APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Aug. 2010, p. 5097–5104 0099-2240/10/$12.00 doi:10.1128/AEM.00411-10 Copyright © 2010, American Society for Microbiology. All Rights Reserved.

Vol. 76, No. 15

Rapid Quantification of Viable Campylobacter Bacteria on Chicken Carcasses, Using Real-Time PCR and Propidium Monoazide Treatment, as a Tool for Quantitative Risk Assessment䌤† M. H. Josefsen,1* C. Lo ¨fstro ¨m,1 T. B. Hansen,1 L. S. Christensen,1 J. E. Olsen,2 and J. Hoorfar1 National Food Institute, Technical University of Denmark, Mørkhøj Bygade 19, 2860 Søborg, Denmark,1 and Department of Veterinary Disease Biology, Faculty of Life Sciences, University of Copenhagen, Stigbøjlen 4, 1870 Frederiksberg C, Denmark2 Received 15 February 2010/Accepted 6 June 2010

A number of intervention strategies against Campylobacter-contaminated poultry focus on postslaughter reduction of the number of cells, emphasizing the need for rapid and reliable quantitative detection of only viable Campylobacter bacteria. We present a new and rapid quantitative approach to the enumeration of food-borne Campylobacter bacteria that combines real-time quantitative PCR (Q-PCR) with simple propidium monoazide (PMA) sample treatment. In less than 3 h, this method generates a signal from only viable and viable but nonculturable (VBNC) Campylobacter bacteria with an intact membrane. The method’s performance was evaluated by assessing the contributions to variability by individual chicken carcass rinse matrices, species of Campylobacter, and differences in efficiency of DNA extraction with differing cell inputs. The method was compared with culture-based enumeration on 50 naturally infected chickens. The cell contents correlated with cycle threshold (CT) values (R2 ⴝ 0.993), with a quantification range of 1 ⴛ 102 to 1 ⴛ 107 CFU/ml. The correlation between the Campylobacter counts obtained by PMA-PCR and culture on naturally contaminated chickens was high (R2 ⴝ 0.844). The amplification efficiency of the Q-PCR method was not affected by the chicken rinse matrix or by the species of Campylobacter. No Q-PCR signals were obtained from artificially inoculated chicken rinse when PMA sample treatment was applied. In conclusion, this study presents a rapid tool for producing reliable quantitative data on viable Campylobacter bacteria in chicken carcass rinse. The proposed method does not detect DNA from dead Campylobacter bacteria but recognizes the infectious potential of the VBNC state and is thereby able to assess the effect of control strategies and provide trustworthy data for risk assessment.

infections, have been published (11, 12, 15, 38, 46). However, since control strategies mostly focus on reduction of the number of bacterial cells on the chicken carcass, the usefulness of these Q-PCR methods for risk assessment could be limited, since they detect all of the Campylobacter bacteria present in a sample, including the dead cells. The Q-PCR method described in the present study quantifies the three major food-borne Campylobacter species (C. jejuni, C. coli, and C. lari), thereby covering all possible prevalence shifts and coinfections. The PCR assay was previously validated according to the Nordic Organization for Validation of Alternative Microbiological Methods (NordVal) and is certified for detection of Campylobacter bacteria in chickens, cloacal swabs, and boot swabs (7). The present study concerns its suitability for the quantification of Campylobacter bacteria in chicken carcass rinse. Furthermore, a propidium monoazide (PMA) sample treatment step has been incorporated into the method (PMA-PCR), ensuring the quantification of only viable cells with intact membranes. PMA can intercalate into the double-helical DNA available from dead cells with compromised membranes, and upon extensive visible light exposure, cross-linking of the two strands of DNA occurs, leaving it unavailable for PCR amplification (30). PMA is a chemical alteration (additional azide group) of propidium iodide (PI), one of the most frequently applied non-membrane-permeating dyes in flow cytometry, and it can be expected to have the same permeating potential as PI (29). This could be of value from a food safety perspective, since PI penetrates only permeabilized

As Campylobacter remains the leading cause of food-borne bacterial gastrointestinal disease in large parts of the developed world (34), much effort is devoted to improving the detection and elimination of the pathogen, especially in poultry. The ultimate goal is to supply consumers with fresh, Campylobacter-free poultry products, but in order to achieve that goal, it is important to gain more insight into the epidemiology of Campylobacter, to make quantitative risk assessments, and to improve control and intervention strategies. Traditional culture-based detection of Campylobacter bacteria, including enrichment, isolation, and confirmation, is a time-consuming procedure requiring 5 to 6 working days (4, 14). Furthermore, bacterial cells may enter a viable but nonculturable (VBNC) state in which they may have the potential to cause human infection (37) but are not detected by the culture method. The introduction of real-time quantitative PCR (Q-PCR) has enabled faster, more sensitive, and less labor-intensive quantitative detection. Q-PCR methods for food-borne Campylobacter jejuni and C. coli in poultry, which is recognized as an important source of human Campylobacter

* Corresponding author. Mailing address: National Food Institute, Technical University of Denmark, Mørkhøj Bygade 19, 2860 Søborg, Denmark. Phone: 45-22844734. Fax: 45-72346001. E-mail: mhjo@food .dtu.dk. † Supplemental material for this article may be found at http://aem .asm.org/. 䌤 Published ahead of print on 18 June 2010. 5097

5098

JOSEFSEN ET AL.

cells and not cells with intact membranes (including the Campylobacter VBNC state), which can still cause infection. Nocker et al. demonstrated that no uptake of PMA occurred in bacterial cells with intact membranes, and PMA was exclusively found in bacteria with compromised membranes (31). PMA sample treatment combined with real-time PCR for detection of viable pathogens has been tested successfully on Listeria monocytogenes and Escherichia coli O157:H7 (31, 36). However, these studies were limited to laboratory-cultured strains and the methods have not been validated on naturally infected samples with the pathogen embedded in a food matrix. This is the first study to establish a correlation between results obtained by PMA-PCR and culture-based enumeration of Campylobacter bacteria for a large number of naturally infected chickens.

