Real-Time PCR Method for Detection of Pathogenic Yersinia ...

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Oct. 2008, p. 6060–6067 0099-2240/08/$08.00⫹0 doi:10.1128/AEM.00405-08 Copyright © 2008, American Society for Microbiology. All Rights Reserved.

Vol. 74, No. 19

Real-Time PCR Method for Detection of Pathogenic Yersinia enterocolitica in Food䌤 S. Thisted Lambertz,1* C. Nilsson,1 S. Hallanvuo,2 and M. Lindblad1 Research and Development Department, National Food Administration, Uppsala, Sweden,1 and Environmental and Food Research Laboratory (TavastLab), Municipal Joint Union for Public Health in Ha ¨meenlinna Region, Ha ¨meenlinna, Finland2 Received 18 February 2008/Accepted 5 August 2008

The current methods for the detection of pathogenic Yersinia enterocolitica bacteria in food are time consuming and inefficient. Therefore, we have developed and evaluated in-house a TaqMan probe-based real-time PCR method for the detection of this pathogen. The complete method comprises overnight enrichment, DNA extraction, and real-time PCR amplification. Also included in the method is an internal amplification control. The selected primer-probe set was designed to use a 163-bp amplicon from the chromosomally located gene ail (attachment and invasion locus). The selectivity of the PCR method was tested with a diverse range (n ⴝ 152) of related and unrelated strains, and no false-negative or false-positive PCR results were obtained. The sensitivity of the PCR amplification was 85 fg purified genomic DNA, equivalent to 10 cells per PCR tube. Following the enrichment of 10 g of various food samples (milk, minced beef, cold-smoked sausage, fish, and carrots), the sensitivity ranged from 0.5 to 55 CFU Y. enterocolitica. Good precision, robustness, and efficiency of the PCR amplification were also established. In addition, the method was tested on naturally contaminated food; in all, 18 out of 125 samples were positive for the ail gene. Since no conventional culture method could be used as a reference method, the PCR products amplified from these samples were positively verified by using conventional PCR and sequencing of the amplicons. A rapid and specific real-time PCR method for the detection of pathogenic Y. enterocolitica bacteria in food, as presented here, provides a superior alternative to the currently available detection methods and makes it possible to identify the foods at risk for Y. enterocolitica contamination.

satisfactorily. The detection of Y. enterocolitica in food can be significantly improved by PCR. The second generation of PCR methodologies, i.e., real-time PCR, which has overcome several limitations of conventional PCR, may be especially useful. In recent years, researchers have developed PCR-based assays for the detection of pathogenic Y. enterocolitica bacteria in food. One conventional PCR assay developed in our laboratory targets the chromosomally located virulence-associated gene ail in a nested PCR format (30). Other researchers have reported on conventional PCR assays designed to target the ail gene, as well as other genes, using single or nested PCR formats (4, 18, 21). Virulence in Y. enterocolitica results from a complex interplay between a series of plasmid- and chromosomally located genes. However, PCR targets located on the virulence plasmid must be considered unsuitable as targets for detection because the plasmid is unstable and easily lost during laboratory treatment (5, 22). Real-time PCR assays, especially those using TaqMan-based probes, provide greater specificity and require less time and labor to complete than conventional PCRs. Real-time PCR assays for the detection of pathogenic Y. enterocolitica bacteria in food using a TaqMan probe have previously been developed (6, 16, 31). However, in some instances these methods are not applicable as a diagnostic tool as they were not rigorously evaluated or lack an internal amplification control. The European Standardization Committee (CEN), in collaboration with the International Organization for Standardization (ISO), is currently preparing a protocol describing the minimal requirements for performance characteristics of mo-

Yersinia enterocolitica causes yersiniosis in many countries in the world (28). Yersiniosis is an internal infection, with the predominant symptoms being diarrhea, fever, abdominal pain, and vomiting. On a worldwide basis, the vast majority of reported cases occur sporadically, with the Y. enterocolitica bioserotype 4/O:3 being mainly responsible for the infections and with the source of infection unsolved. Y. enterocolitica is predominantly considered a foodborne agent, and as the bacterium has the ability to multiply in foods at low temperatures, approaching 0°C, as well as in vacuum-packed and modifiedatmosphere packages, it is of significant concern from a food safety and public health perspective. Y. enterocolitica is widely distributed in nature; however, among a large number of existing Y. enterocolitica bioserotypes, only a few are pathogenic to humans (17, 28). The most commonly encountered bioserotypes isolated from yersiniosis patients worldwide are 4/O:3, 2/O:9, 2/O:5,27, and 1B/O:8. It is well established that pigs are the main reservoir for Y. enterocolitica, and pork is therefore likely to be the most-important vehicle for its transmission to humans, directly or indirectly (17). However, the pathogen has seldom been isolated from food (8). The problem has been identified as methodological, referring to the fact that no traditional culture method works

* Corresponding author. Mailing address: National Food Administration Research and Development Department, P.O. Box 622, SE-751 26 Uppsala, Sweden. Phone: 46 (0)18 17 55 62. Fax: 46 (0)18 17 14 94. E-mail: [email protected]. 䌤 Published ahead of print on 15 August 2008. 6060

PCR DETECTION OF PATHOGENIC Y. ENTEROCOLITICA IN FOOD

VOL. 74, 2008

TABLE 1. Species, bioserotypes, and sources of bacterial strains used for evaluation of the selectivity of the real-time PCR method

Species and bioserotype (no. of strains)

a

No. of strains resulting in amplification of the target fragment

Pathogenic Y. enterocolitica strains (n ⫽ 102) 4/O:3 (81) .....................................................................................81 NT/O:9 (3) ................................................................................... 3 NT/O:8 (8) ................................................................................... 8 NT/O:5,27 (4) .............................................................................. 4 1B/O:18 (1) .................................................................................. 1 NT/O:20 (2) ................................................................................. 2 1B/O:21 (1) .................................................................................. 1 3/O:1,2,3 (1) ................................................................................. 1 5/O:2,3 (1) .................................................................................... 1 Nonpathogenic Y. enterocolitica strains (n ⫽ 14) 1A, diverse (5) ............................................................................. 0 Diverse (9) ................................................................................... 0 Nonpathogenic Yersinia spp. (n ⫽ 11) Y. frederiksenii (7)........................................................................ 0 Y. kristensenii (2) ......................................................................... 0 Y. intermedia (2) .......................................................................... 0 Y. pseudotuberculosis (4)................................................................. 0 Diverse food-related species (21) ................................................. 0 a A total of 152 strains were tested. The results were valid for amplification by using the primers (real 9A and 10A) in combination with the FAM/MGB-labeled probe, as well as with the FAM/TAMRA-labeled probe.

