ANALYTICALBIOCHEMISTRY
Analytical Biochemistry 351 (2006) 84–92
www.elsevier.com/locate/yabio
The use of Endopep–MS for the detection of botulinum toxins A, B, E, and F in serum and stool samples
Suzanne R. Kalb a, Hercules Moura a, Anne E. Boyer a, Lisa G. McWilliams b, James L. Pirkle a, John R. Barr a,¤
a Centers for Disease Control and Prevention, National Center for Environmental Health/Agency for Toxic Substances and Disease Registry, Atlanta, GA 30341, USA
b Battelle Memorial Institute under contract at the Centers for Disease Control and Prevention, National Center for Environmental Health, Atlanta, GA 30341, USA
Received 17 October 2005Available online 3 February 2006
Abstract
Botulinum neurotoxin (BoNT) causes the disease botulism, which can be lethal if untreated. Previous work in our laboratory focusedon developing Endopep–MS, a mass spectrometric-based endopeptidase method for the detection and diVerentiation of BoNT serotypes.We have expanded this eVort to include an antibody capture method to partially purify and concentrate BoNT from serum and stoolextract samples for the Endopep–MS assay. Because complex matrices such as serum and stool contain abundant endogenous proteases,this technique was needed to remove most proteases from the sample while concentrating BoNT from a sample size of 100 to 500 �l to20 �l. When this antibody capture method is combined with the Endopep–MS reaction, limits of detection in 500 �l of spiked humanserum are 10 mouse LD50 (20 mouse LD50/ml) for BoNT A, 0.5 mouse LD50 (1 mouse LD50/ml) for BoNT B, 0.1 mouse LD50 (0.2 mouseLD50/ml) for BoNT E, and 0.5 mouse LD50 (1 mouse LD50/ml) for BoNT F. The limits of detection in spiked stool extracts are somewhathigher due to the high-protease environment of stool extract that also requires use of protease inhibitors. The entire method can be per-formed in as short a time as 4 h.© 2006 Elsevier Inc. All rights reserved.
Keywords: Botulinum neurotoxin; Botulism
Botulinum neurotoxin (BoNT)1 is produced by somespecies of the genus Clostridium, particularly Clostridiumbotulinum, Clostridium butyricum, and Clostridium baratii.BoNTs cause the disease known as botulism, which in mostcases is contracted through ingestion of food containing the
* Corresponding author. Fax: +1 770 488 4609.E-mail address: [email protected] (J.R. Barr).
1 Abbreviations used: BoNT, botulinum neurotoxin; IgG, immunoglobu-lin G; SNAP-25, synaptosome-associated protein; VAMP-2, vesicle-asso-ciated membrane protein 2; ELISA, enzyme-linked immunosorbent assay;Ab, antibody; BSA, bovine serum albumin; PBS, phosphate-buVered sa-line; CDC, Centers for Disease Control and Prevention; AEBSF, 4-(2-ami-noethyl)benzenesulfonyl Xuoride hydrochloride; TFA, triXuoroacetic acid;CHCA, �-cyano-4-hydroxy cinnamic acid; MALDI, matrix-assisted laserdesorption/ionization; LOD, limit of detection; TOF, time-of-Xight.
0003-2697/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.ab.2006.01.027
toxin [1,2]. Other sources of botulism involve colonizationof the bacteria in the gastrointestinal tract of infants orimmunocompromised people and contact of the bacteriawith a wound [1]. Because of its high toxicity, availability,and ease of preparation, BoNT is also a likely agent forbioterrorism [3]. Rapid determination of exposure to BoNTis an important public health goal. Treatment of botulisminvolves administration of equine antineurotoxin immuno-globulin G (IgG) and is most eVective when administeredwithin 24 h after exposure [1]. The treatment must beadministered carefully due to potentially serious negativeside eVects associated with the equine-based treatment.
Our laboratory previously has reported on the develop-ment of an assay for BoNT termed the Endopep–MSmethod [4,5]. In simple matrices, this method has proved to
Use of Endopep–MS for detection of botulinum toxins / S.R. Kalb et al. / Anal. Biochem. 351 (2006) 84–92 85
be successful for detecting all seven BoNT toxin types andinvolves incubation of BoNT with a peptide substrate thatmimics the toxin’s natural target, either synaptosome-asso-ciated protein (SNAP-25) or vesicle-associated membraneprotein 2 (VAMP-2, also called synaptobrevin 2) [6–14].Each BoNT cleaves the peptide substrate in a speciWc toxin-dependent location that is diVerent for each of the BoNTtoxin types [2,4,5]. The reaction mixture then is introducedinto a mass spectrometer, which detects any peptides withinthe mixture and accurately reports the mass of each one.Detection of the peptide cleavage products correspondingto their speciWc toxin-dependent location indicates the pres-ence of a particular BoNT toxin type. If the peptide sub-strate either remains intact or is cleaved in a location otherthan the toxin-speciWc site, then that BoNT toxin type isnot present. Previous publications [4,5] have demonstratedthat this method can detect BoNT at levels comparable toor lower than levels detected with mouse bioassays, the cur-rent standard [15].
