1.25 mM NaH2PO4, 1 mM CaCl2, 1 mM MgCl2, 26 mM NaHCO3 and 10 mM dextrose,
bubbled with 95% O2/5% CO2 (pH 7.4). Slices (300­400 mm thick) were prepared with a
vibratome (Pelco), placed in warm dissection buffer (33­35 8C) for ,30 min, transferred
to a holding chamber containing artificial cerebrospinal fluid (ACSF: 124 mM NaCl, 2 mM
KCl, 1.5 mM MgSO4, 1.25 mM NaH2PO4, 2.5 mM CaCl2, 26 mM NaHCO3, 10 mM
dextrose), and kept at 22­248C for .1 h before use. For experiments, slices were
transferred to the recording chamber and perfused (4.0­4.5 ml min21
) with oxygenated
ACSF at 22­24 8C.
Electrophysiology
Somatic and dendritic whole-cell recordings were made in current-clamp with an
Axopatch 200B amplifier (Axon) using infrared differential interference optics (IR-DIC)
video microscopy. Layer 2/3 pyramidal cells were selected based on morphological and
electrophysiological criteria12,14
. Patch pipettes (somatic: 3­8 MQ; dendritic: 8­20 MQ)
were filled with intracellular solution (120 mM K-gluconate, 10 mM HEPES, 0.1 mM
EGTA, 20 mM KCl, 2 mM MgCl2, 10 mM phosphocreatine, 2 mM ATP, and 0.25 mM
GTP). The mean resting potential was 270.3 ^ 0.7 mV, corrected for the measured liquid
junction potential (6.8 mV). The series resistance was 14.6 ^ 1.6 MQ. Input resistance
(Ri  118.0 ^ 5.4 MQ) was monitored with hyperpolarizing current pulses (50 pA,
100 ms); cells were excluded if Ri changed .30% over the entire experiment12,14
. Data were
filtered at 2 kHz, digitized at 10 kHz and analysed with Clampfit 8 (Axon). EPSPs were
evoked by focal extracellular stimulation (0.01­1 ms, 1­100 V) with small glass bipolar
electrodes. To ensure that the distal (.100 mm from soma) and proximal (,50 mm)
electrodes activated separate synapses, we tested both the linear summation of EPSPs
(Fig. 1b) and paired-pulse depression between inputs. Although consecutive stimulation
through the same electrode induced marked paired-pulse depression of these synapses, no
depression was induced by sequential stimulation through the two electrodes, indicating
that they activated separate synapses. Postsynaptic APs were elicited either with
depolarizing current injection through the recording electrode (1 nA, 1­4 ms; somatic
recordings) or antidromically with an extracellular stimulating electrode near the axonal
initial segment (0.01­1 ms, 5­100 V; dendritic recordings). Presynaptic spike timing was
defined as the onset of EPSP and postsynaptic spike timing was measured at the AP
peak12,14
. Synaptic strength was measured as the initial slope (first 2 ms) of the EPSP. To
measure long-term synaptic modification, a stable baseline of synaptic strength was first
established by 6­12 min of recording with presynaptic stimulation at 0.2 Hz. Synaptic
strength after induction was measured 11­20 min after the end of induction. For
iontophoretic application of glutamate, a sharp microelectrode (150­200 MQ) filled with
250 mM Na
-glutamate was positioned near the apical dendrite (,5 mm). Both the
holding current (1­10 nA) and ejection current (100­300 nA, 0.01­3 ms) were applied
with an amplifier (Getting)3
.
Model simulation
The model consisted of one postsynaptic neuron and 100 presynaptic inputs (50 to the
proximal and 50 to the distal dendrite, see diagram in Fig. 5b). The presynaptic neurons
had a range of response time courses (Fig. 5a) but they were all driven by the same sensory
stimulus, which was a temporally varying random signal (see Supplementary Fig. 6a). The
spike train of the postsynaptic cell was simulated after integrating the synaptic input to
both the distal and proximal dendritic compartments. All connection weights were
initialized to 0.5 plus a small random number (Fig. 5b, upper panel), and were modified
according to the STDP windows measured experimentally. Details of the model are given
in Supplementary Methods.
Received 11 May; accepted 23 December 2004; doi:10.1038/nature03366.
1. Stuart, G. & Spruston, N. Determinants of voltage attenuation in neocortical pyramidal neuron
dendrites. J. Neurosci. 18, 3501­3510 (1998).
2. Larkum, M. E., Zhu, J. J. & Sakmann, B. A new cellular mechanism for coupling inputs arriving at
different cortical layers. Nature 398, 338­341 (1999).
3. Cash, S. & Yuste, R. Linear summation of excitatory inputs by CA1 pyramidal neurons. Neuron 22,
383­394 (1999).
4. Magee, J. C. Dendritic integration of excitatory synaptic input. Nature Rev. Neurosci. 1, 181­190
(2000).
