Calcineurin and CK2 Reciprocally Regulate Synaptic AMPA Receptor Phenotypes via α2δ-1 in Spinal Excitatory Neurons

Abstract

Calcineurin inhibitors, such as cyclosporine and tacrolimus (FK506), are commonly used immunosuppressants for preserving transplanted organs and tissues. However, these drugs can cause severe and persistent pain. GluA2-lacking, calcium-permeable AMPA receptors (CP-AMPARs) are implicated in various neurological disorders, including neuropathic pain. It is unclear whether and how constitutive calcineurin, a Ca²⁺/calmodulin protein phosphatase, controls synaptic CP-AMPARs. In this study, we found that blocking CP-AMPARs with IEM-1460 markedly reduced the amplitude of AMPAR-EPSCs in excitatory neurons expressing vesicular glutamate transporter-2 (VGluT2), but not in inhibitory neurons expressing vesicular GABA transporter, in the spinal cord of FK506-treated male and female mice. FK506 treatment also caused an inward rectification in the current–voltage relationship of AMPAR-EPSCs specifically in VGluT2 neurons. Intrathecal injection of IEM-1460 rapidly alleviated pain hypersensitivity in FK506-treated mice. Furthermore, FK506 treatment substantially increased physical interaction of α2δ-1 with GluA1 and GluA2 in the spinal cord and reduced GluA1/GluA2 heteromers in endoplasmic reticulum-enriched fractions of spinal cords. Correspondingly, inhibiting α2δ-1 with pregabalin, Cacna2d1 genetic knock-out, or disrupting α2δ-1–AMPAR interactions with an α2δ-1 C terminus peptide reversed inward rectification of AMPAR-EPSCs in spinal VGluT2 neurons caused by FK506 treatment. In addition, CK2 inhibition reversed FK506 treatment–induced pain hypersensitivity, α2δ-1 interactions with GluA1 and GluA2, and inward rectification of AMPAR-EPSCs in spinal VGluT2 neurons. Thus, the increased prevalence of synaptic CP-AMPARs in spinal excitatory neurons plays a major role in calcineurin inhibitor-induced pain hypersensitivity. Calcineurin and CK2 antagonistically regulate postsynaptic CP-AMPARs through α2δ-1—mediated GluA1/GluA2 heteromeric assembly in the spinal dorsal horn.

Significance Statement

Clinically used calcineurin inhibitors can cause severe pain, known as calcineurin inhibitor—induced pain syndrome (CIPS). However, its underlying mechanisms remain elusive. This study shows for the first time that calcineurin inhibition caused cell type—specific expression of synaptic Ca²⁺-permeable AMPARs in spinal cord excitatory neurons. Blocking spinal Ca²⁺-permeable AMPARs reduced CIPS. Calcineurin inhibition potentiated the α2δ-1 (previously known as a calcium channel subunit) interaction with GluA1 and GluA2 subunits, disrupting their intracellular assembly in the spinal cord. Additionally, inhibiting spinal CK2 diminished α2δ-1–AMPAR interactions and synaptic Ca²⁺-permeable AMPARs augmented by calcineurin inhibitors. Thus, calcineurin and CK2 dynamically control AMPAR phenotypes in spinal excitatory neurons through α2δ-1—mediated GluA1/GluA2 assembly. Targeting α2δ-1 and CK2 are effective strategies for treating CIPS.

Introduction

Calcineurin is a Ca²⁺/calmodulin protein phosphatase, which is critically involved in immune responses and T-cell activation. Clinically used calcineurin inhibitors, such as tacrolimus (FK506) and cyclosporine A, are highly effective for preserving transplanted organs and tissues and for treating autoimmune diseases. However, extended use of these drugs can cause severe and unexplained pain in patients, often referred to as calcineurin inhibitor-induced pain syndrome (CIPS; Collini et al., 2006; Noda et al., 2008; Kakihana et al., 2012). In addition to its expression in immune T cells, calcineurin is highly expressed in the nervous system, including the dorsal root ganglion and spinal dorsal horn (Strack et al., 1996; Wu et al., 2005). Postsynaptic AMPA receptors (AMPARs), composed of homo- or heterotetramers formed by GluA1–GluA4 subunits (Hollmann and Heinemann, 1994; Mayer, 2006), are predominantly involved in fast excitatory synaptic transmission from primary sensory neurons to spinal dorsal horn neurons. The GluA2 subunit has the foremost influence on AMPAR biophysics, with AMPARs containing the Q/R edited GluA2 being impermeable to Ca²⁺ and displaying a linear current–voltage (I–V) relationship. In contrast, AMPARs lacking the Q/R edited GluA2 are Ca²⁺-permeable AMPARs (CP-AMPARs) and exhibit an inwardly rectifying I–V relationship (Bowie and Mayer, 1995). Traumatic nerve injury reduces calcineurin activity in the spinal cord (Miletic et al., 2015), and a sustained increase in postsynaptic CP-AMPARs of spinal dorsal horn neurons maintains chronic neuropathic pain (Chen et al., 2013; Chen et al., 2019). Calcineurin activity actively controls phosphorylation of both NMDARs and AMPARs (Nishi et al., 2002; Sanderson et al., 2012; Miletic et al., 2015; Zhou et al., 2022a). Although calcineurin inhibition leads to augmented synaptic NMDAR activity in spinal dorsal horn neurons (Chen et al., 2014a; Huang et al., 2022a), whether calcineurin constitutively controls postsynaptic AMPAR plasticity in the spinal dorsal horn is largely unknown.

Distinct populations of excitatory and inhibitory neurons orchestrate nociceptive processing in the spinal cord. Spinal dorsal horn neurons can be broadly classified into two groups: vesicular glutamate transporter-2-expressing (VGluT2) excitatory neurons and vesicular GABA transporter-expressing (VGAT; also termed as vesicular inhibitory amino acid transporter, VIAAT) inhibitory neurons (Todd, 2017; Browne et al., 2020). VGluT2-expressing excitatory dorsal horn neurons relay persistent pain caused by tissue inflammation and nerve injury (Wang et al., 2018), whereas VGAT-expressing inhibitory neurons normally inhibit nociceptive transmission in the dorsal horn (Koga et al., 2017). NMDAR-mediated synaptic plasticity predominantly occurs in spinal VGluT2 neurons in the spinal cord. For example, nerve injury or theta-burst stimulation of primary afferent nerves preferentially induces synaptic NMDAR hyperactivity in spinal VGluT2 neurons to cause persistent pain (Huang et al., 2022b, 2023a). Also, opioid-induced hyperalgesia is associated with augmented synaptic NMDAR activity at primary afferent-excitatory neuron synapses in the spinal cord (Chen et al., 2022). However, how calcineurin controls synaptic AMPAR phenotypes in spinal excitatory and inhibitory neurons remains unknown.

GluA1 and GluA2 subunits in the endoplasmic reticulum (ER) normally have a propensity to form GluA1/GluA2 heterotetramers that promptly traffic to synaptic sites (Mansour et al., 2001; Greger et al., 2002). α2δ-1, previously known as a voltage-gated Ca²⁺ channel subunit, can directly interact with GluA1 and GluA2 subunits through its C terminus (Li et al., 2021; Zhou et al., 2022b). Accordingly, α2δ-1 favors the assembly of GluA1 homotetramers by selectively inhibiting GluA1/GluA2 heteromeric assembly, thereby promoting synaptic CP-AMPARs in the spinal cord in nerve injury-induced neuropathic pain (Li et al., 2021). Because calcineurin actively controls phosphorylation of GluA1 and GluA2 (Ahn and Choe, 2010; Sanderson et al., 2012; Zhou et al., 2024), impaired calcineurin activity may potentiate the interaction of α2δ-1 with phosphorylated GluA1 and GluA2 to impede their subunit assembly and availability of GluA1/GluA2 heterotetramers for synaptic expression. Therefore, in the present study, we investigated whether and how calcineurin controls synaptic AMPAR plasticity in genetically labeled excitatory and inhibitory neurons in the spinal dorsal horn. Our study reveals that calcineurin activity constitutively controls synaptic incorporation of CP-AMPARs in spinal excitatory neurons through regulating α2δ-1–AMPAR interactions and intracellular assembly and availability of GluA1/GluA2 heterotetramers. Additionally, casein kinase 2 (CK2) and calcineurin act antagonistically as key molecular switches to control synaptic AMPAR phenotypes by modulating the α2δ-1 interaction with GluA1 and GluA2. This new information extends our mechanistic understanding of molecular determinants governing the subunit composition of synaptic AMPARs in CIPS and suggests effective strategies for treating this painful condition.

