Synaptic communication and signal processing among sensory cells in taste buds

Abstract

Taste buds (sensory structures embedded in oral epithelium) show a remarkable diversity of transmitters synthesized and secreted locally. The known transmitters accumulate in a cell type selective manner, with 5-HT and noradrenaline being limited to presynaptic cells, GABA being synthesized in both presynaptic and glial-like cells, and acetylcholine and ATP used for signalling by receptor cells. Each of these transmitters participates in local negative or positive feedback circuits that target particular cell types. Overall, the role of ATP is the best elucidated. ATP serves as a principal afferent transmitter, and also is the key trigger for autocrine positive feedback and paracrine circuits that result in potentiation (via adenosine) or inhibition (via GABA or 5-HT). While many of the cellular receptors and mechanisms for these circuits are known, their impact on sensory detection and perception remains to be elaborated in most instances. This brief review examines what is known, and some of the open questions and controversies surrounding the transmitters and circuits of the taste periphery.

Nearly half a century ago, light and electron microscopic studies revealed that taste buds are aggregates of 50–80 bipolar cells lodged in a non-sensory oral epithelium. In contrast, the chemosensory receptors for smell (olfactory sensory neurons) exhibit a loosely distributed pattern. Hence, the question was raised early on: Why are the chemosensory cells for taste aggregated in this fashion, and what functional advantage might accrue from this tight clustering? The answer suggested by a decade of research may be that taste cells exhibit numerous types of paracrine interactions, employing a host of extracellular signalling molecules and receptors, and that these circuits process the afferent sensory signal.

Taste buds and their constituent cells

Cells comprising the taste bud (Fig.1) are quite disparate in their morphologies, in their patterns of gene expression, in their sensitivities to tastants and in other functional properties (Chaudhari & Roper, 2010). The most prevalent of the cells are the so-called type I or glial-like cells that possess thin cytoplasmic wing-like extensions that ensheath the other taste bud cells (Murray, 1993; Pumplin et al. 1997). Their glial morphology is complemented by membrane proteins that orchestrate the reuptake or degradation of several neurotransmitters. Next in order of prevalence are the type II or receptor cells, which serve as chemosensory detectors of sweet, bitter and umami taste stimuli (Chandrashekar et al. 2006). The type III or presynaptic cells are the most neuron-like within the bud, containing clear and dense-cored vesicles clustered near synaptic contacts with nerve fibres (Yee et al. 2001). In addition, the type III cells appear to be the detectors for sour stimuli (Huang et al. 2006, 2008; Chang et al. 2010). Both receptor and presynaptic cells are excitable. Relative to the other taste qualities, we know very little about the identity of salt transducing cells. As little as a decade ago, many authors referred to all cells of the taste bud as ‘taste receptor cells’. It is now abundantly clear only a limited subset, amounting to half or fewer of all cells are indeed receptors.

Figure 1. The three principal constituent cell types in taste buds.

A, confocal micrograph of a taste bud from a Plcb2-GFP mouse. Receptor (type II) cells express GFP (but here, are pseudo-coloured yellow) while presynaptic (type III) cells, immunostained for aromatic amino acid decarboxylase, a 5-HT-synthesizing enzyme, appear green. The dark spaces between the labelled cells are occupied by glial-like (type I) cells that ensheath all the other cells, but are not visualized here. The taste bud resides in oral epithelium (dashed lines). Adapted with permission from J Neurosci (DeFazio, Dvoryanchikov et al. 2006). B, same taste bud, schematized to demonstrate receptor and presynaptic cells in close proximity. The taste bud has a dotted outline while the epithelium is indicated by a dashed outline.

Taste transmitters

The first transmitter to be identified as an afferent transmitter for taste was ATP. Evidence for the afferent role of ATP included the presence of ATP-gated channels, P2X2 and P2X3, in taste bud-associated fibres (Bo et al. 1999), and the loss of taste function in knockout mice lacking such ATP receptors. (Finger et al. 2005). Subsequent studies showed, however, that in addition to lacking postsynaptic responses in taste afferents, taste buds in P2X2/P2X3 double knockout mice do not secrete ATP (Huang et al. 2011), complicating the original interpretation. In addition to ATP, taste stimuli evoke the secretion of a number of additional transmitters, including 5-HT, noradrenaline, acetylcholine and GABA under different circumstances. Receptors for many of these transmitters are found on selected subsets of taste bud cells, and a growing body of evidence suggests extensive paracrine interactions among the bud cells. In this review, I outline a few examples of how such cell–cell interactions may modulate the primary evoked signal and serve to shape the afferent output.

