Toward an International Standard for PCR-Based Detection of Food-Borne Thermotolerant Campylobacters: Assay Development and Analytical Validation

Abstract

As part of a European research project (FOOD-PCR), we developed a standardized and robust PCR detection assay specific for the three most frequently reported food-borne pathogenic Campylobacter species, C. jejuni, C. coli, and C. lari. Fifteen published and unpublished PCR primers targeting the 16S rRNA gene were tested in all possible pairwise combinations, as well as two published primers targeting the 23S rRNA gene. A panel of 150 strains including target and nontarget strains was used in an in-house validation. Only one primer pair, OT1559 plus 18-1, was found to be selective. The inclusivity and exclusivity were 100 and 97%, respectively. In an attempt to find a thermostable DNA polymerase more resistant than Taq to PCR inhibitors present in chicken samples, three DNA polymerases were evaluated. The DNA polymerase Tth was not inhibited at a concentration of 2% (vol/vol) chicken carcass rinse, unlike both Taq DNA polymerase and DyNAzyme. Based on these results, Tth was selected as the most suitable enzyme for the assay. The standardized PCR test described shows potential for use in large-scale screening programs for food-borne Campylobacter species under the assay conditions specified.

The species Campylobacter jejuni, C. coli, and C. lari are among the most frequently reported food-borne pathogens in humans worldwide (7). They can be found in a wide range of foods, including poultry, pig, beef, and seafood products, with chicken meat considered the most common source of human infection (12, 22). Effective methods for detecting these bacteria in foods are important tools for protecting the public health; however, detection of Campylobacter by standard isolation methods is problematic. In samples such as food, the agent may be present in low numbers, and the organisms are relatively sensitive to environmental factors, such as atmospheric oxygen, low pH, dryness, and temperature (22). Consequently, the number of viable Campylobacter cells can be rapidly and substantially reduced during storage or transportation of food samples to testing laboratories (18). Moreover, antibiotics used to improve the selectivity of culture media may inhibit the growth of certain strains if they are sensitive to one or more of the selective agents (8).

The application of culture-independent detection methods such as PCR may help to overcome the aforementioned problems (15). In addition, PCR in general provides faster results than conventional culture and has the potential for automation (9, 27). The latter is necessary for application of the test in large-scale screening programs in which many samples are examined in a short period of time. Many diagnostic laboratories have developed PCR-based methods for pathogen detection (5, 6, 9, 23, 28, 29), but many variables may affect the efficacy of PCR, and the results of tests developed or published by one laboratory can sometimes be difficult to reproduce by other laboratories (21). Moreover, PCR inhibitors originating from food samples may be difficult to overcome in PCR protocols using conventional enzymes: e.g., Taq polymerases (1). This may include testing different DNA polymerases in the matrices chosen for the study with the aim of identifying a polymerase that best overcomes the present inhibitors and validation of an internal amplification control (IAC) to identify false-negative responses. Proper validation based on consensus criteria is therefore an absolute prerequisite for successful adoption of PCR-based diagnostic methodology (10).

One of the aims of the European FOOD-PCR project (www.pcr.dk) was to evaluate and validate noncommercial PCR assays for the specific detection of C. jejuni, C. coli, and C. lari in foods. The present study evaluated 17 published and unpublished PCR primers targeting various rRNA gene regions. An extensive in-house validation was carried out through a new combination of published primers. Furthermore, in order to find a suitable enzyme resistant to inhibition by chicken samples, three DNA polymerases were investigated.

MATERIALS AND METHODS

Terms.

The terms used in this study refer to conventions described in the MicroVal protocol (3). Selectivity was defined as a measure of the degree of response from target and nontarget microorganisms and comprises inclusivity and exclusivity. Inclusivity describes the ability of a method (here PCR) to specifically detect the target pathogen from a wide range of strains, whereas exclusivity is the lack of response from a relevant range of closely related, nontarget strains (10). According to the new International Organization for Standardization (ISO) standard (3, 17), the terms “inclusivity” and “exclusivity” should replace the traditional terms “specificity” and “sensitivity,” which should only be used to express results from diagnostic samples (10).

Bacterial strains, growth conditions, and DNA extraction.

