Functional characterisation of tetanus and botulinum neurotoxins binding domains

INTRODUCTION

The clostridial neurotoxin (CNT) family is composed oftetanus neurotoxin (TeNT) and seven different serotypes ofbotulinum neurotoxins (BoNT) (Minton, 1995). The activeholotoxin is comprised of two fragments, termed heavy (H, 100kDa) and light (L, 50 kDa) chains (Fig. 1A) and blocksneurotransmitter release in vitro and in vivo (Schiavo andMontecucco, 1997). The L chain is responsible for theintracellular activity of CNTs and contains the catalytic domainof the neurotoxins. CNTs are zinc-endoproteases whichspecifically cleave a family of proteins, named SNAREs(soluble NSF attachment protein receptor) (Söllner et al.,1993). These proteins, whose integrity is necessary to sustainregulated secretion, are involved in multiple steps leading tothe docking and fusion of small synaptic vesicles (SSVs) withthe presynaptic plasma membrane and in a variety of otherintracellular trafficking events (Hanson et al., 1997; Hay andScheller, 1997; Robinson and Martin, 1998).

Vesicle-associated membrane protein (VAMP)/synaptobrevinis localised on SSVs and other vesicular compartments and iscleaved by TeNT and four serotypes of BoNT (B, D, F and G)

(Schiavo et al., 1992, 1993a,c, 1994; Yamasaki et al., 1994a,b,c).BoNT/A and E instead cleave synaptosomal-associated proteinof 25 kDa (SNAP-25) (Blasi et al., 1993a; Schiavo et al.,1993a,b), whilst BoNT/C has a dual-specificity for SNAP-25and syntaxin 1 (Blasi et al., 1993b; Schiavo et al., 1995; Foranet al., 1996; Osen Sand et al., 1996; Williamson et al., 1996;Vaidyanathan et al., 1999). Both syntaxin and SNAP-25 aremainly localised on the plasma membrane (Garcia et al., 1995).They are therefore defined as target SNAREs or t-SNAREs,whilst VAMP/synaptobrevin is a vesicular SNARE or v-SNARE. v-and t-SNAREs form a very stable complex, possiblypromoting a strict apposition of the vesicular and target lipidbilayers which is likely to drive membrane fusion (Sutton et al.,1998; Weber et al., 1998). CNTs alter the formation andassembly of this complex by reducing its stability (Hayashiet al., 1994, 1995) and in the case of syntaxin and VAMP/synaptobrevin, by detaching the cytoplasmic portion from theirmembrane anchors (Schiavo and Montecucco, 1997).

Despite the widespread distribution of the cleavableisoforms of VAMP, SNAP-25 and syntaxin, the action of CNTsin vivo is absolutely neurospecific. All the clinical symptomsof botulism can be ascribed to the inhibition of acetylcholine

2715Journal of Cell Science 112, 2715-2724 (1999)Printed in Great Britain © The Company of Biologists Limited 1999JCS4674

Tetanus and botulinum neurotoxins constitute a family ofbacterial protein toxins responsible for two deadlysyndromes in humans (tetanus and botulism, respectively).They bind with high affinity to neurons wherein they causea complete inhibition of evoked neurotransmitter release.Here we report on the cloning, expression and use of therecombinant fragments of the heavy chains of tetanusneurotoxin and botulinum neurotoxin serotypes A, B andE as tools to study the neurospecific binding of theholotoxins. We found that the recombinant 50 kDacarboxy-terminal domains of tetanus and botulinumneurotoxins alone are responsible for the specific bindingand internalisation into spinal cord cells in culture.Moreover, we provide evidence that the recombinantfragments block the internalization of the parentalholotoxins in a dose-dependent manner, as determined by

following the neurotoxin-dependent cleavage of theirtargets VAMP/synaptobrevin and SNAP-25. In addition,the recombinant binding fragments cause a significantdelay in the paralysis induced by the correspondingholotoxin on the mouse phrenic nerve-hemidiaphragmpreparation. Taken together, these results show that thecarboxy-terminal domain of tetanus and botulinumneurotoxins is necessary and sufficient for the binding andinternalisation of these proteins in neurons and open thepossibility to use them as tools for the functionalcharacterisation of the intracellular transport of clostridialneurotoxins.

Key words: Tetanus toxin, Botulinum neurotoxin, Spinal cordneuron, SNARE protein

SUMMARY

Functional characterisation of tetanus and botulinum neurotoxins binding

domains

Giovanna Lalli1,*, Judit Herreros1,*, Shona L. Osborne1, Cesare Montecucco2, Ornella Rossetto2 andGiampietro Schiavo1,‡

1Molecular Neuropathobiology, Imperial Cancer Research Fund, 44 Lincoln’s Inn Fields, London WC2A 3PX, UK2Centro CNR Biomembrane and Dipartimento di Scienze Biomediche, Università di Padova, via Colombo 3, 35121 Padova, Italy*These two authors contributed equally to the work‡Author for correspondence (e-mail: [email protected])

Accepted 3 June; published on WWW 21 July 1999

2716

release at the neuromuscular junction (NMJ), whilst the spasticparalysis of tetanus is due to TeNT targeting to the inhibitoryinterneurons of the spinal cord (Halpern and Neale, 1995;Schiavo and Montecucco, 1997). The neurospecificity and thedifferential trafficking of BoNTs and TeNT appear to be dueto structural elements present at the level of their H chains.These contain two functional domains: a carboxy-terminalportion (HC) that is likely to play a role in the interaction withthe receptor(s), and an amino-terminal part (HN) of 50 kDainvolved in the translocation of the L chain in the cytosol (Fig.1A) (Montecucco and Schiavo, 1993). The recently reportedcrystal structure of TeNT HC domain (Umland et al., 1997;Knapp et al., 1998) and of BoNT/A (Lacy et al., 1998) showthat the HC domains of CNTs are highly related and consist oftwo sub-domains with folds related to those of lectin sugar-binding proteins and the trypsin inhibitor family, respectively.

The presence of two putative binding elements in the HCfragment supports a dual receptor model for the binding ofCNTs to neuronal membranes (Montecucco, 1986). Low (Kdin the nMolar range) and high (Kd, sub-nM) affinity receptorsfor CNTs have been described and experimental evidenceindicates that both lipid and protein receptors are essential forhigh affinity binding (Halpern and Neale, 1995). Whilst thecontribution of certain polysialogangliosides of the G1b seriesto CNTs binding is well documented (Habermann and Dreyer,1986; Halpern and Neale, 1995; Schiavo and Montecucco,1997), the characterisation of their protein receptors at theperipheral and central level has proved elusive.Synaptotagmins, integral proteins of SSVs, have beensuggested as the neuronal receptors for BoNT/B, E and A (Liand Singh, 1998; Nishiki et al., 1994, 1996a,b), thus supportinga model envisaging BoNTs entry at the NMJ via SSV recycling(Matteoli et al., 1996).

Here we report on the expression and functionalcharacterisation of the HC fragments of TeNT and of BoNT/A,B and E which are the three botulinum neurotoxins morefrequently involved in human botulism. We describe thepurification to homogeneity of HC fragments that are able tobind to the surface of murine spinal cord neurons, a modelsystem which contains physiological CNT receptors. Thesefragments antagonise the binding and cellular trafficking of thenative holotoxins, as assayed by the protection of the CNT-mediated SNARE cleavage and by the delay in the onset ofparalysis in the mouse phrenic nerve-hemidiaphragm.