lecular methods developed for the detection of nucleic acid sequences (1). In addition, a European Union research project entitled FOOD-PCR has proposed a strategy and principles for the standardization of PCR methods developed for the detection of foodborne pathogens. A PCR method evaluated in-house should fulfill certain criteria, such as high selectivity, sensitivity, precision, robustness, etc., and an important prerequisite is that an internal amplification control is included in the method (1, 24). The aim of this study was to develop a method for the detection of the pathogenic bioserotypes of Y. enterocolitica present in food by using real-time PCR with TaqMan probes and applying performance criteria suggested by international standardization bodies for a method evaluated in-house. MATERIALS AND METHODS Bacterial strains and growth medium. The species of the 152 bacterial strains used in this study are shown in Table 1. The strains were from human (n ⫽ 81), pig (n ⫽ 16), food (n ⫽ 27), water (n ⫽ 2), and unknown (n ⫽ 26) sources. The pathogenic Y. enterocolitica strains were chosen to represent the most-common bioserotypes associated with human or animal disease, and the nonpathogenic strains are those most commonly recovered from food, human, and environmental sources. The strains were recognized as pathogenic and nonpathogenic by virtue of their origins and/or biochemical classifications. All strains were prepared as follows: pure colonies were grown on nutrient agar (CM3; Oxoid, Hampshire, England) at 30°C overnight. Per sample, a loop of colonies was transferred to a tube containing 200 ␮l of MilliQ water (Millipore, Bedford, MA) and 20 ␮l 0.8 M NaOH. The tubes were incubated for 10 min at 70 to 75°C, and subsequently, 48 ␮l of equal volumes of 0.8 M HCl and 0.1 M Tris (pH 8.3) was added. The samples were mixed and centrifuged. Aliquots of 5 ␮l of each sample were used for the PCR amplification. The reference strain Y. enterocolitica SLV-408 (bioserotype 4/O:3; CCUG 45643), originally isolated from frozen raw

6061

dog food (pig meat) was used for optimization and evaluation of the real-time PCR assay. Terms. Selectivity is a measure of the degree of response from target and nontarget microorganisms (1). It comprises the two terms (i) inclusivity, i.e., detection of the target organism from a wide range of strains, and (ii) exclusivity, i.e., lack of response from a relevant range of closely related nontarget microorganisms. The sensitivity of a PCR assay is the lowest number of cells that can be detected in a single analysis, precision is the closeness of agreement between test results, and robustness refers to the ability of the PCR to withstand small procedural or environmental changes. Finally, accuracy refers to the closeness of agreement between a test’s results and the accepted reference value. DNA purification. When the PCR assay was tested for sensitivity, precision, robustness, and the probability of detecting the target organism, enriched homogenates, broth cultures, or cell suspensions were used and, prior to the PCR amplification, genomic DNA was extracted by using a DNeasy blood and tissue kit (Qiagen Gmbh, Hilden, Germany). The procedure was conducted according to the manufacturer’s protocol. One-milliliter volumes of the homogenates or broth cultures or 100- to 200-␮l volumes of the cell suspensions were used, and the final DNA was resuspended in 200 ␮l kit buffer. Real-time PCR and cycling parameters. To find the primer concentrations that would produce the optimal amplification signal, nine different combinations between 50 and 900 nM of the forward and reverse primer were tested. The optimal PCR conditions were the following: 1⫻ TaqMan universal PCR master mix (contains AmpliTaq Gold DNA polymerase, deoxynucleoside triphosphates, Passive reference 1, and optimized buffer components; Applied Biosystems, Foster City, CA), primers (real 9A and real 10A) to a final concentration of 900 nM, probe to a final concentration of 200 nM, approximately 100 copies of internal amplification control DNA (TaqMan exogenous internal positive control [IPC]; Applied Biosystems, Foster City, CA), 1⫻ internal amplification control mix (Applied Biosystems, Foster City, CA), and finally, 5 ␮l of the sample. Sterile MilliQ water was used to adjust the volume of each reaction mixture to 25 ␮l. The PCR cycling parameters were as follows: initial denaturation of the template DNA at 95°C for 10 min, followed by 45 cycles at 95°C for 15 s and at 60°C for 1 min. In order to prevent carryover contamination, TaqMan universal PCR master mix (Applied Biosystems, Foster City, CA) with the inclusion of dUTP instead of dTTP was used in all tests on naturally contaminated foods presented in this paper; uracil-N-glycosylase was then applied to the reaction mixture to inactivate all undesired amplified PCR products (in an incubation step of 50°C for 2 min which was added to the cycling parameters described above) prior to the amplification reaction. The analyses were performed in 96-well plates (occasionally eight-strips) using an Applied Biosystems 7500 real-time PCR system (Applied Biosystems, Foster City, CA). A nontarget control containing 5 ␮l of MilliQ water instead of DNA was included in each run to detect any PCR fragment contamination. In addition, a nonpatented variant of an internal amplification control based on pUC19 (Fermentas, Germany) was tested. Thus, instead of the IPC system described above, 500 nM of each of the primers (IAC_fw and IAC_re; MWG Biotech, Germany), 200 nM of the accompanying internal amplification control probe (Applied Biosystems, Foster City, CA), and 1 fg of pUC19 were used in each reaction mixture (information below). Internal amplification controls. Two internal amplification controls were tested in this study. (i) A commercially available TaqMan exogenous IPC (Applied Biosystems, Foster City, CA) was used. The reagent kit included primers, a Vicprobe, IPC target DNA, and blocking solution. The IPC target DNA was diluted 10 times to achieve a copy number of approximately 100 per PCR mixture. The PCR product length is not declared to the consumer. (ii) An open formula, pUC19-based internal amplification control developed by Fricker et al. (9) was used. Approximately 50 to 100 copies of target DNA (pUC19) were used per PCR mixture. The size of the pUC19-based internal amplification control was 119 bp. The real-time PCR assay was optimized to work under conditions suitable for the detection of the target organism and not the internal amplification controls; thus, the internal amplification control reactions were performed after the detection of the target DNA. Primer and probe design. One primer set and two alternative probes were used. The primers were R-real 9A, CCCAGTAATCCATAAAGGCTAAC ATAT (27-mer), and F-real 10A, ATGATAACTGGGGAGTAATAGGTTCG (26-mer). The probes were the ail probe, FAM-TGACCAAACTTATTACTGC CATA-MGB, labeled at the 5⬘end with the reporter dye 6-carboxyfluorescein (FAM) and at the 3⬘end with a special chemical compound called minor groove binder (MGB), and the Ye probe, FAM-TCTATGGCAGTAATAAGTTTGGT CACGGTGATCT-TAMRA, labeled at the 5⬘end with FAM but at the 3⬘end with the reporter dye tetramethyl-6-carboxyrhodamine (TAMRA). Two probes were used instead of one in order to offer the user multiple options with a nonpatented, so-called open formula. The primer and probe sequences were

