Tetanus and Botulinum Neurotoxins Are Zinc Proteases Specific for Components of the

Neuroexocytosis Apparatus" GIAMPIETRO SCHIAVO,b ORNELLA ROSSETTO,b

FABIO BENFENATI,' BERNARD POULAIN,d AND CESARE MONTECUCCOb

bCentro C.N.R. Biomembrane and

Dipartimento di Scienze Biomediche Sperimentali

Universita di Padova 35121 Padova, Italy

CDipartimento di Medicina Sperimentale Universita di Roma "Tor Vergata"

Roma, Italy

dLaboratoire de Neurobiologie Cellulaire et Moleculaire CNRS

91 198 Gij-sur- Yvette, France

INTRODUCTION

Tetanus neurotoxin (TeTx) is the causative agent of tetanus, an often fatal spastic paralysis due to the block of neurotransmitter release of the spinal cord inhibitory interneurons.' Botulinum neurotoxins (BoNT, seven different serotypes: A, B, C, D, E, F, and G) cause the flaccid paralysis of botulism via inhibition of acetylcholine release at the neuromuscular junction.' Hence, the opposite clinical symptoms of tetanus and botulism result from the different sites of action of the toxins, with the same cellular mechanism of intoxication.

TeTx and BoNT are produced by toxigenic Clostridia as a single chain, cleaved by proteases at an exposed loop to generate the active di-chain protein (H chain,

This work was supported by grants from the Telethon-Italia and from the CNR Target Project on Biotechnology and Bioinstrumentation.

65

66 ANNALS NEW YORK ACADEMY OF SCIENCES

JJ proteases ( K I O - ~ )

I_

4 L F -------_I a 50 - , ‘.

e . 0 , ; e .

H E h u H

FIGURE 1. Schematic structure and zinc-binding motif of tetanus and botulinum neuro- toxins. The upper part shows the schematic structure and SDS-PAGE profile of the clostrid- ial neurotoxins of tetanus and botulism (H refers to the heavy chain, 100 kDa; L refers to the light chain, 50 kDa). The lower part reports in amino-acid single-letter code the most homologous segment between the seven clostridial neurotoxins (TeTx, BoNT serotypes A, B, C, D, E, and F) whose sequence is available, which contains the zinc-binding motif HEhuH of matrix zinc-endopeptidases, aligned just below (h for hydrophobic; u for uncharged; x for any residue).

100 kDa; L, 50 kDa), depicted in the left part of FIGURE 1. The H chain is involved in the neurospecific cell binding and penetration of the toxins, whereas the L chain is responsible for the block of the neurotransmitter release.’,*

Recently, the c-DNA deduced amino-acid sequences of six BoNT serotypes have been determined and aligned with that of tetanus toxin.3 Despite the very similar mechanism of action, these neurotoxins show a n overall low degree of homology with few segments of close similarity. We noticed that the most conserved segment, located in the central region of the L chain, includes the His-Glu-x-x-His zinc-binding motif of metallo-end~peptidases.~

Here, we show that TeTx and BoNT d o indeed contain a zinc atom, that the zinc ion is essential for TeTx inhibition of neurotransmitter release in Aplysia californica neurons, and that TeTx, BoNT/B, and BoNT/F cleave specifically VAMP/synapto- brevin, a protein component of the small synaptic vesicles.

MATERIALS AND METHODS

TeTx and BoNT were obtained as described p r e v i o u ~ l y ~ . ~ and, before metal deter- mination, were extensively dialyzed at 4°C against 150 mM Tris-chloride (pH 7.4)

SCHIAVO et al.: TETANUS & BOTULISM 67

or 10 mM HEPES-Na and 100 mM sodium chloride (pH 7.0) by using metal-free dialysis tubing (Spectrum, Texas). Samples were analyzed for Zn, Co, Cu, Fe, Mn, and Ni with a Perkin Elmer 4000 atomic absorption flame spectrophotometer with impact bed loading.

Electrophysiological recordings were performed on buccal ganglia of the sea slug Aplysia californica as detailed before.’

