Comparison of Campylobacter jejuni isolates from human, food, veterinary and environmental sources in Iceland using PFGE, MLST and fla-SVR sequencing


Viggó Th. Marteinsson, Matís ohf. Icelandic Food and Biotech R&D, Vínlandsleið 12, 113 Reykjavík, Iceland. E-mail:


Aims: Campylobacter jejuni isolates from various sources in Iceland were genotyped with the aim of assessing the genetic diversity, population structure, source distribution and campylobacter transmission routes to humans.

Methods and Results:  A collection of 584 Campylobacter isolates were collected from clinical cases, food, animals and environment in Iceland in 1999–2002, during a period of national Campylobacter epidemic in Iceland. All isolates were characterized by pulse field gel electrophoresis (PFGE), and selected subset of 52 isolates representing the diversity of the identified PFGE types was further genotyped using multilocus sequence typing (MLST) and fla-SVR sequencing to gain better insight into the population structure.

Conclusions:  The results show a substantial diversity within the Icelandic Campylobacter population. Majority of the human Campylobacter infections originated from domestic chicken and cattle isolates. MLST showed the isolates to be distributed among previously reported and common sequence type complexes in the MLST database.

Significance and Impact of the Study:  The genotyping of Campylobacter from various sources has not previously been reported from Iceland, and the results of the study gave a valuable insight into the population structure of Camp. jejuni in Iceland, source distribution and transmission routes to humans. The geographical isolation of Iceland in the north Atlantic provides new information on Campylobacter population dynamics on a global scale.


Campylobacter jejuni is one of the most frequently reported causes of bacterial gastroenteritis in many developed countries (Altekruse et al. 1999; Petersen and Wedderkopp 2001). Surveys have suggested that the incidence of diseases associated with Campylobacter infections can be as high as 1% of the population per annum (Mead et al. 1999). The reported annual incidence of campylobacteriosis in Iceland over the last 10 years is on average 42 patients per 100 000 inhabitants. Around half of the cases are of domestic origin. The incidence has been steadily declining following an epidemic in 1998–2000, during which time the reported incidence of domestically acquired infections increased from 7 to 118 patients/10 000 inhabitants over the course of these 3 years (Jore et al. 2010).

The bacterium is widespread in the environment constituting a part of the natural intestinal flora of wild and domesticated animals and birds. A wide range of zoonotic and environmental sources have been identified as risk factors for human Campylobacter infection (Altekruse et al. 1999; Petersen and Wedderkopp 2001; Petersen et al. 2001; Neimann et al. 2003; Colles et al. 2008). Outbreaks are well documented but rare, and the majority of cases are sporadic, for which the route of infection is rarely established. The principal route of transmission is thought to be contaminated food, especially chicken, while un-pasteurized milk and surface waters are also recognized as a potential source of infection (Altekruse et al. 1999).

The wide distribution and high levels of genetic diversity and frequent recombination among Camp. jejuni make it difficult to identify transmission sources of sporadic infections to humans. Reliable strain differentiation is necessary for the identification of transmission routes. Various phenotyping and genotyping methods have been developed for Campylobacter strain typing. Pulse field gel electrophoresis (PFGE) is a highly discriminatory method that has shown to be very useful for defining clones and lineages within Campylobacter populations (On et al. 1998; Fitzgerald et al. 2001; Clark et al. 2005) and to successfully confirm Camp. jejuni outbreaks (Levesque et al. 2008).

In recent years, multilocus sequence typing (MLST) has emerged as the state-of-the art method for determining the population structure of various pathogens and an MLST system has been developed for Camp. jejuni (Maiden et al. 1998; Dingle et al. 2001a, 2002). The lower discriminatory power of MLST in comparison with PFGE can be increased by including a more variable locus in the MLST, e.g. the fla-SVR locus which gives the method discriminatory power similar to PFGE and makes it suitable for outbreak studies (Manning et al. 2003; Sails et al. 2003a; Clark et al. 2005; Price et al. 2006).

This paper describes the molecular typing of a collection of Camp. jejuni isolates from various sources in Iceland using three different typing methods; PFGE, MLST and fla-SVR sequencing. The aim of this study was to identify by PFGE typing the distribution of Campylobacter in various animals, environment and food products in Iceland and compare those isolates to others from infected humans to identify the route of transmission. Furthermore, the aim was to subtype a selection of isolates using MLST and fla-SVR sequencing methods to gain better insight into the population structure of Campylobacter in Iceland.