6062

THISTED LAMBERTZ ET AL.

designed manually and calculated using the nearest-neighbor algorithm by using the Primer Express program (version 3.0) to achieve an annealing-extension temperature of 60°C. The specificity of the amplified sequence was tested by a BLAST search in the GenBank database (http:/www.ncbi.nlm.nih.gov). Both probes were of the TaqMan style. All primers and probes were purchased at www.appliedbiosystems.com. Sequencing. PCR products amplified from seven of the pathogenic Y. enterocolitica strains representing five different bioserotypes and from nine of the naturally contaminated food samples were sequenced. The PCR products were purified by using a QIAquick PCR purification kit (Qiagen Gmbh, Hilden, Germany) according to the manufacturer’s protocol before being sequenced at Uppsala Genome Centre (Rudbeck Laboratory, SE-751 85 Uppsala, Sweden). Sequencing was performed in both directions. Sensitivity and probability of detection. The sensitivity of the PCR amplification was determined with purified DNA. Genomic DNA of Y. enterocolitica (purified as described above) was serially diluted 10-fold in sterile MilliQ water and subjected to PCR amplification by using the reaction and cycling parameters described above. The DNA concentration was measured with a NanoDrop ND1000 spectrophotometer at a wavelength of 260 nm. In addition, to determine the probability of detection, a broth culture of Y. enterocolitica SLV-408 grown in brain heart infusion (BHI; Oxoid, Hampshire, England) at 25°C overnight was 10-fold serially diluted in peptone water (Oxoid). Aliquots (0.1 ml) of the serial dilution were plated in duplicate onto nutrient agar plates and grown at 30°C overnight. The colonies were counted to determine the total number of viable Y. enterocolitica CFU. Genomic DNA was extracted (purified as described above) from 1 ml of each of the dilutions presumed to contain 100 to 106 CFU per ml. Per dilution, five aliquots (5 ␮l) and 100 copies of the IPC per aliquot were used for PCR amplification. This experiment was repeated six times, thus resulting in 30 PCR results for each cell concentration. Precision test. The precision of the PCR amplification was determined by analyzing a 10-fold serial dilution of genomic DNA of Y. enterocolitica once each on three consecutive days with the assays conducted by the same person. The samples were prepared as follows: colonies of Y. enterocolitica SLV-408 were suspended in 1 ml of peptone water and genomic DNA released as described above. The DNA concentration was measured with a NanoDrop spectrophotometer at a wavelength of 260 nm. The DNA extracted was 10-fold serially diluted in sterile MilliQ water, and aliquots (5 ␮l) of the appropriate dilutions to achieve 106, 105, 104, 103, 102, 101, and 100 genomic equivalents of Y. enterocolitica DNA were used for PCR amplification. All analyses were performed in the presence of 100 copies of the IPC. In parallel, the 10-fold serial dilution was used to produce a standard curve. Robustness test. The robustness of the PCR amplification was determined by testing four different annealing-extension temperatures, i.e., 58, 60, 62, and 64°C, and by using optimized and suboptimal concentrations, i.e., ⫾20%, of the universal PCR master mix. Each of the conditions was tested in eight replicates on 50 copies of genomic DNA of Y. enterocolitica SLV-408 in the presence of 100 copies of the internal amplification control. All other cycle conditions were kept constant. The samples were analyzed in duplicate. Detection in a background of non-Y. enterocolitica DNA. A pure culture of Y. enterocolitica SLV-408 was grown in BHI broth at 25°C overnight, and genomic DNA was released from 1 ml of the broth (as described above). The DNA concentration was determined spectrophotometrically before 10-fold serial dilution. In parallel, a vial containing a freeze-dried mixture of Enterobacter cloacae (8 ⫻ 103 CFU per ml), Campylobacter jejuni (30 CFU per ml), Escherichia coli O157 (40 CFU per ml), Listeria monocytogenes (60 CFU per ml), and Salmonella enterica serovar Dublin (15 CFU per ml) cells in tryptone soya broth (TSB) was cultured at 25°C overnight. DNA was also extracted from this broth, and the DNA concentration was determined spectrophotometrically. Equal volumes (2.5 ␮l) of this extracted background DNA and various concentrations of DNA from Y. enterocolitica cells ranging from 100 to 106 genomic equivalents (per 2.5 ␮l) were mixed and amplified by PCR as described above. In parallel, the same concentrations, 100 to 106 per 2.5 ␮l, of DNA from Y. enterocolitica cells were subjected to PCR amplification alone, i.e., without the presence of DNA from the background flora. All samples were analyzed in duplicate. Detection in artificially inoculated food. A pure culture of Y. enterocolitica SLV-408 was grown in BHI broth at 25°C overnight. The culture was 10-fold serially diluted in peptone water, and the total number of viable cells was determined by counting viable CFU on nutrient agar plates (as described above). One hundred microliters of the appropriate dilution was used to inoculate five different food items: milk (pasteurized, 1.5% fat), raw minced beef, cold-smoked sausage (heat treated at 65°C for 1 to 3 min), carrots (clean, unpeeled), and raw fish. The food samples were purchased at local stores in Uppsala, Sweden, and transported and stored chilled until analysis, which was started either the same

