Hg1, Novel Peptide Inhibitor Specific for Kv1.3 Channels from First Scorpion Kunitz-type Potassium Channel Toxin Family*

Background: The potassium channel inhibitory activity of scorpion Kunitz-type toxins has not yet been determined. Results: We identified the first scorpion Kunitz-type potassium channel toxin family with three groups and seven members. Conclusion: A novel peptide, Hg1, specific for Kv1.3 channel, was found. Significance: Kunitz-type toxins are a new source to screen and design potential peptides for diagnosing and treating Kv1.3-mediated autoimmune diseases. The potassium channel Kv1.3 is an attractive pharmacological target for autoimmune diseases. Specific peptide inhibitors are key prospects for diagnosing and treating these diseases. Here, we identified the first scorpion Kunitz-type potassium channel toxin family with three groups and seven members. In addition to their function as trypsin inhibitors with dissociation constants of 140 nm for recombinant LmKTT-1a, 160 nm for LmKTT-1b, 124 nm for LmKTT-1c, 136 nm for BmKTT-1, 420 nm for BmKTT-2, 760 nm for BmKTT-3, and 107 nm for Hg1, all seven recombinant scorpion Kunitz-type toxins could block the Kv1.3 channel. Electrophysiological experiments showed that six of seven scorpion toxins inhibited ∼50–80% of Kv1.3 channel currents at a concentration of 1 μm. The exception was rBmKTT-3, which had weak activity. The IC50 values of rBmKTT-1, rBmKTT-2, and rHg1 for Kv1.3 channels were ∼129.7, 371.3, and 6.2 nm, respectively. Further pharmacological experiments indicated that rHg1 was a highly selective Kv1.3 channel inhibitor with weak affinity for other potassium channels. Different from classical Kunitz-type potassium channel toxins with N-terminal regions as the channel-interacting interfaces, the channel-interacting interface of Hg1 was in the C-terminal region. In conclusion, these findings describe the first scorpion Kunitz-type potassium channel toxin family, of which a novel inhibitor, Hg1, is specific for Kv1.3 channels. Their structural and functional diversity strongly suggest that Kunitz-type toxins are a new source to screen and design potential peptides for diagnosing and treating Kv1.3-mediated autoimmune diseases.

Kunitz-type toxins are an ancient and multifunctional toxin family, which have been found in various animal venoms, such as snake, lizard, cattle tick, cone snail, spider, sea anemone, and scorpion (14 -18). Members of this family usually have 50 -70 residues cross-linked by two or three disulfide bridges. Structurally, almost all Kunitz-type toxins adopt the conserved structural fold with two antiparallel ␤-sheets and one or two helical regions (19 -21). Functionally, many Kunitz-type toxins have protease and/or potassium channel inhibiting properties. For example, Kunitz-type toxin bungaruskunin, isolated from snake venom, is a serine protease inhibitor (22), but ␣-dentrotoxin, ␦-dentrotoxin, dentrotoxin K, and dentrotoxin I, also from snake venom, are potent Kv1.1 channel inhibitors (21). Kunitz-type toxins HWTX-XI from spider and APEKTx1, AKC1, AKC2, and AKC3 from sea anemone are bifunctional toxin peptides with both protease and potassium channel-inhibiting properties (20,23,24). Conkunitzin-S1, a 60-residue cone snail Kunitz-type toxin cross-linked by two disulfide bridges, interacts with the shaker potassium channel (19,25). From scorpion venoms, three Kunitz-type toxins, Hg1, SdPI, and SdPI-2 have been isolated, but only SdPI was found to inhibit trypsin (26,27). Until now, the potential potassium channel inhibitory activity of scorpion Kunitz-type toxin has not been determined.
