Targeting the CaVα–CaVβ interaction yields an antagonist of the N-type CaV2.2 channel with broad antinociceptive efficacy

Abstract

Inhibition of voltage-gated calcium (CaV) channels is a potential therapy for many neurological diseases including chronic pain. Neuronal CaV1/CaV2 channels are composed of α, β, γ and α2δ subunits. The β subunits of CaV channels are cytoplasmic proteins that increase the surface expression of the pore-forming α subunit of CaV. We targeted the high-affinity protein–protein interface of CaVβ’s pocket within the CaVα subunit. Structure-based virtual screening of 50,000 small molecule library docked to the β subunit led to the identification of 2-(3,5-dimethylisoxazol-4-yl)-N-((4-((3-phenylpropyl)amino)quinazolin-2-yl)methyl)acetamide (IPPQ). This small molecule bound to CaVβ and inhibited its coupling with N-type voltage-gated calcium (CaV2.2) channels, leading to a reduction in CaV2.2 currents in rat dorsal root ganglion sensory neurons, decreased presynaptic localization of CaV2.2 in vivo, decreased frequency of spontaneous excitatory postsynaptic potentials and miniature excitatory postsynaptic potentials, and inhibited release of the nociceptive neurotransmitter calcitonin gene–related peptide from spinal cord. IPPQ did not target opioid receptors nor did it engage inhibitory G protein–coupled receptor signaling. IPPQ was antinociceptive in naive animals and reversed allodynia and hyperalgesia in models of acute (postsurgical) and neuropathic (spinal nerve ligation, chemotherapy- and gp120-induced peripheral neuropathy, and genome-edited neuropathy) pain. IPPQ did not cause akinesia or motor impairment, a common adverse effect of CaV2.2 targeting drugs, when injected into the brain. IPPQ, a quinazoline analog, represents a novel class of CaV2.2-targeting compounds that may serve as probes to interrogate CaVα–CaVβ function and ultimately be developed as a nonopioid therapeutic for chronic pain.

1. Introduction

The N-type voltage-gated calcium (CaV2.2) channels are critical determinants of increased neuronal excitability and neurotransmission accompanying persistent neuropathic pain.^9,13,18,94 Expressed in the presynaptic termini of primary afferent nociceptors in the spinal cord,⁸³ CaV2.2 represent a control point for synaptic activity. The potential of targeting CaV2.2 has been demonstrated with attenuation of allodynia by CaV2.2-blocking conotoxins⁷² and the altered pain behavior of CaV2.2 knock-out mice.^39,40 CaV2.2 remains a high-value target with several companies developing CaV2.2-targeted compounds.^42,93 Ziconotide (primary alternative to morphine [Prialt])⁴⁷ and gabapentin (Neurontin) directly target different elements of the CaV2.2 complex. These drugs, however, present with problematic side effects, difficult dosing regimens, and have high number needed to treat values.²⁵ Therefore, the development of novel CaV2.2-targeted drugs with improved efficacy and therapeutic index is highly desirable.⁶⁵ We have advanced a strategy targeting protein interactions that modulate CaV2.2 as an alternative to direct channel blockade.^{11,23,26,85,87} Targeting channel regulation may potentially lessen many of the adverse side effects associated with direct channel block. Supporting this concept, we reported that targeting CaV2.2 indirectly with the CBD3 peptide derived from an ancillary regulatory protein (collapsin response mediator protein 2) did not affect memory retrieval, motor coordination, and depression-associated behaviors¹¹ and was not rewarding/addictive.²⁶

Here, we focused on the β subunits of voltage-gated calcium (CaV) channels. CaVβs are cytoplasmic proteins encoded by 4 different genes (β_1–4), including multiple splice variants.²² Their signature functions are to increase the surface expression of the pore-forming α subunit of Ca²⁺ channels^22,68 and regulate biophysical properties of the α subunit of Ca²⁺ channels.^22,68 The sum effect of these ancillary roles is to increase the amount of Ca²⁺ influx within cells. In neurons, Ca²⁺ influx triggers neurotransmitter release, where the amount of transmitter released from a presynaptic terminal is dependent on the amount of Ca²⁺ entering the terminal.^4,70 Further involvement of β subunits in regulating synaptic transmission is also supported by evidence of protein–protein interactions of the CaVβ subunits with the synaptic vesicle release machinery.⁸¹ Thus, the subunit is part of the transmitter release site core complex, at the center of which resides the α subunit of the Ca²⁺ channel.³⁸ Altered CaVβ subunit expression will modulate the function of the α subunit of the calcium channel and can underlie pathological neuronal transmission. In neuropathic pain, CaVβ3 expression is increased and leads to augmented high-voltage-gated Ca²⁺ channel function in small diameter sensory neurons.⁴³ This pathological increase of CaVβ3 amplifies spinal nociceptive neurotransmission and is sufficient to induce pain.⁴³ Thus, pharmacotherapeutic approaches targeting the CaVβ–CaVα interface could restore physiological Ca²⁺ homeostasis and be beneficial for chronic pain management.

We used a computational approach to target a protein–protein interface important for voltage-gated calcium channel activity. We report the identification of the small molecule 2-(3,5-dimethylisoxazol-4-yl)-N-((4-((3-phenylpropyl)amino)quinazolin-2-yl)methyl)acetamide (IPPQ) targeting the CaVβ–CaVα interface. Here, we demonstrate that IPPQ (1) specifically inhibits CaV2.2 (over other subtypes) in sensory neurons, (2) acts at presynaptic sites, (3) blunts release of the pronociceptive neurotransmitter calcitonin gene–related peptide (CGRP), (4) reverses thermal hyperalgesia and mechanical allodynia in acute and neuropathic models of pain, and (5) has no effect on motor activity. This quinazoline analog may be used to mitigate chronic pain by controlling CaV2.2 function.

2. Materials and methods

2.1. Animals and reagents

Adult male Sprague–Dawley rats (225–250 g; Envigo, Indianapolis, IN) were housed in a pathogen-free environment under controlled temperature (23 ± 3°C) and light (12-hour light/12-hour dark cycle; lights on 07:00–19:00) conditions with standard rodent chow and water available ad libitum. All experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the College of Medicine at the University of Arizona. Procedures were conducted in compliance with the Guide for Care and Use of Laboratory Animals published by the National Institutes of Health and the ethical guidelines of the International Association for the Study of Pain. Animals were randomly assigned to the experimental groups, and treatments and experimenters were blinded to that information. Unless otherwise stated, reagents were purchased from Sigma (St. Louis, MO).

2.2. In silico docking and software for structural representation

Virtual screening workflow was performed in Schrodinger Glide molecular modeling software using the X-ray crystal structure of the complex between CaVβ2a subunit and a peptide of the α1c subunit (PDB code:1 T0J).⁷⁹ The α1c subunit peptide was removed, and its binding site was used for docking of 50,000 drug-like small molecule library (molecular weight ≤ 500 Da) available from ChemBridge Inc. The resulting complexes were ranked using Glide XP score and other energy-related terms. Forty-nine compounds were analyzed for their binding in the α-binding pocket on CaVβ2a. Selected compounds were purchased and tested for activity.

2.3. Protein purification

Codon optimized sequence (Escherichia coli) for CaVβ2a residues 203 to 425 was inserted between the restriction sites EcoRI-SalI in pET-28a(+) plasmid (synthesized by Genscript). BL21 cells expressing CaVβ2a were resuspended in 50-mM sodium phosphate pH 7.5, 500-mM NaCl, 10% glycerol, supplemented with complete EDTA-free protease inhibitors (Roche, Basel, Switzerland). Disruption of the bacteria was performed by 2 rounds of high-pressure homogenization at 10,000 PSI with a LM10 microfluidizer (Microfluidics, Westwood, MA), and the lysate was centrifuged 45 minutes at 15,000g at 4°C. The supernatant was loaded on a His-Trap column (GE Healthcare, Uppsala, Sweden) equilibrated with 50-mM HEPES pH 7.5, 10-mM imidazole, 500-mM NaCl, 10% glycerol, 0.5-mM DTT. After a washing step with 50-mM HEPES pH 7.5, 50-mM imidazole, 500-mM NaCl, 10% glycerol, 0.5-mM DTT, CaVβ2a was eluted with a gradient of imidazole. The fractions of interest were loaded on a HiLoad Superdex size exclusion column (GE Healthcare, Chicago, IL) and eluted with a gradient of imidazole. A Pierce assay was used to determine protein concentration using bovine serum albumin as standard. Protein purity was confirmed with SDS-PAGE. The proteins were flash frozen in liquid nitrogen and stored at −80°C.

2.4. Saturation transfer difference nuclear magnetic resonance spectroscopy

One-dimensional ¹H STD NMR⁵⁰ for IPPQ (1 mM in 100-mM sodium phosphate pH 6.9, 10% D₂O) in the absence of protein was acquired at 25°C on a Varian Inova 600-MHz spectrometer equipped with a Varian cold TR/PFG probe. 1D ¹H saturation transfer difference nuclear magnetic resonance (STD NMR)⁴⁹ spectra with a spectral width of 12 ppm were collected for samples containing 100-μM IPPQ and CaVβ2a in phosphate buffer. Saturation transfer difference nuclear magnetic resonance spectra were collected with a spectral width of 12 ppm, 16 K data points, and 3-second repetition delay. A saturation of the protein was achieved by a 2-second train of selective 50-ms Gaussian pulses centered at 0.74 ppm (on resonance) and 30 ppm (off resonance). A 20-ms spin lock was used to suppress the protein signal, followed by the double PFG spin echo to remove residual water signal. We acquired 512 scans per experiment. The on-resonance and off-resonance spectra were acquired interleaved, and the difference spectrum was acquired by phase cycling. Spectra processing and analysis were performed with the VNMRJ 3.2 and MestReNova 7.1.

