α₂δ-1 Gene Deletion Affects Somatosensory Neuron Function and Delays Mechanical Hypersensitivity in Response to Peripheral Nerve Damage

Abstract

The α₂δ-1 subunit of voltage-gated calcium channels is upregulated after sensory nerve injury and is also the therapeutic target of gabapentinoid drugs. It is therefore likely to play a key role in the development of neuropathic pain. In this study, we have examined mice in which α₂δ-1 gene expression is disrupted, to determine whether α₂δ-1 is involved in various modalities of nociception, and for the development of behavioral hypersensitivity after partial sciatic nerve ligation (PSNL). We find that naive α₂δ-1^−/− mice show a marked behavioral deficit in mechanical and cold sensitivity, but no change in thermal nociception threshold. The lower mechanical sensitivity is mirrored by a reduced in vivo electrophysiological response of dorsal horn wide dynamic range neurons. The Ca_V2.2 level is reduced in brain and spinal cord synaptosomes from α₂δ-1^−/− mice, and α₂δ-1^−/− DRG neurons exhibit lower calcium channel current density. Furthermore, a significantly smaller number of DRG neurons respond to the TRPM8 agonist menthol. After PSNL, α₂δ-1^−/− mice show delayed mechanical hypersensitivity, which only develops at 11 d after surgery, whereas in wild-type littermates it is maximal at the earliest time point measured (3 d). There is no compensatory upregulation of α₂δ-2 or α₂δ-3 after PSNL in α₂δ-1^−/− mice, and other transcripts, including neuropeptide Y and activating transcription factor-3, are upregulated normally. Furthermore, the ability of pregabalin to alleviate mechanical hypersensitivity is lost in PSNL α₂δ-1^−/− mice. Thus, α₂δ-1 is essential for rapid development of mechanical hypersensitivity in a nerve injury model of neuropathic pain.

Introduction

Neuropathic pain can arise after disease or a lesion to the somatosensory system and is characterized by hyperalgesia, allodynia, dysesthesia, paresthesia, and often sensory loss (Costigan et al., 2009). The α₂δ-1 ligands gabapentin and pregabalin are effective in the treatment of a range of chronic neuropathic conditions (Moore et al., 2009, 2011). In several different animal models of neuropathic pain caused by peripheral nerve damage, there are many genes, including ion channels, whose expression is altered in injured dorsal root ganglion (DRG) neurons (Wang et al., 2002; Ji and Strichartz, 2004). In particular, there is an upregulation of the calcium channel auxiliary subunit α₂δ-1 (Newton et al., 2001; Wang et al., 2002). A corresponding increase in expression of α₂δ-1 protein is observed in both the affected ganglia and the spinal cord dorsal horn (Li et al., 2004; Bauer et al., 2009). The increase of α₂δ-1 in the spinal cord is the result of an elevation within the presynaptic terminals of the primary afferent neurons, rather than the intrinsic dorsal horn neurons (Bauer et al., 2009), and coincides with the onset of tactile allodynia (Li et al., 2004). In mice engineered to overexpress α₂δ-1, mimicking this aspect of sensory nerve damage, tactile allodynia is present in the absence of nerve damage (Li et al., 2006). Thus, there is a substantial body of evidence that the elevation of the α₂δ-1 subunit is relevant to the development of neuropathic pain, but it is unknown whether it is essential for this process.

One of the main effects of α₂δ subunits on Ca_V1 and Ca_V2 calcium channels is to increase their functional expression, as evidenced by increased calcium current density, and increased channel protein at the plasma membrane (Gurnett et al., 1996; Qin et al., 1998; Wyatt et al., 1998; Barclay et al., 2001; Klugbauer et al., 2003; Cantí et al., 2005), and to increase transmitter release at presynaptic terminals (Hoppa et al., 2012).

In the present study, we have investigated the function of DRG and spinal neurons from mice with targeted disruption of the α₂δ-1 gene (Fuller-Bicer et al., 2009) to examine whether α₂δ-1 is involved in normal primary afferent function, in terms of thermal and mechanical sensitivity, using behavioral and electrophysiological approaches, and in their response to transient receptor potential (TRP) agonists. An important goal was to determine whether the α₂δ-1 subunit is essential for the initiation and maintenance of behavioral hypersensitivity after sensory nerve injury, to shed light on the molecular mechanism(s) responsible for neuropathic pain. Our main findings are that α₂δ-1^−/− mice show a marked behavioral deficit in mechanical and cold sensitivity, but no change in thermal nociception threshold, together with a reduced sensitivity of their DRG neurons in vitro to TRPM8 agonists. The α₂δ-1^−/− mice also exhibit a marked delay in the development of mechanical hypersensitivity and an insensitivity to pregabalin-mediated analgesia, in a nerve injury model of neuropathic pain. These results identify α₂δ-1 as a key component of the mechanical and cold somatosensory pathways.

Materials and Methods

Transgenic mice.

Conventional α₂δ-1 knock-out C57BL/6 mice were obtained (Fuller-Bicer et al., 2009). All experiments were performed on male 9- to 12-week-old (unless otherwise stated) α₂δ-1^−/− mice and α₂δ-1^+/+ littermates, obtained by breeding from heterozygotes. Mice were housed in groups of no more than 5 on a 12 h:12 h light dark cycle; food and water were available ad libitum. All experimental procedures were approved by the United Kingdom Home Office and followed the guidelines of the International Association for the Study of Pain (Zimmermann, 1983). For the in vivo studies, a total of 62 α₂δ-1^+/+ and 53 α₂δ-1^−/− mice were used. All behavioral and in vivo electrophysiological characterizations were performed with the experimenter blind to the genotype and to drug administered.

Mechanical sensitivity.

Mice were placed in isolation inside Perspex chambers on a wire mesh floor and allowed to acclimatize. Mechanical sensitivity was assessed using von Frey filaments (Touch-Test, North Coast Medical) providing forces of 0.07 g (0.69 mN), 0.16 g (1.57 mN), and 0.4 g (3.92 mN). Hairs were tested in ascending order and applied 10 times across left and right hindpaws. Hairs were applied until they buckled for 3–4 s; lifting, biting, and shaking were considered positive responses. Statistical differences between α₂δ-1^+/+ and α₂δ-1^−/− groups were determined with the Mann–Whitney U test.

Cold sensitivity.

Cold sensitivity was tested by applying a drop of acetone, using a modified 0.5 ml syringe with polythene tubing, to the left and right hindpaws. Acetone was applied 5 times, with at least 5 min recovery between applications. The initial response was disregarded as contact withdrawal. The duration of the delayed response (occurring 2–3 s later) was measured over the next 45 s. Flinching, licking, biting, and shaking were considered to be positive behaviors. Data are presented as mean response duration (in seconds). Statistical differences between α₂δ-1^+/+ and α₂δ ^−/− groups were determined with an unpaired Student's t test.

Withdrawal threshold to noxious heat.

Thermal thresholds were determined using an infrared Nd:YAP laser with a wavelength of 1.34 μm (Electronical Engineering). A He–Ne laser illuminated the area to be stimulated with a 3-mm-wide spot. The pulse time was 4 ms. Mice were lightly anesthetized and maintained on 0.8% v/v isofluorane (Baxter) delivered in a 3:2 ratio of nitrous oxide and oxygen. Body temperature was regulated using a homeothermic blanket (Harvard Apparatus). Reflexes were confirmed before each test by pinching between the toes. The laser was aligned across the footpads of the right hindpaw before applying an incremental thermal stimulus starting at 1 J. If no response was observed, the stimulus was increased by 0.5 J. Mice were allowed 3 min to recover between tests. Once a positive withdrawal was observed, the intensity was reduced by 0.25 J, to determine the approximate withdrawal threshold. Thresholds were confirmed by repeating the test on the left hindpaw. The cutoff was set at 3.5 J to prevent tissue damage. Statistical significances were determined using the Mann–Whitney U test.

In vivo electrophysiology.

In vivo electrophysiology was conducted as previously described (Urch and Dickenson, 2003). Mice were anesthetized with intraperitoneal 2.4 g/kg urethane (Sigma) dissolved in 0.9% saline. Once mice were areflexic, a tracheotomy was performed before securing mice in a stereotaxic frame and performing a laminectomy to expose L3-L5 segments of the spinal cord. Extracellular recordings were made from deep dorsal horn lamina V/VI wide dynamic range (WDR) neurons, at depths of 350–700 μm (Watson et al., 2009), with receptive fields on the glabrous skin of the toes, using parylene-coated tungsten electrodes (A-M Systems). Seventeen neurons were characterized from 14 α₂δ-1^+/+ mice. Nineteen neurons were characterized from 15 α₂δ-1^−/− mice.