One hundred fifty strains (mainly Campylobacter spp.) were used in this study (Table 1). These included type, reference, and well-characterized field strains from various sources, including chickens, pigs, and cattle in Denmark, identified by conventional and molecular methods (19). All Campylobacter strains were cultured on 5% calf blood agar plates (CM331; Oxoid, Basingstoke, United Kingdom) under microaerobic conditions (6% O₂, 7% CO₂, 7% H₂, 80% N₂). All non-Campylobacter strains were grown in Luria-Bertani (LB) medium prepared from 5 g of sodium chloride, 5 g of yeast extract (L21; Oxoid) and 10 g of tryptone peptone (211705; Difco, Detroit, Mich.) dissolved in 1,000 ml of distilled water. The pH was adjusted to 7.3 to 7.4. The strains were stored as frozen cell suspensions in LB medium-glycerol (1:1) at −80°C. DNA was extracted from 2- to 3-day-old bacterial growth by using protocol no. 3 of the Easy-DNA kit (K1800-01; Invitrogen, Carlsbad, Calif.).

TABLE 1.

List of strains used for the development and validation of the PCR used in this study^a

No.	Species	Sero/biotype	Strain		No.	Species	Sero/biotype	Strain
1*	C. jejuni	Penner 1	CCUG 10935
2*	C. jejuni	Penner 2	CCUG 10936
3*	C. jejuni	Penner 3	CCUG 10937
4*	C. jejuni	Penner 4	CCUG 10938
5*	C. jejuni	Penner 5	CCUG 10959
6*	C. coli	Penner 5	CCUG 10939
7*	C. jejuni	Penner 6	CCUG 12778
8*	C. jejuni	Penner 7	CCUG 10940
9*	C. jejuni	Penner 8	CCUG 16436
10*	C. jejuni	Penner 9	CCUG 10942
11*	C. jejuni	Penner 10	CCUG 10943
12*	C. jejuni	Penner 11	CCUG 10944
13*	C. jejuni	Penner 12	CCUG 17625
14*	C. jejuni	Penner 13	CCUG 10945
15*	C. coli	Penner 14	CCUG 15360
16*	C. jejuni	Penner 15	CCUG 10946
17*	C. jejuni	Penner 16	CCUG 10947
18*	C. jejuni	Penner 17	CCUG 10948
19*	C. jejuni	Penner 18	CCUG 10949
20*	C. jejuni	Penner 19	CCUG 10950
21*	C. coli	Penner 20	CCUG 10951
22*	C. jejuni	Penner 21	CCUG 10952
23*	C. jejuni	Penner 22	CCUG 10953
24*	C. jejuni	Penner 23	CCUG 10954
25*	C. coli	Penner 24	CCUG 10955
26*	C. coli	Penner 25	CCUG 10956
27*	C. coli	Penner 26	CCUG 10957
28*	C. jejuni	Penner 27	CCUG 10958
29*	C. coli	Penner 28	CCUG 10959
30*	C. jejuni	Penner 29	CCUG 15361
31*	C. coli	Penner 30	CCUG 10960
32*	C. jejuni	Penner 31	CCUG 10961
33*	C. jejuni	Penner 32	CCUG 10962
34*	C. jejuni	Penner 33	CCUG 10963
35*	C. coli	Penner 34	CCUG 10964
36*	C. jejuni	Penner 35	CCUG 10965
37*	C. jejuni	Penner 36	CCUG 10966
38*	C. jejuni	Penner 37	CCUG 10967
39*	C. jejuni	Penner 38	CCUG 10968
40*	C. coli	Penner 39	CCUG 10969
41*	C. jejuni	Penner 40	CCUG 10970
42*	C. jejuni	Penner 41	CCUG 10971
43*	C. jejuni	Penner 42	CCUG 12782
44*	C. jejuni	Penner 43	CCUG 12783
45*	C. jejuni	Penner 44	CCUG 14567
46*	C. jejuni	Penner 45	CCUG 17753
47*	C. coli	Penner 46	CCUG 15362
48*	C. coli	Penner 47	CCUG 17715
49*	C. coli	Penner 48	CCUG 17754
50*	C. coli	Penner 49	CCUG 17755
51*	C. jejuni	Penner 50	CCUG 12790
52*	C. coli	Penner 51	CCUG 12791
53*	C. jejuni	Penner 52	CCUG 12792
54*	C. jejuni	Penner 53	CCUG 15013
55*	C. coli	Penner 54	CCUG 12794