2.5. Microscale thermophoresis

Microscale thermophoresis (MST), a technique monitoring the thermophoretic movement of molecules in optically generated microscopic temperature gradients to allow for analysis of biomolecular interactions was performed as before.^86,87 In MST, increasing concentrations of unlabeled ligand are mixed with the fluorescently labeled biomolecule, kept at constant concentration. Purified Beta2-CaV-His was fluorescently labeled using the His-Tag labeling kit RED-Tris-NTA second generation (Nanotemper, München, Germany) according to the manufacturer’s instructions. Briefly, Beta2-CaV-His was diluted to 200 nM in PBS supplemented with 0.1% Tween-20 (PBS-T buffer) and 0.1% PEG 8000 and mixed with one molar equivalent of the fluorescent NT-647-His-labeling dye and incubated for 30 minutes at room temperature. Labeled Beta2-CaV was then centrifuged at 15,000g for 10 minutes at 4°C. A total of 25 nM of labeled Beta2-CaV-His protein were mixed with increasing concentrations of AID-CaV2.2 peptide in PBS-T, 0.1% PEG 8000 buffer, and incubated 10 minutes at room temperature. The thermophoresis measurements were performed on a Monolith NT.115 (Nanotemper) using MST premium capillaries, at 100% LED at high MST power. Data analysis was performed with the MO Affinity Analysis software (Nanotemper) using the Kd model (standard fitting model derived from law of mass action).

2.6. Primary cell culture of dorsal root ganglia

Dorsal root ganglion (DRG) from all vertebral levels was harvested and dissociated using methods described previously.^56,59,88 Briefly, DRG were dissected from naive animals and treated with collagenase, type I (5 mg/mL), and neutral protease (3.125 mg/mL) in bicarbonate-free DMEM for 45 minutes (Worthington Biochemical, Lakewood, NJ). Mechanical trituration in culture medium (DMEM with 10% fetal bovine serum, penicillin [100 mg/mL], streptomycin [100 U/mL], normocin [0.8 μg/mL], and nerve growth factor [30 ng/mL], Life Technologies, Carlsbad, CA) was then used to dissociate the cells. The cells were then plated in 20 μL of medium on poly-L-lysine/laminin-coated coverslips (BD bioscience, San Jose, CA) and incubated for up to 2 hours before culture media was added to the wells up to a final volume of 1 mL. The cells were then left undisturbed at 37°C (with 5% CO₂) for 12 to 18 hours to allow adhesion.

2.7. Calcium imaging

Dorsal root ganglion neurons were incubated with 3-μM Fura-AM (Cat#F-1221; Life Technologies, stock solution prepared at 1 mM in DMSO, 0.02% pluronic acid, Cat#P-3000MP; Life Technologies) for 30 minutes at 37°C to allow monitoring of changes in intracellular calcium ([Ca²⁺]_c). Recordings were performed in standard Tyrode’s solution (119-mM NaCl, 2.5-mM KCl, 2-mM MgCl₂, 2-mM CaCl₂, 25-mM Na HEPES, 30-mM glucose, pH 7.4,~310 mOsm) at room temperature (~23°C). A 1-minute baseline was acquired followed by brief stimulations (15 seconds) either with a solution (81.5-mM NaCl, 40-mM KCl, 2-mM CaCl₂, 2-Mm MgCl₂, 25-mM HEPES, 30-mM glucose, pH 7.4, ~310 mOsm) to activate low-voltage-activated Ca²⁺ channels (ie, T-type Ca²⁺ channels [CaV3.1-CaV3.3] and R-type [CaV2.3]), or a solution (32-mM NaCl, 90-mM KCl, 2-mM MgCl₂, 2-mM CaCl₂, 25-mM HEPES, 30-mM glucose, pH 7.4, ~310 mOsm) to activate high-voltage-activated Ca²⁺ channels (ie, P/Q-type [CaV2.1], R-type [CaV2.3], N-type [CaV2.2], and L-type [CaV1.1-CaV1.4]). Acquisition was performed using an Nikon Eclipse Ti-U (Melville, NY) inverted microscope, equipped with an objective photometrics-cooled CCD camera CoolSNAP ES² (Roper Scientific, Tucson, AZ) and a Nikon Super Fluor ×10 0.50 objective and operated by the NIS Elements software (version 4.20; Nikon Instruments, Melville, NY). The excitation light was delivered by a Lambda-LS system (Sutter Instruments, Novato, CA). The excitation filters (340 ± 5 and 380 ± 7 nm) were controlled by a Lambda 10 to 2 optical filter change (Sutter Instruments). Fluorescence was recorded through a 505-nm dichroic mirror at 535 ± 25 nm. To reduce photo-bleaching, pictures were taken every ~10 seconds over the duration of the experiment with the minimal exposure time providing sufficient image quality. [Ca²⁺]_c changes were followed using the F₃₄₀/F₃₈₀ ratio calculated after background subtraction from both channels. IPPQ was applied at 20 μM overnight for the initial screening.

2.8. Constellation pharmacology

Constellation pharmacology^56,75,76,88 was performed as described in 2.7, but after baseline recording in Tyrode’s solution, DRG neurons were sequentially stimulated for 15 seconds using the following receptor agonists: 400-nM menthol, 50-μM histamine, 10-μM adenosine triphosphate (ATP), 200-μM allyl isothiocyanate (AITC), 1-mM acetylcholine (Ach), and 100-nM capsaicin diluted in standard Tyrode’s solution. Cells were washed 5 minutes in standard Tyrode’s solution between each agonist, and cell viability was assessed by depolarization-induced Ca²⁺ influx using the excitatory solution for activating high-voltage-activated Ca²⁺ channels at the end of the constellation pharmacology recording. Agonist application was automated using the ValveBank Controller (Automate Scientific, San Francisco, CA). A cell was classified as a “responder” if its maximum fluorescence ratio for 340/380 nm exceeded 10% of its baseline value (calculated as the average fluorescence ratio during the 30 seconds immediately preceding application of the receptor agonist solution).

2.9. MOR-CHO, DOR-CHO, and KOR-CHO cell lines and cell culture

The MOR-CHO cell line was obtained from PerkinElmer (#ES-542-C). Creation of the DOR cell line was reported elsewhere.⁷⁴ For the KOR cell line, an N-terminal 3X-hemagglutinin tagged human KOR expression clone from Genecopoeia was electro-porated into parental CHO cells and selected with 500-μg/mL G418. KOR-expressing cells were subjected to live cell labeling with anti-HA-Alexa488 antibody and triage of the top 2% of the population by flow cytometry was performed. Characterization was performed using immunocytochemistry and Western blot to verify receptor expression and expected signal transduction activation. All 3 cell lines were characterized by saturation radioligand binding with ³H-diprenorphine. Measured K_D were used in competition binding experiments to calculate the K_I (MOR = 5.23 nM; DOR = 0.93 nM; KOR = 1.81 nM; all the mean of N ≥ 3 independent experiments). Cell cultures was performed 50% DMEM 50% F12 media supplemented with 10% heat-inactivated FBS and 1X penicillin/streptomycin (Gibco; Thermo Fisher) at 37°C with 5% CO₂; propagation cultures were further maintained in 500-μg/mL G418. Cultures were propagated for no more than 20 passages before discarding. Cell pellets for experiments were grown in 15-cm plates and harvested in dPBS (Ca²⁺- and Mg²⁺-free) with 5-mM EDTA and stored at −80°C before use.

2.9.1. Competition radioligand binding

Competition radioligand binding experiments were performed as previously reported.⁷⁴ Cell membrane preparations of MOR-, DOR-, or KOR-CHO were combined with a constant concentration of ³H-diprenorphine (MOR = 5.33 nM; DOR = 1.43 nM; KOR = 1.95 nM) and a concentration curve of competitor ligand. Reactions were performed in 200 μL in 96 well plates for 60 minutes at room temperature. Termination of the reaction was obtained by rapid filtration through 96-well format GF/B filter plates (PerkinElmer, Waltham, MA) with cold water, washing, drying, adding of, and Microscint PS (PerkinElmer). Plate reading was performed on a MicroBeta2 96 well format 6 detector scintillation counter (PerkinElmer). The data were normalized to the specific binding caused by ³H-diprenorphine alone (100%) or nonspecific binding determined by a 10-μM concentration of known competitor ligand (0%; MOR = naloxone; DOR = SNC80; KOR = naloxone). K_I values were calculated using the IC₅₀ of each competitor ligand and the previously established K_D of ³H-diprenorphine in each cell line (GraphPad Prism 7.0).

2.10. Whole-cell voltage-clamp electrophysiology

HEK293 cells and acutely dissociated DRG neurons were used for recordings. Experiments were performed at room temperature by using an EPC10 Amplifier-HEKA. Electrodes were pulled from thin-walled borosilicate glass capillaries (Warner Instruments, Hamden, CT) with a P-97 electrode puller (Sutter Instrument) to reach final electrode resistances of 1 to 3 MΩ when filled with internal solutions. External solution was 140-mM Et₄N MeSO₃, 10-mM HEPES, and 5-mM BaCl₂ at pH 7.3 and internal solution 135-mM CsMeSO₃, 5-mM CsCl, 5-mM EGTA, 1-mM MgCl₂, 4-mM MgATP, and 10-mM HEPES at pH 7.3 for HEK293 cells. For DRGs, external solution (at ~315 mOsm, in mM): 10 glucose, 10 HEPES, 10 BaCl_2, 30 TEA-Cl, 110 N-methyl-D-glucamine (NMDG) (pH 7.2 with KOH); internal solution: (at ~305 mOsm, in mM): 5 BAPTA, 5 Mg-ATP, 10 HEPES, 150 CsCl₂ (pH 7.2 with KOH). Isolation of specific current contributions of each calcium channel subtypes was obtained by treating the cells with inhibitors of all other Ca²⁺ channel subtypes. The following compounds were used: ω-conotoxin GVIA (500 nM, N-type voltage-gated Ca²⁺ channel blocker),²⁴ SNX482 (200 nM, R-type voltage-gated Ca²⁺ channel blocker),⁶²ω-agatoxin (200 nM, P/Q-type voltage-gated Ca²⁺ channel blocker),⁵² nifedipine (10 μM, L-type voltage-gated Ca²⁺ channel blocker), and TTA-P2 (1 μM, T-type voltage-gated Ca²⁺ channel blocker).¹⁷ Cells were subjected to current–voltage, inactivation, and use dependence protocols.

