Long-term potentiation of glycinergic synapses triggered by interleukin 1β

Significance

Whereas glycine is one of the three major neurotransmitters in the central nervous system, glycinergic synapses are relatively understudied in intact tissue. Here we provide the first evidence of long-term potentiation (LTP) at mammalian glycinergic synapses. In the spinal cord dorsal horn, glycinergic synapses on inhibitory GABAergic neurons exhibit LTP, occurring rapidly after exposure to the inflammatory cytokine interleukin-1 beta. This form of LTP is mediated by postsynaptic alterations in glycine receptor function. We further show that peripheral inflammation in vivo triggers glycine receptor LTP. Blocking glycine receptor LTP may represent a useful therapeutic strategy in the treatment of inflammatory pain.

Abstract

Long-term potentiation (LTP) is a persistent increase in synaptic strength required for many behavioral adaptations, including learning and memory, visual and somatosensory system functional development, and drug addiction. Recent work has suggested a role for LTP-like phenomena in the processing of nociceptive information in the dorsal horn and in the generation of central sensitization during chronic pain states. Whereas LTP of glutamatergic and GABAergic synapses has been characterized throughout the central nervous system, to our knowledge there have been no reports of LTP at mammalian glycinergic synapses. Glycine receptors (GlyRs) are structurally related to GABA_A receptors and have a similar inhibitory role. Here we report that in the superficial dorsal horn of the spinal cord, glycinergic synapses on inhibitory GABAergic neurons exhibit LTP, occurring rapidly after exposure to the inflammatory cytokine interleukin-1 beta. This form of LTP (GlyR LTP) results from an increase in the number and/or change in biophysical properties of postsynaptic glycine receptors. Notably, formalin-induced peripheral inflammation in vivo potentiates glycinergic synapses on dorsal horn neurons, suggesting that GlyR LTP is triggered during inflammatory peripheral injury. Our results define a previously unidentified mechanism that could disinhibit neurons transmitting nociceptive information and may represent a useful therapeutic target for the treatment of pain.

Glycine mediates fast synaptic inhibition throughout the spinal cord, brainstem, and midbrain, controlling normal motor behavior and rhythm generation, somatosensory processing, auditory and retinal signaling, and coordination of reflex responses (1). Strychnine-sensitive glycine receptors (GlyRs) are pentameric ligand-gated chloride channels of the Cys-loop receptor family that together with GABA_A receptors (GABA_ARs) dynamically interact with the synaptic scaffold protein gephyrin to form inhibitory synapses (1, 2). In the dorsal horn of the spinal cord, glycinergic synapses are essential for nociceptive and tactile sensory processing both during adaptive and pathological pain states (3–7). However, compared with glutamatergic and GABAergic synapses, little is known about the regulation of their synaptic strength. Several studies have examined glycine receptor trafficking in cultured neurons and in heterologous expression systems (8, 9). Intracellular Ca²⁺ appears important in the stabilization of GlyRs at synapses in culture (10), and elevation of intracellular Ca²⁺ can also potently increase glycine receptor single channel openings in cultured cells and in heterologous systems (11). However, the modulation of glycinergic synaptic strength in native tissue remains relatively unexplored.

