Proc. Nati. Acad. Sci. USA Vol. 85, pp. 4910-4913, July 1988 Neurobiology Structure and synthesis of a potent glutamate antagonist in wasp venom (wasp venom toxin/quisqualate receptor) AMIRA T. ELDEFRAWI*, MOHYEE E. ELDEFRAWI*, KATSUHIRO KONNOt, NABIL A. MANSOUR*, Koji NAKANISHIt, EUGENE OLTZt, AND PETER N. R. USHERWOOD§ *Department of Pharmacology & Experimental Therapeutics, School of Medicine, University of Maryland, Baltimore, MD 21201; tDepartment of Chemistry, Columbia University, New York, NY 10027; tDepartment of Plant Protection, College of Agriculture, University of Alexandria, Alexandria, Egypt; and §Department of Zoology, The University of Nottingham, Nottingham NG7 2RD, United Kingdom Communicated by Gilbert Stork, February 19, 1988

ABSTRACT A low molecular weight toxin isolated from leagues have shown that the venom of this wasp contains a the venom ofthe digger wasp Philanthus triangulum, first noted component (termed 8-) that exhibits a number by T. Piek, is a potent antagonist of transmission at quisqua- of pharmacological properties, including open-channel late-sensitive glutamate of locust leg muscle. This blockage of junctional glutamate receptors (16) and extra- philanthotoxin 433 (PTX-433) has been purified, chemically junctional glutamate D-receptors (17) of locust leg muscle, characterized, and subsequently synthesized along with two most of which are quisqualate-sensitive (18). We have puri- closely related analogues. It has a butyryl/tyrosyl/ fied this toxin from the wasp venom, identified its chemical sequence and a molecular weight of 435. Its two analogues, structure, and synthesized the pure toxin, philanthotoxin 433 PTX-343 and PTX-334 (the numerals denoting the number of (PTX-433), which is a potent inhibitor ofthe neurally evoked methylenes between the amino groups ofthe spermine moiety), twitch contraction of locust skeletal muscle. In addition, we are also active on the glutamate of the locust leg describe the synthesis of two pharmacologically active ana- muscle; PTX-334 was more potent and PTX-343 was less potent logues of this toxin (PTX-334 and PTX-343; the numerals than the natural toxin. Such chemicals are useful for studying, denote the number of methylenes between the amino group labeling, and purifying glutamate receptors and may become of the spermine moiety). models for an additional class oftherapeutic drugs and possibly insecticides. MATERIALS AND METHODS Glutamate receptors are believed to be the principal excita- Collection ofWasp Venom and Bioassay. Female Philanthus tory receptors in mammalian brain. Based triangulum F. were collected from the Dakhla oasis in the on the chemicals that activate them, they are generally great Sahara desert in Egypt in the late summer, when the divided into three major subtypes: quisqualate, N-methyl-D- population ofthis wasp is high. The wasps were restrained by aspartate, and kainate. These receptors are involved in chilling at 40C, and their venom sacs and glands with the sting development, , and neuropathology and now are apparati attached (Fig. 1) were removed and placed in liquid suggested to mediate the neurodegenerative consequences of nitrogen before being lyophilized and stored at - 20'C. To hypoxemia, status epilepticus, and Huntington's disease (1- test the biological activity of the crude venom preparation 3). There is considerable interest in developing agents that (water extract of the lyophilized venom glands), it was block glutamate receptors, particularly the N-methyl-D- injected into honeybees. Honeybee workers (1-3 weeks old) aspartate type, because of their anticonvulsant action and were restrained by chilling at 40C, then placed on their backs possible protection from ischemic brain damage (4). Inhibi- in a Lucite holder (16 bees to a holder), injected with 1 1.d of tors of the quisqualate-sensitive receptor in insect skeletal water extract of the venom glands in the ventral thorax muscle cause paralysis (5). behind the first pair of legs, and immediately transferred to Studies of glutamate receptors, in particular those using holding cages supplied with 40% sucrose solution. Controls biochemical techniques, have been made difficult by the received phosphate-buffered Ringer's solution. relative paucity of potent antagonists for these receptor HPLC Fractionation ot Venom Extracts. Venom glands proteins. Selective competitive and noncompetitive antago- were extracted with 50% (vol/vol) CH3CN in H20, and the nists of the N-methyl-D-aspartate receptor have become extracts