Although the numerous stimuli representing
the taste quality of bitterness are known to be transduced through
multiple mechanisms, recent studies have suggested an unpredicted
complexity of the transduction pathways for individual bitter stimuli.
To investigate this notion more thoroughly, a single prototypic bitter
stimulus, caffeine, was studied by using patch-clamp and ratiometric
imaging techniques on dissociated rat taste receptor cells. At
behaviorally relevant concentrations, caffeine produced strong
inhibition of outwardly and inwardly rectifying potassium currents.
Caffeine additionally inhibited calcium current, produced a weaker
inhibition of sodium current, and was without effect on chloride
current. Consistent with its effects on voltage-dependent currents,
caffeine caused a broadening of the action potential and an increase of the input resistance. Caffeine was an effective stimulus for elevation of intracellular calcium. This elevation was concentration dependent, independent of extracellular calcium or ryanodine, and dependent on
intracellular stores as evidenced by thapsigargin treatment. These dual
actions on voltage-activated ionic currents and intracellular calcium
levels suggest that a single taste stimulus, caffeine, utilizes
multiple transduction mechanisms.
 |
INTRODUCTION |
TASTE RECEPTOR CELLS
CONVEY gustatory information from the oral cavity to afferent
nerve fibers that, in turn, relay their activation to the central
nervous system. Our present understanding of these processes suggests
that not only do different transduction mechanisms exist for the
varying qualities of taste but multiple mechanisms exist within a taste
quality (15, 18). Bitter stimuli, in particular, comprise
a particularly heterogeneous group of chemically diverse compounds. For
example, alkaloids, glucosides, divalent cations, methylated or
acetylated carbohydrates, amino acids, and dipeptides are reported to
produce bitter sensations in humans. It is not surprising that a number
of mechanisms might be required for this wide array of chemically
diverse compounds. Moreover, because many bitter compounds are toxic,
there exists an evolutionary advantage to evolve multiple mechanisms to
sense noxious stimuli. The details of these transduction mechanisms are
the subject of this study.
Bitter stimuli have been proposed to utilize transduction pathways that
include receptor-mediated production or inhibition of second-messenger
molecules, modulation of second-messenger molecules by direct
interaction with G proteins or degradative enzymes such as
phosphodiesterase, or direct block of ion channels. Many bitter
stimuli are proposed to be transduced via the gustducin pathway.
Gustducin, a G protein expressed in 20-30% of taste receptor cells, shares considerable homology with transducin in photoreceptors (31). By analogy to transducin, the activation of
gustducin by a seven-transmembrane receptor is suggested to stimulate
phosphodiesterase, thereby lowering cyclic nucleotide concentration and
altering membrane conductance through a cyclic nucleotide-gated ion
channel (25, 35). A family of seven transmembrane
receptors has recently been cloned (2, 22, 30) that appear
to be likely candidates as bitter receptors (8). These
receptors are colocalized with gustducin, and single-taste receptor
cells apparently express multiple members of this family. Moreover,
functional expression of at least one of these receptors suggests that
they may be narrowly tuned (8). The activation of these
receptors is hypothesized to activate gustducin. In vitro biochemical
assays with crude taste receptors have demonstrated the activation of
gustducin by taste stimulation (33, 34), which
subsequently can activate phosphodiesterase in vitro (47).
Mice, deficient in the
-subunit of gustducin, are impaired in
ability to respond behaviorally and neurophysiologically to
particular bitter and sweet stimuli (58), and
transgenic expression of rat
-gustducin restored responsiveness of
gustducin-null mice to bitter and sweet compounds (59).
The mediation of gustducin activation by second messengers proved to be
more complicated than originally proposed. Full resolution of the
heterotrimeric complex of gustducin (23) suggested that the 
-complex, composed of
3 and
13,
could couple to an isoform of phospholipase found in taste cells,
phospholipase C-
2 (46). Indeed,
bitter compounds such as caffeine, denatonium, and strychinine stimulate inositol trisphosphate (IP3) production
(53). Thus the
-subunit of gustducin may couple to
phosphodiesterase, lowering cyclic nucleotides, whereas the

-complex may stimulate phospholipase C-
2, which in
turn elevates IP3 (60).
In addition to the modulation of second messengers such as cAMP and
IP3 by two arms of the gustducin protein, other
transduction mechanisms for bitter stimuli may exist. Modulation of
second messengers may occur in a receptor-independent manner via direct stimulatory interactions on G proteins that further activate downstream transduction cascades (36, 41) or by direct interaction
with enzymes that degrade cyclic nucleotides (28, 42).
Particularly bitter stimuli (caffeine and theophylline) were measured
to increase cGMP, whereas others (strychnine and denatonium) were
ineffective. Ion channels are also thought to be direct targets for
some bitter stimuli, such as the direct block of potassium channels by
quinine in rat taste receptor cells (9) or the activation
of a cationic channel directly gated by bitter stimuli, such as quinine
or denatonium, which results in an inward depolarizing current in frog
taste receptor cells (55, 56). This investigation seeks to
better clarify one widely used bitter compound in particular, caffeine.
Caffeine (1,3,7-trimethylxanthine) is an alkaloid and is closely
related to other methylxanthines such as theophylline
(1,3-dimethylxanthine) and theobromine (3,7-dimethylxanthine). All are
naturally occurring compounds in cocoa beans, cola nuts, coffee, and
tea and are reported as bitter in humans. Although caffeine is commonly
employed in gustatory science as a bitter stimulus, its underlying
transduction mechanisms are essentially unstudied. However, as a
pharmacological tool, caffeine has been well studied in other cell
types. Its actions are mostly circumscribed to three mechanisms
(37): mobilization of intracellular calcium, inhibition of
phosphodiesterases, and antagonism of adenosine receptors.
Stimulus-related activation of all three cellular signaling components
has been reported in taste receptor cells. Elevations of intracellular
calcium have been reported for several bitter stimuli, such as
denatonium (3, 7, 39), as well as for certain sweet
stimuli (4). Various phosphodiesterases have been reported
to be expressed in taste receptor cells (32) and may be
substrates for particularly bitter stimuli (28, 42) or via
activation of the gustducin pathway. Moreover, a role for
stimulus-induced activation of adenosine receptors has been proposed in
taste sensations (49, 50), where activation of adenosine
receptor, by caffeine or other methylxanthines, may act to intensify
the perception of certain sweeteners. Thus it is possible that taste
receptor cells could utilize analogous mechanisms to transduce caffeine
stimulation. The purpose of the present study is to gain insight into
the cellular actions of caffeine by using patch-clamp recording and
ratiometric calcium-imaging techniques. Such observations are
prerequisite to understanding more fully caffeine's transduction
mechanisms in taste receptor cells.
 |
METHODS |
All experiments were performed on isolated taste receptor cells
dissociated from rat circumvallate and foliate papillae
(Sprague-Dawley, 180-350 g) by using whole cell patch-clamp
recording procedures (conventional and perforated-patch methods) or
ratiometric calcium imaging with the fluoroprobe fura 2. Recordings
were conducted at room temperature.
Dissociation procedure.
Taste receptor cells were dissociated from the posterior rat tongue as
previously described (16). Briefly, lingual tissue (circumvallate and foliate papillae) was excised from the tongue after
the animal reached a surgical level of anesthesia achieved with
intramuscular injection of ketamine-acepromazine at 0.09 ml/100 g body
wt [91 mg/ml ketamine (Parke-Davis), 0.09 mg/ml acepromazine
(Vedco)]. Papillae were blocked from tongue tissue and incubated in a
cysteine-activated (1 mg/ml) papain (14 U/ml)-divalent-free bicarbonate-buffered solution for several hours at 32°C in 5% CO2-95% air. Cells were dissociated in a
pseudoextracellular fluid by mild agitation. Some papillae were
maintained in ice-cold extracellular fluid (ECF) solution for
later dissociation. Dissociated taste receptor cells were easily
identified by their elongated or bipolar morphology (16).