Following peripheral injury or inflammation, changes in tactile perception develop, including hyperalgesia (exaggerated pain upon noxious stimulation), allodynia (pain in response to innocuous stimuli), and secondary hyperalgesia (pain spreading beyond the confines of the injured region). Inhibitory interneurons of the spinal dorsal horn have been proposed to gate the flow of innocuous and nociceptive sensory information from the periphery to higher brain centers (12), and supportive evidence for this idea is growing (13–17). Loss of GABAergic/glycinergic inhibition contributes to enhanced transmission of nociceptive signals through the dorsal horn circuit during pain states, resulting in hyperalgesia and allodynia (3, 18–20). For example, polysynaptic A-fiber inputs onto neurokinin 1 receptor (NK1R)-expressing projection neurons become apparent only when GABA_AR and GlyRs are pharmacologically blocked, indicating that under conditions of disinhibition, nonnoxious mechanical stimuli can drive nociceptive-specific projection pathways and elicit allodynia (21). The majority of neurons tested in the dorsal horn receive glycinergic synapses, including lamina I projection neurons, both excitatory and inhibitory interneurons of lamina II (22, 23), and inhibitory glycinergic neurons (24). Given the diversity of afferent targets, it is likely that glycinergic synapses are differentially modulated in a cell type- and subregion-specific manner. For example, during chronic inflammation, prostaglandin E₂ selectively depresses glycinergic synaptic inputs onto nonglycinergic neurons (24). Similarly, peripheral nerve injury suppresses glycinergic inhibition of a specific excitatory interneuron class [protein kinase C (PKC)γ⁺ neurons receiving Aβ fiber inputs], allowing excitatory afferents carrying nonnociceptive tactile information to activate ascending projections of nociceptive pathways that are normally under strong inhibitory control (23).

Both hyperalgesia and allodynia occur within minutes of peripheral inflammation, but the mechanisms underlying these rapid perceptual alterations are poorly understood. The proinflammatory cytokine, IL-1β, is a potent hyperalgesic agent (25–27), contributing both to peripheral and central sensitization after tissue damage (28–31). Following tissue trauma, nerve injury, or inflammation, IL-1β levels are up-regulated in the spinal cord itself (29, 32, 33), and delivery of IL-1β intrathecally increases the activity of superficial dorsal horn neurons that transmit pain signals to the brain (34, 35). Intrathecal delivery of an IL-1 receptor antagonist blocks allodynia in rodent models of inflammatory pain (36, 37). A recent study also found that IL-1β application rapidly potentiated primary afferent (glutamatergic) synapses in dorsal horn slices, through unidentified signaling molecules released from glial cells (38). Here we report that IL-1β rapidly elicits a postsynaptic form of long-term potentiation (LTP) at glycinergic synapses on lamina II inhibitory neurons (GlyR LTP), and that the same glycinergic synapses are potentiated after peripheral inflammation.

Results

IL-1β Potentiates Glycinergic Synapses on Lamina II GABAergic Interneurons.

We recorded evoked glycinergic synaptic currents from neurons in lamina II of the dorsal horn in mouse spinal cord slices. Bath application of IL-1β rapidly potentiated glycinergic inhibitory postsynaptic currents (IPSCs) (IPSC amplitudes: 160.8 ± 11.6% of pre–IL-1β values, n = 7, P < 0.005) (Fig. 1 A and B). In contrast, little or no potentiation of either glutamatergic or GABAergic synapses was observed in similar experiments after application of IL-1β, indicating that IL-1β selectively modulates glycinergic synapses on lamina II neurons (Fig. S1). Although a proportion of the neurons in lamina II are GABAergic, glutamatergic interneurons predominate (39). To focus on GABAergic lamina II interneurons, we recorded specifically from enhanced green fluorescent protein (EGFP)-labeled GABAergic neurons expressing glutamatic acid decarboxylase 65 (GAD65) (40) (Fig. S2). We confirmed that IL-1β produced rapid and robust potentiation of glycinergic currents in this group of identified GABAergic neurons (IPSC amplitudes: 165.0 ± 11.1% of pre–IL-1β values, n = 24, P < 0.0001) (Fig. 1 C and D).

Fig. 1.

IL-1β induces rapid LTP of glycinergic synapses on GAD65-EGFP positive neurons. (A) Representative experiment illustrating that GlyR IPSCs in lamina II neurons are rapidly potentiated by addition of IL-1β (10 ng/mL) to the ACSF bathing the spinal cord slice (IL-1β application denoted by blue bar). (Insets) Averages of 12 IPSCs before and at 20 min after IL-1β application. Calibration: 60 pA, 5 ms. (B) Average responses of nine neurons. Error bars indicate mean ± SEM. (C) A single example showing that GlyR IPSCs recorded exclusively from GABAergic lamina II cells identified by expression of GAD65-EGFP robustly potentiate in response to IL-1β. Calibration for this inset and in all subsequent figures except where noted: 50 pA, 10 ms. (D) Average of 24 cells. (Inset) Differential interference contrast image and fluorescence image of a GAD65 positive neuron in an acutely prepared spinal cord slice.

