Development of charybdotoxin Q18F variant as a selective peptide blocker of neuronal BK(α + β4) channel for the treatment of epileptic seizures

Abstract

Epilepsy is the results from the imbalance between inhibition and excitation in neural circuits, which is mainly treated by some chemical drugs with side effects. Gain‐of‐function of BK channels or knockout of its β4 subunit associates with spontaneous epilepsy. Currently, few reports were published about the efficacy of BK(α + β4) channel modulators in epilepsy prevention. Charybdotoxin is a non‐specific inhibitor of BK and other K⁺ channels. Here, by nuclear magnetic resonance (NMR) and other biochemical techniques, we found that charybdotoxin might interact with the extracellular loop of human β4 subunit (i.e., hβ4‐loop) of BK(α + β4) channel at a molar ratio 4:1 (hβ4‐loop vs. charybdotoxin). Charybdotoxin enhanced its ability to prevent K⁺ current of BK(α + β4 H101Y) channel. The charybdotoxin Q18F variant selectively reduced the neuronal spiking frequency and increased interspike intervals of BK(α + β4) channel by π–π stacking interactions between its residue Phe¹⁸ and residue His¹⁰¹ of hβ4‐loop. Moreover, intrahippocampal infusion of charybdotoxin Q18F variant significantly increased latency time of seizure, reduced seizure duration and seizure numbers on pentylenetetrazole‐induced pre‐sensitized rats, inhibited hippocampal hyperexcitability and c‐Fos expression, and displayed neuroprotective effects on hippocampal neurons. These results implied that charybdotoxin Q18F variant could be potentially used for intractable epilepsy treatment by therapeutically targeting BK(α + β4) channel.

1. INTRODUCTION

Large‐conductance Ca²⁺ and voltage‐gated K⁺ (BK) channels share similar pore structural determinants and sensitivities to toxins with voltage‐dependent K⁺ (K_V) channels. ¹ BK channels have diverse functions, including firing frequency, ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ regulating the action potential (AP) shape and neurotransmitter release. ⁹ BKα channel is composed of four identical pore‐forming α subunits. ¹⁰ The α subunit assembles with four auxiliary β subunits (β1–β4) as well as four γ subunits (γ1–γ4), rendering functional diversity and tissue‐specificity of BK channels by modifying gating properties and pharmacological features. ¹¹ , ¹² The four β subunits have the same topology with two trans‐membrane fragments, connected by one extracellular loop, and with their N‐ and C‐termini locating in the cytoplasm. The β4 subunit is highly expressed in neurons throughout the brain ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ and slows down the kinetics of either activation or deactivation in BK channels. ¹³ , ¹⁴ It induces negative shifts of the conductance‐voltage curves of BK channels at high Ca²⁺ concentration, and positive shifts at low Ca²⁺ concentration. ¹⁸ , ¹⁹ , ²⁰ Dentate granule cells derived from the mice lacking β4 subunit display a gain‐of‐function of BK channels. Gain‐of‐function of BK channels sharpens the action potentials, thus facilitates high‐frequency firing and results in temporal lobe seizures. ⁸ , ²¹ , ²² , ²³ Previous studies have also shown that the inhibition of BK channels suppresses the firing of action potentials by prolonging the duration of after‐hyperpolarization potential (i.e., AHP). ⁸ , ²³

Epilepsy has been characterized by recurrent and unprovoked seizures, which results from an imbalance between inhibition and excitation in neural circuits. It can be suppressed by chemical drugs, such as phenytoin and carbamazepine. ²⁴ Although these anti‐epileptic drugs (AEDs) are effective in inhibiting seizures and improving symptoms, many side effects were reported, including cognitive impairment and emotional change, dysfunction of liver and kidney and even aplastic anemia. ²⁵ Thus, it's indispensable to exploit new high‐efficiently neuron‐targeted AEDs or AED leads with low side effects. BK channels' antagonists were reported to possess anticonvulsant effects in vitro, for example, paxilline can weaken elevated excitability and maintain firing at normal levels. ²⁶ As a specific inhibitor of BK(α + β4) channel, recombinant martentoxin was previously reported by us to efficiently prevent epilepsy by selective interaction with BK(α + β4) channel. ²⁷ Except martentoxin, other BK(α + β4) channel modulators were seldom reported as AED.

Charybdotoxin (ChTX), originally isolated from venom of scorpion Leiurus quinquestriatus hebraeus, ²⁸ is a well‐known blocker of potassium channels. ²⁹ It can block BKα channel through physically occluding the K⁺ conduction pore. BK(α + β1) channel has a binding affinity to ChTX identical to BKα channel, ³⁰ while BK(α + β2) and BK(α + β3) channels have about 30‐fold lower sensitivity to ChTX in equilibrium. ³⁰ , ³¹ In BK(α + β4) channel, the residues Lys¹²⁰, Arg¹²¹ and Lys¹²⁵ in the extracellular loop of human β4 subunit (i.e., hβ4‐loop) specifically impede ChTX to bind with the pore, making BK(α + β4) channel display about 1000‐fold slower association with ChTX than BKα channel. ¹⁵ However, under high concentration of ChTX and long exposure time, ChTX binds to BK (α + β4) channel irreversibly. ¹⁵ Thus, is it possible to develop ChTX variants to selectively inhibit excitation of BK (α + β4) channel for treatment of epileptic seizure? To address this, we here studied how hβ4‐loop interacted with recombinant charybdotoxin (i.e., rChTX), and designed ChTX variant‐Q18F as a potential AED targeting BK(α + β4) channel.

2. RESULTS AND DISCUSSION

2.1. Inhibitory effects of rChTX on BK(α + β4) channel

Isolation and purification of native ChTX from scorpion venom is costly and obtain very rare peptides. We here strived to get biochemically stable and sufficient amount of genetically rChTX as described in the experimental section. The folding and integrality of recombinant peptide were confirmed by NMR technique and MALDI‐TOF mass spectrometry (Figure S1a, b), respectively. The three‐dimensional (3D) solution structure of rChTX was almost identical to that of native form (Figure S2). The pharmacological properties through whole‐cell patch clamp ²⁷ , ³² were presented in Figure 1a. The rChTX at 10 μM dose effectively blocked the K⁺ currents of BK(α + β4) channel (the fraction of unblocked K⁺ currents, I_f ^{rChTX‐WT, BK(α+β4)} = 0.5 ± 0.1, Figure 1a and Table S1), which was weaker than that of native ChTX at the same dose. ¹⁵ The dose response curve was obtained and the IC₅₀ value of rChTX on BK(α + β4) channel was calculated to be 0.9 ± 0.1 μM, with the Hill coefficient equal to 1.6 ± 0.4. Overall, the rChTX has a right global folding and effective bioactivities to human BK(α + β4) channel.

FIGURE 1.

rChTX‐WT interacts with hβ4‐loop and its mutants, and inhibits the outward currents of BK(α + β4) and its mutant channels. (a) Left, representative traces of HEK293T cells expressing BK(α + β4) before and after 10 μM rChTX application. Holding voltage was −80 mV. Testing potential pulse was +80 mV for 200 ms with 300 nM of Ca²⁺ in the pipette solution. Right, dose–response curve of the rChTX inhibition on BK(α + β4) (n = 3–6) was fitted by the Hill equation. The IC₅₀ value was calculated as 0.9 ± 0.1 μM and the Hill coefficient was 1.6 ± 0.4. (b–d) The binding affinities of hβ4‐loop and its variants to rChTX‐WT were measured by MST assay. (a, e–i) Current traces before (black) and after 10 μM rChTX application on BK(α + β4) (red, a), BK(α + β4 Q124A) (orange, e), BK(α + β4 E104K) (green, f), BK(α + β4 E122K) (magenta, g), BK(α + β4 K120E/R121E/K125E) (violet, h) and BK(α + β4 H101Y) (blue, i) channel, respectively. Holding voltage was −80 mV. Testing potential pulse was +80 mV for 200 ms with 300 nM of Ca²⁺ in the pipette solution. (j) Fraction of the currents of BK(α + β4) and mutants channels inhibited by rChTX‐WT (n = 5) *p < .05; **p < .01; ***p < .001. Comparisons were performed among BK(α + β4 Q124A), BK(α + β4 E104K), BK(α + β4 E122K), BK(α + β4 K120E/R121E/K125E), BK(α + β4 H101Y), and BK(α + β4) channels

2.2. rChTX weakly interacts with hβ4‐loop

Our previously reported NMR solution structure of hβ4‐loop ³³ (Figure S1) is valuable for us to probe its potential interactions with rChTX. We firstly titrated rChTX into ¹⁵ N‐labeled hβ4‐loop solution and ran two‐dimensional (2D) ¹H–¹⁵ N HSQC experiments (Figure S1), which demonstrated slight chemical shift of the cross‐peaks in the spectrum. This suggested weak interactions between rChTX and hβ4‐loop, consistent with the value of binding affinity (K _d  = 205.0 ± 12.9 μM) measured by microscale thermophoresis (i.e., MST) assay (Figure 1b). Then, by applying ¹³C and ¹⁵ N isotope double labeled hβ4‐loop protein mixed with unlabeled rChTX peptide, we collected a series of 2D and 3D NMR spectra, and successfully assigned NMR signals of the most backbone and sidechain atoms of bound hβ4‐loop, and of the protons in bound rChTX. Additionally, based on the difference between ¹⁵ N or ¹³C‐edited 3D HSQC‐NOESY spectra with and without ¹⁵ N or ¹³C decoupling ³⁴ (Figure S3), 45 intermolecular NOEs between rChTX and hβ4‐loop were obtained (Table S2). Similar with these observed in recombinant martentoxin, ²⁷ the residues with intermolecular NOEs were found around the toxin, while mainly located on a single, flexible region of hβ4‐loop from Leu¹⁰⁰ to Gln¹²⁶ (including α‐helix α1 and loop L6 in hβ4‐loop). However, we only found and assigned one set of protein's NMR signals, which implied that several hβ4‐loop molecules might encircle symmetrically around one molecular rChTX.