5. Segev, I. & London, M. Untangling dendrites with quantitative models. Science 290, 744­750 (2000).
6. Hausser, M., Major, G. & Stuart, G. J. Differential shunting of EPSPs by action potentials. Science 291,
138­141 (2001).
7. Tamas, G., Szabadics, J. & Somogyi, P. Cell type- and subcellular position-dependent summation of
unitary postsynaptic potentials in neocortical neurons. J. Neurosci. 22, 740­747 (2002).
8. Hausser, M. & Mel, B. Dendrites: bug or feature? Curr. Opin. Neurobiol. 13, 372­383 (2003).
9. Magee, J. C. & Johnston, D. A. synaptically controlled, associative signal for Hebbian plasticity in
hippocampal neurons. Science 275, 209­213 (1997).
10. Markram, H., Lubke, J., Frotscher, M. & Sakmann, B. Regulation of synaptic efficacy by coincidence of
postsynaptic APs and EPSPs. Science 275, 213­215 (1997).
11. Sourdet, V. & Debanne, D. The role of dendritic filtering in associative long-term synaptic plasticity.
Learn. Mem. 6, 422­447 (1999).
12. Feldman, D. E. Timing-based LTP and LTD at vertical inputs to layer II/III pyramidal cells in rat barrel
cortex. Neuron 27, 45­56 (2000).
13. Bi, G. & Poo, M. Synaptic modification by correlated activity: Hebb's postulate revisited. Annu. Rev.
Neurosci. 24, 139­166 (2001).
14. Froemke, R. C. & Dan, Y. Spike-timing-dependent synaptic modification induced by natural spike
trains. Nature 416, 433­438 (2002).
15. Watanabe, S., Hoffman, D. A., Migliore, M. & Johnston, D. Dendritic K channels contribute to
spike-timing dependent long-term potentiation in hippocampal pyramidal neurons. Proc. Natl Acad.
Sci. USA 99, 8366­8371 (2002).
16. Sjostrom, P. J., Turrigiano, G. G. & Nelson, S. B. Neocortical LTD via coincident activation of
presynaptic NMDA and cannabinoid receptors. Neuron 39, 641­654 (2003).
17. Johnston, D. et al. Active dendrites, potassium channels and synaptic plasticity. Phil. Trans. R. Soc.
Lond. B Biol. Sci. 358, 667­674 (2003).
18. Stuart, G. J. & Hausser, M. Dendritic coincidence detection of EPSPs and action potentials. Nature
Neurosci. 4, 63­71 (2001).
19. Golding, N. L., Staff, N. P. & Spruston, N. Dendritic spikes as a mechanism for cooperative long-term
potentiation. Nature 418, 326­331 (2002).
20. Svoboda, K., Helmchen, F., Denk, W. & Tank, D. W. Spread of dendritic excitation in layer 2/3
pyramidal neurons in rat barrel cortex in vivo. Nature Neurosci. 2, 65­73 (1999).
21. Waters, J., Larkum, M., Sakmann, B. & Helmchen, F. Supralinear Ca2
influx into dendritic tufts of
layer 2/3 neocortical pyramidal neurons in vitro and in vivo. J. Neurosci. 22, 8558­8567 (2003).
22. Koester, H. J. & Sakmann, B. Calcium dynamics in single spines during coincident pre- and
postsynaptic activity depend on relative timing of back-propagating action potentials and
subthreshold excitatory postsynaptic potentials. Proc. Natl Acad. Sci. USA 95, 9596­9601 (1998).
23. Zilberter, Y., Kaiser, K. M. & Sakmann, B. Dendritic GABA release depresses excitatory transmission
between layer 2/3 pyramidal and bitufted neurons in rat neocortex. Neuron 24, 979­988 (1999).
24. Rosenmund, C., Feltz, A. & Westbrook, G. L. Calcium-dependent inactivation of synaptic NMDA
receptors in hippocampal neurons. J. Neurophysiol. 73, 427­430 (1995).
25. Tong, G., Shepherd, D. & Jahr, C. E. Synaptic desensitization of NMDA receptors by calcineurin.
Science 267, 1510­1512 (1995).
26. Umemiya, M., Chen, N., Raymond, L. A. & Murphy, T. H. A calcium-dependent feedback mechanism
participates in shaping single NMDA miniature EPSCs. J. Neurosci. 21, 1­9 (2001).
27. Zucker, R. S. Calcium- and activity-dependent synaptic plasticity. Curr. Opin. Neurobiol. 9, 305­313
(1999).
28. Nevian, T. & Sakmann, B. Single spine Ca2
signals evoked by coincident EPSPs and backpropagating
action potentials in spiny stellate cells of layer 4 in the juvenile rat somatosensory barrel cortex.
