CHAPTER 1: Ion Channel Drug Discovery: a Historical Perspective

Ion channels are membrane proteins that control the flow of ions across the cell membrane, are present in the membranes of all cells and make up one of the two traditional classes of ionophoric proteins along with ion transporters. Ion channels are responsible for maintaining resting membrane potential and all electrical signaling; they play a key role in the modulation of intracellular calcium levels crucial for the regulation of many cellular functions events and consequently a fundamental role in many physiological processes.

Ion channels are classified in a number of ways; by gating, i.e. what opens and closes the channels. Voltage-gated ion channels open or close depending on the voltage gradient across the plasma membrane, while ligand-gated ion channels open or close depending on binding of ligands to the channel. Further classifications are by the ion (or ions) that is (are) conducted e.g. sodium, calcium, potassium, proton, chloride or non-selective, or by the duration of the response to stimuli e.g. the transient receptor potential channels (TRP channels). Finally, for example in the potassium channel superfamily, they are classified by the number of pore loops contained in each channel forming subunit. The vast majority of channels form as tetramers of subunits that each contributes a single pore loop to the ion-conducting pore; however, there is a small family of two-pore-domain potassium channels (K2P), where channel subunits have two pore loops each; channels are formed as dimers of these subunits.¹ 

In the same way as “kinome” is used to describe the protein kinase family of enzymes within the genome, the term “chanome” or “channelome” is often used to describe the >300 members of the ion channel family assembled from the ∼500 annotated ion channel proteins as predicted in the human genome. With ion channels being such a large class of targets and involved in many key physiological processes it is surprising that less than twenty percent are currently commercially exploited.

When co-authoring an article on ion channels in 2005,²  examination of the marketed ion channel modulators at that time highlighted that the majority of ion channels targeted up to that point had almost exclusively been ligand- or voltage-gated channels found in excitable tissues such as nerve and muscle. Since then there has been a number of drugs entering the market: retigabine (1), verenicline (2) and ivacaftor (3), and promisingly a considerable number of compounds in phase II/III clinical development with a growing number targeting channels in non-excitable tissues.³  Table 1.1 lists a selection of the marketed ion channel modulators, their use and their specific ion channel target. Figure 1.1 illustrates the sustained commercial interest in the ion channel area as judged by the number of patent filings.

Classic examples of early ion channel modulator discovery are the voltage-gated sodium channel blockers used as local anesthetics and anticonvulsants and the L-type voltage-operated calcium channels blockers (L-VOCCs) which are still in clinical use many years after their first synthesis. Following an early understanding of the structural features of the natural product cocaine (4), the local anesthetic benzocaine (5) was synthesized in 1890, then amylocaine (6) in 1903 and procaine (7) in 1905. This was the start of a classic age of chemistry-driven drug discovery, optimizing compounds for efficacy and duration in animal models resulting in such compounds as procainamide (8) and lignocaine (lidocaine) (9).^4–6  It was not until the late 1950's that these compounds were shown to “inhibit” purported sodium channels which at the time were hypothesized to be integral in nerve impulse transmission.⁷ 

In the anticonvulsant field, phenytoin (10), synthesized as a phenobarbitone analogue in 1937, was introduced for the treatment of epilepsy followed by carbamazepine (11) in the early 1960's and lamotrigine (12) in the late 1980's. During this research compounds were selected primarily using animal models such as the maximal electroshock model in rodents. Later data from electrophysiological experiments on brain slices, in cloned channel systems and disposition studies has led to a better understanding of the mode of action of these compounds,⁸  resulting in the derivation of anticonvulsant pharmacophore models for sodium channel blockers.^9,10 

