Synaptic mechanisms and neuronal oscillations have been proposed to be responsible for neuropathic pain formation. Many studies have also highlighted the important role of electrical synapses in synaptic plasticity and in neuronal oscillations. Thus, electrical synapses may contribute to neuropathic pain generation. However, previous studies have primarily focused on the role of chemical synapses, while ignoring the role of electrical synapses, in neuropathic pain generation.
The authors adopted microinjection, RNA interference techniques, and behavioral tests to verify the link between connexin 36 (Cx36) and neuropathic pain. They also studied the selective Cx36 blocker mefloquine in rat chronic constriction injury and spared nerve injury model of neuropathic pain. Electrophysiologic recordings were used to further confirm the behavioral data.
The authors found that Cx36, which constitutes the neuron–neuron electrical synapses, was up-regulated in the anterior cingulate cortex after nerve injury (n = 5). Meanwhile, Cx36-mediated neuronal oscillations in the gamma frequency range (30 to 80 Hz) (n = 7 to 8) and the neuronal synaptic transmission (n = 13 to 19) were also enhanced. Neuropathic pain was relieved by disrupting Cx36 function or expression in the anterior cingulate cortex. They also found that mefloquine, which are clinically used for treating malaria, affected gamma oscillations and synaptic plasticity, leading to a sustained pain relief in chronic constriction injury and spared nerve injury models (n = 7 to 12).
The electrical synapses blocker mefloquine could affect gamma oscillations and synaptic plasticity in the anterior cingulate cortex and relieve neuropathic pain. Cx36 may be a new therapeutic target for treating chronic pain.
The electrical synapse protein connexin 36 is up-regulated in the anterior cingulate cortex after nerve injury in rats, and gamma-frequency neuronal oscillations are enhanced as well. Mefloquine, an electrical synapse inhibitor, reduces allodynia in the same nerve injury model.
Synapses in the brain can be of either the chemical or electrical type
Electrical synapses have special roles in neuronal oscillations, which may be involved in chronic pain
The electrical synapse protein connexin 36 is up-regulated in the anterior cingulate cortex after nerve injury in rats, and gamma-frequency neuronal oscillations are enhanced as well
Mefloquine, an electrical synapse inhibitor, reduces allodynia in the same nerve injury model
Neuropathic pain, which affects approximately 8% of the population, is a devastating pain syndrome caused by various insults to the nervous system, resulting in chronic spontaneous pain, thermal hyperalgesia, and mechanical allodynia.1,2 Currently available pharmacologic treatments for neuropathic pain have limited therapeutic efficacy. More than two thirds of neuropathic pain patients suffer from insufficient pain relief3 partially because the underlying mechanisms of neuropathic pain development remain poorly understood. Although many studies have confirmed that chemical synapses are crucial for neuropathic pain formation,4–6 little is known concerning the role of electrical synapses in neuropathic pain pathogenesis.
The electrical synapse is a special transmembrane channel that allows for the fast and efficient transduction of electrical signals and the direct exchange of signals between neurons, thus maintaining and regulating cell activity and stability.7 The electrical synapse is composed of two membrane-docked hexameric hemichannels that consist of connexins (Cx) from two adjacent neurons. Approximately 20 types of synaptic Cx proteins exist, and different Cx proteins are named after their respective Cx molecular weights.8 Connexin 36 (Cx36), which appears in neuron–neuron gap junctions (electrical synapses) but not in the gap junctions between astrocytes and/or oligodendrocytes,9 is expressed throughout the brain.10 When Cx36 is knocked out, electrical coupling between neurons is eliminated or profoundly reduced. Thus, Cx36 is an indispensable constituent of electrical coupling between neurons.7
Previous studies have reported that electrical synapses play an important role in generating neuronal oscillations.11,12 A study regarding Cx36-knockout mice demonstrated that the loss of Cx36 disrupts gamma oscillations.11 Application of the general gap junction blocker carbenoxolone disodium salt (Cbx) also significantly reduced the power of gamma neuronal oscillations.13 Moreover, painful stimuli induce the activation of neuronal oscillations in the portion of the cortex associated with pain.14,15 Therefore, electrical synapses may participate in pain processing. Furthermore, deleting Cx36 impairs long-term potentiation of synaptic transmission in the hippocampus,16 and Cbx induces a long-lasting depression of neurotransmission,17 indicating that the function of electrical synapses is associated with synaptic plasticity. Synaptic plasticity that contributes to central sensitization is the key mechanism for neuropathic pain.18–20 Thus, we hypothesized that electrical synapses are crucial for neuropathic pain formation. In the current study, we aimed to study the role of electrical synapses in neuropathic pain pathophysiology.
Materials and Methods
Sprague–Dawley rats aged 9 to 11 weeks and weighing 240 to 340 g (Shanghai Sipper; BK Laboratory Animals Co., Ltd., China) were housed in cages at 24°C and at 50 to 60% humidity with a 12/12 h light/dark cycle and a sufficient food and water supply. Sample size determination was based on previous experience. Rats were randomly divided into different groups. All of the surgical procedures were performed under anesthesia with intraperitoneal injection of pentobarbital sodium (40 mg/kg). The Animal Use and Care Committee for Research and Education, Shanghai Jiao Tong University, Shanghai, China, approved this research protocol.
