I and II (RAI and RAII) LTMRs respond only during the dynamic phases of stimulation and differ in their responsiveness to stimulation frequency and velocity [24].

The functional specialization of amniote LTMR subtypes is thought to arise from two factors. First, it reflects the intrinsic mechanosensory properties of their afferent terminals. Second, it is shaped by non-neuronal cells that associate with afferent terminals to form mechanosensory end-organs in the skin (Box 1). Within glabrous skin, four major mechanosensory end-organs have been described in association with distinct LTMR subtypes: Merkel cells are associated with SAI afferents, Meissner corpuscles with RAI, Pacinian corpuscles with RAII, and Ruffini corpuscles, putatively linked to SAII (Box 2) [24,26].

Box 2. Mechanosensory end-organs of amniote glabrous skin.

Free nerve endings are located in the epidermis and are typically the terminals of HTMRs or polymodal nociceptors in amniote glabrous skin. Although free nerve endings lack a defined accessory structure, recent reports suggest that their sensory properties are modulated by keratinocytes [27–31].

Merkel cells, located in the deepest layer of the epidermis, are specialized oval-shaped epithelial cells with spine-like microvilli on their apical surface. SAI-LTMR afferents lose their myelin sheaths just below the dermal-epidermal junction and arborize into unmyelinated neurites that receive synaptic input from Merkel cells [32]. Within a Merkel cell-neurite complex, both the SAI-LTMR afferent and Merkel cells are mechanosensitive, but their contributions to touch detection differ: the afferent terminal primarily detects the dynamic phase of stimulation, whereas Merkel cells respond to static forces and drive action potential firing in the afferent via chemical synapses [33–38]. The clustered organization of Merkel cell-neurite complexes, together with synaptic signaling, is thought to contribute to the characteristic irregular firing pattern of SAI-LTMR afferents during sustained pressure [32].

Ruffini corpuscles have been proposed to correspond to SAII-LTMR afferents based on correlative electrophysiological and anatomical analyses in cat hairy skin [39]. However, comparable studies in the glabrous skin of other mammalian species have not yielded analogous structures, despite the well-documented presence of regularly firing SAII-LTMR afferents in the corresponding skin regions [40,41].

Meissner corpuscles contain one (RAI-LTMR) or sometimes two (RAI- and SA-LTMR) afferent terminals interdigitated with non-myelinating terminal Schwann cells known as lamellar cells [42–44]. The resultant end-organ is a spherical or slightly elongated structure situated within the dermal papillae of glabrous skin [42,43].

A related but anatomically restricted end-organ is the Krause corpuscle, found only in mucocutaneous skin (e.g., genitalia, tongue). Like Meissner corpuscles, Krause corpuscles are tuned to low-frequency vibrations and are labeled by the same molecular markers. They are also composed of one or two afferent terminals wrapped by lamellar cells but show variable morphology, ranging from simple forms with linear terminals to complex forms with tightly coiled terminals [45].

Pacinian corpuscles are mainly located in the deep dermis, although in some species like mice, they are predominantly found in the periosteum of bones. The sensory core of Pacinian corpuscles is composed of an afferent terminal from an RAII-LTMR surrounded by concentric lamellar cells, and further encapsulated by a multilayered outer core, giving these structures an onion-like appearance in cross-section [43,46].

Both Meissner and Pacinian corpuscles detect transient touch and are tuned to low- and high-frequency vibrations, respectively, contributing to the detection of moving objects and textures, and facilitating the capacity to handle and manipulate tools [24]. Recent studies in mammals and birds have shown that lamellar cells in Meissner and Pacinian corpuscles are also mechanosensitive and excitable, playing active roles in touch detection and contributing to the sensory tuning of the end-organs [42,46–48].

In contrast to the abundant research on mechanosensory end-organs in amniotes, there are relatively few reports documenting end-organ-like structures in fish [2]. This raises the question of whether fish even possess canonical end-organs found in mammals and reptiles. In this section, we will detail current knowledge of mechanoreceptor terminals and end-organ-like structures found in fish skin (Figure 1d).

Myelinated fibers with free nerve endings

Free nerve endings constitute the predominant type of nerve terminals in fish skin. Unlike in amniotes, where free nerve endings are typically reported to arise from unmyelinated afferents and primarily to mediate nociception, fish skin contains myelinated afferents with free nerve endings that likely serve tactile and proprioceptive functions [2]. This organization has been documented across multiple tactile organs and species, especially among teleosts, using both traditional histological methods, including silver impregnation and serial-section electron microscopy, and the recently developed approaches such as whole-mount immunostaining and tissue clearing [10,14,22,49,50].

