Inhibition of N-Type Calcium Channels by Fluorophenoxyanilide Derivatives

A set of fluorophenoxyanilides, designed to be simplified analogues of previously reported ω-conotoxin GVIA mimetics, were prepared and tested for N-type calcium channel inhibition in a SH-SY5Y neuroblastoma FLIPR assay. N-type or Cav2.2 channel is a validated target for the treatment of refractory chronic pain. Despite being significantly less complex than the originally designed mimetics, up to a seven-fold improvement in activity was observed.


Introduction
Neuropathic pain, which results from nerve damage caused by surgery, trauma, infection or disease, often does not respond to existing therapies [1,2]. Various estimates put the proportion of the world's population afflicted by this condition to be at least 3%, with up to 5% of postoperative patients being affected. Safe and effective therapies for neuropathic pain are therefore a major unmet medical need. N-Type calcium channels (Cav2.2 channels) are strongly implicated in chronic and neuropathic pain and their inhibitors have been widely pursued [1][2][3][4]. This approach has been best validated by Ziconotide (Prialt ® ), a synthetic version of the peptide ω-conotoxin MVIIA found in the venom of a fish-hunting marine cone snail Conus magnus. This peptide selectively targets Cav2.2 channels and is one of the very few effective drugs used to treat intractable chronic pain [5]. However, its intrathecal mode of delivery and narrow therapeutic window make it less than ideal as a treatment option.
We as well as others have been developing small-molecule inhibitors of Cav2.2 channels as possible alternatives to Ziconotide . Recently clinical development of Z160 (1, Figure 1), a reformulated form of NP118809 [8], was discontinued after Z160 (1) failed to meet the primary endpoint in Phase II clinical studies [30]. As a result there is only one compound that targets Cav2.2 channels currently in clinical trials for the treatment of chronic pain, CNV2197944, the structure of which is yet to be disclosed [24]. As part of an ongoing program to develop new small-molecule inhibitors of Cav2.2 channels, the pharmacophore of ω-conotoxin GVIA, a 27 residue peptide present in the venom of the cone snail Conus geographus, has been investigated. This peptide binds essentially irreversibly to the Cav2.2 channel, making it unattractive as a therapeutic, however its well-defined structure has facilitated the development of peptidomimetics. A number of such mimetics have been disclosed [25][26][27][28][29], developed using the α,β-bond vector strategy described by Bartlett and Lauri [31], combined with interactive de novo design [32]. In these mimetics the biologically relevant tyrosine, lysine and arginine side chain mimics are projected from a central scaffold, as illustrated by the anthranilamide derivative (2) ( Figure 2) [27,32]. Guided by the results of a radioligand-displacement assay, it has been concluded that with these anthranilamide-based mimetics (Compounds 2-4, Figure 2) the optimum length of the alkyl side chains is n = 7 for the lysine mimic and n = 3 for the arginine mimic. It was also found that the replacement of the phenol functionality with fluorine is well tolerated [25]. The most potent fluorinated analogue in this series was found to be the diguanidino compound (3) while the corresponding diamino compound (4) had comparatively weak activity. This is consistent with previous results where truncation of the arginine side chain mimic to an amine in ω-conotoxin GVIA mimetics is typically detrimental to activity at the N-type channel [25][26][27][28][29].  In order to transition conotoxin mimics towards more drug-like compounds, a number of their physiochemical properties need to be adjusted. Marketed central nervous system (CNS) active drugs, for example, tend to have much lower molecular weights, percentage polar surface areas, total number of nitrogen and oxygen atoms, and hydrogen bond acceptors and donors than are found in mimetics like 2. We have therefore embarked on a program of molecular modifications aimed at improving the physiochemical properties of this class of conotoxin mimics while retaining activity at the N-type calcium channel. A major priority has been to reduce overall molecular weight. Encouraged by favourable results obtained with the simplification of a benzothiazole class of mimetics, which involved the deletion of one of the amino acid side chain mimics [28], a similar strategy has been pursued with the anthranilamides. Thus, in the study described here, the effect on activity of the deletion of the lysine side chain mimic in compounds 2-4 has been investigated, together with the SAR related to the substitution pattern of the central aromatic ring (ortho, meta or para). The amino analogues and their corresponding monoguanidino analogues that were synthesised and tested in this study are shown in Figure 3, compounds 5a-c and 6a-c.

