Design and structural basis of selective 1,4-dihydropyridine inhibitors of the calcium-activated potassium channel K_Ca3.1

Significance

An approach to turn old drugs into new drugs is “selective optimization of a side activity (SOSA).” Here, we modified the 1,4-dihydropyridine nucleus and transformed its side activity (weak blockade of the calcium-activated potassium channel K_Ca 3.1) into the main activity while diminishing the original main activity (blockade of L-type voltage-gated calcium channels). The new 1,4-dihydropyridines block K_Ca 3.1 channels with single-digit nanomolar potency and excellent selectivity, binding in a water-filled cavity in K_Ca 3.1’s pore to impede ion permeation. In proof-of-concept studies, the 1,4-dihydropyridines blocked mutant K_Ca 3.1 channels that cause hereditary xerocytosis and effectively treated ischemic stroke in a rat model. K_Ca 3.1-selective 1,4-dihydropyridines have considerable potential as therapeutics.

Abstract

The 1,4-dihydropyridines, drugs with well-established bioavailability and toxicity profiles, have proven efficacy in treating human hypertension, peripheral vascular disorders, and coronary artery disease. Every 1,4-dihydropyridine in clinical use blocks L-type voltage-gated calcium channels. We now report our development, using selective optimization of a side activity (SOSA), of a class of 1,4-dihydropyridines that selectively and potently inhibit the intermediate-conductance calcium-activated K⁺ channel K_Ca3.1, a validated therapeutic target for diseases affecting many organ systems. One of these 1,4-dihydropyridines, DHP-103, blocked K_Ca3.1 with an IC₅₀ of 6 nM and exhibited exquisite selectivity over calcium channels and a panel of >100 additional molecular targets. Using high-resolution structure determination by cryogenic electron microscopy together with mutagenesis and electrophysiology, we delineated the drug binding pocket for DHP-103 within the water-filled central cavity of the K_Ca3.1 channel pore, where bound drug directly impedes ion permeation. DHP-103 inhibited gain-of-function mutant K_Ca3.1 channels that cause hereditary xerocytosis, suggesting its potential use as a therapeutic for this hemolytic anemia. In a rat model of acute ischemic stroke, the second leading cause of death worldwide, DHP-103 administered 12 h postischemic insult in proof-of-concept studies reduced infarct volume, improved balance beam performance (measure of proprioception) and decreased numbers of activated microglia in infarcted areas. K_Ca3.1-selective 1,4-dihydropyridines hold promise for the many diseases for which K_Ca3.1 has been experimentally confirmed as a therapeutic target.

1,4-dihydropyridines were first introduced as therapeutics in 1975 and now rank among the most prescribed medicines in the world (1, 2). They are widely used to treat human hypertension, peripheral vascular disorders, and coronary artery disease (1, 2). Every 1,4-dihydropyridine approved for therapeutic use in the past five decades blocks L-type voltage-gated calcium channels (Ca_V1.x subfamily) (1, 2). High-resolution structural studies using cryogenic electron microscopy (cryo-EM) reveal that 1,4-dihydropyridines bind in a lipid bilayer-accessible hydrophobic fenestration between repeats III and IV of mammalian Ca_V1.x channels (3, 4). The 1,4-dihydropyridine nucleus has been decorated by diverse substituents to generate analogues against non-Cav1.x targets including receptors (e.g., adenosine receptors) (5) and transporters (e.g., multidrug resistance protein 4) (6). However, these 1,4-dihydropyridines exhibit significantly lower potency and/or selectivity for their specific targets than do 1,4-dihydropyridine Ca_V1.x inhibitors. Here, we describe a class of 1,4-dihydropyridines that block the intermediate-conductance Ca²⁺-activated K⁺ channel K_Ca3.1 (KCNN4, IK_Ca1, SK4, Gardos channel) (7–11) with single-digit nanomolar potency and remarkable selectivity.

The K_Ca3.1 channel functions as a cation counterbalancer, sustaining calcium signaling in B and T lymphocytes, proliferating myofibroblasts, microglia, macrophages, glioblastoma cells, and other cell types through K⁺ efflux-driven hyperpolarization of the plasma membrane (12). The channel is also the primary conduit for conductive K⁺ efflux in red blood cells (10, 11). Despite K_Ca3.1’s broad tissue distribution, the K_Ca3.1-specific triarylmethane inhibitor, Senicapoc, was well tolerated in human trials, suggesting that selective blockade of K_Ca3.1 is safe in vivo (13). Moreover, knockout of the gene encoding K_Ca3.1 in mice does not cause a grossly apparent deleterious phenotype, further supporting the safety in vivo of specific K_Ca3.1 blockade (12). K_Ca3.1 has been experimentally verified in animal models as a therapeutic target for varied diseases including neuroinflammatory diseases (acute ischemic stroke, Alzheimer’s disease, perioperative neurocognitive disorders, spinal cord injury, traumatic brain injury), fibrosis (lung, liver, kidney, heart), inflammatory bowel disease, atherosclerosis, vascular restenosis, hemolytic anemias (hereditary xerocytosis, sickle cell disease), allergic airway disease, secretory diarrheas, and glioblastoma (12, 14). K_Ca3.1-selective 1,4-dihydropyridines may therefore have wide therapeutic use. One of the 1,4-dihydropyridines, DHP-103, blocked K_Ca3.1 potently (IC₅₀ 6 nM) and with >500-fold selectivity over Ca_V1.x channels and >100 additional molecular targets. High-resolution cryo-EM structural studies combined with mutagenesis and electrophysiology revealed that DHP-103 binds in the water-filled central cavity beneath K_Ca3.1’s selectivity filter to block K⁺ permeation through the channel pore. In proof-of-concept efficacy studies, DHP-103 inhibited gain-of-function mutant K_Ca3.1 channels that cause hereditary xerocytosis, a hemolytic anemia, and showed efficacy in a rat model of acute ischemic-reperfusion stroke.

Results and Discussion

Identification of K_Ca3.1 Channel-Selective 1,4-Dihydropyridine Inhibitors.