MATERIALS AND METHODS Experimental design. (i) Optimization and investigation of the quantitative aspect. Preliminary experiments were performed to determine the optimal volume (50, 100, or 500 ml) for rinsing of chicken carcasses, the optimal volume to draw from the chicken carcass rinse (1, 5, or 10 ml) for DNA extraction, if DNA extraction is needed or if Q-PCR can be performed directly on chicken carcass rinse without preceding DNA extraction, and the optimal DNA template volume (5 or 10 ␮l) for Q-PCR. Furthermore, the recovery of Campylobacter bacteria with the rinsing procedure was assessed. First, two chickens determined to be Campylobacter positive by culture were rinsed four times in succession as described below. From each chicken carcass rinse, 10-fold serial dilutions ranging from 100 to 10⫺2 were prepared, and the number of Campylobacter bacteria was determined in duplicate by culture as described below. Second, eight Campylobacter-free chickens were purchased from local retailers and divided into 16 halves. For eight of the halves, the Campylobacter-negative status of the chicken was confirmed by culture of the chicken carcass rinse as described below. The other eight halves were inoculated with 5 ⫻ 102, 5 ⫻ 103, 5 ⫻ 104, and 5 ⫻ 105 CFU C. jejuni (CCUG 11284). The inoculation was performed by placing 10-␮l droplets of bacterial suspension at 25 equally distributed positions on the chicken skin surface. To ensure adherence of the bacterial cells to the chicken skin, the chicken was left unwrapped for 1 h at room temperature, followed by 24 h at 5°C wrapped in a sterile plastic bag. Subsequently, the number of Campylobacter bacteria on the inoculated chickens was determined by culture of the chicken carcass rinse as described below. (ii) Variation attributable to the matrix. A prerequisite for successful quantification of Campylobacter bacteria is to describe the natural variation within the matrix in question. Quantification would be encumbered with uncertainties if large natural variation were observed in the chicken carcass rinse matrix. In order to evaluate this variation, 12 Campylobacter-free chickens of mixed origins were purchased from local retailers. They were rinsed as described below, and the rinse was inoculated with 10-fold dilutions ranging from 1 ⫻ 102 to 1 ⫻ 106 CFU/ml C. jejuni and C. coli, respectively, and subsequent DNA extraction and Q-PCR were performed as described below. (iii) Variation attributable to the individual species. The Q-PCR assay described in this study amplifies the three food-borne species of Campylobacter, C. jejuni, C. coli, and C. lari. In order to quantify all of the species simultaneously without further differentiation, the same sensitivity and amplification efficiency (AE) for all species had to be proven. Hence, standard dilution series made from bacterial cells of the three species were investigated by Q-PCR analysis. C. jejuni (CCUG 11284), C. coli (CCUG 11283), and C. lari (CCUG 23947) were recovered on five blood agar plates each. For each species, the growth was transferred to approximately 10 ml of saline (0.9% NaCl) and the suspension was mixed thoroughly to ensure a homogeneous suspension of bacterial cells. Tenfold serial dilutions were prepared for each species ranging from 10⫺1 to 10⫺9 CFU/ml. The number of bacterial cells in each suspension was determined by duplicate plate spreading on blood agar. Standard rows (ranging from 1 to 1 ⫻ 106 CFU/ml) were produced from the three species. One-milliliter volumes were drawn from the standards, and DNA was extracted as described below and analyzed in duplicate by Q-PCR. This procedure was repeated with two biological replicates, four times for C. jejuni and two times for C. coli and C. lari, to

