|Home | About | Journals | Submit | Contact Us | Français|
Botulinum neurotoxins are produced by the anaerobic bacterium Clostridium botulinum and are divided into seven distinct serotypes (A to G) known to cause botulism in animals and humans. In this study, a multiplexed quantitative real-time PCR assay for the simultaneous detection of the human pathogenic C. botulinum serotypes A, B, E, and F was developed. Based on the TaqMan chemistry, we used five individual primer-probe sets within one PCR, combining both minor groove binder- and locked nucleic acid-containing probes. Each hydrolysis probe was individually labeled with distinguishable fluorochromes, thus enabling discrimination between the serotypes A, B, E, and F. To avoid false-negative results, we designed an internal amplification control, which was simultaneously amplified with the four target genes, thus yielding a pentaplexed PCR approach with 95% detection probabilities between 7 and 287 genome equivalents per PCR. In addition, we developed six individual singleplex real-time PCR assays based on the TaqMan chemistry for the detection of the C. botulinum serotypes A, B, C, D, E, and F. Upon analysis of 42 C. botulinum and 57 non-C. botulinum strains, the singleplex and multiplex PCR assays showed an excellent specificity. Using spiked food samples we were able to detect between 103 and 105 CFU/ml, respectively. Furthermore, we were able to detect C. botulinum in samples from several cases of botulism in Germany. Overall, the pentaplexed assay showed high sensitivity and specificity and allowed for the simultaneous screening and differentiation of specimens for C. botulinum A, B, E, and F.
Botulinum neurotoxins (BoNTs), the causative agents of botulism, are produced by the anaerobic bacterium Clostridium botulinum and are divided into seven serotypes, A to G. While the botulinum neurotoxins BoNT/A, BoNT/B, BoNT/E, and BoNT/F are known to cause botulism in humans, BoNT/C and BoNT/D are frequently associated with botulism in cattle and birds. Despite its toxicity, BoNT/G has not yet been linked to naturally occurring botulism (26).
Botulism is a life-threatening illness caused by food contaminated with BoNT (food-borne botulism), by the uptake and growth of C. botulinum in wounds (wound botulism), or by colonization of the intestinal tract (infant botulism) (14). In addition, C. botulinum and the botulinum neurotoxins are regarded as potential biological warfare agents (8).
The gold standard for the detection of BoNTs from food or clinical samples is still the mouse lethality assay, which is highly sensitive but rather time-consuming. In addition to various immunological assays for BoNT detection, several conventional and real-time PCR-based assays for the individual detection of bont genes have been reported (2, 9-12, 15, 20, 23, 27-30). A major improvement is the simultaneous detection of more than one serotype, which results in a reduction of effort and in the materials used. In recent years, both conventional and real-time PCR-based multiplex assays have been developed for the simultaneous detection of C. botulinum serotypes (1, 6, 22, 24). To date, however, no internally controlled multiplex real-time PCR assay for the simultaneous detection and differentiation of all four serotypes relevant for humans has been reported.
We describe here a highly specific and sensitive multiplex real-time PCR assay based on the 5′-nuclease TaqMan chemistry (17) for the simultaneous detection of the C. botulinum types A, B, E, and F, including an internal amplification control (IAC). Furthermore, we developed six different singleplex assays based on the TaqMan chemistry for the detection of C. botulinum serotypes A to F. Assays were validated on 42 C. botulinum strains, 57 non-C. botulinum strains, on spiked food samples, and on real samples from cases of botulism in Germany.
The bacterial strains used in the present study are listed in Table Table1.1. Clostridial strains were cultured in reinforced clostridia medium (RCM; Sifin, Berlin, Germany) or in tryptone-peptone-glucose-yeast (TPGY) broth for 3 days in an anaerobic workstation (Don Whitley Scientific, Ltd., West Yorkshire, United Kingdom). The titer of the C. botulinum strains 2292 (serotype A), 1029 (serotype B), 1032 (serotype E), and 1033 (serotype F) was determined on blood agar plates. One milliliter of 10-fold dilutions of the cultures was spread on blood agar plates, and colonies were counted after 24 h of incubation, under anaerobic conditions. Bacteria were stored at −20°C in RCM or TPGY broth until use.