J. Neurosci. 24, 1689­1699 (2004).
29. Song, S., Miller, K. D. & Abbott, L. F. Competitive Hebbian learning through spike-timing-dependent
synaptic plasticity. Nature Neurosci. 3, 919­926 (2000).
30. Archie, K. A. & Mel, B. W. A model for intradendritic computation of binocular disparity. Nature
Neurosci. 3, 54­63 (2000).
Supplementary Information accompanies the paper on www.nature.com/nature.
Acknowledgements We thank P. Ascher, G. Bi, L. Chen, M. Frerking, E. Isacoff, R. Kramer and
R. Zucker for helpful discussions, and K. Arendt, K. Borges, N. Caporale and C. Nam for technical
assistance. This work was supported by grants from the National Eye Institute and the Grass
Foundation. R.C.F. is a recipient of the Howard Hughes Predoctoral Fellowship.
Competing interests statement The authors declare that they have no competing financial
interests.
Correspondence and requests for materials should be addressed to Y.D. (ydan@berkeley.edu).
..............................................................
The receptors and coding
logic for bitter taste
Ken L. Mueller1
, Mark A. Hoon2
, Isolde Erlenbach2
,
Jayaram Chandrashekar1
, Charles S. Zuker1
& Nicholas J. P. Ryba2
1
Howard Hughes Medical Institute and Departments of Biology and
Neurosciences, University of California at San Diego, La Jolla, California
92093-0649, USA
2
National Institute of Dental and Craniofacial Research, National Institutes of
Health, Bethesda, Maryland 20892, USA
.............................................................................................................................................................................
The sense of taste provides animals with valuable information
about the nature and quality of food. Bitter taste detection
functions as an important sensory input to warn against the
ingestion of toxic and noxious substances. T2Rs are a family of
approximately 30 highly divergent G-protein-coupled receptors
(GPCRs)1,2
that are selectively expressed in the tongue and palate
epithelium1
and are implicated in bitter taste sensing1­8
. Here we
demonstrate, using a combination of genetic, behavioural and
physiological studies, that T2R receptors are necessary and
sufficient for the detection and perception of bitter compounds,
and show that differences in T2Rs between species (human and
mouse) can determine the selectivity of bitter taste responses. In
addition, we show that mice engineered to express a bitter taste
letters to nature
NATURE | VOL 434 | 10 MARCH 2005 | www.nature.com/nature 225
 2005 Nature Publishing Group

receptor in `sweet cells'9
become strongly attracted to its cognate
bitter tastants, whereas expression of the same receptor (or even a
novel GPCR) in T2R-expressing cells resulted in mice that are
averse to the respective compounds. Together these results
illustrate the fundamental principle of bitter taste coding at the
periphery: dedicated cells act as broadly tuned bitter sensors that
are wired to mediate behavioural aversion.
Two main lines of evidence suggest that T2Rs function as
mammalian bitter taste receptors. First, heterologous expression
assays have shown that several human and rodent T2Rs respond to
bitter tasting compounds: mouse (m)T2R5 is a high affinity
receptor for cycloheximide (Cyx)3
, human (h)T2R16 is a candidate
receptor for b-glucopyranosides (salicin and related compounds)4
,
hT2R14 is a candidate receptor for picrotoxinin6
, and hT2R44 and
hT2R61/hT2R43 are receptors for denatonium, aristolochic acid
and 6-nitrosaccharin7,8
. Second, sequence polymorphisms in T2Rs
have been linked to differences in bitter taste sensitivity in mice3
and
humans5
. To determine if T2Rs and T2R-expressing cells are
necessary and sufficient for bitter taste perception in vivo, we used
several complementary strategies. First, we genetically engineered
mice to express candidate human T2R receptors for tastants that
mice do not respond to, and examined whether introduction of
these candidate bitter receptors endows the animals with an
expanded bitter taste repertoire. Second, we generated genetically
modified mice lacking a specific T2R and studied their behavioural
and physiological responses to bitter compounds. Third, we exam-
ined the specificity of bitter taste responses of animals in which all
sweet, umami and bitter taste function was eliminated and then
selectively restored in T2R-expressing cells. Finally, we functionally
dissected the role of cells and receptors by ectopically expressing a
T2R bitter taste receptor in sweet-sensing9
(T1R) cells.