Materials and Methods

Animal models

All experimental procedures were approved by the Institutional Animal Care and Use Committee at the University of Texas MD Anderson Cancer Center and followed the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Male and female mice (10–14 weeks old) were used in this study, and the genetic background of mice was C57BL/6. Mice were kept at no more than five per cage and had ad libitum access to food and water. The animal housing facility was maintained at 24°C under a 12 h light/dark cycle. CIPS was induced by intraperitoneal injection of FK506 (#3631, Tocris Bioscience; dissolved in dimethyl sulfoxide) at a dose of 3 mg/kg per day for 7 consecutive days, as we reported previously (Chen et al., 2014a; Huang et al., 2020). In this animal model of CIPS, pain hypersensitivity develops within 3 d after starting FK506 treatment and persists for at least another 10 d after discontinuing FK506 treatment (Chen et al., 2014a; Hu et al., 2014). Animals receiving the same volume of vehicle were used as the control group. All behavioral data were collected 3 d after cessation of FK506 treatment.

Cacna2d1^−/− knock-out (KO) mice (Fuller-Bicer et al., 2009) and their wild-type (WT) littermates were obtained by breeding Cacna2d1^+/− heterozygous mice (#6900, Medical Research Council). VGluT2-ires-Cre knock-in mice (#028863), VGAT-ires-Cre knock-in mice (#028862), and tdTomato-floxed mice (#007909) with C57BL/6 genetic background were obtained from The Jackson Laboratory. The VGluT2^Cre/+:tdTomato^flox/flox and VGAT^Cre/+:tdTomato^flox/flox mice were produced by crossing male VGluT2-ires-Cre or VGAT-ires-Cre mice with female tdTomato-floxed mice, respectively (Chen et al., 2022; Huang et al., 2022b). Mouse genotypes were confirmed through genotyping using ear biopsies. The specificity of tdTomato-labeled VGluT2 and VGAT neurons in the spinal dorsal horn has been validated previously (Wang et al., 2018; Browne et al., 2020; Chen et al., 2022).

Electrophysiological recordings in spinal cord slices

Animals were anesthetized with 3% isoflurane, and the lumbar spinal cords were quickly removed via laminectomy. Transverse slices (400 µm thick) of spinal cords were cut using a vibratome (#VT1000s, Leica) and immersed with sucrose-modified artificial cerebrospinal fluid saturated with 95% O₂ and 5% CO₂. The artificial cerebrospinal fluid comprised the following (in mM): 234 sucrose, 26 NaHCO₃, 3.6 KCl, 2.5 CaCl₂, 1.2 MgCl₂, 1.2 NaH₂PO₄, and 25 glucose. The slices were then placed in Krebs’ solution containing the following (in mM): 117 NaCl, 25 NaHCO₃, 3.6 KCl, 2.5 CaCl₂, 1.2 MgCl₂, 1.2 NaH₂PO₄, and 11 glucose. All slices were kept in the continuously oxygenated incubation chamber for at least 1 h at 34°C before being used for recordings.

The spinal cord slices were transferred into a recording chamber and continuously perfused with oxygenated Krebs’ solution at 3 ml/min at 34°C. tdTomato-tagged neurons in the lamina II region were identified with an upright microscope (#BX50 WI, Olympus Optical) equipped with epifluorescence illumination and differential interference contrast optics. Glass recording electrodes (5–8 MΩ) were filled with an internal solution containing the following (in mM): 135 potassium gluconate, 5 KCl, 2 MgCl₂, 0.5 CaCl₂, 5 ATP-Mg, 0.5 Na₂-GTP, 5 EGTA, 5 HEPES, and 10 lidocaine N-ethyl bromide, pH 7.3 (QX314; 280‒300 mOsm). QX314 was used to block voltage-gated Na⁺ channels to suppress action potentials in the postsynaptic neurons. To determine the effect of IEM-1460 on synaptic AMPARs, excitatory synaptic currents (EPSCs) were recorded at a holding potential of −60 mV. The NMDAR antagonist was not used, because NMDARs are minimally open at the holding potential of −60 mV in the presence of 1.2 mM Mg²⁺. Spinal lamina II neurons predominantly receive input from unmyelinated primary afferent fibers (Nakatsuka et al., 2000; Pan et al., 2003). To elicit the release of glutamate from primary afferents, EPSCs of labeled neurons were evoked using a bipolar tungsten electrode via electrical stimulation (0.6 mA, 0.5 ms, and 0.1 Hz) of the ipsilateral dorsal root, which can activate both A and C fibers (Kohno et al., 1999; Li et al., 2002; Zhou et al., 2010). Monosynaptic EPSCs were discerned based on their consistent latency and the absence of conduction failure upon 20 Hz stimulation (Li et al., 2002; Zhou et al., 2012). Blocking AMPARs with 6-cyano-7-nitroquinoxaline-2,3-dione abolishes evoked EPSCs of lamina II neurons without affecting the stimulating artifact (Li et al., 2002; Pan et al., 2002; Pan and Pan, 2004).

To determine the I–V relationship, AMPAR-mediated excitatory postsynaptic currents (AMPAR-EPSCs) were recorded at holding potentials ranging from −70 to +70 mV in 20 mV steps. This was done in the presence of 20 µM (−)-bicuculline methochloride, 1 µM strychnine, and 50 µM dʟ-2-amino-5-phosphonopentanoic acid (AP5) to block GABA_A receptor-, glycine receptor-, and NMDAR-mediated currents, respectively. The pipette solution contained the following (in mM): 145 CsCl, 2.5 NaCl, 4 MgATP, 1 EGTA, 10 HEPES, and 0.1 spermine tetrahydrochloride, pH 7.3 (osmolarity, 300 mOsm). Spermine was included in the intracellular solution to compensate the loss of endogenous polyamine due to intracellular dialysis (Chen et al., 2018). All signals were filtered at 1–2 kHz, processed through a MultiClamp 700B amplifier (Molecular Devices) and digitized at 20 kHz using Digidata 1550B (Molecular Devices).

AP5 (#HB0252) was purchased from Hello Bio. Strychnine (#S0532) and 5,6-dichloro-1-β-d-ribofuranosylbenzimidazole (DRB, #D1916) were obtained from Sigma-Aldrich. Bicuculline (#27802), IEM-1460 (#15623), and pregabalin (#13663) were acquired from Cayman Chemical. The α2δ-1 C-terminal mimicking peptide (VSGLNPSLWSIFGLQFILLWLVSGSRHYLW) and a scrambled control peptide (FGLGWQPWSLSFYLVWSGLILSVLHLIRSN), both fused to the cell-penetrating peptide Tat domain (YGRKKRRQRRR), were synthesized by Synpeptide. All chemicals were dissolved in stock solutions and freshly diluted with Krebs’ solution before recording and delivered at the final concentrations via syringe pumps.