APPL. ENVIRON. MICROBIOL. day or the day after arrival. Five 10-g portions of each food were aseptically transferred to sterile petri dishes and inoculated on the surface with Y. enterocolitica cells to provide cell numbers of 100, 101, 102, 103, and 104 CFU per 10 g of food. The inoculated foods were kept for 0.5 h at 25°C before being mixed with 90 ml TSB supplemented with 0.6% yeast extract and homogenized for 1 min (using Stomacher Seward filter bags). The homogenates were enriched at 25°C for 18 to 20 h. DNA from 1 ml of the homogenates was released (as described above). An aliquot (5 ␮l) of each sample was used for PCR amplification. All samples were analyzed in duplicate. Detection in naturally contaminated food. One hundred five samples of food, comprising raw pork (n ⫽ 25), cold-smoked sausage (n ⫽ 25), fish (n ⫽ 25), carrots (n ⫽ 25), and milk (n ⫽ 5), were purchased from shops in Uppsala during June, July, and August 2007. In addition, 20 raw milk samples were collected in June 2007 from milk cows on two dairy farms located near Uppsala. The food items were transported and stored chilled until analysis, which was started either the same day or the day after arrival. Twenty-five-gram portions of each food were homogenized in 225 ml TSB–0.6% yeast extract and enriched at 25°C for 18 to 20 h. Each homogenate was gently mixed and coarse particles allowed to settle for 15 to 20 min before DNA from 1 ml of the homogenates was released (as described above). An aliquot (5 ␮l) of extracted DNA from each of the samples was used for PCR amplification. Statistical analysis. The Pearson product-moment correlation coefficient (r) was used to examine the relations between threshold cycle (CT) values and log concentrations of purified DNA, as well as between CT values and log numbers of Y. enterocolitica cells (without enrichment). A univariate general linear model (GLM) was used to analyze the relation between numbers of Y. enterocolitica cells inoculated into food, CT values after enrichment, and possible effects of food type on this relation. The coefficient of variation (CV) was used as a measure of intraindividual day-to-day reproducibility in PCR amplifications of different concentrations of DNA.

RESULTS Optimization and efficiency of real-time PCR. When nine different primer concentration combinations were tested, 900 nM of each of the primers was found to produce the optimal amplification signal. A standard curve based on 10-fold serial dilutions of between 101 and 106 genome equivalents of Y. enterocolitica DNA per PCR mixture showed a linear relationship between the log input DNA and CTs. The slope was ⫺3.56, and the efficiency 91%. The threshold line was set to a fluorescence value of 0.01 and was used throughout the study. Selectivity of detection. Real-time PCR amplification of genomic DNA using the primers and probes, reaction conditions, and cycling parameters described here resulted in amplification of the expected 163-bp fragment from all 102 isolates of the pathogenic Y. enterocolitica strains tested, i.e., 100% inclusivity (Table 1). Seven of the amplicons were analyzed on a 2% agarose gel, confirming a single fragment of the expected length compared to a known DNA size marker (Amersham Biosciences, United Kingdom). The identities of the PCR amplicons from these seven isolates were confirmed by sequencing (Table 2). The sequence data showed that the bioserotype 4/O:3, 4/O:9, and 4/O:5,27 strains contained 100% identical nucleotides over the fragments investigated. The amplified 1B/O:8 strain fragment differed from that sequence in four nucleotides. The sequences were subjected to a homology search in BlastN which revealed no identical sequences other than those reported for the ail gene of Yersinia spp. No amplification was noticed with any of the 50 nonpathogenic Y. enterocolitica bioserotypes or other, non-Yersinia species, and thus, the exclusivity was 100%. The results reported here were equally valid for amplification using the primers in combination with either the MGB-labeled probe or the TAMRA-labeled probe.


VOL. 74, 2008

TABLE 2. Nucleotide sequence comparison of ail genes from various sourcesa Strain or food source of nucleotide sequence

No. of identical nucleotides/total no. compared (% identity)b

Strain (bioserotype or serotype) SLV-408 (4/O:3) 959106 (4/O:3) 2096 (O:9) 8238 (O:8)

117/117 (100) 118/118 (100) 106/106 (100) 105/109 (96)

2120 (O:5,27) 2121 (O:5,27) 8239 (O:9)

134/134 (100) 118/118 (100) 107/107 (100)

Food sourcee Minced pork

158/163 (97)

Pork Wienerschnitzel Carrots Carrots

163/163 (100) 115/115 (100) 163/163 (100) 47/49 (96)

Milk Halibut

163/163 (100) 158/163 (97)

Pangasius

158/163 (97)

Warm-smoked salmon

163/163 (100)

Positionsc

Nucleotidesd

6063

TABLE 3. Sensitivity of detection by real-time PCR of Y. enterocolitica SLV-408 genomic DNA in the presence of the internal amplification control IPC No. of Y. enterocolitica genome copies (amt of DNA per PCR mix) 6

MGB-labeled Yersinia probe

IPC probeb

CT valuea

CV (%)

CT valuea

CV (%)

18.17 ⫾ 0.18 21.75 ⫾ 0.58 25.56 ⫾ 0.39 29.24 ⫾ 0.36 32.60 ⫾ 0.71 36.93 ⫾ 0.90 ND ND

1.0 2.7 1.5 1.2 2.2 2.4

29.24 ⫾ 0.31 27.89 ⫾ 0.94 27.23 ⫾ 0.71 27.22 ⫾ 0.78 27.05 ⫾ 1.11 26.35 ⫾ 0.27 25.50 ⫾ 1.20 25.03 ⫾ 0.99

1.1 3.4 2.6 2.9 4.1 1.0 4.7 4.0

2416, 2428, 2461, 2468

C/G, A/G, A/G, A/G

10 105 104 103 102 101 100 0

2416, 2428, 2461, 2468, 2509

T/C, T/C, T/C, G/C, C/T

a The CT value is the fractional cycle number at which the fluorescence passes the threshold. Each CT value is the mean ⫾ 1 standard deviation of the results of three experiments with one replicate in each. ND, not detected. b One hundred copies of the IPC were present.

2416, NT, NT, NT, 2509

T/C, NT, NT, NT, C/T

2416, 2428, 2461, 2468, 2509 2416, 2428, 2461, 2468, 2509

T/C, T/C, T/C, G/C, C/T T/C, T/C, T/C, G/C, C/T

a The ail gene in the published genome (GenBank accession no. AJ605740) for bioserotype 4/O:3 Y. enterocolitica strain SLV-408 was compared with the same sequence amplified from seven culture collection strains and nine naturally contaminated food samples by using the primers real 10A and real 9A and the ail probe. NT, not tested. b The percent identity is approximate. c The position numbers refer to the ail gene. d Variant nucleotide in respective strain/published nucleotide of ail (GenBank accession no. AJ605740). e All food sources except warm-smoked salmon were raw, untreated food.

