α2δ-2 regulates synaptic GluK1 kainate receptors in Purkinje cells and motor coordination

Abstract

Gabapentin and pregabalin are inhibitory ligands of both α2δ-1 and α2δ-2 proteins (also known as subunits of voltage-activated Ca²⁺ channels) and are commonly prescribed for the treatment of neuropathic pain and epilepsy. However, these drugs can cause gait disorders and ataxia through unknown mechanisms. α2δ-2 and GluK1, a glutamate-gated kainate receptor subtype, are coexpressed in cerebellar Purkinje cells. In this study, we used a heterologous expression system and Purkinje cells to investigate the potential role of α2δ-2 in regulating GluK1-containing kainate receptor activity.

Whole-cell patch-clamp recordings showed that α2δ-2 coexpression augmented GluK1, but not GluK2, currents in HEK293 cells, and pregabalin abolished this augmentation. Pregabalin lost its inhibitory effect on GluK1 currents in HEK293 cells expressing both GluK1 and the α2δ-2(R282A) mutant. Blocking GluK1-containing receptors with UBP310 substantially reduced the amplitude of excitatory post-synaptic currents at parallel fibre–Purkinje cell synapses in mice. Also, pregabalin markedly attenuated the amplitude of excitatory post-synaptic currents and currents elicited by ATPA, a selective GluK1 receptor agonist, in Purkinje cells in Cacna2d1 knockout mice. Co-immunoprecipitation assays indicated that α2δ-2, but not α2δ-1, formed a protein complex with GluK1 in cerebellar tissues and HEK293 cells through its C-terminus. Furthermore, α2δ-2 coexpression potentiated surface expression of GluK1 proteins in HEK293 cells, whereas pregabalin reduced GluK1 proteins in cerebellar synaptosomes.

Disrupting α2δ-2–GluK1 interactions using α2δ-2 C-terminus peptide abrogated the potentiating effect of α2δ-2 on GluK1 currents and attenuated the amplitude of GluK1-mediated excitatory post-synaptic currents in Purkinje cells. However, neither pregabalin nor α2δ-2 C-terminus peptide had significant effect on P/Q-type currents in HEK293 cells. Additionally, CRISPR/Cas9-induced conditional knockdown of Cacna2d2 or Grik1 in Purkinje cells, in addition to microinjection of α2δ-2 C-terminus peptide or UBP310 into the cerebellum, substantially impaired beam-walking and rotarod performance in mice.

Our study reveals that α2δ-2 directly interacts with GluK1 independently of its conventional role as a voltage-activated Ca²⁺ channel subunit. α2δ-2 regulates motor coordination by promoting synaptic expression and activity in GluK1-containing kainate receptors in Purkinje cells.

Using cell lines and Purkinje cells, Zhou et al. show that the α2δ-2 protein physically interacts with and promotes synaptic trafficking of GluK1 kainate receptors. Disrupting α2δ-2–GluK1 complexes in the cerebellum impairs motor coordination, providing an explanation for why gabapentinoids can cause gait abnormalities and ataxia.

Introduction

The α2δ proteins (α2δ-1, α2δ-2, α2δ-3 and α2δ-4) are commonly regarded as auxiliary subunits of voltage-activated Ca²⁺ channels (VACCs). Gabapentinoids, such as gabapentin and pregabalin, are widely prescribed for the treatment of epilepsy and neuropathic pain. These drugs bind similarly to the α2δ-1 and α2δ-2 proteins, which are encoded by Cacna2d1 and Cacna2d2, respectively.^1-3 However, gabapentinoids have little effect on VACC activity or VACC-mediated neurotransmitter release from presynaptic terminals.^4-8 Recent studies have revealed that α2δ-1, but not α2δ-2 or α2δ-3, physically interacts with glutamate N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors, regulating their synaptic trafficking and composition independently of VACCs.^6,9-11 Clinical studies indicate that gabapentinoids cause gait abnormalities and ataxia, particularly in the elderly.^12-15 Given that Cacna2d1 knockout mice do not exhibit gait disorders,¹⁶ the adverse effects of gabapentinoids are unlikely to be caused by α2δ-1 inhibition.

The cerebellum plays a crucial role in the accurate control and timing of movement, with the Purkinje cells in the cerebellar cortex being fundamental for motor coordination.^17,18 Purkinje cells use complex ionic mechanisms to perform their computational tasks and generate the output signals needed for the execution of precise movements; thus, gene mutations affecting the ion channels expressed in Purkinje cells usually cause ataxic symptoms.^18,19 Mice with Cacna2d2 mutations or Cacna2d2 knockout display a distinct ataxia phenotype.^20-22 The α2δ-1 and α2δ-2 proteins are differentially distributed across various brain regions. For instance, α2δ-1 shows high expression levels in the hippocampus, hypothalamus and cerebellar granular cells, whereas α2δ-2 shows high expression levels in the habenula, septum, reticular thalamic nucleus and cerebellar Purkinje cells.^23-25 Although α2δ-2 is highly expressed in Purkinje cells, its functional significance remains elusive.

Kainate receptors (KARs), one of the three major classes of ionotropic glutamate receptors, are tetrameric channels made up of GluK1–GluK5 subunits. GluK1–GluK3, which are low-affinity subunits, form functional homomeric and heteromeric KARs. In contrast, GluK4 and GluK5, which are high-affinity subunits, must be co-assembled with GluK1–GluK3 for functional expression. Unlike the broadly expressed AMPA receptors and NMDA receptors in the brain, KARs are expressed only at a subset of glutamatergic synapses.^26-28 In the cerebellum, GluK1 is mainly present in Purkinje cells, whereas GluK2 is most abundant in granule cells.²⁹ GluK3 is poorly expressed in the cerebellum.³⁰ GluK1-containing KARs mediate glutamatergic input to Purkinje cells.³¹ However, the functional significance of GluK1 in Purkinje cells with regard to motor coordination is largely unknown.

To address these gaps in knowledge, we used a heterologous expression system and Purkinje cells to explore the potential role of α2δ-2 in regulating GluK1-containing KAR activity. We discovered that α2δ-2 potentiated GluK1 currents and enhanced GluK1-mediated excitatory synaptic transmission in Purkinje cells. Strikingly, α2δ-2 formed a protein complex with GluK1 through its C-terminal domain, which increased the surface and synaptic expression of GluK1-containing KARs. Furthermore, conditional knockdown of Cacna2d2 or Grik1 in Purkinje cells, in addition to disrupting the α2δ-2–GluK1 interaction in the cerebellum, impaired motor balance and performance. These findings reveal a non-canonical function of α2δ-2 as a key GluK1-interacting protein crucial for controlling motor coordination.

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. The generation of Cacna2d1 knockout mice was previously reported.³ Cas9^Flox/+ mice (strain #026175) and BAC-Pcp2-IRES-Cre mice (Pcp2^Cre/+ mice, strain #010536) were purchased from The Jackson Laboratory. Pcp2^Cre/+ mice express Cre recombinase under the control of the mouse Purkinje cell protein 2 (Pcp2).³² To generate Cas9 ‘knock-in’ in Purkinje cells, we crossed Cas9^Flox/+ mice with Pcp2^Cre/+ mice. Both male and female adult mice were used for electrophysiological and behavioural experiments. Detailed methods can be found in the Supplementary material.