56*	C. jejuni	Penner 55	CCUG 12795
57*	C. coli	Penner 56	CCUG 14537
58*	C. jejuni	Penner 57	CCUG 14538
59*	C. jejuni	Penner 58	CCUG 14539
60*	C. coli	Penner 59	CCUG 14540
61*	C. jejuni	Penner 60	CCUG 14541
62*	C. coli	Penner 61	CCUG 24865
63*	C. jejuni	Penner 62	CCUG 24866
64*	C. jejuni	Penner 63	CCUG 24867
65*	C. jejuni	Penner 64	CCUG 24868
66*	C. jejuni	Penner 65	CCUG 24869
67**	C. jejuni	Type strain	CCUG 11284
68*	C. jejuni	C. jejunib	CCUG 24567
69*	C. jejuni	C. jejunib	CCUG 18265
70*	C. jejuni	C. jejunib	CCUG 18266
71*	C. jejuni	C. jejunib	CCUG 26155
72*	C. jejuni	C. jejunib	CCUG 26152
73*	C. jejuni	C. jejunib	SSI 5384
74**	C. coli	Type strain	CCUG 11283
75*	C. lari	Type strain	CCUG 23947
76**	C. lari	NARTCc	CCUG 15035
77**	C. lari	NARTC	CCUG 12774
78**	C. lari	NARTC	CCUG 23949
79*	C. lari	NARTC	CCUG 19528
80*	C. lari	NARTC	CCUG 23948
81*	C. lari	NARTC	SVS 98-40052
82*	C. lari	UPTCd	CCUG 20707
83*	C. lari	UPTC	CCUG 18267
84*	C. lari	UPTC	CCUG 22395
85*	C. lari	UPTC	CCUG 18294
86*	C. lari	UPTC	CCUG 22396
87*	C. lari	UPTC	LU 6/3BW1
88*	C. lari	UPTC	LU 21/12OC3
89*	C. lari	UPTC	LU 16/1BTG4
90*	C. lari	UPTC	LU 18/3BTG8
91*	C. lari	UPTC	LU 21/12LW18
92*	C. lari	UPTC	LU 16/1OC3
93	C. jejuni	2	7231127-3
94	C. jejuni	4 complex	7231127-2
95	C. coli	46	9831503-2
96	C. coli	59	9831091-1
97	C. hyointestinalis		9930731-1
98	C. jejuni	35	7230701-6
99	C. jejuni	23,36	7230127-3
100	C. coli	51	7231059-1
101	C. coli	5	7231058-1
102	C. hyointestinalis		9930111-1
103	C. jejuni	11	99042253-4
104	C. jejuni	31	9904313-14
105	C. coli	5	9904602-19
106	C. coli	46	9904253-1
107	C. jejuni	29	7231125-2
108	C. jejuni	5	7231120-5
109	C. jejuni	19	723723-2
110	C. jejuni	4 complex	9904253-5
111	C. jejuni	6,7	9930116-6
112	C. coli	30	7230141-6
113	C. coli	46	7231033-1
114	C. coli	NT	9631023-2
115	C. coli	59	7231135-2
116	C. coli	48	9631038-4
117	C. hyointestinalis		7230324-3
118	C. hyointestinalis		9731034-3
119	H. pullorum-like		9831306-5
120	C. lari		9831299-3
121	H. pullorum-like		9831276-8
122**	C. upsaliensis	Type strain	CCUG 14913
123**	C. upsaliensis		CCUG 33890
124**	C. upsaliensis		CCUG 20818
125**	C. upsaliensis		CCUG 23017
126**	C. upsaliensis		CCUG 19559
127**	C. helveticus	Type strain	CCUG 30682
128**	C. helveticus		CCUG 34016
129**	C. helveticus		CCUG 30563
130**	C. helveticus		CCUG 30564
131**	C. helveticus		CCUG 30565
132**	C. helveticus		CCUG 30566
133**	C. helveticus		CCUG 30683
134**	C. helveticus		CCUG 34042
135	C. hyointestinalis	Type strain	CCUG 14169
136	C. lanienae	Type strain	NCTC 13004
137	C. mucosalis	Type strain	CCUG 6822
138	C. fetus	Type strain	CCUG 6823
139	C. concisus	Type strain	CCUG 13144
140	C. curvus	Type strain	CCUG 13146
141	C. showae	Type strain	CCUG 30254
142	C. rectus	Type strain	CCUG 20446
143	C. gracilis	Type strain	CCUG 27720
144	Arcobacter butzleri	Type strain	CCUG 30483
145	Helicobacter pylori		Rigsh. 15893
146	H. pullorum	Type strain	CCUG 33837
147	Escherichia coli		JEO 908149
148	L. monocytogenes		JEO 2268-179
149	Y. enterocolitica		JH2 O:3
150	S. enterica		CCUG 31969

