All chemicals were obtained from Sigma–Aldrich (St. Louis, MO) except where otherwise indicated. Monoclonal antibodies were provided by Dr. James Marks at the University of California, San Francisco. Streptavidin-coupled Dynabeads were purchased from Invitrogen (Lake Success, NY). Botulinum neurotoxin complexes were obtained from Metabiologics (Madison, WI). Crude culture supernatants representing bivalent neurotoxin-secreting strains were produced by incubating subcultures of each strain for 5 days at 30 or 37°C. After centrifugation, supernatants were removed and filtered through 0.22 μm filters. The filtered supernatants were tested for upper limits of toxicity using the mouse bioassay, which indicated that the toxins were all present at concentrations of ≤ 10 μg/mL. Botulinum neurotoxins are highly toxic and appropriate safety measures are required. All BoNT neurotoxins were handled in a class 2 biosafety cabinet equipped with HEPA filters. Fmoc-amino acid derivatives and peptide synthesis reagents were purchased from EMD Chemicals, Inc. (Gibbstown, NJ) or Protein Technologies (Tucson, AZ). Peptides were purchased from Midwest Biotech (Fishers, IN) or prepared in house by a solid phase peptide synthesis method using Fmoc chemistry. The peptide substrate for BoNT/E was recently developed in our laboratory and its sequence will be published at a later date.The sequences of the peptide substrates for BoNT/A, /B, or /F, corresponding cleavage products and their internal standards are listed in the supplementary Table S1.

In-solution or on-bead endopeptidase activity assays were carried out as previously described²⁸. In brief, the reaction was conducted in a 20 μL reaction volume containing toxin, peptide substrate, 10 μM ZnCl₂, 1 mg/mL BSA, 10 mM dithiothreitol, and 200 mM HEPES buffer (pH 7.4) at 37°C for various lengths of time as indicated in the text. Individual peptides or a mixture of the peptide substrates were added to the reaction at the indicated concentration(s). For the in-solution assays the antibody-coated bead incubations were skipped and toxin (BoNT/A, /B, /E, or /F) was added directly into the reaction mixture. For the on-bead assay, the toxin was spiked or was present in buffer, cell culture, serum or stool extract and was incubated with antibodies immobilized on Streptavidin-coupled magnetic beads prior to the activity assessments. After washing, the beads were immersed into the 20 μL reaction solution for the assay as described above. Two microliters of the reaction were removed at various time points (0.5 to 4 hr) and were mixed with 2 μL of an internal standard peptide and 16 μL of α-cyano-4-hydroxy cinnamic acid (CHCA, 5 mg/mL in 50% acetonitrile/0.1% TFA/1 mM ammonium citrate). This mixture (0.7 μL) was then applied onto a MALDI plate and subjected to mass spectrometric analysis. The relative production of the cleavage fragment from any peptide substrate was measured as the ratio of the isotope cluster areas of the MS peak of the N- or C-terminal product versus its internal standard (Table S1). Triplicate samples were spotted on a MALDI plate and analyzed on a 4800 or 5800 MALDI-TOF/TOF instrument (Applied Biosystems, Framingham, MA). Mass spectra of each sample were obtained by scanning from 800 to 3000 m/z in MS-positive ion reflector mode. This instrument uses a Nd-YAG laser at 355 nm, and each spectrum is an average of 2400 laser shots. The data were usually an average of three experiments. CV values for the samples were typically below 20%.

