Detection, identification and quantification of Campylobacter jejuni, coli and lari in food matrices all at once using multiplex qPCR

Genomic DNA from pure bacterial cultures (Table 1) was extracted by thermal lysis. Concentration and purity of the DNA were determined spectrophotometrically using NanoPhotometer™ (Implen, Germany). Only samples whose A₂₆₀/A₂₈₀ ratio ranged from 1.7 to 2.1 were used for further analyses. Real genome copy number was determined using the formula:Genome copies/µ l = (C × N_A × 10^− 9)/(genome length(bp) × M_w), where C is a measured concentration of extracted DNA (ng/μl), N_A is Avogadro’s constant (6.02 × 10²³ molecule/mole) and M_w is molecular weight of 1 bp which is 660 Da [28].

Primers and hydrolyses probes selection was based on previously published studies [18, 28]. Although previously published, selected primers and probes (Table 2) have not been used together in one reaction and therefore it was necessary to perform additional in silico analyses. Their specificity was tested against 30 bacterial genomes (Table 3) using both basic nucleotide BLAST at NCBI (Basic local alignment search tool; National centre for biotechnology information) and FastPCR molecular biology software [38]. Additional more comprehensive analysis for primer pair specificity checking was conducted using Primer-BLAST tool at NCBI. As a database query “Genome (chromosome of all organisms)” was selected and as an organism query was selected “bacteria (taxid: 2)”. In order to enable proper optimization of qPCR protocol for a multiplex platform, necessary determination of primers and probe chemical characteristics was also carried out using FastPCR molecular biology software [38].

In addition to in silico analyses experimental primers’ specificity verification was implemented. Horizontal agarose-gel electrophoresis and melt curve analysis utilizing SYBR Green fluorescent dye (2×Power SYBR Green PCR Master Mix, Applied Biosystems, USA) were performed. All target and non-target Campylobacter species listed in Table 1 were used for these analyses.

Choice of probes and primers concentrations for the first experiments was based on our research on previously published assays where the most common final concentrations were 0.40 μM and 0.20 μM for primers and probes respectively e.g. [22, 24, 39‐42]. Before standard curves were generated, one experiment was carried out with all target and non-target Campylobacter species as well as with other bacterial strains (Table 1) in order to experimentally verify also specificity of probes. After the positive verification only three reference target Campylobacter species (C. jejuni CCM 6212, C. coli CCM 6211 and C. lari CCM 4897) were used for further experiments. Standard curves for singleplex qPCR platform were generated as described in Additional file 1. All samples were run in duplicates and non-template (NTC) and positive controls were included.

Multiplex qPCR was evaluated in two phases. In the first phase, reaction was performed with all components (primers and probes) with DNA from one strain in a tube in order to find out whether undesirable inhibition caused by interaction between components occurs or not. Standard curves were generated and quantification cycles (Cq), y-intercepts, slopes, efficiencies (E), standard deviations (SD), correlation coefficients (R²) and linear ranges were determined using 7500 Software (Applied Biosystems, USA, version 2.0.5), Microsoft Office Excel 2007 (Microsoft, USA, version 2007) and GenEx software (MultiD Analyses AB, Sweden, version GenEx 5 Enterprise).

In the second phase, reaction mixture contained all chemical components as well as mixed DNA sample from all target Campylobacter strains. Standard curves were generated and abovementioned parameters determined using the same approach. In accordance with obtained results the concentrations of individual components and reaction conditions were optimized until satisfying values were obtained (Additional file 1).

Output data were analysed with instrument compliant 7500 Software (Applied Biosystems, USA, version 2.0.5), Microsoft Office Excel 2007 (Microsoft, USA, version 2007) and GenEx software (MultiD Analyses AB, Sweden, version GenEx 5 Enterprise).

Three different types of food matrices were chosen for empirical assay validation – raw chicken wings (local butchery, Prague, Czech Republic), whole frozen chicken without giblets (local hypermarket, Prague, Czech Republic) and fried chicken strips (fast food restaurant, Prague, Czech Republic). All samples were processed immediately after the purchase as described in Additional file 1, and obtained chicken wing rinses, chicken juice and homogenate from fried chicken strips were further examined.

Each sample was divided into four aliquots. One remained unspiked and the rest was artificially contaminated with pure culture of individual reference Campylobacter strain and then serially diluted in order to achieve the range of approximately 10¹-10⁵ CFU/ml. The number of Campylobacter cells used for the spiking was determined on Karmali agar with Campylobacter selective supplement (sodium pyruvate, vancomycin, cefoperazone and cycloheximide; Oxoid, UK) using a drop plate method [43], by our qPCR as well as via genome copy number determination in pure cultures used for spiking [28].