^a*, strains used for testing the four published PCR methods; **, strains used in the preliminary study for identifying the primer pair with the best selectivity.

^bC. jejuni, subsp. doylei strain.

^cNARTC, nalidixic acid-resistant thermophilic Campylobacter.

^dUPTC, urease-positive thermophilic Campylobacter.

Selection of published primers.

rRNA gene sequences of C. jejuni, C. coli, and C. lari consistently share extensive homology, but are more distinct from other Campylobacter spp. (20). Thus, one probe (25) and three primer sets (Table 2) targeting 16S and one primer pair targeting 23S ribosomal DNA (rDNA) (4, 5, 24, 25) were tested on 105 Campylobacter isolates (Table 1). For testing the published PCR assays, the reaction conditions used, including temperature profile and DNA polymerase, were essentially as described in the original publications. The thermocycler used in this and subsequent studies was a GeneAmp PCR system 9700 (Applied Biosystems, Weiterstadt, Germany). After cycling, the PCR amplicons in this and subsequent studies were detected by electrophoresis in 1.8% agarose gels stained with ethidium bromide.

TABLE 2.

Primers used in different combinations to develop the best PCR assay for detection of C. jejuni, C. coli, and C. lari

Target	Primer (sequence)	Reference
23S	Therm1 (5′-TATTCCAATACCAACATTAGT)	6
	Therm4 (5′-CTTCGCTAATGCTAACCC)
16S	6-1 (5′-GTCGAACGATGAAGCTTCTA)	5
	18-1 (5′-TTCCTTAGGTACCGTCAGAA)
16S	CF1 (5′-GGAAGGATGACACTTTTCGGAGCG)	28
	CR2 (5′-TCGCGGTATTGCGTCTCATTGTATATGC)
16S	C442 (5′-GGAGGATGACACTTTTCGGAGC)	29
	C490 (5′-ATTACTGAGATGACTAGCACCCC)
16S	OT1559 (5′-CTGCTTAACACAAGTTGAGTAGG)	25
16S	18-1rev (5′-TTCTGACGGTACCTAAGGAA)	This study
16S	CF1rev (5′-CGCTCCGAAAAGTGTCATCCTCC)	This study
16S	C490rev (5′-GGGGTGGCTAGCCATCTCAGTATT)	This study
16S	JCL-1 (5′-ATAGTTAATCTGCCCTACACAA)	This study
16S	JCL-2 (5′-TCCTTTTCTTAGGGAAGAATTC)	This study
16S	JCL-3 (5′-CGTCAGAATTCTTCCCTAAG)	This study
16S	JCL-4 (5′-AGTTTAGTATTCCGGCTTCGA)	This study
16S	JCL-5 (5′-GATTCCACTGTGGACGGTAA)	This study
IACa	CCL-f (5′-OT1559-GGTTCATGAGGACACCTGAGTT)	This study
	CCR-r (5′-18-1-TATACACTCTCATCCCTCCAAC)	This study

^aIAC, primers used for construction of the internal amplification control.

New primer combinations.