Samples were split into two equal volumes and subjected to the singleplex assay for BoNT/A and the multiplex assay for BoNT/B, /E, or F. The samples and calibrants of BoNT/A, /B, /E and /F having various toxin concentrations (activities) were captured on toxin-specific antibody-bound beads (20 μL bead suspension) followed by washing and immersion of toxin-bound beads into 20 μL reaction solutions which were incubated at 37°C for various time periods. The singleplex reaction solution contained 100 μM of the substrate of BoNT/A (SA) whereas the substrate mixture containing 100 μM SB, 50 μM SE and 100 μM SF was included in multiplex reaction solution. In order to avoid the appearance of premature cleavage reactions, beads and reaction solutions were quickly mixed on ice and the mixtures were then placed at 37° C. After a 30 min incubation period, the initial analysis was conducted by transferring 2 μL aliquots of each reaction solution into a new tube containing 16 μL of CHCA matrix and 2 μL of single or multiple internal standard. At this stage, the internal standards of high concentration were used including 5 μM IS-A (internal standard for BoNT/A cleavage product) for the singleplex assay and a mixture of 5 μM IS-B (for BoNT/B), 2 μM IS-E (for BoNT/E) and 5 μM IS-F (for BoNT/F) (Table S1) for the multiplex assay, respectively. If botulinum neurotoxin(s) were detected and quantitated at this stage, no further assessments were done. Otherwise, the remaining reaction solutions continued their incubation at 37°C until the 4 hour time point was reached and 2-μL aliquots were subjected to the second-stage analysis. The optimized concentration of the internal standards at this stage included 0.5 μM IS-A and the mixture of 1 μM IS-B, 0.4 μM IS-E, and 1 μM IS-F for singleplex and multiplex assay, respectively.

In a previous study for the quantitative detection of BoNT/A and /B activities using ESI mass spectrometry ²⁹, it was observed that a nonlinear response occurs when the toxin concentration reached a certain point, e.g., 100 mouse LD₅₀ (mouse lethal dose determined by standard lethality bioassay) for BoNT/A. This could be attributed to a reduced ratio of the substrate versus toxin, where a hydrolysis reaction no longer conforms to the Michaelis-Menten kinetics, and the formation of cleavage products no longer correlates linearly with BoNT concentration. A possible simple solution is to reduce the reaction time but leave the other parameters unchanged so that the appropriate relative amounts of enzyme and substrate can be maintained. To examine how the time of a BoNT hydrolysis reaction affects the linear dynamic range of BoNT quantification, we conducted toxin cleavage experiments at four different reaction times. Each of the assays was comprised of cleavage reactions with various BoNT/A concentrations ranging from 0.1 to 1000 mouse LD₅₀ where the generation of the C-terminal product (CT-A) formed from the hydrolysis of the BoNT/A substrate (SA) was monitored by MALDI-TOF mass spectrometry. We chose MALDI-TOF rather than LC-ESI mass spectrometry in the present study due to its high throughput capability and significantly reduced sample analysis times compared to the latter. The detection of cleavage products from the reactions with various amounts of BoNT/A at different reaction times was displayed on a logarithmic scale (Figure 1A). Obvious differences were observed among the four data curves obtained from the assays terminated at 0.5, 1, 2, or 4 hours. The reactions, when stopped at 30 minutes, resulted in an apparent linear response in the range of 1 to 1000 -LD₅₀ of BoNT/A toxin. When the reaction times were increased to one hour and higher, some data points at the high toxin concentrations did not fit a linear model, and the curve reached a plateau at the higher concentrations. The apparent bend point (about 100 mouse LD₅₀) for a 4 hour incubation is consistent with the outcome described in a previous ESI-MS study ²⁹. Longer reaction times are important for the analysis, however, because of the improved signal to noise at low toxin concentrations (< 1 - LD₅₀) and the improved detection limits. Figure 1B depicts a similar pattern observed in the assay where the toxin was initially enriched and purified with antibody immobilized on magnetic beads followed by on-bead incubation, which is the Endopep-MS assay platform for the analysis of clinical samples. These results prompted a two-stage approach, in which the aliquots of the cleavage reaction can be analyzed at two different reaction times. For instance, two linear calibration curves could be produced by extracting data points in the above experiments from either the 30 min(1 – 1000 -LD₅₀) or 4 hour reactions (0.1 – 100 LD₅₀) (Figure 1C). By combining these two curves, the amount of BoNT toxin ranging from 0.1 to 1000 LD₅₀ was quantitatively determined. Therefore, it is possible to extend the dynamic range of BoNT detection by preparing two or more calibration curves from measuring different time points in a single experiment. In comparison to a conventional method with serial dilutions of unknown samples to fit one standard curve, one of the potential advantages of a two-stage assay could be to maintain original concentration of an analyte and save precious samples, particularly those containing very low abundant analyte.

