Presynaptic ATP P2X receptors have been proposed to play a role in modulating glutamate release from the first sensory synapse in the spinal cord. Using spinal cord slice preparations and patch-clamp recordings from dorsal horn neurons in lamina V of the rat spinal cord, we showed that the activation of P2X receptors by α,β-methylene-ATP (αβm-ATP) resulted in a large increase in the frequency of spontaneous EPSCs (sEPSCs) and miniature EPSCs (mEPSCs). The increases in mEPSC frequency by αβm-ATP were not blocked by the Ca2+ channel blocker, 30 μmLa3+, but were abolished in a bath solution when Ca2+ was omitted. The increases in mEPSC frequency by αβm-ATP were blocked completely by the P2 receptor antagonist pyridoxalphosphate-6-azophenyl-2′,4′-disulfonic acid (PPADS) at 10 μm. Furthermore, the EPSCs evoked by dorsal root stimulation were potentiated by αβm-ATP as well as by the ecto-ATPase inhibitor ARL67156 and were depressed in the presence of P2 receptor antagonists PPADS (10 μm) and suramin (5 μm). The effects of these compounds on the evoked EPSCs were associated with the changes in glutamate release probability of primary afferent central terminals. Our results indicate that αβm-ATP-sensitive P2X receptors play a significant role in modulating excitatory sensory synaptic transmission in the spinal cord, and the potential role of endogenous ATP is suggested.
- purinergic receptors
- glutamate release
- primary afferent fibers
- patch-clamp technique
- spinal cord slice preparation
ATP P2X receptors, a family of nonselective cation channels gated by extracellular ATP (Jahr and Jessell, 1983; Krishtal et al., 1983; Khakh et al., 2001), may play an important role in somatic sensory transmission. P2X receptors are highly expressed on different functional types of primary sensory neurons (Vulchanova et al., 1997; Xiang et al., 1998; Guo et al., 1999;Li et al., 1999; Petruska et al., 2000). Immunocytochemical evidence shows the presence of different types of P2X receptors in both peripheral and central terminals of primary afferent fibers (Vulchanova et al., 1996, 1998; Guo et al., 1999; Novakovic et al., 1999). Activation of P2X receptors at peripheral sensory nerve endings may initiate sensory impulses and may be associated with nociceptive and non-nociceptive sensory signals (Cook et al., 1997; Sawynok and Reid, 1997; Dowd et al., 1998; Tsuda et al., 2000). At central terminals P2X receptors have been proposed to play a role in modulating glutamate release from sensory synapses, based on a previous study that used a coculture system of dorsal root ganglion (DRG) and dorsal horn (DH) neurons (Gu and MacDermott, 1997). Consistent with this hypothesis, ATP has been shown to have presynaptic effects in lamina II of the DH (Li and Perl, 1995), and activating presynaptic P2X receptors could have facilitative effects on sensory spinal transmission to lamina II of the DH (Li et al., 1998). Khakh and Henderson (1998) also have demonstrated a presynaptic P2X receptor-mediated facilitation of glutamate release at synapses in the brainstem. The role of endogenous ATP in modulating glutamate release from sensory afferent terminals was indicated (Khakh and Henderson, 1998; Li et al., 1998). However, it is currently not clear whether the central terminals of primary afferent fibers connecting to deep lamina DH neurons also express P2X receptors at presynaptic sites and, if so, whether the activation of these receptors may facilitate glutamate release.
To date, at least seven P2X subunits (P2X1 to P2X7) have been cloned (for review, see North and Surprenant, 2000). Except for P2X6, all P2X subunits can form functional homomeric receptors in different heterologous expression systems (Khakh et al., 2001). Among those homomers, P2X1 and P2X3 receptors mediate a rapidly desensitizing response to both ATP and the agonist α,β-methylene-ATP (αβm-ATP). The other five homomers mediate a nondesensitizing response to ATP and are essentially insensitive to αβm-ATP (Khakh et al., 2001). Coexpression of different P2X receptor subunits results in several functional heteromers, including P2X1+5, P2X2+3, P2X2+6, and P2X4+6 (Lewis et al., 1995; Le et al., 1998, 1999; Torres et al., 1998; Haines et al., 1999; King et al., 2000). Of those heteromers, P2X2+6 receptors are insensitive to αβm-ATP (King et al., 2000). Thus, αβm-ATP makes some functional P2X receptors pharmacologically distinguishable from others.