Medications and Reagents
First, mefloquine hydrochloride (Meflo; Sigma-Aldrich, USA, Catalog #M2319) and its isomers were suspended in Tween-80 and then dissolved in normal saline. Once-daily intraperitoneal injections of Meflo or Meflo isomers were administered to the rats on days 0, 1, 2, 3, and 4 after the chronic constriction injury (CCI) or the spared nerve injury (SNI). Both the sham group and the vehicle group received 1% Tween-80 saline. For microinjection, Cbx (100 mM; Sigma-Aldrich, Catalog #C4790) and Meflo (100 mM; dimethyl sulfoxide stock solution) were dissolved in 0.1 M sterile phosphate-buffered saline (PBS), followed by ultrasonification for 10 min. Vehicle animals received similar volumes of 0.1 M sterile PBS or 0.1 M sterile PBS with 1% dimethyl sulfoxide for the Cbx and Meflo vehicles, respectively. (+)-(11S,12R)-Meflo and (−)-(11R,12S)-Meflo were synthesized by the Shanghai Engineering Research Center of Molecular Therapeutics and New Drug Development, Shanghai, China.
Chronic constriction injury model was established in accordance with the method described by Bennett and Xie.21 In brief, the left sciatic nerves were exposed unilaterally after a skin incision was made at the mid-thigh level, followed by blunt dissection at the biceps. Four chromic gut (5-0) ligatures were tied loosely around the nerve proximal to its trifurcation at 1-mm intervals. Sham surgery was performed by exposing the sciatic nerve without ligation.
Spared nerve injury model was established according to the method described by Decosterd and Woolf.22 In brief, an incision was made on the upper edge of the left hind leg of the rat. The muscle was isolated longitudinally to expose the sciatic nerve trunk and its subordinate tibial, peroneal, and sural nerves. The tibial and peroneal nerves were ligated and cut transversely, whereas the thin sural nerve was preserved. Care was taken to avoid excessively stretching the nerves.
Cannulation and Microinjection
Each rat was anesthetized with pentobarbital sodium and secured on a stereotaxic apparatus. A median incision was made to expose the skull of the rat. In accordance with the position determined by the stereotaxic atlas, 28-gauge guide cannulas were bilaterally implanted into the anterior cingulate cortex (ACC) at the following coordinates: 2.7 mm anterior from bregma, ± 0.6 mm lateral to midline, and 2.0 mm ventral from the dura; 0.4 mm anterior from bregma, ± 0.6 mm lateral to midline, and 2.0 mm ventral from the dura. Three stainless steel screws were implanted into the skull around the guide cannulas and fixed with dental acrylic. Dummy cannulas were inserted after surgery and left in place until the infusion day. The animals were housed individually after surgery. Penicillin (80,000 units) was administered for infection prophylaxis 1 day before surgery and 2 days consecutively after surgery. Cerebral microinjection was performed 7 days after cannulation. The dummy cannulas were removed before microinjection. Internal injection cannulas (32 gauge), which extended 1 mm beyond the guide cannulas, were inserted. The solution was injected at a constant rate for longer than 120 s using a microinjector. The injection cannulas were left in place for 3 additional min before being withdrawn. The animals were allowed 30 additional minutes before beginning behavior experimentation. Once-daily, Cbx or Meflo was microinjected into the ACC region of rat brains on days 0, 1, 2, 3, and 4 after CCI.
Mechanical Allodynia Test
The animals were habituated for 2 to 3 days in the test environment before each test. The rats were placed in a plexiglass box with a metal net bottom for 30 min. After habituation to the environment, the hind paw was stimulated using one of a series of von Frey hairs with logarithmically increasing stiffness (0.6, 1, 1.4, 2, 4, 6, 8, 10, and 15 g) (Stoelting Co., USA), which was presented perpendicular to the plantar surface (5 to 6 s for each hair). A positive performance was recorded when the rat escaped the mild pressure from von Frey hairs or raised the hind leg. Dixon up–down method was used to determine the 50% withdrawal threshold.23 The experimenters who conducted the various assessments were blinded to the treatment conditions.
Thermal Hyperalgesia Test
The rats were placed in a plexiglass box on a 3-mm-thick glass plate. After habituation in the box for 30 min, the sole skin of each animal was irradiated with light within a 0.5-cm-diameter circle using a BME-410 thermal radiation stimulator (Institute of Biomedical Engineering, Peking Union Medical College, Tianjin, China) at 10 V and 30 W. The time from irradiation initiation to paw withdrawal was recorded as the paw withdrawal latency value. A cutoff time of 40 s was used to avoid local burn injury. Three measurements were taken for each animal, with a 6- to 8-min interval allowed between trials, and the mean value was used for the analysis.
The open-field test system was applied to pair sets of photo beams to detect movement. All of the animals were acclimated to the testing room for 30 min before the start of the session. The rats were placed individually in the center of the observation box, which had four transparent walls, an open top, and a gray bottom. The total distance that each animal traveled in 15 min was measured using a tracking analysis system (Coulbourn Instruments, USA). This distance was used as a parameter for rat locomotion.
