Bacterial strains

Strain CIP103974 was used to artificially contaminate raw milk samples. A colony grown on a HEYA slant was suspended in 1 ml of Ringer's solution (Mayrhofer Pharmazeutika GmbH&Co KG, Leonding, Austria). Cells were declumped by passing the suspension three times through a disposable insulin syringe (Omnican^®40; B. Braun Melsungen AG, Melsungen, Germany). Debris was removed at 3,000 × g for 1 min. Cells remaining in the supernatant were pelleted at 8,000 × g for 5 min and washed three times in 1 ml Ringer's solution. Again, cells were declumped with a syringe and the cell count in the suspension was determined with a commercial bacterial viability kit (Live/Dead Bac Light; Molecular Probes, Willow Creek, OR, USA) and filtration onto 0.22-μm-pore size, 13-mm black polycarbonate filters (Millipore, Billerica, MA, USA).

DNA isolation of raw milk samples

Separation of MAP from the 98 raw milk samples collected in the Free State of Thuringia in Germany was performed by a method including different detergents, solvents and centrifugation steps [15]. The raw milk samples comprised pre-milk (first streams of milk collected at the beginning of milking) and main milk samples. For each sample, four aliquots of 15 ml of raw milk each were subjected to this protocol. Afterwards, the bacterial pellet was subjected to DNA isolation using the NucleoSpin tissue kit, yielding 100 μl DNA suspension. For each sample, a total of 400 μl DNA suspension (100 μl from each 15 ml aliquot) was concentrated by ethanol precipitation. The DNA pellet was re-suspended in 28 μl ddH₂O. Thus, the DNA from 60 ml milk was eventually concentrated into a 28-μl volume.

Comparative DNA and RNA isolation of raw milk samples and inoculated Dubos broth

The bacterial pellet recovered from 15 ml of artificially contaminated raw milk was either subjected to DNA isolation (NucleoSpin tissue kit) or RNA isolation (High Pure RNA Isolation kit; Roche Diagnostics GmbH, Vienna, Austria), or transfered to a 5-ml volume of BBL™ Dubos broth (Becton Dickinson Austria GmbH, Schwechat, Austria) supplemented with 2 mg/ml Mycobactine J (Synbiotics Europe SA, Lyon, France) after decontamination with 0.1% benzalkonium chloride (Sigma-Aldrich Handels GmbH, Vienna, Austria) [16] and incubated at 37°C. Bacteria were recovered at 8,000 × g for 5 min after incubation for 1 or 2 weeks and subjected to DNA and RNA isolation.

Real-time PCR

The presence of PCR inhibitors in the DNA isolated from milk was tested as previously published [17]. Two-hundred copies of an artificially synthesized 79-bp region of the COCH gene of zebrafish (Danio rerio) (VBC, Genomics, Vienna, Austria) were included in a 25-μl real-time PCR reaction specific for that target. Obtained copy numbers were compared for reactions with or without the addition of isolated DNA from raw milk. A reduction in the obtained copy number or a negative result would indicate the presence of PCR inhibitors in the milk-derived DNA added to the PCR reaction.

Real-time RT-PCR

A 5-μl volume of RNA was reverse-transcribed into complementary DNA (cDNA) using 2 pmol Mptb52.16 specific primers and SuperScript III reverse transcriptase (Invitrogen) according to the manufacturer's instructions. RNase OUT recombinant RNase inhibitor (Invitrogen) was included in the cDNA synthesis reaction. Five microliter aliquots of the 20-μl volume of cDNA template generated were used for real-time PCR analysis as described above. The RT- control included 1.2 μl of un-transcribed RNA, which equals the quantity of RNA in the 5-μl aliquot of cDNA.

Development and optimization of the real-time PCR assay

The objective of the present study was to develop a real-time PCR assay for detection and quantification of MAP in milk, with the additional option of differentiating between viable and dead cells. The gene fragments Mptb52.1, Mptb52.16, Mptb52.33 were selected for the analysis because they are specific for MAP and were shown to be expressed in pure cultures in Dubos broth [14]. Primer and TaqMan^® probes were designed and each assay was tested with DNA isolated from raw milk (Table 2).