To identify novel peptide inhibitors specific for Kv1.3 channels, we screened scorpion Kunitz-type toxins and evaluated their pharmacological activities for potassium channels. By cDNA cloning, bioinformatic analyses, and functional evaluations, we identified the first scorpion Kunitz-type potassium channel toxin family composed of four new members (LmKTT-1c, BmKTT-1, BmKTT-2, and BmKTT-3) and three known members (LmKTT-1a, LmKTT-1b, and Hg1) (26,27). In addition to their functions as trypsin inhibitors, six of the recombinant scorpion Kunitz-type toxins also block 50 -80% of Kv1.3 currents at a concentration of 1 M. The exception was rBm-KTT-3, which had weak activity. Among these peptides, a specific Kv1.3 inhibitor Hg1 was discovered with an IC 50 value of ϳ6.2 Ϯ 1.2 nM. Significantly different from classical Kunitztype potassium channel toxins with the N-terminal region as the channel-interacting interface, Hg1 adopted the C-terminal region as the main channel-interacting interface. Our results describe the first scorpion Kunitz-type potassium channel toxin family, and the identification of the specific Kv1.3 inhibitor Hg1. Kunitz-type toxins are a new group of toxins that can be used to screen and design potential peptides for diagnosing and treating Kv1.3-mediated autoimmune diseases.
Expression and Purification of Scorpion Kunitz-type Toxins-We used the cDNA sequences of LmKTT-1a, LmKTT-1b, and LmKTT-1c from Lychas mucronatus venom gland cDNA libraries and BmKTT-1, BmKTT-2, and BmKTT-3 from scorpion B. martensii cDNA libraries as the templates for PCR to generate respective fragments (27). The PCR product of Hg1 was generated by overlapping PCR. All PCR products were digested with NdeI and XhoI and inserted into expression vector pET-28a. After confirmation by sequencing, the plasmids were transformed into competent Escherichia coli BL21(DE3) cells for expression. QuikChange site-directed mutagenesis kits (Stratagene, Santa Clara, CA) were used for generating the mutants based on the wild-type plasmid pET-28a-Hg1. All mutant plasmids were verified by DNA sequencing before expression. Kunitz-type toxins and mutants were expressed according to our previous protocol (26). For example, the recombinant LmKTT-1a was found to accumulate exclusively in inclusion bodies and was refolded in vitro. Renatured protein was finally purified by HPLC on a C 18 column (10 mm ϫ 250 mm, 5 m Dalian Elite). Peaks were detected at 230 nm. The fraction containing recombinant LmKTT-1a was eluted at 20 -21 min and further analyzed by MALDI-TOF-MS (Voyager-DESTR, Applied Biosystems).
Determination and Modeling of Scorpion Kunitz-type Toxin Structures-The secondary structures of scorpion Kunitztype toxins and mutants with a control peptide BPTI were analyzed by circular dichroism (CD) spectroscopy. All samples were dissolved in water at a concentration of 0.2 mg/ml. Spectra were recorded at 25°C from 250 to 190 nm with a scan rate of 50 nm/min, on a Jasco-810 spectropolarimeter (Jasco Analytical Instruments, Easton, MD). The CD spectra were collected from averaging three scans after subtracting the blank spectrum of water. The three-dimensional structure of Hg1 was modeled using BPTI (PDB 4 code 6PTI) as a template through the SWISS-MODEL server as we have described previously (28).
Molecular Docking-Molecular docking of Hg1 interacting with the Kv1.3 channel was carried our as previous computational approaches (13,29). First, the structure of the Kv1.3 channel was modeled using KcsA (PDB code 1K4C) as a template. Second, molecular docking was performed on the modeled Hg1 peptide and Kv1.3 channels using the ZDOCK program (30), and the docking results were then filtered by scoring combined with detailed mutagenesis information; Finally, a reasonable Hg1-Kv1.3 complex that was consistent with the experimental alanine-scanning mutagenesis was screened out.
Serine Protease Inhibitory Activity Assay-The inhibitory activities of seven Kunitz-type toxins were tested in the presence of serine proteases as described previously (26). Trypsin (bovine pancreatic trypsin; EC 3.4.21.4), ␣-chymotrypsin (bovine pancreatic ␣-chymotrypsin; EC 3.4.21.1), elastase (porcine pancreatic elastase; EC 3.4.21.36), and their respective chromogenic substrates, Na-benzoyl-L-arginine, 4-nitroanilide hydrochloride, N-succinyl-Ala-Ala-Pro-Phe-p-nitroanilide, Nsuccinyl-Ala-Ala-Ala-p-nitroanilide, were purchased from Sigma (Sigma-Aldrich). Trypsin was incubated with various amounts of rHg1 (100 to 400 nM) for 30 min at a final concentration of 400 nM. The reactions were initiated by adding varying concentrations of substrate, ranging from 0.1 to 0.8 mM. The initial rate of p-nitroanilide formation was monitored continuously at 405 nm for 5 min at 25°C. The inhibitory activity of rHg1 was estimated by setting the initial velocity obtained with only protease as 100%. The inhibitory constant (K i ) of the trypsin-inhibitor complex was determined by Lineweaver-Burk plots and further slope replotting. The methods for chymotrypsin and elastase assay were the same, except the final concentration of chymotrypsin was 100 nM.