The Endopep–MS method also has advantages over themouse bioassay, including speed of analysis and no require-ment of use of live animals for the analysis. It also is moresensitive and does not show the cross-reactivity that limitsthe enzyme-linked immunosorbent assay (ELISA), which isthe only other validated method for BoNT detection[16,17]. In a recent validation study, the ELISA showed a1.5% false-positive rate for BoNT A and a 28.6% false-posi-tive rate for BoNT F [16]. The ELISA is based on the recog-nition of toxin antigenic sites and is less sensitive than themouse bioassay with a reported sensitivity of approxi-mately 10 mouse LD50/ml in a 1-day test [16,18]. TheELISA is more rapid than the mouse bioassay, but theELISA is not a functional assay in that it does not measurethe activity of the toxin as the Endopep–MS method does.Measuring the activity of the toxin is important becauseonly active toxin poses a threat to the health and life ofmammals. Positive samples in the ELISA still require con-Wrmation by the mouse bioassay [16].
In its previously reported form, the Endopep–MSmethod alone was not suYcient to detect low levels (<100mouse LD50) of BoNT in a complex matrix such as serum
or stool extract. Antibody (Ab) aYnity methods for theconcentration and isolation of a speciWc protein from acomplex mixture are used frequently in diagnostic proce-dures. Therefore, we employed an Ab aYnity method forpuriWcation and concentration of BoNT from serum andfrom stool extracts. This article reports the successful use ofAb-coated magnetic beads to isolate and concentrateBoNT from spiked serum, spiked stool extract, and clinicalstool extracts combined with the successful use of theEndopep–MS method to detect the BoNT in the samples.
Materials and methods
Materials
BoNT complexes for A, B, E, and F toxin types and poly-clonal rabbit-speciWc IgGs were obtained from Metabiolo-gics (Madison, WI, USA). The BoNT complexes wereprovided at 1�g/ml total protein in 50mM sodium acetate,2mg/ml gelatin, and 3 mg/ml bovine serum albumin (BSA) atpH 4.2. The toxin activities in mouse LD50/ml protein arefrom the supplier as follows: BoNT A at 3£104 mouseLD50/ml, BoNT B at 2£104 mouse LD50/ml, BoNT E at8£101 mouse LD50/ml, and BoNT F at 4£103 mouse LD50/ml. The polyclonal rabbit-speciWc IgGs were supplied byMetabiologics in 150mM potassium phosphate (pH 7.4) atthe supplier-indicated levels: antitype A at 4.61mg/ml, anti-type B at 5.46mg/ml, antitype E at 8.26 mg/ml, and antitypeF at 7.35mg/ml. Dynabeads Protein G were purchased fromDynal (Lake Success, NY, USA) at 1.3g/cm3 in phosphate-buVered saline (PBS, pH 7.4) containing 0.1% Tween 20 and0.02% sodium azide. All chemicals were obtained fromSigma–Aldrich (St. Louis, MO, USA) except where indicatedotherwise. Gelatin phosphate buVer, consisting of bovine gel-atin (2g/L) and sodium phosphate dibasic (anhydrous, 4 g/L),was obtained through the ScientiWc Resources Program atthe Centers for Disease Control and Prevention (CDC,Atlanta, GA, USA). Peptide substrates were synthesized byLos Alamos National Laboratory (Los Alamos, NM, USA)(Table 1). Control serum and stool extracts were collectedfrom anonymous donors, with no demographic information
Table 1Peptide sequences for Endopep–MS method along with observed (M + H)+ of substrate and cleavage products for each serotype
Note. The cleavage site of each substrate is depicted in bold and underlined. NT, N terminal; CT, C terminal.
Peptide Sequence m/z observed
BoNT A substrate Biotin-KGSNRTRIDEANQRATRMLGGK-Biotin 2910.6BoNT A NT product Biotin-KGSNRTRIDEANQ 1714.3BoNT A CT product RATRMLGGK-Biotin 1215.3BoNT B substrate LSELDDRADALQAGASQFETSAAKLK RKYWWKNLK 4037.4BoNT B NT product LSELDDRADALQAGASQ 1759.4BoNT B CT product FETSAAKLKRKYWWKNLK 2296.7BoNT E substrate IIGNLRHMALDMGNEIDTQNRQIDRIMEKAD 3610.6BoNT E NT product IIGNLRHMALDMGNEIDTQNRQIDR 2922.8BoNT E CT product IMEKAD 706.3BoNT F substrate LQQTQAQVDEVVDIMRVNVDKVLERDQKLSELDDRADAL 4496.0BoNT F NT product LQQTQAQVDEVVDIMRVNVDKVLERDQ 3168.8BoNT F CT product KLSELDDRADAL 1345.8
86 Use of Endopep–MS for detection of botulinum toxins / S.R. Kalb et al. / Anal. Biochem. 351 (2006) 84–92
being obtained. Because samples were collected without anyidentiWers or demographic information and pooled, thesecollections were exempt from human subjects review. Posi-tive and negative clinical stool samples with all identiWersremoved were provided by the National Botulism Surveil-lance and Reference Laboratory at the CDC.
IgG binding to protein G beads
The IgG was immobilized to the Dynabeads Protein Gas described in the supplier’s protocol using 50�g of IgGfor one toxin type diluted into 500 �l of PBS for every100�l of Dynabeads Protein G. The IgG-coated Dynabe-ads also were crosslinked using the supplier’s protocol.Crosslinked IgG-coated Dynabeads were stored in PBS–Tween buVer (PBS with 0.05% Tween 20) at 4 °C for up to12 weeks.