Electrical stimulation of WDR neurons was delivered transcutaneously via needles inserted into the receptive field. A train of 16 electrical stimuli (2 ms pulses, 0.5 Hz) was applied at three times the threshold current for C fiber activation. Responses evoked by A (0–50 ms) and C (50–250 ms) fibers were separated and quantified on the basis of latency. Neuronal responses occurring after the C fiber latency band were classed as after discharge. The “input” (a theoretical nonpotentiated response) and the windup (potentiated response) were calculated as follows:

The receptive field was also stimulated using a range of natural stimuli (dynamic brushing, von Frey filaments, Touch-Test, North Coast Medical): 0.4 g (3.92 mN), 1 g (9.81 mN), 4 g (39.23 mN), 8 g (78.45 mN), and 15 g (147.10 mN) and thermal stimuli at 42°C, 45°C, and 48°C applied over a period of 10 s per stimulus and the evoked response quantified. The thermal stimulus was applied with a constant water jet onto the center of the receptive field. The 35°C response was subtracted as it was considered to be predominantly the mechanical component of the stimulus. An acetone drop was applied as an innocuous cooling stimulus, preceded by a water drop to control for concomitant mechanical stimulation during application. Data were captured and analyzed by a Cambridge Electronic Design 1401 interface coupled to Spike 2 software with after stimulus time histogram and rate functions. Statistical differences in fiber threshold, electrical parameters, neuronal responses to cooling, and dynamic brush stimulation were determined using an unpaired Student's t test; Welch's correction was applied where appropriate. Linear regression was performed to examine the difference in the rate of windup after a log transformation. Differences in mechanical and thermal coding were determined using a two-way ANOVA, followed by a Bonferroni correction for multiple paired comparisons. Sphericity was tested with Mauchly's test; Greenhouse-Geiser corrections were used where required.

Partial sciatic nerve ligation (PSNL).

Surgery was performed based on a method described previously (Selzer et al., 1990). Mice were maintained under 2% v/v isofluorane (Baxter) anesthesia delivered in a 3:2 ratio of nitrous oxide and oxygen. Under aseptic conditions, the left sciatic nerve was exposed through blunt dissection of the biceps femoris above the trifurcation of the nerve. Approximately half of the nerve was ligated with a nonabsorbable 7-0 braided silk thread (Ethicon, VetTech). The surrounding muscle was closed with absorbable 6-0 vicryl sutures (Ethicon, VetTech), and the skin with surgical wound clips. Sham surgery was performed in wild-type mice in an identical manner omitting the nerve ligation step. After surgery, the mice were allowed to recover. Foot posture and general behavior of the operated mice were monitored throughout the postoperative period. Two mice with impaired motor activity were omitted from this study. Mechanical and cold hypersensitivity were tested on postoperative days 3, 7, 9, 11, and 14. Mechanical hypersensitivity was compared with the Kruskal–Wallis test, and paired comparisons were made with Dunn's post hoc test followed by a Bonferroni correction. Cold hypersensitivity was compared with a two-way repeated-measures ANOVA, followed by a Bonferroni correction for multiple paired comparisons. Behavioral pharmacology was performed 14 d after injury, with the observer blinded to the treatment. α₂δ-1^+/+ and α₂δ-1^−/− mice were intraperitoneally administered either vehicle (0.9% saline) or 10 mg/kg pregabalin (gift from Pfizer) in a volume of 5 ml/kg. Mechanical hypersensitivity was tested at 30, 60, and 90 min after dosing. Differences to predrug withdrawal frequencies were compared with the Friedman test, followed by a Wilcoxon post hoc and Bonferroni correction for paired comparisons. Differences between vehicle and pregabalin groups were compared with the Mann–Whitney test.

DRG cultures.

DRGs were removed from the entire spine of α₂δ-1^+/+ and α₂δ-1^−/− mice at age 10 weeks, according to Schedule 1 guidelines (Home Office Animals Scientific Procedures Act 1986, United Kingdom). Cell cultures were obtained after enzymatic and mechanical dispersal as described previously (Hendrich et al., 2008). For Ca²⁺-imaging experiments, DRGs were incubated in Hank's basal salt solution containing 1000 U/ml DNase (Invitrogen), 5 mg/ml dispase (Invitrogen), and 2 mg/ml collagenase type 1A (Sigma) for 18 min at 37°C. The partially digested ganglia were then washed and triturated in growth medium (DMEM/F12 with 10% FBS), 2 mm GlutaMAX (Invitrogen), 100 U/ml penicillin, 100 μg/ml streptomycin, and 6 g/L glucose and plated on coverslips coated with poly-l-lysine (Sigma) and laminin (Sigma). For electrophysiological experiments, DRGs were incubated in Hank's basal salt solution containing 100 U/ml DNase, 5 mg/ml dispase, and 2 mg/ml collagenase type 1A for 30 min at 37°C, the growth medium had no glucose added, and laminin was not used as a substrate.

In vitro electrophysiology.

Calcium channel currents in DRG neurons were investigated by whole-cell patch-clamp recording (after 1 DIV). The patch pipette solution contained in mm the following: 140 Cs-aspartate, 5 EGTA, 2 MgCl₂, 0.1 CaCl₂, 2 ATP, 20 HEPES, pH 7.2, 310 mOsm with sucrose. The external solution for recording Ba²⁺ currents contained in mm as follows: 150 tetraethyl-ammonium Br, 3 KCl, 1.0 NaHCO₃, 1.0 MgCl₂, 10 HEPES, 4 glucose, 2 BaCl₂, 0.001 TTX, pH 7.4, 320 mOsm with sucrose. Measurements were made at 30 ms after the start of the voltage step. Analysis was performed as previously described (Cantí et al., 2005). Mean I-V relationships were fitted with a modified Boltzmann equation as follows:

where G_max is the maximum conductance, V_rev is the reversal potential, k is the slope factor, and V_50,act is the voltage for 50% current activation. In experiments in which N-type calcium currents were blocked, cells were preincubated for 15 min in DMEM/F12 (0% FBS) with ω-conotoxin GVIA (1 μm), and recorded in the external solution in the presence of the toxin. I-V relationships recorded in cells from α₂δ-1^+/+ and α₂δ-1^−/− mice, with or without ω-conotoxin GVIA, were compared with two-way ANOVA, followed by post hoc pairwise comparison with Tukey Honest Significant Differences test.

Intracellular calcium imaging.

Cultures of DRG neurons (1 DIV) were loaded with 5 μm fura-2 AM in growth medium for 15 min and then washed in external solution (145 mm NaCl, 2 mm KCl, 5 mm NaHCO₃, 2.5 mm CaCl₂, 1 mm MgCl₂, 10 mm HEPES, 10 mm glucose, pH 7.4, adjusted with NaOH) for 20 min at 37°C. Imaging was performed using an Axiovert 200M microscope (Zeiss) with an ORCA-ER camera (Hamamatsu) at 30°C. Images were captured sequentially (0.5 Hz acquisition rate per paired image) at 340 and 380 nm excitation and 510 nm emission, and the ratio (340/380 nm) was calculated offline for subsequent analysis using Volocity 4 software (Improvision). After a 30 s baseline acquisition, agonists were applied in external solution sequentially in two different experimental series. In the first series, the application order was as follows: 100 μm allyl isothiocyanate (mustard oil [MO], 500 mm stock in DMSO, Sigma) for 20 s, followed by 500 μm (1R, 2S, 5R)-(−)-menthol (ME, 500 mm stock in EtOH, Sigma) for 30 s, followed by 1 μm capsaicin (C or CAPS, 10 mm stock in EtOH, Sigma) for 10 s. In the second series, the application order was as follows: 250 μm ME for 30 s, followed by 100 μm trans-cinnamaldehyde (CA, 500 mm in EtOH, Sigma) for 60 s, followed by 1 μm CAPS for 5 s. Agonist applications were separated by 90 s of washing in external solution. In a separate set of experiments, cells were stimulated with 20 μm WS-12 (20 mm stock in DMSO, Sigma) for 30 s followed by 250 μm ME for 30 s with a 3 min washing period between agonist applications. The elevation of intracellular Ca²⁺ in DRG neurons induced by a 10 s high K⁺ application (85 mm NaCl, 60 mm KCl, 5 mm NaHCO₃, 2.5 mm CaCl₂, 1 mm MgCl₂, 10 mm HEPES, 10 mm glucose, pH 7.4, adjusted with NaOH) 90 s after the last agonist application was used as an indicator of cell viability. Cells not responding to high K⁺ were excluded from the analysis. To quantify, the number of responses for each agonist per coverslip were counted and normalized to the number of cells responding to K⁺ (100%). Responses to ME and CAPS were pooled from experiments using both agonist sequences 1 and 2 (see above). Responses to MO (sequence 1) and CA (sequence 2) were analyzed separately. In total, 982 α₂δ-1^+/+ and 1151 α₂δ-1^−/− cells were analyzed. Peak ratio amplitudes were determined relative to a normalized fluorescence ratio (R = 1.0) measured before agonist application.

Quantitative PCR.

DRGs from the lumbar region 3–5 (L3-L5) were dissected as described previously (Rigaud et al., 2008), from nine sham α₂δ-1^+/+, 7 PSNL α₂δ-1^+/+ and 7 PSNL α₂δ-1^−/− mice, 14 d after surgery from both the ipsilateral and contralateral sides. DRGs were also taken from three α₂δ-1^+/+ and three α₂δ-1^−/− naive mice in the same manner. For quantitative PCR (Q-PCR) and immunoblot assays, the methods used were essentially as described previously (Bauer et al., 2009). Briefly, RNA was extracted from pooled L3-L5 pulverized frozen DRGs using RNeasy columns (QIAGEN), including an on-column DNase step. Reverse transcription was performed using Superscript III reverse transcriptase (Invitrogen) and random hexamer primers (Promega). Q-PCR of sample duplicates was performed with an Applied Biosystems 7500/7500 Fast Real-Time PCR system. The following TaqMan assays with TaqMan Gene Expression Master Mix were used according to the manufacturer's protocol (gene name: assay ID): GAPDH: Mm99999915_g1; CACNA2D1: Mm00486607_m1; CACNA2D2: Mm00457825_m1; CACNA2D3: Mm00486613_m1; ATF3: Mm00476032_m1; TRPV1: Mm01246302_m1; TRPM8: Mm01299593_m1; TRPA1: Mm01227437_m1; PIEZO2: Mm01265861_m1; and CACNB3: Mm00432244_m1.