IL-1β Mediates a Postsynaptic Form of LTP.

After brief IL-1β application, IPSC potentiation persisted for the duration of the recording, suggesting that it represents a novel form of LTP. To rule out the possibility that residual IL-1β in the slice persistently activated its receptors even upon wash out, we followed IL-1β application with an antagonist of the IL-1 receptor (IL-1RN). Blocking the IL-1 receptor had no effect on established potentiation (IPSC amplitudes: 194.4 ± 24.9% of pre–IL-1β values following IL-1RN application; n = 7; P < 0.005) (Fig. 2 A and B), supporting the idea that IL-1β elicits persistent LTP at glycinergic synapses. The same concentration of IL-1RN entirely prevented potentiation when applied before IL-1β, as expected if the potentiation is mediated through the canonical IL-1 receptor (IPSC amplitudes 14–20 min after IL-1β application in the continued presence of IL-1RN: 100.3 ± 6.9% of control values; n = 5; not significant, n.s.) (Fig. 2 C and D).

Fig. 2.

The IL-1 receptor mediates GlyR LTP. (A) Representative example showing that after potentiation by IL-1β, GlyR IPSCs remain potentiated even in the presence of the IL-1 receptor antagonist, IL-1RN (2 mg/mL). (B) Average of seven experiments. (C) Single experiment illustrating that bath application of IL-1RN before IL-1β addition entirely prevented potentiation. (D) Average of five experiments. Error bars indicate mean ± SEM.

Synaptic potentiation can be maintained by a long-lasting increase in neurotransmitter release, or instead by modifications in postsynaptic receptor number or properties. To test whether IL-1β increases postsynaptic glycine receptor sensitivity, we delivered exogenous glycine in the bathing medium instead of using evoked glycine release from nerve terminals. In this experiment, any potentiation must result from enhanced postsynaptic glycine receptor function or number. Pulses of exogenous glycine elicited reproducible inward currents in the recorded neuron. Within 2 min of IL-1β application, however, glycine-induced inward currents increased and remained potentiated for the remainder of the recording (125.3 ± 10.8% of control glycine induced current; n = 8, BSA control; n = 6, IL-1β; P < 0.05) (Fig. 3 A and B). To test the effects of IL-1β on single quanta, we recorded miniature IPSCs (mIPSCs) in GAD65-EGFP–labeled lamina II cells. Exposure to IL-1β increased quantal amplitudes significantly (132.6 ± 12.6% of control amplitudes, measured at 4–24 min after the start of IL-1β application; n = 8; P < 0.05) (Fig. 3 C and D). IL-1β also had no effect on the paired-pulse ratio (Fig. S3A) or on mIPSC kinetics [mean rise time 1.21 ± 0.08 ms vs. 1.20 ± 0.05 ms following IL-1β; n.s.; mean decay time constant (τ_decay) 3.69 ± 0.23 ms vs. 3.91 ± 0.24 ms following IL-1β, n.s.]. Together, our data indicate that IL-1β potentiates glycinergic synapses by increasing glycine receptor number or function specifically in postsynaptic lamina II interneurons.

Fig. 3.

IL-1β potentiates glycinergic synapses by increasing postsynaptic responses to glycine. (A) Bath application of glycine (1 mM, 30 s) induces an inward current that is reproducible; a 2-min application of IL-1β triggers rapid potentiation of exogenous glycine responses. Calibration: 100 pA, 2 min. (B) Averages of six experiments for IL-1β application; eight experiments for BSA control experiments. Experiments for A and B did not use gad2-EGFP mice and therefore may include some non-GABAergic cells. (C) Miniature glycinergic IPSCs in gad2⁺ neurons were pharmacologically isolated and recorded before and after a 10-min exposure to IL-1β. Calibration: 50 pA, 100 ms. (D) Miniature IPSC amplitudes were significantly increased after IL-1β. Error bars indicate mean ± SEM.

