INTRODUCTION

Scorpion venoms contain families of small basic proteins that modify the gating mechanism of Na⁺ channels or block with high affinity K⁺ channels of excitable cells (1, 2). These toxins have been invaluable tools in the identification, purification, structural mapping, and functional characterization of the corresponding ionic channels. The study of other ionic channels have also been aided by toxins from poisonous animals. For instance, snake venoms contain potent neurotoxins that block acetylcholine receptors (3), and snail and spider venoms contain small molecular weight proteins directed against neuronal Ca²⁺ channels (4, 5). A Cl channel-specific blocker peptide was also recently isolated from a scorpion venom (6). Thus, there exist a vast array of natural ligands useful for structural and functional characterization of ionic channels.

The Ca²⁺ release channel of SR¹ constitutes the major pathway for Ca²⁺ release during the process of excitation-contraction coupling in cardiac and skeletal muscle (7). The Ca²⁺ release channel binds the plant alkaloid ryanodine with nanomolar affinity, hence the name ryanodine receptor (RyR). Ryanodine has been an invaluable tool in the structural and functional characterization of RyR. The alkaloid binds to a conformationally sensitive domain on the RyR protein and may be used in binding assays as an index of the functional state of the channel (8, 9). However, ryanodine displays extremely slow dissociation kinetics that make its effect practically irreversible. Furthermore, certain concentrations of ryanodine may open RyRs while others may block them (10), leading to ambiguous results.

From the venom of the African scorpion Pandinus imperator, we isolated Imperatoxin I (IpTx_i), a ~15-kDa protein that inhibited [³H]ryanodine binding to cardiac and skeletal SR by blocking RyR channels (11). At concentrations well above the half-maximal effective concentrations (ED₅₀) exhibited for RyR, IpTx_i did not modify the binding of ligands targeted against other transporters and ionic channels of striated muscle (11). IpTx_i blocked RyR rapidly and reversibly, and when injected in ventricular cells it decreased twitch amplitude and intracellular Ca²⁺ transients, suggesting a selective blockade of Ca²⁺ release from the SR (11).

In this study, we carried out an in-depth analysis of the mechanism of action of IpTx_i on RyRs of cardiac and skeletal muscle. We determined the complete amino acid and nucleotide sequence of IpTx_i and show that, like -bungarotoxins, IpTx_i is a heterodimeric protein composed of a high molecular weight subunit with PLA₂ activity and a small, structurally unrelated subunit. We also show that free fatty acids, lipid products of IpTx_i-PLA₂ activity, are involved in the inhibition of RyR. IpTx_i thus offers an alternative way to block RyR distinct from ryanodine and other ligands that require a physical interaction with the RyR protein.

EXPERIMENTAL PROCEDURES

Protease lysine C (Lys-C) and restriction enzymes were from Boehringer Mannheim. Chemicals and solvents for peptide sequencing were from Millipore Co. Primers for polymerase chain reaction (lambda gt11 forward and reverse; 1218 and 1222) and Vent^R polymerase were from New England BioLabs. DNA sequencing kit (Sequenase version II), was from U. S. Biochemical Corp. Subcloning and sequencing vector (pBluescript phagemid; pKS), M13 20, and M13 reverse sequencing primers were from Stratagene. Brain phosphatidylethanolamine and brain phosphatidylserine were from Avanti Polar Lipids. [³H]Ryanodine was from DuPont NEN. p-Bromophenacyl bromide (pBPB) was from Sigma.

P. imperator venom was obtained by electric stimulation of scorpions maintained alive in the laboratory. Venom (120 mg per batch) was suspended in double distilled water and centrifuged at 15,000 × g for 30 min. The supernatant was applied onto a column (0.9 × 190 cm) of Sephadex G-50 superfine (Pharmacia Biotech Inc.). Fractions were eluted with 20 mM NH₃OAc (pH 4.7) at a flow rate of 20 ml/h. Fraction II containing IpTx_i was applied to a column (0.9 × 30 cm) of carboxymethyl (CM)-cellulose 32 (Whatman) equilibrated with 20 mM NH₃OAc (pH 4.7). Peptides were eluted at a flow rate of 20 ml/h with a linear gradient of 250 ml of 20 mM NH₃OAc (pH 4.7) and 250 ml of the same buffer containing 0.55 M NaCl. Peptides displaying capacity to inhibit [³H]ryanodine binding and phospholipase activity were dialyzed against deionized water (3 × 30 min), concentrated by lyophilization, and injected into a C₄ reverse-phase HPLC column (Vydac). IpTx_i was eluted with a linear gradient of 0-60% acetonitrile in 0.1% trifluoroacetic acid running at 1 ml/min for 60 min. The purity and identity of IpTx_i was confirmed by amino acid sequence analysis, as described for Na⁺ channel-blocking peptides (12). IpTx_i was quantified either by amino acid analysis or by absorbance at 280 nm (A_280 nm) using an extinction coefficient () = 14,952.