ATP and 5-HT: transmitters secreted by different cells

The secretion of transmitters from taste buds has been demonstrated using bulk assays (Finger et al. 2005; Geraedts & Munger, 2013) as well as using transmitter-detecting biosensors, which are highly sensitive and permit cellular resolution (Huang et al. 2005). The biosensors are cells transfected with specific receptors, then used as probes to test transmitter release via Ca²⁺ imaging. ATP biosensor cells, sensitive enough to respond to ≤300 nm ATP, demonstrated unequivocally that taste buds consistently secreted ATP when stimulated with common bitter and/or sweet tastants (Huang et al. 2007). When such taste buds were dissociated into constituent cells and assessed with biosensors, it became clear that individual receptor (type II) cells are the source of the ATP secreted. The observation was consistent with the known expression of G protein coupled taste receptors for sweet, bitter and umami tastants in these receptor cells (Chandrashekar et al. 2006; Chaudhari & Roper, 2010).

Sweet, bitter and umami tastants stimulate Ca²⁺ elevation and the non-vesicular secretion of ATP from receptor cells. Initial steps include the phospholipase C-mediated mobilization of Ca²⁺ and the consequent generation of a receptor potential via the TrpM5 cation channel (Liu & Liman, 2003). The details of the mechanism by which ATP is released continue to be clarified. ATP appears to be released from the cytosol by passing through large pore channels. Both connexins and pannexin 1 (Panx1) were proposed as the conduit (Huang et al. 2007; Romanov et al. 2007). Because Panx1 gating is sensitive to both Ca²⁺ concentration and voltage (Locovei et al. 2006), it is ideally suited to integrate the Ca²⁺ elevation and Trpm5-mediated depolarization signals of transduction. A more recent proposal is that the Calhm1 channel is responsible for non-vesicular ATP release (Taruno et al. 2013), but this channel is dependent on strong depolarizations, insensitive to intracellular Ca²⁺ and appears to require very low extracellular Ca²⁺. Low concentrations of carbenoloxone, an inhibitor of Panx1 (but not Calhm1) channels, was reported to depress ATP release from individual receptor cells (Huang et al. 2007; Murata et al. 2010). A less sensitive assay of ATP secretion from epithelial sheets (Taruno et al. 2013) was insufficient to diagnose which of the channels permeated ATP. On the other hand, taste buds from Panx1 knockout mice were reported to exhibit taste-evoked release of ATP (Romanov et al. 2012), while Calhm1 knockout mice appeared to have depressed ATP release (Taruno et al. 2013). In summary, ATP release from taste cells is not fully explained by the properties of any single channel proposed to date. Further experiments will be needed to resolve the relative roles of connexins, Panx1, Calhm1 and perhaps additional conduits for ATP release from taste cells.

An incompletely resolved question is the extent to which ATP secretion from receptor cells is dependent on action potentials. By recording with a suction electrode, Murata et al. (2010) showed that the quantity of ATP secreted from individual receptor cells matched firing frequency. In contrast, Huang and Roper (2010) reported that taste-evoked ATP secretion measured via biosensors was unperturbed in the presence of tetrodotoxin, and confirmed the cooperative effect of intracellular Ca²⁺ and strong depolarization.

The afferent transmitter used downstream of sour and salty transduction is less worked out than for sweet, bitter and umami tastants that employ G protein coupled taste receptors. At present, there is little evidence of a mechanism linking ATP to afferent transmission for salty or sour taste.

