Tetrapandins, a New Class of Scorpion Toxins That Specifically Inhibit Store-operated Calcium Entry in Human Embryonic Kidney-293 Cells*

Venoms from 14 snakes and four scorpions were screened for inhibitory activities toward store-operated Ca 2 (cid:1) entry (SOCE) in human embryonic kidney-293 cells. An inhibitory activity was found in venom from the African scorpion Pandinus imperator . The active agent of this venom was purified by gel filtration and reverse-phase high pressure liquid chromatography methods. Sequence information on the purified fraction, by automatic Edman degradation and mass spectrometry analysis, identified the activity as being contained in two tetrapeptides, which we have named tetrapandins. We demonstrate that synthesized tetrapandins have inhibitory activity for SOCE in human embryonic kidney-293 cells while having no effect on either thapsigargin-or carbachol-stimulated release of Ca 2 (cid:1) stores. These toxins should be extremely useful in future studies to determine downstream events regulated by SOCE as well as to determine whether multiple pathways exist for thapsigargin-stimulated Ca 2 (cid:1)

A large number of receptors utilize Ca 2ϩ as a second messenger to initiate downstream physiological processes. In the case of G protein-coupled receptors, the Ca 2ϩ response generally is biphasic with an initial rapid spike of Ca 2ϩ , due to release of internal Ca 2ϩ stores, followed by a longer, more sustained response. In some cases, the sustained response takes the form of a stable Ca 2ϩ plateau, while in other cases sustained Ca 2ϩ oscillations are observed; both the Ca 2ϩ plateau and sustained Ca 2ϩ oscillations are dependent on a continued influx of external Ca 2ϩ . Generally this Ca 2ϩ influx is mediated by non-voltage-gated channels (1)(2)(3)(4), and a great deal of effort is being exerted to characterize these channels and the proteins responsible for forming these channels. There appear to be multiple types of non-voltage-gated Ca 2ϩ channels stimulated by agonists of G protein-coupled receptors, one of which, the store-operated channels (SOCs), 1 mediates what was originally referred to as capacitative Ca 2ϩ entry (5) but now often called store-operated Ca 2ϩ entry (SOCE). These channels are activated in response to depletion of internal Ca 2ϩ stores. While there are a number of plausible theories on the mechanism by which store depletion activates these channels (6,7), there is still no complete model for this process. Likewise some progress has been made in identifying members of the TRPC channel family as the likely protein mediators of SOCE, but there is still much controversy in the literature concerning which of the TRPC family members are store-operated and which are receptor-operated. In addition, there is some question as to whether more than one subtype of SOCE might exist (8) as well as a relative dearth of information concerning the downstream consequences of SOCE.
One limitation in the characterization of SOCs and the investigation of the downstream consequences of SOCE has been the lack of an inhibitor with high affinity and high specificity for SOCs. While a number of substances have been used to inhibit SOCE, these agents in general lack the specificity required to fully characterize these channels and to determine unequivocally which downstream physiological events are regulated by Ca 2ϩ entering via SOCs. For example, SKF96365 has frequently been used to inhibit SOCE and to investigate events regulated downstream of SOCE (9). However, SKF96365 has been demonstrated to inhibit Na ϩ channels (10), K ϩ channels (11), and maitotoxin-induced Ca 2ϩ entry (12,13) and to facilitate nicotinic receptor desensitization (14). Other agents utilized as inhibitors of SOCs, such as carboxyamidotriazole (15) and 2-aminoethoxydiphenyl borate (2-APB) (16), have also been demonstrated to have inhibitory effects on other channels and transport molecules with carboxyamidotriazole being reported as an inhibitor of voltage-gated Ca 2ϩ channels (17) and 2-APB being reported to inhibit InsP 3 -mediated Ca 2ϩ release (18 -20), K ϩ currents (21), sarcoplasmic reticulum Ca 2ϩ -ATPase (22), gap junctions (23), and Ca 2ϩ -induced Ca 2ϩ release from isolated mitochondria (24) and even to activate a novel calcium-permeable cation channel (25). Given all of the side effects of existing inhibitors of SOCs, there is a clear need for a potent, specific inhibitor of SOCE.
