LRRC8A constitutively inhibits pain hypersensitivity in rodent models by restraining NMDA receptor activity at spinal cord synapses

Abstract

Amplification of nociceptive transmission due to aberrant N-methyl-d-aspartate receptor (NMDAR) hyperactivity in the spinal cord is a key characteristic of neuropathic pain. However, under normal conditions, both presynaptic and postsynaptic NMDARs in the spinal dorsal horn are largely inactive. The mechanisms restraining synaptic NMDAR activity remain enigmatic. Leucine-rich repeat-containing protein 8A (LRRC8A or SWELL1) is an essential component of volume-regulated anion channels typically involved in regulating cell volume. Here, we report that LRRC8A was highly expressed in dorsal root ganglion (DRG) and spinal dorsal horn neurons of rats. Nerve injury persistently reduced LRRC8A expression in the DRG. siRNA-mediated Lrrc8a knockdown in rats or conditional Lrrc8a knockout in DRG neurons in mice consistently caused a pain hypersensitivity phenotype that was readily reversed by NMDAR antagonists. Correspondingly, Lrrc8a knockdown or conditional Lrrc8a knockout in DRG neurons markedly augmented synaptic localization and activity of NMDARs in the spinal cord. LRRC8A interacted with NMDARs in both rat and human spinal cord tissues primarily through its C-terminal LRR domain, restricting the synaptic trafficking and activity of NMDARs. Furthermore, Lrrc8a^ebo/ebo mutant mice, which lack the LRR domain, exhibited NMDAR-dependent pain hypersensitivity and synaptic NMDAR hyperactivity in the spinal cord. Additionally, intrathecal Lrrc8a gene delivery eliminated nerve injury–induced pain hypersensitivity and synaptic NMDAR hyperactivity in rats. These findings reveal that LRRC8A physically interacts with NMDARs and constitutively restricts their synaptic expression in the spinal cord. Reduced LRRC8A-NMDAR interactions increase synaptic expression of “unleashed” NMDARs, contributing to NMDAR hyperactivity and neuropathic pain in rodent models.

INTRODUCTION

Chronic neuropathic pain is highly prevalent in patients who have undergone major surgery or are affected by herpes zoster, diabetic neuropathy, malignancies, or cancer-related treatments. Because neuropathic pain is often unresponsive to conventional analgesics, understanding its molecular mechanisms is essential for developing new, targeted treatments. N-methyl-d-aspartate receptor (NMDAR)–mediated synaptic plasticity in the spinal cord plays a key role in the development of neuropathic pain (1–4). Although NMDAR antagonists are effective for managing neuropathic pain (5–7), these drugs are often associated with serious adverse effects. Thus, elucidating the mechanisms regulating synaptic NMDAR activity in neuropathic pain is important for identifying new targets and alternative treatment strategies. NMDARs are expressed in dorsal root ganglion (DRG) neurons and their central terminals (8, 9). However, these presynaptic NMDARs are functionally quiescent under normal conditions (3, 4, 10, 11). Also, there is minimal activity of postsynaptic NMDARs in the normal spinal dorsal horn (2, 4, 12). As a result, intrathecal injection of NMDAR antagonists does not affect normal nociception (3, 12–14). In contrast, presynaptic and postsynaptic NMDARs become tonically activated in neuropathic pain, amplifying nociceptive input from primary sensory neurons to excitatory dorsal horn neurons (3, 4, 12). The mechanisms by which these NMDARs remain latent under normal conditions and become tonically active under neuropathic pain conditions are not fully understood.

Volume-regulated anion channels (VRACs) are composed of various leucine-rich repeat-containing protein 8 (LRRC8) family members (15–17). Besides their canonical role in maintaining cell volume, VRACs are also involved in other physiological processes, including cell proliferation, migration, and apoptosis (18). LRRC8A (also known as SWELL1) is the essential subunit of VRACs and forms homo- or heterohexameric channels with at least one of its four paralogs, LRRC8B to LRRC8E (16, 17). LRRC8A is a widely expressed protein with four transmembrane domains, an intracellular N-terminal domain, and a cytoplasmic C terminus with 15 to 17 predicted leucine-rich motifs that constitute a leucine-rich repeat (LRR) domain (19, 20). LRRC8A is implicated in a variety of disease conditions, including stroke, diabetes, and cancer (21–26). Its broad expression in tissues and cells, including neurons (23, 27), suggests that this protein may have additional yet-undiscovered roles in physiology. For instance, LRRC8A functions as a dual sensor of hypoosmolality and low pH in nodose ganglion neurons (27). However, little is known about the role of neuronal LRRC8A in the control of nociceptive transmission.

In this study, we explored the potential role of LRRC8A expressed in primary sensory neurons in the regulation of nociception. We showed that LRRC8A forms a protein complex with NMDARs mainly through its C-terminal LRR domain, thereby constitutively restraining NMDAR synaptic trafficking and activity. In addition, increasing LRRC8A expression in the spinal cord eliminated nerve injury–induced synaptic NMDAR hyperactivity and nociceptive hypersensitivity in a rat model. Our findings uncover a noncanonical role of neuronal LRRC8A in regulating pain chronicity by restricting the synaptic expression of NMDARs in the spinal cord.

RESULTS

LRRC8A is present in DRG and spinal dorsal horn neurons of rats

We first used immunofluorescence labeling to determine the distribution of LRRC8A in the DRG and dorsal spinal cord. Confocal images showed that LRRC8A immunoreactivity was present in 92.4 ± 6.6% of neurons labeled with the neuronal marker NeuN in the rat DRG (Fig. 1A). LRRC8A immunoreactivity was predominantly localized to the cytoplasmic compartments of DRG neurons. Furthermore, LRRC8A immunoreactivity was present in all DRG neurons labeled with IB4 (a marker for nonpeptidergic neurons), CGRP (calcitonin gene-related peptide; a marker for peptidergic neurons), or NF200 (a marker for myelinated neurons) (Fig. 1, A to C).

Fig. 1. Distribution of LRRC8A immunoreactivity in the rat DRG and spinal cord.

(A to C) Representative immunohistochemical images of a rat DRG. Tissue sections (thickness, 30 μm) were stained using LRRC8A (green) and NeuN (blue) antibodies, IB4 (red) (A) LRRC8A (green) and CGRP (red) antibodies (B), or LRRC8A (green) and NF200 (red) antibodies (C). (D) Representative immunohistochemical images of a rat dorsal spinal cord. Tissue sections (thickness, 30 μm) were stained using LRRC8A (green) and NeuN (blue) antibodies and IB4 (red). All images are single confocal optical sections.

In the rat spinal cord, LRRC8A immunoreactivity was distributed in the dorsal horn, particularly the superficial laminae, and was colocalized with NeuN (Fig. 1D). In addition, LRRC8A immunoreactivity was colocalized with IB4 on nerve terminals within laminae I and II of the spinal cord (Fig. 1D). In laminae I and II, as defined by IB4 labeling, LRRC8A immunoreactivity was detected in 97.8 ± 8.3% of NeuN-positive neurons. These findings suggest that LRRC8A is broadly expressed in neurons of both the DRG and the spinal dorsal horn.

Nerve injury causes a sustained reduction in LRRC8A expression in the DRG

Peripheral nerve injury induces down-regulation of antinociceptive genes and up-regulation of pronociceptive genes in the DRG (28, 29). To determine whether nerve injury affects LRRC8A expression in the DRG, we performed L5 and L6 spinal nerve ligation (SNL) in rats, which is commonly used as a rodent model of neuropathic pain. RNA sequencing data indicate that LRRC8A to LRRC8D, but not LRRC8E, are expressed in the DRG and spinal cord (28, 29). Quantitative polymerase chain reaction (PCR) analysis showed that compared with sham surgery, SNL caused a sustained reduction in Lrrc8a transcripts in the DRGs, but not the spinal cords, of rats at 5, 10, and 21 days after surgery (Fig. 2A). However, SNL had no evident effect on the mRNA amount of Lrrc8b to Lrrc8d in the DRG during this period (Fig. 2A).