Sensitivity and probability of detection. The minimum level of detection of the ail target from purified Y. enterocolitica genomic DNA was 85 fg, with a mean CT value of 36.93 (standard deviation ⫽ 0.90). This level of detection was equivalent to 10 Y. enterocolitica cells, based on the assumption that single copies of the target DNA are present in the genomes (Table 3). The analysis of the probability of detection with a 10-foldserially diluted cell suspension of Y. enterocolitica bacteria revealed that there was amplification in 47% (14 out of 30) of the samples with a cell count level of 10 Y. enterocolitica cells per ml and in 97% (29 out of 30) of the samples with a cell count level of 100 Y. enterocolitica cells per ml. Highly significant correlations (Pearson⬘s r ⫽ ⫺1.0; P ⬍ 0.0001) were observed between the CT values and the log concentrations of purified DNA, as well as the log numbers of Y. enterocolitica cells. Precision and robustness. The precision of the PCR amplification is presented in Table 3. The CVs of the mean CT values for the Yersinia MGB-labeled probe ranged from 1.0 to 2.7%. In the test determining the robustness of the PCR assay, it was found that at the annealing-extension temperatures of 58, 60, and 62°C, a 20% increase or decrease in the concentration of the PCR reagents did not affect the detection sensitivity notably, whereas at 64°C, the sensitivity was markedly influenced (Table 4). Influence of a background of non-Y. enterocolitica DNA on detection of target DNA. Y. enterocolitica DNA was detectable

(8.5 ng) (0.85 ng) (0.085 ng) (8.5 pg) (0.85 pg) (85 fg) (8.5 fg)

at all levels tested, ranging from 101 to 106 genome equivalents per PCR mixture, in the presence of a background flora of approximately 107 genome equivalents per PCR mixture. At levels of 104 to 106 genome equivalents of Y. enterocolitica DNA, the detection of the target DNA was positively affected by the presence of the background flora in that the sensitivity increased by 2 CTs. At the levels of 102 to 103 genome equivalents of Y. enterocolitica DNA, detection of the target DNA occurred at the same CT value with and without the background flora. At a target DNA level of 101, the presence of the background flora affected the detection of the target DNA negatively in that the sensitivity decreased by 4 CTs, and finally, at a target DNA level of 100, no detection was recorded. Detection in artificially contaminated food. The results of the PCR amplification showed that 0.5 to 5.5 CFU of Y. enterocolitica artificially inoculated into 10 g of milk and onto 10 g of cold-smoked sausage, respectively, was detected following 18 to 20 h of enrichment. Furthermore, 55 CFU Y. enterocolitica was detected in 10 g of carrots, fish, and minced beef under the same conditions. The mean CT values for milk and cold-smoked sausage were about 10 to 15 cycles lower than for carrots, fish, and minced beef (Table 5). The results of univariate general linear model analysis showed that the CT values were significantly (P ⬍ 0.001) related to the concentrations of Y. enterocolitica cells inoculated into 10 g of food and that the

TABLE 4. Robustness of detection by real-time PCR of 50 copies of Y. enterocolitica SLV-408 genomic DNAa CT value with indicated PCR reagent concnb

Annealingextension temp (°C)

Optimized

20% lower concn

20% greater concn

58 60 62 64

35.57 ⫾ 0.66 35.50 ⫾ 0.71 33.97 ⫾ 0.46 38.01 ⫾ 1.16

34.63 ⫾ 0.27 35.56 ⫾ 0.64 33.54 ⫾ 0.41 35.31 ⫾ 0.49

36.81 ⫾ 0.45 36.08 ⫾ 0.40 34.59 ⫾ 0.49 41.89 ⫾ 1.34

a PCR was performed in the presence of 100 copies of the internal amplification control IPC using optimized or suboptimal (⫾ 20%) PCR reagent concentrations and four annealing-extension temperatures. b The CT value is the fractional cycle number at which the fluorescence passes the threshold. Each CT value is the mean ⫾ 1 standard deviation of the results for eight replicates.

6064


APPL. ENVIRON. MICROBIOL.

TABLE 5. Sensitivity of detection by real-time PCR following enrichment and DNA extraction of Y. enterocolitica SLV-408 bacteria artificially inoculated into various foods Inoculanta or control

5.5 ⫻ 103 5.5 ⫻ 102 5.5 ⫻ 101 5.5 0.5 IPCd Negative controle

CT value with MilliQ water

26.78

CT value with indicated food or controlb Milk

Cold-smoked sausage

Carrots

Fish

Minced beef

15.00 ⫾ 0.12 17.19 ⫾ 0.14 19.17 ⫾ 0.02 22.59 ⫾ 0.31 24.71 ⫾ 0.19 27.52 ⫾ 0.67 ND

14.98 ⫾ 0.06 16.57 ⫾ 0.11 21.03 ⫾ 0.09 23.89 ⫾ 0.19 ND 27.98 ⫾ 0.20 ND

31.73 ⫾ 0.20 34.73 ⫾ 0.94 34.85 ⫾ 0.32 ND ND 29.44 ⫾ 0.55 ND

30.09 ⫾ 0.09 33.12 ⫾ 0.37 35.81c ND ND 29.42 ⫾ 0.25 ND

31.84 ⫾ 0.57 33.82 ⫾ 0.38 38.78 ⫾ 0.82 ND ND 29.48 ⫾ 0.19 ND

a

Ten grams of each food was inoculated with the indicated no. of CFU of Y. enterocolitica. The CT value is the fractional cycle number at which the fluorescence passes the threshold. Each CT value is the mean ⫾ 1 standard deviation of the results of two experiments. ND, not detected. c Only one positive result was obtained with two samples. d 100 copies of the IPC were added to the food matrices and MilliQ water. e The negative amplification control was 5 ␮l sterile water. b

intercept of the regression equations differed significantly (P ⬍ 0.001) between raw food samples (carrots, fish, and minced beef) and prepared food samples (milk and cold-smoked sausage). All uninoculated food samples were negative. The CT values for the IPC when amplified in the uninoculated homogenates were one cycle higher for milk and cold-smoked sausage and three cycles higher for carrots, fish, and minced beef than the CT values for the IPC when amplified in sterile MilliQ water. Detection in naturally contaminated food. Of 125 potentially naturally contaminated food samples collected, there was amplification of the ail gene following 18 to 20 h of enrichment in 6 out of 25 raw pork samples (at CT values between 30.5 and 35.7), 1 out of 25 milk samples (at the CT value 38.4), 4 out of 25 carrot samples (at CT values between 20.4 and 35.3), and 7 out of 25 fish samples (at CT values between 35.6 and 40.7), indicating that pathogenic Y. enterocolitica bacteria were present in these samples. Thus, in all, 14% (18 out of 125) of the samples were PCR positive for the ail gene. The results for nine of these samples (3 raw pork, 1 milk, 2 carrot, and 3 fish) were confirmed by analyzing the amplicons on a 2% agarose gel and by sequencing the amplicons (Table 2). Sequencing was performed twice. For all nine samples, the gel analysis revealed a fragment of the expected length (163 bp), confirmed by comparison to a known DNA size marker (data not shown). The results of the sequence analysis revealed that five of the samples (2 raw pork, 1 carrot, 1 milk, and 1 fish) contained nucleotides that were 100% identical over the fragments investigated to this region in the bioserotype 4/O:3 Y. enterocolitica strain SLV-408 (GenBank accession no. AJ605740). The remaining four amplicons (amplified from 1 raw pork, 1 carrot, and 2 raw fish samples) differed from the sequence in strain SLV-408 (Table 2) but instead showed 100% similarity to the corresponding sequence in the bioserotype 1B/O:8 Y. enterocolitica strain 8081 (GenBank accession no. AM286415). Internal amplification control. When the internal amplification control IPC (Applied Biosystems) was present in the PCR mix at an initial copy number of 100, no differences in the detection sensitivities (CT values) occurred at any of the Y. enterocolitica DNA copy numbers of 106, 105, 104, 103, and 102 per PCR mix. However, the detection sensitivity was slightly

