Effect of extracellular Ca²⁺ on the quinine-activated current of bullfrog taste receptor cells

Abstract

1. The bitter substance quinine activates a cation current from the frog taste receptor cell. We have analysed the noise associated with this current, and the effect of extracellular Ca²⁺ on the current, using whole-cell recording on single dissociated cells.

2. Quinine induced an inward current from the taste receptor cell near the resting potential. The response was accompanied by an increase in current fluctuations. From the variance/mean ratio of the quinine-activated current, the single-channel conductance was estimated to be 12 pS in the nominal absence of extracellular Ca²⁺. In the presence of 1.8 mM Ca²⁺, this conductance decreased to 5 pS. These values broadly agree with those previously obtained from excised, outside-out membrane patches.

3. The dependence of the current on quinine concentration had a K_1/2 of 0.48 mM in the absence of extracellular Ca²⁺, consistent with measurements from excised patches. The K_1/2 value increased to 2.8 mM in 1.8 mM external Ca²⁺. The maximum current induced by quinine was also reduced by about 20% by Ca²⁺.

4. The spectral power density distribution of the quinine-activated current could be described by the sum of two Lorentzian functions, with corner frequencies not substantially different in the absence and presence of 1.8 mM external Ca²⁺.

5. The above results lend further support to the notion that the major component of the response of frog taste receptor cells to quinine comes from an ion channel directly activated by quinine.

In a previous study, we found that the bitter substance quinine induces an inward current in bullfrog taste receptor cells by opening a non-selective cation channel (Tsunenari et al. 1996). Quinine also opens an ion channel on an outside-out membrane patch excised from the same cells, in the absence of cyclic nucleotide, inositol 1,4,5-trisphosphate, Ca²⁺, ATP and GTP on both sides of the membrane (Tsunenari et al. 1999). These observations suggest that quinine stimulates the taste receptor cell by directly opening an ion channel.

In this paper, we describe experiments in which noise analysis was used to study the quinine-induced current in intact cells, in order to compare this more closely with the current observed in excised patches. At the same time, we have examined further the action of external Ca²⁺, which based on our previous experiments has an inhibitory effect on the channel (Tsunenari et al. 1996, 1999).

Preliminary descriptions of these results have been published (Tsunenari & Kaneko, 1997, 1998).

METHODS

Cell isolation

Taste receptor cells were enzymatically isolated from the tongue of bullfrogs (Rana catesbeiana). The care of the animals was in accordance with the Guidelines for the Care and Use of Laboratory Animals of Keio University School of Medicine. The method of cell isolation has previously been described (Tsunenari et al. 1996, modified from the methods of Avenet & Lindemann, 1987, and Miyamoto et al. 1988). Briefly, the animal was chilled on ice (0°C) and killed by double pithing. The tongue was removed and the fungiform papillae dissected out under a microscope. Harvested papillae (approximately 100) were incubated at 27°C for 25 min in a Ca²⁺-free solution to which 2 mm EGTA had been added, followed by a second incubation for 25 min in normal saline solution supplemented with 0.2% collagenase-dispase (Boehringer). After being rinsed twice with Ca²⁺-free solution (no EGTA), the tissue was triturated in normal saline solution with a glass pipette (tip-opening diameter, 0.2 mm). Dissociated cells were stored at 4°C for up to 10 h in normal saline, and plated onto the recording chamber just before recording. The bottom of the chamber was made of a concanavalin A-coated coverslip. Dissociation yielded a mixture of different-shaped cells found in the fungiform papillae, but taste receptor cells were easily identified by their characteristic morphology (Avenet & Lindemann, 1987; Miyamoto et al. 1988). Voltage-dependent Na⁺ current was also used for identification of cell type (Avenet & Lindemann, 1987; Miyamoto et al. 1988; Tsunenari et al. 1996). In some experiments, we stimulated spherically shaped supporting cells of the fungiform papillae, which can be distinguished from taste receptor cells by their spherical shape, by bath superfusion with 7 mm quinine. In five tested cells, no evidence of a quinine-activated conductance was found. This observation was consistent with a previous report with puff application of quinine (Tsunenari et al. 1996).

