Design and Characterization of a Highly Selective Peptide Inhibitor of the Small Conductance Calcium-activated K+Channel, SkCa2*

Apamin-sensitive small conductance calcium-activated potassium channels (SKCa1–3) mediate the slow afterhyperpolarization in neurons, but the molecular identity of the channel has not been defined because of the lack of specific inhibitors. Here we describe the structure-based design of a selective inhibitor of SKCa2. Leiurotoxin I (Lei) and PO5, peptide toxins that share the RXCQ motif, potently blocked human SKCa2 and SKCa3 but not SKCa1, whereas maurotoxin, Pi1, Tsκ, and PO1 were ineffective. Lei blocked these channels more potently than PO5 because of the presence of Ala1, Phe2, and Met7. By replacing Met7 in the RXCQ motif of Lei with the shorter, unnatural, positively charged diaminobutanoic acid (Dab), we generated Lei-Dab7, a selective SKCa2 inhibitor (K d = 3.8 nm) that interacts with residues in the external vestibule of the channel. SKCa3 was rendered sensitive to Lei-Dab7 by replacing His521 with the corresponding SKCa2 residue (Asn367). Intracerebroventricular injection of Lei-Dab7 into mice resulted in no gross central nervous system toxicity at concentrations that specifically blocked SKCa2 homotetramers. Lei-Dab7 will be a useful tool to investigate the functional role of SKCa2 in mammalian tissues.

they promote calcium entry in response to a mitogenic stimulus (7). SK Ca channels have also been implicated in fasciculations in denervated skeletal muscle (8) and in myotonic dystrophy (9). SK Ca channels, products of three phylogenetically related genes SKCa1-3, are found in a variety of tissues including the nervous system (10), liver (11), skeletal and smooth muscle (8,12), adrenal medulla (13), and lymphoid tissues (14,15). In the brain (16), the precise functional role of each channel in specific neuronal pathways has been difficult to elucidate because of the absence of selective blockers. The identity of the specific SK Ca channel(s) responsible for apamin-induced destruction of cerebellar Purkinje neurons (17) and altered seizure threshold (18,19) and for apamin and Lei neurotoxicity (20) is also unclear. The development of inhibitors that target each of these channels selectively would facilitate studies to determine their specific roles in mammalian tissues.
The SKCa2 channel is expressed in the brain (10,16), retina (21), liver (22), melanocytes (expressed sequence tag AA418096), fetal heart (expressed sequence tag AA418000), and human Jurkat T cell line (7,22,23). Precise determination of function has only been possible in Jurkat T lymphocytes in which this channel plays a role in calcium signaling (7,14,24). The existing blockers of SKCa2 channels are not adequately specific to distinguish this channel from other SK Ca subtypes. Apamin, a peptide from bee venom, exhibits only 10-fold selectivity for human SKCa2 over human SKCa1 or SKCa3 (3,22,25,26), while the bisquinolinium cyclophane UCL-1684 blocks SKCa2 and SKCa1 with roughly equivalent potency (25) although it shows some selectivity over SKCa3 (7). The peptide toxin Lei (also known as scyllatoxin) from the scorpion Leiurus quinquestriatus hebraeus, currently the most specific inhibitor of SKCa2, exhibits ϳ200-fold selectivity for human SKCa2 (15) over SKCa1 (25), although its affinity for SKCa3 had not been determined.
This study describes the guided design and electrophysiological characterization of a novel Lei analog that selectively blocks SKCa2 homotetramers with low nanomolar affinity. Our strategy to design a specific inhibitor of the SKCa2 channel exploited an approach used previously to develop selective blockers of Kv1.3 and IKCa1 channels (27,28). In this approach, a potent yet non-selective inhibitor of the target channel is the starting template. Specificity for the desired channel is engineered into the peptide on the basis of experimentally determined differences in its interaction with the target channel and other channels. Using Lei and a closely related peptide PO5 as templates, and by comparing the potencies of the native and mutant toxins for SKCa1, SKCa2, and SKCa3, we delineated the channel-binding surfaces of these toxins. Lei-Dab 7 was designed to target a structural feature unique to the Lei-SKCa2 interaction surface.