Campylobacter jejuni isolates

A total of 584 Camp. jejuni isolates collected from various sources in Iceland between May 1999 and August 2002 were used for the study. Campylobacter jejuni isolates from all sources in Iceland that could be acquired during the study period were taken for PFGE analysis. Clinical isolates were 314, with 275 domestic infections and 39 isolates acquired from Icelandic citizens infected during international travel. Isolates originating from animals, birds and other sources were 270, with 140 from chicken (there of 93 from 47 broiler chickens at retail level), 14 from ducks, 2 from geese, 3 from turkey, 66 from cattle, 15 from sheep, 6 from pig, 1 from dog, 1 from raven and 22 environmental isolates (6 from soil and 16 from rivers or springs).

Pulse field gel electrophoresis, PFGE

PFGE was performed according to the CAMPYNET protocol ( Briefly, isolates were subcultured on blood-agar media at 42°C for 2–3 days under micro-aerobic atmosphere. Bacterial colonies were harvested and re-suspended in 0·5 ml of PIV buffer (1 mol l−1 NaCl, 10 mmol l−1 Tris, 10 mmol l−1 EDTA). About 300 μl of suspension was subsequently mixed with 700 μl of 1% InCert agarose (FME BioProducts, Rockland, ME, USA). The mixture was moulded into plugs using 1-ml syringes as mould and plugs allowed to set at 4°C for c. 10–15 min. The agarose plugs were placed in ESP buffer (0·5 mol l−1 EDTA, 1% Sarcosyl, 1 mg ml−1 Proteinase K) and incubated at 56°C over night. The plugs were washed in TE (10 mmol l−1 Tris, 1 mmol l−1 EDTA) buffer six times for 0·5–1 h and the last wash over night. Thin slices of the agarose plug were digested overnight with SmaI (New England BioLabs, Beverly, MA, USA) at 24°C, and the resulting macrorestriction digests were electrophoresed in 1·4% agarose gel in 10× TBE buffer (0·9 mol l−1 Tris, 0·9 mol l−1 boric acid, 0·5 mol l−1 EDTA) at 6 V cm−1. Pulsing was ramped from 0·5 to 40 s over 22·5 h at 12°C. The gels were stained with ethidium bromide for 7 min, destained in water for 20 min and photographed under UV light with GelDoc2000 photoscanner. A lambda ladder PFG marker (New England BioLabs) was used for fragment size determination.

Multilocus sequence typing, MLST

A subset of 52 isolates from poultry, human, cattle, sheep and environmental origin, selected using a stratified approach, to represent the proportional distribution of PFGE types within the total population, was characterized by MLST using the method described by Dingle et al. (2001a). Briefly, DNA was isolated from bacterial cultures using either Chelex resin (Bio-Rad Laboratories, Hercules, CA, USA) or was recovered using the Promega Magnesil KF, Genomic system (MD1460) DNA isolation kit (Promega Corporation, Madison, WI, USA) in combination with KingFisher magnetic beads automatic DNA isolation instrument (Thermo Labsystems, Waltham, MA, USA) according to the manufacturers recommendations. Alternate primers sets for seven housekeeping genes were used for PCR magnification as described previously (Dingle et al. 2001a). Annealing temperature was lowered 2°C where amplification was not initially successful. Amplification products of the correct size were identified, and PCR products were purified and prepared for sequencing using Exo/Sap (GE Healthcare, Piscataway, NJ, USA). The nucleotide sequence was determined by using appropriate primers and BigDye reaction mix (Applied Biosystems, Foster City, CA, USA) according to the manufacturer recommendations. The reaction products were separated and detected using ABI prism 3730 automated DNA sequencer. The chromatograms were analysed using Sequencher 4.2 (GeneCodes Corp., Ann Arbor, MI, USA). Sequences were trimmed to a corresponding reference sequence for individual gene, and the MLST database was queried (

Fla-SVR sequencing

FlaA short variable region (SVR) sequencing was performed on the same subset of 52 isolates according to the standardized Oxford protocol (Clark et al. 2005). The extracted DNA (same as for MLST) was used as a template in a PCR, using primers described previously (Meinersmann et al. 1997). Amplification products were verified by gel electrophoresis. PCR products were purified and sequenced as described earlier. The sequences were edited to the previously described length of 321 base pairs corresponding to nucleotides 283–603 of the flaA coding sequence (Meinersmann et al. 1997). The sequences were assigned allelic numbers based on the data already present in the Campylobacter fla-SVR database ( For strains with possible new fla-SVR alleles, DNA trace files were submitted to the database administrator for confirmation.