MATERIALS AND METHODS

Vector preparationOligonucleotides with the sequence GATCTTACACCGATATAGAG-ATGAACAGGCTGGGAAAGCGTCGTGCATCTGTTCATATG andAATTCATATGAACAGATGCACGACGCTTTCCCAGCCTGTTCA-TCTCTATATCGGTGTAA were phosphorylated with T4polynucleotide kinase (New England Biolabs) for 30 minutes at 37°Cand then annealed at a final concentration of 1 µM in 40 mM Tris-HCl, pH 7.5, 20 mM MgCl2, 50 mM NaCl. The double strand linkerwas then ligated into the BamHI/EcoRI site of pGEX-4T3 (AmershamPharmacia Biotech). The resulting vector (pGEX-4T3-VSV-G)encodes for glutathione S-transferase (GST) which is fused at thecarboxy terminus, downstream of the thrombin cleavage site, to thesequence YTDIEMNRLGK, corresponding to the vesicular stomatitisvirus protein G epitope (VSV-G) (Gallione and Rose, 1985; Soldati

and Perriard, 1991). This epitope tag is followed by the motifrecognised by protein kinase A (Kaelin et al., 1992). The sameprocedure was followed to prepare the vector pGEX-4T3-HA inwhich the sequence YPYDVPDYA corresponding to the influenzavirus hemagglutinin epitope (HA) replaces the VSV-G epitope. In thiscase, the oligonucleotides GATCTTACCCCTACGACGTGCCCGA-CTACGCCGGATCCCGTCGTGCATCTGTTCATATGCCATGG andAATTCCATGGCATATGAACAGATGCACGACGGGATCCGGCGT-AGTCGGGCACGTCGTAGGGGTAA were inserted in the BamHI/EcoRI site of pGEX-4T3.

PCR amplification and cloningPolymerase chain reaction (PCR) was performed using as a templatecrude genomic DNA from selected toxigenic C. botulinum strains(Fach et al., 1995). The strain P64 which is related to strain 62A (Binzet al., 1990) was used for BoNT/A, NCTC 11219 (Whelan et al.,1992) for BoNT/E and the proteolytic strain B600 for BoNT/B (kindgifts of Dr M. R. Popoff, Institute Pasteur, Paris, France). Thefollowing primers were used: for BoNT/A, residues 845-1295,CCATGGAGTACAGATATACCTTTTCAGCTTTC and GTCGACT-TACAGTGGCCTTTCTCCCCATC (GenBank accession numberM30196); for BoNT/E, residues 820-1252, CCATGGAATAATAG-TATTCCTTTTAAGCTTTCTTC and GTCGACTTATTTTTCTTGC-CATCCATGTTCTTCAG (accession number X62683); for BoNT/B,residues 832-1290, CCATGGAAAACCATTATGCCGTTTGATC-TTTCAA and GTCGACTTATTCAGTCCACCCTTCATCTTTAG.The resulting BoNT/B HC fragment constitutes a new BoNT/B variantand its sequence is now available with the accession numberAJ242628. Following PCR amplification, DNA inserts were sub-cloned into pCR2.1 (Invitrogen), excised with NdeI/XhoI (BoNT/A)and NcoI/SalI (BoNT/B and E) and then ligated into pGEX-4T3-HA(for BoNT/B and E) or pGEX-4T3-VSV-G (BoNT/A). TeNT HC(residues 855-1314; accession number X04436) was obtained in thevector pTTQ8 (Amersham Pharmacia Biotech) from Dr J. Halpern(FDA, Bethesda, Maryland, USA) (Halpern et al., 1990). The insertwas excised with SalI/PstI, blunt-ended with T4 DNA polymerase(New England Biolabs) following manufacturer recommendationsand ligated in the SmaI site of pGEX-4T3-VSV-G.

Expression and purification of recombinant HC fragmentspGEX-4T3-VSV-G and pGEX-4T3-HA vectors containing theappropriate HC insert were used to transform TG1 or BL21(DE3)pLysstrains (Stratagene) and protein expression was induced by incubationfor 4 hours at 30°C in the presence of 0.4 mM isopropyl-β-D-thiogalactoside (IPTG) (Guan and Dixon, 1991). For BoNT/B, TOPP3cells (Stratagene) were used without induction by IPTG. RecombinantGST-fusion proteins were adsorbed on glutathione-agarose beads(Sigma) and sequentially incubated with 50 mM Tris-HCl, pH 7.4, 2mM ATP, 10 mM MgSO4 for 10 minutes at 37°C followed byphosphate-buffered saline containing 0.5 M NaCl and 0.05% Tween-20. The tagged HC fragments were released by thrombin cleavage(Guan and Dixon, 1991). When necessary, the eluted proteins werepurified by metal-chelating chromatography using HiTrap columns(Amersham Pharmacia Biotech) (Rossetto et al., 1992). Briefly, 1 mlHiTrap column was preloaded with 5 ml of 100 mM ZnSO4, followedby 10 ml of distilled water and then equilibrated with 50 mM Hepes-Na, pH 7.4, 150 mM NaCl (buffer A). After loading and extensivewashing with buffer A, the HC fragments were eluted with a lineargradient of imidazole in buffer A (0-25 mM). Pooled fractions weredialysed against 20 mM Hepes-NaOH, pH 7.4, 150 mM NaCl, 10%glycerol, 0.1 mM dithiothreitol (DTT) and, after freezing in liquidnitrogen, stored at −80°C.

Spinal cord cells primary cultureSpinal cords from 14-day foetal mice were isolated, dissociated andplated in 12-well plates (Corning Costar) or on glass coverslips,coated with a mixture of polyornithine and collagen (Fitzgerald,

G. Lalli and others

2717CNT binding fragments interact with neurons

1989; Williamson et al., 1996). Cultures were grown for two weeksin a humidified 10% CO2 incubator at 37°C. Culture medium wasMEM (minimum essential medium with Earle’s salts, with sodiumbicarbonate 3.7 g/l, glucose 6 g/l, pH 7.3-7.4), containing B27supplement (Gibco-BRL), heat-inactivated horse serum (5% v/v)and L-glutamine (2 mM). Five days after plating, 35 µg/ml ofuridine and 15 µg/ml of 5′-fluoro-2′-deoxyuridine (both fromSigma) were added to the medium for 96 hours to block cellproliferation. Medium was changed every 3-4 days.

CNT binding and internalisationSpinal cord cultures were cooled to 4°C, washed twice with 0.1%bovine serum albumin (BSA) in Hanks’ solution (20 mM Hepes-Na,pH 7.4, 0.44 mM KH2PO4, 0.42 mM Na2HPO4, 5.36 mM KCl, 136mM NaCl, 0.81 mM MgSO4, 1.26 mM CaCl2, 6.1 mM glucose) andincubated with recombinant TeNT HC, native TeNT, BoNT/B HC (80nM), BoNT/A HC or BoNT/E HC (200 nM) for 1 hour at 4°C. Inselected samples, native TeNT (20 µM) was pre-incubated at 4°C for15 minutes before addition of the recombinant HC. For internalisationstudies, cells were incubated for 1 hour at 37°C with the recombinantTeNT HC (80 nM) and BoNT/E HC (100 nM) domains diluted inculture medium. After washing, cells were transferred to roomtemperature and fixed in 4% paraformaldehyde, 20% sucrose inphosphate-buffered saline (PBS) without calcium and magnesium for15 minutes. After rinsing twice with PBS, cells were incubated for 20minutes with 50 mM NH4Cl in PBS, washed and then blocked with2% BSA, 0.25% porcine skin gelatin, 0.2% glycine, 15% foetal calfserum in PBS for 1 hour. Coverslips were then incubated for 30-60minutes with mouse anti-VSV-G (1:100) (Soldati and Perriard, 1991),mouse anti-HA (1:50) (Niman et al., 1983) or polyclonal anti-TeNT(1:1,500) antibodies diluted in PBS containing 1% BSA, 0.25%porcine skin gelatin. After rinsing, cells were incubated for 25 minuteswith Alexa 488 goat anti-mouse IgG or Alexa 488 goat anti-rabbitIgG (1:200) (Molecular Probes). To monitor internalisation, cell weretreated with the same solutions with the addition of 0.1% Triton X-100. Coverslips were mounted on slides with Mowiol 4-88 (Harco)and stored at 4°C.

