Scaffold-based design and synthesis of potent N-type calcium channel blockers
Abstract
The therapeutic agents flunarizine and lomerizine exhibit inhibitory activities against a variety of ion channels and neurotransmitter receptors. We have optimized their scaffolds to obtain more selective N-type calcium channel blockers. During this optimization, we discovered NP118809 and NP078585, two potent N-type calcium channel blockers which have good selectivity over L-type calcium channels. Upon intraperitoneal administration both compounds exhibit analgesic activity in a rodent model of inflammatory pain. NP118809 further exhibits a number of favorable preclinical characteristics as they relate to overall pharmacokinetics and minimal off-target activity including the hERG potassium channel.
Voltage-gated ion channel blockers are attractive drug targets for an expanding range of therapeutic indications.1 Voltage-depen- dent calcium channels play a central role in the control of cellular excitability and a number of calcium-dependent cellular functions, including gene transcription and neurotransmitter release.2 Bio- physical and pharmacological studies have identified four subtypes of high voltage activated (HVA) calcium channels that are encoded by a family of seven different a1 subunit proteins (Cav): L-type
(Cav1.1–Cav1.4), N-type (Cav2.2), P/Q-type (Cav2.1) and R-type (Cav2.3) as well as low voltage activated calcium channels, called T-type, encoded by three distinct a1 subunit proteins (Cav3.1–Cav3.3).3 A subset of the HVA calcium channels are located at the presynaptic termini of neurons, where they are directly involved in the regulation of neurotransmitter release. Of particular interest, the transmission of pain signals from periphery to the central nerziconotide®), although the market for Ziconotide® is limited by the requirement for intrathecal delivery.6
The objective of the current drug discovery program was the identification of orally active, selective N-type blockers with an acceptable therapeutic index aimed at the treatment of chronic/ inflammatory pain. Both flunarizine7 and lomerizine8 possess diphenylmethylpiperazine moieties as the basic skeleton (Fig. 1) and initial structure–activity relationship (SAR) investigations indicated that this backbone contributed to calcium channel block- ing activities. In the present report we investigated the structural requirements for selective N-type calcium channel blockade around the flunarizine and lomerizine backbones. Antagonist activ- ity was measured in HEK293 cells transiently transfected with rat brain calcium channel subunits cDNAs (N-type: a1B + a2d-1 + b1b subunits; L-type: a1C + a2d-1 + b1b subunits). After incubation for 24–72 h, whole cell current recordings were performed with barium as the charge carrier as previously described.9 Currents were typically elicited from a holding potential of 100 mV to the peak of current–voltage relation for each channel type. For each calcium channel subtype each compound was examined for blockade by patch clamp analysis on between 3 and 5 cells.
We began by testing the potency of flunarizine and lomerizine against the cloned N-type calcium channel exogenously expressed in HEK cells and found that they both potently inhibited functional N-type channels (estimated IC50 values of 0.08 and 0.09 lM,
respectively; Table 1). However, their inhibitory activities against L-type channels was also high, providing only a 1.7–3.9-fold ratio of selectivity over the two types of high voltage-activated calcium channels (Table 1). A major direction of the SAR optimization was to decrease L-type activity while maintaining the high affinity for N-type calcium channel blockade.
We hypothesized that the benzhydrylpiperazine and trimeth- oxybenzylpiperazine moieties, respectively, could be acting as selectivity elements. With this in mind, we designed and synthe- sized a series of compounds replacing the cinnamyl portion in flu- narizine. We first examined the effect of the methylene spacer between the piperazine and the benzhydryl group (Table 2). Fur- ther, we investigated the effect on N-type calcium channel activity and selectivity by changing the amine into an amide (Table 3). The SAR studies of these chemical series were guided by the whole cell patch clamp analysis of the functional blockade of N-type and L- type channels expressed in HEK cells.
Table 2 shows that the benzhydrylpiperazine backbone with a benzhydryl group on the right hand side resulted in compounds with a significantly poorer degree of N-type blockade (compounds 1a–1e) compared with either parent compounds. Varying the length of the linker between the benzhydryl and the piperazine core resulted in a 10-fold range of N-type affinities suggesting that linker length nonetheless contributed to N-type blocking affinity.
Table 3 shows that the introduction of an amide into the right hand linker was also generally unfavorable compared to the parent compounds. A notable exception to this was compound 2b (NP118809; see Scheme 1) with a total linker length of 3 carbons (n = 1) and which exhibited approximately a 500-fold increased N-type channel blockade compared to compounds in this series with linkers either shorter (n = 0, 2a) or longer (n = 2–4, (2c–e). Examination of the effect of NP118809 on functional L-type cal- cium channel activity showed that this compound was approxi- mately 111-fold more selective for N-type channels.
