Abstract
Extracellular ATP is implicated in numerous sensory processes ranging from the response to pain to the regulation of motility in visceral organs1. The ATP receptor P2X3 is selectively expressed on small diameter sensory neurons2, 3, 4, supporting this hypothesis. Here we show that mice deficient in P2X3 lose the rapidly desensitizing ATP-induced currents in dorsal root ganglion neurons. P2X3 deficiency also causes a reduction in the sustained ATP-induced currents in nodose ganglion neurons. P2X3-null mice have reduced pain-related behaviour in response to injection of ATP and formalin. Significantly, P2X3-null mice exhibit a marked urinary bladder hyporeflexia, characterized by decreased voiding frequency and increased bladder capacity, but normal bladder pressures. Immunohistochemical studies localize P2X3 to nerve fibres innervating the urinary bladder of wild-type mice, and show that loss of P2X3 does not alter sensory neuron innervation density. Thus, P2X3 is critical for peripheral pain responses and afferent pathways controlling urinary bladder volume reflexes. Antagonists to P2X3 may therefore have therapeutic potential in the treatment of disorders of urine storage and voiding such as overactive bladder.
The ion-channel subunit P2X3 is one of seven known subunits
that form homomeric and heteromeric receptors for ATP5. Uniquely,
P2X3 is expressed by a subgroup of small sensory neurons of the
dorsal root and cranial ganglia2, 3, 4. ATP and 
,
-Me-ATP
(a P2X1,3-selective analogue) excite many small calibre afferent
neurons from skin, joints and viscera6. Studies on ATP release6, 7 and P2 receptor antagonism8, 9 support the involvement
of P2X receptors in nociception and the micturition reflex. To investigate
the role of P2X3 in sensory processing, we generated P2X3
-deficient mice carrying a 1-kb deletion of the P2X3 gene
encompassing exon 1 and the initiating codon ATG (Fig. 1a, b).
Figure 1: Targeted disruption of the P2X3 gene and immunolocalization studies.
a, Gene targeting strategy. R, EcoR1; Bg, Bgl I; S,
Sac I; K, Kpn I; X, Xho I; N, Not I; Tk, thymidine
kinase. b, Southern blot of P2X3+/+, P2X
3+/- and P2X3-/- mice using
a 5'-flanking region probe shown in a. c–j,
Colocalization of P2X3 with neuronal markers in DRG, spinal cord
and skin of P2X3+/+ and P2X3-/-
mice. c, d, Transverse sections (10 
m)
of L5 DRG immunostained for P2X3 (green) and P2X2 (red).
In P2X3+/+ DRG, P2X3 and P2X2
immunoreactivities are present in small–medium and medium–large
cells, respectively. P2X2 staining appears unaltered in P2X
3-/- DRG. e–h, Transverse sections
of L5 DRG (15 m; e, f) and lumbar spinal cord
(20 m; g, h) immunostained for P2X3
(red) and IB4 lectin binding (green). In P2X3+/+
DRG, nearly all P2X3-immunoreactive cells bind IB4 (colocalization,
yellow) and IB4 staining appears unaltered in P2X3-/-
DRG. In P2X3+/+ spinal cord, P2X3 and
IB4 staining terminals are co-localized (yellow) in inner lamina II of the
dorsal horn, with apparently normal distribution of IB4 staining in P2X
3-/- spinal cord. i, j, Sections
(15 m) of hindpaw plantar skin immunostained for P2X3 (red)
and the pan-neuronal marker PGP 9.5 (green). In P2X3+/+
skin, P2X3 and PGP 9.5 immunoreactivities are colocalized
(yellow) in some fine epidermal (E) fibres, and to a lesser extent in nerve
bundles in the dermis (D). Epidermal innervation is still evident by PGP 9.5
immunoreactivity in P2X3-/- mice. Scale bars:
c–f, i, j, 25 m; g,
h, 75 m.
Normally, P2X3 is selectively expressed by small sensory neurons marked by the lectin IB4 (refs 4, 10). In P2X3-/- mice, P2X3 immunoreactivity is undetectable in dorsal root ganglion (DRG), spinal cord and peripheral tissues (Fig. 1d, f, h, j). However, staining for IB4 in DRG and spinal cord (Fig. 1e–h), and protein gene product (PGP) 9.5 pan-neuronal staining in the skin (Fig. 1i , j) appears to be unaltered in P2X3 -/- mice. Thus, loss of P2X3 does not appear to affect either peripheral or central innervation patterns. Immunostaining for P2X receptor subunits P2X2, P2X5 and P2X6 was also qualitatively unchanged in DRG, trigeminal and nodose ganglia of P2X 3-/- mice (P2X2 in DRG, Fig. 1c, d; and data not shown), indicating no compensatory changes in other P2X receptor subunits. P2X1, P2X4 and P2X7 receptors were not detected in mouse sensory ganglia.