The comparative C_T (ΔΔC_T) method was used for relative quantification of fold differences (given as the mean ± SEM) of mRNA levels between the ipsilateral and contralateral side from mice that had undergone surgery. For this, data were first normalized for expression of GAPDH mRNA by calculating ΔC_T. Statistical significance between α₂δ-1, α₂δ-2, α₂δ-3, and activating transcription factor 3 (ATF3) and Ca_Vβ3 mRNA levels in sham and PSNL or α₂δ-1^+/+ and α₂δ-1^−/− mice was determined by a one-way ANOVA test with Bonferroni's post hoc test. Statistical significance of differences for TRPM8, TRPA1, TRPV1, and Piezo2 mRNA levels between α₂δ-1^+/+ and α₂δ-1^−/− mice was determined by a one-sample t test (compared with a theoretical value of 1) and by unpaired Student's t test. The ΔΔC_T method was also used for relative quantitation of fold differences of TRP channel mRNA levels from naive mice. The mean value of the left and right DRG samples of each α₂δ-1^−/− and ^+/+ mouse was normalized to the mean value of the α₂δ-1^+/+ samples, and statistical analysis was determined by Student's unpaired t test.

Transfection of tsA-201 cells.

Cells were transfected with cDNA for N-terminal c-myc tagged mouse TRPM8, TRPV1, or TRPA1, together with either α₂δ-1 or β1b, at a ratio of 1:1 using FuGENE according to the manufacturer's protocol (Promega). Myc-TRPM8 was in pcDNA5/FRT, myc-TRPV1 and myc-TRPA1 were in pCMV-myc, and α₂δ-1 and β1b were in pMT2.

Immunoblotting.

Proteins were extracted from DRG neurons, brain, and skeletal muscle from hind leg, using a buffer of PBS with 1 mm EDTA and 1 μm DTT in the presence of protease inhibitors (complete EDTA free, Roche Diagnostics) by homogenization, followed by incubation with 0.2% SDS and 1% Igepal for 40 min on ice. Synaptosomes were prepared from whole brain and spinal cord as previously described (Kato et al., 2007) and solubilized in 50 mm Tris, pH 8, 150 mm NaCl, 1% Igepal, 0.5% Na deoxycholate, 0.1% SDS (Ferron et al., 2008). DRG samples were precleared of IgGs by incubation with Protein-G-agarose beads (Santa Cruz Biotechnology) for 1 h at 4°C, and then deglycosylated with 0.1 U/μl endoglycosidase F (PNGase F, Roche) for 1.5 h at 37°C. After centrifugation (20 min at 13,000 × g at 4°C), samples were loaded onto a 3–8% NuPage Tris/acetate gel (Invitrogen) and proteins were separated by SDS-PAGE. Proteins were transferred to PVDF membranes (Bio-Rad) and, after blocking (500 mm NaCl, 10 mm Tris, pH 7.4, with 3% BSA and 0.5% Igepal), blots were probed with the mouse monoclonal anti-dihydropyridine receptor (α₂-1 subunit) antibody (α₂-1 Ab, 1:2000, Sigma) or in-house Ca_V2.2 polyclonal Ab (1:500; Raghib et al., 2001) or with the mouse monoclonal anti-GAPDH Ab (1:25,000, Ambion) at 4°C overnight. Similarly, for the detection of Myc-tagged TRP channels, in transiently transfected tsA-201 cells, 60 μg of WCL was separated by SDS-PAGE, transferred to PVDF membrane, and probed with the mouse monoclonal anti-Myc Ab (1:1000, Santa Cruz Biotechnology), or with the mouse monoclonal anti-GAPDH Ab (1:25,000, Ambion) for 1 h. The protein–Ab complexes were then labeled with a HRP-conjugated secondary Ab (1:2000, Sigma) for 1 h at room temperature (RT). Bands were detected using the enhanced ECL Plus reagent (GE Healthcare) visualized with a Typhoon 9410 scanner (GE Healthcare). Densitometric quantification of immunoblot bands was performed with ImageQuant software (GE Healthcare) and normalized to the respective GAPDH loading control.

Immunohistochemistry.

L3-L5 DRGs were removed and postfixed for 2 h in 0.1 m phosphate buffer (PB), pH 7.4, with 4% PFA, followed by incubation in PB with 15% sucrose and 0.05% NaN₃ overnight, and finally frozen in optimal cutting temperature (OCT) compound (VWR) before sectioning at 10 μm using a cryostat. Immunofluorescence labeling for α₂δ-1 (mouse monoclonal anti-dihydropyridine receptor Ab (α₂-1 subunit), 1:100, Sigma) and neuropeptide Y (NPY, rabbit polyclonal Ab, 1:1000,Abcam) was performed as described previously (Bauer et al., 2009) with the following alterations. After heat-induced epitope retrieval (10 mm citrate buffer, pH 6.0, 0.05% Tween 20, 95°C for 10 min), the sections were incubated in 1% Triton X-100 in PBS for 1 h at RT and blocking with 10% goat serum in PBS containing 0.1% Triton X-100 for >1 h at RT; and to reduce the background signal resulting from the use of a mouse Ab on mouse tissue, endogenous IgGs were blocked by incubation with the unconjugated goat Fab anti-mouse IgG (H+L) (0.1 mg/ml in PBS, Jackson ImmunoResearch Laboratories) for 1 h at RT. Primary Abs were applied for 2 or 3 d at 4°C. After extensive washes, the samples were incubated with biotin-conjugated goat Fab fragment anti mouse for 2 h at 4°C (1:500, Jackson ImmunoResearch Laboratories), followed by washes and streptavidin-AlexaFluor-488 and anti-rabbit-AlexaFluor-594 for 1 h at RT (both at 1:500, Invitrogen). Nuclei were stained with DAPI (500 nm) before mounting in Vectashield (Vector Laboratories). Staining with the neuronal markers peripherin (mouse monoclonal Ab, 1:500, Sigma) and neurofilament 200 (NF200, rabbit polyclonal Ab, 1:200, Sigma) was performed on sections of L4 DRGs, which were either washed with PB, followed by incubation in PB with 15% sucrose and 0.05% NaN₃ overnight, and finally frozen in OCT or immediately frozen in OCT before being sectioned to a thickness of 10 μm. After fixation for 7 min with 4% PFA in PBS, sections were blocked in 10% goat serum in the presence of 0.3% Triton X-100 for at least 2 h. Primary Abs were applied at 4°C overnight in blocking solution. Sections were then washed with PBS + 0.3% Triton X-100, incubated with biotin-conjugated goat anti-rabbit IgG (1:500, Sigma) in blocking buffer for 2 h at RT, washed with PBS + 0.3% Triton X-100 before incubation with anti-mouse FITC (1:500, Sigma) and streptavidin-AlexaFluor-594 (1:500, Invitrogen) for 1 h. After washing, sections were stained with DAPI (500 nm) before the samples were mounted in VectaShield (Vector Laboratories).

Image acquisition and analysis.

Immunofluorescence labeling of peripherin and NF200 was detected with a Zeiss Axiovert 200M microscope. Images of 14 sections from four α₂δ-1^+/+ and 13 sections from three α₂δ-1^−/− mice were analyzed. For this, the percentage of cells positive for NF200, peripherin, or both neuronal markers was calculated per section. Only intact cells with a clearly visible nucleus were counted. Statistical analysis (Student's unpaired t test) was performed after arcsin transformation of the data. Immunostaining for α₂δ-1 and NPY was visualized using an LSM 510 Meta (Zeiss) confocal microscope. To quantify the NPY staining, images were acquired with constant settings from at least 8 sections (with a minimum distance between sections from the same DRG of 20 μm) from at least three mice. Only cells with a visible nucleus were used for analysis. Images were converted to gray levels using ImageJ software (http://rsb.info.nih.gov/ij) and a threshold determined by using images taken from negative control sections were the primary Ab was omitted during the immunohistochemical procedure. Cells with a gray level intensity higher than the threshold were counted as NPY-positive cells, and the number of NPY-positive cells per section was established. Statistical significance was determined with Student's unpaired t test.

Immunohistochemistry on skin.

Skin biopsies were taken from the left hindpaw of 4 α₂δ-1^+/+ and 4 α₂δ-1^−/− mice. Skin samples were fixed using 4% PFA in 0.1 m PB and then preserved in 15% sucrose before being blocked and further processed into 14 μm sections. Nerve fibers were stained using rabbit anti-PGP 9.5 Ab (1:2000, Ultraclone) and Cy3 anti-rabbit Ab (1:500, Jackson ImmunoResearch Laboratories). Images were acquired using a Zeiss LSM 710 upright confocal microscope, with an EC Plan-Neofluar objective at 40× magnification. The 2 μm interval images were acquired for each section, and the maximum intensity projection tool was used. The samples were processed and analyzed blind with respect to the genotype. The samples were analyzed by the same investigator (J.D.R.), and analysis was performed as explained previously (Ramirez et al., 2012). Briefly, PGP9.5-positive nerve fibers crossing the basal membrane were counted, and a measurement of the length of the sample was then obtained. Intraepidermal nerve fiber density counts are given in number of fibers per millimeter length of epidermis.

