Comparison of genotypes and antibiotic resistance of Campylobacter jejuni isolated from humans and slaughtered chickens in Switzerland


Bożena Korczak, Institute of Veterinary Bacteriology, University of Bern, Laenggass-Str. 122, CH-3001 Bern, Switzerland. E-mail:


Aims:  To get an overview of genotypes and antibiotic resistances in Swiss Campylobacter jejuni implicated in human gastroenteritis and to examine the association with isolates from chickens.

Methods and Results:  Multilocus sequence typing (MLST) and flaB typing were applied to 136 human clinical isolates. Phenotypic resistance to 12 antimicrobials and genotypic resistance to macrolides and quinolones were determined. MLST resulted in 35 known and six new sequence types (ST). The flaB analysis revealed 35 different types, which – in combination with MLST – increased the resolution of the typing approach. Resistance to quinolones, tetracycline and ampicillin was found in 37·5, 33·1 and 8·1% of the isolates, respectively, whereas macrolide resistance was found only once. Genotypic and phenotypic resistance correlated in all cases. A comparison to Camp. jejuni isolated from slaughtered chickens was performed. While 86% of the quinolone-sensitive human isolates showed overlapping MLST-flaB types with those of chicken origin, resistant strains showed only 39% of matching types.

Conclusion:  Mainly quinolone-sensitive Camp. jejuni strains implicated in human campylobacteriosis showed matching genotypes with isolates originating from chickens.

Significance and Impact of the Study:  A large proportion of human cases in Switzerland are likely to originate from domestic chickens, confirming that prevention measures in the poultry production are important.


Campylobacter jejuni lives as a commensal in the digestive tract of many warm-blooded animals, e.g. poultry, wild birds, ruminants, cats and dogs (Horrocks et al. 2009). In humans, however, it is the primary cause of acute bacterial gastroenteritis in many industrialized countries (Allos 2001; Olson et al. 2008). In Switzerland, in 2008, an incidence of 102·3 reports per 100 000 inhabitants was observed, which is one of the highest in Europe (European Food Safety Authority 2010). The disease is mainly caused by Camp. jejuni, which was found in 95% of the cases where species identification was performed (Bruhn 2009).

The infection can be contracted from a variety of sources, including undercooked contaminated meat and meat products, raw milk, contaminated water and close contact with carrier animals (Friedman et al. 2004; Olson et al. 2008). Case–control studies indicate poultry meat as the main risk factor (Wingstrand et al. 2006; Stafford et al. 2008). Studies based on multilocus sequence typing (MLST) also show the strongest association between human and poultry isolates (Wilson et al. 2008; Sheppard et al. 2009). Such data, however, are missing for Switzerland.

As severe or chronic infection necessitates antibiotic therapy, the worldwide increase in the antibiotic resistance to Campylobacter isolates is an issue of concern, particularly resistance to macrolides and quinolones, which are mainly considered for treatment (Allos 2001; Engberg et al. 2001; Alfredson and Korolik 2007; Blaser and Engberg 2008). This increase, as well as the number of resistant Camp. jejuni isolated from humans, has been linked to the frequent use of antimicrobials in poultry production (Alfredson and Korolik 2007). Resistance to these antimicrobials can be detected at the genetic level, where quinolone resistance is mainly determined by the C257T transition in the DNA gyrase gene (gyrA) and macrolide resistance is mostly a result of the point mutation A2075G in the 23S rRNA gene (Alfredson and Korolik 2007). In case of resistance to the aforementioned classes of antimicrobials, tetracycline, chloramphenicol or amoxicillin plus clavulanic acid may be considered for treatment of enteritis, while serious systemic infections necessitate treatment with aminoglycosides (Blaser and Engberg 2008). Therefore, monitoring the resistance pattern of human isolates provides information on the availability of further treatment options.

Data about the prevalence of certain genotypes of Camp. jejuni implicated in human disease and comparison to those present in livestock can help trace the main source of infection. MLST has been found to be especially suitable for this purpose, as it is precise, highly reproducible and allows for a simple worldwide comparison of genotypes (Maiden et al. 1998; Dingle et al. 2001). In addition, sequencing of the short variable region within the flagellin-encoding gene flaB can be used for further differentiation of strains showing the same sequence type (ST) (Mellmann et al. 2004; Dingle et al. 2008; Korczak et al. 2009).

