Bimodal Modulation of the Botulinum Neurotoxin Protein-Conducting Channel

Bimodal modulation of the botulinum neurotoxin protein-conducting channel

Audrey Fischera,1, Yuya Nakaib,1, Lisa M. Eubanksb, Colin M. Clancyc, William H. Teppc, Sabine Pellettc, Tobin J. Dickersonb, Eric A. Johnsonc, Kim D. Jandab,2, and Mauricio Montala,2

aSection of Neurobiology, Division of Biological Sciences, University of California San Diego, La Jolla, CA 92093; bDepartments of Chemistry and Immunology, The Skaggs Institute for Chemical Biology, and Worm Institute for Research and Medicine, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037; and cFood Research Institute, University of Wisconsin, 1925 Willow Drive, Madison, WI 53706

Communicated by Sydney Brenner, The Salk Institute for Biological Studies, La Jolla, CA, December 16, 2008 (received for review November 17, 2008) Clostridium botulinum neurotoxin (BoNT) is the causative agent of internalized toxin across intracellular membranes to reach its botulism, a neuroparalytic disease. We describe here a semisyn- cytosolic targets (7, 8), and that the structure of toosendanin thetic strategy to identify inhibitors based on toosendanin, a would likely preclude it from inhibiting the BoNT/A LC metal- traditional Chinese medicine reported to protect from BoNT intox- loprotease (13) nor BoNT binding to cells (11) suggests that ication. Using a single molecule assay of BoNT serotypes A and E toosendanin could operate by hindering LC translocation out of light chain (LC) translocation through the heavy chain (HC) channel the endosome. Herein, we report that the mechanism of action in neurons, we discovered that toosendanin and its tetrahydrofu- of toosendanin stems from its ability to alter LC translocation ran analog selectively arrest the LC translocation step of intoxica- through the HC channel/chaperone; addition of toosendanin at tion with subnanomolar potency, and increase the unoccluded HC the onset of LC translocation arrests this process, whereas channel propensity to open with micromolar efficacy. The inhibi- exposure to toosendanin after completion of cargo translocation tory profile on LC translocation is accurately recapitulated in 2 modulates HC channel activity, increasing the time the channel different BoNT intoxication assays, namely the mouse protection resides in the open state. These findings disclose an unprece- and the primary rat spinal cord cell assays. Toosendanin has an dented bimodal action of a small molecule on a transmembrane unprecedented dual mode of action on the protein-conducting chaperone: cargo-dependent inhibitor of translocation and car- channel acting as a cargo-dependent inhibitor of translocation and go-free channel activator. as cargo-free channel activator. These results imply that the bimodal modulation by toosendanin depends on the dynamic inter- Results actions between channel and cargo, highlighting their tight inter- Semisynthetic Analysis and Analog Preparation. Based on first play during the progression of LC transit across endosomes. principles, it was not clear how to dissect the complex molecular architecture of toosendanin and elucidate the mechanistic na- natural product ͉ protein translocation ͉ small molecule modulator ture of its antibotulinal properties. Hence, we turned to semi- synthesis to generate a set of rationally designed toosendanin he BoNTs comprise a family of 7 immunologically distinct analogs. Examination of the molecular architecture of toosen- Tproteins synthesized by strains of anaerobic bacteria. These danin reveals 3 key functionalities that could possibly contribute toxins (BoNT/A-G) are the most lethal poisons known, with to the observed biological activity: (i) the hemiacetal bridge BoNT serotype A (BoNT/A) having a LD50 for a 70-kg human spanning C10 and C4 of the core ABCD fused ring system; (ii) of a mere 0.7–0.9 ␮g by inhalation (1). All BoNT serotypes are the heterocyclic furan attached to C17; and (iii) the epoxide at disulfide-linked di-chain proteins consisting of a 50-kDa light C14 and C15 (Fig. 1). Each of these moieties is rare in charac- chain (LC) Zn2ϩ-metalloprotease and a 100-kDa heavy chain terized steroidal structures, with the latter two being common (HC). The HC encompasses the translocation domain (TD) and only to the limonoid family of triterpenoids. Trapping of the the receptor-binding domain (RBD), conventionally denoted as hemiacetal was accomplished by treatment of the parent com- HN and HC (2, 3). The conspicuously specific activity of BoNT pound with N-methylmorpholine N-oxide (NMO) and catalytic to selectively disable synaptic vesicle exocytosis has transformed amount of tetra-n-propylammonium perruthenate (TPAP) (14) this protein into the first bacterial toxin approved by the FDA for to oxidize all alcohols, thus disallowing the open hemiacetal and treatment of a number of diseases characterized by abnormal generating lactone analog 2 (Fig. 1). Heterocyclic furan func- muscle contraction, a blockbuster cosmeceutical, and a highly tionalities are generally considered toxic to mammalian species feared bioweapon (1, 4, 5). Functionally, these clostridial toxins (15). To specifically remove the furan, toosendanin was treated inhibit the release of acetylcholine at neuromuscular junctions with palladium on alumina and hydrogenated using standard through a multistep mechanism that ultimately culminates in the conditions to give tetrahydrofuran analog 3 (THF-toosendanin) cleavage of Soluble N-ethylmaleimide-sensitive fusion protein (Fig. 1). Epoxide moieties are uncommon in natural products in attachment protein receptor proteins, resulting in progressive light of the ensuing high reactivity, and in the case of toosen- flaccid paralysis (6–8). danin, this reactivity should be significantly enhanced due the The major limonoid constituent of the bark of the tree M. toosendan, the triterpenoid toosendanin 1 (Fig. 1), has been reported to possess activities ranging from ascarifuge to anti- Author contributions: A.F., Y.N., L.M.E., W.H.T., S.P., E.A.J., K.D.J., and M.M. designed research; A.F., Y.N., L.M.E., C.M.C., W.H.T., S.P., K.D.J., and M.M. performed research; A.F., botulinum (9–12). Specifically, studies conducted over twenty K.D.J., and M.M. contributed new reagents/analytic tools; A.F., Y.N., L.M.E., W.H.T., S.P., years ago purported that toosendanin could protect monkeys T.J.D., E.A.J., K.D.J., and M.M. analyzed data; and A.F., L.M.E., S.P., T.J.D., E.A.J., K.D.J., and from BoNT/A, BoNT/B, and BoNT/E-induced death in a dose- M.M. wrote the paper. dependent fashion when coadministered with, or several hours The authors declare no conflict of interest. after, neurotoxin administration (9–12). These reports, although 1A.F. and Y.N. contributed equally to this work. mechanistically unclear, were still intriguing in that toosendanin 2To whom correspondence may be addressed. E-mail: [email protected] or mmontal@ucsd. might provide the basis of an unmet challenge, a small molecule This article contains supporting information online at www.pnas.org/cgi/content/full/ therapeutic for the treatment of botulinum intoxication. Given 0812839106/DCSupplemental. that a primary event for intoxication is the translocation of © 2009 by The National Academy of Sciences of the USA

