Typing of Campylobacter jejuni Isolated from Turkey by Genotypic Methods,Antimicrobial Susceptibility,and Virulence Gene Patterns: A Retrospective Study

Abstract

In this retrospective study, typing ability, discriminatory power, and concordance between typing results obtained on 123 Campylobacter jejuni turkey isolates, collected in 1998, within 14 different farms, applying multilocus sequence typing (MLST), pulsed-field gel electrophoresis (PFGE), antibiotic resistance profile, and virulence gene pattern, were assessed and compared. Overall, 33 sequence types, 28 pulsotypes, 10 resistotypes, and 5 pathotypes were identified. MLST and PFGE showed the better discriminatory ability (i.e., Simpson's diversity index >0.90) as well as unidirectional (i.e., Wallace and adjusted Wallace coefficients >0.86) and bidirectional (i.e., adjusted Rand coefficient >0.60) concordance. Moreover, both methods showed a good unidirectional and bidirectional concordance with the resistotype. On the contrary, the congruence of both genotyping methods and resistotype with the pathotype seemed due to chance alone. A clonal relationship was identified among 66.7% of the isolates. Furthermore, 59.7% of the investigated isolates were resistant to two or more antimicrobials and 92% to tetracycline. All the isolates harbored cadF and pldA genes, whereas a flaA gene product and a cdtB gene product were amplified from 85.4% and 79.7% of the isolates, respectively, using the primers designed by Bang et al. (2003). The results of this study clarify the level of genetic diversity among the C. jejuni originating from turkeys. MLST level of correlation with PFGE, resistotype, and pathotype is assessed. This result supports the selection of type and number of typing methods to use in epidemiological studies. Finally, the identification of clonal complexes (i.e., groups of profiles differing by no more than one gene from at least one other profile of the group using the entire Campylobacter MLST database) shared between turkey and human isolates suggests that turkeys could be a possible source of Campylobacter infection.

Introduction

In 2013, the prevalence of Campylobacter within the European Union was 18.4% in the turkey slaughterhouses, 12.8% in the cutting plants, and 10.4% in turkey meat at retail (EFSA, 2015). Two features of Campylobacter infection have hindered investigations into the epidemiology of human campylobacteriosis: (1) human disease is most commonly sporadic, and (2) the bacterium is widespread and can be readily isolated from the intestine of different animals as well as environmental sources, water, and soil (Brown et al., 2004). Therefore, accurate isolate typing is essential to trace back to the source of infection. Among the genotyping methods available for Campylobacter typing, multilocus sequence typing (MLST) exploits the relative conservation in sequence of certain genes in which variations are more likely to be selectively neutral because of their housekeeping functions. This approach is recognized as the “gold standard” typing method for this bacterial genus (Sheppard et al., 2009). In the past, pulsed-field gel electrophoresis (PFGE) was considered the reference method for Campylobacter due to its high discriminatory power (Maslow et al., 1993). The drawbacks of PFGE are the time and labor necessary to perform the protocol properly. Moreover, restriction-based methods are less reproducible and more difficult to interpret than sequence-based methods.

Beside molecular methods, different isolates can be characterized according to their antibiotic resistance profile and pattern in virulence genes. The pathogenesis of Campylobacter infection is not fully elucidated, although flagella-mediated motility, adhesion to intestinal mucosa, and invasion and production of cytotoxin have been identified as virulence determinants (Konkel et al., 1997; Wassenaar, 1997; Datta et al., 2003). Therefore, four genes, each one responsible for one virulence factor, were investigated in this study; flaA, responsible for motility; cadF, for adhesion; pldA, for invasion; and cdtB, for toxin production.

When different typing methods are applied to a set of isolates, they should be compared to verify possible correlations in terms of results obtained in order to avoid duplications in terms of epidemiological information. Carrico et al. (2006) proposed the use of the adjusted Rand coefficient (AR) as a measure for the quantitative assessment of bidirectional correspondence between typing methods, taking into account that agreement may arise by chance alone. Severiano et al. (2011) proposed the adjusted Wallace (AW) and corresponding confidence intervals (CI) as quantitative measures of unidirectional congruence between typing methods.

Since (1) the majority of the available knowledge on typing results concerns broiler isolates, (2) more than one isolate was typed from each farm and animal, and (3) few publications are available on quantitative assessment of the concordance between typing results applied to the same set of isolates, in this research, typing ability, discriminatory power, and concordance between typing results collected on C. jejuni turkey isolates applying MLST, PFGE, antibiotic resistance profile, and virulence gene patterns were assessed and compared.