As yet, there are few studies systematically comparing genotypes and antibiotic resistance of human and chicken isolates from the same time period (Levesque et al. 2008). This approach allows, on the one hand, to assess the contribution of chicken to human campylobacteriosis and, on the other hand, to evaluate the contribution of veterinary antibiotic use to the emergence of resistant Camp. jejuni implicated in human infections. As both macrolides and quinolones are licensed for therapeutical use in poultry farms in Switzerland, strains becoming resistant to these antimicrobials might potentially be transmitted from chickens to humans.

The current study provides an overview of the genotypes and the prevalence of antibiotic resistance of Camp. jejuni responsible for human morbidity in Switzerland. Additionally, more light is shed on the importance of chickens as a source of human campylobacteriosis. For this purpose, MLST and flaB typing were performed with 136 human Camp. jejuni isolates from 2008. Furthermore, fragments of the gyrA and 23S rRNA genes were sequenced to check for the presence of mutations responsible for quinolone and macrolide resistance. In addition, the minimal inhibitory concentrations (MIC) for ciprofloxacin, meropenem, nalidixic acid, amoxicillin/clavulanic acid (2 : 1 ratio), tetracycline, florfenicol, gentamicin, erythromycin, streptomycin, ampicillin, neomycin and chloramphenicol were determined. The data were compared to a recently investigated set of 243 isolates collected during the same year from Swiss broilers at slaughter (Wirz et al. 2010).

Materials and methods

Samples and sample preparation

The 136 Camp. jejuni isolates from human gastroenteritis cases had been collected by the Swiss National Centre for Enteropathogenic Bacteria (NENT) during the period from June to October 2008 representing the peak and thereby the majority of cases during this year (European Food Safety Authority 2010). Phenotypic species confirmation had been performed using routine diagnostic tests including microscopy, oxidase reaction, catalase reaction, hippurate hydrolysis and hydrolysis of indoxyl acetate along with the verification of the expected resistance to cephalothin. Cultures stored at −80°C were grown on tryptone soya agar plates with sheep blood (Oxoid AG, Pratteln, Switzerland) and incubated under microaerophilic conditions at 37°C for 24–72 h, depending on growth. For DNA template preparation, a few colonies from pure cultures were transferred into 450 μl of lysis buffer and incubated at 60°C for 1 h followed by heat treatment at 95°C for 15 min and stored at −20°C until further use (Korczak et al. 2009).

Multiplex PCR and sequencing

A multiplex strategy was applied according to a previously published protocol (Korczak et al. 2009). This involved amplification of the seven classical housekeeping genes used for MLST as well as flaB, 23S rRNA and gyrA. Slight modifications concerning the combination of primers in three amplification groups resulted in one group containing the primers for glnA, tkt, flaB and gyrA, the second group, those for aspA, glmM and 23S rRNA and the third group, those for atpA, gltA and glyA. The enzymatically purified PCR products were sequenced using the same primers as for PCR.

Antimicrobial susceptibility testing

The MIC for ciprofloxacin, meropenem, nalidixic acid, amoxicillin/clavulanic acid (2 : 1 ratio), tetracycline, florfenicol, gentamicin, erythromycin, streptomycin, ampicillin, neomycin and chloramphenicol was determined using the NLV44 Sensititre Custom Plate® (Trek Diagnostic Systems, Cleveland, OH, USA) according to the manufacturer’s instructions. Resistance was defined according to the following clinical breakpoints, which were also used by the Swiss Federal Veterinary Office (BVET) in the 2008 antibiotic resistance monitoring report: ciprofloxacin, ≥4 μg ml−1; meropenem, ≥16 μg ml−1; nalidixic acid, ≥64 μg ml−1; amoxicillin/clavulanic acid, ≥32 μg ml−1; tetracycline, ≥16 μg ml−1; florfenicol, ≥32 μg ml−1; gentamicin, ≥16 μg ml−1; erythromycin, ≥32 μg ml−1; streptomycin, ≥16 μg ml−1; ampicillin, ≥32 μg ml−1; neomycin, ≥16 μg ml−1; and chloramphenicol, ≥32 μg ml−1 (Buettner and Kuhn 2009).

