Abstract
Impaired synaptic inhibition by GABA and glycine contributes to excitatory–inhibitory imbalance in the spinal cord associated with chronic neuropathic pain; however, the underlying mechanisms remain unclear. Here, we investigated how GABAergic and glycinergic inputs regulate synaptic N-methyl-d-aspartate receptor (NMDAR) activity in excitatory and inhibitory neurons of the spinal dorsal horn in male and female mice. Vesicular glutamate transporter 2 (VGluT2)-expressing excitatory neurons and vesicular γ-aminobutyric acid transporter (VGAT)-expressing inhibitory neurons exhibited comparable mixed GABAergic and glycinergic inhibitory postsynaptic currents. Blockade of GABAA receptors with gabazine or glycine receptors with strychnine potentiated NMDAR-mediated miniature excitatory postsynaptic current (mEPSC) frequency, the amplitude of EPSCs monosynaptically evoked from the dorsal root, and puff NMDA currents in VGluT2, but not VGAT, neurons. These effects were abolished by silencing neuronal activity with tetrodotoxin or in Cacna2d1 knock-out (KO) mice. In mice with conditional Grin1 KO in primary sensory neurons (Grin1-cKO), gabazine and strychnine did not affect mEPSC frequency but still enhanced puff NMDA currents in dorsal horn neurons. Furthermore, intrathecal gabazine- or strychnine-induced nociceptive hypersensitivity was diminished by Grin1-cKO, Cacna2d1 KO, or α2δ-1 C-terminus peptide. Additionally, blocking metabotropic glutamate receptor 5 (mGluR5) prevents gabazine- and strychnine-induced increases in NMDAR-mediated mEPSC frequency, evoked EPSCs, and puff NMDA currents in VGluT2 neurons as well as nociceptive hypersensitivity. Our findings reveal that GABAergic and glycinergic inhibition tonically suppresses both presynaptic and postsynaptic NMDAR activity at primary afferent→excitatory neuron synapses. α2δ-1 and mGluR5 are essential for disinhibition-induced nociceptive hypersensitivity and synaptic NMDAR hyperactivity in the spinal cord.
- dorsal root ganglion
- gabapentinoid
- K+-Cl− cotransporter 2 (KCC2)
- neuropathic pain
- synaptic plasticity
- vesicular inhibitory amino acid transporter (VIAAT)
Significance Statement
This study identifies for the first time the specific spinal cord neurons and synapses where inhibitory signals—GABA and glycine—normally suppress glutamate N-methyl-d-aspartate receptor (NMDAR) activity to regulate pain transmission. Loss of this inhibition leads to heightened pain sensitivity by selectively increasing presynaptic and postsynaptic NMDAR activity in excitatory neurons. Eliminating NMDARs from primary sensory neurons or blocking two proteins linked to NMDARs—α2δ-1 and metabotropic glutamate receptor 5 (mGluR5)—markedly reduces this pain hypersensitivity. These findings uncover how disrupted synaptic inhibition drives chronic pain and highlight α2δ-1 and mGluR5 as promising therapeutic targets for restoring excitation–inhibition balance in the spinal cord. This work advances our understanding of key cellular and molecular substrates underlying chronic neuropathic pain.
Introduction
The balance between excitatory and inhibitory synaptic inputs is critical for the integration of somatosensory signals in the spinal dorsal horn. Fast synaptic inhibition in this region is primarily mediated by GABAA and glycine receptors. Intrathecal administration of GABAA or glycine receptor antagonists rapidly induces pain hypersensitivity in rodents (Yaksh, 1989; Yamamoto and Yaksh, 1993). Although both receptor types are expressed in the spinal dorsal horn, only GABAA receptors—but not glycine receptors—are functionally present on dorsal root ganglion (DRG) neurons and their central terminals (White, 1990; Rudomin and Schmidt, 1999; Lorenzo et al., 2014). These DRG neurons normally maintain relatively high intracellular chloride levels due to the selective expression of Na+-K+-2Cl− cotransporter 1 (NKCC1) and the absence of K+-Cl− cotransporter 2 (KCC2; Sung et al., 2000). Consequently, GABAA receptor activation at primary afferent central terminals causes presynaptic depolarization and reduces glutamate release (Rudomin and Schmidt, 1999; Yuan et al., 2009). Nerve injury diminishes expression of GABAA receptor α1–α4 subunits in the DRG (Laumet et al., 2015). Furthermore, glycine receptor mutations result in hyperekplexia and heightened pain sensitivity (Xiong et al., 2014; Vuilleumier et al., 2018). While both GABAergic and glycinergic inputs tonically suppress nociceptive transmission in the spinal cord, the precise cellular and molecular mechanisms are not fully understood.
Increased synaptic N-methyl-d-aspartate receptor (NMDAR) activity in the spinal dorsal horn is a hallmark of chronic neuropathic pain. NMDAR antagonists effectively reduce neuropathic pain in animal models and patients (Chaplan et al., 1997; Correll et al., 2004; Zhou et al., 2012). Nerve injury augments postsynaptic NMDAR activity, promoting central sensitization and nociceptive hypersensitivity (Chen et al., 2014b; Li et al., 2016; Zhang et al., 2021). Additionally, both nerve injury and chemotherapy-induced neuropathy enhance presynaptic NMDAR activity at primary afferent terminals, facilitating glutamate release and spinal excitation (Xie et al., 2016, 2017a; Chen et al., 2018; Deng et al., 2019b). Notably, NMDAR antagonists alleviate pain hypersensitivity induced by intrathecal administration of GABAA or glycine receptor antagonists (Yaksh, 1989; Yamamoto and Yaksh, 1993; Sorkin et al., 1998). Disinhibition-associated nociceptive hypersensitivity is often linked to chloride dysregulation, resulting in unchecked excitatory input (Coull et al., 2003; Zhou et al., 2012; Li et al., 2016). Furthermore, NMDAR overactivation triggers calpain-mediated proteolysis of KCC2, disrupting chloride homeostasis and impairing spinal GABAergic and glycinergic inhibitory control in the spinal cord (Zhou et al., 2012; Li et al., 2016). However, how GABAergic and glycinergic inputs control presynaptic and/or postsynaptic NMDAR activity in the spinal cord remains elusive.
Excitatory neurons expressing vesicular glutamate transporter 2 (VGluT2) and inhibitory neurons expressing the vesicular GABA/glycine transporter (VGAT; also known as vesicular inhibitory amino acid transporter) play distinct roles in spinal sensory processing. VGluT2 neurons in the spinal dorsal horn play a crucial role in the relay of nociceptive information (Zhou et al., 2009; Wang et al., 2018), whereas VGAT-expressing inhibitory neurons tonically inhibit nociceptive transmission (Koga et al., 2017). Synaptic NMDAR hyperactivity in VGluT2, but not VGAT, neurons in the spinal cord are associated with opioid-induced hyperalgesia and chronic pain caused by chemotherapy, calcineurin inhibitors, and traumatic nerve injury (Huang et al., 2020, 2023; Chen et al., 2022). Yet, it is unclear whether GABAergic and glycinergic inhibition differentially regulates synaptic NMDAR activity in these specific neuronal populations.
To address these knowledge gaps, we investigated how tonic GABAergic and glycinergic inputs regulate synaptic NMDAR activity in genetically identified VGluT2 and VGAT neurons in the spinal dorsal horn. We found that blocking GABAA or glycine receptors induces both presynaptic and postsynaptic NMDAR hyperactivity in excitatory, but not inhibitory, neurons. Furthermore, presynaptic NMDARs in primary sensory neurons play a major role in the resulting nociceptive hypersensitivity. Additionally, we demonstrated that both α2δ-1 and metabotropic glutamate receptor 5 (mGluR5) are essential for disinhibition-induced synaptic NMDAR hyperactivity and nociceptive hypersensitivity. These findings advance our mechanistic understanding of the reciprocal relationship between inhibitory control and glutamatergic synaptic plasticity within spinal nociceptive circuits.
