Ligand-specific induction of endocytosis in taste receptor cells

The taste system recognizes and discriminates a large number of taste substances; and in fish, these tastants give rise to acceptance or rejection(Kasumyan and Døving,2003). In mammalian taste buds, three different subtypes of elongated epithelial cells and proliferative basal cells have been described(Murray, 1974). One of these types of taste cells possesses synapses (type III cells, presynaptic cells),another expresses chemosensory transduction proteins (type II cells, light cells, taste receptor cells), and the third class expresses none of these properties (type I cells, dark cells)(Clapp et al., 2006; DeFazio et al., 2006; Tomchik et al., 2007; Yee et al., 2001). It has been stated that cells of the fish taste buds have morphological similarities to those of taste buds of rodent fungiform papillae(Murray, 1974); and a similar classification might be valid for fish taste buds(Reutter et al., 2000; Reutter and Hansen, 2005).

Numerous electron microscopy studies have described vesicles, microtubules and filaments in the apical parts of taste bud cells of mammals and fishes(Farbman, 1965; Ganchrow et al., 1993; Reutter and Hansen, 2005; Reutter and Witt, 1993). These elements are essential for endocytosis and transport of endocytic vesicles. Endocytosis was studied in the sensory cells of frog taste disks with multi-photon microscopy (Li and Lindemann,2003), and the authors found a staining of sensory cells after 24 h exposure of taste buds to the styryl dye FM1-43. It has also been reported that FM1-43 rapidly and specifically labels taste buds by a channel permeation mechanism in mice after subcutaneous injection(Meyers et al., 2003).

In the present study, we demonstrate a ligand-specific induction of endocytosis in taste cells. Such a feature has, to our knowledge, not been demonstrated before. We used brown trout because of the abundance of taste buds in the oral cavity (Hara et al.,1994; Marui et al.,1983) and because of knowledge of substances accepted (stimulants)or rejected (deterrents) when presented to brown trout in food pellets(Kasumyan and Sidorov, 2005). The oral cavity was exposed to taste substances in combination with dyes or to dyes alone. Thus, we stimulated the taste buds in a way that is normal for taste perception. As we see a ligand-specific induction of endocytosis, it seems probable that the stained cells were of the `taste receptor cell' type. Fluorescent styryl dyes (FM dyes) have been frequently used as probes for membrane trafficking and endocytic and exocytic activity because of their physical properties giving an enhanced fluorescence in lipids(Betz et al., 1992a; Betz et al., 1992b; Betz et al., 1996; Brumback et al., 2004; Cochilla et al., 1999). There are several forms of endocytosis and although clathrin-dependent uptake of receptor–ligand complexes has received much attention clathrin-independent forms also exist(Mukherjee et al., 1997; Mayor and Pagano, 2007; Sandvig et al., 2008).

In our experiments, dye reached all parts of taste receptor cells rapidly after apparent endocytic uptake and vesicular transport, since low temperature and disruption of microtubules inhibit ligand-induced staining of the taste receptor cells. An additional finding is that taste receptors for a stimulant and a deterrent seem to be expressed mainly in separate taste receptor cells.

Experimental procedures were approved by the Norwegian Animal Research Authority and were conducted in accordance with the Norwegian Animal Welfare Act of 1974, and the Regulation of Animal Experimentation of 1996. Juvenile brown trout Salmo trutta L. (6–10 g, 5.5–10 cm body length) were obtained from a local fish farm (Oslomarkas fiskeriadministrasjon, OFA, Sørkedalen, Oslo, Norway). The fish were kept in the aquaria facilities at the Department of Molecular Biosciences,University of Oslo at a water temperature of 8°C and at 12 h:12 h L:D photoperiod. The fish were fed twice a week with commercial food pellets. Fish were given an intraperitoneal injection of Hypnodil (metomidatum hydrochloridium; Janssen, Pharmaceutica Beerge, Belgium; 60 mg kg^–1). Anesthetized fish were wrapped in wet paper, adjusted in a cradle belly up. For experiments lasting more than 5 min, two catheters supplied tap water irrigation of the gills. In total 96 fish were used for the physiological studies.