Nociceptive behavioral assessment

To assess tactile allodynia, individual mice were placed in separate chambers with a mesh floor and allowed to acclimate for 30 min. A set of calibrated von Frey filaments (Stoelting) was then vertically applied to the plantar surface of the hindpaw with enough pressure to cause filament bending for 6 s. A response was considered positive if the mouse exhibited a quick withdrawal or flinching of the paw. When a response was observed, the next filament with a slightly lower force was used. If there was no response, the next filament with a slightly higher force was employed. The tactile stimulus strength that produced a 50% probability of paw withdrawal was determined following the “up-down” method (Chaplan et al., 1994; Chen et al., 2001).

To assess mechanical nociception, we measured the withdrawal threshold in response to a noxious pressure stimulus following procedures described in prior studies (Chen et al., 2009; Luo et al., 2020). The Analgesy-Meter device (Ugo Basile) was employed, and a pressure stimulus was applied to the hindpaw by activating a motor that gradually increased the force. As soon as a withdrawal response from the animal was observed, the device was immediately stopped, and the nociceptive threshold was documented.

We evaluated thermal sensitivity using a radiant heat source (IITC Life Science), as we described previously (Sun et al., 2019; Jin et al., 2022). Mice were placed inside a plastic cylinder on a glass plate preheated to 30°C, allowing them to acclimate for 30 min. A radiant light source emitting noxious heat was directed toward the plantar surface of the hindpaw. As soon as the hindpaw rapidly moved away from the heat source, the withdrawal latency was recorded with a timer.

Spinal cord synaptosome and ER preparation

Mice were anesthetized with 3% isoflurane, and the dorsal spinal cords at lumbar L3–L5 levels were rapidly collected via laminectomy. Total tissue proteins were extracted using established methods as described previously (Huang et al., 2020; Zhang et al., 2021). To isolate synaptosomes, we homogenized dorsal spinal cord tissues using ice-cold HEPES-buffered sucrose solution (0.32 mol/L sucrose, 1 mmol/L EGTA, and 4 mmol/L HEPES, pH 7.4) containing a cocktail of protease/phosphatase inhibitors (#78442, Thermo Fisher Scientific). The homogenates were then centrifuged at 2,000 × g at 4°C for 10 min to remove nuclei and large debris, and the supernatant was further centrifuged at 20,000 × g for 30 min to obtain crude synaptosomes. The synaptosomal pellets were subjected to lysis via hypoosmotic shock in ice-cold HEPES buffer containing protease/phosphatase inhibitors for 30 min. The lysates were centrifuged at 25,000 × g at 4°C for 45 min to obtain the synaptosomal fraction (Huang et al., 2020; Zhang et al., 2021). The final protein concentration was determined using a bicinchoninic acid protein assay kit (#5000001, Bio-Rad).

ER-enriched fractions were isolated from the dorsal spinal cord tissues using an ER isolation kit (#ERO100-1KT, Millipore Sigma) as previously described (Li et al., 2021). Briefly, spinal cord tissues (pooled from three mice per sample) were homogenized using an overhead motor (∼200 rpm) in 10-fold 1× Isotonic Extraction Buffer (in mM: 10 HEPES, 250 sucrose, 1 EGTA, and 25 KCl, pH 7.8). The homogenates were centrifuged at 1,000 × g at 4°C for 10 min to remove lipids, nuclei, and large debris. Then the supernatant was recentrifuged at 12,000 × g at 4°C for 15 min. The postmitochondrial fraction was further centrifuged at 100,000 × g at 4°C for 60 min to obtain the pellet that contained ER-enriched fractions.

Coimmunoprecipitation and immunoblotting

Equal amounts of protein samples were incubated with protein G agarose beads (#16-266, EMD Millipore) prebound to the rabbit anti-GluA1 antibody (#AB1504, Millipore) or rabbit anti-α2δ-1 antibody (#ACC-015, Alomone Labs) at 4°C overnight. Protein G beads prebound to rabbit IgG were used as controls. Protein G beads were then washed three times, isolated using sample loading buffer, and were separated by SDS-PAGE with 4–15% gel (#456-1086; Bio-Rad) before being transferred to a PVDF membrane. For immunoblotting, the mouse anti-GluA1 antibody (#N355/1, NeuroMab), mouse anti-GluA2 antibody (#L21/32, NeuroMab), mouse anti-α2δ-1 antibody (#sc271697, Santa Cruz Biotechnology), rabbit anti-GAPDH antibody (#5174, Cell Signaling Technology), and rabbit anti-calreticulin antibody (#12238S, Cell Signaling Technology) were used as primary antibodies (Li et al., 2021; Zhou et al., 2022b). The blotting membranes were incubated with an HRP-conjugated anti-mouse IgG (#7076S, Cell Signaling Technology) or anti-rabbit IgG (#7074S, Cell Signaling Technology), which served as the secondary antibody. The protein bands were visualized using an ECL kit (#34580, Thermo Fisher Scientific) and quantified by an Odyssey Fc Imager (LI-COR Biosciences). For quantification, the input protein bands of GluA1 and GluA2 in isolated ER fractions were normalized to the calreticulin protein band, whereas the input protein bands of GluA1 and GluA2 in tissue lysate samples were normalized to the GAPDH protein band on the same blots. The protein bands of GluA1 and GluA2 in the immunoprecipitation (IP) blots were normalized to those of either GluA1 or α2δ-1 on the same IP blots. The mean values of the band density in vehicle-treated samples were set as 1.

Study design and statistical analysis

Data were presented as the mean ± standard error of the mean (SEM). The investigators were blinded to the mouse genotype and treatment for all experiments. Because we observed no sex differences in the electrophysiological and biochemical assays in this study, data from male and female mice were pooled for statistical analysis. The sample sizes were similar to those in previously published studies (Chen et al., 2014c; Sun et al., 2019; Huang et al., 2020; Li et al., 2021). In electrophysiology experiments, only one neuron was recorded from each spinal cord slice, and at least three mice were used for each recording protocol. The peak amplitudes of evoked EPSCs were quantified by averaging six consecutive traces using pClamp software (Molecular Devices). The rectification index was determined as the ratio of AMPAR-EPSCs amplitude at the holding potential of +50 to −50 mV (Li et al., 2012; Chen et al., 2013, 2019). Series resistance was monitored, and data were excluded if it changed by >15% during the recording. In the AMPAR-EPSC recording dataset, a total of five cells were excluded from the analysis due to poor space clamping. Normal distribution for each dataset was assessed using Shapiro–Wilk test. We used a two-tailed Student's t test for comparisons between two groups. To assess differences among three or more groups, we used one-way or two-way ANOVA with Tukey's or Šidák's post hoc test. Kruskal–Wallis test was used for data that are not normally distributed. All data analyses were performed using Prism (version 10, GraphPad software), and a p value <0.05 was considered statistically significant.

Results

Systemic treatment with FK506 preferentially increases the prevalence of synaptic CP-AMPARs in spinal VGluT2 neurons

Synaptic AMPARs in the spinal cord predominantly consist of GluA1 and GluA2 subunits (Isaac et al., 2007; Chen et al., 2013, 2019). Phosphorylation plays a critical role in the dynamic regulation of synaptic AMPAR composition, and calcineurin actively controls phosphorylation of GluA1 and GluA2 in the brain and spinal dorsal horn (Sanderson et al., 2012; Miletic et al., 2015). We first determined whether systemic administration of FK506, a clinically used specific calcineurin inhibitor, affects the prevalence of synaptic CP-AMPARs in genetically labeled excitatory and inhibitory neurons in the spinal dorsal horn. CIPS was induced by treating mice with intraperitoneal injection of FK506 at a dose of 3 mg/kg per day for 7 consecutive days (Chen et al., 2014a; Huang et al., 2020). Whole-cell voltage-clamp recordings in spinal cord slices revealed that the baseline amplitude of monosynaptically evoked EPSCs in tdTomato-labeled VGluT2 neurons was significantly higher in FK506-treated mice than in vehicle-treated mice (p = 0.0113; F_(3,49)= 2.770; n = 12 neurons in the vehicle group; n = 13 neurons in the FK506 group; Fig. 1A). In FK506-treated animals, an increase in the baseline amplitude of AMPAR-EPSCs in spinal lamina II neurons results from augmented activity of presynaptic NMDARs at primary afferent central terminals (Chen et al., 2014a; Huang et al., 2020). IEM-1460 is a specific open-channel blocker of CP-AMPARs (Twomey et al., 2018). Bath application of 100 µM IEM-1460 for 10 min caused ∼20% reduction in the amplitude of evoked EPSCs in VGluT2 neurons in FK506-treated, but not in vehicle-treated, mice (Fig. 1A).