HC binding to purified polysialogangliosides was monitoredby dot-blot assay (Thomas et al., 1999) by using 0.5 µg ofphosphatidylcholine (PC) (Avanti) or purified gangliosides GM1,GD1b, GT1b and GQ1b (Sigma) adsorbed on nitrocellulose. Afterblocking with 5% dried skimmed milk in PBS, HC fragments (80nM) were diluted in 20 mM Tris-acetate, pH 6.0, 5% milk andincubated for 2 hours at room temperature. Binding was detectedwith anti-VSV-G (1:1,000) and anti-HA (1:200) antibodies,followed by an anti-mouse peroxidase-conjugated IgG andEnhanced Chemi-Luminescence method (ECL, AmershamPharmacia Biotech, UK).

CNT treatment and SNARE analysis in spinal cord cellsSpinal cord cultures were rinsed twice with MEM and incubated for20 hours at 37°C in serum-free culture medium with different CNTs(TeNT and BoNT/A 100 pM, BoNT/E 500 pM, BoNT/B 2.5 nM)purified as previously described (Schiavo and Montecucco, 1995).Some samples were pre-incubated with the recombinant HC (1 nM,10 nM, 100 nM and 1 µM, unless otherwise stated) for 1 hour at 37°Cbefore adding the corresponding CNT. For immunocytochemistry,cells were fixed and treated as described with the addition of 0.1%Triton X-100 in all solutions. SNARE detection was carried out witha polyclonal antibody recognising the carboxy terminus of SNAP 25(1:200) (Osen Sand et al., 1993) or with a monoclonal anti-VAMP-2(1:100) (Edelmann et al., 1995). For western blot analysis, spinal cordcells were washed twice with PBS and then scraped in the same buffer.Proteins were recovered by precipitation with 6.5% trichloroaceticacid (TCA) and analysed by SDS-PAGE containing urea (Söllner etal., 1993). Western blotting was performed by using anti-VAMP-2antibody (1:200) (Edelmann et al., 1995), followed by ECL detection.

For quantitation, protein recovery was normalised using syntaxin 1immunoreactivity as internal standard.

Mouse phrenic nerve-hemidiaphragm recordingMouse phrenic nerve-hemidiaphragms were dissected from animalsweighing about 20 g, mounted in 10 ml of oxygenated (95% CO2 -5% O2) Krebs-Ringer solution containing 11 mM glucose, pH 7.4,and kept at 37°C. The phrenic nerve was stimulated via two ringplatinum electrodes by supramaximal stimuli of 3-6 V amplitude and0.1 millisecond pulse duration with a frequency of 0.1 Hz. Isometricmuscle contraction was monitored via a displacement force transducerconnected to a recorder. In control experiments (without any addedtoxin), the amplitude of muscle contraction under stimulation wasconstant for at least 8 hours. CNTs were added to the bath at aconcentration of 10 nM for TeNT and 0.2 nM for BoNTs underconditions that allow binding to go to completion (Simpson, 1980).For this purpose, tissues were incubated with toxin in physiologicalmedium at 25°C for 60 minutes without nerve stimulation. At the endof incubation, tissues were washed and transferred to 37°C in a bathwithout CNTs. Nerve stimulation was applied until a reduction of90% of the initial muscle twitch was observed. In the competitionassays, mouse phrenic nerve-hemidiaphragms were incubated witheach HC fragment (HC:CNT ratio 100:1 for TeNTand 1,300:1 forBoNTs) for 15 minutes at 25°C followed by a co-incubation with theparental CNT at the same concentration of the control for 60 minutesat 25°C without nerve stimulation. After incubation, all tissues werewashed and paralysis times were monitored. Results were expressedas the time necessary to obtain a 50% reduction of the initial muscleresponse following nerve stimulation. Data are the average of n=3experiments.

RESULTS

Expression and characterisation of TeNT and BoNTsHC fragmentsExperimental evidence suggests that the HC fragment is mainlyresponsible for the specific binding of CNTs to theiracceptor(s) in neurons. However, with the exception of TeNT,the molecular analysis of this process has been hampered bythe lack of suitable protocols for the expression of this domainin a recombinant form. Here we used variants of the expressionvector pGEX, encoding glutathione S-transferase, for thepurification of fusion proteins containing the HC fragments ofTeNT and three BoNT serotypes (A, B and E) (Fig. 1A). Thesenew vectors contain either the VSV-G (pGEX-4T3-VSV-G) orthe HA (pGEX-4T3-HA) epitopes after the thrombin cleavagesite. We adopted a PCR approach using as a template crudegenomic DNA from selected toxigenic C. botulinum strains.Whilst the amplification with BoNT/A and E specific primersgenerated fragments corresponding to the published sequences,the same procedure also allowed the isolation of a newBoNT/B variant with 91% identity to M81186 and Y13630(strain Danish, ATCC 43757) and 88% identity to X71343(strain Eklund, ATCC25765, non-proteolytic) at the amino acidlevel. The neurotoxin corresponding to this new variant is fullytoxic and is immunologically indistinguishable from theclassical B serotype.

The length of the sequences corresponding to the differentHCs was chosen on the basis of sequence alignments and thecrystal structures (Umland et al., 1997; Knapp et al., 1998;Lacy et al., 1998), which highlight the similar organisation ofCNT HC fragments. This structure predicts that both thefolding and the binding activity of the HC domains are

2718

independent of the remainder of the neurotoxin molecule. Totest this hypothesis, we expressed and purified GST-fusionproteins containing the different HC fragments in E. coli (Guanand Dixon, 1991). In the case of TeNT, BoNT/A and E, thisprocedure allowed the isolation of recombinant proteins withan apparent molecular mass ranging from 44 to 47 kDa (Fig.1B, left panel) depending on the serotype, which arespecifically recognised by either anti-VSV-G (Fig. 1B, centralpanel) or anti-HA antibodies (Fig. 1B, right panel). In contrast,the same procedure applied to BoNT/B was not successful andallowed the expression of only a very small amount of HCwithin inclusion bodies. When another E. coli strain (TOPP3,Stratagene) was used for expression, a soluble fusion proteinwas obtained which, after thrombin cleavage, gave a 48 kDahomogeneous band recognised by an anti-HA antibody (Fig.1B, left and right panels).

Binding and internalisation of recombinant HCfragmentsCNTs are known to bind to polysialogangliosides in vitro andthis interaction is mediated by the HC domain (reviewed byHabermann and Dreyer, 1986; Halpern and Neale, 1995). Wetested the competence of our recombinant fragments to bindpolysialogangliosides by using a dot-blot assay (Thomas et al.,1999). As shown in Fig. 1C, TeNT HC interacts with GT1b andto a less extent to GD1b, whereas BoNT/B binds GT1b and GQ1b.No interaction was observed with PC or the monosialilatedganglioside GM1, thus confirming the preference of CNTs fora subset of polysialogangliosides (Habermann and Dreyer,1986; Schengrund et al., 1991; Halpern and Neale, 1995).