We analysed dissociated DRG and nodose ganglion neurons using electrophysiology
(Fig. 2a–d). In P2X3+/+
DRG, 75% (27/38) and 60% (18/30) of neurons tested responded to ATP and ,-Me-ATP,
respectively. ATP-responsive DRG neurons showed either a rapidly
desensitizing inward current (18/27, averaging 0.4 
0.08 nA, Fig. 2a), or a slowly desensitizing response (9/27, averaging
0.4 0.22 nA, data not shown). No P2X3
-/- DRG neurons (0/34) responded to either ATP or ,-Me-ATP
with rapidly desensitizing currents (Fig. 2b). The proportion
of DRG neurons with slowly desensitizing responses to ATP (4/34,12%) was not
significantly different (P > 0.05) from that observed in wild-type
mice, nor was the proportion of cells responding to GABA (
-aminobutyric
acid) or capsaicin (data not shown). Thus, P2X3 homomers appear
to be principally responsible for rapidly desensitizing ATP-activated currents
in DRG. However, P2X2 homomers and P2X2/3 heteromers
may function in a minority of DRG neurons11, 12, 13, 14, 15.
In contrast, in P2X3 wild-type mice, >90% (34/37) of nodose ganglion
neurons responded to ATP and ,-Me-ATP with slowly desensitizing
persistent responses (Fig. 2c). In null-mutant
mice, the proportion of nodose neurons responding to ATP (43/51, 84%)
was similar (P > 0.1). However, the mean amplitude of the response
(3.2 0.6 nA for 100 M ATP, n
= 26) was significantly less (P < 0.05) than that observed
for P2X3+/+ neurons (5.2 0.5 nA
for 100 M ATP, n = 31) (Fig. 2d).
None of the P2X3-/- nodose neurons tested (0/12)
responded to ,-Me-ATP (Fig. 2d). Together
with previous evidence3, 16, these data suggest that nodose
ganglion neurons contain significant proportions of homomeric P2X2
and heteromeric P2X2/3 channels.
Figure 2: Responses to nucleotide agonists and nociceptive behaviour in P2X 3-deficient mice.
a–d, Whole-cell patch-clamp recordings of DRG and nodose
ganglion neurons at a holding potential of -60 mV. a, P2X
3+/+ DRG neurons show rapidly desensitizing inward currents
in response to 10 M ATP and 30 M ,-Me-ATP.
b, P2X3-/- DRG neurons failed to produce a transient
response to either 300 M ATP or 30 M ,-Me-ATP.
c, P2X3+/+ nodose ganglion neurons show
slowly desensitizing inward currents in response to 100 M ATP
and ,-Me-ATP. d, P2X3-/- nodose
ganglion neurons responded to 100 M ATP, but not to 100 M ,-Me-ATP.
e, ATP-evoked behavioural responses. Hindpaw-lifting responses in 2–3-month-old
male P2X3+/+ (filled square) and P2X3
-/- (open square) mice (n = 10 to 15) following intraplantar
injection of varying doses of ATP in a total volume of 20 l.
Double asterisk, P < 0.01 for P2X3+/+
and P2X3-/- mice; analysis of variance. The response
to 500 nmol ATP was also measured 20 min following pre-treatment with a
200 mg per kg body weight intraperitoneal injection of PPADS
in P2X3+/+ (filled circle) and P2X3
-/- (open circle) mice (n = 10 to 15). Asterisk,
P < 0.05 for P2X3+/+ and P2X3
-/- mice; Student's t-test (significance shown only for P2X
3-/-). f, Formalin-induced behavioural responses.
Hindpaw-lifting and licking responses in 3–4-month-old male P2X
3+/+ (open bars) and P2X3-/-
(filled bars) mice (n = 10) following intraplantar injection of 5%
formalin in a total volume of 20 l. Double asterisk, P <
0.01 for P2X3+/+ and P2X3-/-
mice; Student's t-test.