To probe the effect of bioisosteric replacement of carbon on the NP118809 linker, we examined the bioisosteric potential of 3,3- diphenylpropan-1-one bound to the piperazine core, while main- taining the benzhydryl moiety on the opposite side of molecule (Scheme 2). This modification allowed the maximum conservation of the polarity and geometry of the NP118809 (Table 4). While compound 3a maintained favorable inhibitory activity for N-type channels (est. IC50 = 0.15 lM) compared to that of NP118809, the N-type:L-type selectivity profile was significantly lower (12-fold). Further exploration on the linker with compounds 3b–3c showed a trend toward decreased N-type channel blocking affinity (Table 4). Examination of a series of derivatives focused around the lomerizine-based trimethoxybenzylpiperazine moiety resulted in compounds with a generally high degree of N-type channel block- ade. Unlike that for the flunarizine-based series, a longer linker to the benzhydryl group (six carbons total length) did not seem to be deleterious. Furthermore, there was little change in N-type affinity regardless of whether an amide was included in the linker on either side of the piperazine core (Tables 5 and 6). In particular, compound 4a (NP078585; see Scheme 1) exhibited potent N-type channel functional blockade (est. IC50 = 0.11 lM). Examination of
the effect of NP078585 on L-type calcium channels showed that this compound was approximately 25-fold more selective for N- type channels (Table 5).
Replacement of the trimethoxybenzyl with cinnamyl moiety was studied without changing the chain linker lengths (Table 6). Compound 5a had the closest N-type potency compared to NP078585 (est. IC50 = 0.05 lM) but possessed a lower selectivity ratio of L-type to N-type blockade ( 7.8-fold). The other deriva- tives in this series all exhibited sub-micromolar N-type channel affinities but again with no improvement over NP078585 in the L-type to N-type selectivity ratio (Table 6).The two compounds demonstrating high affinity for the N-type channel together with the best selectivity over L-type calcium channels were selected for pharmacokinetic profiling. Examination of pharmacokinetics and oral bioavailability showed that in rats was suggestive of a high degree of protein binding and/or distribu- tion outside of plasma.
Based upon their combined in vitro and pharmacokinetic prop- erties NP078585 and NP118809 were subject to in vivo assay in the rat formalin model inflammatory pain.10 Figure 2 shows that upon intraperitoneal (ip) administration both NP118809 and NP078585 (25 mg/kg) both exhibited significant analgesic activity in the phase IIA portions of the rat formalin model.
In order to further refine the preclinical profiles of the two can- didate molecules their off-target interactions with the hERG potas- sium channel was examined by whole cell patch clamp analysis of HEK cells stably expressing hERG. A single 1 lM dose application of NP078585 was found to significantly block hERG currents (98.7 ± 0.2%, n = 3) suggesting the potential for cardiovascular lia- bility for this agent. Contrastingly, an initial single test dose test of NP118809 (1 lM) resulted in only partial blockade of hERG currents (24.8 ± 0.3%, n = 3). Subsequently, a dose–response profile on hERG tail currents was determined for NP118809. Figure 3 shows the estimated IC50 of NP118809 for the hERG channel extrapolated from the concentration-dependent inhibition curve = 7.4 lM. As such, NP118809 would not be considered a potent blocker of the hERG potassium channel.
Further examining NP118809 at a concentration of 10 lM for off-target activity using a radioligand binding displacement screen of 112 ion channel, receptor and other signaling targets (MDS Phar- ma Services; http://discovery.mdsps.com) resulted in >50% dis- placement of radioligand from only four targets: L-type calcium channel, cannabinoid CB1, l opiate receptor and voltage-gated sodium channel. Subsequent secondary screens for possible func- tional interactions against these targets by NP118809 using representative tissue assays (MDS Pharma Services) showed no significant agonism or antagonism (>50%) at a higher test concen- tration of 30 lM. That there was no observed physiological effect of NP118809 on L-type channel function in a guinea pig model of atrial inotropy while we determined an IC50 ~ 12 lM against the L-type channel by electrophysiological assessment is likely reflec- tive of the highly state-dependent nature of NP118809 channel blockade (data not shown). Voltage- and frequency-dependent blockade would favor drug interaction with a subset of channels under native conditions and are predicted to be important factors towards maximizing the therapeutic window of ion channel block- ers in vivo.11
In summary, SAR studies around the flunarizine and lomerizine diphenylmethylpiperazine backbones produced a series of high affinity N-type calcium channel blockers (<200 nM IC50s) that exhi- bit a significant degree of selectivity (25–111-fold) over the closely related L-type channel. NP078585 and NP118809 were discovered as N-type calcium channel antagonists that exhibit strong analgesic activity in an inflammatory rat model. Both compounds exhibit suitable pharmacokinetic characteristics for animal testing although NP118809 appears to be the more favorable preclinical candidate based upon its lower off-target affinity to the hERG potassium channel and lack of significant interaction against 112 additional molecular targets. Some relevant experimental details have already been published in another form.12