Whilst BoNT/E HC mirrors the binding behaviour of BoNT/B,the HC of BoNT/A presents a much lower interaction withthese glycolipids (not shown).

In order to test the HC fragments in a more physiologicalcontext, their binding was assayed on the most appropriatecellular system presently available to study CNT activity, amixed population of foetal spinal cord cells containingmotoneurons, dorsal root ganglia, glial and stromal cells(Ransom et al., 1977). The neuronal components of this culturedisplay evoked neurotransmitter release which can be blockedby physiological concentrations of CNTs (Williamson et al.,1996), thus indicating the presence of functional CNT bindingsites on their surface.

As shown in Fig. 2, native TeNT and its recombinant HCfragment bind to mouse spinal cord cells in culture with similardistribution. At 4°C, the immunoreactivities of both the VSV-G tagged HC fragment (Fig. 2A) and TeNT (Fig. 2B), presentin the incubation medium at a final concentration of 80 nM,revealed a punctate surface distribution. Membrane staining isnot homogenous and is distributed all along the neurites andon the cell soma. In some cases, labelling has themorphological appearance of neuronal varicosities, areaswhere neurotransmitter release and active endocytosis aresuggested to take place (Bennett et al., 1997; Brain et al., 1997;Bennett, 1998). The staining of HC fragment is specific asdemonstrated by competition with excess of native TeNT (Fig.2C). The absence of staining of the homogenous layer of non-neuronal supporting cells, further demonstrates that TeNT andHC bind selectively to neurons.

The binding of the recombinant HC fragments of BoNT/A,

G. Lalli and others

Fig. 1. Expression of clostridial neurotoxin HC fragments andbinding to polysialogangliosides. (A) Schematicrepresentation of CNTs. The active holotoxin is composed oftwo chains (H and L) held together by a single disulfidebridge. The L chain is responsible for the intracellularproteolytic activity. The H chain, which can be furthersubdivided into two functional subdomains, is involved in theneurospecific binding (HC) and membrane translocation (HN).The HC fragments were tagged with the VSV-G (TeNT andBoNT/A) or with the HA epitope (BoNT/B and /E) at theamino terminus (star) and expressed as GST-fusion proteins.(B) SDS-PAGE profile of the recombinant HC fragmentsstained with Coomassie Blue (left panel, 1 µg/lane) or afterwestern blotting with anti-VSV-G or anti-HA antibodiesfollowed by ECL. (C) The HC fragments of BoNT/B andTeNT bind to polysialogangliosides in a dot-blot assay.Lipids (0.5 µg) were immobilised onto nitrocellulose,overlayed with TeNT and BoNT/B HC (80 nM) and detectedas described in B.

kDa

2719CNT binding fragments interact with neurons

B and E is very similar to that observed with TeNT, as shownin Fig. 2D-F. Staining is concentrated in distinct patches onthe plasma membrane of neurites and the cell soma.Qualitatively, the extent of binding with the three BoNT HCsis lower than the one observed with TeNT. This may reflecta reduced number of membrane acceptors, which in turncould explain the higher concentration of native BoNTsrequired to observe a physiological intracellular effect (seebelow). Alternatively, the reduced staining seen with BoNTHCs could be due to an overall lower binding affinity of thesefragments.

The same cellular system was then used to monitor theability of the HC fragments to be internalised following surfacebinding. After incubation at 37°C, confocal microscopicanalysis revealed that HC immunostaining was concentrated inbright intracellular structures with very few patches stillpresent on the cell surface (Fig. 3). These endocytic vesiclesand their intracellular distribution appear to be identical forBoNT and TeNT HC fragments. Future biochemicalexperiments are necessary to assess the precise nature of thesevesicles and to test the possibility that TeNT and BoNTs use,at least in part, the same internalisation pathway.

Recombinant HC fragments block the binding andintracellular activity of the native CNTs in spinalcord cell culturesIn order to establish that the HC is able to bind to thephysiological receptor of the parental CNT, and thereforerepresents the bona fide binding domain, we tested the differentHC fragments for their ability to inhibit the intracellular activityof the corresponding holotoxin in spinal cord cultures. Asshown in Fig. 4A, the neuronal components of this preparationexpress the SSV protein VAMP/synaptobrevin, which localises

in bright clusters corresponding to synaptic contacts and topoints of accumulation of SSVs. Treatment of the culture for20 hours at 37°C with 100 pM native TeNT caused thecomplete disappearance of VAMP immunostaining (Fig. 4B),following its specific proteolysis by this neurotoxin. Aspreviously reported (Osen Sand et al., 1996; Williamson et al.,1996), TeNT-dependent VAMP ablation did not alter neuronalmorphology nor affect cell survival.

Pre-treatment of the cells with the HC fragment of TeNTpotently inhibited the proteolytic action of the neurotoxin in adose dependent manner (Fig. 4C-F). Complete protection ofVAMP immunostaining was achieved by pre-treatment of theculture with 100 nM of TeNT HC. This result was confirmedby western blotting of the treated cells and analysis of theVAMP immunoreactivity (Fig. 4, inset).

The effect of the HC fragment of BoNT/A, B and E on theintoxication mediated by the native neurotoxin was similarlytested. In this case, spinal cord cells were probed with anantibody against the substrate of the botulinum neurotoxin(VAMP-2 for BoNT/B and SNAP-25 for BoNT/A and E),whose staining disappears upon toxin cleavage (Blasi et al.,1993a; Schiavo et al., 1993a,b). As shown in Fig. 5A and G,SNAP-25 localises along the neurite membrane without aclear concentration at the synapses or at varicosities (Garciaet al., 1995). Cell treatment for 20 hours at 37°C with 100pM of BoNT/A (Fig. 5B) and 500 pM BoNT/E (Fig. 5H)abolished SNAP-25 immunostaining. SNAP-25immunoreactivity was preserved when the cells were pre-incubated with 1 µM of the corresponding HC fragment (Fig.5C,I). BoNT/B was less potent than the other CNTs on thesecells and a concentration of 2.5 nM of BoNT/B was necessaryto achieve complete loss of VAMP immunostaining (Fig. 5E).Under such conditions, only partial protection was elicited by

Fig. 2. Binding of clostridialneurotoxin HC fragments tomouse spinal cord cells.Spinal cord cells wereincubated at 4°C with theHC fragment of TeNT(A,C), native TeNT (B), theHC fragments of BoNT/A(D), of BoNT/B (E) or ofBoNT/E (F). In C, pre-incubation with native TeNTwas able to compete thebinding of the TeNT HC.Binding was detected usinganti-VSV-G (A,C,D), anti-TeNT (B) or anti-HA (E,F)antibodies as described inMaterials and Methods. Bar,5 µm.

2720

pre-incubation of the spinal cord cells with BoNT/B HC (500nM, Fig. 5F). This is consistent with a lower ratio betweenHC and native neurotoxin present in the medium, althoughother explanations are also possible (i.e. the presence ofdifferent conformational isoforms in the preparation of

BoNT/B HC, only a fraction of which are competent forbinding). The immunofluorescence protection experimentswere confirmed by western blot using an anti-VAMP-2antibody and, for SNAP-25, an antibody recognising both theintact protein and the cleaved fragment (not shown).

G. Lalli and others

Fig. 3. Internalisation of HC fragments inmouse spinal cord cells. Spinal cord cellswere incubated with the HC fragment ofTeNT (A,B) or BoNT/E (C,D) for 1 hour at37°C before fixing and permeabilisation.Binding was detected using anti-VSV-G(A,B) or anti-HA (C,D) antibodies asdescribed in Materials and Methods andimages were collected with a confocal laserscanning microscope. (A and C) Two z-sections obtained at 1.5 µm from thesubstrate, whereas B and D display themerged images of the z-sections (14×0.4µm) corresponding to the entire cell. Bars,5 µm.