Fig. 2. Nerve injury causes a sustained reduction in LRRC8A expression in the rat DRG.

(A) Quantitative PCR data of Lrrc8a to Lrrc8d expression in the rat DRG 5, 10, and 21 days after SNL or sham surgery. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control, and the mean value in the sham group was set to 1 (n = 6 rats per group). ***P < 0.001 versus sham group at the same time point (two-way ANOVA followed by Tukey’s post hoc test). (B) Representative immunoblot image (left) and quantification (right) of LRRC8A protein abundance in the DRGs of sham and SNL rats 21 days after surgery (n = 6 rats per group). GAPDH was used as an internal control. *P < 0.05 (two-tailed Student’s t test). (C) Representative current traces (left) and quantification of VRAC currents (right) in DRG neurons from sham control (n = 9 neurons) and SNL (n = 12 neurons) rats 21 days after surgery. **P < 0.01 (two-tailed Student’s t test). Lines above the traces denote the timing of hypotonic solution application. (D) Quantitative PCR data show Lrrc8a expression in the spinal cords of sham control and SNL rats 21 days after surgery (n = 6 rats per group). (E) Representative immunoblot images (left) and quantification of LRRC8A protein abundance (right) in dorsal spinal cords from sham control and SNL rats 21 days after surgery (n = 6 rats per group). Data are shown as means ± SEM.

Furthermore, immunoblotting analysis showed that LRRC8A protein expression in the DRG was reduced in SNL rats compared with sham control rats 21 days after surgery (Fig. 2B). As expected, the VRAC activity of DRG neurons, elicited by hypotonic solution, was also lower in SNL rats than in sham control rats (Fig. 2C). SNL had no effect on Lrrc8a transcripts or LRRC8A protein expression in the dorsal spinal cord (Fig. 2, D and E). These data indicate that nerve injury induces a rapid and long-lasting reduction in LRRC8A expression in the DRG.

siRNA-mediated knockdown of Lrrc8a in the spinal cord and DRG induces NMDAR-dependent pain hypersensitivity

To determine the functional importance of LRRC8A down-regulation by nerve injury, we used intrathecal injection of Lrrc8a-specific small interfering RNA (siRNA; 2 μg/day) and then performed behavioral tests in siRNA-treated rats. Immunoblotting analysis confirmed that the LRRC8A abundance in the DRG and spinal cord was largely reduced in rats treated with Lrrc8a-specific siRNA compared with rats treated with control siRNA (fig. S1, A and B). Furthermore, the VRAC activity in DRG neurons was lower in Lrrc8a-specific siRNA–treated rats than in control siRNA–treated rats (fig. S1C).

Intrathecal injection of Lrrc8a-specific siRNA, but not control siRNA, for 6 consecutive days caused a gradual and profound reduction from baseline withdrawal thresholds in response to tactile, noxious pressure, and thermal stimuli (Fig. 3A), indicating the presence of allodynia and hyperalgesia.

Fig. 3. Knockdown of Lrrc8a in the spinal cord and DRG in rats as well as cKO of Lrrc8a in DRG neurons in mice induces NMDAR-dependent pain-related hypersensitivity in rats.

(A) Time course of changes in tactile, noxious pressure, and heat withdrawal thresholds in rats treated with control siRNA or Lrrc8a-specific siRNA for 6 days (n = 8 rats per group). Arrows denote timing of siRNA injections. *P < 0.05, **P < 0.01, and ***P < 0.001 versus baseline (time 0); ##P < 0.01 and ###P < 0.001 versus control siRNA group at the same time point (two-way ANOVA followed by Tukey’s post hoc test). (B) Effects of intrathecal injection of 5 μg of AP5 on tactile, pressure, and heat withdrawal thresholds in rats 6 days after treatment with control siRNA or Lrrc8a-specific siRNA (n = 8 rats per group). *P < 0.05, **P < 0.01, and ***P < 0.001 versus vehicle-treated Lrrc8a-siRNA group at the same time point (two-way ANOVA followed by Tukey’s post hoc test). (C) Mean changes in tactile, pressure, and heat withdrawal thresholds in WT and Lrrc8a-cKO mice. ***P < 0.001 (n = 8 mice per group, two-tailed Student’s t test). (D) Mean changes in the fall latency measured using Rotarod performance test in WT and Lrrc8a-cKO mice (n = 7 mice per group). (E and F) Effects of intraperitoneal injection of memantine (10 mg/kg) (E) or intrathecal injection of 5 μg of AP5 (F) on tactile, pressure, and heat withdrawal thresholds in Lrrc8a-cKO mice. *P < 0.05, **P < 0.01, and ***P < 0.001 versus vehicle group at the same time point (n = 7 mice per group, two-way ANOVA followed by Tukey’s post hoc test). Data are shown as means ± SEM.

Because increased spinal cord NMDAR activity plays a pivotal role in neuropathic pain (4, 7, 10), we next determined whether pain hypersensitivity caused by LRRC8A down-regulation is mediated by NMDARs. Intrathecal injection of (2R)-amino-5-phosphonopentanoate (AP5; 5 μg), a specific NMDAR antagonist, readily reversed the baseline withdrawal thresholds in rats treated with Lrrc8a-specific siRNA but had no effect in rats treated with control siRNA (Fig. 3B). These findings suggest that diminished LRRC8A expression in the DRG and spinal cord induces pain hypersensitivity through NMDAR activity.

Lrrc8a conditional knockout in DRG neurons induces NMDAR-dependent pain hypersensitivity

To specifically determine the role of constitutive LRRC8A in DRG neurons in nociceptive regulation, we crossed Lrrc8a^flox/flox mice (21) and Avil-Cre–expressing mice, a primary sensory neuron–specific Cre mouse line (30), to generate Lrrc8a-conditional knockout (Lrrc8a-cKO) mice in which Lrrc8a was ablated from primary sensory neurons. The protein abundance of LRRC8A in the DRG, but not the spinal cord, was diminished in Lrrc8a-cKO mice compared with wild-type (WT) control mice (fig. S2, A and B). Also, the amplitude of VRAC currents in DRG neurons was smaller in Lrrc8a-cKO mice than in WT control mice (fig. S2C). Because Avil-Cre–induced target gene knockout occurs in 84% of DRG neurons (31), this explains the incomplete removal of LRRC8A expression in DRG neurons of Lrrc8a-cKO mice.

In addition, double immunofluorescent labeling with LRRC8A and NeuN antibodies in the DRG and spinal cord showed that compared with WT mice, LRRC8A-immunoreactive DRG neurons were mostly absent in Lrrc8a-cKO mice (fig. S2, D and E). LRRC8A immunoreactivity on nerve terminals, but not on NeuN-positive neurons, in spinal laminae I and II was also reduced in Lrrc8a-cKO mice (fig. S2, F and G).

