Discharge effect on pancreatic exocrine secretion produced by toxins purified from Tityus serrulatus scorpion venom.

Three toxic polypeptides were purified from the venom of the Brazilian scorpion Tityus serrulatus by means of gel filtration in Sephadex G-50 and ion-exchange chromatography in carboxymethylcellulose. The peptides are basic molecules with molecular weights in the range of 7000 for which the amino acid compositions and sequences were determined. The effect of the purified peptides on pancreatic exocrine secretion in the guinea pig was studied. Biochemical measurements show that the cells are stimulated by these peptides to discharge their zymogen granules. Light and electron microscopic images confirm the biochemical measurements. At the light microscope level, acinar cells show dramatically fewer zymogen granules than in control pancreas with the appearance of large vacuoles and some loss of morphological integrity. Electron micrographs display apical regions devoid of zymogen granules and condensing vacuoles whereas acinar lumina contain crystalline secretory material. The secretory effect observed in vitro is comparable to that of carbamylcholine and that of the peptidergic secretagogue cholecystokinin-pancreozymin.

The first report on the isolation of a toxic component, named Tityustoxin, from the venom of the scorpion Tityus serrulutus Lutz and Mello, was published in 1966 (13). Subsequently, Toledo and Neves (36) and our group (27,40) have demonstrated the presence of more than one toxic component in the same venom. In fact, the component named gamma was the first toxin from Brazilian scorpion venoms for which the NH2-terminal amino acid sequence was determined (27). Further chemical characterization of the toxins from T. serrulutus was described (5, 28, 30). The specificity and affinity of toxin gamma for Na' channels (3,39) and the effect of toxin 11-9 on voltage-dependent K' channels of squid axon membranes were also documented (7,29). A summary of the known amino acid sequences of various toxic peptides from T. serrulutus was reviewed by Possani in 1984 (26). Along with the chemical characterization of T. serrulutus toxins, some physiological effects of these peptides (17,39) and clinical reports on the actions of the venom in humans as well *This work was partially supported by National Institutes of Health Grant AM27114. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

2699.
)I To whom correspondence should he addressed Tel.: 919-551-as in experimental animals were described (21). Acute pancreatitis in humans stung by these scorpions was also reported (4,38). Since acute pancreatitis is a unique result of a venomous sting in humans, it was our purpose to determine the association between pure protein components isolated from the venom of the Brazilian scorpion, T. serrulutus, and measurements of secretory discharge in guinea pig pancreas in uitro. The work presented here represents both the isolation and characterization of three venom proteins that are active in secretory discharge of zymogen proteins of pancreatic lobules as well as biochemical measurements and a morphological evaluation of secretory discharge.

MATERIALS AND METHODS'
Separation Procedures-Separation of soluble venom was basically performed according to a previous publication by our group (28). Briefly, the procedure includes size-exclusion chromatography and two consecutive steps of weak cation-exchange chromatography on carboxymethylcellulose (CM-cellulose). The first separation was performed in a column (0.9 X 200 cm) of Sephadex G-50 (medium) equilibrated with 20 mM ammonium acetate buffer, pH 4.7, from which four main fractions were obtained. Each of the toxic fractions (I1 to IV) were further separated on CM-cellulose a t two different pH values. The first CM-cellulose column (0.9 X 30 cm) was developed in ammonium acetate buffer, pH 4.7. In this step, toxic components 11-11, 111-8, 111-10, and IV-5 were separated, corresponding to the main toxic components of the venom (28). Each of these components was subsequently purified by rechromatography of the dialyzed fractions through a second CM-cellulose column equilibrated with 50 mM sodium phosphate buffer, pH 6.0. A linear gradient from 0-0.5 M NaCl in the same buffer was used to elute the toxic proteins in homogeneous form.
Enzymatic digestion mixtures of reduced and alkylated proteins were separated into their component peptides by reverse-phase high performance liquid chromatography (HPLC)' on octadecylsilane columns developed with linear gradients of 0.1% trifluoroacetic acid and integrated with up to 60% acetonitrile. Isolated fractions were evaporated to dryness under a stream of nitrogen for solvent removal prior to amino acid analysis and sequenator application.
Lethality Tests-Lethality tests in mice were conducted as described (27). Basically, aliquots were injected intraperitoneally into 20-g mice that were observed for toxic effects. Among the symptoms of toxicity are hyperexcitability, lacrimation, apnea, partial paralysis of the rear limbs, diarrhea, respiratory failure, and death. Acutely intoxicated animals were usually killed according to approved animal protocols before they died of the toxin.