Motor coordination and balance were tested using an accelerating rotarod (Shanghai Jiliang Software Technology, Shanghai, China). This apparatus is composed of a base platform and a rotating horizontal rod with a nonskid surface. The rats were habituated to the apparatus for 30 min to avoid stress during testing. Before testing, the rats were acclimatized to acceleration in six training runs over 3 days. The rod was set to accelerate from 8 to 25 rpm in a 5-min period. The locomotor function was assessed as the performance time on the rod from the start of rotation until the animal fell to the base platform. The mean performance time of the test runs was recorded.
The rats were deeply anesthetized with pentobarbital sodium (50 mg/kg). The bilateral ACC was removed, homogenized in a lysis buffer containing 10 mM Tris-HCl, pH 7.4, 2 mM EDTA, 1% sodium dodecyl sulfate, and a cocktail of protease inhibitors and phosphatase inhibitors, lysed on ice for 30 min, and centrifuged at 4°C for 10 min. The supernatant was removed for protein quantification using a bicinchoninic acid protein assay kit (Beyotime Institute of Biotechnology, Shanghai, China), and 20 μg of protein was loaded onto an 8% sodium dodecyl sulfate–polyacrylamide gel electrophoresis gel for separation. The proteins were transferred to polyvinylidene fluoride membranes at 4°C. After blocking with 5% skim milk for 3 h, membranes were incubated overnight at 4°C with a polyclonal goat antibody against Cx36 (1:400; Santa Cruz Biotechnology, USA, Catalog #sc-14904).24 To ensure equal loading, the membranes were also probed with a 1:3,000 dilution of anti-actin antibody (Sigma Life Science, USA, Catalog #A2668). The results of the Western blotting were quantified using ImageJ software (National Institutes of Health, USA).
We deeply anesthetized the animals with pentobarbital sodium and transcardially perfused the rats with saline, followed by 4% paraformaldehyde. The brain tissues were removed completely and fixed in 4% paraformaldehyde for 4 h, followed by submersion in a 20% cane sugar solution until the tissues sank to the bottom. Then, the brain tissues were sliced into 30 μm sections in a cryostat. The sections were blocked with 5% normal goat serum at room temperature and incubated overnight at 4°C with a polyclonal rabbit antibody against Cx36 (1:50; Santa Cruz Biotechnology, Catalog #sc-25715). Then, the sections were incubated for 1 h at room temperature with Alexa Fluor 488-conjugated donkey anti-rabbit immunoglobulin G (1:300; Life Technologies, USA, Catalog #A-11008) in the dark. The slices were coverslipped with a mixture of 50% (v/v) glycerol and 2.5% (w/v) triethylenediamine in 0.05 M PBS. 4′,6-Diamidino-2-phenylindole was also included in the mounting medium for covisualization of nuclei. The sections were observed under a confocal microscope.
Brain Slice Preparation
Brain slices at the level of the ACC were prepared using standard methods.25 In brief, the animals were anesthetized with pentobarbital sodium, and their brains were quickly excised from the skull and submerged in an ice-cold artificial cerebrospinal fluid (ACSF: 124 mM NaCl, 2.5 mM KCl, 2 mM CaCl2, 1 mM MgSO4, 25 mM NaHCO3, 1 mM NaH2PO4, and 10 mM glucose). After 1 to 2 min of chilling, each brain was trimmed to a block containing the anterior half of the brain. Using a vibroslicer (VT1000 S; Leica Microsystems, USA), coronal slices (350 to 400 μm) were cut from the tissue block in ice-cold ACSF. The slices containing the ACC were transferred to a gas interface-type recording chamber perfused with aerated (95% O2–5% CO2) ACSF (26° to 28°C) at a rate of 1 to 1.5 ml/min using a peristaltic pump-driven or gravity-fed bath perfusion system.