Since none of the published primer sets resulted in the required selectivity (see Results), new primer combinations were tested with 18 strains (Table 1), chosen to identify assays capable of detecting C. jejuni, C. coli, and C. lari, but not the closely related, but not food-borne species C. upsaliensis and C. helveticus. The following PCR mixture (50 μl) was used: 10× PCR buffer, 0.2 mM deoxynucleoside triphosphates (dNTPs) (27-2035-03; Amersham Pharmacia Biotech, Buckinghamshire, United Kingdom), 0.2 μM each primer, 0.4 U of DNA polymerase, and 3 mM MgCl₂ (N808-0010; Applied Biosystems, Nærum, Denmark), and 1.0 μl of target DNA solution. The thermocycling program comprised an initial denaturation (94°C, 2 min) followed by 35 cycles of denaturation at (94°C, 1 min), annealing (55°C, 1 min), and extension at (72°C, 1 min). A final extension cycle (72°C, 4 min) completed the PCR. For preliminary optimization of the PCR and cycling parameters, the type strain of C. jejuni (CCUG 11284) was used. The optimized PCR mixture in 25 μl contained 10× PCR buffer for Tth DNA polymerase (1480022; Roche Applied Science, Hvidovre, Denmark), 0.2 mM dNTP, 0.22 μM primer OT1559, 0.24 μM primer 18-1, 1 U of Tth DNA polymerase (14800322; Roche Applied Science), 5 μg of bovine serum albumin (20 mg/ml) (711454; Roche Applied Science), 2 mM MgCl₂, 0.25 μl (≅10³ copies) of an internal control DNA (described below), and 1 μl of target sample DNA solution (≅100 pg ≅ 5 × 10⁴ copies). All PCRs were made in triplicate in 0.2-ml PCR tubes.

Final standard PCR.

The most selective new combination of primers OT1559 and 18-1 (see Results) was chosen as the final standard test. The final thermocycling program was as follows: initial denaturation 94°C at 2 min; then 35 cycles of denaturation at 94°C for 30 s, annealing at 58°C for 15 s, and extension at 72°C for 30 s; and finally an extension at 72°C for 4 min.

Sequencing.

The 16S rRNA gene sequence from strain CCUG 19559 was determined by a sequencing method as described previously (2). Alignment and numerical comparison of this sequence with GenBank database sequences of the type strains of all 16 Campylobacter species were performed with the program BioNumerics v2.5 (Applied Maths, Kortrijk, Belgium) using both default parameters and those described previously (20).

Construction of internal amplification control.

A 124-bp internal amplification control (IAC) amplicon was constructed based on DNA from the viral hemorrhagic septicema virus (GenBank accession no. X66134). This DNA was chosen since it is not found in food samples and has shown to work well previously (11). The IAC was produced in 50-μl reaction mixtures comprising 10 mM Tris (pH 8.3), 50 mM KCl, 2.5 mM MgCl₂, 0.2 mM dNTP, 0.1 μM each primer, 0.5 U of Taq polymerase (1146165; Roche Applied Science), and 2 μl of target DNA sample. The thermocycling program was as follows: 35 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s. The PCR products were purified from the agarose gel by using the QIAquick gel extraction kit (28704; Qiagen, Hilden, Germany) and finally eluated in 50 μl of sterile water. The fragment was furthermore cloned in plasmid pCR2.1 by using the TA cloning kit and One Shot TOP10 competent cells as recommended by the supplier (K2040-01; Invitrogen, Carlsbad, Calif.). Plasmids were recovered with the QIAprep Spin Miniprep kit (27104; Qiagen).

DNA polymerase and PCR inhibition test.

To identify a DNA polymerase resistant to PCR inhibitors present in chicken samples, a previously standardized PCR assay specific for pathogenic Yersinia enterocolitica was used (14). Different concentrations (from 1 fg/ml to 1 mg/ml) of DNA isolated from Y. enterocolitica Y79 (14) were added to the amplification mixture containing different percentage dilutions (vol/vol) of the chicken rinse sample (Table 3). All chicken samples had been certified free from naturally occurring pathogenic Y. enterocolitica by PCR. Whole chicken carcasses or neck skins were obtained from slaughterhouses or retailers in Denmark or Sweden. The chicken rinse samples comprised whole chickens washed in 500 ml of buffered peptone water (BPW) or sterile saline as described previously (13). Chicken neck skin samples were prepared by adding 10 g of neck skin to 100 ml of BPW or saline, homogenizing it in a stomacher for 30 s, and removing the skin sample. To test the PCR inhibitory effect of these samples, aliquots were added to the PCR mixture in a final concentration of 20% (vol/vol). Also, the inhibition of 10- and 100-fold-diluted chicken carcass rinse samples (respectively, 2 and 0.2% in the PCR mixture) were tested. Two potentially resistant enzymes, along with Taq, were tested (1). DyNAzyme (F501L; Finnzymes, Espoo, Finland), Taq and Tth DNA polymerases and accompanying buffer systems were evaluated for resistance to the inhibitory effect of chicken carcass rinse matrix. For real-time PCR, a LightCycler instrument (Roche Diagnostics) and a real-time assay based on the same Y. enterocolitica primer pair were used. The PCR mix contained 10× buffer supplied with the appropriate DNA polymerase (Taq, Tth, or DyNAzyme), 2.5 U of enzyme, 4 mM MgCl₂, 0.44 μM each primer, 0.2 mM each dNTP, 10,000-times-diluted SYBR Green I (1988131; Roche Applied Science), and 4 μl of sample. The total volume was 20 μl. The amplification conditions included a denaturation step of 1 min at 95°C, followed by 40 cycles of 0.1 s of denaturation at 95°C, 5 s of annealing at 60°C, and 15 s of elongation at 72°C, followed by a single fluorescent measurement and finally 25 s of final elongation. Amplification was followed by a melting curve analysis between 65 and 95°C and finally a cooling step for 1 min at 40°C. During amplification, the fluorescence was measured by using gain setting F1:1 with display mode F1.

