Detection of viable and dead Listeria monocytogenes on gouda-like cheeses by real-time PCR

Authors


K. Rudi, MATFORSK, Norwegian Food Research Institute, Osloveien 1, N-1430 Ås, Norway (e-mail: knut.rudi@matforsk.no).

Abstract

Aims:  Surface contamination by Listeria monocytogenes of gouda-like cheeses during processing represents a potential public health problem. The aim of this work was to develop novel real-time PCR diagnostics to detect the presence of viable, dead or viable but not culturable (VBNC) cells on gouda-like cheeses.

Methods and Results:  We used ethidium monoazide bromide (EMA)-PCR for direct quantification of viable and dead cells, while semiquantitative detection of culturable cells below the PCR detection limit (c. 100 CFU g−1) was obtained by combining growth and real-time PCR. We were able to quantify the fraction of >0·5% viable cells in a background of dead cells by EMA-PCR, given that the viable cell concentration was above the PCR detection limit. The combined growth and real-time PCR complemented the EMA-PCR, and enabled semiquantitative detection of low levels of culturable cells (10 and 100 CFU g−1).

Significance and Impact of the Study:  The significance of this work is that we have developed a novel concept for detection of viable and potentially infectious L. monocytogenes.

Introduction

Real-time quantitative PCR is increasingly being used for pathogen detection and quantification in foods (Norton 2002). There are still two major challenges with the widespread use of PCR for quantitative diagnostics. The detection limit is mainly determined by the amount of material that can be amplified in a single reaction. The other major limitation is the detection of DNA from dead cells (Herman 1997; McKillip et al. 1999; Nogva et al. 2003). This is a particular problem for processed foods or foods subjected to long-time storage.

Listeria monocytogenes is a major problem in food production. This psychrotrophic bacterium is adapted to the chill chain commonly used for food production and storage. Listeria monocytogenes may reach high numbers in foods with long shelf-life (Pearson and Marth 1990). Quantitative diagnostics is important for this bacterium both for control, and to understand contamination and food colonization (Bohnert et al. 1992; Olsen et al. 1995). One hundred viable and infectious L. monocytogenes per gram food has been proposed as an acceptable detection limit from a public health perspective, although there is no worldwide consensus for this (Jay 1996). The reason for controlling L. monocytogenes is that this bacterium can cause the serious disease listeriosis in adults or elderly people with reduced immune system. Pregnant women and their foetuses are also at risk for listeriosis. The mortality rate for listeriosis may exceed 30% during outbreaks (Temple and Nahata 2000).

The contamination of L. monocytogenes in gouda-like cheeses may represent a potential health problem. There are several reports on L. monocytogenes contamination of these products (Jensen et al. 1994; Bachmann and Spahr 1995; Schaffner et al. 2003). Listeria monocytogenes can be introduced through the raw materials, during processing or storage. It does not normally grow in the cheese, but can survive for prolonged periods. The bacteria may also enter a viable but not culturable (VBNC) state where they can no longer be detected by growth-based techniques (Besnard et al. 2000).

The aim of this work was to develop approaches for the detection of low levels (≤100 CFU g−1) of viable (including VBNC) L. monocytogenes on gouda-like cheeses. We addressed the potential presence of dead L. monocytogenes by using a novel real-time PCR-based approach (Nogva et al. 2003) for quantitative differentiation between viable and dead cells. Furthermore, we evaluated the combination of growth-based techniques and real-time PCR in order to obtain semiquantitative information for L. monocytogenes contamination levels below the PCR detection limit (c. 100 CFU g−1).

Materials and methods

Strains and culture conditions

The following L. monocytogenes strains were used in this work; R 54 (Tine BA, Oslo, Norway), MF 182 (Matforsk, Norwegian Food Research Institute, Ås, Norway), MF 228 (Matforsk), MF 830 (Matforsk), MF 1314 (Matforsk), MF 1345 (Matforsk), DSMZ 20600 (Deutsche Sammlung von Mikroorganismen und Zellkulturen, Braunschweig, Germany) and LO28 (Institut Pasteur, Paris, France). All the strains were plated on blood agar. The strain R 54 has been adapted to rifampicin as described in Foegeding et al. (1992) prior to the work presented here. This strain was plated on blood agar containing rifampicin (100 μg ml−1). The liquid media used were either brain–heart infusion (BHI; Oxoid, Basingstoke, UK) medium or Listeria enrichment broth (LEB; Oxoid). Incubations were performed at 30°C in an aerobic atmosphere.