affected when 10 copies of Y. enterocolitica DNA were present in the PCR, as indicated by an increase in the CT value of approximately 1 cycle compared to that obtained in parallel samples without the IPC (data not shown). The results obtained with the pUC19-based internal amplification control in the PCR mix were similar to those reported for the IPC (data not shown).

DISCUSSION In this study, a real-time PCR method for the rapid, specific, and sensitive detection of the pathogenic bioserotypes of Y. enterocolitica in food was developed and evaluated according to criteria for a PCR method suggested by international standardization bodies (1, 24). The all-in-all method consists of an overnight enrichment followed by the extraction of DNA prior to a TaqMan probe-based real-time PCR amplification and can be completed within 1 to 2 working days. The PCR assay was designed to amplify a sequence on the chromosomally located gene ail. The ail gene is an excellent target because it is only found in strains of Yersinia spp. associated with pathogenicity in humans (25) and thus it indicates the presence of the organism and defines the pathogenic subgroup at the same time. At least two variants of the ail gene exist in Y. enterocolitica strains, referred to as American and European (2). One variant is also present in Yersinia pseudotuberculosis. The primer and probe sites for real-time PCR in this study were selected based on knowledge regarding the ail locus acquired from previous work with conventional PCR in our laboratory (30). Whereas the conventional PCR work utilized primer sites on the ail gene that are present in both Y. pseudotuberculosis and Y. enterocolitica, the real-time PCR method was developed to restrict detection to Y. enterocolitica (both variants) only. The results from the selectivity study presented here showed that the chosen primers and probes provided specific detection of a large number of pathogenic Y. enterocolitica strains isolated from infected human patients, animals, foods, and unknown sources. Yersinia species other than Y. enterocolitica and other bacterial species were not detected, suggesting that this newly developed assay is specific

VOL. 74, 2008


and reliable in detecting the pathogenic bioserotypes of Y. enterocolitica (Table 1). Amplicons from seven PCR-positive strains recovered from culture collections were sequenced to reveal the identities of the fragments obtained. The strains were chosen to represent the most-common bioserotypes isolated worldwide from yersiniosis patients. Sequence alignments of the amplified fragments from the bioserotype 4/O:3, 4/O:9, and 4/O:5,27 strains showed 100% similarity to this region in the bioserotype 4/O:3 Y. enterocolitica strain SLV-408 (GenBank accession no. AJ605740) (Table 2). The stability of this gene region has previously been reported (11, 20). The 1B/O:8 nucleotide sequence, on the other hand, differed in the same region in four of the nucleotides; this in turn agreed with the sequence available in the GenBank database (accession no. AM286415.1) for the bioserotype 1B/O:8 Y. enterocolitica strain 8081. Thus, although there is no need for a postamplification analysis when using real-time PCR, by sequencing the amplicons it may be possible to distinguish bioserotype 1B/O:8, which belongs to the mouse-lethal group of strains formerly entitled American, from the mouse-nonlethal and less-virulent European bioserotypes, i.e., the bioserotypes most commonly reported from yersiniosis patients worldwide (4/O:3, 2/O:9, and 2/O:5,27). In addition, amplicons from the naturally contaminated food samples were sequenced (Table 2); out of nine samples investigated, three revealed a nucleotide sequence that was 100% identical to that of the 4/O:3, 2/O:9, and 2/O:5,27 group of strains and the remaining six samples showed 100% sequence similarity to bioserotype 1B/O:8. Whereas pathogenic Y. enterocolitica has previously been isolated from carrots (7), to our knowledge, this is the first time it has been detected in samples of fish. However, these results need to be further confirmed by isolation and biochemical characterization of the bacteria. The minimum level of detection by the real-time PCR was 85 fg purified genomic DNA, equivalent to 10 Y. enterocolitica cells per PCR mixture (Table 3). Comparable results were obtained by other investigators, who reported similar sensitivities for the detection of Y. enterocolitica by real-time PCR, as well as for the detection of other bacterial species (6, 16, 29, 31). It was shown in this study that the probability of detecting the reference strain, SLV-408 (Y. enterocolitica bioserotype 4/O:3), at a cell count level of 10 Y. enterocolitica cells per ml (0.5 CFU per reaction) was 47% and that at a cell count level of 100 Y. enterocolitica cells per ml (5 CFU per reaction mixture) it was 97%. Thus, a consistent level of detection with this TaqMan PCR method was achieved for somewhat more than 102 cells of Y. enterocolitica per ml of a pure bacterial culture. This turned out to be more efficient than other researchers have reported (15, 23). These differences may reflect the different efficiencies of the DNA extraction methods used. The assay showed good precision in the detection of the target DNA, demonstrated by testing a 10-fold serial dilution of Y. enterocolitica genomic DNA on three different occasions. Furthermore, a test for robustness showed that the detection remained robust through small, unfavorable changes in the annealing temperature and the concentration of the PCR reagents (Table 4). The robustness test was set up to simulate variations that could occur in different laboratories. In order to meet the criteria for a fully validated method, a future collaborative trial of this real-time PCR-based method will give fur-