Amino acid analysis was performed on samples hydrolyzed in 6 N HCl with 0.5% phenol at 110 °C in evacuated, sealed tubes as described (13). Reduction of IpTx_i with dithiothreitol and alkylation with iodoacetic acid was performed as described (13). Reduced and alkylated IpTx_i was cleaved with protease Lys-C in 200 µl of 25 mM Tris-HCl (pH 7.2), and 1.0 mM EDTA, at a enzyme:peptide ratio = 1:100. Microsequence determination of native, reduced, and carboxymethylated toxin, and their peptide fragments was carried out with a 6400/6600 MilliGen/Biosearch Prosequencer, using the peptide adsorbed protocol on CD Immobilon membranes.

Two oligonucleotides encoding for two different regions of IpTx_i were synthesized as described (13). Oligonucleotide 1 (5-AC(N)ATGTGGGG(N)AC(N)AA(G/A)TGGTG-3; where N means any nucleotide) encoded for the first 8 amino acids of the amino terminus of the large subunit of IpTx_i. Oligonucleotide 2 (5-GA(G/A)GC(N)GG(N)TA(T/C)GG(N)GC(N)TGGGC-3) encoded for residues 14-21 of the small subunit. Total RNA was purified from the venomous glands (telsons) of 30 scorpions. The telsons were pulverized in liquid nitrogen and poured into 10 ml of GT buffer (4 M guanidinium isothiocyanate, 0.025 M sodium citrate (pH 7.0), 0.5% N-lauryl sarcosine, and 0.03 M -mercaptoethanol), vortexed, and centrifuged at 5,000 × g for 10 min. One ml of 2 M sodium acetate (pH 4) was added to the recovered supernatant. The resulting solution was extracted with 10 ml of water-saturated phenol plus 2 ml of chloroform:isoamyl alcohol (49:1, v:v), allowing to sit on ice for 15 min. After centrifuging at 10,000 × g for 20 min at 4 °C, the aqueous layer was precipitated with 1 volume of isopropyl alcohol, incubated at 20 °C for 1 h, and centrifuged at 10,000 × g for 20 min at 4 °C. The pellet was dissolved in 0.3 ml of GT buffer and 0.3 ml of isopropyl alcohol, incubated at 20 °C for 1 h, and pelleted again. The pellet was washed with 95% ethanol, briefly dried under vacuum, and resuspended in RNase-free water. Messenger RNA was purified following the instructions of the Hybond-mAP protocol (messenger affinity paper; Amersham Corp., RPN.1511). cDNA synthesis was performed as described (13) from 5 µg of mRNA. The cloning of the cDNA library was also performed as described (13). The screening of the cDNA library was performed with oligonucleotides 1 and 2 separately, but the clones detected with oligonucleotide 1 were analyzed first. Inserts of cDNA from positive clones were polymerase chain reaction-amplified using  gt11 forward and reverse primers. Polymerase chain reaction products were purified from agarose gel and ligated into the EcoRV site of pBluescript (pKS) phagemid. The ligation reactions were used to transform Escherichia coli DH5- cells. Plasmid DNA from white colonies was digested with BamHI and HindIII to verify the size of the original inserts. Clones of interest were sequenced using the Sequenase^R kit version 2 (U. S. Biochemical Corp.) on both strands. Oligonucleotides  gt11 forward,  gt11 reverse, M13 -20, and M13 reverse were used for sequencing.

The amino acid sequence of the large and small subunits of IpTx_i were compared with those of other proteins deposited in the protein data base of GenBank (Los Alamos National Laboratory, Los Alamos, NM) by computer analysis using the program Blitz version 1.5 (Biocomputing Research Unit, University of Edinburgh, UK).