Mice and humans have distinctive differences in their sensitivities
to many bitter compounds. For example, several b-glucopyrano-
sides evoke strong bitter taste in humans, yet mice are largely
indifferent to these compounds. Similarly, phenylthiocarbamide
(PTC), a well known bitter tastant often used in human genetic
studies, is ineffective in mice (Fig. 1). We placed the candidate
human receptor for b-glucopyranosides (hT2R16, ref. 4) and PTC
(hT2R38, ref. 5) under the control of a mouse T2R promoter1,10
,
and generated transgenic animals expressing either of these two
candidate taste receptors in T2R-expressing cells. To examine taste
behaviour, we measured taste choices in two-bottle intake prefer-
ence assays, or by direct counting of immediate licking responses in
a multi-channel gustometer (see Methods and ref. 10). Figure 1
shows that T2R-hT2R16 transgenic animals acquire the ability to
detect and respond to phenyl-b-D-glucopyranoside at concen-
trations that closely approximate the human physiological taste
sensitivity range4
. Likewise, expression of hT2R38 confers selective
PTC sensitivity to the engineered mice. These results validate T2Rs
and T2R-expressing cells as mediators of bitter taste perception
in vivo, and demonstrate that T2Rs are sufficient for selective
responses to bitter tastants; they also confirm that hT2R16 is the
b-glucopyranoside receptor4
and hT2R38 is the PTC receptor5
. In
addition, the ability to confer human bitter taste responses on mice
by introduction of human taste receptors illustrates an important
feature of T2Rs and bitter taste: selectivity and sensitivity differences
to bitter compounds between species is probably a reflection of
sequence differences in the respective T2R repertoires.
Next, we used homologous recombination to generate animals
lacking mT2R5, the candidate Cyx receptor3
. If T2R5 is the principal
bitter taste receptor for Cyx, then its knockout should abolish most,
if not all, responses to cycloheximide. To explore the effect of this
knockout in vivo, we recorded tastant-induced action potentials
from one of the principal nerves innervating taste receptor cells of
the tongue. Figure 2 shows that T2R52/2
mice have a dramatic and
selective loss of responses to Cyx. In addition, the animals are no
longer behaviourally averse to Cyx, even at concentrations 100-fold
higher than those required to trigger strong repulsion in wild-type
mice. As expected, responses to sweet, umami, sour and salty
tastants are physiologically and behaviourally comparable to
controls. To further examine the taste repertoire of the T2R52/2
mice, we performed studies of bitter taste against a broad panel of
bitter tastants. T2R52/2
animals retain basically normal responses
to all other bitter tastants tested (Fig. 2 and Supplementary Fig. 2).
These results prove the essential requirement of a T2R receptor for
bitter taste, and together with our T2R mis-expression studies
(Fig. 1) show that defined T2Rs are both necessary and sufficient
for bitter taste sensation. Interestingly, T2R52/2
mice show residual
responses to millimolar levels of cycloheximide, but mice defective
in all bitter taste signalling (for example, phospholipase C b2
knockout (PLCb22/2
) mice, ref. 10) have a complete loss of
Cyx sensitivity (Fig. 2b); we suggest that this residual activity in
T2R52/2
mice reflects the recruitment of lower affinity T2Rs.
Previously, we showed that most T2R receptor genes are co-
expressed in the same subset of taste receptor cells of the tongue and
palate epithelium1
. We interpreted this to mean that individual
T2R-expressing cells act as broadly tuned bitter sensors capable of
responding to a wide diversity of tastants but not necessarily able to
discriminate between them. Recently, we used mice deficient in
sweet, umami and bitter taste (owing to a deletion of the PLCb2
effector molecule10
) to ask whether taste receptor cells are tuned to
single or multiple taste modalities. By selectively rescuing PLCb2
Figure 1 Introduction of human bitter receptors expands the bitter taste repertoire of
mice. Brief-access taste tests measuring immediate lick responses show that normal
FVB/N mice, unlike humans4,5
, do not taste phenyl-b-D-glucopyranoside or PTC (grey). In
contrast, transgenic mice expressing the human T2R taste receptors for phenyl-b-D-
glucopyranoside or PTC in bitter-sensing cells show robust behavioural aversion to these
tastants (red). a, Mice expressing the human T2R16 taste receptor under the control of
the mT2R19 promoter (red) respond to concentrations of phenyl-b-D-glucopyranoside
that taste bitter to humans. b, Transgenic mice expressing human T2R38 (red) are averse
to PTC at concentrations closely mimicking human sensitivity to this tastant. These mice
respond normally to control bitter compounds (data not shown). Values are
means ^ s.e.m. (n  8).
letters to nature
NATURE | VOL 434 | 10 MARCH 2005 | www.nature.com/nature226
 2005 Nature Publishing Group

function only in T2R-expressing cells, we showed that there is
complete functional segregation between attractive (sweet,
umami) and aversive (bitter) tastes10
. We now make use of a similar
strategy to ask whether re-introduction of PLCb2 under the control
of selective T2R-promoters restores complete or only partial bitter
taste. If indeed individual T2R-cells express most bitter taste
receptors, then targeting expression of a PLCb2 `rescue' construct
under the control of any one T2R promoter should be sufficient to
restore normal bitter taste (that is, if the cell can now signal, all the
receptors will function). Conversely, if different T2R-expressing
Figure 2 T2R5 is necessary for cycloheximide taste reception and perception. T2R52/2
mice show a strong and selective impairment in their ability to taste cycloheximide.
