nature NEUROSCIENCE  VOLUME 20 | NUMBER 7 | JULY 2017	 927
a r t ic l e s
Appetite represents an important basis of homeostatic regulation
that drives animals to goal-oriented consummatory behaviors. These
innate behaviors are finely controlled based on internal needs and
external nutrient sensing1,2. For instance, thirsty animals engage
in drinking behavior only when water is sensed at the periphery.
Defining the central and peripheral neural logic underlying appetite
is critical for understanding homeostatic regulations. In mammals,
specific neural populations of circumventricular organs in the brain
sense internal water balance and regulate water appetite3–7, whereas
how animals detect water at the periphery remains unexplored.
Oral sensation serves as an initial sensory checkpoint that evaluates
nutrient factors from the external environment. The mammalian taste
system detects essential nutrients, as well as toxic substances, through
taste receptors and channels expressed in TRCs8–11. For example, low
sodium is sensed by a single class of TRCs expressing the epithelial
sodium channel ENaC. Knocking out of the ENaCα gene abolishes
appetitive salt intake12,13. Similarly, the tastes of l-amino acids, sugars
and bitter substances are recognized by dedicated receptors and
TRCs on the tongue14–17. Acids are detected by a distinct set of TRCs
expressing PKD2L1, a polycystic-kidney-disease-like channel18–20.
In contrast to these basic tastes, it is still controversial whether the
detection of water, another vital nutrient for the body, is mediated by
the taste system in mammals.
Many decades of work have shown that invertebrates such as
Drosophila melanogaster can sense water through a specialized
taste cell population21,22. Recent studies have demonstrated that
PPK28, a member of the DEG/ENaC family, is required for sensory
responses to external water, as well as water-seeking behavior23.
In vertebrate species such as frogs, sheep and cats, water has
been shown to elicit electrophysiological responses in facial nerves
innervating the oral cavity24–26. Moreover, water-induced responses
have been reported in taste-related neurons of the nucleus of the
solitary tract in rodents27. Although the underlying mechanisms is
unknown, these studies suggested that water detection is, in part,
encoded by the taste system.
RESULTS
We reasoned that if water is sensed as taste in mammals, at least two
criteria should be met; first, taste responses to water should be mediated
by specific cellular and molecular substrates in the taste bud and,
second, activation of this pathway should encode a cue for external
water. As an initial step to test these hypotheses, we employed in vivo
extracellular recording from the chorda tympani taste nerves to
explore water responses. By stimulating the tongue with various
solutions, we observed robust nerve responses to deionized water,
along with other basic tastants (Fig. 1a), demonstrating that water
effectively activates the taste system.
How does application of deionized water induce action potentials
in TRCs? The mammalian oral cavity is normally covered with a thin
layer of saliva containing various ions and enzymes28,29. In our experiments,
we used artificially reconstituted saliva made of the ionic components
of normal saliva and found that washing out these ions with
water generated robust nerve responses. These observations indicate
that salivary ions play a key role in causing the responses. Therefore,
we examined the effects of individual saliva components on waterinduced
responses. Intriguingly, we found that the responses to water
were triggered when bicarbonate ions were present in the preceding
solutions: switching from bicarbonate solution to water triggered
robust nerve responses, whereas switching from bicarbonate-free
saliva to water failed to induce responses (Fig. 1b). No other ions in
artificial saliva induced taste responses when changed to water (Fig. 1b
and Supplementary Fig. 1a), although high concentrations of phosphate
caused minor responses (Supplementary Fig. 1b). Indeed,
potassium bicarbonate evoked dose-dependent nerve responses when
1Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, California, USA. 2Institute for Anatomy, University Hospital,
Duisburg-Essen University, Essen, Germany. Correspondence should be addressed to Y.O. (yoka@caltech.edu).
Received 30 March; accepted 30 April; published online 29 May 2017; doi:10.1038/nn.4575
The cellular mechanism for water detection in the
mammalian taste system
Dhruv Zocchi1, Gunther Wennemuth2 & Yuki Oka1
Initiation of drinking behavior relies on both internal state and peripheral water detection. While central neural circuits regulating
thirst have been well studied, it is still unclear how mammals recognize external water. Here we show that acid-sensing taste
receptor cells (TRCs) that were previously suggested as the sour taste sensors also mediate taste responses to water. Genetic
silencing of these TRCs abolished water-evoked responses in taste nerves. Optogenetic self-stimulation of acid-sensing TRCs in
thirsty animals induced robust drinking responses toward light even without water. This behavior was only observed when animals
were water-deprived but not under food- or salt-depleted conditions, indicating that the hedonic value of water-evoked responses
is highly internal-state dependent. Conversely, thirsty animals lacking functional acid-sensing TRCs showed compromised
discrimination between water and nonaqueous fluids. Taken together, this study revealed a function of mammalian acid-sensing
TRCs that provide a cue for external water.
©2017NatureAmerica,Inc.,partofSpringerNature.Allrightsreserved.
928	 VOLUME 20 | NUMBER 7 | JULY 2017  nature NEUROSCIENCE
a r t ic l e s
switched to water while the same concentration of potassium chloride
had no effect (Fig. 1c). Together, these results point out two important
characteristics of taste responses to water: (i) the responses are
induced by washout of saliva with water, mainly mediated by bicarbonate,
and (ii) unlike in invertebrate water detection23, osmolality
change by itself is not the key determinant.