We next examined sensory deficits in P2X3 wild-type and null-mutant
mice. No differences were observed in locomotor activity and rotorod performance
(data not shown). Injection of ATP into the hindpaw evoked a nociceptive behavioural
response (intermittent hindpaw lifting, licking and biting, as in the rat17) in P2X3 wild-type mice that was dose dependent (Fig. 2e). In P2X3 null-mutant mice (
Fig. 2e), responses were significantly decreased by 77% and 45% to
100 and 500 nmol of ATP, respectively. Altered pain responses
were specific to ATP and not seen with intraplantar injections of 30 g
capsaicin (data not shown). Moreover, the non-selective P2 receptor antagonist
pyridoxalphosphate-6-azophenyl-2',4'-disulphonic acid (PPADS)
further reduced the residual hindpaw lifting behaviour of P2X3
-/- mice by about 50% (Fig. 2e). These data
suggest that the pain-producing effects of peripheral ATP in humans18, 19
and animals17, 20 are mainly mediated by P2X3 subunits.
However, some of this response appears to be derived from other P2 receptors.
Responses to noxious thermal and mechanical stimuli were similar in P2X
3 wild-type and null-mutant mice (data not shown). In contrast, pain-related
behaviour was significantly attenuated in both phases of the formalin test
(
50%) in null-mutant mice (Fig. 2f), consistent
with previous work8.
We also investigated the role of P2X3 in urinary bladder sensory
function. Bladder afferent activity during filling drives micturition contractions
mediated by the central nervous system (CNS)21, 22. Therefore,
we monitored these reflex contractions in P2X3 wild-type and null-mutant
mice using two different urodynamic methods. Figure 3a
shows representative cystometrograms from conscious mouse cystometry studies
in which voiding reflexes were measured in response to a continuous intravesical
infusion of saline. P2X3-null mice had significantly decreased
micturition frequencies (mean void intervals 9.0 0.8
min versus 5.3 0.3 min in P2X3
+/+ mice) (Fig. 3b, left), and significantly
increased bladder capacities (mean void volumes 0.41 0.04
ml versus 0.23 0.02 ml in P2X3
+/+ mice) (Fig. 3b, middle), but showed no differences
in bladder pressures recorded at baseline, micturition threshold (not shown)
and micturition peak (Fig. 3b, right). Male and female
mice showed similar urodynamic changes. Additionally, cystometric differences
are not attributable to influence from the 129Sv background (
Fig. 3b).
Figure 3: Bladder cystometry in P2X3-deficient mice.
a, Representative cystometrograms from conscious 5–6-month-old
P2X3+/+ and P2X3-/- mice.
Traces illustrate bladder pressure recorded in response to a constant intravesical
infusion of saline (50 l min-1), and accumulated
void volumes recorded from each micturition (scale bar, 2 min). b,
Void intervals, void volumes and void pressures were quantified for C57BL/6
(n = 7), 129Sv (n = 6), P2X3+/+
(n = 4 males and 4 females) and P2X3-/- (
n = 4 males and 4 females) mice. P2X3-/- mice
had significantly decreased micturition frequencies (increased void interval)
and significantly increased bladder capacities (increased void volume), but
no differences in bladder pressures. Double asterisk, P < 0.01 for
P2X3+/+ and P2X3-/- mice;
analysis of variance. c, Representative acute cystometrograms recorded
from anaesthetized and transurethrally catheterized 5–6-month-old P2X
3+/+ and P2X3-/- mice. Each
cystometrogram consisted of intravesical infusion of saline (20 l min
-1 for 15 min) (scale bar, 2 min). Contractions greater than
10 cm H20 were taken as micturition contractions. d,
Quantification of the average number of contractions per cystometrogram for
P2X3+/+ (n = 8) and P2X3
-/- (n = 11) mice confirms the altered micturition reflex in
P2X3-/- mice. Asterisk, P < 0.05; Student's
t-test.
Acute cystometry carried out under anaesthesia also showed micturition
hyporeflexia in P2X3-null mice. Bladder contractions measured in
response to distension with a fixed infusion rate of saline (
Fig. 3c) resulted in frequent micturition contractions in P2X
3+/+ mice, but virtually no contractions in P2X
3-/- mice, up to the cut-off volume. The average number
of contractions per cystometrogram was significantly reduced in P2X3
-/- mice (0.6 0.38 compared with
4.9 2.37 in P2X3+/+ mice)
(Fig. 3d). Accordingly, the micturition threshold was
significantly increased (0.29 0.01 compared with 0.21 0.02 ml
in P2X3+/+, P < 0.05), and only 23% of
P2X3-/- mice reached the micturition threshold compared
with 75% of P2X3+/+ mice. No differences were observed
in the intravesical pressure at a volume of 0.3 ml (data not shown).