Materials and Methods

Sampling scheme

In March 1998, a total of 196 turkey ceca were collected in 2 slaughterhouses that gathered turkeys coming from 14 different farms located in Emilia-Romagna and Veneto regions (northern Italy). At the slaughterhouse, from each farm, 14 animals, belonging to the same group, were randomly selected from the line. Within 5 h after sampling, approximately 5 g of individual cecal contents were diluted 1:1 into saline solution (0.154 M NaCl), homogenized by vortex mixer, and diluted further 1:10 and 1:100 in the same solution. Ten microliters of these last two dilutions were then streaked onto two plates of Campylobacter Blood-Free Selective Agar (CCDA; Oxoid, Hampshire, UK). The plates were incubated at 37 ± 1°C under microaerobic atmosphere with hydrogen (Bolton et al., 1992) and after 48 h of incubation were checked daily up to 5 days for colonies. From each plate, four colonies of Gram-negative spiral-shaped bacteria that looked like Campylobacter spp. were subcultivated, cloned, submitted to a Campylobacter genus-specific polymerase chain reaction (PCR) (Linton et al., 1996), and stored at −80 ± 3°C for further typing. In 2014, with the beginning of this retrospective study, the isolates were thawed and submitted to a C. jejuni–C. coli-specific multiplex PCR (Denis et al., 1999). Among the C. jejuni isolates, one or more isolates for each animal were selected based on the colony morphological appearance. Overall, 123 C. jejuni isolates were submitted for further investigations.

MLST

The DNA of the C. jejuni isolates was purified using the Chelex^® 100 procedure (Biorad, Milan, Italy) and characterized by MLST as previously described (Dingle et al., 2001). Allele numbers, sequence types (STs), and clonal complexes (CCs) were assigned using the PubMLST database (http://pubmlst.org/campylobacter/) (Jolley and Maiden, 2010). The CCs have been defined as groups of profiles differing by no more than one gene from at least one other profile of the group (Feil, 2004) using the entire Campylobacter MLST database. Novel alleles and STs were coded through submission to the MLST database.

PFGE

The C. jejuni isolates were typed using PFGE with the use of SmaI (Zhou et al., 2011). Band patterns were analyzed with BioNumerics 7.5 software (Applied Maths, Keistraat, Belgium). DNA patterns were compared using the Dice coefficient and the unweighted-pair group method with arithmetic averages, with a 1.0% tolerance limit and 1.0% optimization. Isolates showing a PFGE pattern similarity level of 90% were grouped in the same “PFGE cluster.” The PFGE patterns were classified as pulsotypes.

Antimicrobial susceptibility testing

The resistotype (R-type) was obtained using the E-test method (AB BioMerieux, Solna, Sweden). Mueller-Hinton Agar II (Becton Dickinson, Franklin Lakes, NJ) containing 5% defibrinated sheep blood was used as medium. The antimicrobial agents tested were ciprofloxacin (CIP), enrofloxacin (EN), gentamicin (G), tetracycline (T), chloramphenicol (C), nalidixic acid (NA), ampicillin (AM), and erythromycin (ER). The plates were incubated at 37 ± 1°C under microaerobic atmosphere for 48 h. C. jejuni ATCC 33560, Staphylococcus aureus ATCC 29213 and Escherichia coli ATCC 25922 were used as control strains. NARMS (2012) cutoff values were used for all the antibiotics tested, except for enrofloxacin and ampicillin, for which the Clinical and Laboratory Standards Institute 2008 (CLSI, 2008) cutoff values for Enterobacteriaceae and Campylobacter were used, respectively. The following resistance breakpoints (μg/mL) were considered: ciprofloxacin ≥1, enrofloxacin ≥4, gentamicin ≥4, tetracycline ≥2, chloramphenicol ≥32, nalidixic acid ≥32, ampicillin ≥32, and erythromycin ≥8. The identified R-types were labeled using different numbers.

Detection of virulence genes

The DNA of the C. jejuni isolates was amplified by PCR to detect the presence of toxin and virulence genes (cdtB, cadF, flaA, and pldA), using primers VAT2–WMI-R (5′-GTTAAAATCCCCTGCTATCAACCA-3′; 5′GTTGGCACTTGGAATTTGCAAGGC3′), F2B-R1B (5′-TGGAGGGTAATTTAGATATG-3′; 5′-CTAATACCTAAAGTTGAAAC-3′), flaA-F-flaA-R (5′-GGATTTCGTATTAACACAAATGGTGC3′; 5′-CTGTAGTAATCTTAAAACATTTTG-3′) (Bang et al., 2003), pldA-84-Pld-981 (5′-AAGCTTATGCGTTTTT-3′; 5′ TATAAGGCTTTCTCCA3′) (Datta et al., 2003), respectively. C. jejuni ATCC 33560 was used as positive control and sterile water as negative control. Each pathotype was labeled as P-type.