Small synaptic vesicles (SSV) were isolated from rat cerebral cortex.* Thirty pg of SSV in 30 pL of 5 mM HEPES-Na, 0.3 M glycine, 0.3 M NaCI, and 0.02% NaN3 was treated for 60 minutes at 37°C with 80 ng of native toxin or toxin previously reduced with 10 mM dithiothreitol for 30 minutes at 37°C. Samples were dissolved in 8% SDS, 10 mM Tris-acetate (pH 8.2), and 0.1 mM EDTA and were applied on 13-18070 polyacrylamide gradient gels. The time courses of VAMP proteolysis and production of 12-, 8-, and 7-kDa fragments were determined on Coomassie blue- stained gel by scanning with a dual-wavelength Shimadzu CS-630 densitometer. For sequence analysis, proteins were electroeluted from polyacrylamide gels, applied to ProSpin tubes (Applied Biosystem, California), and sequenced in a pulsed liquid Applied Biosystem Model 477A protein sequencer.

Toxicity of tetanus toxin was determined by injecting intraperitoneally serial dilu- tions of the toxin with phosphate-buffered saline containing 0.1 Yo BSA. Time before death was recorded and mouse lethal doses were calculated as de~cribed.~

RESULTS AND DISCUSSION

Metal Content of Clostridial Neurotoxins

As shown in FIGURE 2, atomic absorption analysis of different clostridial neurotox- ins gave a range of zinc content between 0.75 and 1.05 gramoatoms of zinc ion per toxin mole.’0-’2 A metal content below an integer value is frequently found with metalloproteins due to metal loss during protein purification. Co, Cu, Fe, Mn, and Ni were below detection limits. The zinc atom was associated only with the L chain, the one responsible for the intracellular blockade of neurotransmitter release. Upon incubation with heavy metal chelators, such as EDTA and o-phenanthroline, the neurotoxins release most of their zinc content. This process is perfectly reversible and occurs without protein denaturation because the apotoxins regain their zinc content upon incubation in Zn2+ -containing medium.’0.”

In zinc-endopeptidases, histidines are involved in the metal c ~ o r d i n a t i o n . ~ J ~ - ’ ~ Chemical modification of TeTx and botulinum toxin A, B, and E with diethylpyrocar- bonate (DEPC), a histidine-modifying reagent,I6 indicates that, also in clostridial neurotoxins, two histidines are involved in the zinc-coordination.IoJ’ Parallel chemi- cal modification experiments demonstrate that cysteines do not participate in the zinc-binding.I0.” On the basis of what we know about zinc-proteins, these findings suggest that the metal atom plays a catalytic (rather than a structural) role in clostridial neurotoxin^.^ Dot-blot experiments with radioactive zinc suggest that the zinc atom is bound to the central histidine-rich conserved region of the L chain, containing the zinc-binding motif HExxH.

These results define the clostridial neurotoxins as the first class of metal lot ox in^.'^

ANNALS NEW YORK ACADEMY OF SCIENCES

0) 1.0 - E \

0 n 0

E CI

v

c 0.5 .- Y 0

E N

0

T

T L EDTA A B E I F

TeTx BoNT

FIGURE 2. Zinc content, measured by atomic absorption of clostridial neurotoxins (T, di-chain TeTx; L, L chain of TeTx; A-F, di-chain BoNT). Amount of zinc, in gramoatoms per mole of protein + SD, of four different toxin preparations incubated in the absence (empty columns) or presence of EDTA (dotted column). Co, Cu, Fe, Mn, and Ni were also assayed and were found to be below detection limits.