Fluorescent dye uptake by taste receptor cells was examined at room temperature (20°C), unless otherwise indicated. Dye solutions were applied to the oral cavity of fish at a rate 1 ml min^–1. To examine the rate of styryl dye entry into taste cells, we bathed the oral cavity for∼10 s, 30 s, 1, 2, 5 and 10 min. In all experiments the test substances were dissolved in distilled water. Immediately after the end of application,the oral cavity was rinsed in tap water, the fish was decapitated and the head fixed in paraformaldehyde (4% in 0.1 mol l^–1 phosphate buffer, pH 7.1). To study the temperature dependence of dye uptake,experiments were repeated at +2°C. Long-range transport of endocytic vesicles are often dependent on molecular motors and microtubules which can be disrupted by nocodazole (Schliwa and Woehlke, 2003) so the effect of nocodazole was studied with a pre-exposure for 20 min at 35 μmol l^–1 at room temperature before exposing the oral cavity to a mixture of FM dye and l-cysteine for 5 min. The dextrans were applied at 10 or 20 μmol l^–1.

FM 1-43 and FM4-64 FX were purchased from Molecular Probes (Eugene, OR,USA). l-cysteine, glycine and nocodazole M1404 were purchased from Sigma-Aldrich (St Louis, MO, USA). AM1-43, the fixable analogue to FM1-43 was purchased from Biotium (Hayward, CA, USA). The following fluorescent dextrans were purchased from Invitrogen (Carlsbad, CA, USA): 3 kDa, Texas Red, lysine fixable D3328; 3 kDa Alexa Fluor 488; anionic D-34682; 3 kDa Alexa Fluor 688,D34681, and 40 kDa fluorescein, anionic lysine fixable D1845.

The upper and lower jaws were mounted in Petri dishes to view taste buds on the epithelium. All photographs were made of taste buds in situ. We standardized our examination to the following regions of the oral cavity: the upper and lower lips, 3 mm from the midline to the left and to the right, the anterior 3 mm of the vomer and the anterior 3 mm of the tongue. Taste buds with stained taste receptor cells were counted using a fluorescence microscope(BX50WI, with a 20× water immersion lens; Olympus, Tokyo, Japan)equipped with a digital camera (ProgRes, Jena, Germany). The software program ImageJ (1.37v, Wayne Rasband, NIH, USA) was used to extract information from photographs taken at different depths, usually 2.5 μm. Images were imported and the stacks added in a z-projection with the max intensity parameter. A confocal microscope (Olympus FluoView 1000, BX61W1) was used to observe the details of the taste receptor cells in the taste buds and the possible overlap or colocalization between taste receptor cells in double staining experiments. Images were taken in planes, separated by z-axis steps varying between 0.4 and 2 μm.

Eight brown trout juveniles were placed in individual aerated aquaria(18°C) and trained to take 2% agar pellets, 3.0 mm in length and 1.3 mm in diameter, prepared with Ponceau 4R (red; 5 μmol l^–1). In trial sessions, test pellets with dextran Texas Red (0.5 mmol l^–1), l-cysteine (100 mmol l^–1)or Chironomidae larvae water extract (75 gl^–1) were offered to the fish at random, one by one at intervals of more than 20 min. In total,141 taste trials were performed. The acceptance ratio was calculated as the number of pellets accepted over the number of pellets taken into the mouth.

The taste structures examined were from the anterior 3 mm of the upper and lower lips (left and right side), the vomer, and the tongue in juvenile brown trout. In these regions, there are a total of about 600 taste buds. The tongue and vomer in brown trout have lobes that make it difficult to see all taste buds. We did not attempt to estimate the number of cells in the different taste buds.