APPL. ENVIRON. MICROBIOL. observe the variation between different bacterial suspensions and dilution series. C. jejuni, C. coli, and C. lari have been reported to have three copies of the 16S rRNA gene (21, 43; see also the NCBI Microbial Genomes database [http://www .ncbi.nlm.nih.gov/genomes/MICROBES/microbial_taxfree.html]) which is the target of the primers used in the Q-PCR; therefore, an equimolar ratio of the three species could be assumed. (iv) Real-time PCR signal reduction by PMA. Four series of Campylobacterfree chicken rinse were inoculated with C. jejuni CCUG 11284 to obtain samples containing 1 ⫻ 102, 1 ⫻ 103, 1 ⫻ 104, 1 ⫻ 105, and 1 ⫻ 106 CFU/ml. The Campylobacter cells were subjected to lethal heat treatment at 95°C for 5 min, and PMA sample treatment was performed in two of the series. The absence of viable Campylobacter bacteria in the heat-treated samples was confirmed by culture as described below. All samples were analyzed in duplicate by real-time PCR to investigate the signal reduction capacity of PMA sample treatment. (v) Quantification of viable Campylobacter bacteria in naturally infected chickens. The number of Campylobacter bacteria on 50 chickens from a confirmed Campylobacter-positive flock was quantified in parallel by Q-PCR (with and without PMA sample treatment) and by conventional culture. The chickens were collected immediately after cooling at the abattoir, packed in sterile plastic bags, and kept at 5°C until analyzed (⬍18 h). The chickens were rinsed as described below. Duplicate chicken carcass rinse volumes, with and without PMA sample treatment prior to DNA extraction, were analyzed by Q-PCR in duplicate as described below. From parallel duplicate chicken carcass rinse volumes, 10-fold serial dilutions ranging from 100 to 10⫺2 were prepared and the number of Campylobacter bacteria was determined in duplicate by culture as described below. Bacterial strains and culture conditions. Campylobacter strains were stored at ⫺80°C in LB medium (Statens Serum Institute [SSI], Copenhagen, Denmark) containing 15% glycerol as a cryoprotectant. They were recovered on blood agar (SSI) and isolated on selective solid medium, modified charcoal cefoperazone deoxycholate agar (mCCDA; Oxoid, Greve, Denmark) and Abeyta-Hunt-Bark agar (AHB; Technical University of Denmark, Copenhagen, Denmark). Mueller-Hinton broth (SSI) was used to produce overnight cultures for preparation of the quantification standards and for artificial inoculation of chicken carcass rinse. Incubation was at all times performed at 41.5 ⫾ 1°C under microaerobic conditions (6% O2, 7% CO2, 7% H2, and 80% N2). Chicken carcass rinse. Chicken carcass rinse was prepared according to the recommendations in ISO 6887-2 (2). A whole fresh chicken was placed in a sterile plastic bag and rinsed manually in 50 ml of saline (0.9% NaCl) for 1 min. One-milliliter aliquots were drawn for DNA extraction and subsequent Q-PCR analysis and for culture. Culture-based enumeration. From the chicken carcass rinse, duplicate 10-fold serial dilutions ranging from 100 to 10⫺2 were prepared and 100 ␮l of each dilution was spread in duplicate onto mCCDA and AHB and incubated for 48 h. The agar plates were dried for 30 min prior to use to avoid swarming of the colonies. Two selective agar plates were used, according to the recommendations of ISO 10272-1. Five presumptive Campylobacter colonies per chicken carcass rinse were verified by a validated colony PCR method (23) by dissolving a minimal amount of colony material in 100 ␮l of saline (0.9% NaCl) and analyzing 10 ␮l of this suspension by Q-PCR with a thermal profile including 10 min of primary denaturation to ensure cell wall disruption and DNA accessibility. PMA treatment of samples. PMA (Biotium Inc., Hayward, CA) dissolved in 20% dimethyl sulfoxide (Sigma-Aldrich, Brøndby, Denmark) was added to 1-ml volumes of chicken carcass rinse to a final concentration of 10 ␮g/ml. These were incubated in clear Eppendorf tubes in the dark for 5 min and inverted repeatedly at two points during this period. Following incubation, the Eppendorf tubes were placed on ice and exposed to a 650-W halogen light source (Kaiser Videolight 6; Kaiser Fototechnik, Buchen, Germany) at a distance of 20 cm for 1 min. The Eppendorf tubes were swirled briefly by hand every 15 s and turned over after 30 s of illumination to ensure complete cross-linking of the available DNA and the conversion of free PMA to hydroxylamino propidium. DNA extraction. One-milliliter volumes of chicken carcass rinse/standard were centrifuged at 3,000 ⫻ g for 5 min at 4°C, and DNA extraction was performed on a KingFisher (Thermo Labsystems, Helsinki, Finland) using a DNA isolation kit for blood, stool, cells, and tissue (MagneSil KF, Genomic System; Promega) as specified by the manufacturer. Briefly, the sample pellet was resuspended in 200 ␮l lysis buffer and transferred to wells A, B, and C of a 96-well plate (Thermo Labsystems) containing paramagnetic particles (90 ␮l divided among wells A, B, and C), salt washing buffer (200 ␮l divided between wells D and E), alcohol wash (200 ␮l divided between wells F and G), and 100 ␮l of elution buffer (well H). The DNA extraction program (Genomic_DNA_1) was performed. Ten microliters of the extracted DNA was used as the template in the Q-PCR. In every

VOL. 76, 2010

REAL-TIME PCR QUANTIFICATION OF VIABLE CAMPYLOBACTERS

5099

FIG. 1. Standard curves produced from 10-fold serial dilutions of C. jejuni (CCUG 11284) and C. coli (CCUG 11283) cells ranging from 1 ⫻ 102 to 1 ⫻ 106 CFU/ml chicken carcass rinse in 12 different chicken carcass rinses, showing the variation that can be ascribed to the matrix. Each data point represents three biological and six real-time Q-PCR replicates, with the standard deviation illustrated.

96-well plate (corresponding to 12 DNA extractions), a process control consisting of only DNA extraction reagents was included. Q-PCR analysis. A TaqMan Q-PCR method for the specific detection of food-borne thermotolerant Campylobacter based on the amplification and detection of a 287-bp sequence of the 16S rRNA gene of C. jejuni, C. coli, and C. lari was performed on an Mx3005P (Stratagene, La Jolla, CA) as described previously (18), with the following modifications: 0.6 mM deoxynucleoside triphosphate mixture (Applied Biosystems, Naerum, Denmark), 0.5 ␮M forward and reverse primers, 0.2 g/liter bovine serum albumin (Roche A/S), 0.8 ml/liter glycerol (87%; Merck, Darmstadt, Germany), 75 nM target locked nucleic acid Campylobacter probe 5⬘ [6FAM] CA[⫹T] CC[⫹T] CCA CGC GGC G[⫹T]T GC [BHQ1] 3⬘ (Sigma-Aldrich), 60 nM internal amplification control probe, and 10 ␮l of extracted template DNA. The cycle profile was as follows: initial denaturation at 95°C for 3 min (10 min for colony PCR), followed by 40 cycles of 95°C for 15 s, 60°C for 60 s, and 72°C for 30 s. Fluorescence measurements were obtained online and analyzed with the MxPro-Mx3005P software (version 4.10). The threshold was assigned as described earlier (19). In every Q-PCR analysis, the C. jejuni standard for absolute quantification was included in duplicate. Furthermore, a nontemplate control containing PCRgrade water, a negative DNA control (5 ng of E. coli DNA), and a positive DNA control (5 ng of C. jejuni DNA) were included. Standard curve for absolute quantification. C. jejuni (CCUG 11284) was recovered on blood agar (SSI). A loop full of colony material was transferred to 10 ml of Mueller-Hinton broth and incubated for approximately 16 h. From this suspension, a 10-fold serial dilution was prepared and bacteria were enumerated by spread plating onto blood agar in triplicate. One-milliliter volumes of Campylobacter-free chicken carcass rinse were inoculated with 1 ⫻ 102 to 1 ⫻ 107 CFU C. jejuni (CCUG 11284) from the appropriate dilution, and the DNA was extracted from these as described above. Three biological and six Q-PCR replicates were used to produce the standard curve. Data analysis. (i) Variation attributable to matrix and species. The variation between the standard curves produced in these experiments was evaluated by linear regression (Microsoft Excel 2000). For each standard curve, the number of CFU per milliliter was log10 transformed and defined as the independent variable. The ⌬CT value was defined as the dependent variable. The 95% confidence intervals of estimates of slope and intercept, respectively, were used for the comparison of standard curves. Standard curves with overlapping 95% confidence intervals were not regarded as statistically significantly different. (ii) Standard curve. The standard curve was produced by plotting the CT value obtained in a Q-PCR against the number of CFU (log CFU) in the standards. From this linear relationship, the AE was calculated as follows: AE ⫽ 10(⫺1/slope) ⫺ 1 (22).