Baseline tactile, pressure, and thermal withdrawal thresholds were much lower in Lrrc8a-cKO mice than in WT controls (Fig. 3C). However, motor function, assessed using the Rotarod test, did not differ between the groups (Fig. 3D). Furthermore, intraperitoneal injection of memantine (10 mg/kg), a clinically used NMDAR antagonist, or intrathecal injection of 5 μg of AP5 reversed the reduction in the baseline withdrawal thresholds of Lrrc8a-cKO mice (Fig. 3, E and F). Thus, the pain hypersensitivity phenotype induced by LRRC8A loss of function resembles chronic neuropathic pain caused by nerve injury or chemotherapy (3, 10, 11). These results suggest that LRRC8A proteins in primary sensory neurons constitutively restrain nociceptive hypersensitivity through regulation of NMDARs.

siRNA-mediated knockdown of Lrrc8a in the spinal cord and DRG induces tonic activation of presynaptic and postsynaptic NMDARs in the spinal cord

Spinal dorsal horn neurons are critically involved in processing sensory input from DRG neurons. A critical mechanism of neuropathic pain is increased synaptic NMDAR activity in the spinal cord (3, 10, 11, 32). Because LRRC8A down-regulation elicited NMDAR-dependent pain hypersensitivity, we determined directly whether synaptic NMDAR activity in the spinal cord is altered by intrathecal treatment with Lrrc8a-specific siRNA. We obtained spinal cord slices from rats after treatment with Lrrc8a-specific siRNA or control siRNA for 5 days. We then recorded miniature excitatory postsynaptic currents (mEPSCs), which reflect quantal glutamate release from presynaptic terminals. Treatment with Lrrc8a-specific siRNA, but not control siRNA, increased the baseline frequency, but not amplitude, of glutamatergic mEPSCs in lamina II neurons. Bath application of 50 μM AP5 normalized the increased frequency of mEPSCs in lamina II neurons of Lrrc8a-specific siRNA–treated rats (fig. S3, A to C). In contrast, AP5 had no effect on the frequency or amplitude of mEPSCs in lamina II neurons of control siRNA–treated rats. These data suggest that LRRC8A downregulation potentiates presynaptic NMDAR activity of spinal dorsal horn neurons.

We next determined specifically whether LRRC8A down-regulation increases the activity of NMDARs at primary afferent central terminals. We recorded EPSCs of lamina II neurons monosynaptically evoked from the dorsal root, which represent induced glutamate released from primary afferent terminals (4, 12). The baseline amplitude of evoked EPSCs of lamina II neurons was higher in Lrrc8a-specific siRNA–treated rats than in control siRNA–treated rats (fig. S3, D to F). Bath application of 50 μM AP5 reversed the increased amplitude of EPSCs and markedly increased the paired-pulse ratio (PPR) of evoked EPSCs of lamina II neurons from Lrrc8a-specific siRNA–treated rats. In contrast, AP5 had no effect on the baseline amplitude or PPR of evoked EPSCs in lamina II neurons from control siRNA–treated rats (fig. S3, D to F).

In addition, we determined whether LRRC8A down-regulation affects postsynaptic NMDAR activity in the spinal dorsal horn by recording NMDAR currents elicited by puff application of 100 μM NMDA directly onto lamina II neurons. The amplitude of the inward NMDAR currents of lamina II neurons was substantially larger in Lrrc8a-specific siRNA–treated rats than in control siRNA–treated rats (fig. S3, G and H). Together, these findings suggest that LRRC8A constitutively suppresses the activity of postsynaptic NMDARs and presynaptic NMDARs at the central terminals of primary sensory neurons.

Lrrc8a cKO in DRG neurons induces tonic activation of presynaptic NMDARs in the spinal cord

Subsequently, we specifically determined the potential role of LRRC8A expressed in DRG neurons in the regulation of presynaptic NMDARs at the central terminals of primary afferent nerves. The baseline frequency, but not the amplitude, of mEPSCs in lamina II neurons was higher in Lrrc8a-cKO mice than in WT mice (Fig. 4, A to C). Bath application of 50 μM AP5 reversed the increased baseline frequency of mEPSCs in lamina II neurons in Lrrc8a-cKO mice. Furthermore, the baseline amplitude of EPSCs of lamina II neurons monosynaptically evoked from the dorsal root was larger in Lrrc8a-cKO mice than in WT mice. Bath application of AP5 normalized the increased amplitude of EPSCs and increased the PPR of evoked EPSCs in lamina II neurons from Lrrc8a-cKO mice (Fig. 4, D to F). However, the amplitude of NMDAR current elicited by puff application of 100 μM NMDA to lamina II neurons did not differ between Lrrc8a-cKO mice and WT mice (Fig. 4, G and H), suggesting that postsynaptic NMDAR activity was not affected by conditional knockout of Lrrc8a in DRG neurons. These data suggest that LRRC8A in DRG neurons constitutively inhibits presynaptic NMDARs at the central terminals of DRG neurons.

Fig. 4. cKO of Lrrc8a in DRG neurons induces hyperactivity of presynaptic NMDARs at central terminals in mice.

(A and B) Representative current traces (top row) and cumulative plots (bottom row) show mEPSC frequencies and amplitudes of lamina II neurons from a WT (A) and a Lrrc8a-cKO (B) mouse under baseline condition, after bath application of 50 μM AP5 and after subsequent drug-washout. (C) Average mEPSC frequencies and amplitudes for experiments as in (A) and (B) from WT (n = 11 neurons) and Lrrc8a-cKO (n = 10 neurons) mice. *P < 0.05 and **P < 0.01 (two-way ANOVA followed by Tukey’s post hoc test). (D and E) Representative current traces show evoked monosynaptic single-pulse evoked EPSCs (top rows) and paired-pulse evoked EPSCs used for PPR measurements (bottom rows) in spinal lamina II neurons from a WT and a Lrrc8a-cKO mouse under baseline conditions, after bath application of 50 μM AP5, and after subsequent drug washout. (F) Average EPSC amplitudes and PPRs for experiments as in (D) and (E); WT (n = 11 neurons) and Lrrc8a-cKO (n = 12 neurons) mice. *P < 0.05, **P < 0.01, and ***P < 0.001 (two-way ANOVA followed by Tukey’s post hoc test). (G and H) Representative current traces (G) nd quantification of NMDAR current amplitudes (H) elicited by puff application of 100 μM NMDA to lamina II neurons from WT (n = 11 neurons) and Lrrc8a-cKO (n = 12 neurons) mice. Data are shown as means ± SEM.

LRRC8A, but not LRRC8B, interacts physically with NMDARs

NMDARs continuously cycle into and out of the synaptic site, with their movement regulated by protein-protein interactions (10, 32, 33). We next determined whether LRRC8A directly interacts with NMDAR proteins. We conducted coimmunoprecipitation (co-IP) assays using membrane protein extracts of dorsal spinal cords obtained from rats and human donors. We used a GluN1 antibody for co-IP assays, because GluN1 is an essential subunit of NMDARs. Co-IP analysis revealed that the GluN1 antibody, but not a control immunoglobulin G (IgG), precipitated LRRC8A proteins from the rat and human spinal cords (Fig. 5, A and B). Furthermore, reverse co-IP showed that an LRRC8A antibody, but not IgG, precipitated GluN1, GluN2A, and GluN2B subunits in spinal membrane fractions from rat and human spinal cords (Fig. 5, C and D).

Fig. 5. LRRC8A physically interacts with NMDARs in vivo and in vitro.

(A and B) Co-IP analysis shows the interaction between GluN1 and LRRC8A in the membrane extracts of dorsal spinal cord tissues of rats (A) and humans (B). Proteins were immunoprecipitated first with a mouse GluN1 antibody or IgG. Immunoblotting was performed using a rabbit LRRC8A antibody. Similar data were obtained from three independent experiments. (C and D) Reciprocal co-IP analysis shows the interaction between LRRC8A and NMDAR subunits in the membrane extracts of dorsal spinal cord tissues of rats (C) and humans (D). Proteins were immunoprecipitated first with a mouse LRRC8A antibody or IgG. Immunoblotting was conducted using rabbit GluN1, GluN2A, or GluN2B antibodies. Similar data were obtained from three independent experiments. (E) Co-IP analysis shows the LRRC8A interaction with NMDAR subunits in membrane extracts of HEK293 cells. Lrrc8a-KO HEK293 cells were cotransfected with Flag-LRRC8A and NMDAR subunits as indicated on the left side of the gel images. Proteins were immunoprecipitated first with a FLAG antibody using membrane fractions of HEK293 cells. Immunoblotting was performed by using the antibodies indicated on the right side of the gel images. Similar data were obtained from three independent experiments. (F to H) Co-IP analysis shows that LRRC8A interacted with NMDARs predominantly through its C-terminal LRR domain. GluN1/GluN2A subunits and various Flag-tagged LRRC8A constructs, indicated above the gel images, were coexpressed in Lrrc8a-KO HEK293 cells. Co-IP (using a Flag antibody) and immunoblotting were performed using the antibodies indicated on the right side of the gel images. Similar data were obtained from three independent experiments. (I and J) Original gel images and quantification show the effect of LRR1 peptide on LRRC8A-GluN1 interactions in HEK293 cells (I, n = 11 samples per group) and spinal cord tissues (J, n = 8 rats per group). HEK293 cells were cotransfected with GluN1, GluN2A, and Flag-tagged LRRC8A. Transfected cells or spinal cord slices were treated with 1 μM LRR1 peptide or control peptide for 30 min. The cell lysate was isolated and immunoprecipitated using a Flag or LRRC8A antibody, and immunoblotting (IB) was then performed using the antibodies indicated on the right side of the gel images. **P < 0.01 (one-way ANOVA followed by Dunnett’s post hoc test). Data are shown as means ± SEM.