Statistics.

Statistical analyses were performed using GraphPad Prism 4.03, SPSS version 20 (IBM) or R (R Development Core Team, 2012). For all statistical tests: ***p < 0.001, **p < 0.01, *p < 0.05, not significant p > 0.05, except where actual p values given.

Results

α₂δ-1 protein is not expressed in tissue from α₂δ-1 knock-out mice

We confirmed that α₂δ-1 protein was absent from brain and skeletal muscle of α₂δ-1^−/− mice (Fig. 1A). In L4 + L5 DRGs, α₂δ-1 was present in +/+ mice, and absent from α₂δ-1^−/− littermates, a result confirmed by deglycosylation of the samples, where the expected reduction in MW of the α₂δ-1-immunoreactive band was seen in +/+ DRGs from ∼150 kDa to ∼105 kDa (Fig. 1B, lanes 2 and 4). In brain synaptosomes from α₂δ-1^−/− mice at 10 weeks old, we also found a reduced level of Ca_V2.2 protein compared with littermate α₂δ-1^+/+ tissue (a 25.4 ± 5.8% reduction from four independent experiments, p = 0.022), although this was not evident in whole brain (Fig. 1C). This result is indicative of reduced presynaptic levels of this channel in the absence of α₂δ-1. A similar result was obtained in spinal cord synaptosomes, in which there was a reduction of Ca_V2.2 protein in α₂δ-1^−/− compared with α₂δ-1^+/+ tissue by 56.8 ± 9.9% (from four independent experiments, p = 0.0107; Fig. 1D).

Figure 1.

Immunoblotting for α₂δ-1 and Ca_V2.2 in +/+ and −/− mice tissues. A, Comparison of α₂δ-1 levels in brain (left panel) and skeletal muscle (right panel) in α₂δ-1^−/− (lanes 1, 3) and α₂δ-1^+/+ (lanes 2, 4). A total of 20 μg of protein was loaded/lane. No α₂δ-1 is observed in α₂δ-1^−/− mice. Immunoblot for GAPDH (bottom panel) provides the loading control. B, α₂δ-1 levels in DRGs from α₂δ-1^−/− (lanes 1, 3) and α₂δ-1^+/+ (lanes 2, 4) mice. Material (30 μg of protein/lane) in lanes 3 and 4 has been deglycosylated with PNGase F, showing the characteristic decrease in MW of α₂δ-1 from ∼150 kDa (lane 2) to ∼105 kDa (lane 4). No corresponding bands are observed in α₂δ-1^−/− ganglia. GAPDH (bottom) provides a loading control. Data are representative of n = 3 experiments. C, Ca_V2.2 levels in cerebral synaptosomes (left) and brain WCL (right) from α₂δ-1^−/− (lanes 1, 3) and α₂δ-1^+/+ (lanes 2, 4) mice. GAPDH (bottom) provides a loading control for Ca_V2.2 quantification. Data are representative of n = 4 experiments. D, Ca_V2.2 levels in spinal cord synaptosomes (top) from α₂δ-1^−/− (lane 1) and α₂δ-1^+/+ (lane 2) mice. GAPDH (bottom) provides a loading control for Ca_V2.2 quantification. Data are representative of n = 4 experiments.

Deficits in behavioral responses to mechanical and innocuous cooling stimuli in α₂δ-1^−/− mice

α₂δ-1^−/− mice and wild-type littermate controls were initially examined for behavioral differences in mechanical, cold, and thermal sensitivity. Mechanical sensitivity was tested using von Frey filaments, which were applied across the hindpaws. Significantly fewer withdrawals were observed in α₂δ-1^−/− mice in response to 0.16 g and 0.4 g filaments but not 0.07 g, the lowest weight tested (Fig. 2A). The α₂δ-1^−/− mice also displayed significantly fewer nocifensive behaviors in response to acetone application to the hindpaws as an evaporative cooling stimulus (Fig. 2B). Nociceptive reflexes to thermal stimulation were conducted in lightly anesthetized mice, using infrared laser stimulation, which has been previously demonstrated to stimulate Aδ and C fibers selectively (Sikandar et al., 2013). There was no significant difference in noxious thermal withdrawal thresholds, as determined by infrared laser stimulation, between α₂δ-1^+/+ mice (1.60 ± 0.19 J, n = 10) and α₂δ-1^−/− mice (1.89 ± 0.20 J, n = 9) (p = 0.319, Mann–Whitney U test).

Figure 2.

Behavioral responses of α₂δ-1^+/+ and α₂δ-1^−/− mice to mechanical, cooling, and thermal stimuli applied to the hindpaws. A, Withdrawal frequency to punctate mechanical stimulation applied using von Frey filaments. L, Left hindpaw; R, right hindpaw. n = 14. B, Duration of nocifensive behaviors in response to acetone application (unpaired Student's t test with Welch's correction, n = 12). No significant differences were observed between left and right responses within genotypes. *p < 0.05, significant differences between respective paws between genotypes (Mann–Whitney U test). **p < 0.01, significant differences between respective paws between genotypes (Mann–Whitney U test). Data represent mean ± SEM.

Deficits in sensory coding of lamina V/VI dorsal horn neurons

In vivo electrophysiology was performed to assess the impact of α₂δ-1 deletion on the spinal processing of a range of thermal and mechanical stimuli. All neurons characterized had receptive fields on the glabrous skin of the hind toes and were identified as WDR by confirming responses to light brushing, noxious punctate mechanical stimulation, and thermal stimulation. Neurons were characterized from similar depths, corresponding to deep dorsal horn laminae (α₂δ-1^+/+, 516 ± 26 μm; α₂δ-1^−/−, 542 ± 25 μm).

Deep dorsal horn neurons responded to mechanical and thermal stimulation in an intensity-dependent manner. The reduced behavioral response to punctate mechanical stimulation in α₂δ-1^−/− mice (Fig. 2A) is corroborated by a deficit in mechanical coding of lamina V/VI neurons with significantly reduced neuronal responses to 1, 8, and 15 g stimulation (Fig. 3A). In addition, neuronal responses to dynamic brush stimulation were also attenuated (Fig. 3B, left). In contrast to the behavioral response to acetone-induced cooling (Fig. 2B), deep dorsal horn neuronal responses in α₂δ-1^−/− mice were not different from the α₂δ-1^+/+ controls (Fig. 3B, right). However, only 5 of 17 α₂δ-1^+/+ neurons and 5 of 19 α₂δ-1^−/− neurons were responsive to cooling, consistent with previous reports of minimal cold sensitivity in this temperature range in rat deep dorsal horn neurons (Khasabov et al., 2001). No significant difference in thermal coding of lamina V/VI neurons was observed in α₂δ-1^−/− mice (Fig. 3C), supporting our earlier observation of unaltered thermal withdrawal thresholds.

Figure 3.

Comparison of responses to mechanical, thermal, and electrical stimulation of deep dorsal horn lamina V/VI wide dynamic range neurons in α₂δ-1^+/+ and α₂δ-1^−/− mice. A, Evoked neuronal responses to punctate mechanical stimulation (two-way ANOVA, p < 0.05, followed by Bonferroni post hoc). B, Left, Evoked neuronal responses to dynamic brush stimulation (unpaired Student's t test, p < 0.05). Right, Cooling-evoked neuronal responses (unpaired Student's t test). C, Thermally evoked neuronal responses (two-way ANOVA, p > 0.05). D, Electrical thresholds for activation of A and C fibers (p > 0.05, unpaired Student's t test). E, Windup of deep dorsal horn neurons (16 stimuli, 0.5 Hz, 2 ms pulse), expressed as mean number of action potentials per stimulus number. F, Total action potentials evoked in response to repeated electrical stimulation, separated according to latency: A: 0–50 ms, C: 50–250 ms, PD (post-discharge) >250 ms. Input and windup were calculated as described in Materials and Methods. n for +/+ = 17, n for −/− = 19. *p < 0.05 (unpaired Student's t test). Data represent mean ± SEM.

Electrical thresholds for activation of A and C fibers were indistinguishable between the genotypes (Fig. 3D). After repeated suprathreshold electrical stimulation of the receptive field (16 stimuli, 2 ms pulse, 0.5 Hz), the cumulative total of neuronal events evoked by A and C fiber stimulation was not affected in α₂δ-1^−/− mice (Fig. 3E,F). Lamina V/VI neurons from α₂δ-1^+/+ and α₂δ-1^−/− exhibited a similar rate of windup (Fig. 3E), in addition to a similar total windup (Fig. 3F). Windup is an NMDA-dependent phenomenon where dorsal horn neurons become hyperexcitable after repeated C fiber stimulation (Dickenson and Sullivan, 1987). However, “input,” the nonpotentiated response, more indicative of presynaptic events, and the post-discharge, a property of spinal neurons, were significantly decreased in α₂δ-1^−/− mice (Fig. 3F).

