Taste transduction and channel synapses in taste buds

Sour taste

Extracellular and intracellular protons are the primary stimulant for sour-sensing type III cells. Therefore, H⁺-releasing chemicals (i.e., acids) serve as sour tastants. Despite speculation that ionic taste stimuli are transduced by ion channels, the molecular identity of the sour sensor channel remained elusive until recently. Although many cation channels were proposed as candidates, including ASICs [99], HCNs [83], and TRP channels PKD2L1, and PKD1L3 [34, 36], subsequent studies failed to support their involvement in sour taste [33, 68, 69]. Nevertheless, the complete loss of gustatory nerve responses to acids by the ablation of PKD2L1⁺ cells identified PKD2L1 as a marker gene for sour-tasting type III cells [34]. In the last decade, a series of studies by Emily Liman and colleagues have taken advantage of this molecular identification to reveal an elaborate sour transduction cascade that involves a previously unrecognized H⁺-selective ion channel, otopetrin1 (OTOP1) as the bona fide sour sensor channel (Fig. 1b). The initial event in sour transduction is a H⁺ current in the apical membrane through a unique Zn²⁺-sensitive H⁺-selective conductance that can depolarize the cells sufficiently to drive action potential generation [8, 13]. Comparative transcriptome analysis between PKD2L1⁺ and TRPM5⁺ taste cells identified OTOP1 as a novel H⁺-selective ion channel protein required for the H⁺ current specifically in type III cells [98]. The structure of zebrafish OTOP1 resolved by single-particle cryo-electron microscopy (EM) provided insights into how OTOP1 channels selectively conduct protons [76]. Otop1 knockout (KO) mice had severely attenuated type III cell and gustatory neuron responses to acids, confirming OTOP1 as the sour sensor [92, 114]. The H⁺ influx through OTOP1 results in depolarization and a decrease in intracellular pH (pHⁱ), leading to inhibition of a pH_i-sensitive resting K⁺ conductance mediated by K_ir2.1 channels that enhances depolarization [110]. Type II cells also have K_ir2.1 currents that account for their responsiveness to intracellular acidification [110]. Whereas they are larger than those in type III cells, the higher input resistance of type III cells results in a stronger depolarization. Together, OTOP1 and K_ir2.1 channels work in concert to transduce acid stimuli into a membrane depolarization, triggering action potentials that activate voltage-gated Ca²⁺ channels for Ca²⁺-dependent exocytotic release of neurotransmitters at synapses with gustatory neurons. Although adenosine triphosphate (ATP) and 5-HT have been implicated [28, 41], the primary neurotransmitter linking type III cells and afferent neurons remain to be established. Weak acids such as acetic acid are perceived to be more sour than strong acids at the same pH, likely because protonated forms of weak acids can permeate the plasma membrane and release H⁺ in the cytoplasm, leading to additional intracellular acidification.

Notably, type III cells respond not only to acids but also to carbonation [12], high concentrations of salts [62], and even water [118]. Furthermore, acids stimulate both type III taste cells and trigeminal nociceptors in the oral cavity [95, 105]. Optogenetic activation of type III cells evoked aversive and attractive behaviors in mice, possibly depending on the physiological conditions [106, 118]. Thus, it remains to be resolved how neural outputs from type III cells contribute to our perception of these taste qualities.