2.10.1. Voltage-clamp protocols

Ca²⁺ channel activation was analyzed as a function of current vs voltage, and peak current density was recorded, typically near ~0 to 10 mV. In the current-voltage protocol, DRG neurons were held at resting membrane potential for 5 ms before depolarization by 200-ms voltage steps from −70 mV to +60 mV in 10-mV increments. Normalization of currents to each cell’s capacitance (pF) was performed to allow for collection of current density data to analyze activation of Ca²⁺ channels as a function of current vs voltage as well as peak current density. Dorsal root ganglion neurons were held at −90 mV for 20 ms for the inactivation protocol before depolarizating by 1.5-second voltage steps from −100 to +10 mV in 10-mV increments, followed by a 20-ms pulse at 10 mV before returning to −90 mV for 20 ms. For the use dependence protocol, DRGs were submitted 30 times to this protocol: baseline held at −70 mV for 5 ms before depolarization to 0-mV voltage step lasting 200 ms and followed by a return to −70 mV for 5 ms.

2.11. Preparation of spinal cord slices

Young rats (12–21 days postnatal) were deeply anesthetized using isoflurane, as described previously.⁹⁶ For spinal nerve blocking, 2% lidocaine in a final volume 0.3 mL was injected to both sides of lumbar vertebrae L4 to 5. Laminectomy was performed from midthoracic to low lumbar levels. The spinal cord was harvested and placed in chilled modified artificial cerebrospinal fluid (80-mM NaCl, 2.5-mM KCl, 1.25-mM NaH₂PO₄, 0.5-mM CaCl₂, 3.5-mM MgCl₂, 25-mM NaHCO₃, 75-mM sucrose, 1.3-mM ascorbate, 3.0-mM sodium pyruvate, pH 7.4, ~310 mOsm) oxygenated with 95% O₂ and 5% CO₂. Transverse 350-mm-thick slices were cut on a vibratome (VT1200S; Leica, Nussloch, Germany). Slices were then placed in an oxygenated recording solution (125-mM NaCl, 2.5-mM KCl, 2-mM CaCl₂, 1-mM MgCl₂, 1.25-mM NaH₂PO₄, 26-mM NaHCO₃, 25-mM D-glucose, 1.3-mM ascorbate, 3.0-mM sodium pyruvate, pH 7.4, ~320 mOsm) for at least 1 hour at RT. The slices were then positioned in the recording chamber and with continuous perfusion of oxygenated recording solution at 3 to 4 mL/minute before electrophysiological recordings at RT.

2.11.1. Electrophysiological recording in spinal cord slices by whole-cell patch clamp

Substantia gelatinosa neurons were identified in the slices with an infrared differential interference contrast video microscopy on an upright microscope (FN1; Nikon, Tokyo, Japan) equipped with a 3 40/0.80 water-immersion objective and a CCD camera. Electrodes were obtained by pulling thin-walled borosilicate glass capillaries (Warner Instruments) with a P-97 electrode puller (Sutter Instrument) to reach a final electrode resistances of 6 to 10 MΩ when filled with potassium gluconate–based internal solutions (120-mM potassium gluconate, 20-mM KCl, 2-mM MgCl², 2-mM Na2-ATP, 0.5-mM Na-GTP, 20-mM HEPES, 0.5-mM EGTA, pH 7.28, ~310 mOsm). Membrane potential was held at −60 mV using PatchMaster software and a dual channel EPC10-HEKA amplifier (Lambrecht, Germany).

A giga-ohm seal was then formed, and whole-cell configuration was obtained in voltage-clamp mode. To record spontaneous excitatory postsynaptic currents (sEPSCs), bicuculline methiodide (10 μM) and strychnine (1 μM) were added to the recording solution to block γ-aminobutyric acid–activated and glycine-activated currents. To record miniature EPSCs (mEPSCs), tetrodotoxin (TTX, 1 μM) was also added to the recording solution to block the action potentials. Hyperpolarizing step pulses (5 mV in intensity, 50 ms in duration) were periodically delivered to monitor the series resistance (15–25 MΩ), and recordings were discontinued if the series resistance changed by more than 20%. For each neuron, sEPSCs and mEPSCs were recorded for a total duration of 2 minutes. Drugs were dissolved in artificial cerebrospinal fluid or 0.1% DMSO, and the solution was applied through perfusion. The recordings were obtained within 30 minutes to 3 hours of drug perfusion; no difference was observed between recordings performed earlier vs later in the window after drug perfusion (data not shown). Currents were filtered at 3 kHz and digitized at 5 kHz. Data were further analyzed by the Mini-Analysis Program (Synatosoft Inc, Decatur, GA) to provide spreadsheets for the generation of cumulative probability plots. The amplitude and frequency of sEPSCs and mEPSCs were compared between neurons from animals in control and IPPQ groups.

2.12. Calcitonin gene–related peptide release from lumbar slices

Rats were placed under deep anesthesia with 5% isofluorane and then decapitated. Cervical and lumbar incisions were made to expose the spinal cord. Hydraulic pression was applied with a saline-filled syringe inserted into the lumbar vertebral foramen to extrude the spinal cord. The lumbar region was cut for use in the CGRP release assay. Baseline treatments (#1 and #2) consisted of bathing in Tyrode’s solution. A total of 90-mM KCl were used for excitatory solution (as in calcium imaging experiments) and were paired with the treatment for fraction #4. Four hundred–microliter fractions were collected every 10 minutes for CGRP release measurements. Samples were flash frozen and stored at −20°C. IPPQ (20 μM) or vehicle (0.9% saline) was added to the pretreatment and cotreatment fractions (#3 and 4) for 10 minutes. Enzyme-linked immunosorbent assay (Cat# 589001, Cayman Chemical, Ann Arbor, MI) was used to determine the concentration of CGRP released in the buffer.

2.13. Synapse enrichment and fractionation

Adult rats were anesthesized using isoflurane and decapitated. Spinal cords were removed, and the dorsal horn of the spinal cord was dissected because this structure contains the synapses arising from the DRG. Synaptosomes isolation was performed as described previously.⁶⁴ Fresh tissues were homogenized in ice-cold sucrose 0.32 M, HEPES 10 mM, pH 7.4 buffer. Insoluble material was pelleted by centrifugation at 1000g for 10 minutes. The supernatant was collected and centrifuged at 12000g for 20 minutes to pellet a crude membrane fraction. The membrane fraction was resuspended in a hypotonic buffer (4-mM HEPES, 1-mM EDTA, pH 7.4). Resulting synaptosomes were pelleted by centrifugation at 12000×g for 20 minutes and incubated in 20-mM HEPES, 100-mM NaCl, 0.5% triton X, pH 7.2) for 15 minutes and centrifuged at 12000g for 20 minutes. Sample incubations were performed on ice, and all centrifugations were performed at 4°C. Non-postsynaptic density (non-PSD) membrane fraction, sometimes referred to as the triton-soluble fraction, was obtained from the supernatant. The pellet containing the PSD fraction was solubilized in RIPA buffer (50-mM Tris-HCl, pH 7.4, 50-mM NaCl, 2-mM MgCl₂, 1% [vol/vol] NP40, 0.5% [mass/vol] sodium deoxycholate, 0.1% [mass/vol] SDS) as described previously.²⁶ All buffers were supplemented with phosphatase inhibitors (Cat# B15002, Bimake) and protease inhibitors (Cat# B14002; Bimake, Houston, TX). Non-PSD and PSD fraction’s integrity was verified by immunoblotting. PSD95 was enriched in the PSD fractions while synaptophysin was enriched in non-PSD fractions. BCA protein assay was used to determine protein concentrations.

2.14. Immunoblot preparation and analysis

Indicated samples were loaded on 4% to 20% Novex gels (Cat# EC60285BOX; Thermo Fisher Scientific, Waltham, MA). Proteins were transferred at 120 V for 1 hour using TGS buffer (25-mM Tris pH = 8.5, 192-mM glycine, 0.1% (mass/vol) SDS), 20% (vol/vol) methanol as transfer buffer to 0.45-μm polyvinylidene difluoride membranes (Cat# IPVH00010; Millipore, Billerica, MA), preactivated in 100% methanol. A blocking step was then performed in TBST (50-mM Tris-HCl, pH 7.4, 150-mM NaCl, 0.1% Tween 20), 5% (mass/vol) nonfat dry milk for 1 hour at room temperature, then incubated separately with primary antibodies: CaV2.2 (Cat# TA308673; Origene, Rockville, MD), PSD95 (Cat# MA1–045, Thermo Fisher scientific), synaptophysin (Cat# MAB5258, Thermo Fisher scientific, San Diego, CA), or flotillin (Cat# F1180; Sigma, St. Louis, MO) in TBST, 5% (mass/vol) BSA, at 4°C overnight. After incubation in horseradish peroxidase–conjugated (HRP) secondary antibodies (Jackson Labs) revelation was performed using enhanced luminescence (WBKLS0500; Millipore) before exposure to photographic film. Films were scanned and quantified using Un-Scan-It gel scanning software (version 6.1; Silk Scientific Inc).

2.15. Indwelling intrathecal catheter

Rats were anesthetized with ketamine/xylazine 80/12 mg/kg intraperitoneally (i.p.) (Sigma), and their head was placed in a stereotaxic frame. The cisterna magna was exposed and incised. As previously reported, an 8-cm catheter (PE-10; Stoelting, Wood Dale, IL) was implanted, terminating in the lumbar region of the spinal cord.⁸⁹ Catheters were sutured (using 3–0 silk sutures) into the deep muscle and externalized at the back of the neck. Autoclips were used to close the skin, and other surgeries were performed after a 5- to 7-day recovery period.