Electrophysiological Recordings-The cDNAs encoding mKv1.1, hKv1.2, mKv1.3, and hSKCa3 were provided generously by professor Stephan Grissmer (University of Ulm, Ulm, Germany) and professor George Chandy (University of California, Irvine, CA). We saved the mBKCa channel. All of the channels were subcloned into the pIRES2-EGFP vector (Clontech, Mountain View, CA) and transformed into HEK293 cells. The whole-cell patch clamp was used to measure and record the channel currents according to a previously described procedure (13,31). Peptides were dissolved in stock solutions containing 1% BSA and diluted into solutions containing 0.01% BSA for toxin application in electrophysiological experiments.
Results are shown as mean Ϯ S.E., with n being the number of experiments. The significance between two means was calculated with the Student's t test using Origin software (version 6.0, Microcal Software, Northampton, MA). Differences in the mean values were considered significant at probability Ͻ0.05. Using IGOR software (WaveMetrics, Lake Oswego, OR), concentration versus response relationships were fitted according to the modified Hill equation: I toxin /I control ϭ 1/1 ϩ ([toxin]/ IC 50 ), where I is the peak current, and [toxin] is the concentration of toxin. The parameters to be fitted were concentration at half-maximal effect (IC 50 ).

RESULTS
Primary Structures of Scorpion Kunitz-type Toxins-On the basis of our cDNA libraries and further random sequencing, four new genes encoding Kunitz-type toxins were obtained. Three were isolated from scorpion B. martensii, which were named BmKTT-1, BmKTT-2, and BmKTT-3, and one was isolated from scorpion L. mucronatus and named LmKTT-1c. Combined with the three known Kunitz-type toxins, Hg1, LmKTT-1a, and LmKTT-1b (26,27), seven scorpion Kunitztype toxins can be classified into three groups according to their disulfide bridge patterns (Fig. 1). Hg1 and BmKTT-3 belong to the first group, which adopted classical disulfide pairings similar to HWTX-XI toxin from spider (20), dendrotoxin K from snake (32), and APEKTx1 from sea anemone (23). LmKTT-1a, LmKTT-1b, LmKTT-1c, and BmKTT-1 belong to the second group, which adopted a unique cystine framework that we described in our previous work (26). These toxins lack the normal CysII-CysIV disulfide bonds present in BmKTT-3 from group I but contain two new cysteine residues near the C terminus of the mature peptide. Different from all known Kunitztype toxins from various animals, BmKTT-2 in the third group has eight cysteine residues, which may adopt novel disulfide bonds (17). Our findings demonstrate the molecular diversity of scorpion Kunitz-type toxins.

Preparation and Structural Analysis of Scorpion Kunitz-type
Toxins-To evaluate the function of scorpion Kunitz-type toxins, we obtained seven recombinant toxins as described previously (26,27). Expression and purification of LmKTT-1a toxin were as follows. A His 6 tag and a thrombin cleavage site were fused to LmKTT-1a at the N terminus. Inclusion bodies of LmKTT-1a fusion peptide were suspended in LB medium and refolded successfully in vitro. The soluble protein was further separated by RP-HPLC and SDS-PAGE (supplemental Fig. S1,  A and B). By MALDI-TOF MS, the molecular weight was 8657.8 Da, which is in good agreement with the calculated molecular mass of 8658.6 Da (supplemental Fig. S1C).
BPTI is a classical Kunitz-type peptide (33). By circular dichroism spectroscopy, all seven recombinant scorpion Kunitz-type toxins were found to have similar secondary structures to BPTI (supplemental Fig. S2), which suggests the conserved structures of Kunitz-type peptides.

FIGURE 2. Inhibitory effects of scorpion Kunitz-type toxins on trypsin.