Fig. 4. The recombinant HC fragment of TeNT blocks the binding and intracellular activity of the native neurotoxin. Mouse spinal cord cellswere incubated for 20 hours at 37°C in the absence (A) or in the presence (B-F) of TeNT holotoxin. Before TeNT addition, samples C-F weretreated with increasing amounts of the recombinant TeNT HC (C, 1 nM; D, 10 nM; E, 100 nM; F, 1 µM). Intact VAMP was thenimmunodetected with an anti-VAMP-2 antibody. In parallel samples, cells were recovered and levels of VAMP-2 analysed by western blot(upper right panel) as described in Materials and Methods. Bar, 5 µm.

kDa

2721CNT binding fragments interact with neurons

Recombinant HC fragments antagonise CNT-dependent inhibition of neuromuscular transmissionBoNTs elicit an irreversible paralysis of the well-establishedsystem of the mouse phrenic nerve-hemidiaphragmpreparation. Its onset of paralysis (which is commonlyexpressed as 50% of paralysis time) is dose-and temperature-dependent and ranges from 23 to 37 minutes (Table 1). Due toits site of action, TeNT is less potent on peripheral synapseswith a longer (110 minutes) 50% of paralysis time (Schmitt etal., 1981; Simpson, 1984a,b). We tested the protective activityof the recombinant HC fragments on the CNT-dependentinhibition of neuromuscular transmission by adding them tothe electrophysiological bath. All the recombinant HCfragments antagonised the action of the parental CNT which isseen as a delay in the onset of paralysis (Table 1). The increasein the 50% of paralysis time ranged from 75% for BoNT/A andTeNT to more than 150% in the case of BoNT/B and E. Thisprotective effect of HC is strictly serotype-dependent. In fact,the HC fragment of BoNT/B did not interfere with the paralysisinduced by BoNT/E (not shown), as expected from serotypesnot competing for the same cellular acceptors (Evans et al.,1986; Habermann and Dreyer, 1986).

DISCUSSION

Following the identification of their intracellular targets, CNTs

have been one of the most successful and frequently used toolsin neurobiology and cell biology. CNT-dependent SNAREablation is an effective method to analyse the function of thisfamily of proteins in intact cellular systems (Schiavo andMontecucco, 1997). Despite their importance in the cellintoxication process, considerably less attention has been paidin recent years to the entry and trafficking of these toxins inneuronal cells. A vast literature exists on the preliminarycharacterisation of the in vivo CNT binding sites, but theneuronal receptor(s) of these neurotoxins have not yet beenidentified. One possible explanation is the lack of suitablemolecular biology and biochemical tools. Apart from the

Fig. 5. The recombinant HCfragments of BoNT/A, /B and /Eblock the intracellular activity ofthe corresponding native CNTs.Mouse spinal cord cells wereincubated for 20 hours at 37°C inthe absence (A,D,G) or in thepresence of native BoNT/A (B,C),BoNT/B (E,F) or BoNT/E (H,I).Prior to CNT addition, samples inC, F and I were treated with thecorresponding HC fragment (C,BoNT/A HC, 1 µM; F, BoNT/B HC,0.5 µM; I, BoNT/E HC, 1 µM).Target SNAREs were then detectedwith an anti-SNAP-25 antibodydirected against the carboxyterminus (A-C and G-I) or with ananti-VAMP-2 antibody (D-F) asdescribed in Materials andMethods. Bar, 5 µm.

Table 1. Effect of recombinant HC fragments on the 50%time of paralysis in mouse phrenic nerve-hemidiaphragm

preparations intoxicated with TeNT and BoNTsControl CNT* CNT + HC* (+%)‡

BoNT/A 37±6 65±8 (+75)BoNT/B 23±5 60±9 (+160)BoNTE 24±3 61±7 (+154)TeNT 111±11 195±20 (+75)

*The values given are the 50% paralysis time (in minutes) of tissues (n=3)exposed to CNTs alone or in the presence of their HC fragment. Each valuerepresents the mean ± s.d.

‡Increase in the 50% of paralysis time following pre-treatment with HC(percentage of the control).

2722

analysis of the TeNT HC (Halpern and Loftus, 1993), no furtherdata on the precise definition of the neurotoxin binding domainis currently available for the different BoNT serotypes. Inparticular, the purification of these domains from bacteria hasbeen hampered by low expression levels and relatively highinsolubility. Here we demonstrate the suitability of distinct E.coli expression systems for the production of functionalbinding domains of CNTs which are able to antagonise theactivity of the native neurotoxins in different experimentalsystems.

Despite the relatively low level of sequence homology, thestructures of the HC domain of TeNT and BoNT/A are verysimilar (Umland et al., 1997; Knapp et al., 1998; Lacy et al.,1998), suggesting an analogous function and folding for theuncharacterised BoNT/B and E. In fact, the HC comprises twosimilarly sized sub-domains which have virtually no or verylimited surface interactions with the other half of the H chain.The tertiary structure suggests that this part of the moleculefolds independently and therefore constitutes a good candidatefor functional expression in an heterologous system.

Using variants of the expression vector pGEX, the HCdomain of TeNT, BoNT/A and E were expressed as a solublefusion protein in E. coli. In contrast, no expression of solubleBoNT/B HC was observed in E. coli strains based on the K-12genotype. The reason for this failure is unclear. Expressionlevels are known to be lowered by the insertion of tandem rarecodons into homologous genes (Makoff et al., 1989). Oninspection, the number and type of rare codons (Sharp and Li,1986) in BoNT/B HC are similar to those present in the otherexpressed HCs. However, the HC of BoNT/B does containmultiple triplets of consecutive rare codons concentrated in thefirst third of the sequence, a fact that may explain its poorexpression. To overcome this problem, three non K-12bacterial strains were tested and a homogeneous BoNT/B HCof the expected molecular mass was obtained in the TOPP3strain.

Recombinant HC fragments bind immobilisedpolysialogangliosides and their biological activity was furthertested in various functional assays. The first consisted of thebinding of these domains to mouse embryonic spinal cordcells. This mixed culture contains the cell types responsible forthe uptake of CNTs and the target of their final physiologicalaction in vivo. Recombinant HC fragments of TeNT andBoNT/A, B and E bind efficiently to spinal cord neurons andreveal a punctate staining pattern with areas of high toxinconcentration, suggesting specialised arrangements of CNTcellular receptors. This binding is functional, as demonstratedby the ability of the spinal cord neurons to internalise thedifferent HCs in intracellular vesicular structures upon transferto 37°C. In this regard, these fragments and their mutants willplay a central role for the future dissection of the endocyticpathway of TeNT and BoNTs in isolated neurons.

Recent findings indicate the involvement of synaptotagminsin the cellular recognition of BoNT/A, B and E (Nishiki et al.,1994, 1996a,b; Li and Singh, 1998). The domain acting asBoNTs receptor is the amino-terminal domain ofsynaptotagmin, which is localised in the lumen of the SSV(Schiavo et al., 1998), but becomes accessible to theextracellular medium following the fusion of SSV with thepresynaptic membrane (Matteoli et al., 1992). CNTs mightexploit the SSV exo-endocytic cycle to gain access to the SSV

lumen and then enter the neuronal cytoplasm (Matteoli et al.,1996). Although the distribution of the neurotoxin binding sitesobserved here is compatible with areas of SSV exocytosis, ourexperimental work does not provide direct evidence in supportof this model nor for the role of synaptotagmins as BoNTsreceptors.