Calculation of discriminatory power and congruence among typing methods

The discriminatory indexes (DI) of MLST, PFGE, R-type, and pathotype were calculated using the Simpson's diversity index described by Hunter and Gaston (1988). The Wallace's coefficient (W), the AW coefficient, and the AR coefficient, along with the respective CIs, were all calculated using the Comparing Partitions website (http://darwin.phyloviz.net/ComparingPartitions/index.php?link=Toll). As explained above, W and AW indexes assess the unidirectional concordance between typing methods and AR the bidirectional concordance.

Results

MLST

All 123 isolates were successfully typed by MLST. Five new alleles were discovered in this study. Overall, the isolates were classified in 33 STs: 11 were unique STs, identified in single isolates, whereas 22 were shared between 2 and 15 isolates. The majority of the common STs (i.e., 59.1%) were shared between isolates collected from turkeys reared in the same farm. The remaining STs were shared between isolates collected from animals housed in different farms, located in the same (i.e., 31.8%) or different regions (i.e., 9.1%). In 71.4% of the tested farms, the isolates were characterized by two to four different STs. The most spread ST was ST879, identified among isolates collected in three different farms, followed by STs 50, 227, 3334, 1880, 3476, and 7496 associated with isolates from two different farms. All the remaining STs were identified among isolates coming from a single farm.

PFGE

All 123 isolates were typed by PFGE. Overall, 28 pulsotypes were associated with the isolates genotyped; 13 were unique pulsotypes, identified in single isolates, whereas 15 were shared between 2 and 26 isolates. The majority of the common pulsotypes (i.e., 53.3%) were shared between isolates collected from turkeys reared in the same farm. The remaining pulsotypes were shared between isolates collected from animals housed in different farms, located in the same (i.e., 20%) or different regions (i.e., 26.7%). In 57.1% of the tested farms, the isolates were characterized by two to four different pulsotypes. The most spread pulsotypes were those labeled as 1 and 30, identified among isolates collected in five different farms, followed by pulsotype 11, associated with isolates from three farms, and pulsotypes 3, 14, 15, and 28, associated with isolates from two different farms. All the remaining pulsotypes were identified among isolates coming from a single farm.

Resistotype (R-type)

The distribution of the minimum inhibitory concentration values of the 123 isolates tested is shown in Table 1. All 123 isolates of C. jejuni were sensitive to gentamicin and chloramphenicol, whereas four isolates were sensitive to all antimicrobials tested. Furthermore, 92% of the tested isolates were resistant to tetracycline, 55% to nalidixic acid, 52% to enrofloxacin and ciprofloxacin, and 13% to ampicillin. Finally, one isolate was resistant to erythromycin. Overall, according to the resistance to the tested antimicrobials, 10 different R-types were identified and labeled with numbers. The R-type 1 (T) was found in 35% of the isolates; R-type 2 (CIP-EN-T-NA) in 43%; R-type 3 (CIP-EN-NA) in 2%; R-type 4 (AM) in 2%; R-type 5 (T-NA) in 3%; R-type 6 (T-AM) in 4%; R-type 8 (CIP-EN-T-NA-AM) in 6%; R-types 7 (CIP-EN-T-NA-ER), and 9 (CIP-EN-NA-AM) in 1% of the isolates, respectively; R-type 11 (S), referred to C. jejuni susceptible to all antibiotics tested, in 3% of the isolates.

Table 1.

Distribution of Minimum Inhibitory Concentration (MIC) Values Among the 123 Campylobacter jejuni Turkey Isolates

	Number of C. jejuni isolates with MIC of (μg/mL)
Antimicrobials	≤0.015	0.03	0.06	0.12	0.25	0.5	1	2	4	8	12	16	32	>32	64	128	>128	>256	MIC50	MIC90
Ampicillin							1	28	44	31		3	7		2			7	4	32
Ciprofloxacin			19	22	12	6								64					32	32
Enrofloxacin			21	24	7	5	1			18				47					8	>32
Nalidixic acid							2	22	24	4		3	2					66	>256	>256
Chloramphenicol							9	51	37	25		1							4	8
Erythromycin						1	61	42	18									1	1	4
Gentamicin				17	53	38	7	8											0.25	1
Tetracycline						3	7	5	5	3	2	4	4			4		86	>256	>256

Resistant isolates are represented in bold.