Clones and Mutants-The cloning of human SKCa3 containing 19 polyglutamines in the N terminus (GenBank TM AF031815, AJ251016) and IKCa1 (GenBank TM AF033021) has been reported previously (30 -32). The full coding region of SKCa1 was amplified from human brain total RNA using reverse transcriptase polymerase chain reaction (PCR) with an engineered 5Ј HindIII site near the start codon and a 3Ј BamHI site near the termination codon and was cloned in-frame into the pEGFP-C3 vector (CLONTECH, Palo Alto, CA) to create GFP-SKCa1. PCR was used to generate mutant SKCa3 channels (28). PCR products were digested with KpnI and BamHI and cloned into KpnI and BamHIcut GFP-SKCa3. All clones were verified by sequencing. DNA for transfection was prepared with the QIAGEN (Valencia, CA) Miniprep kit. Human SKCa2 in pcDNA3 was a generous gift from Dr. Bernard Attali (Tel Aviv University, Sackler School of Medicine, Israel).
Transfection of Constructs into Mammalian Cells-COS-7 cells were plated in culture chambers (5 ϫ 10 5 cells/chamber), and 12-24 h later, cells were transiently transfected using FuGene TM 6 (Roche Molecular Biochemicals) with the respective DNA in serum-free OptiMEM medium (Life Technologies, Inc.) as per the manufacturer's recommended protocol. GFP-positive cells were used for electrophysiological studies at 48 h following transfection. Typical transfection efficiencies using this protocol were 40 -70%. PC12 cells were plated overnight on glass coverslips prior to use. Cell lines stably expressing mKv1.1, mKv1.3, and hKv1.5, hSlo, and RBL cells expressing endogenous rKir 2.1 were used for the selectivity screen as described previously (28).
Electrophysiology-Cells were studied in the whole cell configuration of the patch clamp technique. The holding potential in all experiments was Ϫ80 mV. For measurement of IK Ca , SK Ca , and BK Ca currents, we used an internal pipette solution containing (in mM) 145 potassium aspartate, 2 MgCl 2 , 10 HEPES, 10 K 2 EGTA, and 8.5 CaCl 2 (1 M free Ca 2ϩ ), pH 7.2, 290 -310 mOsm. K V currents were recorded with a fluoride-based internal solution. To reduce currents from native chloride channels in COS-7, sodium aspartate Ringer was used as an external solution (in mM) 160 sodium aspartate, 4.5 KCl, 2 CaCl 2 , 1 MgCl 2 , and 5 HEPES, pH 7.4, 290 -310 mOsm. For Jurkat T cells, potassium aspartate Ringer was used as an external solution with K ϩ instead of Na ϩ (164.5 mM K ϩ ). SK Ca currents were elicited by 200-ms voltage ramps from Ϫ120 to 40 mV applied every 10 s, and the reduction of slope conductance at Ϫ80 mV by the toxin was taken as a measure of channel block. BK Ca currents were elicited by 200-ms voltage ramps from Ϫ80 to 80 mV applied every 30 s, and channel block measured at 35 mV. The inward rectifier (rKir 2.1) in RBL cells was studied in sodium aspartate Ringer with a potassium aspartate-based pipette solution containing 50 nM free Ca 2ϩ .

Selective in Vivo Blockade of SKCa2 Channels by Lei-Dab 7 in Mice-
Lei-Dab 7 was administered to 25-g C57/BL6 mice via the intracerebroventricular route, and the LD 50 was determined (33). Groups of four mice per dose were injected with 5 l of the peptide solution containing 0.1% (w/v) bovine serum albumin and 0.9% (w/v) sodium chloride.