Of the seven cloned P2X receptor subunits, six (P2X1 to P2X6) are expressed in primary sensory neurons (Collo et al., 1996; Vulchanova et al., 1996, 1997, 1998; Xiang et al., 1998). In the spinal cord the mRNAs for P2X1, P2X2, P2X4, P2X5, and P2X6 subunits have been found; mRNAs for P2X2, P2X4, and P2X6 are the most abundant (Collo et al., 1996). Immunochemical studies on spinal cord sections with antibodies against P2X1, P2X2, and P2X3 subunits have shown the expression of these subunits on the primary afferent central terminals (Vulchanova et al., 1996, 1997, 1998; Guo et al., 1999). Other P2X receptor subunits also may be expressed at the primary afferent central terminals in the spinal cord. In the present study we have explored the functions of P2X receptors at some sensory synapses in deep laminas (lamina V) of the spinal cord.
MATERIALS AND METHODS
Spinal cord slice preparation. Transverse spinal cord slices (500 μm in thickness) were prepared from L5 spinal cords of rats at postnatal age 11–21 d, as described previously (Nakatsuka et al., 2000). Unless otherwise indicated, slices without dorsal roots attached were used for recordings of spontaneous EPSCs (sEPSCs) and miniature EPSCs (mEPSCs); slices with attached L5 dorsal roots were used for evoked EPSCs (eEPSCs). A spinal cord slice was transferred to a recording chamber (∼0.5 ml) and placed on the stage of an upright infrared-differential interference contrast (IR-DIC) microscope. The slice was superfused (10 ml/min) with Krebs' solution containing (in mm) 117 NaCl, 3.6 KCl, 2.5 CaCl2, 1.2 MgCl2, 1.2 NaH2PO4, 25 NaHCO3, and 11 glucose in 95% O2/5% CO2, pH 7.3, at 22°C.
Preparation of acutely dissociated DRG neurons. Dorsal root ganglia were removed from rats (age of 4–7 weeks) and placed in a 35°C bath solution containing dispase (neutral protease, 5 mg/ml; Boehringer Mannheim, Indianapolis, IN) and collagenase (2 mg/ml; Sigma type 1, St. Louis, MO). After 1 hr of incubation the DRGs were triturated to dissociate the neurons. The dissociated cells were plated on coverslips precoated with poly-l-lysine and allowed to adhere for 1 hr before recording.
Preparation of cultured DRG neurons. Dorsal root ganglia were isolated from rat embryos aged 16 d (E16) in utero, exposed to 0.25% trypsin for 20 min, and dissociated. The dissociated primary sensory neurons were plated on glass coverslips previously prepared with a monolayer of rat cortical astrocytes. At the time of plating, 2.5S NGF (10 ng/ml) and 5-fluoro-2′-deoxyuridine (10 μm) were added, and 2.5S NGF was added once every week when the cells were fed with fresh media. The cultures of 2–3 weeks were used for the experiments. Principles of Laboratory Animal Care (National Institutes of Health publication 86-23; revised, 1985) was followed in all of the tissue preparation procedures described above.
Patch-clamp recordings from spinal cord slices. Lamina regions were identified under a 10× objective, and individual neurons were identified with a 40× objective under IR-DIC microscope. Whole-cell patch-clamp recordings were made from DH neurons with electrodes (∼5 MΩ) filled with a solution containing (in mm) 135 K+-gluconate, 5 KCl, 0.5 CaCl2, 2 MgCl2, 5 EGTA, and 5 HEPES. Signals were amplified and filtered at 2 kHz (Axopatch 200B, Axon Instruments, Foster City, CA) and sampled at 5 kHz. Cells were held at −60 mV, which was close to the reversal potential for GABAA and glycine receptors under the experimental conditions. At this holding potential the outward IPSCs were minimized and usually were undetectable. sEPSCs were recorded in the absence of TTX. mEPSCs were recorded in the presence of 0.5 μm TTX (the term mEPSCs is used for simplicity; “sEPSCs in the presence of TTX” may be a more suitable term). In some experiments mEPSCs were recorded in the presence of 20 μm bicuculline and 2 μmstrychnine in bath solution. αβm-ATP, capsaicin, and other testing compounds described in Results were applied via bath solution. In some experiments the effects of 100 μm ATP on sEPSC frequency were tested in the presence of 50 μm ARL67156 plus 2 mm caffeine. The application intervals for testing compounds were 20 min. At this interval the agonist responses were reproducible. Analyses of sEPSCs and mEPSCs were performed as described previously (Gu and MacDermott, 1997).