Preparation of samples
Stock toxin was diluted to working levels in citrate–BSAbuVer consisting of 9 mg/ml of BSA in 0.1 M citrate buVerat pH 5.5. BoNT is very toxic and so requires appropriatesafety measures. All neurotoxins were handled within aclass 2 biosafety cabinet equipped with HEPA Wlters. Forwork without the IgG-coated beads, the toxin was com-bined with 5 �l of either human serum or human stoolextract at various levels (listed in a later section). For workusing the IgG-coated beads, the toxin was spiked into100�l of either serum or stool extract or 500�l of serum atthe levels listed in a later section. The stool extract was pre-pared by adding 10 to 20 ml of gelatin phosphate buVer to10 to 20 g of stool. The mixture was agitated brieXy andthen centrifuged at 2500 rpm for 30 min. The supernatant ofstool extract was removed, aliquoted, and stored at ¡80 °Cuntil needed.
Sample binding
For samples of 100�l volume, 20�l of IgG-coated beadswas washed Wve times with 100�l each of PBS–Tween buVer.Then 100�l of casein blocking buVer (10 g of casein in 1 L ofPBS) was added to the beads, and the beads were incubatedat 37°C for 30 min without agitation. The casein buVer wasremoved, and the serum or stool extracts were added to thebeads with 5�l of a protease inhibitor mixture consisting of5% casein, 4.5mg/ml of 4-(2-aminoethyl)benzenesulfonylXuoride hydrochloride (AEBSF), 25 mg/ml of 6-aminohexa-noic acid, 3.15 mg/ml of antipain, and 20mg/ml of (2S,3S)-trans-epoxysuccinyl-L-leucylamido-3-methylbutane ethylester. Samples were incubated at 37°C for 2 h. The superna-tants were removed from the beads, which were washed Wvetimes with 100�l each of PBS–Tween buVer, followed by twowashes with 100�l each of water. The same protocol was fol-lowed for samples of 500�l volume except that all reagentswere multiplied by 5, so that 100�l of IgG-coated beads wereused with the 500-�l samples.
Endopep–MS reaction
The reaction was performed as described previously [4,5]with a few modiWcations. In all cases, the Wnal reaction vol-ume was 20 �l; the Wnal concentration of the reaction buVerwas 0.05 M Hepes (pH 7.3), 25 mM dithiothreitol, 20 mMZnCl2, and 1 mg/ml BSA; and the Wnal concentration of thepeptide substrate was 50 pmol/�l. For work without beads,serum or stool extracts and BoNT were combined withreaction buVer, peptide substrate, and 2.5 �l of the proteaseinhibitor cocktail. For experiments with beads and serum,only reaction buVer and peptide substrate were added tothe beads. Experiments with beads and stool extract usedthe protease inhibitor mixture (2.5 �l) in addition to thereaction buVer and peptide substrate. All samples then wereincubated at 37 °C for 4 h. For Ab neutralization experi-ments, BoNT at 100 mouse LD50 was spiked into 18 �l ofthe reaction buVer and 1 �l of either water or speciWc IgG insolution. After 2 h of incubation at 37 °C, 1 �l of the peptidesubstrate was added to the mixture. The samples then wereincubated at 37 °C for 4 h.
MS detection
After incubation, 18 �l of each 20-�l reaction superna-tant was combined with 2 �l of a 1% triXuoroacetic acid(TFA) solution to quench the reaction, and the resultingsolution was desalted and concentrated with C18 minicol-umns (C18 Zip Tips, Millipore, Bedford, MA, USA) usingthe standard protocol. The peptides were eluted in 2 �l of80% acetonitrile/0.1% TFA and mixed with 8 �l of �-cyano-4-hydroxy cinnamic acid (CHCA) at 5 mg/ml in 50% aceto-nitrile/0.1% TFA/1 mM ammonium citrate. We pipetted0.5 �l of this mixture onto each spot of a 192-spot matrix-assisted laser desorption/ionization (MALDI) plate(Applied Biosystems, Framingham, MA, USA). Mass spec-tra of each spot were obtained by scanning from 650 to4500 m/z in MS-positive ion reXector mode on an AppliedBiosystems 4700 Proteomics Analyzer. The instrument usesa nitrogen laser at 337 nm, and each spectrum is an averageof 2400 laser shots.
Results
The Endopep–MS method is a rapid assay to detect anddiVerentiate active BoNT types. Because the mouse bioas-say is the standard method for detection and quantiWcationof BoNT, the toxin usually is quantiWed via activity interms of mouse LD50. The Endopep–MS method also mea-sures the activity of the toxin; therefore, we report here theactivity of the toxin in mouse LD50/ml. Although the activ-ity of toxin varies and the absolute amount of toxinrequired to kill a mouse is not well understood, it is gener-ally estimated that 1 mouse LD50/ml is approximatelyequivalent to 10 to 20 pg/ml [19] or approximately 70 to140 fM. As described below, in clinical matrices such asserum and stool samples, protease activities greatly reduced
Use of Endopep–MS for detection of botulinum toxins / S.R. Kalb et al. / Anal. Biochem. 351 (2006) 84–92 87
the sensitivity of the assay by digesting the substrate andproduct peptides that are essential components of thismethod. To alleviate this problem, we introduced the use ofAbs and protease inhibitors to the method.
IneVectiveness of protease inhibitors alone
One potential choice for overcoming protease activitysuccessfully is to target proteases with protease inhibitors.Although the use of protease inhibitors alone in the Endo-pep–MS method assisted in decreasing the protease activityin complex matrices, it did not enable adequate BoNTdetection in serum and stool samples. This is demonstratedby comparing a control reaction with BoNT B with a reac-tion also containing serum with protease inhibitors. Themass spectrum for the BoNT B reaction should contain apeak corresponding to the intact unreacted substrate at m/z4037.4 as well as peaks corresponding to the toxin cleavageproducts at m/z 1759.5 (N terminal) and 2296.8 (C termi-nal). These three peaks can be seen in Fig. 1A, whichdepicts the mass spectrum of a reaction containing 5 mouseLD50 of BoNT B. An additional peak at m/z 2019.6 corre-sponds to the doubly charged ion of the unreacted peptidesubstrate. Fig. 1B depicts a mass spectrum obtained from areaction consisting of 5 mouse LD50 of BoNT B in 5�l ofserum with protease inhibitors. The peptide substrateappears at m/z 4037.4, but the toxin cleavage products arenot present at m/z 1759.5 and 2296.8. Rather, a large peak isseen at m/z 2739.0, which corresponds to protease activityon the peptide.