To determine whether LRRC8A interacts with NMDARs independently of other neural proteins such as α2δ-1 (10), we used human embryonic kidney (HEK) 293 cells in which endogenously expressed LRRC8A was removed using a CRISPR-Cas9 approach (Lrrc8a-KO HEK293 cells) (21). We conducted co-IP assays using membrane extracts of these Lrrc8a-KO HEK293 cells expressing Flag-tagged LRRC8A together with GluN1, GluN2A, GluN2B, GluN1/GluN2A, or GluN1/GluN2B. The Flag antibody precipitated GluN1, but not GluN2A or GluN2B, in HEK293 cells expressing Flag-tagged LRRC8A with GluN1, GluN2A, or GluN2B alone (Fig. 5E). However, in HEK293 cells expressing Flag-tagged LRRC8A with GluN1/GluN2A or GluN1/GluN2B, the Flag antibody precipitated GluN1, GluN2A, and GluN2B (Fig. 5E).

In addition, we used the proximity ligation assay, a highly sensitive and specific method for detecting and visualizing protein-protein interactions in cells and tissues (34, 35), to confirm the LRRC8A-NMDAR interaction in the rat DRG and spinal cord. Punctate LRRC8A-GluN1 interaction signals were readily detected in NeuN-labeled DRG neurons, particularly within the cytoplasmic compartment (fig. S4A), and in the spinal dorsal horn (fig. S4B). Compared with sham controls, dorsal rhizotomy markedly reduced IB4 labeling in the superficial dorsal horn and diminished LRRC8A-GluN1 interaction signals in laminae I and II, as identified by IB4 labeling (fig. S4, B and C), suggesting that LRRC8A-GluN1 complexes are localized at the central terminals of primary sensory neurons. Reverse co-IP showed that the anti-GluN1 antibody precipitated LRRC8A, but not LRRC8B, in HEK293 cells expressing GluN1/GluN2A with Flag-tagged LRRC8A or Flag-tagged LRRC8B (fig. S5A). Collectively, these findings demonstrate a direct association between LRRC8A and NMDAR proteins both in vivo and in vitro. LRRC8A primarily interacts with functional NMDARs through their obligatory subunit, GluN1.

LRRC8A interacts with NMDARs mainly through its LRR domain

We then attempted to identify which regions of the LRRC8A protein are involved in its interaction with NMDARs. To this end, we coexpressed GluN1/GluN2A with Flag-tagged WT LRRC8A or various Flag-tagged LRRC8A mutants in Lrrc8a-KO HEK293 cells. We initially deleted the intracellular N-terminal domain of LRRC8A (LRRC8A-ΔNT) and expressed Flag-LRRC8A-ΔNT with GluN1/GluN2A in Lrrc8a-KO cells. N-terminal domain deletion did not affect the interaction between Flag-tagged LRRC8A and GluN1 (Fig. 5F). However, when C-terminal truncated Flag-LRRC8A (LRRC8A-ΔCT) was coexpressed with GluN1/GluN2A in Lrrc8a-KO HEK293 cells, the Flag antibody precipitated GluN1 less in cells expressing LRRC8A-ΔCT than in cells expressing WT-LRRC8A (Fig. 5F), suggesting that the C terminus of LRRC8A has a major role in LRRC8A-GluN1 interaction.

LRR-containing proteins are frequently involved in protein-protein interactions (36). The C-terminal LRR domain of LRRC8A may provide docking surfaces for protein-protein interactions for signaling to various protein kinases (37). The LRR domain of LRRC8A has 15 to 17 predicted leucine-rich motifs (19, 20). To identify which leucine-rich motifs are crucial for LRRC8A-NMDAR interaction, we separately deleted nos. 1 to 3, 4 to 6, 7 to 9, 10 to 12, and 13 to 16 within the LRR domain of LRRC8A. We then expressed these Flag-tagged mutants along with GluN1/GluN2A in Lrrc8a-KO HEK293 cells. Co-IP using membrane protein extracts showed that deletion o LRR1 to LRR3 in LRRC8A diminished the interaction between Flag-tagged LRRC8A and GluN1 (Fig. 5G and fig. S5B). To pinpoint which specific LRR is most critical for LRRC8A-NMDAR interaction, we generated a series of truncations where LRR1, LRR2, or LRR3 was deleted individually. Co-IP analysis demonstrated that removing LRR1 substantially reduced LRRC8A-GluN1 interaction (Fig. 5H and fig. S5C).

To further validate direct LRRC8A-NMDAR interactions, we designed a peptide consisting of 22 amino acids that mimicked the LRR1 domain (LELHLFMLSGIPDTVFDLVELE; LRR1 peptide) of LRRC8A. The peptide was fused with the cell-penetrating peptide Tat (YGRKKRRQRRR) to disrupt intracellular LRRC8A-NMDAR interactions. A sequence-scrambled peptide, fused with Tat, was used as the control peptide. Co-IP assays using protein extracts from Lrrc8a-KO HEK293 cells expressing Flag-tagged LRRC8A and GluN1/GluN2A, as well as from rat spinal cord tissues, showed that treatment with the LRR1 peptide (1 μM for 30 min) attenuated LRRC8A-NMDAR interaction (Fig. 5, I and J). However, a Tat-fused peptide mimicking the LRR4 domain (ALHIKFTDIKEIPLWIYSLKTLE; LRR4 peptide) of LRRC8A had no effect on LRRC8A-NMDAR interaction (fig. S6, A and B). In addition, proximity ligation assay showed that LRR1 peptide treatment reduced LRRC8A-GluN1 interaction signals in HEK293 cells (fig. S6, C and D). Together, these results strongly suggest that LRRC8A physically interacts with NMDARs mainly through its LRR1 domain.

Uncoupling LRRC8A-NMDAR interactions induces pain hypersensitivity and hyperactivity of presynaptic and postsynaptic NMDARs in the spinal cord

Next, we used the LRR1 peptide to determine whether disrupting LRRC8A-NMDAR protein complexes influences nociception. Intrathecal injection of 1 μg of Tat-fused LRR1 peptide, but not 1 μg of Tat-fused LRR4 peptide or a scrambled control peptide, induced a rapid and pronounced reduction in tactile, pressure, and thermal withdrawal thresholds in naive rats (Fig. 6A).

Fig. 6. Uncoupling LRRC8A-NMDAR interactions in the DRG and spinal cord induces pain hypersensitivity and synaptic NMDAR hyperactivity in rats.