RESULTS
Lei and PO5 Are Potent Inhibitors of Human SK Ca Channels-Six peptide toxins from scorpion venom have been identified as SK Ca channel blockers based on 125 I-apamin displacement studies on rat brain synaptosomes (34). Fig. 1 shows the sequence alignment of these toxins (35), three of which (Lei, PO5, and Pi1) contain a motif (RXCQ) reported to be important for binding to SK Ca channels. (20,36). Apamin contains an RRCQ sequence (Fig. 1), which has a spatial arrangement similar to that of the RMCQ motif in Lei (Fig. 2).
Lei is reported to block the well characterized human SKCa2 channel in Jurkat T cells with picomolar potency (15,22,23). As the first step in our strategy to design a specific SKCa2 inhibitor, we compared the potency of Lei with that of PO5, a toxin that differs from Lei only at positions 1, 2, 7, and 24 ( Fig.  1), in blocking Jurkat SKCa2. Symmetric internal and external K ϩ solutions were used to induce the inward component of the SKCa2 current. Following break-in with 1 M Ca 2ϩ in the pipette solution, SKCa2 currents were seen at negative potentials. At potentials more positive than Ϫ40 mV, K ϩ currents were mainly due to Kv1.3 with minimal contribution from SKCa2 (Fig. 3A, left). The SKCa2 component was blocked by Lei and PO5, whereas the Kv1.3 current was unaffected (Fig.  3A, left). Dose-response curves showed that Lei blocked Jurkat SKCa2 with K d values consistent with published data on the cloned and native channel (15,22,23) and with ϳ100-fold greater potency than PO5 (Fig. 3B and Table I were next evaluated on cloned SKCa1 and SKCa3 expressed in COS-7 cells. SKCa1 (Fig. 3A, middle) and SKCa3 (Fig. 3A, right) K ϩ currents elicited with 1 M Ca 2ϩ in the pipette solution reversed at Ϫ80 mV in the presence of external sodium Ringer. Lei and PO5 blocked SKCa3 but were ineffective on SKCa1 (Fig. 3A and Table I). Comparison of the dose-response curves of SKCa3 showed Lei to be 25-fold more effective than PO5 (Fig. 3B and Table I). For reasons that remain unclear, four other scorpion toxins (maurotoxin, Pi1, PO1, and Ts) reported to be highly active in 125 I-apamin displacement assays (20, 34, 36) had little or no blocking activity on SKCa2 or SKCa3 (Table I). Thus, the two most potent scorpion peptides, Lei and PO5, exhibited significantly different blocking potencies on SKCa2 and SKCa3 despite differing at only four positions.
Ala 1 , Phe 2 , and Met 7 in Lei Are Responsible for the Enhanced Potency of Lei over PO5-To define the residues responsible for the increased affinity of Lei for SK Ca channels, we replaced each of the four differing residues in PO5 with the corresponding residue in Lei. PO5-V24D blocked SKCa2 and SKCa3 with a potency comparable with native PO5 (Fig. 4A). PO5-T1A, PO5-V2F, and PO5-R7M blocked SKCa2 with potencies approaching that of Lei. The improved potency of all three PO5 mutants to nearly that of Lei suggests that any of the indicated . Only one concentration of PO5 was tested on SKCa1 (shown). Typical SKCa1-3 currents were studied in the whole cell configuration of the patch clamp technique. Recordings were done with 1 M free calcium as the internal pipette solution, and currents were elicited by voltage ramps from Ϫ120 to 40 mV. SKCa1 and SKCa3 were expressed transiently in COS-7 cells. An external solution containing sodium aspartate (5 mM potassium aspartate) was used for the recordings, and the degree of block was measured as the decrease in slope conductance at Ϫ80 mV. Jurkat T-cells were used to assess the effect on endogenous hSKCa2 with symmetric internal and external potassium aspartate (165 mM potassium aspartate).