The extensive abundance of proteases in stool whencompared with serum indicates that stool is an even greaterchallenge. Fig. 1C is the mass spectrum of a reaction con-taining 20 mouse LD50 of BoNT B. In this spectrum, theintact peptide substrate no longer is detectable because ithas been digested by BoNT B into cleavage products, whichare apparent at m/z 1759.4 and 2296.7. In contrast, Fig. 1Dis a mass spectrum obtained from a reaction consisting of20 mouse LD50 of BoNT B in 5�l of stool extract with pro-tease inhibitors. The intact peptide substrate also is notdetectable in this reaction, but neither are the toxin cleav-age products at m/z 1759.5 and 2296.8. The peaks corre-spond to cleavage of the substrate by proteases indigenousto stool.
Endopep–MS method with spiked serum and stool extracts
Because inhibitors alone proved to be ineVective forBoNT analysis of serum and stool, we investigated the useof Ab-coated magnetic beads to concentrate and purify theBoNTs away from these protease-rich matrices. Fig. 2Adepicts the mass spectrum of 10 mouse LD50 of BoNT Aconcentrated from 100 �l of serum (100 mouse LD50/ml)using Ab-coated magnetic beads, followed by a reaction inbuVer with peptide substrate. The C-terminal cleavageproduct is present at m/z 1215.3, detecting the presence ofBoNT A in this sample. A negative control of serum with-out BoNT A did not yield a peak at m/z 1215.3(Supplementary Fig. 1A). Although the blocking and multi-ple wash steps involved in the Ab-binding process greatly
Fig. 1. Mass spectra of 20-�l Endopep–MS reactions. (A) Reaction containing 5 mouse LD50 of BoNT B with no matrix. The substrate peptide and itsdoubly charged ion are at m/z 4037.4 and 2019.6, respectively. The N-terminal cleavage product is at m/z 1759.5, and the C-terminal cleavage product is atm/z 2296.8. (B) Reaction containing 5 mouse LD50 of BoNT B with 5 �l of serum. The high peak at m/z 2739 is from other proteases in the serum. Theintensity of the substrate peptide at m/z 4037.4 also is diminished. (C) Reaction containing 20 mouse LD50 of BoNT B in reaction buVer only. Completecleavage of the substrate peptide was observed, and the only peaks in the spectrum correspond to the N-terminal cleavage product at m/z 1759.4 and theC-terminal cleavage product at m/z 2296.7. (D) Reaction containing 20 mouse LD50 of BoNT B with 5 �l of stool extract. The substrate peptide at m/z4037.4 no longer is detected, and only peaks from cleavage by other proteases were observed.
004404830823027206120061)z/m(ssaM
3.36
0010203040506070809001
% In
tens
ity
5.9571
6.9102
8.6922
4.7304
A
004404830823027206120061)z/m(ssaM
1.902
0010203040506070809001
% In
tens
ity
0.9372 4.7304
B
004404830823027206120061)z/m(ssaM
2.2736
0010203040506070809001
% In
tens
ity
4.9571
7.6922
C
004404830823027206120061)z/m(ssaM
1.131
0010203040506070809001
% In
tens
ity
4.27715.0091
6.7812
5.5081
D
88 Use of Endopep–MS for detection of botulinum toxins / S.R. Kalb et al. / Anal. Biochem. 351 (2006) 84–92
assist in decreasing the amount of proteases present in theWnal reaction, a small degree of protease binding remains.Protease inhibitors help to deter the activity of the unde-sired proteases, but at these levels they do not inhibit themcompletely. When these reactions were attempted withmuch higher levels of protease inhibitors of greater than100£, the BoNT reactions also were inhibited.
We also detected the toxin cleavage products of BoNTB, E, and F in serum samples using Ab-coated beads andprotease inhibitors. In Fig. 2B, the N-terminal toxin cleav-age product is seen at m/z 1758.5 when 0.5 mouse LD50 ofBoNT B from 100 �l of serum (5 mouse LD50/ml) is reactedwith peptide substrate. In Fig. 2C, the N-terminal toxincleavage product is seen at m/z 2922.8 when 0.1 mouse LD50of BoNT E from 100 �l of serum (1 mouse LD50/ml) isreacted with peptide substrate. Finally, in Fig. 2D, theC-terminal toxin cleavage product is seen at m/z 1345.9when 0.5 mouse LD50 of BoNT F from 100 �l of serum (5mouse LD50/ml) is reacted with peptide substrate. All of thenegative controls for BoNT B, E, and F in serum did notyield peaks corresponding to their toxin cleavage products(Supplementary Fig. 1). These spectra demonstrate thatAb-coated beads eVectively concentrate the toxin while dis-carding most of the proteases in serum.
We also applied this method to stool samples. For stoolsamples with higher concentrations of toxin, we achievedresults similar to those in serum with limits of detection(LODs) in 100�l of stool extract listed in Table 2. In allcontrol and spiked samples, stool extracts appeared to havemore nonspeciWc protease activity than serum, and lowerLODs were achieved in all cases with serum. All of the neg-
ative controls for BoNT A, B, E, and F in stool extracts didnot yield peaks corresponding to their toxin cleavage prod-ucts (data not shown).