1330–1335 ͉ PNAS ͉ February 3, 2009 ͉ vol. 106 ͉ no. 5 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0812839106 Downloaded by guest on September 30, 2021 Fig. 1. Semisynthetic preparation of toosendanin analogs. Synthetic mod- iﬁcations were chosen to selectively remove the desired functional group without alteration of the remainder of the molecule.

unfavorable ring strain imparted by the [6,5] CD fused ring system. However, this ring strain can also be exploited to drive a [1,2]-rearrangement and the subsequent formation of a ther- modynamically more stable ketone 4 (Fig. 1). In addition to probing these rare structural features, toosendanin also pos- Fig. 2. In vivo and in vitro tests of toosendanin activity. (A) Toosendanin and sesses other functional groups anticipated to have low in vivo THF-toosendanin extend the time to death of mice challenged with lethal doses of BoNT/A. In all groups, animals (n ϭ 10) were administered the desired stability. In particular, the acetate functionalities at C3 and C12 toosendanin analog (2.5 mM, 0.1 mL, i.v.) immediately followed by BoNT would be labile and could be rapidly hydrolyzed by serum Ͻ challenge (5LD50, i.p.). *, P 0.001 compared with toxin-only control. (B) esterases, including butyrylcholinesterase and liver carboxyles- Western blot analysis of SNAP-25 cleavage by BoNT/A in primary rat spinal cord terases (16). To mimic this reaction, the two acetates were cells treated with the indicated concentrations of toosendanin and 250 pg (5.6 excised without impinging upon the other functionalities under pM) of BoNT/A. (C) Quantitative depiction of inhibition of BoNT/A activity by BIOPHYSICS hydrolysis conditions optimized to selectively hydrolyze the toosendanin in the primary rat spinal cord cells assay. Bands corresponding to Ͻ desired esters without altering other base-labile sites to yield uncleaved and cleaved SNAP-25 were quantiﬁed by densitometry. *, P 0.05 deacylated toosendanin analog 5 (Fig. 1). compared with toxin-only control.