Materials and Methods
Animals
All experimental procedures were approved by the Institutional Animal Care and Use Committee of the University of Texas MD Anderson Cancer Center and adhered to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Adult male and female C57BL/6 mice (8–12 weeks old) were housed in groups of no more than five per cage, with ad libitum access to food and water. Intrathecal injections were administered via lumbar puncture as described in our previous studies (Jin et al., 2023; Huang et al., 2024b).
Conventional Cacna2d1 knock-out (KO; Cacna2d−/−) mice and Cacna2d1+/+ wild-type (WT) littermates were generated by breeding heterozygous Cacna2d1+/− pairs (#6900, Medical Research Council), as previously described (Fuller-Bicer et al., 2009). To generate VGluT2Cre/+:tdTomatoflox/flox and VGATCre/+:tdTomatoflox/flox mice, female tdTomato-floxed mice were crossed with male VGluT2-ires-Cre and VGAT-ires-Cre mice, respectively (Wang et al., 2018; Huang et al., 2022, 2023). The specificity of tdTomato labeling in excitatory VGluT2 and inhibitory VGAT neurons has been validated previously (Koga et al., 2017; Browne et al., 2020; Chen et al., 2022).
Grin1-inducible conditional KO (Grin1-cKO) mice were generated by crossing Grin1flox/flox mice with Avil-CreERT2/+ mice to obtain Avil-CreERT2/+:Grin1flox/flox offspring. To induce Grin1-cKO in adult mice, tamoxifen (#T5648, MilliporeSigma) was dissolved at 20 mg/ml in corn oil and injected intraperitoneally (75 mg/kg/day) for 5 consecutive days. Final experiments were conducted at least 2 weeks after the last injection to allow for Grin1 knockdown (Lau et al., 2011). The following mice were obtained from The Jackson Laboratory: VGluT2-ires-Cre knock-in (#028863), VGAT-ires-Cre knock-in (#028862), tdTomatoflox/flox (#007909), Grin1flox/flox mice (#005246), and Avil-CreERT2/+ (#032027). Genotypes were confirmed via ear biopsy.
Nociceptive assessment
Tactile sensitivity: Mechanical withdrawal thresholds were assessed using calibrated von Frey filaments. Mice were placed individually on a mesh floor within plastic enclosures. Filaments were applied perpendicularly to the plantar surface of the hindpaw for 6 s. If a brisk withdrawal or flinch response was observed, the filament of next lower stiffness was applied. If no response was observed, the filament of next higher stiffness was applied. The withdrawal threshold (in grams) was determined using the up–down method (Chaplan et al., 1997).
Mechanical nociception: Nociceptive thresholds in response to noxious pressure were measured using an Analgesy–Meter (Ugo Basile), as described previously (Chen et al., 2014a,b; Huang et al., 2023). A pointed probe applied increasing pressure to the midplantar surface of the hindpaw while the paw was gently restrained. The threshold was recorded when the animal withdrew its paw.
Thermal nociception: Thermal sensitivity was evaluated using a radiant heat source and thermal testing apparatus (IITC Life Science) as described previously (Zhang et al., 2022; Jin et al., 2023). Mice were acclimated on a 30°C glass surface before testing. A radiant heat stimulus was applied to the plantar surface of the hindpaw, and the latency to paw withdrawal or licking was recorded.
Spinal cord slice preparation
Mice were anesthetized with 3% isoflurane, and the lumbar spinal cords at L3–L5 levels were rapidly exposed and removed via laminectomy. Animals were then killed by 5% isoflurane inhalation followed by decapitation. The spinal cords were immediately transferred to ice-cold artificial cerebrospinal fluid (ACSF) saturated with 95% O2 and 5% CO2, containing the following (in mM): 234 sucrose, 25 glucose, 26 NaHCO3, 3.6 KCl, 1.2 NaH2PO4, 2.5 CaCl2, and 1.2 MgCl2. The tissues were mounted onto a vibratome stage and sectioned into 400-µm-thick transverse slices. The slices were then incubated in oxygenated Krebs’ solution at 34°C for at least 1 h before recordings. The Krebs’ solution contained (in mM) 117 NaCl, 11 glucose, 25 NaHCO3, 3.6 KCl, 1.2 NaH2PO4, 2.5 CaCl2, and 1.2 MgCl2, saturated with 95% O2 and 5% CO2. For electrophysiological recordings, spinal cord slices were placed in a recording chamber and continuously perfused with oxygenated Krebs’ solution at a flow rate of 3 ml/min, maintained at 34°C.
Electrophysiological recordings
Whole-cell voltage–clamp recordings were performed on lamina II neurons, as described previously (Zhou et al., 2007, 2008; Chen et al., 2014a,b). tdTomato-labeled neurons in the spinal lamina II were visualized with epifluorescence and identified using infrared and differential interference contrast optics on an upright microscope (Olympus). Spontaneous inhibitory postsynaptic currents (sIPSCs) were recorded at +10 mV using recording pipettes (4–7 MΩ) filled with an internal solution containing the following (in mM): 110 Cs2SO4, 5 TEA, 5 EGTA, 2 MgCl2, 0.5 CaCl2, 5 HEPES, 5 Mg-ATP, 0.5 Na2-GTP, and 10 QX314 (290–300 mOsm), pH 7.3. QX314 was included to suppress neuronal postsynaptic action potentials.
Excitatory postsynaptic currents (EPSCs) were recorded at a holding potential of −60 mV using recording pipettes (4–7 MΩ) filled with an internal solution containing the following (in mM): 135 K-gluconate, 5 KCl, 5 EGTA, 2 MgCl2, 0.5 CaCl2, 5 HEPES, 5 Mg-ATP, 0.5 Na2-GTP, and 10 QX314 (280–300 mOsm, pH 7.3). Miniature EPSCs (mEPSCs) were recorded in the presence of 0.5 μM tetrodotoxin (TTX). In some experiments, EPSCs were evoked by electrically stimulating the dorsal root (0.6 mA, 0.5 ms, 0.1 Hz) to induce glutamate release from primary afferents. Monosynaptic EPSCs were identified by constant latency and absence of conduction failure during 20 Hz stimulation (Zhou et al., 2010; Chen et al., 2022). Paired-pulse ratio (PPR) was determined using two stimuli delivered 50 ms apart, with PPR calculated as the ratio of the second to the first EPSC (Xie et al., 2016; Zhang et al., 2021).
Postsynaptic NMDAR currents were recorded by focal application of 100 μM NMDA (#M3262, MilliporeSigma) onto the recorded neuron using a puff pipette positioned 150 μm away using a positive pressure ejection system (Toohey Company). To remove voltage-dependent Mg2+ blockade, MgCl2 in the extracellular solution was replaced with CaCl2 (Chen et al., 2014a; Zhou et al., 2021). Signals were processed with a MultiClamp 700B amplifier (Molecular Devices), filtered at 1–2 kHz, and digitized at 10 kHz.
Gabazine (#HB0901), TTX (#HB1035), and 2-amino-5-phosphonopentanoate (AP5; #HB0252) were obtained from Hello Bio. Strychnine (#S8753) was from MilliporeSigma. 2-Methyl-6-(phenylethynyl)pyridine (MPEP; #219911-35-0) was purchased from Cayman Chemical. The α2δ-1 C-terminal mimetic peptide (VSGLN PSLWSIFGLQFILLWLVSGSRHYLW) and scrambled control peptide (FGLGWQPWSLSFYLVWSGLILSVLHLIRSN), both fused to a cell-penetrating Tat sequence (YGRKKRRQRRR), were synthesized by Synpeptide. All drugs were freshly prepared in ACSF and delivered at final concentrations via syringe pumps.