Solutions

The normal saline solution contained (mm): 115 NaCl, 2.5 KCl, 1.8 CaCl₂, 2 Na-Hepes and 2 glucose, pH 7.2. The Ca²⁺-free solution lacked CaCl₂, but no EGTA was added. The Mg²⁺ solution contained 1.8 mm MgCl₂ instead of 1.8 mm CaCl₂. The patch-pipette solution contained (mm): 115 CsCl, 2 MgCl₂, 0.5 CaCl₂, 2 Na-EGTA (ca 100 nm free Ca²⁺), 10 Na-Hepes and 2 Na₂-ATP, pH 7.2. All solutions contained 0.0005% (w/v) phenol red.

Application of the taste stimulus

Quinine hydrochloride (Q-1125; Sigma Chemical Co.) was used as a bitter tastant. It was dissolved in bath solution at 0.1-10 mm, and applied by bath superfusion. The solution outlet consisted of a glass capillary with a 1.2 mm tip diameter, positioned within 100 μm of the recorded cell. For noise-analysis experiments, no air gap was introduced between the control solution and the test solution flowing in the capillary (1 ml min⁻¹). Because the boundary between the two solutions in the tube became blurred as the solutions flowed from the valve to the experimental chamber (50 cm away), the solution change at the recorded cell was not instantaneous (approximately 20 s for a complete change). For dose-response experiments, a small air gap was introduced between the control solution and the test solution in order to avoid solution mixing, and the flow was considerably higher (3 ml min⁻¹). In this condition, a solution change at the recorded cell took less than 1 s to complete.

Electrophysiology

Phase-contrast optics were used on a Nikon TMD inverted microscope. Whole-cell patch-clamp recordings were performed at room temperature (20-26°C). All voltage values have been corrected for liquid junction potentials, which were measured using patch pipettes containing 3 m KCl (Neher, 1992). Membrane current was recorded with a DC to 10 kHz bandwidth and stored on tape. For noise analysis, the tape playback signal was both low-pass filtered at 1 kHz (8-pole Butterworth) for measuring the mean current and band-pass filtered at 1 Hz to 1 kHz for measuring variance. Both signals were digitized at 2 kHz in 500 ms segments by a 12-bit A/D converter.

For calculating spectral power densities, the band-passed data were digitized at 2 kHz in segments of 1024 points (512 ms). The power density spectra were averaged over 40 segments. Also, the power densities were averaged as follows: at 32.2-90.8 Hz, the values at two consecutive frequency points were averaged; at 96.7-214 Hz, four points were averaged; at 226-460 Hz, eight points were averaged; above 460 Hz, 16 points were averaged. Below 32.2 Hz, the points were not averaged. Background noise spectra in the absence of quinine were similarly calculated, and subtracted from those in the presence of quinine.

RESULTS

Single-channel current

Whole-cell recordings under voltage clamp were made from isolated bullfrog taste receptor cells, while the bitter substance quinine was applied by bath superfusion. Because quinine suppresses voltage-dependent K⁺ channels of taste receptor cells (Avenet & Lindemann, 1987; Kinnamon & Roper, 1988; Sugimoto & Teeter, 1991; Chen & Herness, 1997), a pipette solution containing 115 mm CsCl was used, which blocks the K⁺ currents (Tsunenari et al. 1996) and thus minimizes the contribution from this action of quinine. An application of 2 mm quinine induced a sustained inward current (80-280 pA, 6 cells) at a holding potential of -64 mV, which is near the resting potential of these cells (-77 mV; Tsunenari et al. 1996). In Fig. 1A, the current response gradually increased to 160 pA upon quinine application (see Methods). The appearance of the current was accompanied by an increase in current noise (see Fig. 1B for a band-passed trace of the same data). In the rising phase of the current, at least up to 100 pA, the variance (σ²) of the current fluctuations increased linearly with the mean current amplitude (I) (Fig. 1c), as would be expected from a low open probability for the channel. Under these conditions, we have (Stevens, 1972):

1

Figure 1. Whole-cell current response induced by quinine application.