were passed through a reverse-phase HPLC YMC- available during the past few years (6-8), but the search for ODS column 20 x 280 hnm. A 5-95% linear gradient of antagonists of the L-quisqualate-sensitive receptor only re- CN3CN/H20 containing 0.1% CF3COOH was used for 30 cently has shown signs of success (5). Quisqualate-sensitive min at a flow rate of 8 ml/min. The fraction of highest glutamate receptors are distributed widely in excitable tis- pharmacological activity was further purified on a reverse- sues of multicellular animals (9), and recent studies of the phase YMC-ODS column 4 x 280 mm, developed by 15% effects of the venoms of certain wasps and spiders on CH3CN in H20 containing 0.1% CF3COOH for 15 min at a vertebrate and invertebrate and muscle fibers sug- flow rate of 1 ml/min. gest that one source of antagonists for this class of receptor Electrophysiological Studies. The metathoracic retractor might be found in the venoms of some species of predaceous unguis nerve-miscle preparation of the locust Schistocerca arthropods (10-15). The solitary digger wasp Philanthus gregaria was dissected and mounted in a small Perspex bath triangulum F., which is a sphecid wasp that preys on as described (19). The muscle apodeme was attached to a honeybees, manufactures a venom that blocks glutamate Grass FT 10-strain gauge with a short strand ofTerylene, and receptors on locust skeletal muscle (14, 15). Piek and col- the muscle was stretched to maximal body length. The total volume of the bath was about 0.5 ml, and the contents could The publication costs of this article were defrayed in part by page charge be replaced within 1 s. The dissection and setting up proce- payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Abbreviation: PTX, philanthotoxin. 4910 Downloaded by guest on October 1, 2021 Neurobiology: Eldefrawi et al. Proc. Nati. Acad. Sci. USA 85 (1988) 4911 The extract of each batch of 1000 venom glands was fractionated by reverse-phase HPLC, and 30 fractions were collected. Each of the 30 fractions was tested for pharmaco- logical activity on the locust nerve-muscle preparation by using reduction in neurally evoked twitch amplitude as the measure of activity. Ten fractions were pharmacologically active. The most active fraction was the one collected at retention time 13 min (hatched peak in Fig. 2A). Further purification ofthis fraction by reverse-phase HPLC gave four peaks (Fig. 2B); the most pharmacologically active was the major peak (hatched peak in Fig. 2B). This fraction gave 1.1 mg of toxin amorphous powder. The UV spectrum of this fraction, PTX433 (compound 1 of Fig. 3A) had a maximum at 274 nm, which shifted to 290 nm at pH 12, suggesting the presence of a tyrosine residue. This was supported by 1H NMR (250 MHz in 2H20), 6 3.00 (2H, d, J = 7.8 Hz), 4.43 (1H, t, J = 7.8), 6.88 (2H, d, J = FIG. 1. Venom sac (VS), gland (VG), and the sting apparatus 8.7 Hz), 7.18 (2H, d, J = 8.7 Hz). The presence of a butyryl (Stg) dissected from Philanthus triangulum. group was also clear from 1H NMR, 8 0.83 (3H, t, J = 7.2 Hz), 1.57 (2H, quin, J = 7.2 Hz), 2.26 (2H, t, J = 7.2 Hz). dure were performed in continuously flowing saline. The The 1H NMR signals corresponding to six methylenes a to muscle was stimulated indirectly through fine (40-80,um) nitrogen at 8 3.0-3.3 (12H, m) and four methylenes f3 to platinum wire electrodes, insulated to their tips, and placed nitrogen at 8 1.4-1.6 (4H, m) and 2.1-2.2 (4H, m) (20), on the retractor unguis nerve. The venom fractions were together with the FAB-fragment MS [M + M]+ peak at m/z dissolved in locust saline of the following composition: 180 436, showed the remainder ofthe molecule to be a polyamine mM NaCI/10 mM KCI/2 mM CaCI2/10 mM Hepes, pH 6.8. of the spermine type. 1H NMR measured in perdeuterated The nerve muscle preparation was perfused with this saline dimethyl sulfoxide (500 MHz) clarified the connectivity ofthe at a flow rate of 5-10 ml/min at 19°C. butyryl, tyrosyl, and polyamine moieties-namely, two am- ide protons were observed at 8 7.82 and 7.86 as a doublet and RESULTS triplet, respectively, indicating that the former was due to tyrosine and the latter to polyamine. This led to a butyryl/ Honeybees injected with water extracts of the venom glands tyrosyl/polyamine sequence as shown in compounds 1, 2, were paralyzed in a dose-dependent manner. Time for and 3 of Fig. 3A, but since spectroscopic evidence did not recovery from paralysis was 15 ± 3 min and 55 ± 8 min for distinguish among these three possibilities, all