The different dyes when they were applied alone varied greatly in their ability to stain cells. Table 1give the dyes used, the number of cells stained, and the number of taste buds with stained cells. From the table it is evident that fluorescein dextran,Alexa Fluor 488 dextran and Alexa Fluor 680 dextran, stained only a few taste cells. FM1-43 resulted in a moderate number of stained taste cells, whereas the fixable analogue AM1-43 stained more receptor cells than FM1-43. The staining efficiency with the dextrans was dependent upon the conjugated dye. The efficiency of Texas Red and tetramethylrhodamine-conjugated 3 kDa dextran was particularly high, and it is worth noting that more than two taste cells per taste bud were stained.

Exposure of the oral cavity to FM1-43 in distilled water (1 μmol l^–1; 2 min) stained only 40 taste cells in 32 taste buds(Table 1). By contrast,exposing the oral cavity to FM1-43 (1 μmol l^–1) together with the taste stimulant l-cysteine (100 mmol l^–1), stained 175 taste cells in 144 taste buds distributed as follows: the upper lip 31, vomer 51, lower lip 49, and tongue 44.

The number of stained taste cells and taste buds containing stained taste cells increased with the concentration of l-cysteine in the range 0.1–100 mmol l^–1 applied together with 1 μmol l^–1 FM1-43 (Fig. 1). The ratio between taste cells and taste buds in the regions examined varied between 1.2 and 1.4, demonstrating that most taste buds contained only one stained taste cell. The maximum number of stained cells in a single taste bud was 3.

The behavioural threshold for l-cysteine in taste preference tests with agar pellets is about 10 mmol l^–1(Kasumyan and Sidorov, 2005). At this concentration we found 84 taste buds with stained cells. Given the background staining with FM1-43 alone (32 taste buds), it indicates that about 52 taste buds (9%) have to be activated to evoke an appropriate behavioural response to a taste stimulant. At the subthreshold concentration of 1 mmoll^–1, around 27 additional taste buds (5%) were stained,suggesting that these numbers are insufficient for the fish to assess the palatability of the food object and make the decision to swallow it.

Application with a short duration of ∼10 s gave a weak staining of about 20 μm depth of the apical part of taste cells. A 30 s exposure was sufficient to stain most parts of a taste cell(Fig. 2A). It should, however,be noted that it may take some time before the fixative penetrates into the taste bud and reaches the transport processes along the microtubules; thus allowing movements of vesicles for somewhat longer time. With 1, 2, 5 and 10 min exposure before application of fixative, the whole taste cell was stained(Fig. 2).

Fluorescence was distributed throughout the cytosol, and the cell soma was clearly visible. The staining was punctate, i.e. bright spots on a diffuse background (Fig. 2A). The staining made it easy to see the pyriform shape of the taste cells with the apical part being long, slender and curved, and often more intensely staining than the basal part. The cytoplasm around the nucleus was thin. Often the basal part of the cell consisted of a thin protrusion that ended in a small foot (Fig. 3A).

In a single taste bud, with two or more stained cells, these were seldom found close to one another at the nuclear level. The distance between nearest neighbour cell nuclei was around 10 μm.

There are a variety of endocytic mechanisms which have been determined in a number of different tissues (Sandvig and van Deurs, 2005). In mammals, temperatures near 0°C effectively stop all types of endocytosis. In rainbow trout, Oncorhynchus mykiss, and tilapia, Oreochromis nilotica, endocytic activity is strongly reduced at +5°C during the initial lowering the temperature(Padron et al., 2000). We placed the anesthetized fish in iced water, irrigated the gills and exposed the oral cavity to ice cold solutions of 100 mmol l^–1l-cysteine together with 1 μmol l^–1 AM1-43. The temperatures measured in the solution and mouth cavity was +2°C. We observed on the upper jaw (lip and vomer) a considerable reduction from 431 taste cells in 248 taste buds at room temperature to 15 stained taste cells in 14 taste buds at +2°C (Table 2).

Application of nocodazole (see Materials and methods) reduced substantially the number of stained taste cells on the upper lip and vomer from 431 to 111(Table 2). The reduction in the number of stained taste cells in the vomer was less efficient, probably because irrigation of the gills with tap water caused some backflow and rinsing during application of nocodazole. In sum, these results indicate a taste-specific induction of endocytosis, and furthermore that the endocytic vesicles are transported via microtubules toward the taste receptor cell soma.