a b
Water
Saliva
Bitter
Salt
Sour
Umami
Sweet
Normalizedresponse
c
Percentofnormalized
waterresponse
Concentration (mM)
120%
80%
40%
0%
5 10 15 20 25
d
0.4
0.0
W
aterSilicone
oil
Normalizedresponse
Water Saliva
Bitter Salt Sour Umami Sweet
25%NR
20 s
25%NR
20 s
Saliva KHCO3
N
aC
l
KC
l
X (salt)
solution
Water
25%NR
20 s
Water Silicone oil
C
aC
l2
K
2H
PO
4
KH
2PO
4
M
gC
l2
0.0
0.5
1.0
1.5
Saliva
– HCO3
–
Normalizedwaterresponse
KHCO3
NaCl
CaCl2
MgCl2
K2HPO4
KH2PO4
–HCO3
–
Saliva
KCl
0.50
0.25
0.0
–0.25
–0.50
0.75
–0.2
0.8
KHCO3
KCl
Figure 1  Water responses in the mammalian taste system. (a) Water elicits robust responses in chorda tympani taste nerves. Shown are representative
integrated chorda tympani nerve responses to water and five basic tastants (upper) and their quantified data (bottom). Application of water evoked
significant taste responses (n = 7 mice; P = 0.0006, water versus saliva). NR, normalized response. Tastants used were bitter (0.1 mM cycloheximide),
salt (60 mM NaCl), sour (10 mM citric acid), umami (50 mM monopotassium glutamate plus 1 mM inosine monophosphate) and sweet (8 mM acesulfame
potassium). Artificially reconstituted saliva solution (see Online Methods) was used as a base solution for all stimuli. (b) Effects of individual ion
components on water responses. In representative traces (top), gray and blue shades denote each salt solution and water, respectively; the trace for saliva
minus HCO3
– was from a different animal. Average water responses elicited in different salt solutions were quantified (bottom). Removal of bicarbonate
ions elicited water responses while application of water following saliva lacking bicarbonate ions or solutions of other ion species had no effect (n = 3 for
saliva, n = 4 for other solutions; P = 0.0286, KHCO3 versus saliva – HCO3
–). (c) Dose dependence of water responses to potassium bicarbonate. Water
induced larger responses with higher concentrations of potassium bicarbonate while it induced no response with potassium chloride (n = 4). (d) Washout of
saliva with nonaqueous silicone oil did not induce response. Shown are representative traces to water and silicone oil (left) and quantification of responses
(right, n = 4; P = 0.0286, water versus oil). Statistical significance was analyzed with two-tailed Mann-Whitney U-test. Values are means ± s.e.m.
b c
Water
Saliva
Sour
PKD
TeNT
Normalizedresponse
Water
Saliva
Sour
0.0
0.5
1.0
Normalizedresponse
+ amiloride
Salt
Water
– amiloride
0.0
0.5
1.0
Salt
Water
0
1
2
3TeNT
0
1
2
3
Sweet
Salt
Umami
Sour
Bitter
Silenced
a
Normalizedresponse
Trpm5
+/–
Water
Saliva
Bitter
Umami
Sweet
0.0
1.0
Water
Saliva
Bitter
Umami
Sweet
Trpm5
–/–
1.5
0.5
0.0
1.0
1.5
0.5
Water
responses
25%NR
20 s
– Sour– Salt
– Bitter
– Umami
– Sweet
Figure 2  Water activates the acid-sensing taste pathway. (a) Trpm5−/− mice show no responses to bitter, umami or sweet tastants. However, they retain
intact water responses comparable to those of control animals (n = 4 for Trpm5−/− and Trpm5+/−). (b) Application of amiloride (50 µM) completely
blocked sodium responses while it did not exert significant effects on water responses (n = 4; P = 0.2 for water – amiloride versus water + amiloride).
(c) Silencing acid-sensing TRCs (Pkd2l1TeNT mice) eliminated both water responses and acid responses, while control animals (TeNT) show normal
responses to both (n = 5 for saliva in control, n = 6 for the rest; P = 0.0022 for water). Values are means ± s.e.m. Diagrams in the bottom panel show
the type of TRCs silenced in each experiment.
©2017NatureAmerica,Inc.,partofSpringerNature.Allrightsreserved.
nature NEUROSCIENCE  VOLUME 20 | NUMBER 7 | JULY 2017	 929
a r t ic l e s
Under dehydration, animals selectively drink water over other fluids
(for example, oils). If the observed responses are the basis of water
detection, we expect the responses to be highly selective for aqueous
solutions. As hypothesized, application of nonaqueous silicone oil
to the tongue did not evoke any nerve responses compared to water,
indicating that the responses require aqueous medium in the oral
cavity (Fig. 1d).
We next sought to identify the cellular substrate mediating waterinduced
taste responses. Previous studies have identified genetic
markers that specifically label TRCs encoding individual taste qualities11,18,30.
Using these genetic handles, we examined whether taste
responses to water is independent of the previously described five
basic tastes. Transgenic animals lacking TRPM5, a key transduction
channel for umami, sweet and bitter30, were unable to detect these
a
Saliva pH
Percentofresponse
Water
(pH 7.5)
Sour
CA4
+/–
CA4
–/–
Normalizedresponse
b CA4+/–
CA4–/–
c
0.0
0.2
0.4
0.6
Normalizedresponse
+
D
ZA
C
ontrol
+
BZA
C
ontrol
25%NR
10 s
Water
(pH 7.5) 0.8
0.0
0.2
0.4
0.6
0.8
0
1
2
3
100
–100
50
0
–50
6.5 7.0 7.5 8.0
Figure 3  Carbonic anhydrases mediate taste responses to water. (a) Water responses are sensitive to salivary pH. All responses are normalized to the
responses at pH 8 (n = 4; P = 0.0392 for pH 6.5 versus pH 7.5, two-tailed paired t-test). Saliva pH in healthy animals is normally neutral to basic43.
(b) Representative traces from CA4+/− and CA4−/− mice showing responses to neutral water (adjusted to pH 7.5, left). CA4−/− mice have severely reduced
responses to water (n = 5 for CA4−/−, n = 3 for CA4+/−; P = 0.0357, two-tailed Mann-Whitney U-test). NR, normalized response. (c) Water responses
before and after incubation with the carbonic anhydrase inhibitors benzolamide (BZA) or dorzolamide (DZA). Both drugs reduced water responses
(n = 6; P = 0.0159 for benzolamide, n = 9; P = 0.0007 for dorzolamide, two-tailed paired t-test). Values are means ± s.e.m.