Due to weak interactions, unfortunately, we failed to get binding affinity value and accurate binding mole ratio of rChTX with hβ4‐loop directly through isothermal titration calorimetry (ITC) and other biological techniques. It's known that BKα channel is a tetramer composed of the pore‐forming α subunit. ³⁵ , ³⁶ , ³⁷ , ³⁸ In neuron cells of mammals, the α subunit is expressed with auxiliary β4 subunit at a mole ratio 1:1. ³⁹ The cryo‐EM structures of human BKα channel in complex with β4 subunit demonstrates that four β4 subunits encircle around α subunits, forming a tetrameric crown over the pore on the extracellular side. ⁴⁰ Thus, we proposed that rChTX might interact with hβ4‐loop at a 1:4 (rChTX vs hβ4‐loop) stoichiometry.

2.3. Structural model of hβ4‐loop interacted with rChTX

We then calculated the solution structure of four hβ4‐loop molecules in complex with one molecular rChTX using NMR NOE‐derived distance constraints and dihedral angle restraints with XPLOR‐NIH program (Table S3). The final ensemble, which contained 20 structures of hβ4‐loop‐rChTX complexes with the lowest energies, were displayed in Figure S4a, with the root‐mean‐square deviation (i.e., RMSD) values of 0.27 ± 0.10 Å and 0.39 ± 0.12 Å for backbone and heavy atoms in the secondary structural regions, respectively. Four hβ4‐loop molecules (i.e., monomers A, B, C, and D, Figure S4b) formed a rectangle‐like pore with a length close to 28 Å and a width about 23 Å, where one molecular rChTX located, and blocked K⁺ current. The binding interfaces between hβ4‐loop monomer B, monomer C, monomer D and rChTX were confirmed by the chemical cross‐linking of proteins coupled with mass spectrometry (i.e., CXMS) (Figures S5 and S6). From this assay, the residues Gly⁻², Lys²⁷, Lys³¹ of rChTX were close to the residues Lys¹¹² and Lys¹²⁵ in different monomers of hβ4‐loop, respectively, which was consistent with the observation in the complex structural model.

On the whole, rChTX and hβ4‐loop maintained their global folding, while their interacting regions changed local conformations (Figure S7). The side‐chains of residue Glu¹²² in hβ4‐loop monomer A and D formed two hydrogen bonds with the side‐chains of residues Trp¹⁴ and Lys³² of rChTX (Figure S4c, d), and the side‐chains of residue Gln¹²⁴ in hβ4‐loop monomers C and D were 5.3 and 2.8 Å away from the side‐chains of residues Arg¹⁹ and Lys³¹ in rChTX (Figure S4e, f), respectively. Moreover, the negatively charged side‐chains of Glu¹⁰⁴ in all hβ4‐loop monomers formed strong or weak hydrogen bonds with the side‐chains of residues Trp¹⁴, Arg²⁵, Lys²⁷, and Lys³¹ of rChTX (Figure S8), respectively.

To further verify these binding sites, we carried out site‐directed mutagenesis studies. As shown in Figure S9a, the variants did not change the whole folding of hβ4‐loop, detected by circular dichroism (CD) spectra. The mutation from Gln¹²⁴ to Ala¹²⁴ reduced its binding affinity to rChTX‐WT by about 1.5‐fold (Figure 1b and Table S4). Meanwhile, rChTX‐WT almost lost the blockade effects on the outward currents of BK(α + β4 Q124A) channel (Figure 1e and Table S1) compared with BK(α + β4) channel (Figure 1a, j and Table S1). These results further confirmed the roles of Gln¹²⁴ of hβ4‐loop in its binding to rChTX. The variants hβ4‐loop‐E104K and hβ4‐loop‐E122K did not display detectable binding affinities to rChTX (Figure 1c, d), suggesting that Glu104 and Glu122 were both involved in the interactions between rChTX and hβ4‐loop. However, rChTX‐WT demonstrated inhibitory effects on the K⁺ currents of BK(α + β4 E104K) and BK(α + β4 E122K) channels almost similar to that on BK(α + β4) channel (p > .05, Figure 1f, g, j and Table S1). The wash of rChTX on BK(α + β4) or its mutants were also shown in the Figure 1a, e–i. Additionally, three positively charged residues Lys¹²⁰, Arg¹²¹, and Lys¹²⁵ in hβ4‐loop were suggested to impede ChTX entry by repulsive electrostatic interaction. ⁴¹ To validate this, we designed hβ4‐loop‐K120E/R121E/K125E variant, which demonstrated stronger binding affinity to rChTX than wild‐type hβ4‐loop by more than three‐fold (Figure 1b and Table S4). Meanwhile, rChTX‐WT had stronger inhibitory effect on BK(α + β4 K120E/R121E/K125E) channel than that on BK (α + β4) channel (p < .05, Figure 1h, j and Table S1), confirming that repulsive electrostatic interaction weakened the selectivity of ChTX on BK(α + β4) channel.

2.4. Design rChTX‐Q18F variant to enhance inhibition of K⁺ current of BK(α + β4) channel

Previously, four residues Leu⁹⁰, Tyr⁹¹, Thr⁹³, and Glu⁹⁴ of β1 subunit were reported to be critical in conferring high affinity of ChTX to BK(α + β1) channel (which was stronger than BK(α + β4) channel by 50‐fold). ⁴² As shown in the sequence alignment (Figure S10), Leu¹⁰⁰ and Glu¹⁰⁴ are conserved in four β subunits of BK channel. Residues Leu¹⁰⁰, His¹⁰¹, Asp¹⁰³, and Glu¹⁰⁴ in hβ4‐loop corresponded to the residues Leu⁹⁰, Tyr⁹¹, Thr⁹³, and Glu⁹⁴ of β1 subunit. In our structural model, His¹⁰¹ and Glu¹⁰⁴ were involved in the interactions with rChTX (Figures 2 and S8), while Leu¹⁰⁰ and Asp¹⁰³ were not. Thus, besides hβ4‐loop‐E104K mutant, we designed hβ4‐loop‐H101Y variant. However, hβ4‐loop‐H101Y variant was expressed in E coli as inclusion bodies. Fortunately, BK(α + β4 H101Y) channel could be expressed in HEK293T cells. Obviously, rChTX‐WT displayed much higher sensitivity to BK(α + β4 H101Y) channel than BK(α + β4) channel (p < .001, Figure 1i, j and Table S1), which implied that the hydrophilic side‐chain of Tyr¹⁰¹ was important for hβ4‐loop to interact with rChTX. As shown in Figure 2, the side‐chain of residue His¹⁰¹ of hβ4‐loop was close to the hydrophilic sidechains of residues Ser¹⁰, Gln¹⁸, Arg¹⁹, Thr²³, Arg²⁵, Lys²⁷, Lys³², and Arg³⁴. Among these residues, the basic amino acids Arg¹⁹, Lys³², and Arg³⁴ are highly conserved in ChTX, IbTX, and SloTX toxin family members. ⁴¹ Moreover, Lys²⁷ is the binding site of ChTX with BKα channel. Therefore, here, to enhance hydrophobic π–π interaction of rChTX with residue His¹⁰¹ of hβ4‐loop, we just designed three rChTX variants S10F, Q18F, and T23F. At the same time, to testify whether residue Arg²⁵ of rChTX has cation‐𝜋 interactions with His¹⁰¹ of hβ4‐loop, we designed rChTX‐R25E variant.

FIGURE 2.