Sweet, umami, and bitter tastes

Sweet, umami, and bitter tastants are sensed by specific G protein-coupled receptors expressed in dedicated type II cells: sweet, umami, and bitter cells. The TAS1R family members constitute the sweet and umami receptors [57, 58, 117]. Sweet cells use a heterodimer of TAS1R2 and TAS1R3 (TAS1R2/3) for detecting natural and artificial sweeteners, whereas umami cells employ a heterodimer of TAS1R1 and TAS1R3 (TAS1R1/3) for detecting various L-amino acids and nucleotides. Meanwhile, each of the bitter cells in humans and mice expresses ~ 25 and ~ 35 TAS2Rs, respectively to detect numerous, chemically diverse bitter tastants [1, 11]. Expression of sweet, umami, and bitter receptors in taste buds is segregated to distinct cells, although identification of cells expressing all TAS1R subunits has been reported [40]. Downstream of the taste receptors, sweet, umami, and bitter cells share a similar intracellular signaling cascade to transduce chemical stimuli into action potential firing. Binding of tastants to the taste receptors initiates the following signaling cascade (Fig. 1c): (i) activation of phospholipase Cβ2 (PLCβ2) [115], (ii) production of inositol 1,4,5-trisphosphate (IP₃), (iii) an increase in intracellular Ca²⁺ concentration ([Ca²⁺]_i) due to Ca²⁺ release from the endoplasmic reticulum through type 3 IP₃ receptor Ca²⁺ channels (IP₃R3) [31], (iv) Ca²⁺_i-dependent activation [32, 43, 66, 116] of plasma membrane monovalent cation-selective ion channels TRPM5 [65, 115] to generate a graded depolarization, and (v) Na⁺ action potential generation [113]. Of note, previous studies involving Trpm5 KO mice reported a TRPM5-independent Ca²⁺_i-activated ion channel in type II cells with a much lower affinity for Ca²⁺_i than TRPM5 [116] and residual gustatory nerve and behavioral responses to sweet, umami, and bitter [18]. A recent study provided evidence that TRPM4, another Ca²⁺-activated TRP channel, mediated the TRPM5-independent responses [26]. These electrically excited type II cells transmit taste information to the afferent neurons via a channel synapse, as described below.

In addition to the canonical TAS1R-mediated mechanisms, some studies have shown that there are TAS1R-independent receptor mechanisms for sweet and umami tastants. Indeed, Tas1r3 KO mice lost the perception of artificial sweeteners and maintained reduced but significant responses to sugars, especially glucose, and to glutamate [19]. The TAS1R3-independent detection of glucose and glutamate involves sodium-glucose cotransporter 1 (SGLT1) [109, 111] and metabotropic glutamate receptors (mGluR1 and mGluR4) [14, 108], respectively. It is believed that SGLT1-mediated glucose-coupled Na⁺ influx provides a supra-threshold depolarization for action potential generation, based on the observation that the non-metabolizable glucose analog, methyl-a-d-glucopyranoside, mimicked the action of glucose [109]. GLUT family glucose transporters and ATP-sensitive K⁺ channels (K_ATP) may also contribute to depolarization by constituting a metabolic pathway, in which monosaccharides transported into cells by SGLT1 and GLUTs are metabolized to raise the ratio of ATP to ADP, leading to the closure of K_ATP channels and thereby cause membrane depolarization [111]. Since SGLT1, GLUT4, and K_ATP channels are preferentially expressed in TAS1R3⁺ taste cells [111], the SGLT1-dependent sugar detection mechanism involving glucose-coupled Na⁺ influx may explain why a pinch of salt (NaCl) enhances sweetness in food. Curiously, intestinal SGLT1 was recently identified as a glucose sensor that initiates a gut-to-brain post-ingestive sugar sensing pathway critical for the development of sugar preference [86]. Thus, SGLT1 is a key molecule for detecting sugars in both the mouth and gut.

Salty taste

Sodium is an essential mineral because it is the major cation of the extracellular fluid and thereby maintains the body fluid volume homeostasis and various cellular functions. On the other hand, a high dietary intake of sodium is harmful because it is associated with elevated blood pressure, a risk factor for cardiovascular diseases [54, 84]. To control the amount of sodium ingested within a proper range, many mammals are equipped with two salt taste mechanisms: sodium taste [10] and high-salt taste [62].

