Pharmacology of Small- and Intermediate-Conductance Ca²⁺-Activated K⁺ Channels

Abstract

The three small-conductance calcium-activated K_Ca2 channels and the related intermediate-conductance K_Ca3.1 channel are voltage-independent K⁺ channels that mediate calcium-induced membrane hyperpolarization. When intracellular calcium increases in the channel vicinity it calcifies the flexible N-lobe of the channel-bound calmodulin, which then “swings over” to the S4-S5 linker and opens the channel. K_Ca2 and K_Ca3.1 channels are highly druggable and offer multiple binding sites for venom peptides and small molecule blockers as well as for positive and negative gating modulators. In this review we will first briefly summarize the physiological role of K_Ca channels and then discuss the pharmacophores and the mechanism of action of the most commonly used peptidic and small molecule K_Ca2 and K_Ca3.1 modulators. Finally, we will describe the progress that has been made in advancing K_Ca3.1 blockers, and K_Ca2.2 negative and positive gating modulators towards the clinic for neurological and cardiovascular diseases, and discuss the remaining challenges.

INTRODUCTION

Potassium (K⁺) channels are critically involved in regulating fundamental physiological processes such as cellular volume, membrane potential, hormone secretion, calcium signaling, and action potential firing (1). To allow for fine tuning of these processes the human genome contains 78 K⁺ channels. According to the IUPHAR (International Union of Basic and Clinical Pharmacology) Guide to Pharmacology (2) these channels have been grouped based on sequence similarity and the number of their transmembrane domains (TM), which can be 2, 4, 6 or 7. The small- and intermediate-conductance Ca²⁺-activated K⁺ channels belong to the 6TM family and thus resemble the voltage-gated K⁺ channels, with a 4TM voltage sensor domain (VSD) and a 2TM pore domain (3). Like K_V channels, functional K_Ca channels are tetramers and, at least in expression systems, the different family members are able to form heteromultimers in addition to the more common homotetramers (4). However, unlike K_V channels, K_Ca channels have fewer positively charged residues in the S4 segment of their VSD, and are therefore unresponsive to changes in transmembrane voltage and have essentially linear current voltage-relationships at physiological ion gradients (3; 5). Before their cloning, K_Ca2 and K_Ca3.1 channels were referred to as small-conductance (SK) or intermediate-conductance (IK) Ca²⁺-activated K⁺ channels based on their unitary conductance of ~10 pS or ~40 pS in symmetrical K⁺ solutions to differentiate them from the large-conductance BK channel (~200 pS), which is now called K_Ca1.1 (3). This phenomenological nomenclature is still widely used together with the IUPHAR and HUGO nomenclatures, which is why we wanted to introduce it upfront to avoid confusion.

The three K_Ca2 channels, K_Ca2.1 (SK1, KCNN1), K_Ca2.2 (SK2, KCNN2) and K_Ca2.3 (SK3, KCNN3) were cloned in 1996 by the group of John Adelman and are highly homologous across their transmembrane cores (80–90%) but diverge in sequence and length in their N- and C-termini (5). Two years later, the same group demonstrated that the calcium-sensor of the SK channels is calmodulin, which is constitutively bound to the calmodulin binding domain (CaM-BD) in the C-terminus and thus functions as a β-subunit that endows these channels with Ca²⁺ sensitivity (6). The K_Ca3 family only contains a single member, K_Ca3.1 (IK, SK4, KCNN4), which was cloned in 1997 and delegated to its own subfamily because it only is ~40% identical to the three K_Ca2 channels (7; 8). Similar to the K_Ca2 channels, the Ca²⁺-dependent activation of K_Ca3.1 is mediated by calmodulin (9). The reported EC₅₀ values for Ca²⁺ range from 100 to 400 nM for K_Ca3.1 and from 300 to 750 nM for the K_Ca2 channels. This submicromolar Ca²⁺-sensitivity together with their lack of voltage-dependence enables K_Ca channels to be open at relatively negative membrane potentials when intracellular Ca²⁺ is raised in their immediate vicinity and to hyperpolarize towards the K⁺ equilibrium potential of −90 mV. K_Ca channels are accordingly expressed in cells that need to be able to prevent premature action potential generation or sustain Ca²⁺ influx through inward-rectifier Ca²⁺ channels.

EXPRESSION AND PHYSIOLOGICAL FUNCTION OF K_Ca2 CHANNELS

K_Ca2 channels are widely expressed on neurons of the central and peripheral nervous system (10). K_Ca2 currents underlie the so-called medium afterhyperpolarization (mAHP), the second phase of neuronal hyperpolarization following an action potential, and thus regulate intrinsic excitability and spike firing rates. K_Ca2 currents are activated by calcium entering neurons via Ca_V channels activated during the action potential, but may also functionally couple to post-synaptic calcium sources such as NMDA receptors and nicotinic acetylcholine receptors as well as calcium released from intracellular ryanodine or IP₃ receptors (10).