Data analyses

The associations between PFGE types and source were evaluated by chi square test of independence. Statistical significance is reported at three levels: < 0·05, <0·005 and <0·0005. Simpson’s diversity index of discrimination (D) was calculated as previously described by Hunter and Gaston (1988). upgma (unweighted pair group method with arithmetic means) dendogram was constructed using the START2 (sequence type assignment recombinant test) (Jolley et al. 2001).


Diversity of Campylobacter jejuni PFGE genotypes

A total of 584 Campylobacter isolates were subjected to PFGE typing using SmaI in the study. Of these, 45 isolates were nontypeable. The results from the remaining 539 isolates are presented here. Their distribution of typed isolates based on origin and PFGE types are summarized in Fig. 1. A total of 144 PFGE profiles were differentiated among the 539 characterized isolates. Among the 144 observed PFGE types, 80% were represented by only 1 or 2 isolates. The Simpsons diversity index for the domestic Campylobacter isolates was 0·923, indicating considerable diversity in the Icelandic Campylobacter population. A noticeably less genetic diversity of 0·789 was observed from chicken isolates (Table 1). One-third of the foreign-originated isolates belonged to profiles also detected in the domestic population. The ten most common profiles included 57% of the entire set of isolates (Fig. 1b). Two PFGE profiles (types 1 and 7) were dominant and included 34% of the isolates and 22 and 12% for types 1 and 7, respectively. Both types were almost exclusively found in humans and chicken; a single sheep isolate was found to be of type 1. The majority (96%) of the type 1 isolates were detected during 1999 and 2000. Type 7 was detected throughout the investigated period.

Figure 1.

 Distribution of the 539 pulse field gel electrophoresis (PFGE) typed isolates by source (a) and by PFGE types (b). PFGE types detected in five isolates or less are collectively represented as ‘other’. PFGE types indicated by an asterisk (*) are the ten most common types in the population. (a) (inline image) Chicken (24%); (inline image) human (52%); (inline image) cattle (12%); (inline image) pig (1%); (inline image) environment (4%); (inline image) sheep (3%); (inline image) duck (3%) and (inline image) other (1%). (b) (inline image) 1* (22%); (inline image) 7% (12%); (inline image) 15* (4%); (inline image) 8* (4%); (inline image) 9* (4%); (inline image) 116* (4%); (inline image) 59* (3%); (inline image) 7a* (2%); (inline image) 6a* (2%); (inline image) 42* (2%); (inline image) 7b (1%); (inline image) 41 (1%); (inline image) 57 (1%); (inline image) 78 (1%); (inline image) 47 (1%); (inline image) 56 (1%) and (inline image) other (34%).

Table 1.   Genetic diversity (Simpson’s diversity index) of Campylobacter isolates from humans, chicken and cattle
 Origin of isolates
  1. PFGE, pulse field gel electrophoresis.

Isolates no. (total n)28613064541
PFGE profiles no.843031151
Diversity index0·9160·7890·9600·929

PFGE types and source distribution of Campylobacter jejuni from human, chicken, ruminant (cattle and sheep) and environmental isolates

Among the 45 PFGE types detected in more than two isolates, 55% (25) were observed from more than one source. Table 2 shows the percentage of PFGE types from humans, chicken and cattle also associated with other sources. Only 5% of the PFGE types found in chicken were found in cattle isolates. However, 28% of the cattle isolates were also found among chicken. Four PFGE types (15, 42, 59 and 84) were found among isolates of human, chicken and ruminant (cattle and sheep) origin.

Table 2.   The percentage of pulse field gel electrophoresis types found among human, chicken and cattle isolates that were associated with other sources
 Human (%)Chicken (%)Cattle (%)

Categorical analysis revealed that the PFGE types were not independently distributed across sources, humans, poultry, cattle and environment. Further discriminative analysis on the most abundant types was performed, using two-by-two tables to analyse the association between specific PFGE types and individual source. These analyses revealed a number of highly significant associations between specific PFGE types and specific sources, including two exclusive associations (Table 3). PFGE type 116 was exclusively associated with clinical cases, while type 47 was exclusively found in environmental samples. Type 7 was significantly overrepresented in the human isolates, while type 1 was significantly overrepresented among chicken isolates and significantly underrepresented among ruminant and environmental isolates. Various PFGE types (type 15, 41, 42, 56, 57, 59 and 78) were observed to be significantly overrepresented in ruminants, while types 45 and 47 were significantly overrepresented among environmental isolates. PFGE type 9 was significantly overrepresented in chicken and underrepresented in humans.