Spiking experiments

The gouda-like cheese Norvegia was used in the spiking experiments. This gouda-like cheese is produced in Norway (Tine BA). Slices of c. 10 g cheeses were spiked with 50 μl of L. monocytogenes suspension in duplicates for strains MF 182, 228, 830, 1264 and 1345 resulting in a final concentration of c. 10 and 102 CFU g−1, while the spiking was performed in triplicate with 10 CFU g−1 for the strain R 54 for the combined growth and real-time PCR experiments. A range of spiking with 102–104 CFU g−1 of R 54 was used in the optimization of DNA purification. The cultures were spread evenly on the surface and subsequently dried at room temperature for c. 2 h for all spiking experiments.

The slices were packed under vacuum in a polyethylene film package (Multivac; Sepp Hassemüller GmbH, Wolfertschwenden, Germany). Then the packages were stored at 10°C from 1 to 2 weeks to simulate the situation of natural contamination. The spiked material was then transferred to a stomacher bag with 90 ml of prewarmed LEB (45°C) and stomached for 1 min. The samples used for optimization of DNA purification were analysed directly without enrichment. The other samples were incubated at 30°C, and 1 ml of the suspensions were collected for DNA purification after 16 and 24 h for the strains MF 182, 228, 830, 1264 and 1345. The strain R 54, however, was analysed after 6, 12, 16, 20 and 24 h. Plate counts were also conducted in parallel for R 54.

The spiking for the ethidium monoazide bromide (EMA)-PCR was carried out as described above with some modifications. The cheese was spiked with c. 106 CFU g−1 (both viable cells, and cells that have been killed by boiling for 10 min) with the strains LO28, R 54 and DSMZ 20600. The samples were then dried on the surface for c. 2 h, then transferred directly to a stomacher and stomached for 1 min. The homogenate was subsequently analysed following the EMA treatment protocol described below.

EMA treatment

The viable/dead stain EMA was purchased from Molecular Probes Europe BV (Leiden, The Netherlands). This is a molecule that selectively enters cells with compromised membranes, and binds to DNA. EMA was added to samples at a concentration of 100 μg ml−1 in a final volume of 1 ml using transparent microcentrifuge tubes. The samples were then incubated in the dark for 5 min, and subsequently exposed to light for 1 min for covalent binding of EMA to DNA. The light source was an OSRAM SLG 1000 with a 650-W halogen light bulb (Osram, Drammen, Norway), which was placed 20 cm from the sample tubes. The microcentrifuge tubes were then placed on ice prior to light exposure to minimize elevated temperature in the samples. The DNA from the EMA-treated samples was then purified following the DNAeasy DNA purification protocol described below.

DNA purification

DNA for PCR analyses was either purified with the DNAeasy tissue kit (Qiagen, Hilden, Germany), DNA DIRECT (Dynal AS, Oslo, Norway), or the Bugs'n Beads kit (Genpoint AS, Oslo, Norway). The DNA was purified from 1 ml samples. The cells in the samples were pelleted for 5 min at 5000 g in a microcentrifuge (Biofuge Fresco; Kendro Laboratory Products, Asheville, NC, USA). The supernatant was discharged, and the pellet used for further DNA purification. We followed the recommendations by the manufacturers for DNA purification from Gram-positive bacteria. Finally, the purified DNA was resuspended in 100 μl of the respective DNA elution buffers.

Real-time PCR amplification

The real-time quantitative PCR was performed as previously described by Nogva et al. (2000). Amplification reactions (50 μl) contained a DNA sample (2–10 μl); 1x TaqMan buffer A; 0·1 μmL. monocytogenes-specific probe (5′-CGA TTT CAT CCG CGT GTT TCT TTT CG-3′); 0·3 μmL. monocytogenes-specific primers (forward 5′-TGC AAG TCC TAA GAC GCC A-3′ and reverse 5′-CAC TGC ATC TCC GTG GTA TAC TAA-3′), and 2·5 U of AmpliTaq Gold DNA polymerase. All consumables were supplied by Applied Biosystems (Foster City, CA, USA). The PCR mixture was heated to 95°C for 10 min before amplification to denature the template DNA and to activate the DNA polymerase. The amplification profile was as follows: 40 cycles of 95°C for 20 s and 60°C for 1 min. Reactions were performed in the ABI Prism 7700 Sequence Detection System (Applied Biosystems). A fluorescence threshold was chosen which was used to determine the first cycle (CT) where the signals were above the threshold value.