Whole Cell Patch Clamp Recordings
Voltage clamp recordings were obtained from neurons in ACC slices equilibrated for 1 to 8 h in the recording chamber. The micropipettes (tip diameter: 1.5 to 2.0 μm; resistance: 4 to 6 MΩ) were pulled from borosilicate capillaries (P-97; Sutter Instruments, USA) and filled with an internal solution (adjusted to pH 7.2 with KOH) composed of the following: 145 mM K-gluconate, 5 mM NaCl, 1 mM MgCl2, 0.2 mM EGTA, 10 mM HEPES, 2 mM Mg-adenosine triphosphate, 0.1 Na3-guanosine-5'-triphosphate, and 10 phosphocreatine disodium. Evoked excitatory postsynaptic currents (EPSCs) were recorded from layer II/III neurons, and the stimulations were delivered by a bipolar tungsten-stimulating electrode (biphasic square wave, 0.3 ms duration) placed in layer V of the ACC. The α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor–mediated EPSCs were induced by repetitive stimulations at 0.05 Hz, and the neurons were voltage clamped at −70 mV in the presence of DL-2-amino-5-phosphonovaleric acid (50 μM). The internal solution consisted of the following: 140 mM cesium methanesulfonate, 5 mM NaCl, 0.5 mM EGTA, 10 mM HEPES, 2 mM Mg-adenosine triphosphate, 0.1 mM Na3-guanosine-5'-triphosphate, 0.1 mM spermine, 2 mM QX-314 bromide, and 10 mM phosphocreatine disodium (adjusted to pH 7.2 with CsOH). The internal solution was used in rectifying the AMPA receptor–mediated transmission experiment. To calculate the rectification of AMPA receptor–mediated EPSCs, we recorded the currents at holding potentials of −65, 0, and +35 mV. Then, the ratio of the peak EPSC amplitudes at negative (−65 mV) and positive (+35 mV) holding potentials was used as a measure of the rectification index. For miniature EPSC (mEPSC) recordings, 0.5 μM tetrodotoxin was added to the perfusion solution. Picrotoxin (100 μM) was present in all of the experiments to block γ-aminobutyric acid (GABA) type A receptor–mediated inhibitory synaptic currents. Voltage errors resulting from series resistance were compensated off-line for voltage clamp recordings and online for current-clamp recordings using a bridge circuit. The neuron signals, which were amplified using an Axoclamp-700B amplifier (bandwidth filter set at 10 kHz for current clamp and 1 kHz for voltage clamp recordings) were digitized and sampled at 50 μs intervals (Digidata 1440A, pClamp 10.1; Molecular Devices, USA). The series resistance used was 12 to 25 MΩ and was monitored throughout the experiment. Data were discarded when the access resistance changed by more than 15% during the experiment.5
Field Potential Recordings
Local field potentials were recorded using ACSF-filled glass microelectrodes (tip diameter: 2 to 4 μm; resistance: 3 to 5 MΩ) that were placed into layer II/III of the ACC. The data were collected using a MultiClamp 700B amplifier and were low-pass filtered at 2 kHz. Kainate (Sigma Life Science, Catalog #K0250) was bath-applied for 20 min to stabilize gamma oscillations before recording. The original local field potentials were digitally bandpass filtered at 30 to 80 Hz. The gamma power (root mean square) of the filtered signal was calculated by a sliding 25-ms window every 1 ms, and 2 SDs above the background mean power was used as the threshold for detecting gamma episodes.26 Peak frequency and power values were obtained from power spectra generated using Fourier analysis in MATLAB software (MathWorks, USA).
Lentivirus Generation and Injection
The oligonucleotides were taken directly out of a polymerase chain reaction machine, denatured for 10 min at 95°C and annealed at room temperature for more than 30 min. Paired oligonucleotides and the pLL3.7 vector, which was digested with XhoI and HpaI restriction enzymes, were ligated with T4 ligase at 16°C overnight. The ligation product was transformed into stable competent cells. The plasmid containing short hairpin RNA (shRNA)-Gjd2 (gap junction protein, delta 2) was sequenced. pLL3.7-shRNA-Gjd2 was transfected into 293T cells together with Δ8.9 and vesicular stomatitis virus G-protein to package virus particles with Ca2+ reagents. In brief, pLL3.7-shRNA-Gjd2 (10 μg)/Δ8.9 (8 μg)/vesicular stomatitis virus G-protein (6 μg) was mixed with 94 μl CaCl2 (2 M), and the volume was adjusted to 750 μl. Then, the mixture was added to 750 μl 2 × Hank’s Balanced Salt Solution, allowed to sit for 5 min, and then added to 293T cells pretreated with chloroquine (5 μm). This mixture was cultured, with the medium changed to fresh medium every 8 h. After 48 h of culturing, the supernatant was harvested, centrifuged at 24,000 rpm, and discarded. The pellets were virus particles. The viruses were titrated by serial dilution and used to infect 293T cells.
For viral injection, the rats were each anesthetized with pentobarbital sodium and placed in a stereotactic frame. The rats were injected bilaterally with 1 to 1.5 μl of virus into the ACC (2.7 mm anterior from bregma, ± 0.6 mm lateral to midline, and 2.0 mm ventral from the dura; 0.4 mm anterior from bregma, ± 0.6 mm lateral to midline, and 2.0 mm ventral from the dura) using glass microelectrodes at a slow rate (100 to 150 nl/min). The injection microelectrode was slowly withdrawn 5 min after the virus infusion. To calculate transfection efficiency of lentivirus by a green fluorescent protein (GFP) reporter gene, we used ImageJ software to quantify the positive cells of GFP (transfection efficiency = the number of the GFP-positive cells / the number of the 4′,6-diamidino-2-phenylindole–positive cells). All pictures were captured with identical settings. Slices from five rats were quantified, one slice per rat.
Synthesis of the Meflo Isomers
Mefloquine hydrochloric acid (200 g, 0.48 mM) was suspended in 1 M of NaOH solution (2 l) in a separatory funnel. Ethyl acetate (1 l) was added, and the layer was separated after shaking. The aqueous layer was extracted using ethyl acrylate (EA), and the combined organic layers were dried over Na2SO4. A solution of l-O,O-di-p-toluoyl tartaric acid (186 g, 0.48 mM) was added to a solution of mefloquine in EA (5 l). The resulting solution was stoppered and stirred at room temperature overnight. The resulting solid was filtered and washed using EA (4 l) to yield (−)-(11R,12S)-mefloquine l-O,O-di-p-toluoyl tartaric acid salt that was then triturated with boiling EA (4 l) and dried to obtain the solid. The solid (154 g) was suspended in 500 ml methyl alcohol, and 1.25 l of 1 M NaOH was added, followed by 1.25 l water. The mixture was stirred for 1 h. The solid was filtered off, washed by 2 l water, and dried in vacuo to obtain a white solid (74.6 g).