Cell transfection and mutagenesis

HEK293 cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum at 37°C in a 5% CO₂ incubator. Myc-tagged GluK1(Q) or GluK2(Q) was used for the biochemistry experiments. In electrophysiological assays, coexpression of green fluorescent protein (GFP) was used to facilitate the identification of transfected cells.⁶ PolyJet DNA In Vitro Transfection Reagent was used for all transfection experiments. Mutagenesis was conducted using the In-Fusion HD Cloning Kit. A point mutation was introduced by substituting arginine with alanine at the pregabalin binding site of the mouse α2δ-2(R282). For deletion mutagenesis (α2δ-2_ΔVWA), the VWA domain (UniProt: 294–472) of the α2δ-2 subunit was removed.⁶

Preparation of lentiviral vectors

The lentivirus packaging and concentrating procedures followed those described in our previous study.³³ We designed guide RNAs (gRNAs) targeting the mouse Cacna2d2 and Grik1 genes using the ‘GPP sgRNA Design Tool’ online.

Electrophysiological recording in HEK293 cells

Whole-cell patch-clamp recordings were conducted using an EPC-10 amplifier. Glutamate (10 mM) was applied for 5 ms to elicit currents, which were recorded at a holding potential of −60 mV.⁶ The whole-cell VACC current was elicited by applying a depolarizing pulse from a holding potential of −90 to 0 mV for 200 ms.^9,34

Electrophysiological recordings in cerebellum slices

The brains were rapidly removed from isoflurane-anaesthetized mice and were immediately placed in ice-cold sucrose-containing artificial CSF presaturated with 95% O₂ and 5% CO₂. Transverse slices (300 μm thick) of the cerebellums were cut in ice-cold, sucrose-containing artificial CSF and pre-incubated in Krebs solution oxygenated with 95% O₂ and 5% CO₂ at 34°C for ≥1 h before being used for recordings.

Purkinje cells from the vermes (lobe VI) were visualized and identified by the soma size. Excitatory post-synaptic currents (EPSCs) were recorded using whole-cell voltage-clamp techniques. The EPSCs were recorded at a holding potential of −60 mV and evoked from the parallel fibre input to the Purkinje neuron. To identify the parallel fibre-evoked EPSCs,^35,36 the paired-pulse ratio was measured by applying two pulses at an interval of 50 ms. Only cells with EPSCs showing paired-pulse facilitation were chosen for further tests. The post-synaptic GluK1 receptor currents were elicited by puff application of 30 μM (RS)-2-amino-3-(3-hydroxy-5-tert-butylisoxazol-4-yl)propanoic acid (ATPA) directly onto the Purkinje cell at an angle of 30° using a positive-pressure system.

Cerebellar microinjection

Lentiviral vectors expressing specific gRNA sequences were microinjected into the cerebellar vermes of isoflurane-anaesthetized mice using glass pipettes.³⁷ A volume of 500 nl lentiviruses was delivered per injection, and a total of five injections were given. After a 14-day recovery period, the mice underwent behavioural tests, after which tissues were collected.

UBP310 (50 nmol/125 nl), α2δ-2CT peptide (400 pmol/400 nl) or control peptide (400 pmol/400 nl) was microinjected into the cerebellar vermes of mice through an implanted cannula. After a 10 day recovery period, mice were anaesthetized with 2% isoflurane, and a needle was inserted into the guide cannula for cerebellar microinjection. Behavioural tests were conducted 20 min after injection.

Beam-walking and rotarod performance tests

Motor coordination and balance in mice were evaluated by measuring the time required to cross narrow balance beams.^38,39 The mice were trained for 3 days (four trials per day) to cross the 12-mm-square beam. During testing, each animal was given two consecutive trials to cross square beams first and then round beams, in order from the widest to the narrowest. The rotarod test was conducted as previously described.¹⁶ Mice were positioned on an accelerating rotarod in the opposite direction from that of rotation. The rotating speed started at 4 rpm and gradually increased to 40 rpm over a period of 300 s for each trial.

Biotinylation assay

Cell surface biotinylation and isolation of the surface proteins were carried out using the Cell surface Protein Isolation Kit. The surface protein samples were analysed using immunoblotting and probed with rabbit anti-myc antibodies (1:1000 dilution, #2278, Cell Signaling), mouse anti-α2δ-2 antibody (1:1000 dilution, #sc-365911, Santa Cruz Biotechnology) and rabbit anti-Na⁺–K⁺-ATPase antibody (1:1000 dilution, #3010, Cell Signaling).

Cerebellar synaptosomes

The mouse cerebellum slices were homogenized using a Dounce tissue grinder. Synaptic proteins were extracted and enriched using Syn-PER reagent. The homogenate was centrifuged at 1200g for 10 min at 4°C. The supernatant was then centrifuged at 15 000g for 20 min to obtain the crude synaptosomes.⁶ The synaptosomal proteins were probed with rabbit anti-GluK1 antibody (1:500 dilution, #AGC-008, Alomone Labs) or a mouse anti-PSD-95 antibody (1:1000 dilution, #75-348, NeuroMab).

Co-immunoprecipitation and immunoblotting

In co-immunoprecipitation (co-IP) using cerebellum tissues, the mouse anti-α2δ-2 antibody (#sc-365911, Santa Cruz Biotechnology) was used for the pulldown. Rabbit anti-myc antibody (mAb #2278, Cell Signaling) was used for immunoprecipitation in HEK293 cells transfected with myc-tagged proteins. Yellow fluorescent protein (YFP)-tagged chimeric constructs, including α2δ-1_NT(α2δ-2) and α2δ-1_CT(α2δ-2), in which the N- or C-terminus of α2δ-1 was replaced with the corresponding domain from α2δ-2, were generated as we previously described.⁶ The antibodies were pre-incubated with protein G agarose beads at 22°C for 1 h, after which the protein samples were exposed to the antibody-conjugated beads at 4°C overnight.

Co-immunoprecipitated samples from mouse cerebellum tissues were probed using rabbit anti-GluK1, rabbit anti-GluK2 and rabbit anti-α2δ-2 antibodies (1:500 dilution; #AGC-008, #AGC-009 and #ACC-102, respectively; Alomone Labs). For co-IP using HEK293 cells, α2δ-1, α2δ-2 and myc-tagged KAR proteins were detected using mouse anti-α2δ-1, mouse anti-α2δ-2 and mouse anti-myc antibodies (1:1000 dilution; #sc-271697, #sc-365911 and #sc-40, respectively; Santa Cruz Biotechnology), whereas the mouse anti-GFP antibody (1:1000 dilution, #75-132, NeuroMab) was used to probe the YFP-tagged and YFP-tagged α2δ-1 chimeras.

Statistical analyses

Data are presented as means ± standard error of the mean (SEM). Group sizes were predetermined according to published studies in the field.^6,9,40,41 We used Student’s two-tailed t-test to compare the means of two groups and one-way or two-way ANOVA to compare the means of more than two groups. Differences were considered statistically significant if the P-value was <0.05.

