The capsaicin receptor : a heat activated ion channel in the pain pathway

articles
The capsaicin receptor: a
heat-activated ion channel
in the pain pathway
Michael J. Caterina*, Mark A. Schumacher†k, Makoto Tominaga*k, Tobias A. Rosen*, Jon D. Levine‡ & David Julius*
Departments of * Cellular and Molecular Pharmacology, † Anesthesia, and ‡ Medicine, University of California, San Francisco, California 94143-0450, USA
k These authors contributed equally to this study.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Capsaicin, the main pungent ingredient in ‘hot’ chilli peppers, elicits a sensation of burning pain by selectively
activating sensory neurons that convey information about noxious stimuli to the central nervous system. We have used
an expression cloning strategy based on calcium in?ux to isolate a functional cDNA encoding a capsaicin receptor from
sensory neurons. This receptor is a non-selective cation channel that is structurally related to members of the TRP
family of ion channels. The cloned capsaicin receptor is also activated by increases in temperature in the noxious
range, suggesting that it functions as a transducer of painful thermal stimuli in vivo.
Pain is initiated when the peripheral terminals of a subgroup of
has led to its use as a high-af?nity radioligand to visualize saturable,
sensory neurons are activated by noxious chemical, mechanical or
capsaicin- and capsazepine-sensitive binding sites on nociceptors16.
thermal stimuli. These neurons, called nociceptors, transmit infor-
A more detailed understanding of the molecular nature of
mation regarding tissue damage to pain-processing centres in the
capsaicin action and its relationship to endogenous pain signalling
spinal cord and brain1. Nociceptors are characterized, in part, by
mechanisms might be obtained through the cloning of a gene
their sensitivity to capsaicin, a natural product of capsicum peppers
encoding a capsaicin receptor. To achieve this we used a functional
that is the active ingredient of many ‘hot’ and spicy foods. In
screening assay to isolate a cDNA clone that reconstitutes capsaicin
mammals, exposure of nociceptor terminals to capsaicin leads
responsiveness in non-neuronal cells. The deduced amino-acid
initially to excitation of the neuron and the consequent perception
sequence of this clone demonstrates that the capsaicin receptor is
of pain and local release of in?ammatory mediators. With pro-
an integral membrane protein with homology to a family of putative
longed exposure, nociceptor terminals become insensitive to cap-
store-operated calcium channels. The cloned receptor seems to be
saicin, as well as to other noxious stimuli2. This latter phenomenon
of nociceptor desensitization underlies the seemingly paradoxical
use of capsaicin as an analgesic agent in the treatment of painful
disorders ranging from viral and diabetic neuropathies to rheuma-
toid arthritis3,4. Some of this decreased sensitivity to noxious stimuli
may result from reversible changes in the nociceptor, but the long-
term loss of responsiveness can be explained by death of the
nociceptor or destruction of its peripheral terminals following
exposure to capsaicin2,5.
The cellular speci?city of capsaicin action and its ability to evoke
the sensation of burning pain have led to speculation that the target
of capsaicin action plays an important physiological role in the
detection of painful stimuli. Indeed, capsaicin may elicit the
perception of pain by mimicking the actions of a physiological
stimulus or an endogenous ligand produced during tissue injury6.
Although the excitatory and neurotoxic properties of capsaicin
have been used extensively to de?ne and study nociceptive
neurons, its precise mechanism of action has remained elusive.
Electrophysiological7,8 and biochemical9 studies have shown that
capsaicin excites nociceptors by increasing the permeability of the
plasma membrane to cations, but the molecular mechanism under-
lying this phenomenon is unclear. Proposed models range from the
direct perturbation of membrane lipids by hydrophobic capsaicin
Figure 1 Expression cloning of a capsaicin receptor using calcium imaging.
molecules10 to the activation of a speci?c receptor on or within
HEK293 cells transiently transfected with pools of clones from a rodent dorsal root
sensory neurons6. Because capsaicin derivatives show structure–
ganglion (DRG) cDNA library were subjected to microscopic ?uorescent calcium
function relationships and evoke responses in a dose-dependent
imaging before (left) and during (right) treatment with 3 ?M capsaicin. Cells
manner11,12, the existence of a receptor site represents the most likely
transfected with vector alone (pCDNA3; top) exhibited no response to capsaicin.
mechanism. This model has been strengthened by the development
Between 1% and 5% of cells transfected with pool 11 exhibited marked increases
of capsazepine, a competitive capsaicin antagonist13, and by the
in cytoplasmic calcium (middle, arrowheads). This pool was iteratively subdivided
discovery of resiniferatoxin, an extremely potent capsaicin analogue
and reassayed until a single positive clone (VR1) was isolated (bottom). Elevated
from Euphorbia plants that mimics the cellular actions of
relative calcium concentrations are indicated by an increased ratio of Fura-2
capsaicin14,15. The potency of resiniferatoxin at nanomolar quantities
emission at 340 versus 380 nm wavelength excitation (see colour bar).