2.16. Behavioral measurements

2.16.1. Measurement of thermal withdrawal latency

The method of Hargeaves was used. Rats, either naive or after experimentally induced pain (see below), were acclimated to plastic boxes placed on a clear glass plate maintained at 30°C. As per the Hargreaves et al.²⁸ method, a radiant heat source (high-intensity projector lamp) was focused onto the hind-paw plantar surface. When the paw was withdrawn, a motion detector halted the stimulus and recorded the time. A 33.5-second maximal cutoff was used to avoid tissue damage.

2.16.2. Measurement of allodynia

Tactile allodynia (ie, a decreased threshold to paw withdrawal after probing with normally innocuous mechanical stimuli) was assessed by measuring rats’ paw withdrawal threshold in response to probing with a series of fine calibrated filaments (von Frey). Rats were placed in suspended plastic cages with wire mesh floor, and each von Frey filament was applied perpendicularly to the plantar surface of the paw. The “up-and-down” method (sequential increase and decrease of the stimulus strength) was used to determine the withdrawal threshold Dixon’s nonparametric method was used for data analysis, as described by Chaplan et al.^15,21 Data were expressed as the mean withdrawal threshold.

2.17. Pain models

2.17.1. Paw incision model of postoperative pain

A rodent model of surgical pain was generated by plantar incision as previously described.¹⁰ Male rats (Sprague–Dawley) were anesthetized with gas isoflurane. The plantar side of the left hind paw was scrubbed 3 times with betadine followed by 70% alcohol. A 1-cm-long incision, starting 0.5 cm from the heel and extending toward the toes, was made using a number 11 blade. Incision cut through the skin and fascia of the plantar aspect of the hind paw, including the underlying muscle. A longitudinal incision was made in the plantaris muscle leaving both muscle origin and insertion intact. Gentle pressure was applied to facilitate hemostasis, and the skin was closed with 2 mattress sutures using 5–0 nylon attached to a curved needle. Rats were injected with gentamicin (8 mg/kg, subcutaneously) and were allowed to recover from anesthesia before being returned to their home cage. Sham rats were anesthetized, and the left hind paw was scrubbed as previously described, but no incision was made. After surgery, rodents were allowed to recover for 24 hours before measurements of paw withdrawal thresholds and paw withdrawal latencies.

2.17.2. Spinal nerve ligation model of neuropathic pain

Nerve ligation injury induces signs of neuropathic dysesthesias that include thermal hypersensitivity and tactile allodynia. Nerve operations were performed 5 days after intrathecal catheter implantation. After anesthesia with 2% isoflurane in O₂ at 2 L/minute, the skin over the caudal lumbar region was incised, and the muscles retracted. Spinal nerves located at L5 and L6 were exposed and carefully isolated. A tight nerve ligation with 4–0 silk was placed distal to DRG without limiting the use of the left hind paw of the animal. All animals were given 7 days to recover before any testing. Any animal exhibiting signs of motor deficiency was euthanized.

2.17.3. Elevated plus maze test of anxiety-related behaviors

One hour after i.p. injection of IPPQ or its vehicle, mice (CD1) were placed at the center of the elevated plus maze, where the 2 arms intersect and allowed to explore the maze freely for 10 minutes. The apparatus dimensions were as follows: arm length: 35 cm; arm width: 5 cm; and wall height for closed walls: 15 cm. The whole apparatus is elevated 50 cm above ground. Animals movements were recorded by a camera (C310; Logitech, Lausanne, Switzerland) placed above the apparatus and tracked and analyzed using the Any-MAZE software (v6.05; Stoelting). Entry into a zone was defined as 70% of the animal’s body being in the zone. The main parameter analyzed was the time spent into the open arms of the apparatus.

2.17.4. Acute pain models

The tail flick response was evoked either by a light beam (irradiated heat) or by immersion of one-third of the tail in water warmed at 52°C. For the hot-plate test, mice were placed on a metal plate (52°C), and a timer was started. Timer was stopped after the first positive response: jumping, hind paw licking, or hind paw flinching. A 10-second cutoff time was used for tail-flick tests, and a 20-second cutoff time was used for hot-plate test.

2.17.5. HIV-induced sensory neuropathy

A model of HIV-induced sensory neuropathy was produced by intrathecal (i.t.) administration of the glycoprotein gp120⁵¹ from the HIV-1 envelope. Baseline behavioral measurements were obtained 7 days after implantation of an i.t. catheter. Rats were then randomly assigned to 2 groups, treated of vehicle. On days 10, 12, and 14, rats were injected i.t. with either 300 ng of gp120 (Cat#4961, HIV-1 BaL gp120 recombinant protein, NIH-AIDS Reagent program) in a final volume of 20 μL in 0.9% saline and 0.1% BSA or vehicle alone.

2.17.6. Chemotherapy-induced peripheral neuropathy through paclitaxel

Rats were administered pharmaceutical-grade paclitaxel (Cat# P-925–1; Goldbio, Olivette, MO) resuspended at a concentration of 2 mg/mL in 30% 1:1 Cremophor EL: ethanol, 70% saline at 2 mg/kg i.p.. Based on the protocol described by Polomano et al.,⁶⁶ injection were repeated every other day for a total of 4 injections (days 0, 2, 4, and 6), resulting in a final cumulative dose of 8 mg/kg. Animals developed mechanical hyperalgesia within 10 days after the first injection. No abnormal spontaneous behavioral changes in the rats were noted during or after the treatment.

2.17.7. In vivo transfection of CRISPR plasmids for Cas9-mediated editing of Nf1

As described before,^55,58,60 NF1-related pain was induced by CRISPR-/Cas9-mediated editing of the exon 39 of the gene Nf1. The specific gRNA sequence was cloned in the plasmid pLCRISPR-EFS-tRFP²⁹ and was validated in Ref. 55. Lentiviral particle were produced by Viracore (UCSF) and provided concentrated at a titer higher than 10⁷ infectious particles per ml. At day 0, 5 μL of the lentiviral concentrate was injected intrathecally in male Sprague–Dawley rats. After 10 days, rats developed thermal hyperalgesia as described before.^55,58,60

2.18. Rotarod

Rats were trained to walk on a rotating rod (10 rev/minute; Rotamex 4/8 device) with a maximal cutoff time of 180 seconds. Training was initiated by placing the rats on a rotating rod and allowing them to walk until they either fell off or 180 seconds was reached. This process was repeated 6 times. Before treatment (intracerebroventricular [i.c.v.] injection of IPPQ or saline), the rats were run once on a moving rod to establish a baseline value. Assessment consisted of placing the rats on the moving rod and timing until either they fell off or reached a maximum of 180 seconds.

2.19. Statistics

Statistical analyses were performed using GraphPad Prism 7 (GraphPad, La Jolla, CA). Unless stated otherwise, data were obtained from a minimum of 6 independent biological replicates. All data represent the mean ± SEM. The statistical significance of differences between mean values was calculated using either parametric or nonparametric Student t test, analysis of variance followed by post hoc comparisons (Tukey) using GraphPad Prism 7. To determine the type of test to use, all data were first tested for a Gaussian distribution using a D’agostino–Pearson test (GraphPad Prism 7 Software). Statistical significance threshold was set at α < 0.05.

3. Results

3.1. A quinazoline compound (IPPQ) targets the CaVα–CaVβ interface and inhibits depolarization-evoked Ca²⁺ influx in sensory neuron subpopulations

To identify small molecules that disrupt the CaVα–CaVβ interaction, we used a virtual screening approach against the crystal structure of the complex formed between the CaVβ2a subunit and a peptide of the α1c subunit (Fig. 1A).⁷⁹ The alpha-interaction domain (AID) peptide was removed, and its binding site was used for docking 50,000 drug-like small molecules (molecular weight ≤ 500 Da) commercially available from ChemBridge Inc. The resulting complexes were ranked using Glide score and other energy-related terms.²⁷ Forty-nine compounds docked into the α-binding pocket on CaVβ2a. The 49 compounds were screened by Ca²⁺ imaging for their ability to inhibit depolarization-induced calcium influx in rat DRG neurons. Of these compounds, 13 were either insoluble or killed the neurons, and 11 compounds inhibited Ca²⁺ influx by ≥50%. One docked compound, IPPQ, engaged all 3 (V241, I343, and N390) AID “hotspot” residues (Fig. 1B). The structure of IPPQ aligned with 3 residues (W440, I441, and Y437) from the AID (Figs. 1C and E) and specifically bound CaVβ2a protein in saturation transfer difference nuclear magnetic resonance (STD-NMR) (Fig. 1D).

Figure 1.

Structure-guided high-throughput screen of CaVα–CaVβ interaction inhibitors identifies IPPQ (A) Space-filling model of the CaVβ2a subunit showing its α-binding pocket (ABP). A ribbon representation of the CaVα-interaction domain (AID, purple) and the AID’s key interaction residue (M245) are highlighted. (B) Ball-and-stick representation of key ABP residues is shown (ie, V241, I343, and N390). (C) A 3-dimensional representation of IPPQ (red) overlaid on the AIDs reveals shared chemical features, especially at AID residues W440, I441, and Y437. (D) 1D ¹H STD NMR showing on-resonance difference spectrum for IPPQ (red); only regions that yielded a STD signal in the presence of IPPQ are plotted. The 2-dimensional chemical structure of compound IPPQ shows that region a corresponds with the STD signal highlighted. (E) Amino acid alignments of AID residues of the indicated calcium channel subtypes. Residues different from the CaV2.2 core AID sequence are denoted in orange. Asterisks denote the key residues conserved in all AID sequences involved in the interaction with the beta subunits. Microscale thermophoresis was used to determine dissociation constants for binding of AID-CaV2.2 (F), CaV2.1 (G), CaV1.1 (H), CaV1.2/3/4 (I) (500–0.01 nM), and CaV2.3 (J) to Beta2-CaVβ2-His (25 nM) in the presence (filled red circles) or absence of 10-μM IPPQ (open squares). Data are presented as mean ± SEM. The data were all fit with a one-site binding model, and the K_d and r² values are indicated. The data for Beta2-CaVβ2-His and AID-CaV2.1 in the presence of vehicle also fit well with a 2-site binding model (r² = 0.84) but yielded K_ds that were a 1000-fold higher than the one-binding site equation. In the presence of IPPQ, the data in panel (G) had a better fit (r² = 0.74) but yielded K_ds values of 2.3 nM (for the high affinity site) and ~81 mM (for the low-affinity site). (K) Concentration–response curve illustrates that IPPQ inhibits membrane depolarization-evoked Ca²⁺ influx in DRGs (n > 150 cells per concentration from 3 independent platings) in a concentration-dependent manner. IPPQ was applied overnight (15–18 hours). AID, alpha interaction domain; CaV, voltage-gated calcium.