During optimization of the assays, severe interference of the raw milk background (e.g. DNA from microflora as well as DNA from somatic cells) was noted with all three assays when performing agarose gel electrophoresis of the amplicons (data not shown). As these problems were not grossly evident on the real-time PCR amplification curves, but competition for reagents during formation of unspecific amplicons might be detrimental for the detection of low contamination levels, these findings confirm the necessity to perform gel analysis when optimizing real-time PCR assays for environmental samples. Only the Mptb52.16 target could be optimized by increasing the combined annealing/extension temperature and by decreasing the MgCl₂ concentration without sacrificing the efficiency of the real-time PCR reaction. BLAST search indicated that the selected target sequences for Mbtp52.16 specific primers and probes display 100% identity with MAP strains only. Partial identity with different species of other bacteria for each of the oligonucleotides was noted but regarded as not sufficient to enable detection with the selected primers/probe combination. However, it might have been the reason for the observed interference of the raw milk background revealed by agarose gel electrophoresis.

The addition of 50 copies of an IAC to the optimized assay permitted identification of false negative results without exerting a negative effect on the MAP-specific real-time PCR (Figure 1). The amplification efficiency of the real-time PCR reaction with and without the addition of the IAC was 90.5% and 87.2%, respectively.

Since contamination levels in raw milk are low (e.g. l00 cfu/ml in milk from symptomatic and 2-8 cfu/50 ml in milk from asymptomatic cows) [4] and bacteria are not homogenously distributed at low contamination levels, it was decided to pool and concentrate the DNA isolated from four 15-ml aliquots to a final volume of 28 μl. Since 6 μl were transferred to the PCR reaction, a volume of 12.8 ml milk could be analyzed in a single PCR reaction. Performing four PCR reactions per sample permitted to screen a volume of 51 ml milk. It was confirmed that this approach did not lead to the concentration of PCR inhibitors (data not shown). In addition, when analyzing field samples, no false negative results were indicated by the integrated IAC.

Inclusivity and exclusivity testing of pure cultures of MAP (n = 10), non-MAP mycobacteria (n = 22) and raw milk microflora (n = 16) yielded no false positive or false negative results (Table 1), thus confirmed the specificity of the selected target, which was reported to be a drawback of the widely used IS900 target because other mycobacteria harboring similar sequences were observed [11–13]. Exclusivity testing could be expanded further by including members of the Mycobacterium tuberculosis complex and mycobacteria other than M. xenopii, harboring IS900-like elements such as M. chelonae, M. terrae, and M. porcinum [13, 20]. However, since the selected target is different to IS900, specificity issues with those might not be expected.

Detection limit in artificially contaminated raw milk samples

When determining the detection limit, great care was taken to avoid the introduction of large quantities of free DNA into the artificially contaminated milk samples, which might cause a bias towards a lower detection limit determined with a DNA-based method. To avoid overestimation, the bacterial suspension was washed thoroughly [21]. Cells were counted with a microscope because growth-related counts tend to underestimate the number of MAP cells present in a bacterial suspension [4, 22]. Differences ranging up to 2 log scales were observed when counting cells in a counting chamber compared to performing colony counts [21].

Testing different batches of raw milk as well as pasteurized milk and UHT milk to confirm the absence of the MAP-specific target in the milk for artificial contamination experiments revealed a low level of contamination in some of the tested samples. Of two pasteurized milk samples and one UHT milk sample, one of the first and the latter yielded positive results. The contamination level was low: 8.7 × 10^-1 bce/ml pasteurized milk and 2.7 × 10⁰ bce/ml UHT milk. A conventional commercially available PCR assay based on an alternative target (ISMav2) was used to verify the presence of MAP DNA in these samples, and yielded identical results. As all process controls were negative, we could rule out laboratory contamination. Other authors have also observed low numbers of PCR-positive and occasionally even culture-positive pasteurized milk samples [16, 23, 24]. Eventually, a consistently negative batch of raw milk was identified and used for further experiments.