DRG neurons from α₂δ-1^+/+ and α₂δ-1^−/− mice are not different in their expression of neurofilaments

DRG neurons are highly heterogeneous in their sensory properties and physiological function. They can be grouped into distinct subpopulations by many different criteria, for example, according to their differential expression of neurofilaments, a component of the cytoskeleton (Goldstein et al., 1991; Ruscheweyh et al., 2007). We used immunohistochemical methods to visualize the expression of the intermediate filaments peripherin and NF200 in DRG sections from α₂δ-1^+/+ and α₂δ-1^−/− mice. Peripherin is expressed in nonmyelinated nociceptive C fibers, whereas NF200 is expressed in myelinated nociceptive and non-nociceptive A fibers (Ruscheweyh et al., 2007). The majority of cell bodies in α₂δ-1^−/− L4 DRG sections showed positive staining for either peripherin or NF200, with a small number of cells showing staining for both intermediate filaments (Fig. 4A). The proportion of cells that were peripherin-positive (∼60%), NF200-positive (∼40%), or positive for both neurofilaments (∼4%) was not significantly different in L4 DRG sections from α₂δ-1^+/+ compared with α₂δ-1^−/− mice (Fig. 4B), and is similar to the proportional staining reported by others for mice (Quick et al., 2012). These results show that the proportion of unmyelinated C-fibers and myelinated A-fibers is unchanged by the ablation of α₂δ-1.

Figure 4.

Sensory neuronal subpopulations and terminals in the skin are not different between α₂δ-1^+/+ and α₂δ-1^−/− DRGs. A, Representative double immunofluorescence staining for the neuronal intermediate filaments peripherin (green) and NF200 (red) in an L4 DRG section from an α₂δ-1^−/− mouse. The majority of cell bodies are positive for peripherin or NF200, with a small number of cells showing staining for both intermediate filament types (yellow). B, Quantification of peripherin (black bars), NF200 (white bars), and staining for both intermediate filament types (both, gray) in L4 DRGs from α₂δ-1^+/+ and α₂δ-1^−/− mice, shown as percentage of cells stained per section. Data represent mean ± SEM. Statistical analysis was as follows: unpaired Student's t test comparing n = 14 +/+ with n = 13 −/− sections from four mice, respectively; p = 0.727 for peripherin, p = 0.073 for NF200, p = 0.70 for both. C, D, Images of PGP 9.5-positive axons crossing the dermal–epidermal boundary of glabrous skin in α₂δ-1^+/+ (C) and α₂δ-1^−/− (D) mice. The dermal–epidermal junction is represented by the dotted line. Positive nerve profiles (stained with PGP 9.5 in red) crossing the dermal–epidermal junction are counted (arrows). Scale bar, 20 μm.

Normal numbers of sensory neuron axons in the skin of α₂δ-1^−/− mice

We examined the peripheral innervation of sensory fibers and observed no difference in density of PGP 9.5-positive axons crossing the dermal–epidermal boundary, in agreement with the lack of gross differences between DRG neurons in α₂δ-1^+/+ and α₂δ-1^−/− mice. For α₂δ-1^+/+ mice, there were 30.4 ± 1.7 axons/mm of epidermis; and for α₂δ-1^−/− mice, there were 28.6 ± 5.4 axons/mm, crossing the dermal–epidermal boundary (n = 4 mice, 3 sections per mouse; Fig. 4C,D).

DRG neurons from α₂δ-1^+/+ and α₂δ-1^−/− mice differ in their responsiveness to TRP channel agonists

Sensory neurons are broadly divided into mechanoreceptors, thermoreceptors, and nociceptors, and the differential expression of TRP channels is responsible for many of their distinctive somatosensory properties (Hjerling-Leffler et al., 2007). As shown in Figure 5A, we examined these properties by using Ca²⁺ imaging while exposing DRG neurons from α₂δ-1^+/+ and α₂δ-1^−/− mice to sequential application of different TRP channel agonists. In the first set of experiments, MO (100 μm, 20 s) was applied to activate TRPA1, followed by menthol (ME, 500 μm, 30 s) to activate TRPM8, and finally capsaicin (CAPS or C, 1 μm, 10 s) to activate TRPV1. After the final agonist application, cells were stimulated with high potassium (K, 60 mm, 10 s) to assess cell viability. In a second set of experiments, another TRPA1 agonist CA was also applied. The agonist sequence in the second set was 250 μm ME for 30 s, followed by 100 μm CA for 60 s, followed by 1 μm CAPS for 5 s.

Figure 5.

TRPA1, TRPV1, and TRPM8 responses in α₂δ-1^+/+ and α₂δ-1^−/− DRG neurons. A, Representative traces showing changes in cytoplasmic Ca²⁺ activity (fura-2 AM fluorescence ratio: ratio (F₃₄₀/F₃₈₀)) in response to the TRPA1 agonist MO (100 μm, applied for 20 s), the TRPM8 agonist ME (500 μm, 30 s), the TRPV1 agonist CAPS (1 μm, 10 s), and 60 mm K⁺ (K, 10 s) in α₂δ-1^+/+ neurons. B, Quantification of percentage of α₂δ-1^+/+ (black bars) and α₂δ-1^−/− (white bars) DRG neurons responding to MO (100 μm for 20 s, left), CAPS (1 μm for 5 or 10 s, middle), and ME (250 or 500 μm for 30 s, right). The number of responses to each agonist was normalized to number of responses to 60 mm K⁺ per coverslip. Error bars indicate SEM. Statistical analysis was as follows: unpaired Student's t test comparing α₂δ-1^+/+ and α₂δ-1^−/− responses: MO response, p = 0.90 (n = 13 α₂δ-1^+/+ and n = 18 α₂δ-1^−/− coverslips); CAPS response, p = 0.005 (n = 29 α₂δ-1^+/+ and n = 36 α₂δ-1^−/− coverslips); ME response, p = 0.028 (n = 30 α₂δ-1^+/+ and n = 37 α₂δ-1^−/− coverslips). In total, n = 982 α₂δ-1^+/+ and n = 1151 α₂δ-1^−/− cells were analyzed. C, Peak fura-2 AM fluorescence ratio (peak ratio: (F₃₄₀/F₃₈₀)) for MO, CA, CAPS, and ME responses. Error bars indicate SEM. Statistical analysis was as follows: unpaired Student's t test, comparing α₂δ-1^+/+ and α₂δ-1^−/− neurons: MO peak response, p < 0.003 (n = 290 α₂δ-1^+/+, n = 247 α₂δ-1^−/− neurons); CA peak response, p = 0.0002 (n = 230 α₂δ-1^+/+, n = 325 α₂δ-1^−/− neurons); CAPS peak response, p = 0.011 (n = 189 α₂δ-1^+/+, n = 191 α₂δ-1^−/− neurons); ME peak response (250 μm), p = 0.98 (n = 55 α₂δ-1^+/+, n = 49 ^−/− neurons). *p < 0.05. **p < 0.01. ***p < 0.001.

As shown in Figure 5B, the proportion of α₂δ-1^+/+ and α₂δ-1^−/− cells responding to MO (applied in the first agonist sequence) was not significantly different. Furthermore, the proportion of α₂δ-1^+/+ cells and α₂δ-1^−/− cells responding to 100 μm CA in the second agonist sequence was also not significantly different (49.6 ± 4.6% of α₂δ-1^+/+ cells; n = 17 coverslips), compared with 50.6 ± 3.4% of α₂δ-1^−/− cells (n = 19 coverslips). In contrast, a significantly higher proportion of α₂δ-1^−/− cells responded to CAPS (Fig. 5B; a 27% increase, data pooled from both agonist sequences). Furthermore, a significantly lower proportion of α₂δ-1^−/− cells responded to ME (250 or 500 μm for 30 s) compared with α₂δ-1^+/+ cells (Fig. 5B; a 33% decrease, data pooled from both agonist sequences).

To determine whether the changes in proportion of cells responding to CAPS and ME reflected an altered sensitivity to agonist, or a change in the number of DRG neurons expressing these receptors, we next compared the amplitude of TRP channel agonist responses measured in α₂δ-1^+/+ and α₂δ-1^−/− DRG neurons. As shown in Figure 5C, MO, CA, and CAPS all evoked a significantly higher fura-2 AM ratio peak in α₂δ-1^−/− compared with α₂δ-1^+/+ cells. In contrast, there was no significant difference in the peak response between α₂δ-1^−/− and α₂δ-1^+/+ cells evoked by either 250 μm ME (Fig. 5C) or 500 μm ME (ratio in α₂δ-1^+/+ cells: 2.28 ± 0.14, n = 33; ratio in α₂δ-1^−/− cells: 2.51 ± 0.31, n = 21). There was also no significant difference in the peak response between α₂δ-1^−/− and α₂δ-1^+/+ DRGs evoked by 30 s application of 20 μm WS-12, a menthol derivative (Bodding et al., 2007). The ratio was 2.21 ± 0.28 (n = 9) and 2.66 ± 0.13 (n = 19) in α₂δ-1^+/+ and α₂δ-1^−/− cells, respectively (p = 0.11). This suggests that the loss of α₂δ-1 does not affect the function of TRPM8 channels and is not the reason for the smaller number of cells responding to ME. Furthermore, we found no significant change in TRPM8, TRPA1, or TRPV1 mRNA levels, in α₂δ-1^−/− compared with α₂δ-1^+/+ DRGs (Fig. 6A), suggesting that compensatory changes in receptor expression are not responsible for the differences between α₂δ-1^+/+ and α₂δ-1^−/− cells observed in the Ca²⁺ imaging experiments. In addition, there was no effect of α₂δ-1 coexpression on TRPM8 or TRPV1 expression levels when they were coexpressed in tsA-201 cells (Fig. 6B,C). Piezo2 has been shown to be involved in mechanosensation (Coste et al., 2012) and in mechanical allodynia (Eijkelkamp et al., 2013). However, we found that the Piezo2 mRNA levels were not altered in DRGs from α₂δ-1^−/− mice, the level being 94.7 ± 7.8% in α₂δ-1^−/− DRGs, relative to α₂δ-1 DRGs (n = 3; unpaired t test, p = 0.62).