Sodium taste

The remarkable features of sodium taste are the strict selectivity to sodium salts and sensitivity to a diuretic, amiloride, and it typically evokes behavioral attraction. The pioneering work of DeSimone and colleagues [30] and subsequent contributions from many others [75] culminated in the identification of the sensor for sodium taste [10] as the amiloride-sensitive epithelial Na⁺ channel (ENaC), which is composed of α, β, and γ subunits [9, 61] and localizes in apical membranes of taste cells [112]. However, which taste cells are involved, and the details of the intracellular signaling mechanisms underlying sodium taste have remained long-standing enigmas. Sodium taste cells are present in taste buds in fungiform papillae of the anterior tongue but not in circumvallate papillae of the posterior tongue [39, 59]. Until recently, it had been believed that sodium taste cells are a subset of type I cells [15, 37, 56, 79] due to the absence of co-expression of all three ENaC subunits in taste cell populations expressing marker proteins of type II and III cells (TRPM5 and Car4, respectively) [10]. However, the properties of type I cells are apparently not compatible with taste sensor functions, as they are reportedly electrically non-excitable [51]. How are sodium stimuli transduced in sodium taste cells? One study detected an amiloride-sensitive current (I_amiloride), an index of ENaC activity, only in taste cells lacking voltage-gated Na⁺ currents (I_Nav) [102], while others reported cells having both I_amiloride and I_Nav [23, 52]. Thus, it has been unclear whether sodium taste cells are electrically excitable. Furthermore, the involvement of Ca²⁺ signals in sodium taste transduction has been controversial [6, 10], and the neurotransmission mechanisms from sodium taste cells to afferent neurons remained unexplored. A recent report [60] identified the cells responsible for sodium taste and elucidated the mechanisms of sodium taste transduction (Fig. 1c). Patch-clamp electrophysiology and Ca²⁺ imaging in ENaCα⁺ cells from fungiform taste buds of mice expressing GCaMP3 under the ENaCα promoter found I_amiloride in two functionally distinct cell types: electrically excitable and non-excitable ones. In the excitable cells, Na⁺ influx through ENaCs evoked action potentials without changes in [Ca²⁺]_i. Whole-cell current-clamp recordings with strong intracellular Ca²⁺ chelation further demonstrated that I_amiloride alone was sufficient to evoke action potentials without the aid of Ca²⁺ signals. Thus, ENaC-mediated Na⁺ influx alone drives action potential generation. The excitable, but not the non-excitable cells with I_amiloride, expresses the CALHM1/3 ion channel, the neurotransmitter-release channel found in type II cells [48]. Importantly, conditional KO of ENaCα in CALHM1⁺ cells abolished amiloride-sensitive gustatory nerve responses and attenuated behavioral attraction to NaCl, indicating that the excitable cells with I_amiloride mediate appetitive sodium taste. Consistent with earlier studies suggesting a role of CALHM channels in sodium taste [4, 97], global Calhm3 KO mice lost perception of sodium taste, and CALHM1 was localized at points of contact between sodium taste cells, identified as ENaCα⁺ CALHM1⁺ cells, and P2X2⁺ afferent nerves (Fig. 2a), a structural feature of the CALHM1/3 channel synapse in type II cells, as described below. These findings suggested that sodium taste cells employ a channel synapse for neurotransmission to afferent neurons. Together, ENaCα⁺ CALHM1/3⁺ cells constitute sodium taste cells, and ENaC-mediated entry of oral Na⁺ elicits a supra-threshold depolarization for action potentials driving voltage-dependent neurotransmitter release via the CALHM1/3 channel synapse. All steps in sodium transduction are voltage driven and independent of Ca²⁺ signals, in contrast to other types of taste cells. Of note, although the requirement of ENaCα was established by the loss of sodium taste in ENaCα KO animals [10, 60], a recent study failed to find taste cells expressing all three ENaC subunits (α, β, and γ), questioning the stoichiometry of the functional amiloride-sensitive Na⁺ channel in sodium taste cells [44].

Are sodium taste cells, which are distinct from type II and III cells [10], a subset of type I cells? Sodium cells identified as ENaCα⁺ CALHM3⁺ cells do not express NTPDase2, a type I cell marker [60]. Thus, sodium taste cells appear to define a previously unidentified taste cell population in addition to the widely known type I/II/III taste cell classes, suggesting a greater diversity than previously realized in fungiform taste cells. The taste cell classification has mainly been developed in circumvallate taste buds, and it is unclear whether it applies to fungiform taste buds. Further studies are needed to clarify the diversity in fungiform taste cells and how sodium taste cells are classified among them.