One fertile source of ion channel inhibitors has been the venom of poisonous snakes, scorpions, spiders, and sea snails. Peptide toxins have been purified from complex venoms that block Na ϩ channels (26), K ϩ channels (27), voltage-gated Ca 2ϩ channels (28), and receptor-operated channels such as the Nmethyl-D-aspartate receptor (29). From the utilization of these peptide toxins, much has been learned about the subunit composition of these channels, their mode of activation and inactivation, and the downstream events regulated by ion movements through these channels. To date, toxins for non-voltagegated channels such as SOCs have not been identified. Clearly the discovery of a peptide toxin specific for SOCs would allow a level of study of these channels that has not been possible with current inhibitors. In this report, we describe our search for a peptide toxin in venoms of snakes and scorpions, the purification of the toxin from the Pandinus imperator scorpion venom, the identification of the toxin sequence, and the synthesis and functional effect of the peptide toxins we have named tetrapandins. These toxins should be extremely useful in future studies to determine downstream events regulated by SOCE as well to determine whether multiple pathways exist for thapsigarginstimulated Ca 2ϩ entry.
Venom Purification Procedure-All snake species, kept alive at the Kentucky Reptile Zoo (Slade, KY), were milked manually by having them bite into a rubber membrane, which was stretched over a glass funnel. The venom was collected in a glass vial, and the harvest was immediately frozen after extraction and lyophilized within 1 week. The powdered venom from each species was dissolved in 1 ml of double distilled water and kept at Ϫ20°C.
Scorpions of the species P. imperator, kept alive in the laboratory, were milked by electric stimulation, and their venom was dissolved in twice distilled water and centrifuged at 15,000 ϫ g for 15 min. The supernatant was freeze-dried and kept at Ϫ20°C until used. When required, samples of the soluble venom were loaded onto a Sephadex G-50 column (medium size, from Amersham Biosciences) and eluted with 20 mM ammonium acetate buffer, pH 4.7, following earlier published results (30). The active fraction (see below) was further separated using a semipreparative C18 reverse-phase column (Vydac, Hisperia, CA) of an HPLC system using a linear gradient from solution A (0.12% trifluoroacetic acid in water) to 60% solution B (0.10% trifluoroacetic acid in acetonitrile). The active fraction was further subjected to HPLC separation using an analytical Vydac C18 reverse-phase column eluted with the same gradient. Two pure peptides were obtained from this column and were further characterized.
Amino Acid Sequence and Mass Spectrometry Analysis-The peptides obtained from the HPLC system were sequenced using automatic Edman degradation in a ProSequencer apparatus (Beckman LF3000) following the protocol described by the manufacturer. The molecular weights of purified components were determined by means of a Finnigan LCQ Duo ion trap mass spectrometer (San Jose, CA). Confirmation of sequence and molecular masses were obtained using a MALDI-TOF MS apparatus (Voyager DE-PRO, Applied Biosystems, Framingham, MA). Peptides were crystallized using the dried droplet technique and ␣-cyano-4-hydroxycinnamic as matrix. Postsource decay fragment ion spectra were acquired after isolation of the appropriate precursor by using timed ion selection. Fragment ions were refocused onto the detector by stepping the voltage applied to the reflectron and acquiring distinct spectral segments using an Acquiris digitizer at a digitization rate of 500 MHz. Individual spectra were superimposed by using the Data Explorer Version 4.0 software (Applied Biosystems).
Peptide Synthesis and Aggregation-The chemical synthesis of the peptides was performed by the solid phase method of Merrifield (31) using Fmoc (N-(9-fluorenyl)methoxycarbonyl)-derivatized amino acids. NovoSyn TGA resins (600 mg, 0.126 meq) were used for each peptide. Once the peptides were synthesized, they were detached from the resins, and the corresponding protecting groups of the residues were eliminated by means of reactive K (84% trifluoroacetic acid, 5% phenol, 5% H 2 O, 5% thioanisole, 1% dithiothreitol) for 1 h. Deprotected peptides were precipitated with t-butyl methyl ether, washed three times with ether, redissolved in 20% acetic acid, and lyophilized as described. After synthesis, the peptides were purified by HPLC using the same conditions described for purification of the native peptides (see above). Their structures were confirmed by mass spectrometry analysis. The exact amino acid sequence for each one of the peptides was confirmed by dissolving them to a final concentration of 500 pmol/5 l of 50% acetonitrile with 1% acetic acid and directly applying them into the Finnigan LCQ Duo ion trap spectrometer using a Surveyor MS syringe pump delivery system. The eluate at 10 l/min was split to allow only 5% of the sample to enter the nanospray source. The spray voltage was set to start at 1.7 kV, and the capillary temperature was set at 115°C. For MS/MS experiments, the fragmentation source was operated with 25-35V of collision energy and 35-45% (arbitrary units) of normalized collision energy, and the scan was operated with wide band activated. All spectra were obtained in the positive ion mode. The data acquisition and the deconvolution of data were performed with the Xcalibur (Sequest) software. The MS/MS spectra were analyzed manually.