(iii) Quantification of viable Campylobacter bacteria in chickens. The correlation between the results obtained by Q-PCR (with and without PMA sample treatment) and those obtained by culture-based enumeration was evaluated by plotting the log10-transformed number of Campylobacter cell equivalents (CCE) per milliliter obtained from the Q-PCR against the log10-transformed number of CFU per milliliter obtained by duplicate spread plating on mCCDA. For comparison of the abilities of Q-PCRs with and without PMA sample treatment to predict CFU counts on mCCDA, the difference between the log10-transformed bacterial numbers was used as a response variable in a single-factor analysis of variance (Microsoft Excel 2000) comparing Q-PCR with PMA sample treatment to Q-PCR without PMA sample treatment. A P value below 0.05 was considered statistically significant.

RESULTS Optimization. Optimal method performance was obtained when a rinse volume of 50 ml, a sampling volume of 1 ml, and a template DNA volume of 10 ␮l were used (data not shown). Furthermore, it was found necessary to include the DNA extraction step. Recovery of Campylobacter bacteria by rinsing. The rinsing of two chicken carcasses in 50 ml of saline showed that the second rinsing procedure yielded 60% of the amount of Campylobacter bacteria obtained by the first. Performing two subsequent rinsing procedures, 45% and 27% of the initial amount of Campylobacter bacteria were recovered. The recovery of Campylobacter bacteria from eight artificially inoculated chicken carcasses ranged from 55% to 94% and was determined to be 77%, on average. Variation attributable to chicken carcass rinse and Campylobacter species. ⌬CT values were used to produce the standard curves shown in Fig. 1 and 2 (40). To compensate for the variability in the input cell number of the three different species of Campylobacter in the chicken rinses, the CT value of each dilution was subtracted from the CT value obtained for the highest number of input cells. Therefore, the true variation

5100

JOSEFSEN ET AL.

APPL. ENVIRON. MICROBIOL.

FIG. 2. Standard curves produced from 10-fold serial dilutions of C. jejuni (CCUG 11284) (n ⫽ 4), C. coli (CCUG 11283) (n ⫽ 2), and C. lari (CCUG 23947) (n ⫽ 2) ranging from 1 ⫻ 101 to 1 ⫻ 106 CFU/ml showing the variation that can be ascribed to the species at the cell level. Each data point represents two biological and two Q-PCR replicates with the standard deviation illustrated.

between the different chicken rinses and the three species of Campylobacter was not obscured by the inevitable variation in input cell number. As shown in Fig. 1, the contribution to variation from the chicken carcass rinse matrix, based on 15 standard curves of Campylobacter bacteria, in individual chicken carcass rinses was negligible. Statistical comparison of slopes and intercepts of standard curves based on individual chicken rinses showed no difference (P ⬎ 0.05). No difference was observed among standard curves based on the three species, and correspondingly, no significant variance was measured between the standard curves (P ⬎ 0.05), proving equal sensitivity and AE of the method independent of the species (Fig. 2). Real-time PCR signal reduction by PMA. The CT values obtained by real-time PCR for the 10-fold dilution series of heat-treated C. jejuni in chicken rinse (102 to 106 CFU/ml) with and without PMA sample treatment are shown in Fig. 3. No viable Campylobacter bacteria were detectable by culture after heat treatment. The signal reduction for PMA-treated samples was 100%; correspondingly, no signals were obtained from the PMA-treated samples by real-time PCR. Standard curve for absolute quantification. The standard curve produced from DNA extracted from 10-fold cell dilutions in chicken carcass rinse showed an AE of 91%, computed from the slope of the linear relationship between the log10transformed number of CFU per milliliter and the CT value (R2 ⫽ 0.993). The method was shown to be linear over a range of 1 ⫻ 102 to 1 ⫻ 107 CFU/ml chicken carcass rinse, and the limit of quantification was 1 ⫻ 102 CFU/ml. Taking into consideration the volume reduction during DNA extraction, this corresponds to 10 CFU/PCR. From the standard curve, the level of Campylobacter bacteria was expressed by the equation log CFU ⫽ (CT ⫺ 40.8)/⫺3.571. Variation in culture-based enumeration. Two parallel dilutions of chicken carcass rinse were spread plated onto selective agar (mCCDA and AHB) to evaluate the contribution of culture enumeration to the overall variation of the method. The