The primers and probes used here are given in Table Table2.2. Primers and probes were based on the published DNA sequences from GenBank database (http://www.ncbi.nih.gov/GenBank/) for the neurotoxin genes bont/a, bont/b, bont/c, bont/d, bont/e, and bont/f (Table (Table2).2). All primers and LNA probes were obtained from TIB Molbiol (Berlin, Germany) or Sigma-Aldrich (Munich, Germany); MGB probes were obtained from Applied Biosystems (Foster City, CA).
As positive controls and for evaluation of the real-time PCR assays, plasmids containing the PCR target regions pBoNT/A, pBoNT/B, pBoNT/C, pBoNT/D, pBoNT/E, and pBoNT/F were constructed. The amplicon for each real-time PCR was amplified by conventional PCR using the same primers as for the TaqMan PCR (Table (Table2).2). Amplicons were cloned into pCR2.1-TOPO vectors and used for the transformation of TOP10 cells (both from Invitrogen, Karlsruhe, Germany), according to the manufacturer's instructions. Serial dilutions of the plasmids were prepared in H2O (Roth, Karlsruhe, Germany) supplemented with 100 μg of salmon-sperm DNA (AppliChem, Darmstadt, Germany)/ml and stored at −20°C until use.
The plasmid pKoMa2 was used as an IAC for the real-time PCRs. The IAC consists of a chemically synthesized DNA sequence (Table (Table2)2) containing the target sequence for the primers KoMa_F and KoMa_R and the probes KoMa_TM or KoMa_TS (Table (Table2)2) cloned into the pPCR-Script vector (Stratagene, La Jolla, CA).
All real-time PCRs were performed on the LightCycler 480 II system in a total volume of 20 μl using a white LightCycler 480 Multiwell Plate 96 covered with adhesive sealing foil (Roche, Mannheim, Germany). Thermal cycling was done with a two-step PCR protocol: activation of the Taq DNA polymerase at 95°C for 15 min, followed by 45 cycles of 95°C for 15 s and 60°C for 40 s. The fluorescence data were collected at the end of every 60°C step, and runs were analyzed using LightCycler 480 software (Roche). All real-time PCRs were performed as duplicates. Primer and probes used for singleplex and multiplex PCR are listed in Table Table2.2. (i) The multiplex reaction mix contained 2 μl of template; 10 μl of 2× ABsolute QPCR mix (ABgene, Epsom, United Kingdom); and optimized concentrations of primer and probes for C. botulinum serotypes A, B, E, and F and KoMa2 (Table (Table2).2). Also, 75 copies of pKoMa2, as an IAC, were added to a total volume of 20 μl of H2O. (ii) The singleplex reactions contained 2 μl of template, 10 μl of 2× ABsolute QPCR mix, optimized concentrations of primer and probes for C. botulinum serotypes A to F (Table (Table2),2), and H2O added to a final volume of 20 μl.
To determine the performance of the different real-time PCR assays, standard curves were obtained by amplification of 10-fold dilutions of the standard plasmids. The obtained threshold cycles (CT) were plotted against the logarithm of copy number. Analyses were done using the Prism 5.0 Software (GraphPad, La Jolla, CA). The PCR efficiency (E) of the multiplex and singleplex assays was calculated according to the formula E = 10(−1/slope) − 1 (21).
The detection probability for the standard plasmids was determined by multiplex and singleplex PCR, analyzing serial dilutions of the standard plasmids pBoNT/A, pBoNT/B, pBoNT/E, or pBoNT/F. The 95% detection probability was calculated with Prism 5.0, using the cumulative Gaussian distribution model (Table (Table33).
Food samples (vacuum-packed, gilled, smoked mackerel; frozen green beans; vacuum-packed, smoked black pudding with meat) used in the present study were purchased from a local retail shop and stored at −20°C until use. Food samples were homogenized in stomacher bags with an equal volume of gelatin-phosphate buffer (0.2% [wt/vol] gelatin in 28 mM Na2HPO4 [pH 6.2]) using a BagMixer (Interscience, Saint Nom la Bretèche, France). Each homogenate was spiked with 106, 105, 104, or 103 CFU of C. botulinum serotype A (strain 2292), C. botulinum serotype B (strain 1029), C. botulinum serotype E (strain 1032), or C. botulinum serotype F (strain 1033)/ml, respectively.