The synthetic peptide was dissolved in either HBSS or Ca 2ϩ -free HBSS (pH 7.4) and then filter-sterilized. The sterile peptide solutions were either used directly, or they were mixed for 2 days at room temperature on a Glas-Col rotator (Fisher) and then used for calcium imaging experiments.
Ca 2ϩ Imaging-The intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) was measured in cells loaded with the fluorescent indicator fura-2. Cells were plated onto 25-mm coverslips 1 day before the experiment. On the next morning, cells were washed twice with HEPES-buffered HBSS, loaded for 30 min with 5 M fura-2 AM that was dissolved in HBSS supplemented with 1 mg/ml bovine serum albumin ϩ 0.025% Pluronic F-127, and then unloaded in HBSS for another 30 min. The coverslips were mounted as the bottom of a chamber that was placed on the stage of a Nikon Diaphot inverted epifluorescence microscope equipped with a Fluor 10ϫ objective. Cells in the chamber were perfused via an eight-channel syringe system. A suction pipette maintained a constant volume of solution (ϳ0.5 ml) in the chamber.
An InCyt IM2 TM fluorescence imaging system (dual wavelength fluorescence imaging system, Intracellular Imaging Inc., Cincinnati, OH) was used to measure [Ca 2ϩ ] i during the experiment. Excitation light from a xenon light source was alternately passed through 340 and 380 narrow pass filters mounted in a Sutter filter wheel (Lamda 10-C). The 510 nm emissions were captured by a cooled CCD camera (Cohu 4915). The images were transmitted to a computer and processed with the imaging software InCyt IM2 TM Version 4.6. [Ca 2ϩ ] i was calculated by measuring the ratio of the two emission intensities for excitation at 340 and 380 nm. Calcium standard solutions, which were prepared with fura-2 potassium salt, were used to create a graph of fluorescence ratio (F 340 /F 380 ) as a function of Ca 2ϩ concentration (nM). This graph was then used to convert fluorescence ratios in an experiment to calcium concentrations. In experiments in which Ba 2ϩ influx was measured, the data are reported as the ratio (F 340 /F 380 ) since the Ba 2ϩ calibrations curves for fura-2 differ from the calibration curve for Ca 2ϩ .
During the experiment, an averaged response of ϳ800 cells from a single field on each coverslip was represented as one trace. For initial experiments, intracellular Ca 2ϩ stores were depleted in the presence of extracellular Ca 2ϩ , and the effect of the toxin was monitored by measuring the amount of reduction of the Ca 2ϩ plateau. In later experiments where Ba 2ϩ entry was monitored as a measure of the level of SOCE, the thapsigargin-stimulated Ba 2ϩ entry was obtained by subtracting the slope of the Ba 2ϩ leak (before stimulation) from the slope of Ba 2ϩ influx (after stimulation) for each coverslip.
Nominally Ca 2ϩ -free HBSS was prepared by stirring Ca 2ϩ -free, Mg 2ϩ -free, and HCO 3 Ϫ -free HBSS with Chelex-100 beads. After filtering out the Chelex-100 beads, MgCl 2 was added to a final concentration of 1 mM.

