Analytical Biochemistry 396 (2010) 188–193

Contents lists available at ScienceDirect

Analytical Biochemistry

journal homepage: www.elsevier .com/locate /yabio

Rapid product analysis and increased sensitivity for quantitative determinationsof botulinum neurotoxin proteolytic activity

Benjamin Rowe a, James J. Schmidt a, Leonard A. Smith b, S. Ashraf Ahmed a,*

a Department of Cell Biology and Biochemistry, Integrated Toxicology Division, U.S. Army Medical Research Institute of Infectious Diseases, Fort Detrick, MD 21702, USAb Department of Molecular Biology, Integrated Toxicology Division, U.S. Army Medical Research Institute of Infectious Diseases, Fort Detrick, MD 21702, USA

a r t i c l e i n f o

Article history:Available online 24 September 2009

Keywords:HPLCUPLCBotulinum neurotoxinProteaseProduct analysis

0003-2697/$ - see front matter Published by Elsevierdoi:10.1016/j.ab.2009.09.034

* Corresponding author. Fax: +1 301 619 2348.E-mail address: syed.ahmed@amedd.army.mil (S.A

1 Abbreviations used: BoNT, botulinum neurotoxin; LcHPLC, high-performance liquid chromatography; Fenergy transfer; UPLC, ultra-performance liquid chromachain of serotype A/B/C/D; SNAP-25, synaptosomal-aVAMP, vesicle-associated membrane protein; SNAREsensitive factor attachment protein receptor; BoNT/Aserotype A/C/D/F; BSA, bovine serum albumin; TFultraviolet; ESI MS-MS, electrospray ionization tandcapillary electrophoresis; SDS–PAGE, sodium dodecyelectrophoresis; ALISSA, a large immunosorbent suisothiocyanate.

a b s t r a c t

The ultimate molecular action of botulinum neurotoxin (BoNT) is a Zn-dependent endoproteolytic activ-ity on one of the three SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor)proteins. There are seven serotypes (A–G) of BoNT having distinct cleavage sites on the SNARE substrates.The proteolytic activity is located on the N-terminal light chain (Lc) domain and is used extensively as theprimary target toward therapeutic development against botulism. Here we describe an improved methodusing ultra-performance liquid chromatography (UPLC) whereby quantitative data were obtained in1/10th the time using 1/20th the sample and solvent volumes compared with a widely used high-perfor-mance liquid chromatography (HPLC) method. We also synthesized a VAMP (vesicle-associated mem-brane protein)-based peptide containing an intact V1 motif that was efficiently used as a substrate byBoNT/D Lc. Although serotype C1 cleaves the serotype A substrate at a bond separated by only one res-idue, we were able to distinguish the two reactions by UPLC. The new method can accurately quantify aslow as 7 pmol of the peptide substrates for BoNT serotypes A, B, C1, and D. We also report here that thecatalytic efficiency of serotype A can be stimulated 35-fold by the addition of Triton X-100 to the reactionmixture. Combining the use of Triton X-100 with the newly introduced UPLC method, we were able toaccurately detect very low levels of proteolytic activity in a very short time. Sensitivity of the assayand accuracy and rapidity of product analysis should greatly augment efforts in therapeutic development.

Published by Elsevier Inc.

Establishment of a dependable assay method is a prerequisite toany successful therapeutic development. The seven serotypes ofbotulinum neurotoxin (BoNT,1 A–G), produced primarily by Clostrid-ium botulinum, are among the most toxic substances known (for re-views, see Refs. [1,2]) and are potential biothreat agents [3].Currently, no effective countermeasure against botulism is available,but efforts are under way to screen large compound databases forinhibitors of the toxin. Therefore, development of a fast, dependable,and quantitative assay method for the toxin is very important.

BoNT is composed of an N-terminal 50-kDa proteolytic lightchain (Lc) linked with a disulfide bond to the C-terminal 100-kDa

Inc.

. Ahmed)., light chain; Hc, heavy chain;

RET, fluorescence resonancetography; LcA/B/C/C1/D, lightssociated protein of 25 kDa;, soluble N-ethylmaleimide-