The mAHP is absent in K_Ca2.2 knock-out mice, but not mice lacking K_Ca2.1 or K_Ca2.3, revealing that K_Ca2.2 is the dominant subtype responsible for this current (11). Reversely, 10-fold overexpression of K_Ca2.2 increases the mAHP and dampens excitatory postsynaptic potentials (EPSP) making postsynaptic neurons less likely to fire, resulting in mice with impairments in hippocampal learning and memory (12). K_Ca2.3 channels are strongly expressed on dopaminergic substantia nigra neurons, where they maintain regular firing and pace-making control (13) and serotonergic raphe neurons, where they regulate burst firing (14). Pharmacological modulation confirms the role K_Ca2 channels in neurons (Figure 1). The selective K_Ca2 blocker apamin increased intrinsic excitability and firing frequency, while K_Ca2 activators enhance the magnitude of the mAHP and slow down firing rates (10; 15). These altered responses have consequences for long-term potentiation (LTP), the increase in synaptic strength following high-frequency stimulation, which underlies some forms of learning. Inhibiting K_Ca2 activity with apamin improves learning and memory encoding in rodents (16; 17), while increasing K_Ca2 activity with an activator impairs associative learning (18). In the peripheral nervous system, K_Ca2 channels are expressed in DRG primary sensory neurons, where they play a role in nociception (19; 20).

Figure 1.

Physiological role of K_Ca2 and K_Ca3.1 channels.

While all three K_Ca2 channels are found in the nervous system, K_Ca2.3 is the only member found in the endothelial cells of blood vessels, where, together with K_Ca3.1 (21), K_Ca2.3 underlies the endothelium derived hyperpolarization (EDH), a phenomenon that, together with prostacyclin and nitric oxide, controls vessel tone (22). Activation of K_Ca currents on the endothelium leads to hyperpolarization and relaxation of the underlying smooth muscle cells, ultimately reducing blood pressure (Figure 1). K_Ca2 channels are further expressed in the heart (Figure 1) and have been shown to play an important a role in the repolarization of the cardiac action potential, specifically in atrial myocyte and atrioventricular nodes (23; 24). Multiple single nucleotide polymorphisms (SNPs) in K_Ca2.3 have been found to be associated with lone atrial fibrillation (AF) (25; 26). Increasing K_Ca2.3 activity significantly shortens cardiac action potentials resulting in increased susceptibility to AF (27). A K_Ca2.3 splice variant is expressed in liver hepatocytes, where K_Ca2.3 plays a role in cellular responses to metabolic stress (28).

EXPRESSION AND PHYSIOLOGICAL FUNCTION OF K_Ca3.1 CHANNELS

A calcium-activated K⁺ efflux, which was later demonstrated to be mediated by K_Ca3.1 and to contribute to volume regulation and hydration state (29; 30), was fist described in erythrocytes in 1958 (31), which is why K_Ca3.1 is also called the Gárdos channel after the scientist who first described the phenomenon. Heterozygous gain of function mutations in K_Ca3.1 are responsible for erythrocyte dehydration in a subset of patients with hereditary xerocytosis, a disease characterized by hemolytic anemia associated with erythrocyte dehydration (32; 33). In addition to red blood cells (Figure 1), K_Ca3.1 is also widely expressed in cells of the immune system such as T cells (34), B cell (35), mast cells (36), macrophages (37), and microglia (38). The primary role of K_Ca3.1 in the immune cells is to hyperpolarize the cell membrane and create the driving force for the calcium entry that is necessary for activation, proliferation and cytokine production (39). Most, if not all, K_Ca3.1 expression in the brain seems to be localized to microglia, which upregulate K_Ca3.1 after activation in vitro and in vivo (40; 41). Although two studies recently reported that K_Ca3.1 may also be expressed in neurons and contribute to the slow afterhyperpolarization (42; 43), another study presented data that K_Ca3.1 does not contribute to the slow AHP (44). A potential role for K_Ca3.1 in neurons therefore currently remains uncertain. The phenotype of the K_Ca3.1^−/− mouse reinforces the channel’s role in the immune system. T cells from K_Ca3.1^−/− mice show reduced T-cell receptor mediated calcium influx and inflammatory cytokine production and the mice develop less severe colitis (45) and arthritis (46). K_Ca3.1^−/− mice further display blunted IgE-mediated anaphylactic reactions and reduced infarction and neuroinflammation after ischemic stroke (40).

K_Ca3.1 deletion in mice also reduces the EDH response, raises mean arterial blood pressure by 7–9 mm Hg (47) and causes subtle erythrocyte macrocytosis and progressive splenomegaly (48). In secretory epithelia of the lung and gastrointestinal tract K_Ca3.1 works in concert with the Na-K-2Cl cotransporter to facilitate chloride and fluid secretion (49). While K_Ca3.1 channels are not expressed in normal vascular smooth muscle cells, expression is turned on in dedifferentiated vascular smooth muscle cells where K_Ca3.1 activity promotes proliferation and migration, while K_Ca3.1 inhibition reduces atherosclerosis in mice (37) and restenosis in rats and pigs (50; 51). Likewise, K_Ca3.1 drives proliferation and migration in many cancers such as glioblastoma (52; 53), breast or prostate cancer (54; 55), which is why K_Ca3.1 blockers have been proposed to treat diseases that have a proliferative component.