RESULTS
Our initial step was to screen a variety of snake and scorpion venoms in search of an inhibitory effect on SOCE in HEK-293 cells. This was done by investigating the effect of the various venoms on thapsigargin-stimulated Ca 2ϩ entry. HEK-293 cells were stimulated with 1 M thapsigargin in the presence of HBSS. After an initial peak of Ca 2ϩ , the cells establish a stable Ca 2ϩ plateau phase that is rapidly dissipated upon the removal of external Ca 2ϩ , indicating that the plateau is strictly dependent on continued influx of external Ca 2ϩ . Various dilutions of snake and scorpion venoms were added during the Ca 2ϩ plateau phase to test whether an inhibition of Ca 2ϩ entry occurred. We screened a total of 14 snakes (Agkistrodon contortrix contortrix, Agkistrodon contortrix mokeson, Crotalus durissus culminatus, Crotalus durissus cumanensis, Crotalus durissus terrificus, Daboia russelli russelli, Ophiophagus hannah, Naja haje, Naja kaouthia, Naja naja, Naja nivea, Naja pallida, Naja sputatrix, and Bothrops jararaca) and four scorpion (Hadrurus gertschi, Centruroides limpidus limpidus, Centruroides noxius, and P. imperator) venoms and found effects from only one snake venom and one scorpion venom. We undertook an in-depth investigation of the inhibitory agent present in the scorpion venom. The effects of various dilutions of the venom from the African scorpion P. imperator are shown in Fig.  1A. Increasing doses of the P. imperator venom resulted in an enhanced inhibition of the thapsigargin-induced Ca 2ϩ plateau. At this point, we were concerned that the reduction in Ca 2ϩ levels might be due to membrane depolarization that resulted from inhibition or activation of some other ion channel by the venom. Although SOCs are not directly regulated by membrane depolarization, the level of SOCE can be affected by a reduction of the driving force following membrane depolarization. To determine whether this was the case, we investigated whether the addition of valinomycin, which locks the membrane in a hyperpolarized state, would eliminate the effect of the P. imperator venom. The data in Fig. 1B show that the addition of valinomycin led to a slight increase in plateau Ca 2ϩ levels but did not block the effect of the P. imperator venom on Ca 2ϩ plateau levels. We also investigated the effect of the P. imperator venom on thapsigargin-stimulated Ba 2ϩ entry to rule out the possibility that the venom is working on Ca 2ϩ extrusion mechanisms rather than on Ca 2ϩ entry pathways. Since Ba 2ϩ is not pumped by Ca 2ϩ -ATPases (32,33), an inhibition of Ba 2ϩ entry by the venom would indicate that the effect is on a Ca 2ϩ entry pathway. We observed that the venom had a significant inhibitory effect on Ba 2ϩ entry (venom data not shown, but data for purified toxin shown later).
The initial step in characterization of the active agent in the P. imperator venom was to fractionate the venom on a gel filtration column. The Sephadex G-50 column separated several families of components based on molecular sizes (Fig. 2). Tubes corresponding to five subfractions (I-V) were pooled and lyophilized. They were all screened to determine their activity for reducing the thapsigargin-stimulated Ca 2ϩ plateau levels. We observed that only fraction V from the filtration column had an effect on the thapsigargin-stimulated Ca 2ϩ plateau phase. The data in Fig. 3 illustrate the effects of several doses of fraction V (F5) on the Ca 2ϩ plateau level. We performed addi- tional experiments in the presence of valinomycin and found that fraction V still inhibited Ca 2ϩ plateau levels even in the presence of this hyperpolarizing agent (data not shown, but data for purified toxin shown later).
Fraction V was further fractionated by running samples on a reverse-phase HPLC column and eluting samples with increasing strengths of acetonitrile as described under "Experimental Procedures." At least 35 distinct components were resolved by this chromatographic step (Fig. 4) that were collected as 13 individual subfractions for bioassay. Inhibition of the thapsigargin-stimulated Ca 2ϩ plateau phase was observed only with subfraction 7 (Fig. 4, asterisk). Effects of three dilutions of subfraction 7 (SF7) are shown in Fig. 5 where increasing doses of subfraction 7 produced more extensive reduction of the Ca 2ϩ plateau phase.
Subfraction 7 was then submitted to automatic Edman degradation, giving the amino acid sequence Leu-Trp-X-X where X was a mixture of amino acids (Lys, Ser, Gly, and Thr). This was an indication of the presence of more than one peptide in this subfraction 7. For this reason another chromatographic separation was conducted (see Fig. 4, inset), allowing the separation of two peptides for which the sequence was confirmed by postsource decay experiments using a MALDI-TOF mass spectrometer. Two distinct sequences were finally obtained. Peptide 1 eluting at 23.27 min had the amino acid sequence Leu-Trp-Ser-Gly with a molecular mass of 461.5 atomic mass units, and peptide 2 eluting at 23.82 min had the amino acid sequence Leu-Trp-Lys-Thr with a molecular mass of 546.6 atomic mass units. These short peptides are in small abundance in the venom. Peptide 1 is ϳ0.30%, whereas peptide 2 is only 0.17% of the total soluble venom. Significant amounts of peptide analogues containing N-formylkynurenine, an amino acid derivative resulting from Trp oxidation, were also detected in each fraction.