Typically, these myelinated fibers travel in bundles and, together with unmyelinated ones serving other sensory modalities, form nerve plexuses in the dermis [49,50]. As these fibers penetrate the basal membrane at the dermal–epidermal boundary, they shed their myelin sheath and spread in multiple directions, ramifying extensively within the epidermis but failing to reach the most superficial layers [22,49,50]. Most terminals end as free nerve endings located between adjacent epithelial cells; in some reports, epithelial cells partially enclose the terminals, though the nature and functional significance of this interaction remain unclear [14,50–52]. Many intraepidermal fibers contain varicosities rich in mitochondria and, in some cases, vesicular structures proposed to be components of smooth endoplasmic reticulum [50].

Merkel-like cells

Epidermal cells resembling Merkel cells have been identified in distinct tactile organs of lampreys, lungfish, and a limited number of teleost species [53]. However, despite careful examination, they have not been detected in hagfish, cartilaginous fish, most non-teleost bony fish, and in a number of teleost species [2,53]. Unlike mammalian Merkel cells, which often form clusters in the deepest layer of epidermis [24], fish Merkel-like cells are distributed relatively evenly in more superficial epidermal layers [2,15,53].

Historically characterized through electron microscopy studies, fish Merkel-like cells share three major morphological features with their amniote counterparts, albeit with some divergences [2,15,53]. First, they possess spine-like protrusions, or microvilli, with species-dependent variations in shape, disposition, and number. Second, they contain components of the neurosecretory machinery. Whereas mammalian Merkel cells contain dense-core vesicles, some reported fish Merkel-like cells instead have small granules resembling clear vesicles. Third, they form chemical synapses with afferent nerves.

Despite these morphological similarities, it had long remained unclear whether fish Merkel-like cells are developmentally and functionally homologous to those in amniotes. Brown et al. used genetic labeling to show that Merkel-like cells in zebrafish (Danio rerio) [15] express Atoh1a, a transcription factor and a genetic marker of mammalian Merkel cells [54]. Immunostaining further showed that zebrafish Merkel-like cells express molecular components of the neurosecretory machinery. They contain serotonin, one of several neurotransmitters proposed to be used by mammalian Merkel cells to communicate with the afferent [36]. Notably, these cells also take up the ion-channel-activity-dependent dye FM 1–43FX, consistent with the expression of mechanically gated ion channels, as in mammalian Merkel cells [34,37,55,56]. These findings warrant electrophysiological recordings from both Merkel cells and their associated afferents in zebrafish skin to compare their functions with mammalian Merkel cells and SAI afferents, thereby illuminating the functional evolution of cutaneous end-organs that detect sustained pressure.

Corpuscle-like and other types of mechanosensory end-organs

Evidence for corpuscular or encapsulated mechanoreceptors in fish is exceedingly scarce and tenuous [2]. Reports of structures superficially resembling Meissner and Pacinian corpuscles in moray eels [57] and elasmobranch skin [58] are based on limited evidence from two-dimensional histology, with neither their identities nor functions convincingly established.

In cases where electrophysiological recordings have identified cutaneous RA-LTMRs tuned to a broad range of vibratory frequencies—functionally paralleling RAI and RAII afferents innervating Meissner and Pacinian corpuscles—in the Schnauzenorgan of the elephantnose fish, histological analyses using three complementary approaches nonetheless failed to identify corresponding encapsulated structures, instead revealing only myelinated fibers terminating as free nerve endings [22].

Several mechanosensory structures have been reported exclusively in specific fish taxa, highlighting lineage-specific solutions for mechanosensation, even though these structures appear to serve proprioceptive rather than tactile functions. For example, hagfish (such as Myxine glutinosa) possess elongated, lamellated receptors located on the body fascia of the parietal muscles and arranged segmentally along the body. These structures can reach up to 10 mm in length, much longer than any amniote corpuscle, and have been hypothesized to function as stretch receptors, although functional evidence is lacking [59,60]. In cartilaginous fish, Wunderer corpuscles were originally reported as coiled corpuscular endings located at the bases of fins in various shark species [61]. However, subsequent examinations failed to identify a sheath-like capsule surrounding the nerve coil [62,63], and electrophysiological recordings indicated that these receptors are slowly adapting mechanoreceptors functioning as second-order proprioceptors [64]. Similarly, Poloumordwinoff endings, found within the intramuscular connective tissue of paired fins in rays [65,66], morphologically resemble deeply embedded myelinated fibers with free nerve endings and are slowly adapting mechanoreceptors sensitive to tension [67].