CHANNEL STRUCTURE

The K_Ca channel field recently obtained some tremendous structural insights when the group of Roderick MacKinnon solved the full-length cryo-EM structures of K_Ca3.1 in the absence and presence of calcium (56). Unlike K_V1.2, K_Ca3.1 is non-domain swapped and the structure showed four CaMs per channel tetramer, with the CaM C-lobe of each CaM tightly bound to the CaM-BD of each subunit. The CaM N-lobes were only visible in the two open, Ca²⁺-bound states and poorly resolved in the closed, Ca²⁺-free structure suggesting that they are flexible in the absence of Ca²⁺. When Ca²⁺ binds to the N-lobe, it “swings over” to the S4-S5 linker of another subunit and pulls part of the S4-S5 linker, namely the S₄₅A helix downward, thus expanding the S6 helices and opening the pore (56). This structure solved the long-standing conundrum of the K_Ca channel gating symmetry. In 2001, Schumacher et al. had crystallized the K_Ca2.2 channel C-terminal CaM-BD in complex with CaM. This 1.6 Å resolution structure had shown an elongated, anti-parallel dimer of two K_Ca2.2 C-terminal fragments with CaM tightly bound with its C-lobe to two alpha helices connected by a turn from the same channel subunit (57). The CaM N-lobe was “grabbing” the free end of CaM-BD from the other subunit in the dimer suggesting that CaM-BD dimerization might gate K_Ca2 channels (57). However, the 2-fold symmetry suggested by this dimer-of-dimers was difficult to reconcile with the 4-fold symmetry of the pore (58) and the new full-length structure now makes it clear that K_Ca3.1 in fact gates with a more common 4-fold symmetry.

Interestingly, the dimeric crystal has proven very resilient in that it was repeatedly observed in subsequent crystallographic studies addressing the mechanism of action of small molecule K_Ca channel activators and of PIP₂ on K_Ca2.2 channel function (59–62). Several K_Ca channel activators were shown to bind in the interface between the CaM N-lobe and the CaM-BD in this dimeric crystal and the interaction was even confirmed by solution state NMR experiments (62). However, the full-length K_Ca3.1 structure demonstrated that the dimeric crystal is an artefact and suggested that the existing ideas about the binding site of K_Ca2 and K_Ca3.1 activators need to be revised (56). We therefore here show a Rosetta refined (63) K_Ca3.1 model, which is based on open state-1 of the K_Ca3.1 cryo-EM structure, and a K_Ca2.2 homology model to illustrate and discuss the binding sites of the commonly used pharmacological tool compounds in context of this structure.

K_Ca2 AND K_Ca3.1 CHANNEL PHARMACOLOGY

In 1982 the neurotoxic peptide apamin was shown to inhibit some Ca²⁺-activated K⁺ channels (64) and K_Ca channels were therefore typically differentiated into “apamin-sensitive” and “apamin-insensitive” in the 1980s and 1990s. Following their cloning, K_Ca channel pharmacology developed relatively rapidly and the field now has quite a range of peptidic and small-molecule inhibitors as well as positive and negative gating modulators available. Since K_Ca channel pharmacology has been reviewed by us and others a decade ago in great detail (65–67), we here concentrate on the most commonly used modulators and their mechanisms of action.

Venom peptides

The most widely used K_Ca2 channel blocker is the 18-amino acid honey bee venom peptide apamin (64), which is remarkably selective for K_Ca2 channels. Apamin is most potent on K_Ca2.2 (IC₅₀ ~200 pM) and blocks K_Ca2.1 and K_Ca2.3 with 10–50-fold lower affinity (5; 10), while it has no effect on K_Ca3.1. In vitro application of apamin to neurons or brain slices has been instrumental for demonstrating the crucial role of K_Ca2 channels in neuronal excitability (10). While low concentrations of apamin improve cognitive performance in rodents, higher concentrations induce seizures (67; 68). Apamin was initially assumed to be a simple pore blocker but was later found to inhibit K_Ca2 channels through an allosteric mechanism involving an outer pore histidine (69) and residues in the S3-S4 extracellular loop (70), a binding configuration that is recapitulated in our docking pose in the K_Ca2.2 homology model (Figure 2). The larger scorpion toxins scyllatoxin, which is also called leiurotoxin I (71; 72), and tamapin (73) have roughly the same potency as apamin and show comparable preference for K_Ca2.2. A less potent, but highly K_Ca2.2 selective blocker, is the scyllatoxin derivative Lei-Dab⁷ (74), in which one residue is replaced by the unnatural amino acid diaminobutonoic acid (Dab). While Lei-Dab⁷ thus constitutes an even more selective tool to block K_Ca2.2 channels in physiological studies (75), there currently are no natural toxins or analogs that selectively inhibit K_Ca2.1 or K_Ca2.3.

Figure 2.

Rosetta K_Ca2.2 homology model based on the K_Ca3.1 cryo-EM structure (open state 1, pdb: 6cnn). The longer N- and C-terminus of K_Ca2.2 was not modeled. For clarity only two of the four channel subunits are shown (dark gray). Calmodulin is shown in yellow. Potassium ions in the selectivity filter are colored dark purple. The bee venom apamin with its N-atoms colored dark blue and several small molecule modulators are docked where they have been shown to bind by mutagenesis: Apamin in the outer pore, NS8593 in the inner pore (dark green), CM-TMPF (sky blue) in the inner vestibule, and SKA-111 in the interface between the CaM N-lobe and the S₄₅A helix in the S4-S5 linker. The chemical structures of other K_Ca2 blockers, negative gating modulators and activators are colored according to where they have either been shown to bind by mutagenesis or are suspected to bind.