Since the amount of native peptides purified from the venom was too low for an extensive biological characterization and because we knew the amino acid sequences of the peptide toxins, we decided to synthesize both peptides. Approximately 30 mg were obtained for each peptide toxin. They were further separated by HPLC (data not shown) giving a yield of 60% for peptide 1 and 72% for peptide 2. The correct amino acid sequence of the synthetic peptides was confirmed by MS/MS analysis using the ion trap spectrometer as described under "Experimental Procedures." The effects of the synthetic peptides were investigated on SOCE.
Initial studies demonstrated that both peptides had inhibi-tory activities in terms of reducing the thapsigargin-stimulated Ca 2ϩ plateau phase (data not shown). However, the potency of the peptides seemed to be somewhat low. Data for inhibition by peptide 2 showed that a dose of 1 M was required to see measurable inhibition, and maximum inhibition was not observed until a dose of 20 M was reached (Fig. 6A). Since subfraction 7 from the HPLC separation yielded two small peptides, we explored the possibility that these peptides might be associated into a complex that was responsible for the inhibitory activity in the P. imperator venom. We mixed peptide 1 and peptide 2 in a tube and submitted them to constant rotation for various periods of time ranging from hours to days and then tested the potency of the mixture at various time periods. We found that after 2 days of constant rotation a substantially lower dose of peptide was required for inhibition of SOCE. However, this observation was not unique to the mixture of peptide 1 and peptide 2 as we found that placing either peptide 1 alone or peptide 2 alone under similar conditions resulted in a substantial increase in their potency. The data in Fig. 6B show a representative dose response for peptide 2 treated under such conditions. With this material, one can observe a slight inhibitory effect at a peptide dose as low as 1 nM and a maximum effect at a peptide dose of 1 M. At this maximum dose the Ca 2ϩ plateau is reduced by ϳ50%.
To confirm that the modified peptide is inhibiting Ca 2ϩ entry and not stimulating Ca 2ϩ pumps, its effect on Ba 2ϩ entry was measured. For this experiment, we incubated the cells in HBSS and then switched to a Ba 2ϩ -containing, Ca 2ϩ -free medium to obtain a brief measure of Ba 2ϩ leak flux. The cells were then switched to 0 Ca 2ϩ medium followed by 0 Ca 2ϩ with 1 M thapsigargin to deplete internal stores. Following the return of Ca 2ϩ to basal levels, 2 mM Ba 2ϩ was again added to monitor total Ba 2ϩ entry. The SOCE was determined by subtracting the Ba 2ϩ leak from the total Ba 2ϩ entry. As is shown in Fig. 7A, the peptide at a dose of 1 M inhibited Ba 2ϩ entry by ϳ50%. A statistical analysis of a series of experiments at a peptide toxin concentration of 1 M is shown (Fig. 7B). The slope of Ba 2ϩ entry in the control was 0.00072 Ϯ 0.00004 s Ϫ1 (n ϭ 13), and in the presence of modified peptide toxin 2 (1 M) it was 0.00038 Ϯ 0.00002 s Ϫ1 (n ϭ 11). These values are significantly different (p ϭ 0.00005). To determine whether the peptide toxin was having its inhibitory effect by depolarizing the cells and thereby decreasing the driving force for Ca 2ϩ entry, we ran similar experiments in the presence of valinomycin, which locks the membrane potential in a hyperpolarized state. We found that the addition of valinomycin slightly increased the thapsigargin-stimulated Ba 2ϩ entry, consistent with the predicted effect of membrane hyperpolarization. However, in the presence of valinomycin the modified synthetic peptide 2 still inhibited SOCE by a comparable percentage (Fig. 8). In the presence of valinomycin, Ba 2ϩ entry increased to 0.00084 Ϯ 0.00001 s Ϫ1 (n ϭ 4); the addition of modified peptide toxin 2 (1 M) in the presence of valinomycin reduced Ba 2ϩ entry to 0.00046 Ϯ 0.00006 s Ϫ1 (n ϭ 4). The difference in these values is statistically significant (p ϭ 0.03).
To determine the specificity of the peptide toxin, we investigated its effect on another type of Ca 2ϩ entry in HEK-293 cells.