Nature © Macmillan Publishers Ltd 1997
816
NATURE | VOL 389 | 23 OCTOBER 1997

---

articles
expressed exclusively by small-diameter neurons within sensory
(ref. 18), and microscopically examined for capsaicin-evoked
ganglia, providing a de?nitive molecular explanation for the
changes in intracellular calcium levels. A positive pool was identi?ed
remarkable selectivity of capsaicin action. This receptor is also a
(Fig. 1, middle) and iteratively subdivided and reassayed. In this
thermal sensor that is strongly activated when ambient tempera-
way, an individual clone containing a 3-kilobase (kb) cDNA insert
tures are elevated to a range known to elicit pain in humans or pain-
was obtained that, by itself, conferred capsaicin (Fig. 1, bottom) or
associated behaviours in animals. Thus capsaicin elicits burning
resiniferatoxin (not shown) sensitivity to transfected HEK293 cells.
sensations through the activation of a heat-gated ion channel that
Because a vanilloid moiety constitutes an essential chemical com-
is likely to contribute to the detection of painful thermal stimuli
ponent of capsaicin and resiniferatoxin structures, the proposed site
in vivo.
of action of these compounds is more generally referred to as the
vanilloid receptor16. Accordingly, we have named the newly cloned
Expression cloning of receptor cDNA
cDNA VR1, for vanilloid receptor subtype 1.
The lack of speci?c information regarding the molecular structure
of capsaicin receptors prompted us to adopt a functional screening
VR1 and vanilloid receptor pharmacology
strategy for isolating candidate cDNA clones. A mammalian cell
To compare the pharmacological properties of the cloned receptor
expression cloning strategy was devised on the basis of the ability of
to those of native vanilloid sites in sensory ganglia, we expressed
capsaicin to trigger robust calcium in?ux into sensory neurons in
VR1 in Xenopus oocytes and used whole-cell voltage-clamp analysis
vitro9,17. We reasoned that a capsaicin receptor-encoding cDNA
to quantitatively examine the electrophysiological responses to a
might confer upon non-neuronal cells a similar ability to undergo
variety of vanilloid agonists and antagonists. At negative holding
increases in intracellular free calcium upon exposure to capsaicin,
potentials, exposure to capsaicin or resiniferatoxin produced dose-
assuming that capsaicin acts at a proteinaceous site and that a single
dependent inward current responses in VR1-expressing oocytes, but
cDNA can confer sensitivity to capsaicin in a heterologous context.
not in water-injected control cells (Fig. 2a). As observed in sensory
Because capsaicin responsiveness seems to be con?ned to nocicep-
neurons19,20, capsaicin-evoked current responses returned rapidly to
tive neurons with cell bodies that reside within sensory ganglia5, a
baseline after agonist removal, whereas resiniferatoxin responses
cDNA library was constructed from dorsal root ganglion-derived
often failed to recover, even after a prolonged washout period.
messenger RNA. This library was subdivided into pools of approxi-
Half-maximal effective concentrations for these agonists were
mately 16,000 clones, and each pool was transiently transfected into
within an order of magnitude of those reported for native
human embryonic kidney-derived HEK293 cells. Transfected cells
vanilloid receptors8,13, with resiniferatoxin being approximately 20-
were then loaded with the ?uorescent calcium-sensitive dye Fura-2
fold more potent than capsaicin (EC ¼ 39:1 nM and 711.9 nM,
50
a
100
capsaicin (1
resiniferatoxin (0.1
µM)
µM)
80
esponse
60
40
20
% Maximal r
0
-9
-8
-7
-6
-5
200 nA
Agonist (log M)
20 s
b
cap
cap + cpz
cap
100
80
esponse
60
400 nA
20 s
cap
40
cap + RR
cap
20
% Capsaicin r
0
-9
-8
-7
-6
-5
-4
Capsazepine (log M)
1
00 nA
20 s
Figure 2 VR1 responds to puri?ed vanilloids and pepper extracts. a, Activation of
curve is shown to the right (n ¼ 4 independent oocytes for each point). Current
VR1 by capsaicin and resinferatoxin. Left, agonists were applied sequentially to
responses were normalized to that elicited by capsaicin alone in each oocyte.