/C/D/F, botulinum neurotoxinA, trifluoroacetic acid; UV,em mass spectrometry; CE,l sulfate–polyacrylamide gelrface area; FITC, fluorescein

binding–translocating heavy chain (Hc). BoNT enters a human oranimal body and is transported to the peripheral neuronal surface[4]. After translocation through an acidic endosome into the neuro-nal cytosol, BoNT Lc domain, a Zn-dependent metalloprotease,cleaves specific proteins that prevent acetylcholine release, leadingto eventual death of the animal. Traditionally, a mouse lethalityprotocol [5] of injecting a certain amount of the whole BoNT toxininto the animal has remained the ‘‘gold standard” for BoNT assay.However, because of the requirement of special animal and safetyfacilities for this assay, more convenient methods have been devel-oped. These include the use of synthetic peptides as BoNT Lc sub-strates, followed by separation and quantification of the cleavageproducts and quantification by high-performance liquid chroma-tography (HPLC) [6]. This HPLC-based method, requiring approxi-mately 40 min for an analysis, is probably the most dependablequantitative assay available for BoNT. Subsequently, faster fluores-cence resonance energy transfer (FRET) substrate-based continu-ous fluorescence assay and a high-throughput assay have alsobeen described [7,8]. Ease of use and commercial availability ofFRET substrates for BoNT protease activities appear to make thesean attractive option for activity assays. However, FRET-based as-says often exhibit artifacts, can be greatly affected by minorchanges in assay conditions, cannot be used at concentrations

Rapid analysis and increased sensitivity for BoNT / B. Rowe et al. / Anal. Biochem. 396 (2010) 188–193 189

above Km due to the inner filter effect, and must be standardized toa particular fluorimeter [9,10]. Therefore, we sought to minimizethe HPLC time of analysis of products and the synthetic substratesin making it faster yet accurate and quantitative determination ofBoNT protease activity. By employing ultra-performance liquidchromatography (UPLC) and a short column, we cut down the anal-ysis time to 1/10th without compromising sensitivity and accu-racy. By using a nonionic detergent, we also increased thesensitivity of the assay.

Materials and methods

Materials

Recombinant BoNT Lc protease of serotype A (LcA) and serotype B(LcB) were purified as described previously [11–13], and apurification similar to that of serotype D (LcD) will be published else-where. LcC1 was purified as described previously [14]. A SNAP-25(synaptosomal-associated protein of 25 kDa) sequence-derived sub-strate peptide for LcA (SNKTRIDEANQRATKML) [6,15], a VAMP(vesicle-associated membrane protein) sequence-derived sub-strate peptide [16] for LcB (LSELDDRADALQAGASQFETSAAKLKR-KYWWKNLK), and another VAMP sequence-derived substratepeptide for LcD (LQQTQAQVDEVVDIMRVNVDKVLERDQKLSELDD),all having N-terminal acetylation and C-terminal amidation, werecustom synthesized and purified to more than 95% by Quality Con-trolled Biochemical (Hopkinton, MA, USA). The detergents Tween20, Triton X-100, Chaps, sodium cholate, Tergitol 15-5-7, Tergitol40, Tergitol 10, Tergitol 9, Tergitol 7, Ipegal CA-630, and digitonin(Sigma, St. Louis, MO, USA), as well as Nonidet P-40 (American Bio-analytical (Natick, MA, USA), were obtained from commercialsources.

Enzymatic activity assays

Here 1 unit is defined as the amount of enzyme needed to con-vert 1 lmol of substrate into products per milliliter of reactionmixture per minute. The enzymatic assays were based on HPLCor UPLC separation and measurement of the cleaved products orig-inally designed for BoNT/A protease assay with the 17-residue syn-thetic peptide corresponding to residues 187 to 203 of SNAP-25[6,17]. A master reaction mixture lacking the Lc was made, and ali-quots were stored at �20 �C. Stocks of 0.005 to 0.07 mg/ml Lc in50 mM Na–Hepes (pH 7.4) containing 0.05% Tween 20 were storedat �20 �C. Before assay, an Lc stock was thawed and diluted furtherin 50 mM Hepes (pH 7.4) containing 0.2 mg/ml bovine serum albu-min (BSA). At the time of assay, 5 ll of the Lc (see above) wasadded to 25 ll of the thawed master mix to initiate the enzymaticreaction. Components and final concentration in this 30-ll reactionmixture were 0.2 to 0.9 mM substrate peptide, 0.2 mg/ml BSA,0.001 to 0.0026 mg/ml Lc, and 50 mM Na–Hepes (pH 7.4). Whereindicated, the reaction mixture also contained 0.25 mM ZnCl2

and 5 mM dithiothreitol. To monitor any nonenzymatic breakdownof substrate, all enzymatic reactions accompanied a control withno Lc present; no products were detected in the controls.