To record eEPSCs from lamina V, we applied a stimulus (∼120 μA, 0.1 msec) to a dorsal root with a suction electrode. Monosynaptic connection was judged by constant latency of the eEPSCs when multiple stimuli were applied (Nakatsuka et al., 2000). Most lamina V neurons that we recorded had monosynaptic connections with primary afferent fibers (93% from 28 recordings) in our slice preparations. The stimulation condition usually yielded few synaptic failures. In the experiments to determine synaptic potentiation of eEPSCs by P2X receptor activation, we first identified monosynaptic eEPSCs and then gradually reduced the intensity (Malinow and Tsien, 1990) to a level at which some synaptic failures (30–70%) occurred. For higher sensitivity the cells with higher synaptic failure rates were assigned for tests with αβm-ATP or ARL67156, and the cells with lower failure rates were tested with P2X receptor antagonists. Synaptic failure was judged by current levels in the range of current noise baseline of Gaussian distribution (±3 pA). One to three cells were recorded from a slice of each rat.
Patch-clamp recordings from acutely dissociated or cultured DRG neurons. The coverslips with dissociated DRG neurons were mounted in a 0.5 ml recording chamber and placed on the stage of an OlympusIX70 microscope. Cells were perfused continuously with bath solution (22°C) flowing at 1 ml/min. The bath solution contained (in mm) 150 NaCl, 5 KCl, 2 MgCl2, 2 CaCl2, 10 glucose, and 10 HEPES, pH-adjusted to 7.4 with NaOH; osmolarity was adjusted to 320 mOsm with sucrose. Cells were voltage clamped at −70 mV in the whole-cell configuration. Signals were amplified and filtered at 2 kHz (Axopatch 200B) and sampled at 5 kHz. The recording electrode internal solution contained (in mm) 120 KCl, 5 Na2-ATP, 0.4 Na2-GTP, 5 EGTA, 2.25 CaCl2, 5 MgCl2, and 20 HEPES, pH-adjusted to 7.4 by KOH, with an osmolarity of 315–325 mOsm. Recording electrode resistance was between 2.0 and 5.0 MΩ. αβm-ATP (10 μm) or capsaicin (1 μm) was applied rapidly to neurons through a glass tube (inner diameter, ∼500 μm) positioned 1.0 mm away from the cell, and each was applied for 2 sec. The gravity-driven solution flow was controlled electronically by solenoid valves and triggered from a computer.
The potential nonspecific effects of pyridoxalphosphate-6-azophenyl-2′,4′-disulfonic acid (PPADS) on action potentials and Na+ and Ca2+channel activity were examined in cultured DRG neurons. Conditions, including the bath solution and electrode internal solutions for patch-clamp recordings from cultured DRG neurons, were similar to the recordings from acutely dissociated DRG neurons. Under current clamp the action potentials were evoked by current steps of 600 pA for 4 msec. The tests were performed both in normal bath solution and after a 10 min continuous perfusion of 50 μm PPADS in the bath solution. To examine the potential nonspecific effects of PPADS on Na+ and Ca2+ channel activity, we held the cells at −80 mV under perforated voltage-clamp configuration, and the electrode internal solution contained Cs+(Gu et al., 1996). Inward currents with Na+ and Ca2+ channel components (Gu and MacDermott, 1997) were evoked by voltage steps to +10 mV for 200 msec. The tests were performed both in normal bath solution and after a 10 min continuous perfusion of 50 μm PPADS in the bath solution.
ATP, αβm-ATP, ARL67156, PPADS, suramin, capsaicin, caffeine, bicuculline, strychnine, and LaCl3 were purchased from Sigma. CNQX and TTX were purchased from Tocris (St. Louis, MO). Unless otherwise indicated, data represent mean ± SEM; paired Student'st tests were used for statistical comparison, and significance was considered at the p < 0.05 level.