Study design and data analysis
Data are presented as means ± SEM. Sample sizes were determined based on previous publications in the field (Zhou et al., 2007; Huang et al., 2020, 2024a; Chen et al., 2022). Mice were assigned to the control and treatment groups in a 1:1 ratio as they became available. No animal mortality occurred during experiments, and no tests for outliers were conducted. We initially plotted the electrophysiological recordings and behavioral data from male and female mice separately. Because the effects of gabazine and strychnine on nociception and NMDAR activity in VGluT2 and VGAT neurons were substantially overlapping between sexes, we combined data from both male and female mice. Although our study was not specifically powered to detect subtle sex-specific differences, the findings appeared consistent across sexes within our datasets and are in agreement with previously published results (Chen et al., 2022; Huang et al., 2023, 2024a; Jin et al., 2023). Behavioral and electrophysiological experiments were conducted with investigators blinded to genotype and treatment. Evoked EPSCs, PPRs, and NMDA-evoked currents were analyzed using Clampfit 11.0 (Molecular Devices); amplitudes were averaged from four to six consecutive responses. The frequency and amplitude of mEPSCs and sIPSCs were analyzed using MiniAnalysis (Synaptosoft). Only one neuron was recorded per spinal cord slice, and each condition included recordings from at least three animals. Data normality was assessed using the Shapiro–Wilk test. Statistical comparisons between two groups were performed using a two-tailed Student's t test or Mann–Whitney U test. For comparisons involving three or more groups, one-way or two-way ANOVA followed by Tukey's post hoc test was used. A χ2 test was used to compare the proportion of neurons with GABA- and glycine-dominant sIPSCs between VGAT and VGluT2 neurons. All analyses were performed using the Prism software (version 10, GraphPad), with p values <0.05 considered statistically significant.
Results
VGluT2- and VGAT-expressing dorsal horn neurons receive comparable GABAergic and glycinergic inputs
VGluT2-expressing excitatory and VGAT-expressing inhibitory neurons in the spinal cord play opposing roles in nociceptive processing (Koga et al., 2017; Wang et al., 2018). We first examined the potential differences in GABAergic and glycinergic input to VGluT2 and VGAT neurons in the spinal dorsal horn. We generated VGluT2Cre/+:tdTomatoflox/flox and VGATCre/+:tdTomatoflox/flox mice to genetically label VGluT2 and VGAT neurons, respectively, in spinal cord slices. Although bicuculline is commonly used as a specific GABAA receptor antagonist, it also blocks small-conductance calcium–activated potassium channels (Khawaled et al., 1999; Johansson et al., 2001). Thus, we used gabazine and strychnine to selectively block GABAA and glycine receptors, respectively (Young and Snyder, 1973; Heaulme et al., 1987). We first performed whole-cell voltage–clamp recordings of sIPSCs in tdTomato-tagged VGluT2 and VGAT neurons in spinal lamina II. As reported previously (Baba et al., 2000; Zhou et al., 2007, 2008; Jin et al., 2011), two distinct types of sIPSCs were observed, differing in their decay times (Fig. 1A–D). One type exhibited a short duration (20.97 ± 1.77 ms) and was abolished by bath application of 1 μM strychnine, consistent with glycine receptor-mediated transmission. The other type displayed a significantly longer duration (57.28 ± 2.71 ms) and was blocked by 1 μM gabazine, indicating the involvement of GABAA receptors (U = 59; p < 0.001; Fig. 1D). VGluT2 and VGAT neurons had similar baseline frequency and amplitude of mixed sIPSCs (n = 39 VGluT2 neurons; n = 40 VGAT neurons; Fig. 1A–C).
VGluT2 and VGAT neurons in the spinal dorsal horn receive both GABAergic and glycinergic inputs. A, B, Representative recording traces show the effect of bath application of 1 μM gabazine and 1 μM strychnine on sIPSCs of VGluT2 (A) and VGAT (B) neurons in lamina II. C, Summary data show the frequency and amplitude of baseline mixed sIPSCs (n = 39 VGluT2 neurons from 4 mice; n = 40 VGAT neurons from 4 mice) in VGluT2 and VGAT neurons in lamina II. D, Difference in the decay time of GABAergic (n = 40 neurons from 4 mice) and glycinergic (n = 38 neurons from 4 mice) sIPSCs in both VGluT2 and VGAT neurons in lamina II. E, Pie charts show the proportion of GABA-dominant and glycine-dominant sIPSCs in VGluT2 neurons and VGAT neurons in lamina II. Data are shown as means ± SEM. ***p < 0.001 (Mann–Whitney U test).
As previously described (Takazawa et al., 2017), neurons showing a >75% reduction in sIPSC frequency following gabazine application were classified as GABA-dominant, whereas those exhibiting a >75% reduction after strychnine application were classified as glycine-dominant. The proportions of GABA-dominant and glycine-dominant neurons did not differ significantly between VGluT2- and VGAT-expressing neurons (Fig. 1E). These results suggest that both inhibitory and excitatory interneurons in the spinal dorsal horn receive a similar balance of GABAergic and glycinergic inhibitory inputs.
GABAergic and glycinergic inputs tonically inhibit presynaptic NMDAR activity in VGluT2, but not VGAT, dorsal horn neurons
We next determined whether GABAergic and glycinergic inputs regulate presynaptic NMDARs in tdTomato-tagged VGluT2 and VGAT neurons in spinal lamina II. mEPSCs were recorded in the spinal cord slices treated for 30 min with vehicle, 1 μM gabazine, or 1 μM strychnine. The baseline frequency (2.197 ± 0.259 vs 1.111 ± 0.203 Hz; t(19.77) = 4.252; p = 0.0004; two-tailed Student’s t test), but not amplitude, of mEPSCs was significantly higher in VGluT2 than in VGAT neurons (n = 14 neurons per group; Figs. 2, 3).
GABAergic and glycinergic inputs tonically suppress presynaptic NMDAR activity in VGluT2 dorsal horn neurons. A–C, Representative recording traces and cumulative probability plots show the effect of bath application of 50 μM AP5 on mEPSCs of lamina II VGluT2 neurons from vehicle-treated, gabazine-treated, and strychnine-treated slices. D, Summary data show the effect of AP5 on the frequency and amplitude of mEPSCs of lamina II VGluT2 neurons from vehicle-treated (n = 14 neurons from 4 mice), gabazine-treated (GBZ, n = 12 neurons from 4 mice), and strychnine-treated (Stry, n = 12 neurons from 4 mice) slices. Two-way ANOVA showed that there was a significant main effect for gabazine/strychnine treatment (p = 0.0024; F(2,35) = 7.194) and AP5 treatment (p < 0.0001; F(1.737, 60.79) = 27.8) and a significant interaction between the gabazine/strychnine and AP5 treatment (p < 0.0001; F(4,70) = 7.91). Data are shown as means ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001 (two-way ANOVA followed by Tukey's post hoc test).
Tonic GABAergic and glycinergic inputs do not control presynaptic NMDAR activity in VGAT dorsal horn neurons. A–C, Original recording traces and cumulative probability plots show the effect of bath application of 50 μM AP5 on mEPSCs of lamina II VGAT neurons from vehicle-treated, gabazine-treated, and strychnine-treated slices. D, Mean data show the effect of AP5 on the frequency and amplitude of mEPSCs of lamina II VGluT2 neurons from vehicle-treated (n = 14 neurons from 5 mice), gabazine-treated (n = 13 neurons from 5 mice), and strychnine-treated (n = 13 neurons from 5 mice) slices. Data are presented as means ± SEM.