In Vivo Testing of Toosendanin Analogs. The previously reported toosendanin (TSDN) results in gradual preservation of intact, antibotulinal activity of toosendanin was confirmed in a mouse uncleaved SNAP-25 (synaptosomal-associated protein with lethality bioassay for BoNT/A intoxication. This model has been Mr ϭ 25 kDa), the intracellular BoNT/A and BoNT/E substrate, CHEMISTRY well-established as the FDA standard for assessing BoNT po- becoming practically complete above 200 nM (Fig. 2 B and C); tency and is widely used in the study of BoNT antagonists (17). the Effective Dose to achieve 50% of the response (ED ) for Ϸ 50 Toosendanin extended the time to death (TTD) 4-fold from a BoNT/A was 55 Ϯ 2.7 nM. Similar studies conducted with single injection at the highest toosendanin dose (7.1 h in control BoNT/E demonstrated nearly complete SNAP-25 protection at Ͼ animals versus 28 h TTD, using 2.5 mM toosendanin– 1 ␮M(Fig. S2) with the ED50 value for BoNT/E ϭ 289 Ϯ 14.3 equivalent to 7.5 mg/kg) (Fig. 2A and Table S1). Additionally, no nM. To examine whether preservation of SNAP-25 by toosen- toxicity of the compound alone was observed at this dose. danin involves a direct interaction with BoNT/A or BoNT/E Toosendanin analogs (2–5) were tested in the mouse bioassay to before neuronal entry, 3 different protocols were performed in ascertain the specific functional groups of the parent compound parallel. First, BoNT/A or BoNT/E and toosendanin were pre- that are critical for prevention of BoNT-induced death. Of all incubated for 1 h before cell exposure; second, cells were tested synthetic compounds, only 3 had equivalent activity to exposed to BoNT/A or BoNT/E for 2 h before addition of toosendanin and could protect mice from death (Fig. 2A). toosendanin; and third, BoNT/A or BoNT/E and toosendanin were combined and immediately added to the cells. No signif- In Vitro Testing of Toosendanin. Confirmation of the in vivo activity icant difference in inhibition of BoNT/A- or BoNT/E-induced of toosendanin and respective new analogs allowed investiga- SNAP-25 cleavage between the different protocols was ob- tions into the mechanistic nature of the antibotulinal action. served. The inhibition of BoNT/A- and BoNT/E-catalyzed cleav- First, the effects of toosendanin on the recombinant BoNT/A age of SNAP-25 in spinal cord neurons by toosendanin is light chain was undertaken. LC/A catalytic activity was measured consistent with a site of action at the passage of the LC from the using a fluorescence resonance energy transfer assay (18); no interior of the endosome into the cytosol thereby preventing effect was observed on the LC/A protease activity even at mM contact between the protease and its SNARE substrate. The concentrations (Fig. S1). higher potency of toosendanin against BoNT/A versus BoNT/E Accordingly, we investigated the effect of toosendanin, using may arise from the construed slower translocation of the a sensitive and specific spinal cord cell-based assay validated for BoNT/A LC out from the endosome compared with BoNT/E the activity of both BoNT serotypes A and E (19). Exposure of (20, 21). The toosendanin-evoked preservation of neuronal neurons to BoNT/A in presence of increasing concentrations of SNAP-25, an entity that is essential for synaptic vesicle fusion

Fischer et al. PNAS ͉ February 3, 2009 ͉ vol. 106 ͉ no. 5 ͉ 1331 Downloaded by guest on September 30, 2021 Fig. 4. Analysis of toosendanin modulation of BoNT/A channel activity at the onset of LC translocation and after completion of LC translocation. (A) Steady state ␥ as a function of toosendanin concentration for data in S3A;ED50 ϭ 3.8 Ϯ 2.0 nM calculated by sigmoidal ﬁt. (n ϭ 18) (average N per data point ϭ 46,648 events) (B) V1/2 as a function of toosendanin concentration for data in S3B;ED50 ϭ 8.9 Ϯ 1.8 ␮M. (n ϭ 19) (average N per data point ϭ 12,805 events)

conductance intermediates, observed as occluded states, corre- spond to permissible chaperone-cargo conformations populated during protease translocation; concomitantly, the protein- conducting channel progressively conducts more Naϩ around the polypeptide chain before entering an exclusively ion-conductive state (22–24). Transformation of an occluded state characterized by low ␥ intermediates into an unoccluded channel with ␥ ϳ 65 pS only occurs after the LC completes translocation.