More than two decades ago, the 1,4-dihydropyridines nifedipine, nitrendipine, and nimodipine were reported to block the recombinant human K_Ca3.1 channel, albeit at >200-fold lower potency than for blockade of Ca_V1.x channels (15). A few years earlier, a single nitrendipine derivative (compound 29) was shown to inhibit the human erythrocyte K_Ca3.1 channel more potently than the L-type voltage-gated calcium channel of guinea pig ileal longitudinal muscle (16). These studies suggested that K_Ca3.1-selective 1,4-dihydropyridines could be developed. Old drugs can be turned into new drugs through “selective optimization of a side activity (SOSA).” We have now used a SOSA approach (17) to modify the 1,4-dihydropyridine scaffold so as to transform its side activity (weak blockade of K_Ca3.1) into its main activity (potent blockade of K_Ca3.1) while reducing the original main activity (blockade of L-type voltage-gated calcium channels).

We started with commercially available 1,4-dihydropyridines known to inhibit Ca_V1.x channels, some used clinically as drugs, and designed and synthesized a series of 1,4-dihydropyridines. Using patch clamp recordings, we identified 1,4-dihydropyridines that block K_Ca3.1 channels with single-digit nanomolar potency. Fig. 1 A–C and G shows the effects of exemplar 1,4-dihydropyridines on human K_Ca3.1-mediated K⁺ currents in transfected HEK293 cells elicited by voltage ramps (−120 mV to 40 mV from holding potential −80 mV) with 1 µM free Ca²⁺ in the pipette solution. SI Appendix, Fig. S1 A and B lists the analyzed 1,4-dihydropyridines, their moieties at positions R1-R8, and their IC₅₀ values for block of K_Ca3.1. The 1,4-dihydropyridines cluster into four groups based on IC₅₀ values: group 1 (>5 µM), group 2 (1-5 µM), group 3 (<200 nM), and group 4 (<30 nM) (SI Appendix, Fig. S1B). We attribute the order-of-magnitude improvement in the potency of group 3 vs. group 1 to the presence of methyl acetate or propan-2-one at R5 and R6 (SI Appendix, Fig. S1B). Further increases in potency of group 4 1,4-dihydropyridines reflect removal of NO₂ (from either R3 or R4) or a halogen (from R4), and introduction of a halogen (F or Cl) at R1 and CF₃ at R3 (Fig. 1A and SI Appendix, Fig. S1B).

Fig. 1.

Novel 1,4-dihydropyridines block K_Ca3.1 channels significantly more potently than L-type voltage-gated Ca_V1.2 calcium channels. Blockade by three exemplar 1,4-dihydropyridines from groups 4a (DHP-101), 4b (DHP-104), and 4c (DHP-103) on human K_Ca3.1 currents (A–C) and human Ca_V1.2 currents (D–F). For Ca_V1.2 current measurements, we used 0.5 µM nifedipine at the end of each experiment to determine complete block. (G) Concentration–response curves of DHP-101, DHP-104, DHP-103 on K_Ca3.1 and Ca_V1.2 currents are compared (n ≥ 3, mean ± SD).

Since 1,4-dihydropyridines are potent inhibitors of L-type voltage-gated Ca_V1.x channels, we compared block of Ca_V1.2 and K_Ca3.1 currents by these compounds (SI Appendix, Fig. S1C). The group 2 1,4-dihydropyridine DHP-07 blocked Ca_V1.2 channels far more potently than the K_Ca3.1 channel, while 1,4-dihydropyridines from groups 3, 4a, 4b, and 4c inhibited K_Ca3.1 significantly more potently than Ca_V1.2 (Fig. 1 D–G and SI Appendix, Fig. S1C). Group 4b and 4c 1,4-dihydropyridines showed 600- to 2,000-fold selectivity for K_Ca3.1 over Ca_V1.2 (Fig. 1 E–G and SI Appendix, Fig. S1C). To further evaluate selectivity, we examined the ability of group 4c compound DHP-103 (SI Appendix, Fig. S1 A and B) to inhibit [³H]nitrendipine binding to the 1,4-dihydropyridine binding site on Ca_V1.x subfamily channels expressed in the rat cerebral cortex. At 3 µM, DHP-103 caused a ~45% inhibition of [³H]nitrendipine binding (SI Appendix, Fig. S2), which is roughly 500-fold higher than its IC₅₀ (6 nM) for K_Ca3.1 channel block. These data derived from electrophysiological and radionuclide-labeled antagonist binding assays demonstrate that groups 4b and 4c 1,4-dihydropyridines exhibit significant selectivity for K_Ca3.1 channels over Ca_V1.x channels. While both H and CH₃ are tolerated at positions R7 and R8 for potent inhibition of K_Ca3.1, H at both R7 and R8 (group 4c) is preferred for K_Ca3.1-selectivity over Ca_V1.2 (Fig. 1G and SI Appendix, Fig. S1 A–C).

We next measured the ability of group 4c compound, DHP-103, to block other ion channels. DHP-103 showed ≥ 1,000-fold selectivity for K_Ca3.1 over a panel of 13 ion channels, including closely related calcium-activated K⁺ channels (K_Ca1.1, K_Ca2.2, K_Ca2.3), and channels critical for normal cardiac function (K_V11.1/hERG, K_V1.4, K_V1.5, K_V4.2, Na_V1.5) (SI Appendix, Table S1). As further assessment of selectivity, we tested DHP-103 on a panel of 97 molecular targets including channels, transporters, receptors, and kinases (Eurofins pharmacological P9 diversity panel safety screen), and on 5 cytochrome P450 enzymes. DHP-103 (3 µM) had minimal effect on 94 targets, while it caused 40 to 60% inhibition of three targets (progesterone receptor, glucocorticoid receptor, and the DHP-binding site on Ca_V1.x subfamily channels) (SI Appendix, Fig. S2). Since the concentration of DHP-103 tested (3 µM) is 500-fold higher than its IC₅₀ (6 nM) for K_Ca3.1 blockade (Fig. 1G and SI Appendix, Fig. S1), we conclude that DHP-103 exhibits significant selectivity for K_Ca3.1 over all 97 molecular targets. DHP-103 also exhibited >800-fold selectivity for K_Ca3.1 over the five tested cytochrome P450 enzymes (SI Appendix, Table S2). These studies demonstrate that DHP-103 blocks K_Ca3.1 with excellent potency and exquisite selectivity over 116 other therapeutic targets. We attribute DHP-103’s single-digit nanomolar potency and remarkable K_Ca3.1-selectivity to large lipophilic groups (F, CF₃) at R1 and R3, ketoesters at R5 and R6, and H at R4, R7, and R8 (Fig. 1 and SI Appendix, Fig. S1).