Product analyses

The amounts of uncleaved substrate and the products were mea-sured after separation by reverse-phase HPLC (Waters) on a Hi-PoreC18 column (45 � 250 mm, Bio-Rad Laboratories, Hercules, CA, USA)with the Millennium software package (Waters). Solvent A was 0.1%trifluoroacetic acid (TFA), and solvent B was 70% acetonitrile/0.1%TFA. The flow rate was 1.0 ml/min at 25 �C. After the column wasequilibrated with 10% B, the sample was injected and the column

was held at 10% B for 2.5 min. A linear gradient to 36% B over21 min was followed by 100% B for 6 min. The newer system useda Waters Acquity UPLC system equipped with Empower Pro soft-ware employing a 1.7-lm C18 column (2.1 � 50 mm) with 0.1%TFA as solvent A and 70% acetonitrile/0.1% TFA as solvent B at a flowrate of 0.5 ml/min. LcA and LcC substrate and products were resolvedby UPLC with a 0% to 42% B gradient of the solvent over 2 min, fol-lowed by column regeneration for 0.7 min. LcB substrate and prod-ucts were resolved by UPLC with a 0% to 100% B gradient of thesolvents over 2 min, held at 100% B for 0.5 min, followed by columnregeneration for 0.5 min. LcD substrate and products were resolvedby UPLC with a 10% to 25% B gradient of the solvents over 1 min, 25%to 55% B for 0.5 min, held at 55% B for 10 s, 100% B for 1.1 min, fol-lowed by column regeneration for 0.7 min.

Results and discussion

UPLC analyses of substrate and products

The endopeptidase activity of BoNT Lc can be followed by mea-suring the disappearance of substrate or the appearance of eitherproduct. However, any method that allows simultaneous detectionand measurement of both substrate depletion and product accu-mulation allows an assessment of the possibility of substrate usedin any other contaminating reaction. Because peptides strongly ab-sorb in the ultraviolet (UV) region centered at 210 nm, continuousmonitoring of chromatographic eluate of endopeptidase reactionmixtures for UV absorbance should allow simultaneous measure-ment of substrate and products. Fig. 1 shows the UPLC elution pro-files of products and remaining substrates from the reactions ofLcA, LcB, and LcD with their respective substrates. The N- andC-terminal products of the LcA reaction on the SNAP-25 peptideare very well separated, and the whole elution was complete in2 min (Fig. 1A). Identities of the peaks as product and substrate se-quences were determined by electrospray ionization tandem massspectrometry (ESI MS-MS) analyses of the peak fractions. The sam-ple injection volume was 5 ll. A regeneration step of 1 min readiedthe column for the next sample injection. The solvent flow rate inthe column elution and regeneration was 0.5 ml/min, requiring atotal of 1.5 ml of solvent and 3 min for one complete cycle of reac-tion mixture analysis. For comparison, the inset in Fig. 1A showsthe traditionally used HPLC method [6,17] for the analysis of thesame reaction mixture requiring a 28-min analysis and 9 min ofcolumn regeneration time at the expense of 37 ml of solvent. TheUPLC method, therefore, reduced the analysis time 14-fold andthe solvent use approximately 25-fold compared with that neededwith the HPLC method.

Analysis of reaction mixtures of LcB and LcD by the same sol-vent gradient and time as with LcA (Fig. 1A) did not yield good sep-aration of substrates and products. Several trials with differentgradients allowed us to completely and reproducibly resolve thesubstrate and product peaks from these two reactions, althoughboth protocols required a longer analysis time of 3 min (Fig. 1Band C). Nonetheless, assay time was greatly reduced comparedwith the 28 min required for the earlier methods (see insets inFig. 1B and C). Although all three peptides of LcD reaction wereeluted well within 2 min of the UPLC run, the solvent gradientneeded for the separation of substrate and C-terminal product pep-tide necessitated a longer 3-min analysis. However, a much shorteranalysis time of 1 min can be adapted for the LcD reaction analysisby monitoring only the C-terminal product analysis that elutes atapproximately 0.7 min (Fig. 1C). The Lc or BSA proteins as compo-nents of the reaction mixtures elute much later, after the substrateand product peaks and during column regeneration, and thereforedo not interfere with the substrate and product peaks.

Fig. 1. UPLC and HPLC profiles for the elution of substrates and the corresponding products of LcA (A), LcB (B), and LcD (C) reactions. Here 5 ll of partially digested substratepeptides was injected into the UPLC column (2.1 � 50 mm) and eluted at a flow rate of 0.5 ml/min. For comparison, HPLC profiles of the same reaction mixtures are shown inthe insets. For the latter method, 100 ll was injected into the HPLC column (45 � 250 mm) and eluted at a flow rate of 1.0 ml/min.