P2X receptor-mediated enhancement of spontaneous glutamate release
We performed patch-clamp recordings from DH neurons in lamina V of spinal cord slices. Under IR-DIC microscopy lamina regions in the dorsal horn were identified first with a 10× objective (Fig.1 A), and then individual neurons in lamina V were visualized under a 40× objective (Fig. 1 B). To test whether the activation of P2X receptors may release glutamate from afferent central terminals and/or DH interneuron terminals onto lamina V DH neurons, we bath-applied αβm-ATP (100 μm), a metabolic stable ATP analog and selective P2X receptor agonist, to determine its effects on the frequency of sEPSCs. Application of αβm-ATP (100 μm) produced a large increase in sEPSC frequency (Fig. 1 C,D) in most lamina V neurons (Fig.1 E). Of 23 cells that were recorded, 21 cells showed substantial increases in sEPSC frequency, and only two cells showed no increase after αβm-ATP (Fig. 1 E). The overall changes of sEPSC frequency were 355 ± 58% of control (p < 0.05, paired t test;n = 23). The effects lasted for ∼200 sec (202 ± 12 sec; n = 21) after a 60 sec αβm-ATP application (Fig. 1 D). Figure 1 E shows the changes of sEPSC frequency in all of the 23 lamina V neurons that were tested after αβm-ATP application. When αβm-ATP was applied repeatedly at 10 min intervals, it produced similar increases in sEPSC frequency (n = 10) (Fig. 1 F). However, when 100 μm ATP was applied for 60 sec, sEPSC frequency showed either a biphasic change with initially a transient increase (<20 sec) followed by a depression phase (n = 3) (Fig.2 A) or had only a depression phase (n = 5). This was most likely attributable to the rapid metabolism of ATP and the effects of its metabolite adenosine. To test this possibility, we next performed experiments in the presence of the ecto-ATPase inhibitor ARL67156 (50 μm) and the adenosine receptor antagonist caffeine (2 mm) (Salter et al., 1993; Westfall et al., 1996). Under this condition a 60 sec application of ATP (100 μm) produced a nearly threefold increase of sEPSC frequency (270 ± 30%; p < 0.05;n = 10) (Fig. 2 B) that lasted for ∼120 sec (123 ± 6 sec), a result similar to the effects of αβm-ATP on sEPSC frequency. In addition to the increases of sEPSC frequency, ATP also produced small inward currents (11 ± 7 pA) in 2 of 10 recorded neurons (Fig. 2 C). Because the use of ATP may have potentially more complications than the use of αβm-ATP (see Discussion), we used αβm-ATP as the P2X receptor agonist in the remaining study.
We tested whether αβm-ATP-sensitive terminals were connected monosynaptically with most lamina V neurons. This was done by determining the effects of αβm-ATP on sEPSCs in the presence of 500 nm TTX (mEPSCs will be used for simplicity in the following parts). The presence of TTX blocks active interneuronal transmission between DH neurons. Under this condition the bath application of 100 μm αβm-ATP for 60 sec still significantly increased mEPSC frequency (Fig. 3 A,B) in most neurons that were recorded (Fig. 3 C). Of 23 cells that were recorded, 22 of them showed large increases in mEPSC frequency, and one cell had little change (Fig. 3 C). The overall changes of mEPSC frequency were 382 ± 49% of control (p < 0.05, paired t test;n = 23). The effects lasted for ∼170 sec (174 ± 13 sec; n = 22). Of those cells showing the increases in mEPSC frequency by αβm-ATP, 19 of them also were tested with capsaicin (2 μm). mEPSC frequency was not affected by a 60 sec capsaicin application (n = 19) (Fig. 3 B,D).
To determine whether Ca2+ entry through P2X receptors at presynaptic terminals may contribute directly to the increases in spontaneous glutamate release, we preapplied 30 μm La3+ to block the potential Ca2+ entry through voltage-gated Ca2+ channels (Gu and MacDermott, 1997). Under this condition and in the presence of 500 nm TTX, 100 μm αβm-ATP still produced a large increase in the frequency of mEPSCs (273 ± 20% of control; p < 0.05; n = 4) (Fig.4 Aa). On the other hand, when experiments were performed in a 0 Ca2+bath solution (Ca2+ was omitted from normal bath solution), αβm-ATP did not produce a significant increase of mEPSC frequency (117 ± 16% of control; n = 5) (Fig.4 Ab). Whereas αβm-ATP increased mEPSC frequency in almost all lamina V neurons, mEPSC amplitude remained within 99 ± 2% of control (n = 26) (Fig. 4 B). We tested whether the increase of mEPSCs by αβm-ATP could be blocked by the P2X receptor antagonist PPADS (10 μm). In five cells 10 μm αβm-ATP-induced significant increases of mEPSCs (261 ± 25% of control; p < 0.05) in normal bath solution; in these five cells, when 10 μm PPADS was present, the effects of αβm-ATP on mEPSC frequency were abolished completely (99 ± 4% of control) (Fig. 4 C). CNQX (10 μm), a non-NMDA receptor antagonist, was tested on those cells for which αβm-ATP induced increases in the frequency of mEPSCs or sEPSCs. CNQX completely blocked mEPSCs (n= 5) (Fig. 4 D) as well as sEPSCs (data not shown;n = 5). These results suggest that P2X receptors are present at glutamatergic terminals monosynaptically contacting lamina V neurons.