TABLE I
Comparison of competition data on rat brain synaptosomes of indicated toxins (20,34,36)  alterations in the side chains at these positions may allow PO5 to fit more tightly within the channel-binding pocket due to either shorter side-chain size or local change in backbone conformation (37,38). Similar results were obtained with these PO5 mutants on SKCa3 (Fig. 4B). The reverse mutation in Lei (Lei-M7R) reduced potency on both SKCa2 and SKCa3 (data not shown). These results indicate that Ala 1 , Phe 2 , and Met 7 underlie the increased affinity of Lei over PO5 for SKCa2 and SKCa3. These three residues form a localized binding pocket that may represent an important contact point with SK Ca channels (Fig. 5, A and B). Lei-Dab 7 Is a Highly Selective Inhibitor of SKCa2-Positions 6 and 7 of Lei are part of the conserved RXCQ motif (Fig. 1). To define the role of the residues in this motif, a series of Lei mutants was made at positions 6 and 7 to probe the toxinchannel interaction. Charge-neutralization mutations at position 6 that retained size (Arg 6 3 leucine or citrulline) reduced toxin potency 70 -180-fold for both channels (Fig. 6), indicating the need for a charged residue at this position. Substitution of the unbranched lysine at position 6, a mutation that retained charge but decreased size, also reduced toxin affinity for both channels (20 -35-fold), whereas the introduction of the positively charged, bulky branched unnatural amino acid, homoarginine, caused a 1000-fold decrease in toxin potency (Fig.  6). Thus, substitutions at position 6 are not tolerated, and the arginine has the optimum size, charge, and branching required for interaction with SK Ca channels. Furthermore, when residues at positions 6 and 7 were exchanged, the Lei-R6MϩM7R double mutant was considerably less potent than the native toxin on SKCa2 and SKCa3 (K d ϭ 9.5 and 65 nM, respectively), establishing the importance of the relative locations of Arg 6 and Met 7 in the RXCQ motif.
We next turned our attention to position 7 in the RXCQ motif and found our first promising lead. Introduction of positively charged lysine at position 7 (Lei-M7K) yielded a mutant that blocked SKCa2 35-fold more potently than SKCa3 (Fig. 7). In an attempt to further enhance this difference, we generated two additional Lei mutants in which Met 7 was replaced by smaller positively charged unnatural amino acids diaminopropionate (Lei-Dapa 7 ) and diaminobutanoate (Lei-Dab 7 ). Lei-Dapa 7 blocked Jurkat SKCa2 ϳ350-fold more potently than SKCa3, whereas Lei-Dab 7 was ϳ650-fold more effective. Similar results were obtained with the cloned SKCa2 channel expressed in COS-7 cells (Table II). Lei-Dab 7 was also ineffective against hS-KCa1, hIKCa1, hSlo, K V , and Kir channels, establishing its specificity for SKCa2 (Table II).
Asn 367 in the SKCa2 Pore Region Is Important for Lei-Dab 7 Selectivity-Because Lei-Dab 7 and apamin share the critical RXCQ channel-binding motif (Figs. 1 and 2), it is likely that Lei-Dab 7 , like apamin (39), binds to residues in the external S5-Pore-S6 region of the channel. Human SKCa2 and SKCa3 differ at only two positions in the pore region (Fig. 8A). To determine whether one or both these residues contribute to Lei-Dab 7 selectivity (Fig. 8, B and C), we replaced these two residues in SKCa3 (Val 485 and His 521 ) with the corresponding residues of SKCa2 (Ala 331 and Asn 367 ), individually or together. Lei-Dab 7 blocked SKCa3-H521N (K d ϭ 20 Ϯ 4.7 nM) and SKCa3-V485A ϩ H521N (K d ϭ 7.5 Ϯ 1.2 nM) with nearly the same potency as SKCa2 (Fig. 8, D and E, and Table II), whereas SKCa3-V485A did not produce functional channels. In mutant cycle studies with the SKCa3-H521N and SKCa3-V485AϩH521N mutants, residue 7 of Lei-Dab 7 was found to couple tightly with His 521 (⌬⌬G ϭ 2.4), suggesting that these two residues lie in close proximity to each other. We conclude that Lei-Dab 7 binds to the external vestibule of SK Ca channels and that Asn 367 (the residue corresponding to His 521 of hS-KCa3) in SKCa2 contributes to Lei-Dab 7 selectivity.