Microscale thermophoresis—a method that permits analysis of biomolecular interactions—demonstrated that IPPQ reduced the binding of CaVβ2-His to the AID-CaV2.2 peptide from a K_d of 12 to 208 nM (Fig. 1F), reduced the binding of CaVβ2-His to the AID-CaV2.1 peptide from a K_d of 18.3 to 93 nM (Fig. 1G), and reduced the binding of CaVβ2-His to the AID-CaV1.1 peptide from a K_d of 24 nM to 51 nM (Fig. 1H) while binding of CaVβ2-His to the AID peptides from CaV1.2/CaV1.3/CaV1.4 (Fig. 1I) and CaV2.3 (Fig. 1J) was unaffected. Whether the inhibition was achieved through IPPQ binding to the predicted pocket on CaVβ or through allosteric modulation remains to be determined.

IPPQ inhibited depolarization-evoked Ca²⁺ influx in a concentration-dependent manner in DRG sensory neurons with an IC₅₀ of ~9.1 μM (Fig. 1K). Since compounds are known to aggregate at screening-relevant concentrations in every compound library, we queried the Aggregator Advisor database¹⁹ with IPPQ, which revealed no similarity to known aggregators. A similar query of IPPQ in the Zinc15 database revealed no hits to molecules containing PAINS chemotypes. As the Aggregator Advisor calculated a partition coefficient for IPPQ in n-octanol/water (cLogP) value of 3.5, which is in the range reported for many other aggregators,³⁵ dynamic light scattering was performed. Dynamic light scattering curve of IPPQ in the presence of a nonionic detergent (0.1% Tween-20) did not reveal significant colloidal aggregation with only less than 9% of the compound forming particles around a 50-nm radius (supplementary Fig. 1A, available at http://links.lww.com/PAIN/A760). To further control for potential effects of aggregation, we subjected IPPQ to a centrifugation step (15 minutes at 21,000g) and then used the supernatant from the spin-down of IPPQ to perform calcium imaging studies in sensory neurons. There were no differences in the depolarization-triggered calcium responses between cells treated with IPPQ irrespective of whether it was centrifuged or not (supplementary Fig. 1B, available at http://links.lww.com/PAIN/A760). These data, along with the use of nonionic detergent in the MST experiments (Figs. 1F–J) and Tween-80 in the vehicle for all in vivo experiments, triangulate to eliminate colloidal aggregation as a mechanism of IPPQ’s inhibitory effect.

Next, we assessed whether Ca²⁺ influx was inhibited by IPPQ in all classes of DRG neurons (Figs. 2A–D), identified by constellation pharmacology—a method that permits functional “fingerprinting” of neurons to specific receptor agonists.⁷⁶ Sensory neurons were treated with 20 μM of IPPQ and then stimulated with specific receptor agonists allowing us to identify different functional populations of sensory neurons (Figs. 2A–B). IPPQ treatment increased the functional capacity of sensory neurons because they were able to respond to more receptor agonists than vehicle-treated neurons (Fig. 2A). This was correlated with an increased number of neurons responding to ATP and histamine (Fig. 2B). IPPQ decreased the peak response elicited by each receptor agonist (Fig. 2C). Depolarization-evoked Ca²⁺ influx was inhibited by IPPQ in all DRG sensory neuron functional classes (Fig. 2D). These results show that targeting the CaVβ–CaVα interface using IPPQ decreases voltage-gated Ca²⁺ channel activity across a functionally heterogenous sensory neuron subpopulation.

Figure 2.

Constellation pharmacology to identify potential off-target effects of IPPQ. Ca²⁺ imaging from DRGs treated overnight (15–18 hours) with 0.1% DMSO (n = 1152) or 20-μM compound IPPQ (n = 1359). These data include DRGs analyzed over 2 independent experiments with up to 9 trials per experiment. During each trial, >100 DRGs were sequentially stimulated for ~15 seconds (indicated by colored bars) with menthol (400 nM), histamine (50 μM), ATP (10 μM), allyl isothiocyanate (AITC) (200 μM), acetylcholine (1 mM), capsaicin (100 nM), and KCl (90 mM). Only DRGs responsive to membrane depolarization-evoked Ca²⁺ influx by KCl were included. Furthermore, only responses greater than 10% of baseline were included (indicated by filled black circles). (A) Analysis of DRG responses revealed the percentage of cells responding to the indicated number of molecular agonists, independent of which agonists elicited a response, in both DMSO- and IPPQ-treated conditions (n = 184–588, *P < 0.001, z-test.) (B) Analysis of DRG responses revealed the percentage of cells responding to the indicated molecular agonist, independent of any other agonist that also elicited a response, in both DMSO- and IPPQ-treated DRGs (n = 88–693, *P < 0.01, z-test). (C) Bar graph of average peak response to each of the indicated trigger. (D) Bar graph of the average peak response to KCl in DRG neurons responsive to the indicated trigger (n = 1152–1359, *P < 0.05, Student t test vs 0.1% DMSO). DRG, dorsal root ganglion.

3.2. Preferential activity of IPPQ for CaV2.2

Because the CaVα–CaVβ pocket is conserved across the various α and β subunits, we next set out to determine whether IPPQ was selective for any of the isoforms. We began with a heterologous system (human embryonic kidney 293 [HEK293] cells) wherein precise control of subunits could be achieved. Here, we expressed the α subunit of CaV2.2 in combination with CaVβ1–4 subunits in heterologous cells and measured whole cell currents. IPPQ inhibited CaV2.2 currents when coexpressed with CaVβ subunits 1 to 3 (Figs. 3A–D), but not with CaVβ4 (Fig. 3E). The presence of α2δ−1 did not affect the inhibition conferred by IPPQ (Figs. 3C and D). Together, these results demonstrate selectivity of IPPQ for N-type channels with CaVβ subunits 1 to 3, but not CaVβ4.

Figure 3.

IPPQ disrupts α1B–β1b, α1B–β2a, and α1B–β3–α2δ−1 interactions to inhibit Ca²⁺ currents. HEK293 cells expressing either (A) α1B + β1b, (B) α1B + β2a, (C) α1B + β2a + α2δ−1, (D) α1B + β3 + α2δ−1, or (E) α1B + β4 + α2δ−1 were subjected to an activation voltage step protocol to determine their I-V relationship after overnight (15–18 hours) treatment with 0.1% DMSO (black) or 20-μM IPPQ (red). Ca²⁺ currents in α1B + β1b, α1B + β2a, α1B + β2a + α2δ−1, and α1B + β3 + α2δ−1 expressing cells were significantly reduced in IPPQ-treated DRGs compared with 0.1% DMSO-treated counterparts (n = 5–18, *P < 0.05, Kruskal–Wallis test with Dunnett post hoc comparisons).

Next, using whole-cell patch-clamp electrophysiology, we measured Ca²⁺ currents in DRG neurons treated overnight (15–18 hours) with 20 μM of IPPQ (~twice the IC₅₀ of ~9.1 μM) or 0.1% DMSO as a control (Fig. 4A). Consistent with our Ca²⁺ screening, IPPQ inhibited (by ~50%) total Ca²⁺ currents in DRG sensory neurons (Figs. 4B and C), without affecting the biophysical properties (activation and inactivation) of these channels (Figs. 4D and E). Next, using saturating concentrations of selective Ca²⁺ channel inhibitors, we isolated either CaV1 (L-type), CaV2.1 (P/Q-type), CaV2.2 (N-type), or CaV2.3 (R-type) Ca²⁺ currents in DRGs to evaluate the inhibitory potential of IPPQ on each subtype. IPPQ (20 μM) inhibited only N-type Ca²⁺ channels (>70%) but had no effect on the other subtypes (Fig. 4F).

Figure 4.

IPPQ preferentially inhibits Ca²⁺ currents mediated by CaV2.2. (A) Activation (ie, current–voltage [I-V] relationship voltage step protocol). Cells were held at resting membrane potential for 5-ms before depolarization by 200-ms voltage steps from −70 to +60 mV in 10-mV increments. Currents were normalized to each cell’s capacitance (pF). This allowed for collection of current density data to analyze activation of Ca²⁺ channels as a function of current vs voltage and peak current density. (B) Representative Ca²⁺ current traces from DRGs subjected to the activation protocol (shown in A). (C) Summary graph of peak Ca²⁺ current density (pA/pF) from DRGs incubated with 0.1% DMSO or 20-μM IPPQ overnight (15–18 hours) (n = 12; *P < 0.05, Kruskal–Wallis test with Dunnett post hoc comparisons). (D) Inactivation voltage-step protocol. Cells were held at −90 mV for 20 ms before depolarization by 1.5-second voltage steps from −100 to +10 mV in 10-mV increments, followed by a 20-ms pulse at 10 mV before returning to −90 mV for 20 ms. (E) Normalized peak current plotted against its preceding holding potential and fitted with the Boltzmann relation. No significant differences were detected in half-maximal voltage or slope properties of either activation or inactivation between 0.1% DMSO and IPPQ conditions. (F) Summary graph of peak Ca²⁺ current density (pA/pF). Presence of IPPQ did not significantly reduce Ca²⁺ currents pharmacologically isolated through L-type, P/Q-type, or R-type Ca²⁺ channels. Presence of IPPQ significantly reduced N-type Ca²⁺ currents (compared with DMSO-treated DRGs, n = 6–10; *P < 0.05, Kruskal–Wallis test with Dunnett post hoc comparisons).