Two factors should be considered in isotope dilution mass spectrometric (IDMS) quantification analyses such as the quantitative Endopep-MS assay: the amount of internal standard (IS) used in the analysis solution after toxin cleavage and the original concentration of the peptide substrate added to the hydrolysis reaction. In IDMS analysis, the relative or absolute quantity of a peptide is determined by measuring the peak ratio of the peptide and its isotopically labeled internal standard spiked into samples at a fixed concentration. It has been reported that a large difference between the amounts of unlabeled and isotopically labeled peptides affects the precision of a MALDI response peak ratio as well as the linearity of a standard curve ³⁰. If a target peptide or its internal standard is much less abundant than the other, the less abundant one is likely to have a very weak signal or its MS peak could embed into the noise level in a MS spectrum, producing inaccurate or incorrect analyte/standard ratios. Use of less IS, for example, may improve assay sensitivity because the ratio of the cleavage product versus internal standard would not become too great to be accurately determined at very low BoNT quantity. On the other hand, less IS usage may affect accurate determination of high toxin concentration, because the very intense signal of a cleavage product peak may result in an undetectable IS peak. In contrast, more IS may improve the linearity on high toxin concentrations but could adversely reduce assay sensitivity. Therefore, it is necessary to find a balance of the proper concentration of a labeled internal standard in a BoNT quantitative assay. A similar principle can be applied to peptide substrates added in an activity assay as described below.

To determine an optimal amount of an internal standard, we prepared a series of mixtures of a reaction solution, IS and MALDI matrix, to mimic the analysis solution. Various amounts of a synthetic peptide (CT-A, Table S1) derived from the C-terminal cleavage product of BoNT/A were added in the mock reaction solution, whereas a fixed amount of the internal standard (IS-A, Table S1) corresponding to CT-A was added to the analysis solution. Figure 2A shows measured ratios of CT-A and IS-A peak areas as a function of varied CT-A concentration at three fixed IS-A concentrations (1, 10 or 100 μM). It was observed that the data points generated with 1 μM and 10 μM IS-A resulted in measured peak ratios linearly correlated with CT-A concentration while the peak ratios using 100 μM IS-A did not form a linear curve within the range of CT-A products tested. In the latter case, CT-A/IS-A ratio remained almost unchanged when the amount of CT-A become low, indicating the product/IS ratio could not be accurately determined at very low CT-A concentration (below 1 μM), due to very weak CT-A peaks embedded in the noise of the spectra as discussed previously. To achieve a desired level for detection sensitivity and dynamic range, we decided to use different IS amounts in two analysis stages, where less IS used for a short reaction time and more IS for the long reaction time. This did not add an extra step or interfere in the flow of an experiment because the IS was prepared at individual cleavage reaction time points. The optimal IS concentrations were determined for each of the two stages by the analysis of four human-botulism related serotypes of BoNTs (data not shown) and used in the experiments described. For instance, 5 μM of IS-A was applied for the initial, half-hour, stage analysis of BoNT/A and 0.5 μM of IS-A for the second, four-hour, stage.

Similarly, the amount of peptide substrate added in a cleavage reaction can affect the quantitative detection of BoNT toxins by MALDI analysis. Unlike LC-ESI-MS where peptide substrate and cleavage products can be chromatographically separated prior to mass spectrometry, all components in a cleavage reaction, together with an internal standard, are present in a single spot and analyzed simultaneously in a MALDI method. Therefore, the peak intensity of an un-cleaved substrate may impact the detectability of cleavage product and internal standard. As discussed, relatively low concentrations of a peptide substrate in a BoNT assay could give rise to a nonlinear response for samples at high toxin levels. This limitation could be overcome by using more peptide substrate, but a higher amount of substrate could adversely affect the detection of very low amounts of cleavage products in the same way as internal standard versus cleavage product. In addition, the ions of larger amounts of the unreacted peptides in a sample may saturate the MS detector and negatively impact the product/standard ratio. Figure 2B displays the effect of three concentrations (50, 100, and 200 μM) of the peptide substrate (SA) in the activity assays for the detection of BoNT/A toxin. The response of the cleavage product appeared linear in all three sets of experimental data when the BoNT concentrations were lower than 2000 mLD₅₀ (logarithmic value: 3.30). However, significant deviations in product formation were observed when a higher amount of the toxin was present and 50 μM or 100 μM SA was used. In contrast, linear responses could still be achieved at these high toxin levels through the use of 200 μM SA in the reactions. On the other hand, use of high substrate amounts produced a higher limit of detection than with relatively low added peptide substrate (data not shown). Unlike internal standards which can be added at a different concentration for the two individual analytical stages as previously described, it is practical to use a single concentration of substrate in both low- and high-stages for preparing a calibration curve and measuring unknown samples. Thus, the concentrations of individual peptide substrates were optimized for the analysis of each of the four BoNT serotypes tested based on a working quantitation range, and optimal substrate concentrations (100 μM SA, 100 μM SB, 50 μM SE, and 100 μM SF) were used in the experiments described below.