The mouse bioassays for BoNT generally involve injec-tion of 0.5 ml of sample [1,15]. For a more direct compari-son, we analyzed larger sample sizes to enhance thesensitivity of Endopep–MS. Fig. 3 depicts the mass spec-trum obtained from the concentration of 10 mouse LD50 ofBoNT A from 500 �l of serum (20 mouse LD50/ml). Asbefore, the BoNT concentrated on the Ab-coated beadswas used in a 20-�l reaction volume containing peptide sub-strate. The C-terminal cleavage product is visible at m/z1215.3, indicating the presence of BoNT A. The resultsfound for analysis of BoNT B, E, and F in 500 �l of serumwere the same as the results for analysis of 100 �l of serum,thereby allowing us to detect Wve times lower concentrationof toxin in 500 �l of serum than in 100�l of serum (data notshown).
Table 2LODs for BoNT A, B, E, and F in mouse LD50 as determined by mousebioassay and Endopep–MS in buVer, serum, and stool
a Because the mouse bioassay is the standard detection method, theresults of the mouse bioassay deWne the LODs; as a result, all are deWnedand reported as 1 mouse LD50.
b These LODs were determined and reported in [4,5].
Toxin type
Mouse bioassaya
Endopep–MSin buVerb
Endopep–MS in serum
Endopep–MS in stool
A 1 0.01 10 100B 1 0.01 0.5 5E 1 0.08 0.1 0.5F 1 0.01 0.5 5
Fig. 2. Mass spectra of Endopep–MS reactions containing BoNTs concentrated from 100 �l of serum. (A) Reaction containing 10 mouse LD50 of BoNT A(100 mouse LD50/ml). The peak at m/z 1215.3 is the C-terminal BoNT A-dependent cleavage product. (B) Reaction containing 0.5 mouse LD50 of BoNT B(5 mouse LD50/ml). The peak at m/z 1758.5 is the N-terminal BoNT B cleavage product. (C) Reaction containing 0.1 mouse LD50 of BoNT E (1 mouseLD50/ml). The peak at m/z 2922.8 is the N-terminal BoNT E cleavage product. (D) Reaction containing 0.5 mouse LD50 of BoNT F (5 mouse LD50/ml).The peak at m/z 1345.9 is the C-terminal BoNT F cleavage product.
1205 1209 1213 1217 1221 1225Mass (m/z)
86.3
0102030405060708090
100%
Inte
nsity
1215.3A
1750 1754 1758 1762 1766 1770Mass (m/z)
49.0
0102030405060708090
100
% In
tens
ity
1753.6
1758.5
B
2915 2919 2923 2927 2931 2935Mass (m/z)
52.9
0102030405060708090
100
% In
tens
ity
2922.8
C
1335 1339 1343 1347 1351 1355Mass (m/z)
588.9
0102030405060708090
100
% In
tens
ity
1345.9D
Use of Endopep–MS for detection of botulinum toxins / S.R. Kalb et al. / Anal. Biochem. 351 (2006) 84–92 89
BoNT activity neutralization by Ab
Previous work in our laboratory on the development ofthe Endopep–MS method determined that the LOD forBoNT A was similar to the LODs for BoNT B and E. Yetthe LOD for BoNT A was much higher than expected afterAb-coated beads were used to concentrate the BoNT Afrom serum and stool samples. Abs that bind to particularregions, particularly the light chain of BoNT, may be capa-ble of neutralizing the enzymatic activity of the toxin. Poly-clonal Abs that partially inhibit the enzymatic activitywould result in a higher LOD for that particular toxin sero-type (e.g., the higher LOD that was observed for BoNT Acompared with the other BoNT serotypes). Because theEndopep–MS method detects BoNT activity, a logicalsource for the reduced activity and increased LOD was theanti-BoNT Abs.
Therefore, we tested the neutralization of BoNT by thecommercially available polyclonal Ab by comparing reac-tions with and without the polyclonal IgG for BoNT A.Reactions that contained the BoNT A polyclonal Absshowed no detectable product peptides (Fig. 4B), in con-trast to reactions that contained no Abs (Fig. 4A). To showthat these diVerences resulted from the neutralization bythe polyclonal Abs and were not due to the solution inwhich the Abs are stored, we repeated the reactions in thepresence of the polyclonal Abs to BoNT F (Fig. 4C). Noinhibition was seen in the enzymatic activity of BoNT A inthe presence of Abs to BoNT F. This indicates that thecommercially available Abs to BoNT A are partially neu-tralizing the enzymatic activity of BoNT A. In addition, theIgG to BoNT B, E, and F inhibit the corresponding toxinsto a lesser degree (data not shown).
Analysis of stool extracts from patients diagnosed with botulism
We also have applied the Endopep–MS method with Abcapture to the remains of clinical samples from people whohad been diagnosed with botulism. Some of these sampleshad tested as low positive for BoNT B by the mouse bioas-say. All identiWers were removed from the samples, six ofwhich were determined to be positive for BoNT B and three
Fig. 3. Mass spectrum of Endopep–MS reaction containing 10 mouseLD50 of BoNT A (20 mouse LD50/ml) concentrated from 500 �l of serum.