We determined the effect of the peptide toxin on Ca 2ϩ entry stimulated by OAG. While in our previous studies (34) both TRPC1 and TRPC3 appear to mediate SOCE, in later studies we observed that only TRPC3 mediates OAG-stimulated Ca 2ϩ entry (35). Following addition of 1 M modified peptide toxin 2 to cells stimulated with OAG, no effect on the stimulated Ca 2ϩ entry was observed (Fig. 9). Thus, the peptide toxin appears to be highly specific for Ca 2ϩ entry activated by Ca 2ϩ store depletion. As mentioned in the Introduction, 2-APB not only inhibits SOCE but also inhibits release of Ca 2ϩ from internal stores. Thus, the question arose as to whether tetrapandins would also inhibit Ca 2ϩ release from intracellular stores. To investigate this question, we monitored either thapsigargin-stimulated or CCh-stimulated Ca 2ϩ release from intracellular stores. Control cells were incubated for 30 min in HBSS and washed quickly in Ca 2ϩ -free HBSS, and then thapsigargin was added in a Ca 2ϩfree HBSS medium to release Ca 2ϩ from intracellular stores. In parallel, cells were preincubated for 30 min in HBSS containing either 100 M 2-APB or 1 M tetrapandin 2 (both represent maximum doses for inhibition of Ca 2ϩ entry) and washed quickly in Ca 2ϩ -free HBSS, and then thapsigargin was added in Ca 2ϩ -free HBBS in the continued presence of the inhibitors. As can be seen in Fig. 10A, the thapsigargin-induced release of Ca 2ϩ was not modified by preincubation with tetrapandin 2 (right panel), while preincubation of cells with 2-APB blocked a substantial amount of the Ca 2ϩ release (center panel). The   FIG. 3. Inhibition of thapsigarginstimulated Ca 2؉ entry by a partially purified fraction of the P. imperator venom. Fraction V collected from the gel filtration column was diluted in HBSS, and its effect on thapsigargin (TG)-stimulated Ca 2ϩ entry was determined. As described in Fig. 1, once the stable calcium plateau was established, different dilutions of the fraction V (F5) were added, and its effect on the Ca 2ϩ plateau was monitored as a measurement of its effect on Ca 2ϩ entry.
FIG. 4. HPLC separation of fraction V of the P. imperator venom. A 1.0-mg sample of fraction V protein from the Sephadex G-50 column was applied to a semipreparative C18 reverse-phase column, eluted with a linear gradient from solution A (0.12% trifluoroacetic acid in water) to 60% solution B (0.10% trifluoroacetic acid in acetonitrile), and run for 60 min. The active fraction (labeled with an asterisk) was further separated on an analytical C18 reverse-phase column eluted with the same gradient; its profile is shown in the inset of the figure (note that only the first section of the gradient is shown). Two pure peptides were obtained from this column and were further characterized.
average peak height for the thapsigargin control was 143 Ϯ 5.8 nM (n ϭ 8) compared with 143 Ϯ 13.9 nM (n ϭ 8) for tetrapandin 2-treated cells or 84.3 Ϯ 9.9 nM (n ϭ 10, statistically different from control, p Ͻ 0.0002) for 2-APB-treated cells. Similarly, if CCh was utilized to release Ca 2ϩ from InsP 3 -sensitive stores (Fig. 10B), tetrapandin 2 had no effect on the Ca 2ϩ release FIG. 5. Inhibition of thapsigarginstimulated Ca 2؉ entry by an HPLCpurified subfraction of the P. imperator venom. Fraction V from the G-50 filtration column was further fractionated by reverse-phase HPLC, and the eluted subfraction 7 was diluted in HBSS for fura-2 imaging experiments. As described in Fig. 1, once the stable calcium plateau was established, different dilutions of the subfraction 7 (SF7) were added, and its effect on the Ca 2ϩ plateau phase was monitored as a measurement of its effect on Ca 2ϩ entry. TG, thapsigargin. We also wanted to compare the effect of tetrapandins to that of 2-APB on the plateau phase of the CCh-stimulated Ca 2ϩ FIG. 7. Inhibition of thapsigarginstimulated Ba 2؉ entry by modified synthetic toxin peptide. A, HEK-293 cells were incubated in HBSS to establish a base line prior to switching into Ca 2ϩfree HBSS. A Ca 2ϩ -free, 2 mM Ba 2ϩ HBSS solution was added briefly to the cells to measure the Ba 2ϩ leak. Following a return to Ca 2ϩ -free HBSS, cells were stimulated with thapsigargin (TG, 1 M), and after cytosolic Ca 2ϩ returned to basal levels, 2 mM Ba 2ϩ was added. SOCE was defined as the leak-subtracted, TG-stimulated Ba 2ϩ influx. In half of the coverslips, the modified synthetic peptide 2 (1 M) was mixed with 2 mM Ba 2ϩ solution, and it was applied after thapsigargin depletion of Ca 2ϩ stores (thin line). The results obtained were compared with the control (thick line) for which a comparable amount of vehicle was added. B, the Ba 2ϩ entry values were plotted as percentage of control. At a dose of 1 M modified peptide toxin 2, SOCE was reduced by 48 Ϯ 2%. *, significantly different from control (p Ͻ 0.0001).