3.3. Lack of activity of IPPQ for opioid receptors

Inhibition of Ca²⁺ influx in sensory neurons can occur through activation of opioid receptors.³⁶ Consequently, we tested whether opioid receptors could be engaged by IPPQ by competition radioligand binding at all 3 opioid receptors in vitro. We competed IPPQ and a positive control compound (naloxone for MOR and DOR, U50488 for KOR) vs a fixed concentration of ³H-diprenorphine in Chinese hamster ovary (CHO) cells expressing the human μ (MOR), δ (DOR), or κ (KOR) opioid receptor. IPPQ (up to 10 μM) did not bind to the MOR or KOR and only bound to the DOR with very weak affinity (Figs. 5A–C). By contrast, the positive control compounds bound to all 3 targets with expected affinity (Figs. 5A–C). Thus, inhibition of Ca²⁺ influx by IPPQ likely does not involve opioid receptors. Compound IPPQ was evaluated at UNC’s NIMH Psychoactive Drug Screening Program against a battery of 30 receptors known to adversely impact drug effectiveness. No significant binding was observed at 20 μM. Moreover, IPPQ did not affect hERG K⁺ channel activity at 20 μM, which helps to confirm drug safety (data not shown).

Figure 5.

IPPQ does not bind to opioid receptors and does not affect G protein–mediated inhibition of Ca²⁺ current in rat sensory neurons. Competition radioligand binding was performed in CHO cells expressing the human opioids receptors MOR (A), DOR (B), or KOR (C). Compound IPPQ or a positive control compound competed against ³H-diprenorphine in all 3 cell lines. Curves reported as the mean ± SEM of the mean value from each individual experiment in N = 3 independent experiments. The K_I also reported as the mean ± SEM of the individual value from each of N = 3 independent experiments. (A) Compound IPPQ did not bind to the MOR. Naloxone K_I = 33.9 ± 1.7 nM. (B) Compound IPPQ produced 76.2% inhibition at 10 μM at the DOR, with an incomplete curve. This indicates an IC₅₀ >3.3 μM and a K_I >1 μM. Naloxone K_I = 55.7 ± 6.7 nM. (C) Compound IPPQ did not appreciably bind to the KOR. U50488 K_I = 22.4 ± 4.1 nM. (D) Representative current traces following the “double pulse” protocol (top) for control, IPPQ (20 μM) applied for 30 minutes, and norepinephrine (NE, 10 μM). (E) Histograms of ratio of I₂/I_1, measured at 10 msec after the step to −10 mV of DRG neurons treated with 0.1% DMSO, IPPQ, or NE. *P, 0.05, Kruskal–Wallis test with Dunnett post hoc comparisons. DRG, dorsal root ganglion.

CaV2.2 function can be regulated by the G protein–coupled receptor (GPCR) Gi signaling after activation of the μ opioid or the GABA-B receptors.¹ To test whether IPPQ-mediated inhibition of CaV2.2 could involve Gi-coupled GPCR signaling, we used the prepulse facilitation electrophysiological protocol that relies on applying strong depolarizing prepulse to dissociate the interaction between CaV2.2 and the Gβγ subunits.⁹⁵ This triggers a relief of GPCR-mediated inhibition of the calcium channel and allows for a second recording of calcium currents. We did not find any evidence of GPCR-mediated inhibition of CaV2.2 with IPPQ (Figs. 5D and E) thus rejecting the hypothesis that IPPQ acts through activating any inhibitory GPCR signaling. Norepinephrine (NE) was chosen as a control because it has been demonstrated to inhibit Ca²⁺ currents in a voltage-dependent manner.^34,78

3.4. IPPQ inhibits spinal neurotransmission

CaV2.2 is involved in spinal nociceptive neurotransmission³⁰ through the control of presynaptic nociceptive neurotransmitter release from C fibers.^11,30,59,73 IPPQ-mediated inhibition of CaV2.2 from DRG neurons should in turn decrease spontaneous excitatory postsynaptic currents (sEPSCs) measured in the substantia gelatinosa of the dorsal horn of the spinal cord. We recorded sEPSCs from neurons in laminae I-II in this region of the lumbar dorsal horn (Fig. 6A). IPPQ treatment had no effect on the amplitude of the recorded sEPSCs (Figs. 6C and D). However, we observed a decrease of sEPSC frequency after treatment with 20 μM of IPPQ (Figs. 6E and F). We next investigated the effects of IPPQ (20 μM) on miniature EPSCs (mEPSCs) and found that both the frequency and amplitude of mEPSCs were decreased (Figs. 6G–J).

Figure 6.

IPPQ decreases spinal excitatory synaptic transmission. Photomicrograph of slice preparation showing that the substantia gelatinosa (SG) can be identified as a translucent pale band in the superficial dorsal horn enabling positioning of the recording electrode to this region. (A, i) infrared differential interference contrast image (IR-DIC) and (A, ii) image of the same cell (indicated by a purple red box in middle panel) with part of the recording electrode after whole-cell configuration. (B) Representative recording of spontaneous and miniature EPSCs traces of cells from the indicated groups: 0.1% DMSO or 20-μM IPPQ, applied through perfusion system during the experiment within 3 hours. The recordings were obtained within 30 minutes to 3 hours of drug perfusion; no difference was observed between recordings performed earlier vs later in the window after drug perfusion (data not shown). (C and D) The cumulative probability and summary of sEPSC amplitudes. (E and F) The cumulative probability of interevent interval and frequencies of sEPSCs for the indicated groups are shown. IPPQ treatment had no effect on the amplitude but decreased the frequency of the recorded sEPSCs. (G and H) The cumulative probability and summary of mEPSC amplitudes. (I and J) The cumulative probability of interevent interval and frequencies of mEPSCs. Both the frequency and amplitude of mEPSCs were decreased by 20-μM IPPQ. Data are expressed as mean ± SEM from n = 16 to 18 cells per condition. *P < 0.05 (vs DMSO); Student t test.

To determine whether the actions of IPPQ were presynaptic or postsynaptic, we also analyzed EPSCs evoked by a paired-pulse stimulation protocol, which is commonly used to assess changes in presynaptic function.^97,98 At 50-ms intervals, paired-pulse ratios were increased by IPPQ (20 μM) treatment (Figs. 7A and B). These results collectively implicate a presynaptic provenance for the mechanism of IPPQ’s action on spinal synaptic neurotransmission.

Figure 7.

IPPQ increases paired-pulse ratios. (A) Representative traces of paired-pulse ratio (PPR) from the indicated groups: 0.1% DMSO or 20-μM IPPQ. The recordings were obtained within 30 minutes to 3 hours of drug perfusion. (B) Summary of average PPRs from the indicated groups. Data are expressed as mean ± SEM from n = 13 to 17 cells per condition. *P < 0.05 (vs DMSO); Student t test.

3.5. IPPQ reduces CaV2.2 spinal presynaptic localization

IPPQ can inhibit CaV2.2 AID interaction with the beta subunits. We asked whether interfering with this interaction using IPPQ in vivo could affect CaV2.2 presynaptic localization in lumbar dorsal horn of the spinal cord. We first extracted synaptosomes from the dorsal horn of the spinal cord of rats 1 hour after injection with IPPQ (2 μg in 5 μL, intrathecally) and isolated before and after synaptic fractions (Fig. 8A). The fractionation efficiency was verified by western blot where the postsynaptic marker PSD95 was highly enriched in the PSD fraction, whereas the presynaptic marker synaptophysin was only found in the non-PSD fraction (Fig. 8A). Indeed, CaV2.2 was exclusively localized in the presynaptic fraction (Fig. 8A). We then focused on the presynaptic (non-PSD) fraction to evaluate whether IPPQ could change the presynaptic levels of CaV2.2. We found that IPPQ decreased CaV2.2 presynaptic levels (Figs. 8B and C). Notably, IPPQ did not affect the localization of the NaV1.7 voltage-gated sodium channel NaV1.7 (Figs. 8B and C). These results demonstrate that interfering with the CaV2.2 AID/beta subunit interaction specifically reduces the presynaptic localization of the CaV2.2 in vivo.

Figure 8.

IPPQ decreases CaV2.2 presynaptic localization in vivo and reduces evoked CGRP release from spinal cord. (A) Immunoblots showing the integrity of the synaptic fractionation from lumbar dorsal horn of the spinal cord. The non-postsynaptic density (non-PSD) fraction was enriched in the presynaptic marker synaptophysin (Syp), and the PSD fraction was enriched in the postsynaptic marker PSD95. The membrane-associated protein flotillin was used as a loading control. (B) Immunoblots showing the presynaptic CaV2.2 levels in the lumbar dorsal horn of the spinal cord of animals having received IPPQ (2 μg in 5 μL, intrathecally) compared with vehicle (0.1% DMSO). The synaptosomes were prepared from the dorsal horn of the spinal cord of rats 1 hour after injection with IPPQ. Synaptophysin shows the integrity of each fraction. Flotillin is used as a loading control. (C) Bar graph showing decreased CaV2.2 levels at the presynaptic sites of lumbar dorsal horn of the spinal cord in IPPQ-treated animals. Mean ± SEM, *P < 0.05, Mann–Whitney compared with the DMSO vehicle treatment.

3.6. Inhibition of spinal calcitonin gene–related peptide release by IPPQ

CaV2.2 activity controls C-fiber sEPSCs and the release of the nociceptive neurotransmitter CGRP. Because IPPQ inhibited CaV2.2 and sEPSC frequency, we hypothesized that evoked spinal CGRP release would be inhibited by IPPQ. To test this, we measured CGRP release evoked by depolarization (90-mM KCl, a concentration that activates low- and high-voltage-activated Ca²⁺ channels⁸⁴) in an ex vivo preparation of the lumbar region of rat spinal cord. Enzyme-linked immunosorbent assay (ELISA) was used to measure CGRP content; samples were collected every 10 minutes. We chose 10 minutes because this incubation period was sufficient to cause a reduction in depolarization-triggered calcium influx in DRG neurons in vitro (Figs. 9A–C). Basal CGRP levels were ~8.2 pg·mL⁻¹·mg⁻¹ of tissue (Fig. 9D, fractions #1 & 2). Before stimulation, control (0.1% DMSO) or a 20-μM IPPQ was added for 10 minutes (Fig. 9D, fraction #3). Although IPPQ application did not elicit any CGRP release from the spinal cords (Fig. 9D, fraction #3), depolarization evoked CGRP release from spinal cord, which was inhibited by ~63% by compound IPPQ (Fig. 9D, fraction #4). These results show that inhibiting CaV2.2 by targeting the CaVβ–CaVα interface with IPPQ results in decreased depolarization-evoked CGRP release.