In VGluT2 neurons, the baseline frequency of mEPSCs was much greater in slices treated with gabazine (n = 12 neurons) than in slices treated with vehicle (n = 14 neurons; 4.742 ± 0.925 vs 2.197 ± 0.259 Hz; p = 0.0497; F(2,35) = 7.194; Fig. 2A,B,D). However, the amplitude of mEPSCs was comparable in the two groups. Bath application of 50 μM AP5, a specific NMDAR antagonist, for 6 min markedly reduced the frequency of mEPSCs in gabazine-treated slices but had no such effect in vehicle-treated slices (Fig. 2A,B,D). Similarly, treatment with strychnine significantly increased the baseline frequency, but not amplitude, of mEPSCs (6.308 ± 0.969 vs 2.197 ± 0.259 Hz; n = 12 neurons; p = 0.0036; F(2,35) = 7.194; Fig. 2A,C,D). AP5 application rapidly reduced the mEPSC frequency in strychnine-treated slices (Fig. 2A,C,D).
In contrast, VGAT neurons exhibited no significant difference in the baseline frequency or amplitude of mEPSCs between gabazine-treated (n = 13 neurons) and vehicle-treated slices (n = 15 neurons; Fig. 3A,B,D). Likewise, AP5 application had no significant effect on the frequency or amplitude of mEPSCs in VGAT neurons in either group (Fig. 3A,B,D). Similarly, no significant difference was observed between strychnine-treated (n = 13 neurons) and vehicle-treated slices (n = 15 neurons; Fig. 3A,C,D). In addition, AP5 had no significant effect on the frequency or amplitude of mEPSCs in VGAT neurons in strychnine-treated slices (Fig. 3A,C,D). These data suggest that tonic GABAergic and glycinergic inhibition selectively suppresses presynaptic NMDAR activity in excitatory interneurons in the spinal dorsal horn.
Blocking GABAergic and glycinergic inputs augments presynaptic NMDAR activity at primary afferent central terminals synapsing onto VGluT2 neurons
NMDARs at primary afferent central terminals are functionally quiescent under normal conditions but become tonically activated in opioid-induced hyperalgesia and neuropathic pain states (Zhou et al., 2010; Xie et al., 2017b; Deng et al., 2019a; Huang et al., 2023). We thus determined whether GABAergic and glycinergic inputs control NMDAR activity specifically at primary afferent→VGluT2 neuron synapses. In tdTomato-tagged VGluT2 neurons, the baseline amplitude of monosynaptically evoked EPSCs from the dorsal root was significantly larger in gabazine-treated slices than in vehicle-treated slices (489.9 ± 14.74 vs 361.5 ± 21.43 pA; F(2,36) = 6.892; p = 0.0002; n = 13 neurons per group; Fig. 4A,B). Bath application of AP5 for 6 min markedly reduced the EPSC amplitude in gabazine-treated slices (489.9 ± 14.74 vs 394.5 ± 21.18 pA; F(1.82,65.52) = 31.54; p = 0.0001; Fig. 4A,B), but not in vehicle-treated slices. Likewise, treatment with strychnine significantly increased the baseline amplitude of evoked EPSCs in VGluT2 neurons (513.9 ± 26.43 vs 361.5 ± 21.43 pA; F(2,36) = 6.892; p = 0.0005; n = 13 neurons per group, Fig. 4A,B), and this effect was reversed by AP5 application (Fig. 4A,B).
GABAergic and glycinergic inputs inhibit presynaptic NMDARs at primary afferent terminals that synapse with VGluT2, but not VGAT, dorsal horn neurons. A, B, Original recording traces and summary data show the effect of bath application of 50 μM AP5 on the amplitude of EPSCs in lamina II VGluT2 neurons evoked monosynaptically from the dorsal root in vehicle-treated (n = 13 neurons from 4 mice), gabazine-treated (n = 13 neurons from 4 mice), and strychnine-treated (n = 13 neurons from 4 mice) slices. C, D, Representative recording traces and mean data show the effect of AP5 bath application on the PPR of evoked EPSCs in lamina II VGluT2 neurons in vehicle-treated (n = 13 neurons from 4 mice), gabazine-treated (n = 13 neurons from 4 mice), and strychnine-treated (n = 13 neurons from 4 mice) slices. E, F, Representative recording traces and summary data show the effect of 50 μM AP5 on the amplitude of evoked monosynaptic EPSCs in spinal dorsal horn VGAT neurons from vehicle-treated (n = 12 neurons from 5 mice), gabazine-treated (n = 12 neurons from 5 mice), and strychnine-treated slices (n = 12 neurons from 5 mice). G, H, Original recording traces and mean data show the effect of AP5 on the PPR of evoked EPSCs in spinal dorsal horn VGAT neurons from vehicle-treated (n = 12 neurons from 5 mice), gabazine-treated (n = 12 neurons from 5 mice), and strychnine-treated slices (n = 12 neurons from 5 mice). Two-way ANOVA showed that there was a significant main effect for gabazine/strychnine treatment (p = 0.0029; F(2, 36) = 6.892 in B; p = 0.0058; F(2, 36) = 5.959 in D) and AP5 treatment (p < 0.0001; F(1.82, 65.52) = 31.54 in B; p < 0.0001; F(1.532, 55.15) = 29.3 in D). Data are shown as means ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001 (two-way ANOVA followed by Tukey's post hoc test).
VGluT2 neurons in gabazine-treated slices exhibited a significantly reduced PPR of evoked EPSCs, a measure of presynaptic glutamate release probability, compared with vehicle-treated slices (0.707 ± 0.045 vs 0.912 ± 0.047; p = 0.0117; F(2,36) = 5.959; n = 13 neurons per group; Fig. 4C,D). Bath application of AP5 significantly increased the PPR in gabazine-treated slices (0.707 ± 0.045 vs 0.900 ± 0.047; p = 0.0031; F(1.532,55.15) = 29.3; Fig. 4C,D) but had no such effect in vehicle-treated slices. Similarly, VGluT2 neurons in strychnine-treated slices showed reduced PPR compared with vehicle-treated slices (0.699 ± 0.038 vs 0.912 ± 0.047; p = 0.0050; F(2,36) = 5.959; n = 13 neurons per group; Fig. 4C,D), and AP5 reversed the reduction in the PPR in strychnine-treated slices (Fig. 4C,D).
In contrast, VGAT neurons displayed no significant differences in the baseline amplitude of evoked EPSCs between gabazine-treated and vehicle-treated slices (n = 12 neurons per group; Fig. 4E,F). AP5 did not alter the amplitude of evoked EPSCs in either group (Fig. 4E,F). Similarly, treatment with strychnine did not significantly affect the EPSC amplitude in VGAT neurons compared with the vehicle control (n = 12 neurons per group; Fig. 4E,F), nor did AP5 change these responses (Fig. 4E,F).
Furthermore, the baseline PPR of evoked EPSCs in VGAT neurons did not differ significantly among gabazine-treated, strychnine-treated, and vehicle-treated slices (n = 12 neurons per group; Fig. 4G,H). AP5 application did not significantly change the PPR in any treatment condition (n = 12 neurons per group; Fig. 4G,H). Together, these results demonstrate that tonic inhibition by GABAergic and glycinergic synaptic inputs selectively suppresses presynaptic NMDAR activity at primary afferent terminals synapsing onto excitatory neurons in the spinal dorsal horn.
Blocking GABAergic and glycinergic inputs potentiates postsynaptic NMDAR activity selectively in spinal VGluT2 neurons
To determine whether tonic inhibition by GABAergic and glycinergic synaptic inputs regulates postsynaptic NMDAR activity in spinal VGluT2 and VGAT neurons, we recorded NMDAR currents elicited by puff application of 100 μM NMDA directly onto recorded neurons. In VGluT2 neurons, gabazine treatment significantly increased the amplitude of puff-evoked NMDAR currents compared with vehicle-treated controls (266.0 ± 25.83 vs 102.4 ± 13.60 pA; p < 0.0001; F(2,36) = 15.39; n = 13 neurons per group; Fig. 5A,B). Similarly, strychnine treatment also significantly increased the amplitude of puff-evoked NMDAR currents in VGluT2 neurons (220.8 ± 23.22 vs 102.4 ± 13.60 pA; p = 0.0012; F(2,36) = 15.39; n = 13 neurons per group; Fig. 5A,B).