The mother liquor was reduced to dryness and neutralized with a 1 M NaOH solution (1.5 l). After drying over Na2SO4, d-O,O-di-p-toluoyl tartaric acid (111 g, 0.288 mM) in EA (5 l) was added. Then, the solution was stirred at room temperature and left overnight, generating a white solid, which was filtered and washed in EA (2 l) to yield (+)-(11S,12R)-mefloquine d-O,O-di-p-toluoyl tartaric acid salt. This solid was suspended in 500 ml methyl alcohol, and 1.25 l of 1 M NaOH was added, followed by 1.25 l water. The mixture was stirred for 1 h. The solid was filtered off and washed with 2 l of water and dried in vacuo to obtain a white solid (81 g).
The data are expressed as the means ± SEM. A one-way ANOVA with Tukey multiple comparison post hoc tests were used when more than two groups of data were compared, or Student’s t test was used when only two groups were compared. The brain slices were treated as though they were independent of one another when analyzing the electrophysiology data. Behavioral data were analyzed by using a two-way repeated-measures ANOVA that group as a between-groups factor and time as a repeated-measures factor, followed by the Bonferroni post hoc test. SigmaStat 3.5 (Jandel Scientific Software, USA) was used to carry out statistical analyses. The value of the criterion for statistical significance was P less than 0.05 (two-tailed test).
Cx36 Up-regulation and Gamma Oscillation Activation in the ACC after Nerve Injury
The ACC is a key cortical region engaged in neuropathic pain formation.5,27 To explore the role of electrical synapses in the pain pathway, we examined the biochemical and electrophysiologic changes in the ACC after nerve injury. Cx36 expression in the ACC was observed on days 3 and 7 after CCI in our study. Both Western blotting and immunohistochemistry assay showed that Cx36 expression in the ACC was significantly up-regulated 7 days after CCI (fig. 1, A and B; Western blotting in the sham group, 0.78 ± 0.08, n = 5; Western blotting 3 days after CCI, 0.90 ± 0.13, n = 5; Western blotting 7 days after CCI, 1.18 ± 0.20; one-way ANOVA, P = 0.005). To investigate whether the changes in Cx36 expression were a specific phenomenon found only in the ACC, we also examined Cx36 expression levels in an adjacent cortex (i.e., the primary motor cortex) and in a pain-associated area (i.e., the hippocampus). These data demonstrated that Cx36 expression in the primary motor cortex (fig. 1, C and D) and in the hippocampus (fig. 1, E and F) did not change significantly 7 days after CCI.
Cx36 was reported to be responsible for generating gamma-frequency neuronal oscillations.11 Given the importance of gamma oscillations in the perception of pain, we further assessed the consequences of the increase in Cx36 expression on gamma oscillations in the ACC. Previous studies have shown that kainite application resulted in persistent gamma oscillations in the hippocampus and in the entorhinal cortex in vitro.11,13 Similarly, we used ACC slice preparations, and the typical pattern of gamma oscillations was evoked in the presence of 150 nM kainate (fig. 2A). After CCI, the peak frequency of gamma oscillations remained unaffected (fig. 2B). However, a significant increase in the power of gamma oscillatory activity remained 7 days after CCI compared with those values from sham rats (fig. 2C; sham group, 5.09 ± 1.04 × 10−12 V2/Hz, n = 7 slices of four rats; 3 days after CCI, 7.54 ± 1.12 × 10−12 V2/Hz, n = 7 slices of four rats; 7 days after CCI, 11.29 ± 1.49 × 10−12 V2/Hz; one-way ANOVA, P = 0.006, n = 8 slices of four rats). Thus, both Cx36 expression and gamma oscillation power in the ACC increased after CCI.
Disruption of Cx36 Expression in the ACC Attenuates Nerve Injury–induced Neuropathic Pain
To further examine whether Cx36 up-regulation and gamma oscillation activation contributed to neuropathic pain formation, we investigated the relation between electrical synapses and neuropathic pain by altering Cx36 gene expression. Given that changes in motor and cognitive abilities were observed in Cx36 knockout mice,28 we used the local RNA interference (RNAi) technique. By using the Cx36 gene sequences as references, three RNAi sequences were designed to establish a lentiviral vector system to infect shRNA-Cx36 (fig. 3A). shRNA-Cx36 and shRNA-control were microinjected into the ACC before the CCI surgery (fig. 3B). The transfection efficiency of lentivirus was 69.8 ± 2.23% (n = 5). Immunohistochemistry (fig. 3C) and Western blotting (fig. 3D) results indicated that the shRNA-Cx36 group experienced a significant reduction in Cx36 expression in the ACC after CCI compared with the rats in the shRNA-control (control) group (Western blotting in control group, 0.76 ± 0.07, n = 5; Western blotting in shRNA-Cx36 group, 1.26 ± 0.09, n = 5; Student’s t test, P < 0.001). The behavioral test further demonstrated that both the paw withdrawal threshold and the paw withdrawal latency significantly increased in the rats in the shRNA-Cx36 group (fig. 3, E and F).