Ketoesters such as those occupying positions R5 and R6 are generally considered to negatively impact drug stability. We therefore compared the metabolic stability of group 4c 1,4-dihydropyridines in mouse and human liver microsomes to that of verapamil, a calcium channel blocking drug used to treat human hypertension, cardiac arrhythmias, and angina. DHP-103, DHP-110, and DHP-113 exhibited significantly greater metabolic stability than verapamil based on measured t_1/2 and Cl’_int (SI Appendix, Table S3). In vitro biotransformation of DHP-103 in liver microsomes from the mouse, rat, dog, and monkey (1-h duration) resulted in one major metabolite and ten minor metabolites (SI Appendix, Fig. S3 A and B). Furthermore, DHP-103’s pharmacokinetic properties in mice (see later) are similar to Felodipine’s (18), a Ca_V1.x-channel-blocking 1,4-dihydropyridine with ketoesters at R5 and R6 (like DHP-103), which is used to treat human hypertension and stroke.

In summary, we successfully transformed a canonical blocker of L-type voltage-gated calcium channels (Ca_V1.x) into a class of potent and selective inhibitors targeting the intermediate-conductance calcium-activated K_Ca3.1 channel.

DHP-103 Binds in a Water-Filled Cavity in the K_Ca3.1 Pore.

A recent high-resolution cryo-EM structure defined the mechanism of Ca²⁺-calmodulin-dependent activation of K_Ca3.1 (19). However, no high-resolution K_Ca3.1 structure in complex with an inhibitor or activator has been reported. We used cryo-EM structure-determination to identify the DHP-103 binding site in full-length human K_Ca3.1 complexed with calmodulin (CaM) in the presence of 2 mM Ca²⁺. Through cryo-EM single particle reconstruction we obtained a cryo-EM map of the Ca²⁺- and DHP-103-bound K_Ca3.1/CaM complex (henceforth referred to as 103_K_Ca3.1) at 3.45 Å resolution with C4 symmetry and 3.84 Å resolution with C1 symmetry, respectively (SI Appendix, Fig. S4 and Table S4). Our drug-bound 103_K_Ca3.1 structure shows better alignment with the Ca²⁺-bound apo_K_Ca3.1/CaM activation state I (apo_K_Ca3.1_I; Protein Data Bank [PDB]: 6cnn; RMSD: 1.335 Å) than apo_K_Ca3.1 activation state II (apo_K_Ca3.1_II; PDB: 6cno; RMSD: 2.21 Å) (19). The superimposition of the drug-bound 103_K_Ca3.1 and apo_K_Ca3.1_I structures is shown in Fig. 2A. DHP-103 is not displayed in this superimposition, but is shown in the channel pore structure of 103_K_Ca3.1 in Fig. 2B. Four K⁺ (K1 to K4) occupy the selectivity filter in the three structures with an additional K⁺ (K5) in the central cavity of the drug-bound 103_K_Ca3.1 structure (Fig. 2A). The closely spaced ions within the selectivity filter are likely a result of integrated electron densities from multiple K⁺-bound channels, since molecular dynamics simulation of individual channels suggest that water molecules may occupy positions between two K⁺ ions.

Fig. 2.

Binding site for 1,4-dihydropyridine in K_Ca3.1. (A) Comparison of our drug-bound 103_K_Ca3.1 structure (K_Ca3.1, gold; CaM, pink) with apo_K_Ca3.1_1 (pdb:6cnn) (K_Ca3.1, cyan; CaM, gray). The additional K⁺ in the central cavity of our drug-bound 103_K_Ca3.1 structure is identified by a red arrow. (B) EM density of DHP-103 (pink) is shown in the central cavity sandwiched between the innermost K⁺ (K4, green) in the selectivity filter and the additional K⁺ (K5, green) in the central cavity of the drug-bound 103_K_Ca3.1 structure. Sidechains of Thr250, Val275, Thr278, and Val275 (yellow) are labeled. (C) Effect of DHP-103 on K_Ca3.1_wild-type, K_Ca3.1_T250S-V275A, and K_Ca3.1_T212F-V272F K⁺ currents. (D) Concentration–response curves of DHP-103 on K_Ca3.1_wild-type vs. K_Ca3.1_T250S-V275A and K_Ca3.1_T212F-V272F K⁺ currents.

DHP-103’s EM density was seen in maps refined with both C4 and C1 symmetry (Fig. 2B and SI Appendix, Fig. S4) in the water-filled central cavity below the selectivity filter, sandwiched between the innermost K⁺ in the selectivity filter (K4) and the additional K⁺ in the central cavity (K5). DHP-103’s interactions with these two K⁺ ions seem a key determinate of its central cavity binding and action on the K_Ca3.1 channel. Negatively charged activator drugs have been reported to bind below the selectivity filters of K_V11.1, K_2P, and K_Ca1.1 channels and promote K⁺-binding to their central cavities (20). Our drug-bound 103_K_Ca3.1 structure shows direct interactions of permeant ions with a channel-bound inhibitor drug.