Tonic GABAergic and glycinergic inputs suppress postsynaptic NMDAR activity in VGluT2, but not in VGAT, dorsal horn neurons. A, B, Representative recording traces and mean data show currents elicited by puff application of 100 μM NMDA to lamina II VGluT2 neurons from vehicle-treated (n = 13 neurons from 3 mice), gabazine-treated (n = 13 neurons from 3 mice), and strychnine-treated slices (n = 13 neurons from 3 mice). C, D, Representative recording traces and mean data show currents elicited by puff application of 100 μM NMDA to lamina II VGAT neurons from vehicle-treated (n = 12 neurons from 3 mice), gabazine-treated (n = 12 neurons from 3 mice), and strychnine-treated slices (n = 12 neurons from 3 mice). Data are presented as means ± SEM. **p < 0.01; ***p < 0.001 (one-way ANOVA followed by Tukey's post hoc test).
In contrast, VGAT neurons exhibited no significant differences in the amplitude of puff-evoked NMDAR currents among vehicle-, gabazine-, and strychnine-treated groups (n = 12 neurons per group; Fig. 5C,D). These findings indicate that GABAergic and glycinergic inputs preferentially inhibit postsynaptic NMDAR activity in spinal excitatory neurons.
Gabazine- and strychnine-induced synaptic NMDAR hyperactivity in spinal VGluT2 neurons depends on cell excitability
Disinhibition leads to neuronal hyperexcitability, particularly in excitatory neurons of the spinal dorsal horn, contributing to nociceptive hypersensitivity (Sorkin et al., 1998; Coull et al., 2003; Chen et al., 2005; Zhou et al., 2012; Lee et al., 2019). To determine whether the gabazine- and strychnine-induced enhancement of presynaptic NMDAR activity in VGluT2 neurons depends on neuronal excitability, spinal cord slices were treated with 0.5 μM TTX for 15 min followed by incubation with either 1 μM gabazine or 1 μM strychnine in the presence of TTX for 30 min. In slices cotreated with TTX and gabazine (n = 14 neurons) or TTX and strychnine (n = 15 neurons), the baseline frequency and amplitude of mEPSCs of VGluT2 neurons were comparable with those in slices treated with vehicle (Fig. 2 vs Fig. 6A–C). Furthermore, bath application of 50 μM AP5 for 6 min had no significant effect on the frequency or amplitude of mEPSCs in VGluT2 neurons from TTX-cotreated slices (Fig. 6A–C).
Gabazine and strychnine potentiate presynaptic and postsynaptic NMDAR activity via augmented neuronal firing activity in dorsal horn VGluT2 neurons. A, B, Representative recording traces and cumulative probability plots show the effect of bath application of 50 μM AP5 on mEPSCs of lamina II VGluT2 neurons from slices treated with TTX + gabazine (A) or TTX + strychnine (B). C, Summary data show the effect of AP5 on the frequency and amplitude of mEPSCs of lamina II VGluT2 neurons from slices treated with TTX + gabazine (n = 14 neurons from 3 mice) or TTX + strychnine (n = 15 neurons from 3 mice). D, Original recording traces and quantification show currents elicited by puff application of 100 μM NMDA in lamina II VGluT2 neurons from slices treated with TTX + vehicle (n = 13 neurons from 4 mice), TTX + gabazine (n = 13 neurons from 4 mice), or TTX + strychnine (n = 13 neurons from 4 mice). Data are shown as means ± SEM.
Using the same treatment protocol, we examined whether neuronal silencing with TTX blocks the effects of gabazine and strychnine on postsynaptic NMDAR activity in spinal VGluT2 neurons. The amplitude of puff-evoked NMDAR currents in VGluT2 neurons did not differ significantly among TTX + vehicle, TTX + gabazine, and TTX + strychnine groups (n = 13 neurons per group; Fig. 6D,E). These results suggest that neuronal hyperactivity is required for the disinhibition-induced potentiation of both presynaptic and postsynaptic NMDAR activity in spinal excitatory neurons.
Gabazine or strychnine induces postsynaptic NMDAR hyperactivity in the spinal cord independently of NMDARs expressed in primary sensory neurons
Because removing GABAergic and glycinergic inputs similarly augments presynaptic and postsynaptic NMDAR activity in spinal dorsal horn neurons, we investigated whether presynaptic NMDARs expressed in DRG neurons contribute to disinhibition-induced potentiation of postsynaptic NMDAR activity in spinal lamina II neurons. We generated Grin1-cKO mice in which the essential NMDAR subunit GluN1 was conditionally ablated in primary sensory neurons (Chen et al., 2022; Huang et al., 2023). Spinal cord slices from adult Grin1-cKO mice were treated with vehicle, 1 μM gabazine, or 1 μM strychnine for 30 min immediately before electrophysiological recordings. The baseline amplitude of dorsal root–evoked monosynaptic EPSCs in lamina II neurons did not differ significantly among vehicle-, gabazine-, and strychnine-treated groups (n = 12 neurons per group; Fig. 7A,B). Furthermore, bath application of 50 μM AP5 had no significant effect on the amplitude of evoked EPSCs in these neurons (Fig. 7A,B). The PPR of evoked EPSCs was also similar across treatment groups (Fig. 7C,D), and AP5 did not significantly alter the PPR in any condition (Fig. 7C,D). These findings confirm that disinhibition-induced hyperactivity of NMDARs at primary afferent central terminals depends on GluN1 expression in DRG neurons.
GABAergic and glycinergic inputs restrain postsynaptic NMDAR activity independently of presynaptic NMDARs expressed in DRG neurons. A, B, Representative current traces and quantification show the effect of bath application of 50 μM AP5 on EPSCs evoked monosynaptically from dorsal root stimulation in lamina II neurons in slices treated with vehicle (n = 12 neurons from 4 mice), gabazine (n = 12 neurons from 4 mice), or strychnine (n = 12 neurons from 4 mice) from Grin1-cKO mice. C, D, Original recording traces and quantification show the effect of AP5 on the PPR of evoked EPSCs in lamina II neurons in slices treated with vehicle (n = 12 neurons from 4 mice), gabazine (n = 12 neurons from 4 mice), or strychnine (n = 12 neurons from 4 mice) from Grin1-cKO mice. E, Representative recording traces and mean data show currents elicited by puff application of 100 μM NMDA onto lamina II neurons in slices treated with vehicle (n = 14 neurons from 3 mice), gabazine (n = 14 neurons from 3 mice), or strychnine (n = 14 neurons from 3 mice) from Grin1-cKO mice. Data are shown as means ± SEM. ***p < 0.001 (one-way ANOVA followed by Tukey's post hoc test).
Importantly in lamina II neurons from Grin1-cKO mice, treatment with gabazine or strychnine significantly increased the amplitude of puff-evoked NMDAR currents compared with vehicle (n = 14 neurons per group; Fig. 7E,F). This demonstrates that GABAergic and glycinergic inputs regulate postsynaptic NMDAR activity in spinal dorsal horn neurons independently of presynaptic NMDARs expressed in primary sensory neurons.
GABAergic and glycinergic inputs control presynaptic and postsynaptic NMDAR activity in the spinal cord through α2δ-1
Gabapentinoids reduce neuropathic pain by targeting both α2δ-1 and α2δ-2 proteins (Huang et al., 2025). α2δ-1, encoded by the Cacna2d1 gene, is a newly identified regulator of NMDARs and AMPARs in neuropathic pain conditions (Chen et al., 2018, 2019; Li et al., 2021; Zhang et al., 2021; Huang et al., 2022). In contrast, α2δ-2 selectively regulates GluK1-containing kainate receptors in inhibitory neurons (Zhou et al., 2025). To determine whether α2δ-1 is required for gabazine- and strychnine-induced potentiation of presynaptic NMDAR activity in lamina II neurons, we used spinal cord slices from Cacna2d1 KO mice. Slices were treated with vehicle, 1 μM gabazine, or 1 μM strychnine for 30 min prior to whole-cell recordings. In Cacna2d1 KO neurons, both gabazine and strychnine failed to increase significantly the baseline frequency or amplitude of mEPSCs in lamina II neurons (n = 13 neurons per group; Fig. 8A–D). Additionally, bath application of AP5 did not affect the frequency or amplitude of mEPSCs in these neurons, regardless of treatment (Fig. 8A–D).