Phospholipase A₂ activity was determined by the titration method of Shiloah et al. (14), using dilute egg yolk (1:20 in saline solution) as substrate. Liberation of acid was measured at pH 8.0 and 37 °C by titration with 5 mM NaOH under a constant stream of N₂. One unit of enzyme is defined as the amount of enzyme that liberates 1 µmol of free fatty acid/min under the above conditions. Inhibition of phospholipase A₂ with pBPB was carried out as described by Díaz et al. (15). One hundred µl of 200 µM IpTx_i in 35 mM Tris (pH 8.0) were mixed with 10 µl of 2 mM p-BPB in acetone and incubated for 16 h at room temperature under dim light. The reaction was stopped by filtration in a Sephadex G-10 column. pBPB-treated IpTx_i elutes in the void volume of the column, whereas the excess of pBPB is retarded.

The small subunit of IpTx_i was synthesized by the solid phase method of Merrifield (16), using t-butyloxycarbonyl-amino acids. This subunit contains a Cys residue at position 4. To avoid formation of inter-peptide disulfide bonds, Cys⁴ was substituted by Ala (peptide A), Met (peptide M), or Cys (peptide C) protected with 3-nitro-2-pyridinesulfonyl (thiol-protecting group that will not allow formation of disulfide bonds). At the end of the synthesis all three peptides were separated in a C₁₈ reverse-phase HPLC column, using the conditions described above. The purity of the synthetic peptides was confirmed by both amino acid analysis and microsequencing, as described above.

[³H]Ryanodine binding to pig cardiac and rabbit skeletal SR vesicles was carried out for 90 min at 36 °C in 0.1 ml of 0.2 M KCl, 1 mM Na₂EGTA, 0.995 mM CaCl₂, 10 mM Na-Pipes (pH 7.2). The calculated free Ca²⁺ was 10 µM. [³H]Ryanodine (68.4 Ci/mmol) was diluted directly in the incubation medium to a final concentration of 7 nM. Protein concentration was 0.2-0.4 and 0.3-0.5 mg/ml for skeletal and cardiac SR, respectively. Samples were filtered on Whatman GF/B glass fiber filters and washed twice with 5 ml of distilled water. A Brandel M24R cell harvester was used for filtration. Nonspecific binding was determined in the presence of 10 µM unlabeled ryanodine and has been subtracted from each sample.

Recording of single RyR in lipid bilayers was performed as described previously (17). Briefly, a phospholipid bilayer of phosphatidylethanolamine:phosphatidylserine (1:1 dissolved in n-decane to 20 mg/ml) was formed across an aperture of ~300 µm diameter in a delrin cup. The cis chamber (900 µl) was the voltage control side connected to the head stage of a 200 A Axopatch amplifier, and the trans chamber (800 µl) was held at virtual ground. Both chambers were initially filled with 50 mM cesium methane sulfonate and 10 mM Tris/Hepes (pH 7.2). After bilayer formation, cesium methane sulfonate was raised to 300 mM in the cis side, and 100-200 µg of SR vesicles were added. After detection of channel openings, Cs⁺ in the trans chamber was raised to 300 mM to collapse the chemical gradient. Single channel data were collected at steady voltages (30 mV) for 2-5 min. Channel activity was recorded with a 16-bit VCR-based acquisition and storage system at a 10-kHz sampling rate. Signals were analyzed after filtering with an 8-pole Bessel filter at a sampling frequency of 1.5-2 kHz. Data acquisition and analysis were done with Axon Instruments software and hardware (pClamp v6.0, Digidata 200 AD/DA interface).

RESULTS

Purification of IpTx_i from P. imperator venom was performed in three chromatographic steps as described under "Experimental Procedures" and shown in Fig. 1. After fractionation of whole venom in Sephadex G-50 (Fig. 1A), five fractions were collected and assayed for effects on [³H]ryanodine binding. Fraction 2 contained polypeptides in the range of 8-16 kDa and was the only fraction that inhibited [³H]ryanodine binding. The second step in the purification of IpTx_i consisted of ion-exchange chromatography in CM-cellulose (Fig. 1B). The fraction containing IpTx_i (indicated by the arrow) eluted early in the run, when the concentration of NaCl was ~160 mM. Fig. 1C shows the chromatographic profile of IpTx_i after elution from a reverse-phase C₄ HPLC column. Only a minor contaminant was present (labeled with asterisk), which was discarded for further studies of IpTx_i structure. However, amino acid sequence analysis of purified IpTx_i yielded two different amino acids per reaction cycle, with an apparent equivalent stoichiometry. This indicated either that a contaminant was still present at the end of our purification procedure or that IpTx_i was composed of two different peptide subunits.