Presynaptic (type III) cells have long been known to synthesize and package 5-HT (Yee et al. 2001) and more recently, capacitance measurements documented vesicle fusion following depolarization (Vandenbeuch et al. 2010b). The biosensor assay showed that individual presynaptic cells secrete 5-HT when depolarized or stimulated with sour tastants (Huang et al. 2007). Curiously, 5-HT could also be detected following stimulation with sweet and/or bitter stimuli, but only when the presynaptic cells were embedded within intact taste buds. That is, 5-HT release evoked by sweet and bitter tastants is secondary to tastant-evoked ATP from receptor cells. Presynaptic cells appear to contain two separate pools of 5-HT released at different subcellular locations and following distinct pathways of Ca²⁺ elevation (Huang et al. 2005).

Negative and positive feedback within the taste bud

Both 5-HT and ATP have prominent effects within the bud (Fig.2). Receptor cells express membrane receptors for 5-HT and stimulating these receptors strongly inhibits ATP release evoked by tastants (Huang et al. 2009). Furthermore, pharmacologically blocking 5-HT receptors elevated taste-evoked ATP secretion, while prolonging the presence of extracellular 5-HT with reuptake inhibitors dramatically decreased ATP output (Huang et al. 2009). This demonstrated the endogenous function of a negative feedback circuit mediated by 5-HT. Yet, it has remained unclear how such regulation influences afferent signalling from the taste bud. As in other sensory systems, a negative feedback loop may be used to enhance contrast in the presence of taste mixtures, or might result in temporal reshaping of the afferent signal to enhance adaptation. Testing such models of peripheral processing will necessitate single unit recordings downstream of the taste bud, with pharmacological manipulations at the level of the bud itself.

Figure 2. Schematic of the three principal taste cell types and associated transmitters.

Cell types are depicted in the same colour scheme as in Fig.1. The transmitters secreted from each cell type are indicated in the corresponding colour. Solid arrows indicate the source and target of each transmitter according to published experimental evidence. For instance, ATP (yellow) from receptor cells activates afferent fibres. Receptor cells also secrete ACh, while ado is generated extracellularly from secreted ATP. Both ACh and ado act on receptor cells. Where direct evidence is lacking, plausible targets are indicated by dashed arrows. For example, GABA accumulates in both type I and III cells; its action on afferent fibres is unreported to date. One transmitter, glu, is secreted from nerve terminals and acts on presynaptic cells as discussed in the text. ACh, acetylcholine; ado, adenosine; glu, glutamate.

ATP produces a positive feedback loop within the taste bud (Fig.2). The initial observation was that many mature taste cells elevate Ca²⁺ in response to ATP stimulation (Baryshnikov et al. 2003; Hayato et al. 2007). Thus, the small Ca²⁺ signal of taste transduction leads to ATP secretion, which then potentiates both intracellular Ca²⁺ signal and the magnitude of ATP release. While P2Y purinoceptors are prominent on the surface of taste cells (Baryshnikov et al. 2003), P2X receptors have been shown to be critical for the afferent ATP-dependent positive feedback circuit (Huang et al. 2011). Curiously, the ATP-enhanced ATP release is mediated by the same P2X2 receptors on receptor cells as are found on afferent neurons (Huang et al. 2011). Genetic ablation of these P2X2 receptors eliminated ATP secretion under physiological conditions, but transmitter secretion could be rescued if taste stimuli were paired with strong depolarization, effectively mimicking the depolarization produced by cationic currents through P2X2. This autocrine positive feedback loop may enhance ATP secretion and afferent activity above background levels even for weak stimuli. Such a mechanism may serve to tune the sensitivity of the system, for instance permitting detection of particular compounds as nutrition dictates. Such positive feedback may be balanced by negative feedback loops that reduce ATP release within the taste bud (Fig.2).

In addition to the well-studied feedback pathways discussed above, additional transmitters endogenous to the taste bud also appear to regulate the afferent output. For instance, acetylcholine functions as a positive autocrine regulator: it is secreted from receptor cells and enhances the release of the afferent transmitter, ATP (Dando & Roper, 2012). GABA, on the other hand, is secreted within the taste bud from either presynaptic or glial-like cells, or both, and reduces the ATP output of the taste bud (Dvoryanchikov et al. 2011). Finally, glutamate, released from apparently bidirectional nerve fibres (i.e. serving both afferent and modulatory efferent functions, see Fig.2) has been shown to target presynaptic (type III) cells (Vandenbeuch et al. 2010a). Glutamate activation of these type III cells then leads to secretion of 5-HT and consequent inhibition of taste-evoked release of the afferent transmitter, ATP (Huang et al. 2012). The physiologic trigger for glutamate release from afferent axons is unexplored. One possibility is that afferent activity elicited from one taste bud may propagate antidromically in axon branches innervating nearby taste buds and serve to reduce their output.