Figure 1.

FK506 treatment increases the prevalence of synaptic CP-AMPARs in spinal VGluT2 neurons. A, Representative recording traces and quantification show the effect of bath application of 100 µM IEM-1460 on the amplitude of EPSCs evoked monosynaptically from the dorsal root in lamina II VGluT2 neurons in vehicle-treated and FK506-treated mice (n = 12 neurons from 3 mice in the vehicle group; n = 13 neurons from 3 mice in the FK506 group). B, Original recording traces and quantification show the effect of bath application of IEM-1460 on the amplitude of monosynaptically evoked EPSCs in lamina II VGAT neurons in vehicle-treated and FK506-treated mice (n = 14 neurons from 3 mice per group). Data are expressed as mean ± SEM. ^#p < 0.05, ^##p < 0.01 compared with the vehicle-treated groups at the same time point; *p < 0.05 compared with the baseline within the same group (two-way ANOVA followed by Šidák's post hoc test).

In contrast, the baseline amplitude of evoked EPSCs in tdTomato-labeled VGAT neurons in spinal lamina II did not differ significantly between FK506-treated and vehicle-treated mice (n = 14 neurons per group; Fig. 1B). Furthermore, bath application of IEM-1460 had no significant effect on the amplitude of evoked EPSCs in FK506-treated or vehicle-treated mice (Fig. 1B). These results suggest that constitutive calcineurin activity preferentially restrains synaptic expression of CP-AMPARs in spinal excitatory neurons.

Calcineurin inhibition causes cell type-specific expression of postsynaptic CP-AMPARs in spinal VGluT2 neurons

Under normal conditions, the vast majority AMPARs in the spinal dorsal horn synapses are GluA2-containing, Ca²⁺-impermeable AMPARs composed of GluA1/GluA2 heterodimers (Li et al., 2012; Chen et al., 2013). CP-AMPARs display distinct inward rectification at positive holding potentials (Bowie and Mayer, 1995). To further determine the effect of calcineurin inhibition on synaptic CP-AMPARs in spinal excitatory and inhibitory neurons, we treated spinal cord slices with vehicle or 5 µM FK506 for 30 min and then examined the I–V relationship of AMPAR-EPSCs. In all experiments examining the I–V relationship of AMPAR-EPSCs, we bath applied 50 µM AP5 throughout the recording period so that NMDARs are not activated at positive holding potentials. In vehicle-treated slices, whole-cell recordings in spinal cord slices revealed a linear I–V relationship of AMPAR-EPSCs from −70 to +70 mV in tdTomato-labeled VGluT2 neurons in lamina II. However, in FK506-treated slices, AMPAR-EPSCs in VGluT2 neurons exhibited characteristic inward rectification at positive holding potentials from +30 to +70 mV (Fig. 2A,B). The rectification index of AMPAR-EPSCs (I_{+50 mV}/I_{−50 mV}) was significantly lower in FK506-treated VGluT2 neurons than that in vehicle-treated VGluT2 neurons (F_(3,45)= 2.831; p = 0.0492; n = 12 neurons in the vehicle group; n = 13 neurons in the FK506-treated group; Fig. 2E).

Figure 2.

FK506 treatment potentiates postsynaptic CP-AMPARs in spinal VGluT2 neurons. A, B, Original recording traces (A) and quantification (B) show the effect of FK506 treatment on the current–voltage relationship of AMPAR-EPSCs in lamina II VGluT2 neurons (n = 12 neurons from 3 mice in the vehicle group; n = 13 neurons from 3 mice in the FK506 group). C, D, Representative recording traces (C) and quantification (D) show the effect of FK506 treatment on the current–voltage (I–V) relationship (B) of AMPAR-EPSCs in lamina II VGAT neurons (n = 12 neurons from 3 mice per group). E, Rectification index (I_{+50 mV}/I_{−50 mV}) of AMPAR-EPSCs in lamina II VGluT2 and VGAT neurons from mouse spinal cord slices treated with vehicle or 5 µM FK506. Data are shown as mean ± SEM. *p < 0.05 (two-way ANOVA followed by Tukey's post hoc test).

In contrast, the I–V relationship of AMPAR-EPSCs in tdTomato-labeled VGAT neurons in lamina II remained linear from −70 to +70 mV in both vehicle-treated and FK506-treated slices (n = 12 neurons per group, Fig. 2C,D). The rectification index (I_{+50 mV}/I_{−50 mV}) of AMPAR-EPSCs in VGAT neurons showed no significant differences between vehicle-treated and FK506-treated slices (Fig. 2E). These findings indicate that calcineurin inhibition increases the prevalence of postsynaptic CP-AMPARs predominantly in excitatory dorsal horn neurons.

Inhibiting CP-AMPARs at the spinal cord level attenuates calcineurin inhibitor-induced pain hypersensitivity

Increased synaptic NMDAR activity in the spinal dorsal horn contributes to the development of CIPS (Chen et al., 2014a; Huang et al., 2020). To determine the functional significance of synaptic CP-AMPARs in the spinal cord in calcineurin inhibitor-induced pain hypersensitivity, we measured nociceptive withdrawal thresholds in mice treated systemically with FK506 (3 mg/kg per day) or vehicle for 7 d. As reported previously (Chen et al., 2014a), prolonged treatment with FK506 caused a profound reduction in the baseline withdrawal thresholds in response to tactile, pressure, and heat stimuli applied to the hindpaw (n = 8 mice per group; Fig. 3). Intrathecal injection of 5 µg IEM-1460 did not change significantly the tactile, pressure, or heat withdrawal thresholds in vehicle-treated mice. In contrast, treatment with IEM-1460 markedly increased tactile, pressure, and heat withdrawal thresholds in FK506-treated mice (Fig. 3). These in vivo data suggest that increased CP-AMPAR activity in the spinal cord critically maintains pain hypersensitivity caused by the calcineurin inhibitor.

Figure 3.

Inhibiting CP-AMPARs at the spinal cord level attenuates pain hypersensitivity caused by FK506 treatment. A–C, Time course of the paw withdrawal thresholds in response to von Frey filaments (A), pressure (B), and noxious heat (C) stimuli in wild-type mice (n = 8 mice per group) after intrathecal administration of 5 µg IEM-1460 in mice treated systemically with vehicle or FK506 (3 mg/kg per day) for 7 d. Data are expressed as mean ± SEM (n = 8 mice per group). *p < 0.05, **p < 0.01 compared with the baseline within the same group (two-way ANOVA followed by Tukey's post hoc test).