190 Rapid analysis and increased sensitivity for BoNT / B. Rowe et al. / Anal. Biochem. 396 (2010) 188–193

Limits of detection

To quantitatively determine the amount of peptides, the areaunder substrate and product peaks shown in Fig. 1 is normally com-puted automatically by the built-in UPLC software. The extent ofproduct formation is computed as a ratio of the combined productarea to the total area of the substrate and product peaks. Using theoriginal concentration of the substrate peptide, the concentration ofthe products formed can be determined from the area of any of thepeaks. To determine the limit of detection in the UPLC analysis, weused various concentrations of the substrate peptides. The ampli-fied UPLC profile of 7 pmol of the LcA substrate shown in Fig. 2Awas computed automatically. In situations of larger baseline shifts,the peak could be computed manually for accuracy. For compari-son, a UPLC profile of more than 2400 pmol of the same substrateis shown in the inset of Fig. 2A. Linearity of quantitative determina-tion of this peptide was maintained between 0 and 0.5 mM, but forcomparison with other peptides, a range of 0 to 0.2 mM is shown inFig. 2B. Although higher amounts of the substrate can be used, anabsorbance above 1.5 was avoided so as to maintain linearity (seeinset in Fig. 2A). Compared with the 17-residue LcA substrate, thoseof LcB and LcD were much larger, with 35 and 34 residues, respec-tively. As a result, solubilities of these peptides were lower, yet lin-earity of detection could be confidently obtained between 0 and0.2 mM. This limitation of the substrate solubility was not a prob-lem for the enzymatic reaction measurements because kinetic con-stants can be determined at subsaturating substrate concentrations(see below) such as those for BoNT/A and LcA [11,15].

LcC and LcD substrates

Unlike other serotypes acting on only one SNARE protein,BoNT/C can proteolyze both syntaxin and SNAP-25 [18,19]. TheBoNT/C proteolytic site on SNAP-25 is adjacent to that of

BoNT/A. Although a 61-residue SNAP-25 peptide was not cleavedby either BoNT/C1 or LcC1 [20], we recently showed that the 17-mer SNAP-25 LcA substrate peptide was efficiently proteolyzedby LcC1 [14]. Although separated by just one residue, LcA andLcC1 cleavage products from this peptide were reproducibly dis-tinguished from each other by the UPLC analyses (Fig. 3). Dem-onstration of this peptide as a substrate for LcC1 makes it aconvenient dual-purpose substrate for LcA [6] and LcC1 [14].

VAMP is known to have two SNARE motifs (V1 and V2) impor-tant for recognition by BoNT proteases [21]; BoNT/D cleaves the59K–60L bond of the protein. A 31-mer VAMP peptide (residues46–76) lacking nearly all of the V1 motif [22] and a 51-mer VAMPpeptide (residues 44–94) with a partial V1 motif [20], with bothcontaining an intact V2 motif, were not substrates for BoNT/D orLcD. Thus, it is possible that an intact V1 motif is essential for rec-ognition by BoNT/D or LcD. Therefore, we synthesized a 34-merpeptide (32-LQQTQAQVDEVVDIMRVNVDKVLERDQK-LSELDD-65),keeping an intact V1 motif (underlined residues 38–47) and a par-tial V2 motif (underlined residues 63–65 of 63-71). As shown inFig. 1C, the peptide was very efficiently cleaved by LcD. Our resultsproved that an intact V2 motif is not essential for substrate recog-nition by LcD and that at least 6 amino acid residues C terminal tothe scissile bond are sufficient for recognition, although 10 suchresidues were suggested previously [22], and the results supporta less important role of the V2 motif from mutational studies[23]. Interestingly, this sequence is also best recognized by BoNT/F [24]. To determine whether the peptide was useful as a routinesubstrate, we determined its catalytic constants from a Linewe-aver–Burk plot of reaction velocity versus substrate concentration(Fig. 4). The Km of the substrate was calculated as 0.6 mM, and itskcat was calculated as 42/s. For comparison, Km (and kcat) values forthe same substrate catalysis by BoNT/A, LcA, and LcC1 were 5 and1.7 mM (47 and 4.7/s) [15], 3 mM (12/s) [25], and 4.5 mM (5/min)[14], respectively.

Fig. 2. UPLC sensitivity limits of peptide detection. (A) Unambiguous detection of 7 pmol (5 ll of 1.4 lM) of the LcA substrate eluting at 1.58 min is shown in an expandedscale. The inset shows the elution of 2425 pmol in a normal scale. (B) Linearity of the peak area with the amount of injected LcA substrate (h), LcB substrate (s, lower curve),and LcD substrate (s, upper curve). LcA and LcD substrates were monitored at 210 nm, and LcB substrate was monitored at 280 nm (due to the presence of two tryptophansin the latter substrate). Peak areas were integrated from traces like those shown in panel A. LcA, LcB, and LcD substrate measurements were made three, three, and two times,respectively.