the same Xenopus oocyte expressing VR1. Membrane currents were recorded in
(0.6 ?M, open diamond). Current tracing at bottom left shows reversible block of a
the whole-cell voltage-clamp con?guration. Bars denote duration of agonist
capsaicin (0.6 ?M)-evoked response by ruthenium red (RR; 10 ?M). Slash marks
application. Right, concentration–response curve for capsaicin (?lled squares)
denote washout periods of 2 and 12 min, respectively (n ¼ 3). c, Responses to
and resiniferatoxin (open circles). Membrane currents were normalized in each
capsaicin (10 ?M) and extracts derived from four varieties of peppers in oocytes
oocyte to a response obtained with 1 ?M capsaicin and expressed as a percent of
expressing VR1 (30 s application). Bottom right, relative potencies of each pepper
maximal response to capsaicin. Each point represents mean values (?s.e.m.)
extract are plotted (mean ? s:e:m:, n ¼ 3). Values were normalized in each cell to
from ?ve independent oocytes. The Hill equation was used to ?t the response
responses obtained with capsaicin (10 ?M). Extracts evoked no responses in
data. b, Antagonism by capsazepine (cpz) and ruthenium red (RR). Current tracing
water-injected cells. Reported pungencies for pepper varieties (in Scoville units)
at top left shows reversible block of capsaicin (cap; 0.6 ?M) response by
are: Habanero (H), 100,000–300,000; Thai green (T), 50,000–100,000; wax (W),
capsazepine (cpz; 10 ?M) after 2 min pretreatment. Slash marks represent
5,000–10,000; and Poblano verde (P),1,000–1,500 (ref. 23). Capsaicin (C) is rated as
washout periods of 2 and 3 min, respectively (n ¼ 3). A capsazepine inhibition
16 ? 106 units.
Nature © Macmillan Publishers Ltd 1997
NATURE | VOL 389 | 23 OCTOBER 1997
817

---

articles
respectively). Hill coef?cients derived from these analyses (1.95 and
2.08, respectively) suggest that full activation of the receptor
involves the binding of more than one agonist molecule, again
consistent with previously described properties of native vanilloid
receptors8,16.
Capsaicin-evoked responses in
VR1-expressing
oocytes were reversibly blocked by the competitive vanilloid recep-
tor antagonist capsazepine at concentrations (IC ¼ 283:5 nM)
50
that inhibit native receptors13 (Fig. 2b). Another pharmacological
characteristic of vanilloid receptors, is their sensitivity to the non-
competitive antagonist ruthenium red17, which blocked capsaicin-
evoked responses in a reversible manner (Fig. 2b). Responses to
resiniferatoxin (50 nM) were also reversibly antagonized by capsa-
zepine (5 ?M) or ruthenium red (10 ?M) (not shown).
As has been recognized for years, the relative pungencies of
pepper varieties span an enormously wide range, re?ecting, in
part, differences in vanilloid content. Methods for rating peppers
with respect to their relative ‘hotness’ have hitherto relied on
subjective psychophysical assays21 or on the biochemical determi-
nation of capsaicin content22. To further explore the connection
between the biology and biochemistry of vanilloid action, we sought
to determine whether the cloned vanilloid receptor could respond
electrophysiologically to pepper extracts in proportion to their
ability to evoke pain. Ethanol extracts were prepared from several
capsicum varieties and their potencies relative to a saturating dose
of capsaicin (10 ?M) were determined in the oocyte expression
system (Fig. 2c). Indeed, we found that the different ‘hotness’ of
these pepper variants, as determined by subjective psychophysical
ratings23, correlated with their rank order potencies as activators of
VR1.
Figure 3 VR1 is a calcium-permeable, non-selective cation channel. Electro-
physiological properties of capsaicin-activated currents in VR1-transfected
mammalian HEK293 cells. a, VR1 currents are time independent, outwardly
rectifying, and cation speci?c. A capsaicin-evoked inward current response (top)
was analysed using a series of 400-ms step pulses (?100 to +40 mV; middle, left).
Baseline currents (denoted as 1 on top trace) were subtracted from responses in
the presence of agonist (2) to yield a series of agonist-evoked responses at
different holding potentials (middle, right). These responses show outward
recti?cation when plotted as a function of membrane voltage (bottom). Calcium-
free standard bath solution and a caesium aspartate-?lled recording electrode
were used. b, Capsaicin elicits non-selective cation currents in VR1-transfected
cells. Voltage ramps (?100 to +40 mV in 500 ms) were used to generate current–
voltage curves in bath solutions with the indicated cationic compositions.