linear relationship between the replicate culture series was expressed by the equations y ⫽ 1.0004x for mCCDA (R2 ⫽ 0.969) and y ⫽ 1.0154x for AHB (R2 ⫽ 0.935), indicating extremely low variation in the quantitative results obtained by culture. Quantification of viable Campylobacter bacteria in naturally infected chickens. The number of Campylobacter bacteria on 50 chickens sampled postslaughter from a confirmed Campylobacter-positive flock was quantified in parallel by Q-PCR (with and without PMA sample treatment) and conventional culture (Fig. 4; see Table S1 in the supplemental material). By culture, 42 chickens were found to be Campylobacter positive, while this number was 45 for Q-PCR with PMA sample treatment and 48 without PMA sample treatment. Eight culture-negative chickens were found to be Campylobacter positive by Q-PCR; however, the CT value obtained was not within the linear range of the assay and quantification was therefore not possible. The same applied for the four chickens found to be Campylobacter positive by Q-PCR without PMA sample treatment but Campylobacter negative with PMA sample treatment. The CT values obtained from the Campylobacter-positive chicken carcass rinses ranged from 17.6 to 39, corresponding to contamination levels of 25 to 1.5 ⫻ 106 CFU/ml, reflecting a large contamination level variation within the flock. In 36 chicken carcass rinses, a CT value within the linear range of the method was obtained and quantification of Campylobacter bacteria was possible. Except for chickens 1 and 5, the PMA-treated samples gave a reduced signal in Q-PCR. A reduction range of 1 ⫻ 102 to 2.4 ⫻ 106 CCE/ml was observed in PMA-treated samples, and the reduction was positively correlated with the amount of Campylobacter bacteria in the samples. This is also reflected in the obvious linear correlation between the Q-PCR results obtained from both PMA-treated and untreated samples and enumeration by culture (Fig. 5). The data analysis of the culture-based enumeration was based on the average obtained

VOL. 76, 2010

REAL-TIME PCR QUANTIFICATION OF VIABLE CAMPYLOBACTERS

5101

FIG. 3. CT values obtained by real-time PCR on 10-fold dilution series of C. jejuni CCUG 11284 in chicken rinse (102 to 106 CFU/ml) subjected to lethal heat treatment with and without PMA treatment. The curves below the threshold line represent a duplicate analysis of samples that received PMA treatment prior to PCR analysis.

from the duplicate mCCDA plates, since a better correlation was obtained with these than with the AHB plates (data not shown). Analysis of variance comparing Q-PCR (with and without PMA sample treatment) with plate counts on mCCDA showed that Q-PCR with PMA sample treatment produced results that were statistically significantly closer to counts on mCCDA (P ⬍ 0.001) (Fig. 4). DISCUSSION As many advantages as the Q-PCR technology offers regarding pathogen detection in mixed populations, a major drawback has been that PCR detection, operating on a DNA level, cannot distinguish between DNA from viable bacterial cells and that from dead bacterial cells. This is a limitation of PCR technology which is of particular relevance for Campylobacter, since control strategies are focusing increasingly on decontamination, leaving PCR technology unable to assess the potential for food-borne infections. In 2006, European Community regulation 853/2004 permitted the use of substances other than water for decontamination of meat surfaces (3). Both physical and chemical reduction strategies, including treatment with steam, UV light, high hydrostatic pressure, essential oil frac-

tions, acid, and sodium hypochlorite, have been investigated for their Campylobacter reduction potential, with various outcomes (1, 9, 10, 13, 16, 24, 28, 33, 35). DNA from dead bacterial cells has been shown to persist for up to 3 weeks following cell death (20), and it has also been demonstrated that the presence of DNA from dead Campylobacter cells can lead to an overestimation of the number of viable cells and, in some instances, even to false-positive results (45). In the present study, the technique successfully distinguished between viable and dead cells. Although PMA-PCR is claimed to detect “viable” cells (30, 32, 41), viability can be defined in several ways: as having an intact membrane, the ability to metabolize compounds, and/or the ability to maintain a proton gradient between the inner and outer parts of the cell. To further distinguish between these states, the PMA-PCR method can be supplemented with other compounds in the future, e.g., to show the presence of an intact metabolism (31). Despite this uncertainty, PMA-PCR quantification compared favorably with direct culture-based detection of Campylobacter bacteria in this study. The relative specificity of the PMA-PCR method was 100%, and it was shown to be more sensitive than the culture-based method. Eight chickens were found to be

5102

JOSEFSEN ET AL.

APPL. ENVIRON. MICROBIOL.

FIG. 4. Results of Q-PCR, Q-PCR with PMA sample treatment, and culture of naturally infected chickens. The diagram presents the data obtained from the 36 chicken carcass rinses for which a CT value was obtained within the linear range of the Q-PCR method.

Campylobacter positive by Q-PCR but not by culture. According to ISO 20838, these can be regarded as true positives due to the target-specific DNA probe-based PCR response (5). Besides being labor-intensive and time-consuming, culturebased quantification will not detect the VBNC fraction of a given Campylobacter population. The results obtained in the present study suggest that VBNC Campylobacter bacteria were not present in great numbers on the chicken carcasses sampled immediately postslaughter. However, the number of VBNC cells can only be expected to increase as a consequence of food processing and storage, thus presenting a possible diagnostic uncertainty. Furthermore, postslaughter control strategies to reduce the number of viable Campylobacter bacteria on chicken carcasses will contribute to the number of VBNC cells. As long as the infectious potential of the Campylobacter VBNC state is not clarified, quantitative methods for risk assessment should detect these as well. The good correlation demonstrated in this study between counts obtained by Q-PCR with PMA sample treatment and counts obtained by culture-based enumeration enables calculation of the amount of Campylobacter bacteria in naturally infected chicken carcass rinse from a CT value obtained by Q-PCR. It has been demonstrated in several studies comparing Q-PCR with culture-based enumeration that higher counts are produced by Q-PCR, which has been explained by the detection of DNA from dead and VBNC cells (11, 15, 46). This was also the case for the untreated samples in the present study; however, when PMA sample treatment was applied, a Q-PCRbased count lower than the culture-based count was often observed. The reason for this could partly be due to underestimation of the cell input in the standard applied for quantification but also possibly due to overestimation of the number of Campylobacter colonies on the mCCDA plates. Five presumptive Campylobacter colonies from each chicken carcass rinse were subcultured to nonselective medium and subsequently verified by colony PCR. The results indicated that up to 10% of the Campylobacter-like colonies on the mCCDA plates could not be confirmed to be C. jejuni, C. coli, or C. lari, substantiating this theory. Another issue that has to be considered in this regard is that the cell state, and the permeability of the cell