(A) Time course of the effects of intrathecal injection of LRR1 peptide, LRR4 peptide, or control peptide on tactile, pressure, and heat withdrawal thresholds in rats (n = 8 rats per group). *P < 0.05, **P < 0.01, and ***P < 0.001 versus baseline (time 0) in the LRRC8A-LRR 1 peptide group; #P < 0.05 and ###P < 0.001 versus control peptide group at the same time point (two-way ANOVA followed by Tukey’s post hoc test). (B and C) Representative current traces (top row) and cumulative plots (bottom row) show mEPSC frequencies and amplitudes of lamina II neurons under baseline conditions, after bath application of 50 μM AP5, and after subsequent drug-washout in rat spinal cord slices treated with 1 μM control peptide (B) or 1 μM LRR1 peptide (C) for 30 min. (D) Average mEPSC frequencies and amplitudes for experiments as in (B) and (C). Control peptide (n = 15 neurons); LRR1 peptide (n = 16 neurons). ***P < 0.001 (two-way ANOVA followed by Tukey’s post hoc test). (E and F) Representative current traces show evoked monosynaptic single-pulse evoked EPSCs (top rows) and paired-pulse evoked EPSCs used for PPR measurements (bottom rows) in spinal lamina II neurons under baseline conditions, after bath application of 50 μM AP5, and after subsequent drug washout in rat spinal cord slices treated with 1 μM control peptide (E) or 1 μM LRR1 peptide (F) for 30 min. (G) Average EPSC amplitudes and PPRs for experiments as in (E) and (F). Control peptide (n = 13 neurons); LRR1 peptide (n = 14 neurons). *P < 0.05, **P < 0.01, and ***P < 0.001 (two-way ANOVA followed by Tukey’s post hoc test). (H) Representative current traces of NMDAR-E PSCs (top row) and puff NMDA currents (bottom row) from lamina II neurons in spinal cord slices treated with either control peptide or LRR1 peptide. (I) Average amplitudes of NMDAR-E PSCs (control peptide, n = 12 neurons; LRR1 peptide, n = 13 neurons) and puff NMDA currents (control peptide, n = 17 neurons; LRR1 peptide, n = 15 neurons) for experiments as in (H). **P < 0.01 and ***P < 0.001 (two-tailed Student’s t test). Data are shown as means ± SEM.

We then determined whether treatment with LRR1 peptide affects synaptic NMDAR activity in the spinal dorsal horn. The baseline frequency, but not the amplitude, of mEPSCs in lamina II neurons was greater in spinal cord slices treated with 1 μM LRR1 peptide than in slices treated with 1 μM control peptide (Fig. 6, B to D). Bath application of 50 μM AP5 reversed the increased frequency of mEPSCs in lamina II neurons treated with LRR1 peptide. Furthermore, the amplitude of EPSCs of dorsal lamina II neurons monosynaptically evoked from the dorsal root was greater in spinal cord slices treated with LRR1 peptide than in slices treated with the control peptide (Fig. 6, E to G). Bath application of 50 μM AP5 reversed the augmented amplitude of EPSCs and increased the PPR of evoked EPSCs in lamina II neurons treated with LRR1 peptide (Fig. 6, E to G).

In addition, treatment with the LRR1 peptide markedly increased the amplitude of NMDAR-EPSCs in lamina II neurons evoked from dorsal root stimulation (Fig. 6, H and I). Similarly, the amplitude of puff NMDA currents was greater in lamina II neurons treated with the LRR1 peptide than in those treated with the control peptide (Fig. 6, H and I). These findings collectively demonstrate that LRRC8A constitutively suppresses nociceptive hypersensitivity by interacting with and tonically inhibiting synaptic NMDARs in the spinal dorsal horn.

LRRC8A inhibits cell surface expression of NMDARs in vitro and synaptic incorporation of NMDARs in the spinal cords of mice

Because LRRC8A physically interacts with NMDARs, we investigated whether LRRC8A affects surface expression and activity of NMDARs reconstituted in a heterologous expression system. In HEK293 cells expressing GluN1/GluN2A or GluN1/GluN2B subunits, cotransfection with Lrrc8a-specific siRNA substantially increased the NMDAR current density (fig. S7, A and B). Also, we used biotinylation assays to label and isolate the surface proteins of HEK293 cells expressing GluN1/GluN2A or GluN1/GluN2B subunits (10, 33). Cotransfection with Lrrc8a-specific siRNA increased cell surface protein abundance of GluN1, GluN2A, and GluN2B (fig. S7, C and D).

We then investigated whether endogenous LRRC8A controls synaptic expression of NMDARs in vivo. We isolated synaptosomes from dorsal spinal cord tissues of rats intrathecally treated with Lrrc8a-specific siRNA or control siRNA (2 μg/day, for 6 days) and then performed immunoblotting. Treatment with Lrrc8a-specific siRNA increased the protein expression of GluN1, GluN2A, and GluN2B in spinal cord synaptosomes (fig. S7E). In synaptosomes isolated from dorsal spinal cord tissues of WT and Lrrc8a-cKO mice, immunoblotting assays also showed that ablating Lrrc8a in DRG neurons markedly increased the protein expression of GluN1 in spinal cord synaptosomes (fig. S7F). Collectively, these results support the notion that LRRC8A tonically suppresses NMDAR surface expression and their synaptic localization at spinal cord synapses.

The LRR domain of LRRC8A controls NMDAR-dependent nociceptive hypersensitivity as well as synaptic expression and activity of NMDARs in the spinal cord

Because the C-terminal LRR domain of LRRC8A is critically involved in LRRC8A-NMDAR interaction, we investigated whether the LRR domain of LRRC8A regulates pain hypersensitivity. To do so, we took advantage of the available Lrrc8a^ebo/ebo mouse model, which harbors a homozygous 2–base pair frameshift mutation in Lrrc8a that truncates 15 terminal LRR domains (38). Compared with WT mice, Lrrc8a^ebo/ebo mice had much lower baseline tactile, pressure, and thermal withdrawal thresholds (Fig. 7A). In Lrrc8a^ebo/ebo mice, intraperitoneal injection of memantine (10 mg/kg) or intrathecal injection of 5 μg AP5 readily reversed the reduction in tactile, pressure, and thermal withdrawal thresholds (Fig. 7, B and C).

Fig. 7. Mice lacking the C-terminal LRR domain of LRRC8A exhibit NMDAR-dependent pain hypersensitivity and augmented synaptic expression of NMDARs in the spinal cord.

(A) Average tactile, pressure, and heat withdrawal thresholds in WT (n = 8) and Lrrc8a^ebo/ebo (ebo/ebo, n = 7) mice. **P < 0.01 and ***P < 0.001 (two-tailed Student’s t test). (B and C) Time course of the effects of intraperitoneal injection of memantine (10 mg/kg) (B) or intrathecal injection of 5 μg of AP5 (C) on tactile, pressure, and heat withdrawal thresholds in Lrrc8a^ebo/ebo mice (n = 7 mice per group). *P < 0.05, **P < 0.01, and ***P < 0.001 compared with vehicle group at the same time point (two-way ANOVA followed by Tukey’s post hoc test). (D) Representative current traces (top) and average VRAC currents (bottom) in DRG neurons from WT (n = 10 neurons) and Lrrc8a^ebo/ebo (n = 11 neurons) mice. Lines above the traces denote the timing of hypotonic solution application. ***P < 0.001 (two-tailed Student’s t test). (E) Co-IP analysis shows the interaction between LRRC8A and GluN1 in membrane extracts of dorsal spinal cord tissues of WT and Lrrc8a^ebo/ebo mice (n = 6 mice per group). Proteins were immunoprecipitated first with a mouse LRRC8A antibody or IgG. Immunoblotting was performed by using a rabbit GluN1 antibody. ***P < 0.001 (two-tailed Student’s t test). (F) Representative gel images and quantification of GluN1 protein expression in dorsal spinal cord synaptosomes isolated from WT and Lrrc8a^ebo/ebo mice (n = 6 mice per group). PSD-95 was used as an internal control. **P < 0.01 (two-tailed Student’s t test). Data are shown as means ± SEM.

The VRAC current in DRG neurons was largely reduced in Lrrc8a^ebo/ebo mice compared with that in WT mice (Fig. 7D). Co-IP analysis of membrane protein extracts of dorsal spinal cords showed that the LRRC8A antibody precipitated GluN1 proteins much less in Lrrc8a^ebo/ebo mice than in WT mice (Fig. 7E). In addition, immunoblotting of spinal cord synaptosomes showed that the protein expression of GluN1 was markedly increased in Lrrc8a^ebo/ebo mice compared with WT mice (Fig. 7F).