Pathotype (P-type)

All 123 C. jejuni isolates harbored the cadF gene responsible for adherence. The presence of pldA gene, responsible for invasion, was highlighted in all isolates, while the flaA gene, involved in isolate motility, was determined in 105 isolates (85.4%). As regards the cdtB gene, associated with toxin production, it was present in 98 (79.7%) C. jejuni isolates. Overall, five different P-types, labeled as numbers, were identified according to the pathotype (Fig. 1). The P-type 10 (cadF–flaA–pldA–cdtB) was identified in 68% of the isolates, followed by P-types 4 (cadF–flaA–pldA) (16%) and 5 (cdtB–cadF–pldA) (11%). The two remaining P-types, 7 (cadF–pldA) and 3 (cadF– flaA) were identified in four and two isolates, respectively.

FIG. 1.

Distribution of the identified pathotypes (P-type) as determined according to the presence of toxin and virulence genes (i.e., cdtB, cadF, flaA, and pldA). P-type 3 (cadF–flaA); P-type 4 (cadF–flaA–pldA); P-type 5 (cdtB–cadF–pldA); P-type 7 (cadF–pldA); P-type 10 (cadF–flaA–pldA–cdtB).

Isolate discrimination

The DI of the applied typing methods ranged between 0.946 for MLST and 0.499 for P-type (Table 2). Both MLST and PFGE showed a DI higher than 0.90. Therefore, they both show a probability to discriminate C. jejuni isolates different from a genetic point of view higher than 90%. The DIs calculated for each different typing method were significantly different (p < 0.001).

Table 2.

Discriminatory Power (DI) of Each Typing Method

Typing method	N profiles	N isolates with unique profile	N isolates classified in cluster of profiles	Cluster size	DI (CI 95%)
MLST	33	11	112	2–15	0.946^a (0.932–0.961)
PFGE	28	13	110	2–26	0.906^b (0.880–0.932)
R-type	10	2	121	2–53	0.690^c (0.635–0.745)
P-type	5	0	123	2–84	0.499^d (0.404–0.594)

Different superscript letters indicate DI values that are significantly different (p < 0.001).

CI, confidence interval; MLST, multilocus sequence typing; PFGE, pulsed-field gel electrophoresis; P-type, pathotype; R-type, resistotype.

Concordance among typing methods

The Wallace's coefficient (W) and the adjusted Wallace coefficient (AW) concerning unidirectional concordance between MLST, PFGE, R-type, and P-type (Table 3) showed a strong concordance (i.e., W and AW >0.86) between MLST and PFGE, as well as a good concordance (i.e., W and AW >0.73) between MLST and R-type. As explained in the Introduction, this result means that the classification, in terms of different or identical type strains obtained using MLST, can predict that found using PFGE and R-type. A sufficient level of concordance (i.e., W and AW >0.43) was observed also between PFGE and both MLST and R-type (Table 3). Finally, a very low degree of concordance (i.e., W and AW >0.097) was found between R-type and both MLST and PFGE. The P-type did not show concordance with the other typing methods, meaning that agreements between the results provided by P-type and one of the other methods is due to chance only (Table 3). In fact in Table 3, the W_i values calculated for the unidirectional concordance between P-type and MLST, PFGE and R-type are always included in the CI of the W coefficient for each combination of typing methods. Therefore, all the AW values are very low (Table 2). The results on the bidirectional concordance showed an AR >0.60 between MLST and PFGE only (Table 4). This AR value was significantly higher (p < 0.001) than the AR values for PFGE and R-type (AR 0.164) and MLST and R-type (AR 0.171). Finally, all the AR values between P-type and all the other typing methods were significantly lower (p < 0.001) of the AR values between the other combinations of typing methods.

Table 3.