Solutions.
The divalent-free solution for enzymatic incubation was composed of (in
mM) 80 NaCl, 5 KCl, 26 NaHCO3, 2.5 NaH2PO4 · H2O, 20 D-glucose, and 1 EDTA. The standard ECF solution used for
the dissociation procedure included (in mM) 126 NaCl, 1.25 NaH2PO4 · H2O, 5 KCl, 5 NaHEPES, 2 MgCl2, 2 CaCl2, and 10 glucose, with pH adjusted to 7.2-7.4 with HCl. Most experiments were performed by using the perforated-patch configuration of the patch-clamp technique with amphotericin B as the ionophore [400 µg/ml in the intracellular fluid (ICF)]. The composition of the ICF for filling the
pipette consisted of (in mM) 55 KCl, 75 K2SO4,
8 MgCl2, and 10 HEPES. During recording of calcium and
chloride current, KCl and K2SO4 were replaced
by an equivalent amount of CsCl and Cs2SO4. The
composition of ICF for recording inwardly rectifying potassium (KIR) current in conventional whole cell configuration
recording mode consisted of (in mM) 140 KCl, 2 MgCl2, 1 CaCl2, 11 ethylenebis(oxonitrilo)tetraacetate (EGTA), 10 HEPES, and 4 ATP (magnesium salt). The extracellular solution for
recording KIR current consisted of the standard ECF recipe
with the replacement of 95 mM NaCl by an equivalent amount of KCl
(final extracellular potassium concentration of 100 mM). When recording
chloride currents, a potassium-free extracellular solution was employed
for the bath solution; it consisted of (in mM) 126 NaCl, 5 HEPES, 2 CaCl2, 2 MgCl2, and 10 glucose with a final
chloride concentration of 134 mM. The pH of the extracellular recording
solution was adjusted to 7.4 with Tris base. The ECF solution for
recording calcium current consisted of (in mM) 140 NaCl, 20 tetraethylammonium chloride (TEA), 10 HEPES, 5 glucose, 10 KCl, 10 CaCl2, 2 MgCl2, and 5 4-aminopyridine (4-AP),
with pH adjusted to 7.3. In some experiments, 10 mM CaCl2
in ECF was replaced by 20 mM BaCl2.
Whole cell electrophysiological recording.
Micropipettes used for whole cell recording were pulled on a gas-cooled
multistage puller from 1.5-mm-OD borosilicate glass (World Precision
Instruments, Sarasota, FL). Resistances were typically 4-7 M
when filled with ICF and measured in ECF. Junction potentials were
corrected before the electrode contacted the cell. The pipette tip was
positioned to contact the cell membrane, and negative pressure was
applied to its interior to facilitate gigaseal formation. Seal
resistances were typically several decades of gigaohms. After gigaseal
formation, it was necessary to apply further negative pressure to enter
conventional whole cell recording mode. For amphotericin B
perforated-patch recordings, ~30 min were required to reach a stable
level of recording after gigaseal formation. Fast and slow capacitance
compensation was employed as necessary with amplifier controls in both
recording situations. Cell membrane capacitance and uncompensated
series resistance were adjusted to produce optimal transient balancing.
Membrane capacitance was 3-6 pF; series resistance averaged 10 M
in conventional whole cell mode and 20-50 M
in most
amphotericin B perforated-patch-clamp recordings. Low-pass filtering
due to resistance-capacitance coupling was considered minimal. The
product of these factors produces a time constant of 30-300 µs
or a cutoff frequency (1/2
RC, where R is
resistance and C is capacitance) of 1.6-16.6 kHz.
Data were acquired with a high-impedance amplifier equipped by using a
high-resistance feedback head stage (Axopatch-1B, Axon Instruments), a Pentium-based 450-MHz computer, a 12-bit 330-kHz analog-to-digital converter (Digidata 1200, Axon Instruments), and a
commercially available software program (pCLAMP, version 7.0 or 8.01, Axon Instruments). Membrane currents were acquired after low-pass
filtering with a cutoff frequency of 10 kHz (at
3 dB). A
software-driven digital-to-analog converter generated the voltage
protocols. In most situations, currents were measured with voltage
protocols by using standard command step potentials of 80-ms duration
from a holding potential of
80 mV, applied in 10-mV increments, to a
final potential of +90 mV for study of potassium currents or to +40 mV
for calcium currents. A P/4 leak subtraction protocol was employed. For
recording KIR current, the membrane voltage was typically
held at its zero-current potential (in high extracellular potassium),
usually around
3 to
10 mV, and a series of depolarizing or
hyperpolarizing command potentials, in increments of 10 mV, was
applied, ranging from
160 to +30 mV (54). Leak
subtraction was not employed for the study of KIR current.
For recording chloride currents, the membrane potential was held at 0 mV, and a series of 80-ms command potentials in 20-mV increments (range
140 to +120 mV) was applied during acquisition of data with a
sampling rate of 50 µs (20). Leak subtraction was not
employed. To record action potentials, cells were switched from
voltage-clamp to current-clamp mode by transitionally switching to
zero-potential current clamp. Steady-state current was injected to bring the membrane potential to the desired holding potential (usually
80 mV). The software-driven analog-to-digital board generated the current injection protocol to elicit action potentials, typically 2-ms current injections (10).
Exchange of the bathing solution was accomplished with a gravity-fed
perfusion system in the recording chamber with portal and sluice at
opposite ends. Flow rate was ~2 ml/min. Several minutes were allowed
for exchange of bath volume, estimated to be 0.9 ml. The dissociated
cell preparation used in these studies allows stimulation of apical and
basolateral surfaces of the dissociated cell and is unlike that
encountered under in situ conditions. However, this otherwise
significant difference is of less concern for caffeine than for many
other taste stimuli, because caffeine is membrane permeant and, under
in situ situations, can reach intracellular and basolateral sites of
the receptor cell.
Data were analyzed with a combination of off-line software programs
that included a software acquisition suite (pCLAMP, Axon Instruments)
and a technical graphics/analysis program (Origin 6.1, MicroCal
Software). Data obtained from foliate or circumvallate taste receptor
cells were combined, inasmuch as our previous studies detailing ion
currents in these cells never demonstrated any significant differences
between these groups. The value of current before application of drugs
was normalized as 100%. Pooled one-tailed Student's t-test
was used to evaluate the statistical significance of the difference
between means. P < 0.05 was considered to indicate statistical significance. Values are means ± SE.
Calcium imaging.
Intracellular calcium levels in dissociated taste receptor cells were
monitored by using standard ratiometric techniques with the
membrane-permeable calcium-sensitive dye fura 2 and a commercially available software package for data acquisition and analysis
(SimplePCI, Compix, Cranberry Twp, PA). Dissociated posterior taste
receptor cells were loaded with fura 2-AM (5 µM, dissolved in DMSO)
in the presence of a dispersing reagent, 0.05% Pluronic F-127
(dissolved in DMSO), and 1% bovine serum albumin for ~60 min and
then washed with normal ECF for
20 min. Images were acquired with a
charge-coupled device camera (Hamamatsu Orka, Hamamatsu Photonic KK,
Hamamatsu City, Japan) through an oil-immersion ×40 objective lens on
an inverted microscope. For dual-wavelength ratiometric calcium
measurements, pairs of fluorescent images were recorded at 340- or
380-nm excitation. Excitation wavelengths were produced with a
software-driven monochromator (Polychrome II, Photonics, Applied
Scientific Instrumentation, Eugene, OR), and light was collected
through a 510-nm emission filter. Paired images were obtained once
every 10 s during the stimulation period and once per minute
during baseline measurements.