Position analysis of residue His¹⁰¹ in the structure model of hβ4‐loop‐rChTX complex. The residues in rChTX were indicated in yellow, His¹⁰¹ residues of different hβ4‐loop monomers were marked in different colors, respectively

As shown in Figure S9b, these four mutations did not change the global folding of rChTX. Compared with rChTX‐WT, the variants S10F, T23F, and R25E demonstrated slightly weaker binding affinities to hβ4‐loop by about 1.3–2.5 fold (Figure 3a and Table S4). The inhibitory effects of S10F and R25E variants were almost identical to or slightly stronger than that of rChTX‐WT on BK(α + β4) channel (p > .05, Figure 3b, c, d, g and Table S1). The rChTX variant T23F obviously decreased inhibitory abilities of K⁺ outward currents on BK(α + β4) channel (p < .05, Figure 3e, g and Table S1), which further suggested that the residue Thr²³ was essential to pharmacological characteristics of rChTX. Interestingly, rChTX‐Q18F variant demonstrated stronger interactions with hβ4‐loop than rChTX‐WT (Figure 3a and Table S4) by about two‐fold. The time course underlying the inhibition or wash of rChTX‐WT or rChTX‐Q18F on BK(α + β4) channel were displayed in Figure S11. More importantly, the mutation from Gln¹⁸ to Phe¹⁸ in rChTX could remarkably enhance inhibitory effect on BK(α + β4) channel (p < .01, Figure 3f, g and Table S1) by about two‐fold, even better than that of martentoxin (i.e., MarTX, the specific blocker of BK(α + β4) channel with I_f ^{rMarTX, BK(α+β4)} = 0.3 ± 0.1). ²⁷ Thus, the replacement from Gln¹⁸ to Phe¹⁸ (Q18F) in rChTX played a crucial role in modulating the neuroexcitability through specific identification of BK(α + β4) channel. rChTX‐Q18F variant could be a novel anti‐epilepsy drug by efficiently targeting BK(α + β4) channel in neuron system.

FIGURE 3.

rChTX‐WT or its mutants interact with hβ4‐loop and inhibit the outward K⁺ currents of BK(α + β4) channel. (a) The binding affinity of hβ4‐loop to rChTX‐WT or its variants, measured by MST assay. (b–f) Current traces before (black) and after 10 μM rChTX‐WT (red, b), rChTX‐S10F (blue, c), rChTX‐R25E (orange, d), rChTX‐T23F (magenta, e) and rChTX‐Q18F (green, f) application on BK(α + β4) channel, respectively. Holding voltage was −80 mV. Testing potential pulse was +80 mV for 200 ms with 300 nM of free Ca²⁺ in the pipette solution. (g) Inhibitory effects of rChTX‐WT and its mutants on current amplitude of BK(α + β4) channel (n = 4). *p < .05; **p < .01. Comparisons were performed among rChTX‐S10F, rChTX‐R25E, rChTX‐T23F, rChTX‐Q18F, and rChTX‐WT groups

2.5. rChTX‐Q18F variant improves the inhibition selectivity of rChTX on different potassium channels

BK channel as well as its subtypes, Kv1.2 and Kv1.3 channels widely distribute in the nervous system. BK (α + β4) and Kv1.2 channels, mainly expressed in neurons, modulate the repolarization of action potentials. BK (α + β1) and Kv1.3 channels are expressed in astrocytes and microglial cells, respectively. ¹ , ² To identify the inhibition selectivity of rChTX‐Q18F on different K⁺ channel subtypes, human Kv1.2, Kv1.3, BKα, and BK(α + β1) channels were heterologously expressed in HEK293T cells, respectively. In Figures 4a–d and Table S5, the delayed rectifier of the K⁺ currents of human Kv1.2 and Kv1.3 channels, evoked at +40 mV pulse, were potently inhibited by rChTX‐WT at both 1 and 10 μM concentrations, whereas rChTX‐Q18F showed weaker inhibition effects on these two channels than rChTX‐WT. Moreover, we also tested the mouse cerebrospinal fluid containing rChTX‐Q18F on the currents of endogenous Kv1.3 in microglial BV2 cells. It could not inhibit the Kv1.3 currents, while PAP‐1, a non‐peptide blocker, can obviously reduce the currents of endogenous Kv1.3 (Figure S11). Similarly, at 1 or 10 μM concentration, rChTX‐WT strongly blocked the K⁺ current of human BKα channel, while rChTX‐Q18F displayed lower sensitivity to the K⁺ current of BKα channel than rChTX‐WT (Figure 4e–f and Table S5). In contrast, 10 μM rChTX‐Q18F showed higher inhibition activity on BK(α + β4) channel than that on BKα or BK(α + β1) channel (Figure 4e–h and Table S5), suggesting that replacement of Gln¹⁸ by Phe¹⁸ changed the inhibition selectivity of rChTX‐WT on different K⁺ channels. Therefore, rChTX‐Q18F may be a selective blocker of BK(α + β4) channel.

FIGURE 4.

Inhibitory effects of rChTX‐WT and rChTX‐Q18F on different potassium channel subtypes. (a, c, e, and g) Current traces before (black) and after 10 μM rChTX‐WT (red) and rChTX‐Q18F (green) application on (a) Kv1.2, (c) Kv1.3, (e) BKα, and (g) BK(α + β1) channels, respectively. Holding voltage was −80 mV. Testing potential pulse was +80 mV for 200 ms, (b, d, f, h) Inhibition of rChTX‐WT and rChTX‐Q18F on the current amplitude of (b) Kv1.2, (d) Kv1.3, (f) BKα, and (h) BK(α + β1) (n = 6) channels *p < .05; **p < .01

2.6. Modulation of rChTX‐Q18F on spiking frequency and AHP properties

As a classical antagonist of γ‐aminobutyric acid (GABA)_A receptor, pentylenetetrazole (PTZ) is a potent convulsant agent for studying epileptic seizure mechanisms. ⁴³ The PTZ pre‐treatment induces gain‐of‐function of BK channels as well as high firing rates in hippocampal and cortical pyramidal neurons. To test whether the inhibition of neuronal BK(α + β4) channel by rChTX‐WT and its variants reduced the spike output after epileptic seizure, we recorded the spike trains evoked by the current injection in PTZ‐induced pre‐seizure neurons in the absence or in the presence of rChTX‐WT and its variants. Examples of action potential (AP) traces were generated by PTZ‐pretreated hippocampal pyramidal neurons upon their being administrated with blank control saline, rChTX‐WT, rChTX‐Q18F and rChTX‐T23F, respectively (Figure 5a and Table S6).

FIGURE 5.

Effects of rChTX‐WT or its variants on firing properties of hippocampal pyramidal cells evoked by the current injections. (a) APs induced by the current clamp in PTZ‐pretreated neurons exerted with saline (black), rChTX‐WT (red), rChTX‐Q18F (blue) and rChTX‐T23F (magenta). (b) Intervals of inter‐spike preceding the ninth AP during a 300 pA current injection. (c) The number of APs elicited at each current injection. (d) AP width calculated at one‐half height at 300 pA current injections. (e) Time constant of AHP deactivation at 300 pA current injections. (f) Amplitude of the fAHP calculated from the pre‐spike voltage to peak after hyperpolarization at 300 pA current injections. Data were displayed as mean ± SEM. *p < .05, **p < .01, and ***p < .001, respectively, compared with the group treated with the saline (One‐way ANOVA). ^# p < .05, ^## p < .01, and ^### p < .001, respectively, compared with the group treated with rChTX‐WT (One‐way ANOVA)

Generally, APs induced by current injections is limited in the firing frequency probably through two different mechanisms. One is that high current injection, for example, 300 pA trace, leads to the action potential failure in several neurons by the end of the current injection; the other is that the inter‐spike intervals results in fewer APs during increasing the current injection over time. In Figure 5b and Table S6, the saline group (PTZ‐pretreated seizure cells treated with saline) showed very little frequency adaptation of spike, resulting in much short interspike intervals. Compared with the saline group, seizure cells treated with rChTX‐WT or rChTX‐Q18F had a significantly longer interspike intervals. In particular, the interspike intervals rChTX‐Q18F group was obviously longer than that of rChTX‐WT group.

