ω-Conotoxin GVIA Mimetics that Bind and Inhibit Neuronal Cav2.2 Ion Channels

The neuronal voltage-gated N-type calcium channel (Cav2.2) is a validated target for the treatment of neuropathic pain. A small library of anthranilamide-derived ω-Conotoxin GVIA mimetics bearing the diphenylmethylpiperazine moiety were prepared and tested using three experimental measures of calcium channel blockade. These consisted of a 125I-ω-conotoxin GVIA displacement assay, a fluorescence-based calcium response assay with SH-SY5Y neuroblastoma cells, and a whole-cell patch clamp electrophysiology assay with HEK293 cells stably expressing human Cav2.2 channels. A subset of compounds were active in all three assays. This is the first time that compounds designed to be mimics of ω-conotoxin GVIA and found to be active in the 125I-ω-conotoxin GVIA displacement assay have also been shown to block functional ion channels in a dose-dependent manner.

An investigation of the mimic based on the anthranilamide scaffold involved variation of the length of the side chains, the incorporation of two guanidino moieties, and two variations of the diphenyl ether substituent (see Figure 1) [32]. Key findings from this study were that; (a) typically diguanidino compounds [Z = N=(NH 2 ) 2 ] had binding activity in the 6-16 μM range in a radio-labelled ω-conotoxin GVIA displacement assay, whereas diamino compounds [Z = NH 2 ] bound more weakly and; (b) strongest affinity was found with a fluorinated mimic [X = F].
There are a number of compounds that effectively block voltage-gated calcium channels that bear the diphenylmethylpiperazine moiety [34]. Relevant to the current study are Neuromed's NP-118809 (NMED-160) [7,8] and Abbott's A-1048400 [11,12] shown in Figure 2. It was thus decided to prepare analogues of the anthranilamide-based conotoxin mimics where a diphenylmethylpiperazine moiety was incorporated in place of the phenoxyaniline substituent (Figure 3), and test their ability to block neuronal calcium channels in non-functional and, importantly, functional assays.

Radioligand Displacement Assay
The synthesized compounds, diamino (7a-d) and diguanidinium (8a-d), were initially evaluated for their ability to bind to rat brain Ca v 2.2 channels using a previously described radioligand displacement assay [29][30][31][32]39], employing 125 I-labelled ω-conotoxin GVIA. Although not a functional assay, the high selectivity of ω-conotoxin GVIA for Ca v 2.2 channels means that compounds able to displace 125 I-labelled ω-conotoxin GVIA from rat brain homogenate are likely to be functional inhibitors of Ca v 2.2 channels. The results obtained with 7a-d and 8a-d are summarized in Table 1.
The affinities of the compounds that lack the diphenylmethyl functionality (7c, 7d, 8c and 8d) were too weak to be measured by this method, whereas those that do possess that moiety (7a, 7b, 8a and 8b) showed moderate to good binding (see Figure 4). As observed previously with this class of compound [32], the diguandino compounds (8a and 8b) showed strongest binding, comparable to some of the best reported anthranilamides [32], determined by this method. Compounds 7a, 7b, 8a and 8b were therefore selected for follow-up functional assays. Table 1. Ca v 2.2 binding affinities of diamino (7a-d) and diguanidinium (8a-d) anthranilamides, as determined by displacement of 125 I-GVIA from rat brain homogenate (95% confidence intervals are shown in parentheses).

Inhibition of Neuroblastoma Intracellular Calcium Responses, Determined by Fluorescence Measurement of Calcium Flux
The abilities of the diamino (7a and 7b) and diguanidinium (8a and 8b) compounds to inhibit intracellular calcium responses in SH-SY5Y human neuroblastoma cells, in the presence of the L-type calcium channel blocker nifedipine were assessed using a high-throughput Ca 2+ fluorescence assay. It was found that Ca 2+ ion channel responses elicited by KCl-mediated depolarization were partially inhibited by the test compounds at concentrations which did not produce fluorescence addition artefacts (100 µM; Figure 5A), with compound 7b being less efficacious (4.9 ± 1.8% inhibition) than compounds 7a (23.1 ± 1.5% inhibition), 8a (20.3 ± 4.0% inhibition) and 8b (24.4 ± 1.5% inhibition). While these inhibitory effects are not strong, dose-dependent inhibition was observed ( Figure 5B) which allowed the estimation of IC50's for the functional inhibition of intracellular calcium responses in SH-SY5Y human neuroblastoma cells. These results are shown in Table 2. The estimated IC50s from these functional, whole-cell experiments are up to an order of magnitude weaker than those obtained from the radioligand displacement assay. A similar shift in potency is observed with the ω-conotoxins, whose effectiveness is reduced in the presence of physiological Ca 2+ levels [40] and auxiliary subunits [41], which are likely to dissociate from the Ca v α-subunit in membrane preparations used in radioligand displacement assays. It is also plausible that the decreased potency of the test compounds in the functional assays, in addition to effects of co-expressed auxiliary subunits, may have been influenced by relatively short incubation times in the presence of extracellular divalent cations, which has been reported to affect the on-rate of ω-conotoxin block.  Despite giving comparable radioligand displacement assay results to the best previously reported anthranilamides (see Figure 1) [32], the new diphenylmethylpiperazine-substituted compounds (7a, 8a and 8b) showed superior inhibition of intracellular calcium responses in SH-SY5Y cells (data not shown) and hence can be considered an improvement on previous designs.