Figure 6.

TRP channel mRNA and protein expression. A, TRPM8, TRPA1, and TRPV1 mRNA levels in α₂δ-1^−/− (white bar) and α₂δ-1^+/+ (black bar) DRGs (n = 3), normalized to the mean α₂δ-1^+/+ mRNA values. B, C, TRPM8 (B) and TRPV1 (C) expression, after transfection in tsA-201 cells, detected by their associated myc tag (top), was not affected by coexpression of α₂δ-1 (lane 1), compared with a control protein β1b (lane 2). No signal was observed in WCL from untransfected (U/T) tsA-201 cells (lane 3). Bottom, GAPDH loading controls. C, α₂δ-1 expression is also shown (middle). All lanes are from the same gel. Data are representative of n = 2 experiments.

Calcium channel currents are reduced in DRGs from α₂δ-1^−/− compared with α₂δ-1^+/+ mice

Calcium channel currents were recorded using 2 mm Ba²⁺ as the charge carrier, from small and medium DRG neurons, cultured from α₂δ-1^−/− and α₂δ-1^+/+ mice (Fig. 7). Current density–voltage (I-V) relationships were measured 30 ms after the start of depolarization (Fig. 7A,B). The peak current density was 27% smaller at 0 mV and 30% smaller at 10 mV, in α₂δ-1^−/− compared with α₂δ-1^+/+ DRG neurons, but there was no significant shift in the voltage dependence of activation of the currents. In experiments using ω-conotoxin GVIA (1 μm) to block N-type currents (Fig. 7C,D), this toxin inhibited the peak Ba²⁺ currents in α₂δ-1^+/+ DRGs by 51% at 0 mV, (54% at 10 mV), and that in α₂δ-1^−/− DRGs by 55% at 0 mV (52% at 10 mV) (compare Fig. 7A with Fig. 7C), indicating that there is a similar component of N-type current in α₂δ-1^−/− and α₂δ-1^+/+ DRG somata.

Figure 7.

Calcium channel currents in α₂δ-1^+/+ and α₂δ-1^−/− DRG neurons. A, Calcium channel current density-voltage relationship for α₂δ-1^+/+ (closed squares, n = 19 cells, 4 mice) and α₂δ-1^−/− (open circles, n = 29 cells, 5 mice) DRG neurons. Mean data were fit between −60 and 40 mV with a modified Boltzmann relationship (see Materials and Methods). For α₂δ-1^+/+, the V_50,act was −10.2 mV and for α₂δ-1^−/−, the V_50,act was −8.17 mV. B, Representative current traces under the two conditions: α₂δ-1^+/+ (top) and α₂δ-1^−/− (bottom). Top, Voltage protocol. Holding potential −90 mV, test potentials −60 to 10 mV in 10 mV steps. C, Calcium channel current density-voltage relationship for α₂δ-1^+/+ (closed squares, n = 7 cells, 4 mice) and α₂δ-1^−/− (open circles, n = 25 cells, 7 mice) for DRG neurons after preincubation for 15 min with ω-conotoxin GVIA (ω-CTX GVIA, 1 μm); otherwise as in A. For α₂δ-1^+/+, the V_50,act was −7.85 mV and for α₂δ-1^−/−, the V_50,act was −10.3 mV. D, Representative current traces under the two conditions, as in B. Data were compared by two-way ANOVA, followed by the Tukey Honest Significant Differences test. A, C, The statistical significance of the difference in current density between α₂δ-1^+/+ and α₂δ-1^−/− is indicated: *p < 0.05, **p < 0.01. There was no effect of ω-conotoxin GVIA on the difference in current density between the mouse genotypes.

α₂δ-1^−/− mice develop delayed mechanical hypersensitivity after PSNL

We next sought to examine the development of mechanical and cold hypersensitivity in the absence of α₂δ-1, as the upregulation of this subunit has been shown to correlate with the development of neuropathic-like behaviors (Li et al., 2004). Baseline testing was performed 5 d before PSNL surgery and then repeated on postoperative days 3, 7, 9, 11, and 14. After surgery, mice groomed normally and maintained presurgery body weights. The α₂δ-1^+/+ mice developed rapid mechanical hypersensitivity on the injured side after PSNL, with significant increases in withdrawal frequency to the 0.4 g von Frey filament, compared with contralateral responses from day 3 onwards, and to α₂δ-1^+/+ sham responses from day 7 onwards (Fig. 8A). The ipsilateral withdrawal frequencies for α₂δ-1^−/− mice were reduced compared with α₂δ-1^+/+ mice at days 3, 7, 9, and 11 and only differed from contralateral responses from day 11 onwards (Fig. 8B). The α₂δ-1^+/+ sham mice did not display any behavioral changes after surgery for the duration of the 2 week observation period (Fig. 8C).

Figure 8.

Development of mechanical and cold hypersensitivity after PSNL. The same legend applies for all data. ■ represents the ipsilateral side; and □ represents the contralateral side. A, Increased frequency of ipsilateral withdrawals to mechanical stimulation (0.4 g von Frey) compared with contralateral for α₂δ-1^+/+ group. B, Delayed increase in frequency of ipsilateral withdrawals to mechanical stimulation compared with contralateral for α₂δ-1^−/− group. C, Unaltered frequency of ipsilateral withdrawals to mechanical stimulation compared with contralateral for α₂δ-1^+/+ sham group (Kruskal-Wallis, paired comparisons made with Dunn's post hoc test and Bonferroni correction). n for +/+ = 16, n for −/− = 16, n for +/+ sham-operated = 10. D, Increased duration of α₂δ-1^+/+ ipsilateral response to cooling stimulation compared with contralateral. E, Increased duration of α₂δ-1^−/− ipsilateral response to cooling stimulation compared with contralateral. F, Unaltered duration of α₂δ-1^+/+ sham ipsilateral response to cooling stimulation compared with contralateral (two-way repeated-measures ANOVA, paired comparisons made with Bonferroni's post hoc test). n for +/+ = 11, n for −/− = 13, n for +/+ sham = 7. *Statistically significant differences between α₂δ-1^+/+ and α₂δ-1^−/− mice. ⁺Statistically significant differences between α₂δ-1^+/+ and α₂δ-1^+/+ sham groups. ^●Statistically significant differences between ipsilateral and contralateral. G, The 10 mg/kg pregabalin attenuates mechanical hypersensitivity in α₂δ-1^+/+ mice (symbols given in key above H; Friedman test p < 0.01, paired comparisons made with Wilcoxon test). n = 6 for all groups. *Statistically significant difference of response in pregabalin-treated mice, compared with predrug withdrawal frequency. ⁺Statistically significant differences between ipsilateral vehicle-treated and ipsilateral pregabalin-treated responses. *p < 0.05. **p < 0.01. ***p < 0.001. Data represent mean ± SEM. PGB, Pregabalin; Ipsi., ipsilateral; Contra., contralateral. H, Efficacy of pregabalin is lost in α₂δ-1^−/− mice. n for vehicle = 6, n for pregabalin-treated = 7.

Cold hypersensitivity after PSNL was not markedly altered in α₂δ-1^−/− mice, and was an infrequent occurrence after surgery in both genotypes. The α₂δ-1^+/+ ipsilateral responses differed significantly from the contralateral responses on days 11 and 14 (Fig. 8D), α₂δ-1^−/− ipsilateral responses differed significantly from the contralateral responses on days 3, 7, and 9 (Fig. 8E); however, no significant difference was observed between the α₂δ-1^+/+ and α₂δ-1^−/− ipsilateral responses. The α₂δ-1^+/+ sham ipsilateral responses did not differ from the contralateral side (Fig. 8F).

Anti-hyperalgesic efficacy of pregabalin is lost in α₂δ-1^−/− mice after PSNL

The ability of pregabalin to attenuate mechanical hypersensitivity was examined in α₂δ-1^+/+ and α₂δ-1^−/− 14 d after PSNL. Mechanical hypersensitivity was tested before administration of either vehicle or 10 mg/kg pregabalin, and repeated 30, 60, and 90 min after dosing. In α₂δ-1^+/+ mice, pregabalin reduced withdrawal frequency on the injured ipsilateral side compared with predrug values at 30 min after dosing, and to the vehicle-treated group at 30 and 60 min after dosing. Contralateral responses were not altered by vehicle or pregabalin treatment (Fig. 8G). In α₂δ-1^−/− mice, pregabalin did not affect mechanical hypersensitivity compared with predrug withdrawal frequencies or the vehicle-treated group. Contralateral responses were also unaffected by both treatments (Fig. 8H).