We also showed that the HC domains of CNTs inhibit theintracellular proteolytic activity of the native neurotoxins,which was conveniently followed by immunofluorescence andimmunoblotting. This represents a very sensitive approach totest the binding, entry and translocation of native CNTs at aconcentration as low as picoMolar in different cellular systems(Williamson et al., 1996; Schiavo and Montecucco, 1997). Allfour HC fragments effectively inhibited the proteolysis of theparental native neurotoxin in a concentration dependentmanner, indicating that the HC fragments bind to functionalacceptor(s), which mediate the “productive” entry of theneurotoxin into the cell. This result is in contrast to theconclusions of a recent study showing no competition betweenthe binding of native BoNT/A and its H chain, prepared bychemical dissociation, at the murine NMJ (Daniels-Holgateand Dolly, 1996). This preparation, which preserves theintegrity of a fully-developed neuromuscular junction, iswidely used to study the effects of various neurotoxins,including CNTs (Wohlfarth et al., 1997). In our experimentalconditions, all four HC fragments were found to be potentantagonists of the corresponding CNT, causing a significantdelay in the onset of paralysis (75-150%). The discrepancybetween the two results could be due to loss of biologicalfunction of the entire H chain during isolation or alternatively,the isolated fragments may have different conformations oraggregation states which would result in different biologicalactivities.

Although the final concentrations necessary to delay theonset of paralysis are similar for TeNT and BoNT HCs, the ratiobetween HC and the holotoxin is higher for BoNTs than forTeNT (1,300:1 and 100:1, respectively). This reflects the well-documented fact that higher doses of TeNT compared toBoNTs are needed to observe a paralytic effect at the NMJ (10nM vs 0.2 nM) (Dreyer and Schmitt, 1981; Schmitt et al., 1981;Simpson 1984a,b, 1985). Due to the large volume of theelectrophysiological bath, a ratio between HC and holotoxinhigher than 100:1 was possible only for BoNTs. Experimentsperformed with lower excess of BoNT HCs revealed a reducedlevel of protection, possibly due to the sequestration of HC bynon-specific low affinity binding sites present on the tissue.These low affinity sites would release bound HC moleculesonly slowly, resulting in an apparently lower protecting effectof BoNT HCs in this assay.

The specificity of the antagonism of the HC on the paralysiscaused by the native neurotoxin is further demonstrated by itsstrict serotype-dependency. We observed a complete lack ofcompetition with the HC fragment of BoNT/B on the action ofBoNT/E. This result is particularly relevant because it isconsistent with the well-documented lack of competitionbetween BoNT/B and BoNT/E, which do not appear to sharethe same cellular receptor (Habermann and Dreyer, 1986). Thislast result was recently challenged by a series of reportssuggesting that BoNT/A, B and E bind to the SSV proteinssynaptotagmin I and II (Li and Singh, 1998; Nishiki et al.,1994, 1996a,b). The physiological relevance of the interaction

G. Lalli and others

2723CNT binding fragments interact with neurons

between synaptotagmin and CNTs remains controversial(Bakry et al., 1997). The recombinant HC fragments willprovide a valuable tool to verify these conclusions in vitro andin vivo, allowing further investigation into the nature anddistribution of their still uncharacterised neuronal receptors. Inaddition, they will provide an experimental system for theprecise mapping of the HC-receptor protein-protein interactionand the definition of the minimal effector domain which isessential for their possible use as neurospecific protein carrier.

We thank Dr J. Halpern (FDA, Bethesda, Maryland) for providingthe TeNT HC fragment in pTTQ8 vector, Dr M. Popoff (InstitutePasteur, Paris, France) for the kind gift of genomic DNA fromBoNT/A, /B and /E producing strains of C. botulinum and M. Beckerfor the valuable help in preparing BoNT Hcs. We also thank Drs G.Stenbeck, C. Reis e Sousa and S. Tooze for discussion and criticalreading of the manuscript and the referees for excellent suggestions.Work in the laboratories of the authors is supported by HumanFrontier Science Program (J.H.), Telethon-Italia Grant 1068 andBiomed2 BMH4-97-2410 (C.M. and O.R.) and by the ImperialCancer Research Fund.

REFERENCES

Bakry, N. M., Kamata, Y. and Simpson, L. L. (1997). Expression ofbotulinum toxin binding sites in Xenopus oocytes. Infect. Immun. 65, 2225-2232.

Bennett, M. R., Gibson, W. G. and Robinson, J. (1997). Probabilisticsecretion of quanta and the synaptosecretosome hypothesis. Evoked releaseat active zones of varicosities, boutons, and end-plates. Biophys. J. 73, 1815-1829.

Bennett, M. R. (1998). Neurotransmitter release at individual sympatheticvaricosities, boutons. Advan. Pharmacol. 42, 98-101.

Binz, T., Kurazono, H., Wille, M., Frevert, J., Wernars, K. and Niemann,H. (1990). The complete sequence of botulinum neurotoxin type A andcomparison with other clostridial neurotoxins. J. Biol. Chem. 265, 9153-9158.

Blasi, J., Chapman, E. R., Link, E., Binz, T., Yamasaki, S., De Camilli,P., Südhof, T. C., Niemann, H. and Jahn, R. (1993a). Botulinumneurotoxin A selectively cleaves the synaptic protein SNAP-25. Nature365, 160-163.

Blasi, J., Chapman, E. R., Yamasaki, S., Binz, T., Niemann, H. and Jahn,R. (1993b). Botulinum neurotoxin C1 blocks neurotransmitter release bymeans of cleaving HPC-1/syntaxin. EMBO J. 12, 4821-4828.

Brain, K. L., Cottee, L. J. and Bennett, M. R. (1997). Varicosities of singlesympathetic-nerve terminals possess syntaxin zones and differentsynaptotagmin N terminus labeling following stimulation. J. Neurocytol. 26,491-500.

Daniels-Holgate, P. U. and Dolly, J. O. (1996). Productive and non-productive binding of botulinum neurotoxin A to motor nerve endings aredistinguished by its heavy chain. J. Neurosci. Res. 44, 263-271.

Dreyer, F. and Schmitt, A. (1981). Different effects of botulinum A toxin andtetanus toxin on the transmitter releasing process at the mammalianneuromuscular junction. Neurosci. Lett. 26, 307-311.

Edelmann, L., Hanson, P. I., Chapman, E. R. and Jahn, R. (1995).Synaptobrevin binding to synaptophysin: a potential mechanism forcontrolling the exocytotic fusion machine. EMBO J. 14, 224-231.

Evans, D. M., Williams, R. S., Shone, C. C., Hambleton, P., Melling, J. andDolly, J. O. (1986). Botulinum neurotoxin type B. Its purification,radioiodination and interaction with rat-brain synaptosomal membranes.Eur. J. Biochem. 154, 409-416.

Fach, P., Gibert, M., Griffais, R., Guillou, J. P. and Popoff, M. R. (1995).PCR and gene probe identification of botulinum neurotoxin A-, B-, E-, F-,and G-producing Clostridium spp. and evaluation in food samples. Appl.Env. Microbiol. 61, 389-392.

Fitzgerald, S. C. (1989). A Dissection and Tissue Culture Manual of theNervous System. New York: Alan R. Liss Inc.

Foran, P., Lawrence, G. W., Shone, C. C., Foster, K. A. and Dolly, J. O.(1996). Botulinum neurotoxin C1 cleaves both syntaxin and SNAP-25 in

intact and permeabilized chromaffin cells: correlation with its blockade ofcatecholamine release. Biochemistry 35, 2630-2636.