Fig. 3. Toosendanin modulates BoNT/A holotoxin channels. (A) BoNT/A Toosendanin Is a Cargo-Dependent Inhibitor of the BoNT Protein- holotoxin channels in excised patches of Neuro 2A cells; no ion channel activity Conducting Channel. Toosendanin arrests LC translocation, as is detected at the onset of the record, as shown in first segment. The vertical lines indicate gaps to accommodate the full recording in the limited space. shown in Fig. 3B. BoNT/A holotoxin channels were allowed to BoNT/A channel activity begins 20 min after G⍀ seal formation transitioning form in the absence of toosendanin; after onset of low conduc- from low conductance intermediate state (second segment) to the unoc- tance channel activity, toosendanin was added to the trans cluded state after completion of LC translocation (third segment). Channel compartment. Although 0.4 nM toosendanin has no effect on LC opening is indicated by a downward deflection; C and O respectively denote translocation, 4 nM toosendanin persistently arrests channel the closed and open states. (B) Exposure to increasing concentrations of activity at an intermediate step of LC translocation (23, 24). toosendanin after BoNT/A channel insertion and appearance of low conduc- Exposure to higher toosendanin concentrations at this early step tance intermediates (Left) progressively inhibits LC translocation as monitored in translocation progressively inhibits it more effectively and, at by the persistence of the occluded state (Right). (C) Addition of toosendanin ␮ after completion of BoNT/A LC translocation modifies unoccluded BoNT chan- 40 M toosendanin, irreversibly blocks translocation (Fig. 3B nel activity. bottommost image) (23, 24). In sharp contrast, addition of toosendanin after LC translocation has completed unexpectedly results in altered channel and neurotransmitter release, agrees with the in vivo data on the kinetics rather than channel blockade. Although ␥ of the unoc- observed protection of mice from BoNT/A-induced death (Fig. cluded HC channel (␥ ϳ 66 pS) remains constant, the probability 2A and Table S1). of the channel residing in the open state (Po) increases (22). As low as 0.4 ␮M toosendanin augments the Po, promoting a Single-Molecule Assay of Translocation Inhibition. A key step in prolonged residence in the open state, in a dose-dependent intoxication is the translocation of BoNT LC by the BoNT HC manner (Fig. 3C). channel (22–25). We developed an assay to investigate the Analysis of the channel kinetics for BoNT/A exposed to dynamics of translocation focusing on the interactions between toosendanin before and after completion of LC translocation the HC channel/chaperone and its LC cargo for both BoNT/A resolves a unique interaction pattern between toosendanin and and BoNT/E serotypes (23, 24). Using this assay, the transloca- the HC channel. The time course of productive LC translocation tion process is monitored in real time and at the single-molecule exhibits an average half-time for completion (t), estimated from level in excised membrane patches from Neuro 2A cells (23, 24). the transition to high conductance, of 150 Ϯ 25s(Fig. S3). At Translocation requires pH 5.3 on the cis compartment, defined 0.4 nM, toosendanin delays the completion of LC translocation as the compartment containing BoNT, and pH 7.0 on the trans (t ϭ 220 Ϯ 10 s, Fig. S3). Addition of 4 nM toosendanin allows compartment, which is supplemented with the membrane non- progression with a t of ϳ350 s to an intermediate occluded state permeable reductant TCEP, conditions that emulate those characterized by an average ␥ ϳ 35 pS (Fig. 4A and Fig. S3). prevalent across endosomes (23, 24). Translocation is then Above 4 nM, toosendanin aborts translocation blocking the observed as a time-dependent increase in Naϩ conductance (␥) BoNT/A channel in a low conductance, occluded state (22–25). through the HC channel (23, 24), as illustrated for BoNT/A by Toosendanin, therefore, arrests LC/A translocation by the the control experiment shown in Fig. 3A. The time course of ␥ BoNT/A protein-conducting channel with an ED50 value of 4.0 Ϯ change after insertion of BoNT/A holotoxin into the membrane 1.8 nM (Fig. 4A). displays multiple transient intermediate conductances before achieving a ␥ of 67.1 Ϯ 2.0 pS (Fig. 3A Right) (24). This Toosendanin Acts as Activator of the Cargo-Free Protein-Conducting steady-state ␥ is also a hallmark of isolated HC recorded under Channel. Toosendanin increases the unoccluded HC/A channel identical conditions (22); therefore it represents the conduc- Po and shifts the voltage dependence of the channel residing in tance of the cargo-free, protein-conducting channel generated the open state toward more positive membrane potentials: the 1/2 after LC translocation is complete (22–24). We infer that these voltage at which Po ϭ 0.5, V , is approximately Ϫ67 mV in the

1332 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0812839106 Fischer et al. Downloaded by guest on September 30, 2021 Fig. 5. Toosendanin modulates BoNT/E holotoxin channels. (A) After BoNT/E Fig. 6. Toosendanin tetrahydrofuran analog (THF-toosendanin) modulates channel insertion and appearance of low conductance intermediates (Left), BoNT/A holotoxin channels. (A) After BoNT/A channel insertion and appear- addition of 4 nM toosendanin arrests the channel in intermediate states and, ance of low conductance intermediates (Left) addition of 0.4 nM THF- at 400 nM, toosendanin inhibits LC translocation (Right). (B) Steady-state ␥ as toosendanin interrupts LC translocation at intermediate states, and at 40 nM a function of toosendanin concentration for data in A (n ϭ 3 for each THF-toosendanin abrogates it (Right). (B) Steady state ␥ as a function of experimental condition, average N per data point ϭ 7,424 events). (C) Expo- THF-toosendanin concentration for data in A (n ϭ 3 for each experimental sure to toosendanin after completion of LC translocation modulates BoNT/E condition, average N per data point ϭ 21,681 events). (C) Exposure to THF- 1/2 channel activity. (D) V (ﬁlled circles) and Po (open circles) as a function of toosendanin after completion of LC translocation modulates BoNT/A channel 1/2 toosendanin concentration for data in C (n ϭ 3 for each experimental condi- activity. (D) V (ﬁlled circles)and Po (open circles) as a function of THF- tion, average N per data point ϭ 3,092 events). toosendanin concentration for data in C (n ϭ 3 for each experimental condition, average N per data point ϭ 13,304 events).