Whole-cell currents directly evoked by αβm-ATP in sensory neurons
To explore whether αβm-ATP-sensitive terminals were derived mainly from the central terminals of primary afferent fibers that remained in the spinal sections or mainly from terminals of DH interneurons, we determined the expression of αβm-ATP-sensitive P2X receptors on DH neurons and DRG neurons. This was done by examining αβm-ATP-evoked whole-cell currents from DH or DRG neurons. In 101 spinal cord DH neurons, 55 in lamina V and 46 in lamina II, 100 μm αβm-ATP did not induce any detectable whole-cell current directly (Fig. 5 A). We next directly examined αβm-ATP-evoked whole-cell currents in acutely dissociated DRG neurons. Of 20 cells that were recorded, six of them were large DRG neurons (diameter, >50 μm) and had no response to 10 μm αβm-ATP (data not shown). Thirteen cells were small-to-medium DRG neurons and had responses to the application of 10 μm αβm-ATP. Of the 13 αβm-ATP-sensitive DRG neurons, six (cell size 35–45 μm) showed αβm-ATP-evoked P2X currents that had nondesensitizing current components (Fig. 5 B, left). The amplitude of nondesensitizing components was 560 ± 120 pA (n = 6). Interestingly, all of those six cells did not have a measurable response to 1 μm capsaicin (Fig. 5 B,right). For the other seven αβm-ATP-sensitive DRG neurons (cell size 20–30 μm), the αβm-ATP-evoked currents (830 ± 230 pA; n = 7; peak currents) showed rapid and complete desensitization during a 2 sec αβm-ATP application (Fig. 5 C, left). In this type of αβm-ATP-sensitive neurons, 1 μm capsaicin also evoked inward currents (Fig. 5 C, right).
αβm-ATP-sensitive synapses and A-afferent fiber central terminals
We recorded eEPSCs to determine the properties of primary afferent terminals that synapsed with lamina V neurons, using spinal cord slices with attached dorsal roots (Fig.6 A). In the same neurons we also examined the effects of αβm-ATP on the sEPSC frequency. In 12 lamina V cells that showed increases in sEPSC frequency by 100 μm αβm-ATP (306 ± 39% of control) (Fig. 6 B), stimulation of the dorsal root always evoked EPSCs with monosynaptic A-afferent properties (Fig.6 C). Conduction velocities of afferent fibers for those recordings were 2.3 ± 0.2 m/sec (n = 12), within the range of Aδ-afferent fiber conduction velocities for the same age group of rats (Nakatsuka et al., 2000). Latency of eEPSCs after repeated stimulation at different stimulation frequencies was constant in each recording (Fig. 6 C). One cell that showed monosynaptic A-afferent eEPSCs had no response to 100 μm αβm-ATP. Of those 12 cells as shown in Figure 6, eight of them were tested with the bath application of capsaicin (2 μm). Little change in mEPSC frequency was observed in those cells (100 ± 2% of control). The association of αβm-ATP-sensitive terminals and Aδ-afferent terminals was also evident in the experiments below (see Fig. 8).