Figure 9.

IPPQ acutely reduces calcium influx and evoked calcitonin gene–related peptide (CGRP) release from spinal cord. (A) Representative traces of membrane depolarization-evoked Ca²⁺ influx by 90-mM KCl before and after 10 minutes application of either 0.1% DMSO (A) or 20-μM IPPQ (B). (C) Quantification of the peak ratio for the indicated condition. Ten-minute treatment with IPPQ significantly reduced the peak ratio (vs 0.1% DMSO, n = 460–493 cells, *P < 0.05, Student t-test). (D) CGRP released from spinal cord fractions collected every 10 minutes corresponded with the following conditions: 1-baseline, 2-baseline, 3-treatment, 4-treatment + 90-mM KCl, 5 wash. Previous treatment for 10 minutes with compound IPPQ significantly reduced high potassium-evoked CGRP release in spinal cord (vs DMSO, n = 4–5, *P < 0.0001, one-way ANOVA with Dunnett post hoc comparisons). ANOVA, analysis of variance.

3.7. Broad antinociceptive efficacy of IPPQ

The peptide ω-conotoxin GVIA remains a defining ligand for CaV2.2 and is marketed as Prialt (Primary Alternative to Morphine) for relief of chronic and cancer-related pain,⁶⁷ establishing the therapeutic value of targeting CaV2.2. IPPQ increased withdrawal latency to a heat stimulus in naive rats (Fig. 10A), demonstrating antinociceptive potential. Furthermore, CaV2.2 is known to be involved in postsurgical pain,²⁶ neuropathic pain,⁹¹ chemotherapy-induced neuropathy,³⁷ HIV-induced sensory neuropathy,⁵⁷ and neurofibromatosis type 1 (NF1)-related pain.⁶⁰ Thus, we performed an analgesic appraisal of IPPQ on heat and tactile sensitivity of rats in vivo. We chose to administer IPPQ intrathecally (2 μg in 5 μL) to directly reach the site of action of CaV2.2 in the dorsal horn, the presynaptic termini of pain fibers that synapse with spinal cord neurons and require CaV2.2 for the release of neurotransmitter at these synapses. IPPQ reversed mechanical allodynia and thermal hyperalgesia in models of NF1-related pain (Fig. 10B), postsurgical pain (Figs. 10C–D), and neuropathic pain (Figs. 10E–F). IPPQ was also efficient in reversing mechanical allodynia in paclitaxel-induced peripheral neuropathy (Fig. 10G) and HIV-induced sensory neuropathy (Fig. 10H). The reversal lasted about 2 to 3 hours across the behavioral paradigms tested, consistent with a plasma half-life of 2.2 hours for IPPQ (data not shown). These results demonstrate the utility of IPPQ to modulate painful sensations in rodents.

Figure 10.

IPPQ is antinociceptive in naive animals and reverses mechanical allodynia and thermal hyperalgesia in acute and neuropathic pain models. (A) Paw withdrawal latencies (PWLs) of naive rodents to a heat stimulus were significantly increased 1 hour after an intrathecal (i.t.) injection of IPPQ (2 μg) compared to animals injected with vehicle (10% DMSO, 10% Tween-80, 80% saline) (compared with i.t. injection of vehicle (n = 6, *P < 0.05, 2-way ANOVA). (B) PWLs of rodents subjected to targeted Cas9-mediated editing of Nf1 (using i.t. injection) were significantly decreased, but i.t. injections of IPPQ reversed this behavior (vs i.t. injection of vehicle, n = 6, *P < 0.05, 2-way ANOVA). (C) PWLs of rodents that received a paw incision (Sx) on the left hind paw were significantly decreased. Intrathecal injection of IPPQ reversed this nociceptive behavior (compared with i.t. injection of vehicle, n = 6, *P < 0.05, 2-way ANOVA). (D) Paw withdrawal thresholds (PWTs) of rodents that received a paw incision (Sx) on the left hind paw were significantly decreased. Intrathecal injection of IPPQ reversed this nociceptive behavior (compared with i.t. injection of vehicle, n = 6, *P < 0.05, 2-way ANOVA). (E) PWLs of rodents that received a spinal nerve ligation (SNL) on the left hind paw were significantly decreased. Intrathecal injection of IPPQ reversed this nociceptive behavior (compared with i.t. injection of vehicle, n = 6, *P < 0.05, 2-way ANOVA). (F) PWTs of SNL rodents were also significantly decreased. Intrathecal injection of IPPQ also reversed this behavior (vs i.t. injection of vehicle, n = 6, *P < 0.05, 2-way ANOVA). (G) PWTs of rodents that received four 2-mg/kg intraperitoneal (i.p.) injections of paclitaxel (Px) over 12 days were significantly decreased. Intrathecal injection of IPPQ reversed this behavior (vs i.t. injection of vehicle, n = 6, *P < 0.05, 2-way ANOVA). (H) PWTs of rodents were significantly decreased 15 days after receiving 3 i.t. injections of gp-120. Intrathecal injection of IPPQ reversed this behavior (vs i.t. injection of vehicle, n = 6, *P < 0.05, 2-way ANOVA). The experiments were conducted in a blinded fashion. ANOVA, analysis of variance.

3.8. IPPQ does not cause motor impairment

The clinically used CaV2.2 drug Prialt (ziconotide) produces akinesia when injected into the brain.⁴⁸ To test whether IPPQ would elicit a similarly undesirable effect, we used the rotarod test to assess rats’ motor function after an i.c.v. injection of IPPQ (2 μg in 5 μL). Rats were followed for 90 minutes after the injection, and no motor impairment was observed (supplementary Fig. 2, available at http://links.lww.com/PAIN/A760) after IPPQ brain injection. We did not observe other behavioral alterations such as seizures in any of the animals tested.

3.9. IPPQ works through a spinal mechanism and does not affect anxiety

IPPQ was tested in acute thermal pain models (hot-plate or tail-flick) because CaV2.2 blockers have been reported to be more effective against chronic than acute pain.⁴⁷ The tail-flick test is considered predominantly spinal because it can be observed even in animals with spinal cord transection that have severed descending pathways²⁰ or animals under anesthesia, which is not the case for the paw lifting/jumping, twitching, and licking behavior observed in the hot-plate test. When a dose (15 mg/kg) of IPPQ was intraperitoneally administered in acute thermal pain models, the latency times were not affected in the hot-plate test (predominantly supraspinal) but increased in the tail-flick test (predominantly spinal) (Figs. 11A and B). Gabapentin administration (100 mg/kg) was used a control for the hot-plate and tail-flick tests: it increased paw withdrawal latencies in the hot-plate test but was without effect at this dose on tail-flick latencies (Figs. 11C and D), consistent with what has been reported previously.⁵ The lack of effect IPPQ administered intraperitoneally in the hot-plate compared with tail-flick points to a specific action of IPPQ on processes implicated in the tail-flick test, similarly to what has been observed in the α1B^−/− knockout mice⁶⁹ or with specific blockers of CaV2.2.⁵

Figure 11.

IPPQ is analgesic in mice but does not affect anxiety-related behaviors. (A) Paw withdrawal latencies of naive mice in a hot-plate test (52°C) 75 minutes after injection of IPPQ (15 mg/kg) administered intraperitoneally (i.p.) were unaffected (compared with i.p. injection of vehicle: 10% DMSO, 10% Tween-80, 80% saline), n = 12, P = 0.7583, t test with the Welch correction. (B) Tail-flick latencies of naive mice to a hot (52°C) water bath 75 minutes after injection of IPPQ administered i.p. were significantly increased (vs vehicle i.p. injection), n = 12, *P < 0.05, t test with the Welch correction). (C) Paw withdrawal latencies of naive mice to the hot-plate test (52°C) 60 minutes after i.p. injection of gabapentin (100 mg/kg) administered were increased (compared with i.p. injection of vehicle), n = 9, *P < 0.05, Mann–Whitney test. (D) Tail-flick latencies of naive mice to a hot (52°C) water bath 60 minutes after i.p. injection of gabapentin (100 mg/kg) administered were unaffected (vs vehicle i.p. injection), n = 9, P = 0.445, t test with the Welch correction). (E) Naive mice received an i.p. injection of IPPQ (15 mg/kg), or its vehicle and anxiety-related behaviors were assessed during 10 minutes in an elevated plus maze test. Heatmaps of the positions occupied in the elevated plus maze apparatus by treatment groups are represented in (E). (F) Bar graphs of the mean time spent in the open arms of the elevated plus maze. No differences were found between animals treated with IPPQ or with its vehicle (P = 0.6014, t test with the Welch correction, n = 12 mice/group). All experiments were conducted in a blinded fashion.

As it has been previously demonstrated that inhibition of CaV2.2-mediated signaling induces alterations in the neuronal network involved in anxiety-related behaviors,^42,93 we next tested the effects of IPPQ on anxiety using the elevated plus maze test. Time spent in the open arms of the elevated plus maze, time spent in the closed arms, and number of entries in the open arms did not differ between animal treated with IPPQ or its vehicle (Figs. 11E and F), suggesting that IPPQ does not affect anxiety-related behaviors.