Gallione, C. J. and Rose, J. K. (1985). A single amino acid substitution in ahydrophobic domain causes temperature-sensitive cell-surface transport ofa mutant viral glycoprotein. J. Virol. 54, 374-382.

Garcia, E. P., McPherson, P. S., Chilcote, T. J., Takei, K. and De Camilli,P. (1995). rbSec1A and B colocalize with syntaxin 1 and SNAP-25throughout the axon, but are not in a stable complex with syntaxin. J. CellBiol. 129, 105-120.

Guan, K. L. and Dixon, J. E. (1991). Eukaryotic proteins expressed inEscherichia coli: an improved thrombin cleavage and purification procedureof fusion proteins with glutathione S-transferase. Anal. Biochem. 192, 262-267.

Habermann, E. and Dreyer, F. (1986). Clostridial neurotoxins: handling andaction at the cellular and molecular level. Curr. Top. Microbiol. Immunol.129, 93-179.

Halpern, J. L., Habig, W. H., Neale, E. A. and Stibitz, S. (1990). Cloningand expression of functional fragment C of tetanus toxin. Infect. Immun. 58,1004-1009.

Halpern, J. L. and Loftus, A. (1993). Characterization of the receptor-bindingdomain of tetanus toxin. J. Biol. Chem. 268, 11188-11192.

Halpern, J. L. and Neale, E. A. (1995). Neurospecific binding,internalization, and retrograde axonal transport. Curr. Top. Microbiol.Immunol. 195, 221-241.

Hanson, P. I., Heuser, J. E. and Jahn, R. (1997). Neurotransmitter release.Four years of SNARE complexes. Curr. Opin. Neurobiol. 7, 310-315.

Hay, J. C. and Scheller, R. H. (1997). SNAREs and NSF in targetedmembrane fusion. Curr. Opin. Cell Biol. 9, 505-512.

Hayashi, T., McMahon, H., Yamasaki, S., Binz, T., Hata, Y., Südhof, T. C.and Niemann, H. (1994). Synaptic vesicle membrane fusion complex:action of clostridial neurotoxins on assembly. EMBO J. 13, 5051-5061.

Hayashi, T., Yamasaki, S., Nauenburg, S., Binz, T. and Niemann, H.(1995). Disassembly of the reconstituted synaptic vesicle membrane fusioncomplex in vitro. EMBO J. 14, 2317-2325.

Kaelin, W. G. Jr, Krek, W., Sellers, W. R., DeCaprio, J. A., Ajchenbaum,F., Fuchs, C. S., Chittenden, T., Li, Y., Farnham, P. J., Blanar, M. A. etal. (1992). Expression cloning of a cDNA encoding a retinoblastoma-binding protein with E2F-like properties. Cell 70, 351-364.

Knapp, M., Segelke, B. and Rupp, B. (1998). The 1.61 Angstrom structureof the tetanus toxin. Ganglioside binding region: solved by MAD and MIRphase combination. Am. Cryst. Abstract Papers 25, 90.

Lacy, D. B., Tepp, W., Cohen, A. C., DasGupta, B. R. and Stevens, R. C.(1998). Crystal structure of botulinum neuro toxin type A and implicationsfor toxicity. Nature Struct. Biol. 5, 898-902.

Li, L. and Singh, B. R. (1998). Isolation of synaptotagmin as a receptor fortype A and type E botulinum neurotoxin and analysis of their comparativebinding using a new microtiter plate assay. J. Nat. Toxins 7, 215-226.

Makoff, A. J., Oxer, M. D., Romanos, M. A., Fairweather, N. F. andBallantine, S. (1989). Expression of tetanus toxin fragment C in E. coli:high level expression by removing rare codons. Nucl. Acids Res. 17, 10191-10202.

Matteoli, M., Takei, K., Perin, M. S., Südhof, T. C. and De Camilli, P.(1992). Exo-endocytotic recycling of synaptic vesicles in developingprocesses of cultured hippocampal neurons. J. Cell Biol. 117, 849-861.

Matteoli, M., Verderio, C., Rossetto, O., Iezzi, N., Coco, S., Schiavo, G.and Montecucco, M. (1996). Synaptic vesicles endocytosis mediates theentry of tetanus neurotoxin into hippocampal neuron. Proc. Nat. Acad. Sci.USA 93, 13310-13315.

Minton, N. P. (1995). Molecular genetics of clostridial neurotoxins. Curr. Top.Microbiol. Immunol. 195, 161-194.

Montecucco, C. (1986). How do tetanus and botulinum toxins bind toneuronal membranes? Trends Biochem. Sci. 11, 315-317.

Montecucco, C. and Schiavo, G. (1993). Tetanus and botulism neurotoxins:a new group of zinc proteases. Trends Biochem. Sci. 18, 324-327.

Niman, H. L., Houghten, R. A., Walker, L. E., Reisfeld, R. A., Wilson, I.A., Hogle, J. M. and Lerner, R. A. (1983). Generation of protein-reactiveantibodies by short peptides is an event of high frequency: implications forthe structural basis of immune recognition. Proc. Nat. Acad. Sci. USA 80,4949-4953.

Nishiki, T., Kamata, Y., Nemoto, Y., Omori, A., Ito, T., Takahashi, M. andKozaki, S. (1994). Identification of protein receptor for Clostridiumbotulinum type B neurotoxin in rat brain synaptosomes. J. Biol. Chem. 269,10498-10503.

Nishiki, T., Tokuyama, Y., Kamata, Y., Nemoto, Y., Yoshida, A., Sato, K.,

2724

Sekiguchi, M., Takahashi, M. and Kozaki, S. (1996a). The high-affinitybinding of Clostridium botulinum type B neurotoxin to synaptotagmin IIassociated with gangliosides GT1b/GD1a. FEBS Lett. 378, 253-257.

Nishiki, T., Tokuyama, Y., Kamata, Y., Nemoto, Y., Yoshida, A., Sekiguchi,M., Takahashi, M. and Kozaki, S. (1996b). Binding of botulinum type Bneurotoxin to Chinese hamster ovary cells transfected with ratsynaptotagmin II cDNA. Neurosci. Lett. 208, 105-108.

Osen Sand, A., Catsicas, M., Staple, J. K., Jones, K. A., Ayala, G.,Knowles, J., Grenningloh, G. and Catsicas, S. (1993). Inhibition of axonalgrowth by SNAP-25 antisense oligonucleotides in vitro and in vivo. Nature364, 445-448.

Osen Sand, A., Staple, J. K., Naldi, E., Schiavo, G., Rossetto, O.,Petitpierre, S., Malgaroli, A., Montecucco, C. and Catsicas, S. (1996).Common and distinct fusion proteins in axonal growth and transmitterrelease. J. Comp. Neurol. 367, 222-234.

Ransom, B. R., Neale, E., Henkart, M., Bullock, P. N. and Nelson, P. G.(1977). Mouse spinal cord in cell colture. I. Morphology and intrinsicneuronal electrophysiological properties. J. Neurophysiol. 40, 1132-1150.

Robinson, L. J. and Martin, T. F. J. (1998). Docking and fusion inneurosecretion. Curr. Opin. Cell Biol. 10, 483-492.

Rossetto, O., Schiavo, G., Polverino de Laureto, P., Fabbiani, S. andMontecucco, C. (1992). Surface topography of histidine residues of tetanustoxin probed by immobilized-metal-ion affinity chromatography. Biochem.J. 285, 9-12.

Schengrund, C.-L., DasGupta, B. R. and Ringler, N. J. (1991). Binding ofbotulinum and tetanus neurotoxins to ganglioside GT1b and derivativesthereof. J. Neurochem. 57, 1024-1032.