Potencies (IC₅₀s for blockers and negative gating modulators; EC₅₀s for activators): Apamin 60–400 pM, d-tubocurarine 5 μM, dequalinium 200 nM, UCL1684 200 pM, UCL1884 110 pM, BBP 400 nM, NS8593 600 nM, AP14145 1 μM, (–)-B-TMPF 31 nM for K_Ca2.1 and 1 μM for K_Ca2.2, CM-TMPF 24 nM for K_Ca2.1 and 290 nM for K_Ca2.2, SKA-111 8 μM, RA-2 ~100 nM, SKA-31 2 μM, NS309 620 nM, CyPPA 14 μM, NS13001 2 μM.

The best known peptidic blocker of K_Ca3.1 is the 37-amino acid scorpion toxin charybdotoxin, which anchors itself in the outer vestibule of K_Ca3.1 by two salt-bridges while inserting its central lysine residue into the selectivity filter (76) as shown in Figure 3. However, charybdotoxin never was an ideal K_Ca3.1 blocker because it also inhibits K_Ca1.1 (BK) and the voltage-gated Kv1.3 channels, both cross-reactivities, which initially caused confusion concerning the role of K_Ca3.1 in the cardiovascular system and in T cells. Another, somewhat more potent scorpion toxin is maurotoxin (77), which unfortunately cross-reacts to K_V1.2. While these toxins are sometimes used in vitro to obtain a complete biophysical and pharmacological signature of K_Ca3.1, they have not been used as K_Ca3.1 blockers in vivo.

Figure 3.

Rosetta refined model of the K_Ca3.1 cryo-EM structure (open state 1, pdb: 6cnn). For clarity only two of the four channel subunits are shown (dark green). Calmodulin is shown in yellow. Potassium ions in the selectivity filter are colored dark purple. The scorpion toxin charybdotoxin (ChTX) is shown docked into the outer vestibule. Various small molecule modulators are docked were they have been shown to bind by mutagenesis: Senicapoc as a representative triaryl-methane in the inner pore (blue), nifedipine in the fenestration region (pink), and SKA-111 in the interface between the CaM N-lobe and the S₄₅A helix in the S4-S5 linker. The chemical structures of other K_Ca3.1 blockers and activators are colored according to where they have either been shown to bind by mutagenesis or are suspected to bind.

Potencies (IC50s for blockers; EC50s for activators): ChTX 2–28 nM, clotrimazole 70–250 nM, TRAM-34 10–25 nM, senicapoc 11 nM, NS6180 11 nM, nifedipine 0.8–4 μM, 4-phenyl-4H-pyran 8 nM, SKA-111 150 nM, SKA-31 250 nM, NS309 10–30 nM, 1-EBIO 24–80 μM.

Small molecule K_Ca2 Channel Blockers

The key structural feature of selective K_Ca2 blockers is that they carry one or two positive centers, either permanently charged or strongly basic nitrogens, or in some cases acquire a charge by complexing divalent cations. All known K_Ca2 blockers work from the extracellular side and competitively displace radioactively labeled apamin in binding assays. The positive charges are reminiscences of the essential arginines of apamin (dark blue in the apamin structure in Figure 2). Intriguingly, this requirement for positive charges in the pharmacophore parallels classical anti-cholinergic drugs, many of which like d-tubocurarine and dequalinium (Figure 2), also block K_Ca2 channel (78; 79). Dequalinium was used as a starting point for several structure-activity-relationship studies focusing on both the nature of the permanently charged ring and the distance between the positive charges (80; 81), efforts which ultimately lead to the discovery of the high-affinity bis-quinolinium cyclophane K_Ca2 blockers UCL1684 and UCL1848 (Figure 2), which are as potent as apamin in blocking K_Ca2.2 (82; 83). Interestingly, the commonly used permanently charged derivative of the GABA_A receptor antagonist bicucculine, bicucculine methiodide, also blocks K_Ca2 channels as potently as GABA_A receptors (84).

These highly selective but permanently charged K_Ca2 blockers have been mostly used in academia, while there have only been modest activities in the pharmaceutical industry, probably due to the expected complications of permanently charged molecules, such as low permeability across biological membranes. Icagen published a patent on bis-benzimidazoles (USOO7482373B2), which block K_Ca2 channels and displace apamin. These molecules are not permanently charged, but form charged complexes with divalent cations, which probably constitute their active form. Bristol Myers Squibb published a series of 2-aminothiazoles, which also chelate divalent cations (85). Recently, the K_Ca2 channel blocking effect of 2,6-bis(2-benzimidazolyl)pyridine (BBP) (Figure 2), a molecule belonging to this class (86), has been shown to depend on H491 in the extracellular S5-P linker, the same histidine which is also a determinant for apamin binding (69). Despite these quite intriguing approaches, no K_Ca2 blockers have yet entered clinical development. However, due to their potency and good selectivity several of these molecules have been extremely valuable tools for elucidating the role of K_Ca2 channels in cardiac arrhythmias (87; 88) or the endothelium derived hyperpolarization response (89).