10 s
Trials
5
1
–light+light
21 licks
367 licks
20 licks
Light ChR2
+ –
+
+
+
–
10
6
1 s
No.oflicksper5s
Sucrose
–
light
+
light
NaCl
Pkd2l1
ChR2
Control
+ light
– lighta b c d
gf
Licking + light
Feeding
(No.oflicksper5s)
Saltpreference
(No.oflicksper5s)
Laser
Lick-triggered
photostimulation
Lick
detectorPKD Cre
ChR2-EYFPSTOP
Lox Lox
Pkd2l1ChR2
line
Light
Water
Water + light
Numberoflicks
Total 0–3 min 3–10 min
e
0 2 4 6 8 10
0
500
1,000
1,500
2,000
Water
Light
Water + light
Numberoflicksper10min
Time (min)
0
10
20
30
40
10
20
30
40
0
10
20
30
40
0
–
light
+
light
0
500
1,000
1,500
2,000
2,500
Figure 4  Stimulation of acid-sensing TRCs drives drinking responses. (a) Transgenic mice expressing ChR2 in PKD2L1-expressing TRCs (Pkd2l1ChR2)
were subjected to a close-loop self-stimulation experiment in which each lick induces a 1-s laser pulse though an optic fiber placed in an empty water
spout. (b) Photostimulation of PKD2L1-expressing TRCs induced robust drinking responses without water (trials 1–5: blue shading). In the absence of
light, the same water-deprived animal did not show consistent licking (trials 6–10). Each black bar indicates a lick event. (c) Quantification of lightdependent
lick events. Licks were counted during a 5-s window. The number of licks was averaged across trials. The panel shows Pkd2l1ChR2 mice
(n = 6, red bar) and Pkd2l1-Cre control mice (n = 6, black bar) with photostimulation; white bars indicate the number of licks without light (P = 0.0022;
two-tailed Mann-Whitney U-test). (d) Photostimulation induced continuous drinking responses. Shown are plots representative of at least 3 mice
illustrating the drinking responses toward light in thirsty control (top) and Pkd2l1ChR2 animals without (middle) or with light (bottom) during a 1-min
session. Total number of licks is shown at right. (e,f) Food-deprived (e, n = 6) or salt-depleted (f, n = 6) Pkd2l1ChR2 mice did not exhibit appetitive
behavior toward light. NaCl or sucrose solutions (300 mM each) were used as control stimuli. (g) Photostimulation does not satiate animals. Shown
are cumulative number of licks from Pkd2l1ChR2 animals during 10-min behavioral sessions (left). Either water, light, or water + light was given during
a session. Light stimulation induced significantly more total number of licks (n = 3, P = 0.0145; water versus light, two-tailed paired t-test) and licks
during minutes 3–10 (P = 0.0213; water versus light, two-tailed paired t-test). Values are means ± s.e.m.
©2017NatureAmerica,Inc.,partofSpringerNature.Allrightsreserved.
930	 VOLUME 20 | NUMBER 7 | JULY 2017  nature NEUROSCIENCE
a r t ic l e s
three taste modalities (Fig. 2a). In these animals, nerve responses to
water were unaffected and indistinguishable from those in control
animals (Fig. 2a and Supplementary Fig. 2a). Similarly, blocking the
sodium taste sensor, ENaC, by its cognate antagonist (amiloride31)
entirely suppressed sodium-evoked responses, but had no significant
effect on nerve responses to water (Fig. 2b). Finally, we examined
the involvement of acid-sensing TRCs by genetically silencing their
synaptic machinery. To achieve this goal, we used transgenic animals
in which tetanus toxin subunit was targeted to PKD2L1-positive
cells by crossing Cre-dependent tetanus toxin (TeNT) and Pkd2l1-Cre
transgenic lines32 to yield Pkd2l1TeNT mice. Surprisingly, disrupting
synaptic transmission from acid-sensing TRCs resulted in a total loss
of water responses (Fig. 2c) without affecting other taste qualities
except acid (Supplementary Fig. 2b). Taken together, our data reveal
the acid-sensing taste pathway as the cellular substrate underlying
taste responses to water in addition to acids.
What is the mechanisms by which water activates TRCs? Given a
function of PKD2L1-expressing TRCs as acid sensors (Fig. 2c), one
possibility is that the water (saliva washout) stimulus may be converted
to a local pH change, leading to the activation of this population.
Consistent with this possibility, changing pH environment by washout
of buffer with water significantly activated taste nerves (Fig. 1b). In
fact, the amplitude of water responses was highly sensitive to saliva
pH (Fig. 3a): excessive protons in saliva strongly suppressed water
responses, indicating an important role of pH for the responses.
Carbonic anhydrase 4 (CA4) is a membrane-bound enzyme
expressed by acid-sensing TRCs32 and reversibly catalyzes the conversion
of CO2 and water into bicarbonate and protons. We hypothesized
that washout of bicarbonate from saliva drives this reaction, leading to
an increase in local proton production. If this is true, pharmacological
blockade or knockout of CA4 should affect water-induced taste
responses.Infact,micelackingCA4(encodedbyCar4)exhibitedsignificant
and selective reduction in their water responses, although minor
residual responses remained (Fig. 3b and Supplementary Fig. 3a).
Similarly, CA blockers markedly suppressed water responses without
affecting other taste responses (Fig. 3c and Supplementary Fig. 3b).
These results suggest that carbonic anhydrases (mainly CA4) are the
principal detectors that translate water stimuli into the local pH drop
(Supplementary Fig. 3c). This model predicts that responses to water
should have slower kinetics than responses to acids because it requires
an additional step to activate the cells. Our analysis indicate that this
is the case; activation dynamics of taste nerves by water were slower
than those by citric acid (Supplementary Fig. 3d), supporting our
idea that carbonic anhydrase-mediated local pH change is a major
mechanism underlying water responses in TRCs.
Although our electrophysiological data showed that water specifically
activates acid-sensing TRCs, it was still unclear whether
these responses contribute to the detection of water. To address this
question, we used an optogenetic strategy by engineering animals
expressing channelrhodopsin33 (ChR2) in PKD2L1-expressing TRCs
(Pkd2l1ChR2, Fig. 4a). Photostimulating the tongue with blue light in
Pkd2l1ChR2 animals induced time-locked nerve responses, confirming
functional ChR2 expression (Supplementary Fig. 4a). We reasoned
that if we could successfully create an artificial water cue by photostimulation,
thirsty animals should ‘drink’ light. To test this possibility,
we set up a behavioral model in which animals have free access to an
empty bottle attached to an optic fiber that provides touch-based feedback
photostimulation (Fig. 4a). After a 2-d water-restriction regime,
Pkd2l1ChR2 mice exhibited vigorous drinking responses toward light
even in the absence of actual water (Fig. 4b,c). This behavior was
light-intensity dependent (Supplementary Fig. 4b) and was observed
both in a 5-s brief access test (Fig. 4c) and in a 1-min continuous test
(Fig. 4d and Supplementary Videos 1 and 2). In contrast, neither
control animals lacking ChR2 expression nor Pkd2l1ChR2 mice without
photostimulation showed this behavior (Fig. 4d).
If activation of acid-sensing TRCs indeed provides a cue of
water, then animals should be attracted to light only when they are
thirsty. Thus, we explored the effect of photostimulation on various
appetites such as sugar consumption in hungry animals (Fig. 4e)
and salt appetite in sodium-depleted animals (Fig. 4f). As predicted,
animals exhibited no behavioral attraction toward light under hungry
and salt-craving conditions (Fig. 4e,f). These results substantiate
the acid-sensing TRC population as a cellular substrate for external
water detection.