absence of toosendanin and shifts to approximately ϩ63.7 Ϯ 5.4 mV in its presence (Fig. 4B and Fig. S3). Channel activity is values are in fair agreement with the results obtained for drastically modified from being evoked exclusively at negative BoNT/A and BoNT/E on the single molecule translocation assay potentials in absence of toosendanin to being elicited at pro- (Figs. 3–5). gressively more positive potentials with increasing toosendanin BIOPHYSICS concentration. At 40 ␮M toosendanin, transitions to the open Toosendanin Analogs Active in Vivo Recapitulate the Mechanism of state persist at and above ϩ100 mV (Fig. 4B and Fig. S3). the Parent Compound. For BoNT/A, the mouse toxicity bioassay Modulation by toosendanin of unoccluded BoNT/A channel predicted THF-toosendanin (analog 3) to be as active as activity, quantified from measurements of the V1/2, show an toosendanin. As detected by the single molecule assay, 4 nM ED50 ϭ 8.9 Ϯ 1.8 ␮M, a concentration Ϸ2,000-fold above that THF-toosendanin blocks translocation of BoNT/A LC as necessary to arrest LC translocation (Fig. 4B). Thus, toosenda- effectively as toosendanin. However, even at 0.4 nM, a con- nin effectively acts as an activator of the cargo-free protein- centration at which toosendanin is inactive, the THF-

conducting HC/A channel generated after productive cargo toosendanin delays cargo translocation by an average of 17 min CHEMISTRY translocation. (Fig. 6A). The protein-conducting channel remains occluded at an intermediate translocation step characterized with a ␥ ϳ Toosendanin Modulates Both Cargo-Dependent and Cargo-Free Ac- 30 pS (Fig. 6B). In contrast, 4 nM toosendanin is required to tivities of the BoNT/E Protein-Conducting Channel. Toosendanin achieve this effect (Fig. 4A). At 40 nM, THF-toosendanin arrests translocation of BoNT/E LC at 400 nM (Fig. 5A)as aborts translocation at an occluded state with a ␥ ϳ 10 pS (Fig. effectively as BoNT/A LC at 40 nM (Fig. 3B). At 4 nM 6B). Thus, THF-toosendanin is an Ϸ10-fold more potent toosendanin, the BoNT/E protein-conducting channel remains inhibitor of LC/A translocation than toosendanin. Similarly, 40 occluded at an intermediate translocation step characterized nM THF-toosendanin has an effect equivalent to that of 40 with a ␥ ϳ 50 pS (Fig. 5 A and B). At 400 nM, toosendanin aborts ␮M toosendanin on the unoccluded HC/A channel (Fig. 6C). translocation at an occluded state characterized by an average It appears, therefore, to be Ϸ1,000-fold more potent activator ␥ ϳ 24 pS (Fig. 5B). of the cargo-free HC/A channel than toosendanin. THF- Similar to its activity on the unoccluded HC/A channel (Fig. toosendanin modifies the channel gating kinetics in a concen- 1/2 4B), toosendanin increases the HC/E channel Po and shifts the tration-dependent manner shifting the V to more positive voltage dependence of the channel residing in the open state voltages and increasing Po (Fig. 6D). These features are similar toward more positive potentials (Fig. 5C). At 40 ␮M, toosen- to those produced by toosendanin on the HC/A channel (Fig. 4 C danin shifts the BoNT/E HC channel V1/2 from approximately and D). The deacetylated toosendanin, as a positive control, Ϫ45 mV to approximately ϩ65 mV, whereas at 4 nM toosen- displayed neither cargo-dependent nor cargo-free activity on the danin shifts the V1/2 to Ϸ10 mV (Fig. 5D). Thus, the BoNT/E HC BoNT/A protein-conducting channel (data not shown). channel activity evoked at negative potentials is sensitive to toosendanin at nM concentrations, a pattern comparable to that Discussion recorded for the BoNT/A HC channel (Figs. 4B an Fig. S3). An urgent goal is to identify selective and protective agents Of note is the fact that in the highly sensitive spinal cord cell directed to prevent or relieve the neuroparalytic toxic actions of assay, the minimum dose of toosendanin needed to detect BoNTs. Our results suggest that toosendanin is a multiserotype significant preservation of intact SNAP-25 was 8 nM for antibotulinal agent. The basis of the protection from BoNT BoNT/A (Fig. 2 B and C) and 40 nM for BoNT/E (Fig. S2). These intoxication engendered by toosendanin is not apparent from