Unidirectional Concordance Between the Applied Typing Methods

Typing method A	Typing method B	W_i	W _A→B (95% CI)	AW _A→B (95% CI)
MLST	PFGE	0.094	0.879 ^a (0.835–0.923)	0.866^a (0.818–0.915)
MLST	P-type	0.501	0.530^b (0.427–0.633)	0.057^d (0.000–0.264)
MLST	R-type	0.310	0.837 ^a (0.786–0.887)	0.736^a (0.690–0.837)
PFGE	MLST	0.054	0.505 ^b (0.420–0.590)	0.477^c (0.387–0.566)
PFGE	P-type	0.501	0.528^b (0.412–0.643)	0.053^d (0.000–0.284)
PFGE	R-type	0.310	0.613 ^b (0.504–0.722)	0.439^c (0.280–0.598)
P-type	MLST	0.054	0.057^f (0.037–0.077)	0.003^e (0.000–0.024)
P-type	PFGE	0.094	0.099^e (0.066–0.131)	0.005^e (0.000–0.041)
P-type	R-type	0.310	0.319^c (0.254–0.383)	0.012^d (0.000–0.105)
R-type	MLST	0.054	0.145 ^d (0.099–0.191)	0.097^d (0.048–0.145)
R-type	PFGE	0.094	0.185 ^d (0.134–0.236)	0.101^d (0.045–0.157)
R-type	P-type	0.501	0.515^b (0.406–0.624)	0.027^d (0.000–0.245)

Values in bold are those for which the correlation is significant and not due to chance alone.

Different superscript letters indicate that W and AW values are significantly different (p < 0.001).

AW, adjusted Wallace; CI, confidence interval; MLST, multilocus sequence typing; PFGE, pulsed-field gel electrophoresis; P-type, pathotype; R-type, resistotype; W, Wallace's coefficient.

Table 4.

Bidirectional Concordance Between the Applied Typing Methods by Adjusted Rand Coefficient (AR) and Jackknife Pseudovalues (95% CI)

	R-type	PFGE	P-type	MLST
R-type	—	—	—	—
PFGE	0.164^b (0.095–0.232)	—	—	—
P-type	0.017^c (0.000–0.083)	0.010^c (0.000–0.042)	—	—
MLST	0.171^b (0.108–0.234)	0.615^a (0.491–0.740)	0.006^c (0.000–0.026)	—

Different superscript letters indicate AR values significantly different (p < 0.001).

CI, confidence interval; MLST, multilocus sequence typing; PFGE, pulsed-field gel electrophoresis; P-type, pathotype; R-type, resistotype.

Type strains identified

The isolates tested were assigned to different type strains according to their ST, R-type, pulsotype, and P-type (Table 5). Clonal relationships, referred to isolates belonging to the same type strain, were identified for 66.7% of the tested farms (Table 5). They referred mainly to isolates from the same farm. However, isolates belonging to the same type strain isolated in different farms of the same region were also detected (Table 5). Overall, the number of type strains identified within the same farm ranged between 2 and 8.

Table 5.

Type Strains Identified in the Tested Farms ^a

Farm label	Farm location	No. turkeys	No. Campylobacter jejuni isolates	Type strain (no. isolates)
1	Emilia-Romagna	6	6	760-1-1-5 (2)
				1076-1-1-5 (1)
				7489-2-30-4 (1)
				7489-2-30-5 (2)
3	Emilia-Romagna	13	15	367-2-23-5 (11)
				367-2-23-3 (4)
5	Emilia-Romagna	7	7	450-6-22-5 (2)
				3476-1-6-5 (1)
				7493-2-16-5 (3)
				7494-1-11-5 (1)
6	Veneto	10	10	825-2-3-3 (1)
				825-8-18-3 (1)
				2141-8-18-5 (3)
				2141-8-18-4 (1)
				7496-4-28-5 (3)
				7497-1-3-5 (1)
7	Emilia-Romagna	9	11	44-2-7-5 (1)
				44-2-8-5 (1)
				44-2-9-5 (1)
				50-2-2-4 (1)
				50-2-3-3 (1)
				50-2-3-5 (5)
				7490-2-21-5 (1)
15	Emilia-Romagna	7	7	48-1-17-5 (1)
				2807-11-28-5 (3)
				3476-2-5-6 (1)
				7491-1-17-5 (2)
19	Veneto	3	3	50-1-4-3 (1)
				50-1-4-4 (2)
24	Emilia-Romagna	12	14	141-1-1-1 (1)
				141-1-1-4 (2)
				141-1-1-5 (11)
26	Emilia-Romagna	2	2	45-1-30-1 (1)
				45-1-30-2 (1)
27	Emilia-Romagna	10	11	227-1-14-4 (1)
				879-3-1-4 (1)
				879-5-1-4 (1)
				879-1-10-5 (1)
				1962-2-30-5 (4) 3334-2-1-5 (1)
				334-2-15-4 (1)
				3334-2-15-5 (1)
28	Emilia-Romagna	7	7	227-1-14-5 (2)
				879-5-1-5 (2)
				880-7-11-3 (1)
				3334-2-15-5 (1)
				7495-2-12-5 (1)
29	Emilia-Romagna	9	12	879-2-1-4 (1)
				879-2-1-5 (1)
				879-3-1-3 (1)
				879-5-1-5 (1)
				1962-2-30-3 (2)
				1962-2-30-5 (3)
				7500-11-26-5 (1)
				7501-2-13-1 (2)
49	Veneto	4	4	51-9-27-4 (1)
				2854-1-19-5 (1)
				7499-8-24-3 (2)
52	Veneto	11	14	880-1-11-2 (1)
				880-1-11-5 (8)
				7496-1-11-5 (1)
				7497-6-30-3 (1)
				7497-6-30-5 (2)
				7498-2-20-3 (1)