Calcineurin inhibition diminishes GluA1/GluA2 heteromers in the spinal cord

We subsequently attempted to define molecular determinants responsible for the calcineurin inhibitor-induced switch to synaptic CP-AMPARs in the spinal dorsal horn. GluA1/GluA2 heteromers are predominantly present in the spinal dorsal horn, and the synaptic AMPAR composition depends dynamically on the availability of AMPARs assembled intracellularly (Greger et al., 2002). It is possible that calcineurin inhibition promotes synaptic CP-AMPARs in the spinal cord by inhibiting the formation of GluA1/GluA2 heterotetramers. To test this hypothesis, we performed coimmunoprecipitation (co-IP) analysis to measure the protein levels of GluA1/GluA2 complexes in dorsal spinal cord tissues of mice systemically treated with FK506 or vehicle for 7 d. The anti-GluA1 antibody, as opposed to an irrelevant IgG, precipitated GluA2 proteins in the dorsal spinal cord (Fig. 4A). Immunoblotting analysis showed that the GluA1 and GluA2 protein levels in spinal cord lysates were similar between vehicle-treated and FK506-treated mice (n = 8 mice per group; Fig. 4A,B). However, co-IP analyses revealed that the protein levels of GluA1/GluA2 complexes in dorsal spinal cord tissues were significantly lower in FK506-treated mice than those in vehicle-treated mice (n = 8 mice per group; p = 0.0046; t₍₁₄₎= 3.366; Fig. 4A,C). These results suggest that calcineurin inhibition reduces GluA1/GluA2 heterotetramers in the spinal cord.

Figure 4.

Systemic administration of FK506 decreases GluA1/GluA2 heteromers in the spinal cord. A–C, Representative gel images (A) and quantification show the protein levels of GluA1 and GluA2 (B) and GluA1/GluA2 complexes (C) in dorsal spinal cord tissues from mice treated systemically with vehicle or FK506 (3 mg/kg per day) for 7 d (n = 8 mice per group). The input protein bands of GluA1 and GluA2 were normalized to those of the vehicle group on the same blots. D–F, Original gel images (D) and quantification show the protein levels of GluA1 and GluA2 (E) and GluA1/GluA2 complexes (F) in ER-enriched fractions of dorsal spinal cord from mice treated systemically with vehicle or FK506 for 7 d (n = 8 samples per group, each sample contained dorsal spinal cord tissues from 3 mice). Proteins were first immunoprecipitated (IP) with an anti-GluA1 antibody, and immunoblotting was performed using anti-GluA1, anti-GluA2, and anti-calreticulin antibodies. Calreticulin was used as an ER marker. Veh, vehicle; FK, FK506. Data are shown as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed Student's t test).

Calcineurin inhibition impairs GluA1/GluA2 heteromeric assembly in the ER of spinal cords

Synaptic AMPAR composition depends on the availability of AMPARs assembled in the ER (Greger et al., 2002; Li et al., 2021). GluA1 and GluA2 dimers favorably assemble to GluA1/GluA2 heterotetramers in the ER before trafficking to synapses (Mansour et al., 2001; Greger et al., 2002). Thus, we determined whether calcineurin inhibition affects GluA1/GluA2 heteromeric assembly by measuring GluA1/GluA2 complexes in the ER fraction of the dorsal spinal cord. Immunoblotting analysis of ER-enriched fractions from the spinal cord showed that GluA1 protein levels were similar between vehicle-treated and FK506-treated mice (n = 8 samples per group; each sample contained spinal cord tissues from three mice; Fig. 4D,E). However, the protein level of GluA2 in ER-enriched fractions was significantly higher in FK506-treated mice than in vehicle-treated mice (n = 8 samples per group; each sample contained spinal cord tissues from 3 mice; t₍₁₄₎= 2.754; p = 0.0155; Fig. 4D,E). Co-IP assays showed that the GluA1 antibody, but not an irrelevant lgG, pulled down GluA2 in ER-enriched fractions from the spinal cord. Strikingly, treatment with FK506 significantly reduced protein levels of GluA1/GluA2 complexes in ER-enriched fractions compared with those in vehicle-treated mice (n = 8 samples per group; each sample containing spinal cord tissues from three mice; t₍₁₄₎= 6.239; p < 0.0001; Fig. 4D,F). These data suggest that calcineurin inhibition reduces the ability of GluA1 and GluA2 to form heterotetramers and consequently results in GluA2 retention in the ER of the spinal cord.

Calcineurin inhibition potentiates physical interaction of α2δ-1 with GluA1 and GluA2 in the spinal cord

In neuropathic pain induced by traumatic nerve injury, α2δ-1 can directly interact with GluA1 and GluA2, disrupting GluA1/GluA2 heteromeric assembly and promoting the synaptic incorporation of GluA1 homotetramers (CP-AMPARs) in the spinal cord (Li et al., 2021). α2δ-1 is a phospho-binding protein and prefers to interact with phosphorylated proteins (Zhou et al., 2021). Because calcineurin controls the phosphorylation of GluA1 and GluA2 subunits (Ahn and Choe, 2010; Sanderson et al., 2012), we next used co-IP assays to determine whether calcineurin inhibition potentiates α2δ-1–AMPAR interactions in the dorsal spinal cord. An anti-α2δ-1 antibody precipitated both GluA1 and GluA2 proteins in dorsal spinal cord synaptosomes. Immunoblotting analysis of synaptosomes showed that protein levels of GluA1, GluA2, and α2δ-1 were similar between vehicle-treated and FK506-treated mice (n = 8 mice per group; Fig. 5A–C). Systemic treatment with FK506 significantly increased protein levels of both α2δ-1/GluA1 (p = 0.0012; t₍₁₄₎= 4.063; Fig. 5A,D) and α2δ-1/GluA2 protein complexes (n = 8 mice per group; p = 0.0010; t₍₁₄₎= 4.142; n = 8 mice per group; Fig. 5A,D) in spinal cord synaptosomes compared with the vehicle-treated group. These findings suggest that constitutive calcineurin actively controls the α2δ-1 interaction with GluA1 and GluA2 subunits in the spinal cord.

Figure 5.

Calcineurin inhibition potentiates α2δ-1 interactions with GluA1 and GluA2 in the spinal cord. A, C, D, Original gel images (A) and quantification show the protein levels of GluA1, GluA2, and α2δ-1 (C) and α2δ-1–GluA1 and α2δ-1–GluA2 complexes (D) in dorsal spinal cord synaptosomes from mice treated systemically with vehicle or FK506 (3 mg/kg per day) for 7 d (n = 8 mice per group). Proteins were first immunoprecipitated (IP) with an anti-α2δ-1 antibody, and immunoblotting was performed using anti-GluA1, anti-GluA1, and anti-α2δ-1 antibodies. The input protein bands of GluA1, GluA2, and α2δ-1 were normalized to those of the vehicle group on the same blots. B, Representative blotting images show the enrichment of ER and synaptosome (synapt) fractions in spinal cord samples. Calreticulin and PSD95 were used as an ER and synaptic marker, respectively. ER, endoplasmic reticulum; Veh, vehicle; FK, FK506. Data are expressed as mean ± SEM. **p < 0.01, ***p < 0.001 (two-tailed Student's t test).

α2δ-1 is required for calcineurin inhibitor-induced increases in the prevalence of postsynaptic CP-AMPARs in spinal excitatory neurons

Gabapentinoids, such as gabapentin and pregabalin, are inhibitory ligands of α2δ-1 (Fuller-Bicer et al., 2009) and are clinically used to treat chronic neuropathic pain. We used pregabalin to determine the role of α2δ-1 in synaptic incorporation of CP-AMPARs in spinal VGluT2 neurons augmented by the calcineurin inhibitor. To this end, we treated spinal cord slices with 20 µM pregabalin and 5 µM FK506 for 30 min and then examined the I–V relationship of AMPAR-EPSCs in tdTomato-labeled VGluT2 neurons. Remarkably, treatment with pregabalin abolished inward rectifying of AMPAR-EPSCs and increased the rectification index of AMPAR-EPSCs in VGluT2 neurons in FK506-treated slices (n = 16 neurons; Fig. 6A–C). The rectification index of AMPAR-EPSCs in VGluT2 neurons treated with pregabalin and FK506 was similar to the value in vehicle-treated mice (Figs. 2B,E; 6B,C).

Figure 6.