Genomic DNA from bacterial culture was purified with the DNeasy blood and tissue kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. DNA extraction from spiked food samples (Table (Table4)4) or real botulism specimens (Table (Table5)5) was performed with the DNeasy blood and tissue kit using a modified protocol. After centrifugation (10 min, 7,500 × g) of 200 μl of the spiked food slurry, the sediment was dissolved in 180 μl of lysis buffer (20 mM Tris-HCl [pH 8.0], 2 mM EDTA, 1.2% [vol/vol] Triton X-100, 20 mg of lysozyme/ml) and lysed for 30 min at 37°C. After the addition of 25 μl of proteinase K (600 mAU/ml; Qiagen) and 180 μl of ATL buffer (Qiagen), the lysate was incubated for 30 min at 56°C, followed by 15 min at 95°C. The following steps were performed according to the manufacturer's instructions for the purification of DNA from animal tissue. The purified template DNA was eluted twice with 50 μl of AE buffer (Qiagen).
Suspected samples of three cases of human food-borne botulism (cases 1, 2, and 3), one case of infant botulism (case 4), and two cases of animal botulism, one in cattle (case 5) and one in ducks (case 6), were analyzed by TaqMan PCR (Table (Table5).5). The specimens were collected between 2006 and 2009 in Germany. Suspected specimens were transmitted to our laboratory by the relevant German state institutions for food, drug and epizooties, namely, the Landeslabor Brandenburg (case 1), the Landesamt für Verbraucherschutz Sachsen-Anhalt (case 3) or the Institut für Lebensmittel, Arzneimittel, und Tierseuchen in Berlin (cases 2, 5, and 6), respectively. Samples from case 4 originated from the Children's Hospital of the Charité (Berlin, Germany). For enrichment cultures, 5 ml of TPGY broth and/or cooked meat medium (Oxoid, Hampshire, United Kingdom) were inoculated with 500 μl of homogenized samples and incubated anaerobically at 28°C and/or 37°C for 3 to 5 days. After DNA extraction, real-time multiplex and singleplex assays were performed in two independent experiments as described above.
Starting with singleplex real-time PCR approaches, we developed individual assays for the amplification of bont/a, bont/b, bont/c, bont/d, bont/e, and bont/f using MGB containing oligoprobes (Table (Table2).2). Amplification performances of singleplex real-time PCR were determined from the standard curves of the BoNT standard plasmids. As shown in Fig. Fig.1,1, all real-time PCR assays showed a strong linear correlation (R2 > 0.99) between the CT value and the template concentration over a range of 5 orders of magnitude.
In order to develop a multiplex assay for the simultaneous amplification of bont/a, bont/b, bont/e, and bont/f, together with an IAC, it was necessary, for some of the target genes, to select TaqMan probes different from that of the singleplex assays. It was also necessary to include both MGB oligonucleotides and LNA oligonucleotides in order to increase the annealing temperature and to shorten the probe length (Table (Table2).2). The oligoprobes were labeled with different reporter dyes (Cy5, VIC, FAM, TEX, and Cyan500) and were cross validated with the respective forward and reverse primers in a multiplex approach (data not shown). For each bont amplicon, the oligoprobes were tested either coupled to an MGB moiety or modified with LNA bases, and additionally with one of the four different colors Cy5, VIC, FAM, or TEX. The optimal MGB-containing or LNA-containing oligoprobes were selected for each bont amplicon and finally, the quadruplex assay for bont/a, bont/b, bont/e, and bont/f was combined with the IAC assay (i.e., the presence of 75 copies of the synthetic sequence IAC, plasmid pKoMa2). The optimal primer-probe sets for the final pentaplexed approach are given in Table Table2.2. As shown in Fig. Fig.11 for the simultaneous amplification of bont/a, bont/b, bont/e, and bont/f, together with IAC, the performances of the multiplex assay and the respective singleplex assays were almost identical. The slopes for both the multiplex and the singleplex assays ranged from −3.889 to −3.243, resulting in PCR efficiencies of 97.9% versus 90.3%, 96.2% versus 92.9%, 103.4% versus 93.8%, and 90.6% versus 88.6%, respectively, for the multiplex versus singleplex assays for bont/a, bont/b, bont/e, and bont/f. In addition, the fluorescence spectra of the different reporter dyes used in the multiplex assay were well separated and showed no spectral overlap (data not shown).