Seizure cells treated with rChTX‐Q18F cannot continue to evoke APs at higher current injection (300 pA trace) and exhibit significant AP failure (Figure 5a, c, n = 6), which is much different from the saline group. Thus, long inter‐spike intervals, large AP width, slow decay of the AHP and the appearance of AP failures resulted in a substantial decrease in firing rates of neurons. As summarized in Table S6, the rChTX‐WT or rChTX‐Q18F group showed longer inter‐spike intervals, larger AP width (Figure 5d, n = 6), and slower decay of the AHP than the saline group (Figure 5e, n = 6), the appearance of AP failures (Figure 5c, n = 6) led to a substantial decrease in firing rates. Moreover, compared with rChTX‐WT group, rChTX‐Q18F group displayed longer inter‐spike intervals (Figure 5b, n = 6), slower decay of the AHP (Figure 5e, n = 6) and more substantial decrease in firing rates of seizure neurons at each current trace (Figure 5c, n = 6). At 300 pA trace, the AHP sizes of rChTX‐WT, rChTX‐Q18F, and rChTX‐T23F groups were obviously increased compared with that of the saline group (Figure 5f, n = 6), which improved the threshold of neuron firing. Seizure cells treated with rChTX‐T23F had no significant difference from the PTZ‐pretreated seizure cells (saline group) on the inter‐spike intervals (Figure 5b, n = 6) and the decay of the AHP (Figure 5e, n = 6), whereas seizure cells treated with rChTX‐T23F exhibited less AP numbers (Figure 5a) and reduced AP width (Figure 5d, n = 6), compared with the saline group. These results implied that the seizure cells treated with rChTX‐Q18F had an obviously large time constant to achieve the threshold for neuronal firing, which likely explained the lower frequency of neuronal firing in rChTX‐Q18F group compared with saline group.

In order to further verify the specificity of rChTX‐Q18F for BK (α + β4) channel, the β4 subunit has been downregulated by rAAV‐kcnmb4‐shRNA. As shown in Figure S12, the interspike intervals, number of APs and AHP sizes of β4 KD (i.e., kcnmb4 knockdown) group has no significant differences with β4 KD group with administering rChTX‐Q18F. It might be indicated that rChTX‐Q18F could specifically recognize BK (α + β4) channel.

2.7. Anticonvulsant effects of rChTX‐Q18F on PTZ‐induced pre‐sensitized rats

To test possible anticonvulsant effects of rChTX‐WT and its mutants on the generalized seizures in rats with classic properties, ²⁶ , ⁴³ each group of rats was administered with PTZ by intraperitoneal injection (i.p.). During a 24‐h interval, one group was injected with PTZ + saline as a negative control, while other groups were injected with PTZ + rChTX‐WT, PTZ + rChTX‐Q18F, PTZ + rChTX‐T23F, and PTZ + valproic acid (VPA), respectively. A high mortality rate of seizure rats was induced by PTZ injection in the saline group (mortality rate = 30%, n = 9) (Table S7). Administration of rChTX‐WT or rChTX T23F did not remarkably decrease the mortality rate of PTZ‐induced pre‐sensitized rats, which were 27.27% (n = 11) and 10.00% (n = 10), respectively. However, rChTX‐Q18F prevented the PTZ‐injected seizure rats from epileptic death, reducing the mortality rate to 0% (n = 11).

For surviving rats, the administration of rChTX‐Q18F potently reduced the tonic–clonic seizures for 24 h following the initial episode. The latency of group (PTZ + rChTX‐Q18F) was three times longer than that of the negative control group (p < .001). Also, the seizure duration of this group (PTZ + rChTX‐Q18F) was nearly 50% of that in negative control group (PTZ + saline) (p < .01). The rChTX‐Q18F treatment obviously reduced the number of epileptic seizures at high stage (PTZ + rChTX‐Q18F vs. PTZ + saline, p < .05 for stages 3, 4, and 5). On the contrary, the seizure number of the rats treated with rChTX‐WT or rChTX‐T23F had no significant difference from that of the rats treated with saline (p > .05). In addition, administration of rChTX‐WT was remarkably different from rChTX‐Q18F treatment in the latency (p < .05) and in the seizure number at high stage (p < .05). These results confirmed that rChTX‐Q18F effectively suppressed epileptiform activities during the generalized clonic–tonic seizures, which typically associated with clinical seizure‐behaviors. It also implied that rChTX‐Q18F was important to anticonvulsant effects.

2.8. Modulatory effects of rChTX‐Q18F on local field potential responses of seizure rats

To determine whether rChTX‐Q18F could reduce the intensity of epileptic seizure, the effects of rChTX‐Q18F on the power spectral density (PSD) of local field potential (LFP) during PTZ‐induced seizure were directly tested. LFP signals in PTZ pre‐sensitized seizure rats treated with saline, VPA, rChTX‐WT and its variants rChTX‐Q18F and rChTX‐T23F were recorded and compared with one another. As described in Figure 6, LFP activities of seizure rats were visualized through heat maps of spectral density.

FIGURE 6.

Modulation of rChTX‐WT or its mutants on LFP spectral characteristics and PSD of PTZ‐induced generalized tonic–clonic seizures. (a) LFP signals as well as spectral heat maps from a representative seizure rat treated with saline (black), rChTX‐WT (red), rChTX‐Q18F (blue), rChTX‐T23F (magenta), and VPA (green) were displayed, respectively. (b) Spectral plots as well as cumulative distribution curves from a representative seizure rat treated saline, rChTX‐WT, rChTX‐Q18F, rChTX‐T23F, and VPA were shown, respectively. (c) Spectral analysis on PSD values of δ, θ, α, β, and γ waves of PTZ‐induced seizure group and each drug groups (n = 4). Data was shown as mean ± SEM. *p < .05, **p < .01, and ***p < .001, respectively, comparison of the PTZ‐induced seizure group treated with saline with each drug group (One‐way ANOVA). ^## p < .01 comparison of the PTZ‐induced seizure group treated with rChTX‐WT with each drug group (One‐way ANOVA)

Compared with PTZ‐induced seizure rats in the saline group, rChTX‐Q18F treated seizure rats displayed a notable suppression of LFP activity in hippocampus (Figure 6a), whereas rChTX‐WT treatment had no obviously effects on the LFP activities of PTZ‐induced epileptic seizure. In Figure 6b, the PSD‐frequency curve of the saline, rChTX‐WT and rChTX‐T23F groups exhibited a sharp peak at low frequency (δ or θ) wave bands. This sharp peak was absent in the rChTX‐Q18F group, implying that rChTX‐Q18F fundamentally inhibits the neural‐networks responsible for generating the low frequency waves. According to the statistical data (Figure 6c), rChTX‐Q18F treatment significantly reduced the PSD of LFP, compared with the saline administration on δ waves (p < .001, n = 4), θ waves (p < .01, n = 4) and α waves (p < .05, n = 4). However, neither rChTX‐WT nor rChTX‐T23F PSD valves showed obvious difference on α waves with that of the saline group (p > .05, n = 4). Moreover, rChTX‐WT treatment could not suppress the PSD valves on δ waves, which were significantly higher than that of rChTX‐Q18F treatment (p < .01, n = 4). These results supported that replacement from Gln¹⁸ to Phe¹⁸ of rChTX was essential for suppressing neurohyperexcitability.

2.9. rChTX‐Q18F inhibits c‐Fos expression in hippocampus elicited by PTZ

To confirm whether rChTX‐WT and its variants targeted BK(α + β4) channel in hippocampus, we probed the changes of the c‐Fos expression in the hippocampus region elicited by PTZ. Each ipsilateral and contralateral hippocampus section from the groups treated with saline, rChTX‐WT and its variants were shown in Figure 7 and Table S8. In the saline group, predominant Fos‐like protein immune‐reactivity (FLI) neurons were observed in the layer of pyramidal cells of the hippocampal CA1 and CA3 sections, and granular cells of the DG region, which infrequently distributed in other layers. Application of 10 μM rChTX‐WT or its variants caused significant inhibitory effects on the number of FLI positive cells in both ipsilateral and contralateral distinct hippocampal sections. The rChTX‐WT group showed less number of FLI positive neurons than the saline group, which were expressed in CA1, CA3, and DG regions of ipsilateral (Figure 7a1–a3, b1–b3, e, n = 5) and contralateral hippocampus (Figure 7f1–f3, g1–g3, j, n = 5).

FIGURE 7.