α2δ-1 has a pivotal role in postsynaptic CP-AMPARs of spinal dorsal horn neurons induced by FK506 treatment. A, B, Representative recording traces (A) and quantification (B) show the current–voltage relationship of AMPAR-EPSCs in lamina II VGluT2 neurons of wild-type (WT) mice treated with 5 µM FK506 plus 20 µM pregabalin (PGB, n = 16 neurons from 3 mice) and in lamina II neurons of Cacna2d1 knock-out (KO) mice treated with 5 µM FK506 (n = 15 neurons from 3 mice). C, Rectification index (I_{+50 mV}/I_{−50 mV}) of AMPAR-EPSCs in lamina II VGluT2 neurons of WT mice treated with FK506 plus pregabalin and in lamina II neurons of Cacna2d1 KO mice treated with 5 µM FK506. PGB, pregabalin. Data are mean ± SEM.

Because gabapentinoids target both α2δ-1 and α2δ-2 proteins (Marais et al., 2001; Fuller-Bicer et al., 2009), we subsequently used Cacna2d1 (gene encoding α2δ-1) KO mice to validate the role of α2δ-1 in calcineurin inhibitor-induced increases in synaptic CP-AMPARs in the spinal cord. Most of the neurons in spinal lamina II express VGluT2 (Browne et al., 2020; Chen et al., 2022). We examined the I–V relationship of AMPAR-EPSCs in lamina II neurons from Cacna2d1 KO mice treated with 5 µM FK506 for 30 min. Whole-cell recordings in spinal cord slices showed a linear I–V relationship of AMPAR-EPSCs in lamina II neurons of FK506-treated Cacna2d1 KO mice (n = 15 neurons; Fig. 6A,B). The rectification index of AMPAR-EPSCs in lamina II neurons from FK506-treated Cacna2d1 KO mice was similar to that of VGluT2 neurons from wild-type mice treated with pregabalin plus FK506 (Figs. 2B,E; 6B,C). These data provide unambiguous evidence that α2δ-1 is integral to the presence of synaptic CP-AMPARs of excitatory dorsal horn neurons induced by calcineurin inhibitors.

Disrupting α2δ-1–AMPAR interactions attenuates FK506-induced synaptic CP-AMPARs in spinal VGluT2 neurons

α2δ-1 physically interacts with GluA1 and GluA2 via its transmembrane C-terminal domain (an intrinsically disordered region), and a Tat-fused α2δ-1 C-terminal mimicking peptide (α2δ-1CT peptide) effectively disrupts the α2δ-1 interaction with GluA1 and GluA2 (Li et al., 2021). Accordingly, we used α2δ-1CT peptide to determine whether α2δ-1–AMPAR interactions are required for calcineurin inhibitor-induced increases in synaptic CP-AMPARs in spinal excitatory neurons. We incubated spinal cord slices with 5 µM FK506 in the presence of 1 µM α2δ-1CT peptide or 1 µM scrambled control peptide for 30 min and then examined the I–V relationship of AMPAR-EPSCs. Whole-cell recordings of tdTomato-labeled VGluT2 neurons in lamina II showed that treatment with α2δ-1CT peptide, but not the control peptide, abolished the FK506-induced inward rectification of AMPAR-EPSCs and increased the rectification index of AMPAR-EPSCs (n = 14 neurons in control peptide group; n = 15 neurons in α2δ-1CT peptide group; p = 0.0129; t₍₂₈₎= 2.656; Fig. 7A–C). These results suggest that calcineurin inhibition increases the prevalence of synaptic CP-AMPARs in spinal excitatory neurons via α2δ-1–bound AMPARs.

Figure 7.

Disrupting α2δ-1–AMPAR interactions diminishes postsynaptic CP-AMPARs of spinal VGluT2 neurons induced by FK506 treatment. A, B, Original recording traces (A) and quantification (B) show the current–voltage relationship of AMPAR-EPSCs in lamina II VGluT2 neurons treated with 5 µM FK506 in the presence of 1 µM control peptide (n = 14 neurons from 4 mice) or 1 µM α2δ-1CT peptide (n = 15 neurons from 4 mice). C, Rectification index (I_{+50 mV}/I_{−50 mV}) of AMPAR-EPSCs in lamina II VGluT2 neurons treated with FK506 in the presence of control peptide or α2δ-1CT peptide. Data are presented as mean ± SEM. *p < 0.05 (two-tailed Student's t test).

Inhibiting CK2 activity reverses FK506-induced potentiation of α2δ-1 interaction with GluA1 and GluA2 in the spinal cord

The protein kinase CK2 controls the phosphorylation of GluA1 and GluA2 (Lussier et al., 2014). In addition, CK2 inhibition reverses NMDAR hyperactivity in the spinal cord and pain hypersensitivity caused by the calcineurin inhibitor in rats (Hu et al., 2014). In mice treated systemically with FK506 for 7 d, intrathecal injection of 10 µg DRB, a cell-permeable specific CK2 inhibitor (Zandomeni and Weinmann, 1984), readily reversed the reduction in the baseline withdrawal thresholds in response to tactile, pressure, and heat stimuli applied to the hindpaw (n = 8 mice per group; Fig. 8A). In contrast, intrathecal injection of 10 µg DRB had no effect on the withdrawal thresholds in vehicle-treated mice (n = 8 mice per group; Fig. 8A).

Figure 8.

CK2 and calcineurin reciprocally control nociception and α2δ-1 interactions with GluA1 and GluA2 in the spinal cord. A, Time course effects of intrathecal injection of 10 µg DRB or vehicle on the mechanical and heat withdrawal thresholds in mice systemically treated with FK5065 (3 mg/kg per day) or vehicle for 7 d (n = 8 mice per group). *p < 0.05, **p < 0.01, ***p < 0.001 compared with the baseline within the group. ^##p < 0.01, ^###p < 0.001 compared between groups at the same time point (two-way ANOVA with Šidák's post hoc test). B–D, Representative gel images (B) and quantification show the protein levels of GluA1, GluA2, and α2δ-1 (C) and α2δ-1–GluA1 and α2δ-1–GluA2 complexes (D) in dorsal spinal cord tissues from mice treated with FK506 in the presence of DRB or vehicle (n = 8 mice per group). FK, FK506; Veh, vehicle. Data are presented as mean ± SEM. ***p < 0.001 (two-tailed Student's t test).

It is possible that CK2 and calcineurin have opposing roles in regulating AMPAR phosphorylation and the α2δ-1 interaction with GluA1 and GluA2. Therefore, we performed co-IP analysis to measure the protein levels of α2δ-1/GluA1 and α2δ-1/GluA2 complexes in dorsal spinal cord tissues. Spinal cord tissues were removed from FK506-treated mice 30–40 min after intrathecal injection of DRB or vehicle. Co-IP assay showed that total protein levels of GluA1, GluA2, and α2δ-1 in dorsal spinal cord tissues did not differ significantly between mice treated with FK506 plus DRB and mice treated with FK506 plus vehicle (Fig. 8B,C). However, the α2δ-1/GluA1 and α2δ-1/GluA2 protein complexes in dorsal spinal cord tissues were significantly lower in mice treated with FK506 plus DRB than in mice treated with FK506 plus vehicle (n = 8 mice per group; p = 0.0002; t₍₁₄₎ = 5.003 for α2δ-1/GluA1; p = 0.0000; t₍₁₄₎ = 7.050 for α2δ-1/GluA2; Fig. 8B,D). These findings suggest that CK2 and calcineurin reciprocally control the α2δ-1 interaction with GluA1 and GluA2 subunits in the spinal cord.