Results

α2δ-2 enhances GluK1 currents independently of VACCs

In the cerebellum, GluK1 is expressed in Purkinje cells, whereas GluK2 is expressed mainly in granule cells.^29,31,42 Compared with the surface expression of GluK2, the surface expression of GluK1 is limited, resulting in a modest whole-cell current.⁴³ Initially, we used a heterologous expression system to determine whether α2δ-2 regulates GluK1 and GluK2 currents in HEK293 cells, which are devoid of VACC subunits. Because GluK1(R) produces no currents in response to glutamate application, HEK293 cells were transfected with GluK1(Q) or GluK2(Q), either alone or with α2δ-1 or α2δ-2. KAR currents were induced by applying 10 mM glutamate for 5 ms. The current density was much greater in the HEK293 cells expressing GluK2 than in those expressing GluK1 (Fig. 1A and B). Remarkably, coexpression with α2δ-2 profoundly increased GluK1 currents, nearly tripling the current level seen in the cells expressing GluK1 alone [34.88 ± 5.42 pA/pF for GluK1 + α2δ-2, n = 13 cells versus 12.49 ± 1.27 pA/pF for GluK1 alone, n = 14 cells; F(3,47) = 7.363, P = 0.0053; Fig. 1A]. In contrast, coexpression of α2δ-1 did not significantly increase GluK1 currents in HEK293 cells (n = 12 cells; Fig. 1A). Furthermore, the current density did not differ significantly between HEK293 cells expressing GluK2 alone and those expressing both GluK2 and α2δ-2 (n = 10 cells per group; Fig. 1B).

Figure 1.

α2δ-2 enhances GluK1 currents, an effect that is abolished by pregabalin. (A) Representative current traces and mean data show the GluK1 currents, elicited by glutamate (10 mM for 5 ms), in HEK293 cells expressing GluK1 alone or coexpressing GluK1 with α2δ-1, α2δ-2 or the α2δ-2(R282A) mutant [n = 14 cells in the GluK1 alone group, n = 12 cells in the GluK1 + α2δ-1 group, n = 13 cells in the GluK1 + α2δ-2 group and n = 12 cells in the GluK1 + α2δ-2(R282A) group]. (B) Original current traces and mean data show the GluK2 currents in HEK293 cells expressing GluK2 with or without α2δ-2 (n = 10 cells per group). (C) Original current traces and mean data show the current in HEK293 cells expressing GluK1/GluK4 with or without α2δ-2 (n = 13 cells per group). (D) Representative current traces and quantification show the differential effect of pregabalin (PGB) on GluK1 currents in HEK293 cells expressing GluK1 alone and coexpressing GluK1 with wild-type α2δ-2 or the α2δ-2(R282A) mutant [n = 13 cells in the GluK1 + PGB group, n = 12 cells in the GluK1 + α2δ-2 + PGB group and n = 11 cells in the GluK1 + α2δ-2(R282A) + PGB group]. Data are expressed as mean ± standard error of the mean. **P < 0.01, ***P < 0.001 (one-way ANOVA followed by Tukey’s post hoc test in A and D; Student’s two-tailed t-test in B and C).

A low level of GluK4 is expressed in Purkinje cells,⁴⁴ which might form heteromeric GluK1/GluK4 receptors. Hence, we determined whether α2δ-2 coexpression affects GluK1/GluK4 currents in HEK293 cells. The current density was significantly smaller in HEK293 cells expressing GluK1/GluK4 than in those expressing GluK1 alone (Fig. 1A and C). Notably, coexpression of α2δ-2 also significantly increased the current density of GluK1/GluK4 receptors in HEK293 cells [n = 13 cells per group, t(24) = 3.059, P = 0.0054; Fig. 1C].

We then determined whether gabapentinoids block GluK1 currents potentiated by α2δ-2 coexpression. In HEK293 cells expressing GluK1 alone, treatment with 20 μM pregabalin for 30 min had no significant effect on the GluK1 current density (n = 13 cells). However, treatment with pregabalin abrogated the α2δ-2 coexpression-induced potentiation in GluK1 currents (n = 12 cells; Fig. 1D). Gabapentinoids bind mainly to the third arginine (R) within an RRR motif located near the N-terminus of α2δ-1 (R217) and α2δ-2(R282).^45-47 Interestingly, like the wild-type α2δ-2, coexpression of the α2δ-2(R282A) mutant markedly increased the amplitude of GluK1 currents [38.33 ± 8.34 pA/pF, n = 12 cells; F(3,47) = 7.363, P = 0.0015; Fig. 1A]. However, treatment with pregabalin failed to reduce GluK1 currents in HEK293 cells coexpressing GluK1 and the α2δ-2(R282A) mutant (n = 11 cells; Fig. 1D). Collectively, these results indicate that α2δ-2 is capable of potentiating GluK1-containing currents independently of VACCs and that gabapentinoids inhibit GluK1 currents only in the presence of α2δ-2.

α2δ-2 is crucial for GluK1-mediated excitatory post-synaptic currents in Purkinje cells

Both α2δ-2 and GluK1 are expressed in Purkinje cells,^25,29 and post-synaptic GluK1 receptors have a substantial role in mediating excitatory glutamatergic transmission in these cells.³¹ We used pregabalin as an α2δ-2 inhibitory ligand to determine whether α2δ-2 regulates GluK1-mediated glutamatergic synaptic input to Purkinje cells. To exclude the confounding effect of pregabalin on α2δ-1, we obtained cerebellar slices from Cacna2d1 knockout mice and recorded parallel fibre-evoked EPSCs in Purkinje cells. Purkinje cells receive substantial glutamatergic input from parallel fibres,⁴⁸ and parallel fibre–Purkinje cell synapses play a central role in motor learning and coordination.^35,36 Because both AMPA receptors and GluK1-containing KARs are involved in excitatory synaptic transmission in Purkinje cells,³¹ we first blocked AMPA receptor-mediated EPSCs by continuously applying 30 µM GYKI52466, a specific AMPA receptor antagonist.⁴⁹ Bath application of 10 µM UBP310, a selective GluK1 receptor antagonist,⁵⁰ substantially attenuated the amplitude of evoked EPSCs in Purkinje cells [n = 14 neurons, F(5,75) = 6.465, P = 0.0011; Fig. 2A]. Remarkably, pretreatment with 20 µM pregabalin for 30 min markedly reduced the baseline amplitude of evoked EPSCs [n = 13 neurons, F(5,75) = 6.465, P = 0.0114]. In neurons pretreated with pregabalin, subsequent bath application of UBP310 had no further effect on the amplitude of evoked EPSCs (Fig. 2A), suggesting that α2δ-2 activity is essential for the presence of GluK1 receptors at parallel fibre–Purkinje cell synapses.

Figure 2.

α2δ-2 augments synaptic GluK1-mediated excitatory post-synaptic currents and GluK1 currents in Purkinje cells in Cacna2d1 knockout mice. (A) Representative current traces and quantification show the effect of bath application of 10 μM UBP310 on the excitatory post-synaptic currents (EPSCs) of Purkinje cells evoked by parallel fibre stimulation in Cacna2d1 knockout mice (n = 14 neurons in the vehicle group and n = 13 neurons in the pregabalin group). EPSCs were recorded in the presence of 30 µM of GYKI52466. (B) Representative current traces and quantification show the paired-pulse facilitation of EPSCs in Purkinje cells evoked by parallel fibre stimulation in Cacna2d1 knockout mice (n = 15 neurons in the vehicle group and n = 13 neurons in the pregabalin group). (C) Original current traces and mean data show the effect of pregabalin on currents elicited by puff application of 30 μM ATPA to Purkinje cells in Cacna2d1 knockout mice (n = 15 neurons in the vehicle group and n = 13 neurons in the pregabalin group). Cerebellar tissue slices were pretreated with 20 μM of pregabalin or vehicle for 30 min before electrophysiological recordings. Data are expressed as means ± standard error of the mean. *P < 0.05, **P < 0.01 (two-way ANOVA followed by Tukey’s post hoc test for the data in A; Student’s two-tailed t-test for the data in C).