α₂δ-1 mRNA is upregulated after PSNL in α₂δ-1^+/+ DRGs, but other Ca_V auxiliary subunits do not compensate for α₂δ-1 in −/− DRGs

Nerve injury in animal models of chronic pain causes changes in the expression of a plethora of genes and their gene products (Costigan et al., 2002). These changes include the upregulation of α₂δ-1 (Newton et al., 2001). Using quantitative PCR 14 d after PSNL, we found that the α₂δ-1 mRNA level was significantly increased by >100% in ipsilateral L3-L5 DRGs from α₂δ-1^+/+ mice compared with sham-operated α₂δ-1^+/+ mice (Fig. 9A). We next investigated the expression of the other two α₂δ isoforms present in DRGs (i.e., α₂δ-2 and α₂δ-3). We found that their expression level did not change in α₂δ-1^+/+ mice after PSNL compared with α₂δ-1^+/+ mice after sham surgery (Fig. 9B,C, black bars). Moreover, the expression levels of α₂δ-2 and α₂δ-3 mRNA in α₂δ-1^−/− PSNL mice were not significantly different from the levels found in α₂δ-1^+/+ sham and PSNL mice (Figs. 9B,C, white bar compared with black PSNL bar). This indicates that neither of the other two α₂δ isoforms compensates for the absence of α₂δ-1 upregulation after PSNL in α₂δ-1^−/− mice.

Figure 9.

PSNL-mediated ipsilateral changes of mRNA of neuropathy markers in α₂δ-1^+/+ and α₂δ-1^−/− DRGs. The legend applies to all parts of the figure. A, Q-PCR results for α₂δ-1 mRNA levels (fold expression (norm_contra)) in pooled L3-L5 ipsilateral DRGs from sham-operated (n = 9) or PSNL-operated (n = 7) α₂δ-1^+/+ mice (black bars), 14 d after surgery. Data are normalized to the respective contralateral side. Error bars indicate SEM. Statistical analysis was performed using one-way ANOVA with p = 0.003 using Bonferroni's post test. No data are included for the α₂δ-1^−/− DRGs because of the gene disruption. B, Q-PCR results for α₂δ-2 mRNA levels (fold expression (norm_contra)) in pooled L3-L5 ipsilateral DRGs from sham-operated (n = 9) or PSNL-operated (n = 7) α₂δ-1^+/+ mice (black bars) or PSNL-operated α₂δ-1^−/− mice (white bar, n = 7), 14 d after surgery. Data are normalized to the respective contralateral side. Error bars indicate SEM. Statistical analysis: one-way ANOVA with p = 0.11. C, Q-PCR results for α₂δ-3 mRNA levels (fold expression (norm_contra)) in pooled L3-L5 ipsilateral DRGs from sham-operated (n = 9) or PSNL-operated (n = 7) α₂δ-1^+/+ mice (black bars) or PSNL-operated α₂δ-1^−/− mice (white bar, n = 7), 14 d after surgery. Data are normalized to the respective contralateral side. Error bars indicate SEM. Statistical analysis: one-way ANOVA with p = 0.21. D, Q-PCR results for β3 mRNA levels (fold expression (norm_contra)) in pooled L3-L5 ipsilateral DRGs from sham-operated (n = 9) or PSNL-operated (n = 7) α₂δ-1^+/+ mice (black bars) or PSNL-operated α₂δ-1^−/− mice (white bar, n = 6), 14 d after surgery. Data are normalized to the respective contralateral side. Error bars indicate SEM. Statistical analysis: one-way ANOVA with Bonferroni's multiple-comparison test, p > 0.05 for all comparisons. E, Q-PCR results for TRPM8, TRPA1, and TRPV1 mRNA levels (fold expression) in pooled L3-L5 ipsilateral DRGs from PSNL-operated α₂δ-1^+/+ mice (black bars n = 3) or PSNL-operated α₂δ-1^−/− mice (white bars, n = 3), 14 d after surgery. Data are normalized to the respective contralateral side. Error bars indicate SEM. Statistical analysis shows that TRPM8 mRNA ratios for α₂δ-1^+/+ and α₂δ-1^−/− DRGs are significantly less than a theoretical value of 1: †p = 0.0003 and p = 0.0142, respectively (one-sample t test). The difference between α₂δ-1^+/+ and α₂δ-1^−/− TRPM8 ratios is also statistically significant: *p = 0.068 (Student's t test).

Ca_Vβ3, which is the main β subunit in DRGs, and has been found in one study to be upregulated in rat DRGs after SNL (Li et al., 2012); however, here we found the β3 mRNA level was not altered in DRGs by PSNL, either in α₂δ-1^+/+ or α₂δ-1^−/− mice (Fig. 9D).

Investigation of TRP channel mRNA levels in α₂δ-1^+/+ and α₂δ-1^−/− DRGs after PSNL

We compared mRNA levels for TRPM8, TRPA1, and TRPV1 in α₂δ-1^+/+ and α₂δ-1^−/− DRGs, normalizing the data on the ipsilateral side after PSNL to that on the corresponding contralateral side. We found that TRPM8 mRNA was significantly reduced ipsilateral to PSNL compared with the level in the corresponding contralateral DRGs, as previously observed for another nerve injury model (Caspani et al., 2009), but there was a smaller reduction in the α₂δ-1^−/− than in α₂δ-1^+/+ DRGs (Fig. 9E).

ATF3 mRNA and NPY protein are upregulated normally after PSNL in α₂δ-1^−/− DRGs

We next compared the mRNA levels of ATF3, a neuronal marker for nerve injury (Tsujino et al., 2000), in α₂δ-1^+/+ and α₂δ-1^−/− DRGs. We found that, 14 d after PSNL, ATF3 mRNA was significantly upregulated in both α₂δ-1^+/+ and α₂δ-1^−/− DRGs compared with DRGs taken from α₂δ-1^+/+ sham-operated mice (Fig. 10A). Moreover, the upregulation of ATF3 in α₂δ-1^−/− DRGs was not significantly different from the upregulation in α₂δ-1^+/+ tissue.

Figure 10.

PSNL-mediated ipsilateral changes of mRNA and protein levels of neuropathy markers in α₂δ-1^+/+ and α₂δ-1^−/− DRGs. A, Q-PCR results for ATF3 mRNA levels (fold expression) in pooled L3-L5 ipsilateral DRGs from sham-operated (n = 9) or PSNL-operated (n = 7) α₂δ-1^+/+ mice (black bars) or PSNL-operated α₂δ-1^−/− mice (white bar, n = 7) 14 d after surgery. Data are normalized to the respective contralateral side. ***p < 0.001 (one-way ANOVA and Bonferroni's post hoc test). ATF3 mRNA levels in PSNL-operated α₂δ-1^+/+ were not significantly different from PSNL-operated α₂δ-1^−/− mice. B, Representative immunofluorescence images of ipsilateral (ipsi) and contralateral (contra) L4 DRG sections from α₂δ-1^+/+ and α₂δ-1^−/− mice after PSNL. Double immunofluorescence staining for α₂δ-1 (green) and NPY (red); blue represents nuclear staining. Scale bar, 100 μm. C, Quantification of NPY-positive cells per section (number of NPY-positive cells/section) in ipsilateral and contralateral α₂δ-1^+/+ (black bars) and α₂δ-1^−/− (white bars) L3-L5 DRGs 14 d after PSNL. Error bars indicate SEM. Statistical analysis was as follows: unpaired Student's t test comparing number of NPY-positive cells. α₂δ-1^+/+ and α₂δ-1^−/− DRG sections: ipsilateral L3, p = 0.97 (n = 18 α₂δ-1^+/+ sections, n = 19 α₂δ-1^−/− sections); ipsilateral L4, p = 0.84 (n = 22 α₂δ-1^+/+ sections, n = 20 α₂δ-1^−/− sections); ipsilateral L5, p = 0.37 (n = 13 α₂δ-1^+/+ sections, n = 13 α₂δ-1^−/− sections); contra L3, p = 0.27 (n = 11 α₂δ-1^+/+ sections, n = 13 α₂δ-1^−/− sections); contra L4 contra, p = 0.94 (n = 13 α₂δ-1^+/+ sections, n = 12 α₂δ-1^−/− sections); and contra L5, p = 0.39 (n = 8 α₂δ-1^+/+ sections, n = 10 α₂δ-1^−/− sections).

Immunofluorescence staining in DRGs 14 d after PSNL confirmed the ipsilateral upregulation of α₂δ-1 (green) in α₂δ-1^+/+ L4 DRGs compared with the contralateral side (Fig. 10B, upper row, first and third panels). There was no detectable α₂δ-1 staining in sections of ipsilateral or contralateral L4 DRGs from α₂δ-1^−/− mice (Fig. 10B, upper row, second and fourth panels). Staining for neuropeptide Y (NPY, red), another marker for nerve injury (Wakisaka et al., 1992; Ruscheweyh et al., 2007) in the same sections, showed that NPY was upregulated in the ipsilateral DRGs from both α₂δ-1^+/+ and α₂δ-1^−/− mice (Fig. 10B, lower row, first and second panels), whereas the staining was virtually absent from the contralateral side (Fig. 10B, lower row, third and fourth panels). We then quantified the NPY staining by counting the number of NPY-positive cells per section (Fig. 10C). There was no significant difference in the number of NPY-positive cells in ipsilateral and contralateral sections from L3-L5 DRGs from α₂δ-1^+/+ compared with α₂δ-1^−/− mice 14 d after PSNL. The quantification also showed that PSNL mainly affected the L3 and L4 DRG neurons and axons.