DHP-103’s EM density was seen in close proximity to and between the sidechains of Thr250 and Val275 in all four subunits of the K_Ca3.1 homotetramer (Fig. 2B). The four Thr250 sidechains contribute to the binding site for the innermost K⁺ (K4) in K_Ca3.1’s selectivity filter (19, 21). We used mutagenesis and electrophysiology to test DHP-103’s interactions with Thr250 and Val275 as predicted by the structure. Since DHP-103 is 1000-fold less effective vs. K_Ca2.2 (IC₅₀ >10 µM) and K_Ca2.3 (IC₅₀ 6 µM) (SI Appendix, Table S1), we replaced Thr250 and Val275 in K_Ca3.1 with the corresponding residues from K_Ca2.2/K_Ca2.3. DHP-103 blocked the K_Ca3.1_T250S-V275A mutant channel (IC₅₀ 2 µM) with ~333-fold lower potency than the wild-type K_Ca3.1 channel (IC₅₀ 6 nM), confirming that Thr250 and Val275 are essential for the high-affinity binding of DHP-103 to K_Ca3.1 (Fig. 2 C and D). DHP-103’s EM density is also in proximity to Thr278, while ion K5 is near Val282 (Fig. 2B). Although we have not tested the effects of mutating these two residues, it is possible that DHP-103’s interactions with residues Thr250, Val275, Thr278, and the two K⁺ ions K4 and K5, and K5’s interaction with Val282, contribute to both stabilization of DHP-103 binding within the central cavity and to blockade of ion permeation through the pore. Interestingly, earlier mutagenesis and molecular modeling studies predicted that three other potent small molecule inhibitors of K_Ca3.1, the triarylmethane TRAM-34 (22), the benzothiazinone NS6180 (23), and the 4-phenyl-pyran-3 rac-11 (24), each directly blocked K⁺ permeation by interacting with the sidechains of Thr250 and Val275 in the central cavity (21, 25). In contrast, nifedipine, a Ca_V1.x-selective 1,4-dihydropyridine with micromolar potency for K_Ca3.1, was predicted to bind in a lipid bilayer-accessible hydrophobic fenestration between adjacent subunits (21), analogous to the binding site of Ca_V.1x-selective 1,4-dihydropyridines in the L-type Ca_V1.x channel (3, 4). Interaction with Thr250 and Val275 in the central cavity and blockade of ion permeation thus appears to be a feature of small molecule inhibitors (1,4-dihydropyridines, triarylmethanes, benzothiazinones, 4-phenyl-pyran-3) that block K_Ca3.1 with low nanomolar potency.

DHP-103 may reach its binding site within the central cavity in the fully activated (apo_K_Ca3.1_II) channel either by crossing the entire membrane and then entering through the inner gate (10.2 Å diameter), or by traversing half the membrane and accessing the binding site through the lipid bilayer-accessible hydrophobic fenestration between adjacent subunits (9.3 Å diameter) (SI Appendix, Fig. S5). To distinguish between these two possibilities, we narrowed the fenestrae by replacing Thr212 and Val272 with bulky phenylalanines (21), and then measured block of the mutant channel by DHP-103. Narrowing the fenestration with the K_Ca3.1_T212F-V272F mutation caused a small (~4-fold) reduction in DHP-103’s potency (IC₅₀: 23 nM) compared to K_Ca3.1_wild-type (IC₅₀ 6 nM), whereas the K_Ca3.1_T250S-V275A pore mutation reduced DHP-103 potency (IC₅₀: 2 µM) by ~333-fold Fig. 2 C and D). This result potentially excludes the fenestration as DHP-103’s point-of-entry and suggests that the open inner gate in the fully activated state (10.2 Å wide) may be the passage through which DHP-103 reaches it binding site within the central cavity. Interestingly, the inner gate is narrower (7 Å wide) in drug-bound 103_K_Ca3.1 (SI Appendix, Fig. S5), perhaps suggesting that the inner gate narrows after DHP-103 fits into its binding site in the central cavity. These ideas need to be tested by molecular dynamics simulation and kinetic studies.

Since K_Ca3.1 is a calcium-activated K⁺ channel, we examined whether the internal calcium concentration affected DHP-103 potency. Changing the internal free calcium concentration from 0.5 to 10 µM did not affect block by 50 nM DHP-103, a concentration that causes >80% block in the presence of 1 μM Ca²⁺ (Fig. 1 C and G and SI Appendix, Fig. S6). Increasing the external K⁺ concentration from 4.5 to 140 mM did not alter the IC₅₀ for DHP-103 (compare Fig. 1 and Fig. 3). This result excludes displacement of DHP-103 from its binding site in the central cavity by the permeant ion.

Fig. 3.

DHP-103 blocks gain-of-function mutations of human K_Ca3.1 that cause hereditary xerocytosis. (A and B) Side view (A) and extracellular view (B) of apo-K_Ca3.1_1 (PDB: 6cnn) (CaM, gray; K_Ca3.1, light blue) with locations of gain-of-function mutations shown as purple spheres in one subunit. (C) Effect of DHP-103 on K_Ca3.1_wild-type and gain-of-function K_Ca3.1 mutants K_Ca3.1_S314P, K_Ca3.1_A322V, and K_Ca3.1_R352H. (D and E) Concentration–response curves (D) and IC₅₀ values of DHP-103 (E) on K_Ca3.1_wild-type channels and gain-of-function K_Ca3.1 mutant channels K_Ca3.1_S314P, K_Ca3.1_A322V, and K_Ca3.1_R352H (n ≥ 6, mean ± SD; **P < 0.001 vs. K_Ca3.1_wild-type; one-way ANOVA followed by Tukey’s test). (F and G) Block by DHP-103 (F) and TRAM-34 (G) of A23187-stimulated ⁸³Rb⁺ influx into red blood cells from a healthy human donor and HX patients with the heterozygous K_Ca3.1 gain-of-function mutations R352H (F) and V282M (F and G).

Collectively, our cryo-EM, mutagenesis, and electrophysiology data show that binding of DHP-103 in K_Ca3.1’s water-filled central cavity impedes ion permeation. We could not delineate the precise orientation of DHP-103 and the positioning of its moieties (at R1 to R8) within its K_Ca3.1 binding site, as symmetry-mismatch between the fourfold symmetrical K_Ca3.1 channel and the multiple rotational orientations of the flat DHP-103 phenyl ring prevented modeling of DHP-103 into its cryo-EM density.

DHP-103 Blocks K_Ca3.1 Mutants that Cause Hereditary Xerocytosis.