4. Discussion

The present work used a rational structure-based design strategy to identify a new class of nonopioid, analgesic compounds targeting the CaVα–CaVβ interface. The quinazoline analog IPPQ (1) specifically bound to CaVβ, (2) inhibited the biochemical interaction between CaVβ2 and the AID of CaV2.2 and to a lesser extent between CaVβ2 and the AID of CaV2.1, but not other subtypes, (3) and functionally inhibited CaV2.2 function (acutely and chronically) across subtypes of DRG sensory neurons. IPPQ inhibited the presynaptic localization of CaV2.2 in vivo as well as inhibiting spinal neurotransmission, which resulted in decreased neurotransmitter release from the spinal cord. Finally, IPPQ was antinociceptive in naive rats and reversed mechanical allodynia and thermal hyperalgesia in rodent models of acute (postsurgical), neuropathic (spared nerve ligation, paclitaxel, or HIV-induced sensory neuropathy), and genetic (NF1-related) pain. No adverse effects were observed for IPPQ in rats or mice. These results are an instructive example that pharmacological disruption of the CaVα–CaVβ interface can result in a selective, safe, and broadly antinociceptive compound, setting the stage for structure–activity relationship studies to improve on this lead molecule for the development of novel pain therapeutics.

The CaVα–CaVβ interaction shapes activity and trafficking of all subtypes of high-voltage-gated calcium channels (ie, L-, N-, P/Q-, and R-type). Before the crystallization of CaVβ subunits, Wyeth-Ayerst Research used a high-throughput yeast 2-hybrid screen to identity compounds that disrupt α1b and β3 and found WAY141520, which inhibited calcium currents with an IC₅₀ of 95 μM, with some inhibitory activity against CaV2.2 but was not tested against other channels nor was its mechanism of action studied.⁹² Structural studies of CaVβs in early 2000s^16,63,79 with a fragment of the α subunit revealed a high homology between the various α–β subunits, predicting that targeting of this interface would result in nonselective compounds. The details of this interface have been extensively studied in deep thermodynamic detail for all of the CaV1 and CaV2 AIDs and that has defined the hotspots on both sides of this interaction.⁸⁰ Of the 3 “hot spots” identified by the Minor laboratory, V241 makes no contacts with the AID, N390 does not contribute to the energetics of the interaction, and only I343 is important, interacting with one of the core residues that comprise the AID hotspot (CaV1.2 Y437).⁸⁰ Despite these theoretical predictions, IPPQ exhibited selectivity for CaV2.2. IPPQ interacts with CaVα–CaVβ domain mainly at the lipophilic pocket defined by Val 339–Ile 343.⁷⁹ The AID sequence aligns (side chains of critical residues Y437, W440, and I441; Fig. 1C) with the quinazoline ring and phenylpropyl group of IPPQ. We hypothesize that these interactions result in high selectivity for IPPQ. That IPPQ reduced the binding affinity between CaVβ and AID peptides of CaV2.2 (by ~17.3-fold) and CaV2.1 (by ~5-fold) but had no effect on binding to other channel subtypes demonstrates selectivity for the interaction. Although this selectivity was unexpected, the electrophysiology recordings on sensory neurons confirmed a selectivity for CaV2.2 while the electrophysiology recordings in heterologous cells, performed in an independent laboratory, confirmed inhibition through CaV2.2 and CaVβ1b, CaVβ2a, and CaVβ3, irrespective of the presence of the α2δ−1 subunit. Inhibiting CaV2.2 with IPPQ did not show the typical regulation of the channel’s gating property by CaVβs.¹² But this is consistent with a previous report that demonstrated no effect on voltage-dependent activation of high-voltage-activated calcium channels upon deletion of CaVβ3 subunit in DRGs.⁴³ It has also been reported that the orientation of CaVβ relative to the α subunit of CaV2.2 subunit is critical for its regulation of channel activity with insertions or deletions of the AID linker, which are expected to maintain the α-helical structure of the linker but induce a rotation of CaVβ with respect to CaVα1, diminishing CaVβ regulation of activation and inactivation.⁸² Thus, IPPQ could possibly induce conformational changes that result in disrupt the rigidity of the α-helix, which may in turn explain the lack of effects on biophysical properties.

Presynaptic voltage-dependent calcium channels, controlling calcium influx, serve as the trigger for release of glutamate from the terminals of the primary afferent fibers onto spinal dorsal horn neurons.⁴⁵ The interruption of CaVα–CaVβ interaction by IPPQ might result in the inhibition of excitatory synaptic inputs to SG neurons. The decrease of sEPSC frequency suggested that IPPQ inhibited the glutamatergic excitatory inputs by a presynaptic mechanism. It is generally accepted that the changes in frequency of mEPSCs reflects an alteration of presynaptic neurotransmitter release probability, and that the changes in mEPSC amplitude indicate less neurotransmitter content per quanta (per presynaptic vesicle), fewer postsynaptic receptors, or a change in the type of receptors at postsynaptic sites (ie, a change in the single channel conductance of receptors).³² IPPQ also inhibited both frequency and amplitude of mEPSCs. This is consistent with a previous report that specific antagonists for one or more high-threshold-activated Ca²⁺ channels suppress release of transmitter by presynaptic impulses at central synapses, thereby also reducing the postsynaptic response.⁶ Moreover, the PPR protocol, used to investigate the short-term plasticity characteristics of synapse, is a widely accepted approach for assessing the synaptic sites of action.⁷⁷ The increase of PPR induced by paired-pulse stimulation further implicates a presynaptic origin for the mechanism of IPPQ’s action on spinal synaptic neurotransmission. In support of this, we observed decreased levels of spinal presynaptic CaV2.2 after IPPQ treatment, thus reinforcing the assertion that IPPQ acts presynaptically.

That IPPQ was effective and safe in vivo likely stems from the restricted expression patterns of the CaVβ subunits. For example, CaVβ1 is expressed in skeletal muscle,³¹ CaVβ2 is expressed in the heart and at very low levels in the central nervous system (CNS),⁴⁴ CaVβ3 expression is found in smooth muscle and in the CNS (including the spinal cord),⁴⁴ and CaVβ4 is expressed predominantly in the cerebellum.⁴⁴ CaV2.2 is expressed in the brain and in the sensory neurons.¹⁴ Thus, only in the brain and sensory neurons does CaV2.2 expression overlap with that of CaVβ3 and CaVβ4. Our in vitro results demonstrated that only coexpression of the CaVβ3/4 subunits with CaV2.2 resulted in high Ca²⁺ currents while CaVβ1/2 coexpression with CaV2.2 (a configuration that is not physiologically relevant because these subunits are primarily expressed in skeletal or heart muscles) resulted in low Ca²⁺ currents. IPPQ inhibited Ca²⁺ currents when coexpressed with CaVβ1/2/3, but not CaVβ4. Thus in vivo, IPPQ would only target CaVβ3–CaV2.2 coupling because of the overlapping tissue distribution of these subunits. The notion that CaV2.2 functions with CaVβ3 to mediate nociception in vivo is supported by mouse models lacking either CaV2.2⁶⁹ or CaVβ3⁶¹ with these transgenic mice demonstrating resistance to pain sensation. Notably, CaVβ3 expression is increased in neuropathic pain,⁴³ in parallel with increased voltage-gated calcium channel function. Thus, the specific uncoupling of CaVβ3–CaV2.2, but not CaVβ4–CaV2.2, may underpin the calcium channel subtype selectivity of IPPQ and its in vivo efficacy for the reversal of chronic pain behaviors.

A common denominator in the neuropathic models appraised here is the plasticity of CaV2.2: targeted truncation of neurofibromin, achieved by acute CRISPR/Cas9 editing of the Nf1 gene in adult rats, and resulted in increases of CaV2.2 currents⁶⁰; spinal nerve ligation increased CaV2.2 expression in the synaptic plasma membranes of the dorsal horn⁴¹; and the Cacna1b gene encoding CaV2.2 is highly upregulated in a rodent model of HIVSNL treated with gp120 compared with sham-treated controls.⁴⁶ Previous reports have shown attenuation of allodynia by a CaV2.2-blocking compound in the paw incision model of surgical pain²⁶ and the paclitaxel-induced model of chemotherapy-induced peripheral neuropathy.⁸ The broad antinociceptive potential of IPPQ across acute and neuropathic models, in 2 species, can be attributed to its biochemical and functional inhibition of the CaVα–CaVβ interaction, a mechanism distinct from direct block of CaV2.2 by small molecules covering various pharmacophores.⁹⁰

Therapeutic management strategies for chronic pain usually involve the use of opioid molecules. These compounds primarily act on the MOR expressed on DRG sensory neurons to provide analgesia. CaV2.2 can be inhibited downstream of G(i/o)-type G protein–coupled receptor signaling as shown upon morphine to the MOR or the α-conotoxin Vc1.1 binding to be GABA-B receptor.^33,53,54 In neuropathic pain, alternative splicing of CaV2.2 mRNA leads to 2 different CaV2.2 isoforms containing either the exon 37a (e37a) or e37b in sensory neurons.⁷ After a neuropathic injury, expression of e37a is lost, resulting in an enrichment of e37b containing CaV2.2 in sensory neurons.² This switch in alternative splicing is an important event underlying resistance and tolerance to opioids since the remaining e37b containing CaV2.2 is resistant to the inhibition mediated by opioids.³ No alternative splicing has been reported in the binding region between CaVα and CaVβ. Thus, targeting this interface using IPPQ is unlikely to be susceptible to similar tolerance mechanisms in neuropathic pain.

Although the CaV2.2-inhibiting drug ziconotide (Prialt) is effective in chronic and cancer-related pain,⁶⁷ it requires intrathecal dosing to circumvent systemic side effects that include confusion, depression, hallucinations, decreased alertness, somnolence, orthostatic hypotension, and nausea.⁷¹ These are due to the direct blocking of CaV2.2 in the brain. Notably, when Prialt was injected i.c.v. in mice, it induced akinesia that would also decrease rotarod performance.⁴⁸ We observed no effect on rotarod performance in IPPQ-injected rats. If the rats had experienced akinesia, decreased alertness, or orthostatic hypotension, their rotarod performance would have been impaired accordingly. Thus, IPPQ treatment did not induce any of the side effect related to direct CaV2.2 inhibition throughout the brain and exhibits a better safety profile compared with the FDA-approved drug Prialt. Thus, a small molecule, such as IPPQ, devoid of these encumbrances and lacking a nonopioid mechanism would be a welcome addition to the armamentarium of the pain physician for managing chronic pain.