A, inhibitory effects of rHg1 peptide on trypsin with BPTI and BSA as controls. B, inhibitory effects of seven scorpion Kunitz-type toxins at different concentrations on trypsin using the same conditions. Data represent the mean Ϯ S.E. of at least three experiments. chymotrypsin, and elastase. All seven recombinant toxins exhibited apparent inhibitory effects on trypsin (Fig. 2), but no inhibitory effect on chymotrypsin or elastase, even at higher concentrations. Recombinant LmKTT-1a, LmKTT-1b, LmKTT-1c, BmKTT-1, and Hg1 could completely inhibit the trypsin activity at the ratio of 1:1 with a dissociation constant of 140, 160, 124, 136, and 107 nM, respectively (supplemental Fig.  S3). The rBmKTT-2 could completely inhibit the trypsin activity at a ratio of ϳ1.5:1, with a dissociation constant of 420 nM, and rBmKTT-3 could inhibit ϳ85% of the trypsin activity at a ratio of ϳ4:1, with a dissociation constant of 760 nM (Table 1).
Hg1, Selective Inhibitor for Kv1.3 Channels-Based on the pharmacological properties of rHg1, the first potent Kv1.3 channel inhibitor with a Kunitz-type fold, we further investigated its effects on different types of potassium channels. As shown in Fig. 4, rHg1 could inhibit Ͻ50% of the Kv1.1 and Kv1.2 channel currents at a concentration of 1 M (Fig. 4, A and  B), and had little effect on SKCa3 and BKCa channel currents at identical concentrations (Fig. 4, C and D). These data indicate that Hg1 is a selective peptide inhibitor specific for Kv1.3 channels.  Unique Molecular Mechanism of Hg1 Blocking Kv1.3 Channels-By using the alanine-scanning strategy, we investigated the molecular mechanism of Hg1 toxin blocking Kv1.3 channels. As shown in Fig. 5, A-H, there were no apparent effects of His-2, His-3, Asn-4, Arg-5, Leu-9, Leu-10, and Lys-13 residues on the toxin pharmacological activities, whereas there were less conformational changes for toxin mutants (Fig. 5, I and J). These data indicate that Hg1 toxin did not use N-terminal res-idues to inhibit the Kv1.3 channel, which was different from the known molecular mechanism of Kunitz-type toxins such as ␦-dendrotoxin and HWTX-XI. These toxins mainly use N-terminal residues to block Kv1.1 channels (20,34). Among classical animal toxins affecting potassium channels, basic residues as critical residues are common features (35)(36)(37). We then focused the second cluster of basic residues, located at the C terminus al region of Hg1 toxin (Fig. 1) and found the dominant effects of Lys-56, Arg-57, Phe-61, and Lys-63 residues on the toxin affinities when using identical 100 nM concentration of wild-type and mutant Hg1 toxin with less conformational changes (Fig. 6, A-E). The IC 50 values were 582.0 Ϯ 184.5 nM for Hg1-K56A, 305.0 Ϯ 93.7 nM for Hg1-R57A, 360.1 Ϯ 116.5 nM for Hg1-F61A, and 457.9 Ϯ 187.3 nM for Hg1-K63A mutants. Replacement by alanine reduced the ability of the toxin to inhibit Kv1.3 channels by ϳ94-, 49-, 58-, and 74-fold, respectively (Fig. 6F). These structure and function relationships demonstrate that Hg1 toxin mainly uses the C-terminal region as the channel-interacting interface to inhibit Kv1.3 channels.
To further reveal the recognition mechanism of Hg1 peptide toward Kv1.3 channels, a structural model of the Hg1-Kv1.3 complex was obtained through our previous computational approaches (13,34). The importance of four Lys-56, Arg-57, Phe-61, and Lys-63 residues was indicated by the structural analysis. Lys-56 is the pore-blocking residue, which is surrounded by the pore region residues from Kv1.3 channels within a contact distance of 4 Å (Fig. 7A). Phe-61 mainly contacts Pro-280, His-307, Val-309, and Lys-314 residues of the channel A chain and Asp-387 residue of channel D chain by hydrophobic interactions (Fig. 7B). Lys-63 mainly contacts Ser-378, Ser-379, Gly-380, and Asn-382 residues of the channel D chain (Fig. 7C), and Arg-57 mainly contacts Pro-377, Ser-378, Ser-379, His-404, Pro-405, and Val-406 of the channel D chain (Fig. 7D). The structural model of the Hg1-Kv1.3 complex supported the inference that Hg1 mainly used the C-terminal as the channel-interacting surface.