The specificity of primers and probes was evaluated in silico against the published sequences from the GenBank database using the BLAST algorithm. Primer and probe sets were selected to allow amplification of all subtypes of serotypes known thus far: A1, A2, A3, A4, and A5; B1, B2, B3, nonproteolytic B, and bivalent B; E1, E2, E3, E4, E5, and E6; F baratii, F proteolytic, F nonproteolytic, and F bivalent; and C and D, including C/D and D/C mosaic types. To confirm the in silico results, we analyzed 42 C. botulinum strains and 57 non-C. botulinum strains with the singleplex and multiplex real-time PCR assays (Table (Table1).1). The multiplex real-time PCR for the simultaneous detection of bont/a, bont/b, bont/e, and bont/f showed a high specificity for the predicted target. A total of 4 of the 42 tested C. botulinum strains (REB1750, NCTC 2916, NCTC 11199, and NCTC 12265) tested positive for two BoNT genes. As expected, the nontoxic (as determined by mouse assay) C. botulinum A strain 2276 and all of the C. botulinum C and D strains revealed no PCR signal in the multiplex PCR specific for bont/a, bont/b, bont/e, and bont/f.
Similarly, each of the six singleplex real-time PCR was only positive for the predicted target. Strains 2276, 1030, 2141, 2145, and 1739, which had been originally reported to contain the bont gene, gave negative results in the six singleplex reactions. The absence of the bont genes in these cultures was confirmed by mouse assay. Neither the multiplex PCR nor the singleplex PCR analyses of the non-C. botulinum strains was positive for the bont genes.
To define the sensitivity of the multiplex real-time PCR, we determined the detection probability for different copy numbers between 6 and 800 copies of the standard plasmids in the presence of 75 copies of the IAC pKoMa2. The 95% detection probability per reaction analyzed with the multiplex assay was 32.5 copies for pBoNT/A, 7.4 copies for pBoNT/B, 24.0 copies for pBoNT/E, and 287.2 copies for pBoNT/F (Table (Table33).
The obtained detection probabilities of 95% for the standard plasmids analyzed with the singleplex PCR were 31.6 copies for pBoNT/A, 7.4 copies for pBoNT/B, 6.4 copies for pBoNT/C, 6.9 copies for pBoNT/D, 16.8 copies for pBoNT/E, and 122.4 copies for pBoNT/F per reaction (Table (Table33).
We tested the sensitivity of the multiplex PCR assay directly from food samples without prior enrichment by anaerobic culture. We spiked three different food samples (black pudding, green beans, and smoked mackerel) with 10-fold serial dilutions of C. botulinum serotypes A, B, E, and F (103 to 106 CFU/ml), extracted the DNA, and performed the multiplex PCR assay as described above. The detection limits for C. botulinum serotypes A, B, E, and F in smoked mackerel were 105, 103, 104, and 104 CFU/ml, respectively (Table (Table4).4). These detection limits correspond to 400 bacterial cells per PCR of C. botulinum serotype A, 4 cells per PCR for C. botulinum serotype B, and 40 cells per PCR for C. botulinum serotypes E and F. In green beans and black pudding, we were able to detect as little as 104 CFU of C. botulinum serotype A/ml, 103 CFU of C. botulinum serotypes B and E/ml, and 104 CFU of C. botulinum serotype F/ml (Table (Table4).4). This corresponds to detection limits of 40 cells per PCR for C. botulinum serotypes A and F, and 4 cells per PCR for C. botulinum serotypes B and E. All C. botulinum-negative samples returned positive signals specific for the IAC pKoMa2. In addition, we were also able to detect C. botulinum serotypes A, B, E, and F directly from the spiked green beans and black pudding, without previous DNA extraction (data not shown).