1205 1209 1213 1217 1221 1225Mass (m/z)
115.7
0102030405060708090
100%
Inte
nsity
1215.3
of which were determined to be negative by the mouse bioas-say. All nine samples consisted of 100�l of stool extract, andusing the Endopep–MS method, the six samples that previ-ously tested positive for BoNT B by the mouse bioassay alsotested positive for BoNT B. Fig. 5 contains mass spectra ofthese six positive samples. The peaks present at m/z 1759.5 inall spectra and at m/z 2296.8 in four of the six spectra demon-strate the presence of BoNT B in these samples. We presumethat the spectra in Figs. 5C and F do not contain peaks at m/z 2296.8 corresponding to the C-terminal product peptidebecause the product peptide was digested by proteases. How-ever, the presence of only one of the product peptides is suY-
cient to declare a positive result. The spectra in Figs. 5A, B,D, and E contain peaks corresponding to both product pep-tides. The three remaining samples were determined to benegative by both methods (Supplementary Fig. 2). The spec-tra obtained for these samples did not contain peaks at eitherm/z 1759.5 or 2296.8 corresponding to BoNT B cleavage ofthe peptide substrate (data not shown). Therefore, the Endo-pep–MS method successfully identiWed the correct BoNTserotype in remains of clinical stool samples.
Fig. 4. Mass spectra depicting the neutralization of BoNT A by IgG toBoNT A. (A) Reaction containing 100 mouse LD50 of BoNT A with noIgG. (B) Reaction containing 100 mouse LD50 of BoNT A with IgG toBoNT A. (C) Reaction containing 100 mouse LD50 of BoNT A with IgGto BoNT F.
1000 1440 1880 2320 2760 3200Mass (m/z)
449.4
0102030405060708090
100
2910.6
1215.3
1456.3
1714.3
1000 1440 1880 2320 2760 3200Mass (m/z)
898.7
0102030405060708090
100
% In
tens
ity%
Inte
nsity
% In
tens
ity
1456.3
2910.6
1000 1440 1880 2320 2760 3200Mass (m/z)
497.8
0102030405060708090
100
2910.6
1215.3
1456.3
1714.3
A
B
C
90 Use of Endopep–MS for detection of botulinum toxins / S.R. Kalb et al. / Anal. Biochem. 351 (2006) 84–92
Discussion
Before the current work, the Endopep–MS reaction hadbeen carried out in a matrix consisting of a reaction buVerspeciWcally formulated to promote BoNTs’ cleavage prop-erties. The reaction is very eVective in this speciWc buVer inaddition to other buVers, but determining the presence orabsence of BoNT in a matrix other than mass spectrometriccompatible buVers is important.
The enzymatic activity of proteases found in clinicalsamples ensures that the Endopep–MS method in its previ-ously reported form is challenging for several reasons. First,endogenous proteases in complex matrices may digest thepeptide substrate before reaction of the substrate withBoNT, thereby destroying the peptide substrate before theBoNT can act on it. This situation gives peptide reactionproducts in the mass spectrum that do not correspond todigestion of the peptide substrate by BoNT. This would notyield a false positive because the cleavage is not at the siteknown to be cleaved by any of the BoNTs, but it does indi-cate that the substrate peptide has been cleaved and theconcentration available for BoNT reactions would bediminished. This could be incorrectly interpreted to give afalse-negative result unless care was taken to note thereduction of the signal from the unreacted substrate
peptide. In addition, the expected product peptides them-selves may be cleaved by other proteases that could dimin-ish the sensitivity of the Endopep–MS method.
The ability to concentrate the BoNT from a clinical sam-ple using Ab-coated beads is an important advantage. TheMALDI time-of-Xight (TOF) mass spectrometer uses tinysample volumes, and the Endopep–MS method is opti-mized in a 20-�l reaction volume. The sensitivity of themethod increases if larger sample volumes of clinical sam-ples can be used to concentrate the toxin into a smallerreaction volume. The units examined in the 500-�l serumsamples are exact molar equivalents to the units examinedin the 100-�l serum samples. Within the sample volumerange of 100 to 500 �l, the toxin appears to be concentratedto be used in a 20-�l reaction so long as an adequate vol-ume of Ab-coated beads is used. This implies that at leastwithin the volume range of 100 to 500 �l, the Ab aYnityEndopep–MS method is amount (mouse LD50) dependentrather than concentration (mouse LD50/ml) dependent.Therefore, analysis of a larger sample volume (if available)is preferable, permitting concentration of a greater amountof toxin for analysis. This allows for use of smaller amountsof substrate peptides and enables analysis of a larger origi-nal sample by MALDI–TOF MS, thereby lowering theLODs. The Ab aYnity method provides the ability to
Fig. 5. Mass spectra of six clinical stool extract samples (A – F) containing BoNT B as determined by the Endopep–MS method from people who had beendiagnosed with foodborne botulism. These stool extracts previously had tested positive for BoNT B by the mouse bioassay.
004404830823027206120061)z/m(ssaM
5.2702
0010203040506070809001
% In
ten
sity
5.9571
8.6922
7.08814.7304
A
004404830823027206120061)z/m(ssaM
2.5572
0010203040506070809001
% In
ten
sity
5.9571
8.69223.7304
D
004404830823027206120061)z/m(ssaM
6.2193
0010203040506070809001
% In
ten
sity
5.9571
8.6922
3.7304
E
004404830823027206120061)z/m(ssaM
3.735
0010203040506070809001
% In
ten
sity
5.9571
6.56327.0881
7.0961
4.73048.6922
B
004404830823027206120061)z/m(ssaM
1.922
0010203040506070809001
% In
ten
sity
6.56326.0971
5.9571
7.7712
C
004404830823027206120061)z/m(ssaM
3.407
0010203040506070809001
% In
ten
sity
5.9571
6.87716.0961 6.5632
5.5072
F
Use of Endopep–MS for detection of botulinum toxins / S.R. Kalb et al. / Anal. Biochem. 351 (2006) 84–92 91
concentrate the BoNT while removing proteases present incomplex matrices.