Hereditary xerocytosis (HX) (a.k.a dehydrated hereditary stomatocytosis; MIM: 194380) is a disease characterized by erythrocyte dehydration, generally well compensated hemolytic anemia, and variable iron overload, perinatal edema, and susceptibility to thrombotic events (26–35). The disease is linked to mutations of two genes, PIEZO1 (encoding the mechanosensitive nonselective cation channel, PIEZO1) and KCNN4 (encoding K_Ca3.1) (26–35). Gain-of-function mutations of either channel are thought to result in erythrocyte dehydration via excessive loss of K⁺ and water (26–35). Senicapoc, a triarylmethane inhibitor of K_Ca3.1, reduces red blood cell K⁺ loss and dehydration due to gain-of-function mutations of K_Ca3.1 or PIEZO1 (32, 33) and is being evaluated in a clinical trial in HX patients with the V282M mutation (ClinicalTrial.gov: NCT04372498). K_Ca3.1-selective 1,4-dihydropyridines might therefore be useful for treatment of HX if they block disease-causing gain-of-function mutant K_Ca3.1 channels. We tested DHP-103 on HX-causing gain-of-function mutations of K_Ca3.1 located either in the CaM-binding region of K_Ca3.1 (S314P, A322V, R352H) or at the inner gate in the ion conduction pathway (V282M) (Fig. 3 A and B). In patch-clamp experiments, DHP-103 blocked recombinant wild-type and mutant (S314P, A322V, and R352H) K_Ca3.1 channels with low nanomolar potency (IC₅₀ < 20 nM), though mutant A322V was significantly less sensitive than the wild-type K_Ca3.1 channel (Fig. 3 C–E). In calcium ionophore-initiated (A23187) ⁸³Rb⁺ (surrogate for K⁺) flux studies, we compared red blood cells from two HX patients with the heterozygous gain-of-function K_Ca3.1 mutant R352H versus healthy control red blood cells. DHP-103 inhibited Rb⁺ influx into the heterozygous mutant red blood cells as effectively as into wild-type red blood cells, with an IC₅₀ value (Fig. 3 F and G) comparable to the patch-clamp data (Fig. 3 C–E). In contrast, DHP-103 was less effective in inhibiting Rb⁺ influx into red blood cells from two HX patients with the heterozygous gain-of-function K_Ca3.1 mutant V282M than into healthy control red blood cells, whereas a control K_Ca3.1 inhibitor (triarylmethane TRAM-34) (22) was roughly equipotent on mutant and wild-type channels (Fig. 3 F and G). These results from patch-clamp and Rb⁺ flux studies suggest that DHP-103 may have use in the treatment of HX, particularly HX caused by the K_Ca3.1_R352H mutation.

DHP-103 Is Efficacious in a Rat Model of Acute Ischemia–Reperfusion Stroke.

Acute ischemic stroke is the second leading cause of death worldwide (36) and fifth-most common cause of death in the United States (37–40). Large vessel occlusion (LVO) accounts for 24 to 46% of acute ischemic stroke (~80,000 people annually in the United States) and is associated with significantly higher mortality rates than non-LVO ischemic stroke (37, 38, 40, 41). Within hours following the ischemic insult, a strong, sustained inflammatory response develops in the infarcted area and peri-infarct zone, driven primarily by activated microglia/macrophages with contributions from infiltrating T lymphocytes later in the process (37, 38). This neuroinflammation exacerbates the severity of stroke and impairs recovery (38). Endovascular thrombectomy, the standard-of-care treatment for LVO stroke (42), restores tissue-saving blood supply to the compromised area but does little to dampen the initial severity or subsequent progression of neuroinflammation (38). Since K_Ca3.1 blockers suppress ischemia-triggered microglial activation and inhibit infiltration of T cells into infarcted areas (37, 38), we assessed DHP-103’s potential as a therapeutic in a rat model of acute ischemic LVO stroke that recapitulated ischemia followed by restoration of blood flow. We initiated treatment 12 h after ischemia–reperfusion, at the peak of neuroinflammation, to dampen neuroinflammation.

Of particular relevance, DHP-103 crosses the blood–brain barrier in mice and achieves brain concentrations sufficient for K_Ca3.1 target engagement with a t_1/2 of 2.24 to 2.48 h following 25 mg/kg intraperitoneal administration (Fig. 4A and SI Appendix, Table S5A). Note, the brain C_max of 10,458 ng/mL (29 μM) corresponds to an estimated free DHP-103 brain concentration of 1.7 μM (based on mouse plasma protein binding of 94.18%; SI Appendix, Table S6), which ensures maximal, specific K_Ca3.1 blockade with minimal inhibition of brain Ca_V1.x channels. Moreover, both mice and rats show similar pharmacokinetic profiles for DHP-103 following intravenous (2 or 5 mg/kg) or oral administration (50 mg/kg), making it likely that DHP-103 enters the brain in rats and reaches concentrations adequate for substantial blockade of K_Ca3.1 channel activity as it does in mice (Fig. 4B and SI Appendix, Table S5B). After intravenous or oral administration, the plasma C_max ranges from 2 to 8 μM in both mice and rats, with estimated free DHP-103 plasma concentrations of 0.06 to 0.48 μM (based on mouse and human plasma protein binding of 94.18% and 97.76%, respectively; SI Appendix, Table S6) and a respective t_1/2 of 1.15 to 5.15 h (SI Appendix, Table S5B). Furthermore, DHP-103 was well tolerated in a 2-wk repeat-dose toxicology study in mice (25 and 50 mg/kg twice daily, oral administration) where it caused no significant changes in body weight, food consumption, clinical assessment, clinical chemistry, hematological parameters, organ weights, gross organ pathology, or organ histopathology at plasma concentrations sufficient to cause significant blockade of K_Ca3.1 channel activity (SI Appendix, Tables S5B and S7–S12 and Fig. S7). Given the similarity in pharmacokinetic properties of DHP-103 in both mice and rats, and therefore in vivo drug exposure, it is likely that rats tolerate DHP-103 similarly to mice.

Fig. 4.

DHP-103 shows efficacy in a rat model of large vessel occlusion [middle cerebral artery occlusion (MCAO)] acute ischemic-reperfusion stroke. (A) Concentration-time profile of DHP-103 in the plasma (red) and brain (black) of mice following a single intraperitoneal administration of DHP-103 (25 mg/kg; n = 2, mean ± SD) highlighting its brain penetration. Key parameters are shown in SI Appendix, Table S5A. (B) Plasma concentration-time profile of DHP-103 in mice (red) and rats (black) following a single intravenous injection (open circles) or oral administration (filled circles) (n = 3 for each, mean ± SD). Key parameters are shown in SI Appendix, Table S5B. (C) Mean % infarct volume in respective groups (n = 24 for each). Horizontal bars show statistically significant differences between DHP-103- or edaravone-treated MCAO ischemia–reperfusion rats vs. control MCAO ischemia–reperfusion rats (without treatment or administered vehicle) using one-way ANOVA (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). See SI Appendix, Fig. S6A for TTC-stained brain sections. (D) Balance beam test (proprioception) scores for each group on day 0_1 h (n = 8), day 1 (n = 7), day 2 (n = 8), and day 7 (n = 5) after ischemia–reperfusion (mean ± SD; two-way ANOVA of MCAO ischemia–reperfusion rats treated with DHP-103 or edaravone compared to MCAO ischemia–reperfusion control rats that were administered vehicle; *P < 0.1, **P < 0.01, ***P < 0.001, ****P < 0.0001). See SI Appendix, Fig S5 for details about the balance beam test. (E) % of Iba1⁺CD11b⁺ activated microglia in the hippocampus (CA1), caudate-putamen (CPu), primary motor cortex (M1), and somatosensory cortex (S1) in each of three experimental groups: MCAO ischemia–reperfusion without treatment, MCAO ischemia–reperfusion administered vehicle and MCAO ischemia–reperfusion administered DHP-103 (10 mg/kg); (n = 5, mean ± SD; one-way ANOVA; *P < 0.1, **P < 0.01). See SI Appendix, Fig. S6B for representative immunofluorescence micrographs.