Effects of rChTX‐WT or its mutants on c‐Fos expression in hippocampus region elicited by PTZ. (a–d) The c‐Fos expression in the ipsilateral side of CA1 (a1–d1), CA3 (a2–d2), DG (a3–d3) hippocampus region after the administration of saline, rChTX‐WT, rChTX‐Q18F, or rChTX‐T23F, respectively. (e) Statistical histograms of the number of labeled FLI neurons in the ipsilateral side of each hippocampal region after administrating saline, rChTX, rChTX‐Q18F, or rChTX‐T23F (n = 5). (f–i) The c‐Fos expression in the contralateral side of CA1 (f1–i1), CA3 (f2–i2), DG (f3–i3) hippocampus region after the administration of saline, rChTX‐WT, rChTX‐Q18F, or rChTX‐T23F, respectively. (j) Statistical histograms about the number of labeled FLI neuronal cells in the contralateral side of each hippocampal region after the administration of saline, rChTX‐WT, rChTX‐Q18F, or rChTX‐T23F (n = 5). Data were displayed as mean ± SEM. *p < .05, **p < .01, and ***p < .001, compared with the group treated with saline (One‐way ANOVA). ^# p < .05, ^## p < .01, and ^### p < .001, compared with the group treated with rChTX‐WT (One‐way ANOVA)

Treatment with rChTX‐Q18F significantly inhibited the number of FLI neurons at CA1, CA3, DG regions of ipsilateral (Figure 7a1–a3, c1–c3, e, n = 5) and contralateral hippocampus (Figure 7f1–f3, h1–h3, j, n = 5). The application of rChTX‐T23F also obviously induced the number of FLI positive cells in both ipsilateral (Figure 7a1–a3, d1–d3, e, n = 5) and contralateral distinct hippocampal sections (Figure 7f1–f3, 7i1–7i3, j, n = 5), compared with saline application. In particular, rChTX‐Q18F group exhibited significantly smaller FLI number than that of rChTX‐WT at CA1, CA3 and DG regions of ipsilateral hippocampus (Figure 7b1–b3, 7c1–c3, e, n = 5) and of contralateral hippocampus (Figure 7g1–g3, h1–h3, j, n = 5). These results indicate that rChTX‐Q18F potently suppressed the expression of c‐Fos in hippocampus elicited by PTZ pre‐treatments.

2.10. Neuroprotection of rChTX‐Q18F on hippocampal neurons after PTZ‐induced epileptic seizures

Nissl staining of hippocampal sections usually reveal neuron damages caused by PTZ‐induced epileptic seizure. As shown in Figure 8 and Table S9, compared with saline group, rChTX‐WT prevented the PTZ‐induced neuron loss in hippocampal sub‐regions. The number of Nissl stained neurons in rChTX‐WT group was larger than that in saline group at ipsilateral CA1 (Figure 8a1, b1, e, n = 5) and CA3 pyramidal neuronal layers (Figure 8a2, b2, e, n = 5), and DG granular cell layer (Figure 8a3, b3, e, n = 5), and at contralateral CA3 and DG granular cell layer (Figure 8f3, g3, j, n = 5), no effects were observed on contralateral CA1 (Figure 8f1–f2, g1–g2, j, n = 5). The rChTX‐Q18F variant had dramatically protective effects on hippocampal integrity at ipsilateral and contralateral CA1, CA3, and DG regions (Figure 8c1–c3, h1–h3, n = 5). The number of Nissl‐stained neurons in rChTX‐Q18F group was much larger than that in rChTX‐WT group at ipsilateral CA1(Figure 8c1, 8b1, e, n = 5), CA3 pyramidal neuronal layers (Figure 8c2, b2, e, n = 5) and DG granular cell layer (Figure 8c3, b3, e, n = 5) as well as contralateral CA1 (Figure 8h1, g1, j, n = 5), CA3 pyramidal neuronal layers (Figure 8h2, g2, j, n = 5) and DG granular cell layer (Figure 8h3, g3, j, n = 5). Moreover, the number of Nissl stained neurons in rChTX‐T23F group was more than that of saline group not only at ipsilateral CA3 pyramidal neuronal layers (Figure 8d2, a2, e, n = 5) and DG granular cell layers (Figure 8d3, a3, e, n = 5), but also at contralateral CA3 pyramidal neuronal layers (Figure 8i2, f2, j, n = 5) and DG granular cell layer (Figure 8i3, f3, j, n = 5).

FIGURE 8.

Nissl staining micrographs of each experimental group showed the histological integrity of the hippocampus in PTZ‐induced epileptic rats. (a–d) Nissl staining micrographs of the ipsilateral side of CA1 (a1–d1), CA3 (a2–d2), DG (a3–d3) hippocampus region after saline, rChTX‐WT, rChTX‐Q18F, and rChTX‐T23F were administrated, respectively. (e) Statistical histograms of the number of Nissl stained neurons in the ipsilateral side of each hippocampal region after saline, rChTX‐WT, rChTX‐Q18F, and rChTX‐T23F (n = 5) were administrated, respectively. (f–i) Nissl staining micrographs of the contralateral side of CA1 (f1–i1), CA3 (f2–i2), DG (f3–i3) hippocampus region after the administration of saline, rChTX‐WT, rChTX‐Q18F, or rChTX‐T23F. (j) Statistical histograms about the number of Nissl stained neurons in the contralateral side of each hippocampal region after the administration of saline, rChTX‐WT, rChTX‐Q18F, or rChTX‐T23F (n = 5). Data were described as mean ± SEM. *p < .05, **p < .01, and ***p < .001, respectively, compared with group treated with saline (One‐way ANOVA). ^# p < .05, ^## p < .01, and ^### p < .001, respectively, compared to the group treated with rChTX‐WT (One‐way ANOVA)

In addition, to detect neuroprotective effects of rChTX‐Q18F, we tried to test the effects of rChTX‐Q18F on the learning, memory and emotion of mice. Through open field experiments, we found that rChTX‐Q18F could not modulate the distance of activity, time in the central area and average speed of the test mice. In Morris water maze test, rChTX‐Q18F also had no significant effects on the piercing frequency and target‐quadrant dwell‐time (Figure S13, n = 5).

3. MATERIALS AND METHODS

3.1. Expression, purification, and characterization

The hβ4‐loop (KCNMB4, GenBank accession no. AF207992.1, the extracellular loop of human BK β4 subunit, 45–166 aa) and its variants for NMR and binding studies were prepared as described in previous report. ³³ ChTX gene was commercially synthesized in Shanghai Shenggong Biotech Co., China and inserted into a home‐modified plasmid. MBP tag and His₆‐tag were used for soluble expression of ChTX in E coli and easy purification through Ni²⁺ sepharose resin column (GE health Co, USA), respectively. Thrombin cleaved these two tags, resulting in two extra residues ‐GS‐ left in the N‐terminus of ChTX. The purity and molecular weight (MW) of rChTX were confirmed by running SDS‐PAGE gel and MALDI‐TOF mass spectrum (the theoretical MW of rChTX was 4458.04 Da, while its experimentally measured MW was 4456.01 Da). Then, to characterize the overall folding of rChTX, 2D ¹H–¹H COSY, TOCSY and NOESY spectra were conducted on rChTX in 10% D₂O buffer (25 mM NaH₂PO₄, pH 6.80, 100 mM NaCl, 10% D₂O) and 100% D₂O NMR buffer on an Agilent Inova 600 NMR machine, which is equipped with pulsed field gradients and a triple resonances cryoprobe. Its solution structure was determined by common 2D NMR techniques described in previous report, which was similar to that of its native form.

3.2. NMR titration experiments

To verify whether hβ4‐loop can interact with rChTX, we performed NMR titration experiments at NMR buffer condition (25 mM NaH₂PO₄, pH 6.80, 100 mM NaCl, 10% D₂O) on an Agilent Inova 600 NMR spectrometer. The concentration of ¹⁵ N‐labeled hβ4‐loop was 0.42 mM. rChTX was titrated into hβ4‐loop solution at mole ratios (hβ4‐loop vs. rChTX) of 1:0, 1:0.3, 1:0.5, 1:1, 1:2, and 1:4, and then ¹H–¹⁵ N HSQC spectra were conducted. The cross‐peaks in HSQC spectrum at a mole ratio 1:2 were identical to those in the ¹H–¹⁵ N HSQC spectrum at a mole ratio 1:4, thus the complex sample for structural studies was prepared at a mole ratio of 1:2 (hβ4‐loop vs. rChTX).

3.3. Microscale thermophoresis measurements

To obtain binding affinity of rChTX (or its variants) to hβ4‐loop (or its variants), microscale thermophoresis (MST) assay was conducted through a NanoTemper Monolith NT.115 instrument (NanoTemper Technologies, Germany). MST is an immobilization‐free technique to analyze biomolecular interactions. ⁴⁴ In brief, 400 nM wild‐type hβ4‐loop or its variants labeled with NT‐495 fluorophore reagent (from Monolith Protein labeling Kit Blue‐NHS) was incubated with appropriate serial concentration gradient (ranging from 4 mM to 122 nM) of rChTX or its variants for 30 min at 25°C in a buffer of 25 mM NaH₂PO₄, pH 6.8, and 100 mM NaCl. Complex samples were loaded into capillaries treated with NanoTemper hydrophilic solution, and MST measurements were conducted at least three times at 25°C by 40% MST power and 40% excitation power, respectively. Finally, K _d values were determined by itting the data through NanoTemper Analysis (1.2.20 version).