In the current Endopep-MS assay platform, an unknown toxin sample is usually divided into four equal parts and then analyzed in separate activity assays with an single peptide substrate specific for each BoNT serotype A, B, E, or F, in each reaction solution. In other words, four experiments would be performed for any single sample. In order to save valuable samples and to reduce reagent usage and lab efforts needed for preparing calibration curves and sample analysis, we examined the feasibility of developing a multiplex analysis procedure, where a mixture of multiple substrates could be used in a single reaction. For this purpose, combinations of two, three or four peptide substrates for the four BoNT serotypes were mixed in a single reaction, and the cleavage of a serotype-specific substrate in any of the mixtures by an individual BoNT serotype was monitored. Figure 3 shows the effect of a substrate mixture on the detection of the cleavage products of a target substrate, in comparison with the results from the single-substrate reactions.

For the hydrolysis of the BoNT/A substrate (SA), none of the combinations generated more cleavage product (CT-A) than that obtained in the reaction consisting of SA alone (Fig. 3A). While three substrate mixtures yielded 40 – 80% product relative to that of the single substrate, the other four combinations produced less than 10% product in comparison to the single substrate system. Several factors could have contributed to this lower intensity of the CT-A product in these experiments. First, the cleavage efficiency for SA might be decreased by the presence of other BoNT substrate(s) (SB, SE, and SF for BoNT/B, /E, and /F, respectively, Table S1); second, the mass spectrometry signal of CT-A and/or IS-A could be suppressed by the other peptide substrates included in the reactions and/or other foreign internal standards present in the analysis solutions. It was observed that all four substrate mixtures which produced less than 10% product detection included SE, a common foreign component in addition to SA, whereas three other combinations without SE yielded relatively high amount of BoNT/A cleavage product. This was consistent with results of a SE titration experiment in a BoNT/A detection assay where the increasing concentrations of SE caused a decrease in CT-A formation (Fig. 3E), indicating that SE inhibits the cleavage of SA. MS signal suppression of SE on the peak of CT-A and/or IS-A could be ruled out because the relative ratio of IS-A versus other internal standards in subsequent assays involving substrate mixtures remained unchanged (data not shown). This inhibition effect might be attributed to the fact that both peptide substrates, SA and SE, were developed using sequences from the same native protein substrate, SNAP-25, where the cleavage sites for BoNT/A and BoNT/E are only 17 amino acid residues apart. The entire sequence or a partial region of the SE peptide may interfere with the binding of SA to the catalytic domain of the BoNT protein and, hence, reduce its cleavage efficiency. In conclusion, these results revealed that the presence of SE in a substrate mixture negatively impacted the cleavage of the substrate SA; therefore, it is not appropriate to include BoNT/E substrate in the assay for the detection of type A botulinum neurotoxin. On the other hand, it seems not practical to use a single peptide substrate that contains both cleavage sites for BoNT/A and /E because such peptides generated lower sensitivity for both serotypes than the separate, serotype-specific, peptides (SA and SE) did (data not shown). For this reason, we decided to still use two different peptide substrates to achieve the best assay sensitivity for the detection of these two serotypes.

In contrast to BoNT/A, detection of the three other types of BoNTs was not reduced by the presence of other peptide substrates in the mixtures (Figure 3B, 3C, and 3D). In some combinations, the peak intensity of a specific cleavage product was even enhanced. It was interesting to observe that the presence of SB, SE and SF as a mixture in any BoNT type-specific assay either retained or improved the detection of the target products. This result demonstrated that it is viable to include the mixture of the substrates for BoNT/B, /E, /F, but not for BoNT/A, in a single cleavage reaction for the detection of an unknown BoNT sample. In other words, we are able to reduce the number of experiments from four individual assays for any one sample as established in the current Endopep-MS assay to two assays - one individual BoNT/A assay and a multiplexed BoNT/B, /E, and /F assay.