A single subcutaneous injection of FMl-43 in mice brightly labels hair cells, Merkel cells, muscle spindles, taste buds, enteric neurons and primary sensory neurons within the cranial and dorsal root ganglia(Meyers et al., 2003). The authors provided evidence that this cationic dye is able to pass through a number of different ion channels. Given these studies, we applied dextrans with high molecular mass, which are unlikely to pass through ion channels:Texas Red 3 kDa dextran and FITC 40 kDa dextran, together with 100 mmol l^–1l-cysteine. Both types of dextran stained a substantial number of taste receptor cells(Fig. 3). It is unlikely that these large conjugated dextrans would pass through ion channels. Our data suggest that they are internalized by endocytosis.

The basic knowledge of which substances are stimulants and which deterrents in fish is surprisingly fragmentary(Kasumyan and Døving,2003). As seen from Table 1, Texas Red 3 kDa dextran alone was remarkably efficient in staining taste receptor cells in brown trout. These findings provoked a series of taste preference studies with Texas Red dextran to verify that this substance, which stains taste receptor cells, also evoked a taste sensation in brown trout. The behavioural taste experiments demonstrated that the Texas Red-conjugated dextran at a concentration of 0.5 mmol l^–1evoked an acceptance ratio of 6%. In comparison, control pellets, and pellets with l-cysteine or water extract of Chironomidae larvae (75 g l^–1) gave acceptance ratios of 29, 91 and 94%, respectively(Table 3). These findings demonstrated that pellets with Texas Red dextran (0.5 mmol l^–1) were spat out, and this dye is a deterrent for brown trout at a much lower concentration than is effective for the stimulant, l-cysteine (100 mmol l^–1). It is probable that the dye Texas Red alone is responsible for the aversive effect, since dextrans conjugated with other dyes induced a small number of stained taste receptor cells.

Taste substances that induce different behavioural effects ought to stain different taste cells and/or different taste buds if the staining that we see are cells giving rise to taste perception. We carried out several experiments to investigate this assumption.

From previous behavioural experiments it is known that glycine is a deterrent taste substance for brown trout(Kasumyan and Sidorov, 2005). We applied this substance at a concentration of 100 mmol l^–1together with 1 μmol l^–1 AM1-43 to three fish (for 5 min). In three other fish we applied 100 mmol l^–1l-cysteine together with 1 μmol l^–1 AM1-43 (5 min). In the anterior 3 mm of the lips, tongue and vomer we counted the number of taste buds with stained cells, and how many stained cells there were in each taste bud (Table 4). There were no great differences in distribution in the regions investigated, or in the number of stained cells per taste bud.

Given that there are no gross differences in distribution in the regions investigated, one could ask if there are differences in the small scale. Could it be that stimulant and deterrent substances stained taste cells in the same or in separate taste buds? In order to answer this question, we did several experiments with stimulants and deterrents. First, we exposed the oral cavities of two fish first to 100 mmol l^–1 glycine + 1μmol l^–1 AM1-43, then to 100 mmol l^–1l-cysteine + 20 μmol l^–1 Alexa Fluor 680 dextran for 3 min each solution. Inspection under the fluorescence microscope revealed that in these fish, a total of 324 taste buds contained stained cells. Of these taste buds, 164 were stained yellow with 100 mmol l^–1 glycine (deterrent) + 1 μmol l^–1AM1-43, 98 were stained red with 100 mmol l^–1l-cysteine (stimulant) + 20 μmol l^–1 Alexa Fluor 680 dextran, and 62 taste buds had both yellow and red cells. Thus, the majority of taste buds responded either to glycine or to l-cysteine. However, 19.1% of the taste buds had taste cells that stained either yellow for glycine (deterrent) or red for l-cysteine(stimulant) at these concentrations of the applied substances.