Data analysis

Sequence analysis, determination of ST, clonal complex (CC) and antibiotic resistance were performed with the commercial online-based MLST application for Campylobacter provided by SmartGene (Zug, Switzerland), which uses the PubMLST database ( for ST designation. The same database was queried to determine the flaB type. Additionally, sequences were edited in Sequencher ver. 4.6 (GeneCodes, Ann Arbor, MI, USA) and entered into the BioNumerics program ver. 5.10 (Applied Maths NV, Sint-Martens-Latem, Belgium) for cluster analysis using the unweighted pair group method with arithmetic mean (UPGMA). Statistical analyses were conducted with the ncss 2007 software (NCSS, Kaysville, UT, USA) using Fisher’s exact test (two-tailed), with the level of significance set at a P < 0·05, and applying the exact binominal approach for the calculation of confidence intervals (CI). Discriminatory indices for the typing approaches were calculated using Simpson’s index of diversity (Hunter and Gaston 1988). The proportional similarity index (Czekanowski index) was calculated according to Rosef et al. (1985). The data generated in the study were compared to a recently investigated set of 243 isolates collected during 2008 from Swiss broilers at slaughter (Wirz et al. 2010).



All 136 Camp. jejuni strains could be grown in pure culture from frozen stock, and all MLST, flaB, gyrA and 23S rRNA target gene fragments could be successfully generated and analysed. A total of 41 STs were obtained, 35 of which were known STs, encompassing 128 isolates (94%), and six were new STs, encompassing eight isolates (6%). Three of the new STs (ST-4370, ST-4371, ST-4372) could be assigned to CC353, one (ST-4375) was assigned to CC22 and two (ST-4373, ST-4374) could not be assigned to any known CC (, access date 19 July 2010) (Table 1).

Table 1.   Distribution of clonal complexes (CC), sequence types (ST) and associated flaB types of human Campylobacter jejuni isolates from 2008. The number of isolates carrying the C257T transition, which contributes to quinolone resistance, per type is indicated
CCSTflaB typeNumber of isolatesQuinolone-resistant isolates

More than three-quarters of all isolates were grouped within the five most frequent CCs represented by CC21 (28·7%), CC257 (21·3%), CC48 (11·8%), CC206 (9·6%) and CC353 (6·6%). The predominant STs were ST-257 (20·6%), ST-21 (15·4%), ST-48 (11·8%), ST-50 (9·6%) and ST-122 (5·9%). Thus, 63·3% of the isolates were contained within the five most frequent STs, while as many as 25 STs were represented by single isolates. The discriminatory index for MLST was 0·909.

Analysis of the flaB sequences yielded 35 different types, of which 14 were represented by single isolates. The five most frequent flaB types were 103 (24·3%), 16 (14·7%), 36 (6·6%), 301 (6·6%) and 47 (5·2%) and thus amounted to 57·4% of the isolates (Table 1). With 0·907, the discriminatory index for flaB typing was similar to that for MLST.

Analysis of concatenated sequences consisting of the seven MLST and the flaB gene fragments yielded 57 different genotypes for the 136 isolates and increased the discriminatory index to 0·955. The five most frequent types, each including between five and 19 isolates, accounted for 44·1% of the isolates, while 36 types were found only once.

Genotypic antibiotic resistance

The point mutation A2075G (corresponding to A228G in our fragment) in the 23S rRNA gene, which contributes to macrolide resistance, was observed in only one isolate with ST-883 belonging to CC21.

The transition C257T (corresponding to C150T in our fragment) within the gyrA gene leading to quinolone resistance was present in 51 isolates (37·5%; 95% CI: 29·4–46·2%) including the one that showed the macrolide resistance mutation (Table 1). The transition was found with a significantly lower frequency in two of the five most common STs. Specifically, isolates with ST-48 and ST-122 showed the mutation only with frequencies of 12·5 and 0·0%, respectively. ST-572, on the other hand, was associated with quinolone resistance (P = 0·066).