Small molecule K_Ca3.1 Blockers and their Preclinical and Clinical Applications

Apart from low affinity inhibitors like the β-blocker cetiedil (90), a compound that been reported to affect erythrocyte K⁺ fluxes in 1981 (90) and therefore inspired some early medicinal chemistry (67; 91), two compound classes spurred a real interest in finding therapeutically useful small molecule K_Ca3.1 inhibitors in the late 1990s: Dihydropyridines (92; 93) and triaryl-methanes (94). Both pharmacophores had previously been successfully developed as L-type Ca²⁺ channel antagonists for hypertension and as P450 inhibitors for topical antifungals. Following the cloning of K_Ca3.1, scientist at Bayer worked on dihydropyridines (95; 96), a lead optimization resulting in a series of very potent and selective phenyl-pyrans and cyclohexadienes (Figure 3) that showed in vivo efficacy in animal models of traumatic brain injury (97), but never entered clinical development for undisclosed reasons. Following up on work performed by the group of Carlo Brugnara at Harvard showing that clotrimazole reduced erythrocyte dehydration and exerted anti-sickling effects in transgenic mice and in patients with sickle cell disease (98; 99), Icagen pursued the triaryl-methanes and developed the clotrimazole derivative senicapoc. In parallel, one of us used clotrimazole as a template for the design of TRAM-34 (100). In both TRAM-34 and senicapoc (Figure 3) clotrimazole’s toxic effect on the human P450 system have been avoided by substitution of the imidazole ring with either a pyrazole ring or an amide group. While TRAM-34 was not suitable for development, it has become a widely used academic tool compound based on its selectivity and its acceptable pharmacokinetic properties when administered intraperitoneally. For example, TRAM-34 has been utilized to validate K_Ca3.1 as a potential target for vascular restenosis (50), atherosclerosis (37), asthma (101), allograft vasculopathy (102), inflammatory bowel disease (45), ischemic stroke (40), as well as renal (103; 104) and cardiac fibrosis (105). In many of these disease models the pathophysiological relevance of K_Ca3.1 was confirmed by parallel experiments in K_Ca3.1^−/− mice.

Senicapoc, which is orally available and has a long 12-day half-live in humans, entered clinical trials for sickle cell anemia but despite showing good effects on several haematological parameters, it unfortunately failed to meet the predefined primary endpoint, which was reduction in the number of painful crisis, in phase-III clinical studies (106). Based on senicapoc’s efficacy in an asthma model in sheep (107), Icagen subsequently tested senicapoc in two small Phase-II trials for asthma (108) and demonstrated encouraging results in allergic asthma. However, senicapoc did not improve lung function in exercise induced asthma. Following these failures Icagen was purchased by Pfizer and senicapoc deposited in the 2012/13 NIH National Center for Advancing Translational Research (NCAT) library as PF-05416266 making it theoretically available for investigator initiated clinical trials. Senicapoc is currently being “repurposed” by the Pfizer spin-out SpringWorks Therapeutics for the hemolytic anemia disease hereditary xerocytosis, a rare condition caused by gain of function mutations in K_Ca3.1 (109). Another “repurposing” Phase-IIa clinical trial with senicapoc will be conducted by the Alzheimer’s Disease Center at the University of California, Davis. This trial, which is anticipated to start in Fall of 2019, is based on findings that the expression of K_Ca3.1 is increased on microglia in brains from patients with Alzheimer’s disease (AD) and that K_Ca3.1 inhibition with senicapoc reduces inflammation and amyloid-β deposition in mouse models of AD (110).

More recently, several new classes of K_Ca3.1 inhibitors were reported by the pharmaceutical industry. Using a so-called “scaffold hopping” approach (e.g. tetrazole derivatives, US9556132B2) and high-throughput thallium-flux (111) NeuroSearch, a biopharmaceutical company in Denmark, identified a completely new series of benzothiazinone-based K_Ca3.1 blockers in 2013. One of the exemplary compounds, NS6180 (Figure 3), showed efficacy in an animal model of inflammatory bowel disease despite low in vivo exposure (112). Boehringer Ingelheim pursued the closely related fused thiazine-3-ones (US 2015/0232484) and Roche published a patent on 3,4-disubstituted oxazolidinones (WO 2014/067861), which constitute an interesting variation on the triaryl-methane motive (Fig. 3). Nothing specifically was reported on therapeutic indications by the two companies, but a paper co-authored by Boehringer Ingelheim scientists focused on the role of K_Ca3.1 in the process of multinucleation of macrophages and osteoclasts (113) suggesting a possible focus on chronic inflammation or bone-diseases.

Atomistic Mechanism of Action of K_Ca3.1 Blockers

Overall, the various compound classes (Figure 3), which have been reported as potent and selective K_Ca3.1 inhibitors during the last 20 years, are quite remarkable for their chemical diversity. However, on closer inspection a unifying characteristic is the absence of acidic or basic moieties and the presence of two or usually three substituted aryl groups making most of these compounds quite “greasy”, insoluble and prone to suboptimal pharmaceutical properties, such as high plasma protein binding. Remarkably, the majority of the K_Ca3.1 blockers shown in Figure 3 are binding to the same site in the inner pore of K_Ca3.1, just below the selectivity filter. While this canonical site is often touted as not suitable for obtaining subtype selective inhibitors, this seems to be possible for K_Ca3.1, because it is “alone” in its family. Indeed, none of the currently known K_Ca3.1 blockers, cross-react to K_Ca2 channels and the compounds typically also exhibit between 200–1000-fold selectivity over other ion channels.