FIG. 8. Effect of valinomycin on inhibition of SOCE by modified synthetic toxin peptide.
Ba 2ϩ entry experiments similar to those described in Fig. 7 were performed in the presence of 10 M valinomycin. Statistical analysis of the data with the mean value of the leaksubtracted Ba 2ϩ entry representing the amount of SOCE is shown. *, significantly different from control (p Ͻ 0.0001). **, significantly different from valinomycin alone (p Ͻ 0.03).
response. Cells were treated with 100 M CCh in HBSS and allowed to establish a stable plateau phase. In control cells, the HBSS was replaced with Ca 2ϩ -free HBSS to establish that the plateau phase is dependent on a continued influx of Ca 2ϩ (Fig.  11A). For the 2-APB and tetrapandin conditions, these drugs were added during the plateau phase in the continued presence of HBSS. The data in Fig. 11B show that addition of 2-APB resulted in a substantial reduction of the plateau Ca 2ϩ level. A similar response was observed with the addition of tetrapandin 2 (Fig. 11C). The removal of Ca 2ϩ from the external medium brought the Ca 2ϩ value 97 Ϯ 1.3% (n ϭ 5) of the way back down to basal levels. In comparison, 2-APB brought the Ca 2ϩ value 76 Ϯ 3.3% (n ϭ 6) of the way back down to basal levels, while tetrapandin 2 brought the Ca 2ϩ value 70.4 Ϯ 5.0% (n ϭ 9) of the way back down to basal levels. DISCUSSION In this study, we found an inhibitory effect on SOCE by venom from the African scorpion P. imperator. This inhibitory activity appears to be uncommon among venoms from snakes and other scorpions as we found it in only one of 14 snake venoms and in one of four scorpion venoms tested. This finding encouraged us to proceed since it indicated that we were not observing a nonspecific effect of venoms in general, such as membrane depolarization due to toxin action on other ion channels. This was further supported by the finding that the venom still inhibited thapsigargin-stimulated Ca 2ϩ entry even in the presence of valinomycin, which locks the membrane potential in a hyperpolarized state.
We have outlined the steps involved in the purification of a peptide toxin from the P. imperator venom as well those involved in the sequencing and chemical synthesis of a peptide toxin that inhibits SOCE. Two short peptides were found to be present in the HPLC subfraction 7. Both peptides were tetrapeptides (LWSG and LWKT) and were found to share two amino acids in common.
These peptide toxins were named tetrapandins 1 and 2 (peptides 1 and 2, respectively) from the fact that they come from the scorpion of the genus Pandinus and have only four amino acids. This is quite unique among the venom components isolated and characterized thus far in the literature. Most scorpion venoms contain peptides that modify or block the function of ion channels such as voltage-dependent K ϩ and Na ϩ channels, although other venoms also have peptides capable of impairing the function of Cl Ϫ channels or Ca 2ϩ -sensitive ryanodine channels (for a review, see Ref. 36). Actually, from the venom of the same P. imperator, several interesting peptides were isolated previously, such as imperatoxin a and b (for a review, see Ref. 37) and scorpine, a defensin and antimalaria agent (38). The findings described here should not be taken as unusual since of the 100,000 components estimated to exist in scorpion venoms alone, only over 300 components have been studied thus far (36,39,40). Most of the studies reported in the literature deal with peptides from 26 to 76 amino acid residues long, but very few, if any, studies have been performed with the low molecular weight components of these venoms, which elute from the Sephadex G-50 column in fraction V.
Tetrapandins 1 and 2 have a lipophilic segment as well as a hydrophilic segment. The side chains of Leu and Trp are hydrophobic, whereas the C-terminal carboxylic groups and side chains of Lys and hydroxyl groups of Ser and Thr are hydrophilic. Since these peptides have a hydrophobic segment and knowing that biological membranes are permeable to hydrophobic substances, it is possible that these peptides might cross the cell membrane. This would provide the possibility that their effect could be at either an external or internal channel site to inhibit SOCE.