Furthermore, we interfered with the electrical coupling in the ACC using a microinjection of electrical synapse blockers. Both the selective Cx36 blocker mefloquine (Meflo) and the general gap junction blocker Cbx were used in the current study.29 We microinjected Meflo and Cbx locally into the ACC immediately after CCI, and the administration continued for the following 4 days (fig. 4, A and B). Pain behaviors were investigated on days 3, 5, 7, 14, and 21 after CCI. We found that both Meflo (100 mM, 1.6 μl) and Cbx (100 mM, 1.6 μl) were successful in attenuating thermal hyperalgesia and mechanical allodynia when injected locally into the ACC (fig. 4, C–F). Collectively, these data suggest that Cx36 up-regulation in the ACC may be responsible for neuropathic pain formation after CCI.
Systemic Administration of Meflo Attenuates Nerve Injury–induced Neuropathic Pain
Given that the method of local microinjection is not applicable for clinical practice, we further explored whether systemic administration of Meflo can relieve neuropathic pain. Following the same protocol of ACC microinjection, Meflo was intraperitoneal administered to the rats for 5 days (fig. 5A). We found that Meflo (20 and 30 mg/kg) was able to significantly increase the paw withdrawal threshold to mechanical stimuli and the paw withdrawal latency to heat stimuli compared with the vehicle 3 and 21 days after CCI (fig. 5B). Furthermore, this electrical synapses blocker relieved the chronic pain induced by CCI in a dose-dependent manner. To exclude the possibility that the electrical synapse blocker can affect normal pain perception in rats, Meflo was intraperitoneal administered to animals after sham surgery for 5 days consecutively (fig. 5C). We found that Meflo administration did not affect the paw withdrawal threshold to mechanical stimulus (fig. 5D) or the paw withdrawal latency to heat stimuli (fig. 5E). The analgesic effects of the electrical synapse blocker in the current study may be caused by altered rat locomotion. Thus, we performed an open-field test and a rotarod test to exclude the influence of Meflo on the motor functions of the rats. The results demonstrated that Meflo, when administered consecutively for 5 days, did not change the total distance traveled by the rats during the 15-min recording time in the open field (fig. 5F) or the rotarod latency time (fig. 5G). These results demonstrated that Meflo relieved neuropathic pain in the CCI animal model.
The SNI model was also introduced to investigate whether the effectiveness of the electrical synapse blocker on neuropathic pain was specific to the CCI model (fig. 5A). We discovered that Meflo (fig. 5H) administration also significantly attenuated mechanical allodynia in the SNI animal model. In summary, we were able to confirm that the systemic administration of Meflo at the formation stage of neuropathic pain can attenuate pain levels in a sustained manner.
Systemic Administration of Meflo Decreases Gamma Oscillation Power and Synaptic Transmission in the ACC
Gamma oscillations in the pain-associated cortex are particularly related to the experience of pain.14 To further confirm the connection between gamma oscillations and neuropathic pain, we examined the effects of Meflo on gamma oscillations in the ACC. The gamma-frequency oscillatory activity in ACC slices was observed after Meflo administration (fig. 6A). The field recordings demonstrated that Meflo application had no effect on the peak frequency of gamma oscillations (fig. 6B). However, we found that the activation of gamma-frequency oscillatory activity after CCI was almost completely reversed by the selective Cx36 blocker Meflo (fig. 6, A and C; sham group, 5.09 ± 1.04 × 10−12 V2/Hz, n = 7 slices of four rats; CCI group, 11.29 ± 1.49 × 10−12 V2/Hz, n = 8 slices of four rats; Meflo group, 5.11 ± 0.53 × 10−12 V2/Hz, n = 8 slices of four rats; one-way ANOVA, P = 0.002). Thus, gamma oscillation activation in the ACC after CCI may contribute to neuropathic pain generation and Meflo may relieve the pain by reversing the gamma oscillation activation.
Studies have also shown that ACC synaptic plasticity is responsible for neuropathic pain generation.20 To verify whether Cx36 contributes to the synaptic transmission potentiation in the ACC after nerve injury, we also used brain slices to investigate the probability of glutamate release from the presynaptic membranes and AMPA receptor–mediated responses in the postsynaptic membranes of ACC neurons. Consistent with previous studies,5,30 significant enhancements were found in the amplitude (P = 0.0022) and frequency (P = 0.0159) of mEPSCs in the ACC neurons after CCI (d 7 to 14) (fig. 6, D–F and table 1; sham group, n = 14 neurons, four rats; CCI group, n = 15 neurons, four rats, one-way ANOVA). Furthermore, the amplitude (P = 0.0005) and frequency (P = 0.0018) of mEPSCs significantly decreased in the Meflo group (consecutive Meflo intraperitoneal injection for 5 days after CCI and performed on days 7 to 14) compared with the CCI group (fig. 6, D–F and table 1; Meflo group, n = 16 neurons, five rats, one-way ANOVA).