Fischer et al. PNAS ͉ February 3, 2009 ͉ vol. 106 ͉ no. 5 ͉ 1333 Downloaded by guest on September 30, 2021 simple analysis of its complex molecular architecture (Fig. 1). the 7 LC serotypes have been solved, some with small molecules Our approach centered upon the systematic elimination of coordinated in the active site (2, 3, 26–31), making small functional groups that are unique to toosendanin and the molecule interaction modeling a powerful tool (6). Unfortu- limonoid class of steroids, and those moieties that are metabol- nately, these compounds display broad cross-reactivity as the ically labile and easily removed in vivo. Of this set of molecules, active site is highly conserved across cellular proteases and not only the reduction of the furan heterocycle to the corresponding involved in specific substrate recognition. For BoNT/A, the tetrahydrofuran analog led to a retention of biological activity as extended enzyme substrate interface required for optimal sub- measured by extension of the TTD in a murine bioassay (Fig. strate recognition involves an array of substrate binding sites 2A), despite the presence of a diastereomeric pair of com- distant from the active site (29, 32). These exosites may be a more pounds. This finding is particularly intriguing given that other effective target to attenuate the protease;, however, specific functional groups including the C3 and C12 acetates and the serotypes have different substrates, requiring design of 7 unique C14-C15 epoxide would be anticipated to be very labile in vivo, inhibitors. yet their presence is indispensable for the emergence of BoNT Studies focused on early stages of intoxication aim to protection. develop blockers of the interaction between the BoNT RBD Toosendanin arrests translocation by stabilizing nonpermis- (HC) with known receptors (33). This strategy, still in devel- sive conformations of the HC channel-LC complex during the opment, is limited to those serotypes for which the protein early steps of LC translocation (Figs. 3 A and B and 4A for coreceptor has been identified: SV2 for BoNT/A (34, 35) and BoNT/A and Fig. 5 for BoNT/E). In contrast, binding of BoNT/E (36) and synaptotagmin II for BoNT/B (37, 38) and toosendanin to the unoccluded HC channel modulates the gating G (39). The RBD (HC) and LC serotype specificity is attributed kinetics leading to a preponderantly open channel entity (Figs. to interactions with host cellular proteins; in contrast, the TD (HN) 3C,4B, and 5C and D), implying that it would facilitate the functions to chaperone the LC cargo. The TD may function as a catastrophic dissipation of the electrochemical gradients across universal protein translocator; small molecules that inhibit BoNT/A endosomes. This dual modality, the transformation of toosen- channel activity (Figs. 3, 4, and 6) exhibit similar effects on BoNT/E danin from an inhibitor of cargo translocation to an activator of (Fig. 5) thereby eliminating the necessity to develop multiple the protein-conducting channel, is determined by the presence, and entities. by inference, by the conformation of cargo within the chaperone. In light of these previous studies, our work highlights the Toosendanin may inhibit cargo translocation by stabilizing cargo- BoNT channel as a viable therapeutic target for intervention chaperone conformers that, given the transient nature of the transit using small molecules. We envision toosendanin and/or THF- process, are short-lived. In contrast, the cargo-free channel must toosendanin to be developed into natural product-derived lead necessarily expose a different binding site to toosendanin, which compounds with both a prophylactic and therapeutic potential as ultimately stabilizes an open conformation. selective antibotulinal agents that could be deployed to military The single molecule translocation assay dictates that the signal personnel and civilian populations at high risk. arising from the inserted channel occluded by the cargo must be detected before exposure to toosendanin. This protocol insures Materials and Methods that channel formation and insertion has occurred and that the Synthetic procedures for toosendanin analogs and protease activity of BoNT/A intermediate conductances, reporters of the translocation pro- light chain are presented in SI Text. cess, are monitored precisely at the onset of translocation (22–24). The fact that toosendanin acts on the unoccluded HC Materials. Unless otherwise specified, all chemicals were purchased from channel of both BoNT/A and BoNT/E and modulates the Sigma–Aldrich. Purified native BoNT/A and BoNT/E holotoxins were prepared residence time of the channel in the open state indicates that the from C. botulinum strains Hall A hyper (40) and Beluga E (41) as previously interaction of toosendanin is directly on the membrane- described and were purchased from Metabiologics or prepared in E.A.J.’s embedded HC channel. Taken together, we favor the interpre- laboratory. Toosendanin was purchased from Apin Chemicals. tation that toosendanin acts on the regions of the HC channel confined within the lipid bilayer both when occluded by the LC In Vivo Assay. Female CD-1 outbred mice (17–23 g; Harlan Sprague–Dawley) cargo or unoccluded after completion of cargo translocation, this were injected intravenously into the left-hand lateral tail vein with test compounds and, immediately after with BoNT essentially as described (17, 18). notion being valid for both BoNT/A and BoNT/E serotypes. Animals were observed for signs of botulism overnight, and the time of death The BoNT channel is closed under conditions akin to those in minutes was recorded. prevalent across endosomes (22–24). After LC translocation, the closed HC channel conformation precludes the passive dissipa- Primary Rat Spinal Cord Cells Assays. Primary rat spinal cord cells were tion of the electrochemical gradients across endosomes. Micro- prepared and maintained as described in ref. 19. Toosendanin and 250 pg of molar concentrations of toosendanin shift the voltage depen- BoNT/A or BoNT/E (Ϸ20 mouse LD50 Units, and 5.6 pM), toosendanin alone, or dence of the unoccluded HC channel (Figs. 4B and 5D), effecting BoNTs alone were mixed and immediately added to cells, followed by incu- a higher propensity to reside in the open state. Accordingly, bation at 37 °C for 22 h. The cells were lysed and samples were analyzed by dissipation of electrochemical gradients would ensue and disrupt Western blot and densitometry (19). See SI Text for details. endosomal function, potentially harming critical host cellular processes. Interestingly, the modulation of HC channel activity Cell Culture and Patch Clamp Recordings. Excised patches from neuroblastoma by toosendanin occurs at a concentration Ϸ2,000-fold higher Neuro 2A cells in the inside-out configuration were used as described (22–24). Current recordings were obtained under voltage clamp conditions. Records than that at which LC translocation is efficiently arrested. This were acquired at a sampling frequency of 20 kHz and filtered online to 2 kHz bimodal modulation of BoNT channel activity is sufficiently with Gaussian filter. All experiments were conducted at 22 Ϯ 2 °C. Analysis was disparate in concentration as to promote toosendanin use as a performed on single bursts of each experimental record. Only single bursts prophylactic agent. were analyzed because of the random duration of quiescent periods. See SI The HC channel represents a realistic target for inhibition of Text for details. BoNT neurotoxicity. Blocking the BoNT protein-conducting channel with small molecules would abort translocation of the ACKNOWLEDGMENTS. We thank J. Santos, M. Oblatt-Montal, L. Koriazova for protease, thereby abrogating its toxicity. The goal is to identify perceptive comments and G. Boldt and S. Smock for preliminary contributions to this project. This work was supported by National Institutes of Health Grant a single class of compounds that would be effective against all AI 072358 (to K.D.J.), Pacific Southwest Regional Center of Excellence Grant AI BoNT serotypes irrespective of receptor binding, specific pro- 065359 (to M.M. and E.A.J.), and the Skaggs Institute for Chemical Biology (to tease substrate or antigenic properties. The crystal structures of K.D.J.).