Table 3.   Distribution of Campylobacter jejuni pulse field gel electrophoresis (PFGE) types among isolates from human clinical cases, chicken, ruminants and environment
PFGE type†Number of Camp. jejuni isolates‡Total no. of isolates
  1. †PFGE types that were not randomly distributed among sources as determined by categorical analysis are labels with * (indicating P ≤ 0·05), or ** (indicating P ≤ 0·0005).

  2. ‡PFGE types that were significantly higher (+) or lower (−) in number for a specific source than expected based on random distribution as determined by categorical analysis are labelled with * (indicating P ≤ 0·05), ** (indicating P ≤ 0·005) or *** (indicating P ≤ 0·0005).

  3. §Data are for 130 Camp. jejuni isolates from retail chicken and broiler flocks in Iceland between 1999 and 2002.

  4. ¶Data are for 286 Camp. jejuni isolates from human clinical cases in Iceland between 1999 and 2002.

  5. ††Data are for 79 Camp. jejuni isolates from ruminants, cattle and sheep isolated in Iceland between 1999 and 2002.

  6. ‡‡Data are for 22 Camp. jejuni isolates collected from soil and water in Iceland between 1999 and 2002.

1**56 (+)***601 (−)***0 (−)***118
7**1347 (+)***0 (−)***065
9**16 (+)***3 (−)***0020
15*1 (−)*1210 (+)***023
41*02 (−)*5 (+)***07
42*22 (−)*5 (+)***09
45**0012 (+)***7
47**00*06 (+)***6
56033 (+)*06
57*034 (+)**07
59*186 (+)*015
78**025 (+)***07
116**0 (−)*20 (+)***0020

Diversity of Campylobacter jejuni MLST and flaA-SVR genotypes

A subset of 52 isolates was selected for characterization by MLST and fla-SVR sequencing, representing the PFGE genotype and source distribution of the entire sample. The selection included the 13 most prominent PFGE genotypes, covering roughly 60% of the total PFGE diversity. Table 4 shows the diversity of the 52 isolates as depicted by the three different typing methods. All 52 isolates typed could be assigned to clonal complex (CC) although three isolates could not be assigned to sequence type (ST). A total of 21 STs, assigned to nine previously described CCs, were identified among the isolates. Eleven STs represented a single isolate. Two new allele sequences were identified for the pgm and tkt loci, and four new STs previously unreported in the MLST database were identified, two from sheep, one from chicken and one from human isolate, representing 7·7% of the isolates (Table 4).

Table 4.   Comparison of multilocus sequence typing (MLST), pulse field gel electrophoresis (PFGE) and flaA-SVR sequencing for the discrimination of a randomly selected subset of 52 Campylobacter jejuni isolates
IsolateSourceMLSTFla-SVR allelePFGE type
  1. *New alleles and STs.

  2. ST, sequence type; ND, allele not defined; NA, ST not assigned; CC, clonal complex; SVR, short variable region.

E 464/00-1Cattle212132155321842
E 643/99APoultry2221321548921521
E 697/99ASheep212132155321521
E 508/00-1Cattle21213215532110359
E 695/99ASheep4710417385270*45515
E 671/99Poultry47104171454587
E 614/99Poultry1047104171132645217
E 859/99Sheep241271548483215
E 862/99Sheep241271NDNA483242
E 1129/00Poultry1422ND317NA614242
E 695/99BSheep1423432*3173666*614259

All isolates were grouped by MLST among nine previously characterized CCs. The four most prevalent CCs (containing five or more isolates), ST45, ST21, ST48 and ST206, accounted for 84% of all the isolates. The most common of these was the ST45 complex representing 38% of all isolates (with six STs), followed by the ST21 complex representing 28% of the isolates (with two STs). CCs ST48 and ST206 contained 10% of the isolates each with two and three STs, respectively. Within three of these predominant CCs (ST21, ST45 and ST48), a single ST accounted for 50–93% of the isolates. Collectively, these dominant STs accounted for 58% of all the isolates (Table 5). The remaining five CCs contained only one or two isolates (ST61, ST1287, ST177, ST677 and ST682).