The triaryl-methane type K_Ca3.1 blockers clotrimzole, TRAM-34 and senicapoc interact with threonine 250 in the pore loop and valine 275 in S6 as demonstrated by the fact that mutations of these residues completely abolish the sensitivity of K_Ca3.1 to triaryl-methanes (114). Based on a study using the Rosetta molecular modeling suite (115), TRAM-34 anchors itself through hydrophobic interactions with the V275 residues from all four subunits and forms a hydrogen bond to the T250 side chain from one subunit with its pyrazole nitrogen and thus blocks ion conduction by filling the site that would normally be occupied by a K⁺ ion before it enters the selectivity filter. Senicapoc is assuming a similar binding pose (Figure 3) but instead of acting as a hydrogen-bond acceptor like TRAM-34, its amide group functions as a hydrogen-bond donor and interacts with T250 side chains from two subunits (115). Interestingly, the same two mutations that “knock off” triaryl-methane binding, also drastically reduce the affinity of the benzothiazinone NS6180 (112). However, in contrast to TRAM-34 and senicapoc, which are positioned directly under the selectivity filter and interact with all four subunits, NS6180 interacts with the T250 and V275 side chains from only two adjacent subunits (115). However, although NS6180 is “sitting” differently is still overlaps with the pore lumen potassium site (116) and thus seems to act by preventing ion permeation.

In contrast to the triaryl-methanes, the binding site of the dihydropyridine nifedipine has been localized to the fenestration region of K_Ca3.1 (Figure 3), where it binds between the side-chains of T212 in S5 and V272 in S6 from adjacent subunits and has been suggested to stabilize the channel in a non-conducting conformation without directly occluding the pore (115). While this fenestration binding site constitutes a very attractive alternative to the pore site for future design efforts directed towards the identification of K_Ca3.1 inhibitors with improved pharmaceutical properties, a completely unexpected and somewhat “shocking” observation from our group was that the nifedipine isosteric 4-phenyl-pyran (Figure 3), which had been initially described by Bayer (95) and which we resynthesized (115), is binding in the inner pore at the triaryl-methane site and not in the fenestration like its template nifedipine. As explained in detail elsewhere, this finding is consistent with the published structure-activity-relationship of the phenyl-pyrans (95) and the related carba-analogous cyclohexadienes (96). Medicinal chemists generally assume when making isosteric replacements to improve potency and selectivity that the template and the derivatives bind to the same site, which is clearly not the case here and a caution against making assumptions that are not experimentally tested.

Negative Gating Modulators

Negative gating modulation as applied to K_Ca2 channels means an inhibitor that shifts the calcium-activation curve towards higher Ca²⁺ concentrations (in contrast to the left-shifting by positive gating modulators), thereby reducing the apparent Ca²⁺-sensitivity of the channel. The first molecule in this functional class was the aminobenzimidazolone NS8593 (117) and its analogues (118) which show high selectivity for K_Ca2 channels compared to K_Ca3.1. In contrast to most K_Ca2 blockers, this class of molecules is uncharged at physiological pH and therefore more likely to pass biological barriers. NS8593 does not displace radiolabeled apamin and its activity is not reduced by mutations of the extracellular amino acid residues mediating sensitivity to apamin and the small molecule K_Ca2 blockers (see above). Instead, the effect of NS8593 was shown through site directed mutagenesis to depend on the same amino acid positions in the pore region of K_Ca2 (Figure 2) that mediate sensitivity of K_Ca3.1 to TRAM-34 (119) and introduction of just two mutations into K_Ca3.1 could render this normally insensitive channel highly sensitive to NS8593 (119). A closer inspection of the binding pose, however, suggests, that unlike TRAM-34, NS8593 is not completely obstructing the permeation pathway. The fact that gating modulation is possible at this position was hypothesized to be a pharmacological reflection of the previously suggested deep pore gating in K_Ca2 channels (120). The basic characteristics of negative modulators including their pore binding site have recently been confirmed by using the drug candidate AP14145 (Figure 2) from Acesion Pharma, which belongs to the same general class of molecules as NS8593 (121). Interestingly, scientists from Bristol Myers Squibb published another series of molecules, 4-(aminomethylaryl)-pyrrazolopyrimidines, which inhibit K_Ca2 mediated Tl⁺ -fluxes at a site not involving the apamin site (122). Although, not rigorously shown in the paper, these compounds may also act via a negative gating modulatory mechanism. While the compounds described so far do not differentiate among the three K_Ca2 subtypes, subtype selective negative gating modulation was demonstrated for the triazolopyrimidine (−)-B-TPMF (Figure 2), which preferentially inhibited K_Ca2.1 over the other K_Ca2 members and K_Ca3.1, by interaction with Ser293 in K_Ca2.1, a position not previously identified for modulators of K_Ca2 (123). Notably, as described later, this site on K_Ca2.1 can also give rise to positive modulation. Negative gating modulation has recently also been shown to account for the effects of certain dibenzoates, such as RA2 (1,3-phenylenebis(methylene)bis(3-fluoro-4-hydroxybenzoate), which inhibit both K_Ca2.x and K_Ca3.1 channels with similar potency (124), demonstrating that this mode of action is also possible for K_Ca3.1. Although its binding site has never been mapped, the inner vestibule of both K_Ca3.1 and the K_Ca2 channels is large enough to accommodate the molecule.