Amino Acid Composition and Sequence Determination-Amino acid compositions were obtained with acid hydrolysates of purified proteins analyzed with a Durrum/Dionex D-502 automatic amino acid analyzer, as described (27). Prior to amino acid sequence determination, the toxins were either reduced and carboxymethylated or pyridylethylated (27, 28, 30). Determination of half-cystine was obtained after performic acid oxidation (23). Tryptophan content was deduced from sequence data.
Carboxypeptidase digestion for the determination of carboxylterminal amino acid residues was carried out in 50 mM sodium acetate buffer, pH 3.7, with carboxypeptidase P. Release of free amino acids, as a function of time, was determined by amino acid analysis directly from aliquots of the hydrolysis mixture. Usually, 10 nmol of reduced and carboxymethylated toxins (both 111-8 and IV-5) were incubated with enzyme (5% of peptide concentration) a t room temperature (25 "C) for variable lengths of time. An aliquot containing 1 nmol of peptide was withdrawn a t each interval, and hydrochloric acid was added to stop the reaction. Two independent protocols for the time course enzymatic hydrolysis were used. First, the aliquots were taken a t 30 min and 1, 2, and 12 h; second, the aliquots were taken a t intervals of 10 min for the 1st hour, plus samples a t 90 and 120 min. Each sample was dried, resuspended in citrate buffer, pH 2'2, and analyzed.
Sequence determination by automatic Edman degradation (9) was carried out either on a Beckman 890 M microsequencer or on Polybrene-treated glass fiber discs in an automated gas phase sequenator (Model 470A, Applied Biosystems, Inc.) equipped with an on-line sequence identification module, as described (30). Completion of the primary structure required the use of peptide fragments produced by enzymatic digestion.
Fragmentation of reduced and alkylated toxins by cleavage with protease V8 from Staphylococcus aureus (data not shown) and/or trypsin was obtained as described previously (30). Digestion of the reduced alkylated toxins by endoproteinase Asp(N) was performed as described by the supplier (Boehringer Mannheim) a t an enzyme-tosubstrate ratio of 1 to 50 for 16 h unless otherwise indicated. Peptides were separated by HPLC as mentioned above.
Measurement of Exocrine Secretory Discharge-Pancreatic lobules were prepared according to the method described by Scheele and Palade (35), mounted on nylon monofilament discs, and preincubated for 10 min at 0 "C. They were pulse-labeled for 10 min a t 37 "C with 0.4 p M ~-[4,5-'H]leucine (10 pCi/ml) in KRB, then washed with 4.0 mM L-leucine KRB (KRB-chase). Upon transfer to chase incubation flasks containing KRB-chase and the various secretagogues (carbamylcholine or toxins 111-8, 111-10, or IV-5), the lobules were further incubated a t 37 "C in a water bath shaker equipped with a closed gassing hood with a continuous flow of 95% 02,5% CO2.
After a 3-h incubation, the lobules were separated from the medium and homogenized in water. Samples from incubation medium and tissue homogenates were prepared immediately for the radioactivity assay (15) in Aquasol-2 and counted in a Beckman LS 7000 liquid scintillation counter. Exocrine protein discharge was expressed as the percentage of proteins recovered from the incubation medium. Amylase activity was measured according to the Bernfeld method (6) and expressed as milligrams of maltose liberated in 15 min a t 30 "C. Results were expressed as above, i.e. the percentage of amylase released in the medium based on the total amylase distribution in tissue and incubation medium.
Measurement of lactate dehydrogenase activity was carried out on aliquots of the same incubation media to determine the extent of damage to cellular membranes. The spectrophotometric method of Reeves and Fimognari (31) was used.

RESULTS
Purification of Venom Protein Secretagogues-For the isolation of the venom secretagogue proteins we have used essentially the same procedure as described before (28, 30) except that an additional chromatographic step on CM-cellulose equilibrated and run in 50 mM sodium phosphate buffer, pH 6.0, was included. This step improves the quality of the purified toxins, eliminating possible minor contaminants still present after the initial ion-exchange separation. The purity of toxins was verified by electrophoretic mobility using the palanine/acetate urea gel system of Reisfeld et al. (32), as shown in Figs. 1 and 13. Amino acid composition and NH2terminal amino acid sequence determination were additional criteria for assessing the purity of the secretagogue toxins, a s described next.