Caffeine was applied with a pipette perfusion system that allowed focal
application of stimulus by using an eight-barreled pipette controlled
by Teflon valves and channeled into a quartz pipette that was
positioned close to the cell with a micromanipulator (ValueLink 8, Automate Scientific, Oakland, CA). Stimuli were presented against a
slow background perfusion of ECF. This procedure allowed quick focal
application and removal of the stimulus (<10 s). Ratios (340 nm/380
nm) before, during, and after stimulus presentation were taken to
reflect changes of intracellular calcium in response to the stimulus.
Exposure levels at 340- and 380-nm excitation wavelengths were chosen
to produce images well below saturated levels and to optimize ratios. A
60-min loading time of fura 2-AM generally required exposure times of
0.03 s at 340 nm and 0.01 s at 380 nm that subsequently
resulted in a baseline ratio close to 0.7. This allowed optimal
crossing over of 340- and 380-nm wavelengths during times of elevated
calcium and, hence, maximum ratios. Ratios were calculated from the
mean intensity of pixels within a software-defined region of interest
(ROI) within the cell. In these experiments, the ROI was chosen from
the somatal region of the taste receptor cell. Corrected ratios were
background subtracted. Mean intensity from an ROI of background regions
was subtracted from mean intensity value of pixels within ROI of the cell for each wavelength. Pseudocolor was subsequently applied to
gray-scale images of the resulting ratio values calculated from
background-subtracted 340- and 380-nm images. Baseline ratio values,
with which stimulated ratio values were compared, were calculated as
mean values of five to seven data points acquired before stimulus
application at the rate of one point per minute.
 |
RESULTS |
The actions of caffeine were tested on a variety of
voltage-dependent ionic currents isolated from dissociated posterior
taste receptor cells. We previously performed biophysical
characterizations of a number of these currents: voltage-dependent
sodium currents (19), KIR currents
(54), outward potassium currents (10, 21),
and chloride currents (20). These currents are
heterogeneously distributed on a cell-by-cell basis in expression and
magnitude. Of the currents tested in this communication, caffeine had
its strongest effect on potassium and calcium currents. Small
inhibitions were noted on sodium currents, and caffeine was without
effect on chloride currents. In addition, caffeine evoked calcium
release from intracellular stores in a concentration-dependent manner.
Potassium currents.
The bath application of caffeine to dissociated posterior taste
receptor cells produced inhibitions of evoked outward potassium current
that were concentration dependent and reversible. Figure 1A illustrates a
representative family of current traces evoked using whole cell
voltage-clamp recordings from a holding potential of
80 mV with the
perforated-patch technique. Inward sodium and outward potassium
components are clearly evident. Application of 20 mM caffeine to this
cell resulted in an immediate and significant inhibition of the outward
potassium currents with little effect on the inward sodium currents.
These inhibitions were quickly reversed with rinse of the bathing
solution. Previous characterization of these outward potassium currents
(10) demonstrated that they are composed of multiple
components, including delayed-rectifier, A-type, and calcium-activated
potassium currents, and are similar in voltage sensitivity and
inactivation properties to neuronal potassium channels. Only a very
small component of this outward current is carried by chloride
(20). The inhibition of these outward currents by caffeine
application is very similar to that reported by quinine application,
including its magnitude, temporal onset, and reversibility
(9). The inhibition of caffeine on the potassium currents
displayed some voltage dependence. Figure 1B illustrates the
current-voltage (I-V) relationship of the currents illustrated in Fig. 1A. Caffeine caused little inhibition of
the sodium current, whereas the effect on potassium currents was
profound. Although inhibition of potassium current could be measured at all suprathreshold potentials, it was most evident at more depolarized command steps, indicating some voltage dependence to this effect. The
small inhibition of sodium currents was without effect on its
activation, voltage sensitivity, or reversal potential. As expected for
a taste stimulus, caffeine effects were concentration dependent. Data
from a different taste receptor cell stimulated with four
concentrations of caffeine are illustrated in Fig. 1C. In
this cell, response magnitudes for potassium and sodium currents were
recorded over a 60-min period, during which it was stimulated with an
ascending concentration series of caffeine (1, 5, 10, and 20 mM
stimulated for 5.8, 5.5, 4, and 4.2 min, respectively). For potassium
currents, little inhibition was evident at 1 mM, although higher
concentrations produced successively higher-magnitude inhibitions. In
contrast, for sodium current, slight inhibitions were noted at
10 mM.
All inhibitions were reversible. In other cells (data not shown), a
single concentration of caffeine was repeatedly presented and produced
stable and reversible inhibitions. Summarized data for the inhibition
of potassium currents by different caffeine concentrations are
presented in Fig. 1D.

View larger version (37K):
[in this window]
[in a new window]
|
Fig. 1.
Caffeine inhibits voltage-dependent outward currents in dissociated
rat taste receptor cells. A: sample whole cell currents
before, during, and after application of 20 mM caffeine. Voltage
protocol used to evoke the current is shown at top. Caffeine
inhibition usually reached a maximum after 2-4 min of bath
application and was reversible with washout. B:
current-voltage (I-V) plot for potassium and sodium currents
before (filled symbols) and during (open symbols) caffeine application.
Caffeine did not affect the activation threshold of these currents but
reduced their magnitudes at all suprathreshold potentials. Suppression
of potassium current demonstrated some voltage dependence.
C: inhibition to 1, 5, 10, and 20 mM caffeine from a
different cell. Current magnitudes evoked by a test pulse were recorded
over a 60-min period. Horizontal bars, time course of caffeine
application. All actions of caffeine were reversible. D:
summarized data for 1, 5, 10, and 20 mM caffeine. Values are means ± SE of the current magnitude (evoked by a test pulse of 80 to +90
mV) remaining during caffeine presentation. Number of cells for each
concentration is indicated in parentheses. Magnitude of inhibition
increased with increasing caffeine concentration. **P < 0.01.
|
|
Caffeine effects on outward potassium currents were tested at different
holding potentials as an initial assessment of which components of the
outward current might be influenced by this bitter stimulus. For
example, the transient and delayed-rectifier potassium currents are
significantly inactivated at more depolarized holding potentials (e.g.,
50 mV), whereas the calcium-activated potassium current is not.
When inhibitions of outward current produced by 10 mM caffeine were
compared at different holding potentials from the same cell,
inhibitions occurred at both holding potentials (Fig.
2A). Summarized data are
presented in Fig. 2B. Current was inhibited 26 ± 5%
(n = 9) and 39 ± 4% (n = 11) at
80 and
50 mV, respectively. These results are similar to inhibition of outward currents in these cells at these same holding potentials produced by serotonin (17). Similar effects of caffeine
could be demonstrated with the use of the potassium channel inhibitors TEA and 4-AP. We previously established that the concentrations of TEA
and 4-AP used in this study differentially affect sustained and
transient current in taste receptor cells (10); therefore, pretreatment with either agent may provide some evidence into potassium
current subtype affected by caffeine. In particular, TEA inhibition of
outward potassium currents may include cAMP-sensitive outward current
as well as sustained outward current, whereas 4-AP inhibition (at this
concentration) is more specific for transient outward current. As shown
in Fig. 2C, caffeine was capable of further reducing the
magnitude of the outward current in the presence of 1 mM 4-AP or 10 mM
TEA. Summarized data are presented in Fig. 2D for TEA and in
Fig. 2E for 4-AP. Bath application of 10 mM TEA inhibited
potassium current to 31.3% of its original magnitude (n = 5), and, thereafter, 10 mM caffeine with 10 mM TEA
could further inhibit the remaining potassium current to 14.0%
(n = 5). Bath application of 1 mM 4-AP reduced
potassium to 52.7% of its original magnitude (n = 7),
and 10 mM caffeine with 1 mM 4-AP further decreased potassium current
to 25.2% (n = 7). All actions of TEA and caffeine were
reversible. Collectively, these data suggest that caffeine may affect
multiple components of the outward potassium current. A role for
calcium-activated potassium current is most strongly supported by these
data, particularly given caffeine's increased effectiveness at a
holding potential of
50 mV. However, because inhibition could be
demonstrated in all three paradigms, it seems likely, but not directly
proven, that caffeine may exert an inhibitory action at a site common
to potassium channels.