Based on proof-of-concept animal studies demonstrating that NS8593 or AP14145 can terminate atrial fibrillation in rats (125; 126) or even large animals such as pigs (127), negative gating modulators have successfully progressed into clinical development for atrial arrhythmia. According to a press release from Acesion Pharma, a putative analogue of NS8593/AP14145 called AP30663 has recently passed Phase-I clinical trials in human volunteers.

Positive Gating Modulators

Most K_Ca2 and K_Ca3.1 activators are “clean” positive gating modulators meaning that they shift the calcium-activation curve concentration-dependently towards lower intracellular Ca²⁺-concentrations, thereby increasing the apparent Ca²⁺ affinity, but are unable to activate the channels at “0” intracellular Ca²⁺. However, there seem to be exceptions to this simple mechanism in that the existence to “true” activators (123) and potentially superagonists (128) has been suggested. The prototype activators of K_Ca3.1 and K_Ca2 channels are the benzimidazolone 1-EBIO (129) and its more potent derivative dichloro-EBIO (130), which have both played a significant role as pharmacological ex vivo tool compounds in brain slices, endothelia and epithelia, or smooth muscle preparations. However, several drugs that have been on the market for decades, such as the muscle relaxant chlorzoxazone and the ALS drug riluzole are also quite effective K_Ca activators (67; 131), which may well be their major therapeutic action. Dedicated search for more potent and selective activators led to NS309 (132), one of the most potent “pan”- K_Ca3.1/K_Ca2 activators and an important mechanistic tool compound, but not suited for in vivo studies, due to poor pharmacokinetic properties. With the aim of making more selective and potent riluzole-like compounds that could potentially be used in vivo, Sankaranarayanan et al. (133) identified a series of benzothiazoles including SKA-31 and SKA-121, which has improved selectivity for K_Ca3.1 and which has been used to demonstrate that selective K_Ca3.1 activation can lower blood pressure in mice (134).

Also pursuing subtype selectivity, NeuroSearch scientists discovered cyclohexyl-[2-(3,5-dimethyl-pyrazol-1-yl)-6-methyl-pyrimidin-4-yl]-amine (CyPPA) and the more potent analogue (4-chloro-phenyl)-[2-(3,5-dimethyl-pyrazol-1-yl)-9-methyl-9H-purin-6-yl]-amine) (NS13001), which activates K_Ca2.3 and K_Ca2.2 while being inactive on K_Ca2.1 and K_Ca3.1 (75; 135). A different selectivity profile was exemplified by CM-TPMF, which preferentially activates K_Ca2.1 (123) and like the negative gating modulator (−)-B-TPMF depends on a serine residue in S5 for its activity (Figure 2). These subtype selective compounds formed the basis for a collaboration between Saniona, a company continuing the research assets of NeuroSearch, and Ataxion (now Cadent Therapeutics), which finally resulted in selection of CAD-1883 for preclinical development for cerebellar dysfunction. According to the Cadent Therapeutics homepage, CAD-1883 is now in phase-II for essential tremor and for spinocerebellar ataxia. The idea of using K_Ca2 activators for these indications is based on the observations that genetic silencing of K_Ca2 channels in deep cerebellar neurons induces ataxia in mice (136), while treatment of mice with spinocerebellar ataxia type-2 (SCA2) with the K_Ca2 activator NS13001 alleviates motor symptoms and prevents neurodegeneration of Purkinje cells (75).

Atomistic Mechanism of K_Ca Channel Positive Gating Modulators

Mutational studies performed by Pedarzani et al. suggested in 2001 that the binding site of the benzimodazolone-type K_Ca activators is probably located in the C-terminus, close to or within the CaM-BD, since “swapping” the C-terminus of K_Ca3.1 into K_Ca2.2 made this channel as 1-EBIO sensitive as K_Ca3.1 (137). So, when 1-EBIO, NS309 and riluzole were later reported to bind at the interface between the CaM N-lobe and the K_Ca2.2 CaM-BD after being soaked into the above described C-terminal dimeric crystal using x-ray crystallography or in solution state NMR (59–62), this interface was widely assumed to be the binding site of this type of K_Ca activators in both K_Ca2 and K_Ca3.1 channels (62; 138). However, the recently published full-length cryo-EM structure of K_Ca3.1 demonstrated that the analogous segment of the K_Ca3.1 C-terminus, which is designated the C-terminal HC helix, actually forms a coiled coil at the center of the channel (56) and therefore is unlikely to constitute the binding site. In their study the MacKinnon group also proposed that the “real” binding pocket of 1-EBIO is located in the interface between the S₄₅A helix and the CaM N-lobe, in which 1-EBIO is hypothesized to contact L185 in the S₄₅A linker (56) instead of L480 in the C-terminal crystal complex (59), but did not experimentally test this very plausible alternative binding site hypothesis. Our own group recently picked up on this postulate and confirmed through mutagenesis that at least the SKA-type K_Ca channel activators as exemplified by SKA-111 (5-methylnaphtho[1,2-d]thiazol-2-amine) are binding in the interface between the CaM N-lobe and the S₄₅A helix (139). In this interface pocket Rosetta modeling shows that SKA-111 makes van der Waals contacts with S181 and L185 in the S₄₅A helix of K_Ca3.1 but interacts with the same CaM N-lobe residues (M51, E54, and M71) that were previously shown to be involved in binding of the aminothiazole riluzole in the C-terminal crystal dimer (62).