Taken together, existing evidence supports the hypothesis that typical Meissner and Pacinian corpuscles are absent from fish skin and that these structures likely originated later in vertebrate evolution as innovations associated with the transition to terrestrial life.

Tactile encoding

In amniotes, each LTMR subtype preferentially detects and transmits a particular tactile submodality: sustained pressure, transient touch, velocity, and vibration. Together, the specialized LTMRs supply the central nervous system with sensory information that enables rich tactile perception through population coding [24]. Within this framework, SAI afferents encode the intensity and duration of sustained pressure and serve as primary detectors of rough textures and edges; SAII afferents are thought to respond to skin stretch in some species; and RAI and RAII afferents are tuned to low- and high-frequency vibrations, respectively, facilitating object manipulation and discrimination of fine textural features. As mentioned previously, in amniotes LTMR terminals often form multicellular end-organs, which include epithelial or terminal Schwann cells. These non-neuronal components also detect touch, modulate the functionality of the afferent terminal, and thus contribute to the functional tuning (Figure 1c) [42,46–48].

The apparent structural simplicity of mechanoreceptor terminals in fish motivates a set of central questions: do distinct LTMR subtypes with specialized functional properties exist? To what extent can these LTMRs encode the complexity of natural tactile cues? How is such functional specialization achieved given the paucity of multicomponent end-organs typical of amniotes?

Earlier works in several fish species revealed the presence of SA- and RA-LTMRs [4–6,68–72]. Subsequent studies using a broader range of mechanical stimuli extended these observations. Using multi-fiber recordings and spike sorting, Hardy and Hale identified both SA and RA afferents innervating the pectoral fin rays of the round goby (Neogobius melanostomus), a benthic species that frequently contacts the substrate with its pectoral fins [12]. Notably, similar to the vibration-encoding RA afferents in amniotes, these fish RA afferents also exhibit phase-locked responses to vibrations delivered by a rotating drum with periodic gratings, indicating their capacity to encode textural information. However, the afferent terminal structures were not examined in this context, and the tested frequency range did not exceed 28 Hz, limiting direct comparison with RA afferents in amniotes, which are typically associated with corpuscular end-organs and operate over a broader frequency spectrum of up to hundreds of hertz [24,25].

In a systematic comparative study, Tan et al. used a force-controlled indenter to deliver quantitative mechanical stimuli coupled with single-fiber recordings from the somatosensory afferents innervating the Schnauzenorgan skin of the elephantnose fish [22]. This approach revealed three subtypes of fast-conducting fish LTMRs—Aβ SA, Aδ RA, and Aβ RA—with Aδ afferents exhibiting slightly slower conduction velocities. Fish Aβ SA-LTMRs fired continuously during sustained indentation with regular inter-spike intervals, closely resembling mammalian SAII afferents. Aδ and Aβ RA-LTMRs exhibited distinct velocity sensitivity and vibration frequency preferences, functionally analogous to RAI (Meissner) and RAII (Pacinian) afferents in amniotes (Figure 1d).

The apparent mismatch between the functional diversity of LTMRs and the morphological simplicity and uniformity of their terminals raises two related questions. First, how is functional specialization achieved in the absence of corpuscular end-organs? Although the underlying mechanisms remain to be elucidated, potential contributors include subtype-specific expression of ion channels [73,74], less conspicuous morphological features such as terminal branching and local tissue mechanics [75,76], as well as possible functional contributions from unidentified surrounding cells [27–31]. Second, what functional advantages do corpuscular end-organs provide in amniotes? In elephantnose fish, Aδ and Aβ RA-LTMRs detect low- and high-frequency vibrations with peak sensitivities near 30 and 80 Hz, respectively. This functional separation is reminiscent of Meissner and Pacinian corpuscles, which are typically most sensitive to, respectively, 40–60 and 100–400 Hz, but is shifted toward lower frequencies overall [24]. Non-neuronal elements such as lamellar Schwann cells may actively sharpen the frequency tuning of Meissner and Pacinian corpuscles or extend their dynamic range, potentially as adaptations to the skin mechanics and sensory demands of terrestrial life.