Many of the anticonvulsant sodium channel blockers find off-label use in a variety of neuropathic pain conditions¹¹  e.g. compound 13 (CNV-1014802) is a compound from a distinct structural class, shown to be anticonvulsant¹²  and is currently under clinical investigation for neuropathic pain. The compound, structurally related to a number of earlier published compounds (e.g. CO 102862 (14)¹³  and safinamide (15)¹⁴ ) and recent compounds published by workers at Merck (e.g. 16¹⁵ ) is described as a state-dependent sodium channel blocker that exhibits potency and selectivity against the Nav1.7 channel. This molecule was recently granted orphan-drug designation by the US FDA in July 2013 for the treatment of trigeminal neuralgia.¹⁶  Even now we are still to fully understand the precise mode of action of these compounds on the nine members of the voltage-gated sodium channel family (Nav1.1-1.9) that contribute to the range of pharmacology exhibited (see chapter 5 for an extensive review of the Nav channel family).

The recently discovery of the novel anticonvulsant retigabine (1) followed a similar retrospective discovery route. Originally synthesized as an analogue of the opioid analgesic and muscle relaxant flupirtine in 1993 by researchers at ASTA Medica, it was found to be anticonvulsant. Mechanistically this was explained as the result of modulation of the neuronal “M current”. Now with the benefit of molecular insight, retigabine has been shown to be an activator of Kv7.2 (KCNQ2/3). This is covered in detail in chapter 10.

The discovery of L-type calcium channel blockers and their utility in coronary disease occurred by chance in 1963, when it was reported that new compounds such as the phenylalkyamine (17, later named verapamil) mimicked the cardiac effect of simple calcium withdrawal, diminishing calcium-dependent high energy phosphate utilization, contractile force and oxygen requirement. In 1969, the term “calcium antagonist” was given a novel drug designation. In an extensive search for other calcium antagonists, a considerable number of substances that also met these criteria were identified; in 1975 the first dihydropyridines (e.g. nifedipine (18)) were discovered followed by many other members of this class in the following decades, including longer-acting compounds such as amlodipine (19). Also in 1975, a third class of L-type calcium channel blockers was discovered, the benzothiazepine class e.g. diltiazem (20).^17–19 

Ion channel targets in non-excitable tissues remained virtually unexplored in the early period of research apart from the discovery of the sulfonylurea class of anti-diabetic agents that modulate the ATP-sensitive potassium channels (KATP) in pancreatic cells. The sulfonylureas were synthetic off-shoots from the extensive work on sulfonamide antibacterials which were in turn derived from azo dyes (the dye prontosil was found to be a metabolic pro-drug for sulfanilamide). Janbon and co-workers observed that using the sulfonamide isopropylthiadozole (21) to treat typhoid patients caused a high incidence of hypoglycemia and at the same time Loubatières working with the compound in dogs found it to have ‘insulin-like’ properties.²⁰  Compounds such as tolbutamide (22) entered the clinic for the treatment of type-II diabetes in 1956 but are now largely superseded by second-generation agents such as glibenclamide (glyburide) (23).²¹  Almost 40 years after the discovery of the first compounds, the KATP channel in pancreatic β cells was identified as the molecular target for sulfonylureas,²²  and another decade passed before the molecular composition of the pancreatic KATP channel was identified, a complex of the pore-forming inwardly rectifying channel Kir 6.2 and sulfonylurea receptor type 1 (SUR1) regulatory subunit.²³ 