Recording electrodes were ?lled with NaCl. Similar results were obtained with
KCl- or CsCl-?lled electrodes. Replacement of extracellular NaCl (140 mM) with
equimolar KCl or CsCl did not signi?cantly shift reversal potential (Erev ¼
? 0:7 ? 1:2 mV, n ¼ 8; ? 1:5 ? 1:0 mV, n ¼ 9; ? 4:3 ? 0:9 mV, n ¼ 8, respectively;
P =P
=P
K
Na ¼ 0:94; PCa
Na ¼ 0:85). Replacement of extracellular NaCl with isotonic
(112 mM) MgCl2 or CaCl2 shifted Erev to 14:4 ? 0:7 mV (n ¼ 5) or 24:3 ? 2:3 mV
(n ¼ 7), respectively (P
=P
=P
Mg
Na ¼ 4:99;
PCa Na ¼ 9:60). c, Whole-cell current
responses evoked by repeated capsaicin applications show desensitization in
calcium-containing standard bath solution, but not in calcium-free solution.
Capsaicin (1 ?M) was applied every 5 min and CsCl was used as pipette solution.
The ratios of current size at the end of the third application to the peak of the ?rst
application were 95:3 ? 2:6% (n ¼ 3) in calcium-free solution, and 13:0 ? 4:3%
(n ¼ 5) in calcium-containing solution (t-test; P ? 0:00001). d–f, Single-channel
properties of capsaicin-evoked responses. Inside-out (I/O) or outside-out (O/O)
Figure 4 Capsaicin induces death of cells expressing the vanilloid receptor. a,
patches were excised from VR1-transfected cells and analysed in symmetrical
HEK293 cells were transiently transfected with either vector alone (pCDNA3), VR1
140 mM NaCl. d, Traces obtained from a single O/O patch before, during and after
cDNA diluted 1 : 50 in pCDNA3, or VR1 cDNA alone. After 7 h at 37 ?C in the
capsaicin (1 ?M) application to the bath solution (V
presence of capsaicin (3 ?M, black bars) or vehicle (0.3% ethanol, white bars), the
hold ¼ þ40 mV). Note multiple
simultaneous channel openings in the presence of capsaicin. e, Traces obtained
percentage of dead cells was determined. Data represent mean ? s:e:m: of
in the presence of capsaicin at the indicated holding potentials. Broken lines
triplicate determinations from a representative experiment. Asterisks indicate a
indicate the closed-channel level. No agonist-evoked channel activity was seen in
signi?cant difference from ethanol-treated cells (t-test, P ? 0:0001). b, Phase-
cells transfected with vector alone (n ¼ 8, not shown). f, Current–voltage curve of
contrast photomicrographs of parallel cultures transfected with pCDNA3 or VR1
mean single-channel amplitudes (?s.e.m.) calculated from data shown in e, also
(1 : 50) before (left) or 4 h after (right, +CAP) addition of capsaicin (3 ?M). Note the
exhibits pronounced outward recti?cation.
cytoplasmic swelling and eccentric position of cytoplasmic contents (arrows).
Nature © Macmillan Publishers Ltd 1997
818
NATURE | VOL 389 | 23 OCTOBER 1997

---

articles
To explore the possibility that capsaicin mimics the action of
currents in VR1-expressing cells. Current–voltage relations estab-
a known chemical modulator of nociceptor function, we tested
lished for cells bathed in solutions of differing cationic composi-
agents known to activate sensory neurons for their ability to evoke
tions show that VR1 does not discriminate among monovalent
responses in HEK293 cells or oocytes expressing VR1. None of the
cations, but exhibits a notable preference for divalent cations
agents tested gave positive responses, including adenosine tripho-
(permeability
sequence:
Ca2þ ? Mg2þ ? Naþ ? Kþ ? Csþ)
sphate (50 ?M), serotonin (10 ?M), acetylcholine (300 ?M),
(Fig. 3b). The very high relative permeability of VR1 to calcium
bradykinin (1 ?M), substance P (10 ?M), histamine (10 ?M),
ions (P =P
¼ 9:60; P =P
¼ 4:99) exceeds that observed for
Ca
Na
Mg
Na
glutamate (100 ?M), and hypertonic saline (600 mOsm).