Chemical Churacterization of Toxins-The amino acid compositions in Table I show that the toxins are very similar peptides with respect to their molecular weights, half-cystine content, and high percentage of basic amino acid residues.
* Numbers in parentheses represent the nearest integer.
Values for threonine and serine extrapolated to zero time. Determined as pyridylethylcysteine after reduction and alkylation Value after 48 h of hydrolysis. 'Determined from sequence data. 8 Molecular weight calculated with tryptophan from sequence data (recoveries varied from 65 to 95%). and assuming 8 half-cystines. further resolved on CM-cellulose at pH 4.7 (28). The chromatographic profiles of toxin 11-11 of fraction I1 and 111-10 of fraction I11 are eluted at the same ionic strength. This indicates that, on the basis of chromatographic behavior and amino acid composition, these two components are indistinguishable and confirms our original findings (27, 28). Also, the amino acid sequences of both components , as shown next, are identical. Furthermore, these sequences are not distinguishable from that of toxin gamma previously described (5,28). For these reasons, we conclude that toxins 11-11, 111-10, and gamma correspond to the same venom component. Toxin 111-8 has 62 amino acid residues whereas toxin IV-5 has 60, which correspond to molecular weights of 6957 and 6918, respectively. However, their amino acid sequences are distinct (Fig. 2, A and B, of the Miniprint) and also different from toxin gamma, as shown in Fig. 3. The complete amino acid sequence of toxin 111-8 is shown in Fig.  2.4; Fig. 2B shows the sequence for toxin IV-5. These sequences were obtained by automated Edman degradation procedures (9) on the whole proteins following reduction and alkylation of the half-cystine residues (27,28) and by sequence determination of peptides generated by enzymatic cleavage, as indicated in the legends to Figs. 2 and 4 of the Miniprint. Fig. 4 shows the separation of some of the cleaved peptides by reverse-phase HPLC. Small differences also occur between the amino acid compositions shown in Table I and the sequences reported in Fig. 3. The amino acid sequence determined for toxin IV-5 has 1 tyrosine residue more than that calculated by amino acid analysis. This result is not unexpected, since hydrochloric acid hydrolysis of proteins is known to result in a loss of tyrosine. Similar results were obtained for a tyrosine in toxin 111-8. Since the half-cystine content reported in Table I was  values for half-cystine were taken from amino acid sequence data. The COOH-terminal amino acids cystine and lysine were determined from analyses obtained from a time course hydrolysis of reduced and carboxymethylated toxins (111-8 and IV-5). Although the partial (28) and the full amino acid sequences for toxin 111-10 (30) were reported earlier, its sequence determination was repeated in the present work (data not shown).
Sequence Comparisons-In Fig. 3, we compare the sequences of three major toxins from the venom of T. serrulatus.
In comparing these amino acid sequences, artificial gaps (represented by series of dashes) were introduced to enhance similarities. The numbers above the sequences show the 65 possible positions. Boxes enclose areas in which the amino acids are identical. Although a minor segment of the COOHterminal region of toxin IV-5 is still unknown, we have included the expected amino acids, based on the amino acid composition shown in Table I.
Close inspection of these sequences shows that only 21 amino acids are located in identical positions in all three sequences. If we compare toxin gamma with 111-8, there is 69% similarity (49 amino acid residues out of 65 total), whereas toxin gamma compared with IV-5 shows 38% identity

Direct Secretory Effect of Toxins on Pancreatic Lobules-
We determined that venom fractions that were lethal in mice were also the most potent secretagogues in guinea pig pancreatic lobule pulse-chase experiments (10, unpublished data). Therefore, our attention in purification and biochemical characterization of venom proteins for studies of pancreatic secretion was focused on similar fractions. I n uitro dose-response experiments showed that pancreatic lobules respond to stimulation of exocrine discharge by the three purified T. serrulatus toxins a t levels comparable with the effect elicited by the secretagogue carbamylcholine. In At the lowest toxin concentration of 10"' M, secretagogue activity drops to control levels of 5-7% discharge of 3H-secretory proteins. Each of the three toxins produces a characteristic dose-response curve for discharge effect, with toxins 111-8 and IV-5 producing a greater stimulation of pancreatic secretion than did toxin 111-10 at the two highest dose levels. Maximal stimulation of 47% 3H-protein discharge is achieved by toxin 111-10 at a lower dose M), but it never elicits the higher levels of stimulation, 58 and 54%, produced by toxins 111-8 and IV-5, respectively. Although these values are not statistically different at the 95% confidence interval, these differences are reproducible from experiment to experiment. When more toxin 111-10 is present, the secretagogue effect is diminished and does not return to maximal stimulation at the highest dose of M. We have tentatively identified this portion of the dose-response curve as the hypersecretory limb. These results were consistent through repeated experiments (see the legend to Fig. SA). In averaged 50% of the total. The dose-response curve for carbamylcholine (not shown) also demonstrates a hypersecretory limb similar to that determined for toxin 111-10.