Pregabalin treatment had no significant effect on the baseline paired-pulse ratio of evoked EPSCs in Purkinje cells from cerebellar slices of Cacna2d1 knockout mice (Fig. 2B). To determine the role of α2δ-2 in regulating post-synaptic GluK1 currents in Purkinje cells directly, we measured currents elicited by puff application of 30 μM ATPA, a relatively selective GluK1 receptor agonist,⁵¹ onto Purkinje cells in cerebellar slices from Cacna2d1 knockout mice. Pretreatment with 20 µM pregabalin significantly reduced the amplitude of puff ATPA-elicited currents (Fig. 2C). These findings indicate that α2δ-2 tonically controls glutamatergic input to Purkinje cells via augmenting post-synaptic GluK1-containing KARs.

α2δ-2 forms a protein complex with GluK1 in vitro and in vivo

Our experiments above demonstrated that α2δ-2 functionally controls GluK1 receptors in vitro and in vivo. Next, we sought to determine whether α2δ-2 interacts directly with GluK1 proteins. To this end, we performed co-immunoprecipitation (co-IP) assays using protein samples obtained from HEK293 cells coexpressing myc-tagged GluK1 with α2δ-1 or α2δ-2. The pulldown was conducted using an anti-myc antibody, and the precipitates were subsequently probed with an anti-α2δ-1 or anti-α2δ-2 antibody. These co-IP assays showed that GluK1 immunoprecipitated α2δ-2, but not α2δ-1 (Fig. 3A and B). Also, GluK2 weakly precipitated α2δ-2 in HEK293 cells coexpressing myc-tagged GluK2 and α2δ-2 (Fig. 3A).

Figure 3.

α2δ-2 primarily interacts with GluK1 through its C-terminal domain. (A) Co-immunoprecipitation analysis shows the interaction of α2δ-2 with GluK1 or GluK2 proteins in HEK293 cells coexpressing α2δ-2 with myc-tagged GluK1 or myc-tagged GluK2. (B) Co-immunoprecipitation analysis shows the lack of interaction between α2δ-1 and GluK1 or GluK2 in HEK293 cells coexpressing α2δ-1 with myc-tagged GluK1 or myc-tagged GluK2. (C) Co-immunoprecipitation analysis shows the interaction of α2δ-2 and GluK1 proteins in the mouse cerebellum. (D) Co-immunoprecipitation analysis shows the weak interaction between α2δ-2 and GluK2 in the mouse cerebellum. Samples from two different animals were loaded into two lanes in C and D. (E) Co-immunoprecipitation analysis shows the interaction between GluK1 and truncated α2δ-2 lacking the VWA domain in HEK293 cells coexpressing myc-tagged GluK1 with full-length α2δ-2 or α2δ-2_ΔVWA. (F) Co-immunoprecipitation analysis shows the interaction of GluK1 with the α2δ-2 C-terminus, but not the α2δ-2 N-terminus, in HEK293 cells coexpressing myc-tagged GluK1 with wild-type α2δ-1, YFP-tagged α2δ-1_NT(α2δ-2) or YFP-tagged α2δ-1_CT(α2δ-2). In A, B, E and F, proteins extracted from HEK293 cells were immunoprecipitated (IP) initially with a rabbit anti-myc antibody or IgG. Immunoblotting (IB) was then performed using mouse anti-α2δ-1, mouse anti-α2δ-2, mouse anti-GFP or mouse anti-myc antibodies. The anti-GFP antibody cross-reacts with Yellow fluorescent protein (YFP). In C and D, proteins extracted from the mouse cerebellum were immunoprecipitated initially with a mouse anti-α2δ-2 antibody or IgG. Immunoblotting was then performed using rabbit anti-GluK1 or rabbit anti-GluK2 antibodies. IgG and inputs (tissues or cell lysates only, without immunoprecipitation) were used as negative and positive controls, respectively. Similar data were obtained in four independent experiments.

We then performed co-IP assays using mouse cerebellar tissues. The anti-α2δ-2 antibody precipitated GluK1 and only weakly pulled down GluK2 (Fig. 3C and D). These findings align with those of our electrophysiological experiments and suggest that α2δ-2 interacts directly with GluK1 proteins both in vitro and in vivo.

α2δ-2 interacts with GluK1 receptors predominantly via its C-terminal domain

Subsequently, we attempted to identify the molecular determinant involved in the α2δ-2–GluK1 interaction. α2δ interacts with α1 subunits of VACCs through their conserved von Willebrand A (VWA) domains present on α2 proteins.^5,52 To determine whether α2δ-2 interacts with GluK1 via its VWA domain, we initially deleted the VWA domain of α2δ-2 (α2δ-2_ΔVWA). Strikingly, co-IP assays showed that GluK1 similarly precipitated α2δ-2 in HEK293 cells expressing myc-tagged GluK1 with either wild-type α2δ-2 or the α2δ-2_ΔVWA mutant (Fig. 3E), indicating that truncation of the VWA domain had no effect on the α2δ-2–GluK1 interaction.

Gabapentinoids bind to an RRR motif located near the N-terminus of the α2δ subunits.^46,47,53 In contrast, α2δ-1 interacts with AMPA and NMDA receptors via its C-terminus on the δ-1 sequence.^6,10 Because α2δ-1 does not interact with GluK1, we generated chimeric constructs, including α2δ-1_NT(α2δ-2) and α2δ-1_CT(α2δ-2), in which the N- or C-terminus of α2δ-1 was replaced with the corresponding domain from α2δ-2, as described previously.⁶ We expressed these YFP-tagged chimeras individually with myc-tagged GluK1 in HEK293 cells. Co-IP assays revealed that GluK1 precipitated α2δ-1_CT(α2δ-2) but not α2δ-1_NT(α2δ-2) (Fig. 3F). These data strongly suggest that α2δ-2 interacts with GluK1 receptors through its C-terminus.

α2δ-2 promotes surface and synaptic expression of GluK1 proteins

α2δ-1 promotes the surface and synaptic trafficking of NMDA receptors.^6,9,41 Similar to α2δ-1, α2δ-2 is a highly glycosylated protein^54,55 that might also facilitate the surface expression of GluK1 receptors. To test this hypothesis, we conducted biotinylation experiments⁶ to isolate cell surface proteins from HEK293 cells expressing myc-tagged GluK1 or myc-tagged GluK2, with and without α2δ-2. Immunoblotting assays revealed that the surface protein level of GluK1 was much higher in HEK293 cells coexpressing α2δ-2 and GluK1 than in cells expressing GluK1 alone (n = 8 experiments; Fig. 4A and B). However, α2δ-2 coexpression did not significantly affect the surface protein levels of GluK2 (n = 8 experiments; Fig. 4A and B). Furthermore, treatment with 20 µM pregabalin for 30 min diminished the augmentation of GluK1 surface proteins by α2δ-2 coexpression (n = 8 experiments). Treatment with pregabalin also markedly reduced surface protein levels of α2δ-2 (Fig. 4A and B).

Figure 4.