We determined the dose-response relationship for each of the active venom components to assess their potency to inhibit RyRs (Fig. 1D). Whole P. imperator venom and fraction 2 inhibited [³H]ryanodine binding to cardiac SR with a concentration of 20.2 and 4.3 µg/ml, respectively, yielding the half-maximal effect (IC₅₀); IC₅₀ for pure IpTx_i was 0.7 µg/ml. Based on this value and the molecular mass of IpTx_i (~15 kDa), the estimated apparent dissociation constant (K_d) was 46 nM. Essentially identical results were obtained when skeletal SR was used for displacement studies (not shown).

To elucidate whether IpTx_i was composed of two subunits, an aliquot of IpTx_i was reduced and alkylated. The modified toxin was then chromatographed in a Bio-Gel P30 column (Fig. 2A) to eliminate reaction by-products and excess of reagents. Three peaks were obtained (Fig. 2A) and analyzed by SDS-PAGE (Fig. 2B). Peak 1 (labeled Red, abbreviation for reduced) had lower molecular mass than native IpTx_i (labeled Nat); peak 2 could not be resolved in the same gel (not shown), but direct sequencing indicated that it corresponds to a ~3-kDa peptide (Fig. 3A); peak 3 was not peptidic in nature, indicating that it corresponded to the reducing and alkylating reagents. Fig. 2C shows that the apparent molecular mass of native IpTx_i was ~15 kDa and that of the reduced IpTx_i was ~12 kDa. Thus, treatment of IpTx_i with thiol-reducing agents separates a large (~12 kDa) from a small (~3 kDa) subunit.

The reduced and alkylated derivatives of IpTx_i (peak 1 and 2 of Fig. 2A) were sequenced by direct Edman degradation. As shown in Fig. 3A, we identified the first 27 amino acid residues of the large subunit, and all 28 amino acid residues of the small subunit. To identify the remaining residues, the large subunit was cleaved with Lys-C endopeptidase, which yielded three peptide fragments that eluted with different retention times in an HPLC column (data not shown). The amino acid sequence of each of these peptides, obtained also by Edman degradation, is underlined in Fig. 3A.

Two oligodeoxynucleotides synthesized according to the amino acid sequence of stretch 1-8 of the large subunit and stretch 14-21 of the small subunit were used to screen a cDNA library, constructed from the venomous glands of P. imperator scorpions. These oligodeoxynucleotides were used as probes to isolate a full-length IpTx_i cDNA clone. Both probes led to the isolation of the same cDNA clone. Its complete nucleotide sequence is shown in Fig. 3B. This sequence contained an open reading frame of 167 amino acids encompassing 1) a putative signal peptide (first underlined 31 amino acids, positions 31 to 1); 2) the mature ~12-kDa, large subunit of IpTx_i (104 amino acids, positions 1-104); 3) a putative connector pentapeptide (RRLAR, positions 105-109); and 4) the mature small subunit (27 amino acids; positions 110-136). Thus, a single continuous cDNA clone encoded the two polypeptide subunits of IpTx_i. This finding confirmed that IpTx_i is a heterodimeric protein. Since treatment of IpTx_i with thiol-reducing agents (Fig. 2A) effectively separates both subunits, this suggests that they are covalently linked by a disulfide bridge.

A comparison of the amino acid sequence of the two subunits of IpTx_i with sequences available from GenBank revealed intriguing results. First, the small subunit shared no significant similarity with sequences of scorpion peptide blockers of Na⁺, K⁺, or Cl channels, with members of the Kunitz protease inhibitor superfamily, or with any other sequence deposited in the protein data base. Thus, this small subunit constitutes a new class of scorpion peptides. On the other hand, Fig. 4 shows that the large subunit of IpTx_i was 38% homologous to PLA₂ from honey bee (A. mellifera) venom and 35% homologous to PLA₂ from heloderma (H. horridum) venom. These two PLA₂s compose group III of secreted PLA₂ (18), whose main features are low molecular mass (~14 kDa), high disulfide bond content, and strict dependence on Ca²⁺ for lipolytic effect. Since the major criterion for classification is sequence homology more than function, IpTx_i may be classified within this group with the remarkable difference, however, of possessing an accessory protein.