To determine the potential role of the LRR domain of LRRC8A in the control of synaptic NMDAR activity in the spinal dorsal horn, we examined changes in presynaptic and postsynaptic NMDAR activity in spinal cord slices from WT and Lrrc8a^ebo/ebo mice. The baseline frequency, but not the amplitude, of mEPSCs in lamina II neurons was substantially greater in Lrrc8a^ebo/ebo mice than in WT mice (fig. S8, A to C). Bath application of 50 μM AP5 normalized the elevated baseline frequency of mEPSCs in lamina II neurons from Lrrc8a^ebo/ebo mice. Furthermore, the baseline amplitude of EPSCs in lamina II neurons monosynaptically evoked from the dorsal root was much larger in Lrrc8a^ebo/ebo mice than in WT mice. In lamina II neurons from Lrrc8a^ebo/ebo mice, bath application of AP5 markedly reduced the amplitude of evoked EPSCs and increased the PPR of evoked EPSCs (fig. S8, D to F). In addition, the amplitude of NMDAR currents elicited by puff application of 100 μM NMDA directly onto lamina II neurons was greater in Lrrc8a^ebo/ebo mice than in WT mice (fig. S8, G and H). Together, these findings highlight the crucial role of LRRC8A’s C-terminal LRR domain in the constitutive inhibition of nociceptive hyperactivity as well as synaptic expression and activity of NMDARs in the spinal cord.

Intrathecal Lrrc8a gene transfer eliminates nerve injury–induced pain hypersensitivity

Because our study suggests that LRRC8A functions as an antinociceptive regulator by tonically inhibiting synaptic NMDAR activity in the spinal cord, we next determined whether Lrrc8a gene delivery using a lentiviral vector, by intrathecal injection (10, 39), can inhibit nerve injury–induced pain hypersensitivity. We constructed lentiviral vectors encoding rat Lrrc8a (Lrrc8a vector) or enhanced green fluorescent protein (control vector) (fig. S9, A to C). We then performed intrathecal injection of the Lrrc8a vector or control vector 2 weeks after SNL or sham surgery in rats. The Lrrc8a-expressing vector, but not the control vector, gradually and persistently reversed allodynia and hyperalgesia induced by SNL in rats within 3 to 6 weeks after intrathecal injection (Fig. 8A). Intrathecal injection of Lrrc8a vectors also slightly elevated the tactile, pressure, and thermal withdrawal thresholds in sham control rats (fig. S10A).

Fig. 8. Intrathecal Lrrc8a gene delivery eliminates nerve injury–induced pain hypersensitivity and enhanced synaptic NMDAR expression in rats.

(A) Time course of changes in tactile, pressure, and heat withdrawal thresholds of SNL rats after intrathecal injection of Lrrc8a vector or control vector (n = 8 rats per group). *P < 0.05, **P < 0.01, and ***P < 0.001 versus control vector group at the same time point (two-way ANOVA followed by Tukey’s post hoc test). (B and C) Original gel images (B) and quantification (C) show LRRC8A protein abundance in the DRGs of sham and SNL rats after intrathecal injection of Lrrc8a vector (Lv) or control vector (Cv; n = 6 rats per group). *P < 0.05 and ***P < 0.001 (two-way ANOVA followed by Tukey’s post hoc test). (D and E) Representative gel images (D) and quantification (E) show LRRC8A protein abundance in spinal cord tissues of sham and SNL rats treated with Cv or Lv (n = 6 rats per group). **P < 0.01 (two-way ANOVA followed by Tukey’s post hoc test). (F and G) Representative gel images (F) and quantification (G) show GluN1 protein expression in spinal cord synaptosomes isolated from sham and SNL rats treated with Cv or Lv (n = 6 rats per group). ***P < 0.001 (two-way ANOVA followed by Tukey’s post hoc test). (H and I) Representative current traces (H) and quantification (I) show VRAC currents in DRG neurons from sham and SNL rats treated with Cv or Lv (n = 11 neurons per group). Lines above the traces denote the timing of hypotonic solution application. *P < 0.05 and ***P < 0.001 (two-way ANOVA followed by Tukey’s post hoc test). Data are shown as means ± SEM.

Immunoblotting showed that the LRRC8A expression in the DRG was greater in SNL rats treated with Lrrc8a vectors than in SNL rats treated with control vectors (Fig. 8, B and C). In sham control rats, intrathecal injection of Lrrc8a vectors also increased the LRRC8A protein amount in the DRG compared with injected control vectors. Furthermore, intrathecal injection of Lrrc8a vectors substantially increased LRRC8A protein abundance in the dorsal spinal cord in both sham and SNL rats (Fig. 8, D and E). To determine whether increasing LRRC8A expression affects synaptic targeting of NMDARs in the spinal cord, we isolated synaptosomes from dorsal spinal cord tissues from sham or SNL rats treated with Lrrc8a vectors or control vectors. Immunoblotting showed that SNL substantially increased the protein abundance of GluN1 in spinal cord synaptosomes. Treatment with Lrrc8a vectors normalized the increased GluN1 protein amount in spinal synaptosomes of SNL rats but had no effect on GluN1 amount in sham control rats (Fig. 8, F and G).

In addition, the amplitude of VRAC currents in DRG neurons was lower in SNL rats treated with control vectors than in sham rats treated with control vectors (Fig. 8, H and I). Intrathecal treatment with Lrrc8a vectors in SNL rats markedly increased the amplitude of VRAC currents in DRG neurons compared with that in control vector–treated SNL rats. Furthermore, in sham control rats, intrathecal injection of Lrrc8a vectors increased the amplitude of VRAC currents in DRG neurons (Fig. 8, H and I). These results indicate that intrathecal Lrrc8a gene transfer had a profound and long-lasting effect in eliminating chronic pain hypersensitivity caused by nerve injury.

Intrathecal Lrrc8a gene transfer diminishes nerve injury–induced synaptic NMDAR hyperactivity in the spinal cord

Because increased synaptic NMDAR activity in the spinal cord is critically involved in the development of neuropathic pain, we determined whether Lrrc8a gene transfer can reduce synaptic NMDAR hyperactivity caused by nerve injury. We performed electrophysiological recordings in spinal cord slices obtained from sham and SNL rats 6 weeks after intrathecal injection of the control vector or Lrrc8a vector. The baseline frequency of mEPSCs in lamina II neurons was lower in Lrrc8a vector–treated SNL rats than in control vector–treated SNL rats (fig. S10, B to F). Bath application of 50 μM AP5 reversed the increased frequency of mEPSCs in lamina II neurons of control vector–treated SNL rats. In contrast, AP5 had no effect on the frequency of mEPSCs in lamina II neurons from Lrrc8a vector–treated SNL rats.

Furthermore, the baseline amplitude of EPSCs monosynaptically evoked from the dorsal root in spinal lamina II neurons was larger in control vector–treated SNL rats than in control vector–treated sham rats. In lamina II neurons from control vector–treated SNL rats, AP5 markedly reduced the amplitude of evoked EPSCs and increased the PPR of evoked EPSCs (fig. S11, A to E). In contrast, AP5 had no effect on the amplitude of EPSCs or PPR of evoked EPSCs in lamina II neurons from Lrrc8a vector–treated SNL rats.

Last, we recorded postsynaptic NMDAR currents elicited by puff application of NMDA directly to the recorded lamina II neurons. The amplitude of the inward NMDAR currents of lamina II neurons was larger in control vector–treated SNL rats than in control vector–treated sham rats. Treatment with Lrrc8a vectors in SNL rats normalized the amplitude of puff NMDA–elicited currents of lamina II neurons to vector-treated sham control rats (fig. S11F). Together, these findings indicate that increasing LRRC8A expression in the DRG and spinal cord effectively eliminates presynaptic and postsynaptic NMDAR hyperactivity caused by nerve injury.