A, current trace of the quinine-induced response. Current signal was low-pass filtered at 1 kHz. The cell was stimulated by 2 mm quinine (horizontal bar) in normal saline bath solution. Holding potential was -64 mV. B, current fluctuation obtained from the same current signal as in A. Current signal was band-pass filtered at 1 Hz to 1 kHz. The current scale is magnified by five times that in A. The peaks labelled with asterisks in A and B indicate an artifact caused by the manipulation for the change of the bath solution. These parts were not included in the noise analysis. C, variance of current fluctuations as a function of mean current induced by quinine stimulation. Plotted data were taken from the initial phase of records in A and B when mean current was < 100 pA. Current variance and mean current recorded without quinine were subtracted from data of each plot. The slope of the relationship gives the current amplitude of a single channel and has a value of 0.34 pA. The regression line was fitted by the least-squares method.

where i represents the single-channel current amplitude. The linear slope of the plot in Fig. 1C gives a single-channel current of 0.34 pA. From nine experiments, the single-channel current at -64 mV was 0.36 ± 0.03 pA (mean ±s.d.).

To obtain the current-voltage (I-V) relationship of the single-channel current, we repeated the above measurements at various holding potentials (Fig. 2A). The I-V relationship of the single-channel current obtained in this way from two cells was approximately linear, with a reversal potential of +8 mV (Fig. 2B). The single-channel conductance was 5 pS. The non-zero reversal potential reflects an asymmetry of the ion concentrations inside and outside the cell.

Figure 2. Voltage dependence of the single-channel current amplitude.

Cells were bathed in normal saline solution and were stimulated with 2 mm quinine. A, mean-variance relationships of the quinine-induced currents recorded with various holding potentials (V_hold). These data were obtained from one cell. Plotted data were taken in the same way as in Fig. 1C. B, I-V relationship of the single-channel current amplitude. Plotted data were pooled from two cells. Filled circles represent data replotted from A. The regression line had a slope of 5.1 pS. Reversal potential was +8 mV.

Effect of external Ca²⁺ on the single-channel current

The above single-channel current was measured in the presence of 1.8 mm extracellular Ca²⁺. In the nominal absence of external Ca²⁺, the single-channel current amplitude became twice as large, being 0.85 ± 0.12 pA at -64 mV (mean ±s.d., 9 cells; Fig. 3). The corresponding single-channel conductance was 12 pS.

Figure 3. Effect of external Ca²⁺ on the single-channel current amplitude of the response to quinine.

A, current response recorded in normal saline solution containing 1.8 mm Ca²⁺. Current signal in the upper trace was low-pass filtered at 1 kHz. Current signal in the lower trace was band-pass filtered at 1 Hz to 1 kHz. The cell was voltage clamped at -64 mV and stimulated with 2 mm quinine. These current traces were recorded from the same cell as in Fig. 1. B, current response recorded in Ca²⁺-free solution (no added EGTA). The same cell that was recorded in A was stimulated with 0.2 mm quinine. Filtering protocols of current signals were same as those in A. C, mean-variance plots taken from the initial phase of the current responses in A (1.8 mm Ca²⁺) and B (0 mm Ca²⁺). Indicated slopes (in pA) are estimated single-channel current amplitudes. D, comparison of the single-channel current amplitudes estimated in normal saline solution and in Ca²⁺-free solution. The bars show the single-channel current amplitude obtained from mean values of the slopes in the mean-variance relationships. Current responses for the mean-variance plots were recorded from 9 cells stimulated with 2 or 4 mm quinine in normal saline solution (1.8 mm Ca²⁺) and from 9 cells stimulated with 0.2-2 mm quinine in Ca²⁺-free solution (0 mm Ca²⁺). Error bars indicate s.d.

External Mg²⁺ had a similar, but smaller inhibitory effect. The single-channel current in the presence of 1.8 mm MgCl₂ instead of CaCl₂ was 0.46 and 0.48 pA from two experiments.