Tetanus Toxin Inhibition of Neurotransmitter Release in Aplysia Neurons Is Zinc-dependent

The best-characterized system for testing the intracellular activity of clostridial neurotoxins is constituted by the identified cholinergic couples of neurons making synapses in the buccal ganglion of Apbsiu c~lifornicu.~ In this cellular system, it could be directly demonstrated that the L chain of tetanus toxin is the intercellularly active portion of the FIGURE 3A shows that both the native L chain and the Znz+-depleted L chain of tetanus toxin were effective in inhibiting the evoked neurotransmitter release, eventually reaching the same level of inhibition. However, the kinetics of inhibition of the Znz+-depleted L chain was much slower. In contrast, if the Zn2+-depleted L chain was injected in a presynaptic neuron pretreated with heavy metal chelators,I9 no inhibitory activity was recorded for at least 3 hours (filled triangles of FIGURE 3A). These experiments suggest that the Zn2+-depleted L chain is able to slowly regain its zinc ion from the cytoplasmic pool and that bound ZnZ+ is necessary for the L chain inhibition of neurotransmitter release.

The zinc atom of zinc-proteases is directly involved in the cleavage of the peptide b~nd.~**O Phosphoramidon is a very specific inhibitor of zinc-endopeptidases that interacts directly with the zinc atom in the active site. FIGURE 3B shows that the injection of this compound into Aplysia neurons completely prevented the TeTx L chain effect on neurotransmitter release, whereas phosphoramidon alone did not

SCHIAVO et d.: TETANUS & BOTULISM 69

100

s - Q) 50 a U c .- - F m

metal chelators

2. *. ..* . *.. L - c h a i n . . ... .. -...

2.0 . *' :....* . ... .

0 1

- 60 0 60 120 180

FIGURE 3. Inhibition of acetylcholine release induced by intracellular application of the L chain of tetanus toxin. Couples of neurons making identified cholinergic synapses in the buccal ganglion of Aplysiu culifornicu were used. TeTx L chain samples were injected at time zero. Panel A: Native L chain (@) and its Zn2+-depleted form in the presence (A) or absence (0) of Zn2+ chelators. Panel B: Effect of phosphoramidon on the intracellular Ap/ysiu neuron activity of the L chain of tetanus toxin.

affect neurotransmission. The small inhibition of neurotransmitter release detectable long after toxin injection is expected because of diffusion of phosphoramidon out of the nerve cell and dissociation of the toxin-inhibitor complex. Given the high specificity of phosphoramidon inhibition,*' this result is a strong, although indirect, evidence that tetanus toxin is a zinc-endopeptidase.

70 ANNALS NEW YORK ACADEMY OF SCIENCES

FIGURE 4. Structure and proteolytic activity of TeTx, BoNT/B, and BoNT/F on small synaptic vesicles. (A) Sequential activation of clostridial neurotoxins via (i) proteolytic processing of the single chain inactive form to generate the di-chain toxin and (ii) reduction of the interchain disulfide bond. (B) Coomassie blue-stained SDS-PAGE profile of small synaptic vesicles (SSV) alone or after incubation with 80 ng of reduced single chain TeTx (sc-TeTx), di-chain TeTx (TeTx), reduced di-chain TeTx (TeTx red), L chain of TeTx (L chain), and reduced BoNT/B and BoNT/F. The positions of VAMP and of its fragments are indicated by arrows.

Tetanus and Botulinum B and F Neurotoxins Are Proteases Specijc for VAMP

To demonstrate directly a zinc-dependent protease activity of the neurotoxins, many synthetic chromophoric peptide substrates were assayed without success. We

SCHIAVO et al. : TETANUS & BOTULISM 71

100

+VLERDQK LSELD - s v

E c

s 0 1 +DQKL E

5 0 - 0 I U >

BoNT/ F

0 0 30 60 90 120

t i m e ( m i ” )

FIGURE 5. The specific proteolysis of VAMPS mediated by BoNT/F is inhibited by peptides spanning the cleavage site. Small synaptic vesicles were challenged in the presence of reduced BoNT/F alone (0) or after incubation with the peptides DQKL (0) and VLER- DQKLSELD (B) (800 pM final concentration).

tested then several subcellular fractions of nerve tissues and cells known to be sensitive to TeTx and BoNT. A clear-cut result was obtained with small synaptic vesicles (SSV) purified from rat cerebral FIGURE 4B shows that the reduced single chain TeTx and the unreduced di-chain TeTx are ineffective in the modification of the SDS-PAGE protein pattern of highly purified SSV. Only reduced di-chain TeTx and isolated L chain are able to induce the disappearance of a single band of 19 kDa with the concomitant appearance of 7-kDa and 12-kDa fragments. The same result was obtained with reduced di-chain BoNT/B. These results indicate that TeTx, BoNT/B, and BoNT/F need to be cleaved and then reduced in order t o set free the intracellular protease activity (FIGURE 4A). Moreover, these findings imply that clostridial neurotoxins follow the general scheme of activation of bacterial protein toxins with intracellular targets.”