DISCUSSION

Here, we showed that neuronal LRRC8A acts as a powerful regulator of synaptic NMDARs and nociceptive hypersensitivity. Central sensitization, which is the fundamental mechanism of neuropathic pain, results from NMDAR-mediated amplification of nociceptive transmission in the spinal cord (2, 4, 7). Under physiological conditions, presynaptic NMDARs at central terminals of DRG neurons are not functionally active. However, these receptors become tonically activated by endogenous glutamate in neuropathic pain conditions (3, 4, 10, 12). In this study, we found that siRNA knockdown of LRRC8A in the DRG and spinal cord or Lrrc8a-cKO in DRG neurons consistently caused nociceptive hypersensitivity. We showed that blocking NMDARs effectively reversed nociceptive hypersensitivity induced by LRRC8A knockdown or Lrrc8a-cKO in DRG neurons. Supporting the role of NMDARs in this neuropathic pain–like phenotype, we discovered that down-regulation of LRRC8A in the DRG and spinal cord substantially increased the activity of both presynaptic and postsynaptic NMDARs in the spinal dorsal horn. In addition, Lrrc8a-cKO in DRG neurons selectively increased the activity of presynaptic NMDARs at the central terminals of DRG neurons. In a heterologous expression system, LRRC8A knockdown profoundly potentiated the activity of NMDARs reconstituted with GluN1/GluN2A or GluN1/GluN2B subunits. Collectively, these complementary loss-of-function analyses provide compelling evidence that LRRC8A proteins act as constitutive inhibitors of nociceptive hypersensitivity by limiting synaptic NMDAR activity.

Our study reveals a physical interaction between LRRC8A, an anion channel protein, and NMDARs, a glutamate-gated cation channel, in the context of nociceptive regulation. The dynamic physical interactions between proteins in living systems are crucial for their roles in biological processes. Investigating these protein-protein interactions is essential for uncovering unknown protein functions at the molecular level and understanding the complex cellular networks involved in neuropathic pain. Many LRR-containing proteins are involved in diverse protein-protein interactions (36, 37, 40). We showed that the LRRC8A-NMDAR interaction occurred in spinal cord tissues from both rats and humans, suggesting that this interaction is conserved across species. Through a series of truncation and mutagenesis analyses, we identified that LRRC8A interacted with NMDARs predominantly through the LRR1 domain on its C terminus. We demonstrated that the peptide mimicking the LRR1 domain of LRRC8A disrupted the LRRC8A-NMDAR interaction, induced nociceptive hypersensitivity, and increased synaptic NMDAR activity in the spinal cord. These results provide direct functional evidence that physical interactions between LRRC8A and NMDARs play a critical role in constitutively limiting nociceptive hyperactivity and synaptic NMDAR activity. We also found that Lrrc8a^ebo/ebo mutant mice, which lack the C-terminal LRR domain of LRRC8A, exhibited NMDAR-dependent nociceptive hypersensitivity. Furthermore, both presynaptic and postsynaptic NMDAR hyperactivities were present in the spinal dorsal horns of these mutant mice. Collectively, these findings provide compelling in vivo evidence underscoring the importance of LRRC8A in regulating NMDAR-mediated nociceptive hypersensitivity.

Under normal physiological conditions, NMDARs are present but functionally inactive inside primary afferent central terminals (4, 8, 41–43). This inactivity could be attributable to the high expression of LRRC8A proteins, which, as demonstrated in the present study, constitutively restrict synaptic NMDAR expression under basal conditions. Although one study reported that GluN1 cKO in DRG neurons induces neuronal hyperexcitability and pain hypersensitivity (44), these findings have not been replicated and are contradicted by multiple independent studies showing that GluN1 deletion in DRG neurons does not affect normal nociception (4, 9, 32, 41, 42). Another report suggested that presynaptic NMDARs at primary afferent terminals are tonically active and suppress synaptic transmission, based on observations that NMDA application reduced evoked AMPAR-EPSC amplitude, whereas AP5 application increased mEPSC frequency (45). However, exogenous NMDA application can activate peri- and extrasynaptic NMDARs, leading to postsynaptic AMPAR suppression through calcineurin-mediated dephosphorylation and endocytosis (46–48). Thus, the observed NMDA effect on evoked EPSCs likely reflects postsynaptic AMPAR inhibition. In our mEPSC recordings, we did not use γ-aminobutyric acid type A (GABA_A) or glycine receptor antagonists. There is extensive evidence showing that AP5 has no appreciable effect on baseline mEPSCs in spinal lamina I and II neurons from naive rats and mice (3, 4, 12, 32, 39, 41, 43, 49, 50). In contrast, Xie et al. (45) applied AP5 together with high concentrations of GABA_A and glycine receptor antagonists, which are known to induce disinhibition and NMDAR hyperactivity (51, 52). Last, the unchanged baseline nociceptive thresholds in GluN1-cKO mice (45) further argue against a major role for presynaptic NMDARs in suppressing nociceptive transmission.

Our findings provide insight into the regulatory mechanism of NMDAR-mediated synaptic plasticity in neuropathic pain. In this study, we showed that siRNA-induced Lrrc8a knockdown or Lrrc8a-cKO in DRG neurons potentiated synaptic expression of NMDARs in the spinal cord, suggesting that LRRC8A constitutively suppresses NMDAR synaptic trafficking. Through physical interaction with NMDARs, LRRC8A may facilitate their intracellular retention, thereby limiting NMDAR synaptic expression and activity. Thus, abundant LRRC8A in normal DRG neurons may have a critical role in inhibiting chronic pain development by restraining synaptic NMDAR expression. We found that nerve injury selectively reduced LRRC8A expression in the DRG, without affecting other subunits of VRACs. When nerve injury reduces LRRC8A expression in the DRG, NMDARs that are no longer bound to LRRC8A could become “unleashed,” allowing them freely traffic to the central terminal of DRG neurons. This could cause hyperactivity of presynaptic NMDARs and potentiate glutamate release from primary afferent nerves through Ca²⁺ influx (53), representing a potential molecular mechanism by which nerve injury induces hyperactivity of presynaptic NMDARs in the spinal dorsal horn. The NMDAR-restraining action of LRRC8A is unlikely to be limited to DRG neurons, because both presynaptic and postsynaptic NMDAR hyperactivities in the spinal cord were observed after siRNA-induced Lrrc8a knockdown and in Lrrc8a^ebo/ebo mutant mice. Therefore, the transition from LRRC8A-bound NMDARs to LRRC8A-free NMDARs likely constitutes a crucial molecular mechanism for the tonic activation of synaptic NMDARs in the spinal dorsal horn under neuropathic pain conditions.

Recent studies indicate that α2δ-1, commonly known as a voltagegated Ca²⁺ channel subunit, is a crucial protein that preferentially interacts with phosphorylated NMDARs to enhance their synaptic trafficking in the spinal cord in neuropathic pain (4, 10, 33). In this study, we demonstrated that LRRC8A interacted with NMDAR subunits expressed in HEK293 cells, suggesting that LRRC8A-NMDAR interaction is likely independent of other neuronal proteins, including α2δ-1. Furthermore, LRRC8A knockdown increased membrane surface expression of NMDARs in HEK293 cells, indicating that LRRC8A regulates NMDARs independently of α2δ-1. Thus, LRRC8A and α2δ-1 appear to have opposite roles in regulating the synaptic expression and activity of NMDARs. These two proteins may maintain a delicate “push-pull” balance in controlling NMDAR trafficking under normal and pathological conditions. After nerve injury, the combined effects of α2δ-1 up-regulation (10, 49) and LRRC8A down-regulation could enhance the interaction between α2δ-1 and NMDARs while concurrently reducing LRRC8A-NMDAR complexes, thereby potentiating the synaptic trafficking of NMDARs in the spinal dorsal horn. Further studies are needed to delineate how LRRC8A and α2δ-1 cooperate to direct the synaptic expression of NMDARs in vivo.