GABAergic and glycinergic inputs tonically inhibit presynaptic and postsynaptic NMDAR activity in spinal dorsal horn neurons via α2δ-1. A–C, Original recording traces and cumulative probability plots show the effect of bath application of 50 μM AP5 on mEPSCs of lamina II neurons in slices treated with vehicle (A), gabazine (B), or strychnine (C) from Cacna2d1 KO mice. D, Mean data show the effect of AP5 on the frequency and amplitude of mEPSCs of lamina II neurons in slices treated with vehicle (n = 13 neurons from 4 mice), gabazine (n = 13 neurons from 4 mice), or strychnine (n = 13 neurons from 4 mice) from Cacna2d1 KO mice. E, Representative recording traces and mean data show currents elicited by puff application of 100 μM NMDA to lamina II neurons in slices treated with vehicle (n = 13 neurons from 3 mice), gabazine (n = 13 neurons from 3 mice), or strychnine (n = 13 neurons from 3 mice) from Cacna2d1 KO mice. Data are presented as means ± SEM.
To further determine the role of α2δ-1 in postsynaptic NMDAR activity potentiated by gabazine and strychnine in spinal dorsal horn neurons, we examined currents elicited by puff application of 100 μM NMDA to lamina II neurons from Cacna2d1 KO mice. The amplitudes of NMDAR-mediated current did not differ significantly among vehicle-, gabazine-, and strychnine-treated slices (n = 13 neurons per group; Fig. 8E,F). These findings demonstrate that α2δ-1 is essential for disinhibition-induced hyperactivity of both presynaptic and postsynaptic NMDARs in spinal dorsal horn neurons.
NMDARs expressed in DRG neurons and α2δ-1 critically contribute to disinhibition-induced nociceptive hypersensitivity
Intrathecal injection of GABAA and glycine receptor antagonists produces pain hypersensitivity, which can be alleviated by NMDAR antagonism (Yaksh, 1989; Yamamoto and Yaksh, 1993; Sorkin et al., 1998). Based on our electrophysiological findings that both NMDARs in primary sensory neurons and α2δ-1 mediate disinhibition-induced enhancement of glutamatergic input to spinal excitatory neurons, we next investigated whether they also contribute to disinhibition-induced nociceptive hypersensitivity in vivo. We first compared nociceptive hypersensitivity induced by gabazine and strychnine in WT and Grin1-cKO mice. Baseline withdrawal thresholds to von Frey filaments, noxious pressure, and radiant heat stimuli were comparable between WT and Grin1-cKO mice. Gabazine (0.2 μg) or strychnine (0.2 μg) was administered intrathecally, and withdrawal thresholds or latencies were assessed every 15 min until recovery to the baseline. In WT mice, both gabazine and strychnine caused a rapid and large reduction in withdrawal thresholds and latencies, indicating mechanical and thermal hyperalgesia (n = 10 mice per group; Fig. 9A,B). In Grin1-cKO mice, although gabazine and strychnine still reduced withdrawal thresholds and latencies, these effects were diminished compared with WT mice (n = 10 mice per group; Fig. 9A,B).
Spinal GABAergic and glycinergic inputs tonically inhibit pain hypersensitivity via α2δ-1–bound NMDARs and NMDARs expressed in DRG neurons. A, B, Time course of the paw withdrawal thresholds in response to von Frey filaments, pressure, and noxious heat stimuli in WT, Grin1-icKO, and Cacna2d1 KO mice (n = 10 mice per group) after intrathecal administration of 0.2 µg gabazine (A) or 0.2 µg strychnine (B). C, D, Time course of the paw withdrawal thresholds in response to von Frey filaments, pressure, and noxious heat stimuli in WT mice after intrathecal injection of 1 µg control peptide or α2δ-1 CT peptide (n = 8 mice per group) followed by intrathecal administration of 0.2 µg gabazine (C) or 0.2 µg strychnine (D). *p < 0.05; **p < 0.01; ***p < 0.001 compared with the baseline (0 min) within the same group; #p < 0.05; ##p < 0.01; ###p < 0.001 compared with WT or control peptide group at the same time point (two-way ANOVA followed by Tukey's post hoc test). Data are presented as means ± SEM.
Next, we determined nociceptive hypersensitivity induced by gabazine and strychnine in Cacna2d1 KO mice. Remarkably, intrathecal injection of gabazine or strychnine had no significant effect on withdrawal thresholds and latencies in Cacna2d1 KO mice (n = 10 mice per group; Fig. 9A,B). These results indicate that presynaptic NMDARs on primary sensory neurons and α2δ-1 are critical mediators of nociceptive hypersensitivity resulting from the loss of GABAergic and glycinergic inhibition.
α2δ-1–bound NMDARs are essential for nociceptive hypersensitivity induced by both gabazine and strychnine
α2δ-1 interacts directly with NMDARs via its C-terminus, promoting their synaptic expression. The α2δ-1 C-terminus mimicking peptide (α2δ-1 CT peptide) effectively disrupts α2δ-1–NMDAR interactions (Chen et al., 2018; Huang et al., 2025). To determine whether α2δ-1–bound NMDARs are necessary for gabazine- and strychnine-induced nociceptive responses, WT mice received intrathecal injection of either 1 μg α2δ-1 CT peptide or 1 μg scrambled control peptide, followed 10 min later by intrathecal gabazine or strychnine. Paw withdrawal thresholds to tactile, mechanical, and thermal stimuli were measured every 15 min until baseline recovery. Mice treated with the scrambled control peptide exhibited robust and rapid decreases in tactile, mechanical, and thermal withdrawal thresholds following gabazine or strychnine administration (n = 8 mice per group; Fig. 9C,D). In contrast, pretreatment with the α2δ-1 CT peptide diminished gabazine- and strychnine-induced reduction in the withdrawal thresholds and latencies (n = 8 mice per group; Fig. 9C,D). These findings demonstrate that tonic GABAergic and glycinergic inhibition suppresses nociceptive transmission at the spinal cord level through α2δ-1–bound NMDARs.
mGluR5 contributes to gabazine- and strychnine-induced synaptic NMDAR hyperactivity in spinal VGluT2 neurons
Similar to α2δ-1–associated NMDARs, mGluR5 is expressed in both DRG and spinal dorsal horn neurons, where it enhances glutamatergic transmission in models of opioid-induced hyperalgesia and neuropathic pain (Li et al., 2010; Xie et al., 2017b; Jin et al., 2023, 2025). We used the selective mGluR5 antagonist MPEP (Pagano et al., 2000) to determine whether mGluR5 is involved in disinhibition-induced enhancement of presynaptic and postsynaptic NMDAR activity in VGluT2 neurons. Spinal cord slices were pretreated with MPEP (10 μM, 10 min) followed by treatment with either gabazine (1 μM) or strychnine (1 μM) in the presence of MPEP for 30 min. In the presence of MPEP, neither gabazine nor strychnine significantly altered the frequency or amplitude of mEPSCs in lamina II VGluT2 neurons (n = 14 neurons for MPEP + gabazine; n = 13 neurons for MPEP + strychnine; Fig. 10A). Subsequent application of AP5 also had no significant effect on the frequency or amplitude of mEPSC in these neurons from slices treated with MPEP + gabazine or MPEP + strychnine (Fig. 10A).