We occluded the middle cerebral artery (MCAO) with a thread-plug for 60 min (43), then allowed reperfusion for 7 d (SI Appendix, Fig. S8 A and B). We used seven groups in our study, each group consisting of 29 male rats. Control groups included sham-operated rats (blunt dissection, exposure of right common carotid artery, then suture-closure), and rats subjected to MCAO ischemia followed by restoration of blood flow (MCAO ischemia–reperfusion) without further treatment or followed by administration of vehicle (twice daily by intraperitoneal injection for 7 d starting 12 h after reperfusion). Note, the two MCAO ischemia–reperfusion control groups are models for LVO ischemic stroke-treated with standard-of-care approaches to restore tissue-saving blood flow to compromised ischemic brain regions. A fourth control group, received the free-radical quencher edaravone (5 mg/kg, once daily by intraperitoneal injection for 7 d) starting 12 h after MCAO ischemia–reperfusion. Edaravone-Dexborneol, the first innovative drug globally to receive FDA’s breakthrough therapy designation for acute ischemic stroke, is administered to patients within 48 h of onset of acute ischemic stroke after standard-of-care treatment has been administered to restore blood flow to the ischemic brain (44, 45). DHP-103 was administered by intraperitoneal injection (2, 5, or 10 mg/kg twice daily for 7 d) starting 12 h after MCAO ischemia–reperfusion, at the height of neuroinflammation (SI Appendix, Fig. S8 A and B).

Infarct volume was significantly reduced by DHP-103 (2, 5, or 10 mg/kg), with 10 mg/kg decreasing infarct volume by nearly 50% compared to control rats that underwent MCAO ischemia–reperfusion and received either no further treatment or vehicle (Fig. 4C and SI Appendix, Fig. S9A). All three doses of DHP-103 were more effective than edaravone (Fig. 4C). Treatment with the two highest doses of DHP-103 improved balance beam performance (a measure of proprioception, SI Appendix, Fig. S8C) on days 1 and 2 compared to the MCAO-reperfusion vehicle-group, with efficacy exceeding that of edaravone treatment (Fig. 4D). By day 7, proprioception scores had almost returned to baseline, and we were unable to distinguish statistical differences between controls and treated groups (Fig. 4D). We cannot explain why the proprioception scores on day 7 returned to baseline levels (Fig. 4D) in MCAO ischemia–reperfusion control groups, since these rats showed significant infarct size (Fig. 4C) and the presence of activated microglia (Fig. 4E) in infarcted brain regions (Fig. 4E) on day 8.

K_Ca3.1 plays a critical role in calcium signaling, the respiratory burst, migration, proliferation, and nitric oxide production of activated microglia, and inhibitors of the channel suppress microglia-driven neuroinflammation (37, 40, 46). In our studies on a mouse microglial cell line, DHP-103 blocked endogenous K_Ca3.1 channels in the potency range that blocks the recombinant human K_Ca3.1 channel (SI Appendix, Fig. S10 A and B). Moreover, DHP-103 suppressed thapsigargin-induced store-operated calcium entry (SI Appendix, Fig. S10 C–E) at a concentration (1 µM) comparable to the free brain concentration of DHP-103 after IP administration (Fig. 4A and SI Appendix, Table S5A). Guided by these results, we examined whether treatment with DHP-103 reduced the numbers of activated microglia in infarcted regions. We used two markers to identify activated microglia in infarcted brain sections, CD11b (highly expressed on myeloid-lineage cells including microglia) and Iba1 (upregulated during microglial activation). We compared three groups (five animals in each group): 1) MCAO ischemia–reperfusion without treatment, 2) MCAO ischemia–reperfusion administered vehicle, and 3) MCAO ischemia– reperfusion administered DHP-103 (10 mg/kg). Treatment with DHP-103 (10 mg/kg) significantly reduced the number of Iba1⁺CD11b⁺ activated microglia in infarcted regions of the hippocampus (CA1), caudate putamen and striatum (CPu), primary motor cortex (M1), and primary somatosensory cortex (S1) compared to negative controls (Fig. 4E and SI Appendix, Fig. S9B). We did not assess T cell numbers in infarcted regions but suspect that suppression of T cell infiltration contributes to DHP-103’s efficacy, based on the established physiological role of K_Ca3.1 in T lymphocytes (15, 22, 37, 38).

In summary, DHP-103 treatment initiated 12 h after ischemic insult and reestablishment of blood flow to infarcted brain regions reduces infarct volume, improves proprioception scores, and reduces the number of activated microglia in the infarcted brain compared to the MCAO ischemia–reperfusion control groups (without further treatment or with treatment with vehicle). Although our study was done on male rats, our findings should be applicable to females, as other K_Ca3.1 blockers suppress microglia-mediated neuroinflammation equally effectively in males and females in rodent models of Alzheimer’s disease and Parkinson’s disease (47, 48).