most non-selective cation channels, and is similar to values reported
for NMDA-type glutamate receptors and a7 nicotinic acetylcholine
VR1 ion channel has high Ca2+ permeability
receptors (P =P
¼ 10:6 and 20, respectively)24,25, both of which
Ca
Na
To characterize more fully the electrophysiological properties of the
are noted for this property. With all bath solutions examined, an
cloned receptor at both whole-cell and single-channel levels, we
outwardly rectifying current–voltage relation was observed,
performed a series of patch-clamp studies on transfected mamma-
although this feature was less prominent in bath solutions contain-
lian cells expressing VR1. In the whole-cell con?guration, VR1-
ing MgCl2 or CaCl2.
transfected HEK293 cells showed robust inward current responses
In cultured sensory neurons, electrophysiological analyses of
(at a holding potential of ?60 mV) that developed with a short
vanilloid-evoked responses have shown them to be kinetically
latency upon bath application of capsaicin (Fig. 3a). No such
complex and to desensitize with continuous vanilloid exposure20,26.
currents were observed in cells transfected with vector alone (not
This electrophysiological desensitization (which might underlie
shown). In calcium-free medium, the capsaicin-evoked current did
aspects of physiological desensitization produced by vanilloids in
not vary with time, either at a constant holding potential of ?60 mV
vivo) seems to depend, in part, on the presence of extracellular
or during voltage steps from ?100 to +40 mV (in increments of
calcium26,27. Indeed, in the absence of extracellular calcium, capsai-
20 mV) (Fig. 3a). This property enabled us to characterize capsaicin-
cin-evoked responses in VR1-transfected cells showed little or no
mediated currents under steady-state response conditions in sub-
desensitization during prolonged agonist application or with suc-
sequent experiments. Current–voltage relations derived from these
cessive agonist challenges (4:7 ? 2:3% decrease between ?rst and
data show that such responses exhibit prominent outward recti?ca-
third applications, n ¼ 3 (Fig. 3c)). In contrast, responses evoked in
tion resembling that observed in cultured dorsal root ganglion
calcium-containing bath solution consisted of at least two distinct
neurons8 (Fig. 3a, bottom). Because the observed reversal potential
components, one desensitizing (87 ? 4:3% decrease between ?rst
was close to 0 mV (E
¼ 0:5 ? 0:9 mV, n ¼ 13), the capsaicin-
and third applications, n ¼ 5) and one relatively non-desensitizing.
rev
mediated response must involve the opening of a cation-selective
Thus desensitization and multiphasic kinetics of vanilloid-evoked
channel. In sensory neurons, vanilloid-evoked currents are carried
responses can be reproduced without a neuronal context and can be
by a mixture of monovalent and divalent cations7–9, and we there-
distinguished by their dependence on ambient calcium levels.
fore conducted a series of ion substitution experiments to examine
The behaviour of the VR1 response was also examined in
the relative contributions of various cations to capsaicin-evoked
membrane patches excised from transfected cells. In the presence
Figure 5 VR1 resembles store-operated channels. a, Predicted amino-acid
sequence encoded by the vanilloid receptor cDNA VR1. Open boxes delineate
ankyrin repeat domains, black boxes predicted transmembrane domains, the
grey box a possible pore-loop, and ?lled circles predicted protein kinase A
phosphorylation sites. b, Predicted membrane topology and domain structure of
VR1. Outer (o) and inner (i) plasma membrane lea?ets are indicated. c, Alignment
of VR1 with related sequences. Identical residues are in black boxes and
conservative substitutions are in grey. A partial sequence from a human EST is
shown. Other sequences are from members of the putative store-operated
calcium channel family. The Genbank accession number of the Drosophila TRP
protein is p19334; others are indicated.
Nature © Macmillan Publishers Ltd 1997
NATURE | VOL 389 | 23 OCTOBER 1997
819

---

articles
of capsaicin (but not its absence), large and well-resolved currents of
vector alone were not killed by this treatment. The cell death was
unitary amplitude were observed (n ¼ 31; Fig. 3d, e), indicating the
characterized by prominent cytoplasmic swelling, coalescence of
existence of capsaicin-gated ion channels within these patches
cytoplasmic contents, and eventual lysis. Thus VR1 expression in a
whose activation does not depend upon soluble cytoplasmic com-
non-neuronal context can recapitulate the cytotoxicity observed in
ponents. The current–voltage relation at the single-channel level
vanilloid-treated sensory neurons. Staining with Hoechst dye 33342
was almost identical to that established in whole-cell con?guration,
revealed no evidence of the nuclear fragmentation often associated
owing to its outward recti?cation and reversal potential near 0 mV
with apoptotic cell death29 (not shown). Together, these observa-
under similar ionic conditions (Fig. 3f). Unitary conductances of
tions are consistent with necrotic cell death resulting from excessive
76.7 pS at positive potentials and 35.4 pS at negative potentials were
ion in?ux, as has been proposed for vanilloid-induced death of
observed with sodium as the sole charge carrier. These single-
nociceptors7, glutamate-induced excitotoxicity30, and neuro-
channel properties are like those previously described for native
degeneration caused by constitutively activating mutations of
vanilloid receptors8,28. It has been suggested that the site of vanilloid
various ion channels31.