View larger version (33K):
[in this window]
[in a new window]
|
Fig. 2.
Caffeine inhibits several types of potassium currents.
A: magnitude of caffeine inhibition is influenced by the
holding potential. Outward potassium currents were inhibited at holding
potentials of 80 and 50 mV. At 50 mV, which greatly enriches the
contribution of calcium-activated potassium current, caffeine (10 mM)
inhibition is evident. B: summarized data at holding
potentials of 50 and 80 mV. C: calcium-activated
potassium current may additionally be enriched by application of
potassium channel blockers tetraethylammonium (TEA, 10 mM) or
4-aminopyridine (4-AP, 1 mM), which are most effective on
delayed-rectifier and transient potassium currents. In the presence of
either potassium channel blocker, caffeine (Caff) persisted as an
effective inhibitor of the remaining potassium current. D
and E: summarized data for TEA and 4-AP, respectively. Total
inhibition of the potassium current is reported for potassium channel
blocker or potassium channel blocker plus caffeine. **P < 0.01. Nos. in parentheses, no. of cells.
|
|
To determine whether caffeine-induced effects on potassium currents
were dependent on or related to the elevations of intracellular calcium
that are also produced by caffeine stimulation (see below), experiments
with the calcium chelator
1,2-bis(2-aminophenoxy) ethane-N,N,N',N'-tetraacetic acid
(BAPTA) were performed. BAPTA, via pipette administration, and the
membrane-permeant ester BAPTA-AM were employed with similar results.
Caffeine alone inhibited outward potassium current to 47.8 ± 4%
of its original magnitude (n = 26). With the use of
whole cell recording with 10 mM BAPTA in the pipette, 20 mM caffeine
reduced the same current to 51.9 ± 3% of its original magnitude
(n = 6). In separate experiments, with 10-15 min
of preincubation in 10 µM BAPTA-AM, 20 mM caffeine inhibited currents
to 53.7 ± 3% of original magnitude (n = 21). Thus BAPTA appeared to have no significant influence on the
caffeine-induced inhibition of outward potassium current. As a positive
control, BAPTA-AM was tested on calcium-activated potassium current to ensure that its administration was successful.
Caffeine inhibits KIR current.
KIR current is carried by a class of potassium channels
distinct in their electrophysiological and molecular properties from those that produce outward potassium current. This current, which has
its highest conductance at negative potentials, is ubiquitously distributed across taste receptor cells and contributes with other conductances to the resting potential (54). In normal
extracellular potassium concentrations, its conductance is generally
1-2 nS. In the present study, bath application of 10 or 20 mM
caffeine inhibited KIR current (recorded in 100 mM
extracellular potassium). These actions persisted during the
application of caffeine, and washout of caffeine from the bath solution
reversed them. Representative KIR current traces from a
single taste receptor cell are presented before, during, and after
application of 10 mM (6.6-min application) or 20 mM (6.7 min) caffeine
in Fig. 3A. Currents were
inhibited at all potentials, and inhibition displayed little voltage
dependence. Data from a different taste receptor cell for 10 and 20 mM
caffeine are presented in Fig. 3B, which illustrates the
time course of reversibility and a simple dose dependence of inhibition
between the tested concentrations. In most cells, KIR
current was tested more than twice with stable and reversible
inhibitions. Summarized data of KIR current inhibition to
these two concentrations of caffeine are presented in Fig.
3C.

View larger version (45K):
[in this window]
[in a new window]
|
Fig. 3.
Inwardly rectifying potassium (KIR) current is
inhibited by caffeine. A: representative KIR
current traces from a single taste receptor cell before, during, and
after application of 10 mM (top row) or 20 mM (bottom
row) caffeine. Protocol used to evoke currents is shown at
top. Currents were recorded in high extracellular potassium
to enhance current magnitude. B: data from a different taste
receptor cell to 10 and 20 mM caffeine illustrates time course of
reversibility and concentration dependence of the inhibition.
C: summarized data of KIR inhibition to 10 and
20 mM caffeine. Data were inhibited 21 ± 2% (n = 15) and 39 ± 2% (n = 9) by 10 and 20 mM
caffeine, respectively. **P < 0.01.
|
|
Unlike outward potassium currents, KIR current is not
sensitive to inhibition by cAMP but can be inhibited by G-protein
analogs such as guanosine 5'-O-(3-thiotriphosphate)
(personal observations). These observations, considered with the quick
reversibility of these effects, suggest that KIR current
inhibition may not be cyclic nucleotide mediated but could be receptor
linked via G-protein activation. Alternatively, direct block of
KIR current by caffeine has been suggested in other cell
types (11, 57). By either mechanism, inhibition of
KIR current would be expected to depolarize the resting
potential, allowing the resting potential to drift away from the
potassium equilibrium potential as a result of leak conductances.
Additionally, because of the rectifying nature of this conductance,
inhibition would augment its ability to prevent shunting of current
during depolarization.
Caffeine reversibly inhibits calcium currents.
In addition to its action on potassium currents, caffeine also
inhibited calcium currents. Calcium currents are heterogeneously distributed across taste receptor cells, which express a mixture of
T-type, L-type, and/or T- and L-type calcium currents. Calcium currents
were recorded in ECF with an elevated calcium concentration of 10 mM
and a combination of 20 mM TEA and 5 mM 4-AP to inhibit outward
potassium current. Additionally, potassium was replaced with equimolar
cesium in the ICF to inhibit potassium channels internally. Because
these experiments were performed with perforated-patch-clamp recording,
the possibility of decreasing calcium currents as a result of rundown
was minimized. Application of 10 mM caffeine inhibited calcium current
significantly. A representative family of calcium currents is presented
in Fig. 4A before and during caffeine application (6 min). These currents contain transient and
persistent components. In most cases, this inhibition was rapidly
reversed with rinse. The I-V relationship from this cell is
presented in Fig. 4B. Caffeine was noted to inhibit calcium current with little voltage dependence; inhibition was evident over all
suprathreshold potentials. Data from a different cell are presented in
Fig. 4C. The inhibition produced by caffeine was reversible
when the stimulus was rinsed from the bathing solution. (Typical with
the perforated-patch technique, the early portion of the record
demonstrates the settling of the current magnitude as the series
resistance is established.) Summarized data are presented in Fig.
4D.

View larger version (34K):
[in this window]
[in a new window]
|
Fig. 4.
Caffeine reversibly inhibited calcium currents recorded from taste
receptor cells. A: sample family of calcium currents before
and during application of 10 mM caffeine. Protocol used to elicit these
currents is shown at top. Caffeine reversibly inhibited
these currents. B: I-V plot from current traces
in A. Caffeine inhibited calcium current at all
suprathreshold potentials, and peak of the calcium current shifted
slightly to the left. C: time course and reversibility of
caffeine inhibition in a different cell. Current magnitudes to a test
pulse from 80 to 0 mV are plotted over a 30-min period, during which
10 mM caffeine was added to the bathing solution (horizontal bar).