Dose dependence of the quinine response and the effect of external Ca²⁺

We have previously shown that the quinine-induced current is enhanced by as much as 6-fold in a nominally Ca²⁺-free extracellular solution (Tsunenari et al. 1996). To examine this question further, we studied the effect of external Ca²⁺ on the relationship between current and quinine concentration. In the presence of 1.8 mm Ca²⁺ (Fig. 4A), the current activation occurred at higher quinine concentrations and the maximum current was also smaller than that in Ca²⁺-free solution (Fig. 4b). From five experiments, all performed at a holding potential of -64 mV, the half-activating quinine concentration (K_½) was 0.48 mm in nominally zero external Ca²⁺ (Fig. 4b) and 2.8 mm in the presence of 1.8 mm external Ca²⁺ (Fig. 4A). The dose-response relationship was sigmoidal, with a Hill coefficient of 3.2 without Ca²⁺ and 4.5 with 1.8 mm Ca²⁺. Thus, external Ca²⁺ decreases the quinine sensitivity of the channel by almost an order of magnitude. The decrease in maximum current due to Ca²⁺ was by about 20%. This percentage decrease is smaller than the percentage decrease in the single-channel current due to Ca²⁺ described above, and may imply a change in the open probability of the channel caused by Ca²⁺ as well. Because of an initial relaxation of the quinine-induced current in the presence of Ca²⁺, however, the reduction in maximum macroscopic current may be more complex.

Figure 4. Effect of external Ca²⁺ on the dose-response relationship.

A, concentration dependence of the quinine-induced current recorded in normal saline solution. The holding potential was -64 mV. A known concentration of quinine was applied by the faster bath superfusion using an air gap (see Methods). B, concentration dependence of the quinine response recorded in Ca²⁺-free solution (no added EGTA). The currents in A and B were recorded from the same cell. Current and time scales are the same as in A. C, dose-response curves of the quinine-induced current. The response amplitude was measured at a peak current within 10 s from the onset of the current response. The relative response was plotted as a function of quinine concentration. In normal saline solution (1.8 mm Ca²⁺, •), the response to 7 mm quinine (1030 ± 180 pA, mean ±s.d., 5 cells) was taken as 1.0. In the Ca²⁺-free solution (0 mm Ca²⁺, ▪), the response to 2 mm quinine (1270 ± 60 pA, 4 cells) was used for the normalization. The smooth curves were fitted by the Hill equation I/I_max=cⁿ/(cⁿ+K_½ⁿ), where c is the quinine concentration. In normal saline solution (1.8 mm Ca²⁺), K_½= 2.8 mm and n = 4.5. In the Ca²⁺-free solution (0 mm Ca²⁺), K_½= 0.48 mm and n = 3.2.

Spectral analysis of current fluctuations activated by quinine

The decrease in the apparent single-channel conductance could conceivably be due to a flicker block of the channel by Ca²⁺. If so, the spectral density distribution of the quinine-induced current may reflect this effect. Figure 5 shows the power spectra of the quinine-induced current noise obtained in the presence (Fig. 5A) and absence (Fig. 5b) of 1.8 mm external Ca²⁺. The power spectra were fitted by the sum of two Lorentzian functions:

2

Figure 5. Spectral density of the quinine-induced current.

A, power spectrum of the quinine-induced current fluctuations recorded in normal saline solution (1.8 mm Ca²⁺). Analysed data were obtained from the current in Fig. 3A. The cell was stimulated with 2 mm quinine, and the holding potential was -64 mV. The spectrum obtained without quinine was subtracted from the data of each plot. The continuous curve is given by eqn (2) with f₁ and f₂ values of 12 and 120 Hz, respectively, and S₁(0) and S₂(0) values of 1.6 and 0.047 pA² Hz⁻¹. B, power spectrum of the quinine-induced current fluctuations recorded in Ca²⁺-free solution (0 mm Ca²⁺). Data were obtained from the current in Fig. 3B. The same cell as in A was stimulated with 0.2 mm quinine at -64 mV. The spectrum obtained without quinine was subtracted from the data of each plot. The continuous curve is given by eqn (2) with f₁ and f₂ values of 19 and 110 Hz and S₁(0) and S₂(0) values of 6.9 and 0.36 pA² Hz⁻¹. Dotted lines represent the two components in eqn (2).