In order to test the multiplex and singleplex real-time PCR assays on real specimens, we analyzed six suspected food, four stool, and several tissue samples from six different cases of botulism in Germany (Table (Table5).5). All analyses were performed both on DNA extracted directly from the specimens, without prior anaerobic culture or any other enrichment step, and after enrichment by anaerobic culture. The suspected home-cured ham (case 1), the home-made green bean salad (case 2), and home-marinated herring and one stool sample from case 3 tested positive for BoNT/B, BoNT/A, and BoNT/E, respectively, by both multiplex and singleplex PCRs. The stool sample from a case of infant botulism (case 4) prior to antibiotic treatment returned a positive result for C. botulinum serotype A. However, stool specimens taken a week later, after antibiotic treatment, were negative. All tissue specimens from a cow with botulism symptoms (case 5) gave positive results in the singleplex PCR for C. botulinum serotype D. Specimens from a symptomless control animal were negative. In gut and liver specimens from three perished ducks taken during an outbreak of avian botulism (case 6) BoNT/C was identified. All samples that were positive for bont gave positive results without and after enrichment (Table (Table5),5), except for the frozen herring (case 3), which tested positive only after enrichment by anaerobic culture, both with the real-time PCR and in the mouse bioassay. In addition, all of the tested samples demonstrated positive results for the IAC, thus confirming the absence of PCR inhibitors in the extracted DNA. All PCR results were confirmed by mouse bioassays performed at the Robert Koch-Institut, at the Landeslabor Brandenburg (Frankfurt/Oder, Germany) or the Institut für Lebensmittel, Arzneimittel, und Tierseuchen (Berlin, Germany).
In the present study we developed a pentaplexed real-time PCR assay for the simultaneous detection and quantification of bont/a, bont/b, bont/e, and bont/f, including an internal control IAC in comparison to singleplex real-time PCR assays for all relevant bont target genes. We analyzed the assay performance, specificity, and sensitivity and validated the PCR approaches with differently spiked food samples and with specimens from several cases of botulism.
While the detection of single BoNT genes by conventional or real-time PCR methods is well established, a number of PCR assays for the detection of three or more BoNT genes have also been described (11, 13, 15, 23, 27, 28, 29). More recent work has complemented these studies by the description and validation of a conventional multiplex approach for the simultaneous detection of all four human pathogenic bont genes (6, 22). Previous approaches have used degenerated primers to detect more than one BoNT gene. Similarly, our pentaplexed PCR assay uses degenerate primers to accommodate sequence variations within the specific binding regions of bont genes and therefore is able to detect all human pathogenic serotypes and subtypes described thus far (A1, A2, A3, A4, and A5; B1, B2, B3, nonproteolytic B, and bivalent B; E1, E2, E3, E4, E5, and E6; F baratii, proteolytic F, nonproteolytic F, and bivalent F) (5, 7, 16). In addition, the pentaplexed PCR approach correctly identifies a novel A subtype identified in our group in the context of a recent food-borne botulism case (strain Chemnitz, Table Table11 and also unpublished results).
Generally, major advantages of the real-time TaqMan approach compared to the conventional PCR approach (using gel separation of amplified targets) include the increased assay specificity (by a third gene-specific oligoprobe), reduced assay time (online detection of amplification products, no electrophoresis step), and ease of quantification (using defined concentrations of standard plasmids). From a technical perspective, multiplexing a real-time PCR for more than four target genes is challenging for a number of reasons. (i) To date, only a few instruments are able to detect and differentiate more than four different fluorescent reporter dyes, among them the LightCycler 480 II used in the present study. (ii) An increase in the number of primers and probes per PCR at the same time enhances the problem of nonspecific oligonucleotide interactions. Therefore, shorter oligonucleotides with good hybridization performance are necessary. For this purpose, alternative probe chemistries have been introduced, such as MGB- and LNA-containing oligoprobes (19). (iii) Some of these improved oligoprobes can only be purchased coupled to a limited number of fluorescent dyes (e.g., MGB-containing probes can be purchased only with up to three different reporter dyes).