Importantly, while the role of ENaC in rodent salty taste has been established, amiloride-sensitivity of perceived saltiness in humans has been controversial [2, 29, 50, 63, 64, 77, 81, 93, 94]. This may reflect an unconventional subunit composition or subcellular localization of ENaC in human taste cells [82, 104]. Direct evidence supporting ENaC as the principal sensor of human salty taste has yet to be demonstrated [5].

High-salt taste

Alternatively, perceived saltiness is determined by multiple salt taste pathways that are subject to dynamic modulation by experience or the internal state. Indeed, in addition to sodium taste, there is another taste pathway responding to salts called high-salt taste. High-salt taste has a higher threshold concentration than that of sodium taste and is evoked by various salts including NaCl and KCl. This pathway is also known as amiloride-insensitive salty taste because it is not affected by amiloride. In taste buds, the perception of high salt relies on bitter-sensing type II cells and sour-sensing type III cells and typically evokes behavioral avoidance [62]. Recruiting multiple cell types likely involves multiple receptor mechanisms. One or more bitter receptors and carbonic anhydrase 4 were initially suggested in bitter and sour cells, respectively [62]. A more recent study [70] suggested Cl⁻ as a key determinant of high-salt responses in type II cells but ruled out the involvement of Cl⁻ channels and transporters (see also [27]). It has also been suggested that a subset of type III cells responds to high-salt stimuli through multiple mechanisms, including osmotically activated Na⁺-conducting ion channels [42]. Due to these seemingly contradictory findings, the nature of high-salt receptors remains elusive. Adding another layer of complexity, CALHM1/3-dependent, ENaC-independent salt attraction behavior with a high threshold concentration was recently observed in salt-depleted mice [60]. It is unknown whether this attractive high-salt response represents a novel salty taste mechanism or is a result of the central modulation of preference for a known high-salt taste pathway. Identification of high-salt sensors and understanding of modulation and integration of each salt taste pathway at the periphery and in the brain are imperative in order to understand the human salty taste and devise strategies to reduce salt consumption.

Channel synapse

Type II cells transmit taste information to the gustatory neurons by release of neurotransmitter, but the mechanism underlying this neurotransmission was a long-standing mystery. The most widespread mechanism of chemical neurotransmission is that chemicals contained in synaptic vesicles (a.k.a. neurotransmitters) are released into the synaptic cleft by Ca²⁺-activated exocytosis and bind to specific receptor proteins expressed in the plasma membrane of a postsynaptic cell. However, type II cells do not possess essential components of conventional synapses such as synaptic vesicles and SNAP25, a key protein for Ca²⁺-activated transmitter exocytosis [21]. How do taste cells lacking synaptic vesicles mediate neurotransmission?

Discovery of the neurotransmitter-release channel, CALHM1/3

ATP was suggested to be the primary neurotransmitter linking taste cells to sensory nerve fibers. Type II cells release ATP in response to taste stimuli [28, 35, 55, 73, 91], afferent gustatory neurons innervating taste cells express ATP receptors composed of P2X2/P2X3 [7, 49], and double KO of P2X2 and P2X3 abolished taste-evoked responses to sweet, bitter, and umami substances [28]. NTPDase2 is expressed in type I cells and likely contributes to taste perception by degrading extracellular ATP to ADP [3, 101]. Thus, ATP meets most criteria for a neurotransmitter, including release, presence of specific receptors, and mechanisms for clearance [28]. The role of ATP in the sour transmission is unclear. Although P2X2/P2X3 double KO mice lost sour perception [28], ATP release has not yet been detected from type III cells. However, the mechanism for type III cell neurotransmitter release has been well understood. Electrophysiological estimates of the changes in membrane surface area from fluctuations of whole-cell membrane electrical capacitance provided functional evidence that type III cells responded to membrane depolarization by increasing the cell surface area due to the insertion of synaptic vesicles in regulated exocytosis [103]. These functional data are in agreement with anatomical evidence that a conventional vesicular synapse exists in type III cells [17]. However, these functional and anatomical features were not detected in type II cells. The implication was that type II cells use non-exocytotic mechanisms to release ATP, possibly involving ion channels. There are five known classes of ATP-permeable ion channels, including connexin hemichannels, pannexin 1, maxi-anion channels, volume-regulated anion channels, and CALHM channels (reviewed in [88]). Although pannexin 1 was initially proposed as the ATP-release channel in type II cells [20, 35], genetic KO studies [71, 96, 100] precluded its involvement since ATP release from type II cells and taste perception were both intact in Panx1 KO mice. Connexins were also implicated in the type II cell ATP release [73, 74], but their contributions to taste perception have not been tested using KO models. Meanwhile, CALHM1 was identified as an essential component of the ATP release machinery in type II cells [91].