Biology
It has been shown that SH-SY5Y neuroblastoma cells endogenously expressing human Cav2.2 channels allow the rapid screening of potential channel blockers by means of a FLIPR assay [39]. The synthesised compounds, the amino (5a-c) and the guanidinium (6a-c) analogues, as well as the anthranilamide-based mimetic (3), were evaluated for their ability to inhibit Cav2.2 calcium responses in SH-SY5Y cells in the presence of the L-type calcium channel blocker nifedipine. It was found that Ca 2+ ion channel responses elicited by KCl-mediated depolarization were inhibited in a dose-dependent manner (Table 1). Compound 3 only partially inhibited responses at a concentration of 1 mM, resulting in an estimated IC50 value of 1452 µM. In contrast, 5a and 5b fully blocked KCl-induced Ca 2+ responses with IC50s of 46 µM and 35 µM, respectively. In the guanidinium series, 6a and 6b retained weaker activity with IC50 values of 124 µM and 185 µM respectively. [40] Both para substituted compounds, 5c and 6c, were only weakly active and partially inhibited responses with IC50 values of 764 µM and 723 µM, respectively.
Compared to compound 3, compounds 5a-c and 6a-c are significantly less complex and have molecular weights reduced by 33%-45%. It is encouraging, therefore, to find that all but 5c and 6c are considerably more active than 3. A relationship between the substitution pattern around the central aromatic ring and biological activity can also be clearly seen, with the ortho and meta analogues showing considerably stronger activity than the para analogues. It is also interesting to note that the amino compounds 5a-b are three to five fold more active than the guanidino compounds 6a-b.

General Experimental Procedures
Starting materials and reagents were purchased from Sigma-Aldrich (Sydney, Australia) and used without purification. Solvents were dried, when necessary, using standard methods. Normal phase flash chromatography was performed on Merck silica gel No. 9385. Spectra were recorded on a Bruker Av400 or Av600 spectrometer (Fallanden, Switzerland). NMR spectra were referenced to residual solvent peak [chloroform (δH 7.26, δC 77.2), methanol (δH 4 .87, 3.30, δC 49.0)]. The units for all coupling constants (J) are in hertz (Hz). ‡ Denotes signals only observed in 2D NMR. Mass spectrometry (APCI) was performed on a Thermo Scientific Q-Exactive FTMS. High-resolution mass spectra were recorded on a Waters Q-TOF II (Manchester, UK) or Thermo Scientific Q-Exactive FTMS mass spectrometer (Bremen, Germany). Melting points were recorded on a Stuart Scientific Melting Point Apparatus SMP3. Infrared spectra were recorded on a Perkin-Elmer RXI FTIR Spectrometer as thin films. Preparative HPLC was performed on a Waters Prep LC 4000 System using an Alltima C18 column (22 × 250 mm, 5 micron), detection at 237 nm. Mobile phase 12 mL/min 30% CAN/H2O/0.2% TFA isocratic for 135 min then 115 min gradient to 100% ACN containing 0.2% TFA.

N-(3-(4-Fluorophenoxy)phenyl)-4-(3-guanidinopropoxy)benzamide hydrochloride (6b)
The guanidinylated compound 6b was synthesised according to a modified procedure by Bernatowicz et al. [38]. The amine 5b (47 mg, 0.12 mmol), DIPEA ( pre-organisation through restricted rotation might enhance the activity of the ortho and meta analogues (5a, 5b, 6a and 6b). It is also unusual for the amines to be more active than the guanidines in this class of channel blocker. While primary amines typically do not make good drugs, this observation does open the possibility of developing N-type channel blockers capable of crossing the blood brain barrier. Compounds bearing very strongly basic functional groups like guanidine are unlikely to cross the blood brain barrier [45] whereas there are many CNS-active drugs that bear tertiary amines. The mode of action of the compounds reported here is currently being investigated in patch clamp electrophysiology experiments, the results of which will be reported in due course.