Caffeine inhibits the current in a time course consistent with
presentation by bath perfusion and reverses with washout of the
stimulus. D: summarized data from several cells. Current was
inhibited by 36 ± 3.5% in the presence of caffeine and recovered
to 100 ± 4% of its original value. **P < 0.01. Nos. in parentheses, no. of cells.
|
|
Caffeine does not affect chloride currents.
Chloride currents in rat posterior taste receptor cells have been
previously characterized (20). As occurs in other
epithelial cells, taste receptor cells express a heterogeneous array of
chloride currents that display strong outward rectification and contain calcium-dependent and calcium-independent components. Chloride conductance is ubiquitously distributed across taste receptor cells,
and its magnitude is small compared with other ionic conductances (typically 1 nS). These currents are additionally subject to
enhancement by adrenergic agents and likely contribute with
KIR current to the resting potential of taste receptor
cells. In the present experiments, chloride currents were recorded
using the perforated-patch technique; no significant differences were
noted in any of the critical features previously characterized using
the whole cell technique, including tail currents and rectifying in the
positive direction. Currents were isolated by recording in
potassium-free solutions and holding at zero-current potential to
inactivate most voltage-dependent conductances. The cell was held at 0 mV, and a series of depolarizing or hyperpolarizing command pulses, in
20-mV increments, were applied to final voltages of
140 or +120 mV.
Bath application of 10 or 20 mM caffeine did not affect the chloride
current amplitude, temporal course, or duration for positive or
negative evoked currents. A continuous record from a single cell is
presented in Fig. 5A over a
period of 80 min. Current magnitudes to a depolarizing or
hyperpolarizing test pulse were stable when 10 mM (7.4 min) or 20 mM
(7.3 min) caffeine was added to the bathing solution. In 12 tested
cells (Fig. 5B), the current magnitude to an outward
test current was 102.5 ± 1.8% of its original magnitude.
Similarly, the magnitude of these currents in the presence of 20 mM
caffeine was 100.4 ± 1.8% of the current magnitude before
caffeine application (n = 11).

View larger version (15K):
[in this window]
[in a new window]
|
Fig. 5.
Caffeine does not affect chloride current recorded from taste
receptor cells. A: data from a single taste receptor cell
illustrating current magnitudes of inward ( ) and
outward ( ) chloride currents elicited by command
potentials of 140 or +120 mV, respectively, before, during, and after
application of 10 or 20 mM caffeine. Horizontal bar, time course of
caffeine application. Current magnitudes were unaltered by caffeine
stimulation. B: summarized data of chloride current evoked
by command potential of 0-120 mV and normalized to prestimulation
magnitude to 10 and 20 mM caffeine. Nos. in parentheses, no. of
cells.
|
|
Effects of caffeine on membrane resistance and the gustatory action
potential.
These effects of caffeine on voltage-dependent ionic conductances,
particularly those observed on outward potassium currents, suggest that
caffeine may alter membrane resistance and membrane voltage in a manner
that would be physiologically significant. To test this possibility,
input resistance of the membrane and the gustatory action potential
were recorded.
Input membrane resistance was measured in current-clamp recording by
injection of a series of hyperpolarizing and depolarizing currents of
80-ms duration. Figure 6A
illustrates a series of current-clamp recordings and the extrapolated
I-V plot from a taste receptor cell before, 5.5 min after
caffeine application, and 3.5 min after initiation of rinse by using
the perforated-patch technique. Depolarizing current injections were
often suprathreshold and successfully elicited action potentials.
Caffeine significantly increased the input resistance of the cell in a
reversible manner. In this cell, the input resistance was increased by
~25%. The caffeine measurement was taken after 5.5 min of caffeine
superfusion, and the rinse sample after 3.5 min of ECF superfusion.
Summarized data are presented in Fig. 6B. The membrane
resistance was increased to 112 ± 3% of its original value
(n = 13) by application of 10 mM caffeine and to
124 ± 5% of its original value (n = 12) by 20 mM
caffeine. Both changes were statistically significant
(P < 0.01). These results are consistent with the
strong suppression of outward potassium current by caffeine.

View larger version (27K):
[in this window]
[in a new window]
|
Fig. 6.
Caffeine increased input resistance of the taste receptor cell
membrane. Input resistance was measured by intracellular current
injection, under current-clamp recording conditions, and the linear
portion of the resulting I-V relationship in the negative
region, where the influence of active currents is minimal, was
extrapolated. A: resultant voltage traces in response to
current injections before (ECF), during (20 mM caffeine), and after
application of 20 mM caffeine (rinse) in a single cell. As shown in
corresponding I-V plots, input membrane resistance increased
from 1.63 G ( ) to 1.98 G ( ) in
the presence of caffeine and then recovered to 1.52 G
( ) after washout. B: summary of
caffeine-induced increase in input membrane resistance in several
cells. **P < 0.01. Nos. in parentheses, no. of
cells.
|
|
The previously observed inhibitory effects of caffeine on
voltage-dependent currents additionally predict that one should observe
changes in the waveform of the gustatory action potential. To test this
prediction, gustatory action potentials were elicited by current
injection in current-clamp mode. Application of 10 or 20 mM caffeine
caused modulation of the waveform of the action potential. A
representative cell is presented in Fig.
7A. In this cell, the latter
two of five current injections elicited action potentials that had
obvious afterhyperpolarization components. Action potentials in Fig.
7A were elicited 2.7 min after caffeine application or 3.6 min after rinse. We previously established that posterior taste cells
exhibiting action potentials express at least two possible subtypes,
designated fast and slow (10); the action potentials of
Fig. 7A are an example of fast action potentials. Bath
application of 20 mM caffeine decreased the amplitude of the action
potential, prolonged its duration (measured at half-amplitude), and
reduced the amplitude of the afterhyperpolarization potential. These
changes were reversible with washout of caffeine from the bathing
medium. Summarized data to two concentrations of caffeine on the
amplitude of the gustatory action potential are presented in Fig.
7B. The amplitude was reduced to 83 ± 5%
(n = 14) or 81 ± 6% (n = 10) of
its original magnitude by 10 and 20 mM caffeine application,
respectively. Summarized data to two concentrations of caffeine on the
duration (measured at half-amplitude) of the gustatory action potential
are presented in Fig. 7C. The duration was increased to
115 ± 4% (n = 13) or 136 ± 7%
(n = 10) of its original magnitude by 10 and 20 mM
caffeine application, respectively. No noticeable differences were
observed on the effect of caffeine on the waveform of the action
potential when fast and slow action potentials were compared.

View larger version (26K):
[in this window]
[in a new window]
|
Fig. 7.
Caffeine altered the waveform of the gustatory action potential.
A: representative voltage traces from a single taste
receptor cell before, during, and after bath application of 20 mM
caffeine. Current injection protocol used to evoke these voltage
perturbations is shown below middle panel. B:
summarized data for 10 and 20 mM caffeine and amplitude of the
gustatory action potential (AP). C: summarized data for 10 and 20 mM caffeine on duration (measured at half-amplitude) of the
gustatory action potential. *P < 0.05;
**P < 0.01. Nos. in parentheses, no. of cells.
|
|
Caffeine increases intracellular calcium in taste receptor cells.
Caffeine is widely recognized as a releasing agent of calcium from
intracellular pools. With the use of standard ratiometric procedures
with the fluoroprobe fura 2, six concentrations of caffeine, ranging
from 0.01 to 20 mM, were tested in 109 taste receptor cells. Caffeine
elevated intracellular calcium in a majority of tested cells. A typical
response, recorded over an 11-min period, is illustrated in Fig.