isomers were bees injected with 0.2 and 1.2 venom units (a unit is the synthesized. extract from one wasp gland), respectively. Although all bees Chemical synthesis ofthe three isomers is illustrated in Fig. recovered within 1 hr, a dose-dependent mortality was 3 B and C. Details will be reported elsewhere. The protected evident after 24 hr (30%, 80%, and 100%o mortality for bees polyamine 6 was obtained from derivative 4 (21) injected with 0.4, 0.8, and 1.2 venom units). Polyacrylamide through (Fig. 3B) (i) Michael addition to acrylonitrile (76%), disc gel electrophoresis showed that the water extract of the (ii) tert-butoxycarbonyl (Boc) protection (81%), and (iii) venom glands contained a large number of proteins, all of reduction of the nitrile (70%). Further carbobenzoxy protec- which were precipitated by heating the extract at 100'C for 10 tion and Boc-deprotection of 6 yielded partially protected min. The boiled extracts retained full activity when assayed polyamine 7. Deprotection of N-Boc-O-benzyl-L-tyrosine on the locust nerve-muscle preparation or on honeybees. p-nitrophenyl ester 8 (Fig. 3B) with CF3COOH followed by B A PTX-433 3.0 A

2.5: 2 E C C4 2.0- e4J a co 1.5

0.5

0 min min FIG. 2. Fractionation of Philanthus venom by reverse-phase HPLC. (A) Fractionation of lyophilized venom glands, extracted with 50%6 CH3CN in H20; 450 ,ul (representing extracts of 225 wasps) was chromatographed on a YMC-ODS 20 x 280 mM column and developed with a linear aqueous gradient of5% CH3CN/0.1% CF3COOH to 95% CH3CN/0.1% CF3COOH for 30 min at a flow rate of8 ml/min. UV absorption was monitored at 215 nm. (B) Fractionation of the main toxic fraction (hatched peak in A). Downloaded by guest on October 1, 2021 4912 Neurobiology: Eldefrawi et al. Proc. Natl. Acad. Sci. USA 85 (1988)

A CH gested that its action on a locust nerve-muscle preparation was both time- and concentration-dependent. The effects of this toxin on the neurally evoked twitch contraction of the locust retractor unguis muscle (19) were investigated with O H toxin concentrations of 1-10 uM. It was clear from the data 1 PTX-433 NN NN' N-NH NH2 presented in Fig. 4A that PTX-433 exerted a number of H a: H H actions on the locust nerve-muscle syst~m! There was an BuTyr initial reduction in twitch amplitude, which was stimulus- frequency independent. This was followed by a further reduction in the twitch height, the extent of this change being H H H directly proportional to the frequency at which the retractor 2 PTX-334 N N. > unguis nerve was stimulated. Prolontged applications of BuTyr' _ NH2 PTX-433 abolished the twitch. Immediately after removal of the toxin, there was a brief period of repeated ahd prolonged H H contractions in response to a single stimulus applied to the N 3 PTX-343 BuTyr' N N N- NH2 retractor unguis nerve before the twitch slowly returned to H normal. PTX-433 at 10 AM also reduced the response of the retractor unguis muscle to 0.1 mM glutamate by >90%, B which suggests that at least part of the reduction in twitch amplitude was due to the antagonism ot postjunctional, Boc b Buc Boc quisqualate-sensitive glutamate receptors. PTX-334 (Fig. 4B) 5b and PTX-343 influenced the twitch contraction in the same 4 qualitative fashion as PTX-433. In these preliminary studies, 5 PTX-334 seemed to be more potent than PTX-433, but Boc Boc H H PTX-343 was apparently less potent. H2N, N N NHBOC - ' CbzHN%-N. 7 N. NHW 6 7 DISCUSSION I 2 The natural philanthotoxin and its synthetic counterpart PTX-433 and analogs PTX-334 and PTX-343 (Fig. 3) repre- sent an additional class of chemicals that are active biolog- a. b. reagents: CH2-CHCN; (Boc)20; C. LiAIH4; ically and possibly inhibit allosterically the quisqualate- d. CbzCVEt3N; TFA sensitive glutamate receptor in insect skeletal muscle (Fig. 4). They are smaller in molecular weight (Mr, 435) than the toxins isdlated from orb web spider venoms, the argiotoxins (Mr, >600) and easier to synthesize. There are some structural C similarltfes between the philanthotoxins and argiotoxins- i.e., with respect to the presence of a chromophore and a polyamine. However, there are sufficient structural (12, 13, 7. H/Pd-C 23-25) and pharmacological (17, 23) differences between the philanthotoxins and the argiotoxins to warrant their separate a, b 6, TFA, H/Pd-C investigation. 2 The argiotoxins may exhibit some specificity for quisqua- spermine, HiPd-C late-sensitive glutamate receptors, and as such they are 0 N02 3 9 A

* * reagents. a. TFA; b. BuCVEt3N II,10!l11fl ~ I.