TABLE 3.

Effect of inhibition by carcass rinse from chicken on PCR amplification with different DNA polymerases in a Y. enterocolitica PCR assay

DNA polymerase	Rinse sample (%)a	Result with Y. enterocolitica concen (CFU/25-μl reaction tube)d
		Conventional PCRb				Real-time PCRc
				5 × 103	5 × 102	5 × 101	5 × 100	5 × 103	5 × 102	5 × 101	5 × 100
Taq polymerase	Water	++	++	++	+−	++	++	++	−−
	20	−−	−−	−−	−−	−−	−−	−−	−−
	2	+−	−−	−−	−−	++	−−	−−	−−
	0.2	+−	++	+−	−−	++	++	++	−−
Tth polymerase	20	−−	−−	−−	−−	−−	−−	−−	−−
	2	++	++	++	−−	++	++	++	−−
	0.2	++	++	++	−−	++	++	++	−−
DyNAzyme II	20	−−	−−	−−	−−	−−	−−	−−	−−
	2	−−	++	−−	−−	++	++	−−	−−
	0.2	++	++	++	+−	++	++	++	−−

^aCarcass-rinse sample from neck skin of chicken in physiological saline.

^bTwo independent PCR results confirmed by gel electrophoresis.

^cTwo independent PCR results confirmed by melting-curve analysis in a LightCycler instrument.

^dOvernight culture of Y. enterocolitica Y79 in Tris-buffered saline with CFU determination by plating on TGE plates. Dilutions of the cell suspensions were made in physiological saline and added to the PCR mixture to a final amount of 20% (vol/vol).

RESULTS

Selectivity.

The results of the inclusivity and exclusivity tests are presented in Table 4. Only one primer pair, OT1559 plus 18-1, showed adequate selectivity. This primer pair was then tested in PCR against all 150 strains to verify its selectivity. The results showed that the inclusivity was 100%, whereas the exclusivity was 97%; only C. upsaliensis strain CCUG 19559 resulted in a positive PCR amplification. A comparison of the 16S rDNA sequence of strain CCUG 19559 and those of C. upsaliensis, C. jejuni, C. coli, and C. lari showed that the two primer annealing sites in strain CCUG 19559 were identical to sequences of the latter group of species. However, the 16S rRNA sequences of CCUG 19559 differed by 6 and 2 bp, respectively, in the primer-binding region compared to seven published C. upsaliensis 16S rRNA sequences. Nonetheless, the ca. 1,500-bp segment of the CCUG 19559 16S rDNA gene sequence was found to be 98.4% similar to C. upsaliensis 16S rDNA, compared with a corresponding value of 96.5% similarity for other C. jejuni strains.

TABLE 4.