Conclusion

We reveal four important new findings in this study. 1) Using a SOSA approach (17), we transformed 1,4-dihydropyridines, canonical blockers of L-type voltage-gated calcium channels, into potent and selective inhibitors targeting the intermediate-conductance calcium-activated K⁺ channel, K_Ca3.1, a therapeutic target in small and large animals for a diverse list of diseases. One of these, DHP-103 (logP: 3.45), a 1,4-dihydropyridine defined as a New Chemical Entity (NCE) by the criteria of the Chemical Abstract Services (CAS), is marginally more potent (IC₅₀: 6 nM) than other K_Ca3.1 inhibitors including TRAM-34 (25 nM), Senicapoc (11 nM), NS6180 (9 nM), 4-phenyl-4H-pyran rac-11 (8 nM), cyclohexadiene rac-16 (8 nM), tri-fluoro trivanillic ester 13b (19 nM) and fused thiazin-3-ones (30 to 100 nM) (21–23, 49, 50). DHP-103’s >500-fold selectivity for K_Ca3.1 over a panel of 116 other targets, including channels, transporters, receptors, and enzymes, is better characterized than that for the other inhibitors. Furthermore, DHP-103 belongs to a chemical class of compounds (1,4-dihydropyridines) used safely for five decades with well-established bioavailability and toxicity in humans. 2) We used cryo-EM, mutagenesis, and electrophysiology to delineate the binding site of DHP-103 within the K_Ca3.1 channel pore. DHP-103’s direct interactions with residues in the cavity (Thr250, Val275, Thr278) and with permeant ions (K4: innermost K⁺ in the selectivity filter; K5: additional K⁺ in central cavity), and K5’s interaction with Val282, stabilizes the inhibitor at the binding site and impedes ion permeation through the pore. Increasing the external K⁺ concentration does not displace DHP-103 from its binding site, and the potency of DHP-103 is not affected by the internal Ca²⁺ concentration. Our studies support the idea that DHP-103 traverses the entire lipid bilayer, then passes through the open inner gate in the fully activated channel (10.2 Å diameter) to reach K_Ca3.1’s central cavity, and perhaps narrows the inner gate (7 Å diameter) after occupying its binding site. Further testing with kinetic studies and molecular dynamics simulation is required to determine the veracity of this idea. 3) Using electrophysiology and Rb⁺ flux measurements, we showed that DHP-103 inhibits disease-relevant gain-of-function mutant K_Ca3.1 channels (particularly the R352H mutation) that cause hereditary xerocytosis, highlighting its possible use to treat this hemolytic anemia. 4) In proof-of-concept studies in a rat LVO-ischemia–reperfusion stroke model, we demonstrated that DHP-103 effectively reduced infarct size and dampened microglia-mediated neuroinflammation compared to control rats that underwent LVO-ischemia followed by renewal of tissue-saving blood supply to the ischemic brain (modeling standard-of-care therapy). DHP-103’s excellent brain penetration, pharmacokinetic properties, metabolic stability, and tolerability in rodents suggests its possible use as an adjunct treatment for acute LVO ischemic stroke following standard-of-care endovascular thrombectomy to restore blood flow to compromised brain regions. Additional detailed pharmacokinetic, pharmacodynamic, and toxicological studies will be required to develop this series of 1,4-dihydropyridines into therapeutics for K_Ca3.1-related diseases.

Methods

1,4-Dihydropyridines and Other Compounds Used in This Study.

Amlodipine, Manidipine, Nimodipine, Nilvadipine, Cilnidipine, Nifedipine, and Nitrendipine were purchased from Sigma-Aldrich; Levamlodipine, from MedChemExpress; and Edaravone from TargetMol. The methods of synthesis of 1,4-dihydropyridines DHP-56, DHP-57, DHP-79, DHP-84, DHP-87, DHP-101, DHP-102, DHP-103, DHP-104, DHP-105, DHP-106, DHP-108, DHP-109, DHP-110, and DHP-113 are described in SI Appendix, Supplementary Methods.

Screening of 1,4-Dihydropyridines by Electrophysiology.

1,4-dihydropyridines were screened for block of K_Ca3.1 and other channels by patch-clamp using a QPatch HTX automated electrophysiology platform (Sophion, Denmark) and by manual patch-clamp. For QPatch experiments, the giga-seal and whole-cell requirements for the automated electrophysiology were the following: minimum seal resistance of 0.1 GΩ, holding potential −80 mV, holding pressure −20 mbar, positioning pressure −70 mbar. For K_Ca3.1 experiments, the external solution was Na⁺-Ringer’s containing (in mM): 160 NaCl, 10 HEPES, 4.5 KCl, 1 MgCl₂, 2 CaCl₂ (pH 7.2, 310 mOsm). The internal solution contained (in mM): 120 KCl, 10 HEPES, 1.75 MgCl₂, 1 Na₂ATP, 10 EGTA, 8.6 CaCl₂ (1 μM free Ca²⁺), (pH 7.4, 300 mOsm). Following establishment of whole-cell configuration, cells were held at −80 mV and K_Ca3.1 currents elicited by a voltage protocol that held at −80 mV for 20 ms, stepped to −120 mV for 20 ms, ramped from −120 to +40 mV over 200 ms, and then stepped back to −120 mV for 20 ms. This pulse protocol was applied every 10 s. Current slopes (in amperes per second) were measured using the Sophion QPatch software and exported to Microsoft Excel and GraphPad Prism 9 for analysis. Decreases in slope between −85 and −65 mV were used to determine the IC₅₀ for K_Ca3.1 inhibition. Cells with membrane capacitances of 10 to 30 pF were analyzed. Manual patch-clamp studies on K_Ca3.1 (Figs. 1, 2 C and D, and 3 C–E) and other ion channels are described in SI Appendix, Supplementary Methods.

Electrophysiological Studies on Gain-of-Function K_Ca3.1 Mutations that Cause Hereditary Xerocytosis.

Mutations were introduced into human K_Ca3.1 channels using the QuickChange II site-directed mutagenesis kit (Agilent) or through molecular cloning services (Genscript). Wild-type and mutant channel cDNAs constructed in pIRES2-AcGFP1 (Clontech) were transfected into HEK293 cells by the calcium–phosphate method. K_Ca3.1 currents shown in Fig. 3 were recorded 1 to 2 d after transfection, using an inside-out patch configuration with an Axon200B amplifier (Molecular Devices) at room temperature. pClamp 10.5 (Molecular Devices) was used for data acquisition and analysis. Resistance of the patch electrodes ranged from 3 to 5 MΩ. The pipette solution contained (in mM): 140 KCl, 10 HEPES (pH 7.4), 1 MgSO₄. The bath solution contained (in mM): 140 KCl, 10 HEPES (pH 7.2), 1 EGTA, 0.1 Dibromo-BAPTA, and 1 HEDTA, mixed with Ca²⁺ to yield desired free Ca²⁺ concentrations, calculated using Maxchelator (Chris Patton: https://somapp.ucdmc.ucdavis.edu/pharmacology/bers/maxchelator/webmaxc/webmaxcS.htm). The intracellular face of the inside-out patch was initially exposed to bath solution containing 1 μM free Ca²⁺. Channel responses to a range of concentrations of tested compounds were measured in the presence of 1 μM free Ca²⁺. Currents were recorded by repetitive 1-s-voltage ramps from −100 to 100 mV from a holding potential of 0 mV. To construct concentration-dependent inhibition curves, we normalized current amplitudes at −90 mV in the presence of DHP-103 to that in its absence. Normalized currents were plotted as a function of DHP-103 concentration. IC₅₀ values and Hill coefficients were determined by fitting data points to a Hill equation.