Together, functional studies across multiple preparations, tactile organs, and species support the view that fish LTMRs use encoding strategies broadly conserved with those of amniotes, providing functional evidence that fine tactile discrimination may have deep evolutionary origins. Substantial sensory specialization can be implemented at the level of the afferent terminal itself, with corpuscular end-organs providing an additional modulatory layer rather than acting as the sole determinant.

Mechanotransduction

The first step in touch detection is the mechano-electric conversion event carried out by mechanically gated ion channels located in the plasma membrane of tactile afferent terminals and, in some cases, the surrounding cells within associated end-organs. These channels, known as mechanotransducers, open in response to cellular deformation, generating a depolarizing receptor potential that, if sufficient, recruits voltage-gated ion channels to trigger action potentials, which then propagate along the afferent to the central nervous system (Box 1) [26].

Although many ion channels have been shown to exhibit mechanosensitivity, the Piezo family, particularly Piezo2, has emerged as the principal mechanotransducer for light touch across LTMR subtypes of amniotes [77,78]. Piezo homologs are also the primary candidates for mediating tactile mechanotransduction in fish [20,22,79], supporting the idea that Piezo channels are evolutionarily conserved somatosensory mechanotransducers across vertebrates.

In zebrafish, which possess at least four Piezo homologs, piezo2b is expressed in larval somatosensory neurons innervating the head and body skin. Knockdown of piezo2b selectively disrupts larval light-touch responses while leaving noxious mechanical and chemical sensation intact, establishing piezo2b as the key mediator of light-touch detection in the zebrafish larval somatosensory system [79].

Abundant expression of Piezo homologs has also been reported in other fish species, although their functional validation as mechanotransducers is often limited by the lack of genetic tools. In the elephantnose fish, two Piezo2 homologs are abundantly expressed in the somatosensory neurons innervating the Schnauzenorgan [22]. In the sea robin, both piezo1 and piezo2 exhibit robust expression in somatosensory neurons innervating the specialized “walking legs” [20].

At the biophysical level, fish Piezo channels expressed in heterologous systems are functionally similar to their counterparts in amniotes, yet they show notable variation in inactivation kinetics. Channel inactivation determines how rapidly the mechanotransduction currents decay during sustained stimulation, which shapes the temporal profile of receptor potentials and the adaptation patterns of action potential firing. In this context, the elephantnose fish Piezo2-like1 displays an inactivation time constant approximately threefold slower than typical Piezo2 [22]. In zebrafish, Piezo1 inactivates roughly fourfold more slowly than its mammalian ortholog in outside-out patches and can be gated purely by membrane voltage [80].

The diversity of functional responses observed among fish Piezo homologs may contribute to the distinct physiological properties observed across LTMRs. Moreover, fish Piezo homologs provide valuable substrates for structure–function analyses of Piezo channels.

Concluding remarks

Taken together, recent studies have highlighted a remarkable conservation of tactile encoding and mechanotransduction between some fish species and amniotes, although the generalizability of some conclusions remains to be determined [12,15,22]. Anatomically, the apparent absence of corpuscle-like structures in the piscine somatosensory system underscores a key role played by tactile afferent terminals in the detection of different tactile cues, suggesting that the more complex end-organs found in amniotes reflect adaptations to a terrestrial lifestyle. However, there remains the possibility that the functional specificity of fish LTMRs is modulated by epithelial cells in the skin, similar to the role played by keratinocytes in noxious mechanosensation in mammals [27–31]. A key priority in the future is to understand the structural and molecular mechanisms underlying the tactile specialization of different fish LTMR subtypes. The remarkable morphological and ecological diversity of fishes offers a rich resource to address these questions, but realizing its potential will require multidisciplinary and standardized approaches that enable direct comparison both across fish species and between fishes and amniotes. Elucidating the mechanisms of touch detection in fish provides a comparative and evolutionary framework for understanding the overarching principles and specialized mechanistic solutions employed by the somatosensory system to shape tactile functions and meet ecological demands across aquatic and terrestrial vertebrates.

Highlights.

• Fish tactile receptors offer insights into the evolution of the sense of touch

• Tactile receptors innervate scaled and scaleless skin, including specialized organs

• Tactile receptors are functionally diverse despite lacking corpuscle-like structures

• Encoding strategies of fish tactile receptors are largely conserved with amniotes

• Piezo channel homologs contribute to light touch detection across fish species