To our knowledge, the present study is one of the first to describe a multiplex real-time PCR approach that combines both MGB- and LNA-containing oligoprobes within one reaction. Our study describes only the second reported use of a pentaplexed real-time PCR assay, following a recent description of a pentaplexed assay for the simultaneous quantification of respiratory RNA viruses (25). While setting up our pentaplexed assay, we empirically learned that mixing of MGB- and LNA-containing oligoprobes rendered a superior PCR performance, compared to a multiplex assay with LNA probes only. For each target gene, we carefully selected the optimal probe chemistry (MGB or LNA) and reporter dye (Cy5, VIC, FAM, TEX, and Cyan500) in order to obtain concordant results for the pentaplex approach and the optimal singleplex reaction. In the final pentaplexed assay, both the specificity and the sensitivity were equivalent to the singleplex approach: analysis of 42 C. botulinum strains and 57 non-C. botulinum strains using the designed multiplex and singleplex real-time PCR assays showed high specificity for their predicted target, with no observed cross-reactivity.
With respect to the analysis of clinical, environmental, or food samples, the use of an internal amplification control is mandatory for any diagnostic PCR method. An internally controlled PCR would demonstrate failure of the assay, e.g., due to the presence of PCR inhibitors in the complex sample material, and therefore avoids false-negative results (18). Consequently, the use of an IAC is included in different international standards, e.g., in ISO 22174, for the analysis of food. Among all PCR assays published on the detection of C. botulinum, however, only a limited number include the use of an IAC (1, 3, 6, 24). In our pentaplexed PCR approach, the IAC based on a synthetic sequence was simultaneously amplified with the target sequences bont/a, bont/b, bont/e, and bont/f. In order to avoid competition between the bont-specific and the IAC-specific amplification, the IAC plasmid pKoMa2 was optimally used at a low concentration (75 copies per reaction). In our description of the multiplex approach, we followed almost all of the recently defined MIQE guidelines for real-time PCR experiments which serve to standardize experimental practice and to promote consistency between laboratories (4).
When we tested the sensitivity of the multiplex PCR assay with C. botulinum A, B, E, and F spiked into different food, we found a limit of detection of 103 to 105 CFU/ml directly from the spiked food samples, without prior enrichment by anaerobic culture. In addition, we successfully applied the multiplex and singleplex real-time PCR assays to specimens from cases of food-borne, infant, and animal botulism (different food, feces, and different organs). Samples that tested positive for C. botulinum by mouse bioassay concordantly tested positive by our multiplex and singleplex assays. With one exception, the multiplex and singleplex real-time PCR worked with DNA directly extracted from the suspicious specimens, without prior anaerobic culture, thus yielding results faster than the mouse bioassay. Nevertheless, the independent enrichment step by anaerobic culture is recommended for cases where low concentrations of bacteria are present.
Advancing the existing PCR-based detection of BoNT-producing clostridia, the present study describes a pentaplexed real-time PCR assay for the simultaneous detection of bont/a, bont/b, bont/e, and bont/f, including an IAC. Of technical interest is the successful combination of both MGB- and LNA-containing oligoprobes in the current pentaplex approach. The multiplex assay was found to be as sensitive and specific as the independently developed and validated singleplex real-time PCR assays for all relevant bont genes. Furthermore, the pentaplex assay allows the detection, differentiation, and quantification of all serotypes and subtypes associated with human botulism known thus far, including a novel A subtype identified recently in our group. Most importantly, the assay turned out to be very robust and rapid when analyzing food and clinical specimens from cases of botulism. Therefore, this assay will be of great diagnostic value as a rapid screening tool in surveillance studies or in outbreak situations.
We thank Rolf Bergmann (Thüringer Landesamt für Lebensmittelsicherheit und Veterinärmedizin, Bad Langensalza, Germany), Arne Rodloff and Reiner Schaumann (Konsiliarlabor für Anaerobe Bakterien, Leipzig, Germany), and Ulrich Nübel (Robert Koch-Institut, Wernigerode, Germany) for sharing bacterial strains with us. Furthermore, we are very grateful to Angelika Aue and Ulrich Wittstatt from the Institut für Lebensmittel, Arzneimittel und Tierseuchen (Berlin, Germany) for sharing some of the botulism specimens with us. We are indebted to Barbara Biere (Robert Koch-Institut, Berlin, Germany) for the sequence and plasmid for the IAC (pKoMa2) used in this study.
This study was supported by grants from the German Federal Ministry of Health and the Federal Ministry of Education and Research (BiGRUDI project, 13N9601) to B.G.D.
Published ahead of print on 30 April 2010.