Measurement of lactate dehydrogenase activity as an indicator of membrane integrity showed that no leakage of this cytosolic marker enzyme occurred during incubation with the toxins (data not shown). Thus, the secretagogue effects presented here are due to a true stimulation of the exocrine discharge mechanism mediated by the action of each of the toxins on pancreatic acinar cells.
Light dose-response concentration of M was chosen for all three toxins for purposes of comparison. However, according to the secretion curves in Fig. 5, a dose of M is not one that produces a maximum effect for toxin 111-10 but is in the hypersecretory range of activity. In these morphological studies, a separate group of lobules was incubated in a concomitant radiolabeling experiment with the same doses of toxins. From these parallel data we were able to estimate the discharge of secretory protein in the nonradiolabeled flasks under the same incubation conditions.
In control experiments as shown in Fig. 6a, the unstimulated pancreatic acinar cells demonstrate abundant zymogen granules in the apical and Golgi areas of all cells. Cytological enzymatic activity assays performed in parallel with the radiolabeled secretory protein measurements, the dose-response curves for amylase discharge in Fig. 5B retained the same unique features for each of the three toxins. Levels of amylase discharge, however, were never greater than 43% even at the highest doses of toxins.
All experimental discharge assays included measurements of activity stimulated by the muscarinic agonist carbamylcho- Caused by Scorpion Toxins organization is maintained with detectable cellular damage resulting only from the 3-h incubation or fixation procedures. Base-line secretion levels averaged 7%. When exposed to toxin 111-8 at optimal levels of stimulation, the acini in Fig.  6b have lost a large number of zymogen granules with a discharge of labeled secretory protein at 58%. The "scalloped" surfaces of basal areas seen in Fig. 6a are no longer present, and acini seem to be flattened against one another with little or no distinct space between them. In Fig. 6c, under similar stimulation conditions, but with toxin 111-10, zymogen granules are either reduced in number or even absent in some acinar cells in accordance with a 54% secretion of labeled proteins. This lobule portion has retained some of the distinctive basal areas of normal tissue but has developed some large vacuoles (arrows) in the Golgi and basal regions not seen in control specimens. The same is true for Fig. 6d in which tissue has been incubated with toxin IV-5 with a discharge activity of 60% of labeled proteins secreted. All acinar cells shown have some depletion of zymogen granules, and large intracellular vacuoles have appeared in many. All three toxins were as potent as carbamylcholine in their secretagogue effect in vitro. In this experiment, carbamylcholineevoked release of labeled proteins was 60%.
The secretagogue effect of whole T. serrulutus venom on pancreas in uiuo is dramatic, as shown in Fig. 7. In tissue removed at death, most acini have completely discharged their zymogen granules, whereas in those acini retaining granules, the numbers are greatly reduced. No newly formed granules are apparent in the Golgi area, although condensing vacuoles may be present but not visible at the light microscope level. Electron Microscopic Images of Toxin-treated Pancreatic Lobules-In Fig. 8, portions of four cells are shown from an acinus in control tissue incubated for 3 h in vitro. The apical region is densely packed with mature zymogen granules, and the acinar lumen is compressed with numerous microvilli. Discharge of secretory enzymes has not obviously occurred because of the lack of content in this lumen. As seen at the light microscope level, the cells have retained their structural integrity. The lobules in Figs. 9 and 10 have been incubated with toxin 111-8 but are from different specimen blocks of tissue. Most of the zymogen granules in Fig. 9 are discharged, but a number of condensing vacuoles in different stages of maturity remain in the two large Golgi complexes close to the apical plasmalemma. Present in the lumen are both secretory material and membranous debris. The luminal surface shows partial effacement, with the loss of microvilli on part of its acinar plasmalemma and on the plasmalemma of a centroacinar cell. A large acinar lumen in Fig. 10 is distended with both crystalline and fibrous secretory material and with detritus mostly in vesicular form. Microvilli are again absent on some plasmalemma1 surfaces. Endoplasmic reticulum cister-nae are dilated in some areas and the zymogen granule population is reduced. Cellular damage is evident in Fig. 11 in acinar cells modified by toxin 111-10. Zymogen granules are few in number, as are condensing vacuoles. A large amount of fibrous and crystalline secretory material, as well as membranous vesicles, can be seen in the lumen. There is apparently little or no effacement of the acinar luminal surface. The large Golgi complex no longer retains stacks of flattened cisternae but consists mainly of vesicles and some rigid lamellae. In another cell to the right, endoplasmic reticulum cisternae are grossly dilated and may possibly sustain some damage. In Fig. 12 (toxin IV-5), a large acinar lumen distended with secretory content and membranous debris is dramatic evidence that stimulation of these acinar cells has occurred. The few zymogen granules left undischarged are clustered at the apical regions and appear normal in size and shape. Golgi complexes have retained stacks of flattened cisternae, and endoplasmic reticulum cisternae are distended but not disrupted.