Signaling Events Underlying IL-1β–Mediated GlyR LTP.

Postsynaptic Ca²⁺ can increase glycine receptor-mediated currents, both by modulating open channel probability (11) and by increasing the dwell time of glycine receptors at synapses (10). We therefore tested whether intracellular Ca²⁺ is necessary for GlyR LTP by chelating intracellular Ca²⁺ in the postsynaptic GABAergic lamina II cell. Inclusion of BAPTA in the recording pipette entirely blocked GlyR LTP (94.5 ± 4.6% of pre–IL-1β values; n = 5; n.s.), confirming our hypothesis that a rise in postsynaptic intracellular Ca²⁺ in the cell undergoing GlyR LTP is necessary for IL-1β to potentiate glycinergic synapses (Fig. 4A). Furthermore, exposure to thapsigargin, which depletes intracellular Ca²⁺ stores, also prevented IL-1β–induced potentiation (IPSC amplitude: 106.9 ± 11.8% of pre–IL-1β values; n = 8; n.s.) (Fig. 4B). Taken together, these data suggest that intracellular Ca²⁺ microdomains provide a necessary rise in postsynaptic Ca²⁺ concentration required to trigger GlyR LTP.

Fig. 4.

GlyR LTP in GABAergic lamina II neurons requires intracellular Ca²⁺ and p38 MAPK and can be induced by an activator of P2X7 receptors. (A) The Ca²⁺ chelator, BAPTA, was delivered into the postsynaptic cell via the pipette solution; 20 min after initiating the recording, IL-1β application does not potentiate GlyR IPSCs. (B) Bath-applied thapsigargin (10 μM) blocks GlyR LTP. (C) The p38 MAPK antagonist, SB203580 (filled symbols; 20 μM) but not its inactive analog (SB202474, open symbols; 20 μM) blocks GlyR LTP. (D) The P2X7 receptor agonist, Bz-ATP (100 μM), potentiates glycinergic IPSCs on GAD65 positive lamina II neurons. (E) Preapplication of the IL-1β scavenger, IL-1Trap (1.32 nM) prevents Bz-ATP–induced potentiation. (F) Bz-ATP fails to potentiate glycinergic IPSCs in lamina II neurons from P2X7 null mice. These experiments did not use GAD65-EGFP mice and therefore may include some non-GABAergic cells. Error bars indicate mean ± SEM.

How does IL-1β regulate GlyR function at postsynaptic sites? The rapid induction of GlyR LTP indicates that IL-1β might modulate glycinergic synaptic strength directly by binding to postsynaptic IL-1 receptors, followed by downstream activation of the p38 MAPK pathway (41). Consistent with this model, bath application of the selective p38 MAPK inhibitor, SB203580, blocked IL-1β–induced GlyR LTP in GABAergic lamina II neurons (IPSC amplitude: 103.0 ± 10.6% of pre–IL-1β values, n = 8; n.s.), whereas its inactive analog, SB202474, had no effect (205.0 ± 40.6% of pre–IL-1β values; n = 4; P < 0.05) (Fig. 4C).

Activation of P2X7 Receptors Releases IL-1β to Potentiate Glycinergic Synapses on GABAergic Lamina II Cells.

IL-1β is secreted from microglia and astrocytes in response to peripheral inflammation (42–45), and IL-1β release in the spinal cord can be elicited by activation of purinergic P2X7 receptors (45). We hypothesized that when P2X7 receptors are activated, IL-1β released from glia within the spinal cord slice should induce GlyR LTP. As predicted, glycinergic synapses did potentiate after brief exposure to the P2X7 receptor agonist, Bz-ATP (IPSC amplitudes: 135.0 ± 13.0% of pre–Bz-ATP values at 20–26 min after Bz-ATP, n = 11, P < 0.05) (Fig. 4D). Preexposure to an IL-1β scavenging protein (IL-1Trap) entirely prevented the potentiation by Bz-ATP (IPSC amplitudes: 94.0 ± 6.3% of pre-Bz-ATP values at 20–26 min after Bz-ATP in the continuous presence of IL-1Trap, n = 5, n.s.), as expected if Bz-ATP potentiates glycinergic synapses via the release of IL-1β (Fig. 4E). Moreover, Bz-ATP application in P2X7 receptor null mice, animals known to have attenuated inflammatory pain (42), had no effect on glycinergic synapses (IPSC amplitude 92.8 ± 8.0% of pre–Bz-ATP values, n = 4, n.s.) (Fig. 4F). These results suggest that even in the thin slice preparation, activating endogenous P2X7 receptors releases sufficient amounts of IL-1β to potentiate glycinergic synapses.