By electrophoretic mobility and immunoblotting (not shown), this protein was identified as VAMP/synaptobrevin.’’ Also, reduced BoNT/F cleaved VAMP specifi- cally, but at a different peptide bond.’’ The site of cleavage of VAMP was identified at the level of the peptide bond Gln-Lys for BoNT/F and at Gln-Phe for TeTx and BONT/B.’’,~~ BoNT/A and BoNT/E d o not affect the electrophoretic mobility of any proteins of the SSV and their target(s) remains to be identified.

Peptides of various lengths, encompassing the two cleavage sites, were prepared by chemical synthesis. Some of them carried chromophoric groups to facilitate the determination of a zinc-endopeptidase activity. None of them were hydrolyzed by the three neurotoxins. This is not due to lack of binding because these substances

72 ANNALS NEW YORK ACADEMY OF SCIENCES

F Q

al Y) C 0 n ?! 50

0

FIGURE 6. The inhibition of neurotransmitter release in Aplysiu neurons injected with BoNT/B is inhibited by VAMP-2 peptides ASQFETS and QFET. Presynaptic neurons were injected (arrow) with BoNT/B (intracellular concentration of 8 nM) alone (A) or were preincubated with peptides [ASQFETS (0); QFET (A); intracellular concentration, 80 pM].

are very effective inhibitors of the toxin activity both in vitro and in vivo (see below). As an example, FIGURE 5 shows that the peptides DQKL and VLERDQKLSELD, which correspond to segments 57-60 and 53-64, respectively, of rat brain VAMPS, are very effective inhibitors of the VAMP-specific proteolysis mediated by BoNT/F.

Only the VAMP-2 isoform of rat brain was cleaved by the TeTx and BoNT/B, whereas BoNT/F cleaved both isoforms.'2.23 This is most probably due to the fact that rat brain VAMP-1 has a Val in place of the Gln residue of VAMP-2 at the site of TeTx and BoNT/B cleavage. The same situation is present in chickens, whereas both human and mouse VAMP-1 and -2 are identical at the site of TeTx and BoNT/B cleavage. We have speculated that this may help explain the high susceptibility of humans and mice and the resistance of chickens and rats to tetanus and botulism B.24

Two lines of evidence indicate that these neurotoxins act intraneuronally by a specific recognition and cleavage of VAMP/synaptobrevin. FIGURE 6 shows that peptides ASQFETS and QFET, spanning the VAMP cleavage site of TeTx and BoNT/B, injected into Apfysiu neurons, effectively inhibit the intracellular activity of BoNT/B. A search in the sequence data banks shows that the sequence ASQFETS is present only in VAMP. FIGURE 7A shows that a polyclonal antibody, highly specific for VAMP, injected into Aplysia neurons does not interfere with neuroexocytosis. Moreover, it blocks completely the activity of TeTx and BoNT/B added in the extra-

SCHIAVO et al.: TETANUS & BOTULISM 13

0 1 2 3 4

BoNTl B

I 1 I - i o 1 2 3

0

BoNT/A

- i o i 2 3 time ( h )

FIGURE 7. The intracellular injection of anti-VAMP IgG prevents the inhibitory action of BoNT/B, but not of BoNT/A, on the neurotransmitter release in ApIysia californicu neurons. (A) Anti-VAMP IgG does not inhibit the evoked neurotransmitter release in Aplysiu neurons. (B & C) Anti-VAMP IgG was injected into a single presynaptic neuron (O), whereas the other was taken as a control (0). After one hour, BoNT/B (panel B) or BoNT/A (panel C) was bath-applied at the final concentration of 40 nM and postsynaptic responses were recorded as previously described.'