3.4. Chemical cross‐linking assay

To confirm the interaction interfaces between hβ4‐loop and rChTX, we performed the chemical cross‐linking assay, combined with mass spectrometry analysis (CXMS). ⁴⁵ , ⁴⁶ The crosslinking reagent disuccinimidyl suberate (i.e., DSS, which usually links –NH₂ groups in proteins) was first solved in DMSO with a concentration of 20 mM. rChTX, hβ4‐loop and their complex (rChTX vs. hβ4‐loop 2:1) were incubated at 25°C for 4 h with an optimized concentration (1 mM) of DSS, respectively. The total reaction volume was 20 μl, and the final concentration of hβ4‐loop was 0.16 mM in a buffer (100 mM NaCl, 25 mM NaH₂PO₄, and pH 6.8). Subsequently, SDS_PAGE gel was run on the reaction products using non‐crosslinked rChTX and hβ4‐loop as negative controls. In‐solution tryptic proteolysis was further conducted and provided a mixture of loop‐links, inter‐links and mono‐links. After an enrichment step of the peptides being cross‐linked, they were subjected to LC–MS/MS, and the raw MS data was analyzed with pLink software against the database of hβ4‐loop and rChTX sequences. ⁴⁷

3.5. Structural model constructed by NMR

NMR samples include about 0.5 mM uniformly ¹³C,¹⁵ N‐labeled hβ4‐loop and 1 mM rChTX in an NMR buffer (25 mM NaH₂PO₄, 100 mM NaCl, pH 6.80 and 10% D₂O). NMR experiments were conducted at 23°C on an Agilent Inova 600 MHz, 800 MH_Z and Bruker Advance 900 MHz NMR spectrometers in National Center of Protein Sciences, Shanghai, China. To assign the chemical shifts of ¹H, ¹³C, and ¹⁵ N in the backbones and side‐chains of the residues of bound hβ4‐loop, and to collect NOEs, 2D ¹⁵ N‐edited and ¹³C‐edited HSQC, 3D HNCO, HNCA, CBCA(CO)NH, HCCH‐TOCSY, HBHA(CO)NH, HNCACB, HCCH‐COSY, ¹⁵ N‐resolved HSQC‐NOESY for amide protons related NOEs, ¹³C‐resolved HSQC‐NOESY for both aliphatic and aromatic protons related NOEs, and ¹⁵ N‐resolved HSQC‐TOCSY spectra were acquired, respectively. The proton signals of bound rChTX were assigned through 2D ¹⁵ N‐filtered, ¹³C‐filtered and J‐resolved TOCSY and NOESY spectra acquired on ¹⁵ N‐ and ¹³C‐ labeled hβ4‐loop with unlabeled rChTX. To assign NMR signals of the protons in free rChTX, 2D ¹H–¹H COSY, TOCSY and NOESY spectra were acquired on free rChTX in similar NMR buffer. Intermolecular NOEs between hβ4‐loop and rChTX were got by probing the differences between ¹⁵ N‐edited HSQC‐NOESY spectra with and without ¹⁵ N‐decoupling during acquisition, and between ¹³C‐edited HSQC‐NOESY spectra with and without ¹³C‐decoupling during acquisition. ³⁴ All NMR spectra were processed by NMRPipe ⁴⁸ and analyzed with Sparky (http://www.cgl.ucsf.edu/home/sparky/).

The structural calculation was carried out through XPLOR‐2.47 program (NIH version). ⁴⁹ NOE‐derived distance restraints were grouped into strong (1.8–2.9 Å), medium (1.8–3.5 Å), and weak (1.8–6.0 Å) ranges, respectively. Intermolecular NOEs were set as very weak ranges 1.8–8.0 Å. The dihedral angle constraints (i.e., phi and psi) were generated using program TALOS based on the backbone chemical shifts (HN, HA, CO, and CA) of hβ4‐loop. ⁵⁰ Ten iterations of calculation were conducted, where 50 structures were produced in the initial eight iterations. Then, 100 structures were calculated during the last two iterations. Then, 20 structures having the lowest energies were selected out to stand for the final 3D conformers, which did not have NOE violations larger than 0.3 Å and dihedral angle violations larger than 5°, respectively. The structural statistics were accessed by PROCHECK‐NMR ⁵¹ and PROCHECK ⁵¹ (Table S3). The figures about structural information were drawn with PyMOL (http://pymol.org/).

3.6. Circular dichroism spectra

To test the changes in the folding of hβ4‐loop, rChTX and their variants, circular dichroism (CD) spectra were performed at 25°C on JASCO‐715 spectropolarimeter (Jasco International Co., Tokyo, Japan). Data were acquired at 0.1 nm intervals, a bandwidth of 1‐nm, a scan speed of 20 nm/min, and an 0.25 s response time from 250 to 190 nm. The cells of circular quartz with 0.1 and 1 cm path lengths were employed for far‐UV regions. The intensities of CD signals were described as the molar residue ellipticities, which were given in the units of degrees cm² mol⁻¹. The concentrations of protein and rChTX were close to 0.5 mg/ml in ddH₂O.

3.7. Cell culture and transfection

The HEK293T cells were purchased from the cell bank of the Shanghai Institute of Biochemistry and Cell Biology (SIBCB), CAS (Chinese Academy of Sciences). The cells were cultured in DMEM (Dulbecco modified Eagle medium, Life Technologies, USA), which were supplemented with 10% heat‐inactivated FBS (fetal bovine serum, Life Technologies, USA). The dishes were incubated in 95% humidity environment containing, 5% CO₂ (carbon dioxide) at 37°C. The plasmids of potassium channel subtypes including human BKα subunit (KCNMA1, GenBank ID: U23767), BKβ4 subunit (KCNMB4, GenBank ID: U25138), hKv1.2 (KCNA2, GenBank ID: NM012970) and human Kv1.3 (KCNA3, GenBank ID: L23499) are gifts from Davies NW (University of Leicester, UK), Lippiat JD (Leeds university, UK) and Vassilevski AA (Russian Academy of Sciences, Russia), respectively. Before cell transfection, we transferred HEK 293 T cells to 24 well plates (Corning, USA). The cells, at 70–75% confluence, were co‐transfected with 0.5–1 μg potassium channel plasmid per well as well as 0.25 μg pEGFP‐N1 (Life Technologies, USA) per well by administering 1 μl P3000 and 0.75–1 μl Lipofectamine 3,000. The green fluorescence of EGFP was used to identify the transfection efficiency. ⁵²

3.8. Hippocampal neuron separation and primary culture

The hippocampi tissues were dissected from E18 embryonic rats, and enzymatically treated with 0.25% trypsin (Invitrogen, USA) for 10–15 min at 37°C, followed by gentle mixing. The digestion was stopped with and transferred to complete medium that is Dulbecco's modified Eagle's medium (DMEM, Invitrogen, USA) supplemented with 10% fetal bovine serum (FBS, Invitrogen, USA) and 10% F‐12 Nutrient Mixture (F12, Invitrogen, USA) and 1× penicillin–streptomycin (Invitrogen, USA). The tissue was triturated gently with a fire‐polished Pasteur pipet. Dissociated cells were plated at 5 × 10⁴ cells/ml on glass coverslips coated with poly d‐lysine. After 6 h, half of the medium was changed to serum‐free Neurobasal medium with 2% B27 supplement and 2 mM Glutamax‐I. Then, half of the culture medium was replaced with fresh medium every 3 days. Cytosine arabinoside (Invitrogen, USA) was added at 2–4 μM to inhibit glial proliferation. Cultures were maintained in a humidified incubator (5% CO₂ and 95% humidity) at 37°C for 8–14 days. ²⁷

3.9. Whole‐cell recordings

Whole‐cell patch‐clamp recordings were performed by the Axon Multiclamp 700B amplifier (Molecular Devices, USA) or EPC‐10 amplifier (HEKA Eletronik, Germany) as described previously. ⁵³ , ⁵⁴ Briefly, micropipettes from glass capillary tubes (VitalSense, China) were fabricated by P1000 (Sutter, USA) or PC‐10 micropipette Puller (Narishige, Japan), with controlling the resistance to 2–4 MΩ. Data acquisition and stimulation protocols were equipped with Patchmaster software (HEKA Eletronik, Germany) or pCLAMP 10 (Molecular Devices, USA). We omitted the cells of which the seal resistance (Rseal) is below 1 GΩ. To minimize the voltage errors, Rs (series resistance) of cells was compensated to 80–85%. We discarded the cells with an uncompensated Rs above 10 MΩ. The P/4 protocol was used when the leak subtraction was performed during the whole‐cell recording.