P2X receptor-mediated enhancement of glutamate release in responding to the stimulation of Aδ-primary afferent fibers
ATP could be released in DH regions (Salter et al., 1993; Bardoni et al., 1997; Li et al., 1998). If released, will endogenous ATP increase glutamate release from αβm-ATP-sensitive terminals onto lamina V neurons? Repetitive stimulation (20 times at 20 Hz, 1 sec duration) was applied to a dorsal root at Aδ-fiber stimulation intensity (∼120 μA, 0.1 msec). At the end of this repetitive stimulation there was a brief increase in sEPSC frequency that lasted for ∼5 sec. The sEPSC frequency after the stimulation (post-stim) was 260 ± 23% (n = 11) (Fig.7 A,B) of sEPSC frequency before the stimulation (pre-stim). When the same repetitive stimulation was conducted in the presence of 10 μm PPADS, post-stim sEPSC frequency was 171 ± 23% (n = 7) of pre-stim sEPSC frequency, significantly lower than the post-stim sEPSC frequency in normal bath solution (Fig. 7 A,B). Similar experiments also were performed in the presence of 5 μm suramin. The post-stim sEPSC frequency was 187 ± 33% (n = 4; not illustrated as a figure) of pre-stim sEPSC frequency in the presence of suramin, significantly lower than the post-stim sEPSC frequency in normal bath solution. The effects of PPADS or suramin at the concentrations that were used were unlikely to be nonspecific actions on glutamatergic transmission because pre-stim sEPSC frequency was not affected in the presence of 10 μm PPADS (n = 4) (Fig.7 A,B) or 5 μm suramin (data not shown). There was also no change in the basal frequency and amplitude of sEPSCs by 10 μm PPADS (frequency, 101 ± 3% of control; amplitude, 99 ± 1% of control;n = 8) (Fig. 7 C). We performed an analysis of the amplitude of individual electrically evoked EPSCs during the train of dorsal root stimulation under control conditions and in the presence of PPADS. Similar to the finding by Li et al. (1998), we found that there were changes in the paired pulse ratio in the presence of 10 μm PPADS. The paired pulse ratio was 78 ± 16% under control conditions and 192 ± 38% in the presence of PPADS (n = 4; not illustrated as a figure). These results suggested that the effects of PPADS were on the presynaptic sites (Li et al., 1998).
Because some sEPSCs might be generated by action potentials and involved in Ca2+ entry through voltage-gated Ca2+ channels, we determined whether PPADS had effects on the size and shape of action potentials as well as Na+ and Ca2+ channel activity in cultured DRG neurons (Fig. 7 D,E). The size and shape of action potentials elicited by a current step showed identically in normal bath solution and after a 10 min perfusion of cells with bath solution containing 50 μm PPADS (n = 6) (Fig. 7 D). Furthermore, the inward currents of both the Na+ channel component and Ca2+ component (Gu and MacDermott, 1997) were also identical in normal bath solution and in the bath solution containing 50 μm PPADS (n = 6) (Fig.7 E).
Activation of other presynaptic ligand-gated cation channels has been shown to result in either potentiation or depression of the evoked EPSCs at glutamatergic synapses in the brain (MacDermott et al., 1999). Will activation of αβm-ATP-sensitive P2X receptors result in synaptic potentiation of Aδ-afferent glutamatergic transmission to lamina V neurons? Stimuli were applied at a reduced intensity (see Materials and Methods) to a dorsal root, and monosynaptic Aδ-eEPSCs and synaptic failures occurred randomly (Fig.8 Aa,bottom). Glutamate release likely was occurring at a few afferent central terminals during this stimulation protocol, as was evident by a much smaller amplitude of eEPSC (compare with eEPSC size in Fig. 6 C). After the application of 1 μm αβm-ATP for 5 min, the synaptic failure rates that followed the same stimulation protocol were decreased significantly (Fig. 8 Ab, bottom), and the mean amplitude of eEPSCs was increased significantly (Fig.8 Ab, top). When the ecto-ATPase inhibitor ARL67156 (10 μm for 5 min) was used to prevent the breakdown of endogenous ATP, we also observed effects similar to αβm-ATP (Fig. 8 B). Figure 8 Csummarizes the effects of αβm-ATP (n = 4) and ARL67156 (n = 4) on synaptic failure rates and mean amplitudes and shows significant decreases of failure rates (Fig.8 Ca) and increases in mean eEPSC amplitude (Fig.8 Cb). On the other hand, in the presence of 10 μm PPADS or 5 μmsuramin the failure rates increased significantly (Fig.8 D).
In the present study we have provided the electrophysiological evidence for the presence of presynaptic P2X receptors at the central terminals of primary afferent fibers connecting to lamina V neurons of the spinal cord dorsal horn. Our results indicate that the activation of these receptors can result in a robust increase of spontaneous glutamate release, as evidenced by a large increase in the frequency of sEPSCs and mEPSCs. Furthermore, we have provided evidence suggesting that P2X receptor activation can modulate the evoked glutamate release after the stimulation of primary afferent fibers. Our results also suggest that endogenously released ATP, associated with primary sensory fiber stimulation, may enhance spontaneous and evoked glutamate release.