Taste buds
Taste buds
Geniculate ganglion
Geniculate ganglion
Pkd2l1ChR2
Pkd2l1ChR2
;DTR
a
b
Trials
5
1–light+light
10
6
Licking + light
1 s
0
10
20
30
40
No.oflicksper5s
+ light – light
Figure 5  Genetic ablation of ChR2-positive geniculate neurons by
diphtheria toxin. Since geniculate neurons project their dendrites to
TRCs, the ChR2-expressing geniculate population may be activated
by photostimulation of the tongue. We excluded this possibility by
toxin-mediated cell ablation of this population. (a) Representative
immunostaining of taste buds (left) and geniculate ganglia (right) from
Pkd2l1ChR2 and diphtheria toxin-treated animals expressing diphtheria
toxin receptor in the background of Pkd2l1ChR2 (Pkd2l1ChR2;DTR). After
3–4 weeks of recovery, ChR2-EYFP-expressing cells regenerated in taste
buds, but not in geniculate ganglia. Scale bars, 100 µm. (b) Left, lightinduced
licking responses in a toxin-treated Pkd2l1ChR2;DTR animal using
the same behavioral model described in Figure 3b. Photostimulation
of the tongue induced robust drinking (trials 1–5: blue shading) after
ablation of ChR2-positive geniculate neurons. Each black bar indicates a
lick event. Right, quantification of drinking responses during 5 s (n = 3,
P = 0.026, two-tailed paired-t-test). Values are means ± s.e.m.
©2017NatureAmerica,Inc.,partofSpringerNature.Allrightsreserved.
nature NEUROSCIENCE  VOLUME 20 | NUMBER 7 | JULY 2017	 931
a r t ic l e s
We next asked whether the activation of acid-sensing TRCs also
encodes satiation of water drinking. Thirsty animals normally drink
to satiety within few minutes after water becomes available (Fig. 4g).
Remarkably, water-deprived Pkd2l1ChR2 animals showed continuous
and unimpeded licking toward light during entire behavioral sessions
for 10 min (Fig. 4g). However, if water was present, animals stopped
drinking after satiation even with photostimulation (Fig. 4g). Together,
these results clearly demonstrate that activation of acid-sensing TRCs is
sufficient to drive drinking, but does not evoke satiation with water.
In addition to the expression in taste buds, we noticed that ChR2
was ectopically expressed in a small number of geniculate neurons,
secondary taste neurons that innervate taste buds (Supplementary
Fig. 5). To eliminate the possibility that these neurons are involved
in light-induced drinking responses, we expressed Cre-dependent
diphtheria toxin receptor in the background of Pkd2l1ChR2
(Pkd2l1ChR2;DTR, where DTR is derived from simian Hbegf) and
ablated the entire population of PKD2L1-expressing cells by injection
of diphtheria toxin (Fig. 5a). As TRCs but not secondary neurons
regenerate over time, we were able to eliminate the contribution of
ChR2-positive geniculate neurons. After regeneration of TRCs, we
confirmed that Pkd2l1ChR2;DTR mice still showed drinking response
toward light, demonstrating that the behavior is driven by the activity
of TRCs (Fig. 5b).
We next determined the contribution of taste signals to water drinking
behavior. External water is detected through multiple orosensory
systems including taste, temperature and tactile signals34. Even without
taste signals (for example, in Pkd2l1TeNT mice), animals showed
normal spontaneous as well as thirst-induce drinking (Fig. 6a).
Moreover, animals showed no preference regardless of bicarbonate
concentration (Fig. 6b). These data suggest that this taste pathway is
not required for water consumption. Instead, we wondered whether
the taste pathway may help discriminate water from other nonaqueous
liquids. To test this idea, Pkd2l1TeNT and control animals were waterdeprived
and then given a choice between water and low-viscosity silicone
oil in a brief taste preference test. As expected, the control group
showed strong preference to water over silicone oil (Fig. 6a,c and
Supplementary Fig. 6a,b). In contrast, Pkd2l1TeNT animals failed to
show preference between these two fluids and consumed both silicone
oil and water (Fig. 6b,c). In the two-bottle assay, Pkd2l1TeNT animals
consumed more silicone oil but less water compared to the control
group, partly because they drank both fluids and quickly became full.
We observed a similar effect on mineral oil, another tasteless fluid
(Supplementary Fig. 6c), although animals quickly learned to avoid
using nontaste cues such as consistency. Pkd2l1TeNT mice showed normal
attraction to sweet and aversion to bitter compounds (Fig. 6f),
indicating that they retain taste discrimination ability. Thus, signals
via the acid-sensing taste pathway contribute to an appropriate fluid
choice, but not to drinking action per se.
Previous studies have shown that acids activate PKD2L1-expressing
TRCs, and eliminating these cells abolishes acid-evoked taste
a b c Control
WaterSiliconeoil
8
1
8
1
Trials
0
100
200
300
400
500
Numberoflicks
Total
W
aterSilicone
oil
1 s
**
C
ontrol
Preferenceratio(water/oil)
Pkd2l1TeN
T
0 10 20
0
2
4
6
8
Sweetattraction
0.01 0.1 1
0
0.5
1.0
Control
Pkd2l1TeNT
Bitteraversion
Quinine (mM)Acesulfame potassium (mM)
e fd
0
100
200
300
400
500
Numberoflicks
Total
W
aterSilicone
oil
1 s
Pkd2l1
TeNT
Trials
8
1
8
1
WaterSiliconeoil
0
3
6
9
0
2
4
Waterintake(ml)
Ad libitum 24 h WD
0
10
20
30
40
50
No.oflicksper5s
Water + bicarbonate
6
Control
Pkd2l1TeNT
Control
Pkd2l1TeNT
Figure 6  Water-induced taste signals provide a cue for fluid discrimination. Silencing acid-sensing taste pathway disrupts proper fluid choice.
(a) Ad libitum and water deprivation (WD)-induced water intake in Pkd2l1TeNT and control (TeNT) mice. Both genotypes consumed similar amount
of water during 24 h for the ad libitum groups and during 15 min for the WD groups (n = 4 for each genotype). (b) In a two-bottle choice assay,
water-deprived mice (23 h) exhibited no preference toward water and bicarbonate water (25 mM), which does not elicit taste responses to water (n = 4).
(c) In a brief taste preference assay, water-deprived control mice (TeNT) showed strong preference toward water over silicone oil. Left, representative
licking plots for water and silicone oil; fluids were presented for 8 times each. Each black bar indicates a lick event. The number of total licks as well
as licks to each fluid was quantified (n = 6, P < 0.0001 for water versus silicone oil; two-tailed paired t-test). (d) In contrast, Pkd2l1TeNT mice did not
show preference for water and consumed a similar amount of silicone oil (n = 7). (e) Preference between water and oil was quantified as a ratio (n = 7
for Pkd2l1TeNT and n = 6 for control; P = 0.0012, two-tailed Mann-Whitney U-test). (f) Both control and Pkd2l1TeNT mice exhibited dose-dependent
attraction to sweet (acesulfame potassium) and aversion to bitter (quinine), indicating that Pkd2l1TeNT mice retain normal taste discrimination (n = 4
for Pkd2l1TeNT and n = 6 for control). Data are show as a preference ratio to water. Values are means ± s.e.m.