Patch-Clamp Electrophysiology Experiments with HEK293 Cells Expressing Ca v 2.2 Calcium Channels
SH-SY5Y human neuroblastoma cells contain a number of calcium channel subtypes [42][43][44][45]. To specifically measure the effect of the diamino (7a and 7b) and diguanidinium (8a and 8b) anthranilamides on N-type (Ca v 2.2) calcium channel currents, electrophysiological patch-clamp studies were carried out on HEK293 cells stably expressing human Ca v 2.2 (a1 + β3 + α2δ1 subunits). Compounds 7a, 7b and 8a were each tested at two concentrations (30 and 100 µM), whereas 8b was only tested at 30 µM. Representative traces obtained prior to compound application (Control) and following application of compound 7a at 30 and 100 µM are shown in Figure 6A. The time course of washing in and out of compound 7a is shown in Figure 6B, with peak current amplitude being measured at 10 s intervals. All four compounds showed a modest inhibition of the calcium current amplitude; 10%-17% at 30 µM and 20%-27% at 100 µM ( Figure 6C), while exhibiting no shift in the G-V relationship or changes to channel inactivation. To determine an approximate IC 50 value, the data were fit assuming that these compounds completely inhibited the calcium currents, with a Hill coefficient of −1. This gave the predicted half-maximal inhibition (IC 50 's) shown in Table 2 and are consistent with those obtained using the FLIPR-Ca 2+ assay in SH-SY5Y cells. The estimated IC 50 values for 8b obtained from the two techniques are in close agreement and, of all the compounds examined in this study, this compound appears to most effectively block Ca v 2.2 calcium channels. Figure 6. Effects of diamino (7a and 7b) and diguanidinium (8a and 8b) anthranilamides on Ca v 2.2 currents in HEK293 cells. (A) Superimposed whole-cell calcium currents obtained from HEK293 cells stably expressing N-type calcium channels (hCa v 2.2 + β3 + α2δ1), with depolarizing voltage steps from a holding potential of −80 mV to a test potential of +20 mV (150 ms). Representative currents are shown prior to compound 7a application (Control) and in the presence of compound 7a at two concentrations (30 and 100 µM). (B) Time course of inhibition of peak inward current in response to 30 and 100 µM compound 7a. (C) Bar graph of the relative inhibition of peak current amplitude (1 − I compound /I control ) for each compound at concentrations of 30 and 100 µM (number of experiments in parentheses).

General Experimental Procedures
Starting materials and reagents were purchased from Sigma-Aldrich and used without purification. Tetrahydrofuran (THF) was distilled under a nitrogen atmosphere from sodium benzophenone ketal. Dry DCM was obtained after drying over 3 Å sieves. Methanol and DMF were obtained by passage through two alumina columns on the Solvent Dispensing System built by J. C. Meyer and based on the original design by Grubbs and co-workers [46]. Solutions were dried over anhydrous magnesium sulfate (MgSO 4 ) or sodium sulfate (Na 2 SO 4 ). Normal phase flash chromatography was performed on Merck silica gel No. 9385 and reverse phase chromatography was performed using a C18 Chromatorex ® DM1020T column (30 × 40 mm). Spectra were recorded on a Bruker Av200 or Av400 spectrometer. Unless otherwise stated, proton ( 1 H) NMR spectra were acquired at 200 MHz and carbon ( 13 C) NMR spectra at 50 MHz. NMR spectra were referenced to residual solvent peak [chloroform (δ H 7.26, δ C 77.0), methanol (δ H 4.87, 3.30, δ C 49.86)]. The units for all coupling constants (J) are in hertz (Hz). Low resolution mass spectra were recorded on a Micromass Platform spectrometer or VG Platform spectrometer. Accurate mass determinations were carried out at high resolution on an Agilent G1969A LC-TOF system, with reference and mass correction at 4000 V capillary voltage for ESI.