P2X receptors on primary afferent central terminals versus P2X receptors on DH neurons
We have shown that αβm-ATP-sensitive terminals extensively synapse to lamina V neurons. Many of those terminals might be derived from primary afferent fibers. This is consistent with the presence of αβm-ATP-sensitive P2X receptors on DRG neurons. It is supported further by our findings that the eEPSCs after stimulation of primary afferent fibers were potentiated by αβm-ATP and ARL67156 and depressed by PPADS and suramin. The effects of those purine compounds were at the central terminals of primary afferent fibers, because the potentiation of eEPSCs was associated with the increase of presynaptic release probability (Fig. 8).
There is a possibility that some αβm-ATP-sensitive terminals may be derived from DH interneurons. Previously, it was reported that presynaptic P2X receptors were expressed on inhibitory DH neurons; their activation resulted in the release of GABA or glycine (Hugel and Schlichter, 2000; Rhee et al., 2000). Interestingly, presynaptic P2X receptors on inhibitory DH neurons in those studies were found to be αβm-ATP-insensitive. It will be very interesting to know whether P2X receptors are expressed at presynaptic terminals of excitatory DH neurons. In attempting to determine whether αβm-ATP-induced increases in mEPSC frequency might be attributable partially to its direct action on DH neurons, we examined the possible expression of functional αβm-ATP-sensitive P2X receptors in DH neurons. However, after we tested >100 DH neurons, 100 μm αβm-ATP did not evoke any detectable whole-cell current on those neurons. This result may suggest that αβm-ATP-sensitive DH neurons, if present, will be a very small population. This suggestion is also supported by a previous study with acutely dissociated DH neurons (Bardoni et al., 1997). However, we still should not rule out completely the possible presence of αβm-ATP-sensitive P2X receptors at presynaptic terminals of glutamatergic DH neurons.
In this study we also have performed some experiments by using ATP. It appears that, before exogenously applied ATP reached the recorded neurons (usually inside slices ∼70 μm from the surface of the tissue), there had been substantial ATP metabolism that primarily decreased actual ATP concentrations around the recorded cells. However, the ecto-ATPase inhibitor ARL67156 effectively might prevent ATP metabolism in the spinal cord slice. ATP may have complicated effects in the slice preparations, not only because of its metabolism but also because of its direct effects on some DH neurons (see also Salter et al., 1993; Li and Perl, 1995; Bardoni et al., 1997; Li et al., 1998).
αβm-ATP-sensitive terminals and P2X receptor subtypes
Although primary afferent neurons express six P2X subunits (Collo et al., 1996; Xiang et al., 1998), the spinal cord distribution of P2X receptor-expressing afferent terminals has not been well characterized. Immunoreactivity to P2X1, P2X2, and P2X3 subunits is present at primary afferent terminals in superficial laminas (Vulchanova et al., 1996, 1997). The αβm-ATP-sensitive synapses shown in this study could be located in deep laminas, but they also could be located in the superficial laminas because deep lamina neurons extend their dendrites to superficial laminas (Ritz and Greenspan, 1985). P2X3 subunits are expressed extensively at primary afferent terminals innervating inner lamina II, raising a possibility that our αβm-ATP-sensitive terminals are derived from P2X3 receptor-expressing terminals. However, this possibility may be inconsistent with the finding that our αβm-ATP-sensitive afferent terminals are capsaicin-insensitive and are not from C-afferent fibers. Nevertheless, ∼20% of primary afferent neurons did not coexpress P2X3subunits and capsaicin VR1 receptors (Guo et al., 1999). Furthermore, P2X3 subunits might not be expressed exclusively on unmyelinated, IB4-positive sensory neurons as previously thought (Petruska et al., 2000). Thus, we cannot rule out the possibility that P2X3-expressing afferent terminals synapse to lamina V dendrites. If P2X3-expressing afferent terminals were involved in our study, the subtypes of P2X receptors that mediated the augmented glutamate release would most likely be heteromeric P2X2+3 receptors. Heteromeric P2X2+3 receptors respond to αβm-ATP with weak desensitization and probably are expressed on some DRG neurons (Lewis et al., 1995; Burgard et al., 1999; North and Surprenant, 2000). We showed that αβm-ATP produced a sustained increase in spontaneous glutamate release with little or weak desensitization, which appears to be consistent with some properties of heteromeric P2X2+3 receptors. However, it has been shown recently that heteromeric P2X1+5 receptors and heteromeric P2X4+6 receptors in heterologous expression systems also respond to αβm-ATP with a weakly desensitizing current component (Le et al., 1998, 1999; Torres et al., 1998; Haines et al., 1999). A more recent study has provided first evidence suggesting that functional heteromeric P2X1+5 receptors are expressed in native tissues (Surprenant et al., 2000). Thus, these two heteromeric P2X receptors, if expressed on primary afferent fibers, may account for the effects of αβm-ATP shown in this study.