In Vivo Peripheral Inflammation Is Correlated with GlyR LTP.

We next asked whether peripheral inflammation triggers potentiation of glycinergic synapses on GABAergic dorsal horn neurons. Inflammation was produced in the hind paws of mice by injecting formalin, whereas control mice received similar injections of saline, and we measured thermal and mechanical sensitivity. Thirty minutes after injection, thermal sensitivity of saline-injected animals was at basal levels, whereas formalin-injected mice exhibited thermal hyperalgesia consistent with inflammation (secondary hyperalgesia) [paw withdrawal latency 35 min postsaline injection: 107 ± 12.9% of baseline; n = 19; postformalin injection: 63.3 ± 7.7% of baseline, one-way ANOVA with Bonferroni’s test, F_{(3, 32)} = 5.235, P < 0.005; n = 19] (Fig. 5A). Paw edema was notably increased in the formalin-treated group (paw weight/body weight ratio for the saline-treated group: 0.02 ± 0.001; n = 24; formalin-treated group: 0.03 ± 0.002; n = 27, P < 0.0005) (Fig. 5B). Mechanical hypersensitivity (allodynia) was also significantly greater 35 min after formalin [two-way ANOVA, F_{(1, 60)} = 4.158, P < 0.05, Bonferroni post hoc analysis at minute 35 (n = 16, P < 0.05)] (Fig. 5C). At 90 min postinjection, animals from both groups were killed and spinal cord slices prepared. GABAergic neurons from saline-treated mice showed robust GlyR LTP upon bath application of IL-1β. In contrast, in cells from formalin-treated mice, glycinergic synaptic currents did not potentiate with IL-1β (IPSC amplitudes: saline-injected: 176.6 ± 16.7%, n = 9; formalin-injected: 104.1 ± 10.5%, n = 7; P < 0.005) (Fig. 5D). This result is consistent with the hypothesis that inflammation-induced release of IL-1β in vivo had already maximally potentiated (occluded LTP) at these glycinergic synapses. If so, we would expect an increase in mIPSC amplitude during inflammation like that seen after exogenously applied IL-1β (Fig. 3 C and D). Miniature IPSC amplitudes in GAD65 positive neurons from formalin-treated mice were markedly larger than those from saline-treated mice (mIPSC amplitudes, formalin-treated animals: 148.7 ± 12.4% of saline-treated control animals; n = 8, formalin; n = 5, saline; P < 0.05) (Fig. 5 E–G). Together, these data support our hypothesis that inflammation in vivo potentiates glycinergic synapses on lamina II GABAergic neurons.

Fig. 5.

Inflammation of the hind paw elicits thermal hyperalgesia, mechanical allodynia, and potentiates glycinergic IPSCs on lamina II GABAergic neurons. (A) Injection of formalin (red bars) but not saline (white bars) into each hind paw induces secondary hyperalgesia 35 min postinjection. (B) Formalin injection significantly increases paw edema at 90 min. (C) Formalin injection decreases paw withdrawal threshold to von Frey filaments. (D) GABAergic lamina II neurons in slices prepared from saline-treated mice (white) exhibit GlyR LTP in response to IL-1β, whereas in slices from formalin-treated mice, LTP is occluded (red). (E and F) Glycinergic miniature IPSCs recorded from gad2⁺ neurons from formalin-treated mice (red) are larger than those from saline-treated mice. Calibration: 20 pA, 100 ms. (G) Cumulative probability histogram comparing glycinergic IPSC amplitudes in saline-treated (black) and formalin-treated mice (red). Error bars indicate mean ± SEM.