The type strains in bold were identified in different farms; the numbers in bold refer to isolates showing clonal relationships within the same farm.

Each type strain is defined by a combination of four numbers corresponding to the sequence type, pulsotype, resistotype, and pathotype, respectively.

Discussion

In this study, MLST and PFGE showed the better discriminatory ability, followed by R-type and P-type. Moreover, MLST and PFGE showed a good unidirectional and bidirectional concordance. A good unidirectional concordance was also observed between MLST and R-type, PFGE and both MLST and R-type, as well as R-type and the two investigated genotyping methods. These results seem to suggest that MLST only can be applied to trace clonal relationships between isolates. However, the isolates classified in the ST 3476, obtained in farms 5 and 15 displayed different pulsotypes and R-types; therefore, they would be improperly classified as clonal by MLST only. The same consideration can be extended to isolates with ST 50, collected in the farms 7 and 19, as well as to the isolates with ST 7496, from farms 6 and 52. This picture is important in order to approach the analysis of different typing data when available. However, the most promising technique in the analysis of outbreak-associated isolates, leading to faster and more precise source identification and to discriminate between alternative epidemiological hypotheses, is currently represented by whole-genome sequencing (WGS) (Harrison et al., 2013; Revez et al., 2014).

The CC21 identified in 23.4% of the turkey isolates included in this study has been previously described for chicken C. jejuni isolates (Ragimbeau et al., 2009; Sheppard et al., 2009; Colles and Maiden, 2012; Kittl et al., 2013). Moreover, additional CCs, such as CC443 and CC446, associated with 7.2% and 1.6% of the turkey isolates were already identified among chicken strains (Ragimbeau et al., 2008; Kittl et al., 2013). Interestingly, both CC21 and CC443, along with CC206 and CC257, associated with 4.8 and 12.1% of the turkey isolates, respectively, were already identified among human C. jejuni isolates and have been frequently associated with human diseases (Ragimbeau et al., 2008; Sheppard et al., 2009; de Haan et al., 2010; Colles and Maiden, 2012). The identification of CCs shared between turkey and human isolates suggests that turkeys could be a possible source of Campylobacter infection.

The 59.7% of the investigated isolates were resistant to two or more antimicrobials. Although the turkey isolates analyzed in this study were collected before the adoption of EU Regulation 1831/2003 banning the use of antibiotics as growth promoters in animal feeds, high occurrences of multidrug-resistant isolates in C. jejuni isolated from turkeys have been reported more recently. A high prevalence of tetracycline resistance in C. jejuni isolates from chickens and turkeys has been described (Andersen et al., 2006; Gu et al., 2009; Luangtongkum et al., 2009; Zhao et al., 2010). In this study, high resistance rates to (fluoro)quinolones were also observed. High level of tetracycline and (fluoro)quinolones resistances were also reported in C. jejuni human, animals, and food isolates in 2013 (EFSA, 2015).

Concerning the detection of virulence genes, the pathotype results do not correlate with other methods applied in this study; however, this research improves the knowledge on virulence genes of C. jejuni isolates from turkeys (Bang et al., 2004; Thorsness et al., 2008). The occurrence of the virulence genes observed in this study is in agreement with previous surveys performed on C. jejuni isolated from turkeys where cadF, pldA, and flaA genes resulted in the most conserved genes. The majority of our isolates were positive for all four virulence genes, in agreement with that observed by Thorsness et al. (2008), suggesting a potential ability of these Campylobacter to colonize turkeys and a potential capability to induce human Campylobacter infection (Konkel et al., 2000). However, infection is a multifactorial process, and studies on the regulation and expression of these genes in vitro and in vivo will be needed to better understand the role of these genes in the pathogenesis of campylobacteriosis.