Future studies with subtype-selective P2X receptor antagonists such as TNP-ATP (North and Surprenant, 2000; Surprenant et al., 2000; Khakh et al., 2001) may help to reveal the subtype(s) of P2X receptors mediating the augment of glutamate release onto lamina V neurons.
Primary afferent type(s) of αβm-ATP-sensitive terminals
We have provided evidence suggesting that some αβm-ATP-sensitive terminals may be derived from Aδ-afferent fibers. The results that Aδ-eEPSCs can be potentiated after P2X receptor activation directly support this point. Few C-afferent fibers have monosynaptic connections with lamina V neurons (Willis and Coggeshall, 1991), suggesting that the αβm-ATP-sensitive terminals in our study were unlikely to be derived from C-afferent fibers. However, there is a possibility that some Aα/Aβ fibers also may express αβm-ATP-sensitive P2X receptors, because lamina V neurons receive input from large primary afferents as well (Brown, 1982). In our study the stimulation of dorsal root with intensity sufficient to activate Aα/Aβ fibers did not evoke Aα/Aβ monosynaptic responses in lamina V neurons. Similar results have been reported (Yoshimura et al., 1992). This is thought to be attributable to the impairment of large myelinated fibers in the transverse spinal slice preparations because of their descending and ascending path before terminating in lamina V (Brown, 1982). Nevertheless, the involvement of Aα/Aβ-afferent terminals in αβm-ATP-induced glutamate release is discounted by our finding that large-diameter (>50 μm) DRG neurons had little response to αβm-ATP or ATP (see also Li et al., 1999). Aδ-afferent terminals to lamina V are known to carry nociceptive and non-nociceptive sensory information (Willis and Coggeshall, 1991). We have shown that most Aδ-afferent terminals to the lamina V region were αβm-ATP-sensitive/capsaicin-insensitive. This raises a good possibility that at least some nociceptive Aδ-afferent fibers may express αβm-ATP-sensitive P2X receptors at their central terminals. One possible origin of the αβm-ATP-sensitive central terminals could be from Aδ-mechanoreceptors of keen joints because the Aδ-mechanoreceptors were found to be αβm-ATP-sensitive/capsaicin-insensitive (Dowd et al., 1998).
The potential origin of endogenous ATP
Our results suggest that endogenous ATP may be released in the dorsal horn regions during dorsal root stimulation. The simplest interpretation of our results is that ATP was released from primary afferent terminals. The endogenously released ATP then acted on presynaptic P2X receptors on the primary afferent terminals, which in turn enhanced glutamate release probability (Figs. 7, 8). The possible release of ATP from primary afferent central terminals was proposed first by Holton and Holton (1954). Using spinal cord synaptosomal preparations, White et al. (1985) demonstrated the release of ATP by high K+ and suggested that ATP may be released from both primary afferent terminals and dorsal horn interneurons. Consistently, with the use of electrophysiological approaches, ATP was suggested to be released in the spinal cord dorsal horn after focal electric stimulation (Bardoni et al., 1997; Li et al., 1998). It was suggested that the endogenously released ATP might function as a fast synaptic transmitter (Bardoni et al., 1997) or modulator (Li et al., 1998). Jo and Schlichter (1999) showed that ATP could be released from many dorsal horn interneurons. It also has been shown that glutamate can evoke the release of ATP from astrocytes (Queiroz et al., 1999). Thus, in addition to the potential release of endogenous ATP from primary afferent terminals, there is also a possibility that ATP is released from postsynaptic dorsal horn neurons and/or surrounding astrocytes as a consequence of glutamate receptor action on these cells. If the latter hypothesis is true, then ATP is a retrograde signaling molecule in the sensory synapses.
This work was supported by National Institutes of Health Grant NS38254 and Office of Naval Research Grant N00014-01-1-0188 (J.G.G.). We thank A. MacDermott, S. Siegelbaum, D. Price, and B. Cooper for providing thoughtful comments on this manuscript. We appreciate J. Ling for general assistance during this work.
Correspondence should be addressed to Jianguo G. Gu, McKnight Brain Institute of the University of Florida and Division of Neuroscience, Department of Oral Surgery, College of Dentistry, University of Florida, 1600 SW Archer Road, Gainesville, FL 32610. E-mail:.