a, Electrophysiological responses to cycloheximide are essentially eliminated in knockout
mice (red bars), but responses to other tastants are equivalent to those of control
animals10
(grey bars; see Supplementary Information). Shown are integrated chorda
tympani responses normalized to the response to 100 mM NH4Cl (ref. 21). T2R52/2
mice
show a severe selective deficit in their ability to taste cycloheximide (b) but not to other
bitter (c) or sweet, umami, salty or sour tastants (d). b, At concentrations 100-fold higher
than those needed to trigger maximal aversion in control animals, T2R52/2
mice begin to
show behavioural aversion. This probably reflects the recruitment of a low-sensitivity
receptor for cycloheximide; total ablation of the bitter taste system (for example, PLCb22/2
animals; open circles) eliminates this residual response. Values are means ^ s.e.m.;
n  3 (a), n  8 (b­d). See Supplementary Information for responses to additional
tastants in T2R52/2
animals.
Figure 3 T2R-expressing cells are the mediators of bitter taste. a, Taste preferences of
control (black), PLCb22/2
(grey), and transgenic `rescue' lines that express PLCb2
under the control of mT2R5, mT2R32 or mT2R19 (shades of red) were measured
relative to water using a brief-access test against a panel of bitter tastants. The
rescue line mT2R19 was engineered as a two component Tet-on system9
. In the
absence of doxycycline, the mT2R19 rescue mice (open red bars) do not taste bitter
compounds. PLCb2 knockout completely eliminates bitter taste behaviour10
. However,
expression of PLCb2 under the control of each of the three different T2R promoters
restores normal bitter taste to a broad panel of chemically diverse bitter tastants; this
demonstrates broadly tuned bitter sensing for the cells that express T2R5, T2R19
andT2R32, and demonstrates that a large repertoire of T2Rs is expressed in each
T2R-expressing cell1
. Values shown are means ^ s.e.m. (n  6­12). For additional
data and tastants, see Supplementary Fig. 3 b, T2R regulatory sequences drive
PLCb2 expression in T2R-expressing cells. Shown is double labelling of T2R32 driving
PLCb2 expression in a PLCb22/2
background, demonstrating coexpression of PLCb2
(immunoreactivity, green) and T2Rs (in situ hybridization, red); dotted lines highlight
taste receptor cells. The in situ signal is concentrated around the cell nucleus,
whereas immunostaining is primarily cytoplasmic.
letters to nature
NATURE | VOL 434 | 10 MARCH 2005 | www.nature.com/nature 227
 2005 Nature Publishing Group

cells are narrowly tuned to different subsets of bitter tastants11
, then
mice with restricted PLC expression would regain bitter sensitivity
to a discrete repertoire of bitter-tasting compounds.
To address this question we chose three divergent T2Rs mapping
to different chromosomal locations (T2R5, T2R19 and T2R32,
ref. 1) and used their regulatory sequences to drive PLCb2
expression in PLCb22/2
lines. We then tested the engineered
animals against a broad panel of chemically diverse bitter com-
pounds (Fig. 3 and Supplementary Fig. 3). In addition, we designed
one of the three constructs to have an inducible expression system,
in order to determine whether T2R signalling plays a critical role in
the development and/or connectivity of T2R cells. Remarkably, all
of the PLCb2-rescued animals showed completely normal bitter
taste (Fig. 3), including those in which T2R function was restored
only at the adult stage, long after the taste system would have
completed its normal development and wiring programme12
. These
results show that the bitter taste circuitry can be established without
bitter sensory input, and unequivocally demonstrate that individual
T2R cells operate as broadly tuned bitter sensors.
Recently, we generated mice expressing a RASSL13
k-opioid
receptor in sweet cells and showed that these animals become
selectively attracted to the synthetic opioid agonist spiradoline, a
normally tasteless compound9
. Thus, activation of sweet cells, rather
than the sweet taste receptor itself, results in the perception of
sweetness. Does the same logic apply to bitter taste? We tested this
idea by generating mice in which an inducible RASSL receptor was
now targeted to bitter taste cells. Figure 4 shows that non-induced
animals, or wild-type controls treated with doxycycline, are com-
pletely insensitive to the RASSL agonist spiradoline, even at milli-
molar concentrations. In contrast, mice expressing RASSL in bitter
cells show strong aversion to spiradoline. Together, these results
substantiate the coding of both sweet and bitter pathways by
dedicated (that is, labelled) lines. A final corollary emerging from
these findings is that expression of a sweet receptor in bitter cells
should trigger behavioural aversion to sweet tastants, and
expression of a bitter receptor in sweet cells should result in
attraction to the bitter compound. Accordingly, we engineered
mice expressing the bitter receptor for b-glucopyranosides in
sweet cells. Indeed, these mice now display strong attraction to
this family of bitter compounds. Thus, the `taste' of a sweet or a
bitter compound (that is, the perception of sweet and bitter) is a
reflection of the selective activation of T1R-expressing versus T2R-
expressing cells, rather than a property of the receptors or even the
tastant molecules. A
Methods
Gene targeting of T2R5
T2R5 knockout mice were generated by homologous recombination following standard
procedures9,10
. The entire coding sequence of T2R5 was replaced by a reverse-tetracycline
dependent transactivator (rtTA) and a loxP-flanked PGK-neor
cassette (see
Supplementary Fig. 1 for details). Homologous recombination in R1 embryonic stem (ES)
cells was detected by diagnostic Southern hybridization with probes outside the targeting
construct. Two targeted ES clones were injected into C57BL/6 blastocysts. Chimaeric mice
were bred with C57BL/6 mice and progeny backcrossed to C57BL/6 mice for two
generations before establishing a homozygous knockout colony.