6065

ther information on its robustness, expressed as the values for repeatability and reproducibility, as explored at different laboratories. A real-time PCR method is not applicable as a diagnostic tool if no internal amplification control is included (12). Two alternative internal amplification controls were used in this study: one that is commercially available, with limited information given by the manufacturer about the design, and one that is noncommercial, with all information available. Both were noncompetitive variants, i.e., different primer sets were used to amplify the target DNA and the internal amplification control. Noncompetitive variants were chosen instead of competitive variants because they work independently of the target detection system, with the advantage that a single internal amplification control system can be used for many different assays in the same laboratory. In this study, it was observed that in samples containing low target DNA copy numbers, i.e., approximately 101 copies, detection of the target was slightly affected by the presence of the internal amplification control, as indicated by a slightly higher CT value obtained for samples containing the internal amplification control than for those without the internal amplification control. One possible explanation could be that some interaction between the multiple primers occurred during amplification. As recommended by Hoorfar et al. (13), the PCR was optimized only for the target DNA and the IPC reaction was performed after the target DNA reaction, which may have been somewhat subefficient for the IPC reaction. However, as long as the IPC amplicon is detected, it may not be essential to detect it efficiently. Contamination of food by pathogens often occurs at low levels. To avoid false-negative results, a major issue when designing a new detection method is to achieve as high a sensitivity as possible (8). Nevertheless, the use of a highly sensitive PCR assay is often not enough; to enable detection it is necessary to add an enrichment step to the method (3, 7, 14). However, several studies have reported considerable problems in getting Y. enterocolitica bacteria to grow in culture media (8, 19). Fukushima and Gomyoda (10) showed that the growth of a serotype O:3 strain of Y. enterocolitica was suppressed in mixed cultures with Yersinia-related species, especially when the competitors were initially present at 102 to 103 times the amount of the pathogen. Likewise, other members of the Enterobacteriaceae are inhibitory to the pathogenic strain under similar conditions. Enumeration data for Y. enterocolitica bacteria in food are only available from a few reports. In one study, Nesbakken et al. (26) estimated the number of cells in eight samples of raw pork sausage and found Y. enterocolitica bacteria to be present at an average of 439 CFU per gram. Furthermore, Y. enterocolitica bacteria were found in three samples of raw pork cuts at an average of 200 CFU per gram. The results of the real-time PCR method developed here revealed that there was amplification of the ail gene target with an initially low cell count of 55 Y. enterocolitica bacteria per 10 g of food in all the enriched food samples tested (Table 5). This indicates that the real-time PCR method has considerable potential for the detection of Y. enterocolitica bacteria in food even when the pathogen is present at initially low numbers. When adopting a new method, validation by its application to naturally contaminated samples and comparison of the results with those of a reference method are required. Several

6066


reference methods proposed by different national and international organizations exist for Y. enterocolitica, but there is no consensus regarding a standard method. In the United States, the Food Safety and Inspection Service of the U.S. Department of Agriculture (USDA FSIS) proposes three procedures for its detection as there is no universal enrichment schema for the three most-important serotypes in the United States (O:3, O:8, and O:5,27). In the Nordic countries, the methods proposed by the Nordic Committee for Food Analysis (NMKL) and the International Organization for Standardization (ISO), entitled NMKL-117 and EN/ISO 10273, respectively, are the most commonly used. NMKL-117 was recently revised (NMKL-117b), which made it almost identical to the international reference method EN/ISO 10273. There are, however, considerable reservations as to whether these methods are sensitive enough to enable detection of the pathogenic bioserotypes of Y. enterocolitica in food. In a comparison of the ISO 10273:2003 reference method and the real-time PCR described in this current manuscript using food samples artificially inoculated with low levels of Y. enterocolitica bacteria (approximately 10 to 300 CFU/25-g sample), the accuracy of the ISO 10273 method, expressed by the relative sensitivity (27) of the culture method, was only 40%, compared to 95% obtained by using the real-time PCR method (S. Hallanvuo, unpublished data). Furthermore, Fredriksson-Ahomaa and Korkeala (8) concluded that the reported low rate of pathogenic Y. enterocolitica strains that have been isolated from naturally contaminated food samples is most probably due to limited sensitivities of the traditional culture methods; the detection limit is between 103 to 106 CFU or more per g in pork products. The NMKL-117b method was recently collaboratively validated, but the results were not accepted as satisfactory by the NMKL committee. Before taking further action, the committee is waiting for validation of the EN/ISO method, which is planned to take place before the end of 2008. With reference to the present lack of a validated method, the naturally contaminated samples analyzed here were not tested in parallel against a reference method. Instead, to support confidence in the detection results, the PCR-positive results were confirmed by retesting the presumptively positive samples, measuring the size of the PCR products, and sequencing the PCR amplicons (as described above). This novel method, which included enrichment, sample processing, and real-time PCR amplification, exhibited high selectivity, sensitivity, and robustness and can be completed within 1 to 2 working days. This is a substantial improvement over the conventional culture methods that take more than 10 days to complete (EN/ISO and NMKL). The standard curves from the real-time PCR CT values showed that there was a strong correlation between both enriched and unenriched cells and CT values, which can be used to provide potential quantitative estimates of the levels of Y. enterocolitica bacteria in different food items. Higher CT values for the inoculated minced beef, carrots, and fish (raw) than for the inoculated milk and coldsmoked sausage (heat treated) could be due to growth suppression by a competitive background flora present in the raw food samples. Also, PCR amplification of the IPC yielded slightly higher (two to three cycles) CT values for raw food samples, indicating that for these samples, inhibitory substances may have interfered with the amplification process.

APPL. ENVIRON. MICROBIOL.