1334 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0812839106 Fischer et al. Downloaded by guest on September 30, 2021 1. Arnon SS, et al. (2001) Botulinum toxin as a biological weapon: Medical and public 23. Fischer A, Montal M (2007) Crucial role of the disulfide bridge between botulinum health management. J Am Med Assoc 285:1059–1070. neurotoxin light and heavy chains in protease translocation across membranes. J Biol 2. Lacy DB, Tepp W, Cohen AC, DasGupta BR, Stevens RC (1998) Crystal structure of Chem 282:29604–29611. botulinum neurotoxin type A and implications for toxicity. Nat Struct Biol 5:898–902. 24. Fischer A, Montal M (2007) Single molecule detection of intermediates during botu- 3. Swaminathan S, Eswaramoorthy S (2000) Structural analysis of the catalytic and linum neurotoxin translocation across membranes. Proc Natl Acad Sci USA 104:10447– binding sites of Clostridium botulinum neurotoxin B. Nat Struct Biol 7:693–699. 10452. 4. Davletov B, Bajohrs M, Binz T (2005) Beyond BOTOX: Advantages and limitations of 25. Koriazova LK, Montal M (2003) Translocation of botulinum neurotoxin light chain individual botulinum neurotoxins. Trends Neurosci 28:446–452. protease through the heavy chain channel. Nat Struct Biol 10:13–18. 5. Burnett JC, Henchal EA, Schmaljohn AL, Bavari S (2005) The evolving field of biode- 26. Agarwal R, Binz T, Swaminathan S (2005) Structural analysis of botulinum neurotoxin fence: Therapeutic developments and diagnostics. Nat Rev Drug Discov 4:281–297. serotype F light chain: Implications on substrate binding and inhibitor design. Bio- 6. Willis B, Eubanks LM, Dickerson TJ, Janda KD (2008) The strange case of the botulinum chemistry 44:11758–11765. neurotoxin: Using chemistry and biology to modulate the most deadly poison. Angew 27. Agarwal R, Eswaramoorthy S, Kumaran D, Binz T, Swaminathan S (2004) Structural Chem Int Ed Engl 47:8360–8379. analysis of botulinum neurotoxin type E catalytic domain and its mutant Glu2123Gln 7. Schiavo G, Matteoli M, Montecucco C (2000) Neurotoxins affecting neuroexocytosis. reveals the pivotal role of the Glu212 carboxylate in the catalytic pathway. Biochem- Physiol Rev 80:717–766. istry 43:6637–6644. 8. Simpson LL (2004) Identification of the major steps in botulinum toxin action. Annu Rev 28. Arndt JW, Yu W, Bi F, Stevens RC (2005) Crystal structure of botulinum neurotoxin type Pharmacol Toxicol 44:167–193. 9. Jing Z, et al. (1985) The effect of toosendanin on monkey botulism. J Tradit Chin Med G light chain: Serotype divergence in substrate recognition. Biochemistry 44:9574– 5:29–30. 9580. 10. Li PZ, et al. (1982) Therapeutic effect of toosendanin on animal botulism. Chin Tradit 29. Breidenbach MA, Brunger AT (2004) Substrate recognition strategy for botulinum Herb Drugs 13:28–30. neurotoxin serotype A. Nature 432:925–929. 11. Shi YL, Li MF (2007) Biological effects of toosendanin, a triterpenoid extracted from 30. Segelke B, Knapp M, Kadkhodayan S, Balhorn R, Rupp B (2004) Crystal structure of Chinese traditional medicine. Prog Neurobiol 82:1–10. Clostridium botulinum neurotoxin protease in a product-bound state: Evidence for 12. Shi YL, Wang ZF (2004) Cure of experimental botulism and antibotulismic effect of noncanonical zinc protease activity. Proc Natl Acad Sci USA 101:6888–6893. toosendanin. Acta Pharmacol Sin 25:839–848. 