We tested if the small subunit of IpTx_i was responsible for blocking RyR. Reduced and carboxymethylated small subunit (peak 2 of Fig. 2A), at concentrations up to 15 µM, had no effect on the binding of [³H]ryanodine to cardiac or skeletal RyR (data not shown). To discard the presence of spurious reagents as being the cause of negative results, we synthesized the small subunit in three distinctive forms as follows: peptide C was similar to the native small subunit, and peptides A and M were also similar to the native small subunit, except that Cys⁴ was substituted by Ala and Met, respectively. This avoided the formation of interpeptide disulfide bridges caused by oxidation of Cys⁴. The three peptides were HPLC-purified before testing. Table I shows that none of the purified synthetic peptides had significant effect on the binding of [³H]ryanodine to cardiac RyR or on the binding of [³H]PN200-110 to voltage-dependent Ca²⁺ channels of sarcolemma (19).

Table I. Effect of synthetic peptide analogs of the small subunit of ipTxi on the binding of [3H]ryanodine and [3H]PN200-110 to cardiac sarcolemma and SR vesicles Peptide concentration tested was 10 µM. Results are the mean ± S.D. of three independent determinations. Peptide concentration tested was 10 µM. Results are the mean ± S.D. of three independent determinations. [3H]Ryanodine bindinga [3H]PN200-110 bindingb % % Control 100 100 Peptide A 116  ± 8 100  ± 6 Peptide M 110  ± 4 111  ± 12 Peptide C 106  ± 5 111  ± 7 a Binding of [3H]ryanodine to SR vesicles was carried out as described in the text. b Binding of [3H]PN200-110 to cardiac sarcolemma was performed as described (19).

We next investigated if the large subunit of IpTx_i, or its enzymatic activity, was involved in RyR inhibition. The reduced and carboxymethylated form of the large subunit of IpTx_i (peak 1) was incapable of inhibiting [³H]ryanodine binding to cardiac or skeletal RyR. However, this was expected given that internal disulfide bridges are essential for preservation of the toxin's tridimensional structure as well as for expression of its enzymatic activity (20). We thus resorted to pBPB, a covalent modifier of His residues that specifically blocks phospholipase A₂ activity (15) without affecting disulfide bonds. The rate of fatty acid liberation by 1 mol of IpTx_i was 384 ± 26 and 41 ± 12 mol min¹ in control and after incubation with pBPB, respectively (n = 4; data not shown). Thus, treatment of IpTx_i with pBPB decreased substantially its PLA₂ activity. Fig. 5 shows that pBPB-treated IpTx_i decreased dramatically its capacity to inhibit the binding of [³H]ryanodine to cardiac and skeletal RyR (open symbols). This was in contrast to control IpTx_i (a batch of IpTx_i that underwent the same treatment as pBPB-treated IpTx_i, except that pBPB was omitted), which retained its capacity to inhibit cardiac and skeletal RyR with high affinity (filled symbols).

The enzymatic activity of PLA₂ from group III requires Ca²⁺ for catalysis (20). Therefore, we expected that if the inhibition of RyRs by IpTx_i resided in its enzymatic activity, inhibition should be apparent only in the presence of micromolar [Ca²⁺]. Fig. 6 shows that Ca²⁺ was essential for binding of [³H]ryanodine to RyRs and for detection of the IpTx_i effect. The left panel shows the Ca²⁺ dependence of [³H]ryanodine binding to skeletal SR and the effect of IpTx_i. Specific binding in control (open circles) had a threshold for detection at 100 nM [Ca²⁺] (pCa 7) and was optimal at 10-100 µM [Ca²⁺]. Higher [Ca²⁺] inhibited binding, giving rise to a bell-shaped curve. In the presence of IpTx_i (filled circles), the binding curve was dramatically decreased in absolute values. The percentage of IpTx_i inhibition was 5.5, 32, and 71% at pCa 7, 6, and 5, respectively. No further inhibition was observed at higher [Ca²⁺]. Thus, the degree of IpTx_i inhibition increased with [Ca²⁺].

In cardiac SR (Fig. 6, right panel), the Ca²⁺ dependence of [³H]ryanodine binding also had a threshold for detection at pCa 7 and a maximum at pCa 5 (control, open circles). However, unlike skeletal SR, inactivation of binding by high [Ca²⁺] was not pronounced. Binding decreased only by 30% with respect to maximum in cardiac RyRs, and it decreased by 80% in skeletal RyRs. This distinctive response of RyR to Ca²⁺ is also displayed by individual RyR reconstituted in lipid bilayers (8, 21), indicating that the binding assay effectively tracks the activity of the receptor. In the presence of IpTx_i, the binding curve was markedly decreased. The percentage of inhibition was 10, 38, and 83% at pCa 7, 6, and 5, respectively. Thus, although cardiac and skeletal RyRs respond differently to Ca²⁺, the inhibitory effect of IpTx_i increased equally with [Ca²⁺] in both isoforms.