However, these circumstances did not prevent Y. enterocolitica bacteria from being detectable in all five inoculated food items tested here at a level of 55 CFU or less per 10-g sample following enrichment (Table 5). The possibility of detecting the pathogenic bioserotypes of Y. enterocolitica in food at this level would help to determine the foods at risk and, further, the points in the food production process where contamination occurs and where controls could be introduced to reduce or eliminate the pathogen in retail food products. ACKNOWLEDGMENTS We gratefully thank Anna Aspa´n (National Veterinary Institute, Sweden) for commenting on the manuscript and Annelie Lundin (National Food Administration, Sweden) for improving the English. We also thank the farmers for providing us with the raw milk samples. REFERENCES 1. Anonymous. Microbiology of food and animal feeding stuffs—polymerase chain reaction (PCR) for the detection of food-borne pathogens—performance characteristics of molecular detection methods. ISO TC 34/SC 9 N752. International Organization for Standardization, Geneva, Switzerland. Standard in preparation. 2. Beer, K. B., and V. L. Miller. 1992. Amino acid substitutions in naturally occurring variants of Ail result in altered invasion activity. J. Bacteriol. 174:1360–1369. 3. Bhaduri, S. 2006. Enrichment, isolation, and virulence of freeze-stressed plasmid-bearing virulent strains of Yersinia enterocolitica on pork. J. Food Prot. 69:1983–1985. 4. Bhaduri, S., and B. Cottrell. 1997. Direct detection and isolation of plasmidbearing virulent serotypes of Yersinia enterocolitica from various foods. Appl. Environ. Microbiol. 63:4952–4955. 5. Blais, B. W., and L. M. Phillippe. 1995. Comparative analysis of yadA and ail polymerase chain reaction methods for virulent Yersinia enterocolitica. Food Control 6:211–214. 6. Boyapalle, S., I. V. Wesley, H. S. Hurd, and P. Gopal Reddy. 2001. Comparison of culture, multiplex, and 5⬘ nuclease polymerase chain reaction assays for the rapid detection of Yersinia enterocolitica in swine and pork products. J. Food Prot. 64:1352–1361. 7. Cocolin, L., and G. Comi. 2005. Use of a culture-independent molecular method to study the ecology of Yersinia spp. in food. Int. J. Food Microbiol. 105:71–82. 8. Fredriksson-Ahomaa, M., and H. Korkeala. 2003. Low occurrence of pathogenic Yersinia enterocolitica in clinical, food, and environmental samples: a methodological problem. Clin. Microbiol. Rev. 16:220–229. 9. Fricker, M., U. Messelhausser, U. Busch, S. Scherer, and M. Ehling-Schulz. 2007. Diagnostic real-time PCR assays for the detection of emetic Bacillus cereus strains in foods and recent food-borne outbreaks. Appl. Environ. Microbiol. 73:1892–1898. 10. Fukushima, H., and M. Gomyoda. 1986. Growth of Yersinia pseudotuberculosis and Yersinia enterocolitica biotype 3B serotype O3 inhibited on cefsulodin-Irgasan-novobiocin agar. J. Clin. Microbiol. 24:116–120. 11. Gierczynski, R., M. Jagielski, and W. Rastawicki. 2001. The presence of the ail gene in clinical strains of Yersinia enterocolitica isolated from stools in Poland and characteristics of gene variant. Acta Microbiol. Pol. 50:19–25. 12. Hoorfar, J., N. Cook, B. Malorny, M. Wagner, D. De Medici, A. Abdulmawjood, and P. Fach. 2003. Making internal amplification control mandatory for diagnostic PCR. J. Clin. Microbiol. 41:5835. 13. Hoorfar, J., B. Malorny, A. Abdulmawjood, N. Cook, M. Wagner, and P. Fach. 2004. Practical considerations in design of internal amplification controls for diagnostic PCR assays. J. Clin. Microbiol. 42:1863–1868. 14. Hudson, J. A., N. J. King, A. J. Cornelius, T. Bigwood, K. Thom, and S. Monson. 2008. Detection, isolation and enumeration of Yersinia enterocolitica from raw pork. Int. J. Food Microbiol. 123:25–31. 15. 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. 16. Jourdan, A. D., S. C. Johnson, and I. V. Wesley. 2000. Development of a fluorogenic 5⬘ nuclease PCR assay for detection of the ail gene of pathogenic Yersinia enterocolitica. Appl. Environ. Microbiol. 66:3750–3755. 17. Kapperud, G. 1991. Yersinia enterocolitica in food hygiene. Int. J. Food Microbiol. 12:53–65. 18. Kapperud, G., T. Vardund, E. Skjerve, E. Hornes, and T. E. Michaelsen. 1993. Detection of pathogenic Yersinia enterocolitica in foods and water by immunomagnetic separation, nested polymerase chain reactions, and colorimetric detection of amplified DNA. Appl. Environ. Microbiol. 59:2938– 2944.

VOL. 74, 2008


19. Lake, R., A. Hudson, and P. Cressey. 2004. Risk profile: Yersinia enterocolitica in pork. New Zealand Food Safety Authority, Wellington, New Zealand. 20. Lambertz, S. T., K. Granath, M. Fredriksson-Ahomaa, K. E. Johansson, and M. L. Danielsson-Tham. 2007. Evaluation of a combined culture and PCR method (NMKL-163A) for detection of presumptive pathogenic Yersinia enterocolitica in pork products. J. Food Prot. 70:335–340. 21. Lantz, P. G., R. Knutsson, Y. Blixt, W. A. Al Soud, E. Borch, and P. Radstrom. 1998. Detection of pathogenic Yersinia enterocolitica in enrichment media and pork by a multiplex PCR: a study of sample preparation and PCR-inhibitory components. Int. J. Food Microbiol. 45:93–105. 22. Li, H., S. Bhaduri, and W. E. Magee. 1998. Maximizing plasmid stability and production of released proteins in Yersinia enterocolitica. Appl. Environ. Microbiol. 64:1812–1815. 23. Malorny, B., E. Paccassoni, P. Fach, C. Bunge, A. Martin, and R. Helmuth. 2004. Diagnostic real-time PCR for detection of Salmonella in food. Appl. Environ. Microbiol. 70:7046–7052. 24. Malorny, B., P. T. Tassios, P. Radstrom, N. Cook, M. Wagner, and J. Hoorfar. 2003. Standardization of diagnostic PCR for the detection of foodborne pathogens. Int. J. Food Microbiol. 83:39–48. 25. Miller, V. L., J. J. Farmer III, W. E. Hill, and S. Falkow. 1989. The ail locus

26.

27. 28. 29. 30. 31.

6067

is found uniquely in Yersinia enterocolitica serotypes commonly associated with disease. Infect. Immun. 57:121–131. Nesbakken, T., G. Kapperud, K. Dommarsnes, M. Skurnik, and E. Hornes. 1991. Comparative study of a DNA hybridization method and two isolation procedures for detection of Yersinia enterocolitica O:3 in naturally contaminated pork products. Appl. Environ. Microbiol. 57:389–394. Nordic Committee on Food Analysis. 2007. Evaluation of results from qualitative methods. NMKL procedure no. 20. NMKL, Oslo, Norway. Robins-Browne, R. M. 1997. Yersinia enterocolitica, p. 192–215. In M. P. Doyle, L. R. Beuchat, and T. J. Montville (ed.), Food microbiology, fundamentals and frontiers. ASM Press, Washington, DC. Sails, A. D., A. J. Fox, F. J. Bolton, D. R. A. Wareing, and D. L. A. Greenway. 2003. A real-time PCR assay for the detection of Campylobacter jejuni in foods after enrichment culture. Appl. Environ. Microbiol. 69:1383–1390. Thisted Lambertz, S., R. Lindqvist, A. Ballagi Pordany, and M. L. Danielsson Tham. 2000. A combined culture and PCR method for detection of pathogenic Yersinia enterocolitica in food. Int. J. Food Microbiol. 57:63–73. Vishnubhatla, A., D. Y. C. Fung, R. D. Oberst, M. P. Hays, T. G. Nagaraja, and S. J. A. Flood. 2000. Rapid 5⬘ nuclease (TaqMan) assay for detection of virulent strains of Yersinia enterocolitica. Appl. Environ. Microbiol. 66:4131– 4135.