Second, the oral cavity of three fish were exposed, first to 100 mmol l^–1l-cysteine + 1 μmol l^–1AM1-43, then to 100 mmol l^–1 glycine + 2 μmol l^–1 FM4-64 FX each for 3 min. Inspection of the vomer in these fish showed that in total 99 (46.3%) taste buds had only yellow stained cells, 78 (36.4%) had only red stained cells, and 37 (17.3%) had both yellow and red stained cells. Confocal microscopy of some of these taste buds with both yellow and red stained cells demonstrated colocalization of the two dyes in the same cell (Fig. 4).

Third, as the dye Texas Red was found to be a deterrent substance by itself, we also exposed fish first to 20 μmol l^–1 Texas Red, and then to a mixture of the stimulant 100 mmol l^–1l-cysteine together with 20 μmol l^–1 fluorecein 3 kDa dextran for 2 min. Inspection under the fluorescence microscope revealed that in three fish, a total of 281 taste buds contained stained cells. Of these taste buds, 210 had only Texas Red staining, 62 had only fluorescein staining, and nine taste buds had both yellow and red cells. Thus, only 3.2%of the taste buds had taste receptor cells that stained both for l-cysteine (stimulant) and Texas Red dextran (deterrent) at these concentrations of the applied substances(Table 5).

The most important finding in the present study is that there is prominent ligand-specific induced endocytic activity in taste cells of brown trout. The endocytosis is dependent upon the taste substance being present at the taste bud pore. Since the endocytic activity can be identified with styryl dyes that have a more efficient fluorescence in lipid than in water solution, the procedure of including the taste substance together with a dye is an efficient way of labelling activated taste cells. However, all substances present at the taste bud pore could be included in the endocytic vesicles as high molecular mass dextrans with dyes were found to stain the cells. Use of stimulant and deterrent substances revealed that the majority of taste buds had taste cells that stained for either one or the other of the two substances. However, some taste buds had cells that stained for both stimulant and deterrent substances.

In the vertebrate taste bud there are three types of epithelial cells. Several arguments suggest that the cells that stained in our experiments were the so-called receptor cells, or type II cells, following the nomenclature of Murray (Murray, 1974). These cells have the taste receptors and the molecular machinery for downstream processing and thus express the elements of the taste transduction cascade.

There is a vast literature on electron microscopy studies of taste buds in vertebrates, and also taste buds in fishes have been described in great detail(Reutter and Witt, 1993). These electron microscopic examinations indicate that the taste cells contain elements such as vesicles and microtubules that could indicate endocytic processes (Ezeasor, 1982; Fujimoto et al., 1993). Also different types of filaments or filament-associated structures have been found in taste cells, such as actin, espin and keratin(Hofer and Drenckhahn, 1999; Sekerkova et al., 2004; Takeda et al., 1990). It seems reasonable to assume that a similar division might apply also to taste buds in fish (Finger, 2005). Also, the taste receptor cells have a large, round nucleus(Finger, 2008), a characteristic feature of the cells that stained in the present study.

Our data demonstrate that ligands are essential for staining of cells, and one dye (Texas Red) that was particular efficient when tested as control,turned out to be a taste deterrent. The number of stained cells increased with increasing concentration of ligand. These results also support the assumption that the stained cells are related to taste perception.

Finally, our results demonstrate that the majority of taste buds had cells that respond to either stimulants or deterrents, and both types of taste buds are evenly distributed throughout the fish mouth. Thus, the majority of taste cells either expressed receptors for taste stimulants or deterrents, which is a prerequisite if the substances should induce different behavioural effects,i.e. acceptance and rejection, respectively. The fact that the majority of these cells were found in different taste buds, might be related to the species and sets of substances that we used. Ishimaru et al. and Oike et al. examined the types of taste receptors (T1R and T2R) expressed in the taste buds of zebrafish and medaka (Ishimaru et al., 2005; Oike et al.,2007). They suggested that, among vertebrates, attractive tastants are recognized by T1Rs and aversive tastants by T2Rs. They showed that although the T1Rs and T2Rs are expressed in different taste cells, they generally are co-expressed in the same taste bud. These results are in contrast to the present findings in brown trout, showing that stimulants and deterrents, in general, stained receptors cells in different taste buds. By combining the ligand-specific endocytosis with in situ hybridization techniques in the same species, it should be possible to verify these hypotheses.