8. In Fig. 8A, the time-lapse
digital images of a single taste receptor cell represent
background-subtracted ratio values of the raw 340 nm/380 nm images to
which a pseudocolor scheme has been applied. Caffeine was applied to
this cell at 250 s for 4 min. In Fig. 8B, the raw
intensity values for the individual wavelengths are presented. The
baseline intensities were adjusted before data recording through their
exposure times so that the intensity recorded at 380-nm excitation
exceeded that at 340-nm excitation. This procedure resulted in maximal
crossover (and, hence, ratio value) when intracellular calcium levels
increased. Crossover of the individual wavelength intensities was
indeed dramatic when 10 mM caffeine were applied to this cell. The
resulting ratio values of these individual wavelength intensities are
presented in Fig. 8C. The rise in intracellular calcium, as
indicated by the ratio values, was abrupt at the onset of stimulus
presentation and subsequently declined to a tonic value that was
maintained until washout. On washout, ratio values returned to
baseline. The action of caffeine on intracellular free calcium was
reversible. The kinetics of the calcium response caused by caffeine
displayed phasic and tonic components. Typically, intracellular calcium increases produced by caffeine reached maximum levels at the initial application of the stimulus and then gradually decreased to a level
above baseline, although the caffeine was still present in the
solution. After caffeine was removed, the intracellular calcium level
would return to baseline. In general, for repeatable response
magnitudes to occur, an interstimulus interval of
5 min was required.
Ratio responses from one cell to six concentrations of caffeine are
presented in Fig. 9A. This
cell was presented with 0.01, 0.1, 1.0, 5.0, 10, and 20 mM caffeine in
an ascending series over an 80-min test period. Only four
concentrations, 1.0, 5.0, 10, and 20 mM, resulted in measurable
increases of intracellular calcium. Summarized data for four
suprathreshold concentrations of caffeine are presented in Fig.
9B. Taste receptor cells responded to caffeine with
elevations of intracellular calcium in a dose-dependent manner. Data
are presented as averaged ratio values for the baseline period before
stimulus presentation (typically ~5 min) and peak ratio value during
stimulus presentation (Fig. 8C). Ratio values were 1.0 ± 0.19, 1.6 ± 0.12, 1.7 ± 0.12, and 1.9 ± 0.08 (mean ± SE) to 0.1, 1, 5, and 10 mM caffeine, respectively. Not
all cells responded to caffeine, but the percentage of responsive cells
also increased with increasing caffeine concentration. The numbers of
responsive cells were 3 of 25 (12%), 12 of 28 (43%), 13 of 28 (46%),
and 39 of 50 (78%) tested cells for 0.1, 1, 5, and 10 mM caffeine,
respectively.

View larger version (83K):
[in this window]
[in a new window]
|
Fig. 8.
Caffeine increases intracellular calcium in taste receptor cells.
A: pseudocolor representation of a single taste receptor
cell before, during, and after application of 10 mM caffeine. Images
were acquired with the indicated time scale. Caffeine application began
at 250 s. Scale bar indicates relative calcium levels with red as
high and blue as low. B: raw intensity values for 340- and
380-nm excitations for cell shown in A. Horizontal bar,
caffeine application. C: ratio values calculated from data
in B. Horizontal bar, caffeine application. Caffeine evokes
a response with phasic and tonic components, and ratio returns to
baseline with removal of the stimulus.
|
|

View larger version (25K):
[in this window]
[in a new window]
|
Fig. 9.
Concentration-response profile of caffeine-induced
increases of intracellular calcium. A: representative
responses from a single taste receptor cell to 0.01, 0.1, 1, 5, 10, and
20 mM caffeine. Caffeine at 0.01 and 0.1 mM did not evoke increases of
intracellular calcium, whereas 1, 5, 10, and 20 mM caffeine were
suprathreshold concentrations. B: summarized data
(means ± SE) from responsive cells to applications of 0.1, 1, 5, and 10 mM caffeine. Data are presented as ratios before (open bars) and
during (solid bars) caffeine application. Nos. in parentheses, no. of
cells.
|
|
Caffeine response requires intracellular calcium.
A final set of experiments was performed to determine whether
elevations of intracellular calcium were dependent on extracellular calcium or intracellular stores of calcium.
Experiments to determine whether extracellular calcium is required for
caffeine-induced elevations of cytoplasmic calcium were performed by
using a calcium-free ECF with EGTA buffer. A sample response is
presented in Fig. 10A. A
caffeine-sensitive cell was tested with six presentations of 5 mM
caffeine (all at 1 min each) over an 80-min period before and during
bath application of calcium-free ECF. Caffeine continued to elevate
intracellular calcium in the absence of extracellular calcium. Note the
diminution of response magnitude during repeated stimulus presentation
in calcium-free ECF that began to reverse with readdition of
extracellular calcium that is likely due to depletion of intracellular
stores. Figure 10B illustrates summarized data for caffeine
stimulation in calcium-free ECF. The average response magnitude,
normalized to prestimulation baseline (as 100), was 287.51 ± 22.74 in normal ECF and 248.89 ± 26.32 in calcium-free ECF
(n = 7; pooled data to 5 and 10 mM caffeine). The
presence of calcium-free ECF did not affect baseline ratio measurements
(96.9 ± 2). Thus removal of extracellular calcium did not
significantly alter the ability of caffeine to elevate intracellular
calcium levels.

View larger version (30K):
[in this window]
[in a new window]
|
Fig. 10.
Caffeine-induced increases of intracellular calcium
require intracellular calcium stores. A: 6 consecutive
applications of 5 mM caffeine applied in normal extracellular fluid
(ECF), calcium-free ECF, and normal ECF. Eliminating extracellular
calcium did not abolish caffeine responses. B: summarized
data for caffeine stimulation in the presence of calcium-free ECF.
C: response to 10 mM caffeine was completely eliminated when
taste receptor cells were pretreated with 1 µM thapsigargin.
Horizontal bars, caffeine application. D: in another taste
cell, 10 mM caffeine was applied over the same time course shown in
B to control for rundown. E: summarized data for
caffeine stimulation in the presence of thapsigargin. Nos. in
parentheses, no. of cells.
|
|
To test whether caffeine-stimulated elevation of intracellular calcium
is dependent on intracellular stores, taste receptor cells were treated
with 1 µM thapsigargin, an inhibitor of calcium transport into
intracellular stores. Cells were typically perfused with thapsigargin
for 25-40 min before caffeine application (Fig. 10C). A
caffeine-sensitive taste receptor cell was stimulated with 10 mM
caffeine and exposed to 1 µM thapsigargin over a period of 120 min.
Caffeine was completely ineffective as a stimulus after exposure to
thapsigargin. This inhibition of the caffeine response was long
lasting. Results from an additional cell, not exposed to thapsigargin
but tested with caffeine over the same duration, are presented in Fig.
10D to demonstrate that taste receptor cells are able to
maintain responsiveness over this test period. Summarized data with
thapsigargin treatment are presented in Fig. 10E. Before
thapsigargin treatment, the caffeine average response magnitude was
288.26 ± 42.79, whereas after thapsigargin treatment it declined
to 117.78 ± 17.64 (n = 5; pooled data to 5 and 10 mM caffeine). Thapsigargin treatment alone did not significantly affect
baseline ratio measurements (98.3 ± 5%).
Given that caffeine-induced calcium elevations in taste receptor cells
require intracellular stores, the possible involvement of ryanodine
receptors was investigated. Three concentrations of ryanodine were
tested; none resulted in alterations of intracellular calcium.