wall, is not a clear-cut reflection of whether a cell is viable or dead, and PMA could have entered a minor fraction of the culturable cells. The correlation between culture-based enumeration and QPCR was surprisingly high in the present study, whereas it would have been expected to be inferior to the correlation with Q-PCR with PMA sample treatment applied. Sampling of chicken carcasses immediately postslaughter is suspected to explain this finding. All of the chicken carcasses tested had been exposed to the exact same conditions down the slaughter line, which was likely to result in equivalent proportions of Campylobacter cells with compromised membranes on the carcasses. A drawback of the present study is that the quantification limit of the method is 100 CFU/ml of chicken carcass rinse, which does not meet the legal requirement of detection of 1 CFU/25 g (4). Available technologies do not, however, enable separation and concentration of the target organism from the food matrix, resulting in the necessity of a certain level of Campylobacter bacteria before direct detection by Q-PCR (and culture) is possible. The results from the preliminary rinsing experiments also showed that the small rinsing volume of 50 ml failed to recover all of the Campylobacter cells present on the chicken carcasses. The rinsing of artificially inoculated chickens resulted in an average recovery of 77%. Despite efforts to simulate natural infection, it is likely that recovery from naturally infected chickens would be lower due to cells’ being more firmly attached or located within the follicles of the skin. A rinsing volume of 310 ml was used in a study by Jørgensen et al., yet recovery rates similar to those in the present study were observed; 49% and 25% of the yield from the primary rinse were found in the second and third rinsing procedures, respectively (17). These findings have to be considered in estimating the whole carcass contamination level when employing the PMA-PCR method for enumeration of naturally contaminated chicken carcasses. Immunocapture of Campylobacter bacteria prior to PCR, reducing sample volumes from 250 ml to only 200 ␮l, has been conducted but showed poor recovery rates (27). Wolffs et al. described the use of flotation to single out viable and VBNC

VOL. 76, 2010

REAL-TIME PCR QUANTIFICATION OF VIABLE CAMPYLOBACTERS

5103

FIG. 5. Correlation of Campylobacter counts obtained by Q-PCR with (A) and without (B) PMA sample treatment and culture on naturally contaminated chicken carcass rinse samples. The linear regression equations and R2 values are shown in the diagrams.

Campylobacter bacteria prior to Q-PCR. The limit of detection of this flotation-based method was determined to be 8.6 ⫻ 102 CFU/ml, and quantification was possible over a range of 2.6 ⫻ 103 to 2.6 ⫻ 107 CFU/ml of chicken carcass rinse (44). Direct Q-PCR methods for Campylobacter bacteria are applicable as tools for risk assessment and assurance of food safety, as it has been shown that a strong positive correlation exists between the number of Campylobacter bacteria on chickens and the risk of human infection (26, 42). Rosenquist et al. reported that a 2-log reduction in the number of Campylobacter bacteria on chickens can lead to a 30 times lower risk of human infection (39). Based on these recent findings, it is feasible that future legislation will aim at a maximum limit for Campylobacter cell counts, as has been done for other pathogens, such as L. monocytogenes. The advantages of the PCR assay used in this study are its thorough validation in previous comparative and collaborative trials (6) and approval for detection of Campylobacter bacteria in chickens, cloacal swabs, and boot swabs (23). The inclusivity and exclusivity of the primers have been determined for 115 target and 87 nontarget strains to 100% and 97%, respectively (25). The AE, linear range, detection probability, and detection precision of the present PCR assay have been evaluated, and its suitability for quantitative analysis has been confirmed (18). It includes an internal amplification control to avoid false-negative responses and reveal PCR inhibition. Finally,

the standard dilution series for calculation of the level of Campylobacter bacteria was produced from 10-fold cell dilutions (not DNA) in the relevant matrix, from which DNA was subsequently extracted and analyzed by Q-PCR, taking into account that efficiency of DNA extraction can vary substantially with the initial amount of cells (19). In conclusion, this study presents a tool for quantitative detection of food-borne viable Campylobacter bacteria, as measured by membrane integrity, that can be applied to produce accurate and reliable data for risk assessments in chicken carcass rinse. The Q-PCR method recognizes the infectious potential of the VBNC state and is thereby able to assess the outcome and impact of new control strategies. It is currently under consideration in Europe for implementation by a number of reference and research laboratories that provide data for epidemiological studies and risk assessments. ACKNOWLEDGMENTS This work was supported in part by the European Union-funded Integrated Project BIOTRACER (contract 036272) under the 6th RTD Framework. We thank Liselotte Folling for excellent technical assistance and proofreading of the experimental plans and Lantma¨nnen Danpo A/S for kindly providing the samples and for sampling assistance. REFERENCES 1. Anderson, R. C., R. B. Harvey, J. A. Byrd, T. R. Callaway, K. J. Genovese, T. S. Edrington, Y. S. Jung, J. L. McReynolds, and D. J. Nisbet. 2005. Novel

5104

2.

3.

4.

5.

6.

7. 8. 9.

10.

11.

12.

13.

14.

15.

16. 17.

18.

19.

20.

21.

22.

23.