Ca v 2.2 Radioligand Displacement Assay
Radioligand binding assays were run in triplicate in 96-well plates at room temperature as previously described [39]. Each assay contained the test compound, radiolabelled peptide (7 pM 125 I-GVIA) and 8 μg of crude rat brain membrane, added last. All dilutions were made in assay buffer (20 mM HEPES, 75 mM NaCl, 0.2 mM EDTA, 0.2 mM EGTA, 2 μM Leupeptin, 2 μL apoprotinin (to 30 mL assay buffer) and 0.1% BSA, pH 7.4). The final volume in each well was 150 μL. After shaking for 1 h, the membrane was filtered (Wallac, Finland glass fibre filters pre-soaked in 0.6% polyethyleneimine) and washed with 20 mM HEPES, 125 mM NaCl, pH 7.4 on a Tomtec harvester. After addition of scintillant, radioactivity bound to the filter was counted using a 1450 MicroBeta (Wallac, Finland). The data was analyzed using GraphPad Prism 2.0 (GraphPad Software, Inc, San Diego, USA).

Fluorescence Measurement of Calcium Responses
SH-SY5Y cells were plated at a density of 30,000 cells/well on 384-well black-walled imaging plates and loaded for 30 min at 37 °C with Calcium 4 no-wash dye (Molecular Devices, Sunnyvale, CA) diluted in physiological salt solution (PSS; composition: 140 mM NaCl, 11.5 mM glucose, 5.9 mM KCl, 1.4 mM MgCl 2 , 1.2 mM NaH 2 PO 4 , 5 mM NaHCO 3 , 1.8 mM CaCl 2 , 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.4). SH-SY5Y cells represent an attractive model system and express human Ca v 2.2 channels in a physiologically relevant context since in native, systems Ca v channels are most likely co-expressed with auxiliary α2δ and β subunits.
Calcium responses, elicited by addition of 90 mM KCl and 5 mM CaCl 2 in the presence of 10 µM nifedipine, were measured using a FLIPR TETRA fluorescent plate reader (excitation, 470-495 nm; emission, 515-575 nm) after 5 min pre-treatment with test compounds in PSS containing 1.8 mM CaCl 2 . Under these conditions, the Ca 2+ response elicited by addition of KCl/CaCl 2 is mediated predominantly by GVIA-sensitive Ca v 2.2 [43]. ω-Conotoxin CVID (3 µM) was included as a positive control to define maximal inhibition of Ca v 2.2 responses.
Fluorescent responses were plotted as response over baseline using ScreenWorks (Molecular Devices, version 3. 1.1.4). Concentration-response curves of peak calcium responses, normalized to control responses, were generated using GraphPad Prism (Version 4.00, San Diego, California) using a 4-parameter Hill equation with variable Hill slope fitted to the data.

Conclusions
The diphenylmethylpiperazine moiety is less likely to lead to toxic effects than the phenoxyaniline substituent present in previously investigated anthranilamide-based GVIA mimics: Anilides are listed in structural alerts for potential adverse drug effects [48] with hepatotoxicity of anilides in certain cases being linked to P-450 activation to benzoquinoneimines in humans [49]. It was therefore pleasing to see that mimics bearing the diphenylmethylpiperazine pharmacophore (7a, 7b, 8a, 8b) showed very similar activities in the radioligand displacement assay, and improved inhibition of intracellular calcium responses in SH-SY5Y cells, relative to the best of the phenoxyaniline-bearing compounds. Two sets of related compounds that did not possess the diphenylmethyl substituent (7c, 7d, 8c, 8d) were inactive in the radioligand displacement assay, further suggesting that the diphenylmethyl moiety enhances calcium channel affinity in this class of compound.
The radioligand displacement assay employing 125 I-ω-conotoxin GVIA is thought to be a specific measure of binding affinity to Ca v 2.2 channels, because ω-conotoxin GVIA is exceptionally selective for this channel. As this assay does not measure the actual passage of Ca 2+ ions though the channel, there is always the possibility that test compounds could cause the 125 I-ω-conotoxin to be released from the channel through, for example, an allosteric effect, without actually inhibiting the channel. It was therefore also very pleasing to find that ω-conotoxin GVIA mimetics (7c, 7d, 8c, 8d) identified to be active in the radioligand displacement assay also induced functional inhibition of intracellular calcium responses in SH-SY5Y neuroblastoma cells and calcium currents in HEK293 cells stably expressing human Ca v 2.2 channels. These data fill an important gap between binding and downstream functional inhibition on median nerve-stimulated rat vas deferans that we have previously observed [50]. These results therefore validate both the use of ω-conotoxin GVIA as a starting point for the design of Ca v 2.2 channel blockers and the use of the radioligand displacement assay as an initial screening tool for Ca v 2.2 activity.
The results from the functional assays show the difluoro, diguanidino compound 8b to be the most effective Ca v 2.2 channel blocker of the compounds examined in this study. The results obtained with this compound will form the basis for further optimization work, with the aim of improving potency and incorporating more drug-like properties.