Transgenic animals
Transgenic lines were produced by pronuclear injection of zygotes from FVB/N or CB6
(BALB/c  C57BL/6 hybrids) mice. For constructs using the Tet-on inducible system,
tetracycline-dependent gene expression was induced by feeding animals a diet containing
doxycycline (6 g kg21
) (Bio-Serv) for 3 days before and during behavioural testing9
. For
each construct, at least two independent animal lines were generated. The T2R38
transgenic construct used the PAV-taster allele5
. For transgenic constructs using T2R
regulatory sequences, we used the following fragments: T2R5, 210816 to 3 (the location
of the ATG start codon)10
; T2R19, 211012 to 3; T2R32, 29506 to 3. In situ
hyrbridization and immunhistochemistry were carried out as described previously10
.
Behavioural assays
Taste behaviour was assayed using a short-term assay that directly measures taste
preferences by counting immediate licking responses in a multi-channel gustometer
(Davis MS160-Mouse gustometer; DiLog Instruments)14
. Before training and behavioural
testing, all mice with a T2R52/2
or PLCb22/2
background were treated with intranasal
zinc sulphate15
to reduce input through the olfactory system. Mice were trained and tested
as described previously9,10
. Salt attraction to 150 mM NaCl was measured in mice that had
been salt deprived overnight16
. Lick response represents the mean percentage rate at which
mice licked a tested compound relative to their sampling of an appropriate control tastant;
relative responses were scaled to the mean response of control animals. The concentrations
of tastants used for bar graphs were: 300 mM sucrose; 100 mM glutamate  1 mM inosine
monophosphate 0.1 mM amiloride (MSG); 150 mM NaCl (attraction); 150 mM citric
acid; 10 mM cycloheximide; 10 mM quinine; 10 mM denatonium; 10 mM 6-n-propyl-2-
thiouracil (PROP); 10 mM papaverine; 10 mM quinacrine; 1 mM cholchicine, 5 mM
atropine.
Standard two-bottle preference assays were carried out as described previously17
. For
the two-bottle assays in Fig. 4b, consumption relative to total is defined as intake of tastant
divided by total intake (tastant plus water). For mice carrying rtTA and TetO-transgenes,
controls included testing the same mice without induction as well as mice carrying just the
rtTA transgene and exposed to doxycycline.
Nerve recordings
Lingual stimulation and recording procedures were performed as previously described18,19
.
Neural signals were amplified (5,000) with a Grass P511 AC amplifier (Astro-Med),
digitized with a Digidata 1200B A/D converter (Axon Instruments), and integrated with a
time constant of 0.5 s. Taste stimuli were presented at a constant flow rate of 4 ml min21
for 20 s intervals, interspersed by 2 min rinses with artificial saliva20
between presentations.
All data analyses used the integrated response over a 25 s period immediately after the
application of the stimulus. The mean response to 100 mM NH4Cl was used to normalize
responses to each experimental series.
Tastants used for nerve recordings (maximal concentrations) were: 60 mM
acesulfameK (AceK); 100 mM citric acid; 100 mM NaCl; 100 mM NH4Cl; 10 mM
quinine; 1 mM cycloheximide.
Figure 4 Switching the behavioural taste responses of mice by mis-expression of a bitter
taste receptor. a, Expression of RASSL13
in T2R-expressing cells generates animals that
show specific behavioural aversion to spiradoline (red). No responses are seen in control
mice that carry the rtTA transgene but lack the RASSL receptor (open grey circles) or in
non-induced RASSL mice (filled grey circles), even at 100 times the concentration needed
to elicit strong responses in RASSL-expressing animals. b, Transgenic mice expressing
human T2R16 (under control of the mT2R19 promoter) in bitter cells show strong
behavioural aversion to phenyl-b-D-glucopyranoside (open red squares). Expression of
the same receptor in sweet cells (T1R2-expressing cells) generates animals that show
strong attraction for this bitter tastant (filled red circles). Control animals (filled grey circles)
or non-induced experimental animals (open grey circles) show no preference or aversion
to this tastant. Values are means ^ s.e.m. (n  8).
letters to nature
NATURE | VOL 434 | 10 MARCH 2005 | www.nature.com/nature228
 2005 Nature Publishing Group

Received 3 December 2004; accepted 10 January 2005; doi:10.1038/nature03352.