Fig. 3. Comparison of UPLC elution profiles of LcA (dotted line) and LcC1 (solid line)reaction products from the 17-mer SNAP-25 peptide. The sequences SNKTRIDEANQand RATKML are products of the LcA reaction, and the sequences SNKTRIDEANQRand ATKML are products of the LcC1 reaction.

Fig. 4. Lineweaver–Burk plot of LcD reaction. The 30-ll assay reaction mixturescontained 66 nM LcD, 5 mM dithiothreitol, 0.25 mM ZnCl2, 0.2 mg/ml BSA, 50 mMNa–Hepes (pH 7.4), and variable amounts of the substrate. Each data point is anaverage of five assays. The regression coefficient for the line drawn through the datapoints is R = 0.99867.

Fig. 5. Effects of BSA and Tween 20 on the proteolytic activity of LcA. To determinethe effect of BSA on LcA stability, 2.5-ll aliquots containing 5 lg of LcA were storedwithout or with 1.0 mg/ml BSA at �20 �C for 4 days. The aliquots were assayed at37 �C in triplicate in 30-ll reaction mixtures containing the indicated concentra-tions of Tween 20 without or with 0.5 mg/ml BSA. D, LcA-only aliquots assayed inthe absence of BSA; d, LcA-only aliquots assayed with the addition of 0.5 mg/mlBSA; N, LcA (+ BSA) aliquots assayed in the absence of additional BSA; s, LcA (+ BSA)aliquots assayed with further addition of 0.5 mg/ml BSA.

Rapid analysis and increased sensitivity for BoNT / B. Rowe et al. / Anal. Biochem. 396 (2010) 188–193 191

Stimulation of activity by BSA and detergents

Schmidt and Bostian [15] demonstrated that increasing concen-trations of BSA increased the BoNT/A catalysis, and including 1 mg/ml BSA in the reaction mixture increased the catalytic efficiency ofBoNT/A nearly 30-fold. Because stock solutions of proteins are of-ten stored in nonionic detergents as a stabilizer, we investigatedthe effect of Tween 20 on the proteolytic activity of LcA in the

presence and absence of BSA (Fig. 5). In general, including BSAand Tween 20 in the reaction mixture significantly stimulatedthe LcA reaction. Interestingly, the activity also increased withincreasing concentrations of Tween 20 regardless of the presenceof BSA. There was an apparent synergistic effect of BSA and Tween20 up to 1% of the latter concentration. Stimulation by BSA wasmost pronounced at the lowest Tween 20 concentration (Fig. 5),and in the absence of the detergent a nearly 5-fold increase inactivity was observed (Fig. 6). All nonionic detergents, includingthe plant product digitonin, had stimulating effects, but the anionicsodium cholate was inhibitory. Maximum stimulation of LcA activ-ity was displayed by the nonionic detergents Triton X-100, Igepal,and NP-40 (Fig. 6). Like the effect of BSA on BoNT/A activity [15],Triton X-100 improved both Km and Vmax values of the LcA reactionthat resulted in a 35-fold increase in the catalytic efficiency (Fig. 7).

We also investigated the effects of Tween 20 and BSA on the LcDactivity. The detergent (0.007%) alone stimulated the activity6-fold, BSA (0.2 mg/ml) alone stimulated the activity 8-fold, andtogether they stimulated the activity 11-fold. Because all BoNTdomains, including LcA and LcD, are prone to surface denaturation[26], the observed stimulating effects on LcA and LcD activity may

Fig. 6. Effects of 1% detergents on the proteolytic activities of LcA. TGT, Tergitol;NP-40, Nonidet P-40.

Table 1Effects of Triton X-100 on various forms of LcA.

BoNT/A Lc Activity (U/mg) in theabsence of detergent

Activity (U/mg) in thepresence of 1% TritonX-100

Full length, fully active 3.0 32Full length, poorly active 0.3 32Truncated, having

residues 1–420, poorlyactive

0.22 0.32

192 Rapid analysis and increased sensitivity for BoNT / B. Rowe et al. / Anal. Biochem. 396 (2010) 188–193

suggest that the detergents preserve the most active conformationof LcA that was reflected in the activity stimulation. In fact, protec-tion of LcA from precipitation due to surface denaturation was re-cently demonstrated [26]. Except for TGT-7, variations of the effectwithin Tergitols are not significant. The subtle differences in chem-ical structure of these detergents might possibly have differentinfluences on protein conformation. All of the nonionic detergentsare chemically related, with some form of alkyl chain (straight orbranched) linked to a phenyl ring, which in turn is connected toa polyethylene glycol chain of varying lengths via an oxygenbridge. Primary differences are the length and branching of the al-kyl chain component and the length of the polyethylene glycolcomponent.