Responses, normalized to baseline ratios, were 99.1 ± 2%
(n = 11) of prestimulated levels to 50 nM ryanodine,
103.6 ± 3% (n = 5) to 20 µM ryanodine, and
101.8 ± 1% (n = 7) to 100 µM ryanodine.
Ryanodine application was also tested to determine whether it could
block caffeine-induced calcium elevations. Normalized to caffeine
responses before ryanodine application, the caffeine responses after
10-15 min of exposure were 102.5 ± 7% (n = 8) to 50 nM ryanodine, 100.3 ± 10% (n = 5) to 20 µM ryanodine, and 75.7 ± 10% (n = 7) to 100 µM ryanodine. These data suggest a lack of ryanodine receptors in rat
taste receptor cells and agree well with observations reported in
mudpuppy taste cells (39).
Collectively, these data suggest that caffeine-induced elevations of
intracellular calcium are dependent on intracellular calcium stores and
do not require extracellular calcium or ryanodine-sensitive intracellular stores. The decline in response magnitude to repeated stimulus presentations in the absence of extracellular calcium likely
reflects the inability of the cell to restore intracellular pools in
the absence of extracellular calcium. Similarly, small declines in the
caffeine response in the presence of ryanodine are more likely
attributed to sensitization, as observed with repeated caffeine exposures.
 |
DISCUSSION |
The present study establishes that taste receptor cells respond to
the presence of the bitter stimulus caffeine in a multifaceted manner.
Voltage-dependent ionic currents as well as intracellular calcium
levels of posterior rat taste receptor cells were affected. In
particular, caffeine application produced an inhibition of outward
potassium current, KIR current, and calcium current,
whereas chloride currents were unaffected. In addition, caffeine
application was also noted to produce significant elevations of
intracellular calcium. All of these effects were rapid in onset,
persisted during the period of application of caffeine, and recovered
completely on washout, consistent with the expected consequences of
taste stimulation. Moreover, these effects occurred at concentrations known to produce gustatory responses by using electrophysiological or
behavioral measures. For example, the rat threshold for whole nerve
integrated responses is reported to be 10 mM for the glossopharyngeal nerve and 10-30 mM for the chorda tympani (24). This
same study reported 10 mM as the behavioral threshold for aversion in a
two-bottle preference test. Thus the concentrations used in this study,
1-20 mM, compare well with previous data in rat. In fact, the
lowest concentration of caffeine observed to elicit a response was 0.1 mM, which elevated calcium in 12.0% of the tested cells. Most studies
of internal calcium mobilization use higher caffeine concentrations, typically 5-20 mM, suggesting that taste receptor cells may
possess more specialized mechanisms for the detection of this chemical. By comparison, the lowest concentration to inhibit potassium currents in this study was 1 mM. Overall, it appears that caffeine's action on
potassium currents and intracellular calcium occurs via independent mechanisms.
Caffeine effects on potassium currents.
Caffeine inhibited at least two distinct types of potassium currents in
posterior taste receptor cells: an outward potassium current and a
KIR current. Outward potassium currents in posterior rat
taste receptor cells are multifaceted. They are composed of at least
three distinct components: delayed rectifier type, transient A-type,
and calcium-activated potassium currents (10).
KIR current, on the other hand, is carried by a class of
potassium channel distinct from those carrying outward currents. These
channels possess unique electrophysiological, pharmacological, and
molecular properties (54). We previously reported that
~25-30% of the outward potassium current can be inhibited by
cAMP, cAMP analogs, such as 8-(4-chlorophenylthio)-cAMP, or the
phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine
(21). This inhibition is sufficient to alter the waveform
of the action potential, causing its broadening. It is also kinase
dependent, inasmuch as kinase inhibitors, such as H-8
{N-[2-(methylamino)ethyl]-5-isoquinolinesulfonamide} or protein kinase inhibitor (a peptide fragment of the regulatory subunit of protein kinase A), prevented inhibition of potassium currents that would have otherwise been produced by
8-(4-chlorophenylthio)-cAMP or 3-isobutyl-1-methylxanthine. These
observations are particularly relevant to the inhibition of potassium
currents observed in this communication by caffeine.
At concentrations used in this study, caffeine is a potent inhibitor of
phosphodiesterase. Hence, it is plausible to assume that inhibitions of
outward potassium current produced by caffeine or cAMP (e.g., via
phosphodiesterase inhibition) could operate by similar mechanisms. In
fact, the situation is likely more complicated. First, caffeine has
multiple effects on second- messenger systems in taste cells. Caffeine
has been shown to be a potent stimulator of cGMP production in murine
posterior taste receptor cells (44). With the use of
quench-flow analysis, stimulation of taste receptor cells with 25 mM
caffeine was demonstrated to increase cGMP within 50 ms. Additionally,
with the use of the same methods, caffeine (10 mM) was also
demonstrated to increase levels of IP3 (53). Data for cAMP accumulation due to caffeine stimulation are not available. Thus caffeine may increase several second-messenger molecules; the downstream actions of these second messengers are incompletely understood. Our preliminary study of cGMP has shown that
its time course is similar to that of cAMP, but, more importantly, the
percentage of cells tested that responded to cGMP was much less than
the percentage that responded to caffeine (unpublished results).
Second, there exists a discrepancy in the time course of the outward
potassium current inhibition mediated by cyclic nucleotide and that
mediated by caffeine (personal observations). Cyclic nucleotide
inhibition is slow in onset and slowly reversible, whereas that
produced by caffeine is rapid in onset (occurring without discernable
latency with bath application) and easily reversed. In fact, the time
course of inhibition of caffeine more closely resembled that observed
with quinine on these cells (9) or by forskolin
(21). Quinine and forskolin act as direct blockers of
potassium channels. Caffeine, too, has been reported to directly inhibit potassium channels in a number of cell types, including mammalian ventricular myocytes (48), vascular smooth
muscle (38), dissociated chick autonomic ganglion neurons
and pineal cells (43), and rat anterior pituitary cells
(26). The biophysical properties of the caffeine-sensitive
potassium channels are clearly different in these cells, suggesting a
direct effect on conserved pore regions of the channel molecule. It may
well be that caffeine operates like forskolin, which operates via two
mechanisms to inhibit potassium currents: a direct block and cyclic
nucleotide inhibition of potassium channels. Additionally, experiments
with TEA and 4-AP (Fig. 2) suggest that calcium-activated potassium current may be a target of caffeine inhibition, a target different from
that assumed for the cAMP-mediated inhibition of potassium current.
Taken collectively, a more likely possibility is that the predominant
action of caffeine on potassium channels may be an occlusion of its
channel pore in a manner similar to that of TEA and 4-AP, because these
effects were cumulative. Furthermore, caffeine did not change the
inactivation kinetics of potassium channel significantly (personal observations).
In addition to outward potassium currents, caffeine also inhibited
KIR current. This effect may be considered exceptional, inasmuch as KIR current has previously proven to be
recalcitrant to most tested modulations. KIR current is of
particular interest, because, in most cell types, it is a major
participant in the determination of the cell's resting potential.