74 ANNALS NEW YORK ACADEMY OF SCIENCES

cellular medium (FIGURE 7B), but it is ineffective with BoNT/A and BoNT/E, which have an intracellular target different from VAMPzS (FIGURE 7C). This indicates that VAMP is implicated in the impairment of neuroexocytosis brought about by these neurotoxins and that it plays a central role in neuroexocytosis. Indeed, very recently, evidence was provided that VAMP is implicated in the docking of the vesicles to their target membrane via the specific recognition of syntaxin, a protein of the presynaptic membrane.z6 This interaction is mediated by a set of soluble protein factors (NSF and SNAPS) identified by the laboratory of Rothman.”

CONCLUSIONS

Tetanus and botulinum neurotoxins are the first toxins shown to act via a zinc- dependent protease activity. TeTx, BoNT/B, and BoNT/F block neurotransmission by acting at the level of synaptic vesicles. Hence, both the catalytic activity and the target are novel. Their use has allowed the demonstration of the key role played in the process of neuroexocytosis by VAMP, and other fundamental factors will be identified with the other serotypes of botulinum neurotoxins.

[Note Added in Proof: Recently, BoNT/A and BoNT/E have been demonstrated to be zinc-endopeptidases specific for SNAP-25,28,29 a presynaptic membrane protein. Moreover, BoNT/D is specific for VAMP/synaptobrevin 1 and 2 and BoNT/C for syntaxin***’” (in preparation). Hence, all the targets of clostridial neurotoxins contrib- ute to the vesicle docking and fusion multiprotein complex proposed by Sollner et al. 26]

SUMMARY

Tetanus and botulinum neurotoxins bind to nerve cells, penetrate the cytosol, and block neurotransmitter release. Comparison of their amino-acid sequences shows the presence of the highly conserved His-Glu-x-x-His zinc-binding motif of zinc- endopeptidases (HExxH). Atomic absorption measurements of clostridial neurotox- ins show the presence of one atom of zinc/toxin molecule bound to the light chain. The toxin-bound zinc ion is essential for the neurotoxins inhibition of neurotransmit- ter release in Aplysia neurons injected with the toxins. Phosphoramidon, a very specific inhibitor of zinc-endopeptidases, blocks the intracellular activity of the clos- tridial neurotoxins. Highly purified preparations of the light chain of tetanus and botulinum B and F neurotoxins cleaved specifically VAMP/synaptobrevin, an inte- gral membrane protein of small synaptic vesicles, both in vivo and in vitro. From these studies, it can be concluded that the clostridial neurotoxins responsible for tetanus and botulism block neuroexocytosis via the proteolytic cleavage of specific components of the neuroexocytotic machinery.

REFERENCES

1. SIMPSON, L. L., Ed. 1989. Botulinurn Neurotoxin and Tetanus Toxin. Academic Press. New York.

SCHIAVO et al. : TETANUS & BOTULISM 15

2. 3.

4. 5. 6.

7.

8.

9.

10.

11.

12.

13.

14. 15.

16. 17.

18.

19.

20. 21. 22. 23.

24 I

25.

26.

27. 28.

29.

30.

MONTECUCCO, C. 1986. Trends Biochem. Sci. 11: 314-317. NIEMANN, H. 1991. In A Sourcebook of Bacterial Protein Toxins. J. E. Alouf & J. H. Freer,

Eds.: 303-348. Academic Press. New York/London. VALLEE, 13. L. & D. S. AULD. 1990. Biochemistry 29: 5647-5659. SCHIAVO, G., E. PAPINI, G. GENNA & C. MONTECUCCO. 1990. Infect. Immun. 58: 4136-4141. ROSSETTO, o., G. SCHIAVO, P. POLVERINO DE LAURETO, s. FABBIANI & c. MONTECUCCO. 1992.

Biochem. J. 285: 9-12.

& L. TAW. 1990. J. Physiol. (Paris) 84: 247-261.

1378.

338: 99-106.