The results described in Figs. 5 to Fig. 7 suggest that includinghigh concentrations of the nonionic detergents should be able todetect activity of an otherwise poorly active BoNT or its Lc. Suchactivity formulation will be extremely important in detecting lowlevels of the active toxin in environmental samples given theincreasing bioterrorism threats since the anthrax letter attacks of2001. As a test case, we selected a wild-type LcA that had lost morethan 90% activity during prolonged storage and a C-terminallytruncated LcA that had less than 10% activity of a fully activewild-type LcA. As shown in Table 1, including 1% Triton X-100 fullyrestored the activity of the wild-type LcA but had a marginal effecton the inherently low-activity truncated LcA. This result may indi-

Fig. 7. Effects of Triton X-100 on the Km and Vmax values of LcA. The reactionmixtures contained 0.5 to 2.0 mM peptide substrate, 1% Triton X-100, 1.6 lg/mlLcA, and 50 mM Hepes (pH 7.4) and were incubated at 37 �C for 5 min. ALineweaver–Burk plot for the effect of 1% Triton X-100 (solid line) is compared withthe control (dashed line, generated using published data [11]).

cate that the detergent affects the protein structure by stabilizingthe most active conformation, but the truncated Lc cannot achievethat conformation.

By employing UPLC and nonionic detergents, we demonstrateda rapid and sensitive method of determining BoNT proteaseactivities.

Advantage of the method

Since the introduction of our HPLC method using peptide sub-strates [6], several analytical methods for quantitative analysis ofBoNT protease activity have appeared [6–9,15,20,27–29]. Each ofthese methods has its own advantages and disadvantages basedon the particular application. The widely used HPLC method [6] re-quires 28 min and 28 ml of solvent for a single analysis. Using sim-ilar peptide substrates, a capillary electrophoresis (CE) methodrequired 12 to 15 min for analysis, excluding the capillary columnregeneration time for analysis of one sample [20], and a shorteranalysis time of 8 min was also reported by another group [29].Two widely used activity determinations using full-length SNAP-25 substrate employ densitometric analyses of Coomassie-stainedsubstrate and product bands in sodium dodecyl sulfate–polyacryl-amide gel electrophoresis (SDS–PAGE) gels [30] and chromophore-conjugated antibody-stained substrate and products in Westernblots [31] generally require 6 to 24 h to obtain data. Our UPLCmethod described here takes only 2 min of analysis time using1 ml of solvent; thus, it is far superior to the HPLC, CE, SDS–PAGE,and Western blot methods. UPLC analysis requires 1 to 5 ll of sam-ple compared with 20 to 100 ll with HPLC [6,15] and is more sen-sitive, helping to reduce both analysis time and solvent use. Inaddition, it greatly reduces the generation of organic waste.

Two recent sensitive analytical methods are also worth discuss-ing. One report detected BoNT/A at a concentration of 0.1 ng/mlusing CE [28], and another report detected BoNT/A at an unprece-dented attomolar concentration using a large immunosorbent sur-face area (ALISSA) method [27]. Both of these techniques werehighly sensitive and accurate, using fluorescent and dual-labeledBoNT/A peptide substrates, with the latter also using antibodybead-captured BoNT/A. Our UPLC method described here circum-vents the expensive fluorescein isothiocyanate (FITC) and biotinlabeling [28] and antibody bead and FRET labeling [27]. Althoughour group first introduced the use of fluorescence-labeled BoNTsubstrates [7,8], we have limited their use (because of the inner fil-ter effect and often exhibiting artifacts [9,10]) when an alternate,unmodified, and affordable substrate is available for quantitativeactivity determination. The peptides used in this study representsuch unmodified substrates. In addition, our total time of no morethan 10 min (5 min for assay plus 2 min for analysis) is much fasterthan that required with the ALISSA capture (150 min) [27] and CE(>360 min) [28] methods; product analysis alone with the CEmethod required 12 to 20 min. Our assay uses soluble, unmodifiedBoNT protease, whereas the most sensitive ALISSA method usescaptured toxin whose native structure might be subjected to phys-ical distortion by immobilization with antibody. The high sensitiv-

Rapid analysis and increased sensitivity for BoNT / B. Rowe et al. / Anal. Biochem. 396 (2010) 188–193 193

ity of these methods [27,28] is also compromised by long assaytimes. Our UPLC method of analysis was designed to rapidly deter-mine protease activity so as to screen and characterize large num-bers of inhibitors being made available for drug discovery, not todetect low levels of the toxin in biological samples that has beenelegantly demonstrated in the immunological enrichment ofBoNT/A from artificial spiked biological samples [27]. Finally, aper-sample estimated reagent cost of less than $0.30 (U.S.) withour method is 50 times less than the $15 (U.S.) required with theALISSA method [27].