Erythrocytes Used for ⁸³Rb Influx Assay.

Human blood samples were obtained with written informed consent from patients with dehydrated stomatocytosis (hereditary xerocytosis; HX). Blood collected in heparin-containing vacutainer tubes was transported by courier from local collection centers to Beth Israel Deaconess Medical Center. All procedures were HIPAA-compliant and approved by the Institutional Review Board of Beth Israel Deaconess Medical Center. Erythrocyte preparation and erythrocyte ⁸³Rb influx assay are described in SI Appendix, Supplementary Methods.

Middle Cerebral Artery Occlusion (MCAO) Ischemia–Reperfusion Stroke Model in Rats.

MCAO ischemia–reperfusion model studies in rats were conducted by ICE Bioscience (China) with approval of the Ethics and Moral Management Committee of Experimental Animals at Xuzhou Medical University. Male Sprague-Dawley rats weighing 240 to 270 g were purchased from Jinan Pengyue Experimental Animal Breeding Company (China). Focal cerebral ischemia was induced by occlusion of the right middle cerebral artery according to the Longa method (43). Animals were anesthetized with 10% chloral hydrate and a median incision in the neck was made to expose the right common carotid artery, internal carotid artery, and external carotid artery. A slipknot on the common carotid artery, a dead knot on the proximal side of the external carotid artery, and a slipknot on the external carotid artery near the common carotid bifurcation were tied with a silk thread. A small opening between the two knots of the external carotid artery was cut to insert a thread plug into the carotid artery, which was advanced inward into the MCA. The plug was kept in place for 60 min, then removed from the blood vessel to restore blood supply. Reperfusion continued for 7 d and the rats were sacrificed using carbon dioxide asphyxiation on day 8.

We used seven groups for our study, each group consisting of 29 rats: sham-operated rats (blunt dissection, exposure of right common carotid artery, then suture-closure), MCAO ischemia–reperfusion without treatment, MCAO ischemia–reperfusion administered vehicle, MCAO-reperfusion administered edavarone (5 mg/kg), MCAO ischemia–reperfusion administered DHP-103 (2 mg/kg), MCAO ischemia–reperfusion administered DHP-103 (5 mg/kg), and MCAO ischemia–reperfusion administered DHP-103 (10 mg/kg). Treatment began 12 h after reperfusion and continued until day 7. Rats received intraperitoneal injections of vehicle (5% v/v N-Methylpyrrolidone (NMP), 7.5% v/v Solutol HS 15, 50% v/v PEG-400, and 37.5% v/v TPGS [10% w/v]) twice daily, or DHP-103 formulated in vehicle (2, 5, 10 mg/kg, twice daily) or edaravone (5 mg/kg once daily; dissolved in DMSO before formulation in normal saline).

Immediately after reperfusion we randomized 24 of the 29 rats from each group for proprioception assessment. Proprioception was assessed individually using the balance beam test on day 0 (1 h after reperfusion) (n = 8 from each group), day 1 (n = 7), day 2 (n = 8), and day 7 (n = 5) (Fig. 4 D and E). To reduce stress, we used different rats for assessment on different days.

All 24 rats from each group that were used for proprioception assessment were euthanized on day 8 after reperfusion. Their brain slices (2 to 16 mm from frontal pole to occipital pole) were stained with 2,3,5 triphenyltetrazolium (TTC) to determine % infarct volume (Fig. 4F and SI Appendix, Fig. S6A). Infarct area was assessed by ImageJ analysis of scanned TTC-stained whole brain sections.

Immediately after reperfusion we randomized 5 of the 29 rats from each group for assessment of activated microglia. These rats were euthanized on day 8 after reperfusion and their brain sections (hippocampus-CA1, caudate-putamen, primary cortex, and somatosensory cortex) were immunostained for markers of activated microglia (Iba1 and CD11b) (n = 5) (Fig. 4H and SI Appendix, Fig. S6B). Details of immunohistochemical analysis of microglia and leukocyte markers are described in http://www.pnas.org/lookup/doi/10.1073/pnas.2425494122#supplementary-materialsSI Appendix, Supplementary Methods.

Other Methods.

http://www.pnas.org/lookup/doi/10.1073/pnas.2425494122#supplementary-materialsSI Appendix, Supplementary Methods include details of compound synthesis, electrophysiology studies using the manual patch clamp system (Figs. 1–3), electrophysiological studies on the K_Ca3.1 T250S-V275A and K_Ca3.1_T212F-V272F mutants, metabolite stability in mouse and human liver microsomes, biotransformation of DHP-103 in mouse, rat, dog, monkey, and human liver microsomes, purification of K_Ca3.1/CaM protein, cryo-EM structure determination, erythrocyte preparation, erythrocyte Rb⁺ flux, selectivity analysis by competitive binding and enzyme uptake assays, cytochrome P450 enzyme inhibition assays, immunohistochemistry analysis on microglia, electrophysiology studies on mouse microglia, measurement of thapsigargin-induced store-operated calcium entry into mouse microglia, pharmacokinetics, and brain uptake of DHP-103 in mouse and rat.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

The atomic coordinates have been deposited in the Protein Data Bank (PDB) under accession codes 9ED1 (https://doi.org/10.2210/pdb9ED1/pdb) (103_KCa3.1) (51). The cryo-EM map has been deposited in the Electron Microscopy Data Bank (EMDB) under accession codes EMD-47930 (https://www.ebi.ac.uk/pdbe/entry/emdb/EMD-47930) (103_KCa3.1) (51). All study data are included in the article and/or SI Appendix.