Analyses of the real-time EMA-PCR results

The DNA form dead cells with bound EMA cannot be PCR amplified. For the EMA-treated samples we estimated the EMA signal reduction (EMASR), which represents the DNA fraction that cannot be PCR amplified (from dead cells) in the EMA-treated samples.

image

where CTtreat is the CT value for the EMA-treated sample, and CTuntr is the CT for the corresponding untreated sample. E is the amplification efficiency of the L. monocytogenes real-time PCR. This has previously been determined to 0·9 by Nogva et al. (2000). To make correlations between viable cells determined by EMA-PCR against viable cell determinations in known standard mixtures, 10-fold serial dilutions of viable bacteria in a constant background of heat-killed bacteria (boiled for 10 min) were made and subjected to EMA-PCR. If all the DNA in dead cells is inactivated by covalent binding of the EMA molecule and all the DNA in viable cells can be PCR amplified after EMA treatment, then one expects a linear relation between the log fraction of viable cells, and log EMASR. For the empirical data, however, deviations from linearity were observed, particularly when approaching the detection limit of the assay. We used the best-fit polynomial regression formula y = 0·86x3 − 2·1x2 + 0·42x (−2·5 ≤ x ≤ 0) to correct for linear deviations and to estimate the log viable cell fraction (y) from the log of the EMASR (x).

Results and discussion

DNA purification directly from the cheese matrix

We evaluated three different kits for the isolation and detection of L. monocytogenes spiked on cheese. The kits were selected based on the criteria of being adaptable to routine applications. We tested the DNAeasy tissue kit (Qiagen), DNA Direct (Dynal), and Bugs'n Beads (Genpoint).

Listeria monocytogenes R 54, spiked at a concentration of c. 104 CFU g−1 on Norvegia cheese was used in the initial evaluation of the DNA purification. The highest recovery was obtained for the DNAeasy kit (c. 10-fold better yield than for the other kits), as determined by real-time PCR amplification. The reason for the differences between the kits could be that the DNAeasy kit includes proteases to degrade the cheese matrix, and that it is the cheese matrix that influences the recovery. This was confirmed by purifying DNA form the corresponding cultures used for spiking. Here, there were no detectable differences between the three approaches.

The DNAeasy kit was evaluated further with respect to detection limit by analysing dilution series from 102 to 104 CFU g−1. We obtained a detection limit of c. 102·5 CFU g−1 when amplifying 5 μl of the purified DNA in a 50-μl PCR reaction. This is on the border of the theoretical detection limit taking the amount of material analysed in a single PCR reaction (corresponds to 5 mg of the cheese matrix). The DNAeasy approach was therefore used throughout this work.

Separate detection of viable and dead L. monocytogenes

Listeria monocytogenes is found in raw materials and ingredients used for food production (Schlech 2000). Listeria monocytogenes is, for instance, associated with raw and unpasteurized milk (Bachmann and Spahr 1995). The DNA from the bacteria may be present although the bacteria are killed during processing. We have recently developed EMA-PCR which is a novel approach for real-time PCR quantification of viable and dead cells (Nogva et al. 2003). This technology could solve the problem of whether the identified L. monocytogenes are VBNC or dead.

We evaluated the potential of EMA-PCR for quantitative detection of viable and dead L. monocytogenes by diluting viable L. monocytogenes in the background of dead bacteria. A quantitative correlation was determined for a fraction viable relative to dead bacteria down to c. 0·5% (Fig. 1). The relative detection of viable and dead cells are independent of the total amount of bacteria, as long as the fraction of viable cells is above the PCR detection limit. EMA-PCR was evaluated further by spiking the cheese matrix with viable and dead cells. We obtained a good quantitative differentiation between the viable and dead cells on the cheese matrix. The detection limit was in the same range as for the pure culture (Fig. 2). Thus, the cheese matrix did not seem to influence the viable/dead assay.

Figure 1.

Determination of viable cell-fraction using the EMA-PCR. An overnight culture of strain R 54 was adjusted to c. 108 CFU ml−1, and diluted in a background of 108 CFU ml−1 corresponding heat-killed bacteria (boiled for 10 min). The fraction of viable cells, as determined by EMA-PCR, was plotted against cultures with known fractions of viable cells. The broken line depends on the fraction of dead cells in the fresh culture

Figure 2.