Finally, we detected P2X3 immunoreactivity on sensory neurons innervating the suburothelial nerve plexus of wild-type mouse bladder (Fig. 4c), both on small nerve fibres with terminals embedded in the urothelium, as well as on nerve bundles. Bladders from P2X3-null mice showed no P2X3 immunoreactivity (Fig. 4d), but were otherwise histologically normal (Fig. 4a, b), and showed no alterations in sensory innervation patterns as measured by capsaicin receptor (VR-1) immunoreactivity ( Fig. 4e, f).
Figure 4: Immunolocalization in the mouse urinary bladder.
a, b, Haemotoxylin and eosin stained transverse sections
(10 m) showing different layers of the urinary bladder (near trigone):
LU, lumen; U, urothelium; SU, suburothelium; SM, smooth muscle. Loss of P2X
3 does not cause degenerative or hyperplastic changes in P2X3
-/- bladder. c, d, Confocal images of whole mount bladder
exposing the suburothelial sensory nerve plexus. In P2X3
+/+ bladder P2X3 immunoreactivity is detected on small
nerve fibres with terminals embedded in the urothelium, and on a large
nerve bundle (c). In P2X3-/- bladder, P2X
3 immunoreactivity is absent (d), but sensory innervation to
the urinary bladder appears to be intact as evidenced by staining for the
sensory neuron-specific capsaicin (VR-1) receptor (e, f). Scale
bars, 50 m.
Our findings demonstrate the importance of P2X3 receptors in
somatic and visceral sensory function. First, we show that much of the DRG
response to ATP is mediated by homomeric P2X3 receptors, while
in nodose ganglion neurons homomeric P2X2 and heteromeric P2X
2/3 receptors appear most important. Second, our formalin test data
are consistent with a role for ATP activation of P2X3 in mediating
some nociceptive responses to tissue damage. Finally, we show that P2X
3 is critical in regulating micturition reflex excitability. One explanation
for these data is that ATP, released in response to stretch during distension
and filling of the urinary bladder, excites primary afferent voiding circuitry
through direct interaction with P2X3 receptors. ATP is released
from rabbit urothelium in response to stretch7, and P2X
3 is clearly present on nerve fibres innervating urinary bladder23 (Fig. 4). Electrophysiological evidence also
indicates that ,-Me-ATP directly activates and desensitizes mechanosensitive
pelvic afferents arising from rat urinary bladder9. Thus, loss
of P2X3 might impair sensory neuron activity during bladder filling,
raising the volume threshold for activation of the micturition reflex. As
loss of compliance and lowered volume thresholds are a component of many bladder
storage disorders (for example, overactive bladder)24, selective
modulation of P2X3 may provide new therapies. The potential for
similar P2X3 roles in mechanosensation in other hollow organs (for
example, GI tract and lung)25 needs to be explored.
Methods
Physiological studies
F2 and F3
mice were used for in vitro and in vivo studies, respectively.
All experiments were performed blind. Dissociation of neurons and whole-cell
patch-clamp recording was carried out as described previously26.
Agonists were applied rapidly by microperfusion from a 4-barrel manifold controlled
by computer-driven solenoid valves. Exchange of solution around the cell was
complete in less than 100 ms. Time between applications was 2 min
(nodose) and 3.5 min (DRG), allowing sufficient time to achieve reproducible
responses. The minimum detectable response was 20 pA. Traces were acquired
using FETCHEX (pCLAMP V.6.04 software, Axon Instruments), and plotted using
ORIGIN V.4.1 (Microcal). Pain-related responses to injection of ATP into the
hindpaw were measured essentially as described for rat17. The
hindpaw lifting time was measured for a total of 4 min following injection
of ATP. Thermal sensitivity was assessed using a radiant heat stimulus and
tail immersion in a 52 °C water bath. Mechanical sensitivity was
assessed using a set of calibrated von Frey filaments. For the formalin test,
the hindpaw lifting and licking time was measured for a total of 30 min. Conscious
mouse cystometry was performed essentially as described for rat27.
Recovery following catheter implantation was for 7 days, and intravesical
saline infusion was at a rate of 50 l min-1.
For transurethral cystometry, bladder reflexes were assessed in urethane-anesthetized
mice essentially as described for rat28. Each cystometrogram
consisted of intravesical distension to a total volume of 0.3 ml, at
a rate of 20 l min-1. Contractions greater
than 10 cm of H20 were taken as micturition contractions.
Generation of P2X3 receptor-deficient mice and immunohistochemistry methods are described in Supplementary Information.