Table 5. Campylobacter jejuni multilocus sequence typing genotypes among 52 strains isolated from various sources in Iceland (1999–2002)
CCsNo. of different subtypesSource (no of isolates)Genotyping (no of isolates)
  1. ST, sequence type; PFGE, pulse field gel electrophoresis; CC, clonal complex; SVR, short variable region.

ST45197101 1664
ST21144712 254
ST486222  233
ST206532   324
ST612 11  212
ST1772    2111
ST6821    1111
ST6771 1   111
ST12872 1  1222

Fla-SVR sequencing was performed on all isolates to increase the resolution of the MLST analysis. A total of 18 fla-SVR types were observed among the 52 isolates analysed. PCR amplification could not be obtained from two isolates. Majority of the fla-SVR alleles or 67% were associated with one or two isolates. The most commonly detected allele was allele 52 representing 36% of the isolates. The chicken isolates were more homogenous as compared to the human isolates as was the case with entire set of isolates. The 16 human isolates were divided between ten fla-SVR alleles, while the 22 chicken isolates also contained ten alleles.

Association between MLST genotype and source

All STs obtained were queried against previous records in the MLST database (database queried May 2011). All human isolates and 87·5% of the chicken isolates analysed were assigned to four predominant CCs: ST45, ST21, ST48 and ST206. Three additional CCs were observed among the chicken isolates: ST61 (n = 1), ST677 (n = 1) and ST1287 (n = 1). The environmental samples analysed in our study were generally associated with different CCs than observed from livestock and humans. The environmental samples were assigned to the following CCs: ST682, ST1287, ST177 and ST45.

Comparison of genotype diversity by PFGE, MLST and fla-SVR sequencing

PFGE macro-restriction with SmaI from the 52 isolates revealed 13 genotypes, compared to 21 STs by MLST and 18 by fla-SVR sequencing. The discriminatory index (DI) of the genotyping methods varied from 0·875 for MLST to 0·909 for fla-SVR, and 0·886 for PFGE (Table 6). Including the fla-SVR allele as an eighth additional locus to the MLST scheme increased the DI for the method to 0·953 (Table 6).

Table 6.   The discriminatory index (DI) (Simpson’s diversity index) for the three different Campylobacter jejuni genotyping methods
Typing methodNo. of isolatesDITotalNo. of subtypesNo. of isolates in the most common subtype
>10 Isolates>5 Isolates1 Isolate
  1. PFGE, pulse field gel electrophoresis; MLST, multilocus sequence typing; SVR, short variable region.

MLST + SVR500·9532801189
PFGE + SVR500·95433102410

An upgma dendogram was constructed from the 52 Camp. jejuni MLST profiles. The 14 STs from ST21 complex gave four PFGE types and five fla-SVR alleles. The 19 isolates within ST45 complex gave five PFGE types and seven fla-SVR alleles and the six isolates from ST48 complex gave three PFGE types and fla-SVR alleles, respectively (Fig. 2). Many of the isolates indistinguishable by PFGE were further discriminated by MLST, and similarly, a single ST could be associated with different PFGE types (e.g. ST53, ST45 and ST48). There was no simple linear relationship between the typing methods, and PFGE and fla-SVR could predict CCs in 71 and 62% of cases, respectively (data not shown).

Figure 2.

upgma tree constructed from multilocus sequence typing allelic profiles of 52 Campylobacter jejuni isolates by START2 (sequence type analysis and recombinational tests).


The aim of the study was to evaluate the source-dependent distribution and population diversity of Camp. jejuni in Iceland collected during a national epidemic and identifying the main sources of domestic Campylobacter infections in humans. The characterization of Camp. jejuni isolates from multiple sources besides human and chicken has not been previously reported from Iceland. Such characterization is necessary for understanding the subtype diversity and source distribution in the Icelandic Camp. jejuni population.

PFGE revealed significant diversity in the Icelandic Camp. jejuni population. Two PFGE genotypes (type 1 and type 7), however, dominated the population and covered 34% of the total number of isolates. Type 1 was temporally distributed, and 96% of the type 1 isolates were isolated during 1999 and 2000, during which time the type 1 genotype constituted about 70% of the chicken isolates and 42% of the domestic infections. After 2001, this genotype was no longer observed which could possibly be linked to epidemic deflation in this period (Jore et al. 2010).