We here show SKA-111 docked into the S₄₅A helix/CaM N-lobe interface of both K_Ca2.2 (Figure 2) and K_Ca3.1 (Figure 3), fully recognizing that the K_Ca2.2 binding pose is currently not supported by experimental data. Based on the high sequence similarity in the S₄₅A helix between K_Ca3.1 and the three K_Ca2 channels, we would expect the CaM mediated gating and the putative stabilization of the interaction between the CaM N-lobe and the S₄₅A helix by benzothiazole-type K_Ca activators to be similar, even if there are some sequence differences between K_Ca3.1 and K_Ca2 channels. However, in addition to the S₄₅A helix/CaM N-lobe interface, which is present four times in the K_Ca3.1 channel, there are certainly more sites on the cytoplasmic surface of K_Ca channels that could accommodate small molecules and it is feasible that NS309 or the K_Ca2.2/2.3 selective CyPPA and NS13001 are binding at other sites or occupy a different number of sites. Gating modulation is also possible in the transmembrane domain as has been demonstrated by the fact, that a serine residue in S5 is crucial for the action of both the K_Ca2.1 selective positive gating modulator CM-TMPF and the negative modulator (−)-B-TPMF (123).

OUTLOOK

As described here, K_Ca2/3 channels have a relatively well-developed pharmacology and their therapeutic targeting for neurological and cardiovascular diseases is supported by ample preclinical data. The most advanced compound, the K_Ca3.1 blocker senicapoc unfortunately failed in a phase-III clinical trial in sickle cell anemia (106). However, the trial certainly demonstrated that K_Ca3.1 inhibition is safe in humans, and, as described above, senicapoc is currently in the process of being repurposed for the treatment of hereditary xerocytosis and Alzheimer’s disease. For Alzheimer’s disease the therapeutic hypothesis is that K_Ca3.1 inhibition would reduce neuroinflammation by suppressing microglia activation (140). The same hypothesis is used to rationalize repurposing senicapoc for stroke (141) and neuropathic pain (142). Other indications for which repurposing of senicapoc might be worthwhile considering is idiopathic pulmonary fibrosis (143) and glioblastoma (144). While repurposing is an attractive short cut, senicapoc due to its high lipophilicity, low solubility, high plasma protein binding and very long half-live in humans, is not necessarily an “ideal” K_Ca3.1 blocker and it will be interesting to see if the now available full-length K_Ca3.1 structure (56) will revive the interest of the pharmaceutical industry in developing better K_Ca3.1 inhibitors.

The other K_Ca modulators that have recently entered clinical trials, the yet undisclosed negative K_Ca2 channel gating modulator AP30663 for atrial fibrillation and the positive K_Ca2 channel modulator CAD-1883 for cerebellar disorders, have passed phase-I and are currently being tested in patients, which demonstrates the feasibility of balancing benefits and side effects of K_Ca2 channel modulation by optimizing compound properties. AP30663 is intentionally made peripherally restricted presumably by increasing its polarity or by introducing structural elements favoring its extrusion across the blood brain barrier, thereby strongly reducing liability for inducing tremors and seizures that are observed with the brain-penetrant NS8593 in animal studies (121). In contrast, CAD-1883, which is designed for targeting K_Ca2 channels in pacemaker neurons of the cerebellar cortex and deep cerebellar nuclei (primarily K_Ca2.2), is optimized for selectivity and good brain exposure. Similarly, for positive gating modulators targeting peripheral diseases, it will be important to achieve subtype and ideally tissue selectivity. Unselective activators like SKA-31 are useful tool compounds but quickly demonstrated that central K_Ca2 channel mediated sedation and heart rate reduction (145) constitute undesirable side-effects when attempting to target endothelial K_Ca3.1 channels to lower blood pressure, even if the approach is effective in large animals such as dogs (146) and pigs (147). A useful K_Ca3.1 activator for improving endothelial function in hypertension and other cardiovascular disease should therefore ideally be K_Ca3.1 selective and peripherally restricted.

The very significant advances that have recently been made in elucidation the structure of K_Ca3.1 (56) and the resulting improvements in modeling have so far not been used for drug design. All current drug candidates in the K_Ca channel field have been identified by screening or classical medicinal chemistry approaches. However, as more high-resolution protein structures will become available, ideally with K_Ca channel modulators differing in structure, mode of action, and selectivity positioned at their respective pharmacological sites, there is no doubt that this information will be of increasing importance in future drug optimization programs. Immediate questions to solve are, for example, how to explain the already obtained K_Ca2.3/K_Ca2.2 versus K_Ca2.1/K_Ca3.1 selectivity of compounds in the CyPPA/NS13001 series at the atomistic level. Another item on the “wish-list” of the pharmacologist and the drug developer is to gain insights into how to design selective blockers or negative gating modulators for K_Ca2.3, an important channel in all monoaminergic neurons, which could be an important step in developing new drugs for psychiatric diseases. While we believe that K_Ca2.2 or K_Ca2.2/K_Ca2.3 selective activators certainly are also promising for the treatment of dependence on alcohol (148) and other habit forming substances (149), it might be challenging to ever safely translate the beneficial effects of K_Ca2.2 inhibition on learning and memory into the clinic.