GABAergic and glycinergic inputs suppress presynaptic and postsynaptic NMDAR activity in spinal excitatory horn and pain hypersensitivity via mGluR5. A, Mean data show the effect of bath application of 50 μM AP5 on the frequency and amplitude of mEPSCs of lamina II VGluT2 neurons from slices treated with gabazine (n = 14 neurons from 3 mice) or strychnine (n = 13 neurons from 3 mice) in the presence of MPEP. B, Summary data show the effect of AP5 on the amplitude of evoked monosynaptic EPSCs in spinal dorsal horn VGluT2 neurons from slices treated with gabazine (n = 12 neurons from 5 mice) or strychnine (n = 12 neurons from 5 mice) in the presence of MPEP. C, Quantification show the effect of AP5 on the PPR of evoked EPSCs in lamina II VGluT2 neurons slices treated with gabazine (n = 12 neurons from 5 mice) or strychnine (n = 12 neurons from 5 mice) in the presence of MPEP. D, Summary data show the amplitude of currents elicited by puff application of 100 μM NMDA to lamina II VGluT2 neurons from slices treated with gabazine (n = 16 neurons from 3 mice) or strychnine (n = 16 neurons from 3 mice) in the presence of MPEP. E, F, Time course of the paw withdrawal thresholds in response to tactile, pressure, and noxious heat stimuli in mice following intrathecal injection of 60 µg MPEP (n = 8 mice per group) and followed by 0.2 µg gabazine (E) or 0.2 µg strychnine (F). Gabazine alone (WT + gabazine) and strychnine alone (WT + strychnine) data from Figure 9A,B are replotted for comparison. **p < 0.01; ***p < 0.001 compared with the baseline (0 min) within the same group; ###p < 0.001 compared with the gabazine alone or strychnine alone group at the same time point (two-way ANOVA followed by Tukey's post hoc test). Data are presented as means ± SEM.
To determine the contribution of mGluR5 to disinhibition-induced NMDAR hyperactivity specifically at primary afferent central terminals, we measured the amplitude of dorsal root–evoked monosynaptic EPSCs and PPR in lamina II VGluT2 neurons. In the presence of MPEP, neither gabazine nor strychnine significantly affected the amplitude of evoked EPSCs or PPR in these neurons (n = 12 neurons per group; Fig. 10B,C). Furthermore, AP5 had no significant effect on the amplitude of evoked EPSCs or PPR in neurons from MPEP-treated slices (Fig. 10B,C).
Additionally, we assessed whether mGluR5 plays a role in disinhibition-induced postsynaptic NMDAR activity in lamina II VGluT2 neurons. We recorded currents elicited by puff application of 100 μM NMDA onto VGluT2 neurons in slices pretreated with gabazine or strychnine in the presence of MPEP. The amplitude of these currents was similar across slices treated with MPEP + gabazine, MPEP + strychnine (n = 16 neurons per group; Fig. 10D), and vehicle controls (Fig. 5B). These data collectively suggest that mGluR5 is essential for disinhibition-induced presynaptic and postsynaptic NMDAR hyperactivity at primary afferent→excitatory neuron synapses.
mGluR5 is required for nociceptive hypersensitivity induced by both gabazine and strychnine
Finally, to determine whether spinal mGluR5 mediates disinhibition-induced nociceptive hypersensitivity, we intrathecally administered gabazine (0.2 μg) or strychnine (0.2 μg) 10 min after MPEP (60 μg) in WT mice. In mice pretreated with MPEP, neither gabazine nor strychnine significantly reduced paw withdrawal thresholds in response to tactile, pressure, or noxious heat stimuli (n = 8 mice per group; Fig. 10E,F). These findings suggest that mGluR5 at the spinal cord level is a critical mediator of disinhibition-induced nociceptive hypersensitivity.
Discussion
Our study identified several previously unrecognized mechanisms by which impaired synaptic inhibition amplifies nociceptive transmission and drives pain hypersensitivity. We demonstrated that blocking GABAA or glycine receptors induces hyperactivity of both presynaptic and postsynaptic NMDARs specifically in excitatory, but not inhibitory, neurons in the spinal dorsal horn. These effects are critically dependent on neuronal excitability and α2δ-1. Furthermore, NMDARs at primary afferent central terminals play a key role in mediating disinhibition-induced pain hypersensitivity. Additionally, mGluR5 is required for both presynaptic and postsynaptic NMDAR hyperactivity as well as for the resulting pain hypersensitivity induced by gabazine and strychnine. Collectively, these findings provide new insights into how the loss of GABAergic and glycinergic inhibition selectively enhances excitatory drive from primary sensory neurons to spinal glutamatergic neurons via α2δ-1–dependent NMDAR hyperactivity.
While GABA and glycine share the same vesicular transporter, they act on distinct postsynaptic receptors and are regulated by different mechanisms. For example, nitric oxide selectively enhances synaptic release of glycine, but not GABA, in the spinal dorsal horn (Jin et al., 2011). Also, distinct muscarinic receptor subtypes govern synaptic release of GABA and glycine to dorsal horn neurons (Zhang et al., 2005, 2006, 2007; Wang et al., 2006). Stimulation of nociceptive primary afferents generally suppresses GABAergic transmission via group II/III metabotropic glutamate receptors but augments glycinergic transmission through ionotropic glutamate receptors (Zhou et al., 2007, 2008). This pattern suggests a disinhibitory shift in GABAergic tone that facilitates nociceptive signaling, counteracted by a compensatory increase in glycinergic input. We found that functional GABAA and glycine receptors are present in both VGluT2 and VGAT neurons in the lamina II. Although the proportions of GABA- and glycine-dominant sIPSCs did not differ significantly between VGluT2 and VGAT neurons, VGAT neurons have a more depolarized GABA reversal potential compared with VGluT2 neurons (Huang et al., 2024a). Furthermore, spinal VGluT2 neurons receive greater glutamatergic input than VGAT neurons (Huang et al., 2023, 2024a), rendering them more reliant on inhibitory tone to maintain excitatory–inhibitory balance. Disinhibition may disproportionately affect excitatory neurons in the spinal dorsal horn, leading to hyperexcitability (Lee et al., 2019). Thus, spinal excitatory interneurons may be more reliant on GABAergic and glycinergic inhibition to offset strong excitatory input.
We discovered that GABAergic and glycinergic inputs suppress both presynaptic and postsynaptic NMDAR activity selectively at primary afferent→excitatory neuron synapses. Spinal NMDARs play a key role in neuropathic pain (Chaplan et al., 1997; Zhou et al., 2012; Chen et al., 2014b; Xie et al., 2017a). Also, blocking NMDARs reverses pain hypersensitivity induced by GABAA and glycine receptor antagonists or KCC2 inhibitors (Yamamoto and Yaksh, 1993; Sorkin et al., 1998; Huang et al., 2024a). We found that gabazine or strychnine increased the amplitude of puff NMDA–evoked currents in VGluT2, but not VGAT, neurons in the dorsal horn. Furthermore, gabazine and strychnine potentiated AP5-sensitive mEPSC frequency and increased the amplitude of EPSCs evoked by dorsal root stimulation. Interestingly, gabazine and strychnine augmented puff-evoked NMDAR activity in dorsal horn neurons even in Grin1-cKO mice, indicating that GABAergic and glycinergic inhibition controls postsynaptic NMDARs independently of presynaptic NMDARs expressed in DRG neurons. This is consistent with findings that nerve injury also enhances postsynaptic NMDAR activity in the dorsal horn independently of presynaptic NMDARs (Huang et al., 2023).