DNA was isolated from 750 μl of each food sample using commercial PrepSEQ® Spin Sample Preparation Kit with Protocol (Applied Biosystems, USA) in accordance with manufacturer’s recommendations.

qPCR reaction was performed as described in Additional file 1. Briefly, standard curves were constructed using 5 μl of mixed DNA extracted from target Campylobacters. Used DNA concentrations ranged approximately from 10⁰ to 10⁶ genome copies (CFU equivalent) per well for C. jejuni and from 10⁰ to 10⁵ for C. coli as well as for C. lari. In the case of food samples 10 μl of DNA were added into reaction. Simultaneously, all food samples were examined for presence of Campylobacters using drop plate method [43].

First singleplex qPCR served for specificity screen of hydrolysis probes. As a sample DNA isolated from all bacteria listed in Table 1 was used. Three different combinations of primers and probes final concentrations in reaction were tested as follows: 0.40×0.20, 0.30×0.10 and 0.20×0.05 μM respectively. There were no significant differences between Cq values, therefore for further experiments in singleplex platform the combination of concentrations 0.40×0.20 μM were used. No fluorescent signal was detected when non-target DNA was used as a sample. When specificity of probes was positively verified standard curves for target Campylobacters were generated. Quantification cycles and efficiencies were 18.54 and 84.56% for C. jejuni, 25.27 and 87.11% for C. coli, 16.19 and 77.93% for C. lari (Table 4) when 10⁶ genome copies per well used. Because of very high Cq value for C. coli when compared with the others, another combinations of primer and probe concentrations were tested as follows: 0.40×0.20 to 0.50, 0.50×0.20 to 0.50 and 0.80×0.80 μM. However, no significant differences were observed. Another in silico analysis showed one non-complementary base at 3’ end of the forward primer, which was not issue when used in original study [18] where only duplex qPCR was evaluated and quantification cycles ranged between 18.80-23.00 (when 10⁶ genome copies per well used). Therefore 3’ end of original primer was amended by adding a two bases which increased a stability and specificity of annealing step (Additional file 1). Our adjustment caused significant decrease in Cq value to 15.72 (when 10⁶ genome copies per well used) and slight increase in efficiency as well (89.24%) when concentrations of primers and probes in reaction were 0.40×0.20 μM (Table 4). Due to the fact that singleplex platform was mainly performed in order to verify the functionality of the reaction, we proceeded directly to the multiplex without further optimization in order to improve obtained values.

In the first experiment, all components were present in reaction mixture at the same concentration as in the singleplex, but DNA sample in each reaction originated only from individual species (not mixed sample). Considering greater number of components when multiplexing, reaction volume was increased from 25 μl to 30 μl. Results showed that there is no inhibition of the reaction caused by interaction between components. Serial dilutions of DNA were in the range of 10⁰-10⁷ genome copies (CFU equivalents) per well. Quantification cycles and efficiencies were 23.80 and 91.35% for C. jejuni, 23.41 and 95.23% for C. coli and 21.53 and 92.12% for C. lari when 10⁵ genome copies per well used (more details in Additional file 1).

In second phase multiplex with mixed DNA sample was evaluated. First experiment was carried out under the same conditions as the singleplex. Serial dilutions of mixed DNA were in the range of 10⁰-10⁷ genome copies of each strain per well. Although Cq values and efficiencies for C. coli and C. lari were comparable with previous multiplex results (DNA from single strain), it was unambiguous that strong inhibition of amplification occurred in the case of C. jejuni because of a complete disappearance of its PCR product. Therefore conventional multiplex PCR (all three pairs of primers and three probes) with end point horizontal agarose-gel electrophoresis was conducted with two possible combinations of DNA present in sample (C. jejuni×C. coli; C. jejuni×C. lari) in order to determine in which case the problem occurs. Based on results it was found that when all components are present in reaction with DNA sample mixed of C. jejuni and C. lari the amplification of C. jejuni target is affected and the typical PCR product does not form. Having regard to the fact that there was no inhibition due to competition for other reaction components when multiplex with DNA sample from each strain individually was performed, this indicated that there was some interaction between C. jejuni and C. lari DNA even though the trend of C. lari reaction was not affected at all. Considering this fact, another optimization was necessary and various concentrations of C. jejuni and C. lari specific primers and probes were tested (results not shown).

Fully optimized reaction mixture consisted of 0.80 μM C. jejuni, 0.40 μM C. coli and 0.05 μM C. lari primers and 0.20 μM of each probe. Eight points of ten-fold serial dilutions in the range of 10⁰-10⁷ genome copies of each strain per well were used to generate standard curves. Values of quantification cycles and efficiencies are 22.99 and 90.85% for C. jejuni, 20.77 and 96.97% for C. coli, 20.04 and 91.05% for C. lari, when 10⁵ genome copies per well used (Table 4). All reactions were linear over seven orders of magnitude in the range 10¹-10⁷ with potential to cover wider range in higher orders. Detection limits of this assay were determined to be between 6.62-16.10 genome copies/well for C. jejuni, 5.13-6.30 genome copies/well for C. coli and 4.87-5.23 genome copies/well for C. lari. All other parameters are provided in Additional file 1.