Results from the preliminary screening of 26 primer combinations against DNA from 18 isolates and strains

Primer pair	Screening results for:
	Inclusivity (no. of strains)				Exclusivity (no. of strains)
		C. jejuni (n = 1)	C. coli (n = 1)	C. lari (n = 3)	No. of positive target strains/total true positive (%)	C. upsaliensis (n = 5)	C. helveticus (n = 8)	No. of negative nontarget strains/total true negative (%)
6-1 + 18-1	1	1	3	5/5 (100)	5	8	0/13 (0)
CF1 + CR2	1	0	3	4/5 (80)	4	8	1/13 (8)
C442 + C490	1	0	3	4/5 (80)	5	8	0/13 (0)
18-1rev + CR2	1	0	3	4/5 (80)	5	8	0/13 (0)
18-1rev + C490	1	0	3	4/5 (80)	5	8	0/13 (0)
OT1559 + 18-1a	1	1	3	5/5 (100)	1	0	12/13 (92)
OT1559 + C490	1	0	3	4/5 (80)	1	1	11/13 (85)
OT1559 + CR2	1	0	3	4/5 (80)	1	1	11/13 (85)
6-1 + CR2	1	0	3	4/5 (80)	5	8	0/13 (0)
JCL-1 + JCL-3	1	1	3	5/5 (100)	2	6	5/13 (39)
JCL-1 + JCL-4	0	0	0	0/5 (0)	0	0	13/13 (100)
JCL-1 + JCL-5	0	0	0	0/5 (0)	0	0	13/13 (100)
JCL-2 + JCL-4	1	0	2	3/5 (60)	0	1	12/13 (92)
JCL-2 + JCL-5	0	0	0	0/5 (0)	0	1	12/13 (92)
OT1559 + JCL3	1	1	3	5/5 (100)	2	5	6/13 (46)
OT1559 + JCL4	0	0	0	0/5 (0)	0	0	13/13 (100)
OT1559 + JCL5	0	0	0	0/5 (0)	0	1	12/13 (92)
JCL-1 + 18-1	1	1	3	5/5 (100)	2	6	5/13 (39)
C490rev + JCL-4	1	0	3	4/5 (80)	4	7	2/13 (15)
C490rev + JCL-5	1	0	3	4/5 (80)	2	2	9/13 (69)
18-1rev + JCL-4	0	0	0	0/5 (0)	1	1	11/13 (85)
18-1rev + JCL-5	0	0	1	1/5 (20)	1	2	10/13 (77)
6-1 + CF1rev	1	0	0	1/5 (20)	1	8	4/13 (31)
JCL-1 + CF1rev	1	1	3	5/5 (100)	4	8	1/13 (8)
OT1559 + CF1rev	1	1	3	5/5 (100)	2	4	7/13 (54)
C490 + CR2	0	0	0	0/5 (0)	0	0	13/13 (100)

^aNote that the OT1559 + 18-1 primer combination was the most selective new combination and was chosen as the final standard test.

IAC and detection limit.

A dilution series of the purified IAC fragment was made to determine the detection level in the final PCR. The IAC was coamplified with the target DNA (C. jejuni CCUG 11284) at 287 bp (Fig. 1). The detection limits for the IAC were 2.2 × 10⁻¹⁷ g (50 to 100 copies) when it was amplified alone and 2.2 × 10 ¹⁵ g (5 × 10³ copies) when it was amplified together with 10 pg of target DNA by using 35 amplification cycles. The detection limit for the target DNA (C. jejuni CCUG 11284) was 3.1 × 10⁻¹⁴ g (17 copies, assuming a genome size of 1.64 × 10⁶ bp) when it was amplified without IAC.

FIG. 1.

Result of the optimized PCR protocol detecting C. jejuni, C. coli, and C. lari (upper band), including the internal control (lower band). Lanes: M, molecular size marker; 1, positive control (CCUG 11284); 2, negative control (only internal control is added); 3 to 12, different strains of C. jejuni, C. coli, and C. lari; 13 to 20, other Campylobacter and non-Campylobacter species (strains marked with double asterisks in Table 1).

Evaluation of thermostable DNA polymerases.

Undiluted chicken rinse 20% (vol/vol) was found to completely inhibit PCR independent of the concentration of the target or the DNA polymerase was used. Taq and DyNAzyme polymerases were not inhibited when the chicken rinse was added to the PCR at a concentration of 0.2% (vol/vol), while Tth DNA polymerase showed no inhibition at a concentration of 2% (vol/vol) chicken rinse. Similar results were obtained for conventional and real-time PCR. Based on these results, Tth was selected as the most suitable DNA polymerase for the final Campylobacter PCR assay.