The results described here support the surprising finding that there is a massive taste-specific induction of endocytosis in receptor cells of taste buds: there is a substantial reduction of dye uptake at low temperature similar to the reduction of endocytosis in other cell types(Sandvig and Olsnes, 1979). There is also a reduction of dye uptake and staining of taste receptor cells after exposure to nocodazole, which disrupts the microtubules needed for transport of endocytic vesicles. There is a rapid, nocodazole-sensitive transport of vesicles with dye into the basal part of the taste receptor cells, and the uptake is concentration dependent.

Li and Lindemann (Li and Lindemann,2003) showed that in the taste receptor cells of the frog taste disks there is an uptake of FM dyes by endocytosis after 24 h exposure;although without the use of taste substances. In mice, a single subcutaneous injection of FM1-43 brightly labels hair cells, Merkel cells, muscle spindles,taste buds, enteric neurons and primary sensory neurons within the cranial and dorsal root ganglia (Meyers et al.,2003). The authors presented evidence that entry of dye into the hair cells was via transduction channels. However, the possibility that the entry into taste cells was by endocytosis was not discussed. In the present experiments, 3 and 40 kDa dextrans together with l-cysteine were used and extensive staining of the taste receptor cells was found. As the 40 kDa dextran in particular is so large that it would hardly enter the cells through non-selective ion channels, it supports our conclusion that the internalization of stain in taste receptor cells of taste buds is viaendocytosis. The specific induction of endocytosis we describe is initiated by the taste substance presented in the medium applied on the taste bud pore. The use of FM1-43 alone stained a modest number of taste receptor cells. We propose the following scenario as a possible mechanism: the ligand binds to a taste receptor and induces a series of transduction events that finally results in exocytosis of transmitter substance at the basal part of the cell. This process is accompanied by internalization apically, probably of the receptor with ligand. Whether also vesicles without receptors are formed is an open question. There could be induction of different types of endocytosis similar to those previously described at the apical pole of polarized MDCK cells (Sandvig and van Deurs,2005). Vesicles formed are transported via microtubules towards the cell soma. The fate of the internalized membrane will be investigated in future studies.

The successive application of stimulants and deterrents, substances that evoke different feeding behaviours in brown trout, demonstrated selective staining of taste receptor cells in the taste buds. It is important to note that the majority of stained taste receptor cells were in separate taste buds. We seldom observed doubled labelled taste buds, but in some cases confocal images showed co-staining in single receptor cells. A possible conclusion from these observations would be that most taste receptor cells have taste receptors either for a stimulant or a deterrent.

The prominent endocytosis in taste cells is a new observation and there are a multitude of questions related to this process. The massive endocytosis in the peripheral part of the taste cells would require a recycling of the membrane. At present, we do not know how this is achieved. What could be the physiological role of this endocytosis? Could it be that receptor proteins are internalized to reduce that taste perception? Are the receptors recycled at the apical part of the cell? What is the fate of ligands? Is the endocytosis related to adaptation of the physiological response from taste nerve fibres?To what extent will the internalization of activated receptor proteins regulate the homeostasis of this sensory system? These questions will be addressed in future studies.

In other chemosensory cells, as for example the sensory neurons in the olfactory and vomeronasal organs of mice, the structure of the endocytic machinery has been studied and demonstrated to be an uptake of ferritin molecules (Bannister and Dodson,1992). In isolated olfactory sensory neurons of catfish,clathrin-dependent endocytosis was reported to occur after perfusion with l-glutamate together with FM1-43, but not with a mixture of l-alanine, l-arginine, l-cysteine and FM1-43(Rankin et al., 1999). It seems, however, probable that ligand-specific processes similar to those shown by the present in situ study, is common for all chemosensory structures in vertebrates.