α2δ-2 facilitates cell surface and synaptic expression of GluK1 receptors in vitro and in vivo. (A and B) Representative blotting images (A) and quantification (B) show the effect of pregabalin (PGB) on surface protein levels of GluK1 or GluK2 in HEK293 cells coexpressing α2δ-2 with myc-tagged GluK1 or myc-tagged GluK2 (n = 8 independent experiments). Na⁺–K⁺-ATPase, a known membrane protein marker, was used as an internal control on the same gel. (C and D) Representative blotting images (C) and quantification (D) of the protein levels of GluK1 in synaptosomes isolated from the cerebellum of Cacna2d1 knockout mice. PSD-95, a synaptic protein, was used as the internal control for normalizing the protein level on the same gel. The mouse cerebellum was separated into two equal halves, sliced, and incubated with vehicle or 20 μM of PGB for 30 min before synaptosome isolation (n = 10 mice). IB = immunoblotting. Data are expressed as means ± standard error of the mean. **P < 0.01, ***P < 0.001 (one-way ANOVA followed by Tukey’s post hoc test or Student’s two-tailed t-test).

In addition, we isolated synaptosomes from cerebellar tissues of Cacna2d1 knockout mice to determine whether α2δ-2 inhibition affects synaptic GluK1 protein levels. Prior to synaptosome extraction, the cerebellar hemispheres were separated and sliced. One cerebellar hemisphere was incubated with 20 µM pregabalin for 30 min, whereas the other side was treated with vehicle as a control. Subsequently, we quantified synaptic GluK1 protein levels using immunoblotting. Compared with the vehicle group, treatment with pregabalin markedly reduced GluK1 protein levels in the cerebellar synaptosomes [n = 10 mice per group; t(9) = 5.369, P = 0.0005; Fig. 4C and D]. These findings highlight the crucial role of α2δ-2 in potentiating the surface and synaptic trafficking of GluK1 receptors.

Importance of α2δ-2–GluK1 interactions in regulating GluK1 receptor activity in vitro and in vivo

Given the critical involvement of the C-terminus of α2δ-2 in its interaction with GluK1 receptors, we designed two C-terminus peptides (α2δ-2CT peptide A, GASFPPSLGVLVSLQLLLLLGLPPRPQPQV; and α2δ-2CT peptide B, VLVSLQLLLLLGLPPRPQPQVHSFAASRHL) that mimic the C-terminal domain of α2δ-2. These peptides were fused with the cell-penetrating peptide Tat (YGRKKRRQRRR) to disrupt intracellular α2δ-2–GluK1 interactions. We used a sequence-scrambled peptide fused with Tat as the control peptide. Co-IP assays demonstrated that treatment with 1 µM of either α2δ-2CT peptide A or α2δ-2CT peptide B for 30 min greatly reduced α2δ-2 protein levels precipitated by the anti-myc antibody in HEK293 cells cotransfected with α2δ-2 and myc-tagged GluK1 [n = 8 experiments, F(2,21) = 9.999, P = 0.0009; Fig. 5A].

Figure 5.

α2δ-2 enhances GluK1 currents via its physical interaction with GluK1. (A) Representative blotting images and quantification show the effect of α2δ-2CT peptides on the α2δ-2–GluK1 interaction in HEK293 cells coexpressing α2δ-2 with myc-tagged GluK1 (n = 8 independent experiments). HEK293 cells were treated with 1 μM of α2δ-2CT peptides (Pept-A and Pept-B) or 1 μM of control peptide for 30 min. (B) Original current traces and mean data show the effect of treatment with 1 μM of control peptides or 1 μM of α2δ-2CT peptides on GluK1 currents in HEK293 cells (n = 13 cells in the GluK1 group, n = 13 cells in the GluK1/α2δ-2 + control peptide group, n = 12 cells in the GluK1/α2δ-2 + α2δ-2CT pept-A group and n = 12 cells in the GluK1/α2δ-2 + α2δ-2CT pept-B group). (C) Original current traces and quantification show the lack of an effect of pregabalin (PGB) or α2δ-2CT peptide on P/Q-type VACC currents in HEK293 cells coexpressing Cavα1A, Cavβ4 and α2δ-2 (n = 16 cells in the vehicle group, n = 16 cells in the PBG group and n = 18 in the α2δ-2CT peptide group). HEK293 cells were pretreated with 20 μM PGB or 1 μM α2δ-2 CT peptide for 30 min before recordings. Data are expressed as means ± standard error of the mean. **P < 0.01, ***P < 0.001 (one-way ANOVA followed by Tukey’s post hoc test).

Moreover, whole-cell recordings showed that α2δ-2 coexpression markedly increased GluK1 currents in HEK293 cells treated with the control peptide but not in HEK293 cells treated with α2δ-2CT peptides [n = 13 cells for control peptide groups, n = 12 cells for α2δ-2CT peptide A and B groups, F(3,46) = 10.85, P < 0.0001; Fig. 5B]. Because α2δ-2CT peptide B has a slightly more potent effect compared with α2δ-2CT peptide A, we mainly used α2δ-2CT peptide B for our subsequent experiments.

We also determined whether α2δ-2CT peptide and pregabalin inhibit VACC currents. Because P/Q-type channels are responsible for ∼80% of high-threshold VACC currents in Purkinje cells,⁵⁶ we conducted electrophysiological experiments in HEK293 cells transfected with Cav2.1, Cavβ4 and α2δ-2 subunits. Whole-cell recordings showed that neither pregabalin nor α2δ-2CT peptide had any significant effect on reconstituted P/Q-type currents in HEK293 cells (Fig. 5C). These results further demonstrate that α2δ-2 regulates GluK1 currents independently of VACCs.

Next, we determined the functional significance of α2δ-2-bound GluK1 receptors in regulating glutamatergic input to Purkinje cells. Co-IP assays showed that treatment with 1 µM of α2δ-2CT peptide for 30 min significantly reduced GluK1 protein levels precipitated by the anti-α2δ-2 antibody in cerebellar tissues [n = 7 mice per group, F(2,18) =10.79, P = 0.0008; Fig. 6A]. Whole-cell recordings of Purkinje cells in cerebellum tissue slices showed that α2δ-2CT peptide markedly decreased the baseline amplitude of EPSCs evoked by parallel fibre stimulation in the presence of 30 µM GYKI52466 [n = 13 neurons for the control peptide group; n = 15 neurons for the α2δ-2CT peptide group; F(5,78) = 11.47, P = 0.0002; Fig. 6B]. Notably, bath application of UBP310 significantly reduced the amplitude of evoked EPSCs in Purkinje cells in tissue slices treated with the control peptide, but not in slices treated with α2δ-2CT peptide. The baseline paired-pulse ratio of evoked EPSCs in Purkinje cells did not differ significantly between the α2δ-2CT peptide and control peptide groups (Fig. 6C).

Figure 6.