DISCUSSION

Selectivity was the principal criterion used to identify a PCR test for international validation as a tool for rapid and effective detection of C. jejuni, C. coli, and C. lari in foods. We aimed to identify an assay that included all strains of these three predominant food-borne Campylobacter species, but excluded all other species. The four published PCR assays evaluated based upon 16S and 23S rRNA gene sequences lacked the accepted selectivity to food-borne campylobacters, since they also yielded amplicons for C. upsaliensis and C. helveticus. This was considered disadvantageous, since domestic pets are the only known animal reservoir for these taxa, and C. helveticus has not yet been reported from human disease (19). Similar results have been reported with 23S rRNA gene sequence-derived PCR tests first proposed as selective for the identification of C. jejuni and C. coli in a study that emphasized the need for appropriate strain selection in the validation process (21). Subsequently, new combinations of existing primers, together with new primers were tested to improve the selectivity for C. jejuni, C. coli, and C. lari (Table 1). The primer pair with a 100% inclusivity score and the best exclusivity score was then tested in PCR against all 150 strains (Table 1) to assess its overall selectivity (Table 4). However, it was observed that two strains of C. helveticus appeared with a faint, nearly invisible band when an annealing temperature of 55°C was used, but these amplicons were not obtained with the final optimized cycling parameters (30 s at 94°C, 15 s at 58°C, and 30 s at 72°C for 35 cycles). After the change to these conditions, only one C. upsaliensis strain (CCUG 19559) was still detected by the assay. Given that both C. upsaliensis and C. helveticus are highly related to the food-borne species C. jejuni, C. coli, and C. lari (19, 20), this result is not altogether unexpected. A recent publication (4) showed that 28 of 29 hipO-negative Campylobacter isolates possessed 16S rRNA genes that were indistinguishable from those of C. jejuni type strains (based on 16S rRNA restriction fragment length polymorphism data). These hipO-negative isolates were found to be C. coli by the cumulative evidence of six published PCR-based assays, suggesting that speciation data based solely on this gene should be interpreted with caution. Furthermore, four 16S rRNA genes from hipO-negative strains were sequenced, which showed that they were almost identical to C. jejuni type strain 16S rRNA sequences deposited in GenBank. This observation is important given that others have reported problems with phylogenetic analyses of bacterial species based solely on 16S rRNA gene sequence comparison (26).

Moreover, the specific characteristics of the 16S rDNA sequence of CCUG 19559 infer that up to one-third of the gene may have been acquired from C. jejuni in a horizontal gene transfer event, a phenomenon that has attracted substantial credence in recent years (30). However, the fact that most of the C. upsaliensis and C. helveticus strains tested did not give an amplicon in the assay described indicates that the selectivity is acceptable.

Since the assay may be considered as an ISO or European international standard for detection of thermotolerant Campylobacter in food, it was of considerable importance to find the best DNA polymerase enzyme for the assay: i.e., that most resistant to PCR inhibitors naturally occurring in foods and chicken samples in particular. We evaluated DyNAyme, Tth, and Taq for their ability to withstand inhibitors from chicken rinse. To facilitate the evaluation of the effect of the sample matrix only, regardless of the specificity of the selected Campylobacter primers, an already validated PCR assay for detection of pathogenic Y. enterocolitica (14) was used as a test model.

The results indicated Tth to be the DNA polymerase of choice when examining chicken wash samples, since the PCR was substantially less inhibited when this enzyme was used compared with the Taq and DynaZyme polymerases. The improved performance of the PCR assay by use of Tth polymerase was observed in both the conventional and real-time PCR assays studied, which we consider to be an important observation. Real-time PCR assays are becoming of increasing importance in food quality matters, since they assess the level of contamination (and not simply the presence or absence) with a given pathogen. Based on the results obtained in chicken rinse, Tth DNA polymerase was chosen for international validation of the selected PCR assay with the highest specificity. This assay employs a novel combination of two previously published primers, the forward primer OT1559 (5) and the reverse primer 18-1 (25), and amplifies a 287-bp sequence of the 16S gene.

We conclude that the PCR test designed in the present study could form the basis of an accurate, standardized, and robust high-throughput, screening tool for enteropathogenic campylobacters in foods. Results from an international collaborative trial are described elsewhere (16).