JOSEFSEN ET AL.

preharvest strategies involving the use of experimental chlorate preparations and nitro-based compounds to prevent colonization of food-producing animals by foodborne pathogens. Poult. Sci. 84:649–654. Anonymous. 2000. Microbiology of food and animal feeding stuffs. Preparation of test samples, initial suspension and decimal dilution for microbiological examination. Part 2. Specific rules for the preparation of the initial suspension and decimal dilutions of meat and meat products. ISO/CD 6887-2. Association Franc¸aise de Normalisation, Paris, France. Anonymous. 2004. Regulation (EC) no. 853/2004 of the European Parliament and of the Council of 29 April 2004 laying down specific hygiene rules for food of animal origin. Official J. Eur. Union 47:22–82. Anonymous. 2006. Microbiology of food and animal feeding stuffs—horizontal method for detection and enumeration of Campylobacter spp. Part 1: detection method. ISO 10272-1. International Organization for Standardization, Geneva, Switzerland. Anonymous. 2006. Microbiology of food and animal feeding stuffs—polymerase chain reaction (PCR) for the detection of food borne pathogens— requirements for amplification and detection for qualitative methods. ISO 20838. International Organization for Standardization, Geneva, Switzerland. Anonymous. 2009. Protocol for the validation of alternative microbiological methods. Nordic Committee on Food Analysis (NMKL), NordVal, Oslo, Norway. Anonymous. 30 March 2009. NordVal certificate Campylobacter real-time PCR. http://www.nmkl.org/NordVal/Sertifikater/NordVal017.pdf. Reference deleted. Bauermeister, L. J., J. W. Bowers, J. C. Townsend, and S. R. McKee. 2008. Validating the efficacy of peracetic acid mixture as an antimicrobial in poultry chillers. J. Food Prot. 71:1119–1122. Bie`che, C., M. Ritz, O. Tresse, M. Federighi, and M. de Lamballerie. 2009. Impacts of treatment parameters on the inactivation of Campylobacter jejuni by high pressure: a statistical study of main effects and interactions. Lett. Appl. Microbiol. 48:198–202. Botteldoorn, N., E. Van Coillie, V. Piessens, G. Rasschaert, L. Debruyne, M. Heyndrickx, L. Herman, and W. Messens. 2008. Quantification of Campylobacter spp. in chicken carcass rinse by real-time PCR. J. Appl. Microbiol. 105:1909–1918. Debretsion, A., T. Habtemariam, S. Wilson, D. Nganwa, and T. Yehualaeshet. 2007. Real-time PCR assay for rapid detection and quantification of Campylobacter jejuni on chicken rinses from poultry processing plant. Mol. Cell. Probes 21:177–181. Fisher, K., and C. A. Phillips. 2006. The effect of lemon, orange and bergamot essential oils and their components on the survival of Campylobacter jejuni, Escherichia coli O157, Listeria monocytogenes, Bacillus cereus and Staphylococcus aureus in vitro and in food systems. J. Appl. Microbiol. 101:1232–1240. Fitzgerald, C., J. Whichard, and I. Nachamkin. 2008. Diagnosis and antimicrobial susceptibility of Campylobacter species, p. 227–243. In I. Nachamkin, C. M. Szymanski, and M. J. Blaser (ed.), Campylobacter, 3rd ed. ASM Press, Washington, DC. Hong, J., W. K. Jung, J. M. Kim, S. H. Kim, H. C. Koo, J. Ser, and Y. H. Park. 2007. Quantification and differentiation of Campylobacter jejuni and Campylobacter coli in raw chicken meats using a real-time PCR method. J. Food Prot. 70:2015–2022. Isohanni, P. M., and U. Lyhs. 2009. Use of ultraviolet irradiation to reduce Campylobacter jejuni on broiler meat. Poult. Sci. 88:661–668. Jørgensen, F., R. Bailey, S. Williams, P. Henderson, D. R. Wareing, F. J. Bolton, J. A. Frost, L. Ward, and T. J. Humphrey. 2002. Prevalence and numbers of Salmonella and Campylobacter spp. on raw, whole chickens in relation to sampling methods. Int. J. Food Microbiol. 76:151–164. Josefsen, M. H., N. R. Jacobsen, and J. Hoorfar. 2004. Enrichment followed by quantitative PCR both for rapid detection and as a tool for quantitative risk assessment of food-borne thermotolerant campylobacters. Appl. Environ. Microbiol. 70:3588–3592. Josefsen, M. H., M. Krause, F. Hansen, and J. Hoorfar. 2007. Optimization of a 12-hour TaqMan PCR-based method for detection of Salmonella bacteria in meat. Appl. Environ. Microbiol. 73:3040–3048. Josephson, K. L., C. P. Gerba, and I. L. Pepper. 1993. Polymerase chain reaction detection of nonviable bacterial pathogens. Appl. Environ. Microbiol. 59:3513–3515. Kim, N. W., R. Lombard, H. Bingham, E. Hani, H. Louie, D. Ng, and V. L. Chan. 1993. Fine mapping of the three rRNA operons on the updated genomic map of Campylobacter jejuni TGH9011 (ATCC 43431). J. Bacteriol. 175:7468–7470. Klein, D., P. Janda, R. Steinborn, M. Muller, B. Salmons, and W. H. Gunzburg. 1999. Proviral load determination of different feline immunodeficiency virus isolates using real-time polymerase chain reaction: influence of mismatches on quantification. Electrophoresis 20:291–299. Krause, M., M. H. Josefsen, M. Lund, N. R. Jacobsen, L. Brorsen, M. Moos, A. Stockmarr, and J. Hoorfar. 2006. Comparative, collaborative, and on-site

APPL. ENVIRON. MICROBIOL.

24.

25.

26.

27. 28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42. 43. 44.

45.