Phenotypic antibiotic resistance

Phenotypic resistance to ciprofloxacin and erythromycin matched genotypic resistance in all cases.

Resistances to ciprofloxacin (37·5%) and tetracycline (33·1%) were the most prevalent, while there was no resistance to aminoglycosides, amoxicillin plus clavulanic acid, meropenem, florfenicol and chloramphenicol (Table 2). Multiple resistance defined as resistance to two or more classes of antimicrobials was found in 25·0% of the isolates (Table 3). Thereby, both tetracycline and ampicillin resistances were significantly associated with quinolone resistance (P < 0·01).

Table 2.   Antibiotic resistance of Campylobacter jejuni from human cases determined by MIC
AntibioticNumber of resistant strainsPrevalence of resistance (%)95% confidence interval (%)
  1. MIC, minimal inhibitory concentration.

Amoxicillin/clavulanic acid 2 : 1000–2·7
Nalidixic acid5137·529·4–46·2
Table 3.   Prevalence of multiple resistance among human Campylobacter jejuni isolates
Number of resistancesNumber of strainsPrevalence (%)95% confidence interval
1 Class of antibiotics3022·115·4–30·0
2 Classes of antibiotics2518·412·3–25·9
3 Classes of antibiotics85·92·6–11·3
4 Classes of antibiotics10·7 0·0–4·0

Significant differences in antibiotic resistance between STs were also observed. Isolates with ST-122 or ST-257 were significantly (P < 0·01 and P < 0·05, respectively) more likely to be susceptible to all antimicrobials tested than those with other STs. ST-21, on the other hand, showed a significant association with tetracycline resistance (P < 0·01).

Comparison with chicken isolates

The 136 Camp. jejuni isolates from humans were compared to 243 previously investigated Camp. jejuni isolates collected from Swiss chickens at slaughter from March to December 2008, covering the sample period of human isolates (Wirz et al. 2010). Twenty-one STs, including 114 human isolates (83·8%; 95% CI: 76·5–89·6%), overlapped with those found in the chicken sample set. The proportional similarity index (PSI) for MLST was 0·501. A similar overlap was observed for flaB types, with 20 types present in both sample groups, representing a total of 115 human isolates. However, only 107 of these isolates were the same as those found in the ST comparison. The PSI for flaB types was higher than for MLST reaching 0·583.

When comparing the concatenated MLST-flaB sequences, 101 human isolates (74·3%; 95% CI: 66·1–81·4%) still matched a chicken isolate. This number is lower than the 107 isolates having both an ST and a flaB type overlapping with the poultry group because four isolates showed an ST flaB type combination that was not present among the poultry isolates, and two possessed a mutation in the flaB (flaB type 36) or the gltA fragment outside the area used for allele designation. The isolate showing the gltA mutation was also the only one with a macrolide resistance.

To achieve maximum discrimination, the combined sequence of all 10 amplified fragments (glnA, tkt, flaB, gyrA, glmM, aspA, 23S rRNA, glyA, atpA and gltA) was compared between human and broiler isolates. With this approach, 88 human isolates (64·7%; 95% CI: 56·1–72·7%) were identical to at least one chicken isolate.

In addition, quinolone-susceptible and quinolone-resistant human isolates were separately compared to the respective groups of chicken isolates. A significant difference in their correspondence to the chicken isolates was observed. While 94% (95% CI: 87–98%) of the susceptible human isolates showed an ST that was also found in a susceptible broiler isolate, only 53% (95% CI: 38–67%) of the resistant human isolates had an ST also found among the resistant broiler samples. There was also a difference in the PSI, which was 0·512 for the quinolone susceptible and 0·429 for the resistant isolates.

Similar numbers were observed for flaB types with 55% (95% CI: 40–69%) of the resistant and 93% (95% CI: 85–97%) of the susceptible isolates overlapping with the respective broiler groups. The difference in the PSI was more pronounced, with 0·589 for the susceptible and 0·453 for the resistant strains.

Using the approach with concatenated MLST-flaB sequences, 39% (95% CI: 26–54%) of the resistant group and 86% (95% CI: 77–92%) of the susceptible group corresponded to the respective chicken isolates (Fig. 1). However, 16% of the resistant human isolates differed from the susceptible chicken isolates only in the gyrA C257T transition, which is responsible for quinolone resistance.