Inhibiting CK2 activity abolishes FK506-induced postsynaptic CP-AMPARs in spinal VGluT2 neurons

Lastly, we determined whether CK2 acts antagonistically in counteracting calcineurin inhibitor-induced synaptic CP-AMPARs in spinal VGluT2 neurons. We incubated spinal cord slices with 5 µM FK506 in the presence of vehicle or 100 µM DRB for 30 min and then examined the I–V relationship of AMPAR-EPSCs. Whole-cell recordings of tdTomato-labeled VGluT2 neurons in lamina II showed that treatment with DRB normalized the FK506-induced inward rectification of AMPAR-EPSCs and increased the rectification index of AMPAR-EPSCs (n = 15 neurons in the FK506 plus vehicle group; n = 17 neurons in the DRB plus vehicle group; n = 16 neurons in the FK506 plus DRB group; p = 0.0001; H₍₂₎= 18.25; Fig. 9A–C). In contrast, DRB had no significant effect on the linear I–V relationship of AMPAR-EPSCs of VGluT2 neurons in vehicle-treated spinal cord slices (Fig. 9A–C). These data suggest that CK2 and calcineurin exert opposing actions in the control of synaptic CP-AMPARs in spinal excitatory neurons.

Figure 9.

CK2 inhibition reverses postsynaptic CP-AMPARs of spinal VGluT2 neurons potentiated by FK506 treatment. A, B, Original recording traces (A) and quantification (B) show the current–voltage relationship of AMPAR-EPSCs in lamina II VGluT2 neurons treated with 5 µM FK506 plus vehicle (n = 15 neurons from 3 mice), 100 µM DRB plus vehicle (n = 17 neurons from 3 mice), or 5 µM FK506 plus 100 µM DRB (n = 16 neurons from 3 mice). C, Rectification index (I_{+50 mV}/I_{−50 mV}) of AMPAR-EPSCs in lamina II VGluT2 neurons incubated with FK506 in the presence of vehicle or DRB. Data are presented as mean ± SEM. **p < 0.01, ***p < 0.001 (Kruskal–Wallis test followed by Dunn's post hoc test).

Discussion

Our study reveals that calcineurin constitutively controls synaptic AMPAR phenotypes in spinal dorsal horn neurons in a cell type-specific manner. We showed that blocking CP-AMPARs with IEM-1460 reduced ∼20% of the amplitude of AMPAR-EPSCs in VGluT2-expressing excitatory, but not in VGAT-expressing inhibitory, dorsal horn neurons in mice treated with FK506. Furthermore, FK506 treatment induced characteristic inward rectification of AMPAR-EPSCs selectively in VGluT2 dorsal horn neurons. These findings strongly suggest an increased prevalence of postsynaptic CP-AMPARs in spinal excitatory neurons caused by calcineurin inhibition. Because AMPAR currents recorded from GluA1 homotetramers are near 0 at the holding potentials of +50 and +70 mV (Watson et al., 2017; Ge et al., 2019; Li et al., 2021), the limited inward rectification in the I–V relationship of AMPAR-EPSCs also suggests a partial contribution of CP-AMPARs to postsynaptic AMPARs in spinal glutamatergic neurons in mice treated with FK506. Under normal conditions, synaptic AMPARs in the spinal cord primarily comprise GluA1/GluA2 heterotetramers (Chen et al., 2013; Li et al., 2021), which are crucial for maintaining appropriate cytoplasmic Ca²⁺ levels. CP-AMPARs, particularly GluA1 homotetramers, possess enhanced single-channel conductance compared with GluA2-containing AMPARs (Swanson et al., 1997; Feldmeyer et al., 1999). Also, the persistent synaptic expression of CP-AMPARs can lead to excessive Ca²⁺ influx, resulting in strong postsynaptic depolarization and neuronal hyperactivity (Plant et al., 2006; Li et al., 2012). Thus, increased synaptic expression of CP-AMPARs in excitatory neurons can markedly strengthen glutamatergic synaptic transmission in the spinal dorsal horn. Interestingly, potentiated synaptic NMDAR activity in the spinal cord induced by nerve injury and synaptic disinhibition also exhibits the same cell type-specific pattern (Huang et al., 2023a,b), indicating a principal role of spinal excitatory interneurons in the relay and amplification of nociceptive input in various chronic pain conditions. Our findings underscore the importance of intrinsic calcineurin activity in restraining the synaptic integration of CP-AMPARs in spinal excitatory neurons.

Our study provides new evidence about the functional significance of spinal CP-AMPARs in maintaining pain hypersensitivity in CIPS. AMPARs mediate increased glutamatergic input to the spinal dorsal horn, and blocking spinal AMPARs attenuates pain hypersensitivity caused by nerve injury (Chen et al., 2000, 2013). In the present study, we showed that blocking CP-AMPARs with IEM-1460 at the spinal cord level rapidly reversed tactile allodynia and hyperalgesia caused by prolonged treatment with FK506. However, intrathecal injection of IEM-1460 had no effect on nociceptive thresholds in control animals, indicating that CP-AMPAR activity is increased in the spinal cord, sustaining calcineurin inhibitor-induced pain hypersensitivity. Notably, neuropathic pain caused by traumatic nerve injury and diabetic neuropathy is also associated with increased CP-AMPAR activity in the spinal dorsal horn (Chen et al., 2013; Chen et al., 2019). Furthermore, the NMDAR activity is similarly potentiated in the spinal dorsal horn in animal models of CIPS and nerve injury-induced neuropathic pain (Zhou et al., 2012; Chen et al., 2014a, b; Huang et al., 2022a). Thus, CIPS seems to share similar cellular and molecular mechanisms identified for neuropathic pain. Consistent with this notion, patients with CIPS often exhibit symptoms resembling neuropathic pain (Noda et al., 2008; Wei et al., 2018). The incorporation of CP-AMPARs into synapses of spinal excitatory neurons represents a new mechanism underlying CIPS. Unlike NMDARs that require a large depolarization to remove the channel block by Mg²⁺, CP-AMPARs are open and allow Ca²⁺ influx at negative membrane potentials. The sustained Ca²⁺ influx through synaptic CP-AMPARs could trigger certain downstream intracellular signaling, such as calpain-mediated KCC2 proteolysis, thereby diminishing synaptic inhibition (Zhou et al., 2012). Hence, even a small amount of synaptic CP-AMPARs could play a key role in pain hypersensitivity induced by calcineurin inhibitors.

We demonstrated in this study that constitutive calcineurin activity regulates synaptic CP-AMPARs via modifying GluA1/GluA2 heteromeric assembly. The subunit composition of synaptic AMPARs depends largely on the availability of intracellularly assembled AMPAR subunits, which are dynamically controlled by phosphorylation. Constitutive calcineurin activity likely maintains dephosphorylation of both GluA1 and GluA2 subunits. In this regard, calcineurin controls GluA1 phosphorylation in the hippocampus and spinal dorsal horn (Sanderson et al., 2012; Miletic et al., 2015). Also, endogenous calcineurin limits GluA2 phosphorylation in the striatum (Ahn and Choe, 2010). GluA1 and GluA2 normally form heterotetramers in the ER, which are then rapidly transported to synapses (Mansour et al., 2001; Greger et al., 2002). We found that treatment with FK506 substantially reduced GluA1/GluA2 heteromers in the spinal cord. Furthermore, treatment with FK506 diminished GluA1/GluA2 complexes and caused GluA2 retention in the ER of the spinal cord. Unlike GluA1, GluA2 subunits are largely unassembled and ER-retained, and GluA2 can exit from the ER only when assembled with GluA1 (Greger et al., 2003; Isaac et al., 2007). When calcineurin inhibition impairs intracellular assembly of GluA1/GluA2 heteromers, unassembled GluA2 becomes retained in the ER. Our findings suggest that calcineurin likely maintains synaptic dominance of GluA1/GluA2 heterotetramers by minimizing the phosphorylation of GluA1 and GluA2 subunits in spinal dorsal horn neurons.