Selective in Vivo Blockade of Homotetrameric SKCa2 Channels by Lei-Dab 7 Does Not Cause Gross Neurotoxicity-To evaluate the central nervous system effects of specific in vivo blockade of SKCa2 homotetramers, Lei-Dab 7 was administered via the intracerebroventricular route to mice. At a concentration (10 ng (or 300 nM assuming a brain liquid volume of 10 l)) that would selectively block Ͼ99% of SKCa2 homotetramers, no gross central nervous system toxicity was  observed. At a higher concentration (50 ng (or 1500 nM)), all six animals became hyperexcitable, developed convulsions followed by paralysis lasting 15 h, and then fully recovered. At much higher concentrations (80 ng (or 2400 nM)), 50% of the animals died within 2 h, and 100% lethality was observed in 1 h at 100 ng (3 M). The neurotoxicity observed at Lei-Dab 7 concentrations higher than 1500 nM may be due to the blockade of SKCa1 and/or SKCa3 homotetrameric channels (Table II), although we cannot exclude the contribution of heteromultimeric SK Ca channels containing SKCa2 subunits.

DISCUSSION
Functional SK Ca channels are tetramers of SKCa1-3 subunits that can assemble as homo-or heterooligomers (39). Heterooligomers are likely to exist natively given the overlapping patterns of distribution of SKCa1-3 in the central nervous system. SKCa1 and SKCa2 are abundant in the hippocampus, whereas the medial habenula has both SKCa2 and SKCa3 (10,16). The available SK Ca channel inhibitors, including apamin (3,22,25,26) and the UCL compounds (7,25), do not have the requisite selectivity to define the specific functional roles of the homo-and heterooligomeric SK Ca channels present in mammalian tissues. In this study, we used a structure-guided approach to develop Lei-Dab 7 , a novel peptide that blocks SKCa2 homotetramers with nanomolar potency and exhibits 650-fold or greater selectivity over related homooligomeric channels. Heterooligomeric SK Ca channels may exhibit intermediate sensitivity for Lei-Dab 7 compared with their homotetrameric counterparts as has been shown with apamin on recombinant heterotetramers of SKCa1 and SKCa2 (39).
Lei-Dab 7 might help elucidate the contributions of SKCa2 channels in mediating neuronal afterhyperpolarization (3)(4)(5) and in the apamin-induced destruction of cerebellar Purkinje neurons (17) and altered seizure threshold (18,19). Lei-Dab 7 may also help define the role of SKCa2 in the retina, liver, melanocytes, immune system, and fetal heart. The subunit composition of SK Ca heterooligomers in diverse mammalian tissues may also be determined by using Lei-Dab 7 in much the same way as radiolabeled peptide toxins selective for a particular K V subunit have proven useful in assessing the subunit composition of brain heteromeric K V channels (40 -42).
Peptide inhibitors of SK Ca channels form a structurally distinct group and lack the conserved dyad present in peptide blockers of K V , IKCa1, and BK Ca channels that consists of an aromatic residue with a neighboring invariant lysine (43). Therefore, the architecture of the toxin-binding surface in the external vestibule of SK Ca channels may differ from that of K V (44 -47) and IKCa1 (28,48) channels, which are similar to that of the crystallographically defined structure of the bacterial K ϩ channel, KcsA (49,50). The topologies of K V and IKCa1 channels were deduced from toxin mapping studies by identifying multiple contact points between toxins of known structure and these channels. Lei-Dab 7 may similarly be used as a molecular caliper to delineate the architecture of the external vestibule of SKCa2 . Molecular models of SK Ca channels based on complementary mutagenesis studies would facilitate the development of selective and potent inhibitors of SKCa1 and SKCa3.