action may not be con?ned to the extracellular side of the plasma
membrane, owing to the lipophilic nature of these compounds6. We
VR1 resembles TRP-related ion channels
found that capsaicin was able to produce identical responses when
The VR1 cDNA contains an open reading frame of 2,514 nucleo-
added to either side of a patch excised from a cell expressing VR1
tides that encodes a protein of 838 amino acids with a predicted
(Fig. 3e), consistent with the notion that vanilloids can permeate or
relative molecular mass of 95,000 (Mr 95K) (Fig. 5a). Hydrophilicity
cross the lipid bilayer to mediate their effects. A less likely but
analysis suggests that VR1 is a polytopic protein containing six
formally consistent explanation is that vanilloid receptors have
transmembrane domains (predicted to be mostly ?-sheet) with an
functionally equivalent capsaicin-binding sites on both sides of
additional short hydrophobic stretch between transmembrane
the plasma membrane.
regions 5 and 6 (Fig. 5b). The amino-terminal hydrophilic segment
(432 amino acids) contains a relatively proline-rich region followed
Capsaicin kills cells that express VR1
by three ankyrin repeat domains. The carboxy terminus (154 amino
Capsaicin is an excitatory neurotoxin that selectively destroys
acids) contains no recognizable motifs.
primary afferent nociceptors in vivo and in vitro5,9. Is this selective
A homology search of protein databases revealed signi?cant
toxicity solely a re?ection of the speci?city of vanilloid receptor
similarities between VR1 and members of the family of putative
expression, or does it depend on additional properties of sensory
store-operated calcium channels (SOCs), the prototypical members
neurons or their environment? To address this question, we sought
of which include the Drosophila retinal proteins TRP and TRPL32,33
to determine whether capsaicin could kill non-neuronal cells that
(Fig. 5c). Members of this family have been proposed to mediate the
express vanilloid receptors in vitro. We found that, within several
entry of extracellular calcium into cells in response to the depletion
hours of continuous exposure to capsaicin, HEK293 cells trans-
of intracellular calcium stores34. These proteins resemble VR1 with
fected with VR1 died, as determined morphologically and by
respect to their predicted topological organization and the presence
staining with vital dyes (Fig. 4). In contrast, cells transfected with
of multiple N-terminal ankyrin repeats33. There is also striking
Figure 6 Vanilloid receptor expression is restricted to sensory neurons. a,
(kilobases) are shown on the left. b, In situ hybridization detects VR1 expression
Northern blot analysis shows that VR1 transcripts are con?ned to sensory
in a subset of sensory ganglion cells. Adult rat dorsal root ganglia (DRG) and
ganglia. Poly(A)+ RNAs were prepared from adult rats, except for the DRG-P1
trigeminal ganglia (TG) were probed with a digoxigenin-labelled, VR1-derived
sample, which was isolated from dorsal root ganglia of newborn pups. The blot
antisense riboprobe. Positive staining (purple) was con?ned to smaller-diameter
was probed with 32P-labelled VR1 cDNA, then re-probed with a 32P-labelled
cell bodies (arrows) and absent from large-diameter cell bodies (arrowheads).
cyclophilin cDNA to control for loading (bottom). Molecular size markers
Control (sense) riboprobes did not stain these tissues (not shown).
Nature © Macmillan Publishers Ltd 1997
820
NATURE | VOL 389 | 23 OCTOBER 1997

---

articles
amino-acid sequence similarity between VR1 and TRP-related
expressed in trigeminal and dorsal root sensory ganglia, both of
proteins within and adjacent to the sixth transmembrane region,
which contain capsaicin-sensitive neurons (Fig. 6a). This transcript
including the short hydrophobic region between transmembrane
was absent from all other tissues examined, including spinal cord
domains 5 and 6 that may contribute to the ion permeation path33.