1. Adler, E. et al. A novel family of mammalian taste receptors. Cell 100, 693­702 (2000).
2. Matsunami, H., Montmayeur, J. P. & Buck, L. B. A family of candidate taste receptors in human and
mouse. Nature 404, 601­604 (2000).
3. Chandrashekar, J. et al. T2Rs function as bitter taste receptors. Cell 100, 703­711 (2000).
4. Bufe, B., Hofmann, T., Krautwurst, D., Raguse, J. D. & Meyerhof, W. The human TAS2R16 receptor
mediates bitter taste in response to beta-glucopyranosides. Nature Genet. 32, 397­401 (2002).
5. Kim, U. K. et al. Positional cloning of the human quantitative trait locus underlying taste sensitivity to
phenylthiocarbamide. Science 299, 1221­1225 (2003).
6. Behrens, M. et al. The human taste receptor hTAS2R14 responds to a variety of different bitter
compounds. Biochem. Biophys. Res. Commun. 319, 479­485 (2004).
7. Pronin, A. N., Tang, H., Connor, J. & Keung, W. Identification of ligands for two human bitter T2R
receptors. Chem. Senses 29, 583­593 (2004).
8. Kuhn, C. et al. Bitter taste receptors for saccharin and acesulfame K. J. Neurosci. 24, 10260­10265
(2004).
9. Zhao, G. Q. et al. The receptors for mammalian sweet and umami taste. Cell 115, 255­266 (2003).
10. Zhang, Y. et al. Coding of sweet, bitter, and umami tastes: different receptor cells sharing similar
signaling pathways. Cell 112, 293­301 (2003).
11. Caicedo, A. & Roper, S. D. Taste receptor cells that discriminate between bitter stimuli. Science 291,
1557­1560 (2001).
12. Farbman, A. I. Electron microscope study of the developing taste bud in rat fungiform papilla. Dev.
Biol. 11, 110­135 (1965).
13. Coward, P. et al. Controlling signaling with a specifically designed Gi-coupled receptor. Proc. Natl
Acad. Sci. USA 95, 352­357 (1998).
14. Glendinning, J. I., Gresack, J. & Spector, A. C. A high-throughput screening procedure for identifying
mice with aberrant taste and oromotor function. Chem. Senses 27, 461­474 (2002).
15. McBride, K., Slotnick, B. & Margolis, F. L. Does intranasal application of zinc sulfate produce anosmia
in the mouse? An olfactometric and anatomical study. Chem. Senses 28, 659­670 (2003).
16. Franchini, L. F., Rubinstein, M. & Vivas, L. Reduced sodium appetite and increased oxytocin gene
expression in mutant mice lacking beta-endorphin. Neuroscience 121, 875­881 (2003).
17. Nelson, G. et al. Mammalian sweet taste receptors. Cell 106, 381­390 (2001).
18. Dahl, M., Erickson, R. P. & Simon, S. A. Neural responses to bitter compounds in rats. Brain Res. 756,
22­34 (1997).
19. Nelson, G. et al. An amino-acid taste receptor. Nature 416, 199­202 (2002).
20. Danilova, V. & Hellekant, G. Comparison of the responses of the chorda tympani and
glossopharyngeal nerves to taste stimuli in C57BL/6J mice. BMC Neurosci. 4, 5 (2003).
21. Nakashima, K. & Ninomiya, Y. Transduction for sweet taste of saccharin may involve both inositol
1,4,5-trisphosphate and cAMP pathways in the fungiform taste buds in C57BL mice. Cell. Physiol.
Biochem. 9, 90­98 (1999).
Supplementary Information accompanies the paper on www.nature.com/nature.
Acknowledgements We thank F. Liu for help generating T2R5 knockout animals. We are also
grateful to A. Cho and S. Taduru for generation of transgenic lines, and to A. Leslie for preparation
of antibodies. We thank members of the Zuker and Ryba laboratories for valuable comments and
advice. This work was supported in part by a grant from the National Institute on Deafness and
Other Communication Disorders to C.S.Z. C.S.Z. is an investigator of the Howard Hughes
Medical Institute.
Competing interests statement The authors declare that they have no competing financial
interests.
Correspondence and requests for materials should be addressed to C.S.Z.
(charles@flyeye.ucsd.edu) or N.J.P.R. (nick.ryba@nih.gov).
..............................................................