A dilution series of MAP cells in milk containing 2.42 × 10⁵ to 2.42 × 10^-1 MAP cells/ml milk was prepared and analyzed with the real-time PCR assay. The obtained detection limit (2.42 × 10¹ MAP cells/ml milk) was comparable to the results achieved with other real-time PCR assays [22, 25–29]. The majority of the published DNA isolation methods were based on mechanical cell lysis. For the present study we used a non-mechanical method of cell lysis which was successfully applied for real-time PCR-based detection of the Gram-positive bacterial species Listeria monocytogenes in milk [15]. This protocol includes multiple incubation and washing steps at 45°C in combination with solvents and detergents, and maybe permits the recovery of MAP from milk fat as well [25].

Naturally contaminated milk samples

Analysis of 46 pre-milk and 52 main milk samples obtained from farms with a known history of Johne's disease yielded 47.8% MAP-positive pre-milk and 51.9% MAP-positive main milk samples, which is somewhat higher than the data reported by other authors. These farms had cows with a positive blood ELISA, fecal culture, or both, and were participating in a MAP-control program. Twenty-five paired pre-milk and main milk samples were available. Nine of these (36%) were positive in pre-milk only, eight (32%) in main milk only, and one sample (4%) was positive in both types of milk. Some authors observed 32.5% PCR-positive pre-milk samples in a farm with two cows shedding MAP [27], whereas others observed 33% PCR-positive main milk samples under similar circumstances [30]. Lower numbers are usually reported for farms with no cases of diarrhea or weight loss [31]. Comparisons of data are rendered difficult by the fact that different methods were used for DNA isolation and different volumes of milk were analyzed. We analyzed the DNA isolated from 51 ml of milk per sample, whereas other authors had performed PCR analysis on the DNA of 0.1 to 2.5 ml milk [4, 22, 25–27]. Contamination of the pre-milk and the main milk samples ranged from 1 × 10⁰ to 5.1 × 10² bce/51 ml and from 1 × 10⁰ to 6.5 × 10¹ bce/51 ml milk, respectively, with 90.9% of positive pre-milk and 96.3% of positive main milk samples having less than one bce per ml milk. Almost no quantitative data were available for comparison. A few tens to no more than 560 MAP cells/ml milk were detected by real-time PCR in individual milk samples of a herd with a known history of paratuberculosis [27]. Other authors reported less than 100 MAP cells/ml in bulk tank milk, including herds which were MAP positive by environmental culture [25].

Exploring the potential for detection of viable MAP

Detection of RNA might indicate the presence of metabolically active MAP in the sample and could be performed either directly in the sample or after a short incubation period in broth. Transcription of the Mptb52.16 target was analyzed in artificially contaminated 15-ml volumes of raw milk and in 5-ml volumes of Dubos broth inoculated with bacteria collected from the 15-ml volumes of milk. The contamination level of raw milk was 5.1 × 10⁶ MAP cells/15 ml. From this sample, 2.2 × 10⁶ DNA targets and 1.0 × 10⁴ RNA targets were recovered. After one week of incubation in Dubos broth, 8.8 × 10⁵ DNA and 1.3 × 10⁴ RNA targets were identified. The numbers of DNA and RNA targets obtained after an additional week of incubation in Dubos broth was similar (1.0 × 10⁶ DNA and 9.6 × 10³ RNA targets). These data suggest that there was no detectable growth of MAP or increase in the expression of the Mptb52.16-specific target during the observation period. The difference in RNA and DNA target number could indicate a low level of expression of the selected target or low metabolic activity of the MAP cells in general, which could be characterized further at the 16 S rRNA level [32]. On the other hand, the efficiency of RNA isolation might differ from the efficiency of DNA isolation or free DNA could still be transferred into the sample together with the bacterial cells. In addition, insufficient reverse transcription efficiency might have influenced the result. The RT- control indicated the presence of 7.5% DNA in RNA isolated from raw milk. After one week of incubation in Dubos broth, only one of two replicates of the RT- control remained positive on real-time PCR. After the additional week of incubation, no positive signal was identified in the RT- control.