3.10. Current clamp recordings and action potential analysis

Rats were anaesthetized by isoflurane and transcardially perfused with an ice‐cold sucrose‐based saline solution: 110 mM sucrose, 60 mM NaCl, 3 mM KCl, 1.25 mM NaH₂PO₄, 28 mM NaHCO₃, 0.5 mM CaCl₂, 7 mM MgCl₂, 5 mM glucose, saturated with 95% O₂/5% CO₂. After decapitation, brains were removed and placed in the ice‐cold sucrose‐based solution. Hippocampal brain slices (350 μm horizontal sections) were cut using a VT1200 S vibratome (Leica Microsystems Inc., Bannockburn, IL). Slices were incubated in standard artificial cerebrospinal fluid (aCSF): 119 mM NaCl, 2.5 mM KCl, 1.2 mM NaH₂PO₄, 25 mM NaHCO₃, 2.5 mM CaCl₂, 1.3 mM MgCl₂, 10 mM glucose saturated with 95% O₂/5% CO₂ at 32°C for 30 min and then transferred to room temperature for at least 60 min prior to recordings. Somata of hippocampal pyramidal neurons in CA1 region were targeted for whole‐cell recording through borosilicate glass electrodes with a resistance of 4–6 MΩ. Current clamp recordings were considered to be accessible if the tested neurons had an input resistance more than 350 MΩ and a resting membrane potential around −70 mV. The tested neurons, clamped at −80 mV as a holding potential, were stimulated by injection of a step of positive current for 1,000 ms. The hippocampal brain slices from control group or rats pre‐treated with PTZ after 24 h exerted with rChTX and rChTX mutants at 10 min after application, respectively. Spike width was measured at half amplitude of each spike. The difference between the threshold value of spike and the minimum value of the voltage after the AP peak was considered as the measured fAHP size. The decay of AHP was calculated from the peak fAHP to the subsequent 5–10 ms by using an exponential function. The time between the peaks of the action potential was measured as the value of interspike interval.

3.11. Solutions and drugs

The external solution in voltage clamp recordings for BK channels contains: 5 mM KCl, 5 mM HEPES, 135 mM NaCl, 2.5 mM CdCl₂, 1.2 mM MgCl₂, and 10 mM glucose (pH = 7.4, adjusted by NaOH). The standard internal solution contained: 10 mM HEPES, 1 mM EGTA, 117 mM KCl, 2 mM MgATP, 2 mM MgSO₄, and 10 mM NaCl, (pH = 7.2, titrated with KOH), and 300 nM free Ca²⁺ buffered with 10 mM HEDTA, which was determined by calcium electrode. The bath solution for other potassium channels or endogenous Kv1.3 in microglial BV2 cells contains 135 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 1.8 mM CaCl₂, 10 mM HEPES, and 10 mM glucose (pH 7.4, adjusted by NaOH). The pipette solution contains: 130 mM KCl, 0.5 mM MgCl₂, 2 mM MgATP, 10 mM EGTA, and 10 mM HEPES (pH 7.3, adjusted by KOH). The external solution used in current clamp recordings contains: 10 mM HEPES, 2.5 mM KCl, 145 mM NaCl, 1.5 mM CaCl₂, 1.2 mM MgCl₂, and 10 mM glucose (pH = 7.4, adjusted by NaOH). The pipette solution consists: 4 mM MgCl₂, 1 mM CaCl₂, 140 mM KCl, 11 mM EGTA, and 10 mM HEPES (pH = 7.2, adjusted by KOH).

3.12. Animals

The healthy adult male SD (Sprague–Dawley) rats weighing 220–250 g were used in animal experiments, obtained from Experimental Animal Center of CAS in Shanghai. Rodents were kept in an animal culture room, controlled at 20–23°C and a 12 h light–dark cycle. Each five rats were placed in an individual in a cage for free access to water and food. Animal experiments performed in this study complied with ethical considerations of Shanghai University of Traditional Chinese Medicine's Animal Ethic Committee, which is in accordance with the criteria of National Research Council. All the animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Shanghai University of Traditional Chinese Medicine, performed in accordance with the relevant regulations as well as guidelines, which was also approved by the guidelines on ethical standards for investigation of conscious animals.

3.13. Surgery

The rats were anaesthetized by intraperitoneal injection (i.p.) of sodium pentobarbital (40 mg/kg body weight), and then placed in a stainless steel stereotaxic frame (Narishige, Tokyo, Japan). The head of testing rat was shaved and prepared in sterile fashion with povidone‐iodine. A midline incision was performed on the scalp in order to reveal the lambdoid, coronal and sagittal sutures. Then, a guide cannula for hippocampal region microinjection by using the Mini drug delivery pump was implanted at the right lateral dorsal hippocampus (AP‐4.3 mm posterior to bregma, L‐2.2 mm lateral to the midline of the skull and V‐2.5 mm ventral to the dura surface). ²⁷ For local field potentials (LFPs) tests, the recording electrode was also implanted at the same position. The reference wire for LFPs was positioned into the electrode bundle, and the earth electrode was positioned at the front of skull. Electrodes as well as guide cannula were fixed and the wound induced by implanted electrodes and catheters was sutured by dental cement. The tested rats were allowed to recover for at least 72 h. At the end of surgery, tested animals were killed after the brains of them were removed for cresyl violet staining according to the previously reported method. ²⁷ Only the rats with cannulae positioned accurately could be tested for data collection.

For knocking down the BK β4 subunits, rats were anaesthetized by intraperitoneal injection (i.p.) of sodium pentobarbital. After anesthetizing the rats, the microinjector entered the brain and stereotactically located at the hippocampal CA area with the speed of 1 mm/min. After stabilizing for 10 min, the rAAV2‐hSyn‐kcnmb4‐shRNA virus or rAAV2‐hSyn‐scramble‐pA virus (OBIO, China) with carrying the interference sequence of the target gene were injected at a rate of 0.1 μl/min, with a total dose of 200 nl and a titer of 1 × 10¹⁴ vg/ml.

3.14. Seizure behavior observation procedures

For pentylenetetrazole (PTZ)‐induced pre‐sensitized seizure, ²⁶ , ²⁷ tested rats were placed in a 40 × 30 × 50 cm transparent box for 1–2 h, then, anaesthetized in a short time by inhaling ether. rChTX (dissolved in 2 μl saline) or saline was injected into hippocampus through guide cannula by a microinjector. Ten minutes later, 60 mg/kg PTZ was exerted through intraperitoneal injection (i.p). to trigger epileptic seizures. Then, the animals were put back to the box, their behavior was continuously observed for 2 h. 24 h later, one group was injected with saline into the hippocampus. The rest of groups were injected with rChTX, rChTX variants or valproic acid (VPA, positive control), respectively. Then, PTZ was exerted again to induce pre‐sensitized seizures. ²⁷

Both behavior observation and administration of drugs were “double blinded” to eliminate the possibility of human error. The Racine's five‐point scale ⁵⁵ was used to classify seizure‐like animal behavior: Stage 0, no response; Stage 1, ear and facial twitching; Stage 2, myoclonic jerks without upright position; Stage 3, myoclonic jerks, upright position with bilateral forelimb clonus; Stage 4, clonic–tonic seizure; Stage 5, generalized clonic–tonic seizures and loss of postural control. In the first 2 h after PTZ injection, anticonvulsant effects of rChTX were evaluated based on the duration of seizures, latency, seizure numbers and severity stage of seizures. The latency is the average duration between administration of PTZ and the onset of the first epileptic seizure which was above stage 2. An episode of seizure is the duration from the seizure onset to the recovery from epileptic seizure. To quantify the numbers and duration, seizure‐like behavior was divided into four types according to Racine's five‐point scale. The minimal interval of two independent seizures was set for 5 s throughout the quantification of all seizure numbers. Duration and number of seizures were defined as the sums of the multiple seizures for each animal. Seizure severity was evaluated based on maximal seizure score graded from 0 to 5.

3.15. Western blotting

Immunoblotting was performed to assess the protein content. Briefly, the proteins were separated by electrophoresis on 10% SDS‐polyacrylamide gels and transferred onto polyvinylidene difluoride membranes (IPV00010, Millipore, Billerica, MA). The membranes were blocked with 5% skim milk (prepared in 1× Tris‐buffered saline Tween [TBST]) for 1 h at RT, and then incubated with primary antibodies against beta4 antibody (diluted 1:200, Alomone Labs Ltd., Israel) overnight, then washed with phosphate buffered saline plus tween‐20 (PBST, tween‐20 0.05%), and incubated for 1 h at room temperature with a horseradish peroxidase‐conjugated secondary antibody (goat anti‐rabbit IgG, 1:10000, Abcam, UK). Blots were visualized with Chemi DocTM XRS+ Imaging System (Bio‐RAD, USA).