The impact of 5-HT, noradrenaline, glutamate and GABA on the afferent ATP signal was demonstrated primarily using taste mixtures as stimuli. What remains to be examined at a cellular level is whether particular taste qualities are selectively effective at triggering the release of each transmitter, and thus, whether the corresponding feedback circuit might modulate sensitivity or discrimination. Interestingly, manipulating peripheral levels of noradrenaline and serotonin in people were reported to alter specific taste detection thresholds in human subjects (Heath et al. 2006) but the peripheral mechanism underlying this observation was not elucidated.

Adenosine impact on sweet taste

One transmitter that does modify taste in a quality-selective manner is adenosine. The principal transmitter, ATP, is catabolically cleared via the action of several sequential enzymes, beginning with the ecto-ATPase, NTPDase2, localized to the plasma membranes of glial-like cells that ensheath ATP-secreting receptor cells (Bartel et al. 2006; Vandenbeuch et al. 2013). The final dephosphorylated nucleoside in this catabolic pathway, adenosine, is generated in a restricted pattern based on the expression of additional enzymes, NT5E and PAP (Dando et al. 2012). Adora2b receptors for adenosine are found in the taste bud, selectively on sweet-sensing receptor cells and dramatically enhance the afferent output in response to sweet taste stimuli (Dando et al. 2012; Kataoka et al. 2012). This adenosine-mediated potentiation, selective for sweet taste, is endogenously active as shown by decreased responses to only sweet stimuli following either pharmacological block or genetic loss of Adora2b (Dando et al. 2012; Kataoka et al. 2012). Because adenosine in the circulation fluctuates diurnally and under other conditions, one wonders if sweet taste sensitivity may be regulated by systemic adenosine in addition to the local source we have shown.

Concluding remarks

The role of ATP as a primary afferent transmitter for sweet, bitter and umami taste is clear. It is also clear that numerous other transmitters, and ATP itself, are used as local paracrine and autocrine signals many of which can participate in a dynamic interplay between the different cell types in taste buds (Fig.2). These feedback circuits are ideally organized to modulate the afferent output and shape the neural signal. Such reshaping may be temporal (e.g. producing adaptation) or may selectively enhance one component of a mixture (e.g. lateral inhibition). Reduction of bitter taste in the presence of sweet (such as in sweetened coffee) and salty taste in the presence of sour may well reflect peripheral interactions that employ the negative feedback circuits described here.

Current research offers only limited suggestions of such roles for peripheral modulators of taste signals. Could aversive (e.g. bitter) tastes be reduced through pharmacological manipulation of the peripheral modulatory circuits? Can the perceived intensity of a preferred stimulus (e.g. sucrose) be increased to permit the use of lower concentrations? Might the unpleasant taste side effects of certain drugs be caused by unintended manipulation of these peripheral circuits? How is sensory specificity maintained if well-tuned primary responses trigger converging signals in many cells? In addition, what other taste transmitters (e.g. neuropeptides, Geraedts & Munger, 2013) are released during gustatory sensing? Such questions will be fertile ground for future experimental work.

Biography

Nirupa Chaudhari, Ph.D., is a Professor of Physiology and Biophysics at the University of Miami Miller School of Medicine and also is the Director of the Graduate Program in Neurosciences. For the last two decades, her research interests have spanned molecular and physiological aspects of sensory transduction and coding, cell-to-cell communication, cellular diversity and turnover in the peripheral taste system of mammals. Methods have includes highly sensitive gene expression analyses down to the single cell level, transgenic mouse models, and high resolution imaging for both morphological and functional studies. Collaborative studies have included the use of cellular biosensors to detect secreted neurotransmitters.

Additional information

Competing interests

None declared.

Author contributions

N.C. drafted and edited this review article.

Funding

Supported by a grant R01DC006308 from NIH/NIDCD