On the basis of the results described above, a desired workflow for a two-stage multiplex MALDI quantification of multiple BoNT toxins was developed as shown in Figure 4. In this method, a clinical sample can be split for the detection and quantification of BoNT/A in one set and BoNT/B, /E, and /F in another set. Accurate calibration curves can be generated in separate reaction samples including SA alone and a mixture of SB, SE, and SF. The use of two rather than four samples for each evaluation reduces the use of reagents and can enable higher sensitivity by testing larger sample volumes. After the initial 30 min reaction, aliquots of sample and calibrants are withdrawn, mixed with internal standards and MALDI matrix, and subjected to MS analysis. If toxin is detected and quantified at this stage, the assay is stopped, avoiding the need for additional incubation time and reagents. This further improves assay efficiency and significantly decreases the time required for reporting of results. This is particularly important when responding to an outbreak, where every minute may be important for treatment. If toxin is not detectable after 30 minutes, it will be tested after a 4 hour reaction time. This prolonged assay time further improves toxin detection sensitivity while providing results within an eight hour timeframe. Standard curves were prepared in two stages with appropriate concentrations of toxin calibrants such that overlap of calibrant concentrations are included in both calibration curves. This provides consistency of results between the two reaction stages.

Botulinum neurotoxins usually exist in complex biological matrices. Serum and stool are two of the most common clinical specimens collected from botulism patients for disease diagnosis. To evaluate the two-stage method for toxins present in biological matrices, we examined the dynamic range for detection of BoNTs spiked in serum and stool. Figure 5 shows two typical calibration curves for the quantitative detection of a range of BoNT/A concentrations spiked in serum, obtained from either the initial (0.5 hr) or second stage (4 hr) analyses. A linear curve from the 4 hour reactions was achieved with a dynamic range from 0.05 to 50 mouse LD₅₀ in 0.5 mL matrix, while analysis of the 30 minute reactions created a linear curve ranging from 10 to 3000 mouse LD₅₀, demonstrating that 0.05 - 3000 -LD₅₀ BoNT/A in serum matrix can be effectively and quantitatively measured using this two-stage strategy. The amounts of toxin in four QC samples in the two stages were accurately measured with a moderate error range (Table 1). Table 2 summarizes the dynamic ranges of the quantitative detection of botulinum neurotoxins A, B, E and F in either serum or stool matrices. Although the values of the maximum and minimum linear detection limitation differ among the four BoNTs and sometimes between the two matrices, four orders of magnitude in dynamic range were achieved for the quantitation of all four types of botulinum neurotoxins in both biological matrices. By combining the results of the two calibration curves from the initial and second steps, the linear ranges for BoNT/A and BoNT/B were determined to be 0.1 to 6000 and 0.1 to 3000 mouseLD₅₀/mL, respectively. This represents a significant improvement in comparison with the results determined by a previous one-stage (4 hr) method, which yielded the linear ranges of 0.3 to 180 for BoNT/A and 0.3 to 64 mLD₅₀/mL for BoNT/B ²⁹.

This novel two-stage method was applied to a study of temperature-induced change in the BoNT toxin activity of a bivalent bacterial strain. It is known that the production of the toxin proteins in some bivalent strains that express two types of botulinum neurotoxins may be affected by growth temperature during inoculation. The level of the activity of the two toxins produced in a bivalent strain secreting both BoNT/A and BoNT/B (Ba) was evaluated under different temperature conditions. The bacteria were cultured at two temperatures (30°C and 37°C), and BoNT/A and BoNT/B activities were analyzed using this two-stage quantitative Endopep-MS method. As shown in Table 3, the activity of the two different BoNT toxins was successfully determined. The amount of BoNT/A in the samples was too low to be detected after the initial reaction, but was detected and quantified after the 4 hour cleavage reaction. In contrast, the activity levels of BoNT/B were very high and could only be quantitatively determined from the initial stage analysis. This study shows that the production of both BoNT/A and BoNT/B toxins are affected by temperature, with an enhancement of toxin production occurring at 37° C. However, the relative amounts (ratio) of BoNT/A versus BoNT/B remained constant at ~2500:1 B:a. This analysis further demonstrates the feasibility and usefulness of this two-stage quantitative method.