α2δ-2-bound GluK1 receptors mediate glutamatergic input to cerebellar Purkinje cells. (A) Co-immunoprecipitation analysis shows the effect of α2δ-2CT peptides (Pept-A and Pept-B) on the α2δ-2–GluK1 interaction in the mouse cerebellum (n = 7 mice per group). Cerebellar tissue slices were pretreated with 1 μM α2δ-2CT peptides (Pept-A and Pept-B) or 1 μM control peptide for 30 min. Samples from two different animals were used in representative blotting images. (B) Representative current traces and mean data show the effect of bath application of 10 μM UBP310 on the excitatory post-synaptic currents (EPSCs) of Purkinje cells evoked by parallel fibre stimulation in the mouse cerebellar slices pretreated with α2δ-2CT peptide or control peptide (n = 13 neurons in the control peptide group and n = 15 neurons in the α2δ-2CT peptide group). EPSCs were recorded in the presence of 30 µM GYKI52466. (C) Representative current traces and quantification show the paired-pulse facilitation of EPSCs in Purkinje cells evoked by parallel fibre stimulation in cerebellar slices pretreated with α2δ-2CT peptide or control peptide (n = 13 neurons in the control peptide group and n = 15 neurons in the α2δ-2CT peptide group). (D) Original current traces and quantification show the effect of α2δ-2CT peptide on currents elicited by puff application of 30 μM ATPA to Purkinje cells (n = 14 neurons per group). Cerebellar slices were pretreated with 1 μM control peptides or 1 μM α2δ-2CT peptide for 30 min before whole-cell recordings. Data are expressed as means ± SEM. **P < 0.01, ***P < 0.001 (repeated-measures ANOVA followed by Dunnett’s post hoc test in A; two-way ANOVA followed by Tukey’s post hoc test in B; Student’s two-tailed t-test for the data in D).

In addition, the current amplitude elicited by puff ATPA application to Purkinje cells was much smaller in cerebellar tissue slices treated with α2δ-2CT peptide than in slices treated with control peptides [n = 14 neurons per group; t(26) = 3.941, P = 0.0005; Fig. 6D]. These results collectively suggest that post-synaptic α2δ-2-bound GluK1 receptors play a major role in regulating glutamatergic synaptic input to Purkinje cells.

GluK1 in Purkinje cells controls motor balance and performance

Our findings above indicate that GluK1 mediates glutamatergic transmission at parallel fibre–Purkinje cell synapses (Figs 2 and 6). Therefore, we investigated whether GluK1 in the cerebellum regulates motor functions. Our initial study involved microinjection of the selective GluK1 antagonist, UBP310 (50 nmol in 125 nl), or vehicle into the cerebellar vermis of mice. The cerebellar vermis is crucial for coordinating movements of the central body and proximal limbs.^57,58 The balance beam-walking test, a sensitive measure for assessing fine-motor coordination and balance,^38,39 was conducted 20 min after the injection of UBP310 or vehicle. Vehicle-injected mice navigated all square and round beams with ease. In contrast, mice injected with UBP310 exhibited difficulties in traversing these beams, with notably longer times (n = 10 mice for each group; Fig. 7A and B). Also, we evaluated motor learning and coordination using the rotarod test in vehicle-injected and UBP310-injected mice.¹⁶ Both groups of mice showed gradually increased fall latencies on the rotarod over the three consecutive trials (20 min intervals between trials). However, the increase in fall latencies in UBP310-injected mice was significantly attenuated compared with that in vehicle-injected mice (n = 10 mice for each group; Fig. 7C).

Figure 7.

GluK1 in cerebellar Purkinje cells controls motor balance and function. (A–C) Effect of microinjection of UBP310 or vehicle into the cerebellar vermis on the beam-walking (A and B) and rotarod (C) test trials (T1–T3) in mice (n = 10 mice per group). (D) Representative blotting images show the efficacy of three Grik1-specific guide RNAs (gRNAs) on GluK1 protein levels in HEK293 cells cotransfected with GluK1, Cas9 and Grik1 gRNAs. (E) Representative blotting images and quantification show the GluK1 protein levels in the cerebellum of Grik1-cKD and wild-type control (WT Ctrl) mice received microinjection of Grik1-specific gRNA into the cerebellar vermis (n = 6 mice per group). GAPDH was used as an internal control. (F–H) Effect of Grik1-cKD in Purkinje cells on beam walking (F and G) and rotarod (H) test trials (T1–T3) in mice (n = 8 mice in the WT control group and n = 6 mice in the Grik1-cKD group). Data are expressed as mean ± standard error of the mean. Student’s two-tailed t-test was used for the data in A, B and E. Two-way ANOVA followed by Bonferroni’s post hoc test was used for the data in C and H [*P < 0.05, **P < 0.01, ***P < 0.001 versus the corresponding control (T1); ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 versus the vehicle or WT Ctrl group at the same time point]. cKD = conditional knockdown.

Furthermore, to determine specifically whether GluK1 expressed in Purkinje cells plays a role in the control of motor coordination, we used the clustered, regularly interspaced, short palindromic repeats (CRISPR)/Cas9 gene editing approach to conditionally knock down Grik1 (Grik1-cKD) in Purkinje cells. We first crossed Pcp2^Cre/+ mice, which express Cre recombinase under the control of the mouse Purkinje cell protein 2 (Pcp2),^32,59 with Cas9 floxed-STOP mice to induce Cas9 expression in Purkinje cells. Cas9, an RNA-guided endonuclease derived from bacteria, uses base pairing to identify and cleave target DNAs that match the gRNA sequence. By subsequently injecting Grik1-specific gRNA into the cerebellums of Pcp2^Cre/+::Cas9^+/+ mice (mice in which Cas9 was specifically knocked in to target Purkinje cells), we aimed at knocking down GluK1 in Purkinje cells. Cas9^−/− mice, which lack Cas9, were used as the wild-type controls in our experiments. Initially, we screened and selected three effective Grik1-specific gRNAs for Grik1 knockout in HEK293 cells cotransfected with full-length mouse Grik1, lenti-Grik1 gRNA and lenti-Cas9 plasmids. Because the three Grik1-specific gRNAs had similar knockdown efficacy, we used gRNA#1 for the subsequent in vivo experiments (Fig. 7D). We microinjected the lentivirus expressing Grik1-specific gRNA into the cerebellar vermes (500 nl per injection; five injections in total) of Pcp2^Cre/+::Cas9^+/+ and Pcp2^Cre/+::Cas9^−/− mice. Immunoblotting assays showed that, 14 days after gRNA injection, GluK1 protein levels in the cerebellar vermis were significantly lower in the Pcp2^Cre/+::Cas9^+/+ mice compared with the Pcp2^Cre/+::Cas9^−/− mice [n = 6 mice per group, t(10) = 4.406, P = 0.0013; Fig. 7E]. Behavioural tests showed that Grik1-cKD markedly impaired beam-walking and rotarod performance (n = 8 mice in the control group and n = 6 mice in the Grik1-cKD group; Fig. 7F–H). These findings indicate that GluK1 receptors in Purkinje cells play a crucial role in the control of motor coordination.