©2017NatureAmerica,Inc.,partofSpringerNature.Allrightsreserved.
932	 VOLUME 20 | NUMBER 7 | JULY 2017  nature NEUROSCIENCE
a r t ic l e s
nerve responses18,35. Based on these findings, the PKD2L1-expressing
TRC population has been suggested to mediate sour taste and
associated aversive behavior. However, our results show that these
TRCs also mediate water detection that drives appetitive drinking
under thirst. To address this conundrum, we investigated whether
PKD2L1-expressing TRCs are responsible for behavioral aversion to
acid. Control animals with intact acid taste sensors exhibit robust
aversion to citric acid, an acidic tastant (Fig. 7a). In contrast, we
observed similar levels of aversion in Pkd2l1TeNT animals while the
taste nerve responses to acid were eliminated (Fig. 7a). These data
indicate that the taste pathway is minimally required for behavioral
aversion to acids. Conversely, optogenetic stimulation of the same
population with ChR2 in Pkd2l1ChR2 animals triggered robust dosedependent
taste nerve firing (Fig. 7b). These mice, however, showed
no obvious aversion toward water in the presence of light at any intensity
(Fig. 7b). While PKD2L1-expressing TRCs function as acid sensors,
our loss-of-function and gain-of-function experiments strongly
suggest that additional, nontaste pathways mediate the aversive aspect
of sour taste.
DISCUSSION
Peripheral water detection represents an important basis of fluid regulation
in the body. Since Zotterman first described taste responses
to water in frogs several decades ago24, accumulating evidence supports
such a finding in various vertebrate species24–26,36. These studies
implicated specific water detection machinery in the vertebrate taste
system. Here we used in vivo electrophysiology to elucidate mechanisms
underlying water responses in the taste system. We found that
washout of saliva with water activates the acid-sensing taste pathway
through PKD2L1-expressing TRCs. Furthermore, our data imply that
the removal of bicarbonate in saliva leads to local pH change through
the activity of carbonic anhydrases expressed in PKD2L1-expressing
TRCs. Because over 99% by volume of saliva is water, it seems logical
that the taste system has a mechanism to detect the dilution of ions as
a signal of incoming water, rather than sensing water itself. Together,
these studies provide insights into the cellular and molecular basis of
water detection at the periphery.
While stimulation of PKD2L1-expressing TRCs alone can drive
appetitive drinking, our loss-of-function studies in the same population
showed no defect in water consumption (Fig. 6). These observations
suggest that nontaste signals are sufficient to drive normal levels of
water detection and consumption even without taste signals. Although
the physiological importance of PKD2L1-mediated water detection
remains unknown, the sensory redundancy with respect to water may
help maintain body fluid balance even if one sensory pathway fails.
It has been previously demonstrated32,37 that PKD2L1-expressing
TRCs are activated by acids, salts, and CO2, and we now find that
they also respond to water. These stimuli are ultimately converted to
protons that activate the population. Physiologically, these TRCs are
proton sensors mediating acid responses, as shown previously35,38.
What taste information do these TRCs encode? In this study, we found
that optogenetic activation of this taste pathway by light triggered
appetitive licking responses in thirsty animals. Conversely, functional
manipulations of PKD2L1-expressing TRCs had little effect on acidinduced
aversive behavior. These results argue that this taste pathway
detects noxious substances and hence may contribute to aversion
when combined with other sensory signals32,37. However, our study
shows that the activation of this pathway by itself may not directly
encode negative valence. Besides the taste system, other sensory pathways
including the trigeminal system also contribute to orosensation34.
Because various noxious chemicals are known to activate both
taste and trigeminal nerves39, it is conceivable that behavioral aversion
to acids, salt and CO2 are partly mediated by nontaste sensory
pathways40. At the perceptual level, there are two potential models to
explain attractive behavior induced by the acid taste pathway: first,
this population may encode nonaversive sour perception (flavor) and
animals may use it to detect external fluids. Second, these TRCs may
evoke a taste perception distinct from sour (for example, water or
ionic) and sour perception may be transmitted through concurrent
activation of taste and nontaste pathways. Although we cannot distinguish
these possibilities, identifying nontaste sour signals and further
psychophysical studies will enable us to address how taste perception
of water and sour is integrated in the brain.
Notably, behavioral attraction by photostimulation of the acid-sensing
taste pathway was induced only when animals were dehydrated,
but not food or salt depleted. This internal-state dependency is a
distinctive characteristic as compared to basic tastes such as sweet
or bitter that are innately linked to positive and negative values41.
Preferenceratio
a b
Preferenceratio
Normalizedresponse
Normalizedresponse
0 5 10 15 20
Control
Acid preference Light-induced preferenceLight-induced responses
Pkd2l1
TeNT
Control
Pkd2l1
ChR2
0 10 20 30 40 50
Light (mW)
2.5
2.0
1.5
1.0
0.5
0.0
Acid responses
Light (mW) + water
0 10 30 40 5020
1.0
0.5
0.0
0 5 10 15 20
Citric acid (mM)Citric acid (mM)
1.0
0.5
0.0
2.5
2.0
1.5
1.0
0.5
0.0
Figure 7  The acid-sensing taste pathway is not essential for sour aversion. (a) Normalized taste nerve responses to citric acid in Pkd2l1TeNT
and TeNT control animals (left). Control animals (n = 6) showed dose-dependent nerve responses to citric acid while Pkd2l1TeNT (n = 4) mice
showed no response (P = 0.0095, Pkd2l1TeNT versus control at 20 mM citric acid). However, Pkd2l1TeNT and control animals showed similar
levels of aversion toward citric acid (n = 6 for control and n = 4 for Pkd2l1TeNT mice), indicating that Pkd2l1-expressing TRCs are not necessary
for aversive behavior to sourness. (b) Photostimulation of Pkd2l1-expressing TRCs fails to induce aversion. Left, chorda tympani nerve responses
to different intensities of light in Pkd2l1ChR2 and Ai32 littermate controls with only the reporter (n = 3 for 0.04–32.6 mW for Pkd2l1ChR2 mice,
n = 4 for the rest; P = 0.0286, Pkd2l1ChR2 versus control at 48 mW). Right, photostimulation of PKD2L1-expressing TRCs did not change preference
toward water. Mice were given a bottle containing water with an optic fiber for stimulation. Control and Pkd2l1ChR2 mice showed undisturbed drinking
behavior regardless of light intensity (n = 7 for control and n = 6 for Pkd2l1ChR2 mice). Statistical significance was analyzed with two-tailed
Mann-Whitney U-test. Values are means ± s.e.m.