CALHM1 was originally identified as an essential component of a previously uncharacterized ion channel that controls susceptibility to Alzheimer’s disease [24]. Subsequently, it was established as a pore-forming subunit of a plasma membrane non-selective voltage-gated ion channel with a wide pore [47, 80]. Remarkably, activation of CALHM1 leads to translocation of ATP from the cytoplasm to the extracellular milieu, identifying CALHM1 as a novel voltage-gated ATP-release channel [91]. In taste buds, CALHM1 expression was found selectively in type II cells [53], and Calhm1 KO abolished non-selective voltage-gated currents and taste-evoked ATP release in type II cells and severely impaired gustatory nerve and behavioral responses to sweet, bitter, and umami tastes. These findings established CALHM1 as an essential component of the type II cell ATP-release channel [91]. However, the kinetics of voltage-dependent activation of CALHM1 expressed in heterologous systems (τ > 0.5 s) are too slow to be activated by rapid Na⁺ action potentials that trigger ATP release in taste cells [55, 91]. Furthermore, the endogenous taste cell ATP-release channel exhibits considerably faster activation kinetics (τ ~ 10 ms) than those of recombinant CALHM1 [46, 74]. In addition, type II cell ATP release, but not heterologously expressed CALHM1 ion currents, is inhibited by carbenoxolone [20, 35, 47, 55]. Together these resulted suggested that whereas CALHM1 was essential for type II cell-mediated ATP release, other components were likely to also play a role. A recent study [48] discovered that a CALHM1 homolog, CALHM3, hetero-oligomerizes with CALHM1 to form a novel voltage-gated ATP-permeable CALHM1/3 channel with fast activation kinetics (τ ~ 10 ms) and sensitivity to carbenoxolone, which match the properties of the type II cell ATP-release channel. Of note, CALHM1/3 has fast gating kinetics comparable with typical voltage-gated K⁺ channels and is currently the only known ATP-release channel that can potentially be activated by action potentials. In mice, CALHM1 and CALHM3 were co-expressed in type II cells, and Calhm3 KO abolished the endogenous type II cell ATP channel current and taste-evoked ATP release and markedly impaired gustatory nerve and behavioral responses to sweet, umami, and bitter tastes. Thus, a CALHM1/3 channel was established as the bona fide ATP-release channel required for rapid neurotransmission in type II cells (Fig. 1c). More recently, CALHM1/3 was also discovered to mediate neurotransmission of sodium taste [60] (Fig. 1c). This identification of an action potential-activated neurotransmitter-release channel establishes the molecular mechanism of how taste cells lacking synaptic vesicles release neurotransmitters. This represents a new mode of chemical synapse unique to taste cells, which we term “channel synapse” in contrast to the conventional “vesicular synapse” (Fig. 1).