DISCUSSION
The amino acid sequences of Figs. 2, A and B, and 3 show that these toxins have some similarities in their primary structures. Some gaps have to be included to align the half- cystine residues in the same relative positions. It has been suggested that, if the tertiary structure of toxins is conserved by maintaining the position of the disulfide bridges and the a-helix and p-pleated sheet regions, then the same physiological effects are more likely to be expressed (22, 26, 30). This seems to be the case for the toxins of the Brazilian scorpion T. serrulatus. The X-ray diffraction studies conducted with variant 3 from Centruroides sculpturutus (111, a North American scorpion, and with toxin I1 from Androctonus australis Hector, a North African scorpion (12), show that the most prominent structural features highly conserved in these toxins are an a-helical region comprising residues 19-28, a threestranded antiparallel p-sheet containing residues 2-4, 32-37, and 45-51, and three of the four disulfide bridges. The comparison shown in Fig. 3 suggests that similar identical amino acid sequences are conserved in toxins from T. serrulatus.
Further similarities among these toxins are shown by our results on the secretory effect of these toxic peptides in uitro.
Comparisons between control and toxin-stimulated tissue, as described here, reflect qualitative differences. The purpose of this paper is not to present evidence for morphological changes leading to pancreatitis but to verify the potential of the toxins as secretagogues. However, references to some changes other than the discharge of zymogen granules show that there may indeed be instances in toxin-treated tissue of early events related to acinar cell damage leading to the development of pancreatitis. In three of the very few electron microscope studies presented on clinically diagnosed human acute pancreatitis (2, 14, 18), there are manifestations of dilated acinar lumina with some loss of microvilli, the development of intracellular vacuoles, and the presence of intracellular debris and degenerated organelles in lumina. These cellular events also develop during the course of induced interstitial pancreatitis in the rat with hyperstimulatory doses of the peptide secretagogue caerulein (1, 19). Our initial interests in scorpion toxin-mediated pancreatic secretion were based on reports by Waterman (38) and Bartholomew (4), who found that patients stung by the Trinidadian scorpion Tityus trinitatis developed clinical presentations of acute pancreatitis. Machado and da Silveira (21) were able to induce characteristic pathological lesions of acute pancreatitis in dogs administered sublethal doses of T. serrulatus whole venom. Further investigations in this laboratory on the role of Tityus venoms in the causation of pancreatitis must be based on knowledge concerning the mechanism of action of the component protein toxins.
Toxin gamma has been characterized in our work with isolated giant squid axon as a reversible and specific blocker of the voltage-dependent sodium channel (7). Similarly, the effect of toxin gamma on sodium channels of the heart has been described (3,17,39). The two major types of ion channels known to exist in pancreatic acinar cells at this time are a Ca"-activated nonselective monovalent cation channel (K+permeable) and a Ca2+-activated K+-selective channel (24, 25).
Voltage-activated Na+ channels have been studied in insulin secreting cells (34) but have not yet been studied or demonstrated in acinar cells. Whether or not the toxins act directly on acinar cell receptors or ion channels, or indirectly through neuronal stimulation, is of interest in our understanding of how the toxins affect secretory discharge processes. If the secretagogue effect is indeed mediated by a direct modification of ion channel permeability, it is conceivable that this channel would be selective for sodium ions in a way similar to the effects already described in other systems (3,7,17,39). We are presently investigating these possibilities. Finally, another important question that remains to be answered is the possible relevance between the secretory effect of isolated toxins on pancreas and acute pancreatitis observed by whole venom envenomation. We do not know if a synergistic effect is required among several components of the venom for the development of clinical manifestations of pancreatitis.