3.16. Immunohistochemistry

Expression of c‐fos in hippocampus region was evaluated using avidin biotin complex (ABC) after 2 h of behavioral observations. Animals were anesthetized by intraperitoneal injection of pentobarbital sodium (60 mg/kg body weight), injected with 200 ml sterile saline through the left ventricle, and then perfused with 400 ml fixative solution containing 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) for 1–2 h. Brain tissues were taken out and fixed in the same fixative for 12 h, then moved to 20% sucrose solution and soaked until it sank to the container bottom. The tissues were also immersed in 30% sucrose solution until it sank to the bottom of the container. The brain tissues were sliced into 30 μm sections through a constant cold box slicer (CM1950, Leica, Germany) and pasted on the gelatine‐potassium chromium‐sulfate treated glass slides. To process immunohistochemistry, the frozen sections of brain tissue were repeatedly rinsed with 0.01 M PBS (pH 7.4) and incubated with rabbit polyclonal antibody targeting a peptide at N‐terminal of human c‐fos P62 (1:400, diluted in 0.01 M PBS containing 1% BSA, 0.03% sodium azide, 0.3% Triton X‐100, Sc‐52, Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) for 48 h at 4°C. The sections were incubated with biotinylated goat anti‐rabbit IgG (1:200, diluted in 0.01 M PBS, Vector, Burlingame, CA, USA) and avidin–biotin‐peroxidase complex (1:200, diluted in 0.01 M PBS, Vector, USA) at room temperature for 2 h. The nickel‐diaminobenzidine glucose‐oxidase method was used to visualize the reactive products. The c‐fos expression in rat hippocampus was measured by cell counting. The number of c‐fos immunoreactive (FLI) neurons in different sections of hippocampus (CA1, CA3) and dentate gyrus (DG) was calculated. Sections (n = 4–6) of each animal were randomly selected for FLI count, of which the average values were finally calculated. All data were collected blind to drug treatment of each group.

3.17. Nissl staining

In order to calculate the hippocampal neurons loss, the frozen sections were incubated in 0.1% acetic acid solution containing 0.5% cresol violet acetate for 30 min. After being cleared with distilled water and dewatered through 70%, 95% and 100% ethanol, the sections were cleared with xylene and cover‐slipped with DPX mounting medium (Sigma, USA). Histological images were captured with a digital camera (Nikon DS Ri1 camera, Nikon, Japan) coupled to a microscope (Nikon ECLIPSE 80i, Nikon, Japan) at the hippocampal CA and DG regions. Analysis of the neuron count was done based on the visual interpretation.

3.18. Local field potential recordings

The microarray electrode was implanted into the hippocampus region according to sixth edition atlas “The rat brain in stereotaxic coordinates” written by George Paxinos and Charles Watson. To determine the placement of electrodes: AP: 4.3 mm, MR: −2.2 mm, DV: 2.5 mm, with fixed 3–4 screws as the reference electrode was used into the blank area of the skull. The recording electrodes and the base of the electrodes were fixed with dental cement. The LFPs recording were performed after the tested rats were awake. The OmniPlex‐D multichannel recording system (Plexon, USA) was used to record the LFP signals as well as synchronized video. The head mount was connected to a preamplifier tethered to the analog‐digital converter box. The preamplifier was amplified 1000 times, the wave amplitude range was set to −2 to +2 V, the filtering range was set to 1.6–100 Hz. According to Nesquet's sampling theory, the recording frequency of LPFs, the high‐pass filtering and the low‐pass filtering were set to 1 kHz, 50 Hz, and 300 Hz, respectively. The single recording time is no <30 min continuously.

The results of LFPs recording were exported as a *.pl2 file format. The offline sorter v4 software was used for visualization preview. During LFPs analysis, same channel was selected through MATLAB (MathWorks, USA) program to export the recording data. The wavelet transform was used to decompose the signal of different LFP frequency, and get the physiological rhythm of different frequency (δ: ~0–4 Hz, θ: ~4–8 Hz, α: ~8–13 Hz:β: ~13–30 Hz, γ: ~30–100 Hz). ²⁷ , ⁵⁶ The welch method, fast Fourier transform method and hamming window were used to calculate the frequency domain information of LFPs in power spectrum (PS) analysis. The time domain information of energy change was calculated by weighted operation. The calculation formula of power spectral density (PSD) at off‐field potential were listed as follows.

$${\int }_{-\infty }^{+\infty }{x}^2\left(t\right)\mathit{dt}=\frac{1}{2\pi }{\int }_{-\infty }^{+\infty }{\left|X\left(\mathit{jw}\right)\right|}^2\mathit{d\omega }$$

$$\mathrm{P}=\underset{\mathrm{T}\to \infty }{\lim }\frac{1}{2T}{\int }_{-T}^T{x}^2\left(t\right)\mathit{dt}=\frac{1}{2\pi }{\int }_{-\infty }^{+\infty }\underset{T\to \infty }{\lim }\frac{1}{2T}{\left|X_T\left(\omega \right)\right|}^2\mathit{d\omega }$$

3.19. Morris water maze

The Morris water maze (MWM) is a circular pool with a diameter of 120 cm and a height of 60 cm, a water level of 35 cm, and a temperature of 22–24°C. The mice were trained on the MWM with one trial per day for 6 days. A platform was hidden 1 cm below the surface of the water, which was made opaque with white non‐toxic paint. Each trial lasted until the mouse found the platform or for a maximum of 60 s. At the end of each trial, mice were allowed to rest on the platform for 60 s. The time to reach the platform (latency), length of the swim path, and swim speed were recorded semi‐automatically using a video tracking system (Shanghai JiLian Web Tech Ltd.).

3.20. Open‐field testing

The open‐field test is used to measure anxiety‐like behavior in animals. This test used a camera to measure the movement of the test animal in the peripheral and central zones (20 × 20 × 20 cm) of a 42 × 42 × 42 cm polyvinyl chloride box for 10 min. A video tracking program (Stoelting Co., Wood Dale, IL) was used to record and measure the total distance traveled, the time spent in the center of the open‐field arena, and the distance moved in the center of the open‐field arena in each trial. The time spent in the center and distance traveled in the center of the arena were used as measures of anxiety‐like behavior.

3.21. Data analysis

The raw data was analyzed by OriginPro 8.5 (Origin Lab, USA). The data was counted and the results were described as means ± SEM with the number of tests, which were shown as n in the figure legends as well as results. Behavioral and the patch clamp recordings, LFP and behavioral tests were statistically analyzed by using one‐way ANOVA. Two sample Student's test was used if comparisons were restricted to two means. Error probabilities of p < .05, p < .05, or p < .001 were considered statistically significant.

The degree of rChTX effect was termed as fractional outward current (I _f ) using the remaining outward current, where each drug exposure was considered as a fraction of the outward current magnitude of the patch prior to the first drug exposure. Dose–response curve underlying the inhibitory percent of rChTX on outward currents of BK channel was drawn according to the Hill equation I = I _m /(1 + ([toxin]/IC₅₀) ⁿ ). In this equation, I _m stands for maximum enhanced percentage of outward currents of BK, and [toxin] represents the concentration of rChTX or its mutants. IC₅₀ (half‐maximal inhibitory concentration) and n denote the rChTX concentration of half‐maximal inhibitory effect and the Hill coefficient of dose–response curve, respectively.

4. CONCLUSION

In a word, based on the structural model of rChTX with hβ4‐loop and other electrophysiological data, rChTX‐Q18F was successfully designed to selectively inhibit the outward currents of BK(α + β4) channel. Seizure cells treated with rChTX‐Q18F had a significantly long interspike interval with decrease of spike numbers, and could not continue to fire APs at a large current injection. Moreover, rChTX‐Q18F displayed anticonvulsant effects on PTZ‐induced pre‐sensitized rats, modulatory effects on field potential signals, inhibitory effects on hippocampal c‐Fos expression, and neuroprotective effects on hippocampal neurons after PTZ‐induced seizures, respectively. Although rChTX‐Q18F might not be permanent through blood brain barrier, its injection into intracerebroventricular or intrathecal spaces could be achieved by microinfusion pumps.

AUTHOR CONTRIBUTIONS

Xinlian Liu made rChTX samples, performed biochemical studies. Jie Tao, Shuzhang Zhang, Yu Yao, Guoyi Li, and Yonghua Ji did or validated electrophysiological experiments. Chunxi Wang and Wenxian Lan helped analyse NMR and MST data. Hongjuan Xue guided NMR experiments. Chunyang Cao conceptualized whole project, supervised the experiments and wrote manuscript.

CONFLICT OF INTEREST

The potential usage of charybdotoxin Q18F variant as an AED has been applied Chinese patent with an application number 202110969029.0.

Supporting information

Appendix S1: Supporting Information

Liu X, Tao J, Zhang S, Lan W, Yao Y, Wang C, et al. Development of charybdotoxin Q18F variant as a selective peptide blocker of neuronal BK(α + β4) channel for the treatment of epileptic seizures. Protein Science. 2022;31(12):e4506. 10.1002/pro.4506

DATA AVAILABILITY STATEMENT

Data available on request due to privacy/ethical restrictions.