DISCUSSION
The Kv1.3 potassium channel is an attractive pharmacological target for autoimmune diseases, and specific peptide inhibitors are useful tools for diagnosing and treating these diseases (38,39). To screen and design the potent and selective peptide inhibitors, efforts to improve peptide specificity are continuing (11)(12)(13). Kunitz-type toxins are a kind of ancient toxin family that has been identified in many animal venoms, such as those of snake, cone snail, spider, sea anemone, and scorpion (14 -18). Several Kunitz-type toxins have been found to inhibit potassium channels (Table 1).
In this work, we adopted a new strategy to screen novel peptide inhibitor specific for Kv1.3 channels from scorpion Kunitztype toxins. We identified the first scorpion Kunitz-type potassium channel toxin family with three groups and seven members, from which a novel peptide inhibitor, Hg1, specific for Kv1.3 channels, was obtained. Overall, our work provided following unique structural and functional features of the Kunitz-type potassium channel toxin family.
Molecular Diversity of Kunitz-type Potassium Channel Toxins-Combined with the known Kunitz-type toxins affecting potassium channels (Table 1) (21), 17 toxins were found to block potassium channels. Among these, there were significant differences in toxin sequences, sequence lengths, and number and distribution of cysteine residues (20). Most notably, BmKTT-2 toxin from scorpion was found to form four disulfide bridges, which is different from all known Kunitz-type animal toxins (17). The finding of seven addi- tional scorpion Kunitz-type toxins strongly enriched the molecular diversity of Kunitz-type toxins inhibiting potassium channels.
Conserved Structures of Kunitz-type Toxins-Studies have shown that the backbone of dendrotoxin I, a potassium channel blocker from snake venom, superimposes on BPTI with an root mean square deviation of Ͻ1.7 Å (33). Using circular dichroism spectroscopy analyses, we also found structural similarity between BPTI peptide and seven scorpion Kunitz-type toxins, especially with different disulfide bridges (supplemental Fig.  S2). The structural similarity of Hg1 and BPTI was shown by the BPTI structure (PDB code 6PTI) and the Hg1 structural model  . Differential binding interfaces of Kunitz-type toxins blocking potassium channels. A, sequence alignments of Kunitz-type potassium channel toxins Hg1 from scorpion, ␦-DTX from snake, and HWTX-XI from spider. B, the main functional residues of Kunitz-type toxin ␦-DTX (structure modeled with DTX-K as template, PDB code 1DTK) interacting with the Kv1.1 channel. C, the main functional residues of Kunitz-type toxin HWTX-XI (PDB code 2JOT) interacting with the Kv1.1 channel. D, the main functional residues of Kunitz-type toxin Hg1 (structure modeled with BPTI as template, PDB code 6PTI) interacting with Kv1.3 channels.
(supplemental Fig. S4). This indicates that structures were conserved for Kunitz-type toxins.
Functional Diversity of Kunitz-type Potassium Channel Toxins-Previous reports showed several Kunitz-type toxins could affect Kv1.1, Kv1.2, and Shaker channels (Table 1). In this work, we found that scorpion Kunitz-type toxins could inhibit Kv1.3 channels. Except for the scorpion Kunitz-type toxin rBmKTT-3, the other six toxins inhibited ϳ50 -80% of the Kv1.3 currents at a concentration of 1 M (Fig. 3). The IC 50 value of rHg1 for Kv1.3 channels was 6.2 Ϯ 1.2 nM. The functional diversity of Kunitz-type potassium channel toxins will hopefully encourage a functional evaluation of this kind of toxins on additional potassium channels.
In conclusion, we have characterized the first scorpion Kunitz-type potassium channel toxin family with the unique pharmacological property of blocking Kv1.3 channels. From this toxin family, a potent and selective Kv1.3 channel inhibitor, Hg1, was identified. Hg1 is the first Kunitz-type toxin identified that interacts with potassium channels by its C-terminal region as the main channel interacting interface. The structural and functional diversity of these Kunitz-type potassium channel toxins may provide a new source of potassium channel inhibitors used for the diagnosis and treatment of autoimmune disorders.