Type 7 was detected from humans and chickens throughout the 4-year sampling period.

The genotypic diversity of the chicken population was less than that observed in isolates from other sources. Campylobacter populations in chicken have, however, previously been shown to be more diverse than populations from other sources by MLST and PFGE (Dingle et al. 2001a; Manning et al. 2003). The homogenous nature of the chicken population in this study is the result of the common and widespread distribution of the type 1 PFGE genotype among chicken isolates. PFGE type 1 is suspected to be the outbreak strain associated with broiler-related Campylobacter epidemic in Iceland during 1999–2000. The genetic profiles as depicted by PFGE, MLST and fla-SVR sequencing support that. Of the ten PFGE type 1 strains isolated during 1999–2001, all belonged to the same ST (ST53) except for one isolate that had a single nucleotide variance at the gln locus. Furthermore, all the PFGE type 1 strains except for one were assigned to fla-SVR allele 52. The two dominant PFGE types were almost exclusively associated with human and chicken isolates (99·5%). It can be speculated that the majority of type 1 and type 7 strains isolated from humans originated from consumption and handling of contaminated chicken.

A general overlap was observed between the Icelandic veterinary and human Campylobacter populations as has previously been demonstrated from the Icelandic human and chicken populations using fla-SVR sequencing (Callicott et al. 2008) as well as other geographical populations (Dingle et al. 2001a; Manning et al. 2003; Levesque et al. 2008). Chicken, cattle and sheep PFGE types were all found to be associated with human disease, but the strongest association was with chicken. The degree of shared genotypes between chickens, cattle and humans in Iceland observed in this study is in line with previous observations (Hanninen et al. 2000; Dickins et al. 2002; Nadeau et al. 2002; Karenlampi et al. 2003; Hook et al. 2004; Sheppard et al. 2009). In total, 77% of the domestic clinical cases were associated with chicken and cattle, indicating the prominence of these two sources for human infections. Consumption of chicken or cross-contamination has been shown to be the prominent source of Campylobacter infections by risk assessments (Nauta and Havelaar 2007) and observational studies based on molecular typing (Nielsen and Nielsen 1999; Kramer et al. 2000; Wilson et al. 2008; Sheppard et al. 2009), including a study based on fla-SVR sequencing on Icelandic chicken and human isolates collected between 2001 and 2004, where 84% of the human isolates had genotypes also found in broiler flocks (Callicott et al. 2008). In total, clinical strains in our study were found to be genetically related to chicken isolates in 65% of the cases, and when looking solely at retail chicken, 96% of all isolates were found to be associated with clinical strains. PFGE types from environmental samples, sheep and ducks were also associated with clinical isolates in our study, indicating that various sources could be implicated in sporadic cases of human disease.

Some degree of source association was established in the present study. These observations indicate, as previously reported, that there is host-specific distribution of certain Campylobacter genotypes (Manning et al. 2003; Levesque et al. 2008). Specific associations between PFGE types 45 and 47 were observed with environmental samples. PFGE types 15, 41 and 42 were overrepresented in cattle and PFGE type 1 overrepresented in poultry. Substantial genetic differentiation has been reported between isolates from chicken and cattle, and MLST studies indicate that infection between ruminants and chicken does rarely occur (McCarthy et al. 2007). Our results show that PFGE genotypes found in cattle are rarely found in chicken, as only 4·7% of the chicken isolates were found in both sources. On the other hand, 28% of the cattle PFGE types were also found among chicken isolates. This could suggest that a broader variety of Camp. jejuni strains are capable of colonizing chicken as compared to cattle and are in line with the commonly greater diversity reported from chicken Camp. jejuni populations.