Both synthetic peptides have inhibitory action on SOCE when they are freshly prepared, although their potency is somewhat low. However, upon continued rotation for 2 days both peptides greatly increase their potency for inhibition of SOCE. After modification, the peptide toxins can initiate their effects at a dose of 1 nM, which is considerably lower than the 1 M dose required for a threshold effect to be seen with freshly prepared toxin. At present, we are not sure whether the peptides are chemically modified (e.g. air-oxidized) to a new structure or whether they aggregate during the incubation period. It is tempting to speculate that Trp oxidation to N-formylkynurenine might be responsible for such an increase in biological activity since a significant amount of the peptide purified from P. imperator venom showed signs of Trp oxidation. On the other hand, since they are amphiphilic substances (partially hydrophilic and hydrophobic) some kind of aggregation (dimerization or micelle formation) could occur with these peptides that could be responsible for the increased activity seen after agitation at room temperature. Thus, while a number of experiments are still needed to elucidate the exact modification of the synthetic peptides required to obtain the highest potency, we have clearly obtained a preparation that should be quite useful for the further exploration of SOCs. One interesting aspect of this peptide toxin is that at a maximum dose it inhibits only 50% of either the thapsigarginstimulated Ca 2ϩ plateau phase or the thapsigargin-stimulated Ba 2ϩ entry. This is in direct contrast to the effects of 2-APB, which has been shown to inhibit ϳ90% of the thapsigarginstimulated Ba 2ϩ entry in these cells (35). However, 2-APB also is known to inhibit release of Ca 2ϩ from internal stores. Our data indicate that the tetrapandins are more specific than 2-APB in that they have no effect on release of internal Ca 2ϩ stores under conditions where 2-APB dramatically reduces Ca 2ϩ store release. This is true whether the store Ca 2ϩ is released by thapsigargin or by InsP 3 production in response to CCh stimulation (Fig. 10).
The 50% maximal reduction of thapsigargin-stimulated Ca 2ϩ entry may suggest either a partial block of SOCs or full inhibition of one of multiple Ca 2ϩ entry pathways activated by thapsigargin. Previous studies have suggested that there may be more than one pathway activated in response to thapsigargin (8). For example, our previous TRPC3 antisense data (34) indicate that suppression of ϳ60% of the TRPC3 protein reduces SOCE by only 32%. Thus, extrapolation of this data to full TRPC3 protein suppression suggests that TRPC3 may mediate at most 45-50% of the total SOCE. Thus, it is possible that the toxin may be inhibiting in a selective manner either the TRPC3-sensitive or the TRPC3-insensitive portion of the thapsigargin-stimulated Ca 2ϩ entry. We are currently using small interfering RNA technology to try to achieve a more efficient reduction of TRPC3 levels in HEK-293 cells. If we can suppress TRPC3 levels by ϳ90%, then we will be in a better position to address the question of whether tetrapandins are working on the TRPC3-sensitive or TRPC3-insensitive pathway. The alternative possibility, that the toxin is partially blocking the store-operated channel, can be best tested by electrophysiological methods. Electrophysiological experiments to characterize the mechanism for inhibition of SOCE by tetrapandins will be performed in the near future. However, it is also possible for the tetrapandins to work via a mechanism other than direct interaction with the channel pore. There is much evidence for a conformational coupling mechanism for activating SOCs whereby direct interaction between TRPC3 channels and InsP 3 receptors is important. Given the potential for a small peptide with a hydrophobic segment to penetrate the plasma membrane, it is possible that tetrapandins penetrate the plasma membrane and perturb the coupling between TRPC3 channels and InsP 3 receptors. If the small interfering RNA experiments demonstrate that it is the TRPC3-dependent pathway that is inhibited by tetrapandins, this possibility will be explored in depth.
In regard to the potential use of tetrapandins to explore various Ca 2ϩ entry pathways, it was observed that tetrapandins were ineffective in blocking the Ca 2ϩ entry stimulated by OAG (Fig. 9) yet were fairly effective in blocking the CChstimulated Ca 2ϩ plateau phase (Fig. 11). Thus, tetrapandins may be useful in resolving the question of the involvement of SOCs (or subtypes of SOCs) and OAG-stimulated channels in the Ca 2ϩ response stimulated by CCh or other G proteincoupled receptor agonists. On another level, the tetrapandins should be extremely valuable in establishing which downstream physiological responses are regulated by SOCE.