Discussion

This study examined whether α₂δ-1 is involved in baseline somatosensory functions and in development of neuropathic pain. We found that α₂δ-1 knock-out mice show markedly reduced behavioral sensitivity to mechanical and cold stimuli, with no change in withdrawal threshold for noxious heat. α₂δ-1 is expressed in most excitable tissues, including peripheral and central neurons (Cole et al., 2005; Taylor and Garrido, 2008). In this global α₂δ-1 knock-out model, altered behavioral sensitivity could therefore be attributed to differences in supraspinal circuits as well as spinal and somatosensory neurons. Our data strongly support a peripheral and spinal role for α₂δ-1 in modulating sensitivity to these stimuli. The behavioral profile was, in part, reflected by attenuated mechanical responses of dorsal horn neurons, whereas thermal responses were unaffected. The lower presynaptic “input” after electrical stimulation in α₂δ-1^−/− mice likely represents a decrease of synaptic transmitter release from primary afferents projecting to dorsal horn neurons, and the lack of change in windup also supports presynaptic alterations. However, there was no global reduction in primary afferent transmission via A and C-fibers, in terms of evoked responses and thresholds, suggesting no major changes in conduction of afferent nerves.

In addition to its association with calcium channels, α₂δ-1 has also been shown to mediate excitatory synaptogenesis via interaction with thrombospondin (Eroglu et al., 2009). At the spinal level, α₂δ-1 is strongly expressed in presynaptic terminals of primary afferents terminating in the superficial laminae (Bauer et al., 2009), whereas lamina V/VI dorsal horn neurons receive both direct and indirect input from Aβ, Aδ, and C fibers. The present study does not directly address whether there are differences between α₂δ-1^+/+ and α₂δ-1^−/− mice in primary afferent synapse numbers. This hypothesis is certainly feasible and will be examined in the future, although the modality-specific deficits argue against global synaptic dysregulation in the dorsal horn.

It is of interest that the phenotype shown by α₂δ-1^−/− mice (reduced mechanical and cold sensitivity, with no change in sensitivity to noxious heat) shows similarities with that observed in Na_V1.8 knock-out mice (Abrahamsen et al., 2008), and when Na_V1.7 is deleted selectively in Na_V1.8-positive nociceptors (Minett et al., 2012). TRPM8 is the principal detector of cool temperatures in mouse (Bautista et al., 2007; Colburn et al., 2007; Dhaka et al., 2007; Knowlton et al., 2013). Although these neurons constitute a minor proportion of all primary afferents, synaptic transmission in this population may be particularly dependent on α₂δ-1.

To examine the basis for the observed modality-specific differences in response between the α₂δ-1^+/+ and α₂δ-1^−/− mice, we examined the properties of the DRG neurons. Although gross differences were absent, when examining responses to TRP agonists, we found that the proportion of cultured DRG neurons responding to ME was significantly reduced in α₂δ-1^−/− mice, agreeing with the marked reduction in behavioral response to acetone cooling in naive α₂δ-1^−/− mice. Nevertheless, our experiments do not provide evidence that his effect was the result of a reduction in TRPM8 mRNA. Mechanoreceptors have recently been shown to be made up of Piezo proteins (Coste et al., 2012); however, the reduced baseline response to mechanical stimuli in α₂δ-1^−/− mice was not a result of reduced Piezo2 mRNA.

The sensor for noxious heat, TRPV1, is selectively expressed in small-medium DRG neurons (Kobayashi et al., 2005), which also express α₂δ-1 most strongly (Newton et al., 2001; Bauer et al., 2009). Surprisingly, we found an increase in the proportion of DRG neurons in α₂δ-1^−/− mice responding to activation of TRPV1 receptors by CAPS in vitro and a small increase in the peak Ca²⁺ response to CAPS. This difference is in contrast to the lack of change in behavioral threshold for elevated temperatures, and the lack of significant difference in AP response of WDR neurons to thermal stimulation. We hypothesize that this might represent a compensatory upregulation at the cellular level of TRPV1 receptor function in α₂δ-1^−/− DRG neurons, to counteract the reduction in primary afferent transmission resulting from the loss of α₂δ-1 at central terminals. However, we found no change in TRPV1 mRNA level in α₂δ-1^−/− DRGs, suggesting that there is no difference at the level of transcription, indicating that the basis for the increased response to CAPS in vitro may have a translational or post-translational basis. Interestingly, TRPV1 protein has been found to be upregulated in uninjured sensory neurons, after partial nerve injury (Kim et al., 2008).

An alternative hypothesis to explain the reduction in baseline response to cooling and mechanical stimulation is that α₂δ-1 is involved directly in the function of TRPM8 receptors and mechanoreceptors. Whether α₂δ-1, which is transported to peripheral as well as central terminals of DRG neurons (Bauer et al., 2009), is associated directly with any of the channels involved in peripheral mechanosensation or cold sensation will be the subject of future experiments.

We found, as expected, that α₂δ-1^−/− DRG neurons exhibit significantly smaller calcium channel currents, but the proportion of N-type currents in the somata was similar between the genotypes, being 52–55%. DRG neurons have been found to contain a high proportion of N-type currents (50–70%) (Regan et al., 1991; Bell et al., 2004), together with l-type (Scroggs and Fox, 1991), T-type (in some subclasses,Carbone and Lux, 1984; Scroggs and Fox, 1992; Bell et al., 2004), and a small amount of P-type (Saegusa et al., 2001).

We also found a reduced level of Ca_V2.2 protein in brain and spinal cord synaptosomes, in agreement with the hypothesis that α₂δ-1 is important for trafficking Ca_V2 channels to presynaptic terminals (Hoppa et al., 2012). We have recently found that α₂δ subunits play a key role in shaping DRG neuronal action potentials, with overexpression of α₂δ-1 shortening action potential duration (Hoppa et al., 2012). The mechanism may involve the known ability of α₂δ-1 to influence calcium current kinetics (Felix et al., 1997; Cantí et al., 2003; Tuluc et al., 2007) or the coupling of calcium channels to particular K⁺ channels associated with action potential repolarization (Hoppa et al., 2012). DRG neurons show a spectrum of action potential duration, C-nociceptors showing the longest and Aα/β fibers the shortest duration. Action potential duration is inversely correlated with maximal firing frequency (Lawson, 2002). Thus upregulation of α₂δ-1 after nerve injury may allow DRGs to support an increased spontaneous action potential firing frequency and increased transmitter release in a DRG subtype-selective manner, an adaptive response that would be absent from α₂δ-1^−/− mice.

After chronic sensory nerve injury, the α₂δ-1^−/− mice show a marked delay in the development of mechanical hypersensitivity in the affected limb, with a significant difference between the ipsilateral and contralateral side only being observed at 11 d after PSNL, whereas for WT mice this difference was evident at the first time point measured (3 d). This implicates α₂δ-1 in the early establishment of chronic pain, but that other factors must also be involved in the development of central neuroplasticity leading to a long-term neuropathic state. Together with the deficit in baseline mechanical responses in naive α₂δ-1^−/− mice, this suggests that synaptic transmission between the mechanosensitive DRG neurons and dorsal horn neurons is particularly sensitive to the loss of α₂δ-1 from primary afferent terminals.

We previously identified two mechanisms whereby α₂δ subunits increase synaptic transmission (Hoppa et al., 2012). In hippocampal neurons, α₂δ subunits increased the trafficking of Ca_V2.1 from cell body to synaptic boutons, although it was not determined whether more channels were inserted into the plasma membrane, as occurs in expression systems (Hendrich et al., 2008). Second, α₂δ subunits increased synaptic vesicle release probability after an action potential, interpreted as increasing the proximity of channels to active zone release sites, involving local channel organization into slots associated with vesicular release (Hoppa et al., 2012).

There is an increasing body of evidence that the elevation of the α₂δ-1 subunit after nerve injury is key to the mechanism of action of the gabapentinoid drugs in neuropathic pain (Field et al., 2006; Dolphin, 2012). It is therefore likely that these drugs act on multiple α₂δ-mediated trafficking processes (Hendrich et al., 2008; Tran-Van-Minh and Dolphin, 2010), which have different time courses in vivo, an effect on α₂δ-1-associated channel trafficking from DRG somata to presynaptic terminal (Bauer et al., 2009), being a slower process than modification of channel localization/recycling at the active zone.

Another action of gabapentin is to inhibit synaptogenesis via interference with α₂δ-1–thrombospondin interactions (Eroglu et al., 2009), although it is unclear how this mechanism would participate in the observed rapid reduction in mechanical hypersensitivity. In addition, there are other permissive processes for the actions of gabapentinoids (Fehrenbacher et al., 2003; Suzuki et al., 2005; Doyon et al., 2013). Although gabapentinoids display similar affinity for α₂δ-1 and α₂δ-2 (Gong et al., 2001; Marais et al., 2001), pregabalin had no effect on mechanically evoked responses in α₂δ-1^−/− mice, confirming that α₂δ-1 is the molecular target for the anti-hyperalgesic actions of pregabalin (Field et al., 2006). The α₂δ-1^−/− mice now provide another tool to study the mechanism(s) of action of gabapentinoids.