α2δ-2-bound GluK1 receptors in cerebellar Purkinje cells mediate motor coordination

Next, we determined whether α2δ-2 expressed in Purkinje cells plays a role in motor coordination. We again used the CRISPR/Cas9 approach to conditionally knock down Cacna2d2 (Cacna2d2-cKD) in Purkinje cells of Pcp2^Cre/+::Cas9^+/+ mice. We screened and selected two effective, Cacna2d2-specific gRNAs for Cacna2d2 knockdown in HEK293 cells cotransfected with full-length mouse α2δ-2, lenti-Cacna2d2 gRNA and lenti-Cas9 plasmids. Because Cacna2d2-specific gRNA#1 and gRNA#2 had similar efficiency for α2δ-2 knockdown (Fig. 8A), we chose gRNA#1 for the following in vivo study. gRNA#3, targeting an intron of Cacna2d2, did not knock down α2δ-2 in HEK293 cells expressing mouse Cacna2d2 exonic sequences (Fig. 8A). Fourteen days after microinjection of Cacna2d2-specific gRNA (500 nl per injection, five injections in total) into the cerebellar vermes of Pcp2^Cre/+::Cas9^+/+ and Pcp2^Cre/+::Cas9^−/− control mice, immunoblotting assays showed that protein levels of α2δ-2 were significantly reduced in Cacna2d2-cKD mice compared with control mice [n = 6 mice per group; t(10) = 5.418, P = 0.0003; Fig. 8B]. In behavioural tests, control mice walked readily over all square and round beams. However, Cacna2d2-cKD mice struggled to traverse all types of beams, exhibiting significantly longer times (n = 10 mice per group; Fig. 8C and D). Furthermore, both groups of mice showed gradually increased fall latencies over all three trials during rotarod tests. However, compared with control mice, Cacna2d2-cKD mice displayed significantly shorter latencies to fall from the rotating rod (n = 10 mice per group; Fig. 8E). These data indicate that α2δ-2 in Purkinje cells is involved in the coordination of motor function and balance.

Figure 8.

α2δ-2-bound GluK1 receptors in the cerebellum mediate motor coordination. (A) Representative blotting images show the efficacy of three guide RNAs (gRNAs) on α2δ-2 protein levels in HEK293 cells cotransfected with α2δ-2, Cas9 and Cacna2d2 gRNAs. (B) Representative blotting images and quantification show α2δ-2 protein levels in the cerebellum of Cacna2d2-cKD and wild-type control (WT Ctrl) mice after microinjection of Cacna2d2-specific gRNA into the cerebellar vermis (n = 6 mice per group). GAPDH was used as an internal control. (C–E) Effect of Cacna2d2-cKD in Purkinje cells on beam-walking (C and D) and rotarod (E) test trials (T1–T3) in mice (n = 10 mice per group). (F–H) Effect of microinjection of control peptide or α2δ-2CT peptide into the cerebellar vermis on beam-walking (F and G) and rotarod (H) test trials (T1–T3) in mice (n = 14 mice in each group). Data are expressed as means ± SEM. Student’s two-tailed t-test was used for the data in B–D, F and G. Two-way ANOVA followed by Bonferroni’s post hoc test was used for the data in E and H [*P < 0.05, **P < 0.01, ***P < 0.001 versus corresponding control (T1); ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 versus the WT Ctrl or control peptide group at the same time point]. cKD = conditional knockdown.

Lastly, we determined whether α2δ-2-bound GluK1 receptors in the cerebellum are required for motor coordination. We microinjected the α2δ-2CT peptide or control peptide (400 pmol in 400 nl) into the vermes of the mouse cerebellum 20 min before starting the beam-walking and rotarod tests. Compared with the mice injected with the control peptide, the mice injected with the α2δ-2CT peptide exhibited significantly longer latencies in traversing both the square and round beams (n = 14 mice per group; Fig. 8F and G). During the rotarod test trials, both groups of mice displayed a gradual increase in the time spent on the rotating rod. However, the time spent on the rotating rod was significantly shorter for the α2δ-2CT peptide-injected mice than for the control peptide-injected mice in each trial (n = 13 mice per group; Fig. 8H). These results suggest that α2δ-2-bound GluK1 receptors in the cerebellum play a key role in the control of motor coordination.

Discussion

Our study demonstrates a previously unrecognized function of GluK1 receptors in Purkinje cells in the control of motor coordination. Purkinje cells receive excitatory synaptic inputs from climbing fibres and parallel fibres, in addition to inhibitory synaptic inputs from axons of both stellate cells and basket cells.^60-62 The parallel fibre input to Purkinje cells provides information crucial for the motor control of ongoing movements.⁶² Alterations of any synaptic input to Purkinje cells could cause motor incoordination. In this study, we found that blocking GluK1 receptors with UBP310 attenuated the amplitude of EPSCs at parallel fibre–Purkinje cell synapses. Additionally, puff application of ATPA, a GluK1 receptor agonist, onto Purkinje cells elicited a large current, suggesting the presence of post-synaptic GluK1-containing KARs in Purkinje cells. Our findings are consistent with those in a report showing that Grik1 knockout reduces the amplitude of EPSCs in Purkinje cells.³¹ Importantly, our study also demonstrated that microinjection of UBP310 into the cerebellum or conditional knockdown of GluK1 in Purkinje cells markedly impaired motor performance and learning in mice. Our findings collectively indicate that GluK1-containing KARs play a crucial role in motor coordination by their regulation of glutamatergic input to Purkinje cells.

Our study identifies, for the first time, a non-canonical function of α2δ-2, which governs the synaptic expression and activity of GluK1 receptors in Purkinje cells. GluK1, one of the two most important KAR subunits in the adult brain, has limited expression on the cell surface and synapse.^63-65 Although the signal peptide may interact with the N-terminal domain of GluK1 to inhibit surface expression of GluK1,^43,65 the molecular mechanisms governing the synaptic trafficking of GluK1 remain largely unknown. Unlike α2δ-1, which is mainly expressed in the excitatory neurons in the brain and spinal cord, α2δ-2 is predominantly expressed in the inhibitory neurons of these structures.^25,66,67 GABAergic Purkinje cells, the primary output pathway from the cerebellar cortex, abundantly express both α2δ-2 and GluK1 receptors.^25,68 Our use of a heterologous expression system in this study indicated that coexpression of α2δ-2, but not α2δ-1, greatly potentiated GluK1 currents and increased the cell surface expression of GluK1. Strikingly, pregabalin inhibited GluK1 currents only when α2δ-2 was coexpressed. Furthermore, we showed that inhibiting α2δ-2 with pregabalin in Cacna2d1 knockout mice not only attenuated the amplitude of EPSCs and ATPA-elicited currents in Purkinje cells but also reduced the synaptic expression of GluK1 proteins in the cerebellum. These findings indicate that α2δ-2 constitutively promotes or maintains synaptic expression of GluK1 in Purkinje cells. Our discoveries offer new insights into the molecular mechanisms regulating glutamatergic synaptic transmission at parallel fibre–Purkinje cell synapses, revealing a crucial role for α2δ-2 in this process.

A striking finding of our study is that α2δ-2 forms a protein complex with GluK1 receptors and that α2δ-2-bound GluK1 receptors mediate glutamatergic input to Purkinje cells, thereby controlling motor coordination. Our co-IP assays demonstrated that α2δ-2 interacted with GluK1 proteins both in cell lines and in the cerebellum. α2δ proteins associate with BK channels and VACC α1 subunits through their extracellular N-terminus region.^5,52,69 In contrast, using truncation and chimera approaches, we discovered that α2δ-2 interacted with GluK1 through its C-terminal domain. Based on this molecular information, we designed a Tat-fused peptide targeting the C-terminal domain of α2δ-2, which effectively disrupted the α2δ-2–GluK1 association. Full-length α2δ-2 is a highly glycosylated protein that promotes surface trafficking of its interacting protein.^54,55 α2δ-2CT peptide, which mimics the C-terminus of α2δ-2, can impede the interaction of full-length α2δ-2 proteins with GluK1. In this study, we demonstrated that α2δ-2CT peptide significantly reduced baseline EPSCs at parallel fibre–Purkinje cell synapses, suggesting that α2δ-2 is constitutively associated with GluK1 receptors in Purkinje cells in normal conditions. This convergent evidence strongly supports the notion that α2δ-2–GluK1 complexes tonically mediate post-synaptic depolarization and are responsible for carrying glutamatergic synaptic input to Purkinje cells. We also found that conditional knockdown of α2δ-2 in Purkinje cells reduced motor balance and coordination in mice. Importantly, microinjection of the α2δ-2CT peptide into the cerebellum similarly impaired motor learning and balance in mice. Therefore, our findings strongly suggest that the interaction with GluK1, rather than with VACCs, accounts for the crucial role of α2δ-2 in Purkinje cells in regulating motor coordination.