Figure 1.

 Comparison of the multilocus sequence typing flaB types of human and chicken isolates in two groups according to their quinolone resistance, dark grey: isolates differ from chicken isolates, light grey: isolates correspond to chicken isolates.


The aims of the present study have been to provide the first overview on the genotypes and the prevalence of antibiotic resistance of Camp. jejuni implicated in human gastroenteritis in Switzerland. In addition, the possible association between these human isolates and Camp. jejuni isolates collected during the same time from domestic chickens at slaughter was investigated (Wirz et al. 2010).

The most common CCs seen in the set of human isolates were CC21, CC257, CC48, CC206 and CC353, which are also frequently found in human isolates from other countries and are also numerous in the PubMLST database (Manning et al. 2003; Mickan et al. 2007; Levesque et al. 2008; Ragimbeau et al. 2008; Sheppard et al. 2009);, access date 19 July 2010).

Several studies comparing isolates from humans and various food animals found a high overlap of human with both poultry and ruminant isolates (Manning et al. 2003; Levesque et al. 2008; Ragimbeau et al. 2008; Wilson et al. 2008; Sheppard et al. 2009). ST-61, which seems to be the main ST consistently associated with cattle, was only found in two of the human isolates, indicating that cattle might not be a major source of human infection in Switzerland (Manning et al. 2003; Kärenlampi et al. 2007; Ragimbeau et al. 2008; Wilson et al. 2008; Sheppard et al. 2009).

However, a substantial overlap was apparent with recently investigated Swiss broiler isolates, in which ST-45 (14·0%), ST-257 (11·1%), ST-50 (7·4%), ST-21 (7·0%) and ST-48 (7·0%) were found to be predominant (Wirz et al. 2010). Four of these (ST-257, ST-21, ST-48 and ST-50) were also the most prevalent STs in the human sample set. Interestingly, ST-3963, which had been newly described in the Swiss broiler study (Wirz et al. 2010), was also found in one human isolate. This type has not yet been mentioned elsewhere and might therefore be endemic to Switzerland.

Surprisingly, ST-45, which was the most prevalent ST among the chicken isolates, was discovered in only one human isolate. This finding is even more astonishing as other studies have frequently found this ST in human, chicken, cattle and environmental isolates worldwide (Kärenlampi et al. 2007; Levesque et al. 2008; Sopwith et al. 2008; Sheppard et al. 2009). It is noteworthy that a recent study in New Zealand also observed ST-45 to be prevalent in poultry, livestock and wildlife in contrast to a far lower prevalence among human cases (Müllner et al. 2010). Strains of this type seem to be quite heterogeneous, as the 34 isolates of the chicken sample with ST-45 were associated with eight different flaB alleles. This raises the possibility that certain subtypes of ST-45 may be prevalent in animals, especially in poultry, but are of low virulence in humans.

The present study has shown both MLST and flaB typing to be suitable to investigate the epidemiology of Camp. jejuni. The two methods were found to be similar in their discriminatory ability for our set of human isolates with discriminatory indexes of 0·909 and 0·907, respectively. There was, however, no simple linear relationship between MLST and flaB types reflecting the different evolutionary backgrounds of the typing approaches and confirming the findings of previous studies (Wirz et al. 2010). As combining MLST and flaB typing increased the discriminatory index to 0·955, this approach might be advisable to reach the high discriminatory power desirable for short-term epidemiology. In the present study, an additional extended MLST analysis combining all ten fragments (MLST targets, flaB, gyrA, 23S rRNA) was used leading to a further increase in discrimination. It is, however, necessary to note that mutations in genes responsible for antibiotic resistance might rather indicate previous exposure to antimicrobials than phylogenetic relationship.

All four approaches (MLST, flaB typing, MLST-flaB and 10-fragment sequence comparison) indicated a substantial overlap between the human and chicken groups of isolates. Even the high discrimination method using all ten fragments showed 64·7% of the human strains matching broiler strains. This leads to the conclusion that domestic chickens might indeed be the main source of human campylobacteriosis in Switzerland.