To summarize, our assay procedure combines an extremely fastturnaround time, low reagent consumption (and low production oforganic waste), and a very low cost. It is clearly an ideal method forbotulinum protease assays, kinetic studies, development of furtherimproved substrates, and inhibitor characterization.

Acknowledgments

The research described here was supported by the Joint Scienceand Technology Office for Chemical and Biological Defense (JSTO–CBD, 3.10011_06_RD_B and 3.10012_06_RD_B). We thank StephenI. Toth, Matthew L. Ludivico, and Trista Haupt for assistance withthe UPLC setup and enzyme assays; Ernst Brueggemann for themass spectrometry data; and John Cardellina for detergent struc-tures. Opinions, interpretations, conclusions, and recommenda-tions are those of the authors and are not necessarily endorsedby the U.S. Army.

References

[1] C. Montecucco, G. Schiavo, Structure and function of tetanus and botulinumneurotoxins, Q. Rev. Biophys. 28 (1995) 423–472.

[2] G. Schiavo, C.C. Shone, M.K. Bennett, R.H. Scheller, C. Montecucco, Botulinumneurotoxin type C cleaves a single Lys–Ala bond within the carboxyl-terminalregion of syntaxins, J. Biol. Chem. 270 (1995) 10566–10570.

[3] S.S. Arnon, R. Schechter, T.V. Inglesby, D.A. Henderson, J.G. Bartlett, M.S. Ascher,E. Eitzen, A.D. Fine, J. Hauer, M. Layton, S. Lillibridge, M.T. Osterholm, T.O’Toole, G. Parker, T.M. Perl, P.K. Russell, D.L. Swerdlow, K. Tonat, Botulinumtoxin as a biological weapon: medical and public health management, JAMA285 (2001) 1059–1070.

[4] L.L. Simpson, Identification of the major steps in botulinum toxin action, Annu.Rev. Pharmacol. Toxicol. 44 (2004) 167–193.

[5] P.A. Hanna, J. Jankovic, Mouse bioassay versus Western blot assay forbotulinum toxin antibodies: correlation with clinical response, Neurology 50(1998) 1624–1629.

[6] J.J. Schmidt, K.A. Bostian, Proteolysis of synthetic peptides by type A botulinumneurotoxin, J. Protein Chem. 14 (1995) 703–708.

[7] J.J. Schmidt, R.G. Stafford, C.B. Millard, High-throughput assays for botulinumneurotoxin proteolytic activity: serotypes A, B, D, and F, Anal. Biochem. 296(2001) 130–137.

[8] J.J. Schmidt, R.G. Stafford, Fluorigenic substrates for the protease activities ofbotulinum neurotoxins, serotypes A, B, and F, Appl. Environ. Microbiol. 69(2003) 297–303.

[9] H.B. Hines, A.D. Kim, R.G. Stafford, S.S. Badie, E.E. Brueggeman, D.J. Newman, J.J.Schmidt, Use of a recombinant fluorescent substrate with cleavage sites for allbotulinum neurotoxins in high-throughput screening of natural productextracts for inhibitors of serotypes A, B, and E, Appl. Environ. Microbiol. 74(2008) 653–659.

[10] C.G. Knight, Fluorimetric assays of proteolytic enzymes, Methods Enzymol.248 (1995) 18–34.

[11] S.A. Ahmed, P. McPhie, L.A. Smith, Autocatalytically fragmented light chain ofbotulinum A neurotoxin is enzymatically active, Biochemistry 42 (2003)12539–12549.

[12] M.J. Jensen, T.J. Smith, S.A. Ahmed, L.A. Smith, Expression, purification, andefficacy of the type A botulinum neurotoxin catalytic domain fused to twotranslocation domain variants, Toxicon 41 (2003) 691–701.

[13] J. Gilsdorf, N. Gul, L.A. Smith, Expression, purification, and characterization ofClostridium botulinum type B light chain, Protein Expr. Purif. 46 (2006) 256–267.

[14] R. Rawat, S.A. Ahmed, S. Swaminathan, High level expression of the light chainof botulinum neurotoxin serotype C1 and an efficient HPLC assay to monitorits proteolytic activity, Protein Expr. Purif. 60 (2008) 165–169.