Spatial bistability of Dpp­receptor
interactions during Drosophila
dorsal­ventral patterning
Yu-Chiun Wang1
& Edwin L. Ferguson1,2
1
Department of Organismal Biology and Anatomy and 2
Department of Molecular
Genetics and Cell Biology, University of Chicago, Chicago, Illinois 60637, USA
.............................................................................................................................................................................
In many developmental contexts, a locally produced morphogen
specifies positional information by forming a concentration
gradient over a field of cells1
. However, during embryonic
dorsal­ventral patterning in Drosophila, two members of the
bone morphogenetic protein (BMP) family, Decapentaplegic
(Dpp) and Screw (Scw), are broadly transcribed but promote
receptor-mediated signalling in a restricted subset of expressing
cells2­4
. Here we use a novel immunostaining protocol to visualize
receptor-bound BMPs and show that both proteins become
localized to a sharp stripe of dorsal cells. We demonstrate that
proper BMP localization involves two distinct processes. First,
Dpp undergoes directed, long-range extracellular transport. Scw
also undergoes long-range movement, but can do so indepen-
dently of Dpp transport. Second, an intracellular positive feed-
back circuit promotes future ligand binding as a function of
previous signalling strength. These data elicit a model in which
extracellular Dpp transport initially creates a shallow gradient
of BMP binding that is acted on by positive intracellular
feedback to produce two stable states of BMP­receptor inter-
actions, a spatial bistability in which BMP binding and signal-
ling capabilities are high in dorsal-most cells and low in lateral
cells.
The combined activities of both dpp and scw are necessary for
phosphorylation of the BMP signal transducer Mad and ultimately
for the specification of dorsal tissues in the Drosophila embryo3­5
.
Although dpp is expressed uniformly over the dorsal 40% of the
embryonic circumference and scw is ubiquitously expressed, at the
onset of gastrulation (stage 6) phosphorylated Mad (pMad) stain-
ing is restricted to the dorsal 10% of cells that comprise the
extraembryonic amnioserosa2­4
(Fig. 1c). However, the pattern of
pMad staining is dynamic during development. pMad staining is
initially visualized during early and mid-stage 5 as a broad, shallow
gradient centred on the dorsal midline (Fig. 1a, b). Over the 30 min
period between mid-stage 5 and stage 6, the intensity of pMad
staining refines, decreasing laterally but significantly increasing
dorsally6,7
(Fig. 1d).
Because this pattern of BMP signalling does not result from
restricted expression of BMP receptors8­11
, we determined whether
it reflects region-specific ligand­receptor interaction by developing
a technique, called perivitelline injection (PVI), to visualize
secreted, receptor-bound Dpp. We injected anti-green fluorescent
protein (GFP) antibody into the embryonic perivitelline space, the
extracellular space between the cell membrane and the vitelline
membrane, to detect a biologically active, GFP-tagged form12
of
Dpp expressed under control of the even-skipped stripe 2 (eve-st2)
enhancer13
(Fig. 1e). Because injected antibody does not have access
to the cytoplasm, it can only interact with secreted Dpp­GFP. After
fixation and removal of the vitelline membrane, only antibody
bound to Dpp­GFP that is associated with the embryo will be
visualized.
Using PVI, we observed that Dpp­GFP is localized to a stripe on
the dorsal side of the embryo in a pattern distinctly different from
that obtained by conventional immunostaining, which visualizes
Dpp­GFP at its site of production, probably in the secretory
pathway (Fig. 1f­h). PVI initially detects low levels of Dpp­GFP
at mid-stage 5 in a broad dorsal domain centred on the eve-st2
region (Fig. 1g). By early stage 6, a narrow stripe of Dpp­GFP, which
extends over the entire anterior­posterior length of the embryo
(Fig. 1h), closely mirrors the spatial extent of pMad staining
(Fig. 1j). The immunofluorescence appears to be associated with
cell membranes and in punctate structures within cells (Fig. 1i),
representing Dpp­GFP bound to the cell surface or internalized,
presumably in association with its receptor. The ability to detect the
dorsal stripe of Dpp­GFP with conventional immunostaining after
PVI (Fig. 1k) suggests that PVI stabilizes an otherwise transient
process of ligand­receptor interaction and endocytic processing.
Using PVI, we confirmed that an HA-tagged form of Dpp expressed
in its endogenous domain also localizes dorsally (Fig. 1l). These
findings demonstrate that Dpp undergoes long-range, directional
extracellular transport to become localized to the dorsal-most
cells.
Two secreted proteins, the ventro-laterally expressed Short gas-
trulation (Sog) and the dorsally expressed Twisted gastrulation
(Tsg), form a tripartite complex with Dpp that blocks Dpp­receptor
letters to nature
NATURE | VOL 434 | 10 MARCH 2005 | www.nature.com/nature 229
 2005 Nature Publishing Group