Another notable finding from our study is that lentivirus-mediated Lrrc8a gene transfer to the DRG and spinal cord reversed nerve injury–induced pain hypersensitivity and synaptic NMDAR hyperactivity of spinal dorsal horn neurons. Lentiviral vectors have the advantage of predominantly transfecting mature neurons in vivo and are considered safe for microinjection into the brain in humans (54). The profound and sustained effects of Lrrc8a gene transfer are likely due to the ability of lentiviral vectors to induce LRRC8A expression in both spinal dorsal horn and DRG neurons. This increase in LRRC8A-bound NMDARs eliminated nerve injury–induced NMDAR hyperactivity at both presynaptic and postsynaptic sites in the spinal cord, indicating that LRRC8A abundance governs NMDAR activity on both sides of the synapse. Although nerve injury–induced hyperalgesia was abolished within 2 weeks of Lrrc8a gene transfer, the reversal of tactile allodynia was more gradual, taking about 6 weeks to reach its maximum effect. This more rapid alleviation of hyperalgesia compared with allodynia may be attributed to the possibility that the lentiviral vector initially transduces more high-threshold nociceptive neurons than low-threshold sensory neurons (39). Our findings provide additional convergent evidence for the critical role of LRRC8A in constitutively inhibiting nociceptive hypersensitivity and synaptic NMDAR activity in the spinal dorsal horn. Gene therapy holds promise for providing prolonged pain relief without the need for repeated interventions, making it especially suitable for managing chronic pain. However, lentivirus-mediated LRRC8A expression may carry the risk of unintended effects. Given that intrathecal injection through lumbar puncture is a commonly used and minimally invasive procedure, our findings support the potential viability of this therapeutic strategy for patients with severe, treatment-resistant neuropathic pain.

In contrast with the antinociceptive role of LRRC8A demonstrated in our study, a recent report suggests that LRRC8A expressed in spinal microglia could promote neuropathic pain by releasing adenosine triphosphate (ATP) (55). Also, intrathecal injection of dicoumarol, a nonspecific VRAC inhibitor, has been shown to reduce nerve injury–induced mechanical allodynia (55). However, dicoumarol affects multiple molecular targets, including by having anti-inflammatory actions (56–59), making it challenging to attribute its antinociceptive effect specifically to LRRC8A inhibition. In the present study, we consistently observed NMDAR-mediated nociceptive hyperactivity in rats with siRNA-induced knockdown, Lrrc8a-cKO mice, and Lrrc8a^ebo/ebo mutant mice. Our findings strongly suggest that the interaction between LRRC8A and NMDARs is critical for LRRC8A’s role in restraining chronic pain development. LRRC8A knockdown by siRNA and Lrrc8a mutation in Lrrc8a^ebo/ebo mice are not cell type–specific. Moreover, the complete reversal of nerve injury–induced pain hypersensitivity by virus-mediated LRRC8A expression in the DRG and spinal cord indicates that LRRC8A’s antinociceptive effects are not restricted to neurons alone. Therefore, the overall function of LRRC8A in the DRG and spinal cord appears to be the inhibition of nociceptive hypersensitivity, although its effects may vary across different cellular contexts.

Our study has limitations. First, the role of LRRC8A in regulating NMDARs in neuropathic pain was studied only in rodent models, and its clinical relevance remains to be established through further research. Second, the lack of highly specific VRAC blockers prevented direct assessment of LRRC8A-mediated VRAC activity in NMDAR hyperactivity and pain hypersensitivity induced by nerve injury or Lrrc8a knockdown. DRG neurons have a high intracellular Cl⁻ concentration (60), and nerve injury can augment intracellular Cl⁻ concentrations in spinal dorsal horn neurons (1, 39, 61). Therefore, LRRC8A-mediated VRAC activity is likely associated with depolarizing, rather than hyperpolarizing, currents in DRG and spinal dorsal horn neurons in neuropathic pain conditions. In addition, LRRC8A-mediated VRACs are permeable to ATP and glutamate (23, 55). However, the loss of depolarizing Cl⁻ currents or the reduced release of ATP and glutamate from VRACs in DRG and spinal dorsal horn neurons cannot explain the nociceptive hypersensitivity caused by LRRC8A loss of function in our study. Our findings using the LRR1 peptide, which selectively disrupts LRRC8A-NMDAR interaction, suggest that the LRRC8A protein likely controls nociceptive hypersensitivity by directly regulating synaptic NMDARs, rather than through its role in VRAC function.

In summary, our study identified LRRC8A as a regulatory protein of synaptic NMDARs and nociceptive hyperactivity. The loss of LRRC8A’s restraining action on synaptic NMDARs in the DRG and spinal cord may play a crucial role in the development of neuropathic pain. These findings advance our understanding of the homeostatic synaptic plasticity that governs pain chronicity and open the possibility for developing new strategies to attenuate NMDAR hyperactivity without directly blocking NMDAR channels, thereby potentially avoiding the common side effects associated with NMDAR antagonists. Our findings provide proof of principle that intrathecal Lrrc8a gene delivery is highly effective in rodent models of neuropathic pain.

MATERIALS AND METHODS

Study design

This study aimed to determine whether, and by which mechanisms, the LRRC8A protein regulates nociception and synaptic NMDAR activity at the spinal cord level. To address this, we examined the distribution of LRRC8A in the DRG and spinal cord and assessed how nerve injury affects its expression in these tissues. We systematically evaluated nociceptive behaviors in rats with siRNA-mediated Lrrc8a knockdown or lentivirus-mediated overexpression after nerve injury, as well as in mice with Lrrc8a gene mutation or conditional ablation in DRG neurons. Biochemical assays and electrophysiological analyses were conducted to determine how these interventions influence LRRC8A-NMDAR interactions, synaptic expression, and NMDAR activity. Furthermore, truncation and mutagenesis analyses were performed to identify the LRRC8A domain responsible for its interaction with NMDARs. In addition, we developed a specific interfering peptide to elucidate the functional significance of the LRRC8A-NMDAR interaction in regulating nociception and synaptic NMDAR activity.

All the experimental procedures were approved (approval nos. 1231-RN02 and 0660-RN03) by the Institutional Animal Care and Use Committee of University of Texas MD Anderson Cancer Center. The investigators conducting electrophysiological and behavioral experiments were blinded to the treatments and genotypes. The sample sizes used in the study were estimated using an α value of 0.05 and statistical power of 85% and were based on similar studies we published previously (4, 10, 12, 33). All replications referred to in this study are biological replicates. Animals were assigned to control and treatment groups with a 1:1 allocation ratio as they became available; no specific randomization methods were used. The biochemistry, electrophysiology, and behavior experiments were conducted according to established protocols. Detailed methods are listed in the Supplementary Materials. In addition, the LRRC8A-NMDAR interaction was validated by co-IP using frozen human spinal cord tissues provided by the University of Maryland Brain and Tissue Bank.

Statistical analysis

Data are presented as means ± SEM. The D’Agostino-Pearson normality test was used to assess data normality. A two-tailed Student’s t test was used to compare two groups. One-way or two-way analysis of variance (ANOVA), followed by Dunnett’s or Tukey’s post hoc test, was used for comparisons involving more than two groups. Statistical analyses were performed using Prism software (GraphPad Software Inc.). P < 0.05 was considered statistically significant.

Supplementary Material

Other Supplementary Material for this manuscript includes the following:

Data file S1

MDAR Reproducibility Checklist

Data and materials availability:

All data associated with this study are present in the paper or the Supplementary Materials. Lentivirus vectors expressing full-length Lrrc8a and truncated Lrrc8a constructs generated in this study are available from the corresponding author upon signing a material transfer agreement with the University of Texas MD Anderson Cancer Center. Frozen sperm from Lrrc8a^F443* mice are available from the Genetically Engineered Mouse Facility at MD Anderson Cancer Center (Director: F. Delerue). Lrrc8a^flox/flox mice can be obtained from R.S. upon signing a material transfer agreement with Washington University School of Medicine.