A major finding of our study is that α2δ-1 is integral to the calcineurin inhibitor-induced switch from synaptic dominance of Ca²⁺-impermeable AMPARs to CP-AMPARs in the spinal cord. This occurs by impeding intracellular GluA1/GluA2 heteromeric assembly and limiting the availability of GluA1/GluA2 heterotetramers for synaptic expression. Because α2δ-1 predominantly functions as a phospho-binding protein (Zhou et al., 2021), increased AMPAR phosphorylation by calcineurin inhibitors could enhance α2δ-1 association with phosphorylated GluA1 and GluA2 (Zhou et al., 2024). In this study, we found that treatment with FK506 substantially increased the interaction of α2δ-1 with both GluA1 and GluA2 in the spinal cord. α2δ-1 regulates synaptic AMPAR composition in the spinal cord and brain by directly interacting with GluA1 and GluA2 through its C terminus (Li et al., 2021; Zhou et al., 2022b). The α2δ-1/GluA1 and α2δ-1/GluA2 complexes likely cause conformational changes of GluA1 and GluA2 proteins, thereby interrupting normal GluA1/GluA2 heteromeric assembly. In contrast, this α2δ-1–AMPAR interaction does not affect the assembly of GluA1 homotetramers in the ER (Li et al., 2021). Inhibiting α2δ-1 with gabapentinoids or disrupting α2δ-1–AMPAR interactions with α2δ-1CT mimicking peptide can effectively reverse GluA1/GluA2 heteromeric assembly impaired by α2δ-1 (Li et al., 2021). In this study, we demonstrated that pregabalin, Cacna2d1 KO, or the α2δ-1CT peptide consistently abolished the inward rectification of AMPAR-EPSCs in dorsal horn neurons caused by calcineurin inhibition. These findings strongly support the crucial role of α2δ-1 in calcineurin inhibitor-induced synaptic expression of CP-AMPARs in the spinal cord. Given that α2δ-1 is specifically expressed in VGluT2-expressing excitatory neurons in the spinal dorsal horn (Koga et al., 2023), this explains why calcineurin inhibition preferentially promotes synaptic CP-AMPAR in VGluT2, but not in VGAT, dorsal horn neurons.

Another salient finding of our study is that the protein kinase CK2 counteracts the action of calcineurin in the control of synaptic incorporation of CP-AMPARs in spinal excitatory neurons. Traumatic nerve injury increases CK2 activity in the spinal cord, and CK2 inhibition or knockdown at the spinal cord level attenuates neuropathic pain but has no effect on normal nociception (Chen et al., 2014b). We found in this study that intrathecal injection of DRB reversed the pain hypersensitivity induced by the systemic administration of FK506 in mice but did not affect nociceptive thresholds in vehicle-treated mice, suggesting that these two enzymes have opposing roles in regulating nociceptive transmission at the spinal cord level. Numerous serine sites on GluA1 and GluA2 subunits are susceptible to phosphorylation by various protein kinases, including CK2, calmodulin-dependent protein kinase II, protein kinase A, and protein kinase C (Roche et al., 1996; Mammen et al., 1997; Matsuda et al., 1999; Lussier et al., 2014). Notably, CK2 activity regulates the phosphorylation status of GluA1 and GluA2, but not GluA3 or GluA4 (Lussier et al., 2014). In the present study, we showed that treatment with DRB abolished FK506-induced inward rectification of AMPAR-EPSCs in spinal VGluT2 neurons. In addition, DRB treatment attenuated the α2δ-1 interaction with GluA1 and GluA2 in the spinal cord augmented by the calcineurin inhibitor. Our findings collectively suggest that the intrinsic balance between CK2-induced phosphorylation and calcineurin-mediated dephosphorylation governs the α2δ-1-regulated heteromeric assembly of GluA1/GluA2 and the subunit composition of synaptic AMPARs in excitatory dorsal horn neurons.

In addition to physically interacting with AMPARs, α2δ-1 can form a protein complex with NMDARs via its C terminus in the brain and spinal cord (Chen et al., 2018; Luo et al., 2018; Zhou et al., 2018). Also, calcineurin inhibition promotes α2δ-1-dependent synaptic trafficking of NMDARs in the spinal cord and hypothalamus (Huang et al., 2020; Zhou et al., 2023). Intrathecal administration of the NMDAR antagonist or α2δ-1CT peptide reverses pain hypersensitivity induced by calcineurin inhibitors (Chen et al., 2014a; Huang et al., 2020). Because α2δ-1 interacts with both NMDARs and GluA1/GluA2 via its C-terminal domain, α2δ-1CT peptide likely disrupts both α2δ-1–NMDAR and α2δ-1–AMPAR interactions in vivo. Augmented synaptic NMDAR activity normally would be expected to limit synaptic incorporation of CP-AMPARs by increasing Ca²⁺ influx and consequently stimulating calcineurin activity. However, when calcineurin activity is inhibited in CIPS, α2δ-1 has a distinctive ability to elevate intracellular Ca²⁺ levels and enhance glutamatergic input to spinal excitatory neurons by interacting with both NMDARs and GluA1/GluA2. This leads to increased synaptic activity of NMDARs and CP-AMPARs, which could mutually augment the excitability of spinal glutamatergic neurons, thereby amplifying nociceptive transmission to elicit pain hypersensitivity in CIPS. α2δ-1 is expressed in many regions of the brain (Cole et al., 2005). Because altered calcineurin activity in the brain is associated with Down syndrome, neurodegenerative disease, and neurogenic hypertension (Ladner et al., 1996; Hoeffer et al., 2007; Zhou et al., 2022a, 2023), α2δ-1-mediated dual changes in synaptic NMDARs and CP-AMPARs may have a potential role in the development of these disorders.

In summary, our study provides substantial new information indicating that calcineurin constitutively restrains pain hypersensitivity and inhibits synaptic incorporation of CP-AMPARs in spinal excitatory neurons. When calcineurin activity is impaired, α2δ-1 promotes synaptic expression of CP-AMPARs in excitatory dorsal horn neurons via interacting with GluA1 and GluA2 to disrupt their heteromeric assembly and availability of GluA1/GluA2 heterotetramers. Furthermore, calcineurin and CK2 reciprocally regulate the subunit composition of synaptic AMPARs through the tuning of phosphorylation-dependent, dynamic interactions between α2δ-1 and GluA1/GluA2 (Fig. 10). These findings not only advance our mechanistic understanding of the cellular and molecular basis of synaptic AMPAR phonotypes, controlled by opposing activities of calcineurin and CK2, but also suggest new therapeutic targets (e.g., α2δ-1 and CK2) for treating CIPS. Notably, some case reports show that pregabalin and gabapentin effectively alleviate CIPS in patients (Tasoglu et al., 2016; Wei et al., 2018). Accordingly, CK2 inhibitors, gabapentinoids, and α2δ-1CT peptides could restore normal synaptic function and reduce pain syndromes by diminishing aberrant α2δ-1–AMPAR interactions caused by calcineurin inhibitors.

Figure 10.

Schematic shows the potential role of α2δ-1-mediated GluA1/GluA2 heteromeric assembly in reciprocal control of postsynaptic CP-AMPARs by calcineurin and CK2 in spinal excitatory neurons. In spinal excitatory neurons, GluA1 and GluA2 subunits are minimally phosphorylated and typically assemble into GluA1/GluA2 heterotetramers, which are subsequently expressed at postsynaptic membranes. Upon calcineurin inhibition, α2δ-1 interacts directly with phosphorylated GluA1 and GluA2, disrupting heteromeric assembly of GluA1/GluA2 in the ER. This results in partial synaptic incorporation of GluA1 homotetramers, which are Ca²⁺ permeable. Inhibiting CK2 reduces phosphorylation of GluA1 and GluA2 and their interactions with α2δ-1 caused by calcineurin inhibitors, thereby restoring heteromeric assembly of GluA1/GluA2 and synaptic dominance of GluA1/GluA2 heterotetramers in spinal excitatory neurons.