Under physiological conditions, KCC2 activity tonically inhibits synaptic NMDAR activity in spinal VGluT2 neurons (Li et al., 2016; Huang et al., 2024a). Although KCC2 inhibition increases both presynaptic and postsynaptic NMDAR activity in VGluT2 neurons, it does not affect NMDAR activity at primary afferent terminals (Huang et al., 2024a). We showed that GABAA and glycine receptor activity is required to suppress NMDAR activity at primary afferent→excitatory neuron synapses. Nerve injury promotes NMDAR hyperactivity, activating calpain and causing KCC2 cleavage, which disrupts chloride homeostasis and further weakens synaptic inhibition (Zhou et al., 2012). Conversely, impaired KCC2 function enhances excitatory signaling through NMDARs in excitatory neurons, contributing to impaired GABAergic and glycinergic inhibition in neuropathic pain (Li et al., 2016; Huang et al., 2024a). Our findings provide further evidence for the reciprocal relationship between GABAergic and glycinergic inhibitory tone and NMDAR-driven glutamatergic excitatory transmission in the spinal cord.
A key finding of our study is that presynaptic NMDARs in primary sensory neurons play a major role in nociceptive hypersensitivity caused by disinhibition. Spinal presynaptic NMDARs are crucial for opioid-induced hyperalgesia and chronic pain conditions (Zhou et al., 2010; Chen et al., 2019, 2022; Deng et al., 2019b; Zhang et al., 2021; Huang et al., 2023). We showed that nociceptive responses to gabazine and strychnine were profoundly attenuated in Grin1-cKO mice, indicating that presynaptic NMDARs in DRG neurons play a key role in disinhibition-induced pain hypersensitivity. Our study also highlights the importance of neuronal excitability in synaptic NMDAR activity controlled by GABAergic and glycinergic inhibition. Reduced synaptic inhibition allows weak stimuli to trigger firing of dorsal horn neurons, promoting central sensitization (Sorkin et al., 1998; Coull et al., 2003; Chen et al., 2005; Zhou et al., 2012). We found that gabazine and strychnine failed to potentiate presynaptic and postsynaptic NMDAR activity in VGluT2 neurons when action potentials were blocked by TTX, suggesting that increased neuronal firing is necessary for NMDAR hyperactivity at primary afferent→excitatory synapses. Surprisingly, gabazine and strychnine also enhanced the amplitude of AP5-sensitive EPSCs elicited by dorsal root stimulation, despite the absence of glycine receptors on DRG neurons (Lorenzo et al., 2014). Activation of GABAA receptors at primary afferent central terminals typically reduces glutamate release via presynaptic inhibition (Rudomin and Schmidt, 1999; Yuan et al., 2009). However, our data suggest that GABAA receptors expressed at central terminals of DRG neurons may not be directly involved in disinhibition-induced presynaptic NMDAR hyperactivity. Disinhibition-induced hyperexcitability of excitatory interneurons likely triggers a positive feedback loop involving increased synaptic NMDAR activity and glutamate release, sustaining central sensitization in chronic pain.
Our study reveals that α2δ-1 is required for synaptic NMDAR hyperactivity and nociceptive hypersensitivity caused by the loss of GABAergic and glycinergic inhibition in the spinal cord. α2δ-1 has been identified recently as a crucial regulatory protein of NMDARs. Under normal conditions, α2δ-1–NMDAR interactions are minimal due to calcineurin-mediated NMDAR dephosphorylation (Chen et al., 2014a; Huang et al., 2020, 2024b). In neuropathic pain conditions, however, α2δ-1 binds to phosphorylated NMDARs via its C-terminus, promoting their synaptic trafficking and activity (Chen et al., 2018, 2019; Huang et al., 2020; Zhang et al., 2021; Zhou et al., 2021). We showed that gabazine and strychnine failed to potentiate the frequency of mEPSCs or the amplitude of puff NMDA currents in Cacna2d1 KO mice, indicating the α2δ-1 is essential for disinhibition-induced presynaptic and postsynaptic NMDAR hyperactivity in dorsal horn neurons. In the spinal dorsal horn, α2δ-1 expression is detected in VGluT2, but not VGAT, neurons (Koga et al., 2023). This cell-type specificity of α2δ-1 expression likely explains why gabazine and strychnine increased NMDAR activity specifically in VGluT2 neurons and at primary afferent terminals. Furthermore, the absence of pain hypersensitivity induced by strychnine and gabazine in Cacna2d1 KO mice demonstrates that α2δ-1 is essential for NMDAR-dependent nociceptive transmission caused by disinhibition. Additionally, we found that interfering with α2δ-1–NMDAR interactions using an α2δ-1 C-terminus peptide reversed gabazine- and strychnine-induced hypersensitivity.
We demonstrated that mGluR5 activity is another important signaling involved in the control of synaptic NMDAR activity by GABAergic and glycinergic inhibition. In the spinal cord, mGluR5 can form homodimers and directly interact with NMDARs (Jin et al., 2023). mGluR5 contributes to increased NMDAR synaptic expression and activity at primary afferent central terminals in opioid-induced hyperalgesia and neuropathic pain states (Li et al., 2010; Xie et al., 2017b; Jin et al., 2023). We showed that blocking mGluR5 activity with MPEP prevented gabazine- and strychnine-induced presynaptic and postsynaptic NMDAR hyperactivity in VGluT2 neurons. Also, MPEP abrogated the nociceptive hyperactivity induced by these disinhibitory agents. These findings suggest that both mGluR5 and α2δ-1 may form a large signaling complex with NMDARs to facilitate their synaptic localization and/or stabilization to augment nociceptive transmission. mGluR5 activation by glutamate released from excitatory interneurons may trigger protein kinase C–mediated NMDAR phosphorylation and facilitate α2δ-1–NMDAR binding (Zhou et al., 2021; Huang et al., 2025). Alternatively, mGluR5 may directly interact with and stabilize synaptic NMDARs, constituting a positive feedback loop that amplifies nociceptive transmission.
In summary, our findings reveal cell type– and synapse-specific regulation of NMDAR activity by GABAergic and glycinergic inhibition in the spinal nociceptive circuits. We identify increased neuronal excitability, α2δ-1, and mGluR5 as key signaling mechanisms mediating disinhibition-induced NMDAR hyperactivity at primary afferent→excitatory neuron synapses (Fig. 11). This new knowledge enhances our understanding of how disrupted GABAergic and/or glycinergic inputs lead to glutamatergic synaptic plasticity and neuropathic pain. Restoring this inhibitory tone to spinal excitatory neurons, along with inhibition of α2δ-1 and mGluR5, may represent a promising therapeutic strategy to suppress pathological NMDAR activity and alleviate chronic pain.
Schematic illustrating how the loss of GABAergic and glycinergic inhibition leads to synaptic NMDAR hyperactivity at primary afferent→excitatory neuron synapses. Under normal conditions, VGluT2-expressing excitatory neurons in the spinal dorsal horn possess GABAA (GABAAR) and glycine (GlyR) receptors and receive inhibitory inputs from GABAergic and glycinergic interneurons. These inhibitory inputs suppress neuronal excitability and limit synaptic NMDAR activity. Inhibiting the GABAAR or GlyR increases VGluT2 neuronal excitability, leading to enhanced glutamate release that activates mGluR5 located presynaptically on primary afferent terminals and postsynaptically on VGluT2 neurons. The combined action of increased mGluR5 activity and α2δ-1 facilitates NMDAR synaptic targeting and activity at both presynaptic and postsynaptic sites. This NMDAR hyperactivity at primary afferent→VGluT2 neuron synapses selectively amplifies nociceptive input to spinal excitatory neurons. These findings highlight the underlying nociceptive circuitry and the cooperative roles of α2δ-1, mGluR5, and NMDARs in pain hypersensitivity induced by synaptic disinhibition.
Footnotes
This study was supported by grants (NS101880 and NS132398) from the National Institutes of Health and by the Pamela and Wayne Garrison Distinguished Chair Endowment.
The authors declare no competing financial interests.
- Correspondence should be addressed to Shao-Rui Chen at schen{at}mdanderson.org or Hui-Lin Pan at huilinpan{at}mdanderson.org.