46.

validation of a TaqMan PCR method as a tool for certified production of fresh, Campylobacter-free chickens. Appl. Environ. Microbiol. 72:5463–5468. Li, Y., H. Yang, and B. L. Swem. 2002. Effect of high-temperature insideoutside spray on survival of Campylobacter jejuni attached to prechill chicken carcasses. Poult. Sci. 81:1371–1377. Lu ¨beck, P. S., P. Wolffs, S. L. On, P. Ahrens, P. Radstrom, and J. Hoorfar. 2003. Toward an international standard for PCR-based detection of foodborne thermotolerant campylobacters: assay development and analytical validation. Appl. Environ. Microbiol. 69:5664–5669. Messens, W., E. Hartnett, X. Gellynck, J. Viaene, D. Halet, L. Herman, and K. Grijspeerdt. 2007. Quantitative risk assessment of human campylobacteriosis through the consumption of chicken meat in Belgium, p. 167–168. In XVIII European Symposium on the Quality of Poultry Meat and XII European Symposium on the Quality of Eggs and Egg Products, Prague, Czech Republic. Morales-Rayas, R., P. F. Wolffs, and M. W. Griffiths. 2008. Immunocapture and real-time PCR to detect Campylobacter spp. J. Food Prot. 71:2543–2547. Nannapaneni, R., V. I. Chalova, P. G. Crandall, S. C. Ricke, M. G. Johnson, and C. A. O’Bryan. 2009. Campylobacter and Arcobacter species sensitivity to commercial orange oil fractions. Int. J. Food Microbiol. 129:43–49. Nebe-von-Caron, G., P. J. Stephens, C. J. Hewitt, J. R. Powell, and R. A. Badley. 2000. Analysis of bacterial function by multi-colour fluorescence flow cytometry and single cell sorting. J. Microbiol. Methods 42:97–114. Nocker, A., and A. K. Camper. 2009. Novel approaches toward preferential detection of viable cells using nucleic acid amplification techniques. FEMS Microbiol. Lett. 291:137–142. Nocker, A., C. Y. Cheung, and A. K. Camper. 2006. Comparison of propidium monoazide with ethidium monoazide for differentiation of live vs. dead bacteria by selective removal of DNA from dead cells. J. Microbiol. Methods 67:310–320. Nogva, H. K., S. M. Dromtorp, H. Nissen, and K. Rudi. 2003. Ethidium monoazide for DNA-based differentiation of viable and dead bacteria by 5⬘-nuclease PCR. Biotechniques 34:804–813. Northcutt, J., D. Smith, K. D. Ingram, A. Hinton, Jr., and M. Musgrove. 2007. Recovery of bacteria from broiler carcasses after spray washing with acidified electrolyzed water or sodium hypochlorite solutions. Poult. Sci. 86:2239–2244. Olson, K. C., S. Ethelberg, W. Van Pelt, and V. Tauxe. 2008. Epidemiology of Campylobacter jejuni infections in industrialized nations, p. 163–189. In I. Nachamkin, C. M. Szymanski, and M. J. Blaser (ed.), Campylobacter, 3rd ed. ASM Press, Washington, DC. Oyarzabal, O. A., C. Hawk, S. F. Bilgili, C. C. Warf, and G. K. Kemp. 2004. Effects of postchill application of acidified sodium chlorite to control Campylobacter spp. and Escherichia coli on commercial broiler carcasses. J. Food Prot. 67:2288–2291. Pan, Y., and F. Breidt, Jr. 2007. Enumeration of viable Listeria monocytogenes cells by real-time PCR with propidium monoazide and ethidium monoazide in the presence of dead cells. Appl. Environ. Microbiol. 73:8028– 8031. Rollins, D. M., and R. R. Colwell. 1986. Viable but nonculturable stage of Campylobacter jejuni and its role in survival in the natural aquatic environment. Appl. Environ. Microbiol. 52:531–538. Ro ¨nner, A. C., and H. Lindmark. 2007. Quantitative detection of Campylobacter jejuni on fresh chicken carcasses by real-time PCR. J. Food Prot. 70:1373–1378. Rosenquist, H., N. L. Nielsen, H. M. Sommer, B. Norrung, and B. B. Christensen. 2003. Quantitative risk assessment of human campylobacteriosis associated with thermophilic Campylobacter species in chickens. Int. J. Food Microbiol. 83:87–103. Rudi, K., H. K. Hoidal, T. Katla, B. K. Johansen, J. Nordal, and K. S. Jakobsen. 2004. Direct real-time PCR quantification of Campylobacter jejuni in chicken fecal and cecal samples by integrated cell concentration and DNA purification. Appl. Environ. Microbiol. 70:790–797. Rudi, K., B. Moen, S. M. Dromtorp, and A. L. Holck. 2005. Use of ethidium monoazide and PCR in combination for quantification of viable and dead cells in complex samples. Appl. Environ. Microbiol. 71:1018–1024. Stern, N. J., and S. Pretanik. 2006. Counts of Campylobacter spp. on U.S. broiler carcasses. J. Food Prot. 69:1034–1039. Taylor, D. E., M. Eaton, W. Yan, and N. Chang. 1992. Genome maps of Campylobacter jejuni and Campylobacter coli. J. Bacteriol. 174:2332–2337. Wolffs, P., B. Norling, J. Hoorfar, M. Griffiths, and P. Radstrom. 2005. Quantification of Campylobacter spp. in chicken rinse samples by using flotation prior to real-time PCR. Appl. Environ. Microbiol. 71:5759–5764. Wolffs, P., B. Norling, and P. Radstrom. 2005. Risk assessment of falsepositive quantitative real-time PCR results in food, due to detection of DNA originating from dead cells. J. Microbiol. Methods 60:315–323. Yang, C., Y. Jiang, K. Huang, C. Zhu, and Y. Yin. 2003. Application of real-time PCR for quantitative detection of Campylobacter jejuni in poultry, milk and environmental water. FEMS Immunol. Med. Microbiol. 38:265– 271.