©2017NatureAmerica,Inc.,partofSpringerNature.Allrightsreserved.
nature NEUROSCIENCE  VOLUME 20 | NUMBER 7 | JULY 2017	 933
a r t ic l e s
While the mechanisms of this valence change are unknown, hypothalamic
and reward circuits are likely involved in this process5,42. Future
studies with neural manipulations of peripheral water pathway and
central thirst circuits should help address how appetite and the
hedonic value of water are encoded in the brain.
Methods
Methods, including statements of data availability and any associated
accession codes and references, are available in the online version of
the paper.
Note: Any Supplementary Information and Source Data files are available in the
online version of the paper.
Acknowledgments
We thank B. Ho for help with mouse husbandry. We also thank K. Scott, M. Meister
and D.J. Anderson for helpful suggestions. We thank C.S. Zuker (Columbia)
and N. Ryba (NIDCR) for generously sharing Pkd2l1-Cre and TRPM5 knockout
transgenic animals, H. Matsunami (Duke) for PKD2L1 antibody, S. Lee for
technical support and members of the Oka laboratory for comments. This work
was supported by Startup funds from the President and Provost of California
Institute of Technology and the Biology and Biological Engineering Division of
California Institute of Technology. Y.O. is also supported by the Searle Scholars
Program, the Mallinckrodt Foundation, the Okawa Foundation, the McKnight
Foundation and the Klingenstein-Simons Foundation. Support was provided by
DFG WE 2344/9-1 to G.W. Y.O. have disclosed these methods and findings to the
Caltech Office of Technology Transfer.
AUTHOR CONTRIBUTIONS
D.Z. and Y.O. conceived the research program. D.Z. and Y.O. designed and carried
out the experiments and analyzed data. G.W. maintained and provided CA4
knockout animals. D.Z. analyzed data and, together with Y.O., wrote the paper.
Y.O. supervised the entire work.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Reprints and permissions information is available online at http://www.nature.com/
reprints/index.html. Publisher’s note: Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
1.	 McKinley, M.J. & Johnson, A.K. The physiological regulation of thirst and fluid
intake. News Physiol. Sci. 19, 1–6 (2004).
2.	 Sternson, S.M. Hypothalamic survival circuits: blueprints for purposive behaviors.
Neuron 77, 810–824 (2013).
3.	 Abbott, S.B., Machado, N.L., Geerling, J.C. & Saper, C.B. Reciprocal control of
drinking behavior by median preoptic neurons in mice. J. Neurosci. 36, 8228–8237
(2016).
4.	 McKinley, M.J. et al. The sensory circumventricular organs of the mammalian brain.
Adv. Anat. Embryol. Cell Biol. 172, 1–122 (2003).
5.	 Oka, Y., Ye, M. & Zuker, C.S. Thirst driving and suppressing signals encoded by
distinct neural populations in the brain. Nature 520, 349–352 (2015).
6.	 Stricker, E.M. & Sved, A.F. Thirst. Nutrition 16, 821–826 (2000).
7.	 Young, J.K. Hunger, Thirst, Sex, and Sleep: How the Brain Controls Our Passions
(Rowman & Littlefield, 2012).
8.	 Chaudhari, N. & Roper, S.D. The cell biology of taste. J. Cell Biol. 190, 285–296
(2010).
9.	 Finger, T.E. Evolution of taste and solitary chemoreceptor cell systems. Brain Behav.
Evol. 50, 234–243 (1997).
10.	Liman, E.R., Zhang, Y.V. & Montell, C. Peripheral coding of taste. Neuron 81,
984–1000 (2014).
11.	Yarmolinsky, D.A., Zuker, C.S. & Ryba, N.J. Common sense about taste: from
mammals to insects. Cell 139, 234–244 (2009).
12.	Chandrashekar, J. et al. The cells and peripheral representation of sodium taste in
mice. Nature 464, 297–301 (2010).
13.	Shigemura, N. et al. Amiloride-sensitive NaCl taste responses are associated with
genetic variation of ENaC alpha-subunit in mice. Am. J. Physiol. Regul. Integr.
Comp. Physiol. 294, R66–R75 (2008).
14.	Matsunami, H., Montmayeur, J.P. & Buck, L.B. A family of candidate taste receptors
in human and mouse. Nature 404, 601–604 (2000).
15.	Mueller, K.L. et al. The receptors and coding logic for bitter taste. Nature 434,
225–229 (2005).
16.	Nelson, G. et al. An amino-acid taste receptor. Nature 416, 199–202 (2002).
17.	Nelson, G. et al. Mammalian sweet taste receptors. Cell 106, 381–390
(2001).
18.	Huang, A.L. et al. The cells and logic for mammalian sour taste detection. Nature
442, 934–938 (2006).
19.	Ishimaru, Y. et al. Transient receptor potential family members PKD1L3 and
PKD2L1 form a candidate sour taste receptor. Proc. Natl. Acad. Sci. USA 103,
12569–12574 (2006).
20.	Ye, W. et al. The K+ channel KIR2.1 functions in tandem with proton influx to
mediate sour taste transduction. Proc. Natl. Acad. Sci. USA 113, E229–E238
(2016).
21.	Evans, D.R. & Mellon, D. Jr. Electrophysiological studies of a water receptor
associated with the taste sensilla of the blow-fly. J. Gen. Physiol. 45, 487–500
(1962).
22.	Wolbarsht, M.L. Water taste in Phormia. Science 125, 1248 (1957).
23.	Cameron, P., Hiroi, M., Ngai, J. & Scott, K. The molecular basis for water taste in
Drosophila. Nature 465, 91–95 (2010).
24.	Liljestrand, G. & Zotterman, Y. The water taste in mammals. Acta Physiol. Scand.
32, 291–303 (1954).
25.	Shingai, T. Ionic mechanism of water receptors in the laryngeal mucosa of the
rabbit. Jpn. J. Physiol. 27, 27–42 (1977).
26.	Shingai, T. & Beidler, L.M. Response characteristics of three taste nerves in mice.
Brain Res. 335, 245–249 (1985).
27.	Rosen, A.M., Roussin, A.T. & Di Lorenzo, P.M. Water as an independent taste
modality. Front. Neurosci. 4, 175 (2010).