Conclusions

The results of this study show that the population of C. jejuni isolates colonizing turkeys includes different-type strains within the same farm, which must be discriminated in order to make source attribution in case of human campylobacteriosis link to turkey or turkey products. Therefore, a representative number of animals and isolates must be collected during epidemiological investigations. Even if we already entered the WGS era, whenever an alternative typing method must be selected for typing C. jejuni, MLST is confirmed as the best choice. In fact, in this study MLST shows the best discriminatory ability and good correlation with PFGE and R-type. Even if this result seems to indicate to avoid running all methods on the same isolates, isolates sharing the same MLST profile with different PFGE and R-type were identified.

Footnotes

Disclosure Statement

No competing financial interests exist.

References

Andersen

, Saadbye

, Shukri

, Rosenquist

, Nielsen

, Boel

. Antimicrobial resistance among Campylobacter jejuni isolated from raw poultry meat at retail level in Denmark. Int J Food Microbiol, 2006; 107:250–255.

Bang

, Moller

, Scheutz

, Pedersen

, Handberg

, Madsen

. PCR detection of seven virulence and toxin genes of Campylobacter jejuni and Campylobacter coli isolates from Danish pigs and cattle and cytolethal distending toxin production of the isolates. J Appl Microbiol, 2003; 94:1003–1014.

Bang

, Bork

, Nielsen

, Scheutz

, Pedersen

, Madsen

. Detection of seven virulence and toxin genes of Campylobacter jejuni isolates from Danish turkeys by PCR and cytolethal distending toxin production of the isolates. J Food Prot, 2004; 67:2171–2177.

Bolton

, Wareing

DRA

, Skirrow

, Hutchinson

. Identification and biotyping of campylobacters. In: Identification Methods in Applied Environmental Microbiology. Board

, Jones

, Skinner

(eds.). Oxford: Blackwell Scientific Publications, 1992, pp. 151–161.

Brown

, Christensen

, Clough

, Diggle

, Hart

, Hazel

, Kemp

, Leatherbarrow

, Moore

, Sutherst

, Turner

, Williams

, Wright

, French

. Frequency and spatial distribution of environmental Campylobacter spp. Appl Environ Microbiol, 2004; 70:6501–6511.

Carrico

, Silva-Costa

, Melo-Cristina

, Pinto

, de Lencastre

, Almeida

, Ramirez

. Illustration of a common framework for relating multiple typing methods by application to macrolide-resistant Streptococcus pyogenes. J Clin Microbiol, 2006; 44:2524–2532.

[CLSI] Clinical and Laboratory Standards Institute. Performance Standards for Antimicrobial Disk Dilution Susceptibility Tests for Bacteria Isolated from Animals. Approved Standard, 3 ^rd ed. CLSI Document M31-A3. Wayne, PA: CLSI, 2008.

Colles

, Maiden

MCJ

. Campylobacter sequence typing databases: Applications and future prospects. Microbiology, 2012; 158:2695–2709.

Datta

, Niwa

, Itoh

. Prevalence of strains isolated from humans, poultry meat and broiler and bovine faeces. J Med Microbiol, 2003; 52:345–348.

10.

de Haan

, Kivisto

, Hakkinen

, Rautelin

, Hanninen

. Decreasing trend of overlapping multilocus sequence types between human and chicken Campylobacter jejuni isolates over a decade in Finland. Appl Environ Microbiol, 2010; 76:5228–5236.

11.

Denis

, Soumet

, Rivoal

, Ermel

, Blivet

, Salvat

, Colin

. Development of a m-PCR assay for simultaneous identification of Campylobacter jejuni and C. coli . Lett Appl Microbiol, 1999; 29:406–410.

12.

Dingle

, Colles

, Wareing

, Ure

, Fox

, Bolton

, Bootsma

, Willems

, Urwin

, Maiden

. Multilocus sequence typing system for Campylobacter jejuni . J Clin Microbiol, 2001; 39:14–23.

13.

[EFSA] European Food Safety Authority and European Centre for Disease Prevention and Control. The European Union Summary Report on Trends and Sources of Zoonoses, Zoonotic Agents and Food-borne Outbreaks in 2013. EFSA J, 2015; 13:74–85.

14.

Feil

. Small change: Keeping pace with microevolution. Nat Rev Microbiol, 2004; 2:483–495.

15.

, Siletzky

, Wright

, Islam

, Kathariou

. Antimicrobial susceptibility profiles and strain type diversity of Campylobacter jejuni isolates from turkeys in Eastern North Carolina. Appl Environ Microbiol, 2009; 75:474–482.