The 52 isolates for molecular subtyping using MLST and fla-SVR sequencing were selected to represent the diversity of the entire subset as depicted by PFGE typing. Genotyping by MLST sequencing has not previously been published from Icelandic Campylobacter isolates. The isolates were assigned to nine previously described CCs. Five of these nine CCs (ST21, ST45, ST48, ST206 and ST61) are among the six lineages previously most strongly associated with human disease (Dingle et al. 2002). Four of these CCs (ST21, ST45, ST48 and ST206) dominated the sample, constituting 84% of the entire subset. All four CCs are large in the MLST database and well documented from various sources and different geographical regions. ST21 is the most common CC in the MLST database both among clinical and animal isolates. It is thought to form a complex group of related genotypes along with ST complexes 206 and 48 that are widely distributed and capable of colonizing variety of hosts (Dingle et al. 2001a, 2002). In total, ST21 complex and related CCs account for 48% of the entire subset of isolates. The remaining five CCs detected in our study collectively represented 16% of the isolates. Two isolates belonged to ST61 complex, one from sheep and the other from poultry. The ST61 complex has previously mainly been associated with sheep, cattle and clinical isolates and is uncommon in chicken. Only 2% of ST61 complex genotypes in the MLST database are associated with chicken. ST682 and ST177 complexes were found in environmental samples, while complexes ST1287 and ST677 were found in broiler chicken. The ST677 complex has been associated with a variety of sources, including chickens and chicken offal. The remaining three CCs have previously been most strongly associated with environmental sources and wild birds in the MLST database. The ST1287 complex has not been detected in broiler chicken before and has been exclusively associated with wild birds. The ST682 and ST177 complexes have mainly been isolated from wild birds and sandy beaches although the latter has also been linked to sporadic cases in humans. A caveat on representativeness of the isolates in the MLST database should be made in relation to the discussion described earlier. Our analysis and comparison of STs and isolate origin in the MLST database are subject to the representativeness of the isolates in the database and the continuous renewal of the database as new isolates, and datasets are added.

The CCs in our study showed considerable divergence from PFGE banding pattern and fla-SVR genotype. CCs in weakly clonal bacterial populations as that of Campylobacter are considered to be derived from common ancestors (Holmes et al. 1999). Overall, there was considerable PFGE type and fla-SVR diversity within CCs, but neither was reliable indicators of genetic relatedness in terms of clonal lineages. This is in line with previous observations for fla-SVR (Dingle et al. 2001b, 2002; Price et al. 2006; Levesque et al. 2008) and PFGE (Sails et al. 2003b; Djordjevic et al. 2007; Levesque et al. 2008; Griekspoor et al. 2010). PFGE and fla-SVR sequencing are highly discriminatory genotyping methods and valuable for resolving relationships between closely related isolates. The rapid rate of recombination and genomic rearrangements reported within the Camp. jejuni genome and the hypervariable nature of the fla-SVR locus, however, make them ill suited for establishing population structure and studying long-term epidemiology (Hanninen et al. 1998; Wassenaar et al. 1998; Sails et al. 2003a). Applying a more variable typing scheme as the fla-SVR locus to the MLST housekeeping genes can, however, substantially increase the discriminatory ability of the genotyping method and its applicability for short-term epidemiological studies on Camp. jejuni and possibly facilitate source tracing and outbreak investigations (Dingle et al. 2001b; Michaud et al. 2001; Sails et al. 2003a; Clark et al. 2005). The discrimination of MLST in our selected isolates was 0·875 and comparable to that of PFGE which was 0·886. This reduced discrimination, as compared to that of the entire set of isolates, is because of PFGE type 1 being overrepresented in our sample. PFGE type 1 isolates all belonged to ST21 complex resulting in decreased resolution of isolates. Adding the highly variable fla-SVR as an eighth allele for the MLST analysis, however, resulted in a substantial increase in strain discrimination, increasing the DI from 0·875 to 0·953, allowing it to be useful for differentiating epidemiologically unrelated isolates.

To conclude, PFGE analysis of the Camp. jejuni population from various sources in Iceland reveals substantial genetic diversity. Two PFGE genotypes exclusive to humans and chicken dominated the samples, and type 1 can be suspected to be the outbreak genotype of 1999–2000. Clinical cases were strongly associated with PFGE types from chicken isolates as well as cattle isolates. MLST and fla-SVR sequence typing of a selected subset of isolates allowed the comparison of isolates with the international MLST database which has not performed with Icelandic Campylobacter isolates before. The population was found to consist of common geographically and temporally distributed CCs. Four dominant CCs that all are predominant in the MLST database and associated with different sources accounted for 84% of the isolates, and ST21-related complexes accounted for 48% of the isolates. Although the number of isolates subjected to MLST was limited, the method proved highly applicable, gave valuable insight into the structure of the Icelandic Camp. jejuni population and has provided motivation for further regional and global comparison with extended data sets.


This work was supported by a grant from the Icelandic National Research Council.