and brain. A much smaller RNA species (?1.5 kb) was detected in
Outside these regions, VR1 shares little sequence similarity with
the kidney, but it is unclear whether this transcript could encode a
TRP family members, suggesting that its evolutionary relationship
functional VR1 protein. In situ hybridization histochemistry was
to these proteins is distant. Given the high permeability of VR1 to
used to assess the cellular pattern of VR1 expression within sensory
calcium ions, we nonetheless considered the possibility that it might
ganglia (Fig. 6b). These experiments clearly show that within dorsal
function as a SOC. To test this, we examined calcium-dependent
root and trigeminal ganglia, VR1 expression predominates in a
inward currents in VR1-expressing oocytes whose intracellular
subset of neurons with small diameters. This is in keeping with the
calcium stores had been depleted by treatment with the compound
observation that most capsaicin-sensitive neurons have cell bodies
thapsigargin. In water-injected control oocytes, a clear depletion-
of relatively small to medium size5,27. In contrast to the prominent
induced current was seen, as previously described35 (not shown). In
expression of VR1 transcripts in neurons of the dorsal root gang-
VR1-expressing oocytes, no quantitative or qualitative differences
lion, no visible signal was observed in the adjacent spinal cord dorsal
were observed in this response (not shown). Moreover, application
horn (not shown). Although binding sites for radiolabelled resini-
of SKF 96365 (20 ?M), an inhibitor of depletion-stimulated cal-
feratoxin have been detected in the dorsal horn, they are believed to
cium entry36, had no effect on capsaicin-evoked currents in VR1-
reside on presynaptic terminals that project from primary nocicep-
expressing oocytes (not shown). Thus VR1 does not seem to be a
tors with cell bodies located in the dorsal root ganglia27. Our results
functional SOC under these circumstances.
support this interpretation. Two other tissues that have been
An expressed sequence tag (EST) database homology search
proposed to express capsaicin receptors are the nodose ganglion,
revealed several human clones with a high degree of similarity to
which contains cell bodies of visceral nociceptors27, and the preoptic
VR1 at both the nucleotide and predicted amino-acid levels. The
area of the hypothalamus2, which is involved in thermoregulation37.
similarity of one of these clones to the corresponding region of VR1
By using in situ hybridization methods, we did not detect VR1
(Fig. 5c) is extremely high (68% amino-acid identity and 84%
expression at either location. Although these tissues might express
similarity within the region shown), suggesting that it is likely to be
VR1 at levels below the detection limit of our assay, vanilloid
the human VR1 orthologue or a closely related subtype. Human
responsiveness here might be conferred by distinct VR subtypes.
EST clones corresponding to other domains of VR1 show compar-
Indeed, VR heterogeneity has been proposed on the basis of
able degrees of similarity (not shown), and could represent frag-
biochemical studies4,16.
ments of the same human transcript.
Vanilloid receptor activated by noxious heat
Sensory neuron-speci?c expression of VR1
The ‘burning’ quality of vanilloid-induced pain suggests that
The highly selective nature of capsaicin action suggests that vanil-
vanilloids and heat may evoke painful responses through a
loid receptors serve as speci?c molecular markers for nociceptive
common molecular pathway. We therefore explored the effects of
neurons. Indeed, northern blot analysis showed that a mRNA
elevated temperature on VR1 activity. In initial studies, transfected
species of approximately 4 kb is prominently and exclusively
HEK293 cells were subjected to ?uorescent calcium imaging during
Figure 7 VR1 is activated by noxious thermal stimuli. a, HEK293 cells transiently
transfected with VR1, but not vector alone (pcDNA3), exhibit a pronounced
increase in cytoplasmic free calcium in response to heat. Cells were analysed
before and immediately after addition of heated buffer (300 ?l CIB at 65 ?C was
applied to cells in 150?l CIB at 22 ?C). Under these conditions, cells were
transiently exposed to a peak temperature of ?45 ?C. Relative calcium
concentrations are indicated by the colour bar, as in Fig. 1. b, Whole-cell patch-
clamp analysis (Vhold ¼ ? 60 mV) of VR1-transfected HEK293 cells reveals inward
current responses to both heat and capsaicin. The temperature of the bath
b
medium was raised from 22 to 48 ?C (heat), and then restored to 22 ?C, after
heat
capsaicin
which capsaicin (0.5 ?M) was added to the bath (left trace). Ionic conditions were
heat
2
identical to those described in Fig. 3a. Voltage ramps (?100 to +40 mV in 500 ms)
capsaicin
1
)
were applied before, between and during responses. Stimulus-induced current–
A
Voltage (mV)
voltage relations are shown on the right. c, VR1 expressed in Xenopus oocytes is
t

(
n
n
-100 -80 -60 -40
20
40
activated by noxious but not innocuous heat. Oocytes injected with either VR1
r
r
e
-1
u
C
cRNA or water were subjected to two-electrode voltage-clamp while the perfu-
h
c
-2
sate temperature was raised from 22.7 ?C to the level indicated, then held constant
1 nA 10 s
for 60 s. The magnitudes of the resulting inward currents are shown as the
c
d
mean ? s:e:m: (VR1, n ¼ 8; water, n ¼ 6 independent cells). The asterisk indicates
a signi?cant difference from water-injected oocytes (t-test, P ? 0:0005). d, Ruthe-
*
VR1
2,000
nium red (RR) inhibits heat-evoked responses in VR1-expressing oocytes. The
VR1
cap
heat
heat + RR
heat
current tracings shown were generated from representative VR1- or water-
1,500
injected oocytes during successive applications of the indicated stimuli. VR1-
1,000
ent (nA)
injected oocytes exhibited the following mean inward current responses ?s.e.m.