16.

Harrison

, Paterson

, Holden

, Larsen

, Stegger

, Larsen

, Petersen

, Skov

, Christensen

, Bak Zeuthen

, Heltberg

, Harris

, Zadoks

, Parkhill

, Peacock

, Holmes

. Whole genome sequencing identifies zoonotic transmission of MRSA isolates with the novel mecA homologue mecC. EMBO Mol Med, 2013; 5:509–515.

17.

Hunter

, Gaston

. Numerical index of the discriminatory ability of typing systems: An application of Simpson's index of diversity. J Clin Microbiol, 1988; 26:2465–2466.

18.

Jolley

, Maiden

MCJ

. BIGSdb: Scalable analysis of bacterial genome variation at the population level. BMC Bioinformatics, 2010; 11:595.

19.

Kittl

, Heckel

, Korczak

, Kuhnert

. Source attribution of human Campylobacter isolates by MLST and fla-typing and association of genotypes with quinolone resistance. PLoS One, 8 2013; e81796.

20.

Konkel

, Garvis

, Tipton

, Anderson

, Cieplak

. Identification and molecular cloning of a gene encoding a fibronectin-binding protein (CadF) from Campylobacter jejuni . Mol Microbiol, 1997; 24:953–963.

21.

Konkel

, Joens

, Mixter

. Molecular characterization of Campylobacter jejuni virulence determinants. In: Campylobacter, 2 ^nd ed. Nachamkin

, Blaser

. Washington, DC: American Society for Microbiology, 2000, pp. 217–240.

22.

Linton

, Owen

, Stanley

. Rapid identification by PCR of the genus Campylobacter and of five Campylobacter species enteropathogenic for man and animals. Res Microbiol, 1996; 147:707–718.

23.

Luangtongkum

, Jeon

, Han

, Plummer

, Logue

, Zhang

. Antibiotic resistance in Campylobacter: Emergence, transmission and persistence. Future Microbiol, 2009; 4:189–200.

24.

Maslow

, Mulligan

, Arbeit

. Molecular epidemiology: Application of contemporary techniques to the typing of microorganisms. Clin Infect Dis, 1993; 17:153–162.

25.

NARMS—National Antimicrobial resistance Monitoring System: Enteric bacteria.. 2012. Annual Human Isolates Report. Available at: http://www.cdc.gov/narms/reports/annual-human-isolates-report-2012.html, accessed July 16, 2014 .

26.

Ragimbeau

, Schneider

, Losch

, Even

, Mossong

. Multilocus sequence typing, pulsed-field gel electrophoresis, and fla short variable region typing of clonal complexes of Campylobacter jejuni strains of human, bovine, and poultry origins in Luxembourg. Appl Environ Microbiol, 2008; 74:7715–7722.

27.

Revez

, Llarena

, Schott

, Kuusi

, Hakkinen

, Kivistö

, Hänninen

, Rossi

. Genome analysis of Campylobacter jejuni strains isolated from a waterborne outbreak. BMC Genomics, 2014; 15:768.

28.

Severiano

, Pinto

, Ramirez

, Carriço

. Adjusted Wallace coefficient as a measure of congruence between typing methods. J Clin Microbiol, 2011; 49:3997–4000.

29.

Sheppard

, Dallas

, Strachan

, MacRae

, McCarthy

, Wilson

, Gormley

, Falush

, Ogden

, Maiden

, Forbes

. Campylobacter genotyping to determine the source of human infection. Clin Infect Dis, 2009; 48:1072–1078.

30.

Thorsness

, Sherwood

, Danzeisen

, Doetkott

, Logue

. Baseline Campylobacter prevalence at a new turkey production facility in north Dakota. J Food Prot, 2008; 71:2295–2300.

31.

Wassenaar

. Toxin production of Campylobacter spp. Clin Microbiol Rev, 1997; 10:466–476.

32.

Zhao

, Young

, Tong

, Abbott

, Womack

, Friedman

, McDermott

. Antimicrobial resistance of Campylobacter isolates from retail meat in the United States between 2002 and 2007. Appl Environ Microbiol, 2010; 76:7949–7956.

33.

Zhou

, Hussain

, Liles

, Arias

, Backert

, Kieninger

, Oyarzabal

. A simplified and cost-effective enrichment protocol for the isolation of Campylobacter spp. from retail broiler meat without microaerobic incubation. BMC Microbiol, 2011; 11:175.