where S(f) is power spectral density, and S₁(0) and S₂(0) are the zero-frequency asymptotes of the two Lorentzian functions with corner frequencies f₁ and f₂. The best-fit f₁ and f₂ values were 16.4 ± 2.6 and 108 ± 10 Hz with 1.8 mm Ca²⁺ (5 cells), and 22.5 ± 5.8 and 110 ± 12 Hz without Ca²⁺ (4 cells). The difference in corner frequencies between the two situations may not be statistically significant. Furthermore, the increase would appear to be too small to be consistent with a flicker block by external Ca²⁺. Perhaps the main increase in spectral power due to Ca²⁺ occurs at frequencies beyond our 1 kHz setting of the low-pass filter. Another possibility is that Ca²⁺ may decrease the rate of ion transport by binding to an allosteric site outside the channel pore. This idea would be consistent with an apparent lack of voltage dependency of the channel conductance (Fig. 2b).

DISCUSSION

A previous study has demonstrated that the bitter substance quinine leads to the activation of a cation channel in bullfrog taste receptor cells (Tsunenari et al. 1996). In the present study, we analysed the whole-cell current with noise analysis and found that the underlying single-channel conductance was similar to that previously observed in excised membrane patches in both the absence and the presence of external Ca²⁺. The whole-cell experiments give a conductance value of 12 pS in the absence and 5 pS in the presence of 1.8 mm extracellular Ca²⁺; in excised-patch experiments, the corresponding values were 9 and 4.5 pS, respectively. Likewise, in the absence of external Ca²⁺, the sensitivity to quinine measured in the present experiments (K_½= 0.48 mm) agrees with that measured in excised patches (K_½= 0.52 mm); the dose-response relationship in the presence of Ca²⁺ has not been measured in excised patches. Finally, as we have reported previously, the ion selectivities measured under whole-cell and excised-patch recording conditions are also similar, with P_Na:P_K:P_Cs= 1:0.5:0.42 for the whole-cell current (Tsunenari et al. 1996) and 1:0.48:0.39 in the excised patch (Tsunenari et al. 1999). Taken together, these results suggest that the same channel was studied in the two sets of experiments. Since quinine can open the channel in an excised patch apparently without the involvement of a G protein, common second messengers such as cyclic nucleotides and inositol trisphosphate, or protein phosphorylation (because ATP is not required), the channel is presumably directly gated by quinine (Tsunenari et al. 1999). The channel activity of excised patches has been observed from both the apical end and the soma of the cell, although the density distribution remains to be measured (Tsunenari et al. 1999). We suppose that the channels on the apical membrane can be activated in situ by bitter tastants in the oral cavity. As to the possible function of the basolateral channels, it would have to depend on whether the tastants can access them.

Our experiments suggest that extracellular Ca²⁺ decreases the apparent affinity of the channel for quinine by almost one order of magnitude. The site of this Ca²⁺ action could be extracellular or intracellular. An intracellular action would be more interesting because it could provide a negative-feedback control on bitter taste transduction. Such an action would require the channel to be Ca²⁺ permeable, a question currently under study. In gustatory-nerve recordings, external Ca²⁺ has also been found to suppress the response to quinine (Yoshii et al. 1982). Whether this phenomenon is identical to that studied here is still unclear.

Under our experimental conditions, the response of the frog taste receptor cell to quinine appears to consist largely of a channel directly opened by quinine. However, other workers have reported, and studied in some detail, G protein-coupled pathways that are activated by bitter sensation (Akabas et al. 1988; Hwang et al. 1990; Spielman et al. 1994, 1996). In particular, the G protein gustducin is thought to rather specifically mediate the bitter taste response (McLaughlin et al. 1992; Ruiz-Avila et al. 1995). More recently, a family of what appears to be bitter taste receptors has also been identified in the mammalian genome (Adler et al. 2000; Chandrashekar et al. 2000). Likewise, the responses of taste receptor cells to sugars and amino acids are generally thought to involve seven-transmembrane-helix receptors and G proteins (Béhé et al. 1990; Chaudhari et al. 2000). On the other hand, channels directly gated by taste substances have also recently been reported to underlie amino acid and sugar taste (Kumazawa et al. 1998; Murakami & Kijima, 2000). Thus, in taste transduction, as in synaptic transmission in general, ionotropic receptors and metabotropic receptors may function in parallel with each other.