In the 1970's and 80's ion channel research was driven by the pharmacological discovery and classification of ion channel targets in tissue using a variety of synthetic tool molecules designed from endogenous ligands or pharmacologically active natural products. The discovery of membrane-permeable fluorometric indicator dyes allowed the measurement of intracellular calcium levels and membrane potential and the development of the patch-clamp technique allowed individual ion channel activity to be measured for the first time. Data could therefore be provided that gave kinetic and biophysical information to aid the design of potent and functionally selective agents. The discovery stories associated with the 5HT₃ and the neuromuscular nicotinic acetylcholine receptors (nAChRs) have been well documented and resulted in a considerable number of therapeutically useful drugs, e.g. the antiemetic agent ondansetron (24) by workers at Glaxo (now GSK)^24,25  and the neuromuscular blocking agent atracurium (25) by academic workers at Strathclyde University in collaboration with Wellcome (now GSK).²⁶  The latter was designed from consideration of other bi-quaternary compounds and d-tuborcurarine, an alkaloid from the South American arrow poison curare, all used as neuroblocking agents at that point. The elimination of atracurium is not dependent on metabolic breakdown by esterases, unlike other agents, and was found to be under polymorphic control thus causing unpredictable duration of muscular relaxation between patients. Instead atracurium was designed to rely on pH dependent Hoffman elimination, which provided highly predictable onset and duration of action across patients.

Natural product research still remains an important source of ion channel modulators (see chapter 12) either as leads for medicinal chemistry programs or as drugs in their own right. The recently launched varenicline (2), a partial agonist of the α4β2 nicotinic acetylcholine receptor, utilized therapeutically in smoking cessation, was designed from (−)-cytisine (26) following observations that it was a partial agonist and antagonized the receptor response to its endogenous neurotransmitter, acetylcholine.²⁷ 

The cone snail peptide ziconotide (27) entered clinical use in the last decade for the treatment of severe pain and is a highly potent blocker of the N-type voltage-gated calcium channel. It is delivered as an infusion into the cerebrospinal fluid using an intrathecal pump system.²⁸  Continuing with the larger molecule theme, antibodies represent a growing field of research for finding ion channel modulators; this subject is covered in chapter 13.

In the 1990's advances in molecular pharmacology enabled the recombinant expression of channel targets in heterologous cellular backgrounds that could be used in high throughput screening modes. Chapter 2 covers the impact and advances in high throughput screening in the ion channel field. VX-770 (ivacaftor) (3) was approved by the FDA in January 2012 for the treatment of cystic fibrosis in patients with the G551D mutation and probably represents the first marketed compound to be derived from a high throughput screen hit. This molecule stemmed from work carried out at Vertex using cell-based fluorescence membrane potential assays in high-throughput mode to identify potentiators and correctors of CFTR function. Early hits from the potentiator and corrector screens are exemplified by compounds VRT-422 (28) and VRT-532 (29), respectively.^29,30  Workers at Genzyme utilized a similar approach to identify hits such as 30.³¹ 

Another approach towards the treatment of cystic fibrosis is the blockade of the epithelial sodium channel (ENaC) with research focusing on topical administration into the lung of analogues of the potassium sparing diuretic amiloride (31), developed by Merck in the 1960's.³²  Amiloride itself has been studied in a number of clinical trials with conflicting results; the lack of efficacy of amiloride may be due to its short duration of action: it is cleared rapidly from the airways and its effect on lower airway potential difference lasts for only approximately 30 minutes.³³  Consequently, research efforts have focused on the design of longer acting analogues; this work is described in chapter 7. Recently, amiloride has also been shown to be a non-selective inhibitor of acid-sensing ion channels (ASICs) and exhibits a modest effect in rat pain models at high concentrations. Structural modification has provided more potent analogues, with particular focus on ASIC3 as the specific family member targeted for blocking chronic inflammatory pain.³⁴ 

The ionotropic glutamate receptors (iGluRs) comprising of the AMPA, NMDA and kainate families have received much attention as targets for a variety of CNS disorders with a large number of compounds in clinical trial.³  Perampanel (32), a non-competitive AMPA-type glutamate receptor antagonist, was recently reported to be well-tolerated and effective in reducing seizure frequency in partial onset epilepsy versus placebo.³⁵  Chapter 6 describes an approach to discovery of allosteric modulators of this target where co-crystallization of hits derived from screening with a ligand binding domain construct was used to aid structural based design.