Structure of the channel synapse

Recently, super-resolution immunohistochemical analyses identified the localization of CALHM1 in the basolateral membrane of both type II cells and sodium taste cells in discrete puncta localized close to P2X2-expressing afferent nerve fibers [38, 60, 72]. Furthermore, ultrastructural immunohistochemistry revealed that large “atypical” mitochondria are consistently juxtaposed closely to membrane areas accumulating CALHM1 [72]. Pharmacological inhibition of ATP production in mitochondria blocked taste cell ATP release, suggesting that the immediate source of released ATP was the atypical mitochondrion rather than a cytoplasmic pool [72]. The close apposition of the ATP source, ATP-release channel, and ATP receptors defines the structure of the channel synapse, and is thought to enable stable, spatially localized neurotransmitter release in the absence of synaptic vesicles (Fig. 2). This spatial arrangement of the three components is unlikely to be a product of chance and likely involves scaffold proteins, as suggested by the uniform distances between the mitochondrial outer membrane and plasma membrane (20~40 nm) and between a taste cell and a nerve (10~15 nm) [72]. Identification of these scaffold proteins will facilitate understanding of the channel synapse formation.

Structure of CALHM channels

The CALHM channel family comprises six members, CALHM1-6 [24], among which CALHM1, CALHM2, and CALHM3 have been implicated in physiological and pathological processes, including taste neurotransmission [48, 60, 89, 91], neuronal excitability [47], Alzheimer’s disease [24], and depression [45]. Recent studies by single-particle cryo-EM have revealed structures of several CALHM channels and provided insights into how CALHMs function as ion channels [16, 22, 25, 67, 85, 107]. As previously predicted [24, 47, 80, 87, 90], protomers of all CALHM channels harbor four transmembrane (TM) helices and cytoplasmic C-terminal helices (CTH). This membrane topology is reminiscent of other large pore-forming channel proteins such as connexins, innexins, pannexin1, and LRRC8. However, viewed from the extracellular side, the anticlockwise arrangement of TM1, TM2, TM3, and TM4 helices in CALHMs distinguishes them from other large-pore channels, all of which have a clockwise arrangement. CALHMs exhibit various oligomeric assemblies from octamers to dodecamers and consequently different sizes, although the physiological relevance of the diverse oligomeric assemblies is not well understood [16, 22, 25, 67, 85, 107]. Unlike the hexameric assembly previously suggested for human CALHM1 and CALHM1/3, three studies revealed octamer assembly of killifish CALHM1 [22] (Fig. 3a), chicken CALHM1 [85], and zebrafish CALHM1 [67]. These studies suggest that human CALHM1 is also likely an octamer since the residues responsible for subunit interactions are highly conserved in the CALHM1 orthologs [22, 85]. Furthermore, based on the sequence similarity between CALHM1 and CALHM3, the CALHM1/3 channel is speculated to be a hetero-octamer [22]. The pore architecture has been resolved in the highest detail in killifish CALHM1, where the N-terminal helix (NTH) preceding TM1 was successfully visualized [22]. Killifish CALHM1 has a channel pore along a central axis perpendicular to the membrane, and the NTH forms a single constriction site in the pore with the narrowest diameter of 15.7 Å (Fig. 3b). NTH-truncated mutant channels showed enhanced ATP-release activities, supporting the role of the NTH in channel pore architecture [22]. The observed diameter is close to the value functionally estimated for human CALHM1 (~ 14 Å) [80] and is larger than the size of an ATP molecule (~ 12 Å). Human CALHM1 allows permeation of both cations and anions with relative permeabilities of P_Ca:P_Na:P_K:P_Cl = 11:1:1.17:0.56 [47]. The pore-facing residues in the NTH are neutral, which may account for the weak charge selectivity. Thus, the observed pore architecture of killifish CALHM1 is consistent with the functional pore properties of human CALHM1. These results provide important insights into the structure and ATP permeation of the neurotransmitter-release channel in taste cells, CALHM1/3. Nevertheless, the structure of the heteromeric CALHM1/3 awaits resolution. Also, CALHM1 and CALHM3 do not possess canonical voltage-sensor domains, but CALHM1/3 is rapidly activated by depolarization [48]. Further studies are required to identify the structural basis of the voltage-dependent gating of CALHM channels. Finally, although a putative closed-state structure of CALHM2 has been reported [16], the mechanisms involved in closing CALHM1 and CALHM1/3 also await resolution.