Schiavo, G., Benfenati, F., Poulain, B., Rossetto, O., Polverino de Laureto,P., DasGupta, B. R. and Montecucco, C. (1992). Tetanus and botulinumB neurotoxins block neurotransmitter release by proteolytic cleavage ofsynaptobrevin. Nature 359, 832-835.

Schiavo, G., Rossetto, O., Catsicas, S., Polverino de Laureto, P., DasGupta,B. R., Benfenati, F. and Montecucco, C. (1993a). Identification of thenerve terminal targets of botulinum neurotoxin serotypes A, D, and E. J.Biol. Chem. 268, 23784-23787.

Schiavo, G., Santucci, A., Dasgupta, B. R., Mehta, P. P., Jontes, J.,Benfenati, F., Wilson, M. C. and Montecucco, C. (1993b). Botulinumneurotoxins serotypes A and E cleave SNAP-25 at distinct COOH-terminalpeptide bonds. FEBS Lett. 335, 99-103.

Schiavo, G., Shone, C. C., Rossetto, O., Alexander, F. C. and Montecucco,C. (1993c). Botulinum neurotoxin serotype F is a zinc endopeptidasespecific for VAMP/synaptobrevin. J. Biol. Chem. 268, 11516-11519.

Schiavo, G., Malizio, C., Trimble, W. S., Polverino de Laureto, P., Milan,G., Sugiyama, H., Johnson, E. A. and Montecucco, C. (1994). BotulinumG neurotoxin cleaves VAMP/synaptobrevin at a single Ala-Ala peptidebond. J. Biol. Chem. 269, 20213-20216.

Schiavo, G. and Montecucco, C. (1995). Tetanus and botulism neurotoxins:isolation and assay. Meth. Enzymol. 248, 643-652.

Schiavo, G., Shone, C. C., Bennett, M. K., Scheller, R. H. and Montecucco,C. (1995). Botulinum neurotoxin type C cleaves a single Lys-Ala bondwithin the carboxyl-terminal region of syntaxins. J. Biol. Chem. 270, 10566-10570.

Schiavo, G. and Montecucco, M. (1997). Clostridial neurotoxins. In BacterialToxins. Tools in Cell Biology and Pharmacology (ed. K. Aktories), pp. 169-186. London: Chapman & Hall.

Schiavo, G., Osborne, S. L. and Sgouros, J. G. (1998). Synaptotagmins:more isoforms than functions. Biochem. Biophys. Res. Commun. 248, 1-8.

Schmitt, A., Dreyer, F. and John, C. (1981). At least three sequential stepsare involved in the tetanus toxin-induced block of neuromusculartransmission. Naunyn Schmiedebergs Arch. Pharmacol. 317, 326-330.

Sharp, P. M. and Li, W. H. (1986). Codon usage in regulatory genes in

Escherichia coli does not reflect selection for ‘rare’ codons. Nucl. Acids Res.14, 7737-7749.

Simpson, L. L. (1980). Kinetic studies on the interaction between botulinumtoxin type A and the cholinergic neuromuscular junction. J. Pharmacol. Exp.Ther. 212, 16-21.

Simpson, L. L. (1984a). The binding fragment from tetanus toxin antagonizesthe neuromuscular blocking actions of botulinum toxin. J. Pharmacol. Exp.Ther. 229, 182-187.

Simpson, L. L. (1984b). Fragment C of tetanus toxin antagonizes theneuromuscular blocking properties of native tetanus toxin. J. Pharmacol.Exp. Ther. 228, 600-604.

Simpson, L. L. (1985). Pharmacological experiments on the binding andinternalization of the 50,000 dalton carboxy terminus of tetanus toxin atthe cholinergic neuromuscular junction. J. Pharmacol. Exp. Ther. 234,100-105.

Soldati, T. and Perriard, J.-C. (1991). Intracompartimental sorting ofessential myosin light chains: molecular dissection and in vivo monitoringby epitope tagging. Cell 66, 277-289.

Söllner, T., Whiteheart, S. W., Brunner, M., Erdjument-Bromage, H.,Geromanos, S., Tempst, P. and Rothman, J. E. (1993). SNAP receptorsimplicated in vesicle targeting and fusion. Nature 362, 318-324.

Sutton, R. B., Fasshauer, D., Jahn, R. and Brunger, A. T. (1998). Crystal-structure of a SNARE complex involved in synaptic exocytosis at 2.4angstrom resolution. Nature 395, 347-353.

Thomas, C. L., Steel, J., Prestwich, G. D. and Schiavo, G. (1999).Generation of phosphatidylinositol-specific antibodies and theircharacterisation. Biochem. Soc. Transact. 27, 32-36.

Umland, T. C., Wingert, L. M., Swaminathan, S., Furey, W. F., Schmidt,J. J. and Sax, M. (1997). Structure of the receptor binding fragment Hc oftetanus toxin. Nature Struct. Biol. 4, 788-792.

Vaidyanathan, V. V., Yoshino, K., Jahnz, M., Dorries, C., Bade, S.,Nauenburg, S., Niemann, H. and Binz, T. (1999). Proteolysis of SNAP-25 isoforms by botulinum neurotoxin types A, C, and E: Domains and aminoacid residues controlling the formation of enzyme-substrate complexes andcleavage. J. Neurochem. 72, 327-337.

Weber, T., Zemelman, B. V., McNew, J. A., Westermann, B., Gmachl, M.J., Parlati, F., Söllner, T. H. and Rothman, J. E. (1998). SNAREpins:minimal machinery for membrane fusion. Cell 92, 759-772.

Whelan, S. M., Elmore, M. J., Bodsworth, N. J., Atkinson, T. and Minton,N. P. (1992). The complete amino acid sequence of the Clostridiumbotulinum type E neurotoxin deduced by nucleotide sequence analysis ofthe encoding gene. Eur. J. Biochem. 204, 657-667.

Williamson, L. C., Halpern, J. L., Montecucco, C., Brown, J. E. and Neale,E. A. (1996). Clostridial neurotoxins and substrate proteolysis in intactneurons: botulinum neurotoxin C acts on synaptosomal-associated proteinof 25 kDa. J. Biol. Chem. 271, 7694-7699.

Wohlfarth, K., Goschel, H., Frevert, J., Dengler, R. and Bigalke, H. (1997).Botulinum A toxins: units versus units. Naunyn Schmiedebergs Arch.Pharmacol. 355, 335-340.

Yamasaki, S., Baumeister, A., Binz, T., Blasi, J., Link, E., Cornille, F.,Roques, B., Fykse, E. M., Südhof, T. C., Jahn, R. et al. (1994a). Cleavageof members of the synaptobrevin/VAMP family by types D and F botulinalneurotoxins and tetanus toxin. J. Biol. Chem. 269, 12764-12772.

Yamasaki, S., Binz, T., Hayashi, T., Szabo, E., Yamasaki, N., Eklund, M.,Jahn, R. and Niemann, H. (1994b). Botulinum neurotoxin type Gproteolyses the Ala81-Ala82 bond of rat synaptobrevin 2. Biochem. Biophys.Res. Commun. 200, 829-835.

Yamasaki, S., Hu, Y., Binz, T., Kalkuhl, A., Kurazono, H., Tamura, T.,Jahn, R., Kandel, E. and Niemann, H. (1994c). Synaptobrevin/vesicle-associated membrane protein (VAMP) of Aplysia californica: structure andproteolysis by tetanus toxin and botulinal neurotoxins type D and F. Proc.Nat. Acad. Sci. USA 91, 4688-4692.

G. Lalli and others