28.	Edgar, W.M. & O’Mullane, D.M. Saliva and Oral Health (British Dental Association,
1996).
29.	Schneyer, L.H., Young, J.A. & Schneyer, C.A. Salivary secretion of electrolytes.
Physiol. Rev. 52, 720–777 (1972).
30.	Zhang, Y. et al. Coding of sweet, bitter, and umami tastes: different receptor cells
sharing similar signaling pathways. Cell 112, 293–301 (2003).
31.	Halpern, B.P. Amiloride and vertebrate gustatory responses to NaCl. Neurosci.
Biobehav. Rev. 23, 5–47 (1998).
32.	Chandrashekar, J. et al. The taste of carbonation. Science 326, 443–445
(2009).
33.	Boyden, E.S., Zhang, F., Bamberg, E., Nagel, G. & Deisseroth, K. Millisecondtimescale,
genetically targeted optical control of neural activity. Nat. Neurosci. 8,
1263–1268 (2005).
34.	Zald, D.H. & Pardo, J.V. Cortical activation induced by intraoral stimulation with
water in humans. Chem. Senses 25, 267–275 (2000).
35.	Bushman, J.D., Ye, W. & Liman, E.R. A proton current associated with sour taste:
distribution and functional properties. FASEB J. 29, 3014–3026 (2015).
36.	Hanamori, T. Effects of various ion transport inhibitors on the water response in
the superior laryngeal nerve in rats. Chem. Senses 26, 897–903 (2001).
37.	Oka, Y., Butnaru, M., von Buchholtz, L., Ryba, N.J. & Zuker, C.S. High salt recruits
aversive taste pathways. Nature 494, 472–475 (2013).
38.	Chang, R.B., Waters, H. & Liman, E.R. A proton current drives action potentials
in genetically identified sour taste cells. Proc. Natl. Acad. Sci. USA 107,
22320–22325 (2010).
39.	Dessirier, J.M., O’Mahony, M., Iodi-Carstens, M. & Carstens, E. Sensory properties
of citric acid: psychophysical evidence for sensitization, self-desensitization, crossdesensitization
and cross-stimulus-induced recovery following capsaicin. Chem.
Senses 25, 769–780 (2000).
40.	Finger, T.E. et al. ATP signaling is crucial for communication from taste buds to
gustatory nerves. Science 310, 1495–1499 (2005).
41.	Peng, Y. et al. Sweet and bitter taste in the brain of awake behaving animals.
Nature 527, 512–515 (2015).
42.	Betley, J.N. et al. Neurons for hunger and thirst transmit a negative-valence teaching
signal. Nature 521, 180–185 (2015).
43.	Baliga, S., Muglikar, S. & Kale, R. Salivary pH: a diagnostic biomarker. J. Indian
Soc. Periodontol. 17, 461–465 (2013).
©2017NatureAmerica,Inc.,partofSpringerNature.Allrightsreserved.
nature NEUROSCIENCE doi:10.1038/nn.4575
ONLINE METHODS
Animals. All procedures were carried out in accordance with US National
Institutes of Health (NIH) guidelines for the care and use of laboratory animals
and received approval from the Caltech Institutional Animal Care and Use
Committees (IACUC; approval #1694-14G). C57BL/6ByJ (B6, stock number
00664), Ai32 (stock number 012569) and Rosa26iDTR (stock number 08040)
mice were obtained from the Jackson Laboratory. Transgenic lines used were
Trmp5 knockout (ref. 30), Pkd2l1-Cre (ref. 18) and Rosa26-flox-TeNT (ref. 44), as
described previously. For optogenetic experiments, Pkd2l1ChR2 mice were generated
by crossing Pkd2l1-Cre and Ai32 lines. Rosa26-flox-DTR was crossed with the
Pkd2l1ChR2 line to generate Pkd2l1ChR2;DTR mice for cell-ablation experiments.
Mice used for data collection were both males and females, at least 6 weeks of
age. Animals were group-housed in a temperature-controlled environment with
a 13-h light and 11-h dark cycle and ad libitum access to food and water unless
otherwise noted. To minimize the use of animals, we used the same group of
animals for multiple behavioral tests.
Nerve recordings. Mice were anesthetized with pentobarbital (100 mg/kg)
and placed in a head-fixation apparatus. Body temperature was monitored
and regulated using a closed-loop heating system. Chorda tympani
taste nerve recordings were performed as previously described12,37. Briefly,
animals were given a tracheotomy to prevent suffocation and the right branch
of the chorda tympani nerve was exposed. A high-impedance tungsten
electrode was hooked onto the nerve and a drop of halocarbon oil was dropped
inside the surgical cavity. Stimuli were delivered using a pressurized perfusion
system (AutoMate Scientific) to keep a constant flow rate. Stimuli used were
60 mM NaCl (salt), 10 mM citric acid (sour), 8 mM acesulfame potassium
(sweet), 0.1 mM cycloheximide (bitter), 50 mM monopotassium glutamate
plus 1 mM inosine monophosphate (umami), 5-centistoke silicone oil (Aldrich)
and mineral oil (Aldrich). Deionized water was either filtered (Elaga, PURELAB
flex) or purchased (ultra-purified water, Invitrogen 10977-015). All solutions
were used at room temperature. Nerve responses during the 20-s tastant
stimulation were integrated and analyzed. Responses in each recording session
were normalized to 8 mM acesulfame potassium (Figs. 1, 2b,c,e,f and 5),
10 mM citric acid (Fig. 2a), and to 25 mM KHCO3 (Fig. 1c). For Figure
1b, individual responses were normalized to the average 8 mM acesulfame
potassium responses across entire sessions. The artificial saliva composition
was as follows: 4 mM NaCl, 10 mM KCl, 6 mM KHCO3, 6 mM NaHCO3,
0.5 mM CaCl2, 0.5 mM MgCl2, 0.24 mM K2HPO4, 0.24 mM KH2PO4. pH
was held between 7.4 and 7.6. For Figure 3a, the pH of artificial saliva was
adjusted between 6 and 8. To stabilize pH of water in CA4 experiments
(Fig. 3b), we added 2 mM KHCO3 and adjusted pH to 7.5. Each tastant stimulation
was followed by intervals at least for 40 s. Pharmacological experiments
were conducted as follows: 50 µM amiloride was dissolved into tastant stimulus
solutions and delivered via the pressurized perfusion system. Tastant solutions
containing amiloride were presented for 20 s, preceded and followed
by 5-s incubation periods with saliva containing the same concentration of
amiloride. The oral cavity was incubated with a membrane-impermeable carbonic
anhydrase blocker, benzolamide (650 µM), or a membrane-permeable
blocker, dorzolamide (0.5%), in water for 7 min before washing out with saliva
as described previously32.