EMBO J . 11: 3577-3583.

Chem. 267: 23479-23483.

Biol. Chem. 268: 11516-11519.

1972. Nature 238: 37-41.

POULAIN, B., M. MOCHIDA, J. D. F. WADSWORTH, U . WELLER, E. HABERMANN, J. 0. DOLLY

HUTTNER, W. B., W. SCHIEBLER, P. GREENCARD & P . DE CAMILLI. 1983. J . Cell Biol. 96: 1374-

WELLER, U., F. MAULER & E. HABERMANN. 1988. Naunyn-Schmiedeberg’s Arch. Pharmacol.

SCHIAVO, G., B. POULAIN, 0. ROSSETTO, F. BENFENATI, L. TAUC & C. MONTECUCCO. 1992.

SCHIAVO, G., 0. ROSSETTO, A. SANTUCCI, B. R. DASGUPTA & C. MONTECUCCO. 1992. J. Biol.

SCHIAVO, G., C. C. SHONE, 0. ROSSETTO, F. C. G. ALEXANDER & C. MONTECUCCO. 1993. J.

MATTHEWS, B. W., J. N. JANSONIUS, P. M. COLMAN, B. P. SCHOENBORN & D. DUPOURQUE.

THAYER, M. M., K. M. FLAHERTY & D. B. MCKAY. 1991. J . Biol. Chem. 266: 2864-2871. BODE, W., F. X. GOMIS-RUTH, R. HUBER, R. ZWILLINC & W. STOCKER. 1992. Nature 358: 164-

MILES, E. W. 1977. Methods Enzymol. 47: 431-442. SCHIAVO, G., B. POULAIN, F. BENFENATI, B. R. DASGUPTA & C. MONTECUCCO. 1993. Trends

MOCHIDA, S., B. POULAIN, U. EISEL, T. BINZ, H. KURAZONO, H. NIEMANN & L. TAUC. 1990.

ARSLAN, P., F. DI VIRGILIO, M. BELTRAME, R. Y. TSIEN & T. POZZAN. 1985. J. Biol. Chem.

MATTHEWS. B. W. 1988. Acc. Chem. Res. 21: 333-340. WEAVER, L. H., W. R. KESTER & B. W. MATTHEWS. 1977. J. Mol. Biol. 114: 119-132. BENFENATI, F., M. BAHLER, R. JAHN & P. GREENGARD. 1989. J. Cell Biol. 108: 1863-1872. SCHIAVO, G., F. BENFENATI, B. POULAIN, 0. ROSSETTO, P. POLVERINO DE LAURETO, B. R.

PATARNELLO, T., L. BARGELLONI, 0. ROSSETTO, G. SCHIAVO & C. MONTECUCCO. 1993. Nature

POULAIN, B., 0. ROSSETTO, F. DELOYE, G. SCHIAVO, L. TALK & C. MONTECUCCO. 1993. J .

167.

Microbiol. 1: 170-174.

Proc. Natl. Acad. Sci. U S A . 87: 7844-7848.

260: 2719-2727,

DASGUPTA & C. MONTECUCCO. 1992. Nature 359 832-835.

364: 581-582.

Neurochem. 61: 1175-1178.

TEMPST & J. E. ROTHMAN. 1993. Nature 362: 318-324. SOLLNER, T., s. w. WHITEHEART, M. BRUNNER, H. ERDJUMENT-BROMACE, s. GEROMANOS, P.

ROTHMAN, J . E. & L. ORCI. 1992. Nature 355: 409-415. SCHIAVO, G., 0. ROSSETTO, S. CATSICAS, P. POLVERINO DE LAURETO, B. R. DASGUPTA, F.

BENFENATI & C. MONTECUCCO. 1993. J. Biol. Chem. 268: in press.

NIEMAN & R. JAHN. 1993. Nature 365: 160-163. BLASI, J., E. R. CHAPMAN, E. LINK, T. BINZ, S. YAMASAKI, P. DE CAMILLI, T. C. SUDHOF, H.

HUTTNER, W. B. 1993. Nature 365: 104-105.