Hence, inhibition of KIR current acts to depolarize the
membrane potential and, in excitable cells, may elicit an action
potential. In posterior taste cells, it has been demonstrated that
KIR current is a major contributor to the resting potential
(54). In taste cells, the KIR conductance is
~50% maximal at values of the potassium equilibrium potential, and
barium, an effective inhibitor of KIR current, depolarizes
the zero-current potential, whereas TEA, a weak blocker, is only mildly
effective and 4-AP, an ineffective blocker of KIR current,
is without effect. Hence, its inhibition would be suspected to be a
prime target for initial transduction cascades to alter the membrane
potential and subsequently elicit an action potential. Surprisingly,
however, many tested tastants did not alter this current (personal
observations). Additionally, the application of cAMP, a proposed
transduction agent in bitter and sweet cascades, was similarly
ineffective in altering KIR current (personal
observations). As with outward potassium currents, direct block of
KIR channels by caffeine has also been reported (11,
57). In isolated guinea pig ventricular myocytes, 10 mM caffeine
consistently reduced the slope of the I-V relation of
KIR current. Loading cells with BAPTA to suppress
intracellular calcium increase did not prevent this effect of caffeine.
Thus caffeine, which produced a large and easily reversible inhibition
of KIR current, could easily depolarize the taste receptor
cell through its action on this current, possibly via a direct block of
the inwardly rectifying channel.
Caffeine actions on intracellular calcium.
Among the more complicated effects of caffeine stimulation on taste
receptor cells are its modulations of cellular calcium levels. An
inhibition of calcium current and a release of intracellular calcium
were observed. So little is known of regulatory mechanisms of
intracellular calcium in taste receptor cells that it is premature to
speculate regarding mechanism. One of the first issues to be resolved
for a more complete understanding of caffeine's actions on calcium is
determination of the type of intracellular calcium store that is
caffeine sensitive in taste receptor cells. Our data suggest that taste
receptor cells have IP3-sensitive calcium stores but do not
possess ryanodine-sensitive calcium stores. Caffeine is a well-known
agonist of the latter, whereas it has actually been reported to inhibit
the former (6, 52). For example, in salivary and
pancreatic acinar cells, caffeine has been reported to activate
ryanodine receptors, leading to calcium release, but to inhibit
IP3 receptors and calcium release from IP3-sensitive calcium stores (51). Some
precedent data are available in mudpuppy taste cells, where different
bitter stimuli have been demonstrated to operate via
IP3-sensitive calcium stores or from a novel
IP3- and ryanodine-insensitive store (39). The
latter was observed with concomitant inhibition of voltage-gated ionic currents, not unlike those observed in this communication with caffeine. Hence, calcium stores in taste receptor cells are likely to
be complex, if not novel.
Caffeine's seemingly opposite actions on intracellular calcium,
releasing calcium from intracellular stores, while simultaneously causing an inhibition of calcium influx from extracellular sources, have been observed in other cell types and may represent an integrated mechanism for caffeine regulation of total intracellular calcium from
plasma membrane and endoplasmic reticulum. Caffeine-mediated transient
increase in intracellular calcium, similar to that observed in taste
receptor cells, has been reported in a multitude of other cell types
and reflects immediate calcium release from intracellular stores
(13, 14, 40). Calcium release from caffeine-sensitive and
IP3-gated stores inactivates calcium channels in other cell types, such as ventricular myocytes (1), pituitary
GH3 cells (26), and chromaffin cells
(29). Caffeine similarly inhibits calcium current and
increases intracellular calcium in rod photoreceptors (27). In these cells, it was concluded that an
intracellular calcium-dependent mechanism, triggered by caffeine, led
to suppression of the calcium current, although a smaller component
could be attributed to a direct effect of caffeine's action on the
calcium channel. One key component was that, if barium, rather than
calcium, were used as a charge carrier through voltage-gated calcium
channels, then caffeine was without effect on the calcium current. The
entrance of barium, rather than calcium, is without effect on
intracellular stores or in priming ryanodine receptors. We performed
the same experiment on taste receptor cells and found the same result: that using barium as a current carrier prevented any inhibition of the
calcium current by caffeine (personal observations). This suggests that
calcium release from the intracellular calcium pool might mediate the
suppression of the calcium channel by caffeine. Together, the degree of
interregulation of plasma membrane and endoplasmic reticulum calcium
channels and the degree of store loading suggest that these coincident
events may be a regulatory mechanism of caffeine's action. In neurons,
these two membrane systems have been proposed as a binary membrane
system for overall calcium regulation (5). An important
factor in determining the sensitivity of IP3 or ryanodine
receptors is the content of calcium in the lumen of the endoplasmic
reticulum. These receptors become primed via previous entry of calcium
from the plasma membrane voltage-gated calcium channels. Hence, the
degree of store loading is dependent on previous activation of the
cell, a form of "short-term memory." Thus the complicated interplay
among calcium release from caffeine-sensitive stores, calcium influx
through plasma membrane calcium channels, and prior caffeine
stimulation may play critical roles in determining the response to this stimulus.
Putative transduction mechanisms of caffeine.
To transduce the presence of caffeine, taste receptor cells appear to
utilize the same signal transduction mechanisms that serve as major
cellular targets of caffeine's actions in other cell types. Most
prominently, these include caffeine's inhibition of cyclic nucleotide
phosphodiesterases and caffeine's release of calcium from
intracellular stores. Additionally, the direct blockage of potassium
channels by caffeine may play an important role. The inhibition of
phosphodiesterase and release of intracellular calcium are well
established in the transduction mechanisms for bitter stimuli. As such,
it seems reasonable that these known actions of caffeine on other cell
types and the role of these same mechanisms in bitter transduction
likely serve as the cellular basis for the multiple effects of caffeine
observed on taste receptor cells in this study.
A general notion that has received empirical and theoretical support is
the closure of a resting potassium channel conductance by the taste
cells as an early transduction mechanism, resulting in depolarization
of the membrane potential on stimulus arrival. The membrane
depolarization subsequently affects more dynamic changes in
conductance, such as eliciting an action potential, which in turn cause
the activation of calcium required for synaptic transmission.
Additionally, the action potential may be modulated by changes in
voltage-gated ionic currents. The present data would agree with such an
interpretative view for caffeine. One level of complication in
interpreting steady-state alterations of current is distinguishing
among those effects on voltage-dependent conductances that might be
employed to transduce early events in the transduction cascade, such as
the arrival of the stimulus and those changes in voltage-gated
conductances that might be responsible for later events, such as
transmitter release. The inhibition of KIR current by
caffeine could clearly serve as a mechanism for the former. The roles
that modulation of second-messenger systems may play in the
transduction mechanisms associated with caffeine stimulation of taste
receptor cells await further study. Direct evidence exists for two such
second-messenger systems. Quench-flow analysis has shown that caffeine
stimulation produced measurable increases in IP3 production
(53) as well as cGMP production (44). Of these, elevation of cGMP was more robust. However, IP3 may
play a significant role in releasing intracellular calcium from the caffeine-sensitive intracellular store. The actions of cGMP and intracellular calcium stores in taste receptor cells are not well understood.
In overview, the transduction mechanisms underlying the various effects
observed in this study suggest a complex interplay of multiple
transduction pathways. Caffeine's multiple actions on electrical
properties of taste receptor cells may be exemplary of tastants in
general. One set of mechanisms may act in early events to depolarize
the membrane potential and produce an action potential. Other actions,
such as second-messenger production and internal calcium release, may
act to modulate, perhaps in a tastant-specific manner, the subsequent
production of a train of action potentials. This view suggests that
action potentials play essential signaling mechanisms within the bud,
likely for activating the afferent nerve as well as cell-to-cell
communication within the bud.
Address for reprint requests and other correspondence:
M. S. Herness, College of Dentistry, Ohio State University,
305 West 12th Ave., PO Box 182357, Columbus, OH 43218-2357 (E-mail:
herness.1{at}osu.edu).
The costs of publication of this
article were defrayed in part by the
payment of page charges. The article
must therefore be hereby marked
"advertisement"
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 16 July 2001; accepted in final form 8 March 2002.