A disconcertingly large proportion (37·5%) of the human isolates was found to be resistant to quinolones. The C257T transition in gyrA appears to be the main mechanism as all phenotypically resistant isolates showed this mutation. Interestingly, the 37·5% of resistant human isolates were significantly greater than the 18·9% found in Swiss broiler isolates (Wirz et al. 2010). In addition, the quinolone-susceptible human isolates showed a far greater overlap with the respective chicken isolates than did the resistant ones. This was most apparent when comparing the combined MLST-flaB types, which corresponded to those of the chicken isolates in 86% of the susceptible and only 39% of the resistant human isolates (Fig. 1). However, an additional 16% of the resistant isolates differed from the chicken isolates only in the gyrA C257T transition, which is responsible for quinolone resistance. These cases, nevertheless, might actually originate from chickens, assuming that the Camp. jejuni strains involved acquired the mutation upon infection and subsequent antibiotic treatment: a plausible scenario as resistance has been shown to emerge rapidly under treatment with quinolones in both chickens and humans (Adler-Mosca et al. 1991; Wretlind et al. 1992; McDermott et al. 2002; Luo et al. 2003). However, verification of this assumption was not possible because of a lack of information about antibiotic treatment prior to stool sampling of the patients included in this study.

Nevertheless, our findings indicate that infections with quinolone-resistant Camp. jejuni in Switzerland are less likely to have domestic chicken as a major source than infections with susceptible strains. This observation is consistent with the results of three case–control studies from Canada, Denmark and the United Kingdom where consumption of chicken has also been found not to be a risk factor for quinolone-resistant Campylobacter infection (Engberg et al. 2004; Johnson et al. 2008; Evans et al. 2009). Nonetheless, 39% of the resistant human strains showed a full match with genotypes found in slaughtered chickens. Thus, quinolone use at Swiss poultry farms might contribute to the increased resistance of human isolates; it seems, however, that there are other yet unknown factors. Foreign travel might be one explanation, as it has been recognized as a major risk factor for infections with quinolone-resistant strains in other industrialized countries (Engberg et al. 2004; Johnson et al. 2008; Evans et al. 2009). This interpretation would correspond to the observation that the human isolates with genotypes other than those of the chicken isolates were rather diverse. Furthermore, the rate of quinolone resistance is lower among Swiss chicken isolates than that in most other European countries, which might be potential travel destinations (European Food Safety Authority 2010). Unfortunately, information concerning the travel history of the patients selected for this study was unavailable; thus, the answer to this question must await further research.

Tetracycline resistance was also pronounced among the human isolates (33·1%). In addition, it was significantly higher than tetracycline resistance of isolates from slaughtered chickens from the same time period (19·1%) (Buettner and Kuhn 2009). Isolates with MLST-flaB types that matched those of chickens were less likely to be resistant to tetracycline even though the association was not significant (P = 0·094). Nevertheless, infections contracted abroad might explain this observation as has been proposed for quinolone resistance, especially as tetracycline resistance is higher among poultry isolates in those European countries that are popular travel destinations (European Food Safety Authority 2010).

The prevalence of ampicillin resistance was slightly lower among human than among chicken isolates (8·1 vs 13·0%) but no significant difference was apparent (Buettner and Kuhn 2009).

In contrast to quinolone and tetracycline resistance, macrolide resistance among human Camp. jejuni isolates remains low in most countries, Thailand being an exception with 12% (Gibreel and Taylor 2006). In the present study, only one isolate was resistant, confirming this observation for Switzerland. Therefore, macrolides might be the drugs of choice for initial treatment before results of susceptibility tests can be obtained.

In conclusion, it is highly probable that domestic chickens are the source of the majority of human campylobacteriosis cases in Switzerland. This, however, does not hold true for infections with quinolone and maybe tetracycline-resistant strains where other yet unknown sources, such as foreign travel, appear to be predominant.


We thank Dr Gudrun Overesch for constructive discussions and Elisabeth Lüthi, Andreas Thomann and Yvonne Schlatter for technical help. This research was supported by a Swiss Federal Veterinary Office grant (1.08.12).