* FIG. 3. The chemical structures and synthesis bf the natural philanthotoxin, PTX-433, and two isomers, PTX-334 and PTX-343. (A) The structures of the three toxins. (B) Synthesis ofinterrhediates 25 pM PTX-433 100 S of compounds 1 and 2. (C) The final steps in synthesis of the three toxins.

acylation with butyryl chloride gave key intermediate 9 in 85% yield. Coupling of 9 with protected polyamines 7 and 6 and commercial spermine (1,12-diamino-4,9-diazadodecane) (22) at ca. 65% yield was followed by deprotection, gi'ing 25 pM PTX-334 PTX-433 and analogs 2 and 3 at cd. 80% yield. Synthetic material derived from 7 was found to be identical with the FIG. 4. Effects of PTX-433 (A) and PTX-334 (B) on the neurally natural product in all respects (1H NkIR, MS, CD, HPLC, evoked twitch contraction of locust metathoracic retractor unguis arid biological activities). Thus, the chemical structure of the muscle. A and B are data from different nerve-muscle preparations major naturally occurring philanthotoxin is 1, which we dissected from the same adult female locust (Schistocerca gregaria). The nerve-muscle preparations were supelfuged with standard locust designated PTX-433, the numerals denoting the number of saline (23) for 30 min before the toxins were applied. The retractor methylene groups between the amino groups of the spermine unguis nerve was stimulated with a single, brief(0.1 s), supramaximal moiety from left to right. All three of the synthetic end stimulus applied at a constant lowfrequency (0.1 Hz) before and after products were biologically active, PTX-334 having a higher toxin application (in locust saline); however, during the period of potency than the natural PTX-433 toxin. toxin application, the stimulation frequency was sometimes in- Preliminary pharmacological studies with PTX-433 sug- creased temporarily to 0.6 Hz (*). Downloaded by guest on October 1, 2021 Neurobiology: Eldefrawi et al. Proc. Natl. Acad. Sci. USA 85 (1988) 4913 attracting the interest of both the pharmaceutical and pesti- 6. Watkins, J. C. & Evans, R. H. (1981) Annu. Rev. Pharmacol. cide industries as well as academic neuroscientists seeking Toxicol. 21, 165-204. probes for studying this class of glutamate receptors. It 7. Meldrum B. (1987) in Neurotoxins and Their Pharmacological Implications, ed. Jenner, P. (Raven, New York), pp. 133-152. remains to be established whether the philanthotoxins are 8. Kemp, J. A., Foster, A. C. & Olverman, H. J. (1987) Trend specific antagonists of only the quisqualate-sensitive gluta- NeuroSci. 10, 265-272. mate receptor in insect muscle or whether they are as potent 9. Usherwood, P. N. R. (1978) Adv. Comp. Physiol. Biochem. 7, when applied to vertebrate central quisqualate receptors. It 222-309. is also of significant interest to determine if these philantho- 10. Kawai, A., Miwa, T. & Abe, T. (1982) Brain Res. 247, 169-171. toxins inhibit the N-methyl-D-aspartate receptor. More than 11. Boden, P., Duce, I. R. & Usherwood, P. N. R. (1984) J. Br. one structure is biologically active, and modifications oftheir Pharmacol. 