Discussion

LTP at glutamatergic C-fiber synapses onto lamina I projection neurons has been extensively studied as a cellular mechanism accompanying hyperalgesia (18, 46–48), but relatively little is known about plasticity at dorsal horn inhibitory synapses (49). To our knowledge, the LTP we describe here is the first example of LTP at glycinergic synapses in the mammalian CNS. Our results from GABAergic lamina II neurons indicate a postsynaptic locus of potentiation, requiring either increased numbers of postsynaptic glycine receptors, or long-lasting modulation of glycine channel function. Intracellular Ca²⁺ in the postsynaptic cell and p38 MAPK are both required for GlyR LTP. Finally, we found that potentiation of the same glycinergic synapses accompanies peripheral inflammation, suggesting a possible role for GlyR LTP in inflammatory pain. Local release of endogenous IL-1β elicited with an ATP analog also triggered GlyR LTP, and ATP released in the dorsal horn as a result of peripheral inflammation may therefore mediate the local release of IL-1β that modifies synaptic strength in vivo.

Glycine Receptor Trafficking and Channel Properties Modulated by Ca²⁺.

The regulation of glycine receptor trafficking by Ca²⁺-dependent processes has been described using elegant single particle tracking approaches (10, 50, 51). IL-1β can rapidly increase intracellular Ca²⁺ in astrocytes and neurons (52, 53). Combining these observations, we speculate that after IL-1β activates the IL-1R, there is a resulting rise in intracellular Ca²⁺ that increases the scaffolding of glycine receptors to gephyrin at synaptic sites. GlyR LTP may require Ca²⁺-mediated glycine receptor delivery to synapses, as chelating postsynaptic Ca²⁺ was sufficient to impair LTP induction. Alternatively, IL-1β may increase channel open probability rather than number; glycinergic currents and single-channel openings are transiently increased by Ca²⁺ (11), or by a number of neuromodulators downstream of intracellular Ca²⁺, including PKC, calcium/calmodulin-dependent protein kinase II, and endocannabinoids (11, 54–62). Similar signaling processes may mediate LTP described at goldfish Mauthner cell glycinergic synapses, where underlying mechanisms have not been investigated (63).

IL-1β and GlyR LTP.

The most extensively studied signaling cascades downstream from the IL-1 receptor generally lead to transcription factor activation and altered gene expression. The time course of GlyR LTP requires instead a rapid signaling pathway likely to involve p38 MAPK, a kinase strongly implicated in pain processing in the dorsal horn, and which rapidly enhances voltage-gated sodium currents in nociceptors in response to IL-1β (31). Although our data cannot rule out the possibility that IL-1β releases another signaling molecule that itself up-regulates glycine receptors in neurons, as suggested for excitatory dorsal horn synaptic potentiation (38), the rapid time course of GlyR LTP makes this less likely. IL-1β can modulate synaptic currents in unidentified superficial dorsal horn neurons, transiently increasing (64) or decreasing IPSC amplitudes (35), and synaptic potentiation of excitatory C-fiber synapses on lamina I neurons requires the release of IL-1β (38) during peripheral inflammation. Furthermore, intrathecal activation of glial cells produces mechanical allodynia and thermal hyperalgesia within 20 min, which are prevented by blocking IL-1 receptors intrathecally (65). GlyR LTP may contribute to the rapid nociceptive effects of intrathecal IL-1β (34, 36), a scenario consistent with the block of mechanical hyperalgesia by IL-1RN (66), the IL-1 receptor antagonist that blocked GlyR LTP in our experiments.

Inhibitory Synapses in Nociception.