The membranes of SR vesicles contain different classes of phospholipids that may serve as substrates for the PLA₂ activity of IpTx_i. We reasoned that if the inhibitory properties of IpTx_i were at least partly due to its enzymatic activity, then inhibition should be observed by incubating RyR with supernatant of IpTx_i-treated SR vesicles. Control SR vesicles were diluted in binding medium to 1 mg/ml and incubated at 36 °C with 100 nM IpTx_i. After 30 min, the IpTx_i-treated SR vesicles were pelleted at 32,000 × g for 15 min, and a clear supernatant was obtained. The supernatant was then tested for its capacity to inhibit the binding of [³H]ryanodine to cardiac SR vesicles. A control tube that contained only IpTx_i in binding medium without SR vesicles was run in parallel to determine the percentage of inhibition caused by the toxin alone. Fig. 7 shows that increasing concentrations of IpTx_i-treated SR supernatant inhibited the binding of [³H]ryanodine dose-dependently (filled circles). This inhibition could not be attributed to IpTx_i because the control supernatant reduced binding only marginally (open circles). For instance, in the most extreme case, 10 µl of IpTx_i-treated SR supernatant inhibited 100% of specific binding; the estimated concentration of IpTx_i in this tube was 10 nM, which inhibited binding only by 22%. Hence, the lipolytic activity of IpTx_i releases from SR vesicles a phospholipid product that is capable of inhibiting RyR activity.

To rule out the possibility that inhibition of RyR by IpTx_i was caused by an irreversible disruption of the channel protein, we incubated SR vesicles (1 mg/ml) with water (control) or with 1 µM IpTx_i (IpTx_i-treated SR). After 30 min at 36 °C, an aliquot was taken from each sample. Then, both SR samples were washed twice, pelleted, and prepared for [³H]ryanodine binding. Fig. 8 shows that binding of [³H]ryanodine to IpTx_i-treated vesicles was decreased with respect to control after 30 min of incubation. However, these vesicles displayed essentially the same binding as control after washing off IpTx_i. Thus, RyRs recover their capacity to bind [³H]ryanodine after treatment with IpTx_i. Assuming that the [³H]ryanodine binding assay followed changes in channel gating (8-10, 17, 21), the changes described above suggest that IpTx_i released a phospholipid product that bound to RyRs (or a closely associated regulatory protein) and decreased open probability (p_o); after removal of the phospholipid product, p_o could recover to control levels.

To test directly the above hypothesis and to gain insight on the manner in which IpTx_i inhibits RyRs, we reconstituted swine cardiac RyR in planar lipid bilayers, as described (11, 17). Fig. 9 shows traces from continuous recording at 30 mV holding potential in the absence (control) and the presence of increasing concentrations of IpTx_i. Channel activity was monitored over 80 s in each condition, and the mean p_o was obtained from current histograms. The traces show that IpTx_i blocks RyR dose-dependently: at low concentrations (200 nM) IpTx_i decreases the lifetime of the open events, converting long openings into brief, frequently unresolved open events; at higher concentrations, both long and brief openings progressively decrease in frequency until they disappear. A quantitative description of this effect is presented in the open time histograms of Fig. 9B. In control, 2,034 events could be fitted with three exponentials with mean open time () = 0.7 ms (66%), 1.9 ms (29%), and 9.8 ms (5%). In the presence of 500 nM IpTx_i, the histogram for 801 openings collected during the same period as control was monoexponential with  = 0.8 ms. Both the decrease in the frequency and duration of the open events contributed to the decrease in p_o. Fig. 9C shows the log dose-response relation of p_o as a function of IpTx_i concentration. The IC₅₀ for IpTx_i inhibition was 140 nM. Since this value is in fairly good agreement with that determined in binding experiments (50-80 nM), this indicates that [³H]ryanodine, at the concentrations used, does not interfere with the inhibitory capacity of IpTx_i. More importantly, the results show that the lipolytic product of IpTx_i is capable of interacting directly with the RyR or with a closely associated protein that regulates channel gating.