water
(n ¼ 5): capsaicin (cap, 1
Curr
500
?M), 1;221 ? 148 nA; heat (50 ?C), 2;009 ? 134 nA; heat
water
plus RR (10 ?M), 243 ? 47 nA. Inhibition by RR was signi?cant (88 ? 2%, n ¼ 5;
0
paired t-test, P
20
30
40
50
µ
A
cap
heat
heat + RR
? 0:0001). In the absence of RR, no diminution in current was
1
Temperature (oC)
observed with successive heat pulses (not shown). Water-injected oocytes
40 s
showed no response to capsaicin and much smaller responses to heat
(338 ? 101 nA, n ¼ 5). RR inhibited these responses by only 21 ? 26% (n ¼ 5;
paired t-test, P ? 0:1).
Nature © Macmillan Publishers Ltd 1997
NATURE | VOL 389 | 23 OCTOBER 1997
821

---

articles
a sudden increase in ambient temperature from 22 ?C to ?45 ?C.
VR1-expressing oocytes to heat, whereas the smaller response seen
Under these conditions, cells transfected with vector alone exhibited
in control cells was reduced by only 21 ? 26% (n ¼ 5; Fig. 7d).
only a mild, diffuse change in cytoplasmic free calcium (Fig. 7a,
Taken together, these observations strongly support the hypothesis
left). In contrast, a large proportion of cells expressing VR1
that VR1 is activated by noxious, but not innocuous, heat.
exhibited a pronounced increase in calcium levels within seconds
of heat treatment (Fig. 7a, right). These responses subsided within a
Protons may be endogenous modulators of VR1
few minutes, and a subsequent challenge with capsaicin produced a
A reduction in tissue pH resulting from infection, in?ammation or
characteristic calcium response (not shown), suggesting that the
ischaemia can produce pain in mammals. It has therefore been
response to heat is a speci?c signalling event and not a consequence
proposed that protons might act as endogenous activators or
of nonspeci?c membrane perturbation or disruption of cell integ-
modulators of vanilloid receptors38–40. To address this possibility,
rity. To determine whether speci?c heat-evoked membrane currents
we examined the effects of hydrogen ions on the cloned vanilloid
are associated with this phenomenon, VR1-expressing cells were
receptor using the oocyte expression system. We investigated
examined using patch-clamp methods. Exposure of these cells to a
whether an abrupt reduction in bath solution pH, from 7.6 to 5.5,
rapid increase in temperature (22 ?C to ?48 ?C in 25 s, monitored
was suf?cient to activate VR1 in the absence of capsaicin. Fewer
using an in-bath thermistor) produced large inward currents
than 10% of VR1-expressing oocytes treated in this way exhibited a
(791 ? 235 pA at ?60 mV, n ¼ 9) that were typically similar in
large inward current (not shown), suggesting that hydrogen ions
amplitude to that evoked by a subsequent application of capsaicin at
alone cannot ef?ciently activate this protein. We next assessed the
500 nM (Fig. 7b). Both heat- and vanilloid-evoked responses
effect of reduced pH on the responsiveness of VR1 to capsaicin.
showed outward recti?cation, suggesting that they are mediated
VR1-expressing oocytes were treated with a submaximal concen-
by the same entity (Fig. 7b). By comparison, thermally evoked
tration of capsaicin (300 nM) at pH 7.6 (Fig. 8). Once their current
responses of control, vector-transfected cells were much smaller
responses reached a relatively stable plateau, the oocytes were
(131 ? 23 pA, n ¼ 8) and exhibited no recti?cation (not shown).
exposed to a solution containing the same concentration of capsai-
The heat response in VR1-transfected cells desensitized during
cin at pH 6.3. Under these conditions, the inward current rapidly
stimulus application, whereas that observed in vector-transfected
increased to a new plateau up to ?vefold greater in magnitude than
cells did not. These results suggest t