This method can detect BoNT B, E, and F in serum atlevels similar to the level of the mouse bioassay. Eventhough the sensitivity of Endopep–MS is lower in stoolextracts than in serum, it can adequately detect BoNT inclinical stool samples. The main advantage of testing stoolsamples from patients is that stool samples tend to containhigher amounts of BoNT than do serum samples [20,21],with typical levels of 101 to 105 mouse LD50/g in the stool ofinfants [20]. Indeed, when 10 mouse LD50 of BoNT A wasspiked into 100 �l of stool extract, this reaction did notyield any evidence of BoNT A, whereas this amount oftoxin spiked into serum did yield a positive result. However,given the abundant nature of endogenous proteases presentin stool, the Abs appear to concentrate the BoNT andremove many of the proteases in stool. Viable C. botulinumoften can be cultured from stool samples, and the resultingculture supernatant can be tested for BoNT. We plan tocarefully compare spiked samples with the mouse bioassayto learn the true sensitivity of Endopep–MS in severalmatrices, including serum and stool extracts.
By incorporating immunoaYnity puriWcation of BoNTs,the Endopep–MS method now can detect as little as 20mouse LD50/ml for BoNT A in human serum. Because theIgG to BoNT A inhibits the activity of BoNT A, this cap-ture technique in its current status is not optimal for sensi-tivity. We are obtaining Abs speciWc to the heavy chain ofBoNT A that should bind the toxin without neutralizingthe enzymatic activity. Therefore, we may determine a sig-niWcantly lower LOD for BoNT A. The LODs of all fourtoxin types tested are likely to decrease when using IgGsthat bind to the toxin without inhibiting the endopeptidaseactivity of the toxin. Because the current detection limits forBoNT B, E, and F are 1, 0.2, and 1 mouse LD50/ml, respec-tively, we are optimistic that with nonneutralizing Abs wewill achieve detection limits near or below the detectionlimit of the mouse bioassay for BoNT A as well.
This analysis can be performed easily in an 8-h work day.The duration of the various procedures can be described asfollows. First, 15 min is allowed for material gathering,Ab–bead aliquoting, and adding sample to the Ab-coatedbeads. Following a 30-min incubation of the Ab-coatedbeads in casein buVer, the Ab-coated beads are incubatedwith the sample for 2 h, allowing the BoNT in the samples tobind the Abs. Then 15 min is allowed for bead washing andthe addition of the reagents (reaction buVer and peptide sub-strate) to the beads. The Endopep–MS reaction is performedfor 4 h, although with high levels of BoNT the reaction needonly last for 30 min or less, as demonstrated in [22]. Finally,30 min is allowed for sample cleanup and mass spectrometricanalysis of the sample. This totals to a maximum of 7.5 h forthe detection and type identiWcation of BoNT in a sample. Ifthe Endopep–MS reaction is performed for only 30 min, thistotals to a maximum of 4 h.
In conclusion, we have found that Ab binding beforeEndopep–MS for determination of BoNT in complex
matrices, such as serum and stool extract, is essential. Theseresults demonstrate the ability of Ab capture Endopep–MSto bind small amounts of toxin (in some cases <1 mouseLD50/ml) from serum or stool extracts and concentratethem to retain activity in a 20-�l reaction volume. The over-all method is very sensitive. The use of Abs appears to neg-atively aVect the activity of some of the toxins, particularlyBoNT A. Current work in our laboratory involving theacquisition of IgG to BoNT that eVectively binds the toxinwithout inhibiting it should dramatically improve the sensi-tivity of this method. In addition, future direction for thiswork will involve multiplexing the method. In previouspublications [4,5], we have reported the ability of the Endo-pep–MS method to be multiplexed, which involves testingone sample aliquot for four toxin types. We intend toextend multiplexing to the Ab capture step so that one clin-ical sample can be tested for four toxin types at once. Toachieve high-throughput status, we also intend to automatethis method. The bottleneck in this method is the samplebinding/matrix removal step. With automation of this stepin particular, processing hundreds of samples in an 8-hwork day might be possible, thereby increasing thethroughput of this method to be more than adequate forrapid examination of food or clinical samples.
Acknowledgments
The authors appreciate the donation of botulinum neu-rotoxin complexes from Metabiologics and the donation ofpositive and negative patient stool samples from theNational Botulism Surveillance and Reference Laboratoryat the Centers for Disease Control and Prevention,National Center for Infectious Diseases, Division of Bacte-rial and Mycotic Diseases, Foodborne and Diarrheal Dis-eases Branch.
Appendix A. Supplementary data
Supplementary data associated with this article can befound, in the online version, at doi:10.1016/j.ab.2006.01.027.
References
[1] Centers for Disease Control and Prevention, Botulism in the UnitedStates, 1899–1996: Handbook for Epidemiologists, Clinicians, andLaboratory Workers, Centers for Disease Control and Prevention,Atlanta, GA, 1998.
[2] G. Schiavo, M. Matteoli, C. Montecucco, Neurotoxins aVecting neu-roexocytosis, Physiol. Rev. 80 (2000) 717–766.