Another family of ion channels that has received much interest over the past decade is that of the transient receptor potential channels (TRP channels) mentioned in the introduction. They are divided into two groups: group 1 containing the TRPC, TRPV, TRPM, TRPN and TRPA sub-groups and group 2 containing the TRPP and TRPML sub-groups. The fascinating prospect for modulation of these channels is that they mediate a variety of key bodily sensations such as pain, hot or cold, tastes, pressure, and aspects of vision.³⁶  The search for antagonists of TRPV1 as potential analgesics has been a major endeavor for many researchers and is covered in chapter 9.

One channel that has received more attention than almost any other is the hERG channel. HERG refers to the human Ether-à-go-go-Related Gene, alternately known as KCNH2, that codes for the potassium channel alpha subunit Kv11.1. This channel is involved in the mediation of the repolarizing current (I_Kr) in the cardiac action potential. HERG is actually a therapeutic target with compounds such as dofetilide (33), a class III antiarrhythmic agent, used for the maintenance of sinus rhythm.³⁷  Unfortunately the main attention has not been for positive reasons but as a target associated with off-target pharmacology (a so-called “anti-target”) manifesting itself as life threatening drug-induced long-QT syndrome. Terfenadine (34) is perhaps the most well-known example, used widely as an antihistamine until reports of long-QT syndrome prompted its withdrawal. It was found to be a hERG blocker that under normal circumstances was rapidly metabolized by CYP3A4 on first-pass to an active metabolite that did not block hERG. However, many other drugs (e.g. erythromycin) and some foods (e.g. grapefruit juice) inhibit the function of CYP3A4 thus preventing the metabolism of terfenadine, leading to accumulation in the plasma and inevitable cardiac side-affects. The active metabolite of terfenadine, fexofenadine (35), is now marketed as an antihistamine in its own right and gave a structural insight into how to avoid hERG affects.³⁸  The structure–activity relationships of hERG modulation have been widely studied; a growing pharmacophoric understanding and careful pharmacological screening now aid drug design in avoiding hERG associated liabilities.^39,40  A full discussion on hERG is presented in chapter 11.

Since Rod Mackinnon and Peter Agre received the 2003 Nobel Prize for chemistry for discoveries concerning channels in cell membranes and for the discovery of water channels, respectively,⁴¹  there has been an increasing numbers of ion channel crystal structures solved, with an expectation that many more structures will be determined in the coming years. The form and composition of native ion channels has also been under scrutiny and it is clear now that ion channels are not isolated ion-conducting pores but are integral components of structural and signaling complexes in cells. They do not just function in isolation but have closely associated accessory scaffolding sub-units and proteins, which are important for the functionality of the native channel, not only in terms of activation but also in terms of ensuring correct assembly, trafficking, insertion and retrieval.⁴²  Chapter 4 gives further insights into the study of ion channel structure.

Despite the availability of cellular systems containing cloned channel targets and the growing amount of information regarding ion channel structure and function, the pursuit of ion channel targets in high throughput screening mode has not progressed as well as other target classes, such as GPCRs and enzymes. Often screens use indirect assay readouts, such as ion-sensitive and potentiometric dyes, radiotracer flux and binding assays; although much higher throughput, all have well-documented limitations, such as sensitivity and a high potential for false positives.⁴³  Electrophysiology represents the highest fidelity option for ion channel screening, but was exceptionally low throughput, time consuming and required technically skilled operators. However, the advent of automated electrophysiology offers the opportunity of running direct electrophysiology-based HTS campaigns; this is covered in chapter 3.

Ion channel research has come a long way in the last 100 or so years, providing a wealth of novel and important medicines for the clinic. Continuing scientific and technological advances in the field of screening, structural-based design, the increasing chemical space of modulators available (low molecular weight, peptides, antibodies etc.) coupled with the availability of genetic information, raises the exciting prospect of taking us into a new more predictive and hopefully productive era of ion channel drug discovery.

Many thanks to Andrew Davis for carrying out the searches to provide the data expressed in Figure 1.1.