PLA₂ catalyzes the hydrolysis of the S_n2 fatty acyl bond of phospholipids to release free fatty acids and lysophospholipids (22). We tested the effect of each of these lipids separately on RyR activity. Fig. 10A shows that lysophosphatidylcholine (lyso-PC), at concentrations up to 300 µM, does not modify substantially the open lifetime or the unitary conductance of cardiac RyR. On the other hand, linoleic acid, an 18-carbon unsaturated fatty acid, blocks RyR at lower concentrations and in a manner that is reminiscent of the blockade produced by IpTx_i. Addition of 30 µM linoleic acid to the cytosolic face of the channel decreases the lifetime and the frequency of open events; similar to IpTx_i, the fatty acid induces the appearance of openings that are too fast to be resolved by our recording bandwidth.

The potency of several fatty acids to inhibit RyRs was determined by their capacity to inhibit [³H]ryanodine binding to cardiac SR vesicles (Fig. 10B). Palmitic acid, a 16-carbon saturated fatty acid, partially inhibited RyR but at concentrations 100 µM. Arachidonic acid, linoleic acid, and oleic acid, unsaturated fatty acids abundant in SR membranes of cardiac and skeletal muscle (23), totally inhibited RyR with an IC₅₀ = 25, 55, and 70 µM, respectively. In agreement with single channel results, lyso-PC and lysophosphatidylethanolamine were unable to modify [³H]ryanodine binding. Thus, not all PLA₂ products are capable of inhibiting RyRs. Among the fatty acids, the potency to inhibit RyRs increases with the carbon chain length and the number of unsaturations.

DISCUSSION

In the present study, we determined the complete amino acid and nucleotide sequence of IpTx_i, a heterodimeric protein from the venom of the scorpion P. imperator, and unraveled the molecular mechanism by which it inhibits RyRs of cardiac and skeletal muscle.

The deduced amino acid sequence from the cDNA encoding IpTx_i and the amino acid sequence determined by Edman degradation strongly suggests that IpTx_i is synthesized as a precursor of the pre-pro form. Putative signal and connector peptides must be removed to produce mature IpTx_i. The proposed signal peptide does have a large content of acidic residues (8 out of the last 18 residues are acidic), a property that is rarely seen in signal peptides. For this reason, we cannot exclude whether the proposed signal peptide codes for another peptide before coding for the large subunit of IpTx_i. Regarding the connecting pentapeptide Arg-Arg-Leu-Ala-Arg (positions 105 to 109 in the cDNA sequence of Fig. 3), we gather that several enzymes must be required to eliminate it from the mature protein. Initially, a cleavage must occur at Arg¹⁰⁹ (monobasic site), as it occurs in the maturation of prosomatostatin and other prohormones (24). Two characteristics distinguish this cleavage site and both are present in the cDNA of IpTx_i: (i) Arg (or other basic residue) is immediately preceded by Ala or Leu (or both); (ii) Arg is located three or five residues upstream the basic amino acid involved in the cleavage. Next, another cleavage must occur at Arg¹⁰⁶ (dibasic site) as is the case during the maturation of diverse peptidic hormones (25). A third enzyme with carboxypeptidase activity must intervene to remove Arg¹⁰⁵ and Arg¹⁰⁶. Because of this complicated pattern of maturation, we assume that the linking of the small and large subunit of IpTx_i must be important for the function of IpTx_i.

Mature IpTx_i is composed of a ~3-kDa peptide covalently linked to a ~12-kDa peptide with PLA₂ activity. The large subunit of IpTx_i conserves the most important substructures of secreted PLA₂ from group III (Fig. 4). The His/Asp pair, essential for Ca²⁺ binding, is in position 33/34 of IpTx_i and is preceded in the three PLA₂s by a Cys-Cys-Arg motif; the amino terminus (roughly residues 4-12 of IpTx_i), presumably involved in substrate binding, is also highly conserved in the three PLA₂s; lastly, the Cys residues, essential for maintaining proper folding through disulfide bonds, may be used in this group as the elemental frame to align homologous sequences and to identify stretches of sequence that have been inserted or deleted through evolution. Only Cys¹⁰¹ of IpTx_i (relative position 104 in Fig. 4) does not match with Cys residues present at the carboxyl terminus of the other PLA₂ shown. It is likely, therefore, that Cys¹⁰¹ is involved in the disulfide bridging with Cys⁴ of the small subunit.