[3] S.S. Arnon, R. Schechter, T.V. Inglesby, D.A. Henderson, J.G. Bart-lett, M.S. Ascher, E. Eitzen, A.D. Fine, J. Hauer, M. Layton, S. Lilli-bridge, M.T. Osterholm, R. O’Toole, G. Parker, T.M. Perl, P.K.Russell, D.L. Swerdlow, K. Tonat, for Working Group on CivilianBiodefense, Botulinum toxin as a biological weapon: Medical andpublic health management, JAMA 285 (2001) 1059–1070.
[4] J.R. Barr, H. Moura, A.E. Boyer, A.R. WoolWtt, S.R. Kalb, A. Pavlop-oulos, L.G. McWilliams, J.G. Schmidt, R.A. Martinez, D.L. Ashley,Fast detection and diVerentiation of botulinum neurotoxins A, B, E,and F using Endopep–MS, a mass spectrometry-based method,Emerg. Infect. Dis. 11 (2005) 1578–1583.
92 Use of Endopep–MS for detection of botulinum toxins / S.R. Kalb et al. / Anal. Biochem. 351 (2006) 84–92
[5] A.E. Boyer, H. Moura, A.R. WoolWtt, S.R. Kalb, A. Pavlopoulos, L.G.McWilliams, J.G. Schmidt, J.R. Barr, From the mouse to the massspectrometer: novel detection and diVerentiation of the endoprotein-ase activities of botulinum neurotoxins A–G by mass spectrometry,Anal. Chem. 77 (2005) 3916–3924.
[6] C. Anne, F. Cornille, C. Lenoir, B.P. Roques, High-throughput Xuoro-genic assay for determination of botulinum type B neurotoxin prote-ase activity, Anal. Biochem. 291 (2001) 253–261.
[7] T.A.N. Ekong, I.M. Feavers, D. Sesardic, Recombinant SNAP-25is an eVective substrate for Clostridium botulinum type A toxinendopeptidase activity in vitro, Microbiology 143 (1997) 3337–3347.
[8] B. Hallis, B.A.F. James, C.C. Shone, Development of novel assays forbotulinum type A and B neurotoxins based on their endopeptidaseactivities, J. Clin. Microbiol. 34 (1996) 1934–1938.
[9] J.J. Schmidt, K.A. Bostian, Proteolysis of synthetic peptides by type Abotulinum neurotoxin, J. Protein Chem. 14 (1995) 703–708.
[10] J.J. Schmidt, R.G. StaVord, C.B. Millard, High-throughput assays forbotulinum neurotoxin proteolytic activity: Serotypes A, B, D, and F,Anal. Biochem. 296 (2001) 130–137.
[11] J.J. Schmidt, R.G. StaVord, Fluorigenic substrates for the proteaseactivities of botulinum neurotoxins, serotypes A, B, and F, Appl.Environ. Microbiol. 69 (2003) 297–303.
[12] C.C. Shone, A.K. Roberts, Peptide substrate speciWcity and propertiesof the zinc-endopeptidase activity of botulinum type B neurotoxin,Eur. J. Biochem. 228 (1994) 263–270.
[13] M. Wictome, K. Newton, K. Jameson, B. Hallis, P. Dunnigan, E. Mac-kay, S. Clarke, R. Taylor, J. Gaze, K. Foster, C. Shone, Developmentof an in vitro bioassay for Clostridium botulinum type B neurotoxin infoods that is more sensitive than the mouse bioassay, Appl. Environ.Microbiol. 65 (1999) 3787–3792.
[14] M. Wictome, K.A. Newton, K. Jameson, P. Dunnigan, S. Clarke, J.Gaze, A. Tauk, K.A. Foster, C.C. Shone, Development of in vitroassays for the detection of botulinum toxins in foods, FEMS Immu-nol. Med. Microbiol. 24 (1999) 319–323.
[15] D.A. Kautter, H.M. Solomon, Collaborative study of a method forthe detection of Clostridium botulinum and its toxins in foods, J.Assoc. Anal. Chem. 60 (1977) 541–545.
[16] J. Ferreira, S. Maslanka, E. Johnson, M. Goodnough, Detection ofbotulinal neurotoxins A, B, E, and F by ampliWed enzyme-linkedimmunosorbent assay: Collaborative study, J. AOAC Intl. 86 (2003)314–331.
[17] J.L. Ferreira, S.J. Eliasberg, P. Edmonds, M.A. Harrison, Comparisonof the mouse bioassay and enzyme-linked immunosorbent assay pro-cedures for the detection of type A botulinal toxin in food, J. FoodProtection 67 (2004) 203–206.
[18] AOAC International, OYcial Methods of Analysis, 17th ed., AOACInternational, Gaithersburg, MD, 2000.
[19] S. Notermans, J. Nagel, Assays for botulinum and tetanus toxins, in:L.L. Simpson (Ed.), Botulinum Neurotoxin and Tetanus Toxin, Aca-demic Press, San Diego, 1989, pp. 319–331.
[20] C.L. Hatheway, L.M. McCroskey, Examination of feces and serumfor diagnosis of infant botulism in 336 patients, J. Clin. Microbiol. 25(1987) 2334–2338.
[21] J.C. Paton, A.J. Lawrence, I.M. Steven, Quantities of Clostridium bot-ulinum organisms and toxin in feces and presence of Clostridium botu-linum toxin in the serum of an infant with botulism, J. Clin. Microbiol.17 (1983) 13–15.
[22] S.R. Kalb, M.C. Goodnough, C.J. Malizio, J.L. Pirkle, J.R. Barr,Detection of botulinum neurotoxin A in a spiked milk sample withsubtype identiWcation through toxin proteomics, Anal. Chem. 77(2005) 6140–6146.