To further determine whether the presynaptic and postsynaptic transmissions were decreased by Meflo, paired-pulse facilitation (PPF) and inward rectification were examined in ACC neurons in the current study. PPF is a form of short-term synaptic plasticity commonly used as a measure of presynaptic function in which the response to the second stimulus is increased as a result of residual calcium in the presynaptic terminal after the first stimulus.31 PPF was observed at stimulus intervals of 30, 60, and 120 ms in this study. A significant reduction in PPF was observed in the ACC neurons after nerve ligation.5 Inward rectification indicated better Ca2+ permission of AMPA receptors and enhanced postsynaptic transmission.32,33 Previous studies have demonstrated that inwardly rectifying properties of AMPA receptors occurred as a result of an alteration to their subunit composition in ACC neurons after nerve injury.5,33,34 We found that both PPF (fig. 6G and table 2; PPF in CCI group, n = 18 neurons, four rats; PPF in Meflo group, n = 19 neurons, five rats, two-way ANOVA, P = 0.018) and inward rectification (fig. 6H and table 1; CCI group, n = 13 neurons, four rats; Meflo group, n = 13 neurons, four rats, one-way ANOVA, P = 0.0465) had significant differences between the Meflo group and the CCI group. However, no significant difference was observed between the Meflo group and the sham group (fig. 6, G and H). These data demonstrated that Meflo reversed potentiated synaptic transmission after nerve injury. Taken together, these results suggested that Meflo normalized gamma oscillation activation and decreased potentiated synaptic transmission in CCI animal model.
Effects of the Isomers of the Antimalarial Drug Meflo on Neuropathic Pain
Meflo is composed of two isomers, (+)-(11S,12R)-Meflo and (−)-(11R,12S)-Meflo (fig. 7A).35 We further examined which isomer was responsible for the pain relief by Meflo injection. The rats were intraperitoneal administered (+)-(11S,12R)-Meflo or (−)-(11R,12S)-Meflo immediately after CCI for 5 days consecutively (fig. 7B). We discovered that (−)-(11R12S)-Meflo (20 mg/kg) was able to significantly decrease the paw withdrawal threshold to mechanical stimuli and the paw withdrawal latency to heat stimuli (fig. 7, C and D). However, (+)-(11S,12R)-Meflo (20 mg/kg) had no effect on neuropathic pain induced by CCI (fig. 7, C and D). Thus, (−)-(11R,12S)-Meflo may have potential as a new therapeutic for neuropathic pain.
The results presented in this study confirm the relation between the electrical synapses of the central nervous system and the formation of neuropathic pain. Evidence of the association between gap junctions and neuropathic pain formation is quite sparse. Previous studies primarily focused on gap junctions in the peripheral nerves and in glial cells. Cx37 mRNA levels have been reported to increase proximally and distally to injured locations in rat sciatic nerves after injury, and this increase lasted for 21 days.36 Several studies have demonstrated that a close association exists between interglial connexin (Cx43) and neuropathic pain formation. An orofacial neuropathic pain model was used to prove that the selective silencing of Cx43 expression on the dorsal root ganglion of the trigeminal nerve could attenuate neuropathic pain after infraorbital nerve injury.37 Another study reported that heat hyperalgesia and mechanical allodynia induced by spinal cord injury were prevented in transgenic mice with Cx43/Cx30 deletions but were fully developed in transgenic mice with only the Cx30 deletion.38 Whether electrical synapses contributed to neuropathic pain formation remains unclear.
In the current study, we demonstrated that systemic application of Meflo relieved neuropathic pain by both behavioral tests and electrophysiologic experiments. Meflo was invented at the Walter Reed Army Institute of Research in the 1970s and was marketed as Lariam and Mephaquine. Meflo has been used as an antimalarial drug since its approval by Food and Drug Administration in 1989 for systemic oral use in humans; thus, its safety has been well established. Meflo is effective on various types of human and animal malaria and can control Plasmodium falciparum that are resistant to chloroquine.39 In summary, the current study attempted to provide evidence regarding the new clinical indications for this drug. Moreover, we found that (−)-(11R, 12S)-Meflo, which is a Meflo isomer, is promising as a new drug candidate for neuropathic pain. However, its effectiveness and the adverse effects under the effective dosage must be confirmed in further clinical studies. Once established after nerve injury, the symptoms of neuropathic pain are often resistant to conventional approaches of pain treatment. Thus, we applied the electrical synapse blocker during the early phase of neuropathic pain development. Our research may provide a new therapeutic strategy that could prevent the development of this disease during the early phase after nerve injury.
Our data demonstrated that Cx36 expression increased in the ACC region after nerve injury. Cx36, which is the major neuronal connexin, has been shown to modulate the gamma-frequency (30 to 80 Hz) oscillations.11 The gamma oscillations in cortical brain regions may be a possible mechanism for various cognitive processes, such as pain perception, attention, and memory formation.14,40,41 However, the involvement of ACC gamma oscillations in nerve injury-induced neuropathic pain has not been elucidated. Using brain slice experiments, we confirmed that ACC gamma oscillations could be evoked in a manner similar to the entorhinal cortex and hippocampus in the presence of nanomolar concentrations of kainate. We found that a significant increase in ACC gamma oscillation power was found after nerve injury, along with the up-regulation of Cx36 and an increase in the symptoms of neuropathic pain. Electrical coupling enhancement between neurons in the ACC might lead to a better synchronization of the functions of Cx36 gap junction channels, resulting in improved precision of action potential timing and of the rhythmicity of oscillation.11 The outcome of these changes would be reflected by gamma oscillation activation in the ACC. We further demonstrated that the selective Cx36 blocker Meflo could reverse the activation of gamma-frequency oscillatory activity and relieve the behavioral sensitization induced by CCI. Thus, ACC gamma oscillations might play a prominent role in the pathogenesis of nerve injury-induced neuropathic pain.