There has been considerable interest in the idea that the removal of GABAergic/glycinergic inhibition contributes to nociception and central sensitization resulting from inflammation, peripheral nerve damage, or experimental C-fiber excitation (2, 3, 6, 12, 21). Dorsal horn functional circuitry is incompletely mapped out, but lamina II GABAergic interneurons are ideally located to gate the flow of nociceptive information from the periphery to supraspinal areas. Previous studies have shown that these interneurons receive both high-threshold and low-threshold primary afferent input (67), and innervate NK1R-expressing projection neurons (21, 68), therefore being positioned to provide inhibitory control over dorsal horn nociceptive circuitry. We hypothesize that IL-1β released from microglia or astrocytes rapidly potentiates glycinergic synapses on GABAergic lamina II interneurons that normally inhibit the passage of nociceptive signals to the brain; after inflammation, the nociception-transmitting neurons will in turn be rapidly disinhibited via this mechanism, enhancing the perceived pain and contributing to hyperalgesia and allodynia (44). GlyR LTP may operate in tandem with LTP at excitatory primary afferent synapses, causing disinhibition and opening the “gate” to enhanced pain signaling, both by unmasking A-fiber inputs onto lamina I and III projection neurons (21), and reducing the excitatory drive to GABAergic lamina II neurons (69). The dorsal horn circuit is complex, and more work will be required to fully understand the role of GlyR LTP in nociception; in particular, we need to determine which other dorsal horn neurons can also undergo GlyR LTP.

Levels of IL-1β are up-regulated in a variety of rodent pain models over a time course of days or weeks, and it is possible that glycine receptor levels in lamina II interneurons temporally follow IL-1β expression. Glycine receptors have been suggested as an important component of pain signaling that might be amenable to drug therapy. However, blocking all glycine receptors with intrathecal administration of strychnine, for example, promotes pain hypersensitivity (70). Whereas this global block of glycine receptors in the entire network degrades somatosensory processing and elicits pain, modulation of synaptic strength in specific selected circuit elements provides a powerful tool available to control signal flow without affecting all neurons at once. Although global glycine receptor blockade is not a viable strategy for treating pain (3, 70), allosteric modulation of GlyRs (1, 6, 71) or the signaling cascade that underlies GlyR LTP may yield novel drug targets useful in treating intractable pain.

Methods

Electrophysiology.

Lumbar spinal cord slices (L4–L6) were prepared from C57Bl6 mice, GAD65-EGFP mice, and P2X7^−/− mice (P15–P32) and were perfused at ∼2 mL/min at room temperature with artificial cerebrospinal fluid (ACSF). Whole-cell voltage-clamp experiments were made from lamina II neurons using patch pipettes filled with a KCl-based internal solution. For glycinergic IPSCs, ACSF contained: 6,7-dinitroquinoxaline-2,3-dione (DNQX; 10 μM) and bicuculline (10 μM), to block AMPA and GABA_A receptors; the remaining synaptic currents were entirely blocked by 1 μM strychnine (Fig. S3B). For excitatory postsynaptic currents, ACSF contained picrotoxin (100 μM). Glycine-mediated IPSCs were stimulated at 0.1 Hz using a stainless steel stimulating electrode placed lateral to the recording site within lamina II. mIPSC were recorded at −70 mV in ACSF at 28–30 °C with 4 mM Sr²⁺ replacing Ca²⁺, and eight stimuli at 40 Hz were used to elicit asynchronously released strontium mIPSCs. Asynchronous mIPSCs were measured during the 400-ms period following the synchronous response and analyzed by an experimenter blind to animal treatment.

Behavior.

Mice were tested for latency to paw flick or licking behavior on a hot plate (53 °C) or for paw withdrawal using von Frey filaments; measurements were made before a formalin (0.25 cc, 5%) or saline plantar injection into both hind paws. Saline-treated animals were interleaved with formalin-treated littermates. N values represent the number of animals; one slice per animal was used for slice experiments.

Analysis and Statistics.

Data are presented as means ± SEM of the percent change in IPSC amplitude. Potentiation was measured at 14–20 min following application of IL-1β or other manipulation except as noted. Significance was determined using a two-tailed Student t test or one-way ANOVA with a significance level of P < 0.05. mIPSC amplitudes were statistically compared using the Student t test and Kolmogorov–Smirnoff test.