Real-time EMA-PCR for Listeria monocytogenes LO28 (a), R 54 (b) and DSMZ 20600 (c) spiked on cheese. The cheese matrix was spiked with 106 CFU g−1 viable or heat-killed (boiled for 10 min) L. monocytogenes. Black bars represent killed bacteria, while white bars represent viable bacteria. The error bars represent standard deviations from three independent replicates

Potential natural contamination by L. monocytogenes

We detected low levels of L. monocytogenes DNA (corresponding to c. 104 CFU g−1) in one of 18 unspiked samples. However, no bacteria were recovered by growth-based techniques in any of these samples. We used the EMA-PCR to evaluate if the DNA was from VBNC or dead bacteria. The fraction of viable bacteria was determined to be <1% (the detection limit of the direct real-time PCR analyses). We thus concluded that the DNA in the cheese sample was from dead L. monocytogenes. This had not been possible to determine with the current technology (Barer and Harwood 1999).

Quantitative detection of L. monocytogenes on a cheese matrix by combined growth and real-time PCR

We tested the possibility of combining growth and real-time PCR for semiquantitative detection of low levels of L. monocytogenes in the cheese matrix (Fitter et al. 1992).

First, we analysed samples spiked with L. monocytogenes R 54 for making a standard curve. The growth of L. monocytogenes was exponential in the LEB medium for the tested time-ranges (Fig. 3a). This was also reflected from the CT values for the given time-points using real-time quantitative PCR (Fig. 3b). The conclusions from this experiment are that it is possible to obtain semiquantitative relations between growth and real-time quantitative PCR, and that the growth did not seem to be affected by the drying or storage under nongrowth conditions.

Figure 3.

Correlation of incubation time with CFU (a) and CT values (b) respectively. Three independent time-series were conducted for strain R 54. The r2 for was 0·99 both for the CFU (a) and the CT (b) regression curves

The next question we addressed was whether it was possible to use the standard curve for the R 54 strain to predict low levels of other L. monocytogenes strains. This was performed by spiking the cheese matrix with five strains at known concentrations (10 and 100 CFU g−1). The real-time PCR CT values were then determined after 16 and 20 h of growth. The standard curve for L. monocytogenes R 54 was used to predict the original amounts of L. monocytogenes in the samples from the CT values obtained (Table 1). The overall predictability was 100·8±0·2 CFU g−1 after 16 h and 101·2±0·2 CFU g−1 after 20 h for the samples spiked with c. 10 CFU g−1. The corresponding values for the samples spiked with c. 102 CFU g−1 were 101·8±0·1 and 102·1±0·2 for 16 and 20 h respectively.

Table 1.  Combination of growth and real-time PCR for estimating low amounts of Listeria monocytogenes on spiked cheese
StrainSpiked (CFU g−1)* Real-time PCR (CT)Estimated (CFU g−1)†
16 h20 h16 h20 h
  1. *The error for the CFU counts was c.±25%.

  2. †The estimates were made from the real-time PCR results.

MF 1821012723101101·1
1022420101·9102
MF 2281012824100·7100·9
1022420101·9102
MF 8301012722101101·4
1022419101·9102·3
MF 12641012822100·7101·4
1022419101·9102·3
MF 13451012823100·7101·1
1022520101·6102
Mean ± SD10127·6 ± 0·519·6 ± 0·5100·8±0·2101·2±0·2
Mean ± SD10224·3 ± 0·422·8 ± 0·8101·8±0·1102·1±0·2

Our conclusion is that the combination of growth and real-time PCR is a promising approach for obtaining semiquantitative information about low levels of L. monocytogenes on gouda-like cheeses. It did not seem that the background flora on the cheese samples influenced the growth of L. monocytogenes. Furthermore, there was low variation in the generation or lag times for the strains analysed. Incubation for up to 14 days on the cheese matrix prior to analysis did not affect the quantitative results.

Future perspectives on the quantitative diagnostics of L. monocytogenes

Quantitative measurements of low levels of viable L. monocytogenes will be very important in our future understanding of listeriosis. Now, we can more accurately measure contamination levels, for example, in foods suspected as sources of outbreaks. Also, as described here, novel quantitative diagnostics will help in the evaluation of foods recommended as safe for humans at risk for listeriosis. We have the possibility to rule out the presence of VBNC bacteria, which until now has been a continuous enigma in food diagnostics.

Acknowledgements

This work was supported by Tine BA and a Norwegian research levy on agricultural products.

Ancillary