Analysis of activation kinetics of taste nerves. Activation kinetics of nerve
responses was analyzed using similar methods as previously described45. The
time points of 25% and 75% of maximum amplitude in each response were determined
using Matlab, and the rise time was calculated as the difference between
these two points (SupplementaryFig.3d). The ratio of water to sour rising-phase
slopes was calculated as the slope of the line connecting the points at 25% to 75%
of the peak amplitude of the response.
Taste preference assays. Animals were tested in a custom gustometer to measure
taste preference as previously described12. Solutions were presented to
animals for 60 s per trial. Each behavioral session comprised 10 to 30 trials,
depending on the number of tastants tested. Each stimulus was presented at
least 5 times in one session. Presentations automatically terminated 5 s after the
first lick. The number of licks in each of these 5-s windows were counted and
then averaged across the session. Each animal was tested for up to 3 sessions
with the same taste repertoire. We freshly prepared solutions for each behavioral
experiment to minimize contamination by other sensory cues such as odors.
Prior to all behavioral experiments, mice were water restricted for 23–46 h.
For experiments that extended more than 23 h, animals were provided with
0.5 mL of water after 23 h. For sucrose appetite assays, animals had no access
to food for the 23 h before the experiments (Fig. 4e). For salt appetite assays
(Fig. 4f), mice were injected with furosemide as previously described37 and
were kept for 23 h on a low-sodium diet (Envigo 90228) with free access to
water. For sucrose and salt appetite experiments, we used 300 mM sucrose
and NaCl. Before testing with photostimulation, animals were pretrained to
drink these solutions.
For drinking assays for silicone oil and water (Fig. 6 and Supplementary
Fig. 6), animals were pretrained to drink water and oil in a gustometer
before testing. In 5-s brief access test (Fig. 6), oil and water were presented
in the same behavioral session at least five times each. In the 5-min assay
(Supplementary Fig. 6), individual fluids were tested on separate dates; water
tests were normally followed by silicone oil tests because some animals were
euthanized after the silicone oil assays because of dehydration. We noted that
both control and Pkd2l1TeNT animals preferred water in a long-term ad libitum
drinking assay.
For quantifying spontaneous and thirst-driven drinking, animals were individually
placed in their home cages and water intake was monitored for 24 h
(ad libitum) or for 15 min (after water restriction for 23 h).
Cell ablation by injection of diphtheria toxin. Pkd2l1ChR2;DTR mice were given
an intramuscular injection of diphtheria toxin fragment A (20 µg/kg BW per
day, Sigma D0564) for 2 consecutive days. Expression of ChR2 on the tongue
was monitored before, during and after injections to assess amount of ablation.
After the 2-d injection regime, mice were housed 3–4 weeks before being used
for behavioral experiments to allow regeneration of TRCs.
Optogenetic stimulation. Photostimulation of the tongue was performed using
a gustometer as described above. Animals were subjected to a brief access taste
preference assay as follows: (i) two empty bottles with and without an optic
fiber or (ii) solutions with and without an optic fiber. Blue laser pulses (430–490
nm, Shanghai Lasers and Optics Century Co.) were delivered through an optic
fiber (1 mm diameter, ThorLabs) using a pulse generator (World Precision
Instruments). Every lick triggered a laser pulse of 1 s duration in a closedloop
manner. The laser power was kept at 48 mW (measured at the tip) unless
otherwise noted.
Photostimulation of taste nerves. The same surgery for nerve recording was
performed as outlined above. For optogenetic stimulation, the tip of an optic
fiber was placed a few millimeters from the tongue while nerve responses from
Pkd2l1ChR were recorded. Trains of light pulses of 20 s duration were flashed
onto the tongue. Each 20-s stimulation window was followed by a 40-s interval.
Light trains were delivered at 8 Hz, with each pulse of 40 ms duration. The frequency
and duration was determined based on the licking behavior of mice in
our behavioral assays.
Histology. Animals were euthanized with CO2 followed by cervical dislocation
and perfused with PBS followed by 4% PFA. Tongues were removed and
kept in 4% PFA for 12 h followed by 30% sucrose in PBS for overnight. Frozen
sections of 20 µm thickness were prepared on slides and were postfixed with
4% PFA for 15 min. Then the samples were incubated with blocking buffer
(10% donkey serum, 0.2% Triton-X) for 2 h before incubation with primary
antibodies. Primary antibodies used were goat anti-CA4 (R&D Systems, 1:500,
AF2414), anti-GFP (Abcam, 1:1000, ab6673) and rabbit anti-PLC-β2 (Santa
Cruz Biotechnologies, 1:500, sc-206). Rabbit PKD2L1 antibody19 (1:500) was
a generous gift from H. Matsunami at Duke University. Secondary antibodies
were donkey anti-goat Cy3 (Jackson ImmunoResearch, 1:500, 705-165-147) and
donkey anti-rabbit Alexa-488 (Jackson ImmunoResearch, 1:500, 711-545-152),
donkey anti-chicken Alexa-488 (Jackson ImmunoResearch, 1:500, 703-545-155).
For each representative image, the experiments were successfully repeated at
least three times.
©2017NatureAmerica,Inc.,partofSpringerNature.Allrightsreserved.
nature NEUROSCIENCEdoi:10.1038/nn.4575
Statistical analysis. Depending on the experimental design, we used either
two-tailed Mann-Whitney U-test or paired t-test. Values indicate the number
of animals used for the experiments. No statistical methods were used to predetermine
sample sizes, but our sample sizes are similar to those reported in
previous publications5,12. Data distribution was assumed to be normal, but this
was not formally tested. No randomization was used for the data collection. Data
collection and analysis were not performed blind to the conditions of the experiments,
although key behavioral experiments such as Figure 4 were repeated by
other laboratory members in a blind fashion. Data points were excluded from
the analysis if surgery was unsuccessful.
Data and code availability. The data that support the findings of this study are
provided in source data for Figures 1, 3, 4, 6 and 7, and code is available from
the corresponding author upon reasonable request.
A Supplementary Methods Checklist is available
44.	Yamamoto, M. et al. Reversible suppression of glutamatergic neurotransmission of
cerebellar granule cells in vivo by genetically manipulated expression of tetanus
neurotoxin light chain. J. Neurosci. 23, 6759–6767 (2003).
45.	Storm, J.F. Action potential repolarization and a fast after-hyperpolarization in rat
hippocampal pyramidal cells. J. Physiol. (Lond.) 385, 733–759 (1987).
©2017NatureAmerica,Inc.,partofSpringerNature.Allrightsreserved.