31. Silvaggi NR, et al. (2007) Structures of Clostridium botulinum neurotoxin serotype A 13. Zhou JY, Wang ZF, Ren XM, Tang MZ, Shi YL (2003) Antagonism of botulinum toxin light chain complexed with small-molecule inhibitors highlight active-site flexibility. type A-induced cleavage of SNAP-25 in rat cerebral synaptosome by toosendanin. FEBS Chem Biol 14:533–542. Lett 555:375–379. 32. Chen S, Kim JJ, Barbieri JT (2007) Mechanism of substrate recognition by botulinum 14. Ley SV, Norman J, Griffith WP, Marsden SP (1994) Tetrapropylammonium perruthe- neurotoxin serotype A. J Biol Chem 282:9621–9627. nate, Pr4NϩRuO4-, TPAP: A catalytic oxidant for organic synthesis. Synthesis 1994:639– 33. Eswaramoorthy S, Kumaran D, Swaminathan S (2001) Crystallographic evidence for 666. doxorubicin binding to the receptor-binding site in Clostridium botulinum neurotoxin 15. Nelson SD, McClanahan RH, Thomassen D, Gordon WP, Knebel N (1992) Investigations B. Acta Crystallogr D 57:1743–1746. ϩ of mechanisms of reactive metabolite formation from (R)-( )-pulegone. Xenobiotica 34. Dong M, et al. (2006) SV2 is the protein receptor for botulinum neurotoxin A. Science 22:1157–1164. 312:592–596. 16. Potter PM, Wadkins RM (2006) Carboxylesterases—detoxifying enzymes and targets 35. Mahrhold S, Rummel A, Bigalke H, Davletov B, Binz T (2006) The synaptic vesicle protein for drug therapy. Curr Med Chem 13:1045–1055. 2C mediates the uptake of botulinum neurotoxin A into phrenic nerves. FEBS Lett 17. Hatheway CH, Snyder JD, Seals JE, Edell TA, Lewis GE, Jr (1984) Antitoxin levels in 580:2011–2014. botulism patients treated with trivalent equine botulism antitoxin to toxin types A, B, 36. Dong M, et al. (2008) Glycosylated SV2A and SV2B mediate the entry of botulinum and E. J Infect Dis 150:407–412. neurotoxin E into neurons. Mol Biol Cell 19:5226–5237. 18. Eubanks LM, et al. (2007) An in vitro and in vivo disconnect uncovered through 37. Chai Q, et al. (2006) Structural basis of cell surface receptor recognition by botulinum high-throughput identification of botulinum neurotoxin A antagonists. Proc Natl neurotoxin B. Nature 444:1096–1100. Acad Sci USA 104:2602–2607. 38. Jin R, Rummel A, Binz T, Brunger AT (2006) Botulinum neurotoxin B recognizes its 19. Pellett S, Tepp WH, Clancy CM, Borodic GE, Johnson EA (2007) A neuronal cell-based protein receptor with high affinity and specificity. Nature 444:1092–1095. botulinum neurotoxin assay for highly sensitive and specific detection of neutralizing BIOPHYSICS serum antibodies. FEBS Lett 581:4803–4808. 39. Rummel A, Karnath T, Henke T, Bigalke H, Binz T (2004) Synaptotagmins I and II act as 20. Keller JE, Cai F, Neale EA (2004) Uptake of botulinum neurotoxin into cultured neurons. nerve cell receptors for botulinum neurotoxin G. J Biol Chem 279:30865–30870. Biochemistry 43:526–532. 40. Malizio CJ, Goodnough MC, Johnson EA (2000) Purification of Clostridium botulinum 21. Wang J, et al. (2008) Novel chimeras of botulinum neurotoxins A and E unveil type A neurotoxin. Methods Mol Biol 145:27–39. contributions from the binding, translocation, and protease domains to their func- 41. Prabakaran S, Tepp W, DasGupta BR (2001) Botulinum neurotoxin types B and E: tional characteristics. J Biol Chem 283:16993–17002. Purification, limited proteolysis by endoproteinase Glu-C and pepsin, and comparison 22. Fischer A, Montal M (2006) Characterization of Clostridial botulinum neurotoxin of their identified cleaved sites relative to the three-dimensional structure of type A channels in neuroblastoma cells. Neurotox Res 9:93–100. neurotoxin. Toxicon 39:1515. CHEMISTRY

Fischer et al. PNAS ͉ February 3, 2009 ͉ vol. 106 ͉ no. 5 ͉ 1335 Downloaded by guest on September 30, 2021