Our present results also provide some specific evidence explaining how Cx36 and gamma oscillations participate in neuropathic pain development. Gamma oscillation has been supposed to facilitate the neuronal communications by the temporal precision of coincident synaptic inputs,40,42,43 which may result in synaptic transmission strengthening. Moreover, potentiated synaptic transmission in the ACC was found to be the key mechanism for neuropathic pain. Thus, given the close link between synaptic transmission and pathological pain, we further demonstrated that the specific electrical synapse blocker Meflo could decrease the chemical excitatory synaptic transmission by electrophysiologic techniques, which might lead to the attenuation of neuropathic pain. Both presynaptic and postsynaptic changes were observed in excitatory synapses within the ACC after CCI. The changes in PPF and in the frequencies of mEPSCs in ACC neurons after nerve injury represent enhanced presynaptic transmitter release. However, after Meflo injection, these alterations were primarily restored. In addition, the postsynaptic functions of excitatory synapses within the ACC, including the amplitude of mEPSCs and the function of AMPA receptors, were also evaluated. Similarly, Meflo prevented the increase in mEPSC amplitudes after CCI. The biophysical properties of AMPA receptors depend on their subunit composition.44,45 Long-term potentiation of synaptic transmission was absent from the hippocampus in the adult GluA1 knockout mouse.46 Because of nerve injury, the AMPA receptor GluR1 subunit increased in the membrane fraction.5 This alteration of subunit composition in ACC neurons after nerve injury led to an increase in the rectification index of AMPA receptor–mediated transmission,5 which may denote better Ca2+ permission of AMPA receptors and potentiation in synaptic transmission. Thus, the rectification index was a reflection of synaptic plasticity. In this study, we identified whether the inwardly rectifying properties of AMPA receptors in ACC neurons after nerve injury could be restored by inhibiting electrical synapses. Our results demonstrated that the specific electrical synapse blocker Meflo could significantly decrease the rectification index of AMPA receptors in ACC neurons after nerve injury, which might result in less Ca2+ permission of the AMPA receptors and decreased synaptic transmission.
In addition, we interfered with Cx36 expression in the ACC region using the RNAi technique and found that neuropathic pain was relieved after blocking Cx36 expression in the ACC region. Therefore, the increased Cx36 expression in the ACC after nerve injury might lead to gamma oscillation activation and then to synaptic transmission potentiation, which might further result in neuropathic pain generation after nerve injury.
A recent study showed decreased Cx36 expression in spinal dorsal horn resulted in the development of neuropathic pain.47 It is very interesting that the change of Cx36 in spinal dorsal horn and ACC is different during the neuropathic pain. According to their results, Cx36 was colocalized with glycinergic neurons but not GABAergic neurons in spinal dorsal horn. However, Ma et al.48 indicated that Cx36 was expressed almost specifically in GABAergic neurons of the adult rat neocortex. Therefore, the function of Cx36 in spinal dorsal horn and neocortex may be different. We speculate that the up-regulation of Cx36 in ACC is a high-level central regulation for peripheral pain. As we discussed in our study, the persistent up-regulation of Cx36 may lead to a potentiation of chemical synapses and result in central sensitization finally. This may be one of the reasons why available treatments for neuropathic pain have limited therapeutic efficacy.
There are also some limitations of the study need to be addressed. We did not conduct studies of neuropathic pain using the oral form of Meflo. For Meflo to have any utility in clinical settings, an oral application is necessary. Intragastric administration of Meflo could ensure the accuracy of the administered dose. However, injuries caused by gavage are not uncommon.49 Moreover, repeated gavage may lead to stress in rats that would affect the function of the ACC. Thus, accurate behavioral data of oral administration could not provide by our study. We found that Meflo could relieve mechanical pain in CCI model for 21 days at least, whereas it could not in SNI model. The two nerve injury models examined in our study differ considerably. Previous study also reported the excitatory synaptic transmission changed differently in CCI and SNI models.50 These facts may account for the different pain relief effects of Meflo on two animal models. Moreover, our future research will try to figure out the distinct effects of Meflo on the synaptic transmission in different animal models. This may further explain how Cx36 involve in the synaptic transmission and neuropathic pain.
The authors thank Qing-Wen Zeng, Ph.D. (Laboratory of Brain Function and Behavior, Institute of Brain Functional Genomics, East China Normal University, Shanghai, China), for his excellent technical support in the field of behavior experiment, and Wen-Hao Hu, Ph.D. (Department of Chemistry, East China Normal University), for the assistance in compound synthesis.
The work was supported by two grants from the National Natural Science Funds of China, Beijing, China (grant nos. 81000787 and 81171169). This work was also supported in part by a grant from the Shanghai Municipal Government, Shanghai, China (grant no. 20114217).
The authors declare no competing interests.