The discrepancy in GluK1 currents affected by α2δ-2 between HEK293 cells and Purkinje cells is likely to arise from the differing proportions of α2δ-2-bound and α2δ-2-free GluK1 receptors present in these cell types. Our data suggest that HEK293 cells used in the recordings express both α2δ-2-bound and α2δ-2-free GluK1 receptors on their surface. In contrast, treatment with pregabalin or the α2δ-2CT peptide abolished GluK1-mediated EPSCs at parallel fibre–Purkinje cell synapses, indicating that α2δ-2 is integral to GluK1 receptor activity at these synapses. Furthermore, it should be noted that not all GluK1 receptors in Purkinje cells are bound to α2δ-2, and pregabalin or the α2δ-2CT peptide specifically targets α2δ-2-bound GluK1 receptors at the synapses. Because puff application of ATPA activates both synaptic (α2δ-2-bound) and extrasynaptic (largely α2δ-2-free) GluK1 receptors, this could explain why treatment with pregabalin or the α2δ-2CT peptide only partly attenuated puff ATPA-elicited currents in Purkinje cells.

Our findings challenge the prevailing notion that α2δ-2 strictly regulates VACCs, which has been the focal point of most studies on α2δ-2.^5,20,21 Intriguingly, many γ proteins, initially regarded as subunits of VACCs, are now recognized as regulatory proteins of ionotropic glutamate receptors. For example, the γ2 (stargazin), γ7 and γ8 subunits, which are known as transmembrane AMPA receptor regulatory proteins, directly interact with and regulate AMPA receptors.^70-72 α2δ-1, through direct interactions, also facilitates synaptic expression of NMDA and calcium-permeable AMPA receptors in the spinal cord and brain.^6,9,10,73 The therapeutic actions of gabapentinoids in neuropathic pain predominantly involve the inhibition of α2δ-1-bound NMDA and calcium-permeable AMPA receptors in the spinal cord.^6,10,74,75 Notably, gabapentin has no effect on VACC activity or VACC-mediated neurotransmitter release in the brain or spinal cord.^4-8,16 In the present study, we observed that treatment with pregabalin or α2δ-2CT peptide did not reduce P/Q-type VACC currents, which represent the major subtype of VACCs in Purkinje cells.⁵⁶

Previous studies have shown that the trafficking and biophysical properties of GluK1 and GluK2 are regulated directly by neuropilin and tolloid-like (Neto) proteins.^63,76,77 In hippocampal neurons, Neto1 accelerates GluK1 desensitization, whereas Neto2 slows GluK1 desensitization and promotes synaptic expression of GluK1.^63,76 The loss of Neto1 reduces hippocampal synaptic GluK2 proteins by 50%.⁷⁸ In the cerebellum, where Neto2 is abundant, the loss of Neto2 decreases synaptic GluK2 proteins by ∼40%.⁷⁹ However, the function of Neto2-coupled GluK2 in the cerebellum has not been determined explicitly. Our identification of α2δ-2 as a unique, KAR-interacting protein sheds new light on the molecular composition, heterogeneity and synaptic function of native KARs. Because the coexpression of α2δ-2 and GluK1 is not limited to the cerebellum, our findings provide a crucial foundation for future research to explore the functional significance of α2δ-2-bound GluK1 receptors in other brain regions.

Our study also sheds light on the possible molecular mechanisms underlying the adverse effects of gabapentinoids, which frequently manifest as gait abnormalities and ataxia.^12,13,80 These drugs exhibit similar binding affinities for α2δ-1 and α2δ-2.^2,81 Although the role of α2δ-1 in the therapeutic effects of gabapentinoids on neuropathic pain and epilepsy is well documented,^3,47,53 the involvement of α2δ-2 in the various effects of gabapentinoids remains enigmatic. Mutations in Cacna2d2 are linked to cerebellar ataxia and significantly reduce gabapentin binding in cerebellar membranes.²⁰ Notably, the R217 mutation (R217A) in an RRR motif in α2δ-1, or the equivalent R282A mutation in mouse Cacna2d2 (equal to the R279A mutation in the human gene), diminishes gabapentin binding.^46,47,53 The R217A mutation in α2δ-1 reduces pregabalin binding in vivo, leading to the loss of its analgesic and antiepileptic efficacy.^47,53 In our study, we showed that α2δ-2 coexpression failed to potentiate GluK1 currents in the presence of α2δ-2CT peptide. Furthermore, pregabalin lost its inhibitory effect on GluK1 currents in cell lines expressing both GluK1 and the α2δ-2(R282A) mutant. These findings suggest that the impact of gabapentinoids on motor coordination is likely to stem from their ‘on-target’ inhibitory action on α2δ-2-bound GluK1 receptors. Although our focus was on α2δ-2-bound GluK1 receptors in the cerebellum, this receptor complex might also be involved in the adverse effect of gabapentinoids in other brain regions. In addition, Cacna2d2 mutations reduce VACC activity in Purkinje cells.^20,21 Therefore, although gait disorders caused by gabapentinoids might result primarily from the reduced activity of α2δ-2-bound GluK1 receptors, severe ataxia associated with Cacna2d2 mutation or conventional knockout is likely to involve the impaired activity of both VACCs and GluK1 receptors across multiple brain regions.

Conclusion

In summary, our study unveils a novel role of α2δ-2 in regulating synaptic GluK1-containing KAR activity in Purkinje cells. α2δ-2 interacts directly with GluK1 through its C-terminus, promoting or maintaining the synaptic expression of GluK1 receptors at parallel fibre–Purkinje cell synapses (Supplementary Fig. 1). This new information advances our mechanistic understanding of the physiological role of α2δ-2 in regulating GluK1-containing KARs and motor coordination. Therefore, both α2δ-1 and α2δ-2 proteins are pivotal in regulating glutamatergic synaptic transmission and plasticity through their direct interactions with distinct ionotropic glutamate receptors. Considering that gabapentinoids inhibit both α2δ-1 and α2δ-2, therapeutics (such as biologics or small molecules) that specifically target the α2δ-1 C-terminus could be beneficial for treating neuropathic pain and epilepsy while mitigating the motor function-related adverse effects associated with gabapentinoids. Alternatively, developing peripherally restricted gabapentinoids could help to avoid their CNS adverse effects.⁸²

Data availability

Data are available from the corresponding author upon reasonable request.

Funding

This study was supported by the National Institutes of Health (grants NS101880 and NS132398) and by the Pamela and Wayne Garrison Distinguished Chair Endowment (to H.-L.P.).

Competing interests

The authors report no competing interests.

Supplementary material

Supplementary material is available at Brain online.