The presence in IpTx_i of a PLA₂ subunit covalently linked to a small, structurally unrelated peptide is reminiscent of the bipartite arrangement of -bungarotoxins (26), a group of snake neurotoxins that inhibit neurotransmitter release (27). Each -bungarotoxin dimer has a PLA₂ subunit covalently bound to a smaller subunit related to Kunitz-type protease inhibitors. Members of the Kunitz superfamily bind to and block voltage-dependent K⁺ and Ca²⁺ channels (28). Thus, although the small subunit of IpTx_i showed no homology to members of the Kunitz superfamily, its distinctive arrangement within the toxin suggested that it might act as a blocker. For this reason, it was surprising at first to realize that the small subunit was not directly responsible for inhibiting RyRs. Direct addition of the small subunit to our [³H]ryanodine binding assays failed to inhibit RyRs, as did a synthetic peptide with amino acid sequence similar to that of the small subunit and two additional derivatives designed to avoid interpeptide disulfide bridge formation (Table I). Hence, the target site for the small subunit, or its biological function, remains to be determined.

Given our failure to detect inhibition by the small subunit, we turned our attention to the large subunit of IpTx_i. In a reduced and carboxymethylated form, the large subunit of IpTx_i failed to inhibit RyRs (not shown). However, this was not surprising given that disulfide bridges are essential to maintain the toxin's three-dimensional structure (20). It was the treatment of IpTx_i with pBPB, a covalent modifier of His residues that inhibits PLA₂ activity without affecting disulfide bridges (15), that decreased substantially the capacity of IpTx_i to inhibit RyRs (Fig. 5). This suggested that a lipid product of PLA₂ activity was involved in the inhibition of RyRs. Three separate lines of evidence supported this notion. 1) Inhibition of RyR by IpTx_i was favored by the presence of Ca²⁺ (1 µM; Fig. 6), as expected from the Ca²⁺ dependence of enzymatic activity of PLA₂ from group III and several other groups (20). 2) The supernatant of IpTx_i-treated SR vesicles could inhibit RyRs even before IpTx_i was present in a concentration large enough to do so (Fig. 7); this indicated that RyR inhibition was brought about by a lipid product of IpTx_i released from SR vesicles during the incubation period. 3) The kinetics of RyR inhibition by IpTx_i were mimicked by direct addition of linoleic acid (but not of lyso-PC) to the cytoplasmic side of the channel (Figs. 9 and 10); other long chain, unsaturated fatty acids could also inhibit RyRs (Fig. 10). Thus, a potential scenario for inhibition of RyR by IpTx_i is as follows: in the presence of micromolar [Ca²⁺], the large subunit of IpTx_i catalyzes the hydrolysis of the S_n2 fatty acyl bond of the phospholipids of the SR membrane. The reaction yields free fatty acids and lysophospholipids; the former are released into the incubation medium, and the latter may be liberated or remain embedded into the SR membrane. Free fatty acids bind to RyR or to a closely associated protein that controls gating. At low concentrations, they produce an incomplete block of RyR; higher concentrations gradually block the RyR completely, giving rise to the dose-response relationship of channel activity and [³H]ryanodine binding versus IpTx_i concentration of Figs. 1, 5, and 9.

PLA₂s are abundant components of snakes, scorpions, and bee venoms and often constitute the main toxic component to mammals. Although all PLA₂s induce a variety of pathological symptoms including neurotoxicity and myotoxicity (29), they differ in their mechanism of action and in their molecular targets. For example, the snake neurotoxin crotoxin is composed of a PLA₂ subunit and an inhibitory subunit, which keeps the phospholipase inactive until binding to a presynaptic receptor triggers the dissociation of the inhibitory subunit (30). In contrast, notexin from Netechis scutatus scutatus is a PLA₂ without an associated subunit that produces muscle paralysis by binding to a specific receptor in the neuromuscular junction (31). IpTx_i is the first example of a scorpion toxin in which a PLA₂ is found chaperoned by a smaller, structurally unrelated subunit. In a previous study (11), we found that several transporters and ion channels of striated muscle including the inositol trisphosphate receptor, the muscarinic receptor, voltage-sensitive Na⁺ channels, and the Ca²⁺-ATPase of SR were insensitive to micromolar concentrations of IpTx_i. These results argue against an indiscriminate effect on key molecules of excitation-contraction coupling by the lipolytic action of IpTx_i. However, an important question that still remains to be answered concerns the presence of the small subunit of IpTx_i. What is its molecular target? What function does it perform to enhance the PLA₂ activity or to suppress it while in transit to the specific receptor? Regardless of the primary site affected, a major contribution of this study was to establish the fact that the molecular mechanism involved in the toxicity of IpTx_i will most likely involve abnormalities in intracellular Ca²⁺ mobilization due to blockade of RyRs.