[15] J.J. Schmidt, K.A. Bostian, Endoproteinase activity of type A botulinumneurotoxin: substrate requirements and activation by serum albumin, J.Protein Chem. 16 (1997) 19–26.

[16] P. Foran, C.C. Shone, J.O. Dolly, Differences in the protease activities of tetanusand botulinum B toxins revealed by the cleavage of vesicle-associatedmembrane protein and various sized fragments, Biochemistry 33 (1994)15365–15374.

[17] S.A. Ahmed, L.A. Smith, Light chain of botulinum A neurotoxin expressed as aninclusion body from a synthetic gene is catalytically and functionally active, J.Protein Chem. 19 (2000) 475–487.

[18] P. Foran, G.W. Lawrence, C.C. Shone, K.A. Foster, J.O. Dolly, Botulinumneurotoxin C1 cleaves both syntaxin and SNAP-25 in intact andpermeabilized chromaffin cells: correlation with its blockade ofcatecholamine release, Biochemistry 35 (1996) 2630–2636.

[19] L.C. Williamson, J.L. Halpern, C. Montecucco, J.E. Brown, E.A. Neale, Clostridialneurotoxins and substrate proteolysis in intact neurons: botulinumneurotoxin C acts on synaptosomal-associated protein of 25 kDa, J. Biol.Chem. 271 (1996) 7694–7699.

[20] A.L. Purcell, H.M. Hoard-Fruchey, A capillary electrophoresis method to assaycatalytic activity of botulinum neurotoxin serotypes: implications forsubstrate specificity, Anal. Biochem. 366 (2007) 207–217.

[21] O. Rossetto, G. Schiavo, C. Montecucco, B. Poulain, F. Deloye, L. Lozzi,C.C. Shone, SNARE motif and neurotoxins, Nature 372 (1994) 415–416.

[22] S. Yamasaki, A. Baumeister, T. Binz, J. Blasi, E. Link, F. Cornille, B. Roques, E.M.Fykse, T.C. Sudhof, R. Jahn, H. Niemann, Cleavage of members of thesynaptobrevin/VAMP family by types D and F botulinal neurotoxins andtetanus toxin, J. Biol. Chem. 269 (1994) 12764–12772.

[23] R. Pellizzari, S. Mason, C.C. Shone, C. Montecucco, The interaction of synapticvesicle-associated membrane protein/synaptobrevin with botulinumneurotoxins D and F, FEBS Lett. 409 (1997) 339–342.

[24] J.J. Schmidt, R.G. Stafford, Botulinum neurotoxin serotype F: identification ofsubstrate recognition requirements and development of inhibitors with lownanomolar affinity, Biochemistry 44 (2005) 4067–4073.

[25] S.A. Ahmed, M.A. Olson, M.L. Ludivico, J. Gilsdorf, L.A. Smith, Identification ofresidues surrounding the active site of type A botulinum neurotoxin importantfor substrate recognition and catalytic activity, Protein J. 27 (2008) 151–162.

[26] S.I. Toth, L.A. Smith, S.A. Ahmed, Extreme sensitivity of botulinum neurotoxindomains towards mild agitation, J. Pharm. Sci. 98 (2009) 3302–3311.

[27] K. Bagramyan, J.R. Barash, S.S. Arnon, M. Kalkum, Attomolar detection ofbotulinum toxin type A in complex biological matrices, PLoS ONE 3 (2008)e2041.

[28] H.P. Wei, Y. Kang, Z.P. Zhang, Z.Q. Cui, Y.F. Zhou, X.E. Zhang, Micellarelectrokinetic chromatography and laser induced fluorescence detection ofbotulinum neurotoxin type A activity using a dual-labelled substrate, Int. J.Environ. Anal. Chem. 88 (2008) 947–956.

[29] T.D. Laing, A.J. Marenco, D.M. Moore, G.J. Moore, D.C. Mah, W.E. Lee, Capillaryelectrophoresis laser-induced fluorescence for screening combinatorialpeptide libraries in assays of botulinum neurotoxin A, J. Chromatogr. B 843(2006) 240–246.

[30] S. Cai, B.R. Singh, Role of the disulfide cleavage induced molten globule state oftype A botulinum neurotoxin in its endopeptidase activity, Biochemistry 40(2001) 15327–15333.

[31] R.V. Kukreja, S. Sharma, S. Cai, B.R. Singh, Role of two active site Glu residues inthe molecular action of botulinum neurotoxin endopeptidase, Biochim.Biophys. Acta 1774 (2007) 213–222.