83, 221P. 12. Bateman, A., Boden, P., Dell, A., Duce, I. R., Quicke, D. L. J. structures may produce selective antagonists of the different & Usherwood, P. N. R. (1985) Brain Res. 339, 237-244. types of glutamate receptors. 13. Grishin, L. G., Voldova, T. M., Arsoniev, A., Reshetova, A. S., Onorprienko, V. V., Magazanic, L. G., Antonov, S. M. Note Added in Proof. Since submitting this manuscript, we have & Fedorova, I. M. (1986) Bioorg. Khim. 12, 1121-1124. noted the announcement at a meeting in the United Kingdom in April 14. Piek, T., Mantel, P. & Engels, E. (1971) Comp. Gen. Pharma- 1988 by T. Piek (Farmacologisch Laboratorium, Universitat van col. 2, 317-331. Amsterdam, The Netherlands) of the structure of 8-PTX, which is 15. Piek, T. & Njio, K. D. (1975) Toxicon 13, 199-201. identical to that of PTX-433. 16. Piek, T., Mantel, P. & Jas, H. (1980) J. Insect Physiol. 26, 345- 349. 17. Clark, R. B., Donaldson, P. L., Gration, K. A. F., Lambert, We thank Drs. Mamdouh Idriss and Shebl Sherby of the Univer- J. J., Piek, T., Ramsey, R. L., Spanjer, W. & Usherwood, sity ofAlexandria, Egypt, for their help in the collection of wasps and P. N. R. (1982) Brain Res. 241, 105-114. dissection. The studies were supported in part by National Science 18. Gration, K. A. F., Clark, R. B. & Usherwood, P. N. R. (1979) Foundation Grant INT-8610138 (M.E.), National Institutes ofHealth Brain Res. 171, 360-364. Research Grants Al 10187 (to K.N.) and ES 02594 (to A.T.E.), North 19. Usherwood, P. N. R. & Machili, P. (1968) J. Exp. Biol. 49, 341- Atlantic Treaty Organization Research Grant 662/86 (to P.N.R.U. 361. and A.T.E.), and a grant from Science and Engineering Research 20. Ohshima, T. (1979) J. Biol. Chem. 254, 8720-8722. Council (U.K.) (to P.N.R.U.). 21. Humora, M. & Quick, J. (1970) Org. Chem. 44, 1166-1168. 22. Hashimoto, Y., Endo, Y., Skudo, K., Aramaki, Y., Kawai, N. 1. Cotman, C. W. & Iversen, L. L. (1987) Trends Neurosci. 10, & Nakajima, T. (1987) Tetrahedron Lett. 28, 3511-3514. 263-265. 23. Budd, T., Clinton, P., Dell, A., Duce, I. R., Johnson, S. J., 2. Robinson, M. B. & Coyle, J. T. (1987) FASEB J. 1, 446-455. Quicke, D. L. J., Taylor, G. W., Usherwood, P. N. R. & 3. Silverstein, F. S., Torke, L., Barks, J. & Johnston, M. V. Usoh, G. (1988) Brain Res., in press. (1987) Dev. Brain Res. 34, 33-39. 24. Adams, M. E., Candy, R. L., Enderlin, F. E., Fu, T. E., 4. Wong, E. H. F., Kemp, J. A., Priestley, T., Knight, A. R., Jarema, M. A., Li, J. P., Miller, C. A., Schooley, D. A., Woodruff, G. N. & Iversen, L. L. (1986) Proc. Nati. Acad. Sci Shapiro, M. J. & Venema, V. J. (1987) Biochem. Biophys. Res. USA 83, 7104-7108. Commun. 348, 678-683. 5. Usherwood, P. N. R. (1987) in Neurotoxins and Their Phar- 25. Aramaki, Y., Yasuhara, T., Higashijima, T., Yoshioka, M., macological Implications, ed. Jenner, P. (Raven, New York), Miwa, A., Kawai, N. & Nakajima, T. (1986) Proc. Jpn. Acad. pp. 133-152. Ser. B 62, 359-362. Downloaded by guest on October 1, 2021